lable at ScienceDirect

Biomaterials xxx (2010) 1e9

Contents lists avai

Biomaterials

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

Dynamic cell behavior on shape memory polymer substrates

Kevin A. Davis a,b, Kelly A. Burke b,c, Patrick T. Mather a,b, James H. Henderson a,b,*

aDepartment of Biomedical and Chemical Engineering, 121 Link Hall, Syracuse University, Syracuse, NY 13244, USAb Syracuse Biomaterials Institute, 318 Bowne Hall, Syracuse University, Syracuse, NY 13244, USAcDepartment of Macromolecular Science and Engineering, 2100 Adelbert Road, Case Western Reserve University, Cleveland, OH 44106, USA

a r t i c l e i n f o

Article history:Received 15 November 2010Accepted 1 December 2010Available online xxx

Keywords:Cell cultureShape memoryThermally responsive materialSurface topography

* Corresponding author. Syracuse BiomaterialsSyracuse University, Syracuse, NY 13244, USA. Tel.: þ443 7724.

E-mail address: jhhender@syr.edu (J.H. Henderson

0142-9612/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.biomaterials.2010.12.006

Please cite this article in press as: Davis KA, ej.biomaterials.2010.12.006

a b s t r a c t

Cell culture substrates of defined topography have emerged as powerful tools with which to investigatecell mechanobiology, but current technologies only allow passive control of substrate properties. Here wepresent a thermo-responsive cell culture system that uses shape memory polymer (SMP) substrates thatare programmed to change surface topography during cell culture. Our hypothesis was that a shape-memory-activated change in substrate topography could be used to control cell behavior. To test thishypothesis, we embossed an initially flat SMP substrate to produce a temporary topography of parallelmicron-scale grooves. After plating cells on the substrate, we triggered shape memory activation usinga change in temperature tailored to be compatible with mammalian cell culture, thereby causingtopographic transformation back to the original flat surface. We found that the programmed erasure ofsubstrate topography caused a decrease in cell alignment as evidenced by an increase in angulardispersion with corresponding remodeling of the actin cytoskeleton. Cell viability remained greater than95% before and after topography change and temperature increase. These results demonstrate control ofcell behavior through shape-memory-activated topographic changes and introduce the use of active cellculture SMP substrates for investigation of mechanotransduction, cell biomechanical function, and cellsoft-matter physics.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Cells are capable of surveying the mechanical properties of theirsurrounding environment [1]. This capacity is critical to events ofembryogenesis, postnatal tissue development and maintenance,and the onset of certain pathological conditions [2e4]. In thesynthetic realm, substrates of defined topography or elasticmodulus have emerged as powerful tools in the investigation of theunderlying cellular mechanisms. In particular, mesoscale, micro-scale, and nanoscale patterns of substrate topography have beenshown to direct cell alignment, cell adhesion, and cell tractionforces [5e12], while isotropic and anisotropic substrate elasticityhave been used to direct cell lineage specification, growth, andmigration [1,13,14]. Defined topographies have also found applica-tion in biomedical device design, for example to enhance implantcell adhesion and tissue ingrowth [15]. These findings haveunderscored the potential for substrates to control and assay themechanical interactions between cells and their physical

Institute, 318 Bowne Hall,1 315 443 9739; fax: þ1 315

).

All rights reserved.

t al., Dynamic cell behavior on

environment during cell culture; however, the substrates used todate have generally been passive, with substrate properties thatcould not be programmed to change significantly during culture.This physical stasis has limited the potential of substrates to controlcells in culture and, therefore, to advance both fundamentalunderstanding of cell biology and cell-based bioengineeringapplications.

To overcome the limitations of physically static systems, therehas been growing interest in the development of stimuli-respon-sive biomaterials for cell culture, with several notable areas ofprogress. Poly(N-isopropylacrylamide) (PIPAAm) has been used torelease cell layers [16] that can be used in laminar tissue engi-neering [17] by taking advantage of the material’s lower criticalsolution temperature (LCST) at 32 �C and a change in hydropho-bicity around this temperature. PIPAAm’s thermo-responsivebehavior has also been exploited in copolymer hydrogels designedto apply equibiaxial stretching to encapsulated cells when thetemperature is lowered to 25 �C [18]. Elastin-like polypeptides(ELPs) are stimuli-responsive biomaterials that can exhibit an LCSTin a physiologically relevant range [19] and have been used tocapture and release proteins at surfaces [20], which can be used tocontrol cell adhesion. Thermo-responsive ELP hydrogels for cellculture or tissue engineering applications have also been

shapememory polymer substrates, Biomaterials (2010), doi:10.1016/

K.A. Davis et al. / Biomaterials xxx (2010) 1e92

developed [21,22]. To allow for spatial control of responsivebiomaterials, UV-induced crosslink degradation has been used ina photo-responsive biomaterial to investigate the effects ofsubstrate modulus change during cell culture [23]. This approachhas been extended to three-dimensional spatial control with thedevelopment of a PEG-based hydrogel and two-photon photodegradation system to study cell response to selective removal ofcell-adhesion sites [24].

