Hydrogel-based 3D semi-circular microvascular bbb-on-chip with integrated sensors for barrier assessment through TEER measurements

MASTER’S THESIS

HYDROGEL-BASED 3D SEMI-CIRCULAR MICROVAS- CULAR BBB-ON-CHIP WITH INTEGRATED SENSORS FOR BARRIER FUNCTION ASSESSMENT THROUGH TEER MEASUREMENTS

L.E. de Heus

Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS) BIOS Lab on a Chip

Exam committee:

dr. L.I. Segerink (chairman)

ir. M.P. Tibbe (daily supervisor)

dr.ir. A.M. Leferink (External member)

“Not everything that can be counted counts, and not everything that counts can be counted.”

Albert Einstein

Abstract

Background: Realistic models that optimally mimic the blood-brain barrier (BBB) are essential for researches regarding drug passage across the barrier that aim for de- veloping adequate neuropharmaceuticals. Over the past decade, in vitro BBB models have evolved from static two-dimensional (2D) Transwell systems towards dynamic hydrogel-based three-dimensional (3D) BBB-on-chip systems with physiologically relevant matrix stiffness. However, this next generation BBB models lacks integra- tion of a measurement setup for assessing barrier function using trans endothelial electrical resistance (TEER). Here, we propose a hydrogel-based 3D microvascular BBB-on-chip with integrated sensors for barrier function assessment through TEER.

Method: Although a fully functional device was not achieved during the scope of this thesis, an initial concept design was optimized in an iterative process, af- ter which individual components were developed an experimentally tested with preliminary studies. Hyaluronic-based hydrogels suitable for this application were identified by their ability to adopt channel features from a mould and subsequently seal on a chemically functionalized glass substrate. To investigate the suitability of the selected hydrogel for cell culture purposes, human cerebral endothelial cells (hCMEC/D3) were cultured on hyaluronic-based hydrogels complemented with gelatin to promote cell adhesion. Cell behaviour was quantified and qualified by assessing metabolic activity and cell morphology, respectively. Finally, the devel- oped sensor comprised of electrodes incorporated in a glass slide was characterized using hydrogel-based electrolytes with increasing conductivity.

Results and discussion: The optimized novel hydrogel-based chip design proved appropriate for imprinting hydrogels with channel features that subsequently could be aligned with electrodes integrated in a glass substrate. Sealing of the open chan- nels on the glass substrate was not achieved. hCMEC/D3 cultured on 5% hyaluronic- based hydrogel complemented with gelatin at a final concentration of both 0.5 and 1.0 mg/ml showed a healthy characteristic cobblestone-like morphology with simi- lar cell viability. Measured resistance values roughly followed the trend for increas- ing conductivity and electrode spacing as expected from the theoretical cell constant, however appeared to lose sensitivity for high conductivity values.

Conclusion: During the scope of this thesis, first steps were made in achieving a

hydrogel-based microfluidic chip with integrated sensors. Open channels could be

fabricated in hyaluronic-based hydrogel, which proved suitable for endothelial cell

culture. Moreover, preliminary tests with the electrode configurations showed its

potential for ultimately TEER measurements on chip. Future research on channel

sealing is vital for completing the device such that it facilitates cell culture and TEER

measurements.

Acknowledgements

I would like to thank a lot of different people for their help during my master’s

thesis. First of all, I would like to thank Albert van den Berg for giving me the

opportunity of writing my master’s thesis in the BIOS Lab-on-a-Chip group. I would

like to thank all the members of the examination committee, Loes Segerink, Anne

Leferink and Martijn Tibbe. During the meetings with Loes, I always gained a lot

of knowledge but mostly interest in a for me new topic, electrodes, which I really

appreciate. I would like to thank Anne for her overall guidance and sometimes

the push in the right direction, which I am very grateful for. And ofcourse, a big

thanks to Martijn. I really enjoyed working with you! For me, it was the perfect

combination between doing research while at the same time having a lot of fun and

ofcourse listening to the best music in the labs. You always kept me motivated and

focused whenever I needed. Besides the committee, I would like to thank Johan

Bomer and Hans de Boer who helped me with the design and fabrication of the

electrodes and holders. I would like to thank everyone of the BIOS Lab-on-a-Chip

group for the laughs and inspiring talks in the coffee corner, during the trips to the

fish stand, the weekly football matches, or any other activity during this amazing

time. And last but definitely not least, I would like to thank Rob, who had to cope

with me during the last months.

Contents

Abstract v

Acknowledgements vii

1 Introduction 1

1.1 Report structure . . . . 3

2 Theory 5 2.1 Hyaluronic-based hydrogel . . . . 5

2.2 Fabrication of hydrogel-based microfluidic chips using soft-lithography in tissue engineering . . . . 5

2.3 TEER . . . . 7

2.3.1 In vivo TEER measurements . . . . 7

2.3.2 In vitro TEER measurements . . . . 9

Cell constant . . . 11

3 Design optimization 15 3.1 Hydrogel-based microfluidic chip . . . 15

3.1.1 Initial concept . . . 15

3.1.2 Observed limitations . . . 15

3.1.3 Proposed adaptations . . . 16

3.2 Electrode configuration . . . 17

3.3 Optimized novel design . . . 19

3.3.1 Strengths and limitations . . . 19

4 Hydrogel chip fabrication by soft-lithography 23 4.1 Introduction . . . 23

4.2 Experimental methods . . . 24

4.2.1 Silicon mould fabrication . . . 24

4.2.2 Hydrogel preparation . . . 24

4.2.3 Surface functionalization . . . 26

4.2.4 Hydrogel chip fabrication . . . 26

4.2.5 Analysis channel feature transfer . . . 27

4.2.6 Analysis sealing hydrogel channels . . . 27

4.3 Results and Discussion . . . 28

4.3.1 Hydrogel channel fabrication . . . 28

4.3.2 Channel feature transfer . . . 28

4.3.3 Sealing hydrogel channels . . . 32

4.4 Conclusion . . . 33

5 Endothelial cell morphology and viability on hyaluronic-based hydrogel 35 5.1 Introduction . . . 35

5.2 Experimental methods . . . 36

5.2.1 Hydrogel preparation . . . 36

5.2.2 hCMEC/D3 cell culture . . . 36

5.2.3 Fluorescent staining . . . 36

5.2.4 Viability assay . . . 36

5.3 Results and Discussion . . . 37

5.4 Conclusion . . . 40

6 Electrode characterisation 41 6.1 Introduction . . . 41

6.2 Experimental methods . . . 41

6.2.1 Electrode fabrication . . . 41

6.2.2 Hydrogel preparation . . . 43

6.2.3 Two terminal impedance spectroscopy . . . 43

6.2.4 Data analysis . . . 43

6.3 Results and Discussion . . . 44

6.3.1 Theoretical cell constant . . . 44

6.3.2 Experimental resistance . . . 44

Experimental vs. theoretical cell constant . . . 47

Resistance vs. ion concentration . . . 48

Resistance vs. electrode spacing . . . 49

6.4 Conclusion . . . 52

7 Future outlook 53 7.1 Channel sealing . . . 53

7.1.1 Sacrificial hydrogel element . . . 53

7.1.2 Glass functionalization with molecular hydrogel coating . . . . 54

7.2 TEER measurements . . . 55

7.3 Implementation controllable shear stress variable . . . 56

8 Conclusion 57 Bibliography 59 A Hydrogel chip fabrication 63 A.1 Silicon mould fabrication . . . 63

A.2 Exploring limited sealing of hyaluronic-based hydrogel on tyramine

functionalized glass slide . . . 63

B Electrode characterisation 65 B.1 Reconstruction K

from Linderholm et al. . . . 65 B.2 Raw impedance data . . . 66 B.3 COMSOL model electrode characterisation . . . 69

