Microfluidic-based 3D cell culture for studies of biophysical and
ARCHNES
biochemical regulation of endothelial function
by
Vernella V. V. Vickerman
Masters of Science in Chemical Engineering Practice (M.S. CEP)
Massachusetts Institute of Technology, Cambridge, MA, 2006
Bachelors of Science in Chemical Engineering (B.S.)
Howard University, Washington D.C., 2003
SUBMITTED TO THE DEPARTMENT OF CHEMICAL ENGINEERING IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY IN CHEMICAL ENGINEERING
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
JUNE 2012
@ 2012 Vernella V. V. Vickerman. All rights reserved.
The author hereby grants to MIT permission to reproduce and to
distribute publicly paper and electronic copies of this thesis
document in whole or in part in any medium now known or
hereafter created
Signature of Author:
Department of Chemical Engineering, MIT
March 12, 2012
Certified by:
Roger D. Kamm
Cecil and Ida Green Distinguished Professor of Biological and Mechanical Engineering, MIT
Thesis Supervisor
Certified by:
Paula T. Hammond
David H. Koch Professor in Engineering
Department of Chemical Engineering, MIT
Thesis Supervisor
Accepted by:
William M. Deen
Professor of Chemical Engineering
Chairman, Committee for Graduate Students
Thesis Committee
Advisor:
Roger D. Kamm
Cecil and Ida Green Distinguished Professor of Biological and Mechanical Engineering
Director of EBICS
Massachusetts Institute of Technology
Advisor:
Paula T. Hammond
David H. Koch Professor in Engineering
Department of Chemical Engineering
Massachusetts Institute of Technology
Associate Editor, ACS Nano
David H. Koch Institute for Integrative Cancer Research
Chair:
William M. Deen
Carbon P. Dubbs Professor and Graduate Officer
Department of Chemical Engineering
Chairman, Committee for Graduate Students
Massachusetts Institute of Technology
Member:
Jeffrey T. Borenstein
Director Biomedical Engineering Center
Charles Stark Draper Laboratory
Member:
Carlos E. Semino
Assistant Professor of Bioengineering
Universitat Ramon Llull
2
Microfluidic-based 3D cell culture for studies of biophysical and biochemical
regulation of endothelium function
by
Vernella V. V. Vickerman
Submitted to the Department of Chemical Engineering
March 12, 2012 in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy in Chemical Engineering
Abstract
New and more biologically relevant in vitro models are needed for use in drug
development, regenerative medicine, and fundamental scientific investigations. The
ultimate challenge lies in replicating the native cell/tissue environment ex vivo. Certain key
features of living tissues such as the three dimensionality, biophysical and biochemical
microenvironment cannot be readily replicated in traditional culture platforms. Moreover,
the capability for multi-parameter manipulation, on a single platform, with the optical
resolution to monitor the dynamics of individual cells or small populations is lacking.
In this thesis, we developed a novel multiparameter microfluidic-based cell culture
platform. The system permits 2D or 3D culture of cells on/in biologically-derived or
synthetic hydrogel scaffolds and allows for controlled flow rates, pressure and
concentration gradients while directly visualizing cellular response. In addition to the realtime and post-fixation imaging using optical microscopy, methods were developed to
extend post-fixation analysis to transmission electron microscopy (TEM).
The platform was subsequently used to demonstrate for the first time, two microfluidicbased 3D in vitro assays with direct relevance to tumor development and glaucoma. For the
first assay, biochemical induced sprouting was demonstrated. Endothelial cells sprout from
an intact monolayer to form multicellular capillary-like structures. Furthermore, using
time-lapse microscopy the cellular dynamics during sprouting angiogenesis were observed
with great detail, showing tip cell dynamics, cell division events and lumen formation. Of
particular relevance to tissue engineering community, we demonstrated that endothelial
cells when cultured for several days can assemble into vascular networks with open,
perfusable lumen. Using this new system, we present novel findings and results supporting
a potential mechanism for flow-mediated mechanical regulation of angiogenesis by
3
transendothelial fluid flow. We demonstrate that flow direction is sufficient to define an
angiogenic ON or OFF state. The balance is tipped by forces generated at mechano-sensitive
cell-matrix adhesions involving FAK-mediated signaling. These results provide one
explanation for the bias towards angiogenesis occurring from the venous side of the
circulation. For the second assay, an aqueous humor (AH) outflow model was developed.
Subsequent proof-of-concept experiments confirmed its capability for studying the role of
the inner wall endothelium in the regulation of AH outflow dynamics.
Thesis Supervisor: Roger D. Kamm
Title: Cecil and Ida Green Distinguished Professor of Biological and Mechanical
Engineering, MIT
Thesis Supervisor: Paula T. Hammond
Title: David H. Koch Professor in Engineering Department of Chemical Engineering, MIT
David H. Koch Institute for Integrative Cancer Research
4
DEDICATION
To my Family
The two most Influential Women in my life
Amanda Evans - "mama"
Janett Evans - "mommy"
My Devoted Husband
James Paul Kelley
My Adorable Daughter
Vahnyah Vickerman Kelley
Thank YOU
For
Your Sacrifices
YourLove
And
Your Endless Support!
"The most authentic thing about us is our capacity to create, to overcome, to endure, to
transform, to love and to be greater than our suffering."
-Ben Okri
5
Biographical Note
Vernella Vickerman
Education
e
Chemical Engineering Doctoral Candidate, Massachusetts Institute of Technology,
Cambridge, MA
*
*
Master of Science in Chemical Engineering Practice (MSCEP), Massachusetts Institute of
Technology
Bachelor of Science (B.S.) in Chemical Engineering at Howard University, Washington
DC (summa cum laude GPA: 4.0/4.0)
Industrial Experience
*
"
Project Consultant Glaxo-Smith- Kline Pharmaceuticals (GSK, England). Drug Particle
Characterization and Dissolution Kinetics
Project Consultant. General Mills (GM, Minnesota, USA). Dough Rheology and Process
Parameter Optimization
Research Experience
"
*
*
*
Graduate Researcher, Chemical Engineering, MIT.
Thesis Advisors: Dr. Roger D. Kamm and Dr. Paula Hammond
Undergraduate Researcher, Chemical Engineering Department, Howard University.
Advisor: Dr. Ramesh Chawla
Research Project: Soil Remediation, Trichloroethylene (TCE) dissolution kinetics and
reaction kinetics in soil and aqueous systems.
Undergraduate Researcher, Howard University Physical Chemistry Lab
Advisor: Dr. Helen de Clercq
Research Project: Ion Optics for Cross Beam Apparatus
Undergraduate Researcher, Howard University Chemistry Laser Laboratory.
Advisors: Dr. Joshua Halpern & Dr. Helen de Clercq
Research Project: Photoablation of Polymer Films: Characterization of potential
polymer substrate for the support and deployment of Solar Sails.
Teaching Experience
*
Teacher's Assistant for 10.25: Industrial Chemistry and Chemical Process Pathways,
MIT
"
"
"
Teacher's Assistant for Chemical Engineering Analysis, Howard University
Teacher's Assistant for Chemical Engineering Calculation, Howard University
Personal Tutor: Algebra, Geometry, Differential Equations
6
Students Supervised
*
e
e
*
e
*
*
Johnathan Cyr, currently graduate student at BU
Cathy Nguyen, NECO Program. Currently graduate student at SCCO
Joseph Atnafu, MIT UROP
Pablo Rosado, currently Graduate Student at Berkeley
Christine Yu, currently Medical Student at BU
John Casey, AMGEN UROP Program. Currently MIT Graduate Student
Emily Yiqi Li, MIT UROP. Currently a Consultant at Oliver Wyman
Other Positions Held
*
"
"
"
Project Consultant. Kamm Lab NUS, Singapore
MIT Summer Research Program (MSRP) Group Leader and Program Assistant
Public Service Fellow, Supervisor Don Monahan; Project: Methods to integrate Reading
in Science Curriculum - Create a Database of Supplementary Material for 7th and 8th
Grade Cambridge Public School Science Curriculum
Howard University - Baltimore/Washington Area Community College Chemistry REU
Program Research Mentor. Advisor: Dr. Helen de Clercq
Supervised and provided support for a group of community college REU students.
Awards and Honors
e
"
"
e
e
"
"
*
*
*
*
"
*
"
"
*
MD Anderson TRAMCEL Best Poster Award
Molecular, Cell & Tissue Biomechanics Trainee
NIBIB Research Grant Supplement
MIT Public Service Fellowship
Keith and Helen Rumbel Fellowship
Tau Beta Pi Fellowship
National Capital Section AIChE and Exxon Mobile Research and Engineering Award
Award of Excellence by Howard University Chemical Engineering Faculty
Howard University Academic Achievement Highest GPA
Tau Beta Pi Record Scholar
Merck Index award
Tau Beta Pi Scholarship for College of Engineering, Architecture and Computer Science
sophomore of the Year
Tau Beta Pi Scholarship for College of Engineering, Architecture and Computer Science
freshman of the year
National Deans' List
Howard University Trustee Scholarship
Archie Alexander Scholarship
7
Publication List
1.
Vickerman V, Blundo J, Chung S, Kamm RD. Design, fabrication and implementation of a
novel multiparameter control microfluidic platform for three-dimensional cell culture and
real-time imaging. Lab Chip, 2008, 8, 1468-1477. (LF.=6.478;2008),selected Top 10 Articles
for Lab on a Chip in September (with 779 access), [Cited by 66]
2. Chung S, Sudo S, Mack PJ, Wan C-R, Vickerman V, Kamm RD. Cell migration into scaffold
under coculture conditions in a microfluidic platform. Lab Chip, 2009, 9(2):269-75.
(I.F.=6.478;2008), selected Top 10 Articlesfor Lab on a Chip in January(with 589 access),
[Cited by 74]
3.
Mack PJ, Zhang Y, Chung S, Vickerman V, Kamm RD, Garcia-Cardena G. Biomechanical
regulation of endothelium-dependent events critical for adaptive remodeling. J Biol Chem.
4.
Sudo R, Chung S, Zervantonakis IK, Vickerman V, Toshimitsu Y, Griffith LG, Kamm RD.
Transport mediated angiogenesis in 3D epithelial coculture. FASEB J, 2009,23, 7, 2155-
2009, 284(13):8412-8420. (IF.=5.52;2008), [Cited by 16]
2164. (LF.=7.049; 2008), [Cited by 31]
5.
Abdul Rahim NA, McDaniel W, Bardon K, Srinivasan S, Vickerman V, So PTC, Ho Moon J.
Conjugated Polymer Nanoparticles for Two-Photon Imaging of Endothelial Cells in a Tissue
Model. Advanced Materials,2009,21(34):3492-3496 (May 2009). (IF.=8.191; 2008), [Cited
by 14]
6.
Chung S, Sudo R, Vickerman V, Zervantonakis IK, Kamm RD, Microfluidic platforms for
studies of angiogenesis, cell migration, and cell-cell interactions, Annals of Biomedical
Engineering,2010, 38(3), 1164-1177 (LF.=2.605; 2008), [Cited by 13]
7.
8.
Vickerman V, Kim C, Kamm RD MicrofluidicDevicefor Angiogenesis in Mechanical and
Chemical Signaling in Angiogenesis Book Chapter 2011 Submitted
Vickerman V, Kamm RD, Mechanical regulation of angiogenesis Under review
Conference Proceedings and Presentation
1.
2.
3.
4.
5.
Vickerman Kelley V, Chung S, Wan CR, Mack P, Kamm RD, 2007, Microfluidic Bioreactors
for Studying Angiogenesis, The Annual Hilton Head Workshop, Hilton Head, GA, 05/2007,
Poster Presentation
Vickerman Kelley V, Kamm RD, 2007, Microfluidics Bioreactor: A Platform for Studying
Capillary Morphogenesis in Response to Biochemical and Biophysical Cues, SBC2007,
Keystone, Colorado, Podium Presentation (PhD Paper competition - Honorable Mention)
Vickerman Kelley V, Kamm RD, 2007, Microfluidics Bioreactor: A Platform for Studying
Capillary Morphogenesis in Response to Biochemical and Biophysical Cues, NIBIB Grant
Trainees Meeting, Keystone, Colorado, Poster Presentation
Vickerman Kelley V, Kamm RD, 2007, In vitro microfluidic platform for the study of
sprouting angiogenesis, BMES2007, Los Angeles, California, Podium Presentation
Casey JP, Vickerman Kelley V, Kamm RD, 2008, Optimizing Vascular Morphogenesis in a
Self-Assembling Peptide Hydrogel, IBE 2008, PosterPresentation
8
6.
I.Zervantonakis, S.Chung, R.Sudo, V.Vickerman, P.Mack, C.Wan, R.D.Kamm, 2008, A Novel
Microfluidic Bioreactor for Studying Capillary Morphogenesis, MEMS@MIT, Cambridge, MA,
04/2008
7.
R.D.Kamm, V.Vickerman, S.Chung, R.Sudo, I.Zervantonakis, 2008, Regulating Capillary
Morphogenesis in a Microfluidic System, BME2008, London, UK, keynote lecture
8. R.D.Kamm, V.Vickerman, S.Chung, I.Zervantonakis, S.de Valence, R.Sudo, 2008, Microfluidic
systems for mimicking in vivo conditions, ICBME2008, Singapore,12/2 008, keynote lecture
9. R.Sudo, S.Chung, I.Zervantonakis, V.Vickerman, Y.Toshimitsu, L.Griffith, R.D.Kamm, 2008,
Microfluidic platform for the three-dimensional coculture of hepatocytes and endothelial
cells, BMES 2008 Annual Fall Meeting, St. Louis, MO, 10/2008, PosterPresentation
10. P.J.Mack, Y.Zhang, S.Chung, V.Vickerman, R.D.Kamm, G.Garcia-Cardena, 2008, Collateral
Flow Regulates Endothelial-dependent Events Critical for Adaptive Remodeling, NIBIB
Grant Trainees Meeting, Bethesda, MD, 06/2008, Poster Presentation
11. R.D.Kamm, R.Sudo, V.Vickerman & S.Chung, 2009, Using microfluidics to study
coordinated cell population behavior, ICCB 2009, Bertinoro (Forli), Italy, 9/16-9/18, invited
speaker
12. V. Vickerman, J. Cyr, C. Yu, H. Gong, R Kamm, 2010, Direct Imaging of Giant Vacuole
Dynamics of Schlemm's Canal Endothelial Cells Using a Novel In Vitro Microfluidics-Based
3D Cell Culture System, ARVO 2010, Ft. Lauderdale, FL, 5/2010, Podium Presentation
13. V. Vickerman, RD Kamm, 2010, A Flow-gated angiogenesis switch, MIT Biological
Engineering Annual Retreat, Newport, RI, 03/2010, Poster Presentation
14. V. Vickerman, RD Kamm, 2010, A Flow-gated angiogenesis switch, NIBIB Grant Trainees
Meeting, Bethesda, MD, 06/2010, Poster Presentation
15. V. Vickerman, RD Kamm, 2010, A Flow-gated angiogenesis switch, WCB 2010, Singapore,
08/2010, Podium Presentation
16. V. Vickerman, RD Kamm, 2010, Eye-on-a-chip: Applications for Glaucoma Research, MIT
BioMAN, 11/2010, Poster Presentation
17. V. Vickerman, RD Kamm, 2011, Investigating the effects of the biophysical
microenvironment on tumor angiogenesis, SPRBM 2011, Poster Presentation, MD
Anderson TRAMCEL Best Poster Award.
18. R.D. Kamm, V. Vickerman, W Polacheck, 2011, Going against the flow:
Mechanotransduction in the interstitium, BMES 2011, Keynote Presentation.
PATENTS AND TECHNOLOGY DISCLOSURE
*
PCT Application: Three-dimensional microfluidic platforms and methods of use thereof.
Filed April 8, 2009 based on a Technology Disclosure filed April 8, 2008. Published
October 15, 2009.
*
Technology Disclosure: Surface modifications, patterning of surface modifications, and
novel manufacturing methods for a microfluidic platform containing a threedimensional scaffold. Filed Oct 6,2009.
SCIENTIFIC AND PROFESSIONAL SOCIETIES
*
*
Society for Physical Regulation in Biology and Medicine
The Association for Research in Vision and Ophthalmology (ARVO)
9
e
e
*
*
*
e
e
e
e
e
National Organization for the Advancement of Black Chemist and Chemical Engineers
NOBCChe - MIT Chemical Engineering Representative
Member of NOBCChe
President, Academy of Courageous Minority Engineers (ACME)
Biomedical Engineering Society (BMES)
Golden Key International Honour Society
Vice President of American Institute of Chemical Engineers (AIChE)
Recording Secretary of Tau Beta Pi DC Alpha Chapter
Member of Tau Beta Pi Honor Society
Member of National Society of Black Engineers (NSBE)
Member of AIChE
VOLUNTEER WORK
"
e
*
e
e
ACCESS: A Community in Chemical Engineering Select Symposium, MIT. Program
targets under-represented minority students, and provides an overview of the potential
benefits of a graduate chemical engineering degree.
Roxbury YMCA: Design Summit, interactive workshop for young people interested in
design, landscaping and urban planning.
MIT ESP: SPARK. Program targets middle and high school student.
Department of Chemical Engineering and ODGE Graduate Recruiting
iMATH after school program, mentors help 8th graders to understand and appreciate
math.
10
Acknowledgement
"I will lift up my eyes unto the hillsfrom whence cometh my help. My help cometh from the Lord, which
made heaven and earth." - Psalms 121:1:2
It would not have been possible to complete this journey without the support of the people around me. I
would like to thank my thesis advisors Roger D. Kamm and Paula T. Hammond for their support. I am
especially thankful to my primary advisor, Roger for giving me the opportunity to work in his lab and
scientific freedom to pursue my passion. I am grateful to my thesis committee members Jeffrey T.
Borenstein, William M. Deen and Carlos E. Semino for their time and dedication.
To my professors at Howard University Ramesh C.Chawla, Joseph N. Cannon, John P. Tharakan, William
E. Collins, Robert
J. Lutz, M. Gopala Rao, Mobolaji E. Aluko, Helen L. deClercq, Joshua B. Halpern, Jesse M.
Nicholson and Oladapo Barake thank you for preparing me for this journey.
To the current and past members of the Kamm Lab, thank you. I am especially grateful to Alisha SarangSieminski. Thank you for your constant and endless support throughout the years.
To my student accountability group, ACME, thank you creating a sense of community. Aisha, Reggie,
Legena, Melva, Zenzi, Melissa, Obi, Joy, Isaac, Rhonda, Robbin, Shauni, Lincoln, Eric thanks for sharing
your journey with me.
To Dean Blanche Staton, thank you for your wisdom, vision, love and support.
To my "adopted" family and relatives, Leisa, Aunt Becky, Uncle Barry, Twila, Bruce, Grandma Stutzie,
Grandpa Ollen thank you for love and prayers. To Nigel Neil, thank you for you love and support.
Finally to my family mommy (Janett), mama (Amanda), Paul and Vahnyah words cannot adequately
express my appreciation and gratitude. I dedicate this thesis to you.
11
Blank Page
12
Table of Contents
Thesis Committee
2
Abstract
3
DEDICATION
5
Biographical Note
6
Acknowledgement
11
List of Figures
17
List of Tables
19
Chapter 1:
a.
Introduction and Background Information
20
Endothelial Cells
i.
20
Vascular Endothelial Microenvironment
Endothelial Mechanotransduction
ii.
1.1.1
21
23
Implication of Endothelial Cells in Disease Pathogenesis
26
b.
Motivation for in vitro model systems
27
c.
Microfabrication Biological Sciences *
30
i.
Advent of Microfluidics
30
ii.
Microfluidic devices for angiogenesis
Microfluidics for tissue engineering applications
31
iii.
d.
41
Thesis Aims and Overview
Chapter 2:
41
Microfluidic-based3D cell culture platformsfor biologicalstudies**
45
a.
Chapter Abstract
45
b.
Introduction
46
Material and Methods
49
c.
i.
ii.
iii.
iv.
v.
vi.
vii.
viii.
d.
Microfluidic Bioreactor Design
Device Fabrication and Surface Modification
Scaffold Loading Microinjection System and Device Assembly
Demonstration of Concentration Gradient across Gel in p1FD with and without Channel Flow
Cell Culture and in vitro capillary morphogenesis models
49
Characterization of capillary morphogenesis and tube-like structures
Microvascular Endothelial Cells Sprouting Movies
54
50
50
52
53
55
Cytoskeleton and Nuclei Staining
56
Results
i.
ii.
iii.
iv.
3D microfluidic bioreactor (ptFD)
Concentration gradient maintained in ltFD
Microfluidic cultures, model system: capillary morphogenesis
ECs invade 3D matrix to form capillary-like structures reminiscent of sprouting angiogenesis
56
56
59
60
__
62
13
v.
vi.
ECs suspended in 3D collagen gels form capillary-like structures
2D surface migration promotes formation of tube-like structures
64
65
e.
Discussion and Conclusions
67
f.
Acknowledgements
69
g.
Supplementary Material
70
i.
Optimization of gel cage geometry
70
ii.
PDMS surface treatment facilitates scaffold microinjection
70
iii.
71
iv.
Finite element analysis for diffusion of a non-reactive solute in microfluidic device
Simultaneous control for fluid flow and gradients
v.
vi.
Time-Lapse Video-Microscopy#
Supplementary Figures
73
74
71
Appendix I: Implementation of microfluidic-based 3D cell culture platform in coculture
and imaging studies
80
Appendix III: Biochemical Regulation of Angiogenesis: Vascularization of Scaffolds in a
Microfluidic Cell Culture Platform
vii.
Biomaterial Scaffold
82
82
Appendix II: Microfluidic Device Design Gallery
92
Chapter 3:
A Model for Aqueous Humor Outflow across the Inner Wall of Schlemm's
Canal: Direct Imaging of Giant Vacuole Dynamics using an in vitro Microfluidic-based
Platform
93
a.
Introduction
93
b.
Background
i.
Aqueous humor circulatory pathway
ii. Aqueous Humor Outflow and Glaucoma
iii. The Inner Wall of Schlemm's Canal
iv. Juxtacanalicular Tissue (JCT)
v.
Ex vivo aqueous humor outflow models
94
94
95
96
98
100
c.
Materials and Methods *
i.
Cell Handling, Maintenance and Culture
106
106
ii.
3D Matrix
106
iii.
iv.
v.
vi.
vii.
hSCEC monolayer formation and perfusion
Visualization of Giant Vacuole Dynamics
Evaluation of Fluid Hydrodynamic Filtration Pattern
Effect of pharmacological inhibitor Y27632
Post fixation Imaging
107
108
109
110
110
d.
Results
113
14
i.
3D Matrices for hSCEC culture: Collagen type I and Matrigel
113
ii.
Primary Schlemm's Canal endothelial cells on 3D matrix as a model for AH outflow and formation
of Giant Vacuole-like Structures
116
iii.
Real-time Visualization of Giant Vacuole Dynamics
118
iv.
v.
vi.
Pharmacological Inhibition
Hydrodynamic Filtration Pattern
Ultrastructure by new TEM method for microfluidic samples
120
122
124
e.
Discussion and Conclusions
127
f.
Appendix: Supplementary Material
130
g.
Appendix: hSCEC Maintenance and Microfluidic Protocol
138
Chapter 4:
Biomechanical regulation of angiogenesis **
151
a.
Chapter Abstract
151
b.
Introduction
152
Materials and Methods
154
c.
d.
e.
i.
Cell Culture
154
ii.
Microfluidic - based cell culture platform
154
iii.
Imaging
155
iv.
EC monolayer formation and characterization
155
v.
In vitro transendothelial flow angiogenesis assay
156
vi.
Evaluation of signal transduction at cell-matrix adhesions
156
vii.
Localization of signaling protein
viii.
Pharmacological Inhibition
ix. Antibody Labeling and Counterstaining
157
158
159
Results
i.
Formation of a functional EC monolayer
ii. Switching response evoked by transendothelial flow
159
159
162
iii.
164
iv.
Multiple signaling pathways are involved in B-A flow induced sprouting angiogenesis
B-A flow activation occurs via FAK-mediated signaling
v.
Transendothelial flow direction differentially affects VE-cadherin localization
168
vi.
vii.
Transendothelial flow direction differentially affects actin cytoskeletal organization
VE-Cadherin delocalization is mediated by Src
169
171
166
Discussion and Conclusions
i.
Flow direction acts as an angiogenic switch
ii. EC activation is due to mechanical stimulation
172
173
173
iii.
B-A Flow increases FAK-mediated signaling at cell-matrix adhesions
B-A flow induced VE-Cadherin delocalization and angiogenesis are mediated by Src
174
iv.
v.
vi.
Reorganization of actin cytoskeleton by transendothelial flow
Force estimates acting on the monolayer
177
177
176
15
vii.
f.
Summary Proposed Model
Supplementary Information
i.
Characterization of EC monolayer transport properties
178
180
180
g.
Appendix IV: Concentration Polarization
191
h.
Appendix: Immunolabeling Protocol for Microfluidic Samples
199
Chapter 5:
Concluding Remarks and Future Directions
201
a.
Summary of Thesis Contributions
201
b.
Future Research Directions
206
c.
Closing Thoughts
209
Chapter 6:
Works Cited
210
16
List of Figures
Figure 1-1: Vascular endothelial m icroenvironm ent ..........................................................................
Figure 1-2: Cellular M echanotransduction .........................................................................................
22
25
Figure 1-3: Role of Endothelial Cells in Atherosclerosis and Metastatic Cancer.............................27
Figure 1-4: H istory of cell culture .......................................................................................................
29
Figure 1-5: Schem atic of Microfluidic Device Fabrication: .................................................................
Figure 1-6: M icrofluidic-based Angiogenesis Assays .........................................................................
Figure 2-1: Schematic of [tFD and scaffold loading protocol. ...........................................................
Figure 2-2: Experim ental design and device capabilities ...................................................................
31
40
58
61
Figure 2-3: Microvascular endothelial cells sprout to form capillary - like structures...................64
Figure 2-4: 2D migration promotes sheet-like and tube-like structures...........................................66
Figure 2-5: Single suspended microvascular endothelial cells form multi-cellular structures..........67
Figu re 2-6 : Gel cage geom etry..................................................................................................................75
Figure 2-7: Scaffold delivery into microfluidic device via microinjection procedure. ....................
76
Figure 2-8: Distribution of a non-reactive solute in scaffold cage....................................................77
Figure 2-9: Evolution on the concentration profile across the "gel cage" with a constant flow rate in
m icro flu id ic ch an n els................................................................................................................................78
Figure 2-10: Evolution of liquid pressure differential across "gel cage" induces interstitial flow
through the three-dim ensional scaffold. .............................................................................................
79
Figure 2-11: Diverse Applicability of Microfluidic-based 3D cell culture platform.........................80
Figure 2-12: 3D encapsulation assay ...................................................................................................
86
Figure 2-13: HMVEC Sprouting Assay - RAD16-1 and RAD16-I plus PRG & KLT............................87
Figure 2-14: Capillary morphogenesis on RAD16-I/PRG/ KLT self-assembling peptide gels........88
Figure 2-15: Single cell migration through RAD16-I/PRG/KLT functionalized self-assembling
p e p tid e g e ls ................................................................................................................................................
89
Figure 2-16: Example of microfluidic design and mask layout .......................................................
92
Figure 3-1: Conventional flow pathway of aqueous humor...............................................................
Figure 3-2: Two main types of glaucoma, open-angle and angle-closure glaucoma .......................
95
Figure 3-3: Inner W all of Schlem m 's Canal .........................................................................................
Figure 3-4: Juxtacanalicular Tissue.....................................................................................................
98
Figure 3-7: Adhesion dynamics of hSCEC to pure Matrigel hydrogels...........................................
Figure 3-8: Adhesion of hSCEC to Collagen/Matrigel Blended gels ...............................................
114
96
99
Figure 3-5: Ex vivo anterior segm ent perfusion m odel........................................................................
102
Figure 3-6: In vitro SCE m onolayer perfusion m odel...........................................................................104
115
Figure 3-9: Comparison of Giant Vacuole morphology from in vivo and in vitro studies..................117
Figure 3-10: Giant Vacuole-like structure dynamics: Perfusion at constant flow rate......................119
Figure 3-11: Giant Vacuole-like structure dynamics: Perfusion at constant pressure......................120
17
Figure 3-12: Effect of Y27632 on giant vacuole-like structures ..........................................................
Figure 3-13: Variation in tracer labeling pattern .................................................................................
121
123
Figu re 3 -14 : B ead T races ........................................................................................................................
124
Figure 3-15: Light and Transmission Electron Microscopy.................................................................126
Figure 3-16: hSCEC spreading on collagen gels of different concentration........................................130
Figure 3-17: hSCEC monolayer extends membrane projections in underlying matrix.....................131
Figure 3-18: Expression profile of junction proteins in cultures hSCEC ............................................
133
Figure 3-19: Pressure gradient sensitivity of GVL structures .............................................................
134
Figure 3-20: Evaluation of bead passage through 3D matrix in "gel-cage" region ............................ 135
Figure 3-21: Evaluation of bead passage through 3D matrix ..............................................................
136
Figure 3-22: Experim ental estim ation of gel perm eability..................................................................137
Figure 3-23: Constant Flow Rate and Pressure Setups ........................................................................
147
Figure 4-1: Microfluidic-based 3D cell culture system for studies of the effects of transendothelial
flow on sprouting angiogenesis from an endothelial monolayer........................................................161
Figure 4-2: Transendothelial flow direction modulates EC monolayer phenotype...........................163
Figure 4-3: Inhibition of B-A flow induced angiogenesis by small molecule pharmacological
in h ib ito rs............................................................
. . . .............................................................................
16 5
Figure 4-4: B-A flow promotes activation of integrin-mediated adhesion in endothelial monolayers.
...............................................................
... -----.........................................................................................
16 7
Figure 4-5: B-A flow induces delocalization of VE-cadherin ...............................................................
169
Figure 4-6: B-A flow direction affects distribution of cortical actin....................................................170
Figure 4-7: Src mediates B-A flow induced remodeling at cell-cell junction......................................172
Figure 4-8: Expression profile of adherens (VE-Cadherin, RED) and tight (ZO-1, GREEN) junctional
proteins in HMVEC cultured on glass substrate (static culture) .........................................................
185
Figure 4-9: Sprouting response of microvascular (HMVEC) and macrovascular (HUVEC) endothelial
cells to B -A flow .. . . . ..................................
..........................................................................................
186
Figure 4-10: Estimation of EC Monolayer Diffusional Permeability (Pd)............................................187
Figure 4-11: Bead Tracer Method for Flow Characterization..............................................................188
Figure 4-12: Negative controls and secondary antibody cross reaction test. ....................................
Figure 4-13: Device geom etry in COM SOL ............................................................................................
190
Figure 4-16: Predicted Relationship between Polarization Distance and Pecletendo .........................
Figure 4-17: FEM solution for solute concentration as a function of distance...................................
196
192
Figure 4-14: Solute drag reflection coefficient .....................................................................................
194
Figure 4-15: Predicted Relationship between EC Monolayer Sieving Coefficient and Peclet Number
............................................................................... ---.......
. . . ................................................................
19 5
197
18
List of Tables
Table 1-1: Comparison of Blood, Lymphatic and Schlemm's canal endothelial cells ......................
Table 1-2: Pro- and anti- Angiogenesis Factors .................................................................................
21
35
Table 1-3: Angiogenesis related diseases............................................................................................
Table 2-1: Dimensions for microfluidic network...............................................................................
36
Table 4-1: FAK and p-FAK Y397 Antibody Evaluation.........................................................................
189
74
19
Chapter 1: Introduction and Background Information
a. Endothelial Cells
Endothelial cells (EC) form the innermost cellular lining of the circulatory and lymphatic systems.
Consequently, they are in direct contact with blood (blood/vascular EC), lymph (lymphatic EC) and
circulating cells (e.g. neutrophils) (1). In the eye, there is a special class of EC known as Schlemm's
canal endothelial cells that are in contact with neither blood nor lymph but a special fluid known as
aqueous humor (AH) which functions as the blood analog for ocular tissues. Table 1.1 (2) provides
a comparison of different class of EC with regards to structural characteristics, transmural flow
direction and "priority" during development. Within the vascular tree however there is some
structural heterogeneity in the EC monolayer and are classified as continuous, fenestrated or
discontinuous (1). Furthermore, transmural flow direction is reversed at postcapillary venules and
potentially within tumor microenvironments.
The embryonic origin of vascular and lymphatic endothelial and the regulatory mechanisms
underlying cell fate determination is widely agreed upon (3). Differentiation of hemangioblasts
gives rise to angioblast (endothelial progenitor cells) which subsequently produces endothelial
cells. Following arterial-venous specification a subpopulation of venous cells further differentiates
to produce lympatic endothelial cells. The origin of Schlemm's canal endothelial cells has also been
debated. Due to recent compelling evidence it is now accepted that they originate from the
intrascleral venous plexus and not the neural crest as previously believed (2).
20
Comparison of BECs, LECs, and SCECs
BEC
Endotheial monolayer
cond m
Basement membrane
Pericytes
Basement membrane
thickness
Monolayer thickness
interendothelial junctions
Cell-cell heterotypic
junctions
Extracellular structures
Flow direction
Development
Comparison of BEC(.
feat ure.
Continuous
Yes
50nm
0.2-0.3 pm
Tight junctions,
adherens junctions.
gap juncdons
Paicytes
None
Apical to basal
Primary
SCEC
LEC
Contnuous
Diondnuous
No
Condnma
60-80 nm
400nm
0. 1-0.6pm
Macula adherens
No
s 1 pm
None
Tight junctions
gap junctons,
adhrs junctims
JCT
Anchoring filaments
Basal-to-pial
SeCOndary
Tendoms septs
a1sal-S-picar
Secondary
Lt Cs. and SCECs. Highlighted seti ons indicae the EC type thati is more similar to SCEC for ech given
Ramos et al., J Glaucoma (2007) 16. 391-405
Table 1-1: Comparison of Blood, Lymphatic and Schlemm's canal endothelial cells
Table reproduced from Ramos et al., 2007 (2).
i.
Vascular Endothelial Microenvironment
Hemodynamic Environment. Endothelial cells are naturally found in a mechanically active
environment. In vivo, vascular endothelial cells constitute the lumen of blood vessels and in small
capillaries they compose the vessel with some pericyte coverage. Consequently, ECs are exposed to
various mechanical forces including shear stress (blood flow), compressive (blood pressure acting
perpendicular to vessel wall) and tensile (cyclic stretch from pulsatile flow) forces.
Chemical Environment. Products of circulating cells, interstitial cells and components of plasma
give rise to a complex chemical vascular environment (Figure 1-1). In neoplastic tissues, tumor cells
produce a cocktail of additional factors that influences endothelial phenotype.
21
Other factors. The structure of blood vessels varies along the vascular tree which gives rise to
different local environments. Arteries, arterioles, venules and veins are composed to three layers
(a) Intima: the endothelium and associated basement membrane; (b) Media: smooth muscle cells
and elastic connective tissue and (c) Adventitia: connective tissue containing collagen and elastin.
Blood
Endothelium
Smooth muscle
- Red blood cells
- Extracellular
- Signaling molecules
-
1eukocytes
- Platelets
matrix
- Cell-cell
contacts
- Signaling molecules
Blood vessel
- Shear stress
TV
Figure 1-1: Vascular endothelial microenvironment
Image and text reproduced from van der Meer et al. 2009 (4). Schematic overview of a blood vessel
and the endothelial cell microenvironment. The inner wall of a blood vessel (left) consists of a layer
of endothelial cells that are embedded in a three-dimensional microenvironment (right). This
environment consists of cell biological, biochemical, and physical stimuli, such as red and white
blood cells, signaling molecules, and shear stress, respectively. Mimicking this complex
microenvironment in vitro is a major challenge in vascular research.
22
ii.
Endothelial Mechanotransduction
Mechanotransduction refers to the mechanisms by which mechanical forces are converted to
biochemical signals. Most tissues and cells have the capability to sense and respond to mechanical
stimulation. Mechanotransduction signaling is vital in mechanically stressed tissues including
muscle, bone, cartilage and blood vessels (5). The underlying mechanisms have been investigated in
sensory cells (6); cardiac myocytes (7); endothelial cells (8), (9), (10); vascular smooth muscle cells
(11) and tumor cells (12). In addition cells are also able to generate stresses. Moreover, mechanical
stress influences a wide range of biological processes important for embryonic development (13)
and adult physiology (for review on vascular physiology see (14)) and diseases including deafness,
atherosclerosis, heart failure, glaucoma and cancer to name a few (see (15), (5) excellent reviews).
At the cellular level mechanical forces impact cell signaling and functions important for adhesion
apoptosis, proliferation, migration and differentiation (5).
Mechanotransduction in endothelial cells have been studied extensively with both in vitro and in
vivo models (10). Studies have shown that endothelial cells alter their morphology, growth rate and
metabolism in response to fluid shear stress (Ueda et al., 2004; Chen et al., 1999; reviewed in
Davies 1995). Moreover, shear stress profiles (e.g. pulsatile flow vs. oscillatory flow) differential
affect cell function (16), (17), (18). For example, oscillatory flow characterized by low mean shear
stress and high shear stress gradient was found to up-regulate P-selectin, ICAM-1 and MCP-1 mRNA
levels in bovine aortic endothelial cells (BAEC) and promote monocyte adhesion when compared to
pulsatile flow (17).
23
1.
Cellular Mechanosensors
There is a growing list of mechanosensory domains and mediators of mechanotransduction that are
available to cells (Figure 1-2), including (5), (14): (I) Cell Membrane (e.g. Stretch-Activated
Channels, membrane receptors, and caveolae), (II) Cell-matrix adhesions (e.g. Integrin, Focal
Adhesion Complexes), (III) Cell-cell adhesion (e.g. Cadherins), (IV)
Glycocalyx, primary cilium, streocilia),
components
(V) Cytoskeleton, (VI)
Surface processes
Cell Nucleus and (VII)
(e.g.
ECM
(e.g. fibronectin). Furthermore there are interactions between these different
mechanosensors, for example shear-stress evokes cross-talk between integrins, ion channels and Gproteins in endothelial cells (19).
24
o0 Calcium and other ions
e0 Cell-signalling molecules
and transcription factors
00 Extracellular ligands
Fluid shear stress
Stretch-activated
ion channels
complex
ECM
|Cell-surface
receptor
Tissue strain
Nature Reviews Molecular Cell Biology
Figure 1-2: Cellular Mechanotransduction
Image and text reproduced from Jaalouk and Lammerding Nature Reviews Molecular Cell
Biology (5). Several biological components, not mutually exclusive, have been proposed to act as
cellular mechanosensors and are schematically depicted in a representative cell (see figure). a I
Stretch-activated ion channels. b I In endothelial cell, the glycocalyx. c I Cell-cell junctional
receptors. d I Extracellular matrix (ECM)-cell focal adhesions. e I Force-induced unfolding of ECM
protein. f I Intracellular strain induced conformational change in cytoskeletal elements. g I Nucleus.
h I Compression of intercellular space alter effective concentration of autocrine and paracrine
signalling molecules.
25
1.1.1
Implication of Endothelial Cells in Disease Pathogenesis
Among the top 5 leading causes of death in the U.S. for 2009 are diseases of the heart, malignant
neoplasms (cancer) and cerebrovascular diseases (20). For vascular-related diseases endothelial
dysfunction is reportedly a common denominator (21). Endothelial cells are not a passive barrier
between blood/lymph/AH and tissues, in contrast they perform many vital functional roles
important for tissue health and repair and maintaining vascular homeostasis (10). For example,
nitric oxide (NO) produced by vascular endothelial cells promotes vasodilation and inhibits
inflammation, thrombosis and smooth muscle cell proliferation all important defensive action
against atherosclerosis (Figure 1-3A) (22) a major risk factor for heart attacks, strokes and
peripheral vascular disease.
On the other hand, the primary cause of death from malignant
neoplasms is metastasis (Figure 1-3B)- the process by which cancer cells spread throughout the
body to establish new colonies in remote organs and tissues via three main routes of dissemination
which includes lymphatic vessels, blood vessels and serosal surfaces (23). It is widely accepted that
endothelial cells promote metastasis formation by enhancing the growth of primary tumor via
angiogenesis (discussed in details later) or by facilitating tumor cell invasion by honing tumor cells
to blood and lymphatic vessels (24).
26
A
Endothelial cells and Atherosclerosis: Roles in
Endothelial cells and Metastatic Cancer: Roles in
Leukocyte Adhesion and Plaque Development
Extravasation, Intravasation and Angiogenesis
FAWOt
BMmgtiond
*
~
4
Nopowradlon
~
Eftam~awn-Cwcueatoo
Gr-ng metasta
Jackson, Shaun P. 2011.
Nature Medicine 17 (11) 1423-1436
Bacac M. Stamenkovic I.2008.
