NADPH Oxidase Mediates Membrane Androgen Receptor–Induced Neurodegeneration

Abstract

Oxidative stress (OS) is a common characteristic of several neurodegenerative disorders, including Parkinson disease (PD). PD is more prevalent in men than in women, indicating the possible involvement of androgens. Androgens can have either neuroprotective or neurodamaging effects, depending on the presence of OS. Specifically, in an OS environment, androgens via a membrane-associated androgen receptor (mAR) exacerbate OS-induced damage. To investigate the role of androgens on OS signaling and neurodegeneration, the effects of testosterone and androgen receptor activation on the major OS signaling cascades, the reduced form of NAD phosphate (NADPH) oxidase (NOX)1 and NOX2 and the Gα_q/inositol trisphosphate receptor (InsP₃R), were examined. To create an OS environment, an immortalized neuronal cell line was exposed to H₂O₂ prior to cell-permeable/cell-impermeable androgens. Different inhibitors were used to examine the role of G proteins, mAR, InsP₃R, and NOX1/2 on OS generation and cell viability. Both testosterone and DHT/3-O-carboxymethyloxime (DHT)–BSA increased H₂O₂-induced OS and cell death, indicating the involvement of an mAR. Furthermore, classical AR antagonists did not block testosterone’s negative effects in an OS environment. Because there are no known antagonists specific for mARs, an AR protein degrader, ASC-J9, was used to block mAR action. ASC-J9 blocked testosterone’s negative effects. To determine OS-related signaling mediated by mAR, this study examined NOX1, NOX2, Gα_q. NOX1, NOX2, and the Gα_q complex with mAR. Only NOX inhibition blocked testosterone-induced cell loss and OS. No effects of blocking either Gα_q or G protein activation were observed on testosterone’s negative effects. These results indicate that androgen-induced OS is via the mAR–NOX complex and not the mAR–Gα_q complex.

Parkinson disease (PD) is the second most common neurodegenerative disease (1), and the prevalence of this disorder is expected to increase to ∼2 million by the year 2030 (2). The two major hallmark neuropathological characteristics of PD are the presence of Lewy bodies and the loss of dopaminergic neurons in the substantia nigra pars compacta of the brain. The established motor symptoms (resting tremor, rigidity, bradykinesia) manifest when ∼80% of the dopaminergic neurons within the substantia nigra are lost (3, 4). Because no laboratory biomarker or imaging exists for PD diagnosis, diagnosis of PD is a clinical diagnosis that requires the presence of at least two established motor symptoms.

Although the etiology of PD remains elusive (5), oxidative stress (OS) has been linked to PD development and progression (6). OS is an early event in PD pathogenesis and can precede dopaminergic neurodegeneration in the substantia nigra (7–9). The substantia nigra pars compacta is rich in dopaminergic neurons (∼25,000 cells) (10–12), and this composition may underlie its vulnerability to OS insults. Dopamine and its metabolites are highly reactive owing to the auto-oxidation of dopamine that can generate free radicals, leading to cellular vulnerability of the substantia nigra (13, 14). In addition to dopamine auto-oxidation, major cellular OS signaling cascades can contribute to dopaminergic neuronal vulnerability, such as the reduced form of NAD phosphate (NADPH) oxidase (NOX) and calcium (Ca²⁺) neurotoxicity (15, 16). NOX1 and NOX2 isoforms can increase OS generation and play a role in neurodegeneration (17–21), including dopaminergic neuronal loss in PD (22, 23). Indeed, NOX can affect Ca²⁺ release and vice versa (24–33). For example, OS can modulate ryanodine receptors and inositol trisphosphate receptors (InsP₃Rs) that are involved in intracellular Ca²⁺ release from the endoplasmic reticulum via the canonical G protein–coupled receptor (GPCR) Gα_q signaling pathway (30–33). Furthermore, NOX inhibition can block Ca²⁺ signaling (28). Because these signaling pathways can induce a feed-forward loop, dysregulation could lead to substantia nigra vulnerability and, ultimately, PD. Therefore, determining mechanisms that can affect this feed-forward loop is paramount.

Sex differences in PD incidence and prevalence have also been recognized (34–38). Men are 1.5- to 2-fold more likely to develop PD than are women (36, 39, 40). Furthermore, sex differences have been observed in OS signaling. Overall, men have a higher level of circulating OS and higher levels of NOX1 and NOX2 than do women (41–45). In fact, sex hormones (estrogens, androgens) can influence OS generation. It is well established that estrogens are protective against OS insults (46, 47). However, the role of androgens, such as testosterone and DHT, in neurodegeneration is not extensively studied.

Current findings on the effects of androgens on cells are equivocal, wherein androgens were found to be protective or damaging to cells (48–52). Low testosterone is a risk factor for neurodegenerative disorders in men (53–55). However, other studies have found that testosterone can induce neuronal loss (56–60). Our laboratory observed that testosterone can have either neuroprotective or neurodamaging effects, depending on the presence of OS in the cellular environment (58, 60). Specifically, we found that testosterone is an oxidative stressor in both in vivo and in vitro studies (58–64). Depending on the level of OS in the cell, testosterone-induced OS could be neuroprotective via a preconditioning mechanism (58, 65, 66) or damaging by exacerbating existing OS damage (58–60, 64, 67–69). Using cell-impermeable androgens (e.g., testosterone bound to BSA), we found that testosterone can increase cellular OS via a membrane-associated androgen receptor (mAR), specifically an androgen receptor (AR) variant, AR45, that is missing its regulatory N-terminal domain (58, 60, 70). However, the specific OS signaling pathways initiated by mAR are not fully understood. To address this gap in knowledge, this study investigates the mechanisms underlying mAR-induced neurodegeneration in the N27 dopaminergic cell line. Specifically, the role of NOX1/NOX2 and Gα_q/InsP₃R in mediating androgen-induced neurodegeneration in an OS environment was investigated.

