Involvement of phosphatase and tensin homolog deleted from chromosome 10 in rodent model of neuropathic pain

Background Many cancer research studies have extensively examined the phosphatase and tensin homolog deleted from chromosome 10 (PTEN) pathway. There are only few reports that suggest that PTEN might affect pain; however, there is still a lack of evidence to show the role of PTEN for modulating pain. Here, we report a role for PTEN in a rodent model of neuropathic pain. Results We found that chronic constriction injury (CCI) surgery in rats could elicit downregulation of spinal PTEN as well as upregulation of phosphorylated PTEN (phospho-PTEN) and phosphorylated mammalian target of rapamycin (phospho-mTOR). After examining such changes in endogenous PTEN in neuropathic rats, we explored the effects of modulating the spinal PTEN pathway on nociceptive behaviors. The normal rats exhibited mechanical allodynia after intrathecal (i.t.) injection of adenovirus-mediated PTEN antisense oligonucleotide (Ad-antisense PTEN). These data indicate the importance of downregulation of spinal PTEN for nociception. Moreover, upregulation of spinal PTEN by i.t. adenovirus-mediated PTEN (Ad-PTEN) significantly prevented CCI-induced development of nociceptive sensitization, thermal hyperalgesia, mechanical allodynia, cold allodynia, and weight-bearing deficits in neuropathic rats. Furthermore, upregulation of spinal PTEN by i.t. Ad-PTEN significantly attenuated CCI-induced microglia and astrocyte activation, upregulation of tumor necrosis factor-α (TNF-α) and phospho-mTOR, and downregulation of PTEN in neuropathic rats 14 days post injury. Conclusions These findings demonstrate that PTEN plays a key, beneficial role in a rodent model of neuropathic pain.


Background
Pain affects 1.5 billion people globally, including 116 million people in the USA and 164 million people in Europe and Israel combined [1]. The 2009 global pain market was estimated to be over US $50 billion [2]. Previous studies indicated that chronic pain occurs in about 20% of the general population [3,4], and the prevalence of neuropathic pain is 6.9% [4]. In a review of 174 trials published, Finnerup et al. [5] reported that there are no drug treatments available that can relieve all neuropathic pain conditions. Moreover, the detailed mechanisms underlying neuropathic pain still remain unclear.
The phosphatase and tensin homolog deleted from chromosome 10 (PTEN) is a tumor suppressor of phosphatased activity [6]. PTEN has been studied extensively through cancer research [7,8]; however, there are only few reports that suggest that PTEN might affect pain [9]. Goebbels et al. have demonstrated that by targeted disruption of Pten in Schwann cells that causes focal hypermyelination in the PNS and is associated with progressive peripheral neuropathy in mice [9]. However, there is still a lack of evidence to show the role of PTEN for modulating pain. PTEN is considered an upstream inhibitory mediator of mammalian target of rapamycin (mTOR) [10][11][12]. Several studies have demonstrated that inhibition of the spinal mTOR pathway can attenuate nociception in neuropathic pain [13,14]. We suspect that PTEN upregulation may have a therapeutic effect on neuropathic pain.
Here, we use a rat model of chronic constriction injury (CCI)-induced neuropathic pain combined with an intrathecal (i.t.) delivery system to determine whether central PTEN plays a role in neuropathic pain. We examined changes in endogenous spinal PTEN in neuropathic rats using our adenovirus-mediated target gene overexpression system. This system has been shown to be stable and effective in rodent models, especially adenovirusmediated PTEN (Ad-PTEN) in in vitro mouse models [15] and in both ex vivo [16] and in vivo [17] rat models. Additionally, we examined whether modulating the spinal PTEN pathway affected nociceptive behaviors that were measured reliably with behavioral pain assays [18,19]. Spinal neuroinflammation may accelerate central sensitization and promote the development and maintenance of neuropathic pain [20,21]. Furthermore, spinal neuroinflammation in CCI is characterized by microglial and astrocytic activation and increased expression of the proinflammatory mediator tumor necrosis factor-α (TNF-α) [19,22]. Many studies have further demonstrated that inhibiting microglial and astrocytic activation can have analgesic effects [19,[23][24][25]. Similarly, TNF-α reportedly plays key roles in neuropathic pain [26,27], whereas, inhibition of spinal TNF-α inhibits neuropathic pain behavior [28]. In the present study, we examine whether PTEN affects spinal microglial and astrocytic activation and upregulation of TNF-α accompanied the nociceptive behaviors in CCI rats. In addition, we examine whether PTEN affects downstream signaling of mTOR at the spinal level and in the neuropathic state.

