The Relevance of Neuroactive Steroids in Schizophrenia, Depression, and Anxiety Disorders

Abstract

1. Neuroactive steroids are steroid hormones that exert rapid, nongenomic effects at ligand-gated ion channels. There is increasing awareness of the possible role of these steroids in the pathology and manifestation of symptoms of psychiatric disorders. The aim of this paper is to review the current knowledge of neuroactive steroid functioning in the central nervous system, and to assess the role of neuroactive steroids in the pathophysiology and treatment of symptoms of schizophrenia, depression, and anxiety disorders. Particular emphasis will be placed on GABA_A receptor modulation, given the extensive knowledge of the interactions between this receptor complex, neuroactive steroids, and psychiatric illness.

2. A brief description of neuroactive steroid metabolism is followed by a discussion of the interactions of neuroactive steroids with acute and chronic stress and the HPA axis. Preclinical and clinical studies related to psychiatric disorders that have been conducted on neuroactive steroids are also described.

3. Plasma concentrations of some neuroactive steroids are altered in individuals suffering from schizophrenia, depression, or anxiety disorders compared to values in healthy controls. Some drugs used to treat these disorders have been reported to alter plasma and brain concentrations in clinical and preclinical studies, respectively.

4. Further research is warranted into the role of neuroactive steroids in the pathophysiology of psychiatric illnesses and the possible role of these steroids in the successful treatment of these disorders.

INTRODUCTION

The term ‘‘neurosteroids’’ was defined in 1981 by Baulieu and coworkers (Corpechot et al., 1981; Baulieu, 1997), who found that a number of steroid hormones existed in higher concentrations in the nervous system than in the plasma, and that these steroids were synthesized in the brain. It is now known that these neurosteroids can be metabolized in the brain from precursor compounds originating from endocrine sources, and can also be synthesized de novo in the brain from cholesterol. Neuroactive steroids are steroids that produce rapid, nongenomic actions in the brain, generally through actions on ligand- or voltage-gated channels (Stoffel-Wagner, 2001). These steroid hormones differ significantly from classic steroid hormones, which are synthesized primarily in peripheral structures and act mainly upon intracellular receptors, ultimately resulting in long-lasting, genomic effects (Truss and Beato, 1988). Neuroactive steroids can either positively or negatively modulate the activity of a number of ligand-gated ion channel-associated receptors including γ-aminobutyric acid type A (GABA_A), serotonin type 3 (5-HT₃), glycine, and glutamate N-methyl-d-aspartate (NMDA), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA), and kainate receptors. Examples of neuroactive steroids include (a) the pregnane steroids 4-pregnene-3,20-dione (progesterone; P), 3β-OH-5-pregnen-20-one (pregnenolone; Preg), pregnenolone sulfate (PregS), 3α-OH-5β-pregnan-20-one (pregnanolone), 3α-hydroxy-5α-pregnan-20-one (allopregnanolone; ALLO), 3α,21-dihydroxy-5α-pregnan-20-one (allotetrahydrodeoxycorticosterone; THDOC), and dihydrodeoxycorticosterone (DHDOC); and (b) the androstane steroids OH-androst-5-en-20-one (dehydroepiandrosterone; DHEA), and DHEA-sulfate (DHEAS) (see Rupprecht, 2003 for review). The structures of some of these steroids are illustrated (Fig. 1).

Fig. 1.

Chemical structures of neuroactive steroids frequently referred to in this review.

Neuroactive steroids are essential for the proper development and functioning of the adult brain and play a major role in the stress response. Based on plasma and cerebral spinal fluid (CSF) level studies in humans and on studies on the effects of drugs on brain and plasma levels in laboratory animals, it appears likely that these steroids may contribute to the pathology and symptoms of some psychiatric illnesses and may be affected by drugs used to treat these disorders (Le Melledo and Baker, 2002; Marx et al., 2003; Dubrovsky, 2005; Pinna et al., 2006). Consistent with this, some neuroactive steroids have been reported to possess anxiolytic, antidepressant, antipsychotic, anticonvulsant, ataxic, and/or anesthetic properties in animals (Crawley et al., 1986; Landgren et al., 1987; Belelli et al., 1990; Bitran et al., 1993; Korneyev and Costa, 1996). At the cellular level, they may exert neuroprotective effects (both in vitro and in vivo), promote neurogenesis, and aid in the regeneration of neurons following injury, including the promotion of myelin formation (Schumacher et al., 1996; Waters et al., 1997; Kimonides et al., 1998; Mayo et al., 2001; Keller et al., 2004; Schumacher et al., 2004). A significant role for some neuroactive steroids in cognition has also been observed in animals, as assessed by performance on a number of memory tasks (Flood et al., 1992; Mayo et al., 1993; Vallee et al., 1997). Given these findings, it is not surprising that a great deal of research is now being devoted to the role of neuroactive steroids in a number of psychiatric disorders, including attention deficit hyperactivity disorder, premenstrual dysphoric disorder, anxiety disorders, depression, and schizophrenia (Romeo et al., 1998; Girdler et al., 2001; Strous et al., 2001; Le Melledo and Baker, 2002; Pisu and Serra, 2004).

This review will focus on the putative role of neuroactive steroids in a number of psychiatric illnesses, as well as how fluctuations in their concentrations may contribute to the therapeutic effects of pharmacological treatments. Neuroactive steroid actions at the GABA_A receptor complex will be emphasized, given the extensive amount of research devoted to the actions of neuroactive steroids at this receptor and its relevance to the pathophysiology of mental disorders.

NEUROACTIVE STEROID FUNCTION IN THE CENTRAL NERVOUS SYSTEM

Neurosteroidogenesis occurs in the mitochondria of neurons and glial cells in the central and peripheral nervous systems. In the rate-limiting step of neurosteroidogenesis, brain cholesterol (Dietschy and Turley, 2001) is transported through the mitochondrial membrane by the concurrent actions of both the steroidogenic acute regulatory (StAR) protein and peripheral benzodiazepine receptors (PBRs) (Hauet et al., 2005). Cholesterol is then converted to pregnenolone in the inner mitochondrial membrane by the P450 side chain cleavage (P450scc) enzyme. Pregnenolone is further metabolized to the pregnane and androstane steroids by P450 and nonP450 enzymes present in different cellular compartments. 3α-Hydroxysteroid dehydrogenase (3α-HSD) and 5α-reductase (5α-R) are responsible for the conversion of P and DHDOC to ALLO and THDOC, respectively (Robel et al., 1999). The action of 3α-HSD is reversible, and the direction of the reaction is dependent upon the cellular conditions. Studies have suggested that increasing GABAergic neurotransmission, through activation of the GABA_A receptor complex in the hypothalamus, inhibits the activity of these steroidogenic enzymes (Do-Rego et al., 2000). Given that a number neuroactive steroids enhance GABA_A receptor activity, it has been suggested that, in this manner, neuroactive steroids may regulate their own biosynthesis (Do-Rego et al., 2000).

Neuroactive steroids act on a variety of receptors associated with ligand-gated ion channels. For example, the 5-HT₃ receptor has been shown to be negatively modulated by ALLO, PregS, P, and 17α- and 17β-estradiol (Wetzel et al., 1998), the NMDA receptor positively modulated by PregS (Wu et al., 1991; Irwin et al., 1992), and the σ₁ receptor positively modulated by DHEA and DHEAS and negatively modulated by P. However, neuroactive steroid modulation of the GABA_A receptor has received the most attention. ALLO and THDOC are among the most potent known positive allosteric modulators of the GABA_A receptor, whereas DHEA, DHEAS, and PregS are negative modulators at this receptor complex (Demirgoren et al., 1991; Park-Chung et al., 1999). Neuroactive steroids can either increase or decrease the GABA-evoked Cl⁻ current by altering the duration and frequency of the channel open time (Lambert et al., 1995). It is thought that in this way, they alter the length of the refractory period (after depolarization), thus altering the excitability of the cell (Mellon and Griffin, 2002). At higher doses (higher than physiological levels), ALLO and THDOC can exert positive modulatory effects in the absence of GABA (Puia et al., 1990). As neuroactive steroids may modulate the activity of a particular receptor in opposite ways, knowledge of levels of individual steroids to one another is important to determine the net effect. A table summarizing modulatory effects of neuroactive steroids on different receptors is provided in the recent review by Dubrovsky (2005).

