Androgens selectively protect against apoptosis in hippocampal neurones

Abstract

Androgens can protect neurones from injury, but androgen neuroprotection is not well characterised in terms of either specificity or mechanism. Here, we compared the ability of androgens to protect neurones against a panel of insults, empirically determined to induce cell death by apoptotic or non-apoptotic mechanisms. Three criteria defining, but not inclusive of apoptosis are: protection by caspase inhibition, protection by protein synthesis inhibition, and presence of pyknotic nuclei. According to these criteria, β-amyloid, staurosporine, and Apoptosis Activator II induced cell death involving apoptosis, while hydrogen peroxide (H₂O₂), iron, calcium ionophore, and 3-nitropropionic acid induced cell death featuring non-apoptotic characteristics. Pretreatment of hippocampal neurones with testosterone or dihydrotestosterone attenuated cell death induced by β-amyloid, staurosporine, and Apoptosis Activator II, but none of the other insults. The anti-oxidant Trolox did not reduce cell death induced by β-amyloid, staurosporine, and Apoptosis Activator II, but did protect against H₂O₂ and iron. Similarly, a supra-physiological concentration of oestrogen reduced cell death induced by H₂O₂ and iron, an effect not observed with androgens. We also show that activation of oestrogen pathways was not necessary for androgen neuroprotection. These data suggest that androgens directly activate a neuroprotective mechanism specific to inhibition of cell death involving apoptosis. Determining the specificity of androgen neuroprotection may enable the development of androgen compounds for the treatment of neurodegenerative disorders.

INTRODUCTION

Men experience a significant decrease in levels of testosterone in blood (1) and brain (2) due to normal aging. Age-related androgen loss in men adversely affects muscle and bone mass, sexual arousal, sperm production, and brain functions such as mood, memory, and cognition (1). Recent data also suggest that low levels of testosterone in aging men may be one of several risk factors for the development of Alzheimer’s disease (AD) (2,3). Androgens have many beneficial actions in the CNS, which the loss of may contribute to age-related neurological deficits and AD. For example, testosterone decreases levels of β-amyloid (Aβ), a protein that is a key contributor to AD pathogenesis (4,5). In addition, androgens are positive regulators of neuronal plasticity in the spinal nucleus of the bulbocavernosus (6), excitability in the CA1 region of hippocampus (7), and spine density in hippocampus (8). Androgens also prevent retraction (9) or increase the length (10) and size (11) of neurites from motor neurones. Other neurotrophic effects of testosterone include cell differentiation (12), neurogenesis (13–15), and development of neurones in the hippocampus (16) and motor (17–19) and autonomic (20) systems.

One important cellular action of androgens is regulation of neurone viability. During development, testosterone and its oestrogen (17β-oestradiol) and androgen (dihydrotestosterone) metabolites determine neurone number in specific sexually dimorphic nuclei via regulation of apoptosis (21, 22). Further, androgens regulate survival of central and peripheral motoneurones following injury (23). For example, androgens increase the survival of motor neurones in newborn rats after cranial nerve crush (24). In addition, androgens increase the speed of regeneration of injured axons of motor neurones in young and adult rats (25, 26). Androgens are also endogenous regulators of viability in neurones challenged with toxic insults in adult animals. Adult male rats and mice depleted of endogenous androgens by orchidectomy exhibit increased vulnerability to hippocampal neurone loss induced by excitotoxins (27, 28), an effect that can be reversed by treatment with DHT (28). In primary neurone culture paradigms, testosterone and related androgens attenuate cell death induced by serum deprivation (29), Aβ (30–32), and hydrogen peroxide (H₂O₂) (33). However, androgens can fail to protect neurons and even exacerbate injury in response to some forms of injury such as ischemia (34,35), mitochondrial toxin 3-nitropropionic acid (3-NP) (36), and muscimol-induced excitotoxicity (37).

