lable at ScienceDirect

Biomaterials 35 (2014) 3999e4004

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

The effect of retinal pigment epithelial cell patch size on growth factorexpression

Elizabeth Vargis a,b,1, Cristen B. Peterson c, Jennifer L. Morrell-Falvey c, Scott T. Retterer a,c,*,Charles Patrick Collier a,**aCenter for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USAb Joint Institute of Biological Sciences, University of Tennessee, Knoxville, TN 37996, USAcBiosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

a r t i c l e i n f o

Article history:Received 13 November 2013Accepted 8 January 2014Available online 30 January 2014

Keywords:Microcontact printingRetinal pigment epithelialTight junctionsVEGFIn vitro modelMicropatterning

* Corresponding author. Center for Nanophase MNational Laboratory, Oak Ridge, TN 37831, USA. Tel.: þ574 5345.** Corresponding author. Tel.: þ1 865 576 3638; fax

E-mail addresses: rettererst@ornl.gov (S.T. R(C.P. Collier).1 Current institution: Department of Biological Eng

sity, Logan, UT 84322-4105, USA.

0142-9612/$ e see front matter � 2014 Elsevier Ltd.http://dx.doi.org/10.1016/j.biomaterials.2014.01.016

a b s t r a c t

The spatial organization of retinal pigment epithelial (RPE) cells grown in culture was controlled usingmicropatterning techniques in order to examine the effect of patch size on cell health and differentiation.Understanding this effect is a critical step in the development of multiplexed high throughput fluidicassays and provides a model for replicating disease states associated with the deterioration of retinaltissue during age-related macular degeneration (AMD). Microcontact printing of fibronectin on poly-styrene and glass substrates was used to promote cell attachment, forming RPE patches of controlled sizeand shape. These colonies mimic the effect of atrophy and loss-of-function that occurs in the retinaduring degenerative diseases such as AMD. After 72 h of cell growth, levels of vascular endothelialgrowth factor (VEGF), an important biomarker of AMD, were measured. Cells were counted andmorphological indicators of cell viability and tight junction formation were assessed via fluorescencemicroscopy. Up to a twofold increase of VEGF expression per cell was measured as colony size decreased,suggesting that the local microenvironment of, and connections between, RPE cells influences growthfactor expression leading to the initiation and progression of diseases such as AMD.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Controlling the spatial organization and growth of cells is animportant step in understanding cell behavior and modeling dis-ease processes in vitro. Monitoring the expression of importantproteins within these controlled environments can lead to newinsights that inform our understanding of how disease processesare initiated and develop over time. While many of these questionshave been studied using conventional cell culturing methods, thecontinued development of physiologically relevant models that canreplicate the metabolic, mechanical and barrier properties offunctional tissues remains important [1,2]. Microscale surfacechemical patterning offers one method of controlling the growth of

aterials Sciences, Oak Ridge1 865 405 4066; fax: þ1 865

: þ1 865 574 1753.etterer), colliercp@ornl.gov

ineering, Utah State Univer-

All rights reserved.

cells in vitro. Previous studies have evaluated the effect of extra-cellular matrix (ECM) composition on cell size and shape as well asthe secretion of additional ECM by growing cells onmicropatternedstructures [3,4]. The impact of micropatterning on function andmorphology has been investigated on many types of cells includingkeratinocytes, neuronal cells, fibroblasts, macrophages, stem,epithelial, and cancer cells [3e6]. Such studies are critical tocombining directed growth techniques with high throughputmicrofluidic assays, where individual patches of cells can beaddressed, manipulated and studied in parallel [7e9]. Many dis-eases, such as liver failure, osteoporosis, and cancer, have beenstudied using micropatterning techniques. These studies have ledto the development of techniques for controlling the self-assemblyof vascular structures and liver tissues, manipulating the mechan-ical forces and stress placed on osteoblasts, and demonstratingdifferences in morphology and motility between normal andcancerous cells [10e13]. In this paper, micropatterning techniquesare used to create cellular ‘patches’ that emulate defects in naturalsystems which can be used to examine the impact of local cellecelland celleenvironment interactions on the loss of barrier propertiesas well as changes in the biomarker expression of retinal pigmentepithelial cells.

