Preparation and assessment of an optimized multichannel acellular nerve allograft for peripheral nerve regeneration

Abstract Peripheral nerve regeneration after injury is still a clinical problem. The application of autologous nerve grafting, the gold standard treatment, is greatly restricted. Acellular nerve allografts (ANAs) are considered promising alternatives, but they are difficult to achieve satisfactory therapeutic outcomes, which may be attributed to their compact inherent ultrastructure and substantial loss of extracellular matrix (ECM) components. Regarding these deficiencies, this study developed an optimized multichannel ANA by a modified decellularization method. These innovative ANAs were demonstrated to retain more ECM bioactive molecules and regenerative factors, with effective elimination of cellular antigens. The presence of microchannels with larger pore size allowed ANAs to gain higher porosity and better swelling performance, which improves their internal ultrastructure. Their mechanical properties were more similar to those of native nerves. Moreover, the optimized ANAs exhibited good biocompatibility and possessed significant advantages in supporting the proliferation and migration of Schwann cells in vitro. The in vivo results further confirmed their superior capacity to promote axon regrowth and myelination as well as restore innervation of target muscles, leading to better functional recovery than the conventional ANAs. Overall, this study demonstrates that the optimized multichannel ANAs have great potential for clinical application and offer new insight into the further improvement of ANAs.

by a modified decellularization method. These innovative ANAs were demonstrated to retain more ECM bioactive molecules and regenerative factors, with effective elimination of cellular antigens. The presence of microchannels with larger pore size allowed ANAs to gain higher porosity and better swelling performance, which improves their internal ultrastructure. Their mechanical properties were more similar to those of native nerves. Moreover, the optimized ANAs exhibited good biocompatibility and possessed significant advantages in supporting the proliferation and migration of Schwann cells in vitro. The in vivo results further confirmed their superior capacity to promote axon regrowth and myelination as well as restore innervation of target muscles, leading to better functional recovery than the conventional ANAs.
Overall, this study demonstrates that the optimized multichannel ANAs have great potential for clinical application and offer new insight into the further improvement of ANAs.

