Although α-synuclein (α-SYN) aggregation is a hallmark of sporadic and familial Parkinson's disease (PD), it is not known how it contributes to early events of PD pathogenesis such as oxidative and inflammatory stress. Here, we addressed this question in a new animal model based on stereotaxic delivery of an adeno-associated viral vector (rAAV) for expression of human α-SYN in the ventral midbrain of mice lacking the transcription factor Nrf2 (Nrf2−/−). Two months after surgery, Nrf2−/− mice exhibited exacerbated degeneration of nigral dopaminergic neurons and increased dystrophic dendrites, reminiscent of Lewy neurites, which correlated with impaired proteasome gene expression and activity. Dopaminergic neuron loss was associated with an increase in neuroinflammation and gliosis that were intensified in Nrf2−/− mice. In response to exogenously added α-SYN, Nrf2−/− microglia failed to activate the expression of two anti-inflammatory genes, heme oxygenase-1 (HO-1) and nicotinamide adenine dinucleotide phosphate quinone oxidorreductase-1 (NQO1). This impaired Nrf2 response correlated with a shift in the microglial activation profile, towards increased production of proinflammatory markers, IL-6, IL-1β and iNOS and reduced phagocytic capacity of fluorescent beads, and lower messenger RNA levels for TAM receptors Axl and Mer. Postmortem brain tissue samples from patients in early- to middle-stage progression of PD showed increased HO-1 expression in astrocytes and microglia, suggesting an attempt of the diseased brain to compensate these hallmarks of PD through activation of the Nrf2 pathway. This study demonstrates that α-SYN and Nrf2 deficiency cooperate on protein aggregation, neuroinflammation and neuronal death and provides a bifactorial animal model to study early-stage PD.
The causes that lead to Parkinson's disease (PD) remain largely unknown, but a strong correlation exists between the extent of the neurodegeneration and the amount, structure, subcellular location and function of α-synuclein (α-SYN). In idiopathic PD, α-SYN accumulates in Lewy bodies and dystrophic neurites, which are thought to be the underlying pathology leading to neurodegeneration and to progression of the clinical symptoms (1). Three point mutations (A53T, A30P and E46K) in the α-SYN gene as well as duplication and triplication of the wild-type gene have been found to cause familial forms of PD (2,3). Although its exact role in neuron physiology and pathology is ill defined, evidence suggests that overexpression of wild-type or the above-mentioned point mutants of α-SYN might lead to a toxic gain of function related to alterations in axonal transport, oxidative stress and neuroinflammation (4–10).
Unfortunately, most animal models used to study PD, such as 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) and other toxin-based models, do not faithfully reproduce the α-synucleinopathy of PD and therefore they have a limited value to understand the human pathology and to develop disease-modifying therapies (11). As a result, some observations made in these neurotoxin models might be misleading. For example, glial cell-derived neurotrophic factor administered either as recombinant protein or via viral vectors has been found to be highly efficient in protecting nigral dopamine (DA) neurons from neurodegeneration induced by MPTP or 6-hydoxydopamine (12–14), but it has failed to provide protection against α-SYN overexpression (15,16).
In recent years, the proteinopathy of PD has been reproduced partially by overexpressing human α-SYN in the rodent midbrain by means of stereotaxic delivery of an adeno-associated viral vector (rAAV), serotype 6 in our case (17,18). This new animal model reproduces important aspects of human PD such as slow and progressive development of axonal pathology and inflammation and appearance of α-SYN-positive cytoplasmic inclusions (19). The morphology of α-SYN-containing dystrophic axons is remarkably similar to those observed in brains from PD patients (20,21). However, the rAAV6-α-SYN model is characterized by a mild loss of tyrosine hydroxylase (TH)-positive neurons of the Substantia nigra (SN) in the range of 50% in rats (18) and 20% in mice (22), as observed at 2–3 months after injection. Therefore, this slow and progressive model suggests that α-SYN overexpression may require at least one additional “hit” to enhance α-synucleinopathy and induce more robust neuroanatomical alterations. In this study, we hypothesize that such a hit may include impaired proteostasis and prominent or long-lasting inflammation.
We and others have demonstrated that the transcription factor Nrf2 (NF-E2-related factor 2) modulates microglial activation to prevent an excessive anti-inflammatory response in the MPTP mouse paradigm (23–25) and in Drosophila it has been reported that upregulation of the Nrf2 pathway by overexpressing Nrf2 or its DNA-binding dimerization partner, Maf-S, restores the locomotor activity of α-SYN-expressing flies (26). Nrf2 regulates the cell response against oxidative and inflammatory stress and controls the expression of more than 100 genes involved in antioxidant and detoxification reactions including those encoding heme oxigenase-1 (HO-1), NAD(P)H quinone reductase 1 (NQO1) and enzymes related to glutathione metabolism such as glutathione S-transferase, γ-glutamyl cysteine ligase, glutathione peroxidase, glutathione reductase, etc. (27). Moreover, some preliminary observations indicate that adaptation to oxidative stress includes Nrf2-dependent induction of at least some proteasome genes, increasing the cell capacity to degrade oxidized proteins (28,29). Therefore, the concept of detoxification genes regulated by Nrf2 might be expanded to consider the proteasome genes as well.
Several evidences suggest a very relevant role of this transcription factor or its target detoxification genes in the development of PD. Thus, Nrf2 was strongly expressed in the nucleus of the surviving nigral neurons in PD (30), indicating that Nrf2 was induced, but maybe it was insufficient to protect neurons from degeneration. More recently, an association was found between an haplotype in the Nrf2-codifying NFE2L2 gene, including the promoter, and risk to develop PD (31,32). Although not so compelling, Nrf2 activity decreases with age which is the main risk factor for PD (33). Despite this evidence, it remains to be determined whether impairment of Nrf2 cooperates with α-SYN pathology and its implication in the progression of the disease.
