- Email: [email protected]
Multiple-channel scaffolds to promote spinal cord axon regeneration
ARTICLE IN PRESS
Biomaterials 27 (2006) 419–429 www.elsevier.com/locate/biomaterials
Multiple-channel scaffolds to promote spinal cord axon regeneration Michael J. Moorea, Jonathan A. Friedmanb, Eric B. Lewellync, Sara M. Mantilaa, Aaron J. Krychd, Syed Ameenuddinc, Andrew M. Knightc, Lichun Lua,e, Bradford L. Currierd,e, Robert J. Spinnerd,f, Richard W. Marshd,f, Anthony J. Windebankc,d, Michael J. Yaszemskia,d,e, a
Department of Physiology and Biomedical Engineering, Mayo Clinic College of Medicine, Rochester, MN 55905, USA b Section of Neurosurgery, Dartmouth Hitchcock Medical Center, Lebanon, NH 03756, USA c Department of Neurology, Mayo Clinic College of Medicine, Rochester, MN 55905, USA d Mayo Medical School, Mayo Clinic College of Medicine, Rochester, MN 55905, USA e Department of Orthopedic Surgery, Mayo Clinic, MS 3-75, 200 First Street, SW, Rochester, MN 55905, USA f Department of Neurologic Surgery, Mayo Clinic College of Medicine, 200 First Street, SW, Rochester, MN 55905, USA Received 30 June 2005; accepted 27 July 2005 Available online 31 August 2005
Abstract As molecular, cellular, and tissue-level treatments for spinal cord injury are discovered, it is likely that combinations of such treatments will be necessary to elicit functional recovery in animal models or patients. We describe multiple-channel, biodegradable scaffolds that serve as the basis for a model to investigate simultaneously the effects on axon regeneration of scaffold architecture, transplanted cells, and locally delivered molecular agents. Poly(lactic-co-glycolic acid) (PLGA) with copolymer ratio 85:15 was used for these initial experiments. Injection molding with rapid solvent evaporation resulted in scaffolds with a plurality of distinct channels running parallel along the length of the scaffolds. The feasibility of creating scaffolds with various channel sizes and geometries was demonstrated. Walls separating open channels were found to possess void fractions as high as 89%, with accessible void fractions as high as 90% through connections 220 mm or larger. Scaffolds degraded in vitro over a period of 30 weeks, over which time-sustained delivery of a surrogate drug was observed for 12 weeks. Primary neonatal Schwann cells were distributed in the channels of the scaffold and remained viable in tissue culture for at least 48 h. Schwann-cell containing scaffolds implanted into transected adult rat spinal cords contained regenerating axons at one month post-operation. Axon regeneration was demonstrated by three-dimensional reconstruction of serial histological sections. r 2005 Elsevier Ltd. All rights reserved. Keywords: Nerve tissue engineering; Scaffold; Microstructure; Schwann cell; Drug release; Image analysis
1. Introduction Advances in neuroscience over the past 2 decades begin to offer hope for spinal cord injury (SCI) victims. Since the demonstration in 1980 that central nervous system (CNS) axons have the capacity to regenerate Corresponding author. Department of Orthopedic Surgery, Mayo Clinic, MS 3-75, 200 First Street SW, Rochester, MN 55905, USA. Tel.: +1 507 284 2267; fax: +1 507 284-5075. E-mail address: [email protected] (M.J. Yaszemski).
0142-9612/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2005.07.045
within peripheral nervous system (PNS) grafts [1], much has been accomplished toward understanding factors that contribute to a physiologically permissive environment. Mechanisms of injury, of regeneration, and of inhibition to regeneration are being delineated, and several promising treatment strategies have arisen. Transplantation of a variety of cell types, including Schwann cells [2,3], olfactory ensheathing glial cells [4,5], or neural stem cells [6,7], has resulted in axon regeneration and limited functional improvement after spinal cord injury in rats. Molecular therapies that work
ARTICLE IN PRESS 420
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
to promote regeneration, such as administration of neurotrophins [8–10], and those that target deleterious inhibition of regeneration, such as chondroitinase ABC [11–15], have also yielded favorable results. Synthetic biomaterials have been investigated for their ability to reconstruct spinal cord tissue architecture, to provide guidance for regenerating axons, and to prevent the infiltration of scar tissue [16–19]. A comprehensive review of neural regeneration strategies was recently provided by Schmidt and Baier Leach [20]. Despite recent advances, the limited demonstration of functional improvement in animal models has prevented advancement of any regenerative therapy to clinical use. This may be due in large part to the multifaceted nature of spinal cord injuries, which presents a major challenge to therapeutic development. Primary mechanical trauma to the cord induces secondary injury consisting of a complex cascade of molecular events that lead to the loss of myelin and the formation of a glial scar [21,22]. Therefore, in order for viable treatment strategies to be realized clinically, it is likely that combinations of current therapeutic approaches must be used. Indeed, combinatorial approaches have already shown promise in animal models. For example, administration of neurotrophins enhanced axon growth into Schwann-cell seeded guidance channels and increased integration into the graft–host interface [23]. Synergistic effects on CNS axon regeneration have been demonstrated when strategies promoting regeneration and antagonizing inhibition were used simultaneously [24,25]. An elegant solution may lie in the design of a bioartificial graft that targets injury mechanisms at the molecular, cellular, and tissue levels. Biodegradable polymers can simultaneously provide a tissue scaffold, a cell delivery vehicle, and a reservoir for sustained drug delivery [16]. This integrative approach suggests a possible treatment strategy and may serve as an in vivo model for studying optimization of various combinations of treatments. We describe techniques for producing biodegradable polymer scaffolds with parallel-channel architecture that can be systematically modified. They may be seeded with multiple cell types arranged spatially in anatomically relevant locations and may serve as a vehicle for sustained drug delivery. We quantitatively describe the scaffolds’ architecture, their in vitro degradation profile, their drug delivery characteristics, their biocompatibility with Schwann cells in culture, and their promotion of in vivo axon regeneration.
evaporation technique. Cylindrical, Teflon molds, with diameter 3.0 mm were fitted with Delrin spacers containing an array of seven, uniformly spaced, 508- or 660-mm stainless-steel wires (Malin, Cleveland, OH), as shown in Fig. 1A. The wire arrays were spray-coated with a minimal amount of Ease Release 200 (Mann Formulated Products, Easton, PA) mold lubricant to facilitate removal. A concentrated solution of poly(D,L-lactic-co-glycolic acid) (PLGA, Alkermes, Cambridge, MA), with copolymer ratio 85:15 lactide:glycolide and number average molecular weight (Mn) 75,000, was made by adding 1.0 g PLGA to 2.0 ml dichloromethane (DCM) in a glass vial and shaking vigorously for 3 h. The viscous PLGA solution was injected with a syringe through a 16-gauge hypodermic needle from the bottom of each mold until solution was seen escaping through the top. Polymer-filled molds were vacuum-dried for at least 24 h with a high-vacuum pump (VP 190, Savant Instruments, Holbrook, NY) connected to a condensation trap (RT 4104, Savant Instruments). Vacuum drying removed the solvent and created pores in the walls of the scaffolds. Scaffolds to be used in cell culture or in vivo experiments were washed in ethanol for 30 min with gentle shaking both to disinfect the scaffolds and to remove any residual mold lubricant. Scaffolds were again vacuum-dried for 24 h to remove the ethanol, then sealed in sterilized glass vials and stored desiccated at 4 1C until further use. Creation of scaffolds with more complex, biomimetic architecture was demonstrated by fabrication of molds using computer-aided design (CAD) and solid freeform fabrication (SFF). A PatternMaster three-dimensional (3-D), piezo-inkjet printer (Solidscape, Merrimack, NH) was used to create wax molds from models designed manually with ProEngineer software (Parametric Technology Corp., Needham, MA). Fig. 1B shows one wax mold manufactured by the PatternMaster. Injection of PLGA polymer solution followed by vacuum-evaporation of solvent resulted in PLGA scaffolds 1 cm in diameter. These were used to test the principle that complex architecture mimicking the structure of the human spinal cord can be fabricated.
2. Materials and methods
Fig. 1. Injection molds for fabrication of multiple-channel scaffolds. (A) Cut-away view of parallel-wire injection molding apparatus. Mold inner diameter ¼ 3.0 mm. Polymer solution was injected with a syringe through an injection canal, as shown in the model. (B) Photograph of wax mold fabricated with PatternMaster for creation of scaffold by injection of PLGA. Parallel cross bars are for support of inner mold parts. Mold inner diameter ¼ 1 cm.
2.1. Scaffold fabrication Biodegradable scaffolds with controlled, parallel-channel architecture were fabricated by an injection molding, solvent
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
421
2.2. Morphological characterization
diameter ¼ 0.50 mm:
Scaffold architecture was viewed with X-ray microcomputed tomography (micro-CT or mCT). A custom-built mCT scanner [26], housed in the Mayo Physiological Imaging Research Laboratory, provided 3-D interior images which were viewed and manipulated with Analyze AVW (Mayo Foundation, Rochester, MN) [27], a comprehensive image display and analysis software package developed by the Mayo Biomedical Imaging Resource. Void fraction and accessible void fraction were measured from mCT images by a method described in detail elsewhere [28]. In summary, this program uses morphological operations to estimate the percentage of the void volume of a scaffold that maintains connections to the outside air through openings of a given minimum size.
ph 2 (2) ðD nd 2 Þ ¼ 85:4 mm3 . 4 If porosity is taken as the overall porosity, including that contributed by the pores and the channels, then V in Eq. (1) is merely the volume of the 3.0-mm-diameter and 15.0-mm-long cylindrical mold, i.e. 106 mm3. Scaffolds were made as in the above section, except that the amount of PLGA added was varied to create solutions of increasing polymer concentration. Solutions of 30%, 40%, 50%, 60%, and 70% (w/v) PLGA were made by dissolving in DCM. The resulting scaffolds were weighed on a Denver Instruments (Arvada, CO) M-220D electronic balance, and their void fraction estimated by the calculations in Eqs. (1) and (2).
