Immobilized laminin concentration gradients on electrospun fiber scaffolds for controlled neurite outgrowth Nicole E. Zander and Thomas P. Beebe Jr. Citation: Biointerphases 9, 011003 (2014); doi: 10.1116/1.4857295 View online: http://dx.doi.org/10.1116/1.4857295 View Table of Contents: http://scitation.aip.org/content/avs/journal/bip/9/1?ver=pdfcov Published by the AVS: Science & Technology of Materials, Interfaces, and Processing Articles you may be interested in Optical scattering in electrospun poly(ε-caprolactone) tissue scaffolds J. Laser Appl. 26, 032004 (2014); 10.2351/1.4870675 A cell-laden nanofiber/hydrogel composite structure with tough-soft mechanical property Appl. Phys. Lett. 102, 211914 (2013); 10.1063/1.4808082 A controlled biochemical release device with embedded nanofluidic channels Appl. Phys. Lett. 100, 153510 (2012); 10.1063/1.4704143 Autoclaving as a chemical-free process to stabilize recombinant silk-elastinlike protein polymer nanofibers Appl. Phys. Lett. 98, 263702 (2011); 10.1063/1.3604786 Embedding of magnetic nanoparticles in polycaprolactone nanofiber scaffolds to facilitate bone healing and regeneration J. Appl. Phys. 107, 09B307 (2010); 10.1063/1.3357340

Immobilized laminin concentration gradients on electrospun fiber scaffolds for controlled neurite outgrowth Nicole E. Zandera) US Army Research Laboratory, Weapons and Materials Research Directorate, Aberdeen Proving Ground, Maryland 21005

Thomas P. Beebe, Jr. Department of Chemistry and Biochemistry, University of Delaware, Newark, Delaware 19716

(Received 31 October 2013; accepted 3 December 2013; published 8 January 2014) Neuronal process growth is guided by extrinsic environmental cues such as extracellular matrix (ECM) proteins. Recent reports have described that the growth cone extension is superior across gradients of the ECM protein laminin compared to growth across uniformly distributed laminin. In this work, the authors have prepared gradients of laminin on aligned electrospun nanofibers for use as substrates for neuronal growth. The substrates therefore presented both topographical and chemical guidance cues. Step gradients were prepared by the controlled robotic immersion of plasma-treated polycaprolactone fibers reacted with N-hydroxysuccinimide into the protein solution. The gradients were analyzed using x-ray photoelectron spectroscopy and confocal laser scanning microscopy. Gradients with a dynamic range of protein concentrations were successfully generated and neurite outgrowth was evaluated using neuronlike pheochromocytoma cell line 12 (PC12) cells. After 10 days of culture, PC12 neurite lengths varied from 32.7 6 14.2 lm to 76.3 6 9.1 lm across the protein concentration gradient. Neurite lengths at the highest concentration end of the gradient were significantly longer than neurite lengths observed for cells cultured on samples with uniform protein coverage. Gradients were prepared both in the fiber direction and transverse to the fiber direction. Neurites preferentially aligned with the fiber direction in both cases indicating that fiber alignment has a more dominant role in controlling neurite orientation, compared to the chemical C 2014 American Vacuum Society. [http://dx.doi.org/10.1116/1.4857295] gradient. V

I. BACKGROUND Gradients are important for driving biological processes and exist in many forms in the natural world. Plants, such as bamboo, and bone and teeth have mechanical property gradients, while soft tissues and interfaces exhibit chemical or physicochemical gradients.1 Gradients exist within the extracellular matrix (ECM) architecture where they control cell migration, elongation, differentiation, and signaling.2–10 Chemical gradients are known to direct cells during embryogenesis; mechanical gradients in bone control the differentiation of mesenchymal cells; and electrochemical gradients are responsible for diffusion across cell membranes. Chemical gradients are also important for wound healing, immune response, controlling cell adhesion, and providing axonal guidance.11–17 Cell–ECM interactions are controlled by ligand-receptor binding mediated by transmembrane integrin proteins. Beltran et al. controlled the adhesion of myoblasts with fibronectin concentration gradients on poly (ethylacrylate) surfaces,18 while Biran et al. found a similar response for dorsal root ganglion (DRG) neurons on such gradients.19 It is well known that nerve cell axonal growth cones rely on navigation cues to find their targets. One of the fundamental issues preventing control of nerve regeneration is a lack of understanding how growing axons determine their path to establish specific connections in the nervous system.20 Ramon y Cajal suggested a)

Electronic mail: [email protected]

011003-1 Biointerphases 9(1), March 2014

that axons navigate by chemotaxis, while Sperry postulated that gradients of specific molecules could guide axons.20,21 Laminin is an ECM protein which constitutes a major part of the basement membrane. Laminin contains the peptide sequences tyrosine-isoleucine-glycine-serine-arginine and arginine-glycine-aspartic acid, known to influence cell adhesion and migration, as well as the sequence isoleucinelysine-valine-alanine-valine, which promotes neurite outgrowth.22–28 For these reasons, much work in the nerve regeneration field focuses on the use of either the whole protein laminin or laminin peptides as substrates to explore nerve growth and differentiation. These substrates may be used alone or with other growth factors [e.g., nerve growth factor (NGF)] or growth inhibitors (e.g., chondrotin sulfate proteoglycan). Deister found that DRG neurite extension was correlated in a dose-dependent manner to laminin concentration.29 Dodla et al. found that DRG neurons extended axons faster on laminin-1 gradients compared to samples with isotropic laminin concentration.30 Adams et al. found that a laminin peptide gradient could steer growth cones of DRG neurons.12 Cao et al. and Kapur et al. saw a similar effect with NGF gradients and neuronlike pheochromocytoma cell line 12 (PC12) cells.31–33 Thus, anisotropy in scaffold design may enable faster nerve regeneration compared to scaffolds with constant protein or growth factor coverage. Bellamkonda et al. constructed agarose hydrogels with gradients of laminin-1 and NGF to successfully repair rat sciatic nerves with up to 20 mm gaps.30 Isotropic scaffolds with the