Why androgens are neuroprotective against some insults, but not others is unclear. Our previous findings show that androgen neuroprotection involves a mitogen-activated protein kinase/extracellular signal-regulated kinase(MAPK/ERK) signalling pathway that functionally inactivates the pro-apoptosis factor Bcl-2-associated death protein (Bad) (32). These data suggest that androgens regulate vulnerability to apoptosis -related mechanisms and thus, androgens may selectively protect neurones against insults that induce cell death involving apoptosis pathways. To investigate this hypothesis, we evaluated the effects of androgens on the viability of neurones challenged with a panel of insults, some of which were empirically determined to induce an apoptotic type of cell death. Specifically, we used seven different neurotoxins: i) Aβ, an aggregating peptide involved in the initiation and promotion of neurodegeneration in AD (38); ii) staurosporine, a general protein kinase inhibitor commonly used to induce apoptosis (39); iii) Apoptosis Activator II, a cell permeable compound that promotes apoptosis by activating caspases in a cytochrome c- and Apaf-1-dependent manner (32,40); iv) the pro-oxidant hydrogen peroxide (H₂O₂) (41, 42); v) iron (ferrous/ferric chloride, FeCl_2/3), an oxidant that acts via Fenton chemistry (41, 43); vi) calcium ionophore A23187 (44); and vii) 3-nitropropionic acid (3-NP), an irreversible inhibitor of succinate dehydrogenase/mitochondrial complex II (45).

MATERIALS AND METHODS

Materials

Testosterone, dihydrotestosterone (DHT), 17β-oestradiol (E2), and 5α-androstane-3β, 17β-diol (3β-diol) were acquired from Steraloids (Newport, RI). zVAD-fmk, cycloheximide, Trolox, staurosporine, hydrogen peroxide (H₂O₂), ferrous/ferric chloride (FeCl_2/3), A23187, and 3-nitropropionic acid (3-NP) were purchased from Sigma-Aldrich (St. Louis, MO). β-Amyloid (Aβ) peptide 25–35 was obtained from Bachem (Torrance, CA) and Apoptosis Activator II was acquired from Calbiochem (San Diego, CA).

Neurone cultures

Neurones (~ 95% neuronal, ~5% astroglial) were cultured from embryonic day 18 Sprague-Dawley rat pups (n ≥ 6 pups per preparation) using a standard protocol, as earlier described (32, 46). In brief, hippocampi were dissected, and then dissociated both enzymatically (0.125% trypsin-EDTA, 37°C, 10 min) and mechanically. The cell suspension was filtered through a 40 μm strainer (Falcon, Franklin Lakes, NJ), and then diluted in serum-free Dulbecco’s modified Eagle medium (DMEM) with 20 mM HEPES, 100μg/ml transferrin, 5μg/ml insulin, 100 μM putrescine, and 30 nM selenium added. Cells were plated at a density of 3.75 × 10⁴ cells/cm² in 48-well plates (NUNC, Naperville, IL) previously treated with poly-L-lysine (0.05 mg/ml) and used for experiments beginning 3 d after plating. Cultures were kept at 37°C in a humidified incubator with 95% room air/5% CO₂.

Culture treatments

Cultures were treated with the steroid hormones testosterone (10 nM or 10 μM), DHT (10 nM or 10 μM), or E2 (10 μM) or vehicle (0.1% ethanol) for 2 h before and during 24 h exposure to toxin. Some cultures were similarly treated with the following inhibitors, delivered at previously established effective concentrations, or vehicle (0.1% dimethyl sulfoxide, DMSO): zVAD-fmk (50 μM) (47, 48), cycloheximide (10 μg/ml) (49), Trolox 250 (μM) (50). The toxins included aggregated Aβ25–35 (0–50 μM), prepared as previously described (32,46), staurosporine (0–0.5 μM), Apoptosis Activator II (0–7 μM), H₂O₂ (0–25 μM), FeCl_2/3 (0–2.5μM), A23187 (0–350 nM), and 3-NP (0–2.5 mM). Steroids, toxins, and inhibitors were solubilised in 100% ethanol or DMSO, and diluted in culture medium to a final ethanol or DMSO concentration of ≤ 0.1%; vehicle controls consisted of ethanol and/or DMSO at 0.1%.

Cell viability

Neuronal viability was determined by standard cell-counting procedures previously described (32, 46, 51). All viable cells within the defined field of a microscope reticle grid (final magnification 300X) were counted using a manual mechanical counter by an experimenter blinded to condition. Cells were scored viable on the basis of both positive staining with the vital dye calcein acetoxymethyl ester (Molecular Probes, Eugene, OR) and the morphological criterion of a smooth, spherical soma. Counts of viable cells were made in four non-overlapping fields per culture well (in a predetermined pattern maintained across all experiments) with each condition represented by 3 separate wells. The number of viable cells counted per well for vehicle-treated control conditions ranged from 100–200. All experiments were repeated in at least 3 independent culture preparations. Raw cell count data were statistically analyzed with one-way ANOVA, followed by between group comparisons using the Fisher LSD test (significance indicated by P < 0.05). Cell viability is presented graphically as a percentage of live cells in the vehicle-treated control condition.

