Three-dimensional immobilization of proteins within agarose

hydrogels using two-photon chemistry

by

Ryan Gavin Wylie

A thesis submitted in conformity with the requirements for the degree of Doctorate of Philosophy

Department of Chemistry University of Toronto

© Copyright by Ryan Gavin Wylie 2011

ii

Three-dimensional immobilization of proteins within agarose

hydrogels using two-photon chemistry

Ryan Wylie

Doctor of Philosophy

Department of Chemistry University of Toronto

2011

Abstract

Three-dimensional biomolecule patterned hydrogels provide cellular microenvironments that

mimic in vivo conditions. We are particularly interested in the fabrication of materials to spatially

control stem cell differentiation towards the creation of tissue analogues. To this end, we have

designed a 3D protein patterning system where differentiation factors were immobilized within

distinct volumes through two-photon chemistry, which provides 3D control since the excitation

volume is limited to the focal point of the laser. Agarose hydrogels were modified with 6-bromo-

7-hydroxy-coumarin (Bhc) protected amines or thiols, which upon two-photon excitation are

deprotected in defined volumes yielding reactive amines or thiols. Fibroblast growth factor-2

(FGF-2) was immobilized onto agarose-thiol-Bhc through either disulfide bond formation with

agarose thiols or the physical interaction between human serum albumin (HSA) and the albumin

binding domain (ABD). The use of biological binding pairs also provides mild immobilization

conditions, minimizing the risk for bioactivity loss. Similarly, two differentiation factors for

retinal stem progenitor cells were simultaneously immobilized: 1) ciliary neurotrophic factor

(CNTF); and 2) N-terminal sonic hedgehog (SHH). Maleimide modified binding proteins, such

as maleimide-streptavidin; react with exposed thiols, yielding 3D patterns of covalently

immobilized streptavidin in agarose hydrogels. Growth factors are then introduced as fusion

iii

proteins with binding domains, such as biotin-CNTF, for complexation and thus 3D

immobilization. By combining multiple binding systems with two-photon patterning, we were

able to simultaneously 3D immobilize proteins towards the creation biomimetic hydrogels.

iv

Acknowledgments

I would like to thank Prof. Molly Shoichet for all her guidance and help during the course of my

graduate studies. I am especially grateful for the freedom provided to explore new avenues of

research. My time in the Shoichet lab has not only further my knowledge of science but taught

me how to think critically. I would also like to thank Prof. Cindi Morshead for all the invaluable

insight and direction towards the completion of my PhD. I am also grateful to my committee

member Prof. Gilbert Walker for encouraging me to think beyond the direct scope of my project

and providing insight for future directions. I especially like to thank Dr. Karen Maxwell for her

priceless help in protein engineering and expression. I also greatly appreciate the invaluable

advice I received from Diane Bona for molecular biology, protein expression and purification.

I would also like to thank past and present Shoichet and Morshead lab members. I am

particularly grateful to Dr. Jordan Wosnick, for his help in establishing the two-photon chemistry

necessary for my project. I greatly appreciate all the biological help I have received from Dr.

Mike Cooke, Shoeb Ahsan and Yukie Aizawa. I would also like to thank all the members of the

Shoichet group for all their politically correct encouragement during the course of my studies.

v

Table of Contents

 

Acknowledgments .......................................................................................................................... iv 

Table of Contents ............................................................................................................................ v 

List of Figures ................................................................................................................................ xi 

List of Tables ............................................................................................................................. xviii 

1  Introduction ................................................................................................................................ 1 

1.1  Rationale ............................................................................................................................. 1 

1.2  Hypothesis and objectives ................................................................................................... 2 

1.3  In vitro cell culture .............................................................................................................. 3 

1.3.1  Limitations of 2D cell culture ................................................................................. 4 

1.3.2  Extracellular environment ....................................................................................... 5 

1.4  3D cell culture ..................................................................................................................... 7 

1.4.1  Scaffolds for 3D cell culture ................................................................................... 7 

1.4.2  Elucidation of mechanical cues through 3D cultures ............................................. 9 

1.4.3  Elucidation of chemical cues through 3D cultures ............................................... 10 

1.5  Peptide and Protein Immobilization .................................................................................. 11 

1.5.1  Immobilization through covalent bonds ............................................................... 11 

1.5.2  Immobilization using physical interactions .......................................................... 12 

1.5.3  Enzyme catalyzed immobilization ........................................................................ 14 

1.5.4  Spatial control of biomolecule immobilization ..................................................... 14 

1.6  Two-photon excitation ...................................................................................................... 15 

1.6.1  Cross-section for two-photon absorption .............................................................. 17 

1.6.2  Photoiniators/sensitizers for biological applications ............................................ 18 

vi

1.6.3  Two-photon uncaging and photolabile protecting groups .................................... 19 

1.7  Three-dimensional photopatterning of biomolecules within hydrogels ........................ 22 

1.7.1  Scaffolds for 3D patterning ................................................................................... 22 

1.7.2  Biomolecule patterning in hydrogels .................................................................... 23 

1.7.3  Photopolymerization ............................................................................................. 24 

1.7.4  Photo-grafting ....................................................................................................... 26 

1.7.5  Photo-uncaging ..................................................................................................... 28 

1.8  Stem and progenitor cells and patterned hydrogels .......................................................... 29 

1.8.1  Adult neural stem cells .......................................................................................... 29 

1.8.2  Adult retinal precursor cells .................................................................................. 30 

1.8.3  Cell penetration into hydrogels – proteolytic and non-proteolytic migration ....... 31 

1.8.4  Retina as a tissue engineering target ..................................................................... 34 

1.9  Summary of research ........................................................................................................ 35 

2  Two-photon micropatterning of amines within an agarose hydrogel* .................................... 37 

2.1  Abstract ............................................................................................................................. 37 

2.2  Introduction ....................................................................................................................... 37 

2.3  Materials and Methods ...................................................................................................... 39 

2.3.1  Materials ............................................................................................................... 39 

2.3.2  Synthesis of Boc-protected amino coumarin (3). ................................................. 40 

2.3.3  Synthesis of aminocoumarin (4). .......................................................................... 40 

2.3.4  Synthesis of aminocoumarin agarose (5). ............................................................. 41 

2.3.5  Photo-uncaging of aminocoumarin agarose with UV light. ................................. 41 

2.3.6  Two-photon Irradiation of Aminocoumarin agarose hydrogels. .......................... 41 

2.3.7  Preparation of 1 wt% aminocoumarin agarose hydrogels for amine visualization with CBQCA. .................................................................................. 42 

2.4  Results and Discussion ..................................................................................................... 42 

vii

2.4.1  Synthesis of aminocoumarin 3 .............................................................................. 43 

2.4.2  Modification of agarose with amine-protected coumarin ..................................... 44 

2.4.3  Degree of substitution of aminocoumarin agarose ............................................... 44 

2.4.4  Photo-uncaging of aminocoumarin agarose .......................................................... 45 

2.4.5  Two-photon Irradiation of aminocoumarin agarose hydrogels ............................ 48 

2.5  Conclusion ........................................................................................................................ 52 

3  Three-dimensional spatial patterning of proteins in hydrogels* .............................................. 53 

3.1  Abstract ............................................................................................................................. 53 

3.2  Introduction ....................................................................................................................... 54 

3.3  Materials and Methods ...................................................................................................... 57 

3.3.1  Materials ............................................................................................................... 57 

3.3.2  Preparation of agarose-thiol-Bhc gels ................................................................... 57 

3.3.3  Expression and purification of FGF2-ABD .......................................................... 58 

3.3.4  Labeling of FGF2-ABD with Alexa 546 .............................................................. 59 

3.3.5  Addition of maleimide to HSA ............................................................................. 59 

3.3.6  Bioactivity of recombinant FGF2-ABD ............................................................... 60 

3.3.7  Photo-patterning and Imaging ............................................................................... 60 

3.3.8  Patterning FGF2-ABD-SH to Agarose-SH through disulfide bonds .................... 60 

3.3.9  Patterning FGF2-ABD to Agarose-HSA .............................................................. 61 

3.3.10  Testing the stability of FGF2-ABD pattern with HSA ......................................... 61 

3.3.11  Quantification of FGF2-ABD ............................................................................... 61 

3.3.12  Statistical analysis ................................................................................................. 62 

3.4  Results ............................................................................................................................... 63 

3.4.1  Synthesis and characterization of FGF2-ABD and mal-HSA .............................. 63 

3.4.2  Immobilization of FGF2 using disulfide bonds .................................................... 65 

3.4.3  Immobilization of FGF2 using HSA/ABD ........................................................... 68 

viii

3.4.4  Stability of HSA/FGF2-ABD complex ................................................................. 68 

3.5  Discussion ......................................................................................................................... 70 

3.6  Conclusion ........................................................................................................................ 73 

4  Three-dimensional, spatially-controlled simultaneous patterning of multiple growth factors in hydrogels* ................................................................................................................ 75 

4.1  Abstract ............................................................................................................................. 75 

4.2  Introduction ....................................................................................................................... 75 

4.3  Materials and Methods ...................................................................................................... 78 

4.3.1  Materials ............................................................................................................... 78 

4.3.2  Photo-patterning and Imaging. .............................................................................. 79 

4.3.3  Patterning SHH-barstar. ........................................................................................ 80 

4.3.4  Patterning biotin-CNTF. ....................................................................................... 80 

4.3.5  Dual Patterning. .................................................................................................... 81 

4.3.6  Migration of NPCs into SHH/RGD channel. ........................................................ 81 

4.3.7  Conversion of fluorescence intensity to concentration for bartar-SHH-488 and biotin-CNTF-633 .................................................................................................. 83 

4.3.8  Stability study for immobilized SHH using barnase-barstar ................................ 83 

4.3.9  Preparation of coumarin sulfide agarose ............................................................... 83 

4.3.10  Plasmid design ...................................................................................................... 84 

4.3.11  Expression and purification of barnase ................................................................. 84 

4.3.12  Synthesis of maleimide-barnase with sulfo-SMCC .............................................. 85 

4.3.13  Expression, purification and labeling of barstar-SHH .......................................... 85 

4.3.14  Expression, purification and labeling of biotin-CNTF ......................................... 86 

4.3.15  Preparation of gels for bioactivity assay ............................................................... 87 

4.3.16  Obtaining retinal precursor cells ........................................................................... 88 

4.3.17  Plating of cells for bioactivity studies ................................................................... 89 

4.3.18  Cell survival analysis with PicoGreen .................................................................. 89 

ix

4.3.19  Cell survival analysis with Live/Dead staining .................................................... 89 

4.3.20  Gene Expression Assays ....................................................................................... 89 

4.3.21  Immunocytochemistry .......................................................................................... 90 

4.4  Results and Discussion ..................................................................................................... 90 

4.4.1  3D immobilization of SHH using barnase-barstar ................................................ 90 

4.4.2  3D immobilization of CNTF using streptavidin-biotin ........................................ 95 

4.4.3  Simultaneous immobilization of SHH and CNTF ................................................ 96 

4.4.4  Immobilized SHH and CNTF are bioactive ........................................................ 101 

4.4.5  NSPCs migrate into patterns of SHH .................................................................. 104 

4.5  Conclusion ...................................................................................................................... 104 

5  Discussion .............................................................................................................................. 106 

5.1  3D photochemical patterning in Agarose hydrogels ....................................................... 106 

5.1.1  Agarose as a scaffold for 3D biochemical patterning ......................................... 106 

5.1.2  Bhc photocage ..................................................................................................... 107 

5.1.3  Two-photon patterning in aminocoumarin agarose ............................................ 108 

5.2  3D protein immobilization .............................................................................................. 109 

5.2.1  Immobilization of proteins through disulfide bonds ........................................... 111 

5.2.2  Immobilization of maleimide molecules ............................................................ 111 

5.2.3  Immobilization through physical interactions .................................................... 112 

5.2.4  Simultaneous Immobilization of Proteins ........................................................... 114 

5.3  Bioactivity of immobilized factors ................................................................................. 115 

5.4  Migration into agarose hydrogels ................................................................................... 116 

5.5  3D protein patterning and regenerative medicine ........................................................... 117 

6  Conclusions ............................................................................................................................ 118 

6.1  Achievements of Objectives ........................................................................................... 118 

6.2  Major contributions ......................................................................................................... 120 

x

7  Recommendations for future work ......................................................................................... 122 

7.1  Protein patterning in different hydrogels ........................................................................ 122 

7.2  Increased migration coupled with matrix degradation .................................................... 123 

7.3  Three-dimensional differentiation of stem cells ............................................................. 124 

8  References .............................................................................................................................. 126 

Appendix A: Abbreviation .......................................................................................................... 142 

Appendix B: Supplemental Figures ............................................................................................ 144 

Appendix C: Magnetic cell seeding ............................................................................................ 147 

Copyright Acknowledgements .................................................................................................... 149 

xi

List of Figures

Figure 1-1: Schematic depicting the complexity of the extracellular matrix surrounding cells in

vivo. The ECM contains many insoluble structural components and soluble factors that influence

cell processes. Two-main functions of the ECM are shown: 1) integrin (tan) binding to adhesion

sites (green); and 2) growth factor (red circles) sequestering. Copyright Wiley-VCH Verlag

GmbH & Co. KGaA. Reproduced with permission. [12] ............................................................... 5

Figure 1-2: Comparing the excitation volume of one and two-photon irradiation of a fluorescein

solution. One-photon excitation (bottom laser) results in the excitation of fluorescein throughout

the path of the laser. The two-photon excitation volume is limited to the focal point of the laser

(top laser). Reproduced with permission from Prof. Kevin Belfield [80]. ................................... 15

Figure 1-3: Excitation of fluorescent molecules with microscope equipped with a pulsed multi-

photon laser. Laser pulses are concentrated at the focal point to achieve the intensity of light

needed for two-photon excitation. Reproduced with permission from Dr. Michael Davidson[84].

....................................................................................................................................................... 16

Figure 1-4: Methodology for the 3D patterning of biomolecules in hydrogels using two-photon

lasers. The site of immobilization is controlled by the path of the TP laser focal point. .............. 24

Figure 1-5: Photopolymerization of fluorescently labeled BSA and IKVAV in hydrogels. (a)

Fluorescently labeled BSA was photocrosslinked within an agarose hydrogel at different laser

intensities (left to right: low to high intensity). The increased fluorescent signal in the samples

exposed to high laser intensity indicates that the amount polymerized can be controlled by

varying the laser intensity. (b) Fluorescent BSA (green) and IKVAV (red) were sequentially

patterned within the same hydrogel. Copyright Wiley-VCH Verlag GmbH & Co. KGaA.

Reproduced with permission. [87] ................................................................................................ 25

Figure 1-6: Maleimide-Alexa Fluor 488 and 546 were sequentially 3D immobilized in coumarin-

sulfide modified agarose hydrogels, (a) oblique and (b) side view. The layered structure of green

squares (488) and red circles (546; ~ 50 µm in diameter) demonstrate the high spatial achievable

through TP patterning. Reproduced with permission [113]. ......................................................... 29

xii

Figure 2-1: Synthesis of aminocoumarin ...................................................................................... 43

Figure 2-2: Synthesis of aminocoumarin agarose ......................................................................... 43

Figure 2-3: Photo-induced deprotection of agarose amines ......................................................... 44

Figure 2-4: Detection of primary amines using CBQCA after the irradiation of an

aminocoumarin agarose solution. Samples irradiated with a UV lamp at long wavelength (365

nm) showed significantly greater fluorescence when compared to samples that were not

irradiated. The increase in fluorescence from the irradiated samples confirms the production of

primary amines upon irradiation. .................................................................................................. 45

Figure 2-5: A 50 by 50 µm box was patterned 40 μm below the surface of the gel. The yield of

reaction (percent of coumarin photocleavage and pmol of amines) was determined by measuring

the decrease in coumarin fluorescence within the patterned region. The change in fluorescence

intensity of coumarin was measured over the patterned region through the first 100 μm. The box

was scanned three times with the pulsed Ti-Sapphire laser set to 740 nm. (a) The yield of

coumarin deprotection by two-photon irradiation was then calculated as a function of depth by

comparing the change in coumarin fluorescence in the patterned region to a non-patterned

region. (b) The amount of amines in piocomoles as a function of depth within the patterned

region. Confocal micrographs of coumarin fluorescence are shown at : (c) 20 μm, (d) 30μm, (e)

40μm, (f) 50μm and (g) 60μm below the surface. ....................................................................... 48

Figure 2-6: Confocal image of patterned aminocoumarin agarose hydrogel visualized using the

fluorescence of coumarin. A series of boxes was patterned into the hydrogel using two-photon

excitation. The first box on the left was scanned 7 times, the second 9 times, the third 11 times

and the fourth 13 times. As the number of scans increased, the fluorescence observed decreased

due to greater photocleavage of coumarin. The fine lines located between the boxes are due to

the laser scanning on the confocal microscope. The microscope scanned the region bordered by

the fine lines but only irradiated in the region of the boxes by modulating the laser intensity;

however, the intensity of the laser outside of the boxes is still sufficient to produce the fine lines

observed. (The image was enhanced for clarity only using in photoshop.) .................................. 50

Figure 2-7: Confocal image showing the presence of amines within the patterned regions, the

boxes correspond to those in Figure 2-6. The amine reactive fluorescent probe CBQCA was used

xiii

to detect the uncaged amines. The bright boxes represent the fluorescence from the CBQCA

amine complex. The box on the left was irradiated with 7 scans and each subsequent box (to the

right) was scanned two more times. (The image was enhanced for clarity only in photoshop.) .. 50

Figure 3-1: Scheme for the 3D immobilization of FGF2-ABD to agarose through either disulfide

bonds or HSA/ABD physical interaction. (a) Schematic diagram demonstrating the 3D photo-

deprotection of thiols in agarose-thiol-coumarin for the coupling of FGF-2. (b) FGF2-ABD was

immobilized to agarose-thiol through disulfides bonds. Thiols are deprotected by two-photon

excitation of coumarin (740 nm), which subsequently form disulfide bonds with free cysteines

on FGF2-ABD. (c) FGF2-ABD was immobilized using the physical binding pair of HSA/ABD.

Maleimide-HSA was immobilized through two-photon irradiation of agarose as in (b), followed

by the addition of FGF2-ABD, which selectively binds with immobilized HSA. ....................... 62

Figure 3-2: Expression, purification and bioactivity of FGF2-ABD. (a) Protein sequence of the

expressed FGF2-ABD with the FGF-2 at the N-terminus (red) and the ABD at the C-terminus

orange) with a spacer (black) in between the two sequences to minimize interdomain

interactions. (b) SDS-PAGE protein electrophoresis of purified FGF2-ABD shows that a pure

sample (indicated with an arrow) with the proper MW (27,336 g/mol) was expressed. (c) FGF2-

ABD was determined to be bioactive by counting the number of neurospheres formed from

NSPCs after 7 d of culture. NSPCs were cultured as single cells in a 48 well plate in the

presence of varying concentrations of FGF2-ABD or commercial FGF-2. Bioactivity of FGF2-

ABD was similar to the commercial FGF-2 (mean±standard deviation shown, n=3 for each

condition, one-way ANOVA Tukey’s post-test, p < 0.05). .......................................................... 64

Figure 3-3: 3D immobilization of FGF2-ABD-546 through disulfide bonds. (a) Confocal

micrograph of a series of squares with varying concentrations of FGF2-ABD-546. 10 squares

were patterned 500 µm below the surface of the hydrogel with 5 to 50 laser scans (scale bar: 100

µm). (b) The concentration of FGF2-ABD was quantified by converting the fluorescence

intensity of each square using a calibration curve. A range of 8.5±2.9 to 58.7±12.9 nM of FGF2-

ABD was immobilized (mean±standard deviation shown, n=3 for each condition). (c) The

fluorescence z-axis profile of the squares for 30 and 50 scans was measured to determine the

axial resolution. A resolution of approximately 40 µm was achieved for each square

(mean±standard deviation shown, n=3 for each condition). ......................................................... 66

xiv

Figure 3-4: 3D immobilization of FGF2-ABD-546 using the physical binding interaction of

HSA and ABD. (a) Scheme for the immobilization of FGF2-ABD by first immobilization mal-

HSA to agarose-thiol using two-photon irradiation. (b) Confocal micrograph of immobilized

FGF2-ABD-546 in a series of 10 squares scanned 5 to 50 times. Fluorescence increased as a

function of scan number (scale bar: 100µm) . (c) The concentration of FGF2-ABD was

quantified by converting the fluorescence intensity of each square using a calibration curve. A

range of 77.9±15.1 to 189.1±20.6 nM of FGF2-ABD was immobilized (mean±standard deviation

shown, n=3 for each condition). (d) The fluorescence z-axis profile was determined for squares

scanned 30 and 50 times to determine the axial resolution. A resolution of approximately 40 µm

was achieved for each square (mean±standard deviation shown, n=3 for each condition). ......... 67

Figure 3-5: Immobilized FGF2-ABD-546 complexed with HSA is stable in PBS in both the

presence and absence of soluble HSA. The fluorescence intensity of the sample having squares

scanned 50 times was immersed in 30 mL of () PBS or () PBS with 10 mg/ml HSA was

followed over time. No significant difference was observed between the conditions (PBS versus

PBS with 10 mg/ml) at any time point, indicating the complex is stable in the presence of soluble

HSA (mean±standard deviation shown, n=3 for each condition, unpaired t test, p < 0.05). The

complex was also determined to be stable over time since no significant difference in

fluorescence was observed between any timepoints for the same condition (ANOVA with

Tukey's post hoc analysis, p < 0.05). ............................................................................................ 69

Figure 4-1: Method for the simultaneous immobilization of SHH and CNTF. Maleimide barnase

( ) is immobilized using two-photon photochemistry and a femtosecond laser. The hydrogel is

then washed in buffer to remove unbound mal-barnase. Next maleimide streptavidin ( ) is

immobilized using two photon irradiation followed by another washing step. The fusion proteins

barstar-SHH ( ) and biotin-CNTF ( ) are soaked into the gel and bind to barnase and

streptavidin, respectively. After washing out excess protein, both SHH and CNTF are

simultaneously and independently immobilized in three-dimensions. ......................................... 77

Figure 4-2: 3D immobilization of barstar-SHH-488 using barnase-barstar. (a) Maleimide-barnase

( ) was immobilized in a coumarin-sulfide agarose gel, followed by the addition of barstar-SHH

( ) modified with Alexa 488. (b) 10 different squares 100 x 100 µm having heights of 20-40 µm

were patterned 400 µm below the surface of the gel, with each square being scanned a different

xv

amount from 10 to 46 scans. Scale bar: 100 µm. (c) The amount of barstar-SHH-488

immobilized per number of scans was quantified by measuring the fluorescence from each box

and compared against a standard curve of coumarin-sulfide agarose gels with known

concentrations of Alexa 488 (mean ± s.d., n=3). (d) The z-axis profile of fluorescence of barstar-

SHH-488 for boxes with 10, 26 and 46 scans was plotted, with the maximum intensity centered

at 0 µm. ......................................................................................................................................... 91

Figure 4-3: Stability of SHH pattern using barnase-barstar immobilization. The amount of SHH

immobilized in the pattern from Fig 3 was recalculated after soaking the gels in PBS pH 7.4 for

14 days at room temperature using the same procedure as previously described. No significant

difference in immobilized SHH over time was observed, demonstrating that the pattern remains

stable over 14 days (mean ± s.d., n=3). ........................................................................................ 93

Figure 4-4: 3D immobilization of biotin-CNTF-633 using biotin-streptavidin. (a) Maleimide-

streptavidin ( ) was immobilized in a coumarin-sulfide agarose gel, followed by the addition of

biotin-CNTF ( ) modified with Alexa 633. (b) 10 different regions of boxes 100 x 100 µm

having heights of 40-80 µm were patterned 400 µm below the surface of the gel, with each

region being scanned a different amount from 1 to 19 scans. Scale bar: 100 µm. (c) The amount

of biotin-CNTF-633 immobilized per number of scan was quantified by measuring the

fluorescence from each box and compared against a standard curve of coumarin-sulfide agarose

gels with known concentrations of Alexa 633 (mean ± s.d., n=3). (d) The z-axis profile of

fluorescence of biotin-CNTF-633 for boxes with 1, 9 and 19 scans was plotted, with the

maximum intensity centered at 0 µm. ........................................................................................... 95

Figure 4-5: Representative figures for the simultaneous 3D patterning of biotin-CNTF-633 and

barstar-SHH-488. Mal-barnase was patterned in layers in the shape of a truncated (green) circle

400, 500, 600 and 700 µm below the surface the hydrogel with 40 scans per layer. Mal-

streptavidin was then patterned in a smaller (red) oval shape inserted into the truncated circle of

the mal-barnase pattern. The oval was patterned with 15 scans in four layers, following the

identical method for mal-streptavidin. Barstar-SHH-488 and biotin-CNTF-633 were immobilized

by simply immersing the hydrogel in solutions of the proteins. (a) A confocal micrograph

showing the loss of coumarin fluorescence of the layer at 400 µm from patterning of mal-barnase

and mal-streptavidin: scale bar: 100 µm. (b) A confocal micrograph of the layer at 400 µm

xvi

demonstrating the localization of barstar-SHH-488 and biotin-CNTF-633 to the volumes

patterned: scale bar: 100 µm. (c) 3D projection of the reconstructed stack using image J 3D

viewer rotated to see the layers. (d) Same projection as (c) viewed from a different angle (biotin-

CNTF-633 in red; barstar-SHH-488 in green). ............................................................................. 97

Figure 4-6: SHH and CNTF signaling pathways are activated in RPCs that are cultured on

immobilized SHH and CNTF, respectively. (a) RPCs were assayed for the presence of the ptch1

receptor in the SHH pathway using RT-PCR. (b) RPCs upregulate a key SHH signaling

mediator, gli2, in response to immobilized SHH as assayed by RT-PCR. (c) No cytotoxic effect

was found by comparing the survival of RPCs cultured on agarose-barnase-SHH (with GRGDS),

agarose-barnase (with GRGDS) and agarose-GRGDS. Cell numbers were measured after 7 d by

total dsDNA content using the PicoGreen assay (mean ± s.d., n=5 with 5,000 cells per gel). No

significant difference was observed between groups using one-way ANOVA with Tukey’s post-

hoc analysis (p > 0.05). (d) RPCs were assayed for the presence of CNTF receptor, CNTFR, by

RT-PCR. (e) RPCs respond biologically to immobilized CNTF. This was determined through

immunostaining for phosphorylated STAT-3, a protein activated through phosphorylation upon

CNTF ligand binding to CNTFR. RPCs cultured on gels with either immobilized CNTF or

soluble CNTF both stained positive for STAT-3P, whereas gels with only streptavidin and

GRGDS did not stain for STAT-3P. The percentage of immunostained cells was calculated, as

written below each series of images, and shown to be not statistically different (p>0.05, n=5

samples, mean ± s.d.). (f) The survival of RPCs cultured on agarose-streptavidin-CNTF (with

GRGDS), agarose-streptavidin (GRGDS) and agarose-GRGDS was similar. Cell numbers were

quantified after 7 d by the amount of dsDNA present using the PicoGreen assay (mean ± s.d.,

n=5 with 5,000 cells per gel). No significance difference was observed between any groups using

one-way ANOVA with Tukey’s post-hoc analysis (p > 0.05). ................................................... 100

Figure 4-7: NPCs migrate into a channel of SHH with RGD. (a) Quantification of the

concentration profile of SHH-488 as a function of depth within the hydrogel from the surface of

the gel to a depth of 100 µm. (b) Brightfield image of SHH/RGD channel show that NPCs have

migrated into the agarose gel after 14 d to a depth of 85 µm. (c) Brightfield image of RGD only

channel show that only minimal migration was observed within the hydrogel after 14 d to a depth

of 20 µm. Mostly processes were observed within the gel. (d) Confocal micrograph of

SHH/RGD channel emphasize migration of NSPCs expressing YFP into the agarose gel. All

xvii

scale bars represents 50 µm. For all cell images the white dashed line represents the surface of

the gel. ......................................................................................................................................... 103 

xviii

List of Tables

Table 1.1: Common polymer scaffolds in 3D cell culture and tissue engineering………………..8

Table 1.2: Common photoinitiators or photosensitizers for TP polymerization…………….…...18

Table 1.3: Common photocages for biomolecules……………………………………………….20

1

1 Introduction

1.1 Rationale

The in vitro study of mammalian cells has provided invaluable insight on cellular and tissue

function. Traditional in vitro cell culture has focused on the two-dimensional (2D) study of cells

cultured on flat surfaces. Although many important findings have been and will continue to be

discovered using 2D culture systems, it is becoming apparent that many cellular functions in vivo

are influenced by cues in the three-dimensional (3D) microenvironment – such as those provided

by other cells, the extracellular matrix proteins, presentation of growth factors, mechanical

properties, among others [1]. To study cellular processes in environments similar to those in

vivo, it is necessary to design artificial scaffolds that mimic the cellular microenvironment. Until

recently biomaterials did not mimic chemical and structural composition of native tissues,

limiting our ability to accurately study cells in vitro. Furthermore, a better understanding of cell-

matrix interactions from 3D culture systems will push the field towards the creation of

multicellular structures for applications in tissue engineering.

To accurately mimic the extracellular matrix in native tissues, scaffolds must be designed with

spatially tunable chemical properties. Photochemistry has been established as a useful technique

to spatially control the incorporation of biomolecule in hydrogels. For example, Luo and

Shoichet photo-patterned adhesion peptide with a He-Cd laser within hydrogels to guide neurite

outgrowth from dorsal root ganglion cells [2]. Although, traditional one-photon chemistry does

not provide the necessary 3D control since excitation occurs throughout the entire laser path. The

work presented herein uses two-photon chemistry, which provides micron resolution since the

excitation and thus reaction volume is limited to the laser’s focal point.

2

Two-photon chemistry combined with protein engineering provides the versatility necessary for

the immobilization of multiple proteins. While two-photon irradiation yields spatial control,

protein engineering provides the ability to immobilize proteins simultaneously under mild

conditions to preserve bioactivity. Proteins can be recombinantly modified with highly specific

binding domains that physically bind to photo-patterned regions of the hydrogel. Orthogonal

binding systems can be developed to work in concert for the immobilization of multiple proteins.

Furthermore, the system is applicable to numerous proteins since most protein can be expressed

with a binding domain.

We are particularly interested in the 3D immobilization of growth factors to spatially control the

differentiation of stem cells within hydrogels for the treatment of degenerative diseases. To this

end, we have designed an orthogonal 3D protein patterning system where differentiation factors

are immobilized in distinct volumes to spatially control the differentiation of stem cells. We are

particularly interested in creating biomaterials for the treatment of retinal diseases, and have

focused on the immobilization of known differentiation factors for retinal stem progenitor cells

(RSPCs). The retina was chosen as a target since it is a spatially defined cellular structure that is

~ 100-130 µm thick, which is amenable to two-photon patterning. Two-photon patterning is ideal

for small structures that require high spatial resolution such as the retina.

1.2 Hypothesis and objectives

The stated hypothesis governing the body of work is:

Multiple bioactive growth factors can be three-dimensionally and simultaneously immobilized in

distinct volumes within a hydrogel using physical interactions.

3

To test this hypothesis, the following objectives were set:

1. Design a matrix for 3D immobilization of biomolecules

a. Select a suitable matrix

b. Develop a patterning methodology

2. 3D immobilize proteins with a matrix

a. Directly through covalent bonds

b. Indirectly through physical interactions

3. Immobilize 2 differentiation factors simultaneously in a 3D matrix

a. Develop multiple complementary binding partners

b. Quantify amount immobilized

c. Test bioactivity of immobilized factors

1.3 In vitro cell culture

The first attempt to study cells isolated from an organism were made by Ross Harrison in the late

1800s and early 1900s [3]. Tissue culture, more commonly referred to as cell culture today, was

originally developed as a tool to study tissue growth. Ross Harrison is attributed with the

development of the cell culture technique, being the first to successfully culture and experiment

on cells in vitro. The breakthrough experiment by Harrison was a result of his research focusing

on the development of peripheral nerves during embryogenesis. Amazingly, Harrison realized

that nerve outgrowths could not occur in liquid medium and required a solid substrate, probably

the first use of a biomaterial. His first successful in vitro cell culture experiment is referred to as

the hanging drop [4]. He gelled an explant of embryonic frog neural tube fragments within a

lymph and gelatin clot on the surface of a coverslip. The sample was then inverted onto a glass

slide with a depression and nerve fiber outgrowth was observed. This was the first demonstration

4

of directly studying any cellular activity in a controlled fashion. Since then cell culture has

resulted in many advances from the development of vaccines[5, 6] to the production of numerous

drugs [6, 7].

Modern cell culture has evolved to primarily focus on the culturing of cells on two-dimensional

hard surfaces such as tissue-culture polystyrene. Cells can be investigated in a controlled fashion

to determine the effects of chemical and physical cues on cells as well as cell-cell interactions.

Traditional cell culture experiments are still producing important findings today. For example,

Discher et al., Wang et al. and Chen et al. have used 2D surfaces with different stiffness to prove

that substrate stiffness is important for stem cell differentiation and not just the number of

adhesion sites between the cell and the matrix as had been proposed previously [8-10].

1.3.1 Limitations of 2D cell culture

Although cell culture is vital for the investigation of cell physiology, it has become apparent that

2D cultures are not always representative of cells in vivo, which experience 3D environments.

