Blood vessel remodeling in late stage of vascular network reconstruction is essential for peripheral nerve regeneration

Abstract One of the bottlenecks of advanced study on tissue engineering in regenerative medicine is rapid and functional vascularization. For a deeper comprehension of vascularization, the exhaustive, dynamic, and three‐dimensional depiction of perfused vascular network reconstruction during peripheral nerve regeneration was performed using Micro‐CT scanning. The 10 mm defect of sciatic nerve in rat was bridged by the autologous or tissue engineered nerve. The blood vessel anastomosis between nerve stumps and autologous nerve accomplished at 4 days to 1 week after surgery, which was a sufficient basis for the mature vascular network re‐establishment. The stronger ability for sprouting angiogenesis and vascular remodeling of autologous nerve compared with tissue engineered nerve was revealed. However, common phases of vascularization in peripheral nerve regeneration were painted: hypoxic initiation, sprouting angiogenesis, and remodeling and maturation. The effect of less‐concerned vascular remodeling on nerve regeneration was further analyzed after nerve crush injury. The blockage of vascular remodeling in late stage by VEGF injection significantly inhibited axons and myelin sheaths regeneration, which attenuated the impulse conduction toward reinnervated muscles. It was illustrated that a large amount of immature blood vessels rather than necessary vascular remodeling elevated local inflammation level in nerve regeneration microenvironment. The figures inspired us to understand the close connections between vascularization and peripheral nerve regeneration from a broader dimension to achieve better constructions, regulations and repair effects of tissue engineered nerves in clinic.

nerve stumps and autologous nerve accomplished at 4 days to 1 week after surgery, which was a sufficient basis for the mature vascular network re-establishment. The stronger ability for sprouting angiogenesis and vascular remodeling of autologous nerve compared with tissue engineered nerve was revealed. However, common phases of vascularization in peripheral nerve regeneration were painted: hypoxic initiation, sprouting angiogenesis, and remodeling and maturation. The effect of lessconcerned vascular remodeling on nerve regeneration was further analyzed after nerve crush injury. The blockage of vascular remodeling in late stage by VEGF injection significantly inhibited axons and myelin sheaths regeneration, which attenuated the impulse conduction toward reinnervated muscles. It was illustrated that a large amount of immature blood vessels rather than necessary vascular remodeling elevated local inflammation level in nerve regeneration microenvironment. The figures inspired us to understand the close connections between vascularization and peripheral nerve regeneration from a broader dimension to achieve better constructions, regulations and repair effects of tissue engineered nerves in clinic.

K E Y W O R D S
blood vessel three-dimensional reconstruction, inflammation, peripheral nerve regeneration, tissue engineered nerve, vascular remodeling, vascularization researches is not only to repair the defect morphologically, but also to restore the physiological function, in which vascular network reconstruction is a premise and the basis for graft survival. So far, one of the key points and bottleneck problems is the vascularization of tissue engineered tissues and organs. [6][7][8][9] Blood vessels are regenerated, once the maximum free diffusion distance of oxygen about 200 μm is exceeded. [10][11][12] Therefore, the rapid establishment of an effective blood circulation with the surrounding tissues after implantation is crucial for the successful regeneration.
The same is true for peripheral nerve regeneration studies. Construction strategies of tissue engineered nerves have developed from only biomaterial nerve conduits to tissue engineered nerves involving supporting cells, growth factors or cell matrices. [13][14][15][16][17] Unfortunately, in both animal experiments and clinical trials, repair effects of tissue engineered nerves under existing strategies are difficult to fully achieve or even surpass those of autologous nerves. Autologous nerve transplantations are still the gold standard for clinical peripheral nerve defect repairs. [18][19][20][21] It was confirmed that migrations of Schwann cells were guided by vascular endothelial cells in peripheral nerve regeneration, which proved the close relationship between angiogenesis and nerve regeneration in morphology. 22 Meanwhile the intimate collaboration between peripheral nerve regeneration and angiogenesis at the molecular regulation level was demonstrated. 23 It is widely realized that vascular network reconstruction is an important part of nerve regeneration microenvironment, deficiency of which is the key constraint for advanced improvement of repair effects of tissue engineered nerves. 24,25 Few basic and systematic researches on vascularization in peripheral nerve regeneration were performed. Alternatively, most studies focused on application explorations. Supplements of proangiogenic factors or incorporations of endothelial cells and stem cells were adopted to promote the neovascularization of tissue engineered nerve. [26][27][28][29][30][31] Yet it is far from enough to only conduct application explorations for breakthrough of angiogenesis bottleneck to more effectively promote peripheral nerve regeneration. How vascular networks are reconstructed step by step in peripheral nerve regeneration? Are the more blood vessels the better peripheral nerve regeneration in different phases? What are the differences of vascularization in nerve regeneration between tissue engineered and autologous nerve grafts? Lots of important processes and key regulations are still unclear. Fundamental researches in-depth are required to reveal the entire vascular network reconstruction during peripheral nerve regeneration including details, as well as its effects on regeneration microenvironment.
Vascularization of autologous nerve repair, as a gold standard in clinic, has important inspirations and great references. In our research, the tissue engineered nerve was constructed with Schwann cells differentiated from skin-derived precursors (SKP-SCs) as supporting cells and chitosan nerve conduits combined with silk fibroin fibers as scaffolds. Allogeneic cells improve the local nerve regeneration microenvironment by secreting a variety of growth factors during their survival period, thereby promoting peripheral nerve regeneration well. The sciatic nerve defect in rat was bridged by the tissue engineered nerve or the autologous nerve to observe the vascularization in peripheral nerve regeneration. Based on the detailed and objective depiction, the main stages of vascular network reconstruction were further analyzed to reveal the important influences on peripheral nerve regeneration. The results painted a panoramic picture of the similarities and differences in vascular network reconstruction by different repair methods. More importantly, it is recognized that vascularization in peripheral nerve regeneration is quite complicated, and each stage directly play an important role on nerve regeneration. Especially, the tight link between nerve regeneration and vascular remodeling was identified. This study provides a more comprehensive and in-depth understanding of vascularization and microenvironment in peripheral nerve regeneration, which is the theoretical basis and new inspiration to precisely regulate the construction of vascularized tissue engineered nerve for better clinical peripheral nerve injury repair.