The GABA_A receptor is a ligand-gated Cl⁻ channel and is primarily inhibitory in nature; it is a heteromeric pentamer with a number of possible subunit combinations (for example, α₁–α₆, β₁–β₄, γ₁–γ₄, ρ₁–ρ₃, δ, ɛ, and θ subunits exist (Olsen and Betz, 2006)). The subunit combination of each particular receptor determines its pharmacological and physiological properties (Smith, 2002). The GABA_A receptor has many binding sites for depressant/sedative agents, such as benzodiazepines, barbiturates, anesthetics, and anticonvulsant drugs. Each of these appear to have a unique binding site on the receptor surface, determined by the subunit composition of the receptor. The binding site for benzodiazepines, for example, is formed by α and γ subunits (Scholtz et al., 1996) (with the exception of the α₄ and α₆ subunits, which are benzodiazepine-insensitive) (Smith, 2001). GABA binds to its binding site in the cleft between α and β subunits (Smith, 2002).

The binding sites of neuroactive steroids on the GABA_A receptor have not been definitively characterized, although the α and β subunits were shown to mediate sensitivity of the GABA_A receptor to neuroactive steroids (Maitra and Reynolds, 1999). GABA_A receptors containing the ɛ subunit were shown to be insensitive to the modulatory effects of the pregnane steroids (Davies et al., 1997), whereas the presence of the δ subunit altered the modulatory effects of neuroactive steroids (Zhu et al., 1996; Belelli et al., 2002; Wohlfarth et al., 2002; Belelli and Lambert, 2005). Evidence further suggests that there are multiple steroid binding sites at the GABA_A receptor (Gee et al., 1988; Park-Chung et al., 1999; Ueno et al., 2004), indicating that the ratio of positive and negative modulating steroids may play an important role in regulating overall GABA_A receptor activity. Neuroactive steroid binding sites also appear to be distinct from the binding sites of other GABA_A receptor modulators (such as benzodiazepines and barbiturates) (Gee et al., 1987, 1988; Morrow et al., 1990), with the possible exception of DHEAS, which has been suggested to act at the picrotoxin site (Sousa and Ticku, 1997).

NEUROACTIVE STEROIDS AND STRESS

Activation of the hypothalamic-pituitary-adrenal (HPA) axis initiates the behavioural and physiological responses to stress. Corticotropin releasing hormone (CRH) and arginine-vasopressin (AVP), released from the paraventricular nucleus of the hypothalamus, stimulate the release of adrenocorticotrophic hormone (ACTH) from the pituitary gland, which then stimulates the release of glucocorticoids (primarily cortisol in humans, corticosterone (CORT) in rats) from the adrenal cortex. Negative feedback mechanisms exist at various sites of the HPA axis to regulate its activity, with mineralocorticoid and glucocorticoid receptors being the primary sites of regulation.

Neuroactive steroids play an important role in both acute and chronic stress. Because stress often plays a significant role in the manifestation or exacerbation of symptoms in psychiatric illness, the role of neuroactive steroids in the modulation of the stress response may be relevant to the manifestation of these disorders.

Acute Stress

Acute stress induced in animals by forced swimming, mild foot shock, or carbon dioxide exposure transiently alters the concentration of some neuroactive steroids in both the brain and plasma. Along with CORT, P, Preg, ALLO, THDOC, and dihydroprogesterone concentrations are significantly elevated in the brain and plasma of acutely-stressed animals compared to nonstressed controls (Purdy et al., 1991; Barbaccia et al., 1994, 1996a,b, c,1997, 1998; Vallee et al., 2000; Higashi et al., 2005). Acute stress is also associated with a downregulation of GABA_A receptor functioning, shown in rodents by increased ³⁵S-TBPS binding, decreased GABA_A receptor density, decreased [³H] GABA binding, decreased Cl⁻ flux in brain synaptoneurosomes, and through a variety of behavioural tests (Biggio et al., 1981; Biggio, 1983; Havoundjian et al., 1986; Trullas et al., 1987; Concas et al., 1988; Drugan et al., 1989). Reduction or complete blockade of GABAergic neurotransmission also mimics the physiological and behavioural effects of stress in rodents, inducing anxiety-like behaviour (Crestani et al., 1999; Kash et al., 1999) and increasing brain and plasma concentrations of cortisol, ALLO, and THDOC (Barbaccia et al., 1996c, 1997, 1998). These effects are antagonized by prior administration of positive modulators of the GABA_A receptor (Barbaccia et al., 1996c, 1998).

Stress-induced increases in P, ALLO, and THDOC may be endogenous events that serve to protect against the effects of stress (Purdy et al., 1991; Majewska, 1992; Paul and Purdy, 1992); increased concentrations of ALLO and THDOC counteract the inhibitory effect of stress on GABAergic transmission, likely leading to normalization of HPA axis activity. The inhibitory effects of GABAergic transmission on HPA axis activity are well documented, and an increase in the concentration of ALLO and THDOC would increase the GABAergic tone mediating HPA axis activity. Several lines of evidence support an involvement of neuroactive steroids in the regulation of HPA axis activity. ALLO significantly alters the biosynthesis and release of CRH (Patchev et al., 1994) and increases the basal release of AVP via GABA_A receptor activation (Hansen et al., 2003), indicating that the pituitary gland is a target of ALLO. It has been reported that ALLO administration attenuates the ACTH and CORT response to emotional stress in rats (Patchev et al., 1996). Furthermore, central administration of ALLO antiserum enhances the CORT response to cold swimming stress (Guo et al., 1995), and administration of finasteride (which blocks ALLO synthesis) increases the magnitude and duration of the dopamine response to acute stress (Dazzi et al., 2002). Behavioral tests have also confirmed the role of ALLO in mediating the stress response, as ALLO administration has been shown to counteract the anxiogenic effects of CRH in rats (Patchev et al., 1994). Interestingly, ALLO positively modulates GABA_A receptor function more efficiently when GABA levels are low (Concas et al., 1996), such as during periods of acute stress. Together these results suggest that ALLO may be particularly efficient in regulating the physiological response to stress by reducing HPA axis activity through its actions at the GABA_A receptor.

THDOC is another potent positive modulator of the GABA_A receptor, and has been suggested to be a neuroactive steroid related to homeostatic mechanisms responding to stress (Reddy, 2003); aside from its modulatory effects on the GABA_A receptor following acute stress, evidence for a role of THDOC in the maintenence of normal brain functioning is lacking. Given that the majority of this steroid is of adrenal origin, the possibility that THDOC is a primarily stress-related steroid cannot be discounted. Neonatal administration of THDOC during maternal separation resulted in reduced anxiety-like behaviour in adulthood and decreased CRH concentrations in the hypothalamus (Patchev et al., 1997), suggesting that THDOC may play a role in the development of the HPA axis.

