SCALABLE HUMAN INTESTINE MODEL WITH PERFUSABLE VASCULATURE 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



  

 

SCALABLE HUMAN INTESTINE MODEL WITH ACCESSIBLE LUMEN AND 

PERFUSABLE BRANCHED VASCULATURE 

 

BY: KRISTEN L. HAYWARD, B.SC. 

 

A Thesis Submitted to the School of Graduate Studies in Partial Fulfillment of the 

Requirements for the Degree Master of Applied Science 

McMaster University © Copyright by Kristen L. Hayward, July 2021 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 ii 

McMaster University MASTER OF APPLIED SCIENCE (2021) Hamilton, Ontario 

(Chemical Engineering) 

TITLE: Scalable Human Intestine Model with Accessible Lumen and Perfusable 

Branched Vasculature  

AUTHOR: Kristen L. Hayward, Hons. B.Sc. (York University) 

SUPERVISOR: Boyang Zhang, Ph.D. 

NUMBER OF PAGES: xv, 154  

 

 

 

Some sentences in this work have been reproduced with permission from Hayward, K. L., 

Kouthouridis, S. & Zhang, B. Organ-on-a-Chip Systems for Modeling Pathological Tissue 

Morphogenesis Associated with Fibrosis and Cancer. ACS Biomaterials Science & 
Engineering, doi:10.1021/acsbiomaterials.0c01089 (2020). Copyright 2020 American 

Chemical Society. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

iii 

Lay Abstract 

Two-dimensional cell culture and animal models inadequately represent human drug 

metabolism and diseases like inflammatory bowel disease and colorectal cancer. The 

objective of this work is to develop a more physiologically relevant human intestine model. 

Using fabrication techniques pioneered by the semiconductor industry, a custom organ-on-

a-chip platform in the format of a 384-well plate was developed. This platform is 

compatible with standard laboratory equipment and practices and can accommodate up to 

128 human intestine models comprised of the intestinal epithelium and associated network 

of blood vessels. In this platform, the cells of the intestinal epithelium and vasculature are 

supported by a network of natural proteins. This allows processes like vessel growth to be 

modelled in this platform. Vessel growth plays a key role in the progression of 

inflammatory bowel disease and cancer, and this model could help scientists better 

understand these diseases.  

 
 
 
 
 
 
 
 
 
 
 
 
 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

iv 

Abstract  

Two-dimensional cell culture and animal models inadequately represent human 

pharmacokinetics and diseases like inflammatory bowel disease and colorectal cancer. This 

means missed diagnostic and therapeutic opportunities, high drug attrition rates, and a 

portfolio of approved drugs that underdeliver the desired benefits to patient outcomes. This 

encourages the development of a more physiologically relevant intestine model. The 

objective of this work was to develop a 384-well plate organ-on-a-chip platform, 

IFlowPlate™, that can accommodate up to 128 human intestine models with accessible 

lumens and perfusable branched vasculature in an ECM environment. Fibrin-Matrigel® was 

used a structurally supportive and biologically instructive substrate that enabled: (1) 

prolonged cell culture (at least 15 days) with routine refreshment of aprotinin-supplemented 

medium, (2) formation of a confluent Caco-2 monolayer with barrier function, and (3) de 

novo assembly of a vascular network with barrier function. A fluorescent dextran 

permeability assay was used for in situ real-time measurements of epithelial barrier 

function in a high-throughput manner. Mixed co-culture of endothelial cells and fibroblasts 

in fibrin-Matrigel® resulted in the formation of an interconnected network of patent vessels 

that retained an albumin surrogate tracer within the luminal space indicating endothelial 

barrier function. To improve the success rate of anastomoses between living vessels and 

fluidic channels, the modification of inherently hydrophobic PDMS and polystyrene 

culture surfaces with ECM protein was explored. To address the limitations of a cancer cell 

line-derived intestine model, the replacement of Caco-2 cells with biopsied-derived colon 

organoid cells was investigated. Different gel formulations were assessed for their ability 

to induce colon organoid fragments to form monolayers. Finally, the incorporation of 

multiscale intestinal topography and luminal flow was considered through a modified 

approach to plate fabrication, whereby moulded alginate is embedded in ECM and 

sacrificed to generate a scaffold. Work to make the moulded alginate more robust is 

presented.  

 

 

 
 
 
 
 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

v 

Acknowledgements  
 
 

 I would like to thank my supervisor, Dr. Boyang Zhang, for introducing me to 

organ-on-a-chip technology and giving me an opportunity to engage in challenging and 

innovative work. I greatly value the breadth of knowledge and skills I have cultivated by 

working at the intersection of engineering, biomaterials science, and cell biology. I would 

also like to thank my examining committee members, Dr. Li Xi and Dr. Sandeep Raha, for 

the time and effort they extended in this role. 

 Thank you to all the past and present members of the Zhang Lab that I had the 

pleasure of working with. I appreciate that sharing was an important pillar of our workplace 

culture. We shared more than just a Google calendar. We shared knowledge, resources, the 

joy of each other’s successes, laughs, but thankfully never coronavirus.  

 I would like to express my gratitude for the financial support I received during my 

graduate studies in the form of the Queen Elizabeth II Graduate Scholarship in Science and 

Technology and the Clifton W. Sherman Scholarship.  

 Many thanks to the patients who consent to the biobanking of collected tissue for 

research. Your contribution is invaluable for advancing scientific understanding of health 

and disease. 

 To my extraordinary parents, Judy and Terry, this accomplishment is as much yours 

as it is mine. Thank you for all your sacrifices, love, and support that made this possible.  

  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

vi 

Table of Contents  
 
Lay Abstract  ......................................................................................................... iii  
 
Abstract  ................................................................................................................ iv 
 
Acknowledgements  ............................................................................................... v 
 
List of Figures  ....................................................................................................... x 
 
List of Tables  ...................................................................................................... xiii 
 
List of Abbreviations  ......................................................................................... xiv 
 
Declaration of Academic Achievement  ............................................................. xv 
 
1. Introduction  ...................................................................................................... 1 
 
2. Literature Review  ............................................................................................. 3 
 
2.1 Intestine Physiology  ........................................................................................ 3 
 
2.2 In Vitro Models  ............................................................................................... 4 
 
2.2.1 2D Cell Line Cultures  .................................................................................. 4 
 
2.2.2 Organoids  ..................................................................................................... 6 
 
2.2.3 Microstructured Scaffolds  .......................................................................... 6 
 
2.2.4 Organ-on-a-chip  ........................................................................................... 8 
 
3. Materials and Methods  .................................................................................. 12 
 
3.1 IFlowPlate™ Fabrication  .............................................................................. 12 
  

     3.1.1 Photolithography for SU-8 Master Template Fabrication  ............... 13 
 
     3.1.2 Soft Lithography for PDMS Mould 1 Fabrication  ............................ 14 
           
     3.1.3 Replica Moulding for Fabrication of Epoxy and PDMS Moulds  ..... 14 
            
     3.1.4 Silanization  ............................................................................................ 19 
 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

vii 

     3.1.5 Injection Moulding to Pattern Styrene Plate Bottom  ........................ 21 
       
     3.1.6 IFlowPlate™ Assembly and Disinfection  ............................................. 24 
      
3.2 SynoPlate™ Fabrication  ................................................................................ 26 
 
3.3 SynoPlate™ Cell Culture  .............................................................................. 33 
 
3.4 IFlowPlate™ Cell Culture  ............................................................................. 35 
 
     3.4.1 Cells and Culture Conditions  .............................................................. 36 
 
          Endothelial Cells ......................................................................................... 36 
 
          Fibroblasts ................................................................................................... 37 
 
          Caco-2 Cells  ............................................................................................... 38 
 
          Cell Passage Number  ................................................................................. 38 
 
          Culture Conditions for Cell Lines  .............................................................. 39 
 
     3.4.2 Hydrogels  ............................................................................................... 39 
 
          Fibrin  .......................................................................................................... 42 
 
          Matrigel®  ..................................................................................................... 44 
 
          Fibrin-Matrigel®  ......................................................................................... 46 
 
          Collagen  ...................................................................................................... 47 
 
          Gelatin  ........................................................................................................ 51 
 
     3.4.3 Cell Culture in IFlowPlate™  ................................................................. 52 
 
          Selection of Microfluidic Systems  ............................................................... 52 
 
          Temporary Sealing of Microfluidic Channels with Gelatin  ....................... 53 
 
          Cell Harvest and Encapsulation  ................................................................. 53 
 
          Replacement of Gelatin with Culture Medium  ........................................... 55 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

viii 

          Endothelialization of Microfluidic Channels with GFP-HUVECs .............. 58 

          Seeding Caco-2 Cells on Gel Surface  ......................................................... 58 

          Gravity-Driven Flow  .................................................................................. 58 
 
3.5 Characterization of Epithelial Barrier  ....................................................... 59 
 
     3.5.1 Formation of Caco-2 Confluent Monolayer on Fibrin-Matrigel®  .... 59 
 

     3.5.2 Immunofluorescence Staining  ............................................................. 59 
 
     3.5.3 Paracellular Permeability to Fluorescent-Dextrans  .......................... 61 
 
     3.5.4 Transepithelial Electrical Resistance (TEER)  ................................... 70 
 

3.6 Characterization of Self-Assembled Vessels  .............................................. 71 
 
     3.6.1 Paracellular Permeability of Vessels to 70 kDa TRITC-dextran  ..... 72 
 

     3.6.2 Image Acquisition  ................................................................................. 72 

 

3.7 Coating IFlowPlate™ Microfluidic Channels with ECM Protein  ............. 73 
 

3.8 Colon Organoid-Derived Monolayer Experiment  ..................................... 76 
 

     3.8.1 Source of Colon Organoids  .................................................................. 76 
 

     3.8.2 Seeding Colon Organoid Fragments on Collagen Gel Surface  ........ 76 
 
     3.8.3 Colon Organoid Culture Conditions .................................................... 77 
 
Chapter 4: Results and Discussion  .................................................................... 78 
 
4.1 IFlowPlate™ Avascular Intestine Model  ..................................................... 78 
 

     4.1.1 Modelling Intestine Epithelium with Caco-2 Cell Line  ..................... 78 
 

          Caco-2 Cells Polarize and Form a Functional Barrier  ............................. 78 
 
          Caco-2 Cells Form Confluent Monolayers on Fibrin-Matrigel®  ............... 81 
 
          Caco-2 Cells Express E-cadherin  ............................................................... 85 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

ix 

     4.1.2 Barrier Function  ................................................................................... 86 
 

          Paracellular Permeability to Fluorescent-Dextrans  .................................. 86 
 

          Transepithelial Electrical Resistance (TEER  ............................................. 92 
 
4.2 IFlowPlate™ Vascular Intestine Model  ....................................................... 96 
 
     4.2.1 Self-Assembled Vascular Networks  .................................................... 98 
 
     4.2.2 Anastomoses of Self-Assembled Vessels with Fluidic Channels  ..... 104 
 
          Sealing Gel  ............................................................................................... 104 

