Development of Biomimetic Microfluidic Adhesive Substrates
for Cell Separation
by
Chia-Hua Lee
CMASSACHUSETTS
B.S., National Taiwan University, Taiwan (2003)
M.S., National Taiwan University, Taiwan (2005)
INSTt1UTE
OF TECHNOLOGY
OF.TECNOLOG
MAY 14211
M.S., Massachusetts Institute of Technology, USA (2008)
LiBRARES
Submitted to the Department of Materials Science and Engineering
in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 2014
© 2014 Massachusetts Institute of Technology. All rights reserved
Signature of Author
Department of Materials Science and Engineering
January 10 th014
Certified by
Rohit Karnik
Associate Professor of Mechanical Engineering
Thesis Co-advisor
Krystyn J. Van Vliet
V
Associat Pr fessor of Materials Science and Engineering
Thesis Co-advisor
Accepted by
Chair, Depart
ommittee on Graduate Student
Development of Biomimetic Microfluidic Adhesive Substrates
for Cell Separation
by
Chia-Hua Lee
Submitted to the Department of Materials Science and Engineering on January
10 th
2014
in partial fulfillment of the requirements for the Degree of Doctor of Philosophy in
Materials Science and Engineering
ABSTRACT
Cell separation is important in medical, biological research, clinical therapy, diagnostics
and many other areas. The conventional methods of cell sorting have limited applications
due to sophisticated equipment settings, high costs, or time-intensive and labor-intensive
processing steps. Inspired from natural cell sorting system-cell rolling, a novel
microfluidic device design was proposed for point-of-care and point-of-use applications.
It relies on interaction of cells with biomimetic adhesive substrates comprising multiple
inclined, asymmetric bands of weak adhesive molecules. Such device design allows
continuous sorting of cells without irreversible capture of cells. To realize such device,
comprehensive studies of how cells settle onto the substrate, how cells capture by the
substrate, the effect of substrate parameters on separation potential, and selection of
adhesion molecules are needed to optimize device performance.
In this thesis, first, how the cells settle and how they are captured by the receptors were
studied using HL60 cells as a model leukocyte cell line and P-selectin as a model
receptor. Settling distance of HL60 cells under different shear stresses inside microfluidic
channels was identified from the study of convection velocity of cells at different position
along the channel. The results show that settling distance increases with increasing shear
stress. Cell capture was then quantified by characterizing how far settled HL60 cells
travelled before they were captured by P-selectin molecules, defined as the attachment
distance. Cumulative probabilities of attachment distance of cells at different shear
stresses revealed that increasing shear stress results in exponential increase of the
attachment distance of cells by receptor molecules. An empirical model was developed to
predict capture probability by an inclined receptor band and the prediction value was
verified by experimental data from a device.
3
Second, a patterning method involving microcontact printing was developed to create
biomimetic adhesive substrates comprising multiple inclined receptor bands of P-selectin
molecules. The patterned substrates were then used to study how transport of HL60 cells
can be controlled by the substrate parameters including pattern inclination angle with
respect to shear flow direction, shear stress magnitude, and P-selectin incubation
concentration. The effects of substrate parameters were quantified in terms of the edge
tracking length, lateral displacement, and the rolling velocity. The edge inclination angle
was identified as the strongest modulator of edge tracking length on a single band for
captured cells. To study optimization of the device design, experimental data of cell
settling, cell attachment, and edge tracking length were integrated into a model to predict
device performance including device capture efficiency and total lateral displacement.
General guidelines for microfluidic device design were established based on the results
from the model: smaller band width, edge angle of 15-20', and lower shear stress.
Finally, to develop new specific receptor-ligand systems, M13 pVIII and pIll phage
libraries were used for selecting peptides with affinity to CD4 proteins. Screened phage
from pVIII library was immobilized on the gold surface and capture efficiency of CD4+
cells were characterized. The interaction between selectin phage and CD4+ cells were
demonstrated to be CD4-dependent. Moreover, the selected phage from pIII library and
the corresponding synthetic peptides were demonstrated to exhibit specificity to CD4
proteins.
In summary, this thesis focuses on development of biomimetic adhesive substrates for
microfluidic devices involving transient interactions between the cells and the receptorpatterned substrates. How cells flow and get captured by patterned biomimetic substrates
inside the microfluidic channels, how substrates parameters affect cell rolling trajectories
and device performance, and how to identify new receptor-ligand systems were discussed
in this thesis. This study has led to realization of a microfluidic device for separating
neutrophils from blood. This microfluidic system provides continuous sorting without
irreversible capture of cells, and is believed to be an effective method that can potentially
be used in many point-of-care applications.
Keywords: microfluidics, cell separation, cell rolling, selectin, biopanning, M13
Thesis Committee:
Prof. Rohit Karnik, Associate Professor of Mechanical Engineering, MIT. (Thesis Co-advisor)
Prof. Krystyn J. Van Vliet, Associate Professor of Materials Science and Engineering, MIT.
(Thesis Co-advisor)
Professor Angela M. Belcher, Professor of Materials Science and Engineering, MIT.
Professor Darrell J. Irvine, Professor of Materials Science and Engineering, MIT.
Prof. Jeffrey M. Karp, Associate Professor of Medicine, Harvard Medical School, Brigham and
Women's Hospital.
4
Acknowledgements
My time at MIT has been an incredible journey. I owe thanks to many people for their
guidance, support, encouragement, patience, and friendship.
I would like to express my gratitude to my PhD thesis advisors (Prof. Rohit Karnik and
Prof. Krystyn Van Vliet) and my committee members (Prof. Angela Belcher, Prof.
Jeffrey Karp and Professor Darrell Irvine) for all their support and guidance, which
without, this thesis could not have been achieved. Prof. Kamik, especially, has been an
exceptional role model and mentor to me. He has been always there to listen and give
advice with his patience. His unflinching encouragement and support have helped me
overcome many crisis situations and finish this dissertation. I am also deeply grateful to
Prof. Van Vliet for the discussions that have helped me sort out a lot of technical details
of my work in the past few years. She has left her mark on me with her professionalism
and encouragement of my work. Moreover, my thesis work would not have been possible
without the support from Prof. Belcher. It has been an extraordinary experience for me to
work in her lab, too. Her insightful comments and constructive criticisms have helped me
in all the time of my research work. Prof. Karp has offered fresh perspectives on this
work, which have motivated its completion. Prof. Irvine has offered many advices over
the course of the thesis and for that I am grateful as well.
I am lucky to have interactions with three amazing groups of people (the Kamik group,
the VV group and the Belcher group) that I have ever had the opportunity to work with.
In particular, Suman in the Kamik group trained me in nearly every critical area and
offered much help and guidance to me. Chong also provided invaluable assistance for my
research, too. Jongho, Marco, Jong-Min, Sean, Tarun, Mike, Sunandin, Sung-Yong, HuiYing, Minsoung, Mohamed, who joined the Kamik group well after me, have been
valuable friends and colleagues throughout. I also greatly appreciate that the VV group
members always give me their full support. From Belcher group, I especially owe much
to Nimrod, Rana, and Gaelen, who have all actively contributed to my work with
suggestions and feedback. I also thank other Belcher members who have offered much
help to me.
There have been innumerable others at MIT who have affected my life here and I would
like to express my gratitude to them as well: Prof. Karen Gleason, Prof. Sang-Gook Kim,
and Prof. Samuel Allen, Richa, Yin, Jane, Yvonne, Liang-Yi, Yi-Chun, Po-Yen, Alina,
Ting-Yen, Chia-Hou, Alice, Sabrina, Hao-Wei, David, Yu-Chung, Wei-Shan, Lei,
Xiaoting, Hyeongho, Benita, Geetha, and Meri. Moreover, I would like to thank HisnYing, Ivy, Kevin, Ying-Chieh, Jiun-Tai, and Min-Chen for their friendship for years.
Last, but not the least, I would like to specially thank my family: my grandparents, my
parents, Chung-Yi, Wan-Yu, and Yueh-Hsun for their love and support.
5
This page is intentionally left blank.
6
1
Introduction.....................................................................19
1.1
Background and Motivation...........................................................................
19
1.1.1
Surface Marker-Based Cell Separation .............................................
20
1.1.2
Low Affinity Based Separation ........................................................
23
1.1.2.1 C ell Rolling ..........................................................................
24
1.1.2.2 Cell Rolling-Based Separation.............................................28
2
1.2
Scope of the Thesis W ork .............................................................................
31
1.3
Chapter Reference ........................................................................................
32
Free-Flow Travel and Capture of HL60 Cells inside Microfluidic
Channels............................................................................37
2.1
Introduction ....................................................................................................
37
2.2
Part A. Cell Settling inside Microfluidic Channels......................................
40
Fabrication of Substrate and Microfluidic Channel .........................
40
2.2.1.1 M aterials...............................................................................
40
2.2.1
2.2.1.2 Fabrication of Substrate Incorporated PDMS channel.........41
2.2.2
Convection Velocity Experiment ......................................................
41
2.2.2.1 C ell Culture ..........................................................................
41
2.2.2.2 Experim ent Setup .................................................................
41
2.2.2.3 Results and Data Analysis....................................................42
2.3
Part B. Attachment Distance of the Cells in the Microfluidic Channel........46
2.3.1
Fabrication of Substrate and Microfluidic Channel .........................
46
2.3.1.1 M aterials...............................................................................
46
2.3.1.2 Fabrication of Substrate Incorporated PDMS channel.........47
2.3.2
2.3.1.3 Substrate Characterization ...................................................
47
Cell Capture Experiment .................................................................
48
2.3.2.1 Cell Culture and Experiment Setup.......................................48
2.3.2.2 D ata A nalysis ........................................................................
2.3.3
48
Flow Cytometry Analysis.................................................................54
2.4
C onclusion ....................................................................................................
55
2.5
Chapter Reference ........................................................................................
56
7
3 Lateral Displacement of HL60 Cells Rolling on Inclined P-Selectin
Adhesive Bands.....................................................................59
3.1
Introduction ...................................................................................................
59
3.2
Fabrication of P-selectin-Patterned Substrates.............................................
61
3.2.1
M aterials..........................................................................................
61
3.2.2
Fabrication of PDMS Stamp .............................................................
61
3.2.3
Fabrication of Patterned Substrates .................................................
62
3.2.4
Substrate Characterization...............................................................
63
3.3
Cell Rolling on Asymmetric Patterns in a Flow Chamber............................64
3.3.1
C ell C ulture......................................................................................
64
3.3.2
Experim ent Setup ............................................................................
64
3.3.3
Results and Data Analysis ...............................................................
65
3.3.3.1 Effect of Edge Angle on the Rolling Behavior of HL60 Cells 68
3.3.3.2 Effect of Shear Stress on Rolling Behavior of HL60 Cells ..... 70
3.3.3.3 Effect of P-selectin Incubation Concentration on Rolling
Behavior of HL60 Cells ........................................................
72
3.3.3.4 Detachment of Cells Rolling Along an Edge Can be Described
by a Poisson Process. ...........................................................
3.4
73
Optimization of Device Design ....................................................................
80
3.4.1
Effect of Band Width (w) on Device Performance...........................84
3.4.2
Effect of Edge Inclination Angle on Device Performance ...............
3.4.3
Effect of Shear Stress on Device Performance.................................87
86
3.5
C onclusion ......................................................................................................
91
3.6
Chapter
92
........................................................................................
4 Identification of Adhesion Molecules with Weak Affinity via Phage
Display from M13 pVIII Library ..............................................................
97
4.1
Introduction ...................................................................................................
97
4.2
Biopanning of pVIII Library ........................................................................
99
4.2.1
Phage display library of random peptides ........................................
99
4.2.2
Biopanning Procedure ........................................................................
8
100
4.2.3
Phage Titering.....................................................................................102
4.2.3.1 Lysogeny Broth (LB) Medium Preparation............102
4.2.3.2 Tetracyline (Tet) Stock Solution Preparation ........................ 102
4.2.3.3 IPTG/Xgal Stock Solution Preparation..................................102
4.2.3.4 LB/IPTG/Xgal Plates Preparation..........................................102
4.2.3.5 Agarose Top...........................................................................102
4.2.3.6 Titration Procedure.................................................................103
4.2.4
Amplification of M13 bactriophage ...................................................
104
4.2.4.1 PEG/NaCl Stock Solution Preparation...................................104
4.2.4.2 Amplification between Rounds of Panning ...........................
4.2.5
DNA extraction for sequence identification.......................................105
4.2.6
Biopanning Results.............................................................................107
4.3
Immobilization of phage on Au-coated glass slide .........................................
4.4
Substrate Characterization...............................................................................110
4.5
Cell Capture Experiments ...............................................................................
4.6
104
109
112
4 .5.1
C ell Culture.........................................................................................112
4.5.2
Cell Staining .......................................................................................
112
4.5.3
Cell Capture Experiment in Flow Chamber .......................................
113
Cell Capture Efficiency...................................................................................114
114
4.6.1
Phage-engineered Surface ..................................................................
4.6.2
CD4-blocked Phage-engineered Surface............................................116
4.7
Peptide Sequences Characterization of pVIII Library ....................................
117
4 .8
C onclu sion .......................................................................................................
120
4.9
Chapter Reference ...........................................................................................
120
5 Identification of Adhesion Molecules with Weak Affinity via Phage
Display from M 13 pII Library ...............................................................
5.1
Introduction .....................................................................................................
5.2
Biopanning of pIll Library..............................................................................125
5.3
5.2.1
Biopanning Procedure ........................................................................
5.2.2
Biopanning Results.............................................................................127
Enzyme-linked Immunosorbent Assay (ELISA) ............................................
9
123
123
125
130
5.3.1
Surface Density of CD4 on Well Plate as the Target of Biopanning .130
5.3.1.1 Materials.................................................................................130
5.3.2
5.3.1.2 M ethod ...................................................................................
131
5.3.1.3 ELISA Analysis .....................................................................
131
Specificity of the Selected Phage to CD4 Protein ..............................
132
5.3.2.1 M aterials.................................................................................132
5.3.2.2 M ethods..................................................................................132
5.3.2.3 ELISA Analysis .....................................................................
5.4
Specificity of the Synthetic Peptide to CD4 protein .......................................
137
5.4.1
Synthetic peptide ................................................................................
137
5.4.2
Conjugation of Peptide to the Alexa Fluor 488 Dye .......................... 137
5.4.3
Purification of Dye-conjugated Peptides............................................138
5.4.4
Staining the CD4-functionalized Beads with the Dye-conjugated
P eptid es ..............................................................................................
5.4.5
13 9
Characterization of Binding Specificity between Peptide and the
Functionalized Beads .........................................................................
6
134
5.5
Conclusion.......................................................................................................143
5.6
Reference.........................................................................................................143
Conclusion and Future W ork.............................................................145
10
140
List of Figures
Figure 1-1 Multistep leukocyte adhesion cascade in response to inflammation.
Selectins initiate capture and rolling of the leukocytes. Integrins are involved in adhesion
and cause the cells to arrest and transmigrate into the underlying tissues .................... 24
Figure 1-2 Selectins and their major ligands. Three selectin family members (P-, E-,
and L-selectin) and their corresponding ligands are shown in this figure. These three .... 25
Figure 1-3 Force-dependent lifetimes of single bonds of PSGL-1/ P-selectin (blue)
and PSGL-1/ L-selectin (purple) under different shear stress. The catch-to-slip bond
transition under increasing forces has been observed: the lifetime increases (catch bond)
first until it reaches a maximal value and then decreases (slip bond)..........................26
Figure 1-4 Scanning electron microscope (SEM) images of neutrophils. (a)
Localization of PSGL-1 on microvilli. A human neutrophil was incubated with a
mixture of the anti-PSGL-1 antibodies, followed by a rabbit anti-mouse antibody, and
then with protein A-gold of 5 nm diameter. The arrow shows the gold nanoparticles (b)
low magnification of tethered neutrophils. The neutrophil cell has multiple projections
named microvilli on its surface. (c) high magnification of tethered neutrophils. The
neutrophil was rolling on P-selectin under shear stress 2 dyn/cm 2 . The arrow shows a
28
tether. Scare bars are all referred to 1 m. ..................................................................
Figure 1-5 Effect of a receptor pattern on the motion of a rolling cell. Tracks of HL60
cells rolling on patterned P-selectin edge show that patterning can direct the trajectories
of rolling cells along the edges (only long tracks are shown on the right)....................29
Figure 1-6 Schematic diagram of a device for separation of cells. Cells rolling along
band edges are laterally displaced into the adjacent buffer stream, resulting in separation.
Pink lines indicate adhesion molecule-functionalized regions. Red and blue circles are
interacting cells and non-interacting cells, respectively...............................................30
Figure 2-1 Schematic diagrams of cells travelling in a microfluidic channel. (a) The
illustration of a cell entering the microfluidic channel and travelling in both x and y
directions. The settling distance, xs, is defined as how far a cell will travel before it
settles close enough to the bottom surface. (b) Illustration of cells passing the Pselectin-coated area. A cell may be captured from the flowing stream at an attachment
39
distance, /a, after it enters the P-selectin-coated area. ...................................................
Figure 2-2(a) Convection velocity (u,,) of the cells at the different horizontal
position. Convection velocity divided by shear stress decreases and reaches a minimum
value as the cells travel along the channel. Data are presented as the mean and the
standard deviation of the convection velocity obtained from 20 cells under each
experiment condition. Different markers and colors represent different shear stresses. (b)
11
of the cells at different horizontal positions (c) Variation of the convection velocity
of the settled cells with the shear stress. Inset summaries the mean and the standard
deviation of the convection velocity of the cells at position x = 13 cm. The relationship
between u, and T is well described by a linear fit (dashed line). .................................
46
Figure 2-3 Fluorescent images of the substrate. The bright area corresponds to FITCB SA -coated area................................................................................................................47
Figure 2-4 HL60 cells being captured onto the P-selectin-coated area at the shear
stress of 0.25 dyn/cm 2. Some cells were observed to flow out of the field of view........49
Figure 2-5 The attachment distances of HL60 cells at the shear stress of 0.25
dyn/cm 2. P-selectin edge is located at 0 gin. The data obtained by analyzing 65 images
acquired at 1 fps using a customized Matlab code........................................................49
Figure 2-6 Attachment distance of HL60 cells at different shear stresses of 0.25-1.0
dyn/cm 2. Representative results are shown for only one experiment for each shear stress.
(a) Distribution of attachment distance at different shear stresses (0.25-1.0 dyn/cm 2 ). (b)
Cumulative probability (CP) of attachment distance at different shear stresses (0.25-1.0
dyn/cm 2) ..........................................................................................................
51
Figure 3-1 Illustration of a typical cell rolling trajectory along the receptor pattern
inclined an angle (a) to the fluid flow direction. The cell binds within the receptor line,
and rolls in the direction of shear flow toward the pattern edge. The cell then tracks the
edge to define an edge tracking length 4, resulting in a net lateral displacement d, before
detaching to continue along the direction of fluid flow before possible reattachment and
rolling along a new receptor line. Cell rolling velocity v, along within the receptorfunctionalized line in the x-direction of fluid flow can be distinguished from the velocity
Ve along the line edge, where ve,yis lateral velocity (the vertical component perpendicular
to the streamlines and parallel to the lateral displacement, d)........................61
Figure 3-2 Schematic diagram for patterning of P-selectin on a gold substrate
involving microcontact printing. Step 1: Selective deposition of PEG molecules on the
gold surface. Step 2: Filling in of the uncoated surface with P-selectin. .................. 62
Figure 3-3 Characterization of P-selectin patterned substrates. AFM images of 10
pm wide P-selectin lines separated by 15 gm wide PEG bands (after step 2), displaying63
Figure 3-4 Rectangular flow chamber for cell rolling on multiple asymmetric
pattern study. The flow chamber was attached to the P-selectin-patterned substrate via a
vacuum manifold to assemble the device. The thickness of the gasket determines the
height of the chamber, here, height = 0.005 inch)...........................................................65
Figure 3-5 The interaction of HL60 cells with P-selectin/PEG patterned substrate in
the flow chamber. The substrate was prepared using large line-patterned stamp (LS).
The edge inclination angle was 100 and the shear stress was 0.5 dyn/cm 2 . The HL60 cells
only interact with the P-selectin patterned region, not PEG patterned region, which
confirms the anti-fouling property of PEG........................................66
12
Figure 3-6 Tracks of HL60 cells rolling on P-selectin patterns. The tracks (blue lines)
on P-selectin lines (pink) were obtained by analyzing 300 images acquired at 1 fps using
a customized Matlab code. The edge inclination angle was 100 and the shear stress was
0.5 dyn/cm 2 . Inset shows a track corresponding to a cell that first rolled inside the Pselectin line (green) in the direction of fluid flow and then tracked along the edge (black).
................... 68
....... ............................
.........
Figure 3-7 Effect of edge inclination angle a on rolling behavior of HL60 cells at a
fluid shear stress magnitude of 0.5 dyn/cm 2 . Variation of (a) edge tracking length, 1e;
(b) lateral displacement, d; (c) rolling velocities v, and ve within the P-selectin lines and
on the edge, respectively; and (d) lateral velocity, ve,,y (component of the edge rolling
velocity in the direction of lateral displacement). Error bars represent one standard
deviation, with n = 3 replicate experiments for each condition................................70
Figure 3-8 Effect of shear stress T on rolling behavior of HL60 cells at an edge
inclination angle of 50. Variation of (a) edge tracking length, le; (b) lateral displacement,
d; (c) rolling velocities v, and ve within the P-selectin lines and on the edge, respectively;
and (d) lateral velocity, ve,,y. The effect of shear stress on rolling behavior of HL60 cells is
not statistically significant as observed in (a), (b), (c) and (d). Error bars represent one
standard deviation, where n = 3 replicate experiments for each condition...................71
Figure 3-9 Effect of P-selectin incubation concentration on rolling behavior of HL60
cells. Variation of (a) edge tracking length, le; (b) lateral displacement, d; (c) rolling
velocities v, and ve within the P-selectin lines and on the edge, respectively; and (d)
lateral velocity, vey at edge inclination angles a = 100 and 20' and shear stress of 0.5
dyn/cm2 . Error bars represent one standard deviation, where n = 3 replicate experiments
.......... 73
for each condition................... ................................
Figure 3-10 Detachment of cells rolling along an edge is well described by a Poisson
distribution. (a) Cumulative distribution function of edge tracking lengths le (filled
triangles) was fitted to a Poisson distribution described by Eq. 3-1 (solid lines). Insets
show the frequency distribution of the experimentally measured edge tracking lengths,
along with that predicted by the Poisson distribution fit to the CDF (solid lines). Colors
indicate different inclination angles a of the receptor pattern. Representative results are
shown for only one experiment for each a. (b) Variation of the average value of k with
edge inclination angle is well described by a linear fit on a semi-log plot (solid line). (c)
Variation with the edge inclination angle of the average value of the lateral displacement
(solid line) obtained from the empirical fit shown in (b) along with the experimental
results (open circles). Error bars in (b) and (c) represent one standard deviation. Shear
stress is 0.5 dyn/cm 2 . . . ............................ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Figure 3-11 Prediction of cell trajectories on a receptor-patterned substrate (a)
Probability distribution of net lateral displacement of HL-60 cells after rolling on three
consecutive bands of P-selectin patterns as obtained from Monte Carlo simulations
(shaded area) and experimental observations. a = 200 (shear stress 0.5 dyn/cm 2 ). (b)
Prediction of the downstream distribution of HL60 cells rolling on patterned P-selectin in
79
a separation device. ..............................................................
13
Figure 3-12 Optimization device design. (a) Schematic diagram of cell settling,
capture, rolling on the patterns. (b) Schematic diagram of a device. Device capture
efficiency, total lateral displacement, and required chamber length can be modeled based
on different device design parameters..........................................................................
81
Figure 3-13 (a) Schematic diagram showing cell hopping onto downstream band. (b)
Variation of effective chamber length on total lateral displacement when hopping
events were (p, = 1.0) and were not (p, = pf = 0.03) considered for the pattern design
of a = 150, w = 10 tm, g = 10 pIm, and r= 0.25 dyn/cm2
.... . . . . . . .
.....
84
Figure 3-14 The relationship between device capture efficiency (Pd) and effective
chamber length (xe) and the relationship between total lateral displacement (d,) and
effective chamber length (xe) when w varies from 10 ptm to 50 pIm with a = 15*, g =
10 stm, and r = 0.25 dyn/cm2 ........
. .. ... ....... ........ ...... .. ..... .... . . . .85
Figure 3-15 The relationship between device capture efficiency (Pd) and effective
chamber length (xe) and the relationship between total lateral displacement (d) and
effective chamber length (xe) when a varies from 5 0 to 250 with w = g =10 pim, and r
. . . . . . . . . . . . . . . . .. . .
. . . . .. . . . . . . .
. . . . . . . . . . 87
= 0.25 dyn/cm 2 . . . . .
Figure 3-16 The relationship between device capture efficiency (Pd) and effective
chamber length (xe) and the relationship between total lateral displacement (d) and
effective chamber length (xe) when i-varies from 0.25 dyn/cm 2 to 1.0 dyn/cm 2 with a
= 15 0 and w = g =10 pm.................................................89
Figure 3-17 Device optimization parameters including edge inclination angle a,
band width w, gap width g, and shear stress T, for a miniaturized device of a
chamber length of L, with device capture efficiency Pd, and total lateral
displacement d,. x, is settling distance and xe is effective chamber length. ...........
91
Figure 4-1 Schematic diagram of M13 bacteriophage and the summary of coat
protein....................................................
.......... ...................... ......................
... 98
Figure 4-2 Schematic diagram for identifying peptide motifs of affinity to CD4
protein by biopanning using pVIII library. The pVIII library was exposed to the CD4coated well plate. Unbound phage was then washed off from the target and the bound
phage was eluted from the surface by acid solution. The eluted phage was amplified to
serve as the sub-library for the next round of the panning process. The peptide sequences
of the eluted and amplified phage was identified by titering and DNA sequencing. The
panning process was repeated until the consensus peptides were found. ....................... 101
Figure 4-3 Titration of phage on LB/IPTG/Xgal plates (dilution range is from 101 to
104). The phage was titered on LB/IPTF/Xgal plates and the blue plaque was then
randomly selected for sequencing.....................................
..................................... 104
Figure 4-4 Schematic diagram of DNA extraction for sequence identification. DNAs
of the phage was extracted from the bacteria. The bacteria were lysed and the DNAs were
isolated from the lysed bacteria and then eluted using QlAprep Spin Miniprep Kit....106
14
Figure 4-5 Schematic diagram for immobilizing phage on a gold substrate. Step 1:
Functionalization of DSP onto the gold surface. Step 2: Immobilization of phage on the
reactive gold surface...............................................................110
Figure 4-6 Surface characterization of phage-engineered substrates. AFM height
images of phage immobilized on the gold-coated substrate. All the images are 1 Om X
1Opm....................................................................................-----------------.......................112
Figure 4-7 Fluorescent microscope images of the cells on the RK1-engineered gold
substrate (A) before the flow was started (B) after the flow was started for 60 s. The
white dots are stained K562 cells and black dots are non-stained SUPTI cells.............114
Figure 4-8 Capture efficiencies of RK1-RK12-immobilized substrates at shear stress
of 0.1 dyn/cm 2. Error bars represent one standard deviation, where n = 4 replicate
experiments for each condition......................................................115
Figure 4-9 Capture efficiencies of RK1-immobilized substrates. The effect of
different shear stress on the capture efficiencies of the cells on the RK1-immobilized
substrates. Error bars represent one standard deviation, where n = 4 replicate experiments
for each condition................................................116
Figure 4-10 Microscope images of the SUPT1 cells captured on the CD4-blocked
RK1-engineered gold substrate (a) before the flow was started (b) after the flow was
started for 60 sec..............................................................................117
Figure 5-1 (a) CD4 protein physically adsorbed on regular polystyrene surface. (b)
His-tagged CD4 protein immobilized on Ni (II)-NTA-coated surface with specific
orientation.....................................................-------------------------------------------------------------..124
Figure 5-2 Schematic diagram for identifying peptide motifs of affinity to CD4 by
biopanning using pIII library. The pIII library was exposed to the CD4-coated well
plate. Unbound phage was then washed off from the target and the bound phage was.. 126
Figure 5-3 Schematic diagram of ELISA procedure. CD4-His protein was
immobilized on Ni (II) coated well plate. The phage was then exposed to the CD4-coated
well plate. HRP-conjugated anti-M13 antibody binds specifically to the target (pVIII
protein of the phage). Several washing steps were repeated between each ELISA step to
remove unbound materials. Substrate (TMB) was added and the colored end product
from enzyme-substrate reaction is produced. The color signal correlates to the amount of
... 134
the phage present in the well plate. .....................................................................
Figure 5-4 ELISA assay output. (a) Qualitative results. Representative results are
shown for only one experiment at each condition. Wells in column 1, 3, and 5 were
coated with CD4 and then BSA. Wells in column 2, 4, and 6 were coated with only BSA
as control. Phage with different input concentration were added. Phage AMB1 were
incubated in column 1 and 2. Phage AMB2 were incubated in column 3 and 4. WT phage
was incubated in column 5 and 6; (b) Quantitative results. The relationship between the
absorbance of the colored end product and the input concentration of the phage. Error
15
bars represent one standard deviation, with n = 3 replicate experiments for each
condition.......................--...........................-- - -. -----...............................................
136
Figure 5-5 Reaction of the Alexa Fluor@ 488 C5 Maleimide with a thiol-peptide. .137
Figure 5-6 Collected drained samples. The first five fractions were collected before the
dye-conjugated peptide solution drained through and thus contained mostly buffer
(colorless). The orange color came from dye-conjugated peptide. The non-conjugated dye
shows yellow color
.......
- . ....
...
-........ ...............................................
138
Figure 5-7 Schematic diagram for staining the CD4-coated beads with dyeconjugated peptide. The Ni(II) beads were incubated with CD4-His and AF 647 conjugated BSA, separately. The beads were then mixed and incubated with BSA to
block non-specific interaction. The mixed CD4 and AF 647-B SA-coated beads were then
incubated with AF 488-conjugated peptide. Several washing steps were repeated between
each step to remove unbound materials before the samples were imaged......................140
Figure 5-8 Fluorescence microscope analysis. (a-b) CD4-immobilized beads. Images
of (a) bright field and (b) FITC channel. CD4-immobilized beads do not show
autofluorescence. (c-d) BSA-coated beads that were incubated with AF 488 dyeconjugated peptides. Images of (c) bright field and (d) FITC channel. AF 488 dyeconjugated peptides do not show non-specific binding to BSA-coated beads. (e-f) The
mixture of the CD4-immobilized beads and the AF 647 dye-BSA-coated beads
stained with AF 488 dye-conjugated peptides. Images of (e) bright field; (f) Cy5
channel; and (g) FITC channel; (h) is the Cy5/FITC channel merged image. AF 488 dye
shows bright, green fluorescence and AF 647 shows bright, red fluorescence. Beads
number 1, 2, 6, and 11 were coated with BSA and beads number 3, 4, 5, 7, 8, 9, and 10
........................ ..............---------------............................
.....................
142
Figure 6-1 Experimental design for characterizing hopping behavior of cells. Cells
flow over parallel receptor bands with different gap widths aligned perpendicular to flow
direction. The minimum gap width without making cell cross over the patterns can be
identified.................
................
.............
.......
..... .. 147
16
List of Tables
Table 1-1 Kinetics and affinity of selectin-ligand and antibody-antigen interactions
.................................................... 23
.....................................
Table 2-1 Comparison of the average values and standard deviation of A and <Ia>
2
obtained from curve fitting at different shear stresses of 0.25-1.0 dyn/cm . n = 3
replicate experiments. Data were obtained by fitting to two empirical equations with
51
experimentally measured attachment distance .............................................................
Table 3-1 Comparison of the experimental average of 1e, Poisson average value of k,
and the Poisson mean value ' from the empirical fit. n = 3 replicate experiments.... 76
Table 3-2 Prediction of required effective chamber lengths (xe). The range of xe
which target at Pd - 90%, dt : 850 ptm and both criteria (Pd > 90% and d, ? 850 Pim)2
when w varies from 10 pm to 50 pin with a = 150, g = 10 ptm, and T = 0.25 dyn/cm .
...............................................
........