In this study, we have analyzed the main hallmarks of human PD in Nrf2-knokcout mice submitted to stereotaxic delivery of rAAV6-α-SYN in the ventral midbrain. We report that α-SYN overexpression as primary insult and Nrf2 deficiency as sensitive background cooperate to worsen protein aggregation, neuronal death and neuroinflammation.
Genetic deletion of the transcription factor Nrf2 aggravates nigral dopaminergic cell death elicited by α-SYN overexpression
First, we developed a parkinsonian mouse model based on combination of α-SYN overexpression and impaired Nrf2 response. This was achieved by stereotaxic delivery of a recombinant adeno-associated vector for expression of human α-SYN (rAAV6-α-SYN) in the ventral midbrain of Nrf2-knockout mice (Nrf2−/−) and wild-type littermates (Nrf2+/+). Two microliters of viral vector (at 4.0 × 1012 GC/ml) were injected into the right ventral midbrain, just dorsal to the location of SN. Nrf2+/+ and Nrf2−/− mice selected for this study expressed human α-SYN in the striatum and ventral midbrain with similar efficiency. However, there was a tendency to see fewer α-SYN-stained neurons in Nrf2−/− mice (Fig. 1A). Double immunofluorescence staining indicated that most cells expressing human α-SYN were also labeled with anti-TH antibodies, therefore confirming the expression of this rAAV6-α-SYN vector in nigral dopaminergic neurons (Fig. 1B). Moreover, human α-SYN was anterogradely transported to the striatum along the nigrostriatal fibers in both mouse groups (Fig. 1C). Similar results were obtained with an rAAV6 vector expressing green fluorescence protein (GFP) (data not shown). These results are in agreement with previous work published using these vectors (34–36) and validate these genotypes as a model to compare the role of Nrf2 in α-synucleinopathy.
We evaluated dopaminergic neuron death at 8 weeks after stereotaxic injection, which corresponds to the time point when neuropathological findings in this model of slow PD progression are intense (15,17,18). In Nrf2+/+ mice, α-SYN induced a small reduction of nigral dopaminergic neurons in the injected right side in comparison with the non-injected left side as determined by immunohistochemical staining of TH+ neurons (Fig. 2A). In contrast, in Nrf2−/− mice we observed exacerbated loss of TH+ neuron bodies (Fig. 2B) and dendritic fibers at the ventral side of SN (Fig. 2C and D). Stereological analyses corroborated these observations, indicating that Nrf2−/− mice suffered 23% greater nigral dopaminergic neuron loss than Nrf2+/+ mice (Fig. 2I).
We found similar TH-staining of striatal efferent fibers at the injected and non-injected sides of both Nrf2+/+ and Nrf2−/− mice, as detected by immunohistochemistry (Fig. 2E–H; densitometric analyses are shown in Supplementary Material, Fig. S1) and immunoblot of striatal protein lysates with the anti-TH antibody (Supplementary Material, Fig. S1). These results were somewhat unexpected because there was anterograde transport of human α-SYN to the striatum (Fig. 1C). In addition, in Nrf2−/− mice striatal DA exhibited only a minor non-significant decrease in the lesioned side compared with the non-lesioned side (as shown in Fig. 2J). 3,4-Dihydroxyphenylacetic acid (DOPAC), a major DA metabolite, and the ratio DOPAC/DA, indicative of DA turnover, did not show a statistically significant difference, although there was a trend of increase in the α-SYN overexpressing Nrf2−/− mice (Fig. 2K and L). Homovanillic acid levels did not show a significant difference either (data not shown). Consistent with these observations, we did not observe a behavioral motor asymmetry between lesioned and non-lesioned groups when they were submitted to the corridor test (37) or to the standard apomorphine test (data not shown). These results suggest that surviving dopaminergic neurons might sprout to maintain efferent fiber density at sufficiently high levels to compensate for the small neuronal loss (38). To determine whether there are plasticity-related changes, we performed immunofluorescence staining for synaptophysin, as a synaptic marker, in the striatum (39). Animals showed increased expression of synaptophysin in the side injected with rAAV6-α-SYN, in comparison with the rAAV6-GFP-injected controls (Supplementary Material, Fig. S2). These results suggest that this model is particularly useful to study early pre-symptomatic stages of parkinsonian neurodegeneration and, at the same time, the fact that there was more sensibility to α-SYN toxicity in the SN of Nrf2−/− mice also suggests the relevance of this transcription factor in protection against α-synucleinopathy.
Nrf2−/− mice exhibit an exacerbated aggregation of α-SYN in dendrites and mild reduction of proteasome subunits
Overexpressed α-SYN accumulated densely in dopaminergic fibers of both genotypes. However, Nrf2−/− mice exhibited many thick dendrites that were heavily loaded with α-SYN (Fig. 3B) compared with Nrf2+/+ mice (Fig. 3A). This pathologic accumulation of α-SYN was similar to that observed in dystrophic Lewy neurites found in PD patients (Fig. 3C). These results are in line with the notion that α-SYN degradation is a limiting step in PD pathology but, more importantly, they highlight the relevance of Nrf2 in proteostasis.
One possible cause for α-SYN aggregation is impairment in the ubiquitin-proteasome system. To determine whether Nrf2 might regulate the expression of proteasome genes, we measured messenger RNA (mRNA) levels of the beta subunit PSMB7 of the catalytic core 20S proteasome and the ATPase PSMC3 and PSMC4 subunits of the 19S regulator proteasome in mouse embryo fibroblasts (MEFs) obtained from Nrf2+/+ and Nrf2−/− mice. MEFs from Nrf2−/− mice exhibited lower basal mRNA levels of subunits PSMB7, PSMC3 and PSMC4 (Supplementary Material, Fig. S4A) although the difference was too small to be detected by the immunoblot (data not shown). In response to the proteasome inhibitor MG132, which stabilizes Nrf2 protein levels and also stimulates expression of proteasome genes (40), we found that Nrf2−/−-derived MEFs consistently showed a lesser increase in the mRNA of those subunits compared with Nrf2+/+-derived MEFs (Supplementary Material, Fig. S4A). Moreover, catalytic chymotrypsin and peptidyl glutamyl peptide-hydrolyzing proteasome activities were lower in extracts from Nrf2−/− MEFs than in control ones (Supplementary Material, Fig. S4B).