1934-8630/2014/9(1)/011003/9/$30.00

C 2014 American Vacuum Society V

011003-1

011003-2 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

same concentration could only effectively repair 10 mm gaps. Growing axons and neurites are not only guided by chemical cues, but also topographical cues. Due to the known importance of contact-guidance in the regeneration of nerve tissue, the majority of research has focused on scaffolds that can act as a biomaterial bridge across the scar region.34,35 Fibers have the advantage of controlled alignment over directionally isotropic materials such as most hydrogels. Aligned nanofibers collected on a rotating mandrel (formed by the electrospinning process) exhibit both good fiber alignment and high surfacearea-to-volume ratios, which are favorable for cell attachment and surface modification to improve cell adhesion.36–38 Gradients on planar surfaces or in hydrogels are formed by a variety of techniques including controlled diffusion, postcoating, microfluidic mixing, photo-crosslinking, and controlled dipping.3 Gradient surface functionalization of 3-D nanofiber scaffolds is a relatively new concept that is just beginning to be explored. Shi et al. presented a method to make continuous gradients on nanofibers by controlling the rate of wetting of the fibers by the modifying protein solution.39 In this work, we present methods to generate both step and continuous gradients of covalently bound laminin on plasma-treated polycaprolactone (PCL) fibers in order to control neurite outgrowth for use as nerve regeneration scaffolds. Our method, with a mechanically controlled translation stage, utilizes high-fidelity reactions to control protein immobilization. Gradients of protein concentration were explored both parallel and orthogonal to the fiber orientation, and the effect of wicking was also investigated. The scaffolds were examined with x-ray photoelectron spectroscopy (XPS) and confocal laser scanning microscopy (CLSM), and the regional effect of protein concentration on the neurite outgrowth of PC12 cells was determined. Although protein gradients were anticipated to be three-dimensional and all CLSM imaging was conducted using z-stacks throughout the mat thickness, we did not specifically probe their effect on cellular infiltration. We have shown in previous work that PC12 cells are able to infiltrate the mats used in these experiments up to ca. 500 lm in thickness.40 II. METHODS A. Materials

Polycaprolactone, 3-mm pellets with an average molecular weight of 40 kDa, was obtained from Polysciences, Inc. Anhydrous N,N-dimethylformamide (DMF), mouse laminin, Tween 20, N-(3-dimethylaminopropyl)-N0 -ethylcarbondiimide (EDC), N-hydroxy succinimide (NHS), and 2-(Nmorpholino)ethanesulfonic acid (MES) were obtained from Sigma-Aldrich. Dichloromethane and phosphate buffered saline (PBS) were obtained from Fisher Scientific. Laminin-488 was obtained from Cytoskeleton. Dulbecco Modified Eagle’s Medium (DMEM), Roswell Park Memorial Institute (RPMI) 1640, calf serum, horse serum, and NGF (7S) were obtained from Fisher Scientific. Antibiotic/antimycotic was obtained from Cellgro (cat. 30-004-C1). Rhodamine phalloidin was obtained from Invitrogen. Biointerphases, Vol. 9, No. 1, March 2014

011003-2

B. Fabrication of electrospun fibers

A 20 wt. % solution of PCL was prepared by dissolving PCL in a 50:50 (w/w) mixture of anhydrous DMF and dichloromethane and stirring at RT overnight. PCL fibers were electrospun using a custom built set-up consisting of a syringe pump (Aladdin AL-1000) and a rotating 6-in.-wide mandrel. The mandrel was connected to a motor (Dayton 115 V AC–DC) with a speed controller, allowing the mandrel to rotate between 0 and 8000 rpm. A 10-mL syringe was filled with polymer solution and fed through a 21-gauge stainless steel needle at flow rates of 1–5 mL/h with an applied potential of þ18.5 kV at the needle. The gap between the needle and the collector was fixed at 7 in., and the collector was set to an applied potential of 3 kV. Fibers were dried under vacuum at RT overnight before characterization and functionalization. C. Electrospun fiber modification

For the covalent attachment of protein to the fibers, vacuum dried PCL fibers were cut to 144 mm2 and plasma treated in air using an inductively coupled radio-frequency (RF) plasma cleaner (Harrick PDC-32-G) for 5 min at a RF power of 18 W to introduce carboxylic functionalities to the surface of the fibers. Plasma treated fibers were then immersed in a MES buffer containing of 5 mg/mL EDC and 5 mg/mL NHS for 1 h at RT.41 Fiber mats were then rinsed with MES buffer, adhered to 18 mm glass coverslips with double-stick tape, and attached to the sample holder in the custom-built robotic gradient maker. The LexanV wells were filled with ca. 4 mL of a 50 lg/mL laminin solution in PBS. A variable-speed stepper motor was used to control the immersion rate (0.25–3.8 mm/min) to produce either continuous or step gradients. To control the humidity and prevent evaporation of the protein solution, the sample stage and solution reservoir were enclosed in a Lexan chamber with a saturated aqueous environment. Figure 1(a) displays a schematic of the gradient-maker showing the sample holder, solution well, as well as the variable-speed motor and pulley system, which controls the immersion of sample stage. Figure 1(b) presents a schematic of a step gradient sample, consisting of six regions of theoretical constant protein coverage that are 2 mm in depth (immersion axis) and 12 mm in the horizontal direction. The sample stage was advanced rapidly to a depth of 2 mm, held for time t1 in the protein solution, and then advanced rapidly to 4 mm, where is was held for time t2. This process was repeated to produce a total of six constant-protein-coverage regions of dimensions 12 mm  2 mm, consisting of protein–surface reaction times (from bottom to top of mat) of 8 h, 1 h, 15 min, 5 min, 2 min, and 0 min. Fiber mats with constant protein coverage were prepared by immersing the fiber mats in a 24-well plate in the protein solution (10–50 lg/mL) for the specified time period. After the set time period, both gradient and uniform coverage samples were removed from the protein solution, and the mats were washed in a 0.05% Tween 20 solution in PBS with R

