Repurposing steroidogenesis inhibitors for the therapy of neuropsychiatric disorders: Promises and caveats

Abstract

Steroids exert a profound influence on behavioral reactivity, by modulating the functions of most neurotransmitters and shaping the impact of stress and sex-related variables on neural processes. This background - as well as the observation that most neuroactive steroids (including sex hormones, glucocorticoids and neurosteroids) are synthetized and metabolized by overlapping enzymatic machineries - points to steroidogenic pathways as a powerful source of targets for neuropsychiatric disorders.

Inhibitors of steroidogenic enzymes have been developed and approved for a broad range of genitourinary and endocrine dysfunctions, opening to new opportunities to repurpose these drugs for the treatment of mental problems. In line with this idea, preliminary clinical and preclinical results from our group have shown that inhibitors of key steroidogenic enzymes, such as 5α-reductase and 17,20 desmolase-lyase, may have therapeutic efficacy in specific behavioral disorders associated with dopaminergic hyperfunction. While the lack of specificity of these effects raises potential concerns about endocrine adverse events, these initial findings suggest that steroidogenesis modulators with greater brain specificity may hold significant potential for the development of alternative therapies for psychiatric problems.

1. Introduction

Over the past two decades, progress in neuroscientific research has led to significant improvements in our understanding of the genetic and pathophysiological basis of all major neuropsychiatric disorders. This exponential growth, however, has failed to translate into significant advances in the development of novel treatments (Caraci et al., 2017, Friedman, 2013, Hyman, 2012, Millan et al., 2015, O’Brien et al., 2014). A telling example of this gap is provided by the 2017 data on research and development of novel drugs: the number of new treatments for mental illnesses amounted to roughly 140 – a noticeable contrast with the 240 immuno-oncological therapies and 260 vaccines reported in the same year (PhRMA, 2017). This discrepancy is even more striking, given the high prevalence rate and socioeconomic burden of psychiatric illnesses across the world. Mental disorders are estimated to affect 18–25% of the population of Western countries and impose a global cost of $2.5 trillion (Insel, 2015, Kessler et al., 2009, Merikangas et al., 2010, NIMH, 2016).

A key limitation in the development of novel psychotropic medications lies in the challenge of translating results from animal models of psychiatric diseases to clinical trials. In turn, this barrier partially reflects the lack of precise biological criteria of reference and quantitative markers in the DSM-5 and other current diagnostic frameworks. In the attempt to overcome this deadlock and give new impetus to the design of new therapies, the NIH developed a new alternative function-based framework for mental disorders, the Research Domain Criteria (RDoC) (Cuthbert and Insel, 2013, Insel et al., 2010). Although the focus of RDoC on psychopathology has been generally regarded as highly meritorious, this initiative has also generated substantial pushback from several researchers and clinicians, who have questioned the pragmatic value of their classifications, as well as the applicability of these taxonomic systems to our current therapeutic setting (Weinberger et al., 2015).

Any serious attempt to reduce the distance between phenomenological and pathophysiological perspectives in psychiatry will require the implementation of translational strategies that may focus on novel neurochemical pathways. This type of experimentation can be realistically afforded only by testing the neuropsychiatric impact of various categories of drugs already approved for clinical use.

This strategy holds the potential of enriching our knowledge on the generalizability of preclinical findings to clinical results and enhance the predictive validity of our animal models. In addition, this approach may be critical to expand the current pharmacopoeia and yield potentially important economic advantages - a non-negligible fact, given that the average cost needed to bring a new drug to the market is currently estimated at $2.56 billion (DiMasi et al., 2016).

One of the most promising areas for drug repurposing in psychiatry comes from endocrine targets. The burgeoning field of psychoneuroendocrinology has extensively documented that brain functions are directly and indirectly modulated by interactions with peripheral glands, and the actions of hormones in the brain can have profound implications in psychopathology. For example, the pathogenesis and clinical course of most psychiatric disturbances are shaped by stress response, which is extensively contributed by glucocorticoids and other endogenous steroids. Accordingly, neuroactive steroids modulate the function and signaling of all major brain neurotransmitter systems, by interacting with multiple targets, including ion channels as well as membrane-bound and intra-cytoplasmic receptors (Reddy and Jian, 2010).