On the basis of these observations, we analyzed the mRNA levels of PSMB7, PSMC3 and PSMC4 because it has been reported that the levels of some proteasome subunits are decreased in PD patients (41–43). The three mRNAs exhibited a lower basal abundance in Nrf2−/− compared with Nrf2+/+ mice although the difference did not reach statistical significance (Fig. 3D). To study more specifically the effect of α-SYN on the transduced neurons without the background of other cell types of the brain parenchyma, we analyzed by immunohistochemistry the protein levels of PSMB7 (Fig. 3E). We observed a baseline level of expression in most cells, detected as scattered dots in the cytoplasm and nucleus, but, very importantly, all neurons that overexpressed α-SYN presented more intense PSMB7 staining. Interestingly, in Nrf2−/− mice, the induction of PSMB7 was drastically impaired in comparison to Nrf2+/+ mice. As a control, the expression of PSMB7 in Nrf2−/− and Nrf2+/+ mice injected with rAAV6-GFP was found to be similarly low in both cases (Supplementary Material, Fig. S3). These results suggest that proteasome adaptation to degrade high levels of α-SYN is a limiting step in Nrf2−/− mice, resulting in α-SYN accumulation in dystrophic neurites.
Phosphorylation of α-SYN on Ser129 has been reported in swollen neurites and nuclei (44) and it has been suggested to accumulate as a result of low proteasome-mediated degradation (45). So, we further analyzed the phosphorylation state of α-SYN. Immunofluorescence analysis detected phospho-Ser129-α-SYN only at the side of the SN that had been injected with rAAV6-α-SYN. We did not find positive staining at either the rAAV6-GFP injected or the non-injected sides (data not shown). When we compared the phospho-Ser129-α-SYN staining between Nrf2+/+ and Nrf2−/− mice, we detected significantly enhanced staining in Nrf2−/− neurons (Fig. 4). These results do not explain whether the post-translational modification of α-SYN is preventing its degradation or whether impaired proteasome activity leads to phosphorylation of available α-SYN, but altogether they indicate that α-synucleinopathy is aggravated in Nrf2−/− mice due at least in part to impaired regulation of some proteasome genes.
α-SYN overexpression worsens glial activation at the SN of Nrf2−/− mice
Next, we analyzed the effect of Nrf2 deficiency on the profiles of astroglial and microglial activation in response to α-SYN. Double immunofluorescence with anti-α-SYN and anti-glial fibrillary acidic protein (GFAP) antibodies in 30 µm thick coronal midbrain sections demonstrated astrocyte activation in both mouse genotypes. Nrf2−/− mice exhibited more astrocytes than Nrf2+/+ littermates in the area of α-SYN overexpression at the SN. These astrocytes displayed enlarged bodies and ramifications consistent with activation (Fig. 5A). Quantification of these morphological changes showed higher number of B-type astrocytes in Nrf2−/− than in Nrf2+/+ mice (Fig. 5B). Regarding microglia, double immunofluorescence with anti-α-SYN and anti-Iba1 antibodies of SN sections indicated that Nrf2−/− microglia displayed more enlarged bodies with pseudo-amoeboid shape in the area of α-SYN overexpression in comparison to the Nrf2+/+ microglia (Fig. 5C). To quantify these differences, we differentiated four microglial states according to (46): type A, resting microglia; type B, initiation of microglial activation; type C, activated but non-phagocytic; type D, activated phagocytic. Quantification of the different phenotypes showed that in the Nrf2−/− mice there was an increase in B and C types, indicating exacerbated microglial activation (Fig. 5D). Glial changes were attributed to α-SYN expression because injection of an rAAV6-GFP control vector did not lead to reactive astrogliosis or microgliosis at the same area (Fig. 5A and B). Also, we observed that Nrf2−/− mice presented increased baseline number of microglial cells, as reported previously (24). Taken together, these results indicate that Nrf2 deficiency aggravates the gliosis caused by α-SYN overexpression in the surrounding area of degeneration.
Impaired uptake of α-SYN and increased proinflammatory markers in Nrf2−/−-deficient astrocytes
To determine whether glial changes might be a consequence of α-SYN stimulation rather than just a mere consequence of neuron damage, we analyzed the effect of recombinant α-SYN on glial cultures. Most cultured astrocytes from both Nrf2+/+ and Nrf2−/− mice exhibited a flat polygonal or spindle-like morphology as expected (47) (Fig. 6). Immunocytochemistry analysis demonstrated that neither Nrf2+/+ nor Nrf2−/− astrocytes expressed α-SYN at basal levels. Upon treatment with 1 µm α-SYN for 8 h, both astrocyte types showed α-SYN+ inclusions and some rounded and condensed cells, suggesting that α-SYN was somehow internalized by astrocytes in agreement with Lee et al. (48) and that it is toxic to this cell type. Nevertheless, there was no clear difference between both genotypes. In order to determine whether α-SYN could just adhere non-specifically to astrocytes, we compared the effect of adding monomeric and oligomeric α-SYN to astrocytes. Confocal microscopy showed that monomeric α-SYN was inside the astrocytes (Fig. 6), whereas oligomeric α-SYN was attached to the cell surface (Supplementary Material, Fig. S5). Interestingly, in response to α-SYN, there was no induction of Nrf2-dependent genes, such as those encoding HO-1 or NQO1 (Fig. 7A and B) in either astrocyte type. Nrf2+/+ astrocytes released low levels of proinflammatory cytokines IL-6, IL-1β and iNOS that were similar or slightly higher in Nrf2−/− astrocytes (Fig. 7C–E, respectively). On the other hand, Nrf2−/− astrocytes exhibited ∼50% less baseline expression of IL-4 than Nrf2+/+ astrocytes and were insensitive to α-SYN (Fig. 7F). Taken together, these data indicate that, contrary to the in vivo model where Nrf2−/− mice exhibited enhanced astrogliosis, α-SYN in culture is similarly captured and induces similar astrocyte activation in both genotypes (see Discussion).