Animals
We housed the male Wistar rats (260 to 285 g) for free access to food and water in a temperature-controlled (22°C ± 1°C) and light-cycle-controlled (12-h light/12-h dark) room. After the approval by the National Sun Yatsen University and Use Committee, we conformed to the Guiding Principles in the Care and Use of Animals of the American Physiology Society to use rats throughout the experiments. For surgery and drug injections, all rats were anesthetized under isoflurane inhalation (2%). Then, for preventing infection during the surgery, all rats received intramuscularly postoperative injection of veterin (cefazolin; 0.17 g/kg). Our every effort in experimental design and execution was on the purpose of minimizing the suffering and number of rats we used.

Induction of peripheral mononeuropathy (CCI)
As described by Bennett and Xie [29] and our previous studies [19,23], we performed the surgery of CCI to the right sciatic nerve of rats, to expose the right sciatic nerve of rats (at mid-thigh level), to dissect a 5-mm-long nerve segment of the sciatic nerve, to place four loose ligatures (4 to 0 chromic gut) around the sciatic nerve (with 1-mm intervals), and then to suture muscle and skin incision layer by layer. For the sham-operated rats, the surgery was performed only to expose the right sciatic nerve but without ligation.

Implantation of i.t. catheters
Using the method described in Yaksh and Rudy [30] and our previous studies [19,23], we inserted i.t. catheter (PE5 tubes: 9-cm long, 0.008-in. inner diameter, 0.014in. outer diameter; Spectranetics, Colorado Springs, CO, USA) to the lumbar enlargement of the spinal cord via the atlanto-occipital membrane at the base of the rat's skull down, and then externalized and fixed one end of the i.t. catheter to the cranial aspect of the rat's head for spinal drug administration. Because the dead volume of i.t. catheter was 3.5 μL, to ensure complete drug delivery, an i.t. artificial cerebrospinal fluid (aCSF) flush (10 μL) followed all i.t. injections. The composition of aCSF is as follows: 151.1 mM Na + , 2.6 mM K + , 1.3 mM Ca 2+ , 0.9 mM Mg 2+ , 122.7 mM Cl − , 21.0 mM HCO 3 − , 2.5 mM HPO 4 2− , and 3.5 mM dextrose and bubbled with 5% CO 2 in 95% O 2 for adjusting the final pH to 7.3. Five days after implantation of i.t. catheters, we excluded rats with i.t. catheter that had the fresh blood in the CSF or exhibit of gross neurological injury from the following experiments. According to the method described in other previous study [31] and our previous studies [19,23], we evaluated the locomotor function of rats using the Basso, Beattie, and Bresnahan (BBB) locomotor scale [32].
Nociceptive behavioral testing (i.) Thermal hyperalgesia test: After we placed each rat into each compartment of clear plastic cages onto an elevated glass platform, we then used an IITC analgesiometer (IITC Inc., Woodland Hills, CA, USA) to test thermal hyperalgesia as described previously by Hargreaves et al. [33] and our previous study [18,34] to position a radiant heat source with low-intensity heat (active intensity = 25) onto the middle of the plantar surface of the rat, with a cutoff time of 30 s, to measure the paw withdrawal latency (PWL; in seconds) of the rat until the rat showed a positive sign of pain behavior (licking or withdrawal).
(ii.) Mechanical allodynia test: After we set each rat into each compartment of clear plastic cages onto an elevated metal mesh floor for easy access to the rats' paws, we measured hindpaw withdrawal threshold (PWT; in g) to assess mechanical allodynia using calibrated von Frey filaments (Stoelting, Wood Dale, IL, USA). Then, we apply a series of von Frey filaments of logarithmically incremental stiffness to the midplantar region of the rat hindpaw by Chaplan's 'up-down' method to determine the closest filament to the threshold of pain response (licking or withdrawal) of the rat as described previously by Chaplan et al. [35] and our previous studies [18,19].
(iii.) Cold allodynia test: As described previously by our previous study [18], after we placed each rat in the individual plastic compartments on an elevated metal mesh floor, cold allodynia response (acetone response score; in point) of the rat was monitored during 1 min following acetone stimulus (25 μl) onto the center of the plantar surface of rats' hindpaw. Modified from 4-point scale of previous studies [36,37], acetone response scores of the paw were graded according to a 6-point scale: 0, repeated flicking with persistent licking in 2 s following acetone stimulus; 1, prolonged withdrawal or repeated flicking in 2 s following acetone stimulus; 2, quick and more violent withdrawal, flick, or stamp in 2 s following acetone stimulus; 3, quick withdrawal, flick, or stamp in 2 s following acetone stimulus; 4, withdrawal, flick, or stamp more than 2 s following acetone stimulus; and 5, no response. Then, to sum the four individual scores for obtaining acetone response score of the paw of each rat. The minimum possible total score of the paw could be 0 point (the rat exhibits repeated flicking and licking of paws on each of the four trials), and the maximum possible total score of the paw could be 20 points (the rat exhibits no response to any of the four trials).
(iv.) Weight-bearing test: To let the rat place its hindpaws onto the two force transducers of the incapacitance tester (Singa Technology Corporation, Taoyuan, Taiwan, Republic of China) for measuring the hindpaw weightbearing deficits (change in hindpaw weight distribution; in g) as described in our previous study [18,38]. Under normal conditions, the naïve rat distributes weight equally between both hind limbs, but after inducing inflammation of one hind limb (like CCI), the rat redistributes weight for lowering weight-bearing of the affected limb [39]. Change in hindpaw weight distribution of the rat was expressed as a difference by subtracting the affected limb (the right hind limb; the ipsilateral side) from the normal limb (the left hind limb; the contralateral side) measured at the same time.

