ATF6alpha Promotes Astroglial Activation and Neuronal Survival in a Chronic Mouse Model of Parkinson’s Disease

Accumulating evidence suggests a crucial role for the unfolded protein response (UPR) in Parkinson’s disease (PD). In this study, we investigated the relevance of the UPR in a mouse model of chronic MPTP/probenecid (MPTP/P) injection, which causes severe and persistent degeneration of dopaminergic neurons. Enhanced activation of the UPR branches, including ATF6α and PERK/eIF2α/ATF4, was observed after MPTP/P injections into mice. Deletion of the ATF6α gene accelerated neuronal degeneration and ubiquitin accumulation relatively early in the MPTP/P injection course. Surprisingly, astroglial activation was strongly suppressed, and production of the brain-derived neurotrophic factor (BDNF) and anti-oxidative genes, such as heme oxygenase-1 (HO-1) and xCT, in astrocytes were reduced in ATF6α −/− mice after MPTP/P injections. Decreased BDNF expression in ATF6α −/− mice was associated with decreased expression of GRP78, an ATF6α-dependent molecular chaperone in the ER. Decreased HO-1 and xCT levels were associated with decreased expression of the ATF4-dependent pro-apoptotic gene CHOP. Consistent with these results, administration of the UPR-activating reagent tangeretin (5,6,7,8,4′-pentamethoxyflavone; IN19) into mice enhanced the expression of UPR-target genes in both dopaminergic neurons and astrocytes, and promoted neuronal survival after MPTP/P injections. These results suggest that the UPR is activated in a mouse model of chronic MPTP/P injection, and contributes to the survival of nigrostriatal dopaminergic neurons, in part, through activated astrocytes.


Introduction
Parkinson's disease (PD) is a progressive neurodegenerative disease pathologically characterized by the selective loss of nigrostriatal dopaminergic neurons and the presence of protein aggregates, known as Lewy bodies [1]. Although the etiology of PD is not fully understood, several genetic and environmental factors have been discovered that are utilized to model PD in experimental animals [2]. 1-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), an inhibitor of mitochondrial complex I, induced human Parkinsonism [3], and has, therefore, been widely used to generate a PD model in animals. MPTP is highly lipophilic and crosses the blood-brain barrier. In the brain, it is converted to the active form, 1-methyl-4-phenylpyridinium (MPP + ) by astrocytes, and is taken up by dopaminergic neurons through the dopamine transporter (DAT). Accumulating evidence suggests that, compared with acute and subacute MPTP injection, chronic administration of MPTP with probenecid (MPTP/P) into mice causes more severe neurodegeneration and dopamine (DA) depletion that resembles human PD [4,5].
Although MPTP causes oxidative stress and energy depletion because of impaired mitochondrial function, recent studies suggest that MPTP also causes endoplasmic reticulum (ER) stress, a type of intracellular stress that is characterized by the accumulation of unfolded proteins in the ER. ER stress occurs when cells are in conditions such as glucose starvation (energy depletion), oxygen deprivation, protein modification inhibition, and disturbance of Ca 2+ homeostasis. Eukaryotic cells respond to ER stress by activating a set of pathways known as the unfolded protein response (UPR) [6]. In mammals, the UPR is transmitted through 3 types of sensor proteins; double-stranded RNA-activated protein kinase (PKR)-like ER kinase (PERK), inositol-requiring enzyme 1a (Ire1a), and activating transcription factor 6a (ATF6a) [6]. Ire1a and ATF6a downstream genes include molecular chaperones in the ER, such as glucose-regulated protein78 (GRP78), and oxygen-regulated protein 150 (ORP150), and ER-associated degradation (ERAD) molecules such as Derlins, ER degradation enhancing alpha-mannosidase-like protein (EDEM), and homocysteine-inducible endoplasmic reticulum stress protein (Herp). In contrast, PERK downstream genes include eukaryotic translation initiation factor 2 (eIF2a), which suppresses general protein synthesis to reduce protein loads into the ER, and activating transcription factor 4 (ATF4), which upregulates the expression of anti-oxidative genes such as heme oxygenase 1 (HO-1) and cystine/glutamate antiporter (xCT). PERK also upregulates the pro-apoptotic transcriptional factor C/EBP homologous protein (CHOP) [7]. Cell culture models [8,9] and the acute MPTP injection models [10,11] demonstrated the UPR has important roles in promoting neuronal survival against MPTP neurotoxicity. Furthermore, a recent report demonstrated that MPP + -associated oxidative stress enhanced the interaction between phosphorylated p38 mitogen-activated protein kinase (p38MAPK) and ATF6a, causing increased transcriptional activity of ATF6a [12]. These findings suggest an important communication between the oxidative stress response and the UPR in PD pathogenesis.