2.3. Modulation of porosity
2.4. Scaffold degradation
For the purpose of clarity, scaffold channels are defined as the voids left by the stainless-steel wires, walls as the porous, polymer-occupying regions between the channels, and pores as the voids within the walls of the scaffold left by evaporated solvent (see Fig. 2). Wall porosity was estimated from mold dimensions, scaffold weights, polymer density, and Eq. (1) below, where V ¼ mold void volume, r ¼ density of PLGA ¼ 1.2 mg/mm3, and m ¼ measured mass (weight) of scaffold: m 100%. (1) wall porosity ð%Þ ¼ 1 Vr
Individual scaffolds were heat-sealed in pouches made from 10-mm, nylon mesh sheets (Spectrum Labs, Rancho Dominguez, CA) and vacuum-dried for 24 h before obtaining tare weights. Dried samples were then immersed in 20 ml phosphate-buffered saline (PBS) at pH 7.4 and incubated at 37 1C. All solutions were replaced, and their pH monitored, weekly. At regular intervals, for a total of 30 weeks, three samples were removed, washed repeatedly in distilled, deionized water, and vacuum-dried for 24 h. Dried pouches were immersed in 1 ml tetrahydrofuran (THF) and shaken overnight to dissolve the scaffolds. Empty pouches were then removed, washed repeatedly in DCM, vacuum-dried overnight, and weighed. Degradation was characterized by mass loss, molecular weight loss, and morphological change. Mass loss was determined by weighing dried samples at a given time point (mt) and comparing to the initial mass (mo): mt 100%. (3) mass loss ð%Þ ¼ 1 mo
The mold void volume V is given by Eq. (2), where h is the mold height ¼ 15.0 mm, D is the mold diameter ¼ 3.0 mm, n is the number of wires ¼ 7, and d is the wire
V¼
Polymer molecular weight loss was characterized by gel permeation chromatography (GPC) with a Waters 717 Plus Autosampler GPC system (Waters, Milford, MA) connected to a model 515 HPLC pump and model 2410 refractive index detector. The columns consisted of a Styragel HT guard column (7.8 300 mm, Waters) in series with a Styragel HR 4E column (7.8 300 mm, Waters). Degraded and undegraded scaffolds, for comparison, were measured by injecting 20 ml of each sample and eluting at 1 ml/min flow rate in THF. Monodisperse polystyrene standards (Polysciences, Warrington, PA) with Mn of 0.474, 6.69, 18.6, and 38 kD and polydispersities of less than 1.1 were used to construct the calibration curve. Morphological changes were characterized by monitoring degrading samples using mCT. 2.5. Release of model drug
Fig. 2. PLGA scaffolds created from parallel wire molds. Top panels: photographs, and bottom panels: mCT slices. Left: cross-sectional view, and right: longitudinal view. Each scale bar ¼ 500 mm.
Fluorescein isothyocyanate-dextran (FITC-D, average Mw 167 kD, Sigma, St. Louis, MO) served as a model drug for estimating release kinetics of proteins in a similar molecular weight range. After dissolution of PLGA in DCM, 50 mg FITC-D was added to the solution and stirred vigorously before vortexing for an additional 3 h. The polymer/solvent/
ARTICLE IN PRESS 422
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
drug mixture was then injected into molds and vacuum-dried for 24 h. Dried scaffolds were weighed, heat-sealed in pouches made from Spectra/Mesh nylon filters (10 mm mesh), then placed in 1.0 ml PBS at pH 7.4 and 37 1C. At each time point, the nylon pouches were removed and placed in 1.0 ml of fresh PBS, and the absorbance at 490 nm of the conditioned supernatant was measured on a spectrophotometer (SPECTRAmax PLUS 384, Molecular Devices, Sunnyvale, CA) and compared with a calibration curve of known FITC-D concentrations. Drug loading was estimated for each scaffold by calculating the mass ratio of FITC-D to the PLGA used, then multiplying this ratio by the mass of each scaffold. The released FITC-D was calculated as the cumulative percentage of the initial loading for each individual scaffold.
2.6. Schwann cell culture and evaluation All experiments involving animals were approved by the Institutional Animal Care and Use Committee (IACUC). Male, transgenic Sprague-Dawley rats expressing green fluorescent protein (GFP) (Genome Information Research Center, Osaka, Japan) were mated with female SpragueDawley rats (Charles River Laboratories, Wilmington, MA). GFP-positive rat pups were identified with an ultraviolet lamp. Primary neonatal Schwann cells were obtained by methods similar to those first described by Brockes et al. [29]. Two- to five-day-old rat pups, both GFP-positive and negative, were anesthetized by intraperitoneal injection of 0.05–0.1 ml sodium pentobarbital. Sciatic nerves from both legs were isolated, sectioned into small pieces (1 mm3), and digested enzymatically for 45 min with 0.25% trypsin and 0.03% collagenase in Hank’s balanced salts solution. After digestion, cells were pelleted at 800 rpm in a Beckman TJ-6 centrifuge for 5 min, the supernatant was removed, and cells were resuspended in 5 ml of DMEM/F12 containing 10% fetal bovine serum (FBS) and 100 units/ml penicillin/streptomycin. Cells were dissociated by repeatedly passing them through a narrowed Pasteur pipette, then added to a laminin-coated 35-mm dish, and incubated at 37 1C and 5% CO2 for 3 days. Schwann cells were identified by characteristic elongated spindle body shape, a prominent nucleus, and bi- or tripolar extensions. These findings were confirmed by immunostaining in GFP-negative animals. Culture dishes were rinsed with PBS, fixed with 1:1 methanol/acetate solution, labeled with anti-S-100 polyclonal antibody (NeoMarkers, Fremont, CA), biotinylated anti-rabbit secondary antibody (Dako, Carpinteria, CA), streptavidin HRP (Dako), and stained with diaminobenzene (DAB) (Sigma). Cells were detached with trypsin-EDTA for 2 min, resuspended in chilled Matrigel (BD Biosciences, Bedford, MA) at a target concentration of approximately 1.2 108 cells/ml [30,31], and then added to each channel of the scaffold with a gel-loading pipette tip. The volume of the cell suspension added was on the order of 1 ml/channel. Cell-loaded scaffolds were cut to 2 mm in length and incubated in DMEM/F12 medium with 10% FBS for 48 h prior to microscopic evaluation in vitro, or 24 h prior to implantation in vivo. Scaffolds loaded with cells suspended in Matrigel were fixed in 4% paraformaldehyde and cut into 0.2 mm sections on a Reichert-Jung (Leica Microsystems, Germany) Biocut
microtome. Sections were viewed with fluorescence optical microscopy (Zeiss Axiovert 35, Germany), and also processed for transmission electron microscopy (JEOL 1200EX, Peabody, MA). 2.7. Animal surgery and post-operative care Female Sprague-Dawley rats (Charles River Laboratories) weighing 300–350 g were deeply anesthetized with intraperitoneal injection of Ketamine 80 mg/kg (Fort Dodge Animal Health, Fort Dodge, IA) and Xylazine 5 mg/kg (Lloyd Laboratories, Shenandoah, IA). Animals were placed on a heating pad constantly maintained at 37 1C during surgery, and ophthalmic ointment was used to prevent drying of the eyes. A 4-cm incision was made along the dorsal midline. The paraspinal muscles were dissected subperiosteally from the spinous processes and laminae. Multiple laminectomies were performed with sharp scissors at levels T8–T10 while ensuring that the facet joints were not violated. The posterior aspect of the spinal cord was exposed and the dura was cut vertically using microforceps and microscissors. An angled microscissors was used to transect fully the spinal cord, and a 1.0-mm segment of spinal cord at level T9 was removed. The stumps were retracted making a 2.0-mm gap in the spinal cord. Complete transection was confirmed by irrigating the site with lactated Ringer’s solution until the bottom of the canal was visible. Biodegradable implants containing either Matrigel alone or Schwann cells suspended in Matrigel were inserted into the canal, making sure that the severed ends of the cord fit tightly. Control animals received no implants. Exposed spinal cords with or without scaffolds were covered with muscles and fascia. Before closing the wound, the spine was stabilized by suturing two 2-mm 2-cm, non-biodegradable polypropylene shards along the spinous processes of T4–T10. A Zeiss (Germany) S3B microsurgical microscope was used during the entire surgical procedure. Animals were injected with 5.0 ml lactated Ringer’s solution subcutaneously after surgery and allowed to recuperate. Buprenex 0.05 mg/kg was given subcutaneously for 3–4 days to minimize pain. Baytril 65 mg/kg was given intramuscularly for 1 week and as needed thereafter to prevent infection. All rats were kept in low-sided cages to enable them easy access to food and water, and bedding was changed frequently to prevent decubitus ulcers. Daily post-operative care was undertaken by maintaining and monitoring body temperature, general fluid and electrolyte balance, and cardiovascular homeostasis by direct observation. Bladders were manually expressed twice daily for the duration of the experiment. 2.8. Tissue preparation Animals were sacrificed at the end of 1 month. Animals were deeply anesthetized with 0.3 ml sodium pentobarbital and transcardially perfused with 100–120 ml chilled PBS followed by 120–150 ml 4% paraformaldehyde. After perfusion, the spinal column was removed and post-fixed for 24 h in the same fixative. The vertebrae were carefully cut away, and isolated spinal cords with or without scaffolds were soaked for 48 h in 4% paraformaldehyde and stored in PBS at 4 1C until they
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
423
were processed for histology and immunohistochemistry. Spinal cords with scaffolds were dehydrated in a mixture of ethanol and Citrisolv (Fisher Scientific, Pittsburg, PA), to prevent dissolution of scaffold, and embedded in paraffin. Tissue blocks were cut into 8 mm thick sections (Leica Microsystems) and stained with mouse monoclonal antineurofilament antibody against phosphorylated neurofilament-H (clone 2F11, Dako). 2.9. 3-D image reconstruction Digital photomicrographs were obtained at 2.5 and 20 using a Zeiss Axioplan II microscope equipped with an AxioCam digital camera. All registration, segmentation, and rendering operations were performed with Analyze AVW software (Mayo Foundation) [27] using methods discussed in greater detail elsewhere [32]. Serial histological images were registered automatically with an automatic registration algorithm. Individual slices were then registered manually to correct for errors in alignment. Regions corresponding to channels or axon bundles were segmented from surrounding scaffold and tissue regions using automatic multispectral analysis with a mode filter. Image segmentation was further altered manually with the use of image editing and morphological operations. Sixteen consecutive slices were reconstructed and rendered assuming isotropic voxels, and reconstructed images were displayed orthographically with wedging to reveal internal channel or regenerating axon geometry.
Fig. 3. Modulation of void fraction. Upper panel contains inverted, summed-voxel projections representative of scaffolds made with varying polymer concentrations. These concentrations correspond to those plotted in the lower panel, where void fractions measured gravimetrically are plotted against polymer concentrations. Error bars indicate7standard deviation; n ¼ 5. Note broken ordinate axis.
