HYDROGEL BASED MEMBRANES
FOR BIOPHARMACEUTICAL AND
BIOMEDICAL APPLICATIONS
i
HYDROGEL BASED MEMBRANES FOR
BIOPHARMACEUTICAL AND BIOMEDICAL
APPLICATIONS
By SEUNG MI YOO, MASc.
A Thesis Submitted to the School of Graduate Studies in Partial Fulfillment of the
Requirements for the Degree Doctor of Philosophy
McMaster University © Copyright by SEUNG MI YOO, April 2014
II
DOCTOR OF PHILOSOPHY (2014)
McMaster University
(Chemical Engineering)
Hamilton, Ontario
TITLE:
HYDROGEL BASED MEMBRANES FOR
BIOPHARMACEUTICAL AND BIOMEDICAL
APPLICATIONS
AUTHOR:
SEUNG MI YOO, MASc. (University of Waterloo)
SUPERVISOR:
Professor Raja Ghosh
NUMBER OF PAGES:
xxv, 141
iii
Abstract
Membrane technology has been actively used as a separation tool in the chemical,
environmental, and biopharmaceutical industries for several decades. As membrane
quality requirement in the industry has increased, efforts have been directed towards
enhancement in mechanical strength, chemical durability and functionality of membranes.
One of the approaches for membrane quality enhancement is based on the combination of
hydrogel technology with membrane technology. This thesis focused on the application
and development of hydrogel based membranes, notably hydrophilized PVDF
(polyvinylidene fluoride) membrane for hydrophobic interaction membrane
chromatography; the fabrication of paper-hydrogel composite membranes for membrane
chromatography; development of a technique for coating alginate (a natural hydrogel) on
the outer surface of a hollow fiber membrane for potential application in bioreactors and
the use of hollow fiber membranes as mold for fabrication calcium alginate fibers for
biomedical and tissue engineering applications.
A membrane chromatography-based polishing technique was developed for removing
leached protein-A and aggregates from monoclonal antibody (mAb). A commercial
synthetic membrane that is known to be hydrophilized by hydrogel grafting was
employed to develop this polishing process that resulted in highly pure mAb, free from
aggregates and protein-A. This mAb polishing technique could easily be integrated with a
hydrophobic interaction membrane chromatography based mAb purification process.
iv
A paper-hydrogel composite membrane was developed as an inexpensive alternative to
commercial synthetic membranes used for carrying hydrophobic interaction membrane
chromatography. Poly(N-vinylcaprolactam) or PVCL hydrogel was coated on Whatman
filter paper to prepare these membranes. These environment responsive membrane which
responded to changes in salt concentration, gave excellent fractionation of multicomponent protein mixtures. As case study, a mixture of immunoglobulin G, human
serum albumin and insulin was fractionated.
A technique for modifying the surface of synthetic hollow fiber membranes with alginate
(a natural hydrogel) was developed. This manner of surface modification led to the
improvement in membrane mass transport. The alginate was cross-linked on the outer
surface of the membrane by diffusion of the cross-linker (calcium ions) through the
membrane pores. The calcium alginate coating layer was characterized by optical and
transmission electron microscopy, contact angle measurement, hydraulic permeability
measurement and by examining solute transport.
Hollow and solid calcium alginate fibers were fabricated using a novel hollow fiber
membrane based moulding technique. The pore present on the hollow fiber membrane
served as the reservoir for the calcium chloride solution with cross-linked the alginate
within the lumen. The calcium alginate fibers produced were characterized by optical,
transmission electron, and scanning electron microscopy. Cell immobilization
experiments were carried out to demonstrate biocompatibility and potential for tissue
engineering applications.
v
Acknowledgements
I would like to express my sincere gratitude to all those who supported my research
leading to this thesis.
First, I would like to express my deepest gratitude to my research supervisor, Dr. Raja
Ghosh, he has been a tremendous mentor for me. I would like to thank him for
encouraging my research projects. This led me to grow more creative and developed my
thinking about the project. His advice on both research, as well as on my career have
been priceless. Also, I appreciate his efforts to give me the opportunities to experience
worldwide conferences; it made me understand the relationship between research and
industry. Without his guidance and support this dissertation would not be possible.
I would also like to acknowledge my supervisory committee membranes, Dr. James
Dickson, Dr. Ravi Selvaganapathy for their valuable advice and encouragement by
participating in my annual committee meetings. Dr. Todd Hoare is also acknowledged for
kindly allowing me access to his facilities.
I would like to acknowledge the funding support from Natural Sciences and Engineering
Research Council of Canada, mAbnet research funding. I also thank my personal
scholarship:
International Excellence Awards, and best TA awards. Also, the
ambassadorial scholarship from Rotary international helped me to explore a new country
with fantastic Rotarians.
vi
Dennis, Lu, Mukesh and Zheng taught me the proper use of experimental machinery and
guided my grad student life. Bioseparation research group, Annabel, Evan, Si, Rahul, and
Pedram, for their warm hearts and always providing a happy environment in the lab.
Special thanks to summer students, Ryan Sungho, Steven, Simon, who made for a fun
and dynamic environment while we were doing research. My dear colleagues, Scott, Matt,
Emilia, Daryl, Chris, Lina, and Wein whom I thank for sharing their knowledge of
different research areas and their friendship. Moreover, I would like to thank my dearest
and sweetest friends Meng, Yesol, and Soomin for their support throughout my education.
Dr. JungBok Lee who educated me with his biology backgrounds and cell culture practise
for the last two years without any reward.
I would like to give my thanks to the staff in Chemical Engineering Department, Kathy
Goodram, Lynn Falkiner, Melissa Vasil, and Cathie Roberts for their assistance. Special
acknowledge to Paul Gatt for fabricating various custom designed membrane modules.
Finally, I would like to express my deep gratitude to my family, mom, dad, and my
younger brother for all of the sacrifices that you have made on my behalf. Your prayer for
me was what sustained me thus far. And my Canadian family, Kathi and Ray Smith, for
without their support and belief in me, I could not have even started my graduate studies
in Canada.
vii
Table of Contents
Chapter 1. Introduction of membranes with hydrogels....................................................... 1
1.1 Membrane technology .............................................................................................. 1
1.2 Membrane chromatography ...................................................................................... 3
1.3 Hydrogel based membrane...................................................................................... 10
1.4 Thesis outline .......................................................................................................... 12
1.5 References ............................................................................................................... 16
Chapter 2. Simultaneous removal of leached protein-A and aggregates from monoclonal
antibody using hydrophobic interaction membrane chromatography............................... 20
2.1 Abstract ................................................................................................................... 21
2.2 Introduction ............................................................................................................. 22
2.3 Materials and Methods ............................................................................................ 27
2.3.1 Materials .......................................................................................................... 27
2.3.2 Methods............................................................................................................ 28
2.4 Results and Discussion ............................................................................................ 29
2.5 Conclusion............................................................................................................... 41
2.6 References ............................................................................................................... 42
Chapter 3. High-resolution hydrophobic interaction chromatographic separation of multiprotein mixture using paper-based composite membranes .............................................. 47
viii
3.1 Abstract ................................................................................................................... 48
3.2 Introduction ............................................................................................................. 49
3.3 Materials and Methods ............................................................................................ 55
3.3.1 Materials .......................................................................................................... 55
3.3.2 Separation of protein mixture with HIC butyl column .................................... 56
3.3.3 Paper based composite membrane preparation ................................................ 56
3.3.4 Membrane chromatography set-up .................................................................. 57
3.3.5 Hydraulic testing of membranes ...................................................................... 58
3.3.6 SDS-PAGE ...................................................................................................... 58
3.3.7 SEC-HPLC ....................................................................................................... 58
3.4 Results and Discussions .......................................................................................... 59
3.4.1 Optimization of protein separation with resin based hydrophobic interaction
chromatography ........................................................................................................ 59
3.4.2 Composite paper based membrane characterizations (salt responsive and
protein binding/elution test) ...................................................................................... 63
3.4.3 Process optimization (elution mode) ............................................................... 67
3.5 Conclusions ............................................................................................................. 75
3.6 References ............................................................................................................... 77
Chapter 4. Hollow fiber membrane surface coating with natural hydrogel for bioreactor 81
ix
4.1 Abstract ................................................................................................................... 82
4.2 Introduction ............................................................................................................. 83
4.3 Materials and Methods ............................................................................................ 87
4.3.1 Materials ........................................................................................................... 87
4.3.2 Methods............................................................................................................ 88
4.4 Results and Discussions .......................................................................................... 92
4.5 Conclusions ........................................................................................................... 105
Chapter 5. Fabrication of alginate fibers using a microporous membrane based molding
technique ........................................................................................................................ 111
5.1 Abstract ................................................................................................................. 112
5.2 Introduction ........................................................................................................... 113
5.3 Materials and Methods .......................................................................................... 118
5.3.1 Fabrication of calcium alginate hollow fibers ............................................... 119
5.3.2 Microscopy .................................................................................................... 121
5.3.3 Human umbilical vein endothelial cell culture .............................................. 122
5.4 Results and Discussions ........................................................................................ 123
5.5 Conclusion............................................................................................................. 135
Chapter 6 Contributions and recommendations .............................................................. 140
6.1 Membrane chromatography .....................................Error! Bookmark not defined.
x
6.2 Paper based membrane chromatography .................Error! Bookmark not defined.
6.3 Hollow fiber modification with natural hydrogel ....Error! Bookmark not defined.
6.4 Calcium alginate fiber fabrication using hollow fiber membrane Error! Bookmark
not defined.
xi
List of Figures
Figures in chapters 2, 3, 4 and 5 are as they appear in published paper, submitted
manuscript, or manuscript with pending submission.
Chapter 1
Figure 1.1
The comparison of mass transport and resolution in bead based column
chromatography versus membrane chromatography *adopted from [1]
Figure 1.2
The summary of individual research project. A) mAb polishing with
PVDF membrane chromatography B) multi-proteins separation with paper based
composite membrane C) calcium alginate coated on hollow fiber membrane
extracapillary surface D) calcium alginate fiber fabrication with hollow fiber membrane
molding technique
Chapter 2
Figure 1
Schematic diagram for monoclonal antibody polishing based on
hydrophobic interaction membrane chromatography
Figure 2
Effect of salt concentration on the binding of hIgG1-CD4 on
hydrophilized PVDF membrane (eluting buffer: 20 mM sodium phosphate, pH 7.0;
binding buffer: eluting buffer + different concentrations of ammonium sulfate (1) 1.2 M,
(2) 1.3 M, (3) 1.4 M and, (4) 1.5 M; hIgG-CD 4 concentration: 0.1 mg/mL, injection
volume: 500 μL, flow rate: 2 mL/min)
xii
Figure 3
Effect of salt concentration on the binding of hIgG1-CD4 “dimers” on
hydrophilized PVDF membrane (eluting buffer: 20 mM sodium phosphate, pH 7.0;
binding buffer: eluting buffer + different concentrations of ammonium sulfate (1) 1.5 M,
(2) 1.4 M, (3) 1.3 M and, (4) 1.2 M; hIgG1-CD 4 “dimers” concentration: 0.1 mg/mL,
injection volume: 500 μL, flow rate: 2 mL/min)
Figure 4
Effect of salt concentration on the binding of protein-A on hydrophilized
PVDF membrane (eluting buffer: 20 mM sodium phosphate, pH 7; binding buffer:
eluting buffer + different concentrations of ammonium sulfate (1) 1.5 M, (2) 3 M;
protein-A concentration: 0.1 mg/mL; injection volume: 500 μL, flow rate: 2 mL/min)
Figure 5
Effect of protein A amount on HIMC chromatograms obtained using mAb
– protein-A mixtures (membrane: hydrophilized PVDF; eluting buffer: 20 mM sodium
phosphate, pH 7.0; binding buffer: eluting buffer + 1.37 M ammonium sulfate;
chromatogram (1): 0.002 mg protein A + 0.035 mg mAb; chromatogram (2): 0.0015 mg
protein A + 0.035 mg mAb; chromatogram (3): 0.001 mg protein A + 0.035 mg mAb;
chromatogram (4): 0.0005 mg protein A + 0.035 mg mAb; injection volume: 500 μL,
flow rate: 2 mL/min)
Figure 6
Polishing of mAb by hydrophobic interaction membrane chromatography
(membrane: hydrophilized PVDF; eluting buffer: 20 mM sodium phosphate, pH 7.0;
binding buffer: eluting buffer + 1.37 M ammonium sulfate; feed: 0.04 mg mAb monomer
+ 0.005 mg “dimers” + 0.0025 mg protein-A; injection volume: 500 μL, flow rate: 2
mL/min)
xiii
Figure 7
Size exclusion chromatography of samples obtained from the HIMC based
mAb polishing experiment (1: pure mAb monomer, 2: “dimers”, 3: simulated feed, 4:
flow through, 5: eluate; column: Ultrahydrogel SEC column, mobile phase: 0.25 M
sodium chloride solution in water, flow rate: 0.3 mL/min)
Figure 8
Coomassie blue stained SDS-PAGE gel (10%, reducing) obtained with
samples from the HIMC based mAb polishing experiment (M: marker; lane 1: flow
through, lane 2: eluate, lane 3: simulated feed; lane 4: pure protein-A; lane 5: “dimers”;
lane 6: mAb monomer)
Chapter 3
Figure 1
Effect of salt concentration on the binding of three proteins (hIgG, HSA,
and Insulin) on butyl Sepharose TM HIC (eluting buffer: 20mM sodium phosphate, pH 7;
binding buffer: eluting buffer+ different concentrations of ammonium sulphate – (1) 1.3
M, (2) 1.5 M; model protein concentration: hIgG 0.21 mg+ HSA 0.35 mg+ Insulin 0.14
mg in 500μL injection volume)
Figure 2
Coomassie blue stained SDS-PAGE gel (10%, non-reducing) obtained
with samples from the butyl Sepharose TM HIC. Multi-protein mixture (M: marker; lane
1: hIgG, lane 2: HSA, lane 3: insulin, lane 4: feed –three protein mixture, lane 5: flow
through-peak a, lane 6: first elute-peak b, lane 7: second elute-peak c)
Figure 3
Pressure profile in buffer change experiment carried out using filter paper
based composite membranes (flow rate: 1.0 mL/min; 1.5 M ammonium sulphate in 20
xiv
mM sodium phosphate buffer, 3M of sodium chloride in 20 mM sodium phosphate buffer,
pH 7.0)
Figure 4
Protein binding effect test using composite membrane, hIgG 0.5mg, and
insulin 0.3 mg in 500 μL injection loop(flow rate: 1.0 mL/min; 0.6M and 0.8M of
ammonium sulphate in 20mM sodium phosphate buffer, pH 7.0)
Figure 5
Effect of salt concentration on the binding of three model proteins on
PVCL composite membrane with 20 mL of gradient (eluting buffer: 20 mm sodium
phosphate, pH 7.0; binding suffer: eluting buffer + 1.3 M ammonium sulphate, injected
proteins concentration: IgG 0.21 mg+ HSA 0.35 mg+ Insulin 0.14 mg in 500μL injection
volume, total protein loading 0.7 mg/0.25 mL of bed volume, flow rate: 1.0 mL/min)
Figure 6
Size exclusion chromatography of samples obtained from the HIMC based
multi-proteins separation a) pure standard samples 1: hIgG, 2: HSA, 3: Insulin, 4: feed –
(hIgG+ HSA+ Insulin), b) fractions from figure 5, 1: flow through, 2: first eluting peak 3:
second eluting peak; column: TSKgel G3000SWXL, mobile phase: 0.25 M sodium
chloride solution in 20 mM of sodium phosphate, flow rate: 0.3 mL/min)
Figure 7
Coomassie blue stained SDS-PAGE gel (10%, reducing) obtained with
samples from the HIMC based multi-protein mixture (M: marker; lane 1: hIgG, lane 2:
HSA, lane 3: insulin, lane 4: feed –three protein mixture, lane 5: flow through, lane 6:
first elute peak, lane 7: second elute peak
xv
Figure 8
Effect of salt concentration on the binding of three model proteins on
pVCLM composite membrane with step change 0-70-100 % (eluting buffer: 20 mm
sodium phosphate, pH 7.0; binding suffer: eluting buffer + 1.3 M ammonium sulphate,
injected proteins concentration: IgG 0.21 mg+ HSA 0.35 mg+ Insulin 0.14 mg in 500μL
injection volume, flow rate: 1.0 mL/min)
Figure 9
Size exclusion chromatography of samples obtained from the HIMC based
multi-proteins separation a) pure standard samples 1: hIgG, 2: HSA, 3: Insulin, 4: feed –
(hIgG+ HSA+ Insulin), b) fractions from figure 8, 1: flow through, 2: first elute peak, 3:
second elute peak, column: TSKgel G3000SWXL, mobile phase: 0.25 M sodium chloride
solution in 20 mM of sodium phosphate, flow rate: 0.3 mL/min)
Figure 10
Coomassie blue stained SDS-PAGE gel (10%, reducing) obtained with
samples from Figure 8 (M: marker; lane 1: hIgG, lane 2: HSA, lane 3: insulin, lane 4:
feed –three protein mixture, lane 5: flow through, lane 6: first elute peak, lane 7: second
elute peak)
Chapter 4
Figure 1
The method for fabricating calcium alginate coating on ultrafiltration
hollow fiber membrane as a substrate. A) cross section view of hollow fiber membrane
lumen filled with water or 50 v/v% ethanol for pre-wetting step, B) after removal of
water or 50 v/v% ethanol, introduce of sodium alginate solution in shell side at the mean
xvi
while calcium chloride solution enter the membrane lumen, C) cross-linking of alginate
radially outward diffusion of calcium ions
Figure 2
Experimental set-up for membrane characterization on hydraulic
permeability at constant pressure, single PES hollow fiber membrane with effective
length 30 cm, membrane module with effective inner volume 23.57 mL, and data
acquisition (pressure and mass per time) with Logger pro.
Figure 3
Experimental set-up for DMSO (300 g/L) diffusion experiment, PES
ultrafiltration hollow fiber membrane with 30 cm effective length, Extracapillary space
effective volume 5.89 mL, the pressure data collected and analyzed in Logger Pro (flow
rate: 0.00866ml/min, pressure 101 kPa )
Figure 4
Magnified side view images of the uncoated/coated PES hollow fiber
membrane. Images were obtained by light microscopy with a magnification factor of 25.
Coating conditions: 1 wt% Alginate concentration, 0.1 M calcium chloride concentration
in 50 v/v% ethanol, and 40 min cross-linking time
Figure 5
Fabrication of calcium alginate coated hollow fiber membrane
controllability test. Fabrication conditions: 1wt% alginate concentration, 0.1M of calcium
chloride concentration, varied cross-linking time 0, 5, 10, 15, 20, and 40 min.
Figure 6
Hollow fiber membrane coating with calcium alginate hydrogel A) Dew:
calcium chloride solution pass through membrane pores B) 5 min cross-linking process C)
xvii
15 min cross-linking time. Fabrication condition: sodium alginate 1 wt%, calcium
chloride 0.1 M in water, PES hollow fiber membrane 150 kDa MWCO.
Figure 7
TEM image obtained with the hollow fiber and composite membrane
fabricated with 1 wt% sodium alginate and 0.1 M calcium chloride on PES ultrafiltration
membrane, with a cross-linking time of 30 min. (A) control sample: uncoated
membrane,25000 X magnification (B) 1 wt % of sodium alginate and 0.1 M calcium
alginate in water, 5000 X magnification (C) bigger magnification of sample B, 20000X
(D) 0.1M of calcium chloride solution in 50 v/v EtOH %, 20000 X magnification
Figure 8
Calcium alginate polymerization study with different calcium chloride
solvents (A: alginate + water based calcium chloride, B: alginate + 50v/v% ethanol based
calcium chloride, C: alginate + 100 % ethanol, D: ethanol + alginate, E: sample D vertical
pose), 1 wt% sodium alginate of 1mL, 0.1 M calcium chloride 300 μL
Figure 9
Fabrication of calcium alginate coated on PES hollow fiber membrane
(Alginate concentration 1wt%, calcium chloride concentration 0.1M and 1 min crosslinking time) optical microscope 25 times.
Figure 10
Water contact angle analysis with PES hollow fiber membrane (left), and
calcium alginate coated membrane (right). Picture obtained by digital camera.
Figure 11
Measurement of water flux and hydraulic permeability of uncoated/coated
PES hollow fiber membrane, operating pressure at 109.45 kPa (uncoated membrane) and
110.56 kPa (coated membrane) with constant.
xviii
Figure 12
Effect of coating hollow fiber membrane on DMSO mass transport with
Fig 3 experimental set up.
Chapter 5
Figure 1
The method for fabricating calcium alginate hollow fibers using porous
hollow fiber membrane as mold (A) cross sectional view of hollow fiber membrane with
pores filled with calcium chloride solution, (B) sodium alginate solution being filled
inside the lumen of the hollow fiber membrane, (C) cross-linking of alginate by radially
inward diffusion of calcium ions, (D) removal of un-cross-linked alginate core with
calcium chloride solution, and (E) expulsion of calcium alginate hollow fiber from mold.
Figure 2
Sequence of events in calcium alginate hollow fiber molding process (A)
sweating of calcium chloride on the outer wall of the hollow fiber membrane indicating
that the pores are filled, (B) removal of un-cross-linked sodium alginate core from the
mold using calcium chloride solution, (C) expulsion of calcium alginate hollow fiber
from mold using water, (D) alginate hollow fiber after recovery from mold.
Figure 3
Alginate fibers fabricated using hollow fiber membrane based molding
technique: (A) Fibers made with 2.5% sodium alginate, 0.1 M calcium chloride solution
as cross-linker using (from the top) Accurel Q3/2, Tetronic and Accurel PP S6/2 hollow
fiber membrane molds respectively. The cross-linking times were 10, 30 and 60 seconds
respectively for the small, medium and large diameter hollow fibers. (B)Solid and (C)
hollow fibers fabricated with 5% sodium alginate, 1 M calcium chloride solution as
xix
cross-linker, using Accurel PP S6/2 membrane as mold. The cross-linking times were 600
seconds and 60 seconds respectively.
Figure 4
Magnified cross-sectional and side view images of the alginate hollow
fiber shown in Figure 3 C. Images were obtained with the fiber immersed in water by
light microscopy with a magnification factor of 25.
Figure 5
Confocal laser scanning (CLS) microscopy images of the FITC-dextran -
alginate composite hollow fiber, (A) z-stack images obtained along the axis at sequential
depths of 200μm, (B) five times magnification, and (C) twenty-time magnification
Figure 6
Scanning electron microscopy (SEM) images of calcium alginate hollow
fiber (400X magnification) fabricated with 2.5 % sodium alginate and 1 M calcium
chloride, using Accurel PP S6/2 membrane as mold, with a cross-linking time of 60
seconds.
Figure 7
TEM image obtained with the hollow fiber described in Figure 6 at (A)
5000X magnification, and (B) 75000X magnification
Figure 8
Effect of sodium alginate concentration on dimension of alginate fiber
(mold: Accurel PP S6/2, calcium chloride concentration: 1 M; cross-linking time: 60
seconds)
xx
Figure 9
Effect of calcium chloride concentration on dimension of alginate fiber
(mold: Accurel PP S6/2, sodium alginate concentration: 5%, cross-linking time: 60
seconds)
Figure 10
Effect of cross-linking time on dimensions of alginate fiber (mold:
Accurel PP S6/2, sodium alginate concentration: 5%, calcium chloride concentration: 1
M)
Figure11
Light microscopy image of HUVEC embedded alginate hollow fiber
(image obtained one day after fabrication, magnification: 100 X, sodium alginate
concentration: 1%, mold: Tetronic, calcium chloride concentration: 0.1 M, cross-linking
time: 30 seconds)
List of Tables
Tables in chapter 2, 3, 4 and 5 are as they appear in published paper, submitted
manuscript, or manuscript with pending submission.
Chapter 1
Table 1. Membrane specifications in this dissertation
Chapter 2
Table 1. Different combinations of protein-A and mAb used in HIMC experiments
represented in Fig. 5.
xxi
Chapter 4
Table 1. PES hollow fiber membrane coated with calcium alginate; 0.1 M calcium
chloride solution was prepared in 50 v/v% ethanol, 1 wt % sodium alginate, in different
cross-linking time
Nomenclature
E1
First eluted peak
E2
Second eluted peak
FT
Flow through
HSA
Human serum albumin
HIC
hydrophobic interaction chromatography
HIMC
hydrophobic interaction membrane chromatography
HPLC
High pressure liquid chromatography
kDa
kilo Dalton
LCST
Lower critical solution temperature
MWCO
Molecular weight cut-off
PEG
Poly(ethylene glycol)
xxii
PVCL
Poly(N-vinylcaprolactam)
PVDF
Polyvinylidene fluoride
SDS-PAGE
Sodium dodecyl sulfate polyacrylamide gel electrophoresis
SEC
Size exclusion chromatography
T
Temperature (°C)
P
Pressure (kPa)
xxiii
Preface
This Ph.D. thesis is organized in a sandwich style based on the following published,
submitted and prepared articles:
1. S.M. Yoo, R. Ghosh, Simultaneous removal of leached protein-A and aggregates from
monoclonal antibody using hydrophobic interaction membrane chromatography,
J.Membr. Sci, 390-391 (2012) 263-269. This article was made into Chapter 2.
2. S.M. Yoo, R. S.H. Park, R. Ghosh, High-resolution hydrophobic interaction
chromatographic separation of multi-protein mixture using paper-based composite
membranes. Manuscript is in preparation. This article was made into Chapter 3.
3. S.M. Yoo, R. Ghosh, Hollow fiber membrane surface coating with natural hydrogel for
bioreactor, Manuscript in preparation. This article was made into Chapter 4.
4. S.M. Yoo, R. Ghosh, Fabrication of alginate fibers using a microporous membrane
based molding technique, in submitted in Biochem. Eng. J. Manuscript # BEJ-D-1400034R1. This article was made into Chapter 5.
All the other articles were prepared by Seung Mi Yoo. Dr. Ghosh provided guidance in
research direction and project planning, and also helped in manuscript preparation. Ryan
Sung Ho Park conducted initial experiments, membrane characterization and protein
binding, as a project student for the project described in Chapter 3. The work reported in
this dissertation was undertaken from January 2013 to April 2014.
xxiv
Chapter 1. Introduction of membranes with hydrogels
1.1 Membrane technology
Membrane technology has experienced tremendous application in chemical and
biopharmaceutical engineering industries. The first significant industrial application of
membrane technology took place in 1960s, and ever since then it has contributed
significantly to improvement in quality of life [1]. There are various definitions for the
term ‘membrane’ depending on the application it is being used, for example simply
filtering particles or selectively transporting one or more substances. For a biochemical
engineer who has studied downstream processing, a membrane is a physical separation
barrier with porous structure which requires either pressure or another driving force to
selectively separate materials. A membrane is generally defined as “a thin barrier or film
through which solvent and solutes are selectively transported” [2]. In the
biopharmaceutical industry membranes have been used for protein concentration,
microorganism and virus removal, buffer exchange, and cell recovery [3]. A much
simpler mechanism is utilized in the water treatment industry, where dispersed particles
are removed by size exclusion or sieving [4]. There are many membrane manufactures
involved in the production of different types of membrane elements and configurations
that are used for various applications. The major three international membrane companies
whose products have been used in my research are Merck Millipore, and Membrana
1
GmbH. Microfiltration and ultrafiltration membranes from these companies that have
been used in my projects are made of organic polymers such as hydrophilized
polyvinylidene fluoride (PVDF) (Chapter 2), cellulose (Chapter 3), polyethersulfone
(PES) (Chapter 4 and 5), polypropylene (PP) (Chapter 5), polysulfone (PS), cellulose
acetate (CA), and regenerated cellulose (see Table 1). Membranes are generally available
in three geometries: flat sheet (Chapter 2 and 3), hollow fiber (Chapter 4 and 5), and
tubular (not discussed in this thesis). The advantage of a flat sheet membrane is that it can
be utilized for various types of operations in different module configurations. For
example, a flat sheet membrane could be cut into a desired shape (square, circular, or
rectangular) and size to fit a module for dead-end, cross flow, or co/counter current flow
separation, whereas hollow fiber and tubular membrane are available pre-assembled
within membrane modules.