We initiated this study by estimating the UPR activation status after chronic MPTP/P injections into wild-type mice. We compared neurodegeneration, protein aggregation, and glial activation levels between wild-type and ATF6a 2/2 mice. Finally, we estimated the neuroprotective property of the UPRactivating compound, tangeretin (5,6,7,8,IN19), in the mouse model of chronic MPTP/P injection.

Ethics Statement
All animal care and handling procedures were approved by the Animal Care and Use Committee of Kanazawa University (No. 71241-1).

Mice and Chronic MPTP/P Injection PD Model
ATF6a 2/2 mice were generated as described previously [14], and backcrossed to the C57BL/6 strain more than 8 times. Wildtype and ATF6a 2/2 male mice (aged 12-15 weeks and weighing 26-30 g) were used for the experiments. The chronic MPTP/P injection PD model was created as described previously with some modifications. In brief, mice were administered MPTP (20 mg/kg in saline, subcutaneously) and probenecid (250 mg/kg in DMSO, intraperitoneally) twice a week for 5 weeks [4]. At the indicated times, brain samples were prepared for histological analyses, RT-PCR/quantitative real time RT-PCR (qRT-PCR) and Western blotting as described. In some experiments, mice were administered tangeretin (10 mg/kg, per oral, in saline including 10% Cremophore EL and 10% DMSO) or the dissolving solution (vehicle) 24 h and 2 h before MPTP/P injections [11].

RT-PCR and Quantitative Real Time RT-PCR (qRT-PCR)
Total RNA was extracted from the ventral midbrain or caudate putamen (CPu) of each mouse using RNAzolHRT (Molecular Research Center Inc, Cincinnati, OH). RT reactions containing 1 mg of total RNA were performed using PrimeScript (Takara, Shiga, Japan). The individual cDNA species were amplified in a reaction mixture containing 1 unit of Taq DNA polymerase (Takara) and specific primers for ATF6, GRP78, ORP150, ATF4, HO-1, CHOP, X-box binding protein 1 (XBP-1), Sec61b, GFAP, Iba1, andb-actin as described previously. In some experiments, cDNA derived from cultured astrocytes treated with tunicamycin for 8 h [15] was used as a positive control for XBP1 activation. For qRT-PCR, cDNA was amplified with THUNDERBIRD TM SYBR qPCRH Mix (TOYOBO CO, LTD, Osaka, Japan) by using specific primers for brain-derived neurotrophic factor (BDNF), leukemia inhibitory factor (LIF), interleukin-6 (IL-6), GRP78, ORP150, CHOP, xCT, manganese superoxide dismutase (MnSOD), and b-actin. The comparative Ct method was used for data analyses with MxPro 4.10 (Agilent Technologies, Santa Clara, CA). Values for each gene were normalized to b-actin expression levels.

Histological and Immunohistochemical Analyses
Brains were removed from mice after perfusion with 4% paraformaldehyde, and postfixed in the same fixative for 4 hours at 4uC. After cryoprotected in 30% sucrose, brains were cut in serial coronal 10 mm-thick sections containing the CPu (from Bregma+1.34 mm to Bregma+0.26 mm) and the midbrain covering the whole SNpc (from Bregma-2.80 mm to Bregma-3.80 mm) on a cryostat, and mounted in series on ten slides (around ten sections were mounted on each slide). One out of these ten slides, representing a set of sections 100 mm apart, were processed for immunohistochemistry, and the negative control, in which the primary antibody was omitted, was performed in parallel with each procedure. Primary antibodies used were; anti-TH (Sigma), anti-GRP78, anti-HO-1, anti-GLT-1, anti-Ubiquitin (StressGen), anti-cleaved caspase 3 (Cell Signaling Technology, Danvers, MA), anti-BDNF, anti-GFAP, anti-Iba1. In some cases, the cell nucleus was visualized with DAPI (Sigma), and Cresyl violet (Sigma) was used for counterstaining. Appropriate Alexa Fluor 488, Cy3conjugated IgG or peroxidase-conjugated IgG was used as a secondary antibody. Confocal images were obtained by using Nikkon EZ-C1. In the process of apoptosis, cleaved caspase 3 was observed both in the cytosol and nucleus. Expression of GRP78, ORP150, HO-1, and BDNF was immunohistochemically detected mainly in the cell body of neurons and/or astrocytes, and that of GLT-1 was detected in the process of astrocytes.