Etched Silicon Master

PDMS, Silicone Mold

50µg/mL Fibronectin

PS Cell Culture Dish

0.7% (w/v) Pluronic F-127 SolutionR

a

b

c

d

e

f

g

h

Fig. 1. Substrate functionalization and cell seeding is accomplished by molding aPDMS (silicone) stamp from an etched silicon master (a,b). The stamp is ‘inked’ using afibronectin solution, allowed to dry and is then placed in contact with a polystyrene orglass culture dish (cef). The surface was blocked with Pluornic� F-127 prior to seedingRPE cells onto the fibronectin-patterned substrate (g,h).

E. Vargis et al. / Biomaterials 35 (2014) 3999e40044000

Retinal pigment epithelial cells (RPEs) are polarized and highlyspecialized neural cells that provide metabolic support to photo-receptor cells (rods and cones). They are an important componentof the retina with a central role in normal vision as well as theinitiation and progression of retinal diseases such as age-relatedmacular degeneration (AMD). When RPE cells do not behaveappropriately or die as the result of repeated exposure to reactiveoxidative species, the formation of drusen (acellular debris, dryAMD) or retinal neovascularization (wet AMD) can occur [14]. DryAMD is currentlymanaged by vitamin supplementation and dietarymodifications. For patients with wet AMD, expensive anti-angiogenic (i.e. anti-vascular endothelial growth factor (VEGF))therapy is the most common treatment option [15].

Anti-VEGF therapy is used to treat wet AMD because of thecentral role that VEGF plays in neovascularization. As a signalingprotein, VEGF is needed for angiogenesis and neovascularization innormal processes such as embryonic development and the for-mation of new muscles after exercise. It has been implicated incancer malignancy as well as diseases of the retina, including dia-betic retinopathy and AMD. In the eye, RPE cells are the primaryretinal source of VEGF [16] and also express VEGF receptors [17,18].VEGF, in turn, impacts tight junction formation of RPE cells, alteringthe permeability of the epithelial barrier between the retina and itsblood supply [17].

The goal of this study is to replicate the effects of disease andloss of function within the retina by using micropatterning tech-niques to direct the growth of RPE cells into discrete patches andmeasuring the effect of patch size on VEGF expression. RPE cellswere added to micropatterned fibronectin patches on tissue-culture treated polystyrene dishes with a hydrophobicehydro-philic block-copolymer as a blocking agent. Quantitative imageanalysis of RPE patch morphology and tight junction formationwasperformed. VEGF expression was measured from ensembles of RPEpatches of varying size and was compared to VEGF expression inconfluent layers of RPE cells using enzyme-linked immunosorbentassay (ELISA). Levels of VEGF were measured as a function of cellnumber and patch area. Changes in cell density, cell size, ormorphologywithin the RPE patches were examined. Hypothesizingthat VEGF expression was linked to global VEGF concentrations inculture, VEGF expression from RPE patches of varying size werequantified following VEGF agonist administration. This study il-lustrates how direct cell patterning can be used to influence thesecretion of VEGF in RPE tissue models and highlights a path to-wards mimicking the effects of tissue damage or atrophy in cellculture.

2. Methods

2.1. Fibronectin printing

Microcontact printing was used to pattern fibronectin (Sigma Aldrich, St. Louis,MO) onto substrates. Stamps with patches of 100, 200, 300, 400 and 500 mmdiameter were made out of poly(dimethylsiloxane) (PDMS), prepared from a Sylgard184 silicone elastomer kit (Dow Corning, Midland, MI), using soft lithography [7].The PDMS stamps were sterilized in 70% ethanol and “inked” with a solution of50 mg/mL fibronectin overnight. The fibronectin layer was transferred to the sub-strate by placing the stamp in conformal contact with it for 1 min (Fig. 1). Afterpassivation with 0.7% wt/vol Pluornic� F-127 (Sigma Aldrich), to prevent proteinadsorption and cell attachment to unpatterned regions of the substrate, cells wereplated at a concentration of 1�106 cells/mL. Thirty minutes after seeding, the mediawas replaced to prevent non-specific cell attachment.