Materials and Methods

Reagents

The AR antagonist bicalutamide (no. B9061) was obtained from ApexBio Tech. AR degrader (ASC J9, HY-15194) and NOX2 inhibitor (no. GSK2795039) were purchased from MedChem Express. Apocynin (no. 4663) and diphenyleneiodonium chloride (DPI; no. 0504) were obtained from Tocris Bioscience. NOX1 inhibitor (no. ML171) was purchased from EMD Millipore. InsP₃R inhibitor (2-APB; no. 100065) and Gα_q G protein inhibitor (BIM-46187; no. 533299) were purchased from Calbiochem. GDPβS trilithium salt (no. G7637) was obtained from Sigma-Aldrich/Millipore. Antibodies, including MOX1 (sc-25545 for NOX1) (71), NOX2 (sc-130543) (72), goat anti-rabbit (sc-2004) (73), goat anti-mouse (sc-2005) (74), AR C19 (sc-815) (75), and Gα_q (sc-365906) (76), were from Santa Cruz Biotechnology. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibody (GTX627408) (77) was purchased from GeneTex. Dimethyl sulfoxide (DMSO) was obtained from VWR International, and PBS was from Quality Biological. A Pierce™ bicinchoninic acid protein assay kit was purchased from Thermo Scientific. Protein gels (no. 456-1093) and Mini-PROTEAN TGX (no. 456-9036) were acquired from Bio-Rad Laboratories. Protein A–Sepharose (CL-4B) was obtained from GE Healthcare. RPMI 1640 was purchased from HyClone. Penicillin/streptomycin and TrypLE select (10×) were acquired from Gibco. Fetal bovine serum (FBS) and l-glutamine were obtained from Corning, and charcoal-stripped FBS was from Atlanta Biologicals. Tert-butyl H₂O₂ (A13926) and 3-(4,5-dimethylthiazol-2-yl)-2,5-dimethyltetrazolium bromide (MTT; no. L11939) were purchased from Alfa Aesar. A fluorescent thiol detection kit (no. FLTHIO100) was obtained from Cell Technology. Super Signal West Femto chemiluminescent substrate was purchased from Thermo Scientific. Testosterone (no. A6950-000) and DHT/3-O-carboxymethyloxime/BSA (DHT-BSA, no. A2574-050) were obtained from Steraloids. Testosterone was made from a stock solution in 100% ethanol, whereas DHT-BSA was prepared in media. All inhibitors, except Gα_q protein, were made from a stock solution in DMSO. Final concentrations of DMSO and ethanol were <0.001% in all vehicle controls and treatment groups.

Cell culture

The immortalized 1RB3AN₂₇ (N27) dopaminergic neuronal cell line was harvested from fetal female rat mesencephalic tissue (78). This cell line expresses tyrosine hydroxylase (TH), a marker for dopaminergic neurons, and a membrane-associated splice variant of the AR, AR45, that is missing the regulatory N terminus domain (58, 69, 70, 79–81). N27 cells were cultured as previously published (60). Cells were grown in RPMI 1640 medium supplemented with 10% FBS and 1% penicillin (100 U/mL) and streptomycin (100 μg/mL) at 37°C in 5% CO₂. Cells were seeded at a density of 6 × 10⁴ per well in 96-well cell culture plates (for cell viability assay), 6 × 10⁴ per 100-mm dish (for OS assays), and 1 × 10⁶ per 100-mm dish (for coimmunoprecipitation and Western blot studies). To ensure the quality and integrity of the N27 cell line, we only used passages between 16 and 19 for all experiments. We also characterized these cells based on morphology, doubling time, and a well-characterized response to tert-butyl H₂O₂ and testosterone (58, 60).

Experimental design

Cells were plated into 96-well cell culture plates at a density of 6.0 × 10⁴ cells per well in RPMI 1640 media supplemented with 10% FBS and 1% penicillin/streptomycin and incubated for 24 hours at 37°C, 5% CO₂. At ∼80% confluency, media were changed to RPMI 1640 medium with charcoal-stripped FBS to remove confounding variables (e.g., hormones) from the serum. Maintaining cells in charcoal-stripped FBS does not negatively impact cell viability, as the media still contain salt, glucose, amino acids, and other nutrients (58, 60, 82). Inhibitors (and associated vehicle controls) used in this study were applied 2 hours prior to H₂O₂. Cells were exposed to 10 μM H₂O₂ for 2 hours. This concentration of H₂O₂ is sufficient to achieve 20% cell loss, which is necessary to achieve the OS threshold for androgens to negatively impact cells. After H₂O₂ exposure, androgens were applied to cells for 4 hours. Based on our previous studies (58, 60), androgen concentrations used in this study were 100 nM testosterone and 500 nM DHT-BSA. A fivefold increase in the DHT-BSA concentration (i.e., 500 nM) was used to account for decreased DHT receptor binding due to the structure of DHT-BSA (11 DHT molecules bound to 1 BSA molecule). Furthermore, these androgen concentrations are appropriate to examine the AR splice variant, AR45, which exhibits a similar affinity as the full-length AR to androgens (83).