Spinal immunohistofluorescence analysis
For reducing variations in immunohistochemical procedures, we mounted the lumbar spinal tissues from the different groups of rats into the same OCT block, sectioned these spinal tissues together using a cryostat at −30°C (HM550; Microm, Waldorf, Germany) and performed the following spinal immunohistofluorescence analysis, with a modified method described in Sung et al. [40] and our previous studies [18,23] For immunostaining analysis, we use a Leica DM-6000 CS fluorescence microscope (Leica Instruments Inc., Wetzlar, Germany) for examination of these stained spinal sections, and then use a SPOT Xplorer Digital camera (Diagnostic Instruments, Inc., Sterling Heights, MI, USA) for photographing all immunofluorescence images of phospho-PTEN, PTEN, phospho-mTOR, OX-42, GFAP, and TNF-α, respectively. We measured the pixel values of the immunoreactivepositive area (using three sections per rat) by Image J software (National Institutes of Health, Bethesda, MD, USA). Spinal neurons located in the superficial laminae, laminae I to III, respond to nociceptive stimuli and directly participate in the transmission of nociception to the brain [41], and thereby the superficial laminae play a more important role in neuropathic pain than the deep laminae. Therefore, in accordance to the method used for neuropathic rodents [41][42][43][44], we focused on quantifying the immunoreactivity of the targeted protein in the superficial laminae of the spinal cord. The immunofluorescence data were expressed as a percentage change compared to shamoperated or sham-operated plus i.t. vehicle group, which were considered to be 100%. For double-immunofluorescent staining of PTEN and neuronal marker, the spinal sections were incubated with a mixture of anti-PTEN (1:200 dilution) and anti-Neuronal Nuclei (NeuN) (neuronal-specific nuclear protein; neuronal marker, 1:500, Alexa Fluor 488 conjugated antibody, cat. MAB377X, EMD Millipore, Temecula, CA, USA; monoclonal mouse antibody) antibodies overnight at 4°C, and then followed by DyLight 549-conjugated anti-rabbit IgG antibody (1:400 dilution) for 40 min at room temperature. On the other hand, for double-immunofluorescent staining of PTEN and microglial marker or astrocytic marker, the spinal sections were incubated with a mixture of anti-PTEN (1:200 dilution) and OX-42 (1:200 dilution) or GFAP (1:200 dilution) antibodies overnight at 4°C, and then followed by a mixture of Alexa Fluor 488-conjugated anti-mouse IgG antibody (1:400 dilution) and DyLight 549-conjugated anti-rabbit IgG antibody (1:400 dilution) for 40 min at room temperature. For double-immunofluorescent staining of anti-TNF-α or phospho-mTOR and astrocytic marker, the spinal sections were incubated with a mixture of anti-TNF-α (1:200 dilution) or anti-phospho-mTOR (1:200 dilution) and GFAP (1:200 dilution) antibodies overnight at 4°C, and then followed by a mixture of DyLight 549-conjugated antirabbit IgG antibody (1:400 dilution) and Alexa Fluor 633conjugated goat anti-mouse IgG antibody (1:400 dilution, cat. A21052; Life Technologies Corporation) for 40 min at room temperature. The double-immunostaining images were examined and acquired with Leica TCS SP5 II confocal microscope (Leica Instruments Inc., Wetzlar, Germany). We set the color of TNF-α or phospho-mTOR for 549-nm excitation line as pseudo red and set the color of GFAP for 633 nm excitation line as pseudo green. When co-localization of two proteins occurred, the merge of above two colors, pseudo red (TNF-α or phospho-mTOR) and pseudo green (GFAP), yielded yellow color.