Membranes are also categorized based on their pore sizes into micro-, ultra-, nanofiltrations, and reverse osmosis. Micro- and ultra- filtration are most extensively used for
waste water treatment and for biopharmaceutical production. Nanofiltration and reverse
osmosis are used in clinical setting such as viral removal or plasma-derived product
separation [43] and drinking water purification units. Commercial microfiltration
membranes commonly have 0.45, 0.2, 0.1 μm pore sizes which are able to reject
microorganism from feed solutions. Using this rejection characteristic, buffers and other
solutions can be clarified. Ultrafiltration is mainly employed for protein concentration
2
and buffer exchanges prior to analysis. Table 1 shows the specifications of the membrane
that were used in the different projects discussed in this thesis.
Table 1 Membrane specifications in this dissertation
Flat
sheet
Hollow
fiber
Material
Catalogue
information
Modification
method/base
material
Thickness
Manufacture
Morphology
Hydrophilized
PDVF
GVWP
14250
Grafted
125 µm
Millipore
asymmetric
Composite
membrane
Fabricated
Coating PVCL
on filter paper
200 µm
Fabricated
symmetric
Polypropylene
Accurel PP
S6/2
Polypropylene
Membrana
Isotropic
Accurel PP
Q3/2
Polypropylene
Membrana
Isotropic
Tetronic
Hydrophilized
PES
1.8 mm
i.d./2 mm
o.d.
0.6 mm
i.d./1mm
o.d.
0.8 mm
i.d./ 1.3
mm o.d.
Hydranautics
Anisotropic
Polyethersulfone
The membrane industry has been producing high quality membrane for more than 50
years. However, while some product lines have remained the same, the applications and
operating conditions required better mechanical/ chemical durability, high flux, less
fouling, and easy to operation. To solve these problems, many researchers and
membrane developers are still seeking for new types of membrane such as hydrogel
based composite membrane, which will be discussed in considerable detail in this thesis.
1.2 Membrane chromatography
Various bioseparation techniques are used in the downstream processes of the
biopharmaceutical industry. Column chromatography has been used extensively in
3
production lines for the separation and purification of biological samples with high
resolution. For example, various types of column chromatography are used in the capture
(affinity chromatography), purification (hydrophobic interaction chromatography), and
polishing (ion exchange chromatography) steps of antibody purification [5]. However,
because the separation relies mainly on the diffusion of samples through pores in a soft
gel matrix, the procedure requires long contact times with large volumes of mobile phase.
In addition, when this soft material is placed in a column, the drop in pressure across the
column results in the deformation of the beads. These disadvantages limit the scalability
of this technique. Therefore, studies have focused on advancing the technique by
improving the diversity of the support matrix, narrow pore distribution, and ligand
chemistry. Although new modifications in the method, such as combining different kinds
of column chromatography for more complex protein separations have been investigated,
the operating conditions have a narrow window, and possible target proteins are very
limited. A simulated moving bed (SMB) technique for multi-protein separations was
developed in the early 2000s [6]. However, this technique has limited applications. The
operating condition requires the ratio of the individual protein or amino acid residence
time versus the zone flow rate to be calculated and estimated from many parameters.
Moreover, it requires tedious optimization procedures for particular target proteins.
Membrane chromatography provides an attractive and efficient mass transport process,
driven predominantly by the principle of convection flow, as illustrated in Figure 1.1. The
membrane adsorbers that are used in this technique possess high-throughput abilities.
4
These membrane stacks create void channels for the mobile phase to pass, allowing
selective separation of biological substances. Membrane chromatography incorporates
aspects of membrane filtration (high-throughput) and packed-bed column
chromatography (high-resolution) [7, 8]. The combination of membrane and column
chromatography was developed to achieve better separation with smaller footprints,
shorter process times, less buffer usage, and more flexibility in terms of configuration and
operation. Furthermore, the limitations associated with conventional column
chromatography, such as sudden drops in pressures, may be minimized. The relatively
rigid microporous structure of the membrane allows for a high flux and minimal batch-tobatch variations [2].
Figure 1.1 Comparison of the mass transport process and resolution in bead-based
column versus membrane chromatography. *adopted from [9].
5
Membrane chromatography operates on the same principle as column chromatography.
The affinity moiety is immobilized on the membrane substrate and serves as the
stationary phase. The mobile phase (protein mixture or biological sample) is then passed
over the stationary phase. The sample interacts with the stationary phase and binds to the
membrane surface, while the impurities remain in the flow-through. The sample is then
eluted from the surface of the membrane by changing the operating conditions such that
the interaction between the sample and the affinity moiety is broken. These membranes
may be used as alternatives to protein-A/G affinity column chromatography for the
purification of antibodies [10, 11]. In addition, ion exchange membrane chromatography
may be used to remove impurities such as host cell protein (HCP), DNA, and viruses
from biological samples [12, 13].
A number of recent studies have demonstrated the use of hydrophobic interaction
membrane chromatography (HIMC) [14–17] for the purification of monoclonal
antibodies (mAbs) [18–26]. Ghosh (2001) used a polyvinylidene fluoride (PVDF)
membrane to separate CAMPATH-1G monoclonal antibodies and bovine serum albumin
(BSA) using hydrophobic interactions [18]. Similarly, human plasma proteins, human
immunoglobulin G (hIgG) and human serum albumin (has) could be separated using a
PVDF membrane (pore size, 1 μm) [27]. A paper-polyethylene glycol (PEG)
interpenetrating polymer network-based composite membrane was also developed as an
inexpensive alternative to commercial membranes [22]. The PEG polymer undergoes a
phase transition when treated with high concentrations an anti-chaotropic salt; this
6
characteristic feature of the polymer was adopted to separate and purify mAb on the basis
of its hydrophobic interaction with the PEG polymer.
Membranes have not only been examined for their adsorption capacity and resolution but
also for their binding mechanisms, which were investigated using rabbi IgG and proteinA conjugates [21]. The binding sites of these conjugates on the membrane surface were
identified by digesting them with papain and pepsin. Because protein-A recognizes the Fc
domain of rabbit IgG, the Fab domain and the hinge region of IgG are available for
interaction with the membrane. This binding mechanism was analyzed using a confocal
laser-scanning microscope.
In their studies, Huang et al. (2009) coated a PVDF membrane with an environmentally
responsive polymer (poly-N-vinylcaprolactam [PVCL]) for use in HIMC [24]. Pereira et
al. (2010) demonstrated that a modified Sartobind® epoxy membrane supported with a
linear alkyl chain ligand could be used for plasmid DNA purification [23]. The alkyl
ligand layer increased the adsorption of RNA on the membrane, as expected. Other
studies investigated the use of composite membranes such as paper-based membranes for
the purification of mAb [25]. Environmentally responsive PVCL was grafted on
regenerated cellulose membranes using the atom-transfer radical-polymerization (ATRP)
technique; the binding capacity per column volume for IgG and HSA were then analyzed
[25, 26]. Two parameters of composite membranes may be manipulated by environment
stimuli: the range of binding capacity and the permeability.
7
Membrane chromatography is widely used commercially for the purification and
fractionation of biological samples in the laboratory as well as the pilot-scale. Mustang®
ion exchange membrane (Pall Life Sciences), Sartobind® Phenyl, and Sartobind® S/Q
(Sartorius AG) are available commercially. Mustang® Q or S (or Sartobind® S or Q) are
based on the principles of ion exchange chromatography. Similarly, Sartorius also
manufactures an HIMC membrane composed of regenerated cellulose with covalently
attached phenyl ligands, called Sartobind® Phenyl [28]. This commercial product is prestacked within the cartridge for convenience; however, certain aspects such as fluid
distribution within the membrane module and scale-up with respect to bed volume need
to be improved. Moreover, commercially available products are expensive to use,
whereas conventional chromatography columns may be regenerated and reused
approximately 100 times, depending on the chemistry involved.
Two operating modes are used in membrane chromatography: the flow-through mode and
the bind-elute mode. The flow-through mode has been embraced to remove trace amounts
of unwanted substances from mAb purifications [12, 20]. This mode has the advantage of
simple operating conditions, where the desired products are collected in the eluate and
trace amounts of impurities are entrapped within the membrane matrix. This can be
achieved without changing the mobile fluid conditions and does not require a membrane
regeneration step. The flow-through mode is excellent for use as the final polishing step
for biopharmaceuticals; however, the selectivity of the membrane has to be high enough
to separate the product from the impurities and the membrane quality and operating
8
conditions have to be precise. In mAb production, a protein-A affinity column is the main
capturing tool used to separate mAb from HCP, DNA, or viruses. The protein-A ligand
binds strongly to the Fc domain of mAbs; therefore, the capture step is the most crucial
step in mAb production. Despite the benefits associated with the use of protein-A column
process, the low pH conditions required for dissociation cause the mAbs to aggregate and
leach protein-A ligand. Therefore, an ion exchange membrane chromatography technique
was developed for the large-scale production of mAbs [12]. At neutral pH, the impurities
and mAbs have different charges (negative and positive, respectively). Therefore, anion
exchange membrane chromatography may be used to purify mAbs. Knudsen et al. (2001)
reported that cation and anion exchange membranes may be utilized for the removal of
potential contaminants and impurities. However, because cation exchange membrane
chromatography has to be operated under two operating steps (bind and elute), the
operating cost increases compared to anion exchange membrane chromatography
(available commercially as Sartobind® Q) [12]. The economic aspect of using membrane
chromatography was reviewed by Zhou and Tressel in 2006 [29]. Buffer usage may be
lowered by as much as 95% with membrane chromatography compared to packed-bed
anion exchange chromatography. The disposability of membrane columns eliminates the
packing, washing, and regeneration steps of column chromatography, thereby reducing
processing time and labor costs. A long-term cost calculation analysis of membrane
versus resin-based column chromatography revealed that membrane chromatography is
cheaper than column chromatography.
9
The bind-elute operating mode has been actively investigated for use in both academia
and industry. This operating mode requires the target substance to be bound on the
membrane, followed by elution in one or multiple steps or, alternatively, in a gradient.
However, membranes (in the absence of hydrogel-modifying treatments) have a lower
binding capacity than resin-based columns; therefore, improvements in binding capacity
were required. The most common application of the bind-elute mode is in affinity
membrane chromatography, which has replaced protein-A/G column chromatography for
the purification of mAb [10, 11]. The bind-elute mode has also been applied in HIMC for
the separation of proteins [18,19,21,22,25]. Recently, hydrogel coated/grafted membranes
are being actively investigated because they provide improved binding capacities. The
development of a hydrogel composite membrane is discussed in the next section.
1.3 Hydrogel based membrane
Synthetic membranes are used widely in the environmental and biopharmaceutical
industries. Membrane adsorbers, which are traditionally organic-based materials, have
been developed using the emerging hydrogel technology for enhanced performance and
functionality. Two types of hydrogels have been studied for use in hydrophobic
interaction, ion exchange, and affinity membrane chromatography. One is the synthetic,
polymer-based hydrogel, which may be composed of PEG, poly (N-isopropylacrylamide)
(PNIPAM), PVCL, polyester, poly lactic acid (PLA), poly lactic-co-glycolic acid
(PLGA), and polycarbonate. The other type comprises naturally derived hydrogels, such
10
as those based on dextran, alginate, collagen, fibrin, and gelatin [30]. Hydrogels are
cross-linked water-soluble polymers with hydrophilic pendent groups such as amides and
carboxylic acids in their structure. The combination of the hydrophobic membrane
substrate and the hydrophilic hydrogel resulted in the development of environmentally
responsive membranes, also termed “smart-membranes.”
Various environmentally responsive membranes have been developed by many
researchers [14, 21, 24, 26, 31–33]. Such membranes change their physical structure,
charge, or affinity in response to changes in pH [34], temperature [33], and salt
concentrations [32]. By manipulating the tunable characteristics of an environmentally
responsive membrane, complex or multi-component separations may be achieved easily.
PNIPAM has been actively researched for cross-linking or interpenetrating membrane
substrates to make a three dimensional formation [33, 35, 36]. This hydrogel network can
change its physical and chemical properties upon receiving the various environmental
stimuli mentioned above. Rao et al. (2002) constructed a hybrid membrane based on
silica substrate with PNIPAM encapsulated within it using a sol-gel process. The
permeability of the membrane changed with temperature. However, after PNIPAM
toxicity became problematic in biological applications, PVCL and PEG were investigated
as alternatives to PNIPAM [37, 38]. PVCL is also a thermo-responsive polymer with a
low critical solution temperature (LCST). It is relatively less toxic with enhanced
physical strength. It was studied extensively for the fabrication of composite membranes.
However, in terms of practical applications, controlling salt concentration is much
11
simpler than controlling the temperature of the mobile phase and its environment to
achieve switchable hydrophobic interactions between proteins and polymers [24].
Therefore, an inexpensive composite membrane with cellulose filter paper was developed
for protein separations. Yu et al. (2008) demonstrated the use of PEG-coated cellulose
paper for the separation of mAbs from Chinese hamster ovary (CHO) cell culture [22].
Mah and Ghosh (2010) demonstrated the feasibility of a composite membrane composed
of PVCL on Whatman filter paper for the purification of IgG from simulated CHO cell
culture. In addition to coating, hydrogels may be grafted on membrane substrates for
better membrane resolution [39–42]. Although the grafting method is more complicated,
involving the use of various chemicals (acids and bases) and long periods of pretreatment before polymerization, the ligand length and density may be tailored for exact
applications [15]. Himstedt et al. (2013) grew PVCL polymer chains on a regenerated
cellulose membrane. BSA and IgG, which are often used as model proteins, were
employed to test the adsorption as well as the binding capacity of the membrane [26].
1.4 Thesis outline
This thesis consists of six chapters. Chapter 1 is an introduction to membrane technology,
membrane modifications, and membrane chromatography which will also touch on all the
research projects in chapter 2, 3, 4, 5. Four chapters could be categorized into two key
compartments: flat membrane and hollow fiber membrane configurations. All of the
12
projects involved the combination of both hydrogels and membranes. Chapter 2 and
chapter 3 focus on membrane chromatography for various protein separation applications.
Chapter 4 and 5 are studied on membrane modification with natural hydrogel for tissue
engineering applications. The summary of each project is shown in Figure 1.2.
Figure 1.2
The summary of individual research project. A) mAb polishing with
PVDF membrane chromatography B) multi-proteins separation with paper based
composite membrane C) calcium alginate coated on hollow fiber membrane
extracapillary surface D) calcium alginate fiber fabrication with hollow fiber membrane
molding technique
Chapter 2 describes the purification of a monoclonal antibody from its aggregates and
leached protein-A using membrane chromatography. In this project, the membrane
chromatography consisted of a hydrophilized PVDF membrane which is commercially
available in the market as an ultrafiltration membrane. A PEG hydrogel, which had
environment responsiveness to salt present in solution, was grafted into the pores of the
hydrophilized PVDF membrane substrate [29]. This hydrogel, which was based on an
13
LCST polymer, had a phase transition controlled by salt concentration. With stimuli
responsiveness, apparent protein and grafted polymer hydrophobic interaction degree
changed. Manipulating the salt concentration made the pure monoclonal antibody
separate from impurities in flow-through mode. This work was published in Journal of
Membrane Science.
Chapter 3 presents the fractionation of plasma proteins, hIgG, and HSA, and also a
popular hormone, insulin, using a paper based composite membrane. A composite
membrane was fabricated using poly (N-vinylcaprolactam) polymer coated on Whatman
cellulose filter paper. Membrane chromatography was tested for environment responsive
characteristics under different ionic strength, and the transmembrane pressure was a good
indication on the hydrogel states. The collapsed and swollen states of hydrogel induced a
hydrophobic difference on membrane matrix. By manipulating this characteristic, protein
had experienced different interaction with the membrane. This work is ready for journal
submission.
Chapter 4 investigated a hollow fiber membrane extracapillary surface coated with
calcium alginate. The polymerization method is novel and the thickness of the coating
layer is highly controllable using a gentle and precise procedure. The cross-linkers are
passively diffused through the membrane pores and come into contact with the crosslinking alginate in extracapillary space. This composite membrane was studied for
membrane performance on hydraulic permeability and solute mass transport rate. The
hydrophilic characteristic was tested using water contact angle. The TEM analysis was
14
obtained for calcium alginate morphologies. This work also was written in manuscript for
journal submission.
Chapter 5 shows new application of micro- or ultra-filtration hollow fiber membrane. The
membrane pores were utilized for storage of cross-linking reagent, calcium ion, when
monomer (sodium alginate) solution contacted the polymerization process occurred. A
hollow fiber membrane is used as the mold for fabricating these calcium alginate fibers.
This process was used for membranes of varying diameters, sizes, and geometries. This
work has been submitted in Biochemical Engineering Journal.
As a whole, the research reported in this thesis could lead to four journal publications.
One is published (Yoo and Ghosh 2012, Journal of Membrane Science), one is submitted
(in Biochemical Engineering Journal, BEJ-D-14-00034), and two more papers are ready
submission to journals.
15
1.5
References
[1]
R.Baker, Overview of membrane science and technology, in: Membr. Technol.
Appl., 2nd ed., Wiley, England, 2004: pp1-14.
[2]
R. Ghosh, Membranes, in: Protein bioseparation using ultrafiltration, Imperial
College Press, London, UK, 2003: pp 19.
[3]
P. Gagnon, Purification tools for monoclonal antibodies by mixed-mode
chromatography, in: Process scale purification of antibodies, U. Gottschalk, ed.,
John Wiley & Sons, New York, 2009 pp 125-144.
[4]
J.C. Schrotter, B. Bozkaya-Schrotter, Current and emerging membrane processes
for water treatment, in: Membrane for water treatment, K.V. Penemann, S.P.
Nunes, John Wiley & Sons, Weinheim, 2010 pp 53-91.
[5]
B. Kelley, G. Blank, A. Lee, Downstream processing go monoclonal antibodies:
Current Practices and future opportunities. In: Process scale purification of
antibodies. John Wiley & Sons, Hoboken, NJ, 2009 pp1-23.
[6]
D. Wu, Y. Xie, Z. Ma, N.L. Wang, Design of simulated moving bed
chromatography for amino acid separations, 5885 (1998) 4023-4035.
[7]
J. Thommes, M.R. Kula, Membrane chromatography-an integrative concept in the
downstream processing of proteins, Biotechnol. Prog. 11 (1995) 359-367.
[8]
J.A. Gerstner, R. Hamilton, S.M. Cramer, Membrane chromatographic protein
separations systems for high-throughout, 596 (1992) 173-180.
[9]
R. Ghosh, Protein separation using membrane chromatography: opportunities and
challenges, J. Chromatogr. A. 952 (2002) 13-27.
[10] H. Zou, Q. Luo, D. Zhou, Affinity membrane chromatography for the analysis and
purification of proteins, J. Biochem. Biophys. Methods. 49 (2001) 199-240.
[11] L.R. Castilho, F.B. Anspach. W.D. Deckwer, Comparison of affinity membranes
for the purification of immunoglobulins, J. Membr. Sci. 207 (2002) 253-264.
[12] H.L. Knudsen, R.L. Fahrner, Y. Xu, L.A. Norling, G.S. Blank, Membrane ionexchange chromatography for process-scale antibody purification, 907 (2001) 145154.
16
[13] M. Phillips, J. Cormier, J. Ferrence, C. Dowd, R. Kiss, H. Lutz, J. Carter,
Performance of a membrane adsorber for trace impurity removal in biotechnology
manufacturing, J. Chromatogr. A. 1078 (2005) 74-82.
[14] J. Queiroz, C.T. Tomaz, J.M. Cabral, Hydrophobic interaction chromatography of
proteins, J. Biotechnol. 87 (2001) 143-159.
[15] A. Ladiwala, F. Xia, Q. Luo, C.M. Breneman, S.M. Cramer, Investigation of
protein retention and selectivity in HIC systems using quantitative structure
retention relationship models, Biotechnol. Bioeng. 93 (2006) 836–50.
[16] D.K. Roper, E.N. Lightfoot, Separation of biomolecules using adsorptive
membranes, J. Chromatogr. A. 702 (1995) 3–26.
[17] R. van Reis, A. Zydney, Membrane separations in biotechnology, Curr. Opin.
Biotechol. 12 (2001) 208-211.
[18] R. Ghosh, Separation of proteins using hydrophobic interaction membrane
chromatography, J. Chromatogr. A. 923 (2001) 59–64.
[19] R. Ghosh, L. Wang, Purification of humanized monoclonal antibody by
hydrophobic interaction membrane chromatography, J. Chromatogr. A. 1107
(2006) 104–9.
[20] S.M. Yoo, R. Ghosh, Simultaneous removal of leached protein-A and aggregates
from monoclonal antibody using hydrophobic interaction membrane
chromatography, J. Membr. Sci. 390-391 (2012) 263–269.
[21] X. Sun, D. Yu, R. Ghosh, Study of hydrophobic interaction based binding of
immunoglobulin G on synthetic membranes, J. Membr. Sci. 344 (2009) 165–171.
[22] D. Yu, X. Chen, R. Pelton, R. Ghosh, Paper-PEG-based membranes for
hydrophobic interaction chromatography: purification of monoclonal antibody,
Biotechnol. Bioeng. 99 (2008) 1434–42.
[23] L.R. Pereira, D.M.F. Prazeres, M. Mateus, Hydrophobic interaction membrane
chromatography for plasmid DNA purification: Design and optimization, J. Sep.
Sci. 33 (2010) 1175–84.
[24] R. Huang, K.Z. Mah, M. Malta, L.K. Kostanski, C.D.M. Filipe, R. Ghosh,
Chromatographic separation of proteins using hydrophobic membrane shielded
with an environment-responsive hydrogel, J. Membr. Sci. 345 (2009) 177–182.
17
[25] K.Z. Mah, R. Ghosh, Paper-based composite lyotropic salt-responsive membranes
for chromatographic separation of proteins, J. Membr. Sci. 360 (2010) 149–154.
[26] H.H. Himstedt, X. Qian, J.R. Weaver, S.R. Wickramasinghe, Responsive
membranes for hydrophobic interaction chromatography, J. Memb. Sci. 447 (2013)
335–344.
[27] R. Ghosh, Fractionation of human plasma proteins by hydrophobic interaction
membrane chromatography, J. Memb. Sci. 260 (2005) 112–118.
[28] N. Fraud, R. Faber, C. Kiss, Novel membrane adsorber for hydrophobic interaction
chromatography, in: Proceedings of the 236th ACS National Meeting, Philadelphia,
PA, USA, 2008.
[29] J.X. Zhou, T. Tressel, Basic concepts in Q membrane chromatography for largescale antibody production, Biotechnol. Prog. (2006) 341-349.
[30] K.Y. Lee, D.J. Mooney, Hydrogels for Tissue Engineering, Chem. Rev. 101 (2001)
1869–1880.
[31] R. Huang, L.K. Kostanski, C.D.M. Filipe, R. Ghosh, Environment-responsive
hydrogel-based ultrafiltration membranes for protein bioseparation, J. Memb. Sci.
336 (2009) 42–49.
[32] L.K. Kostanski, R. Huang, R. Ghosh, C.D.M. Filipe, Biocompatible poly(Nvinyllactam)-based materials with environmentally-responsive permeability, J.
Biomater. Sci. Polym. Ed. 19 (2008) 275–90.
[33] D, Crespy, M. Rossi, Temperature-responsive polymers with LCST in the
physiological range and their applications in textiles, Polym. Int. 1468 (2007)
1461-1468.
[34] N.M. Ranjha, J. Mudassir, Swelling and aspirin release study: cross-linked pHsensitive vinyl acetate-co-acrylic acid (VAC-co-AA) hydrogels, Drug Dev. Ind.
Pharm. 34 (2008) 512–521.
[35] R. Chennamsetty, I. Escobar, X. Xu, Characterization of commercial water
treatment membranes modified via ion beam irradiation, Desalination. 188 (2006)
203–212.
[36] G.V.R. Rao, M.E. Krug, S. Balamurugan, H. Xu, Q. Xu, G.P. Lo, Synthesis and
Characterization of Silica - Poly ( N -isopropylacrylamide ) Hybrid Membranes :
Switchable Molecular Filters, Chem. Mater. (2002) 5075–5080.
18
[37] H. Vihola, A. Laukkanen, L. Valtola, H. Tenhu, J. Hirvonen, Cytotoxicity of
thermosensitive polymers poly(N-isopropylacrylamide), poly(N-vinylcaprolactam)
and amphiphilically modified poly(N-vinylcaprolactam)., Biomaterials. 26 (2005)
3055–3064.
[38] J. Kopecek, Hydrogel biomaterials: a smart future?, Biomaterials. 28 (2007) 5185–
5192.
[39] Y.J. Shen, G.X. Wu, Y.B. Fan, H. Zhong, L.L. Wu, S.L. Zhang, X.H. Zhao, W.J.
Zhang, Performances of biological aerated filter employing hollow fiber
membrane segments of surface-improved poly (sulfone) as biofilm carriers, J.
Environ, Sci. (China) 19 (2007) 811-817.
[40] K. Asfardjani, Y. Secui, Y. Aurelle, N. Abidine, Effect of plasma treatments on
wettability of polysulfone and polyetherimide, 43 (1991) 271–281.
[41] D.S. Wavhal, E.R. Fisher, Membrane surface modification by plasma-induced
polymerization of acrylamide for improved surface properties and reduced protein
fouling, Langmuir, 19 (2003) 79-85.
[42] J. Chen, J. Li, C. Chen, Surface modification of polyvinylidene fluoride (PVDF)
membrane by low-temperature plasma with grafting styrene, Plasma Sci. Technol.
11 (2009) 42-47.
[43] T. Burnouf, M. Radosevich, Nanofiltration of plasma-derived biopharmaceutical
products, Haemophilia 9 (2003) 24-37
19
Chapter 2
Simultaneous removal of leached protein-A and aggregates
from monoclonal antibody using hydrophobic interaction
membrane chromatography1
1
This chapter is version of the paper published in Journal of Membrane Science, 390-391: 263-269 (2012)
by Seung Mi Yoo, Raja Ghosh.
