Biomaterials 22 (2001) 1263}1269

Oriented Schwann cell growth on micropatternedbiodegradable polymer substrates

Cheryl Miller�, Howard Shanks�, Anthony Witt�, Gregory Rutkowski�,Surya Mallapragada��*

�Biomedical Engineering Program, Iowa State University, Ames, IA 50011, USA�Microelectronics Research Center, Iowa State University, Ames, IA 50011, USA

�Department of Chemical Engineering, Iowa State University, 1035 Sweeney Hall, Ames, IA 50011, USA

Received 22 May 2000; accepted 16 August 2000

Abstract

This paper investigates the in#uence of substrate-mediated chemical and physical guidance on the growth and alignment ofSchwann cells in vitro. Novel techniques were developed to fabricate microgrooves with adsorbed proteins on biodegradable polymersubstrates made of poly(D,L-lactic acid). Compression molding and solvent-casting were used to transfer micropatterns from quartzand silicon substrates onto biodegradable polymer "lms. Laminin was selectively adsorbed onto the grooves and rat sciatic Schwanncells were seeded on the substrates. Laminin was found to improve adhesion of Schwann cells on the substrates. The microgrooveswere found to cause the Schwann cells to align along the direction of the grooves. The groove width in#uenced Schwann cellalignment the most, while groove depth did not seem to play a signi"cant role. The degradation of the grooves in the solvent cast "lmswas much slower than those in the compression-molded "lms, making them the preferred substrates for Schwann cell cul-ture. � 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Schwann cells; Micropatterned polymers; Biodegradable polymers; Poly(D,L-lactic acid); Nerve regeneration

1. Introduction

The nervous system is composed of neurons and glial,or satellite cells. Glial cells in the peripheral nervoussystem are Schwann cells. The neurons carry signalsbetween the brain and the rest of the body, while theSchwann cells support the neurons while enhancing thespeed of electrical signals and produce proteins essentialfor neuron growth [1,2]. Because Schwann cells areknown to enhance the regeneration of axons [3}6], theymake an attractive addition to arti"cial nerve grafts [7].However, no work has been done in merging the use ofSchwann cells with micropatterned biodegradable poly-mer substrates to provide guidance to regenerating axonsat the cellular level.Various cells recognize three-dimensional geometricalcon"guration on the surface of substrates and their

*Corresponding author. Tel.: #1-515-294-7407; fax: #1-515-294-2689.E-mail address: suryakm@iastate.edu (S. Mallapragada).

growth can be guided and controlled by fabricatingmicrogrooves on substrate surfaces [8}11]. The cellsexhibit sensitivity to the dimensions of the microgrooves[12]. Most of the microgrooves have been fabricatedon inorganic substrates, such as micromachined siliconchips [13] which are not very desirable for implantation.Micropatterned regions on glass coverslips with adsor-bed laminin have been demonstrated to provide chemicalguidance for axonal outgrowth [14]. This paper focuseson methods to produce microgrooves on biodegradablepolymer surfaces, and the in#uence of physical as well aschemical guidance cues provided by these micropat-terned substrates on the growth and alignment ofSchwann cells in vitro.

2. Materials and methods

2.1. Microgrooved substrate fabrication

Poly(D,L-lactic acid) (PDLA) (Birmingham Polymers,AL) with an inherent viscosity of 0.69 dl/g in chloroform

0142-9612/01/$ - see front matter � 2001 Elsevier Science Ltd. All rights reserved.PII: S 0 1 4 2 - 9 6 1 2 ( 0 0 ) 0 0 2 7 8 - 7

Fig. 1. Micropatterning of quartz templates: (a) quartz slide ( ); (b)micropatterned chrome mask ( ) placed on the quartz slide and sub-jected to reactive ion etching; (c) etched quartz slide with chrome ontop; (d) etched quartz slide with the chrome removed. ( ) represents thegrooves in the quartz slide.

