Preclinical study of peripheral nerve regeneration using nerve guidance conduits based on polyhydroxyalkanaotes

Abstract Nerve guidance conduits (NGCs) are used as an alternative to the “gold standard” nerve autografting, preventing the need for surgical intervention required to harvest autologous nerves. However, the regeneration outcomes achieved with the current NGCs are only comparable with autografting when the gap is short (less than 10 mm). In the present study, we have developed NGCs made from a blend of polyhydroxyalkanoates, a family of natural resorbable polymers. Hollow NGCs made from a 75:25 poly(3‐hydroxyoctanoate)/poly(3‐hydroxybutyrate) blend (PHA‐NGCs) were manufactured using dip‐molding. These PHA‐NGCs showed appropriate flexibility for peripheral nerve regeneration. In vitro cell studies performed using RT4‐D6P2T rat Schwann cell line confirmed that the material is capable of sustaining cell proliferation and adhesion. PHA‐NGCs were then implanted in vivo to repair 10 mm gaps of the median nerve of female Wistar rats for 12 weeks. Functional evaluation of the regenerated nerve using the grasping test showed that PHA‐NGCs displayed similar motor recovery as the autograft, starting from week 7. Additionally, nerve cross‐sectional area, density and number of myelinated cells, as well as axon diameter, fiber diameter, myelin thickness and g‐ratio obtained using the PHA‐NGCs were found comparable to an autograft. This preclinical data confirmed that the PHA‐NGCs are indeed highly promising candidates for peripheral nerve regeneration.

tremendous advances in surgical techniques and tissue engineering, the prognosis of PNI is still poor.
PNIs may be caused by acute compression, laceration, or penetrating trauma resulting in the loss of sensory function, motor function or both. Nerve regeneration and recovery of nerve function depend on the type of nerve fiber injury, patient age, site of the lesion, length of the defect, level of damage of the surrounding tissues, and availability of neurotrophic factors. 4 In the mildest kind of nerve injury, neurapraxia, the continuity of endoneurial tubes is preserved, and recovery occurs without Wallerian degeneration. However, neurotmesis and axonotmesis involve the loss of axonal continuity and the distal segment of injury undergoes Wallerian degeneration. Neurotmesis is the most severe type of nerve fiber injury including stretch injuries and laceration. 4,5 When the nerve gap is less than 5 mm, peripheral nerves can regenerate spontaneously with the support of Schwann cells (SCs), that promote a beneficial environment for axonal growth. As a response to denervation, SCs located in the distal axonal segment secrete a range of growth factors to facilitate regenerating axons to reach their sensory end organ or target muscle. In this case, end-to-end epineurial neurorrhaphy is suitable if tension free coaptation can be achieved after suturing the two stumps. For more severe injuries, implantation of an autologous nerve graft (autograft) is the gold standard procedure. However, nerve autograft may potentially involve further complications including scar tissue formation, donor site morbidity, lack of donor nerves and aberrant regeneration. 6,7 Therefore, new therapeutic strategies for peripheral nerve repair have focused on the development of nerve guidance conduits (NGCs) as alternatives to nerve autografts.
Current commercially available NGCs exhibit considerable drawbacks. For example, synthetic bioresorbable NGCs may produce an immune response, scar tissue, and release of by-products that are detrimental for the regeneration process. Nonbiodegradable NGCs involve a second surgery for conduit removal, comprising an additional disadvantage. 8 Hence, a diversity of materials, nanostructures and biochemical factors have been investigated in attempts to improve the performance of NGCs. [9][10][11] Bioresorbable materials are preferred over non-bioresorbable materials since they prevent both, chronic nerve compression and fibrotic reactions; and have been shown to produce a reduced risk of neuromas. 8,12 Although NGCs made from polymers of natural origin have shown a reduced immune reaction, the regeneration outcomes are not as good as the autograft when the gaps are longer than 10 mm.
Polyhydroxyalkanoates (PHAs), polymers of bacterial-origin, are gaining increasing popularity, since they exhibit high biocompatibility, and tuneable biodegradability and mechanical properties. 13 Studies have shown that D-3-hydroxybutyric acid (3HB), a natural constituent of blood, 14 is a degradation product of some PHAs, which contributes to their high biocompatibility. Moreover, PHAs have shown superior biocompatibility with neuronal cells compared to the widely used synthetic polymers, polycaprolactone (PCL) 6 and PLA. 15,16 PHAs exhibit properties that may overcome some of the limitations of the available NGCs including controllable surface erosion, lower acidity of their degradation products and longer-lasting stability compared to their synthetic counterparts.
Although P(3HB) have previously displayed satisfactory nerve regeneration, its mechanical properties are unsuitable for peripheral nerve repair. 17,18 To overcome this limitation, we have fabricated NGCs by the dip molding technique using the biodegradable PHAblend 75:25 P(3HO)/P(3HB), which has been shown to possess the required flexibility and biocompatibility for this application. 16 In the present study, we have carried out, for the first time, preclinical assessment of novel PHA blend-based NGCs, PHA-NGCs, for regeneration of median nerve gaps and functional repair by using the tubulation technique, with significantly promising results.