Cell viability

Cell viability was determined by an MTT assay as previously described (60). Briefly, N27 cells were plated into 96-well cell culture plates at a density of 6.0 × 10⁴ cells/mL in each well. Treatments were carried out as shown in Fig. 1. Cell viability was examined by aspirating the media and incubating the cells in each well with 100 μL of phenol red–free RPMI 1640 medium with charcoal-stripped FBS, followed by 20 μL of 5 mg/mL MTT at 37°C and 5% CO₂ for a 3-hour period. Absorbance was read at 595 nm. Experiments were replicated at least three times on three separate plates; each n is a mean of eight wells per treatment group on one plate. The colorimetric intensity is directly proportional to the number of live cells.

Figure 1.

Experimental design. N27 cells were plated in 96-well cell culture plates and incubated at 37°C in 5% CO₂ for 24 h. Two-hour pretreatment was conducted for all inhibitors and AR degrader. At 80% confluency, cells were exposed to the OS or, H₂O₂, for 4 h followed by testosterone treatment.

Oxidative stress

A fluorescent thiol detection kit was used to measure the levels of reduced thiols, as previously published (60, 69). Cells were plated in 100-mm × 20-mm cell culture dishes at a density of 6.0 × 10⁴ cells per plate. Similar to our cell viability experiments, at 80% confluency, cells were treated with the inhibitor or vehicle for 2 hours prior to H₂O₂ exposure, but they were only exposed to testosterone for 2 hours instead of 4 hours. This method allows OS to be measured prior to cell loss. Afterward, cells were lysed with a mild lysis buffer (10× TrypLE select) on ice, collected into tubes, centrifuged for 1 minute at high speed, and then trypsin was removed. Fluorescent thiol lysis buffer (200 µL) was added to each tube, homogenized, and centrifuged (10,000 rpm) for 5 minutes. Supernatants were collected. Reduced thiols, an inverse measure of OS, were quantified in the cell lysates (1.0 × 10⁵ cells/mL). At least three independent experiments were performed, and fluorescence was measured at 488 nm excitation and 525 nm emission wavelengths.

Cell lysates and homogenization

Following our experimental design (Fig. 1), cells were placed on ice, washed with PBS, and lysed using Nonidet P-40 lysis buffer with a cocktail of dithiothreitol (1 µM), EDTA (1 mM), and phosphatase and protease inhibitors (1:100). The lysates were homogenized and centrifuged at 12,753 rpm at 4°C for 20 minutes. The supernatant was then removed and assayed for protein concentration. Protein concentrations were measured by using the Pierce bicinchoninic acid protein assay kit, according to the manufacturer’s instructions.

Coimmunoprecipitation

For coimmunoprecipitation studies to determine protein–protein interactions of proteins with molecular masses ∼40 to 50 kDa, the protocol was slightly modified to move the IgG bands from 45 kDa to 100 kDa. Cell lysate in Nonidet P-40 lysis buffer was incubated with various primary antibodies overnight at 4°C with agitation. The next day, the cell lysate plus primary antibody mixture was placed on ice, and a bead slurry (protein A–Sepharose beads with PBS) was added. The mixture was incubated at 4°C for 4 hours, centrifuged at maximum speed for 2 minutes, and the supernatant was aspirated. To separate the proteins attached to the primary antibody from the beads, 2× Laemmli sample loading buffer, containing β-mercaptoethanol, was added to the mixture and allowed to incubate at 37°C for 30 minutes in a water bath. After centrifugation at high speed for 2 minutes, protein–protein interactions were determined by Western blotting using respective antibodies specific to the proteins of interest.

Western blot

For all experiments 20 µg of protein was loaded on either Bio-Rad AnykD or 4% to 20% precast gels, except for coimmunoprecipitation that used 37 5µg of protein. Experiments were performed according to our previously published protocols (70). Proteins were separated by SDS-PAGE at room temperature at 25 mA. Next, proteins were transferred overnight at 50 V onto a polyvinylidene difluoride membrane at 4°C. Membranes were quickly washed under agitation with Tris-buffered saline with Tween 20 (TBST) and incubated in 5% nonfat milk in TBST at room temperature to block nonspecific binding. Afterward, the membranes were incubated with specific primary antibodies (1:500 dilution for MOX1, NOX2, and ARC19, and 1:1000 dilution for Gα_q) (71, 72, 75, 76) in TBST with 1% nonfat milk overnight at 4°C. The membranes were washed twice with TBST for 10 minutes on a shaker and incubated for 30 minutes in the corresponding secondary antibody (1:1000 dilution for goat anti-rabbit and goat anti-mouse) (73, 74) in TBST with 1% nonfat milk at room temperature. After washing membranes twice for 10 minutes, the protein bands were detected using a Super Signal West Femto chemiluminescence assay. Protein band intensities were imaged using the Syngene G:BOX system together with FluorChem HD2 AIC software. Protein band densities were measured through densitometry with National Institutes of Health ImageJ densitometer software and normalized to GAPDH (1:10,000 dilution).

Statistical analysis

All analyses were performed using IBM SPSS Statistics version 21 software. Results were expressed as mean ± SEM, and a P value ≤0.05 indicates statistically significant differences. Comparisons were made by two- or three-way ANOVA using inhibitors, oxidative stressor, and hormone as independent factors. Fisher least significant difference post hoc analysis was used to assess differences between the various groups. Each experiment was replicated at least three times with different cell cultures.