Assessment of nuclear morphology

Nuclear morphology was assessed with the membrane permeable, nucleic acid stain SYTO 11 (Molecular Probes). Briefly, cultures were treated with 1 μM SYTO 11, incubated for 30 min at 37°C in the dark, and nuclear morphology examined using fluorescence microscopy. Neurones were observed for pyknotic nuclei, features of cells undergoing apoptosis (52,53). Degenerating neurones that had condensed nuclei and did not display these features of apoptosis (53) were classified as non-apoptotic. Cultures were evaluated for nuclear morphology by a researcher blinded to experimental condition.

RESULTS

Aβ, staurosporine, and Apoptosis Activator II induce neuronal apoptosis

To begin investigating the hypothesis that androgens selectively protect against apoptosis, we determined whether apoptosis contributes to cell death induced by seven different toxins: Aβ, staurosporine, Apoptosis Activator II, H₂O₂, FeCl_2/3, A23187, and 3-NP. First, we evaluated the effect of two pharmacological inhibitors of apoptosis on the extent of cell death. Specifically, we tested the ability of the general caspase inhibitor zVAD-fmk and the translation inhibitor cycloheximide to attenuate cell death because apoptosis generally involves both activation of caspases and induction of protein synthesis (54,55). Exposure of cultures to 50 μM zVAD-fmk for 2 h before and during insult exposure significantly attenuated cell death due to Aβ, staurosporine, and Apoptosis Activator II (Fig. 1A). In contrast, zVAD-fmk treatment failed to inhibit cell death induced by H₂O₂, FeCl_2/3, A23187, and 3-NP (Fig. 1B). Treatment of cultures with 10μg/ml cyclohexamide also significantly protected against Aβ, staurosporine, and Apoptosis Activator II toxicity (Fig. 1C), but not against H₂O₂, FeCl_2/3, A23187, and 3-NP (Fig. 1D).

Fig. 1.

Pharmacologic inhibition of caspases and protein synthesis attenuates cell death induced by Aβ, staurosporine, and Apoptosis Activator II. Neuroneal cultures were treated with 50 μM zVAD-fmk, 10 μg/ml cycloheximide (CHX), or vehicle for 2 h, then exposed to 50 μM Aβ, 0.4 μM staurosporine (STS), 3 μM Apoptosis Activator II (AAII), 25 μM H₂O₂, 2.5 μM FeCl_2/3, 200 nM A23187, 2.5 mM 3-NP, or vehicle for 24 h, and processed for cell viability. Caspase inhibition with zVAD-fmk reduces cell death induced by (a) Aβ, staurosporine, and Apoptosis Activator II, but not (b) H₂O₂, FeCl_2/3, A23187, and 3-NP. Similarly, cycloheximide attenuates neurotoxicity caused by (c) Aβ, staurosporine, and Apoptosis Activator II, but not (d) H₂O₂, FeCl_2/3, A23187, and 3-NP. Data show mean cell viability (± SEM) pooled from 3 independent experiments (n = 3). * P < 0.05 relative to the vehicle-treated control condition.

To further investigate the contribution of apoptosis in cell death induced by the different toxins, we qualitatively assessed cultures for the presence of pyknotic nuclei, a morphological characteristic of apoptosis (52). Using the nucleic acid stain SYTO 11 to examine nuclear morphology, we found that vehicle-treated neurones exhibited round, smooth, and large nuclei with uniform staining (Fig. 2A). As cells undergo apoptosis, the chromatin condenses and the nuclei break down into small pyknotic spheres (52). After 24 h exposure to 50 μM Aβ (Fig. 2B), 0.4 μM staurosporine (Fig. 2C), or 3 μM Apoptosis Activator II (Fig. 2D), cultures showed numerous neurones with pyknotic nuclei suggestive of cell death involving apoptosis. In contrast, 24 h treatment with 25 μM H₂O₂ (Fig. 2E), 2.5 μM FeCl_2/3 (Fig. 2F), 200 nM calcium ionophore A23187 (Fig. 2G), or 2.5 mM 3-NP (Fig. 2H) resulted in predominately condensed nuclei and essentially no pyknotic nuclei.

Fig. 2.