| Manufacturing of NGCs
PHA-NGCs were fabricated from P(3HO)/P(3HB) 75/25 blend whose biocompatibility with neuronal cells was previously assessed by Lizarraga-Valderrama et al. 6 The NGCs were made by a multi-dip molding process using a solution of P(3HO) and P(3HB) mixture (mass ratio 75 to 25) with a total polymer concentration of 6 wt% in chloroform. PTL-MMB02 Programmable Dip Coater (MTI Corporation, Richmond, CA) was used for mandrel dipping into the polymer solution with a speed of 200 mm/min. NGCs were produced by using a six-mandrel tool with mandrels of 1.8 mm outer diameter, along with a matching vial holder. Dip molding was conducted through five coating cycles consisting of five dips, resulting in a total of 25 dips. The drying time between dips in a coating cycle was 30 sec while the drying time between cycles was 4 min. After completing the coating cycles, the tubes were kept at room temperature for 3 days to complete solvent evaporation. Tubes were removed from the mandrels and stored for 5 weeks at room temperature (aged NGCs) before all the tests were carried out. All the tubes were previously sputter-coated with a 20 nm film of palladium using a Polaron E5000 sputter coater. The operating pressure of the sputter coating was 5 Â 10 À5 bar with a deposition current of 20 mA for a duration of 90 s. The images were then recorded at 5 kV using the FEI software.

| Mechanical analysis
Tensile testing was carried out using a 5942 Testing Systems (Instron, High Wycombe, UK) equipped with a 500N load cell at room temperature. NGCs of total length around 40 mm were fixed in rubber-coated grips with the separation distance between the grips of 23 mm. Metal mandrels were inserted into the NGCs from both sides to the full gripping length (approximately 8 mm). Deformation rate was set to 10 mm per min. The average values for four specimens were calculated.

| Thermal analysis
Thermal transitions of NGC polymer blends were studied using DSC 214 Polyma (Netzsch, Germany), equipped with Intracooler IC70 cooling system. Scanning was conducted between À70 and 200 C at a heating rate of 10 C/min under the flow of nitrogen at 60 mL/min. Samples of known history (aged for 5 weeks at room temperature) were used in the DSC studies. Therefore, all thermal transitions were analyzed for the first heating which provided properties for the conditioned material. Enthalpy of fusion for each component of the polymer blend was normalized to a corresponding mass fraction.  incision from the nipple to the elbow, the median nerve was isolated to establish a defect in the middle of the exposed part, immediately followed by nerve repair according to the experimental groups. In the experimental group PHA blend NGC, a median nerve segment was removed and 10 mm-long conduits were inserted and sutured to the nerve stumps with one 9/0 epineural stitch to each nerve ends. In the Autograft group, the median nerve segments were removed, reversed (distal-proximal) and sutured with three 9/0 epineural stitch to the nerve stumps of the same nerve ( Figure 1).