Western blot analysis
Following our published method [22,34,45], we performed Western blotting analysis on the ipsilateral side of the lumbar spinal dorsal samples from the rats. For Western blotting analysis, we collected and washed spinal dorsal samples from rats with ice-cold PBS, and then used a Polytron homogenizer (5 cycles of 10 s at 3,000 rpm) to homogenize in ice-cold lysis buffer (1 μg/ml aprotinin, 50 mM Tris, 150 mM NaCl, 100 μg/ml phenylmethylsulfonyl fluoride, 1% Triton X-100, pH 7.5). After centrifuging at 20,000 × g for 60 min at 4°C, we retained this supernatant for Western blot analysis of phospho-PTEN and PTEN. Modified from the method of Lowry et al. [46], we determined the protein concentrations of the supernatant with the DC protein assay kit (Bio-Rad, Hercules, CA, USA). We added an equal volume of sample buffer (2% 2-mercaptoethanol, 10% glycerol, 0.1% bromophenol blue, 50 mM Tris-HCl, pH 7.2, and 2% sodium dodecyl sulfate (SDS)) to the supernatant. Then, we electrophoresed the proteins of supernatant through a tricine SDS-polyacrylamide gel at 150 V for 90 min and transferred the proteins within the gel to a polyvinylidene difluoride membrane (PVDF membrane; Immobilon-P, Millipore, 0.45-μM pore size) in transfer buffer (380 mM glycine, 1% SDS, 50 mM Tris-HCl, 20% methanol) at 125 mA overnight at 4°C. With blocking PVDF membrane for 1 h at room temperature using 5% non-fat dry milk in Tris-buffered saline (TTBS; 137 mM NaCl, 0.1% Tween 20, 20 mM Tris-HCl, pH 7.4), we then incubated the PVDF membrane with antibodies against phospho-PTEN (Ser380; 1:1,000 dilution) or PTEN (1:1,000 dilution) proteins for 180 min at room temperature. The antibodies recognized immunoreactive bands of phospho-PTEN (approximately 54 kDa) and PTEN (approximately 47 kDa) protein, which were visualized with enhanced chemiluminescence (ECL kit; Millipore) and photographed using the UVP BioChemi imaging system (UVP LLC, Upland, CA, USA), respectively. Finally, we performed the relative densitometric quantification of the immunoreactive bands of phospho-PTEN and PTEN protein with LabWorks 4.0 software (UVP LLC, Upland, CA, USA), and relative variations between the bands of the sham-operated or shamoperated plus i.t. vehicle group and the other groups were calculated using the same image. In addition, we reprobed the PVDF membranes using an anti-β-actin antibody (1:2,500 dilution; catalog no. A5441; Sigma Co., Ltd., St Louis, MO, USA; monoclonal mouse antibody), which was the loading control.

Data and statistical analysis
We showed all following data as means ± standard error on the mean (SEM). For statistical analysis, we calculated differences between groups of rats by a one-way analysis of variance (ANOVA), used the Student-Newman-Keuls post hoc test, and then defined the criterion for statistical significance as P < 0.05.