The mechanism by which stress causes increased concentrations of neuroactive steroids remains unclear. Alterations in the concentration of steroidogenic enzymes are unlikely, since stress-induced increases in neuroactive steroids are evident within 5 min of the onset of stress, and reach nearly 200% of control values within 30 min (Purdy et al., 1991); stimulating the transcription of the genes encoding for these enzymes would be much slower. Instead, a rapid change in the rate of steroidogenesis may occur at the level of cholesterol transport from the outer mitochondrial membrane to the inner mitochondrial membrane (Stocco and Clark, 1996). This requires the actions of both PBRs and StAR. Acute stress has been reported to upregulate the density of PBRs in the cortex, hippocampus (Basile et al., 1987; Rago et al., 1989; Avital et al., 2001), and blood platelets (Droogleever Fortuyn et al., 2004), an effect that could lead to a rapid increase in steroidogenesis in these regions. Indeed, acute stress-induced increases in blood platelet PBR density were correlated with plasma ALLO concentration (Droogleever Fortuyn et al., 2004). However, PBR density remains elevated 72 h following the onset of stress (Avital et al., 2001), whereas stress-induced increases in ALLO and THDOC concentrations return to control values within 120 min of acute stress onset (Barbaccia et al., 2001b). These temporal discrepancies imply that while changes in PBR density may contribute to the stress-induced changes in the concentrations of some neuroactive steroids, other mechanisms independent of PBR activity are also likely involved. Furthermore, reports on acute stress and PBR density are inconsistent; acute stress was also shown to decrease PBR density in the adrenal glands (Avital et al., 2001). Little research has been conducted on the effects of acute stress on StAR protein expression, although rats that were water deprived for 48 h exhibited increased StAR concentrations in the adrenal glands (Ulrich-Lai and Engeland, 2002). Furthermore, other events known to increase neuroactive steroid concentrations, such as acute ethanol administration (Kim et al., 2003), upregulate StAR expression in the brain, and upregulation of StAR protein expression has, in fact, been shown to result in increased pregnenolone synthesis (Clark et al., 1994). Therefore, stress-induced increases in StAR might contribute to increases in neuroactive steroid concentrations. Whether upregulation of StAR expression occurs in the brain immediately preceding acute stress is presently not known, and warrants investigation.

Stimulation of the HPA axis may also lead to increases in some neuroactive steroids. Stress-induced decreases in GABAergic transmission likely result in disinhibition of the HPA axis, resulting in increased hypothalamic and adrenal output. In this manner, neuroactive steroids of adrenal origin may be upregulated. Consistent with this, adrenalectomized and gonadectomized animals do not exhibit a stress-induced increase in neuroactive steroid concentrations (Purdy et al., 1991; Barbaccia et al., 1997).

Chronic Stress

The effects of chronic stress on neuroactive steroid concentrations have been reported to differ from those of acute stress. Mild chronic stress, induced in rodents by social isolation, has been shown to cause a significant decrease in P, Preg, ALLO, and THDOC in both the brain and plasma compared to group-housed controls (Matsumoto et al., 1999; Serra et al., 2000; Dong et al., 2001; Serra et al., 2004; Matsumoto et al., 2005), whereas DHEA levels remain unchanged (Serra et al., 2000). Neuroactive steroid changes were not evident after 48 h of chronic stress, but decreases were present after 7 days of chronic stress, and remained so after 30 days (Serra et al., 2000). Chronic stress is also associated with a decrease in GABA_A receptor function (Serra et al., 2000), possibly leading to the well-established disinhibition of HPA axis activity observed following chronic stress (Bartanusz et al., 1993; Herman et al., 1995; Aguilera and Rabadan-Diehl, 2000).

The mechanism by which chronic stress decreases the concentration of these neuroactive steroids is unclear. Social isolation has been shown to downregulate brain 5α-R mRNA expression (Dong et al., 2001), although this likely does not fully explain the alterations in steroids observed, as concentrations of all neuroactive steroids are not reduced following chronic stress (Barbaccia et al., 2001b). An alteration in the secretory ability of the adrenal gland is also unlikely, since administration of an acute stressor in chronically stressed animals causes a significant increase in Preg and ALLO. Interestingly, social isolation decreases the brain concentration of PBRs (Serra et al., 2004), while chronic food deprivation stress significantly decreases (and repeated swim stress tends to decrease) the density of PBRs in the adrenal gland (Weizman et al., 1990; Burgin et al., 1996). This decrease in PBR density may contribute to the reduced synthesis of neuroactive steroids in these regions. However, the reduction in PBR density was shown to be evident 24 h after repeated swim stress, whereas neuroactive steroid changes were not evident for 48 h (Avital et al., 2001), suggesting that other mechanisms contribute to the reduction in neuroactive steroid concentration. Expression of the StAR protein may be altered in response to chronic stress, although no data on this subject have been reported.

Alternatively, altered activity of the HPA axis induced by chronic stress may result in dysregulated neuroactive steroid concentrations. While the importance of the adrenal glands in acute stress-induced increases in neuroactive steroids has been established, the effects of chronic stress on HPA axis functioning are complex, and tend to depend upon the type of stressor. However, it has been shown in rats that chronic stress both upregulates CRH mRNA in the brain (Makino et al., 1995; Albeck et al., 1997) and activates CRH-releasing neurons (Rivier and Vale, 1987; de Goeij et al., 1991; Ojima et al., 1995), and that chronic exposure to CRH in vitro results in the desensitization of pituitary cells to its stimulating effect on ACTH release (Hoffman et al., 1985). Decreased ACTH release may result in reduced adrenal gland activity, and thus decreased synthesis and basal secretion of some neuroactive steroids synthesized in the adrenal glands (Serra et al., 2000). Interestingly, ACTH has been shown to induce steroidogenesis; thus decreased basal ACTH release due to chronic stress might result in reduced steroidogenesis. This possibility is further supported by the finding that chronic stress results in HPA axis hyperresponsiveness, as chronically stressed animals subjected to a subsequent acute stressor exhibit significantly greater plasma and brain elevations in P, Preg, ALLO, and THDOC compared to nonchronically stressed animals subjected to the same acute stressor (Barbaccia et al., 2001b).

NEUROACTIVE STEROIDS IN SCHIZOPHRENIA

The role of stress in the development and exacerbation of psychosis in schizophrenia has been demonstrated, and several studies have shown abnormalities in the functioning of the HPA axis in these patients (Gispen-de Wied, 2000). Abnormal dopaminergic activity has been well documented in schizophrenia (Seeman and Kapur, 2000). Deficits in the GABAergic (Blum and Mann, 2002; Wassef et al., 2003; Costa et al., 2004) system have been well documented in schizophrenia, and in recent years there has been a growing body of evidence in support of deficits in GABAergic and glutamatergic activity in schizophrenia (Blum and Mann, 2002; Coyle et al., 2003; Konradi and Heckers, 2003; Wassef et al., 2003; Costa et al., 2004). Neuroactive steroids modulate the activity of all of these systems, both directly and indirectly, and therefore may contribute to the pathophysiology of the illness. Regulation of the concentrations of these steroids may also play a role in the therapeutic benefits of antipsychotic drugs, particularly those whose actions involve the GABA_A receptor complex (see also Shulman and Tibbo, 2005).

Progesterone and its metabolites may be dysregulated in schizophrenia. While some studies have reported comparable levels of P in individuals with schizophrenia and in controls (Shirayama et al., 2002), it has been shown elsewhere that males with chronic schizophrenia exhibited a trend for elevated plasma P levels compared to healthy controls, and that following metabolic stress induced by injection of 2-deoxyglucose, schizophrenic patients exhibited a significantly greater increase in plasma P compared to healthy controls (Breier and Buchanan, 1992). Plasma P levels have also been shown to be lower in schizophrenia than in healthy controls (Taherianfard and Shariaty, 2004), consistent with the finding that the density of PBRs is decreased in some brain regions in schizophrenia (Kurumaji et al., 1997). Preliminary evidence also suggests that in nonmedicated first-episode patients, plasma Preg and ALLO may be slightly, although not significantly, decreased and that Preg may be negatively correlated with illness severity (Marx et al., 2004). Increased plasma ALLO was also shown to be associated with aggression and hostility in a small number of individuals with schizophrenia (Spalletta et al., 2005).