Image Quantification
Quantification of the RT-PCR, Western Blotting, and immunohistochemical analyses were performed using Image J (version 1.42, Wayne Rasband, National Institutes of Health). The number of TH positive neurons in the SNpc was counted in five representative sections out of ten sections mounted on one slide, which covered the whole SNpc. Statistical analyses were performed using Bonferroni/Dunn test following a one-way ANOVA.  To evaluate the activation status of the 3 major UPR pathways after MPTP/P injections, we analyzed the expressions of genes involved in each pathway by using RT-PCR ( Fig.1 B I, II, III). The transcripts of ATF6a and its downstream genes, such as GRP78 and ORP150, were enhanced 2.5, 2.3, and 2.4 fold, respectively, 8 h after the first MPTP/P injection. The increased expression tended to last until at least fifth MPTP/P injection ( Fig. 1 B I). Transcripts of ATF4 and its downstream genes, such as CHOP and HO-1, which is also a target of NF-E2-related factor 2 (Nrf2), were enhanced 2.3, 2.5, and 7.6 fold, respectively, 8 h after the first MPTP/P injection. However, the expression levels dropped to 1.7, 1.4, and 3.9 fold, respectively, 24 h after the first injection ( Fig. 1 B II). In contrast, the Ire1/XBP-1 pathway, and XBP-1, which can be detected by the altered splicing pattern [6], was not activated. In addition, the downstream gene Sec61b was not upregulated after MPTP/P injection ( Fig. 1 B III).

Accelerated Neurodegeneration and Ub Accumulation in ATF6a 2/2 Mice after MPTP/P Injections
To evaluate the neuroprotective role of the UPR in the chronic MPTP/P injection model, we immunehistochemically compared nigrostriatal neuronal degeneration between wild-type and ATF6a 2/2 mice ( Fig. 2 A I, II). In the control condition (without MPTP/P administration), the number of TH-positive neurons in the SNpc and the intensity of TH in the CPu were not significantly different among the wild-type and ATF6a-deficient mice. In contrast, in the early course of MPTP/P injections (2 nd and 3 rd injections), the number of TH-positive neurons in the SNpc and the intensity of TH in the CPu were significantly lower in ATF6a 2/2 mice compared to wild-type mice. Consistent with these results, higher numbers of activated caspase-3-positive, THpositive neurons were observed in ATF6a 2/2 mice (74%) compared to wild-type mice (47%; Fig. 2 A III). The specificity of the antibody and the appropriate immunoreactivity of the antigen were confirmed by the negative control experiment where primary antibody was omitted (Fig. S 2 A) and the serial photograph of the confocal images ( Fig. S 2 B), respectively. In the later injections (6 th -10 th injections), however, the nigrostriatal dopaminergic neurons had degenerated to similar levels in both cohorts ( Fig.2 A I, II). Egawa et al. recently demonstrated the presence of Ubpositive inclusions in ATF6a 2/2 mice after acute MPTP injection [12]. Therefore, we assessed Ub accumulation in our model. In the control condition, slight Ub immunoreactivity in the cell body was observed in TH-positive neurons of both cohorts ( Fig. 2 B I, upper row). However, after the 3 rd MPTP/P injection, Ub accumulation was observed in the degenerating dopaminergic neurons in the ATF6a 2/2 SNpc, but not in the wild-type SNpc ( Fig. 2 B I, arrows). In 29% (4 of 14) of ATF6a 2/2 mice, Ubpositive inclusions were observed in the neurons after 10 th MPTP/ P injection (Fig. 2 B II), but not in wild-type neurons (data not shown). These inclusions were negative or very weak for asynuclein immunoreactivity (Fig. 2 B III), as previously described [12]. Impaired Astroglial Activation in ATF6a 2/2 mice after MPTP/P Injections Accelerated neurodegeneration was observed in ATF6a 2/2 mice predominantly when glial cells were activated (Fig. S1). Therefore, glial activation was assessed in wild-type and ATF6a 2/2 mice. Immunohistochemical analyses in the SN and CPu revealed that in the control condition, GFAP-positive astrocytes were sparsely observed in both mouse cohorts ( Fig. 3 A I, II, arrows). However, in the period from the 2 nd to 3 rd MPTP/P