C Future outlook 71

C.1 COMSOL model TEER . . . 71 C.2 COMSOL model shear stress . . . 73

D 3D view final designs 75

List of Abbreviations

BBB Blood-brain barrier

TEER Trans Eendothelial Electrical Resistance

2D two-dimensional

2D three-dimensional ECM Extracellular matrix

hCMEC/D3 human Cerebral Microvascular Endothelial Cell /D3 cell line EC Endothelial Cells

HA-ta Hyaluronic Acid - tyramine Dex-ta Dextran - tyramine

Gel-ta Gelatin - tyramine HRP Horse Radish Peroxidase PBS Phosphate Buffered Saline

RT Room Temperature

PDMS Polydimethylsiloxane

PMMA Poly (methyl methacrylate)

Chapter 1

Introduction

The world’s population’s average age and life expectancy is increasing, accompa- nied with the rising trend in prevalence of chronic diseases, such as neurodegener- ative diseases. According to Alzheimer’s Disease International, 46.8 million peo- ple were estimated to suffer from Alzheimer’s disease in 2015, a number that is expected to increase even more. [1] Consequently, the development of adequate neuropharmaceuticals is required. However, drug delivery into the central nervous system (CNS) exposes a major challenge, since molecular diffusion into the brain tissue is hindered by the so called blood-brain barrier (BBB). This barrier partitions brain interstitial fluid from peripheral blood. It is composed of highly specialized endothelial cells forming the brain capillaries that are closely interacting with astro- cytes, pericytes, microglia and a base membrane. Restricted uptake of compounds is highly controlled by the endothelial cells with their characteristic continuous tight junctions, lack of fenestrae, low pinocytic activity and high concentration of efflux transporters. [2] Therefore, the identification of neuropharmaceuticals suitable for drug delivery across the BBB remains a major challenge.

As studies regarding the drug development towards neuropharmaceuticals con- tinue, realistic models are essential to investigate drug passage across the barrier.

Currently, the gold standard for testing drug compounds in pharmaceutical indus- try are in vivo animal models. [3] Advantages of animal models are that they provide an elaborate view on the overall response, including the cellular, tissue, organ and systemic level. Nevertheless, as a result of species-to-species variations, successful outcomes of animal studies do not guarantee similar results in human studies. [4]

In addition, animal studies are costly, labour-intensive and ethically questionable.

Therefore, researchers focus on developing in vitro BBB models that both anatomi- cally and physically mimic the BBB.

Due to the BBB’s complexity, a lot of research is done on simplified in vitro mod-

els. In the past years, multiple Transwell

systems aiming at mimicking the BBB

have been investigated and developed. In these systems, endothelial cells are seeded

in the upper compartment and astrocytes in the lower compartment, partitioned by a

porous membrane. Morphological and physiological characteristics of the cells and

the entire barrier, such as the permeability and transendothelial electrical resistance

(TEER) could be determined, allowing proper evaluation of the BBB model. [5], [6]

Although Transwell

systems are easy to use, facilitate various assays to quantify or qualify barrier functionality and are relatively cheap, they lack, amongst others, the implementation of flow generated shear stress, one of the main factors responsible for endothelium alignment and tightness in the BBB. [7], [8]

BBB-on-chip systems are able to address the aforementioned shortcoming. Gen- erally, these chips are comprised of two perpendicular channels, representing the apical and basolateral compartment, separated by a semi-permeable membrane on which partitioned two-dimensional (2D) growth of endothelial cells and astrocytes or pericytes exposed to shear stress is facilitated. Similar to the Transwell

set up, these systems allow for easy morphological characterization of cells using mi- croscopy and barrier function assessment through TEER measurements and perme- ability assays. [9], [10] Further improvements on this design have been reported by Sellgren et al. [11] They inserted collagen hydrogel loaded with astrocytes in the basolateral channel to mimic the physically relevant stiffness of the brain extracel- lular matrix (ECM), which is known to affect astrocyte activation and subsequently their ability to modulate BBB function. [12] However, one major limitation of this general BBB-on-chip design is the creation of a 2D endothelium instead of a three- dimensional (3D) vessel-like structure.

Currently, the focus in BBB-on-chip designs increasingly shifts towards integrat- ing 3D structures, mostly by using hydrogels that have stiffnesses in the same order of magnitude as brain ECM. Brain microvasculature has predominantly been mim- icked by circular shaped channels in collagen hydrogels produced by viscous finger patterning and using microneedles. [13], [14] These designs enabled implementa- tion of shear stress and allowed qualification and quantification of barrier functions by imaging and permeability assays using fluorescently labelled compounds. How- ever, TEER measurements on these 3D brain microvasculatures on chip have not been reported to date.

In this study, a hydrogel-based 3D semi-circular microvascular BBB-on-chip is

proposed, with integrated sensors for barrier function assessment through TEER

measurements. Using soft-lithography, open channel features are aimed to be trans-

ferred into a hyaluronic-based hydrogel and sealed on a glass substrate. Upon en-

capsulating astrocytes or pericytes in the hydrogel, this chip would allow for a co-

culture with human brain endothelial cells lining the channel wall. The total de-

sign enables visualization of morphological and physical characterizations using im-

munofluorescence staining of cell specific proteins whereas permeability assays pro-

vide insight in the diffusion through the endothelial cell layer. Moreover, incorpo-

ration of electrodes in the glass substrate aims at facilitating a standardized method

for assessing barrier function by measuring TEER according to the leaky cable the-

ory as performed by Crone et al. [15] During the scope of this thesis, first steps were

made in achieving this goal. We have developed a combination of devices that allow

for channel feature transfer in hyaluronic-based hydrogels that subsequently could

be aligned with electrodes integrated in a glass substrate. The hydrogel proved

suitable for conventional endothelial cell culture purposes, which was concluded from the characteristic cobblestone morphology and active remodeling capacity of the hCMEC/D3 cell line. Subsequently, preliminary tests were performed for char- acterization of the proposed electrode configuration and showed to potentially be suited for impedance spectroscopy measurements in the developed device. Future research on channel sealing is crucial for completing the device such that it facili- tates cell culture in the hydrogel-based microfluidic chip and subsequently allows for performing TEER measurements.

