Cerebral Injury Promotes Regeneration of The Brachial Plexus


 Background: Our previous trial of contralateral seventh cervical nerve transfer (CC7) for spastic arm paralysis suggested that regeneration of the C7 nerve seems to be faster in patients undergoing nerve transfer due to cerebral injury compared to patients receiving surgery due to brachial plexus injury (BPI). This finding needs to be further verified, and the underlying mechanism remains largely unknown. The present study compared C7 regeneration between two groups of patients and animal models. Proteomics was utilized to reveal the mechanism mediating the promotion of C7 nerve regeneration.Methods: We assessed Tinel’s sign in the arm after C7 transfer and compared the latency to the appearance of Tinel’s sign to evaluate regeneration. Traumatic brain injury (TBI) and BPI were induced in C57 mice to establish injury models. C7 regeneration was assessed by electrophysiology and histology. Enrichment analysis of the differentially expressed proteins identified by proteomics suggested altered inflammation related to TBI. qPCR and histology were performed to assess the inflammatory environment in the C7 nerve. We evaluated the influence of the contributing factor serum amyloid protein A1 (SAA1) on the regenerative ability and inflammatory response of the C7 nerve by electrophysiology, qPCR and histology.Results: Faster C7 regeneration was identified in patients, which was further confirmed in mouse models by electrophysiological recordings and histology. Altered systemic inflammation, which led to increased M2 macrophage activation, may represent an underlying mechanism of increased regeneration. In mice, SAA1 facilitated C7 regeneration and interfered with macrophage polarization in vivo.Conclusions: Altered inflammation promoted the regenerative capacity of the C7 nerve by altering macrophage behavior, and SAA1 may be a therapeutic target to improve the recovery of injured peripheral nerves.


Background
The peripheral nervous system can be affected by different types of injuries, including blunt trauma, traction injury, freeze injury, and chemical injury, which give rise to functional losses and even disability (1). Although axons in the peripheral nervous system have a remarkable capacity for regeneration, clinical improvement consistently requires a long period of time, and functional recovery is usually incomplete (2). Even after peripheral nerve surgery, approximately 25% of patients experience reduced quality of life and remain out of the workforce for more than 1.5 years (3). In our previous study, regeneration of the seventh cervical nerve (C7) seemed to be faster was in patients subjected to contralateral C7 transfer (CC7) for the treatment of spastic arm paralysis due to cerebral injury (4) compared to those who received C7 transfer due to brachial plexus injury (BPI) (5,6). In a rat model, craniocerebral injury was reported to promote regeneration of the sciatic nerve ganglia after nerve transection (7). Based on this interesting nding, we aimed to decipher how cerebral injury contributes to peripheral nerve regeneration.
In response to injury, myelinating Schwann cells dissociate from distal nerve segments. While degenerated nerve fragments are cleared by recruited neutrophils and macrophages, Schwann cells adopt a repair phenotype and work in concert with phagocytes. As degenerated substances are inhibitory to nerve regeneration, the timely removal of myelin and axonal debris is essential for optimal regeneration (8) . Although the in ammatory response is required for the timely removal of myelin and axonal debris, which inhibit axonal outgrowth, optimal nerve regeneration also requires good resolution of so-called neuroin ammation. During this period, macrophages in the distal segment shift from the M1 to the M2 phenotype, which is anti-in ammatory and promotes repair (9,10).
Serum amyloid A (SAA) refers to a group of proteins consisting of four members. Induced expression of SAA1 and SAA2 by hepatocytes accounts for the majority of the increase in SAA levels in the plasma following stroke and traumatic brain injury (TBI). In contrast, SAA3 is the major form of SAA in in ammatory sites, and SAA4 is constitutively expressed and is not inducible (11). In the traditional view, SAA is a proin ammatory molecule that is induced by in ammatory cytokines, including IL-1β and IL-6, and exerts physiological activity by interacting with multiple proin ammatory receptors, such as FPR2, TLR2 and TLR4. This concept was largely challenged by the fact that SAA not only upregulates the expression of anti-in ammatory cytokines (IL-10 and IL-1rn) but also polarizes macrophages toward the M2 phenotype. To date, substantial evidence suggests that the role of SAA in in ammation is regulatory and homeostatic rather than simply proin ammatory (12)(13)(14)(15).
The present study aimed to verify the in uence of cerebral injury on C7 regeneration in both patients and animals. Furthermore, we investigated the mechanism underlying the bene cial effect of cerebral injury on C7 regeneration. By using proteomics, ELISA and immunohistochemical staining techniques, we documented a further increase in SAA1 expression in mice subjected to TBI and BPI compared with mice subjected to BPI alone. The effect of SAA1 on C7 recovery was explored by applying a commercial recombinant SAA1 protein. Histological examination was subsequently performed to evaluate morphological recovery by measuring the regeneration of C7 nerve, while functional recovery was assessed by electrophysiological recordings. Mechanistically, the extent of the in ammatory response and macrophage phenotype were determined at various time points after BPI.