Results

mAR interacts with NOX1, NOX2, and Gα_q

Previously, we discovered that the mAR, AR45 splice variant, was localized to lipid rafts within the plasma membrane (70), indicating that the mAR is a component of a communication hub that could facilitate its interaction with membrane-associated proteins to initiate cellular signaling (e.g., OS signaling). To determine whether the mAR (i.e., AR45) complexes with membrane-associated NOX1/NOX2 and Gα_q/InsP₃R OS signaling cascades, we performed coimmunoprecipitation on N27 cell lysate. To immunoprecipate specific proteins of interest, primary antibodies for NOX1 (i.e., MOX1), NOX2, and AR45 (i.e., AR-C19) (71, 72, 75) were used. Next, we probed for protein–protein interactions by using primary antibodies for Gα_q, NOX1, NOX2, and AR45 (71, 72, 75, 76), which showed AR45 complexed with NOX1 and NOX2. Consistent with our prior publication, AR45 complexed with Gα_q (70). Although AR45 coupled with Gα_q, no protein interactions between NOX1 or NOX2 with Gα_q were observed (Fig. 2). Only the 45-kDa AR45 protein was observed, indicating that the coimmunoprecipitation studies were determining protein interactions with mAR.

Figure 2.

Coimmunoprecipitation of AR45, NOX1, and NOX2 to determine the protein complex. NOX1, NOX2, and AR45 proteins were precipitated using specific antibodies and then probed with respective antibodies for NOX1, NOX2, and AR45 to determine protein–protein interactions. NOX1, NOX2, and AR45 interact to form a protein complex. (A) Gα_q couples with AR45 but does not interact with either NOX1 or NOX2. (B) Testosterone can bind to the mAR, AR45, and activate multiple signaling pathways via its protein interactions with NOX1, NOX2, and Gα_q in a lipid raft. Caveolin (purple), flotillin (blue), and phospholipids (orange) are shown. Co-IP, coimmunoprecipitation; WB, Western blot.

Testosterone’s detrimental effects are not mediated through the classical genomic pathway

We previously published that flutamide, a classical AR antagonist, did not block testosterone’s negative effects (58). In this study, we strengthened our findings using a classical AR antagonist, bicalutamide. Bicalutamide binds to the ligand-binding pocket of AR and inhibits AR by failing to induce the correct conformational change (84, 85). Therefore, to confirm that testosterone’s detrimental effects are not mediated through the classical cytosolic AR, bicalutamide AR antagonist was used. As expected, testosterone did not affect cell viability (Fig. 3A), and H₂O₂ significantly decreased cell viability (F_1,16 = 799.2, P < 0.05). In the presence of OS, testosterone further decreased cell viability, as indicated by a significant interaction between OS and hormone (F_1,16 = 165, P < 0.05). The AR antagonist bicalutamide had no effect on cell viability, regardless of the OS environment. Furthermore, bicalutamide did not block testosterone’s negative actions on cell viability in an OS environment.

Figure 3.

Testosterone’s detrimental effects are not mediated through the classical genomic pathway. Testosterone alone does not affect cell viability. H₂O₂ induced ∼20% cell loss, which was exacerbated by testosterone. (A) The AR antagonist bicalutamide did not block testosterone’s negative effects in an OS environment. (B) Testosterone did not alter AR45 expression. The AR degrader, ASC J9, significantly decreased the expression of AR45, irrespective of the presence of testosterone. (C) ASC J9 blocked testosterone-induced cell loss in an OS environment, but it did not influence H₂O₂-induced cell loss. Results are reported as mean ± SEM. Results were determined by ANOVA followed by a Fisher least significant difference post hoc test. *P ≤ 0.05 vs all groups; ^#P ≤ 0.05 vs control; ⁺P ≤ 0.05 vs HT groups. B, bicalutamide; C, vehicle control; H, H₂O₂; HT, posttreatment T; J9, ASC J9; T, 100 nM testosterone.

Because classical AR antagonists did not influence testosterone’s effects, an AR degrader, ASC J9, was used to degrade both cytosolic and mARs. ASC J9 selectively promotes AR degradation by disrupting the interaction between AR and AR coregulators, without impacting AR mRNA expression (86). To ensure that ASC J9 degraded mAR protein (45 kDa) expression, ASC J9 was used with and without testosterone, as previous reports stated that testosterone can stabilize AR (70, 87, 88). Consistent with our prior findings (70), testosterone did not alter mAR protein expression (Fig. 3B). In contrast, ASC J9 significantly reduced mAR expression (F_1,8 = 10.6, P < 0.05), regardless of testosterone exposure (Fig. 3B). To determine whether degrading the mAR impacts testosterone’s negative effects in an OS environment, N27 cells were pretreated with ASC J9 for 2 hours prior to exposure to H₂O₂ and testosterone (Fig. 3C). As expected, significant negative effects from OS exposure on cell viability were observed (F_1,28 = 167.1, P < 0.05). Additionally, significant interactions between oxidative stressor and testosterone treatment (F_1,28 = 8.2, P < 0.05) and between oxidative stressor, hormone, and degrader (F_1,26 = 4.5, P < 0.05) were observed. Specifically, H₂O₂ induced ∼20% cell loss, and testosterone exacerbated H₂O₂-induced cell loss. The AR degrader blocked testosterone’s negative effects on cell viability in an OS environment. However, the AR degrader did not impact H₂O₂-induced cell loss.