1.1 Report structure

In this master thesis, Chapter 2 will elaborate on theory about previous work regard-

ing the hyaluronic-based hydrogel and hydrogel-based microfluidics in general, and

some background on methods to determine the TEER of endothelium in vivo as well

as in vitro. In Chapter 3, the design of the hydrogel-based chip fabrication method

is shortly discussed together with the design of the integrated electrodes. Since a

complete functional hydrogel-based microfluidic chip was not achieved during the

scope of this thesis, experiments done on imprinting channel structures in hydro-

gel together with channel sealing, endothelial cell morphology and viability on the

hydrogel and characterisation of the electrode configuration are addressed individ-

ually in Chapter 4, 5 and 6, respectively, each with individual results, discussions

and conclusions. The final chapter gives an outlook on future work.

Chapter 2

Theory

2.1 Hyaluronic-based hydrogel

Hydrogel-based BBB models are mostly comprised of collagen type-I. [14], [16] How- ever, this abundant ECM component is hardly present in the brain tissue. Therefore, a more relevant hydrogel type is hyaluronic acid, one of the main components of the brain ECM. [17] To date, various applications of hyaluronic acid (HA)-based hydrogels in tissue engineering have been reported, with an enormous variety of modifications and crosslinking mechanisms that allow for highly tunable hydrogel properties suitable for, amongst others, cartilage repair [18]. The hyaluronic-based hydrogel utilized in this thesis is composed of dextran and hyaluronic acid both functionalized with a tyramine group (Dex-ta and HA-ta), which allows for enzy- matic crosslinking by horse radish peroxidase (HRP) with (H

O

) as oxidant of HRP.

[18]

2.2 Fabrication of hydrogel-based microfluidic chips using soft-lithography in tissue engineering

Fabrication of channel features in hydrogel-based microfluidic chips has not only been investigated for blood-brain barrier model purposes, but also for creating in vitro microvessels in general. Micromoulding methods used to obtain channel fea- tures in a hydrogel are typically stamping, casting or injecting hydrogel in a mould, after which the imprinted open channels are sealed by various mechanisms includ- ing shortly melting the imprinted hydrogels [19], chemical crosslinking [20] or ap- plying pressure [16].

Ling et al. developed an agarose hydrogel microfluidic chip by moulding agarose on a SU-8 silicon wafer patterned with channel features. [19] (Figure 2.1a) Subse- quent heating of the moulded agarose layer as well as an additional flat agarose layer to a temperature of 71

C for only 3 seconds, enabled them to seal the open channels. The total method allowed for the creation of hydrogel channels with di- mensions as small as 50 μm by 70 μm, encapsulating amongst others hepatocytes.

Zheng et al. created a device in which collagen type-I was imprinted with mi-

crofluidic structures upon injecting the solution in a holder with on the bottom a

F

2.1: Examples of hydrogel-based microfluidic chips fabri- cated using soft-lithography. Adapted from A) Ling et al. [19], B)

Zheng et al. [16] and C) Qiu et al. [20].

patterned PDMS stamp. [16] (Figure 2.1b,i) In this holder, an additional outlet was integrated to allow trapped air to escape and two steel pins were inserted to define the inlet and outlet for the hydrogel channels. After gelation, the open channels were mechanically sealed on a flat collagen type-I layer (ii) by applying pressure and cells were seeded inside the hydrogel channels (iii). Medium was provided through a PDMS layer placed on the inlet and oulet that served as reservoir (iv). This design allowed for studying endothelium permeability and angiogenesis in 100 μm x 100 μm hydrogel channels.

A variation of this method was developed by Qiu et al. (Figure 2.1c) [20] Their agarose-gelatin hydrogel mixture was casted in a silicon master at 65

C, physically crosslinked at 4

C and transferred onto a 200 μm hydrogel layer that was crosslinked with a silane functionalized glass slide. The total system was finalized by chemi- cally crosslinking all layers via gelatin through carbodiimide chemistry. Using this method, they managed to fabricate microvascular-sized hydrogel channels with more complex branching geometries mimicking the capillary beds with channel diame- ters down to 20 μm. This hydrogel-based microfluidic chip facilitated endothelial cell culture in physiological relevant matrix stiffnesses, channel geometries and di- mensions corresponding to the in vivo microvascular environment.

Although numerous variations of hydrogel-based microfluidic chip devices have been reported, a common factor can be recognized in the fact that open channel structures are predominantly sealed on an additional flat layer of the same hydro- gel. The latter approach is inapplicable for the hydrogel-based microfluidic chip as proposed in this thesis, since the TEER measurement sensor requires the open hy- drogel channels to be in direct contact with the electrodes incorporated in the glass substrate. Therefore, an alternative method for sealing of the hydrogel channels is required.

2.3 TEER

2.3.1 In vivo TEER measurements

First in situ TEER values have already been reported back in 1981 and 1982 by Crone

et al. [15] They have performed a measurement technique that enabled them to

determine the electrical resistance along a frog brain capillary by injecting a small

alternating current and measuring the voltage drop along the capillary. The cur-

rent injection and voltage measuring electrode circuits were isolated by using a four

electrode set up with the current injecting and voltage measuring glass micro elec-

trodes inserted into the brain capillary and the corresponding silver/silver chloride

(Ag/AgCl) reference electrodes connected on the brain surface. An overview of

their experimental setup is depicted in Figure 2.2. This method is based on the leaky

cable theory that states that an electric current traveling through an electrolyte cylin-

der covered with an isolating layer also leaks through that ion permeable isolating

F

2.2: Experimental set up as perfomed by Crone et al. [15].

TEER measurements of frog brain capillaries was done using a cur- rent injecting and potential measuring electrical circuit. Integration of a lock-in amplifier allowed detection of the signal’s potential with a frequency corresponding to the 2.5 Hz current pulse as was injected

into the brain microvessel. Copied from [15].

layer. When applying this general definition to a microvessel, it can be interpreted as the intracapillary potential along a blood vessel decaying due to ionic permeabil- ity of the endothelial cells lining the vessel wall, thereby following an exponential function:

V ( x ) = V ( 0 ) e

(2.1)

where V(x) is the intracapillary potential at distance x from the current injecting electrode (V(0)) and λ is the so called length constant. This constant is a measure for how rapidly the potential drops in longitudinal direction and is related to the TEER, blood resistivity (ρ), cross sectional area (A) and circumference (C) of the cor- responding vessel according to the following equation:

TEER = ^ρ

A λ

C = ^ρ

πr

λ

2πr (2.2)

with r the radius of the blood vessel. Simplification of the original three dimen-

sional (3D) phenomena to this one dimensional (1D) model is justified when several

assumptions hold true. These include amongst others a radially negligible poten-

tial difference and a well conducting electrolyte surrounding the capillary. Thus, the

leaky cable theory can be used for estimating TEER values of small diameter blood

vessels only. The latter assumption is experimentally corrected for by superfusing

the investigated capillaries with a highly conductive solution.

In summary, Crone et al. proved that TEER values determined using the leaky cable theory are an indicative for endothelial barrier tightness. However, since sim- plified BBB models are predominantly 2D and in vitro, a translation of this 3D in vivo measurement method towards a 2D in vitro measurement set-up was required.

