Inhibition of Microglia-Derived Oxidative Stress by Ciliary Neurotrophic Factor Protects Dopamine Neurons In Vivo from MPP+ Neurotoxicity

We demonstrated that capsaicin (CAP), an agonist of transient receptor potential vanilloid subtype 1 (TRPV1), inhibits microglia activation and microglia-derived oxidative stress in the substantia nigra (SN) of MPP+-lesioned rat. However, the detailed mechanisms how microglia-derived oxidative stress is regulated by CAP remain to be determined. Here we report that ciliary neurotrophic factor (CNTF) endogenously produced by CAP-activated astrocytes through TRPV1, but not microglia, inhibits microglial activation and microglia-derived oxidative stress, as assessed by OX-6 and OX-42 immunostaining and hydroethidine staining, respectively, resulting in neuroprotection. The significant increase in levels of CNTF receptor alpha (CNTFRα) expression was evident on microglia in the MPP+-lesioned rat SN and the observed beneficial effects of CNTF was abolished by treatment with CNTF receptor neutralizing antibody. It is therefore likely that CNTF can exert its effect via CNTFRα on microglia, which rescues dopamine neurons in the SN of MPP+-lesioned rats and ameliorates amphetamine-induced rotations. Immunohistochemical analysis revealed also a significantly increased expression of CNTFRα on microglia in the SN from human Parkinson’s disease patients compared with age-matched controls, indicating that these findings may have relevance to the disease. These data suggest that CNTF originated from TRPV1 activated astrocytes may be beneficial to treat neurodegenerative disease associated with neuro-inflammation such as Parkinson’s disease.


Introduction
Microglia are the resident macrophages in the brain and spinal cord [1]. Many studies have demonstrated that activated microglia play a pivotal role in the cause and progression of Parkinson's   Given the neurotoxicity of activated microglia on DA neurons in the MPP + -lesioned SN [9], we hypothesized that MPP + neurotoxicity is in parallel with microglial activation in vivo. We found that MPP + -induced degeneration of DA neurons was accompanied by microglial activation as visualized by OX-6 (marker for activated microglia) and OX-42 (general marker for microglia) immunostaining in the SN ( Figure 1K,L). CAP treatment decreased the immunoreactivity of OX-6 + and OX-42 + cells in the SN at 2 weeks post MPP + injection compared to vehicle-treated MPP + -lesioned SN ( Figure 1Q,R).