          Bidirectional Anastomoses ........................................................................ 108 

          Coating IFlowPlate™ Microfluidic Channels with ECM Protein ............. 114 

4.3 Enhancing Physiological Relevance of IFlowPlate™ Intestine Model  .... 121 
 

     4.3.1 IFlowPlate™ Primary Intestine Model  .............................................. 121 
 
     4.3.2 SynoPlate™ Intestine Model  ............................................................... 129 
 
Chapter 5: Conclusion  ...................................................................................... 135 
 
5.1 Summary ...................................................................................................... 135 
 
5.2 Prospective Application of IFlowPlate™ Intestine Model  ....................... 141 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

x 

List of Figures 
 
 
Figure No. Figure Title 

 

Page 

Figure 1 
 

Anatomy of the small and large intestine 

 

10 

Figure 2 
 

Timeline figure showing important milestones in the history of in 

vitro intestine model development 

 

11 

Figure 3 

 
Schematic showing Part A and B of IFlowPlate™ fabrication 

 

17 

Figure 4 
 

Schematic showing Part C of IFlowPlate™ fabrication 

 

18 

Figure 5 Schematic showing the possible arrangements of Pluronic® tri-

block co-polymers on a hydrophobic PDMS substrate 

 

22 

Figure 6 
 

Exploded diagram showing one of the 40 microfluidic systems of 

the intestine SynoPlate™ 

 

27 

Figure 7 
 

Schematic showing Part A of SynoPlate™ fabrication 

 

31 

Figure 8 Structure and swelling behaviour of calcium-alginate gel 

 

32 

Figure 9 
 

Schematic showing Part B of SynoPlate™ fabrication 

 

33 

Figure 10 Exploded diagram showing one of the 128 microfluidic systems of 

IFlowPlate™ 

 

36 

Figure 11 Cross-linkable natural polymers can reversibly form hydrogels in 

response to physical, chemical, and enzymatic stimuli 

 

42 

Figure 12 
 

1 mg/mL collagen prepared by neutralization of acid-solubilized 

concentrated rat-tail collagen type I 

 

49 

Figure 13 Schematic showing stepwise procedure for establishing a vascular 

tissue model in an IFlowPlate™ microfluidic system 

 

57 

 
 
 
 
 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

xi 

Figure 14 Schematic overview of the permeability assay procedure for assessing 

barrier function in IFlowPlate™ intestine model systems 

 

63 

Figure 15 
 

Two-fold serial dilution of 0.5 mg/mL FITC- and TRITC-dextran 

solution 

 

67 

Figure 16 
 
 

Day 7 standard curves of fluorescence intensity as a function of the 

amount of FITC- and TRITC-dextran 
68 

Figure 17 
 
 

Day 15 standard curves of fluorescence intensity as a function of the 

amount of FITC- and TRITC-dextran 

69 

Figure 18 
 
 

Exploded diagram showing the procedure for measuring the 

transepithelial electrical resistance of a Caco-2 monolayer in 

IFlowPlate™ using the EVOM2 

 

71 

Figure 19 
 

 

PDMS has high surface hydrophobicity and requires surface 

modification treatment to increase hydrophilicity and facilitate cell 

adhesion 

 

74 

Figure 20 
 
 

Caco-2 cells cultured on non-porous polystyrene substrates 

spontaneously form domes (arrows) indicating polarization and 

barrier function 

 

79 

Figure 21 
 
 

Caco-2 cells form hallmark cobblestone epithelium on fibrin-

Matrigel® and in mixed media that supports vasculature growth 

 

84 

Figure 22 
 

Caco-2 cells stained for junction marker E-cadherin to assess barrier 

integrity 

 

85 

Figure 23 
 
 

Microscopy images showing the permeability of a representative 

subconfluent (~80 % confluent) Caco-2 monolayer to high and low 

molecular weight fluorescent dextrans 

 

89 

Figure 24 
 

Microscopy images showing the permeability of a representative 

postconfluent (+8 days from confluence) Caco-2 monolayer to high 

and low molecular weight fluorescent dextrans 

 

90 

Figure 25 
 

Quantification of amount of eluted fluorescent dextran as an indicator 

of the barrier function of Caco-2 monolayers in IFlowPlate™ 

 

91 

Figure 26 In-house fabricated IFlowPlate™ microfluidic systems exhibit 
inconsistent geometries 

94 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

xii 

Figure 27 
 

Fibrin gel-encapsulated GFP-HUVECs self-assemble into a vascular 

network by Day 6 of culture in IFlowPlate™ 

 

101 

Figure 28 
 
 

70 kDa TRITC-dextran is transported and retained in lumens 

indicating the formation of patent vessels with a functional 

endothelial barrier 

 

103 

Figure 29 
 
 

Fibrin pre-gel cast in the middle well leaks into fluidic channels of 

IFlowPlate™ microfluidic systems and compromises perfusion of 

vascular networks 

 

106 

Figure 30 
 
 

Schematic showing that anastomoses between living vessels 

embedded in gel and fluidic channels of IFlowPlate™ microfluidic 

systems can be induced bidirectionally 

 

110 

Figure 31 Encapsulated GFP-HUVECs migrate from the gel into fluidic 

channels and form a “leaky” anastomotic connection with only one 

of the fluidic channels 

 

111 

Figure 32 GFP-HUVECs loaded into fluidic channels of IFlowPlate™ 

microfluidic systems insufficiently endothelialize channels under 

perfusion culture conditions 

 

112 

Figure 33 GFP-HUVECs loaded into fluidic channels of IFlowPlate™ 

microfluidic systems insufficiently endothelialize channels under 

static culture conditions 

 

113 

Figure 34 IFlowPlate™ microfluidic systems were coated with ECM protein to 

modify inherently hydrophobic PDMS and polystyrene surfaces and 

enhance GFP-HUVEC adhesion 

 

117 

Figure 35 Migration of gel-encapsulated GFP-HUVECs from tissue chamber 

into ECM-coated lateral fluidic channels in response to a strong 

chemoattractant gradient 

 

118 

Figure 36 Patient-derived colon organoid fragments seeded on top of collagen 

hydrogels to assess substrate-dependent monolayer formation 
 

124 

Figure 37 Microscopy images of polystyrene patterned with alginate via 

injection moulding using a two-level or single-level PDMS template 

134 

 

 

 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

xiii 

List of Tables 
 
 
Table No. Table Title 

 
Page 

Table 1 
 

Results of pilot study testing the measurement of electrical 

resistance of Caco-2 monolayers in IFlowPlate™ as a non-invasive 

method for assessing barrier function 

 

93 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

xiv 

List of Abbreviations  
 

ECM extracellular matrix 

PEGDM poly(ethylene glycol) dimethyl ether 

PDMS polydimethylsiloxane  

UV ultraviolet 

HEPA high-efficiency particulate air 

dH2O distilled water 

PEO poly(ethylene oxide)  

PPO poly(propylene oxide) 

CMC critical micelle concentration 

GFP-HUVECs green-fluorescent protein-human umbilical vein endothelial 

cells 

ECGM-2 endothelial cell growth medium 2 

DMEM Dulbecco’s modified Eagle’s medium 

FBS fetal bovine serum 

G α-L-guluronic acid 

D-PBS Dulbecco’s phosphate buffered saline 

BSA bovine serum albumin 

EHS Engelbreth-Holm-Swarm 

TEER transepithelial electrical resistance 

RGD Arg-Gly-Asp 

RCF relative centrifugal force 

FITC fluorescein isothiocyanate 

TRITC tetramethylrhodamine isothiocyanate 

PFA paraformaldehyde 

PI propidium iodide 

VEGF vascular endothelial growth factor 

MP microplastic 

PCBs polychlorinated biphenyls 

 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

xv 

Declaration of Academic Achievement 
 

I, Kristen Hayward, declare this thesis to be my own work, apart from the contributions 

outlined below.  

Dr. Boyang Zhang conceptualized IFlowPlate™ and SynoPlate™ and drew their designs 

using AutoCAD.  

Feng Zhang fabricated SU-8 master templates on silicon wafers in a clean room facility at 

the University of Toronto.  

Feng Zhang and Mandeep Kaur Marway fabricated IFlowPlate™ epoxy moulds.  

Shravanthi Rajasekar dissociated and seeded colon organoids.  

Scientific drawings were created with BioRender.com 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 1 

Chapter 1: Introduction  
  

 Angiogenesis is a hallmark of inflammatory bowel disorders1 and cancer.2 

Angiogenesis plays a critical role in the proliferation and metastasis of colorectal cancer.2 

Colorectal cancer is among the top three most prevalent cancers worldwide and is the third 

leading cause of cancer-related deaths.2 New vessels (1) supply the growing tumour with 

oxygen and nutrients to meet its increasing metabolic demands and (2) permit the 

dissemination of tumour cells to secondary sites; thus, contributing to metastasis. Not 

surprisingly, cancers with a poor prognosis are correlated with increasing production of 

angiogenic factors and increased density of tumour vasculature.3 For these reasons, 

antiangiogenic therapies represent an attractive approach to treating cancer. However, 

despite the widespread attention antiangiogenic therapies have received from the research 

community, they have not delivered the anticipated improvements to patient outcomes.4 

Possible reasons for this are that antiangiogenic compounds can 1) promote metastasis 

through hypoxia-induced upregulation of paracrine factors that promote cancer cell motility 

and invasion,5 and/or 2) reinforce tumour circumvention of angiogenesis whereby tumours 

use alternative strategies for gaining access to blood supply, such as vasculogenic mimicry6 

or vessel cooption.4  

 There is also growing interest in therapeutically targeting the tumour extracellular 

matrix (ECM) because it might represent less of a moving target than the genetically 

unstable cancer cells which frequently acquire new genetic alterations as they divide – a 

mechanism that can lead to drug resistance.7-9 While targeting the tumour ECM or the 

fibroblasts that produce/remodel it could represent an effective cancer therapy, current 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 2 

understanding of how the ECM contributes to cancer morphogenesis is limited. 

Deciphering the complex bidirectional relationships between cells and the ECM 

encourages the reductionist approach of in vitro models. 

 Experimentally tractable human intestine models that can model angiogenesis and 

the ECM with stromal cells, have the potential to improve scientific understanding of 

complex intestinal diseases, and facilitate the development of more effective therapies. The 

most well-established intestine-on-a-chip systems (Section 2.3.4) are configured as two 

stacked planar channels which are made of PDMS and separated by a synthetic membrane. 

In the upper compartment, epithelial cells are cultured, and in the lower compartment, 

endothelial cells are cultured. Two notable limitations of these traditional organ-on-a-chip 

systems are: (1) they are low-throughput and poorly compatible with standard laboratory 

equipment and practices, and (2) they impose non-physiological constraints and thus are 

not suited for modelling tissue morphogenesis (e.g., angiogenesis) which occurs in the 

context of a three-dimensional ECM environment. 