2.2. Cell culture

ARPE-19 cells (American Type Culture Collections, Manassas, VA) were main-tained and passaged in standard culture dishes (Corning; Sigma Aldrich) usingDMEM/F12 media (Lonza, Allendale, NJ) with 10% fetal bovine serum (Gibco,Carlsbad, CA) and 20 mL/L L-glutamine-penicillin-G-streptomycin (2 mM, 100 U/mL,0.1 mg/mL, respectively, Gibco) in a humidified incubator at 37 �C in 5% CO2. Cellswere obtained at passage 11 and used through passage 30.

2.3. Enzyme-linked immunoassays

Aliquots of media from culture wells containing the stamped substrates werecollected (100 mL). VEGF concentrations were assayed using a human VEGF ELISA kit(Invitrogen, Carlsbad, CA) using a BioTek Synergy 2 microplate reader. The outputwas reported in pg/mL to account for any volume changes during the experiment.VEGF expression was measured at regular intervals: 4, 24, 30 48, 54, 72 h after thecells were seeded onto the patches.

2.4. Immunohistochemistry and analysis

Substrates with patterned RPE patches were washed in PBS and fixed in 2%paraformaldehyde (Sigma Aldrich) at 4 �C for 1 h. The samples were washed 2 timesin PBS at RT and permeabilized by exposure to 0.2% Triton X-100 (VWR, WestChester, PA) for 30 min at 4 �C. Samples were washed and incubated with primaryantibody (mouse anti-ZO-1, diluted 1:100, Invitrogen) for 1 h at 4 �C. After washing,the samples were incubated with secondary antibody (Alexa 488 goat anti-mouse,diluted 1:500; Invitrogen) for 30 min at RT. The samples were then stained withHoechst 33342 and examined using a Zeiss LSM710 confocal microscope and Zen2010 (Zeiss) software.

Image analysis was performed using ImageJ (NIH). First, images were convertedfrom grayscale to black and white (binary) images reflecting stained and unstainedareas from the fluorescent micrographs. The image threshold was determined usingthe triangle method within ImageJ and a set of cells from each sample was outlinedas a Region of Interest (ROI). Parameters such as area and number of cells wereextracted using the ‘Analyze Particles.’ function. A Student’s t-test was used todetermine if significant differences in cell size and tight junction formation werepresent (p < 0.05).

0

20

40

60

80

100

Conflu

ent

Cell

s 100 µ

m

200 µ

m

300 µ

m

400 µ

m

500 µ

m

No Cell

s

% C

ell C

over

age

Sample Type

Fig. 3. Percent cell coverage observed for each patch size. Error bars indicate onestandard deviation.

Table 1Patterning Efficiency, measured after 72 h.

Sample type Mean patterning efficiency Mean cell size (mm2)

Confluent cells 99.6 � 1.7% 80 � 1.2100 mm 63.1 � 1.9% 81 � 0.9200 mm 76.2 � 1.3% 80 � 1.1300 mm 83.7 � 0.9% 78 � 0.8400 mm 87.3 � 1.8% 78 � 2.3500 mm 98.6 � 0.8% 79 � 1.4

E. Vargis et al. / Biomaterials 35 (2014) 3999e4004 4001

2.5. Agonist

Micropatterned patches of RPE cells were treated with 5 ng/mL VEGF-E (aspecific viral VEGF analog targeting VEGF-R2; Fitzgerald, Concord, MA) after 20 h inculture. VEGF expression was measured before the addition of the VEGF-E agonist(at 4 h) and after (24, 30, 48, 54, 72 h).

3. Results

Fig. 2 shows fluorescent images of the 5 stamps after fluorescentfibronectin inking (a), bright field images 30 min after the additionof cells (b), and fluorescent images of nuclei (c) and tight-junctionformation (d) after 72 h. Patch sizes and their correspondingdiameter and area were verified by measuring the diameter of thefluorescent fibronectin patches (Fig. 2a) in ImageJ (NIH). Cellcoverage was measured from at least 10 separate patterns in eacharray (Fig. 3) and normalized by dividing by the total pattern area tocalculate the percentage of cell coverage. Table 1 displays the meanfilling efficiency and the mean cell size. Although fewer cells perunit area were present in the smaller patterns, the average cell sizedid not change as a function of pattern size. The ‘unpatterned’samples corresponded to cells grown on 35mmpetri dishes treatedwith fibronectin.