011003-3 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

011003-3

processed using Casa XPS software. All reported atomic percentages are the average of n ¼ 2 measurements on a minimum of three replicate samples. E. Conjugated laminin assay

FIG. 1. (Color online) Fabrication of protein gradients on electrospun nanofibers. (a) Robotic gradient-maker with gear-driven sample stage for controlled immersion. Fiber samples were mounted on glass coverslips and attached to the sample stage. The variable-speed motor and pulley system allowed for controlled immersion of the fiber samples in the protein well to generate either step or continuous gradients. (b) Schematic representation of a step protein gradient. The step gradients consisted of 0, 2, 5, 15, 60, and 480 min reaction times.

gentle shaking for 30 min to remove physically adsorbed proteins. Mats were then washed thoroughly with PBS and sterilized overnight by immersing in a sterile solution of 2% antibiotic/antimycotic in PBS. Samples were kept sterile for cell culture studies or rinsed thoroughly with deionized water and dried for characterization. Experiments to demonstrate the covalent attachment of laminin and the removal of physically adsorbed proteins are detailed in the Supporting Information.42 D. Characterization of scaffolds

The morphology of the fiber scaffolds was examined using a field emission scanning electron microscope (SEM) (Hitachi S-4700) in the secondary electron mode, using a mixture of upper and lower detectors. An accelerating voltage of 0.6 kV was maintained in order to prevent surface damage to the substrate. Before observation, the samples were first coated with gold using a sputter coater (Hummer XP Sputtering System, Anatech LTD). Several areas were imaged in order to examine the uniformity of the fiber diameters and alignment. Fiber diameters were measured using image analysis software (Image J v 1.34, National Institutes of Health). Surface compositional analysis was performed using a Kratos Axis Ultra 165 XPS system equipped with a hemispherical analyzer. Sampling areas of 1 mm  0.5 mm were irradiated with a 140 W monochromatic Al Ka (1486.7 eV) beam and a take-off angle of 90 . The XPS chamber pressure was maintained between 109 and 1010 Torr. Elemental high resolution scans were conducted with a 20 eV pass energy for the C 1s, O 1s, and N 1s core levels. A value of 284.6 eV for the hydrocarbon C 1s core level was used as the calibration energy for the binding energy scale. Data were Biointerphases, Vol. 9, No. 1, March 2014

The concentration of covalently attached laminin in the step gradient samples was probed using a fluorescently tagged laminin protein. Samples were prepared as previously described, with the entire sample stage covered with a box to keep samples in the dark during the reaction. After the reaction, the samples were then rinsed with Tween and PBS as previously detailed. The immunostained samples were imaged using confocal laser scanning microscopy on a Zeiss LSM5 Pascal, as has been previously described.40,43,44 Detector gain and amplifier offset were kept constant for all samples to enable semiquantitative comparison. The protein was imaged with the 488 nm laser in channel 1, while the fibers were imaged using the 543 nm laser in reflection mode in channel 2. A minimum of n ¼ 5 random areas for each of three replicate samples was imaged using the 20 objective. Z-stacks were collected throughout the width of the mat (ca. 800 lm) with a 160 lm stack size. The stacks were then compressed into one plane using the projection feature in the Zeiss software. The first angle was set to 0, and the number of angles and difference angle were set to 1. The maximum intensity of the compressed z-stack was determined using Image J and normalized to the maximum intensity of the compressed z-stack for the same area of reflected fibers. F. Culture of PC12 cells

PC12 cells derived from the pheochromocytoma of the rat adrenal medulla were used in these experiments. PC12 cells are widely used in in vitro studies and undergo neuronlike differentiation when treated with nerve growth factor. Therefore, they are a useful cell line to probe the effect surface chemistry on neurite outgrowth.41,45,46 PC12 cells were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated horse serum and 5% fetal bovine serum and 1% antibiotic/antimycotic “complete medium” at 37  C and 5% CO2. Sterile fiber mats in 24-well plates were incubated in serum-free medium 2 h prior to seeding cells. For differentiation studies, cells at a passage number between 20 and 30 were seeded at a density of 10 000 cells/well in high-glucose DMEM with 1% horse serum, 0.5% calf serum, and 1% antibiotic/antimycotic “differentiation medium.” After 24 h, 100 ng/mL NGF was added to the differentiation medium. For cell adhesion studies, cells were seeded at a density of 10 000 cells/well in high-glucose DMEM with 1% horse serum, 0.5% calf serum, and 1% antibiotic/antimycotic medium. Cells were fixed after 10 days and stained and analyzed in the same manner as described in Sec. II G. G. Neurite outgrowth study

After day-10 of seeding cells in differentiation medium supplemented with 100 ng/mL NGF, neurite outgrowth was characterized by staining actin filaments with phalloidin and