 The objective of this work is to address the shortcomings of gut-on-a-chip systems 

through the development of a scalable vascularized intestine model in a custom 384-well 

microfluidic platform termed IFlowPlate™. This platform is (1) free of non-physiological 

constraints and incorporates an ECM-based hydrogel that supports tissue morphogenesis 

and incorporation of stromal cells, (2) compatible with standard laboratory equipment and 

practices, and (3) has the potential to deliver high-throughput experimental capabilities. To 

make judicious use of resources and to make it easier to diagnose and address problems, 

the development of an intestine model in IFlowPlate™ was pursued incrementally. This 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 3 

work can be organized into three aims: (1) develop a Caco-2 epithelial barrier model, (2) 

develop a gel-embedded network of perfusable branched vasculature, and (3) enhance the 

physiological relevance of the model by a) incorporating primary intestine tissue, b) 

luminal flow, and c) multiscale intestine topography (i.e., tubular structure and crypts). Part 

b) and c) of the third aim requires reconfiguration of the IFlowPlate™ platform and a 

modified biofabrication approach. To reflect this departure, this platform is trademarked as 

SynoPlate™.  

 

Chapter 2: Literature Review  

 

2.1 Intestine Physiology 
 

 The lower gastrointestinal tract is comprised of the stomach, small intestine, and 

large intestine.10 The epithelial surfaces of the small and large intestine have distinct 

microstructures that support their specialized functions (Figure 1).10 Consistent with its 

role in nutrient absorption, the surface area of the small intestine is maximized by repeating 

units of luminal projections, villi, and invaginations, crypts.10 The epithelium of the large 

intestine possesses only crypts.10 This is consistent with its role in housing a highly dense 

and diverse community of microbes that ferment indigestible carbohydrates and produce 

absorbable metabolites essential for overall host health.10 A muscular layer surrounds the 

intestine and through a series of wave-like contractions, referred to as peristalsis, 

food/waste is simultaneously mixed and propelled unidirectionally along the 

gastrointestinal tract.10  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 4 

 The intestinal epithelium serves as a barrier between the luminal contents and the 

body.10 The lumen environment can be very harsh, and this requires the rapid turnover of 

epithelial cells.10,11 To support rapid tissue regeneration, stem cells are compartmentalized 

in sheltered niches at the base of crypts.10,11  The differentiated cells of the intestine 

originate from stem cells and migrate up the crypt to the villi (small intestine) or luminal 

surface (large intestine) where they can perform their special functions.10 Absorptive cells 

are the most abundant cell type in the intestine.10 They are covered in microvilli to 

maximize the absorptive surface area.10 Other cell types include mucus-secreting goblet 

cells, hormone-secreting enteroendocrine cells, and in the small intestine, specifically, 

microfold cells which present luminal antigens.10   

 
2.2 In Vitro Models 
 
 In this section, important milestones in the history of in vitro intestine model 

development (Figure 2) will be discussed. 

 
2.2.1 2D Cell Line Cultures 
 
 The failure to sustain differentiated normal primary intestinal epithelial cells ex vivo 

has led to the extensive use of cell lines established from gastrointestinal tumours.12,13 The 

Caco-2 cell line is the most well-established and widely used intestinal epithelial cell line 

for modelling the intestinal epithelium because it exhibits spontaneous differentiation and 

barrier and transport functions.13-16  

 In the body, intestinal epithelial cells are fed nutrients from their basolateral side 

surface which is in contact with the ECM facing the blood supply.17 When cultured on an 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 5 

impermeable substrate (e.g., polystyrene or glass) (Figure 2), Caco-2 cells are fed from 

their apical luminal surface and this can induce abnormal phenotypes.17 Transwell culture 

systems represent a more physiological model, as they are based on a porous membrane 

insert that is inserted into wells of standard multi-well cell culture plate. In these systems, 

Caco-2 cells are cultured on top of the insert allowing both the apical and basolateral 

surfaces access to nutrients leading to improved morphological and functional 

differentiation (Figure 2).13 Moreover, these systems can be used to assess transport across 

the epithelial monolayer. 

 Oral drug administration is the most common and most preferred method of drug 

administration because it does not require trained medical personnel, is non-invasive, and 

has high patient compliance.18,19  The major site of metabolism of orally administered drugs 

is the intestine.20,21 Drug metabolism in the intestine is largely determined by the types and 

abundance of cytochrome P-450 enzymes which are different between mice and humans 

and thus undermine the efficacy of preclinical animal models to predict human 

pharmacokinetics.20,21 For this reason, Caco-2 transwell systems are routinely employed by 

the pharmaceutical industry to model the human small intestine.20 However, the 

extrapolation of data from these models to predict drug metabolism and drug-drug 

interactions in humans is challenged by the low expression of cytochrome P-450 

enzymes.20  

 Other limitations of Caco-2 models include: (1) they do not not distinctly represent 

the small intestine or colon, rather they display characteristics of both,13 (2) they can form 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 6 

multilayer regions that resemble polyp-like masses,22 and (3) having originated from a 

tumour, they cannot be used to study the transformation of precancerous cells.  

 
2.2.2 Organoids 
 
 Organoid models rely on intestinal stem cells that reside in the crypts of biopsied 

intestinal tissue to generate a self-organizing miniaturized intestine that possesses all the 

differentiated cell types of the native epithelium.23 Intestine stem cells are cultured in a 3D 

Matrigel® matrix.23 This growth environment promotes the formation of polarized cysts 

with a central lumen (Figure 2).24 The closed, cyst-like morphology characteristic of 

intestinal organoids complicates their use in studies of nutrient transport, drug absorption, 

and microbe-epithelium interactions that require direct access to the inside of the organoid 

which is analogous to the intestinal lumen.25,26  Moreover, the accumulation of secreted 

material and shed apoptosed cells inside of the organoid is non-physiological and 

necessitates their regular passaging to maintain them in culture.27,28 In the body, 

debris/waste that is shed into the intestinal lumen is continuously cleared by luminal flow. 

To overcome the limitations presented by the closed, spherical geometry of organoids, 

model systems that “open-up” 3D organoids into 2D monolayers with an accessible lumen 

have been developed.25-27,29 

 
 
2.3.3 Microstructured Scaffolds 
 
 Work in the field of mechanobiology supports the existence of a multiscale 

structural hierarchy from organ structure to intracellular structure (i.e., the cytoskeleton) 

that physically integrates mechanical forces with the nucleus, suggesting a mechanism by 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 7 

which tissue geometry can regulate gene expression.30 Tissue geometry can also exert its 

effects on cell phenotype and behaviour in a mechanical-independent manner by defining 

cell proximity and the local concentrations of secreted factors.31-33 Reconstituting the 3D 

geometry of the native intestine has been shown to increase mucin production and improve 

metabolic function of cultured Caco-2 cells, compared to 2D models.34   

 A significant limitation of 2D cell culture models and organoids is their lack of 

representation of native intestine 3D geometries. For example, a distinguishing feature of 

the small intestine is the presence of villus structures which are finger-like luminal 

projections that increase the absorptive surface area of the small intestine. Small intestine 

organoids inappropriately resemble colon organoids in that no villus structures are present.  

 Villous atrophy is a characteristic of many intestinal disorders, such as celiac 

disease, Crohn’s disease,35 and acute radiation syndrome,36 and is diagnosed by visual 

inspection of tissue obtained from a biopsy. Villous atrophy results in malabsorption 

leading to nutrient deficiencies, weight loss, and diarrhea.35 Modelling villous atrophy can 

facilitate the development of medical counter measures.37  

 A typical approach to creating intestine models with villus and/or crypt structures, 

is the fabrication of a hydrogel mould upon which epithelial cells can be seeded. Laser 

ablation and soft lithography have been used to generate a PDMS template for moulding 

calcium-alginate into a secondary dissolvable template.38 This calcium-alginate template 

was used to mould collagen gel into villi, before it was sacrificed to generate the final 

microstructured collagen scaffold which can be placed on a transwell insert (Figure 2).34,38 

In contrast to the subtractive manufacturing approach just described, Creff et al. used a 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 8 

high-resolution additive manufacturing approach to reproduce the complex villus-crypt 

geometries of the small intestine39 Specifically, they employed 3D stereolithography 

printing whereby a UV laser is directed in a specific path across a photopolymerizable 

hydrogel to induce cross-linking of polymers into a hardened layer.39  

 

2.2.4 Organ-on-a-chip 
 

Since scientists began culturing cells ex vivo, the focus has been on the chemical 

environment of cells. By manipulating the chemical composition of the cell culture 

medium, gene expression can be regulated to control cell survival, proliferation, and 

differentiation. Organ form and function emerge from the dynamic and complex 

interactions between cells and their surroundings. This urges the development of in vitro 

models that not only represent the chemical environment, but also the diversity and spatial 

organization of cell types, and the physical (e.g., ECM microstructure) and mechanical 

(e.g., flow and cyclic strain) environment. This has been accomplished by the convergence 

of biological principles with engineering to create organ-on-a-chip systems. The wide 

appeal garnered by organ-on-a-chip systems is largely for two reasons, (1) they can model 

human physiology/pathophysiology more closely than 2D cell culture and animal models 

and (2) they offer unprecedented control over the components of living systems that are 

represented and thus make it easier to disentangle cause and effect relationships.  

The first and most well-established organ-on-a-chip model of the human intestine, 

Gut-on-a-Chip,40 (Figure 2) is a PDMS-based device with two stacked perfusable planar 

channels for independent 2D culture of epithelium and endothelium. Epithelial cells are 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 9 

cultured on top of an ECM-coated porous membrane that separates the upper and lower 

compartment.40 Cyclic suction applied to lateral hollow vacuum chambers is used to mimic 

peristaltic motions of the gut.40 It has been shown that when cultured in a mechanically 

active environment reminiscent of the living intestine, Caco-2 cells can be stimulated to 

undergo spontaneous morphogenesis into villi-like structures and differentiate into four 

different epithelial cell types of the small intestine – absorptive, enteroendocrine, Paneth 

and Goblet cells.41 

 To enhance the physiological relevance of the Gut-on-a-Chip cultured with Caco-2 

cells, historically independent organoid and organ-on-a-chip approaches to tissue 

modelling were combined to create a new category of in vitro model – the organoid-on-a-

chip. Compared to Caco-2 chip models, intestinal organoid-on-a-chip models have gene 

expression profiles that more closely resemble the native human intestine.20,27 For example, 

the Duodenum Intestine-Chip showed levels of CYP3A4 expression comparable to levels 

observed in the human adult duodenum and thus could be used to better predict human 

pharmacokinetics.20 Moreover, intestinal organoid-on-a-chip models exhibit higher 

functional fidelity.27 The Intestine-Chip (Figure 2) displayed a 10 times higher 

concentration of mucus which plays a critical role in tissue homeostasis.27 It has also been 

shown that when cultured in the Intestine-Chip, organoid-derived duodenal cells display 

gene expression profiles that are more similar to the living human duodenum than duodenal 

organoids.27 This suggests that recapitulation of the native cellular environment which 

includes endothelium and mechanical stimuli, plays an important role in modulating 

cellular gene expression. 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 10 