at 303C was chosen for fabricating the substrates as it isa resorbable, biocompatible polymer that is used exten-sively in biomedical applications. Conventional micro-lithographic techniques cannot be directly applied tothese biodegradable polymers because the processingtechniques accelerate the degradation process. Micro-grooved PDLA "lms were produced by imprinting pat-terns onto the polymer surface from a microgroovedquartz substrate.A chrome lithography mask was produced with thedesired micron scale patterns using conventional litho-graphic techniques. The mask was deposited onto a 2-in.diameter quartz slide or a 4-in. diameter silicon wafer ina vacuum chamber purged with argon at a pressure ofless than 1�Torr. Quartz substrates were etched usingreactive ion etching (RIE) through the mask, leavingbehind long rectangular areas capped by chrome (Fig. 1).RIE is a dry developed process with an etch rate of about30nm/min and a resolution of 1.5�m. After the chromewas removed from the quartz substrate, the quartz sub-strate was used as a micro die to transfer the geometricmicrogrooves to the biodegradable polymer.Compression-molded "lms of PDLAwere obtained bycompressing solid PDLA powder in a Carver press at500 psi pressure at 603C for 20min. They were thenplaced on the quartz die and heated between two platesat 503C for 5min. Solvent cast "lms were produced bycasting solutions of PDLA in chloroform and spin cast-ing them onto the micropatterned silicon wafers. Anaqueous PVA (Elvanol�, duPont, Wilmington, DE) solu-tion (6%w/v) was spun on a Si wafer at 4000 rpm for1min and allowed to dry for 4 h. The PDLA solution inchloroform (30%w/v) was then spun on the substrate at1800 rpm for 1min. After drying for 24 h, these "lms were#oated o! the silicon wafers onto the surface of water andused.The substrates (1 cm�1 cm surface area) were immer-sed in 5ml of Dulbecco's Modi"ed Eagle's Medium(DMEM) for varying durations at 373C and examinedusing scanning electron microscopy (SEM) to ascertainthe degradation of the substrates with time. The substra-tes were sputter coated (SEM Coating Unit E5100 SeriesII, Polaron Instruments Inc., Wartford Hertfordshire,UK) with 35nm of conductive metal (60% palladium,40% gold). The substrates were imaged with a JEOLJSM-5800 low-vacuum SEM.

2.2. Laminin adsorption

A solution of laminin (Sigma, St. Louis, MO) in phos-phate-bu!ered saline (PBS) (100�l laminin/ml PBS) wasadsorbed onto the surface of the micropatterned substra-tes by placing a drop on the substrate. After 10min, thelaminin was removed with a pipette placed perpendicularto the surface. The excess laminin was allowed to dryfor 4 h. It was hypothesized that the laminin would

be mainly concentrated in the grooves due to surfacetension e!ects.To observe the laminin distribution on the surface,83�g/ml FITC-anti laminin conjugate solution wasadded to the substrate and incubated for 1 h at 373C.The substrate was then rinsed with PBS and 25�g/mlAlexaFluor 488 rabbit anti-FITC IgG antibody solution(Molecular Probes, Eugene, OR) was added. The sub-strate was incubated again for 1 h at 373C and rinsed withPBS. The same procedure was repeated after adding15�g/ml AlexaFluor 488 goat anti-rabbit IgG conjugatesolution to the substrate and the #uorescence on thesubstrate was examined using a microscope witha 490nm cuto! "lter to observe the laminin distributionon the surface.

2.3. Schwann cell isolation

Schwann cells obtained from continuous cell lines arevery di!erent from cells isolated from primary cultures.Continuous cell lines exhibit a non-Schwann cell-likemorphology and do not associate with neurons [15].Therefore primary cultures were used for the studies.Schwann cells were obtained using the method developedby Morrisey and co-workers [16]. Sprague}Dawley rats,16}20 days old, were anesthetized and then decapitated.The skin was cut back revealing muscle tissue by the hindlimbs. The sciatic nerve was removed and kept in chilledGey's balance salt solution (Gibco) supplemented with6.5mg/ml glucose (Sigma). The epineurium, connectivetissue, and blood vessels were removed using "ne forcepsand the nerve was cut into pieces approximately 1mm inlength.The nerve pieces were placed in 60mm tissue cultureplates (Fisher Scienti"c). Approximately 0.5ml of chickenplasma solution (5mg/ml) was spread evenly over thebottom of the plate. Nerve pieces were placed in the dish