011003-4 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

011003-4

analyzing with CLSM. To prepare samples for CLSM, the fiber scaffolds were rinsed thoroughly with PBS and fixed in 4% paraformaldehyde in PBS for 20 min. Scaffolds were rinsed with PBS and cells were permeabilized with 0.2% Triton X-100 for 10 min. Nonspecific labeling was prevented by incubating samples in a blocking buffer composed of 1% bovine serum albumin in PBS for 20 min. Samples were then immersed in rhodamine phalloidin (1:200) in blocking buffer for 1 h. Samples were rinsed thoroughly with PBS and kept in the dark at 4  C until analysis. Samples were imaged on a Zeiss LSM5 Pascal equipped with Epiplan-Neofluar lenses. The cells in the scaffolds were imaged with a 543-nm laser. A minimum of n ¼ 5 random areas for each of a minimum of three replicate fiber mat samples were imaged using the 10 and 20 objectives. Neurite length was measured using the Zeiss LSM software (v 4.2.0.121). H. Statistics

All data are expressed as mean 6 standard deviation unless noted. One-way repeated measures of analysis of variance with post-hoc Tukey means comparison tests, unpaired Student’s t-tests, and prediction bands were conducted with a significance level of p < 0.05, using Origin v 8.5. One experiment was conducted for each of the XPS assays and the conjugated laminin assay. Three independent experiments were conducted for the cellular assays in the fiber direction (one in the transverse direction). A minimum of three replicate samples were used for all experiments. Data points were collected every ca. 1 mm across the gradient samples. Thus, a minimum of ca. 36 data points were collected for each sample type. All data points are displayed in Figs. 2–5.

FIG. 3. (Color online) Step gradients of covalently attached laminin on aligned electrospun polycaprolactone fibers. Fibers were prepared as described in Sec. II C. Six regions of 12 mm  2 mm were generated by immersing 2 mm of the mat into the protein well and holding for a set period of time and then rapidly advancing 2 mm to form the next step region. This process was repeated to generate six regions with t ¼ 0, 2, 5, 15, 60, and 480 min. Red line denotes centroid of data while blue lines denote 90% confidence limits. Insets are XPS N 1 s spectra at different positions on the fiber gradient as denoted by the x-axis. Scale bars represent 30 CPS. (a) Fibers were immersed in the fiber direction and (b) fibers were immersed transverse to the fiber direction.

III. RESULTS A. Morphology and chemical characterization of uniform-protein-coverage nanofibrous scaffolds

FIG. 2. (Color online) Capillary wicking effects on aligned electrospun polycaprolactone fibers. The bottom 3 mm of the fiber mats were immersed in 50 lg/mL laminin for 8 h. The portion of the fiber mat immersed is between the dashed lines. Biointerphases, Vol. 9, No. 1, March 2014

The electrospinning technique was utilized to generate aligned nanofibrous scaffolds with fiber diameters of 155 6 70 nm (n ¼ 50). An SEM image of a representative fiber mat and a histogram of fiber diameters are displayed in Fig. 6. The attachment of protein did not significantly alter the fiber diameter [138 6 90 nm (n ¼ 54)], as we have

011003-5 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

FIG. 4. (Color online) Analysis of laminin protein step gradient on aligned polycaprolactone electrospun nanofibers using CLSM. Fibers were prepared as described using laminin conjugated with a fluorescent dye. CLSM zstacks were collected, compressed into one plane of maximum intensity and analyzed using the histogram feature in Image J. The mean intensity was normalized to the fiber area determined from the reflected fibers. Red line denotes centroid of data while blue lines denote 90% confidence limits. The insets are CLSM images of the fiber mats at positions along the gradient as denoted in the x-axis. Image size is 400 lm  400 lm.

previously reported.40 SEM images before and after protein attachment can be found in the Supporting Information. Nanofiber surfaces were modified with the protein laminin, studied as a function of protein solution concentration (10–50 lg/mL) and reaction time. The resulting surface coverage was quantified by XPS measurements, and a comparison of the nitrogen-to-carbon ratio (N/C) is displayed in

FIG. 5. (Color online) PC12 neurite outgrowth on aligned polycaprolactone electrospun fibers with protein concentration step gradient. Black squares denote neurite lengths on step gradients in the fiber direction. Red circles denote neurite lengths on step gradient in the transverse fiber direction. Biointerphases, Vol. 9, No. 1, March 2014

011003-5

Fig. 7. There is a steep rise in protein coverage in the first ca. 2 h of reaction time. A slowing of protein accumulation occurs between ca. 2 and 24 h. In previous studies, it was determined one protein monolayer coverage was obtained during this timeframe.43 Proteins in the first monolayer are assumed to be covalently attached and fully covering the fiber mat surface. Additional deposition after 24 h is most likely due to multilayer protein coverage. Proteins in this region are most likely physically adsorbed to the surface rather than covalently attached. The surface-equivalent protein concentration was determined previously for a series of different protein solution concentrations using a fluorescence assay, and ranged from 0 to 1.35 nM/mg fibers for the 50 lg/mL concentration/24 h timepoint.43 Protein coverage on the fibers depends not only on reaction time but also on protein solution concentration. Higher protein solution concentrations led to increased XPS N/C ratios and higher amounts of protein on the fiber surface. Thus, both approaches (varying solution concentration or time) are valid methods for controlling protein deposition. B. Chemical characterization of gradient nanofibrous scaffolds