Exploitation of shape memory polymers (SMPs) has also beenproposed as a possible solution to current limitations of static,unresponsive substrates, medical devices [25], and tissue engi-neering scaffolds [26,27]. SMPs are a class of active materials thathave the ability to memorize a permanent shape through cross-linking, be manipulated and then fixed to a temporary shape by animmobilizing transition, such as vitrification or crystallization, andthen later recover to the permanent shape by a triggering event,such as a thermal, electrical, or solvent activation [28]. Thisphenomenon is known as one way shape memory, since activity iswitnessed in one direction. Although most previous work hascentered on bulk one way shape memory behavior, surface (ortopographical) shape memory is also possible [29,30]. In the case ofthermal triggering, an SMP frozen in a stressed configuration willreturn to its equilibrium configuration under the action of network(entropic) elasticity upon heating through the glass transition ormelting point, termed the recovery temperature. Traditionalbiocompatible SMPs, including crosslinked (to confer shapememory properties) PCL, PLGA, and polystyrene, feature recoverytemperatures too high for cells to maintain viability during thethermal trigger [28]. Considerable effort has been focused on thisissue of thermal cytocompatibility and, very recently, havingobserved dynamic mechanical characteristics indicating thepotential for shape memory behavior [31], one of us has reportedbulk and surface shape memory behavior of a commercially avail-able, non-cytotoxic [32] optical adhesive [33], Norland OpticalAdhesive 63 (NOA-63, Norland Products, Cranbury, NJ, USA), witha recovery temperature that can be tuned by the crosslinkingconditions [33].

The goal of this study was to develop temperature-responsiveSMP substrates that can be programmed to change topographyunder cell culture conditions.We further sought to test theworkinghypothesis that a shape-memory-activated change in substratetopography can be used to control cell behavior. To test thishypothesis, we programmed SMP substrates to transition froma micron-scale grooved surface to a flat surface and analyzed cellalignment before and after the topographic transition.

2. Materials and methods

2.1. Substrate preparation

To enable surface shape memory during cell culture, we prepared a glassy SMPfrom the NOA-63 formulation (Lot # 111). To prepare shape memory substrates intheir permanent, flat surface, NOA-63 was used as received and injected betweenan aluminum block and a glass slide with a 1 mm thick Teflon spacer(Supplementary Data Fig. S1). NOA-63 is supplied as a photocrosslinkable solvent-free prepolymer with a photoinitiator and, although its exact composition is notdisclosed by the supplier, it is known to be a polyurethane that is end-linked bya thiol-based crosslinker using photoinitiated thiol-ene crosslinking [34]. NOA-63exhibits one way shape memory around its glass transition temperature (Tg;Supplementary Data Fig. S2) [33]; moreover, the Tg of NOA-63 can be manipulatedby controlling the temperature during photo-curing, which alters the vitrificationevent. The effective Tg of NOA-63 when wet (e.g., during cell culture) is lower thanthe dry Tg due to plasticization by water. To address the shift in Tg caused byhydration, we adjusted the wet Tg to fall in a cell-compatible range, approximately30 �Ce37 �C, by choosing a cure temperature of 125 �C, which resulted in a dry Tg(51 �C) above the cell-compatible range. The prepolymer was photocured for20 min on a hot plate at 125 �C with a 1000 W Spectroline SB-100PC 365 nm UVlight source placed 6.5 cm above the top glass surface of the mold described above.The film was removed from the mold, placed on the hot plate, and post-cured with

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior onj.biomaterials.2010.12.006

heat and UV light for an additional 3.75 h with the light placed 6.5 cm above theNOA-63 surface. The cured material was stored in a desiccator at �20 �C untilfurther processing. Next, the NOA-63 films were readied for embossing by cuttinginto approximately 3 mm � 3 mm pieces. These pieces were arranged on an epoxyembosser (Supplementary Data Fig. S3), which was the negative impression ofa vinyl record (Supplementary Data Method 1), and compressed using a Carver3851-0 hydraulic press with heating platens. A compressive stress of 4.9 MPa wasapplied to the samples at 90 �C and held for 1 min. Water cooling of the platenswas then started and the embossing stress was released at 20 �C, after approxi-mately 5 min of cooling. Samples prepared in this manner thus had a temporarilygrooved topography that could be triggered to transition to the permanent flatsurface by heating. To produce a static control substrate with a permanentlygrooved topography (i.e., to produce a substrate that has the grooved topographyafter curing, instead of after embossing into a temporary shape), we replicatedequilibrated samples (described under Substrate Surface Equilibration) in silicone(Supplementary Data Method 2), which then served as a mold to preparepermanently grooved SMPs. NOA-63 was poured onto the mold and curedfollowing the heat and UV protocol described above. Finally, to produce a staticcontrol substrate with a flat topography, temporarily grooved samples wererecovered in air at 90 �C for 5 min before sterilization to recover the permanent flattopography. Sample topography was measured by profilometry (SupplementaryData Method 3) before sterilization.