The most apparent difference between 2D and 3D cultures is cell morphology. In 2D cultures

cells adhere and spread on a flat surface, whereas cells in vivo are surrounded by the extracellular

matrix and can spread/adhere in all directions. Furthermore, cell shape has also been shown to

affect cell survival, proliferation and differentiation, solidifying the need for 3D cultures. For

example, Chen et al. demonstrated that the differentiation of human messenchymal stem cells

(hMSCs) is dependent on cell shape. Cell morphology was controlled by micro-patterning

adhesive ligands on surfaces to form islands of adhesion of varying sizes, thus controlling the

region over which a cell could spread. They observed that hMSCs cultured on smaller islands

(1024 µm2) preferentially differentiated into adipocytes whereas osteoblasts were favored on

larger islands (10000 µm2) [11]. In vivo cells are surrounded by a complex environment that

5

manipulates cellular processes; therefore the development of 3D in vitro culture systems that

mimic the extracellular environment is crucial for the understanding of cellular activities.

1.3.2 Extracellular environment

Figure 1-1: Schematic depicting the complexity of the extracellular matrix surrounding

cells in vivo. The ECM contains many insoluble structural components and soluble factors

that influence cell processes. Two-main functions of the ECM are shown: 1) integrin (tan)

binding to adhesion sites (green); and 2) growth factor (red circles) sequestering. Copyright

Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission. [12]

The extracellular environment comprises insoluble and soluble macromolecules manufactured by

cells [13]. The physical structure of the extracellular microenvironment, referred to as the

extracellular matrix (ECM), is formed by 3 classes of hydrated insoluble macromolecules: 1)

fibrillar proteins (collagen), 2) glycoproteins (laminin and fibronectin) and 3) proteoglycans [13,

14]. Beyond structural properties, insoluble macromolecules also contain signaling sequences

which bind to cell surface receptors, such as cell adhesion sites (Figure 1-1). Soluble

macromolecules within the ECM primarily serve as signaling molecules, and are categorized into

6

4 main groups: 1) growth factors, 2) cytokines, 3) chemokines and 4) hormones [13, 15]. The

ECM sequesters these soluble factors and regulates their bioactivity by controlling their

distribution, activation and presentation (Figure 1-1). Together all the components of the ECM

act as a signaling structure with spatial and temporal control.

The chemical composition of the ECM also determines the mechanical properties of different

tissues. The physical characteristics of different tissues are generally compared by their elastic

modulus, which is related to their stiffness. For instance, the elastic modulus of brain tissue (~ 1

kPa) is much lower than that of muscle (~10 kPa) and bone (~ 100 kPa) [16]. Bone ECM is

highly abundant in fibril proteins such as collagen [17], which contributes to the high stiffness of

the tissue. Conversely, the ECM of the brain is mostly devoid of fibril proteins and rich in

lectins, proteoglycans with a lectin and hyaluronic acid binding domain [18], yielding a

relatively weak ECM.

The ECM contains two sources of chemical signals: 1) signaling sequences of the insoluble

matrix; and 2) sequestered biochemicals that interact with the matrix. Adhesion is the most

common cellular interaction with the insoluble matrix; cell-surface receptors such as integrins

and cadherins bind to specific domains of the ECM and activate signaling pathways based on the

chemical composition and orientation (physical structure) of ECM components [15]. ECM

adhesion domains are vital since all cells, except circulating blood cells, must be anchored to a

matrix to remain viable. The ECM also binds many soluble biofactors, acting as a scaffold or

reservoir controlling their presentation and exposure levels to cells [19]. Sequestered factors,

such as growth factors, can act as solid-phase ligands or be released in bursts during ECM

degradation [20]. For example, fibroblast growth factor-2, FGF-2, remains bound to heparin

sulfate proteoglycans acting as solid-phase signaling ligands [15]. The ECM is a complex system

7

with many different signals controlling proliferation [21], viability [22], survival [22],

differentiation [23] and migration [15, 24].

1.4 3D cell culture

Many overlapping signals are received by cells in tissue from the ECM, making it difficult to

independently isolate and investigate different signals. In vitro cell culture in 3D matrices can

recapitulate in vivo environments where variables can be independently controlled for cellular

studies. For instance, the effects of matrix stiffness on stem cell differentiation have been studied

using artificial scaffolds. This is not possible in vivo, since it would be impossible to isolate the

effects of one variable, matrix stiffness [25]. Therefore, 3D cell culture provides conditions that

can elucidate cellular processes.

1.4.1 Scaffolds for 3D cell culture

Because the ECM influences cell fate, building scaffolds that can reproduce different aspects and

functions of the ECM is a major focus in 3D cell culture. Considerable effort has focused on the

development of versatile systems that can spatially present biological signals to cells. Hydrogels,

water swollen crosslinked polymer networks, are the most common scaffold used in tissue

engineering because they closely mimic the extracellular matrix both chemically and physically

[14]. Furthermore, hydrogels are mostly water, greater than 98% for most applications, which

allows for efficient nutrient transfer. Hydrogels also offer the advantage of being easily

chemically tuned [2, 26], for the incorporation of specific biochemical cues. A variety of natural

and synthetic polymers have now been used for various applications in tissue engineering.

Natural polymers derived from the ECM have the advantage of being intrinsically

biocompatible, and contain signaling domains to promote cell survival and viability. Whereas,

8

synthetic hydrogels are blank canvas’ that do not contain signaling domains, making them ideal

for the incorporation of biochemical signals. Table 1.1 lists common hydrogels used to

recapitulate different environments.

Table 1.1: Commonly used polymer scaffolds in 3D cell culture and tissue engineering

Polymer scaffolds Targeted tissue References

Agarose

Neural tissue

Spinal cord [2, 27, 28]

Hyaluronan

Cartilage

Skin

Neural tissues

Adipose tissues

[29-33]

Alginate

Skin

Myocardial and cardiac tissues

Liver

Bone and cartilage

Neural tissues

[34-36]

Chitosan

Neural tissues

Bone and cartilage

Liver

Ligaments and tendons

Skin

Vascular tissues

[37-41]

9

Poly(ethylene glycol)

Cartilage

Bone

Neural

Vascular tissue

[42-45]

Poly(lactic-glycolic) acid

Neural

Cartilage

Bone

[46-48]

1.4.2 Elucidation of mechanical cues through 3D cultures

Culturing cells within hydrogels has elucidated the relationship between cellular activities and

mechanical environments. To study the effect of mechanical environments, cells are cultured

within biocompatible hydrogels with different properties, with matrix stiffness being the most

studied. Gel stiffness is most commonly varied by altering the crosslink density and has been

shown to be an important factor for a number of cellular activities including differentiation.

Leipzig and Shoichet showed that the differentiation profile of NSPCs from the subventricular

zone on chitosan hydrogels varies in relationship to the stiffness of the gels [49]. It was

demonstrated that differentiation towards neurons was enhanced when cultured on soft surfaces

with a Young’s elastic modulus (EY) of less than 0.1kPa whereas oligodendrocyte differentiation

was preferred on stiffer surfaces (EY > 7kPa). Schaffer and Healy have produced similar results,

where NSPCs from the hippocampus were cultured within poly(acrylamide) gels of with varying

elasticities and discovered that neurons were favored on soft gels (0.1-0.5 Pa) and glial

differentiation was favored on stiff gels (1-10 kPa) [50]. Therefore, 3D cell culture has been

vital in understanding the relationship between mechanical stiffness and cell fate.

10

1.4.3 Elucidation of chemical cues through 3D cultures

The chemical modification of hydrogels to mimic the signaling environment of the ECM

provides insight into numerous cellular activities including adhesion, survival, migration, and

differentiation. To this end, hydrogels have been modified to contain a number of biochemical

signals to elucidate the relationship between cell activities and matrix composition [12]. To

mimic the adhesive properties of the ECM, biologically derived molecules such as collagen [51],

laminin [52], fibronectin [53] and short peptide sequences short peptide sequences have been

incorporated. From these studies, it was found that the incorporation of adhesion sites is not only

necessary to control adhesion but often important to cell survival. Most cells are anchorage-

dependent and must adhere to avoid anoikis, programmed cell death in the absence of cell-matrix

interactions. Cellular migration has been studied in hydrogels through the incorporation of

chemoattractant gradients and degradation sites. West et al. demonstrated that vascular smooth

muscle cells migrated down concentration gradients of immobilized fibroblast growth factor 2

(FGF-2) within PEG hydrogels [54]. Proteolytic sites have also been incorporated into hydrogels,

to study the effects of matrix degradation on migration [55]. Furthermore, it has been

demonstrated that growth factors can act as solid phase ligands for the differentiation of stem

cells. For example, immobilized vascular endothelial growth factor (VEGF) in agarose gels has

been shown to guide the differentiation of mouse embryonic stem cells toward blood progenitor

cells [56]. 3D cell culture has already proven to be a useful tool for the elucidation of matrix

composition and cellular functions.

The research presented here focuses on the incorporation of proteins within hydrogels to direct

stem cell differentiation. Therefore, the following sections will discuss different strategies that

have been employed to chemically modify hydrogels with peptides and proteins.

11

1.5 Peptide and Protein Immobilization

1.5.1 Immobilization through covalent bonds

The most common methods for protein immobilization take advantage of the natural reactive

groups in proteins: 1) amines; 2) carboxylic acids; and 3) thiols. Amines are commonly found on

the surface of proteins, and are frequently coupled to hydrogels containing carboxylic acids with

carbodiimide crosslinkers such as 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) for

amide bond formation [57]. Other common strategies for amine coupling include reaction with

N-hydroxysuccinimide-activated esters (NHS-esters), epoxides, sulfonyl chlorides, aldehydes,

isocyanates, and isothiocyanates [58]. Proteins with free carboxylic acids can also be

immobilized onto scaffolds with amines or hydrazides forming amide bonds. All of these

reactions are pH dependent and water sensitive, making amine couplings difficult in water-

swollen hydrogels. Furthermore, amine and carboxylic acid couplings can result in protein

crosslinking since many proteins contain both groups.

Cysteine thiols are another common coupling site for immobilization. Proteins with reactive

thiols can be easily immobilized in thiolated hydrogels through disulfide bonds [59]. Hydrogels

can also be functionalized with thiol reactive groups such maleimides, iodo/bromoacetates,

epoxides and acryloys [60]. These reactions are usually performed near neutral pH and are thus

less susceptible to hydrolysis. These methods can also be applied to proteins without cysteines.

For example, amines can be converted into thiols using Traut’s reagent [61], or cysteine can be

recombinantly added to the protein sequence.

Click reactions are increasingly popular in the biomaterials field since covalent bonds are formed

without by-products, simplifying purification. One of the reactions developed, Huisgen

12

cycloaddition, where a triazole is formed by reacting an alkyne with an azide has been used for

both polymer crosslinking and protein immobilization [62]. The reaction requires copper (I) as a

catalyst, which is cytotoxic [63]. A copper-free variant of the Huisgen cycloaddition has been

recently developed using monofluorinated cyclooctine (MOFO)[64] or a di-fluorinated

cyclooctyne moiety (DIFO3)[26, 64]. In this case, the reaction is driven by the cyclooctyne’s

high ring strain and electron withdrawing fluorine substituents and not a catalyst. Maleimides

have also been used in Diels-Alder reactions with furans [65]. Shoichet et al. demonstrated that

maleimide modified protein could be coupled to nanoparticles functionalized with furans [66]. A

Diels-Alder reaction has also been used to immobilized cyclopentadiene modified adhesion

peptides to quinine functionalized surfaces[67].

Click reactions involving thiols are frequently used since they are naturally occurring functional

groups in peptides. A number of groups have performed a Michael-type addition of thiols to

maleimides for the immobilization of peptides and proteins in hydrogels [2, 68, 69]. The reaction

is performed with a pH between 6.5 and 7.5, such that the majority of amines on the protein are

protonated and thus unreactive. Amines can react with maleimides, although at a pH of 7 the rate

of reaction between a thiol and maleimide is 1000 times faster than that with an amine [70].

Thiols have also been utilized in radical mediated thiol-ene click reactions. Anseth et al. have

demonstrated that peptides containing cysteines can be immobilized in hydrogels functionalized

with an alkene when exposed to light in the presence of a photoiniator [71].

1.5.2 Immobilization using physical interactions

Protein-protein or protein-small molecule physical interactions have been used for protein

immobilization. These interactions are highly specific, thus controlling the region of the protein

used for immobilization [72]. Polyhistine tags, commonly used for immobilization, are a repeat

13

of 6 histidines placed at either the N or C-terminus of the protein, which chelates to certain

transitions metals such as Ni2+. In this case, the hydrogel is modified with the chelator

nitrilotriacetic acid (NTA) and then treated with a solution of Ni2+. Proteins with histidine tags

then chelate with the Ni2+ on the hydrogel. Other affinity tags initially produced for protein

purification have also been utilized. Since they were designed as purification tags, the binding is

reversible and therefore have relatively weak binding with dissociation constants (Kd) on the

order of 1 to 10µM [60]. Therefore immobilization with these methods cannot be considered

permanent and are not ideal for applications requiring long-term protein immobilization.

The avidin/streptavidin-biotin interaction is the strongest known natural physical interaction with

a Kd ~ 10-15 M [73], making it ideal for protein immobilization. Furthermore, avidin and

streptavidin are extremely stable towards heat [74], denaturants, pH variations and proteolysis

[60]. Usually, peptides or proteins are modified with biotin that are then immobilized within

avidin or streptavidin modified hydrogels. For example, biotin modified interferon gamma(IFN-

γ) was immobilized in a streptavidin-chitosan hydrogel to guide the differentiation of neural stem

progenitor cells towards neurons [75]. Proteins are usually biotinylated by reacting amines with

an NHS-ester-biotin, carboxylic acids with an amine-biotin to form an amide, or incorporating a

sequence recognized by the enzyme biotin ligase for the enzymatic addition of biotin [76].

Beyond protein-small molecule interaction, protein-protein interactions have also been used for

immobilization. For instance, Tirrell and co-workers demonstrated that a leucine zipper pair, ZE

and ZR, with Kd ~ 10-15 M are also able to immobilize proteins on 2D surfaces [77]. ZE was

covalently immobilized onto a glass surface, followed by the addition of a fusion protein of

either green fluorescent protein (GFP) or glutathione-S-transferase (GST) with ZR. GFP and

GST were used as proof of principle and not for direct biological applications. Therefore, strong

14

physical binding events (Kd ~ 10-15M) that occur at physiologic conditions are useful tools for

protein immobilization.

1.5.3 Enzyme catalyzed immobilization

Enzymatic reactions have been used to immobilize proteins within pre-functionalized hydrogels.

In this case, hydrogels and proteins are modified with substrates for enzyme mediated

crosslinking reactions. Transferases catalyze the transfer of functional groups from one molecule

to another, thereby forming a covalent bond for protein grafting. For example,

phosphopantetheinyl transferases catalyze a crosslinking reaction between short recognition

sequence with coenzyme A[78]. Therefore, proteins with the recognition sequence are

immobilized on to coenzyme A modified surfaces. Other enzymes such as transpeptidases have

also been used for protein immobilization. Sortase A cleaves the threonine-glycine bond of its

recognition sequence LPXTG, and catalyzes the formation of a peptide bond with an N-terminal

pentaglycine sequence[23]. Therefore, proteins with a C-terminal LPXTG tag can be

immobilized to a surface containing a pentaglycine sequence using Sortase A[79].

1.5.4 Spatial control of biomolecule immobilization

3D patterns are necessary to mimic the spatial distribution of biomolecules in the ECM. To this

end, Luo and Shoichet pioneered the use of photochemistry to three-dimensionally pattern

biomolecules within transparent hydrogels. Agarose was modified with photosensitive S-2-

nitrobenzyl-cysteines, which undergo deprotection upon irradiation. The newly exposed thiols

were then reacted with mal-GRGDS, an adhesion peptide. Immobilization was spatially

controlled by using a focused laser (He-Cd 325nm) for the creation of 3D cylinders of RGD. The

RGD channels were then shown to guide neurite outgrowth from dorsal root ganglion cells

15

(DRGs). This was the first demonstration of spatially controlling cellular activity within a

hydrogel using photochemistry.

UV irradiation does not provide the necessary resolution to create chemically complex hydrogels

to mimic in vivo environments. Spatial resolution in photo-patterning is governed by the

excitation volume, which spans the entire length of the laser beam. Therefore, molecules will be

immobilized throughout the entire depth of the hydrogel. Resolution can be increased by using a

multi-photon laser where the excitation volume is limited to the focal point of the laser (Figure

1-2).

1.6 Two-photon excitation

Figure 1-2: Comparing the excitation volume of one and two-photon irradiation of a

fluorescein solution. One-photon excitation (bottom laser) results in the excitation of

fluorescein throughout the path of the laser. The two-photon excitation volume is limited to

the focal point of the laser (top laser). Reproduced with permission from Prof. Kevin

Belfield [80].

It is simplest to first consider the excitation of a fluorophore to explain two-photon (TP)

excitation. For example, the traditional one-photon excitation of fluorescein can be achieved with

a 380nm laser. This can be seen in Figure 1.2 (bottom), where a fluorescein solution is excited

throughout the entire path of the laser. Excitation can also be achieved using a TP laser at 760nm

16

where the excitation is limited to the focal point of the laser (Figure 1-2) [81]. In TP excitation,

fluorescein must absorb two photons simultaneously. This can only occur in a region of high

laser intensity, the focal point of the laser. Therefore, the high resolution of TP excitation is

achieved by restricting the volume of excitation.

The probability that a molecule will absorb two-photons is a function of both the spatial and

temporal overlap of photons [82]. After the absorption of one photon, the molecule enters a

virtual intermediate state with a lifetime of less than a femtosecond. Therefore, the second

photon must be absorbed within a femtosecond, which only occurs in regions of high photon

concentration. To achieve TP excitation, photons from femtosecond pulsed lasers are focused

through objectives of high numerical aperture. Even then, the only region with a sufficient

concentration of photons for excitation is the focal point of the laser [83]. Figure 1-3

demonstrates how the pulses for a focused laser result in a confined region of excitation.

Figure 1-3: Excitation of fluorescent molecules with microscope equipped with a pulsed

multi-photon laser. Laser pulses are concentrated at the focal point to achieve the intensity

of light needed for two-photon excitation. Reproduced with permission from Dr. Michael

Davidson[84].

17

1.6.1 Cross-section for two-photon absorption

The most common method to evaluate two-photon absorption properties of molecules is to

compare their TP absorbance cross-sections [85]. TP cross-sections are measurable properties of

light absorbing molecules, and are related to the amount of photons absorbed per second and

intensity of light [86].

In one photon absorption, the number of photons absorbed per second (NA) is given by:

NA (photon/s) = δ (cm2) I (photons/cm2s)

Where δ is the cross-section for one-photon absorption and I is the intensity of light. The cross-

section represents the effective area over which a single molecule can absorb light. For one-

photon absorption the cross-section is approximately the size of the fluorophore. Therefore, the

cross-section is determined by measuring the number of photon absorbed for a given intensity of

light.

In two-photon absorption the number of photons absorbed per second is given by:

NA2 (photons/s) = δ2 (cm4s/photon) I2 (photons/cm2s)2

The value for δ2 is expressed in Gopper-Mayer (GM) units, where 1 GM = 10-50cm4s/photon

[85]. δ2 are experimentally determined and used as a measure for TP activity, where molecules

with good TP absorption properties have high δ2.

Two main classes of two-photon active molecules have been utilized for the 3D patterning of

biomolecules: 1) photoiniators or photosensitizers [26, 87]; and 2) photocages [2, 88].

18

1.6.2 Photoiniators/sensitizers for biological applications

Photoinitiators or photosensitizers are used for radical polymerization, which result in the

polymerization of proteins into 3D structures or the grafting of biomolecules to functionalized

hydrogels. TP induced polymerization of acrylates [89], vinyl ether [90], thiol-ene monomers

[26, 71] and proteins [87] have been previously demonstrated. All of these polymerizations rely

on photoinitiators or sensitizers that are excited by TP light [91]. Photoinitiators are generally

small molecules that absorb light forming radicals upon excitation that initiate or propagate a

reaction. Photosensitizers are used to increase the reaction rate by absorbing and transferring

radiation to photoinitiators. Since common photoinitiators have low δ2, much research is

currently focused on developing new initiators with improved photochemical properties. With

higher δ2, TP polymerization will be achieved with lower laser power and decreased irradiation

time. This will not only decrease fabrication time, but also minimize optical damage of sensitive

materials. Table 1 lists common commercially available photoiniators and sensitizers used for

3D polymerization.

Table 1.2: Common photoiniators or photosensitizers for TP polymerization

Compound TP λ (nm) Approx. TP-cross-section (δ2, GM)

Irgacure 184 530 23 [92]

Irgacure 261 530 20 [92]

Irgacure 651 530 28 [92]

Irgacure 754 530 21 [92]

19

Irgacure OXE01 660 31 [92]

Irgacure 2959* 720[26] Unknown

Darocure MBF 530 27 [92]

Darocure 1173 530 20 [92]

Methylene Blue 740 7 [93]

Rose Bengal 800 10[94]

Eosin 800 10 [94]

Erythrosin 800 10 [94]

*Even though Irgacure 2959 has not been well characterized, it is widely used in the bioengineering field because of

its relatively high water solubility [95].

1.6.3 Two-photon uncaging and photolabile protecting groups

This research focuses on the uncaging of functional groups similar to the Luo and Shoichet work

described above, where a deprotection reaction occurs as a result of excitation by light. TP

uncaging sensitivity is described by the uncaging action cross-section δu (expressed in GM),

which is the product of δ2 and the uncaging quantum yield Qu [85].

δu = δ2 · Qu

The quantum yield of uncaging is defined as the ratio of uncaged product over amount of

photons absorbed. Generally, the photolabile protecting group should have a δu of at least 0.1

20

GM to be considered TP active. Table 1.2 list common photocages which have been used for

biological applications.

Table 1.3: Common photocages for biomolecules (Adapted from [96])

Caging group Caged molecule δu (GM) Ref.

2-Nitrobenzyl

Acetate 0.03 (740 nm) [85]

Ca2+ 0.6 (720 nm) [97]

O

NO2

O O

O

NHR

Coumarin 0.68 (740 nm) [98]

6-Bromo-7-hydroxycoumarin-4ylmethyl (Bhc)

Glutamate 0.89 (740 nm) [85]

Glutamate 0.95 (740 nm) [85]

Ketone or aldehyde 0.51–1.23 (740 nm) [99]

21

8-Bromo-7-Hydroxyquinoline (BHQ)

Acetate 0.59 (740 nm) [100]

Phosphate 0.43 (740 nm) [101]

o-Nitrophenylalkyl

Glutamate 0.45 (800 nm) [102]

Nitrobenzyl based molecules were not considered TP active until recently. Interestingly, two

nitrobenzyl based photocages have been shown to be two-photon active with δu ~ 0.6-0.7 GM.

Nitrodibenzylfuran (NDBF) derivative of EGTA was used as a cage to release Ca2+ upon

irradiation [97]. Similarly, 1-(2-nitrophenyl)ethyl (NPE) was used as a fluorescent cage for

coumarin, by covalently linking to the coumarin hydroxyl. After photolysis the coumarin

fluorescence increased 200-fold [98]. Although some examples of nitrobenzyl based two-photon

cages exists, the one used by Luo and Shoichet, 2-nitrobenzyl, was not two-photon active.

Therefore, alternatives needed to be explored.

Tsien et al. developed a coumarin based photolabile protection group, brominated 7-

hydroxycoumarin4-ylmethyls (Bhc), with a δu of ~ 1GM for the two-photon uncaging of

carboxylic acids, amines and phosphates [85]. They further demonstrated that they were able to

perform the uncaging reaction in tissue slices to study the glutamate sensitivity of neurons with

22

three-dimensional control. This was the first demonstration of a versatile biologically relevant

two-photon cage. Dore et al. have further demonstrated that the Bhc-diols can be used as a

protecting group for aldehydes and ketones [99]. Wosnick and Shoichet have used Bhc as the

only TP cage for sulfhydryls [88].

Dore et al. have also developed another two-photon cage for carboxylic acids, 8-bromo-7-

hydroxyquinoline (BHQ) [100]. The authors compared the photochemical properties and

stability of BHQ and Bhc as a photocage for carboxylic acids, BHQ-OAc and Bhc-OAc. Both

molecules have a maximum absorbance of ~370nm, although the δu of Bhc-OAc was greater

than that of BHQ-OAc (0.72 to 0.59 GM). Furthermore, Bhc-OAc was less susceptible to

hydrolysis when not exposed to light. The hydrolysis of both molecules followed a simple

decaying exponential curve, where Bhc-OAc and BHQ-OAc had a time constant of 517 and 70.9

h, respectively. Even though Bhc has better spectral characteristics, BHQ will be useful in a

number of biological applications due to its relatively high water solubility. Furthermore, BHQ

is a versatile cage having been used as a cage for phosphate and diols [101].

1.7 Three-dimensional photopatterning of biomolecules within

hydrogels

1.7.1 Scaffolds for 3D patterning

For biochemical photopatterning studies, hydrogels must have the following properties: 1)

transparent; 2) biologically inert; 3) minimal non-specific absorption; and 4) be reactive for

biomolecule immobilization. Hydrogels must be transparent not only to perform

photochemistry, but also for cell imaging purposes. Because the goal of patterning is to tailor in

23

bioactive molecules, hydrogels with inherent biological properties must be avoided. For instance,

adhesion sites cannot be patterned in collagen hydrogels since collagen is naturally adhesive for

most cells. The gel also must not absorb biomolecules, since specific localization would be

impossible. Therefore, hydrogels with charges should be avoided to limit protein absorption

through ionic interactions. The gel should also have functional groups readily available for

modification with groups for the photopatterning of biomolecules.

The only two polymers used in the 3D bio-patterning field that meet the above criteria are

polyethylene glycol (PEG) and agarose. Currently, most PEG hydrogels are formed using click

chemistry. For example the Anseth group has utlilized hydrogels made by reacting four arm PEG

tera-azide with alkyne (di-fluorinated cyclooctine) di-functionalized peptides in aqueous

conditions at 37°C [26]. In PEG hydrogels the crosslinker, in this case a peptide, must be

synthesized with a functional group for photochemical immobilization of biomolecules.

Similarly, West et al. have used photocrosslinked PEG hydrogels, where diacrylates PEG

polymers were crosslinked with a photoiniators [103]. The unreacted acrylates are then used as

grafting sites for biomolecules. Agarose has also been used because of its simple thermal

gelation mechanism, and ease of chemical modification due to its hydroxyl groups [2, 88].

1.7.2 Biomolecule patterning in hydrogels

3D patterns of biomolecules are created by photochemically reacting peptides or proteins to

hydrogels using a two-photon (TP) laser. Since the reaction only occurs at the focal point, 3D

patterns are created by controlling the location of the focal point in the gel. This is accomplished

by either moving the focal point of the laser or moving the hydrogel (Figure 1-4). The methods

currently used for 3D photopatterning can be categorized into three groups: 1)

photopolymerization; 2) photografting; and 3) photo-uncaging.

24

Figure 1-4: Methodology for the 3D patterning of biomolecules in hydrogels using two-

photon lasers. The site of immobilization is controlled by the path of the TP laser focal

point.

1.7.3 Photopolymerization

Two-photon (TP) polymerization involves the use of a photoinitiator that produces radicals in the

presence of infrared light for radical polymerization. Once excited the photoiniator produces

singlet oxygen, which then reacts with an amine and aromatic groups of the protein [104]. The

resulting protein radical will then react with other proteins resulting in polymerization [105,

106].

Schmidt and Shear have demonstrated that 3D structures of protein constructs can be

incorporated into hydrogels [87]. Fluorescently labeled BSA was crosslinked within agarose or

hyaluronic acid hydrogels using the photoiniator methylene blue with TP irradiation (Figure

1-5). The hydrogels were first soaked with fluorescent BSA and photoiniator, followed by

patterning with a multi-photon laser. The gel was washed to remove any unreacted protein and

then imaged. This method has been expanded to include the immobilization of biotinylated

peptides such as IKVAV [87]. First biotinylated BSA was patterned as above, followed by the

25

addition of neutravidin, which binds up to 4 biotins. The resulting neutravidin pattern was then

free to bind biotinylated peptides. They further demonstrated that two different molecules can be

sequentially patterned within the same hydrogel. Figure 1-5 shows a 3D pattern with fluorescent

BSA (green) and IKVAV (red) immobilized in distinct volumes. Fluorescent BSA was first

immobilized, followed by the patterning of biotinylated fluorescent BSA for the immobilization

of biotin-IKVAV.

Figure 1-5: Photopolymerization of fluorescently labeled BSA and IKVAV in hydrogels. (a)

Fluorescently labeled BSA was photocrosslinked within an agarose hydrogel at different

laser intensities (left to right: low to high intensity). The increased fluorescent signal in the

samples exposed to high laser intensity indicates that the amount polymerized can be

controlled by varying the laser intensity. (b) Fluorescent BSA (green) and IKVAV (red)

were sequentially patterned within the same hydrogel. Copyright Wiley-VCH Verlag

GmbH & Co. KGaA. Reproduced with permission. [87]

Campagnola et al. used a similar technique to immobilize 3D structures of BSA, fibronectin,

fibrinogen, collagen, laminin, concavilin and alkaline phosphatase with photoiniators such as

rose Bengal [104, 107-109]. Importantly, they demonstrated that the polymerized proteins

retained their activity, fibroblasts adhered and spread on 3D patterns of fibronectin [106].

Furthermore, patterns of alkaline phosphatase retained enzymatic activity after being crosslinked,

indicating that harsh reaction conditions such as radical polymerization is not necessarily

a b

26

detrimental to protein function. Campagnola is further developing this technique to fabricate

fibers of laminin as a model to study adhesion and migration of ovarian cancer cells [109].

The radical photopolymerization of proteins has had great success in the field of biomaterials,

although several issues still need to be addressed. First, many small molecule photoiniators are

toxic and cannot be used in the presence of cells. Second, one of the main motivations for 3D

patterns of proteins was to create three-dimensional culture environments for cells. The 3D

patterns created through photopolymerization create protein fibers. The cells may view these

fibers as 2D surfaces and not 3D environments. It is also hard to imagine that all proteins will

remain bioactive after photopolymerization. Therefore the system used by Schmidt and Shear

where biotinylated peptides are grafted onto the polymerized fibers could prove advantageous,

since the bioactive protein is not photopolymerized.

1.7.4 Photo-grafting

Photo-grafting involves the crosslinking of the hydrogel and the biomolecule in the presence of a

photoiniator. This method will not create fibers of proteins as with the photopolymerization

method described above since the biomolecules are immobilized directly onto the hydrogel.

West et al. have demonstrated a 3D grafting method in polyethylene glycol (PEG) hydrogels

using photoiniators [103, 110]. First PEG diacrylate (PEGDA) hydrogels were formed by

crosslinking PEGDA in the presence of the photoiniator 2,2-dimethoxy-2-phenyl-acetophenone

(DMAP) in N-vinylpyrrolidone (NVP) under UV light (365nm). After gelation the remaining

acrylates are used as attachment sites for biomolecules. A solution of fluorescently labeled

acryloyl (ACRL)-PEG-peptide (RGDS) with DMAP in NVP was soaked into the gel. Using a

rastering TP laser, the gel was selectively 3D patterned with the RGDS peptide. The amount of

peptide immobilized was controlled by varying the irradiation exposure time, beam intensity or

27

laser scan speed. Building on this technology West and Hoffman have demonstrated that three

different peptides can be 3D patterned within the same hydrogel[103]. As proof of concept, three

different fluorescently labeled monoacrylate-PEG-RGDS (Alexa Fluor 488, 532 and 633) were

sequentially patterned into the hydrogel. They demonstrated that this system can create 3D

patterns of RGD in collagenase degradable PEG hydrogels that guided fibroblast migration,

where the fibroblast were encapsulated within a fibrin clot within the gel[110]. Similarly,

endothelial cells were shown to undergo tubulogenesis in channels of RGD and vascular

endothelial growth factor (VEGF)[111].

The Anseth group has also demonstrated patterning in PEG hydrogels using thiol-ene chemistry

[26]. The hydrogels are formed by a copper free click reaction between PEG tetra-azide and a

difunctional di-fluorinated cyclooctyne peptide crosslinker. This reaction is performed under

physiological conditions and has been used to encapsulate cells with no apparent toxicity. The

peptide crosslinker incorporated an alkene as a reaction site for the immobilization of thiol

containing peptides. In the presence of a photoactive hydrogen abstracting initiator, thiols are

deprotonated to thiyl radicals for reaction with alkenes in the gel. This system allows any thiol-

containing molecule to be patterned. As in other systems, the amount of peptide immobilized is

controlled by the amount of TP irradiation. The system has also been used to immobilize three

different peptides in the same gel through sequential reactions [26]. Furthermore, fibroblasts

were shown to preferentially migrate to regions patterned with RGD indicating the peptides

remain bioactive after immobilization and cellular migration can be controlled with 3D

precision.

28

1.7.5 Photo-uncaging

Building on work by Luo and Shoichet, agarose hydrogels with either primary amines [112] or

thiols [88] protected with the photocage Bhc were synthesized. Work by Luo and Shoichet

utilized a nitrobenzyl photocage for thiols for the creation of 3D patterns [2]. As mentioned

early, nitrobenzyl has a poor uncaging cross-section δu and would not function efficiently for TP

uncaging. Therefore, nitrobenzyl was replaced with Bhc, which has previously been

demonstrated to be an efficient TP photocage for biological applications. The caging of primary

amines with Bhc within an agarose hydrogel is described in Chapter 2.