After 72 h, cells were fixed and stained for the presence of Zona-occludens 1 (ZO-1) as an indicator of tight junction formation(Fig. 4). ImageJ was used to assess the total area of tight junctionsformed within each pattern. In order to understand tight junctionformation within the patterned compared to unpatterned samples,the results were reported as a percent comparison relative to thetight junctions formed in a monolayer of RPE cells of an unpat-terned sample of equal area. The average tight junction formationin smaller pattern sizes was lower and more variable (63� 10% ZO-1, 100 mm diameter pattern; 74 � 7%, 200 mm pattern) compared tothe larger pattern sizes (82e91%). Differences among the sampleswere not statistically significant.

Total VEGF expression was first measured from each array ofmicropatterned areas after 72 h using an ELISA kit and is reportedas the VEGF concentration per cell based on nuclear counts (Fig. 5).

Fig. 2. Verification of stamp formation and cell growth on patches. (a) Fluorescent fibronecontrast levels were increased on all the images to enhance the visualization of stamps ofpatches after 24 h. (c) After 72 h, nuclei and tight junctions (ZO-1, D) were stained, h(aec) ¼ 100 mm; grey scale bares (d) ¼ 20 mm.

After 72 h, the VEGF concentration per cell was inversely propor-tional to the patch size, i.e. RPE cells from smaller patchesexpressed higher levels of VEGF per cell. A monolayer of cells(unpatterned) and no cells on polystyrene surfaces were used as

ctin was used initially to form the fibronectin patches for cell growth. Brightness and300 mm and 200 mm diameter. (b) Bright-field images of cells growing on fibronectinighlighting variations in cell density among various patch sizes. White scale bars

0

20

40

60

80

100

100 200 300 400 500 % C

over

age

of Z

O-1

per

1 p

atte

rn

Pattern Size ( m)

Fig. 4. Tight junction (ZO-1) coverage after 72 h. Each measurement was normalizedto the tight junction coverage of confluent layer of cells. Error bars indicate onestandard deviation.

0

0.5

1

1.5

2

Conflu

ent

Cell

s

VEG

F C

once

ntra

tion

(pg/

mL)

per

cel

l

Sample Type 10

0 µm

200 µ

m

300 µ

m

400 µ

m

500 µ

m

No Cell

s

Fig. 5. VEGF concentration per cell was calculated for different patch sizes fromsamples taken after 72 h of culture. Measurements were made using a VEGF ELISA kit.Imaging with nuclear staining was used to facilitate cell counts. Error bars indicate onestandard deviation.

E. Vargis et al. / Biomaterials 35 (2014) 3999e40044002

positive and negative controls, respectively. To understand how theconcentration of VEGF changes over time, VEGF expression levelswere also measured at various intervals (Fig. 6; 4, 24, 30, 48, 54,72 h). While the level of VEGF was similar 4e24 h after the cellswere seeded across the different pattern sizes, a significant differ-ence was observed after 48 h, with smaller patches producing ahigher amount of VEGF than larger patches or unpatterned samples(Fig. 6).

Cells growing in small micropatterns may experience differentlocal concentrations of VEGF compared to cells in larger patches,and thus alter VEGF expression levels to compensate. To determineif the higher levels of VEGF expression observed in cells grown insmall patches was the result of cells responding to lower initialoverall levels of VEGF in the RPEmicroenvironment, a VEGF agonist,VEGF-E, was added to the patterned surfaces at a concentration of5 ng/mL, 5� the amount of VEGF observed at 72 h time points, 20 hafter the initial cell seeding. VEGF-Ewas derived from a non-humansource and was therefore not detected with the VEGF ELISA. Theconcentration of VEGF after the addition of the agonist wasmeasured at the same intervals as previous experiments and is

reported as the % change in VEGF expression per cell for each RPEpatch size (Fig. 7). While the VEGF expression from the cells in thepositive control and the patterns of 500 and 400 mm diameter didnot change as a result of the additional VEGF-E, the VEGF levelsfrom the 100, 200 and 300 mm patterns decreased, reaching levelsper cell closer to the larger pattern sizes.

