TGF-β1 is one of the most potent endogenous immune modulators of inflammation. The molecular mechanism of its anti-inflammatory effect on the activation of the transcription factor NF-κB has been well-studied; however, the potential effects of TGF-β1 on other proinflammatory signaling pathways is less clear. In this study, using the well-established LPS and the 1-methyl-4-phenylpyridinium-mediated models of Parkinson’s disease, we demonstrate that TGF-β1 exerts significant neuroprotection in both models via its anti-inflammatory properties. The neuroprotective effects of TGF-β1 are mainly attributed to its ability to inhibit the production of reactive oxygen species from microglia during their activation or reactivation. Moreover, we demonstrate that TGF-β1 inhibited LPS-induced NADPH oxidase (PHOX) subunit p47phox translocation from the cytosol to the membrane in microglia within 10 min. Mechanistic studies show that TGF-β1 fails to protect dopaminergic neurons in cultures from PHOX knockout mice, and significantly reduced LPS-induced translocation of the PHOX cytosolic subunit p47phox to the cell membrane. In addition, LPS-induced ERK phosphorylation and subsequent Ser345 phosphorylation on p47phox were significantly inhibited by TGF-β1 pretreatment. Taken together, our results show that TGF-β1 exerted potent anti-inflammatory and neuroprotective properties, either through the prevention of the direct activation of microglia by LPS, or indirectly through the inhibition of reactive microgliosis elicited by 1-methyl-4-phenylpyridinium. The molecular mechanisms of TGF-β1-mediated anti-inflammatory properties is through the inhibition of PHOX activity by preventing the ERK-dependent phosphorylation of Ser345 on p47phox in microglia to reduce oxidase activities induced by LPS.
Substantial evidence now demonstrates that microglia-mediated inflammatory processes play an important role in the pathogenesis of several neurodegenerative diseases, including Parkinson’s disease (PD),3 Alzheimer’s disease, multiple sclerosis, and the AIDS dementia complex (1, 2, 3). Microglia are the resident mononuclear phagocyte population within the CNS, and these cells share many phenotypical and functional characteristics with macrophages, suggesting that microglia participate in innate immune responses in the brain. Although activation of microglia serves an important protective function in immune surveillance by removing foreign microorganisms (4), overactivation of microglia followed by overproduction of proinflammatory factors has been shown to result in neuronal death in the brain (5, 6). The midbrain region that encompasses the substantia nigra is particularly rich in microglia (7), and therefore activation of nigral microglia and release of these proinflammatory neurotoxic factors may be crucial components of the degenerative process of dopaminergic (DA) neurons in PD. Because these proinflammatory neurotoxic factors also exhibit immunoregulatory functions necessary for normal immune responses, the microglial response to inflammatory stimuli must be tightly regulated to avoid overactivation and disastrous neurotoxic consequences (2).
TGF-β is a pleiotropic cytokine that plays a critical role in control of cell growth, differentiation, inflammation, cell chemotaxis, apoptosis, and hematopoiesis. Numerous in vitro studies have shown TGF-β can protect neurons from cell death induced by glutamate excitotoxicity (8), chemical hypoxia (9), apoptosis (10), and oxidative injury (10). In vivo studies have shown that TGF-β suppresses the progression of experimental autoimmune encephalomyelitis (11), and that rTGF-β delivered intracerebrally or via virus vectors protects animals against brain injury induced by ischemic (12), excitotoxic (9), and oxidative stress (13). Although TGF-β has been strongly implicated as a neuroprotective factor, it has also been reported that TGF-β can trigger neuronal cell death under certain conditions (14, 15). In addition, the molecular mechanism underlying its neuroprotection has not been clearly elucidated. Although several reports indicated TGF-β has direct protective effects on neurons, other studies reported the neuroprotective effect of TGF-β is mediated through glia cells, such as astrocytes or microglia (16, 17). Further assessment of the functional contribution of TGF-β critically depends on the elucidation of downstream secondary signaling mechanisms, which might offer interesting targets for the development of pharmacological drugs for the treatment of both acute and chronic CNS pathologies.