injection, the features of astroglial activation (enlarged cell bodies and thick processes) in the SN and CPu were observed more frequently in wild-type mice compared to ATF6a 2/2 mice ( Fig. 3 A I, II, 2 nd and 3 rd rows). In the wild-type SN, astrocytes became enlarged in the SN pars reticulata (SNpr) first (arrowheads), and then penetrated into the SNpc (asterisks), but ATF6a 2/2 astrocytes were not enlarged after MPTP/P injections. In the CPu, wild-type astrocytes near the lateral ventricle (arrows) and corpus callosum (data not shown) became enlarged and, almost simultaneously, spread over the CPu, but again, ATF6a 2/2 astrocytes were not enlarged after MPTP/P injections. Consistent with the immunohistochemical observations, Western blot analyses revealed enhanced GFAP expression in wild-type mice, but not in ATF6a 2/2 mice, after the 2 nd and 3 rd MPTP/P injections ( Fig.4 C I). In contrast to high levels of astroglial activation, microglial activation was modest in this model, and the differences in the microglia morphology were not clear between wild-type and ATF6a 2/2 mice after the 2 nd MPTP/P injection (Fig. 3 A II).
Activated astrocytes contribute to neuroprotection in several ways, including neurotrophic factor synthesis, enhancement of anti-oxidative systems, and glutamate metabolism [16,17]. Therefore, we compared the expression of BDNF (a neurotrophic factor), HO-1 (an anti-oxidative gene), and GLT-1 (a glutamate transporter) in wild-type and ATF6a 2/2 mice. Immunohistochemical analyses revealed that BDNF and HO-1 expression (Fig. 3 B I, II), but not GLT-1 expression (Fig. S2 C), were higher after MPTP/P injections in wild-type astrocytes compared with ATF6a 2/2 astrocytes in the CPu.

Reduced UPR Levels and Gene Expression in ATF6a 2/2 Astrocytes after MPTP/P Injections
To determine whether impaired astroglial activation was associated with reduced UPR levels in ATF6a 2/2 mice, expression of astrocyte-derived neurotrophic factor (BDNF), anti-3-positive, TH-positive neurons. The relative number of activated caspase 3-positive, TH-positive neurons in the SNpc are also shown in the graph. Values shown are the mean 6 S.D. Scale bars = 50 mm (I), 100 mm (II), 30 mm (III). *P,0.05, compared between wild-type and ATF6a 2/2 mice (n = 4). B, Ub accumulation (I) and intracellular inclusion bodies (II) in ATF6a 2/2 mice after MPTP/P injections. (I) Brain sections, including the SN, from mice that were injected with or without MPTP/P were immunostainined with the TH and Ub antibodies. Nuclei were stained with DAPI. Note that in ATF6a 2/2 mice, Ub accumulation was observed in cells with reduced TH immunoreactivity (arrows). (II, III) Brain sections, including the SN, from ATF6a 2/ 2 mice that were injected 10 times with MPTP/P were immunostained with the Ub (II) or a-synuclein antibody, followed by counterstaining with cresyl violet. Arrows indicate Ub-positive inclusion bodies. Scale bars = 30 mm (I), 20 mm (II, III). doi:10.1371/journal.pone.0047950.g002 oxidative genes (HO-1, xCT, and MnSOD), astrogliosis-inducing factors (LIF and IL-6), and the glutamate transporter (GLT-1) was compared with the expression of ATF6a-dependent molecular chaperones in the ER (GRP78 and ORP150) and the PERK/ ATF4-dependent pro-apoptotic transcriptional factor (CHOP). Expression of the BDNF protein in ATF6a 2/2 mice was 62% and 68% of that in wild-type mice in the control condition and after the 3 rd MPTP/P injection, respectively ( Fig. 4 A I). Similarly, expression of the GRP78 protein in ATF6a 2/2 mice was 60% and 55% of that in wild-type mice in the control condition and after the 3 rd MPTP/P injection, respectively ( Fig. 4 A I). The expression level of BDNF, GRP78, and ORP150 mRNA in ATF6a 2/2 mice was 71%, 77% and 79% of that in wild-type mice in the control condition, and 150%, 74%, and 83% at twenty-four hours after the 1 st MPTP/P injection ( Fig. 4 A II). These results suggest that the deletion of the ATF6a gene may suppress BDNF expression by transcriptional (as in the control condition) and translational/posttranslational (as in the MPTP/Pinjected conditions) mechanisms.