| INTRODUCTION
Peripheral nerve injury (PNI) is the primary type of traumatic damage in the nervous system, affecting over one million people worldwide annually with a high prevalence among younger individuals. 1,2 Currently, autologous nerve grafting is still considered the "gold standard" for peripheral nerve repair. 3 However, the clinical application of autologous nerves is greatly restricted by the inherent drawbacks, such as limited availability, morbidity at donor site, and size mismatch. 4,5 In recent years, numerous tissue engineering nerve grafts made by various synthetic and natural biomaterials have been devised. 6,7 Because of the insufficient neurorestorative capacity of synthetic biomaterials, they are gradually replaced by natural biomaterials. 8,9 However, natural biomaterials often contain only one or a few purified extracellular matrix (ECM) components, thus they cannot completely reproduce the complex constitution of ECM related to multiple functions during regeneration. 10 This makes it difficult for artificial nerve grafts to establish an ideal repair microenvironment. Therefore, acellular nerve allografts (ANAs) have long gained a lot of attention and were regarded as one of the most promising alternatives to autologous nerves. 11 During the decellularization process, the cellular components capable of inducing deleterious immune responses are removed, but the heterogeneous compositions and inherent ultrastructure of ECM are preserved, which can provide a favorable substrate for nerve regeneration. 12 Several studies have reported that ANAs have consistently performed better than other types of artificial grafts. 13,14 So far, there have been commercial ANAs approved by China National Medical Products Administration and U.S. Food and Drug Administration. 15,16 A variety of physical, chemical, and biological approaches, as a single or combined treatment, have already been developed to prepare ANAs. 17 Among them, the detergent-based methods devised by Sondell et al. and Hudson et al. have been extensively recognized and studied. The Sondell method includes repeated treatments with Triton X-100 followed by sodium deoxycholate (SDC). This method was confirmed to clear cells and myelin sheaths effectively, and the yielded allografts achieved good nerve regeneration after transplantation. 18 Although nonionic detergent Triton X-100 is described to cause minimal damage to the native tissues, 19 SDC, an ionic detergent, is considered to be one of the most potent and disruptive chemical agents.
Subsequently, Hudson et al. proposed a milder chemical treatment consisting of several steps repeated twice as well. 20 The Hudson method has been shown not only to reliably remove cellular components, but more importantly, to achieve superior results in maintaining the ECM intact. This may be attributed to the introduction of the zwitterionic detergents sulfobetaine À10 and À16 in combination with the ionic detergent Triton X-200. The zwitterionic detergents have properties of both ionic and nonionic detergents, and cause mild damage to ECM during decellularization. 21 Unfortunately, Triton X-200, the key detergent of the Hudson method, has been discontinued from the only manufacturer, making it unavailable for further investigation. 22 Nevertheless, it is worth mentioning that this foundation leads to a wide range of work optimizing the decellularization of nerve grafts.
Currently, despite the Sondell method being proposed early, it is still a well-accepted conventional decellularization protocol and has been extensively used. 23 Nevertheless, it has to be mentioned that repeated rinsing in SDC for a long time contributes to the elimination of cellular antigens, but it will inevitably result in substantial loss of ECM components and biomolecules, such as collagen, laminin, fibronectin, and growth factors, which may significantly impair the neuroregenerative function of ANAs. 24 In addition, it is worth noting that the threedimensional organization within peripheral nerves is very compact with low porosity, and mainly consists of small basal lamina tubes. 21 However, the Sondell method is incapable of remodeling the inherent ultrastructure of native nerves. And the basal lamina tubes remaining within the conventional ANAs do not change remarkably after decellularization. 18,20 It was reported that the pore size of basal lamina tubes (5-10 μm) is similar to or even smaller than the typical size of regenerative cells (10-15 μm). The basal lamina tubes appear to be not large enough to meet the needs of abundant cell migration and axon extension, which may compromise nerve repair. 25 More than that, inadequate porosity of ANAs may restrain the diffusion of tissue fluids after transplantation. For this reason, Schwann cells generally fail to completely infiltrate into ANAs, resulting in markedly inferior outcomes. 26 At present, the conventional ANAs are believed to be incompetent in supporting nerve regeneration across long segmental defects (>3 cm). 27 Considering these drawbacks in the conventional ANAs, targeted optimization of nerve decellularization methods is a highly prospective research direction, which may open up further opportunities to enhance the proregenerative capacity of ANAs.
In this study, an innovative ANA with optimized axial multiple channels was prepared to deal with the existing problems of the conventional ANAs. A modified decellularization method was developed to obtain the multichannel allografts, which includes a combination of chemical, physical, and enzymatic treatments. To minimize the disruption of bioactive components during decellularization, this newly devised protocol starts with an osmotic shock generated by hypotonic and hypertonic solution, followed by alternating treatments of low concentration Triton X-100 and 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS). CHAPS, as a zwitterionic detergent, can remove cellular components efficiently with minor disruptive effects upon ECM similar to Triton X-100. 28,29 The aim of these gentle treatments is to better coordinate the balance of cellular antigen removal and ECM component preservation. Furthermore, to address the shortcomings of inadequate basal lamina tube size and low porosity of the conventional ANAs, unidirectional freeze-drying was introduced in the processing procedure. In our previous study, unidirectional freeze-drying has been shown to impart large axially aligned microchannels in ANAs, thereby improving their dense internal ultrastructure. 30,31 Notably, the freeze-drying technique itself can be used as a physical decellularization method, effectively lysing cells within tissues and organs. 32,33 Last, nuclease treatments were carried out to ensure adequate clearance of residual antigenic components within ANAs, allowing them to meet the criteria for in vivo transplantation. The novel ANAs prepared by the modified decellularization method are expected to yield an optimized multichannel architecture as well as achieve superior preservation of ECM components and regenerative factors, resulting in a significant enhancement in neurorestorative effects. As far as we know, there are only a few similar reports on multi-aspect optimization of ANAs to date. In the present study, the multichannel ANAs were thoroughly characterized on ultrastructural, biochemical, and biomechanical properties in comparison to the conventional ANAs prepared by the Sondell method. Furthermore, their biocompatibility and functionalities on peripheral nerve regeneration were investigated via an in vitro system and a rat sciatic nerve defect model.

| Animals and ethics
All animal work in this study was approved by the Institutional Animal Care and Use Committee (IACUC) of China Medical University (CMU2019192) and was performed in accordance with Guides for the Care and Use of Laboratory Animals from the Chinese Ministry of Public Health and U.S. National Institutes of Health. Adult male Sprague Dawley rats (250 ± 20 g) were obtained from Changsheng Bio Inc (Liaoning, China). Rats had access to 12-h light/dark cycles and standard water and food. They were transported and kept more than 7 days before surgery. All surgical procedures were performed under general anesthesia with intraperitoneal injection of sodium pentobarbital (40 mg/kg body weight).