2.2. Substrate surface equilibration

SMPs based on Tg triggers (Class I SMPs [35]) exhibit a temperature range overwhich recovery from the temporary to the permanent shape can occur. For a ther-mally triggered SMP surface to function as an active cell culture substrate, thesubstrate must maintain a stable temporary topography during an initial phase ofcell culture and then recover its permanent topography following a temperatureincrease. We developed such an equilibration protocol and found that approxi-mately 30 h of equilibration at 30 �C in complete growth medium (but before cell-seeding) was required to reach a relatively stable topography where topographicfeatures of initial amplitude 25.6 � 0.8 mm recovered to 12.6 � 1.5 mm. After another28.5 h at 30 �C the amplitude decreased 4.5 mm to 8.1�1.1 mm (Supplementary DataFig. S4) for an average rate of change of 0.16 mm h�1.

2.3. Cell culture

Prior to active cell culture experiments, C3H/10T1/2 mouse embryonic fibro-blasts were expanded in complete growth medium: Basal Medium Eagle supple-mented with 10% fetal bovine serum (Invitrogen, Gibco, Lot #502439), 1% penicillin/streptomycin, and 1% GlutaMAX (Invitrogen) in a 37 �C humidified incubator with5% CO2. Cells were cultured on a T182 flask with 35 mL of complete growthmedium.Themediumwas changed every four days and cells were passaged at 80% confluenceusing 0.25% trypsin-EDTA solution. Cells were collected at passage 14 for active cellculture experiments.

2.4. Active cell culture experiments

To determine whether a change in topography can be used to control cellbehavior, we plated C3H/10T1/2 mouse embryonic fibroblasts on the temporarytopography and analyzed cell alignment before and after triggering the shapememory activated topographic transition. Temporarily grooved samples, groovedcontrol samples, and flat control samples were cleaned by vortexing in 70 % ethanolfor 20 s and then sterilized by exposure to UV light for 12 h (Supplementary DataMethod 4). Samples were then placed in a 96-well plate and 150 mL of completegrowth medium was added. The plates were placed in a 30 �C incubator for 30 h toequilibrate the temporarily grooved samples. After 30 h, the existing medium wasremoved and replaced with 150 mL of a cell solution at a concentration of20,000 cells mL�1 in complete growth medium. Cells on temporarily groovedsamples, grooved control samples, and flat control samples were stained and imagedafter one of two endpoints: before transition (9.5 h in a 30 �C incubator) and aftertransition (9.5 h in a 30 �C incubator followed by 19 h in a 37 �C incubator). Threecured NOA-63 substrates were prepared identically. For each such substrate, twosets of temporarily grooved samples were embossed by independent shape fixingcycles. The active cell culture experiment was run for each cycle of each substrate foran n of 6, except where contamination caused a reduction in sample number asindicated.

2.5. Cell alignment quantification

Cells were stained with Cellmask Deep Red Plasma Membrane stain (Invi-trogen). Cellmask stain was used at 5 mL mL�1 and all incubation steps were per-formed at 30 �C. Fixed Cellmask stained samples were mounted in Prolong Gold(Invitrogen). Cell morphometric data was quantified by manually tracing celloutlines (30 isolated cells per replicate) from Cellmask stained samples using ImageJv1.43 (U.S. NIH http://rsb.info.nih.gov/ij/). The traces were filled and binary imageswere created. The Analyze Particles function was used to determine cell area (A),

shapememory polymer substrates, Biomaterials (2010), doi:10.1016/

9.5 h 30 °C 19 h 37 °C

Equilibrated shape Before transition After transition

0 0 1200 µm

15 µm

5kV x140 100µm 15 26 SEI 5kV x140 100µm 15 26 SEI 5kV x140 100µm 15 26 SEI

Fig. 1. Active cell culture substrate transitions from a grooved topography to a flat surface. Cells were plated on an equilibrated temporarily grooved sample and allowed to attachfor 9.5 h. The substrate was then triggered to transition to the flat surface by moving the samples to a 37 �C incubator. After 19 h, the cells were assayed again to detect any changesdue to the topographic transition. Traces below images are profilometry scans of representative samples. SEM images are representative samples without cells to highlight thechange in topography.

Fig. 2. Change in substrate topography controls cell behavior. a, Cells stained with Cellmask plasma membrane stain on a temporary grooved topography are aligned with groovedirection (white arrow) before transition. After transition, cells are randomly oriented. The angular histogram shows small dispersion and long average mean resultant vector length(orange line extending radially from origin), R, before transition. After transition, cell angles are widely dispersed and the average R has shortened significantly (p ¼ 0.001, n ¼ 6).Bins are 12� and bin length is normalized by the bin with the highest frequency of cell angles. Color indicates contribution from each sample (n ¼ 6). Scale bar is 200 mm. Traces areas in Fig. 1. b, Cells on flat control substrates are randomly oriented before and after temperature increase with large dispersion and short average R (n ¼ 6). c, Cells on groovedcontrol substrates are aligned with groove direction before and after temperature increase with small dispersion and long average R (n ¼ 4).

K.A. Davis et al. / Biomaterials xxx (2010) 1e9 3

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior on shapememory polymer substrates, Biomaterials (2010), doi:10.1016/j.biomaterials.2010.12.006

K.A. Davis et al. / Biomaterials xxx (2010) 1e94

perimeter (p), shape index (p2/[4pA]), and angle. Cell angles were defined from 0� to180� . Cell angles on grooved samples were rotationally shifted so that the groovedirection corresponded to an angle of 90� . Cell alignment was quantified by themean resultant vector length (R) calculated using the vector summation of indi-vidual cell angles treated as unit vectors and normalized by total cell number [36].Uniformly distributed cell angles feature an R of w0.63 when defined from 0� to180� , and cells aligned parallel to a biasing angle have an R of 1.0.

