Introduction

Nerve gaps from segmental tissue loss due to injury or surgical resection are routinely repaired by autogenous nerve grafts. However, this currently accepted technique results in relatively poor recovery at the expense of normal donor nerves.14,24 Although several clinical alternatives are available for peripheral nerve gap repair, including allografts, autogenous grafts and tubularization, these methods provide suboptimal functional recovery compared to autograft repairs.37 Tubularization techniques have been extensively studied as an “off the shelf” alternative,8,25 and control randomized clinical trials have demonstrated promising results with biodegradable tube repair.52 In particular, conduits made of polyglycolic acid40 and type I collagen53 have demonstrated clinical benefit for nerve regeneration through short nerve gaps. However, despite the progress made on improved structural materials for the nerve tubes,11,36 functional recovery remains incomplete.32

The rat sciatic nerve originates from the spinal segments L4–L6 and it has been estimated that, at midthigh, the nerve is composed of approximately 27,000 axons; 6% myelinated motor axons, 23% and 48% are myelinated and unmyelinated sensory axons, respectively.44 A critical factor contributing to the lack of functional recovery after gap injuries seems to be the inappropriate pathways taken by regenerating axons through and distal from the nerve graft or conduits, which compromise functional recovery by erroneously innervating aberrant targets and making dysfunctional connections.26 Attempts to improve directed axonal growth have included the deployment of luminal fillers,7 bioabsorbable filaments,34,56 multi-luminal conduits,20,53 and electrical stimulation of the proximal stump.1 These strategies have resulted in promising but modest improvements on nerve regeneration and functional recovery.18,45

Here we evaluated whether linearly restricting the regeneration pathway within a biosynthetic nerve implant (BNI) through multiple longitudinally oriented collagen-filled microchannels, would enhance the regenerative process and the recovery of function after nerve gap repair. Using the rat sciatic nerve injury model, we demonstrated that axonal regeneration of motor and sensory neurons through the BNI may be comparable to the autologous nerve repair method and able to mediate recovery of reflexive behavior.

Methods

The Biosynthetic Nerve Implant

The BNI was designed to mediate fascicular-like growth of axons in peripheral nerve gap repair by directing the growth of the regenerating axons through multiple agarose microchannels filled with collagen (Fig. 1a). Agarose has been previously shown to be biocompatible in long-term studies, and so was chosen here to provide support to the collagen-filled channels.15,47 The BNIs were reproducibly prepared by a casting device that accommodates an external Micro-Renathane® tube (Braintree Scientific, Inc; OD 3 mm, ID 1.75 mm, and length of 12 mm) and manually perforated to increase nutrient and gas exchange (250 μm perforations, every 1 mm on all sides of the tube). A brush made of metal fibers (250 and 500 μm diameter) was inserted through the tube into a “loading well”. The device was then cleaned with 70% ethanol and UV sterilized for 1 h. A 1.5% sterile agarose solution was then injected into the tube, filling the lumen, and allowed to polymerize. Subsequently, liquid collagen or a cell/collagen suspension maintained at 4 °C was added into the loading well (Fig. 1b), then drawn into the lumen of the microchannels by the negative pressure created when removing the fibers from the polymerized agarose (Fig. 1c). The BNIs are then placed in the incubator for 15 min at 37 °C which induces the polymerization of the collagen gel and the stabilization of the matrix and cells in the BNI microchannels.

Figure 1
figure 1

Simultaneous fabrication and cell-seeding of multi-luminal nerve guides. (a) Illustration of the biosynthetic nerve implant (BNI). (b) Fabrication method consisting of: (i) inserting a brush made of metal fibers through polyurethane tubing and into a compartment filled with collagen or cells (black dots), (ii) polymerization of agarose in the tube, (iii) subsequent removal of the fibers from the well creates multiple microchannels simultaneously filled with collagen and/or cells. (c) The BNI external tubing and multi-channel collagen fillers. (d) Photograph of a BNI seeded with GFP+ Schwann cells after 48 h in culture showing even cell dispersion and survival. (e) Shows a higher magnification of GFP-expressing Schwann cells in (d), illustrating the normal fusiform morphology of those cells in the BNI. (f) Merged 3D stack reconstruction of β-tubulin-positive DRG axons growing through collagen. Scale bars: (c) 1.2 mm, (d) 0.6 mm, (e) 125 μm, (f) 90 μm