| Nerve harvest and decellularization
Sciatic nerves (about 30 mm in length) were harvested aseptically. After the removal of adipose and connective tissues around the epineurium, nerves were then transferred to fresh phosphate buffered saline (PBS) and preserved at 4 C until further use (for maximum of 24 h).
The obtained nerves were trimmed into 10 mm segments and randomly assigned to three groups as follows.
Native nerves (Native): This group contains unprocessed sciatic nerves as a control for all analyses.
Sondell ANAs (S-ANA): As a standard for comparison, the allografts in this group were generated using the decellularization method previously proposed by Sondell et al. 18 (Figure 1a). Briefly, nerve segments were immersed in distilled water for 7 h, followed by two successive chemical extraction processes in 3% (v/v) Triton X-100 (Sigma-Aldrich, USA) for 12 h and 4% (w/v) SDC (Sigma-Aldrich) for 24 h at room temperature under constant agitation (120 rpm).
Multichannel ANAs (M-ANA): This group contains the allografts decellularized by the modified method devised in our lab (Figure 1a).
The nerve segments were first incubated in 6% (w/v) sodium chloride for 12 h, followed by rinsing in distilled water for 12 h. Next, these segments were treated with 2% (v/v) Triton X-100 for 12 h, followed by washing three times with distilled water (15 min each), then transferred to 6% (w/v) CHAPS (Sigma-Aldrich) for 12 h incubation.
This treatment was repeated with the two detergents. All procedures were at room temperature and under constant agitation (120 rpm).
After washing in distilled water (three times for 15 min each), unidirectional freeze-drying was conducted to form optimized multichannel ultrastructure. In this step, a custom-made silicon rubber mold and a stainless steel plate were applied. The mold serves to hold nerve segments perpendicular to the plate and insulate the surrounding heat transfer, thus ensuring the heat transfer along the long axis of nerves. The stainless steel plate was precooled (À80 C) to establish unidirectional temperature gradient inside the mold, which allows longitudinal ice crystals to form within nerves ( Figure 1b).
Briefly, the procedure is to first put nerve segments into the prefabricated grooves in one side of the mold, and then close the mold.
Subsequently, the mold containing nerve segments was placed on the precooled stainless steel plate, and rapidly transferred to a freezer (À40 C). After a 1 h hold period, the mold was moved to a freeze dryer and lyophilized for 24 h to remove ice crystals and generate microchannels ( Figure S1).
After rehydration in PBS overnight, the nerve segments were subjected to an enzymatic treatment with DNase (50 U/ml) and RNase (5 U/ml) at 37 C for 12 h under constant agitation (120 rpm).
After thoroughly washing in PBS (three times for 1 h each), the processed nerve segments of each group were stored in PBS at 4 C until use.

| Histological analysis
The samples of both Native and ANAs were fixed in 4% paraformaldehyde, dehydrated with ethanol, hyalinized with xylene, embedded in paraffin, and cut into 5 μm sections. Transverse sections were stained with hematoxylin and eosin (HE) to assess the removal of cellular components. Masson trichrome staining and Picrosirius Red (PSR) were conducted in longitudinal sections to analyze the morphological and componential characterization of collagen fibers.

| Biochemical analysis
Residual DNA content and size were measured to evaluate the extent of decellularization. Total DNA was extracted from samples using a genomic DNA extraction kit (Takara, Japan) and quantified using a (b) The schematic diagram of the ultrastructural optimization principle of unidirectional freeze-drying. The hematoxylin and eosin (HE) staining (ca, scale bar = 20 μm) was conducted to assess the removal of cell components. The Masson trichrome staining (cb, scale bar = 100 μm) and the Picrosirius Red (PSR) staining (cc, scale bar = 50 μm) revealed the collagen integrity and distribution after decellularization. The DNA remnants were analyzed further by agarose gel electrophoresis (d) and detection of DNA content (e). The detection of total collagen (f) and sulfated GAGs (g) illustrated the effects of different decellularization methods on ECM components. Data are expressed as the mean ± SD (n = 5). *p < 0.05 compared to native nerves, # p < 0.05 compared to S-ANA. NanoPhotometer N50 (IMPLEN, Germany). The size of the extracted DNA fragments was detected by 1% agarose gel electrophoresis.
The amount of collagen and sulfated glycosaminoglycans (GAGs) was measured to assess the abundance of ECM components. Total collagen content was determined by quantifying the hydroxyproline content with a hydroxyproline assay kit (Sigma-Aldrich). Because hydroxyproline is largely restricted to collagen, it serves as an indicator for the amount of collagen in tissues. Total collagen content was calculated based on a hydroxyproline-to-collagen ratio of 1:7.69. In addition, sulfated GAGs content was quantified by 1,9-dimethylmethylene blue dye-binding assay using a Blyscan kit (Biocolor, UK).
Each procedure was conducted according to the manufacturer's instructions. All biochemical measurements were normalized to tissue dry weight and five samples were used for each group.