Wosnick and Shoichet demonstrated for the first time a two-photon active photocage for

sulfhydryls, utilizing Bhc [88]. Previously, Bhc was only shown to be an effective cage for

amines, carboxylic acids, phosphates, ketones and aldehydes [85, 99]. Furthermore, 3D

patterning of thiol reactive molecules was achieved in agarose hydrogels modified with Bhc

caged thiol. A sulfide protected Bhc amine was synthesized by reacting 6-Bromo-4-

chloromethyl-7-hydroxycoumarin with commercially available Boc-protected

mercaptoethylamine. Tsien et al. previously published the procedure for the synthesis of 1[85].

After deprotection, 3 was grafted to agarose polymer chains using carbodiimide chemistry

yielding a substitution rate of 0.2 molecules of 3 for each repeat unit of agarose. Agarose

hydrogels with Bhc protected thiols were then irradiated in the presence maleimide (mal)-Alexa

Fluor 488 to demonstrate 3D control of irradiation. As with all TP patterning methods, lateral

resolution of a few microns was achieved. Although less control is afforded over axial resolution

since the focal point of a laser is elliptical, this is governed by the optics used. In this case with a

20x lens (NA 0.4), an axial resolution of ~20 µm was achieved. Similar to the system developed

by West and Anseth, multiple different molecules can be independently patterned within the

29

same hydrogel. Two fluorophores, mal-Alexa Fluor 488 and mal-Alexa Fluor 546 (Figure 1-6),

were sequentially patterned within the same hydrogel.

Figure 1-6: Maleimide-Alexa Fluor 488 and 546 were sequentially 3D immobilized in

coumarin-sulfide modified agarose hydrogels, (a) oblique and (b) side view. The layered

structure of green squares (488) and red circles (546; ~ 50 µm in diameter) demonstrate the

high spatial achievable through TP patterning. Reproduced with permission [113].

1.8 Stem and progenitor cells and patterned hydrogels

The research conducted for this thesis was towards the design of biomaterials for stem/progenitor

cells for applications in tissue engineering. The long term goal of this research is to synthesize

biomaterials to control spatially control the differentiation of stem/progenitor cells and thus

cellular organization within hydrogels. This can then lead to the engineering of tissues de novo

for the replacement or regeneration of tissues. Therefore the following section will discuss neural

and retinal stem cells and methods for their incorporation into protein patterned hydrogels.

1.8.1 Adult neural stem cells

Neural stem progenitor cells (NSPCs) are multipotent proliferative cells that give rise to all three

cell types of the central nervous system: astrocytes, oligodendrocytes and neurons. Weiss and

Reynolds were the first to isolate adult NSPCs from the mouse brain striatum [114], since then

30

NSPCs have also been discovered in the dentate gyrus of the hippocampus [115], cells adjacent

to the central canal of the spinal cord [116] and the subventricular zone [117, 118]. NSPCs are

expanded by culturing in the presence of epidermal growth factor (EGF) and fibroblast growth

factor-2 (FGF-2). Cell culture experiments have identified a wide range of methods for the

directed differentiation of NSPCs. For instance, NSPCs cultured in the presence of interferon-

gamma (IFN-γ) [119], platelet derived growth factor-AA (PDGF) [120] or CNTF [121] will

preferentially give rise to neurons, oligodendrocytes or astrocytes, respectively. The fact that

NSPCs can proliferate and are multipotent makes them an ideal candidate for the regeneration of

neural tissue. In this work, the migration of NSPCs into patterned hydrogels was investigated.

1.8.2 Adult retinal precursor cells

Retinal stem cells (RSCs) self-renew and are multipotent, meaning they can give rise to the cell

types of the retina given the proper environmental cues. RSCs, similar to other stem cells, also

give rise to retinal progenitor cells (RPCs) that are still capable of differentiating into different

cell types but will no longer self-replicate indefinitely. Precursor cells isolated from the eye will

contain a mixed population of retinal stem and progenitor cells (RSPCs). These cells have the

capacity to differentiate into the 7 types of retinal cells: 1) rod photoreceptors; 2) cone

photoreceptors; 3) amacrine cells; 4) bipolar cells; 5) horizontal cells; 6) ganglion cells; and 7)

Müller glia [122]. van der Kooy et al. were the first to report adult mammalian RSPCs [122]. The

cells were isolated from the pigmented cells of the ciliary margin in the murine eye and shown to

proliferate in vitro and self-renew giving rise to clonally derived spheres. The number of spheres

can also be expanded in vitro in the presence of fibroblast growth factor-2 (FGF-2). Importantly,

the cells were shown to differentiate into many cell types of the retina when cultured under

31

differentiation conditions. RSPCs have also been isolated from the human eye [123], making

RSPCs a relevant choice for medical applications.

The differentiation profile of RSPCs with soluble factors, growth factors, has almost exclusively

been elucidated through in vitro 2D cell culture. Several chemical cues have been identified from

the literature as differentiation factors. Ciliary neurotrophic factor (CNTF) differentiates RSPCs

into either bipolar cells [124, 125] or Müller glia [125, 126] depending on concentration. N-

terminal sonic hedgehog (SHH) has been shown to increase the number of rod photoreceptors

[127]. An increase in ganglion cells has been observed with bone morphogenetic protein 4

(BMP-4) [128].

1.8.3 Cell penetration into hydrogels – proteolytic and non-proteolytic

migration

The lack of cell migration into hydrogel constructs is one of the main limitations facing the tissue

engineering field. Cell migration is important for both in vitro experiments where cells must be

seeded into hydrogels, and in vivo where endogenous cells must interact with biomaterial

implants. Although important, the field is just beginning to address the complex factors that

influence 3D cell migration. For hydrogels with 3D immobilized proteins, cells must be

incorporated after the patterning process. If cells are pre-encapsulated they will be exposed to

soluble protein during the patterning process, and this may interfere with the spatial control of

biological activity. Small factors such as peptides which readily diffuse through hydrogels may

be patterned in the presence of cells, since the soluble form will only be present for a limited

amount of time[26]. Although, larger macromolecules, such as proteins, may take several hours

to days to diffuse out of hydrogels.

32

Cells migrate through a 3D matrix either through matrix degradation (proteolytic) or amoeboid-

like mechanisms (non-proteolytic). Proteolytic migration occurs when cells express proteases

that degrade the surrounding matrix. Once degraded, cells can migrate through degraded regions.

Non-proteolytic migration occurs without matrix degradation but rather matrix deformation or

displacement. Therefore, non-proteolytic migration will only occur in soft matrices, which can

be deformed.

Proteolytic degradation of the ECM is controlled by proteases such as matrix metalloproteinases

(MMPs), plasmin, and elastases. MMPs are a family of secreted or membrane bound zinc

dependent proteases, which collectively can degrade ECM components such as collagen,

proteogycans, and glycoprotein. Similarly, plasmins and elastases degrade fibrin and elastin

respectively. A number of groups have taken advantage of protease secretion for cell migration

in hydrogels. West et al. demonstrated that cell migration and protease activity are directly linked

in fibroblast migration within PEG hydrogels [129]. PEG hydrogels were crosslinked with a

peptide corresponding to the collagenase cleavage sequence (LGPA). The peptide was modified

with two Bodipy groups, such that fluorescence is increased after the sequence is cleaved due to

FRET. Therefore, the enzymatic degradation of the matrix can be followed in three-dimensions

and related to cell migration. It was found that fibroblast migration within PEG hydrogels was

directly coupled with collagenase activity, migration only occurred within degraded regions.

West has further developed their system to control cell migration using 3D patterns of RGD, an

adhesive peptide. In this case fibroblasts only migrated within areas containing the adhesive

ligand. Therefore, both matrix degradation and adhesion are necessary for migration. Anseth et

al. have also incorporated a similar system into PEG hydrogels formed by non-cytotoxic copper

free click reactions [26]. The crosslinker contained a collagenase degradable sequence. Again,

33

3T3 fibroblasts only migrated into regions that were enzymatically degraded with an adhesive

ligand.

As mentioned earlier non-proteolytic migration also occurs with the natural ECM. This method

of migration is limited by hydrogel pore sizes which are generally smaller than cellular

dimensions. For example, work by Hubbell has shown that the pore size of PEG hydrogels are

around 25 nm whereas collagen hydrogels have pores reaching 7.4 µm [130]. The average pore

diameter of 0.5 to 1 wt % agarose hydrogels ranges from 300 to 500 nm [131]. Interestingly,

Lutolf et al. have recently published an article showing that preosteoblast MC3T3-E1 cells can

migrate within non-degradable PEG hydrogels [132].

Lutolf et al. studied the migration of MC3T3-E1 cells in both degradable and non-degradable

hydrogels to compare proteolytic and non-proteolytic migration. In the weak gels (storage

moduli of ~ 100 Pa) cell migration occurred at the same rate in both gels, indicating that non-

proteolytic migration was the major contributor. The authors hypothesized that cells can migrate

without degradation through matrix deformation. The authors verified their hypothesis by

fluorescently labeling the hydrogel. In proteolytic migration, non-fluorescent volumes were

created indicating that the cells degraded the matrix in their migration tract. Although in non-

proteolytic migration, no void volumes were observed and increased fluorescence was observed

around the cells. Therefore, the cells were compressing the matrix around them to produce a

physical volume for migration. These experiments demonstrate that non-proteolytic cell

migration does occur in hydrogels depending on matrix stiffness.

34

1.8.4 Retina as a tissue engineering target

The retina is divided into three main layers of cells with the photoreceptors on top, followed by

the bipolar cells and then the ganglion cells. The horizontal cells lie at the interface of the

photoreceptors and bipolar cells whereas the amacrine cells are at the junction of the bipolar and

ganglion cells. Müller glia cells span the entire retina, and isolate the neural cells from one

another except at the synapses. The entire multi-cellular structure spans only 100 to 130µm in

humans [133]. Cone cells are responsible for colour vision and function best in light, whereas

rods have evolved from cones to function in situations of low light intensity. Both photoreceptors

transfer electrical signals directly to bipolar cells [134]. The horizontal cells, the least abundant

retinal cell type, are present at the junction of photoreceptors, and provide feedback to cones and

rods to enhance contrast between light and dark regions. In other words horizontal cells can vary

signal responses to overall levels of illumination [134]. Bipolar cells transmit electrical signals

from the photoreceptors to the ganglion cells either directly or indirectly through amacrine cells

[134]. Amacrine cells form the majority of the synapses with ganglion cells, and act as a

modulator of transmitted signals between bipolar and ganglion cells through feedback loops.

Their main function is to control ganglion cell responses. Ganglion cells for the optic nerve

transmit signals from bipolar and amacrine cells to the brain [134].

Because of the retina’s relatively simple layered structured, we investigated the design of a

simplified, biomimetic retina. To achieve an engineered retina, we propose to spatially guide the

differentiation of RSPCs within hydrogels using a 3D protein patterning system. Therefore by

localizing differentiation factors in select regions, the cell type of that region can be controlled.

35

1.9 Summary of research

The following chapters will discuss the development of a protein patterning systems with 3D

control. This system was designed as a tool for the creation of complex protein patterns to

control cellular activities in 3D. A two-photon patterning system for hydrogels is described in

Chapter 2[112], where agarose was modified with Bhc protected amines, which are deprotected

after exposure to light. Because Bhc is two-photon active (δu ~ 1), we were able to 3D control

the location of deprotection of amines within the hydrogel using a multi-photon laser. Using this

method, chemical patterns were created in gels with micron resolution.

Chapter 3 demonstrates the ability to pattern FGF-2 within Bhc-thiol agarose gels using two

different methods. The first procedure immobilized FGF-2 directly to uncaged agarose thiols

through disulfide bonds. This method would function for any protein that contains free

cysteines; FGF-2 contains two accessible cysteines. The second immobilization procedure took

advantage the strong physical interaction between human serum albumin (HSA) and the albumin

binding domain (ABD). Thiol reactive mal-HSA was first photo-patterned into the gel, followed

by the addition of FGF-2 as a fusion protein with ABD. This system was developed as a

universal system for protein immobilization, where any protein can be immobilized once

expressed as an ABD fusion protein. Furthermore, both immobilization procedures occur under

mild conditions to limit bioactivity loss.

The simultaneous patterning of proteins for the creation of complex patterns was described in

Chapter 4. In this case, sonic hedgehog (SHH) and ciliary neurotrophic factor (CNTF) were

immobilized using the binding partners barnase/barstar and streptavidin/biotin, respectively. The

gels were sequentially patterned with maleimide-barnase and maleimide-streptavidin, followed

36

by the addition of the fusion proteins barstar-SHH and biotin-CNTF. This system allows for the

simultaneous immobilization of proteins at the final hydrogel fabrication step. Therefore, the

proteins are not exposed to any potentially harmful patterning conditions, which limit bioactivity

loss.

37

2 Two-photon micropatterning of amines within an agarose hydrogel*

*This chapter was published in the Journal of Materials Chemistry. Wylie, R. G.; Shoichet, M. S., Two-photon micropatterning of amines within an agarose hydrogel. Journal of Materials Chemistry 2008, 18 (23), 2716-2721.

2.1 Abstract

The ability to create three-dimensional micropatterns within polymeric materials is applicable in

a wide number of fields, from photonic bandgaps to tissue engineering. We are particularly

interested in three-dimensional chemical patterning of soft materials with a view towards their

use in regenerative medicine. To this end, we created three dimensional micropatterns of amines

within an agarose hydrogel using two-photon patterning. Agarose was first modified with caged

amines, using a derivative of 6-bromo-7-hydroxycoumarin, which upon two-photon excitation,

cleaved the coumarin molecule thereby yielding primary amine-functionalized agarose. Three

dimensional micropatterns were achieved because the excitation / deprotection reaction was

limited to the focal volume of the two-photon laser absorbance. The three-dimensional amines

serve as reactive sites for further water-based chemistry and may also render agarose cell

adhesive in those amine-containing volumes.

2.2 Introduction

Micropatterning has become a rapidly expanding field in areas such as microelectronics,

photonics, tissue engineering and microfluidics[135, 136]. The ability to create complex patterns

on the micron scale is crucial for the design of future materials. Photolithography is one of the

more common methods to create micropatterns, although these techniques have been limited to

38

the generation of two-dimensional (2D) structures. For biomaterials applications, for example,

amines have been patterned on the polymeric scaffolds for the immobilization of

oligonucleotides[137], proteins[138] and carbohydrate microarrays[139].

Since the creation of patterns using photolithography is governed by the excitation of molecules

with a laser, it is possible to increase the spatial resolution and decrease the excitation volume by

utilizing two-photon irradiation. For two-photon excitation, a molecule must absorb two or more

photons simultaneously in order to reach the excited state, requiring a high intensity of light.

This intensity can be achieved using a pulsed laser that is focused through a microscope lens

where the photons being emitted are equal to half the energy required for excitation. In this case

excitation is limited to the focal point of the laser since the absorption of two or more photons

depends non-linearly on the light intensity[140]. Therefore two-photon lithography provides the

spatial control needed for 3D patterning thereby overcoming the limitations associated with

traditional 2D photolithography. A variety of different 3D microstructures have been produced

using this technique including conductive metal/polymer hybrid devices[140], as well as micro-

chains and springs[141]. Typically a laser is used to control the polymerization (or crosslinking)

of polymers through the excitation of a radical initiator.

In contrast to the methods discussed above, the micropatterning described herein involves

chemical modification of hydrogel scaffolds by the two-photon uncaging of functional groups.

This technique results in minimal changes to the material’s mechanical and structural properties

while modifying the local chemical environment through the placement of specific functional

groups[2, 28, 88, 142]. Using the spatial control associated with two photon irradiation, we now

demonstrate, for the first time, how primary amines can be three-dimensionally patterned within

agarose hydrogels. The natural polymer agarose was chosen as the scaffold since it is a

39

transparent three-dimensionally networked hydrogel that has hydroxyl groups available for

chemical modification and is itself non-adhesive to cells, thereby allowing this functionality to

be dialed in through chemical modification. Three-dimensional (3D) micropatterning of amine

groups in agarose is desirable because the amine functional group is stable in water, serving as a

site itself for either cellular interaction or further modification with cell-specific molecules.

Herein we describe the modification of agarose scaffolds with coumarin-caged amines that are

deprotected upon irradiation with a pulsed laser to yield primary amines. 6-bromo-7-

hydroxycoumarin was chosen as the photolabile group since it is known as a photocage for

amines and is two-photon active[143]. By selectively positioning the focal point of the pulsed

laser, the location, volume and concentration of free amines within the agarose scaffold can be

three-dimensionally controlled. Two-photon patterning of amines could prove useful in

materials engineering by supplying the spatial control and chemical modifications needed for the

construction of complex materials, by either covalent or non-covalent (electrostatic) interactions.

2.3 Materials and Methods

2.3.1 Materials

All reagents were used as received unless otherwise noted. Agarose type IX-A, carbonyl

diimidazole, dimethylaminopyridine, triethylamine and sodium cyanide were purchased for

Sigma-Aldrich (Oakville, ON, Canada). Dichloromethane and trifluoroacetic acid were

purchased from Caledon Labs (Georgetown, ON, Canada). CBQCA was purchased from

Invitrogen Inc. (CA, USA).

40

2.3.2 Synthesis of Boc-protected amino coumarin (3).

A solution of compound 1 (500 mg, 1.84 mmol), carbonyl diimidazole (359 mg, 2.21 mmol) and

dimethylaminopyridine (450 mg, 3.68 mmol) in 150 ml of dichloromethane was stirred under

nitrogen in the dark at room temperature for 3 hours. Tert-butyl 2-aminoethylcarbamate 2 (354

mg, 2.21 mmol) was then added and stirred for an additional 24 hours. The solution was washed

with 100 ml of a 10% citric acid solution, distilled water and brine. After drying over

magnesium sulfate the solution was concentrated to yield a light yellow solid. The product was

purified by reverse phase preparative HPLC using a gradient mixture of acetonitrile to water

(10%-80%) to yield compound 3 as an off-white solid (240 mg, 28.5%). mp 170°C. 1H NMR

(400 MHz, DMSO-d6): 1.35 (s, 9H), 3.00 (m, 4H), 5.24 (s, 2H), 6.18 (s, 1H), 6.87 (m, 1H), 6.89

(s, 1H), 7.51 (m, 1H), 7.86 (s, 1H), 11.47 (s, 1H). 13C NMR (400 MHz, Acetone-d6): δ 27.9,

40.4, 41.4, 61.2, 78.6, 103.8, 106.2, 109.7, 111.7, 128.8, 150.6, 154.7, 155.8, 157.4, 159.7. ESI-

MS (M+): 456.1 (calc: 456.05). Found: C, 47.0; H, 4.4; N, 6.2. Calc. for C18H21BrN2O7: C,

47.3; H, 4.6; N, 6.1%.

2.3.3 Synthesis of aminocoumarin (4).

200 mg of 3 was stirred in 10 ml of a 10:1 (v/v) mixture of dichloromethane and trifluoroacetic

acid for 24 hours. The solution was concentrated, redissolved in water and lyophilized yielding

to an off white solid (347 mg, 100%). mp 184°C. 1H NMR (400 MHz, DMSO-d6): δ 3.07 (t, 2H,

J = 5.85Hz), 3.39 (t, 2H, J = 5.91Hz), 4.99 (s, 2H), 6.08 (s, 1H), 6.57 (s, 1H), 7.47 (s, 1H). 13C

NMR (400 MHz, Acetone-d6): δ 38.6, 47.5, 61.5, 103.8, 106.2, 109.3, 111.3, 128.4, 150.5,

154.7, 156.0, 157.7, 159.7. ESI-MS (M+): 357.0082 (calc: 357.0080). Found: C, 37.9; H, 3.1; N,

6.0. Calc. for C15H14BrF3N2O7: C, 38.2; H, 3.0; N, 6.0%.

41

2.3.4 Synthesis of aminocoumarin agarose (5).

A solution of agarose (250mg) and carbonyl diimidazole (35 mg, 0.21 mmol) in 100 ml of

DMSO was stirred under nitrogen for 3 hours. Compound 4 (50 mg, 0.14 mmol) was added and

the solution was stirred for an additional 24 hours. The solution was dialyzed (MW cut-off of

3,500) and lyophilized to yield a white solid (yield 190mg, 76%).

2.3.5 Photo-uncaging of aminocoumarin agarose with UV light.

A 100μL solution of aminocoumarin agarose (10mg/ml) was irradiated with a Rayonet UV

reactor (8 RMR-3500 UV tubes with an intensity of ~0.05 mW/cm2) under long wavelength

(365nm) for 30 minutes. After irradiation 50ul of a 1mg/ml solution of CBQCA in DMSO and

50mM sodium cyanide solution in TES pH 8.5 was added to the aminocoumarin agarose solution

and left at room temperature for 30 minutes. Non-irradiated samples were prepared the same as

the irradiated samples except they were not exposed to UV light. The fluorescence was then

measured using a fluorescent plate reader with an excitation and emission wavelength of 465nm

and 560nm respectively.

2.3.6 Two-photon Irradiation of Aminocoumarin agarose hydrogels.

1 weight percent agarose hydrogels were irradiated with a confocal microscope equipped with a

femtosecond Ti-Saphire laser, 20X/0.5NA objective and an electronic stage. To view the

hydrogels using coumarin fluorescence, the laser was set to 740nm with an offset of 75% and a

gain of 0% and a scanning speed of 400 Hz using the Leica confocal software. The focal point of

the laser was positioned within the gel by moving the stage. A region to be patterned was

selected by creating a region of interest using the Leica confocal software. The intensity of the

42

laser was increased by setting the offset to 75% and the gain to 65%, and the region of interest

was scanned. In order to visualize the pattern the intensity of the laser was lowered by setting the

offset to 75% and the gain to 0%. The depth profile was created by taking a picture every micron

for the first 100µm below the surface of the gel. The intensity of coumarin fluorescence was then

measured for each picture in the position of the patterned square and was compared to a non-

patterned region. The change in coumarin fluorescence was then plotted as a function of depth to

give the yield for amine deprotection (figure 2).

2.3.7 Preparation of 1 wt% aminocoumarin agarose hydrogels for amine

visualization with CBQCA.

70 μL of a 10 mg/ml solution of CBQCA in DMSO and 132 μL of a 50 mM solution of sodium

cyanide in water were added to 600 μL of 1.35 wt% solution of 5 in water. 50 μL of this solution

was pipetted into ~70 μL chambers on a glass slide and placed at 4°C for 2 hours for gelation.

The patterns were created as mentioned above. The amine patterns were visualized using an

excitation wavelength of 442nm (HeCd laser) and an emission wavelength of 560nm.

2.4 Results and Discussion

In order to create 3D micropatterned amine-functionalized hydrogels, agarose was first

chemically modified with coumarin-caged amines, dissolved in water and then cast into a mold.

By cooling the ultra-low gelling temperature agarose to 4°C for 2 hours, a hydrogen-bonded

crosslinked gel resulted which was then patterned with a pulsed laser, yielding distinct chemical

volumes of amine groups.

43

Figure 2-1: Synthesis of aminocoumarin

O

HO

HO

O OO

O

OH

O

HO

Agarose

CDI, DMAP

DMSOO

HO

HO

O OO

O

OH

O

OO

N

N

O

HO

HO

O OO

O

OH

O

O

O

O

OBr

OH

O

HN

HN

O

4

5

Figure 2-2: Synthesis of aminocoumarin agarose

2.4.1 Synthesis of aminocoumarin 3

The coumarin caged amine was synthesized for attachment to agarose according to Figure 2-1. 6-

bromo-7-hydroxymethylcoumarin[143] 1 and tert-butyl 2-aminoethylcarbamate[144] 2 were

synthesized according to published literature procedures. Compound 3 was synthesized by

reacting 6-bromo-7-hydroxymethylcoumarin with carbonyl diimidazole followed by the addition

of tert-butyl 2-aminoethylcarbamate in dichloromethane. The product was purified by reverse

phase preparative HPLC. Compound 3 was deprotected in a 1:10 solution of trifluoroacetic acid

and dichloromethane to yield aminocoumarin, compound 4.

44

Figure 2-3: Photo-induced deprotection of agarose amines

2.4.2 Modification of agarose with amine-protected coumarin

Aminocoumarin modified agarose was synthesized, as shown in Figure 2-2, by activating

agarose with carbonyl diimidazole prior to reaction with aminocoumarin 4 (from Figure 2-1).

The modified agarose was purified by dialysis and lyophilized to yield a white solid, 5.

2.4.3 Degree of substitution of aminocoumarin agarose

The amount of aminocoumarin bound to agarose was determined by measuring the absorbance

of the coumarin moiety at 370 nm: 0.100 mol of aminocoumarin was bound per mol of agarose

monomer as calculated relative to a standard curve. To determine whether aminocoumarin was

covalently bound or physically adsorbed to agarose, a control experiment was conducted where 3

and agarose were co-dissolved in DMSO and allowed to react as described for the covalent

modification except in the absence of the carbonyl diimidazole coupling agent. Unbound

aminocoumarin was removed by dialysis prior to measuring the absorbance at 370 nm where it

was determined that 0.0037 mol of aminocoumarin was present per mole of agarose monomer.

45

By subtracting the physically-adsorbed aminocoumarin (0.0037) from the total aminocoumarin

(0.100) measured, we determined that there were 0.0963 moles of aminocoumarin per mole of

agarose monomer covalently immobilized, yielding a degree of substitution of 9.63%.

Figure 2-4: Detection of primary amines using CBQCA after the irradiation of an

aminocoumarin agarose solution. Samples irradiated with a UV lamp at long wavelength

(365 nm) showed significantly greater fluorescence when compared to samples that were

not irradiated. The increase in fluorescence from the irradiated samples confirms the

production of primary amines upon irradiation.

2.4.4 Photo-uncaging of aminocoumarin agarose

To demonstrate that amines are photocaged within the agarose hydrogels, samples of

aminocoumarin agarose were irradiated with UV light and compared to samples that were not

irradiated. The excitation of 6-bromo-7-hydroxycoumarin caged amines results in the cleavage

between the carbon and oxygen producing carbamic acid, which then undergoes decarboxylation

to yield a primary amine (Figure 2-3)[145]. The fluorogenic probe 3-(4-carboxybenzoyl)-2-

46

quinolinecarboxaldehyde (CBQCA) was used to confirm the production of amines after

photoirradiation of aminocoumarin agarose. CBQCA with sodium cyanide forms a fluorescent

complex with primary amines, having excitation and emission wavelengths of 465 nm and 560

nm, respectively[146]. A 100 μL solution of aminocoumarin agarose (10 mg/ml) was irradiated

with a UV lamp under long wavelength (365 nm) for 30 minutes. After irradiation a solution of

CBQCA in DMSO and sodium cyanide was added to the aminocoumarin agarose solution and

left at room temperature for 30 minutes. Figure 2-4 shows that the irradiated samples produce a

stronger fluorescent signal than non-irradiated samples indicating the production of primary

amines after irradiation.

47

48

Figure 2-5: A 50 by 50 µm box was patterned 40 μm below the surface of the gel. The yield

of reaction (percent of coumarin photocleavage and pmol of amines) was determined by

measuring the decrease in coumarin fluorescence within the patterned region. The change

in fluorescence intensity of coumarin was measured over the patterned region through the

first 100 μm. The box was scanned three times with the pulsed Ti-Sapphire laser set to 740

nm. (a) The yield of coumarin deprotection by two-photon irradiation was then calculated

as a function of depth by comparing the change in coumarin fluorescence in the patterned

region to a non-patterned region. (b) The amount of amines in piocomoles as a function of

depth within the patterned region. Confocal micrographs of coumarin fluorescence are

shown at : (c) 20 μm, (d) 30μm, (e) 40μm, (f) 50μm and (g) 60μm below the surface.

2.4.5 Two-photon Irradiation of aminocoumarin agarose hydrogels

To demonstrate three dimensional patterning, a 50 μL 1 wt% agarose hydrogel of 5 was

irradiated using a femtosecond Ti-Sapphire pulsed laser on a Leica TCS SP2 confocal

microscope equipped with a 20X/0.5NA objective. An excitation wavelength of 740 nm for two

photon activation was selected since the one photon maximum absorbance of aminocoumarin

occurs at 370 nm; two-photon excitation requires each photon to be half the energy or twice the

wavelength as those for one photon irradiation. The focal point of the pulsed laser was directed

40 μm below the surface of the gel and a region of interest (ROI) of 50 m x 50 m was

selected. The laser will only irradiate in the ROI. The ROI was then scanned 3 times with the

pulsed Ti-Sapphire laser with an offset of 75% and a gain of 65%.

The gel was then viewed using coumarin fluorescence with the Ti-Sapphire laser set to low

intensity with an offset of 75% and a gain of 0%, thereby ensuring little further aminocoumarin

deprotection at the low intensity laser setting. Figure 2-5e shows a dark box 50 μm by 50 μm at

a depth of 40 µm. The dark region indicates the lack of coumarin fluorescence and therefore the

deprotection of aminocoumarin agarose.

49

To demonstrate three-dimensional patterning, the percent yield for coumarin photocleavage was

calculated as a function of depth by measuring the change in coumarin fluorescence within the

hydrogel using the confocal microscope (Figure 2-5). The patterned region, a 50 μm by 50 μm

box, was selected and the change in coumarin fluorescence was measured every micron from the

top of the gel to 100 μm below the surface of the gel. Figure 2-5a shows that the cleavage of

coumarin begins around 20 μm, reaches a maximum at 40 μm and decreases back to close to

zero at 60 μm. Coumarin cleavage occurs at wherever two-photon absorption is possible. The

maximum of coumarin cleavage occurs at 40 µm since the laser was focused at that depth for

two-photon irradiation; therefore the highest concentration of photons is at a depth of 40 µm. As

you move away from the centre of the focal point the probability of two-photon absorption

decreases thus lowering the amount of coumarin cleaved. Confocal micrographs at depths of 20,

30, 40, 50 and 60 μm are shown in Figure 2-5. No box is visible at depths of 20 and 60 μm

because less than 5% deprotection occurred. A box is clearly visible at the point of maximum

coumarin cleavage, 33%, at 40 μm below the surface of the gel. At 30 and 50 µm below the

surface of the gel, the box is still visible where 17% deprotection occurred. Therefore a box was

patterned with approximate dimensions of 50 x 50 x 40 μm within the aminocoumarin agarose

hydrogel, the depth of the box was determined from figure 2-5a where coumarin cleavage occurs

between 20 µm to 60 µm. 40 µm represents the minimum size of the pattern in the z-direction

(depth); however, boxes with a larger z-dimension (> 40µm) can be created by irradiating the gel

at multiple depths.

The molar amount of free amines in the patterned region was calculated using the coumarin

cleavage yield and the substitution rate of aminocoumarin on agarose calculated in section 2.3.

Figure 2-5b demonstrates that the amount of free amines in the box varies in the picomole range

for a given depth.

50

Figure 2-6: Confocal image of patterned aminocoumarin agarose hydrogel visualized using

the fluorescence of coumarin. A series of boxes was patterned into the hydrogel using two-

photon excitation. The first box on the left was scanned 7 times, the second 9 times, the

third 11 times and the fourth 13 times. As the number of scans increased, the fluorescence

observed decreased due to greater photocleavage of coumarin. The fine lines located

between the boxes are due to the laser scanning on the confocal microscope. The

microscope scanned the region bordered by the fine lines but only irradiated in the region

of the boxes by modulating the laser intensity; however, the intensity of the laser outside of

the boxes is still sufficient to produce the fine lines observed. (The image was enhanced for

clarity only using in photoshop.)

Figure 2-7: Confocal image showing the presence of amines within the patterned regions,

the boxes correspond to those in Figure 2-6. The amine reactive fluorescent probe CBQCA

was used to detect the uncaged amines. The bright boxes represent the fluorescence from

the CBQCA amine complex. The box on the left was irradiated with 7 scans and each

subsequent box (to the right) was scanned two more times. (The image was enhanced for

clarity only in photoshop.)

51

A second pattern of a series of boxes was patterned with 25 μm between each box. The amount

of irradiation per box was controlled by the number of scans per box, which can be used to

control the concentration of amine functionality. The first box, on the left in Figure 2-6 and 2-7,

was scanned 7 times with two scans added to each subsequent box.

After the irradiation was complete, the gel was imaged for the fluorescence of the coumarin

moiety at 450 nm on the confocal microscope. A decrease in fluorescence was observed in the

areas that were irradiated, resulting in a pattern of dark boxes (Figure 2-6). The dimensions of

the individual boxes were determined to be ~75 μm wide, ~75 μm long and ~40 μm high, which

corresponded to the volume of the gel that was irradiated with the pulsed laser. The decrease in

fluorescence results from the coumarin moiety being either photobleached or cleaved from the

agarose to produce free amine.

To confirm the presence of amines within the box patterns, CBQCA with sodium cyanide was

added to the gel. To enhance the reaction of agarose amine groups with CBQCA, triethylamine

was also added to increase the pH of the gels and thus the reactivity of the primary amines. The

fluorescence of CBQCA within the gel was then visualized using a HeCd laser at 442 nm (Figure

2-7) and confirmed the presence of amines in the patterned volumes. The low CBQCA

fluorescence intensity reflected the picomolar amount of uncaged amine groups present (Figure

2-5). While the concentration of amine groups is low, it is sufficiently high for biomaterial

applications where femtomolar concentrations of peptides have been shown to promote a cellular

response.[147] Notwithstanding the weak fluorescent signal from CBQCA, the irradiation of

aminocoumarin hydrogels with a pulsed laser resulted in the selective deprotection and

micropatterning of defined volumes of amine groups in agarose. The reaction of these amine

52

groups with CBQCA demonstrates the capacity of these amine groups for further modification

and is useful for imaging.