α-SYN activates the NF-κB and Nrf2 pathways
To determine whether α-SYN induces microglial activation and the pathways implicated, we first used the microglial cell line BV2. We focused on the crosstalk between p65-NF-κB and Nrf2, which are master regulators of proinflammatory (iNOS, IL-1β, IL-6 and others) and antioxidant (HO-1, NQO1 and others) genes, respectively. α-SYN increased p65 protein levels in a time-dependent manner and reached a maximum after 30 min (Fig. 8A and B). Similarly, α-SYN also increased Nrf2 protein levels after 2 h (Fig. 8A and B). In addition, α-SYN treatment for 4 and 8 h induced the mRNA expression of HO-1, a prototype Nrf2-regulated gene, and IL-1β, IL-6 and iNOS, all targets of NF-κB (Fig. 8C). At the protein level, α-SYN required 8 h to increase HO-1, although iNOS and Iba-1 were already induced by 4 h (Fig. 8D and E). These data indicate that in microglial cells, α-SYN activates the NF-κB proinflammatory pathway in the short-term and the Nrf2 anti-inflammatory pathway in a long-term.
Nrf2 deficiency alters microglial morphology and proinflammatory cytokine production in response to α-SYN
We further investigated the effect of α-SYN in primary microglia from Nrf2+/+ and Nrf2−/− mice. Immunofluorescence analysis demonstrated that Nrf2+/+-derived microglia expressed low but detectable levels of endogenous α-SYN, which corroborates previous findings (10). Nrf2−/− microglia did not present any apparent alteration in endogenous α-SYN expression (Fig. 9). When Nrf2+/+ microglia was exposed to α-SYN (1 µm, 8 h), we observed a progressive transition from the rod-shaped to the amoeboid morphology (Fig. 9). Nrf2−/− microglia presented an amoeboid shape and vacuolated cytoplasm with some spine-like structures and in some cells there was a tendency to become round and condensed. To determine the mechanism of interaction of α-SYN and microglial cells, we analyzed uptake of monomeric and oligomeric α-SYN. Confocal microscopy showed that monomeric α-SYN was inside the microglia (Fig. 9), whereas oligomeric α-SYN was just adhered to the cell surface (Supplementary Material, Fig. S6).
To further define the modulator role of Nrf2 in the microglial inflammatory response to α-SYN, we determined by quantitative real-time polymerase chain reaction (qRT-PCR) mRNA levels of HO-1 and NQO1, which are regulated by Nrf2. Untreated Nrf2−/− microglia expressed lower basal levels of both enzymes than Nrf2+/+ microglia, as expected (Fig. 10A and B). When Nrf2+/+ microglia was treated with α-SYN, there was an increase in mRNA levels of both HO-1 and NQO1 after 8 h but not in Nrf2−/− microglia (Fig. 10A and B). Furthermore, we analyzed mRNA levels of proinflammatory cytokines IL-6 and IL-1β and iNOS. In Nrf2+/+ microglia, these factors reached maximal levels at 4 h and were maintained for at least 8 h, but in the case of Nrf2−/− microglia, these proinflammatory factors continued to rise even after 8 h (Fig. 10C–E), indicating that Nrf2 modulates the proinflammatory response to α-SYN in microglia.
To study the phagocytic response to α-SYN, microglia from Nrf2+/+ and Nrf2−/− mice were incubated for 2 h with fluorescent-labeled polystyrene beads in the absence or presence of 1 µm α-SYN. Phagocytic activity was measured as a function of the amount of intracellular beads (Fig. 10F). At basal levels, Nrf2−/− microglia showed impaired phagocytosis compared with Nrf2+/+ cells. In addition, α-SYN enhanced phagocytosis in Nrf2+/+ and Nrf2−/− microglia, although the Nrf2−/− microglia was less efficient (Fig. 10F). It has been reported that TAM receptor tyrosine kinases, Axl, Mer and Tyro3, are implicated in microglial phagocytosis and inflammatory gene expression (49). Interestingly, at basal levels, Nrf2 deficiency significantly decreased mRNA levels of Axl (Fig. 10G) and Mer (Fig. 10H), whereas the change on Tyro3 was not statistically significant (Fig. 10I). When microglial cells were treated with α-SYN, the Axl receptor showed a diminished induction in Nrf2−/− microglia, compared with Nrf2+/+. These results suggest that Nrf2 deficiency impairs microglial phagocytosis and provide an explanation to the observation reported in Fig. 9, indicating that Nrf2+/+ had greater α-SYN aggregates than Nrf2−/− microglia. Altogether, these data indicate that α-SYN induced the activation of the Nrf2 pathway in microglia and that Nrf2 participates in the acquisition of both pro- and anti-inflammatory profiles, favoring its scavenging over its inflammatory role of microglia.