Evidence suggests that the androstanes DHEA and DHEAS may also be dysregulated in schizophrenia, although results are conflicting. Although it has been reported that plasma levels of these steroids do not differ between patients with schizophrenia and controls (Brophy et al., 1983; Harris et al., 2001; Shirayama et al., 2002), older studies reported reduced plasma DHEA concentrations in schizophrenics (Tourney and Hatfield, 1972; Oertel et al., 1974), particularly in the morning (Tourney and Erb, 1979). More recent studies, arguably more accurate (due to improved methodological techniques), have reported elevated plasma levels of DHEA and DHEAS in severely psychotic male subjects (Howard, 1992), medicated patients with chronic schizophrenia (Oades and Schepker, 1994; di Michele et al., 2005) and in nonmedicated first-episode patients (Strous et al., 2004) compared to controls. Abnormal plasma diurnal rhythms of DHEA, but not DHEAS, were observed in schizophrenia such that patients were distinguished from controls with 100% accuracy (Erb et al., 1981). It has been suggested that cortisol/DHEA or cortisol/DHEAS ratios may be more appropriate measures than DHEA or DHEAS alone (Hechter et al., 1997), and elevated ratios of cortisol/DHEA and cortisol/DHEAS have also been reported in schizophrenics compared to controls (Ritsner et al., 2004). Interestingly, however, DHEAS levels have been shown to be negatively correlated with aggression and symptom severity in nonmedicated first-episode patients (Strous et al., 2004), consistent with the observation in chronic patients that low serum DHEA levels and a lower ratio of DHEA/cortisol were associated with more severe symptomotology (Harris et al., 2001). Paradoxically, then, it appears that high levels of circulating DHEA and/or DHEAS are evident in schizophrenia, yet among this population higher steroid concentrations are associated with superior functioning.

Dysregulated steroid concentrations may contribute to the physiological and/or functional abberations evident in the schizophrenic brain. Decreased concentrations of the positively-modulating GABAergic steroids ALLO and THDOC, together with elevated concentrations of negatively-modulating GABAergic steroids DHEA and DHEAS, may contribute to the deficits in GABAergic functioning reported in schizophrenia. Hyperactivity of the serotonergic system in schizophrenia may also contribute to certain aspects of the illness (Akhondzadeh, 2001). Like some antipsychotic drugs (such as clozapine and haloperidol) (Hermann et al., 1996; Rammes et al., 2004), P and ALLO are functional antagonists of the 5-HT₃ receptor (Wetzel et al., 1998), and decreased concentrations of these steroids may contribute to excessive serotonergic activity. It has also been suggested that σ₁ receptors may be involved in the pathogenesis of schizophrenia (Hayashi and Su, 2004), and it is possible that elevated concentrations of DHEA and DHEAS, positive modulators of the σ₁ receptor (Monnet et al., 1995; Maurice et al., 1997), together with decreased concentrations of P, a σ₁ receptor antagonist (Monnet et al., 1995), may be contributing factors in schizophrenia. Finally, DHEA and DHEAS have been shown to stimulate NMDA receptor activity via the σ₁ receptor (Bergeron et al., 1996); therefore excessive σ₁ receptor activation may lead to hyperactivity of the NMDA receptor, and to the downregualtion of NMDA receptor density reported in schizophrenia (see for example Gao et al., 2000).

Antipsychotic Potential of Neuroactive Steroids

Progesterone and its metabolites exhibit antipsychotic properties, suggesting that decreased concentrations of these steroids may confer vulnerability to psychosis and increased concentrations may provide therapeutic benefit. In several behavioural paradigms, P and its metabolites have been shown to possess similar properties to antipsychotic drugs; as observed after clozapine and haloperidol administration (Bristow et al., 1997; Khisti et al., 2002), intracerebroventricular administration of ALLO inhibited amphetamine-induced motor hyperactivity (Khisti et al., 2002). P and ALLO administered intraperitoneally and intracerebroventricularly, respectively, also inhibited the conditioned avoidance response in rodents (Ugale et al., 2004), an effect also observed following intraperitoneal injection of olanzapine, risperidone, and haloperidol (Wadenberg et al., 2001; Ugale et al., 2004). Furthermore, P reverses the disruptive effects of apomorphine on prepulse inhibition in rodents (Rupprecht et al., 1999), and intraperitoneal injection of P and intracerebroventricular injection of ALLO antagonize apomorphine-induced climbing behaviour in rodents (Khisti et al., 2002; Ugale et al., 2004), a behavioural test proposed to be indicative of antipsychotic activity (Costall et al., 1978). These antipsychotic-like effects are likely due to interactions between pregnane steroids, the GABAergic system and the dopaminergic system; dopamine-mediated behaviours are inhibited by ALLO administration, suggesting that ALLO inhibits dopaminergic neurotransmission (Khisti et al., 2002). This possibility is supported by studies showing that neuroactive steroids that positively modulate the GABA_A receptor decrease K⁺-evoked [³H]DA release from the nucleus accumbens, whereas negative modulators of this receptor increase DA release in this region (Jaworska-Feil et al., 1998). Furthermore, intraventricular administration of THDOC dose-dependently reduced DA metabolites in the prefrontal cortex (Grobin et al., 1992), and intraventricular administration of ALLO decreased DA content of the prefrontal cortex and nucleus accumbens (Motzo et al., 1996). Therefore, the antipsychotic-like effects of ALLO may arise from blockade of DAergic transmission. Moreover, it has been postulated that P enhances the neuroprotective effects of estrogen against psychosis in schizophrenia by enhancing the expression of D₅ receptors (Lee et al., 2001), whose upregulation may play a role in the improvement of negative symptoms of the illness (Sanfilipo et al., 1996).

The notion of pregnane steroids possessing antipsychotic properties is further substantiated by gender differences, given that females have higher circulating concentrations of pregnane steroids than do males. In general, females are more likely to be responsive to antipsychotic medication, and tend to have less severe forms of the illness than males. Also, females tend to develop schizophrenia at a later age than do males, and females show a second period of increased incidence of the illness at the time of menopause (when the concentrations of circulating pregnane steroids are significantly reduced) (Hafner et al., 1993).

Consistent with the hypothesis that neuroactive steroids may be protective against some symptoms of psychosis, several atypical antipsychotic drugs increase pregnane steroid concentrations in rodents. Acute administration of clozapine and olanzapine, but not haloperidol, sulpiride, or risperidone, to rats caused a significant, dose-dependent increase in plasma and brain levels of P, ALLO, and THDOC (Marx et al., 2000; Barbaccia et al., 2001a; Marx et al., 2003) to an extent shown to potentiate the activity of the GABA_A receptor (Morrow et al., 1987). The increases in steroid concentrations were transient; while cortical and striatal P, ALLO, and THDOC were significantly increased 45, and to a lesser extent, 90 min following drug injection, steroid concentrations returned to baseline values 180 min after clozapine injection (Barbaccia et al., 2001a). Interestingly, drug administration to adrenalectomized (ADX) animals abolished the drug-induced increase in steroid concentration (Marx et al., 2003), implicating the adrenal glands in antipsychotic drug-induced increases in neuroactive steroids. Furthermore, acute challenge with clozapine in animals chronically treated with clozapine induced a significant increase in pregnane steroids in most animals (Barbaccia et al., 2001a). The possible importance of drug-induced increases in pregnane steroids in the action of antipsychotic drugs was highlighted by a study investigating the olanzapine-induced increases of ALLO. It was shown that both olanzapine and ALLO blocked conditioned avoidance response and apomorphine-induced climbing behaviour in rodents, and that subeffective doses of olanzapine and ALLO combined lead to antipsychotic-like effects (Ugale et al., 2004). Furthermore, blockade of olanzapine-induced increases in brain levels of ALLO abolished these antipsychotic-like effects, as did pharmacological blockade of the GABA_A receptor (Ugale et al., 2004). These results indicate that drug-induced increases in ALLO concentrations modulate the activity of the GABA_A receptor, and together these effects are vital for at least some of the antipsychotic-like effects of olanzapine.