Increasing evidence has shown that oxidative stress plays a very important role in PD (18, 19), and that anti-inflammatories that inhibit the oxidative stress response can be neuroprotective (20, 21, 22). The primary mediator of the oxidative stress response in microglial cells is the enzyme NADPH oxidase (PHOX), and upon LPS stimulation, microglia cells are known to produce reactive oxygen species (ROS) via PHOX activation (23). PHOX is a multicomponent enzyme consisting of a membrane-associated cytochrome b558 (composed of two subunits: gp91phox and p22phox) and the cytosolic components: p47phox, p67phox, p40phox, and a small GTPase rac2 (24). Previous studies have shown that activation of PHOX activity requires p47phox phosphorylation, a protein that has an important role in translocation of cytosolic components to cytochrome b558, as well as in the assembly and activation of PHOX (25). Previous reports have demonstrated that TGF-β1 can regulate ROS production in hepatocytes and microglia (26, 27, 28, 29). However, the mechanism by which TGF-β1 inhibits ROS production in microglial cells, and its role in regulating the interplay between microglia and neurons in chronic CNS inflammation, remains to be determined.
The main purpose of this study was to elucidate the molecular mechanism underlying TGF-β1-elicited neuroprotection. Using primary rat mesencephalic neuron-glia cultures, we show that TGF-β1 has significant protective effects on both LPS- and 1-methyl-4-phenylpyridinium (MPP+)-induced DA neurotoxicity through its inhibition of microglia activation and reactivation. We found that TGF-β1 is acting to inhibit PHOX activity by inhibiting the translocation of the p47phox subunit of the NADPH oxidase enzyme to the cellular membrane, resulting in the loss of superoxide production by activated microglia and the inhibition of a wide array of proinflammatory mediators produced by activated microglia. The mechanism of action of TGF-β1 is due to the inhibition of ERK phosphorylation, resulting in the loss of p47phox phosphorylation at Ser345. These results offer new insights into our understanding of the action of TGF-β1 on microglia, and on the etiology and eventual therapeutic treatment of neurodegenerative diseases such as PD.
Materials and Methods
PHOX-deficient (gp91phox−/−) and wild-type C57BL/6J (gp91phox+/+) mice were obtained from The Jackson Laboratory. The PHOX−/− mutation is maintained in the C57BL/6 background; thus, C57BL/6 mice were used as control animals. Breeding of the mice was performed to achieve timed pregnancy with the accuracy of ±0.5 days. Timed-pregnant Fisher F344 rats were obtained from Charles River Laboratories. Housing and breeding of the animals were performed in strict accordance with the National Institutes of Health guidelines.
Recombinant human TGF-β1 was obtained from R&D Systems. LPS (Escherichia coli strain O111:B4) was purchased from Calbiochem. Cell culture reagents were obtained from Invitrogen. [3H]DA (30 Ci/mM) was obtained from PerkinElmer Life Sciences, and the mAb against the CR3 complement receptor (OX-42) was purchased from Chemicon International. The polyclonal anti-tyrosine hydroxylase Ab was a gift from Dr. J. Reinhard (GlaxoSmithKline, Research Triangle Park, NC). The Vectastain ABC kit and biotinylated secondary Abs were purchased from Vector Laboratories. The fluorescence probe dichlorodihydrofluorescein diacetate (DCFH-DA) was obtained from Calbiochem. Rabbit anti-p47phox was purchased from Upstate Biotechnology. FITC-conjugated goat anti-rabbit IgG was obtained from Jackson ImmunoResearch Laboratories. Rabbit anti-GAPDH was obtained from Abcam. Mouse anti-gp91phox was purchased from BD Transduction Laboratories. U0126 was purchased from Biomol.
Microglial cell line
The rat microglia HAPI cell line was a gift from Dr. J. R. Connor (Pennsylvania State University, Hershey, PA) (30). Briefly, cells were maintained at 37°C in DMEM supplemented with 10% FBS, 50 U/ml penicillin, and 50 μg/ml streptomycin in a humidified incubator with 5% CO2 and 95% air. The HAPI cell was derived from rat primary microglia-enriched cultures and retains the phenotypic and morphological characteristics of microglia, including phagocytic activity, expression of the receptor for isolectin B4 from Griffonia simplicifolia, as well as the specific microglial markers OX-42 (complement 3 receptor) and glucose transport protein 5. Gene expression of TNF-α and inducible NO synthase can be induced by treatment with LPS.
Primary mesencephalic neuron-glia cultures
Neuron-glia cultures were prepared from the ventral mesencephalic tissues of embryonic day 14–15 rats or day 13–14 mice, as described previously (31, 32). Briefly, dissociated cells were seeded at 1 × 105/well and 5 × 105/well in poly-d-lysine-coated 96- and 24-well plates, respectively. Cells were maintained at 37°C in a humidified atmosphere of 5% CO2 and 95% air, in MEM containing 10% FBS, 10% horse serum, 1 g/L glucose, 2 mM l-glutamine, 1 mM sodium pyruvate, 100 μM nonessential amino acids, 50 U/ml penicillin, and 50 μg/ml streptomycin. Seven-day-old cultures were used for drug treatments. At the time of treatment, immunocytochemical analysis indicated that the rat neuron-glia cultures were made up of 11% microglia, 48% astrocytes, 41% neurons, and 1% tyrosine hydroxylase immunoreactive (TH-IR) neurons. The composition of the neuron-glia cultures of PHOX-deficient mice was very similar to that of the wild-type mice, which consist of 12% microglia, 48% astrocytes, 40% neurons, and 1% TH-IR neurons.