 Most recently, an intestine organoid-on-a-chip system that incorporates a 

microstructured ECM scaffold has been reported.28 Laser ablation was used to generate 

crypt-like cavities in a natural collagen-Matrigel® hydrogel (Figure 2). Sox2 staining 

showed the physiological spatial segregation of stem cells to the base of crypts.28 

 
 

 
Figure 1. Anatomy of the small and large intestine. The epithelial surfaces of the small 

and large intestine have distinct microstructures that support their specialized functions. 
Stem cells (green) are enriched at the bases of crypts and differentiated cells (white) along 

the villi (small intestine) or luminal (large intestine) surface. The generic structure of the 

capillary bed within the intestinal mucosa is shown. Arterial (red) and veinous (blue) 

circulation run parallel to each other. Lacteals are shown in yellow. Figure reproduced 

from Ref. 10. 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 11 

 
 

Figure 2. Timeline figure showing important milestones in the history of in vitro intestine model development. Organ-
on-a-chip figures reproduced from Refs. 40 (Gut-on-a-Chip), 27 (Intestine-Chip, emulatebio.com), 28 (scaffold-guided 
homeostatic mini-intestine). 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 12 

Chapter 3: Materials and Methods 
 
 
3.1 IFlowPlate™ Fabrication  
 
 IFlowPlate™ is a custom 384-well plate with 128 independent microfluidic systems 

comprised of three-well units. The wells of each unit are connected by microfluidic 

channels allowing for the flow and exchange of fluid between wells. ECM pre-gel solution 

(with/without cells) is cast in the middle well and fluid flows from/into adjacent reservoir 

wells through the hydrogel. This is representative of interstitial flow in the body which is 

the flow of fluid around interstitial cells and through the ECM that surrounds tissues.42 

Interstitial flow through ECM differs from open-channel flow through a conduit, such as a 

microfluidic channel or blood vessel, as fluid moves in all directions through the ECM and 

flow velocity is retarded by the high flow resistance of the ECM.42 “IFlowPlate™” is a 

portmanteau of the words “interstitial flow” and “plate.” 

 IFlowPlate™ is comprised of a standard black 384-well bottomless plate (Cat. # 

82051-544, Greiner Bio-One) and a 11.6 cm x 7.5 cm x 0.081 cm (length x width x height) 

polystyrene sheet that serves as the plate bottom. The polystyrene sheet is patterned with a 

sacrificial polymer – poly(ethylene glycol) dimethyl ether (PEGDM, Mn ~2,000, Cat. # 

445908, Sigma-Aldrich) using injection moulding. PEGDM is injected into a 

polydimethylsiloxane (PDMS) elastomer mould (SYLGARD™ 184 silicone elastomer 

clear, Cat. # 4019862, Dow) of an array of 128 channel-like grooves that bridge three 

adjacent wells. The PEGDM-patterned polystyrene sheet is glued to the bottomless plate 

and the PEGDM is subsequently dissolved with water to generate channels that connect 

three adjacent wells. This method represents a relatively inexpensive and technically facile 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 13 

approach to integrating microfluidics with multiwell plates of different configurations (e.g., 

6-well to 384-well plates) to deliver high-throughput experimental capabilities. The 

IFlowPlate™ fabrication method is described in detail in the sections below and a schematic 

illustrating the stepwise process is shown in Figures 3 and 4.  

 

3.1.1 Photolithography for SU-8 Master Template Fabrication 
 

 The design for IFlowPlate™ was drawn using computer-aided software (AutoCAD, 

Autodesk). A photomask that met the specifications of the design was manufactured by an 

external supplier and used to fabricate the SU-8 master template on a silicon wafer 

(diameter: 150 mm, thickness: 0.65 mm, UniversityWafer, Inc., South Boston, MA) via 

conventional photolithography. 

 A thin film of light-sensitive epoxy-based material, SU-8 2050 negative photoresist 

(Cat. # Y1110720500L1GL, MicroChem, Westborough, MA) was spin-coated onto the 

silicon wafer and soft baked (Figure 3A, step 1). The patterned photomask was applied to 

the surface of the photoresist and the wafer was exposed to ultraviolet (UV) light (Figure 

3A, step 2). Portions of the photoresist that were not masked were cross-linked by UV light 

and rendered insoluble by the liquid developer which the SU-8 was immersed in following 

a post-exposure hard bake step (Figure 3A, step 3). Fabrication of the SU-8 master 

template was done at the University of Toronto in a clean room equipped with a high-

efficiency particulate air (HEPA) filter and UV filters. The SU-8 master template was 

placed inside a square bioassay dish (245 mm x 245 mm, Cat. # 29186-491, Corning) and 

secured using adhesive tape.  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 14 

3.1.2 Soft Lithography for PDMS Mould 1 Fabrication 

 Unlike photolithography which relies on light exposure to transfer a pattern to a 

substrate, soft lithography relies on the physical deformation of a soft, elastomeric material 

(e.g., PDMS). To replicate the inverse of the SU-8 master features in PDMS, PDMS 

(SYLGARD™ 184) comprised of five parts by weight elastomer base and one part curing 

agent was poured on top of the SU-8 wafer in a bioassay dish. The PDMS was degassed 

under vacuum (−0.09 MPa gauge pressure, 30 minutes), cured at room temperature for 48 

hours, and was cut and peeled from the SU-8 wafer (Figure 3B, step 1). This PDMS mould 

will hereafter be referred to as PDMS mould 1 to distinguish it from the PDMS moulds 

used as templates for injection moulding. 

 

3.1.3 Replica Moulding for Fabrication of Epoxy and PDMS Moulds 

 PDMS mould 1 replicates the inverse features of the SU-8 master template and was 

used to create an epoxy mould that replicates the SU-8 master features in a process referred 

to as replica moulding. With the feature surface facing up, PDMS mould 1 was cut using a 

single edge blade. A 2 mm biopsy punch (Cat. # 21909-132, Integra York PA, Inc.) was 

used to punch a hole at the ends of each channel (Figure 3B, step 2). Punch waste was 

pushed through the holes using the fine tips of precision stainless steel tweezers and 

discarded. The holes in PDMS mould 1 generated posts in the epoxy mould which in turn 

generated holes in the cast and demoulded PDMS used as a template to pattern the 

polystyrene plate bottom. The holes in the PDMS template serve as ports for the injection 

of PEGDM and permit the escape of air during filling of microchannels with PEGDM.  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 15 

 The featureless surface of PDMS mould 1 was cleaned of debris using adhesive 

tape and activated using a BD-20A high frequency plasma generator (Cat. # 12011A, 

Electro-Technic Products, Chicago, IL) before being bonded to a cleaned and activated 

PDMS (SYLGARD™ 184) slab (Figure 3B, step 3). This PDMS slab was prepared by 

pouring 90 g of a mixture of 5:1 by weight PDMS elastomer base to curing agent into an 

empty square bioassay dish (245 mm x 245 mm, Cat. # 29186-491, Corning). After curing 

at room temperature, the PDMS was peeled from the dish and a single edge blade was used 

to cut a PDMS slab to the same length and width as PDMS mould 1.  The two-layer PDMS 

mould 1 was treated with vapored trichloro (1H, 1H, 2H, 2H-perfluorooctyl) silane to 

deposit a “nonstick” film that would prevent irreversible bonding of the PDMS with epoxy 

(Section 3.1.5).  

 To fabricate the epoxy mould, a 15 cm x 9 cm x 5 cm (length x width x height) 

cardboard rectangular box was lined with a smooth aluminum foil insert. 18 g of PDMS 

(SYLGARD™ 184), prepared by mixing the elastomer base and curing agent in a ratio of 

5:1 by weight, was poured into the box. The box was rotated to coat the bottom and half 

the height of the side walls with a thin film of PDMS. The liquid PDMS was partially cured 

at room temperature to achieve a semi-solid state with a viscosity and stickiness that 

resembles that of glue. Both faces (feature and featureless) of the silanized two-layer PDMS 

mould 1 were cleaned with adhesive tape to lift any debris. With the feature surface facing 

up, PDMS mould 1 was placed in the middle of the box on top of the glue-like PDMS. 

Following overnight incubation at room temperature, the glue-like PDMS was fully cured. 

80 mL of equal parts by volume of epoxy resin and hardener (EasyCast® clear casting 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 16 

epoxy, Cat. # 33016, Environmental Technology, Inc., Galesburg, MI) was vigorously 

mixed and cast on top of PDMS mould 1. After degassing under vacuum (- 0.09 MPa gauge 

pressure, 10 minutes), unpopped air bubbles that had risen to the surface of the resin were 

lifted off with a dry wooden craft stick and the epoxy was left to cure at room temperature. 

The epoxy mould was released from the aluminum foil and PDMS mould 1 was lifted from 

the epoxy mould using a double-ended microtapered stainless steel spatula and discarded 

(Figure 3B, steps 4 – 5). The epoxy mould was silanized with vapored trichloro (1H, 1H, 

2H, 2H-perfluorooctyl) silane (See 3.1.5) and subsequently used to create PDMS moulds 

with holes and channel-like grooves to be used as templates for patterning the polystyrene 

plate bottom.  

 The PDMS mould for patterning the polystyrene IFlowPlate™ bottom was made by 

pouring PDMS (SYLGARD™ 184) comprised of thirty parts by weight elastomer base and 

one part curing agent into the IFlowPlate™ epoxy mould. The cast PDMS was degassed 

under vacuum (−0.09 MPa gauge pressure) for 1 hour to decrease the solubility of gas in 

the PDMS and remove air bubbles which could lead to defects in the cured mold. The 

PDMS mould was cured at 47 °C overnight and subsequently demoulded (Figure 3B, steps 

6 – 8).  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 17 

 

 

 

128 
features 

Figure 3. Schematic showing Part A and B of IFlowPlate™ fabrication. (A) 
Photolithography to generate SU-8 master template with 128 positive features on silicon 
wafer. (B) Soft lithography and replica moulding to generate PDMS moulds with negative 
features for patterning styrene sheets via injection moulding (See Part C, Fig. 4).  

(3) 

Posts          Holes 

Groove Ridge 

(5) (8) 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 18 

 

 

 

 

Figure 4. Schematic showing Part C of IFlowPlate™ fabrication (Figure 1 Cont’d). 
(C) Stepwise process to pattern a styrene sheet (plate bottom) with PEGDM via injection 
moulding and bond it to a 384-well bottomless plate to generate a custom 384-well 
microfluidic platform termed IFlowPlate™.  IFlowPlate™ has 128 independent three-well 
microfluidic units. Channels between wells permit the flow and exchange of fluids.  
 