Regarding the anti-oxidative genes, expression of the PERK/ ATF4-dependent genes, HO-1 and xCT, was lower in ATF6a 2/ 2 mice than wild-type mice, at both the protein and mRNA level after MPTP/P injection, although they were not significantly different in the control conditions ( Fig. 4 B I, II). Similar results were obtained with CHOP ( Fig. 4 B I, II). In contrast, expression of the PERK/ATF4-independent anti-oxidative gene MnSOD was not significantly different between wild-type and ATF6a 2/2 mice, either in the control or after MPTP/P injection (data not shown).
Several cytokines and growth factors secreted from damaged neurons or other cells in the brain can induce astroglial activation [18]. Therefore, we examined the expression level of these genes in wild-type and ATF6a 2/2 mice. LIF and IL-6 expression was significantly enhanced, both at the protein ( Fig. 4 C I, II) and mRNA ( Fig. 4 C III) level, after MPTP/P injection in wild-type mice, but not in ATF6a 2/2 mice. Importantly, increased LIF and IL-6 expression in wild-type mice preceded GFAP upregulation ( Fig. 4 C I). Although reduced expression of LIF and IL-6 is likely not associated with reduced GRP78 or CHOP expression in ATF6a 2/2 mice, these results suggest that ATF6a may transcriptionally regulate the expression of astrogliosis-inducing factors after MPTP/P injection.
Consistent with the immunohistochemical results, GLT-1 expression was not significantly different between wild-type and ATF6a 2/2 mice in the control condition or after MPTP/P injection (Fig. S2 D).

UPR Activation and Neuroprotection by IN19
We previously reported that IN19 activated the UPR and protected dopaminergic neurons against acute MPTP administration [11]. In this study, we evaluated the neuroprotective property of IN19 in the chronic MPTP/P injection model. Immunohistochemical analyses revealed that IN19 administration (86IN19 administration p.o./2 weeks) upregulated the expression of ORP150 (Fig. 5 A I, II) and GRP78 (Fig. S3 A) in TH-positive dopaminergic neurons and GFAP-positive astrocytes without reducing the number of TH-positive neurons or the intensity of TH. RT-PCR analyses revealed enhanced activation of ATFa and PERK/ATF4 pathways, but not of the Ire1/XBP1 pathway, in IN19-administrated mice (Fig. 5 B I, II, III). Unlike MPTP/P administration, IN19 administration did not upregulate GFAP or Iba1 (Fig. 5 B IV), suggesting that the effect of IN19 on UPR was not mediated by general neuronal damage. IN19 also enhanced eIF2a phosphorylation in dopaminergic neurons (Fig. S3 B), as previously described [11]. Next, we assessed the neuroprotective property of IN19 after MPTP/P injections. When mice were given IN19 (10 mg/kg, p.o. in saline, including 10% Cremophore EL and 10% DMSO) 24 h and 2 h before MPTP/P injection, the number of TH-positive neurons in the SN and the intensity of TH in the CPu were significantly increased (Fig. 6 A). Consistently, the number of activated caspase 3-positive, TH-positive neurons (Fig. 6 B) decreased in the SN, and expression of BDNF in the CPu increased in the astrocytes of mice given IN19 after MPTP/P injection ( Fig. 6 C I, II). Importantly, expression of GFAP in the CPu also mildly, but significantly, increased in mice given IN19 after MPTP/P injection ( Fig. 6 C I, II), suggesting that IN19 may protect dopaminergic neurons, at least in part, through the activated astrocytes after MPTP/P administration.