The potential of steroids as a class of therapeutic targets is highlighted by their fundamental role both in supporting stress response as well as sex differences in brain functions. Prompted by this premise, our group and others have begun studying the potential applications of steroidogenic inhibitors in neuropsychiatry. Blockers of steroidogenic enzymes have been extensively used to treat a variety of medical conditions that depend on endogenous steroid dysregulations, such as breast and prostate cancers, metabolic diseases, benign prostatic hyperplasia, and hair loss. However, the well-recognized overlap between neurobiological and endocrine functions, and the multifaceted role of steroids in shaping psychopathology points to the possibility that these drugs may help reveal novel treatments for mental conditions. In this article, we will briefly outline the theoretical assumptions by which steroidogenesis may be a rich source of novel therapeutic targets and review the emerging applications of this concept to psychiatry.

2. Steroidogenic pathways: enzymes and inhibitors

Steroidogenesis involves a series of sequential enzymatic reactions responsible for the synthesis and metabolism of cholesterol-derivative steroids, including bile acids, hormones (androgens, progestogens, estrogens, glucocorticoids and mineralo-corticoids) and neurosteroids (i.e., steroids directly synthesized in the brain). Brain functions are influenced by steroid hormones produced in the gonads and adrenal glands, as well as neurosteroids synthesized ex novo in neurons and glia (for details on the steroidogenic properties of the brain, see Celotti et al., 1997, Lephart and Husmann, 1993, Martini and Melcangi, 1991). The first step of steroidogenesis is the transfer of cholesterol (obtained by either de novo synthesis or catabolism of lipoproteins) to mitochondria, which is initiated by the steroidogenic acute regulatory protein (StAR). Cholesterol is then converted into pregnenolone via three consecutive monooxygenase reactions by the enzyme cytochrome P450 side chain cleavage (CYP11A1). In turn, pregnenolone has multiple metabolic fates, which differ depending on the specific makeup of enzymatic machinery available in the intracellular milieu. Despite the richness of steroid metabolites, these metabolic pathways are served by a relatively small number of families of enzymes, including 3β-hydroxysteroid dehydrogenase (3β-HSD), 5α-reductase (5αR), 3α-hydroxysteroid dehydrogenase (3α-HSD), 17,20 desmolase-lyase (CYP17A1), 17β-hydroxysteroid dehydrogenase (17β-HSD) (Fig. 1), and aromatase. In the following paragraphs, we will briefly outline the key characteristics of these enzymes, with emphasis on their CNS expression and function. Furthermore, we will discuss the available inhibitors for each of these targets and point to their potential neurobiological effects.

Fig. 1.

Simplified schematization of the pathways for the biosynthesis of neuroactive steroids from pregnenolone. For details about the role of each enzyme, see text.

2.1. 3β-Hydroxysteroid dehydrogenase/Δ^5–4 isomerase (3β-HSD)

The reaction catalyzed by 3β-HSD leads to the oxidation and isomerization of Δ⁵-3β-hydroxysteroids, such as pregnenolone and 17α-hydroxypregnenolone, into their Δ⁴-ketosteroid derivatives progesterone and 17α-hydroxyprogesterone (Mason et al., 1997, Simard et al., 1996), using NAD⁺ or NADP⁺ as electron acceptors. The two human 3β-HSD isozymes, encoded by HSD3B1 and HSD3B2 genes, differ by tissue distribution: type 1 is primarily found in placenta, mammary glands, small intestine and skin, while type 2 is abundant in adrenal glands and gonads (Rhéaume et al., 1991). In situ hybridization studies have revealed the presence of 3β-HSD (particularly type 1) transcript across several brain regions, such as olfactory tubercles, cerebral cortex, striatum, hippocampus, thalamus, hypothalamus and cerebellum. Notably, the expression of 3β-HSD is affected by stress, as well as other environmental factors such as dietary manipulations. As the conversion from pregnenolone to progesterone is necessary for the synthesis of all major steroids, the effects of 3β-HSD inhibition are wide-ranging and can lead to profound metabolic toxicity, due to the reduction of cortisol and aldosterone synthesis. The 3β-HSD inhibitors that could be considered for therapeutic use, trilostane and epostane, inhibit 3β-HSD1 with much higher affinity than 3β-HSD2 (Thomas et al., 2005, Thomas et al., 2002). The advantages of a relatively high selectivity ratio for type 1 are the predominant effect in the mammary gland and/or brain, and the lower degree of interference with the production of gluco - and mineralocorticoids in the adrenal glands. Trilostane has been particularly proposed as an alternative treatment for postmenopausal breast cancer resistant to first- and second-line anti-estrogenic therapy (Puddefoot et al., 2006). In general, however, the expected effects of 3β-HSD on the steroidogenic tree raise important concerns of toxic liability, also with respect to their potential applications in neuropsychiatry.