TOP BOTTOM 

(6) 
(8) 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 19 

3.1.4 Silanization  

 To prevent bonding of PDMS with epoxy, a “nonstick” film of vapored trichloro 

(1H, 1H, 2H, 2H-perfluorooctyl) silane was deposited on (1) the two-layer PDMS mould 

used to make the epoxy mould (i.e., PDMS mould 1), and (2) the epoxy mould used to 

make PDMS moulds for patterning polystyrene plate bottoms. A square bioassay dish (245 

mm x 245 mm, Cat. # 29186-491, Corning) with lid was used to create a closed chamber 

in which the liquid silane would vaporize at room temperature to generate a silane 

atmosphere. Due to the toxicity of trichloro (1H, 1H, 2H, 2H-perfluorooctyl) silane, 

silanization was performed inside a chemical fume hood. The procedures for silanizing 

two-layer PDMS moulds and epoxy moulds were nearly identical apart from how the 

moulds were secured to the square bioassay dish. PDMS moulds were secured to the dish 

using double-sided adhesive tape placed on the featureless surface of the two-layer mould. 

To secure epoxy moulds to the dish, masking tape was placed along the long and short 

edges of the mould. Two PDMS/epoxy moulds were secured to the dish one above the other 

in landscape orientation. The inside of the lid was covered with aluminum foil and 90 µl of 

silane was deposited on the aluminum foil in equally spaced beads. Using a pipette tip, 

silane beads were streaked in a grid-like pattern for even exposure of mould surfaces to 

evaporated silane. To create the chamber, the dish was put inside the lid so that the moulds 

were suspended above the silane. Silanization time was 45 minutes and 1 hour for PDMS 

and epoxy moulds, respectively.  

 Using the procedure described, the vapor phase silane formed an opaque, white-

coloured film on PDMS mould 1 indicating multi-layer deposition of silane molecules on 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 20 

the PDMS surface. This proved to be problematic for downstream fabrication steps as the 

epoxy moulds generated from the silanized two-layer PDMS mould 1 also had an opaque, 

white-coloured film, as did the PDMS moulds generated from the epoxy moulds. These 

PDMS moulds were used as templates for patterning the polystyrene plate bottom via 

injection moulding and their precise alignment with a reference 384-well bottomless plate 

(See 3.1.6) relies on being able to see through a transparent mould to the reference plate. 

Moreover, polystyrene sheets patterned with PEGDM using opaque PDMS moulds showed 

substantial PEGDM leakage. It is speculated that the thick film of silane and/or mould of 

the surface roughness on the PDMS mould undermines bonding with the polystyrene sheet 

(See 3.1.6).  Lastly, excess silane can fill in the negative features (e.g., holes) of a template 

mould and negatively change the dimensions of positive features (e.g., posts) of the replica 

mould.43 This is particularly relevant for nanoscale features.   

 While silanes are known to chemically adsorb to material surfaces and form self-

assembled monolayers, this process is usually highly variable.43 Manipulating the volume 

of silane, duration of silanization, and ambient pressure by performing the reaction in a 

vacuum chamber/desiccator can help to improve the quality and consistency of silane 

deposition.43 In addition to manipulating these parameters, a post-silanization processing 

step to remove silane can also help to achieve the desired result. Indeed, it was discovered 

serendipitously that adhesive tape works well to lift excess layers of silane molecules from 

PDMS mould 1 and restore its transparency. It is speculated that the ease with which the 

silane molecules could be removed is due to the disorganized multi-layer arrangement of 

silane molecules.43 Therefore, following the removal of PDMS mould 1 from the silane 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 21 

chamber, overlapping pieces of adhesive tape were placed on the feature face of the mould, 

smoothed over, and peeled off after 2 hours. This was repeated once more as lines of silane 

remained after the first tape application. These lines corresponded to the borders of tape 

pieces suggesting that the borders might be less adhesive than the interior of the tape.  

 

3.1.5 Injection Moulding to Pattern Polystyrene Plate Bottom  

 Using a box cutter, a 0.081 cm optically transparent polystyrene sheet (Cat. # 

V16010, Jerry’s Artarama) was cut to 11.5 cm x 7.6 cm to fit the dimensions of a black 

bottomless 384-well plate (Cat. # 82051-544, Greiner Bio-One). Using a single edge blade, 

the perimeter of the PDMS mould was trimmed to fit the dimensions of the polystyrene 

sheet and then bathed for 30 minutes in a 5% (w/v) solution of Pluronic® F-127 (Cat. # 

P2443, Sigma-Aldrich) in distilled water (dH2O). With the mould submerged, features were 

tapped with fingertips to simultaneously purge inlet/outlet holes and microchannels of 

trapped air and fill them with amphiphilic Pluronic® F-127 solution (Figure 4, step 1). 

Pluronics® are tri-block co-polymers comprised of poly(ethylene oxide) (PEO) and 

poly(propylene oxide) (PPO).44 These co-polymers self-assemble with the hydrophobic 

PPO block interfacing with the hydrophobic PDMS substrate and the hydrophilic PEO 

blocks interfacing with the water layer.44,45 This arrangement improves the hydrophilicity 

of the PDMS and deters other molecules from interfacing with the substrate via the large 

excluded volume and configurational entropic repulsion mechanisms.44,45 Previous studies 

have shown that the success of Pluronic® surface treatment is dependent on the solution 

concentration, and a concentration that does not exceed 5% - the critical micelle 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 22 

concentration (CMC), is ideal to achieve a surface saturated with exposed PEO blocks 

(Figure 5a).44 At concentrations below the CMC there is incomplete surface coverage 

(Figure 5b). At concentrations above the CMC, Pluronic® co-polymers can form bilayers 

with poorly exposed PEO blocks (Figure 5c), or micelle/aggregates (Figure 5d) that can 

be rinsed away. Modification of the PDMS mould surface chemistry using 5% Pluronic® 

F-127 solution was ideal for preventing irreversible bonding of the PDMS mould with the 

injected PEGDM or the polystyrene sheet.  

 

 

 

 

 

 

 

 

 

 

 

 The PDMS mould was rinsed with dH2O and placed on top of low-linting absorbent 

tissue to dry. A 11.5 cm x 7.6 cm x 0.081 cm polystyrene sheet was plasma treated to 

functionalize the surface and enhance bonding of the PEGDM with the polystyrene sheet. 

Adhesive tape was used to lift debris off the feature surface of the dry PDMS mould before 

Figure 5. Schematic showing the possible arrangements of Pluronic® tri-block co-
polymers on a hydrophobic PDMS substrate. (a) high density monolayer, (b) low 
density monolayer, (c) bilayer, (d) micelle/aggregates. Figure is reproduced from Ref. 1. 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 23 

it was superimposed onto the functionalized polystyrene sheet. For precise alignment of the 

mould features with wells, the polystyrene sheet was placed on a bottomless 384-well plate 

for reference of well positions. The PDMS mould was bonded with the polystyrene sheet 

(Figure 4, step 2). The device was lifted off the reference bottomless 384-well plate and 

exposed to vacuum (- 0.04 MPa gauge pressure) for 15 minutes to remove air trapped 

between the PDMS mould and the polystyrene sheet.  

 The polystyrene sheet was patterned with poly (ethylene glycol) dimethyl ether 

(PEGDM, Mn ~2,000, Cat. # 445908-50G, Sigma-Aldrich) using injection moulding. The 

water solubility of PEGDM makes it attractive for use in applications requiring a sacrificial 

polymer, such as the fabrication of IFlowPlate™. Using a 10 mL syringe with a cropped 19 

G needle, liquid PEGDM (melted at 65 °C) was injected into the inlet and outlet holes of 

the PDMS mould of an array of channel-like grooves that connect three adjacent wells 

(Figure 4, step 3). Because PEGDM quickly solidifies at room temperature, following 

injection, the device was placed in a 65 °C oven to return the PEGDM to a liquid state. 

With the PEGDM in a liquid state, the device was pulse centrifuged at 1000 RPM, and then 

incubated in a 65 °C vacuum oven at  −0.05 MPa gauge pressure for 30 minutes. After 

incubation, the polystyrene sheet was pulse centrifuged at 1000 RPM and incubated at 65 

°C for 30 minutes. This technique is a modification of a technique referred to as “vacuum 

filling” described in microfluidics literature.46 When the polystyrene sheet is exposed to 

vacuum, displaced air in the channels will escape through the PEGDM liquid or the gas-

permeable PDMS walls. Eventually, a pressure equilibrium will be reached and the 

pressure in the channels will be low. When the device is returned to atmospheric pressure, 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 24 

there is higher pressure at the surface of the mould compared to inside the channels. This 

positive pressure differential pushes the PEGDM into the channels.  

 After channels were filled, excess PEGDM was removed from the inlet and outlet 

holes using a multichannel pipette attached to a vacuum aspirator. To minimize the risk of 

inadvertently aspirating PEGDM from the channel-like grooves, the pipette tips were 

oriented in the opposite direction of channels. To accelerate solidification of PEGDM, the 

polystyrene sheet was incubated at –20 °C for 10 minutes. The PDMS mould was peeled 

from the styrene sheet (Figure 4, steps 4 – 5) revealing the patterned 300 µm x 150 µm 

PEGDM features. Localized debonding/incomplete bonding at the PDMS-polystyrene 

interface due to contaminants or air bubbles can result in PEGDM leakage. Patterned 

polystyrene sheets were inspected prior to assembly of IFlowPlate™. Leaked PEGDM on 

the back and edges of the polystyrene sheet was cleaned with a small, wetted polyfoam 

paint brush. For more delicate cleaning of leaked PEGDM around patterned features, a 

wetted pointed end tortillon (i.e., paper pencil used by artists for blending) was used. 

 

3.1.6 IFlowPlate™ Assembly and Disinfection 

 To assemble IFlowPlate™, a black 384-well bottomless plate was bonded to the 

PEGDM-patterned polystyrene sheet using high-viscosity PDMS (SYLGARD™ 186 

silicone elastomer clear, Cat. # 2137054, Dow) comprised of 10 parts by weight elastomer 

base and one part curing agent. PDMS is widely used in biomedical applications because 

it has many desirable properties including optical transparency, biocompatibility, and gas 

permeability.47 It is because of these properties that PDMS was chosen to bond the 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 25 

polystyrene plate bottom to the 384-well bottomless plate. The PDMS glue encases the 

PEGDM and when the PEGDM is dissolved, the PDMS will form the walls of the 

microfluidic channel, and thus be in direct contact with culture medium and cells.  

 5 g of high-viscosity PDMS (SYLGARD™ 186) was evenly spread in a thin layer 

across a glass slide (large glass microscope slide, Cat. # 260233, Ted Pella, Inc., Redding, 

CA) that was cut to the approximate size of the polystyrene sheet. For greater control over 

the amount of glue spread on the glass slide and thus transferred to the bottomless plate, 

two pieces of masking tape were layered on each long edge of the glass slide before 

spreading glue. The glass slide was placed on a plasma-treated 384-well bottomless plate 

to facilitate the transfer of the PDMS glue to the bottomless plate. The increased surface 

wettability of the plasma-treated bottomless plate improves the spread of the PDMS glue. 

The glass slide was lifted off of the bottomless plate and the PEGDM-patterned polystyrene 

sheet was aligned with the bottomless plate and then dropped into position. Gentle 

compression was used to bond the polystyrene sheet to the 384-well bottomless plate. A 

clean glass slide was placed on top of the polystyrene sheet and clamped to the assembled 

IFlowPlate™ using four extra-large paper clamps. After overnight curing of the PDMS glue 

at room temperature, the clamps and glass slide were removed (Figure 4, step 6). Glass 

slides with PDMS were cleaned with acetone and re-used. 