.......................------------
86
Table 3-3 Prediction of required effective chamber lengths (xe). The range of xe
which target at Pd ! 90%, d, ? 850 ptm and both criteria (Pd > 90% and d, f 850 pim)
2 ... . ..... . . ... .... 87
when a varies from 5 0 to 254 with w = g =10 pim, and r =0.25 dyn/cm
Table 3-4 Prediction of required effective chamber lengths (xe). The range of xe
which target at Pd !90%, dt ? 850 gim and both criteria (Pd > 90% and d, ? 850 stm)
when r varies from 0.25 dyn/cm 2 to 1.0 dyn/cm 2 with a = 151 and w = g =10 pm.....89
Table 4-1 Input and output concentration of phage and the recovery of each round
of screening. The concentration is defined as the number of the phage per well. For each
round of screening, four wells coated with CD4 were used...................................107
Table 4-2 CD4 protein binding sequences from biopanning of pVIII library.........107
Table 4-3 Selected peptide sequences and their properties. (negatively charged amino
acids in red; positively charged amino acids in blue; polar amino acid without charged in
yellow; hydrophobic amino acid in gray; aromatic amino acids in green; Tyrosine in
109
...---------------------------............................
............................
orange)................................
Table 4-4 Peptide sequences from biopanning of pVIII library. (96 samples were
o........................117
----------------------..................
selected)..................
Table 5-1 Binding affinity (dissociation constant (KD) of oligo-histidine peptides to
-- - -- -.... 125
Ni(II)....................................................................---
17
Table 5-2 Input and output concentration of phage and the recovery of each round
of screening. The concentration is defined as the number of the phage per well. For each
round of screening, four wells coated with CD4 were used............................................127
Table 5-3 CD4 protein binding sequences from biopanning of pIII library............128
Table 5-4 Selected peptide sequences and their properties. (negatively charged amino
acids in red; positively charged amino acids in blue; polar amino acid without charged in
yellow; hydrophobic amino acid in gray; aromatic amino acid in green; Histidine in
orange).............................................................................................................................130
18
1
1.1
Introduction
Background and Motivation
Separation and analysis of heterogeneous cell populations are important for diagnostic
and therapeutic applications, and for elucidating the biology of rare cell types. Numerous
cell-sorting techniques have been developed, with sorting characteristics based on
physical properties of the cells, such as size, density, and deformability [1-4]. However,
these physical separation methods may not be sufficiently specific enough to distinguish
similar cell types. Surface maker-based cell separation/analysis techniques allow cells to
be distinguished via biochemical differences (protein expression), and can confer higher
specificity. Some standard techniques include fluorescence-activated cell sorting (FACS),
magnetic-activated cell sorting (MACS), and cell affinity chromatography (CAC).
Among these, FACS or flow cytometry has been the gold standard of cell separation for
years. However, sophisticated equipment, and time-intensive and labor-intensive sample
preparation processes limit the use of most methods. For example, white blood cell
counts and leukocyte differential are routinely used in clinical diagnostics [5]. Blood
counts are usually performed using automated counters to identify major blood cell types
and separation of subgroups of immune cells are done through FACS [5]; however, flow
cytometry is typically not available in the situation with resource-limited settings. Also,
transportation of clinical samples from patients to labs and multiple sample processing
steps are time-consuming and labor-intensive. Microfluidic devices offer a promising
platform for cell separation in which all of the processes can be compacted in one single
step. Small amount of samples required and portability of microfluidic devices facilitate
the application in point-of-care testing. However, most of surface-marker-based
microfluidic devices use antibodies to capture cells with specific antigens and therefore
separate cells. The antibodies with high affinity to their antigen limit the usability of
these microfluidic devices since retrieval of the captured cells is difficult. Therefore,
there exists a need for developing techniques of rapid cell separation based on specific
surface markers but weak adhesion which will allow cells to be retrieved and will be
useful for many applications. A number of molecules in nature exhibit weak, but
relatively specific adhesive interactions. For example, selectins mediate leukocyte rolling
19
which is a process in inflammatory or immune responses [6-9]. Selectin molecules-based
separation has been studied before; however, it was limited to a batch process [10].
Karnik et al have demonstrated that patterning of selectin molecules inside the
microfluidic device create the possibility of continuous sorting of cells by transient, weak
interaction between cell and patterned selectin molecules [11]. Comprehensive studies of
cell settling, cell attachment, pattern geometry, flow condition, and selection of adhesion
molecules are needed in order to design and realize these devices.
1.1.1
Surface Marker-Based Cell Separation
Surface markers are typical proteins expressed on the surface of cells. Cluster of
differentiation (CD) molecules are a group of surface markers mostly expressed on
leukocytes and associated with immune function or properties. The cells can be defined
based on which CD molecules expressed on their surface. Using multiple CD markers
allows the cells to be defined more specifically. For most of existing surface markerbased cell sorting techniques, the surface cluster of differentiation proteins (antigens) are
identified via binding of conjugate proteins (antibodies). An important advantage of
surface marker-based cell separation is that it can be used for sorting the cell populations
of the same size or density. For example, CD4 and CD8 proteins are well known surface
markers for helper T lymphocytes
(T cells) and suppressor/cytotoxic T cells,
respectively[12]. The diagnosis of HIV disease relies on the efficient separation of human
CD4+ T lymphocytes (T cells) from whole blood. The most general way to distinguish
CD4+ T cells and CD8+ T cells is to identify them by antibodies that can specifically
bind to CD4 or CD8 proteins expressed on the T cells. The antibodies can be conjugated
to fluorescent molecules or magnetic beads, or can be immobilized onto columns or
microchips. The methodologies or operation procedures will differ, depending on the
materials to which the antibodies are conjugated or immobilized. The most common
surface marker-based technologies are: fluorescence-activated cell sorting (FACS),
magnetic-activated cell sorting (MACS), cell affinity chromatography (CAC), and
antibody-based microfluidic devices (ABM) [1-3, 13], which are outlined below.
20
Fluorescence-Activated
Cell Sorting (FACS). FACS is the gold standard cell
subpopulation separation technique [14, 15]: Typically, specific cell surface markers of
the cells labeled by antibodies. Those antibodies are labeled with fluorescent molecules
and the cells are then identified by fluorescent intensity corresponding to the number of
antigens on each cell. By collecting the information such as fluorescent intensity of each
cell, the machine can identify and then separate cells of different antigen surface density.
FACS has several technical advantages, such as single-cell-level sensitivity, high
throughput (~10,000 cells/sec) of separation/counting, and ability to track four to five
different cell surface markers at the same time [15]. However, the capital investment and
maintenance of equipment, and the multiple sample preparation steps, hinder the
applications
in point-of-care tests or in resource-limited
settings environment.
Furthermore, cell viability after ejection from the machine, as well as the potential need
to remove antibody labels before human use of such sorted cells, remain major issues,
that limit the use of FACS in clinical or translational applications [14, 15].
Magnetic-Activated Cell Sorting (MACS). Similar to FACS, in MACS, the surface
markers of the cells are recognized by the antibodies. These antibodies are coated onto
the surface of magnetic nanoparticles [16]. The cells can be sorted by applying a strong
magnetic field that collects bead-bound cells of higher antigen density or affinity.
Compared with FACS, by which the cells are sorted one by one, MACS can separate the
cells in serial (continuous mode) or in parallel (batch mode), resulting in higher
throughput (more than 1011 cells per hour [2, 13]). Another advantage is that cells can be
sorted positively or negatively depending on whether targeted or nontargeted cells are
labeled. However, there are several drawbacks for this method, such as the timeconsuming, sensitive process of labeling cells onto the antibodies-coated magnetic beads,
complications of removing magnetic labels, and difficulty of sorting more than two cell
populations at the same time.
Cellular Affinity Chromatography (CAC). Affinity chromatography has been widely
used for separating biological molecules, such as peptides, proteins and enzymes [17-22].
Affinity chromatography can also be applied in cell separation by using different column
21
design [23-26]. Cells can be sorted by being passed via fluid flow through a column
packed with antibody-coated beads or other solid supports. Cells with affinity to the
immobilized antibody will be to be selectively retained; other cells will be passed through
unretarded. One advantage of CAC is that the sample preparation is simple. For example,
there are no complicated processes of labeling cells with fluorescent dye or magnetic
beads [1]. However, it is often difficult to release the bound cells from the beads; elution
is typically achieved by applying high shear flow or adding competitive agents [27].
Adhesion-based Microfluidics Devices for Cell Sorting (ABM). Similar to CAC,
sorting of the cells by adhesion-based microfluidics devices is achieved by binding cells
of interest onto the devices where the specific molecules are immobilized. The adhesive
molecules (usually antibodies) can recognize the surface markers present on the surface
of the cells. The separation can be completed in a single processing step. The untargeted
cells are allowed to flow through without interacting with the adhesive molecules, while
the targeted cells are captured inside the chip. The sensitivity of this method is enhanced
because of the larger surface area to volume ratio in a microscale chip than in macroscale
setup. Moreover, the residence time inside the microchips is shorter than that of CAC.
The low-cost microchips provide advantages of flexibility for point-of-care diagnostics.
However, the difficulty in retrieving the captured cells makes analysis and enumeration
sorted cells difficult.
Most major surface marker-based cell separation techniques involve using antibodies that
target corresponding antigens expressed on the cell surfaces. However, removal of the
labels without changing cellular functions and phenotypes is one of the major issues that
hinder the use of these methods. Additionally, these methods are then limited to isolation
of cells with distinct antigens and for which purified antibodies are readily available.
There exists a growing need for technologies that do not require modification or labeling
of the sorted cells. Receptors with high specificity but weak affinity resulting in transient
interaction with conjugate molecules on the cells may be used to fulfill the requirement.
22
1.1.2
Low Affinity Based Separation
It has been widely studied that typical antibody-antigen interactions show high affinity
where on-rate (k,,) is in the rage of 10'~108 M-'s-1 , off-rate (korr) is in the rage of 102
_10-4 s-1 and dissociation constant (Kd) is in the range of 0.3~1.5 pM [28-30]. The high
affinity property results in firm adhesion between antibody and antigen, which limits the
usability of these microfluidic devices since retrieval of the captured cells is difficult. In
order to develop microfluidic devices that allow cells to be retrieved, adhesion molecules
with weak but specific interaction is needed. In nature, a number of molecules exhibit
unique kinetics and affinity that allows the cells to be specifically identified, but weakly
captured. For example, selectins and their ligands mediate inflammatory or immune
responses [6-9] and allow leukocytes to be captured from the blood stream and roll on the
vascular endothelium. There are three types of selectins: (1) P-selectin (CD62P), (2) Eselectin (CD62E), and (3) L-selectin (CD62L) [6, 31-33]. Table 1-1 lists kinetics and
affinities of selectin-ligand and antibody-antigen interactions [34-36]. Compared with
typical antibody-antigen interaction, the moderate ko. and high koff of the selectin-ligand
interactions suggest that the adhesive bond can be formed and be broken quickly and
therefore the cells of ligand molecules can roll on the receptor molecules. Using this
weak but specific adhesion opens possibility of separating cells without labeling and
allowing cells to be retrieved.
Table 1-1 Kinetics and affinity of selectin-ligand and antibody-antigen interactions
Receptor and Ligand
kon(M
1
s1)
koff(s-)
Ka(M
1
Kd(PM)
)
6
Reference
P-selectin /PSGL1
(0.93 - 4.4)x10 6
1.4
L-selectin /PSGL1
2x10 5
10
2x10 4
50
[35]
E-selectin /ESL1
7.4x10 4
4.6
1.6x10 4
62
[36]
Antibody/Antigen
10 ~ 108
10-2 10-4
23
0.66-3.1x10
107
_ 1012
1.5
0.3
10-6
101
[34, 35]
[28-30]
1.1.2.1
Cell Rolling
Cell rolling is mainly observed as one of the steps in recruiting leukocytes from
circulating blood into the sites of inflammation, as shown in Figure 1-1 [37]. Rolling is
initiated first, followed by integrin-dependent adhesion and transmigration to the site of
inflammation. Cell rolling involves transient interactions between the leukocyte and the
vascular endothelium (blood vessel lining) under fluid shear in a physiological process.
The interaction is mediated by glycoprotein receptors known as selectins and their
ligands. Rolling mechanism also plays an important role in homing of hematopoietic
stem cells and metastasis of tumor cells [6-9].
Capture
Rolng
salyl Lewis X
Integrins ITuimin
Selectins
ICAM-1
VCAM-
PECAM
VAP-
Figure 1-1 Multistep leukocyte adhesion cascade in response to inflammation.
Selectins initiate capture and rolling of the leukocytes. Integrins are involved in adhesion
and cause the cells to arrest and transmigrate into the underlying tissues [37].
Selectin Family. Selectins including P-selectin, E-selectin and L-selectin and their
ligands mediate leukocyte rolling interaction [6, 31-33]. P-selectin molecules are
expressed by endothelial cells and platelets [38]. E-selectin molecules are expressed on
endothelial cells and L-selectin molecules are expressed on leukocytes. Figure 1-2 lists
selectins and their major ligands [6]. The major motifs of selectin molecules are: Ca+2_
dependent lectin domain, epidermal growth factor (EGF) like domain, short consensus
repeat domain (SCR), transmembrane motif, and a short intracellular cytoplasmic tail as
shown in Figure 1-2. Each selectin molecule mediates adhesion of rolling cells through
the interaction of its lectin domain with sialyl Lewis x (sLex) ligand in the presence of
extracellular Ca . Several posttranslational modifications (glycosylation and sulfation)
24
of the sLex motif have been identified for being able to preferentially bind to specific
type of selectin [31]. For example, attachment of sLeX to the O-linked glycan of the Pselectin glycoprotein ligand-1 (PSGL-1) and the sulfation of N-terminal tyrosine in
PSGL-1 are required for P-selectin binding [39].
Aeftated Plaftiet
L
--
P-selecin
TELOkct
ewPSGL-1
Actb*a&W ondotheftl eN
LeuM"cye
PSGL-1
P-selectin
CD4
E-selactin -
Un~
F
%ESL-1
I
WicllLu
At
.1aiW
GIyCAM-1
PNAd
4m4
-
MoYte
L-eelectln
CD34
L.Podocalyxin Laeukocy"
L-eFeadn
==4PSGL-1
Lectin domain OEGF domain COnSMnSuS rmpt
E GNAc M GNAC @ Gal A Fuc * SIS SSO
fNr-9ycan
I O-ycan
Figure 1-2 Selectins and their major ligands. Three selectin family members (P-, E-,
and L-selectin) and their corresponding ligands are shown in this figure. These three
25
selectin molecules share similar structure: N-terminal Ca+2-dependent lectin domain,
epidermal growth factor like domain, short consensus repeat domain (SCR),
transmembrane domain, and a short intracellular cytoplasmic tail (last two parts are not
highlighted here). CD, cluster designation; EGF, epidermal growth factor; ESL-1, Eselectin ligand-1; GlyCAM-1, glycosylated cell adhesion molecule-1; GalNAc, Nacetylglucosamine; Gal, galactose; Fuc, fucose; Sia, sialic acid. [61
The force exerted from the flow on the cells may affect the lifetimes of selectin-ligand
bonds by changing their off-rates. Bond lifetime of a receptor-ligand bond may increase
(catch bonds) or decrease (slip bonds) [6, 40-45]. Lifetime is equal to the reciprocal of
off-rate, l1/kff. The catch-to-slip bond transition under increasing forces has been
observed experimentally for PSGL-l/ P-selectin and PSGL-l/ L-selectin as shown in
Figure 1-3 [6, 40, 41, 46]. The external force has a biphasic effect on selectin-ligand
interactions: the lifetime increases (catch bond) first until it reaches a maximal value and
then decreases (slip bond). The catch bond behavior was observed over wider range of
external force for L-selectin (up to 50 pN per bond) compared with PSGL-l/ P-selectin
(up to 20 pN per bond) [40, 41]. In contrast, only slip bonds were observed for
selectin/antibody interaction under external force [40, 41].
05
0.40.4 -
E
P-selectln/PSGL-1
*#
03-
0.2
L-selectin/PSGL-1
0.05 -
0
0.4
0.8
1.2
1.
2
Wall shear stress (dyn cm- 2 )
Figure 1-3 Force-dependent lifetimes of single bonds of PSGL-1/ P-selectin (blue)
and PSGL-1/ L-selectin (purple) under different shear stress. The catch-to-slip bond
transition under increasing forces has been observed: the lifetime increases (catch bond)
first until it reaches a maximal value and then decreases (slip bond). [61
26
Contributions to Rolling Motion. Cell rolling is a complex phenomenon involving
multiple mechanical forces ranging from nanoscale to microscale [46]. For example,
kinetics of receptor and ligand interaction [47], clustering of receptors [48, 49],
elongation of microvilli [47, 48, 50-52], deformation of the cells [53],
and the fluid
forces acting on the cells [54, 55] can all perturb cell adhesion durations.
Cellular properties that modulate cell rolling. It has been demonstrated that selectin(or selectin ligand-) coated microspheres can roll on the substrate immobilized with
selectin ligand (or selectin). As those microspheres do not exhibit other biological
features of cell rolling (microvilli extension, cell deformability and ligand cluster which
will be discussed later), this suggests that molecular properties of selectin and its ligand
are sufficient enough for supporting rolling [56-58]. We note that, compared to such
beads, cells are deformable. Deformability of the cells enlarges the contact area between
the bottom of the rolling cells and the endothelial cells, which results in a higher number
of selectin-ligand bonds [59]. Additionally, the surfaces of cells are highly irregular. As
compared to polymeric beads, for example, neutrophils (the most abundant leukocytes in
the peripheral blood of human) have multiple microvilli and it has been observed that
PSGL-1 and L-selectin are localized on the tip of microvilli on neutrophils as shown in
(Figure 1-4(a)) [60, 61]. The bond cluster on the microvilli enhances tethering by
increasing the contact with the receptors on the endothelial cells [62]. Tethers extend
under shear flow by stretching microvilli (Figure 1-4(b) and (c)) [48] and the force on
selectin-ligand bonds can be reduced by stretching tethered mivrovilli [63]. Most of the
tethers retract after the selectin-ligand bond breaks [50].
27
Figure 1-4 Scanning electron microscope (SEM) images of neutrophils. (a)
Localization of PSGL-1 on microvilli. A human neutrophil was incubated with a
mixture of the anti-PSGL-1 antibodies, followed by a rabbit anti-mouse antibody, and
then with protein A-gold of 5 nm diameter. The arrow shows the gold nanoparticles (b)
low magnification of tethered neutrophils. The neutrophil cell has multiple projections
named microvilli on its surface. (c) high magnification of tethered neutrophils. The
neutrophil was rolling on P-selectin under shear stress 2 dyn/cm 2 . The arrow shows a
tether. Scare bars are all referred to 1 pim. (Images are adapted from [48], and [60].)
1.1.2.2
Cell Rolling-Based Separation
Several groups have attempted to mimic physiological process of leukocyte rolling to
develop cell separation techniques [64-70].
In contrast to conventional high affinity-
antibody-based methods, separation based on immobilized receptors involved in cell
rolling obviates labeling and label removal steps, enabling facile elution with minimized
alteration of viability or function post-sorting [64-66]. For example, Chang et al. have
demonstrated E-selectin-coated micropillar arrays for partially fractionating two cell
types (HL60 and U937) based on the difference in transient interactions with the
receptors [64]. Choi at al. also demonstrated that HL60 cells can be separation from
K562 cells by using three dimensional P-selectin-coated inclined microstructure which
promotes rolling of HL60 cells in the lateral direction [68]. Karnik et al. have
demonstrated the transient interaction of HL60 cells and an inclined P-selectin band as
shown in Figure 1-5 [11]. The P-selectin molecules were coated on the substate where the
28
edge of P-selectin area was inclined. The rolling cells can be displaced orthogonally to
the direction of the flow, which opens new possibilities for separation of cells by
diverting the rolling direction of cells.
(a)
(b)j
Figure 1-5 Effect of a receptor pattern on the motion of a rolling cell. Tracks of HL60
cells rolling on patterned P-selectin edge show that patterning can direct the trajectories
of rolling cells along the edges (only long tracks are shown on the right). [11]
Figure 1-6 shows the concept of a device that adapts this observation, based on
controlling the trajectories of the rolling cells along the edges of multiple receptor bands
[70]. On the bottom of the device, multiple inclined, asymmetric adhesive bands are
coated. The interacting cells from the continuously flowing sample can be captured on
the adhesive bands and then roll on them. The asymmetry of the band orientation with
respect to the flow direction alters the trajectory of the interacting cells, such that they
track along the edge of the band, get displaced into the buffer stream and get sorted out.
The non-interacting continue with the flow trajectory and stay in the sample stream. Such
devices design with biomimetic adhesive substrates enables cell separation in a simple,
continuous-flow, label-free manner. To realize the devices, a comprehensive study of
interaction of cell and substrate is required. Cell settling and attachment onto substrates,
the effect of substrate parameters on separation potential, and selection of adhesion
molecules for the substrates will be discussed in this thesis later in order to maximize
separation potential of cell rolling microfluidic devices.
29
Buffer
Cells
#Sortedeeel
(interacting cells)
Aoo
Cells
Adhesive substrate
%ee
**:Other Cells
(non interacting cells)
Figure 1-6 Schematic diagram of a device for separation of cells. Cells rolling along
band edges are laterally displaced into the adjacent buffer stream, resulting in separation.
Pink lines indicate adhesion molecule-functionalized regions. Red and blue circles are
interacting cells and non-interacting cells, respectively.
Since cell rolling phenomenon is exhibited in vivo or in vitro by several types of cells
including leukocytes, hematopoietic stem cells, and some cancer cells [6-9], those types
of cells have potential to be separated by steering them on biomimetic adhesive bands
using the device as described in Figure 1-6. Such devices can be suitable for point-of care
applications, especially in non-industrialized settings, which require device simplicity
and minimized cell processing steps. Furthermore, continuous flow sorting can promote
separation efficiency and selectivity. Another advantage is that observation of cellsurface interactions on immobilized receptors offers the opportunity for label-free
enumeration of cells [71]. Separation methods for therapeutics or subsequent analysis of
cells need to preserve the cell phenotype and such devices allow invasive separation that
can promise for these applications. Cell rolling separation devices with multiple adhesive
bands also offer potential for direct analysis of the cell phenotype or receptor expression.
In addition, co-immobilization of other cell type-specific molecules along with selectins
can significantly influence cell rolling [72, 73], which opens the possibility of further
tuning the specificity of separation and analysis based on the expression of a pertinent
receptor on the cell. On the other hand, although directing rolling trajectories of cells by
patterning of weak adhesion molecules onto the microfluidic device is promising for
label-free cell sorting, this technology is limited to the molecules known for enabling cell
rolling. To apply this technique to general surface markers, new receptors other than the
existing molecules need to be identified. Biopanning is an affinity selection technique
30
used for identifying peptides that bind to a target of affinity within micromolar range
[74], which potentially can be used to identify new receptor with weak adhesion to the
cells. The possibility of using biopanning to select receptors for cell separation will be
studied in this thesis.
1.2
Scope of the Thesis Work
To successfully develop rolling adhesion microfluidic devices that can deflect cells by
inclined receptor bands as shown in Figure 1-6, a biomimetic adhesive substrate is the
key. Therefore, the goal to this thesis work is to develop biomimetic adhesive substrates
for cell separation in microfluidic devices. My three main aims are:
-
Aim 1: To understand transport and attachment of cells to biomimetic adhesive
substratesin microfluidic channels.
Chapter II discusses how the cells settle and how they are captured by receptor
functionalized surface under fluid shear flow. In Part A, to quantify cell settling,
settling distance under different shear stresses is characterized from the study of
convection velocity of cells at different position along the channel. In part B, to
quantify cell capture, cumulative probabilities of cell attachment distance and average
attachment distance of cells at different shear stresses is analyzed. An empirical
model is developed to predict capture probability by an inclined receptor band.
-
Aim 2: To investigate effects of biomimetic adhesive substrate parameters on cell
separationpotential
In Chapter III, a patterning method involving microcontact printing is developed to
study how cell transport can be controlled on inclined receptor bands. The effects of
edge inclination angle with respect to shear flow direction (a = 50, 100, 150 and 20'),
shear stress magnitude (r = 0.5, 1, 1.5 and 2.0 dyn/cm 2 ), and P-selectin incubation
concentration (15, 20, and 25 ptg/mL) are quantified in terms of the edge tracking
length, lateral displacement, and rolling velocity. A methodology is developed to
predict device performance to enabled successful design of separation device.
31
-
Aim 3: to extend this approach to other and novel cell surface markers
Beyond the PSGL-1 on leukocytes, this adhesive band patterning approach may be
extensible to tether cell surface targets, such as CD4 proteins, that is indicative of
disease state. M13 pVIII (Chapter IV) and pIll (Chapter V) phage libraries are used
for selecting peptides with affinity to CD4 proteins. In Chapter IV, selected phage is
immobilized on the gold surface and capture efficiency of CD4+ cells is
characterized. The interaction between selectin phage and CD4+ cells is demonstrated
to be CD4-dependent. In Chapter V, specificity of selected phage clone and synthetic
peptides to CD4 proteins is confirmed.
1.3 Chapter Reference
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
M. Radisic, R. K. Iyer, and S. K. Murthy, Micro- and Nanotechnology in Cell
Separation.Int J Nanomedicine, 2006. 1(1): p. 3.
A. A. Bhagat, H. Bow, H. W. Hou, S. J. Tan, J. Han, and C. T. Lim, Microfluidics
for Cell Separation. Med Biol Eng Comput, 2010. 48(10): p. 999.
M. J. Tomlinson, S. Tomlinson, X. B. Yang, and J. Kirkham, Cell Separation:
Terminology and Practical Considerations. J Tissue Eng, 2013. 4: p.
2041731412472690.
D. R. Gossett, W. M. Weaver, A. J. Mach, S. C. Hur, H. T. Tse, W. Lee, H.
Amini, and D. Di Carlo, Label-Free Cell Separation and Sorting in Microfluidic
Systems. Anal Bioanal Chem, 2010. 397(8): p. 3249.
R. A. McPherson, M. R. Pincus, and J. B. Henry, Henry's ClinicalDiagnosisand
Management by Laboratory Methods. 21st ed, 2007, Philadelphia: Saunders
Elsevier. p. p.
R. P. McEver and C. Zhu, Rolling Cell Adhesion. Annu Rev Cell Dev Biol, 2010.
26: p. 363.
P. J. Quesenberry and P. S. Becker, Stem Cell Homing: Rolling, Crawling, and
Nesting. Proc Natl Acad Sci U S A, 1998. 95(26): p. 15155.
R. Giavazzi, M. Foppolo, R. Dossi, and A. Remuzzi, Rolling and Adhesion of
Human Tumor Cells on Vascular Endothelium under Physiological Flow
Conditions. J Clin Invest, 1993. 92(6): p. 3038.
T. Lapidot, A. Dar, and 0. Kollet, How Do Stem Cells Find Their Way Home?
Blood, 2005. 106(6): p. 1901.
A. D. Hughes, J. Mattison, L. T. Western, J. D. Powderly, B. T. Greene, and M.
R. King, Microtube Device for Selectin-Mediated Capture of Viable Circulating
Tumor Cellsfrom Blood. Clin Chem, 2012. 58(5): p. 846.
32
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
R. Karnik, S. Hong, H. Zhang, Y. Mei, D. G. Anderson, J. M. Karp, and R.
Langer, Nanomechanical Control of Cell Rolling in Two Dimensions through
Surface PatterningofReceptors. Nano Lett, 2008. 8(4): p. 1153.
T. A. Springer, Adhesion Receptors of the Immune System. Nature, 1990.
346(6283): p. 425.
A. Thiel, A. Scheffold, and A. Radbruch, Immunomagnetic Cell Sorting--Pushing
the Limits. Immunotechnology, 1998. 4(2): p. 89.
W. A. Bonner, H. R. Hulett, R. G. Sweet, and L. A. Herzenberg, Fluorescence
Activated Cell Sorting. Rev Sci Instrum, 1972. 43(3): p. 404.
L. Bonetta, Flow Cytometry Smaller and Better. Nature Methods, 2005. 2(10): p.
785.
S. Miltenyi, W. Muller, W. Weichel, and A. Radbruch, High GradientMagnetic
Cell Separationwith Macs. Cytometry, 1990. 11(2): p. 231.
P. Cuatrecasas, M. Wilchek, and C. B. Anfinsen, Selective Enzyme Purificationby
Affinity Chromatography.Proc Nati Acad Sci U S A, 1968. 61(2): p. 636.
A. H. Ross, D. Baltimore, and H. N. Eisen, Phosphotyrosine-ContainingProteins
Isolated by Affinity Chromatography with Antibodies to a Synthetic Hapten.
Nature, 1981. 294(5842): p. 654.
M. C. Smith, T. C. Furman, T. D. Ingolia, and C. Pidgeon, Chelating PeptideImmobilized Metal Ion Affinity Chromatography. A New Concept in Affinity
ChromatographyforRecombinantProteins. J Biol Chem, 1988. 263(15): p. 7211.
H. Gadgil, S. A. Oak, and H. W. Jarrett, Affinity Purification of DNA-Binding
Proteins.J Biochem Biophys Methods, 2001. 49(1-3): p. 607.
C. Tozzi, L. Anfossi, and G. Giraudi, Affinity ChromatographyTechniques Based
on the Immobilisation of Peptides Exhibiting Specific Binding Activity. J
Chromatogr B Analyt Technol Biomed Life Sci, 2003. 797(1-2): p. 289.
S. A. Camperi, N. B. Lannucci, G. J. Albanesi, M. Oggero Eberhardt, M.
Etcheverrigaray, A. Messeguer, F. Albericio, and 0. Cascone, Monoclonal
Antibody Purificationby Affinity Chromatographywith Ligands Derivedfrom the
Screening of Peptide Combinatory Libraries. Biotechnol Lett, 2003. 25(18): p.
1545.
K. Berg, C. A. Ogburn, K. Paucker, K. E. Mogensen, and K. Cantell, Affinity
Chromatography of Human Leukocyte and Diploid Cell Interferons on
Sepharose-BoundAntibodies. J Immunol, 1975. 114(2 Pt 1): p. 640.
M. M. Baran, D. M. Allen, S. R. Russell, M. E. Scheetz, 2nd, and J. F. Monthony,
Cell Sorting Using a Universally Applicable Affinity Chromatography Matrix:
Solid-Phase Anti-Fluorescein Isothiocyanate Antibody. J Immunol Methods,
1982. 53(3): p. 321.
C. M. Hertz, D. J. Graves, D. A. Lauffenburger, and F. T. Serota, Use of Cell
Affinity Chromatography for Separation of Lymphocyte Subpopulations.
Biotechnol Bioeng, 1985. 27(5): p. 603.
A. Kumar and A. Srivastava, Cell Separation Using Cryogel-Based Affinity
Chromatography.Nat Protoc, 2010. 5(11): p. 1737.
E. Mandrusov, A. Houng, E. Klein, and E. F. Leonard, Membrane-Based Cell
Affinity Chromatographyto Retrieve Viable Cells. Biotechnol Prog, 1995. 11(2):
p. 208.
33
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
S. Lin, S.-Y. Lee, C.-C. Lin, and C.-K. Lee, Determination of Binding Constant
and Stoichiometry for Antibody-Antigen Interaction with Surface Plasmon
Resonance. Current Proteomics, 2006. 3(12): p. 271.
C. A. Borrebaeck, A. C. Malmborg, C. Furebring, A. Michaelsson, S. Ward, L.
Danielsson, and M. Ohlin, Kinetic Analysis of Recombinant Antibody-Antigen
Interactions: Relation between Structural Domains and Antigen Binding.
Biotechnology (N Y), 1992. 10(6): p. 697.
S. Hearty, P. Leonard, and R. O'Kennedy, MeasuringAntibody-Antigen Binding
Kinetics Using Surface Plasmon Resonance. Methods Mol Biol, 2012. 907: p.
411.
S. M. Lucila and U. H. von Andrian, Immunological Adhesion and Homing
Molecules, in Els200 1, John Wiley & Sons, Ltd.
L. A. Lasky, Selectin-Carbohydrate Interactions and the Initiation of the
Inflammatory Response. Annual Review of Biochemistry, 1995. 64: p. 113.
R. P. McEver, Selectins: Lectins That Initiate Cell Adhesion under Flow. Curr
Opin Cell Biol, 2002. 14(5): p. 581.
P. Mehta, R. D. Cummings, and R. P. McEver, Affinity and Kinetic Analysis ofPSelectin Binding to P-Selectin Glycoprotein Ligand-1. J Biol Chem, 1998.