Patients
The study included two groups of 16 subjects, with 8 patients in each group. All the patients received standard rehabilitation treatment in our department after CC7 surgery. One group included patients with spastic paralysis of one upper limb caused by brain injury (such as stroke or trauma.), and the other group included patients who underwent nerve transfer due to BPI. All the participants underwent Tinel's sign examination once per month after surgery to assess the regeneration of the C7 nerve. The institutional review board of Huashan Hospital approved the study, and each participant provided written informed consent.

Experimental animals
The mouse experiments were approved by the Animal Care and Use Committee of Huashan Hospital and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Adult mice (male, 8-10 weeks) were purchased from the Shanghai SLAC Laboratory (http://www.lascn.com/SupplyDemand/Site/Index.aspx?id=77). The mice were housed at 22°C and 50% humidity on a 12-h light/dark cycle. All experiments were conducted during the light phase, and the mice were anesthetized with sodium pentobarbital (80 mg/kg, ip).

Controlled cortical impact
The procedures were similar to those described in our previous study (16,17). An anesthetized mouse was xed in a stereotactic frame, the scalp was opened, and the skull was cleaned. A 3.0-mm-diameter hole was made in the skull over the motor cortex of the left hemisphere (anteroposterior = +1.0 mm to -2.0 mm, mediolateral = 0.5 mm to 3.5 mm). A head trauma contusion device (TBI-0310 Impactor, Precision Systems and Instrumentation) was equipped with a 3.0-mm-diameter impact tip. The injury was delivered with a speed of 3.5 m/s, a depth of 1.0 mm, and a dwell time of 150 ms. For the sham TBI operation, a hole was made in the skull, and then the skin was sutured without induction of cortex injury.

Brachial plexus injury
Two weeks after TBI or sham TBI surgery, brachial plexus injury was induced. An anesthetized mouse was placed in the beach-chair position, and the brachial plexus was exposed as described in our previous study (18). The C5-T1 nerves on the right side were transected, and only the C7 nerve was sutured.

Electrophysiology
An electrophysiology device (Dantec Keypoint G4, Natus Medical Inc.) was utilized for electromyography. On the indicated days, the mice were anesthetized, and the sutured C7 nerve was re-exposed. A bipolar electrode was placed on the proximal side of the C7 coaptation site to deliver the stimulation, and the recording needle electrodes were placed on the triceps brachii. Waves with an amplitude of 1.0 mA and a width of 0.04 ms were delivered, and the average amplitudes of the compound muscle action potentials (CMAPs) were recorded to assess C7 regeneration.

Histological studies
To evaluate nerve regeneration, immuno uorescence was performed to detect the expression of neuro lament (to label axonal bers) and myelin protein zero (MPZ, to indicate myelination) in formalinxed, para n-embedded longitudinal sections of the C7 nerve. For each nerve, 3-µm-thick sections were cut, and then the sections were depara nized, boiled in 10 mM sodium citrate buffer for epitope retrieval and permeabilized with 0.5% Triton X-100. After incubation for two hours in blocking buffer, the sections were incubated with primary antibodies against neuro lament (Sigma, N4142, 1:100) and MPZ (Abcam, ab134439, 1:1000) at 4°C overnight. Afterwards, the sections were washed three times and then incubated with secondary antibody (Abcam, ab150077, 1:500 for neuro lament; Abcam, ab150176, 1:500 for MPZ) at room temperature for one hour. After multiple washing steps, the specimens were counterstained with 1 µg/ml DAPI (SouthernBio, 0100-20 was used to visualize antibody binding. ImageJ software was used for image analysis. Five random images per slide were taken to assess protein localization.