2.6. Cell viability

Cells were stained with LIVE/DEAD cell viability stain (Invitrogen). LIVE/DEADstain was used at a concentration of 2 mM for both Calcien AM and EthD-1. Sampleswere stained following manufacturer protocol, except that the incubation stepswere performed at 30 �C. Cell viability was calculated by counting the number of liveand dead cells for a 2052 mm� 1539 mm area of the sample and dividing the numberof live cells by the total number of cells.

2.7. Actin cytoskeleton staining

Cells were stained with Alexa Fluor 647 conjugated phalloidin (Invitrogen),enabling imaging of the actin cytoskeleton. Fixed phalloidin stained samples weremounted in Prolong Gold.

2.8. Cell imaging

LIVE/DEAD and Cellmask stained cells were imaged on a Leica DMI 4000Binverted microscope with a Leica DFC 340FX camera using a 5x/0.15 NA objective.Phalloidin stained cells were imaged on a Zeiss LSM 710 confocal laser scanningmicroscope using a 20x/0.8 NA air or a 40x/1.30 NA oil objective with Zeiss Immersol518 F immersion oil. ImageJ was used to apply the red or green lookup table to allmicrographs. Histogram stretching was applied to Cellmask images in order tomaximize image contrast.

2.9. SMP recovery kinetics

Temporarily grooved samples were incubated simultaneously with cell-seededsamples and removed from the heated incubators at the pre-selected intervals todetermine substrate topography recovery kinetics (Supplementary Data Fig. S4).Topographic amplitude was measured for each sample by profilometry before andafter thermal treatment (Supplementary Data Method 3). Relative amplitude wascalculated as the average amplitude measured after thermal treatment divided bythe average amplitude measured after embossing, this measure ranging from 1.0 forno recovery to 0.0 for complete recovery. Representative samples were also imagedusing scanning electron microscopy (Supplementary Data Method 5) for qualitativecomparison.

2.10. Statistics

The 95% confidence intervals of the mean average resultant vector lengthwere calculated using a bootstrap method (Supplementary Data Method 6). A twofactor generalized linear model permutation-based ANOVA was used to compareall groups with controls followed by multiple comparisons. The two factor ANOVA

1.05

*1.000.950.900.850.800.750.700.650.600.55

Temporarygrooved

Flatcontrol

Groovedcontrol

Before Transition

Fig. 3. Cell angle dispersion increases following topographic transition. Box plot of average Rconfidence intervals for all grooved samples and red (bottommost) shading indicates the tdotted line indicates greatest possible R, alignment to a single angle, and red dashed line indwith median center line; diamonds indicate mean; capped whiskers indicate 95% confidencvalues. Star indicates significance p ¼ 0.001 (n ¼ 6, permutation test). A two-factor generalithis thermal protocol (9.5 h at 30 �C followed by 19 h at 37 �C) subject to either initial to(p ¼ 0.001, two-tailed), indicating that for the given thermal protocol, the initial topograph

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior onj.biomaterials.2010.12.006

test was performed using the DISTLM software and permutation tests wereperformed using a Minitab macro for multiple comparisons (Supplementary DataMethod 7).

3. Results

3.1. Substrate material properties

The SMP substrate employed in this study featured a stepwisedecrease in tensile storage modulus at about 51 �C, with a glassymodulus of about 2200 MPa below this temperature and a rubberyplateau of about 25 MPa above this temperature (SupplementaryData Fig. S5). Embossing to fix a temporary topography producedrounded peaks with a peak to trough amplitude of 25.6 � 0.8 mm.

3.2. Active cell culture experiments

Cells were seeded on equilibrated temporarily groovedsamples and allowed to adhere and spread for 9.5 h at 30 �Cbefore imaging. The topographic change was activated andallowed to recover to completion for a second 9.5 h time periodand cell morphology was allowed to stabilize for a final 9.5 h timeperiod before final imaging (Fig. 1). Shape memory activationinduced a topographic transition (Supplementary Data Movie 1)that was found to cause cells to change from preferential align-ment along the grooves to a more random, unaligned orientation(Fig. 2a). Prior to shape memory activation, cells were elongatedand aligned parallel to the direction of the grooves with anaverage R of 0.85 � 0.04 (Fig. 3, Supplementary Data Fig. S6).Following shape memory activation, however, cells becamerandomly oriented with an R of 0.67 � 0.05 (p ¼ 0.001, n ¼ 6).Control groups in which the substrate did not change topographyconfirmed that the observed change in cell alignment was a resultof the change in topography rather than the change in tempera-ture (Fig. 2b,c, Supplementary Data Fig. S7). Angular histogramsshow clearly the large dispersion of cell angles on flat andrecovered-to-flat samples and the small dispersion of cell angleson grooved and temporarily grooved samples (SupplementaryData Fig. S8). In contrast to the observed changes in cell align-ment, the projected cell area, cell perimeter, and cell shape indexwere found to be similar for all groups (data not shown) and weobserved qualitatively that cell shapes were similar among allgrooved and flat groups regardless of alignment angle, except for