4. Discussion

The micropatterned surfaces were characterized to verify theirtrue area and the amount of cells that adhered to each pattern. Themeasured diameter (Table 2) found from the fluorescent fibro-nectin patches (Fig. 2a) was used in all calculations of area to ac-count for small variations in the PDMS mold and the stamping.Variations in the intensity of the fluorescent fibronectin were mostlikely due to differences in stamping pressure. The size and numberof cells in each pattern size was then determined using the nuclearstained images (Fig. 2c) and dividing by the total area of eachpattern (Fig. 3). Fig. 3 and Table 1 show that after 72 h, althoughcells of similar sizes were seeded onto each pattern, the meanfilling efficiencies of smaller pattern sizes were lower than forlarger patch sizes. This may have been due to the lower cell densityresulting in lower cell viability or the mechanistic effects limitingthe cells from adhering to smaller pattern sizes [2,13]. The size andmorphology of the RPE cells was verified by measuring mean cellsize (Table 1) and the percent of tight junction formation relative tounpatterned cells for each patch size (Fig. 4). Differences betweenpatches were not statistically significant. These results suggest thatalthough there were fewer cells on smaller patterns, the RPE cellswere similar in size, morphology and function across patch sizes.

That the RPE cells in small patches expressed higher levels ofVEGF per cell suggests that these cells may function to maintain aconsistent level of VEGF within their local microenvironment. Cellsin smaller patches respond by expressing higher amounts of VEGFto maintain basal VEGF levels (Figs. 5 and 6). RPE cells in largerpatches can maintain the same basal levels of VEGF expression byexpressing smaller amounts of VEGF per cell. Previous studies haveshown that VEGF is a modulator of the barrier function in theretinal endothelial and RPE layers [19,20]. The expression of VEGFhas similarly been characterized in excised retinal tissue. Humanspecimens with early and advanced AMD have significantlyincreased levels of VEGF compared to normal [21]. Normal RPEmonolayers can also express higher levels of VEGF under hypoxicconditions [22].

Following the addition of the VEGF agonist to the micro-patterned cultures, levels of VEGF expression per cell for the smallpatch sizes approached the levels of the unpatterned and largerpatch sizes (Fig. 7). Adding the agonist effectively increased theamount of VEGF detected by the cultured cells in each sample. TheVEGF levels per cell obtained from smaller patch sizes (100, 200,300 mm) decreased after the VEGF agonist was added. We hy-pothesize that cells within these smaller patterns reduce VEGFexpression levels because of the increased levels of VEGF detectedwithin their local environment. The patches of larger sizes werealready exposed to higher levels of VEGF in their microenvironmentand therefore showed no change in expression levels when a VEGFagonist was added. As a result, the cells cultured in larger patchesmaintained their basal VEGF expression.

Microcontact printing provides a tool for manipulating thespatial organization of retinal pigment epithelial cells as a means ofexploring the impact of atrophy, tissue damage, or loss-of-functionwithin the retina. Combined with targeted biochemical measure-ments, these patterned growth techniques provide a potentiallypowerful way of examining the links between growth factorexpression, tissue reorganization and the progression of AMD. The

0

0.4

0.8

1.2

1.6

2

4 24 30 48 54 72

VEG

F C

once

ntra

tion

per c

ell (

pg/m

L)

Time (hours)

100 m 200 m 300 m 400 m 500 m Confluent Cells No Cells

Fig. 6. Time course of VEGF expression per cell measured at 4, 24, 30, 48, and 72 h.Error bars indicate one standard deviation.

-25

-20

-15

-10

-5

0

5 4 24 30 48 54 72

% c

hang

e in

VEG

F Ex

pres

sion

Per

Cel

l Fo

llow

ing

the

Add

ition

of A

goni

st

Time (hours)

100 m 200 m 300 m 400 m 500 m Confluent Cells

Fig. 7. Effect of agonist added after 20 h in culture on VEGF expression, displayed as a %change in VEGF expression compared to VEGF expression without the addition of anagonist. Error bars indicate one standard deviation.