2.3.2 In vitro TEER measurements

TEER measurements have already been widely implemented in in vitro BBB mod- els to assess the barrier tightness. In 2D semi-permeable membrane based systems, TEER measurements are performed based on Ohm’s law, where a voltage is applied between electrodes positioned on both sides of the membrane and the resulting cur- rent is measured. Subsequently, the resistance can be calculated using Equation 2.3:

R

= ^U

I (2.3)

where U is the applied voltage, I the resulting current and R

the total resistance of the system which is reflected by the endothelium and the semipermeable membrane:

R

= R

− R

(2.4)

indicating that these TEER measurements require a reference measurement to deter- mine the resistance of the semipermeable membrane. Finally, from this data TEER values can be obtained, which scale linearly with the surface area (A) of the semiper- meable membrane and endothelium and are mostly reported in Ωcm

:

TEER = R

· A (2.5)

However, since the electrical behaviour of a cell is frequency dependent, impedance

spectroscopy provides further insight in the endothelium under investigation. As

shown in Figure 2.3a, the applied current can pass paracellular through the tight

junctions and transcellular through the cell layer, of which the latter pathway offers a

parallel resistive and capacitive pathway as a result of the lipid bilayer. Additionally,

a capacitive and resistive component are resulting from the electrode double layer

and electrolyte resistivity, respectively. The impedance spectrum of such a system

is depicted in Figure 2.3b, from which the characteristic double layer capacitance

(C

), the TEER plateau, lipid bilayer capacitance (C

) and resistance of the medium

(R

) can be recognized.

F

2.3: A) Schematic representation of the electrical circuit com- ponents contributing to the impedance as measured across an en- dothelium, further specified in a paracellular and transcellular path- way. B) Simplified electrical circuit (2) with corresponding impedance spectrum (1). R

: resistance of the medium, R

resistance of the medium, C

: capacitance cell layer, C

: capacitance double

layer electrodes. Adapted from [21].

Practical implementation of TEER measurements in Transwell

systems involves manually inserting electrodes in both fluidic compartments, whereas microfluidic system often have integrated electrodes on either side of the semi-permeable mem- brane in the channels. For the latter application, the position of the electrodes in the microfluidic channels is of great importance, preferably, as close to the semi- permeable membrane as possible to reduce the total resistance of the system. The resistance of an electrolyte (R

) in a microfluidic channels can be estimated by:

R

= ρ

l

A

(2.6)

with ρ

the electrolyte resistivity and l

and A

the microfluidic

channel length and cross sectional area, respectively, which indicates that the small

cross sectional area result in high resistances.

Cell constant

Results from impedance spectroscopy measurements not solely reflect the capaci- tance and resistance of the material, but also the electrode configuration and geom- etry of the sensor. The influence of the latter is summarized in one term, namely the cell constant (K

), which is proportional to the measured resistance (R

) and the material’s conductivity (σ

), assuming a homogeneous electrolyte and infinite thin electrodes:

R

= ^K

σ

(2.7)

Electrode configurations integrated in microfluidic set-ups are mostly parallel or coplanar. For parallel electrodes of the same surface area (A) at a perpendicular distance (D), the cell constant can simply be described by:

K

= ^R

ρ

= ^D

A (2.8)

However, the cell constant for two coplanar electrodes situated in a microfluidic channel in direct contact with medium can be described according to Equation 2.9 and 2.10. [22]

K

= ² L

K ( k )

K ( k

) ^(2.9)

K ( k ) =

Z

1

1

p ( ₁ − t

)( ₁ − k

t

) ^{dt (2.10)} with L the length of the electrodes and k and k’ dependent on the spacing between the electrodes (s) and width of the electrodes (w) as expressed in Equation 2.11 and 2.12:

k = ^s

s + w (2.11)

k

= ^p 1 − k

(2.12)

It has to be noted that these equations are based on the assumption of a semi-infinite medium on top of the electrodes. However, this is not the case in most microfluidic channels, where the electrode widths and spacing are not negligible in comparison to the channel height. A more accurate approximation of the cell constant was pre- sented by Lindholm et al. which allowed compensation for the channel height. [23]

For this, two conformal mappings are performed prior to using the final Schwarz-

Christoffel transformation as already presented in Equation 2.10. First, the sensor

geometry is simplified to a two dimensional (2D) design in the Z-plane with real

and imaginary as illustrated in Figure 2.4a. Integration of the height of the chan-

nel is done by replacing the left part of the symmetric channel with a conductor at

the centre (Figure 2.4b). Subsequently, the first transformation is performed to map all coordinates onto the real axis of the first intermediate plane, the U-plane, by us- ing a sine function (Figure 2.4c). The Schwarz-Christoffel transformation requires electrodes to be of the same width and symmetric around the origin of its plane.

Since this is not the case for the geometry as mapped on the U-plane, an additional mapping is needed. The coordinates of the U-plane are transformed using a bilinear function to the second intermediate plane, the V-plane (Figure 2.4d). The constants A, B, C and D need to be determined such that the previously stated requirements are met. This results in an expression for v4, of which the reciprocal is the modulus for the Schwarz-Christoffel transformation from the V-plane.

F

2.4: An overview of conformal mapping transformations of the electrode geometry (A) required to determine the cell constant theoretically. Due to symmetry, the left part of the channel can re- placed by a conductor (B, z1 and z2) with the same height as the mi- crofluidic channel (h). Second, a sine transformation is done to map all coordinates on the real part of the intermediate U-plane (C). Bilin- earization is then required to bring back symmetry (D), to then allow for a final mapping procedure into a plane parallel conductor (E) us-

ing Schwarz Christoffel mapping. Adapted from [23].

v

= ( u

− u

)( u

− u

)

u

( u

+ u

− 2u

) + 2 p

( u

− u

)( u

− u

)( u

− u

)( u

− u

) + u

u

+ u

(− 2u

+ u

) (2.13)

k = ¹

v

(2.14)

and can be filled in in Equation 2.10. The final step to determine the cell constant

is multiplying Equation 2.9 by a factor 2 to compensate for the fact that these math- ematical expressions only describe half of the microfluidic channel, resulting in the cell constants that can be described by Equation 2.15 (per transversal length, elec- trode length L) or 2.16 (dimensionless), respectively.

K

= ⁴ L

K ( k )

K ( k

) ^(2.15)

K

= 4 K ( k )

K ( k

) ^(2.16)

Chapter 3

Design optimization

The project started with a conceptual design of a hydrogel-based microfluidic chip for mimicking blood vessels with expansion possibilities to perform TEER measure- ments. During this project, this concept was optimized resulting in the novel chip design discussed in this report. First, the optimization of the concept is elaborated on, structured in 1) initial concept, 2) observed limitations and 3) proposed adapta- tions. Second, the design of the electrode configuration for the TEER measurements is discussed. Finally, all proposed adaptations are integrated into the final design, which is presented at the end of this chapter.