The role of G protein and InsP₃R in testosterone-induced neurodegeneration

Because AR45 complexes with Gα_q (Fig. 2A) (70), it is of interest to determine whether Gα_q plays a role in testosterone’s damaging effects in an OS environment. The canonical GPCR Gα_q signaling pathway, which is involved in intracellular Ca²⁺ release from the endoplasmic reticulum via ryanodine receptors and InsP₃Rs, can increase OS (89). To examine this pathway, G protein activity was blocked with GDPβS trilithium, Gα_q with BIM-46187, and InsP₃R was blocked with 2-APB. In Fig. 4A using GDPβS trilithium, a GDP analog, to block G protein activity, significant effects of oxidative stressor (F_1,29 = 206.8, P < 0.05), hormone (F_1,29 = 51, P < 0.05), and an interaction between oxidative stressor and hormone (F_1,29 = 40.6, P < 0.05) were observed. As expected, we observed no effects of testosterone alone and significant effects of H₂O₂ on the cell viability ∼20% cell loss. Testosterone, in the presence of an oxidative stressor, caused a further decrease in cell viability. However, GDPβS trilithium neither blocked H₂O₂’s effects nor the negative effects of testosterone in an OS environment.

Figure 4.

The role of G protein and InsP₃R receptor in testosterone-induced neurodegeneration. (A and B) GDPβS trilithium, a GDP analog, and BIM-46187, a Gα_q inhibitor, did not protect the cells from testosterone’s detrimental effects in an OS environment. (C) The InsP₃R inhibitor 2-APB was able to block testosterone’s damaging effects in an OS environment. Results are reported as mean ± SEM. Results were determined by ANOVA followed by a Fisher least significant difference post hoc test. *P ≤ 0.05 vs all groups; ^#P ≤ 0.05 vs control; ⁺P ≤ 0.05 vs HT groups. C, vehicle control; G, GDPβS trilithium; Gq, BIM-46187; H, H₂O₂; HT, posttreatment T; I, 2-APB; T, 100 nM testosterone.

Next, Gα_q was blocked with BIM-46187. We found significant effects of oxidative stressor (F_1,23 = 303.1, P < 0.05), hormone (F_1,23 = 70.8, P < 0.05), and an interaction between oxidative stressor and hormone (F_1,23 = 65.4, P < 0.05) (Fig. 4B). Again, no effects of testosterone alone were observed. H₂O₂ significantly induced ∼20% cell loss and testosterone worsened H₂O₂’s damaging effects. Similar to the GDP analog, the Gα_q inhibitor did not block testosterone’s detrimental effects in an OS environment. These results indicate that testosterone’s damaging effects via the mAR are not through the Gα_q protein pathway, although AR45 complexes with Gα_q.

Lastly, we looked at the effects of InsP₃R inhibition on testosterone’s negative effects on cell viability. We found significant effects of oxidative stressor (F_1,48 = 314.10, P < 0.05), hormone (F_1,48 = 58.8, P < 0.05), inhibitor (F_1,48 = 35.1, P < 0.05), an interaction between oxidative stressor and hormone (F_1,23 = 65.4, P < 0.05), and an interaction between oxidative stressor, hormone, and InsP₃R inhibitor (F_1,48 = 27, P < 0.05) (Fig. 4C). Consistent with our prior studies, testosterone alone did not influence cell viability, whereas H₂O₂ did. Testosterone exacerbated H₂O₂’s detrimental effects on the cells. InsP₃R inhibition with 2-APB, alone, had no effect on cells, regardless of OS environment. In contrast to earlier observations with GDPβS trilithium and BIM-46187, 2-APB was able to protect the cells from testosterone’s negative effects in an OS environment. This finding of InsP₃R signaling, which mediates intracellular Ca²⁺ release, extends our previous work showing that mAR’s negative effects on OS and cell viability in N27 cells was influenced by Ca²⁺ influx into the mitochondria (58).

The effects of NOX inhibitor on OS generation and cell loss

To determine whether NOX is involved in mAR-induced OS, we used the frequently used NOX inhibitors, apocynin and DPI, at various concentrations based on their IC50 (10 μM and 39 nM, respectively) (90, 91). DPI was toxic to the cells, regardless of the concentration (10 to 80 nM) used. However, apocynin did not impact cell viability in any of the doses (3 to 20 μM) investigated (Fig. 5A). Therefore, apocynin was used as a NOX inhibitor in the subsequent experiments.

Figure 5.

The effects of nonspecific NOX inhibitor on OS generation and cell loss. (A) DPI was toxic to N27 cells, regardless of the concentration used. Apocynin did not show toxicity. (B) Testosterone alone increased OS generation, as evidenced by decreased reduced thiols. Apocynin blocked testosterone’s effects on OS generation. (C) Testosterone alone had no effect on cell viability. Apocynin did not protect cells from H₂O₂’s effects. Apocynin blocked testosterone-induced cell loss in an OS environment. (D) Similarly, apocynin blocked DHT-BSA exacerbation of H₂O₂-induced cell loss. Results are reported as mean ± SEM. Results were determined by ANOVA followed by a Fisher least significant difference post hoc test. *P ≤ 0.05 vs all groups; ^#P ≤ 0.05 vs control; ⁺P ≤ 0.05 vs HT groups. A, apocynin; C, vehicle control; H, H₂O₂; HT, posttreatment testosterone; T, 100 nM testosterone, 500 nm DHT-BSA.

OS was measured by quantifying reduced thiols (an inverse measure of OS). Significant effects of oxidative stressor (F_1,16 = 124.2, P < 0.05), hormone (F_1,16 = 28.7, P < 0.05), inhibitor (F_1,16 = 19.3, P < 0.05), an interaction between oxidative stressor and hormone (F_1,16 = 3.7, P < 0.05), and an interaction between oxidative stressor, hormone, and inhibitor (F_1,16 = 2.3, P < 0.05) were observed (Fig. 5B). Consistent with our prior publications (58–60, 63), testosterone is an oxidative stressor. Both testosterone and H₂O₂ significantly decreased the level of reduced thiols, indicating an increase in OS. Furthermore, testosterone exacerbated H₂O₂-induced OS by further decreasing reduced thiols. The NOX inhibitor apocynin did not alter H₂O₂-induced cell loss, indicating that H₂O₂ increases OS via a non-NOX mechanism. However, apocynin blocked the testosterone-induced OS, indicating that NOX mediates testosterone-induced OS generation.