Cell death induced by Aβ, staurosporine, and Apoptosis Activator II is characterised by pyknotic nuclei. Cultures were treated with vehicle, 50 μM Aβ, 0.4 μM staurosporine (STS), 3 μM Apoptosis Activator II (AAII), 25 μM H₂O₂, 2.5 μM FeCl_2/3, 200 nM A23187, or 2.5 mM 3-NP for 24 h. Representative images show nuclear changes visualised with membrane-permeable nucleic acid stain SYTO 11. (a) Vehicle-treated control neurones show normal morphology (arrowheads). (b) Aβ, (c) staurosporine, and (d) Apoptosis Activator II treatment induce pyknotic nuclei in neurones (arrows), while (e) H₂O₂, (f) FeCl_2/3, (g) A23187, and (h) 3-NP induce condensed nuclei (asterisks).

Androgens are neuroprotective against insults that induce apoptosis

If androgen neuroprotection is specific to cell death that involves apoptosis, then only those insults empirically determined to induce apoptosis (i.e., Aβ, staurosporine, Apoptosis Activator II) should be attenuated by androgen treatment. To investigate this prediction, we compared the ability of testosterone to reduce cell death induced by the panel of seven toxins. Cultures were pretreated with 10 nM testosterone, a physiological concentration that we have previously determined exerts maximum levels of neuroprotection (30). Following 2 h testosterone pretreatment, cultures were exposed to increasing concentrations of each toxin, and then analyzed for cell viability by counts of live cells. We observed that androgens protected against neuronal death induced by Aβ, staurosporine, and Apoptosis Activator II across a range of toxin concentrations. Exposure for 24 h to Aβ (0–25 μM), staurosporine (0–0.5 μM), Apoptosis Activator II (0–7 μM), H₂O₂ (0–25 μM), FeCl_2/3 (0–2.5 μM), A23187 (0–350 nM), or 3-NP (0–2.5 mM) decreased cell viability with increasing concentration of toxin compared to vehicle-treated controls (Fig. 3). Pretreatment with 10 nM testosterone significantly reduced toxicity caused by Aβ (Fig. 3A), staurosporine (Fig. 3B), and Apoptosis Activator II (Fig. 3C) at most insult concentrations. In contrast, testosterone pretreatment did not attenuate cell death induced by H₂O₂ (Fig. 3D), FeCl_2/3 (Fig. 3E), A23187 (Fig. 3F), or 3-NP (Fig. 3G).

Fig. 3.

Testosterone is neuroprotective against cell death involving apoptosis. Cultures were treated with 10 nM testosterone (T) or vehicle for 2 h, exposed to increasing concentrations of (a) Aβ25–35, (b) staurosporine (STS), (c) Apoptosis Activator II (AAII), (d) H₂O₂, (e) FeCl_2/3, (f) A23187, or (g) 3-NP for 24 h, and processed for cell viability. Data show mean cell viability (± SEM) pooled from 3 independent experiments (n = 3). * P < 0.05 relative to the matched vehicle-treated condition.

In addition to acting directly on androgen receptors, testosterone is a prohormone that is metabolised in brain into several biologically active hormones. In particular, testosterone is converted to the oestrogen 17β-oestradiol (E2), which acts on oestrogen receptors, and the androgen dihydrotestosterone (DHT), which activates androgen but not oestrogen receptors. Further, DHT can be metabolised to 5α-androstane-3β, 17β-diol (3β-diol), which has been shown to interact with and activate oestrogen receptor β (ERβ) with an affinity similar to E2 (56). Our prior work has shown that testosterone neuroprotection does not involve oestrogen pathways, but rather is dependent upon activation of androgen receptors (30,32), which are expressed in our hippocampal neuron system (32) (and unpublished observations). To confirm an androgen mechanism, we first evaluated the ability of DHT to protect against the seven toxins. We observed that 10 nM DHT induced a pattern of protection parallel to that of testosterone, significantly reducing cell death induced by Aβ, staurosporine, and Apoptosis Activator II (Fig. 4A), but not affecting toxicity resulting from H₂O₂, FeCl_2/3, A23187, or 3-NP (Fig. 4B). Although we found that 10 nM E2 also significantly protected against the insults Aβ and Apoptosis Activator II (Fig. 4D), the DHT metabolite 3β-diol at 1 nM and 10 nM concentrations was not neuroprotective (Fig. 4C). Further, addition of the oestrogen receptor antagonist ICI 182,780 1 h prior to DHT pretreatment did not attenuate DHT protection (Fig. 4E).