Proteomics and data analysis
Four weeks after BPI, serum was obtained from three mice from each group and prepared with the PreOmics "iST" Kit (P.O. 00001, PreOmics GmbH) according to the kit manual. After TMT labeling and high pH reverse-phase separation (ACQUITY UPLC Peptide C18 column, 2.1 mm x 150 mm, 1.7 μm), the samples were measured using an LC-MS instrument (Triple TOF 5600, AB SCIEX, USA).
Raw MS les were processed by MaxQuant version 1.5.2 (http://www.maxquant.org). The precursor mass and fragment mass were identi ed with initial mass tolerances of 6 ppm and 20 ppm, respectively. The minimal length was set to 7 amino acids, and the maximum length was set to two miscleavages. The "match between runs" option was selected when performing the MS runs, and a 20-second time window for retention was utilized for matching. Proteins that matched the reverse database were ltered out. The threshold for peptide and protein identi cation was set to FDR-corrected 0.01.
The pro le was then clustered based on Euclidean distances with average linkage using the modi ed function Clustergram in MATLAB R2014b, and a heatmap was also constructed in MATLAB.
We used a volcano plot to identify signi cantly differentially expressed proteins, with the X axis indicating the log2-based fold change and the Y axis representing the negative log10 of the p-value calculated from a two-tailed t-test. The red points in the upper left panel are signi cantly upregulated proteins, while the blue points are signi cantly downregulated proteins. Cutoff values of 1.5-fold for upregulated proteins and of 0.67-fold for downregulated proteins were used to identify differentially expressed proteins between the TBI+BPI mice and BPI mice (P < 0.05 for the t-test).
For bioinformatics analysis, differentially expressed proteins were subjected to ingenuity pathway analysis (IPA; Qiagen, Hilden, Germany) to elucidate their biological signi cance. Biological information obtained from the analysis is presented as biological function and disease and network data. For function and disease analysis, the threshold value was set to FDR-corrected P < 0.001. The gure showing enrichment data was constructed by the R programming language. Networks with scores greater than 30 were considered signi cant, and the gure showing the network was constructed by IPA.

PCR
Prior to nerve harvesting, the mice were perfused with cold PBS to remove the blood. RNA was isolated using an RNA extraction kit (TIANGEN, DP419) according to the manufacturer's instructions. Afterwards, the RNA was transcribed into cDNA using the PrimeScript™ RT Reagent Kit (Takara, RR047A). Finally, gene expression was analyzed by real-time PCR using PrimeScript™ RT Master Mix (Takara, RR036A) in a LightCycler® 480 instrument (Roche Diagnostics GmbH, Mannheim, Germany)

ELISA analysis
The dynamic levels of SAA1, which was identi ed by bioinformatics analysis, were further determined by ELISA. At 0, 2, 4, and 6 weeks post-BPI, serum was obtained from 3 mice in each group (the TBI+BPI and BPI groups). Commercial ELISA kits were purchased from Life Diagnostics (West Chester, PA, USA), and the protein concentrations were measured according to the manufacturer's instructions.
SAA1 treatment SAA1 (Peprotech, Rocky Hill, USA) was administered as previously described (14). Mice were injected subcutaneously with 120 µg SAA1 per kg body weight immediately after nerve injury and every day thereafter. PBS without SAA1 served as the vehicle control.

Statistical analysis
Statistical analysis was performed using SPSS 22.0. The Wilcoxon rank sum test (U test) was used to compare the time for the C7 nerve to regenerate to the target muscle (Tinel's sign) between the two groups of patients. Two-way ANOVA followed by the Holm-Šidák post hoc test was utilized to compare dynamic nerve recovery between the two groups of mice. A U test was performed to assess the in uence of SAA1 on nerve regeneration and the differences in SAA1 localization, cytokine expression and macrophage polarization. The differences were considered statistically signi cant if P < 0.05.