Evidence for induction of Nrf2 activity at the SN pars compacta in human PD
To determine the relevance of Nrf2-regulated antioxidant gene enzymes in modulation of inflammation in PD, we compared the levels of HO-1 in biopsies from asymptomatic subjects and patients with α-synucleinopathy in a medium stage of PD progression (Braak stages 4 and 5). In PD patients with α-synucleinopathy, demonstrated by formation of Lewy bodies and dystrophic Lewy dendrites (data not shown), we analyzed coexpression of both HO-1 and either GFAP for astrocytes or Iba1 for microglia, compared with control asymptomatic patients. It was not possible to perform double labeling due to different retrieval requirements for each antigen, HO-1, GFAP and Iba-1, but we circumvented this problem by analyzing two adjacent 4 µm thick sections that in many cases showed part of the same cells. As shown in Figure 11B, PD tissue showed increased astro- and microgliosis compared with control tissue (Fig. 11A). At higher magnification, we observed that some activated astrocytes with thick bodies and branches evidenced by GFAP staining were also positive for HO-1 staining in PD tissue (Fig. 11C and D). Similarly, some clusters of Iba1-stained microglia were also positive for HO-1 expression (Fig. 11E and F). We quantified these observations in inmunoblots of human tissues from the SN comparing control versus PD patients. Enhanced HO-1 protein levels were observed in PD patients which correlate with increased expression of GFAP and Iba1 (Fig. 12), indicating that reactive astrogliosis and microglial activation were accompanied by augmented levels of the enzyme HO-1. These results suggest that some glial cells upregulate the Nrf2 response possibly in an attempt to modulate the oxidative stress and inflammation that result at least in part from α-SYN interaction with glia.
The study of α-synucleinopathy and its contribution to PD etiopathology have been hampered by the lack of reliable animal models. The recent use of rAAV vectors for human α-SYN expression has provided a powerful new tool to reassess the contribution of proteinopathy, axonal dysfunction, neuron cell death, oxidative stress and neuroinflammation in PD. Moreover, the slow and progressive development of Parkinsonism in this animal model provides an excellent opportunity to analyze the cooperation of other factors that may participate in disease initiation and progression. In this study, we have analyzed the relevance of impaired Nrf2 transcriptional activity as an adjuvant factor. We have found that Nrf2 deficiency aggravates α-SYN pathology in neuronal bodies and dystrophic neurites and exacerbates gliosis, inflammation and dopaminergic neuron death.
To our knowledge, this is the first study where the proteinopathy associated with PD is analyzed in connection with Nrf2 deficiency. We found that basal mRNA levels of several proteasome subunits such as PSMB7, PSMC3 and PSMC4 were slightly decreased in ventral midbrain of Nrf2−/− mice compared with Nrf2+/+ mice. Moreover, Nrf2−/−-derived MEFs showed impaired expression of proteasome genes and proteasome catalytic activity when they were treated with the proteasome inhibitor MG132, which stabilizes Nrf2 protein levels and also stimulates proteasome gene expression (40). These observations provide a possible interpretation to the reported decrease of proteasome activity and low levels of 20S and 26S subunits in PD brains (50–52). Microarray studies (53,54) indicate that several 20S and 26S proteasome subunits might be transcriptionally upregulated by Nrf2, and a slight increase in the expression of some proteasome genes has been reported under chemical induction or genetic upregulation of Nrf2 (28). In line with those reports, we found that some proteasome genes, including PSMB7, PSMC3 and PSMC4, were regulated by Nrf2 because in Nrf2−/− mice baseline levels of expression were slightly lower. Moreover, α-SYN overexpression led to increased levels of a PSM7 in Nrf2+/+ mice but not in Nrf2−/− mice, further demonstrating a role of Nrf2 in proteasome regulation under the α-SYN challenge.
Gliosis and neuroinflammation are events associated with early PD that can be replicated in the rAAV6-α-SYN model. Here, we report that Nrf2−/− mice presented exacerbated gliosis and inflammation in response to α-SYN. We found that both Nrf2+/+ and Nrf2−/− astrocytes captured α-SYN in a way that resembles astrocyte α-synucleinopathy reported in sporadic PD (55) and diffuse Lewy body disease (56). Astrocytes from both genotypes increased expression of proinflammatory genes encoding for IL-6, IL-1β and iNOS to a similar extent. This essentially similar response is consistent with the lack of Nrf2-dependent induction of HO-1 and NQO1, suggesting that in astrocytes α-SYN does not provoke a gross release of reactive oxygen species (ROS) and does not activate the Nrf2 response. This observation appears to be different from the observed induction of HO-1 in postmortem PD brains (Fig. 10). We speculate that induction of the Nrf2 response in the astrocytes is not a direct consequence of α-SYN signaling but most likely results from interaction with other molecules released to the brain parenchyma.
The effect of α-SYN on microglial cultures was more evident than on astrocytes. Here, α-SYN enhanced microglial phagocytosis in Nrf2+/+ animals, in agreement with Park et al. (57), but phagocytosis in Nrf2−/− microglia was significantly reduced as determined by counting captured fluorescent beads. These results indicated that Nrf2 is also involved in the phagocytic process in concordance with the findings that macrophages from Nrf2−/− mice showed a lower phagocytosis index (58). But more importantly, we describe for the first time that, two of the three TAM receptors, Mer and Axl, were significantly downregulated in Nrf2−/− mice and that Axl expression was impaired in response to α-SYN. The TAM family of receptor protein tyrosine kinases plays a pivotal role in phagocytosis. Particularly Mer has been shown to be involved in clearance of the apoptotic cells by peripheral macrophages and dentritic cells (59). Although different cell types appear to require different combinations of TAM receptors, regarding microglia it has been described that Gas6 (Growth arrest specific gene 6) induces Axl/Mer family to stimulate phagocytosis and repress transcriptional expression of proinflammatory cytokines, including iNOS and IL-1β (49). Therefore, our results uncover an unexplored mechanism by which Nrf2 participates in modulation of microglial activation, disfavoring classical proinflammatory and supporting alternative pro-phagocytic phenotypes.