AR Annu. Rev. Pathol.Mech. Dis. 3:221-47
Figure 1-3: Role of Endothelial Cells in Atherosclerosis and Metastatic Cancer
Involvement of endothelial cells in (A) atherosclerosis (image reproduced from Jackson 2011 (25))
and (B) metastasis (image reproduced from Bacac 2008 (23)).
b. Motivation for in vitro model systems
Ross G. Harrison, one of the pioneers of cell culture models wrote in 1907, "the immediate objective
of the following experiment was to obtain a method by which the end of a growing nerve could be
brought under direct observation while alive, in order that a correct conception might be had
regarding what takes place as the nerve fiber extends during embryonic development from the
nerve center out to the periphery." Like many scientific pursuits his work was motivated by an
27
existing debate - regarding the growth and development of the nervous system - and would only be
settled by irrefutable experimental findings. A glimpse of the process would reveal clues to
underlying mechanisms which could prove or disprove his theory. The potential for new scientific
breakthrough was quickly realized an in the years to follow isolated cells and tissue were used as
model systems to investigate processes relating to health and disease. The ultimate challenge lies in
replicating the native cell/tissue environment ex vivo.
Moreover, to improve human health, the translation of basic research from the bench top to the
clinics or bedside is a necessary yet an enormous challenge. For example, in the area of drug
discovery only a few agents survive the many hurdles from conception to the realization of
clinically useful therapies. In several instances, this repeat failure has been attributed to limitations
in model systems. Physiologically relevant and well controlled models that mimic the in vivo
microenvironment would enable better understanding and effective evaluation of new therapies. In
vivo models are physiologically relevant but inherently complex, hard to control and difficult to
decouple, potentially confounding results. In vitro models can provide more control but in most
cases lack physiological relevance. For example, in a majority of early endothelial functional studies
cells are cultured on noncompliant substrates which are coated with thin layers of adhesion
proteins. While very valuable for elucidating certain aspects, for example the mechanism of
mechanotransduction, in vivo, the endothelium interacts with a viscoelastic three dimensional
substrate-basement membrane. Consequently, continued improvements on in vitro models are
necessary to more closely replicate the native tissue/cell microenvironment.
28
THE HISTORY OF CELL CULTURE
ArnsToTLE liEOPHRM~US
sPortANTors
-
PLANT
Figure 1-4: History of cell culture
Image and text reproduced from Witkowski 1979 (26). A diagram drawn up by P. R. White to
illustrate the history of tissue culture and the relationships between different workers in the field.
It is divided vertically in areas of interest, with "animal" subdivided into "materials and methods"
(left) and "nutrients" (right).
29
c. Microfabrication Biological Sciences
*
(* some of the material in this section has been submitted as a part of a book chapter)
i.
Advent of Microfluidics
Microfluidics emerged as a multidisciplinary research field since its inception at Stanford University
where the technology was first applied to the fabrication of gas chromatographic air analyzers (27)
and for designing the nozzle component for the first inkjet printers by IBM (28). The foundation for
microfluidic fabrication was laid by the microelectronics industry that developed fabrication
techniques for creating high-resolution features in microelectronic components. Early microfluidic
systems were fabricated in silicon or glass using standard photolithography (other lithographic
techniques include electron lithography, X-ray lithography, ion lithography) and chemical etching
methods (29). In recent years, there has been a shift towards elastomeric materials that permits
rapid prototyping of microfluidic systems, discussed further in section 3 and illustrated in Figure
1-5 (30), (31).
Current applications for microfluidic devices span multiple disciplines, with applications in
biotechnology and biochemical processing (32); clinical and forensic analysis (33); combinatorial
chemistry (34); systems biology (35); tissue engineering (36), (37), cell-based biosensors (38);
diagnostics and personalized medicine (39); and embryo production (40). The biology community
has greatly benefited from these advances, and progress in microfabrication technologies has paved
the way for new approaches to manipulate and observe cells in microenvironments that more
closely mimic in vivo conditions. Microfluidic-based cell culture systems provide new capabilities
30
for continuous monitoring of dynamic processes, such as angiogenesis, at high spatial and temporal
resolution in a controlled microenvironment (41).
(a)
Silicon wafer
(b)
The coated SU-8
photoresist
uv
uv
uv
(c)
ka*a
Silicon wafer is placed on
a spin coater.
SU-8 photoresist is coated
onto silicon wafer.
For the patterning, SU-8
photoresist is exposed with UV.
Petri dish
-&-(g)
(e)
niastei nnold
Patterned SU-8
@
After baking and development,
SU-8 photoresist is developed.
(h)
IMF*
SU-8 master mold is placed
in Petri dish.
(d)
(d
PDMS is poured onto master
mold in the Petri dish
Iq
After curing, the PDMS slab is
peeled off from the master molds.
0I
~,S[
~icic
Punched PDMS slab is bonded
by air plasma treatment.
Figure 1-5: Schematic of Microfluidic Device Fabrication:
The microfluidic chip is fabricated using an SU-8 fabrication method (a-d) and replica molding (gh). (a) A silicon wafer is placed on a spin-coater to achieve the desired SU-8 resist film thickness. (b)
SU-8 photoresist is coated onto the silicon wafer. (c) For patterning, the SU 8 photoresist on the
silicon wafer is exposed to conventional UV (350-400 nm) radiation using a photomask. (d) After
baking and development, the SU-8 resist is developed with SU-8 developer, and an SU-8 master
mold is made. (e) The SU-8 master mold is placed in a large Petri dish for PDMS preparation. (f)
PDMS is poured onto the master mold in the Petri dish. (g) After curing, the PDMS replica is peeled
away from the master mold. (h) For bonding, the PDMS slabs are bonded with a cover slip by air
plasma treatment.
ii.
Microfluidic devices for angiogenesis
1.
Motivation
Angiogenesis is important in health and disease. During embryonic development, for example,
angiogenesis facilitates the expansion of the primitive vascular plexus, thus meeting the demands
31
for oxygen and nutrient requirements during embryogenesis. In healthy adults physiologic
angiogenesis occurs during endometrial and placental formation, wound granulation after injury
and hair follicle vascularization (42). Furthermore, angiogenesis has been implicated in and directly
linked to the pathogenesis of a growing list of diseases (43), (44). Given the far-reaching impact, it
is not surprising that the radical theory of angiogenesis proposed by Folkman during the 1970's has
spawned an entirely new field of both basic and translational research. Moreover, there is
considerable interest in understanding how to modulate vascularization and insights from
angiogenesis are used to design new therapies for cardiovascular diseases (45) or more boldly, the
in vitro creation of vascularized tissues or organs by implementing tissue engineering approaches
for replacement therapies.
2.
The Biology of Angiogenesis
a. Angiogenesis in development
Although the processes by which microvascular networks form are described in detail in other
chapters, a brief background is useful as motivation for in vitro studies in microfluidics to follow.
The embryonic vasculature is formed by two distinct processes, vasculogenesis - the de novo vessel
formation from endothelial progenitors, angioblasts- and angiogenesis - the expansion of a preexisting vascular network which occurs during the later stages of development. Interestingly, these
two processes of vascularization occur in distinct embryonic regions which are defined by the three
germ layers: ectoderm, mesoderm and endoderm. Vasculogenesis, which gives rise to the primitive
vascular
plexus, occurs
splanchnopleural
in tissue of a splanchnopleural
mesoderm),
while
angiogenesis
origin
predominantly
(includes
occurs
endoderm
in
tissues
and
of a
somatopleural origin (ectoderm and somatopleural mesoderm) (46).
32
b.
The pathways of angiogenic sprouting and network formation
Angiogenesis is a complex, multi-step process involving a series of well delineated steps. Once
endothelial cells have acquired an angiogenic phenotype, the following processes occur: (a)
protease production increases facilitating degradation of basement membrane; (b) directional cues
initiate migration towards the angiogenic stimulus; (c) proliferation; (d) tube formation; and (e)
maturation. The regulatory chemical signals produced at the onset of and during angiogenesis
originate from many cell types including tumor cells, fibroblast, keratinocytes and macrophages.
Up-regulation of pro-angiogenic factors and the simultaneous down-regulation of endogenous
angiogenesis inhibitors are normally triggered when the tissue environment becomes hypoxic (47),
(48) or inflammatory (49).
Oxygen sensing is important for many biological processes including development, pH
homeostasis and angiogenesis (50). Critical molecular mediators of hypoxia and cellular oxygensignaling pathways, hypoxia-inducible factor (HIFs), are known to directly activate the expression
of such pro-angiogenic factors as VEGF, PDGF-B, Ang-1, Ang-2 and receptors VEGFR-1, VEGFR-2 and
Tie-2 (48), (51), (see also Table 1-2). In particular, hypoxia induces a dramatic increase in VEGF
messenger RNA levels (52). VEGF is essential for embryonic vasculogenesis and angiogenesis, as
demonstrated by the observation that VEGF gene inactivation is lethal (53), (54). In vitro, VEGF
promotes EC proliferation, migration, differentiation and capillary formation. It is also widely
appreciated that tumor cells secrete angiogenesis-related proteins under hypoxic conditions (55),
(56). Recently, a linear correlation was established between hypoxic growth conditions and the
expression levels of eight angiogenesis-related proteins including VEGF, IL-8, PDGF-AA, PDGFAA/BB, TGF-P1, TGF-p2, EGF, and IP-10 (57). In addition, stromal cells promote angiogenesis (58),
(59) and also enhance vascular maturation (60).
33
Inflammatory cytokines and chemokines represent another group of signaling molecules
with critical roles in angiogenesis regulation during tumor growth, wound healing and ischemia
(49). Furthermore, chronic inflammatory conditions such as rheumatoid arthritis (RA)
are
considered as angiogenic diseases, as the excessive neovascularization contributes to their
pathogenesis.. Pro-inflammatory factors such as TNF-a, IL-8/CXCL8 and SDF-1/CXCL12 are known
angiogenic mediators while IL-4, LIF and PF4/CXCL4 exert an inhibitory role (61), (62).
For
extensive reviews on additional topics related to angiogenesis in inflammation see (63).
Positive/
Negative
Soluble Ligands/
Receptors
Impact on Endothelial Cells
Notes
VEGF
Chemotactic (64)
EC migration, proliferation,
apoptosis, differentiation
Targets: EC specific mitogen
Sources: Smooth
muscle cells,
keratinocytes, macrophages, tumor
cells
Gene Regulation: Oxygen tension,
Regulators
Pro-
VEGFR-1
VEGFR-2
1/KDR)
(Flt-1)
(Flk-
anti-
angiogenic
Factors
growth
factor
inflammatory
Angiopoietin-1
in
Tie-2
formation, survival
in
vivo:
Maturation,
recruitment of mural cells.
vitro:
EC
migration,
tube
(TFG-alpah,PDGF),
cytokines
(IL-1-
alpha,IL-6); Hormones (TSH, ACTH);
oncogenes (ras) CS
(66)
signal
FGF
Chemokinetic
Synergistically with VEGF.
bFGF binds FGFR-1
T EC proliferation
Broad spectrum, multiple cell types
aFGF
PDGF-BB
PDGFR-cta
(64)
T EC migration,
pretreatment
PDGFR-sp
PDGFR-ap
IL-8/CXCL8
Chemotactic
CXCR1
T EC proliferation
CXCR2
T EC migration
require
bFGF
Targets: VSMCs/pericytes mitogen
and chemoattractant.
bFGF sensitizes EC (67)
ELR-positive chemokine (68)
T MMP
Stromal cell-derived
factor 1
production
Chemotactic
in vivo; dermal angiogenesis
SDF-1/CXCL12
recruitment of EPCs
(69)
CXCR4
TGF-O
Induces
EC
apoptosis
T bFGF
(70)
expression, T VEGF synthesis
Hepatocyte growth
factor/scatter factor
HGF/SF
T EC proliferation
I EC migration
Down-regulate TSP-1 in tumor cells
(71)
Met
34
Antiangiogenic
Factors
Angiostatin
Endostatin
Inhibit EC proliferation
Inhibit EC proliferation
Interleukins
(IL-4,
IL-13)
Platelet
factor-4
(PF4)
CXCR3-B
Thrombospondin-1
TSP-1
Vasostatin
Inhibit
Inhibit
Inhibit
Inhibit
38kDa plasminogen fragment (72)
20kDa C-terminal collagen XVIII
fragment (73)
Cytokine (74), (75)
EC migration
tube formation
EC proliferation
EC migration
ELR-negative chemokine (76)
Inhibit EC migration
(77)
Inhibit EC proliferation
Fragment of calreticulin (78)
Table 1-2: Pro- and anti- Angiogenesis Factors
A partial list of biochemical factor that are known to be pro- and anti- angiogenesis.
c.
Angiogenesis in disease processes
In healthy adults, endothelial cells remain virtually quiescent - the result of a strict balance between
pro- and anti-angiogenic factors.
The transition from a quiescent to an angiogenic phenotype
occurs when this balance is disrupted and pro-angiogenic factors dominate. While physiologic
angiogenesis occurs in a tightly regulated fashion, deregulation of this process contributes to the
progression of a variety of diseases (Table 1-3).
The pathogenesis of the diseases may be
characterized by either an excessive or insufficient growth of blood vessels. Interestingly, while the
cellular and molecular regulator of normal and pathological angiogenesis have much in common,
the latter is characterized by uncontrolled, aberrant vascular growth (79). For example, VEGF
signaling plays a pivotal role in normal and tumor angiogenesis, however tumor vasculature is
chaotic and structurally abnormal (79).
35
Angiogenesis
Diseases
Notes/References
Excessive/Abnormal
Angiogenesis
Cancer
Hemangiomas
Retinopathies:
Proliferative
ischemic retinopathies: proliferative
diabetic
retinopathy
(PDR),
retinopathy of prematurity (ROP).
Age-related macular degeneration
Blindness, excessive pre-retinal
blood vessel growth leading to
retinal detachment. (80)
(AMD). Retinoblastoma
Atherosclerosis
Inflammatory rheumatic diseases:
Inflammatory disease. (81), (82),
(83)
(84)
Rheumatoid arthritis (RA)
Psoriasis
Chronic Inflammatory disease of
skin and small
Asthma
Endometriosis
Insufficient
Angiogenesis
Chronic
liver disease:
portal
hypertension
(PH),
fibrosis,
cirrhosis,
non-alcoholic
steatohepatitis
(NASH),
hepatocellular carcinoma
Coronary artery disease
Ischemic Stroke
Chronic wounds
Inflammatory rheumatic diseases:
Systemic sclerosis or systemic
joints.
(85)
Chronic Inflammatory disease:
(86)
Gynecological
disorder,
proliferation
of endometrial
glands and stroma outside
uterine cavity. (87)
(88)
(89)
(90)
(91)
(84)
scleroderma (SSc)
Table 1-3: Angiogenesis related diseases
A partial list of diseases characterized by either excessive or insufficient angiogenesis.
d. Angiogenesis in Tissue Repair
During tissue repair, for example in wound healing, angiogenesis is necessary for generating a
functional granulated tissue. In healthy individuals, this phase of wound repair progresses
naturally. However in certain situations, for example in diabetes, growth factor deficiencies,
36
impaired keratinocyte and fibroblast migration and proliferation, and accumulation of antiangiogenic glycation end-products in their tissues impair the angiogenic response and subsequent
tissue repair (92).
Furthermore, tissue damage is inevitable in diseases characterized by
inadequate vascular perfusion and interventions to correct vascular insufficiency are required. One
approach, therapeutic angiogenesis, has been extensively explored. Therapeutic angiogenesis is the
delivery of exogenous factors (e.g. small molecules, genes or cells) or the use of mechanical devices
to stimulate neovascularization, restore form and function to tissues (42)]. Growth factor therapy
with recombinant
proteins, gene
therapy using
non-viral
(e.g. plasmid
DNA,
liposomes,
nanoparticles) or viral (e.g. recombinant retrovirus, adenovirus, adeno-associated virus, Herpes
simplex virus type-1) techniques to deliver pro-angiogenic factors, tissue-engineered products,
hyperbaric oxygen and negative pressure wound therapy are current avenues for stimulating
angiogenesis (93), (42). Recombinant growth factors including VEGF, bFGF and PDGF-BB have
been used in animal models of chronic limb ischemia (94), (95). More recently, using plasmid-based
gene delivery systems, local intramuscular administration of FGF-1 was shown to be safe and
improved amputation-free survival in patients with critical limb ischemia (45). Hyperbaric oxygen
therapy (HBOT), the intermittent exposure of patients to 100% oxygen at pressures above 1
atmosphere, has been known to promote angiogenesis and collagen synthesis but the underlying
mechanisms remain unclear (96).
Another strategy, negative pressure therapy has been used
extensively in clinical management of wounds (97). The underlying mechanism by which topical
negative pressure stimulates angiogenesis has been investigated using in vitro methods and linked
to increased endothelial cell migration and proliferation (98), (99).
Angiogenesis is clearly a complex process requiring the coordination of multiple cell types and
integration of a host of chemical and mechanical microenvironmental signals. Despite these
37
enormous challenges small successes provide a map for future directions and motivation for
further pursuits in the application of angiogenesis principles for tissue repair and regeneration.
3.
Traditional in vitro and in vivo models of angiogenesis
Over time, investigators have implemented different types of in vivo and in vitro models in an effort
to recapitulate natural angiogenesis. The requirement for more affordable, reliable, reproducible
and well-characterized model systems has contributed to vast progress over the years. Despite
these advances, there still remain challenges with either approach, evident by the limited success in
the translation of basic research to the clinic or bedside. Nevertheless, the research community is
aware of these shortcomings, and has identified the extensive criteria that must be met (100) for
successfully recreating angiogenesis in the lab. Thus, the field continues to evolve, as newer models
are developed and older ones are refined.
In vivo systems:
Three main types of in vivo angiogenesis assay have been described; (1)
microcirculatory preparations in chick embryo and rodents; (2)
recruitment of vessels by
biocompatible polymer matrix implants; and excision of vascularized tissues (see (100) for an
extensive review]. One of the earliest in vivo angiogenesis models dates back to the late 1930s,
where Ide et al. (101)
demonstrated the vascularization
of an implant of Brown-Pearce
epitheliaoma using the transparent rabbit ear window developed by Sandison (102). Since then,
several other in vivo assays have been developed including cranial windows, chick chorioallantoic
membrane (CAM), corneal micropocket assays among others (100).
In vitro systems: In vitro assays traditionally take what could be viewed as a minimalistic
approach; the angiogenesis cascade is decomposed and investigated as the sum of its individual
steps, namely migration, proliferation and tube formation. Two research groups, Jaffe et al. and
38
Gimbrone et al. were among the first to report of the successful long-term culture of endothelial
cells in vitro (103), (104) . However, it was not until methods for the successful culture and clonal
expansion of endothelial cells (ECs) in vitro (105) were developed that the first in vitro
angiogenesis assay was established. In 1980, Folkman and Haudenschild demonstrated that cloned
capillary ECs cultured on a gelatinized Cuprak dish in tumour-conditioned medium could initiate
angiogenesis (106). Since then a variety of in vitro models have been developed in an attempt to
simulate and analyze the process of neovascularization. These in vitro assays can be broadly
classified as two-dimensional (2D) or three-dimensional (3D). In 2D models, cells are plated on
culture surfaces that are coated with thin layers of adhesion proteins whereas in 3D models cells
are cultured on or in 3D matrices. In 3D cultures, cells are able to invade or migrate within the
matrix, which better recapitulates the 3D nature of the in vivo microenvironment. A common theme
involves the culture of ECs in the presence of different extracellular matrix components and
angiogenic factors. Although these models have been extremely valuable for understanding several
aspects of the cellular and molecular mechanisms operative in angiogenesis, they lack the capability
of including other important factors such as chemical gradients, surface shear stress and interstitial
flows, as discussed in more detail later.
Physiologically relevant and well-controlled models that better mimic normal and pathological
angiogenesis would narrow the gap between bench-top discoveries and clinical applications. In vivo
models have physiological relevance yet inherently lack a high level of control. On the other hand, in
vitro models have the potential for greater degree of control, yet lack critical elements of the in vivo
microenvironment. There remains much room for improvement, and this has motivated many to
explore microfluidic methods in the search for greater in vivo relevance in an in vitro model.
39
4. MicrofluidicAngiogenesisAssays
Recently, the number of microfluidic-based 3D angiogenesis assays has increased (Figure 1-6).
Advances in microfluidic technologies have permitted new approaches for manipulating and
monitoring cells while allowing for the controlled simulation of factors known to influence vascular
sprouting in vivo. (For additional details see chapter 2)
A
Vickerman et al., Lab chip 2008
- fluid
environment
Barkefors et al., Lab chip 2009
3D hydrogel
-
lia
nnels
EC Sprout
PDMS
a
3D Matrix
Sprout in Microfluidic System
N
Chung et al., Lab Chip 2009
E
Shamloo and Heilshorn Lab chip 2010
Chung et al., Advance Material 2009
OOWWrs-n
Figure 1-6: Microfluidic-based Angiogenesis Assays
Examples of published microfluidic-based platforms used in different angiogenesis studies (41),
(107), (108), (109) and (110).
40
iii.
Microfluidics for tissue engineering applications
There is an increasing interest in understanding how to modulate vascularization in vitro for tissue
engineering applications. The development of biological substitutes that restore, maintain and
improve tissue function (111) holds promise as a solution to tissue failure and end-stage organ
failure. One of the major challenges of engineering larger tissue, however, is the need to create a
microvascular perfusion network to the newly generated or implanted tissue (112), (113), (114).
The microvasculature serves to supply oxygen and nutrients to the target tissue as well as to
remove metabolic waste. Tissues in the body overcome issues of oxygen and nutrient distribution
by being in close proximity ~ 100 pim of capillary beds. There is currently active research to
develop methods to engineer microvascular network with diameters in the range of small arterioles
(100 [tm) and capillaries (10pm) or tissue-engineered construct with an inherent vasculature
which would serve to connect to host vascular system upon transplantation (115), (113), (116),
(117). In this regard, knowledge gained from angiogenesis research can be used to develop new
strategies for vascular tissue engineering applications.
d. Thesis Aims and Overview
Thesis Aims
Aim 1: Microfluidic-based 3D Cell Culture Platform Design and Fabrication. To design and fabricate
an in vitro microfluidic system that: serves as a microfabricated substrate for 2D and 3D seeding of
cells in biomaterials; permits the delivery of controlled flow of biological fluids; pressure gradients
across biological scaffold; growth factor concentration gradients and facilitates real time
monitoring of cells' response to their microenvironment.
41
Aim 1b: Vascularization of Biomaterials in a Microfluidic-based Cell Culture Platform. To
demonstrate feasibility of microfluidic platform for the growth of vascular networks using synthetic
or biological-derived scaffolds.
Aim 2: Biophysical Regulation of Angiogenesis. To investigate potential mechanisms for basal-toapical transendothelial flow induced angiogenesis
Aim 3: A 3D in vitro model for aqueous humor outflow: Transport across the inner wall
endothelium. To develop a 3D in vitro model for aqueous humor outflow dynamics to characterize
giant-vacuole formation and visualize fluid transport pathway across Schlemm's canal endothelial
cells in response to basal-to-apical flow.
Thesis Overview
This thesis utilizes microfabrication technology to develop microfluidic-based platforms and
methods for facilitating three-dimensional (3D) culture of cells/tissue with control of physical and
chemical microenvironment. The main focus lays in the development and subsequent application
for different biological assays motivated by research relating to vascular tissue engineering,
glaucoma and cancer in the context of but not limited to endothelial cells.
CHAPTER TWO describes the development of a novel multiparameter control microfluidic-based
3D cell culture platform with the flexibility to control both the biochemical and biophysical
microenvironment. The system permits 2D or 3D culture of cells on/in biologically derived (e.g.
collagen, Matrige TM ) or synthetic hydrogel scaffolds (e.g. RAD16-I) and allow for controlled flow
rates, pressure and concentration gradients while directly visualizing cellular response. In addition
42
to the real-time and post-fixation imaging using optical microscopy, methods were developed to
extend post-fixation analysis to transmission electron microscopy (TEM).
The platform was subsequently used to demonstrate for the first time, the angiogenesis in a
microfluidics-based 3D cell culture system. With proof-of-concept experiments we demonstrated
that the cellular dynamics during biochemically induced sprouting angiogenesis can be observed
with great details. Furthermore, of particular relevance to tissue engineering community, we
demonstrated that endothelial cells when cultured for several days can assemble into vascular
networks with open, perfusable lumen structures.
Results from work with different collaborators which further demonstrates the broad applicability
of the platform developed in this thesis are presented in APPENDIX I of CHAPTER TWO.
The vascularization potential of a class of synthetic peptide hydrogels was also examined within the
microfluidic-based 3D cell-culture device, APPENDIX II of CHAPTER TWO. Experiments with
synthetic peptides (RAD16I and RAD16II) for the endothelial sprouting assay demonstrated that
the angiogenic response, in particular, the degree of invasion and sprout formation, was very low
compared to the widely used rat tail type I collagen gels. However, functionalized peptides PRG
(RGD binding motif) and KLT (mimics VEGF helix region, activate VEGF receptor) when added to
RAD161 promoted HMVEC invasion and the formation of capillary-like structures.
CHAPTER THREE further demonstrates broad applicability of the microfluidic-based 3D cell culture
platform developed in CHAPTER TWO. The application of the microfluidic-based 3D cell culture
platform for a glaucoma relevant assay is established. In this chapter, we report the development of
the first in vitro cell-based 3D aqueous humor (AH) outflow model and subsequent proof-of-
43
concept experiments demonstrating
its capability for studying the role of the inner wall
endothelium in regulating AH outflow dynamics.
In, CHAPTER FOUR the potential mechanisms by which basal-to-apical transendothelial flow
promote sprouting angiogenesis is examined. This work was partially motivated by the observation
that in vivo in tumor angiogenesis new vascular sprouts emerge predominantly from the low
pressure venous circulation. A proposed model for basal-to-apical flow induced angiogenesis was
presented based on evidence from protein localization, phosphorylation and pharmacological
studies.
Finally, concluding remarks, summary of thesis contribution and future research directions are
presented in CHAPTER FIVE.
44
Chapter 2: Microfluidic-based 3D cell culture platforms for
biological studies**
**Material in this chapter was originally published in a peer-review journal: Vickerman et al, "Design,
fabrication and implementation of a novel multiparameter control microfluidic platform for threedimensional cell culture and real-time imaging". Lab Chip, 2008, 8, 1468-1477. (selected Top 10 Articlesfor
Lab on a Chip in September (with 779 access), [Cited by 73 - 03/2012]
a. Chapter Abstract
New and more biologically relevant in vitro models are needed for use in drug development,
regenerative medicine, and fundamental scientific investigation. While the importance of the
extracellular microenvironment is clear, the ability to investigate the effects of physiologically
relevant biophysical and biochemical factors is restricted in traditional cell culture platforms.
Moreover, the versatility for multi-parameter manipulation, on a single platform, with the optical
resolution to monitor the dynamics of individual cells or small population is lacking. Here we
introduce a microfluidic platform for 3D cell culture in biologically derived or synthetic hydrogels
with the capability to monitor cellular dynamics in response to changes in their microenvironment.
Direct scaffold microinjection, was employed to incorporate 3D matrices into microfluidic devices.
Our system geometry permits a unique window for studying directional migration, e.g. sprouting
angiogenesis, since sprouts grow predominantly in the microscopic viewing plane. In this study, we
demonstrate the ability to generate chemical gradients, surface shear, interstitial flow, and image
cells in situ. Three different capillary morphogenesis assays are demonstrated.
Human adult
dermal microvascular endothelial cells (HMVEC-ad) were maintained in culture for up to 7 days
during which they formed open lumen-like structures which was confirmed with confocal
microscopy and by perfusion with florescent microspheres. In the sprouting assay, time-lapse
movies revealed cellular mechanisms and dynamics (filopodial projection/retraction, directional
45
migration, cell division and lumen formation) during tip-cell invasion of underlying 3D matrix and
subsequent lumen formation.
b. Introduction
Progress in microfabrication technologies has paved the way for new approaches to manipulate
and monitor cells in an environment that closely mimics in vivo conditions.
The in vivo
microenvironment is characterized by short distances to nutrient flow, a 3-dimensional (3D)
scaffold, and the presence of multiple cell types in proper spatial relationship. These features are
readily achievable by current microfabrication techniques and can be combined with appropriate
cell culture protocols. The major advantages of micro-scale cell culture systems or microfluidic
bioreactors are their ability to use small quantities of cells and reagents, precise control of spatial
and temporal environment and high resolution visualization of cellular events in real-time.
Microfluidic devices of varying levels of complexity are rapidly gaining popularity within the cell
biology community (38), (118), (119), (120). Most are 2D in the sense that the cells are plated on
top of a coated substrate.
Patterned substrates have been studied as well as surfaces with
topological variations (121), (122), (123).
These are useful for studying cell confinement,
alignment and cell-cell cell-surface or receptor-ligand interactions. Researchers now generally
agree, however, that many critical biological subtleties may be missed in 2D cell culture models
(124), (125), (126) and are seeking alternative approaches that incorporate a 3D gel or matrix.
The in vivo microenvironment is an integration of a complex array of global as well as local
stimuli and interactions that can be categorized as either biochemical or biophysical in nature.
46
Biochemical factors include the steady or time-dependent presentation of soluble factors as well as
functionalization of the 3D scaffold the cells reside either within or on top of. Some of the known
biophysical factors are the stiffness, density and pore size of the scaffold material, and features
related to flow, either surface shear stress or interstitial flow around the cells in the matrix or
through a surface layer of, say, endothelial or epithelial cells.
While the importance of the extracellular microenvironment is clear, the ability to
investigate the effects of physiologically relevant biophysical and biochemical factors is restricted
in existing 3D cell culture platforms. Current 3D in vitro models range from traditional multi-well
plate cultures (127), (128) to macrofluidic systems (129), (130), (131). Most are limited in the
scope of biochemical or biophysical factors that can be controlled, and often cannot be imaged at
high resolution during the course of an experiment. Better resolution has been achieved in microscale systems, for example, 3D gel microscope slide assays or cells cultured on 2D microfluidic
channels, but with an obvious compromise in the range of parameter control and/or, physiological
relevance. Recent advances in 3D microfluidic systems include the works of Koh and Pishko who
use photo-patterning techniques to fabricate cell-containing hydrogel microstructures inside
microfluidic channels (132) and Desai and Tan who took a microfluidic multilayer patterning
approach to created biomimetic 3D structures (133). Paguirigan and Beebe fabricated a novel
microfluidic device from gelatin which was used to culture mammary gland epithelial cell which
organized to form a more in vivo like 3-dimesional structures as opposed to cell culture on tissue
culture plastic (134).
In this study, we develop a novel microfluidic bioreactor (ptFBR) system for 3D cell culture
and demonstrate its capability for studying capillary morphogenesis. Capillary morphogenesis is an
important process that occurs in both physiological (e.g. wound healing and development) and
47
pathological (e.g. tumor growth) conditions. In addition, one of the major challenges facing the field
of tissue engineering is creating highly vascularized tissue with a functional vascular network. Each
of these processes has been extensively studied and shown to be regulated by a wide spectrum of
biophysical factors and sequenced biochemical cues.
Due to system constraints, however, each
individual study typically focuses on the effects of one or a small number of factors, making it
difficult to comprehend their potential interactions. Moreover, since capillary sprouting occurs
perpendicular to an existing monolayer, it is advantageous to have the capability to view the
monolayer from the side, so that capillary growth occurs in the viewing plane. To provide greater
flexibility and a higher degree of control, we have developed an in vitro ptFBR designed to permit:
(1) 3D cell culture, (2) the delivery of controlled time-dependent delivery of biochemical agents,
uniformly or with a spatial gradient, (3) the generation of pressure gradients and flows, both
surface and interstitial, (4) advantageous viewing angle perpendicular to a formed monolayer and
(5) real-time monitoring of cells in response to changes in their microenvironment.
Here we demonstrate the capability of our system for studying capillary morphogenesis;
however it can be used in the study of other cell types either as single cultures or co-cultures. We
know of no in vitro angiogenesis models and microfluidic systems which couple microfabricated
technology with the basic concepts of traditional 3D in vitro angiogenesis models and that are
geared towards promoting and studying capillary morphogenesis
in response the relevant
biochemical and biophysical features present in vivo.
48
c. Material and Methods
i.
Microfluidic Bioreactor Design
In order to satisfy these stated design criteria, a [tFD was created (Figure 2-1a) consisting of two
parallel microfluidic channels for delivering cell culture medium and a central "gel cage" with a
micro-pillar array for housing cell-seeded scaffolds. One primary objective was to facilitate the
incorporation of most commonly used injectable cell culture scaffolds (hydrogels), which may be
biologically derived (collagen gels or MatrigelT M ) or synthetic (self-assembling peptide gels) (135).
Gel cage geometry was optimized to permit the use of a range of hydrogel concentrations, with the
posts providing physical support to prevent the fracture of brittle gels and provide stability for soft
gels. The microfluidic network and gel cage were cast in PDMS which was bonded to glass after
loading the scaffold (see details below). With the scaffold in place the two media channels are
isolated from each other, communicating only through the gel containing region. This permits the
establishment of gradients in both pressure and chemotactic factors across the gel. Either static or
dynamic cultures can be established by integration with macro-scale plumbing. In dynamic
cultures, the bioreactor chip is connected to an external flow circuit which allows the control of
both surface shear stress and interstitial flow perpendicular to a cultured monolayer or past cells
suspended in the gel. The external platform is equipped with two isolated media reservoirs having
adjustable hydrostatic heads (producing a pressure gradient across the gel) at the inlet, two-way
valves (to isolate the system when necessary) and outlet connectors to a programmable push/pull
syringe pump (PHD 2000, Harvard Apparatus). Pressure gradients can also be generated by
connecting reservoirs directly to the PDMS wafer. The entire ptFBR platform was designed to be
49
small enough to fit on conventional microscope stage to allow real-time monitoring of cells over the
course of an experiment.
ii.
Device Fabrication and Surface Modification
The design of the microfluidic network was created in AutoCAD (Autodesk, San Rafael, CA) with the
dimensions of the microfluidic channels, gel "cage" and micro-pillars as provided in table 1. A
transparency mask was created from the CAD file with a minimum geometric feature size of
approximately
100
ptm and printed by a high-resolution printer
(PageWorks, MA).
This
transparency mask was used in photolithography of SU-8 photoresist to create a silicon wafer
master. Microfluidic devices were made by replica molding (31) polydimethylsiloxane (PDMS)
(Dow Corning, USA) and curing the degassed elastomer mix (10:1, base: curing agent) against the
silicon master in an 80 'C oven for 2 hours. Polymerized PDMS wafers were peeled off the silicon
master, individual bioreactor wafers (35 mm diameter, 0.8-1cm height) cut out and inlets and
outlets cored down to microfluidic channels using a sharpened flat-ended 16 gauge needle. Prior to
cell culture PDMS wafers were cleaned and sterilized at 120 'C for 20 minutes in a wet cycle
followed by a dry cycle at 120 'C for 35 minutes (20 min sterilization/ 15min dry). Next, the PDMS
surface was rendered hydrophilic by exposure to air plasma to facilitate scaffold loading (detailed
below). Briefly, sterilized wafers were placed on trays in plasma cleaner (Harrick, CA) chamber
(pattern side up). A pump down cycle (-
2 minutes) was initiated followed by irradiation for 2
minutes with pink plasma. Surface treated wafers were stored in sterile container and used within
0.5 - 2 hrs following plasma treatment.
iii.
Scaffold Loading Microinjection System and Device Assembly
50
A microinjection station was created to load the cell culture scaffold (sub ptL volumes) into the
device under aseptic conditions (Figure 2-1c). The system components included a manual
micromanipulator (MN-151 Joystick Micromanipulator with H-7 Pipette Holder, NARISHIGE, NY),
microliter syringe (Hamilton, 62RNR, 2.5 pL SYR, 22s/2"/3, VWR), digital microscope (Big Blue
QX5, COMPUVISOR.COM, TX) (all housed in a laminar flow hood) and a monitor for visual guidance.
The MN-151 joystick feature provided control of micro-scale adjustment in the XY plane with
additional coarse adjustments along the X, Y and Z axes.
Scaffold Microinjection. Sterilized PDMS wafers with their surfaces rendered hydrophilic as
described above are positioned on the microscope stage (patterned surface upward) with the "gel
cage" in clear view on the video monitor (Figure 2-1b). The tip of the microliter syringe (pre-loaded
with pre-polymer solution), attached to micromanipulator, is positioned a few microns above of the
"gel cage" and a small droplet of the pre-polymer solution is created manually and lowered until the
droplet first makes contact with the micro-pillars. Droplet size is controlled such that its diameter is
approximately equal to half the width of the "gel cage". Small droplets are created just above the gel
cage, lowered and dispensed; this process is repeated until the gel cage is full.
Scaffold Loading and Device Assembly. Gel pre-polymer solution (collagen type I, rat tail in these
experiments) is microinjected into the "gel cage" (details above); fluidic channels are sealed with a
clean glass cover slip (35 mm, VWR) and secured with a mechanical clamp (dynamic cultures when
connected to external platform). This is repeated for multiple wafers at a time. After scaffold
injection, assembled PDMS wafers are placed in a secondary humidified container, to prevent the
hydrogels from drying out. Gels are allowed to polymerize for 30 min at 371C in a humidified
incubator.
51
iv.
Demonstration of Concentration Gradient across Gel in pFD with and without
Channel Flow
Collagen gels were formed as previously described. Following polymerization, microfluidic
channels were filled with cell culture media (without supplements). Gradient studies were
performed under static (no flow) (Figure 2-2c) and dynamic (flow through the channels)
conditions (detailed in supplementary material) with the media in one microchannel replaced by a
dilute solution of fluorescent dextran (40 kDa, Invitrogen) at an initial concentration of 20 [tg ml-'.
Fluorescent intensity was visualized with a Nikon TE300 microscope (Nikon Instruments Inc., NY).
A series of fluorescent images (4X magnification) of the gel region were acquired with a
Hamamatsu camera (Hamamatsu, Japan) using Openlab (Improvision, MA) data acquisition
software and stored for further analysis. Images were processed to obtain the changes in
fluorescent intensity across the gel at each time point. Image processing of time-lapse fluorescent
images was performed using a custom written ImageJ (NIH public-domain image processing
software) macro and MATLAB (MathWorks, MA) code. Briefly, parallel diagonal lines were drawn
across the gel region from the sink to the source channel (Figure 2-2a) in each fluorescent image. A
clear path was defined which avoided the PDMS micro-pillars. Pixel intensities and corresponding
location from the "source" channel were recorded for these line sections. Average fluorescent
intensities were calculated for pixels at the same distances from the dextran channel for all pixels
across the length of the gel. At each time point a plot of the normalized average intensity profile
across the gel was generated. Variations in fluorescent intensity as a function of time were plotted
at fixed locations within the gel region and compared to theoretical predictions (detailed in
supplementary material) obtained from a finite element model generated in FEMLAB (Comsol,
USA).
52
v.
Cell Culture and in vitro capillary morphogenesis models
All cell cultures were maintained in a humidified incubator at 5% CO2 and 37'C. Human adult
dermal microvascular endothelial cells (HMVEC-ad, LONZA, USA) were propagated in EGM-2MV
media system with 5 % fetal bovine serum (LONZA formerly CAMBREX, USA). Cells were expanded
on collagen-coated flasks and used at passages 6-8.
3D encapsulation in collagen gels. For 3D encapsulation studies, cells well suspended at 1
x
106
cells ml-1 in ice-cold liquid type I rat tail collagen with a final gel concentration of 2mg ml-1. Liquid
collagen was prepared by adding collagen stock solution to a mixture of 1oX PBS, 1M NaOH and
tissue culture grade water to obtain a 2.5mg ml-1 solution. A predetermined volume of high density
cell suspension was then mixed with the collagen solution to obtain the required seeding density.
The collagen/cell mixture was loaded in a microliter syringe and gels cast as previously described.
Following gelation, microfluidic channels were filled with cell culture medium and incubated for 24
hours. To demonstrate the effects of biochemical factors, cells were cultured under static conditions
with complete media (control) or media enriched with pro-angiogenic factors (bFGF, VEGF and
PMA all at 50 ng ml-1). Cells maintained in cultures for several days in complete media or media
supplemented with bFGF/VEGF/PMA cocktail which were replenished at 24 hour time points. At
the end of an experiment samples were fixed, tagged with florescent markers and imaged (see
details below).
Surface shear stress on 3D encapsulated cells. To demonstrate the effect of biophysical stimuli,
cells were subjected to small levels of surface shear stress. EC populated gels were formed and
53
culture as previously described. A pressure differential (50 Pa) was imposed across the gel by
varying the difference in the height of culture media in the reservoir columns (Figure 2-2c).
Endothelial cell monolayer formation. Two different cell seeding protocols were used to control
the substrate on which ECs initially formed confluent monolayers, namely 2D and 3D substrate
monolayer seeding. Collagen gels were formed as previously described. Following gelation,
microfluidic channels were filled with a 2 mg ml-1 fibronectin coating solution and incubated
overnight. Prior to cell seeding the coating solution was replaced with complete media and
equilibrated for another 2 - 4 hours. A cell suspension of 2-3 x 106 cells ml-1 was flowed into one
microfluidic channel and the cells permitted to adhere to the rigid glass or compliant gel surface as
they settle out of suspension by gravity. Endothelial cells were cultured for 24 - 48 hours on the
rigid (2D monolayer seeding) or compliant (3D monolayer seeding) surface before further
treatment. Pro-angiogenic factors were either presented as a gradient or at uniform concentration.