3.1 Hydrogel-based microfluidic chip

3.1.1 Initial concept

The design and fabrication process of the hydrogel-based chip as illustrated in Fig- ure 3.1 formed the starting point of this project. It is comprised of two separate com- ponents: 1) a 3D printed holder with a lowered plateau in which a silicon mould with channel features for imprinting channel structures in hydrogel is placed; 2) a polydimethylsiloxane (PDMS) mould obtained by replica moulding using a mi- cromilled poly-(methyl methacrylate) (PMMA) master, which snugly fits into the mould holder. The injection set-up is assembled by transferring the PDMS mould onto the chip holder, thereby creating a cavity which sets the boundaries for the hy- drogel that is injected with a needle through the PDMS (Figure 3.1a, b). After gela- tion, the PDMS mould containing hydrogel with imprinted channel features was lifted from the mould holder and transferred onto a glass slide. Glass was the mate- rial of choice since it enables imaging of the hydrogel channels as well as electrode incorporation.

3.1.2 Observed limitations

It proved difficult to fabricate reproducible hydrogels with imprinted channel fea-

tures and to subsequently seal these channels on glass using this method. Com-

plications in hydrogel injection were caused by the flexibility of the PDMS and the

entrapped air in the cavity that compromised the geometry of the hydrogel channel

F

3.1: A hydrogel mixture is injected into injection setup, com- prised of the silicon mould holder below and the gel support on top (B). The channel structure is imprinted into the hydrogel by soft- lithography using a silicon mould with positive channel features. Af- ter gelation, the top part is lifted-off, leaving open hydrogel channels (C). The hydrogel chip can be placed on a glass slide with or without

electrodes to seal the channels (D).

features. The negative effect of the flexibility was that an uneven pressure distribu- tion across the PDMS mould during manual injection would result in deformation of the mould. Consequently, the PDMS mould would release from the mould holder so that the cavity was not properly sealed and additional air could enter the cavity.

Moreover, air entrapped in the cavity upon assembling the injection set-up could not escape in this design. All compromised the geometry of the hydrogel channels.

Second, complications in sealing the hydrogel channels were caused by the mechan- ical instability of the hydrogel on glass and the lack of binding between glass and the hydrogel. Upon addressing the hydrogel channels with a pipette tip, the flexi- ble PDMS mould would deform the hydrogel, resulting in the hydrogel to release from the glass surface and leaking its content. The latter could not be prevented by chemical functionalization of the glass surface to promote gel adhesion. In addition to aforementioned limitations, a self-contained constraint is the fact that manually aligning the channel to the sensor would be inaccurate. Hence, successful integra- tion of a sensor requires electrode alignment with the hydrogel channel.

3.1.3 Proposed adaptations

Since avoiding air entrapment in the cavity appeared a prerequisite, the geometry

of the cavity was changed. The optimized geometry resembles a tunnel with inlet

for the hydrogel and a outlet for the air. This inlet for hydrogel injection is designed

to tightly fit a pipette tip to prevent backflow. To accommodate for the flexibility

of the gel support, PDMS was replaced by PMMA. To mechanically stabilize the in-

jection set up and prevent additional air from entering the cavity, the gel support

was completely encapsulated by the mould holder and firmly secured by screws to

prevent both horizontal and vertical movement. A similar solution was proposed

to mechanically stabilize the hydrogel on the glass slide, by designing an additional

glass holder which simultaneously would ensure alignment of the hydrogel channel

with the electrodes on the glass slide. Additional inlets were incorporated in the gel

support for addressing the hydrogel channels, which needed to be perfectly sealed to prevent backflow.

3.2 Electrode configuration

Initial concept theory

From experiments performed by Crone et al. it was learned that an optimal distance between the current injecting and most distant potential measuring electrode ap- proximates two length constants for the specific capillary, resulting in an intercapil- lary potential of 10% of the initial potential at the current injecting electrode. [24] The length constant as expected for a semi-circular hydrogel channel (r

= 30 μm) containing cell culture medium (ρ

medium

= 58.8 Ωcm) and lined with hCMEC/D3 cells (TEER = 22 Ωcm

) sealed on a glass surface can be calculated based on Equa- tion 2.2, thereby assuming no ion leakage through the glass substrate. [9] By rewrit- ing the latter equation and filling in the stated values for each variable, the length constant can be approximated at:

λ = s

TEERπr

ρ

2πr

= 237µm (3.1)

Similar to experiments conducted by Crone et al. the goal is to perform potential measurements at three distances from the current injecting electrode to be able to observe an exponential potential decay. Since this would imply that only a small channel distance could be monitored with one electrode set, integration of multiple electrode sets to allow for determining TEER values across different segments in the hydrogel channel was desirable.

First electrode design

All previously stated considerations are integrated into the final electrode design as depicted in Figure 3.2. In total, three electrode sets were located across the hydrogel channel. Each of those electrode sets was comprised of four platinum (Pt) electrodes for current injection and potential measurements, and two Ag/AgCl electrodes as reference electrodes, one per electrical circuit. The center of the small contact areas at the end of each electrode are aligned below the centerline of the hydrogel chan- nel. From this moment on, those ends of the electrodes aligned with the channels are further referred to as the "electrodes", whereas the bigger contact areas on the sides are referred to as "contacts".

Expected limitations

Ag/AgCl is known to be suited as reference electrode since it has a high exchange

current density, is reversible and non-polarizable. However, failure of the relatively

thin electrodes might occur due to consumption of Ag or AgCl, which would require

F

3.2: Electrode set-up on glass slide. Six Ag/AgCl (green) and twelve Pt electrodes (light blue) deposited on glass and covered by an additional insulating layer of SiO

(dark blue), thereby ensuring ex- posure of electrodes at the desired locations solely inside the channel of the gel and at the contact areas. Electrode numbers are indicated in the contact areas. Electrode dimensions: width = 10μm, length con- tact area in channel = 50 μm, contact area = 1.5 x 2.5 mm, total length electrode = 13.5 - 8.5 mm, height Pt = 135 nm, height Ag/AgCl = 500 nm, spacing Pt electrodes = 150 μm, horizontal spacing Ag/AgCl-Pt electrodes = 600 μm, spacing Ag/AgCl electrodes perpendicular to

channel = 200 μm.

replacement. [25]

3.3 Optimized novel design

The proposed adaptations were processed and integrated in the optimized design, which consists of an hydrogel injection setup and a measurement setup as depicted in Figure 3.3d, e respectively. The injection setup is comprised 1) the Delrin mould holder with a lowered plateau in which an interchangeable silicone mould with channel features is placed (Figure3.3a), 2) a PMMA gel support structure with on the bottom two rectangular open tunnels. The tunnel has two sets of shafts, the outer one for addressing the tunnel for injecting hydrogel and the inner ones for address- ing the hydrogel channels. Additional shafts are integrated at the sides and aligned with the contact areas of the electrodes to allow connection of the electrode through pins, and 3) a combination of fixation plates and screws to mechanically stabilize the gel support in the mould holder (Figure 3.3b). The measurement setup is comprised of 1) the same gel support, fixation plates and screws (Figure 3.3b). Added to these is 2) a new Delrin glass holder with a window for imaging purposes and 3) the glass substrate with integrated sensor and connector pins (Figure 3.3c). A crossectional view of the cavity for hydrogel in the total injection set up is shown in Figure 3.4.