In Nrf2+/+ but not in Nrf2−/− microglia, α-SYN stimulated the expression of Nrf2-regulated genes such as those coding HO-1 and NQO1. Moreover, in Nrf2−/− microglia, α-SYN exacerbated a proinflammatory profile as demonstrated by increased expression of IL-6, IL-1β and iNOS. These results indicate a role for Nrf2 in down-modulation of microglial proinflammatory phenotype. To get more insights into the mechanistic connection, we used the microglial cell line BV-2. Here, α-SYN induced both the NF-κB proinflammatory and Nrf2 anti-inflammatory responses but with different kinetics. Thus, although the protein levels of p65-NF-κB increased within 30 min, Nrf2 induction was delayed to ∼2–3 h. Regarding gene expression, iNOS protein levels were increased at 4 h, whereas HO-1 protein levels required 8 h. The different kinetics suggest that NF-κB activation is an early and direct event of α-SYN signaling, whereas Nrf2 is a secondary event that participates in a later negative loop of NF-κB regulation. We speculate that at the same time that α-SYN activates the NF-κB signaling pathway, it induces assembling and upregulation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, therefore promoting oxidative stress on the surrounding environment and in the microglia itself. Activation of NADPH oxidase would trigger a second wave of events that would lead to Nrf2 activation. Consistent with our hypothesis, it has been reported that α-SYN induces ROS production in microglia by NADPH oxidase activation which is linked to direct induction of macrophage antigen-1 receptor (7,60).
To determine the relevance of Nrf2-regulated genes in the human PD pathology, we analyzed HO-1 expression in postmortem biopsies of patients in a middle stage of PD progression, as a reporter of Nrf2 induction. Direct analysis of Nrf2 protein was not possible due to the low basal level of expression of this protein and to the poor quality of available mouse Nrf2 antibodies. The specificity of immunodetection with anti-HO-1 antibodies has been previously demonstrated by lack of signal staining in brain of HO-1 knockout mice (61). In control asymptomatic subjects, HO-1 expression was below the level of detection. However, in PD patients we observed a weak staining in dopaminergic neurons and astrocytes in agreement with a previous report (62), but not decorating Lewy bodies. More importantly for this study, we show for the first time that clusters of microglial cells are HO-1+ in PD patients in areas close to dystrophic dopaminergic neurons. The increase in HO-1 staining observed in these cell types suggests the presence of exacerbated oxidative and inflammatory stress. At the same time, the fact that HO-1 induction was not generalized among all cells in SN, either neurons, astrocytes or microglia, suggests that induction of the Nrf2 response was not sufficient to elicit an efficient protection. These results are consistent with reports indicating that Nrf2 activity declines with age (33,63), which is the main risk factor for PD.
Several hallmarks of PD such as mitochondrial generation of oxidative stress, association with environmental toxins, proteinopathy and early inflammation are all targeted by Nrf2 through control of redox homeostasis, biotransformation, proteasome regulation and inhibition of NF-κB, respectively. However, because Nrf2 has a tremendous impact on many aspects of cell and tissue homeostasis, it is anticipated that its effects will not be specific for α-SYN-induced neuron damage, gliosis and inflammation. In this regard, Nrf2 might fall in the category of defensive transcription factors, such as HIF-1 in hypoxia or ATF-4 in unfolded protein response, which participate in cell and tissue damage control. Therefore, the activation of Nrf2 should provide a therapeutic benefit for PD, but also for other diseases that show signs of oxidative stress and low-grade chronic inflammation such as Alzheimer's disease or Huntington's disease. At the same time, because of its relatively unspecific impact on brain protection, targeting Nrf2 might also contribute to amelioration of non-motor symptoms of PD by protecting non-dopaminergic pathways.
MATERIALS AND METHODS
Animals and treatments
Colonies of Nrf2 knockout (Nrf2−/−) mice and wild-type (Nrf2+/+) littermates were established from funders provided by Dr Masayuki Yamamoto (Tohoku University Graduate School of Medicine, Sendai, Japan) (64). All animal protocols were approved by the Ethical Committee for Research of the Universidad Autónoma de Madrid following institutional, Spanish and European guidelines [Boletín Oficial del Estado (BOE) of 18 March 1988; and 86/609/EEC, 2003/65/EC European Council Directives].
Production of the viral vectors and surgical procedures
Transfer plasmids for pseudotyped rAAV6 vector production were prepared by cloning a human synapsin 1 promoter driving GFP or α-SYN into pTR-UF20 plasmid. Woodchuck hepatitis virus post-translational regulatory element and poly A signal were inserted downstream to the gene of interest. The protocol for viral vector production was previously described (15). Both GFP and α-SYN vector batches yielded 4.0 × 1013 genome copy/ml titers. The stock preparations were diluted 10-fold for in vivo injections. Viral vector injections were performed under ketamine/xylazine anesthesia (8 mg/kg ketamine and 1.2 mg/kg xylazine) on adult mice weighing 20–25 g. Surgery was performed using a stereotaxic frame (Stoelting, Wood Dale, IL, USA) and a 5 µl Hamilton syringe fitted with a pulled glass capillary tube (outer diameter of 60–80 µm). Animals received a single 2 µl injection into the right SN at the following coordinates: −2.5 mm posterior and −1.4 mm lateral to bregma, and −4.5 mm ventral relative to dura, calculated according to the mouse atlas of Paxinos and Franklin (65). Vector stocks were injected at a rate of 0.4 µl/min and the needle was left in position for an additional 5 min after the infusion was completed before being slowly retracted. Animals were then sutured and returned to their cage.
High-performance liquid chromatography determination of DA and DA metabolites
After decapitation, the brains were rapidly removed, and the striatum was dissected out. The dissected tissue was rapidly frozen on dry ice and kept at −80°C until further processing. At the time of analysis, tissue samples were homogenized in 0.1m perchloric acid and centrifuged at 10 000 rpm for 10 min before being filtered through a polyvinylidene fluoride filter (0.45 μm; Uni-filter) and spun down for an additional 3 min at 10 000 rpm. Ten microliters of the final filtrate of each sample was injected by a cooled Alexys AS 100 autosampler into Alexys Monoamine Analyzer (Antec, Leyden, the Netherlands) consisting of a DECADE II electrochemical detector and VT-3 electrochemical flow cell. The mobile phase (50 mm phosphoric acid, 50 mm citric acid, 8 mm NaCl, 0.1 mm EDTA, 12.5% methanol, 600 mg/l octane sulfate, pH 3.1) passed through a 1 × 150 mm column with 3 μm particle size (ALF-115) (Antec) at a flow rate of 100 μl/min for determination of DA and DOPAC.