2.2. 5α-reductase (5αR)

The 5αR family includes five proteins, namely the three main types of 5α-reductase (5αR1, 5αR2, and 5αR3), glycoprotein synaptic 2 (GPSN2) and glycoprotein synaptic 2-like (GPSN2L) (Paba et al., 2011). These enzymes catalyze the saturation of the 4,5-double bond of the A ring of Δ^4–3 ketosteroid substrates, with the aid of cofactor NADPH. 5αRs convert testosterone into 5α-dihydrotestosterone (DHT), the most potent androgen, which is responsible for the acquisition of secondary sexual characteristics in males (Lauber and Lichtensteiger, 1996, Melcangi et al., 1998, Poletti et al., 1997, Torres and Ortega, 2003). The same enzymes also catalyze the conversion of progesterone to 5α-dihydroprogesterone (DHP), as well as the degradation of glucocorticoids (such as corticosterone, cortisol and deoxycorticosterone) into their 5α-reduced metabolites. DHT, DHP and other 5α-reduced steroids are further converted into key neurosteroids and androgens. In addition to these properties, 5αR3 also serves the conversion of polyprenol into dolichol, which is required for N-glycosylation of proteins (Cantagrel et al., 2010). The two best-characterized steroidogenic 5αR isoenzymes, 5αR1 and 5αR2, diverge by biochemical and molecular properties, substrate specificity and anatomical distribution. A thorough overview of the differences between these enzymes is beyond the scope of this article (the interested reader is referred to Paba et al., 2011). It should be noted, however, that the affinity of 5αR1 for testosterone and progesterone is much lower than that of 5αR2, possibly indicating separate roles of these two enzymes in relation to different levels of circulating precursors.

5αR1 is abundantly expressed in the skin, sebaceous and sweat glands, neurons of central and peripheral nervous systems, and adrenal glands. In the rodent brain, 5αR1 has been detected in the olfactory bulb, pyramidal cells of the layers 2,3 and 5, the pyramidal and granule cells in CA1, CA3 and dentate gyrus of the hippocampus, the output neurons of basolateral amygdala, the medium spiny neurons in the striatum, the dorsomedial and reticular nuclei neurons in the thalamus, the Purkinje cells of the cerebellar cortex. In addition, 5αR1 is typically localized in the cytosol of type I astrocytes and oligodendrocytes in most brain regions (Kiyokage et al., 2005, Pelletier et al., 1994, Tsuruo et al., 1996).

5αR2 is mainly expressed in tissues and organs of the male urogenital tract (including prostate, epididymis, testicles and seminal vesicles), as well as genital skin, hair follicles and liver. 5αR2 has also been detected in several brain regions of adult rodents and humans, albeit at lower levels than 5αR1. We showed that 5αR2 is widely distributed across most key regions of the adult rat brain, including cortex, olfactory bulb, hippocampus and cerebellum (Castelli et al., 2013). The cell distribution of 5αR2 is strikingly divergent from that of 5αR1, as it is only expressed in neurons but absent in glial cells.

The two 5αR inhibitors approved for human use, finasteride and dutasteride, are 4-azasteroids that block the enzyme via the formation of covalent adducts with NADH, which bind to the enzyme in a virtually irreversible fashion, with a very slow turnover (T_1/2 ≈ 30 days) (Paba et al., 2011). Finasteride is a potent competitive inhibitor of 5αR2 (IC₅₀ = 0.16 nM) but has limited affinity for 5αR1 (IC₅₀ = 81 nM). In men, treatment with finasteride (5 mg/day) reduces DHT serum levels by 71% (Drake et al., 1999). Conversely, dutasteride is more potent than finasteride and inhibits both isoforms with comparable efficacy (Clark et al., 2004). The two drugs also differ with respect to their pharmacokinetic properties: finasteride has a half-life of approximately six hours in men aged 18–60, which increases with age up to approximately eight hours in men; conversely, dutasteride has approximately 260 h in men aged 50–69 years and approximately 300 h in men aged over 70 (Gisleskog et al., 1999).