A similar paradigm was observed for the cell viability assay, except that testosterone alone had no effect on cell viability. Significant effects of oxidative stressor (F_1,23 = 13.4, P < 0.05), an interaction between oxidative stressor and hormone (F_1,23 = 21.4, P < 0.05), and an interaction between oxidative stressor, hormone, and inhibitor (F_1,23 = 10.3, P < 0.05) were observed (Fig. 5C). Specifically, H₂O₂ decreased cell viability, inducing ∼20% cell loss. Testosterone exacerbated H₂O₂-induced cell death, which was blocked by apocynin. Apocynin had no effect on cell viability, regardless of OS exposure. A similar pattern was observed using DHT-BSA. In Fig. 5D, we show significant effects of oxidative stressor (F_1,21 = 513.9, P < 0.05), hormone (F_1,21 = 81.8, P < 0.05), inhibitor (F_1,21 = 57.8, P < 0.05), and an interaction between oxidative stressor and hormone (F_1,21 = 17, P < 0.05). H₂O₂ induced ∼20% cell loss, and DHT-BSA further increased H₂O₂-induced cell loss. Apocynin did not alter H₂O₂’s effects on cell viability. However, apocynin blocked DHT-BSA’s detrimental effects in the presence of OS, implying the involvement of mAR and NOX on androgen’s negative effects in an OS environment.

Because apocynin is a nonspecific NOX inhibitor (92, 93), the role of NOX1 and NOX2 in testosterone-mediated cell loss was examined. Cells were pretreated with a selective NOX1 inhibitor, ML171 (94, 95), followed by H₂O₂ and testosterone. Significant effects of oxidative stressor (F_1,47 = 321.5, P < 0.05), hormone (F_1,47 = 60.9, P < 0.05), inhibitor (F_1,47 = 8.9, P < 0.05), an interaction between oxidative stressor and hormone (F_1,47 = 64.1, P < 0.05), and an interaction between oxidative stressor, hormone, and inhibitor (F_1,47 = 9.3, P < 0.05) were observed. Similar to previous observations, H₂O₂ decreased cell viability, and testosterone further exacerbated H₂O₂-induced cell loss. ML171, however, did not alter cell viability by itself or in the presence of testosterone. Furthermore, ML171 did not protect the cells from H₂O₂. Contrary to observations with apocynin, ML171 partially protected against testosterone-induced cell loss in an OS environment (Fig. 6A).

Figure 6.

The effects of selective NOX inhibitors on testosterone-induced cell loss. (A) The selective NOX1 inhibitor, ML171, had no effect on cell viability. H₂O₂-induced cell loss was not blocked by ML171. ML171 partially blocked testosterone-induced cell loss in the presence of OS. (B) The selective NOX2 inhibitor, GSK2795039, had no effect on cell viability. H₂O₂-induced cell loss was not blocked by GSK2795039. GSK2795039 partially blocked testosterone-induced cell loss in the presence of OS. Results are reported as mean ± SEM. Results were determined by ANOVA followed by a Fisher least significant difference post hoc test. *P ≤ 0.05 vs all groups; ^#P ≤ 0.05 vs control; ⁺P ≤ 0.05 vs HT groups. C, vehicle control; H, H₂O₂; HT, posttreatment T; N, ML 171; N2, GSK2795039; T, 100 nM testosterone, 500 nm DHT-BSA.

The role of NOX2 in testosterone-induced neurodegeneration was further examined using a selective NOX2 inhibitor, GSK2795039. Again, H₂O₂ decreased cell viability, and testosterone further exacerbated H₂O₂-induced cell loss. However, GSK2795039, alone or in the presence of testosterone, did not alter cell viability. GSK2795039 did not protect the cells from H₂O₂, indicating that GSK2795039 at this concentration does not exhibit general antioxidant properties. Similar to observations with ML171, GSK2795039 partially blocked testosterone-induced cell loss in the presence of H₂O₂. Significant effects of oxidative stressor (F_1,28 = 233.5, P < 0.05), hormone (F_1,28 = 34.6, P < 0.05), inhibitor (F_1,28 = 4.3, P < 0.05), an interaction between oxidative stressor and hormone (F_1,28 = 24.5, P < 0.05), and an interaction between oxidative stressor, hormone, and inhibitor (F_1,28 = 4.6, P < 0.05) were observed (Fig. 6B). Based on these findings using ML171 and GSK2795039, cells were exposed to both NOX1 and NOX2 inhibitors with the aim of achieving a full protection from testosterone’s neurotoxic effects in an OS environment. Interestingly, antioxidant effects in the H₂O₂ group were observed, indicating nonspecific antioxidant properties. This general antioxidant effect may be due to the phenolic structure of these inhibitors, which can have antioxidant effects at higher levels (96–98). Hence, we were unable to examine the combined effects of these inhibitors on testosterone-induced cell loss in an OS environment.

The effects of AR degradation on NOX1 and NOX2 protein expression

Because the mAR (i.e., AR45) complexed with NOX1 and NOX2, we wanted to ensure that mAR protein degradation did not alter NOX1 and NOX2 protein expression. Two-hour pretreatment of N27 cells with the AR degrader, ASC J9, did not alter NOX1 or NOX2 expression, irrespective of OS or testosterone exposure (Fig. 7A). Although it appears mAR degradation with ASC J9 affects NOX2 expression, there was no statistical significance across all treatment groups (F_1,24 = 3.95, P = 0.059) (Fig. 7B).