 To dissolve the PEGDM and reveal channels that connect units of three adjacent 

wells, dH2O was added to the wells (Figure 4, step 7). The plate was pulse centrifuged at 

40 G to simultaneously fill wells/channels with water and purge trapped air. The plate was 

placed on a rocker and incubated at 37 °C for 1 – 2 days before the PEGDM-dH2O solution 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 26 

was aspirated out. To disinfect IFlowPlate™ in preparation for cell culture, the plate was 

placed inside a square bioassay dish inside the biosafety cabinet. The bioassay dish, and 

IFlowPlate™ lid and wells were filled with 70% (v/v) ethanol. The bioassay dish was 

covered with the lid to minimize evaporative loss of ethanol and was left undisturbed for 

1.5 to 2 hours (Figure 4, step 8).  

 Fabrication flaws that affect the performance of IFlowPlate™ microfluidic systems 

should be identified prior to cell culture. Each fabricated IFlowPlate™ was inspected for 

glue-related defects, such as too little or too much glue, that could result in leaking wells 

or obstructed channels, respectively. Firstly, the open connection between three-well units 

was confirmed by aspirating liquid from the middle well and observing a negative change 

in the liquid level in the connected inlet and outlet wells. Secondly, dried plates were 

inspected for incomplete seals around wells and channels. The alphanumeric codes of three-

well units that met quality standards were recorded and these microfluidic systems were 

available for cell culture. To avoid edge of plate effects, microfluidic systems at the plate 

periphery were not selected for cell culture. 

 

3.2 SynoPlate™ Fabrication 

 SynoPlate™ is a custom 384-well plate with 40 independent microfluidic systems. 

Each microfluidic system is a nine-well unit comprised of a triad of three-well units. Aside 

from having three times the number of interconnected three-well units, the most notable 

attribute of SynoPlate™ that distinguishes it from IFlowPlate™, is the integration of 

biologically inspired tissue-specific scaffolds. These scaffolds are tubular cavities that are 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 27 

patterned in a fibrin-based hydrogel and serve as a compliant boundary that guides the 

assembly of cells into tissues that resemble their in vivo counterparts. Cell attachment to 

the ECM substrate supports open-channel flow through the cell-lined conduits. Each three-

well unit is comprised of a dedicated inlet and outlet reservoir well connected via channels 

to a tissue scaffold in the centre well where the triad of three-well units converge. This 

configuration enables three compartmentalized, yet biologically interactive tissues with 

unique architectures, cell types, media requirements, and opportunities for independent 

sampling of effluent. The intestine SynoPlate™ has three independently addressable tissue 

scaffolds that guide the assembly of cells into a tubular intestine with crypt-like 

invaginations, and a branching vascular network above and below the intestine tube 

(Figure 6).  

 

 

 

 

  

 

 

 

 

  

Intestine SynoPlate™ 
(Bottom) 

Centre Well 
(Zoomed) 

9-well 
Microphysiological 

System 

Upper Vascular 
Network 

Lower Vascular 
Network 

Intestine Tube 

Figure 6. Exploded diagram showing one of the 40 microfluidic systems of the intestine 
SynoPlate™. Each microfluidic system is comprised of a triad of three-well units. Each 
three-well unit has a dedicated inlet and outlet reservoir well connected via channels to a 
tissue scaffold in a shared centre well. This configuration allows tissue scaffolds to be 
populated with different cell types (e.g., endothelial or intestinal epithelial cells) that can be 
nourished with different media. Tissues converge in the centre well (zoomed view) to form 
a microscale organ with intestinal tissue and supportive vasculature embedded in an 
extracellular matrix. Figure adapted from a grant proposal prepared and submitted by 
Boyang Zhang, Ph.D.  
 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 28 

 SynoPlate™ fabrication is like IFlowPlate™ fabrication with a few key differences 

that will be highlighted in the text and accompanying figures of this section. The 11.6 cm 

x 7.5 cm x 0.081 cm polystyrene sheet that serves as the bottom of SynoPlate™ was 

sequentially patterned with two different sacrificial materials. First, the polystyrene sheet 

was patterned with a 3% (w/v) solution of sodium alginate (Alginic acid sodium salt from 

brown algae, low viscosity, Cat. # A1112, Sigma-Aldrich) in dH2O. Alginate is a 

polysaccharide derived from marine algae and is composed of 1,4-linked β-D-mannuronic 

acid (M) and α-L-guluronic acid (G) residues.48,49 Sodium alginate is dissolved in water to 

form a viscous solution that can be gelled at room temperature in the presence of divalent 

metal cations, such as  Cu2+, Ba2+, and Ca2+.48,49 With the SynoPlate™ PDMS mould bonded 

with the polystyrene sheet, the device was submerged in a bath of 3% (w/v) sodium alginate 

solution (Figure 7, step 2). Features were filled with alginate using the vacuum filling 

approach described in Section 3.1.6, with all steps occurring at room temperature. To 

minimize spillage of alginate solution inside the centrifuge during pulse centrifugation, a 

custom lid was made for the one-well plate into which the alginate solution was poured. To 

prepare this lid, 41 g of PDMS (SYLGARD™ 184) comprised of forty parts by weight 

elastomer base and one part curing agent was poured into the one-well plate lid and cured 

at room temperature. Using three elastic bands, the custom PDMS lid was secured to the 

one-well plate filled with alginate. After a 40-minute incubation of the device in the alginate 

bath at atmospheric pressure, all features were filled (Figure 7, step 2).  The alginate 

solution was discarded, and excess alginate was aspirated from inlet and outlet holes using 

a multichannel pipette attached to a vacuum aspirator.  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 29 

 The device was submerged in a bath of 5.5% (w/v) calcium chloride (Calcium 

chloride dihydrate, MW 147.01, Cat. # 223506, Sigma-Aldrich) and pulse centrifuged at 

1000 RPM. Unpopped air bubbles at the surface of the inlet/outlet holes were eliminated 

by injecting calcium chloride solution into the inlet/outlet holes using a syringe and 19 G 

needle. In the presence of divalent calcium ions, alginate polymers are cross-linked, and a 

sol-gel transition occurs almost immediately (Figure 7, step 3 – 4). The gelation takes place 

via an ion-exchange mechanism in which monovalent sodium cations are exchanged for 

divalent cations that induce association between GG blocks of neighbouring polymer 

chains.48,49 The resulting structure of the calcium-alginate gel is popularly represented by 

the “egg-box model” (Figure 8A).49 Among different divalent metal ions, such as copper 

and barium, the ability for alginate to bind calcium is comparatively weaker.48 Calcium was 

chosen to induce gelation because it allows rehydration/swelling in the presence of a 

sodium ion-containing solution (e.g., fibrin pre-gel solution) to occur with ease – this is 

desirable for our application. After overnight incubation at room temperature, the calcium 

chloride solution was discarded, and the device was cleaned by briefly immersing it in a 

dH2O bath to gently lift excess calcium-alginate gel from the PDMS and back surface of 

the device for removal. Excess water was absorbed with tissue before the device was placed 

inside the fume hood under a stream of compressed air to accelerate dehydration of the 

calcium-alginate gel. After 48 hours of drying, calcium-alginate features were 

microscopically examined (Figure 7, step 5).  Dehydrated calcium-alginate features 

exhibited volumetric shrinkage which has been attributed to the formation of “egg-box 

multimers” resulting in more tightly packed polymers.50 During air-drying, the increase in 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 30 

calcium ion concentration induces the association of junction zones (i.e., where two 

alginate chains cross-link to form an “egg-box dimer”) and therefore, the formation of 

“egg-box multimers” (Figure 8B).49,50  Dehydrated calcium-alginate features patterned on 

the polystyrene sheet were encapsulated with PEGDM using injection moulding as 

described in Section 3.1.6. SynoPlate™ was assembled as described in Section 3.1.7.  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 31 

 

 

Unfilled 

Filled 

Dehydrated Ca2+-alginate gel 

Figure 7. Schematic showing Part A of SynoPlate™ fabrication. Stepwise process 
to sequentially pattern a styrene sheet with alginate and PEGDM to generate a custom 
384-well microfluidic platform termed SynoPlate™.  

PEGDM-encapsulated  
Ca2+-alginate gel 

 

(2)  (3)  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 32 

 

 

 

 

 

 

 

 

Figure 8. Structure and swelling behaviour of calcium-alginate gel. Schematic 
illustration of the (A) egg-box model for calcium-alginate gel and (B) association and 
disassociation of junction zones in calcium-alginate gel during dehydration and 
rehydration, respectively. (C) Photomicrograph of the swelling behaviour of a calcium-
alginate fibre in saline solution, 200X. Figure (A) is reproduced from Ref. 49. Figure 
(B) is reproduced from Ref. 50. Figure (C) is reproduced from Ref. 48. 

A 

B C 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 33 

 

 

 

 

3.3 SynoPlate™ Cell Culture  

 SynoPlate™ was not used for any cell culture experiments; however, this paragraph 

will provide a brief description of how the plate is intended to be used as it gives context 

for work to minimize breakage of patterned calcium-alginate features (See Section 4.3.2) 

To prepare SynoPlate™ for cell culture, PEGDM is dissolved with dH2O and the plate is 

disinfected with 70% (v/v) ethanol (Figure 9, steps 10 – 11). Acellular fibrin pre-gel 

solution is cast in the centre well and encapsulates the patterned calcium-alginate feature. 

Fibrin pre-gel solution is a source of sodium ions as its primary constituent, fibrinogen, is 

dissolved in 1 X Dulbecco’s phosphate buffered saline (D-PBS, without calcium or 

magnesium, Cat. # 311-425-CL, WISENT Inc., Saint-Bruno, Quebec). In the presence of 

competing monovalent sodium ions which are non-crosslinking, the weak electrostatic egg-

box dimer-dimer interactions are readily disrupted and the calcium-alginate gel is 

Figure 9. Schematic showing Part B of SynoPlate™ fabrication. To assemble 
SynoPlate™, alginate/PEGDM-patterned styrene sheet is bonded to a standard 384-well 
bottomless plate using high viscosity PDMS as glue. Water is added to the wells to dissolve 
PEGDM. Finally, SynoPlate™ is disinfected with 70% (v/v) ethanol in preparation for cell 
culture.  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 34 

rehydrated and swells (Figure 8C).50 As a result, the calcium-alginate feature, initially 

patterned in two-dimensions, lifts from the polystyrene plate bottom and changes volume 

and shape in three dimensions inside of the fibrin gel.  

 The appeal of this approach, which combines 2D patterning with the stochasticity 

of swelling kinetics, is that it yields microtissues of highly reproducible shapes and sizes 

while introducing some anatomical variability. This parallels the robustness and plasticity 

that is characteristic of tissue morphogenesis during human development to generate organs 

of stereotypical, but not identical size and shape. As a high-throughput platform that could 

be applied to testing the safety and efficacy of drugs – a “clinical-trial-on-a-plate”, the 

inherent variability in SynoPlate™ microtissues is an attribute that is embraced as it 

simulates inter-individual anatomical variation. The importance of representing 

population-level variation is underscored by the drug approval process which involves 

large-scale randomized clinical trials to better estimate population-level treatment effects.  