Fig. 4.

DHT protects against cell death involving apoptosis by an oestrogen-independent mechanism. To examine neuroprotection, cultures were treated with hormone or vehicle for 2 h, exposed to 50 μM Aβ, 0.4 μM staurosporine (STS), 3 μM Apoptosis Activator II (AAII), 25 μM H₂O₂, 2.5 μM FeCl_2/3, 200 nM A23187, 2.5 mM 3-NP, or vehicle for 24 h, and processed for cell viability. Pretreatment with 10 nM DHT reduced neuronal death induced by apoptosis activators (a) Aβ, staurosporine, and Apoptosis Activator II, but not the non-apoptotic insults (b) H₂O₂, FeCl_2/3, A23187, and 3-NP. In parallel experiments, cell death by the apoptosis activators Aβ, staurosporine, and Apoptosis Activator II was significantly reduced by pretreatment with 10 nM 17β-oestradiol (E2) (d) but not by 1 nM (grey bars) or 10 nM (black bars) concentrations of the DHT metabolite (3β-diol) (c). (e) Neuroprotection afforded by 10 nM DHT was not affected by 1 h pretreatment with 1 μM ICI 182,780, (ICI) an oestrogen receptor antagonist. Data show mean cell viability (± SEM) pooled from 3 independent experiments (n = 3). * P < 0.05 relative to the matched vehicle-treated condition.

Androgens do not protect against oxidative stressors

To investigate the role of oxidative stress in the mechanism of toxin-induced cell death, we assessed the ability of the free radical scavenger Trolox to reduce toxicity induced by the different toxins. Confirming its well established antioxidant properties, treatment of cultures with 250 μM Trolox for 2 h before and during toxin exposure attenuated cell death induced by oxidative stressors H₂O₂ and FeCl_2/3 (Fig. 5B), but not by Aβ, staurosporine, and Apoptosis Activator II (Fig. 5A) nor by A23187 and 3-NP (Fig. 5B).

Fig. 5.

Antioxidants attenuate neuronal death induced by H₂O₂ and FeCl_2/3. Cultures were treated with 250 μM Trolox or vehicle for 2 h, exposed to 50 μM Aβ, 0.4 μM staurosporine (STS), 3 μM Apoptosis Activator II (AAII), 25 μM H₂O₂, 2.5 μM FeCl_2/3, 200 nM A23187, 2.5 mM 3-NP, or vehicle for 24 h, and processed for cell viability. Trolox reduced cell death caused by the oxidative insults (b) H₂O₂ and FeCl_2/3, but not by (a) Aβ, staurosporine, and Apoptosis Activator II nor by (b) A23187 and 3-NP. Data show mean cell viability (± SEM) pooled from 3 independent experiments (n = 3). * P < 0.05 relative to the matched vehicle-treated condition.

Interestingly, oestrogen exerts neuroprotection at micromolar levels by a direct anti-oxidant mechanism (57, 58). To determine if micromolar concentrations of androgens similarly protects, we tested the ability of 10 μM testosterone and 10 μM DHT to attenuate neuronal death induced by oxidative stressors. Consistent with prior reports (57, 58), we found that treatment of cultures with 10 μM E2 for 2 h before and during insult exposure resulted in protection against toxicity induced by H₂O₂ and FeCl_2/3 (Fig. 6D), but not by the other insults (Fig. 6A, D). In comparison, 10 μM testosterone and 10 μM DHT were not protective against the oxidative stressors H₂O₂ and FeCl_2/3 in our paradigm (Fig. 6E, F). The supraphysiological, 10 μM concentration of testosterone and DHT did not increase the efficacy of neuroprotection against apoptosis activator Aβ, staurosporine, and Apoptosis Activator II (Fig. 6B, C) in comparison to the physiological 10 nM concentration (Fig. 3A–C, 4A).

Fig. 6.