Due to its symmetrical features, the gel support fits in the mould and glass holder such that mirroring has no influence. The injection setup is assembled by placing the silicon mould in the lowered plateau in the Delrin mould holder after which the PMMA gel support is placed on top of it, all secured by two plates and four screws. Hydrogel is injected into both tunnels through one of the outer inlets by using a pipette tip. After gelation, the PMMA gel support is lifted from the mould holder and transferred to the glass slide with integrated electrodes in the Delrin glass holder, thereby aligning the electrodes along the centerline of the hydrogel channels.

The connector pins are inserted in the shafts of the gel support to enable connection with the contacts. Hydrogel channels can be addressed by glass capillaries installed in the shafts aligned with the channel features on the silicon mould, thereby facili- tating static and ultimately dynamic cell culture when connecting the capillaries to a pump.

3.3.1 Strengths and limitations

The new cavity geometry in combination with a rigid PMMA gel support resulted

in a more reliable standardized hydrogel injection method with equally distributed

pressure across the cavity and only required 30 μl instead of 120 μl per hydrogel,

which is economically favourable. The combination of the injection and measure-

ment set up allows for electrode alignment in the hydrogel channel. Moreover, dy-

namic cell culture can be facilitated by connecting the capillaries to a pump.

F

3.3: Individual components of the gel injection and mea- surement set up: A) a silicon mould with microchannel features and Delrin mould holder with lowered plateau perfectly fitting the sil- icon mould, B) a transparent PMMA gel support with two tunnels as cavities for hydrogel (1mm x 1mm x 20 mm in width, height and length respectively), C) a Delrin glass holder with open window for imaging and a glass slide with incorporated electrodes. D) The com- plete gel injection set-up is assembled by placing the silicon mould in the predefined area in the Delrin mould holder after which the PMMA gel support is placed on top of it, all secured by two plates and four screws. Hydrogel is injected into both tunnels through its inlets present at the extremities of the tunnel. E) After gelation, the gel support is removed from the total setup and transferred onto the glass slide with integrated electrodes in the Delrin glass holder. Elec- trodes can be addressed through the connector pins perfectly aligned

above the electrode’s contact areas.

F

3.4: Cross-sectional view of gel injection set up, additionally zoomed in on the tunnel placed on the silicon mould containing the

channel features.

Nevertheless, this design also has some shortcomings. First, the small width of the

tunnel in the new cavity geometry may be a limiting factor in the creation of broader,

more complex channels geometries such as branching channels mimicking the inter-

weaving capillary beds. Since the tunnel width is dependent on the width of the

injection tool, in this case a pipette tip, this limitation can be overcome by designing

a customized injection tool for hydrogel injection. Second, despite mechanical stabi-

lizing the hydrogel, channel sealing remains an uncertainty and might be improved

by chemical functionalization of the glass slide, see Chapter 4.5.

Chapter 4

Hydrogel chip fabrication by soft-lithography

4.1 Introduction

For optimal hydrogel chip fabrication, two functional requirements have to be met:

1) the channel fabrication method should result in reproducible channels with con- trollable channel dimensions and 2) the created hydrogel channels should be sealed on the glass surface, thereby preventing leakage of the fluid it contains. A factor influencing the first requirement is the elastic modulus of the hydrogel which is amongst others dependent on hydrogel polymer concentration. Various successful results have been reported for 3% agarose hydrogel with elastic moduli reported varying from 19-32 kPa. [19], [26] However, since the young’s modulus for brain tissue is reported to be approximately 800 Pa [27], lower hydrogel stiffnesses are desirable. To investigate the suitability of using a HA-based hydrogel for creating 3D vessel structures by the proposed soft-lithography technique the performance of HA-ta/Dex-ta hydrogel with polymer concentrations 2.5%, 5% and 10% was tested.

Subsequently, a comparison is made with 1%, 2% and 3% agarose hydrogels. Cor- responding elastic moduli as known from literature are presented in Table 4.1. Per- formance of the different hydrogels was quantified by measuring the channel width as created in the hydrogels compared to the feature dimensions of the used silicon mould. The most suitable hyaluronic-based hydrogel would be the lowest polymer concentration at which reproducible transfer of channel features into the hydrogel still is achievable.

Subsequently, the second requirement was aimed to be achieved by chemical

functionalization of the glass slide with tyramine groups to create a chemical bond

between the hydrogel and glass surface. Sealing of the channels on glass was as-

sessed by flowing microbeads of various sizes through the channels utilizing the

capillaries integrated in the gel holder. The different microbead sizes would indi-

cate whether the obtained sealing was sufficient for cells, estimated to be ± 2μm in

diameter, to be seeded inside the channel without ending up in between the gel and

glass.

T

4.1: An overview of the reported elastic moduli for the investi- gated hydrogels. Data of 2.5% and 5% HA-ta/Dex-ta is yet unknown.

Hydrogel Elastic modulus (kPa) Agarose 1% 3.6 [28]

Agarose 2% 10.6 [28]

Agarose 3% 19-32 [19]

HA-ta/Dex-ta 2.5% - HA-ta/Dex-ta 5% - HA-ta/Dex-ta 10% 15 [18]

4.2 Experimental methods

4.2.1 Silicon mould fabrication

Fabrication of the silicon mould with positive channel topographies was done by soft-lithography, as illustrated in Figure 4.1. In short, silicon wafers were primed with hexamethyldisilazane (HMDS) to create a hydrophobic surface followed by spin-coating a layer of positive photoresist AZ

40XT at various layer thicknesses depending on the desired height of the channels (details available in Table A.1).

Then, the wafer was soft baked at 126

C for 7 minutes. The channel features were transferred into the AZ

40XT layer by exposure to UV light for 40 seconds in a mask aligner followed by a post exposure bake of 100 seconds at 105

C. Subsequently, the wafer was developed for approximately 3 minutes in AZ

726 metal ion free (MIF) developer solution. To create the semi-circular crosssection of the channel features, the AZ

40XT was reflowed at 140

C for 60 seconds. The latter step was left out of the process flow for square or rectangular shaped channels.

4.2.2 Hydrogel preparation

Both hydrogel types were prepared in vials at room temperature (RT). 1%, 2% and 3% agarose was prepared by dissolving agarose powder (low EEO, Sigma-Aldrich, St. Louis, U.S.A.) in phosphate buffered saline (PBS, Gibco) upon heating in the microwave, as described by the supplier. Hydrogel composed of HA-ta and Dex- ta was prepared as previously described by Wennink et al. [18] with an additional gelatin-tyramine (Gel-ta) component to promote endothelial cell proliferation. In short, standard procedure for preparing the hyaluronic based hydrogels required mixing of three solutions containing the HA-ta/Dex-ta/Gel-ta polymers, the HRP and H

O

at a ratio of respectively 8:1:1 volume parts. For a 2.5%, 5% and 10%

hydrogel, HA-ta, Dex-ta and Gel-ta were combined 1:1:0.05, 1:1:0.1 and 1:1:0.2 to

obtain polymer solutions of 3.125%, 6.25% and 12.5% w/v in PBS, respectively. Sub-

sequently, HRP (stock solution 311 U/ml in PBS) was added to a final concentration

of 0.5 U/ml, after which the mixture was gently shaken and stored at 4

C overnight

to allow all components to distribute evenly. Prior to use, the hydrogel mixture was

F

4.1: Process flow for fabrication of the silicon moulds with AZ40

XT channel topographies. A silicon wafer (1) was covered with a spincoated layer of positive photoresist (AZ40

XT) (2), which was exposed to UV light through a photomask (3), developed (4) and

reflowed (5).

finalized by adding freshly made H

O

(30wt%, Sigma) at a final concentration of

0.015% w/v to initiate crosslinking. HA-ta, Dex-ta and Gel-ta powders, were a gen-

erous donation from the Developmental BioEngineering research group, University

of Twente, Enschede, the Netherlands.