Primary mesencephalic neuron-enriched cultures
Midbrain neuron-enriched cultures were established as described previously (21). Briefly, 24 h after seeding, cytosine β-d-arabinocide was added to a final concentration of 10 μM to suppress glial proliferation. Three days later, the medium was removed and replaced with maintenance medium. Cells were used for drug treatments 7 days after initial seeding. Routinely, the 7-day-old neuron-enriched cultures, which normally contain <0.1% microglia, and <3–5% astrocytes, were used for treatment. Among the neuronal population (Neu-N-immunoreactive neurons), 2.7–3.9% were DA neurons (TH-IR-positive neurons).
Primary midbrain neuron-astroglia cocultures
Rat primary neuron-astroglia cocultures were obtained by suppressing microglial proliferation with 1.5 mM L-leucine methyl ester 24 h after seeding the cells, as described previously (21). Three days later, cultures were changed back to maintenance medium and used for treatment 7 days after initial seeding. The cultures stained with F4/80 Ab showed <0.1% microglia.
Primary microglia-enriched cultures
Rat microglia-enriched cultures, with a purity of >98%, were prepared from whole brains of 1-day-old Fischer 344 rat pups as described previously (33). For superoxide assays, 105 cells were grown overnight in 96-well culture plates before use.
[3H]DA uptake assay
[3H]DA uptake assays were performed as described (33). Briefly, cells were incubated for 20 min at 37°C with 1 μM [3H]DA in Krebs-Ringer buffer (16 mM sodium phosphate, 119 mM NaCl, 4.7 mM KCl, 1.8 mM CaCl2, 1.2 mM MgSO4, 1.3 mM EDTA, and 5.6 mM glucose (pH 7.4)). Cells were washed with ice-cold Krebs-Ringer buffer three times, after which the cells were collected in 1 N NaOH. Radioactivity was determined by liquid scintillation counting. Nonspecific DA uptake observed in the presence of mazindol (10 μM) was subtracted.
Immunostaining was performed as previously described (32). Briefly, formaldehyde (3.7%)-fixed cultures were treated with 1% hydrogen peroxide followed by sequential incubation with blocking solution, after which the cells were incubated overnight at 4°C with Abs against TH (1/20,000). Cells were incubated with biotinylated secondary Ab for 2 h followed by incubation with ABC reagents for 40 min. Color was developed with 3,3′-diaminobenzidine. For morphological analysis, the images were recorded with an inverted microscope (Nikon) connected to a charge-coupled device camera (DAGE-MTI) operated with MetaMorph software (Universal Imaging). For visual counting of TH-IR neurons, nine representative areas per well of the 24-well plate were counted under the microscope at ×100 magnification by three individuals. The average of these scores was reported.
The production of superoxide was determined by measuring the superoxide dismutase (SOD)-inhibitable reduction of the tetrazolium salt WST-1 (34, 35). Neuron-glia or microglia-enriched cultures in 96-well culture plates were washed twice with HBSS without phenol red. Cultures were then incubated at 37°C for 30 min with vehicle control (water) or TGF-β1 in HBSS (50 μl/well). Then, 50 μl of HBSS with and without SOD (50 U/ml, final concentration) was added to each well along with 50 μl of WST-1 (1 mM) in HBSS, and 50 μl of vehicle or LPS (10 ng/ml). To measure superoxide production induced by MPP+, 7-day-old mesencephalic neuron-glia cultures grown in 96-well plates were treated with TGF-β1 in the presence and absence of MPP+, or vehicle alone in 150 μl of phenol red-free treatment medium. Four days after treatment, 50 μl of HBSS with and without SOD (50 U/ml, final concentration) were added to each well along with 50 μl of WST-1 (1 mM) in HBSS. Fifteen minutes later, absorbance at 450 nm was read with a SpectraMax Plus microplate spectrophotometer (Molecular Devices). The difference in absorbance observed in the presence and absence of SOD was considered to be the amount of superoxide produced, and results were expressed as the percentage of vehicle-treated control cultures.