The clinical applications of finasteride and dutasteride rely on their ability to reduce DHT. This steroid plays a pivotal role in the development of prostatic tissue, as well as in the regulation of hair growth in the scalp. Excess DHT leads to the development and/or progression of benign prostatic hyperplasia (BPH) and androgenic alopecia. For these reasons, 5αR inhibitors have been approved for these conditions (dutasteride is not approved for alopecia due to its extremely long half-life).

Initial trials also explored the possibility that 5αR inhibitors may be used as a chemopreventive agent for prostate cancer, but, while finasteride led to a decrease in prostate cancer by 24.8%, this compound was also associated with a significant increase in risk for high-grade prostate cancer (Etzioni et al., 2005).

In addition, novel selective 5αR1 inhibitors have been developed, including the steroid MK-386, the non-steroid FK-143 and the noncompetitive inhibitor bexlosteride (LY-191704). MK-386 selectively reduces DHT levels in sebum without affecting its concentrations in semen (Schwartz et al., 1997). Unlike the competitive inhibition of finasteride, FK-143 showed robust inhibition of both human and rat 5αR1 in a noncompetitive fashion (Hirosumi et al., 1995). Bexlosteride was developed as a treatment for prostate cancer (Hirsch et al., 1993, Neubauer et al., 1996); in Phase I and II clinical trials, the substance was found to be safe and well-tolerated. However, Phase III clinical trials showed no efficacy for prostate cancer, and testing of the drug was discontinued (ADIS, 2000). Another 5αR inhibitor that has recently emerged as a popular treatment for prostatic hyperplasia and genitourinary disorders is the extract of saw palmetto (Serenoa repens) berries (Di Silverio et al., 1998, Steiner et al., 1994, Sultan et al., 1984, Vacherot et al., 2000; for a review see Suzuki et al., 2009). Among the derivatives of this extract, SPET-085 appears to be the most promising 5αR2 inhibitor, with greater bioactivity and efficacy on the prostate (Pais, 2010).

Aside from the reduction of DHT synthesis, another well-documented effect of these drugs in animal models is the suppression of neurosteroids, including allopregnanolone and tetrahydrodeoxycorticosterone (THDOC). These steroids play a key role in the ontogeny of stress response. While clinical research on these compounds is still rudimentary, rich evidence show that, by modulating GABA-A receptor functions, these steroids are extremely important in stress coping, as they facilitate resiliency and promote the suppression of anxiety that derives from acute stress (Gunn et al., 2015).

2.3. 3α-HSD

3α-HSDs are aldo-keto reductases, a group of enzymes that convert 3-ketosteroids into their corresponding 3α-hydroxy metabolites (using NADPH as cofactor). For example, 3α-HSD converts DHT into the less potent androgen 5α-androstane-3α,17β-diol (3α-diol); furthermore, the same enzyme transforms DHP and deoxycorticosterone into the neurosteroids allopregnanolone and THDOC, neurosteroids exhibiting anxiolytic and anticonvulsant activities through positive modulation of GABA-A receptors (Majewska, 1992, Melcangi et al., 2005, Patte-Mensah et al., 2005, Purdy et al., 1991).

3α-HSDs are important in the formation of bile acids and in the metabolism of several xenobiotics. Four human isoforms have been isolated to date: type 1 (AKR1C4), type 3 (AKR1C2), type 5 (AKR1C3) and 20α,3α-HSD (AKR1C1) (Dufort et al., 1996, Khanna et al., 1995, Lin et al., 1997). Each isoenzyme has a characteristic tissue distribution and repertoire of catalytic activity and substrate selectivity:

• •3α-HSD type 1 is abundant in the liver and plays a key role in the detoxification of xenobiotics containing a keto-group (Barski et al., 2008, Penning et al., 2004, Penning et al., 2000). This process leads to the formation of a hydroxy derivative, which can be then conjugated to a sulfo- or glucuronosyl-group to promote elimination.