Results
Subjects with cerebral injury obtained faster C7 regeneration As shown in Figure 1A, regeneration of the C7 nerve after surgery was faster in patients with cerebral injury than in patients who underwent CC7 surgery due to BPI. Furthermore, promoted recovery of C7 injury was observed in mice with the combination of BPI and TBI compared with mice with BPI alone ( Figure 1B-D), as measured by the amplitudes of triceps brachii CMAPs. Histological examinations were further performed to evaluate C7 regeneration at the structural level at various time points after C7 transection. Both immuno uorescence staining of longitudinal sections and electronic microscopy analysis of cross-sections suggested that morphological regeneration was accelerated in mice subjected to TBI+BPI compared to mice subjected to BPI alone ( Figure 1E and 1F). These results con rmed the cerebral injury-induced increase in peripheral nerve regeneration capacity, but its underlying mechanisms remain largely unknown.

Serum amyloid protein A1 expression was systemically induced by traumatic brain injury
After cerebral injury, both the systemic stress response and an increase in the permeability of the bloodbrain barrier result in signi cant changes in the systemic circulation. We next carried out proteomics analysis to evaluate the difference in systemic response between mice with TBI+BPI and those with BPI alone (Figure 2A), and a total of 135 TBI-related differentially expressed proteins were detected ( Figure  2B). Both biological function and disease enrichment analysis and network analysis revealed enrichment of in ammation-and acute phase response-related pathways ( Figure 2C and 2D). Furthermore, the protein SAA1, the level of which is signi cantly increased in mice with TBI+BPI, was identi ed by both analyses (Figure 2C and 2D). These analyses suggested that SAA1 is a candidate regulator underlying the cerebral injury-mediated increase in C7 regeneration capacity. Subsequently, ELISA was performed at various time points after C7. As shown in Figure 2E, the serum levels of SAA1 were increased in the TBI+BPI group compared with the BPI group; the levels peaked at 2 weeks, and the increase was sustained at 4 weeks. Consistently, an increase in the protein levels of SAA1 was also observed in the tissue around the injured nerve, as determined by immunohistochemical staining (Figure 2F and 2G). Although previous studies have demonstrated that SAA1 is primarily produced by the liver and secreted into the circulation upon stress, we cannot rule out the possibility that SAA1 may be generated in situ. The mRNA levels of SAA1 in the tissue around the injured nerve were determined by RT-PCR, and no signi cant difference was detected between the two groups ( Figure 2H). Thus, our results suggested that cerebral injury may systemically upregulate the expression of SAAs, which in ltrate into the tissue around the injured nerve and may contribute to peripheral nerve regeneration by modulating in ammation.

Cerebral injury modulated in ammation and macrophage phenotypes in injured peripheral nerves
To explore the mechanism of improved peripheral nerve regeneration triggered by cerebral injury, we assessed the expression of in ammatory mediators and macrophage phenotypes. Total RNA was isolated from tissue segments around the C7 nerve from the indicated groups at different time points after BPI. The expression of in ammatory cytokines, including IL-1β, IL-6, MCP-1, and Fizz1, was strongly induced by BPI ( Figure 3A). Speci cally, the mRNA levels of IL-1β and IL-6 reached a maximum at 2 weeks after BPI and subsequently dropped. The expression of MCP-1, a chemoattractant for macrophages, was slightly elevated at 1 and 2 weeks and then returned to baseline soon after. The induction of Fizz1 expression, an M2 macrophage-associated gene, was increased at 2 weeks and gradually reduced to basal levels at 4 weeks after BPI. Compared with BPI, TBI had no in uence on the maximum expression of IL-1β and IL-6 at 2 weeks after BPI but accelerated reductions in the levels of these cytokines at 4 weeks. No signi cant difference in the level of MCP-1 was detected between the two groups. Strikingly, the BPI-mediated increase in Fizz1 levels was further enhanced by TBI ( Figure 3A). Immuno uorescence staining was performed to determine the phenotype of the macrophages. M1 activation (CD86: F4/80) peaked at 1 week after BPI and gradually decreased subsequently. Although no signi cant difference in M1 activation was detected between the groups at any time point, a decrease in M1 macrophage activation seemed to be promoted by TBI ( Figure 3B and 3C). More importantly, TBI signi cantly promoted M2 activation of macrophages around the injured C7 nerve ( Figure 3D and 3E). Therefore, TBI may contribute to peripheral nerve regeneration by ne tuning of the proin ammatory response and induction of the macrophage M2 phenotype.