273(49): p. 32506.
A. G. Klopocki, T. Yago, P. Mehta, J. Yang, T. Wu, A. Leppanen, N. V. Bovin,
R. D. Cummings, C. Zhu, and R. P. McEver, Replacing a Lectin Domain Residue
in L-Selectin Enhances Binding to P-Selectin Glycoprotein Ligand-1 but Not to 6Sulfo-Sialyl Lewis X J Biol Chem, 2008. 283(17): p. 11493.
M. K. Wild, M. C. Huang, U. Schulze-Horsel, P. A. van der Merwe, and D.
Vestweber, Affinity, Kinetics, and Thermodynamics of E-Selectin Binding to ESelectin Ligand-1. J Biol Chem, 2001. 276(34): p. 31602.
K. Noda, S. Nakao, S. Ishida, and T. Ishibashi, Leukocyte Adhesion Molecules in
Diabetic Retinopathy. J Ophthalmol, 2012. 2012: p. 279037.
G. A. Zimmerman, Two by Two: The Pairings of P-Selectin and P-Selectin
Glycoprotein Ligand 1. Proc Natl Acad Sci U S A, 2001. 98(18): p. 10023.
P. P. Wilkins, K. L. Moore, R. P. McEver, and R. D. Cummings, Tyrosine
Sulfation of P-Selectin Glycoprotein Ligand-1 Is Required for High Affinity
Binding to P-Selectin. J Biol Chem, 1995. 270(39): p. 22677.
K. K. Sarangapani, T. Yago, A. G. Klopocki, M. B. Lawrence, C. B. Fieger, S. D.
Rosen, R. P. McEver, and C. Zhu, Low Force DeceleratesL-Selectin Dissociation
from P-Selectin Glycoprotein Ligand-1 and Endoglycan. J Biol Chem, 2004.
279(3): p. 2291.
B. T. Marshall, M. Long, J. W. Piper, T. Yago, R. P. McEver, and C. Zhu, Direct
Observation of Catch Bonds Involving Cell-Adhesion Molecules. Nature, 2003.
423(6936): p. 190.
R. Alon, S. Chen, K. D. Puri, E. B. Finger, and T. A. Springer, The Kinetics ofLSelectin Tethers and the Mechanics of Selectin-Mediated Rolling. J Cell Biol,
1997. 138(5): p. 1169.
M. Dembo, D. C. Torney, K. Saxman, and D. Hammer, The Reaction-Limited
Kinetics of Membrane-to-Surface Adhesion and Detachment. Proc R Soc Lond B
Biol Sci, 1988. 234(1274): p. 55.
34
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
W. Thomas, Catch Bonds in Adhesion. Annu Rev Biomed Eng, 2008. 10: p. 39.
R. Alon, S. Chen, R. Fuhlbrigge, K. D. Puri, and T. A. Springer, The Kinetics and
Shear Threshold of Transient and Rolling Interactions of L-Selectin with Its
Ligand on Leukocytes. Proc Natl Acad Sci U S A, 1998. 95(20): p. 11631.
P. Sundd, M. K. Pospieszalska, L. S. Cheung, K. Konstantopoulos, and K. Ley,
Biomechanics ofLeukocyte Rolling. Biorheology, 2011. 48(1): p. 1.
P. Pawar, S. Jadhav, C. D. Eggleton, and K. Konstantopoulos, Roles of Cell and
Microvillus Deformation and Receptor-Ligand Binding Kinetics in Cell Rolling.
Am J Physiol Heart Circ Physiol, 2008. 295(4): p. H1439.
E. Y. Park, M. J. Smith, E. S. Stropp, K. R. Snapp, J. A. DiVietro, W. F. Walker,
D. W. Schmidtke, S. L. Diamond, and M. B. Lawrence, Comparison of Psgl-1
Microbead and Neutrophil Rolling: Microvillus Elongation Stabilizes P-Selectin
Bond Clusters. Biophys J, 2002. 82(4): p. 1835.
G. Schumacher, U. Bakowsky, C. Gege, R. R. Schmidt, U. Rothe, and G. Bendas,
Lessons Learnedfrom Clustering of FluorinatedGlycolipids on Selectin Ligand
Function in Cell Rolling. Biochemistry, 2006. 45(9): p. 2894.
V. Ramachandran, M. Williams, T. Yago, D. W. Schmidtke, and R. P. McEver,
Dynamic Alterations of Membrane Tethers Stabilize Leukocyte Rolling on PSelectin. Proc Natl Acad Sci U S A, 2004. 101(37): p. 13519.
M. K. Pospieszalska and K. Ley, Dynamics of Microvillus Extension and Tether
Formationin Rolling Leukocytes. Cell Mol Bioeng, 2009. 2(2): p. 207.
J. Y. Shao, H. P. Ting-Beall, and R. M. Hochmuth, Static and Dynamic Lengths
ofNeutrophil Microvilli. Proc Natl Acad Sci U S A, 1998. 95(12): p. 6797.
C. Dong and X. X. Lei, Biomechanics of Cell Rolling: Shear Flow, Cell-Surface
Adhesion, and Cell Deformability.Journal of Biomechanics, 2000. 33(1): p. 35.
A. J. Goldman, R. G. Cox, and H. A. Brenner, Slow Viscous Motion of a Sphere
Parallel to a Plane Wall. I. Motion through a Quiescent Fluid. Chemical
Engineering Science, 1967. 22: p. 637.
A. J. Glodman, R. G. Cox, and H. A. Brenner, Slow Viscous Motion of a Sphere
Parallelto a Plane Wall. Ii. Couette Flow. Chemical Engineering Science, 1967.
22: p. 653.
A. W. Greenberg, D. K. Brunk, and D. A. Hammer, Cell-Free Rolling Mediated
by L-Selectin and Sialyl Lewis(X) Reveals the Shear Threshold Effect. Biophys J,
2000. 79(5): p. 2391.
A. 0. Eniola, P. J. Willcox, and D. A. Hammer, Interplay between Rolling and
Firm Adhesion Elucidatedwith a Cell-Free System Engineered with Two Distinct
Receptor-LigandPairs.Biophys J, 2003. 85(4): p. 2720.
D. K. Brunk and D. A. Hammer, Quantifying Rolling Adhesion with a Cell-Free
Assay: E-Selectin and Its CarbohydrateLigands. Biophys J, 1997. 72(6): p. 2820.
X. Lei, M. B. Lawrence, and C. Dong, Influence of Cell Deformation on
Leukocyte Rolling Adhesion in Shear Flow. J Biomech Eng, 1999. 121(6): p. 636.
K. L. Moore, K. D. Patel, R. E. Bruehl, F. Li, D. A. Johnson, H. S. Lichenstein, R.
D. Cummings, D. F. Bainton, and R. P. McEver, P-Selectin Glycoprotein Ligand1 Mediates Rolling of Human Neutrophils on P-Selectin. J Cell Biol, 1995.
128(4): p. 661.
35
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
D. F. Bainton, Distinct Granule Populations in Human Neutrophils and
Lysosomal Organelles Identified by Immuno-Electron Microscopy. J Immunol
Methods, 1999. 232(1-2): p. 153.
U. H. von Andrian, S. R. Hasslen, R. D. Nelson, S. L. Erlandsen, and E. C.
Butcher, A Central Role for Microvillous Receptor Presentation in Leukocyte
Adhesion under Flow. Cell, 1995. 82(6): p. 989.
T. Yago, A. Leppanen, H. Y. Qiu, W. D. Marcus, M. U. Nollert, C. Zhu, R. D.
Cummings, and R. P. McEver, Distinct Molecular and Cellular Contributions to
Stabilizing Selectin-MediatedRolling under Flow. Journal of Cell Biology, 2002.
158(4): p. 787.
W. C. Chang, L. P. Lee, and D. Liepmann, Biomimetic Technique for AdhesionBased Collection and Separation of Cells in a Microfluidic Channel. Lab on a
Chip, 2005. 5(1): p. 64.
A. W. Greenberg and D. A. Hammer, Cell Separation Mediated by Differential
Rolling Adhesion. Biotechnology and Bioengineering, 2001. 73(2): p. 111.
D. D. Nalayanda, M. Kalukanimuttam, and D. W. Schmidtke, Micropatterned
Surfaces for Controlling Cell Adhesion and Rolling under Flow. Biomed
Microdevices, 2007. 9(2): p. 207.
C. Edington, H. Murata, R. Koepsel, J. Andersen, S. Eom, T. Kanade, A. C.
Balazs, G. Kolmakov, C. Kline, D. McKeel, Z. Liron, and A. J. Russell, Tailoring
the Trajectory of Cell Rolling with Cytotactic Surfaces. Langmuir, 2011. 27(24):
p. 15345.
S. Choi, J. M. Karp, and R. Karnik, Cell Sorting by Deterministic Cell Rolling.
Lab Chip, 2012. 12(8): p. 1427.
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Studying Cell
Rolling Trajectorieson Asymmetric Receptor Patterns.J Vis Exp, 2011(48).
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Examining the
Lateral Displacement of H160 Cells Rolling on Asymmetric P-Selectin Patterns.
Langmuir, 2011. 27(1): p. 240.
X. H. Cheng, D. Irimia, M. Dixon, K. Sekine, U. Demirci, L. Zamir, R. G.
Tompkins, W. Rodriguez, and M. Toner, A Microfluidic Device for Practical
Label-Free Cd4+TCell Counting of Hiv-Infected Subjects. Lab on a Chip, 2007.
7(2): p. 170.
S. Q. Chen, R. Alon, R. C. Fuhlbrigge, and T. A. Springer, Rolling and Transient
Tethering of Leukocytes on Antibodies Reveal Specializations of Selectins.
Proceedings of the National Academy of Sciences of the United States of
America, 1997. 94(7): p. 3172.
N. Charles, J. L. Liesveld, and M. R. King, Investigating the Feasibility of Stem
Cell Enrichment Mediated by Immobilized Selectins. Biotechnology Progress,
2007. 23(6): p. 1463.
B. K. Kay, J. Kasanov, and M. Yamabhai, Screening Phage-Displayed
CombinatorialPeptide Libraries.Methods, 2001. 24(3): p. 240.
36
2
Free-Flow
Travel and Capture of HL60 Cells
inside
Microfluidic Channels
2.1
Introduction
Chapter Two aims to study how cells travel in microfluidic channels before they are
drawn out of the flowing stream by affinity-mediated cell capture. How the cells are
captured on the substrate is important because these parameters influence the design and
performance of microfluidic devices for cell separation.
Adhesive receptor molecules exhibiting weak interaction with cells can recruit cells out
from a continuously flowing stream resulting in cell rolling and tracking along the
patterns [1-11]. Continuous label-free devices for the separation or analysis of cells based
on transient interactions with low affinity receptor-coated bottom surfaces within the
device have attracted a lot of attention recently [9-14]. When a cell enters the
microfluidic channel where the flow is in the low-Reynold-number regime (the inertial
effects are negligible), it travels in both x and y directions with the flow: (a) in y (vertical)
direction, the cell settles to the bottom of the channel because of the gravity while
experiencing frictional force and buoyant force in the opposition direction; (b) in x
(horizontal) direction, the cell travels with the flowing stream by the convection force
before it is captured by the adhesive receptors coated on the bottom of the device [15, 16].
The design of the length of the channel should be long enough to ensure the majority of
the cells can settle under the influence of gravity while it travels by convection until it
starts to interact with the receptor molecules at the bottom surface of the channel.
Accordingly, it is important to know how far the cell travels before it reaches the bottom
surface. This distance is defined as the settling distance, xs, as shown in Figure 2-1(a).
Rolling may be initiated after the cell is captured onto the substrate.
Cell capture in bio-functionalized microchannels under shear flow has been studied [1520]. For example, Tempelman et al have studied the effect of shear rate on the adhesion
of anti-dinitrophenol (DNP)- IgE sensitized cells on to the DNP-coated area in the flow
37
chamber: the cells were allowed to settle before they entered the DNP-coated area and it
was observed that the density of adherent cells decreases with distance down the
chamber; however, only the percentage of adhesion of the cells within the field of view
was qualitatively analyzed [18]. On the other hand, Cheung et al have analyzed the
spatial distributions of captured circulating tumor cells (CTCs) along the anti-EpCAMfunctionalized microchannels and results show that at the shear stress of 0.207 dyn/cm 2,
the most probable location for the cells to be captured is about x/L=0.28, where x is the
locations from the inlet and L is the total length of the channel (=32 cm). However, in
their system, the whole microchannel was coated with antibodies and the cells were not
settled at x/L =0 [17].
In Chapter One, a separation device with multiple inclined receptor bands has been
proposed (Figure 1-6). The interacting cell can be captured onto the patterns and roll on
them. The asymmetry of the patterns with respect to the flow alters the trajectory of the
interacting cells. The interacting cells can track along the edge of the pattern and get
displaced into the buffer stream and are thus separated from the non-interacting cells.
Since multiple patterns are designed on the substrate, it is important to understand the
capture process on each inclined pattern from the free flowing stream. For example, what
is the attachment distance, /a, which is defined as how far a settled cell will travel from
the beginning of a receptor-coated area until it is captured and starts to roll on the
substrate (see Figure 2-1(b)). This parameter relates to the design of the width of pattern
and further plays an important role in the performance of the device. On the other hand,
the distribution of the attachment distance is unknown and it is not fully understood how
the attachment distance changes with shear stress magnitudes. Therefore, systematic
study of free-flow travel and the recruitment process of the cells from the flow onto the
substrate inside the microfluidic channels is prerequisite for the design and the
development of microfluidic devices involving transient interactions between the cells
and the receptor-coated substrate for analysis or separation of the cells. The problem is
also of interest in cell biology research, for example, seeding of cells in microchannels
for tissue culture [21].
38
(a)
inlet,
(b)
x,: settling distance
Se ed cell
[R] Receptor
5 cm
Q
tAttachment Di ance, 1.
* Attached cells
vomw Flowing cells
Figure 2-1 Schematic diagrams of cells travelling in a microfluidic channel. (a) The
illustration of a cell entering the microfluidic channel and travelling in both x and y
directions. The settling distance, x,, is defined as how far a cell will travel before it
settles close enough to the bottom surface. (b) Illustration of cells passing the Pselectin-coated area. A cell may be captured from the flowing stream at an attachment
distance, I., after it enters the P-selectin-coated area.
In Chapter Two, the role of fluid flow on cell adhesion will be studied using simple shear
flow in a parallel-plate microchannel. A long microfluidic channel envisioned in the top
of Figure 2-1(b) was designed for the study of how the HL60 cells travel with the free
flowing stream and be captured onto the P-selectin-coated substrate. HL60 cells and Pselectin molecules have been widely used as a model system for leukocyte rolling study
39
[22-26]. The surfaces of HL60 cells express P-selectin glycoprotein ligand-1 (PSGL-1),
which binds reversibly to the receptor P-selectin to enable rolling in vivo and in vitro
[22]. The long microlfuidic channel was first used to study the settling distance (x,) under
different shear stresses. The changes of the convection velocity of the cells while they
travel along the channel were analyzed to quantify how far the cells travel before settling
under the gravity. In the second part, the P-selectin molecules were coated at the end
region of the channel (see Figure 2-1(b)) to ensure the cells have enough time to settle
before they start to interact with the receptor molecules at the bottom surface. The
attachment distances (Ia) of the cells under different shear stresses were quantified and an
empirical relation between the characteristic attachment distance and the shear stress was
established based on the experimental results.
2.2 Part A. Cell Settling inside Microfluidic Channels
Part A aims to quantitatively analyze free-flow travel of the cells inside microfluidic
devices and identify the settling distance (x,) under different shear stresses. A long
microchannel (-
15 cm) was designed for this study. The convection velocities of the
cells at different location along the channel were analyzed. The convection velocity was
expected to decrease as the cells settle to the surface under the gravity and x, is identified
as how far the cells travel before settling.
2.2.1
2.2.1.1
Fabrication of Substrate and Microfluidic Channel
Materials
Petri dishes were purchased from BD FalconTM. Polydimethylsiloxane
(PDMS;
Sylgard@184) was purchased from Dow Coming. Bovine serum albumin (BSA,
Rockland Immunochemicals, Inc.) was diluted in 150 mM NaCl Dulbecco's phosphate
buffered saline (DPBS, Mediatech Inc.). All materials employed in this study were used
without further purification unless specified otherwise.
40
2.2.1.2
Fabrication of Substrate Incorporated PDMS channel
The Petri dish was incubated with 3-wt% BSA in DPBS for 3 h at room temperature. The
Petri dish was then rinsed with DPBS and kept wet to serve as the substrate. A silicon
master mold that defines the microfluidic channel used in this study was a gift from
Chong Shen in Prof. Kamik's group at MIT. The microfluidic channel was cast in PDMS
from the silicon master mold. The height of this 15 cm-long channel is 95 pim and the
width is 2 mm. The PDMS channel was then attached to the BSA-coated substrate via a
vacuum manifold to assemble the device.
2.2.2
2.2.2.1
Convection Velocity Experiment
Cell Culture
HL60 myeloid cells (ATCC) were maintained in Iscove's modified Dulbecco's medium
supplemented
with 20%
fetal bovine
serum (FBS)
according to the ATCC
recommendation. Cells were cultured in 75 cm 2 polystyrene tissue culture flasks (BD
Falcon) with the cell density maintained between 105 - 106 cells/mL. (passage number
39, 40 and 41 were used in this study).
2.2.2.2
Experiment Setup
A suspension of the HL60 cells (~ 105 cells/mL) was allowed to flow into the assembled
microfluidic channel as described above at room temperature of 24.5 C. A syringe pump
(World Precision Instruments (WPI), SP230IW) was used to generate different flow rates
of between 5 and 20 pL/min, with corresponding shear stresses of 0.25-1.0 dyn/cm 2
(-0.025-0.1 Pa). The flow was laminar (Reynold's number Re ~ 0.1-0.4), and the shear
stress r was calculated using the plane Poiseuille flow equation r=6pQ/wh 2 , where p is
the viscosity,
Q is volumetric
flow rate, w is width of the flow chamber, and h is height
of the channel. An inverted microscope (Nikon TE2000-U) with a mounted camera
(Andor iXon 885) was used to record the flowing HL60 cells using a 4x objective,
typically at a rate of 1 frame/s for durations of 60-180 s with exposure time of 0.05-1.0 s.
41
The videos recording the cells traveling in the free flowing stream were taken at different
positions away from the inlet (2, 4, 6, 8, 10, 12, and 13 cm) under different shear stresses.
2.2.2.3
Results and Data Analysis
For each experiment of specific position and shear stress, 20 cells were selected randomly
from the video. The position in x-coordinate (xi) of each selected cell in the free stream at
time t4 was identified manually by ImageJ. The convection velocity (u,,) of a cell is
defined as
u
= X'"(t)m
X,(tn)
;m>n
(Eq. 2-1)
tm -tn
The convection velocity of the cell in the free flowing stream is assumed to be the same
as the velocity profile of steady state laminar flow for Newtonian fluid, which varies
quadratically with the height from the bottom surface [27]:
QY-j
hW
h
where
4Uy
h
I
Yiy
h
h
(Eq. 2-2)
Q:
volumetric flow rate
w:
width of the channel
h:
height of channel
y :height
of the center of the cell from the bottom surface of the channel
of the flow
rate
shear
y7:
Umax: maximum convection velocity of flow
However, when the cell (radius, r) is settled to the height less than 4r (y < 4r), the
convection velocity is reduced by the hydrodynamic wall effect as described by the
Goldman equation [15, 16, 28]:
U=~(
h W
1-- 5 r
16 y
42
YY
(r)
16 y
(Eq. 2-3)
Figure 2-2(a) summaries the convection velocities of the cells in the free flowing stream
(y-axis) at different horizontal positions away from the inlet (x-axis). Wall shear stress
varies from 0.25 to 1.0 dyn/cm 2 . Data are presented as the mean velocity and the
standard deviation of 20 cells for each experiment condition. This convection velocity
can be used to indicate whether a cell is close to the surface or not: the cell travels at
maximum convection velocity near the half height of the channel; as the cell keeps
settling down, the convection velocity decreases until u, reaches a minimum value. As
the cell reaches the bottom surface (y ~ r), the Goldman equation can be simplified as:
U,--
11r
11
ry--T
16
16
U
C
(1 XT
or
or
(Eq. 2-4)
\ 16)(t
Eq. 2-4 suggests that as y ~ r, the convection velocity is roughly proportional to the shear
rate, y) (or shear stress r; for Newtonian fluid with viscosity of p, shear stress is defined
as r = p Y).
Figure 2-2(b) summarizes the values of
u /
II)(T)]
of cells at different
horizontal positions The results show that for all the different shear stresses, the values of
u
/ I1)(-r) decreases with the travelling distance and reaches a minimum around
the value of 5 [pm], which is roughly consistent with the Goldman equation (as y ~ r ~
5p m, u
/
~)]
4.95 [pm]). This suggests that the average convection velocity of
the cells after settling increases linearly with the increasing shear stress within the tested
range. The settling distance, xs, can be identified from Figure 2-2(a) or Figure 2-2(b). For
the shear stresses of 0.25, 0.5 and 0.75 dyn/cm 2 , x, is between 4 and 6 cm. For the shear
stress of 1.0 dyn/cm 2 , x, is around 10 - 12 cm. For the design of the continuous
microfluidic cells separation device based on the interaction between the cells and the
receptor-coated bottom surfaces, x, is be a key parameter to ensure most of the cells are
able to interact with the substrate. Moreover, the average of the convection velocity of the
43
settled cells increases linearly with increasing shear stress (see Figure 2-2 (c)). The
average convection velocity scales with the shear stress [28], which indicates shear stress
is a controlling parameter for convective transport of the cells inside the microfluidic
device. Yago et al observed similar linear relationship when the microspheres flowed
inside the flow chamber [20].
44
2000
(a)
2
T (dyn/cM
1500
)
-0-O.25
E
-0-.75
S1000
500
0
12
10
8
6
4
2
Horizontal Position from the inlet, x (pm)
3
(b)
14
25
. (dyn/cM2 )
20
+0.25
+0.5
=1.11~
15
+0.75
10
5
0
0
(c)
12
10
8
6
4
2
Horizontal Position from the inlet, x (pm)
500
(dyn/cm 2)
Uc, (pm/s)
(Avg ± SD)
0.25
0.5
0.75
1.0
104± 5
205± 5
292 ± 5
391 ± 5
T
400
300
200
1
0
- .P
394r
R= 0.99
100
0
0
0.25
0.75
0.5
T (dyn/cm2)
45
1
14
Figure 2-2(a) Convection velocity (u,) of the cells at the different horizontal
position. Convection velocity divided by shear stress decreases and reaches a minimum
value as the cells travel along the channel. Data are presented as the mean and the
standard deviation of the convection velocity obtained from 20 cells under each
experiment condition. Different markers and colors represent different shear stresses. (b)
[u, ] /[(I I/ 16)(T /i)] of the cells at different horizontal positions (c) Variation of the
convection velocity of the settled cells with the shear stress. Inset summaries the mean
and the standard deviation of the convection velocity of the cells at position x = 13 cm.
The relationship between uc, and T is well described by a linear fit (dashed line).
2.3 Part B. Attachment Distance of the Cells in the Microfluidic
Channel
Part B aims to quantitatively analyze capture process of cells onto the receptor-coated
microfluidic devices under different shear stresses. HL60 cells were flowed into the 15cm long microchannel and the P-selectin molecules were coated only at the end region
(13-14cm) of the bottom of the microchannel to ensure the cells have enough time to
settle before starting to interact with the receptor molecules on the bottom surface.
Capture process can be studied by analyzing the distribution of the attachment distance
('a), the distance between the starting point of the P-selectin area and the position where
the cells were captured.
2.3.1
2.3.1.1
Fabrication of Substrate and Microfluidic Channel
Materials
The Petri dish was purchased from BD FalconTM. Polydimethylsiloxane (PDMS;
Sylgard@184) was purchased from Dow Coming. Recombinant human P-selectin (R&D
Systems Inc.) and fluorescein isothiocyanate (FITC)-conjugated BSA (Sigma-Aldrich)
were diluted in 150 mM NaCl Dulbecco's phosphate buffered saline (DPBS, Mediatech
Inc.). Perfusion chambers were purchased from Electron Microscopy Sciences. All
materials employed in this study were used without further purification unless specified
otherwise.
46
2.3.1.2
Fabrication of Substrate Incorporated PDMS channel
The Petri dish was incubated with P-selectin at 5 pg/mL in DPBS using a perfusion
chamber for 3 h at room temperature. The surface was then incubated with FITCconjugated BSA (1 mg/mL in DPBS) at room temperature for 15 min to identify the edge
of P-selectin-coated area. The surface was then backfilled with BSA (3wt% in DPBS) at
room temperature for 1 h to block nonspecific interactions. The Petri dish was then rinsed
with DPBS and kept wet to serve as the substrate. The PDMS channel (the same design
as being used in Part A) was then assembled with the substrate via a vacuum manifold.
The P-selectin-coated area was aligned near the end of the microfluidic channel as
illustrated in Figure 2-1(b). The P-selectin-coated area started from the position around
13-14 cm away from the inlet (which is larger than the settling distance (x,)) to ensure
the cells had enough time to settle down.
2.3.1.3
Substrate Characterization
An inverted fluorescent microscope (Nikon TE2000-U) was used to characterize the
substrate. Well-defined edge of P-selectin-coated region can be identified as shown in
Figure 2-3. The bright area corresponds to FITC-BSA-coated area.
inlet
outlet
tin
Figure 2-3 Fluorescent images of the substrate. The bright area corresponds to FITCBSA-coated area.
47
2.3.2
2.3.2.1
Cell Capture Experiment
Cell Culture and Experiment Setup
The setup of the cell capture experiment was similar to what has been described in
section 2.2.2 (the convection velocity experiment). The videos were taken near the area
where the P-selectin-coated area started. The HL60 cells may be captured onto the Pselectin-coated area to initiate rolling after they enter this region. The position where the
cells captured from the free flowing stream onto the surface was defined as the
attachment distance, /a (see Figure 2-1(b)). The wall shear stresses were varied from 0.25
to 1.0 dyn/cm 2. For each shear stress magnitude, three independent experiments were
performed.
2.3.2.2
Data Analysis
The image sequences were analyzed using a customized Matlab (Mathworks, Inc.)
program that utilized particle-tracking freeware to detect the captured cells [29]. The total
number of the cells entering the P-selectin-coated area during the certain time period
within the field of view was counted manually.
Figure 2-4 is an example showing one frame extracted from a video taken under shear
stress of 0.25 dyn/cm 2 . After the cells entered the receptor-coated area, they can be drawn
out from the free stream and be captured onto the surface. It is clearly shown that some
of the cells flowing into the field of the view flowed out of the field of view. The distance
where a cell was captured onto the surface from the start of the receptor coated area is
defined as the attachment distance (la). The average attachment distance is useful for
quantitatively understanding the capture process. Figure 2-5 shows an example of the
distribution of the attachment distances of HL60 cells at shear stress of 0.25 dyn/cm 2. The
cells flowed into the field of view at different time. This figure was obtained by analyzing
65 images acquired at 1 fps using a customized Matlab code. It was observed that the
attachment distance has a distribution, which is consistent with previous observations [18,
48
19]. It is more insightful to understand the distribution of the attachment distance, since it
could be used for studying the spatial distribution of the cells inside a microfluidic
channel.
ng out of
of view
n
P+ P-sekwctn coated as
x-O
Figure 2-4 HL60 cells being captured onto the P-selectin-coated area at the shear
stress of 0.25 dyn/cm 2 . Some cells were observed to flow out of the field of view.
80
6
.0
Poe0
a'
E 4
%
0,%0*
0
*1=
*
0
0 Go
2
0
500
1000
1500
2000
Attachment Distance (pm)
Figure 2-5 The attachment distances of HL60 cells at the shear stress of 0.25
dyn/cm 2. P-selectin edge is located at 0 lim. The data obtained by analyzing 65 images
acquired at 1 fps using a customized Matlab code.
Figure 2-6(a) shows the frequency of the attachment distance at different shear stresses
2
(0.25-1.0 dyn/cm 2). For shear stresses of 0.25 and 0.5 dyn/cm , the distributions exhibit a
decaying characteristic. Cumulative probability (CP) was calculated form the data and fit
it with an empirical equation:
49
CP(x) = A l - exp
7
(Eq. 2-5)
Here, </4> is the characteristic attachment distance and in the ideal case, parameter A will
approach the value of one. A and </a> obtained by fitting the cumulative probabilities
obtained from experimental data to Eq. 2-5 for different shear stresses are summarized in
Table 2-1. Three independent experiments were performed for each shear stress and data
are presented as mean and one standard deviation. Eq. 2-5 fits well for T = 0.25 and 0.5
dyn/cm 2 . Interestingly, the average values of A for T = 0.25 and 0.5 dyn/cm 2 have close
results (0.5±0.1 and 0.5±0.1, respectively). However, for T = 0.75 and 1.0 dyn/cm 2 , the
plateaus in Figure 2-6(b) were not observed due to insufficient data so it is difficult to do
curve fitting using Eq. 2-5. To further investigate the empirical relation between <4,> and
T, we assumed that the value of A fort = 0.75 and 1.0 dyn/cm 2 is the same as that for -=
0.25 and 0.5 dyn/cm 2 :
HL60 cells may exhibit cellular heterogeneity. HL60 cells with higher PSGL-1
expression level may have small <Ia> and other cells with lower PSGL- 1 expression level
may have large </a>. The fraction of the cells with small </a> is "A" and the fraction of
the cells with large <I,> is "1-A." Limited by the field of view of the camera, only the
cells with small <la> can be analyzed under the shear stress we are interested (0.25 and
1.0 dyn/cm 2).
The empirical equation was then simplified as:
CP(x) =0.5 1- exp -
(Eq. 2-6)
Here, 0.5 was determined by averaging the mean values of A from -u = 0.25 and 0.5
dyn/cm 2 . <4"> obtained by fitting the cumulative probabilities to Eq. 2-6 for different
shear stresses are summarized in Table 2-1. The values of <a> obtained from Eq. 2-5 and
50
Eq. 2-6 for
T =
0.25 and 0.5 dyn/cm 2 were compared and the maximum difference was
found to be within 4% (data are not shown in the Table 2-1).
(a)
I
2
0.25 dyn/cm
U
0
GJ
2000
1000
o
1000
2000
0*
a)
15
U-
8
0.75 dyn/cm
2
1.0 dyn/cm
10
00
1000
1
2000
00
500
1000
1500
2000
Attachment Distance (pm)
(b)
0.5
0.25 dyn/cm2m
0.4
.5 dyn/cM2
-
'u 0.3
0
0.2
LI
0.75 dyn/cM2
1.0 dyn/cM2
0.1
0
0
500
1000
1500
2000
Attachment Distance (pm)
Figure 2-6 Attachment distance of HL60 cells at different shear stresses of 0.25-1.0
dyn/cm2. Representative results are shown for only one experiment for each shear stress.
(a) Distribution of attachment distance at different shear stresses (0.25-1.0 dyn/cm 2). (b)
Cumulative probability (CP) of attachment distance at different shear stresses (0.25-1.0
dyn/cm2).
Table 2-1 Comparison of the average values and standard deviation of A and <J.>
obtained from curve fitting at different shear stresses of 0.25-1.0 dyn/cm 2. n = 3
51
replicate experiments. Data were obtained by fitting to two empirical equations with
experimentally measured attachment distance.
P umba i ve CP(x)- A 1- exp ))
Probability
.1,.)
T(dyn/cm 2)
0.25
0.5
0.75
1
0.25
0.5
0.75
1-
Avg±SD
0.5±0.1
0.5±0.1
A
<I?
pm
Probability CP(x) -0.5
T (dyn/cm 2)
0.25
0.5
0.75
1
620±100
1500±100
-
1- exp -
Avg±SD
<a>
pm
650±80
1480±490
4840±930
15580±9000
Interestingly, the average value of </,> obtained by fitting the cumulative probability to
Eq. 2-6 showed an exponential increase with increasing shear stress (Figure 2-7). This
observation provides an empirical relation between the shear stress and the average
characteristic attachment distance under the shear stress of 0.25-1.0 dyn/cm 2 . As a cell
settles to contact the substrate, new cell-substrate adhesion may form at the contact area
between the cell and receptor-coated surface. Higher flow rate results in higher shear
stress and faster convection velocity. Higher convection velocity near the bottom of the
substrate produces more frequent collisions between the cells and the substrate but short
the contact time before the cell moves forward [8, 30]. There is no justification why an
exponential relation was observed. However, the experimental results imply that the
effect of shear stress on the characteristic attachment distance becomes more significant
at higher shear stress, and likely originates from the nature of the molecular interactions
rather than hydrodynamic effects since the velocity of settled cells remains linear with the
shear stress (Figure 2-2(c)).