20
2.1
Abstract
Protein-A affinity chromatography is the standard capture technique used for purification
of monoclonal antibodies (mAbs). A major problem with this technique is its inability to
remove monoclonal antibody aggregates from the monomeric form of the antibody.
Moreover, the elution of mAbs from a protein-A column is carried out using acidic
conditions which in turn could accelerate the formation of antibody aggregates and
causes leaching of protein-A from its supporting media. These impurities have to be
removed using appropriate separation methods before the mAb can be used for
therapeutic applications. There is currently a limited repertoire of polishing techniques
which simultaneously remove both mAb aggregates and protein-A. In this paper, we
describe a polishing method based on hydrophobic interaction membrane
chromatography for simultaneously removing these impurities. The flow-through mode
was used whereby monomeric mAb flowed through a membrane stack while impurities
were retained by reversible adsorption. These impurities were subsequently eluted by
lowering the salt concentration and the membrane stack was thus regenerated. The
operating conditions for this method were systematically optimized using pure mAb,
aggregates, protein-A, and mAb/protein-A complexes. The mechanisms by which
aggregates and leached protein-A were simultaneously removed are hypothesized.
21
2.2
Introduction
Monoclonal antibodies or mAbs were first produced in the 1970s [1]. These are now
extensively used in biomedical research, and for a range of therapeutic and diagnostic
applications [2]. MAbs are usually purified using either protein-A or protein-G based
affinity column chromatography as the principal capture technique [3, 4]. Protein-A,
which was originally isolated from a bacteria recognizes and binds the Fc domain of the
immunoglobulin G or IgG class mAbs at neutral pH [3]. The use of protein-A affinity
chromatography results in highly selective mAb purification and there is typically around
90 - 95% removal of impurities such as media proteins, host cell protein, nucleic acids
(mainly DNA) and endotoxin [5]. When the mAb containing feed solution is passed
through a protein-A column, the impurities are removed in the flow through while the
mAb is retained within the column. The column-bound mAb is then eluted by lowering
the pH, this acidic condition being suitable for inactivating viral contaminants. The mAb
yield in protein-A affinity chromatography tends to be quite high.
Despite these obvious advantages, the use of protein-A affinity chromatography for mAb
purification throws up some challenges that need to be addressed. Firstly, protein-A does
not discriminate between the mAb monomer and its aggregated forms as the Fc region is
not affected by the aggregation process [6, 7]. Consequently, the mAb aggregates that are
formed during cell culture and the earlier downstream processing steps such as
centrifugation or microfiltration are not removed in the flow through and are found to coelute from the protein-A column along with the mAb monomer. Secondly, the lowering
22
of pH during elution accelerates mAb aggregation [8]. Aggregated forms of mAb have
lower biological activity and can in some cases be toxic or immunogenic [6, 9]. Thirdly,
protein-A is known to leach out in small amounts from its support matrix, more so under
acidic conditions [10]. This leachate and its fragments are immunotoxic and their
presence in therapeutic products is highly undesirable [11-13]. For this reason, polishing
of protein-A affinity chromatography purified mAb become absolutely essential, i.e. for
the removal of product related impurities such as mAb aggregates and process related
impurities such as leached protein-A and its fragments.
Aggregated forms of mAbs are commonly removed by size exclusion chromatography
(or SEC) [14, 6]. The dimer and higher aggregates being larger are excluded from the
pores present in the chromatographic media and consequently move faster through the
column than the mAb monomer. The use of SEC for both preparative and analytical scale
separation of mAb aggregates has been reported [14, 6]. Some of the problems of using
SEC for preparative separation are low speed, poor scalability and limited resolution.
Researchers have tried to address these by combining SEC with other technique [15, 16].
Techniques such as field flow fractionation (or FFF) [17] and hydroxyapatite (or HA)
based column chromatography have also been used for removing aggregated forms of
monoclonal antibody [18, 19]. The use of ion exchange chromatography in the presence
of polyelectrolyte additives such as polyvinylsulfonic acid, polyacrylic acid and
polystyrenesulfonic acid for removing mAb aggregates has been reported [20]. Wang et
al. [7] used hydrophobic interaction membrane chromatography (or HIMC) for separating
23
mAb monomer from its dimer based on their hydrophobicity difference. In a subsequent
paper [21] the feasibility of fractionating different aggregated forms such as dimer, trimer,
tetramer and pentamer from each other using the same technique was also demonstrated.
There are several other more recent reports on the use of hydrophobic interaction
chromatography or hydrophobic charge induction chromatography employing both
columns and membranes for the removal of mAb aggregates [22-26]. The use of the
hydrophobic charged induction mode led to improvement in binding capacity, step yield,
and impurity removal rate [24]. In addition, less amount of salt was required when
compared with standard hydrophobic interaction chromatography.
The leaching of protein-A which co-elutes along with the mAb was systematically
studied by Bloom et al. [27]. They also examined the relationship between protein-A
contamination level and column usage. Protein-A is typically leached in small amounts
and its detection can in some cases be challenging. Steindl et al. [28] used sandwich-type
enzyme linked immusorbent assay to measure extremely low concentrations of leached
protein-A. However in most cases, standard techniques such as sodium dodecyl sulphate
polyacrylamide gel electrophoresis (or SDS-PAGE) are considered sensitive enough to
detect presence of leached protein-A which is co-eluted along with a huge excess of mAb
and is therefore likely to be present as complex with one or several mAb molecules. To
measure the exact concentration of protein-A, such complexes need to be first dissociated
by boiling in the presence of SDS [28] or by employing microwave [29]. Leached
protein-A and its fragments could be removed using anion exchange chromatography [27,
24
30], cation exchange chromatography [10, 30], and hydroxyapatite chromatography [18].
Some researchers have attempted to combine mAb aggregate and protein-A removal in a
single step. Zhou et al. [31] showed that cation exchange chromatography using hybrid
gradients of pH and conductivity resulted in removal of mAb aggregates and leached
protein-A while Kelley et al. [30] used upgraded anion exchange chromatography for
achieving the same objective. However, there is a limited repertoire of polishing
techniques which simultaneously remove both mAb aggregates and protein-A and
therefore new and efficient polishing techniques would be highly desirable.
Sun et al. [32] while investigating the mechanism of IgG binding on hydrophobic
surfaces, found that protein-A on its own did not interact with a hydrophobic membrane
except at very high salt concentrations, i.e. protein-A was quite hydrophilic when
compared with IgG. Interestingly though, IgG – protein-A complexes were found to be
significantly more hydrophobic than free IgG. Any protein-A leached out from affinity
media and co-eluted with mAb would, as reported by earlier workers [27, 28], be present
complexed with mAb. As observed by Sun et al. [32], such complexes are likely to be
more hydrophobic than the mAb. Based on this, we hypothesized that leached protein-A
could be removed from mAb using hydrophobic interaction chromatography. Also, by
proper optimization, removal of both mAb aggregates and protein-A could potentially be
combined into a single step.
25
Figure 1
Schematic diagram for monoclonal antibody polishing based on
hydrophobic interaction membrane chromatography
This paper discusses a HIMC based polishing method for simultaneous removal of
aggregates and leached protein-A from mAb. The working hypothesis for this method is
shown in Figure 1. The feed solution for this polishing step is mAb eluted from a proteinA column using acidic buffer. In addition to mAb, the feed would therefore also be
expected to contain some aggregates (represented by the dimer in the figure) and leached
protein-A (which would form complex with mAb as indicated in the figure). While
protein-A can bind up to five IgG molecules [33] and the figure shows just two of the
possible complexed forms for the sake of simplicity. In step 1, salt was added to the feed
solution to obtain conditions suitable for binding impurities on a hydrophobic membrane
stack. In step 2, the mAb being less hydrophobic than the impurities was obtained in the
26
flow through while the aggregates and mAb – protein-A complexes remained bound to
the membrane stack. In step 3, the membrane-bound impurities were eluted by lowering
salt concentration and the membrane stack was thus regenerated. This polishing method
was systematically optimized and the results obtained are discussed.
2.3
Materials and Methods
2.3.1 Materials
Hydrophilized polyvinylidene fluoride (PVDF) membrane discs with 0.22 μm nominal
pore diameter (catalogue no. GVWP 14250) were purchased from Millipore, Billerica,
MA, USA. The average membrane thickness was 125 μm. Humanized mAb (hIgG1-CD4,
batch no.10) and “dimers” (mixture of hIgG1-CD4 monomer and dimer, batch no. 6)
were kindly donated by the Therapeutic Antibody Center, University of Oxford, UK.
Pure hIgG1-CD4 stock solution had a total protein concentration of 4.76 mg/mL. The
“dimers” stock solution had a total protein concentration of 1.29 mg/mL and contained
slightly less than 50% monomer, 50% dimer and trace amounts of higher aggregates.
Sodium phosphate monobasic (catalogue no. S0751), sodium phosphate dibasic
(catalogue no. S0876), ammonium sulphate (catalogue no. A4418), and recombinant
protein A from Staphylococcus aureus expressed in Escherichia coli (catalogue no.
P7837) were purchased from Sigma-Aldrich, St. Louis, MO, USA. All buffers and test
solutions were prepared using ultrapure water (18.2 MΩ-cm) water obtained from a
27
DiamondTM NANOpure water purification unit (Barnstead International, Dubuque, IA,
USA). The buffers were filtered through a 0.45 μm microfiltration membrane for
clarification and degassing. Protein solutions were clarified by centrifugation at 10000
revolutions per minute for 10 minutes.
2.3.2 Methods
The HIMC experiments were carried using a stack of six hydrophilized PVDF
membranes (18 mm effective diameter) housed in a module which was designed and
fabricated in-house [34]. The module was integrated with an AKTA Prime liquid
chromatography system (GE Healthcare Bioscience, Uppsala, Sweden). UV absorbance
at 280nm, pH, conductivity and pressure data was collected during the chromatography
experiments and logged into a computer. The feed solutions used in the HIMC
experiments were adjusted to the appropriate ammonium sulphate concentration to
facilitate binding. The membrane bound material was eluted with ammonium sulphate
free buffer using appropriate gradients. Samples collected during these experiments were
analyzed by size exclusion chromatography (SEC) and sodium dodecyl sulphate
polyacrylamide gel electrophoresis (SDS-PAGE) [35].
SEC analysis was carried out with an Ultrahydrogel SEC column (catalogue number
WAT011530, 7.8mm i.d., 30cm length, Waters, Milford, Massachusetts, USA) using an
HPLC system (Prepstar 218, Varian Canada, Mississauga, Ontario). In these SEC
experiments, sodium chloride solution in water (0.25 M) was used as mobile phase at a
flow rate 0.3 mL/min. A 100 μL loop was used for sample injection. Reducing SDS28
PAGE (10% gel) was run using a Hoefer MiniVE vertical electrophoresis unit (GE
healthcare Bio-Sciences, Montreal, Quebec, Canada). Prior to loading on the gel, samples
were boiled in SDS and dithiothreitol (DTT) containing buffer to obtain individual
protein subunits and to break the disulfide bonds. The electrophoresis experiments were
run with normal polarity at 120 V (~35 mA current). Protein bands on the gel were
visualized by staining with Coomassie blue dye.
2.4
Results and Discussion
Figure 2
Effect of salt concentration on the binding of hIgG1-CD4 on
hydrophilized PVDF membrane (eluting buffer: 20 mM sodium phosphate, pH 7.0;
binding buffer: eluting buffer + different concentrations of ammonium sulfate (1) 1.2 M,
(2) 1.3 M, (3) 1.4 M and, (4) 1.5 M; hIgG-CD 4 concentration: 0.1 mg/mL, injection
volume: 500 μL, flow rate: 2 mL/min)
29
Figure 2 summarizes the effect of ammonium sulphate concentration on the binding of
pure mAb on a stack of six hydrophilized PVDF membrane discs. In the four experiments
represented in the figure, the concentration of ammonium sulphate in the binding buffer
was varied between 1.2 -1.5 M. Previous work based on the same membrane and mAb [7]
had shown that the presence of 1.5 M ammonium sulphate concentration in the binding
buffer resulted in strong binding of both mAb and its dimer. Therefore 1.5 M ammonium
sulphate was set as the maximum salt concentration (chromatogram 1) and the salt
concentration in the binding buffer was reduced in steps of 0.1 M (chromatograms 2, 3,
and 4 respectively). In all four cases, elution from the membrane stack was effected by
using negative salt gradient going from 0 to 100% eluting buffer, applied over 30 mL of
buffer flow. As expected, in the presence of 1.5 M salt, the entire amount of mAb
injected bound to the membrane stack (as indicated by the absence of a flow through
peaks in chromatogram 1). The membrane-bound mAb could easily be recovered in the
form of a sharp peak (51.2 mL retention volume) by switching over to the eluting buffer.
However, no mAb binding took place on the membrane stack when the ammonium
sulphate concentration in the binding buffer was lowered (i.e. to 1.4, 1.3 and 1.2 M,
corresponding to chromatograms 2, 3 and 4 respectively) as evident from the presence of
flow through peaks in these chromatograms. The shape of the flow through peak changed
with change in salt concentration: being quite broad at 1.4 M, slightly sharper at 1.3 M,
and very sharp at 1.2 M. This indicates that at 1.4 M salt concentration, the mAb
interacted to a certain extent with the membrane stack but this was not strong enough for
binding. The interaction decreased a bit at 1.3 M salt concentration and decreased even
30
further at 1.2 M salt concentration. Such change in the level of interaction is also
indicated by the difference in retention volumes of the flow through peaks, these being
13.1, 6.7 and 1.5 mL respectively at 1.4, 1.3 and 1.2 M salt concentrations.
Figure 3
Effect of salt concentration on the binding of hIgG1-CD4 “dimers” on
hydrophilized PVDF membrane (eluting buffer: 20 mM sodium phosphate, pH 7.0;
binding buffer: eluting buffer + different concentrations of ammonium sulfate (1) 1.5 M,
(2) 1.4 M, (3) 1.3 M and, (4) 1.2 M; hIgG1-CD 4 “dimers” concentration: 0.1 mg/mL,
injection volume: 500 μL, flow rate: 2 mL/min)
Figure 3 shows the effect of ammonium sulphate concentration in the binding buffer on
chromatograms obtained with the “dimers”, i.e. the mAb monomer-aggregate mixture. At
1.5 M ammonium sulphate concentration, almost all components present in the injected
sample bound to the membrane stack as indicated by the tiny flow through peak (see
chromatogram 1). When elution was carried out using a 30 mL linear gradient form 0 to
31
100% eluting buffer, the mAb monomer and dimer were observed as sharp peaks
corresponding to 50.5 and 54.1 mL retention volumes respectively. The small peaks
observed after the dimer peak correspond to higher aggregates which were known to be
present in small amounts in the “dimers” sample. When the binding buffer contained 1.4
M ammonium sulphate, the mAb monomer was obtained in the flow through as a very
broad peak corresponding to 7.4 mL retention volume while the eluted dimer peak shifted
in the forward direction to 51.4 mL (see chromatogram 2).When the salt concentration in
the binding buffer was further reduced to 1.3 M, the mAb monomer was obtained as a
very sharp flow through peak corresponding to a retention volume of 1.3 mL (see
chromatogram 3). The reduction in the size of the dimer peak (49.7 mL retention volume)
clearly indicates that some of the dimer went through the membrane stack along with the
monomer without binding. This unbound dimer fraction was presumably responsible for
the shoulder following the principal flow through peak. The higher aggregates were still
able to bind to the membrane stack and were eluted after the dimer peak. When the
ammonium sulphate concentration in the binding buffer was further reduced to 1.2 M,
both mAb monomer and dimer flowed through corresponding to peaks having retention
volumes of 1.3 and 3.8 mL respectively. However, the higher aggregates were still able to
bind to the membrane stack and were obtained as eluted peaks (47 – 53 mL retention
volume). The separation strategy for mAb polishing is to bind the mAb dimer and higher
aggregates on the membrane while obtaining the mAb monomer in the flow through. The
above results suggest that in order to achieve this, the ammonium sulphate concentration
in the binding buffer would have to be between 1.3 and 1.4 M. The main criteria for mAb
32
polishing step is to maximize the removal of aggregates while maintaining high monomer
recovery. The data shown in Figures 2 and 3 seem to suggest that the recovery of mAb
would not be significantly affected by salt concentration in the above range. However, the
binding capacity of aggregates could be anticipated to increase with salt concentration.
Based on these considerations, 1.37 M ammonium sulphate concentration was chosen for
the binding buffer for mAb polishing.
Figure 4
Effect of salt concentration on the binding of protein-A on hydrophilized
PVDF membrane (eluting buffer: 20 mM sodium phosphate, pH 7; binding buffer:
eluting buffer + different concentrations of ammonium sulfate (1) 1.5 M, (2) 3 M;
protein-A concentration: 0.1 mg/mL; injection volume: 500 μL, flow rate: 2 mL/min)
Figure 4 compares the binding of protein-A on hydrophilized PVDF membrane at two
different ammonium sulphate concentrations in the binding buffer, i.e. 1.5 and 3 M. At
the lower salt concentration (chromatogram 1) there was no protein-A binding as evident
33
from the absence of an eluted peak. However, at 3 M salt concentration, some protein-A
binding took place, the bound material being eluted as two peaks (retention volumes of
61.2 and 63.8 mL respectively). The presence of the two eluted peaks presumably
indicates that some aggregation of protein-A took place at 3 M ammonium sulphate
concentration. These results are consistent with those reported by Sun et al. [32], i.e.
protein-A is quite hydrophilic and does not bind on a hydrophobic surface except at very
high salt concentration.
Table 1 Different combinations of protein-A and mAb used in HIMC experiments represented in
Figure 5.
mAb mass (mg)
Protein-A mass (mg)
Protein-A: mAb
Protein-A
(mass ratio)
concentration (mole
fraction)
Chromatogram 1
0.035
0.002
0.057
0.16
Chromatogram 2
0.035
0.0015
0.043
0.125
Chromatogram 3
0.035
0.001
0.029
0.086
Chromatogram 4
0.035
0.0005
0.014
0.045
34
Figure 5
Effect of protein A amount on HIMC chromatograms obtained using mAb
– protein-A mixtures (membrane: hydrophilized PVDF; eluting buffer: 20 mM sodium
phosphate, pH 7.0; binding buffer: eluting buffer + 1.37 M ammonium sulfate;
chromatogram (1): 0.002 mg protein A + 0.035 mg mAb; chromatogram (2): 0.0015 mg
protein A + 0.035 mg mAb; chromatogram (3): 0.001 mg protein A + 0.035 mg mAb;
chromatogram (4): 0.0005 mg protein A + 0.035 mg mAb; injection volume: 500 μL,
flow rate: 2 mL/min)*
Sun et al. [32] reported that when protein-A was present in a solution along with excess
rabbit IgG, it formed complexes which were more hydrophobic than the free IgG.
Whether mAb –protein-A complexes behaved in a similar way was examined by carrying
out HIMC experiments with different combinations of mAb and protein-A in the feed
solution. The ammonium sulphate concentration in the binding buffer used in these
experiments was 1.37 M. Figure 5 shows the chromatograms obtained with the different
mAb – protein-A mixtures. In these experiments the amount of mAb in the sample was
kept fixed while the amount of protein-A was varied. Table 1 shows the mAb and
35
protein-A contents, the mass ratio and the protein-A concentration (expressed as mole
fraction) corresponding to the mixtures used to obtain chromatograms 1-4 shown in
Figure 5. As the amount of protein-A in the sample was increased, the area of the flow
through peak decreased and there was a corresponding increase in the amount of material
eluted from the membrane stack. The reason for this is that increasing the protein-A
concentration increased the amount of hydrophobic complexes in the feed solution and
consequently more material bound to the membrane. These results clearly demonstrate
that mAb – protein-A complexes are more hydrophobic than either mAb or protein-A,
this being consistent with results obtained with protein-A and rabbit IgG mixtures [32].
The above results also show that HIMC using hydrophilized PVDF membrane stack
could potentially be used to remove protein-A from mAb samples. The presence of
multiple eluted peaks suggests that different types of complexes were formed. In theory
one molecule of protein-A can bind up to five molecules of IgG [33].
36
Figure 6
Polishing of mAb by hydrophobic interaction membrane chromatography
(membrane: hydrophilized PVDF; eluting buffer: 20 mM sodium phosphate, pH 7.0;
binding buffer: eluting buffer + 1.37 M ammonium sulfate; feed: 0.04 mg mAb monomer
+ 0.005 mg “dimers” + 0.0025 mg protein-A; injection volume: 500 μL, flow rate: 2
mL/min)
The feasibility of polishing mAb using hydrophilized PVDF membrane was examined
using simulated feed consisting of an mAb solution prepared in 1.37 M ammonium
sulphate spiked with appropriate amounts of “dimers” and protein-A. The conductivity of
the feed solution was 156 mS/cm and the membrane stack consisted of six discs. The
injected feed sample (500 μL) which contained 0.04 mg of mAb monomer, 0.005 mg
“dimers”, and 0.0025 mg of protein-A represents the worst case scenario for mAb eluted
from a protein-A column. Figure 6 shows the chromatogram obtained in the above
separation experiment. A 30 mL linear gradient was employed to elute membrane-bound
material. The flow through peak (3.1 mL retention volume) consisted of free mAb while
37
the eluted peaks (50 – 55 mL retention volume) consisted of aggregates and mAb –
protein-A complexes. Sample corresponding to the two peaks were collected and
analysed by SEC and SDS-PAGE.
Figure 7
Size exclusion chromatography of samples obtained from the HIMC based
mAb polishing experiment (1: pure mAb monomer, 2: “dimers”, 3: simulated feed, 4:
flow through, 5: eluate; column: Ultrahydrogel SEC column, mobile phase: 0.25 M
sodium chloride solution in water, flow rate: 0.3 mL/min)
Figure 7 shows the SEC chromatograms obtained with the mAb monomer, “dimers”,
simulated feed, flow through, and eluate fractions collected during the mAb polishing
experiment. The mAb monomer standard showed one peak at 28 min retention time while
the “dimers” showed two peaks, the first corresponding to the mAb dimer (26 min
retention time) and the second corresponding to the mAb monomer. The simulated feed
solution showed two major peaks, one corresponding to the monomer, and the other
38
having a retention time of 24 min and corresponding to a much bigger molecule
(approximately 450 kDa) than the dimer (approximately 300 kDa). Quite clearly, the
presence of protein-A in the simulated feed resulted in the formation of large complexes
containing more than one mAb molecules. While protein-A has five mAb binding sites
[33], findings reported in earlier research paper seem to suggest that this theoretical
maximum is rarely reached. The static binding capacity of IgG subclass 1 on protein-A
has been found to be in the 3 [35] to 3.3 [36] range. The flow through obtained in the
polishing experiment contained the mAb monomer only as evident from the single peak
of 28 min retention time. The eluate gave two peaks, the first corresponding to the mAb –
protein-A complexes and the second corresponding to the mAb monomer. The above
results clearly show that while the polished mAb sample was pure, the mAb monomer
could not be totally recovered in the flow through. Based on material balance, the mAb
recovery was determined to be slightly above 80%.
39
Figure 8
Coomassie blue stained SDS-PAGE gel (10%, reducing) obtained with
samples from the HIMC based mAb polishing experiment (M: marker; lane 1: flow
through, lane 2: eluate, lane 3: simulated feed; lane 4: pure protein-A; lane 5: “dimers”;
lane 6: mAb monomer)
Figure 8 shows the stained SDS-PAGE gel obtained under reducing conditions with
samples from the mAb polishing experiment. The molecular weight marker proteins for
calibration were loaded in the “M” lane. The pure mAb monomer standard (hIgG1-CD4)
was loaded in lane 6 and the “dimers” sample was loaded in lane 5. The two bands in lane
6 correspond to the heavy and light chains the mAb splits into when boiled in SDS and
DTT containing buffer. The same bands were also obtained with the “dimers” sample as
the aggregated forms of the mAb were first disaggregated by the action of SDS followed
by their splitting into the heavy and light chains by the action of DTT. Pure protein-A
was loaded in lane 4 and this gave a single band slightly higher than the 43 kDa
40
molecular marker protein. The simulated feed sample was loaded in lane 3 and this gave
three bands corresponding to the heavy and light chains of mAb, and that of protein-A.
The flow through sample (lane 1) gave the two signature bands of the mAb heavy and
light chains while the eluate (lane 2) gave the protein-A band in addition to these. These
SDS-PAGE results conclusively prove that protein-A is indeed removed by the HIMC
based mAb polishing technique. The results discussed above clearly demonstrate that
hydrophobic interaction membrane chromatography could be used for polishing proteinA purified monoclonal antibody.
2.5
Conclusion
The experimental results discussed in this paper unambiguously demonstrate the
feasibility of using hydrophobic interaction membrane chromatography for removing
aggregates and leached protein-A from monoclonal antibody. The aggregates were
removed by selective adsorption on the membrane as these were more hydrophobic than
the monomeric form of the monoclonal antibody while protein-A was removed based on
the fact that it formed relatively hydrophobic complexes with the monoclonal antibody.
Under optimized solution conditions, these impurities could be simultaneously removed.
The monoclonal antibody was obtained in the flow through while the impurities remained
bound to the membrane and could subsequently be eluted by lowering salt concentration.
41
2.6
References
[1]
G. Kohler, C. Milstein, Derivation of specific antibody-producing tissue culture
and tumor lines by cell fusion, Eur. J. Immunol. 6 (1976) 511-519.
[2]
T.A. Waldmann, Monoclonal antibodies in diagnosis and therapy, Science 252
(1991) 1657-1662.
[3]
P.L. Ey, S.J. Prowse, C.R. Jenkin, Isolation of pure IgG1, IgG 2a, IgG 2b
immunoglobulins from mouse serum using protein A sepharose,
Immunochemistry 15 (1978) 429.
[4]
R.L. Fahrner, D.H. Whitney, M. Vanderlaan, G.S. Blank, Performance
comparison of protein A affinity-chromatography sorbents for purifying
recombinant monoclonal antibodies, Biotechnol. Appl. Biochem. 30 (1999) 121128.
[5]
S. Hober, K. Nord, M. Linhult, Protein A chromatography for antibody
purification, J. Chromatogr., B 848 (2007) 40-47.
[6]
J. Phillips, A. Drumm, P. Harrison, P. Bird, K. Bhamra, E. Berrie, G. Hale,
Manufacture and quality control of CAMPATH-1 antibodies for clinical trials,
Cytotherapy 3 (2001) 233-242.
[7]
L. Wang, J. Hale, R. Ghosh, Non-size-based membrane chromatographic
separation and analysis of monoclonal antibody aggregates, Anal. Chem. 78 (2006)
6863-6867.
[8]
T. Arakawa, J.S. Philo, K. Tsumoto, R. Yumioka, D. Ejima, Elution of antibodies
from a protein-A column by aqueous arginine solutions, Protein Expres Purif. 36
(2004) 244-248.
42
[9]
E.M. Cromwell, E. Hilario, F. Jacobson, Protein aggregation and bioprocessing,
AAPS J 8 (2006) 66.