SAA1 treatment promoted peripheral nerve regeneration
To date, it is unclear whether SAA can increase peripheral nerve regeneration capacity. To answer this question, we set up a six-week treatment protocol using recombinant human SAA (Figure 4a). The control group received an equivalent volume of PBS. Electrophysiological recordings and histological examination were performed to evaluate peripheral nerve recovery at the functional and morphological levels. As expected, SAA1 treatment promoted C7 regeneration after BPI ( Figure 4B, 4C and 4D). Collectively, the results indicate that upregulation of SAA1 expression may be responsible for the bene cial role of TBI in peripheral nerve regeneration. The effects of SAA1 administration on the in ammatory response and macrophage phenotype in peripheral nerves subjected to BPI were further determined. SAA1 had no in uence on the maximum expression of IL-1β, IL-6 and MCP-1 after BPI ( Figure   4E). The nerve injury-induced M1 macrophage phenotype was also comparable between these two groups (data not shown). In contrast, the resolution of the proin ammatory response was accelerated by SAA1 administration, as evidenced by decreased levels of IL-1β and IL-6 ( Figure 4E). Moreover, SAA1 treatment resulted in higher mRNA levels of Fizz1 and M2 activation than the vehicle control ( Figure 4E, 4F and 4G). Collectively, these ndings suggest that the effects of SAA1 on the nerve injury-induced immune response and macrophage phenotype are similar to those of TBI, providing further evidence that TBI may promote peripheral nerve regeneration via an SAA1-dependent mechanism.