Full size image

Cell survival inside the agarose microchannels was confirmed by the long-term survival evaluation (up to 15 days in vitro) and growth of green fluorescent protein (GFP) labeled cells seeded in the BNI (Figs. 1d, 1e). The ability of the agarose microchannels to linearly entice axonal regeneration within the BNI was confirmed by the growth of neonatal (P0–P4) murine DRGs from one end of the tube (Fig. 1f). Neurons were cultured in Neurobasal/B27/l-glutamine (2 mM) media without neurotrophins for 48–72 h. For visualization, explants were rinsed with PBS, fixed for 45 min in 4% paraformaldehyde (PFA), and processed for immunocytochemical staining. These data confirmed the ability of the BNI to support gas and nutrient exchange, thus providing a viable structural regenerative scaffold for nerve regeneration.

Dorsal Root Ganglion Explants Culture

Dorsal root ganglia (DRG) were obtained from neonate (P0–P3) rats. The animals were anesthetized by hypothermia and killed. The lumbar spinal cord was exposed and the L2–L6 DRGs were isolated, transferred to hank’s buffered media (Gibco, Carlsbad, CA) and cleaned from dorsal and ventral roots using tungsten needles. The DRGs were then affixed at one end of the multi-luminal hydrogel using 50 μL of growth factor reduced extracellular matrix (ECM; Matrigel, BD Biosciences, San Jose, CA) polymerized for 5 min at 37 °C. Neurobasal medium supplemented with l-glutamine, B-27 and penicillin/streptomycin (Gibco) was then used as culture media.

Schwann Cell Culture

Schwann cells were obtained as previously described.16 Briefly, adult rat sciatic nerve explants with removed epineurium were cultured for 2 weeks to allow for fibroblast migration. The explants were then enzymatically digested using Collagenase (ICN Biomedicals, Aurora, OH) followed by 1% Trypsin (bovine pancreas trypsin; Sigma) at 37 °C. The cells were cultured in serum-enriched Dulbecco’s Modified Eagle’s Media (DMEM/10% fetal bovine serum) supplemented with forskolin, pituitary gland extract, and herregulin. Green fluorescent protein (GFP) expression was obtained by transducing the Schwann cells with a GFP-expressing adenovirus (1 × 106 plaque forming units/μL) obtained as previously reported.38,39

Animals

48 Lewis rats (Harlan Sprague-Dawley Inc, Indianapolis, IN) were used in this study. The animals were divided equally into four groups and then separated for tract-tracing and morphometric analysis (n = 4–6 per group/evaluation method). The experimental groups included: (1) non-injured controls, (2) double resection achieved by sciatic nerve transection and re-suturing, (3) sciatic nerve transection repaired with a tube filled with collagen, and (4) sciatic nerve transection tube repaired with 14 collagen-filled agarose microchannels (BNI). The initial six animals of this study were implanted with 7-microchannel BNIs before 14 microchannel BNIs became available. The animals were maintained under conditions of controlled light and temperature. Food and water were available ad libitum. Institutional Animal Care and Research Advisory Committee regulations were observed for surgical, behavioral, and care procedures.

Surgical Procedure

Under anesthesia (ketamine 87 mg/kg/medetomidine 13 mg/kg), the left sciatic nerve was exposed through an upper lateral thigh incision. In the “double resection” group we avoided the morbidity associated with harvesting the contralateral nerve to repair the tissue gap, which defines a true autograft. Instead, we resorted to an approximation in which the sciatic nerve was sectioned, creating a 10 mm gap, and sutured back in place. To prevent nerve retraction, the suture was threaded through the nerve prior to segmentation and the proximal and distal segments were cut and sutured sequentially. The expected result from this control is better than a true autograft as the nerve segments are perfectly aligned, but the repair method still results in a 10 mm denervation gap injury. Thus, the double resection repair served as a more stringent control than the traditional autograft method. In all the experimental tubularization groups a 5–7 mm segment was excised proximal to the trifurcation of the sciatic nerve into the tibial, sural, and common peroneal nerves, and allowed to retract (1–2 mm). We then placed 12 mm Micro-Renathane® tubing over the underlying muscle and secured it with a 6-0 suture. A 1 mm segment of the proximal and distal nerve ends was inserted into the tubing and secured in place using a 10-0 nylon suture, thus rendering a 10 mm gap defect. The overlying muscle was then sutured and the skin stapled. Post-operatively the animals received atipamezole 1 mg/kg and recovered for 10 weeks.