For this assay VEGF (10 - 50 ng ml-1) and S1P (250 nM) were used to promote morphogenesis.
Shear stress on endothelial monolayer. Flows can also be generated within either or both of the
channels for the purpose of either maintaining a stable gradient across the gel region or subjecting
the cell monolayer to surface shear stress. To generate channel flow, the inlets are connected to
constant pressure reservoirs and the outlets to a syringe pump set to maintain a constant flow rate
(time-varying flows are also possible, but have not yet been tested). (See supplementary material
for details.)
vi.
Characterization of capillary morphogenesis and tube-like structures
54
Phase-contrast, epifluoresence and confocal microscopy were used to characterize capillary
morphogenesis and the 3D morphology of EC structures. Florescent and phase contrast images
were acquired on Nikon TE300 microscope equipped with a Hamamatsu camera and Openlab
image acquisition software. Time-lapse images were taken of live samples every 12 - 24 hours with
phase contrast microscopy. At the end of an experiment, samples were fixed with 4.0 %
paraformaldehyde (PFA) and tagged with a fluorescent marker for actin cytoskeleton and cell
nuclei. Confocal images were collected using spinning disk confocal microscope (Zeiss Axiovert
200M) furnished with Imaging Suite (PerkinsElmer Life Science) acquisition software. A series of
100 optical serial sections (1 ptm thick) were obtained. The aligned images were stacked and
rendered for 3D visualization using Imaris (Bitplane, MN).
vii.
Microvascular Endothelial Cells Sprouting Movies
To demonstrate the capability to monitor cells in real-time, time-lapse movies were recorded of
endothelial cells during sprouting angiogenesis. An endothelial monolayer was formed on collagen
gel as previously described. The ptFD was kept in a custom built environmental control chamber at
37 9C and 5% C0 2 and cells visualized with a Zeiss inverted microscope. To minimize evaporation
during the course of the experiment, media reservoirs (with zero height differentials) were
connected directly at each inlet and outlet ports. The device was then placed in a secondary
container with a humidified local environment and cut-out glass window in the bottom for
visualization. Bright-field images were taken with the AxioCam MRm (Carl Zeiss) (at single optical
plane) at 2 minutes interval with AxioVision image acquisition software.
55
viii.
Cytoskeleton and Nuclei Staining
F-actin distribution and number of cells involved in "capillary - like" network or tube structures
were assessed after 2-7 days culture in the ptFD. F-actin and nuclei staining were performed after
fixation with 4.0 % PFA (30 minutes). The fixed samples were rinsed twice with 1X phosphate
buffered saline (PBS), treated with 0.1 % Triton-X (1-2 minutes), rinsed with 1X PBS followed by
the infusion of a mixture of DAPI and rhodamine phalloidin (30 minutes) and a final wash step with
1X PBS.
d. Results
i.
3D microfluidic bioreactor (FFD)
A new ptFD was designed and fabricated from PDMS using standard soft lithography and replica
molding techniques. To facilitate precise control of the fluidic microenvironment (shear levels and
soluble factors), the device consists of two parallel flow channels and a central gel cage which was
designed to immobilize injectable scaffolds (soft hydrogels) for cell culture. A staggered array of
micro-pillars is incorporated in gel cage to provide mechanical stabilization for the scaffolds which
can biologically-derived or synthetic. We have found that this specific design provides optimal
mechanical support (supplementary material Figure 2-6) for gels and can sustain pressure
differentials in excess of several cm of water. With the gel in place the two microfluidic channels are
essentially isolated from each except for the diffusion or convection of soluble factors through the
porous scaffold from one channel to the other.
Creating a 3D in vitro microenvironment. Hydrogel scaffolds are directly microinjected in to the
CGCC. The microinjection protocol provided the ability to load minute volumes of scaffold material,
56
with or without cells, directly in the designated space without the need for the alternative perfusion
loading.
PDMS surface modification. PDMS was chosen for the device fabrication because of its
documented biocompatible and excellent optical transparency. However, after polymerization flat
PDMS surfaces are hydrophobic and exhibit poor wetability, which was not amenable to scaffold
confinement. Scaffold injection into untreated PDMS devices resulted in gels that exuded into the
channels and often did not fill the gel cage, resulting in small bubbles adjacent to the micro-pillars.
Fortunately, the hydrophobicity of PDMS is tunable and PDMS surfaces can be temporarily
rendered
hydrophilic by exposure to air plasma (136). Subsequent to plasma treatment,
hydrophobic recovery time is dependent upon the method of preparation and storage. Recent work
has shown thermal aging, longer oxidation time and storage in nitrogen to be effective in delaying
the recovery of hydrophobicity (137). Here PDMS was surface treated with air plasma for 2
minutes, longer than is typically required for immediate bonding to glass. To maintain the seal
when connected to macro-scale plumbing, typical microfluidic devices are permanently bonded or
vacuum sealed (138) to glass or a layer of PDMS to prevent leaks. Here neither technique could be
used since it was necessary to have direct access to the gel cage for scaffold injection. Instead, we
found that the plasma treatment used to contain scaffold spreading and filling was sufficient to
promote bonding to glass. We have found that the strength of the bond between PDMS and glass
increases over time and have experienced minimal problems with leaks.
57
Figure 2-1: Schematic of pFD and scaffold loading protocol.
(a) The pFD consists of two parallel channels for controlling fluid flow and a central gel cage for 3D
cell culture. One key feature of the gel cage is the staggered array of micro-pillars which provide
mechanical support for soft hydrogels. (b) Scaffold loading protocol. PDMS wafers were fabricated
as described in Materials and methods. PDMS wafers were cleaned, sterilized and surface treated
(step 1). Pre-polymer solution was precisely microinjected in the gel cage. Hydrophilic PDMS
surface permitted proper gel spreading and filling (step 2). Microfluidic channels were defined by
sealing device with glass cover slip. (c) Microinjection station in sterile laminar flow cell culture
hood. Precise scaffold loading can be achieved with a custom made microinjection setup which
includes digital microscope and xyz micromanipulator with microliter syringe. (d) Micrograph of
assembled [tFD showing media channels and collagen partially stained with coomassie blue. Both
biologically derived and synthetic matrices used for traditional macro-scale 3D cell culture can be
precisely immobilized in gel cage. With the gel in place media channels are virtually isolated except
for diffusive or convective transport of molecules through the porous matrix.
58
ii.
Concentration gradient maintained in ItFD
A typical time course of the concentration profile in the gel cage following the introduction of a
fluorescent dextran to one channel is shown in Figure 2-2a. Normalized florescence intensity (C Cmin)/(Cmax - Cmin) in the gel is plotted as a function of normalized distance (X/Xmax) from the dextran
(40 kDa) 'source' channel . Depending on the width of the gel region, a steady state concentration
profile was reached in the order of an hour. For gradients established under static (no flow)
conditions, we have demonstrated the ability to sustain a stable gradient for up to 40 h (without
replenishing either the source or sink channels). While a linear gradient can be maintained, there is
depletion and accumulation of dextran in the source and sink channels, respectively. In order to
maintain a constant concentration, it is necessary to continuously replenish the sink and source
channels. This was achieved by establishing a slow flow (10 pl min-1) in the two microfluidic
channels, one with control fluid and the other containing fluorescent dextran, to maintain a
gradient across the gel (Figure 2-9) for several hours. These results demonstrate our ability to
simultaneously control surface shear stress (on the surface of the endothelial monolayer) and
gradients in our device. In this study, 40 kDa dextran (non-reactive solute) was chosen because it is
similar in size to several growth factors of interest including VEGF, bFGF and IGF (139).
The experimental results for gradients under static conditions were compared to results from a
finite element (Figure 2-8) model, assuming a diffusion coefficient of 4
x
10-7 cm
2
s-1 with the gel
region. Fig. 2d shows the comparison between the theoretical results (solid lines) and experimental
results for two gel cage geometries. This value for diffusivity agrees well with the range of values
reported in the literature (140), (141). The ability to generate gradients of soluble factors across a
3-dimensional matrix provides the potential to simulate physiologically relevant mechanisms
during directional migration including sprouting angiogenesis, tumor metastasis and immune
59
response. The dynamic motility of migratory cells can be probed in a controlled microenvironment
and monitored in real-time. In addition, the spatial and temporal presentation of such factors
provides another level of control which would be physiologically relevant but not possible in most
current systems. Saadi et al. however, have recently demonstrated the ability to generate
concentration gradients in their microfluidic ladder chamber (142).
iii.
Microfluidic cultures, model system: capillary morphogenesis
As a case study we used our ptFD system to study microvascular endothelial morphogenesis. The
primary mechanism by which new blood vessels or capillaries are formed in vivo, angiogenesis
(143), involves a series of well delineated steps including matrix degradation, cell migration,
proliferation and lumen formation. This is a tightly regulated process that is affected by metabolic
stress (144), (145), mechanical stresses (146), (147), (148), soluble factors (149) and ECM matrix
components (150), (151). With this wide range of stimuli and morphological responses they elicit
this serves as a model system to demonstrate the unique capabilities of our [FD.
60
q
'U
(a)Demonstration of Gradient across gel "cage"
(b) 3D pFBR Cell
Seeding Modes
PDMS
0.8
(c) Cell culture configurations
scaffold
Cells on 3D
scaffold
droplet
micro-pillar
Cells in 3D
0.6
media
"ereservoirs
0
EL0.4
PDMS wafer in
mechanical
clamp
Cells on pchannel
E_0.2
glass
cover slip
Normalized Distance
1
-0.5
e 50 Pa
0.9
5
0.4
*100Pa-Expt.1
0.8
X
100 Pa - Expt. 2
1 0.7
CC
*
100.6
0 .3
xK
0.6
0-
0.2
*
.
*e
*c*
OA
)K)
J 0.3
6
0.2
0.1
20.1
2
0
0
Time (minutes)
0
0.2
0.4
0.6
Normalized Time
0.8
1
Figure 2-2: Experimental design and device capabilities
(a) Demonstration of gradient across gel cage. Collagen gels were formed as described in Materials
and methods. Fluorescent dextran (40kDa) was used to demonstrate the capability of generation
gradients in the ptFD. Shown are the time-course of fluorescent intensity and concentration of
dextran (used to simulate nonreactive solute within that size range) across the gel 'cage'. Plot (a)
shows representative experimental curves plotted for over 40 h. (b) Schematic of cell culture
assays. (top) EC sprouting assay. Cells are cultured on a 3-dimensional gel with physiological
relevant polarity. (middle) 3D encapsulation assay. Cells are suspended in the gel and are initially
separated from each other. (bottom) 2D migration assay. Cells form monolayer predominantly on
61
glass substrate (non-compliant) coated with ECM material (fibronectin). (c) Cell culture flow
configurations. (1) For static cultures droplets of media are place on inlet and outlet ports. Devices
are kept in local high humidity (Petri dish with water) secondary container in the incubator. (2)
Setup used for imposing pressure gradient across gel cage, differential in height of liquid reservoirs.
(3) Microfluidic platform. Schematic of platform used for generating physiological levels of shear
stress in micro-channels. (d) Plots represent values of normalized intensity at fixed locations in the
gel region, solid lines are theoretical predictions and symbols (circles (open-middle, solid-near sink
channel) and squares for devices D and C, respectively) (Table 2-1) are for experimental results
(detailed in the supplementary material section). (e) Experimental results for the evolution of
normalized pressure differential
(dP/dPmax) for generating interstitial flow through three-
dimensional scaffold in ptFD (detailed in the supplementary material section), values in Pa indicate
initial pressure differential.
iv.
ECs invade 3D matrix to form capillary-like structures reminiscent of sprouting
angiogenesis
HMVEC-ad were cultured until sub-confluence on a collagen coated flask in EGM-2MV complete
media, harvested and subsequently cultured in the ptFD. HMVEC-ad remains viable for a period of
several days. Within a few hours following cell seeding EC form a monolayer on collagen gels. In
order to demonstrate the capability to better characterize and study the cellular mechanism during
sprout formation time-lapse movies were made. In traditional sprouting models this capability is
limited since cells are viewed through the monolayer. Here the directional sprouting and migration
occurs in the microscopic viewing plane. As a result we were able to capture the dynamics that
occurs which have not been demonstrated in other systems. Figure 2-3, panels (a)-(i) show the
sequence of morphological changes that occurs. Time-lapse imaging shows a "lead-cell" as it
invades the underlying 3D collagen matrix. In case of single sprout formation (supplementary
movie 1, available online DOI: 10.1039/B802395F); the lead-cell extends filopodial projections into
the underlying matrix while the neighboring ECs on the monolayer remain non-invasive. Cell
invasion follows a period of dynamic projection and retraction of filopodia while maintaining
62
contact with the monolayer and remain highly polarized. An initial root-like structure is formed in
the direction of migration that persists for several minutes with more dynamic smaller extensions.
Subsequent morphological changes included increased penetration depth, filopodial diameter and
translocation of cell from monolayer (evident by movement of nucleus) (Figure 2-3c & d) followed
by conical structure (onset of lumen formation). The invading cells subsequently form sprouts with
open lumen structures (Figure 2-3(h) shows a particle trapped in the newly developing lumen).
With this system, we were able to demonstrate all the sequential cellular mechanisms that occur
during sprouting angiogenesis in vivo. This provides a very powerful tool for future studies with the
capability to answer and clarify some of the controversy in this area of research.
Figure 2-3(j) shows a representative result for long term cultures .ECs maintained in cultured for
several days form multi-cellular capillary-like structures. Endpoint F-actin and DAPI labeling shows
the organization and the complexity of these structures. However, capillaries maintained under
static conditions regress and lose their connection to the monolayer. One of the hallmarks of
capillary formation is development lumen structures. To demonstrate the presence of open lumens,
florescent microspheres were added to the channel on the apical surface of the monolayer. Figure
2-3, panels (k) and (m) provides confirmation of open lumen (microspheres collected in structure).
63
Figure 2-3: Microvascular endothelial cells sprout to form capillary - like structures.
(a)-(i). Time-lapse images demonstrating cellular mechanism during early sprouting process
(movie in supplementary material online DOI: 10.1039/B802395F). Microvascular endothelial cells
were cultured to form monolayer on collagen gel according to Materials and methods. Directional
migration and sprouting occurs in the microscopic view plane. Cells were cultured in VEGF
enriched media (10 ng ml-1) with a gradient in S1P (250 nM). Time indicated in hours and minutes
refers to time after stimulation with VEGF and S1P. Scale bars shown in (a)-(d) represent 20 [1m,
those in (e)-(i) 50 [im. (a) Initial root-like structure forms from membrane protrusion just below
monolayer. (b) Filopodia lengthens to form highly branched structure. (c) Onset of nucleus
translocation into filopodia. (d) Filopodia further thickens, nucleus completely translocated into
filopodia and forms cone-like structure. (e) Formation of lumen-like structure behind polarized cell.
(e) Cell division on monolayer. (f)-(i) Single sprout elongation, lumen-like structure lengths behind
highly polarized cell . Particle trapped in lumen demonstrates it is a open structure. (j) EC
monolayer cultured for several days (here 6 days) with soluble angiogenic stimuli (VEGF 10 ng ml-1
enriched media, with a gradient in S1P (250 nM)) form complex multi-cellular capillary-like
structure. End point sample tagged with fluorescent marker shows actin cytoskeleton (orange) and
nuclei (green). (k)-(l) Another small microvascular sprout, with green fluorescent microspheres
collected in lumen space.
v.
ECs suspended in 3D collagen gels form capillary-like structures
The culture of EC populated collagen gels have been previously studied in macro-scale systems
(152) but not yet in a microscale device. Endpoint results are shown in Figure 2-5. Isolated cells
64
culture in 3D formed multi-cellular chords and EC rings. To demonstrate the effect of biochemical
stimuli, 3D encapsulated ECs were culture in media supplemented with bFGF, VEGF and PMA
(Figure 2-5b). As expected, there was a drastically difference in the morphology compared to
control samples (Figure 2-5a). In control sample, cells migrate and organize to form isolated multicellular ring-like structure. Cells stimulated with pro-angiogenic factors remodel to form complex
interconnected multi-cellular capillary - like structures. In the presence of interstitial flow ECs form
multi-cellular structures within the gel and the monolayer at the gel/liquid interface (Figure 2-5c).
vi.
2D surface migration promotes formation of tube-like structures
Microvascular
ECs
cultured
as
shown
in
Figure
2-2b
(bottom)
experienced
extensive
morphogenesis. ECs on the fibronectin coated channel retain their characteristic cobblestone
phenotype while a remarkable difference in morphology was evident at the gel surface (Figure
2-4a). Prior to sheet or tube formation, the cells migrated as a contiguous structure into the gel
region with a prominent increase in vacuoles and blebs (Figure 2-4b & 4c). These structures were
highly dynamic but eventually evolved into more stable sheets (Figure 2-4d) and tubes (Figure 2-4e
& 4f).
Serial sections of fixed samples obtained from confocal imaging and subsequent 3D
reconstruction of EC networks confirm the presence of circular and flattened lumen-like structures
which extend throughout the length of the vessel. The existence of continuous lumens is further
demonstrated by flowing beads through the vessels under a small pressure drop (supplementary
movie 2, available online DOI: 10.1039/B802395F). Some bead can be observed flowing all the way
across the gel cage and others collect at necked-down regions in the vascular structure.
65
1
m ic r,-, -bi-ad
olow
Figure 2-4: 2D migration promotes sheet-like and tube-like structures.
ECs cultured on 2D substrate next to 3D collagen gel migrated into gel region to form sheet-like and
tube-like structures. (a) EC stimulated with a gradient in S1P (250 nM) and media enriched with
VEGF (10 ng ml- 1). Results show a clear transition from cobble-stone to multi-cellular well defined
individual 2D tube structures with highly polarized tips. Scale bar shown represents 100 im. (b)(e) The morphogenesis of EC can be tracked over several days. Here, a sequence of micrographs
depicting morphological changes during the formation of EC tube-like structures over a 7 day
period. Scale bars represent 100 im. Cells migrate into gel region with a prominent increase in
vacuolar structures that precede tube structure development. (f) Cells were fixed and stained for
actin (yellow) and nuclei (blue). Merged images show complex multi-cellular organization for
tubular and flatten sheet-like structure. Scale bar shown represents 100 lIm. (g)-(i) Confocal
images of tube-like structure in panel (f), boxed region. 3D reconstruction of serial sections
confirms the presence of open lumen structure. (j)-(k) Branched flattened tubular structure with
fluorescent microspheres in lumen. Scale bar shown represents 250 pm.
66
Figure 2-5: Single suspended microvascular endothelial cells form multi-cellular
structures.
Fixed samples of microvascular endothelial cells encapsulated in collagen gels cultured for 4 days.
Scale bar shown represents 125 Rm. Cells were stimulated with angiogenic factors or switch to
interstitial flow after 24 h after cell seeding. (a)-(c) Micrographs of fixed samples stained for actin
cytoskeleton (orange) and nuclei (green). (a)-(b) Images show the effect of biochemical stimuli. (a)
Cell cultured in complete EGM-2MV medium, (b) complete EGM-2MV medium supplemented with
VEGF/bFGF/PMA cocktail (all at 50 ng ml-' final concentration). Results showed a drastic
difference in EC morphology and extent of multi-cellular structures organization. Control samples
form EC rings that are mostly isolated from each other, while EC cultured with potent proangiogenic factors organize to form cellular cords. (c) Interstitial flow (flow direction indicated by
arrow bottom to top), with complete medium EGM-2MV. ECs form multi-cellular structures and
monolayer at gel/medium interface.
e. Discussion and Conclusions
In vitro models offer enormous potential for gaining new insights into essential but complex in vivo
phenomena. Angiogenesis is one example that has been widely investigated because of its profound
physiological and pathological importance. The main challenge has been to provide a model system that accurately mimics key features of the in vivo microenvironment yet provides means for
continuous observation and control. The growth of new blood vessels involves a cascade of events
67
that are a result of molecular, cellular and mechanical interactions in a spatiotemporally controlled
environment. As a result in vitro systems should be able to accommodate these physiologically
relevant features.
We have developed a novel microfluidic (ptFD) system that for the first time is capable of mimicking
many of these complexities on a single in vitro platform. The integration of 3D scaffolds and
microfluidic networks permits the control of the fluidic environment with the benefit of a more in
vivo -like 3D micro-architecture and high-quality imaging capabilities for dynamic studies. With
this platform we are able to control (1) surface shear stress, (2) interstitial flow through the matrix
(3) gradients in non-reactive solutes, (4) properties of cell culture scaffold and (5) simultaneously
monitor cells in real-time.
We have presented a new procedure for introducing 3D matrices into microfabricated systems
using surface treatment followed by gel microinjection. Our gel cage geometry incorporating a
staggered array of micro-pillars is the key enabling feature of this design. Soft hydrogels can be
mechanically supported against differentials in pressure across the gel.
We demonstrate the flexibility of this system in three separate EC capillary morphogenesis assays.
The capillary-like structures that form in these three assays differ drastically with the most
promising and physiologically relevant being the 3D sprouting assay (Figure 2-3). In 3D
encapsulated assay (Figure 2-5); ECs form multi-cellular cords and isolated ring structures, while
ECs that predominately migrate in a 2D manner (against hard substrate into 3D matrix region)
form tube and sheet-like structures (Figure 2-4).
Finally, we have also demonstrated the ability to monitor cells in real-time by recording time-lapse
movies of endothelial morphogenesis. We find that the ability to follow and capture these dynamic
processes in time provides insights not possible through single images taken on the order of hours
or days apart.
68
While the original motivation and design specifications for the 3D-[tFBR were tailored to study the
morphogenesis of ECs in response to biochemical and biophysical stimulation, variations on this
basic platform can be used for a variety of other cell experiments. Some examples include studies of
cell migration through 3D matrices and co- or multi-cell culture systems to create the appropriate
modes of cell-cell communication necessary for the creation of more complex tissues.
f. Acknowledgements
The authors thank Carlos Semino and Alisha L Sieminski for discussions about angiogenesis
experiments, Jeffrey Borenstein and Ernest Kim for insights regarding microfabrication techniques,
Alexandre Micoulet for assistance with time-lapse imaging and Hyungsuk Lee for help with confocal
microscopy. This work was supported by funding from Draper Laboratory (Cambridge, MA).VV is
supported by NIBIB (EB003805 and a research supplement grant to VV).
69
g. Supplementary Material
i.
Optimization of gel cage geometry
The "gel cage" contains an array of micro-pillars which serves two main functions. Firstly, it permits
the delivery and precise localization of small volumes of injectable scaffold material. During
microinjection the liquid scaffold in confined within this region due to surface tension. Secondly,
once the liquid scaffold has polymerized the micro-pillars provide additional mechanical support.
The mechanical stability of hydrogels in the piFD is dependent on the geometry of the "gel cage"
region. In previous designs which consisted of a "regular array" of micro-pillars, gels would fracture
(Supplementary Fig. 1c) as a result of small pressure differentials across the 'gel cage". The
optimized geometry that is presented in this paper, lends mechanical stability and the gels used in
the present study are able to withstand pressure differences up to
ii.
-
200 Pa.
PDMS surface treatment facilitates scaffold microinjection
Supplemental Figure 2 shows a water droplet on (a) untreated PDMS and (b) PDMS that was
treated with air plasma. A reduction in the contact angle (125" to 21") allows for favorable wetting
dynamics which permits scaffold spreading and filling during microinjection of gel solution.
Supplementary Fig.2 c-f shows typical scaffold loading results for PDMS surfaces that have been
rendered hydrophilic. Optimal microinjection results were obtained for contact angles between 200
- 40L (values obtained from hydrophobic recovery data for water droplet on flat PDMS surface
which corresponds to approximately 0.5 to 2 hours following surface modification).
70
iii.
Finite element analysis for diffusion of a non-reactive solute in microfluidic
device
Supplementary Figure 3 shows simulation results for the distribution of a non-reactive solute (in
our gradient experiments, 40kDa dextran) in the "gel cage" region. Dirichlet boundary conditions
were prescribed at the inlets for both sink
(Csink =
0) and source
(Csource =
1) channels, continuity at
the scaffold interface and zero-flux condition at all other surfaces. A diffusion coefficient ratio
(Dgei/Dfluid)
of 0.1 was assumed for finite element simulations. Normalized intensity values along a
line (drawn diagonally to avoid micro-pillars) in the gel region were obtained, the resultant
normalized intensity from two fixed points on this line (indicated by * and ** in supplementary
Figure 3) were recorded for different simulation times. The results obtained (two different gel cage
geometries) were compared to experimental results (obtained in a similar manner) in Figure 2d of
the main text.
iv.
Simultaneous control for fluid flow and gradients
To demonstrate the multi-parameter control capability of our microfluidic platform, gradients
studies where performed with convective flow (hence surface shear stress e.g. on a monolayer) in
the microfluidic channels. Here the fluid flow (fixed flow rate controlled by a syringe pump) is
perpendicular to the diffusion direction across both lateral surfaces of the gel cage, thus constantly
replenishing the sink and source channels to a fixed concentration.
The setup consists of two
reservoirs that are connected at the base by a long tube (3mm ID Tygon tube, Cole Palmer) which
facilitates rapid equilibration of reservoir fluid height thus eliminating any pressure differential
across the gel region. This ensures that the evolution of the concentration gradient in the gel region
is due to diffusion and not convection through the gel. In addition, the equilibration tube is long
71
enough to avoid contamination of the sink reservoir with fluorescent dextran. A similar procedure
to the one presented in the main text was used to analyze fluorescent micrographs. Supplemental
Figure 4 shows the results from these experiments. Panel a, shows results (from a single device) of
the time evolution of the concentration profile up to 6 hours. A steady state profile is reached and
maintained for several hours. To visually demonstrate that there is flow in the channel (while
establishing a gradient), fluorescent particles were added to the source reservoir (at t = 15 hours)
and fluorescent images taken at 1 second intervals. Supplemental Figure 4b shows the localized
increase in fluorescence intensity associated with the streaks made by the moving particles within
the channels. The baseline intensity in the channel remains relatively constant as well as the
sustained gradient (intensities obtained along the dashed line) in the gel. Application of Interstitial
flow through three dimensional scaffolds
Supplementary Figure 5 shows the reservoir set-up for imposing pressure differentials across the
three-dimensional gel scaffold. The ability to isolate the two microfluidic channels is demonstrated
in panel (a), which shows the microfluidic device with two different color fluid streams. Threedimensional collagen gels were formed as described in the main text. To ensure easy visualization
of liquid column levels, green and red food coloring dye were added to PBS. The upstream (with
higher liquid column) reservoir (2.5cm tall Tygon tube, 4mm ID, connected to reducing connector
1/8"X 1/16" ID, Cole Palmer) was colored red and the downstream colored green. A the beginning
of each experiment equal volumes of PBS were added to the upstream and downstream reservoirs,
and the system was allowed to equilibrate for approximately 2 hours before imposing the pressure
differential. Following equilibration, the pressure differential was set (100 Pa or 50 Pa), and
mineral oil (M5904, Sigma) was added to all liquid columns to prevent evaporation (these
experiments were done outside of the incubator). Supplemental Figure 5b-e shows images of the
72
liquid reservoirs at different time points during the course of the experiment. The devices were
oriented such that the downstream and upstream reservoirs were directly in front of each other.
The resulting contrast (due to the difference in colors) made it relatively easy to track the evolution
of the liquid column levels. The dashed lines indicate the levels of upstream and downstream liquid
levels. Figure 2d in the main text shows the result from three separate experiments; two with an
initial pressure differential of 100 Pa and one with 50 Pa.
v.
Time-Lapse Video-Microscopy#
(#movies can be found online at RSC Publishing website under Supplementary Info here:
http://pubs.rsc.org/en/Content/ArticleLanding/2008/LC/b802395f)
Microvascular endothelial cells sprouting movie
Supplementary movie 1 and lb shows the dynamics during sprouting from an intact
endothelial cell monolayer.
Movie 1: Initial invasion starts -6 hours after stimulation with growth factors. Highlights include
filopodial dynamics and nucleus translocation into protruding filopodia. Note that structures are in
various degrees of focus due to their 3-dimensional nature.
Movie 1b: This is a continuation of movie 1. Highlights include observation of cell proliferation on
monolayer and lumen-like structure formation and progression
Vacuole dynamics during 2D migration
Movie 2
73
Perfusion of vascular structures with florescent beads
Movie 3: Shows fluorescent beads flowing through a network of endothelial tube structures. A
suspension of beads was perfused through the channel in which cells were initially seeded. The
endothelial network that developed (after several days in culture) resulted in a direct connection
between the two channels thus allowing perfusion of the beads through the vascular network.
vi.
Supplementary Figures
Device label
Micro-pillars
(pm)
Gel cage
(pm)
Channel height
(pm)
A
10Ox 100
1000 x 1100
120
B
10Ox 100
500 x 1100
120
C
250 x 250
1250 x 1750
240
D*
250 x 250
750 x 1750
240
Table 2-1: Dimensions for microfluidic network.
Microfluidic channels are 500 im wide x 2 cm long. * The gel cage for device D is asymmetric with
short side measuring 500 ptm. The number of micro-pillars in gel cage differs for each device: A-28,
B-16, C-8 and D-5.
74
Figure 2-6: Gel cage geometry.
Micrographs showing gel cage layouts, "regular array" - (a) and (c) and "staggered array" - (b) and
(d) of the VFD "gel cage" (micro-pillar 100 pim x 100 jim). In panels (c) and (d) the gel cage is filled
with a self-assembling hydrogel.
75
-urn
I
I
I
I
L
Figure 2-7: Scaffold delivery into microfluidic device via microinjection procedure.
Water droplet on (a) native PDMS (hydrophobic) and (b) plasma treated PMDS (hydrophilic).
Panels (c)-(f) illustrate microinjection procedure, (f) channels sealed with glass coverslip; sharp
line at the liquid-air interface is due to light diffraction (image taken prior to gelation at 37*C).
76
(a)(b)
Csn
-
09
-
0-7
I-0-3
~~02
0_
02
0
02
04
0A
0A
12
1
250 pnj
_=~
Figure 2-8: Distribution of a non-reactive solute in scaffold cage.
(a) 2D surface map showing results from finite element solution. Concentration of dextran obtained
by assuming diffusion coefficient ratio (Dgei/Diuid) = 0.1 and Dgei = 4 X 10' cm 2 s-. (b) Concentration
distribution in gel region. Dashed line indicates the path (avoiding micro-pillars) from the sink to
source
channel
for which concentrations
are used to compare
to experimental
results.
Concentrations at points * and ** along diagonal line were used to generate Figure 2(d) in main
text.
77
(b)
(a)
1-
140
S
0.810
c
- -
U
U
6----
-.
*
gel
120-
5 min
min
30 min
60 min
90 min
180 min
360 min
100-
*
80
C
w
*60
iN 0.4
h0.2
6U
40
20 -
z
Normalized Distance
Normalized Distance
Figure 2-9: Evolution on the concentration profile across the "gel cage" with a
constant flow rate in microfluidic channels.
(a) Normalized average fluorescent intensity. (b) Transients obtained within a few seconds of each
other. Spikes in intensity indicate streaks made by particles that flow past the gel region. Dashed
line indicates the path along gel and channel for which intensities are recorded. Red and green stars
(*) on micrographs (same device at two different time points) indicate the position of beads and
corresponding spike in intensity; arrow denotes flow direction.
78
Figure 2-10: Evolution of liquid pressure differential across "gel cage" induces
interstitial flow through the three-dimensional scaffold.
(a) Photo of microfluidic device, two microfluidic streams separated by three-dimensional gel
scaffold. (b) Photos of liquid level in reservoir at four different time points during the course of the
experiment (50 hrs). Dashed line indicates the position of the upstream (red) and downstream
(green) reservoir liquid levels. Contrast is due to the orientation the device (reservoirs directly in
front each other). Vertical arrows indicate the pressure differential across the scaffold.
79
Appendix I: Implementation of microfluidic-based 3D cell culture platform
in coculture and imaging studies
The microfluidic based cell culture platform developed in this thesis and subsequent modifications
has formed the basis for additional studies (Figure 2-11).
AU
Epithelial co-culture Model: Hepatocytes and EC
Sudo et al., FASEB J2009
rn
*
Co-culture Model for ECs expressing KLF2 and SMC
Mack et al., Journal of Biological Chemistry 2009
Orn
UU
U
Co-culture Model for ECs with
different cell types
Chung et al., Lab Chip 2009
Conjugated Polymer Nanopartides for
Two-Photon Imaging of ECs
Abdul Rahim et al, Advanced Materials 2009
Figure 2-11: Diverse Applicability of Microfluidic-based 3D cell culture platform
Images were reproduced from (A) Sudo et al., (153), (B) Mack et al., (18), (C) Chung et al., (108),
and Abdul Rahim et al., (154)
Work done with this platform by (A) Sudo et al., demonstrated its applicability for generating 3D
tissue-like structures from primary hepatocytes (isolated from rat liver). The functionality of the
80
engineered tissue was demonstrated by visualizing metabolites (fluorescein diacetate - FD) that
was secreted into bile canaliculi structures and ethoexyresorufin O-dealkylase (EROD) activity. (B)
Mack et al., investigated the effect of endothelial KFL2 expression on smooth muscle cell (SMC)
migration. Under coculture conditions, the expression of transcription factor KFL2 was found to
significantly reduce SMC migration. In the same report, KFL2 was unregulated in endothelial cells
subjected to a shear stress profile that is characteristic of an adaptive remodeling coronary
collateral waveform. (C) Chung et al., further demonstrated the applicability for coculture studies
with relevance to cancer cell (MTLn3 - adenocarcinoma cells and U87MG - glioblastoma cells) or
smooth muscle cell (10 T
1/2
- smooth muscle cell precursor) endothelial paracrine interactions.
Under culture conditions, MTLn3 cancer cell was found to promote endothelial migration while
U87MG had minimal effect whereas, 10 T
cells suppressed endothelial migration. (D) Abdul
Rahim et al., demonstrate the applicability for evaluating a new fluorescent probe for two-photon
imaging application. Two-photon imaging has been identified as a promising tool for in vivo
imaging which holds the potential for high sensitivity, greater tissue penetration depth, reduced
photodamage to tissue and lower instrument associated costs (154).
Additional studies (155), (109), not directly related to this thesis has also used similar platforms
and methods developed in this thesis demonstrating broad applicability for different biological
studies.
81
Appendix III: Biochemical Regulation of Angiogenesis: Vascularization of
Scaffolds in a Microfluidic Cell Culture Platform
vii.
Biomaterial Scaffold
Background In formation
A critical element in most tissue engineering approaches is the biomaterial scaffold which would
serve to mimic many roles of the extracellular matrixes (ECM) found in tissues (156), serving as a
structural support while providing biochemical cues to regulate cell function (157). The physical
and chemical properties of the scaffold are critical, for example for anchorage-dependent cells
adhesion is vital for cell survival and in endothelial cells integrin-mediated adhesion may trigger
different signaling pathways depending on the attachment motif (158), (159) available and the
ability to generate traction forces (158), (160), (152), (148).
Scaffolds for in vitro angiogenesis studies and potential vascular-related tissue engineering with
angiogenic potential (ability support vascular ingrowth) with physical (elastic modulus, pore size,
fiber dimensions) properties and chemical composition that can be controlled by design are highly
desirable. Naturally-derived biomaterials (e.g. MatrigelT M , collagen and fibrin gels) have been
scaffold of choice for most in vitro angiogenesis models. However animal-derived biomaterials may
not be suitable for tissue engineering applications such as the development of highly vascularized
scaffolds; since they could potentially carry and transmit harmful pathogens (161) as well as
associated batch-to-batch variability.
Synthetic biomaterials may provide an alternative to
naturally derived matrices in light of their associated risks. Moreover, synthetic scaffolds are
flexible for design and optimization.
82
Biomaterials composed of self-assembling oligopeptide has been identified (162), (161) as
potential candidate for tissue engineering applications. These oligopeptide consist of alternating
hydrophilic and hydrophobic amino acid residues. The first member of this family, EAK16
(AEAEAKAKEAAEAKAK),
was originally discovered in a region of alternating hydrophobic and
hydrophilic residue in a yeast protein known as zuotin (163), (164). Since their discovery, a
number of self-assembly peptide systems (e.g. RAD16-I, RAD16-II, PRG, KLT, VEVK9, VEVK12) have
been designed and developed (163), (165), (166). These oligopeptides are highly soluble in pure
water and have the tendency to form stable P-sheet structures (163). Upon the addition of
monovalent cations of physiological salt solution these oligopeptides spontaneously assemble to
form matrices consisting of interwoven nanofibers measuring 10-20nm and pore size of 50-200nm
in diameter (162).
These peptide gels have been shown to support neurite outgrowth and formation of
functional synapses (167), chondrocytes development and ECM synthesis (168) and hepatocyte
differentiation (169). Another important characteristic of these peptide biomaterials is that their
mechanical (stiffness) and biochemical properties can be controlled by the manipulation of the
peptide concentration and adhesion motifs. Studies have also been done with functionalized
peptide scaffolds (170). In particular the RAD16-I class of oligopeptides has been modified with
short functional amino acid sequences PRG (2-unit RGD) and KLT (mimicking VEGF helix region)
which have been shown to endothelial cell survival, proliferation, migration and morphological
differentiation in 3D encapsulation and gel sandwich assays (166) (these results were published at
the same time as results from our microfluidic studies).
83
Methods
Gel preparation and monolayer formation on RAD16-I and RAD16-I/PRG/KLT self-assembling
peptide gels
Two different types of synthetic peptide gels were used in these experiments, RAD16-I (unmodified
form; gift from PuraMatrix 3DM Inc.) and mixture of RAD16-1 and modified peptides KLT, Ac(RADA) 4G4KLTWQELYQLKYKGI-CONH
2
and PRG, Ac-(RADA) 4GPRGDSGYRGDS-CONH 2 (gift from
Zhang lab). RAD16-I gel-forming mixture was prepared by first combining 1% RAD16-I stock
solution with 20% dextran solution (tissue culture grade, D8802; Sigma-Aldrich) at a 1:1 volume
ratio. To initiate gelation, equal volumes of 1X PBS (containing Ca2+and Mg2 +)and RAD16-I/dextran
solution were mixed together to obtain a final concentration of 0.25% RAD16-I. Blended gelforming solution (RAD16-I/PRG/KLT) was prepared by combining 0.5%RAD16-I mixture (i.e. 1%
RAD16-I plus dextran solution) with equal volume PRG/KLT/PBS solution (i.e. PRG/KLT plus 2X
PBS at 1:1 volume ratio).
Gel-forming mixtures of 0.25% RAD16-I and RAD16-l/PRG/KLT were loaded into PDMS
microfluidic devices as previously described and incubated at 37C and 5% C02 for at least 8 hours.
Devices were kept in a secondary high humidity container - "humidity box" as before to prevent
gels from drying out. This incubation period is much longer compared to collagen gels (30-35
minutes), however this is necessary for forming uniform gels since higher concentrations of PBS
was found to interface with the gelation process. Following polymerization, peptide gels were
incubated with cell culture medium (EGM-2MV) overnight.
HMVEC monolayers were generated in the device as previously described. Briefly, HMVEC cell
suspension (-4 x10 6 cells/ml) was prepared and added to one of the microfluidic channels, devices
84
inverted to allow cells to settle onto or near the gel surface. Once monolayers were formed, cell
culture medium was replaced with experimental medium (details included with results).
During the course of the experiment, phase contrast/bright field images (on a Nikon Eclipse Ti-U;
with DS-2MV camera using NIS Elements D image acquisition software (Nikon Instruments Inc., NY
USA)) were taken to monitor capillary morphogenesis.
Results and Discussion
Results are presented in the form of represented micrographs for "optimal" culture conditions
obtained with regards to gel concentration, gel composition and experimental media. Initial
preliminary experiments were done by suspending HUVEC in RAD16-I gels (using a different
gelation protocol from the one described above)-3D encapsulation assay (Figure 2-12). The 3D
encapsulation method was previously used in our lab by Sieminski et al., (152) and more recently
by Wang et al., to evaluate the angiogenic potential of "designer self-assembling peptide nanofiber
scaffolds" (166). In these studies "macro" size gels were used which permitted a wide range of
concentration; however lowering concentration gels which correlate to significantly lower stiffness
(e.g. for RAD16-I, 0.1% gel measured 46 Pa; and 0.287% gel measured 400 Pa) was found to
support extensive capillary-like network formation (152). However using the "old" gelation
protocol (shorter time for peptide polymerization before incubation with cell culture media) which
permitted 3D encapsulation assay did not allow for gel concentration within that range. As a result
a new protocol (one described above) was developed for forming peptide gels in the microfluidic
device. This protocol called for much longer incubation time (t ~ 8 hour, cells suspended in gels as
with 3D assay would not remain viable) as a result subsequent experiments were done with
monolayer sprouting assay. Furthermore, capillary sprouts formed from an intact monolayer are
85
readily perfusable (41) and has a similar polarity to in vivo and might be a superior strategy for
engineering patent, perfusable pre-formed vascular networks.