While antipsychotic treatment affects plasma and brain levels of neuroactive steroids in rodents, it has been shown in humans that morning plasma levels of ALLO and THDOC were not altered in treatment-resistant patients treated long-term with clozapine, although clinical improvement was observed (Monteleone et al., 2004). While these results might indicate that the therapeutic benefit of clozapine may not be related to alterations in circulating levels of pregnane steroids, the possibility of drug-induced alterations in diurnal rhythms of neuroactive steroids remains unexplored. Furthermore, a considerable amount of time had elapsed between the time of drug administration and the time of steroid measurement (i.e. more acute effects on neuroactive steroid concentrations were not investigated). Clinical improvement may also be associated with the downstream effects of altered neuroactive steroid concentrations rather than their plasma concentrations per se. For example, transient increases in pregnane steroids induced by clozapine and olanzapine (see above) may lead to longer-term increases GABAergic transmission in various brain regions. Transiently increased concentrations of some neuroactive steroids could also lead to changes in GABA_A receptor subunit composition. Indeed, alterations in mRNA levels of certain subunits have been demonstrated in individuals with schizophrenia (Huntsman et al., 1998; Ohnuma et al., 1999), and abnormalities in specific GABA_A receptor–ligand binding in individuals with schizophrenia compared to healthy controls (see Blum and Mann, 2002 for review) may also reflect altered GABA_A receptor compositions.

Physiological and behavioural evidence suggesting that androstane steroids may possess antipsychotic properties is lacking, although it has been observed in individuals with schizophrenia that DHEA administration alone or in conjunction with antipsychotic drugs leads to the improvement of negative, anxiety, and depressive symptoms of the illness (Strauss et al., 1952; Strous et al., 2003). However, animals treated with clozapine for 8 days and sacrificed two hours after the last drug injection exhibited significantly decreased cortical DHEA and DHEAS compared to animals treated with haloperidol or vehicle (Nechmad et al., 2003a). This result supports preliminary findings in a small number of chronic patients taking clozapine alone that dosages were negatively correlated with plasma DHEA levels (di Michele et al., 2005). If elevated concentrations of DHEA and DHEAS are a feature of schizophrenia, it is possible that a reduction of these elevated steroid concentrations may play a role in the clinical effectiveness of atypical antipsychotic drugs. How administration of DHEA in patients, then, can improve symptoms is unclear, though it has been observed that DHEA administration in humans is accompanied by an increase in plasma P and ALLO (Stomati et al., 2000; Nadjafi-Triebsch et al., 2003), whose antipsychotic-like effects have been described (see above). A DHEA-induced increase in GABAergic activity via these steroids may contribute to the clinical benefit of DHEA in patients with schizophrenia. Alternatively, clozapine-induced decreases in DHEA and DHEAS may be irrelevant to its clinical efficacy.

Like neuroactive steroids, some antipsychotic drugs alter GABAergic transmission. While clozapine has been shown in vitro to be an antagonist at the GABA_A receptor, chronic administration of clozapine increases GABA turnover in the substantia nigra (Marco et al., 1978), and chronic clozapine and olanzapine treatment have been shown to downregulate GABA_A receptor density in the rat hippocampus and temporal cortex (Farnbach-Pralong et al., 1998). These results suggest that clozapine causes an overall increase in GABAergic activity. A role of neuroactive steroids in this apparent increase in GABAergic transmission cannot be excluded, as some atypical antipsychotic drugs increase levels of positively-modulating GABA_A receptor neuroactive steroids and decrease levels of negatively-modulating steroids (Marx et al., 2000; Barbaccia et al., 2001a; Marx et al., 2003; Nechmad et al., 2003a). Because GABAergic deficits have been widely demonstrated in the brains of individuals with schizophrenia, increasing GABAergic transmission may lead to the improvement of some symptoms. Consistent with this, clinical deterioration has been observed in patients switching from clozapine, an antipsychotic drug with relatively prominent GABAergic effects, to risperidone, an antipsychotic drug with less potent effects on GABAergic transmission (Verghese et al., 1996).

Other Neurotransmitter Systems Affected by Altered Neuroactive Steroid Concentrations

Deficits in glutamatergic activity have been reported in schizophrenia, and are thought to contribute to some aspects of the illness (Coyle et al., 2003). DHEA and DHEAS indirectly stimulate NMDA receptor activity, and DHEA has been reported to potentiate glutamate release both in vitro and in vivo (Lhullier et al., 2004). Administration of DHEA may produce beneficial effects by increasing glutamatergic transmission via increasing activity of the NMDA receptor. Indeed, potentiation of NMDA receptor function with glycine in individuals treated with antipsychotic drugs has been reported to lead to a significant improvement of negative symptoms, as well as positive and cognitive symptoms of schizophrenia (Heresco-Levy et al., 2004). These hypotheses are not consistent with the finding that clozapine decreased cortical DHEA and DHEAS concentrations. However, antipsychotic drugs also have direct effects on NMDA receptors, and therefore any effects on glutamatergic transmission by neuroactive steroids may be secondary to these modulatory actions. Alternatively, the reduction in DHEA and DHEAS by clozapine may exert therapeutic benefits independent of the glutamatergic system completely, or may not contribute to the therapeutic benefits of the drug at all.

Evidence also suggests that σ₁ receptors may be involved in the pathology of certain deficits in schizophrenia, particularly those involving dyregulated glutamatergic transmission (Hayashi and Su, 2004). Several selective σ₁ receptor antagonists such have shown antipsychotic potential (Frieboes et al., 1997; Huber et al., 1999), although results have varied (Volz and Stoll, 2004). While ALLO is devoid of σ₁ receptor activity, P is a σ₁ receptor antagonist, and DHEA, DHEAS, and PregS are σ₁ receptor agonists. Abnormal concentrations of these steroids in schizophrenia may contribute to some of the symptoms observed in the illness. Furthermore, elevated brain P concentrations and decreased DHEA and DHEAS concentrations induced by certain atypical antipsychotic drugs may result in reduced σ₁ receptor activity, which may contribute to the beneficial effects of antipsychotic drugs.

NEUROACTIVE STEROIDS AND DEPRESSION

There is extensive evidence suggesting that the GABAergic system is dysfunctional in individuals suffering from depression. Several clinical studies have shown that GABA concentrations in the CSF (Gold et al., 1980; Gerner and Hare, 1981; Kasa et al., 1982; Berrettini et al., 1983; Gerner et al., 1984) and plasma (Petty and Schlesser, 1981; Petty and Sherman, 1984; Petty et al., 1990, 1992) are decreased in depressed patients compared to healthy controls (but see Rode et al., 1991). Neuroimaging studies have found decreased GABA levels in the occipital cortex (Sanacora et al., 1999; Sanacora et al., 2004), and postmortem studies have shown decreased glutamic acid decarboxylase (GAD) activity (Fatemi et al., 2005) and increased GABA_A receptor density in the brains of depressed patients compared to controls (though not all studies have shown consistent findings (Guidotti et al., 2000)). Furthermore, successful treatment of depressive symptoms, either pharmacologically (Sanacora et al., 2002) or with ECT (Sanacora et al., 2003), has been reported to be accompanied by increased GABA levels in the occipital cortex. These results suggest that decreased GABAergic activity may play a role in the pathophysiology and/or expression of symptoms in depressed patients.