Behavioral Testing

Animals were tested for recovery of motor function using the Digit Abduction Score (DAS) assay which semi-quantitatively measures (distal) muscle weakness,9 and was used to evaluate motor axon reinnervation. Briefly, the animals were tail-suspended to elicit hind limb extension and digit abduction. The extended hind limbs were photographed each week and digit abduction scored on a five-point scale (0 = normal to 4 = maximal reduction in digit abduction and leg extension) by four independent observers blinded to the treatment. The distance of toe spread was measured from the images obtained during tail suspension (n = 5–6 in each group).

To evaluate locomotor recovery, gait was observed using the CatWalk as described previously.21 Briefly, light from an encased source enters a glass plate such that the light is internally reflected except at points where an object makes contact with the glass. Rats were enticed to walk along the glass surface and their footprints recorded by a digital video camera.

Retrograde Tracing

A subset of animals (n = 4 per group) was evaluated for anatomical regeneration using a fluorescent retrograde tract-tracer from the sciatic nerve distal to the implant. Animals were anesthetized and the sciatic nerve distal to the repair site was exposed and transected. FluoroGold (FG: Fluorochrome, Englewook, CO, USA) crystals were placed on the cut end of the regenerated nerve for 10 min. The nerve stump was then carefully rinsed and repositioned in the leg after which the soft tissues and skin were surgically closed. These animals were allowed to survive for 6 days prior to tissue harvesting. Animals were euthanized with an overdose of pentobarbital and perfused with PBS followed by 4% PFA. The lumbar spinal cords were cryosectioned and FG positive cells with clear nuclei were counted in the dorsal root ganglion and ventral horn at L3–L6 levels which provide sensorimotor information from the hind limb.

Immunostaining

Cells or tissues were incubated for 1 h at room temperature and with gentle agitation in a combination of primary antibodies against acetylated β-tubulin (1:200) and S-100 (1:500; Sigma-Aldrich, St. Louis, MO), for the identification of axons and Schwann cells, respectively. Visualization was achieved by tissue incubation in Cy2- and Cy3-conjugated secondary antibodies (1:400 1 h RT; Jackson Labs, West Grove, PA). Neurotrace (1:250; Invitrogen, Carlsbad, CA) was used as fluorescent Nissl counterstain. The staining was evaluated using a Zeiss Pascal confocal microscope.

Electron Microscopy and Histomorphometry

In a subset of animals (n = 6 per group) a segment of the regenerated nerve was harvested and post-fixed in 2% glutaraldehyde/1% PFA/0.15 M sodium cacodylate, pH 7.2 at 4 °C overnight, rinsed, stained in 2% uranyl acetate, dehydrated, and infused in propylene oxide/Durcupan (25/75; Fluka Chemika-BioChemika, Ronkonkoma, NY) for 1 h at room temperature. Nerves were then embedded in fresh Durcupan resin and polymerized 24–36 h at 65 °C. 1 μm thick sections were stained with Toluidine blue. Thin sections were viewed at 60 kV and photographed on a JEOL 100 CX conventional transmission electron microscope. For quantification, 21 pictures were randomly taken of each nerve cross-section covering 1575 μm2 per picture, and totaling 0.033 mm2 in sampling area per animal. A MACRO (Zeiss, Co.) was written to evaluate the number of myelinated axons, axon diameter and myelin thickness in each electron micrograph, which was validated by direct comparison with measurements obtained manually. The number of unmyelinated axons was estimated manually from photographic prints.

Statistical Analysis

Raw data were analyzed by ANOVA followed by Neuman–Keuls multiple comparison post hoc test with an α level of 0.05 (Prism 4; GraphPad Software Inc.).