 After gelation of the fibrin pre-gel solution, the re-hydrated calcium-alginate gel is 

degraded by perfusing D-PBS through the tissue networks. In the presence of a high 

concentration of salt ions, “egg-box dimers” are disrupted and the alginate polymers un-

crosslink and dissolve in aqueous solution. After the system is flushed, a moulded cavity 

within the fibrin gel is revealed. These independently addressable tissue-specific cavities 

are populated with endothelial and intestinal epithelial cells to form a 3D microtissue with 

perfusable branched vasculature and an intestine tube. Overall, the (1) water insolubility of 

the cross-linked alginate which enables the calcium-alginate gel to remain intact during 

PEGDM dissolution in water and 70% (v/v) ethanol disinfection, (2) 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 35 

dehydration/rehydration dynamics and (3) mild degradation conditions, make calcium-

alginate gel an idealistic sacrificial material for the described application.  

 

3.4 IFlowPlate™ Cell Culture 

 IFlowPlate™ is an array of 128 microfluidic systems each consisting of two 

reservoir wells connected via channels to a middle tissue well with a fibrin-based 

hydrogel51 (Figure 10). Hydrogel-encapsulated endothelial cells self-assemble into a 

vascular network. Anastomotic connections of vessels with microfluidic channels enables 

perfusion of the vascular network. The reservoir wells hold culture medium that is passively 

and bidirectionally circulated between the three wells using gravity-driven flow sustained 

by placing the plate on an interval rocker. Intestine cells are cultured as a monolayer on the 

surface of the hydrogel to give access to the luminal surface and emulate the anatomical 

organization of the intestine and associated microcirculation.  

 

 

 

 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 36 

 

 

 

 

 

 

3.4.1 Cells and Culture Conditions 

 

Endothelial Cells  

 Green fluorescent protein-expressing human umbilical vein endothelial cells (GFP-

HUVECs) were purchased from Angio-Proteomie (Cat. # cAP-0001GFP) and cultured in 

endothelial cell growth medium 2 (ECGM-2, Cat. # C-22111, PromoCell). HUVECs are 

the most widely studied and commonly used human endothelial cell type.52 Of note, more 

than half of the studies involving endothelial cell assembly in fibrin gel have used 

HUVECs.52 Given the interspecies differences in endothelial cell biology, the popularity of 

HUVECs can be explained by their human relevance and the ease with which they can be 

Reservoir  

IFlowPlate™ 

Reservoir 

Channel 

Hydrogel 

Figure 10. Exploded diagram showing one of the 128 microfluidic systems of 
IFlowPlate™. Each microfluidic system is comprised of three adjacent wells connected by 
channels. The middle well contains an ECM-based hydrogel. The height of the medium 
column in reservoir wells is manipulated to establish a hydrostatic pressure gradient that 
stimulates the passive flow of fluid down the gradient (i.e., from higher pressure to lower 
pressure). For indefinite perfusion, the plate is placed on an interval rocker platform that 
continuously re-establishes the medium height differential between reservoir wells. Figure 
adapted from Ref. 51.  



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 37 

acquired and cultured.52 Indeed, HUVECs are isolated from the veins of umbilical cords 

which are frequently discarded as biomedical waste, and they can be expanded to very large 

numbers in vitro. The stable expression of GFP by HUVECs enables real-time fluorescence 

imaging of live cells for continuous monitoring of vascular network formation and 

anastomoses with channels. For these reasons, GFP-HUVECs were exclusively used for 

the development of IFlowPlate™ vascular models. 

 

Fibroblasts 

 Human normal lung fibroblasts (ATCC® Cat. # PCS-201-013™, American Type 

Culture Collection, Manassas, VA) were cultured in Dulbecco’s modified Eagle’s medium 

(DMEM, Cat. # 319-005-CL, WISENT Inc., Saint-Bruno, Quebec) supplemented with 

10% (v/v) fetal bovine serum (FBS, Cat. # 098-150, WISENT Inc., Saint-Bruno, Quebec), 

1% (v/v) 1M HEPES (Cat. # 330-050-EL, WISENT Inc., Saint-Bruno, Quebec) and 1% 

(v/v) Penicillin-Streptomycin (Pen-Strep, 450-200-EL, WISENT Inc., Saint-Bruno, 

Quebec). Human normal lung fibroblasts are commonly cocultured with endothelial cells 

to support the spontaneous formation of microvascular networks in vitro.53-55 Fibroblasts 

stimulate the formation of microvascular networks and sustain them through the release of 

paracrine factors (e.g., hepatocyte growth factor), deposition of ECM proteins,54,55 and 

pericyte-like behaviour, as demonstrated by their close physical association with 

endothelial tubes.53,56 For these reasons, IFlowPlate™ vascular models were established by 

coculturing human normal lung fibroblasts and GFP-HUVECs in a 3D fibrin-based matrix. 

 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 38 

Caco-2 Cells 

 For modelling the intestine, the human intestinal colorectal adenocarcinoma cell 

line Caco-2 (C2BBe1, ATCC® Cat. # CRL-2102™, American Type Culture Collection, 

Manassas, VA) was used. The original Caco-2 cell line represents a heterogenous 

population of cells of which only a subset assemble a brush border upon reaching 

confluence.57 The presence of an apical brush-border membrane is a morphological 

indicator of a polarized and differentiated intestinal epithelial cell.58 Caco-2 BBe1 cells 

were cloned from the parental Caco-2 cell line to create a homogenous population of brush 

border expressing (BBe) cells.57 Relative to their sister clone Caco-2 BBe2, BBe1 cells 

achieve confluency with less culture time and are more easily dissociated in the presence 

of trypsin during cell passaging.57 For proof-of-concept studies in a nascent microfluidic 

platform, such as IFlowPlate™, the retention of some stem cell-like properties makes Caco-

2 cells a valuable surrogate for primary intestinal tissue which is more difficult to acquire 

and resource intensive and costly to culture. Caco-2 BBe1 cells, hereafter referred to as 

Caco-2 cells, were cultured in DMEM (with glucose and L-glutamine) supplemented with 

10% (v/v) FBS, 1% (v/v) 1M HEPES, and 1% (v/v) Pen-Strep. This formulation will 

hereafter be referred to as complete DMEM.  

 

Cell Passage Number  

 In all experiments, cells between passage 2 and 4 were used, as passage number is 

associated with phenotypic and genotypic changes which could have experimentally 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 39 

relevant consequences. For example, it has been reported that HUVECs older than passage 

5 form tip cells and intracellular vacuoles less frequently and tubulogenesis is delayed.59  

 A study of Caco-2 characteristics and passage number has shown that as passage 

number increases, growth rate, transepithelial electrical resistance (TEER) and sucrase 

activity also increase.22 Another study reported that early (P22)  and late (P198)  passage 

Caco-2 cells can differ as much as 50-fold in their glucose consumption rate and expression 

of sucrase-isomaltase.60  

  

Culture Conditions for Cell Lines 

 Cells were cultured in a humidified 37 °C, 5% CO2 incubator. Cells were expanded 

in T75 flasks until approximately 80% confluent. For GFP-HUVEC expansion, T75 flasks 

were coated with 10 mL of 0.2% bovine skin, type B gelatin (Cat. # A9418, Sigma-

Aldrich,) in D-PBS for 30 minutes at 37 °C.  

 For IFlowPlateTM bi- and tri-culture experiments involving GFP-HUVECs and 

fibroblasts, and GFP-HUVECs, fibroblasts, and Caco-2 cells, respectively, a 1:1 mixture 

by volume of ECGM-2 and complete DMEM was used for culture. For all experiments 

with fibrin gel, cell culture medium was supplemented with 1% (v/v) 2 mg/mL aprotinin 

(Aprotinin, Bovine Lung, Crystalline, Cat. # 616370, Sigma-Aldrich) in dH2O to slow 

protease-mediated fibrin degradation.  

 

3.4.2 Hydrogels 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 40 

 Hydrogels are three-dimensional networks of hydrophilic polymers that can hold 

large amounts of water.61 The cross-linkable nature of the polymers prevents their 

dissolution and maintains the three-dimensional structure of the hydrogel.61 The high water 

content and insolubility of the cross-linked polymers gives hydrogels physical 

characteristics similar to ECM of soft tissues making them well-suited for applications in 

research (e.g., 3D cell culture) and medicine (e.g., contact lenses, drug delivery systems, 

and scaffolds for tissue engineering).62,63  

 Hydrogels can be broadly classified as natural, synthetic, or semi-synthetic 

according to the origin of polymer constituents.63 Natural polymers derived from living 

tissue present high biofunctionality.62 Natural polymers (1) assemble into supramolecular 

architectures (e.g., fibres) that resemble the topography of native ECM, (2) contain cell-

binding motifs that support cell adhesion, migration, and activation of intracellular 

signaling pathways, and (3) are recognized and metabolically processed by cells allowing 

for cell-mediated remodelling and release of sequestered bioactive molecules (e.g., growth 

factors).62,64 Therefore, natural hydrogels can provide a bioinstructive environment in 

which cells can proliferate, differentiate, and develop into tissues that more closely 

resemble the native tissue.62  

 The structural and mechanical properties of natural hydrogels can be manipulated 

to elicit different cellular responses. For example, the microstructure (e.g., fibre thickness 

and density) of fibrin gels is determined by fibrinogen concentration and gelation time 

which is significantly affected by thrombin concentration.65 Manipulation of these 

parameters can be used to modulate matrix (1) pore size, and thus transport of 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 41 

macromolecules and nutrients for growth, survival, and cell-cell communication, and (2) 

stiffness.65,66 The effect of fibrin stiffness on cell behaviour has been widely investigated. 

Broguiere et al. showed that intestinal organoid expansion, defined as the frequency of cyst 

formation, was optimal in lower fibrinogen concentrations which corresponded to lower 

stiffness.67 Duong et al. showed that based on proliferation and spreading morphology, 

fibroblasts favoured fibrin matrices of lower stiffness.66 

 In this work, a diversity of cross-linkable natural polymers that can reversibly form 

gels in response to physical (e.g., temperature), chemical (e.g., ions, pH), or enzymatic 

stimuli (e.g., thrombin) were used (Figure 11).  Importantly, while distinct triggers are used 

to initiate polymerization, the polymers have in common polymerization conditions mild 

enough to take place in the presence of cells without inducing cytotoxic effects. Section 3.2 

introduced alginate which undergoes a reversible sol-gel transition in the presence of 

divalent metal ions (Figure 11A). Alginate hydrogel was used as a sacrificial material to 

mould fibrin gel into pre-defined geometries that when populated with cells guide the 

formation of 3D microtissues. This section will discuss the use of natural hydrogels, such 

as fibrin, Matrigel®, and collagen, as mimetics of soft tissue ECM for cell culture. Particular 

attention will be paid to the preparation of a fibrin-Matrigel® composite gel to serve as a 

suitable scaffold for establishing a vascular intestine model in IFlowPlate™. The application 

of gelatin to temporarily seal IFlowPlate™ microfluidic channels will also be discussed. 