Discussion
The bene cial effect of cerebral injury on peripheral nerve regeneration was reported in a previous study, but the underlying mechanisms remain unknown (7). In the present study, a faster C7 regeneration related to cerebral injury was identi ed in patients receiving CC7 treatment, and TBI enhanced morphological regeneration and functional recovery of the brachial plexus nerve in mouse models. Intriguingly, the increase in SAA1 expression in both the plasma and local tissue was more obvious in the BPI+TBI group than in the BPI group. After BPI, SAA1 administration contributed to the regeneration of the brachial plexus nerve. Mechanistically, the bene cial effect of SAA1 on peripheral nerve regeneration may mainly depend on proper regulation of the in ammatory reaction and the phenotype of M1/M2 macrophages in the lesioned nerve.
A growing amount of evidence has demonstrated that mutual regulation exists between the central and peripheral nervous systems upon injury. After peripheral nerve injury, neuronal plasticity, glial hypertrophy and even adult neurogenesis can take place in the adult central nervous system (19)(20)(21). However, limited studies have investigated the regulatory effect of the central nervous system on the injured peripheral nervous system. Our previous study implied a faster regeneration of peripheral nerves in patients who underwent CC7 to treat spastic arm paralysis due to cerebral injury. In addition, rats with combined sciatic and craniocerebral injury exhibited better functional and morphological recovery of the sciatic nerve than controls subjected to sciatic nerve injury alone (7). While these ndings support the concept that cerebral injury contributes to regeneration after peripheral nerve injury, the precise mechanisms underlying this phenomenon remain unclear. Numerous studies have documented that TBI facilitates fracture healing by increasing the levels of various cytokines and growth factors (22,23). Thus, we presumed that cerebral injury may contribute to peripheral nerve recovery in a similar manner.
Early activation of the in ammatory response after peripheral nerve injury, called neuroin ammation, may result in both bene cial and detrimental effects (24). Following nerve injury, neuroin ammation is essential for degeneration and clearance of the distal nerve segment, which is termed Wallerian degeneration. However, a prolonged and excessive in ammatory response exacerbates peripheral nerve injury and inhibits neurite outgrowth. Both resident macrophages and in ltrating macrophages play important roles in neuroin ammation of the peripheral nervous system. In the early phase, the majority macrophages exhibit an M1 phenotype, which is associated with proin ammatory functions, neurodegeneration and myelin phagocytosis (8). In addition, downregulation of the expression of M1related factors by pharmacological or genetic approaches compromises Wallerian degeneration and subsequent regeneration. In the present study, the total number of macrophages in injured peripheral nervous system was comparable between mice subjected to TBI and controls. Furthermore, neither the M1-activatied macrophages nor M1-associated gene expression was altered by TBI prior to 2 weeks after BPI. According to these results, TBI does not participate in the proliferation of resident macrophages, the in ltration of circulating macrophages or macrophage phenotype in the early phase. Thus, we hypothesize that TBI may have no in uence on Wallerian degeneration after peripheral nervous injury.
In contrast to the role of M1 macrophages in Wallerian degeneration, M2 macrophages are involved in the nerve regenerative process. Typically, M1 macrophages are associated with proin ammatory and neurodegenerative characteristics, while M2 macrophages are associated with anti-in ammatory and reparative functions in the late phase of in ammation (8). In our study, TBI induced an increase in the M2 activation of macrophages 2 weeks after BPI. Considering that the total number of macrophages was not signi cantly altered by TBI, it is reasonable to speculate that BPI does not in uence the proliferation of resident macrophages or the in ltration of circulating macrophages but contributes to the phenotypic switch of M1 to M2 macrophages. Due to the neurotoxic effect of M1 macrophages on cultured neuronal cells, M2 macrophages have been reported to promote neurite outgrowth in vitro (8,25). An elevated number of M2 macrophages is also related to accelerated structural and functional recovery after peripheral nerve injury in vivo (26). Thus, our data suggest that cerebral injury promotes peripheral nerve regeneration by facilitating the M1-to-M2 transition of macrophages.
To explore the molecular mechanism by which cerebral injury regulates the macrophage phenotype in the peripheral nervous system, proteomics assays combined with bioinformatics analyses were performed to identify the underlying molecules. Among the diverse molecules in the circulation that were signi cantly altered by TBI, SAA1, the protein expression of which was upregulated in the serum and tissue around the injured nerve, stands out as a promising candidate. However, the mRNA levels of SAA1 were not in uenced by TBI. These ndings are in agreement with the classic notion that SAA1 is primarily produced by the liver and secreted into the circulation upon stress. In a previous study, SAA1 was shown to have a potential effect on the induction of the M2 macrophage phenotype both in vitro and in vivo (27,28). By using recombinant human SAA1, we provided further evidence that SAA1 administration contributes to the resolution of the proin ammatory response, facilitates the M1-to-M2 transition of macrophages and promotes peripheral nerve regeneration. As documented by previous investigations, neurotrophic factors released by macrophages may underlie the promotion of nerve regeneration (29,30). Cattin and colleagues recently found that peripheral nerve regeneration is a multicellular process initiated by hypoxia. Among multiple cell types, only macrophages can sense hypoxic stimuli and induce neovascularization by stabilizing HIF-1a and secreting VEGF-A. The newly formed blood vessels provide a scaffold for Schwann cell-mediated peripheral nerve regeneration (9). Considering that the proangiogenic effect of M2 macrophages is well documented in other tissues and under pathological conditions, we reason that M2 macrophages may also play a similar role in the regeneration of peripheral nerves (31). However, further studies are needed to verify our hypothesis.
In the present study, we found that TBI enhances morphological regeneration and functional recovery of the C7 nerve. In addition, the bene cial effect of cerebral injury on peripheral nerve regeneration is at least partly mediated by upregulation of SAA1 expression, which contributes to the resolution of the proin ammatory response and facilitates the M1-to-M2 transition of macrophages. These ndings imply that SAA1 represents a potential therapeutic target for the promotion of peripheral nerve regeneration.

Conclusions
CC7 surgery provides an opportunity to gain insight into the interaction between the brain and peripheral nervous system. The circulation is one system that connects them and altered neuroin ammation at least partly accounts for the in uence of cerebral injury on peripheral nerve regeneration. Modulation of neuroin ammation represents an approach to enhance nerve repair, and SAA1 may be a promising therapeutic target. Abbreviations CC7, contralateral seventh cervical nerve transfer; BPI, brachial plexus injury; TBI, traumatic brain injury; SAA1, serum amyloid protein A1; C7, seventh cervical nerve; CMAPs, compound muscle action potentials; MPZ, myelin protein zero; IPA, ingenuity pathway analysis.

Declarations
Ethics approval and consent to participate The institutional review board of Huashan Hospital approved the study on patients, and each participant provided written informed consent. The mouse experiments were approved by the Animal Care and Use Committee of Huashan Hospital and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.