Results

Fascicular-Like Nerve Regeneration Through Collagen-Filled Hydrogel Microchannels

Multi-luminal nerve regeneration was observed in all BNI repairs of 10 mm sciatic nerve gap injuries (Fig. 2). Compared to the double resection (Fig. 2a) and the simple tube (Fig. 2b) repair methods, collagen-loaded BNIs with either 7 (Figs. 2c, 2e, 2g) or 14-microchannels (Figs. 2d, 2f, 2h) promoted fascicular-like nerve repair though out the length of the implant. In all the available microchannels, the regenerated nerve tissue appeared similar in thickness. Some evidence suggestive of vascularization of the BNI nerve fascicles was observed inside each micro-channel (Figs. 2e, 2h). The perforations made throughout the length of the external tube allowed the migration of mesenchymal cells between the luminar wall of the external tube and the periphery of the agarose explant itself, forming a cell layer with evidence of vascularization, not unlike the normal nerve perineurium (Figs. 2f, 2h). No gross evidence of chronic inflammation or tissue reaction was observed in any of the BNI-implanted animals.

Figure 2
figure 2

Multi-luminal nerve repair through the BNI. Adult rat sciatic nerves are shown at 10 weeks post-gap repair either by double resection (a), collagen-filled tubularization (b), collagen-loaded 7-channel BNI with external tubing (c, e, g), and 14-channel BNI without the external tubing (d, f, h). Nerve regeneration was observed in the BNI-repaired animals through the available microchannels. Vascularization is indicated in both the intraluminal nerve regenerate (arrowheads) as well as in the outer mesenchymal membrane (arrows). Scale bars: (a) 2 mm, (e) 400 μm

Full size image

Toluidine histology in thin sections confirmed that, similar to double resection (Fig. 3a) or collagen-filled tube (Fig. 3b) repairs, multi-luminal BNI nerve repair mediated normal nerve regrowth (Figs. 3c, 3d). The regenerated nerves in each micro-channel contained numerous axons surrounded by a perineurium-like outer membrane highly resembling the fascicular architecture of the normal nerves (Figs. 3c, 3d). Electron microscopic evaluation of the regenerated tissue showed that the perineurium, a 6–8 myofibrotic cell layer wrapping each nerve fascicle and characterized by their pinocytotic vesicular content and tight junctions (Fig. 4a), was present in all repaired groups. However, in contrast to the highly condensed perineurium observed in uninjured controls, animals with double resection repair showed a less condensed perineurium (Fig. 4c), which qualitatively appeared slightly more disorganized in the tube/collagen and BNI-regenerated groups (Figs. 4e, 4g). When considering remyelination, all the repaired groups appeared to have a comparable proportion of myelinated and unmyelinated fibers (Figs. 4d, 4f, 4h), which at 10 weeks after injury were clearly not as mature (defined by myelin thickness and number of myelinated axons) as those observed in uninjured controls (Fig. 4b).

Figure 3
figure 3

Toludine blue-stained sections of double resection (a), collagen-filled tube (b), and BNI-repaired (c, d) sciatic nerves. A mesenchymal layer was observed to cover the outer surface of the BNI hydrogels (arrow). Higher magnification of the insert in (c), (d) shows a perineurium-like layer (arrows), blood vessels (arrowheads), and densely packed axons regenerating within the BNI microchannels. Scale bars: (a–c) 400 μm, (d) 50 μm

Full size image
Figure 4
figure 4

Electron microscopy photographs of uninjured normal controls (a, b), and those repaired by double resection (c, d), collagen-loaded tube (e, f), and 14-channel BNI (g, h), 10 weeks after sciatic nerve transection. The outer 6–8 cell layer perineurium in the uninjured control is densely packed (a). Reconstitution of this myofibrotic membrane in the double resection (c), tube/collagen (e), and BNI (g) show different levels of reconstitution of the perineurium in each repair method. Scale bar = 10 μm

Full size image

Quantification of nerve regeneration, which was estimated by tracing the areas containing visible nerve growth, showed comparable axonal growth among uninjured, double resection and animals repaired with simple tubes filled with collagen (Fig. 5a). In contrast, the regenerated area observed in animals repaired with the 14-channel BNI was significantly reduced, reflecting the limited cross-sectional area through the openings of the microchannels available for nerve regeneration (approximately 34% of that available in the collagen-filled tube). Direct quantification of the regenerating nerve fibers from EM images over a 0.033 mm2 sampling area (see methods) revealed a 2–4 fold increase in the number of unmyelinated axons in all repair groups compared to uninjured controls (Fig. 5b), while the highest number of unmyelinated axons was present in the BNI group, no statistical difference was found between those in the collagen tube and the BNI collagen groups. The number of myelinated axons was also significantly increased in the double resection and BNI groups, compared to the tube/collagen repairs (5-fold and 3-fold, respectively, Fig. 5c). When the myelin thickness was evaluated (Fig. 5d), an approximately 30–60% reduction compared to uninjured controls was observed in all repaired groups. This proportion was lowest in the BNI-repaired animals. Evaluation of the axon diameter distribution in all repair groups (Fig. 5e) revealed that both the double resection and the BNI groups contained the highest number of small-diameter axons (<6 μm).