• •3α-HSD type 3 is expressed in several steroidogenic organs and tissues, including the prostate, skin, adrenal glands and brain (Penning et al., 2004, Penning et al., 2000). It is also involved in the metabolism of xenobiotics and bile synthesis; however, unlike type 1, it is involved in hormonal regulation within endocrine organs and serve allopregnanolone synthesis in the brain as well as steroid hormones in the adrenal glands (Griffin and Mellon, 1999).

• •Type 5 serves a fundamental role in the production of adrenal androgens during adrenarche, converting androstenedione into testosterone (as well as DHT into 3α-diol, albeit with lower efficiency) (Barski et al., 2008). Dysregulations of this enzyme are linked to breast carcinoma, endometrial hyperplasia, endometrial carcinoma, and prostate carcinoma (Ji et al., 2003).

• •20α, 3α HSD (AKR1C1) catalyzes the 20α-reduction of pregnanes, in fact, type 3 3α-HSD inactivates DHT, the most potent androgen, whereas 20α-HSD inactivates progesterone, a female hormone. AKR1C1 together with type 3 is the most abundant in the human brain and protects androgen-producing cells against overstimulation from female hormones (Penning et al., 2000).

In general, it is still unclear whether 3α-HSD inhibitors may have clinical potential for neuropsychiatric disorders. Given that the reaction mediated by this enzyme is bidirectional, it is difficult to predict the outcome of its reaction. In addition, all available 3α-HSD inhibitors, such as the non-steroidal anti-inflammatory drug (NSAID) indomethacin, are non-specific (Barski et al., 2008, Penning and Talalay, 1983). However, it is worth mentioning that the stimulation of 3α-HSD by some antidepressants, such as fluoxetine, fluvoxamine and sertraline, has been deemed extremely important for their antidepressant properties (Griffin and Mellon, 1999, Pinna, 2010).

2.4. CYP17A1

CYP17A1 (steroid 17α-hydroxylase/17,20 lyase) is a member of the CYP450 family, mainly localized in the endoplasmic reticulum. This enzyme has two distinct activity: a 17α-hydroxylase capacity, which is required for the synthesis of glucocorticoids; and a 17,20-lyase activity, which, in coordination with 17α-hydroxylase, serves the synthesis of androgens and the conversion of pregnenolone into dehydroepiandrosterone (DHEA) (Nakajin et al., 1981; for a review see Akhtar et al., 2005). Indeed, the expression of CYP17A1 in the mammalian brain was first theorized based on the detection of high DHEA levels in the CNS. Further studies showed that CYP17A1 is expressed both in oligodendrocytes, astrocytes as well as in neurons. The main inhibitors of CYP17A1, abiraterone and ketoconazole, have clinical efficacy against prostate cancer, insofar as they reduce the synthesis of androgens and, indirectly, the activation of androgen receptors (Pont et al., 1982, Vasaitis et al., 2011). Abiraterone selectively blocks the synthesis of androgens in major steroidogenic organs, including testis, adrenal glands, and prostate, without inducing adrenal insufficiency. It is clinically used as a second-line therapy for metastatic castration-resistant prostate cancer (mCRPC) (Udhane et al., 2016; for a review see Gartrell and Saad, 2015). Ketoconazole, an imidazole antifungal agent, has lower potency and specificity on CYP17A1, as inhibits several additional CYP450 enzymes, including CYP11A1, CYP11B1, and CYP11B2 (Daniel and Newell-Price, 2015, Feelders and Hofland, 2013), triggering a cascade of distinct endocrinological changes. The reduction in cortisol activates a negative feedback loop, stimulating adrenocorticotropic hormone (ACTH) production. There is potential for mineralocorticoid excess, which is a possible dose-limiting toxicity of abiraterone acetate in some patients; however, the incidence of mineralocorticoid excess is generally abrogated by concomitant administration of prednisone (or dexamethasone) or aldosterone antagonists, such as eplerenone or spironolactone. The most common symptoms of this condition observed with abiraterone acetate administration were hypokalemia, hypertension, and edema, reported in patients in the phase III studies.