Astrocytic TRPV1 Knockdown by shRNA Alters DA Neuronal Survival and Microglial Activation In Vivo in MPP + -Lesioned Rat
As astrocytic TRPV1 activation by CAP contributes to DA neuronal survival in vivo [13], we wondered whether it could exert neuroprotection by regulating microglial activation. To test this hypothesis, astrocytic TRPV1 function was selectively and efficiently inhibited by a lentivirus carrying a small hairpin-forming interference RNA (shRNA) targeted against TRPV1 (shTRPV1) [13] (also see Section 4.10). The virus also contained DNA encoding a fluorescent marker, EGFP, which permitted visualization of the location and amount of viral infection.
To examine the selectivity and efficacy of shRNA in vivo, we stereotaxically injected MPP + and control scrambled shRNA (shCtrl) in the SN (see Section 4.3). We observed EGFP expression of shCtrl mainly within astrocytes 7 days post MPP + injection ( Figure S1A,B). shTRPV1 selectively and efficiently inhibited TRPV1 expression primarily within astrocytes in the SN of MPP + -lesioned rat ( Figure S1C,D). To evaluate the role of astrocytic TRPV1 on DA neurons in the SN, control scrambled shRNA (shCtrl) or shTRPV1 was ipsilaterally injected into the SN immediately after the unilateral MFB injection of MPP + . In the CAP-treated MPP + -lesioned rats, shTRPV1 decreased the density of TH + fibers by 54% in the striatum ( As astrocytic TRPV1 activation by CAP contributes to DA neuronal survival in vivo [13], we wondered whether it could exert neuroprotection by regulating microglial activation. To test this hypothesis, astrocytic TRPV1 function was selectively and efficiently inhibited by a lentivirus carrying a small hairpin-forming interference RNA (shRNA) targeted against TRPV1 (shTRPV1) [13] (also see Section 4.10). The virus also contained DNA encoding a fluorescent marker, EGFP, which permitted visualization of the location and amount of viral infection.
To examine the selectivity and efficacy of shRNA in vivo, we stereotaxically injected MPP + and control scrambled shRNA (shCtrl) in the SN (see Section 4. 3). We observed EGFP expression of shCtrl mainly within astrocytes 7 days post MPP + injection ( Figure S1A,B). shTRPV1 selectively and efficiently inhibited TRPV1 expression primarily within astrocytes in the SN of MPP + -lesioned rat ( Figure S1C,D). To evaluate the role of astrocytic TRPV1 on DA neurons in the SN, control scrambled shRNA (shCtrl) or shTRPV1 was ipsilaterally injected into the SN immediately after the unilateral MFB injection of MPP + . In the CAP-treated MPP + -lesioned rats, shTRPV1 decreased the density of TH + fibers by 54% in the striatum (Figure 2A,G,M), and the numbers of TH + cells by 69% and Nissl + cells by 76% in the SN ( Figure 2B-D,H-J,N) as well as reducing functional recovery ( Figure 2O) compared to shCtrl.
In parallel with aggravation of DA neuronal loss and behavioral deficits by shTRPV1, astrocytic TRPV1 deficiency significantly increased immunostaining intensity of OX-6 + ( Figure 2K) and OX-42 + ( Figure 2L) cells in the SN compared with shCtrl-treated control rats. Collectively, these results indicate that inhibition of microglial activation by astrocytic TRPV1 activation might contribute to neuroprotection and motor recovery.  In parallel with aggravation of DA neuronal loss and behavioral deficits by shTRPV1, astrocytic TRPV1 deficiency significantly increased immunostaining intensity of OX-6 + ( Figure 2K) and OX-42 + ( Figure 2L) cells in the SN compared with shCtrl-treated control rats. Collectively, these results indicate that inhibition of microglial activation by astrocytic TRPV1 activation might contribute to neuroprotection and motor recovery. As CNTF has neuro-protective properties [37], and inhibits microglial activation [29] and/or microglia-derived oxidative stress [25], CNTF expression was assessed in the absence or presence of CAP in the SN in vivo of MPP + -lesioned rat. At 2 weeks post MPP + injection, analysis by immunohistochemical staining showed that CNTF expression in GFAP + astrocytes was significantly higher in the rat SN compared to control ( Figure 3A,B). CNTF expression in TH + neurons was significantly reduced ( Figure 3C,D), whereas basal level of CNTF expression in OX-42 + microglia was unchanged ( Figure 3E,F) compare to control. In CAP-treated MPP + -lesioned SN, CNTF expression in both GFAP + astrocytes ( Figure 3A,B) and TH + neurons ( Figure 3C,D) was significantly increased, whereas CNTF expression in OX-42 + microglia was relatively unchanged ( Figure 3E,F) compared to vehicle-treated MPP + -lesioned SN. Collectively, the results indicate that CAP induces expression of CNTF originated from astrocytes, but not microglia. Knockdown of astrocytic TRPV1 reduced CNTF expression in the absence or presence of CAP in the SN of MPP + -lesioned rat (data not shown), supporting our recent findings of astrocytic TRPV1 activation-derived CNTF expression [13].