1264 C. Miller et al. / Biomaterials 22 (2001) 1263}1269

spaced 2mm apart. One hundred microliters of thrombin(Sigma) (10 units/ml) was added to the plate and mixedwith chicken plasma. Fibrin from the chicken plasma wasallowed to clot for about 5min before adding media. Themedia used for the cell cultures was DMEM (Gibco) with10% v/v fetal bovine serum (FBS) (Gibco) and 5�l gen-tamicin/ml medium (Sigma).Media was changed every 2 days. As the nerve piecesdegenerated, "broblasts began to spread onto the plate.When enough "broblasts spread onto the plate in 5}7days, the nerve pieces were transplanted onto new tissueculture plates using the above procedure. After 3 or4 transplantations, the cells spreading from the nervepieces consisted mostly of Schwann cells. At this pointthe nerve pieces were dissociated and then incubated for1 h at 373C. The media and cell debris were removed andthe remaining cells were resuspended in 1ml of freshmedia. The cell suspension was placed in 75 cm� tissueculture #ask (Fisher Scienti"c) and enough media wasadded to make 10ml.

2.4. Schwann cell purixcation and seeding

Schwann cell cultures were puri"ed using a methodadapted from the work of Rutkowski and co-workers[17]. The puri"cation was achieved by removing the cellsfrom a 75 cm� T-#ask and adding 2ml anti-Thy 1.1media to the Schwann cells for 30min and adding 1.5mlrabbit sera complement for another 30min. The cellswere resuspended and fed with DMEM/10% FBS sup-plemented with 0.5mM forskolin, 0.5mM isobutylmethyl-xanthate (IBMX) and 0.1�g/ml human heregulin-�1(EGF domain) (R&D Systems). This combinationreduces "broblasts while greatly enhancing the prolifer-ation of the Schwann cells. Cells were not passaged morethan 2 times because subculturing the cells more thantwice leads to immortalization of Schwann cells [18].Schwann cell numbers were determined by trypan blueexclusion using a hemacytometer. Counts were done intriplicate and averages for total cell numbers and viabil-ity were calculated.Schwann cells were seeded onto the etched poly-mer surfaces at initial densities varying from 5000 to400,000 cells/cm� and incubated at 373C. Photomicro-graphs were taken daily and positioning of the Schwanncells along the grooves was veri"ed by microscopy.DiIC

��(3) (Molecular Probes, Eugene, OR) (3.7�l/ml of

media) was seen to be quite e!ective in staining the cellsand helping to distinguish them from the background bythe #uorescence. SYTO23 (Molecular Probes, Eugene,OR) (3.7�l/ml of media) was also used to stain theSchwann cells.The e!ect of substrate processing conditions, initialcell seeding density, microgroove width and depth, andthe laminin in the grooves, on the adhesion and align-ment of Schwann cells was ascertained. Alignment was

determined by whether the Schwann cell maintainedalignment in the direction of the groove without crossingfrom one groove to another. The cells were counted onan Olympus IMT2-RFC microscope using an IMT2-DMGdichroic mirror unit since DiI has an absorption of549 nm and a #uorescence emission of 565 nm. TheSchwann cells were counted for cells aligned in the direc-tion of the grooves and cells not aligned using a simpleF-test to compare the two population variances.