F

4.2: An overview of the chemical functionalization of a glass substrate with tyramine. Steps include: 1) oxygen plasma treatment;

2) incubation with APTES; 3) which is hydrolyzed upon dissolving in water so that ethanol groups leave, allowing its -OH groups to re- act with hydroxyl groups on glass surface; 4) incubation with NHS- tyramine at pH ≈ 8.5; 5) resulting in an amide bond between primary

amine of APTES and tyramine.

4.2.3 Surface functionalization

The surface modification procedure is illustrated in Figure 4.2. The first step in- volved activation of the glass surface with hydroxyl groups by oxygen plasma treat- ment of the glass substrates (24x40 mm, Waldemar Knittel Glasbearbeitungs GmbH) with the Femto Science Cute plasma cleaner. Subsequently, glass slides were sub- merged in 3% v/v (3-aminopropyl)triethoxysilane (APTES, ≥ 98%, Sigma-Aldrich) solution in ultra pure water (PURELAB

flex 3, Elga) for 30 minutes at RT to coat the glass with primary amines after which the substrates were washed thoroughly by rinsing with water, thereby ensuring removal of unbound silanes. Tyramine groups were then chemically bonded onto the primary amines via NHS-ester chem- istry with a NHS-tyramine. For this, NHS-tyramine (3-(4-hydroxypheyl) propionic acid N-hydroxysuccinimide ester (Sigma) was first dissolved in d imethyl sulfox- ide (DMSO) after which the solution was diluted in a 100 mM bicarbonate reaction buffer (pH=8.5) to a final concentrations of 0.5, 1.0, 2.0 and 5.0 mg/ml. Next, the glass slides were submerged in this solution for 15 minutes at RT, rinsed with ultra pure water and used on the same day.

4.2.4 Hydrogel chip fabrication

The PMMA gel support was finalized by fixation of the inlet tubes and capillar-

ies (355/101 μm outer/inner diameter, molex Polymicro Technologies

) with UV

curable glue (nr. 81, Norland Products INC.) to ensure a proper sealing. Prior to as-

sembly of the total injection set-up as illustrated in Figure 3.3, the silicon moulds

with channel features were pretreated with oxygen plasma (Femto Science Cute

plasma cleaner) followed by vapour phase deposition of perfluorodecyltrichlorosi- lane (FDTS, abcr GmbH, Germany) to prevent hydrogel from sticking to the sur- face. Hydrogel solutions were injected into the tunnels via the outer inlets by using pipette tips and allowed to solidify for 10 minutes at RT. After disassembling the gel injection set-up, the silicon mould that had remained attached to the PMMA gel support was carefully lifted from the hydrogel. To prevent trapped air bubbles in- side the hydrogel channels, the capillaries were flushed prior to assembly of the gel holder whilst keeping the hydrogel hydrated by covering the gel in PBS to prevent dehydration. The gel support was transferred onto a glass slide with or without incorporated electrodes and subsequently placed in the Delrin glass holder to com- plete the total gel holder set-up. Afterwards, the pipette tips containing crosslinked hydrogel were replaced with pipette tips filled with PBS to ensure the hydrogels to remain hydrated. The system was allowed to equilibrate for 30 minutes prior to further proceedings.

4.2.5 Analysis channel feature transfer

Silicon moulds with three types of channel features were available: semi-circular channels with diamaters of 40 μm and 60 μm as well as square channels with a width of 50 μm. The channel height of 27 μm was constant for all chips. Performance of the different hydrogels was quantified by measuring the channel width as created in the hydrogels and subsequently comparing those values to the feature dimensions of the used silicon chip. Hydrogels were imaged with the EVOS FL Cell Imaging System (ThermoFischer Scientific) and analyzed with Leica Application Suite software (LAS version 4.12.0, Leica Microsystems). Measurements were performed in triplicates from which an average channel width with corresponding standard deviation was calculated per hydrogel type.

4.2.6 Analysis sealing hydrogel channels

Hyaluronic-based hydrogels were specifically transferred onto a tyramine function-

alized glass slide covered with an additional droplet of H

O

to enable crosslinking

between the hydrogel and the glass slide, whereas agarose hydrogels were placed

on non-treated glass slides. Sealing of the channels was functionally tested by ad-

dressing the hydrogel channels through the capillaries integrated in the measure-

ment set-up with 2 μm and 10 μm microbeads (Polybead

polystyrene black dyed

microspheres, Polysciences, Inc.).

4.3 Results and Discussion

4.3.1 Hydrogel channel fabrication

The developed injection set-up offered a method for transferring channel features of the silicon mould into 1%, 2% and 3% agarose and 5% and 10% hyaluronic-based hydrogels without being compromised by the presence of air bubbles, as depicted in Figure 4.3a,c. The 2.5% HA-ta/Dex-ta/Gel-ta appeared to lack stiffness required to obtain the channel structure. Subsequently, the created hydrogel channels could be aligned with the electrodes incorporated in the glass slide by transferring the PMMA gel support into the glass holder (Figure 4.3b,d).

Occasionally, some abnormalities in hydrogel geometries were observed as de- picted in Figure 4.4. The first type of deformities were related to the AZ

40XT channel features. The silicon moulds were re-usable up to at least twenty times.

After that, some deformities were observed in hydrogel channels which could be traced back to impairments of the AZ

40XT channel features on the silicon mould (Figure 4.4a,b). The second type of deformations however, could not be explained that easily. They were hypothesized to arise from structural changes of dissolved hydrogel components when stored at 4

C over time. These might affect the silane bonds of the FDTS coating in such a manner that gelation is impaired, hence the waving channel structures. For this reason, it is advised to store dissolved compo- nents at 4

C up to seven days at most or, preferably, prepare fresh solutions prior to each experiment.

4.3.2 Channel feature transfer

Although the proposed method proved successful in fabrication of hydrogel chan- nels in general, reproducibility is questionable. The reproducibility was quantified by comparing the width of the AZ

40XT channel features with the width of the successfully fabricated hydrogel channels as shown in Figure 4.5. First, it can be remarked that an exact transfer of the channel features into the hydrogel was not achieved. As an exception to the aforementioned observation, the 50 μm feature with squared crosssection appeared to be transferred 1:1 in all agarose hydrogels.