2.5. 17β-HSD

The 17β-HSD family includes 14 enzymes that catalyze the stereospecific oxidation/reduction of androgens and estrogens at carbon 17β. Except for type 5, which is an aldo-keto reductase, all enzymes belong to the short-chain dehydrogenase/reductase family. Each specific variant has a predominantly reductive or oxidative activity, which plays a critical role in the regulation of the synthesis of androgens and estrogens. Specifically, 17β-reduction converts androstenedione and estrone to testosterone and estradiol, respectively; 17β-oxidation serves the opposite reaction, and can protect cells from sex hormonal overstimulation. From this perspective, it can be understood that alterations in the regulation or expression of 17β-HSD enzymes have been implicated in hormone-dependent cancers, as well as metabolic and sexual disorders (Moghrabi and Andersson, 1998). Selective inhibition of oxidative and reductive 17β-HSD can be an attractive mechanism to alter local sex hormone distributions.

All 17β-HSD isoforms are present in the human temporal lobe of children and adults (Stoffel-Wagner et al., 1999). The importance of 17β-HSD in the maintenance of physiological levels of androgens and estrogens is supported by its wide distribution in the human tissues (Martel et al., 1994, Martel et al., 1992). Thus, this enzyme contributes to the fine regulation of sex hormones in the brain. Although active efforts are in place to develop selective 17β-HSD inhibitors, none of them are currently under clinical investigation.

2.6. Aromatase

Aromatase is a member of the cytochrome P450 superfamily, which catalyzes the rate-limiting step for the synthesis of estrogens from androgen precursors. It consists of a specific glycoprotein and a membrane-bound, heme-containing NADPH-cytochrome P450 reductase with a steroid-binding pocket (for a review see Simpson et al., 2002). The catalytic reaction includes three subsequent hydroxylations of their 19-methyl group of androgens, followed by demethylation and aromatization of the A ring.

Aromatase is localized in the endoplasmic reticulum of most tissues, where it is subjected to tight regulation from hormones, glutamate and other mediators. In the CNS, aromatase has been detected in the hypothalamus and limbic system (Naftolin et al., 1972, Naftolin et al., 1971a, Naftolin et al., 1971b). In the brain, aromatase is primarily involved in the aromatization of circulating or locally produced testosterone. This essential process is involved in several physiological and pathological processes, including the early masculinization of the brain (Sasano et al., 1998), neurogenesis and neurite outgrowth (Kretz et al., 2004, Rune et al., 2006, von Schassen et al., 2006), neural plasticity and neuroprotection (Azcoitia et al., 2001, Bender et al., 2017, Diotel et al., 2013, Saldanha et al., 2009). Furthermore, aromatase has been shown to modulate emotional states and cognitive functions (Fink et al., 1999).

While aromatase is primarily expressed in neurons, it can be reactively expressed in astrocytes (Yague et al., 2006) in response to pathological conditions, such as anoxia and ischemia (Zhong et al., 2017), to promote the synthesis of neuroprotective estrogens (Duncan and Saldanha, 2011, Garcia-Segura et al., 1999).

Three generations of aromatase inhibitors have been developed, with subsequent increase in specificity for the aromatase enzyme (Carlini et al., 2005, Miller, 1989). First-generation (aminoglutethimide) and second-generation (fadrozole, formestane, and vorozole) aromatase inhibitors show low selectivity and decreased aldosterone and cortisol production in addition to aromatase (for a review see Wiseman and Goa, 1996). Side effects included lethargy, vertigo, rash and mood disturbance. Third-generation inhibitors approved for clinical use in the US include the reversible, nonsteroidal agents anastrozole and letrozole, and the irreversible steroidal inhibitor exemestane. These drugs have higher efficacy than earlier blockers and are the preferred first-line treatment for metastatic hormone receptor-positive cancers and for adjuvant use in postmenopausal women before or after surgery for estrogen-related positive breast cancer (Buzdar et al., 2002; for a review see Buzdar and Howell, 2001).

The efficacy of aromatase inhibitors in breast cancer is based on their ability to block the synthesis of estrogens, which promote the growth and survival of breast epithelial cells by binding and activating estrogen receptors (ER) α. Although aromatase blockers are well-tolerated, their applicability in neuropsychiatric therapy appears limited, given that the reduction of estrogens is linked to cognitive impairment and greater vulnerability to excitotoxicity (Azcoitia et al., 2001, Garcia-Segura et al., 2003) and risk of neurodegeneration. For example, aromatase inhibition potentiates the nigrostriatal damage after exposure to a dopaminergic neurotoxin in models of Parkinson’s disease (Morale et al., 2008). In addition, lack of aromatase in mice has been associated with greater aggression, as well as depressive-like and perseverative behaviors (Hill et al., 2007, Trainor et al., 2006).