Neuroactive Steroids in Depression

Abnormal circulating levels of some pregnane steroids may be associated with depression. Although P levels were shown to be in the normal range (George et al., 1994; Romeo et al., 1998), individuals with major depression were shown to have lower CSF and plasma concentrations of ALLO (Romeo et al., 1998; Uzunova et al., 1998; Strohle et al., 1999; Strohle et al., 2000), and lower plasma levels of its isomer 3α,5β-THP (Romeo et al., 1998), compared to healthy controls. CSF levels of Preg have also been shown to be decreased in patients with depression compared to controls (George et al., 1994), though other studies have found contradictory results. Conversely, plasma concentrations of THDOC and its precursor DHDOC were shown to be increased in individuals suffering from depression relative to healthy controls (Strohle et al., 1999, 2000). 3β,5α-THP, an isomer of ALLO thought to negatively modulate the GABA_A receptor (Prince and Simmonds, 1992; Maitra and Reynolds, 1998), was also increased in the plasma of individuals suffering from depression relative to controls (Romeo et al., 1998). A role for pregnane steroids in depression is further substantiated by the finding that CSF ALLO concentrations were negatively correlated with illness severity (Uzunova et al., 1998). Patients with major depression did not have altered platelet PBR density compared to healthy controls (Weizman et al., 1995). Uzunova et al have recently written a comprehensive review on the relevance of 3α-reduced neuroactive steroids to depression.

Abnormal concentrations of DHEA and/or DHEAS may also be present in individuals with depression, though results are conflicting. It has been reported that DHEA and DHEAS concentrations in individuals suffering from depression are not different from healthy controls (Osran et al., 1993; Romeo et al., 1998; Young et al., 2002), whereas others have observed both elevated (Tollefson et al., 1990; Heuser et al., 1998; Takebayashi et al., 1998; Assies et al., 2004) and decreased (Scott et al., 1999; Michael et al., 2000; Schmidt et al., 2002; Poor et al., 2004, 2005) plasma, urinary, or salivary DHEA or DHEAS compared to healthy controls. It has been reported that psychotic depressed patients have elevated plasma DHEA and DHEAS concentrations compared to controls (Hansen et al., 1982; Maayan et al., 2000). Cortisol/DHEA and cortisol/DHEAS ratios have been reported to be higher in depressed patients than in controls (Scott et al., 1999; Young et al., 2002) at some, but not all timepoints over a 24-h period (Goodyer et al., 1998), indicating that the conflicting findings regarding DHEA and DHEAS concentrations and their ratios with cortisol (Takebayashi et al., 1998) may be related to the time of sample collection. As DHEA and DHEAS concentrations decline with age, conflicting findings may also result from differing ages of those studied. Unlike ALLO, no correlation between DHEA or DHEAS and illness severity has been observed (Scott et al., 1999; Maayan et al., 2000).

Animal studies also imply a role for neuroactive steroids in depression. Proestrus female rats (having high endogenous hippocampal ALLO concentrations) subjected to the forced swim test exhibited less depressive behaviour than diestrus female rats and male rats (having lower endogenous hippocampal ALLO), and this effect was abolished by the pharmacological blockade of ALLO synthesis (Frye and Walf, 2002). Administration of finasteride, an inhibitor of 5α-R, decreased plasma and hippocampal ALLO concentrations, and induced depressive-like behaviour in the forced swim test (Frye and Walf, 2004). Olfactory bulbectomy (OBX) has been shown to produce depressive-like symptoms in animals and is a widely-used animal model for depression. In OBX rats, brain ALLO concentrations are significantly decreased in the amygdala and frontal cortex, and significantly increased in the whole cortex compared to sham-operated controls (Uzunova et al., 2003), indicating that a region-specific dysregulation of ALLO synthesis and/or concentration may be a feature of depression.

Abnormal plasma and/or CSF concentrations of some steroids might reflect abnormal steroid concentrations in the brain, which may contribute to the pathology and/or symptoms of depression. Decreased concentrations of ALLO and its isomer 3α,5β-THP, both positive modulators of the GABA_A receptor, together with increased concentrations of 3β,5α-THP, DHEA, and DHEAS, negative modulators of the GABA_A receptor, may contribute to the downregulation of GABAergic activity observed in depression, and may be related to some of the symptoms characteristic of the illness.

Depression has been associated with dysfunctioning of the HPA axis; specifically hypercortisolemia, increased plasma CRH, AVP and ACTH concentrations, and nonresponse to the dexamethasone suppression test (Barden, 2004). The interaction between ALLO and the HPA axis has been discussed (see stress section), and indicates that decreased concentrations of this neuroactive steroid, perhaps via a decrease in GABAergic tone, may lead to disinhibition of the HPA axis, and thus contribute to hyperactivity of this system.

Antidepressant Potential of Neuroactive Steroids

Some neuroactive steroids have been shown to possess antidepressant properties. ALLO exhibits antidepressant-like effects in the forced swim test in both rats and mice, effects likely mediated by the GABAergic system (Khisti and Chopde, 2000; Khisti et al., 2000; Molina-Hernandez et al., 2004, 2005). Consistent with this, other positive allosteric modulators of the GABA_A receptor have been shown to exert antidepressant-like effects in animal models of depression and to have antidepressant actions in clinical trials (Lloyd et al., 1983; Weiss et al., 1986). DHEA, DHEAS and PregS also exhibit antidepressant-like effects in the forced swimming test (Prasad et al., 1997; Reddy et al., 1998; Urani et al., 2001; Maayan et al., 2005), effects possibly mediated through the σ₁ receptor (Reddy et al., 1998; Urani et al., 2001). It has been reported that σ₁ receptor ligands possess antidepressant properties in animal models of depression, and have been tested in clinical trials in individuals suffering from depression (Bermack and Debonnel, 2005). Administration of DHEA itself has also been reported to allieviate the symptoms of depression in a clinical population (Wolkowitz et al., 1997; Wolkowitz et al., 1999; Schmidt et al., 2005).

Clinical benefits of antidepressant drugs are not observed until several weeks after initiation of treatment. Given that some neuroactive steroids may be endogenous antidepressant compounds, it is possible that effective treatment of depression results, at least in part, from therapy-induced changes in the concentrations of these steroids. Indeed, antidepressant drugs alter neuroactive steroid concentrations. Acute administration of the selective serotonin reuptake inhibitor (SSRI) fluoxetine increased brain ALLO (Uzunov et al., 1996; Serra et al., 2001; Pinna et al., 2004) and decreased brain 5α-DHP concentrations in rats (Uzunov et al., 1996), and either failed to alter (Uzunov et al., 1996) or increased (Serra et al., 2001) brain Preg, P, and DHEA, and plasma ALLO, Preg, P, and THDOC. Acute administration of the SSRI paroxetine to rats elicited a small but significant increase in brain ALLO, whereas acute imipramine did not alter brain ALLO concentrations (Uzunov et al., 1996). A recent review further suggests an important role for some neuroactive steroids in the therapeutic effects of SSRIs, and proposes that many of the therapeutic effects of these drugs may in fact be the result of increased GABAergic transmission secondary to changes in neuroactive steroid levels (Pinna et al., 2006). In fact, these authors propose that the term “SSRIs” could be replaced by “selective brain steroidogenic stimulants” (Pinna et al., 2006).

Chronic antidepressant treatment in animals is a more clinically relevant paradigm to determine drug effects on neuroactive steroid concentrations, yet few chronic studies have been conducted. Paroxetine treatment for 21 days (but not 9 days) in rats resulted in an increase in brain ALLO synthesis 28 h after the last injection, an effect shown to be independent of peripheral steroidogenesis (Nechmad et al., 2003b). However, chronic fluoxetine treatment (15 days) in rats reduced both cortical and plasma Preg, P, ALLO, and THDOC 48 h after the last drug injection (Serra et al., 2001). These conflicting results may arise from methodological differences between the two studies, such as the time between the last drug injection and sample collection. In addition, comprehensive dose–response studies were not conducted. Further studies regarding chronic effects on brain and plamsa neuroactive steroid concentrations are warranted. In the OBX animal model for depression, chronic treatment with fluoxetine, desipramine, sertraline, or venlafaxine reversed the decrease in cortical ALLO observed following OBX in rats (Uzunova et al., 2004).