Recoverdto flat

Flatcontrol

Groovedcontrol

After Transition

from angular histogram data. Orange (uppermost) shading indicates the total range ofotal range of confidence intervals for flat and recovered to flat samples. Orange dash-icates a uniform distribution of angles from 0� to 180� . Boxes display interquartile rangee interval of the mean; and uncapped whiskers indicate minimum and maximum datazed linear model, permutation-based ANOVA (Supplementary Data Method 7) test thatpography (flat or grooved) had identical distributions of R was statistically significanty had a significant effect on R.

shapememory polymer substrates, Biomaterials (2010), doi:10.1016/

Fig. 4. Cell actin cytoskeleton rearranges following topographic transition. a, Confocal images of cells stained with phalloidin on a temporary grooved topography show micro-filaments aligned with groove direction (white arrow) before transition and temperature increase. After transition, microfilaments have rearranged, are randomly oriented, andexhibit apparent stress fiber formation. b, Cells on flat control substrates show randomly oriented microfilaments before and after temperature increase. c, Cells on grooved controlsubstrates show microfilaments aligned with groove direction before and after temperature increase. Scale bar is 100 mm. Traces are as in Fig. 1.

K.A. Davis et al. / Biomaterials xxx (2010) 1e9 5

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior on shapememory polymer substrates, Biomaterials (2010), doi:10.1016/j.biomaterials.2010.12.006

Fig. 5. Cell viability is high on active cell culture substrates. All groups show many viable (green) cells and very few dead (red) cells. The high background is from substrate autofluorescence in both channels, allowing the grooves on grooved control and temporarily grooved substrates to be visualized. Scale bar is 200 mm.

K.A. Davis et al. / Biomaterials xxx (2010) 1e96

some cells that are very highly elongated along the direction ofthe grooves. The lack of apparent difference may also be due tothe effect of the 2D projected image of cells on the 3D micron-scale topography.

Supplementary videos related to this article can be found at doi:10.1016/j.biomaterials.2010.12.006.

3.3. Actin cytoskeleton remodeling

Analysis of microfilament organization revealed profoundremodeling after shape memory activation. Microfilaments thatinitially aligned along the direction of the grooves reorganized to

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior onj.biomaterials.2010.12.006

a randomly distributed organization following transition to the flatsurface (Fig. 4a). The observed microfilament organization beforeand after transition was comparable to that observed in cells ongrooved and flat static controls, respectively (Fig. 4b,c). Cells onstatic controls exhibited no comparable reorganization followingthe temperature change (Supplementary Data Fig. S9).

3.4. Cell viability

For all groups, cell viability remained greater than 95% and nocell detachment was observed (Fig. 5; Supplementary Data FigsS10,S11), including those cases where SMP substrates with

shapememory polymer substrates, Biomaterials (2010), doi:10.1016/

K.A. Davis et al. / Biomaterials xxx (2010) 1e9 7

adherent cells went through a simultaneous topography andtemperature change.

3.5. SMP recovery kinetics

Profilometry analysis of unseeded SMP samples collected inparallel with cell experiments revealed the topographic recoverykinetics experienced by the cells. Activation of the SMP substrate byincreasing the temperature to 37 �C resulted in rapid onset oftopographic change. Equilibrated temporarily grooved samples(Fig. 1, Equilibrated Shape; Supplementary Data Fig. S12), of initialamplitude 12.6 � 1.5 mm showed an amplitude decrease of 3.4 mmto an amplitude of 9.2� 1.0 mm after 9.5 h at 30 �C. Upon heating to37 �C by incubator transfer, the amplitude decreased still further to0.2 � 0.1 mm within 9.5 h. No detectable decrease over the next9.5 h was observed (Fig. 6). Thus, adherent cells were expected toexperience a large and rapid substrate topography change (flat-tening) when heated from 30 �C to 37 �C. The final topography had0.2 mm deep and 40 mm wide grooves spaced 80 mm apart(Supplementary Data Fig. S4).

4. Discussion

Here we have demonstrated a temperature-responsive cellculture system, active cell culture substrates, which can be used tocontrol cell behavior via surface shape memory. The work pre-sented here demonstrates the first application of surface shapememory to cell culture. Shape memory activation caused changesin cell morphology and remodeling of the actin cytoskeleton whilemaintaining excellent cell viability. Previous research has firmlyestablished substrate topography as a useful tool for investigatingmultiple aspects of cell biology, including cell motility, cellecellinteraction, the effect of cell shape on differentiation and cellviability, and cell traction. The approach described in this paperexpands the potential of this tool by allowing the unprecedentedability to apply programmed changes in topography during culture.