Immunohistochemistry and antibodies used on mouse and human tissues
Immunohistochemistry in mice was performed on 30µm thick coronal brain sections with a standard avidin–biotin immunohistochemical protocol as previously described (24). Primary antibodies were: rabbit anti-TH (1:1000) (Chemicon International, Inc., Temecula, CA, USA); rabbit anti-α-SYN specific for human protein (1:2000) (Millipore Iberica, Madrid, Spain), mouse anti-GFP (1:1000) (Sigma-Aldrich, St Louis, USA). Secondary biotinylated secondary antisera (Vector Labs, Burlingame, CA, USA) was developed using diaminobenzidine (DAB).
Postmortem brain tissues were obtained from two controls (age 58 and 71 years) and four PD patients (ages 62–75 years) within a 12 h postmortem interval, according to standard procedures of Banco de Tejidos de la Fundación CIEN (Madrid, Spain). The control subjects had no background of neuropsychiatric disease and a full neuropathological examination on paraffin-embedded tissue excluded relevant brain pathology. In PD patients, diagnosis was confirmed by hematoxylin and eosin and α-SYN staining on paraffin-embedded tissue sections. All PDs were categorized within Braak stages 4 and 5 for Lewy body pathology progression (1). Immunohistochemistry in human SN was done on 4 µm thick paraffin embedded samples. Primary antibodies were incubated with Dako REAL antibody diluents (Dako Diagnostics, Spain) for 1 h at 22°C; primary antibodies were used as follows: anti-TH (1:200), anti-α-SYN (1:2000), anti-GFAP that recognizes the human protein (1:1000) and anti-HO-1 (1:200) all from Millipore Iberica; anti-Iba-1 (1:500, Wako Chemicals, Neuss, Germany). After washing, the sections were incubated with the secondary biotinylated antiserum (Vector) at 22°C and then with the ABC kit system, and developed using DAB. In the case of HO-1, the secondary antibody was EnVision+ System-HRP (Dako Diagnostics).
The total number of TH+ neurons in the SN was estimated using an unbiased stereological quantification method by employing the optical fractionator principle (66,67). All quantifications were done after blinding the identity of the sections by a coding system and then excluding the samples that had lower than 60% transgene transduction in the SN. For the analysis, every fourth section throughout SN was used and the borders for the region of interest were defined by using a 4× objective. Cell counting was performed as previously described by Decressac et al. (15). The error coefficient attributable to the sampling was calculated according to Gundersen and Jensen (68), and values ≤0.10 were accepted.
Immunofluorescence and antibodies
The protocol was previously described by Rojo et al. (24). Primary antibodies were: rabbit anti-Iba1 (1:100, Wako Chemicals), rabbit anti-GFAP that recognizes the mouse protein (1:500, Dako Diagnostics), mouse anti-α-SYN that recognizes both mouse and human proteins (1:1000, BD Biosciences, Spain), mouse anti-TH (1:250, Millipore Iberica), rabbit anti-synaptophysin (1:250, Synaptic Systems GmbH, Germany), rabbit anti-phospho-Ser129-α-SYN (1:150, Abcam, Cambridge, UK), mouse anti-PSMB7 (1:500, Enzo Life Sciences, USA) and mouse anti-GFP (1:1000, Sigma-Aldrich). Secondary antibodies were: Alexa Fluor 546 goat anti-mouse, Alexa 546 goat anti-rabbit and Alexa Fluor 488 goat anti-mouse at 1:500 dilutions (Life Technologies, Madrid, Spain). Control sections were treated following identical protocols but omitting the primary antibody.
Neonatal (P0–P2) mouse cortex from Nrf2+/+ and Nrf2−/− were mechanically dissociated and the cells were seeded onto 75 cm2 flasks in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS) and penicillin/streptomycin. After 2 weeks in culture, flasks were trypsinized and separated using CD11b MicroBeads for magnetic cell sorting (MACS Miltenyi Biotec, Germany). Microglial and astroglial cultures were at least 99% pure, as judged by immunocytochemical criteria. Medium was changed to DMEM serum-free without antibiotics 16 h before treatment. Then cell cultures were treated with α-SYN (Sigma-Aldrich, Spain), as indicated. For experiments in Supplementary Material, Figure S6, α-SYN was previously incubated in phosphate-buffered saline (PBS) at 37°C for 7 days to obtain oligomers. BV2 microglial cells were cultured in RPMI 1640 medium supplemented with 10% FCS and 80 µg/ml gentamicin. Cells were changed to RPMI serum-free without antibiotics 16 h before addition of α-SYN. Embryonic skin fibroblasts from mice (MEFs) from Nrf2+/+ and Nrf2−/− mice were isolated around embryonic day 13.5 and grown in DMEM with 15% bovine growth serum (Hyclone, Thermo Scientific, USA).
Human and mouse brain extracts
Frozen postmortem brain tissues were obtained from two control (age 58 and 71 years) and three PD patients (age range: 73–85 years) within a 5 h postmortem interval, according to the standardized procedures of Banco de Tejidos de la Fundación CIEN (Madrid, Spain). Human SN and striatal mouse lysates were prepared in radio-immunoprecipitation assay buffer [25 mm Tris–HCl, pH 7.6, 150 mm NaCl, 1 mm ethylene glycol tetraacetic acid, 1% Igepal, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 1 mm α-toluenesulfonyl fluoride (PMSF), 1 mm Na3VO4, 1 mm NaF, 1μg/ml aprotinin, 1 μg/ml leupeptin and 1 μg/ml pepstatin]. Fifty micrograms of protein were loaded for SDS–polyacrylamide gel electrophoresis (PAGE) electrophoresis. The primary antibodies used were anti-p65-NF-κB (1:2000, Calbiochem, Merck Chemicals International, USA), anti-Nrf2 (1:2000, generous gift of Dr John Hayes, Biomedical Research Institute, Ninewells Hospital and Medical School, University of Dundee, Scotland, UK), anti-HO-1 (1:1000, Millipore Iberica), anti-iNOS and anti-Iba1 (1:1000, Abcam, Cambridge, UK), anti-GFAP (1:10 000, Sigma-Aldrich) and anti-β-actin (1:2000, Santa Cruz Biotechnology, Santa Cruz, CA, USA). Cell lysates were resolved in SDS–PAGE and transferred to Immobilon-P membranes (Millipore Iberica). These membranes were analyzed using the primary antibodies indicated above and appropriate peroxidase-conjugated secondary antibodies. Proteins were detected by enhanced chemiluminescence (Amersham).