52
10000
-
A
1000
og<Ia>= 1.9T+2.3
V
R2 = 0.99
100
0
0.25 0.5 0.75
1
Shear Stress,T (dyn/cm 2)
Figure 1- 1 Variation of the average value of </4> with shear stress. The average
values of <1.> were obtained from the empirical fit (Eq. 2-6). The relationship between
<la> and T is well described by a linear fit on a semi-log plot (dashed line). Error bars
represent one standard deviation.
The probability of a cell being captured by a pattern from the free flow can be obtained
from the empirical equation as described above. Figure 2-8 shows the schematic diagram
of a cell travelling over a single inclined pattern of width of w, with respect to the flow of
angle of a. After the cell enter the pattern area, it can be captured within the area of
width of x = wisina. Eq. 2-5 can be modified as:
CP(x) = A I -exp -
.
(Eq. 2-7)
For a cell travelling under a shear stress of 0.5 dyn/cm 2 over an inclined pattern of width
of 15 pim, at the inclination angle of a = 150 with respect to the flow, the probability of
the cells being captured by this pattern from the free flowing stream is ~ 0.02 (A = 0.5
and </a> = 1480). This value is consistent with Bose et al's study which reported the
experimental probability of 0.015 - 0.038 at this specific pattern geometry and shear
stress [14].
53
Starting point
l
Cell%
Fw
Flow
Figure 1- 2 Schematic diagram of a cell travelling over a single inclined pattern. A
cell travels over the pattern of width w and an inclination angle a with respect to the flow.
After the cell enters the pattern area, it can be captured within a distance of x = wsina.
The probability of the cell being captured by this pattern from the free flowing stream is
CP(x').
2.3.3
Flow Cytometry Analysis
The work of flow cytometry analysis was done in collaboration with Suman Bose (MIT).
10 % fetal bovine serum (FBS, ATCC) and 0.1% sodium azide (Sigma-Aldrich) in
phosphate buffered saline (PBS, Mediatech Inc) were prepared to serve as FACS buffer.
PE (phycoerythrin)- conjugated anti-PSGL-1 antibody (BD Pharmingen) was used to
stain PSGL-l surface expression from HL-60 cells. Mouse IgG1,
C isotype (BD
Pharmingen) was used as the control. Two antibodies were diluted separately in FACS
Buffer to the final concentration of 1 pg/mL. HL60 cells were transferred from the flask
into two 15 mL conical tubes and were spun at 1700 rpm for 5 min. The supernatant was
discarded and the pellets were suspended in diluted anti-PSGL 1-PE and isotype antibody
solutions separately to a final concentration of 106 cell/mL. The suspended cells were
then incubated inside the ice bath for 30 minutes. The cells were then spun down at 1700
rpm for 5 min and the pellets were then suspended in FACS Buffer. The samples were
then analyzed by flow cytometry machine (LSRFortessa).
54
Heterogeneity in the cell population has been proposed to have an impact on the cell
adhesion under hydrodynamic flow [18, 19]. The flow cytometry results (see Figure 2-9)
suggest that the expression of PSGL-1 on HL60 cells has broad distribution, which is
consistent with Gee et al's observation [31]. The broad distribution of PSGL-1
expression indicates that there is some degree of cell-to-cell difference. Such cell
population heterogeneity may result in the distribution of the attachment distance and the
cells with higher expression level of PSGL-1 may have shorter attachment distance and
vice versa.
If
HL60-PSGL1-PE
HL60-iso
0
103
104
10s
Intensity
Figure 1- 3 Flow cytometry analysis. PE conjugated anti-PSGL- 1 antibody was used to
stain PSGL-1 surface expression from HL-60 cells. Mouse IgGi, K isotype antibody was
used as the control. (Adapted from Suman Bose).
2.4 Conclusion
In summary, the cells settle down under the influence of gravity after they enter the
microfluidic channel and then travel along the channel with the free stream. The
experimental results show that the average convection velocity of the cells with the flow
decreases until it reaches a minimum value when the cells are close to the bottom, which
is consistent with the Goldman equation. The minimum convection velocity is an
indicator when the cell are settled and the position where the convection velocity reaches
a minimum value or the cells settle to the bottom of the substrate is defined as the settling
distance (xs). Quantification of the settling distance is useful for the design of the
continuous microfluidic cell separation device based on the interaction with receptorcoated bottom surfaces. Moreover, and the shear stress has been identified as the
controlling parameter of convective transport. In addition, the nature of cell recruitment
55
from the flow onto the receptor-coated area has been studied by analyzing the attachment
distance (l), which is defined as the distance between the position where the cell is
captured and the position where the pattern starts. Although the study was limited by the
size of the field of view, the experimental results suggest that high shear stress results in
long attachment distance and the effect of shear stress on the characteristic attachment
distance become more significant at higher shear stress since the characteristic
attachment distance increase exponentially with increasing shear stress. The empirical
relationship between cumulative probability of the cells being captured and the position
of the channel was established. The empirical relationship was then applied to study the
probability of a cell being captured on a single inclined pattern from the free flow and the
result is consistent with experimental observation.
2.5 Chapter Reference
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
G. I. Bell, Models for Specific Adhesion of Cells to Cells. Science, 1978.
200(4342): p. 618.
R. Alon, D. A. Hammer, and T. A. Springer, Lifetime of the P-SelectinCarbohydrate Bond and Its Response to Tensile Force in Hydrodynamic Flow.
Nature, 1995. 374(6522): p. 539.
D. N. Granger and P. Kubes, The Microcirculation and Inflammation Modulation of Leukocyte-Endothelial Cell-Adhesion. Journal of Leukocyte
Biology, 1994. 55(5): p. 662.
L. A. Lasky, Selectin-Carbohydrate Interactions and the Initiation of the
Inflammatory Response. Annual Review of Biochemistry, 1995. 64: p. 113.
B. T. Marshall, M. Long, J. W. Piper, T. Yago, R. P. McEver, and C. Zhu, Direct
Observation of Catch Bonds Involving Cell-Adhesion Molecules. Nature, 2003.
423(6936): p. 190.
T. F. Tedder, D. A. Steeber, A. Chen, and P. Engel, The Selectins - Vascular
Adhesion Molecules. Faseb Journal, 1995. 9(10): p. 866.
G. A. Zimmerman, Two by Two: The Pairings of P-Selectin and P-Selectin
Glycoprotein Ligand 1. Proc Natl Acad Sci U S A, 2001. 98(18): p. 10023.
R. P. McEver and C. Zhu, Rolling Cell Adhesion. Annu Rev Cell Dev Biol, 2010.
26: p. 363.
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Studying Cell
Rolling Trajectorieson Asymmetric Receptor Patterns.J Vis Exp, 2011(48).
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Examining the
Lateral Displacement of H160 Cells Rolling on Asymmetric P-Selectin Patterns.
Langmuir, 2011. 27(1): p. 240.
56
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
R. Karnik, S. Hong, H. Zhang, Y. Mei, D. G. Anderson, J. M. Karp, and R.
Langer, Nanomechanical Control of Cell Rolling in Two Dimensions through
Surface PatterningofReceptors. Nano Lett, 2008. 8(4): p. 1153.
S. Choi, J. M. Karp, and R. Karnik, Cell Sorting by Deterministic Cell Rolling.
Lab Chip, 2012. 12(8): p. 1427.
C. Edington, H. Murata, R. Koepsel, J. Andersen, S. Eom, T. Kanade, A. C.
Balazs, G. Kolmakov, C. Kline, D. McKeel, Z. Liron, and A. J. Russell, Tailoring
the Trajectory of Cell Rolling with Cytotactic Surfaces. Langmuir, 2011. 27(24):
p. 15345.
S. Bose, R. Singh, C. Shen, M. Hanewich-Hollatz, C. H. Lee, D. M. Dorfman, J.
M. Karp, and R. Karnik, Affinity Flow Fractionation of Cells Via Transient
Interactionswith Asymmetric Molecular Patterns. Scientific Reports, 2013(3): p.
2329.
Y. Zhang and S. Neelamegham, Estimating the Efficiency of Cell Capture and
Arrest in Flow Chambers: Study of Neutrophil Binding Via E-Selectin and Icam1. Biophys J, 2002. 83(4): p. 1934.
Y. Zhang and S. Neelamegham, An Analysis Tool to Quantify the Efficiency of
Cell Tethering and Firm-Adhesion in the Parallel-Plate Flow Chamber. J
Immunol Methods, 2003. 278(1-2): p. 305.
L. S. Cheung, X. Zheng, L. Wang, R. Guzman, J. A. Schroeder, R. L. Heimark, J.
C. Baygents, and Y. Zohar, Kinematics of Specifically Captured Circulating
Tumor Cells in Bio-Functionalized Microchannels. Microelectromechanical
System 2010. 19(4): p. 752.
L. A. Tempelman and D. A. Hammer, Receptor-Mediated Binding of IgeSensitized Rat Basophilic Leukemia Cells to Antigen-Coated Substrates under
Hydrodynamic Flow. Biophys J, 1994. 66(4): p. 1231.
D. G. Swift, R. G. Posner, and D. A. Hammer, Kinetics of Adhesion of IgeSensitized Rat Basophilic Leukemia Cells to Surface-Immobilized Antigen in
Couette Flow. Biophys J, 1998. 75(5): p. 2597.
T. Yago, V. I. Zarnitsyna, A. G. Klopocki, R. P. McEver, and C. Zhu, Transport
Governs Flow-Enhanced Cell Tethering through L-Selectin at Threshold Shear.
Biophys J, 2007. 92(1): p. 330.
E. W. K. Young and D. J. Beebe, Fundamentals of Microfluidic Cell Culture in
ControlledMicroenvironments. Chemical Society Reviews, 2010. 39(3): p. 1036.
K. E. Norman, K. L. Moore, R. P. Mcever, and K. Ley, Leukocyte Rolling in- Vivo
Is Mediated by P-Selectin Glycoprotein Ligand-1. Blood, 1995. 86(12): p. 4417.
M. B. Lawrence, G. S. Kansas, E. J. Kunkel, and K. Ley, Threshold Levels of
Fluid Shear Promote Leukocyte Adhesion through Selectins (Cd62l,PE). Journal
of Cell Biology, 1997. 136(3): p. 717.
C. Dong and X. X. Lei, Biomechanics of Cell Rolling: Shear Flow, Cell-Surface
Adhesion, and Cell Deformability.Journal of Biomechanics, 2000. 33(1): p. 35.
L. Wu, B. T. Xiao, X. L. Jia, Y. Zhang, S. Q. Lu, J. Chen, and M. Long, Impact of
CarrierStiffness and Microtopology on Two-Dimensional Kinetics of P-Selectin
and P-Selectin Glycoprotein Ligand-1 (Psgl-1) Interactions. Journal of Biological
Chemistry, 2007. 282(13): p. 9846.
57
[26]
[27]
[28]
[29]
[30]
[31]
W. C. Chang, L. P. Lee, and D. Liepmann, Biomimetic Techniquefor AdhesionBased Collection and Separation of Cells in a Microfluidic Channel. Lab on a
Chip, 2005. 5(1): p. 64.
R. B. Bird, W. E. Stewart, and E. N. Lightfoot, TransportPhenomena, 1960, New
York: Wiley. p. 45.
A. J. Glodman, R. G. Cox, and H. A. Brenner, Slow Viscous Motion of a Sphere
Parallelto a Plane Wall. Ii. Couette Flow. Chemical Engineering Science, 1967.
22: p. 653.
http://physics.georgetown.edu/matlab/.
K. C. Chang and D. A. Hammer, The Forward Rate of Binding of SurfaceTethered Reactants: Effect of Relative Motion between Two Surfaces. Biophys J,
1999. 76(3): p. 1280.
D. J. Gee, L. K. Wright, J. Zimmermann, K. Cole, K. Soule, and M. Ubowski,
Dimethylsulfoxide Exposure Modulates Hl-60 Cell Rolling Interactions. Biosci
Rep, 2012. 32(4): p. 375.
58
3
Lateral Displacement of HL60 Cells Rolling on Inclined PSelectin Adhesive Bands'
3.1 Introduction
Understanding the cell rolling behaviors on the receptor bands is important to develop the
separation devices based on transient cell-receptor interaction. Chapter Three will discuss
the development of the substrates with the receptor bands and the studies of the cells
rolling over the patterned substrates.
Reversible binding between glycoprotein receptors known as selectins expressed on
vascular endothelial cell surface and the selectin-ligands expressed on the cell surface
result in cell rolling on the vasculature under shear flow [1-8]. Karnik et al have
demonstrated that a single inclined P-selectin receptor pattern can be used to direct the
trajectories of cells rolling HL60 cells on the surface [9]; this finding opens the
possibility of microfluidic devices for simple, continuous-flow, and label-free separation
and analysis of cells [10-12]. The concept of the separation device with multiple inclined
receptor bands has been shown in Chapter 1 (Figure 1-6). Such devices have the potential
for rapid diagnostic applications including detection of sepsis and inflammatory
conditions that alter cell rolling behavior [13-15], separation and enumeration of different
leukocytes from blood using rolling ligands with appropriate specificities [16], and labelfree analysis of the expression of cell surface ligands. Since cell rolling is non-invasive
and mimics the natural cell homing process, these devices are also promising for
separation or analysis of subpopulations of hematopoietic or mesenchymal stem cells and
enrichment of circulating rare cells [17, 18].
Figure 3-1 shows a schematic illustration of the trajectories of cells rolling on a patterned
substrate. Cells that track along the edges of these patterns are laterally displaced into the
adjacent buffer stream, potentially enabling their separation. Quantitative understanding
Part of the content in this chapter is based on the work published in Langmuir, 2011.
27(1), 240-249.
59
of the nature of cell rolling trajectories and lateral displacements on receptor-patterned
substrates is prerequisite for realization of such devices. Although Karnik el al have
demonstrated that HL60 cells could track along P-selectin, the effect of flow conditions
and pattern inclination angle on cell rolling trajectories was not studied [9]. Furthermore,
the distribution of edge tracking lengths and lateral displacements was not analyzed due
to insufficient data obtained from a single patterned edge. Therefore, how the parameters
of cell rolling relate to cell rolling trajectories and lateral displacements for such
asymmetric patterns is poorly understood. For example, it is unknown to what extent cell
rolling is affected by shear stress magnitudes or pattern inclination angles. Further, the
nature of the detachment of cells after tracking along an edge is not known, and is it not
yet established whether detachment of cells from such patterns is a random process that is
unaltered by interaction history with the pattern itself. Systematic study of cell rolling
trajectories along such well-defined receptor bands is therefore prerequisite for
addressing these issues and enabling the development of label-free devices for separation
or analysis of cells such as the device envisioned in Figure 1-6.
In Chapter III, cell rolling trajectories of rolling HL60 cells onto P-selectin patterned
substrate were studied. HL60 cell is widely used as a model to study leukocyte rolling
[19-23]. Its surfaces express a specific ligand named P-selectin glycoprotein ligand-1
(PSGL-1), which binds reversibly to the receptor P-selectin to enable rolling in vivo and
in vitro [19]. A technique based on microcontact printing (pCP) to pattern alternating
ptm-scale lines of adhesive P-selectin regions with passivating poly(ethylene glycol)
regions on a gold substrate was developed. The edge tracking lengths and rolling
velocities of HL60 cells on these patterned substrates within a flow chamber at different
edge inclination angles and shear stress magnitudes, and P-selectin incubation density
were quantified in this Chapter. The distribution of edge tracking lengths and observation
of re-attachment of cells were incorporated into a computational simulation tool
developed by Suman Bose (MIT) to predict the trajectories of cells on a patterned
substrate [10].
60
Figure 3-1 Illustration of a typical cell rolling trajectory along the receptor pattern
inclined an angle (a) to the fluid flow direction. The cell binds within the receptor line,
and rolls in the direction of shear flow toward the pattern edge. The cell then tracks the
edge to define an edge tracking length 1e, resulting in a net lateral displacement d, before
detaching to continue along the direction of fluid flow before possible reattachment and
rolling along a new receptor line. Cell rolling velocity v, along within the receptorfunctionalized line in the x-direction of fluid flow can be distinguished from the velocity
ve along the line edge, where ve,,y is lateral velocity (the vertical component perpendicular
to the streamlines and parallel to the lateral displacement, d).
3.2 Fabrication of P-selectin-Patterned Substrates
3.2.1
Materials
Polydimethylsiloxane (PDMS; Sylgardl 84) was purchased from Dow Corning. Goldcoated glass slides were purchased from EMF Corp. All slides were cleaned with piranha
solution prior to use (3:1 mixture of sulfuric acid (Sigma-Aldrich) to 30% hydrogen
peroxide
(Sigma-Aldrich)).
(l-Mercaptoundec-11-yl)tetra(ethylene
glycol)
(PEG
alkanethiol; HS(CH 2)1 (OCH 2 CH 2)40H; Sigma-Aldrich) was diluted in absolute ethanol
(Pharmco-AAPER) at a concentration of 5 mM for microcontact printing. Recombinant
human P-selectin (R&D Systems Inc.) and bovine serum albumin (BSA, Rockland
Immunochemicals, Inc.) were diluted in 150 mM NaCl Dulbecco's phosphate buffered
saline (DPBS, Mediatech Inc.). All materials employed in this study were used without
further purification unless otherwise specified.
3.2.2
Fabrication of PDMS Stamp
Microcontact printing stamps that defined the receptor pattern were fabricated in PDMS
using an SU-8 molding process. The small line-patterned stamp (SS) (15 ptm line width
61
and 10 pm spacing between adjacent lines) was used to characterize the patterning
process. The medium line-patterned stamp (MS) (70 gm stamping regions spaced 50 pm
apart) and the large line-patterned stamp (LS) (100 gm stamping regions spaced 100 jim
apart) were used study cell rolling behavior.
3.2.3
Fabrication of Patterned Substrates
A schematic diagram of the patterning process is shown in Figure 3-2. Step 1:
Microcontact printing (pCP) was used to form a alternating self-assembled monolayer
(SAMs) of PEG molecules on the gold substrate. The PDMS stamp was first inked with
PEG solution in ethanol (5 mM), dried, and pressed onto the surface to be patterned for
40 s. The surface was then rinsed with ethanol and dried under a stream of N2. Step 2:
After selective deposition of PEG molecules, the substrates were incubated in P-selectin
solution (15 jg/mL in DPBS, unless stated otherwise) using a perfusion chamber
(Electron Microscopy Sciences) for 3 h at room temperature to pattern the remaining
areas with P-selectin. The surfaces were then backfilled with BSA (1 mg/mL in DPBS)
for 1 h to block non-specific interactions.
Wstep
1
PEGG
PEG
Step 2
P4kd
-eed
PEG moeueP-selectin
Figure 3-2 Schematic diagram for patterning of P-selectin on a gold substrate
involving microcontact printing. Step 1: Selective deposition of PEG molecules on the
gold surface. Step 2: Filling in of the uncoated surface with P-selectin.
Direct microcontact printing (sCP) of proteins has been used widely to control the
geometry of protein patterns on various planar surfaces [24-29], including printing of Pselectin to study neutrophil rolling [30]. However, it is possible that the protein will
denature or lose bioactivity during the PDMS stamping process [31]. Additionally,
transfer of the stamp material (PDMS) from the stamp to the surface during pCP can
contaminate the patterned areas [32]. In the present method, after selective deposition of
62
PEG molecules, the gold substrate is patterned with P-selectin in the solution phase so the
possibility of denaturation due to protein drying can be ruled out.
3.2.4
Substrate Characterization
Atomic force microscopy (Veeco Dimension 3100; Tapping mode; scan rate: 1 Hz) and
scanning electron microscopy (JEOL 6700; acceleration voltage 3.5 kV) were used to
characterize the patterned surface geometry. All substrates for AFM and SEM
characterization were placed in a vacuum chamber overnight before imaging to minimize
residual solvent on the surface; no further coating was employed for SEM imaging.
AFM images of P-selectin patterned using the small line-patterned stamp (SS) show
clearly defined 10 im-wide lines of P-selectin with well-resolved, straight edges (Figure
3-3 (a), (b)). The medium line-patterned stamp (MS) was used to prepare surfaces for
studying edge track lengths and rolling velocities. The sharp contrast between P-selectin
regions and PEG regions confirms that the resulting patterns had well-defined edges over
large areas as revealed by SEM (Figure 3-3 (c), (d)).
Figure 3-3 Characterization of P-selectin patterned substrates. AFM images of 10
[tm wide P-selectin lines separated by 15 ptm wide PEG bands (after step 2), displaying
63
the contrast between P-selectin and PEG regions in (a) height and (b) phase, respectively.
The phase image indicates a difference between the mechanical properties of the surface
in the two regions. SEM images of surfaces after PEG printing (c) (step 1) and after Pselectin adsorption (d) (step 2), respectively, showing uniformity of patterning (brighter
areas correspond to PEG regions). Scale bars are 5 pm in (a) and (b) and 100 tm in (c)
and (d).
3.3
3.3.1
Cell Rolling on Asymmetric Patterns in a Flow Chamber
Cell Culture
HL60 myeloid cells (ATCC) were maintained in Iscove's Modified Dulbecco's Medium
supplemented with 20 % fetal bovine serum (FBS) as per ATCC recommendation. Cells
were cultured in 75 cm 2 polystyrene tissue culture flasks (BD Falcon) with the cell
density maintained between 105-106 cells/mL (passage number 26, 27 and 29 were used
in this study).
3.3.2
Experiment Setup
A suspension of HL60 cells was flowed over the patterned surfaces in a rectangular flow
chamber (Glycotech, Inc; width w = 1.0 cm; length = 6 cm; height h = 0.005 inch) as
shown in Figure 3-4 with inclination angles of the receptor pattern of either a = 50, 100,
150 or 200 (see Figure 3-1) at room temperature of 24.5"C. A syringe pump (World
Precision Instruments (WPI), SP230IW) was used to generate different flow rates
between 75 and 300 pL/min, with corresponding shear stresses of 0.5-2.0 dynes/cm 2
(~0.05 to 0.2 Pa). Flow was laminar (Reynold's number Re ~ 0.1-3) and shear stress r
was calculated using the plane Poiseuille flow equation T = 6pQ/wh 2 , where p is the
viscosity, Q is volumetric flow rate, w is width of the flow chamber, and h is height of
the flow chamber. For confirming the non-fouling properties of PEG, the injected cell
concentration was ~106 cells/mL (high density). For the studies of edge tracking lengths
and rolling velocity, the injected cell concentration was ~105 cells/mL (low density). An
inverted microscope (Nikon TE2000-U) with a mounted camera (Andor iXon 885) was
used to record HL60 rolling interactions with adhesive P-selectin substrates using a 4x
64
objective, typically at a rate of 1 frame per second for durations of 300 s with exposure
time of 0.01-0.05 s. For each shear stress magnitude, pattern inclination angle, and Pselectin concentration, three independent experiments were performed. Data are
presented as mean and standard deviation of the average values obtained from each
experiment.
Figure 3-4 Rectangular flow chamber for cell rolling on multiple asymmetric
pattern study. The flow chamber was attached to the P-selectin-patterned substrate via a
vacuum manifold to assemble the device. The thickness of the gasket determines the
height of the chamber, here, height = 0.005 inch).
3.3.3
Results and Data Analysis
Poly(ethylene glycol) (PEG) has been widely used in developing surfaces that prevent
fouling coming from the proteins [33, 34]. Although previous studies suggests that only
the high molecular weigh PEG (MW > 3500 Da) can resist protein adsorption [35], Prime
and Whitesides demonstrated that self-assembled monolayers (SAMs) made from shortchain alkanethiols such as HS(CH 2)1 1(OCH 2CH 2 )nOH can be used in designing highly
protein-resistant surfaces and they also reported that the number of the ethylene oxide
groups can be as few as two [36, 37]. In this chapter, HS(CH 2)1 1(OCH 2CH 2)40H was
used. To confirm the anti-fouling properties of PEG, a suspension of HL60 cells with
high density (10
6 cells/mL)
was flowed over the patterned surfaces (prepared by LS) in
the flow chamber. The HL60 cells were observed to roll specifically in the P-selectin
patterned regions as shown in Figure 3-5. These results confirm that the PEGfunctionalized regions on either side of P-selectin lines could block P-selectin adsorption
and that P-selectin molecules retained their activity after being adsorbed to the exposed
gold. In addition, experiments to characterize cell adhesion on PEG surfaces and surfaces
65
coated with BSA were performed, and none of cell-surface interactions was observed,
further confirming that the observed cell-rolling interactions were due to P-selectin.
PEG region
P-selectin region
Figure 3-5 The interaction of HL60 cells with P-selectin/PEG patterned substrate in
the flow chamber. The substrate was prepared using large line-patterned stamp (LS).
The edge inclination angle was 100 and the shear stress was 0.5 dyn/cm 2. The HL60 cells
only interact with the P-selectin patterned region, not PEG patterned region, which
confirms the anti-fouling property of PEG.
For studies of the effects of edge inclination angle, shear stress, and the P-selectin
incubation density on the edge tracking lengths and rolling velocities, the patterned
substrates were prepared by medium line-patterned stamp (MS). The P-selectin patterned
substrates were incorporated into a flow chamber for studying the rolling behavior of
HL60 cells under different condition. The injected cell concentration was ~105 cells/mL
(low density) to prevent cell-cell interaction
Tracking of Cells. The image sequences were analyzed using a customized Matlab
(Mathworks, Inc.) program that utilized a particle tracking freeware [38] to detect the
cells and generate tracks along the patterned line edges. The procedure used to generate
tracks has been described in Karnik et al's previous work [39]. A tracking criterion that
required the cell displacement between consecutive frames was set to be less than 10 pim.
Tracks with total length <40 ptm were not included in analysis of the distribution of cell
responses, as these short tracks predominantly represented artifacts on the substrate such
66
as pinholes and in some cases unlinked fragments of a single track. These settings
allowed for selective tracking of only the rolling cells, since velocities of the non-adhered
freely flowing cells were much higher. Tracks generated by the software were randomly
selected and inspected manually by comparing with the images to ascertain their
accuracy. Figure 3-6 shows an example of actual tracks of HL60 cells rolling on Pselectin patterns at an edge angle of 100, shear stress of 0.5 dyn/cm 2 , and incubation
concentration of 15 [tg/mL. The tracks clearly depict the change in direction of rolling
HL60 cells that occurred when the cells encountered the inclined edges.
Identification of Tracks that Encountered an Edge. Positions of the patterned edges
were identified using the difference in contrast between the PEG and P-selectin regions
as imaged using optical microscopy. The procedure used to identify tracks that
encountered the edge was developed by Suman Bose and described in Langmuir 2011,
27(1) 240-249 [10]. Tracks with endpoints within 10 pm of the nearest edge were
identified as having encountered an edge. As shown in Figure 3-6 inset, the tracks were
subdivided into two segments at intersection point Pi - one that represented rolling inside
the P-selectin line (green line), and another representing rolling on the edge (black line).
The length travelled along the patterned line edge was defined as edge tracking length, 4,
and calculated from the distance between the Pi and the end of the track. Rolling velocity
was calculated by dividing the length of the segment by the time taken to traverse it. For
each experiment, an array of edge tracking length (le), velocity on edge (vp) and velocity
on plain region (vp) with one data set corresponding to each track was generated and used
for further analysis.
67
Figure 3-6 Tracks of HL60 cells rolling on P-selectin patterns. The tracks (blue lines)
on P-selectin lines (pink) were obtained by analyzing 300 images acquired at 1 fps using
a customized Matlab code. The edge inclination angle was 100 and the shear stress was
0.5 dyn/cm 2 . Inset shows a track corresponding to a cell that first rolled inside the Pselectin line (green) in the direction of fluid flow and then tracked along the edge (black).
Hopping was observed in our study: After cells detached form the edges of the band,
most of them will reattach onto successive downstream band. However, the customized
Matlab program used in this study can only identify the tracks of the cells on a single
band but not multiple bands. The following studies of the effect of substrate parameters
including edge angle, shear stress, and P-selectin incubation density on the rolling
behavior of HL60 cells are based on the data extracted from the tracks on a single band.
3.3.3.1
Effect of Edge Angle on the Rolling Behavior of HL60 Cells
The effect of the edge inclination angle a subtended by edges of the P-selectin lines with
respect to the direction of fluid flow on the edge tracking length 4, the lateral
displacement d, and the rolling velocities v, an ve, respectively, is shown in Figure 3-7. At
a=5', HL60 cells rolled an average distance of more than 135 pm along the edges before
detachment at a shear stress of 0.5 dyn/cm2 . Figure 3-7(a) shows that, as the edge angle
was increased from 5* to 20' in 50 increments, the average edge tracking length 1e
68
decreased significantly (ANOVA, F = 18.403, p = 0.001). In other words, the ability of
the cells to roll along the edges was reduced with increasing edge inclination angle (and
increasing component of the fluid force on the cell directed away from the edge).
Comparison of data pairs (post-hoc t-test) demonstrated
statistically
significant
differences in 1e for every 5' increase in a. In contrast, Figure 3-7(b) shows that the
lateral displacement d
(ANOVA, F = 3.075, p
=
=
1e sina, did not show a significant trend with increasing a
0.091), and varied from 7.0 to 12.5 pm over this range of a.
However, a statistically significant difference was observed between the lateral
displacements at a =10 and 200 (post hoc t-test). This behavior can be understood in that
although sina increases with increasing edge inclination angle, edge tracking length 1e
concomitantly decreases. It is the magnitude of this lateral displacement that is more
relevant to separation of cells by rolling on such a patterned substrate.
In addition to altering the direction of cell rolling, asymmetric receptor bands can also
alter the rolling velocity of the cells [39]. To examine the effect of a on rolling velocity,
we quantified the average rolling velocity of cells within and along the edges of the Pselectin lines as a function of a at a fixed shear stress magnitude of 0.5 dyn/cm 2 (Figure
3-7(c)). The rolling velocity within the P-selectin line was in the range of v, = 2.9 - 3.6
tm/s, and was always less than that along the edge regions (ve = 4.4 - 6.0 pm/s). This can
be understood in terms of the expectation that, as a increases, surface area of interaction
and adhesion resistance to rolling between the cell and the surface decreases, leading to
an increase in rolling velocity. Pairwise (t-test) statistical analyses show a significant
increase in ve at large edge angles (15' and 20') compared to v,, consistent with previous
observations [39]. The average rolling velocity on the edge ve increased from 4.4 ptm/s
for an edge inclination angle of 5' to 6.0 pm/s at an edge inclination angle of 200, though
this trend did not reach a degree of statistical significance (ANOVA, F = 3.55, p=0.067).
In contrast, ve,, (lateral velocity, defined previously as the edge velocity component in the
direction of lateral displacement d) increased significantly from 0.4 pm/s to 2.1 pm/s as
a increased from 5" to 200. Thus, receptor bands characterized by large edge inclination
angles (a=15', 20') led to greater lateral displacement of cells over a given rolling
duration.
69
(a)
(b)
300
p= 0.0055
w 250
, 20
p= 0.0697
2
p= 0.01
D=0.0423
15
200
p= 0.007
150
a.
100
p= 0.027
s0
10
5
0
5
(C)
10
15
Edge Inclination angle, a ()
V.
p=
p.
P80.16
p3.0O88
r_
=L
5
(d)
8
6
20
4
p=0v19p= 0.043
0.11
~
15
10
Edge Incdhon angle, a()
0
pu 0.0005
3
p=
4
71
20
0.0007
p= 0.0023
p= 0.024
2
pp=0.017
0
1
2
A
0
5
10
15
20
Edge inclination angle, a (*)
5
10
15
20
Edge inclination angle, a (*)
Figure 3-7 Effect of edge inclination angle a on rolling behavior of HL60 cells at a
fluid shear stress magnitude of 0.5 dyn/cm2. Variation of (a) edge tracking length, le;
(b) lateral displacement, d; (c) rolling velocities v, and v, within the P-selectin lines and
on the edge, respectively; and (d) lateral velocity, ve,, (component of the edge rolling
velocity in the direction of lateral displacement). Error bars represent one standard
deviation, with n = 3 replicate experiments for each condition.