This first successful demonstration of shape memory activationat 37 �C (body temperature) with attached and viable cells wasachieved through manipulation of an end-linked thiol-ene system,and there will now be several immediate opportunities forimprovement and diversification of active cell culture substratesthrough materials design. First, researchers have demonstratedshape memory activation triggered by phase transitions in semi-crystalline polymers [37] and liquid crystalline elastomers [38],with light activated and degraded crosslinks [39], and with Joule-

0

0.2

0.4

0.6

0 5 10

Rel

ativ

e am

plitu

de

Tim

Trigger shapememory activatio

Plate Cells

Fig. 6. SMP activation results in a rapid topographic change. Equilibrated (recovered 52% fromthat were incubated for 9.5 h at 30 �C (blackC) showed an amplitude decrease of 3.4 mm todecrease of 9.0 mm to 0.2 � 0.1 mm within 9.5 h (red :), and no detectable decrease overRelative amplitude (defined as the final amplitude/initial embossed amplitude) over time is ficells experience a large rapid decrease in amplitude (red line) when moved to 37 �C. The(Supplementary Data Fig. S4). Inset traces are profilometry profiles of representative samplethe length of the traces is 1200 mm.

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior onj.biomaterials.2010.12.006

heating from electrical current [40] and near infrared heating fordoped Tg-based systems [41]. Use of such triggers may allow forisothermal recovery or may reduce the cell culture temperaturechange needed to trigger meaningful recovery. Second, the shaperecovery is slower the lower themodulus of the activated SMP state[42], so recovery could be prolonged by designing SMPs of lowermodulus. Indeed, comparing cell behavior for varying shapememory recovery rates may provide distinct insights into cellmechanobiology and the theoretical understanding of the mecha-nisms underlying the soft-matter physics of cells. In the presentwork, for example, we observed that the actin cytoskeletonremodels profoundly after topography change, which suggests Rhoactivation [43]. Third, increasing fixed strain is expected to increaseshape recovery rate [42], so the temporal recovery profiles willdiffer for different fixed topographies. Lastly, the temperature atwhich an SMP is deformed or fixed will influence the temperatureat which recovery begins [44], suggesting yet another level of activecell culture control.

The ability of SMPs to undergo programmed changes in shape isentirely distinct in several important ways from other stimuli-responsive biomaterials based on photo degradation, hydrogelswelling, or LCSTs. For example, SMPs are unique in that they areable to store energy that can later be triggered to do work, they donot undergo changes in surface or bulk chemistry during thetransition, and they are able to change modulus and molecularorientation [45] in addition to changing shape. Furthermore, SMPscan also be fabricated into a variety of morphologies by a number ofmethods such as electrospinning, casting, fiber drawing, saltleaching, direct writing, UV patterning, and photolithography toallow for a wide range of bulk or topographic transitions such aspillars or microwells; and shape fixation of SMPs can be accom-plished by custom mechanical devices, manual embossing, micro-and nanoindentation [30], or mechanical testing apparatus. Theseunique properties of SMPs allow for dynamic control of modulusand shape which can be used to accomplish and extend the capa-bilities of many existing stimulieresponsive systems. For Tg-basedSMPs, the modulus decreases by as many as 3 orders of magnitude,which is reversed by decreasing the temperature below Tg. Topo-graphical transitions may also allow for release of confluent celllayers, which could also be patterned spatially simply by fabricationof an appropriate embosser. The SMPs can be stretched orcompressed to apply uni-or biaxial strain to cells without the needfor expensive or custom mechanical devices. Finally, with appro-priate design, a single substrate could simultaneously exhibit manyof these capabilities.

15 20 25 30e (h)

*

n

initial embossed amplitude) temporarily grooved samples of amplitude 12.6 � 1.5 mm9.2 � 1.0 mm. Once heated by incubator transfer to 37 �C, samples showed an amplitudethe final 9.5 h was observed. Error bars represent one standard deviation (n ¼ 4e6).t well by a multiple exponential decay (Supplementary Data Method 8). Thus, adherentfinal topography (*) has 0.2 mm deep and 40 mm wide grooves spaced 80 mm aparts. Scale for all traces is the same. The maximum represented amplitude is 14.9 mm and

shapememory polymer substrates, Biomaterials (2010), doi:10.1016/

K.A. Davis et al. / Biomaterials xxx (2010) 1e98

An intriguing extension of the present work would be applica-tion of the shape memory principles and approaches presentedhere to three-dimensional cell culture, particularly tissue engi-neering. The stasis of most current tissue engineering scaffoldsprevents investigation of dynamic biological processes and hinderstherapeutic applications. The potential for SMPs to address thesechallenges was recognized as early as 2006 [26], but no approach todate has successfully employed an SMP as a tissue engineeringscaffold. Cui et al. [46] achieved recovery of a porous SMP at 37 �C,but no cell culture was performed. To evaluate the suitability ofa poly(e-caprolactone)dimethacrylate network as a tissue engi-neering scaffold, Neuss et al. [27] cultured cells on a contractingfilm of the SMP, but cell viability was poor due to the need to heat toa hyperthermic 54 �C to trigger shape change. In the present study,cells remained healthy for extended periods of time at both a pre-activation temperature and the activation temperature and whileundergoing a change in topography, supporting the feasibility ofthe development of SMP scaffolds.

5. Conclusions

We have developed active cell culture substrates that provideprogrammed topographical changes and showed that thesubstrates can be used to control cell alignment. Alignment of thecell body and microfilaments along the grooves of the temporarytopography was significantly reduced upon transition to a flatpermanent topography. Furthermore, we presented an SMP systemand programming cycle that can provide a topography changebetween two topographies that have distinct affects on cellmorphology under cell culture conditions while maintaining highcell viability and attachment, which should enable the study of cellresponses to dynamic mechanical stimuli.