The time course of changes of endogenous spinal PTEN in neuropathic rats
To explore any potential changes in the spinal PTEN pathway in neuropathic rats, we prepared the lumbar spinal tissues from the following eight groups of rats: (1) 1 day after sham operation, (2) 3 days after sham operation, (3) 7 days after sham operation, (4) 14 days after sham operation, (5) 1 day after CCI, (6) 3 days after CCI, (7) 7 days after CCI, and (8) 14 days after CCI. We performed a spinal immunohistofluorescence analysis using three antibodies against phospho-PTEN, PTEN, and phospho-mTOR to evaluate the expression of the PTEN pathway in the ipsilateral side of the lumbar spinal dorsal gray matter in neuropathic rats. Compared with the sham-operated group ( Figure 1A,D), immunoreactivity increased for phospho-PTEN but decreased for PTEN, 7 ( Figure 1B,E) and 14 days ( Figure 1C,F) after CCI. Similar to the findings in a previous study [47], phospho-mTOR immunoreactivity increased 7 ( Figure 1H) and 14 days ( Figure 1I) after CCI, as compared with that of the shamoperated group ( Figure 1G). Quantification of the immunoreactivity results further confirmed phospho-PTEN ( Figure 1J) and phospho-mTOR ( Figure 1L) upregulation, as well as PTEN downregulation ( Figure 1K), 3, 7, and 14 days after CCI. Further, we used Western blot analysis to verify changes in endogenous PTEN in the dorsal horn of the lumbar spinal cord (Figure 2A). The Western blot analysis revealed upregulation of phospho-PTEN ( Figure 2B) and downregulation of PTEN ( Figure 2C) 14 days after CCI surgery. To examine the cellular specificity of endogenous PTEN in the dorsal horn of the lumbar spinal cord, we prepared lumbar spinal tissues from the sham-operated group for double-immunofluorescent staining. Confocal double-immunostaining images of the lumbar spinal dorsal gray matter of sham-operated rats further confirmed that most PTEN signals were more often co-localized with GFAP-positive cells (astrocytes, Figure 2E) than NeuN-positive (neuronal cells; Figure 2D) or OX-42-positive cells (microglial cells, Figure 2F).

The effect of i.t. Ad-antisense PTEN on nociceptive responses
To explore the effects of modulating spinal PTEN pathway on nociceptive behaviors in normal rats, we prepared the following five groups of rats: (1)

The inhibitory effects of Ad-PTEN on nociception in neuropathic rats
To verify that Ad-PTEN has the ability to upregulate production of spinal PTEN, we prepared the following three groups of rats: (1) naïve plus i.  the immunoreactivity results ( Figure 4D) further confirmed that PTEN upregulation in the naïve plus i.t. Ad-PTEN group 14 days after i.t. injection. Confocal double-immunostaining images of the lumbar spinal dorsal gray matter further confirmed most PTEN signals were more often co-localized with GFAP-positive cells (astrocytes; Figure 4F) than NeuN-positive (neuronal cells; Figure 4E) or OX-42-positive cells (microglial cells; Figure 4G) in the naïve rats administered i.t. Ad-PTEN.
To examine whether spinal PTEN upregulation induced with i.t. Ad-PTEN affected neuropathic pain behaviors, we prepared the following three groups of rats: (1) CCI plus i.t. vehicle, (2) CCI plus i.t. Ad-GFP, and (3) CCI plus i.t. Ad-PTEN. Five days after the i.t. catheters were implanted, the rats underwent CCI surgery, and the vehicle, Ad-GFP, or Ad-PTEN was administered i.t. immediately afterwards. Similar to previous findings [48], CCI promoted the development and maintenance of nociceptive behaviors in rats. Compared with the vehicle group, i.t. Ad-GFP did not significantly affect PWL ( Figure 5A), PWT ( Figure 5B), acetone response score ( Figure 5C), and weight-bearing deficits ( Figure 5D) for the ipsilateral hindpaw of CCI rats. Compared with i.t. Ad-GFP group, i.t. Ad-PTEN significantly attenuated CCI-induced thermal hyperalgesia ( Figure 5A), mechanical allodynia ( Figure 5B), cold allodynia ( Figure 5C), and weight-bearing deficits ( Figure 5D) up to 14 days post injury. In contrast, similar to previous studies [49,50], the contralateral hindpaw in CCI rats did not show pain-related behavior (       Figure 10B) and phospho-mTOR (red; Figure 10E) were primarily co-localized with astrocytes in the CCI group, and i.t. Ad-PTEN significantly attenuated CCI-induced upregulation of spinal TNF-α ( Figure 10C) and phospho-mTOR ( Figure 10F) immunoreactivity in astrocytes.