Discussion
In this study, we first demonstrated the activation of the UPR in a chronic MPTP/P injection model. Of the 3 UPR branches, the ATF6a and PERK/eIF2a/ATF4 pathways were preferentially activated after MPTP/P injections (Fig. 1 B). We also observed a trend that the PERK/eIF2a/ATF4 pathway was highly activated after the 1 st MPTP/P injection (8 h after injection; Fig. 1 B II), but the ATF6 pathway was activated for longer periods over the course of the MPTP/P injections (1 st through 5 th injections; Fig. 1 B I). These results are consistent with those of previous reports demonstrating differential activation between the 3 UPR branches after PD-related stresses caused by MPP + or 6-OHDA in cultured cells [9,19]. Taken together with a recent report, which demonstrated a direct link after MPP + treatment between p38 MAP kinase and ATF6a [12], these findings suggest critical roles for the ATF6a and PERK/eIF2a/ATF4 pathways as defense systems against PD-related neurotoxins.
Analyses of wild-type and ATF6a 2/2 mice showed accelerated degeneration of the nigrostriatal neurons in ATF6a 2/2 mice (Fig. 2 A I, II, III) after the earlier MPTP/P injections (2 nd and 3 rd injections), but not after the later injections (6 th through10 th injections). Similarly, Ub accumulation was observed in ATF6a 2/2 dopaminergic neurons after the early MPTP/P injections (2 nd and 3 rd injections; Fig. 2 B I). However, Ub-positive inclusions, which were abundantly observed in ATF6a 2/2 mice after acute MPTP injection [12], were observed only in 29% of ATF6a 2/2 mice after the last injection (10 th injection; Fig. 2 B II). These results suggest that ATF6a may contribute to neuronal survival and protein aggregation regulation in the early stages, but not in the late stages, of PD. Regarding the cell death pathways Figure 3. Impaired astroglial activation in ATF6a2/2 mice after MPTP/P injections. A, GFAP, TH, and Iba1 expression after MPTP/P injections. Brain sections, including the SN (I) or CPu, (II) from wild-type and ATF6a2/2 mice that were injected with or without MPTP/P were immunostained with GFAP, TH, and Iba1 antibodies. Nuclei were stained with DAPI. Arrows and arrowheads indicate non-activated and activated astrocytes, respectively. Asterisks indicate activated astrocytes penetrating into the SNpc. Scale bars = 30 mm. B, BDNF (I) and HO-1 (II) expression in astrocytes after MPTP/P injections. Brain sections, including the CPu, from wild-type and ATF6a 2/2 mice that were injected with or without MPTP/P were immunostained with the BDNF, HO-1, and GFAP antibodies. Nuclei were stained with DAPI. The relative intensity of BDNF (I) or HO-1 (II) in the GFAP-positive cells is shown in the graph. The intensity of the signals derived from wild-type mice without MPTP/P injection is designated as one. Values shown are the mean 6 S.D. *P,0.05, compared between wild-type and ATF6a2/2 mice (n = 4). Scale bars = 30 mm. doi:10.1371/journal.pone.0047950.g003 involved in our model, increased expression of activated caspase-3 was observed in ATF6a 2/2 dopaminergic neurons after MPTP/P injections. However, the expression of CHOP, a medi-ator of ER stress-induced cell death, was reduced in ATF6a 2/2 mice compared with wild-type mice after MPTP/P injections ( Fig. 4 A II, C II). These data suggest that the accelerated neuronal Figure 4. Reduced UPR levels and gene expression in ATF6a2/2 mice after MPTP/P injections. Protein expression of neurotrophic factor (A I), anti-oxidative genes (B I), astrogliosis-inducing factor (C I, II), and the UPR-target genes (A I, B I). Protein extracts from brains (CPu) of wild-type or ATF6a 2/2 mice that were injected or not injected with MPTP/P were subjected to Western blotting with the indicated antibodies (A I, B I, C I), or subjected to IL-6 ELISA (C II). Relative intensities are shown in the graphs. The intensity of the genes from wild-type mice without MPTP/P administration is designated as one. Values shown are the mean 6 