[10]
J.N. Carter-Franklin, C. Victa, P. McDonald, R. Fahrner, Fragment of protein A
eluted during protein A affinity chromatography, J. Chromatogr., A 1163 (2007)
105-111.
[11]
G.J. Ventura, A.U. Buzdar, S. Kau, B. Lichtiger, G.N. Hortobagyl, Clinical trial
of plasma perfusion over immobilized Staphylococcal protein A in metastatic
breast cancer, Cancer Treat. Rep. 71 (1987) 411-413.
[12]
K.L. Carson KL, Flexibility- the guiding principle for antibody manufacturing,
Nat. Biotechnol. 23 (2005) 1054-1058.
[13]
S. Ghose, M. Jin, J. Liu, J. Hickey, Integrated polishing steps for monoclonal
antibody purification. In: Gottschalk U, editor. Process scale purification of
antibodies. New Jersey: John Wiley & Sons, Inc. New Jersey, 2009, P 145-164.
[14]
A. Litzen, J.K. Walter, H. Krischollek, K.G. Wahlund, Separation and
quantitation of monoclonal antibody aggregates by asymmetrical flow field-flow
fractionation and comparison to gel permeation chromatography, Anal. Biochem.
212 (1993) 469–480.
[15]
A.M. Azevedo, P.A.J. Rosa, I.F. Ferreira, M.R. Aires-Barros, Integrated process
for the purification of antibodies combining aqueous two-phase extraction,
hydrophobic interaction chromatography and size-exclusion chromatography, J.
Chromatogr., A 1213(2008) 154-161.
[16]
J.P. Gabrielson, M.L. Brader, A.H. Pekar, K.B. Mathis, G. Winter, J.F. Carpenter,
T.W. Randolph, Quantitation of aggregate levels in a recombinant humanized
monoclonal antibody formulation by size-exclusion chromatography,
43
asymmetrical flow field flow fractionation, and sedimentation velocity, J. Pharm.
Sci. 96 (2007) 268-279.
[17]
J. Pollastrini, Y. Jiang, Separation and characterization of protein aggregates and
particles by field flow fractionation, Curr. Pharm. Biotechnol. 10 (2009) 382-90.
[18]
P. Gagnon, P. Ng, J. Zhen, C. Aberin, J. He, H. Mekosh, L. Cummings, S. Zaidi,
R. Richieri, A ceramic hydroxyapatite-based purification platform: simultaneous
removal of leached protein A, aggregates, DNA, and Endotoxins from MAbs,
BioProcess Int. 4 (2006) 50-60.
[19]
P. Gagnon, Improved antibody aggregate removal by hydroxyapatite
chromatography in the presence of polyethylene glycol, J. Immunol. Methods 336
(2008) 222-228.
[20]
P. McDonald, C. Victa, J.N. Carter-Franklin, R. Fahrner, Selective antibody
precipitation using polyelectrolytes: A novel approach to the purification of
monoclonal antibodies, Biotechnol. Bioeng. 102 (2008) 1141-1151.
[21]
L. Wang, R. Ghosh, Fractionation of monoclonal antibody aggregates using
membrane chromatography, J. Membr. Sci. 318 (2008) 311-316.
[22]
Y. Lu, B. Williamson, R. Gillespie, Recent advancement in application of
hydrophobic interaction chromatography for antibody aggregate removal in
industrial purification process, Curr. Pharm. Biotechnol. 10 (2009) 427–433.
[23]
I. Kostareva, F. Hung, C. Campbell, Purification of antibody heteropolymers
using hydrophobic interaction chromatography, J. Chromatogr., A 1177(2008)
254-264.
[24]
J. Chen, J. Tetrault, A. Ley, Comparison of standard and new generation
hydrophobic interaction chromatography resins in the monoclonal antibody
purification process. J. Chromatogr., A 1177 (2008) 272-281.
44
[25]
M. Kuczewski, N. Fraud, R. Faber, G. Zarbis-Papastoitis, Development of a
polishing step using a hydrophobic interaction membrane adsorber with a
PER.C6-Drived recombinant antibody, Biotechnol. Bioeng. 105 (2010) 296-305.
[26]
J. Chen, J. Tetrault, Y. Zhang, A. Wasserman, G. Conley, M. DiLeo, E. Haimes,
A.E. Nixon, A. Ley, The distinctive separation attributes of mixed-mode resins
and their application in monoclonal antibody downstream purification process, J.
Chromatogr., A 1217 (2010) 216-224.
[27]
J.W. Bloom, M.F. Wong, G. Mitra, Detection and reduction of protein A
contamination in immobilized protein A purification monoclonal antibody
preparations, J. Immunol. Methods 117 (1989) 83-89.
[28]
F. Steindl, C. Armbruster, R. Hahn, C. Armbruster, H.W.D. Katinger, A simple
method to quantify staphylococcal protein A in the presence of human or animal
IgG in various samples, J. Immunol. Methods 235 (2000) 61-69.
[29]
J. Zhu-Shimoni, F. Gunawan, A. Thomas, M. Vanderlaan, J. Stults, Trace level
analysis of leached protein A in bioprocess samples without interference from the
large excess of rhMAb IgG, J. Immunol. Methods 341 (2009) 59-67.
[30]
B.D. Kelley, S.A. Tobler, P. Brown, J.L. Coffman, R. Godavarti, T. Iskra, M.
Switzer, S. Vunnum, Weak partitioning chromatography for anion exchange
purification of monoclonal antibodies, Biotechnol. Bioeng. 101 (2008) 553-566.
[31]
J.X. Zhou, S. Dermawan, F. Solamo, G. Flynn, R. Stenson, T. Tressel, S. Guhan,
pH-conductivity hybrid gradient cation-exchange chromatography for processscale monoclonal antibody purification, J. Chromatogr., A 1175 (2007) 69-80.
[32]
X. Sun, D. Yu, R. Ghosh, Study of hydrophobic interaction based binding of
immunoglobulin G on synthetic membranes, J. Membr. Sci. 344 (2009)165-171.
45
[33]
H. Hjelm, J. Sjodahl, J. Sjoquist, Immunologically active and structurally similar
fragments of protein-A from Staphylococcus aureus, Eur. J. Biochem. 57 (1975)
395–403.
[34]
R. Ghosh, T. Wong, Effect of module design on the efficiency of membrane
chromatographic separation processes, J. Membr. Sci. 281 (2006) 532 – 540.
[35]
S. Ghose, B. Hubbard, S.M. Carmer, Binding capacity differences for antibodies
and Fc-Fusion proteins on protein A chromatographic materials. Biotechnol.
Bioeng. 96 (2007) 768-779.
[35]
U.K. Laemmli, Cleavage of structural proteins during the assembly of the head of
bacteriophage T4, Nature 227 (1970) 680–685.
[36]
A. Jungbauer, R. Hahn, Engineering Protein A affinity chromatography, Curr.
Opin. Drug Discovery Dev. 7 (2004) 248.
46
Chapter 3
High-resolution hydrophobic interaction chromatographic
separation of multi-protein mixture using paper-based
composite membranes 2
2
This chapter is an edited version of the manuscript written by Seung Mi Yoo, Ryan Sungho Park, and
Raja Ghosh. Ryan Sungho Park was a project student who conducted membrane fabrication and membrane
characterization with different salts. Figure 3 and 4 was carried by Ryan Sungho Park. All his work was
conducted under Seung Mi co-supervision.
47
3.1 Abstract
Biopharmaceutical industry can benefit from usage of numerous protein separation
methods. In this study, paper based poly (N-vinylcaprolactam) (PVCL) hydrogel
composite membrane was developed for multi-protein separation purposes. These
membranes showed environment responsiveness that bind and elute proteins depending
on their hydrophobicity with variations in buffer salt concentration. Human
immunoglobulin G (hIgG), human serum albumin (HSA) and insulin were adopted to
demonstrate the membrane performance. The mixture of these three proteins was
introduced to HIC butyl column and the composite membrane for comparison. The result
indicated superior resolution via usage of the composite membrane over the HIC butyl
column. Optimized operations of gradient and stepwise change of eluting buffer
concentration demonstrated successful protein separation. HPLC-SEC and SDS-PAGE
were implemented to analyze the fractionates. The flexibility of designing and
constructing the composite membrane is able to allow the biopharmaceutical industry to
utilize this method for multiple purposes such as drug delivery vehicle and polishing.
48
3.2 Introduction
Protein separation in the biopharmaceutical industry is dominated by resin-based column
chromatography based on the principles of hydrophobic interaction [1], ion exchange [2],
affinity [3, 4], and size exclusion [5]. Because of its extensive applications in downstream
processes, resin-based column chromatography has been developed with improved
diversity in support matrix, ligand chemistry, and narrow pore distribution. Various
techniques have been investigated for complex protein separations. Although many
efforts have been made to improve the chromatographic resolution of protein separation,
resin-based chromatography has a number of limitations. In this method, the separation of
components primarily relies on pore diffusion, which requires sufficient interaction time
between the mobile and stationary phases. The long process time, consequently, requires
extra buffer consumption and creates high process costs. Moreover, the drop in pressure
that occurs across the column poses a challenging issue because silica and soft hydrogels
are easily deformed under the operating pressure.
Membrane adsorbents have been utilized as selective barriers according to their sizes to
remove products and process-related impurities in the biochemical engineering industry
for the last 60 years [6, 7]. Advances in membrane science have improved the usability of
adsorbents. They serve as surfaces and pores in affinity, charged, and
hydrophobic/hydrophilic membranes, in addition to their traditional applications as filters
and molecular weight cut-offs (MWCO) membranes. Because traditional bead-based
chromatography exhibited some process limitations, membranes have been adopted as
49
alternative separation tools, later called membrane chromatography [8]. Several
membranes are stacked in a module, where the membrane void space creates flow
channels. Membrane chromatography has been actively investigated since the 1980s [9].
Membrane chromatography has brought several advancements in the field of
biopharmaceutical engineering. The development of membrane technologies resulted in
simple scaling-up procedures, lower pressure drops across the membrane cartridge, and
larger loading volumes compared to column chromatography. It also resulted in great
improvement in chromatographic resolution. Moreover, the convection flow-driven
separation mechanism reduces processing time and consequently, buffer consumption,
thereby leaving a small footprint [10]. Owing to these advantages of membrane
chromatography, Sartorius commercialized regenerated cellulose membrane substrates
with covalently attached phenyl ligands, called Sartobind® Phenyl, which are used for
hydrophobic interaction chromatography. Pereira et al. (2010) demonstrated the
purification of plasmid DNA using Sartobind® by manipulating the concentration of
ammonium sulfate [11].
Methods used for the modification of membrane surfaces, such as plasma [12] and
chemical treatment [13], ion beam irradiation [14, 15], physical adsorption (coating) [16–
19], and grafting [20–23] have been extensively investigated. In particular, hydrogel
coating or grafting has been actively investigated for stimuli-responsive membranes that
undergo phase transition when subjected to changes in temperature, pH, and salt
concentration. These membranes are called “smart-membranes” because of their
50
responsive properties. PNIPAM has been actively researched for cross-linking or
interpenetration on membrane substrates to make a three dimensional formation [14, 24,
25]. This hydrogel network can modify its physical and chemical properties upon
encountering environmental stimuli. PVCL and PEG have also drawn interest in the field
of biochemical engineering because of their superior biocompatibility and lower toxicity
when compared to PNIPAM [26, 27].
PVCL hydrogel, like PNIPAM, is a thermo-responsive polymer that exhibits
hydrophobic-hydrophilic phase transition above its LCST at 37 °C. It swells above its
LCST and collapses below its LCST [28]. The presence of salt in the mobile phase tends
to decrease the LCST of PVCL. Moreover, the degree of polymerization of Nvinylcaprolactam can also affect its LCST [29]. Under experimental operating conditions,
at approximately 25 °C and high salt concentrations, PVCL exists in its collapsed state
and attracts proteins for adsorption. On the other hand, under conditions of low ionic
strength and temperatures above its LCST, it swells to its original state and releases the
adsorbed proteins.
Hydrophobic interaction chromatography involves the interaction between hydrophobic
patches present on proteins and hydrophobic ligands on the stationary phase [30]. Even
though it has been used for the separation of proteins for several decades, the mechanism
underlying hydrophobic interaction is still controversial. The “salting-out” theory is the
most widely accepted theory that explains hydrophobic interaction chromatography
(HIC). According to this theory, surface residues of proteins form hydrogen bonds with
51
water molecules, resulting in the formation of a water-film around the protein, thereby
preventing protein aggregation and precipitation. The ability to form hydrogen bonds
varies with protein sequence and structure, as well as folding configuration. The apparent
hydrophobicity of a protein depends on the nature and distribution of the amino acids on
the protein surface (hydrophobic index/hydropathy). As salt is added to the aqueous
solution, the protein loses the bound water molecules, which exposes it to the
hydrophobic stationary phase. The initial mobile phase in HIC requires high salt
concentrations to promote hydrophobic interaction between the protein and the stationary
phase. Upon interaction with a salt (stimuli), a more favorable energetic state is generated,
which results in a decreased Gibbs energy for the protein. Different salts increase the
interaction between the protein and the hydrophobic stationary phase to different extents.
The Hofmeister series ranks common ions according to their abilities to increase the
strength of water tension around proteins. This chart is commonly consulted for the
purification of proteins using the precipitation technique.
The increment in molal surface tension by various salts is shown below in the decreasing
order [31, 32].
MgCl2 > Na2SO4 > K2SO4 > (NH4)2SO4 > MgSO4 > Na2HPO4 > NaCl > LiCl > KSCN
This order is only valid for water tension, and may vary for different proteins. It has been
shown that anions dominantly control the hydrophobic strength of water tension.
Therefore, the concentration of salt in the buffer is an important factor that requires
careful consideration during a HIC experiment.
52
There have been numerous studies on the separation of protein mixtures by membrane
chromatography using the principles of ion exchange chromatography (IEX) and HIC.
Gerstner et al. (1992) reported that carboxymethyl (CM) and diethylaminoethyl (DEAE)
membranes could be used to separate a three-component mixture using an ion-exchange
moiety [33]. Castilho et al. (2002) demonstrated the purification of IgG from
immobilized protein-A ligands using polymer (polyvinyl alcohol [PVA])-coated nylon
membranes [34]. Yu et al. (2008) demonstrated that PEG-cellulose paper could be used to
separate mAbs from a CHO cell culture broth [35]. Kostanski et al. (2008) characterized
PVCL hydrogel-coated glass microfiber membranes to evaluate its environmental
responsiveness under different ionic strengths. The membranes were characterized by
evaluating their mass gain, infrared spectra, and media permeability. [36]. Huang et al.
(2009) designed a commercial microfiltration membrane (microporous PVDF) crosslinked with a polymer (PVCL) for the separation of equine ferritin and hIgG using
hydrophobic interactions [37]. Mah and Ghosh (2010) demonstrated the feasibility of
using Whatman filter paper coated with PVCL polymer to make inexpensive composite
membranes [17]. In a recent study, Himstedt et al. (2013) grew a PVCL polymer chain on
a regenerated cellulose membrane. The grafting method that is used to layer the hydrogel
on the membrane surface has a few advantages. First, the length of the polymer chains
can be controlled easily. In addition, this is a well-defined polymerization method
compared to random coating on the membrane substrate. The model proteins, BSA and
IgG, were used to test the adsorption as well as the binding capacity of the membrane
[37].
53
In this study, we coated PVCL hydrogel on a Whatman filter paper membrane for use in
stimulus-based hydrophobic interaction membrane chromatography [17]. Our preliminary
results were published in 2010 [17]. We demonstrated that environmentally responsive
PVCL composite membranes have a lower critical solution temperature in the presence of
ammonium sulfate. On the basis of these results, we hypothesized that a paper-based
composite membrane might be able to fractionate a multi-protein mixture by hydrophobic
interaction membrane chromatography. After consulting the Hofmeister series,
ammonium sulfate and sodium chloride were tested for their ability to influence
membrane responsiveness.
This paper discusses the fractionation of plasma proteins, HSA, hIgG, and a popular
hormone, insulin, using a “custom made” PVCL composite membrane chromatography
system based on hydrophobic interactions. HSA and hIgG are the two most abundant
proteins in human blood plasma and occur in a ratio of 7:2. In addition, insulin is an
important molecule in the pharmaceutical industry. Furthermore, the differences in their
hydrophobicities are very subtle [38]. To demonstrate the large capacity of the composite
membranes, bioseparation was tested for a large amount of protein. These membranes
allowed high flow rates while maintaining relatively low pressures. Moreover, they had a
high selectivity for proteins. Fractionated proteins were collected and analyzed via high
performance liquid chromatography-size exclusion chromatography (HPLC-SEC) and
sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The present
54
work aims to demonstrate the bioseparation of a multi-protein mixture and address some
challenges that are encountered when using paper-based composite membranes.
3.3 Materials and Methods
3.3.1 Materials
Whatman filter paper purchased (no.5, catalogue # 28462-166) from VWR and was used
for support matrix on composite membrane. Monomer, N-vinylcaprolactam (catalogue
#415464), and cross-linker, methylene-bisacrylamide (catalogue #146072), UV-initiator,
diphenyl (2,4,6-trimethylbenzoyl)-phosphine oxide/2-hydroxy-2-methylpropiophenone,
catalogue #405663) was purchased from Sigma-Aldrich, St. Louise, MO, USA. The
HPLC-grad n-isopropanol (catalogue # AH323-4) was purchased from Fisher Scientific,
Ottawa, ON, Canada. Human immunoglobulin G (hIgG) (≥95% purity, catalogue
#14506), Insulin from bovine pancreas (catalogue number I6634), sodium phosphate
(mono- and dibasic) (catalogue # S0751 and S0876), ammonium sulphate (catalogue
#A4418), and sodium chloride (catalogue # S5886) were purchased from Sigma-Aldrich,
St. Louise, MO, USA. Purified human serum albumin (HSA) was kindly donated by the
Scottish Blood Transfusion Service, Edinburgh, UK. All buffers and protein solutions
were prepared using ultrapure water (18.2 MΩ-cm) obtained from Diamond NANOpure
water purification unit (Barnstead International, Dubuque, IA, USA) and were filtered
through a 0.45 μm microfiltration membrane for clearing and degassing.
55
3.3.2 Separation of protein mixture with HIC butyl column
The protein mixture was consisted with human immunoglobulin, serum albumin, and
insulin with the mass ratio of 1.5: 2.5: 1. The protein mixture was loaded at the flow rate
of 1 mL/min onto a 1 mL column of butyl HIC (GE healthcare bioscience, Uppsala,
Sweden) where integrated with AKTA prime liquid chromatography system (GE
healthcare bioscience, Uppsala, Sweden). This process was obtained for comparison of
protein resolutions. The protein sample loading condition was equilibrated in 1.5 or 1.3
M ammonium sulphate buffer (pH 7) composed of 20 mM of sodium phosphate. The
loading condition was determined by optimization of individual protein bindings prior.
After protein mixture injected into AKTA prime system along with binding buffer,
stabilization time has given. Once the chromatogram became the plateau, a linear
gradient of decreasing ammonium sulphate concentration (1.5 or 1.3 to 1 M) was applied
over 20 mL of sodium phosphate. All the flow through and eluting proteins were
collected for further protein analysis. The desalting and concentrating of those fractions
was conducted with Amicon ultra centrifugal (3 kDa NMWCO) filter devices (Millipore
Corporation, Bedford, MA).
3.3.3 Paper based composite membrane preparation
The Whatman filter paper were cut in 20 mm diameter discs and conditioned for
measuring the mass of paper discs in constant temperature and humidity (23 °C and 50 %
relative humidity) room. The polymer solution was prepared in 3g of n-isopropanol with
different molar ratio of monomer (30%, 35% and 40%) and cross-linkers (2%, 3%, and
56
4%) and 0.2 of UV-initiator; the recipes were adopted from Mah and Ghosh 2010 [17].
The conditioned filter papers were immersed into the coating (polymer) solution for 5
min. The individual discs were sandwiched within laminating pouches and pressed with
rollers to seal filter paper discs entrapped polymer solution. This process also eliminated
any possible of oxygen inhibition on polymerization and well distribution of polymer
solution within membrane pores. The sealed discs were placed in UV chamber with UVirradiation at 350-400 nm for 20 min (3.3 mW/cm2 intensity). After fully polymerized
membrane discs were removed from laminating pouches and rinsed with DI water in
agitator at 37 °C for 24 hours. The composite membrane discs were weighed after dried
in paper conditioning room (23 °C and 50 % relative humidity). The mass gain and
thickness of composite membrane were measured with 20 discs.
3.3.4 Membrane chromatography set-up
Five same type composite membranes (20-mm diameter) were stacked within a customdesigned membrane module [17] which was then integrated with an AKTA prime liquid
chromatography system (GE Healthcare Biosciences, BAIE D'urfe, QC, Canada). The
UV absorbance (at 280 nm), conductivity and pH of the effluent stream from the module
as well as the system pressure data were logged into a computer using Prime View.
Appropriate loops were used to inject the feed samples into the membrane module. All
membrane chromatography experiments were carried out at ambient temperature (24 °C).
57
3.3.5 Hydraulic testing of membranes
Membrane stacks were tested for hydraulic permeability at 1 mL/min flow rate using two
buffers: 20 mM sodium phosphate pH 7.0 (buffer B) and 1.1 M ammonium sulphate
prepared in 20 mM sodium phosphate pH 7.0 (buffer A). The flow of buffers A and B
through the membrane stack was alternated several times and the backpressure was
recorded.
3.3.6 SDS-PAGE
Non-reducing SDS-PAGE (10% gel) was run by Hoefer MiniVE vertical electrophoresis
unit (GE healthcare Bioscience, Montreal, Quebec, Canada). Prior to loading on the gel,
samples were boiled in SDS containing buffer. A constant current 25mA/gel for 1.5 h in
separating gel was applied [39]. Protein bands on the gel were stained with 0.1%
Coomassie Brilliant Blue R-250 for 30 min at room temperature, followed by destaining
in 10 % (v/v) acetic acid and 30 % (v/v) methanol in water. The gel picture was obtained
using Bio-Imaging MiniBis Pto system (24-25-PR) purchased from DNR-Imaging
System, Jerusalem, Israel.
3.3.7 SEC-HPLC
SEC analysis was carried out with a TSKgel G4000 SWXL column (catalogue number
08542, 7.8 mm i.d., 30 cm length, TOSOH Bioscience LLC, King of Prussia, PA, USA)
using HPLC system (Prepstar 218, Varian Canada, Missussuage, Ontario). In these SEC
experiments, mobile phase; 0.25M of sodium chloride in 20 mM of sodium phosphate
solution at pH 7, was carried out at a flow rate of 0.5mL/min. Each analysis was injected
58
with 50 μL volume. The mobile phase was filtered through a 0.1 μm microfiltration
membrane for clearing and degassing.
3.4 Results and Discussions
3.4.1 Optimization of protein separation with resin based hydrophobic interaction
chromatography
Figure 1
Effect of salt concentration on the binding of three proteins (hIgG, HSA,
and Insulin) on butyl Sepharose TM HIC (eluting buffer: 20mM sodium phosphate, pH 7;
binding buffer: eluting buffer+ different concentrations of ammonium sulphate – (1) 1.3
M, (2) 1.5 M; model protein concentration: hIgG 0.21 mg+ HSA 0.35 mg+ Insulin 0.14
mg in 500μL injection volume)
59
The suitability of the membranes as chromatographic support had to be confirmed by
assessing their performance by challenging with the same proteins binding and eluting
condition on both bead-based HIC butyl column and membranes. The first
chromatographic runs were done by using pH 7.0, 1.3 M ammonium sulphate in 20 mM
sodium phosphate as the binding buffer and pH 7.0, 20 mM sodium phosphate as the
elution buffer. This is illustrated in Figure 1. The elution was done by negative gradient
over 20mL of buffer flow on HIC butyl bead-based column. The injected protein
solution was previously conditioned by adding ammonium sulphate salt up to 1.3M. The
second chromatographic runs were done by identical chromatographic conditions except
the binding buffer was 1.5 M ammonium sulphate. The maximum binding condition was
limited by insulin aggregation; at ammonium sulphate concentration of 1.5 M, the insulin
protein solution began to become cloudy which can induce negative effects on the
hydrophobicity of the interaction between the protein and butyl column. Higher
concentration of lyotropic salt takes up water molecules and consequently insulin protein
molecules form dimer to hexamers (reversible aggregates)[35]. The injection protein
sample (500 μL) containing 0.21 mg of hIgG, 0.35 mg of HSA, and 0.14 mg of insulin;
accounting to a total protein amount of 0.7 mg per 1 mL of HIC butyl column bed
volume. Both of chromatogram 1 and 2 had flow through peaks (1-2 mL retention
volume) and elution peaks (11-25 mL retention volume) obtained by negative salt
concentration. As expected, chromatogram 2 has smaller flow through protein peak while
more protein was eluted at 11-25 mL effluent volume since higher salt concentration
induced more hydrophobic interaction between proteins and stationary phase (butyl
60
column). The flow through peak (peak a) and elute peaks (peak b and c) samples in the
case of 1.3 M salt concentration (chromatogram 1) were collected for SDS-PAGE
analysis.
Figure 2
Coomassie blue stained SDS-PAGE gel (10%, non-reducing) obtained
with samples from the butyl Sepharose TM HIC. Multi-protein mixture (M: marker; lane
1: hIgG, lane 2: HSA, lane 3: insulin, lane 4: feed –three protein mixture, lane 5: flow
through-peak a, lane 6: first elute-peak b, lane 7: second elute-peak c)
The feed, flow through and eluted protein samples were analyzed by 10 % non-reducing
SDS-PAGE, the results were shown in Figure 2. The molecular marker loaded “M” lane
for reference and three standard samples were loaded in lane 1 (hIgG), lane 2 (HSA), and
lane 3 (insulin). The hIgG (150 kDa) had a band on the top of gel while HSA has two
bands at 150 kDa and 67 kDa due to the traceable amount of hIgG present in plasma
stock solution. The molecular weight of insulin is 5.8 kDa which appeared below 10 kDa
61
protein marker. In the figure, the color of the insulin band was not clearly shown; it had
silver color while other proteins were stained by Coomassie blue. Because insulin
naturally contained metal (Zn), its band on the gel is not blue. The feed sample (mixture
of three proteins), the collected samples (peak a, b, and c) were loaded in lane 4, 5, 6 and
7 respectively. For the feed sample (lane 4), three bands were observed corresponding to
hIgG, HSA, and insulin. The gel picture for the flow through (lane 5) fraction of the butyl
HIC experiment showed two faint bands which corresponded to the hIgG and HSA
respectively. The first eluate (peak b) showed a similar band as the flow through fraction
with a darker band while second eluate (peak c) contained all three proteins (hIgG, HSA,
and insulin). It clearly showed separation of these three proteins with HIC butyl column
is not the best choice.