| Western blot analysis
To evaluate the effects of different decellularization methods on the preservation of main ECM molecules, laminin, fibronectin, collagen I and IV in M-ANA and S-ANA were detected by immunoblotting.
Total protein from samples was extracted and quantified using a BCA protein kit (Thermo Fisher Scientific, USA). Equal amounts of protein were separated by electrophoresis on two sodium dodecyl sulfate polyacrylamide gels. One gel was stained with Coomassie Brilliant Blue R-250 to quantify total protein, serving as a control for loading.
The other one was transferred to a PVDF membrane. After blocking, the membrane was incubated with the anti-laminin, anti-fibronectin, anti-collagen I and IV antibody (1:1000, Abcam, USA) overnight at 4 C, followed by the second antibody (1:10000, Abcam) for 1 h at room temperature. Protein bands were visualized with an enhanced chemiluminescence kit (Thermo Fisher Scientific) and quantified by gray scale analysis using Image J (National Institutes of Health, USA).
All detections were independently repeated three times.

| Enzyme-linked immunosorbent assay
To analyze the effects of different decellularization methods on the retention of critical regenerative factors, enzyme-linked immunosorbent assays (ELISAs) were conducted to detect nerve growth factor (NGF), vascular endothelial growth factor (VEGF), and brain-derived neurotrophic factor (BDNF) in M-ANA and S-ANA. ELISA kits (Thermo Fisher Scientific) were used according to the manufacturer's instructions. The concentrations of NGF, VEGF, and BDNF were calculated from the standard curve. All detections were independently repeated three times.

| Scanning electron microscopy analysis
Scanning electron microscopy (SEM) was performed to reveal the ultrastructural features of each group. Samples were fixed in 2.5% glutaraldehyde for 2 h and postfixed in 1% osmic acid for 30 min at 4 C.
Then, samples were quenched in liquid nitrogen and broken to achieve fracture surfaces. After dehydration through a gradient of ethanol and critical point drying, cross sections were sprayed with platinum and observed with an SEM (VEGA3, TESCAN, Czech). SEM images were used to characterize internal morphology and distribution of pore size in each group. Five samples for each group were analyzed and five random high power visual fields were selected in each sample, located in the upper, lower, middle, left, and right part of the cross sections. Over 200 pores per group were measured to obtain distribution and average of pore size. The pore size was measured by Image-Pro Plus (Media Cybernetics, USA) and expressed as the mean of its shortest and longest diameters.

| Porosity test
The porosity of each group was determined by the gas-ethanol replacement method. Briefly, the lyophilized samples were immersed in definite volume (V 0 ) of anhydrous ethanol in a graduated cylinder for 10 min, the total volume of the ethanol-impregnated samples and ethanol was recorded as V 1 . Then, the samples were taken out and the volume of residual ethanol was recorded as V 2 . Five samples were examined for each group. The porosity was calculated by the following formula:

| Swelling test
The swelling behavior of each group was tested using a rehydration study. The samples were first lyophilized to obtain a constant dry weight (W 0 ). Then, the lyophilized samples were immersed in PBS for 24 h at room temperature. After taking the samples out of PBS, the excess water was removed by filter papers from the surface to achieve a constant swelling weight (W s ). Five samples were examined for each group.
The swelling ratio was calculated by the following formula:

| Biomechanical test
To assess the biomechanical properties of each group, a tensile test was carried out using a dynamic electromechanical testing instrument (Tianjin, China). Tensile uniaxial stress was applied parallel to the longitudinal axis of samples at a constant strain rate of 10 mm/min. Samples were kept moist during testing. Five samples were examined for each group.
To detect tensile properties, both ends of samples were mounted on custom clamps fitted with sandpapers, leaving a distance of 10 mm between the clamps. Each sample was stretched to complete tensile failure.
Young's modulus, stress at fracture, and strain at fracture were measured.
To measure suture retention strength, every sample was sutured between two stumps of fresh rat sciatic nerves with 8-0 nylon sutures. The suture was pierced through epineurium at 1 mm from the edge and knotted at least seven times to ensure no slippage. The other side of the fresh nerve was clamped into the testing instrument.
Suture retention strength was defined as the maximum load when sutures were pulled out of epineurium.  Table S1. All detections were independently repeated three times.

| Gait analysis
Gait analysis was performed to evaluate motor functional recovery.
After dipping their hind feet in ink, the rats of each group were placed in a clear corridor with a dark shelter at the end. Paw length (PL), toe spread (TS), and intermediary TS (ITS) of the normal (N) and experimental (E) feet were measured. The sciatic functional index (SFI) was calculated according to the following formula 34 :

| Electrophysiological analysis
Electrophysiological analysis was conducted to assess nerve functional recovery by the measurement of compound muscle action potentials (CMAPs). The surgical sites of sciatic nerves were exposed under anesthesia. The electrical stimuli (stimulus intensity = 6 mA; frequency = 1 Hz; duration = 1 ms) were applied to the proximal and distal ends of each graft sequentially, and the CMAPs of ipsilateral anterior tibial muscle were recorded. CMAP amplitude was expressed as the ratio of surgical side to normal side. Motor nerve conduction velocity (MCV) was calculated based on latency and the distance between the two stimulating sites.

| Muscle wet weight evaluation
After the electrophysiological test, the rats were euthanized by sodium pentobarbital overdose. Then, the bilateral anterior tibial muscles of each group were carefully dissected and immediately weighed to evaluate muscle recovery. The ratio of muscle wet weight was expressed as the percentage of surgical side to normal side.

| Toluidine blue staining
Toluidine blue staining was performed to assess newborn axon myelination. The nerve segments of central grafts and distal stumps were fixed in 2.5% glutaraldehyde, followed by postfixation in 1% osmium tetroxide. After dehydration in a graded series of ethanol, the processed segments were embedded in Epon 812 epoxy resin.
Transverse semi-thin sections of central grafts and distal stumps were cut using an ultramicrotome and stained with toluidine blue.
The stained sections were observed under a light microscope (BX53, Olympus, Japan), and five random fields were selected for each semi-thin section to determine the density of myelinated axons by Image J.

| Statistical analysis
Data were presented as mean ± standard deviation (SD) and were statistically analyzed using GraphPad Prism 5.0 (GraphPad Software, USA). Differences between two groups were analyzed with Student's t test, and differences among three or more groups were analyzed with one-way or two-way ANOVA followed by Tukey's post hoc test for multiple comparisons. A value of p < 0.05 was considered statistically significant.

| ECM morphological characterization
The ECM morphology of ANAs was characterized by Masson trichrome staining, PSR staining, and SEM (Figures 1c and 3a). Native Collagen I, S-ANA exposed more Collagen III after decellularization.
In M-ANA, it was observed the presence of the optimized multichannel ultrastructure. The collagen fibers exhibited a loose distribution. There were many continuous microchannels aligned axially between parallel fibers. Similar to S-ANA, a large amount of Collagen III can be seen uniformly distributed around Collagen I in M-ANA.

| ECM componential characterization
To assess the remaining ECM components after decellularization, this study has quantitatively examined the amount of total collagen and

| Swelling characterization
The swelling characterization was analyzed to assess the ability of water absorption. In comparison with Native, M-ANA and S-ANA achieved a significantly higher swelling ratio following processing (p < 0.05). Moreover, the swelling ratio was remarkably higher in M-ANA than in S-ANA (p < 0.05) (Figure 3e).

| Biomechanical properties
The biomechanical test was conducted to determine the changing properties of ANAs following different decellularization processes.
The Young's modulus and stress at fracture in M-ANA were not markedly different from those in Native (p > 0.05). However, S-ANA exhibited a significant increase compared to Native (p < 0.05), and was significantly higher than M-ANA (p < 0.05). In addition, there was no significant difference in the strain at fracture and suture retention strength among the three groups (p > 0.05) (Figure 4).