1
lOOpm
Figure 2-12: 3D encapsulation assay
HUVEC suspended in RAD16-I self-assembling peptide (A) and rat tail collagen type I (B) gels. Scale
bar 100 [tm.
Monolayer Sprouting Assay
The functionalized peptides found to have similar structural properties to RAD16-I sequence
however with a notable increase in HUVEC attachment, viability, proliferation and migration (166).
Consistent with this report we found that HMVEC attachment to RAD16-I/PRG/KLT was found to
be consistently better than unmodified RAD16-I gels. More often, clusters of rounded cells or gaps
would be found on monolayers formed RAD16-I (Figure 2-13 (A) inset).
86
A
RAD16-I
B
RAD16-1 plus PRG & KLT
Figure 2-13: HMVEC Sprouting Assay - RAD16-1 and RAD16-I plus PRG & KLT
Micrographs showing HMVEC monolayer 0.25 % RAD16-I (A) and 0.25% RAD16-l/PRG/KLT
peptide gel 24 hrs after stimulation with VEGF gradient. Monolayers were formed as described in
methods. Promote migration and capillary morphogenesis, EGM-2M medium was supplemented
with 50 ng/ml bFGF and 250 nM S1P and added to apical channel. To establish a gradient in VEGF,
VEGF was added to EGM-2MV containing both bFGF and SiP (both at the indicated concentration)
to a final concentration of 40 ng/ml VEGF. This cocktail was then added to the basal channel
opposite the monolayer. Scale bar 100 um
Furthermore, monolayers cultured in EGM-2MV supplemented with 50 ng/ml bFGF and 250 nM
SiP on RAD16-I/PRG/KLT gels form extensive sprout structures measuring over 100 prm (Figure
2-13 (B)) in response to VEGF gradient compared to RAD16-I gels. This is consistent since
attachment and cell survival are clear prerequisites for sprout formation. Occasionally, in regions
where cell cultures are observed on 0.25% RAD16-I gels shallow membrane projections are
observed (Figure 2-13 (A) inset). It is foreseeable that the mechanism for sprout elongation from a
monolayer and capillary morphogenesis in singly suspended cells in a 3D gel might be different.
87
The most obvious mechanistic difference involves the active penetration of their underlying matrix
to emerge from the monolayer (see additional discussion below).
RAD16-I plus PRG & KLT
E
72hrs.
Figure 2-14: Capillary morphogenesis on RAD16-I/PRG/ KLT self-assembling peptide
gels
Micrographs of HMVECs sprouts emerging from monolayers generated on RAD16-1/PRG/KLT selfassembling peptide gel at 24 (A), 42 (B) and 72 hrs. (C). Methods and experimental medium
composition as before. Scale bar 100 pim.
88
One important criterion for successfully generating vascular network is the growth, survival and
eventual maturation/stabilization of the network. Examination of sprouts generated on RAD16I/PRG/KLT gels over the course of several days revealed that sprouts narrow (Figure 2-14(B)) and
are subsequently pruned leaving a portion attached (Figure 2-14(C)) to the monolayer followed by
single cell migration through the 3D matrix (Figure 2-14(C) and (A-B)).
1
I
IN
Figure 2-15: Single cell migration through RAD16-I/PRG/KLT functionalized selfassembling peptide gels
89
Micrographs of a single HMVEC cell migrating in 3D through RAD16-I/PRG/KLT (A) showing higher
magnification (B-C) of regions indicated by arrows in (A), showing cell "tracks". (A) Boxed inset
shows another cell "stopped in its tracks". Scale bar 100 prm.
The mechanism but which cells from monolayer penetrate the underlying peptide gel, elongates to
form longer sprouts or migrate as single cell through the 3D matrix in unclear. In samples with
singly migrating cells, migration "tracks" are clearly visible in phase contrast images (Figure
2-15(A)) even more so at higher magnification with increased contrast (Figure 2-15(B)). The
peptide sequence for RAD16-I, PRG or KLT was not designed with protease cleavable sites that are
present in later designer peptides (165). These newer scaffolds with MMP-2 cleavable motif were
found to exhibit favorable conditions thus permitting accelerated 3D migration which was
independent of scaffold stiffness (165). In our experiments, it is likely that cells are able to tunnel
their way through and push the matrix fibers away to create a passage for migration. However
further investigation in necessary to answer this sufficiently question, possibly visualizing matrix
using reflectance microscopy or ultrastructure obtained from TEM to visualize matrix fiber
orientation in the vicinity of sprouts or migrating cells. It would also be interesting to compare the
sprouting response for RAD16-I/PRG/KLT and MMP-2 cleavable peptides. We anticipate that these
peptides would support more extensive EC invasion and sprout elongation.
Concluding Remarks
Preliminary work with peptide (RAD16I and RAD16II) for the endothelial sprouting assay soon
demonstrated that the angiogenic response, in particular, the degree of invasion and sprout
formation, was very low compared to the widely used rat tail type I collagen gels. Recently we
revisited these sprouting experiments with self-assembling peptide gels as the scaffold, and
90
developed a new protocol which permitted the loading of lower concentration gels (almost 10
times lower concentration with correspondingly lower stiffness (152)). With these softer gels EC
invade to form capillary-like structures, a significant improvement.
RAD16-I/PRG/KLT
self-assembling
peptide
gels permitted
HMVEC
to invade from
intact
monolayers to form capillary-like structures. This response was an improvement to cultures with
RAD16-I under same culture conditions.
Acknowledgement
The work presented in this section (Appendix III) was done in collaboration with MIT Summer
Research Program (MSRP) student, Pablo Rosado who worked under my supervision during the
summer of 2008.
91
Appendix II: Microfluidic Device Design Gallery
Figure 2-16: Example of microfluidic design and mask layout
(A) Layout of microfluidic device for creating transparency mask and silicon wafer master. (B)-(C)
Higher magnification of individual devices with narrow (B, Top) and wide (B, bottom) "gel-cage",
showing a staggered rectangular array. In the final fabricated product (PDMS device) the resulting
"gel-cage" consists of an array of post which confines a liquid pre-polymer mixture due to surface
tension effects and provides mechanical stability for the polymerized gel against pressure
differential across the gel region.
92
Chapter 3: A Model for Aqueous Humor Outflow across the Inner
Wall of Schlemm's Canal: Direct Imaging of Giant Vacuole
Dynamics using an in vitro Microfluidic-based Platform
a. Introduction
Primary open angle glaucoma (POAG) is thought to occur as a result of an obstruction to aqueous
humor (AH) outflow somewhere in the vicinity of the endothelium lining the inner wall of
Schlemm's canal (SC). The mechanism of passage has been a source of considerable speculation and
direct observations with in vivo models remain a challenge. Intercellular and giant vacuole (GV)
pores have been implicated as the primary pathway for AH outflow. Nevertheless, the flow
mediated mechanisms that regulate giant-vacuole formation and subsequent aqueous humor
outflow dynamics across the inner wall remain largely uncharacterized. Previous efforts to
elucidate potential mechanisms have been impaired by the inability to directly visualize the flow of
AH across this narrow layer. In the present study we developed a novel 3D in vitro model of
aqueous humor dynamics and examined the formation of GVs in response to basal-to-apical flow
across the endothelial monolayer. An accurate understanding of the underlying mechanisms that
modulate outflow resistance at the inner wall would allow for the development of strategies to
modulate intraocular pressure and potentially treat POAG. We have developed a novel microfluidicbased physiologically relevant 3D cell culture system which can be used as a research tool of a
platform for drug screening with application to but not limited to glaucoma therapy.
93
b. Background
i.
Aqueous humor circulatory pathway
Aqueous humor (AH) is a clear fluid and is the blood analog of the eye that provides nutrients and
removes metabolic waste among other functions to maintain homeostasis of ocular tissues (171).
AH is produced by non-pigmented epithelial cells of the ciliary processes from which it flows freely
over the lens, through the pupil and into the anterior chamber. From here AH leaves the eye via two
routes (a) The conventional outflow pathway via the trabecular meshwork and (b) The nonconventional/uveoscleral outflow pathway via the uveal meshwork (see (172)for a recent review).
The conventional outflow pathway (Figure 3-1A) accounts for the bulk of AH drainage. In this
route, AH exits the anterior chamber, flows into the trabecular meshwork across the endothelium
of the inner wall and into Schlemm's canal (Figure 3-1B). Once in the canal AH enters the systemic
venous circulation via connector channels that connect to the aqueous veins. From the aqueous
veins AH flows into the episcleral veins located near the surface of the eye.
AI
Caalef
W all
InnerJC
Trabecular M eshwork
T
Images copied from: (A) http://sierraeyeassociates.com/glaucoma.htm
and (B) Ramos et al., J Glaucoma (2007) 16. 391-405
94
Figure 3-1: Conventional flow pathway of aqueous humor
Diagram showing drainage of AH via the Conventional Outflow Pathway (A). AH produced by
ciliary processes in the posterior chamber flows in to the anterior chamber and exits through the
trabecular meshwork and Schlemm's canal. Cartoon showing the structure of the Trabecular
meshwork, the route taken by AH as it exits the anterior chamber (B).
ii.
Aqueous Humor Outflow and Glaucoma
Under normal conditions the rate of AH production (-
2.0-2.75 p1/min) is equal to the rate of
drainage which ensures normal IOP levels. A malfunction in production or drainage mechanism
shifts the normal equilibrium (10-21 mmHg) in IOP to hypertensive levels (> 21 mmHg) a known
critical risk factor for glaucoma - a group of eye disease afflicting over 70 million people worldwide
of which over 4 million are Americans (171). This elevation in pressures, which is characteristic of
Glaucoma, ultimately leads to optic nerve damage and blindness. Furthermore, glaucoma is the
second leading cause of blindness. Since there is no cure, there are only disease management
strategies which include medication and/or surgery. Experimental studies have shown that the
probability of having glaucoma increases with increased IOP levels (173). Current therapies are
focused on lowering IOP. Consequently, the mechanisms regulating AH drainage is highly relevant.
Glaucoma is an umbrella term for the many manifestations of the disease which include primary
open-angle (POAG), angle-closure, normal-tension, congenital glaucoma and other types which are
variants of open-angle and angle-closure glaucoma [Glaucoma Research Foundation]. Figure 3-2
shows AH flow pattern in the two main types. The most common form, POAG, is caused by a
malfunction in AH drainage.
95
In POAG cases, there is a notable increase in the trabecular outflow resistance. This increase is
resistance is attributed to cellular (e.g. increase expression of myocilin/TIGR and aB crystalline)
and the associated morphological (e.g. trabecular beam thickening, increased sheath derived
plaques and narrowing of intertrabecular channels) changes within the region (174), (175).
Furthermore, POAG patients have been shown to have a reduction in the number of pore structures,
important features in AH transport across the inner wall endothelium (176).
rainage
Drainage Canals
uid Row
Iris
/
Cornea
Crnna"
-
\
\
'
Drainage Canals
\
images: http://www.glaucoma.org/uploads/poag_illus.jpg
Figure 3-2: Two main types of glaucoma, open-angle and angle-closure glaucoma
Diagram showing AH fluid flow pattern in the two main types of glaucoma (A) open angle and (B)
closed angle. Nomenclature "open" or "closed" is based on the configuration of the region between
the iris and the cornea known as "Angle".
iii.
The Inner Wall of Schlemm's Canal
96
It is now widely accepted that major outflow resistance sites to AH include juxtacanalicular tissue
(JCT) and the endothelium of the inner wall of Schlemm's Canal (177). However there is an ongoing
debate over the individual contribution of the inner wall endothelium in regulating outflow facility
(see (178) for an excellent review). Nonetheless, there is a growing body of evidence that suggest
that the inner wall influences outflow facility (178). The continuous endothelium lies on a
discontinuous basal lamina. Figure 3-3 A is typical scanning electron micrograph of the region and
shows the connectivity between the IW and the JCT. Passage across the IW is believed to occur via
outflow structure includes giant vacuoles and intercellular pores (Figure 3-3B, C). The regulation
of such structures would consequently impact the hydraulic conductivity, outflow facility and
ultimately IOP.
U
Images: Ethier C.R. Exp. Eye Res. (2002)74. 161-172
97
Figure 3-3: Inner Wall of Schlemm's Canal
Images reproduced from published literature (178). (A) Cross sectional view of Schlemm's canal
showing inner wall (B) En face view of Inner Wall (C) Cross sectional view of Inner wall showing
Giant vacuole structure (GV)
iv.
Juxtacanalicular Tissue (JCT)
The JCT (Figure 3-4) forms a loose connective tissue comprised of fibroblast-like cells surrounded
by extracellular matrix (various collagens isoforms, laminin, fibronectin, elastin), matricellular
proteins (thrombospondin-1 and SPARC)
,
growth factors (TGF-j2, CTGF, BMP7, BMP4) and
proteoglycans (versican, syndecans)(excellent reviews by Tamm 2009, (179); Acott and Kelley
2008, (180)).
98
Schiemm's Canal
Sciera
Figure 3-4: juxtacanalicular Tissue
Images and text reproduced from Acott and Kelley 2008 (180). Diagram of the outflow pathway
and juxtacanalicular region. The lower portion shows a stylized view of the TM and the upper inset
shows as expanded view of the JCT region.
99
v.
Ex vivo aqueous humor outflow models
Organ Perfusion
Two main organ perfusion AH outflow models include the intact enucleated eye and anterior
segments. These have been widely used by many laboratories because they more closely mimic the
in vivo condition thus permitting easier extrapolation of experimental results.
Intact Enucleated Eye. Perfusion of whole eyes is a well-established method dating back to
the 1950's (181), (182). Donor eyes from different species (e.g. human, bovine, porcine, rabbit) are
typically enucleated within 24 hours of death.
To preserve freshness and ensure integrity of
tissues, eyes are stored at 4'C in a moist chamber and used for AH perfusion experiments shortly
afterwards (within a few days). Perfusion fluid typically used includes cell culture medium (183),
Barany's solution (balanced salt solution containing glucose) (184). Eyes are perfused via the
anterior or posterior chamber using small gauge needles or corneal fitting which is connected to
external plumbing and pressure controls for quantitative evaluation of outflow facility (C). The
value C is obtained from fluid flow rate (F) and steady state pressure (P) and is equal to the ratio
F/P. For morphological studies, at the end of the experiment, tissues are dissected, processed and
examined using confocal, light or electron microscopy.
This model has been widely used in a
variety of glaucoma motivated studies, however is not conducive to long-term studies which might
be the case in some pharmacological studies where drug effect occurs over several hours.
Anterior segment. The perfusion-cultured anterior segment model provides an alternative
model for longer term eye perfusion studies. Figure 3-5 shows perfusion setup and flow schemes
developed by Johnson and Tschumper (185). In their initial report, trabecular tissues remained
100
viable during cultures for up to 4 weeks. In this method, freshly enucleated eyes are bisected at the
equator and the lens, iris and vitreous humor removed. The anterior segment is subsequently
mounted to custom made perfusion devices and sealed to form a closed anterior segment. Cultures
are maintained in a humidified culture chamber at 37 C and 5% C0 2 with constant perfusion rate to
match physiological levels (through the trabecular
2.5 pl/min). Intraocular pressure is established and flows routed
meshwork and into
Schlemm's canal.
Several
investigators
have
subsequently used this model (186), (187). This setup has been valuable for screening drug
candidates and to evaluate potential therapy that could target the trabecular meshwork to
modulate outflow function. For example, TGF-beta2 treatment was shown to reduce outflow
facility, by promoting accumulation of extracellular material and the length of inner wall available
for flow (186). Interestingly, TGF-beta2 concentrations in AH of patients with PAOG is significantly
increased (188). In another study, with a combination of Qdot labeling and RNAi silencing, the
proteoglycan versican has been implicated as an important component which impacts outflow
resistance (187). Despite its value, the anterior segment perfusion model has some limitations
when tissue viability in considered, consequently the need for 'real time' indicator of tissue
segment viability such as central corneal thickness has been recommended (189). Furthermore,
this method does not permit in situ dynamic monitoring of cellular responses to chemical or
mechanical perturbation.
101
U
Images: (AB) Johnson D.H. and Tschumper R.C., Invest Ophthalmol.Vis. Scl. (1987) 28 945-953
and (C)http://www.mavoclinic.org/images/glaucoma-research-human-conventional-outflow-2col.g
Figure 3-5: Ex vivo anterior segment perfusion model
(A) Ex vivo perfusion device. (B) Device with enucleated eye. (C) (left) Schematic of perfusion setup
and flow pattern, (right) pressure monitor.
In vitro cell culture perfusion models - A potential to narrow current in vitro/ in vivo gap
SCEC culture models provide an additional degree of flexibility by permitting studies of the inner
wall endothelium in "isolation" from other tissues in the trabecular outflow pathway. In the late
1980's Perkins and colleagues reported a perfusion culture system, (190) (Figure 3-6A) which was
subsequently used for simulating flow across the inner wall endothelium (191). In this study,
isolated SCE were cultured on filter membrane which was housed in a custom designed flow
102
chamber. This setup permitted precise flow rate and pressure controls but was not compatible with
real-time imaging. Recently, over two decades later, an improvement on this original concept was
reported by Pedrigi et al (192)(Figure 3-6B). This latest design is compatible with real-time
imaging; however continuous viewing of the monolayer is limited by the need to "refocus" and
requires confocal microscopy. Despite these advance, the current models are limited by their ability
for incorporation of physiologically relevant 3D matrix environment.
The trabecular outflow pathway which includes the trabecular meshwork (uveal and corneoscleral
meshwork), JCT and IW endothelium is a complex environment (excellent reviews (180), (179)).
One of the strengths of in vitro models is that this inherent in vivo complexity can be simplified.
Within this context, the studies of the inner wall in "isolation" are prime examples. Nevertheless,
the ability to add layers of complexity is highly desirable in a sense "rebuilding" the in vivo
environment which would allow synergistic interactions to be captured. For example, the three
dimensional region that lies adjacent to the inner wall endothelium, the
JCT, is in intimate contact
and a local source of communication and could potentially impact its function relating to AH
outflow. Cells within the JCT region are known to extend processed which form gap and adherens
junction with endothelial cells of the inner wall (193). Likewise, endothelial cells form projections
that extend in the
JCT. Grierson concluded that "cells in the trabecular wall form a synchronized
system" (193). The paradigm of synergistic interactions between cells, flow and ECM is gaining
popularity, at least in current discussion (177), (180), (178), (194).
103
- - -
- -
- -
- -
- - -
- -
- -
- -
t
Perkins et al., Invest. Ophthalmol. Vis. Sc. (1988) 29. 1836-1846.
------------------
Computer
Direction ofj
Perfusion
Filter Membrane
Cover Glass
Cel71s
Membrane Insert
Adapter
Objective -
Objective
Pedrigi et al., Exp Eye Res. (2011) 92.57-66.
Figure 3-6: In vitro SCE monolayer perfusion model
Images reproduced from (190) and (192). (A) Filter-based perfusion systems without continuous in
situ imaging and (B) live cell imaging capability. Monolayers are cultured and perfused on a 2D
filter surface.
There is sufficient evidence which would support the need for in vitro models that bring together
cells, matrix and flow in a single experimental platform. TM cells are known to response to
mechanical stretching, laser irradiation and pro-inflammatory cytokines by secreting enzymes and
cytokines that modulates both cells and ECM function within the conventional AH outflow pathway
(195). For example, studies show that TM cells actively respond to mechanical stimulation caused
104
by fluid flow or stretch to alter gene expression and protein production which impact vascular
permeability, cytoskeleton reorganization and extracellular matrix remodeling (196), (197) to
name a few. Furthermore, SCECs alter gene expression and permeability when cultured in
conditioned media obtained from to laser irradiated TM cells (198). On the endothelial side, Ethier
et al., reports that Schlemm's canal endothelial cells in vivo are subjected to significant shear stress
levels which impacts cell cytoskeletal remodeling (199). Additionally, SCECs produce factors that
affect TM cell function (198). Despite the limitations of current cell-based models for shear stress,
3D matrix and potential paracrine or juxtacrine interaction on in co-culture studies current models
have proven valuable for elucidating cellular and molecular mechanisms within the conventional
outflow tissue.
105
c. Materials and Methods *
(* more detailed protocols are included in the appendix section of this chapter)
i.
Cell Handling, Maintenance and Culture
Primary Schlemm's canal endothelial cells and cell culture maintenance protocols were kindly
provided by Dan Stamer (University of Arizona).
Cell Isolation. hSCEC isolation protocol from cadaveric eyes was developed by Stamer et al. (200).
Briefly, hSCECs were isolated by cannulating Schlemm's canal with a gelatin-coated suture (sterile
nylon monofilament) and tissue segments containing sutures cultured for 3 weeks. The gelatincoated suture served as a substrate that permitted the infiltration of SCEC from the canal wall. Cell
laden sutures were removed from the canal and cells expanded in tissue culture plates (Stamer et
al., 1998, (200) provides additional details).
Cell Culture. For experiments, frozen cell stock was thawed and seeded in T12
or T25 tissue
culture flask. Cells were cultured on collagen coated (50 ig/ml) tissue culture flasks in Dulbecco's
Modified Eagle Medium (DMEM, low glucose; 11885-084, Invitrogen, Chicago IL) supplemented
with
10 % fetal bovine serum (FBS certified, 16000044;
streptomycin-glutamine
(10378-016;
Invitrogen)
and 1%
penicillin-
Invitrogen). Cultures were maintained in a humidified
environment at 5% CO2 and 37'C.
ii.
3D Matrix
Since there were no known prior published studies of hSCEC cultured on hydrogels, experiments
were conducted to evaluate hSCEC adhesion to 3D hydrogels. Furthermore cell attachment to hard
substrates coated with ECM protein does not guarantee attachment and spreading on soft
106
compliable gels of a similar composition. For these studies rat tail collagen and Matrigel'" were
selected as good candidates because of prior experience working with these gels which have
supported the attachment of other vascular ECs. Liquid rat tail collagen type I (354236 or 354249 the high concentration product, BD Biosciences, Chicago IL) and BD Matrigel'" (354230; basement
membrane matrix, growth factor reduced, BD Biosciences) pre-polymer solution was prepared
according to product specifications.
Gels formed from pure rat tail collagen type I, Matrigel or
collagen/Matrigel blends were used. Blended gel-forming mixtures were prepared by combining
collagen pre-polymer solution with freshly thawed Matrigel"
to make different ratios by volume
gels. Two criteria were used in evaluating the best options for a 3D matrix namely (1) ability to
support cell attachment and monolayer formation and (2) permit microbeads passage (see bead
tracer experiments below) with and without an endothelial monolayer on one gel surface. For cell
adhesion experiments, hydrogels were form (100-120 p1/well, 30 - 35 minutes incubation at 370C)
in glass bottom petri dishes (P35G-0-7-C; MATTEK CORP, Ashland MA). Gels were equilibrated with
cell culture medium, hSCEC cell suspension and incubated at 370C and 5% C02. To observe cell
dynamics during attachment, time-lapse images were recorded (Zeiss Axiovert 200, Carl Zeiss,
Germany) otherwise cultures were maintained in humidified incubator for at least 6 hours and
imaged (Nikon Eclipse Ti-U).
iii.
hSCEC monolayer formation and perfusion
For perfusion studies, hSCEC monolayer was generated on 3D hydrogel (selected for cell
attachment studies) in a microfluidic-based cell culture device. The capability of controlling the
biochemical environment and flow was previously demonstrated as described in detail in (41).
Briefly, the design includes two independent microfluidic channels that are separated by a central
region - "gel cage" for housing injectable hydrogels (e.g. synthetic peptide, MatrigelT
",
collagen)
107
allowing for three-dimensional (3D) matrices. This geometry permits the establishment of pressure
gradients across the scaffold and small interstitial fluid flows through the 3D matrix. Human
Schlemm's canal endothelial cell (hSCEC) monolayers were generated on 3D collagen/Matrigel
matrix by perfusing single cell suspension through one channel, allowing the cells to settle along the
collagen surface followed by 2-3 days of static culture.
Perfusion Studies (additional details available in appendix). Two main types of
perfusion studies were conducted to demonstrate proof-of-concept for (1) monitoring giant vacuole
dynamics and (2) visualizing fluid hydrodynamic filtration patterns. Flows across the endothelial
monolayer were generated at constant flow rate (0.3 pl/min) or known pressure (liquid column
height limited by current choice of portable on-chip reservoirs -450 Pa) by using a programmable
push/pull syringe pump (PHD 2000, Harvard Apparatus) or small fluid reservoirs respectively
unless otherwise noted. At constant flow rate the equivalent rate per monolayer surface area was
50 pl/min/cm 2 which was selected to be within reasonable range compared to physiologic rates
(190). For known pressure differential setup, the relative heights of the liquid columns in
reservoirs connected to basal and apical channels were defined. In both setups, flow was generated
in the basal-to-apical direction with respect to the monolayer which is similar to the in vivo
situation where AH filters through the trabecular meshwork and flows across the endothelium of
the inner wall.
iv.
Visualization of Giant Vacuole Dynamics
Visual demonstration of GV formation and dynamics in live cells is critical for understanding
mechanisms of formation. Multiple imaging modalities were evaluated for visualizing B-A flow
induced GV-like structure formation including phase contrast and fluorescent microscopy. For live
fluorescent imaging, hSCEC monolayers were labeled with live cell tracking dyes (Dil; D-282,
108
Molecular Probe, Eugene OR) prior to perfusion. Other live cell tracking dyes were also evaluated
(see appendix). For long-term studies (several hours) monolayers were monitored on a microscope
housed in an environmental chamber with CO2 levels and temperature controls. For phase/bright
field monolayers were visualized on a Nikon Eclipse Ti-U (Nikon Instruments Inc., NY USA) with a
DS-2MV (Nikon Instruments Inc., NY USA) camera using NIS Elements D image acquisition
software. For higher resolution images a confocal microscope was used (Olympus FluoView
1000).
Morphometric Analysis of Giant Vacuoles. Time-lapse fluorescent micrographs were
analyzed using ImageJ macro to measure changes in GV cross-sectional area. To determine GV
dimensions a built-in area selection tool was used to trace GVs ("dome" shaped structure on
monolayer). Data from each image frame was recorded and reported as a function of time to
demonstrate the dynamics of GVs.
v.
Evaluation of Fluid Hydrodynamic Filtration Pattern
Epifluorescent (Nikon TE300 equipped with a Hamamatsu, ORCA-ER camera and OpenLab imaging
software) and confocal (Olympus FluoView 1000) microscopy were evaluated for used in fluid
filtration pattern studies and to track microbeads passage in relation to GV-like structures.
Hydrodynamic filtration pattern would provide evidence for where fluid is being "funneled"
through the monolayer. A bead tracer method similar to that used by Gong and colleagues (201)
was used to label fluid filtration pattern in the collagen matrix upstream the cultured SCEC
monolayers. With this method however, bead trajectories can also be tracked in situ which has not
been done previously. hSCEC monolayer were formed on collagen or Matrigel/collagen blended
gels as previously described. For these studies fluorescently tagged microbeads (200nm) were
109
used. Cell culture medium (unless otherwise stated) containing microbeads was used in reservoirs
connected to fluidic ports leading to the basal channel (channel upstream of the monolayer). A
pressure differential and B-A flow was initiated as previously described. Time-lapse images were
recorded and fluorescent micrographs were analyzed in ImageJ (public-domain image processing
software, U.S. National Institute of Health, Bethesda, MD) using built-in particle tracking macro.
vi.
Effect of pharmacological inhibitor Y27632
Current glaucoma treatment strategies involve lowering IOP levels. Medications to lower the
pressure in the eye are typically administered topically in the form of eye drops, orally or
intravenously. The common theme is to reduce the production or increase the outflow facility of the
flow of AH. Consequently, the demonstrated capability for evaluating the effect of drugs on the
function cultured hSCEC within a 3D culture environment would be of clinical relevance. For
example, in vivo and in vitro studies with pharmacological inhibitor Y27632 was found to increase
outflow facility in bovine, rabbit, monkey porcine eyes (201), (202). As a result, to demonstrate the
capability for drug screening studies in our microfluidic-based AH outflow model, hSCEC monolayer
were incubated with Rho kinase inhibitor, Y27632 (688000, EMD4 Biosciences, San Diego CA) prior
to and during B-A perfusion. In this case, the effect of drug treatment was assessed by evaluating
giant vacuole formation. For end point imaging, at the end of the experiment monolayer was
perfusion fixed and stained for actin cytoskeleton (phalloidin) and nuclei (DAPI) (detailed below).
vii.
Post fixation Imaging
To evaluate potential for traditional post-processing and ultrastructural visualization frequently
used in morphological studies of tissue samples in perfusion studies with enucleated eyes and
anterior chamber
segment microfluidic samples were prepared for transmission
electron
microscopy (TEM).
110
Sample preparation for Light and Transmission Electron Microscopy (TEM). Briefly, glass
coverslips were coated with a thin layer of PDMS (details below). Coated glass slides were used in
place of glass coverslips which are used to seal the PDMS device. All other setup and experimental
procedures remain unchanged. All samples were fixed and processed for TEM in situ by perfusion
of fixative and solutions via microfluidic channels. At the end of experiment, samples were fixed
with Karnovsky's fixative (2.5% glutaraldehyde and 2% paraformaldehyde in phosphate buffer, pH
7.4). Samples were post-fixed with 1% osmium tetroxide (Electron Microscopy Sciences, Hatfield,
PA) and 1.5% potassium ferrocyanide (Fisher Scientific Company, New Jersey) for 1 hour,
dehydrated in cold ethanol and embedded in Epon-Araldite (Electron Microscopy Sciences, Hatfield,
PA). Samples embedded in plastic were removed from the PDMS device and re-embedded in EponAraldite. Ultrathin sections were cut with an ultramicrotome, counterstained with uranyl acetate
(Fisher Scientific Company, New Jersey) and imaged by TEM (Model 300, Philips Eindhoven, The
Netherlands).
PDMS coated coverslips. To facilitate removal of fixed sample from microfluidic devices
after embedding in plastic, glass coverslips were coated with PDMS (to use in place of glass
coverslips). PDMS pre-polymer mixture was made by mixing 10 parts PDMS base solution with 1
part curing agent. A spin coater was used to create a thin layer of PDMS on top of the glass surface.
To create thin coats, individual glass coverslips were positioned in the centre on the spin coater
chuck, vacuum sealed and a small droplet of the degassed PDMS mixture deposited on the centre of
the coverslip. To create a uniform flat coating, excess material is spun-off (cycle 1: x seconds at x
rpm and cycle 2: y seconds at y rpm) and PDMS-coverslips baked in the oven at 80'C for at least 2
hours.
111
Immunocytochemistry for Confocal Microscopy. For high resolution fluorescent visualization
monolayers in flow samples were perfused-fixed and counter-stained for examination with confocal
microscopy. PFA fixed samples (4% PFA) were rinsed twice with 1X PBS, permeabilized with 0.1%
Triton-X and counterstained with 4', 6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, Atlanta,
GA) to identify nuclei and phalloidin (Invitrogen, Chicago, IL) to label F-actin.
112
d. Results
i.
3D Matrices for hSCEC culture: Collagen type I and Matrigel
Unlike previous studies, here isolated hSCEC are grown on 3D hydrogels instead of 2D filter
membranes.
Criteria I: Cell attachment and monolayer formation. To evaluate potential matrix
candidates for monolayer perfusion studies, hSCEC were cultured on collagen and Matrigel'"-based
gels. We found that hSCEC attach and spread on collagen, Matrigel and collagen/Matrigel blends.
However, monolayers formed only on pure collagen and collagen/Matrigel blended hydrogels. The
main protein components of Matrigel'
are laminin and collagen IV (61%, 30% respectively; BD
Biosciences website). For integrin mediated adhesions the corresponding integrin pairs are alp1 or
a2p1 for collagen and a3pl, a6p1 or a6p4 for laminin. Recently, cultured hSCEC were shown to
abundantly express collagen-specific (a2) and laminin-specific (3,
a6 and
p4) integrin subunits
(203). On pure Matrigel hydrogels, cells would attach and spread initially but would form clusters
of cells due to subsequent detachment (Figure 3-7). Presumably, the observed response is due to
difference in gel stiffness. This notion is supported by the observation that cells retain spread
morphology
on glass surfaces having a thin coating of Matrigel
and published
reports
demonstrating that cultured hSCEC express the necessary receptors to support integrin mediated
adhesions (203). Other cell types, for example NIH3T3 fibroblast and bovine aortic endothelial cells
(BAEC)
exhibit substrate stiffness dependent cell spreading response; where the degree of
spreading as measure by surface area increases with substrate stiffness (158). However, this
response was found to be eliminated when cell-cell contacts were established. Here, hSCEC
spreading and attachment was not improved by increased cell density and by establishing
113
subsequent cell-cell contacts (Figure 3-7, Figure 3-8) Based on these studies a decision was made
to use pure collagen and collagen/Matrigel'
blended gels.
Figure 3-7: Adhesion dynamics of hSCEC to pure Matrigel hydrogels
Micrograph of hSCEC cultured on pure Matrigel gels. Sequences taken from timelapse movie
recorded after single cells suspension of hSCEC were seeded on gel. (I.) Cells are initially rounded.
(II.) Cells begin to elongate, spread out onto matrix and make cell-cell connections. (III. - V.) Cell
sheet/monolayer contracts (VII.) Cell spheroid - cells detach from matrix.
Collagen/Matrigel'"
blended gels were qualitatively better for promoting longer-term
cell
attachment at higher ratio of collagen (Figure 3-8). Furthermore, preliminary experiments
suggests hSCEC barrier function to high molecular weight dextran (150kDa; Invitrogen) was higher
for 75:25 compared to 50:50 collagen/Matrigel blended gels as assessed by concentration
114
difference across the monolayer. In subsequent experiments pure collagen and collagen/Matrigel"
blended gel at 75:25 ratio were used.
Collagen/Matrigel Blended gels
Figure 3-8: Adhesion of hSCEC to Collagen/Matrigel Blended gels
Micrograph of hSCEC cultured on pure collagen, Matrigel and collagen/Matrigel blended gels. Label
(0:100) indicates the ratio by volume of collagen to Matrigel. (Top row) Pure gels, cells are rounded
on pure Matrigel whereas on collagen they are spread. (Bottom row) Blended gels, progressively
better attachment and spreading.
Criteria II: Tracer bead passage (additional details in appendix). Bead passage depends
on matrix density and pore size. Consequently to conduct filtration pattern experiments a trade off
in matrix concentration/density (correlated to matrix stiffness) was necessary, since hSCEC
adhesion is better (see appendix: Figure 3-16; spread area of gels of different concentration) on
stiffer gels whereas easy bead passage (see appendix: Figure 3-20, Figure 3-21) requires less
115
dense (correlates with lower stiffness) gels. For pure collagen gels concentrations of 2.5-3.0 mg/ml
both accommodated monolayer formation and easy passage and traceability of 200nm beads with
and without a monolayer.
ii.
Primary Schlemm's Canal endothelial cells on 3D matrix as a model for AH
outflow and formation of Giant Vacuole-like Structures
1.
Variability of hSCEC response on 3D matrix
A majority of the microfluidic perfusion experiments and reported results are for hSCEC
monolayers formed on collagen gels. Prior to perfusion, monolayers where visualized with phase
microscope to examine cell attachment and completeness of monolayer. Interestingly, in some
cases hSCEC monolayer were not quiescent and extended membrane projections (Figure 3-17)
several hundred microns penetrating the underlying 3D matrix. While seemingly unfavorable, in
vivo inner wall endothelial cells and cells within the underlying JCT region are known to extend
processes which form gap and adherens junction with each other (180).
Furthermore, hSCEC and
human trabecular meshwork cells (HTM) cultured in vitro form cell processes that extended into
filter membrane below the monolayer (190), (191). In more extreme cases, single hSCEC migrate
from monolayer into the 3D matrix, which present a challenge for 3D culture models and might be a
sign of differentiation while in culture (see appendix for additional details). Recently, reports from
the Stamer lab (hSCEC source) alluded to this in the context of laminin expression profile. They
observed that the "immature" protein (laminin-332) is expressed by hSCEC and TM cells cultured
in vitro but not expressed in situ in the human outflow tissue (only the "mature" version laminin511 is present) , from which they concluded that this marker may reveal that cultured cells are in
116
"an angiogenic state" (203). Their observations and conclusion is consistent with the migratory
phenotype that is observed in some hSCEC.
U
Human enucleated eye perfusion model
E
Microfluidic -based AH perfusion model
Image: Hann et al., Invest Ophthalmol. Vis. Sci. (2005) 46. 1-7
current thesis work
current thesis work
Figure 3-9: Comparison of Giant Vacuole morphology from in vivo and in vitro studies
Micrographs of Giant Vacuole structures obtained from (A) human enucleated eye perfusion model
(image reproduced from Hann et al., 2005) and (B, C) microfluidic-based 3D cell culture AH
perfusion model. Here, hSCEC monolayers were formed and perfused with a reservoir pressure
setup as described in Materials and Methods (with further details in appendix). In (B) cells were
stained with a Dii and image taken while cells were alive. (C) Fixed samples stained for actin (Red,
phalloidin) to visualize cytoskeleton and DNA to identify the cell nuclei (Blue, DAPI).
117
2.
Cultured hSCEC monolayer on 3D matrix form giant vacuole-like structures
Stable monolayers perfused via constant flow rate or pressure differential in the basal-to-apical
direction remained attached to the gel and formed giant vacuole structures (Figure 3-10, Figure
3-11). This is consistent with previous reports for hSCEC cultured in vitro on filter membrane
(191), (192). These structures can readily be observed with a light or fluorescent microscope.
Furthermore, the observed morphology is similar to those in previous studies (Figure 3-9),
featuring a "smooth-walled, round, oval, reniform or crescentic spaces" (184). These pressure
sensitive structures are visible on SCEC monolayers as early as 30 minutes of B-A perfusion;
however there is an inherent variability in responsiveness to B-A perfusion between SCEC from
different donors as was also observed by (192). Fixed samples of GVL structures stained for actin
cytoskeleton shows a diffuse labeling pattern at base and faint labeling along the "dome" (Figure
3-9C) which could be indicative of cytoskeletal remodeling in that region.
These results demonstrate that our microfluidic-based AH outflow model, like others is capable of
replicating in vivo observation with regards to GV formation on inner wall.
iii.
Real-time Visualization of Giant Vacuole Dynamics
Previous models have been limited in their ability to continuous monitor GV formation during the
course of an experiment. Giant vacuole formation and dynamics was observed under constant
velocity and constant pressure conditions. Figure 3-10 shows the growth of typical GV-like
structure over a 15 min interval. Longer term GV dynamics are shown in Figure 3-11. For GV size
quantification, cross-sectional area was calculated in a fixed plane of view which does not take into
account potential movement of GV in or out of the plane of view during the course of an experiment.
High resolution time-lapse confocal images, Figure 3-11 shows potential details of GV-like
118
structure formation documenting potential initiation, growth and fusion phases. Here smaller
structures are formed and subsequently coalesce to form larger structures. Also as expected,
retrograde perfusion results in the partial collapse of these structures (appendix: Figure 3-19).
Figure 3-10: Giant Vacuole-like structure dynamics: Perfusion at constant flow rate
Micrograph of hSCEC monolayer showing Giant Vacuole-like structures (A). Quantification of shortterm dynamics of GV (B). Here images were taken on Zeiss Axiovert 200M confocal microscope at
568nm.
119
1200
Growth
8 00 -
-!
Fusio0
S600
-
S400
-
U 200
-
0
Tim (mintes
100
200
300
Time (minutes)
Figure 3-11: Giant Vacuole-like structure dynamics: Perfusion at constant pressure
Confocal micrographs showing Giant Vacuole-like structure formation (A (I.) - (VII.)). Monolayer
labeled with Ca 2+ dye. Initial signs of GV formation evident by subtle changes in membrane (boxed
region in II.). GV emerge (III.), grow (IV.) and fuse (VI.) to form even larger structures.
Quantification of GV dynamics (B), sudden increase in area is due to GV fusion.
iv.
Pharmacological Inhibition
In vitro screening of potential outflow regulating drugs would be beneficial for the development of
new therapies while investigating contributing mechanism at the cellular and molecular levels. For
example, the disruption of cytoskeleton
or cytoskeletal
contractility with pharmacological
inhibitors including cytochalasins (disrupts actin cytoskeleton), latrunculins (disassembly of actin
120
filaments), H-7 or Y27632 (inhibits actomyosin-driven contractility) has been shown to reduce
aqueous humor resistance and intraocular pressure (204), (201), (202). To demonstrate our model
capability for investigating drug effect with cellular level details Y27632 was used. In preliminary
experiments, this Rho kinase inhibitor promotes GV-like structure formation (Figure 3-12) which
would be consistent with the in vivo observation of an increase in outflow facility following
treatment and increase GV formation in enucleated bovine eye model (201).