S.D. *P,0.05, compared with mice without MPTP/P administration. # P,0.05, compared between wild-type and ATF6a 2/2 brains (n = 4). Transcripts of neurotrophic factor (A II), anti-oxidative genes (B II), astrogliosis-inducing factors (C III), and the UPR-target genes (A II, B II). Total RNA (1mg) isolated from wild-type or ATF6a 2/2 brains (CPu) at indicated times after 1 st MPTP/P injection was subjected to qRT-PCR with specific primers for the indicated genes. The relative intensity of the genes from wild-type mice not administered MPTP/P is designated as one. Values shown are the mean 6 S.D. *P,0.05, **P,0.01, compared with the mice without MPTP/P administration. # P,0.05, compared between wild-type and ATF6a 2/2 brains (n = 4). doi:10.1371/journal.pone.0047950.g004  death in ATF6a 2/2 dopaminergic neurons after MPTP/P injection may include ER stress-induced, CHOP-independent neuronal death. This is consistent with the finding of a previous report demonstrating that null mutation of CHOP did not protect against neuronal loss in a chronic MPTP/P model [20].
Interestingly, astroglial activation was strongly suppressed, and biosynthesis of the neurotrophic factor BDNF and the antioxidative gene heme oxygenase-1 (HO-1) was reduced in ATF6a 2/2 mice after MPTP/P injections ( Fig. 3 B I, II). Astrocytes are ubiquitous in the brain, and upon central nervous system insult, undergo molecular and morphological changes, referred to as reactive astrogliosis or astroglial activation [18]. Activated astrocytes enhance the neuronal survival by secreting neurotrophic factors or antioxidants, as well as by reducing glutamate levels in extracellular spaces [16,17]. In our model, reduced BDNF expression was observed in ATF6a 2/2 mice and was associated with reduced GRP78 expression (Fig. 4 A, B). These results were consistent with those of our previous reports that ATF6adependent molecular chaperones, such as GRP78 and ORP150, promote the maturation of neurotrophic factors in the ER [21,22]. Expression of the PERK/ATF4-dependent anti-oxidative genes HO-1 and xCT, but not the PERK/ATF4-independent antioxidative gene MnSOD was also reduced at both the protein and mRNA level in ATF6a 2/2 mice after MPTP/P injections (Fig. 4  A). These results suggest that some of PERK/ATF4-dependent anti-oxidative genes are also transcriptionally regulated by ATF6a, similar to PERK/ATF4-dependent pro-apoptotic gene CHOP [23]. Although it is currently unknown which steps in astroglial activation are impaired in ATF6a 2/2 mice, it is possible that extracellular signals including LIF and IL-6, from damaged neurons or other cells in the brain, are too low to promote activation in ATF6a 2/2 astrocytes. Alternatively, it is also possible that enhanced levels of ER stress in ATF6a 2/2 astrocytes compromised intracellular signals which are important for the astroglial activation, as was recently reported for hepatocytes [24].
Consistent with the results from ATF6a 2/2 mice, administration of the IN19 to wild-type mice enhanced UPR-target gene expression, including ORP150 and GRP78, in both nigrastriatal neurons and astrocytes, and facilitated neuronal survival after MPTP/P injection. These results are consistent with those of previous reports demonstrating that IN19 can distribute into the brain after oral administration [25], and protect cells in both the ER stress model and acute MPTP injection model [11,25]. Although IN19 alone did not cause astrogliosis (Fig. 5 B IV), IN19 administered in the course of MPTP/P injections enhanced expression of GFAP ( Fig. 6 C I, II) mildly, but significantly, suggesting that IN19 may protect dopaminergic neurons, at least in part, through the activated astrocytes after MPTP/P administration. A recent report demonstrated that Salubrinal, a com-pound that regulates ER stress by activating the eIF2a/ATF4 pathway, attenuated disease manifestation in the A53T asynuclein-overexpressed PD model [26]. These results emphasize the protective role of the UPR in PD.