| Schwann cell penetration analysis
The penetration behavior of Schwann cells was analyzed to determine the effects of optimized ultrastructure on the functional characteristics of ANAs. After culturing for 3 days, no significant cell penetration

| Motor function evaluation
SFI is a key indicator to assess the motor functional recovery following repair. It was found that each group showed an improvement in SFI with time. However, no obvious difference among ANG, S-ANA, and M-ANA was observed over the first 6 weeks (p > 0.05). From 8 weeks, M-ANA exhibited higher SFI than S-ANA (p < 0.05), but was still significantly lower compared to ANG (p < 0.05) (Figure 6b).

| Electrophysiological analysis
Electrophysiological analysis allows for a meaningful assessment of the effective reinnervation of grafts able to deliver electrical stimuli to the target organs. At 6 weeks, the MCV of M-ANA was significantly superior to that of S-ANA (p < 0.05), and showed no significant difference compared to ANG (p > 0.05) (Figure 6c). In terms of CMAP amplitude, there was no obvious difference between M-ANA and S-ANA (p > 0.05), and they were significantly lower than ANG (p < 0.05). At 12 weeks, the MCV and CMAP amplitude of M-ANA were remarkably greater than those of S-ANA (p < 0.05), but were still inferior to those of ANG (p < 0.05) ( Figure 6d).

| Muscle recovery evaluation
The wet weight ratio of anterior tibial muscle reflects the degree of muscle recovery after repair. The results showed that the wet weight ratios of all groups increased over time. At 6 weeks, there was no significant difference among the three groups (p > 0.05). At 12 weeks, M-ANA had a higher wet weight ratio compared to S-ANA (p < 0.05), but was significantly lower than ANG (p < 0.05) (Figure 6e).

| Analysis of axon regrowth and remyelination
Axon regrowth in newborn nerves was analyzed by immunofluorescence detection labeled with NF200 and S-100 (Figure 7a). At