2.5 Conclusion

A coumarin caged amine was synthesized and immobilized onto agarose gels, which upon two-

photon excitation, resulted in cleavage of the coumarin moiety yielding primary amines. Using a

pulsed laser, spatially defined volumes of micropatterned amine cubes were patterned into

hydrogels of the modified agarose. Using fluorescent CBQCA, we proved the success of the

uncaging chemistry while demonstrating the capacity of these amine groups for further

modification. This first demonstration of 3D micropatterned volumes of amine functional

groups within transparent polymeric hydrogels is currently being explored for cell guidance in

the context of tissue engineering and regenerative medicine.

53

3 Three-dimensional spatial patterning of proteins in hydrogels*

*This chapter was published in Biomacromolecules. Wylie, R. G.; Shoichet, M. S., Three-Dimensional Spatial Patterning of Proteins in Hydrogels. Biomacromolecules 2011. DOI: 10.1021/bm201037j

3.1 Abstract

The ability to create 3D matrices approaching the chemical complexity of the extracellular

matrix is crucial for both the elucidation of fundamental biological phenomena and tissue

engineered biological constructs. To this end, we designed a system where proteins can be

photochemically patterned in three-dimensions within hydrogels under physiological conditions.

Fibroblast growth factor-2 (FGF-2) was immobilized within agarose hydrogels that were

modified with two-photon labile 6-bromo-7-hydroxycoumarin-protected thiols. Two different

methods were developed for FGF-2 immobilization. The first procedure relies on the protein

containing free cysteines for the formation of disulfide bonds with photo-exposed agarose-thiols.

The second procedure takes advantage of the femtomolar binding partners (KD ~ 10-14 M),

human serum albumin (HSA) and albumin binding domain (ABD). Here HSA-maleimide was

chemically bound to photo-exposed agarose thiols and then the FGF2-ABD fusion protein was

added to form a stable complex with the immobilized HSA. The use of orthogonal, physical

binding pairs allows protein immobilization under mild conditions, and is broadly applicable to

any protein expressed as an ABD fusion.

54

3.2 Introduction

The development of techniques to synthesize matrices mimicking the three-dimensional (3D)

signaling environment in vivo is crucial to accurately study cellular activities [12]. Although

many important findings have been and will continue to be discovered using two-dimensional

(2D) cultures systems, many cellular functions are influenced by the spatial arrangement of the

3D microenvironment – such as those provided by other cells, the extracellular matrix proteins,

presentation of growth factors, mechanical properties, among others[1] – that cannot be

adequately represented in a 2D system. Hydrogels have proven to be useful substrates to study

biochemical cues that depend on the 3D environment[12, 14, 148]; however, current hydrogels

lack the chemical complexity required to mimic in vivo environments. To address these

limitations, researchers have been developing 3D patterning systems for biomolecules. The

majority of work has focused on the immobilization of adhesive peptides within hydrogels to

influence cell migration and morphology[2, 26]. Current research is focused on the development

of methods for protein immobilization for more complex studies, such as vascular endothelial

growth factor (VEGF) gradients to encourage vascularization[149]. The creation of protein

patterns in complex and versatile systems needs to be developed for 3D protein immobilization.

Patterning technologies will be useful tools for 3D cell culture to better understand cellular

behavior in vivo.

Hydrogels are commonly used as 3D scaffolds since they can be designed to mimic the

extracellular matrix (ECM) both chemically and physically[12]. Natural hydrogels made from

components of the ECM such as collagen, elastin, and fibrin are frequently used because they are

intrinsically biocompatible. Furthermore, these hydrogels contain bioactive elements such as

adhesion sites, which can promote cell survival and proliferation[12, 150]. However, hydrogels

55

with natural bioactive sites are not optimal for patterning experiments because they contain

intrinsic biological signals. For example, specific adhesion sites cannot be patterned into

collagen hydrogels since it is highly cell adhesive in nature. The optimal hydrogel would be a

blank canvas where biochemical signals can be incorporated in 3D through biomolecule

immobilization. Natural hydrogels, from non-mammalian sources such as alginate and agarose,

and synthetic hydrogels, such as polyethylene glycol (PEG) and poly(N-isopropyl acrylamide)

(poly-NIPAAM), are biochemically inert, and thus provide an ideal blank canvas for biochemical

patterning.

Proteoglycan-binding proteins, such as fibroblast growth factor-2 (FGF-2), are particularly

amenable to immobilization strategies since they are presented by the ECM [151]. Proteoglycans

are heavily glycosylated proteins that form part of the insoluble matrix of the ECM and contain

binding sites for many growth factors and cytokines. For example, heparin sulfate proteoglycans

have been shown to sequester growth factors both for the establishment of reservoirs and the

creation of spatial gradients[15, 152]. Interestingly, many proteins are presented to cells from

the ECM as solid-phase ligands, thereby providing a rationale for our protein immobilization

strategy[15]. FGF-2 was chosen as a model protein for immobilization since it naturally exists as

an immobilized signaling molecule[15], and is implicated in many cellular processes including

differentiation[153], regeneration[154], wound repair[155] and vascularization[156, 157].

Immobilized FGF-2 has also been shown to promote several cellular activities including

proliferation[158], migration[54], and cell organization[151]. Because of this breadth of activity,

3D patterns of FGF-2 could have wide applicability in the elucidation of biological pathways and

phenomena.

56

A number of photochemical methods have been utilized to achieve 3D patterns of proteins in

hydrogels using either photoinitiators or photocages. For example, the first 3D patterned peptide

hydrogel was designed with maleimide-modified peptides that readily reacted with deprotected

photo-caged thiols in distinct volumes[149]. Thiol-containing biomolecules, such as RGDSC,

were patterned into alkene containing hydrogels using thiol-ene chemistry. In the presence of a

photoactive hydrogen abstracting initiator, thiols are deprotonated to thiyl radicals for reaction

with alkenes in the gel[26, 159, 160]. Acrylate-modified peptides and proteins have also been

patterned into hydrogels containing acrylates groups using two-photon active initiators[110, 161,

162]. 3D patterns of proteins have already proven useful in tissue engineering. For example, 3D

immobilized VEGF encourages tubulogenesis of endothelial cells for the creation of

vasculature[149, 163]. Moreover, immobilized EphrinA1 also encourages microvascularization,

further solidifying the rationale for immobilized proteins in tissue engineering[164].

We designed a system that achieves 3D immobilization of proteins without the need of chemical

crosslinker within transparent hydrogels, which could prove advantageous in the preservation of

bioactivity. This system could then be used to study cellular processes or as a tool in tissue

engineering to guide cell fate in 3D. To this end, FGF-2 was immobilized within agarose

hydrogels using two-photon chemistry, which provides the necessary 3D control since the

excitation and thus reaction volume is limited to the focal point of the laser[81]. Agarose was

modified with 6-bromo-7-hydroxycoumarin (Bhc) protected thiols, which are deprotected upon

excitation to yield reactive thiols and were subsequently used for the immobilization of proteins

(Figure 3-1). Bhc was chosen as the photocage over more common cages such as 2-nitrobenzyl

since it has a larger two-photon uncaging cross-section[85], thus increases the yield of the photo-

uncaging. Bhc has previously been used in cell culture experiments without any measurable

toxicity[85, 149]. FGF-2 was immobilized through either disulfide bonds or the strong physical

57

interaction between human serum albumin (HSA) and the albumin binding domain (ABD)

(Figure 3-1). Immobilization of proteins using disulfide bonds has previously been established as

an efficient method[59]. Strong physical interactions have also been shown to be stable and

useful for protein immobilization[77].

3.3 Materials and Methods

3.3.1 Materials

BL21 (DE3) E. coli was purchased from New England Biolabs (Ipswich, MA). Isopropyl β-D-1-

thiogalactopyranoside, ampicillin and terrific broth were purchased from Bioshop Canada Inc.

(Burlington, ON). Mouse fibroblast growth factor-2, human serum albumin (fatty acid free) and

anti-foam 204 were purchased from Sigma-Aldrich (Oakville, ON). Ni-NTA agarose was

purchased from Qiagen (Valencia, CA). Sulfosuccinimidyl-4-(N-maleimidomethyl)

cyclohexane-1-carboxylate was purchased from Thermo Scientific (Waltman, MA). Alexa

Fluor® 546 C5 maleimide was purchased from Invitrogen (Carlsbad, CA). The plasmid coding

for FGF2-ABD in pET-21a(+) was purchased from Genscript (Piscataway, NJ).

3.3.2 Preparation of agarose-thiol-Bhc gels

Coumarin-sulfide agarose was prepared as previously reported with minor modifications[113].

400 mg of agarose in 40 ml of DMSO was mixed with 180 mg of carbonyl diimidazole for 2 h,

followed by the addition of 100 mg of coumarin sulfide with 5 drops of triethylamine. After 24

h the product was purified by dialysis against water, yielding a substitution rate of 2.8% (2.8% of

the agarose repeat units were modified with coumarin sulfide). Gels were formed in chambers

58

comprise of o-rings (outer diameter: 0.7 mm, inner diameter: 0.5 mm, and height: 0.2 mm)

attached to coverslips.

3.3.3 Expression and purification of FGF2-ABD

A pET-21a(+) plasmid (Novagen) coding for a fusion protein of mouse FGF-2 and a modified

ABD[165] with a 6X histidine tag for Ni-NTA purification was transformed into chemically

competent BL21 (DE3) E. coli. Protein expression was conducted in a 1.8L culture consisting of

85.7 g of terrific broth, 14.4 mL of glycerol, 6 drops of anti-foam 204 and 180 mg of ampicillin.

E. coli was grown at 37 °C with air sparging until an OD600 of 0.8, followed by the addition of

342 mg of isopropyl β-D-1-thiogalactopyranoside (IPTG) while lowering the temperature to 16

°C. After 20 h, the cells were pelleted by centrifugation at 12,227 G for 10 min (Beckman

coulter centrifuge Avanti J-26 with rotor JLA-8.1000), resuspended in 60 mL of buffer (50 mM

Tris pH 7.5, 500 mM NaCl, 5 mM imidazole) and sonicated for 10 min at 30% amplitude with a

pulse of 2 s (Misonix S-4000 Sonicator Ultrasonic Processor equipped with a Dual Horn probe).

The slurry was centrifuged at 45,000 G for 15 min at 4 °C (Beckman coulter centrifuge Avanti J-

26 with rotor JA-25.50). The liquid fraction was incubated with 2 mL of nickel-nitrilotriacetic

acid (Ni-NTA) resin solution for 15 min at 4 °C. The resin was collected in a column with a

glass frit and washed 10 x 10 mL with 50 mM Tris (pH 7.5, 500 mM NaCl, 30 mM imidazole)

and eluted with 50 mM Tris (pH 7.5, 500 mM NaCl, 250 mM imidazole). The protein solution

was then loaded onto a heparin column (HiTrap Heparin HP 1mL, GE healthcare) and further

purified over a NaCl gradient from 0 mM to 2 M in phosphate buffer pH 7.3. 2 mg of pure

protein was obtained after size-exclusion chromatography (SEC) in 10 mM phosphate buffer (pH

7.3, 250mM NaCl) using fast protein liquid chromatography (FPLC, Superdex 75 HR 10/30,

59

AKTA Explorer 10, Amersham Pharmacia). Protein concentrations were determined by

absorbance at 280 nm using an extinction coefficient of 18490 M-1cm-1 and a MW of 27.336 da.

3.3.4 Labeling of FGF2-ABD with Alexa 546

A solution consisting of 500 µL of 1 mg/ml of FGF2-ABD in 10 mM phosphate buffer (pH 7.3,

250mM NaCl) and 10 µL of 10 mg/ml maleimide Alexa Fluor 546 in DMSO was mixed for 2 h

at room temperature. The protein was purified using 200 µL of Ni-NTA resin, washed with 10 x

1 mL of 50 mM Tris (pH 7.5, 500 mM NaCl, 30 mM imidazole) and eluted in 250 µL of 50 mM

Tris (pH 7.5, 500 mM NaCl, 250 mM imidazole) yielding 220 µg of FGF2-ABD-546. The

substitution rate was determined to be 0.61 mol of Alexa 546 per mol of protein calculated

according to Invitrogen[166]. Briefly, the absorbance at 546 nm of the protein solution was

measured to determine the concentration of Alexa Fluor 546. Protein concentration was then

determined using the absorbance at 280 nm, the absorbance contribution of Alexa Fluor 546 at

280 nm was removed. The concentration of Alexa Fluor 546 was divided by the protein

concentration to determine the substitution ratio.

3.3.5 Addition of maleimide to HSA

A solution consisting of 500 µL of a 5 mg/ml solution of HSA in PBS and 100 µL of 35 mg/ml

sulfo-SMCC in DMSO was mixed for 2 h. The protein was purified by SEC (FPLC, Superdex

200 prep grade HiLoad 16/60, AKTA Explorer 10, Amersham Pharmacia) with PBS (pH 6.8) as

the running buffer yielding 2.2 mg of maleimide (mal) HSA. The protein solution was

concentrated to 4 mg/ml by centrifuge concentration (Vivaspin 20 10kDa, GE Healthcare,

Piscataway, NJ). Solutions were stored at -80°C until further use.

60

3.3.6 Bioactivity of recombinant FGF2-ABD

Mouse neural stem progenitor cells (NSPCs) were isolated from the subventricular zone[167];

5,000 cells (passage 2) were plated per well in a 48 well plate in serum free media (DMEM/F12

with 20 ng/ml of EGF and 2 µg/ml of heparin) with varying concentrations (0 to 1.2 nM) of

either FGF2-ABD or commercial FGF-2. The numbers of neurospheres greater than 100 µm in

diameter were counted for each condition after 7 d of culture as a measure of bioactivity.

3.3.7 Photo-patterning and Imaging

All patterns were created and imaged on a Leica TCS-SP2 confocal microscope equipped with a

Green HeNe laser (1.2mW; 543), multi-photon Mai Tai laser, a 20x objective (NA = 0.4) and an

electronic stage. The multi-photon laser was set to 740 nm with an offset of 75% and gain of 0%

for visualization and an offset of 75% and gain of 43% for patterning. The uncaging of thiols

can be immediately visualized by the loss of fluorescence from Bhc. Patterns of FGF2-ABD

were visualized with the following settings: laser 543 at 100%, wavelengths 560 to 700 nm

collected, photomultiplier tube at 845 V and 6 scans on average per image. Leica software

version 2.5.1227a was used for the visualization and fluorescence quantification.

3.3.8 Patterning FGF2-ABD-SH to Agarose-SH through disulfide bonds

Gels of 1 wt% agarose-thiol-Bhc in PBS (pH 6.8; 20 µL) were patterned by selecting a region of

interest of 100 x 100 µm forming a square 500 µm below the gel. A series of squares was

patterned with varying number of lasers scans from 5 to 50 scans. After washing the gels in PBS

(pH 7.4) for 24 h, 100 µL of 0.15 mg/ml of FGF2-ABD-546 was placed on top of the gel for 16

h at 4 °C. The gels were the washed for 2 d in 200 mL of PBS (pH 7.4) with daily buffer

61

replacement. Z-stacks spanning 100 µm in depth were constructed to determine the axial

fluorescent profile.

3.3.9 Patterning FGF2-ABD to Agarose-HSA

20 µL of 1 wt% agarose-thiol-Bhc with 2 mg/ml of mal-HSA in PBS (pH 6.8) were irradiated in

the same manner as above. After removing excess mal-HSA by soaking the gels in PBS (pH

7.4) with 5 mM β-mercaptoethanol for 1d, FGF2-ABD-546 was introduced as above. Therefore

immobilization occurred in PBS buffer at pH 7.4 with 1 mM β-mercaptoethanol, thereby

ensuring only agarose-HSA would react with FGF2-ABD.

3.3.10 Testing the stability of FGF2-ABD pattern with HSA

The fluorescence intensity of 100 x 100 µm squares patterned with 50 scans of 2-photon

exposure was followed over 8 d in PBS with and without soluble HSA. Gels were soaked either

in 30 mL of PBS or 30 mL of PBS with 10 mg/ml of HSA. The fluorescence intensity of the

patterns was measured on days 0, 2, 5 and 8. Changes in fluorescence were compared by

normalizing to day 0.

3.3.11 Quantification of FGF2-ABD

To convert the fluorescence intensity into protein concentration, a calibration curve was

constructed for Alexa Fluor 546. 1 wt% agarose-thiol-Bhc hydrogel with concentrations of 0,

5.75, 11.5, 23, 29, 45, 70 and 110 nM were imaged at 500 µm below the surface of the gel. The

calibration curve along with the known number of 546 tags per protein was used to calculate the

protein concentration for each square.

62

3.3.12 Statistical analysis

All statistics were performed using the software GraphPad Prism 5 (La Jolla, CA, USA).

Differences among groups were either assessed by T-test or ANOVA with Tukey's post hoc

analysis. All data is presented as mean ± SD.

Figure 3-1: Scheme for the 3D immobilization of FGF2-ABD to agarose through either

disulfide bonds or HSA/ABD physical interaction. (a) Schematic diagram demonstrating

the 3D photo-deprotection of thiols in agarose-thiol-coumarin for the coupling of FGF-2.

(b) FGF2-ABD was immobilized to agarose-thiol through disulfides bonds. Thiols are

deprotected by two-photon excitation of coumarin (740 nm), which subsequently form

disulfide bonds with free cysteines on FGF2-ABD. (c) FGF2-ABD was immobilized using

the physical binding pair of HSA/ABD. Maleimide-HSA was immobilized through two-

photon irradiation of agarose as in (b), followed by the addition of FGF2-ABD, which

selectively binds with immobilized HSA.

63

3.4 Results

3.4.1 Synthesis and characterization of FGF2-ABD and mal-HSA

In this study, FGF2-ABD was expressed from E. coli transformed with a pET-21a(+) plasmid

coding for the protein sequence (Figure 3-2a). To increase the percentage of protein in the

soluble fraction, the expression was performed at 16 °C. At 37 °C, the majority of the protein

remains in the insoluble fraction. After expression and collection of the soluble fraction, the

protein was purified using a combination of 3 columns: 1) Ni-NTA, 2) heparin, and 3) a size-

exclusion chromatography (SEC) column. The purity of FGF2-ABD was confirmed by sodium

dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE, Figure 3-2b).

Recombinantly expressed FGF2-ABD was determined to be bioactive by comparison to

commercially available FGF2 using the neurosphere assay (Figure 3-2c). Here, bioactive FGF2

is required for neurosphere formation from NSPCs seeded as single cells[167]. As FGF2

concentration increased, so did the number of neurospheres. There was no significant difference

in number of neurospheres between our expressed FGF2-ABD and commercial FGF2 at any

concentration (one-way ANOVA Tukey’s post-test, p < 0.05).

Mal-HSA was synthesized by reacting HSA with sulfo-SMCC[70] and purified by SEC. We

observed no evidence of protein crosslinking; mal-HSA eluted at the same time as HSA by SEC.

The degree of modification was determined by MALDI-TOF; unmodified HSA produced a

primary peak of 66,524 g/mol whereas modified HSA produced a broad peak with an average

molar mass of 71,071 g/mol. The addition of a maleimide group would increase the mass by 219

g/mol, it was calculated that an average of 20.8 maleimides were added per HSA molecule.

64

Since HSA contains 59 lysines[168], and assuming only lysine primary amines reacted, we

estimate that approximately one in three lysines were modified.

Figure 3-2: Expression, purification and bioactivity of FGF2-ABD. (a) Protein sequence of

the expressed FGF2-ABD with the FGF-2 at the N-terminus (red) and the ABD at the C-

terminus orange) with a spacer (black) in between the two sequences to minimize

interdomain interactions. (b) SDS-PAGE protein electrophoresis of purified FGF2-ABD

shows that a pure sample (indicated with an arrow) with the proper MW (27,336 g/mol)

was expressed. (c) FGF2-ABD was determined to be bioactive by counting the number of

neurospheres formed from NSPCs after 7 d of culture. NSPCs were cultured as single cells

in a 48 well plate in the presence of varying concentrations of FGF2-ABD or commercial

FGF-2. Bioactivity of FGF2-ABD was similar to the commercial FGF-2 (mean±standard

deviation shown, n=3 for each condition, one-way ANOVA Tukey’s post-test, p < 0.05).

65

3.4.2 Immobilization of FGF2 using disulfide bonds

The 3D localization of FGF2-ABD-546 was successfully achieved by introducing the protein

within a two-photon patterned agarose-thiol hydrogel. While the ABD was not necessary for the

formation of disulfide bonds, keeping the FGF2-ABD constant across both patterning

methodologies allowed us to compare groups. A pattern of squares with different fluorescent

intensities was created 500 µm below the surface of the gel (Figure 3-3a) by raster scanning

across a given volume a set number of times. As observed, the fluorescence intensity increases

with number of scans because there are increasing numbers of deprotected thiols available to

react with FGF2-thiols. A range of 8.5±2.9 to 58.7±12.9 nM was immobilized by increasing the

number of laser scans from 5 to 50 (Figure 3-3b). The axial (z-axis) fluorescent profile of boxes

scanned 30, 40 and 50 times was quantified and yielded a Gaussian distribution over a 40 µm

(Figure 3-3c). To demonstrate the requirement of thiols for immobilization, we attempted to

pattern of streptavidin which does not contain accessible cysteines. No pattern resulted indicating

the necessity of free thiols on the protein for successful immobilization.

66

Figure 3-3: 3D immobilization of FGF2-ABD-546 through disulfide bonds. (a) Confocal

micrograph of a series of squares with varying concentrations of FGF2-ABD-546. 10

squares were patterned 500 µm below the surface of the hydrogel with 5 to 50 laser scans

(scale bar: 100 µm). (b) The concentration of FGF2-ABD was quantified by converting the

fluorescence intensity of each square using a calibration curve. A range of 8.5±2.9 to

58.7±12.9 nM of FGF2-ABD was immobilized (mean±standard deviation shown, n=3 for

each condition). (c) The fluorescence z-axis profile of the squares for 30 and 50 scans was

measured to determine the axial resolution. A resolution of approximately 40 µm was

achieved for each square (mean±standard deviation shown, n=3 for each condition).

67

Figure 3-4: 3D immobilization of FGF2-ABD-546 using the physical binding interaction of

HSA and ABD. (a) Scheme for the immobilization of FGF2-ABD by first immobilization

mal-HSA to agarose-thiol using two-photon irradiation. (b) Confocal micrograph of

immobilized FGF2-ABD-546 in a series of 10 squares scanned 5 to 50 times. Fluorescence

increased as a function of scan number (scale bar: 100µm) . (c) The concentration of FGF2-

ABD was quantified by converting the fluorescence intensity of each square using a

calibration curve. A range of 77.9±15.1 to 189.1±20.6 nM of FGF2-ABD was immobilized

(mean±standard deviation shown, n=3 for each condition). (d) The fluorescence z-axis

profile was determined for squares scanned 30 and 50 times to determine the axial

resolution. A resolution of approximately 40 µm was achieved for each square

(mean±standard deviation shown, n=3 for each condition).

68

3.4.3 Immobilization of FGF2 using HSA/ABD

The second immobilization strategy of FGF-2 took advantage of physical binding partners

designed for proteins without disulfide bonds. FGF2-ABD was immobilized by first photo-

patterning mal-HSA within distinct volumes of Bhc-photo-cleaved agarose-thiol (Figure 3-4a).

The gel was washed in PBS (pH 7.4) with 5 mM β-mercaptoethanol to quench any unreacted

maleimides. FGF2-ABD-546 was then added, followed by a washing step before visualization.

Using this method, a series of boxes was created, with the fluorescent intensity increasing with

the number of raster scans (Figure 3-4b). By comparing fluorescence to a calibration curve, we

calculated that concentrations ranging from 77.9±15.1 and 189.1±20.6 nM of FGF2 were

immobilized by increasing the number of laser scans from 5 to 50 (Figure 3-4c). As was

described for the disulfide modification strategy, the fluorescence profile in the z-axis of the

squares for 30, 40 and 50 scans showed an axial resolution of approximately 40 µm for the

FGF2-ABD/HSA-agarose strategy (Figure 3-4d).

3.4.4 Stability of HSA/FGF2-ABD complex

Given the prevalence of albumin in cell culture media and the fact that HSA/ABD is a non-

covalent bond, the stability of the immobilized FGF2-ABD complex with agarose-HSA was

investigated in both the presence and absence of soluble HSA. The fluorescence intensity of the

samples that had boxes scanned 50 times were followed over time for a total of 8 d (Figure 3-5).

With the fluorescence at day 0 set to 100%, the depletion of fluorescent intensity was followed in

the presence of 10 mg/ml HSA dissolved in PBS or simply PBS alone. To ensure stability of the

pattern under cell culture conditions, the concentration of soluble HSA (10 mg/ml) used was

greater than that commonly found in cultures containing serum (media with 10% serum contains

69

~ 2 mg/ml of albumin)[169]. Over the 8 d period, the fluorescence intensity decreased to

80.6±16.1% when immersed in PBS and to 77.5±6.6% when immersed in HSA. No significant

difference between samples in PBS or 10 mg/ml HSA in PBS at any timepoint was observed

(mean±standard deviation shown, n=3 for each condition, unpaired t-test, p < 0.05), leading us to

conclude that the FGF2-ABD/agarose-HSA interaction is not influenced by the presence of

soluble HSA. The decrease in fluorescence over time for both conditions is not statistically

significant indicating that the complex is stable over the time period tested (p < 0.05).

0 2 4 6 8 100

20

40

60

80

100

120PBSPBS with 10mg/ml HSA

Days

Rel

ativ

e F

luo

rese

nce

to

day

0

Figure 3-5: Immobilized FGF2-ABD-546 complexed with HSA is stable in PBS in both the

presence and absence of soluble HSA. The fluorescence intensity of the sample having

squares scanned 50 times was immersed in 30 mL of () PBS or () PBS with 10 mg/ml

HSA was followed over time. No significant difference was observed between the

conditions (PBS versus PBS with 10 mg/ml) at any time point, indicating the complex is

stable in the presence of soluble HSA (mean±standard deviation shown, n=3 for each

condition, unpaired t test, p < 0.05). The complex was also determined to be stable over

time since no significant difference in fluorescence was observed between any timepoints

for the same condition (ANOVA with Tukey's post hoc analysis, p < 0.05).

70

3.5 Discussion

The patterning of proteins within two-photon active hydrogels offers a platform for the

construction of 3D biomimetic environments for cell culture. In this study, we developed two

methods for the immobilization of FGF2 with agarose-thiol-Bhc: 1) through the formation of

disulfide bonds and 2) using the physical interaction between HSA and ABD (Figure 3-1). For

the first strategy, we took advantage of the two free cysteines of FGF-2, which are not in the

active site and are available to form disulfide bonds with photo-exposed agarose-thiol

groups[170, 171]. For the second strategy, we took advantage of the orthogonal binding between

HSA and ABD by first immobilizing maleimide-HSA to photo-deprotected thiols by Michael-

type addition and then introducing the FGF2-ABD as a fusion protein. Both systems allowed

FGF-2 to be immobilized under mild conditions to maintain bioactivity, and can be used as a

model for the 3D immobilization of numerous proteins.

Proteins with free cysteines, such as FGF-2, can be directly photo-patterned in hydrogels with

available thiol reactive groups, as was demonstrated herein with agarose thiol-Bhc hydrogels in a

single step. The concentration of agarose-thiols available to react for protein conjugation is

independent of the protein being conjugated and dependent only on the photopatterning. The 3D

localization of FGF2-ABD-546, labeled for visualization with Alexa 546, was successfully

achieved by simply introducing the protein within a two-photon patterned agarose-thiol

hydrogel. As observed, the immobilized concentration can be tailored for the introduction of

gradients, which are useful for cell migration[54] (Figure 3-3b). 8.5±2.9 nM represents the

lowest detectable concentration of FGF2-ABD-546 as visualized with the confocal microscope.

Lower concentrations could be quantified using an instrument with greater sensitivity. Since

each square was scanned on only one plane, the z-axis fluorescent profile represents the axial

71

resolution. In this case, a Gaussian distribution was observed spanning approximately 40 µm.

Moreover, immobilization using disulfide bonds is also applicable to proteins without free

cysteines by either converting amines into thiols through reaction with Traut’s reagent[172] or

by the recombinant incorporation of cysteines into the peptide sequence. Therefore, this system

is as a versatile model for 3D protein immobilization.

Proteins can also be engineered with peptide binding domains, such ABD, for immobilization.

The equilibrium dissociation constant (KD) for the wild-type sequence of ABD for HSA is only

1.2 nM[165], similar to reversible interactions used for protein purification such as the FLAG tag

with an anti-FLAG antibody[173]. Nygren et al. have engineered a number of ABDs with

varying affinities for HSA reaching femtomolar affinity[165], which is necessary to form stable

complexes for immobilization experiments. Therefore, we incorporated the ABD sequence with

the strongest affinity (or a very low dissociation constant, KD ~ 10-14 M) at the C-terminus of

FGF-2. This interaction has a similar binding affinity as biotin-streptavidin, which has been

successfully used in protein immobilization studies[75]. A spacer of 28 residues was

incorporated between the FGF2-terminus and ABD to minimize interference between the two

domains during expression and binding events (Figure 3-2a). This methodology is advantageous

over disulfide bond immobilization for proteins that require cysteines for activity or cannot be

modified to contain thiols.

FGF-2 was immobilized with 3D control using the HSA and ABD physical interaction. After

first immobilizing mal-HSA, FGF2-ABD-546 was introduced and immobilized within the

irradiated volumes. As shown for disulfide bond immobilization, the amount immobilized was

dependent on the number of laser scans. Although in this case, a higher concentration of protein

was immobilized for the same number of scans when compared to FGF2 immobilization using

72

disulfide bonds. The increase is most likely a result of the immobilization procedure. For the

disulfide bond method, the gels were patterned first followed by the introduction of FGF2.

Therefore, some photo-deprotected thiols in agarose may have reacted with other functional

groups, such as forming disulfide bonds between agarose thiols in the patterned region. For the

HSA/ABD method, mal-HSA was present at the time of patterning, which limited the formation

of disulfide crosslinks within the gel and increased the amount of free thiols for reaction. . For

both systems, the concentration of protein immobilized was linear with irradiation indicating that

only a fraction of thiols were used. This was expected since the gels have a thiol-Bhc

concentration of 820 µM, which is much higher than the nanomolar range of protein

immobilized. Importantly, the axial resolution (~ 40 µm) remained the same between the two

systems, which indicates that the resolution is solely dependent on the volume of excitation and

not the immobilization strategy.

Immobilization of proteins using the physical binding pair HSA and ABD produces a stable

complex. The presence of soluble HSA did not influence the FGF2-ABD pattern, indicating this

method can be used in conditions containing albumin such as in vitro cell culture. If the FGF2-

ABD complex with HSA was disassembling, the amount of immobilized protein would decrease

faster in the samples stored in HSA solutions than those stored in PBS. Soluble HSA would

prevent reattachment of FGF2-ABD to agarose-immobilized HSA since it is at a much higher

concentration. The similar changes in FGF2-ABD observed between samples stored in HAS and

PBS indicates a very slow disassociation of the complex. This result is consistent with the

reported dissociation rate constant of HSA/ABD (kd: 1.5 x 10-6 s-1)[165] which is similar to that

of biotin/streptavidin (kd: 6.8 x 10-5 s-1)[174]. The femtomolar dissociation constants and very

low dissociation rate constants are commonly referred to as a “quasi-covalent interaction”[175].

73

The use of disulfide bonds or physical binding pairs with two-photon irradiation allows for 3D

protein immobilization under mild conditions. Traditional 3D photo-patterning techniques have

relied on the use of chemical crosslinkers or photoiniators, which can negatively influence

bioactivity or cell viability. The methods described above, allow for the immobilization of

proteins in buffers without the use of potentially cytotoxic organic molecules. The use of binding

domains also directs the site of immobilization on the protein molecule, which in turn can control

protein orientation for optimal bioactivity. In this case, ABD was placed at the C-terminus

leaving the N-terminus of the protein free for receptor binding. For another protein, where the

binding site is at the C-terminus, ABD could have been placed at the N-terminus, thereby leaving

the C-terminus free. Thus, the molecular location of immobilization can be dialed-in depending

on which protein region is needed for receptor binding. The ABD could have been placed at

either terminus of FGF-2, since both the N and C-termini are not involved in receptor binding

(PBD: IEV2)[176]. Furthermore, immobilization using binding systems is versatile since it can

be applied to any protein expressed as a fusion protein with ABD.

3.6 Conclusion

The ability to 3D pattern proteins within hydrogels provides a useful tool to the fields of cell

biology where the 3D microenvironment is known to influence cell fate. To this end, we have

developed two separate techniques where proteins were 3D-patterned with the ability to vary

immobilized concentrations. The disulfide bond system provides a direct immobilization method

for proteins with free cysteines. This simple method requires only the native protein for

immobilization with light and will thus facilitate 3D protein patterning for non-experts within the

fields of regenerative medicine. The HSA and ABD system provides a versatile immobilization

method and is applicable to any protein since it is not dependent on any intrinsic property of the

74

biologically relevant protein. The incorporation of ABD provides additional control over the

orientation of the immobilized proteins. Furthermore, both systems provide immobilization

methods without the need of crosslinking agents that could be detrimental to protein bioactivity

or cell viability. These methods should have broad applicability in research involving

biochemical patterning.