3.3.3.2
Effect of Shear Stress on Rolling Behavior of HL60 Cells
Figure 3-8 summarizes the edge tracking length l,, the lateral displacement d, and the
rolling velocities v, an v, as a function of the shear stress (,r = 0.5 dyn/cm 2 to 2.0
dyn/cm 2 ) at a fixed edge angle of 5*. The edge tracking length le varied in the range of
118.6-173.1
im over r = 0.5 to 2.0 dyn/cm 2 . However, there was no statistically
significant effect of shear stress on le (ANOVA, F=2.119, p=O.176) (Figure 3-8(a)). The
lateral displacement d varied between 10.3-15.1
ptm, again with no statistically
significant dependence on the shear stress (Figure 3-8(b)). Similarly, Figure 3-8(c) shows
that the rolling velocity within the P-selectin lines and on the edge also did not vary
70
significantly with shear stress (ANOVA, p=0.917 and p=0.165, respectively). This lack
of dependence on shear stress is not surprising, given that cell rolling involves
mechanisms at the cellular and molecular levels to regulate the rolling response over a
range of shear stresses [40]. Similar independence of rolling velocity with shear stress has
been observed before in the case of HL60 cells rolling on E-selectin [41, 42]: the rolling
velocity of HL60 cells has been observed to increase with shear stress at low shear stress
(r <0.5 dyn/cm 2 ) and reach a plateau at higher shear stresses. Our experiments indicate
that similar to rolling within the P-selectin line, shear stress also does not have a
significant effect on the rolling behavior of HL60 cells on asymmetrically patterned
edges within x = 0.5 to 2.0 dyn/cm 2.
(b)
(a)
250
200
.- ZU
E
15
150
10
k
100
5
~50
W
0
0
0.5
(c)
8
6
0.5
2.0
1.0
1.5
Shear stress, T (dyn/cm2)
(d)
o.*p-0.033
V13
Ip. 0.088
p 0.012
4
p= 0. 3
1.5
2.0
1.0
2
Shear stress, Y(dyn/cM )
0.6
0.5
~
0.4
0.3
4
0.2
S2
0.1
-'0
0.5
1.5
1.0
Shear stress, Y (dyn/cm 2)
0.5
2.0
1.0
1.5
2.0
2
Shear stress, t (dyn/cm )
Figure 3-8 Effect of shear stress T on rolling behavior of HL60 cells at an edge
inclination angle of 5'. Variation of (a) edge tracking length, 1,; (b) lateral displacement,
d; (c) rolling velocities v, and v, within the P-selectin lines and on the edge, respectively;
and (d) lateral velocity, ve,,y. The effect of shear stress on rolling behavior of HL60 cells is
not statistically significant as observed in (a), (b), (c) and (d). Error bars represent one
standard deviation, where n = 3 replicate experiments for each condition.
71
3.3.3.3
Effect of P-selectin Incubation Concentration on Rolling Behavior of HL60
Cells
Apart from shear stress and edge inclination angle, the P-selectin density on the surface
may be expected to affect the trajectories of cells rolling on the inclined edges. Increasing
P-selectin density on a glass surface under high shear stresses (~20 dyn/cm 2) has been
observed to decrease the rolling velocity of HL60 cells, but receptor density did not
modulate the rolling velocity under low shear stress (< 2 dyn/cm 2 ) [21]. When the Pselectin incubation concentration was varied in the range of 5 to 30 ptg/rnL while
maintaining an incubation time of 3 h, no adhesion (5, 10 pg/mL), non-uniform rolling
(11, 12, 13, and 14 pg/mL), robust rolling (15, 20, and 25 gg/mL), and firm adhesion at
30 pg/mL for a shear stress of 0.5 dyn/cm 2 were observed. Thus, the range of P-selectin
incubation concentrations that resulted in a useful rolling response was 15 to 25 pg/mL.
Therefore, the edge tracking length and cell rolling velocities in this range of P-selectin
concentrations were then characterized as shown in Figure 3-9. Interestingly, there was
no significant change in rolling velocity with change of P-selectin incubation
concentration: the average rolling velocities inside the bands (vp) were 2.78±0.57,
2.44±0.64, and 2.67±0.67 pm/s for P-selectin concentrations of 15, 20, and 25 pg/mL at
shear stress of 0.5 dyn/cm 2 , respectively. Similarly, there is no significant change in the
cell behavior along the edge including edge tracking length 1e, lateral displacement d,
edge rolling velocity ve, and lateral velocity ve,,y.
These results are in agreement with
previous observation of less change in rolling variation with a variation of ligand density
under low shear stress (< 2 dyn/cm 2 ) by Dong et al [21]. These results indicate that when
P-selectin is directly immobilized on a gold substrate, the rolling behavior along the edge
cannot be controlled as easily as that by changing the edge inclination angle.
72
(a)
(b)
~80
a= 100
a= 200
~60
15
12
E a= 100
*= 200
n
9
CD 40
6
20
3
0
25
20
15
P-selectin Incubation concentration (pg/mL)
15
20
25
P-selectin Incubation concentration (pg/mL)
(c)
(d)
8
3
o=10
a a= 10*
= 20'
a =20*
7A
2
0
0
25
20
15
P-selectin incubation concentration (pg/mL)
25
20
15
P-selectin incubation concentration (pg/mi)
Figure 3-9 Effect of P-selectin incubation concentration on rolling behavior of HL60
cells. Variation of (a) edge tracking length, le; (b) lateral displacement, d; (c) rolling
velocities v, and v, within the P-selectin lines and on the edge, respectively; and (d)
lateral velocity, ve,y at edge inclination angles a = 100 and 200 and shear stress of 0.5
dyn/cm 2 . Error bars represent one standard deviation, where n = 3 replicate experiments
for each condition.
3.3.3.4
Detachment of Cells Rolling Along an Edge Can be Described by a Poisson
Process.
While the average edge tracking length and lateral displacement are useful to elucidate
the effects of edge angle, shear stress, and P-selectin incubation concentration on cell
rolling, knowledge of the distributions of the distance rolled along the edge and the
lateral displacement is important for predicting the eventual distribution of a population
of cells in a separation device. Similarly, this knowledge is required to elucidate the
number of edge tracking events that must be observed to distinguish between cell
phenotypes by observing rolling on the patterned substrates. Therefore, the distribution of
the edge tracking lengths was examined, with the aim of developing a model that would
73
serve as a tool to predict cell rolling trajectories and their spread, with direct implications
on analysis resolution and separation efficiency.
For all the experiments conducted, distributions of the edge tracking length exhibited a
decaying exponential characteristic similar to that expected from a Poisson process as
shown in Figure 3-10. Poisson process occurs when individual events within the process
are random in nature with uniform probability of occurrence independent of history. It is
well known that cell rolling is a stochastic process involving discrete adhesive
interactions between the cell and the surface [43-46]. Rolling cells continuously form
adhesive contacts with the surface with receptors localized at the tips of extensible
structures known as microvilli [47]. In the case of cells rolling along an edge, the region
of overlap between the cell's contact area and the receptor-coated region is the site where
new cell-surface adhesions are formed. As the cell tracks along an edge, new cell-surface
adhesions are continuously formed in the region of overlap between the cell's contact
area and the receptor-coated region. The cell will detach from the surface when a new
adhesive interaction fails to form before the last adhesive interaction is broken under
shear flow. If the probability of formation of new adhesive interactions in a given period
is constant (i.e., independent of the past rolling history of each cell and/or uncorrelated
among cells within a population), detachment is expected to be a random event and the
distance traveled by each cell along the edge is expected to follow a Poisson distribution.
To test this hypothesis, the cumulative distribution function (CDF) was calculated from
the data and fit it with a Poisson distribution (Figure 3-10) given by:
C(le)= I-exp(
L
(Eq. 3-1)
Here, k is the mean value of the Poisson distribution. In the ideal case where detachment
is a random process, k will approach the mean value of the edge tracking length. The
CDF does not involve any arbitrary bin widths and therefore allows for an objective
comparison of the actual distribution of the edge tracking lengths with that predicted for a
Poisson process (Eq. 3-1). The CDFs displayed in Figure 3-10 do not begin at the origin
74
because edge tracking lengths <10 pm could not be resolved by the present video frame
rate and magnification; these unresolved lengths were used for calculation of the CDF
itself but were not used for fitting the Poisson function. The Poisson distribution
described by Eq. 3-1 well fit the CDF for all edge inclination angles considered, and the
mean value of k obtained by fitting to the CDF also accurately predicted the observed
edge tracking length histograms (Figure 3-10(a), inset) at different edge inclination
angles. Additionally, to confirm that the process of detachment of cells from the edge was
indeed a Poisson process for a given cell, k under different experimental conditions was
calculated and it was compared to the measured average edge tracking lengths (Table 31). A strong correspondence between k and the average edge tracking length was found
from the experiments; even at different shear stress magnitudes, the maximum difference
was found to be within 7%. These data suggest the fact that detachment of these HL60
cells rolling along an edge is indeed a random process.
Interestingly, the average value of k obtained by fitting the CDF to the Poisson
distribution showed an exponential decrease with increasing edge inclination angle
(Figure 3-10(b)). This observation provides an empirical relation between the edge angle
and the average edge tracking length, while the distribution of the lengths itself can be
modeled using a Poisson process. The Poisson mean values
'
obtained from the
empirical fit (in Figure 3-10(b)) are summarized in Table 3-1 and they were compared to
the measured average edge tracking lengths. The maximum difference was found to be
within 5% which suggests that this exponential dependence combined with modeling cell
detachment as a Poisson process can be used to interpolate the average edge tracking
length as well as its distribution for any edge angle between 5' and 20'.
The prediction of the variation of lateral displacement with edge inclination angle was
made based on this exponential fitting that agreed well with the experimental data (Figure
3-10(c)). Interestingly, the curve predicts an optimal edge inclination angle between 50
and 100 that maximizes the lateral displacement for rolling along a single edge. This
prediction explains why a statistically significant difference was not observed between
lateral displacements at edge inclination angles of 5' and 100 (Figure3-6). The average
75
lateral displacement exhibits a maximal value at the edge inclination angle between 5'
and 10'.
(a)
-
1.0
2G .
W0
1150
I
U
20
40
10
20
0
0.5
150
-
C.)
0
200
400
1V
200
200
15*
400
2W
1501
100
100
so
50
W
0
Edge Ucking length,
0.0
200
0
400
800
Edge traddng length, Is (Jim)
(b)
(c)
logX = -0.126a+5.489
R2=0.9%62
100
100
)
1000
800
15
10F
5
O Experkmental data
Pmedictio from Poisson process
--
10
50
le (Am)
0
5
15
10
Edge inclination angle, a 0
0
20
5
20
10
15
Edge inclination angle, a o
25
30
Figure 3-10 Detachment of cells rolling along an edge is well described by a Poisson
distribution. (a) Cumulative distribution function of edge tracking lengths l, (filled
triangles) was fitted to a Poisson distribution described by Eq. 3-1 (solid lines). Insets
show the frequency distribution of the experimentally measured edge tracking lengths,
along with that predicted by the Poisson distribution fit to the CDF (solid lines). Colors
indicate different inclination angles a of the receptor pattern. Representative results are
shown for only one experiment for each a. (b) Variation of the average value of k with
edge inclination angle is well described by a linear fit on a semi-log plot (solid line). (c)
Variation with the edge inclination angle of the average value of the lateral displacement
(solid line) obtained from the empirical fit shown in (b) along with the experimental
results (open circles). Error bars in (b) and (c) represent one standard deviation. Shear
stress is 0.5 dyn/cm 2.
Table 3-1 Comparison of the experimental average of 1e, Poisson average value of k,
and the Poisson mean value ' from the empirical fit. n = 3 replicate experiments.
76
Data were obtained by fitting to a Poisson distribution with experimentally measured
average edge tracking length. Shear stress is 0.5 dyn/cm 2 .
Experimental
Poisson mean value
Edge angle
average
Poisson mean
% Error
from the empirical fit
% Error
a (0)
/,±SD (jim)
A±SD (pm)
100(A-I./)I. (pm)
A'(pm)
100(A'-,/)I. (pm)
5
135.5±36.3
134.1±33.2
-0.98%
128.9
-4.87%
10
72.2±17.2
66.2±21.7
-8.35%
68.6
-4.92%
15
37.2±7.6
34.7±9.4
-6.60%
36.6
-1.72%
20
20.3±3.7
20.3±3.9
0.11%
19.5
-4.08%
Prediction of cell trajectories on a receptor-patterned substrate. The experimental
results obtained in this study enable development of a simulation tool for predicting the
trajectories and distribution of cells rolling across a substrate patterned with asymmetric
P-selectin lines. The experimental observations were incorporated into a simulation of the
lateral displacement of HL60 cells as they encounter multiple edges of P-selectin. A
Monte Carlo simulation was developed and performed by Suman Bose; the detailed
procedure is described in Langmuir 2011, 27(1) 240-249 [10]. The simulated results
agreed well with the experimental distribution as shown in Figure 3-11(a) despite the low
number of experimental data points available. Although a comparison of distribution over
more number of edges is desirable, the study was limited by the field of view of the
camera, which could only accommodate approximately three edges of the patterned
bands.
A second Monte Carlo simulation was performed by Suman Bose again to examine the
net lateral displacement and resulting distributions of the cells (injected at a particular
location (0, 0)) at a longitudinal distance of 1 cm downstream of the origin (similar to
Figure 1-6). The detailed procedure of this simulation can be found at Langmuir 2011,
27(1) 240-249 [10]. In the simulation study, the device was assumed to be patterned with
parallel P-selectin lines having a width of 50 im and a gap of 70 ptm between adjacent
lines (similar to the pattern used in experiments). The cells were assumed to roll on all
the P-selectin lines that they encountered with the edge tracking lengths described by a
Poisson distribution with the mean I as obtained earlier. Figure 3-11(b) summarizes the
77
probability density functions of the net lateral displacement, defined here as the sum of
all displacements that a given cell acquired over the longitudinal travel distance of 1 cm.
The net lateral displacement increased with increasing edge inclination angle a from 50
to 150, with little change as a increased from 15' to 200. The net lateral displacement
showed maximum sensitivity to the edge inclination angles between 50 to 100. Increasing
the edge inclination angle a increases the total number of edges encountered by the cells,
but simultaneously decreases the mean lateral displacement on each edge (ksina) (Figure
3-10(c)). These opposing effects resulted in the net lateral displacement exhibiting an
increase till a = 150, and little change between 150 and 200. With increasing number of
edges encountered, the distribution of cells is seen to approach a Gaussian distribution,
consistent with the central limit theorem. Interestingly, although a = 100 and 200 yield
similar net lateral displacements, it is noteworthy that the Gaussian spread is smaller for
a = 200 due to the larger number of edges encountered. It is noteworthy that although the
distribution of cells approaches a Gaussian form, it is not exactly Gaussian as some of the
cells in the distribution are rolling along the edge; these cells create "spikes" in the
distribution at specific positions where the edges intersect the axis (at a longitudinal
travel distance of 1 cm).
78
(a)
0.05
0aSimulated Result
0.04
51
V
Expreimental Result
0.03
0.02
(L
0.01
LL-I
-
0.00
20
0
rm
100
60
80
40
Net lateral displacement (pm)
(b)
120
50
(81.64 ± 21.95 pm)
0.01
0.00 I-"
I
0.01
0.00 I
100
(157.34± 38.30 pm)
150
(184.81± 37.18 pm)
0.01
C.
0.00 i
200
(190.89±33.59 pm)
0.01
0.00
0
50
200
250
100
150
Net lateral displacement (pm)
300-300
Figure 3-11 Prediction of cell trajectories on a receptor-patterned substrate (a)
Probability distribution of net lateral displacement of HL-60 cells after rolling on three
consecutive bands of P-selectin patterns as obtained from Monte Carlo simulations
(shaded area) and experimental observations. a = 200 (shear stress 0.5 dyn/cm 2). (b)
Prediction of the downstream distribution of HL60 cells rolling on patterned P-selectin in
a separation device. (Courtesy of Suman Bose)
The study of the effects of substrate parameters on the cell rolling trajectories has led to
realization of a microfluidic device with multiple P-selectin bands of edge inclination
79
angle of 150 and this device has been demonstrated to be able to separate neutrophils
directly from the blood [12].
3.4 Optimization of Device Design
Next, the results of cell rolling parameterization are used to predict optimal design
features of the cell separation device that will maximize separation of a cell
subpopulation that binds and rolls transiently along the inclined band pattern. Chapter II
discussed how cells flowed inside the microchannel and settling distance (xs) was
identified as a function of shear stress. An empirical equation was developed to predict
cell capture probability from the free stream of fluid, as shown in Figure 3-12 (a) (left). In
the first part of Chapter III, the edge tracking length for rolling cells on a single adhesive
band was characterized. The effective chamber length (xe) was defined as the chamber
length required for settled cells to be captured by the receptors and then deflected as
shown in Figure 3-12 (a) (right). Both settling distance (x,) and effective chamber length
(xe) should be considered when chamber length is designed. Based on experimental data
from Chapter II and Chapter III, device capture efficiency (Pd), total lateral displacement
(d,), and chamber length for a P-selectin-patterned microfluidic system can be predicted.
The prediction enables study of sensitivity of device design parameters including edge
inclination angle (a), band width (w), and shear stress
80
(T).
pf= capture probability
fron free stream by an
incli ned pattern
(a)
x,= settling distance
L
(b)
4
V
x, = effecttve chamber length
L =chamber length= x,+x,
Pd= device capture efficiency
2LJ
W
"P
dt=total lateral
displacement
Device design (a, w, g, r)
Figure 3-12 Optimization device design. (a) Schematic diagram of cell settling,
capture, rolling on the patterns. (b) Schematic diagram of a device. Device capture
efficiency, total lateral displacement, and required chamber length can be modeled based
on different device design parameters.
Device performance such as device capture efficiency (Pd), and total lateral displacement
(dt) depends on two features: the number of the bands comprising the inclined pattern of a
microfluidic device, and the capture efficiency of the cells by receptor bands. The
number of the bands (n) inside a microfluidic device can be determined by:
n ~Xe
x
(Eq. 3-2)
sin a
where
x,
is the effective chamber length, w is the band width, g is pattern gap, and a is
edge angle as shown in Figure 3-12(a).
81
Device capture efficiency (Pd) is the capture efficiency of cells by a device or capture
efficiency by "n" number of bands patterned inside the device and is defined as:
pd
=1-(Ipf)"
(Eq. 3-3)
where pf is the capture probability from the free stream by a single receptor band. The
empirical equation (Eq. 2-7) developed in Chapter II can be applied to calculate pf here.
To study total lateral displacement of a cell interacting with multiple bands, hopping of
cells from one band to the next needs to be considered. This unique hopping behavior
may affect the capture efficiency of the cells by receptor bands. The effect of hopping
behavior can be studied by considering two extreme cases: (1) The first case is not to
consider hopping, which means recapture probability by downstream bands (p,) is equal
to capture probability from the free stream (pr). (2) The second case is to consider
hopping and assume p,= 1.0 as shown in Figure 3-13 (a). This case implies that all cells
hop to other downstream bands when detached from upstream bands
When hopping events are not considered, the capture probability by any one of the bands
inside the device is pf. The capture probability byj bands when a cell is travelling inside
a device of n bands is Ci x pfj x (1- pf)"-1 . Total lateral displacement (d) when
hopping events are not considered can be derived as:
d, =< d>x (>Cj
x Pfj'x (1 - Pf)n-ji
xj
f
(j=0
(Eq. 3-4)
where <d> is the average lateral displacement of a cell along a single band edge. The
value of <d> assumed here was obtained from experimental observations discussed in
this chapter. C' representsj combinations of n elements.
Assuming cells will continue hopping and be recaptured by successive downstream bands
after they detached from the edge of upstream bands (p, = 1.0), the probability of
82
captured by the first band is pf and the capture probability by ith band is (1- Pj )I1 X p1 .
The number of the bands that a cell will roll on after it is captured by
ih
band is (n +1- i)
. Therefore, the total lateral displacement (d,) can be derived as:
d, =< d > x ( -p)''
xpf
(Eq. 3-5)
x(n +1-i)
i=1
For the device design of band edge inclination angle a = 150, band width w = 10 pm, gap
width g = 10 gm and shear stress
T =
0.25 dyn/cm 2 , the probability of attachment of cells
from the free stream can be predicted to be 0.03 from Eq. 2-7. Considering two cases: (1)
P, = 1.0 and (2) pr = pr = 0.03, the predicted total lateral displacements at different
effective chamber length are shown in Figure 3-13(b). It is clear that the hopping plays a
critical role in enhancing the overall lateral displacement. The total lateral displacement
can increase by several orders of magnitude depending on the chamber length.
(a)
P,= recapture probability
by downstream pattern
(hopping probability)
P1 = cap ture
probab ility from free
stream by an pattern
dt, total lateral
displacement
"W %VWIMWU"=IV
VW
(b)
E
I
-" 6000
"..
4000
C'
2000
Pr Pf= 0.03
--%185gm
0
PS
0
10
20
30
40
Effective Chamber Length, x, (cm)
83
50
Figure 3-13 (a) Schematic diagram showing cell hopping onto downstream band. (b)
Variation of effective chamber length on total lateral displacement when hopping
events were (pr = 1.0) and were not (pr = pf = 0.03) considered for the pattern design
of a = 150, w = 10 ptm, g = 10 pIm, and T = 0.25 dyn/cm 2 .
The objective of this section is to study optimization of the device design (a, w, and T) by
integrating experimental data into device performance prediction. Since cell hopping
among adhesive bands was observed from the experiments with HL60 cells in the flow
chamber (Section 3-3), it is reasonable to consider hopping in this prediction study and to
simplify the system, p, is considered to be 1.0 in the prediction model discussed in this
chapter. The effect of a, w, or T on the device performance including device capture
efficiency (Pd) and total lateral displacement (dt) can then be predicted by Eq. 3-3 and
Eq.3-5, respectively.
Since simple miniaturization and massive parallelization
enable high-throughout
microfluidic systems and enhance separation potential, one of the aims for many
microfluidic systems is to minimize chamber length [48, 49]. Therefore, the minimum
required chamber length (xe) for targeting certain Pd and d, was identified as a function of
a, w, or r. The minimum x, was used as the indicator to determine the sensitivity and
performance of the P-selectin patterned device. In other words, device design and
operation parameters that would minimize device length and maximize separation of a
specific cell population were quantified.
3.4.1
Effect of Band Width (w) on Device Performance
The device capture efficiency (Pd) and the total lateral displacement (d) at different
effective chamber length (xe) were predicted by Eq. 3-3 and Eq.3-5, respectively, at w
varying from 10 pm to 50 tm and a = 15', g = 10 pm, and r =0.25 dyn/cm 2. The band
width around 10 pm was the limit achievable via the microcontact printing used in this
thesis. The predictions are shown in Figure 3-14. Pd increases as the adhesive band width
increases, while d, decreases as the band width increases. To target at Pd > 90% and dt >
850 pm, for example, the required
xe
that meet each criterion and both criteria are listed
84
in Table 3-2. The criteria of device capture efficiency of 90% and total lateral
displacement of 850 pm used in the prediction of optimal design were chosen arbitrarily
to illustrate the effect of substrate parameters on device performance. It is clearly shown
that the required effective chamber length that meets both criteria increases as band width
increases. This study suggests that to simply miniaturize and massively parallelize the
separation system, the band width should be minimized. However, the value of the
minimum band width (w.), which allows the stable edge tracking, is not identified in this
thesis. This value is expected to be related to the adhesive footprint (effective contact
area) of the cells. For example, the neutrophil footprint is on the order of 10 pm [50].
Thus, although there is motivation to decrease patterned band width below the limits
achievable via the microcontact printing used in this thesis, target widths will also be
limited by cell-adhesive band interactions that require further study.
(a,w, g, r)=(1s,10,10, O.s)
(15, 20, 10, O.S)
(15, 30, 10,0.5)
(15, 40, 10, O.5)
(15, 50,10,0.5)
-8,
4-OQ
+
-Q+
4000 -
100
E
80
3000
E
_
- 2000
9JA
~40/
e20
j
a
-1000
0
0
0
4 8 12 16 20 24 28 32
Effective Chamber Length, x, (cm)
Figure 3-14 The relationship between device capture efficiency (Pd) and effective
chamber length (x,) and the relationship between total lateral displacement (dt) and
effective chamber length (xe) when w varies from 10 pm to 50 pm with a = 150, g =
10 ptm, and T = 0.25 dyn/cm 2 .
85
Table 3-2 Prediction of required effective chamber lengths (x,). The range of x,
which target at Pd > 90%, d, ? 850 pm and both criteria (Pd 90% and d, ? 850 pm)
when w varies from 10 pm to 50 pm with a = 15*, g = 10 pm, and r = 0.25 dyn/cm 2 .
3.4.2
w (Pm)
x, (cm) @ Pd = 90%
x, (cm) @dt = 850 gm
Required x, (cm) that
meet both Pd (90%)
and dt (850 pm)
15
20
30
40
50
6-8
4-6
4-6
4-6
4-6
8-10
12-14
14-16
18-20
22-24
8-10
12-14
14-16
18-20
22-24
Effect of Edge Inclination Angle on Device Performance
The device capture efficiency (Pd) and the total lateral displacement (dt) at different
effective chamber lengths (xe) were predicted when a varied from 50 to 250 with fixed w
= g =10 pm, and T = 0.25 dyn/cm 2 and shown in Figure 3-15. The Pd is the same for all
edge inclination angles considered, since n and pf are the same for all cases. The
maximum d, per unit effective chamber length was identified to be between a = 150 and
200. To target at Pd > 90% and d, > 850 pm, for example, the required x, that meet each
criterion and both criteria are listed in Table 3-3. It is clearly shown that the required
effective chamber length that meet both criteria is minimized at a = 15'-20'. This
conclusion is consistent with the results from Monte Carlo simulations described in
3.3.3.4 and suggests that using a device design of a = 15'-20' can maximize separation
potentials.
86
(a, w,gr)=(5, 10, 10,0.s) -
+
(10,10, 10,0.5)
(15, 10, 10,0.s)-r
(20,10,10,0.5)-e
(25,10,10,05) -0+
4000E
100
.4
80
3000
C
E
ZI 60
2000
ILU
u 20
-
000
1000.g
> 0
0
-i
0 4 8 12 16 20 24 28 32
Effective Chamber Length, x, (cm)
Figure 3-15 The relationship between device capture efficiency (Pd) and effective
chamber length (xe) and the relationship between total lateral displacement (d,) and
effective chamber length (xe) when a varies from 5 * to 250 with w = g =10 stm, and T
= 0.25 dyn/cm 2 .
Table 3-3 Prediction of required effective chamber lengths (xe). The range of x,
which target at Pd ? 90%, d, 850 stm and both criteria (Pd ? 90% and d, > 850 sIm)
when a varies from 5 0 to 250 with w = g =10 pm, and r = 0.25 dyn/cm2 .
Required x, (cm) that
a (0)
Xe (cm) @ Pd = 90%
Xe (cm) @dt = 850
pm
meet both Pd (90%)
and dt (850 ptm)
10
15
20
25
3.4.3
10-12
8-10
8-10
10-12
4-6
4-6
4-6
4-6
10-12
8-10
8-10
10-12
Effect of Shear Stress on Device Performance
Figure 3-16 shows the relationship between device capture efficiency (Pd) and effective
chamber length (xe), and the relationship between total lateral displacement (d,) and
effective chamber length (xe), when r varies = 0.25 dyn/cm 2 to 1.0 dyn/cm 2 with fixed a
= 150 and w = g =10 gm. Both Pd and d, decrease as shear stress increases. To target Pd >
90% and dt > 850 im, for example, the required effective chamber lengths that meet each
87
criterion and both criteria are listed in Table 3-4. It is shown that the required effective
chamber length that meets both criteria increases by about 140 cm from T =0.25 dyn/cm 2
to r = 1.0 dyn/cm 2 for this particular design target.
On the other hand, in the study of cell settling in the microfluidic channel (section 2.2),
the result shows the settling distance increases as the shear stress increases. The total
chamber length (L) is considered as the effective chamber length plus the settling
distance as shown in Figure 3-12 (b). Both of the studies (in this section and section 2.2)
clearly indicate that separation chamber length is very sensitive to shear stress.
Specifically, lower shear stress results in higher capture efficiency and smaller settling
distance and therefore better separation performance for P-selectin-based substrates.
However, vacuum leaks made it difficult to examine shear stress below 0.25 dyn/cm 2 for
this particular experimental setup used in this thesis. Although the optimum shear stress
was not identified in this study, the shear stress of 0.25 dyn/cm 2 was experimentally
identified to ensure cell rolling and separation occurs in current experimental system.
On the other hand, it has been widely studied that Selectin molecules mediate shear
threshold phenomenon. This phenomenon through P-selectin or E-selectin is less evident
than through L-selectin [51-53]. Therefore, the conclusion drawn from current study
system that smaller shear stress results in better device performance may not be valid for
L-selectin-based systems. For example, Alon et al have demonstrated that leukocytes
only accumulated on the L-selectin-coated surface above a threshold shear of 0.3 dyn/cm 2
and as shear stress increased above this threshold, the number of accumulation of cells
increased and reached a maximum at an optimal shear about 2 dyn/cm 2 , and then
decreased as shear increased further [53]. It can be inferred that the capture efficiency of
a device would be affected by shear threshold phenomenon for a L-selectin-based system
and it may maximize at shear stress around 2 dyn/cm 2 . However, the settling distance
increases with increasing shear stress, which is independent of the molecules coated on
the substrates and shear stress, may affect edge tracking length significantly. Therefore,
to identify the optimum shear stress for a L-selectin-patterned microfluidic system, the
88
studies of cell capture by L-selectin-coated-substrates and edge tracking length at
different shear stress by inclined L-selectin bands need to be analyzed.
(a, w, g, r)= (15, 10,
(15, 10,
(15, 10,
(15, 10,
(a)
10, 0.25)*+'f
10, 0.50)+410, 0.75) "Q"
10, 1.0) +-
S100
80
160
40
20
U
0
0
4
8
12 16 20 24 28 32
Effective Chamber Length, x, (cm)
(b)
,~4000
E
3000
2000
o 1000
10
q
0
0
-0
4 8 12 16 20 24 28 32
Effective Chamber Length, x, (cm)
Figure 3-16 The relationship between device capture efficiency (Pd) and effective
chamber length (xe) and the relationship between total lateral displacement (d,) and
effective chamber length (xe) when T varies from 0.25 dyn/cm 2 to 1.0 dyn/cm 2 with a
= 150 and w = g =10 pim.
Table 3-4 Prediction of required effective chamber lengths (xe). The range of x,
which target at Pd > 90%, dt > 850 sm and both criteria (Pd > 90% and d, ? 850 Pim)
when T varies from 0.25 dyn/cm 2 to 1.0 dyn/cm 2 with a = 15* and w = g =10 pm.
89
r (dyn/cm 2 )
Xe
(cm) @ Pd = 90%
0.25
0.5
0.75
1
x,
(cm) @d, = 850 rn
6-8
Required x, (cm) that
meet both Pd (90%)
and dt (850 prm)
8-10
12-14
44-46
142-144
10-12
18-20
30-32
8-10
12-14
44-46
142-144
Optimization design parameters including edge inclination angle a, band width w, gap
width g, and shear stress T, for a miniaturized device of a chamber length of L, with
device capture efficiency Pd, and total lateral displacement d, were studied in this thesis
(Figure 3-17). Some future guidelines for the device optimization based on the studies of
optimization design parameters are:
1.
Decreasing shear stress increases device capture efficiency, which in turn
minimizes chamber length.
2. Increasing probability of attachment from the free stream can enhance device
capture efficiency. For example, using flow-focusing design to force cells to
interact with substrates.
3.
Device capture efficiency and total lateral displacement can be increased by
increasing the number of bands. Decreasing band width and/or gap width
increases the number of bands within a device for a given chamber length.
4. Further reductions in band width and gap width require additional studies of
minimums required for both efficient edge tracking and hopping.
5. Optimum edge inclination angle is a = 15'-20', which allows maximizing total
lateral displacement for a given chamber length or minimizing chamber length for
targeting at certain total lateral displacement.
90
Device design
w, g, r)
Pd=
device capture efficiency
hopping
d =total lateral
displacement
(r = shear stress)
L = chamber length = x,+x,
Figure 3-17 Device optimization parameters including edge inclination angle a,
band width w, gap width g, and shear stress T, for a miniaturized device of a
chamber length of L, with device capture efficiency P, and total lateral
displacement d,. x, is settling distance and xe is effective chamber length.