Figure 7.

The effects of AR degradation on NOX1 and NOX2 protein expression. (A and B) Degrading the AR45, using ASC J9, did not affect NOX1 or NOX2 expression. Data are expressed as a normalized ratio of protein band density of (A) NOX1 and (B) NOX2 against GAPDH and presented as mean ± SD. Results were determined by ANOVA followed by a Fisher least significant difference post hoc test. C, vehicle control; H, H₂O₂; HT, posttreatment T; J9, ASC J9; T, 100 nM testosterone.

Discussion

More men than women are affected by PD, regardless of age (35, 40, 69, 99). Sex hormones have been proposed to be involved in this disparity. Interestingly, the AR in a ligand-dependent manner can induce transcriptional activity of the TH gene (100) and increase TH activity (101). Thus, androgens can modulate dopaminergic neurons and dopamine synthesis, and ultimately OS generation via auto-oxidation of dopamine (13, 14). This pathway may underlie androgen neuroprotection via OS preconditioning (58) and the increased incidence in PD for men. Notably, AR is expressed in the substantia nigra pars compacta, the region affected by PD, and not in the substantia nigra pars reticulata (102). ARs are known to mediate important physiological and biological actions, including mood and libido (103–105), sexual reproductive functions (106), and learning and memory (107, 108). Indeed, in men with PD, decreased testosterone has been associated with poor cognitive abilities (109), reduced sexual libido, and erectile dysfunction (110–112). Testosterone replacement therapy in men with PD significantly improved motor and nonmotor symptoms (113, 114). However, in light of reports that estrogen via the estrogen receptor α can also upregulate TH activity (115, 116), androgen-induced transcriptional regulation of TH does not fully explain the increased PD symptoms after disease onset observed in men compared with women (117–119). Therefore, in this study, other OS signaling cascades that can be modulated by androgens, such as NOX1, NOX2, and Gα_q/InsP₃R were examined.

Low levels of androgens (1 nmol/L) via a nongenomic mechanism can exacerbate OS–induced damage in dopaminergic neurons (58, 60), supporting the role of a mAR in testosterone’s negative effects in an OS environment. The function of mAR is poorly understood, but it is proposed to mediate fast, nongenomic actions of androgens (120–122) and is unaffected by classical AR antagonists (58). Furthermore, it is unclear whether the mAR is a full-length AR, truncated AR, or an unknown AR localized to the plasma membrane. Recently, our laboratory discovered the presence of an AR splice variant, AR45, localized to lipid rafts within the substantia nigra pars compacta and the dopaminergic N27 cells (70). AR45 lacks the N-terminal regulatory domain, resulting in a molecular mass of 45 kDa, unlike full-length AR with a molecular mass of 110 kDa (84, 106). Neither age nor testosterone affected AR45 expression (70). In this study, degradation of AR45 using ASC J9, unlike classical AR antagonists, did block mAR exacerbation of OS–induced damage, supporting the role of AR45 as an mAR. This is of interest, as currently there are no known mAR antagonists. To the best of our knowledge, this is the first work to show inhibition of an mAR in the central nervous system.

Based on previous publications showing that AR45 interacts with Gα_q protein within the lipid rafts (70) and that testosterone increases intracellular Ca²⁺ release in N27 cells (58), the role of the Gα_q/InsP₃R/Ca²⁺ pathway on mAR AR45-induced neurodegeneration was determined. Although AR45 complexes with Gα_q protein, Gα_q does not influence mAR-induced neurodegeneration, as evidenced by no effects from either Gα_q protein inhibitor or GDP analog. Interestingly, inhibition of InsP₃R, which mediates intracellular Ca²⁺ release, did block mAR-induced neurodegeneration. Clinical studies found therapeutic benefits of blocking Ca²⁺. Dihydropyridine Ca²⁺ channel blockers have been shown to decrease PD risk, progression, and even mortality (123–128). Furthermore, there is an ongoing exploratory phase 3 clinical trial of isradipine (dihydropyridine Ca²⁺ channel blocker), which may provide better insights into the association between the Ca²⁺ pathway and PD and its therapeutic potential (129).

In addition to the canonical GPCR Gα_q pathway modulation of InsP₃R for intracellular Ca²⁺ release from the endoplasmic reticulum (30–33), NOX can affect InsP₃R-mediated Ca²⁺ release (24–33, 130). There are seven members of the NOX family: NOX1, NOX2, NOX3, NOX4, NOX5, Duox 1, and Duox 2. It is well known that the NOX enzyme is associated with dopaminergic neurons (131, 132) and plays an important role in normal cell physiology, including differentiation, proliferation, and immune response (133–135). Indeed, several studies have reported that NOX is the predominant source of reactive oxygen species in brain cells (16, 22), aside from the mitochondria. Dysregulation of NOX can induce neuronal dysfunction. NOX1, NOX2, and NOX4 are expressed in neurons (136–139). In fact, NOX1 and NOX2 have been shown to be expressed in dopaminergic neurons from PD patients (20, 140). Postmortem substantia nigra tissue from PD patients showed higher NOX2 expression than did individuals without PD (131). In several PD experimental models [e.g., 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, rotenone, and 6-hydroxydopamine (AR) toxins], including the N27 dopaminergic cell line, NOX1 and NOX2 can mediate dopaminergic neuronal death (20, 22, 141–147).