Intracellular ROS assay
Intracellular ROS were determined by using a DCFH-DA assay as described previously with minor modifications (36). DCFH-DA enters cells passively and is deacetylated by esterase to nonfluorescent DCFH. DCFH reacts with ROS to form DCF; the fluorescent product DCFH-DA was dissolved in methanol at 10 mM and was diluted 500-fold in HBSS to give DCFH-DA at 20 μM. The cells were exposed to DCFH-DA for 1 h and then treated with HBSS containing the corresponding concentrations of LPS for 2 h. The fluorescence was read immediately at wavelengths of 485 nm for excitation and 530 nm for emission using a SpectraMax Gemini XS fluorescence microplate reader (Molecular Devices). The experimental value minus the value of the control group was interpreted as the increase in intracellular ROS.
Nitrite and TNF (TNF-α assays)
The production of NO was determined by measuring the accumulated levels of nitrite in the supernatant with Griess reagent and the release of TNF-α was measured with a rat TNF-α ELISA kit from R&D Systems, as described (21).
HAPI cells seeded in dish at 5 × 104 cells/well were treated with LPS for 10 min in the presence and absence of TGF-β1 pretreatment for 1 h. Cells were fixed with 3.7% paraformaldehyde in PBS for 10 min. After washing with PBS, cells were incubated with rabbit polyclonal Ab against p47phox (0.5 μg/ml). Cells were then washed and incubated with FITC-conjugated goat anti-rabbit Ab. Focal planes spaced at 0.4-μm intervals were imaged with a Zeiss 510 laser scanning confocal microscope (63 × PlanApo 1.4 numerical aperture objective) equipped with LSM510 digital imaging software.
Membrane fractionation and Western blot analysis
Membrane fractionation was performed as described (37). HAPI cells were lysed in hypotonic lysis buffer (1 mM EGTA, 1 mM EDTA, 10 mM β-glycerophosphate, 10 mM NaF, 1 mM sodium orthovanadate, 2 mM MgCl2, 10 mM DTT, 1 mM PMSF, and 10 μg/ml each leupeptin, aprotinin, and pepstatin A), incubated on ice for 30 min, and then subjected to Dounce homogenization (20∼25 stokes, tight pestle A). The lysates were loaded onto sucrose in lysis buffer (final 0.5 M) and centrifuged at 1,600 × g for 15 min; the supernatant above the sucrose gradient was centrifuged at 150,000 × g for 30 min. The pellets solubilized in 1% Nonidet P-40 hypotonic lysis buffer were used as membranous fraction. Equal amounts of protein (20 μg/lane) were separated by 4∼12% Bis-Tris Nu-PAGE gel and transferred to polyvinylidene difluoride membranes (Novex). Membranes were blocked with 5% nonfat milk and incubated with rabbit anti-p47phox Ab (1/2000 dilution) or mouse anti-gp91phox (1/2000 dilution) for 1 h at 25°C. HRP-linked anti-rabbit or mouse IgG (1/3000 dilution) for 1 h at 25°C, ECL+Plus reagents (Amersham Biosciences) were used as a detection system. For detection the phosphorylation of MAPK, membranes were blocked with 5% nonfat milk, washed three times followed by incubation with either anti-phospho-ERK1/2 Ab, -ERK1/2 Ab, -phospho-p38 Ab, -p38 Ab, -phospho-JNK Ab, or-JNK Ab at 1/1000 dilution (Cell Signaling Technology) overnight at 4°C. Anti-rabbit HRP-linked secondary Ab (1/2000 dilution) was incubated for 1 h at 25°C. The same detection system as above was used.
The data were presented as the means ± SE. For multiple comparisons of groups, two- or three-way ANOVA was used. Statistical significance between groups was assessed by paired Student’s t test, followed by Bonferroni correction using the JMP program (SAS Institute). A value of p < 0.05 was considered statistically significant.