CNTFRα Neutralization Alters Neuroprotection and Microglial Activation in the SN In Vivo in MPP + -Lesioned Rat
As CNTF primarily functions via CNTFRα [30], we examined if CNTF could exert its effect on microglia through CNTFRα expressed on microglia. Immunohistochemical analysis showed a significant increase in the level of CNTFRα expression in OX-42 + cells in the SN of MPP + -lesioned rats, compared with control ( Figure 4A-C). CNTFRα in Iba-1 + microglia was also significantly expressed in the SN of PD brain compared to age-matched human control brain (Figure 4D-F; Table 1), suggesting the possible relevance to PD. Accordingly, we tested the effects of CNTFRα on microglial activation. CNTFRα neutralizing antibody (CNTFRαNAb) was unilaterally injected to block CNTF actions in the ipsilateral rat SN at 1 week after unilateral MFB injection of MPP + . Non-specific IgG was used as a control. CAP-induced neuro-protection and motor recovery were diminished as assessed by the density of TH + fibers by 38% in the striatum ( Figure 5A,G,M), and the numbers of TH + cells by 39% and Nissl + cells by 63% in the SN ( Figure 5B-D,H-J,N) and amphetamine-induced rotations ( Figure 5O) in MPP + -lesioned rats treated with CNTFRαNAb compared with non-specific IgG control rats. In parallel, in the CAP-treated MPP + -lesioned rats, CNTFRαNAb was found to increase immunostaining intensity of OX-6 + ( Figure 5K) and OX-42 + ( Figure 5L) cells in the SN, compared with non-specific IgG-treated respective control rats ( Figure 5E,F). Taken together, these results indicate that CAP-induced astrocytic TRPV1-derived CNTF attenuates microglial activation through microglial CNTFRα.
higher in the rat SN compared to control ( Figure 3A,B). CNTF expression in TH + neurons was significantly reduced ( Figure 3C,D), whereas basal level of CNTF expression in OX-42 + microglia was unchanged ( Figure 3E,F) compare to control. In CAP-treated MPP + -lesioned SN, CNTF expression in both GFAP + astrocytes ( Figure 3A,B) and TH + neurons ( Figure 3C,D) was significantly increased, whereas CNTF expression in OX-42 + microglia was relatively unchanged ( Figure 3E,F) compared to vehicle-treated MPP + -lesioned SN. Collectively, the results indicate that CAP induces expression of CNTF originated from astrocytes, but not microglia. Knockdown of astrocytic TRPV1 reduced CNTF expression in the absence or presence of CAP in the SN of MPP + -lesioned rat (data not shown), supporting our recent findings of astrocytic TRPV1 activation-derived CNTF expression [13]. As CNTF primarily functions via CNTFRα [30], we examined if CNTF could exert its effect on microglia through CNTFRα expressed on microglia. Immunohistochemical analysis showed a significant increase in the level of CNTFRα expression in OX-42 + cells in the SN of MPP + -lesioned rats, compared with control ( Figure 4A-C). CNTFRα in Iba-1 + microglia was also significantly expressed in the SN of PD brain compared to age-matched human control brain (Figure 4D-F; Table  1), suggesting the possible relevance to PD.   Accordingly, we tested the effects of CNTFRα on microglial activation. CNTFRα neutralizing antibody (CNTFRαNAb) was unilaterally injected to block CNTF actions in the ipsilateral rat SN at 1 week after unilateral MFB injection of MPP + . Non-specific IgG was used as a control. CAP-induced neuro-protection and motor recovery were diminished as assessed by the density of TH + fibers by 38% in the striatum ( Figure 5A,G,M), and the numbers of TH + cells by 39% and Nissl + cells by 63% in the SN ( Figure 5B-D,H-J,N) and amphetamine-induced rotations ( Figure 5O) in MPP + -lesioned rats treated with CNTFRαNAb compared with non-specific IgG control rats. In parallel, in the CAPtreated MPP + -lesioned rats, CNTFRαNAb was found to increase immunostaining intensity of OX-6 + ( Figure 5K) and OX-42 + ( Figure 5L) cells in the SN, compared with non-specific IgG-treated respective control rats ( Figure 5E,F). Taken together, these results indicate that CAP-induced astrocytic TRPV1derived CNTF attenuates microglial activation through microglial CNTFRα.