3.5
Conclusion
In summary, adhesive patterned substrates with multiple well-defined P-selectin patterns
have been designed by a facile, versatile method utilizing microcontact printing. These
substrates were incorporated in a flow chamber for studying HL60 cell rolling behavior
including quantification of edge tracking lengths and cell rolling velocities. Among the
parameters (edge inclination angle, shear stress, and P-selectin incubation density)
considered, the pattern edge inclination angle modulated the cell rolling trajectory most
strongly; in fact, the edge tracking length decreased exponentially with increasing edge
inclination angle. In addition, the nature of cell rolling and detachment along the edge
was consistent with a Poisson process. This correlation suggests that detachment of
rolling cells from receptor-functionalized edges is a random process that is not disrupted
measurably by cell-surface interactions over the device timescales considered.
Experimental characterization of cell rolling enabled the development of a computational
Monte Carlo tool to simulate the trajectories of cells, and showed that the optimum edge
inclination angle is 15'-20'. Experimental characterization of edge tracking length on a
single edge (in Chapter III) combined with study of capture probability from the free
stream (in Chapter II) allowed the device performance to be predicted. The prediction
results suggest that reducing shear stress, decreasing band width (and/or gap width), and
91
using an optimum edge angle of 15 -20'
can enhance overall efficiency of cell
separation for a miniaturized device.
3.6 Chapter Reference
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
G. I. Bell, Models for Specific Adhesion of Cells to Cells. Science, 1978.
200(4342): p. 618.
R. Alon, D. A. Hammer, and T. A. Springer, Lifetime of the P-SelectinCarbohydrate Bond and Its Response to Tensile Force in Hydrodynamic Flow.
Nature, 1995. 374(6522): p. 539.
D. N. Granger and P. Kubes, The Microcirculation and Inflammation Modulation of Leukocyte-Endothelial Cell-Adhesion. Journal of Leukocyte
Biology, 1994. 55(5): p. 662.
L. A. Lasky, Selectin-Carbohydrate Interactions and the Initiation of the
Inflammatory Response. Annual Review of Biochemistry, 1995. 64: p. 113.
B. T. Marshall, M. Long, J. W. Piper, T. Yago, R. P. McEver, and C. Zhu, Direct
Observation of Catch Bonds Involving Cell-Adhesion Molecules. Nature, 2003.
423(6936): p. 190.
T. F. Tedder, D. A. Steeber, A. Chen, and P. Engel, The Selectins - Vascular
Adhesion Molecules. Faseb Journal, 1995. 9(10): p. 866.
G. A. Zimmerman, Two by Two: The Pairings of P-Selectin and P-Selectin
Glycoprotein Ligand 1. Proc Natl Acad Sci U S A, 2001. 98(18): p. 10023.
R. P. McEver and C. Zhu, Rolling Cell Adhesion. Annu Rev Cell Dev Biol, 2010.
26: p. 363.
R. Karnik, S. Hong, H. Zhang, Y. Mei, D. G. Anderson, J. M. Karp, and R.
Langer, Nanomechanical Control of Cell Rolling in Two Dimensions through
Surface PatterningofReceptors. Nano Lett, 2008. 8(4): p. 1153.
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Examining the
Lateral Displacement of H160 Cells Rolling on Asymmetric P-Selectin Patterns.
Langmuir, 2011. 27(1): p. 240.
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Studying Cell
Rolling Trajectorieson Asymmetric Receptor Patterns.J Vis Exp, 2011(48).
S. Bose, R. Singh, C. Shen, M. Hanewich-Hollatz, C. H. Lee, D. M. Dorfman, J.
M. Karp, and R. Karnik, Affinity Flow Fractionation of Cells Via Transient
Interactionswith Asymmetric Molecular Patterns. Scientific Reports, 2013(3): p.
2329.
G. C. Ibbotson, C. Doig, J. Kaur, V. Gill, L. Ostrovsky, T. Fairhead, and P.
Kubes, Functional Alpha4-Integrin: A Newly Identified Pathway of Neutrophil
Recruitment in CriticallyIll Septic Patients.Nat Med, 2001. 7(4): p. 465.
K. L. Davenpeck, M. E. Brummet, S. A. Hudson, R. J. Mayer, and B. S. Bochner,
Activation of Human Leukocytes Reduces Surface P-Selectin Glycoprotein
Ligand-1 (Psgl-1, Cd162) and Adhesion to P-Selectin in Vitro. Journal of
Immunology, 2000. 165(5): p. 2764.
92
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
D. E. Lorant, R. P. Mcever, T. M. Mcintyre, K. L. Moore, S. M. Prescott, and G.
A. Zimmerman, Activation of Polymorphonuclear Leukocytes Reduces Their
Adhesion to P-Selectin and Causes Redistribution of Ligands for P-Selectin on
Their Surfaces. Journal of Clinical Investigation, 1995. 96(1): p. 171.
P. H. Reinhardt and P. Kubes, Differential Leukocyte Recruitment from Whole
Blood Via EndothelialAdhesion Molecules under Shear Conditions. Blood, 1998.
92(12): p. 4691.
N. Charles, J. L. Liesveld, and M. R. King, Investigating the Feasibilityof Stem
Cell Enrichment Mediated by Immobilized Selectins. Biotechnology Progress,
2007. 23(6): p. 1463.
S. D. Narasipura, J. C. Wojciechowski, N. Charles, J. L. Liesveld, and M. R.
King, P-Selectin-Coated Microtube for Enrichment of Cd34(+) Hematopoietic
Stem and Progenitor Cells from Human Bone Marrow Hematology. Clinical
Chemistry, 2008. 54(1): p. 77.
K. E. Norman, K. L. Moore, R. P. Mcever, and K. Ley, Leukocyte Rolling in-Vivo
Is Mediated by P-Selectin Glycoprotein Ligand-1. Blood, 1995. 86(12): p. 4417.
M. B. Lawrence, G. S. Kansas, E. J. Kunkel, and K. Ley, Threshold Levels of
Fluid Shear Promote Leukocyte Adhesion through Selectins (Cd62l,PE). Journal
of Cell Biology, 1997. 136(3): p. 717.
C. Dong and X. X. Lei, Biomechanics of Cell Rolling: Shear Flow, Cell-Surface
Adhesion, and Cell Deformability.Journal of Biomechanics, 2000. 33(1): p. 35.
L. Wu, B. T. Xiao, X. L. Jia, Y. Zhang, S. Q. Lu, J. Chen, and M. Long, Impact of
CarrierStiffness and Microtopology on Two-Dimensional Kinetics of P-Selectin
and P-Selectin Glycoprotein Ligand-1 (Psgl-1) Interactions.Journal of Biological
Chemistry, 2007. 282(13): p. 9846.
W. C. Chang, L. P. Lee, and D. Liepmann, Biomimetic Technique for AdhesionBased Collection and Separation of Cells in a Microfluidic Channel. Lab on a
Chip, 2005. 5(1): p. 64.
C. D. James, R. C. Davis, L. Kam, H. G. Craighead, M. Isaacson, J. N. Turner,
and W. Shain, Patterned Protein Layers on Solid Substrates by Thin Stamp
MicrocontactPrinting.Langmuir, 1998. 14(4): p. 741.
A. Bernard, E. Delamarche, H. Schmid, B. Michel, H. R. Bosshard, and H.
Biebuyck, PrintingPatternsofProteins.Langmuir, 1998. 14(9): p. 2225.
J. L. Tan, J. Tien, and C. S. Chen, MicrocontactPrintingof Proteins on Mixed
Self-Assembled Monolayers. Langmuir, 2002. 18(2): p. 519.
D. J. Graber, T. J. Zieziulewicz, D. A. Lawrence, W. Shain, and J. N. Turner,
Antigen Binding Specificity of Antibodies Patterned by Microcontact Printing.
Langmuir, 2003. 19(13): p. 5431.
J. 0. Foley, E. Fu, L. J. Gamble, and P. Yager, MicrocontactPrintedAntibodies
on Gold Surfaces: Function, Uniformity, and Silicone Contamination. Langmuir,
2008. 24(7): p. 3628.
M. Ghosh, C. Alves, Z. Tong, K. Tettey, K. Konstantopoulos, and K. J. Stebe,
Multifunctional Surfaces with Discrete Functionalized Regions for Biological
Applications. Langmuir, 2008. 24(15): p. 8134.
93
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
D. Lee and M. R. King, MicrocontactPrintingof P-Selectin Increases the Rate of
Neutrophil Recruitment under Shear Flow. Biotechnology Progress, 2008. 24(5):
p. 1052.
A. Folch and M. Toner, Microengineering of Cellular Interactions. Annual
Review of Biomedical Engineering, 2000. 2: p. 227.
K. Glasmastar, J. Gold, A. S. Andersson, D. S. Sutherland, and B. Kasemo,
Silicone Transfer During Microcontact Printing. Langmuir, 2003. 19(13): p.
5475.
I. Banerjee, R. C. Pangule, and R. S. Kane, Antifouling Coatings: Recent
Developments in the Design of Surfaces That Prevent Fouling by Proteins,
Bacteria, and Marine Organisms. Adv Mater, 2011. 23(6): p. 690.
M. Salta, J. A. Wharton, P. Stoodley, S. P. Dennington, L. R. Goodes, S.
Werwinski, U. Mart, R. J. Wood, and K. R. Stokes, Designing Biomimetic
Antifouling Surfaces. Philos Trans A Math Phys Eng Sci, 2010. 368(1929): p.
4729.
W. R. Gombotz, G. H. Wang, T. A. Horbett, and A. S. Hoffman, Protein
Adsorption to Poly(Ethylene Oxide) Surfaces. J Biomed Mater Res, 1991. 25(12):
p. 1547.
K. L. Prime and G. M. Whitesides, Adsorption of Proteins onto Surfaces
Containing End-Attached Oligo(Ethylene Oxide): A Model System Using SelfAssembled Monolayers. J. Am. Chem. Soc., 1993. 115(23): p. 10714.
K. L. Prime and G. M. Whitesides, Self-Assembled Organic Monolayers: Model
Systems for Studying Adsorption of Proteins at Surfaces. Science, 1991.
252(5009): p. 1164.
http://physics.georgetown.edu/matlab/.
R. Karnik, S. Hong, H. Zhang, Y. Mei, D. G. Anderson, J. M. Karp, and R.
Langer, Nanomechanical Control of Cell Rolling in Two Dimensions through
Surface PatterningofReceptors. Nano Letters, 2008. 8(4): p. 1153.
T. Yago, A. Leppanen, H. Y. Qiu, W. D. Marcus, M. U. Nollert, C. Zhu, R. D.
Cummings, and R. P. McEver, Distinct Molecular and Cellular Contributions to
Stabilizing Selectin-MediatedRolling under Flow. Journal of Cell Biology, 2002.
158(4): p. 787.
S. Q. Chen, R. Alon, R. C. Fuhlbrigge, and T. A. Springer, Rolling and Transient
Tethering of Leukocytes on Antibodies Reveal Specializations of Selectins.
Proceedings of the National Academy of Sciences of the United States of
America, 1997. 94(7): p. 3172.
M. B. Lawrence and T. A. Springer, Neutrophils Roll on E-Selectin. J Immunol,
1993. 151(11): p. 6338.
M. J. Smith, E. L. Berg, and M. B. Lawrence, A Direct Comparison of SelectinMediated Transient, Adhesive Events Using High Temporal Resolution.
Biophysical Journal, 1999. 77(6): p. 3371.
K. C. Chang and D. A. Hammer, Adhesive Dynamics Simulations of SialylLewis(X)/E-Selectin-Mediated Rolling in a Cell-FreeSystem. Biophysical Journal,
2000. 79(4): p. 1891.
94
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
S. Jadhav, C. D. Eggleton, and K. Konstantopoulos, A 3-D ComputationalModel
Predicts That Cell Deformation Affects Selectin-Mediated Leukocyte Rolling.
Biophysical Journal, 2005. 88(1): p. 96.
C. B. Korn and U. S. Schwarz, Dynamic States of Cells Adhering in Shear Flow:
From Slipping to Rolling. Physical Review E, 2008. 77(4).
J. Y. Shao and R. M. Hochmuth, Micropipette Suctionfor MeasuringPiconewton
Forces of Adhesion and Tether Formation from Neutrophil Membranes.
Biophysical Journal, 1996. 71(5): p. 2892.
S. Vyawahare, A. D. Griffiths, and C. A. Merten, Miniaturization and
Parallelizationof Biological and ChemicalAssays in Microfluidic Devices. Chem
Biol, 2010. 17(10): p. 1052.
S. Baratchi, K. Khoshmanesh, C. Sacristan, D. Depoil, D. Wlodkowic, P.
McIntyre, and A. Mitchell, Immunology on Chip: Promises and Opportunities.
Biotechnol Adv, 2013.
P. Sundd, E. Gutierrez, M. K. Pospieszalska, H. Zhang, A. Groisman, and K. Ley,
Quantitative Dynamic Footprinting Microscopy Reveals Mechanisms of
Neutrophil Rolling. Nat Meth, 2010. 7(10): p. 821.
E. B. Finger, K. D. Puri, R. Alon, M. B. Lawrence, U. H. von Andrian, and T. A.
Springer, Adhesion through L-Selectin Requires a Threshold Hydrodynamic
Shear. Nature, 1996. 379(6562): p. 266.
T. Yago, V. I. Zarnitsyna, A. G. Klopocki, R. P. McEver, and C. Zhu, Transport
Governs Flow-Enhanced Cell Tethering through L-Selectin at Threshold Shear.
Biophys J, 2007. 92(1): p. 330.
R. Alon, S. Chen, R. Fuhlbrigge, K. D. Puri, and T. A. Springer, The Kinetics and
Shear Threshold of Transient and Rolling Interactions of L-Selectin with Its
Ligand on Leukocytes. Proc Natl Acad Sci U S A, 1998. 95(20): p. 11631.
95
This page is intentionally left blank
96
4 Identification of Adhesion Molecules with Weak Affinity via
Phage Display from M13 pVIII Library
4.1
Introduction
Chapter IV aims to discuss and demonstrate how to identify specific receptors with low
affinity to the target molecules to broaden the application range for cell separation
devices based on transient cell-receptor interaction.
Selectin is one of the families of cell adhesion molecules (CAMs). Three known
members (P-, E- and L-selectin) of the selectin family have similar calcium-dependent
lectin-like domain, which recognizes certain carbohydrate groups [1-4]. For example, Pselectin expressed on endothelial cells binds to P-selectin Glycoprotein Ligand- 1 (PSGL1) highly expressed on the neutrophils during the leukocytes adhesion cascade [5, 6]. The
nature of P-selectin's low affinity to PSGL-1 allows the leukocytes to be captured and
roll on the blood vessel wall and enables the feasibility of label-free netrophil separation
via cell rolling on P-selectin patterns in microfluidic devices [7-10]. However, there
exists a need for applying this technology to different receptors other than selectin family.
The antibody is commonly used for identifying an unique target, called antigen, and it has
been reported to show high specificity and high affinity (nanomolar ~subnanomolar
range) [11-13]. Antibody-based microfluidic devices were reported to show high-purity
capture of cells [14, 15]. However, high affinity resulting in stable and strong binding of
the cells makes it difficult retrieving captured cells within the device which hinders some
application where the viable cells are required, and prevents easy enumeration of cells.
Biopanning is an affinity selection technique used for identifying peptides that bind to a
target. Selected peptides have been reported to show affinity to the target within
micromolar range [16], which opens the possibility of using selected peptides as weak
affinity receptors for cell separation based on cell rolling. One of the most commonly
used libraries is the M13 phage display library. M13 is a well-known filamentous
97
bacteriophage or virus and it is harmless to mammalian cells [ 17]. The length of the M 13
phage is around 880 nm and the diameter is around 6-7 nm. M13 has five coat proteins
(pIII, pVI, pVII, pVIII, and pIX). Figure 4-1 shows the schematic diagram of M13
bacteriophage and the summary of the residues and number of copies of each type of coat
protein. Among the five coat proteins, pVIII is the major one, which has 2,700 copies per
phage. The whole genome of M13 has been revealed and studied for years. Its coat
protein groups can be independently altered by engineering the viral genome. Because of
this advantage, the M13 phage can be easily engineered to serve as a powerful biological
toolkit. M13 phage display library is one of the examples [18-23]. A phage display
library can be constructed by displaying random peptides on the certain position of the
proteins of the phage. For example, pVIII library is the phage pool where the random
peptides were inserted on pVIII coat protein. The phage library is then exposed to the
target to find out which peptide sequence has affinity to the target.
pVll, piX
P1
pVi,. pill
--
6-7 nm
880 nm
Protein
Residues
# of Copies
pVll
33
5
piX
32
-5
pVil
73
2,700
pill
427
3-5
pVI
112
~5
Figure 4-1 Schematic diagram of M13 bacteriophage and the summary of coat
protein. (adapted from Dr. Chung-Yi Chiang from Prof. Angela Belcher's lab)
In Chapter IV, biopanning using pVIII library was studied to investigate whether it can
be used to identify molecules with low affinity with CD4 chosen as the target. CD4 is a
type of protein molecules expressed on the surface of the mature Th cells (helper T cells),
named CD4+ T cells [24]. It is a receptor for Glycoprotein gpl20 of human
immunodeficiency virus (HIV) [25]. Binding of gp120 to CD4 is the first step of viral
98
entry, which leads to the fusion of viral and cell membranes. As the virus infects the cells,
the number of CD4+ T cells decreases resulting in weakening the immune system. CD4+
T cells count is also an indicator of the success or the failure of anti-retroviral therapy
(ARV) [26]. In addition to application of diagnosis and immune monitoring, sorting and
analysis of pure, viable CD4+ T cells can be applied in studies on proliferation, apoptosis
and regulation of CD4+ T cells [27-29].
In this Chapter, pVIII library was applied to select phage with affinity to CD4 protein.
The selected phage was chemically immobilized on a gold substrate and the capture
efficiencies of the CD4 expressing cells (SUPTI) and non-CD4 expressing cells (K562)
were quantified. The phage-engineered substrate was blocked with CD4 to demonstrate
the specificity of the selected phage to CD4.
4.2
4.2.1
Biopanning of pVIII Library
Phage display library of random peptides
The phage display library is a pool of bacteriophage of millions of random peptide
sequences with different functionalities [18-21, 23]. Synthetic oligonucleotides of fixed
length and random condons are fused to genes VIII of M13 vectors. The fusions are
expressed as random peptide on the capsid or shell proteins of M 13 to form phage display
library. The libraries are then used to screen peptides for targeting the materials of
interest. To select peptides for targeting the materials of interest, the phage library is
exposed to the target. Phage with affinity to the target can bind to the target. Unbound
phage is then washed off from the target and the bound phage is then eluted from the
target surface by changing the pH value or temperature. The sequences of the peptide
motifs on the eluted phage are then characterized. The eluted phage is amplified to serve
as the sub-library for the next round of the panning process. The panning process is
repeated until the consensus peptides are found.
99
4.2.2
Biopanning Procedure
Biopanning with the M13 pVIII library was conducted in collaboration with Dr. Nimrod
Heldman and Dr. Rana Gosh in Prof. Angela M. Belcher's lab at MIT (DMSE and BE).
Materials. Recombinant human CD4 receptor protein was purchased from Protein
Science Corp and diluted into 100 pg/mL in phosphate buffered saline (PBS, Mediatech
Inc). The pVIII library used in this chapter was constructed by Dr. Paul Widboom
(Belcher's group at MIT) and the diversity of this particular library is 106. Bovine serum
albumin (BSA) was purchased from Rockland Immunochemicals, Inc. Tris-HC buffer
(Mallinckrodt Chemicals) was titrated to 1 M, pH=9.1. Glycine-HCl (Sigma-Aldrich)
was titrated to 0.2 M, pH=2.2. 96-well standard microplate were purchased from
Falcon@.
Methods. A schematic diagram of identifying peptide motif with affinity to CD4 by
biopanning method is shown in Figure 4-2. 400 pL of the diluted CD4 protein solution
was added into 4 wells (100 pL/well; 10 pg CD4/well) of 96-well standard microplate.
The plate was incubated overnight at 4'C in the refrigerator. Next day, the plate was
taken out to room temperature. The CD4 solution was poured off. The wells were cleaned
by filling each well with PBS and then slapped face down onto the clean paper towel ten
times. Blocking solution was prepared by dissolving 3wt% BSA in PBS. 1200 ptL of the
diluted BSA solution was added into the CD4-coated wells (300 pL/well) to block nonspecific interaction and kept at room temperature for 3 hours. The BSA solution was
poured off and the wells were cleaned again by filling each well with PBS and then
slapped face down onto the clean paper towel 10 times.
400 pL phage solution containing 400 copies of each sequence was prepared and added
to four wells (100 pL/well; 100 copies of each sequence /well). The microplate was then
kept overnight in the 4'C refrigerator. Next day, the microplate was taken out to room
temperature. The unbound phage solution was poured off. The wells were cleaned by
filling each well with PBS and then slapping it face down onto the clean paper towel ten
times before bound phage was eluted. After the washing steps, 100 pL of Glycine-HCl
100
(pH=2.2; 0.2 M) containing 1 mg/mL BSA were added to each well and incubated for 10
minutes to elute the bound phage. The elution process was then repeated once. All the
eluted phage in Glycine-HCl (eluate) were added to 150 ptL Tris-HCl (1 M, pH=9.1). The
eluate was titered and the sequences of the peptide motif on the phage was sequenced.
The phage solution was then amplified and purified to serve as the sub-library. The sublibrary was titered again and the sequences of the peptide motif on the sub-library were
sequenced. The sub-library was then used for the next round of the panning process. The
panning process was repeated three times. For the second round and the third round of
biopanning, the initial input concentration of phage is 1011 per well (4 wells) and
sequential steps were the same as described in the first round of biopanning.
pVIII Phage Library
_Livrsity-16)
Amp
&P
Ug
tion
cation
P P miUnbound
Phages
DNA sequencing
Figure 4-2 Schematic diagram for identifying peptide motifs of affinity to CD4
protein by biopanning using pVIII library. The pVIII library was exposed to the CD4coated well plate. Unbound phage was then washed off from the target and the bound
phage was eluted from the surface by acid solution. The eluted phage was amplified to
serve as the sub-library for the next round of the panning process. The peptide sequences
of the eluted and amplified phage was identified by titering and DNA sequencing. The
panning process was repeated until the consensus peptides were found.
101
4.2.3
4.2.3.1
Phage Titering
Lysogeny Broth (LB) Medium Preparation
1 L Lysogeny broth (LB) medium were prepared by mixing 25 g Difco LB Broth (BD
Biosciences) and 1 L water. The medium was then autoclaved and stored at room
temperature for further use.
4.2.3.2 Tetracyline (Tet) Stock Solution Preparation
Tetracycline (OmniPur) was dissolved in Methanol (Pharmco-AAPER) to a final
concentration of 20 mg/mL. The Tet stock solution was then stored in dark at -20'C for
further use.
4.2.3.3
IPTG/Xgal Stock Solution Preparation
1.25g isopropyl P-D-thiogalactoside (IPTG, Sigma-Aldrich) and ig 5-Bromo-4-chloro-3indolyl- P-D-galactoside
(Xgal,
Sigma-Aldrich)
were
mixed
in
25
mL
Dimethylformamide (Sigma-Aldrich) and stored at 4C in the dark for further use.
4.2.3.4
LB/IPTG/Xgal Plates Preparation
15 g agar (BD Biosciences) and 1 L LB medium were mixed, autoclaved, and then cooled
to <70'C. 1 mL Tet stock solution and 1 mL pre-made IPXG/Xgal were add to agar/LB
solution. The solution was then poured on Petri dishes (BD Falcon) (1 5mL per dish). The
dishes were then served as LB/IPTG/Xgal plates after they were solidified at room
temperature. The LB/IPTG/Xgal plates were sealed and stored at 4C in the dark for
further use.
4.2.3.5
Agarose Top
5 g Bacto-Tryptone (BD Biosciences), 2.5 g yeast extract (BD Biosciences), 2.5 g NaCl
(Sigma-Aldrich), 0.5 g MgCl 2 6H 2 0 (Alfa Aesar), 3.5 g Agarose (OmniPur) were
102
dissolved and mixed in 500 mL water. The solution was then autoclaved and stored at
room temperature for further use.
4.2.3.6
Titration Procedure
The phage was titered on LB/IPTF/Xgal plates. 5 mL LB medium and a single colony of
ER2738 (New England Biolabs) or XLlBlue (Agilent Technologies) were incubated
overnight at 37*C with shaking serving as overnight cell culture. 10-fold serial dilutions
of the phage in LB medium were prepared. For unamplified panning eluates, the dilution
range is 101-104. For amplified phage supernatants, the dilution range is 108-1014.
Agarose Top were pre-melted at 50'C. LB/IPTG/Xgal plates were pre-warmed at 37.5'C.
10 gL of each dilution were added to 200 pL of overnight cell culture to infect cells and
incubated at room temperature for 5 minutes. The infected cells were then added to 3 mL
pre-warmed Agarose Top and all the solution were poured onto the LB XGal/IPTG
plates. After the plates were cooled and the Agarose Top was solidified at room
temperature, the plates were then incubated overnight at 37 'C. Next day, the blue
plaques on the plates (as shown in Figure 4-3) were randomly picked for sequencing.
103
Figure 4-3 Titration of phage on LB/IPTG/Xgal plates (dilution range is from 10' to
104). The phage was titered on LB/IPTF/Xgal plates and the blue plaque was then
randomly selected for sequencing.
4.2.4
4.2.4.1
Amplification of M13 bactriophage
PEG/NaCl Stock Solution Preparation
200 g PEG8000 (Amreco) and 146 g NaCl (Sigma-Aldrich) were dissolved and mixed in
1L water. The stock solution was then autoclaved and stored at room temperature for
further use.
4.2.4.2
Amplification between Rounds of Panning
400 pL overnight cell culture, 40 pL Tet stock solution, and 400 pL the phage eluates
were added to 40 mL autoclaved LB media. The culture solution was then incubated at
37'C for 4.5 to 5 hours with shaking (MaxQ 4000 shaker, Thermo Scientific). The
culture solution was transferred to a centrifuge tube and spun for 20 minutes at 3500 rpm
(Beckman Coulter JA25.5) at 4 'C. The supernatant was then transferred to a fresh tube.
1/6 volume of PEG/NaCl stock solution were added to the tube with supernatant and kept
at 4'C overnight to allow the phage to precipitate. Next day, PEG precipitate was spun
down for 20 minutes at 10,000 rpm at 4'C. The supernatant was then discarded. The
pellet was suspended in 1 mL PBS and then transferred to 1.5 mL microcentrifuge tube.
The solution was then spun for 5 minutes at 3500 rpm in a table-top microcentrifuge
(Beckman Coulter) to pellet residual cells. The supernatant was then transferred to fresh
microcentrifuge tube and re-precipitated with 1/6 volume of PEG/NaCl at ice for 2 hours.
PEG precipitate was then spun down for 20 minutes at 10,000 rpm. The supernatant was
then discarded. The pellet was suspended in 100 piL PBS to serve as sub-library. The
phage concentration was characterized by spectrophotometer (Nanodrop@ ND 1000). The
relationship between phage concentration and absorption is given by the following
equation derived by G. Smith [30]:
104
phages/ mLIL(A
=
number
where
(A2 , - A 320 )x 6.-10"
(26 2 ,6 -A 32 )x 6.-10"
30)x10JA29-A30x-017-
of
nucleotides / phage
7200
bases! phage
(Eq. 4-1)
A2 69 : the absorbance at 269 nm which reflects the DNA content in the solution
A 32 0 : the absorbance at 320 nm which corrects for the naturally high baseline
value of the solution
The amplified phage sub-library was then titered for sequencing as described above.
4.2.5
DNA extraction for sequence identification
After each round of biopanning and amplification, DNA samples of phage was prepared
for sequencing and the DNA sequences were then translated into peptide sequences to see
if there are multiple occurrences of the same sequence, which indicates dominant binders.
Plaque Amplification. A blue plaque on LB/IPTG/Xgal plates were picked randomly by
sterile pipette tip and were then transferred to a tube containing diluted culture (mixture
of 30 piL overnight cell culture, 3 piL Tet stock solution, and 3.5mL LB). The culture
solution was then incubated in the 37'C for 4.5 to 5 hours with shaking. Multiple plaques
were picked.
DNA Extraction. The schematic diagram is shown in Figure 4-4. All the buffers and
columns used for DNA extraction were purchased from Qiagen (QlAprep Spin Miniprep
Kit) without further purification. The culture amplified solution was transferred to a
centrifuge tube and spun for 20 minutes at 3500 rpm. The supernatant was discarded. The
pelleted bacterial cells were resuspended in 250 piL Buffer P1 and then transferred to a
microcentrifuge tube. 250 ptL Buffer P2 was added to the solution and the tube was
gently inverted 5 times to mix. 350 tL Buffer N3 was then added to the solution and the
tube was inverted gently 5 times. The solution was then spun 10 minutes at 13000 rpm in
a table-top microcentrifuge (Beckman Coulter). The supernatants were then applied to the
QUAprep spin column. The solution was then spun 1 minute at 13000 rpm and the flow-
105
through was discarded. 0.75 mL Buffer PE was added to QlAprep spin column and the
solution was spun 1 minute at 13000rpm. The flow-through was discarded and the
column was placed in the microcentrifuge and spun down for 1 minute to remove residual
Buffer PE. QlAprep column was then placed in a clean 1.5 mL microcentrifuge tube and
50 pL Buffer EB (10 mM Tris-Cl, pH8.5) was added to the column to elute DNA. The
solution was then spun 1 minute at 13000 rpm. The sample for sequencing was prepared
by adding 8.8 pL eluted DNA to 3.2 pL 96 gIll sequencing primer (5'-.
HOCCC
TCA
TAG TTA GCG TAA CG -3', 1 pmol/pl; New England BioLabs). The samples were
then analyzed by polymerase chain reaction (PCR, LightCycler 480 Real-time PCR
system,
Roche
Applied
Science).
The
DNA
sequences
obtained
from PCR
characterization of the phage was then translated into peptide sequences.
Suspend
bacteria in
P buffer
Spin down
bacteria
Lyse
bacteria
using P2
Neutralize
the sarnple
using N3
buffer
buffer
Spin down,
apply
MLSFAAAYDTNMDSD
PAKAAFNSLQASATEY
supernatants
Into QIA-
IGYAWAMVVVIVGAT
IGIKLFKKFTSKAS
coun
Translation
into amino
acids
PCR
sequencing
Elute DNA
by adding
EB buffer
Discard flow-though,
wash with PE buffer,
centrifuge, and discard
flow-through
Figure 4-4 Schematic diagram of DNA extraction for sequence identification. DNAs
of the phage was extracted from the bacteria. The bacteria were lysed and the DNAs were
isolated from the lysed bacteria and then eluted using QlAprep Spin Miniprep Kit.
106
4.2.6
Biopanning Results
The number of phage per well of input and output solutions and the recovery for each
round of screening are summarized in Table 4-1. The large recovery from the first round
may come form the nonspecific binding among phage in the library. The enrichment was
observed between the second and the third round. The peptide sequences of the phage
screened from the the first round of panning, the first round of amplification, the second
round of panning, the second round of amplification, and the the third round of panning
are summarized in Table 4-2. Although some clones appears multiple times, they may be
fast growers. For example, VNTTVSGE didn't appear in the first and the second round of
biopanning but appear twice (out of ten samples) after the second amplification. To
explore wider range of possibility of identifying specific peptide motifs, twelve clones
were selected and were named as RK1, RK2, RK3.... and RK12. . Peptide properties of
RK1-RK12 are shown in Table 4-3 (negatively charged amino acids in red; positively
charged amino acids in blue; polar amino acid without charged in yellow; hydrophobic
amino acid in gray; aromatic amino acids in green; Tyrosine in orange). RK3 and RK12
show similar charge distribution. RK4, RK7, RK8, and RK9 also have similar charge
distribution. Each clone was then amplified as described above for cell capture
experiment.
Table 4-1 Input and output concentration of phage and the recovery of each round
of screening. The concentration is defined as the number of the phage per well. For each
round of screening, four wells coated with CD4 were used.