Antidepressant drug-induced changes in neuroactive steroid concentrations have been observed in the clinical population, providing further support a role for steroids in the pathology of depression. Successful pharmacological treatment of depression with fluoxetine (50 days) was associated with increased plasma ALLO and 3α,5β-THP levels and decreased plasma levels of the negative allosteric modulator 3β,5α-THP such that concentrations of these steroids were indistinguishable from healthy controls (Romeo et al., 1998; Strohle et al., 2000). Fluoxetine or fluvoxamine treatment resulted in an elevation of CSF ALLO that was positively correlated with improvement in mood (Uzunova et al., 1998). Treatment with tricyclic antidepressants (TCAs) achieved similar changes in plasma ALLO, 3α,5β-THP, and 3β,5α-THP levels (Romeo et al., 1998; Strohle et al., 1999). Fluoxetine treatment was also associated with decreased THDOC concentrations (Strohle et al., 2000), while treatment with TCAs had no effect on THDOC levels (Strohle et al., 1999). Plasma P, Preg, DHEA, 5α-DHP, and 5α-DHDOC were not altered after successful treatment with either SSRIs or TCAs (Romeo et al., 1998; Strohle et al., 1999). It has also been shown that treatment for 35 days with either amitriptyline or paroxetine resulted in a significant decline in plasma DHEAS concentrations, an effect that was more prominent in patients taking amitriptyline than paroxetine (Deuschle et al., 2004). Given that ALLO and 3α,5β-THP appear to be elevated in patients with depression, whereas THDOC and perhaps DHEA and DHEAS are decreased, it is possible that drug-induced regulation of these steroid concentrations contribute to the clinical effectiveness of these drugs.

The mechanism by which antidepressant drugs alter neuroactive steroid concentrations remains unclear. These drugs are thought to regulate HPA axis activity, and modulation of this system could result in altered neuroactive steroid concentrations. However, acute fluoxetine administration to rats elicited an increase in ALLO in ADX animals similar to sham-operated animals (Uzunov et al., 1996), suggesting a mechanism independent of the adrenal glands, and thus independent of HPA axis activation. Alternatively, antidepressant-induced changes in serotonin and noradrenaline systems may be related to changes in steroid metabolism, as it was shown in vitro that β-adrenergic (Morita et al., 2004) and 5-HT_2A (Morita et al., 2005) receptor stimulation induce gene expression of 5α-R in glial cells. Furthermore, subeffective doses of several serotonergic agents were shown to potentiate the antidepressant effects of ALLO in rats (Khisti and Chopde, 2000). However, fluoxetine was shown to increase brain ALLO levels independently of fluoxetine-induced changes in extracellular 5-HT (Guidotti and Costa, 1998), and the abilities of fluoxetine, paroxetine, and imipramine to increase cortical ALLO were not correlated with their abilities to inhibit the 5-HT transporter (Uzunov et al., 1996). Furthermore, fluoxetine-induced increases in ALLO were observed following pharmacological depletion of serotonin stores, indicating that in this case the mechanism of SSRI-induced increase in ALLO is independent of the drug’s ability to increase serotonin availability (Khisti and Chopde, 2000). These data suggest that the acute effects of antidepressant drugs on neuroactive steroids are not related to changes in serotonergic transmission; whether serotonin activity affects neuroactive steroid concentrations after chronic administration of antidepressant drugs via genomic mechanisms remains unknown.

The theory that antidepressant drugs alter the activity of steroidogenic enzymes appears to be plausible. Although this hypothesis was discounted on the grounds that fluoxetine induced uneven ALLO increases (and even ALLO decreases) in different brain regions (Uzunov et al., 1996), it should be noted that many isoforms of steroidogenic enzymes exist in humans, each with distinct enzymatic activities and specific locations within the brain. SSRIs alter the activities of specific isoforms of 3α-HSD; paroxetine, for example, altered the activity of the human type III enzyme more potently than the human type II (rat 3α-HSD) (Griffin and Mellon, 1999). Animal studies (Nechmad et al., 2003b) and ex vivo (Uzunov et al., 1996) and in vitro (Griffin and Mellon, 1999) experiments have also suggested that SSRI treatment shifts the activity of 3α-HSD such that the conversion of 5α-DHP to ALLO and 3α,5β-THP (and perhaps the conversion of DHDOC to THDOC) is enhanced. Fluoxetine may further inhibit the oxidation of ALLO to 5α-DHP (Uzunov et al., 1996; Guidotti and Costa, 1998). It is possible, then, that specific antidepressant drugs exert specific effects on the activities of specific isoforms of steroidogenic enzymes, resulting in region-specific changes in neuroactive steroid metabolism.

Unlike antidepressant drugs, it appears that nonpharmacological treatments for depression investigated to date do not alter plasma pregnane steroid concentrations. Clinical improvement observed following electroconvulsive therapy (ECT) (Baghai et al., 2005), repetitive transcranial magnetic stimulation (Padberg et al., 2002), or partial sleep deprivation (Schule et al., 2004) was not accompanied by any changes in plasma ALLO, 3α,5β-THP, 3β,5α-THP, or P concentrations. However, while DHEA/cortisol or DHEAS/cortisol ratios were not altered following ECT (Maayan et al., 2000), elevated plasma and urinary DHEAS concentrations following successful treatment of depressive symptoms with ECT have been reported (Ferguson et al., 1964; Maayan et al., 2000). Interestingly, it was noted that individuals that responded to ECT had significantly lower basal DHEAS concentrations than did responders, indicating that basal DHEAS concentrations may predict successful response to ECT in depressed individuals (Maayan et al., 2000). This hypothesis received further support in an animal study (Maayan et al., 2005). The implications of these results are not clear, although it was suggested that the opposing actions of DHEA/DHEAS and ECT on σ₁ receptor activity may be involved (Maayan et al., 2005), as DHEA and DHEAS are σ₁ receptor agonists, while ECT is thought to down-regulate σ₁ receptors (Nowak, 1996).

Other Neurotransmitter Systems Affected by Altered Neuroactive Steroid Concentrations

Deficits in noradrenergic activity have also been implicated in depression, and abnormal concentrations of neuroactive steroids may indirectly contribute to this. For example, GABA has been shown to regulate noradrenergic activity in the locus coeruleus (Aston-Jones et al., 2004; Koga et al., 2005) and median preoptic nucleus (Sakamaki et al., 2004). Furthermore, activation of the GABA_A receptor was shown to increase noradrenaline release in the rat cortex and hippocampus (Suzdak and Gianutsos, 1985). Decreased GABAergic tone in the brains of individuals with depression, resulting from abnormal concentrations of steroids modulating this neurotransmitter system, may result in decreased noradrenergic tone in the brain, contributing to the symptoms of depression. DHEA and DHEAS, via their actions at σ₁ receptors, were shown to modulate NMDA-evoked noradrenergic release (Monnet et al., 1995). Dysregulation of DHEA and DHEAS may therefore also lead to deficits in noradrenergic transmission.

NEUROACTIVE STEROIDS AND ANXIETY DISORDERS

The involvement of neuroactive steroids in anxiety disorders (i.e. generalized social phobia, generalized anxiety disorder, panic disorder, and posttraumatic stress disorder) has not been studied as extensively as in schizophrenia and depression.