E. Vargis et al. / Biomaterials 35 (2014) 3999e4004 4003

results of this paper are suggestive of the effects of tissue atrophyon biomarker expression and the behavior of cells. While retinalcells and AMD were the focus of this study, similar models can bedeveloped to mimic other cell types and disease states associatedwith tissue atrophy. For example, a number of diseases of the brainlead to cerebral atrophy, such as Alzheimer’s, multiple sclerosis,and Cushing’s syndrome [23e25]. Similar models can be used tostudy common treatments and other types of therapies. CurrentAMD therapy relies on monthly anti-VEGF injections to block the

Table 2Dimensions of patterns measured from PDMS stamps and fluorescent fibronectin.Area calculated based on fluorescent fibronectin pattern.

Nominal patterndiameter (mm)

Diameter of PDMSpattern (mm)

Diameter offibronectionpattern (mm)

Measuredpattern area(mm2)

100 100.1 � 0.5 99.3 � 0.3 70 � 5200 204.2 � 1.4 201.1 � 1.2 340 � 20300 299.7 � 2.6 298.6 � 2.3 710 � 40400 401.3 � 1.6 399.7 � 1.4 1210 � 110500 503.2 � 2.9 502.2 � 1.9 1860 � 240

downstream effects of VEGF overexpression, but not the over-expression of VEGF itself. The results presented here demonstratethat it may also be useful to treat the cells before they begin toabnormally overexpress biomarkers such as VEGF. Other growthfactors that act as VEGF-antagonists, such as pigment-epitheliumderived growth factor (PEDF), may promote normal cell behaviorand reduce the abnormal and disease-causing results of VEGFoverexpression that occur in retinal cells during AMD [17]. Studyingother treatments and diseases using models similar to the onepresented here may lead to a greater understanding of both thedisease and potential avenues of treatment.

5. Conclusion

In this study, micropatterned surfaces were used to replicate theeffect of atrophic and leaky RPE cells on VEGF expression. Thismethod offers a unique combination of controlling the spatialgrowth of a cell population and interrogating changes in growthfactor expression that may result from the specific growth patterns.The results from this study suggest that as larger areas of atrophicor missing cells surround RPE cells, they begin to secrete largeramounts of VEGF per cell, an important factor necessary for theneovasculogenesis that commonly occurs during macular degen-eration. The micropatterned surfaces provide a means of studyingthe loss of barrier function in the RPE cells of the eye, providing anenhanced understanding of retinal disease progression. Futureresearch will incorporate high throughput microfluidic devices toperform real-time local dosing and measurements of proteinexpression from small patches of cells aligned to more complexmicrofluidic architectures.

Acknowledgments

A portion of this research was conducted at the Center forNanophase Materials Sciences, which is sponsored at Oak RidgeNational Laboratory by the Scientific User Facilities Division, Officeof Basic Energy Sciences, U.S. Department of Energy.

References

[1] Falconnet D, Csucs G, Grandin HM, Textor M. Surface engineering approachesto micropattern surfaces for cell-based assays. Biomaterials 2006;27(16):3044e63.

[2] Ohl A, Schröder K. Plasma-induced chemical micropatterning for cell culturingapplications: a brief review. Surf Coatings Technol 1999;116:820e30.

[3] Gomez AM, Valdivia HH, Cheng H, Lederer MR, Santana LF, Cannell MB, et al.Defective excitation-contraction coupling in experimental cardiac hypertro-phy and heart failure. Science 1997;276(5313):800e6.

[4] Singhvi R, Kumar A, Lopez GP, Stephanopoulos GN, Wang DI, Whitesides GM,et al. Engineering cell shape and function. Science 1994;264(5159):696e8.

[5] Frimat JP, Menne H, Michels A, Kittel S, Kettler R, Borgmann S, et al. Plasmastencilling methods for cell patterning. Anal Bioanal Chem 2009;395(3):601e9.

[6] Bauwens CL, Peerani R, Niebruegge S, Woodhouse KA, Kumacheva E,Husain M, et al. Control of human embryonic stem cell colony and aggregatesize heterogeneity influences differentiation trajectories. Stem Cells2008;26(9):2300e10.