Round of screening
Input phage
(#/well)
Output phage
(#/well)
Recovery
(Input/Output)
108
1.66x10-s
7.12x10-8
9.5x10-6
1
2
1011
1.66x1O3
7.12x10 3
3
1011
9.5x105
Table 4-2 CD4 protein binding sequences from biopanning of pVIII library.
107
Occurrence
Sequence
Round1
panning
A
A
A
D
D
D
D
D
D
D
G
G
G
Q
S
Y
G
N
N
P
V
W
Y
D
G
L
D
Y
D
K
G
L
N
Y
N
D
N
D
G
T
G
T
Y
V
Q
I
N
G
L
R
S
D
S
S
N
L
S
Y
Y
S
P
M
L
P
D
S
D
M
N
S
G
N
I
G
N
G
G
T
L
V
D
G
I
D
D
D
T
N
T
M
M
P
S
S
E
P
S
S
T
E
A
E
P
S
Round1
amplification
A
D
D
D
G
V
V
Q
S
W
W
G
N
S
D
Y
N
S
N
S
T
T
S
G
D
F
Q
D
S
T
P
A
T
G
L
S
S
G
G
A
I
S
L
M
T
G
D
D
R
P
P
E
E
T
T
E
A
A
D
D
E
E
V
V
V
S
Y
S
P
S
N
N
S
Y
D
Y
Y
F
R
P
S
S
T
T
T
S
Y
S
G
Q
N
T
M
N
T
A
S
S
G
G
G
N
D
S
D
D
G
I
S
D
S
S
M
G
V
T
D
E
E
E
M
P
A
P
E
T
T
D
G
V
V
V
T
G
N
N
S
Y
D
S
T
S
S
S
Q
T
N
T
P
G
V
G
S
G
I
S
S
L
M
D
G
E
P
P
T
E
T
V
S
S
S
M
G
D
T
V
S
S
S
M
G
D
T
A
A
P
Y
D
D
N
T
D
M
S
I
N
S
P
E
Round 2
panning
Round 2
amplification
Round 3
panning
Q
108
X2
X2
X2
X2
X3
X2
Table 4-3 Selected peptide sequences and their properties. (negatively charged amino
acids in red; positively charged amino acids in blue; polar amino acid without charged in
yellow; hydrophobic amino acid in gray; aromatic amino acids in green; Tyrosine in
orange).
RK
2
RK
3
RK
6
S
RK7S
RK8
RK 8
RK
RK
RK
10
11
12
S
S
S
P
NT
$
T
T
S
P
4.3 Immobilization of phage on Au-coated glass slide
Gold-coated glass slides were purchased from EMF Corp. All slides were cleaned with
piranha solution prior to use (3:1 mixture of sulfuric acid (Sigma-Aldrich) and 30%
hydrogen peroxide (Sigma-Aldrich)). A schematic diagram of the immobilization of
phage on Au-coated glass slides is shown in Figure 4-5. Step 1: Dithiobis (succinimidyl
propionate) (DSP, Sigma-Aldrich) was diluted in absolute ethanol (Pharmco-AAPER) to
a concentration of 5 mM. The cleaned Au-coated slides were inbucated in 5 mM DPS for
3 hours at room temperature with shaking. The surface was then rinsed with ethanol and
dried under a stream of N2. Step 2: Each clone was diluted in PBS to a concentration of
1013 /mL. The substrates were then incubated in phage solution (101 3 /mL in PBS, unless
stated otherwise) using a perfusion chamber (Electron Microscopy Sciences) at 4'C
overnight. The surfaces were then backfilled with BSA (3 wt% in PBS) for 3 h to block
nonspecific interactions.
109
DSP
d94
'T'-
Au-coated glass
slide
Step 1
Step 2
Figure 4-5 Schematic diagram for immobilizing phage on a gold substrate. Step 1:
Functionalization of DSP onto the gold surface. Step 2: Immobilization of phage on the
reactive gold surface.
4.4 Substrate Characterization
Atomic force microscopy (Veeco Dimension 3100, tapping mode, 1 Hz scan rate was
used to characterize the surface. All substrates for AFM characterization were placed in a
vacuum chamber overnight before imaging to minimize residual solvent on the surface.
AFM height images are shown in Figure 4-6. Although all the samples were prepared at
the incubation concentration of 1013 phage/mL, the surface density of the phage varies
among the samples. For example, RK4, 6, 8 and 9 show bundle-like structures. It is
possible that those clones have more hydrophobic motifs and tend to aggregate because
of hydrophobic interaction. However, the surface density of RK6 is not as high as them
although RK6 has similar charge distribution as RK4, RK8, and RK9. On the other hand,
RK1 1 and RK12 show very low density on the gold substrates and those clones have
more negatively charged motifs. A single phage may repel one another because of
negative charges and may result in low surface density of the phage on the gold
substrates.
110
III
wuo
WU
S"7T
wu 0
WU SIT
wu 0
WU
SIT
wu 0
wu S
wu 0
wu 0
wu ST
wu S*L-
wu 0
wu S-LwuS
wu 0
WU ST
15 nm
(j) RK10
0 nm
15 nm
0 nm
20 nm
)
0 nm
17.5 nm
Onm
Figure 4-6 Surface characterization of phage-engineered substrates. AFM height
images of phage immobilized on the gold-coated substrate. All the images are 1 Om X
10 m.
4.5
4.5.1
Cell Capture Experiments
Cell Culture
SUPTI cells and K562 cells (both purchased from ATCC) were maintained in Iscove's
modified Dulbecco's medium supplemented with 20% fetal bovine serum according to
the ATCC recommendation. Cells were cultured in 75 cm2 polystyrene tissue culture
flasks (BD Falcon@) with the cell density maintained between 105- 106 cells/mL.
4.5.2
Cell Staining
All the cells were stained in sterile condition (inside the Bio-Hood) before the
experiments. K562 cells were transferred from the flash into a 15mL conical tube and
were spun at 1700 rpm for 5 min. The supernatant was discarded and the pellet was
112
suspended in PBS at a final concentration of 106 cell/mL. The Cell trackerM Green
(CMFDA, 5-Chloromethylfluorescein Diacetate; Life Technologies) was then added to
the solution to the final concentration of 1pM and incubated inside the water bath of
37.5'C for 20 min. The cells were spun down at 1800 rpm for 5 min and the pellet was
then suspended in appropriate cell media to a final concentration of 106 cells/mL. The
solution was then transferred to the wells of 6-well plate (Falcon@) (2 mL per well) and
incubated inside the incubator for 1 h at 37.5'C with 5% CO 2. The solution was
transferred to 15mL centrifuge tube and spun down at 1800 rpm for 5 min. The
supernatant was discarded and the pellet was suspended in the media to the concentration
of 106 cells/mL. The solution was then transferred into the well plate and kept in the
incubator at 37.5'C with 5% CO 2 for further use.
4.5.3
Cell Capture Experiment in Flow Chamber
An inverted fluorescent microscope (Nikon TE2000-U) with a mounted camera (Andor
iXon 885) was used to record the interaction of cells and the phage-immobilized
substrates using a 4x objective, typically at a rate of 1 frame/s with exposure time of
0.1-1 s. A suspension of cells was allowed to flow over the phage-coated surfaces in a
rectangular flow chamber (Glycotech, Inc.; width w=1.0 cm, length l=6 cm, and height
h=0.005 in.) at room temperature of 24.5'C. A syringe pump (World Precision
Instruments (WPI), SP230IW) was used to generate different flow rates of between 15
and 75 iL/min, with corresponding shear stresses of 0.1-0.5 dyn/cm 2 (~ 0.01-0.05 Pa).
The flow was laminar (Reynold's number Re ~ 0.1-3), and the shear stress T was
calculated using the plane Poiseuille flow equation r=6puQ/wh 2 , where p is the kinematic
viscosity,
Q is the volumetric
flow rate, w is the width of the flow chamber, and h is the
height of the flow chamber. The cells were first flowed slowly (flow rate ~ 15 pL/min)
into the flow chamber until they were shown on the window of the view. The flow was
then stopped for 5 min to allow the cells to settle down and interact with the substrate.
The flow was started again and parts of the cells were detached from the surface.
113
4.6 Cell Capture Efficiency
4.6.1
Phage-engineered Surface
A suspension of the mixture of roughly equal amounts of non-stained SUPTI cells and
stained K562 cells (~ 105 cells/mL) were flowed over the phage-engineered substrate. The
fluorescent images of the cells inside the flow chamber before the flow started and after
the flow started over the RK1-immobilized substrate are shown in Figure 4-7. The
number of the settled cells before the flow started and the number of the cells attached on
the substrate after the flow started for 60 s were counted.
(a) Before starting the flow
(b) After starting the flow
1~(=O.1
25
dyn/cm2)
20p
Figure 4-7 Fluorescent microscope images of the cells on the RK1-engineered gold
substrate (A) before the flow was started (B) after the flow was started for 60 s. The
white dots are stained K562 cells and black dots are non-stained SUPTI cells.
The capture efficiency (CE) was defined as:
N
CE - -R-xl00%
where
NR
Ns:
(Eq. 4-2)
: Number of the cells remaining on the substrates after the flow started
Number of the cells settling on the substrates before the flow started
Figure 4-8 shows the results of capture efficiency (CE) of different clones (RK1-RK12)
114
at shear stress of 0.1 dyn/cm 2 . Since SUPTI is CD4 expressing cells and K562 is nonCD4 expressing cells, ideally, the SUPTI CE is higher than K562 CE. However, for
some clones (RK2, RK3, RK5, RK7, RK8, RK10, RK11, and RK12), the K562 CE is
higher than that of SUPTI CE. In some cases, RK 4, RK6, and RK9, for example, there is
no significant difference between SUPTI CE and K562 CE although the average values
of CE of SUPTI is higher than that of K562. Among these clones, RKl (DSYSTSMP)
has most difference between the SUPTI and K562 capture efficiencies; however, the
difference was not statistically significant (p=0.06). On the other hand, there is no strong
evidence showing that surface density of phage relates to capture efficiency.
100
l JSUPT1
U K562
80
60
L
4
-J-
20
0
1
2
3
4
5
6
7
8
9
10
11
12
Clone Number
Figure 4-8 Capture efficiencies of RK1-RK12-immobilized substrates at shear stress
of 0.1 dyn/cm2. Error bars represent one standard deviation, where n = 4 replicate
experiments for each condition.
To study the effect of shear flow on the CE, RK1 were immobilized on the substrate and
shear stresses of 0.1, 0.2, 0.3, 0.4, and 0.5 dyn/cm 2 were generated by syringe pump as
described before. Figure 4-9(a) shows the SUPTI and K562 capture efficiencies at
different shear stresses. For both SUPT1 and K562, CEs decrease with increasing shear
stresses. However, difference between SUPTI and K562 CEs does not become more
significant as the shear stress increases.
115
100
0 SUPT1
*EK562
.8
(U1
0
60
-
ul
a)
p=0.06
40
p=0.30
20
0
0.1
r-i
p=0.24
p=0.47
p=O.50
0.2
0.3
0.4
0.5
Shear Stress,
T
(dyn/cm 2)
Figure 4-9 Capture efficiencies of RK1-immobilized substrates. The effect of
different shear stress on the capture efficiencies of the cells on the RKl-immobilized
substrates. Error bars represent one standard deviation, where n = 4 replicate experiments
for each condition.
4.6.2
CD4-blocked Phage-engineered Surface
To confirm that the interaction between the phage and the cells was specific to CD4. The
CD4 proteins were used to block the interaction of phage to the cells. The phageimmobilized slides were incubated with CD4 proteins (5gg/mL) at room temperature for
3 h and then incorporated with the flow chamber. A suspension of SUPT1 cells (~ 105
cells/mL) was flowed over the CD4-blocked RK1-engineered substrate. The microscopic
images of the cells on the substrate inside the flow chamber before the flow started and
after the flow started over the RK1-immobilized substrate are shown in Figure 4-10. It
was observed that the cell-surface interaction was reduced by blocking the surface with
CD4, further confirming that the previously observed interaction was due to CD4.
116
(a) Before starting the flow
Figure 4-10 Microscope images of the SUPT1 cells captured on the CD4-blocked
RK1-engineered gold substrate (a) before the flow was started (b) after the flow was
started for 60 sec.
4.7 Peptide Sequences Characterization of pVIII Library
The sequence characterization of pVIII library was done after the three rounds of
biopanning were completed to identify the possible cause for unsuccessful results. 96
samples were selected and sequenced (the procedure is the same as described before in
section 4.2) and the results are shown in Table 4-4. Some unexpected biases have been
found. For example, RK 7(VSSNGSET) appeared three times (out of 96 samples). This
indicates that this library may not be a good screening pool.
Table 4-4 Peptide sequences from biopanning of pVIII library. (96 samples were
selected)
117
Sequence
Sequence
D
N
S
M
E
G
S
G
D
D
G
D
S
D
M
T
D
M
N
M
S
S
V
E
M
p
p
A
K
p
K
E
E
K
S
D
E
p
p
p
Y
G
N
M
Occurrence
Occurrence
Note
Note
X3
RK2
X2
RK10
P
X2
D
G
p
w
A
V
K
L
A
P
V
A
S
S
N
D
I
Y
A
S
Y
S
D
Y
Y
A
S
A
S
D
M
D
S
D
L
M
P
E
S
T
M
P
S
N
T
L
S
N
M
I
P
N
D
s
N
T
S
A
X3
I
I
118
RK5
X2
0
m
V
m
Ci M
-i
x o
x
-I
M) K
()
Z
M)
j
0
V (A) Z
V)
r-
m
z Zo 0
0 v)--j
- (An (A
-a
-Z-I--Zn
ZZ
m
-
m
C)nQi
KM m
)-0Z
0
0
-<
V-C)(
-- -0<
(An Z
0m
0
MM0
4>
Ln
(A
-n
mn
,
-- 1
C
>
*-<
m
V
>
o
o
m
0
0
Z
V
-<
-<
-<
m L
.-u -<
> *)<
o
mm m mm
>
C)C)C)C)C)C)C)C)C)C)C)C)nm
M
< < < <<
m(A-Ir-Im
<
C)
A A(A-
ZZA
-V v Z
< <<
<<
(A (A
L
< <<
-.) -V
<<
-<
4.8 Conclusion
In this chapter, the goal was to identify specific phage (or peptides) with affinity to CD4
and being able to capture CD4-expressing cells by phage-engineered substrate. 12 clones
of phage was selected by biopanning using pVIII library as the screening pool and the
CD4 protein as the target. Cell-capture experiments were performed on phage-engineered
surfaces and the cells can be eluted from the surface by shear flow. 12 clones show
different surface morphology after they were immobilized on the surface. Among 12
clones, RKl (DSYSTSMP) has most difference between the SUPTI and K562 capture
efficiencies, although this was not statistically significant (p = 0.06). It was observed that
the RK1 phage interaction is specific to CD4 expressed on the cells; however, the SUPTI
capture efficiency of RK1 is only around 21% and this may not be able to provide
sufficient capture efficiency for cell separation application.
There are two possible reasons why the biopanning used in this chapter was not
successful: (1) pVIII library used for biopanning has bias; (2) CD4 immobilized onto the
well plate as the target for biopanning may not be oriented correctly and actively. It has
been shown that the pVIII library used in this chapter is not a good screening pool
because of the present of biased clones. A library containing a wide diversity of
sequences without biases is needed for screening process to increase the possibility of
selecting specific binders. Moreover, the biopanning target for this study, CD4 protein,
was immobilized on the polystyrene surface. The immobilization of proteins on the
polystyrene has been studied extensively and the interaction is mainly based on physical
adsorption [31]. The CD4 may be adsorbed on the polystyrene well plate without specific
orientation. Proteins may be denatured after they are physically adsorbed on the surface
[31, 32]. To serve as the target for biopanning, CD4 should be immobilized onto the
substrate with correct orientation.
4.9
[1]
Chapter Reference
R. P. McEver and C. Zhu, Rolling Cell Adhesion. Annu Rev Cell Dev Biol, 2010.
26: p. 363.
120
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
S. M. Lucila and U. H. von Andrian, Immunological Adhesion and Homing
Molecules, in Els200 1, John Wiley & Sons, Ltd.
L. A. Lasky, Selectin-Carbohydrate Interactions and the Initiation of the
Inflammatory Response. Annual Review of Biochemistry, 1995. 64: p. 113.
R. P. McEver, Selectins: Lectins That Initiate Cell Adhesion under Flow. Curr
Opin Cell Biol, 2002. 14(5): p. 581.
K. L. Moore, K. D. Patel, R. E. Bruehl, F. Li, D. A. Johnson, H. S. Lichenstein, R.
D. Cummings, D. F. Bainton, and R. P. McEver, P-Selectin Glycoprotein Ligand1 Mediates Rolling of Human Neutrophils on P-Selectin. J Cell Biol, 1995.
128(4): p. 661.
G. A. Zimmerman, Two by Two: The Pairings of P-Selectin and P-Selectin
Glycoprotein Ligand 1. Proc Natl Acad Sci U S A, 2001. 98(18): p. 10023.
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Examining the
Lateral Displacement of H160 Cells Rolling on Asymmetric P-Selectin Patterns.
Langmuir, 2011. 27(1): p. 240.
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Karnik, Studying Cell
Rolling Trajectories on Asymmetric Receptor Patterns.J Vis Exp, 2011(48).
C. Edington, H. Murata, R. Koepsel, J. Andersen, S. Eom, T. Kanade, A. C.
Balazs, G. Kolmakov, C. Kline, D. McKeel, Z. Liron, and A. J. Russell, Tailoring
the Trajectory of Cell Rolling with Cytotactic Surfaces. Langmuir, 2011. 27(24):
p. 15345.
S. Choi, J. M. Karp, and R. Karnik, Cell Sorting by Deterministic Cell Rolling.
Lab Chip, 2012. 12(8): p. 1427.
S. Lin, S.-Y. Lee, C.-C. Lin, and C.-K. Lee, Determinationof Binding Constant
and Stoichiometry for Antibody-Antigen Interaction with Surface Plasmon
Resonance. Current Proteomics, 2006. 3(12): p. 271.
C. A. Borrebaeck, A. C. Malmborg, C. Furebring, A. Michaelsson, S. Ward, L.
Danielsson, and M. Ohlin, Kinetic Analysis of Recombinant Antibody-Antigen
Interactions: Relation between Structural Domains and Antigen Binding.
Biotechnology (N Y), 1992. 10(6): p. 697.
S. Hearty, P. Leonard, and R. O'Kennedy, MeasuringAntibody-Antigen Binding
Kinetics Using Surface Plasmon Resonance. Methods Mol Biol, 2012. 907: p.
411.
K. T. Kotz, W. Xiao, C. Miller-Graziano, W. J. Qian, A. Russom, E. A. Warner,
L. L. Moldawer, A. De, P. E. Bankey, B. 0. Petritis, D. G. Camp, 2nd, A. E.
Rosenbach, J. Goverman, S. P. Fagan, B. H. Brownstein, D. Irimia, W. Xu, J.
Wilhelmy, M. N. Mindrinos, R. D. Smith, R. W. Davis, R. G. Tompkins, and M.
Toner, Clinical Microfluidicsfor Neutrophil Genomics and Proteomics. Nat Med,
2010. 16(9): p. 1042.
E. A. Warner, K. T. Kotz, R. F. Ungaro, A. S. Abouhamze, M. C. Lopez, A. G.
Cuenca, K. M. Kelly-Scumpia, C. Moreno, K. A. O'Malley, J. D. Lanz, H. V.
Baker, L. C. Martin, M. Toner, R. G. Tompkins, P. A. Efron, and L. L. Moldawer,
Microfluidics-Based Capture of Human Neutrophils for Expression Analysis in
Blood and BronchoalveolarLavage. Lab Invest, 2011. 91(12): p. 1787.
B. K. Kay, J. Kasanov, and M. Yamabhai, Screening Phage-Displayed
CombinatorialPeptideLibraries.Methods, 2001. 24(3): p. 240.
121
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
M. A. Hemminga, W. L. Vos, P. V. Nazarov, R. B. Koehorst, C. J. Wolfs, R. B.
Spruijt, and D. Stopar, Viruses: Incredible Nanomachines. New Advances with
FilamentousPhages. Eur Biophys J, 2010. 39(4): p. 541.
D. R. Wilson and B. B. Finlay, Phage Display: Applications, Innovations, and
Issues in Phage and Host Biology. Can J Microbiol, 1998. 44(4): p. 313.
D. J. Rodi and L. Makowski, Phage-DisplayTechnology--Findinga Needle in a
Vast Molecular Haystack. Curr Opin Biotechnol, 1999. 10(1): p. 87.
S. S. Sidhu, EngineeringM13for Phage Display. Biomol Eng, 2001. 18(2): p. 57.
M. Szardenings, Phage Display of Random Peptide Libraries: Applications,
Limits, and Potential.J Recept Signal Transduct Res, 2003. 23(4): p. 307.
U. 0. Seker and H. V. Demir, Material Binding Peptidesfor Nanotechnology.
Molecules, 2011. 16(2): p. 1426.
U. Kriplani and B. K. Kay, Selecting Peptides for Use in Nanoscale Materials
Using Phage-DisplayedCombinatorialPeptide Libraries. Curr Opin Biotechnol,
2005. 16(4): p. 470.
R. V. Luckheeram, R. Zhou, A. D. Verma, and B. Xia, Cd4(+)T Cells:
DifferentiationandFunctions. Clin Dev Immunol, 2012. 2012: p. 925135.
M. W. Cloyd, J. J. Chen, P. Adeqboyega, and L. Wang, How Does Hiv Cause
Depletion of Cd4 Lymphocytes? A Mechanism Involving Virus Signaling through
Its Cellular Receptors. Curr Mol Med, 2001. 1(5): p. 545.
L. N. Makoae, C. J. Portillo, L. R. Uys, P. S. Dlamini, M. Greeff, M. Chirwa, T.
W. Kohi, J. Naidoo, J. Mullan, D. Wantland, K. Durrheim, and W. L. Holzemer,
The Impact of Taking or Not Taking Arvs on Hiv Stigma as Reported by Persons
Living with Hiv Infection in Five African Countries. AIDS Care, 2009. 21(11): p.
1357.
M. Xydia, Y. Ge, U. Quitsch, and P. Beckhove, Cd4Ol Co-Stimulationfrom Cd8+
to Cd4+ Effector Memory T Cells Supports Cd4+ Expansion. Immunol Cell Biol,
2011. 89(6): p. 670.
A. H. Patki, S. P. Zielske, S. F. Sieg, and M. M. Lederman, PreferentialS Phase
Entry and Apoptosis of Cd4(+) T Lymphocytes of Hiv-1 -Infected Patients after in
Vitro Cultivation. Clin Immunol, 2000. 97(3): p. 241.
L. Pan, W. J. Weng, L. H. Xu, J. Wei, and J. P. Fang, Separation and
Amplification of Cd4(+)Cd25(+) Regulatory T Cells from Sensitized Mice.
Zhongguo Shi Yan Xue Ye Xue Za Zhi, 2012. 20(2): p. 500.
http://www.biosci.missouri.edu/smithGp.
J. E. Butler, L. Ni, R. Nessler, K. S. Joshi, M. Suter, B. Rosenberg, J. Chang, W.
R. Brown, and L. A. Cantarero, The Physicaland FunctionalBehavior of Capture
Antibodies Adsorbed on Polystyrene. J Immunol Methods, 1992. 150(1-2): p. 77.
V. V. Hlady and J. Buijs, Protein Adsorption on Solid Surfaces. Curr Opin
Biotechnol, 1996. 7(1): p. 72.
122
5
Identification of Adhesion Molecules with Weak Affinity via
Phage Display from M13 pIII Library
5.1
Introduction
In the previous chapter (Chapter IV) CD4 adhesion peptides selected from biopanning
using M13 pVIII library were studied. The selected peptides were demonstrated to
exhibit CD4-dependent adhesion. This chapter discusses study of the selected peptides
obtained from bacteriophage-based biopanning, using a new protocol modified from the
one studied in Chapter IV. This approach demonstrates the binding specificity of the
selected peptides to the CD4 protein.
Two main changes in the biopanning protocol discussed in this chapter are:
(1)
the
screening pool, and (2) the target and the immobilization method used in the biopanning.
Chapter IV established that the specific pVIII library used was not a good screening pool
because of the presence of biased clones. In this chapter, a commercially available M13
pIll library is applied in selecting peptides with affinity to CD4 protein. This pIll library
for M13 phage has been shown to contain a wide diversity of peptide sequences without
population biases [1]. On the other hand, the CD4 protein used in Chapter IV as the
biopanning target was immobilized on polystyrene well plates. The immobilization of
proteins on the polystyrene has been studied extensively, and the interaction is mainly
based on physical adsorption [2]. Thus, it is possible that the CD4 proteins may be
adsorbed on the polystyrene well plate without specific orientation, as shown in Figure 51 (a). Proteins may also be denatured after they are physically adsorbed on the surface [2,
3]. To serve as the target for biopanning, CD4 should be immobilized onto the substrate
with optimal orientation with respect to ligand binding. Nickel (1)-nitrilotriacetic acid
(NTA) chelation has been widely used in isolation and purification of histidine-tagged
proteins [4-9]. The coordination interaction between the metal ion and the histidine
residue provides the possibility of appropriate orientation of the histidine-tagged protein
on the metal ion coated substrates. On the other hand, CD4 has four extracellular domains
(D1,
D2, D3, and D4) where D1 is the outmost. Using peptides selected against
123
uppermost DI and D2 domains may increase capture efficiency of CD4+ cells. Moreover,
DI and D2 domains play an important role in Envelope Glycoprotein GP120 (gp120)
binding during Human Immunodeficiency Virus Type 1 (HIV-1) infection [10-12].
Indentifying a peptide sequence that binds specifically to CD4 presented by T cells may
enable development of HIV monitoring via T cell counts [13, 14].
(a)
CD4
(Dr-D4)
_
Regular PS well plate
(b)
H
9H
9
R-N -C-N-CH-C-R
62
&2
(His-tagged
D1 and D2)
Ni (II) well plate
NNNf)-NTA
Figure 5-1 (a) CD4 protein physically adsorbed on regular polystyrene surface. (b)
His-tagged CD4 protein immobilized on Ni (II)-NTA-coated surface with specific
orientation.
Nickel(II)-nitrilotriacetic acid (NTA) chelation has been widely used in isolation and
purification of proteins tagged with oligo-histidine (His) [4-9]. Figure 5-1(b, left) shows
the chelation between the histidine residues and the Ni(II)-NTA. The NTA can
coordinate the Ni(II) ion with four valencies, and two remaining coordination sites are
available for interaction with the imidazole ring from the histidine residues [7]. Knecht et
al. have comparatively analyzed the binding properties of histidine(His)-tag with various
lengths based on the surface plasmon resonance (SPR) measurements (see Table 5-1) [8].
The association constant (KD) values decrease from 62.7 pM for dihistidine to 0.014 pM
for hexahistidine; however, a further increase in the length of the histidine does not
improve the affinity: the KD values increase from 0.016 FM for heptahistidine to 0.07 pM
for decahistidine [8]. Hence, hexahistidine has been empirically identified as the most
124
effective affinity tag and was used here.
Table 5-1 Binding affinity (dissociation constant (KD) of oligo-histidine peptides to
Ni(II). [81
Oligo-His Length
Kd (pM)
H-(His) 2-OH
62.7 ±
3.4
H-(His) 3-OH
2.23 ±
0.15
H-(His) 4-OH
0.313
± 0.031
H-(His) 5-OH
0.024
± 0.002
H-(His) 6-OH
0.014
± 0.001
H-(His) 7-OH
0.016
± 0.001
H-(His) 8 -OH
0.030
± 0.002
H-(His)9 -OH
0.047
± 0.002
H-(His)10-OH
0.070 ± 0.007
In this Chapter, hexa(His)-tagged CD4 proteins were immobilized onto the Ni(IL)-NTA
coated substrates to serve as the target as shown in Figure 5-1(b) and M13 pIll library
were used as the screening pool for biopanning. ELISA analysis indicates that the
selected phage have specificity to CD4 proteins compared with wild type phage. The
specificity of the corresponding synthetic peptide to the CD4 proteins was confirmed by
staining the CD4-immobilized beads with dye-conjugated peptide.
5.2 Biopanning of pIII Library
Biopanning with the M13 pIII library was conducted in collaboration with Dr. Nimrod
Heldman and Dr. Rana Gosh in Prof. Angela M. Belcher's lab at MIT (DMSE and BE).
5.2.1
Biopanning Procedure
Materials. Hexa(Histidine) tagged recombinant human CD4 molecules were purchased
from Creative Biomart. and was diluted into 1 Opg/mL in phosphate buffered saline (PBS,
Mediatech Inc). The phage screening pool was the PhD.-7TM pIII library (New England
125
Biolab) of the diversity ~10 9. Nickel(II) coated 96-well plates were purchased from
Thermo Scientific. Other materials employed in this chapter can be referred to the text in
Chapter IV unless otherwise specified.
Methods.
A schematic diagram of identifying peptide motif with affinity to CD4 by biopanning
using M13 pIll library is shown in Figure 5-2. The methods (including biopanning
procedure, phage titerring, phage amplification, and DNA extraction) were the same as
described in Chapter IV (section 4.2), except:
-
CD4 is Histidine tagged;
-
96-well plate is Ni(II) coated;
-
Screening pool is PhD.-7TM pIII library;
-
Four rounds of biopaning were completed;
-
For the first round of biopanning, the initial input concentration of phage was 100
copies per sequence in each well (1.28x 1011 phage per well and 4 wells in total).
pill Phage Library
(Diversity -109)
CD4 protein
late
W
Am
g
tion
&Pcation
UN
DNA sequencing
Bound P ages
Unbound Phages
Figure 5-2 Schematic diagram for identifying peptide motifs of affinity to CD4 by
biopanning using pIII library. The pIll library was exposed to the CD4-coated well
plate. Unbound phage was then washed off from the target and the bound phage was
126
eluted from the surface by acid solution. The eluted phage was amplified to serve as the
sub-library for the next round of the panning process. The peptide sequences of the eluted
and amplified phage was identified by titering and DNA sequencing. The panning
process was repeated until the consensus peptides were found.
5.2.2
Biopanning Results
The number of phage per well of input and output solutions and the recovery for each
round of screening are summarized in Table 5-2. The enrichment was observed from the
first round to the fourth round of biopanning for positive selection (i.e., adhesive
sequences). The sequences screened from four panning rounds are listed in Table 5-3.
After the fourth round, there are four peptide motifs with consensus sequences: (1)
FYSHSAETV (2) KVWSLDH (3) KLWTLDW and (4) SPSTHWK. FYSHSAETV is
wild type (WT) phage. Although SPSTHWK appeared multiple times (3 times out of 36
samples) in the fourth round of biopanning, it is identified as a fast grower: it appears
once (out of 29 times) in the third round of biopanning and four times (out of 12 samples)
in the third amplification. The possibility of occurrences increases greatly after
amplification so there is a high possibility that SPSTHWK is a fast grower. Two
remaining clones were then selected (KVWSLDH and KLWTLDW) and named as
AMB1 and AMB2, respectively, and amplified for further study. Peptide properties of
AMB 1 and AMB2 are shown in Table 5-4. It is interesting that the AMB 1 and AMB2
have similar charge distribution. The first six amino acids in the order are positively
charged, hydrophobic, aromatic, polar, hydrophobic, and negatively charged for both
phage. This indicates that they may have similar binding interaction to CD4 proteins.
Table 5-2 Input and output concentration of phage and the recovery of each round
of screening. The concentration is defined as the number of the phage per well. For each
round of screening, four wells coated with CD4 were used.
Round of screening
1
2
3
4
Input phage
(#/well)
1.28x10"
Output phage
(#/well)
2.38x10 6
Recovery
(Input/Output)
5.57x10-6
1011
2.85x106
2.85x1O-s
1011
1.02x10
6
1.02x10-5
4.75x10
6
4.75x10-5
1011
127
Table 5-3 CD4 protein binding sequences from biopanning of pIII library.