3. Results and discussion

An e!ective technique was developed for fabricatingmicrogrooves with various pattern sizes and spacings onbiodegradable PDLA "lms using both compressionmolding as well as solvent casting techniques. To ourknowledge, this is the "rst time that this simple bute!ective technique has been used for fabricating pat-terned "lms of biodegradable polymers with selectiveadhesion of proteins. These techniques can be used formost biodegradable polymers. SEM images of the quartzdie and the microgrooved compression-molded polymersubstrates are shown in Fig. 2a and b. The averagethickness of the compression-molded polymer "lms was0.5mm. The average thickness of microgrooved solventcast polymer "lms (Fig. 2c) was about 200 �m.Due to the slightly elevated temperatures and pres-sures that the compression-molded "lms were subjectedto, they were found to degrade much faster than thesolvent cast "lms. This degradation is seen from the SEMimages of the substrates immersed in media at 373C forvarious durations. SEM images of compression-moldedand solvent-cast "lms with initial groove depths of 3 �mimmersed in media for 2 days are seen in Fig. 3a and b. Inthe compression-molded "lms, the microgrooves haveeroded to a signi"cant extent and the groove depth ismuch smaller than the original depth of 3�m. However,in the same time period, the solvent cast "lms show littleor no degradation. At longer time periods of 2 weeks, thegrooves completely disappear in the compression-molded "lms while the solvent cast "lms retain thegroove size and shape (Fig. 4a and b). It was found thatthe groove depth does not decrease signi"cantly for thesolvent cast "lms until after 28 days (Fig. 5). If these "lmsare to be e!ective in guiding peripheral nerve regenera-tion [19], the grooves are required to provide axonalalignment for periods of longer than 4 weeks. Based onthis requirement, the solvent cast "lms were found to bepreferable compared to the compression-molded "lmsand were used for most of the experiments. However, thecompression-molded "lms are useful in investigating thecontinued alignment of cells even after the grooves disap-pear due to degradation. Moreover, it is very di$cult toproduce uniformly thick solvent-cast "lms using thistechnique. Therefore the compression-molding technique

C. Miller et al. / Biomaterials 22 (2001) 1263}1269 1265

Fig. 2. SEM images of (a) patterned quartz die with dimensions of 10/20/3.3�m (bar"20 �m) and (b) compression-molded PDLA "lm with groovedimensions of 10/20/3.3�m (bar"10�m) and (c) solvent cast PDLA "lm with groove dimensions of 10/20/4�m (bar"200�m). The dimensionsrepresent width of the groove/spacing between the grooves/depth of the grooves in �m.

Fig. 3. SEM images of (a) compression-molded (bar"10 �m) and (b) solvent cast PDLA "lms (bar"10�m) with initial groove dimensions of10/20/3.3�m after 2 days in culture.

can be used in applications where thicker "lms arerequired.The adhesion of Schwann cells to smooth substrateswith and without adsorbed laminin on the surface wascompared. The presence of laminin (100�g/ml of media)was found to improve the initial cell adhesion signi"-cantly. 29450$1560 cells adhered initially to each sub-strate adsorbed with laminin and 3700$165 cellsadhered to each substrate without laminin (standard

errors for 18 substrates are reported here). As hypo-thesized, the laminin adsorption technique concentratedthe laminin in the grooves as seen in Fig. 6 and helpedincrease cell adhesion in the grooves. All further studieswere conducted with laminin adsorbed in the grooves.Schwann cells were seeded on the polymer substratesand microgroove dimensions such as groove width anddepth were optimized. The groove width is the key para-meter for alignment of the Schwann cells. The width of

1266 C. Miller et al. / Biomaterials 22 (2001) 1263}1269

Fig. 4. SEM images of (a) compression-molded (bar"10 �m) and (b) solvent cast PDLA "lms (bar"10�m) with initial groove dimensions of10/10/3.3�m after 2 weeks in culture.

Fig. 5. SEM image of a solvent cast PDLA "lm with initial groovedimensions of 10/20/2�m after 4 weeks in culture. Bar"10�m.

Fig. 6. Distribution of laminin in microgrooves on a solvent castPDLA "lm with groove dimensions of 10/20/3�m. Bar"30�m.