Proteasome activity assays
Proteasome activity was assessed in MEF extracts using fluorogenic peptides. Briefly, cells were sonicated in ice-cold buffer [50 mm 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.54, 150 mm NaCl, 5 mm EDTA, 0.1 mm leupeptin, 1 μg/ml pepstatin A and 1 mm PMSF] and extracts prepared by centrifugation for 15 min at 3000g. Proteasome-specific peptidase activities were assayed by monitoring the production of 7-amino-4-methylcoumarin (AMC) from the following fluorogenic peptides (Enzo Life Science): 100 μm Suc-LLVY-AMC (chymotrypsin-like), 100 μm Boc-LLR-AMC (trypsin-like) and 100μm Z-LLE-AMC (PGPH or caspase-like activity) in 20 mm HEPES, pH 7.4. Reactions were started by adding an aliquot of cellular extract and the fluorescence of released AMC (excitation, 380 nm; emission, 460 nm) was monitored continuously at 37°C in a standard microtiter plate fluorometer (Synergy™ HT, Bio-Tek Instruments GmbH, Friedrichshall, Germany). Background activity (caused by non-proteasomal degradation) was determined by the addition of the proteasome inhibitor MG132 at a final concentration of 100μm. Assays were calibrated using standard solutions of free fluorophores. Substrate consumption at the end of incubation never exceeded 1%.
Astrocytes and microglia were seeded in 24-well plates. After 24 h, cells were shifted to serum-free medium without antibiotics. After treatment with 1 µm α-SYN for 8 h, cells were washed with cold PBS and fixed with 4% paraformaldehyde for 10 min. Cells were permeabilized with 0.25% Nonidet P-40 (Sigma-Aldrich) for 10 min and incubated with primary antibodies for 1.5 h at 37°C in a humidified box. Secondary antibodies were incubated for 45 min at 37°C in the same conditions. Cells were counterstained with 4′,6′-diamino-2-phenylindole (DAPI) (Molecular Probes, Leiden, the Netherlands). Fluorescent images were captured using appropriate filters in a Leica DMIRE2TCS SP2 confocal microscope (Leica, Nussloch, Germany). Primary antibodies were: rabbit anti-GFAP (1:500), rabbit anti-Iba1 (1:100) and anti-α-SYN (1:1000). Secondary antibodies were: Alexa Fluor 546 goat anti-mouse and Alexa 546 goat anti-rabbit.
Microglial cells from Nrf2+/+ and Nrf2−/− mice were collected as described above and 150 000 cells were plated on coverslips for 16 h. Then the medium was replaced with serum-free DMEM without antibiotics for 24 h before adding fluorescent microspheres (150 microspheres per cell) (FluoSpheres Polystyrene Microspheres, Invitrogen, Madrid, Spain) and incubation for 2 h. Then the cells were washed with PBS, fixed with 4% paraformaldehyde and stained with DAPI. The images were captured using 90i Nikon microscope (Nikon, Montreal, Que., Canada) at 40×.
Analysis of mRNA levels by quantitative real-time PCR
Total RNA from microglial and astroglial primary cultures and from BV2 cells was extracted using TRIzol reagent according to the manufacturer’s instructions (Invitrogen). One microgram of RNA from each experimental condition was treated with DNase (Invitrogen) and reverse-transcribed using 4 µl of high-capacity RNA-to-cDNA Master Mix (Applied Biosystems, Foster City, CA, USA). For real-time PCR analysis, we performed the method previously described by Rojo et al. (24). Primer sequences are shown in Table 1. To ensure that equal amounts of cDNA were added to the PCR, the β-actin housekeeping gene was amplified. Data analysis is based on the ΔΔCt method with normalization of the raw data to housekeeping genes as described in the manufacturer’s manual (Applied Biosystems, Life Technologies, Spain). All PCRs were performed in triplicates.
|Gene product||Forward primer||Reverse primer|
|Gene product||Forward primer||Reverse primer|
Data are presented as mean ± SEM. One- and two-way analyses of variance with post hoc Newman–Keuls test and Bonferroni's test were used, as appropriate. A Student's t-test was used to assess differences between groups.
This work was supported by grants SAF2010-172218 [to A.C.] from the Spanish MICINN, “Bolsa de Investigación L'Oreal-UNESCO 2010” [to I.L.B.], Swedish Research Council (K2009-61-20945-03-1), European Community's Seventh Framework Programme FP7-HEALTH-2009 under grant agreement no. 241791 (MEFOPA) and from the European Research Council under grant agreement no. ERC-2009-StG 242932 (TreatPD) [to D.K.]. I.L.B. is recipient of a Ramón y Cajal contract (MICINN-RYC) and A.U. is a recipient of European Union Marie Curie Actions Research Training Network Program in Nervous System Repair (MRTN-CT-2003-504636).
We thank Rosa Ana Ramírez, Johanna Troya Balseca, Ana Belén Rebolledo, Hongyan Liu and Ulla Samuelsson for their technical support.
Conflict of interest statement. None declared.
- parkinson disease
- nitric oxide synthase
- rna, messenger
- surgical procedures, operative
- transcription factor
- multicatalytic endopeptidase complex
- dopaminergic neuron
- symptom aggravating factors
- protein aggregation, pathological