Round 1
panning
Sequence
F
L
G
H
G
S
G
S
L
R
S
H
T
L
V
G
V
L
Y
L
Y
S
Ocurec
Occurrence
N
H
A
L
V
Y
E
W
N
P
Y
H
P
R
I
H
Q
K
I
V
K
V
D
M
N
A
L
P
L
S
K
I
S
Round I
amplification
N
G
V
R
L
Y
L
T
G
L
F
P
K
S
T
S
K
R
A
H
G
S
K
G
N
S
D
T
G
H
R
V
P
T
R
K
L
A
L
Round 2
panning
L
P
X2
T
T
w
N
P
Q
N
K
L
S
X2
R
H
R
L
V
L
Round 2
amplification
128
X2
X2
Round 3
panning
Round 3
amplification
Sequence
SeauenceOcunc
H
A
A
P
L
A
M
A
D
F
V
D
L
D
I
H
E
Y
P
F
A
F
H
K
I
H
I
E
L
L
P
L
S
L
M
R
D
N
R
E
N
H
M
N
N
w
Q
N
H
S
S
D
I
S
F
K
S
P
C
I
S
T
V
T
N
T
T
P
V
D
L
A
N
S
x
Y
H
K
Y
N
L
Occurence
Y
H
L
K
R
T
S
Q
Y
L
S
H
w
A
T
S
T
E
T
K
T
w
T
K
K
T
S
V
E
L
P
S
s
T
T
G
T
T
T
A
E
L
A
Q
H
R
T
Y
T
T
V
G
Y
H
Y
N
R
Q
L
M
E
N
T
S
M
M
H
G
P
L
G
V
V
R
A
V
I
I
Q
S
T
T
x
T
R
G
L
x
w
L
K
A
T
L
w
A
w
E
E
F
N
Q
S
Y
Round 4
panning
H
w
T
S
G
S
H
T
Y
H
Q
P
S
T
R
w
T
F
N
0
T
Q
A
L
L
S
D
D
L
A
A
D
F
H
K
K
K
L
N
R
R
S
S
T
T
T
T
V
W
H
w
N
F
G
S
V
129
X2
T
V
X2
X4
T
V
X4
X7
X6
X3
Table 5-4 Selected peptide sequences and their properties. (negatively charged amino
acids in red; positively charged amino acids in blue; polar amino acid without charged in
yellow; hydrophobic amino acid in gray; aromatic amino acid in green; Histidine in
orange).
AMB 1
AMB 2
S
VT
L
5.3 Enzyme-linked Immunosorbent Assay (ELISA)
The enzyme-linked immunosorbent assay (ELISA) is a common technique to detect the
presence of antigen [15-17]. After the antigen is immobilized on the surface (usually a
well plate), the detection antibody conjugated to an enzyme (or the antibody can be
detected by a secondary antibody which is conjugated to an enzyme) is added, forming a
complex with the antigen. Several washing steps are repeated between each ELISA step
to remove unbound materials. Finally, the plate is developed by adding and enzymatic
substrate to produce a colored end product, which provides visible signal. The signal
correlates to the amount of antigen present in the well plate and can be qualitatively
analyzed. ELISA was used here to test the above peptides for specific binding to CD4
proteins
5.3.1
5.3.1.1
Surface Density of CD4 on Well Plate as the Target of Biopanning
Materials
Anti-CD4 antibody (MEM- 115, a mouse monoclonal IgG 2a; Santa Cruz Biotechnology,
Inc) was prepared at a 1:1000 dilution in PBS (with 2% BSA) and this antibody detects
an epitope in the DI domain of CD4. Rabbit Anti-Mouse IgG 2 a - horseradish peroxidase
(HRP) was phurased from Life technologies and was prepared at a 1:1000 dilution in
PBS (with 2% BSA). Turbo 3,3',5,5'-Tetramethylbenzidine (TMB) was purchased
from Thermo Scientific. IM sulfuric acid (H2 SO 4) was purchased from Sigma-Aldrich.
130
5.3.1.2
Method
ELISA was used to characterize the surface density of CD4 protein on well plates, which
were served as the target of biopanning in previous and this chapters. Preparation of the
CD4-coated well plates was discussed in the sections of 4.2.2 and 5.2.1. Three wells of
each immobilization method were prepared. 100 pL anti-CD4 IgG 2a dilutions were then
added to into each well and then incubated for 1 h at room temperature. The antibody
solution was poured off. The wells were then cleaned by filling each well with PBS and
slapped face down onto the clean paper towel ten times. Next, 100 pL anti-Mouse IgG 2 a
-
HRP dilutions were then added to into each well and then incubated for 1 h at room
temperature. The wells were then washed ten times. Next, 100 gL Turbo TMB was
added to each well and then incubated for 30 min at room temperature. 100 pl 1 M
H2SO 4 was then added to each well. TMB substrates yield a blue color when detecting
HRP and the color changes to yellow with the addition of H2 SO 4 at the maximum
absorbance wavelength of 450 nm.
5.3.1.3
ELISA Analysis
The intensities of coloration at wavelength of 450 nm (yellow) were measured using
microplate reader (Tecan M1000) and the results were compared with the intensity
characterized
from the enzyme substrate
reaction with different antibody-HRP
concentrations. The surface density for the CD4 (with Dl-D4 domains) physically
adsorbed on the polystyrene well plate used in previous chapter (Chapter IV) was
calculated to be around 4.7-9.3x 109 /cm 2 . The surface density for the target used in this
chapter was about 3.4-6.8x 10'2 /cm 2 . The results suggests that the coordination
interaction between the Ni(II) and the histidine provided appropriate orientation of
histidine-tagged CD4 protein on the Ni(II) substrates and greatly enhanced the binding
density of the CD4 protein on substrate by about three orders of magnitude.
131
5.3.2
Specificity of the Selected Phage to CD4 Protein
5.3.2.1
Materials
Bovine serum albumin (BSA, Rockland Immunochemicals, Inc.) solution was prepared
by BSA in PBS to a final concentration of 3 wt%. AMBI, AMB2 and Wild type (WT)
phage was amplified (procedure was the same as what has been described in section 4.2)
for this study. HRP-conjugated anti-M13 monoclonal antibody (GE Healthcare) was
prepared at a 1:10000 dilution in PBS (with 2% BSA); this antibody recognizes and
epitope on pVIII protein of M13 phage. Other materials can be referred to the text in
sections of 5.2.1 and 5.3.1.1.
5.3.2.2
Methods
ELISA was used to confirm the specific affinity of the screened phage to the CD4 protein
molecules. Three independent experiments were performed. Quantitative data were
presented as mean and standard deviation of the values obtained from each experiment. A
schematic diagram of ELISA procedure is shown in Figure 5-3. The detailed procedure is
described below:
Immobilization of CD4 protein on Ni(II) well plate. The diluted CD4-His protein
solution was added into Ni(II) well plate in the well number Al-Gi, A3-G3 and A5-G5
(100 pL/well; 1 pgCD4 /well). The plate was incubated overnight at 4'C in the
refrigerator. Next day, the plate was taken out to room temperature. The CD4-His
solution was poured off. The wells were cleaned by filling each well with PBS and then
slapped face down onto the clean paper towel ten times. The BSA solution was then
added into Al-Gl, A2-G2, A3-G3, A4-G4, A5-G5, and A6-G6 (300 pL/well) to block
non-specific interaction and kept at room temperature for 3 hours. The BSA solution was
poured off and the wells were cleaned again by filling each well with PBS and then
slapped face down onto the clean paper towel 10 times.
Exposure of the phage onto the CD4-coated well plate. Different dilutions (1013, 101,
101,
1010, 101, and 108/mL) of phage solutions were prepared and added to the coated
132
wells (100 pL/well). Column 1 and column 2 of the plate were incubated with AMB 1.
Column 3 and Column 4 were incubated with AMB2. Column 5 and Column 6 were
incubated with wild type (WT) phage. In Row A, B, C, D, E, F, G, the phage
concentration is 1012, 1011
10,10
,
108, 10 7 , and 0 per well, respectively. The
microplate was then kept overnight in the 4'C refrigerator. Next day, the microplate was
taken out to the room temperature. The wells were cleaned by filling each well with BSA
solution and slapped face down onto the clean paper towel 5 times. The plate was then
cleaned by filling each well with PBS and slapped face down onto the clean paper towel
10 times.
Detection of phage by anti-M13-HRP antibody. 100 tL Anti-M13-HRP dilutions were
added to into each well (Al-Gi, A2-G2, A3-G3, A4-G4, A5-G5, and A6-G6) and then
incubated for 1 h at room temperature. The antibody solution was poured off. The wells
were then cleaned by filling each well with PBS and slapped face down onto the clean
paper towel ten times.
Enzyme-substrate reaction. 100 ptL Turbo TMB was added to each well (Al-G1, A2G2, A3-G3, A4-G4, A5-G5, and A6-G6) and then incubated for 30 minutes at room
temperature. The TMB substrate yielded a blue color when HPR were detected. Next,
100 ptL 1 M sulfuric acid was then added to each well (Al-Gi, A2-G2, A3-G3, A4-G4,
A5-G5, and A6-G6). The solution changed it color to yellow after sulfuric acid was
added.
133
Sam
NI(lI) surface
CD4-HIs
Phages
Colored product
HRP-conjugated
o0
TMB substrate
AntI-M13
00
0 sa
TMB substrate
Figure 5-3 Schematic diagram of ELISA procedure. CD4-His protein was
immobilized on Ni (II) coated well plate. The phage was then exposed to the CD4-coated
well plate. HRP-conjugated anti-M13 antibody binds specifically to the target (pVIII
protein of the phage). Several washing steps were repeated between each ELISA step to
remove unbound materials. Substrate (TMB) was added and the colored end product
from enzyme-substrate reaction is produced. The color signal correlates to the amount of
the phage present in the well plate.
5.3.2.3
ELISA Analysis
ELISA analysis was used to identify the affinity of the selected phage (AMBl and
AMB2) to CD4 proteins. The ELISA qualitative and quantitative results are shown in
Figure 5-4(a) and Figure 5-4(b), respectively. Three 96-well plates were used to perform
the experiment. For each well plate, column 1, 3, and 5 were coated with CD4 proteins
then BSA, while column 2, 4, and 6 were coated with BSA only as control. Three clones
134
of phage (AMB 1, AMB2, and WT) were then added into the wells. AMB 1 phage was
added to column 1 and 2. AMB2 phage was added to column 3, and 4. WT phage used as
control was added to column 5 and 6. The ELISA qualitative results are shown in Figure
5-4(a). The phage was titered and the input concentration were labeled in the Figure 54(a). It was observed that the wells coated with CD4 protein are darker than the wells
coated with only BSA and the color of the well becomes darker as the input concentration
of the phage increases. The darker wells indicate higher concentration of anti-M13-HRP,
which also means higher concentration of phage. To qualitatively compare the AMB1,
AMB2 and WT phage, the intensities of coloration at wavelength of 450 nm (yellow)
were measured using microplate reader (Tecan M1000); The ELISA quantitative results
are shown in Figure 5-4(b). The detection limits for AMB1 and AMB2 are around
101-109 peptides per well. At the input concentration of 1012 per well, high absorbance
may result from non-specific interaction between the phage and the coated molecules. It
can be seen from the comparisons of the results of three types of phage at the input
concentration around 1014-10" per well, the binding of AMBI and AMB2 to CD4
protein is more effective than that of WT phage. Moreover, it was demonstrated that
AMB 1 and AMB2 are more specific to CD4 proteins than to the control molecules (BSA).
These results also suggest that the biopanning process results in good affinity selection.
135
AMBi
(a)
Phage conc.
[number/well]
:CD4/
AMB2
ICD4/
IsA BSA
BS
s
A
I CD4/
I'
BSA BSA
IA
A
B
1010
C
109
D
103
E
10"
F
0
G
(b)
-0- AMB1-CD4/BSA
00 ABM1-BSA
AMB2-CD4/BSA
+ ABM2-BSA
WT-CD4/BSA
WT-BSA
1
0.8
C
0.6
.0
0.4
A.
0.2
0
6
7
g
9
12
13
Log [phage conc]/well
Figure 5-4 ELISA assay output. (a) Qualitative results. Representative results are
shown for only one experiment at each condition. Wells in column 1, 3, and 5 were
coated with CD4 and then BSA. Wells in column 2, 4, and 6 were coated with only BSA
as control. Phage with different input concentration were added. Phage AMBi were
incubated in column 1 and 2. Phage AMB2 were incubated in column 3 and 4. WT phage
was incubated in column 5 and 6; (b) Quantitative results. The relationship between the
absorbance of the colored end product and the input concentration of the phage. Error
bars represent one standard deviation, with n = 3 replicate experiments for each
condition.
136
5.4
5.4.1
Specificity of the Synthetic Peptide to CD4 protein
Synthetic peptide
The custom peptide was synthesized by GenScript Corp. The sequence of the peptide is
KLWTLDWGGGS-(CH 2 CH 2O) 6 -C (from N-terminus to C-terminus). KLWTLDW is the
sequence of AMB2. The sequence of GGGS was designed to function as the spacer. The
six repeat units of poly(ethylene oxide) were added to prevent further non-specific
interaction between the peptide and other molecules. Cysteine containing of the thiol
functional group at the C-terminus was designed for further use in functionalization.
5.4.2
Conjugation of Peptide to the Alexa Fluor 488 Dye
Alexa Fluor@ 488 (AF 488) C5 Maleimide (C 3oH 2 5N4 NaOi 2 S2 ) is thiol-reactive and was
purchased from Life Technologies. Figure 5-5 shows the reaction scheme of the dyemaleimide with the thiol-peptide. The peptide stock solution was made by dissolving the
peptide in PBS to a concentration of 100 pM. AF 488 C5 Maleimide was added to the
peptide stock solution to a concentration of 1 mM. The reaction was then allowed to
proceed for 2 h at room temperature. This AF 488 dye-conjugated peptide solution may
contain free AF 488 dye, non-conjugated peptide and AF 488 dye-conjugated peptide and
needs further purification.
o
Alexa Fluor'
F0
Figure 5-5 Reaction of the Alexa Fluor@ 488 C5 Maleimide with a thiol-peptide.
137
5.4.3
Purification of Dye-conjugated Peptides
A polyacrylamide desalting column (gravity-flow column; Thermo Scientific) was used
for recovering the molecules of the molecular weight larger than 1800 Da and the small
compounds can be removed fully. The molecular weight of the peptide is 1684 and the
molecular weight of the dye-conjugated peptide is 2405 and thus this column can be used
for separating the dye-conjugated peptide form the solution. Some PBS buffer was first
added into the column to equilibrate the column. The AF 488 dye-conjugated peptide
solution (-5mL) was then added to the column. Several collection tubes were used to
collect the drained samples (see Figure 5-6).
Figure 5-6 Collected drained samples. The first five fractions were collected before the
dye-conjugated peptide solution drained through and thus contained mostly buffer
(colorless). The orange color came from dye-conjugated peptide. The non-conjugated dye
shows yellow color.
Fraction #8 was selected as the stock solution for further study. The concentration of this
AF 488 dye-conjugated peptide was characterized by spectrophotometer (Nanodrop@
ND1000). The concentration was determined to be 250 gM by Eq. 5-1 (Beer-Lambert
Law).
Awhere
A:
C:
S:
C:
-c-l
the absorbance at 488 nm
extinction coefficient (71000 M-1 cm~)
the concentration of the dye-conjugated peptide
path length
138
(Eq. 5-1)
5.4.4
Staining the CD4-functionalized Beads with the Dye-conjugated Peptides
Materials. Ni(II)-NTA agarose beads (average bead size ~ 34 pm) were purchased from
QLAgen. Alexa Fluor 647@ conjugated bovine albumin serum (AF 647 dye-conjugated
BSA) were purchased from Life Technologies and diluted to 5 mg/mL in PBS. Blocking
solution was prepared by dissolving BSA in PBS to a final concentration of 5 wt%.
Incubating BSA-coated Ni(II) Beads with Dye-conjugated Peptides (as Control)
Ni(II) beads were spun down and the supernatant was discarded. The pellet was
suspended in suspended in blocking solution (BSA, 5 mg/mL in PBS) and incubated at
room temperature for 3 h. Several washing steps were then repeated to remove free BSA.
The beads were then incubated with 10 piM dye-conjugated peptide at room temperature
for 1 h. Several washing steps were then repeated to remove unbound peptides.
Functionalization of Ni(II) Beads with CD4 Protein
Ni(II) beads were transferred into 1 mL microtube and were spun down for 15-30 s. The
supernatant was discarded and the pellet was suspended in CD4-His solution (100 pg/mL
in PBS) at a final concentration of 106 beads/mL. The solution was then incubated at 4'C
overnight. Several washing steps were then repeated to remove unbound CD4-His
molecules. The beads were then incubated in blocking solution at room temperature for 3
h.
Functionalization of Ni(II) Beads with Dye-conjugated BSA Protein (as Control)
Ni(II) beads were spun down and the supernatant was discarded. The pellet was
suspended in AF 647 dye-conjugated BSA (5 mg/mL in PBS) at a final concentration of
106 beads/mL. The solution was then incubated at room temperature for 10 min. The
beads were then spun down again and the pellet was then suspended in blocking solution
and incubated at room temperature for 3 h.
Staining Functionalized Beads with Dye-conjugated Peptides
139
The specificity of the synthetic peptides to the CD4 protein was confirmed by staining the
functionalized beads with dye-conjugated peptides. The schematic diagram is shown in
Figure 5-7. Two types of functionalized beads (CD4-coated beads and AF 647-BSAcoated beads) were mixed and incubated with BSA for 3 h at room temperature to block
non-specific interaction. Several washing steps were applied and the mixed beads were
then incubated with 10 pM AF 488-conjugated peptides at room temperature for 1 h.
Several washing steps were then repeated to remove unbound peptides.
AF 488-conjugated peptide
KLWTLDWGGGS-(CH 2CH 20)sW
CD4-HIs
---
BSA
00
BSA-AF 647
Figure 5-7 Schematic diagram for staining the CD4-coated beads with dyeconjugated peptide. The Ni(II) beads were incubated with CD4-His and AF 647 conjugated BSA, separately. The beads were then mixed and incubated with BSA to
block non-specific interaction. The mixed CD4 and AF 647-BSA-coated beads were then
incubated with AF 488-conjugated peptide. Several washing steps were repeated between
each step to remove unbound materials before the samples were imaged.
5.4.5
Characterization
of
Binding
Specificity
between
Peptide
and
the
Functionalized Beads
An inverted fluorescence microscope (Nikon TE2000-U) with a mounted camera (Andor
iXon 885) was used to examine autofluorescence of CD4-immobilized beads, and
binding specificity of between peptides and the functionalized beads. A 40x objective
was used. Different wavelengths were applied to image the mixture of the beads (see
Figure 5-8). To identify if CD4 protein shows autofluorescence, CD4-immobilized beads
were first imaged under bright field (Figure 5-8 (a)) and FITC channel (Figure 5-8 (b)).
No autofluorescence was observed from FITC channel. Moreover, to characterize if
140
selected peptides bind to BSA, BSA-coated beads stained with AF 488 dye-conjugated
peptides were imaged under bright field (Figure 5-8 (c)) and FITC channel (Figure 5-8
(d)). No fluorescence was observed from FITC channel, which suggests that there was no
binding between BSA-coated beads and peptides. Finally, to confirm the specificity of
selected peptide to the CD4 protein, mixed beads (beads coated with CD4 protein and
beads coated with AF 647 dye-conjugated BSA) were stained with AF 488 dyeconjugated peptides and then imaged (Figure 5-8(e) ~ Figure 5-9(h)). Eleven such beads
were first imaged under bright field (Figure 5-8(e)) and the Cy5 channel (Figure 5-8(f)).
By comparing images of Figure 5-8(e) and Figure 5-8(f), beads number 1, 2, 6, and 11
were identified as the ones coated with AF 647 dye-conjugated BSA, which also suggests
that number 3, 4, 5, 7, 8, 9, and 10 were coated with CD4 proteins. The mixed beads were
then imaged under FITC channel Figure 5-9(g) and it shows the AF 488 dye-conjugated
peptides bound to the CD4 coated beads (number 3, 4, 5, 7, 8, 9, and 10). Figure 5-9(h) is
the Cy5/FITC channel merged image. These images indicated that selected peptide bound
to only CD4-immobilized beads, which confirm the specificity of screened peptides to
the CD4 protein.
141
(a)
(C)
(d)
(e)
(f)
(h)
(g)
100
Figure 5-8 Fluorescence microscope analysis. (a-b) CD4-immobilized beads. Images
of (a) bright field and (b) FITC channel. CD4-immobilized beads do not show
autofluorescence. (c-d) BSA-coated beads that were incubated with AF 488 dyeconjugated peptides. Images of (c) bright field and (d) FITC channel. AF 488 dyeconjugated peptides do not show non-specific binding to BSA-coated beads. (e-f) The
mixture of the CD4-immobilized beads and the AF 647 dye-BSA-coated beads
stained with AF 488 dye-conjugated peptides. Images of (e) bright field; (f) Cy5
channel; and (g) FITC channel; (h) is the Cy5/FITC channel merged image. AF 488 dye
shows bright, green fluorescence and AF 647 shows bright, red fluorescence. Beads
number 1, 2, 6, and 11 were coated with BSA and beads number 3, 4, 5, 7, 8, 9, and 10
142
were immobilized with CD4 proteins. The dye-conjugated peptide bound only to CD4immobilized beads.
5.5 Conclusion
Biopanning using the M13 pIII library as screening pool and His-tagged CD4 on Ni(II)
substrates as target were performed. The modified immobilization method resulted in an
increase in the surface density of the target. Two consensus peptides were found, and
selected phage and corresponding synthetic peptide were demonstrated to show specific
binding to CD4 proteins. This initial identification of peptides specific to CD4 proteins
prompts further studies to demonstrate that CD4+ cells or CD4 protein-functionalized
beads can roll or be captured on peptide-functionalized surfaces.
5.6 Reference
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
New England Biolabs, Ph.D. - 7TM Phage Display Peptide Library Kit Instruction
Manual.
J. E. Butler, L. Ni, R. Nessler, K. S. Joshi, M. Suter, B. Rosenberg, J. Chang, W.
R. Brown, and L. A. Cantarero, The Physicaland FunctionalBehavior of Capture
Antibodies Adsorbed on Polystyrene. J Immunol Methods, 1992. 150(1-2): p. 77.
V. V. Hlady and J. Buijs, Protein Adsorption on Solid Surfaces. Curr Opin
Biotechnol, 1996. 7(1): p. 72.
K. Terpe, Overview of Tag Protein Fusions: From Molecular and Biochemical
Fundamentalsto Commercial Systems. Appl Microbiol Biotechnol, 2003. 60(5):
p. 523.
L. Nieba, S. E. Nieba-Axmann, A. Persson, M. Hamalainen, F. Edebratt, A.
Hansson, J. Lidholm, K. Magnusson, A. F. Karlsson, and A. Pluckthun, Biacore
Analysis of Histidine-TaggedProteins Using a Chelating Nta Sensor Chip. Anal
Biochem, 1997. 252(2): p. 217.
X. Y. Dong, X. D. Feng, and Y. Sun, His-Tagged ProteinPurificationby MetalChelate Affinity Extraction with Nickel-Chelate Reverse Micelles. Biotechnol
Prog, 2010. 26(4): p. 1088.
H. Block, B. Maertens, A. Spriestersbach, N. Brinker, J. Kubicek, R. Fabis, J.
Labahn, and F. Schafer, Immobilized-Metal Affinity Chromatography (Imac): A
Review. Methods Enzymol, 2009. 463: p. 439.
S. Knecht, D. Ricklin, A. N. Eberle, and B. Ernst, Oligohis-Tags:Mechanisms of
Binding to Ni2+-Nta Surfaces. J Mol Recognit, 2009. 22(4): p. 270.
143
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
M. De, S. Rana, and V. M. Rotello, Nickel-Ion-Mediated Control of the
Stoichiometry of His-Tagged Protein/Nanoparticle Interactions. Macromol
Biosci, 2009. 9(2): p. 174.
L. Li, X. Shi, Q. Lu, S. Zhang, X. Wang, X. Jiang, Y. Liu, G. Wang, W. Zhu, R.
Lei, and H. Wu, Role of Human Cd4 Dld2 Domain in Hiv-1 Infection. Immunol
Invest, 2013. 42(2): p. 106.
L. J. Matthias, P. T. Yam, X. M. Jiang, N. Vandegraaff, P. Li, P. Poumbourios, N.
Donoghue, and P. J. Hogg, Disulfide Exchange in Domain 2 of Cd4 Is Required
for Entry of Hiv-1. Nat Immunol, 2002. 3(8): p. 727.
D. Sharma, M. M. Balamurali, K. Chakraborty, S. Kumaran, S. Jeganathan, U.
Rashid, P. Ingallinella, and R. Varadarajan, Protein Minimization of the Gp120
Binding Region ofHuman Cd4. Biochemistry, 2005. 44(49): p. 16192.
X. H. Cheng, D. Irimia, M. Dixon, K. Sekine, U. Demirci, L. Zamir, R. G.
Tompkins, W. Rodriguez, and M. Toner, A Microfluidic Device for Practical
Label-Free Cd4+ T Cell Counting of Hiv-Infected Subjects. Lab on a Chip, 2007.
7(2): p. 170.
X. Cheng, A. Gupta, C. Chen, R. G. Tompkins, W. Rodriguez, and M. Toner,
Enhancing the Performance of a Point-of-CareCd4+ T-Cell Counting Microchip
through Monocyte Depletionfor Hiv/Aids Diagnostics. Lab Chip, 2009. 9(10): p.
1357.
A. Voller, A. Bartlett, and D. E. Bidwell, Enzyme Immunoassays with Special
Reference to Elisa Techniques. J Clin Pathol, 1978. 31(6): p. 507.
A. Voller, The Enzyme-Linked Immunosorbent Assay (Elisa) (Theory, Technique
and Applications). Ric Clin Lab, 1978. 8(4): p. 289.
R. M. Lequin, Enzyme Immunoassay (Eia)/Enzyme-Linked ImmunosorbentAssay
(Elisa). Clin Chem, 2005. 51(12): p. 2415.
144
6 Conclusion and Future Work
Weak but specific interaction play an important role in many biological and physiological
processes such as bacterial adhesion [1], cell homing [2, 3], and immune surveillance [1,
4]. Selectin and their ligand interaction resulting in rolling motion of leukocytes, stem
cells and cancer cells has been widely studied. Patterning selectin with multiple inclined
bands inside microfluidic devices opens new possibilities for separating cells by diverting
the rolling direction of cells[5]. Therefore, systematic study of how cells flow inside the
microfluidic channel, how cells get captured onto the patterned substrates, how substrates
parameters affect cell rolling trajectories or separation potential, and how to identify new
receptor-ligand system is prerequisite for the design of the substrates involving transient
interactions between the cells and the receptor-patterned substrates. In this thesis, HL60
cells and P-selectin were used as a model system to study development of adhesive
substrates and biopanning was used as the strategy to identify new receptor-ligand
system. The specific contributions of the thesis are:
-
Identified shear stress controlled the transport of cells inside microfluidic device
and capture efficiency of cells by receptor-functionalized substrates from the free
stream.
-
Developed empirical equations to predict capture efficiency by adhesive
substrates.
-
Developed a standard protocol using microcontact printing to create biomimetic
adhesive substrate.
-
First observation of hopping of cells on successive inclined patterns.
-
Identified edge inclination angle affected edge tracking length of rolling cells on a
single receptor band significantly.
-
Developed a standard approach to predict device performance and provided
guidelines for future device design
-
Identified anti-CD4 peptides, which potentially could enable cell rolling, using
biopanning.
145
The biomimetic adhesive microfluidic substrates for a new method to isolate cells in a
continuous manner based on transient interaction of specific ligands on cell surface with
inclined receptor bands have been developed and this study has led to realization of
microfluidic device done by Bose el al for sorting leukocytes directly from blood using
P-selectins in continuous flow [6].
The fact that separation length is most sensitive to shear stress since shear stress limits
settling distance and capture efficiency of cells by P-selectin from the free stream in this
technique. This thesis work suggests to lower shear stress in order to enhance separation
potential for P-selectin-patterned microfluidic substrates. Moreover, it has been
demonstrated that shear stress has no significant effect on edge tracking length after cells
were captured. However, these statements may not be true for other adhesion molecules.
For example, for L-selectin-based microfluidic devices, lower shear stress may not result
in increasing capture efficiency because of shear threshold phenomenon. Detail studies of
cell attachment and substrate parameters are needed for developing microfluidic systems
based on new adhesion molecules.
On the other hand, hopping behavior over successive patterns was observed in both flow
chamber and device [6, 7] and it greatly increases the overall total lateral displacement of
deflected cells, and therefore enhances device separation efficiency. However, the
mechanism why cells hopped after they detached from the edge was not clear. What
parameters may affect hopping behavior has not been identified. Moreover, study of how
to increase hopping probability would help enhance device performance. For example,
will decreasing gap width significantly increase the hopping probability? If yes, what will
be the minimum gap width which leads to maximum hopping probability without making
cell cross over the patterns? Therefore, understanding and quantification of hopping
behavior would be useful for development of device design. To characterize hopping
behavior, parallel receptor bands aligned perpendicular to flow direction can be designed
as shown in Figure 6-1. Receptor bands are patterned on the substrate with different gap
widths. The cells flow over the pattern and the minimum gap width without making cell
cross over the pattern can be identified by studying the interaction between cells and the
146
pattern. Studying instantaneous flowing velocity of cells after they detach from the
receptor bands may be interesting, too. For example, how the instantaneous flowing
velocity changes from the moment the cell detach until it captures by a downstream band.
Such study may help clarify the mechanism of hopping behavior of cells.
I
0
I
I
U-
)ping
pi
I
0.
Z
CL
0
Q
U_
G1
G3
G2
(G1 < G2 < G3)
Figure 6-1 Experimental design for characterizing hopping behavior of cells. Cells
flow over parallel receptor bands with different gap widths aligned perpendicular to flow
direction. The minimum gap width without making cell cross over the patterns can be
identified.
Since hopping probability was observed to be higher than capture probability from the
free stream, increasing capture probability from the free stream will enhance device
performance. 3D vertical hydrodynamic flow-focusing device was developed to
synthesize polymer particle [8]. Similar design may help force cells to interact with
substrates and therefore increase capture efficiency. Settling distance may reduce by such
design as well.
Using only known biological molecules such as selectins [9, 10] that enable cell to roll
limits the application of biomimetic adhesive substrates studied in this thesis. To explore
potential application for current technique, design or evolution of synthetic adhesion
molecules against arbitrary surface marker is needed. While the present work does not
147
directly address the feasibility of separating different types of cells that interact with
synthetic adhesion molecules, it lays the groundwork for future studies in this direction.
Reference:
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
W. E. Thomas, V. Vogel, and E. Sokurenko, Biophysics of Catch Bonds. Annu
Rev Biophys, 2008. 37: p. 399.
P. J. Quesenberry and P. S. Becker, Stem Cell Homing: Rolling, Crawling, and
Nesting. Proc Natl Acad Sci U S A, 1998. 95(26): p. 15155.
T. Lapidot, A. Dar, and 0. Kollet, How Do Stem Cells Find Their Way Home?
Blood, 2005. 106(6): p. 1901.
R. P. McEver and C. Zhu, Rolling Cell Adhesion. Annu Rev Cell Dev Biol, 2010.
26: p. 363.
R. Karnik, S. Hong, H. Zhang, Y. Mei, D. G. Anderson, J. M. Karp, and R.
Langer, Nanomechanical Control of Cell Rolling in Two Dimensions through
Surface PatterningofReceptors. Nano Lett, 2008. 8(4): p. 1153.
S. Bose, R. Singh, C. Shen, M. Hanewich-Hollatz, C. H. Lee, D. M. Dorfman, J.
M. Karp, and R. Karnik, Affinity Flow Fractionationof Cells Via Transient
Interactionswith Asymmetric Molecular Patterns.Scientific Reports, 2013(3): p.
2329.
C. H. Lee, S. Bose, K. J. Van Vliet, J. M. Karp, and R. Kamik, Examining the
Lateral Displacement of H160 Cells Rolling on Asymmetric P-Selectin Patterns.
Langmuir, 2011. 27(1): p. 240.
M. Rhee, P. M. Valencia, M. I. Rodriguez, R. Langer, 0. C. Farokhzad, and R.
Karnik, Synthesis of Size-Tunable Polymeric Nanoparticles Enabled by 3d
Hydrodynamic Flow Focusing in Single-Layer Microchannels. Adv Mater, 2011.
23(12): p. H79.
M. E. Taylor and K. Drickamer, Paradigmsfor Glycan-Binding Receptors in Cell
Adhesion. Curr Opin Cell Biol, 2007. 19(5): p. 572.
S. Hakomori, Le(X) and Related Structures as Adhesion Molecules. Histochem J,
1992. 24(11): p. 771.
148