3. Results 3.1. Scaffold morphology Scaffolds fabricated with the injection molding technique possessed macro-architecture consistent with the mold design, as shown by the photographs in Fig. 2. The scaffolds’ geometrical and dimensional properties were reproducible and were maintained with a fair degree of manual manipulation, such as what might be expected during neurosurgical implantation. Micro-CT analysis revealed the seven large channels that remained intact throughout the length of each scaffold, with highly porous walls separating the channels, as shown in Fig. 2. Pores within the walls, formed by the evaporating solvent, exhibited a wide range of sizes, with apparent longitudinal isotropy and radial anisotropy. Channel diameters were measured with image analysis and found to be slightly less than that of the wires that formed them. Wires which were 510 mm in diameter created channels with diameters of 45079 mm, and those which were 660 mm created channels 600710 mm in diameter. 3.2. Modulation of void fraction Fabrication of scaffolds with differential concentrations of PLGA solutions yielded scaffolds with varying
void fractions. Qualitative differences in void fraction were apparent from mCT images, as indicated by the inverted, summed-voxel projections in Fig. 3. The projections shown are similar to conventional X-rays, but in these negative images, PLGA appears dark. Gravimetric void fraction measurement confirmed quantitatively a nearly linear trend in void fraction as a function of polymer concentration. The void fraction of the walls displays a similar linear trend as the overall void fraction, though as expected, the wall void fraction is less than the overall, due to the large volume occupied by the open channels. Analysis of mCT images confirmed changes in pore interconnectivity as a function of the overall void fraction. As a measure of pore interconnectivity, the accessible void fraction was estimated for a range of minimum connection sizes for three of the groups with modulated porosity. This quantity is an estimate of the fraction of void volume that is connected to either the channels or the ends of the scaffold, through connections of the minimum size indicated. The groups chosen possessed a high overall void fraction (88.6%, measured gravimetrically), a medium void fraction (83.4%), and a low void fraction (78.7%). The data in Fig. 4 show that as the void fraction increases, the accessible void fraction for a given minimum connection size also
ARTICLE IN PRESS 424
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
increases, i.e. a higher porosity yields a higher interconnectivity. 3.3. Scaffold degradation Scaffolds in simulated physiologic conditions were found to degrade as expected for PLGA with the given copolymer ratio and molecular weight. As shown by the degradation profile in Fig. 5, scaffold molecular weight decreased steadily from 100% of the initial molecular weight down to 5% by 26 weeks of degradation. In
contrast, little or no mass loss was observed for the first 20 weeks, followed by a period of more precipitous mass loss over the next 10 weeks. The observed sigmoidal mass loss behavior is typical for PLGA degradation. Only slight changes in morphology were apparent during the period when negligible change in mass was observed. As mass loss occurred, however, the morphology of the degrading scaffolds changed drastically. Fig. 6 shows transverse and longitudinal cross-sections of a scaffold after 8 weeks of degradation and another scaffold after 24 weeks. The scaffold degraded for 8 weeks appears to possess slight distortions of external geometry and channel geometry, as compared to undegraded specimens (Fig. 2). The scaffold degraded for 24 weeks, on the other hand, shows evidence of pronounced deformations in geometry and dimension, as seen in the lower panel of Fig. 6. Marked reduction in void fraction is apparent qualitatively, and the seven channels no longer appear to run contiguously throughout the length of the scaffold. 3.4. Release of model drug
Fig. 4. Accessible void fractions for three selected scaffold groups. Accessible void volume, as a percent of the overall void volume, is given for scaffold groups with high (88.6%), medium (83.4%), and low (78.7%) overall void fractions, measured gravimetrically, over a range of minimum connection sizes. Error bars indicate7standard deviation; n ¼ 4. Note broken ordinate axis.
Upon visual inspection, mechanical mixing of FITCD within the polymer scaffolds was an effective method for scaffold loading, though dispersal appeared to be inhomogeneous. Fig. 5 shows the FITC-D release profile compared with the mass and molecular weight degradation profiles. Release of the model drug into PBS commenced with an initial burst release over the first 48 h, followed by a period of more steady release for 4 weeks. Sustained release continued, though even more slowly, for an additional 8 weeks, after which little or no
Fig. 5. Degradation and FITC-D release profile. Mass or molecular weight change is indicated as a percentage of initial value at time zero; error bars indicate7standard deviation, n ¼ 3. FITC-D release is indicated as a cumulative percentage of loaded drug; means of two simultaneous experiments, each with n ¼ 5, are plotted with error bars indicating standard error of the mean.
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
425
release was detected for the remainder of the degradation period. 3.5. Schwann cell culture Harvesting and seeding of Schwann cells produced viable primary cultures that were approximately 90% pure, the remainder being fibroblasts. As seen in Fig. 7A, cells staining positive for S-100 were accompanied by relatively few fibroblasts, and possessed morphology consistent with that of Schwann cells. Suspension of primary cells in Matrigel was relatively facile, and injection of cell suspensions into individual channels resulted in a distribution of viable cells within distinct channels of the scaffold. Fig. 7B shows GFP-labeled Schwann cells densely populating the seven channels of the scaffold. Further evidence for the identity and viability of the cells in the scaffold is given in Fig. 7C, which is a transmission electron micrograph (TEM) of a Schwann cell. This image shows loosely packed chromatin with areas of condensation in the sub-membranous regions of the nucleus. Fig. 6. Morphology of degraded scaffolds. Top panel shows transverse and longitudinal sections of a scaffold after 8 weeks of degradation. Bottom panel, after 24 weeks. Top and bottom panels do not represent identical specimens. Each scale bar ¼ 500 mm.
3.6. Axon regeneration Scaffolds containing Schwann cells suspended in Matrigel promoted axon regeneration at 1 month after
Fig. 7. Schwann cells isolated from neonatal rat pups and cultured in vitro. (A) Phase contrast micrograph of primary Schwann cell culture stained with S-100. (B) Light micrograph of a transverse section through the scaffold, showing distribution of GFP-positive cells suspended in Matrigel within each of the 7 channels. Some scaffold autofluorescence is apparent; bar ¼ 500 mm. (C) TEM of Schwann cell with indicative morphology and contained in scaffold; scale bar ¼ 2 mm.
ARTICLE IN PRESS 426
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
implantation. Animals receiving no treatment, or those which received scaffolds with no Schwann cells possessed few or no regenerating axons (data not shown). Histology revealed the channels of the scaffold containing ample axon regeneration, which was apparent throughout the complete length of the scaffold. As shown in Fig. 8, the channels were found distributed approximately as arranged by the scaffold, with scaffold and tissue in the spaces separating the channels. Channels generally contained a centralized core of tissue containing axons and capillaries, surrounded by circum-
Fig. 8. Histological images of tissue cables within scaffold channels. (A) Histological cross section near rostral end of scaffold showing tissue cables within all seven channels; bar ¼ 500 mm and applies to both A and B. Boxed region is shown in greater detail in D. (B) Histological cross section shown in A with channels segmented and displayed in color. (C) Orthographic, wedged view of 16 serial slices registered, segmented, and reconstructed to reveal 3-D structure of channels. (D) Histological cross section of boxed region in A; scale bar ¼ 100 mm. (E) Orthographic, wedged view of reconstructed channel shown in D, with segmented axon bundles displayed in color.
ferential fibrous tissue. Macrophages could be identified, likely due to their engulfment of neurofilament-stained material. The 3-D reconstructions shown in Fig. 8C and D show clearly the arrangement of the seven channels containing regenerating axons, as well as live axon bundles within the channels.
4. Discussion The fabrication technique described was chosen for its versatility and simplicity. Virtually any polymer, degradable or inert, could conceivably be used, along with an appropriate solvent if necessary. The scaffold components experience only temperatures and pressures at or below ambient in this injection molding process. Thus, bioactive molecules can be incorporated without risk of denaturation or degradation due to high heat or pressure. A variety of techniques for the production of pores can be used either in addition to, or in lieu of, the solvent evaporation procedure. Further, manipulation of longitudinal geometry is achieved easily with this technique. Variations on the multiple-channel design have been reported for use in the regenerating spinal cord [33,34], and designs similar to that presented here have been described for use in peripheral nerve regeneration [35,36]. Other authors have suggested that scaffold architecture may play a role in maximizing axon regeneration in the spinal cord [17,37], but the optimum architecture as of yet remains unknown. The parallel, multichannel geometry described in this paper may provide certain advantages. First, the number, size, and arrangement of the channels are easily manipulated by choice of mold design. This may allow for the systematic manipulation of scaffold architecture and the testing of its impact, if any, on axon regeneration. Further, because the channels are distinct, as opposed to a global, isotropically oriented pore network, transplantation of cells into specific regions of the scaffold is achievable. Using either the simple parallel-wire approach, or the more complex SFF technique, it appears feasible to design scaffolds that may be used to transplant cells into appropriate regions of the scaffolds’ cross sections. These regions may correspond, for instance, to gray and white matter, ascending and descending pathways, and even distinct white matter tracts. The multiple channel design may help prevent collapse of the conduit containing a regenerating tissue cable, a phenomenon sometimes observed in degradable, hollow tubes [19]. Also, the global distribution of therapeutic agent is ideally suited for delivery to regenerating tissue in each channel, even in regions that would be inaccessible by diffusion from the perimeter alone. Rapid solvent evaporation, used in this case for the formation of voids between the parallel channels, was
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
shown to be an effective method for fabrication of scaffolds with a high void fraction. Effectual, though limited, control over the void structure was demonstrated by the simple variation of polymer concentration, as shown in Fig. 3. Void fraction was inversely proportional to the mass of polymer used for a given solvent volume, i.e. the polymer concentration. High void fraction may be desirable as it allows the scaffold to occupy a large space with a relatively low mass of synthetic material, but a very high void fraction may result in poor mechanical properties and a limited polymer mass from which to deliver biological molecules. Sustained drug delivery and scaffold permeability both depend on void interconnectivity for access to a diffusible medium. Void interconnectivity, as measured by the accessible void fraction (Fig. 4), increased with the void fraction. Consequently, the factors that will determine the optimum void fraction, are the mechanical properties of the scaffold, mass of scaffold required for sustained delivery, and void interconnectivity. Scaffold geometry, as revealed by mCT (Fig. 6), remained quite static for at least 8 weeks, and then changed drastically as the mass loss rate increased. It is difficult to decipher what impact this may have on axon regeneration in vivo. The images shown in this study were acquired after degrading samples were dried, so some of the architectural transformation may have been due to the drying process itself. Previous work by other investigators has shown that axon regeneration into biodegradable conduits can be observed as early as 2 weeks [38]. In the present set of experiments, axon regeneration similar to that shown in Fig. 8 was present throughout the length of the scaffold at 4 weeks after implantation, which is well before substantial changes in scaffold geometry were observed in this study. It is possible that once tissue has traversed the length of a channel, subsequent alterations in structure due to degradation may not impede further growth. Release of the surrogate drug progressed in concert with scaffold degradation, as shown in Fig. 5. A rather high variability was observed, indicated by standard error bars in the figure, and this is probably due to the inhomogeneity observed in the distribution of FITC-D within scaffolds. The optimal release profile will depend on the particular drug being employed, but in this example, sustained release was accomplished for a period of about 12 weeks. Of course, maintenance of a released molecule’s biologic activity will be necessary in order to elicit a desired response. This will need to be assessed independently for each biomolecule considered as a candidate for controlled release in this system. The scaffold used in this approach is not designed to guide regenerating fibers by providing an oriented surface on which to grow, as proposed by other investigators [17,33,37,39], but rather as a template for the arrangement of cells that may attract and guide
427
regenerating axons. Drugs may be incorporated to provide added tropic attraction, for neuroprotection, or to reduce the inhibitory effects of secondary injury. However, the basis for promotion of axon regeneration is envisioned to be in transplanted cells. Therefore, successful Schwann cell harvesting, suspension, and distribution into scaffold channels is of critical importance in this model. Regeneration may be affected by a multitude of variables, including cell culture purity, cell density, distribution within the matrix (i.e. Matrigel), and distribution within the channels. Images in Fig. 7 indicate that Schwann cell harvest, suspension in Matrigel, and loading into distinct channels is feasible, and that cells survive in culture under physiologic conditions within the scaffold. The ability of the cells to promote axon regeneration will depend further on their survival throughout the transplantation process of the cell-seeded scaffold into the injured spinal cord. Axon regeneration in this study was visualized by 3-D reconstruction of serial histological sections, an approach that appears useful for evaluation of regenerating tissue architecture. Histological sections alone reveal much about internal regeneration, but deciphering of 3D characteristics is often difficult. Reconstructed data may assist in determining factors that will affect the 3-D arrangement of regenerating tissue, and may provide clues regarding functional changes. The experiments outlined in this study suggest a model for simultaneous, in vivo testing of a multitude of factors, and combinations thereof, that may influence axon regeneration in the transected rat spinal cord. Aspects of this model may be systematically manipulated to study effects of the structural, cellular, and molecular microenvironments surrounding the injury site. As novel factors promoting axon regeneration are identified, variations and combinations may be investigated to assess interactions among those factors. Of course, if successful axon regeneration is accompanied by functional recovery, the model would immediately suggest a possible treatment strategy.