Micromolar concentrations of oestrogen, but not androgens can protect against cell death induced by oxidative stressors. Cultures were pretreated for 2 h with vehicle or a supra-physiological concentration (10 μM) of 17β-oestradiol (E2) (a, d), testosterone (b, e), or dihydrotestosterone (DHT), (c, f) then exposed to 50 μM Aβ, 0.4 μM staurosporine (STS), 3 μM Apoptosis Activator II (AAII), 25 μM H₂O₂, 2.5 μM FeCl_2/3, 200 nM A23187, 2.5 mM 3-NP, or vehicle for 24 h, and processed for cell viability. E2 reduced cell death induced by (d) oxidative stressor H₂O₂, FeCl_2/3, but not by (a) apoptotic (Aβ, staurosporine, Apoptosis Activator II) or other non-apoptotic (d) (A23187, 3-NP) insults. Testosterone protection against (b) apoptotic (Aβ, staurosporine, Apoptosis Activator II) and (d) non-apoptotic (H₂O₂, FeCl_2/3, A23187, 3-NP) insults is not enhanced at supra-physiological concentration. Similar results are seen with DHT against (c) apoptotic and (f) non-apoptotic insults. Data show mean cell viability (± SEM) pooled from 3 independent experiments (n = 3). * P < 0.05 relative to the matched vehicle-treated condition.

DISCUSSION

Although androgens are established regulators of neurone survival, the specificity of this action is unclear. In this study, we sought to further characterise androgen neuroprotection by determining what types of insults androgens protect against. Because our prior work demonstrated that androgen neuroprotection involves MAPK/ERK-dependent regulation of the apoptosis-related protein Bad (32), we hypothesised that androgen protection may be specific to cell death involving apoptosis. Consistent with this possibility, our results demonstrate that testosterone and its active androgen metabolite DHT selectively attenuate neuronal death induced by the apoptosis-inducing insults Aβ, staurosporine, and Apoptosis Activator II. Exposure of primary hippocampal neurone cultures to Aβ, staurosporine, or Apoptosis Activator II resulted in pyknotic nuclei and was dependent upon caspase activation and protein synthesis, as evidenced by inhibition of cell death with zVAD-fmk and cycloheximide, respectively. Pyknotic nuclei (52) and involvement of caspases and protein synthesis (53–55) are established components of apoptosis. Conversely, neither testosterone nor DHT reduced cell death caused by exposure to the insults H₂O₂, FeCl_2/3, A23187, and 3-NP. In our paradigm, these insults induced a form of cell death inconsistent with apoptosis as indicated by an absence of pyknotic nuclei and lack of protection resulting from inhibition of caspase activity and protein synthesis. Interestingly, cell death may involve multiple pathways that are not completely blocked by targeted protective agents, as evidenced by our findings that toxicity induced by Aβ, staurosporine, and Apoptosis Activator II was only partially reduced by caspase inhibition, protein synthesis inhibition, and androgens.

Whether specific insults induce apoptosis varies across different culture paradigms. The role of apoptotic pathways may depend not only on the specific insult, but also on insult severity and cellular factors including energy level (59). Consistent with our results, previous studies showed that apoptosis is the predominant form of cell death induced by Aβ (60), staurosporine (39), and Apoptosis Activator II (40). In some paradigms, Aβ can induce non-apoptotic cell death (61). Although H₂O₂, FeCl_2/3, A23187, and 3-NP induced non-apoptotic cell death in our paradigm, these insults can induce apoptosis depending on culture and treatment conditions. For example, calcium ionophore A23187 induces apoptosis at nanomolar concentrations, but necrosis at micromolar concentrations (44). 3-NP can also induce apoptotic and non-apoptotic death in cultured neurones depending on the presence or absence of astrocytes (45) and the level of glutamate (62). Similarly, H₂O₂ (41,42) and FeCl_2/3 (41) induce apoptosis in some culture paradigms. Thus, our classification of insults as either involving apoptosis is expected to vary according to paradigm. Consequently, we predict that the ability of androgens to protect against Aβ, staurosporine, Apoptosis Activator II, H₂O₂, FeCl_2/3, A23187, 3-NP, and other insults may similarly vary across paradigms depending on the relative involvement of apoptosis. Thus, we conclude that androgen neuroprotection is not necessarily insult-specific, but rather specific to cell death that predominantly involves apoptosis.