Schwann cells varies from 5 to 10 �m and patterns widthsor spacings of 10}20�m were found to be optimal for thealignment of Schwann cells. Smaller widths did not pro-mote orientation because the groove widths were smallerthan the cells, while larger widths were much larger thanthe cell size and therefore did not promote alignment.Substrates with groove widths of 10�m and groovedepths of 1.4, 1.5, 18, 1.9, 2.3, 3.1 and 3.3�m were fab-ricated and seeded with Schwann cells. For cell seedingdensities of 50,000 cells/cm�, 100% alignment of theSchwann cells along the grooves (Fig. 7) was observed forgroove depths varying from 1.4 to 3.3�m. This alignmentwas maintained on the solvent cast "lms even after2 weeks in culture. In contrast, 72% alignment alonga major axis was observed for Schwann cells seeded onsmooth substrates (Fig. 8). This is due to the inherenttendency of Schwann cells to form columns [20]. How-ever, the direction of orientation cannot be controlled onsmooth substrates. Therefore the grooves were found tosigni"cantly promote the alignment of Schwann cells inthe desired directions.

The initial cell seeding density was found to have ane!ect on the alignment of the Schwann cells. Initial cellseeding densities of 50,000 cells/cm� were found to cause100% alignment of the Schwann cells, while lower seed-ing densities were not as e!ective in promoting align-ment. Substrates with groove widths of 10�m and groovedepths of 1.9�m seeded with Schwann cells at an initialseeding density of 25,000 cells/cm� had 87$10.5%(standard deviation for 12 substrates reported here) of thecells aligned along the grooves. This lower alignment rateis probably due to lesser probability of cell-to-cell com-munication caused by lower cell densities.DiI and SYTO23 were found to be very e!ective instaining Schwann cells and distinguishing them from thebackground by #uorescence. Fig. 9 shows a substrateseeded with Schwann cells with horizontal grooves onthe left side of the substrate and with no grooves on theright side. The Schwann cells are stained with SYTO23.The Schwann cells on the left side of the substrate areall oriented along the grooves, while the cells on theright side are not oriented in any particular direction.

C. Miller et al. / Biomaterials 22 (2001) 1263}1269 1267

Fig. 7. Alignment of Schwann cells along microgrooves on a PDLAcompression-molded "lm (initial groove dimensions of 10/20/1.5�m)after 2 days in culture. Bar"50 �m.

Fig. 8. Schwann cells on a smooth compression-molded PDLA sub-strate after 2 days in culture. Bar"50�m.

Fig. 9. Relative alignment of Schwann cells stained with SYTO23 ona PDLA substrate with microgrooves (10/10/4�m) to the left of thearrow and a smooth surface to the right of the arrow. Image was takenafter 2 days in culture. Bar"60 �m.

Therefore, the microgrooved substrates with adsorbedlaminin were found to be e!ective in promoting theadhesion and alignment of Schwann cells in vitro.

4. Conclusions

Novel patterning techniques based on reactive ionetching coupled with compression molding or solventcasting were developed to produce microgrooved biode-gradable polymer substrates. The microgrooves degrademore slowly in solvent-cast "lms compared to the com-pression-molded "lms. Laminin adsorbed to the bottomof the grooves was found to increase the cell adhesionsigni"cantly. The fabrication technique used was foundto concentrate the laminin adsorption in the grooves.Groove width was found to be a signi"cant factor inpromoting Schwann cell alignment, and widths and spac-ings of 10}20�m were found to be optimal. Groovedepths of 1.4�m or greater with initial cell seeding densit-

ies of greater than 50,000 cells/cm� caused completelong-term alignment of the Schwann cells along thegrooves. Lower initial cell seeding densities were not ase!ective in promoting Schwann cell alignment due todecreased cell-to-cell communications. Microgroovedbiodegradable polymer substrates selectively adsorbedwith laminin were found to provide physical and chem-ical guidance to Schwann cells and enhance theiradhesion and alignment on the substrates in vitro. Thesesubstrates are of potential signi"cance in promoting peri-pheral nerve regeneration.

Acknowledgements

Financial support from an NSF POWRE grant (BES9973287) (to SKM) and a grant from the US Departmentof Energy (W-7405-Eng-82) is gratefully acknowledged.The authors would also like to thank Dr. Srdija Jeftinijain the Neuroscience Program at Iowa State Universityand Dr. Carole Heath at Immunex for helpful commentsand suggestions.

References

[1] Bunge RP. Expanding roles for the Schwann cell: ensheathment,myelination, trophism and regeneration. Curr Biol 1993;3:805}9.