CNTF Derived from TRPV1 Activated Astrocytes Inhibits Microglial ROS Production in the SN In Vivo in MPP + -Lesioned Rat
Activated microglia generates ROS, which may impose oxidative stress on DA neurons in the SN of PD animal models [38,39] and in the SN of patients with PD [3,4,7], leading to DA neuronal cell death. As CNTF seems to exert its neuro-protection through inhibiting microglial activation, we hypothesized that CNTF could inhibit activated microglia-derived ROS production via CNTFRα on microglia. To test this, hydro-ethidine histochemistry was performed for in situ visualization of O 2 − production. MPP + significantly increased O 2 − production (visualized as the fluorescent product of oxidized hydroethidine, i.e., ethidium accumulation) compared to control ( Figure 6A). Microglia-derived O 2 − production was then visualized using double immunofluorescence staining, which showed co-localization of hydroethidine staining within OX-42 + cells ( Figure 6B). In the CAP-treated MPP + -lesioned rats, shTRPV1 significantly increased O 2 − production in microglia of the SN ( Figure 6A,C,D), compared with shCtrl-injected control rats at 2 weeks post MPP + injection. The CAP-induced reduction of O 2 − production in microglia was significantly decreased in the SN of MPP + -lesioned rats treated with CNTFRαNAb compared with non-specific IgG as a control ( Figure 6A,C,D). Taken together, these results indicate that CNTF derived from TRPV1 activated astrocytes inhibits microglial ROS production through CNTFRα on microglia in the SN of MPP + -lesioned rat.
MPP + -lesioned rats, shTRPV1 significantly increased O2 − production in microglia of the SN ( Figure  6A,C,D), compared with shCtrl-injected control rats at 2 weeks post MPP + injection. The CAPinduced reduction of O2 − production in microglia was significantly decreased in the SN of MPP +lesioned rats treated with CNTFRαNAb compared with non-specific IgG as a control ( Figure  6A,C,D). Taken together, these results indicate that CNTF derived from TRPV1 activated astrocytes inhibits microglial ROS production through CNTFRα on microglia in the SN of MPP + -lesioned rat.

Discussion
The present study demonstrates that CNTF originated from TRPV1 activated astrocytes inhibits microglial activation and activated microglia-derived oxidative stress through CNTFRα on microglia and rescues DA neurons in the SN of MPP + -lesioned rats, an animal model of PD (Figure 7). Intriguingly, compared to control, microglia show no change in expression of endogenous CNTF with a significant increase in CNTFRα in MPP + -lesioned rat SN in vivo. Immunohistochemical analysis revealed a significant increase in levels of CNTFRα expression on microglia in postmortem SN tissue from humans with PD, indicating that these findings may have relevance to this disease.
Many studies on PD pathology suggest that glial activation along with neuro-inflammatory processes, like microglial activation, astrogliosis and infiltrated immune cells, contribute to the initiation or progression of PD [38]. DA neurons in the SN are particularly susceptible to oxidative stress because of severe depletion of antioxidants by accumulation of iron and ageing [40][41][42]. Activated microglia generates ROS, which may impose oxidative stress on DA neurons in the SN of both patients with PD [3,4,7] and MPP + -lesioned rats [9], leading to DA neuronal cell death. We recently reported that CAP contributes to DA neuronal survival by inhibiting microglia-derived oxidative stress in the SN of MPP + -lesioned rats [9]. This study also showed that CAP neuro-protection was masked by pharmacological inhibition of TRPV1 with capsazepine, an antagonist of TRPV1, suggesting TRPV1 involvement. This hypothesis was further confirmed and strongly supported by the present data showing that the CAP-induced neuro-protection and behavioral recovery attained through inhibiting microglia-derived oxidative stress are abolished by genetic knockdown of TRPV1 with shTRPV1 [13]. Consistent with our recent report [13], the present study demonstrated selective inhibition of astrocytic TRPV1 function by shTRPV1, indicating that astrocytic TRPV1 activation, in particular, contributes to inhibiting microglia-derived oxidative stress in the SN of MPP + -lesioned rats.