These results
demonstrate that our microfluidic-based AH model can be used to explore new glaucoma drug
therapies designed to target the inner wall endothelium and would complement current in vivo or
traditional ocular perfusion in vitro investigations by providing additional details at the cellular
level.
without Y27632
with Y27632
Figure 3-12: Effect of Y27632 on giant vacuole-like structures
Confocal micrographs of hSCEC monolayer (A) without and (B) with Y27632. Fixed samples were
stained for actin (Red, phalloidin).
121
v.
Hydrodynamic Filtration Pattern
Previous studies to label aqueous humor filtration pattern across the inner wall has been done
using enucleated eye perfusion models (183), (201). A characteristic, segmental or non-uniform
flow pattern is typical of the TM due to regions of preferential flow (183). However details of the
underlying mechanism remain unclear. Current in vitro outflow models (191), (192) are not
conducive to these experiments because they lack a 3D matrix. Furthermore, in situ bead tracking is
limited due to design-related imaging constraints (the need to refocus since beads would move out
of viewing plane). With the current model, beads and monolayer can be observed/monitor at single
focal plane without the need for refocusing. Proof-of-concept experiments done with our current
setup, demonstrates both uniform and non-uniform tracer labeling patterns were observed.
Interestingly bead accumulation upstream of monolayer seems to least in regions where GV-like
structure is present (Figure 3-13) or near cell-cell junction (thinnest regions of monolayer)
(Figure 3-13). Moreover, with this technique single 200nm tracer beads were also observed
crossing hSCEC monolayer near giant vacuole structures (Figure 3-14A). In some instances beads
were observed lingering Figure 3-14 A(ii) in regions where they co-localized with GV-like
structures before they were released Figure 3-14 A(iii) into the fluid channel. Furthermore bead
trajectories which indicated the path taken by fluid can be reconstructed Figure 3-14B.
122
'IU
Monolayer with microbeads
Microbeads
C
Figure 3-13: Variation in tracer labeling pattern
Confocal micrographs showing microbead tracer labeling pattern upstream of hSCEC monolayer.
(A-C) Images taken of live cells stained with Ca 2 dye. (Arrows) regions with least bead labeling,
that co-localize with Giant Vacuole-like structures (**).
123
Figure 3-14: Bead Traces
Confocal micrograph showing microbeads traversing hSCEC monolayer in the vicinity of Giant
Vacuole-like structure ((A) I. - (A) IV.). Boxed area, region of interest showing individual bead when
trapped (II.) then released (III.) into fluid stream. (B). In a separate device, shown are individual
traces made by microbeads in gel upstream of hSCEC monolayer in a region close (I.) or far (II.)
from the monolayer. Time-lapsed fluorescent micrographs where taken of the gel region and traces
generated in ImageJ using a built-in macro.
vi.
Ultrastructure by new TEM method for microfluidic samples
Light and transmission electron micrographs can be successfully prepared from microfluidic
samples despite the small sample dimensions (Figure 3-15). This capability is important because
TEM permits significantly higher resolution compared to light microscopes. For example, cellular
pores and details of GVL structures could be visualized with greater details. As a proof-of-concept,
124
monolayers cultured under static conditions were processed for examination via TEM. Detail
features of cell layer and underlying matrix are preserved (Figure 3-15 A-E) showing different
junctional morphology, intercellular organelles and matrix fibers. TEM preparation of hSCEC
(Figure 3-15B) and BEC (Figure 3-15C) are shown.
125
*t
U
4P
Figure 3-15: Light and Transmission Electron Microscopy
Light micrographs of gel region with an EC monolayer (A). Transmission electron micrograph of
hSCEC and BEC monolayers (B-E). Structure resembling a pore between two cells (B inset). (C, D)
Cell-cell junction morphology. (E) Higher magnification showing details of cell cytoplasm.
126
e. Discussion and Conclusions
Intraocular pressure (IOP) remains the only treatable risk factor for glaucoma. Pharmacological
treatment is a popular alternative to ocular surgery or laser treatment and work by modulating
aqueous humor production and/or drainage. As a result, a fundamental question with clinical
relevance pertains to the factors that regulate AH drainage. For this reason, in vitro models systems
have proven and continue to be invaluable tools for investigating details of the physiology of the
trabecular meshwork and aid in the advancement of new therapies. However, there exists a gap
between current in vitro models and in vivo native tissue which provides an opportunity for further
model improvements.
We developed
a new in vitro cell culture-based
model and conducted proof-of-principle
experiments which demonstrate its applicability for investigating AH outflow dynamics. A key
feature of the in vivo inner wall endothelium, giant vacuoles, was replicated in our system and its
dynamics could be monitored in real-time. Furthermore the mechanism of aqueous humor outflow
dynamics across the inner wall remain largely uncharacterized however with our system it is now
possible to label fluid path as demonstrated by hydrodynamic filtration pattern experiments.
Consequently, for the first time fluid flow can be visualized in a model of the inner wall endothelium
in real-time thus providing a unique tool for future investigations.
Furthermore, drug effects can be examined. Treatment with rho-associated kinase inhibitor,
Y27632 seems to promote giant vacuole formation. This is consistent with recent finding, where
cells exposed to Y27632 exhibited a dose dependent cell softening (205). Presumably cells that are
softer are more deformable and would more readily form giant vacuole structures. These
127
observations along with in vivo finding of increased outflow facility following Y27632 treatment are
consistent.
Gap in Replicating the Inner Wall Microenvironment
Unlike previously reported models, here hSCEC are cultured on 3D hydrogels gel instead of 2D filter
membranes for perfusion. It is widely accepted that in vitro culture dimensionality is important and
that 3D cultures narrows the gap between in vitro models and live tissue (206), (124). Of particular
importance is the region
JCT region that lies adjacent to the IW endothelium can only be modeled in
cultures with 3D capability. Consequently, this aspect of the trabecular outflow pathway cannot be
readily addressed with current models (191), (192). Moreover, independent studies taken together
highlights the various synergistic interactions that occur with the trabecular outflow tissue that
ultimately impacts the regulation of AH outflow dynamics (177), (180), (178), (194), (193), (195),
(196), (197), (198), (199). Following this line of reasoning, our microfluidic-based platform sets the
stage future investigations where an integrative approach can be taken to investigate the
physiology and pathophysiology of trabecular outflow tissue. Consequently, this study is a nontrivial improvement to currently available cell-based perfusion models (191), (192).
Real-Time Visualization of Giant Vacuole Formation and Hydrodynamic Filtration Pattern
Real-time visualization is a coveted capability when designing in vitro models of biological systems.
In this case, the in situ visualization of the formation giant vacuoles by hSCEC is important for
investigating mechanism of formation. To our knowledge, this was only achieved recently by one
other group (192) , where they used a 2D filter-based perfusion setup and cells are visualized en
face. With this system they were able to readily capture GV dynamics as assessed by lateral
movement and growth of "void" in the plane of the monolayer. With our system, however the
128
natural viewing plane produces a cross-sectional view of the monolayer and subsequent GVs that
are formed in response to a basal-to-apical pressure gradient. This view is advantageous because
the entire thickness of the monolayer can be visualized at one plane which would provide greater
details on the mechanism of GV formation.
Studies of giant vacuoles suggest that these pressure sensitive structures are indeed markers for
region of active flow across the inner wall (184). In agreement with other studies, cultured hSCEC
cultured monolayer forms giant vacuole-like structures in response to basal-to-apical perfusion.
Furthermore, tracer beads were observed crossing the endothelium in the vicinity of giant vacuole
structures. Time-lapse microscopy studies suggest that giant vacuoles might not a ballooning of the
inner wall endothelium as previously thought by some in the field, since in the giant vacuoles
examined gross cell detachment was not evident. But instead, might be due to subtle changes at the
basal surface of the cell.
Collectively, these proof-of-concept experiments demonstrates that this platform presents a
promising tool for investigating the effects cellular and molecular mechanisms which are
responsible for modulating aqueous humor outflow facility.
129
f. Appendix: Supplementary Material
hSCEC Spread Area on Different Concentration Gel
3mg/ml collagen
6mg/ml collagen
I
0.25
B.
0.2
CZ
<
0.15
0.1
0.05
0
1 mg/mI
3 mg/mI
6 mg/ml
Concentration of Collagen Gel
Figure 3-16: hSCEC spreading on collagen gels of different concentration
(A) Fluorescent micrographs of hSCEC on pure rat tail collagen type I of different concentrations (1,
3 and 6 mg/ml). (B) Quantification of cell spreading on gels of different concentration. Fluorescent
micrographs were analyzed in ImageJ; average cell area (arbitrary units) was calculated for cells on
gels of different concentration. Arbitrary cell area increases with increased gel concentration.
130
Potential Challenges of 3D Culture?
Figure 3-17: hSCEC monolayer extends membrane projections in underlying matrix
Micrograph showing cross-sectional view of hSCEC monolayer on rat tail collagen type I gel (6.0
mg/ml) cultured in microfluidic device. Filopodia/Invadopodia-like membrane projections extend
from hSCEC monolayer (at top of image) into the underlying collagen matrix to form branched
structures.
131
Expression of junction Proteins
The expression profile of junction proteins (adherens, gap and tight junction) in cultured hSCEC
was
analyzed
by
immunofluorescence.
Fixed
hSCEC
monolayers
were
prepared
for
immunocytochemistry as previously described. A summary of the results is shown in Figure 3-18.
132
Jun1lctionl Pr~oteinsI
DAPI]
VE - Cadherin
PECAM
E - Cadherin
Claudin
Connexin 43
zo - 1
Figure 3-18: Expression profile of junction proteins in cultures hSCEC
Fluorescent micrographs of hSCEC (on glass) immunolabeled with a panel of antibodies for cell-cell
junction proteins (VE-Cadherin, PECAM, E-Cadherin, Claudin, Connexin 43 and ZO-1).
133
Pressure gradient sensitivity of giant vacuole-like structures
Figure 3-19: Pressure gradient sensitivity of GVL structures
Micrographs showing the dynamics of GVL structures in response to the direction of the applied
pressure gradient and the resultant fluid flow across the hSCEC monolayer. White arrows indicate
the direction of the pressure gradient (arrow's tail-to-head corresponds to high-to-low pressure).
(A) Micrograph showing GVL structures (indicated by asterisks) formed by hSCEC in one region of
microfluidic device. Here pressure gradient and flow directions (basal-to-apical) are similar to the
in vivo condition. (B) Micrograph showing the same region in (A) when pressure gradient and flow
direction are reversed. Asterisks label the position of one of the GVL structures that have
disappeared/collapsed in response to change on pressure gradient.
Bead size selection
Preliminary experiments were conducted to determine bead size and density for use in tracer bead
experiments
for determining gel permeability and fluid hydrodynamic filtration patterns.
Fluorescently-tagged microbeads of different sizes (20 nm, 100 nm, 200 nm and 500 nm) were
evaluated for ease of passage and traceability first in gels (i.e. different collagen concentrations and
collagen/Matrigel blends) without a monolayer and finally with an hSCEC monolayer. Fluid flow
through gels was established with reservoirs (see pressure setup details below). Low resolution
phase and fluorescent micrographs from a typical experiment is shown in Figure 3-20.
134
Fluorescenit Imag1(es (11. - VI.)
Before perfusion with beads
After perfusion with beads
Bead solution
VI.
vof
a
W
Figure 3-20: Evaluation of bead passage through 3D matrix in "gel-cage" region
Phase (1) and fluorescent (II - VI) micrographs showing the "gel-cage" region of the microfluidic
device. Micrographs illustrate the sequence of steps during a typical preliminary experiment to
evaluate bead passage and suitability for subsequent studies (e.g. measuring gel permeability in
situ, labeling hydrodynamic filtration pattern in the case when a monolayer is present).
Micrograph(s) taken (I) before and (III-VI) after perfusion with media containing beads. (VI) Inset,
higher magnification showing typical bead trajectory in micrographs taken at high exposure times.
For quantitative experiments higher resolution images (> 20X objective) of gel region is necessary,
the inset of Figure 3-20 VI shows typical trace that is obtained at higher magnification. Bead sizes
of 20 nm and 100 nm were found to pass freely through 3.0 mg/ml collagen gels but individual
beads were not easily traceable whereas 500nm beads frequently got trapped in the gel (Figure
135
3-21). Based on the result from these experiments 200nm bead size was selected for subsequent
experiments.
Figure 3-21: Evaluation of bead passage through 3D matrix
Fluorescent micrographs showing microbeads in gel region of microfluidic device (I. -III.).
Consecutive snap shots showing 500 nm microbeads trapped in 3.0mg/ml collagen gel as a result
this bead size would not be suitable for bead tracer experiments. Scale bar 100 [m.
136
2E-1 3
1.8E-13
1.6E-13
1.4E-13
S1.2E-1 3
I
*OW
ASOW
40M
4NOW
1E-13
8E-14
CL
6E-14
4E-14
2E-14
0-
0
2 mg/ml
2.5 mg/m I
4
--
6 mg/ml
75/25 blend
Figure 3-22: Experimental estimation of gel permeability
Graph showing gel permeability as a function of concentration for pure rat tail collagen type 1 (2,
2.5 & 6 mg/ml) and collagen/Matrigel" blended gels (75/25 blend, where ratio indicate percent by
volume; 75% pure collagen (6.0 mg/ml) and 25% Matrigel" (undiluted)). Permeability was
determined experimentally using a bead tracer method (see section 4.6 for additional details).
(Inset) Micrographs of "gel-cage" during a typical bead tracer experiment and higher magnification
image showing a single trace obtained for images taken with high exposure setting.
137
g. Appendix: hSCEC Maintenance and Microfluidic Protocol
Human Schlemm's canal endothelial cells (hSCEC) Maintenance
(Original protocols for general hSCEC culture were provided by Kristin Perkumas from Stamer
Lab at University ofArizona. This protocol was modified to include additional details and
notes.)
Flasks containing proliferating cells were shipped overnight. Upon arrival cell medium was
refreshed, flask incubated and subsequently expanded to generated sufficient cells for future
experiments.
Freezing down hSCEC cells
Only healthy looking cells should be frozen down. Cells should be at approximately 80% confluence,
look healthy and not older than passage 3. Examine cells to make sure they are not too large or
contain a lot of vesicles.
Procedure:
1.
Remove media and replace with PBS (w/o Ca 2+ or Mg 2 +). Gently, rock flask to wash cells.
Aspirate PBS.
2.
Add warm trypsin to cells in flask, approximately 0.5ml for a T25 flask. Tilt flask back and
forth so that entire cell surface is covered with trypsin. Cells generally detach quickly, as a
result you do not need to incubate at 37C. Monitor cell detachment with microscope. Tap
flask gently to help dislodge cells that remain attached. Do not wait too long in an effort to
recover every cell from the flask. Work quickly; you want to have vial cells at the end of the
freezing protocol.
3.
Add 9.5 ml of warm cell culture medium (DMEM supplemented with 10% FBS and 1x PSG)
to cells/trypsin suspension.
4. Transfer contents of flask to a 15ml tube and centrifuge for 5 minutes at 600 rpm. hSCEC
cells pellet very easily.
5.
Spray down tube with ethanol and return tube with pellet to sterile hood. Be careful not to
knock the tube as this might dislodge your pellet and you could potentially lose it.
6.
Carefully aspirate all media/trypsin supernatant.
138
7.
Resuspend cell pellet in chilled 900 pil of FBS. To generate a uniform suspension without
need for excessive pipetting add only 300 [d to cell pellet and gently tap tube to "break up"
pellet. Add the remaining 600 [il and pipette gently.
8.
Transfer cell suspension cryo vials that contain 100 il of DMSO. Generally you will be
freezing down multiple flasks which would require that multiple vials containing DMSO are
ready for this step.
9.
Cap vials and mix gently.
10. Place vials on ice for 5 minutes.
11. Transfer vials to freezer box and store at -20 C for 4 hours. Move freezer box and cryo vials
with cells to -80'C for 4 hours and finally to liquid nitrogen storage tank.
Thawing hSCEC cells
(You may want to thaw in to a T12
if you don't have a lot of cells)
Procedure:
1.
Before removing frozen cell stock from liquid nitrogen storage warm cell culture medium
(DMEM supplemented with 10% FBS and 1x PSG) to 37C.
2.
Add 5ml of medium to collagen coated and labeled flask.
3.
Remove cryo vial from storage, warm vial in a water bath at 37 0C until contents of the starts
to melt. You should see an ice pellet surround by liquid. (Take care that the lid is not below
the water level to avoid any chance of potential contamination).
4.
Spray and wipe down vial with ethanol. Return to sterile hood.
5.
Add 1ml of warm media to cell suspension and ice pellet (this should thaw the remaining ice
pellet).
6.
Transfer contents to the prepared T25 flask containing warm cell culture medium (you
might want to keep the flask in the incubator until you reach this step).
7.
Rinse the cryo vial with another 1 ml of warm medium and add to flask with cells.
8.
Incubate flask containing cells at 37"C and 5% CO2 for 1.5 to 3 hours. After approximately 2
hrs check flask to monitor cell attachment. They should look like fried eggs initially. It is
important to be GENTLE and not disturb cells during this attachment process.
9.
Once a majority of the cells have started to attach (- 2hours) remove the thawing media
and transfer to another flask to allow other cells to attach. Add warm cell culture media to
139
cells in the original flask. Do not pipette media directly over cells, instead gently run media
down the side of the flask then return flask.
10. Return both flasks to the incubator. (In most cases the cells in the second flask are not as
viable)
PDMS Devices Fabrication
Materials and Equipment
1.
2.
3.
4.
5.
6.
7.
8.
SU-8 Wafer (attached to bottom of oven safe Petri dish)
Sylgard 184 Silicone, PDMS Kit (base and curing agent); cat# 184 SIL ELAST KIT 0.5kg
Degasser w/ house vacuum connection line
Clean Disposable cup & Stirrer
Digital scale
Oven
Razor/Scalpel
PDMS corer (35mm diameter)
Procedure
1.
Weigh PDMS base and curing agent in disposable cup at 10:1 ratio.
2.
Mix base and curing agent vigorously and thoroughly. A well-mixed batch should have a lot
of bubbles.
3.
Degas mixture in vacuum sealed degasser. 30 minutes should be sufficient for 110 g mixture
(i.e. 100 g base plus 10 g curing agent). It might be necessary to vent chamber a couple
times to prevent PDMS mixture from overflowing. During the first 5 minutes, mixture will
foam over.
4.
Pour degassed PDMS mixture over SU-8 wafer to a depth of approximately 0.8-1.0 cm
(thickness does not need to be exact, however thicker devices are much easier to handle
and ensures delamination does not occur when attaching external plumbing).
5.
DO NOT return wafer and PDMS mixture to degasser.
6.
Leave wafer and PDMS out on bench for a few minutes (-5 minutes should be fine). Surface
bubbles equilibrate quickly. You may use a tooth pick (lab supplied) or sharp point to
dislodge any bubbles that may be stuck to small pattern features.
7.
Bake PDMS in oven at 80 C for at least 2 hours.
8.
Remove the cured PDMS from the oven and leave it out on the bench to COOL.
140
9.
Once cooled use a scalpel to cut out PDMS slab.
10. Punch out individual patterns with PDMS corer. This should be done on a clean surface with
pattern side up, to minimize dirt and debris from sticking to pattern surface. Keep
individual device in a clean container (large Petri dishes work well) with pattern side facing
upwards.
11. Core ports as needed (to avoid getting dust in ports, core holes when you need the devices).
Additional Notes:
1.
Do not fabricate and store PDMS to far in advance. Old devices tend not to bond to glass or
PDMS-coated coverslips very well.
2.
Use a sharp corer to punch out ports. A dull corer will cause PDMS to tear and may cause
leaks when connecting external plumbing (reservoirs, tubing etc.)
Cleaning and Sterilization
Material and Equipment
1.
Individual PDMS devices (do forget to punch out ports)
2.
Scotch tape
3.
Large Beaker
4. Water (Millipore or from DI line)
Procedure
1.
Clean PDMS surfaces thoroughly with scotch tape. Ensure surfaces are free of
dust/particles.
2.
Place PDMS into large beaker containing water, cover with aluminum foil. You may also
prepare "humidity box" which will be needed during hydrogel loading/gelation process.
Add -400mL of water to empty pipette boxes (1000 pl pipette tip box with tray works well
- fits 6 devices).
3.
Place beaker with PDMS and "humidity box" into secondary bin and autoclave for 20
minutes on wet cycle.
4.
At the end of wet cycle, remove bin, promptly drain water off PDMS devices and transfer to
empty pipette tip boxes (NOT "humidity boxes"). Ensure pattern side face upwards. Set
141
"humidity box" aside to cool then transfer to incubator (HOT humidity box directly from the
autoclave will disturb the set point temperature in the incubator).
5.
In preparation for second autoclave cycle, load glass cover slips into clean empty pipette
box. Place pipette tip boxes with PDMS devices and glass coverslips in secondary bin and
transfer to autoclave. Run dry autoclave cycle (20min/15min).
6.
At the end of dry cycle, place pipetted tip boxes in laminar flow hood in tissue culture room.
Open lids and let devices and coverslip cool down for approximately 10 minutes.
Additional Notes:
1.
After the wet autoclave step do not leave devices to sit in water for a long time. The PDMS
devices will absorb water and appear cloudy. Do not proceed to gel loading step with cloudy
devices as this will make the gel loading step difficult. Dry out device in oven; maintain
sterile seal.
PDMS Surface Treatment
Material and Equipment
1.
Plasma cleaner w/ vacuum pump setup
2.
Slides (tray for holding devices during plasma etching)
3.
PDMS devices (cleaned and sterilized)
4.
Tweezers
5.
Ethanol
6.
Timer
7.
Gloved hands
Procedure
1.
Turn on plasma cleaner, spray and wipe down outside of chamber (NOT INSIDE). Spray and
wipe down working area and tweezers. Spray and wipe down trays for holding devices.
2.
Etching cycle is 2 minutes pump down followed by 2 minutes of irradiation with
purple/pink air plasma. Run one cycle with trays to warm up the machine. At the end of the
cycle, vent chamber quickly.
142
3.
Remove trays and load devices with pattern side facing upwards. Place trays with devices in
plasma etcher chamber (Do not overcrowd chamber). Run etching cycle, 2 minutes pump
down followed by 2 minutes irradiation with plasma. Check plasma color periodically to
ensure you have a stable color.
4.
At the end of each cycle, vent chamber and quickly transfer devices back to pipette tip box.
5.
Repeat etching cycle until finished.
Gel Loading by Macro- or Microinjection
Materials and Equipment
1.
Microinjection System (Computer monitor, Microscope (Digital Blue Toy microscope), Glass
slide-PDMS holder, 3 axis (xzy) micromanipulator with microsyringe holder, magnetic base
and plate)
2.
Hydrogel pre-polymer mix on ICE
3.
Microliter syringe (already cleaned)
4.
Plasma etched/treated PDMS devices
5.
Sterilized glass coverslips
6.
Cell culture grade water
7.
"Humidity box" (just out of 37C incubator)
Procedure
1.
Setup microinjection system in the hood, start microscope program and check for a live
feed. Spray and wipe down surface a knob with ethanol.
2.
Flush needle with cold cell culture water to ensure no bubbles are in the needle.
3.
Position PDMS device on microscope stage with pattern surface facing upward. Focus and
center the "gel-cage" by monitoring the live feed on the computer monitor.
4.
Load microliter syringe with hydrogel pre-polymer mix and discard (eppendorf tubes are
handy for this purpose). Repeat this a couple times.
5.
Load microliter syringe with pre-polymer mixture and position the syringe tip above the
PDMS gel-cage in preparation for gel loading.
6.
Generate a small droplet of pre-polymer mixture at the tip of the needle, and fill gel cage
drop wise. For initial drop, lower syringe needle until the droplet makes contact with the
143
PDMS surface (center in "gel-cage"); fill by carefully creating small droplets and bringing
them to the surface by maneuvering the manipulator xyz axis/joystick as necessary.
7.
WORK quickly; this becomes second nature with practice.
8.
Lower microscope stage, remove tray with device and seal channels and "gel-cage" using a
glass coverslip. Push down on coverslip to ensure a good seal, starting at the edges of the
device.
9.
Transfer sealed devices loaded with pre-polymer mixture to the "humidity box". Repeat
steps 3, 5-9 for a couple more devices then incubate at 37 C for 30-35 minutes. If you
experience significant resistance while dispensing droplets perform steps 2 and 4 before
loading mixture into devices.
10. Following gel polymerization, remove humidity box containing device from the incubator
and transfer to hood. Wipe off any moister from glass slide surface and transfer individual
devices to a clean petri dish (large Petri dish works well for when handling multiple dives
simultaneously).
11. Inspect gels under microscope, fibers should be visible and is an indication of a well formed
gel. Back in the hood, carefully and slowing fill microfluidic channels with cell culture
medium by advancing the liquid front pass the gel cage. Rapid introduction of fluid might
result in bubbles near the gel (i.e. when approach the gel quickly, the fluid front is slower at
the gel surface than the PDMS wall causing the front not to advance uniformly thus trapping
air at the gel channel interface).
12. Repeat the media filling process for each device. Fill enough media just up to the levels of
the port.
13. Store PDMS device in the incubator overnight in preparation for cell loading step. Overnight
incubation before further handling helps to prevent leaks between glass coverslip and
PDMS surface since bond strength increases with time.
Human Schlemm's canal endothelial monolayer formation
Materials and Equipment
1.
Proliferation flask of hSCEC
2.
Cell culture medium
3.
PBS (w/o Ca 2 + or Mg 2 +)
144
4. Warm trypsin
5.
Cell seeding box (Pipette tip box with glass slides
Procedure
1.
Prepare single cell suspension (1.0
x
106 cell/ml)
a.
Aspirate media from flash and wash with PBS (5ml for T25 flask)
b.
Add 0.5ml warm trypsin, tilt flask to distribute over cells and tap gently to help cell
detachment. Cell detachment from flask is quick (< 1min) and does not require the
typical incubation step during trypsinization.
c.
Add 4.5ml of warm cell culture media to inhibit trypsin, transfer contents to 15ml
tube and centrifuge at 600 rpm for 5 minutes.
d. Aspirate media and resuspend cells in cell culture media at the desired
concentration.
2.
Transfer PDMS devices from incubator to hood. Aspirate media droplet and flush channels
with warm media.
3.
Pipette cell suspension, tilt device and add to one port while keeping device in titled
orientation (this allow cells to quickly accumulate near gel surface).
4.
Transfer individual device to cell seeding box -keep tilted orientation.
5.
Repeat 3-4 for all devices.
6.
Incubate cell seeding box and culture devices in this orientation for at least 2-4 hours at
37C and 5% C0 2 .
7.
Transfer cell seeding box from incubator to hood, wipe any moisture from glass coverslip
surface and return devices to upright position in a clean Petri dish. Quickly examine cell
attachment to gel under microscope. Flush channel once with warm cell culture media and
add droplets to each port.
8.
Maintain hSCEC in static culture for 2-3 days. A daily media exchanges is required.
Additional Notes
1.
Scalloped or concave gel interface does not work well for generating hSCEC monolayer, a
flat surface or slight overfilled gel works better.
145
Human Schlemm's canal endothelial monolayer Derfusion
Material and Equipment
1.
hSCEC monolayer in device
2.
Live cell dye, Dil (D-282, Molecular Probe) gift Michael Murrel
3.
Microscope w/ environmental chamber
4.
cell culture medium
5.
Reservoirs
6.
Push/Pull syringe pump
7.
Syringe (new or sterilized)
8.
Tubing (sterilized)
Live Cell Staining
Prior to setting up flow stain hSCEC monolayer with live cell dye (- 2 hrs).
1.
Prepare warm cell culture medium (in 37 C water bath) with (Dil at 1:1000 dilution) and
without cell staining dye.
2.
Remove PDMS device with hSCEC monolayer from incubator and transfer to hood.
3.
Aspirate media droplets from ports and perfuse channel with warm cell culture media
without dye (to wash out any dead cells from channel).
4.
Perfuse apical channel twice with medium supplemented with dye.
5.
Add droplets to ports (i.e. media with dye to apical channel ports and media without dye to
basal channel ports) and incubator for approximately 25 minutes.
6.
At the end of the staining period, in the hood remove droplets and perfuse apical and basal
channels with medium without dye. Wash apical channels thoroughly by perfusing and
exchanging media 2-3 times.
146
AM,
6
0
"-
Figure 3-23: Constant Flow Rate and Pressure Setups
Fluid circuit layout for constant flow rate (A) and pressure (B) setups for generating
transendothelial flow in AH outflow model
Constant flow rate setup
To generate flows rate within the physiologic range use the in vivo estimate of AH outflow rate
(-2.5 pm/min) and projected surface are of Schlemm's canal to calculate an estimated flow rate per
unit surface area.
Procedure
1. Fill syringe with cell culture medium.
147
2.
Attach 3-way valves, reducing connector and tubing to syringes. Figure x-x show an
example of parts connections and setup.
3.
Remove device (monolayer stained with live cell dye for fluorescent imaging) from
incubator. In hood, aspirate droplets and flush channel with cell culture medium.
4.
Connect inflow tubing to basal channel ports and outflow reservoirs to apical channel ports
(with spill of port) to device.
5.
Transfer setup to microscope equipped with an environmental chamber (not shown above).
6.
Setup image acquisition program and ROI on monolayer.
7.
Start syringe pump.
8.
Immediately start image acquisition. Survey any monolayer response at lower
magnification when possible and magnify. for better resolution as necessary.
Reservoir pressure setup
An example of the setup for defining the pressure drop across the hSCEC monolayer is shown in
Figure 3-23B. In this setup, the pressure drop across the inner wall is used as a guide for
experimental pressure levels.
Procedure
1.
Fill reservoirs (4 per device) partially with cell culture medium via reducing connector
(check to ensure that small bubbles are not trapped in connector).
2.
Transfer devices with monolayer from incubator to the hood. Aspirate droplets from ports
and perfuse channel with warm cell culture media.
3.
Carefully connect reservoirs one at a time while making sure not to trap any bubble in
connector or ports. Top off liquid in the connector region of the reservoir.
4. Add warm media to reservoir connected to the basal channel ports. Here the relative
difference in the liquid level in the reservoir establishes pressure differential across the
hSCEC monolayer thus generating flows.
5.
For live cell imaging, transfer setup microscope stage equipped with an environmental
chamber otherwise return setup to the incubator (5% C0 2 and 37 C).
148
6.
Setup up image acquisition software and ROI on monolayer.
7.
Start image acquisition. Survey any monolayer response at lower magnification when
possible and magnify for better resolution as necessary.
Hydrodynamic Filtration Pattern
The perfusion setup for hydrodynamic filtration pattern with tracer beads is the same as described
above except that cell culture media in the basal channel and reservoirs connected to basal channel
is replace with cell culture media containing microbeads.
Beads size selection. Initial experiments were done with 500nm, 200nm, 100nm and 20
nm beads to examine bead passage through gel. 500nm beads frequently got trapped and 100 nm
beads were too small for easy tracking. Based on results 200nm bead size was selected.
Material and Equipment
1.
hSCEC monolayer in device
2.
Live cell dye, Dil (D-282, red-orange, Molecular Probe) gift Michael Murrel
3.
Microscope w/ environmental chamber
4.
Warm cell culture medium
5.
200nm microbeads (green)
6.
Reservoirs
Procedure
1.
Fill reservoirs partially with cell culture medium via reducing connector. Two reservoirs
with only media and two with media containing microbeads (1:10000 dilution). Check to
ensure that small bubbles are not trapped in connector.
2.
Transfer devices with monolayer from incubator to the hood. Aspirate droplets from ports
and perfuse channel with warm cell culture media.
3.
Carefully connect reservoirs (connect apical reservoirs first) one at a time while making
sure not to trap any bubble in connector or ports. Top off liquid in the connector region of
the reservoir with appropriate media.
149
4.
Add warm media containing microbeads to reservoir connected to the basal channel ports.
Here the relative difference in the liquid level in the reservoir establishes pressure
differential across the hSCEC monolayer thus generating flows.
5.
Transfer setup microscope (Epifluorescent or Confocal). An environmental chamber is
needed for long term tracking experiment. In these studies, imaging with epifluorescent
microscope works well for visualizing bead traces as fluorescent streaks (high exposure
acquisition).
6.
Acquire time-lapse images.
7.
At the end of the experiment transfer micrographs to ImageJ for further analysis (e.g.
reconstructing bead tracks, determining bead velocities and patterns)
150
Chapter 4: Biomechanical regulation of angiogenesis **
**Material in this chapter has been submitted to a peer-review journal
a. Chapter Abstract
A bias towards angiogenesis from the venous circulation has long been known, but its cause
remains unclear. Here we explore the possibility that high interstitial pressure in tumors and the
resultant net filtration pressure gradient that would induce flow from the interstitium into the
venous circulation or lymphatics could also be an important mechanical regulator of angiogenesis.
The objective of this study was to test the hypothesis that basal-to-apical (B-A) transendothelial
flow promotes angiogenesis and to investigate potential mechanisms. Macro- and microvascular
endothelial monolayers were cultured on type I collagen gels in a microfluidic cell culture device
and subjected to apical-to-basal (A-B) and B-A transendothelial flows. Samples were perfusion fixed
and analyzed for morphological responses, localization and degree of phosphorylation of certain
signaling proteins. Application of B-A, but not A-B flow, to cultured endothelial monolayers was
found to promote capillary morphogenesis and resulted in distinct localization patterns of VECadherin and increased FAK phosphorylation. These results suggest that B-A flow triggers the
transition of vascular endothelial cells from a quiescent to invasive phenotype and that the flowmediated response involves signaling at cell-cell and cell-matrix interfaces. These results support
the hypothesis that transendothelial pressure gradients resulting in B-A flow promotes sprouting
angiogenesis and are consistent with early observations that tumor angiogenesis occurs from the
venous side of the circulation.
151
b. Introduction
Angiogenesis - the formation of new blood vessels from a pre-existing parent vessel - is commonly
defined by the imbalance of soluble pro- and anti-angiogenic factors (44).
The biochemical
regulation which involves the binding of soluble ligands to endothelial cell (EC) surface receptors
and subsequent downstream signalling have been painstakingly mapped out by decades of research
(143), (207), (208). The in vivo endothelial microenvironment however, is a complex integration of
both biochemical and biomechanical factors, which together promote either a quiescent or
angiogenic phenotype. The relative importance of biochemical or biomechanical stimuli in the
regulation of endothelial fate and function remains unclear. Nonetheless, it is widely accepted that
mechanical forces regulate endothelial cell growth, differentiation, motility, protein synthesis and
gene expression (209).
Studies of the impact of mechanical stimulation due to fluid flow on cultured endothelial
monolayers traditionally focus on the effects of shear stress - an important regulator of vascular
tone, homeostasis, inflammatory and immune response. In vivo, endothelial cells are subjected to
both surface shear flow as well as transendothelial flow - fluid filtration across the endothelium caused by pressure differentials between luminal/apical and abluminal/basal endothelial surfaces.
Apical-to-basal (A-B) transendothelial flow has been shown to inhibit the transmigration of
neutrophils across human umbilical vein EC (HUVEC) cultured on polycarbonate filters (210), alter
EC transport properties (211) and influence capillary morphogenesis (131) (212), (213), although
the mechanisms for this remain to be elucidated. Early studies of tumor angiogenesis made the
observation that new vessels emerge predominantly from venules (214). However, the direction of
transendothelial flow, which is also physiologically relevant, has only recently been investigated in
the context of endothelial function or sprouting angiogenesis (212), (213).
152
High interstitial pressure is a hallmark of neoplastic tissue (215), (216) and gradients in
interstitial pressure are thought to play an important role in a variety of developmental processes
(217). With high interstitial pressure induced by the solid tumor, neighboring blood vessels from
the lower pressure circulation are potentially affected by the resulting transmural pressure, causing
vessel collapse and potentially influencing the rate of transendothelial flow. The resultant net
filtration pressure gradient would induce flow from the interstitium into the venous circulation or
lymphatics. Endothelial cells lining these vessels would be subjected to transendothelial pressures
with higher pressure on the basal-surface compared to the apical surface. Under these conditions
basal-to-apical (B-A) transendothelial flow is highly probable.
The objective of this study was to study the role that B-A transendothelial flow exerts on
sprouting angiogenesis and to investigate potential mechanisms. In this study, we present evidence
of the initial signalling cascade that is initiated by B-A flow and propose links based on our results
and
previously
published
work that supports
an
argument
for mechanically-stimulated
angiogenesis. We find that FAK-mediated signalling accompanied by extensive remodelling of cellcell
junctions
and
redistribution
of the
actin cytoskeleton
contributes
to the effect
of
transendothelial flow on vascular sprouting.
153
c. Materials and Methods
i.
Cell Culture
Human dermal microvascular endothelial cells (HMVEC - cc-2643, LONZA, Walkersville, MD) and
human umbilical vein endothelial cells (HUVEC - cc-2617, LONZA) were expanded on collagencoated flasks in EGM-2MV (cc-3202, LONZA) and EGM-2 (cc-3162, LONZA) medium, respectively.
Cells were cultured in a humidified incubator at 37 'C and 5% C02 . All experiments were conducted
with passages 4-6 cells. Monolayer maintenance medium was EGM2MV without hydrocortisone.
Experimental medium was maintenance medium supplemented with VEGF165 (293-VE-010, R&D
Systems Inc., Minneapolis, MN) or VEGF121 (4644-VS-010, R&D Systems Inc.) where indicated.
ii.
Microfluidic - based cell culture platform
A microfluidic-based cell culture system with the demonstrated capability of controlling the
biochemical environment and flow was used in these studies, as described in detail in (41). Briefly,
the design includes two independent microfluidic channels that are separated by a central region "gel cage" for housing injectable hydrogels (synthetic peptide, Matrigel, collagen) allowing for
simultaneous culture on two-dimensional (2D) surfaces and within three-dimensional (3D)
matrices. This geometry permits the establishment of pressure gradients and small interstitial fluid
flows through the 3D matrix. The device is fabricated using standard soft lithographic techniques
from Polydimethylsiloxane (PDMS - 184 SIL ELAST KIT, Ellsworth Adhesives, Germantown WI), a
widely used biocompatible and optically transparent elastomer for microfluidics-based cell culture
devices (30). For a detailed description of device assembly and operation see Vickerman et al 2008
(41). Interstitial flow is established by means of inserting small reservoirs into the inlet and outlet
154
ports of the channels, and regulating and monitoring the relative heights of the liquid columns (41),
(153), thereby allowing both flow rate and pressure drop to be determined by visual inspection
(details provided in Supplementary Information).
iii.
Imaging
Multiple imaging modalities were used in this study. Phase-contrast (Zeiss Axiovert 200, Carl Zeiss,
Germany), epifluoresence (Nikon TE300, Nikon Instruments Inc., NY), confocal (Olympus FluoView
1000 , Olympus America, Center Valley, PA and Carl Zeiss 510, Axiovert 200M Laser Scanning
Microscope, Carl Zeiss, Germany) and transmission electron microscopy (Model 300, Philips,
Eindhoven, The Netherlands) were used to characterize microenvironmental properties, cellular
morphology, protein localization and phosphorylation state. EC monolayers grow perpendicular to
the imaging plane which readily permits sprout visualization and imaging.
iv.
EC monolayer formation and characterization
In the current study, rat tail collagen type I was used as scaffold. Liquid rat tail collagen type I prepolymer solution (354236, BD Biosciences) was prepared according to product specifications to
obtain a final gel concentration of 2.5 mg/ml. Collagen gels and endothelium monolayers were
formed in the microfluidic devices as previously described (41). Briefly, collagen pre-polymer
solution was microinjected into the gel-cage, sealed with a glass coverslip and polymerized at 37 'C
for 30-35 min. Following collagen polymerization, gels were incubated overnight in monolayer
maintenance medium. Cell suspension was perfused through one microfluidic channel, device tilted
and cells allowed to attach to the collagen gel surface, spread and form a monolayer. Diffusional
permeability, Pa, was measured by monitoring the concentration of fluorescent dextrans introduced
155
via the endothelial-lined
channel under steady-state
conditions, as described in detail in
Supplementary Information. Evaluation of EC expression and localization of adherens and tight
junction proteins was done by immunolabeling with purified rabbit polyclonal VE-Cadherin (ALX-
210-232-C100, ENZO Life Sciences, Uniondale, PA) and mouse monoclonal ZO-1 (339110,
Invitrogen, Chicago, IL) primary antibodies (details below). Labeled monolayers were imaged with
a confocal microscope. In separate samples, further ultrastructural characterization was done using
transmission electron microscopy (TEM) (details provided in Supplementary Information).
v.
In vitro transendothelial flow angiogenesis assay
Basal-to-apical and A-B trans-endothelial flows were generated by imposing a higher liquid column
(typically 80 mm) on the basal or apical surface of the EC monolayer, respectively. The monolayer
was perfused with experimental medium for 24 hours whereas non-flow or static conditions were
used as control. At the end of an experiment, the monolayer was fixed with 4% paraformaldehyde
(PFA), stained with 4', 6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, Atlanta, GA) for nuclei
and phalloidin (Alexa Fluor 488-phalloidin, A12379; Invitrogen, Chicago, IL) for actin and stored for
further processing. Fixed samples were imaged with a confocal microscope (Olympus FluoView
1000) to generate stacks for further quantification. Length, number of sprouts and total number of
cells that invaded in 3D gel were used as simple metric for quantification (155). "True sprouts"
were characterized
as capillary-like structures that contained at least one nucleus. Similar
experiments were conducted to study the effect of pharmacological inhibition (see details below).
vi.