[2] Tortora GJ. In: Principles of human anatomy: nervous system.New York: Harper Collins, 1992. p. 459}66.

[3] Guenard V, Kleitman N, Morrissey TK, Bunge RP, Aebischer P.Syngeneic Schwann cells derived from adult nerves seeded insemipermeable guidance channels enhance peripheral nerve re-generation. J Neurosci 1992;12:3310}20.

[4] Kleitman N, Bunge RP. The Schwann cell: morphology anddevelopment. In:Waxmann SG, Kocisis JD, Stys PK, editors. Theaxon: structure, function, and pathophysiology. London: OxfordUniversity Press, 1995. p. 97}115.

1268 C. Miller et al. / Biomaterials 22 (2001) 1263}1269

[5] Berthold CH, Rydmark C. Morphology of a normal peripheralnerve. In: Waxmann SG, Kocisis JD, Stys PK, editors. The axon:structure, function, and pathophysiology. London: Oxford Uni-versity Press, 1995. p. 13}48.

[6] Salzer JL. Mechanisms of adhesion between axons, glial cells. In:Waxmann SG, Kocisis JD, Stys PK, editors. The axon: structure,function, and pathophysiology. London: Oxford University Press,1995. p. 164}84.

[7] Heath CA, Rutkowski G. Bioarti"cial nerve grafts in peripheralnerve regeneration. Trend Biotechnol 1998;16:163}8.

[8] Weiss P. Experiments of cell and axon orientation in vitro: therole of colloidal exudates in tissue organization. J Exp Zool1945;63:401}50.

[9] Turner DC, Lawson J, Dollenmeier P, Ehrissman R, Chiquet M.Guidance of myogenic cell migration by oriented deposits of"bronectin. Dev Biol 1983;95:497}604.

[10] Clark P, Connolly P, Curtis SG, Dow JAT, Wilkinson CDW.Topographical control of cell behavior. I. Simple step cues.Develop 1987;99:439}48.

[11] Dow JA, Clark P, Connolly P, Curtis ASG, Wilkinson CDW.Novel methods for guidance and monitoring of single cell andsimple networks in culture. J Cell Sci 1987;8:55}79.

[12] Clark P, Britland S, Connolly P. Growth cone guidance andneuron morphology on micropatterned laminin surfaces. J CellSci 1993;105:203}12.

[13] Zhao Q, Drott J, Laurell T, Wallman L, Lindstrom K, BjurstenLM, Lundborg G, Montelius L, Danielsen N. Rat sciatic nerveregeneration through micromachined silicon chip. Biomaterials1997;18:75}80.

[14] Tai HC, Buettner HM. Neurite outgrowth and growth conemorphology on micropatterned surfaces. Biotechnol Prog 1998;14:364}70.

[15] Yoshimura T, Kobayashi T, Goda S, Goto I. Inhibition of theproliferation of cultured immortalized Schwann cells by forskolinwith a decreased basal level of diacycloglycerol. Neurochem Res1994;19:735}41.

[16] Morrissey T, Kleitman N, Bunge R. Isolation and functionalcharacterization of Schwann cells derived from adult peripheralnerve. J Neurosci 1991;11:2433}42.

[17] Rutkowski JL. Puri"cation and expansion of human Schwanncell in vitro. Nature Med 1995;1:80}3.

[18] Eccleston PA, Mirsky R, Jessen KR. Spontaneous immortaliz-ation of Schwann cells in culture: short-term cultured Schwanncells secrete growth inhibitory activity. Develop 1991;112:33}42.

[19] Mallapragada SK, Heath CA, Shanks H, Miller CA. Patternedconduits and methods for nerve regeneration. US ProvisionalPatent No. 60/198,370 (pending).

[20] Gilmore SA, Sims TJ. Glial cells and regeneration in the peri-pheral nervous system. In: Ketternmann H, Ransom BR, editors.Neuroglia. London: Oxford University Press, 1995. p. 829}41.

C. Miller et al. / Biomaterials 22 (2001) 1263}1269 1269