Discussion
The present study demonstrates that CNTF originated from TRPV1 activated astrocytes inhibits microglial activation and activated microglia-derived oxidative stress through CNTFRα on microglia and rescues DA neurons in the SN of MPP + -lesioned rats, an animal model of PD (Figure 7). Intriguingly, compared to control, microglia show no change in expression of endogenous CNTF with a significant increase in CNTFRα in MPP + -lesioned rat SN in vivo. Immunohistochemical analysis revealed a significant increase in levels of CNTFRα expression on microglia in postmortem SN tissue from humans with PD, indicating that these findings may have relevance to this disease. Many studies on PD pathology suggest that glial activation along with neuro-inflammatory processes, like microglial activation, astrogliosis and infiltrated immune cells, contribute to the initiation or progression of PD [38]. DA neurons in the SN are particularly susceptible to oxidative stress because of severe depletion of antioxidants by accumulation of iron and ageing [40][41][42]. Activated microglia generates ROS, which may impose oxidative stress on DA neurons in the SN of both patients with PD [3,4,7] and MPP + -lesioned rats [9], leading to DA neuronal cell death. We recently reported that CAP contributes to DA neuronal survival by inhibiting microglia-derived oxidative stress in the SN of MPP + -lesioned rats [9]. This study also showed that CAP neuroprotection was masked by pharmacological inhibition of TRPV1 with capsazepine, an antagonist of TRPV1, suggesting TRPV1 involvement. This hypothesis was further confirmed and strongly supported by the present data showing that the CAP-induced neuro-protection and behavioral recovery attained through inhibiting microglia-derived oxidative stress are abolished by genetic knockdown of TRPV1 with shTRPV1 [13]. Consistent with our recent report [13], the present study demonstrated selective inhibition of astrocytic TRPV1 function by shTRPV1, indicating that astrocytic TRPV1 activation, in particular, contributes to inhibiting microglia-derived oxidative stress in the SN of MPP + -lesioned rats. CNTF, expressed in astrocytes upon brain injury [21,43,44] has neuro-protective properties [37], inhibiting microglial activation [29], microglia-derived oxidative stress [25] or both. Consistent with our recent data [9,13], CNTF was expressed on astrocytes in the SN of MPP + -lesioned rats and treatment with CNTFRαNAb reduced CAP-induced neuro-protection. Accompanying the lack of neuro-protection was a reduction of CAP effects on microglia-derived oxidative stress, suggesting the inhibitory effects of astrocytic CNTF on microglial activation.
Although our data point to the beneficial effects of CNTF, originating from astrocytes, in particular, the possibility remains that CAP-induced TRPV1 activation produces microglia-originated CNTF, resulting in neuro-protection and inhibition of microglial activation-mediated oxidative stress. This hypothesis is supported by several lines of evidence showing that TRPV1 is expressed on microglia [45], particularly in the SN of MPP + -lesioned rats [13] and that microglia produces CNTF following several types of stimulation, including light-induced retinal degeneration [22] and cuprizone-induced demyelination [23]. However, our immunohistochemical analysis revealed little CNTF expression in the SN microglia of MPP + -lesioned rats with or without CAP administration. It is therefore likely that CNTF is primarily expressed in astrocytes, but not in microglia under our experimental conditions.
Although CNTF appeared to exert neuro-protection via CNTFRα located on DA neurons [13], several in vitro studies have shown CNTFRα expression on microglia [25,33]. The present study demonstrated a substantial expression of CNTFRα on microglia in the SN of MPP + -lesioned rats and a CNTFRαNAb-induced reduction in CAP effects, suggesting that CNTFRα located on microglia could be involved in neuronal survival through inhibition of microglial activation. This microglial CNTFRα-mediated effect would be in addition or parallel to CNTF actions on DA neurons. This observation appears to be important under pathological conditions, given that microglia-derived oxidative stress plays a pivotal role in the initiation of PD, its progression, or both. Therefore, it is likely that CNTF derived from CAP-induced astrocytic TRPV1 activation can act as an endogenous neurotrophic factor, contributing both directly and indirectly to DA neuronal survival in the MPP + rat model of PD.
We have recently shown that CNTFRα was expressed in TH + cells in the SN of intact rat, and CNTFRα expression in TH + cells was significantly decreased in the SN of MPP + -lesioned rat without change of total CNTFRα expression [13]. Regarding this, in the present study, the total expression of CNTFRα is not changed, whereas there is significantly increased CNTFRα expression within OX-42 + cells in the SN of MPP + -lesioned rat. Accordingly, increased level of CNTFRα expression in OX-42 + cells may reflect decreased levels of CNTFRα expression in TH + cells in the SN of MPP + -lesioned rat.
Taken together with our recent data (TRPV1 and CNTF expression on astrocytes, and CNTFRα expression on DA neurons) [13], CNTFRα expression on microglia in the SN of both humans with PD and MPP + -lesioned rats suggests that endogenous CNTF derived from TRPV1 activated astrocytes has therapeutic potential to restore motor function in the treatment of PD by inhibiting activated microglia-mediated oxidative stress and/or preventing the progressive degeneration of DA neurons.

Animals
All

Human Samples
Human brain tissues (age-matched control and Parkinson's disease subjects obtained from the Victoria Brain Bank Network (VBBN), Table 1) were prepared to immunohistochemical staining [13]. Human tissue experiments were done in accordance with the Institutional Review Board of Kyung Hee University Hospital (KMC IRB 1234-04, 30 November 2012).