Evaluation of signal transduction at cell-matrix adhesions
Transendothelial
flow direction-related difference in signalling at cell-matrix adhesion was
156
determined by quantifying levels of phosphorylation of FAK on Tyr 397 compared to total FAK.
Antibody pairs were validated (methods described in Supplementary Material; Table 4-1 and
Figure 4-12) before use for quantification. Endothelial cell monolayers were subjected to A-B, B-A
flow or static conditions, perfusion fixed and immunostained with mouse monoclonal anti-FAK
(Clone 4.47, Millipore) and rabbit polyclonal anti-p-FAK Y397 (ab4803, Abcam Inc.) and
appropriate secondary antibodies (detailed below). Samples were also counterstained with DAPI to
identify nuclei. Doubly immunolabeled samples were imaged with a confocal microscope (Carl Zeiss
510, Axiovert 200M Laser Scanning Microscope) equipped with a multi-track channel system, to
generate stacks (at least 40 slices, 1 tm thick) for p-FAK, FAK and DAPI. ImageJ (U.S. National
Institute of Health, Bethesda, MD) (218) was used to quantify florescent intensity for FAK and pFAK Y397. Average p-FAK/FAK ratio was then calculated for each stack. Average ratios for A-B and
B-A flow were normalized by averages from static controls. For quantification, confocal settings
were kept the same for all samples. Reported p-values were obtained from Student's t-test analysis.
vii.
Localization of signaling protein
On the assumption that the morphological changes observed at later times were a consequence of
signaling activated much earlier, shorter term experiments were also conducted to study the effects
of flow direction on signaling protein localization. HMVEC monolayers were perfused in either B-A
or A-B direction for 2 hrs and perfusion fixed while maintaining the initial flow direction. Samples
were subsequently immunolabeled (details below) with VE-cadherin antibody (ALX-210-232-C100,
ENZO Life Sciences, Uniondale, PA) and imaged with a confocal microscope (Carl Zeiss 510,
Axiovert 200M Laser Scanning Microscope). A series of at least 40 optical serial sections were
obtained. Each confocal slice yields a cross-sectional view of the monolayer which is formed
157
perpendicular to the optical imaging plane. Enface views showing protein localization on the
monolayer was generated from projection of image stacks using LSM Image Browser (Carl Zeiss,
Germany) software.
Fluorescent line intensity profiles were obtained for quantitative representation of VE-cadherin
distribution. Cross-sectional images of the monolayer were obtained by confocal imaging and
processed in ImageJ (U.S. National Institute of Health, Bethesda, MD). Fluorescent intensity was
obtained along a horizontal line defined between two adjacent cell-cell junctions.
viii.
Pharmacological Inhibition
Two sets of pharmacological experiments were conducted: (1) long term, to evaluate the effect of
various inhibitors on EC migration or sprouting angiogenesis, and (2) short term, to determine the
effect on flow-induced protein localization. For these experiments 3D sprouting assays were
repeated in the presence of the following small molecules: Genistein (100 [tM; 345834, EMD
Chemicals Inc., Gibbstown, NJ ), a tyrosine kinase inhibitor (219), PP2 (10 [tM; 529576, EMD4
Biosciences, San Diego, CA), a Src inhibitor (220), Y27632 (50 piM; 688000, EMD4 Biosciences), a
Rho-associated kinase (ROCK) inhibitor (221), Heparinase III (15 mU/ml; H8891, Sigma-Aldrich,
Atlanta, GA) and L-NAME (100 liM; N5751, Sigma-Aldrich, Atlanta, GA), a nitric oxide synthase
(NOS) inhibitor (222). Many dynamic processes and molecular interactions in cells are mediated by
protein phosphorylation and Genistein is a widely used tyrosine kinase inhibitor. Rho proteins and
their effectors play essential roles in regulating cytoskeletal events critical for cell migration. The
main downstream RhoA effector, ROCK, controls actomyosin contractility which is important for 3D
cell migration. ROCK inhibition with Y27632 treatment decreases invasive potential of cancer cells
(223) (224). Furthermore, tumor-derived endothelial cells reportedly record constitutively high
158
levels of ROCK (225). A subset of the pharmacological inhibitors was used to further investigate the
effects of inhibition on protein localization. In all studies inhibitors were incubated 2 hours prior to
and throughout flow treatment. As before, monolayers were perfusion fixed with 4% PFA,
immunostained and imaged by confocal microscopy.
ix.
Antibody Labeling and Counterstaining
For antibody labeling, fixed samples were rinsed twice with 1X PBS, permeabilized with 0.1%
Triton-X, rinsed twice with 1X PBS and incubated at room temperature for 2 hrs in blocking buffer 10% BSA in PBS. Following the blocking step, samples were then rinsed once with rinse buffer 0.5% BSA. Primary and secondary antibodies were diluted in rinse buffer. Blocked samples were
incubated with primary antibodies overnight at 4 C. Subsequently, samples were then washed
thoroughly with rinse buffer (3-times, 30 min rinse incubation at room temperature) and incubated
with secondary antibodies in the dark for 2 hrs. Following indirect antibody labeling, samples were
washed thoroughly with 1X PBS and counterstained at the indicated dilution with DAPI (1:1000) to
identify nuclei and phalloidin (1:50) to label F-actin. In some cases, fixed and Triton-X treated
samples were stained with only DAPI and phalloidin thus the blocking, antibody incubation and
associated wash steps were not required.
d. Results
i.
Functional
Formation of a functional EC monolayer
monolayers
with barrier function to fluid and macromolecules
establishment of transendothelial pressure gradients and flow.
are vital for
In order to characterize the
159
functionality of the monolayers cultured in our microfluidic device, both the permeability of a
40kDa dextran (typical size of growth factors) and the expression of junction protein were
evaluated. EC cultured for 24 hrs on type 1 collagen expressed both tight and adherens junction
protein as indicated by immunofluorescence staining for ZO-1 and VE-Cadherin, respectively
(Figure 4-1). The expression profile is similar to that of ECs cultured on glass (supplementary
material Figure 4-8). Ultrastructural analysis of the monolayer from TEM micrographs shows
typical junction morphology of extended membrane adhesions [Figure 4-1C].
Monolayer functionality was also assessed by testing barrier function of monolayers cultured for
24 hours using a fluorescent 40kDa dextran and computing diffusional permeability. An average
Pd
value of 9.7 x 10- m/s was obtained which is comparable to values reported in the literature for
monolayers cultured in vitro (226). Together these results demonstrate that EC formed a functional
barrier on the collagen gels within microfluidic device and is capable of supporting transendothelial
pressure gradients (see also Discussion).
160
C h a nn e|
Gel
apical-to-basal Flow
basal-to-apical Flow
VE-cadherin
Dapi
VE-cadherin
ZO - 1
Dapi
Dapi
Merged
Figure 4-1: Microfluidic-based 3D cell culture system for studies of the effects of
transendothelial flow on sprouting angiogenesis from an endothelial monolayer.
(A) Layout of microfluidic-based 3D cell culture system. Design features include "gel-cage" loaded
with collagen gel (pink), an array of pillars (white squares) for mechanically supporting the gel and
two parallel fluidic channels (denoted apical- and basal at pressures P 1 and P 2, respectively). Single
cell suspension is perfused through the apical channel and an endothelial monolayer grows to
confluence. (Dashed arrows denote direction of flow). Transendothelial flow is established in either
A-B (red) or B-A (green) direction by applying a pressure gradient across the gel-cage. (B) Confocal
images of endothelial monolayer showing coverage on gel and channel surfaces, immunostained
with anti-VE-cadherin (Red) and nuclei (Blue). To confirm that EC form physiologically relevant
161
monolayers the quality of cell-cell junction was evaluated. (B) Localization of tight junction and
adherens junction proteins was analyzed by confocal microscopy (Magnification 40x). En face view
of doubly labeled monolayer (i) VE-Cadherin, (ii) ZO-1 and (iii) merged images indicating
colocalization at cell-cell contacts. Inset shows cross-sectional views. (C) Ultrastructure of cell-cell
junction of monolayer cultured in the device was analyzed by transmission electron microscopy
(TEM). Micrograph of EC cell-cell contact shows typical junction morphology.
ii.
Switching response evoked by transendothelial flow
Prior studies in our lab have demonstrated the use of our microfluidic-based cell culture device for
chemokine or biochemically induced sprouting angiogenesis assays. In the current work, the impact
of a biomechanical stimulus - transendothelial flow - on sprouting angiogenesis is investigated.
Confluent monolayers of HMVEC were cultured on 3-D collagen gels and subjected to A-B or B-A
flow for 24 hrs. B-A flow induced an angiogenic response (Figure 4-2A, right) while monolayers
subjected
to
A-B
flow remained
quiescent
[Figure
4-2A, left].
In
addition
to
sprout
formation/capillary morphogenesis, B-A flow also induced/triggered invadopodia (extensive and
highly branched filopodia-like projections which are actin rich membrane projections) that extend
deep into the 3D gel. Further morphogenesis of these sprout precursors result in the formation of
true sprouts once the nucleus has translocated from the monolayer to the elongating stalk. The
dynamics of this process is captured using time-lapsed video microscopy (see sequential images
from movie in Figure 4-2G for five time points).
In order to determine if these flow-induced effects were unique to microvascular cells, separate
experiments were conducted using HUVECs. Similar to microvascular cells, cells of a macrovascular
origin form sprouts in response to B-A flow (see Supplementary material Figure 4-9).
162
,
MB-A Flow
36
AA-B
Flow
UStatic
13
0
0
Num. of Invasive Cells
*;
0
0
Num. of Sprouts
60
50
40
n HUVEC
30
mHMVEC
20
10
Num. of Invasive Num. of Sprouts
Cells B-A Flow
*
(i) - 30 min
B-A Flow
(ii) -110 min
iii) - 168 min
(iv)- 242 min
(v)- 270 min
Figure 4-2: Transendothelial flow direction modulates EC monolayer phenotype
(Red arrow indicates flow direction). HMVECs are cultured on rat tail collagen type I gels and
subjected to B-A, A-B flow and static control. (A) Fluorescent micrograph of HMVEC monolayer
subjected to A-B (LEFT) and B-A (RIGHT) transendothelial flow (green-actin cytoskeleton, bluenucleus). Scale bar = 20 prm. (B) Quantification of the asymmetric response that is observed. The
average number of cells that have migrated into the collagen gel as well as average number of
sprouts per device is reported. (Note: Some cells also migrate as single cells and are not considered
as sprouts). (C) Comparison of B-A induced sprouting in HUVECs (macrovascular EC) and HMVECs
(microvascular EC).) (D) Confocal image of invadopodia and nucleus deformation during the
formation of true spouts. (G) Time-lapse images from video micrograph during B-A flow induced
sprouting angiogenesis. Endothelial cells project invadopodia in response to B-A flow initiation and
subsequently invade the underlying collagen gel to form sprouts. Scale bar = 20um
163
iii.
Multiple signaling pathways are involved in B-A flow induced sprouting
angiogenesis
In order to identify potential signaling pathways that might be involved in B-A transendothelial
flow induced angiogenesis, we first used a panel of pharmacological inhibitors to target signaling
pathways common to angiogenesis.
Protein phosphorylation is a common post-translational
modification that affects protein function or state of activation. Tyrosine phosphorylation
inhibition, with a natural protein kinase inhibitor, Genistein, blocked invadopodia formation and
invasion (Figure 4-3), suggesting that B-A flow-induced sprouting angiogenesis is tyrosine
phosphorylation-dependent. Treatment with PP2, a specific Src family kinase inhibitor, decreased
EC invasion and sprout formation. Rho/ROCK signaling has been reported to be important for
angiogenic processes including EC migration, survival and permeability (227). Inhibition with
specific ROCK inhibitor Y27632 blocked B-A transendothelial flow induced EC invasion and "true"
sprout formation. Instead, Y27632 treatment resulted in extensive invadopodia devoid of nuclei.
These results are consistent with reports that Y27632 blocks VEGF-mediated angiogenesis in
retinal explants, EC migration and lumen-containing tube-like structures in vitro (227). Our results
demonstrate that invadopodia are necessary for EC invasion and sprout formation but their
appearance does not guarantee the formation of true sprouts. Transendothelial flow has been
reported to upregulate NO production (210). The NO pathway has been extensively studied in the
context of cells exposed to A-B transendothelial flow or transmigration studies. For these reasons,
the effect of NO production using the NO inhibitor LNAME was used to investigate the potential role
of NO production in B-A flow induced angiogenesis. Consistent with this line of reasoning, NO
inhibitor LNAME was also found to reduce B-A mediated EC invasion into collagen gels. The cell
surface glycocalyx has been known to mediate shear response in endothelial (228), (5) and vascular
smooth muscle cells (229). To determine the potential role of heparan sulfate proteoglycans
164
(HSPGs), cells were treated with heparinase III and subjected to B-A flow. Similar to other
inhibitors, digestion of glycocalyx reduced B-A flow mediated EC invasion.
100
90
____
____
___
___
2
___
_
80
70
60
0
_____
____
____40
____
20
10
Contml
Genistein
Y27632
PP2
Hepannase
LNAME
Pharmacological Inhibitors
Ilep 1Irinae
I\
\\l
Figure 4-3: Inhibition of B-A flow induced angiogenesis by small molecule
pharmacological inhibitors.
Multiple potential signaling pathways are involved in B-A flow induced sprouting angiogenesis. EC
monolayers were generated as described in Methods. Monolayers were incubated for 2 hours with
experimental medium containing pharmacological inhibitors prior to and during B-A perfusion.
Monolayers were perfused for a total of 24 hours, fixed with 4% PFA and stained for actin and
nuclei with phalloidin and DAPI respectively. Samples were subsequently imaged and the degree of
inhibition quantified. (A) Panel of representative confocal micrographs of monolayers subjected to
B-A flow in the presence of inhibitors for Rho kinase (Y27633, 50ptM), Tyrosine phosphorylation
(Genistein, 100pM), Src kinase (PP2, 10ptM), Enzyme specific to heparan sulfate - glycocalyx
component (Heparinase III, 15 mU/ml) and Nitric oxide (LNAME, 100tM). (B) Quantification of the
inhibitory effects of pharmacological inhibitors.
165
iv.
B-A flow activation occurs via FAK-mediated signaling
Pharmacological studies provided some clues into signaling pathways that might potentially be
involved in B-A flow induced sprouting angiogenesis. However further investigation was warranted
to identify the initiating event linking flow to signal transduction. To gain deeper insight into the
observed switch response we postulated that ECs adherent to a matrix can sense the direction of
transendothelial flow and that the observed difference in migration and sprouting angiogenesis
could be attributed to differential integrin activation. Differential activation via cell-matrix adhesion
receptors (e.g. integrins, HSPGs) was determined by quantifying the ratio of FAK Y397 to total FAK
for the two flow directions relative to static control. Transendothelial flow direction induced
different levels of FAK activation as measured by the FAK Y397 to FAK ratio [Figure 4-4].
Monolayers subjected to B-A flow consistently recorded significantly higher levels of the FAK
Y397/FAK ratio compared to those exposed to A-B flow (p=0.0003) and compared to controls
(p=0.0005), whereas A-B flow and control conditions were not significantly different. These results
demonstrate that flow direction differentially affects FAK-mediated signaling.
166
Force|
- --
-
A pical-to-Basal
-
Apical
----
surface
'I
Basal
surface
----.-
TEF
HAtt
--
Basal-to-Apical
Integrins
- -
-
Extracellular matrix
(collagen type 1)
Blue: Dapi nuclear stain
Green: pFAKY397
U
150
a)
100
U
a)
s-i
0
zb
so
4-J
0
0
4
8
12
16
Distance across cell (um)
CU
D
1.6
T
0
4-J
1.4
o
0
-
a
1.2
1
2
0.8
>
5
4
3
<z
0.6
1
0
-J
2 hrs
0.4
0.2
6 hrs
Flow Duration
I
St
0
Static
Control
A-B Flow
l
B-A Flow
-j
I
Figure 4-4: B-A flow promotes activation of integrin-mediated adhesion in
endothelial monolayers.
(A) Schematic of B-A flow induced forces at cell-matrix adhesion. EC monolayers were subjected to
2 and6 hrs of static, apical-to-basal and B-A flow conditions. Fixed samples were labeled with antiFAK and p-FAK Y397 antibodies. (B) En face view of monolayer subjected to apical-to-basal (TOP)
and B-A (BOTTOM) transendothelial flows labeled with p-FAK Y397 antibody. (C) Confocal image
showing cross-sectional view of EC monolayer subjected to B-A flow and labeled with p-FAK Y397
167
antibody (LEFT) and fluorescent line intensity profile (RIGHT) from apical to basal cell surface
showing asymmetry of p-FAK Y397 distribution towards cell-matrix interface. To evaluate FAK
activation the ratio of p-FAK to total FAK was obtained for monolayers under static, A-B and B-A
flow treatments. (D) (LEFT) Quantification of pFAK Y397/FAK ratio. Values are reported relative to
static controls. B-A flow samples are significantly different compared to apical-to-basal and static
treatments (p-values 0.0003 and 0.0005 respectively). No significant difference between static and
apical-to-basal treatment (p value 0.44). (RIGHT) Relative levels of p-FAK Y397 at 2 and 6 hours.
v.
Transendothelial flow direction differentially affects VE-cadherin localization
Since signal transmitted via cell-matrix receptor, for example through integrins has been shown to
regulate junctional remodeling (230), the effect of flow direction on endothelial cell-cell junction
protein distribution was assessed. Confluent monolayers were perfused for 2 hrs in either apicalto-basal or B-A flow direction, perfusion fixed and stained for junction protein to capture early
signs of differential protein distribution (Figure 4-5). B-A flow caused delocalization of VEcadherin from cell-cell junctions, while samples subjected to A-B flow retained well-defined
junctional labeling. Compared to static controls there was no noticeable change in VE-cadherin
localization that could be attributed to A-B flow. For the case of A-B flow, fluorescent intensity line
profiles across the cell (Figure 4-5C) show distinct peaks and valleys corresponding to high and
low VE-cadherin expression at cell-cell junctions and in the cytoplasm, respectively. A similar
analysis for monolayers subjected to B-A flow yields, by comparison, a relatively uniform profile.
These results demonstrate that flow direction differentially affects adherens junction remodeling
and confirms the plasticity of cell-cell contacts, an important prerequisite for migration and
angiogenesis.
168
line profile
Cross-sectional View
En Face View
e
e
300
S250
-
200 200
-Basal-
3-
~
Basal-to---
Apical-to
150
100
50
0
1
0.5
Normalized Cell-Cell Junction to
junction
Distance
Figure 4-5: B-A flow induces delocalization of VE-cadherin
Functional monolayers were generated as described in Methods. HMVEC monolayer was subjected
to 2 hrs of static, A-B and B-A flow treatments. (A) Enface view of HMVEC monolayer subjected to
A-B (TOP) or B-A (BOTTOM) transendothelial flow. (B) (TOP) Schematic of procedure for obtaining
intensity profile. (BOTTOM) Fluorescent intensity line profile across cell. RED: B-A flow and BLACK:
A-B. Intensity profile was obtained according to the schematic. B-A flow caused diffuse VE-cadherin
labeling pattern while A-B flow maintains typical VE-cadherin pattern characteristic of well-defined
junctions.
vi.
Transendothelial flow direction differentially affects actin cytoskeletal
organization
Because forces acting on focal adhesions often lead to recruitment or rearrangement of various
intracellular proteins
including
actin
(231),
(232),
(233), we investigated
the effect of
transendothelial flow direction on actin distribution. In addition, since cell-cell junctions are
mechanically coupled to actin cytoskeleton, we suspected that reorganization of VE-cadherin
complexes might also result in remodeling of the actin cytoskeleton. Transendothelial flow
direction exerted a strong influence on the intracellular distribution of F-actin as revealed by
169
confocal sections of phalloidin stained monolayers (Figure 4-6). Monolayers subjected to A-B flow
exhibited dense actin labeling at cell-cell contacts, whereas monolayers exposed to B-A flow
exhibited a more diffuse/disorganized staining pattern. In addition B-A flow induced clustering of
actin near the basal surface from which "hair-like" structures emerged (Figure 4-6C (ii)), at the
tips or filopodia/invadapodia (Figure 4-6C (iii)) and the base of newly forming sprouts (Figure
4-6C (i)).
Cross-sectional View
B-A
Flow
06
apical surface to
basal distribution of
actin
I
04
i
0.2
0
L
0
0.5
1
1
.6-4
0.9
0.8
0
.C
0.7
0.6
0.5
0) 0.4
E
0
z
0.3
0.2
0.1
0
0
0.5
1
Normalized distance across cell
U
3D
View
Figure 4-6: B-A flow direction affects distribution of cortical actin
(A) Confocal section of HMVEC monolayer exposed to A-B (TOP) and B-A (BOTTOM) flow. Samples
were fixed and stained with phalloidin to label actin cytoskeleton. (B) Fluorescent line intensity
profile of actin localization across cell for A-B (BLUE) and B-A (RED) flow directions; (INSET) apical
170
surface to basal surface distribution of actin for monolayer subjected to B-A flow, showing notable
clustering of actin towards basal surface. (C) (LEFT) 3D view of a sprouting monolayer stained with
phalloidin (RIGHT) Magnification of boxed regions highlighting actin localization and clustering
towards basal surfaces.
vii.
VE-Cadherin delocalization is mediated by Src
Src is important in mediating communication between focal adhesions and cell-cell junctions (234),
and has been implicated in angiogenesis in vivo and in vitro (235). Our results show clustering of
phosphorylated Src at locations in monolayers from which sprouts emerged (Figure 4-7A). Src is
reportedly a key mediator in signal transduction between integrin-mediated adhesions and
cadherin mediated cell-cell contacts (234). Furthermore upon integrin activation, Src associates
with FAK in focal complexes and VE-cadherin is known to possess a Src phosphorylation site on Tyr
658 (236), (237). We therefore wanted to determine whether B-A flow induced VE-cadherin
redistribution at cell-cell contact was mediated by Src. Treating monolayers with PP2 prior to and
during the application of B-A flow, we found that this Src-family kinase specific inhibitor was able
to substantially reduce B-A flow induced VE-cadherin delocalization (Figure 4-7B), demonstrating
that B-A flow induced VE-cadherin delocalization at cell-cell junction is mediated by Src.
171
B
En Face View
I
U.
Figure 4-7: Src mediates B-A flow induced remodeling at cell-cell junction.
(A) Reconstruction of confocal images showing 3D views of B-A flow induced sprouting monolayer
stained with anti-p-Src Y416 antibody (GREEN) and DAPI (BLUE). (i-ii) View from basal surface,
showing clustering at membrane projections (white arrow) and tip cell (inset); (iii-iv) View from
apical surface showing p-Src clustering where sprout emerges from monolayer, around lumen
(inset). (B) B-A flow induced VE-cadherin delocalization from junctional complex is mediated by
Src. Images show enface view of monolayer subjected to B-A flow (i) without PP2 (control) (ii) with
Src inhibitor, PP2 treatment and stained for VE-cadherin (RED) and DAPI (BLUE).
e. Discussion and Conclusions
Cells are continuously experiencing external mechanical perturbations within their environment
and mechanical forces play an important role in physiological as well pathological conditions. Of
particular interest is the role that mechanical forces play in the tumor microenvironment. Much of
the work published on endothelial mechanotransduction addresses cardiovascular diseases while
172
tumor vascularization studies typically focus on the chemo-regulation in the recruitment of new
blood vessels. Here we explore flow-mediated regulation of angiogenesis in a tumor environment
and demonstrate that sprouting angiogenesis can be promoted by B-A transendothelial flow
through a process that is mediated by integrin activation. This work confirms that the tumor
microenvironment is mechanically dynamic and that mechanical stress may be an important factor
in the initiation of angiogenic sprouts.
i.
Flow direction acts as an angiogenic switch
We investigated the effect of transendothelial flow direction on sprouting angiogenesis and
observed a switch-like response in which B-A flow promotes angiogenesis but the reverse flow
direction, A-B, did not. Our results therefore suggest that transendothelial flow direction acts as an
angiogenicswitch. B-A flow promotes angiogenesis in micro- and macro- vascular endothelial cells defining an angiogenic ON state while the alternate flow direction - apical-to-basal - maintains the
OFF state. This behavior is consistent with the early observation that neovascularization in tumors
originate from venules and post-capillary venules (214).
ii.
EC activation is due to mechanical stimulation
The observed angiogenic response could be due to either biochemical or biomechanical stimuli. We
first explored the hypothesis that a migratory stimulus could arise due to concentration
polarization of soluble or matrix-bound growth factors (e.g. VEGF) arising from B-A flow in
combination with the low permeability of the endothelial monolayer. Experiments at saturation
levels of VEGF165 and VEGF121 (the non-binding isoform) (results not shown) confirmed that the
influence of flow direction was unaltered.
Furthermore, intuition, confirmed by numerical
simulation, tells us that although B-A flow would lead to higher concentrations at the basal
173
membrane, it would also produce a gradient that would discourage sprouting in that the
concentration would fall in the direction of cell migration. Thus, although we cannot completely
rule out the possibility that biochemical effects contribute to our observations; this evidence
suggests that factors other than biochemical gradients are responsible.
iii.
B-A Flow increases FAK-mediated signaling at cell-matrix adhesions
Based on the low hydraulic permeability of the endothelial monolayer, a simple force balance
dictates that the pressure difference associated with B-A flow would give rise to a net force acting
to lift the cells off the hydrogel surface. Since separation was not observed, presumably the cell
matrix adhesions, for example a1p1 or
a2@1
integrins in the case of type I collagen or HSPGs must be
capable of supporting the force arising from this pressure drop.
Conversely, in A-B flow, the cell
can be supported by direct physical contact with the gel; to the extent that the cell adhesion
receptor-collagen bonds support the load, it would be compressive rather than tensile. It is well
established that integrins transduce signals from the extracellular matrix (238), and that
mechanical force leads to integrin activation (239) resulting from integrin conformational change
(240) or clustering (241) (242). While there is insufficient evidence to irrefutably name the specific
cell-matrix adhesion mechanosensor, based on evidence in the literature and models describing the
signalling events following integrin activation it is tempting to speculate that the observed
difference due to flow direction is integrin-mediated. It is widely accepted that integrin activation
leads subsequently to phosphorylation of Tyr 397 in focal adhesion kinase (FAK).Nevertheless we
cannot completely rule out potential contributions due to non-integrin mediated signalling. Our
observations that B-A flow induces a significant increase in the ratio of p-FAK Y397 to total FAK
when compared to A-B or static conditions therefore supports the role of FAK-mediated signalling
in the cellular response due increased autophosphorylation and potential integrin activation. A
174
recent study by Shi et al., (229) reports on FAK-mediated signaling downstream of HSPGs induced
by interstitial flow for vascular smooth muscle cells (SMCs) suspended in collagen gel. Here they
propose a cooperative interaction between integrin and HSPGs with the latter being the main signal
transducer. A potential role for HSPGs is also plausible in our current study on the basis of the
observed inhibitory effect of heparinase treatment on B-A flow induced cell invasion. Unlike in our
current study, Shi et al., (229) did not observe changes in FAK phosphorylation on Tyr 397, instead
flow mediated difference were observed at Tyr 925. Due to differences in cell type (SMCs vs. EC),
seeding configuration (3D suspended cells vs. monolayer) and other differences in experimental
methods it is difficult to reconcile the differences in FAK-mediated signaling and subsequent
conceptual models. Nevertheless, crosstalk between signaling pathways downstream of individual
mechanosensors is highly probable. Consequently, further investigated is necessary to identify
potential mechanosensor(s) and to determine the involvement of integrins, HSPGs or both in the
current model. We postulate that at the molecular level, B-A flow produces tension at the site of
cell-matrix adhesion and subsequent activation of an undetermined mechanosensor as reflected by
FAK Y397 levels. Src, which translocates to cell-matrix adhesions following integrin activation or
FAK autophosphorylation on Tyr 397, subsequently becomes activated. Following this initial event
Src can further phosphorylates FAK on multiple other sites including Tyr 576 and Tyr 577 in the
activation loop, which promotes optimal FAK activity (236). We propose that this is the major
difference between flow directions that results in the observed angiogenic response.
Besides physically tethering cells to the extracellular matrix, integrins are important regulators of
cell growth, survival and migration during angiogenesis (243). Collagen type I engages both aisi
and a21
integrins; key regulators of VEGF-induced angiogenesis (244). It is conceivable that a
similar angiogenic signaling cascade is triggered due mechanical activation of integrins ligated to
collagen gels during B-A transendothelial flow.
175
iv.
B-A flow induced VE-Cadherin delocalization and angiogenesis are mediated by
Src
VE-Cadherin engages in homophilic interactions between neighboring cells and its presence at cellcell junctions is typical of a quiescent endothelial phenotype. We demonstrate here that B-A flow
promotes delocalization of VE-cadherin from cell-cell junctions, while static or apical-to-basal flow
does not. While this could be a direct consequence of forces acting at the cell-cell junction, we
postulated that this junctional remodeling is due instead to signaling initiated by integrin
activation. VE-cadherin intercellular adhesive activity is regulated by cytoplasmic signaling events
involving catenins, which can be abolished by tyrosine phosphorylation of both catenin and VEcadherin. Disruption of cadherin-catenin complexes causes the destabilization of intercellular
junctions (245). Remodeling of adherens junction can occur due to VE-cadherin internalization,
enzymatic cleavage or kinase phosphorylation. Tyrosine phosphorylation of Y658 or Y731 on VEcadherin prevents the binding of p120- and
p-catenin, respectively (246). VE-cadherin retention at
adherens junctions requires association with p120 (247). B-A flow can therefore initiate a signaling
cascade that causes disruption of adhesion complexes and subsequently delocalization of VEcadherin. Furthermore, VE-cadherin is a substrate for Src (246) and is phosphorylated on Y685
(237). Additionally, phosphorylation of VE-cadherin is inhibited in Src-deficient mice (248). For Src
to function, the Src kinase domain must be unmasked. Src is activated by both growth-factor
initiated signals (249) and those of a mechanical origin (239). VE-Cadherin complex disruption has
been shown to enhance angiogenesis (250). While the detailed events leading to Src activation and
subsequent association with VE-cadherin are beyond the scope of this paper, presumably, Src
activation follows from B-A flow induced integrin activation. We also showed in pharmacological
studies that VE-cadherin delocalization is Src-mediated; inhibition with PP2 blocks VE-cadherin
176
delocalization during B-A flow. Based on these findings we infer that VE-cadherin delocalization
from intercellular junctions was likely due to phosphorylation of VE-cadherin which was mediated
by Src and downstream of integrin activation.
v.
Reorganization of actin cytoskeleton by transendothelial flow
The cytoskeleton is a key player in mechanotransduction (251) and invasive cell migration, which
involves dynamic remodeling of actin cytoskeleton, is essential for angiogenesis. FAK promotes cell
motility by activating regulators of cytoskeletal dynamics (e.g. Rac1) (252) and the intercellular
adhesion complex is mechanically coupled to the actin cytoskeleton. It therefore follows that
reorganization of VE-cadherin would affect distribution of actin at junctions. Direct evidence of
cortical actin remodeling in the endothelial monolayer during sprouting angiogenesis is not
available. Wang and colleagues reported VE-cadherin and actin remodeling when cultured cells
were treated with ECM coated beads (230). Moreover, parallels can be drawn between sprouting
angiogenesis and epithelial-to-mesenchymal transition (EMT) where cells transition from a stable
endothelial (angiogenesis) or epithelial (see (253) for review) monolayer to acquire a migratory
phenotype.
Remodeling of cortical actin to actin stress fibers is a distinct feature of migratory
mesenchymal cells (254). Epithelial cells exhibit cortical actin while transformed mesenchymal
cells do not (255). Moreover, Src kinase activity is required for protrusion dynamics specifically;
phosphorylation of FAK on Tyr 925 is directly mediated by Src and is necessary for the dynamic
regulation of matrix adhesion during cell migration (256).
vi.
Force estimates acting on the monolayer
Numerous studies have investigated the forces necessary to elicit a mechanoresponse, so it is useful
to compare the levels of force acting on the monolayer to those associated with other
177
mechanotransduction phenomena. While it is difficult to determine the pressure drop across the
monolayer as distinct from that across the gel-monolayer combination, based on a comparison of
the flow through the matrix with and without a monolayer present, we estimate that under our
experimental conditions, the monolayer supports a pressure of -15
Pa.
Using this value in
combination with an estimated cell area of -500 [tm2, we arrive at a value of -7 nN supported by
each cell; this level of force is well in excess of that required by endothelial cell activation by shear
stress (-0.5 nN) (8), (257) or by direct application of force by pulling on cell-tethered beads (-1.0
nN) (258). Therefore, in terms of force magnitude alone, it is not surprising that B-A flow activates
mechanotransduction pathways.
vii.
Summary Proposed Model
Collectively, these results suggest that B-A flow triggers the transition of vascular endothelial cells
from a quiescent to a migratory phenotype. We present evidence of a signaling cascade that is
initiated at focal adhesions and subsequently transduced across the plasma membrane prompting
changes at cell-cell junctions and within the actin cytoskeleton. We propose that B-A flow produces
mechanical stress at cell-matrix adhesions, which leads to FAK-mediated signaling, Src-dependent
cell-cell junction remodeling and delocalization of VE-cadherin and cytoskeleton reorganization
which promotes a transition from a quiescent to an invasive/angiogenic phenotype. Multiple proangiogenic signaling cascades follow FAK activation and VE-cadherin remodeling which would
trigger endothelial cell invasion and angiogenesis. Furthermore, Src-induced deregulation at
cadherin junctions in cancer cells require integrin signaling (234), and invadopodia dynamics
during migration requires Src-specific phosphorylation of FAK (256).
In vivo, new blood vessels predominantly emerge from postcapillary venules which is the lower
pressure side of the circulation and would be more susceptible to B-A transendothelial flows in a
178
high pressure tumor environment. Following this line of reasoning, B-A flow could also contribute
to tumor angiogenesis.
It is intriguing that similar results can be captured in our microfluidic
system, which opens the opportunity for future investigation
of angiogenesis where both
biochemical and biomechanical environment is recapitulated.
Acknowledgements
The authors thank Haiyan Gong for TEM imaging and use of their confocal imaging facility. We
thank Choong Kim for help with3D graphics and Carlos Semino for helpful discussion. This work
was supported by funding from NIBIB (EB003805) NSF SCT (CBET-0939511) and NSF-EFRI
(0735997).
179
f. Supplementary Information
i.
Characterization of EC monolayer transport properties
Diffusional permeability (Pd)
In order to characterize the functionality of the monolayers cultured in our microfluidic device EC
barrier function to dextran was evaluated. For diffusive permeability estimation, monolayers were
cultured under static conditions. EC monolayers were formed as described above. Experimental
medium in the apical channel was replaced with medium containing FITC-labeled dextran (Figure
4-10). Fluorescent images were obtained using an inverted microscope (Nikon TE300, Nikon
Instruments Inc., NY) and micrographs acquired with a Hamamatsu ORCA-ER camera (Hamamatsu,
Japan) using OpenLab image acquisition software at least 4 hours following incubation with
dextran. We assume that the transport of dextran across the endothelial monolayer was due to
passive paracellular transport from the apical channel. With this assumption
Pd
was estimated
based on the diffusive flux of dextran across the monolayer:
J=
I
=
Ddextran *
Pd
*
AC)
(Ax gei
(ACEC monolayer)
Where Pd is the diffusional permeability, ACEcmonolayer is the concentration difference across the
monolayer, Ddextran denotes the diffusion coefficient of dextran,
(-)gel
the concentration gradient
180
of dextran in the gel region and
J
the flux from the apical channel across the monolayer. A
fluorescence intensity profile was obtained in ImageJ (U.S. National Institute of Health, Bethesda,
MD) (Figure 4-10B) from which ACEc monolayer and M
coefficient for 40kDa dextran, 4
x 10-11
el were calculated. The diffusion
m 2 /s, was used (41).
Flow rate measurement and Pressure drop Estimation
Fluid velocities in the gel and pressure drop across the EC monolayer were estimated using a bead
tracer method (details below).
First, the Darcy permeability of collagen gel (k) without an EC
monolayer was calculated from Darcy's Equation for flow through a porous matrix.
v
= - k
VP
Ki
Darcy's Equation
where VP denotes the imposed pressure gradient, pt denotes fluid viscosity and v is the fluid
velocity which is obtained experimentally from measurements of tracer particle velocity. As before,
fluid reservoirs were used to impose a pressure differential across the gel region. To visualize fluid
passage through the gel, fluorescent microspheres were added to the upstream reservoirs and
tracer velocities, v determined as a function of a known pressure drop, APtot, across the collagen gel
alone. Separate experiments were repeated with collagen gels covered by an EC monolayer to
obtain new values for v. In these experiments the upstream reservoir was connected to the basal
channel. Here, the collagen gel and the endothelial monolayer were modeled as two resistors in
series, and the total pressure drop, APtot, is the sum of the pressure drops across the gel (APgel)
and EC monolayer (APEc monolayer)181
Vtot = APgel + APEc monolayer
From Darcy's equation, the resistance, Rgel =
IL
-L
where L is the width of the gel in the flow
direction. From this, the pressure drop across the EC monolayer, APEc monolayer, can be estimated.
A
EC monolayer =Atot
-
Vgel with EC monolayerRgel
Bead Tracer Method. FITC-labeled microspheres (200nm in diameter) were used as fluorescent
tracer particles. Fluorescent traces of microspheres flowing through the gel were visualized on a
Nikon TE300 inverted microscope and time-lapse micrographs acquired (Hamamatsu, ORCA-ER,
Japan) and analyzed using OpenLab image acquisition software. Traces were obtained by using a
long exposure during image acquisition which gives rise to streaks indicating bead path through the
gel. While the flow is 3-dimensional, we used particles that remained in focus over their entire
trajectory, thus ensuring that the flow was primarily in a single plane. Subsequently, tracer bead
velocities were calculated from streak lengths and exposure times. Streak lengths were measure in
ImageJ (public-domain image processing software, U.S. National Institute of Health, Bethesda, MD).
Time-lapse Video Microscopy
Time-lapse movies were recorded of endothelial cells during B-A flow induced sprouting
angiogenesis. Endothelial monolayers were formed on collagen gel and B-A flow established as
182
described in METHODS. EC morphogenesis was visualized with an inverted microscope (Zeiss
Axiovert 200, Carl Zeiss, Germany) equipped with an environmental chamber at 37 C and 5% C02.
Images were acquired at 2 min intervals with the AxioCam MRm (Carl Zeiss, Germany) using
AxioVision image acquisition software.
Transmission Electron Microscopy (TEM)
A method for processing microfluidic samples for TEM was developed.
Briefly, glass coverslips
were coated with a thin layer of PDMS. Coated glass slides were used in place of glass coverslips
which are used to seal the PDMS device. All other setup and experimental procedures remain
unchanged. All samples were fixed and processed for TEM in situ by perfusion of fixative and
solutions via microfluidic channels. At the end of experiment, samples were fixed with Karnovsky's
fixative (2.5% glutaraldehyde and 2% paraformaldehyde in phosphate buffer, pH 7.4). Samples
were post-fixed with 1% osmium tetroxide (Electron Microscopy Sciences, Hatfield, PA) and 1.5%
potassium ferrocyanide (Fisher Scientific Company, New Jersey) for 1 hour, dehydrated in cold
ethanol and embedded in Epon-Araldite (Electron Microscopy Sciences, Hatfield, PA). Samples
embedded in plastic were removed from the PDMS device and re-embedded in Epon-Araldite.
Ultrathin sections were cut with an ultramicrotome, counterstained with uranyl acetate (Fisher
Scientific Company, New Jersey) and imaged by TEM (Model 300, Philips Eindhoven, The
Netherlands).
p-FAK Y397 and FAK Antibody Pair Validation
HMVECs were used in all antibody validation experiments. HMVECs were cultured on glass bottom
dishes, fixed with 4% PFA and double stained (see METHODS) with pFAK and FAK antibody pair.
Four different primary antibody pairs were evaluated to find a "compatible" pair (Table S1).
183
Compatibility was determined by visual inspection with selection criteria which were based on the
presence of the classic staining pattern and colocalization of pFAK with FAK in focal adhesion
complexes. Stained samples were imaged (Nikon TE300, Nikon Instruments Inc., NY; Hamamatsu
Orca ER, Japan) and analyzed (OpenLab Software, Improvision, MA). Higher resolution images were
obtained with a confocal microscope (LSM 510 Zeiss Axiovert 200M v4.0). Additionally, a
requirement that the antibody binding epitope did not occur in the region of Tyr 397 was also
imposed. Where available, information regarding binding epitope was obtained from the product
supplier. The selected antibody pair was then subjected to a secondary cross-reaction test. Here
HMVEC samples were processed for immunolabeling according to METHODS, labeled with primary
antibody "A" and incubated with the secondary antibody for primary antibody "B". For e.g. EC
labeled with mouse anti-FAK was incubated with a Goat anti-rabbit secondary.