| Surgery
PHA NGCs were immersed in sterile saline for at least 5 min before implantation. Animals were sacrificed by anesthetic overdose after 6 weeks for a qualitative observation of the ongoing regeneration (n = 3 PHA-NGC) or after 12 weeks for quantitative analysis of nerve regeneration (n = 6 PHA-NGC and n = 6 Autograft).

| Functional evaluation of the regenerated nerve: The grasping test
The grasping test was performed to estimate the functional recovery after nerve reconstruction. The analysis was carried out every 2-3 weeks until the animal was sacrificed (12 weeks after surgery) following the same procedure previously described by Papalia et al. 19 and Ronchi et al. 20

| Immunohistochemistry
The regenerated nerve inside the conduit (for the sample withdrawn  with a DFC320 digital camera and an IM50 image manager system (Leica Microsystems, Wetzlar, Germany) was used for section analysis.
With the same ultramicrotome, ultra-thin sections (70 nm thick) were cut and stained with saturated aqueous solution of uranyl acetate and lead citrate. Sections were analyzed using a JEM-1010 transmission electron microscope (JEOL, Tokyo, Japan) equipped with a Mega-View-III digital camera and a Soft-Imaging-System (SIS, Münster, Germany) for the computerized acquisition of the images.

| Quantitative analysis of nerve regeneration: Stereological and morphometrical analysis
In order to quantify myelinated nerve fibers with high resolution light microscopy, one toluidine blue stained semi-thin section was selected and the total cross-sectional area of the whole nerve was measured.
Thirteen to fifteen sampling fields were selected using a systematic random sampling protocol, as previously described by Geuna 21 and Geuna et al. 22 In each sampling field, a two-dimensional dissector procedure was adopted to cope with the edge effect. 22 Mean fiber density, total fiber number, fiber and axon diameter, myelin thickness and g-ratio were then estimated.  Our preliminary experiments demonstrated that due to the high molecular weight of the natural PHAs it was not possible to prepare processable solutions with the concentrations which would allow fabrication of tubes with sufficient wall thickness after a single dipping. Therefore, we adopted a multi-dip molding process using a solu-  Figure S1). However, the gaps between the inclusions of the dispersed phase and continuous phase are not evident which implies good adhesion between the phases. The NGCs were characterized by a high flexibility inherited from the elastomeric P(3HO) (SI, Figure S1). These NGCs could withstand deformations up to 150% which was larger than rat sciatic nerves can resist. Ultimate tensile strength and Young's modulus of the aged NGCs were found to be 6 and 35 MPa, respectively (Table 1). The Young's modulus of the NGC is 60 times higher than rat sciatic nerve (Table 1). Although the stiffness of NGCs is significantly higher compared to that of the rat sciatic nerves, it should be taken into consideration that the NGCs were tested in a dry state and the stiffness is expected to decrease for the wet NGCs after implantation. However, these changes cannot be significant since PHAs, particularly P(3HO), are hydrophobic polymers and their swelling in aqueous media is limited. Also, higher stiffness of NGCs can be beneficial to prevent the collapse of the hollow structure after implantation.

| In vitro proliferation and cell morphology assay
In vitro proliferation and cell morphology assays, using the RT4-D6P2T cell line, were performed to evaluate the ability of the glial cells tested to make direct contact with the substrate represented by PHA-NGC and consequently to determine the biocompatibility and the biomimetic properties. RT4-D6P2T proliferation was significantly higher on the control compared to PHA-NGCs ( Figure 1h). However, RT4-D6P2T morphology at 2 DIV was very well organized and cell dimensions significantly higher on the PHA-NGCs compared to the control (Figure 1i).

| Surgical procedure
The rat median nerve repair model was used as a preclinical test were also immunolabeled for neurofilament and S100 protein, respectively. Figure 3h shows the distribution of nerve fibers (green) and Schwann cells (red) within the nerve cross-section.