Second, the coefficients of determination (R

) are relatively high, implying that the

trend lines are a good model for predicting the hydrogel channel width dependent

on the used mould dimensions. Again, an exception in the overall trend is observed,

in this case for the 3% agarose hydrogel. The latter is striking since this specific

hydrogel has been used elaborately in literature [19], [26] and was therefore hypoth-

esized to ensure best feature transfer. Third, standard deviations of the measured

channel widths corresponding to the semi-circular cross sections (40 μm and 60 μm)

are seemingly bigger compared to the squared cross section (50 μm). Moreover, the

average measured channel width for the 40 μm channels are exceeding the expected

values for most hydrogels, whereas the opposite holds true for the 60 μm channels.

F

4.3: An image of the actual injection (A) and measurement set-up (B), with zoomed view of a silicon mould with channel fea- tures (30 μm radius) and glass slide with incorporated electrodes.

Connector pins were soldered to wires that enabled connecting the electrodes to a measurement apparatus. Using both set-ups allowed for hydrogel channel fabrication (C, 50 μm radius) and subsequently alignment of the electrodes underneath the channel upon transferring the PMMA gel support in the glass holder (D, zoomed view presented

in E). Scale bars represent: C = 1000 μm; D and E = 400 μm.

F

4.4: Abnormalities in hydrogel channel structures: A) defor- mities in specific segments of hydrogel channels could be traced back to B) impaired channel features on the silicon mould; C) deformed channel geometries were observed when using hyaluronic-based hy- drogel components that had been dissolved and subsequently stored

for a period longer than seven days. Scale bars represent 100 μm.

The used method for quantifying channel feature transfer has some limitations, which might explain unexpected results. First, limitations can be related to the use of the, at that time, readily available silicon moulds and their channel features. These investigated channel features differed regarding their cross sectional area, which were square (50 μm) and semi-circular shaped (40 and 60 μm). Both shapes might be- have differently when subjected to vertical pressure such as applied in the measure- ment set up. Upon applying vertical pressure, a square shaped channel would start collapsing while maintaining its width due to a vertical load distribution, whereas a semi-circular shaped channel would have a horizontal load distribution result- ing in variations of the measured channel width. The latter solely holds true when free movement of the hydrogel on the glass surface is allowed (even when limited), which is the case for hydrogels that are not chemically anchored to the glass slide as observed for the majority of the hydrogels. This will be further elaborated on in Section 4.3.3. To assess this effect, cross sectional imaging of the channel is desired.

Related to this topic, due to a lack of channel sealing on the glass, channel bound-

aries might be positioned in other focus planes which would affect the measured

channel width. As final shortcoming, the range in channel width was limited. For

F

4.5: A,B) Measurements of the AZ

40XT channel features and hydrogel channels, respectively, were performed using the LAS software. C, D) The average hydrogel channel width (channel di- mension) is plotted against the width of the channel feature (mould dimension), together with a linear line to indicate the expected be- haviour (y=x) and for each investigated hydrogel a corresponding

trend line (n=3).

a better overview it is desirable to have a broader spectrum of channel widths, for instance 20-100 μm which corresponds to precapillary arteriole dimensions that are part of the BBB microvasculature together with the capillaries ( ≤ 10 μm). [29]

To eliminate these limitation, one could repeat these measurements with a fresh,

carefully selected set of silicon moulds. All with the same shape and height features

and a broader range of diameters. Optional is repeating the measurements twice,

using circular and square shaped features. Interestingly, although brain endothelial

cells are typically exposed to high curvatures in the capillaries, endothelial cells’ ca-

pacity to remodel square cross sectional hydrogel channels towards elliptical shapes

might question the need for semi-circular hydrogel channels. [16]

F

4.6: Microbeads with diameters of 2 μm (C) and 10 μm (A,B) visualized hydrogel channel sealing on glass slides. Some segments were sealed sufficiently (A,C), whereas others did not (B). Channels are indicated in each subfigure with an

|{z}

. Scale bars represent: A and B = 200 μm, C = 400 μm.

4.3.3 Sealing hydrogel channels

Sealing of the hydrogel channels on the glass slide was visualized by flushing the channels with two solutions containing 2 μm and 10 μm microbeads, respectively.

In general, a proper sealing of the channels could not be achieved, independent of the used hydrogel type or the concentration of the tyramine coating (Figure 4.6b).

Occasionally, applying pressure through the measurement set-up would suffice and microbeads could be flushed through (segments of) the hydrogel channels as shown in Figure 4.6a,c. The latter is hypothesized to affect the sealing in two ways. First, upon applying pressure, the hydrogel structure is mechanically stabilized. Second, the hydrogel will absorb liquid from the PBS reservoir until it reaches a maximum swollen state as determined by the boundaries set by the gel support and the glass slide, thereby sealing the channels. Figure 4.6c illustrates this observation, where the 2 μm beads can be found inside the channel and underneath the PMMA gel sup- port due to a rupture elsewhere in the hydrogel (segment not shown), whereas they are not present underneath the hydrogel that is tightly pressed onto the glass slide.

This implies that proper control of the hydrogel swelling behaviour offers possi- bilities in channel sealing on the glass slide. Nevertheless, although this might be sufficient for cell culture purposes, performing TEER measurements requires fully sealed channels without any liquid leakage.

To explore what prevented the hydrogel from chemically binding to the glass

slide, two factors were questioned and examined: 1) the chemically functionalized

glass slide and 2) the presence of free reactive tyramine groups on the hydrogel

surface after crosslinking required for subsequent crosslinking with the functional-

ized glass. (Appendix A.2) First, successful chemical functionalization of the glass

slide with tyramine using the aforementioned protocol was confirmed by allowing

hyaluronic-based hydrogel to crosslink directly on the treated glass surface, which

resulted in proper attachment of the hydrogel onto the glass slides. The second

factor was tested by indirect crosslinking of hyaluronic-based hydrogels comprised

of decreasing H

O

concentrations on similarly treated glass slides. However, hy- drogels crosslinked with the lowest H

O

concentration that would still allow for transfer of channel features, did not chemically bind to the tyramine treated glass slide upon indirect crosslinking.

Since indirect crosslinking of hydrogel on a substrate other than the same hy- drogel has not been reported to our knowledge, comparison of the obtained results could not be done. However, inspiration for sealing open hydrogel channels on a glass surface may be found in other methodologies used for creating hydrogel mi- crovasculature, based on sacrificial gel structures or applying a molecular coating of the hydrogel on the glass slide. These options will be further elaborated on in Chapter 7.1.

4.4 Conclusion

The developed injection set-up offered a method for transferring channel features

of the silicon mould into agarose and hyaluronic-based hydrogels without being

compromised by the rise of air bubbles. A most suitable hyaluronic-based hydro-

gel for imprinting purposes could not be identified since channel feature transfer

from the silicon mould to the either tested hydrogels did not occur 1:1. However,

obtained trend lines between the channel and mould dimensions as experimentally

determined appeared a good model for predicting the hydrogel channel width de-

pendent on the mould dimension and used hydrogel type. Sealing of the hydrogel

channels was not achieved. Hyaluronic-based hydrogels with sufficient stiffness for

imprinting channel features through the proposed soft-lithography method could

not be further enzymatically crosslinked on tyramine coated glass slides.