Supplementary movie: Dynamics during B-A transendothelial flow induced
sprouting from an intact EC monolayer (link not yet available at the time of thesis
submission)
184
Supplementary Figures
VE-cadherin
zo - 1
Dapi
Dapi
Merged
Figure 4-8: Expression profile of adherens (VE-Cadherin, RED) and tight (ZO-1,
GREEN) junctional proteins in HMVEC cultured on glass substrate (static culture)
185
12
-
10
-
SMHUVEC
HMVEC
6 S4~
CD ;' DCD 0 CD
C,
D W
Sprout Length (pm)
Figure 4-9: Sprouting response of microvascular (HMVEC) and macrovascular
(HUVEC) endothelial cells to B-A flow.
(A) Confocal images of HMVEC (LEFT) and HUVEC (RIGHT) (B) Distribution of sprout lengths for
HMVEC and HUVEC monolayers subjected to B-A flow.
186
A
B.
intensity profile
105
dextran
moleculesmolculsCO
apical
*
channel
,
*
6
115
95
5
EC
monolayer
AC
EKmonolayer
85
7
S75
C 65
:oll gen
gel
Dextran ...
gradient in gel
55
-
45
0
50
100
150
200
250
300
350
400
Distance (pm)
Figure 4-10: Estimation of EC Monolayer Diffusional Permeability (Pd).
Barrier function to macromolecule was evaluated using 40kDa fluorescently-tagged
dextran. (A) Schematic of diffusional permeability experimental setup. Medium containing
FITC-tagged dextran is introduced into the apical channel. Intensity profiles are obtained
perpendicular to the monolayer (black dashed-line) (B) A typical concentration profile of
dextran along a lone perpendicular to the monolayer that is used to calculate Pd.
187
A.
microfluic device
B
Streak micrograph
Figure 4-11: Bead Tracer Method for Flow Characterization.
(A) Low resolution fluorescent micrograph of gel cage region during a typical bead tracer
experiments. White arrow indicates fluid flow direction and white spots are fluorescently labeled
microbeads. (B) Higher resolution micrograph taken at long exposure time showing typical
fluorescent streaks for flow velocity calculation.
188
Anti-FAK
Anti-FAK (Y397)
Par
Antibody
Pair #
Anti-FAK
antibodies
Anti- FAK (Y397)
antibodies
1
Mouse monoclonal
ab105917
Rabbit polyclonal
ab4803
2
Rabbit monoclonal
ab76496
Mouse monoclonal
Millipore 05-1140
3
Rabbit monoclonal
ab40794
Mouse monoclonal
Millipore 05-1140
4
Mouse monoclonal
Millipore clone 4.47
Rabbit polyclonal
ab4803
Table 4-1: FAK and p-FAK Y397 Antibody Evaluation.
(LEFT) Antibody pairs screened. (RIGHT) Fluorescent micrographs of HMVEC stained for FAK and
p-FAK Y397 proteins using four different pairs.
189
A
488/ 568 Channel
Merged
Figure 4-12: Negative controls and secondary antibody cross reaction test.
(A) Confocal images of HMVEC monolayer incubated with Alexa Fluor 488 and 568 secondary
antibodies and DAPI (BLUE) (Negative control test). (B) Confocal images of HMVEC monolayer for
antibody cross reaction test. (TOP ROW) Mouse anti-FAK antibody incubated with Rabbit
secondary and (BOTTOM ROW) Rabbit p-FAK Y#(& antibody incubated with Mouse secondary
antibody.
190
g. Appendix IV: Concentration Polarization
Concentration Polarization
The transport of macromolecules across the EC monolayer can be hindered and thus result in an
effective concentration difference across the monolayer. To determine if this phenomenon was
applicable to this study experiments were performed to evaluate potential polarization across the
cultured HMVEC monolayer. EC monolayers were formed as before (METHODS). Experimental
medium was supplemented with FITC-labeled dextran and incubated with monolayer for 2 hours.
This was done to ensure homogeneous dextran concentration (simulating uniform growth factor
conditions prior to perfusion). Perfusate was also experimental medium containing dextran.
Monolayers were perfused in either A-B or B-A flow directions. Fluorescent micrographs were
taken after at least 2 hours of perfusion. ImageJ was used for micrograph analysis. In ImageJ
fluorescent line intensity were obtained for lines drawn perpendicular to the EC monolayer
extending from the apical channel across the endothelial monolayer and through the collagen gel.
Profiles were visually inspected for any noticeable intensity/concentration changes across the EC
monolayer.
Concentration Polarization: Transport Model in FEMLAB. Furthermore, a simplified model was
formulated to describe the transport of a freely diffusible solute across the EC monolayer using
commercial finite element software, FEMLAB. In FEMLAB the mass and momentum transport
problem was solved using Brinkman Equations and Convection and Diffusion Chemical Engineering
Modules.
191
Model Geometry. The model geometry and dimensions for the specific microfluidic device was
generated in COMSOL (Figure 4-13).
U
.;'AA
-44lUp
W
ke
k
QW
n
-U
t
PDMS POST
EC
monolayer
qL
Gel
channel
I-
PDMS POST
Figure 4-13: Device geometry in COMSOL
Microfluidic device geometry (showing region of interest, "gel-cage" and endothelial monolayer)
generated in COMSOL for a mass and momentum Finite Element Model of flow across "gel cage"
with an endothelial monolayer on one surface of the gel. (A) Snapshot of COMSOL window showing
geometry in mesh mode. (B) Boxed region in (A) at higher magnification.
Boundary Conditions. Pressure boundary conditions were specified at the inlet and outlet ports.
For example, in simulations for basal-to-apical transendothelial flow, the pressure was specified at
the basal channel inlets (Po) and apical channel outlets (P1) ports where Po is greater than P1.
Concentration boundary conditions were specified at the inlet ports.
Gel and Endothelial Monolayer Domain. For the purpose of this model, the endothelial monolayer
was treated as a thin layer with hydraulic permeability (Lp) and diffusive permeability (Pd) which
were used to calculate model constants. The gel region was modeled as a porous matrix with an
192
associated Darcy permeability
(Kgei)
which was
determined experimentally. The diffusion
coefficient (D) for 40kDa dextran in collagen matrix was also obtained experimentally.
Drag reflection coefficient (sigma, a). In COMSOL, the selectivity of the endothelial monolayer based
on molecular radius was defined by multiplying the x-velocity component in the convection
diffusion module by (1-a). Values for a range from 0 to 1, where a value of 1 implies that the
membrane is impermeable and 0 implies that the solute crosses as easily as water (259). Typical
published experimental values and the model prediction for a based on molecular radius is shown
in Figure 4-14. For 40kDa dextran, a predicted average value for a of 0.4 is obtained (260).
1
0.9
C
0.8
a)
C.)
4-
ci)
0
C.)
C
0
0
+
0.7
0
0.6
a) 0.5
U)
0)
~0
'7
0.37 - 0.43
0.4
Eqn (18) Bassingthwaighte r = 7.9 nm
U)
--
0.3
0
U)
-<)
0.2
-
l
+
0.1
0
+
_
I
0
1
- r
=7 nm
Expt. Data Kellen & Bassingthwaighte 20 03
70kDa dextran - BAEC, Demaio et. al 20 04
Albumin - PAEC, Suttorp et. al 1988
7OkDa dextran - BREC, Demaio et. al 2C0
04
Eqn (18) 4OkDa dextran
. . . I I I I i I I I I I I II I I I I I I I I I I I I
2
3
4
5
6
7
8
r molecular radius nm
193
Figure 4-14: Solute drag reflection coefficient
Graph (solid line and experimental data points) reproduced from (260), (261). Curve for rp = 7 nm
was generated based on equation 18 in (260), single pore analysis. Additional data points were
obtained from experimental results published in (262) and (263). Estimated values of a for 40kDa
dextran were obtained based on rs - 2.5 nm and rp = 7 nm and 7.9 nm.
This value was not experimentally validated, but seems reasonable based on values reported (262)
for larger molecular radii and more mature in vitro EC monolayers. It is reasonable to expect that
monolayers cultured for shorter times with less mature junctions would be "leakier" and have a
lower value for a (as shown in Figure 4-14, corresponding to a shift downwards for larger pore
size, rp). Nonetheless, simulations were done with a range of values for a, to quantitatively
determine any correlation with polarization distance and sieving coefficient (ratio
Cbasal/Capical
where C is solute concentration).
194
10
Pecletmax
= UmaxPd
o B-A Flow sigma 0.95
* B-A Flow sigma 0.8
+ B-A Flow sigma 0.5
a=0.8
a
* A-B Flow sigma 0.95
* B-A Flow sigma 0.8 Kgel low
B-A Flow sigma 0.8 Dgel high
a = 0.5
0
0
0
0.01
0.001
0.0001
*
++
:+
-+
+
++
0.1
1
' 10
100
1000
Peclet #
Peend 0=(1-sigma)*(UfijPa)
U.I
Figure 4-15: Predicted Relationship between EC Monolayer Sieving Coefficient and
Peclet Number
Result from FEMLAB simulations showing the predicted experimental range of the sieving
coefficient (shaded region). Vertical dashed line indicates maximum Peclet number based on
experimental estimates of monolayer diffusional permeability (Pd) and maximum filtration velocity
(Umax).
The main objective of these "in silico" experiments was to examine the operating range for flow
experiments with regards to concentration difference across the EC monolayer, here represented
by sieving coefficient
(Cbasal/Capical)
monolayer parameters,
function of
Peendo
Peendo =
The Peclet number (Peendo) was defined in terms of endothelial
(1-a)*(UfiIt/Pd). The predicted correlation for sieving coefficient as a
for different values of a is summarized in Figure 4-15. Based on a a of 0.4 and
195
maximum
Peendo)
~ 6, the predicted operating range was obtained (green boxed region) Figure
4-15.
400
Pd= 0.5
Ca
CaCU
350
300
0
Pd.:
-+-1.OOE-08
-U-1.OOE-06
1.OOE-04
-+"-1.00E-05
1.OOE-07
250
200
CO
150
Ca
100
0
50
0
0.001
. I . II- !
0.01
.
. . . ... i
0.1
I I . .- -i
. . . . . ... i
1
10
. . . . . ... i
100
I . I . . .. ,
1000
Pecletendo
Peendo = (1-(Y) Ufilt/Pd
Figure 4-16: Predicted Relationship between Polarization Distance and Pecletendo
Result from FEMLAB simulation showing concentration polarization boundary layer thickness as a
function of Peclet number based on endothelial monolayer Pendo during B-A flow for a = 0.5 and
indicated Pd values.
The polarization distance was also determined as a function of Peendo) and the results summarized
in Figure 4-16. As expected, the concentration polarization distance decreases as Peendo increases.
196
Results from concentration polarization experiments however did not show polarization of the EC
monolayer due to flow and suggest that there is no selectivity to 40 kDa dextran. The discrepancy
between model prediction and experimental results may be due to imprecise model parameters.
For example estimates for a were not validated using the current experimental setup. However,
intuition, confirmed by numerical simulation, tells us that although B-A flow would lead to higher
concentrations at the basal membrane (i.e. for a "selective" monolayer), it would also produce a
gradient that would discourage sprouting in that the concentration would fall in the direction of cell
migration Figure 4-17.
1.30E-06
monolayer
1.20E-06
=0.5
Peendo
S1.10E-06
ed
--
0.775
2.57
0--3.375
0
1.OOE-06
U 9.OOE-07
0
8.OOE-07
7.OOE-07
6.OOE-07
-
0.0086
'
'
0.0087
0.0088
0.0089
0.009
0.0091
0.0092
0.0093
Distance (m)
Figure 4-17: FEM solution for solute concentration as a function of distance
197
Concentration polarization results obtained from FEMLAB simulations of basal-to-apical flow with
the indicated endothelial monolayer properties (a = 0.5 and Peend 0.775, 2.57 and 3.375). (Inset) A
schematic that shows a section of the endothelial monolayer, gel region and dashed-line for
evaluating solute concentration as a function of distance.
198
h. Appendix: Immunolabeling Protocol for Microfluidic Samples
1.
Fix samples with 4% PFA - 30 minutes.
2.
Wash thoroughly with 1X PBS (w/o Ca
3.
Day 1: Primary Antibody
a.
2
+ or
Mg2 +).
Permeabilize sample with 0.1% Triton-X for 10 minutes (handle on 4 devices at a
time - perfuse 60ul through each port)
b.
Wash with 1X PBS - twice
c.
Block with 10% BSA (100mg/ml e.g. 1.Lg BSA plus 11ml IX PBS)
i. Perfuse channels twice with blocking solution
ii.
Incubate for 2 hours at room temperature
d.
Wash once with 0.5% BSA (e.g. 0.5 ml of 10% BSA plus 9.5 ml IX PBS)
e.
Antibody Incubation
i. Recommended dilution for primary antibody is 1:100
ii.
Prepare antibody in 0.5% BSA
iii.
For multiple labeling, prepare cocktail of different antibodies raised in
different species (e.g. Rabbit VE Cadherin, Mouse PECAM, Goat Vinculin etc.)
iv.
Perfuse channels once with 60 microliter of antibody cocktail and add
extra at ports to prevent samples from drying out
v. Incubate for at least 16 hours at 4 C
4.
Day 2: Secondary Antibody
a.
Wash samples thoroughly with 0.5% BSA
i. Perfuse channel a couple times between each wash step
ii. Wash 3X with 30 minute incubation at room temperature
b.
Secondary antibody incubation
i. Recommended dilution for secondary antibody is 1:200
ii.
Prepare antibody in 0.5% BSA (in the dark)
iii.
For multiple labeling, prepare cocktail of different species and color
conjugated secondaries (e.g. donkey anti Rabbit-Alexa 488, anti Mouse-Alexa
568 etc.)
iv. Incubate for 2 hours at room temperature (cover samples with Al foil)
199
c.
Wash samples thoroughly with 0.5% BSA
i. Perfuse channel a couple times between each wash step
ii. Wash 3X with 30 minute incubation at room temperature (final wash step
use 1X PBS)
d.
Final Step: Counter stain
i. Prepare Dapi (1:1000) and Phalloidin (1:50) in PBS
ii. Incubate for 1 hour at room temperature
200
Chapter 5: Concluding Remarks and Future Directions
Ross G. Harrison, one of the pioneers of cell culture models wrote in 1907, "the immediate objective
of the following experiment was to obtain a method by which the end of a growing nerve could be
brought under direct observation while alive, in order that a correct conception might be had
regardingwhat takes place as the nerve fiber extends during embryonic development from the nerve
center out to the periphery." Like many scientific pursuits his work was motivated by an existing
debate - regarding the growth and development of the nervous system - and would only be settled
by irrefutable experimental findings. A glimpse of the process would reveal clues to underlying
mechanisms which could prove or disprove his theory.
The potential for new scientific
breakthrough was quickly realized and in the years to follow isolated cells and tissue were used as
model systems to investigate processes relating to health and disease. The ultimate challenge lies in
replicating the native cell/tissue environment ex vivo. The work in this thesis shares a similar
motivation and sought to contribute to the growing field of in vitro model development for scientific
advancement and demonstrate their implementation in studies relating to vascular tissue
engineering, vascular mechanobiology, ophthalmology and cancer research.
a. Summary of Thesis Contributions
This thesis addresses the need for physiologically relevant in vitro models that replicates critical
components
of the in vivo microenvironment. It documents the development of a novel
microfluidic- based 3D cell culture platform and demonstrates its subsequent use in different
biological studies.
201
Novel microfluidic-based 3D cell culture technology (subject of CHAPTER TWO)
In vitro models offer enormous potential for gaining new insights into complex in vivo
phenomena. One of the central contributions of this thesis is the development of novel microfluidicbased 3D cell culture platforms which was demonstrated to be an important tool for different
biological studies.
We have developed a novel microfluidic-based system that for the first time is capable of
mimicking many of these complexities on a single in vitro platform. The integration of 3D scaffolds
(biologically-derived or synthetic) and microfluidic networks permits the control of the fluidic
environment with the benefit of a more in vivo -like 3D micro-architecture and high-quality
imaging capabilities for dynamic studies. Due to the unique construction of the system, fluid
composition and pressure can be controlled on the two sides of a central gel region, making
possible a platform that is able to control (1) surface shear stress, (2) interstitial flow through the
matrix (3)
gradients in chemoattractants, (4) properties
of cell culture scaffold and (5)
simultaneously monitor cells in real-time.
With this platform, biochemically induced angiogenic sprouting from an endothelial
monolayer formed on a gel such as collagen and peptide was demonstrated. We were able to
visualize in real-time growing vascular network including tip cell dynamics, cell division and
migration. The ability to follow and capture these dynamic processes in real time provides insights
not possible through single images taken on the order of hours or days apart. Furthermore, of
particular relevance to tissue engineering community, we demonstrated that endothelial cells when
cultured for several days can assemble into vascular networks with open, perfusable lumen
structures.
202
The vascularization potential of a class of synthetic peptide hydrogels was also examined
within the microfluidic-based 3D cell-culture device, (APPENDIX II). Experiments with synthetic
peptides (RAD16I and RAD16II) for the endothelial sprouting assay demonstrated that the
angiogenic response, in particular, the degree of invasion and sprout formation, was very low
compared to the widely used rat tail type I collagen gels. However, functionalized peptides PRG
(RGD binding motif) and KLT (mimics VEGF helix region, activate VEGF receptor) when added to
RAD161 promoted HMVEC invasion and the formation of capillary-like structures.
To our knowledge, this thesis is the first to demonstrate the creation of a vascular network
and subsequent perfusion in a microfluidics-based 3D cell culture system. This work is being
continued as a part of the newly established Science and Technology Center at MIT (Director, Dr.
Roger Kamm) to generate biomachines. Given the large interest in understanding how to modulate
vascularization, the results obtained from this work can potentially be of interest for applications in
tissue engineering where pre-vascularized scaffolds may be essential. Alternatively, vascular
networks could be generated and then used as a test bed to study interactions with different cells
types. For example, the interaction of an in vitro generated endothelium with circulating cancer or
immune cells.
The physiological relevance of 3D cultures and the numerous advantages offered by
miniaturization have opened the doors for several biological studies with application to a wide
variety of diseases. This platform technology has been used to launch several new studies in cell
migration, interactions between endothelial cells and tumor cells, axonal guidance, and endothelial
cell interactions with mesenchymal stem cells.
The established platform and subsequent
modifications have formed the basis for newly funded programs from, NCI ICMAT, GRI, NUS
Innovation award, and much fruitful collaboration (APPENDIX I).
203
Model for Aqueous Humor Outflow (subject of CHAPTER THREE)
We developed a new in vitro cell culture-based model and conducted proof-of-principle
experiments which demonstrate its applicability for investigating AH outflow dynamics. A key
feature of the in vivo inner wall endothelium, giant vacuoles, was replicated in our system and its
dynamics could be monitored in real-time. Furthermore the mechanism of aqueous humor outflow
dynamics across the inner wall remain largely uncharacterized however with our system it is now
possible to label fluid path as demonstrated by hydrodynamic filtration pattern experiments.
Consequently, for the first time fluid flow can be visualized in a model of the inner wall endothelium
in real-time thus providing a unique tool for future investigations.
Furthermore, unlike previously reported models, here hSCEC are cultured on 3D hydrogels
instead of 2D filter membranes for perfusion. It is widely accepted that in vitro culture
dimensionality is important and that 3D cultures narrows the gap between in vitro models and live
tissue (206), (124). Of particular importance is the juxtacanalicular region that lies adjacent to the
IW endothelium can only be modeled in cultures with 3D capability. Consequently, this aspect of
the trabecular outflow pathway cannot be readily addressed with current models (191), (192).
Moreover, independent studies highlight the various synergistic interactions that occur with the
trabecular outflow tissue that ultimately impacts the regulation of AH outflow dynamics (177),
(180), (178), (194), (193), (195), (196), (197), (198), (199). Following this line of reasoning, our
microfluidic-based platform sets the stage for future investigations where an integrative approach
can be taken to investigate the physiology and pathophysiology of trabecular outflow tissue.
Consequently, this study is a non-trivial improvement to currently available cell-based perfusion
models (191), (192).
204
Moreover, new and effective strategies to modulate intraocular pressure and potentially
treat POAG are needed. This platform presents a promising tool for investigating the effects of
pharmacological drugs on cellular mechanisms which are responsible for modulating aqueous
humor outflow facility.
Mechanical regulation of angiogenesis (subject of CHAPTER FOUR)
Many
cellular
processes
are
regulated
by
mechanical
forces
exerted
via
their
microenvironment. In angiogenesis, soluble pro- and anti-angiogenic molecules are potent
regulators. The in vivo scenario, however, encompasses the integrative effects of a population of
endothelial cells influenced by both chemical and mechanical cues as they are induced to form
sprouts. Here, with the aid of microfluidic technology, we present a novel mechanism for flowmediated mechanical regulation of angiogenesis by transendothelial fluid flow. We demonstrate
that flow direction is sufficient to define an angiogenic ON or OFF state. The balance is tipped by
forces generated at mechano-sensitive integrin-mediated adhesions. These results provide one
explanation for the bias towards angiogenesis occurring from the venous side of the circulation.
This thesis also contributes to the growing knowledge of the basic mechanobiology of
endothelial cells with potential implication in tumor angiogenesis. The biochemical regulation of
tumor angiogenesis is the most common school of thought. This work highlights that the underlying
mechanism is more complex and that endothelial cells are continuously integrating both
biochemical and biophysical cues. The synergy that exists increases the complexity for studying
diseases (e. g. cancer) ex vivo and emphasizes the need for flexible multiparameter controllable
platform without which progress towards successful therapies would be greatly hampered.
205
b. Future Research Directions
Mechanobiology of basal-to-apical flow induced angiogenesis
A proposed model for basal-to-apical flow induced angiogenesis was presented based on evidence
from protein localization, phosphorylation and pharmacological studies. The following future
studies would provide a deeper understanding of underlying molecular mechanisms.
1.
Identify transendothelial flow induced phosphorylation site on VE-Cadherin and additional
sites on FAK using anti-bodies and subsequent protein localization analysis.
2.
Explore signaling directly downstream of FAK and VE-Cadherin that would promote a
migratory phenotype, for example P120.
3.
FRET technology is widely used for mechanobiology studies. By utilizing currently available
FRET biosensors would provide spatial and temporal
visualization
of intracellular
molecular interactions as it relates to mechanotransduction process involved in basal-toapical flow induced angiogenesis. For example success with FRET-based Src/FAK, Cadherin,
Rac1 GTPase biosensors in live cells has been demonstrated (264), (265), (266).
4.
Cells are equipped with an array mechanosensors with potential for cross-talk and
cooperative interactions. Consequently, the potential role of other mechanosensors in
basal-to-apical
flow induced angiogenesis for example heparin sulfate proteoglycans
(HSPGs) warrants further investigation. Recently, cell surface glycocalyx HSPGs in concert
with integrin-mediated cell-matrix adhesions has been shown to mediate flow induced
signaling of smooth muscle cells (SMC) suspended in 3D collagen gels (229).
Aqueous humor outflow model
206
A new model and subsequent proof-of-concept experiments demonstrating formation of giant
vacuole and labeling fluid path with bead tracers for reconstructing hydrodynamic filtration
pattern were present. Work provides framework for model refinement and future experiments,
including:
1.
Quantitative studies of giant vacuole formation by human Schlemm's canal endothelial cells
from normal and glaucomatous eyes.
2.
Utilize transmission electron microscopy (TEM) to analyze details of morphological changes
involved in giant vacuole and pore formation from microfluidic samples.
3.
Explore the potential effects of paracrine signaling between hSCEC and trabecular
meshwork cells (isolated from normal or glaucomatous eyes) by designing co-cultures
studies. Here hSCEC can be cultured as before with TM cells suspend in the 3D gel.
Interesting questions to explore: (a) does co-culture affect hSCEC monolayer quiescence?
(a) Quantitatively, does co-culture (w/ normal or glaucomatous TM) affect giant vacuole
formation by hSCEC?
Vascularization Studies
Microfluidic-based angiogenesis/vascularization assays are beneficial because compared to macroscale studies only a small amount of reagent is needed. Studies with self-assembling RAD16-I
peptide gels and functionalized peptides PRG (RGD binding motifO/ KLT (mimics VEGF helix region,
activate VEGF receptor) demonstrated an improvement in angiogenic potential for gel containing
the functionalized peptides. Future studies with the MMP-2 cleavable variant may reveal even
greater potential for generating vascularized
synthetic
scaffolds. Furthermore, subsequent
strategies to stabilize networks should be explored. In a separate preliminary study, 1OT
smooth
207
muscle cell precursor seemed to stabilize vascular structures generated by HMVEC in collagen gels.
This co-culture or alternatively, media conditioned by 10T 1/2 cells may be explored further.
Microfluidic Platform
"Make everything as simple as possible, but not simpler." - Albert Einstein. Ease of use and
applicability as a tool for conducting routine biological experiments dictate that the complexity of
these devices must be kept to a minimum. Nevertheless, the microfluidic system presented here
could be further enhanced by:
1.
Implementing additional strategies to increase throughput. For example the use of
photocrosslinkable polymers, patterned masks and on-chip photopolymerization strategies
(267) could make gel loading process high-throughput. Furthermore, this strategy could
potentially provide an option for preloading gels in commercially manufactured devices.
However, issue regarding feasibility and shelf-life should be explored further.
2.
Expanding device functionality, for example hypoxia studies with capability for control and
generation and detection of simple or complex gradients of gas (e.g. oxygen) (268). With
these expanded capabilities, further investigation into the competing effects (as it may exist
in native tissues), for example of flow and gradients in oxygen and growth factors on cell
function.
3.
Upgrading to multiplexed systems including for example on-chip detection units or modules
for quantitative analysis (ideally with spatial and temporal resolution) of cell secreted
factors. Approaches using antibodies and optical detection methods developed by Love and
colleagues (269) have reported success. An alternative strategy including the use of singlewalled carbon nanotubes as biosensors has been described (270).
4.
Integration with a computerized flow or pressure control and monitoring platform.
208
c. Closing Thoughts
"No man ever steps in the same river twice, for it's not the same riverand he's not the same man."
-Heraclitus
The journey, a cycle, the pursuit of knowledge... Robert Browning describes it perfectly "how sad
and bad and mad it was - but then, how it was sweet."
209
Chapter 6: Works Cited
1. Feletou, Michel. The Endothelium. Part 1: Multiple Functions of the Endothelial Cells-Focus on
Endothelium-Derived Vasoactive Mediators. San Rafael (CA) : Morgan & Claypool Life Sciences
Publishers, 2011.
2. Schlemm's canal endothelia, lymphatic, or blood vasculature. Ramos, R. F., et al., et al. 4, 2007, J
Glaucoma, Vol. 16, pp. 391-405.
3. Specification of arterial, venous, and lymphatic endothelial cells during embryonic development.
Kume, Tsutomu. 5, 2010, Histol Histopathol., Vol. 25, pp. 637-646.
4. Microfluidic Technology in Vascular Research. van der Meer, A. D., et al., et al. 2009, Journal of
Biomedicine and Biotechnology, Vol. 2009, pp. 1-10.
5. Mechanotransduction gone awry. Jaalouk, Diana E. and Lammerding, Jan. 2009, Nature Reviews
Molecular Cell Biology, Vol. 10, pp. 63-73.
6. The Micromachinery of Mechanotransduction in Hair Cells. Vollrath, Melissa A., Kwan, Kelvin Y. and
Corey, David P. 2007, Annual Review of Neuroscience, Vol. 30, pp. 339-365.
7. Mechanotransduction in cardiac myocytes. Lammerding, J, Kamm, R. D. and Lee, R. T. 2004, Ann. NY
Acad. Sci., Vol. 1015, pp. 53-70.
8. Flow-Mediated Endotheilal Mechanotransduction. Davies, P.F. 3, 1995, Physiol Rev, Vol. 75, pp. 519560.
9. Molecular basis of the effect of shear stress on vascular endothelial cells. Li, Y. S., Haga, J. H. and
Chien, S. 2005, J. Biomech, Vol. 38, pp. 1949-1971.
10. Mechanotransduction and endothelial cell homeostasis: the wisdom of the cell. Chien, Shu. 3, 2007,
AJP-Heart, Vol. 292, pp. H1209-H1224.
11. Fluid Flow Mechanotransduction in Vascular Smooth Muscle Cells and Fibroblasts. Shi, Zhong-Dong
and Tarbell, John M. 6, 2011, Ann Biomed Eng., Vol. 39, pp. 1608-1619.
12. Cell tension, matrix mechanics, and cancer development. Huang, Sui and |ngber, Donald E. 3, 2005,
Cancer Cell., Vol. 8, pp. 175-176.
13. Mechanotransduction in development: a growing rolefor contractility. Wozniak, Michele A. and
Chen, Christopher S. 2009, Nature Reviews Molecular Cell Biology, Vol. 10, pp. 34-42.
210
14. Mechanotransduction in vascular physiology and atherogenesis. Hahn, Cornelia and Schwartz,
Martin A. 1, 2009, Nat Rev Mol Cell Biol., Vol. 10, pp. 53-62.
15. Mechanobiology and diseases of mechanotransduction. Ingber, Donald E. 2003, Ann Med , Vol. 35,
pp. 1-14.
16. Laminar Shear Stress Mechanism by Which Endothelial Cells Tranduce and Atheroprotective Force.
Traub, Oren and Berk, Bradford C. 1998, Arteriosclerosis, Thrombosis and Vascular Biology, Vol. 18, pp.
677-685.
17. Monocyte recruitment to endothelial cells in response to oscillatory shear stress. Hsiai, TK, et al., et
al. 2003, FASEB J, Vol. 17, pp. 1648-1657.
18. Biomechanical Regulation of Endothelium-dependent Events Criticalfor Adaptive Remodeling. Mack,
Peter J., et al., et al. 13, 2009, Journal of Biological Chemistry, Vol. 284.
19. Interactions of mechanotransduction pathways. Labrador, V, et al., et al. 2003, Biorheology, Vol. 40,
pp. 47-52.
20. Deaths: Preliminary Datafor 2009. Kochanek, Kenneth D., et al., et al. 4, 2011, National Vital
Statistics Reports, Vol. 59, pp. 1-51.
21. Oxidative stress and endothelial dysfunction in cerebrovascular disease. Chrissobolis, Sophocles, et
al., et al. 16, 2011, Front Biosci, Vol. 1, pp. 1733-1745.
22. Endothelial Function: From Vascular Biology to Clinical Applications. Behrendt, Dominik and Ganz,
Peter. 2002, Am J Cardiol, Vol. 90, pp. 40L-48L.
23. Metastatic Cancer Cell. Bacac, Marina and Stamenkovic, Ivan. 2008, Annu. Rev. Pathol. Mech. Dis.,
Vol. 3, pp. 221-247.
24. Role of the Endothelium during Tumor Cell Metastasis: Is the Endothelium a Barrier or a Promoter for
Cell Invasion and Metastasis? Mierke, Claudia Tanja. 2008, Journal of Biophysics, Vol. 2008.
25. Arterial thrombosis - insidious, unpredictable and deadly. Jackson, Shaun P. 11, 2011, Nature
Medicine, Vol. 17, pp. 1423-1436.
26. Alexis Carrel And The Mysticism of Tissue Culture. Witkowski, J.A. 1979, Medical History, Vol. 23, pp.
279-296.
27. A gas chromatographic air analyzer fabricated on a silicon wafer. Terry, S.C., Jernab, J.H. and Angell,
J.B. 12, 1979, IEEE Trans. Electron. Devices ,Vol. 26, pp. 1880-1886.
211
28. Inkjet printing nozzle arrays etched in silicon. Bassous, E, Taub, H.H. and Kuhn, L. 1977, Appl. Phys.
Lett., Vol. 31, pp. 135-137.
29. Manipulation and Flow of Biological Fluids in Straight Channels Micromachined in Silicon. Wilding,
Peter, et al., et al. 1994, Clin. Chem, Vol. 40, pp. 43-47.
30. Rapid Prototyping of Microfluidic Systems in Poly(dimethylsiloxane). Duffy, David C., et al., et al. 23,
1998, Anal. Chem., Vol. 70, pp. 4974-4984.
31. Fabrication of microfluidic sytem in poly(dimethylsiloxane). McDonald, J Cooper, et al., et al. 2000,
Electrophoresis, Vol. 21, pp. 27-40.
32. Microfabricated devices in biotechnology and biochemical processing. Chovan, T and Guttman, A. 3,
2002, Trends Biotechnol., Vol. 20, pp. 116-122.
33. Microfluidic chips for clinical andforensic analysis. Verpoorte, E. 5, 2002, Electrophoresis., Vol. 23,
pp. 677-712.
34. Microfluidic combinatorial chemistry. Watts, P and Haswell, SJ. 3, 2003, Curr Opin Chem Biol, Vol. 7,
pp. 380-387.
35. Microfluidics-based system biology. Breslauer, DN, Lee, PJ and Lee, LP. 2, 2006, Mol Biosyst, Vol. 2,
pp. 97-112.
36. Microfabrication and microfluidics for tissue engineering: sate of the art and future opportunities.
Andersson, H and van den Berg, A. 2, 2004, Lab Chip, Vol. 4, pp. 98-103.
37. Microscale technology for tissue engineering and biology. Khademhosseini, A, et al., et al. 8, 2006,
Proc Natl Acad Sci USA, Vol. 103, pp. 2480-2487.
38. Integration of cell culture and microfabrication technology. Park, TH and Shuler, ML. 2, 2003,
Biotechnol Prog., Vol. 19, pp. 243-253.
39. Applications of biochips:from diagnostics to personalized medicine. Jain, KK. 3, 2004, Curr Opin Drug
Discov Devel, Vol. 7, pp. 285-289.
40. Application of sexed semen technology to in vitro embryo production in cattle. Wheeler, MB, et al.,
et al. 1, 2006, Theriogenology, Vol. 65, pp. 219-227.
41. Design, fabrication and implementation of a novel multi-parameter control microfluidic platform for
three-dimensional cell culture and real-time imaging. Vickerman, Vernella, et al., et al. s.l. : Lab on a
Chip, 2008.
212
42. The Role of Therapeutic Angiogenesis in Tissue Repair and Regeneration. Li, William W, et al., et al.
2005, Adv Skin Wound Care, Vol. 18, pp. 491-500.
43. Angiogenesis in life, disease and medicine. Carmeliet, Peter. 2005, Nature, Vol. 438, pp. 932-936.
44. Angiogenesis in health and disease. Carmeliet, Peter. s.l. : Nature Medicine, 2003.
45. Therapeutic angiogenesis with intramuscular NV1FGF improves amputation-free survival in patients
with critical limb ischemia. Nikol, S, et al., et al. 2008, Mol Ther., Vol. 16, pp. 972-978.
46. Relationship between vasculogenesis, angiogenesis and haemopoiesis during avian ontogeny.
Pardanaud, L, Yassine, F and Dieterlen-Lievre, F. 1989, Development, Vol. 105, pp. 473-485.
47. Mechanisms of adaptive angiogenesis to tissue hypoxia. Fong, Guo-Hua. 2008, Angiogenesis, Vol. 11,
pp. 121-140.
48. The role of hypoxia-inducible factors in tumorigenesis. Rankin, EB and Giaccia, AJ. 2008, Cell Death
Differ, Vol. 15, pp. 678-685.
49. Analysis of the role of chemokines in angiogenesis. Bernardini, G, et al., et al. 2003.
50. Hypoxia-Mediated Biological Control. Cassavaugh, Jessica and Lounsbury, Karen. 2011, J Cell
Biochem, Vol. 112, pp. 735-744.
51. Regulation of angiogenesis by hypoxia and hypoxia-inducible factors. Hickey, MM and Simon, MC.
2006, Curr Top Dev Biol, Vol. 76, pp. 217-257.
52. Vascular endothelial growth factor induced by hypoxia may mediate hypoxia-initiated angiogenesis.
Shweiki, Dorit, et al., et al. 1992, Nature, Vol. 359, pp. 843-845.
53. Abnormal blood vessel development and lethality in embryos lacking a single VEGF alllele. Carmeliet,
P, et al., et al. s.l. : 380, 1996, Nature, pp. 435-439.
54. Heterozygous embryonic lethality induced by targeted inactivation of the VEGF gene. Ferrara, N, et
al., et al. 1996, Nature, Vol. 380, pp. 439-442.
55. Hypoxic Tumor Cell Modulates Its Microenvironment to Enhance Angiogenic and Metastatic Potential
by Secretion of Proteins and Exosomes. Park, Jung Eun, et al., et al. 6, 2010, Mol Cell Proteomics, Vol. 9,
pp. 1085-1099.
56. Molecular responses to hypoxia in tumor cells. Kunz, Manfred and Ibrahim, Saleh M. 23, 2003,
Molecular Cancer, Vol. 2.
213
57. Comparison of angiogenesis-related factor expression in primary tumor cultures under normal and
hypoxic growth conditions. Heinzman, JM, Browe, SL and Bush, JE. 11, 2008, Cancer Cell |nt., Vol. 8.
58. Tumor induction of VEGF promoter activity in stromal cells. Fukamara, D, et al., et al. 1998, Cell, Vol.
94, pp. 715-725.
59. Stromal Cells Promote Angiogenesis and Growth of Human Prostate Tumors in a Differential Reactive
Stroma (DRS) Xenograft Model. Tuxhorn, Jennifer A, et al., et al. 2002, Cancer Research
,
Vol. 62, pp.
3298-3307.
60. Blood vessel maturation:vascular development comes of age. Darland, Diane C and D'Amore,
Patricia A. 2, 1999, J Clin Invest., Vol. 103, pp. 157-158.
61. Szekanecz, Zoltan and Koch, Alisa E. Chemokines and cytokines in inflammatory angiogenesis. [book
auth.] Michael P Seed and David A Walsh. Angiogenesis in Inflammation; Mechanisms and Clinical
Correlates. s.l. : Birkhauser Basel, 2008.
62. The role of chemokines and their receptor in angiogenesis. Kiefer, Friedemann and Siekmann, Arndt
F. 2011, Cell. Mol. Life Sci, Vol. 68, pp. 2811-2830.
63. Seed, Michael P and Walsh, David A. Angiogenesis in Inflamation; Mechanisms and Clinical
Correlates. Germany : Birkhauser Basel, 2008.
64. Differential endothelial migration and proliferation to basic fibroblast growth factor and vascular
endothelial growth factor. Yoshida, A, Anand-Apte, B and Zetter, BR. 1996, Growth Factors, pp. 57-64.
65. Vascular Endothelial Growth Factor: Basic Science and Clinical Progress. Ferrara, Napoleone. 4, 2004,
Endocrine Reviews, Vol. 25, pp. 581-611.
66. The enigmatic role of angiopoietin-1 in tumor angiogenesis. Metheny-Barlow, Linda J and Li, LuYuan. 5, 2003, Cell Research, Vol. 13, pp. 309-317.
67. Angiogenic factors FGF2 and PDGF-BB synergistically promote murine tumor neovascularization and
metastasis. Nissen, U, et al., et al. 2007, J Clin Invest., pp. 2766-2777.
68. IL-8 directly enhanced endothelial cell survival, proliferation, and matrix metalloproteinases
production and regulated angiogenesis. Li, A, et al., et al. 6, 2003, J Immunol., Vol. 170, pp. 3369-3376.
69. Stromal cell-derived factor 1 promotes angiogenesis via a heme oxygenase 1- dependent mechanism.
Deshane, Jessy, et al., et al. 3, 2007, JEM, Vol. 204, pp. 605-618.
70. VEGF, a prosurvivalfactor, acts in concert with TGF-beta to induce endothelial cell apoptosis. Ferrari,
G, et al., et al. 2006, Proc Natl Acad Sci USA, pp. 17260-17265.
214
71. Hepatocyte growth factor/scatter factor mediates angiogenesis through positive VEGF and negative
thrombospondin 1 regulation. Zhang, Yu-Wen, Su, Yanli,Volpert,Olga V and Vande Woude, George F.
22, 2003, Proc Natl Acad USA, Vol. 100, pp. 12718-12723.
72. Angiostatin: a novel angiogenesis inhibitor that mediates the suppression of metastases by a Lewis
lung carcinoma. O'Reilly, MS, et al., et al. 2, 1994, Cell., Vol. 79, pp. 315-328.
73. Endostatin; an endogenoud inhibitor of angiogenesis and tumor growth. O'Reilly, MS, et al., et al. 2,
1997, Cell., Vol. 88.
74. Inhibition of angiogenesis by interleukin 4. Volpert, OV, et al., et al. 6, 1998, J Exp Med., Vol. 188.
75. IL-13 attenuates vascular tube formation via JAK2-STAT6 pathway. Nishimura, Y, et al., et al. 3,
2008, Circ J., Vol. 72.
76. Plateletfactor 4: an inhibitor of angiogenesis. Bikfalvi, A. 3, 2004, Semin Thromb Hemost., Vol. 30,
pp. 379-385.