[7] Tan JL, Liu W, Nelson CM, Raghavan S, Chen CS. Simple approach to micro-pattern cells on common culture substrates by tuning substrate wettability.Tissue Eng 2004;10(5e6):865e72.

[8] Anderson DE, Hinds MT. Endothelial cell micropatterning: methods, effects,and applications. Ann Biomed Eng 2011;39(9):2329e45.

[9] Colpo P, Ruiz A, Ceriotti L, Rossi F. Surface functionalization for protein and cellpatterning. Adv Biochem Eng Biotechnol 2010;117:109e30.

[10] Nahmias Y, Odde DJ. Micropatterning of living cells by laser-guided directwriting: application to fabrication of hepatic-endothelial sinusoid-like struc-tures. Nat Protoc 2006;1(5):2288e96.

[11] Moraes C, Wang G, Sun Y, Simmons CA. A microfabricated platform for high-throughput unconfined compression of micropatterned biomaterial arrays.Biomaterials 2010;31(3):577e84.

[12] Ruiz SA, Chen CS. Emergence of patterned stem cell differentiation withinmulticellular structures. Stem Cells 2008;26(11):2921e7.

E. Vargis et al. / Biomaterials 35 (2014) 3999e40044004

[13] Tzvetkova-Chevolleau T, Stephanou A, Fuard D, Ohayon J, Schiavone P,Tracqui P. The motility of normal and cancer cells in response to the combinedinfluence of the substrate rigidity and anisotropic microstructure. Bio-materials 2008;29(10):1541e51.

[14] Chopdar A, Chakravarthy U, Verma D. Age related macular degeneration. BMJ2003;326(7387):485e8.

[15] Klettner A, Roider J. Treating age-related macular degeneration e interactionof VEGF-antagonists with their target. Mini Rev Med Chem 2009;9(9):1127e35.

[16] Kvanta A. Expression and regulation of vascular endothelial growth factor inchoroidal fibroblasts. Curr Eye Res 1995;14(11):1015e20.

[17] Ablonczy Z, Prakasam A, Fant J, Fauq A, Crosson C, Sambamurti K. Pigmentepithelium-derived factor maintains retinal pigment epithelium function byinhibiting vascular endothelial growth factor-R2 signaling through gamma-secretase. J Biol Chem 2009;284(44):30177e86.

[18] Ohno-Matsui K, Yoshida T, Futagami S, Yasuzumi K, Shimada N, Kojima A,et al. Patchy atrophy and lacquer cracks predispose to the development ofchoroidal neovascularisation in pathological myopia. Br J Ophthalmol2003;87(5):570e3.

[19] Ablonczy Z, Crosson CE. VEGF modulation of retinal pigment epitheliumresistance. Exp Eye Res 2007;85(6):762e71.

[20] Georgiadis A, Tschernutter M, Bainbridge JW, Balaggan KS, Mowat F, West EL,et al. The tight junction associated signalling proteins ZO-1 and ZONABregulate retinal pigment epithelium homeostasis in mice. PLoS One2010;5(12):e15730.

[21] Kliffen M, Sharma HS, Mooy CM, Kerkvliet S, de Jong PT. Increased expressionof angiogenic growth factors in age-related maculopathy. Br J Ophthalmol1997;81(2):154e62.

[22] Witmer AN, Vrensen GF, Van Noorden CJ, Schlingemann RO. Vascular endo-thelial growth factors and angiogenesis in eye disease. Prog Retin Eye Res2003;22(1):1e29.

[23] Chetelat G, Villemagne VL, Bourgeat P, Pike KE, Jones G, Ames D, et al. Rela-tionship between atrophy and beta-amyloid deposition in Alzheimer disease.Ann Neurol 2010;67(3):317e24.

[24] Chard DT, Griffin CM, Parker GJ, Kapoor R, Thompson AJ, Miller DH. Brainatrophy in clinically early relapsing-remitting multiple sclerosis. Brain2002;125(Pt 2):327e37.

[25] Ragnarsson O, Berglund P, Eder DN, Johannsson G. Long-term cognitive im-pairments and attentional deficits in patients with Cushing’s disease andcortisol-producing adrenal adenoma in remission. J Clin Endocrinol Metab2012;97(9):E1640e8.