75

4 Three-dimensional, spatially-controlled simultaneous patterning of multiple growth factors in hydrogels*

*This chapter was published in Nature Materials. Wylie, R. G.; Ahsan, S.; Aizawa, Y.; Maxwell, K. L.; Morshead, C. M.; Shoichet, M. S., Spatially controlled simultaneous patterning of multiple growth factors in three-dimensional hydrogels. Nat Mater 2011, 10 (10), 799-806.

4.1 Abstract

Three-dimensional (3D) protein-patterned scaffolds provide a more biomimetic environment for

cell culture than traditional two-dimensional surfaces, but simultaneous 3D protein patterning

has proven difficult. We developed a method to spatially control the immobilization of different

growth factors in distinct volumes in 3D hydrogels, and to specifically guide differentiation of

stem/progenitor cells therein. Stem cell differentiation factors, sonic hedgehog (SHH) and

ciliary neurotrophic factor (CNTF), were simultaneously immobilized utilizing orthogonal

physical binding pairs, barnase-barstar and streptavidin-biotin, respectively. Barnase and

streptavidin were sequentially immobilized using two-photon chemistry for subsequent

concurrent complexation with fusion proteins barstar-SHH and biotin-CNTF, resulting in

bioactive 3D patterned hydrogels. The technique should be broadly applicable to the patterning

of a wide range of proteins.

4.2 Introduction

The ability to localize proteins within 3D scaffolds is critical for spatial control of cellular

activities such as cell migration, differentiation and proliferation[2, 26, 110]. Neural tissue,

such as the retina, is defined by a laminar structure, comprised of multiple cellular layers within

a depth of 100-130 m.[177] Taking advantage of a new multiphoton patterning technique,

76

scaffolds can be created with chemically-defined volumes with micron resolution, providing a

framework for controlled growth and differentiation of stem/progenitor cells (collectively termed

precursor cells). Chemically defined volumes of growth factors were selected based on literature

precedence to promote differentiation of retinal precursor cells to mature cell types: N-terminal

sonic hedgehog (SHH) for rod photoreceptors[178]; and ciliary neurotrophic factor (CNTF) for

bipolar cells[124, 179] or Müller glia[126, 179]. Importantly, several factors, including SHH

and CNTF, have been previously shown to remain active when immobilized[56, 120, 180-182].

Furthermore, since SHH is a chemoattractant for adult neural precursor cells (NPCs)[183], we

investigated cell migration within 3D photochemically-patterned, immobilized SHH gradient

hydrogels.

To achieve broad applicability, several criteria for protein patterning in hydrogels were required:

(1) protein localization must be controlled in three dimensions; (2) multiple proteins must be

immobilized simultaneously to avoid protein inactivation over multiple immobilization and

washing steps; and (3) the system must be applicable to a wide range of proteins. Satisfying

these conditions would yield hydrogels with any desired 3D configuration of bioactive proteins

as a basis for engineered tissue constructs.

We took advantage of the orthogonal chemistry of peptide binding pairs[184, 185], to design

hydrogels with 3D immobilized proteins. The binding peptides were first immobilized in defined

volumes in the hydrogel using two-photon chemistry[112, 113]. Then the candidate factors,

expressed as fusion proteins containing the corresponding binding partner, were immobilized.

The system is applicable to many proteins by using fusion proteins, where one end of the

molecule contains the binding domain and the other a biologically active protein, such as a

growth factor that guides precursor cell differentiation. The physical binding pairs used herein

77

were barnase-barstar[186, 187] and streptavidin-biotin[188] because they form strong complexes

with Kds of 10-14 M and 10-15 M, respectively. Barnase and streptavidin are ideal anchoring

proteins because they are stable, which is not true for all proteins, yet critical during the

fabrication steps of the hydrogel. For example, streptavidin has a high thermal stability with a

melting temperature of 75°C[74]; and barnase readily refolds in the event that it is

denatured[189].

Figure 4-1: Method for the simultaneous immobilization of SHH and CNTF. Maleimide

barnase ( ) is immobilized using two-photon photochemistry and a femtosecond laser. The

hydrogel is then washed in buffer to remove unbound mal-barnase. Next maleimide

streptavidin ( ) is immobilized using two photon irradiation followed by another washing

step. The fusion proteins barstar-SHH ( ) and biotin-CNTF ( ) are soaked into the gel

and bind to barnase and streptavidin, respectively. After washing out excess protein, both

SHH and CNTF are simultaneously and independently immobilized in three-dimensions.

An agarose hydrogel was modified with coumarin-caged thiols which, upon two photon

irradiation, is uncaged to yield reactive thiol groups[113]. Agarose was chosen as the hydrogel

because it is a transparent scaffold, which is critical to multi-photon chemical patterning[113].

Furthermore, the diffusion of proteins through agarose is sufficient for their introduction and 3D

78

immobilization[190]. 6-Bromo-7-hydroxycoumarin was chosen as the thiol protecting group

since it has been previously used with cells and tissue slices [85]. Thiols were selected as the

reactive sites because they were previously shown to be both effective for biomolecule

immobilization and non-cytotoxic[2, 191]. The thiols act as anchoring sites for the sequential

immobilization of both barnase and streptavidin which were modified to contain thiol reactive

maleimides. Two-photon patterning affords 3D control because of the limited excitation volume.

As shown schematically in Figure 4-1, maleimide (mal)-barnase is photochemically immobilized

in the hydrogel, washed to remove unreacted mal-barnase, and then mal-streptavidin is

immobilized using the same photochemistry, but in spatially distinct volumes. After washing out

the unreacted mal-streptavidin, the fusion proteins barstar-SHH and biotin-CNTF are soaked into

the gel simultaneously, and specifically bind to immobilized barnase and streptavidin,

respectively. A final washing step is performed to remove unbound proteins, yielding two

immobilized bioactive factors in spatially defined volumes within the agarose hydrogel (see

Methods). This methodology was tested for each factor individually, facilitating quantification

and bioactivity of the immobilized protein, and then for both factors together, demonstrating the

power of this technique for simultaneous protein immobilization.

4.3 Materials and Methods

4.3.1 Materials

The plasmid for SHH-barstar in a pET-21a(+) vector was purchased from Genscript (NJ, USA).

Sodium chloride, imidazole, Terrific broth, Lennox broth, guanidine hydrochloride, Tris,

ampicillin and kanomycin were purchased from Bioshop (ON, Canada). Maleimide-streptavidin,

agarose type IX-A and anti-foam 204 were purchased from Sigma-Aldrich (ON, Canada). Sulfo-

79

SMCC, 96-well chamber slides and all dialysis cassettes were purchased from Thermo-Fisher

(Pittsburgh, PA, USA). FluoReporter biotin quantification assay kit, NHS-Alexa-Fluor-488,

NHS-Alexa-Fluor-633, Quant-iT PicoGreen dsDNA Assay kit, Live/Dead Viability Assay,

Superscript II cDNA synthesis kit, Platinum taq DNA Polymerase High Fidelity, 568-Alexa-

Fluor donkey secondary antibody and Hoechst 33258 were purchased from Invitrogen (CA,

USA). Anti-Phospho-Stat3-P tyro705 antibody was purchased from Cell Signaling Technology

(MA, USA). Mouse recombinant N-terminal sonic hedgehog and rat recombinant ciliary

neurotrophic factor were purchased from R&D systems (MN, USA). All dialysis membranes

were purchased from Spectrum Labs (CA, USA). RNeasy Mini kit was purchased from Qiagen

(MD, USA). Ni-NTA resin was purchased from Qiagen (ON, Canada). Vivaspin protein

concentrators were purchased from GE Healthcare (Buckinghamshire, UK). Biotin ligase was

purchased from Avidity (CO, USA). CD1 mice were purchased from Charles River (MA, USA).

DMSO was purchased from Caledon (ON, Canada). Mal-GRGDS was purchased from AnaSpec

(CA, USA). E. coli BL21 (DE3) was purchased from New England Biolabs (MA, USA).

4.3.2 Photo-patterning and Imaging.

All patterns were created and imaged on a Leica TCS-SP2 confocal microscope equipped with

an Argon (50mW; 458, 476, 488, 514nm), Red HeNe (10mW; 633nm), a multi-photon Mai Tai

laser using a 20x objective (NA = 0.4) and an electronic stage. For patterning experiments, the

multi-photon laser was set to 740 nm with an offset of 75% and gain of 0% for visualization and

an offset of 75% and gain of 43% for patterning. One scan of a 100 µm x 100 µm square takes

1.28 seconds. A typical patterned hydrogel of 10 squares (Figures 2 and 3) took between 2 and 6

minutes. The maximum length scale that can be achieved (maximum depth of patterning) is

limited by the working distance of the lens (15 mm). Leica software version 2.5.1227a was used

80

for the visualization and fluorescence quantification. Z-stacks and 3D images were constructed

using Image J.

4.3.3 Patterning SHH-barstar.

25 l of 1 wt% coumarin sulfide agarose gels with 0.15 mg/ml of mal-barnase was patterned and

reacted for 2 h at RT in a humidity chamber. The series of boxes was created by selecting a 100

µm by 100 µm squares, having a height of ~20-40 µm, 400 µm below the surface of the

hydrogel. Using a macro, the first box was scanned 10 times followed by an additional 4 scans

for each subsequent box. The gels were then washed in 200 ml of PBS for 1 d. 20 l of 0.3

mg/ml SHH-barstar-488 was placed on top of the gel and left overnight at RT. The gels were

washed in PBS pH 7.4 for 2 d, changing the PBS daily. For the quantification, a z-stack was

imaged spanning 182 µm with 2 µm steps and 6 scans per slice. The 458, 476, 488 nm excitation

wavelengths were set to 100% and the gain of the PMT was 602 with wavelengths from 500 to

590 nm collected.

4.3.4 Patterning biotin-CNTF.

25 l of 1wt% coumarin sulfide agarose gels with 1 mg/ml of mal-streptavidin was patterned and

reacted for 2 h at room temperature (RT) in a humidity chamber. A series of boxes was created

by selecting a 100 µm by 100 µm square, having a height of ~40-80 µm, 400 µm below the

surface of the hydrogel. Using a macro the first box was scanned once followed by an additional

2 scans for each subsequent box. The gels were then washed in 200 ml of PBS pH 7.4 for 1 d. 20

l of 0.54 mg/ml biotin-CNTF-633 was placed on top of the gel and left for 16 h at RT. The gels

were washed again in PBS pH 7.4 for 2 d, changing the PBS once. For the quantification, a z-

stack was imaged spanning 163.2 µm with 2 µm steps and 6 scans per slice. The 633 nm

81

excitation wavelength was set to 100% and the gain for the photomultiplier tube (PMT) at 591

with wavelengths from 640 to 750 nm collected.

4.3.5 Dual Patterning.

25 µl gel of 1 wt% coumarin-sulfide agarose with 0.15 mg/ml of mal-barnase was patterned.

The truncated circle was selected, after which the region was scanned 40 times. This was

repeated 3 additional times with each 100 µm below the previous pattern to construct the layered

pattern. After 2 h in a humidity chamber, the gel was washed in PBS pH 6.8 for 2 h. A solution

of 20 µl of 2 mg/ml of maleimidopropionic acid in PBS pH 6.8 was added on top of the gel.

After 16 h the gel was washed for 1 d in PBS pH 6.8. 20 µl of 2 mg/ml mal-streptavidin in PBS

pH 6.8 was added on top of the gel at 4°C. After 16 h, the gel was patterned again by selecting an

oval region fitting into the truncated circle and scanned 15 times. This was repeated for each

layer of the barnase pattern. The gel was then washed in 200 ml of PBS pH 7.4 for 1 d. A 20 µl

solution of 0.3 mg/ml of both barstar-SHH-488 and biotin-CNTF-633 was added on top of the

gel. After 1 d, the gel was washed for 2 d in PBS pH 7.4 changing the buffer once. A 327.6 µm

stack was constructed with 2.1 µm spacing between slices using the following settings: lasers

458, 476, 488 and 633 nm set to 100%; 6 scans per slice; collected wavelengths of 500-590 nm

and 640-800 nm; and PMTs of 683 and 589 for green and red channels, respectively.

4.3.6 Migration of NPCs into SHH/RGD channel.

NPCs placed on top of hydrogels with patterns consisting of a SHH gradient with GRGDS or

GRGDS alone. After 14 d, hydrogels were imaged and cell migration into the hydrogel was

compared between conditions. 50 µl of 0.3 wt% coumarin-sulfide agarose gel with 0.15 mg/ml

mal-barnase (PBS 6.8) in a glass cuvette (interior length: 10 mm; interior width: 2 mm) coated

82

with Sigmacote was patterned through the glass cuvette viewing the gel from the side. At 650

µm from the side of the gel, a 300 µm x 300 µm box was selected such that 100 µm was above

the surface of the gel. A gradient was then created by scanning the box 10 times, followed by

moving the box further into the gel by 100 µm and scanning 11 times. This was repeated 6 times,

adding one scan at each repeat. The entire procedure was repeated every 15 µm further from the

side of the gel until we reach a distance of 845µm from the side of the gel. The final SHH

channel of 300 µm x 200 µm x 300 µm took 20.8 minutes to pattern. Then 50 µl of 1 mg/ml mal-

GRGDS (PBS 6.8) was added on top of the gel and left overnight at 4°C. The region that was

patterned with barnase was irradiated to co-localize GRGDS with barnase. Again a 300 µm x

300 µm box was selected and the region was scanned 20 times followed by moving 10 µm

further from the side of the gel and repeating over a distance of 190 µm. Another adjacent

column was created to cover the entire barnase pattern. The gel was then washed for 2 d, and

then 0.5 mg/ml of SHH-488 was added on top and left for 16 h. The gel was then washed for 3 d.

The SHH-488 pattern was imaged at 750 µm with the following settings: excitation wavelengths

458, 476, 488 at 100%, PMT of 700, and 20 scans. The gradient was quantified by comparing the

fluorescent profile to a standard curve of known 488 concentrations. Gels were first soaked in

serum free media with EGF, FGF and heparin, and 20,000 YFP-expressing NPCs (passage 2;

derived from the adult mouse subventricular zone[167]) were plated on top of the gel in 300 µl

of media. The media was replaced every 3 d. Cells were imaged after 14 d. The same procedure

was performed for the gel with only a GRGDS pattern without the barnase patterning step.

83

4.3.7 Conversion of fluorescence intensity to concentration for bartar-

SHH-488 and biotin-CNTF-633

To convert the fluorescence into concentration of protein, two calibration curves were

constructed for NHS-Alexa 488 and NHS-Alexa 633. For Alexa 488, 1 wt% coumarin sulfide

agarose hydrogels with concentrations of 0, 5, 20, 100, 250, 500, 600 nM were imaged at 400

µm below the surface of the gel. For Alexa 633, 1 wt% agarose hydrogels with concentrations of

0, 5, 20, 100, 250, 500, 600, 700, 800, 1000 nM were imaged using the same settings used for

protein quantification at 400 µm below the surface of the gel. The calibration curve along with

the known number of 488 or 633 tags per protein was used to calculate the protein concentration.

4.3.8 Stability study for immobilized SHH using barnase-barstar

The gels used for the quantification of immobilized SHH were soaked in 50 ml of PBS for 14 d,

and reimaged using the same settings as described for the original quantification of SHH.

4.3.9 Preparation of coumarin sulfide agarose

Coumarin-sulfide agarose was prepared as previously reported with minor modifications[113].

For the patterning experiments, 400 mg of agarose in 40 ml of DMSO was mixed with 180 mg of

carbonyl diimidazole for 2 h, followed by the addition of 100 mg of coumarin sulfide with 5

drops of triethylamine. After 24 h the product was purified by dialysis against water, yielding a

substitution rate of 2.8% (2.8% of the agarose repeat units were modified with coumarin sulfide).

The coumarin-sulfide agarose for the bioactivity experiments was prepared exactly as described

by Wosnick et al[113], yielding a substitution rate of 0.5%.

84

4.3.10 Plasmid design

Mouse CNTF was expressed from pET24d with the biotinylation sequence and histidine tag at

the C-terminus. Barnase along with barstar from Bacillus amyloliquefaciens was cloned into a

modified pET15b plasmid with the addition of 21 amino acids at the N-terminus containing a

hexahistidine tag followed by a TEV cleavage site. Mouse SHH was expressed as a fusion

protein with SHH at the N-terminus followed by a spacer, barstar and a histidine tag.

4.3.11 Expression and purification of barnase

The plasmid pMT1002 was acquired from Dr. Hartley through addgene[192]. The DNA

encoding for the barnase and barstar was amplified by PCR using the primers 5’-

TTGTATTTCCAGGGCGCACAGGTTATCAACACGTTTG-3’ and 5’-

CAAGCTTCGTCATCAAAGAAAGTATGATGGTGATG-3’. The amplified product was

cloned (ligation independent cloning) into a modified pET15b plasmid with the addition of 21

amino acids at the N-terminus containing a hexahistidine tag followed by a TEV cleavage site.

The construct was verified through sequencing. The plasmid was transformed into BL21 (DE3)

E. coli. A starting culture was shaken overnight at 37°C in 20 ml of LB with 100 µg/ml of

ampicillin, and then transferred into a 1.8 L solution of terrific broth, 100 µg/ml of ampicillin

and 6 drops of anti-foam 204. The culture was placed at 37°C with an air bubbler until an OD600

of 0.8 when 1.8 ml of 190 mg/ml of IPTG was added and the protein was expressed for 5 h at

37°C.

The bacteria pellet was collected by centrifugation for 10 min at 12,227 g (Beckman coulter

centrifuge Avanti J-26 with rotor JLA-8.1000) and resuspended in 40 ml of denaturing buffer (6

M Guanidine, 100 mM NaH2PO4, 10 mM Tris, 10 mM imidazole, pH 8.0) and shaken overnight

85

at room temperature. The lysate was centrifuged for 30 min at 48,384 g (Beckman coulter

centrifuge Avanti J-26 with rotor JA-25.50). The supernatant was collected and mixed with 2 ml

of Ni-NTA resin for 15 min at room temperature. The resin was then washed with 4 x 20 ml of

denaturing buffer and eluted with 2 x 15 ml of elution buffer (6 M Guanidine·HCl, 200 mM

acetic acid). The protein was then refolded by dialysis in PBS pH 7.4. The solution was

centrifuged at 4,000 RPM for 15 min to remove precipitated protein, and further purified by SEC

(AKTA design FPLC by Amersham Pharmacia biotech with a HiLoad 16/60 Superdex 200 prep

grade column) yielding 2.4 mg of soluble barnase. An extinction coefficient and MW of 27310

and 14.8kDa, respectively, was used for barnase.

4.3.12 Synthesis of maleimide-barnase with sulfo-SMCC

500 µl of 0.6 mg/ml solution of barnase in PBS pH 7.4 was mixed with 100 µl of 4 mg/ml of

sulfo-SMCC solution in ddH2O for 30 min at room temperature. A precipitate was removed by

centrifuging for 10 min at 16,160g (Beckmann Coulter, Microfuge 16). The protein was purified

by dialysis at 4°C against PBS pH 6.8 using a dialysis cassette (MWCO 3,500). The sample was

concentrated to a final concentration of 0.3 mg/ml and stored at -80°C.

4.3.13 Expression, purification and labeling of barstar-SHH

A plasmid was purchased from Genscript, which coded for a fusion protein with SHH at the N-

terminus, followed by a spacer (EFPKPSTPPGSSGGAP)[186], barstar and a pentahistidine tag

at the C-terminus in a pET21a(+) plasmid. The SHH-bartar plasmid was transformed into BL21

(DE3) E. coli and expressed. A starting culture was shaken overnight at 37°C in 20 ml of LB

with 100 µg/ml of ampicillin, then transferred into a 1.8 L solution of terrific broth with 100

µg/ml of ampicillin and 6 drops of anti-foam 204. The culture was placed at 37°C with an air

86

bubbler until an OD600 of 0.8. The temperature was then reduced to 16°C and 1.8 ml of a 190

mg/ml IPTG solution was added. After 18 h, bacteria was collected by centrifugation at 12,227 g

(Beckman coulter centrifuge Avanti J-26 with rotor JLA-8.1000) for 10 min at 4°C and

resuspended in binding buffer (50 mM Tris pH 7.5, 500 mM NaCl, 5 mM imidazole) to a total

volume of 60 ml. The sample was divided into 2 vials of 30 ml and each was lysed by probe

sonication (Misonix S-4000 Sonicator Ultrasonic Processor equipped with a Dual Horn probe)

with 30% amplitude and an on/off pulse of 2 s for a total sonication time of 5 min. The lysates

were centrifuged at 48,384g (Beckman coulter centrifuge Avanti J-26 with rotor JA-25.50) for

15 min at 4°C, the soluble fraction was collected and mixed with 2 ml of Ni-NTA resin at 4°C

for 15 min. The resin was washed with 10 x 10 ml of wash buffer (50 mM Tris pH 7.5, 500 mM

NaCl, 30 mM imidazole) and eluted with elution buffer (50 mM Tris pH 7.5, 500 mM NaCl, 250

mM imidazole). The protein solution was dialyzed against 4 L of PBS pH 7.4 changing the

buffer once. 3.7 mg of SHH-barstar was collected. A 1 ml solution of 1 mg/ml barstar-SHH in

PBS was mixed at room temperature with 30 µl of 5 mg/ml NHS-Alexa-488 in PBS. After 1 h an

additional 40 µl of 5 mg/ml NHS-488 was added. After an additional hour the protein was

purified by adding 200 µl of NTA-Ni resin, washing with 10 x 1 ml of wash buffer, and eluted

with elution buffer. The substitution rate was determined to 5.8 mol of Alexa 488 per mol of

protein calculated as explained by Invitrogen[166]. An extinction coefficient and MW of

46940M-1cm-1 and 32.3kDa, respectively, was used for barstar-SHH.

4.3.14 Expression, purification and labeling of biotin-CNTF

CNTF was expressed as previously described[193]. Briefly, the protein was purified using a Ni-

NTA column followed by dialysis against PBS pH 7.4 and further purified by size-exclusion

chromatography (SEC) (AKTA design FPLC by Amersham Pharmacia biotech with a HiLoad

87

16/60 Superdex 200 prep grade column) with PBS pH 7.4 as the running buffer. CNTF was

biotinylated with a biotin protein ligase kit according to manufacturer’s protocol. Biotinylated

CNTF was purified by dialysis for 48 h in a dialysis cassette (MWCO 10 kDa) for 48 h against

4L PBS pH 7.4 changing the buffer once after 24 h. The degree of biotinylation was determined

to be 100% using the FluoReporter biotin quantification assay kit. 0.5 ml of a 0.8 mg/ml solution

of biotin-CNTF was mixed with 20 µl of a 10 mg/ml solution of NHS-Alexa-633 at room

temperature, after one hour another 20 µl of a 10 mg/ml solution of NHS-Alexa-633 was added.

After an additional hour, the protein was purified by dialysis for 48 h against 4 L PBS pH 7.4

using a dialysis cassette (MWCO 10kDa) changing the buffer after 24 h, yielding a substitution

rate of 2.9 moles of Alexa-633 per mole of biotin-CNTF, calculated as described by

Invitrogen[166]. An extinction coefficient and MW of 33570 M-1cm-1 and 28.0 kDa,

respectively, was used for biotin-CNTF.

4.3.15 Preparation of gels for bioactivity assay

All gels and solutions were prepared in PBS pH 7.4. A 2 wt% sulfide-coumarin agarose (with

0.5% of agarose repeat units modified with coumarin sulfide) solution was irradiated for 10 min

using a UV reactor (Rayonet by The southern New England Ultraviolet Company with 365nm

UV lamp).

For streptavidin gels, 781 µl of the irradiated coumarin-sulfide agarose solution was mixed with

250 µl of a 2 mg/ml solution of maleimide streptavidin, 5.7 µl of a 1 mg/ml maleimide GRGDS

solution, and 213 µl of PBS. Gels were prepared by pipetting 50 µl of the solution prepared

above into wells in a 96 well chamber slide. The gels were left at room temperature for 1 h, and

then placed at 4°C for an additional hour to gel. The gels were washed in 250 ml of PBS at 4°C

for 4 d, changing the buffer daily.

88

For CNTF immobilized gels, 75 µl of a 100 µg/ml of biotin-CNTF solution was added on top of

agarose-GRGDS-streptavidin gels from above, and left overnight at room temperature. The gels

were then washed again for 4 d changing the buffer daily. The amount of biotin-CNTF

immobilized was calculated using biotin-CNTF-633. The fluorescence from the gel was

measured using the confocal as described before and compared to a calibration curve. The

concentration of immobilized CNTF was calculated to be 6.12±0.91 nM.

Agarose-RGD-barnase gels were prepared as follows: 781 µl of the irradiated 2 wt% coumarin-

sulfide agarose solution was mixed with 250 µl of a 0.2 mg/ml maleimide-barnase solution, 5.7

µl of a 1 mg/ml maleimide-RGD solution and 213.3 µl of PBS. The gels were then prepared

using the same method as agarose-streptavidin gels.

For SHH immobilized gels, 75 µl of a 100 µg/ml solution of SHH-barstar was added on top of

agarose-GRGDS-barnase gels and treated the same way as for the immobilization of CNTF. The

amount immobilized SHH was quantified using SHH-barstar-488. The concentration of SHH

immobilized was calculated to be 4.22±0.27 nM.

For agarose-GRGDS only gels, 1.25 ml of irradiated 2 wt% coumarin-sulfide agarose was mixed

with 24 µl of mal-GRGDS and 726 µl of PBS. The gels were treated the same as for agarose-

GRGDS-streptavidin gels.

4.3.16 Obtaining retinal precursor cells

RPCs were obtained as described by Tropepe[122]. Briefly, the ciliary margin of adult CD1 mice

were dissected, dissociated and plated in serum-free media with 20 ng/mL FGF-2 and 2000

U/mL Heparin sulphate for 7 d at 37 °C, 5% CO2. After 7 d, clonal spheres were collected,

dissociated and single cell suspensions were used for further experiments.

89

4.3.17 Plating of cells for bioactivity studies

5,000 RPCs were plated on each agarose gel for 7 d in serum-free media. RPCs were cultured

with 10 ng/mL mouse recombinant N-terminal sonic hedgehog peptide or 10 ng/mL rat

recombinant ciliary neutrophic factor or in the absence of soluble growth factors.

4.3.18 Cell survival analysis with PicoGreen

After 7 d of culture, RPCs were lysed using 0.3% Triton-X and vortexed thoroughly. To quantify

dsDNA content, the Quant-iT PicoGreen dsDNA Assay kit was used and the manufacturer’s

instructions were followed.

4.3.19 Cell survival analysis with Live/Dead staining

After 7 d of culture, RPCs were stained using the Live/Dead® Cell Viability Assay according to

manufacturing protocols. Cells were then imaged on a fluorescent microscope.

4.3.20 Gene Expression Assays

RT-PCR was used to assay for specific gene expression. RNA was isolated from RPCs after 7 d

of culture using RNeasy Mini kit as per the manufacturer’s instructions. The RNA was reverse

transcribed into cDNA using Superscript II cDNA synthesis kit following the manufacturer’s

instructions. For PCR reactions, Platinum Taq DNA Polymerase High Fidelity was used with the

following primers at a final concentration of 800 nM: gli2: FWD: 5’-

CACAGGGCGGGCACAAGAT-3’, REV: 5’-GGAGGGCAGTGTCAAGGAA-3’, 18S rRNA:

FWD: 5’-GTAACCCGTTGAACCCCAT-3’, REV: 5’-CCATCCAATCGGTAGTAGCG-3’,

ptch1: FWD: 5’-AATTCTCGACTCACTCGTCCA-3’, REV: 5’-

90

CTCCTCATATTTGGGGCCTT-3’, cntfr: FWD: 5’-TGGACTGTGTTTCTGCGTGT-3’, REV:

5’-TGGAGAACAGCTGGTGGTAA-3’. The PCR reaction was set at 95 oC for 5 min, then 33

cycles of 95 oC for 15 s, 60 oC for 30 s, 72 oC for 30 s and then a final 72 oC for 10 min step

performed on a GeneAmp PCR System 9700 (Applied Biosystems, Foster City, CA). Samples

were run on 2wt% gels in TE buffer at 100V for 30 minutes.

4.3.21 Immunocytochemistry

RPCs were fixed after 4 d of culture with fresh 4% paraformaldehyde. RPCs were incubated with

100% ethanol for 30 s at room temperature and then washed with 1x Stockholm’s PBS. The

cells were then blocked with 10% normal donkey serum for 1 h at room temperature. They were

incubated with polyclonal anti-Phospho-Stat3-P (Tyro705) antibody overnight at 4 °C. The cells

were then incubated with a 568-AlexaFluor donkey secondary antibody for 1 h at 37 °C, washed,

counterstained with Hoechst 33258 and imaged. Cells were also stained with Hoechst.

Photographs taken from 4 random quadrants. The images were enhanced using Image J software

to highlight cell nuclei staining for Hoechst stain and Stat-3-phosphorylated. Random selection

of images was used, and the number of nuclei and the number of Stat-3-phosphorylated were

counted. An average of the percentage of stat-3-phosphorylated positive cells per image was

taken (n=5, mean ± s.d.).

4.4 Results and Discussion

4.4.1 3D immobilization of SHH using barnase-barstar

To spatially control the immobilization of SHH to agarose, barstar-SHH and barnase-agarose

were synthesized and then the two reacted. The fusion protein barstar-SHH was expressed in E.

91

coli and then labeled with the fluorescent Alexa 488 prior to reaction with agarose-barnase. To

synthesize the latter, barnase was expressed in E. coli[187, 192], followed by modification with

sulfo-SMCC, yielding mal-barnase that reacted readily with deprotected agarose-coumarin-

sulfides (Figure 4-2a).

Figure 4-2: 3D immobilization of barstar-SHH-488 using barnase-barstar. (a) Maleimide-

barnase ( ) was immobilized in a coumarin-sulfide agarose gel, followed by the addition of

barstar-SHH ( ) modified with Alexa 488. (b) 10 different squares 100 x 100 µm having

heights of 20-40 µm were patterned 400 µm below the surface of the gel, with each square

being scanned a different amount from 10 to 46 scans. Scale bar: 100 µm. (c) The amount

of barstar-SHH-488 immobilized per number of scans was quantified by measuring the

fluorescence from each box and compared against a standard curve of coumarin-sulfide

agarose gels with known concentrations of Alexa 488 (mean ± s.d., n=3). (d) The z-axis

profile of fluorescence of barstar-SHH-488 for boxes with 10, 26 and 46 scans was plotted,

with the maximum intensity centered at 0 µm.

92

The synthesis of barnase-agarose was a multi-step process, involving first synthesis of coumarin-

protected agarose-sulfide[113], selective deprotection (in defined volumes) of the coumarin

groups by pulsed Ti-sapphire laser and finally reaction of these reactive agarose sulfides with

mal-barnase. A 1 wt% (wt/vol) coumarin sulfide agarose hydrogel was synthesized with 2.8% of

the agarose repeat units modified with coumarin-sulfide and then mixed with mal-barnase, prior

to casting as a gel. By controlling the exposure of defined agarose volumes to the Ti-Sapphire

pulsed laser at 740 nm, the amount of photo-exposed agarose-sulfide groups and thus the amount

of mal-barnase immobilized was varied. Moreover, by simply varying the number of laser scans

over a selected volume, the concentration of SHH-barstar immobilized to streptavidin-barnase

was controlled.

As shown in Figure 4-2, barstar-SHH was immobilized in spatially defined volumes of agarose-

barnase. A series of squares, patterned 400 µm below the surface of the hydrogel, demonstrates

that the amount of irradiation correlates with the amount of protein immobilized. Ten boxes of

100 µm by 100 µm by ~40 µm (depth) were patterned in the presence of mal-barnase with an

increasing number of laser scans, from 10 to 46 per box, washed to remove unreacted mal-

barnase and cleaved coumarin molecules, and then immersed in a solution of barstar-SHH-488

for 16 hours. The gel was thoroughly washed to remove excess barstar-SHH-488 and visualized

using confocal microscopy (Figure 4-2b). The fluorescence intensity was converted to amount

immobilized by comparison to a calibration curve and plotted as a function of the number of

scans in a given volume (Figure 4-2c). Interestingly, after 10 scans, 12 nM of SHH were

immobilized and after 46 laser scans, 134 nM of SHH were immobilized. A linear relationship

between number of scans and amount of immobilized SHH was observed. 12 nM represented the

lowest concentration that could be imaged, not the lowest concentration that can be immobilized,

since the quantification was restricted by the detection limit of the confocal microscope. The

93

concentration of immobilized SHH can be further controlled by both the substitution rate of

coumarin sulfide on agarose and varying the laser intensity. The distribution of immobilized

barstar-SHH in the z-plane within the agarose gel was demonstrated by plotting the fluorescence

profile of barstar-SHH-488 against distance along the z-axis. The point of irradiation (maximum

fluorescence) was arbitrarily set to 0 microns for graphical representation. The fluorescent

intensity profile of the z-axis for all boxes spans approximately 40 µm (Figure 4-2d), which

represents the best resolution that can be achieved in the z-axis. The profile of fluorescence along

the z-axis broadened with scan number because the excitation volume of the two-photon laser

increased with amount of irradiation; however, a 5 µm resolution can be achieved in the x/y

plane when patterning fluorescent molecules[113]. These data demonstrate that the coumarin-

sulfide photochemistry combined with the barnase-barstar system allows 3D immobilization of

proteins.

Figure 4-3: Stability of SHH pattern using barnase-barstar immobilization. The amount of

SHH immobilized in the pattern from Fig 3 was recalculated after soaking the gels in PBS

pH 7.4 for 14 days at room temperature using the same procedure as previously described.