5. Conclusions The data shown in the present work indicate that a simple injection molding/solvent evaporation technique may be used to produce biodegradable scaffolds with multiple-channel geometry. Scaffolds fabricated via these techniques possessed a relatively high void fraction, which could be modulated by changing the initial polymer concentration. Void space interconnectivity was found to increase as void fraction increased. Scaffolds degraded predictably in vitro, during which time sustained drug release occurred. Distinct channels of the scaffold could be seeded individually with Schwann cells, which survived for at least 48 h in vitro.
ARTICLE IN PRESS 428
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
Schwann-cell containing scaffolds promoted axon regeneration in vivo, and regenerating tissue architecture could be described with 3-D reconstructions of histological sections. These scaffolds may serve as the basis for an in vivo model used to determine the effects of the structural, cellular, and molecular environments on regeneration of spinal cord axons.
Acknowledgements Funding support was provided in part by the Mayo Foundation and the National Institutes of Health through grants AR45871 to Dr. Yaszemski’s laboratory, EB02390 to Dr. Windebank’s laboratory, and predoctoral fellowship NS45544 to the first author. The authors would also like to thank Mr. James Gruetzmacher for GPC analysis, Ms. LouAnn Gross for microscopy sample processing, Mr. Fred Schultz for assistance with mold design and manufacturing, Mr. Adam Ellison for assistance with CAD, and the staff of the Mayo Biomedical Imaging Resource for assistance with image processing. References [1] Richardson PM, McGuinness UM, Aguayo AJ. Axons from CNS neurons regenerate into PNS grafts. Nature 1980;284:264–5. [2] Kuhlengel KR, Bunge MB, Bunge RP, Burton H. Implantation of cultured sensory neurons and Schwann cells into lesioned neonatal rat spinal cord. II. Implant characteristics and examination of corticospinal tract growth. J Comp Neurol 1990;293: 74–91. [3] Li Y, Raisman G. Schwann cells induce sprouting in motor and sensory axons in the adult rat spinal cord. J Neurosci 1994; 14:4050–63. [4] Li Y, Field PM, Raisman G. Repair of adult rat corticospinal tract by transplants of olfactory ensheathing cells. Science 1997; 277:2000–2. [5] Ramon-Cueto A, Plant GW, Avila J, Bunge MB. Long-distance axonal regeneration in the transected adult rat spinal cord is promoted by olfactory ensheathing glia transplants. J Neurosci 1998;18:3803–15. [6] McDonald JW, Liu XZ, Qu Y, Liu S, Mickey SK, Turetsky D, Gottlieb DI, Choi DW. Transplanted embryonic stem cells survive, differentiate and promote recovery in injured rat spinal cord. Nat Med 1999;5:1410–2. [7] Ogawa Y, Sawamoto K, Miyata T, Miyao S, Watanabe M, Nakamura M, Bregman BS, Koike M, Uchiyama Y, Toyama Y, Okano H. Transplantation of in vitro-expanded fetal neural progenitor cells results in neurogenesis and functional recovery after spinal cord contusion injury in adult rats. J Neurosci Res 2002;69:925–33. [8] Fernandez E, Pallini R, Lauretti L, Mercanti D, Serra A, Calissano P. Spinal cord transection in adult rats: effects of local infusion of nerve growth factor on the corticospinal tract axons. Neurosurgery 1993;33:889–93. [9] Schnell L, Schneider R, Kolbeck R, Barde YA, Schwab ME. Neurotrophin-3 enhances sprouting of corticospinal tract during development and after adult spinal cord lesion. Nature 1994;367: 170–3.
[10] Namiki J, Kojima A, Tator CH. Effect of brain-derived neurotrophic factor, nerve growth factor, and neurotrophin-3 on functional recovery and regeneration after spinal cord injury in adult rats. J Neurotrauma 2000;17:1219–31. [11] Zuo J, Neubauer D, Dyess K, Ferguson TA, Muir D. Degradation of chondroitin sulfate proteoglycan enhances the neuritepromoting potential of spinal cord tissue. Exp Neurol 1998;154: 654–62. [12] Yick LW, Wu W, So KF, Yip HK, Shum DK. Chondroitinase ABC promotes axonal regeneration of Clarke’s neurons after spinal cord injury. Neuroreport 2000;11:1063–7. [13] Yick LW, Cheung PT, So KF, Wu W. Axonal regeneration of Clarke’s neurons beyond the spinal cord injury scar after treatment with chondroitinase ABC. Exp Neurol 2003;182: 160–8. [14] Lemons ML, Howland DR, Anderson DK. Chondroitin sulfate proteoglycan immunoreactivity increases following spinal cord injury and transplantation. Exp Neurol 1999;160:51–65. [15] Bradbury EJ, Moon LD, Popat RJ, King VR, Bennett GS, Patel PN, Fawcett JW, McMahon SB. Chondroitinase ABC promotes functional recovery after spinal cord injury. Nature 2002;416: 636–40. [16] Friedman JA, Windebank AJ, Moore MJ, Spinner RJ, Currier BL, Yaszemski MJ. Biodegradable polymer grafts for surgical repair of the injured spinal cord. Neurosurgery 2002;51:742–51 discussion 51-2. [17] Teng YD, Lavik EB, Qu X, Park KI, Ourednik J, Zurakowski D, Langer R, Snyder EY. Functional recovery following traumatic spinal cord injury mediated by a unique polymer scaffold seeded with neural stem cells. Proc Natl Acad Sci USA 2002;99:3024–9. [18] Xu XM, Zhang SX, Li H, Aebischer P, Bunge MB. Regrowth of axons into the distal spinal cord through a Schwann-cell-seeded mini-channel implanted into hemisected adult rat spinal cord. Eur J Neurosci 1999;11:1723–40. [19] Oudega M, Gautier SE, Chapon P, Fragoso M, Bates ML, Parel JM, Bunge MB. Axonal regeneration into Schwann cell grafts within resorbable poly(alpha-hydroxyacid) guidance channels in the adult rat spinal cord. Biomaterials 2001;22:1125–36. [20] Schmidt CE, Leach JB. Neural tissue engineering: strategies for repair and regeneration. Annu Rev Biomed Eng 2003;5:293–347. [21] Profyris C, Cheema SS, Zang D, Azari MF, Boyle K, Petratos S. Degenerative and regenerative mechanisms governing spinal cord injury. Neurobiol Dis 2004;15:415–36. [22] Silver J, Miller JH. Regeneration beyond the glial scar. Nat Rev Neurosci 2004;5:146–56. [23] Bamber NI, Li H, Lu X, Oudega M, Aebischer P, Xu XM. Neurotrophins BDNF and NT-3 promote axonal re-entry into the distal host spinal cord through Schwann cell-seeded minichannels. Eur J Neurosci 2001;13:257–68. [24] Tropea D, Caleo M, Maffei L. Synergistic effects of brain-derived neurotrophic factor and chondroitinase ABC on retinal fiber sprouting after denervation of the superior colliculus in adult rats. J Neurosci 2003;23:7034–44. [25] Chau CH, Shum DK, Li H, Pei J, Lui YY, Wirthlin L, Chan YS, Xu XM. Chondroitinase ABC enhances axonal regrowth through Schwann cell-seeded guidance channels after spinal cord injury. Faseb J 2004;18:194–6. [26] Jorgensen SM, Demirkaya O, Ritman EL. Three-dimensional imaging of vasculature and parenchyma in intact rodent organs with X-ray micro-CT. Am J Physiol 1998;275:H1103–14. [27] Robb RA, Hanson DP, Karwoski RA, Larson AG, Workman EL, Stacy MC. Analyze: a comprehensive, operator-interactive software package for multidimensional medical image display and analysis. Comput Med Imaging Graph 1989;13:433–54. [28] Moore MJ, Jabbari E, Ritman EL, Lu L, Currier BL, Windebank AJ, Yaszemski MJ. Quantitative analysis of interconnectivity of
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
[29]
[30]
[31]
[32]
[33]
porous biodegradable scaffolds with micro-CT. J Biomed Mater Res A 2004;71:258–67. Brockes JP, Fields KL, Raff MC. Studies on cultured rat Schwann cells. I. Establishment of purified populations from cultures of peripheral nerve. Brain Res 1979;165:105–18. Guenard V, Kleitman N, Morrissey TK, Bunge RP, Aebischer P. Syngeneic Schwann cells derived from adult nerves seeded in semipermeable guidance channels enhance peripheral nerve regeneration. J Neurosci 1992;12:3310–20. Xu XM, Guenard V, Kleitman N, Bunge MB. Axonal regeneration into Schwann cell-seeded guidance channels grafted into transected adult rat spinal cord. J Comp Neurol 1995;351: 145–60. Mantila SM, Camp JJ, Krych AJ, Robb RA. Modeling the functional repair of nervous tissue in spinal cord injury. In: Galloway Jr. RL (Ed.), Medical imaging 2004, visualization, image-guided procedures, and display 5367, San Diego, CA, 2004. p. 491–9. Patist CM, Mulder MB, Gautier SE, Maquet V, Jerome R, Oudega M. Freeze-dried poly(D,L-lactic acid) macroporous guidance scaffolds impregnated with brain-derived neurotrophic factor in the transected adult rat thoracic spinal cord. Biomaterials 2004;25:1569–82.
429
[34] Stokols S, Tuszynski MH. The fabrication and characterization of linearly oriented nerve guidance scaffolds for spinal cord injury. Biomaterials 2004;25:5839–46. [35] Sundback C, Hadlock T, Cheney M, Vacanti J. Manufacture of porous polymer nerve conduits by a novel low-pressure injection molding process. Biomaterials 2003;24:819–30. [36] Bender MD, Bennett JM, Waddell RL, Doctor JS, Marra KG. Multi-channeled biodegradable polymer/CultiSpher composite nerve guides. Biomaterials 2004;25:1269–78. [37] Blacher S, Maquet V, Schils F, Martin D, Schoenen J, Moonen G, Jerome R, Pirard JP. Image analysis of the axonal ingrowth into poly(D,L-lactide) porous scaffolds in relation to the 3-D porous structure. Biomaterials 2003;24:1033–40. [38] Gautier SE, Oudega M, Fragoso M, Chapon P, Plant GW, Bunge MB, Parel JM. Poly(alpha-hydroxyacids) for application in the spinal cord: resorbability and biocompatibility with adult rat Schwann cells and spinal cord. J Biomed Mater Res 1998;42: 642–54. [39] Maquet V, Martin D, Scholtes F, Franzen R, Schoenen J, Moonen G, Jerome R. Poly(D,L-lactide) foams modified by poly(ethylene oxide)-block-poly(D,L-lactide) copolymers and aFGF: in vitro and in vivo evaluation for spinal cord regeneration. Biomaterials 2001;22:1137–46.