The N27 cell line expresses NOX1, NOX2, NOX4, Duox 1, and Duox 2 homologs of NOX (20), and this study confirms the presence of NOX1 and NOX2 in N27 cells. This study extends these findings by showing that NOX1 and NOX2 complex with AR45. To determine the role of NOX on mAR neurodegeneration, two widely used nonspecific small molecule NOX inhibitors, apocynin and DPI, were used (91, 142, 148, 149), as both have shown neuroprotective benefits in neurodegenerative diseases (150–155). Although prior studies using DPI observed protection from 6-OHDA–induced OS in N27 cells (20), it was not appropriate for our studies, as DPI was toxic to N27 cells. DPI toxicity has been observed in other cells such as N11 glial cells and cultured cells isolated from rat heart, whereby DPI blocked a major antioxidant pathway, resulting in increased OS and apoptosis (156, 157). In contrast, apocynin proved to be nontoxic in this cell line, making it an ideal NOX inhibitor. This is consistent with publications reporting the effectiveness, yet low toxicity, of apocynin in both in vitro and in vivo models (91, 158). Prior studies have shown that apocynin decreased NOX activation, inflammation, and apoptosis in experimental models of PD (132, 159). Indeed, in this study apocynin blocked testosterone’s neurodamaging effects in an OS environment, suggesting that both NOX1 and NOX2 are involved. Apocynin had no influence on H₂O₂-induced cell loss. This effect was not unexpected, as tert-butyl H₂O₂ exerts its effects on OS via two pathways: cytochrome P450 for peroxyl and alkoxyl radical production, and glutathione peroxidase for lipid peroxidation (160). Importantly, note that sublethal concentrations of tert-butyl H₂O₂ that do not affect NOX protein expression were used in this study.

Apocynin’s mechanism of action is controversial. It can act on NOX1 and NOX2 (21), as well as other flavoprotein enzymes (92, 161, 162). Because increased expression of NOX1, and not NOX2, has been shown in response to 6-OHDA and rotenone PD toxins in N27 cells (20, 147), this study examined both NOX1 and NOX2. NOX1 was examined by using a NOX1-specific inhibitor, ML171, which does not affect NOX2 activity (94, 163). ML171, an unsubstituted phenothiazine has shown very high potency in blocking NOX1-dependent reactive oxygen species generation (94). Contrary to observations with apocynin, this selective NOX1 inhibitor only partially inhibited testosterone’s negative actions in an OS environment in N27 cells. These results indicate that another NOX subunit is involved, such as NOX2, that also complexes with AR45. Indeed, selective NOX2 inhibition by GSK2795039 partially blocked testosterone-induced cell loss in an OS environment. Although NOX2 is generally associated with astrocytes and microglia (16, 131, 164, 165), NOX2 is expressed in dopaminergic neurons (20, 140) and can influence OS (144). Unlike apocynin that is orally bioavailable (166–168), in vivo use of NOX2 inhibitors is limited due to the poor oral bioavailability and the need for intravenous administration (163).

Because there has been a lack of success of mitigating mitochondrial-associated OS in clinical trials (169), NOX inhibitors may be potential therapeutics for PD. The NOX1, NOX4, and NOX5 inhibitor, GKT137831, is orally bioavailable and well tolerated in humans. Furthermore, this is the only NOX inhibitor that has progressed into clinical trials with >170 patients (170, 171). Apocynin has also been proposed as a possible therapeutic. Using the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine PD model, apocynin was shown to protect dopaminergic neurons and improve motor function in nonhuman primates (172). Although apocynin has not been tested in patients with neurodegeneration, it decreased OS production in asthmatic patients (173). These preclinical and clinical studies appear promising. Because NOX1 and NOX2 complex with AR45, NOX inhibition may be a potential therapeutic to slow PD progression in men or in postmenopausal women who exhibit an androgenic hormone profile compared with premenopausal women (174, 175).

In conclusion, this study provides evidence that (i) AR45 located in lipid rafts complexes with NOX1, NOX2, and Gα_q; (ii) Gα_q does not complex with NOX1 and NOX2; (iii) androgen’s negative effects are mediated via AR45 and NOX; (iv) Gα_q does not mediate testosterone’s negative effects; and (v) InsP₃R is involved in testosterone-induced neurodegeneration (Fig. 8). The findings of this study help identify key players (e.g., mAR, NOX, and InsP₃R-mediated Ca²⁺) in testosterone-induced neurodegeneration that could serve as potential therapeutic targets for PD.

Figure 8.

Working model. Membrane AR (i.e., AR45) resides in a lipid raft within the plasma membrane and complexes with NOX1, NOX2, and Gα_q. Testosterone can bind and activate AR45, which in turn stimulates NOX1 and NOX2, resulting in OS generation. Alternatively, the AR45–NOX complex can upregulate InsP₃R activity to increase intracellular calcium release, resulting in increased OS. Dysregulation of this pathway can lead to neurotoxicity, such as during PD. The AR45–Gα_q pathway is not involved in testosterone-induced cell loss in OS environments. PLC, Phospholipase C.

Glossary

Abbreviations:

6-OHDA

6-hydroxydopamine

AR

androgen receptor

DHT-BSA

DHT/3-O-carboxymethyloxime/BSA

DMSO

dimethyl sulfoxide

DPI

diphenyleneiodonium chloride

FBS

fetal bovine serum

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

GPCR

G protein–coupled receptor

InsP₃R

inositol trisphosphate receptor

mAR

membrane-associated androgen receptor

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-dimethyltetrazolium bromide

NADPH

reduced form of NAD phosphate

NOX

reduced form of NAD phosphate oxidase

OS

oxidative stress

PD

Parkinson disease

TBST

Tris-buffered saline with Tween 20

TH

tyrosine hydroxylase