No significant difference in immobilized SHH over time was observed, demonstrating that

the pattern remains stable over 14 days (mean ± s.d., n=3).

94

The stability of agarose-immobilized SHH was investigated by measuring the SHH

concentration before and after immersion in PBS. After 14 days in PBS, there was no statistically

significant change in concentration of SHH-agarose (Figure 4-3), indicating that the barnase-

barstar interaction is a suitable physical interaction for stable protein immobilization. This

complements the streptavidin-biotin complex, which has been previously shown to be effective

for stable biomolecule immobilization in hydrogels[113, 194].

95

Figure 4-4: 3D immobilization of biotin-CNTF-633 using biotin-streptavidin. (a)

Maleimide-streptavidin ( ) was immobilized in a coumarin-sulfide agarose gel, followed by

the addition of biotin-CNTF ( ) modified with Alexa 633. (b) 10 different regions of boxes

100 x 100 µm having heights of 40-80 µm were patterned 400 µm below the surface of the

gel, with each region being scanned a different amount from 1 to 19 scans. Scale bar: 100

µm. (c) The amount of biotin-CNTF-633 immobilized per number of scan was quantified

by measuring the fluorescence from each box and compared against a standard curve of

coumarin-sulfide agarose gels with known concentrations of Alexa 633 (mean ± s.d., n=3).

(d) The z-axis profile of fluorescence of biotin-CNTF-633 for boxes with 1, 9 and 19 scans

was plotted, with the maximum intensity centered at 0 µm.

4.4.2 3D immobilization of CNTF using streptavidin-biotin

To spatially control the immobilization of CNTF to agarose, biotin-CNTF and streptavidin-

agarose were synthesized and then the two reacted (Figure 4-4a), following a similar overall

strategy as described with barstar-SHH and barnase-agarose. Mouse CNTF with the biotinylation

sequence, GLNDIFEAQKIEWHE[195], and a histidine tag at the C-terminus was expressed in

E. coli from a pET-24d vector[193] and purified prior to biotinylation with the E. coli

biotinylation enzyme. The protein was then covalently-modified with the fluorescent Alexa 633

tag to allow visualization once immobilized to agarose-streptavidin.

To visualize 3D immobilized fluorescently-tagged CNTF, a series of boxes was patterned in the

agarose gel, with scans varying from 1 to 19 (Figure 4-4b). As was observed in the barstar-

barnase system, increasing the number of scans with the multiphoton Ti-sapphire pulsed laser (at

740 nm) led to increased immobilized CNTF, based on increased agarose-sulphides available to

bind mal-streptavidin and then biotin-CNTF. The concentration of immobilized CNTF was

determined by fluorescence to vary between 20 nM to 80 nM as a function of scan number

96

(Figure 4-4c). The fluorescence profile along the z-axis for biotin-CNTF was quantified using

the same method described for barstar-SHH. The fluorescence intensity profile of the z-axis for

the box increased with number of scans from approximately 40 m for 1 scan to 100 m for 19

scans, with the peak fluorescence centered at 0 µm (Figure 4-4d). The resolution in the z-axis

decreased (broadened) in the streptavidin/biotin systems compared to that of barnase/barstar.

Since the same photochemistry was used for both methods, the difference in resolution is likely a

consequence of the binding interactions having different valencies.

Interestingly, similar amounts of barnase (11.9±2.3 nM) and streptavidin (15.8±4.8 nM) were

immobilized with comparable laser scan numbers of 10 and 9, respectively, yet significantly

different concentrations of SHH and CNTF were immobilized due to the different binding

capacities of barnase-barstar and streptavidin-biotin. For example, the 10 scans required to

immobilize barnase resulted in 11.9±2.3 nM of barstar-SHH whereas the 9 scans used to

immobilize streptavidin resulted in 63.0±4.8 nM of biotin-CNTF. Streptavidin has four binding

sites for biotin[196, 197] whereas barnase has only one binding site for barstar, thus accounting,

in part, for the different concentrations of immobilized factors. In this way, we were able to vary

the multivalency of grafted proteins between our systems, which has previously been shown to

be important for protein potency on linear polymer chains[198].

4.4.3 Simultaneous immobilization of SHH and CNTF

The simultaneous 3D immobilization of proteins was achieved with SHH and CNTF by taking

advantage of orthogonal chemistry with the selective protein binding pairs, barnase-barstar and

streptavidin-biotin (Figure 4-1). By first immobilizing mal-barnase to distinct volumes of

coumarin-deprotected agarose sulfide, washing extensively, and then repeating with mal-

97

streptavidin, the agarose hydrogel was synthesized for 3D patterning with the barstar and biotin

binding partners. A coumarin-sulfide agarose hydrogel with mal-barnase in PBS pH 6.8 was

patterned with a

Figure 4-5: Representative figures for the simultaneous 3D patterning of biotin-CNTF-633

and barstar-SHH-488. Mal-barnase was patterned in layers in the shape of a truncated

(green) circle 400, 500, 600 and 700 µm below the surface the hydrogel with 40 scans per

layer. Mal-streptavidin was then patterned in a smaller (red) oval shape inserted into the

truncated circle of the mal-barnase pattern. The oval was patterned with 15 scans in four

layers, following the identical method for mal-streptavidin. Barstar-SHH-488 and biotin-

CNTF-633 were immobilized by simply immersing the hydrogel in solutions of the proteins.

(a) A confocal micrograph showing the loss of coumarin fluorescence of the layer at 400 µm

from patterning of mal-barnase and mal-streptavidin: scale bar: 100 µm. (b) A confocal

micrograph of the layer at 400 µm demonstrating the localization of barstar-SHH-488 and

biotin-CNTF-633 to the volumes patterned: scale bar: 100 µm. (c) 3D projection of the

reconstructed stack using image J 3D viewer rotated to see the layers. (d) Same projection

as (c) viewed from a different angle (biotin-CNTF-633 in red; barstar-SHH-488 in green).

98

truncated circular shape, figuratively representing the globe of an eye. A series of four identical

shapes were patterned into the agarose hydrogel by scanning 40 times across this shape with the

Ti-sapphire laser: each layer was separated by 100 m in depth, with the deepest layer being

700 m below the surface of the gel. The gel was washed thoroughly in PBS pH 6.8 to remove

any unreacted mal-barnase and then re-patterned in the presence of mal-streptavidin. To

immobilize mal-barnase in the appropriate volume, attaining the desired final shape, the loss of

coumarin fluorescence due to mal-barnase immobilization was first imaged with the confocal

microscope. The laser was then focused in the proper volume to immobilize mal-streptavidin

(Figure 4-5a). The mal-streptavidin was patterned in an oval shape, within the pocket of the

truncated circle of mal-barnase, figuratively representing the lens of the eye. The mal-

streptavidin was immobilized in four distinct volumes with 15 scans per layer of the Ti-sapphire

multiphoton laser. More scans were used to immobilize mal-barnase than mal-streptavidin,

based on the results with single protein patterned volumes.

The 3D patterned agarose hydrogel, with distinct volumes of barnase and streptavidin, was

simultaneously modified with barstar-SHH-488 and biotin-CNTF-633 by simply immersing the

hydrogel in a solution containing both proteins. Importantly, barstar-SHH and biotin-CNTF

were selectively immobilized in distinct volumes, following the agarose-immobilized patterns of

barnase and streptavidin, respectively.

Figure 4-5b shows a confocal micrograph of the first layer of the pattern, with barstar-SHH

(green) and biotin-CNTF-633 (red). This image overlaps with that of the loss of coumarin

fluorescence from the patterning steps with the laser, demonstrating spatial control through

multi-photon irradiation. The 3D reconstructed view, showing each of the four patterned volume

layers, demonstrates our ability to pattern both proteins simultaneously in three-dimensions

99

(Figure 4-5c,d). The fluorescence intensity for each layer of SHH-488 appears to decrease with

depth even though the same number of scans was used: the fluorescence at a depth of 400 µm in

the gel is more intense than that at 700 µm because the laser intensity during both excitation and

emission is attenuated as a function of depth due to increased scattering of light. The decrease in

fluorescence was not observed for CNTF-633 (Supplementary Fig. 1), indicating that the two-

photon patterning remains consistent over the depths investigated. The attenuated fluorescence

observed for SHH-488 was not observed for CNTF-633 likely because longer wavelengths

scatter less light than shorter wavelengths. This suggests that the majority of fluorescence loss

for SHH-488 as a function of depth is not from the photochemistry used for immobilization, but

rather an artifact of the imaging process (Supplementary Fig. 1).

The beauty of this technique is its simplicity. It can be applied to a broad range of proteins for

multiple simultaneous patterning and, importantly, can be achieved with the multiphoton laser of

a confocal microscope. Any protein that can be expressed as a fusion protein, with the

appropriate binding partner, can be immobilized. Therefore this system can be applied to

numerous applications involving 3D cell culture. We demonstrated the strength of this technique

with two proteins, but recognize that more proteins can be immobilized simultaneously with the

immobilization of other binding partners. Furthermore, we have demonstrated that concentration

gradients of proteins can be patterned (Figure 4-2, Figure 4-4), which is useful for cell guidance.

Since all but the final washing step is complete prior to protein immobilization, the risk of

denaturing or degrading the proteins during immobilization is significantly diminished. If the

proteins were immobilized sequentially, then those proteins immobilized first may be completely

or partially inactive by the time the final protein is immobilized. Having the proteins added at

the final step, we are more confident in their bioactivity. Thus simultaneous protein

immobilization obviates numerous sequential immobilization and washing steps, which had been

100

the state of the art prior to this study. Moreover, because the immobilization is governed by

specific physical interactions, many potential side reactions are eliminated. For example, by

reacting agarose-thiols with maleimide binding partners, side reactions, such as disulfide bonds

between proteins and agarose-thiols, are obviated.

Figure 4-6: SHH and CNTF signaling pathways are activated in RPCs that are cultured on

immobilized SHH and CNTF, respectively. (a) RPCs were assayed for the presence of the

ptch1 receptor in the SHH pathway using RT-PCR. (b) RPCs upregulate a key SHH

signaling mediator, gli2, in response to immobilized SHH as assayed by RT-PCR. (c) No

cytotoxic effect was found by comparing the survival of RPCs cultured on agarose-barnase-

SHH (with GRGDS), agarose-barnase (with GRGDS) and agarose-GRGDS. Cell numbers

were measured after 7 d by total dsDNA content using the PicoGreen assay (mean ± s.d.,

n=5 with 5,000 cells per gel). No significant difference was observed between groups using

101

one-way ANOVA with Tukey’s post-hoc analysis (p > 0.05). (d) RPCs were assayed for the

presence of CNTF receptor, CNTFR, by RT-PCR. (e) RPCs respond biologically to

immobilized CNTF. This was determined through immunostaining for phosphorylated

STAT-3, a protein activated through phosphorylation upon CNTF ligand binding to

CNTFR. RPCs cultured on gels with either immobilized CNTF or soluble CNTF both

stained positive for STAT-3P, whereas gels with only streptavidin and GRGDS did not

stain for STAT-3P. The percentage of immunostained cells was calculated, as written below

each series of images, and shown to be not statistically different (p>0.05, n=5 samples,

mean ± s.d.). (f) The survival of RPCs cultured on agarose-streptavidin-CNTF (with

GRGDS), agarose-streptavidin (GRGDS) and agarose-GRGDS was similar. Cell numbers

were quantified after 7 d by the amount of dsDNA present using the PicoGreen assay

(mean ± s.d., n=5 with 5,000 cells per gel). No significance difference was observed between

any groups using one-way ANOVA with Tukey’s post-hoc analysis (p > 0.05).

4.4.4 Immobilized SHH and CNTF are bioactive

Having demonstrated the patterning chemistry, we tested the bioactivity of the immobilized

proteins. Since we are interested in the nervous system, we tested bioactivity with retinal

precursor cells (RPCs) derived from the ciliary margin of the adult mouse retina[122]. We

examined the activation of SHH and CNTF signaling pathways in cells exposed to the

immobilized proteins by plating RPCs on the surface of functionalized hydrogels. All agarose

hydrogel scaffolds were chemically modified with the cell-adhesion peptide, GRGDS[199],

because agarose itself is non-adhesive and the bioactivity and cell survival assays to test for

possible cytotoxic effects could not be performed on a non-cell-adhesive substrate[2, 120].

RPCs were first shown to express a SHH receptor, ptch-1, by RT-PCR (Figure 4-6a), which upon

SHH binding leads to the upregulation of the transcription factor, gli2[200]. To test the

bioactivity of the immobilized SHH fusion-protein, RPCs were screened for the expression of

gli2 by RT-PCR. Gli2 expression was evident for both agarose-barnase-barstar-SHH (with

102

GRGDS) and agarose-GRGDS plus soluble wild-type SHH, but not for agarose-barnase (with

GRGDS only) (Figure 4-6b). These data demonstrate that the immobilized SHH fusion-protein,

expressed in E. coli, is bioactive. Assays for potential cytotoxicity compared cell survival on

agarose-barnase-barstar-SHH (with GRGDS) to agarose-barnase (with GRGDS) and agarose-

GRGDS. The PicoGreen assay[201] was used to determine the relative amount of double

stranded DNA as a measure of viable cells after 7 days in culture. The number of viable cells in

all groups was not significantly different (p>0.05), demonstrating no toxic effect of the

immobilization method used for SHH in the cultures (Figure 4-6c). A live/dead stain (calcein

AM/ethidium homodier-1) of the RPCs demonstrated that the percent of live cells varied

between 70 and 80% with no statistical significance between groups, further demonstrating the

coupling method is non-toxic (Supplementary Fig. 2). It is important to realize that we

anticipated a decrease in cell number relative to the number plated, even after 7 d of culture,

because many of the cells are lost during plating[202].

To test the bioactivity of agarose-immobilized CNTF, RPCs were first shown to express the

CNTF receptor (CNTFR) by RT-PCR (Figure 4-6d). To monitor CNTFR activation, we followed

the expression of phosphorylated STAT-3, a well-known downstream effector of CNTF-CNTFR

binding[126, 179, 203]. Importantly, RPCs stained positive with anti-phospho-STAT-3 for

immobilized and soluble CNTF and not for controls that lacked CNTF (Figure 4-6e). These

results demonstrate that our CNTF fusion-protein expressed in E. coli remained bioactive after

immobilization and was able to activate the CNTF signaling pathway. Cell viability assays to

test for potential cytotoxicity were performed using the PicoGreen and live/dead assays, and

demonstrated that RPCs cultured for 7 d on agarose-streptavidin-biotin-CNTF (with GRGDS) vs.

agarose-streptavidin (with GRGDS) vs. agarose-GRGDS had similar cell viability (p>0.05,

Figure 4-6f; Supplementary Fig. 3). Moreover, the RPC viability on the CNTF gels was not

103

significantly different from that on the SHH gels (p>0.05). These results confirm that barstar-

SHH and biotin-CNTF remain bioactive after immobilization, capable of downstream signaling

similar to that observed with soluble controls. Our data are consistent with previous results that

demonstrate SHH and CNTF remain bioactive after immobilization, albeit with different

chemistry and other cell types[181, 182].

Figure 4-7: NPCs migrate into a channel of SHH with RGD. (a) Quantification of the

concentration profile of SHH-488 as a function of depth within the hydrogel from the

surface of the gel to a depth of 100 µm. (b) Brightfield image of SHH/RGD channel show

that NPCs have migrated into the agarose gel after 14 d to a depth of 85 µm. (c) Brightfield

image of RGD only channel show that only minimal migration was observed within the

hydrogel after 14 d to a depth of 20 µm. Mostly processes were observed within the gel. (d)

Confocal micrograph of SHH/RGD channel emphasize migration of NSPCs expressing

YFP into the agarose gel. All scale bars represents 50 µm. For all cell images the white

dashed line represents the surface of the gel.

104

4.4.5 NSPCs migrate into patterns of SHH

To gain greater insight into cell migration into the patterned agarose hydrogel, an immobilized

gradient of barstar-SHH in GRGDS-agarose was photochemically patterned using an identical

synthetic procedure. Using Alexa-488 SHH, the protein gradient was quantified over the first

100 m, relative to a calibration curve, to be 100 to 500 ng/ml (Figure 4-7a). Neural precursor

cells (NPCs) from the subventricular zone, which are known to migrate along an SHH

gradient[183], were plated on SHH-gradient GRGDS-agarose gels and compared to GRGDS-

agarose. Interestingly, we observed cellular migration predominantly into patterns that contained

SHH gradients (to a depth of 85 m, Figure 4-7b), which is a depth appropriate for thin tissues,

such as the retina. Only limited migration was observed into channels with only GRGDS (to a

depth of 20 m, Figure 4-7c). To facilitate visualization of the migrating cells, yellow

fluorescent protein expressing NPCs were shown to migrate into SHH-gradient GRGDS-agarose

gels (Figure 4-7d). These data demonstrate that 3D penetration of cells into photochemically-

patterned agarose gels is facilitated with chemoattractant molecules, such as SHH. These data are

consistent with previous results, where an immobilized vascular endothelial growth factor

concentration gradient was required to guide endothelial cells into an agarose hydrogel[191].

4.5 Conclusion

Defining the cellular microenvironment is becoming increasingly important as we design in vitro

systems to better predict in vivo response and engineer de novo tissues. Designing the three-

dimensional scaffold with the appropriate chemical and physical properties is the first step to

understanding the cues important to cell survival and stimulation. For example, these 3D

biomimetic hydrogels can be used to begin to emulate the complexity of the stem cell niche,

105

presenting ECM-mimetics, growth factors and mechanical stimuli similar to native tissue while

also allowing co-culture of multiple cell types. As the 3D imaging tools continue to advance,

probing cells in 3D will be facilitated and open high throughput (or high content) screening

protocols to advanced 3D patterned hydrogels. Moreover, 3D protein patterning has applications

in regenerative medicine where tissues are engineered in vitro prior to transplantation. Here the

agarose hydrogel serves as a blank palette in which proteins were patterned, demonstrating our

ability to create chemically-complex scaffolds that will be used ultimately to spatially guide cell

fate.

106

5 Discussion

A versatile two-photon patterning system was developed for the 3D patterning of proteins in

hydrogels. Using a combination of photochemistry and protein engineering, we were able to

design a 3D protein immobilization strategy that is applicable to many proteins, functions under

mild conditions and can simultaneously immobilize multiple proteins. Previously, 3D chemical

patterns consisted of small molecules such as short peptides or more recently patterns consisting

of one protein. The system here provides the tools to spatially control the location of multiple

proteins in hydrogels to better mimic the in vivo environment. The following sections will

discuss the patterning technology and its applications.

5.1 3D photochemical patterning in Agarose hydrogels

5.1.1 Agarose as a scaffold for 3D biochemical patterning

Agarose was shown to be a suitable matrix for photopatterning of proteins since it is transparent,

bioinert and provides a blank slate for the incorporation of biochemical cues. The gels were

shown to be transparent to the pulsed two-photon laser (740 nm), with patterns being created up

to 800 µm below the surface of the hydrogel (Figure 4-5). This was more than adequate, since

our system was developed for the design biomaterials for the retina, which is only ~ 100µm

thick. Agarose does not contain any inherent biochemical properties that would limit the type of

biochemical cues that could be patterned. In the studies investigating the migration of NSPCs

into SHH/RGD gradients, the NSPCs only adhered to the portions of the gels containing RGD

and did not interact with unpatterned agarose. In other words, NSPC adhesion sites were

successfully patterned into agarose hydrogels. This could not be performed in gels that are

107

intrinsically adhesive for NSPCs, making agarose an ideal choice for patterning experiments.

Furthermore, agarose did not limit the diffusion of biomacromolecules for immobilization. All

the proteins immobilized in these studies diffused within the agarose hydrogels, as was expected

since agarose has little effect on protein diffusion. Zhang et al. demonstrated that the diffusivity

of bovine serum albumin (BSA) in 0.5 wt.% agarose (8.21 x 10-7 cm2/s) is similar to that of BSA

in infinite solution (9.35 x 10-7 cm2/s) [190]. Diffusion time for proteins in hydrogels depend on

the thickness of the gel and the size of the macromolecule, which is particularly important for

washing steps. Diffusion time is directly related to the square of the gel thickness (L2).

Therefore, the length of washing steps can be lowered by decreasing the thickness of the gel. It

was also noted that proteins around 30-60 kDa could be washed from 1.5 mm gels within 24 h,

whereas larger proteins (~150 kDa) would take up to 3 or 4 days. Therefore, careful attention

must be taken when determining washing times for different gel thickness’ and protein size.

Furthermore, cells were shown to penetrate and migrate down gradients of chemoattractants in

agarose hydrogels (Figure 4-7), solidifying agarose as a relevant bio-scaffold since cells can

penetrate into biochemical patterns (discussed further in section 5.4).

5.1.2 Bhc photocage

6-bromo-7-hydroxycoumarin (Bhc) is a non-toxic two-photon active photocage for amines and

thiols that functions in aqueous environments. A number of TP active photocages exist; although

very few are efficient and useful for biological applications. In these studies Bhc was shown to

be an ideal photo-cage for patterning in hydrogels since it is water soluble and versatile, being

able to cage both amines and thiols. Furthermore, the photodeprotection reaction occurs in

aqueous environments at physiological pH. Bhc is pH dependent where the hydroxyl must be

deprotonated, otherwise Bhc is non-fluorescent and will not undergo photo-deprotection. Under

108

physiological conditions Bhc remains active since more than 90% of the cage is ionized (pKa =

6.2) [204]. Recently, Kotsur et al. have synthesized a Bhc derivative,

bis(carboxymethyl)aminomethyl-substituted Bhc, with a pka of 4.9, thus increasing the pH range

of Bhc as a photocage [205]. Also, no cytotoxic effects were observed when RPCs or NSPCs

were cultured on top or within Bhc-hydrogels. After 2 weeks in culture, the gels remained

fluorescent in the coumarin channel indicating that the photolabile groups are stable. We did not

observe hydrolysis of Bhc caged amines or thiols when protected from light in our patterning

system even though Bhc caged esters were previously shown to undergo hydrolysis. This is not

suprising since esters are known to hydrolyze whereas carbamate (Bhc-amine) and sulphur-

carbon bonds (Bhc-thiol) are stable in aqueous conditions.

5.1.3 Two-photon patterning in aminocoumarin agarose

3D patterns of amines were created in agarose hydrogels using two-photon irradiation. The work

presented in Chapter 2 as well as that by Wosnick and Shoichet were the first demonstrations of

3D functional group patterning in hydrogels. The deprotection reaction of Bhc protected amines

was limited to the excitation volume of the laser resulting in boxes of free amines. Because

amines are photo-deprotected, the concentration of free amines within the patterns can be easily

controlled by varying the laser exposure. Although, high lateral resolution (xy plane) can be

achieved through two-photon irradiation [88], axial resolution is highly dependent on the optics.

To investigate axial resolution, boxes were scanned on a single plane in the gel and the z-axis

(axial) profile of deprotection was quantified by measuring the loss of Bhc fluorescence. The

deprotection profile along the z-axis had a Gaussian distribution and was confined to a ~40 µm

region. The resolution of the patterning system could be improved by using a lens with a higher

numerical aperture (NA). Although, increasing the NA of the lens will decrease the focal

109

distance and thus limit the size of hydrogels that can be patterned. Therefore, a trade-off is made

between resolution and working distance. All studies were performed using a 20x objective with

a NA of 0.4 since it provided the resolution needed for our applications and provided a large

working distance of ~ 15 mm.

The 3D deprotection of functional groups such as amines can control cells in 3D by encouraging

cell adhesion or acting as anchoring sites for biomolecules. The unveiling of primary amines will

result in positive charges, which can promote cell adhesion. Cells are negatively charged and will

preferentially interact with positively charged scaffolds [206]. Scaffolds with charges will also

absorb cell adhesive proteins from the media through ionic interactions, which further

encourages cell adhesion. For example, Nakanishi et al. showed that patterns of amines on 2D

surfaces can direct cell adhesion[207]. Amine bearing glass substrates were PEGylated using a

photocleavable linker. PEG was selectively removed from the surface through exposure to UV

light, thus creating patterned of free amines. The authors found that HeLa cells only attached in

regions containing free amines, confirming that patterns of amines can spatially direct cell

attachment. The work presented here (Chapter 2) has created a method for the 3D localization of

amines, which will allow for more complex studies to control the adhesion of cells in 3D.

Amines can also be used as anchoring sites for biomolecules. For instance, peptides with

carboxylic acids would react with the uncaged amines in the presence of a carbodiimide such as

EDC.

5.2 3D protein immobilization

The work discussed here provides a number of methods for the immobilization of proteins under

mild conditions. Previously, protein immobilization lacked spatial control were patterns usually

110

spanned hundreds of microns[2]. To increase spatial control, this work utilized two-photon

chemistry which resulted in the immobilization of proteins with micron precision as discussed in

the following sections. Furthermore, the use of biological binding partners has allowed for the

creation of complex patterns of multiple proteins that form under mild conditions, thus

preserving bioactivity.

The thiol patterning system developed by Wosnick and Shoichet was chosen for protein

immobilization to take advantage of both disulfide bond formation and the efficient click

reaction between thiols and maleimides [2]. Acceptor molecules for amines are typically more

susceptible to hydrolysis than thiol reactive molecules. Therefore, patterns of thiols in hydrogels

provide a more efficient method for protein immobilization than amine containing gels.

Furthermore, proteins usually contain numerous free amines that would compete or interfere

with immobilization reactions onto aminated agarose, whereas free protein thiols are much less

common. Thiols also provide other chemical grafting options beyond the ones described in this

work, increasing the versatility of the thiol patterning system. Biomolecule immobilization could

also be accomplished by reacting with iodoacetyl containing molecules or through thiolene click

reactions [26]. Unpublished work by the Shoichet group demonstrates the ability to pattern

iodoacetyl fluorescent molecules in hyaluronic (HA) acid hydrogels modified with Bhc caged

thiols. It should also be noted that the amount of protein immobilized was directly (linear

relationship) related to the amount of irradiation for most systems. Since the amount of protein

immobilized is controlled by the coumarin deprotection reaction, the photochemical reaction

showed linear dependence. This is not surprising since only a small fraction of the coumarin

groups were uncaged for protein immobilization (nM versus µM). Therefore the concentration

during the irradiation process did not change significantly resulting in a constant rate of

deprotection yielding the linear relationship between irradiation and protein immobilization.

111

5.2.1 Immobilization of proteins through disulfide bonds

This work immobilized FGF-2 through the formation of disulfide bonds, demonstrating that

proteins with free cysteines can be directly immobilized in a single step without chemical

modification. This is advantageous since the chemical modification of proteins can have a

detrimental effect on protein activity. For example, FGF-2 was found to be inactive when

modified with maleimide-Alexa 546 (data not shown). The fluorescent protein was not able to

proliferate NSPCs, whereas controls of unmodified FGF-2 increased the proliferation of the

cells. The system is also able to control the concentration of FGF-2 immobilized by controlling

the amount of Bhc deprotection by varying laser exposure. The system is also efficient since

proteins are immobilized in a single step without the need for any other reagents. FGF-2 was

simply immobilized by soaking the protein into pre-patterned hydrogels. In this case, the protein,

FGF-2, was not modified or exposed to chemical agents for immobilization. Other methods,

including those presented here and in the literature require either the modification of the protein,

or the use of crosslinkers such as photoinitiators. Disulfide bond patterning can also be extended

to proteins that do not contain free cysteines. Surface amines can be converted into thiols using

Traut’s reagent, or cysteines can be added recombinantly using molecular biology techniques.

5.2.2 Immobilization of maleimide molecules

Maleimide modified proteins, HSA, streptavidin and barnase, were immobilized as binding

partners for the physical immobilization of FGF-2, CNTF and SHH. The immobilization process

should be carried out at slightly acidic pH to avoid hydrolysis of the maleimide group. For

example, the immobilization of mal-streptavidin failed when it was soaked into gels at pH 7.4.

Whereas, the patterning was successful when the pH was lowered to 6.8, indicating that the

112

maleimide group is unstable at pH 7.4. Maleimides undergo ring-opening hydrolysis to an

unreactive product under basic conditions. Other than lowering the pH, hydrolysis could also be

prevented by using maleimide groups with greater water stability such as maleimides with

proximal aliphatic cyclohexane rings (SMCC). Alternatively, thiol reactive groups less

susceptible to hydrolysis, such as iodoacetyls, could be used.

5.2.3 Immobilization through physical interactions

Immobilization of proteins through physical interactions is a versatile and highly specific method

that occurs under mild conditions. Physical interactions have been traditionally used for the

immobilization of proteins onto surfaces. The work presented here takes advantage of those

methods for the creation of complex 3D protein patterns. The systems developed (HSA/ABD,

barnase/bartar and streptavidin/biotin) were designed to function with any protein, where

proteins of interest can be expressed with the appropriate binding partner at their N or C-

terminus. Furthermore, the physical interactions form very stable and specific complexes in

neutral buffers at room temperature, which minimizes potential bioactivity loss. Furthermore, the

molecular site of immobilization can be optimized for bioactivity by the placement of the

binding domain in the peptide sequence.

Strong physical binding interactions must be chosen carefully since not all are suitable for 3D

immobilization of proteins. For example, the 3D immobilization of EGF using the high affinity

(KD ~ 10-15M) leucine zippers ZE and ZR required extensive irradiation that resulted in poor

results, even though Tirrell et al. successfully immobilized GFP using the same binding pairs on

2D surfaces. EGF patterning was attempted by first immobilizing mal-ZE, followed by the

addition of EGF-ZR. Although EGF was immobilized using this system, excessive irradiation

was needed when compared to the other binding systems (HSA/ABD, barnase/barstar and

113

biotin/streptavidin). Scans of between 200 and 300 were necessary to detect any immobilized

EGF, whereas only about 10 scans were needed for the detection of proteins using the other

systems. Immobilization using the binding pair ZE and ZR is complicated by the fact that both

ZE and ZR form homodimers with KDs in the micromolar range. Therefore, some of the

immobilized ZE will exist in the homodimer form, limiting the number of available binding sites

for EGF-ZR. The system did result in the immobilization of EGF-ZR, although the amount of

irradiation needed was deemed too high for further studies.

Multi-valent binding partners increase the concentration of immobilized protein. Using proteins

with multiple binding sites such as streptavidin, results in higher concentrations of proteins for

the same amount of irradiation as compared to mono-valent systems. It is also possible to

increase the valency of other binding systems. For instance, dibarnase molecules where two

barnase sequences are fused together would increase the valency of the barnase-barstar system

from 1 to 2. Dibarnase molecules have previously been expressed and shown to bind two barstar

peptides [186]. Moreover, the use of multivalent systems will decrease the amount of irradiation

needed and thus the time required to construct patterned hydrogels. This is particularly important

for two-photon patterning techniques, which have long processing times since only a small

volume can be excited at a time. Therefore, the use of multivalent binding systems will allow the

construction of larger constructs.

Immobilization of proteins through physical interactions could be further investigated as a

method with temporal control. In this work strong physical interactions were used for long term

immobilization. It is possible to imagine using a similar system with weaker interactions where

the complex would disassemble over a period of hours, days or weeks. Therefore, proteins would

only be present for a pre-determined amount of time. This could prove useful to mimic

114

environments where proteins are only present for a certain amount of time or to study the

different signally characteristics between proteins that can and can’t be internalized.

5.2.4 Simultaneous Immobilization of Proteins

Using the binding pairs of barnase-barstar and streptavidin-biotin, CNTF and SHH were

simultaneously immobilized within distinct volumes in agarose hydrogels. After introducing

barnase and streptavidin into defined volumes, barstar-SHH and biotin-CNTF formed complexes

only with their specific binding partners. The patterns of SHH and CNTF corresponded perfectly

with the regions containing barnase and streptavidin, demonstrating that the system is able to

work in concert and is highly specific.

Simultaneous immobilization of proteins limits their exposure to harsh patterning conditions

limiting bioactivity loss. Previously, multiple molecules could only be patterned within the same

hydrogel through sequentially patterning. Consider the sequential immobilization of molecules A

and B. Molecule A is first photochemically patterned, the hydrogel is washed to remove excess

A, followed by the introduction of molecule B for photopatterning and a final washing step.

Therefore, molecule A would be exposed to multiple patterning and washing steps. This system

works well with stable molecules such as small molecules or short peptides; however, larger

proteins may lose bioactivity during the sequential patterning process. With that in mind, we

designed our patterning system with stable binding factors, barnase and streptavidin, that can

withstand sequential patterning for the simultaneous immobilization of SHH and CNTF. Barnase

is known to refold, and streptavidin has high thermal and chemical stability. Using this method,

the biologically relevant proteins, SHH and CNTF, are immobilized at the final stage and are not

exposed to multiple patterning and washing steps.

115

The ability to simultaneously immobilize proteins is crucial for the design of chemically defined

hydrogels. This will allow for the design of materials that can recapitulate in vivo environments,

and thus the investigation of complex biological phenomena. Because immobilization is only

dependent on the binding domain and not the biologically relevant protein, this system can be

predictably used for the immobilization of numerous proteins. Therefore, this system can be used

as a platform technology for the design of protein patterned hydrogels.

5.3 Bioactivity of immobilized factors

Many proteins remain active after immobilization, indicating that cellular uptake is not necessary

for all proteins. Before conducting protein patterning experiments, it is best to verify that the

proteins of interest remain active in the solid-phase form. Luckily, many proteins have been

immobilized on surfaces and literature reviews can usually resolve this issue. For example, all

the proteins used in this thesis (FGF2, SHH and CNTF) were previously shown to remain active.