Figure 5
figure 5

Quantification of the total area of nerve regeneration indicates that BNI-repair offered limited available area for nerve repair compared to the other methods (a). Quantification of unmyelinated (b) and myelinated (c) axons. Separate analysis of myelin thickness (d) and axon diameter distribution (e), shows increased numbers of 4–6 μm axons in both the double resection and the BNI groups, and reduced myelination in all repaired groups compared to the normal animals (n = 6 per group; error bars = standard deviation). *p ≤ 0.05; **p ≤ 0.01

Full size image

Overall, the morphometric studies revealed multi-luminal nerve regeneration through the BNI that seemed to favor small unmyelinated axons and those with thinner myelin. This result might indicate that the collagen-filled BNI entices primarily small sensory fibers over high caliber motor axons. Alternatively, it is possible that the restricted growth and the high axon density in the BNI microchannels might have limited axon maturation and myelination, resulting in an increased number of fibers with thin or no myelin. To distinguish between these two possibilities, we evaluated the specific neuron subtypes of regenerated axons in the different repair groups.

Preferential Sensory Regeneration Through the BNI

A cohort of animals (n = 4 per group) underwent Fluoro-Gold (FG) tract-tracing of the sciatic nerve distal to the graft. Numerous FG+ cells were visualized in the sensory dorsal root ganglia (DRG; Fig. 6a) and in the ventral motoneuron pool of the spinal cord (VMN; Fig. 6b). Quantification of FG+ sensory neurons in the DRG was statistically comparable among all the groups (Fig. 6d). In contrast, the number of FG+ VMN in the spinal cord showed a significant reduction in the BNI-repair animals (approximately 20–45%) compared to the uninjured, double resection and tube/collagen groups (Fig. 6c). Given that the available area for axonal regeneration in the BNIs is about 60% less than in the other repair methods, it would seem that the BNI supported disproportionally greater amounts of sensory axon regeneration. While both sensory and motor axons were tract-traced distal to the graft in all repaired nerves, the collagen-filled BNI seemed to favor sensory neurons.

Figure 6
figure 6

Tract-tracing of regenerated neurons in the spinal cord and dorsal root ganglia. Regenerated motoneurons in the sensory neurons in the DRG (a) and ventral motor neurons (VMN) of the spinal cord (b) were visualized with Nissl staining and identified by FluoroGold (FG) tracing of axons distal to the grafted implant. Quantification of FG+ cells in the VMN (c) and DRG (d) (n = 4 animals per group; error bars = standard deviation). *p ≤ 0.01. Scale bar = 50 μm

Full size image

BNI-Repair Mediates Partial Motor Functional Improvement

To determine the extent to which these animals could use these regenerated axons, we investigated post-repair behavioral recovery as well. The Digit Abduction Score (DAS) assay was used to evaluate motor behavioral recovery9 (Figs. 7a, 7b). Baseline measurements were normal (score of 0) for all treatment groups and significantly increased to the worst score (4) immediately after injury to the sciatic nerve. The recovery of animals with double resection was noted as early as 5 weeks post-injury and reached their best score (0.5) at 7 weeks post-injury. Conversely, those repaired with either a tube/collagen or BNI, reached their best score at 8 weeks post-injury, with slight improvement to a score of 1.5 at week 10 in the collagen/tube group. The similar behavioral recovery observed in the collagen-filled tube and the BNI, despite the difference in available regeneration area (1.5 and 0.5 mm2; respectively) suggests the possibility of increased efficiency in axonal regeneration by the multiluminar BNI-repair method.