62
3.4.2 Composite paper based membrane characterizations (salt responsive and
protein binding/elution test)
Figure 3
Pressure profile in buffer change experiment carried out using filter paper
based composite membranes (flow rate: 1.0 mL/min; 1.5 M ammonium sulphate in 20
mM sodium phosphate buffer, 3M of sodium chloride in 20 mM sodium phosphate buffer,
pH 7.0)
Based on previous work [17], among the 5 recipes of composite membranes, a
composition of 40 mass % monomer and 2 mole % cross-linker was selected for protein
separation experiment due to the reasonable hydraulic permeability (2.55 X 10-6 m Pa-1 s1
per m of bed height) and the saturation hIgG binding capacity (12 mg/mL bed volume)
at 1.1M ammonium sulphate. The five-membrane stack was placed in customized
membrane module and its transmembrane pressure (ΔP) was tested under two different
salts as shown in Figure 3. One salt is ammonium sulphate and the other is sodium
chloride which was intended for the potential usage on biological system application.
63
Since the PVCL hydrogel coating on filter paper has lower critical solution temperature
(LCST) of 32 °C, the polymer layer collapses in a high ionic strength solution when
operated at room temperature [41, 42]. Salt induces the hydrogel to collapse from swollen
salts as it explains in introduction section. It also causes the increase in the void space
within membrane pores and shortens tortuosity. The transmembrane pressure is a good
indication of the hydrogel state as well as hydrophobicity of membrane surface.
As expected membrane back pressure was as high as 0.046 MPa (ammonium sulphate)
and 0.038 MPa with 2.5 X 10-3 (Ns/m2) (sodium chloride) shown in Figure 3. As more
salt was introduced to the system as conductivity indicated, the membrane back pressure
dramatically decreased to approximately 0.035 MPa (in both salt cases) since the
hydrogel was collapsed. Under salt free environment, PVCL polymers are swollen to
induced protein desorption mode. On the other hand, when the salt is present, PVCL
polymer can collapse to the cellulose filter papers; in the collapsed condition more
favorable for protein adsorption. Regardless of the responsive properties of PVCL,
proteins are able to pass through the membrane because the pore size is much bigger than
the protein hydrodynamic radii. In contrast, non-coated membrane (Whatman filter paper
itself) has no environment response by salt additives. As the salt concentration increases,
the viscosity of mobile phase also increases (higher salt buffer cause higher
transmembrane pressure due to the salt liquid viscosity). For example, sodium chloride
solution has dynamic viscosity difference between 1 to 2.5 X10-3 N s/m2 at 0-3 M
concentration. The environmental responsiveness was manipulated for multi-component
64
separation with membrane chromatography [17,19,43]. Most previous works were done
only with ammonium sulphate solution. However, in addition to ammonium sulphate,
sodium chloride solution was also used in this study. The membranes are responsive to
sodium chloride as well although the pressure difference is not as high as ammonium
sulphate solution. This requires further investigation of appropriate model protein with
strong hydrophobicity also not for separation applications (not shown in this dissertation).
Figure 4
Protein binding effect test using composite membrane, hIgG 0.5mg, and
insulin 0.3 mg in 500 μL injection loop(flow rate: 1.0 mL/min; 0.6M and 0.8M of
ammonium sulphate in 20mM sodium phosphate buffer, pH 7.0)
Individual protein (hIgG and insulin) binding and elution was studied under different
ammonium sulphate concentration as shown in Figure 4. Insulin (0.3mg) was loaded at
65
0.6 M of ammonium sulphate as binding condition; however, UV absorbance indicated
unbound insulin flowed through the membrane system which indicates it was not
hydrophobic enough for insulin molecules. The second chromatogram indicates 0.8 M of
ammonium sulphate is strong enough for insulin and membrane interactions, and after
elution buffer applied to the system, insulin eluted as shown in Figure 4a. Similarly,
hydrophobicity of hIgG (0.5mg) was tested at 0.6M and 0.8Mammonium sulphate
binding condition. The protein hydropathy index is commonly used as a reference for
estimation of proteins and their stationary interactions. Each amino acid has hydrophobic
scale and summation of each numbers indicates hydrophobicity/hydrophilicity of proteins.
According to the hydropathy index, all three proteins have values below zero which
means they are hydrophobic. However, there are limitations for assessing degrees of
hydrophobicity scale on each protein, because they have three dimensional structures in
the liquid solution [44]. Based on individual protein binding experiment as shown in
Figure 4, three protein mixtures were investigated on fractionations with paper based
composite membrane using hydrophobic interactions.
66
3.4.3 Process optimization (elution mode)
Figure 5
Effect of salt concentration on the binding of three model proteins on
PVCL composite membrane with 20 mL of gradient (eluting buffer: 20 mm sodium
phosphate, pH 7.0; binding suffer: eluting buffer + 1.3 M ammonium sulphate, injected
proteins concentration: IgG 0.21 mg+ HSA 0.35 mg+ Insulin 0.14 mg in 500μL injection
volume, total protein loading 0.7 mg/0.25 mL of bed volume, flow rate: 1.0 mL/min)
Salt-responsive membrane chromatography was applied to the separation of proteins. As
shown in Figure 5, the separation of a mixture of three proteins, hIgG (0.21 mg) and HSA
(0.35 mg) and insulin (0.14 mg) was achieved by changing the salt concentration. The
total protein mass loaded was 0.7 mg in 0.25 mL membrane stack volume. The saturation
binding capacity of hIgG at 1.1 M of ammonium sulphate was 12 mg/mL [17]. The
binding buffer consists of 20 mM sodium phosphate (pH 7.0) containing 1.3 M of
ammonium sulphate. This solution passed through the system containing the membrane
67
stack. According to Figure 4, under 1.3 M of ammonium sulphate concentration, both
hIgG and insulin are bound to the membrane stack while HSA has no interactions with
the membrane. Elution was carried out by a low salt condition which was 20 mM sodium
phosphate for 20 mL of linear gradient. The elution order of the three proteins reflects
their hydrophobic properties; less hydrophobic proteins are eluted earlier and more
hydrophobic proteins are eluted from membrane later. Similar study was done with
hydrophobic interaction column chromatography (HIC) in order to verify the protein
hydrophobicity by its retention time [38]. As discussed in section 3.2, the property of the
surface of the stationary phase (PVCL coated membrane) was altered from hydrophobic
to hydrophilic. At effluent volume of 1-8 mL, the flow through peak was observed, and
the two elution peaks were observed at 18-28 mL and 28-38 mL of effluent volume.
When compared to other commercialized membranes (i.e. PVDF[43]),the resolution of
chromatogram had protein release for a longer period of time. The samples in accordance
with the peaks (FT, E1 and E2) were collected, and analysed for their protein contents by
SEC-HPLC and SDS-PAGE.
68
Figure 6
Size exclusion chromatography of samples obtained from the HIMC based
multi-proteins separation a) pure standard samples 1: hIgG, 2: HSA, 3: Insulin, 4: feed –
(hIgG+ HSA+ Insulin), b) fractions from figure 5, 1: flow through, 2: first eluting peak 3:
second eluting peak; column: TSKgel G3000SWXL, mobile phase: 0.25 M sodium
chloride solution in 20 mM of sodium phosphate, flow rate: 0.3 mL/min)
Figure 6 shows the SEC chromatography obtained from standard samples and fractions
from the gradient mode membrane separation. The standard sample chromatograms
shows that the retention times of hIgG, HSA, and insulin were 13.5 min, 15 min, and 26
min respectively. The chromatogram for the feed solution showed three peaks, fist
shoulder corresponded to hIgG, while the second peak had a retention time at 15 min
which corresponded to HSA, and the broad peak at 26 min corresponded to insulin. The
flow through and the two elute peaks were collected during the 20 mL gradient separation
condition. The flow through peak in the separation experiment contained the HSA; this
69
was shown by SEC analysis with a single peak of 15 min retention time. The first elute
peak showed one peak in SEC analysis at 13 min indicating only hIgG was present; while
in the second elute fraction, two small and one large peak at 13 min, 15 min, and 26 min
were observed. Therefore, by using gradient elution, HSA can be resolved in flow
through while hIgG can be collected in first eluted peak. However, the second elute peak
contained some residual hIgG and HSA remained with insulin.
Figure 7
Coomassie blue stained SDS-PAGE gel (10%, reducing) obtained with
samples from the HIMC based multi-protein mixture (M: marker; lane 1: hIgG, lane 2:
HSA, lane 3: insulin, lane 4: feed –three protein mixture, lane 5: flow through, lane 6:
first elute peak, lane 7: second elute peak
The flow through and eluate protein fractions from the PVCL coated membrane
chromatography separation were analyzed with 10 % non-reducing PAGE. Figure 7
shows the flow through and two eluates collected from 20 mL gradient elution mode. The
70
molecular marker was loaded at “M” lane for reference and three standard samples were
loaded in lane 1 (hIgG), lane 2 (HSA), and lane 3 (insulin). The feed sample (mixture of
three proteins) and the three fractions were loaded in lane 4, 5, 6 and 7 respectively. The
PAGE gel was more sensitive than SEC since HSA in SEC data did not show hIgG peak.
The multi-protein feed sample was loaded in lane 4, which had three bands on the gel
corresponded to hIgG, HSA, and insulin respectively. Lane 5 contained the flow through;
faint hIgG and dark broad HSA bands were observed. Lane 6 was loaded with the first
eluate sample (peak E1) which mostly contained hIgG. The second eluate (lane 7) was
analyzed and it seemed to have a small amount of hIgG though the primary content was
insulin. Same as the SEC analysis data, with gradient mode elution, trace amount of
unseparated proteins were still observed in fractions. This indicated there are still rooms
for improvement in resolution of protein separation.
71
Figure 8
Effect of salt concentration on the binding of three model proteins on
PVCL composite membrane with step change 0-70-100 % (eluting buffer: 20 mm sodium
phosphate, pH 7.0; binding suffer: eluting buffer + 1.3 M ammonium sulphate, injected
proteins concentration: IgG 0.21 mg+ HSA 0.35 mg+ Insulin 0.14 mg in 500μL injection
volume, flow rate: 1.0 mL/min)
The attempt to separate multi-proteins using PVCL composite membrane with gradient
operation mode has been described in section 3.3.1. In the following sections, the elution
was carried out by step change of buffer salt concentration for comparison. Figure 8
showed multi-protein separation using coated membranes with 1.3 M binding buffer and
the elution was carried out by step change of elution buffer; 0 to 70 % and 70 to 100% of
20 mM of phosphate. The flow through (FT) peak was shown at 1-8 mL of retention
volume and fist elution peak (E1) was collected at 20 to 26 mL. The second elution peak
(E2) appeared as a broad shoulder peak from 32 to 38 mL. As expected, the shapes of the
peaks were sharper than the ones from gradient elution experiment due to the PVCL
72
hydrogel phase transition by sudden salt removal. All three fractions were collected for
further protein analysis, HPLC-SEC and SDS-PAGE.
Figure 9
Size exclusion chromatography of samples obtained from the HIMC based
multi-proteins separation a) pure standard samples 1: hIgG, 2: HSA, 3: Insulin, 4: feed –
(hIgG+ HSA+ Insulin), b) fractions from figure 8, 1: flow through, 2: first elute peak, 3:
second elute peak, column: TSKgel G3000SWXL, mobile phase: 0.25 M sodium chloride
solution in 20 mM of sodium phosphate, flow rate: 0.3 mL/min)
Figure 9 represents the obtained chromatograms, for the flow through, and the two elute
fractions collected during the hydrophobic interaction membrane chromatography (HIMC)
stepwise (0-70-100 % of eluting buffer) separation. The flow through fraction
chromatogram (FT) has a single peak at 15 min which corresponded to HSA, and one
peak was observed for the first eluate fractionation (E1) at 13 min which corresponded to
73
hIgG. The last chromatogram, which was for the second eluate (E2), also had only one
peak at 26 min indicating only insulin was present in this fraction. This was strong
evidence showing that HSA protein (among three proteins) was able to flow through the
hydrogel coated membrane chromatography completely without any interaction with the
membrane since there is no HSA observed in any of the eluate fraction; meanwhile, hIgG
and insulin were bound to the membrane until the change in environment to lower salt
concentration. The hIgG’s elution was followed by insulin in the stepwise elution.
Comparing with the linear gradient mode, the stepwise elution has a better resolution as
shown in SEC-HPLC analysis.
Figure 10
Coomassie blue stained SDS-PAGE gel (10%, reducing) obtained with
samples from Figure 8 (M: marker; lane 1: hIgG, lane 2: HSA, lane 3: insulin, lane 4:
feed –three protein mixture, lane 5: flow through, lane 6: first elute peak, lane 7: second
elute peak)
74
The fraction samples collected from stepwise separation experiment were analysed by
SDS-PAGE, and the results are shown in Figure 10. Reference molecular ladder was
loaded in lane “M” and standard protein samples were loaded in lane 1 (hIgG), lane 2
(HSA), and lane 3 (insulin). The feed (mixture of three) sample was loaded in lane 4.
Lane 5 showed the flow through (FT) fraction which was collected from stepwise
separation only for the HSA content. While lane 6 (E1) showed primarily hIgG which
was collected from the step change of 0 – 70 % elution buffer. Finally, lane 7 (E2)
indicated the primary content in the second elution peak was insulin with a trace amount
of hIgG, and this was collected from the step change of 70 – 100 % elution buffer.
Compared with the separation of hIgG, HSA, and insulin using gradient elution, the
resolution of stepwise elution improved significantly. However, in real pharmaceutical
industry, it is challenging to operate a sharp change in salt concentration; it is likely that
the linear gradient condition is more suitable for industry scale operation. This research
project showed the feasibility of optimized operating condition; it also showed that PVCL
coated membrane was able to separate multi-component proteins.
3.5 Conclusions
Hydrophobic interaction membrane chromatography has been actively developed for
performance and cost reduced membrane. Here, poly (N-vinylcaprolactam) (PVCL)
hydrogel was coated on filter paper to obtain a responsive membrane since PVCL
hydrogel is LCST polymer. Lyotropic salts were lowering LCST at the same time being a
75
trigger for phase transition. As more ionic strength applies to membrane environment,
PVCL polymer deformed and collapsed. By manipulating responsive properties,
switchable membrane hydrophobicity (physical and chemical properties) successfully
demonstrated for three proteins separation. HIC butyl column was tested along with the
composite membrane in order to compare the resolution. The least hydrophobic protein
(HSA) flowed through at 20 mM of sodium phosphate in 1.3 M of ammonium sulphate.
The second most hydrophobic protein (hIgG) was eluted after 70% of stepwise elution
buffer obtained to the membrane chromatography system. Lastly, the most hydrophobic
protein (insulin) was eluted when 100 % of eluting buffer applied in the membrane
chromatography which the responsive membrane undergoes hydrated state. Optimized
elution mode which is stepwise change in elution buffer concentration, had the best
resolution as visualized in HPLC-SEC and SDS-PAGE protein analysis. The composite
membrane experienced some of fouling problems, and less sensitive phase transition than
grafted commercialized membrane (i.e. hydrophilized PVDF) [43]. Nevertheless, the
composite membrane has more freedom to tailor with degree of hydrogel crosslinking
and membrane size or shape, and higher permeability good for mass process unit such as
waste water treatment [45]. Product driven design is necessary depending on the
applications. For example, hydrophobic protein interaction for drug delivery vehicle
requires high content of hydrogel. On the other hand, hydraulic permeate flux
achievement in polishing step can implement less hydrogel content membranes. This cost
effective development of hydrogel based membrane chromatography could significantly
benefit the downstream purification or polishing step in biopharmaceutical industry.
76
3.6
References
[1]
S. Hierten, Some general aspects of hydrophobic interaction chromatography, J.
Chromatogr. A. 87 (1973) 325-331.
[2]
P.R. Levison, C. Mumford, M. Streater, A. Brandt-Nielsen, N.D. Pathirana, S.E.
Badger, Performance comparison of low-pressure ion-exchange chromatography
media for protein separation, J. Chromatogr. A. 760 (1997) 151–158.
[3]
H. Lakhiari, D. Muller, Purification of IgG and insulin on supports grafted by
sialic acid developing “thiophilic-like” interactions, J. Chromatogr. B. Analyt.
Technol. Biomed. Life Sci. 818 (2005) 53–9.
[4]
R. Hahn, R. Schlegel, A. Jungbauer, Comparison of protein A affinity sorbents,
790 (2003) 35–51.
[5]
J. Porath, P. Flodin, Gel filtration: A mrthod for desalting and group separation,
Nature. 183 (1959) 1657–1659.
[6]
R.W. Baker, Membrane technology. Encyclopedia of polymer science and
technology, John Wiley & Sons, Inc., Membrane technology & research, Inc.,
Menlo park, California, 2001.
[7]
D.K. Roper, E.N. Lightfoot, Separation of biomolecules using adsorptive
membranes, J. Chromatogr. A. 702 (1995) 3–26.
[8]
R. van Reis, a Zydney, Membrane separations in biotechnology., Curr. Opin.
Biotechnol. 12 (2001) 208–11.
[9]
S.R. Eisenberg, A.J. Grodzinsky, Electrically modulated membrane permeability, J.
Memb. Sci. 19 (1984) 173–194.
[10] R. Ghosh, Protein separation using membrane chromatography: opportunities and
challenges., J. Chromatogr. A. 952 (2002) 13–27.
[11] L.R. Pereira, D.M.F. Prazeres, M. Mateus, Hydrophobic interaction membrane
chromatography for plasmid DNA purification: Design and optimization., J. Sep.
Sci. 33 (2010) 1175–84.
[12] K. Asfardjani, Y. Secui, Y. Aurelle, N. Abidine, Effect of Plasma Treatments on
Wettability of Polysulfone and Polyetherimide, 43 (1991) 271–281.
77
[13] D. Mukherjee, A. Kulkarni, W.N. Gill, Chemical treatment for improved
performance of reverse osmosis membranes, 104 (1996) 239–249.
[14] R. Chennamsetty, I. Escobar, X. Xu, Characterization of commercial water
treatment membranes modified via ion beam irradiation, Desalination. 188 (2006)
203–212.
[15] K. Good, I. Escobar, X. Xu, M. Coleman, M. Ponting, Modification of commercial
water treatment membranes by ion beam irradiation, Desalination. 146 (2002)
259–264.
[16] J. Lei, M. Ulbricht, Macroinitiator-mediated photoreactive coating of membrane
surfaces with antifouling hydrogel layers, J. Memb. Sci. 455 (2014) 207–218.
[17] K.Z. Mah, R. Ghosh, Paper-based composite lyotropic salt-responsive membranes
for chromatographic separation of proteins, J. Memb. Sci. 360 (2010) 149–154.
[18] R. Huang, K.Z. Mah, M. Malta, L.K. Kostanski, C.D.M. Filipe, R. Ghosh,
Chromatographic separation of proteins using hydrophobic membrane shielded
with an environment-responsive hydrogel, J. Memb. Sci. 345 (2009) 177–182.
[19] R. Huang, L.K. Kostanski, C.D.M. Filipe, R. Ghosh, Environment-responsive
hydrogel-based ultrafiltration membranes for protein bioseparation, J. Memb. Sci.
336 (2009) 42–49.
[20] J.E. Kilduff, S. Mattaraj, M. Zhou, G. Belfort, Kinetics of Membrane Flux Decline:
The Role of Natural Colloids and Mitigation via Membrane Surface Modification,
J. Nanoparticle Res. 7 (2005) 525–544.
[21] N. Hilal, V. Kochkodan, L. Al-Khatib, T. Levadna, Surface modified polymeric
membranes to reduce (bio)fouling: a microbiological study using E. coli,
Desalination. 167 (2004) 293–300.
[22] J. Pieracci, J. V Crivello, G. Belfort, Photochemical modification of 10 kDa
polyethersulfone ultrafltration membranes for reduction of biofouling, J. Memb.
Sci. 156 (1999) 223–240.
[23] M. Taniguchi, G. Belfort, Low protein fouling synthetic membranes by UVassisted surface grafting modification: varying monomer type, J. Memb. Sci. 231
(2004) 147–157.
78
[24] M. Rossi, D. Crespy, Temperature-responsive polymers with LCST in the
physiological range and their applications in textiles TEMPERATURE IN THE
PHYSIOLOGICAL, Polym. Int. 1468 (2007) 1461–1468.
[25] G.V.R. Rao, M.E. Krug, S. Balamurugan, H. Xu, Q. Xu, G.P. Lo, Synthesis and
Characterization of Silica - Poly ( N -isopropylacrylamide ) Hybrid Membranes :
Switchable Molecular Filters, Chem. Mater. (2002) 5075–5080.
[26] H. Vihola, A. Laukkanen, L. Valtola, H. Tenhu, J. Hirvonen, Cytotoxicity of
thermosensitive polymers poly(N-isopropylacrylamide), poly(N-vinylcaprolactam)
and amphiphilically modified poly(N-vinylcaprolactam), Biomaterials. 26 (2005)
3055–3064.
[27] J. Kopecek, Hydrogel biomaterials: a smart future?, Biomaterials. 28 (2007) 5185–
92.
[28] M. Prabaharan, J.J. Grailer, D.A. Steeber, S. Gong, Stimuli-responsive chitosangraft-poly (N-vinylcaprolactam) as a promising material for controlled
hydrophobic drug delivery, Macromol. Biosci., 8 (2008), 843–851
[29] N.S. Ieong, M. Hasan, D.J. Phillips, Y. Saaka, R.K. O’Reilly, M.I. Gibson,
Polymers with molecular weight dependent LCSTs are essential for cooperative
behaviour, Polym. Chem., 3 (2012) 794–799.
[30] W.R. Melander, D. Corradini, C. Horvath, Salt-mediated retention of proteins in
hydrophobic-interaction chromatography. Application of solvophobic theory, J
Chromatogr, 317 (1984) 67-85.
[31] Y. Zhang, P.S. Cremer, Interactions between macromolecules and ions: the
Hofmeister series, Curr Opin, Chem, Biol, 10 (2006) 658-663.
[32] F. Hofmeister, Zur lehre von der wirkung der salze. Naunyn-Schmiedeberg’s
Archives of Pharmacology 24 (1888) 247-260.
[33] J.A. Gerstner, R. Hamilton, S.M. Cramer, Membrane chromatographic protein
separations systems for high-throughput, 596 (1992) 173–180.
[34] L.R. Castilho, F.B. Anspach, W.-D. Deckwer, Comparison of affinity membranes
for the purification of immunoglobulins, J. Memb. Sci. 207 (2002) 253–264.
[35] D. Yu, X. Chen, R. Pelton, R. Ghosh, Paper-PEG-based membranes for
hydrophobic interaction chromatography: purification of monoclonal antibody.,
Biotechnol. Bioeng. 99 (2008) 1434–42.
79
[36] L.K. Kostanski, R. Huang, R. Ghosh, C.D.M. Filipe, Biocompatible poly(Nvinyllactam)-based materials with environmentally-responsive permeability., J.
Biomater. Sci. Polym. Ed. 19 (2008) 275–90.
[37] H.H. Himstedt, X. Qian, J.R. Weaver, S.R. Wickramasinghe, Responsive
membranes for hydrophobic interaction chromatography, J. Memb. Sci. 447 (2013)
335–344.
[38] A. Ladiwala, F. Xia, Q. Luo, C.M. Breneman, S.M. Cramer, Investigation of
protein retention and selectivity in HIC systems using quantitative structure
retention relationship models., Biotechnol. Bioeng. 93 (2006) 836–50.
[39] U.K. Laemmli, Cleavage of structural proteins during the assembly of the head of
bacteriophage T4, Nature. 227 (1970) 680–685.
[40] W. Wang, Protein aggregation and its inhibition in biopharmaceutics., Int. J.
Pharm. 289 (2005) 1–30.
[41] S. Saeki, N. Kuwahara, M. Nakata, M. Kaneko, Phase separation of poly (ethylene
glycol) -water-salt systems, 18 (1977) 1027–1031.
[42] S. Peng, C. Wu, Surfactant Effect on pH and Temperature Sensitivities of Poly ( N
-vinylcaprolactam- co -sodium acrylate ) Microgels, (2001) 568–571.
[43] S.M. Yoo, R. Ghosh, Simultaneous removal of leached protein-A and aggregates
from monoclonal antibody using hydrophobic interaction membrane
chromatography, J. Memb. Sci. 390-391 (2012) 263–269.
[44] A. Mahn, M.E. Lienqueo, J.C. Salgado, Methods of calculating protein
hydrophobicity and their application in developing correlations to predict
hydrophobic interaction chromatography retention., J. Chromatogr. A. 1216 (2009)
1838–1844.
[45] E.G. Mah, Fabrication of paper based thermo-responsive membranes and
investigation for their use in adsorption of emerging water contaminants,
McMaster dissertation, 2012.
80
Chapter 4
Coating the surface of hollow fiber membranes with a natural
hydrogel for use in bioreactors3
3
This chapter is an edited version of a manuscript written by Seung Mi Yoo and Raja Ghosh. This
manuscript is ready to journal submission.
81
4.1
Abstract
The use of hollow fiber membrane bioreactor in biotechnology and tissue engineering is
rapidly expanding. However there are some drawbacks of hollow fiber membrane for cell
culture applications such as protein or small molecules fouling rate which lead to mass
transport limitations. There arises a need for a bioreactor where cells can be grown
efficiently with well define mass transport. This work focused on PES hollow fiber
membrane surface coating with calcium alginate hydrogel is to overcome membrane
bioreactor limitations mentioned above. Calcium alginate is well defined and studied for
a long time since its excellent biocompatibility. We consider here a hydrophobic hollow
fiber membrane coating with natural hydrogel method. Alginate coat was made by
calcium ions diffused from membrane lumen to extracapillary space through membrane
pores. Uniformly coated hydrogel was obtained by passive diffusion which was a gentle
and precise fabrication method. This fabrication method is novel, fully controllable with
simple and easy to manipulate fabricating conditions. Different cross-linker solvents and
reaction time were investigated. The developed composite hollow fiber membrane was
characterized by TEM for morphology. Hydrophilic test was conducted with water
contact angle and hydraulic permeability. In addition, optical microscope was used for
identifying the calcium alginate thickness. Our results imply calcium alginate coated
hollow fiber membrane can be used as a membrane bioreactor with three dimensional
scaffold or ultrafiltration system.
82
4.2
Introduction
Hollow fiber membrane bioreactors (HFMB) have been widely used for cell culture in the
biopharmaceutical and tissue engineering fields as well as for biomass filtration in water
treatment units since 1972 [1, 2]. A typical hollow fiber membrane set-up consists of
membranes arranged in a bundle and assembled within a module. This configuration
provides a high surface area over volume ratio and therefore, has applications in
continuous-perfusion culture systems. The set-up has been developed for use in
mammalian cell culture [3, 4], microbial culture for ethanol production [5], mAb
production [6–8], and enzymatic reactions [9, 10]. HFMB is also highly used for the
purification of drinking water [11], in industry and municipal water treatment units [12],
and dairy food processing [13].
The HFMB set-up has several advantages over static culture set-ups such as multi-wall
plate or T-flask, including the ability to achieve high cell density. Static culture is a type
of batch reactor in which nutrients, oxygen concentration, pH, and the accumulation of
toxic metabolites within the flask over time follow a gradient structure [14]. In this
method, there is a risk of contamination when the medium is being changed. Therefore,
HFMB has been adopted in cell culture to provide or remove nutrients, growth factors,
and metabolites through the porous wall of the membrane in a continuous manner [15].