TGF-β1 is neuroprotective against LPS- and MPP+-induced DA neurotoxicity
We first sought to determine whether LPS-induced inflammation by glial cells leading to destruction of DA neurons could be inhibited by the anti-inflammatory cytokine TGF-β1. Mesencephalic neuron-glia cultures were pretreated with TGF-β1 for 1 h and then stimulated with LPS for 7 days. The degeneration of DA neurons was then determined by [3H]DA uptake assay and numeration of the TH-IR neurons. The [3H]DA uptake assay showed that LPS treatment reduced the capacity of the cultures to take up DA by ∼70% and this LPS-induced inhibition was prevented by a dose-dependent pretreatment with TGF-β1 (Fig. 1,A). Although addition of 0.03 ng/ml TGF-β1 showed slight but insignificant neuroprotection, addition of 0.3, 3, or 30 ng/ml TGF-β1 to cultures shows that the LPS-mediated decrease in DA uptake was significantly restored. Concentrations above 3 ng/ml TGF-β1 did not show any additional increase in neuroprotection (Fig. 1,A). Consistent with a previous report, TGF-β1 alone at 3 ng/ml show a slight but significant neurotrophic effect as measured by DA uptake in these cultures, likely through the activation of astrocytes (17). A similar protective effect was observed with TGF-β1 when counting the number of DA neurons after immunostaining (Fig. 1,B). Thus, LPS-induced loss of DA neurons was prevented by TGF-β1 pretreatment, morphological inspection revealed that LPS treatment not only decreased the number of DA neurons, but also caused a loss of neuronal processes, and these characteristics were also reversed by low-dose TGF-β1 pretreatment (Fig. 1 C).
To investigate whether TGF-β1 mediates its protective function by inhibiting LPS-induced inflammatory response by glial cells, we treated neuron-glia cultures with MPP+, the active metabolite of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, which, unlike LPS, is known to kill DA neurons directly. Our previous work has shown that neuronal death induced by MPP+ is the result of both direct cytotoxic effects on DA neurons by MPP+, as well as by reactive microgliosis induced by toxic factors released by dying neurons (22, 38, 39). Following treatment with 0.2 μM MPP+ for 7 days, DA uptake was reduced to 50% in neuron-glia cultures (Fig. 2,A). These results show that TGF-β1 significantly protects DA neurons from MPP+-mediated toxicity (Fig. 2 A).
To further understand how TGF-β1 protects DA neurons from the cytopathic effects of MPP+, we treated cultures of rat midbrain lacking all glial cells (N, neuron-enriched cultures) or cultures containing neurons and astrocytes but not microglia (NA, neuron-astrocyte enriched cultures) and measured DA neurons uptake 7 days after treatment. DA uptake was reduced to ∼40% in these neuron-enriched cultures (Fig. 2 B). However, TGF-β1 was not capable of protecting DA neurons from MPP+-induced toxicity. These results demonstrate that microglia, not neurons or astroglia, serve as the target of TGF-β1-mediated neuroprotection against MPP+-induced toxicity, and that TGF-β1 is capable of inhibiting this reactive microgliosis.
TGF-β1 inhibits the expression of ROS and other proinflammatory factors by activated microglia
Overactivation of microglia through the direct effects of LPS on microglia or by reactive microgliosis via the death of neurons produces an array of proinflammatory mediators, including ROS, which is the pivotal product mediating inflammation-related neurotoxicity (23, 40). To test the effect of TGF-β1 on the generation of ROS by microglia, primary cells were pretreated with TGF-β1, then exposed to either LPS or MPP+. TGF-β1 significantly reduced LPS-mediated extracellular superoxide production (Fig. 3,A) and intracellular ROS concentrations (Fig. 3,B). Likewise, treatment of cultures with TGF-β1 significantly inhibited the MPP+-induced superoxide production resulting from reactive microgliosis (Fig. 3 C). Taken together, these results demonstrated that TGF-β1 can inhibit ROS free radical production by microglia induced either directly by LPS stimulation or indirectly by MPP+-mediated reactive microgliosis.
We also investigated the effect of TGF-β1 on other inflammatory markers by measuring proinflammatory mediators in microglia-enriched cultures stimulated with LPS. We used LPS to activate microglia because reactive microgliosis induced by MPP+ does not result in measurable NO or proinflammatory cytokine production in neuron-glia mixed cultures. TGF-β1 at as low as 0.03 ng/ml significantly inhibited LPS-induced production of TNF-α and NO, as shown in Fig. 3, D and E. However, no neuroprotection of DA neurons was seen at this dose, suggesting that the major source of neurotoxicity in these cultures was not through TNF-α or NO-mediated toxicity but rather through the production of ROS.
PHOX plays an important role in TGF-β1-mediated protection against LPS-induced neurodegeneration
The results mentioned above indicate that TGF-β1 significantly reduced both LPS and MPP+-induced production of superoxide. To further investigate the role of ROS in TGF-β1-elicited neuroprotection, we used neuron-glia cells from mice deficient in gp91, the catalytic subunit of PHOX, which is the key enzyme required for the production of ROS. Neuron-glia cultures were prepared from PHOX−/− knockout and PHOX+/+ wild-type mice and, as shown in Fig. 4, LPS treatment of neuron-glia cultures prepared from PHOX+/+ mice substantially reduced [3H]DA uptake. Similarly, TGF-β1 significantly attenuated the decrease in [3H]DA uptake (Fig. 4). In contrast, a three-way ANOVA analysis showed significant differences between wild-type and knockout mouse (p = 0.0213). Although LPS treatment also showed a significant albeit smaller reduction in [3H]DA uptake capacity in PHOX−/− mice, paired Student’s t test—followed by Bonferroni correction—results indicated that TGF-β1 failed to show any protective effect on DA neurons from these mice (p = 0.36). These results demonstrate that PHOX plays a key role in TGF-β1-mediated neuroprotection.