An assay with PC12 cells showed a substrate-dependent response in neurite outgrowth rates correlated to laminin concentration.43 Based on this work, and the desire to fabricate next-generation anisotropic scaffold materials, we developed methods to form protein gradients on the surface of the fibers. Selected protein–fiber reaction time points were chosen to generate step gradient samples and were based on the effect of constant-coverage samples at certain XPS N/C ratios on the neurite outgrowth of PC12 cells.43 The longest reaction time needed to generate such samples was 8 h. Since we were concerned about the effect of capillary wicking on the hydrophilic plasma-treated fibers, especially at long time periods, we conducted an experiment to examine wicking effects. The results are displayed in Fig. 2. In these experiments, approximately 3 mm of the sample was immersed in 50 lg/mL laminin for 8 h in a humidity controlled chamber. The data points to the right of the dashed line in the figure arise from the immersed portion of the mat. The XPS N/C is between ca. 8 and 12 for the immersed portion of the mat. These values range between ca. 3 and 12 for the adjacent nonimmersed 3 mm. Beyond the adjacent 3 mm, the XPS N/C values drop off to between 0 and ca. 4. Thus, there is evidence of wicking occurring, and it affects approximately 25% of the mat length (2–3 mm). This explains the previous results, and the protein observed in the “zero hour” time points on both the continuous and step gradient samples. Six regions with reaction times of 0, 2, 5, 15, 60, and 480 min were selected to form the step gradient samples. Thus, for the 12 mm long samples, each region of constant protein coverage was 2 mm in length. Step gradients were formed in both the fiber direction and transverse to the fiber direction. The XPS analysis of the step gradient is displayed

011003-6 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

011003-6

FIG. 6. Scanning electron micrograph of electrospun PCL fibers. (a) Aligned PCL fibers, (b) histogram of (a) (n ¼ 50). Scale bar denotes 5 lm.

in Fig. 3. The centroid of the XPS N/C data, denoted by the red line in the figure, ranged between ca. 2 and 11 for the gradients formed in the fiber direction [Fig. 3(a)], with a linear slope of 0.66 (linear least squares fit). The dynamic range of the data was slightly reduced for the gradients transverse to the fiber direction [Fig. 3(b)]. The centroid of the XPS N/C ranged between ca. 3 and 8, with a linear slope of 0.49. Six replicate samples were evaluated for the gradients formed in the fiber direction (n ¼ 73), while three replicates were evaluated for the transverse fiber direction gradients (n ¼ 39). The R2 coefficient was 0.70 for the former samples, compared to 0.44 for the latter samples, indicating a better fit of the data. Graphs of the step gradient data plotted in time overlaid with the Figure 7 data are displayed in the Supporting Information. Fairly good correlation is observed although step gradient N/C ratios are slightly lower than the N/C for uniform coverage samples for the first 4 h.

FIG. 7. (Color online) Protein–fiber surface reaction kinetics for electrospun polycaprolactone fibers reacted with laminin as determined by x-ray photoelectron spectroscopy. Fibers were plasma treated and reacted with Nhydroxysuccinimide in EDC. Fibers were then reacted for specified time periods with laminin at four different concentrations: 10, 25, 37, and 50 lg/mL, as denoted by the inset in the figure. Note 0 h time point denotes ca. 30 s reaction time. Error bars denote mean 6 standard deviation (n ¼ 6). Biointerphases, Vol. 9, No. 1, March 2014

C. Fluorescent imaging of gradient nanofibrous scaffolds

In order to provide another means to characterize the formation of the gradients, the step gradients were prepared using laminin conjugated to a fluorophore to provide a tag for fluorescent imaging. In order to normalize the fluorescent intensity of the laminin on the fiber mats, a channel consisting of the reflected fiber image was collected in order to account for fiber area in each image. The results are shown in Fig. 4. The normalized fluorescent intensity of the CLSM z-stacks increased from about 0.5 to 3.5 along the gradient of increasing protein concentration. Three replicate samples were evaluated, and all data points are displayed in the figure (n ¼ 36). The R2 coefficient was 0.73, similar to the XPS data. D. Interaction of PC12 cells with nanofibrous scaffolds

PC12 cells were cultured on native PCL fibers, PCL fibers with constant protein coverage and PCL fibers with step concentration gradients. Cellular adhesion and neurite outgrowth were evaluated after 10 days in culture. The cell coverage was determined by an image area analysis of CLSM images at different positions on the gradient. The cell coverage ranged from ca. 5% at the 0 mm position to ca. 50% at the 12 mm position of the gradient. The cell coverage increased in a linear fashion across the gradient, with more cells adhered on regions with higher protein concentration and hence more cell adhesive motifs (see Supporting Information). Cell coverage at the 0 mm gradient position (region not directly exposed to protein solution) was similar to coverage observed on native PCL fibers.44 The morphology of the PC12 cells varied greatly at different positions along the gradient. As illustrated in Fig. 8, at the zero position with essentially no protein coverage, the cells are mostly clumped together and not well adhered or differentiated. As the protein concentration increased, the spreading of the cells improved, and length of the neurites and number of cells differentiated increased [Figs. 8(b) and 8(c)].

011003-7 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

011003-7

FIG. 8. Cell morphology of PC12 on aligned polycaprolactone (PCL) electrospun fiber scaffolds. (a) Low protein concentration region of step gradient on PCL fibers, (b) middle protein concentration region of step gradient on PCL fibers, (c) high protein concentration region of step gradient on PCL fibers, (d) unmodified PCL fibers, (e) PCL fibers with constant protein coverage (50 lg/mL, 8 h). Scale bars denote 100 lm. Arrows show selected locations of neurites.