Many other proteins have also been shown to remain active after immobilization including NGF

[208], PDGF [120], VEGF [191], EGF [209], and IFN-γ [75], demonstrating the breadth

immobilization can have in the biomaterials field. Furthermore, some proteins have higher

stability when immobilized compared to the soluble form. Protein grafting onto surfaces or

within hydrogels prevents aggregation and increases the thermal stability of the peptide [210].

In this study, the bioactivity of the immobilized forms of SHH and CNTF was confirmed. CNTF,

as expected, increased the phosphorylation of STAT3 in RPCs when compared to controls

without CNTF. The bioactivity of SHH was confirmed using both RPCs and NSPCs. RPCs

upregulated the transcription factor Gli2 in the presence of immobilized SHH. NSPCs responded

to a gradient of SHH, a chemoattractant for NSPCs, and migrated within patterns of the factor.

116

5.4 Migration into agarose hydrogels

Patterning systems require a method to introduce cells into the hydrogel after protein

immobilization. If cells were encapsulated within the hydrogel prior to patterning, they would be

exposed to soluble protein during the immobilization process. Encapsulation of cells is therefore

not realistic since the goal is to selectively signal cells with 3D protein patterns. Therefore,

strategies were investigated to introduce cells after the patterning process.

Cells can only migrate through non-proteolytic methods since agarose is not biodegradable. It

has previously been established that weak gels encourage migration [132], hence agarose

hydrogels with the lowest possible concentration were investigated. Low concentration gels

provide a lower physical barrier and thus encourage non-proteolytic cell migration by allowing

matrix deformation. The optimal agarose gel concentration was found to be 0.3 wt%; weaker

gels fell apart during patterning or cell culture.

Cells plated on top of hydrogels will migrate into channels of bio-factors. Luo and Shoichet

demonstrated that neurites could extend within agarose hydrogels containing channels modified

with adhesion sites. Since patterns of adhesive factors did not encourage retinal or neural stem

cell migration into agarose gels, patterns of a chemoattractant were investigated. As shown in

Figure 4-7, SHH gradients were able to encourage NSC migration into the gel. In this case, cells

penetrated approximately 100 µm into the gel, making this system useful in application requiring

thin biomaterials. Therefore, this system could be used towards retinal tissue engineering since

the retina is 100 to 130 µm thick in humans.

If greater cell migration is required, a combination of strategies will need to be pursued

concurrently. It is now evident that cells migrate through proteolytic degradation, matrix

117

deformation and biochemical cues. Hydrogels that allow for all three methods will result in the

greatest amount of cell migration and thus hydrogel penetration. Therefore, weak degradable

hydrogels with patterns of cell adhesive and chemoattractant molecules would be optimal for cell

migration of longer distances.

5.5 3D protein patterning and regenerative medicine

The ability to control the chemical environment around cells is important for stem cell biology,

especially for the design of an artificial stem cell niche. Stem cell therapies will require the

production of large quantities of cells. Although, current expansion techniques are limited to a

few passages since cells lose their multipotency over time. Therefore engineering an artificial

niche to keep stem cells multipotent and proliferative is very important. Our 3D patterning

system could be useful for the design of matrices to mimic the stem cell niche, which provides a

favorable environment for proliferation and maintenance of multipotency. Furthermore,

investigating the interplay between the ECM and stem cells will unlock fundamental biological

understanding.

The most exciting applications for patterned scaffolds are in tissue engineering. The ability to

create constructs to replace or repair tissue would be transformative in a number of diseases.

Current research in the field uses hydrogels in vascular reconstruction [45], retina regeneration

[211], nerve repair [212], and bone formation [213]. The critical step for biomaterial synthesis is

to mimic the natural environment such that cells behave in a controlled and predictable fashion.

Our system has the ability to chemically modify hydrogels mimicking the complexity of the

natural ECM, thus providing a method to make complex biomaterials to control cell fate.

118

6 Conclusions

6.1 Achievements of Objectives

The original hypothesis for this work was:

Multiple bioactive growth factors can be three-dimensionally and simultaneously immobilized

within a hydrogel scaffold.

Several objectives were set to test this hypothesis and are revisited here with a summary of the

work presented to meet these objectives.

1. Design a matrix for 3D immobilization of molecules

A three-dimensional patterning system was designed within agarose hydrogels using two-photon

chemistry. Agarose was modified to contain amines protected with the photolabile group 6-

bromo-7-hydroxycoumarin. Upon irradiation the protecting group is removed to yield amines

within defined volumes. The reaction was visualized by reacting the uncaged amines with the

dye CBQCA; the reaction of CBQCA with primary amines forms a distinct fluorescent molecule.

Because of the small volume associated with two-photon excitation, the deprotection was 3D

controlled with micron precision. Therefore, a matrix was designed where molecules can be

immobilized with 3D control. These data were presented in Chapter 2 and published in the

Journal of Materials Chemistry[112].

2. 3D immobilize proteins with a matrix

FGF-2 was 3D immobilized in agarose hydrogels through disulfide bonds and protein binding

pairs. The immobilization of proteins was achieved using a similar system as described in

objective 1, except thiols were protected with 6-bromo-7-hydroxycoumarin instead of amines.

119

Taking advantage of the 3D deprotection of thiols the protein FGF-2 was directly immobilized

through disulfide bonds. FGF-2 contains two free cysteines not necessary for bioactivity, thus

making it amenable to disulfide bond formation with thiolated agarose. Since this method is only

useful for proteins with cysteines, a second universal method was developed using the physical

interaction between human serum albumin (HSA) and an albumin binding domain (ABD).

Immobilization was achieved my first immobilizing HSA followed by the addition of FGF-2 as a

fusion protein with ABD. The fusion protein bioactivity was similar to FGF-2, demonstrating

that the incorporation of binding partners does not influence activity. The use of physical

interactions provides an immobilization strategy that functions under mild conditions while

being applicable to a number of proteins. The stability of the HSA-ABD complex was confirmed

by following the change in fluorescence of a FGF-ABD pattern over time. These data were

presented in Chapter 2 and submitted to Biomacromolecules.

3. Immobilization of 2 differentiation factors simultaneously in a 3D matrix

Sonic hedgehog (SHH) and ciliary neurotrophic factor (CNTF) were simultaneously

immobilized using the physical interactions between barnase/barstar and streptavidin/biotin,

respectively. These interactions were chosen since they are capable of working simultaneously

and under mild conditions (PBS). Furthermore, barnase and streptavidin are stable proteins

making them ideal candidates for photochemical patterning. Hydrogels were prepared for

immobilization by first patterning barnase and streptavidin in distinct volumes. Fusion proteins

of SHH-barstar of CNTF-biotin were then added and immobilized through physical interactions

with the aforementioned binding partners. This demonstrates that two separate proteins can be

simultaneously immobilized in distinct volumes using strong physical interactions. The

immobilized factors were also determined to be bioactive, SHH upregulated the transcription

factor Gli2 and CNTF resulted in STAT3 phosphorylation in RPCs as expected. SHH activity

120

was further demonstrated as a chemoattractant for the migration of NSPCs into the patterned

hydrogel. These data were presented in Chapter 4 and accepted for publication in Nature

Materials.

6.2 Major contributions

In this thesis, it was demonstrated that complex proteins patterns could be incorporated into

agarose hydrogels for applications in tissue engineering. To this end, a two-photon patterning

system was developed where reactive groups are protected with a multi-photon active

photolabile group, which is removed upon excitation. The photo-deprotection reaction is

confined to the focal point of the laser. Furthermore, the amount of deprotection can be

controlled by varying the amount of irradiation.

The patterning system served as the basis for the controlled immobilization of proteins under

mild conditions. Proteins were either immobilized through disulfide bonds or physical

interactions. The patterning process provides techniques for immobilization under mild

conditions, without the need for reactive chemicals such as crosslinkers. The mild nature

provides an environment that minimizes protein bioactivity loss and cellular toxicity.

Furthermore, the photochemistry allows for the creation of complex protein patterns with

varying concentrations.

The combination of photochemistry and protein engineering allowed for an orthogonal

patterning system applicable to many proteins. The system is versatile since any protein can be

immobilized by the recombinant incorporation of a binding partner. The use of binding partners,

barnase/barstar and streptavidin/biotin, allowed for the simultaneous immobilization of proteins

preventing the exposures of signaling proteins to multiples steps associated with sequential

121

patterning to limit bioactivity loss. Moreover, the systems can control the concentration and

location of the immobilized proteins independently.

122

7 Recommendations for future work

7.1 Protein patterning in different hydrogels

The protein patterning system was designed as a tool for the creation of matrices to study and

control cellular behaviour in 3D. Therefore, the system was developed to be translatable into

other hydrogels. This is important since the content of the ECM varies between tissues, and thus

different scaffolds will be needed to mimic different tissues. For instance, connective tissue

contains a large concentration of collagen whereas the ECM of the brain has a high proportion of

proteoglycans. In order to create representative scaffolds of various tissues, it would be

interesting to test the patterning system in various hydrogels.

Hydrogels must have the following criteria for 3D patterning: 1) transparent, 2) stable and 3)

minimal non-specific adsorption of proteins. Hydrogels must be transparent for the wavelength

of irradiation for efficient excitation of Bhc. Second, the gels must not degrade during the

patterning process, meaning they must be stable in buffers near neutral pH. Proteins must not

adsorb to the hydrogel as this would greatly hinder the ability to create patterns. Charged

hydrogels that are known to absorb proteins, such as laminin, through non-covalent interactions

must be avoided or washing procedures must be established. Recent work in the Shoichet group

has used a similar system as presented here for the patterning of proteins in negatively charged

hyaluronic acid gels. Although, washing times had to be increased to remove all the non-

immobilized proteins from the gel.

123

7.2 Increased migration coupled with matrix degradation

Hydrogels other than agarose could also prove advantageous for increased cell penetration into

the gels. Agarose is non-degradable limiting the cells ability to remodel the matrix to promote

cell migration. Weak degradable matrixes allow for proteolytic and non-proteolytic migration. In

this case, cells would be able to respond to chemoattractants (as NSPCs did with SHH) while

degrading the scaffold and lowering the physical barrier imposed by the matrix.

Matrix degradation could interfere with 3D patterning since it would degrade patterned regions

turning immobilized proteins into soluble proteins; strategies must be developed to overcome

this fundamental problem. One possibility would be to use a semi-degradable matrix, where

some crosslinks are cleaved to decrease gel stiffness but the immobilized proteins remain

immobilized to the matrix. This could be accomplished by crosslinking a gel with two different

molecules with only one being degradable. For instance, PEG hydrogels could be crosslinked

with a mixture of MMP degradable peptides and non-degradable peptides. In this system, cell

secreted enzymes or an external stimulus would cleave the degradable crosslinks producing a

weaker gel more amenable to cell migration without losing the 3D patterns in the gel. The effect

of hydrogel swelling on patterns after partial degradation would need to be characterized.

Anseth et al. have designed a system for the controlled degradation of hydrogels. This system

utilized a photo-degradable crosslinker, thereby controlling matrix degradation with a focused

laser. The method has the advantage of controlling the location, degree and time of degradation. I

believe a similar system combined with our protein patterning system would be advantageous for

applications in tissue engineering. Not only would 3D patterned proteins influence cellular

activity such as differentiation but cell migration could also be 3D controlled by selectively

124

degrading the matrix. In other words, the placement of cells and their activity would be 3D

controlled.

A combined method of photochemical and proteolytic degradation would externally control the

location of degradation while the rate of degradation would be governed by cells. This system

would yield spatial control over migration, while allowing the cells to remodel the matrix at their

own pace, which mimic’s native proteolytic migration. In this case, proteolytic sites incorporated

into the hydrogel would be caged with a photolabile group. The sequence would only be

recognized by the corresponding protease after photo-uncaging; only irradiated volumes could be

proteolytically degraded. Therefore, the cellular migration is spatially controlled by

photodeprotection but the rate of degradation is controlled by the interaction between the cells in

the matrix. The materials described above would yield spatial control over cell migration, which

is advantageous to guide cellular localization.

7.3 Three-dimensional differentiation of stem cells

As mentioned in Chapter 1, a long-term goal of the project is to spatially guide stem cell

differentiation in 3D patterned hydrogels for tissue engineering applications. Therefore, cells

must be seeded into hydrogels patterned with multiple growth factors in distinct volumes. The

cells would then differentiate corresponding to their biochemical environment. For example,

NSPCs within a PDGF-AA pattern would preferentially differentiate into oligodentrocytes[120].

Stem cell migration into growth factor patterns is the next step towards 3D stem cell

differentiation. Cell penetration can be achieved using gradients of chemoattractants,

demonstrated in Chapter 4, or techniques as discussed above. As mentioned above, cell

migration or matrix degradation must not destroy the growth factor patterns before stem cell

125

differentiation. Therefore, elegant systems must be developed where the lifetime of the patterns

are sufficient for stem cell differentiation.

Immunohistochemistry can be used to follow the differentiation profile of stem cells exposed to

patterns of immobilized growth factor to determined the optimal chemical environment. Staining

in 3D hydrogels is not trivial, and requires many troubleshooting steps. For example, the

incubation time of antibodies will need to be increased to account for diffusion rates within the

hydrogel. Once cells are seeded into patterns and staining techniques are established, the exact

chemical environment necessary for differentiation must be elucidated. At first, we will need to

determine which factors are needed and at what concentration. Since the patterning system

controls the immobilized concentration, stem cell differentiation can be easily determined as a

function of concentration of different factors. Furthermore, the versatility of the system allows

for the immobilization of various proteins for efficient screening of protein combinations. After

successful spatial differentiation, this technology can be applied to the design of complex cellular

structures for applications in tissue engineering.

 

126

8 References

1. Owen, S.C. and M.S. Shoichet, Design of three dimensional biomimetic scaffolds. Journal of Biomedical Materials Research Part A, 2010. 94(4): p. 1321-1331.

2. Luo, Y. and M.S. Shoichet, A photolabile hydrogel for guided three-dimensional cell growth and migration. Nature materials, 2004. 3(4): p. 249-253.

3. Harrison, R.G., The reaction of embryonic cells to solid structures. Journal of Experimental Zoology, 1914. 17(4): p. 521-544.

4. Carleton, H., Tissue culture: A critical summary. Journal of Experimental Biology, 1923. 1(1): p. 131.

5. Strady, A., et al., Antibody persistence following preexposure regimens of cell-culture rabies vaccines: 10-year follow-up and proposal for a new booster policy. Journal of Infectious Diseases, 1998. 177(5): p. 1290.

6. Chu, L. and D.K. Robinson, Industrial choices for protein production by large-scale cell culture. Current opinion in biotechnology, 2001. 12(2): p. 180-187.

7. Wurm, F.M., Production of recombinant protein therapeutics in cultivated mammalian cells. Nature biotechnology, 2004. 22(11): p. 1393-1398.

8. Engler, A.J., et al., Matrix elasticity directs stem cell lineage specification. Cell, 2006. 126(4): p. 677-689.

9. Discher, D.E., P. Janmey, and Y. Wang, Tissue cells feel and respond to the stiffness of their substrate. Science, 2005. 310(5751): p. 1139.

10. Fu, J., et al., Mechanical regulation of cell function with geometrically modulated elastomeric substrates. Nat Meth, 2010. 7(9): p. 733-736.

11. McBeath, R., et al., Cell shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment. Developmental cell, 2004. 6(4): p. 483-495.

12. Tibbitt, M.W. and K.S. Anseth, Hydrogels as extracellular matrix mimics for 3D cell culture. Biotechnology and bioengineering, 2009. 103(4): p. 655-663.

13. Fon, D., et al., Tissue Engineering of Organs: Brain Tissues. Tissue Engineering, 2011: p. 457-492.

14. Geckil, H., et al., Engineering hydrogels as extracellular matrix mimics. Nanomedicine, 2010. 5(3): p. 469-484.

15. Hynes, R.O., The extracellular matrix: not just pretty fibrils. Science, 2009. 326(5957): p. 1216.

127

16. Rehfeldt, F., et al., Cell responses to the mechanochemical microenvironment--implications for regenerative medicine and drug delivery. Advanced drug delivery reviews, 2007. 59(13): p. 1329-1339.

17. Hay, E.D., Cell biology of extracellular matrix. 1991: Springer Us.

18. Ruoslahti, E., Brain extracellular matrix. Glycobiology, 1996. 6(5): p. 489-492.

19. Li, J., Y.P. Zhang, and R.S. Kirsner, Angiogenesis in wound repair: angiogenic growth factors and the extracellular matrix. Microscopy research and technique, 2003. 60(1): p. 107-114.

20. Dinbergs, I.D., L. Brown, and E.R. Edelman, Cellular response to transforming growth factor- 1 and basic fibroblast growth factor depends on release kinetics and extracellular matrix interactions. Journal of Biological Chemistry, 1996. 271(47): p. 29822.

21. Gospodarowicz, D., D. Delgado, and I. Vlodavsky, Permissive effect of the extracellular matrix on cell proliferation in vitro. Proceedings of the National Academy of Sciences of the United States of America, 1980. 77(7): p. 4094.

22. Marastoni, S., et al., Extracellular matrix: a matter of life and death. Connective Tissue Research, 2008. 49(3-4): p. 203-206.

23. Samantaray, S., et al., Peptide−Sugar Ligation Catalyzed by Transpeptidase Sortase:  A Facile Approach to Neoglycoconjugate Synthesis. Journal of the American Chemical Society, 2008. 130(7): p. 2132-2133.

24. Reing, J.E., et al., Degradation products of extracellular matrix affect cell migration and proliferation. Tissue Engineering Part A, 2008. 15(3): p. 605-614.

25. Badylak, S.F., D.O. Freytes, and T.W. Gilbert, Extracellular matrix as a biological scaffold material: structure and function. Acta biomaterialia, 2009. 5(1): p. 1-13.

26. DeForest, C.A., B.D. Polizzotti, and K.S. Anseth, Sequential click reactions for synthesizing and patterning three-dimensional cell microenvironments. Nature materials, 2009. 8(8): p. 659-664.

27. Gillies, G., et al., A spinal cord surrogate with nanoscale porosity for in vitro simulations of restorative neurosurgical techniques. Nanotechnology, 2002. 13: p. 587.

28. Luo, Y. and M.S. Shoichet, Light-activated immobilization of biomolecules to agarose hydrogels for controlled cellular response. Biomacromolecules, 2004. 5(6): p. 2315-2323.

29. Kim, H.D. and R.F. Valentini, Retention and activity of BMP 2 in hyaluronic acid based scaffolds in vitro. Journal of biomedical materials research, 2002. 59(3): p. 573-584.

128

30. Aigner, J., et al., Cartilage tissue engineering with novel nonwoven structured biomaterial based on hyaluronic acid benzyl ester. Journal of biomedical materials research, 1998. 42(2): p. 172-181.

31. Choi, Y.S., et al., Studies on gelatin containing artificial skin: II. Preparation and characterization of cross linked gelatin hyaluronate sponge. Journal of biomedical materials research, 1999. 48(5): p. 631-639.

32. Hu, M., et al., Three dimensional hyaluronic acid grafts promote healing and reduce scar formation in skin incision wounds. Journal of Biomedical Materials Research Part B: Applied Biomaterials, 2003. 67(1): p. 586-592.

33. Duranti, F., et al., Injectable hyaluronic acid gel for soft tissue augmentation: A clinical and histological study. Dermatologic surgery, 1998. 24(12): p. 1317-1325.

34. Alsberg, E., et al., Cell-interactive alginate hydrogels for bone tissue engineering. Journal of Dental Research, 2001. 80(11): p. 2025.

35. Rowley, J.A., G. Madlambayan, and D.J. Mooney, Alginate hydrogels as synthetic extracellular matrix materials. Biomaterials, 1999. 20(1): p. 45-53.

36. Dhoot, N.O., et al., Peptide modified alginate surfaces as a growth permissive substrate for neurite outgrowth. Journal of Biomedical Materials Research Part A, 2004. 71(2): p. 191-200.

37. Yu, L.M.Y., K. Kazazian, and M.S. Shoichet, Peptide surface modification of methacrylamide chitosan for neural tissue engineering applications. Journal of Biomedical Materials Research Part A, 2007. 82(1): p. 243-255.

38. Li, J., et al., Investigation of MC3T3-E1 cell behavior on the surface of GRGDS-coupled chitosan. Biomacromolecules, 2006. 7(4): p. 1112-1123.

39. Mochizuki, M., et al., Integrin dependent cell behavior on ECM peptide conjugated chitosan membranes. Peptide Science, 2007. 88(2): p. 122-130.

40. Elçin, M., V. Dixit, and G. Gitnic, Hepatocyte attachment on biodegradable modified chitosan membranes: in vitro evaluation for the development of liver organoids. Artificial organs, 1998. 22(10): p. 837-846.

41. Ma, L., et al., Collagen/chitosan porous scaffolds with improved biostability for skin tissue engineering. Biomaterials, 2003. 24(26): p. 4833-4841.

42. Park, Y., et al., Bovine primary chondrocyte culture in synthetic matrix metalloproteinase-sensitive poly (ethylene glycol)-based hydrogels as a scaffold for cartilage repair. Tissue Engineering, 2004. 10(3-4): p. 515-522.

43. Burdick, J.A. and K.S. Anseth, Photoencapsulation of osteoblasts in injectable RGD-modified PEG hydrogels for bone tissue engineering. Biomaterials, 2002. 23(22): p. 4315-4323.

129

44. Mahoney, M.J. and K.S. Anseth, Three-dimensional growth and function of neural tissue in degradable polyethylene glycol hydrogels. Biomaterials, 2006. 27(10): p. 2265-2274.

45. Seliktar, D., et al., MMP 2 sensitive, VEGF bearing bioactive hydrogels for promotion of vascular healing. Journal of Biomedical Materials Research Part A, 2004. 68(4): p. 704-716.

46. Bini, T., et al., Poly (l-lactide-co-glycolide) biodegradable microfibers and electrospun nanofibers for nerve tissue engineering: an in vitro study. Journal of materials science, 2006. 41(19): p. 6453-6459.

47. Uematsu, K., et al., Cartilage regeneration using mesenchymal stem cells and a three-dimensional poly-lactic-glycolic acid (PLGA) scaffold. Biomaterials, 2005. 26(20): p. 4273-4279.

48. Holy, C.E., M.S. Shoichet, and J.E. Davies, Engineering three-dimensional bone tissue in vitro using biodegradable scaffolds: investigating initial cell-seeding density and culture period. Journal of biomedical materials research, 2000. 51(3): p. 376-382.

49. Leipzig, N.D. and M.S. Shoichet, The effect of substrate stiffness on adult neural stem cell behavior. Biomaterials, 2009. 30(36): p. 6867-6878.

50. Saha, K., et al., Substrate modulus directs neural stem cell behavior. Biophysical journal, 2008. 95(9): p. 4426-4438.

51. Helary, C., et al., Concentrated collagen hydrogels as dermal substitutes. Biomaterials, 2010. 31(3): p. 481-490.

52. Suri, S. and C.E. Schmidt, Cell-laden hydrogel constructs of hyaluronic acid, collagen, and laminin for neural tissue engineering. Tissue Engineering Part A, 2010. 16(5): p. 1703-1716.

53. Seidlits, S.K., et al., Fibronectin-hyaluronic acid composite hydrogels for three-dimensional endothelial cell culture. Acta biomaterialia, 2011.

54. DeLong, S.A., J.J. Moon, and J.L. West, Covalently immobilized gradients of bFGF on hydrogel scaffolds for directed cell migration. Biomaterials, 2005. 26(16): p. 3227-3234.

55. Lee, S.-H., et al., Proteolytically Degradable Hydrogels with a Fluorogenic Substrate for Studies of Cellular Proteolytic Activity and Migration. Biotechnology Progress, 2005. 21(6): p. 1736-1741.

56. Rahman, N., et al., The Use of Vascular Endothelial Growth Factor Functionalized Agarose to Guide Pluripotent Stem Cell Aggregates Toward Blood Progenitor Cells. Biomaterials, 2010.

57. Valeur, E. and M. Bradley, Amide bond formation: beyond the myth of coupling reagents. Chem. Soc. Rev., 2008. 38(2): p. 606-631.

130

58. Sletten, E.M. and C.R. Bertozzi, Bioorthogonal chemistry: fishing for selectivity in a sea of functionality. Angewandte Chemie International Edition, 2009. 48(38): p. 6974-6998.

59. Rusmini, F., Z. Zhong, and J. Feijen, Protein immobilization strategies for protein biochips. Biomacromolecules, 2007. 8(6): p. 1775-1789.

60. Wong, L.S., F. Khan, and J. Micklefield, Selective covalent protein immobilization: strategies and applications. Chemical reviews, 2009. 109(9): p. 4025-4053.

61. Werle, M. and M. Hoffer, Glutathione and thiolated chitosan inhibit multidrug resistance P-glycoprotein activity in excised small intestine. Journal of controlled release, 2006. 111(1-2): p. 41-46.

62. Nandivada, H., X. Jiang, and J. Lahann, Click chemistry: versatility and control in the hands of materials scientists. Adv. Mater, 2007. 19(17): p. 2197-2208.

63. Dameron, C.T. and M.D. Harrison, Mechanisms for protection against copper toxicity. The American journal of clinical nutrition, 1998. 67(5): p. 1091S.

64. Johnson, J.A., et al., Copper-free click chemistry for the in situ crosslinking of photodegradable star polymers. Chem. Commun., 2008(26): p. 3064-3066.

65. Nimmo, C.M., S.C. Owen, and M.S. Shoichet, Diels- Alder Click Cross-Linked Hyaluronic Acid Hydrogels for Tissue Engineering. Biomacromolecules, 2011.

66. Shi, M., et al., Immuno Polymeric Nanoparticles by Diels–Alder Chemistry. Angewandte Chemie, 2007. 119(32): p. 6238-6243.

67. Dillmore, W.S., M.N. Yousaf, and M. Mrksich, A photochemical method for patterning the immobilization of ligands and cells to self-assembled monolayers. Langmuir, 2004. 20(17): p. 7223-7231.

68. Kosif, I., et al., Fabrication of Maleimide Containing Thiol Reactive Hydrogels via Diels- Alder/Retro-Diels- Alder Strategy. Macromolecules, 2010. 43(9): p. 4140-4148.

69. Hu, B.H., J. Su, and P.B. Messersmith, Hydrogels cross-linked by native chemical ligation. Biomacromolecules, 2009. 10(8): p. 2194-2200.

70. Hermanson, G.T., Bioconjugate techniques. 1996: Academic Pr.

71. DeForest, C.A., E.A. Sims, and K.S. Anseth, Peptide-Functionalized Click Hydrogels with Independently Tunable Mechanics and Chemical Functionality for 3D Cell Culture. Chemistry of Materials, 2003.

72. Saleemuddin, M., Bioaffinity based immobilization of enzymes. Thermal Biosensors, Bioactivity, Bioaffinitty, 1999: p. 203-226.

73. Weber, P.C., et al., Structural origins of high-affinity biotin binding to streptavidin. Science, 1989. 243(4887): p. 85.

131

74. González, M., et al., Interaction of biotin with streptavidin. Journal of Biological Chemistry, 1997. 272(17): p. 11288.

75. Leipzig, N.D., et al., Differentiation of neural stem cells in three-dimensional growth factor-immobilized chitosan hydrogel scaffolds. Biomaterials, 2011. 32(1): p. 57-64.

76. Cull, M.G. and P.J. Schatz, [26] Biotinylation of proteins in vivo and in vitro using small peptide tags. Methods in Enzymology, 2000. 326: p. 430-440.

77. Zhang, K., M.R. Diehl, and D.A. Tirrell, Artificial Polypeptide Scaffold for Protein Immobilization. Journal of the American Chemical Society, 2005. 127(29): p. 10136-10137.

78. Wong, L.S., J. Thirlway, and J. Micklefield, Direct Site-Selective Covalent Protein Immobilization Catalyzed by a Phosphopantetheinyl Transferase. Journal of the American Chemical Society, 2008. 130(37): p. 12456-12464.

79. Clow, F., J. Fraser, and T. Proft, Immobilization of proteins to biacore sensor chips using &lt;i&gt;Staphylococcus aureus&lt;/i&gt; sortase A. Biotechnology Letters, 2008. 30(9): p. 1603-1607.

80. Belfield, K.; Available from: http://belfield.cos.ucf.edu/one%20vs%20two-photon%20excitation.html.

81. Xu, C. and W.W. Webb, Measurement of two-photon excitation cross sections of molecular fluorophores with data from 690 to 1050 nm. Journal of the Optical Society of America-B-Optical Physics, 1996. 13(3): p. 481-491.

82. Belfield, K.D., et al., Synthesis of new two-photon absorbing fluorene derivatives via Cu-mediated Ullmann condensations. The Journal of Organic Chemistry, 2000. 65(15): p. 4475-4481.

83. Helmchen, F. and W. Denk, Deep tissue two-photon microscopy. Nature Methods, 2005. 2(12): p. 932-940.

84. Davidson, M.; Available from: http://micro.magnet.fsu.edu/primer/techniques/fluorescence/multiphoton/multiphotonintro.html.

85. Furuta, T., et al., Brominated 7-hydroxycoumarin-4-ylmethyls: photolabile protecting groups with biologically useful cross-sections for two photon photolysis. Proceedings of the National Academy of Sciences of the United States of America, 1999. 96(4): p. 1193.

86. Lakowicz, J.R. and B.R. Masters, Principles of fluorescence spectroscopy. Journal of Biomedical Optics, 2008. 13: p. 029901.

87. Seidlits, S.K., C.E. Schmidt, and J.B. Shear, High Resolution Patterning of Hydrogels in Three Dimensions using Direct Write Photofabrication for Cell Guidance. Advanced Functional Materials, 2009. 19(22): p. 3543-3551.

132

88. Wosnick, J.H. and M.S. Shoichet, Three-dimensional chemical patterning of transparent hydrogels. Chemistry of Materials, 2007. 20(1): p. 55-60.

89. Li, L. and J.T. Fourkas, Multiphoton polymerization. Materials Today, 2007. 10(6): p. 30-37.

90. Yagci, Y., S. Jockusch, and N.J. Turro, Photoinitiated polymerization: advances, challenges, and opportunities. Macromolecules, 2010.

91. Fouassier, J.P., et al., Dyes as Photoinitiators or Photosensitizers of Polymerization Reactions. Materials, 2010. 3(12): p. 5130-5142.

92. Schafer, K.J., et al., Two-photon absorption cross-sections of common photoinitiators. Journal of Photochemistry and Photobiology A: Chemistry, 2004. 162(2-3): p. 497-502.

93. Penzkofer, A. and W. Leupacher, S 0-S 1 two photon absorption dynamics of organic dye solutions. Optical and quantum electronics, 1987. 19(6): p. 327-349.

94. Campagnola, P.J., et al., 3-dimensional submicron polymerization of acrylamide by multiphoton excitation of xanthene dyes. Macromolecules, 2000. 33(5): p. 1511-1513.

95. Armentrout, R.S. and C.L. McCormick, Water soluble polymers. 76. Electrolyte responsive cyclocopolymers with sulfobetaine units exhibiting polyelectrolyte or polyampholyte behavior in aqueous media. Macromolecules, 2000. 33(2): p. 419-424.

96. Warther, D., et al., Two-photon uncaging: New prospects in neuroscience and cellular biology. Bioorganic & Medicinal Chemistry, 2010. 18(22): p. 7753-7758.

97. Momotake, A., et al., The nitrodibenzofuran chromophore: a new caging group for ultra-efficient photolysis in living cells. Nature Methods, 2006. 3(1): p. 35-40.

98. Zhao, Y.R., et al., New caged coumarin fluorophores with extraordinary uncaging cross sections suitable for biological imaging applications. Journal of the American Chemical Society, 2004. 126(14): p. 4653-4663.

99. Lu, M., et al., Bhc-diol as a photolabile protecting group for aldehydes and ketones. Organic Letters, 2003. 5(12): p. 2119-2122.

100. Fedoryak, O.D. and T.M. Dore, Brominated hydroxyquinoline as a photolabile protecting group with sensitivity to multiphoton excitation. Organic Letters, 2002. 4(20): p. 3419-3422.

101. Zhu, Y., et al., 8-Bromo-7-hydroxyquinoline as a photoremovable protecting group for physiological use: mechanism and scope. Journal of the American Chemical Society, 2006. 128(13): p. 4267-4276.

102. Gug, S., et al., Photolabile Glutamate Protecting Group with High One and Two Photon Uncaging Efficiencies. ChemBioChem, 2008. 9(8): p. 1303-1307.

133

103. Hoffmann, J.C. and J.L. West, Three-dimensional photolithographic patterning of multiple bioactive ligands in poly (ethylene glycol) hydrogels. Soft Matter, 2010. 6(20): p. 5056-5063.

104. Basu, S., et al., Multiphoton excited fabrication of collagen matrixes cross-linked by a modified benzophenone dimer: Bioactivity and enzymatic degradation. Biomacromolecules, 2005. 6(3): p. 1465-1474.

105. Balasubramanian, D., X. Du, and J.S. Zigler, THE REACTION OF SINGLET OXYGEN WITH PROTEINS, WITH SPECIAL REFERENCE TO CRYSTALLINS. Photochemistry and Photobiology, 1990. 52(4): p. 761-768.

106. Pins, G.D., et al., Multiphoton excited fabricated nano and micro patterned extracellular matrix proteins direct cellular morphology. Journal of Biomedical Materials Research Part A, 2006. 78A(1): p. 194-204.

107. Pins, G.D., et al., Multiphoton excited fabricated nano and micro patterned extracellular matrix proteins direct cellular morphology. Journal of Biomedical Materials Research Part A, 2006. 78(1): p. 194-204.