In conclusion, we found that the UPR branches were activated in a mouse model of chronic MPTP/P injection, and they contributed to nigrostriatal neuronal survival, at least in part, through activated astrocytes. Further studies to dissect the neuronglial association through the UPR should provide novel therapeutic window for PD and other neurodegenerative diseases. Figure S1 Astrocyte and microglia activation in a mouse model of chronic MPTP/P injection. Total RNA (1 mg) isolated from brain samples, including the ventral midbrain, after MPTP/P injections was subjected to RT-PCR with specific primers for GFAP (activated astrocytes) and Iba1 (activated microglia) as described in Fig. 1 B. The relative intensity of the bands derived from mice without MPTP/P administration is designated as one. Values shown are the mean 6 S.D. *P,0.05, **P,0.01 compared with mice not administered MPTP/P (n = 4). (TIF) Figure S2 Immunohistochemical analysis of wild-type and ATF6a 2/2 brains after MPTP/P injections. A, Negative control experiment. Brain sections, including the SN from wild-type and ATF6a 2/2 mice after MPTP/P injection, were incubated with mouse anti-TH antibody, followed by incubation with both anti-rabbit Alexa Fluor 488 and Cy3conjugated anti-mouse IgG. Scale bars = 30 mm (low mag.), 15 mm (high mag.).B, Serial photograph of activated caspase 3 in wild-type mice after MPTP/P injections. Brain sections, including the SN from wild-type mice after MPTP/P injection, were immunostained with TH and activated caspase 3 antibodies. The nuclei are stained with DAPI. Scale bar = 15 mm. C, Immunohistochemical analyses of GLT-1. Brain sections, including the CPu, from wild-type and ATF6a 2/2 mice after MPTP/P injections were immunostained with GLT-1 and GFAP antibodies. Nuclei were stained with DAPI. Scale bar = 30 mm. D, Western blot analyses for GLT-1. Protein extracts from brains (CPu) of wild-type and ATF6a 2/2 mice that were injected or not injected with MPTP/P were subjected to Western blot with the GLT-1 antibody. (TIF) Figure S3 UPR activation and astrogliosis after tangeretin (IN19) administration. GRP78 expression (A), eIF2a activation (B) in the SN. Brain sections, including the SN, from wild-type mice administered or not administered IN19 for 2 weeks (4 times/week) were immunostained with the GRP78, phosphorylated eIF2a, and TH antibodies. Arrows indicate activated compared between vehicle-and IN19-administered mice (n = 4). Scale bars = 100 mm (SN), 200 mm (CPu). B, The effect of IN19 on caspase-3 activation after MPTP/P injections. Brain sections, including the SN from wild-type and ATF6a 2/2 mice that were injected with MPTP/P, in the presence or absence of IN19, were immunostained with the activated caspase 3 and TH antibodies. Nuclei were stained with DAPI. The relative number of activated caspase 3-positive, TH-positive neurons are shown in the graph. Values shown are the mean 6 S.D. *P,0.05, **P,0.01, compared with mice without MPTP/P administration. # P,0.05, compared between vehicle-and IN19-administered mice (n = 4). Scale bar = 30 mm. C, The effect of IN19 on BDNF expression. (I) Brain sections, including the CPu, from wild-type mice that were injected or not injected with MPTP/P, in the presence or absence of IN19, were immunostained with the BDNF and GFAP antibodies. Nuclei were stained with DAPI. The relative intensity of BDNF or GFAP is shown in the graph. The intensity of the signals derived from vehicle-administered, not MPTP/P-injected mice, is designated as one. Values shown are the mean 6 S.D. *P,0.05, compared with mice without MPTP/P administration, # P,0.05, compared between vehicle-and IN19-administered mice (n = 4). Scale bar = 20 mm. (II) Protein extracts from the CPu of wild-type mice that were injected or not injected with MPTP/P, in the presence or absence of IN19, were subjected to Western blotting with the indicated antibodies. Relative intensities are shown in the graphs. The intensity of the signals derived from vehicle-administered mice, not injected with MPTP/P, is designated as one. Values shown are the mean 6 S.D. *P,0.05, compared with mice without MPTP/P administration. # P,0.05, compared between vehicle-and IN19-administered mice (n = 4).