The neurite outgrowth of PC12 cells was accessed on both uniform protein coverage scaffolds, as well as scaffolds with protein step gradients. The constant protein coverage samples were prepared with the same protein concentration and at the longest reaction time used in the formation of the step gradients (50 lg/mL, 8 h). Thus, it was expected that similar neurite lengths as at the higher protein concentration end of the gradient samples would be observed. In addition, in order to understand if the 8 h reaction time at RT (used for the step gradients) had an adverse effect on the laminin protein bioactivity, constant coverage samples were prepared both using fresh laminin solutions and laminin solutions that had been previously exposed to RT for 8 h. The neurite lengths for the cells grown on the scaffolds prepared with fresh protein solution (46.96 6 22.56 lm, n ¼ 88) were not found to be statistically different from those grown on scaffolds with the protein exposed to RT (41.43 6 21.03 lm, n ¼ 151). Thus, laminin conformation and reactivity were not significantly altered by the ambient temperature for the specified time period. Figure 5 displays the average neurite lengths of PC12 cells at each position on the step gradient samples. Results from both the fiber direction (n ¼ 9 replicates) and transverse fiber direction samples (n ¼ 3 replicates) were included. The neurite lengths in the fiber direction ranged from 32.7 6 14.2 lm at the lowest concentration step region to 76.3 6 9.1 lm at the highest concentration step region (black squares). Neurite lengths on gradients formed transverse to the fiber direction gave similar results with a range from 43.6 6 15.3 lm to 75.1 6 9.3 lm (red circles). Positive controls consisting of PCL fibers with uniform coverage of covalently attached laminin (50 lg/mL, 8 h) were included in the experiments. (See Supporting Information for neurite Biointerphases, Vol. 9, No. 1, March 2014

lengths separated by individual experiment.) In both cases, neurite lengths were longer at the highest protein concentration end of the gradient when compared to samples with uniform equivalent protein coverage (43.48 6 21.7 lm, n ¼ 4 replicates). One major difference observed between the two sample types was the neurite orientation direction, displayed in Fig. 9. The neurites of cells grown on fibers with protein gradients in the fiber direction, aligned with both the protein gradient and the fiber direction. Yet, neurites of cells grown on fibers aligned transverse to the protein gradient direction, preferentially oriented with the fiber orientation. IV. DISCUSSION Axonal growth in developing nervous systems is known to be guided by both diffusible and immobilized chemical

FIG. 9. Confocal laser scanning microscopy images of PC12 cells on aligned polycaprolactone electrospun fibers with protein concentration step gradients. (a) Protein gradient in the fiber direction, and (b) protein gradient transverse to the fiber direction. Arrows denote fiber direction. Scale bars denote 100 lm.

011003-8 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

gradients. Thus, it is of great interest to prepare scaffolds with such gradients to aid in vitro and in vivo nerve regeneration. Besides the utility of each gradient sample having the potential to be a multivariate sample, recent research has demonstrated improved axonal guidance control, neurite elongation, and cellular adhesion on such samples.19,30 In this work, we have developed a method to form step gradients (series of regions with constant local protein coverage and steplike fashion increase from one region to the next). Due to the plasma treatment step in the protein attachment mechanism, which introduces carboxylic groups that can form an amide bond with the protein’s amine groups, the fiber mats are fairly hydrophilic during the protein immersion step. Although we were initially concerned that capillary action or wicking up the aligned fibers would make it challenging to generate dynamic gradients, this was proven unfounded. We first focused on generating continuous gradients to minimize this effect. We surmised that since the protein coverage was constantly changing, small additions of protein due to wicking would not be noticeable. Another advantage of continuous gradients is the size-scale of the gradient. Since the protein coverage is continually changing, the gradient is, in theory, on the nanometer or micrometer size-scale, mimicking natural gradients in tissues. Although we were able to form continuous gradients, the range in protein coverage was quite small due to limitations in our gradient production hardware (40 min reaction time limit). Based on previous neurite outgrowth assays with PC12 cells, a larger dynamic range was necessary to observe significant changes in neurite lengths.43 Since we did not have the requisite equipment to generate longer reaction time continuous gradients, we turned our attention to generating step gradients. Bellamkonda et al. found that step gradients (5 mm steps) were actually more effective than continuous gradients in the peripheral nerve regeneration in rats.30 We examined gradients both in the fiber direction and transverse to see if there would be a difference in terms of protein accumulation. We were also interested in probing whether the neurites would orient in the direction of the fiber axis or protein gradient, when given a choice. Indeed, we found that a larger dynamic range and smaller spread of the XPS N/C data was achieved for the gradients formed in the fiber direction compared to the transverse direction. We believe the wicking was more pronounced along the fiber axis and aided in the protein–fiber surface reaction, but at the same time did not overpower the gradient formation. We expected to observe improved neurite outgrowth on the gradient samples prepared in the fiber direction due to the higher XPS N/C ratios and hence more protein on the surface. But, in fact, the difference between neurite lengths on the two samples types was not significant. Thus, the difference in the amount of protein was not large enough to produce an observable effect on neurite outgrowth. In terms of neurite orientation, the fiber direction was found to be the dominant controlling factor, and thus, chemical gradients played less of a role in providing contact guidance cues for Biointerphases, Vol. 9, No. 1, March 2014