Despite these significant advantages of HFMB, there are some limitations with respect to
its mass transport ability [16]. In this method, cells are suspended in the extracapillary
space and nutrients are provided through the membrane lumen. Nutrients diffuse through
83
the membrane pore driven by concentration differences, while the metabolites diffuse
from the extracapillary side into the lumen. However, cells that are present at a large
distance from the membrane have limited access to nutrients and oxygen, compared to
cells that are present close to the membrane pore. In addition, these biological substances
(nutrients and cytokines) tend to attract cells near the surface of the membrane, causing
the cells to adhere on the surface. This results in differences in the physiological and
biological states of cells at different distances from the membrane pores. Moreover, mass
transport across the membrane is inhibited by the dense layer of cells that grow near the
outer surface of the hollow fiber membrane. This increases the necrosis of cells within the
HFMB. Therefore, HFMB is suitable only for bulk cell production within short periods
(approximately 7 days) [17]. In addition, various components in water, such as natural
organic matter (NOM) and proteins, cause fouling of the membrane [18, 19]. This
decreases the flux and renders the process energy-intensive [20, 21].
Some recent studies have focused on improving the usability of HFMB in cell culture
applications. To this end, three-dimensional tissue engineered constructs with scaffolds
were proposed. Hollow fiber membranes have a capillary-like structure; on top of the
membrane surface, a hydrophilic polymer was mixed with the cells for better inoculation
in the extracapillary space. For example, Ye et al (2007) studied fibroblast cells by
mixing them with collagen and Cytodex 1 microcarriers and injecting them into the
extracapillary space of a cellulose acetate hollow fiber membrane. This inoculation
treatment to generate a composite 3D scaffold resulted in a high cell density and
84
consistent lactate production. Ellis and Chaudhuri (2007) [22] developed a poly(lactic-coglycolide) (PLGA) hollow fiber membrane that can be used as clinical-scale bone
constructs. This bioreactor enhances regeneration of bone tissues and provides an
environment similar to the human circulatory system. Antwiler (2008) [23] demonstrated
that the surface of a hollow fiber membrane may be treated with platelet lysate (PL),
plasma, and fibronectin (FN), which are cell adhesion molecules, simply by dipping the
membrane into the solution. Balmert et al. (2011) [24] evaluated the various circuit
configurations applicable to bioreactors for measuring the volume shift of inner and outer
fluids. The technical challenges involved in the process were resolved by designing
complex tube circuits for better operating conditions. This configuration prevents shock
due to changes in pressure and spikes in nutrient concentrations. Salzman (1999) [25]
developed an immobilized enzyme reactor (by analytical modeling and numerical
simulations) to obtain high conversions with intermediate flow rates and short operation
times. Ye et al. (2006) [26] demonstrated nutrient transport in hollow fiber membrane
bioreactors for 3D bone tissue.
Efforts are being made to construct a bioreactor with improved membrane filtration and
mass transport abilities. Polypropylene (PP), polyvinylidene fluoride (PVDF),
polytetrafluoroethylene (PTFE), polysulfone (PS), and polyethersulfone (PES) materials
are commonly used as separation layers because of their strong mechanical and chemical
strengths. However, these materials are inherently hydrophobic. Such membranes were
made more hydrophilic by using methods such as treatment with plasma [27, 28], grafting,
85
or coating [28–33], and blend modification [34, 35]. When pre-treated with ultra-violet
rays, plasma, or chemical reagents, functional monomers grow on the membrane surface
by initiating the active site [32, 36]. Although the membrane surface can be modified to
be more hydrophilic, surface treatments are energy-intensive and require long
modification times. Moreover, there is a high risk of un-polymerized monomer being left
behind on the membrane, which leads to increased process costs [37].
The calcium alginate-coated PES hollow fiber membrane might be more cell-friendly,
and have enhanced water permeability and mass transport abilities. Overall threedimension cell culture scaffolds provide high cell viability and targeted cell functionally
[38, 39]. Sodium alginate is a natural polymer extracted from brown algae. It is a
polysaccharide with (1-4)-linked β-D-mannuronate (M) and α-L-guluronic (G) residues.
The ratio and sequence of M and G can be varied by treatment with chemical reagents
during the extraction process [40, 41]. The alginate gel is cross-linked by divalent cations
such as Ca2+, Ba2+, and Sr2+, which link the G residues [42]. Calcium alginate is a wellknown agent that is used for cell encapsulation, a process that involves dropping the
alginate along with the cells (or drugs) into a calcium chloride bath. Within a specific
cross-linking time, the cells or drugs are embedded in the calcium alginate matrix and are
ready for further applications [43]. In addition, because the hydrogel is able to hold large
volumes of water, it has been used as extracellular matrix in wound healing and tissue
engineering [44]. It has low immunogenicity and toxicity, and an easy and simple crosslinking process; therefore, it has applications in the field of biochemical engineering.
86
In this study, we introduced alginate polymer onto a polyethersulfone (PES) hollow fiber
membrane by cross-linking calcium ions diffusing through the membrane pores to
improve membrane hydrophilicity. The degree of hydrophobicity was studied by
evaluating the water contact angle. With this simple and gentle fabrication method, we
achieved precisely controlled polymerization; moreover, the membrane was found to be
biocompatible. The water permeability of this composite membrane was investigated by
applying a positive pressure on the shell side of the membrane to drive water molecules
into the membrane lumen. A small solute (dimethyl sulfoxide [DMSO]) was used to
assess mass transfer limits for uncoated and coated membranes. DMSO is a common
solvent used in chemical engineering. It is also used as a cryoprotectant to preserve living
cells [44]. Coated membranes were analyzed by light and transmission electron
microscopy (TEM). To the best of our knowledge, there is no other report that describes
the coating of calcium alginate coated on a hollow fiber membrane using a simple crosslinker diffusion method.
4.3
Materials and Methods
4.3.1
Materials
Sodium alginate (W201502), antipyrine ® (A5882), and DMSO (D4540) were purchased
from Sigma-Aldrich, St. Louis, MO, USA. Calcium chloride (C77-500) and ethanol
(BP2818) were obtained from VWR, Mississauga, ON, Canada. Hydrophilic
87
polyethersulfone (PES) hollow fiber membranes (Tetronic, 0.8mm i.d., 1.2 mm o.d., 150
kDa MWCO) was kindly donated by Hydranautics Inc., Japan. All tested solutions and
reagents were prepared using ultra-pure water (18.2 MΩ-cm) obtained from a Diamond
NANOpure water purification unit (Barnstead International, Dubuque, IA, USA), filtered
just prior using by 0.2 μm pore size of Acrodisc® 25 mm syringe filter membranes (Pall
Life Science, QC, Canada).
4.3.2
A
Methods
B
C
D
Figure 1
The method for fabricating calcium alginate coating on ultrafiltration
hollow fiber membrane as a substrate. A) cross section view of hollow fiber membrane
lumen filled with water or 50 v/v% ethanol for pre-wetting step, B) after removal of
water or 50 v/v% ethanol, introduce of sodium alginate solution in shell side at the mean
while calcium chloride solution enter the membrane lumen, C) cross-linking of alginate
radially outward diffusion of calcium ions
88
Sodium alginate solution (1 wt %) was prepared in ultrapure water. 0.1 M of calcium
chloride solution was prepared in two solvents; one is water and the other is 50 v/v %
ethanol. The solution was allowed to stand for 24 hours at 4 °C before use in order to
evenly disperse the sodium alginate and to remove any entrapped air bubbles. Figure 1
shows some of steps involved in the coating process. The lumen of the hollow fiber
membrane was filled with the base solvent (i.e. water or ethanol) for pre-wetting inner
lumen of membrane presented in Figure 1A. This step could help better access of the
calcium chloride solution into the hydrophobic (PES membrane) membrane pores. After
removing base solvent from membrane lumen, the external and internal surface was
patted-dry and placed in membrane module for coating process. This was done to prevent
irregular cross-linking of alginate. The lumen was then filled with calcium chloride
solution meanwhile shell side was filled with sodium alginate solution for the required
duration in a given experiment (Figure 1B). Calcium ions passively diffused towards the
alginate solution and ionically cross-linked alginates to be calcium alginate gel on the
PES hollow fiber membrane (Figure 1C). After the given cross-linking duration,
unreacted sodium alginate solution was rinsed using water (Figure 1D).
Alginate coated hollow fiber membrane for tested by TEM were fixed with 2%
Glutaraldehyde (2% v/v) in 0.1M sodium cacodylate buffer pH 7.4. The sample were
then rinsed two times in buffer and post-fixed in 1 % osmium tetroxide in 0.1 M sodium
cacodylate buffer for 1 hour. The samples were then dehydrated using graded ethanol
series (i.e. 50%, 70%, 70%, 95% and 100%). The final dehydration for TEM samples was
89
done in 100 % propylene glycol (PPG). Infiltration with Spurr’s resin was carried out
using graded series (i.e. PPG:Spurr’s resin ratios of 2:1, 1:1, 1:2, and lastly 100 % Spurr’s
resin) with rotation of the samples in between solution changes. The samples were
transferred to embedding molds with were then filled with fresh 100 % Spurr’s resin and
polymerized overnight in a 60°C oven. Thin sections were cut using a Leica UCT
Ultramicrotome and picked up onto Cu grids. The sections were post-stained with uranyl
acetate and lead citrate and then viewed in a JEOL JEM 1200 EX TEMSCAN
transmission electron microscope (JEOL, Peabody, MA, USA) operated at an
accelerating voltage of 80kV.
Figure 2
Experimental set-up for membrane characterization on hydraulic
permeability at constant pressure, single PES hollow fiber membrane with effective
length 30 cm, membrane module with effective inner volume 23.57 mL, and data
acquisition (pressure and mass per time) with Logger pro.
90
Figure 2 shows the set-up used for the ultrafiltration experiment. The single PES hollow
fiber was cut in 30 cm and fixed in the membrane module which was immersed within
tap water for 30 min, for both coated and uncoated membranes. The content of feed tank
was kept under pressure using compressed air. The pressure was applied in the feed tank
as well as shell side of membrane module kept constant over the experiment. The
permeate was collected by balance in every minute and was used to determine the
permeate flux. Gas pressure sensor (Vernier software & Technology, Mississauga,
Canada) and balance were connected to the computer for data acquisition using Lab Pro
and Logger Pro (Vernier software & Technology, Mississauga, Canada).
Figure 3
Experimental set-up for DMSO (300 g/L) diffusion experiment, PES
ultrafiltration hollow fiber membrane with 30 cm effective length, Extracapillary space
effective volume 5.89 mL, the pressure data collected and analyzed in Logger Pro (flow
rate: 0.00866ml/min, pressure 101 kPa )
91
Figure 3 shows the set-up used for the test solute and drug diffusivity experiment. The
single PES hollow fiber placed in the membrane module. The feeder solutions (DMSO)
were placed outside the hollow fibers (i.e. also referred to as extra-capillary space) while
the receiver solution pumped through the membrane lumen with lowest flow rate
(0.00866 ml/min) by two peristaltic pumps (Ismatec®, Cole-Parmer Instrument Company,
Vernon Hills, IL, USA). Inner flow solution was recycled back to the feed tank which
was kept well mixed. Pressure data was monitored by pressure sensor and inner solution
sample was collected every 5min for UV absorbance. DMSO and antipyrine was
analysed by NanoDrop 2000C spectrophotometer (Thermo Fisher Scientific, Waltham,
MA, USA)
4.4
Results and Discussions
Figure 4
Magnified side view images of the uncoated/coated PES hollow fiber
membrane. Images were obtained by light microscopy with a magnification factor of 25.
Coating conditions: 1 wt% Alginate concentration, 0.1 M calcium chloride concentration
in 50 v/v% ethanol, and 40 min cross-linking time
92
Figure 4 shows one example of fabrication of composite membrane carried out with
ultrafiltration PES hollow fiber membrane using calcium alginate hydrogel. The
experiment was carried out by 1 wt% of sodium alginate with cross-linked for 40 min by
0.1 M of calcium alginate solution in 50 v/v% EtOH. Top of the figure had uncoated PES
hollow fiber membrane side view and the bottom sample was calcium alginate coated
PES hollow fiber membrane. Natural hydrogels, such as alginate, collagen, agarose, and
chitosan, have lower mechanical strength than synthetic hydrogels [45]. One of the
features of hydrogel is environmental responsiveness under salt, pH, and temperature [46].
In this experiment, not only the coating process but also the tests with solutions that had
salt presented for reversible collapse and swell of hydrogel were done. Once calcium
alginate coated membrane was immersed into the 1M of ammonium sulphate solution,
the hydrogel was collapsed. However, the collapsed hydrogel was not re-hydrated
irreversibly, after water applied on dehydrated sample. As we expected, salt took the
water body from the calcium alginate structure. At the mean time all the connections
between the polymers broke. Under microscopy, the damaged hydrogel seemd like flakes
on the membrane surface, which is not shown in figure.
93
Figure 5
Fabrication of calcium alginate coated hollow fiber membrane
controllability test. Fabrication conditions: 1wt% alginate concentration, 0.1M of calcium
chloride concentration, varied cross-linking time 0, 5, 10, 15, 20, and 40 min.
The calcium alginate coating process was systemically controllable using different crosslinking times shown in Figure 5. This experiment was accommodated only with calcium
chloride in 50 v/v % ethanol solution because calcium alginate solution in water based
was not able to be verified for the hydrogel thickness below the level (25 times zoom) of
light microscopes. The six of membranes were aligned side by side; the uncoated
membrane (A), the once coated for 5 min (B), 10 min (C), 15 min (D), 20 min (E), and 40
min (F) with 0.1 M of calcium chloride cross-linker in 50 v/v % of ethanol. All the
samples were cross-linked and put in storage in 0.1 M of calcium chloride solution for
preventing deformation of hydrogel. Under the light microscope, samples were took
snapshots for measuring the thickness using (Image J) software, which are presented in
Table 1. The correlation between cross-linking time and thickness of hydrogel well fitted
94
first order (0-20 min). As a result of this experiment, the controllability of the amount of
hydrogel deposit on membrane surface is high enough for various applications.
Table 1. PES hollow fiber membrane coated with calcium alginate; 0.1 M calcium
chloride solution was prepared in 50 v/v% ethanol, 1 wt % sodium alginate, in different
cross-linking time
Time
Outer Diameter (n=4)
0 min
1.3 mm
5 min
1.53 ± 0.025 mm
10 min
1.74 ± 0.015 mm
15 min
1.89 ± 0.018 mm
20 min
2.34 ± 0.02 mm
40 min
2.99 mm
Figure 6
Hollow fiber membrane coating with calcium alginate hydrogel A) Dew:
calcium chloride solution pass through membrane pores B) 5 min cross-linking process C)
15 min cross-linking time. Fabrication condition: sodium alginate 1 wt%, calcium
chloride 0.1 M in water, PES hollow fiber membrane 150 kDa MWCO.
95
Figure 6 shows another method for coating technique which the calcium chloride solution
fully filled up under pressure in the membrane pores (Figure 6A). At t = 0, calcium ions
are ready to cross-link alginate while the passive diffusion method requires time for
calcium ions travel from inner membrane pore entrance to outer pore exit to meet alginate
molecules. The difference is only where the calcium ion starts polymerization, the rest of
fabricate conditions are the constant: diffusion coefficient of calcium chloride (in water
10-9m2/s) [47], tortuosity of pores, porosity of the membrane, concentration differences,
and temperature. Because the calcium ions are ready to cross-linking, the calcium
alginate gel are much thicker (Figure 6B). According to Cuadro et al. 2012 the estimated
gelation of calcium alginate time was 40 sec for 2% of alginate and 0.5 % calcium
chloride with 200 μm radius cylinder [48]. Calcium alginate polymerization is a rapid
process if cross-linker availability is high enough. For this reason prefilled calcium
chloride method showed more calcium alginate cross-linked. However, the roughness can
be another point drawing consideration for various applications. Two samples in Figure 6
B and C showed different surface contours. The ultrafiltration hollow fiber membrane is
in asymmetric conformation due to the extrusion manufacturing method, in which the
pores presented unevenly on the hollow fiber wall compared to microfiltration membrane.
The rough surface is a good indication of rapid calcium chloride polymerizing alginate
from uneven location whereas passive diffusion process lead to controlled slow
polymerization from membrane pores. This mass transport may counter the diffusion of
alginate solution into membrane pores and slow calcium ions contacting, which generate
calcium alginate gel morphologies. From this experiment, the application requires rapid
96
and thin layer of calcium alginate coating on the membrane surface for tissue engineering
or immobilize cells in short period time at initial seeding step and it has more merits than
passive method.
Figure 7
TEM image obtained with the hollow fiber and composite membrane
fabricated with 1 wt% sodium alginate and 0.1 M calcium chloride on PES ultrafiltration
membrane, with a cross-linking time of 30 min. (A) control sample: uncoated
membrane,25000 X magnification (B) 1 wt % of sodium alginate and 0.1 M calcium
alginate in water, 5000 X magnification (C) bigger magnification of sample B, 20000X
(D) 0.1M of calcium chloride solution in 50 v/v EtOH %, 20000 X magnification
The calcium alginate coating morphology was observed by TEM analysis as shown in
Figure 7. The samples were prepared three cases; one is uncoated membrane (A), second
97
is cross-linker solution dissolved in water (B and C), third one is cross-linker solution
dissolved in 50 v/v% ethanol. Only the solvents for cross-linking agent were varied;
concentrations of alginate, calcium chloride, and cross-linking time were kept constant.
The control (uncoated PES hollow fiber membrane cross section) specimen is presented
in Figure 7A. According to the degree of shade, there are three sections. Resin, top layer
of membrane extracapillary side, and inner membrane were well indicated as expected.
Figure 7 B and C are the same piece with different magnifications. The coated hydrogel
thickness presented thinner than fresh piece because the preparation processes involved
non-polar solvent caused dehydration of calcium alginate gel. Under higher magnification,
hydrogel pore size and distribution is different in distance. The calcium alginate gel pore
size has been gradually decreased toward to the edge of the outer alginate gel. Whereas
ethanol based cross-linking cases (D) showed opposite trend. There are two possibilities
to consider. One is calcium ion diffusivity in different solvent interactions with alginate
base solution. Another point is ethanol wets membrane inner and pores faster than water
so this enhances the calcium ions travel to alginate solution faster. The second possibility
is alginate precipitating by solvent. Hassan reported much work about cross-linked
alginate hydrogel where the forms go through the sol-gel state [49-51]. For better
explanation of trend of calcium alginate formation, the compatibility of alginate solution
and ethanol was investigated in Figure 8.
98
Figure 8
Calcium alginate polymerization study with different calcium chloride
solvents (A: alginate + water based calcium chloride, B: alginate + 50v/v% ethanol based
calcium chloride, C: alginate + 100 % ethanol, D: ethanol + alginate, E: sample D vertical
pose), 1 wt% sodium alginate of 1mL, 0.1 M calcium chloride 300 μL
Figure 8 illustrated four different cases of alginate and calcium chloride solutions.
Concentration of alginate solution with 1ml mixed with calcium alginate solution 0.1M in
water based (A), 50 v/v% ethanol (B). In comparison of two samples A and B, the
mixture of alginate and calcium chloride well mixed in A, while two phases separate right
after two liquid meet. Furthermore 100% of ethanol was injected in alginate solution (C)
and rapidly alginate solution had chunk of sol with many bubbles. It suggested that
ethanol induced the alginate into dehydrated state and precipitated first and after the
solution was well mixed, aggregated alginate disappeared. Alginate aggregation was
carried out in large volume of ethanol with small amount of alginate solution (D and E).
This transition of alginate was well supported for Figure 7. From simple step of calcium
alginate gel made of water based cross-linker case in Figure 7 C, this is a simple passive
diffusion through the membrane pores and cross-linked alginate G residues. Moreover
ethanol based cross-linker was one more step involved, first ethanol diffuse radically
toward to shell side of alginate solution. Precipitate or aggregate alginate near the
99
membrane surface then calcium ions tides each alginate G residues. This is the reason
why ethanol based calcium alginate had denser pores placed near the membrane surface.
Figure 9
Fabrication of calcium alginate coated on PES hollow fiber membrane
(Alginate concentration 1wt%, calcium chloride concentration 0.1M and 1 min crosslinking time) optical microscope 25 times.
Figure 9 presented the fabrication of calcium alginate composite membrane that was
carried out with 1 wt% of alginate, 0.1 M of calcium chloride solution in water. The
cross-linking was done for 1min. Since the short exposure time to cross-linking agent, the
calcium alginate hydrogel generated was too thin to measure under optical microscope.
However, top sample is corresponding to uncoated membrane with rough/scratched skin
while bottom sample has clean and smooth surface. Even though calcium alginate coating
layer is not able to observe, light reflections of the samples are different. For more
membrane characterization in terms of hydrophilized surface assessment, contact angle
analysis was carried out.
100
Figure 10
Water contact angle analysis with PES hollow fiber membrane (left), and
calcium alginate coated membrane (right). Picture obtained by digital camera.
Contact angle was tested to identify the hydrophobicity of coated and uncoated
membrane surfaces. The same samples were used from Figure 10. Contact angle can
provide information about the interfacial forces of water and membrane surfaces shown
in Figure 8. Left side of sample is uncoated PES membrane that is relatively hydrophobic
than right side of sample. Coated membrane is friendlier to water which pulls the water
toward to the membrane surface, which is indicated by the water contour changes. This
wetting effect is an indication of the membrane surface hydrophilicity improvement.
101
50
45
40
35
30
25
20
15
10
5
0
Uncoated membrane
coated membrane
2 4 6 8 10121416182022242628
Filtration Time (min)
Figure 11
Measurement of water flux and hydraulic permeability of uncoated/coated
PES hollow fiber membrane, operating pressure at 109.45 kPa (uncoated membrane) and
110.56 kPa (coated membrane) with constant (one data set).
Figure 11 shows the water flux and hydraulic permeability of the uncoated and coated
hollow fiber membranes. The membrane was prepared in the same condition with Figure
9, and 10 with 1 min cross-linking time of 1wt% alginate concentration and 0.1 M of
calcium chloride. The experimental set up is explained in Figure 2. The water flux was
calculated by
𝑊𝑎𝑡𝑒𝑟 𝑓𝑙𝑢𝑥 =
Permeated water volume
Membrane area × Filtration time
102
as shown in Figure 11. Since the pressure applied by compressed air was not exactly
same value in two cases, hydraulic permeability can be a factor to normalize to overlook
the membrane filtration behaviour with consideration of pressure term. The hydraulic
permeability was obtained by
𝑃𝑒𝑟𝑚𝑒𝑎𝑏𝑖𝑙𝑖𝑡𝑦 =
Flux × Viscosity of water
Transmembrane pressure
[50], where the viscosity of water is 0.000145 pa.s. The hydraulic permeability also had
the trend to decrease over the filtration volume. Coated membrane has similar water flux
behaviour. There are two possibilities for this phenomenon. One is fouling behaviour
over filtration time. The second one is coated hydrogel was in fully hydrated state prior
the filtration process and then the pressure applied from outside to inside made hydrogel
collapse, which increased the resistance of permeate flux. The water permeability of
coated membrane had higher than uncoated membrane because calcium alginate was not
working as a barrier, on the contrary it helps penetration of water using hydrophilic
characteristic of the hydrogel. Sawada et al. 2012 reported PES hollow fiber membrane
was grafted on acrylamide incubating silver nanoparticles. It shows hydrophilicity
improved by contact angle measure value however the water permeability was decreased
by grafting layer [53]. It is suggested that even one more layer of hydrophilic hydrogel
presented on the membrane surface, the hydraulic permeability does not change
membrane performance.
103
Figure 12
Effect of coating hollow fiber membrane on DMSO mass transport with
Fig 3 experimental set up (one data set).
Figure 12 shows the effect of coating hollow fiber membrane by solute diffusion from
shell side of membrane to inner lumen. It was obtained with 300 g/L of DMSO feed
solution which slowly diffused through the membrane pores into the circulated water in
inner lumen during 200 minutes the circulation ran with 0.00866ml/min flow rate.
Samples were collected from the permeate reservoir by 10 uL for monitoring DMSO
concentration. The increase of DMSO concentration in membrane lumen is good
indicator for small solute transport through membrane pores. At the beginning of
diffusion experiment (0-100 min), the DMSO diffusion rate was higher because the
concentration difference between two sides (inner and outer) of membrane was high. In
this experiment, pressure is atmosphere and it means the concentration difference of shell
104
and lumen side is mass transport driving forces. After 100 min of diffusion run, the flux
of DMSO decreased and reached a plateau. This experiment suggested the calcium
alginate coated membrane had similar mass transport behaviour.
4.5
Conclusions
Calcium alginate hydrogel was successfully coated on PES hollow fiber membrane by
passive diffusion of cross-linking agent through membrane pores. This coating technique
is gentle, precisely controllable on hydrogel thickness, as well as simple in experimental
set-up. Effect of cross-linking agent solvent was studied on thickness and morphology of
calcium alginate by TEM analysis. Calcium alginate coated membrane was characterized
by water contact angle as well as optical microscope. Since calcium alginate has been
used in tissue engineering with good biocompatibility, this technique will offer many
advantages: tissue engineering application for three dimensional MBR with similar
hydraulic permeability, material diffusion rate, and better cell adhesion effect due to the
hydrophilic coated on membrane. Calcium alginate coating made water favorable for
easy permeation and this characteristic helps the water purification process, too. However,
there are lots of room to improve this technique. Each proteins or simulated cell culture
medium can be tested with coated membrane for tissue engineering aspect, and several
coated membranes put as a bundle for tests of the water filtration performance since
uncoated and coated membrane indicated similar hydraulic permeability.
105
4.6
Reference
[1]
A.S. Michaels, C.R. Robertson, S.N. Cohen, Microbiological methods using
hollow fiber membrane reactor, (1984) US4440853
[2]
R.A. Knazek, P.M. Gullino, P.O. Kohler, R.L. Dedrick, Cell Culture on Artificial
Capillaries: An Approach to Tissue Growth in vitro, Science. 178 (1972) 65–67.
[3]
D.Q. Wu, G.L. Zhang, C. Shen, Q. Zhao, H. Li, Q. Meng, Evaluation of diffusion
in gel entrapment cell culture within hollow fibers., World J. Gastroenterol. 11
(2005) 1599–604.
[4]
H. Gloeckner, H.D. Lemke, New miniaturized hollow-fiber bioreactor for in vivo
like cell culture, cell expansion, and production of cell-derived products,
Biotechnol. Prog. 17 (2001) 828–31.
[5]
D. Inloes, D. Taylor, Ethanol production by Saccharomyces cerevisiae
immobilized in hollow-fiber membrane bioreactors, Appl. Environ. Microbiol. 46
(1983) 264–278.
[6]
P.J. Yazaki, L. Shively, C. Clark, C.W. Cheung, W. Le, B. Szpikowska, J.E.
Shively, A.A. Raubitschek, A.M. Wu, Mammalian expression and hollow fiber
bioreactor production of recombinant anti-CEA diabody and minibody for clinical
applications, J. Immunol. Methods. 253 (2001) 195–208.