Discussion
The role of endogenous PTEN in neuropathic and normal rats Previous neuroscience studies have focused on the role of PTEN in axon regeneration [11,51,52], Alzheimer's disease [53], Parkinson's disease [54], ischemic brain injury [12], and spinal cord injury [52,[55][56][57]. To the best of our knowledge, the present report is the first to explore the role of spinal PTEN in pain, especially neuropathic pain. Using a spinal immunohistofluorescence assay system and Western blot analysis, we found that CCI could elicit downregulation of ipsilateral dorsal spinal PTEN in rats (Figures 1 and 2). Phosphorylation of Ser-380 is known to suppress phosphatase activity of PTEN [58]. We also demonstrated that upregulation of phospho-PTEN was associated with downregulation of PTEN 14 days after CCI surgery (Figures 1 and 2). Phosphorylation of Ser-2448 mTOR is a biomarker for Relative band intensities of PTEN and phospho-mTOR were quantified by densitometry and indicated as the percent change relative to that for the sham-operated 7 days post injury group (100%). Western blotting revealed that PTEN and phospho-mTOR protein expression were detectable in the sham-operated group, but CCI-induced downregulation of PTEN and upregulation of phospho-mTOR protein on day 7 and day 14 after CCI surgery were both attenuated by i.t. Ad-PTEN. Western blotting of β-actin was performed to verify that equivalent amounts of protein were loaded in each lane. Each bar in (B) and (C) represents the mean ± SEM with three rats per group. *P < 0.05 compared with sham-operated 7 days post injury group. # P < 0.05 between CCI plus i.t. Ad-PTEN group compared with the same time points in CCI plus i.t. Ad-GFP group. Ad-GFP, adenovirus-mediated green fluorescent protein; Ad-PTEN, adenovirus-mediated phosphatase and tensin homolog deleted from chromosome 10; CCI, chronic constriction injury; p-mTOR, phosphorylated mammalian target of rapamycin.
the activation status of mTOR [59], a downstream inhibitory target of PTEN [10][11][12]. However, there is a lack of information regarding the changes in spinal phospho-mTOR in CCI rats. Indeed, we found that upregulation of phospho-mTOR ( Figure 1L) was associated with upregulation of phospho-PTEN ( Figure 1J) after CCI surgery. After examining the changes in endogenous PTEN levels in neuropathic rats, we explored the potential effects of modulating spinal PTEN pathway on neuropathic pain. We observed that rats exhibited mechanical allodynia after i.t. injection of Ad-antisense PTEN ( Figure 3B), which confirms that downregulation of spinal PTEN is important for nociception, and therefore, we suspect that spinal PTEN exerts beneficial effects against pain.

The analgesic effects of spinal PTEN
Clinical neuropathic pain syndromes are characterized by evoked pain (hyperalgesia and allodynia) or spontaneous pain [60]. We have confirmed these nociceptive behaviors in the rat CCI model, including thermal hyperalgesia ( Figure 5A), mechanical and cold allodynia ( Figure 5B,C), and weight-bearing deficits (spontaneous pain; Figure 5D). Ad-PTEN administered i.t. significantly prevented CCI-induced development of pain behaviors in neuropathic rats, including thermal hyperalgesia, mechanical allodynia, and cold allodynia accompanied with weight-bearing deficits ( Figure 5). Central sensitization within the dorsal horn of the spinal cord could contribute to the hypersensitive pain behaviors commonly observed with neuropathic pain [61]. Activation of microglia and astrocytes could dominate spinal neuroinflammation [62,63] and accelerate central sensitization as well as the subsequent development and maintenance of neuropathic pain [20,21]. Upregulation of OX-42 (microglial marker) and GFAP (astrocytic marker) immunoreactivity in the spinal dorsal horn are known indicators of elevated nociceptive states [23][24][25][64][65][66][67][68][69], which was further confirmed in the CCI rat model by our spinal immunohistofluorescence data (Figure 7). Many studies have demonstrated that inhibition of microglial and astrocytic activation exerts analgesic effects [19,[23][24][25]. Although minocycline, a microglial inhibitor, can prevent neuropathic pain, it cannot reverse established neuropathic pain, thereby showing the important role of spinal microglia in the development phase but not in the maintenance phase of neuropathic pain [25,70,71]. However, it is still controversial whether spinal microglia contributes to the maintenance of nociception in neuropathy. As previously reported by Stuesse et al., OX-42 activity substantially peaked at around day 7 after nerve injury, and this increase was maintained for 35 days post-injury [68]. Several studies also suggest that microglial activation is probably important for the maintenance as well as the early development of neuropathic pain [31,72,73]. A series of studies have shown that astrocytic activation is implicated in the maintenance of neuropathic pain [64][65][66][67]69], indicating the key role of astrocytes in maintaining neuropathic pain [74,75]. Activated astrocytes are also reported to be important contributors of neuropathic pain development [76]. Our previous studies [19,42,77] and our present findings also indicate that nociceptive behaviors accompany spinal microglial and astrocytic activation (Figure 7) in CCI rats. Moreover, our present findings also demonstrate that the upregulation of spinal PTEN by i.t. Ad-PTEN can significantly attenuate CCI-induced microglial and astrocytic activation in neuropathic rats (Figure 7).