| Tissue engineered nerve construction
Tissue engineered nerves were constructed in vitro including chitosan nerve conduits inserted with silk fibroin fibers as scaffolds and SKP-SCs as supporting cells. In brief, the chitin/chitosan (Nantong Xincheng Biochemical, Nantong, China) mixtures were injected into stainless-steel casting molds, which were then sealed and placed at À12 C for 2-4 h. Then the conduits were lyophilized under a 35-45 mTorr vacuum for 20 h after rinsing. The porous chitosan conduits were 2 mm inner diameter, 3 mm outer diameter. 32 Skin-derived precursors (SKPs) of newborn Sprague-Dawley (SD) rats were isolated and differentiated to SKP-SCs, then amplificated in vitro. 33,34 The SKP-SCs and scaffolds were co-cultured with 37 C and 5% CO 2 for sufficient contact. Briefly, SKP-SCs resuspended in Dulbecco's modified Eagle's medium (DMEM) were added to the silk fibroin fibers and chitosan nerve conduits followed initial adhesion for 4-6 h, which was performed again after the scaffolds were turned over. The artificial nerves and SKP-SCs were co-cultured in DMEM for 2 days and then in DMEM supplemented with 50 μg/ml ascorbic acid (Sigma) for additional 12 days. The cells homogeneously adhered to the surface of fiber and conduit. The final cell density was 10 6 /ml. 35 Tissue engineered nerves were stored in NS following rinse twice with NS.  Table 1). The rats were housed in a temperature-controlled environment and allowed food and water ad libitum. The Administration Committee approved all experimental protocols of Experimental Animals, following the guidelines of the Institutional Animal Care and Use Committee, Nantong University, China (Inspection No: 20180301-009). The SD rats were deeply anesthetized with an intraperitoneal injection of a compound anesthetic. 20 The skin and muscle were incised to expose the sciatic nerve at the left mid-thigh. An 8 mm segment of the sciatic nerve was resected to produce a 10 mm gap after slight retraction. 36 The nerve defect was bridged by the tissue engineered nerve (TEN) or autologous nerve with a 180 reversal of the ipsilateral sciatic nerve after dissection (Auto). Compared with the defect model, the expected good regeneration of crush model is more convenient for the nerve regeneration study. For the nerve crush model, the sciatic nerve was exposed carefully and crushed 3 mm for 30 s in the middle segments using hemostatic forceps. The 6 μl of 1 μg/ml vascular endothelial growth factor (VEGF) recombinant protein was injected perineurally at the injury site immediately after nerve injury. In the control group, to exclude possible influences of the injection operation and pressure on nerve regeneration, the 6 μl of saline was also injected perineurally. 37 Then, the muscle layer and skin were closed with sutures. After surgery, the animals were placed in warmed cages.