Biomaterials 27 (2006) 419–429 www.elsevier.com/locate/biomaterials
Multiple-channel scaffolds to promote spinal cord axon regeneration Michael J. Moorea, Jonathan A. Friedmanb, Eric B. Lewellync, Sara M. Mantilaa, Aaron J. Krychd, Syed Ameenuddinc, Andrew M. Knightc, Lichun Lua,e, Bradford L. Currierd,e, Robert J. Spinnerd,f, Richard W. Marshd,f, Anthony J. Windebankc,d, Michael J. Yaszemskia,d,e, a
Department of Physiology and Biomedical Engineering, Mayo Clinic College of Medicine, Rochester, MN 55905, USA b Section of Neurosurgery, Dartmouth Hitchcock Medical Center, Lebanon, NH 03756, USA c Department of Neurology, Mayo Clinic College of Medicine, Rochester, MN 55905, USA d Mayo Medical School, Mayo Clinic College of Medicine, Rochester, MN 55905, USA e Department of Orthopedic Surgery, Mayo Clinic, MS 3-75, 200 First Street, SW, Rochester, MN 55905, USA f Department of Neurologic Surgery, Mayo Clinic College of Medicine, 200 First Street, SW, Rochester, MN 55905, USA Received 30 June 2005; accepted 27 July 2005 Available online 31 August 2005
Abstract As molecular, cellular, and tissue-level treatments for spinal cord injury are discovered, it is likely that combinations of such treatments will be necessary to elicit functional recovery in animal models or patients. We describe multiple-channel, biodegradable scaffolds that serve as the basis for a model to investigate simultaneously the effects on axon regeneration of scaffold architecture, transplanted cells, and locally delivered molecular agents. Poly(lactic-co-glycolic acid) (PLGA) with copolymer ratio 85:15 was used for these initial experiments. Injection molding with rapid solvent evaporation resulted in scaffolds with a plurality of distinct channels running parallel along the length of the scaffolds. The feasibility of creating scaffolds with various channel sizes and geometries was demonstrated. Walls separating open channels were found to possess void fractions as high as 89%, with accessible void fractions as high as 90% through connections 220 mm or larger. Scaffolds degraded in vitro over a period of 30 weeks, over which time-sustained delivery of a surrogate drug was observed for 12 weeks. Primary neonatal Schwann cells were distributed in the channels of the scaffold and remained viable in tissue culture for at least 48 h. Schwann-cell containing scaffolds implanted into transected adult rat spinal cords contained regenerating axons at one month post-operation. Axon regeneration was demonstrated by three-dimensional reconstruction of serial histological sections. r 2005 Elsevier Ltd. All rights reserved. Keywords: Nerve tissue engineering; Scaffold; Microstructure; Schwann cell; Drug release; Image analysis
1. Introduction Advances in neuroscience over the past 2 decades begin to offer hope for spinal cord injury (SCI) victims. Since the demonstration in 1980 that central nervous system (CNS) axons have the capacity to regenerate Corresponding author. Department of Orthopedic Surgery, Mayo Clinic, MS 3-75, 200 First Street SW, Rochester, MN 55905, USA. Tel.: +1 507 284 2267; fax: +1 507 284-5075. E-mail address: [email protected] (M.J. Yaszemski).
0142-9612/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2005.07.045
within peripheral nervous system (PNS) grafts [1], much has been accomplished toward understanding factors that contribute to a physiologically permissive environment. Mechanisms of injury, of regeneration, and of inhibition to regeneration are being delineated, and several promising treatment strategies have arisen. Transplantation of a variety of cell types, including Schwann cells [2,3], olfactory ensheathing glial cells [4,5], or neural stem cells [6,7], has resulted in axon regeneration and limited functional improvement after spinal cord injury in rats. Molecular therapies that work
ARTICLE IN PRESS 420
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
to promote regeneration, such as administration of neurotrophins [8–10], and those that target deleterious inhibition of regeneration, such as chondroitinase ABC [11–15], have also yielded favorable results. Synthetic biomaterials have been investigated for their ability to reconstruct spinal cord tissue architecture, to provide guidance for regenerating axons, and to prevent the infiltration of scar tissue [16–19]. A comprehensive review of neural regeneration strategies was recently provided by Schmidt and Baier Leach [20]. Despite recent advances, the limited demonstration of functional improvement in animal models has prevented advancement of any regenerative therapy to clinical use. This may be due in large part to the multifaceted nature of spinal cord injuries, which presents a major challenge to therapeutic development. Primary mechanical trauma to the cord induces secondary injury consisting of a complex cascade of molecular events that lead to the loss of myelin and the formation of a glial scar [21,22]. Therefore, in order for viable treatment strategies to be realized clinically, it is likely that combinations of current therapeutic approaches must be used. Indeed, combinatorial approaches have already shown promise in animal models. For example, administration of neurotrophins enhanced axon growth into Schwann-cell seeded guidance channels and increased integration into the graft–host interface [23]. Synergistic effects on CNS axon regeneration have been demonstrated when strategies promoting regeneration and antagonizing inhibition were used simultaneously [24,25]. An elegant solution may lie in the design of a bioartificial graft that targets injury mechanisms at the molecular, cellular, and tissue levels. Biodegradable polymers can simultaneously provide a tissue scaffold, a cell delivery vehicle, and a reservoir for sustained drug delivery [16]. This integrative approach suggests a possible treatment strategy and may serve as an in vivo model for studying optimization of various combinations of treatments. We describe techniques for producing biodegradable polymer scaffolds with parallel-channel architecture that can be systematically modified. They may be seeded with multiple cell types arranged spatially in anatomically relevant locations and may serve as a vehicle for sustained drug delivery. We quantitatively describe the scaffolds’ architecture, their in vitro degradation profile, their drug delivery characteristics, their biocompatibility with Schwann cells in culture, and their promotion of in vivo axon regeneration.
evaporation technique. Cylindrical, Teflon molds, with diameter 3.0 mm were fitted with Delrin spacers containing an array of seven, uniformly spaced, 508- or 660-mm stainless-steel wires (Malin, Cleveland, OH), as shown in Fig. 1A. The wire arrays were spray-coated with a minimal amount of Ease Release 200 (Mann Formulated Products, Easton, PA) mold lubricant to facilitate removal. A concentrated solution of poly(D,L-lactic-co-glycolic acid) (PLGA, Alkermes, Cambridge, MA), with copolymer ratio 85:15 lactide:glycolide and number average molecular weight (Mn) 75,000, was made by adding 1.0 g PLGA to 2.0 ml dichloromethane (DCM) in a glass vial and shaking vigorously for 3 h. The viscous PLGA solution was injected with a syringe through a 16-gauge hypodermic needle from the bottom of each mold until solution was seen escaping through the top. Polymer-filled molds were vacuum-dried for at least 24 h with a high-vacuum pump (VP 190, Savant Instruments, Holbrook, NY) connected to a condensation trap (RT 4104, Savant Instruments). Vacuum drying removed the solvent and created pores in the walls of the scaffolds. Scaffolds to be used in cell culture or in vivo experiments were washed in ethanol for 30 min with gentle shaking both to disinfect the scaffolds and to remove any residual mold lubricant. Scaffolds were again vacuum-dried for 24 h to remove the ethanol, then sealed in sterilized glass vials and stored desiccated at 4 1C until further use. Creation of scaffolds with more complex, biomimetic architecture was demonstrated by fabrication of molds using computer-aided design (CAD) and solid freeform fabrication (SFF). A PatternMaster three-dimensional (3-D), piezo-inkjet printer (Solidscape, Merrimack, NH) was used to create wax molds from models designed manually with ProEngineer software (Parametric Technology Corp., Needham, MA). Fig. 1B shows one wax mold manufactured by the PatternMaster. Injection of PLGA polymer solution followed by vacuum-evaporation of solvent resulted in PLGA scaffolds 1 cm in diameter. These were used to test the principle that complex architecture mimicking the structure of the human spinal cord can be fabricated.
2. Materials and methods
Fig. 1. Injection molds for fabrication of multiple-channel scaffolds. (A) Cut-away view of parallel-wire injection molding apparatus. Mold inner diameter ¼ 3.0 mm. Polymer solution was injected with a syringe through an injection canal, as shown in the model. (B) Photograph of wax mold fabricated with PatternMaster for creation of scaffold by injection of PLGA. Parallel cross bars are for support of inner mold parts. Mold inner diameter ¼ 1 cm.
2.1. Scaffold fabrication Biodegradable scaffolds with controlled, parallel-channel architecture were fabricated by an injection molding, solvent
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
421
2.2. Morphological characterization
diameter ¼ 0.50 mm:
Scaffold architecture was viewed with X-ray microcomputed tomography (micro-CT or mCT). A custom-built mCT scanner [26], housed in the Mayo Physiological Imaging Research Laboratory, provided 3-D interior images which were viewed and manipulated with Analyze AVW (Mayo Foundation, Rochester, MN) [27], a comprehensive image display and analysis software package developed by the Mayo Biomedical Imaging Resource. Void fraction and accessible void fraction were measured from mCT images by a method described in detail elsewhere [28]. In summary, this program uses morphological operations to estimate the percentage of the void volume of a scaffold that maintains connections to the outside air through openings of a given minimum size.
ph 2 (2) ðD nd 2 Þ ¼ 85:4 mm3 . 4 If porosity is taken as the overall porosity, including that contributed by the pores and the channels, then V in Eq. (1) is merely the volume of the 3.0-mm-diameter and 15.0-mm-long cylindrical mold, i.e. 106 mm3. Scaffolds were made as in the above section, except that the amount of PLGA added was varied to create solutions of increasing polymer concentration. Solutions of 30%, 40%, 50%, 60%, and 70% (w/v) PLGA were made by dissolving in DCM. The resulting scaffolds were weighed on a Denver Instruments (Arvada, CO) M-220D electronic balance, and their void fraction estimated by the calculations in Eqs. (1) and (2).