Stereotaxic Surgery and Tissue Preparation
Stereotaxic surgery under chloral hydrate was performed as described [13]. Using coordinates relative to the bregma, stereotaxic injections of MPP + (right medial forebrain bundle (MFB); A/P -3.  [46]. Lentivirus was injected immediately after MPP + injection. CNTFRαNAb was injected at 1 week post MPP + injection. As described [47], animals were anesthetized with chloral hydrate (360 mg/kg, i.p. injection) at the indicated time points after injection and transcardially perfused and fixed with 4% paraformaldehyde dissolved in 0.1 M phosphate buffer (PB). Brains were removed from the skull, post fixed overnight at 4 • C in buffered 4% paraformaldehyde, and stored at 4 • C in 30% sucrose solution until they sank. Brains were frozen sectioned using a sliding microtome into 40 µm coronal sections and collected in six separate series of SN and striatum.

Capsaicin (CAP) Injection
Before choosing route and dose of CAP injection used here, we measured behavioral changes to avoid painful activation (behavior) of peripheral TRPV1-containing neurons ( Table 2). All rats intraperitoneally received a single injection of capsaicin (1, 5, and 10 mg/kg) per day for 7 days. Based on behavioral symptoms and previous results [13,48], CAP (1 mg/kg, i.p. injection; a single injection/day for 7 days, Sigma) was injected at 1 week and 1 day post MPP + injection. Last for more than 10 min after capsaicin injection for 7 days Some of rat (N = 9) was dead in less than 1 min +, slight response; ++, medium response; +++, strong response.

Rotational Behavior Test
As described [13], D-Amphetamine (5 mg/kg, i.p. injection) was used to monitor ipsilateral rotation in rats with unilaterally MPP + administration. Rats that exhibited ipsilateral rotations, indicative of an effective lesion were randomly selected for treatment with the CAP (1 mg/kg, i.p. injection) or vehicle each day for 7 days. The ipsilateral rotations were counted for 1 h. Rats were transcardially perfused for immunostaining at 30 min after the last rotation experiment.

Stereological Estimation
As described [13], the total number of TH + neurons were counted in the various animal groups using the optical fractionator method performed on a bright field microscope (Olympus Optical, BX51, Tokyo, Japan) using Stereo Investigator software (MBF Bioscience, Williston, VT, USA). This unbiased stereological method of cell counting is not affected by either the reference volume SN or the size of the counted elements (neurons). According to Paxinos and Watson [46], all assessments were analyzed in six separated sections in the front and the rear of each target area. As described previously [47], the quantification area for SN are -4.56 mm to -6.60 mm from bregma in the total area.

Morphological Analysis
Optical densities of the TH + striatal fibers were measured using Science Lab 2001 Image Gauge (Fujifilm, Tokyo, Japan) [13]. According to Paxinos and Watson [46], all assessments were analyzed in six separated sections in the front and the rear of each target area. The quantification areas for striatal fibers are +2.16 mm to +0.12 mm from bregma in the total area.

Image J Analysis
Imaging data were analyzed in Image J (National Institutes of Health) as described recently [13]. Image J with co-localization plugin was used to quantify immunofluorescence and with color deconvolution plugin was used to quantify chromogenic signal intensity on image.

In Situ Detection of O 2
− and O 2 − -Derived Oxidants As previously described [9], hydroethidine histochemistry was performed for in situ visualization of O 2 − and O 2 − -derived oxidants. At 2 weeks post MPP + injection, hydroethidine (1 mg/mL in PBS containing 1% dimethyl sulfoxide; Sigma) was administered through tail vein. After 45 min, the animals were transcardially perfused, fixed and brain tissues (40 µm thickness) were mounted on gelatin-coated slides. The oxidized hydroethidine product, ethidium, examined by confocal microscopy and merged with OX-42 antibody for double-immunofluorescence staining.

Statistical Analysis
All values are expressed as mean standard error of the mean. Statistical significance (p < 0.05 for all analysis) was assessed by One way ANOVA Newman-Keuls analyses and Student unpaired t-test (GraphPad Software, San Diego, CA, USA). F values were provided in each figure legends.