Acknowledgements

This material is based upon work supported by a grant awardedto JHH and PTM: NSF Grant No. DMR-0907578. We thank XiaofanLuo for advice inmaterial preparation.We thank Cassandra Gormanfor helpwith data analysis and presentation.We thankDr. Hyune-JuKim for discussions regarding statistical analysis. We also thank theSyracuse University Physics Department for use of the profilometerand Syracuse University for the Syracuse Biomaterials Institutefacilities.

Appendix

Figures with essential color discrimination. Figs. 1e6 in thisarticle have parts that are difficult to interpret in black and white.The full color images can be found in the online version, at doi:10.1016/j.biomaterials.2010.12.006.

Appendix. Supplementary data

Supplementary data associated with this article can be found inthe online version at doi:10.1016/j.biomaterials.2010.12.006.

References

[1] Pelham RJ, Wang YL. Cell locomotion and focal adhesions are regulated bysubstrate flexibility. Proc Natl Acad Sci U S A 1997;94(25):13661e5.

[2] Ingber DE. Mechanical control of tissue morphogenesis during embryologicaldevelopment. Int J Dev Biol 2006;50(2e3):255e66.

[3] Patwari P, Lee RT. Mechanical control of tissue morphogenesis. Circul Res2008;103(3):234e43.

[4] Jaalouk DE, Lammerding J. Mechanotransduction gone awry. Nat Rev Mol CellBiol 2009;10(1):63e73.

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior onj.biomaterials.2010.12.006

[5] Tan JL, Tien J, Pirone DM, Gray DS, Bhadriraju K, Chen CS. Cells lying on a bedof microneedles: an approach to isolate mechanical force. Proc Natl Acad Sci US A 2003;100(4):1484e9.

[6] Teixeira AI, Nealey PF, Murphy CJ. Responses of human keratocytes to micro-and nanostructured substrates. J Biomed Mater Res A 2004;71A(3):369e76.

[7] Zhao Y, Zhang X. Cellular mechanics study in cardiac myocytes using PDMSpillars array. Sens Actuators A Phys 2006;125(2):398e404.

[8] Addae-Mensah KA, Kassebaum NJ, Bowers MJ, Reiserer RS, Rosenthal SJ,Moore PE, et al. A flexible, quantum dot-labeled cantilever post array forstudying cellular microforces. Sens Actuators A Phys 2007;136(1):385e97.

[9] Lam MT, Clem WC, Takayama S. Reversible on-demand cell alignment usingreconfigurable microtopography. Biomaterials 2008;29(11):1705e12.

[10] du Roure O, Saez A, Buguin A, Austin RH, Chavrier P, Siberzan P, et al.Force mapping in epithelial cell migration. Proc Natl Acad Sci U S A 2005;102(7):2390e5.

[11] Stevens MM, George JH. Exploring and engineering the cell surface interface.Science 2005;310(5751):1135e8.

[12] Yang M, Sniadecki N, Chen C. Geometric considerations of micro- to nanoscaleelastomeric post arrays to study cellular traction forces. Adv Mater 2007;19(20):3119e23.

[13] Saez A, Ghibaudo M, Buguin A, Silberzan P, Ladoux B. Rigidity-driven growthand migration of epithelial cells on microstructured anisotropic substrates.Proc Natl Acad Sci U S A 2007;104(20):8281e6.

[14] Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem celllineage specification. Cell 2006;126(4):677e89.

[15] Anselme K. Osteoblast adhesion on biomaterials. Biomaterials 2000;21(7):667e81.

[16] Takezawa T, Mori Y, Yoshizato K. Cell culture on a thermo-responsive polymersurface. Nat Biotech 1990;8(9):854e6.

[17] Yang J, Yamato M, Sekine H, Sekiya S, Tsuda Y, Ohashi K, et al. Tissue engi-neering using laminar cellular assemblies. Adv Mater 2009;21(32e33):3404e9.

[18] Yamaki K, Harada I, Goto M, Cho C-S, Akaike T. Regulation of cellularmorphology using temperature-responsive hydrogel for integrin-mediatedmechanical force stimulation. Biomaterials 2009;30(7):1421e7.

[19] Nuhn H, Klok H-A. Secondary structure formation and LCST behavior of shortelastin-like peptides. Biomacromolecules 2008;9(10):2755e63.

[20] Hyun J, Lee W-K, Nath N, Chilkoti A, Zauscher S. Capture and release ofproteins on the nanoscale by stimuli-responsive elastin-like polypeptide"switches". J Am Chem Soc 2004;126(23):7330e5.

[21] Martin L, Alonso M, Girotti A, Arias FJ, Rodriguez-Cabello JC. Synthesis andcharacterization of macroporous thermosensitive hydrogels from recombi-nant elastin-like polymers. Biomacromolecules 2009;10(11):3015e22.

[22] Lim DW, Nettles DL, Setton LA, Chilkoti A. In situ cross-linking of elastin-likepolypeptide block copolymers for tissue repair. Biomacromolecules 2007;9(1):222e30.