Despite abundant evidence that androgens are neuroprotective (3), there are also situations in which androgens do not increase neurone viability. For example, although oestrogen protects female mice from injury of nigrostriatal dopaminergic neurones caused by methamphetamines (63) or methyl-4-phenyl-1, 2, 3, 6-tetrahydropyridine (MPTP) (64), parallel studies with male mice fail to show a neuroprotective effect of testosterone (63, 64). One explanation for this apparent discrepancy is evidence that classic apoptosis may not be the primary degenerative mechanism resulting from methamphetamine (65) or MPTP (66), although other data suggest apoptotic mechanisms (67, 68). In the absence of a caspase-dependent apoptotic mechanism, our findings would predict that androgens would not protect against dopaminergic neuronal injury. Of course other factors, including regional or cell population differences in androgen responsiveness, may also contribute to observed differences in androgen neuroprotective ability and mechanism. For example, androgens regulate cell number through the apoptotic gene Bax in certain neurones of the bed of nucleus of the stria terminalis, spinal nucleus of the bulbocavernosus, and anteroventral periventricular nucleus of the hypothalamus, but not via Bax in vasopressin-expressing neurones of those regions (69, 70).

Interestingly, a few reports indicate that androgens can actually exacerbate some types of neural injury. In neurone culture paradigms, testosterone can be directly toxic at supraphysiological, micromolar concentrations (71) and increase excitotoxic injury at both micromolar (72) and nanomolar (37) concentrations. Toxic androgen actions may show sex specificity with more robust effects in neuron cultures derived from males (37); our use of cultures from a mixture of male and female rat pups precludes evaluation of whether protective androgen actions are also sex-specific. In animal models, striatal 3-NP lesions are increased by testosterone and reduced by oestrogen in ovariectomised female rats (36). Also, testosterone worsens neural injury caused by middle cerebral artery occlusion model of ischemia-reperfusion (73). Cheng et al. (35) found that DHT increases ischemic infarction volume, perhaps through DHT-regulation of apoptosis-related genes sphingosine kinase 1, B-cell leukemia/lymphoma 2 related proteinA1, and inhibitor of apoptosis protein 1. However, infarction volume may be dose-dependent, with lower androgen doses protecting against and higher doses exacerbating ischemic insults (74). One possibility is that the same signaling pathways that contribute to androgen neuroprotection against some insults may potentiate other insults. For example, our prior work demonstrated that androgen neuroprotection involves activation of a MAPK/ERK signaling pathway (32). However, activation of MAPK/ERK signaling can also drive neuronal cell death in several paradigms (75), including ischemia-reperfusion models (76). Androgen activation of protective versus cell death-promoting pathways may also depend on the relative contributions of intracellular versus membrane-associated androgen receptors (77). Alternatively, as in the case of spinal and bulbar muscular atrophy, androgen-dependent degeneration may result from androgen interaction with androgen receptors containing extended polyglutamine repeat (78). Thus, although our data suggest a straightforward association of androgen neuroprotection with apoptotic insults, the relationship is likely more complex.

Our results also suggest that androgen neuroprotection may not involve direct anti-oxidant action previously demonstrated with oestrogen. Consistent with earlier reports (57,79), we found that supraphysiological, micromolar concentrations of oestrogen are neuroprotective against established oxidative insults. In contrast, micromolar concentrations of testosterone and DHT neither enhanced androgen neuroprotection nor protected against cell death induced by the oxidative stressors H₂O₂ and FeCl_2/3. The observed difference between oestrogen and the androgens testosterone and DHT in terms of anti-oxidant action may reflect the presence of a phenol group on the A ring of oestrogen (79) that is not found on either testosterone or DHT. A phenol-containing structure is common to many anti-oxidants, including vitamin E, and contributes to the neutralization of reactive oxygen species. A prior report demonstrated that testosterone protected cultured cerebellar neurones from cell death caused by the pro-oxidant H₂O₂ (33). Importantly, the data did not indicate a direct anti-oxidant androgen action, but rather a genomic mechanism involving increased expression of the anti-oxidant enzyme catalase (33). However, whether H₂O₂ induced apoptosis in this cerebellar neurone paradigm was not determined, and therefore we cannot discount the idea that androgens can protect against oxidative injury that involves apoptosis as the primary mechanism of cell death. Nonetheless, it is noteworthy that androgen neuroprotection may involve regulation of oxidative stress via induction of antioxidant defenses (80–82).

An important cellular function of androgens is promotion of neurone viability, not only during development (21,83), but also in adulthood (28). Apoptosis plays a key role in the regulation of neuronal survival in both development and age-related neurodegenerative diseases. In this study, we have established that androgens selectively protect cultured neurones from cell death involving apoptosis. As evidence accumulates linking age-related androgen depletion with enhanced risk for neurodegenerative disorders (3), continued understanding of androgen neuroprotection will be necessary to successfully exploit the therapeutic potential of androgen compounds for the treatment and or prevention of neurodegenerative diseases.