Figure 7
figure 7

Functional recovery mediated by repaired peripheral nerves. Reflexive motor function was evaluated using the digit abduction scoring assay and by measuring the distance of toe spread (a, b). Locomotor function was evaluated by catwalk analysis (c). The paw on the left of each image was the injured side of the animal and all show poor toe spread with plantar placement of the paw during stepping

Full size image

Surprisingly, when we evaluated locomotor function by the CatWalk assay we observed that none of the repaired animals, including those with double resection repair, regained the voluntary use of the repaired limbs (Fig. 7c). This result suggests that while nerve regeneration can be achieved through gap injuries, allowing the animals to reflexively use their limbs, the use of the repaired extremity for some motor tasks may require additional rehabilitation therapy. It is also possible that sensory regeneration made walking painful or was insufficient to provide normal feedback to spinal locomotor neural circuitry. Likewise, this behavioral testing could not rule out learned disuse despite preserved reflex behavior. Given that the extent of toe spread is similar between BNI and collagen tube animals, and the poor toe spread during locomotion is similar between all groups (Fig. 7), there may be some other, as yet undetermined, reason common to all these animals that might explain why one toe spread behavior and not another (during locomotion) recovers after injury and repair.

Discussion

Fascicular-Like Gap Repair by the Multiluminar BNI

Despite recent progress in the engineering of nerve grafts, alternate biosynthetic designs have not replaced the gold-standard autograft repair.45 Multi-luminal conduits have been proposed for peripheral nerve regeneration to better resemble the natural morphology of multiple fascicular compartments in the peripheral nerve.19 However, current fabrication methods of multi-luminal nerve scaffolds require rather complicated techniques, and the evidence on functional regenerative efficacy is either incomplete or lacking for the different proposed designs.10,20,31,48

Here we report a simple and reproducible method for the fabrication of multi-luminal biosynthetic nerve implants that compartmentalize axon regeneration and provide a permissive collagen intraluminal milieu within the channels to entice linear nerve growth. The agarose-made channels restricted the wandering of axons and prevented axonal crossing between them, while the intraluminal collagen provided the regenerative matrix needed for axonal growth. These characteristics resulted in efficient axonal regeneration in the BNI, as demonstrated by the increased number of both myelinated and unmyelinated axons observed in the microchannels despite the significantly reduced area available for regeneration in the BNI compared to regular tubularization or double resection repair. This design reproducibly supported fascicular-like nerve repair while bridging a 10 mm gap lesion in the rat sciatic nerve, mediating partial functional recovery similar to that observed in the collagen tube and in some cases, indistinguishable from double resection repaired animals.

The biomimetic nerve repair mediated by the BNI was further suggested by the formation of perineurial-like cell layers wrapping each of the nerve cables within the microchannels. In addition, mesenchymal cells were observed to migrate inside the tubing and formed a seemingly vascularized layer external to the agarose multilumen, resembling the regeneration of an epineurium in the BNI-implanted animals. While regeneration of a functional epineurium in the BNI would require direct testing of the regained integrity of the nerve–blood–brain barrier, the anatomical evidence obtained in this study is supportive of this possibility. This is relevant since the potential clinical application of the BNI design will eventually rely on the use of biodegradable external tubing which, after reabsorption, will benefit from the formation of a protective and vascularized layer external to the multi-luminal agarose filler (i.e., a perineurium-like layer). This is particularly important as regeneration of the epi and perineurium has been recognized as necessary in achieving normal nerve conduction restoration in regenerated peripheral nerves.29,42,51

Luminar Collagen Seems to Favor Sensory Neuron Regeneration

Peripheral nerve regeneration from the proximal stump is known to be attracted towards the distal end of the transected nerve through target-derived diffusible signals and to be influenced by the growth substrate.46,50 The resulting regenerated nerve, therefore, might be composed predominantly of motor fibers4 or of sensory axons,17 depending on the integrated interpretation of the diffusible signals and contact mediated guidance cues presented to them. Most studies utilizing multi-luminal nerve repair methods have described the innervation of motor axons,11,12 while the extent of sensory fiber regeneration has not been fully characterized. In this study, we directly evaluated the number of motor and sensory axons that regenerated through the BNI. We observed a 15–30% reduction in the number of regenerating motoneurons (uncorrected for the reduced available regeneration area), but a normal number of re-growing sensory neurons, compared to the double resection and collagen-filled tubularization repair methods despite the reduced area for regeneration.