2.3. Modulation of porosity
2.4. Scaffold degradation
For the purpose of clarity, scaffold channels are defined as the voids left by the stainless-steel wires, walls as the porous, polymer-occupying regions between the channels, and pores as the voids within the walls of the scaffold left by evaporated solvent (see Fig. 2). Wall porosity was estimated from mold dimensions, scaffold weights, polymer density, and Eq. (1) below, where V ¼ mold void volume, r ¼ density of PLGA ¼ 1.2 mg/mm3, and m ¼ measured mass (weight) of scaffold: m 100%. (1) wall porosity ð%Þ ¼ 1 Vr
Individual scaffolds were heat-sealed in pouches made from 10-mm, nylon mesh sheets (Spectrum Labs, Rancho Dominguez, CA) and vacuum-dried for 24 h before obtaining tare weights. Dried samples were then immersed in 20 ml phosphate-buffered saline (PBS) at pH 7.4 and incubated at 37 1C. All solutions were replaced, and their pH monitored, weekly. At regular intervals, for a total of 30 weeks, three samples were removed, washed repeatedly in distilled, deionized water, and vacuum-dried for 24 h. Dried pouches were immersed in 1 ml tetrahydrofuran (THF) and shaken overnight to dissolve the scaffolds. Empty pouches were then removed, washed repeatedly in DCM, vacuum-dried overnight, and weighed. Degradation was characterized by mass loss, molecular weight loss, and morphological change. Mass loss was determined by weighing dried samples at a given time point (mt) and comparing to the initial mass (mo): mt 100%. (3) mass loss ð%Þ ¼ 1 mo
The mold void volume V is given by Eq. (2), where h is the mold height ¼ 15.0 mm, D is the mold diameter ¼ 3.0 mm, n is the number of wires ¼ 7, and d is the wire
V¼
Polymer molecular weight loss was characterized by gel permeation chromatography (GPC) with a Waters 717 Plus Autosampler GPC system (Waters, Milford, MA) connected to a model 515 HPLC pump and model 2410 refractive index detector. The columns consisted of a Styragel HT guard column (7.8 300 mm, Waters) in series with a Styragel HR 4E column (7.8 300 mm, Waters). Degraded and undegraded scaffolds, for comparison, were measured by injecting 20 ml of each sample and eluting at 1 ml/min flow rate in THF. Monodisperse polystyrene standards (Polysciences, Warrington, PA) with Mn of 0.474, 6.69, 18.6, and 38 kD and polydispersities of less than 1.1 were used to construct the calibration curve. Morphological changes were characterized by monitoring degrading samples using mCT. 2.5. Release of model drug
Fig. 2. PLGA scaffolds created from parallel wire molds. Top panels: photographs, and bottom panels: mCT slices. Left: cross-sectional view, and right: longitudinal view. Each scale bar ¼ 500 mm.
Fluorescein isothyocyanate-dextran (FITC-D, average Mw 167 kD, Sigma, St. Louis, MO) served as a model drug for estimating release kinetics of proteins in a similar molecular weight range. After dissolution of PLGA in DCM, 50 mg FITC-D was added to the solution and stirred vigorously before vortexing for an additional 3 h. The polymer/solvent/
ARTICLE IN PRESS 422
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
drug mixture was then injected into molds and vacuum-dried for 24 h. Dried scaffolds were weighed, heat-sealed in pouches made from Spectra/Mesh nylon filters (10 mm mesh), then placed in 1.0 ml PBS at pH 7.4 and 37 1C. At each time point, the nylon pouches were removed and placed in 1.0 ml of fresh PBS, and the absorbance at 490 nm of the conditioned supernatant was measured on a spectrophotometer (SPECTRAmax PLUS 384, Molecular Devices, Sunnyvale, CA) and compared with a calibration curve of known FITC-D concentrations. Drug loading was estimated for each scaffold by calculating the mass ratio of FITC-D to the PLGA used, then multiplying this ratio by the mass of each scaffold. The released FITC-D was calculated as the cumulative percentage of the initial loading for each individual scaffold.
2.6. Schwann cell culture and evaluation All experiments involving animals were approved by the Institutional Animal Care and Use Committee (IACUC). Male, transgenic Sprague-Dawley rats expressing green fluorescent protein (GFP) (Genome Information Research Center, Osaka, Japan) were mated with female SpragueDawley rats (Charles River Laboratories, Wilmington, MA). GFP-positive rat pups were identified with an ultraviolet lamp. Primary neonatal Schwann cells were obtained by methods similar to those first described by Brockes et al. [29]. Two- to five-day-old rat pups, both GFP-positive and negative, were anesthetized by intraperitoneal injection of 0.05–0.1 ml sodium pentobarbital. Sciatic nerves from both legs were isolated, sectioned into small pieces (1 mm3), and digested enzymatically for 45 min with 0.25% trypsin and 0.03% collagenase in Hank’s balanced salts solution. After digestion, cells were pelleted at 800 rpm in a Beckman TJ-6 centrifuge for 5 min, the supernatant was removed, and cells were resuspended in 5 ml of DMEM/F12 containing 10% fetal bovine serum (FBS) and 100 units/ml penicillin/streptomycin. Cells were dissociated by repeatedly passing them through a narrowed Pasteur pipette, then added to a laminin-coated 35-mm dish, and incubated at 37 1C and 5% CO2 for 3 days. Schwann cells were identified by characteristic elongated spindle body shape, a prominent nucleus, and bi- or tripolar extensions. These findings were confirmed by immunostaining in GFP-negative animals. Culture dishes were rinsed with PBS, fixed with 1:1 methanol/acetate solution, labeled with anti-S-100 polyclonal antibody (NeoMarkers, Fremont, CA), biotinylated anti-rabbit secondary antibody (Dako, Carpinteria, CA), streptavidin HRP (Dako), and stained with diaminobenzene (DAB) (Sigma). Cells were detached with trypsin-EDTA for 2 min, resuspended in chilled Matrigel (BD Biosciences, Bedford, MA) at a target concentration of approximately 1.2 108 cells/ml [30,31], and then added to each channel of the scaffold with a gel-loading pipette tip. The volume of the cell suspension added was on the order of 1 ml/channel. Cell-loaded scaffolds were cut to 2 mm in length and incubated in DMEM/F12 medium with 10% FBS for 48 h prior to microscopic evaluation in vitro, or 24 h prior to implantation in vivo. Scaffolds loaded with cells suspended in Matrigel were fixed in 4% paraformaldehyde and cut into 0.2 mm sections on a Reichert-Jung (Leica Microsystems, Germany) Biocut
microtome. Sections were viewed with fluorescence optical microscopy (Zeiss Axiovert 35, Germany), and also processed for transmission electron microscopy (JEOL 1200EX, Peabody, MA). 2.7. Animal surgery and post-operative care Female Sprague-Dawley rats (Charles River Laboratories) weighing 300–350 g were deeply anesthetized with intraperitoneal injection of Ketamine 80 mg/kg (Fort Dodge Animal Health, Fort Dodge, IA) and Xylazine 5 mg/kg (Lloyd Laboratories, Shenandoah, IA). Animals were placed on a heating pad constantly maintained at 37 1C during surgery, and ophthalmic ointment was used to prevent drying of the eyes. A 4-cm incision was made along the dorsal midline. The paraspinal muscles were dissected subperiosteally from the spinous processes and laminae. Multiple laminectomies were performed with sharp scissors at levels T8–T10 while ensuring that the facet joints were not violated. The posterior aspect of the spinal cord was exposed and the dura was cut vertically using microforceps and microscissors. An angled microscissors was used to transect fully the spinal cord, and a 1.0-mm segment of spinal cord at level T9 was removed. The stumps were retracted making a 2.0-mm gap in the spinal cord. Complete transection was confirmed by irrigating the site with lactated Ringer’s solution until the bottom of the canal was visible. Biodegradable implants containing either Matrigel alone or Schwann cells suspended in Matrigel were inserted into the canal, making sure that the severed ends of the cord fit tightly. Control animals received no implants. Exposed spinal cords with or without scaffolds were covered with muscles and fascia. Before closing the wound, the spine was stabilized by suturing two 2-mm 2-cm, non-biodegradable polypropylene shards along the spinous processes of T4–T10. A Zeiss (Germany) S3B microsurgical microscope was used during the entire surgical procedure. Animals were injected with 5.0 ml lactated Ringer’s solution subcutaneously after surgery and allowed to recuperate. Buprenex 0.05 mg/kg was given subcutaneously for 3–4 days to minimize pain. Baytril 65 mg/kg was given intramuscularly for 1 week and as needed thereafter to prevent infection. All rats were kept in low-sided cages to enable them easy access to food and water, and bedding was changed frequently to prevent decubitus ulcers. Daily post-operative care was undertaken by maintaining and monitoring body temperature, general fluid and electrolyte balance, and cardiovascular homeostasis by direct observation. Bladders were manually expressed twice daily for the duration of the experiment. 2.8. Tissue preparation Animals were sacrificed at the end of 1 month. Animals were deeply anesthetized with 0.3 ml sodium pentobarbital and transcardially perfused with 100–120 ml chilled PBS followed by 120–150 ml 4% paraformaldehyde. After perfusion, the spinal column was removed and post-fixed for 24 h in the same fixative. The vertebrae were carefully cut away, and isolated spinal cords with or without scaffolds were soaked for 48 h in 4% paraformaldehyde and stored in PBS at 4 1C until they
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
423
were processed for histology and immunohistochemistry. Spinal cords with scaffolds were dehydrated in a mixture of ethanol and Citrisolv (Fisher Scientific, Pittsburg, PA), to prevent dissolution of scaffold, and embedded in paraffin. Tissue blocks were cut into 8 mm thick sections (Leica Microsystems) and stained with mouse monoclonal antineurofilament antibody against phosphorylated neurofilament-H (clone 2F11, Dako). 2.9. 3-D image reconstruction Digital photomicrographs were obtained at 2.5 and 20 using a Zeiss Axioplan II microscope equipped with an AxioCam digital camera. All registration, segmentation, and rendering operations were performed with Analyze AVW software (Mayo Foundation) [27] using methods discussed in greater detail elsewhere [32]. Serial histological images were registered automatically with an automatic registration algorithm. Individual slices were then registered manually to correct for errors in alignment. Regions corresponding to channels or axon bundles were segmented from surrounding scaffold and tissue regions using automatic multispectral analysis with a mode filter. Image segmentation was further altered manually with the use of image editing and morphological operations. Sixteen consecutive slices were reconstructed and rendered assuming isotropic voxels, and reconstructed images were displayed orthographically with wedging to reveal internal channel or regenerating axon geometry.
Fig. 3. Modulation of void fraction. Upper panel contains inverted, summed-voxel projections representative of scaffolds made with varying polymer concentrations. These concentrations correspond to those plotted in the lower panel, where void fractions measured gravimetrically are plotted against polymer concentrations. Error bars indicate7standard deviation; n ¼ 5. Note broken ordinate axis.