[23] Frey MT, Wang Y-l. A photo-modulatable material for probing cellularresponses to substrate rigidity. Soft Matter 2009;5(9):1918e24.

[24] Tibbitt MW, Kloxin AM, Dyamenahalli KU, Anseth KS. Controlled two-photonphotodegradation of PEG hydrogels to study and manipulate subcellularinteractions on soft materials. Soft Matter 2010;6(20):5100e8.

[25] Small IV W, Singhal P, Wilson TS, Maitland DJ. Biomedical applications ofthermally activated shape memory polymers. J Mater Chem 2010;20(17):3356e66.

[26] Rickert D, Lendlein A, Peters I, Moses M, Franke R-P. Biocompatibility testingof novel multifunctional polymeric biomaterials for tissue engineeringapplications in head and neck surgery: an overview. Eur Arch Otorhinolar-yngol 2006;263(3):215e22.

[27] Neuss S, Blomenkamp I, Stainforth R, Boltersdorf D, Jansen M, Butz N, et al. Theuse of a shape-memory poly([epsilon]-caprolactone)dimethacrylate networkas a tissue engineering scaffold. Biomaterials 2009;30(9):1697e705.

[28] Lendlein A, Behl M, Hiebl B, Wischke C. Shape-memory polymers as a tech-nology platform for biomedical applications. Expert Rev Med Devic 2010;7(3):357e79.

[29] Shaw GA, Trethewey JS, Johnson AD, Drugan WJ, Crone WC. Thermo-mechanical high-density data storage in a metallic material via the shape-memory effect. Adv Mater 2005;17(9):1123e7.

[30] Nelson BA, King WP, Gall K. Shape recovery of nanoscale imprints in a ther-moset "shape memory" polymer. Appl Phys Lett 2005;86(10):1e3. 103108.

[31] Park J, Kim YS, Hammond PT. Chemically nanopatterned surfaces using poly-electrolytes and ultraviolet-cured hard molds. Nano Lett 2005;5(7):1347e50.

[32] Zurgil N, Afrimzon E, Deutsch A, Namer Y, Shafran Y, Sobolev M, et al. Polymerlive-cell array for real-time kinetic imaging of immune cells. Biomaterials2010;31(18):5022e9.

[33] DiOrio AM, Luo X, Lee KM, Mather PT. A functionally graded shape memorypolymer. Soft Matter 2011;7(1):68e74.

[34] Tang L, Lee N. Micro-perforated elastomeric poly(dimethylsiloxane) maskfabricated using high-aspect-ratio micro-pillar arrays for spatially definedsurface modification: an unconventional method for establishing a microarrayplatform. Anal Bioanal Chem 2009;394(4):1227e32.

[35] Liu C, Qin H, Mather PT. Review of progress in shape-memory polymers.J Mater Chem 2007;17(16):1543e58.

[36] Fujita S, Ono D, Ohshima M, Iwata H. Supercritical CO2-assisted embossing forstudying cell behaviour on microtextured surfaces. Biomaterials 2008;29(34):4494e500.

shapememory polymer substrates, Biomaterials (2010), doi:10.1016/

K.A. Davis et al. / Biomaterials xxx (2010) 1e9 9

[37] Lee KM, Knight PT, Chung T, Mather PT. Polycaprolactone-POSS chemical/physical double networks. Macromolecules 2008;41(13):4730e8.

[38] Rousseau IA, Mather PT. Shape memory effect exhibited by smectic-c liquidcrystalline elastomers. J Am Chem Soc 2003;125(50):15300e1.

[39] Lendlein A, Jiang H, Junger O, Langer R. Light-induced shape-memory poly-mers. Nature 2005;434(7035):879e82.

[40] Sahoo NG, Jung YC, Goo NS, Cho JW. Conducting shape memory polyurethane-polypyrrole composites for an electroactive actuator. Macromol Mater Eng2005;290(11):1049e55.

[41] Small W, Metzger MF, Wilson TS, Maitland DJ. Laser-activated shapememory polymer microactuator for thrombus removal following ischemicstroke: preliminary in vitro analysis. Ieee J Sel Top Quant 2005;11(4):892e901.

Please cite this article in press as: Davis KA, et al., Dynamic cell behavior onj.biomaterials.2010.12.006

[42] Morshedian J, Khonakdar HA, Rasouli S. Modeling of shape memory inductionand recovery in heat-shrinkable polymers. Macromol Theory Simul 2005;14(7):428e34.

[43] Hall A. Rho GTPases and the actin cytoskeleton. Science 1998;279(5350):509e14.

[44] Gall K, Yakacki CM, Liu Y, Shandas R, Willett N, Anseth KS. Thermomechanicsof the shape memory effect in polymers for biomedical applications. J BiomedMater Res A 2005;73A(3):339e48.

[45] Burke KA, Mather PT. Soft shape memory in main-chain liquid crystallineelastomers. J Mater Chem 2010;20(17):3449e57.

[46] Cui J, Kratz K, Heuchel M, Hiebl B, Lendlein A. Mechanically active scaffoldsfrom radio-opaque shape-memory polymer-based composites. Polym AdvTechnol. in press; doi:10.1002/pat.1733.

shapememory polymer substrates, Biomaterials (2010), doi:10.1016/