The majority of spinal cells affected by PTEN
The expression and localization of PTEN in pain-related regions of the central nervous system are not completely understood. Our confocal double-immunostaining images of the lumbar spinal dorsal gray matter further confirmed that astrocytes are a major source of PTEN expression as compared to neurons and microglia in sham-operated rats ( Figure 2). But certainly, the possibility that neurons and microglia could also be the source of PTEN expression cannot be excluded. This finding supports the hypothesis that the spinal astrocytic response plays a crucial role in neuropathic pain [19]. Ad-PTEN administered via i.t. injection significantly upregulated spinal PTEN immunoreactivity ( Figure 4D). Previous studies have shown that the majority of spinal cells with adenovirus-mediated target gene overexpression were astrocytes [78,79]. Similar to the above studies, the present confocal results showed that spinal PTEN was primarily co-localized with astrocytes in the i.t. Ad-PTEN group (Figure 4E,F,G). Moreover, our present findings also showed that i.t. Ad-PTEN significantly attenuated CCI-induced astrocytic activation in neuropathic rats (Figure 7). This phenomenon was partially echoed by the finding that PTEN loss resulted in hypertrophy and increased proliferation of astrocytes in vivo [80], which are two characteristics of activated astrocytes [81]. Based on the above reports and the present results, we suggest that the prevention of neuropathic pain by i.t. Ad-PTEN may be mainly associated with its inhibitory effects on activated astrocytes, which play key roles in nociceptive hypersensitization. Previous studies have demonstrated that microglia could interact with astrocytes, causing downstream biochemical and behavioral responses to noxious stimuli [82][83][84]. Additionally, mediators released by activated microglia induce spinal astrocyte activation, and blocking these mediators attenuates astrocyte activation and pain [82][83][84][85]. Furthermore, mediators released by activated astrocytes, in turn, also activate spinal microglia in neuropathic or focal brain-injured rodents [86,87]. These findings from previous studies suggest that spinal microglia-astrocyte interactions could promote nociceptive responses ( Figure 11A). Therefore, in this study, we hypothesized that the effect of the upregulation of spinal PTEN by i.t. Ad-PTEN on attenuating CCI-induced microglial activation may be secondary to the direct inhibition of activated astrocytes through suppressing the astrocytemicroglia interaction by reducing the release of astrocytic mediators ( Figure 11B).