While the mechanism of the apparent preferential growth of sensory fibers through the BNI is not known, previous reports suggest that the collagen matrix might be particularly instructive to sensory neurons. The work by Yoshii and colleagues seems to support this notion, as they reported that the linearization of collagen fibers within a tube promoted the regeneration of myelinated neurons with a smaller diameter, consistent with known morphology of sensory neurons.57 However, other factors might also play a role such as primary and secondary injury response differences in the various repair methods, variable physical properties (i.e., mechanical stiffness of the collagen), adhesion, and availability and concentration of soluble growth factors.

This finding suggests that other instructive molecules such as NCAM, LI, and HNK1 should be incorporated into individual lumens of the BNI in order to influence the regeneration modality in each of the microchannels. In the peripheral nervous system, cell adhesion molecules L1 and N-CAM are expressed by Schwann cells and provide axonal regeneration selectively.27 Similarly, HNK1 is a carbohydrate that has been reported to provide selectivity for the regeneration of motoneurons.43 Therefore, the incorporation of specific instructive molecules in the different microchannels will provide more control over the type of neuron that is elicited to regenerate through the microchannels and thus the type of fascicle (i.e., motor or sensory) intended for regeneration.

Increasing the Regeneration Efficacy in Multiluminar Nerve Repair

Although axonal regeneration was demonstrated in the BNI gap repair, the total available area for regeneration and the functional recovery mediated by this method were suboptimal compared to that in double resection-repaired animals. A number of modifications can be introduced into multi-luminal nerve repair designs to improve its regenerative potential. The development of different multi-luminal geometric configurations for the microchannels specifically aimed at maximizing the total available area for regeneration seem granted in future studies, as it will permit an increase in the total number of regenerated axons, possibly providing greater functional recovery. We have estimated that a modification of the BNI multilumen from 14 channels (250 μm internal diameter each), to eight channels; seven in the periphery (300 μm internal diameter), and a central one of 500 μm in diameter, increases the total available area from 34 to 70%. Furthermore, other materials can be incorporated or tested in lieu of the agarose or collagen, such as alginate, fibronectin, or synthetic ECM.28,33,49,54

In addition, functional recovery after peripheral nerve injury requires the accurate regeneration of axons to their original distal targets. This can be further improved in the BNI by the incorporation of controlled neurotrophin release3,30,55 and/or Schwann cells within the lumen of the microchannels.35 However, while these strategies have been the focus of much attention in the neuroregenerative field,5,6 major obstacles remain to be overcome before they can have a practical impact in the clinical repair of nerve gaps. Among them are: (a) the definition of the specific neurotrophic factors, alone or in combination, that will benefit the regeneration of motor and modality specific sensory neurons, (b) the development of methods to extend the spatio-temporal growth factor delivery distal to the repair site needed to ensure proper target innervation, and (c) the implementation of Schwann cell survival strategies, as this is highly compromised at grafting, particularly in areas of traumatic injury, edema, and inflammation.22

Additional strategies to be considered include electrical stimulation and treadmill exercise which have been shown to improve axon regeneration and functional outcomes following peripheral nerve injury.2,13,41 These strategies are particularly relevant considering the results of our CatWalk study, in which none of the repaired animals learned to reuse their paws during stepping, despite clear functional recovery of distal digit reflex function. Since it has been estimated that the rat sciatic nerve regenerates at a rate of 1–3 mm/day, reaching the leg muscles in 28 days post-injury,23 the lack of normal plantar placement during the CatWalk test, is unlikely the result of a lack of target innervations. Conversely, the substantial functional recovery of the toe-spreading reflex in all groups, combined with the lack of normal function in all animal groups during the CatWalk, strongly suggest that while the neurons were able to reach their targets, the animals were not using them during locomotion. This result underscores the limitations of individual functional testing and the importance of limb disuse abnormal behavior after nerve repair. Perhaps a combination of treadmill training and electrical stimulation will encourage not only correct reinnervation of the lower limb, but also promote the reuse of the reinnervated target muscles.

Conclusions

In summary, we demonstrate that fascicular-like nerve gap repair can be achieved by compartmentalizing axonal regeneration and that this method will likely offer additional alternatives to control the regenerative process in the peripheral or central nervous system by the presentation of modality specific guidance signals in individual microchannels.