[7]
L.R. Jackson, L.J. Trudel, J.G. Fox, N.S. Lipman, Evaluation of hollow fiber
bioreactors as an alternative to murine ascites production for small scale
monoclonal antibody production, J. Immunol. Methods. 189 (1996) 217–31.
[8]
D. Lowrey, S. Murphy, R.A. Goffe, A comparison of monoclonal antibody
productivity in different hollow fiber bioreactors, J. Biotechnol. 36 (1994) 35–8.
[9]
P. Ye, Z.K. Xu, A.F. Che, J. Wu, P. Seta, Chitosan-tethered poly(acrylonitrile-comaleic acid) hollow fiber membrane for lipase immobilization, Biomaterials. 26
(2005) 6394–403.
[10] H. Kawakita, K. Sugita, K. Saito, M. Tamada, T. Sugo, H. Kawamoto, Production
of cycloisomaltooligosaccharides from dextran using enzyme immobilized in
multilayers onto porous membranes, Biotechnol. Prog. 18 (2002) 465–9.
106
[11] S. Nakatsuka, I. Nakate, T. Miyano, Drinking water treatment by using
ultrafiltration hollow fiber membranes, Desalination. 106 (1996) 55–61.
[12] C. Lubello, R. Gori, Membrane bio-reactor for textile wastewater treatment plant
upgrading, water science and technology, J. Int. Assoc. Water Pollut. Res. 52
(2005) 91–98.
[13] Y. Pouliot, Membrane processes in dairy technology – from a simple idea to
worldwide panacea, Int. Dairy J. 18 (2008) 735–740.
[14] L.K. Nielsen, Bioreactors for hematopoietic cell culture, Annu. Rev. Biomed. Eng.
1 (1999) 129–52.
[15] U. Marx, H. Matthes, A. Nagel, R. Baehr, Application of a hollow fiber membrane
cell culture system in medicine, Am. Biotechnol. Lab. (1993) 26.
[16] A. Ouyang, S. T. Yang, A two-stage perfusion fibrous bed bioreactor system for
mass production of embryonic stem cells., Expert Opin. Biol. Ther. 8 (2008) 895–
909.
[17] H. Ye, Z. Xia, D.J.P. Ferguson, J.T. Triffitt, Z. Cui, Studies on the use of hollow
fibre membrane bioreactors for tissue generation by using rat bone marrow
fibroblastic cells and a composite scaffold., J. Mater. Sci. Mater. Med. 18 (2007)
641–8.
[18] H. Evenblij, J.H. van der Graaf, Occurrence of EPS in activated sludge from a
membrane bioreactor treating municipal wastewater, Water Sci. Technol. 50 (2004)
293–300.
[19] N. Park, B. Kwon, I.S. Kim, J.W. Cho, Biofouling potential of various NF
membranes with respect to bacteria and their soluble microbial products (SMP):
characterizations, flux decline and transport parameters, J. Membr. Sci. 258 (2005)
43–54.
[20] A. Bjørkøy, L. Fiksdal, Characterization of biofouling on hollow fiber membranes
using confocal laser scanning microcscopy and image analysis, Desalination. 245
(2009) 474–484.
[21] I. Chang, P. Le Clech, B. Jefferson, S. Judd, Membrane Fouling in Membrane
Bioreactors for Wastewater Treatment, (2002) 1018–1029.
[22] M. Ellis, J. Chaudhuri, Poly (lactic-co-glycolic acid) hollow fibre membranes for
use as a tissue engineering scaffold, Biotechnol. Bioeng. 96 (2007) 177–187.
107
[23]
G.D. Antwiler, Improved bioreactor surfaces, (2008) WO2008124229 A2
[24] S.C. Balmert, D. McKeel, F. Triolo, B. Gridelli, K. Zeilinger, R. Bornemann,J.C.
Gerlach, Perfusion circuit concepts for hollow-fiber bioreactors used as in vitro
cell production systems or ex vivo bioartificial organs, Int J Artif Organs. 34 (2011)
410–421.
[25] G. Salzman, R. Tadmor, S. Guzy, S. Sideman, N. Lotan, Hollow fiber enzymic
reactors for a two substrate process: analytical modeling and numerical simulations,
Chem. Eng. Process. Process Intensif. 38 (1999) 289–299.
[26] H. Ye, D.B. Das, J.T. Triffitt, Z. Cui, Modelling nutrient transport in hollow fibre
membrane bioreactors for growing three-dimensional bone tissue, J. Memb. Sci.
272 (2006) 169–178.
[27] D.S. Wavhal, E.R. Fisher, Membrane surface modification by plasma-induced
Polymerization of Acrylamide for Improved Surface Properties and Reduced
Protein Fouling, Langmuir. 19 (2003) 79–85.
[28] H.Y. Yu, H. Lei, Z.K. Xu, Enhencement of the flux for polypropylene hollow fiber
membrane in a submerged membrane-bioreactor by surface modification, J.
Environ. Sci. (China). 18 (2006) 1050–1055.
[29] A. Bhattacharya, B.N. Misra, Grafting: a versatile means to modify
polymersTechniques, factors and applications, Prog. Polym. Sci. 29 (2004) 767–
814.
[30] W. Sun, J. Liu, H. Chu, B. Dong, Pretreatment and Membrane Hydrophilic
Modification to Reduce Membrane Fouling, Membranes (Basel). 3 (2013) 226–
241.
[31] Y.J. Shen, G.X. Wu, Y.B. Fan, H. Zhong, L.L.Wu, S.L. Zhang, X.H. Zhao, W.J.
Zhang, Performances of biological aerated filter employing hollow fiber
membrane segments of surface-improved poly (sulfone) as biofilm carriers, J.
Environ. Sci. (China). 19 (2007) 811–7.
[32] M. Taniguchi, G. Belfort, Low protein fouling synthetic membranes by UVassisted surface grafting modification: varying monomer type, J. Memb. Sci. 231
(2004) 147–157.
108
[33] W. Junfu, Z. Kongyin, W. Lei, Z. Huan, C. Li, Z. Zhengpu, PVDF Hollow fiber
Membrane with High Flux and High Rejection Ratio Prepared by Irradiation
Induced Grafting of PAA, Macromol. Symp. 297 (2010) 231–239.
[34] Q.P. Pham, U. Sharma, A.G. Mikos, Electrospinning of polymeric nanofibers for
tissue engineering applications: a review., Tissue Eng. 12 (2006) 1197–211.
[35] C. Liu, R. Bai, Adsorptive removal of copper ions with highly porous
chitosan/cellulose acetate blend hollow fiber membranes, J. Memb. Sci. 284 (2006)
313–322.
[36] J. Pieracci, J. V Crivello, G. Belfort, Photochemical modification of 10 kDa
polyethersulfone ultrafltration membranes for reduction of biofouling, J. Memb.
Sci. 156 (1999) 223–240.
[37] H. Vihola, A. Laukkanen, L. Valtola, H. Tenhu, J. Hirvonen, Cytotoxicity of
thermosensitive polymers poly(N-isopropylacrylamide), poly(N-vinylcaprolactam)
and amphiphilically modified poly(N-vinylcaprolactam)., Biomaterials. 26 (2005)
3055–3064.
[38] Y. Luo, G. Engelmayr, D.T. Auguste, S. Ferreira, J.M. Karp, R. Saigal, R. Langer,
3-dimensional scaffolds, in Principles of tissue engineering, 3rd edition, R. Lanza,
R. Langer, J.P. Vacanti, Elsevier/Academic press, 2007, 359-373.
[39] M. Lovett, K. Lee, A. Edwards, D.L. Kaplan, Vascularization strategies for tissue
engineering., Tissue Eng. Part B. Rev. 15 (2009) 353–70.
[40] G.T. Grant, E.R. Morris, D.A. Rees, P.J.C. Smith, D. Thom, Biological
interactions between polysaccharides and divalent cations: The egg-box model,
FBES Lett. 32 (1973) 195–198.
[41] K.I. Draget, G. Skjåk-Braek, O. Smidsrød, Alginate based new materials, Int. J.
Biol. Macromol. 21 (1997) 47–55.
[42] O. Smidsrød, G. Skjåk-Braek, Alginate as immobilization matrix for cells, Trends
Biotechnol. 8 (1990) 71–8.
[43] Y.M. Kolambkar, K.M. Dupont, J.D. Boerckel, N. Huebsch, D.J. Mooney, D.W.
Hutmacher, R.E. Guldberg, An alginate-based hybrid system for growth factor
delivery in the functional repair of large bone defects, Biomaterials. 32 (2011) 65–
74.
109
[44] D. Pegg, Cryopreservation and freeze-drying protocols, in: J. Day, G. Stacey
(Eds.), Methods in Molecular Biology (Humana Press), 2007: pp. 39–57.
[45] K.Y. Lee, D.J. Mooney, Hydrogels for Tissue Engineering, Chem. Rev. 101 (2001)
1869–1880.
[46] J. Kopecek, Hydrogel biomaterials: a smart future?, Biomaterials. 28 (2007) 5185–
5192.
[47] A.C.F. Ribeiro, M.C.F. Barros, A.S.N. Teles, A.J.M. Valente, V.M.M. Lobo,
A.J.F.N. Sobral, M.A. Esteso, Diffusion coefficients and electrical conductivities
for calcium chloride aqueous solutions at 298.15K and 310.15K, Electrochim.
Acta. 54 (2008) 192–196.
[48] T.R. Cuadros, O. Skurtys, J.M. Aguilera, Mechanical properties of calcium
alginate fibers produced with a microfluidic device, Carbohydr. Polym. 89 (2012)
1198–1206.
[49] R. Hassan, A. Gobouri, I. Zaafarany, S.A. Kingdom, Kinetics and mechanism of
Sol-Gel transformation between sodium alginate anionic polyelectrolyte and some
alkaline earth metal ions with formation of coordination biopolymer ionotropic
polymembrane hydrogels of capillary structure, ABB (2013) 47-56.
[50] M. Mahmoud, R. Abdel-Kader, M. Hassanein, S. Saleh, S. Botros, Antipyrine
clearance in comparison to conventional liver function tests in hepatitis C virus
patients, Eur. J. Pharmacol. 569 (2007) 222–7.
[51] R. Hassan, F.Tirkistani, I. Zaafarany, A. Fawzy, M. Khairy, S. Iqbal, Polymeric
biomaterial hydrogels. I. Behavior of some ionotropic cross-linked metal-alginate
hydrogels especially copper-alginate membranes in some organic solvents and
buffer solutions, ABB 3 (2012) 845-854.
[52] J. Kestin, M. Sokolov, W. A. Wakeham, Viscosity of liquid water in the range
−8 °C to 150 °C, J. Phys. Chem. Ref. Data. 7 (1978) 941.
[53] I. Sawada, R. Fachrul, T. Ito, Y. Ohmukai, T. Maruyama, H. Matsuyama,
Development of a hydrophilic polymer membrane containing silver nanoparticles
with both organic antifouling and antibacterial properties, J. Memb. Sci. 387-388
(2012) 1–6.
110
Chapter 5
Fabrication of alginate fibers using a microporous membrane
based molding technique 4
4
This chapter is organized based on a manuscript written by Seung Mi Yoo and Raja Ghosh. This work has
been submitted and under the minor revision process to Biochemical Engineering Journals (Manuscript
#BEJ-D-14-00034)
111
5.1
Abstract
Tubular structures made of alginate and similar hydrogels have been extensively
investigated for drug delivery, tissue engineering and other biomedical applications.
Alginate fibers are usually fabricated using complicated techniques such as coaxial flow
extrusion or electro-spinning. This paper discusses the fabrication of hollow and solid
alginate fibers using a simple membrane based molding technique. A hollow-fiber
microfiltration membrane served both as the mold as well as reservoir for the crosslinking agent. The pores of the membrane were first filled with the cross-linker (i.e.
calcium chloride) solution, after which the lumen was filled with sodium alginate solution.
The calcium ions diffused from the membrane pores into the lumen, cross-linking the
alginate in a radially inward direction. Solid alginate fibers were obtained by crosslinking all the alginate within the lumen, while hollow fibers were fabricated by pushing
out un-cross-linked alginate from the central core using calcium chloride solution. The
alginate fibers fabricated as described above were expelled from the membrane mold
using water under pressure. These were then characterized by optical, scanning electron,
and transmission electron microscopy. The most attractive features of this fabrication
method are its simplicity and flexibility with regards to alginate concentration. The fibers
obtained were straight with excellent definition and uniform thickness, and could be
fabricated in a highly reproducible manner. Other complicated solid and hollow 3dimensional structures suitable of biomedical, and indeed other applications could also
potentially be fabricated using similar membrane-based molding techniques.
112
5.2
Introduction
Hydrogels can hold large amounts of water within their structures and are widely used for
a range of pharmaceutical and biomedical applications [1,2]. Synthetic hydrogels such as
cross-linked polyethylene glycol (PEG), poly (N-isopropylacrylamide) (PNIPAM) and
poly-hydroxyethylmethacrylate (PHEMA) can be tailored to specific chain lengths and
degree of cross-linking, and can also be functionalized to make them stimuli-responsive,
an attribute extremely useful for controlled or sustained release of drugs [3]. However,
un-polymerized monomer residues can be highly toxic, and their complete removal can
prove to be challenging. Consequently, natural hydrogels continue to be widely used for
drug delivery, cell culture and tissue engineering applications, where toxicity due to
monomers and cross-linkers could be a major issue [4]. Natural hydrogels include
biopolymers such as alginate, collagen, hyaluronic acid (HA), fibrin, agarose, chitosan,
and combinations of two or more of these [5]. Alginate, which is obtained from brown
algae is the most popular amongst these, having been widely used since the 1980s, due to
its low immunogenicity and low toxicity, for cell encapsulation, drug delivery, woundhealing, and tissue engineering [6-8]. It is a polysaccharide with (1-4)-linked β-Dmannuronate (M) and α-L guluronic acid (G) residues, with the sequence of M and G
residues varying, depending on the source. The gelation of alginate is induced by the
presence of divalent cations such as Ba2+, Sr2+, and Ca2+ which interact with the G
residues [6, 7]. Due to the involvement of the G residue in cross-linking, the G/M ratio
affects the mechanical strength and biodegradability of alginate [7, 8]. The formation of
calcium alginate gel has been described based on an “egg-box model” [8].
113
Spherical calcium alginate beads have been widely used for cell encapsulation and drug
delivery due to the simple fabrication method, i.e. drop-wise addition of sodium alginate
solution into cross-linker (calcium chloride) solution [7]. The material to be encapsulated
is typically pre-mixed with the sodium alginate solution prior to cross-linking. Some of
the drawbacks linked with spherical alginate beads and their fabrication method are a)
loss of cells or drug into the cross-linking solution, b) the so called “burst-effect”
whereby there is an initial rapid release of drug from the beads to any recipient solution,
and c) nutrient mass transport limitations at the centre of the beads which could lead to
cell necrosis [9-12]. The drop-wise bead production methods also leads to size
distributions, and this makes precise drug dosing difficult. A critical limiting factor for
the spherical geometry is that the surface area to volume ratio is determined by the radius,
thereby limiting flexibility in designing structures. Alternative geometries such as tubes,
rods and cylinders have been shown to have some advantages in certain drug delivery and
tissue engineering applications [13]. With the cylindrical form, dosage can be very
precisely controlled based on length and diameter. Also, the surface area to volume ratio
is decoupled from the radius and can be flexibly manipulated by altering both the radius
and length of the structure. Moreover, in certain applications such as nerve and vascular
tissue engineering, the fiber is the only suitable scaffold geometry.
Electro-spinning, one of the well-established methods for fabricating fibers, has attracted
much attention due the ease with which fine fibers with diameters ranging from 10 nm to
100 micrometers can be produced using both synthetic and natural hydrogel materials
114
[14-20]. However, this technique involves the application of high voltages, e.g. 10~15
kV[17], 20 kV [18], 13~15 kV [19], and 10~50 kV [20], which could potentially lead to
cell mortality in cell immobilization and tissue engineering applications. During cellembedded fiber fabrication by electro-spinning, cell death was as high as 60 % [21].
Moreover, electro-spinning is not particularly suitable for fabrication of fibres with outer
diameter greater than 100 micrometer [17, 19, 22]. Another major problem with electrospinning of alginate is that the method is restricted to the use of alginate solutions of
lower than 2% concentration due to viscosity issues, and consequently fibers with low
mechanical strengths are produced[22].
Extrusion based technique have also been widely used to produce alginate fibers. For
instance, Li et al [11] reported a very simple technique for producing alginate cylinders
by directly injecting sodium alginate into calcium chloride solution. Improved versions of
this technique involving sophisticated multichannel nozzles, suitable for fabricating more
complicated structures have been subsequently reported [23-25]. These techniques were
suitable for producing 100~800 μm diameters fibres by manipulating variables such as
reagent concentration and extrusion velocity, and by using sheath solutions [26]. The
combination of extrusion and microfluidic techniques have been shown to enhance the
consistency of fiber production [27-29]. However higher concentrate alginate solutions
cannot be satisfactorily handled by microfluidic systems as their high viscosity makes it
difficult to balance the central flow and sheath flows, which in turn leads to nonuniformity in fiber dimension. Moreover the high shear stress observed in such system
115
could be problematic in cell immobilization and tissue engineering applications. In an
effort to reduce shear stress, a roller assisted microfluidic system was developed whereby
alginate fibres produced within microfluidic channels were drawn out in a controlled
manner [30]. Other techniques reported for fabrication of alginate fibers include 3D
printing [31], laser-assisted printing [32] and the use of centrifugal force [33]. Boland et
al. [34] have reviewed the use of 3D printing in tissue engineering including the
fabrication of alginate structures.
In this paper, we describe a molding technique, using which hollow and solid alginate
fibers can be fabricated under gentle and sterile conditions. A hollow-fiber micro- or
ultra-filtration membrane is used as the mold for fabricating these alginate fibers. The
pores present on the hollow fiber membrane served as the reservoir for calcium chloride
solution was used to cross-link alginate within the fiber lumen. The method is
summarized in Figure 1. As shown in A, the pores of the hollow fiber were first filled
with calcium chloride solution. The lumen of the hollow fiber was then filled with
sodium alginate solution (B). The divalent calcium ions diffused from the pores into the
lumen, cross-linking the alginate in a radially inward direction (C). For producing solid
fibers, the alginate held within the lumen was allowed to fully cross-link before expelling
it from the hollow-fiber membrane. For producing alginate hollow fibers, the crosslinking was allowed to take place till the desired wall thickness was attained followed by
removal of uncross-linked alginate from the interior core using calcium chloride solution
(D). This step also resulted in the curing of the inner surface of the alginate hollow fiber,
116
as well as delamination of alginate fiber from substrate membrane by dehydration which
was then recovered by expulsion from the mold under pressure using water or calcium
chloride solution (Figure 1 E). Alginate fibers of different sizes and geometries were
fabricated using the above technique. Effects of variables such as alginate concentration,
cross-linker concentration and cross-linking time on the fiber dimension were examined
to demonstrate the controllability of the fabrication technique. The alginate fibers
produced were tested using techniques such as scanning electron microscopy (SEM),
tunneling electron microscopy (TEM), and light and fluorescence microscopy. The
feasibility of using the molding techniques for tissue engineering and cell immobilization
applications was demonstrated by fabricating alginate hollow fibers containing live
human umbilical vein endothelial cells (HUVEC).
117
Figure 1 The method for fabricating calcium alginate hollow fibers using a porous
hollow fiber membrane as a mold (A) cross sectional view of hollow fiber membrane
with pores filled with calcium chloride solution, (B) sodium alginate solution being filled
inside the lumen of the hollow fiber membrane, (C) cross-linking of alginate by radially
inward diffusion of calcium ions, (D) removal of un-cross-linked alginate core with
calcium chloride solution, and (E) expulsion of calcium alginate hollow fiber from mold.
5.3
Materials and Methods
Sodium alginate (W201502), and FITC-dextran (46947; excitation 490 nm, emission 520
nm) were purchased from Sigma-Aldrich, St. Louis, MO, USA. Calcium chloride (C77500) was obtained from VWR, Mississauga, ON, Canada. Polypropylene hollow fiber
microfiltration membranes (Accurel PP S6/2, 1.8 mm i.d., 2 mm o.d., 0.2 μm pore size;
Accurel Q3/2, 0.6 mm i.d., 1 mm o.d., 0.2 micron pore size) were purchased from
Membrana, Wuppertal, Germany. Hydrophilic polyethersulfone hollow fiber
118
ultrafiltration membranes (Tetronic, 0.8 mm i.d., 1.2 mm o.d., 150 kDa MWCO) was
kindly donated by Hydranautics Inc., Japan. All test solutions and reagents were prepared
using ultra-pure water (18.2 mΩ-cm) obtained from a Diamond Nanopure water
purification unit (Barnstead International, Dubuque, IA, USA), vacuum-filtered just prior
to use using 0.45 μm pore size cellulose acetate membrane (Nalgene Nunc International,
Rochester, NY, USA).
5.3.1 Fabrication of calcium alginate hollow fibers
Figure 2
Sequence of events in calcium alginate hollow fiber molding process (A)
sweating of calcium chloride on the outer wall of the hollow fiber membrane indicating
that the pores are filled, (B) removal of un-cross-linked sodium alginate core from the
mold using calcium chloride solution, (C) expulsion of calcium alginate hollow fiber
from mold using water, (D) alginate hollow fiber after recovery from mold.
119
Sodium alginate solutions of different concentration (e.g. 1, 2.5, 5 wt %) were prepared in
ultrapure water. The solutions were allowed to stand for 24 hours at 4°C before use in
order to evenly disperse the polymer and to remove any entrapped air bubbles. Calcium
chloride (cross-linker) solutions of different concentrations (e.g. 0.1, 0.5, 1 M) were also
prepared in ultrapure water. All the materials used for alginate fiber fabrication were
sterilized and transferred to a sterile laminar flow cabinet. Figure 2 shows some of the
steps involved in the fabrication process. The lumen of the hollow fiber membrane was
first filled with calcium chloride under pressure to fill up the pores present on the
membrane. Sweating on the outer surface of the membrane (see A) confirmed that the
membrane pores were filled with cross-linker. The calcium chloride solution was emptied
from the lumen, the external surface was patted-dry and air was passed through the lumen
to remove excess calcium chloride solution from the inner surface. This was done to
prevent irregular cross-linking of alginate during the fabrication process. The lumen was
then filled with sodium alginate solution which was maintained inside for the required
duration in a given experiment. For preparing alginate hollow fibers, uncross-linked
material present in the core was pushed out using calcium chloride solution (see B). The
fiber fabricated within the hollow fiber membrane was expelled using water under
pressure (see C). The filling, rinsing and pressurising steps described above were carried
out using sterile plastic syringes.
120
5.3.2 Microscopy
Alginate fiber samples for testing by TEM and SEM were fixed with 2% Glutaraldehyde
(2% v/v) in 0.1 M sodium cacodylate buffer pH 7.4. The samples were then rinsed two
times in buffer and post-fixed in 1 % osmium tetroxide in 0.1 M sodium cacodylate
buffer for 1 hour. The samples were then dehydrated using graded ethanol series (i.e.
50%, 70%, 70%, 95%, 95%, and 100%). The final dehydration for TEM samples was
done in 100% propylene glycol (PPG). Infiltration with Spurr's resin was carried out
using graded series (i.e. PPG:Spurr’s resin ratios of 2:1, 1:1, 1:2, and lastly100% Spurr's
resin) with rotation of the samples in between solution changes. The samples were
transferred to embedding moulds which were then filled with fresh 100% Spurr's resin
and polymerized overnight in a 60 °C oven. Thin sections were cut using a Leica UCT
Ultramicrotome and picked up onto Cu grids. The sections were post-stained with uranyl
acetate and lead citrate and then viewed in a JEOL JEM 1200 EX TEMSCAN
transmission electron microscope (JEOL, Peabody, MA, USA) operated at an
accelerating voltage of 80kV. The samples for SEM were dehydrated in 100% ethanol
and then critical point dried. These were then mounted onto SEM stubs, sputter-coated
with gold and then viewed in a Tescan Vega II LSU scanning electron microscope
(Tescan USA, PA) operated at 20kV. A florescence inverted microscope- Zeiss Axiovert
100M microscope was used for confocal laser scanning microscopy of alginate fibers. All
light microscopy images were acquired with a Zeiss AxioCam HRc camera. AxioVision
software was used for transferring and analyzing images.
121
5.3.3 Human umbilical vein endothelial cell culture
Human umbilical vein endothelial cells or HUVEC (C-003-5C) were purchased from
Gibco Life Technology (Burlington ON, Canada). HUVEC was cultured by
supplementing low serum growth factors (EGM-2, CC-4176) in endothelial cell basal
medium containing 1% v/v penicillin-streptomycin (P/S) (Cat #15070063). The cells
were cultured at 37 °C in 5 % carbon dioxide and 100 % relative humidity. The media
was changed every second day. Once the cells attained 100 percent confluence, cells were
washed with 12 mL of 0.25% Trypsin/EDTA (Invitrogen, 12604021) for 3 min to detach
the cells from the inner surface of the T-75 culture flask used for cell culture. The cell
suspension thus obtained was transferred to a sterile 15 mL falcon tube and centrifuged at
14,000 rpm for 5 min. The cell pellet which had approximately 9 × 106 cells/mL was
dispersed in sterile sodium alginate solution to fabricate cell-containing alginate fibers.
122
5.4
Results and Discussions
Figure 3
Alginate fibers fabricated using hollow fiber membrane based molding
technique: (A) Fibers made with 2.5% sodium alginate, 0.1 M calcium chloride solution
as cross-linker using (from the top) Accurel Q3/2, Tetronic and Accurel PP S6/2 hollow
fiber membrane molds respectively. The cross-linking times were 10, 30 and 60 seconds
respectively for the small, medium and large diameter hollow fibers. (B)Solid and (C)
hollow fibers fabricated with 5% sodium alginate, 1 M calcium chloride solution as
cross-linker, using Accurel PP S6/2 membrane as mold. The cross-linking times were 600
seconds and 60 seconds respectively.
Figure 3 shows some of the solid and hollow alginate fibers fabricated using different
hollow fiber membranes as molds, using different reagents and cross-linking time. The
hollow fibers shown in Figure 3A were made using 2.5% sodium alginate solution, and
0.1 M calcium chloride solution as cross-linker. The cross-linking times were 10, 30 and
60 seconds respectively for the small, medium and large diameter hollow fibers. The
123
small diameter hollow fiber (200 µm o.d., 50 µm i.d.) was molded using Accurel Q3/2
membrane, the medium diameter hollow fiber (600 µm o.d., 375 µm i.d.) using Tetronic
membrane, and the large diameter hollow fiber (1.6 mm o.d., 0.6 mm i.d.) using Accurel
PP S6/2 membrane. The solid and hollow fibers shown in Figures 3B and 3C respectively
were fabricated using 5% sodium alginate with 1 M calcium chloride solution as crosslinker, using Accurel PP S6/2 membrane as mold. The only difference between the two
was with regards to cross-linking time. While the solid fibers were obtained by a 600
second cross-linking process, the hollow fibers were obtained by 60 seconds of crosslinking. As expected, the outer diameter of the alginate fiber was dependent on the inner
diameter of the hollow fiber membrane mold. Also, for a given mold, the inner diameter
could be controlled through the cross-linking time and hollow or solid fibers, as required
could be produced. In addition to the differences in dimension, these fibers also differed
significantly in terms of their rigidity. The fibers produced using 2.5% sodium alginate
solution was softer than those produced using 5% alginate. The fibers could be produced
in a highly reproducible manner, simply by controlling the reagent concentrations and
cross-linking time. The individual solid fibers shown in Figure 3B were separately
fabricated and were found to have fairly consistent outer diameter of 1.35 ± 0.08 mm
(n=8). Similarly, the hollow fibers shown in Figure 3C were separately fabricated and
found to have consistent outer diameter of 1.42± 0.06 mm (n=8) and inner diameter of
0.65± 0.004 mm (n=8) mm. The slightly lower outer diameter of the solid fiber suggests
that the greater extent of cross-linking caused some inward shrinkage of the structure.