| Electrophysiology detection
Under deep anesthesia, the crushed sciatic nerve was re-exposed.
Electrical stimuli were applied to the nerve trunk at distal and proximal ends of crushed segment sequentially. Compound muscle action potentials (CMAPs) were recorded on the gastrocnemius belly. The detection of normal CMAPs was performed at the uninjured contralateral side. 38

| Stereoscopic observation and histological assessment
The rats were deeply anesthetized with an intraperitoneal injection after surgery at different time points. The tissue engineered nerves or autologous nerves were fully dissociated and exposed. The bridge segments were placed in the vision of stereomicroscope (AZ100, Nikon). The surface blood vessels were focused and photographed. In addition, the tissue engineered nerves, autologous nerves and crushed nerves were harvested, fixed and frozen sliced into cross sections followed by hematoxylin-eosin (HE) staining. The regenerated nerves after crush were also performed to a special trichrome staining, for which three main dyes (hematoxylin, Fast Green FCF and Chromotro-pe2R) were used. 16 The slices stained by HE and trichrome were observed and photographed in light microscope (AxioImager M2, Zeiss). The three random fields with high magnification (Â400) of crushed nerves per animal were selected for the inflammation analysis.

| Immunofluorescence
The sections in the middle segments of crushed nerves were blocked with 5% goat serum for 1 h at 37 C, incubated with primary anti-   Nuclei were marked using Hoechst 33342 (1:5000 dilution, Life Technologies). Images were acquired under fluorescence microscopy (Zeiss). The three random fields with high magnification (Â400) per animal were selected for the statistical analysis.

| Transmission electron microscope
The regenerated nerves of crushed segments were collected, postfixed in 4% glutaraldehyde, and embedded in Epon 812 epoxy resin (Sigma). 19 Ultrathin sections were obtained and stained with lead citrate and uranyl acetate. The morphology of nerves was observed under a transmission electron microscope (JEOL Ltd., Tokyo, Japan). The three random fields with low magnification (Â1.2 k) per animal were used for myelinated nerve fiber statistics involving the g-ratio (nerve axon diameter/ nerve fiber diameter), average diameter of axons and average thickness of myelin. The five random fields with high magnification (Â20.0 k) per animal were selected for the analysis of myelin sheath layers.

| Statistical analysis
The data were presented as means ± SD. The sample sizes for statistical analysis were displayed in Table 1. Comparisons between two groups were carried out with Student's t test using Graph-Pad Prism

| Sprouting angiogenesis based on vessel reuse of the autologous nerve
A small amount of sprouting growth of blood vessels was observed from both sides of nerve stumps at 1 day after the nerve defect bridged by an autologous nerve (Figure 3a). Immediately at 4 days to 1 week after surgery, the blood vessels of bilateral nerve endings had anastomosed with those originally in bridge segment that were longitudinally parallel blood vessels rather than new sprouting micro vessels to establish an effective circulation quickly (Figure 3a).  1c and 3c). The diameter distribution at 4, 8, and 12 weeks of later stages was closer to the normal nerve than that at 1 day, 1 week, 2 weeks and 3 weeks (Figures 1c and 3c).The surface and inner blood vessels of the autologous nerve in vascularization were also displayed by stereoscope and HE staining. The reperfusion of blood vessel in graft segment was demonstrated at 4 days and 1 week of the autologous nerve (Figure 3e). In terms of the quantity, connectivity, and spatial distribution except for diameter, the tissue engineered nerve was significantly different from normal nerve, while the autologous nerve was close to normal. Despite the numerous differences in detail, it was a common process including a stage with blood vessel increase followed by a stage with blood vessel decrease. Three phases of vascular network reconstruction were summarized: hypoxic initiation, sprouting angiogenesis, and remodeling and maturation ( Figure 4g). Particularly, blood vessel remodeling was a late stage of vascularization in peripheral nerve regeneration, which co-existed in two repair methods.

| Inhibition of peripheral nerve regeneration following vascular remodeling blockage
Is the vascular remodeling in late stage of vascularization during peripheral nerve regeneration only a natural and insignificant continuation of sprouting angiogenesis, or is it also directly related to nerve regeneration? The influence of peripheral nerve regeneration after vascular remodeling blockage was further observed and analyzed ( Figure 5a). A remodeling process with reduced blood vessels after nerve crush injury was also displayed (Figure 5b). The number of blood vessels was continuously maintained at a high level after VEGF injection, which was significantly different from that of the saline injection (Figures 5c,d and S1). Meanwhile, it was noted that a large number of blood vessels that disrupted the vascular remodeling were mostly small or immature vessels without surrounding smooth muscle cells (Figure 5h). On the contrary, the blood vessels in late stage of nerve regeneration in the control group showed greater maturity ( Figure 5h). Then the nerve regeneration with vascular remodeling blockage was compared and analyzed. Regarding the number and size of regenerated axons and myelin sheaths, the loss of vascular remodeling severely inhibits nerve regeneration (Figure 5c and S1). It was calculated that the number of axons, areas of axons and myelins subjected to VEGF treatment were significantly less than those in the control group (Figure 5c, 5e-g). The ultrastructure of regenerated nerve further illustrated the close connection between vascular remodeling and nerve regeneration. The average myelin thickness and number of myelin layers were obviously reduced caused by the loss of vascular remodeling (Figure 6a-d, f,g). In addition, the inhibition of vascular remodeling directly affected the average diameter of regenerated axons, although the difference between two groups was not statistically significant (Figure 6a-d, e). The difference of g-ratio was also revealed, which had no statistical significance due to the simultaneous decrease in axon diameter and myelin thickness (Figure 6h).
Correspondingly, the effect of vascular remodeling disruption on nerve regeneration was also reflected in the target muscle reinnervation. The significant differences in CMAPs of gastrocnemius between the two groups were detected ( Figure S2).
The representative photographs of neural trichrome staining were displayed, and the fields inside the rectangle were magnified. Due to the local injection of pro-angiogenic factors, the number of new blood vessels in the VEGF group was significantly higher than that in the control group. Compared to the control group, the regenerated nerve fibers were significantly reduced in both number and size in the VEGF group. Myelins appear red. Axons and connective tissues appear green. Nuclei appear purple-blue. Arrows indicated the blood vessels.