77. A tumor suppressor-dependent inhibitor of angiogenesis is immunologically and functionally
indistinguishable from afragment of thrombospondin. Good, DJ, et al., et al. 17, 1990, Proc Natl Aca Sci
USA, Vol. 87, pp. 6624-6628.
78. Vasostatin, a calreticulin fragment, inhibits angiogenesis and suppresses tumor growth. Pike, SE, et
al., et al. 12, 1998, J Exp Med., Vol. 188, pp. 2349-2356.
79. Why are tumour blood vessels abnormal and why is it important to know? Nagy, JA, et al., et al.
2009, British Journal of Cancer, Vol. 100, pp. 865-869.
80. Proliferative retinopathies: Angiogenesis that blinds. Sapieha, Przemyslaw, et al., et al. 2010, The
International Journal of Biochemistry & Cell Biology, Vol. 42, pp. 5-12.
81. Angiogenesis as a risk factor for plaque vulnerability. Di Stefano, R, Felice, F and Balbarini, A. 10,
2009, Curr Pharm Des., Vol. 15, pp. 1095-1106.
82. Angiogenesis in atherosclerosis: gathering evidence beyond speculation. Moulton, Karen S. 5, 2006,
Current Opinion in Lipidology, Vol. 17, pp. 548-555.
83. Atherosclerotic Plaque Progression and Vulnerability to Rupture: Angiogenesis as a Source of
Intraplaque Hemorrhage. Virmani, Renu, et al., et al. 2005, Arterioscler Thromb Vasc Biol, Vol. 25, pp.
2054-2061.
215
84. Vasculopathy and disordered angiogenesis in selected rheumatic diseases: rheumatoid arthritis and
systemic sclerosis. Koch, Alisa E and Distler, Oliver. Suppl 2, 2007, Arthritis Research & Therapy 2007,
Vol. 9, p. S3 (doi:10.1186/ar2187).
85. Angiogenesis drives psoriasis pathogenesis. Heidenreich, Regina, Rocken, Martin and Ghoreschi,
Kamran. 2009, Int. J. Exp. PAth., Vol. 90, pp. 232-248.
86. Angiogenesis in asthma. Ribatti, D, et al., et al. 2009, Clinical & Experimental Allergy, Vol. 39, pp.
1815-1821.
87. Mechanistic and Therapeutic Implications of Angiogenesis in Endometriosis. Taylor, Robert N, et al.,
et al. 2, 2009, Reprod Sci., Vol. 16, pp. 140-146.
88. Angiogenesis in chronic liver disease and its complications. Coulon, Stephanie, et al., et al. 2010,
Liver International, pp. 146-162.
89. Therapuetic Angiogenesis for Coronary Artery Disease. Freedman, Saul Benedict and Isner, Jeffery
M. 2002, Ann Intern Med., Vol. 136, pp. 54-71.
90. The Role of Angiogenesis in Damage and Recovery from Ischemic Stroke. Arenillas, Juan F, et al., et
al. 2007, Current Treatment Options in Cardiovascular Medicine, Vol. 9, pp. 205-212.
91. Cui, Tengjiao, Kirsner, Robert S and Jie, Li. Angiogenesis in Chronic Wounds. [book auth.] Chandan K
Sen. Advances in Wound Care: Volume 1. s.l. : Mary Ann Liebert, Inc., 2010, Vol. 1, pp. 347-352.
92. Cellular and molecular basis of wound healing in diabetes. Brem, Harold and Tomic-Canic, Marjana.
2007, J Clin Invest., Vol. 117, pp. 1219-1222.
93. Gene Therapy and Wound Healing. Eming, Sabine A, Krieg, Thomas and Davidson, Jeffrey M. 2007,
Clin Dermatol., Vol. 25, pp. 79-92.
94. Therapeutic angiogenesis: a single intra-arterial bolus of vascular endothelial growth factor
augments sollateral vesselformation in a rabbit ischemic hind-limb model. Takeshita, S, et at., et al.
1994, J Clin Invest, Vol. 93, pp. 662-670.
95. Angiogenic synergy,vascular stability and improvement of hind-limb ischemia by a combination
PDGF-BB and FGF-2. Cao, R, et al., et al. 2003, Nat Med, Vol. 9, pp. 604-613.
96. The use of hyperbaric oxygen therapy to treat chronic wounds: A review. Thackham, JA, McElwain,
DL and Long, RJ. 2008, Wound Repair Regen., Vol. 16, pp. 321-330.
97. Negative pressure wound therapy: Unusual and Innovative Applications. Cipolla, James, et al., et al.
2008, OPUS 12 Scientist, Vol. 2.
216
98. Topical negative pressure stimulates endothelial migration and proliferation: a suggested mechanism
for improved integration of Integra. Baldwin, C, et al., et al. 2009, Ann Plast Surg., Vol. 62, pp. 92-96.
99. In vitro optimisation of topical negative pressure regimens forangiogenesis into synthetic dermal
replacements. Potter, MJ, et al., et al. 2008, Burns., Vol. 34, pp. 164-174.
100. Quantitative angiogenesis assays: Progress and Problems. Jain, Rakesh K., et al., et al. 11, 1997,
Nature Medicine, Vol. 3, pp. 1203-1208.
101. Vascularization of the Brown-Pearce rabbit epithelioma transplant as seen in the transparent ear
chamber. Ide, A. G., Baker, N. H. and Warren, S. L. 1939, Ameri. J. Roentgenol., Vol. 42, pp. 891-899.
102. A new method for the microscopic study of living growing tissues by the introduction of a
transparent chamber in the rabbit's ear. Sandison, J. C. 1924, Anat. Rec., Vol. 28, pp. 281-287.
103. Culture of human endothelial cells derived from umbilical veins. Identification by morphologic and
immunologic criteria. Jaffe, E. A., et al., et al. 11, 1973, J Clin Invest., Vol. 52, pp. 2745-2756.
104. Human Vascular Endothelial Cells In Culture. Growth and DNA Synthesis. Gimbrone Jr., Michael A.,
Cotran, Ramzi S. and Folkman, Judah. 1974, The Journal of Cell Biology, Vol. 60, pp. 673-684.
105. Long-term culture of capillary endothelial cells. Folkman, J., Haudenschild, C. C. and Zetter, B. R.
10, 1979, Proc. Nati. Acad. Sci. USA., Vol. 76, pp. 5217-5221.
106. Angiogenesis in vitro. Folkman, J. and Haudenschild, C. 11, 1980, Nature, Vol. 288, pp. 551-556.
107. A fluidic device to study directional angiogenesis in complex tissue and organ culture models.
Barkefors, Irmeli, et al., et al. 2009, Lab Chip, Vol. 9, pp. 529-535.
108. Cell migration into scaffolds under co-culture conditions in a microfluidic platform. Chung, Seok, et
al., et al. 2009, Lab Chip, Vol. 9, pp. 269-275.
109. Surface-Treatment-Induced Three-Dimensional Capillary Morphogenesis in a Microfluidic Platform.
Chung, Seok, et al., et al. 2009, Adv. Mater., Vol. 21, pp. 4863-4867.
110. Matrix density mediates polarization and lumen formation of endothelial sprouts in VEGF gradients.
Shamloo, Amir and Heilshorn, Sarah C. 2010, Lab Chip, Vol. 10, pp. 3061-3068.
111. Tissue Engineering. Langer, Robert and Vacanti, Joseph P. 5110, 1993, Science, Vol. 260, pp. 920926.
112. The roles of tissue engineering and vascularization in the development of micro-vascular networks:
a review. Kannan, Ruben Y., et al., et al. 14, 2005, Biomaterials, Vol. 26, pp. 1857-1875.
217
113. Tissue engineering: Creation of long-lasting blood vessels. Koike, Naoto, et al., et al. 2004, Nature,
Vol. 428, pp. 138-139.
114. Tissue engineering-current challenges and expanding opportunities. Griffith, L.G. and Naughton, G.
5557, 2002, Science, Vol. 295, pp. 1009-1014.
115. Engineered blood vessel networks connect to host vasculature via wrapping-and-tapping
anastomosis. Cheng, Gang, et al., et al. 17, 2011, Blood, Vol. 118, pp. 4740-4749.
116. Engraftment of a vascularized human skin equivalent. Schechner, Jeffery S., et al., et al. 15, 2003,
FASEB J, Vol. 17, pp. 2250-2256.
117. Vascularization and engraftment of a human skin substitute using circulating progenitor cell-derived
endothelial cells. Shepherd, Benjamin R., et al., et al. 10, 2006, FASEB J, Vol. 20, pp. 1739-1741.
118. Biology on a chip: microfabrication for studying the behavior of cultured cells. Li, N., Tourovskaia, A.
and Folch, A. 5-6, 2003, Crit Rev Biomed Eng., Vol. 31, pp. 423-488.
119. Amperometric determination of nitric oxide derived from pulmonary artery endothelial cells
immobilized in a microchip channel. Spence, D. M., et al., et al. 11, 2004, Analyst., Vol. 129, pp. 9951000.
120. Deformation-induced release of ATP from erythrocytes in a poly(dimethylsiloxane)-based microchip
with channels that mimic resistance vessels. Price, A. K., et al., et al. 16, 2004, Anal Chem., Vol. 76, pp.
4849-4855.
121. Patterned deposition of cells and protein onto surfaces by using three-dimensional microfluidic
systems. Chiu, D. T., et al., et al. 6, 2000, Proc Natl Acad Sci U S A., Vol. 97, pp. 2408-2413.
122. Fabrication of patterned mutlicomponent protein gradients and gradient array using microfluidic
depletion. Fosser, K. A. and Nuzzo, R. G. 21, 2003, Anal Chem., Vol. 75, pp. 5775-5782.
123. Microfabrication platform for studying stem cellfates. Chin, V. I., et al., et al. 3, 2004, Biotechnol
Bioeng., Vol. 88, pp. 399-415.
124. Taking cell-matrix adhesions to the third dimension. Cukierman, E., et al., et al. 5547, 2001,
Science, Vol. 294, pp. 1708-1712.
125. Mechanobiology in the third dimension. Pedersen, J. A. and Swartz, M. A. 11, 2005, Ann Biomed
Eng., Vol. 33, pp. 1469-1490.
126. Cell culture: biology's new dimension. Abbott, A. 6951, 2003, Nature, Vol. 424, pp. 870-872.
218
127. RGD-dependent vacuolation and lumen formation observed during endothelial cell morphogenesis
in three-dimensional fibrin matrices involves the alpha(v)beta(3) and alpha(5)beta(1) integrins. Bayless,
K.J., Salazar, R. and Davis, G. E. 5, 2000, Am J Pathol., Vol. 156, pp. 1673-1683.
128. Sphingosine-1-phophate markedly induces matrix metalloproteinase and integrin-dependent human
endothelial cell invasion and lumen formation in three-dimensional collagen and fibrin matrices. Bayless,
K. J. and Davis, G. E. 4, 2003, Biochem Biophys Res Commun. , Vol. 312, pp. 903-913.
129. Effect of shear stress on microvascular network formation of endothelial cells with in vitro threedimensional model. Ueda, A., et al., et al. 3, 2004, Am J Physiol Heart Circ Physiol., Vol. 287, pp. H994H1002.
130. Interstialflow as a guidefor lymphangiogenesis. Boardman, K. C. and Swartz, M. A. 7, 2003, Circ
Res., Vol. 92, pp. 801-808.
131. Autocrine EGF receptor activation mediates endothelial migration and vascular morphogenesis
induced by VEGF under interstitialflow. Semino, Carlos E, Kamm, Roger D and Lauffenburger, Douglas
A. 3, 2006, Exp Cell Res., Vol. 312, pp. 289-298.
132. Fabrication of cell-containing hydrogel microstructures inside microfluidic devices that can be used
as cell-based biosensors. Koh, W. G. and Pishko, M. V. 8, 2006, Anal Bioanal Chem., Vol. 385, pp. 13891397.
133. Microscale multilayer cocultures for biomimetic blood vessels. Tan, W. and Desai, T. A. 2, 2005, J
Biomed Mater Res A., Vol. 72, pp. 146-160.
134. Gelatin based microfluidic devices for cell culture. Paguirigan, A. and Beebe, D. J. 3, 2006, Lab Chip,
Vol. 6, pp. 407-413.
135. Design of nanostuctured biological materials through self-assembly of peptides and proteins. Zhang,
S., et al., et al. 6, 2002, Curr Opin Chem Biol., Vol. 6, pp. 865-871.
136. Long-term stability of plasma oxidized PDMS surfaces. Kim, B., ET, K. P. and Papautsky, I. 2004,
Conf Proc IEEE Eng Med Biol Soc., Vol. 7, pp. 5013-5016.
137. Thermal aging and reduced hydrophobic recovery of polydimethylsiloxane. Eddington, David T.,
Puccinelli, John P. and Beebe, David J. 1, 2006, Sensors and Actuators B: Chemical, Vol. 114, pp. 170172.
138. Vascular mimetics based on microfluidics for imaging the leukocyte--endothelial inflammatory
response. Schaff, U. Y., et al., et al. 4, 2007, Lab Chip, Vol. 7, pp. 448-456.
219
139. The role of polypeptide growth factors in human carcinomas: new targets for a novel
pharmacological approach. Favoni, R. E. and de Cupis, A. 2, 2000, Pharmacol Rev., Vol. 52, pp. 179-206.
140. Intracellular macromolecular mobility measured by fluorescence recovery after photobleaching with
confocal laser scanning microscopes. Braga, J., Desterro, J. M. and Carmo-Fonseca, M. 10, 2004, Mol
Biol Cell., Vol. 15, pp. 4749-4760.
141. Diffusion and convection in collagen gels: implication for transport in the tumor interstitium.
Ramanujan, S., et al., et al. 3, 202, Biophys J., Vol. 83, pp. 1650-1660.
142. Generation of stable concentration gradients in 2D and 3D environment using a microfluidic ladder
chamber. Saadi, W., et al., et al. 5, 2007, Biomed Microdevices., Vol. 9, pp. 627-635.
143. Angiogenesis in cancervascular rhematoid and other disease. Folkman, Judah. s.l. : Nature, 1995,
Nature Medicine, Vol. 1.
144. Oxygen-sensing in tumors. Verma, A. 4, 2006, Curr Opin Clin Nutr Metab Care., Vol. 9, pp. 366-378.
145. Hypoxia-associated proteins. Graven, K. K. and Farber, H. W. 2, 1995, New Horiz., Vol. 3, pp. 208218.
146. Laminar shear inhibits tubule formation and migration of endothelial cells by an angiopoietin-2
dependent mechanism. Tressel, S. L., et al., et al. 10, 2007, Arterioscler Thromb Vasc Biol., Vol. 27, pp.
2150-2156.
147. Impact of mechanical stress and tension-stress on angiogenesis in wound healing. Wang, H. Q. and
Zhao, L. 2, 2006, Chin J Traumatol., Vol. 9, pp. 118-124.
148. Tensionalforce in fibrillar extracellular matrices control directional capillary sprouting. Korff,
Thomas and Augustin, Hellmut G. 1999, Journal of Cell Science, Vol. 112, pp. 3249-3258.
149. Molecular mechanisms of blood vessel growth. Conway, E. M., Collen, D. and Carmeliet, P. 3, 2001,
Cardiovasc Res., Vol. 49, pp. 507-521.
150. How does extracellular matrix control capillary mophogenesis? lngber, D. E. and Folkman, J. 5,
1989, Cell., Vol. 58, pp. 803-805.
151. Mechanochemical switching between growth and differentiation during fibroblast growth factorstimulated angiogenesis in vitro: role of extracellular matrix. |ngber, D. E. and Folkman, J. 1, 1989, J Cell
Biol., Vol. 109, pp. 317-330.
152. The Stiffness of Three-dimensional Ionic Self-assembling Peptide Gels Affects the Extent of Capillarylike Network Formation. Sieminski, A. L., et al., et al. 2, 2007, Cell Biochem. Biophys., Vol. 49, pp. 73-83.
220
153. Transport-mediated angiogenesis in 3D epithelial coculture. Sudo, Ryo, et al., et al. 7, 2009, FASEB,
Vol. 23, pp. 2155-2164.
154. Conjugated Polymer Nanoparticles for Two-Photon Imaging of Endothelial Cells in a Tissue Model.
Abdul Rahim, Nur Aida, et al., et al. 34, 2009, Advanced Materials, Vol. 21, pp. 3492-3496.
155. Dettermining Cell Fate Transition Probabilities to VEGF/Ang 1 Levels: Relating Computational Model
to Microfluidic Angiogenesis Studies. Das, Anusuya, et al., et al. 4, 2010, Cellular and Molecular
Bioengineering, Vol. 3, pp. 345-360.
156. Hydrogelsfor Tissue Engineering. Lee, K. Y. and Mooney, David J. 7, 2001, Chemical Reviews, Vol.
101, pp. 1869-1879.
157. Adhesion events in angiogenesis. Eliceiri, Brian P. and Cheresh, David A. 2001, Current Opinions in
Cell Biology, Vol. 13, pp. 563-568.
158. Effects of Substrate Stiffness on Cell Morphology, Cytoskeletal Structure, and Adhesion. Yeung,
Tony, et al., et al. 2005, Cell Motility and the Cytoskeleton, Vol. 60, pp. 24-34.
159. Role of Laminin and Basement Membrane in the Morphological Differentiation of Human
Endothelial Cells into Capillary-like Structures. Kubota, Yasuo, et al., et al. 1988, The Journal of Cell
Biology, Vol. 107, pp. 1589-1598.
160. The relative magnitudes of endothelialforce generation and matrix stifness modulate capillary
mophogenesis in vitro. Sieminski, A. L., Hebbel, R. P. and Gooch, K. J. 2004, Experimental Cell Research,
Vol. 297, pp. 574-584.
161. Novel peptide-based biomaterial scaffolds for tissue engineering. Holmes, Todd C. 1, 2002, Trends
in Biotechnology, Vol. 20, pp. 16-21.
162. Design of nanostructured biological materials through self-assembly of peptides and proteins.
Zhang, Shuguang, et al., et al. 2002, Current Opinion in Chemical Biology, Vol. 6, pp. 865-871.
163. Self-complementary oligopeptide matrices support mammalian cell attachment. Zhang, Shuguang,
et al., et al. 1995, Biomaterials, Vol. 16, pp. 1385-1393.
164. Spontaneous assembly of a sel-complementary oligopeptide to form a stable macroscopic
membrane. Zhang, Shuguang, et al., et al. 1993, Proc. Natl. Acad. Sci., Vol. 90, pp. 3334-3338.
165. Functionalized scaffolds of shorter self-assembling peptides containing MMP-2 cleavable motif
promote fibroblast proliferation and significantly accelerate 3-D cell migration independent of scaffold
stiffness. Kumuda, Yoshiyuki, Hammond, Nathan A. and Zhang, Shuguang. 2010, Soft Matter, Vol. 6,
pp. 5073-5079.
221
166. Designer functionalized self-assembling peptide nanofiber scaffolds for growth, migration, and
tubulogenesis of human umbilical vein endothelial cells. Wang, Xiumei, Horii, Akihiro and Zhang,
Shuguang. 2008, Soft Matter, Vol. 4, pp. 2388-2395.
167. Extensive neurite outgrowth and active synapse formation on self-assembling peptide scaffolds.
Holmes, Todd C., et al., et al. 12, 2000, Proc. Nati. Acad. Sci., Vol. 97, pp. 6728-6733.
168. Self-assembling peptide hydrogelfosters chondrocyte extracellular matrix production and cell
division: Implication for catilage tissue repair. Kisiday, J., et al., et al. 15, 2002, Proc. NatI. Acad. Sci., Vol.
99, pp. 9996-10001.
169. Functional differentiation of hepatocyte-like spheroid structures from putative liver progenitor cells
in three-dimensional peptide scaffolds. Semino, Carlos E., et al., et al. 4-5, 2003, Differentiation, Vol. 71,
pp. 262-270.
170. The effect offunctionalized self-assembling peptide scaffolds on human aortic endothelial cell
function. Genove, Elsa, et al., et al. 16, 2005, Biomaterials, Vol. 26, pp. 3341-3351.
171. Aqueous Humor Dynamics: A Review. Goel, Manik, et al., et al. 2010, The Open Ophthalmology
Journal, Vol. 4, pp. 52-59.
172. Uveoscleral outflow-- a review. Alm, A. and Nilsson, S. F. 4, 2009, Exp Eye Res., Vol. 88, pp. 760768.
173. The probability of having glaucoma at different lOP levels. Davanger, Martin, Ringvold, Amund and
Blika, Sigmund. 5, 1991, Acta Ophthalmologica, Vol. 69, pp. 565-568.
174. Structural changes of the trabecular meshwork in different kinds of glaucoma. Tektas, Ozan-Yuksel
and Lutjen-Drecoll, Elke. 4, 2009, Exp Eye Res., Vol. 88, pp. 769-775.
175. Morphological changes in glaucomatous eyes and the role of TGFbeta2for the pathogenesis of the
disease. Lutjen-Drecoll. 1, 2005, Exp Eye Res., Vol. 81, pp. 1-4.
176. The Pore Density in the Inner Wall Endothelium of Schlemm's Canal of Glaucomatous Eyes. Johnson,
Mark, et al., et al. 9, 2002, Invest. Ophthalmol. Vis. Sci., Vol. 43, pp. 2950-2955.
177. What controls aqueous humour outflow resistance? Johnson, Mark. 4, 2006, Exp Eye Res., Vol. 82,
pp. 545-557.
178. The Inner Wall of Schlemm's Canal. Ethier, C. Ross. 2, 2002, Exp. Eye Res., Vol. 74, pp. 161-172.
179. The trabecular meshwork outflow pathways: Structural andfunctional aspects. Tamm, Ernst R. 4,
2009, Exp. Eye Res., Vol. 88, pp. 648-655.
222
180. Extracellular matrix in the trabecular meshwork. Acott, Ted S. and Kelley, Mary J. 4, 2008, Exp. Eye
Res., Vol. 86, pp. 543-561.
181. Experiments on Aqueous Outflow in Human and Rabbit Eyes. Moi, ivar. 3, 1958, Acta
Ophthalmologica, Vol. 36, pp. 387-393.
182. Experimental Aqueous Perfusion in Enucleated Human Eyes. Grant, W. Morton. 6, 1963, Arch
Ophthalmol., Vol. 69, pp. 783-801.
183. Cationic Ferritin and Segmental Flow through the Trabecular Meshwork. Hann, Cheryl R., Bahler,
Cindy K. and Johnson, Douglas H. 2005, Invest Ophthalmol Vis Sci., Vol. 46, pp. 1-7.
184. Giant Vacuoles Are Found Preferentially near Collector Channels. Parc, Christine E., Johnson,
Douglas H. and Brilakis, Harilaos S. 2000, Invest Ophthalmol Vis Sci., Vol. 41, pp. 2984-2990.
185. Human Trabecular Meshwork Organ Culture. Johnson, Douglas H. and Tschumper, Renee C. 1987,
Invest Ophthalmol Vis Sci, Vol. 28, pp. 945-953.
186. Effects of TGF-beta2 in perfused human eyes. Gottanka, J., et al., et al. 1, 2004, Invest Ophthalmol
Vis Sci., Vol. 45, pp. 153-158.
187. Segmental Versican Expression in the Trabecular Meshwork and Involvement in Outflow Facility.
Keller, Kate E., et al., et al. 8, 2011, Invest. Ophthalmol. Vis. Sci., Vol. 52, pp. 5049-5057.
188. Aqueous Humor in Glaucomatous Eyes Contains an Increased Level of TGF-beta2. Tripathi, Ramesh
C., et al., et al. 6, 1994, Exp. Eye Res., Vol. 59, pp. 723-728.
189. Rate of change in central corneal thickness: a viability indicator for conventional drainage tissues in
organ culture. Wan, Z., et al., et al. 6, 2006, Exp. Eye Res., Vol. 82, pp. 1086-1093.
190. Trabecular Meshwork Cells Grown on Filters. Conductivity and Cytochalasin Effects. Perkins, Todd
W., et al., et al. 12, 1988, Invest Ophtalmol Vis Sci., Vol. 29, pp. 1836-1846.
191. Endothelial of Schlemm's canal and trabecular meshwork: distinct molecular, functional, and
anatomic features. Alvarado, Jorge A. 3, 2004, Am J Physiol Cell Physiol, Vol. 286, pp. C621-C634.
192. A model of giant vacuole dynamics in human Schlemm's canal endothelial cells. Pedrigi, Ryan M., et
al., et al. 1, 2011, Exp. Eye Res., Vol. 92, pp. 57-66.
193. Associations between the Cells of the Wall of Schlemm's Canal. Grierson, I., et al., et al. 1978,
AlbrechtVon Graefes Arch. Klin. Exp. Ophthalmol., Vol. 208, pp. 33-47.
223
194. The changing paradigm of outflow resistance generation: towards synergistic models of JCT and
inner wall endothlelium. Overby, D. R., Stamer, W. D. and Johnson, M. 4, 2009, Exp. Eye Res., Vol. 88,
pp. 656-670.
195. Constitutive secretion of chemokines by cultured human trabecular meshwork cells. Shifera, A. S., et
al., et al. 1, 2010, Exp. Eye Res., Vol. 91, pp. 42-47.
196. Genes Upregulated in the Human Trabecular Meshwork in Response to Elevated Intraocular
Pressure. Gonzalez, Pedro, Epstein, David L.and Borras, Teresa. 2, 2000, Invest. Ophthalmol. Vis Sci.,
Vol. 41, pp. 352-361.
197. Effects of Mechanical Stretching on Trabecular Matrix Metalloproteinases. Bradley, John M. B., et
al., et al. 7, 2001, Invest. Ophthalmol. Vis. Sci., Vol. 42, pp. 1505-1513.
198. A new insight into the cellular regulation of aqueous outflow: how trabecular meshwork endothelial
cells drive a mechanism that regulates the permeability of Schlemm's canal endothelial cells. Alvarado, J.
A., et al., et al. 11, 2005, Br J Ophthalmol, Vol. 89, pp. 1500-1505.
199. Biomechanics of Schlemm's Canal Endothelial Cells: Influence on F-Actin Architecture. Ethier, C.
Ross, Read, A. Thomas and Chan, Darren. 4, 2004, Biophysical Journal, Vol. 87, pp. 2828-2837.
200. Isolation, culture, and characterization of endothelial cells from Schlemm's canal. Stainer, W. D., et
al., et al. 10, 1998, Invest. Ophthalmol. Vis. Sci., Vol. 39, pp. 1804-1812.
201. The mechanism of increasing outflowfacility by rho-kinase inhibition with Y-27632. Lu, Zhaozeng, et
al., et al. 2, 2008, Vol. 86, pp. 271-281.
202. Effects of the Rho kinase inhibitor Y27-632 and the phosphatase inhibitor calyculin A on outflow
facility in monkeys. Tian, B and Kaufman, P. L. 2, 2005, Exp. Eye Res., Vol. 80, pp. 215-225.
203. Structural basement membrane components and corresponding integrins in Schlemm's canal
endothelia. VanderWyst, S. S., et al., et al. 2011, Mol Vis., Vol. 17, pp. 199-209.
204. Cytoskeletal Involvement in the Regulation of Aqueous Humor Outflow. Tian, Baohe, et al., et al. 3,
2000, Invest. Ophthalmol. Vis. Sci., Vol. 41, pp. 619-623.
205. Mechanical responsiveness of the endothelial cell of Schlemm's canal: scope, variability and its
potential role in controlling aqueous humor outflow. Zhou, E. H., et al., et al. 2011, J. R. Soc. Interface.
206. The third dimentsion bridges the gap between cell culture and live tissue. Pampaloni, Francesco,
Reynaud, Emmanuel G. and Stelzer, Ernst H. K. 2007, Nature Reviews Molecular Cell Biology, Vol. 8, pp.
839-845.
224
207. Mechanisms of angiogenesis. Risau, Werner. s.l. : Nature, 1997, Nature, pp. 671-674.
208. Molecular mechanisms and clinical appkications of angiogenesis. Carmeliet, Peter and Jain, Rakesh
K. 2011, Nature, pp. 298-307.
209. Effects of Mechaniccal Forces on Signal Transduction and Gene Expression in Endothelial Cells.
Chien, Shu, Li, Song and Shyy, John Y-J. 1998, Hypertension.
210. Transendothelialflow inhibits neutrophil transmigration through a nitric oxide-dependent
mechanism:potential role fro cleft shear stress. Burns, Alan R, et al., et al. 2007, Am J Physiol Heart Circ
Physiol, Vol. 293, pp. H2904-H2910.
211. Regulation of hydraulic conductivity in response to sustained changes in pressure. Kim, Min-ho,
Harris, Norman R and Tarbell, John M. 2005, Am J Physiol Heart Circ Physiol, Vol. 289, pp. H2551H2558.
212. A Flow-gatedAngiogenesis Switch. Vickerman, Vernella and Kamm, Roger D. Singapore: s.n.,
2010. World Congress of Biomechanics.
213. Fluidforces control endothelial sprouting. Song, Jonathan W and Munn, Lance L. 2011.
214. Migration and proliferation of endothlelial cells in preformed cells and newly formed blood vessels
during tumor angiogenesis. Ausprunk, DH and Folkman, J. 1, 1977, Microvasc Res., Vol. 14, pp. 53-65.
215. Transport of Molecules in the Tumor Intertitium: A Review. Jain, Rakesh K. 1987, Cancer Res, Vol.
47, pp. 3039-3051.
216. High interstitialfluid pressure - an obstacle in cancer therapy. Heldin, Carl-Henrik, et al., et al.
2004, Nature Reviews Cancer, Vol. 4, pp. 806-813.
217. A driving force for change: interstitialflow as a morphoregulator. Rutkowski, JM and Swartz, MA.
1, 2007, Trends Cell Biol., Vol. 17, pp. 44-50.
218. Image Processing with Imagei. Abramoff, M.D., Magalhaes, P.J. and Ram, S.J. 7, 2004,
Biophotonics International, Vol. 11, pp. 36-42.
219. Genistein, a Specific Inhibitor of Tyrosine-specific Protein Kinases. Akiyama, Tetsu, et al., et al.
1987, J Biol Chem., pp. 5592-5595.
220. Discovery of a novel, potent, and Srcfamily-selective tyrosine kinase inhibitor. Study of Lck- and
FynT-dependent T cell activation. Hanke, J.H., et al., et al. 1996, J. Biol.Chem., Vol. 271, pp. 695-701.
221. Calcium sensitization of smooth muscle mediated by a Rho-associated protein kinase in
hypertension. Uehata, M, et al., et al. 6654, 1997, Nature, Vol. 389, pp. 990-994.
225
222. Characterization of three inhibitors of endothelial nitric oxide synthase in vitro and in vivo. Rees,
D.D, et al., et al. 3, 1990, Br J Pharmacol, Vol. 101, pp. 746-752.
223. ROCK- and Myosin-Dependent Matrix Deformation Enables Protease-Independent Tumor-Cell
Invasion In Vivo. Wyckoff, Jeffrey B, et al., et al. 2006, Curr Biol, pp. 1515-1523.
224. Pharmacological inhibition of Rho-kinase signaling with Y-27632 blocks melanoma tumor growth.
Routhier, Alissa, et al., et al. 2010, Vol. 23, pp. 861-867.
225. Tumor-derived endothelial cells exhibit aberrant Rho-mediated mechanosensing and abnormal
angiogenesis in vitro. Ghosh, Kaustabh, et al., et al. 2008, Proc Natl Acad Sci USA, pp. 11305-11310.
226. Effect of Vascular Endothelial Growth Factor on Cultured Endothelial Cell Monolayer Transport
Properties. Chang, Yong S., et al., et al. 2000, Microvascular Research, Vol. 59, pp. 265-277.
227. RhoA/ROCK signaling is essentialfor multiple aspects of VEGF-mediated angiogenesis. Bryan, Brad
A., et al., et al. 2010, FASEB J., Vol. 24, pp. 3186-3195.
228. Heparan Sulfate Proteoglycan Is a Mechanosensor on Endothelial Cells. Florian, Jeffry A., et al., et
al. 2003, Circulation Research, Vol. 93, pp. e136-e142.
229. Heparan Sulfate Proteoglycans Mediate Interstitial Flow Mechanotransduction Regulating MMP-13
Expression and Cell Motility via FAK-ERK in 3D Collagen. Shi, Z-D, Wang, H and Tarbell, JM. 1, 2011, PLoS
ONE, Vol. 6, p. e15956.
230. Integrin regulate VE-cadherin and catenins: Dependence of this regulation on Src, but not on Ras.
Wang, Yingxiao, et al., et al. s.l.: PNAS, 2005.
231. The relationship between force and focal complex development. Galbraith, Catherine G., Yamada,
Kenneth M. and Sheetz, Michael P. 4, 2002, J Cell Biol, Vol. 159, pp. 695-705.
232. Cell Stiffening in Response to External Stress is Correlated to Actin Recruitment. Icard-Arcizet,
Delphine, et al., et al. 7, 2008, Biophys J., Vol. 94, pp. 2906-2913.
233. Cyclic Force Applied to FAs Induces Actin Recruitment Depending on the Dynamic Loading Pattern.
Ueki, Y., Sakamoto, N. and Sato, M. 2010, Open Biomed Eng J., Vol. 4, pp. 129-134.
234. Src-induced de-regulation of E-cadherin in colon cancer cells requires integrin signaling. Avizienyte,
E. 2002, Nature Cell Biol., Vol. 4, pp. 632-638.
235. Inactivation of Src family kinases inhibit angiogenesis in vivo: implications for a mechanism
involving organization of the actin cytoskeleton. Kilarski, Witold W, Jura, Natalia and Gerwins, Par. 1,
2003, Experimental Cell Research, Vol. 291, pp. 70-82.
226
236. Phosphospecific antibodies revealfocal adhesion kinase activation loop phosphorylation in nascent
and mature focal adhesions and requirement for the autophosphorylation site. Ruest, PJ, et al., et al.1,
2000, Cell Growth Differ., Vol. 11, pp. 41-48.
237. Src kinase phosphorylates vascular endothelial-cadherin in response tovascular endothelial growth
factor:identification of tyrosine 685 as the unique target site VE-cadherin Tyr685: a target site for Src
kinae. Wallez, Y, et al., et al. 2007, Oncogene.
238. Signal transduction from the Extracellular Matrix. Juliano, R.L. and Haskill, S. 1993, J Cell Biol, pp.
577-585.
239. Visualizing the mechanical activation of Src. Wang, Yingxiao, et al., et al. 2005, Nature, pp. 10401045.
240. Integrin Activation and Matrix Binding Mediated Cellular Responses to Mechanical Stretch.
Katsumi, Akira, et al., et al. 2005, Journal of Biological Chemistry.
241. pp125FAK, a structurally distinctive protein tyrosine kinase associated with focal adhesions.
Schaller, M.D., et al., et al. 11, 1992, Proc. Nati. Acad. Sci. USA, Vol. 89, pp. 5192-5196.
242. Cell Adhesion or Integrin CLustering Increases Phosphorylation of a Focal Adhesion-associated
Tyrosine Kinase. Kornberg, L, et al., et al. 33, 1992, The Journal of Biological Chemistry, Vol. 267, pp.
23439-23442.
243. Integrins in angiogenesis and lymphangiogenesis. Avraamides, Christie J, Garmy-Susini, Barbara
and Varner, Judith A. 2008, Nature Reviews, Vol. 8.
244. Angiogenesis promoted by vascular endothelial growth factor: regulation through alphalbetal and
alpha2betal integrins. Senger, D.R. 1997, Proc. Natl Acad. Sci. USA, pp. 13612-13617.
245. Cavallaro, U and Christofori, G. 2004, Nature Review Cancer.
246. Tyrosine phosphorylation of VE-cadherin prevents binding of p120- and beta-catenin and maintains
the cellular mesenchymal state. Potter, MD, Barbero, S and Cheresh, DA. 2005, J Biol Chem.
247. Phosphorylation of VE-cadherin controls endothelial phenotypes via p120-catenin coupling and
Rac1 activation. Hatanaka, K, Simons, M and Murakami, M. 2011, Am J Physiol Heart Circ Physiol, pp.
H162-H172.
248. Pathophysiological consequences of VEGF-induced vascular permeability. Weis, S.M. and Cheresh,
D.A. 2005, Nature, pp. 497-504.
227
249. Epidermal growth factor-induced activation and translocation of c-Src to the cytoskeleton depends
on the actin binding domain of the EGF-receptor. van der Heyden, Marcel A.G, et al., et al. 1998.
250. The role of cadherin-catenin-cytoskeleton complex in angiogenesis:antisense oliogonucleotide of
plakoglobin promotes angiogenesis in vitro, and protein kinace C (PKC) enhances angiogenesis through
the plakoglobib signaling pathway. Nagashima, H, et al., et al. 1997, Heart Vessels.
251. Mechanotransduction: All Signals Point to Cytoskeleton, Matrix, and Integrins. Alenghat, Francis J
and Ingber, Donald E. 2002, STKE.
252. Integrins in invasive growth. Brakebusch, Cord, et al., et al. 2002.
253. Epithelial-mesenchymal transition in tumour progression. Thiery, JP. 6, 2002, Nat. Rev Cancer, Vol.
2, pp. 442-454.
254. TGFbetal promotes actin cytoskeleton reorganization and migratory phenotype in epithelial
trachael cells in primary culture. Boland, Sonja, et al., et al. 1996, Journal of Cell Science, Vol. 109, pp.
2207-2219.
255. Complete reversal of epithelial to mesenchymal transition requires inhibition of both ZEB expression
and the Rho pathway. Das, Shreyas, et al., et al. 2009, BMC Cell Biology.
256. Identification of Src-Specific Phosphorylation Site on Focal Adhesion Kinase: Dissection of the Role of
Src SH2 and Catalytic Functions and Their Consequences for Tumor Cell Behavior. Brunton, Valerie G., et
al., et al. 2005, Cancer Res, Vol. 65, pp. 1335-1342.
257. Effects of Disturbed Flow on Vascular Endothelium: Pathophysiological Basis and Clinical
Perspectives. Chiu, Jeng-Jiann and Chien, Shu. 1, 2011, Physiol Rev, Vol. 91, pp. 327-387.
258. Force-induced focal adhesion translocation: effects offorce amplitude and frequency. Mack, P.J., et
al., et al. 2004, Am J Physiol Cell Physiol , Vol. 287, pp. C954-C962.
259. Integral-mass balance methodfor determination of solvent drag reflection coefficient. Wolf,
Matthew B., Watson, Philip D. and Scott Il, Duncan R. C. 1987, Am. J. Physiol., Vol. 253, pp. H194-H204.
260. A Practical Extension of Hydrodynamic Theory of Porous Transport for Hydrophilic Solutes.
Bassingthwaighte, James B. 2, 2006, Microcirculation, Vol. 13, pp. 111-118.
261. Transient transcapillary exchange of water driven by osmoticforces in the heart. Kellen, Michael R.
and Bassingthwaighte, James B. 2003, Am J Physiol Heart Circ Physiol, Vol. 285, pp. H1317-H1331.
262. VEGF increases paracellular transport without altering the solvent-drag reflection coefficient.
DeMaio, Lucas, et al., et al. 3, 2004, Microvascular Research, Vol. 68, pp. 295-302.
228
263. Bacterial exotoxins and endothelial permeability for water and albumin in vitro. Suttorp, Norbert,
et al., et al. 3, 1988, Am J Physiol Cell Physiol, Vol. 255, pp. C368-C376.
264. Live Cell Imaging of Src/FAK Signaling by FRET. Seong, Jihye, Lu, Shaoying and Wang, Yingxiao. 4,
2011, Cell Mol Bioeng., Vol. 2, pp. 138-147.
265. FRET and mechanobiology. Wang, Yingxiao and Wang, Ning. 2009, Integr. Biol., Vol. 1, pp. 565-573.
266. Mosser, Eric A. Visualization of caderin-cadherin association in living cells. Dissertation (Ph.D.). s.l.
California Institute of Technology, 2007.
267. Reduced UVlight scattering in PDMS microfluidic devices. Seiffert, Sebastian, et al., et al. 2011, Lab
Chip, Vol. 11, pp. 966-968.
268. Oxygen gradients for open well cellular cultures via microfluidic substrates. Lo, Joe F., Sinkala, Elly
and Eddington, David T. 2010, Lab Chip, Vol. 10, pp. 2394-2401.
269. A microengraving method for rapid selection of single cells producing antigen-specific antibodies.
Love, J. Christopher, et al., et al. 6, 2006, Nature Biotechnology, Vol. 24, pp. 703-707.
270. Near-infrared fluorescent sensors based on single-walled carbon nanotubes for life sciences
applications. Boghossian, AA, et al., et al. 7, 2011, ChemSusChem., Vol. 4, pp. 848-863.
271. Rapid Prototyping of Microfluidic Systems in Poly(dimethylsiloxane). Duffy, DC, et al., et al. 1998,
Anal Chem., Vol. 70, pp. 4974-4984.
272. Tumor Angiogenesis: Therapeutic Implications. Folkman, Judah. 1971, N Engl J Med, Vol. 285, pp.
1182-1186.
229