3. Results 3.1. Scaffold morphology Scaffolds fabricated with the injection molding technique possessed macro-architecture consistent with the mold design, as shown by the photographs in Fig. 2. The scaffolds’ geometrical and dimensional properties were reproducible and were maintained with a fair degree of manual manipulation, such as what might be expected during neurosurgical implantation. Micro-CT analysis revealed the seven large channels that remained intact throughout the length of each scaffold, with highly porous walls separating the channels, as shown in Fig. 2. Pores within the walls, formed by the evaporating solvent, exhibited a wide range of sizes, with apparent longitudinal isotropy and radial anisotropy. Channel diameters were measured with image analysis and found to be slightly less than that of the wires that formed them. Wires which were 510 mm in diameter created channels with diameters of 45079 mm, and those which were 660 mm created channels 600710 mm in diameter. 3.2. Modulation of void fraction Fabrication of scaffolds with differential concentrations of PLGA solutions yielded scaffolds with varying
void fractions. Qualitative differences in void fraction were apparent from mCT images, as indicated by the inverted, summed-voxel projections in Fig. 3. The projections shown are similar to conventional X-rays, but in these negative images, PLGA appears dark. Gravimetric void fraction measurement confirmed quantitatively a nearly linear trend in void fraction as a function of polymer concentration. The void fraction of the walls displays a similar linear trend as the overall void fraction, though as expected, the wall void fraction is less than the overall, due to the large volume occupied by the open channels. Analysis of mCT images confirmed changes in pore interconnectivity as a function of the overall void fraction. As a measure of pore interconnectivity, the accessible void fraction was estimated for a range of minimum connection sizes for three of the groups with modulated porosity. This quantity is an estimate of the fraction of void volume that is connected to either the channels or the ends of the scaffold, through connections of the minimum size indicated. The groups chosen possessed a high overall void fraction (88.6%, measured gravimetrically), a medium void fraction (83.4%), and a low void fraction (78.7%). The data in Fig. 4 show that as the void fraction increases, the accessible void fraction for a given minimum connection size also
ARTICLE IN PRESS 424
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
increases, i.e. a higher porosity yields a higher interconnectivity. 3.3. Scaffold degradation Scaffolds in simulated physiologic conditions were found to degrade as expected for PLGA with the given copolymer ratio and molecular weight. As shown by the degradation profile in Fig. 5, scaffold molecular weight decreased steadily from 100% of the initial molecular weight down to 5% by 26 weeks of degradation. In
contrast, little or no mass loss was observed for the first 20 weeks, followed by a period of more precipitous mass loss over the next 10 weeks. The observed sigmoidal mass loss behavior is typical for PLGA degradation. Only slight changes in morphology were apparent during the period when negligible change in mass was observed. As mass loss occurred, however, the morphology of the degrading scaffolds changed drastically. Fig. 6 shows transverse and longitudinal cross-sections of a scaffold after 8 weeks of degradation and another scaffold after 24 weeks. The scaffold degraded for 8 weeks appears to possess slight distortions of external geometry and channel geometry, as compared to undegraded specimens (Fig. 2). The scaffold degraded for 24 weeks, on the other hand, shows evidence of pronounced deformations in geometry and dimension, as seen in the lower panel of Fig. 6. Marked reduction in void fraction is apparent qualitatively, and the seven channels no longer appear to run contiguously throughout the length of the scaffold. 3.4. Release of model drug
Fig. 4. Accessible void fractions for three selected scaffold groups. Accessible void volume, as a percent of the overall void volume, is given for scaffold groups with high (88.6%), medium (83.4%), and low (78.7%) overall void fractions, measured gravimetrically, over a range of minimum connection sizes. Error bars indicate7standard deviation; n ¼ 4. Note broken ordinate axis.
Upon visual inspection, mechanical mixing of FITCD within the polymer scaffolds was an effective method for scaffold loading, though dispersal appeared to be inhomogeneous. Fig. 5 shows the FITC-D release profile compared with the mass and molecular weight degradation profiles. Release of the model drug into PBS commenced with an initial burst release over the first 48 h, followed by a period of more steady release for 4 weeks. Sustained release continued, though even more slowly, for an additional 8 weeks, after which little or no
Fig. 5. Degradation and FITC-D release profile. Mass or molecular weight change is indicated as a percentage of initial value at time zero; error bars indicate7standard deviation, n ¼ 3. FITC-D release is indicated as a cumulative percentage of loaded drug; means of two simultaneous experiments, each with n ¼ 5, are plotted with error bars indicating standard error of the mean.
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
425
release was detected for the remainder of the degradation period. 3.5. Schwann cell culture Harvesting and seeding of Schwann cells produced viable primary cultures that were approximately 90% pure, the remainder being fibroblasts. As seen in Fig. 7A, cells staining positive for S-100 were accompanied by relatively few fibroblasts, and possessed morphology consistent with that of Schwann cells. Suspension of primary cells in Matrigel was relatively facile, and injection of cell suspensions into individual channels resulted in a distribution of viable cells within distinct channels of the scaffold. Fig. 7B shows GFP-labeled Schwann cells densely populating the seven channels of the scaffold. Further evidence for the identity and viability of the cells in the scaffold is given in Fig. 7C, which is a transmission electron micrograph (TEM) of a Schwann cell. This image shows loosely packed chromatin with areas of condensation in the sub-membranous regions of the nucleus. Fig. 6. Morphology of degraded scaffolds. Top panel shows transverse and longitudinal sections of a scaffold after 8 weeks of degradation. Bottom panel, after 24 weeks. Top and bottom panels do not represent identical specimens. Each scale bar ¼ 500 mm.
3.6. Axon regeneration Scaffolds containing Schwann cells suspended in Matrigel promoted axon regeneration at 1 month after
Fig. 7. Schwann cells isolated from neonatal rat pups and cultured in vitro. (A) Phase contrast micrograph of primary Schwann cell culture stained with S-100. (B) Light micrograph of a transverse section through the scaffold, showing distribution of GFP-positive cells suspended in Matrigel within each of the 7 channels. Some scaffold autofluorescence is apparent; bar ¼ 500 mm. (C) TEM of Schwann cell with indicative morphology and contained in scaffold; scale bar ¼ 2 mm.
ARTICLE IN PRESS 426
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
implantation. Animals receiving no treatment, or those which received scaffolds with no Schwann cells possessed few or no regenerating axons (data not shown). Histology revealed the channels of the scaffold containing ample axon regeneration, which was apparent throughout the complete length of the scaffold. As shown in Fig. 8, the channels were found distributed approximately as arranged by the scaffold, with scaffold and tissue in the spaces separating the channels. Channels generally contained a centralized core of tissue containing axons and capillaries, surrounded by circum-
Fig. 8. Histological images of tissue cables within scaffold channels. (A) Histological cross section near rostral end of scaffold showing tissue cables within all seven channels; bar ¼ 500 mm and applies to both A and B. Boxed region is shown in greater detail in D. (B) Histological cross section shown in A with channels segmented and displayed in color. (C) Orthographic, wedged view of 16 serial slices registered, segmented, and reconstructed to reveal 3-D structure of channels. (D) Histological cross section of boxed region in A; scale bar ¼ 100 mm. (E) Orthographic, wedged view of reconstructed channel shown in D, with segmented axon bundles displayed in color.
ferential fibrous tissue. Macrophages could be identified, likely due to their engulfment of neurofilament-stained material. The 3-D reconstructions shown in Fig. 8C and D show clearly the arrangement of the seven channels containing regenerating axons, as well as live axon bundles within the channels.
4. Discussion The fabrication technique described was chosen for its versatility and simplicity. Virtually any polymer, degradable or inert, could conceivably be used, along with an appropriate solvent if necessary. The scaffold components experience only temperatures and pressures at or below ambient in this injection molding process. Thus, bioactive molecules can be incorporated without risk of denaturation or degradation due to high heat or pressure. A variety of techniques for the production of pores can be used either in addition to, or in lieu of, the solvent evaporation procedure. Further, manipulation of longitudinal geometry is achieved easily with this technique. Variations on the multiple-channel design have been reported for use in the regenerating spinal cord [33,34], and designs similar to that presented here have been described for use in peripheral nerve regeneration [35,36]. Other authors have suggested that scaffold architecture may play a role in maximizing axon regeneration in the spinal cord [17,37], but the optimum architecture as of yet remains unknown. The parallel, multichannel geometry described in this paper may provide certain advantages. First, the number, size, and arrangement of the channels are easily manipulated by choice of mold design. This may allow for the systematic manipulation of scaffold architecture and the testing of its impact, if any, on axon regeneration. Further, because the channels are distinct, as opposed to a global, isotropically oriented pore network, transplantation of cells into specific regions of the scaffold is achievable. Using either the simple parallel-wire approach, or the more complex SFF technique, it appears feasible to design scaffolds that may be used to transplant cells into appropriate regions of the scaffolds’ cross sections. These regions may correspond, for instance, to gray and white matter, ascending and descending pathways, and even distinct white matter tracts. The multiple channel design may help prevent collapse of the conduit containing a regenerating tissue cable, a phenomenon sometimes observed in degradable, hollow tubes [19]. Also, the global distribution of therapeutic agent is ideally suited for delivery to regenerating tissue in each channel, even in regions that would be inaccessible by diffusion from the perimeter alone. Rapid solvent evaporation, used in this case for the formation of voids between the parallel channels, was
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
shown to be an effective method for fabrication of scaffolds with a high void fraction. Effectual, though limited, control over the void structure was demonstrated by the simple variation of polymer concentration, as shown in Fig. 3. Void fraction was inversely proportional to the mass of polymer used for a given solvent volume, i.e. the polymer concentration. High void fraction may be desirable as it allows the scaffold to occupy a large space with a relatively low mass of synthetic material, but a very high void fraction may result in poor mechanical properties and a limited polymer mass from which to deliver biological molecules. Sustained drug delivery and scaffold permeability both depend on void interconnectivity for access to a diffusible medium. Void interconnectivity, as measured by the accessible void fraction (Fig. 4), increased with the void fraction. Consequently, the factors that will determine the optimum void fraction, are the mechanical properties of the scaffold, mass of scaffold required for sustained delivery, and void interconnectivity. Scaffold geometry, as revealed by mCT (Fig. 6), remained quite static for at least 8 weeks, and then changed drastically as the mass loss rate increased. It is difficult to decipher what impact this may have on axon regeneration in vivo. The images shown in this study were acquired after degrading samples were dried, so some of the architectural transformation may have been due to the drying process itself. Previous work by other investigators has shown that axon regeneration into biodegradable conduits can be observed as early as 2 weeks [38]. In the present set of experiments, axon regeneration similar to that shown in Fig. 8 was present throughout the length of the scaffold at 4 weeks after implantation, which is well before substantial changes in scaffold geometry were observed in this study. It is possible that once tissue has traversed the length of a channel, subsequent alterations in structure due to degradation may not impede further growth. Release of the surrogate drug progressed in concert with scaffold degradation, as shown in Fig. 5. A rather high variability was observed, indicated by standard error bars in the figure, and this is probably due to the inhomogeneity observed in the distribution of FITC-D within scaffolds. The optimal release profile will depend on the particular drug being employed, but in this example, sustained release was accomplished for a period of about 12 weeks. Of course, maintenance of a released molecule’s biologic activity will be necessary in order to elicit a desired response. This will need to be assessed independently for each biomolecule considered as a candidate for controlled release in this system. The scaffold used in this approach is not designed to guide regenerating fibers by providing an oriented surface on which to grow, as proposed by other investigators [17,33,37,39], but rather as a template for the arrangement of cells that may attract and guide
427
regenerating axons. Drugs may be incorporated to provide added tropic attraction, for neuroprotection, or to reduce the inhibitory effects of secondary injury. However, the basis for promotion of axon regeneration is envisioned to be in transplanted cells. Therefore, successful Schwann cell harvesting, suspension, and distribution into scaffold channels is of critical importance in this model. Regeneration may be affected by a multitude of variables, including cell culture purity, cell density, distribution within the matrix (i.e. Matrigel), and distribution within the channels. Images in Fig. 7 indicate that Schwann cell harvest, suspension in Matrigel, and loading into distinct channels is feasible, and that cells survive in culture under physiologic conditions within the scaffold. The ability of the cells to promote axon regeneration will depend further on their survival throughout the transplantation process of the cell-seeded scaffold into the injured spinal cord. Axon regeneration in this study was visualized by 3-D reconstruction of serial histological sections, an approach that appears useful for evaluation of regenerating tissue architecture. Histological sections alone reveal much about internal regeneration, but deciphering of 3D characteristics is often difficult. Reconstructed data may assist in determining factors that will affect the 3-D arrangement of regenerating tissue, and may provide clues regarding functional changes. The experiments outlined in this study suggest a model for simultaneous, in vivo testing of a multitude of factors, and combinations thereof, that may influence axon regeneration in the transected rat spinal cord. Aspects of this model may be systematically manipulated to study effects of the structural, cellular, and molecular microenvironments surrounding the injury site. As novel factors promoting axon regeneration are identified, variations and combinations may be investigated to assess interactions among those factors. Of course, if successful axon regeneration is accompanied by functional recovery, the model would immediately suggest a possible treatment strategy.