108. Pitts, J.D., et al., New Photoactivators for Multiphoton Excited Three dimensional Submicron Cross linking of Proteins: Bovine Serum Albumin and Type 1 Collagen¶†. Photochemistry and Photobiology, 2002. 76(2): p. 135-144.

109. Chen, X., et al., Adhesion and migration of ovarian cancer cells on crosslinked laminin fibers nanofabricated by multiphoton excited photochemistry. Integrative Biology, 2009. 1(7): p. 469-476.

110. Lee, S.H., J.J. Moon, and J.L. West, Three-dimensional micropatterning of bioactive hydrogels via two-photon laser scanning photolithography for guided 3D cell migration. Biomaterials, 2008. 29(20): p. 2962-2968.

111. Leslie-Barbick, J.E., et al., Micron-Scale Spatially Patterned, Covalently Immobilized Vascular Endothelial Growth Factor on Hydrogels Accelerates Endothelial Tubulogenesis and Increases Cellular Angiogenic Responses. Tissue Engineering Part A, 2010. 17(1-2): p. 221-229.

112. Wylie, R.G. and M.S. Shoichet, Two-photon micropatterning of amines within an agarose hydrogel. Journal of Materials Chemistry, 2008. 18(23): p. 2716-2721.

113. Wosnick, J.H. and M.S. Shoichet, Three-dimensional chemical Patterning of transparent hydrogels. Chemistry of Materials, 2008. 20(1): p. 55-60.

114. Reynolds, B.A. and S. Weiss, Generation of neurons and astrocytes from isolated cells of the adult mammalian central nervous system. Science, 1992. 255(5052): p. 1707.

115. Roy, N.S., et al., In vitro neurogenesis by progenitor cells isolated from the adult human hippocampus. Nature medicine, 2000. 6(3): p. 271-277.

134

116. Weiss, S., et al., Multipotent CNS stem cells are present in the adult mammalian spinal cord and ventricular neuroaxis. The Journal of neuroscience, 1996. 16(23): p. 7599.

117. Morshead, C.M., et al., Neural stem cells in the adult mammalian forebrain: a relatively quiescent subpopulation of subependymal cells. Neuron, 1994. 13(5): p. 1071-1082.

118. Palmer, T.D., J. Takahashi, and F.H. Gage, The Adult Rat Hippocampus Contains Primordial Neural Stem Cells* 1. Molecular and Cellular Neuroscience, 1997. 8(6): p. 389-404.

119. Wong, G., Y. Goldshmit, and A.M. Turnley, Interferon-[gamma] but not TNF [alpha] promotes neuronal differentiation and neurite outgrowth of murine adult neural stem cells. Experimental neurology, 2004. 187(1): p. 171-177.

120. Aizawa, Y., et al., The effect of immobilized platelet derived growth factor AA on neural stem/progenitor cell differentiation on cell-adhesive hydrogels. Biomaterials, 2008. 29(35): p. 4676-4683.

121. Hermanson, O., K. Jepsen, and M.G. Rosenfeld, N-CoR controls differentiation of neural stem cells into astrocytes. Nature, 2002. 419(6910): p. 934-939.

122. Tropepe, V., et al., Retinal stem cells in the adult mammalian eye. Science, 2000. 287(5460): p. 2032.

123. Coles, B.L.K., et al., Facile isolation and the characterization of human retinal stem cells. Proceedings of the National Academy of Sciences of the United States of America, 2004. 101(44): p. 15772.

124. Zahir, T., H. Klassen, and M.J. Young, Effects of ciliary neurotrophic factor on differentiation of late retinal progenitor cells. Stem Cells, 2005. 23(3): p. 424-432.

125. Bhattacharya, S., et al., Ciliary Neurotrophic Factor Mediated Signaling Regulates Neuronal Versus Glial Differentiation of Retinal Stem Cells/Progenitors by Concentration Dependent Recruitment of Mitogen Activated Protein Kinase and Janus Kinase Signal Transducer and Activator of Transcription Pathways in Conjunction with Notch Signaling. Stem Cells, 2008. 26(10): p. 2611-2624.

126. Goureau, O., K.D. Rhee, and X.J. Yang, Ciliary neurotrophic factor promotes muller glia differentiation from the postnatal retinal progenitor pool. Developmental neuroscience, 2004. 26(5-6): p. 359-370.

127. Levine, E.M., et al., Sonic hedgehog promotes rod photoreceptor differentiation in mammalian retinal cells in vitro. The Journal of neuroscience, 1997. 17(16): p. 6277.

128. Du, Y., Q. Xiao, and H.K. Yip, Regulation of Retinal Progenitor Cell Differentiation by Bone Morphogenetic Protein 4 Is Mediated by the Smad/Id Cascade. Investigative Ophthalmology & Visual Science, 2010. 51(7): p. 3764.

135

129. Lee, S.H., et al., Poly (ethylene glycol) hydrogels conjugated with a collagenase-sensitive fluorogenic substrate to visualize collagenase activity during three-dimensional cell migration. Biomaterials, 2007. 28(20): p. 3163-3170.

130. Raeber, G., M. Lutolf, and J. Hubbell, Molecularly engineered PEG hydrogels: a novel model system for proteolytically mediated cell migration. Biophysical journal, 2005. 89(2): p. 1374-1388.

131. Bellamkonda, R., et al., Hydrogel based three dimensional matrix for neural cells. Journal of biomedical materials research, 1995. 29(5): p. 663-671.

132. Ehrbar, M., et al., Elucidating the Role of Matrix Stiffness in 3D Cell Migration and Remodeling. Biophysical journal, 2011. 100(2): p. 284-293.

133. Pires, I., et al., Retinal thickness in eyes with mild nonproliferative retinopathy in patients with type 2 diabetes mellitus: comparison of measurements obtained by retinal thickness analysis and optical coherence tomography. Arch Ophthalmol, 2002. 120(10): p. 1301-6.

134. Masland, R.H., The fundamental plan of the retina. Nat Neurosci, 2001. 4(9): p. 877-886.

135. Marder, S.R., J.L. Bredas, and J.W. Perry, Materials for multiphoton 3D microfabrication. Mrs Bulletin, 2007. 32(7): p. 561-565.

136. Bellamkonda, R., J.P. Ranieri, and P. Aebischer, Laminin oligopeptide derivatized agarose gels allow three-dimensional neurite extension in vitro. J Neurosci Res, 1995. 41(4): p. 501-9.

137. Diaz-Quijada, G.A. and D.D.M. Wayner, A simple approach to micropatterning and surface modification of poly(dimethylsiloxane). Langmuir, 2004. 20(22): p. 9607-9611.

138. Chang, Y., et al., Sub-micrometer patterning of proteins by electric lithography. Langmuir, 2007. 23(8): p. 4112-4.

139. Carroll, G.T., et al., Photochemical micropatterning of carbohydrates on a surface. Langmuir, 2006. 22(6): p. 2899-905.

140. Farrer, R.A., et al., Selective functionalization of 3-D polymer microstructures. Journal of the American Chemical Society, 2006. 128(6): p. 1796-1797.

141. Kawata, S. and H.B. Sun, Two-photon photopolymerization as a tool for making micro-devices. Applied Surface Science, 2003. 208: p. 153-158.

142. Musoke, P. and M.S. Shoichet, Anisotropic three-dimensional peptide channels guide neurite outgrowth within a biodegradable hydrogel matrix. Biomed. Mater., 2006. 1: p. 162-169.

143. Furuta, T., et al., Brominated 7-hydroxycoumarin-4-ylmethyls: Photolabile protecting groups with biologically useful cross-sections for two photon photolysis. Proceedings Of

136

The National Academy Of Sciences Of The United States Of America, 1999. 96(4): p. 1193-1200.

144. Krapcho, A.P. and C.S. Kuell, Mono-Protected Diamines - N-Tert-Butoxycarbonyl-Alpha,Omega-Alkanediamines From Alpha,Omega-Alkanediamines. Synthetic Communications, 1990. 20(16): p. 2559-2564.

145. Rossi, F.M., et al., N-Nmoc-L-glutamate, a new caged glutamate with high chemical stability and low pre-photolysis activity. J Biol Chem, 1997. 272(52): p. 32933-9.

146. Liu, J.P., et al., Design Of 3-(4-Carboxybenzoyl)-2-Quinolinecarboxaldehyde As A Reagent For Ultrasensitive Determination Of Primary Amines By Capillary Electrophoresis Using Laser Fluorescence Detection. Analytical Chemistry, 1991. 63(5): p. 408-412.

147. Tong, Y.W. and M.S. Shoichet, Peptide surface modification of poly(tetrafluoroethylene-co-hexafluoropropylene) enhances its interaction with central nervous system neurons. J Biomed Mater Res, 1998. 42(1): p. 85-95.

148. Kloxin, A.M., et al., Mechanical properties of cellularly responsive hydrogels and their experimental determination. Advanced Materials, 2010. 22(31): p. 3484-3494.

149. Aizawa, Y., R. Wylie, and M. Shoichet, Endothelial cell guidance in 3-D patterned scaffolds. Advanced Materials, 2010. 22: p. 4831-4835.

150. Hersel, U., C. Dahmen, and H. Kessler, RGD modified polymers: biomaterials for stimulated cell adhesion and beyond. Biomaterials, 2003. 24(24): p. 4385-4415.

151. Campbell, P.G., et al., Engineered spatial patterns of FGF-2 immobilized on fibrin direct cell organization. Biomaterials, 2005. 26(33): p. 6762-6770.

152. Nur-E-Kamal, A., et al., Covalently attached FGF-2 to three-dimensional polyamide nanofibrillar surfaces demonstrates enhanced biological stability and activity. Molecular and Cellular Biochemistry, 2008. 309(1): p. 157-166.

153. Ito, T., et al., FGF-2 increases osteogenic and chondrogenic differentiation potentials of human mesenchymal stem cells by inactivation of TGF- signaling. Cytotechnology, 2008. 56(1): p. 1-7.

154. Rabchevsky, A., et al., Basic fibroblast growth factor (bFGF) enhances functional recovery following severe spinal cord injury to the rat. Experimental neurology, 2000. 164(2): p. 280-291.

155. Schultz, G.S. and A. Wysocki, Interactions between extracellular matrix and growth factors in wound healing. Wound Repair and Regeneration, 2009. 17(2): p. 153-162.

156. Montesano, R., et al., Basic fibroblast growth factor induces angiogenesis in vitro. Proceedings of the National Academy of Sciences of the United States of America, 1986. 83(19): p. 7297.

137

157. Marra, K.G., et al., FGF-2 enhances vascularization for adipose tissue engineering. Plastic and reconstructive surgery, 2008. 121(4): p. 1153.

158. Miller, E.D., et al., Dose-dependent cell growth in response to concentration modulated patterns of FGF-2 printed on fibrin. Biomaterials, 2006. 27(10): p. 2213-2221.

159. DeForest, C.A., E.A. Sims, and K.S. Anseth, Peptide-Functionalized Click Hydrogels with Independently Tunable Mechanics and Chemical Functionality for 3D Cell Culture. Chemistry of Materials, 2010. 22: p. 4783–4790.

160. Fairbanks, B.D., et al., A versatile synthetic extracellular matrix mimic via thiol-norbornene photopolymerization. Advanced Materials, 2009. 21(48): p. 5005-5010.

161. Hahn, M.S., J.S. Miller, and J.L. West, Three Dimensional Biochemical and Biomechanical Patterning of Hydrogels for Guiding Cell Behavior. Advanced Materials, 2006. 18(20): p. 2679-2684.

162. Hoffmann, J.C. and J.L. West, Three-dimensional photolithographic patterning of multiple bioactive ligands in poly(ethylene glycol) hydrogels. Soft Matter, 2010. 6(20): p. 5056-5063.

163. Leslie-Barbick, J.E., et al., Micron-scale spatially patterned, covalently immobilized vascular endothelial growth factor on hydrogels accelerates endothelial tubulogenesis and increases cellular angiogenic responses. Tissue Engineering - Part A, 2011. 17(1-2): p. 221-229.

164. Saik, J.E., et al., Biomimetic hydrogels with immobilized EphrinA1 for therapeutic angiogenesis. Biomacromolecules, 2011. 12(7): p. 2715-2722.

165. Jonsson, A., et al., Engineering of a femtomolar affinity binding protein to human serum albumin. Protein Engineering Design and Selection, 2008. 21(8): p. 515.

166. Invitrogen, Amine reactive probes.

167. Chiasson, B.J., et al., Adult mammalian forebrain ependymal and subependymal cells demonstrate proliferative potential, but only subependymal cells have neural stem cell characteristics. The Journal of neuroscience, 1999. 19(11): p. 4462.

168. Hein, K.L., et al., Crystallographic analysis reveals a unique lidocaine binding site on human serum albumin. Journal of structural biology, 2010. 171(3): p. 353-360.

169. Tognon, G., et al., Fetal bovine serum, but not human serum, inhibits the in vitro cytotoxicity of ET-743 (Yondelis, trabectedin). Cancer chemotherapy and pharmacology, 2004. 53(1): p. 89-90.

170. Kang, C.E., C.H. Tator, and M.S. Shoichet, Poly (ethylene glycol) modification enhances penetration of fibroblast growth factor 2 to injured spinal cord tissue from an intrathecal delivery system. Journal of Controlled Release, 2010. 144(1): p. 25-31.

138

171. Gasparian, M., et al., Overexpression in Escherichia coli and purification of human fibroblast growth factor (FGF-2). Biochemistry (Moscow), 2009. 74(2): p. 221-225.

172. McCall, M.J., H. Diril, and C.F. Meares, Simplified method for conjugating macrocyclic bifunctional chelating agents to antibodies via 2-iminothiolane. Bioconjugate Chemistry, 1990. 1(3): p. 222-226.

173. Parisot, J., et al., Use of azobenzene amino acids as photo responsive conformational switches to regulate antibody–antigen interaction. Journal of separation science, 2009. 32(10): p. 1613-1624.

174. Chivers, C.E., et al., A streptavidin variant with slower biotin dissociation and increased mechanostability. Nature methods, 2010. 7(5): p. 391.

175. Nguyen, D., et al., Molecular Mechanism of Apolipoprotein E Binding to Lipoprotein Particles†. Biochemistry, 2009. 48(13): p. 3025-3032.

176. Plotnikov, A.N., et al., Crystal structures of two FGF-FGFR complexes reveal the determinants of ligand-receptor specificity. Cell, 2000. 101(4): p. 413-424.

177. Pires, I., et al., Retinal thickness in eyes with mild nonproliferative retinopathy in patients with type 2 diabetes mellitus: comparison of measurements obtained by retinal thickness analysis and optical coherence tomography. Archives of Ophthalmology, 2002. 120(10): p. 1301.

178. Levine, E.M., et al., Sonic hedgehog promotes rod photoreceptor differentiation in mammalian retinal cells in vitro. Journal of Neuroscience, 1997. 17(16): p. 6277.

179. Bhattacharya, S., et al., Ciliary neurotrophic factor-mediated signaling regulates neuronal versus glial differentiation of retinal stem cells/progenitors by concentration-dependent recruitment of mitogen-activated protein kinase and Janus kinase-signal transducer and activator of transcription pathways in conjunction with Notch signaling. Stem Cells, 2008. 26(10): p. 2611-2624.

180. Leipzig, N.D., et al., Functional immobilization of interferon-gamma induces neuronal differentiation of neural stem cells. Journal of Biomedical Materials Research Part A. 93(2): p. 625-633.

181. Ho, J.E., et al., Immobilized sonic hedgehog N-terminal signaling domain enhances differentiation of bone marrow-derived mesenchymal stem cells. Journal of Biomedical Materials Research Part A, 2007. 83(4): p. 1200-1208.

182. Nakajima, M., et al., Combinatorial protein display for the cell-based screening of biomaterials that direct neural stem cell differentiation. Biomaterials, 2007. 28(6): p. 1048-1060.

183. Angot, E., et al., Chemoattractive activity of sonic hedgehog in the adult subventricular zone modulates the number of neural precursors reaching the olfactory bulb. Stem Cells, 2008. 26(9): p. 2311-2320.

139

184. Zhang, K.C., M.R. Diehl, and D.A. Tirrell, Artificial polypeptide scaffold for protein immobilization. Journal of the American Chemical Society, 2005. 127(29): p. 10136-10137.

185. Zhang, K.C., A. Sugawara, and D.A. Tirrell, Generation of Surface-Bound Multicomponent Protein Gradients. Chembiochem, 2009. 10(16): p. 2617-2619.

186. Deyev, S.M., et al., Design of multivalent complexes using the barnase· barstar module. Nature biotechnology, 2003. 21(12): p. 1486-1492.

187. Hartley, R.W., Barnase-barstar interaction. Methods in enzymology, 2001. 341: p. 599.

188. Weber, P.C., et al., Structural origins of high-affinity biotin binding to streptavidin. Science, 1989. 243(4887): p. 85.

189. Khan, F., et al., The kinetic pathway of folding of barnase. Journal of molecular biology, 2003. 333(1): p. 169-186.

190. Liang, S., et al., Protein diffusion in agarose hydrogel in situ measured by improved refractive index method. Journal of controlled release, 2006. 115(2): p. 189-196.

191. Aizawa, Y., R. Wylie, and M. Shoichet, Endothelial Cell Guidance in 3D Patterned Scaffolds. Advanced Materials, 2010. 22: p. 4831-4835.

192. Hartley, R.W., et al., Barstar inhibits extracellular ribonucleases of Streptomyces and allows their production from recombinant genes. Protein and Peptide Letters, 1996. 3(4): p. 225-231.

193. Cognet, I., et al., Expression of biologically active mouse ciliary neutrophic factor (CNTF) and soluble CNTFRalpha in Escherichia coli and characterization of their functional specificities. European cytokine network, 2004. 15: p. 255-262.

194. Lagunas, A., E.M. Comelles, and J. Samitier, Universal Chemical Gradient Platforms Using Poly(methyl methacrylate) Based on teh Biotin-Streptavidin Interactions for Biological Application. Langmuir, 2010.

195. Cull, M.G. and P.J. Schatz, Biotinylation of proteins in vivo and in vitro using small peptide tags. Methods in enzymology, 2000. 326: p. 430-440.

196. Howarth, M., et al., A monovalent streptavidin with a single femtomolar biotin binding site. Nature methods, 2006. 3(4): p. 267-273.

197. Chaiet, L. and F.J. Wolf, The properties of streptavidin, a biotin-binding protein produced by Streptomycetes. Archives of Biochemistry and Biophysics, 1964. 106: p. 1-5.

198. Wall, S.T., et al., Multivalency of Sonic hedgehog conjugated to linear polymer chains modulates protein potency. Bioconjugate chemistry, 2008. 19(4): p. 806-812.

140

199. Ruoslahti, E. and M.D. Pierschbacher, New perspectives in cell adhesion: RGD and integrins. Science, 1987. 238(4826): p. 491.

200. Mill, P., et al., Sonic hedgehog-dependent activation of Gli2 is essential for embryonic hair follicle development. Genes & development, 2003. 17(2): p. 282.

201. Singer, V., et al., Characterization of PicoGreen reagent and development of a fluorescence-based solution assay for double-stranded DNA quantitation. Analytical Biochemistry, 1997. 249(2): p. 228-238.

202. Coles, B.L.K., et al., Loss of retinal progenitor cells leads to an increase in the retinal stem cell population in vivo. European Journal of Neuroscience, 2006. 23(1): p. 75-82.

203. Peterson, W.M., et al., Ciliary neurotrophic factor and stress stimuli activate the Jak-STAT pathway in retinal neurons and glia. Journal of Neuroscience, 2000. 20(11): p. 4081.

204. Furuta, T., et al., Bhc cNMPs as either Water Soluble or Membrane Permeant Photoreleasable Cyclic Nucleotides for both One and Two Photon Excitation. ChemBioChem, 2004. 5(8): p. 1119-1128.

205. Hagen, V., et al., [8-[Bis (carboxymethyl) aminomethyl]-6-bromo-7-hydroxycoumarin-4-yl] methyl Moieties as Photoremovable Protecting Groups for Compounds with COOH, NH2, OH, and C O Functions. The Journal of Organic Chemistry, 2010. 75(9): p. 2790-2797.

206. Freier, T., et al., Controlling cell adhesion and degradation of chitosan films by N-acetylation. Biomaterials, 2005. 26(29): p. 5872-5878.

207. Kaneko, S., et al., Photocontrol of cell adhesion on amino-bearing surfaces by reversible conjugation of poly(ethylene glycol) via a photocleavable linker. Physical Chemistry Chemical Physics, 2011. 13(9): p. 4051-4059.

208. Yu, L.M.Y., F.D. Miller, and M.S. Shoichet, The use of immobilized neurotrophins to support neuron survival and guide nerve fiber growth in compartmentalized chambers. Biomaterials, 2010. 31(27): p. 6987-6999.

209. Chen, G. and Y. Ito, Gradient micropattern immobilization of EGF to investigate the effect of artificial juxtacrine stimulation. Biomaterials, 2001. 22(18): p. 2453-2457.

210. Mateo, C., et al., Improvement of enzyme activity, stability and selectivity via immobilization techniques. Enzyme and Microbial Technology, 2007. 40(6): p. 1451-1463.

211. Ballios, B.G., et al., A hydrogel-based stem cell delivery system to treat retinal degenerative diseases. Biomaterials, 2010. 31(9): p. 2555-2564.

212. Orive, G., et al., Biomaterials for promoting brain protection, repair and regeneration. Nature Reviews Neuroscience, 2009. 10(9): p. 682-692.

141

213. Mieszawska, A.J. and D.L. Kaplan, Smart biomaterials- regulating cell behavior through signaling molecules. BMC biology, 2010. 8(1): p. 59.

142

Appendix A: Abbreviation

2D: Two-dimensional

3D: Three-dimensional

ABD: Albumin Binding Domain

Bhc: 6-bromo-7-hydroxycoumarin

BHQ: 8-bromo-7-hydroxyquinoline

CBQCA: 3-(4-carboxybenzoyl)-2-quinolinecarboxaldehyde

CNTF: Ciliary Neurotrophic Factor

DRG: Dorsal Root Ganglion

ECM: Extracellular Matrix

EDC: 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide

EGF: Epidermal Growth Factor

FGF-2: Fibroblast Growth Factor-2

FPLC: Fast Protein Liquid Chromatography

HSA: Human Serum Albumin

IFN-γ: Interferon gamma

Kd: Dissociation constant

Mal: Maleimide

MMP: Matrix Metalloproteinase

NA: Numerical Aperture

NGF: Nerve Growth Factor

NHS: n-hydroxysuccinimide

NSPC: Neural Stem Progenitor Cell

143

NTA: Nitrilotriacetic acid

PDGF: Platelet-derived Growth Factor

PEG: Polyethylene glycol

Qu: Uncaging quantum yield

RPC: Retinal Progenitor Cell

RSC: Retinal Stem Cell

RSPCs: Retinal Stem Progenitor Cells

SDS-PAGE: Sodium dodecyl sulfate polyacrylamide gel electrophoresis

SEC: Size-exclusion chromatography

SHH: N-terminal Sonic Hedgehog

Sulfo-SMCC: Sulfosuccinimidyl-4-(N-maleimidomethyl)cyclohexane-1-carboxylate

TP: Two-photon

VEGF: Vascular Endothelial Growth Factor

δ: Cross-section for one-photon absorption

δ2: Cross-section for two-photon absorption

δu: Uncaging cross-section

144

Appendix B: Supplemental Figures

Supplemental Figure 1: Fluorescence intensity profile of SHH-488 and CNTF-633 as a

function of depth for the dual pattern presented in Fig 4. (a) The fluorescence profile

for SHH-488 showing the four layers of the pattern in Fig 4. The intensity

decreases for each layer even though each layer was patterned with the same

amount of irradiation due to the scattering of light during the imaging process. (b)

The fluorescence profile of CNTF-633 also shows the four layers of the pattern.

The fluorescence intensity did not decrease as much as SHH-488 because 633

scatters less light than 488. This demonstrates that the decrease in fluorescence

of SHH-488 results from the imaging process, and not from the immobilization,

using two-photon chemistry. If the decrease in fluorescence was from the

immobilization process, a decrease in fluorescence for both SHH-488 and CNTF-

633 would have been observed. In a separate experiment where Alexa-488 and

Alexa-633 were immobilized at defined depths, we observed a similar decrease in

fluorescent intensity for Alexa-488 only (and not Alexa-633) as a function of

depth, confirming our scattering hypothesis.

145

Supplemental Figure 2: RPCs survive on agarose hydrogels with immobilized

SHH using the barnase/barstar system. (a) Micrographs of RPCs cultured on top of

agarose-barnase-SHH (with GRGDS), agarose-barnase (with GRGDS) and GRGDS

alone. Cells were stained using the Live/Dead (calcein AM/ethidium homodimer-1)

assay. Micrographs showing brightfield, live cells (green), and dead cells (red) are

shown. (b) Percent of live cells was quantified for each condition (mean ± s.d., n=6 gels

with 5,000 cells per gel). No significant difference was observed between any groups

using one-way ANOVA with Tukey’s post-hoc analysis (p > 0.05).

146

Supplemental Figure 3: RPCs survive on agarose hydrogels with immobilized

CNTF using the streptavidin/biotin system. (a) Micrographs of RPCs cultured on top

of agarose-streptavidin-CNTF (with GRGDS), agarose-streptavidin (with GRGDS) and

GRGDS alone. Cells were stained using the Live/Dead (calcein AM/ethidium

homodimer-1) assay. Micrographs showing brightfield, live cells (green), and dead cells

(red) are shown. (b) Percent of live cells was quantified for each condition (mean ± s.d.,

n=6 gels with 5,000 cells per gel). No significant difference was observed between any

groups using one-way ANOVA with Tukey’s post-hoc analysis (p > 0.05).

147

Appendix C: Magnetic cell seeding

NSPCs with magnetic nanoparticles were seeded into agarose hydrogels using a magnetic field.

NSPCs were first cultured in the presence of dextran coated iron nanoparticles (250 nm), which

resulted in cellular uptake of the particles. The cells were then plated on top of 0.5 wt. % agarose

hydrogels and cultured on a 1” rare earth magnet. Interestingly, it was found that cells only

penetrated the gels cultured in the presence of a magnet field. Therefore the cells were able to

migrate through the gel only with the additional magnetic force. Figure C1 demonstrates that

cells in the presence of the magnetic field penetrated the gel to a depth of at least 150 µm. In the

absence of the field no cellular migration within the gel was observed. Therefore magnetic fields

can be investigated as an external stimulus to encourage cell penetration into patterned

hydrogels.

Figure C1: NSPCs with magnetic nanoparticles penetrated into agarose-RGD hydrogels in the

presence of a magnetic field after 7 d. The surface of the gel is indicated by the black dashed

line.

148

Method

NSCs were cultured in the presence of 0.25mg/ml of submicron magnetic particles for 2 days in

NBM media with EGF, FGF and heparin. Cells were then washed three times in 1%FBS NBM

before plating. 100,000 cells were then plated on 400ul of RGD agarose gels of 0.5 wt.% in a

cuvette (1 x 1 cm base) with rare earth magnet of 1” in diameter underneath. Controls were

cultured in the absence of a magnet. The cells were cultured in 1% FBS NBM for 7 d.

149

Copyright Acknowledgements

Chapters 2, 3, and 4 were adapted from published original research articles. Proper copyright

permissions were granted by the respective publishers. These works were primarily written by

Ryan Wylie.

150

Ryan Wylie

160 College Rm530 Toronto Ontario M5S 3E1 Canada

Phone: 416-978-0343 Email: ryan.wylie@utoronto.ca

EDUCATION

Sept 2005-Present: University of Toronto, Doctoral Candidate in Chemistry. Three-dimensional patterning of hydrogels for tissue engineering. Supervisor: Prof. Molly Shoichet.

Sept 2001-May 2005: Concordia University, BSc. in Biochemistry (Honours).

PUBLICATIONS

Peer-Reviewed 10- Wylie, R. G.; Shoichet, M. S., Three-Dimensional Spatial Patterning of Proteins in

Hydrogels. Biomacromolecules 2011. doi: 10.1021/bm201037j 9- Wylie RG, Ashan S, Aizawa Y, Maxwell KL, Morshead CM, Shoichet MS. Three-

dimensional, spatially-controlled simultaneous patterning of multiple growth factors in hydrogels. Nature Materials 2011, 10 (10), 799-806.

8- Leipzig ND, Wylie RG, Kim H, Shoichet MS. Differentiation of neural stem cells in three-dimensional growth factor-immobilized chitosan hydrogel scaffolds. Biomaterials 32, 57-64 (2011) doi:10.1016/j.biomaterials.2010.09.031

7- Wang Y, Cooke MJ, Lapitsky Y, Wylie RG, Sahewsky N, Corbett D, Morshead CM,

Shoichet MS. Transport of epidermal growth factor in the stroke-injured brain. Journal of Controlled Release 149, 225-235 (2011). doi:10.1016/j.jconrel.2010.10.022

6- Aizawa Y, Wylie RG, Shoichet MS. Endothelial Cell Guidance in 3D Patterned Scaffolds.

Advanced Materials, 22, 4831-4835 (2010). doi: 10.1002/adma.201001855 5- Rahman N, Purpura, KA, Wylie RG, Zandstra PW, Shoichet MS. The use of vascular

endothelial growth factor functionalized agarose to guide pluripotent stem cell aggregates toward blood progenitor cells. Biomaterials 31, 8262-8270 (2010). doi: 10.1016/j.biomaterials.2010.07.040

4- Taerum T, Lukoyanova O, Wylie RG, Perepichka, DF. Synthesis, Polymerization, and

Unusual Properties of New Star-Shaped Thiophene Oligomers. Organic Letters 11, 3230-3233 (2009). doi: 10.1021/ol901127q

3- Wylie RG, Shoichet, MS. Two-photon micropatterning of amines within an agarose

hydrogel. Journal of Materials Chemistry 18, 2716-2721 (2008). doi: 10.1039/B718431J

151

Patent

2- Wosnick, J.; Wylie, R.; Shoichet, M.S. U.S. Provisional Patent Application, 2008 "Three-Dimensional Patterned Hydrogels" Patent Pub. No. 2008/028630

Book Chapter 1- Lévesque, S.; Wylie, R.; Aizawa, Y.; Shoichet, M.S. 2008 "Peptide Modification of

Polysaccharide Scaffolds for Targeted Cell Signaling" in Handbook of Natural-based Polymers for Biomedical Applications, Ch. 9, pp. 260-87, edited by R.L. Reis, Woodhead Publishing Ltd, UK.

ORAL PRESENTATIONS

Wylie RG, Ahsan S, Maxwell K, Morshead C, Shoichet MS. Tissue Engineered 3D Patterned Hydrogels. ACS National Meeting 2010. San Francisco, CA. Wylie RG, Wosnick JH, Ashan S, Morshead C, Shoichet MS. Femtosecond Light Patterned Hydrogels for Tissue Engineering. TERMIS-NA 2009. San Diego, CA.

Wylie RG, Shoichet MS. Femtosecond Laser Patterning of Hydrogels for Tissue Engineering. 34th Québec-Ontario Physical Organic Mini-Symposium 2006. Montréal, QC.

POSTER PRESENTATIONS

Wylie RG, Wosnick JH, Maxwell K, Shoichet MS. Photo-patterning of matrices to spatially control cellular activity. Stem cell network AGM 2008. Vancouver, BC.

Ahsan S, Wylie RG, Shoichet MS, Morshead C. Creating Retinal Tissue in 3D with Defined Factors using Adult Retinal Stem Cells. TERMIS-NA 2008. San Diego, CA.

Wylie RG, Wosnick J, Ashan S, Morshead C, Shoichet MS. Femtosecond light patterned hydrogels for the guidance of retinal progenitor cells. Vision Science Reseach Day 2008. Toronto, ON.

Wylie RG, Wosnick J, Shoichet MS. Femtosecond light patterned hydrogels for tissue engineering with stem cells. Stem cell network AGM 2007. Toronto, ON.

Wylie RG, Wosnick J, Morshead C, Shoichet MS. Femtosecond light patterned hydrogels for tissue engineering. Ontario Centre of Excellence Discovery 2007. Toronto, ON.

Taerum TA, Wylie RG, Perepichka DF. Synthesis of building blocks for 2-D conjugated polymers. ACS meeting 2007. Boston, MA.

Wylie RG, Wosnick J, Miller RJD, Morshead C, Shoichet MS. Femtosecond Light Pattern Neural Networks: New Concepts for Regenerative Medicine. Ontario Centre of Excellence Discovery 2006. Toronto, ON.

152

AWARDS

FQRNT – Doctoral Research Scholarship 2007-2009 Vision Science - Doctoral Research Scholarship 2006-2009 Recruitment Award (U of T) Spring 2005 FQRNT - Master’s Research Scholarship 2005-2007 NSER C USRA (INRS-EMT) Summer 2004 NSERC USRA (Methylgene) Fall 2003 NSERC USRA (Delmar) Summer 2002

INDUSTRIAL RESEARCH EXPERIENCE

May 2004-Aug 2005: Institut National de la Recherche Scientific - Énergie, Matériaux et Télécommunication (Varennes, QC). Synthesis of thiophene oligomers.

Sept 2003-Dec 2003: Methylgene Inc. (Montréal, QC). Design and synthesis of β-lactamase inhibitors.

May 2002-Apr 2003: Delmar Inc. (Montréal, QC). Optimization of chemical reactions and purification of Paclitaxel from the Canadian yew tree.