TGF-β1 inhibits LPS-induced translocation of the cytosolic subunits p47phox of PHOX to the cellular membrane
It has previously been shown that activation of the PHOX enzyme requires that the cytosolic component p47phox be phosphorylated, and subsequently translocated along with the p67phox component to the plasma membrane, where they associate with cytochrome b558 to assemble into an active enzyme complex (24). Because TGF-β1 showed potent effects on extracellular ROS production, we sought to determine whether TGF-β1 inhibits PHOX activation by preventing the phosphorylation and translocation of the PHOX cytoplasmic subunit p47phox from cytosol to membrane in the microglial cell line HAPI following LPS stimulation. Using confocal light scanning microscopy, we observed that LPS initiated the translocation of cytosolic p47phox, and that within 10 min after LPS stimulation the majority of the p47phox-FITC fluorescence signal was found clustered on the membrane (Fig. 5,A, panel II). The addition of TGF-β1 at 3 ng/ml significantly prevented the translocation of p47phox (Fig. 5,A, panel III). In cells treated with vehicle (Fig. 5,A, panel I) or TGF-β1 alone (Fig. 5,A, panel IV), p47phox were found localized mainly in the cytosol. Consistent with the results of the confocal study, Western blot assay clearly showed the majority of p47phox in the cellular membrane fraction of HAPI cells within 10 min after LPS treatment, while following TGF-β1 pretreatment we see p47phox remaining mostly within the cytosolic fraction. In contrast, the level of gp91 found in the cellular membrane fraction was unaltered by TGF-β1 treatment (Fig. 5 B). Therefore, it is likely that TGF-β1-mediated inhibition of superoxide production by LPS is primarily through the inhibition of p47phox translocation to the cellular membrane.
TGF-β1 suppresses LPS-induced ERK-regulated p47phox phosphorylation
Phosphorylation of p47phox is one of the key intracellular events associated with PHOX activation, and Ser345 phosphorylation of p47phox by the MAPK protein ERK plays a critical role in the potentiation of PHOX activation by proinflammatory agents (41). Therefore, we first examined the levels of LPS-induced p47phox phosphorylation using the anti-phospho-Ser345-p47phox Ab. Enriched microglial cells were incubated with LPS for 10 min, and the levels of phosphorylation of Ser345 in p47phox were analyzed by Western blot assay. In the absence of LPS, weak basal phosphorylation of p47phox was detected, and significant phosphorylation of p47phox was observed 10 min following LPS stimulation. LPS-induced phosphorylation of p47phox was significantly inhibited by TGF-β1 pretreatment (Fig. 6,A). These data suggest that the TGF-β1 inhibits Ser345 phosphorylation on p47phox in microglia to reduce oxidase activities induced by LPS. Furthermore, we determined whether TGF-β1 mediated the inhibition of Ser345 phosphorylation on p47phox through the inhibition of ERK. TGF-β1 pretreatment significantly inhibited LPS-induced ERK1 and ERK2 phosphorylation 5 min after LPS stimulation (Fig. 6 B). These data suggest that TGF-β1 inhibits LPS-induced ROS production in microglia by inhibiting the phosphorylation of ERK1/2 and the subsequent phosphorylation of p47phox at Ser345, which in turn resulted in the reduction in translocation of p47phox to the cellular membrane.
ERK1/2 activation is required for LPS-induced DA neurotoxicity, microglial superoxide production, and p47phox translocation
Because TGF-β1 significantly inhibited LPS-induced ERK phosphorylation, we further tested the role of ERK on LPS-induced DA neurotoxicity, microglial superoxide production, and p47phox translocation using the specific ERK inhibitor U0126. Mesencephalic neuron-glia cultures were pretreated with U0126 for 30 min and then stimulated with LPS for 7 days. The [3H]DA uptake assay showed that LPS treatment reduced the capacity of the cultures to take up DA to ∼50% of the vehicle control and this LPS-induced reduction was prevented by U0126 pretreatment in a dose-dependent manner (Fig. 7 A). At 1 and 5 μM U0126, the LPS-induced decrease in DA uptake was significantly restored, while U0126 alone at 5 μM did not affect DA uptake levels.