Circulating levels of pregnane steroids in individuals suffering from anxiety disorders may be different from healthy controls, although results are conflicting and appear to depend upon the particular disorder. Elevated plasma ALLO has been reported in patients with panic disorder compared to matched controls (Strohle et al., 2002); however, other researchers did not observe significant differences in circulating ALLO between patients with generalized anxiety disorder, generalized social phobia or panic disorder and healthy volunteers (Semeniuk et al., 2001; Heydari and Le Melledo, 2002; Le Melledo and Baker, 2002; Brambilla et al., 2005). Preg and its sulfate ester may also be altered in anxiety disorders. In women suffering from anxiety-depressive disorder and treated with fluoxetine, it was reported that plasma PregS concentrations were significantly elevated compared to healthy volunteers in both phases of the menstrual cycle, while Preg was elevated in patients only in the follicular phase (Bicikova et al., 2000). Elevated concentrations of these steroids may result in increased neuronal excitability via interactions with the NMDA receptor, causing anxiety. However, in unmedicated patients with social phobia, Preg concentrations were shown to be similar to those of healthy controls (Laufer et al., 2005), and in patients with generalized anxiety disorder and generalized social phobia, but not with panic disorder, it was shown that plasma levels of PregS were lower than those of healthy controls (Semeniuk et al., 2001; Heydari and Le Melledo, 2002). Le Melledo and Baker (2002) have suggested that the decreased PregS concentrations in the latter disorders could be a homeostatic attempt to decrease anxiogenic activity by reducing the negative modulation of the GABA_A receptor and reducing the positive modulation of the NMDA receptor. Finally, in unmedicated patients with social phobia neither DHEA or DHEAS were altered compared to healthy controls (Spivak et al., 2000; Heydari and Le Melledo, 2002; Brambilla et al., 2003; Laufer et al., 2005). Individuals suffering from posttraumatic stress disorder have been reported to have higher plasma DHEA and DHEAS levels than healthy controls (Spivak et al., 2000).

A role for modulatory actions of neuroactive steroids on the GABA_A receptor in anxiety disorders is further substantiated by induction of panic in patients with panic disorder via administration of anxiogenic substances such as pentagastrin or cholecystokinin tetrapeptide (CCK-4). Individuals with panic disorder exhibited a significant decrease in the positive GABA_A receptor modulators ALLO and 3α,5β-THP in response to panicogenic sodium lactate/CCK-4 administration and an increase in the negative GABA_A receptor modulator 3β,5α-THP, whereas these levels were not altered in healthy volunteers (Strohle et al., 2003). Similarly, induction of panic with pentagastrin in males with panic disorder resulted in an increase in plasma DHEA and a trend for increased ALLO, while no alterations in steroid levels were observed following panic induction in healthy volunteers (Tait et al., 2002). Interestingly, panic induction with CCK-4 was shown to elevate plasma THDOC levels in healthy volunteers, accompanied by an increase in plasma ACTH and cortisol (Eser et al., 2005). Further studies are needed to delineate whether these changes in neuroactive steroid levels are due to panic or to the biochemical cascade associated with stress.

Anxiolytic Potential of Neuroactive Steroids

Animal studies have shown that positive modulation of the GABA_A receptor results in anxiolytic activity, whereas negative modulation of this receptor produces anxiogenic activity (Mayo et al., 1999). Consistent with this notion, steroids such as P, ALLO, and THDOC have been reported to produce anxiolytic effects in rodent behavioural tests including the elevated plus maze, the Geller–Seifter test, and the burying behaviour paradigm (Mellon and Griffin, 2002). Preg was shown to induce anxiogenic behaviour, and PregS exerted biphasic effects on anxiety-like behaviour (Melchior and Ritzmann, 1994b). Interestingly, DHEA was shown to possess anxiolytic properties in animal models of anxiety (Melchior and Ritzmann, 1994a), despite the fact that it negatively modulates the GABA_A receptor. Altered levels of these steroids may lead to the anxiety-like behaviour observed in individuals suffering from anxiety disorders, and regulation of these steroids may be therapeutically relevant to the treatment of these disorders.

Evidence also suggests that regulation of neuroactive steroid concentrations may contribute to the mechanism of action of anti-anxiety drugs. A recent report indicated that the anxiolytic drug etifoxine increases plasma and brain levels of P, ALLO, and Preg independently of the adrenal glands (Verleye et al., 2005). The drug-induced increase in ALLO was accompanied by an increase in PBRs, suggesting that etifoxine may increase neurosteroidogenesis via increased PBR density.

Together, these data from laboratory animal experiments and from clinical studies support a potential role for neuroactive steroids in the symptomatology and treatment of anxiety disorders. Results to date suggest that anxiety disorders have in common some dysregulations involving neuroactive steriods but that each disorder may display its own unique pattern of dysregulation.

PREMENSTRUAL SYNDROME AND PREMENSTRUAL DYSPHORIC DISORDER

Neuroactive steroid levels in disorders having both anxiety and mood components, such as premenstrual syndrome (PMS) and premenstrual dysphoric disorder (PMDD) have also been studied. These are somatopsychic conditions occurring in response to fluctuations in sex steroids associated with ovulatory menstrual cycles (Berga et al., 2005). GABAergic components of these disorders have been described (Halbreich et al., 1996; Sundstrom et al., 1997; Sundstrom et al., 1998). Decreased concentrations of ALLO in subjects with PMS or PMDD compared to matched controls have been reported (Wang et al., 1996; Rapkin et al., 1997; Bicikova et al., 1998; Monteleone et al., 2000; Lombardi et al., 2004) (but see Schmidt et al., 1994; Wang et al., 1996; Epperson et al., 2002), and it was suggested that low ALLO concentrations may contribute to behavioural symptoms such as anxiety, mood and sleep disorders and impaired attention observed in these disorders (Rapkin et al., 1997). Women suffering from PMDD were also shown to exhibit elevated P and Preg levels compared to healthy controls in the late luteal phase of the menstrual cycle (Epperson et al., 2002). SSRI antidepressants have been shown to be useful in rapidly treating the symptoms of PMS and PMDD, and it has been proposed that SSRI-induced increases in P metabolites may be involved in this therapeutic effect (Czlonkowska et al., 2003; Pearlstein, 2002).

CONCLUSIONS

Considerable preclinical and clinical evidence indicate that neuroactive steroids may play important roles in the etiology and treatment of a number of psychiatric disorders, including schizophrenia and mood and anxiety disorders. It is important in future studies to include measurements of as many neuroactive steroids as possible in individual studies since the modulatory effect of an increase in levels of one neuroactive steroid may be offset by increases in levels of another steroid with the opposite modulatory action or by decreases in levels of another steroid with the same modulatory action at the level of the GABA_A receptor (or at other receptors such as the NMDA receptor). Although many of the investigations on neuroactive steroids have been focused on their possible interactions with the GABA_A receptor, the actions of these steroids at other receptor complexes (e.g., NMDA, 5-HT₃, and σ) must also be considered, both in terms of how these systems in turn affect GABAergic transmission, and how they may play a direct role in altering mood and behaviour. Given the relationship between HPA axis function and psychiatric illness, further studies are warranted to determine the role of the action of neuroactive steroids at these receptors on HPA axis functioning, and vice versa. As more comprehensive information regarding the actions of these steroids becomes available, it will be of considerable interest to determine if, and how, neuroactive steroids may be common factors linking several of these systems. The studies reviewed here also highlight the importance of biochemical challenges (such as panic challenges), whereby compensatory changes in neuroactive steroid levels may lead to a better understanding of the role of these steroids in pathological behaviour. Furthermore, it is important to ascertain if successful drug or psychotherapeutic treatment is associated with normalization of neuroactive steroid dysfunction observed pretreatment. Given the apparent importance of neuroactive steroids in a number of functions in the body, it will also be of interest in future to determine what role these fascinating compounds may have in the well known strong association between physical health (e.g., cardiac function) and mental health.