| Functional recovery
To assess the functional recovery of the median nerve, the grasping test was performed in all operated animals at week 2, 4, 5, 7, 9, and understood multicellular response. However, this process might be considerably hindered when injury gaps are longer than 5 mm. 8 The natural regenerative process in peripheral nerves involve distinct stages lead by various cells including neurons, SCs, macrophages, fibroblasts, endothelial cells, and neutrophils. 24,25 Following the severing of the nerve, the stumps retract and secrete plasma exudate rich in neurotrophic factors and ECM precursor molecules including fibrinogen and factor XIII. 25,26 A "bridge" composed of perineurial cells, inflammatory cells, fibroblasts and matrix is formed to reconnect the two nerve stumps. 25 When regeneration occurs within a hollow tube, initially an acellular fibrin cable forms between the two stumps, followed by migration of SCs, endothelial cells and fibroblasts along the cable. 26 This fibrin cable normally forms within 1 week across a noncritical 10 mm gap in rat models. 26,27 After cellular migration, degradation and removal of the fibrin cable lead to the onset of remyelination by switching from progenitor-like phenotype SCs to a more mature "myelinating" phenotype. As a result, these mature SCs wrap around the regenerated axons to form the myelin sheath. 26 The regenerative process inside the PHA-NGCs followed the same phases, with the formation of a thin regenerated nerve inside the PHA-NGCs after 6 weeks and still a thin (but bigger as compared to 6 weeks) regenerated nerve after 12 weeks. In both time points the cross section of the regenerated nerve appeared to be thinner as compared to both the diameter of the conduit and the diameter of the nerve stumps, but increased over time. Also, in both time points the regenerated nerve grew in the middle of the conduit. In our experience, this is a normal process in nerve regeneration: regenerated nerve fibers start to grow where the fibrin cable was located and appear to have a thin cross-section, at least for the first weeks/ months. We cannot exclude the fact that the space between the regenerating fibers and the inner wall of the PHA-NGCs was colonized by some material (in particular ECM), but unfortunately, after the removal of the conduit for subsequent analysis, this material was lost.
Indeed, to perform quantitative stereological and morphometrical analyses on resin-embedded nerves, we removed the PHA-NGCs before the fixation with glutaraldehyde. Resin embedding and toluidine blue staining of nerve cross section allowed to perform a reproducible and standardized assessment of the degree of nerve regeneration, by preserving the fine structure of the nerve tissue and by providing high quality and clear detailed images of nerve fibers (and in particular of the myelin sheath). 28 On the other hand, by removing the PHA-NGCs, we were not able to investigate the direct interaction between the regenerated nerve and the wall of the conduit. We will endeavor to carry this out in future studies.
In vitro tests performed on the PHA-NGCs depicted that the glial RT4-D6P2T cells showed a very well-organized morphology and proliferated efficiently on this regenerative substrate. In particular, the phalloidin staining allowed to highlight a marked organization of the actin filaments and an increase in the area occupied by the cells.  behavioral method allowed the qualitative and quantitative evaluation of the flexor function in rat median nerve. Grasping response is a complex sensory-motor response integrating sensory afferents with motor afferents through the cerebral cortex. 33 Flexion of the fingers depends on two nerves, the median and the tibial nerves located in the forelimb and in the hindlimb, respectively. 33 A common feature between humans and rats is the possession of prehensile forelimbs, forepaws and digits resulting in remarkable similarities between the two species. 34 These striking resemblances makes the grasping test a powerful translational tool. 35

| CONCLUSIONS
The results of the present study demonstrate, for the first time, that NGCs made using the bioresorbable PHA-blend 75:25 P(3HO)/ P(3HB) can successfully sustain cell proliferation and adhesion in vitro and nerve regeneration across a 10 mm median nerve defect in vivo. The conduit has proven to be biocompatible with the surrounding tissue, since no signs of inflammation or scar tissue formation were found. Also, our PHA-NGCs, with a diameter of 1.8 mm is suitable for human nerve size, especially for digital nerve repairs, which are the most frequently severed peripheral nerves. Moreover, this hollow NGC could provide an excellent scaffold to design and develop engineered nerve grafts used to repair longer nerve gaps in the future, together with gene/cell therapy approaches. Further investigation of this NGC will focus on optimization of the conduit structure and properties, including wall permeability and biodegradation in vivo.

ACKNOWLEDGMENTS
The authors would like to acknowledge the Department of Clinical