124
Figure 4
Magnified side view (A) and cross-sectional view (B) images of the
alginate hollow fiber shown in Figure 3 C. Images were obtained with the fiber immersed
in water by light microscopy with a magnification factor of 25.
Figure 4 shows magnified cross-section and side view images of the alginate hollow fiber
shown in Figure 3 C. These images were obtained under immersed condition (in water)
by light microscopy with a magnification factor of 25. The outer diameter of the fiber
was 1.54 mm and the inner diameter was 0.64 mm. The slightly higher outer diameter
(with respected to that reported for Figure 3 C in the previous paragraph) was either due
to optical aberration or due to fiber expansion due to hydration. Fibers obtained by
spinning and extrusion techniques generally have bends, twists, turns, and other types of
imperfections. This is usually due to the flow processes involved in these techniques.
Figures 3 and 4 clearly show that alginate fibers obtained using the molding technique
had excellent definition and uniformity of appearance, and were straight. The inner and
outer surfaces were smooth and the fibre shape was perfectly circular. The inner and
outer diameter, and therefore the thickness of the fiber were uniform along the length of
the fiber. A particularly attractive feature of this molding technique is that the alginate
125
fibers can be fabricated under sterile conditions without too much complication.
Moreover, the fabrication was carried out under low pressure and stress conditions
without the need for electric current. Also, the data shown in the previous paragraph
clearly show the high degree to reproducibility with which the fibers could be molded.
During alginate fiber fabrication by extrusion, the nascent fiber remains exposed to a
large body of liquid and this increases the possibility of leakage of material being
immobilized, i.e. cell or drugs. By contrast, during fiber fabrication using the hollow fiber
membrane based molding technique no such exposure to bulk liquid takes place and
consequently the loss of material are expected to be significantly lower. All these
attributes would be particularly appealing in biomedical applications such as fabrication
of inserts, soft delivery and connecting tubes, tissue engineering scaffolds and drugdelivery devices. The main disadvantage of the molding technique in its current form is
that it is not suitable to producing large quantities of fibers. Also, the maximum length of
the fiber is restricted to the length of the mold, i.e. the hollow fiber membrane.
126
Figure 5
Confocal laser scanning (CLS) microscopy images of the FITC-dextran alginate composite hollow fiber, (A) z-stack images obtained along the Z-axis at
sequential depths of 200μm, (B) five times magnification, and (C) twenty-time
magnification
For imaging the interior of the fiber, FITC-dextran was added to the alginate solution
prior to fiber fabrication. The sodium alginate concentration used in this experiment was
2.5 %, the FITC-dextran concentration was 2 mg/mL, the calcium chloride concentration
was 1 M, an Accurel PP S6/2 membrane was used as mold, and the cross-linking was
carried out for 60 seconds. The alginate hollow fiber thus produced was physically stiffer
and more opaque than the fibers produced using sodium alginate alone. Most likely,
dextran reinforced the structure of the alginate fiber and made it stronger. Figure 5 shows
the images of the composite alginate fiber obtained by confocal laser scanning
127
microscopy. Figure 5A shows the z-stack images obtained along the axis at sequential
depths of 200μm while B and C show respectively five-times and twenty-time magnified
images of the same. The addition of the fluorescent additive facilitated a better view of
the interior of the alginate fiber. The fiber had uniform circular cross-sectional area, with
consistent in inner and outer diameter along the axis as evident from the z-stack images.
The inner and outer surfaces of the fiber were smooth and the fluorescent additive was
more or less uniformly distributed, except for a circular band of low fluorescent intensity
close to the outer surface, and some intense fluorescent patches which were clearly
visible under high magnification.
Figure 6
Scanning electron microscopy (SEM) images of calcium alginate hollow
fiber (400 X magnifications) fabricated with 2.5 % sodium alginate and 1 M calcium
chloride, using Accurel PP S6/2 membrane as mold, with a cross-linking time of 60
seconds.
128
Figure 7
TEM image obtained with the hollow fiber described in Figure 6 at (A)
5000 X magnification, and (B) 75000X magnification
Figure 6 shows scanning electron microscopy (SEM) images of calcium alginate hollow
fibers (at 400 X magnifications) fabricated using 2.5 % sodium alginate, 1 M calcium
chloride solution, Accurel PP S6/2 membrane as mold, and a cross-linking time of 60
seconds. The outer diameter of the hollow fiber under imaging condition was 560 μm
while the wall thickness was 140 μm. Consistent with the earlier imaging techniques, the
inner curved surface of the fiber was smooth and of perfectly circular curvature. The
distortion on the outer surface was caused by the tweezers used to handle and process the
fiber for imaging. The bulk material comprising the fiber appears to be homogeneous, but
a skin-like structure is observed on the inner surface. Figure 7 shows the TEM image
obtained with the same hollow fiber at two different magnifications (i.e. 5000X and
129
75000X respectively for A and B). The sponge-like porous structure of calcium alginate
is evident at both levels of magnification.
Figure 8
Effect of sodium alginate concentration on dimension of alginate fiber
(mold: Accurel PP S6/2, calcium chloride concentration: 1 M; cross-linking time: 60
seconds)
Figure 8 shows the effect of sodium alginate concentration on alginate hollow fiber
dimension. These experiments were carried out using Accurel PP S6/2 membrane as
mold with 1 M calcium chloride solution as cross-linker, the molding time being 60
seconds. While the dimension of the alginate hollow fiber did not show much
responsiveness to sodium alginate concentration, there were significant differences in the
strength and appearance of material obtained. The fiber obtained with 1 % alginate (data
not shown in Figure 8) was quite fragile and difficult to measure due to its highly flexible
130
nature. The fiber obtained with 2.5% alginate was much more defined but the external
surface was much rougher that that obtained using 5% sodium alginate solution. The
strongest and most well defined fiber was obtained with 5% alginate. These fibers was
also more opaque and of higher stiffness. Their slightly lower outer diameter was
possibly due to dehydration during the gel formation, typically observed during crosslinking of highly concentrated polymer solutions [29].
Figure 9
Effect of calcium chloride concentration on dimension of alginate fiber
(mold: Accurel PP S6/2, sodium alginate concentration: 5%, cross-linking time: 60
seconds)
Figure 9 shows the effect of calcium chloride concentration on the dimension of the
alginate fibers produced with 5% sodium alginate and 60 seconds of cross-linking time,
using Accurel PP S6/2 membrane as mold. The cross-linker concentration had a far more
131
significant effect on the fiber dimension than the alginate concentration. The outer
diameter increased very significantly when the calcium chloride concentration was
decreased from 1 M to 0.5 M, and increased even further when the concentration was
decreased to 0.1 M. The inner diameter also increased when the cross-linker
concentration was decreased. However, the thickness increased with increase in crosslinker concentration. While the increase in hollow fiber wall thickness with calcium
chloride concentration was due to greater cross-linker penetration into sodium alginate
driven by greater concentration gradient, the effects on the inner and outer diameter are
harder to explain. A plausible explanation is that the higher calcium chloride
concentration caused shrinkage of the alginate structure by dehydration, leading to fibers
of lower outer and inner diameters.
132
Figure 10
Effect of cross-linking time on dimensions of alginate fiber (mold:
Accurel PP S6/2, sodium alginate concentration: 5%, calcium chloride concentration: 1
M)
Figure 10 shows the effect of cross-linking time on the dimension of alginate fibers
obtained using Accurel PP S6/2 membrane as mold with 5 % sodium alginate, and 1 M
calcium chloride as cross-linker. While the outer diameter of the alginate fiber did not
change appreciably; the inner diameter did decrease with increase in cross-linking time
due to greater penetration of calcium ion within sodium alginate held within the mold.
While the difference in inner diameter between 30 and 60 second cross-linking was not
very significant, solid fiber was obtained by 600 seconds of cross-linking.
133
Figure11
Light microscopy image of HUVEC embedded alginate hollow fiber
(image obtained one day after fabrication, magnification: 100 X, sodium alginate
concentration: 1%, mold: Tetronic, calcium chloride concentration: 0.1 M, cross-linking
time: 30 seconds)
The feasibility of immobilizing cells was demonstrated by fabricating HUVEC embedded
calcium alginate hollow fibers under sterile condition. The fabrication process was
similar to that described in the previous paragraphs. Sodium alginate solution (2%)
prepared in culture medium was mixed in 1:1 ratio with cells suspension. The mixture
thus obtained was found to be less viscous than 1% sodium alginate solution prepared in
water. This was possibly due to dilution caused by water released from cells due to
osmotic effect. Tetronic hollow fiber membrane (0.8 mm i.d., 1.2 mm o.d.) was used as
mold and cross-linking was carried out for 30 seconds using 0.1 M calcium chloride.
Figure 11 shows the HUVEC embedded alginate hollow fiber obtained, the image being
taken one day after fabrication by light microscopy using 100 X magnification. The cells
134
were uniformly dispersed and appeared to be viable. The cells appear bright in the center
due to reflected light. Image analysis indicated that the density of immobilized cells was
4 × 106 cells /mL. The tissue or cell mass was not succeed during one week incubation
period, HUVECs was not sufficient choice for immobilized system. Relatively HUVECs
are sensitive on culture environment compared to fibroblast or neural cell line. Also the
matrix composition can be stiff for the cell migration. For the future reference, the
feasibility test can be conducted with microorganism or mammalian cells.
The above results clearly demonstrate the suitability of the hollow fiber membrane based
molding method for fabricating very well defined hollow or solid alginate fibers. The
method as discussed is not suitable for large-scale production, but may be developed
further for continuous extrusion of molded fibers. The technique as discussed in this
paper will be particularly attractive in applications where small-scale precision fiber
fabrication under sterile and gentle conditions is required.
5.5
Conclusion
Calcium alginate fibers, both solid and hollow, could be fabricated by using a hollow
fiber membrane based molding technique. Cross-linking agent (calcium chloride) held
within the pores of the membrane diffused into the sodium alginate held within the
membrane, cross-linking the alginate in an inward direction. The outer diameter of these
fibers depended primarily on the inner diameter of the mold while the inner diameter
135
could be controlled based on the cross-linking time. Prolonging the cross-linking time
resulted in the formation of solid alginate fibers. Factors such as alginate concentration
and calcium chloride concentration were found to affect the dimension and physical
properties of the fabricated fibers. The inner and outer surfaces of the hollow fibers were
very well defined and the fibres were perfectly circular. The inner and outer diameter,
and therefore the thickness of the fiber were uniform along the length of the fiber which
were free from imperfections such as bends, twists, and turns and could be fabricated
with high degree of reproducibility. The fibers could be fabricated under sterile condition
and the technique did not involve the use of high pressure, high shear stress or electric
current. The molding technique is therefore particularly appealing for producing inserts
and soft tubes for biomedical applications, tissue engineering scaffolds and drug-delivery
devices. However, the technique in its current form it is not suitable to producing large
quantities of fibers and the maximum length of the fiber is restricted to the length of the
mold.
136
5.6
References
[1]
A.S. Hoffman, A. Afrassiabi, L.C. Dong, Thermally reversible hydrogels: II.
Delivery and selctive removal of substances from aquesous solutions, J. Control.
Release. 4 (1986) 213-222.
[2]
G. Skjåk-Braek, Alginates: biosynthesis and some structure-function relationships
relevant to biomedical and biotechnological applications, Biochem. Soc. T. 20
(1992) 27–33.
[3]
B. V. Slaughter, S.S. Khurshid, O.Z. Fisher, A. Khademhosseini, N.A. Peppas,
Hydrogels in regenerative medicine, Adv. Mater. 21 (2009) 3307–3329.
[4]
S.J. Bryant, C.R. Nuttelman, K.S. Anseth, Cytocompatibility of UV and visible
light photoinitiating systems on cultured NIH / 3T3 fibroblasts in vitro, J. Biomater.
Sci. 11 (2000) 439–457.
[5]
K.Y. Lee, D.J. Mooney, Hydrogels for Tissue Engineering, Chem. Rev. 101 (2001)
1869–1880.
[6]
K.I. Draget, G. Skjåk-Braek, O. Smidsrød, Alginate based new materials, Int. J.
Biol. Macromol. 21 (1997) 47–55.
[7]
O. Smidsrød, G. Skjåk-Braek, Alginate as immobilization matrix for cells, Trends
Biotechnol. 8 (1990) 71–78.
[8]
G.T. Grant, E.R. Morris, D.A. Rees, P.J.C. Smith, D. Thom, Biological
interactions between polysaccharides and divalent cations: The egg-box model,
FBES Lett. 32 (1973) 195–198.
[9]
Y. Yeo, K. Park, Control of encapsulation efficiency and initial burst in polymeric
microparticle systems, Arch. Pharm. Res. 27 (2004) 1–12.
[10] X. Huang, C.S. Brazel, On the importance and mechanisms of burst release in
matrix-controlled drug delivery systems, J. Control. Release. 73 (2001) 121–136.
[11] R.H. Li, D.H. Altreuter, F.T. Gentile, Transport characterization of hydrogel
matrices for cell encapsulation, Biotechnol. Bioeng. 50 (1996) 365–373.
[12] R. Zarzycki, Z. Modrzejewska, K. Nawrotek, Drug release from hydrogel matrices,
Eco.Chem. Eng. 17 (2010) 117–136.
[13] N. Doshi, B. Prabhakarpandian, A. Rea-Ramsey, K. Pant, S. Sundaram, S.
Mitragotri, Flow and adhesion of drug carriers in blood vessels depend on their
137
shape: a study using model synthetic microvascular networks, J. Control. Release.
146 (2010) 196–202.
[14] Y. Dzenis, Material science. Spinning continuous fibers for nanotechnology,
Science. 304 (2004) 1917–1919.
[15] D.H. Reneker, I. Chun, Nanometre diameter fibres of polymer, produced by
electrospinning, Nanotechnology. 7 (1996) 216–223.
[16] M. Bognitzki, W. Czado, T. Frese, A.Schaper, M. Hellwig, M. Steinhart, A.
Greiner, J.H. Wendorff, Nanostructured fibers via electrospinning, Adv. Mater. 13
(2001) 70–72.
[17] C.A. Bonino, M.D. Krebs, C.D. Saquing, S.I. Jeong, K.L. Shearer, E. Alsberg, S.A.
Khan, Electrospinning alginate-based nanofibers: From blends to crosslinked low
molecular weight alginate-only systems, Carbohydr. Polym. 85 (2011) 111–119.
[18] D. Fang, Y. Liu, S. Jiang, J. Nie, G. Ma, Effect of intermolecular interaction on
electrospinning of sodium alginate, Carbohydr. Polym. 85 (2011) 276–279.
[19] J.J. Chang, Y.H. Lee, M.H. Wu, M.C. Yang, C.T. Chien, Preparation of
electrospun alginate fibers with chitosan sheath, Carbohydr. Polym. 87 (2012)
2357–2361.
[20] L. S. Barbarash, J. L. Elgudin, V. V. Sevostyanova, A. S.Golovkin, Tissueengineered vascular graft and its fabrication approach (2013) WO 2013151463 A2
[21] N. Traitcheva, H. Berg, Electroporation and alternating current cause membrane
permeation of photodynamic cytotoxins yielding necrosis and apoptosis of cancer
cells, Bioelectrochemistry. 79 (2010) 257–260.
[22] N. Bhattarai, Z. Li, D. Edmondson, M. Zhang, Alginate-based nanofibrous
scaffolds: structural, mechanical, and biological properties, Adv. Mater. 18 (2006)
1463–1467.
[23] T. Takei, S. Sakai, H. Ijima, K. Kawakami, Development of mammalian cellenclosing calcium-alginate hydrogel fibers in a co-flowing stream, Biotechnol. J. 1
(2006) 1014–1017.
[24] T. Takei, S. Yamaguchi, S. Sakai, H. Ijima, K. Kawakami, Novel technique for
fabricating double-layered tubular constructs consisting of two vascular cell types
in collagen gels used as templates for three-dimensional tissues, J. Biosci. Bioeng.
104 (2007) 435–438.
138
[25] T. Takei, N. Kishihara, S. Sakai, K. Kawakami, Novel technique to control inner
and outer diameter of calcium-alginate hydrogel hollow microfibers, and
immobilization of mammalian cells, Biochem. Eng. J. 49 (2010) 143–147.
[26] M. Hu, R. Deng, K.M. Schumacher, M. Kurisawa, H. Ye, K. Purnamawati, J.Y.
Ying, Hydrodynamic spinning of hydrogel fibers, Biomaterials. 31 (2010) 863–
869.
[27] H. Onoe, T. Okitsu, A. Itou, M. Kato-negishi, R. Gojo, D. Kiriya, K. Sato, S.
Miura, S. Iwanaga, K. Kuribayashi-shigetomi, Y.T. Matsunaga, Y. Shimoyama, S.
Takeuchi, Metre-long cell-laden microfibers exhibit tissue morphologies and
functions, Nat. Mater. 12 (2013) 584–590.
[28] S.J. Shin, J.Y. Park, J.Y. Lee, H. Park, Y.D. Park, K.B. Lee, C.M. Whang, S.H.
Lee, “On the fly” continuous generation of alginate fibers using a microfluidic
device, Langmuir. 23 (2007) 9104–9108.
[29] T.R. Cuadros, O. Skurtys, J.M. Aguilera, Mechanical properties of calcium
alginate fibers produced with a microfluidic device, Carbohydr. Polym. 89 (2012)
1198–1206.
[30] J. Su, Y. Zheng, H. Wu, Generation of alginate microfibers with a roller-assisted
microfluidic system., Lab Chip. 9 (2009) 996–1001.
[31] S. Sugiura, T. Oda, Y. Aoyagi, M. Satake, N. Ohkohchi, M. Nakajima, Tubular gel
fabrication and cell encapsulation in laminar flow stream formed by
microfabricated nozzle array, Lab Chip. 8 (2008) 1255–1257.
[32] J. Yan, Y. Huang, D.B. Chrisey, Laser-assisted printing of alginate long tubes and
annular constructs., Biofabrication. 5 (2013) 1-8.
[33] R.T. Weitz, L. Harnau, S. Rauschenbach, M. Burghard, K. Kern, Polymer
nanofibers via nozzle-free centrifugal spinning, Nano Lett. 8 (2008) 1187–1191.
[34] T. Boland, T. Xu, B. Damon, X. Cui, Application of inkjet printing to tissue
engineering., Biotechnol. J. 1 (2006) 910–917.
139
Chapter 6 Contributions and Recommendations
The following was achieved through the research work described in this thesis.
6.1 Monoclonal Antibody Polishing Process Deign Based Membrane
Chromatography
PVDF membranes were studied for environmentally responsive characteristic under salt
concentrations [1- 5]. This membrane was suggested as bioseparation tool using different
apparent hydrophobic interactions between protein and membrane matrix. IgG and its
aggregates [1], and rabbit IgG and protein-A conjugated rabbit IgG [2] were used as
model proteins for assessing subtle hydrophobic difference. Based on earlier studies,
PVDF membrane chromatography was utilized for designing the monoclonal antibody
polishing process. The model proteins were choose according to industry practice, which
were monoclonal antibody, di-, tri-mer antibody aggregates, and conjugates with proteinA. Individual protein hydrophobic interaction with PVDF membrane chromatography
was investigated, and then feed solution (mixture of three proteins) was applied on
hydrophobic interaction membrane chromatography with optimized operating conditions.
A polishing step was built up with flow-through mode that allowed product (monoclonal
antibody) to flow through the membrane chromatographic system and two impurities
(aggregates and protein-A conjugates) were trapped in it.
140
This case study could afford many opportunities in the last step of impurities removal of
monoclonal antibody production line as a polishing step. It is suitable for fewer amounts
of impurities bound on membrane with high flow rate. It is even more beneficial from an
industries perspective, as the process does not need to be stopped in the middle to switch
buffers, because the fluid with final product will be collected in flow through stream. The
membrane chromatography adsorbent can be used up by calculating the amount of
impurities reached up to 80% of membrane binding capacity. This membrane adsorbent
can be regenerated by lowering salt concentration if it is necessary.
The second contribution would be the feasibility of this process development applied in
other biopharmaceutical products. As chapter 1 listed the advantage of membrane
chromatography, the biopharmaceutical product process development will be guided
toward to membrane chromatographic separation. With relative less membrane
chromatography technologies established time compared to resin-based column
chromatography, the growth of membrane chromatography performance is fast enough
which will be applied in biopharmaceutical production line.
For the future work can be suggested below.

Membrane binding capacity can be improved by adding hydrogel coating/grafting
on the membrane substrate.
141
6.2 inexpensive membrane chromatography with paper based composite membrane
The development of a PVCL composite membrane which is applied to membrane
chromatography has been assessed. In earlier work [6], the interests of composite
membrane were more focused on protein adsorbent. In this thesis chapter 3 achieved
PVCL composite membranes were adopted as a protein separation tool using its
environmentally responsive behaviour. Two major plasma proteins (hIgG and HSA) and
the popular hormone (insulin) were choosing for model protein. It flowed into the
composite membrane chromatography system at 1.3M of ammonium sulphate
concentration. First, the least hydrophobic protein, HSA, passed through the membrane
matrix without interaction. The remaining two proteins, hIgG and insulin, were eluted by
reducing salt concentration incrementally.
From this work, “custom made” hydrogel coated on filter paper membrane with an
inexpensive fabrication price, controllable of cross-linking degree, and flexible
geometries could be benefit both manufacturing and research areas. For comparison for
the economical aspect in chapter 2 and 3, commercially available hydrogel grafted
membrane (PVDF) for membrane chromatographic application increases the total cost of
production.
For the future work can be suggested below.

PVCL hydrogel stability needs to be improved over period of time. Composite
membrane has to be stored in liquid solution for the prevention of dehydration
which led breakage of polymer chain.
142

Protein fouling problem was presented after repeating separation runs. Several
experimental runs indicated pressure increases due to the protein fouling.
6.3 Hollow fiber membrane modification with natural hydrogel
The fouling proteins or biomolecules on hollow fiber membrane surface is the chronic
problem in membrane bioreactor and water treatment applications. Due to the
concentration polarization, the mass transport gradually lowers in comparison to initial
operating conditions. The initial motivation behind his project was prevent biological
(cell or cell debris) substance fouling/blocking on extra-capillary space, as it hindered
mass transport of nutrients and metabolite counter diffusion in membrane bioreactor. A
calcium alginate hydrogel was coated on single hollow fiber membrane for investigating
hydrophilized membrane surface which is friendlier to cell adhesion but lower fouling
behaviour.
There are three points of novelty in this work. The first is that the fabrication technique of
calcium alginate hydrogel is new in that cross-linking agents were stored within
membrane pores and diffused out to sodium alginate in controlled manner. This process
leads to the calcium alginate hydrogel coating being uniform and smooth along the entire
membrane length. Second, ethanol facilitates calcium chloride solution, showing different
hydrogel morphologies while having a sterile effect for biomedical applications. Third,
changes of hydrophilic characteristic membrane surface had positive effect on water
permeation, and DMSO mass transfer. This work has been investigated on each of these
143
individual components. It has much potential to be used in biomedical or
biopharmaceutical industry.
For the future work can be suggested below.

Biocompatibility test can be obtained with cells (i.e. T-cells, bacterial for certain
drug productions).

In this project, only ultrafiltration hollow fiber membrane was test. Microfiltration
hollow fiber membrane also can be a good candidate.

Hydrophobicity can be varying by different membrane materials such as
polyethersulfone, polysulfone, polypropylene, and PVDF hollow fiber membrane.
6.4 Calcium alginate fiber fabrication using hollow fiber membrane
Alginate hollow fibers are commonly produced by a complicated experimental set-up and
a large amount of samples (including immobilized cells). Since the size of set-up the
fabrication process is large, it is difficult to keep in sterile condition. Some of these
techniques, particularly electrospinning and two nozzle extrusion, also expose the cells to
harsh conditions, leading to an increased cell death rate. In this work we present a novel
method for fabricating calcium alginate on hollow fiber membranes. The membrane was
not used as typical separation barrier; it was adopted as a mold where it provides the
reservoir for cross-linking agents. Using this manner, it is easy to control the dimension
such as diameter, length, thickness of fibers, and morphology of fiber can be manipulated
by concentration of sodium alginate, cross-linking agent, and cross linking duration.
144
This opens up a new scope in membrane usage in tissue engineering area. In this thesis,
only the fabrication process was obtained with different sizes of fibers and several
characterizations of these fibers were performed with various imaging techniques.
Biocompatibility test was obtained with human umbilical vein endothelial cells.
For the future work can be suggested below.

Further research involving cell immobilization (i.e. using human urinary bladder
cell lines, epithelial cells, or human microvascular endothelial cells)

Biodegradation rate and cell growth rate within calcium alginate fiber matrix must
be undertaken for future work.

Seeding method can be embedded (chapter 5) or spreading on the surface of
calcium alginate fiber. Cell adhesion molecule can be treated on the surface such
as gelatin, collagen or serum.

In vitro experiments would check for functionality, leading to potential in vivo
(mouse) experiments.
145
6.5
References
[1]
L. Wang, J. Hale, R. Ghosh, Non-size-based membrane chromatographic
separation and analysis of monoclonal antibody aggregates, Anal. Chem. 78 (2006)
6863-6867
[2]
X. Sun, D. Yu, R. Ghosh, Study of hydrophobic interaction based binding of
immunoglobulin G on synthetic membranes, J. Membr. Sci. 344 (2009)165-171
[3]
L. Wang, R. Ghosh, Purification of humanized monoclonal antibody by
hydrophobic interaction membrane chromatography, J. Chromatogr. A 1107 (2006)
104-109.
[4]
R.Ghosh, Separation of human albumin and IgG by a membrane-based integrated
bioseparation technique involving simultaneous precipitation, microfiltration and
membrane adsorption, J. Membr. Sci. 237 (2004) 109-117
[5]
X. Sun, D. Yu, R. Ghosh, Study of hydrophobic interaction based binding of
immunoglobulin G on synthetic membranes, J. Membr. Sci. 344 (2009) 165-171.
[6]
K.Z. Mah, R. Ghosh, Paper-based composite lyotropic salt-responsive membranes
for chromatographic separation of proteins, J. Memb. Sci. 360 (2010) 149–154.
146