Consistent with the neuroprotection data, U0126 at 1 and 5 μM, significantly attenuated the LPS-induced superoxide production in microglia (Fig. 7,B). Furthermore, we stimulated rat HAPI microglia cells with LPS, isolated the membrane fraction, and ran Western blot analysis to measure p47phox translocation. We found that while the majority of p47phox was located in the cellular membrane fraction of stimulated HAPI cells, pretreatment of stimulated HAPI cells with U0126 prevented the translocation of p47phox to the cellular membrane, but rather it remained mostly within the cytosolic fraction. In contrast, the level of gp91 found in the cellular membrane fraction was unaltered by U0126 treatment (Fig. 7,C). Quantitatively analyzing the intensity of the p47phox and gp91phox bands in Western blot, we found a significant inhibitory effect of U0126 on LPS-induced p47phox translocation (Fig. 7 D). These results suggest that ERK phosphorylation is required for LPS-induced p47phox phosphorylation, its translocation to the cellular membrane, and ultimately for cellular ROS production resulting in the toxicity of DA-producing neurons.
Inflammation-induced degeneration of DA neurons in mesencephalic neuron-glia cultures is a useful in vitro model for studying the mechanism and identifying the potential therapeutic application in PD (2). Using the well-characterized models of LPS and MPP+-induced neurodegeneration, we sought to identify the mechanism by which TGF-β1, a major anti-inflammatory cytokine, mediates the neuroprotection of inflammation-induced neurotoxicity in vitro. Our results showed that TGF-β1 exerted potent effects in inhibiting LPS and MPP+-induced inflammation and neuronal destruction through the inhibition of oxidative stress responses in microglial cells. Three salient features of this protective role of TGF-β1 were observed in this study: 1) TGF-β1 exerts potent anti-inflammatory and neuroprotective effects through the inhibition of both direct microglial activation by LPS, and reactive microgliosis elicited by MPP+; 2) TGF-β1 decreases NADPH oxidase-mediated superoxide production mainly through the inhibition of p47phox translocation to the cellular membrane, a novel site of action for the neuroprotective effect of TGF-β1; and 3) the inhibition of p47phox translocation is mediated through the inhibition of PHOX subunit p47phox phosphorylation at Ser345 via suppression of the ERK-signaling pathway.
Our results showed that TGF-β1 has protective effects in both the LPS and the MPP+ model of PD even though the target of these two agents is different. LPS leads to the direct activation of microglia, which results in death of DA neurons through the production of inflammatory mediators. Reports from our laboratory and others have shown that MPP+ can cause reactive microgliosis, and that oxidative stress is involved in MPTP/MPP+-induced neurotoxicity (19, 40, 42). Even though MPP+ directly damages DA neurons, we show that TGF-β1 still provides significant neuroprotective effects through the inhibition of reactive microgliosis in neuron-glia cultures. However, when microglial cells are removed, TGF-β1 can no longer show any protective effects on DA neurons, suggesting that TGF-β1 does not work directly on DA neurons, but rather indirectly by inhibiting activated microglia which contribute to additional neurotoxicity by producing toxic inflammatory mediators. Based on our current evidence, we propose that the anti-inflammatory effect of TGF-β1 is capable of inhibiting both LPS-induced microglial activation and MPP+-induced reactive microgliosis, and its ultimate result is to suppress the inflammation that mediates chronic neurodegeneration in PD.
Our studies indicated that TGF-β1 functions in neuroprotection by inhibiting the initial events in the inflammatory response, the activation of oxidative stress response, and the subsequent production of inflammatory mediators TNF-α and NO. It has been shown that DA neurons in the substantia nigra are uniquely vulnerable to oxidative stress due to lower antioxidant capacity, increased accumulation of iron, high content of dopamine auto-oxidative metabolites, and high density of microglia in the substantia nigra (7, 43, 44). The fact that TGF-β1 significantly inhibits the production of superoxide induced by LPS within a few minutes after stimulation led us to examine this factor in greater details by using PHOX-deficient mice. The findings that TGF-β1 could significantly lessen the LPS-induced DA uptake reduction in cells from wild-type mice, but has no significant protective effect on cells from PHOX−/− mice (Fig. 4) strongly support the contention that the protective effect of TGF-β1 is most likely mediated through the inhibition of PHOX activity.