Potential analgesic mechanisms of PTEN
Previous studies have found that the activation of spinal mTOR is required for CCI-induced pain hypersensitivity [47]. Inhibition of the spinal mTOR pathway by i.t. rapamycin [13] or spinally administered rapamycin ester analogue CCI-779 [14] attenuates mechanical hyperalgesia in neuropathic rats. The above studies suggested that the anti-nociceptive effects produced by inhibition of the spinal mTOR pathway in neuropathic rats are consistent with a deleterious role for mTOR in neuropathic rats. In the present study, we observed upregulation of phospho-mTOR following CCI surgery ( Figure 1L). The findings of cancer research studies have indicated that PTEN is an upstream inhibitory mediator of mTOR, which also has been confirmed in the central nervous system (CNS) [10][11][12]. Our results indicated that i.t. Ad-PTEN significantly attenuated CCI-induced upregulation of spinal phospho-mTOR immunoreactivity (Figures 8 and 9). Moreover, similar to the systemic anti-inflammatory effects of intra-articular Ad-PTEN [17], i.t. Ad-PTEN also has central anti-inflammatory effects and can inhibit CCI-induced upregulation of spinal TNF-α (Figure 7). TNF-α, a neuropathic pain-related cytokine [26], caused thermal hyperalgesia and mechanical allodynia in rats [27]. Moreover, neuropathic mechanical allodynia in rats was blocked by i.t. injection of a TNF-soluble receptor, a TNF antagonist [28]. In the present study, the confocal results from CCI rats showed that upregulated phospho-mTOR and TNF-α were localized predominantly to astrocytes, and this upregulation was inhibited by i.t. Ad-PTEN ( Figure 10). Rapamycin, a specific inhibitor of mTOR, was reported to inhibit TNF-α release from human vascular smooth muscle cells in vitro [88]. Other studies have reported that rapamycin inhibition of mTOR in microglia [89] and astrocytes [90] produces anti-inflammatory effects, which suggests that inhibiting mTOR can reduce neuroinflammation [91]. Thus, according to the previous and present findings, we conclude that one analgesic mechanism of PTEN may be inhibition of the astrocytic inflammatory response (upregulated TNF-α) via inhibition of phospho-mTOR ( Figure 11B).

Advantages and future studies of PTEN for neuropathic pain
We suspect that the upregulation of PTEN may be therapeutic for neuropathic pain, a neuroinflammationrelated disorder. Our present results demonstrate that PTEN exerts beneficial effects against neuropathic pain, which highlights three future research directions. First, it is important and valuable to explore the functions of PTEN through neuroscience research, especially neuroinflammation-related disorder mediated by activated astrocytes. However, a few in vitro [89] and in vivo [92] studies showed that inhibition of mTOR may modulate neurons or microglial activation. Therefore, the possibility that i.t. Ad-PTEN directly affects spinal neurons or microglia cannot be excluded. Second, Sims et al. [93] considered the adenovirus to be an efficient and safe vector for CNS gene delivery, although there are limitations to adenoviral-mediated PTEN gene therapy. The individuals treated with adenoviral vectors were previously reported to have experienced systemic reactions including fever, chill, and hypertension [48]. Although i.t. Ad-PTEN did not significantly affect the response to thermal hyperalgesia, mechanical allodynia, or cold allodynia tests on the contralateral hindpaw in CCI rats ( Figure 6). Additionally, i.t. Ad-PTEN-treated Figure 11 Schematic representation of the possible mechanisms of PTEN involvement in neuropathic pain. (A) Peripheral nerve injury could downregulate the spinal astrocytic PTEN, which leads to the upregulation of phospho-mTOR and TNF-α and results in neuroinflammation and nociceptive sensitization. In addition, spinal microglia-astrocyte interactions also promote nociceptive responses. (B) Through the upregulation of spinal astrocytic PTEN by i.t. Ad-PTEN, peripheral nerve injury-induced upregulation of phospho-mTOR and TNF-α is inhibited, thereby inhibiting neuroinflammation and nociceptive sensitization. The effect of the upregulation of spinal PTEN by i.t. Ad-PTEN on attenuating peripheral nerve injury-induced microglial activation may be through suppressing the astrocyte-microglia interaction by reducing the astrocytic mediator TNF-α. Ad-PTEN, adenovirus-mediated phosphatase and tensin homolog deleted from chromosome 10; p-mTOR, phosphorylated mammalian target of rapamycin; TNF, tumor necrosis factor. normal rats and CCI rats did not exhibit any obvious abnormal external behavior (including locomotor function). The minimal effective dose and the duration of PTEN gene therapy still need to be determined in future research. Third, there is no known PTEN activator for neuropathic pain, and upregulation of PTEN may serve as an in vitro index to screen compounds that may be potential neuropathic pain therapies.

Conclusions
This study is a novel step toward understanding the benefits of PTEN on neuropathic pain and provides a foundation for future studies utilizing upregulation of PTEN in combinational therapies for neuropathic pain.