011003-8

extending neurites in our study. Most studies that have examined the effect of chemical and topographical cues have looked at their synergistic effect.47 The separation of the two effects has proven to be confounding. Richardson found a complex relationship between the response of DRG neurites to micropatterned laminin stripes and Schwann cell replica topography. Neurites oriented preferentially with the chemical cues when the stripe width was 50 lm, and with the topographical cues when the width was 500 lm.48 Gomez et al. found prepared microfabricated channels of 1–2 lm in width next to immobilized nerve growth factor and laminin. Topographical cues were preferred 70% of the time in these assays.49,50 Forciniti et al. developed a model to provide insight into the aforementioned choice assay. Their findings agree closely with Richardson. Neurites oriented with the chemical cues when the topographical feature size was greater than ca. 40 lm.51 In our research, the fiber diameters were 155 6 70 nm and thus much smaller than the feature size threshold identified in Forciniti’s model. Hence, the preferential alignment with the fiber axis over the laminin concentration gradient agrees with the previous research. Based on previous research, we anticipated that the neurite outgrowth would be enhanced on the anisotropic protein concentration gradient samples compared to uniform coverage samples (with protein concentration equivalent to highest protein concentration in the gradient).29 Indeed, our results confirm this trend in which scaffolds with a protein gradient were superior to uniform protein coverage scaffolds. From our cell adhesion assay, we observed that cell coverage and number of cells adhered increased with increasing protein concentration. Thus, it is possible that increased cell–cell communication and production of growth factors could have contributed to this result. V. CONCLUSIONS Electrospun fiber scaffolds with protein concentration gradients of laminin were preparing using a custom-built system consisting of a linear translation stage and stepper motor for controlled immersion of the fiber mats. Step gradients consisting of six constant-protein-coverage regions with reaction times up to 8 h showed a good range of protein concentration. The gradients formed in the fiber direction yielded better gradients than those formed transverse to the fiber direction based on XPS analysis, most likely due to the enhanced wicking effect for the former. Neurite outgrowth rates of PC12 cells were similar regardless of the fiber direction. Cellular adhesion and neurite outgrowth varied significantly across the gradient, based on the local protein concentration. Neurites oriented along the fiber axis for both sample types, indicating that the topographical cues were a more dominant factor in neurite orientation than chemical cues. Scaffolds with protein concentration gradients yielded enhanced neurite outgrowth over uniform coverage samples. Thus, these protein concentration gradient scaffolds provide not only a high level of control of neurite outgrowth based on regional protein concentration, but also improve the

011003-9 N. E. Zander and T. P. Beebe, Jr.: Immobilized laminin concentration gradients on electrospun fiber scaffolds

outgrowth rates compared to conventional uniform protein coverage scaffolds. Based on our results, these materials could offer better outcomes for the treatment spinal cord and peripheral nerve injuries. It would be interesting to compare step gradients to continuous gradients formed over an 8 h period to probe the effect of submicron scale protein concentration changes. The exploration of other types of gradients could also be worthwhile, such as mechanical property gradients (stiffness) since this is known to influence neurite outgrowth.52–54 In addition, roughness gradients have also been reported to control neurite outgrowth.55 It is believed that such gradient materials will serve as next-generation scaffolds.56 ACKNOWLEDGMENTS The authors declare that they have no competing interests. N.Z. conceived of the idea, prepared and characterized the samples. T.B. participated in the design of experiments, data analysis, and helped draft the manuscript. Both authors read and approved the final manuscript. 1

S. Sant, M. J. Hancock, J. P. Donnelly, D. Iyer, and A. Khademhosseini, Can. J. Chem. Eng. 88, 899 (2010). 2 M. Singh, C. Berkland, and M. S. Detamore, Tissue Eng., Part B Rev. 14, 341 (2008). 3 B. Li, Y. X. Ma, S. Wang, and P. M. Moran, Biomaterials 26, 1487 (2005). 4 R. T. Petty, H. W. Li, J. H. Maduram, R. Ismagilov, and M. Mrksich, J. Am. Chem. Soc. 129, 8966 (2007). 5 S. J. Singer and A. Kupfer, Annu. Rev. Cell Biol. 2, 337 (1986). 6 L. Y. Liu, B. D. Ratner, E. H. Sage, and S. Y. Jiang, Langmuir 23, 11168 (2007). 7 J. T. Smith, J. K. Tomfohr, M. C. Wells, T. P. Beebe, T. B. Kepler, and W. M. Reichert, Langmuir 20, 8279 (2004). 8 D. E. Willis, E. A. van Niekerk, Y. Sasaki, M. Mesngon, T. I. Merianda, G. G. Williams, M. Kendall, D. S. Smith, G. J. Bassell, and J. L. Twiss, J. Cell Biol. 178, 965 (2007). 9 H. J. Song and M. M. Poo, Nat. Cell Biol. 3, E81 (2001). 10 C. ffrench-Constant and H. Colognato, Trends Cell Biol. 14, 678 (2004). 11 J. L. Roam, H. Xu, P. K. Nguyen, and D. L. Elbert, Biomaterials 31, 8642 (2010). 12 D. N. Adams, E. Y. C. Kao, C. L. Hypolite, M. D. Distefano, W. S. Hu, and P. C. Letourneau, J. Neurobiol. 62, 134 (2005). 13 B. K. Mueller, Annu. Rev. Neurosci. 22, 351 (1999). 14 S. K. W. Dertinger, X. Y. Jiang, Z. Y. Li, V. N. Murthy, and G. M. Whitesides, Proc. Natl. Acad. Sci. U.S.A. 99, 12542 (2002). 15 R. C. Gunawan, J. Silvestre, H. R. Gaskins, P. J.A. Kenis, and D. E. Leckband, Langmuir 22, 4250 (2006). 16 S. Kramer, H. Xie, J. Gaff, J. R. Williamson, A. G. Tkachenko, N. Nouri, D. A. Feldheim, and D. L. Feldheim, J. Am. Chem. Soc. 126, 5388 (2004). 17 K. Moore, M. Macsween, and M. Shoichet, Tissue Eng. 12, 267 (2006). 18 J. Ballester-Beltran, M. Cantini, M. Lebourg, P. Rico, D. Moratal, A. J. Garcia, and M. Salmeron-Sanchez, J. Mater. Sci.: Mater. Med. 23, 195 (2012). 19 R. Biran, K. Webb, M. D. Noble, and P. A. Tresco, J. Biomed. Mater. Res. 55, 1 (2001). 20 S. R. Y. Cajal, Anat. Anz 5, 609 (1890). 21 R. W. Sperry, Proc. Natl. Acad. Sci. U.S.A. 50, 703 (1963). 22 L. Yao, S. G. Wang, W. J. Cui, R. Sherlock, C. O’Connell, G. Damodaran, A. Gorman, A. Windebank, and A. Pandit, Acta Biomater. 5, 580 (2009).