Although our rat primary midbrain cultures consist of a variety of different types of cells, it is clear that the primary source of LPS-induced ROS in these cultures is the microglial cell. This notion was supported by our previous reports, which indicated that LPS fails to produce extracellular superoxide in neuron-glia cultures devoid of microglial cells, or in cultures of enriched microglia prepared from PHOX−/− mice (23). Thus, these findings indicate that PHOX is the key enzyme involved in superoxide production in these cultures. Activation of PHOX in microglia not only increases the production of superoxide, but indirectly increases the intracellular ROS concentration, possibly through the conversion of superoxide to H2O2, which is membrane permeable. Increase of intracellular ROS can intensify the activation of NF-κB, which leads to higher TNF-α production (2, 23). In addition, it was reported that PHOX inhibitors prevented LPS/IFN-γ-induced degradation of IκBα, and thus, inhibited the activation of NF-κB (45). However, the ability to activate NF-κB-dependent genes such as TNF-α in PHOX−/− cells suggests that PHOX plays an important but not exclusive role in regulating inflammation in microglial cells. These data are consistent with the notion that inhibition of NF-κB activity by TGF-β1 may be mediated at least in part through its inhibition of PHOX, and that PHOX is the major target of the anti-inflammatory activity and neuroprotective effects of TGF-β1.
It is known that translocation of the cytosolic components p47phox, p67phox, p40phox, and rac2 to the plasma membrane is required for the activation of PHOX (25). The phosphorylation of Ser345 of p47phox by proinflammatory agents enhances this translocation event (41). While investigating the mechanism by which TGF-β1 inhibits PHOX activity, we found that TGF-β1 significantly inhibits this LPS-induced p47phox phosphorylation at Ser345, resulting in the inhibition of p47phox translocation. As Ser345 is located in the MAPK consensus sequence, we investigated whether TGF-β1 inhibited components of the MAPK-signaling pathway, and our results indicate that TGF-β1 shows significant inhibitory effect on LPS-induced ERK phosphorylation, but not p38 or JNK (data not shown).Furthermore, a specific ERK inhibitor, U0126, showed strong inhibitory effects against LPS-induced neurodegeneration, superoxide production, and p47phox translocation, suggesting a central role for ERK in these effects (Fig. 7). These findings, coupled with our previous findings on the role of ERK in PHOX activation (22), strongly suggest that it is ERK that regulates p47phox phosphorylation and is the crucial target for TGF-β1-mediated inhibition of PHOX activation. Previously, we have shown that ERK is a crucial mediator of GM-CSF-induced activation of Ser345 on p47phox in neutrophils (41), and it appears that ERK also plays a central in the LPS-induced phosphorylation of Ser345 on p47phox in microglial cells. Taken together, we suggest that TGF-β1 inhibits LPS-induced ROS production in microglia by inhibiting p47phox phosphorylation and translocation, which is regulated by the ERK-signaling pathway.
These results suggest a central role for microglia in the pathogenesis of PD, and that by limiting their proinflammatory response with anti-inflammatory mediator TGF-β1, we can significantly inhibit the neurotoxicity associated with this disease. In addition, we have previously shown that IL-10 also has similarly potent neuroprotective properties (46), and the combination of both TGF-β1 and IL-10 might even act in concert to regulate chronic inflammation in the brain. It has recently been shown that regulatory T cells, a significant source of both IL-10 and TGF-β1, show strong therapeutic efficacy in the treatment of neuroinflammation (47). It is yet to be determined the exact role TGF-β1 plays in the physiological regulation of chronic CNS inflammation in PD, and how TGF-β1 may synergize with other anti-inflammatory mediators to regulate microglia-mediated neurotoxicity. In addition, the level of the immune response in the brain necessary to effectively control infections without resulting in neuropathology is yet to be understood. Consequently, much work remains to determine whether TGF-β1, either therapeutically delivered or produced in the CNS, can play an important anti-inflammatory role by limiting the activation of microglia and promoting a well-regulated immune response that is lacking during the pathophysiology of CNS disorders.
We thank Dr. Salvador Nares (University of North Carolina, Chapel Hill, NC) for helpful suggestions for this report.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health (NIH) Grant DE-13079 from the National Institute for Dental and Craniofacial Research, and was also supported in part by the Intramural Research Program of the NIH/National Institute on Environmental Health Sciences.
Abbreviations used in this paper: PD, Parkinson’s disease; DA, dopaminergic; PHOX, NADPH oxidase; ROS, reactive oxygen species; MPP+, 1-methyl-4-phenylpyridinium; DCFH-DA, dichlorodihydrofluorescein diacetate; TH-IR, tyrosine hydroxylase-immunoreactive; SOD, superoxide dismutase.