5. Conclusions The data shown in the present work indicate that a simple injection molding/solvent evaporation technique may be used to produce biodegradable scaffolds with multiple-channel geometry. Scaffolds fabricated via these techniques possessed a relatively high void fraction, which could be modulated by changing the initial polymer concentration. Void space interconnectivity was found to increase as void fraction increased. Scaffolds degraded predictably in vitro, during which time sustained drug release occurred. Distinct channels of the scaffold could be seeded individually with Schwann cells, which survived for at least 48 h in vitro.
ARTICLE IN PRESS 428
M.J. Moore et al. / Biomaterials 27 (2006) 419–429
Schwann-cell containing scaffolds promoted axon regeneration in vivo, and regenerating tissue architecture could be described with 3-D reconstructions of histological sections. These scaffolds may serve as the basis for an in vivo model used to determine the effects of the structural, cellular, and molecular environments on regeneration of spinal cord axons.
Acknowledgements Funding support was provided in part by the Mayo Foundation and the National Institutes of Health through grants AR45871 to Dr. Yaszemski’s laboratory, EB02390 to Dr. Windebank’s laboratory, and predoctoral fellowship NS45544 to the first author. The authors would also like to thank Mr. James Gruetzmacher for GPC analysis, Ms. LouAnn Gross for microscopy sample processing, Mr. Fred Schultz for assistance with mold design and manufacturing, Mr. Adam Ellison for assistance with CAD, and the staff of the Mayo Biomedical Imaging Resource for assistance with image processing. References [1] Richardson PM, McGuinness UM, Aguayo AJ. Axons from CNS neurons regenerate into PNS grafts. Nature 1980;284:264–5. [2] Kuhlengel KR, Bunge MB, Bunge RP, Burton H. Implantation of cultured sensory neurons and Schwann cells into lesioned neonatal rat spinal cord. II. Implant characteristics and examination of corticospinal tract growth. J Comp Neurol 1990;293: 74–91. [3] Li Y, Raisman G. Schwann cells induce sprouting in motor and sensory axons in the adult rat spinal cord. J Neurosci 1994; 14:4050–63. [4] Li Y, Field PM, Raisman G. Repair of adult rat corticospinal tract by transplants of olfactory ensheathing cells. Science 1997; 277:2000–2. [5] Ramon-Cueto A, Plant GW, Avila J, Bunge MB. Long-distance axonal regeneration in the transected adult rat spinal cord is promoted by olfactory ensheathing glia transplants. J Neurosci 1998;18:3803–15. [6] McDonald JW, Liu XZ, Qu Y, Liu S, Mickey SK, Turetsky D, Gottlieb DI, Choi DW. Transplanted embryonic stem cells survive, differentiate and promote recovery in injured rat spinal cord. Nat Med 1999;5:1410–2. [7] Ogawa Y, Sawamoto K, Miyata T, Miyao S, Watanabe M, Nakamura M, Bregman BS, Koike M, Uchiyama Y, Toyama Y, Okano H. Transplantation of in vitro-expanded fetal neural progenitor cells results in neurogenesis and functional recovery after spinal cord contusion injury in adult rats. J Neurosci Res 2002;69:925–33. [8] Fernandez E, Pallini R, Lauretti L, Mercanti D, Serra A, Calissano P. Spinal cord transection in adult rats: effects of local infusion of nerve growth factor on the corticospinal tract axons. Neurosurgery 1993;33:889–93. [9] Schnell L, Schneider R, Kolbeck R, Barde YA, Schwab ME. Neurotrophin-3 enhances sprouting of corticospinal tract during development and after adult spinal cord lesion. Nature 1994;367: 170–3.
[10] Namiki J, Kojima A, Tator CH. Effect of brain-derived neurotrophic factor, nerve growth factor, and neurotrophin-3 on functional recovery and regeneration after spinal cord injury in adult rats. J Neurotrauma 2000;17:1219–31. [11] Zuo J, Neubauer D, Dyess K, Ferguson TA, Muir D. Degradation of chondroitin sulfate proteoglycan enhances the neuritepromoting potential of spinal cord tissue. Exp Neurol 1998;154: 654–62. [12] Yick LW, Wu W, So KF, Yip HK, Shum DK. Chondroitinase ABC promotes axonal regeneration of Clarke’s neurons after spinal cord injury. Neuroreport 2000;11:1063–7. [13] Yick LW, Cheung PT, So KF, Wu W. Axonal regeneration of Clarke’s neurons beyond the spinal cord injury scar after treatment with chondroitinase ABC. Exp Neurol 2003;182: 160–8. [14] Lemons ML, Howland DR, Anderson DK. Chondroitin sulfate proteoglycan immunoreactivity increases following spinal cord injury and transplantation. Exp Neurol 1999;160:51–65. [15] Bradbury EJ, Moon LD, Popat RJ, King VR, Bennett GS, Patel PN, Fawcett JW, McMahon SB. Chondroitinase ABC promotes functional recovery after spinal cord injury. Nature 2002;416: 636–40. [16] Friedman JA, Windebank AJ, Moore MJ, Spinner RJ, Currier BL, Yaszemski MJ. Biodegradable polymer grafts for surgical repair of the injured spinal cord. Neurosurgery 2002;51:742–51 discussion 51-2. [17] Teng YD, Lavik EB, Qu X, Park KI, Ourednik J, Zurakowski D, Langer R, Snyder EY. Functional recovery following traumatic spinal cord injury mediated by a unique polymer scaffold seeded with neural stem cells. Proc Natl Acad Sci USA 2002;99:3024–9. [18] Xu XM, Zhang SX, Li H, Aebischer P, Bunge MB. Regrowth of axons into the distal spinal cord through a Schwann-cell-seeded mini-channel implanted into hemisected adult rat spinal cord. Eur J Neurosci 1999;11:1723–40. [19] Oudega M, Gautier SE, Chapon P, Fragoso M, Bates ML, Parel JM, Bunge MB. Axonal regeneration into Schwann cell grafts within resorbable poly(alpha-hydroxyacid) guidance channels in the adult rat spinal cord. Biomaterials 2001;22:1125–36. [20] Schmidt CE, Leach JB. Neural tissue engineering: strategies for repair and regeneration. Annu Rev Biomed Eng 2003;5:293–347. [21] Profyris C, Cheema SS, Zang D, Azari MF, Boyle K, Petratos S. Degenerative and regenerative mechanisms governing spinal cord injury. Neurobiol Dis 2004;15:415–36. [22] Silver J, Miller JH. Regeneration beyond the glial scar. Nat Rev Neurosci 2004;5:146–56. [23] Bamber NI, Li H, Lu X, Oudega M, Aebischer P, Xu XM. Neurotrophins BDNF and NT-3 promote axonal re-entry into the distal host spinal cord through Schwann cell-seeded minichannels. Eur J Neurosci 2001;13:257–68. [24] Tropea D, Caleo M, Maffei L. Synergistic effects of brain-derived neurotrophic factor and chondroitinase ABC on retinal fiber sprouting after denervation of the superior colliculus in adult rats. J Neurosci 2003;23:7034–44. [25] Chau CH, Shum DK, Li H, Pei J, Lui YY, Wirthlin L, Chan YS, Xu XM. Chondroitinase ABC enhances axonal regrowth through Schwann cell-seeded guidance channels after spinal cord injury. Faseb J 2004;18:194–6. [26] Jorgensen SM, Demirkaya O, Ritman EL. Three-dimensional imaging of vasculature and parenchyma in intact rodent organs with X-ray micro-CT. Am J Physiol 1998;275:H1103–14. [27] Robb RA, Hanson DP, Karwoski RA, Larson AG, Workman EL, Stacy MC. Analyze: a comprehensive, operator-interactive software package for multidimensional medical image display and analysis. Comput Med Imaging Graph 1989;13:433–54. [28] Moore MJ, Jabbari E, Ritman EL, Lu L, Currier BL, Windebank AJ, Yaszemski MJ. Quantitative analysis of interconnectivity of
ARTICLE IN PRESS M.J. Moore et al. / Biomaterials 27 (2006) 419–429
[29]
[30]
[31]
[32]
[33]
porous biodegradable scaffolds with micro-CT. J Biomed Mater Res A 2004;71:258–67. Brockes JP, Fields KL, Raff MC. Studies on cultured rat Schwann cells. I. Establishment of purified populations from cultures of peripheral nerve. Brain Res 1979;165:105–18. Guenard V, Kleitman N, Morrissey TK, Bunge RP, Aebischer P. Syngeneic Schwann cells derived from adult nerves seeded in semipermeable guidance channels enhance peripheral nerve regeneration. J Neurosci 1992;12:3310–20. Xu XM, Guenard V, Kleitman N, Bunge MB. Axonal regeneration into Schwann cell-seeded guidance channels grafted into transected adult rat spinal cord. J Comp Neurol 1995;351: 145–60. Mantila SM, Camp JJ, Krych AJ, Robb RA. Modeling the functional repair of nervous tissue in spinal cord injury. In: Galloway Jr. RL (Ed.), Medical imaging 2004, visualization, image-guided procedures, and display 5367, San Diego, CA, 2004. p. 491–9. Patist CM, Mulder MB, Gautier SE, Maquet V, Jerome R, Oudega M. Freeze-dried poly(D,L-lactic acid) macroporous guidance scaffolds impregnated with brain-derived neurotrophic factor in the transected adult rat thoracic spinal cord. Biomaterials 2004;25:1569–82.
429
[34] Stokols S, Tuszynski MH. The fabrication and characterization of linearly oriented nerve guidance scaffolds for spinal cord injury. Biomaterials 2004;25:5839–46. [35] Sundback C, Hadlock T, Cheney M, Vacanti J. Manufacture of porous polymer nerve conduits by a novel low-pressure injection molding process. Biomaterials 2003;24:819–30. [36] Bender MD, Bennett JM, Waddell RL, Doctor JS, Marra KG. Multi-channeled biodegradable polymer/CultiSpher composite nerve guides. Biomaterials 2004;25:1269–78. [37] Blacher S, Maquet V, Schils F, Martin D, Schoenen J, Moonen G, Jerome R, Pirard JP. Image analysis of the axonal ingrowth into poly(D,L-lactide) porous scaffolds in relation to the 3-D porous structure. Biomaterials 2003;24:1033–40. [38] Gautier SE, Oudega M, Fragoso M, Chapon P, Plant GW, Bunge MB, Parel JM. Poly(alpha-hydroxyacids) for application in the spinal cord: resorbability and biocompatibility with adult rat Schwann cells and spinal cord. J Biomed Mater Res 1998;42: 642–54. [39] Maquet V, Martin D, Scholtes F, Franzen R, Schoenen J, Moonen G, Jerome R. Poly(D,L-lactide) foams modified by poly(ethylene oxide)-block-poly(D,L-lactide) copolymers and aFGF: in vitro and in vivo evaluation for spinal cord regeneration. Biomaterials 2001;22:1137–46.