| Higher inflammation of regenerative microenvironment due to continuous sprouting angiogenesis
The level of inflammation in peripheral nerve regeneration microenvironment is critical to nerve regeneration outcome. The number of main inflammatory cells of the regenerated nerves in crushed segments was counted and analyzed. First of all, the total number of macrophages displayed significant differences between two groups (Figure 7a,c). More M1 macrophages existed in the VEGF group, although the difference from that in the control group was not statistically significant (Figure 7a,d). Whereas, the significant decrease in the number of M2 macrophages in the VEGF group resulted in more total macrophages in the control group (Figure 7a,e). Hereafter, the in-depth analysis of two subtypes of macrophages corresponding to the inflammatory or regenerative environment was conducted. [39][40][41] The results indicated that the loss of vascular remodeling significantly increased the ratio of M1 macrophages, while the ratio of M2 macrophages was significantly lower than that of the control group

| DISCUSSION
The entire progress including details of vascular network reconstruction during peripheral nerve regeneration is quite unclear, which restricts the further improvement of tissue engineered peripheral nerve. This is the reason that prompted us to carry out such comprehensive and systematic basic research. It is believed that nerve regeneration and vascularization are closely linked. 23,38 Although the addition of more factors may increase the threshold of clinical applications, tissue engineered nerves incorporating allogeneic SKP-SCs were designed to paint the vascularization process due to the potential to better repair peripheral nerve defects. 42 The blood vessel threedimensional reconstruction by Microfil perfusion and Micro-CT scanning was selected, which was widely used in vascular research. [43][44][45] The advantage of Microfil contrast agent lies in its good filling of blood vessels and its small shrinkage after curing. 35 Scale bar, 20 μm. (c) Immunofluorescence staining of the regenerated nerves with VEGF or saline injection. A large number of new blood vessels continued to sprout due to VEGF injection, which led to significant reduction in peripheral nerve regeneration. Blood vessels were red (CD34+). Axons were white (NF+). myelins were green (S100+). Nuclei were stained using Hoechst (blue). Scale bar, 200 and 20 μm, respectively. (d-g) Histograms of the area of blood vessels, number of axons, area of axons and area of myelins (n = 3). There were significant statistical differences between the control and VEGF groups. **p < 0.01. (h) Immunofluorescence staining of the smooth muscle of blood vessels. The number of blood vessels surrounding smooth muscles in control group was more than that in VEGF group. Blood vessels were red (CD34+). smooth muscles were green (α-SMA+). Nuclei were stained using Hoechst (blue). Scale bar, 50 and 20 μm, respectively.
reconstruction of functional blood perfusion during nerve regeneration was revealed by the Microfil perfusion.
It was reported that increased angiogenesis leads to improved nerve regeneration to some extent in a few studies. 48-50 Based on the description of vascular network of normal nerve, the process of vascularization in peripheral nerve regeneration was analyzed and compared between the tissue engineered nerve and autologous nerve. The different patterns of vascular network reconstruction in two repair approaches were revealed in details. The anastomosis of blood vessels and the recanalization of blood flow of autologous nerve were fairly rapid at 4 days to 1 week after surgery, which was the guidance for Schwann cell migration and axon extension. 22 Meanwhile it was interesting that the blood vessel volume of autologous nerve at the time of vascular anastomosis at 4 days was larger than that of the normal nerve, suggesting that there may be a stage of vascular cavity expansion, whose role and significance need further research. Moreover, blood vessels of autologous nerve at 1 week were less than normal revealing that maybe just the part of original blood vessels anastomosed to be the basis of sprouting angiogenesis, which is another unknown and complex question. Compared with the tissue engineered nerve, the reuse of original blood vessels of autologous nerve aroused more rapid, efficient and mature vascular network reconstruction. It is likely that the differences in vascular network reconstruction are one of key factors affecting repair effects of tissue