 

 

 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 42 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fibrin 

 In vivo, fibrin plays a central role in haemostasis and wound healing making it well-

suited for use in vitro as a scaffold for the development of microvasculature.52,68 Following 

vascular injury, provisional fibrin matrices are rapidly assembled from fibrinogen, an 

abundant protein of blood plasma.68 Fibrin possesses cell-binding motifs that support the 

attachment of many cell types integral to the wound healing response, including platelets 

Figure 11. Cross-linkable natural polymers can reversibly form hydrogels in response 
to physical, chemical, and enzymatic stimuli. (A)  Alginate cross-links in the presence 
of divalent calcium ions via an ion-exchange mechanism. Figure reproduced from Ref. 42. 
(B) Enzyme-activated polymerization of fibrin occurs when thrombin cleaves A 
fibrinopeptides (FpA) exposing knobs that interact with complementary holes on 
neighbouring fibrin monomers. (C) Hierarchical self-assembly of collagen I occurs at 
neutral pH. (D) Intermolecular crosslinks between aqueous gelatin (from bovine skin) form 
upon cooling (< 30 °C). Matrigel® (not shown) is also representative of a thermoresponsive 
natural hydrogel.   

C pH

D TemperatureB Enzyme

A Ions

Alginate
Calcium

Gelatin



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 43 

that form clots to stop bleeding, immune cells that phagocytize dead/damaged tissue, 

fibroblasts that produce and deposit components of the basement membrane, and 

endothelial cells that rebuild the vasculature.52,68,69  

 From a biological performance perspective, the natural capacity of fibrin to promote 

the development of blood vessels is a large part of its appeal for use in endothelial assembly 

models. However, in selecting a hydrogel for IFlowPlate™, fibrin also had to be considered 

from a practical application perspective. The suspension of GFP-HUVECs and fibroblasts 

in a liquid pre-cursor solution requires that the gelation process is cell friendly. As shown 

in Figure 11B, polymerization of fibrin is initiated by the cleavage of A fibrinopeptides 

(FpA) by thrombin.70 The removal of A fibrinopeptides exposes knobs ‘A’ that are 

complementary to holes ‘a’. Knob-hole interactions give rise to fibrin oligomers which 

grow to become protofibrils that laterally aggregate to generate fibres.70 The enzyme-

activated polymerization of fibrin can occur at room temperature under physiological 

conditions making the process compatible with cells. Moreover, these conditions are 

compatible with the conditions required for gelation of Matrigel® which is a component of 

the fibrin-based gel used in IFlowPlate™, and gelatin which is loaded into fluidic channels, 

undergoes a sol-gel transition, and is used to prevent entry of the fibrin-based pre-gel 

solution. Another important feature of fibrin is the ability to slow its degradation by 

supplementing culture medium with aprotinin, a small molecule that is a competitive 

inhibitor for the active site of the protease plasmin.55,71,72 This preserves fibrin architecture 

for an extended period and allows for long-term cell culture. 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 44 

 Stock solutions of fibrinogen were prepared by dissolving fibrinogen from human 

plasma (Cat. # F3879, Sigma-Aldrich) in D-PBS to a concentration of 30 mg/mL or 10 

mg/mL. Stock solutions of 10 U/mL thrombin from bovine plasma (Cat. # T4648, Sigma-

Aldrich) were prepared by dissolving thrombin in 0.1% (wt/v) BSA. Stock solutions of 

0.1% (wt/v) BSA (Cat. # A9418, Sigma-Aldrich) were prepared by dissolving BSA in D-

PBS. Stock solutions of 2 mg/mL aprotinin (Bovine Lung, Crystalline, Cat. # 616370, 

Sigma-Aldrich) were prepared by dissolving aprotinin in dH2O. Stock solutions were sterile 

filtered and stored at −20 °C as single use aliquots. Culture medium was supplemented 

with 1% (v/v) aprotinin to inhibit fibrinolysis.  

 

Matrigel®  

 In the body, the basolateral surface of epithelial and endothelial cell monolayers is 

separated from connective tissue by the basement membrane.73 The basement membrane is 

a 50 to 100 nm layer of specialized extracellular matrix that provides structural support to 

cells and can affect changes in cell behaviour via outside-in signalling.73 The most abundant 

components of basement membrane are laminin, type IV collagen, nidogen and 

perlecan.73,74  

 The mouse Engelbreth-Holm-Swarm (EHS) tumour produces large amounts of 

basement membrane components that can be extracted, solubilized, and assembled into a 

gel when warmed to 24 to 37 °C.73,74 The stability and structural integrity of basement 

membrane depends on the interaction of its major components. Collagen IV and laminin 

individually self-assemble into networks that are linked by nidogen and perlecan.73 Sterile 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 45 

EHS-derived basement membrane commercialized as “Matrigel®”, a portmanteau of the 

words “matrix” and “gel”, is commonly used as scaffold to support cell survival and 

differentiation ex vivo.73,74   

 Pioneers in organoid development Sato et al. first reported that isolated intestinal 

stem cells, which are prone to anoikis outside of the normal tissue context, could be 

successfully cultured in Matrigel® to generate organoids that display the hallmarks of 

intestinal epithelium (i.e., crypt-villus compartmentalization, cell type composition, and 

self-renewal capacity).75 Matrigel® is rich in laminin – a key component of intestinal crypt 

basement membrane.75 Recent studies using soluble RGD (Arg-Gly-Asp) peptides as a 

competitive inhibitor suggest that integrin binding to RGD sites on matrix components, 

such as laminin, is necessary for organoid growth.76 Fibrin also contains RGD domains; 

however, in fibrin alone, colony formation efficiencies were significantly lower compared 

to fibrin gels supplemented with Matrigel®.76 Doubling or tripling RGD concentration by 

coupling RGD ligands to the fibrin matrix had no effect on organoid growth suggesting that 

RGD concentration on fibrin was optimal.76 It was hypothesized that Matrigel® possessed 

one or more components that provide the biological signals necessary for organoid 

growth.76 Further experiments using purified preparations of the main components of 

Matrigel® identified laminin as the major biological signalling molecule that is required for 

organoid growth.76  

 Matrigel® is a complex mixture of ill-defined chemical composition and high batch 

to batch variability. In-depth proteomic analysis of Matrigel® samples has revealed that 

Matrigel® is an assortment of identifiable and unidentifiable species with a range of 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 46 

molecular weights.77 In addition to the structurally important ECM proteins that constitute 

the bulk of Matrigel® (e.g., laminin and collagen IV), there are numerous other proteins 

including growth factors, transcription factors, binding proteins, and proteins with 

unknown roles in cell culture.77 These proteins may play a supportive role in the survival 

and growth of certain cell types, such as stem cells, that are more sensitive to traditional 

culture conditions.44 However, the ill-defined chemical composition and high batch to batch 

variability of Matrigel® could complicate the interpretation of experimental results and 

limit the application of  Matrigel® cell cultures.76-78  

 Compared to standard Matrigel®, growth factor-reduced Matrigel® is a more 

defined Matrigel® preparation that has been purified to a greater extent to reduce the 

abundance of growth factors (e.g., basic fibroblast growth factor, epidermal growth factor, 

insulin-like growth factor 1, transforming growth factor beta, etc.).77 Single use aliquots of 

undiluted phenol red-free growth factor-reduced Matrigel® (9.06 mg/mL, lot 0055015, Cat. 

# CACB356231, Corning®) were stored at –20 °C and thawed on ice before use. 

 

Fibrin-Matrigel®   

 Fibrin-Matrigel® composite gels were prepared by mixing 5 mg/mL fibrinogen 

(with or without cells) with 10% (v/v) phenol red-free growth-factor reduced Matrigel®. 10 

mg/mL fibrinogen stock solution stored at –20 °C was warmed in a water bath and diluted 

with D-PBS to a final working concentration of 5 mg/mL. Matrigel® (9.06 mg/mL, lot 

0055015) was thawed on ice and used undiluted. 10 U/mL thrombin stock solution stored 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 47 

at –20 °C was thawed at room temperature and diluted with 0.1% (wt/v) BSA to a final 

working concentration of 1 U/mL.  

 125 µl of 5 mg/mL fibrinogen was dispensed into a microcentrifuge tube. To this, 

12.5 µl of cold, undiluted Matrigel® was added. Matrigel® was kept cold to prevent 

premature gelation at warmer temperatures. 25 µl of 1 U/mL thrombin was added to the 

fibrinogen-Matrigel® solution to initiate polymerization. The final concentrations of 

fibrinogen and Matrigel® were 3.85 mg/mL and 7.7% (v/v), respectively.  

 

Collagen 

 Human primary intestine monolayer models have been accomplished by culturing 

crypts/stem cells on collagen hydrogels.25,28 Collagen I is commonly used for cellular 

scaffolds because of its biocompatibility and  availability which can be attributed to its 

abundance in natural ECM and relative ease of extraction with minimal contamination by 

other proteins.79 Figure 11C shows a schematic illustration of the hierarchical assembly of 

a collagen fibre from collagen I polypeptide chains. Three collagen I polypeptide chains 

twist together to form a triple helix structure referred to as tropocollagen.79,80 Tropocollagen 

molecules assemble into a collagen fibril, and many collagen fibrils bundle together to form 

a collagen fibre.79,80 Finally, crosslinking of collagen fibres generates a matrix structure 

that in the presence of a water-based solvent will swell to produce a hydrogel.79  

 Polymerization of acid-solubilized collagen polypeptides is initiated upon 

neutralization of the solution.79 Collagen polymerization progresses in two phases: the 

nucleation phase during which collagen molecules assemble into fibrils, and the growth 



M.A.Sc. Thesis – K.L. Hayward; McMaster University – Chemical Engineering 

 48 

phase during which crosslinking occurs.81 Reaction kinetics are temperature-dependent and 

can have a significant effect on hydrogel properties and cellular response.79,81,82 At higher 

temperatures, polymerization rate is accelerated resulting in a shorter nucleation phase and 

fibres with reduced diameter.79,81,82 Conversely, lower temperatures prolong the nucleation 

phase producing thicker fibres.81,82 Xie et al. found that compared to collagen gels 

polymerized at 4 °C, collagen gels polymerized at 37 °C form thinner, softer fibres that 

support increased spreading and proliferation of human mesenchymal stem cells.82 

Collagen gels polymerized at 37 °C have been reported to support the formation of a self-

renewing primary intestinal epithelial monolayer.25,28 In the interest of replicating these 

results for the development of a primary intestine model with an accessible lumen in 

IFlowPlate™, collagen gels were prepared and polymerized at 37 °C. 

 Collagen gels were prepared from acid-solubilized 10.80 mg/mL rat tail collagen 

type I (Cat. # CACB354249, Corning®). Neutralization solution was prepared with 10X 

Medi