Biointerphases, Vol. 9, No. 1, March 2014

23

011003-9

R. A. Neal, S. G. McClugage, M. C. Link, L. S. Sefcik, R. C. Ogle, and E. A. Botchwey, Tissue Eng. Part C-Methods 15, 11 (2009). 24 J. Graf, Y. Iwamoto, M. Sasaki, G. R. Martin, H. K. Kleinman, F. A. Robey, and Y. Yamada, Cell 48, 989 (1987). 25 Y. Iwamoto, F. A. Robey, J. Graf, M. Sasaki, H. K. Kleinman, Y. Yamada, and G. R. Martin, Science 238, 1132 (1987). 26 S. K. Powell and H. K. Kleinman, Int. J. Biochem. Cell B 29, 401 (1997). 27 K. Tashiro, G. C. Sephel, B. Weeks, M. Sasaki, G. R. Martin, H. K. Kleinman, and Y. Yamada, J. Biol. Chem. 264, 16174 (1989). 28 S. Y. Boateng, S. S. Lateef, W. Mosley, T. J. Hartman, L. Hanley, and B. Russel, Am. J. Physiol.: Cell Physiol. 288, C30 (2005). 29 C. Deister, S. Aljabari, and C. E. Schmidt, J. Biomater. Sci., Polym. Ed. 18, 983 (2007). 30 M. C. Dodla and R. V. Bellamkonda, Biomaterials 29, 33 (2008). 31 X. Cao and M. S. Shoichet, Neuroscience 103, 831 (2001). 32 X. Cao and M. S. Shoichet, Neuroscience 122, 381 (2003). 33 T. A. Kapur and M. S. Shoichet, J. Biomed. Mater. Res. A 68A, 235 (2004). 34 T. B. Bini, S. J. Gao, T. C. Tan, S. Wang, A. Lim, L. B. Hai, and S. Ramakrishna, Nanotechnology 15, 1459 (2004). 35 J. M. Corey, D. Y. Lin, D. C. Martin, and E. L. Feldman, Ann. Neurol. 58, S65 (2005). 36 H. Q. Cao, T. Liu, and S. Y. Chew, Adv. Drug Deliv. Rev. 61, 1055 (2009). 37 D. Grafahrend, J. L. Calvet, K. Klinkhammer, J. Salber, P. D. Dalton, M. Moller, and D. Klee, Biotechnol. Bioeng. 101, 609 (2008). 38 W. S. Li, Y. Guo, H. Wang, D. J. Shi, C. F. Liang, Z. P. Ye, F. Qing, and J. Gong, J. Mater. Sci.: Mater. Med. 19, 847 (2008). 39 J. Shi, L. Wang, F. Zhang, H. Li, L. Lei, L. Liu, and Y. Chen, ACS Appl. Mater. Interfaces 2, 1025 (2010). 40 N. E. Zander, J. A. Orlicki, A. M. Rawlett, and T. P. Beebe, J. Mater. Sci.: Mater. Med. 24, 179 (2013). 41 H. S. Koh, T. Yong, C. K. Chan, and S. Ramakrishna, Biomaterials 29, 3574 (2008). 42 See supplementary material at http://dx.doi.org/10.1116/1.4857295 for covalent immobilization of laminin and removal of unconjugated laminin, immobilization kinetics of laminin step gradients, adhesion of PC12 cells, PC12 neurite outgrowth data separated by individual experiment. 43 N. E. Zander, J. A. Orlicki, A. M. Rawlett, and T. P. Beebe, ACS Appl. Mater. Interfaces 4, 2074 (2012). 44 N. Zander, J. Orlicki, A. Rawlett, and T. Beebe, Biointerphases 5, 149 (2010). 45 S. Y. Chew, J. Wen, E. K. F. Yim, and K. W. Leong, Biomacromolecules 6, 2017 (2005). 46 J. Y. Lee, C. A. Bashur, A. S. Goldstein, and C. E. Schmidt, Biomaterials 30, 4325 (2009). 47 A. Kundu, L. Micholt, S. Friedrich, D. R. Rand, C. Bartic, D. Braeken, and A. Levchenko, Lab Chip 13, 3070 (2013). 48 J. A. Richardson, “Neurite response to topographical and molecular cues,” Ph. D. dissertation [Online], Brown University, Providence, RI, https:// repository.library.brown.edu/fedora/objects/bdr:297581/datastreams/PDF/ content (Accessed 26 November 2013) (2012). 49 N. Gomez, S. Chem, and C. E. Schmidt, J. R. Soc. Interface 4, 223 (2007). 50 N. Gomez, Y. Lu, S. Chen, and C. E. Schmidt, Biomaterials 28, 271 (2007). 51 L. Forciniti, C. E. Schmidt, and M. H. Zaman, Ann. Biomed. Eng. 37, 363 (2009). 52 J. B. Leach, X. Q. Brown, J. G. Jacot, P. A. DiMilla, and J. Y. Wong, J. Neural Eng. 4, 26 (2007). 53 M. L. Previtera, C. G. Langhammer, and B. L. Firestein, J. Biosci. Bioeng. 110, 459 (2010). 54 S. Y. Chou, C. M. Cheng, C. C. Chen, and P. R. LeDuc, Soft Matter 7, 9871 (2011). 55 X. R. Li, M. R. MacEwan, J. W. Xie, D. Siewe, X. Y. Yuan, and Y. N. Xia, Adv. Funct. Mater. 20,1632 (2010). 56 R. V. Bellamkonda, Biomaterials 27, 3515 (2006).