| DISCUSSION
ANAs are considered the potentially ideal nerve grafts for PNI treatment due to the preservation of ECM components and structure, as well as low immunogenicity. 35 However, many current decellularization methods rely on harsh detergents to extract cellular antigens, which are destructive to the tissue matrix. 36  The immunogenicity of allogeneic nerves is mainly derived from cells. The immunogenicity of neural ECM is extremely low or even negligible. 43 The effective clearance of cellular components is a requisite for the preparation of ANAs. In this study, our new protocol achieved excellent decellularization outcomes similar to the Sondell method. M-ANA exhibited an absence of cell nuclei and over 95% reduction in DNA content to less than 50 ng/mg dry weight compared to native nerves, with no DNA fragments >200 bp in length, which meets the criteria for acellular tissues proposed by Crapo et al. 44 The presence of well-distributed essential ECM in ANAs provides the optimal substrate for Schwann cell migration and subsequent axon regrowth. 45 Therefore, the preservation of ECM molecular composition is a critical requirement for decellularization. Collagen, as a major structural protein, constitutes about 49% of the total proteins of peripheral nerves. The ECM derived from nerves is predominantly composed of Collagen I that is involved in the fibril formation. 46 while maintaining ECM integrity. 49,50 GAGs are important ECM components in the peripheral nervous system that are implicated in the regulation of axon growth and guidance. 51 The sulfated GAGs are covalently bound to core proteins of chondroitin sulfate proteoglycans (CSPGs) in their natural state. 52 Due to their large size and negative charge, these sulfated GAGs can bind to growth-promoting ECM molecules and suppress their function, which imparts axonal inhibitory activity to CSPGs. 53 In the present study, the biochemical quantification revealed significant decreases of sulfated GAGs in M-ANA and S-ANA compared to native nerves. However, there were more sulfated GAGs found in M-ANA than in S-ANA, suggesting that the sulfated GAGs removal efficiency of the modified decellularization method is inferior to the conventional method, probably due to the application of SDC in the Sondell protocol. SDC, as a potent detergent, could lead to greater disruption of matrix components, especially GAGs. 54 In the study of Philips et al., it can be observed that the decellularization method they devised incorporating Triton X-100, DNase, and RNase but without SDC retained more sulfated GAGs than the Sondell method as well. 54 Notably, although previous studies have demonstrated that the reduction of sulfated GAGs in ANAs by enzymatic treatment facilitated the repair of damaged nerves, excess degradation of sulfated GAGs has been noted to cause uncontrolled axonal sprouting outside nerve fascicles, resulting in inappropriate reinnervation. [55][56][57] Moderate removal of sulfated GAGs may be a more feasible strategy to optimize ANAs rather than complete GAGs elimination.
The compact ultrastructure of ANAs derived from natural nerves may be optimal for supporting and protecting fully grown nerve fibers, but is theoretically less ideal for PNI repair. 25 We have previously ANAs should provide sufficient mechanical strength for fixation and resistance to external forces. It should be noted that the mechanical environment can influence the behavior and functionality of neural cells within ANAs. It is necessary to match the approximate mechanical properties of target tissues. 63,64 The biomechanical analysis revealed that Young's modulus and stress at fracture of M-ANA appeared to be elevated compared to native nerves, but not in a significant manner. However, these parameters in S-ANA exhibited significantly higher levels than native nerves, suggesting an obvious difference in the effects of these two decellularization methods on the mechanical properties of allografts. Collagen fibers play a vital role in nerve elasticity. 65 After decellularization, collagen fiber network would lose their intrinsic wave-like pattern, resulting in the relaxation of collagen fibers. 66,67 Moreover, the removal of cells could enhance the mobility of collagen fibers, which allows them to reorient toward the direction of applied strain. 68 These lead to changes in the mechanical features of ANAs. Previous studies also reported increased stiffness and ultimate stress in the allografts produced by the Sondell method. 69 This may also be closely related to the disruption of ECM due to long-time repeated treatments of SDC, especially the massive loss of GAGs. 54 In contrast, the modified decellularization method was able to better maintain ECM and generate the multichannel ultrastructure, which endows the prepared allografts with mechanical performance closer to native nerves. Notably, the ultrastructure remodeling did not significantly alter the ability of ANAs to withstand suturing. The suture retention strength of M-ANA was higher than the generally accepted adequate suture strength for implantation (2 N), 70 with no significant difference to S-ANA and native nerves.
Nerve regeneration takes place through the extension of Schwann cells rather than axon growth. 23 Schwann cells guide and support the newborn axons, and exert key regulatory effects in the recovery process. 71 The competence to support the viability and bio-  74 Thicker myelin sheaths give rise to faster conduction. In addition, given that there is a trade-off between myelin thickness and axon diameter due to spatial constraints, the parameter G-ratio, reflecting their relationship, is also one of the reliable predictors to evaluate axon remyelination. 75,76 G-ratio is informative of the underlying myelin ultrastructure, and could indicate the relative efficiency and maximal conduction velocity of axons. 77 The optimal G-ratio is thought to be around 0.6 in the peripheral nervous system. 78 Myelinated axons with G-ratio close to the ideal value may achieve higher conduction velocity and lower energy consumption, which contributes to the recovery of nerve function after injury. 79 Indeed, MCV is demonstrated to be related to the myelination rate and regenerative axon level. 80 The electrophysiological results revealed that M-ANA did show a significant increase in MCV, closely resembling ANG, which is consistent with the histomorphological results. Notably, the ANAs produced using the Hudson method that better preserves ECM components and architecture were shown to have no significant advantage over S-ANA in the early stage of nerve repair. 23  However, we also identified that there still exist gaps in regenerative performance between the optimized ANAs and autologous nerves. It requires a long way to transfer from bench to bedside. The novel ANAs can be recellularized to achieve further functionalization.
The integration of neurotrophic factors may also be a feasible solution for improvement. In this study, unidirectional freeze-drying was employed to impart axially aligned channels to ANAs. This multichannel structure could be further manipulated by modifying the freezing process, such as altering cold source and regulating temperature gradient, which may yield different impacts on nerve repair. More studies are required to determine the optimal ultrastructure for a growth-permissive microenvironment. Moreover, the variations in nerve dimension and structure between different species are nonnegligible factors, which can greatly affect the successfulness of the decellularization process. This newly developed decellularization protocol should still be adapted and optimized to meet the needs of preparing human ANAs.

| CONCLUSIONS
In conclusion, we developed a novel ANA with optimized multichannel ultrastructure using a modified decellularization method described in this study. These unique ANAs had significant advantages in the retention of main ECM bioactive molecules and regenerative factors, while eliminating over 95% of cellular components. The presence of the microchannels with larger pore size allowed them to obtain higher porosity and swelling