3. Neuropsychiatric effects of 5αR inhibitors: therapeutic potential and caveats

As mentioned above, steroids serve a key role in influencing the impact of environmental variables – and particularly stress - on behavior. Thus, it is not surprising that these compounds have extensive interactions with most neurotransmitter systems. Our group and others have particularly focused on the interactions between steroids and dopamine, a central neurotransmitter in stress coping and resiliency (Cabib and Puglisi-Allegra, 2012, Isingrini et al., 2016). Extensive preclinical evidence has shown that this class of “non-canonical” fast-acting steroids exerts a pleiomorphic influence on dopamine signaling (Di Paolo, 1994, Sánchez et al., 2010). For example, progesterone, testosterone, and 17β-estradiol have all been shown to modify both dopamine release and turnover across various brain regions in male and female rodents (Sánchez et al., 2010). On the other hand, allopregnanolone dose-dependently modulates dopamine efflux in rat brain, in which at low doses enhanced extracellular dopamine levels in nucleus accumbens (Rougé-Pont et al., 2002) while at high doses reducing extracellular dopamine levels in medial prefrontal cortex and nucleus accumbens (Dazzi et al., 2002, Motzo et al., 1996). Conversely, DHEA and its sulfated ester (DHEAS) increase dopamine release from cell cultures (Charalampopoulos et al., 2005, Murray and Gillies, 1997); furthermore, DHEA has been shown to reduce dopamine D2 receptor activation (Catalina et al., 2002, Suárez et al., 2005), while DHEAS increases D1 receptor-dependent PKA activity (Dong et al., 2007).

Converging evidence has shown that excessive activation of the dopaminergic system results in a broad range of neuropsychiatric problems, ranging from mania to decision-making problems and addiction, as well as Tourette syndrome and dyskinetic manifestations. Conditions linked to this hyperdopaminergic state are often problematic to resolve, given that dopaminergic blockers can cause extrapyramidal symptoms and cognitive dulling. As shown above, targeting neuroactive steroids may be an effective avenue to modulate dopaminergic responses and associated mental illnesses.

3.1. Therapeutic potential of 5αR inhibitors in neuropsychiatric disorders

To study the impact of neuroactive steroids on dopaminergic neurotransmission, our group has primarily focused on the effects of 5αR inhibitors. The focus on 5αR was primarily motivated by the importance of this enzyme in catalyzing the rate-limiting step of neurosteroidogenesis; furthermore, as outlined in the previous section, 5αR blockers have additional advantages, including the lack of overt toxicity. In line with this hypothesis, our group began investigating the role of this enzyme in the behavioral regulation mediated by dopaminergic agonists.

Our first finding was that high doses of finasteride and dutasteride prevented the hyperlocomotion and stereotyped behavior (two well-validated assays to measure behavioral outcomes of dopaminergic activation) induced by dopaminergic agonists, including d-amphetamine, a dopamine release enhancer, and apomorphine, a potent, non-specific dopamine receptor agonist (Bortolato et al., 2008). Our emphasis was particularly placed on information processing, and specifically on prepulse inhibition (PPI) of the acoustic startle reflex. PPI is an operational index of sensorimotor gating, a process aimed at filtering out irrelevant or redundant information in the brain, which is reduced by overactivity mesolimbic dopaminergic activity (Swerdlow et al., 1992, Swerdlow et al., 1990).

Sensorimotor gating is disrupted in schizophrenia and other disorders (Braff et al., 2001), such as mania (Perry et al., 2001), Tourette syndrome (Castellanos et al., 1996, Swerdlow et al., 2006), and obsessive-compulsive disorder (Hoenig et al., 2005). The unique advantage of PPI lies in the possibility of testing similar testing protocols in psychiatric patients and animal models. The high face and predictive validity of PPI as a tool for the assessment of neurobehavioral deficits related to schizophrenia is highlighted by the fact that, in animals, administration of dopaminergic agonists significantly reduces this index, in a fashion sensitive to both typical and atypical antipsychotic drugs (Geyer et al., 2001). Unlike these drugs, however, finasteride failed to induce extrapyramidal effects (Bortolato et al., 2008), highlighting its interesting potential as an alternative to other antipsychotic therapies.

In further studies, we endeavored to qualify the neural substrates of finasteride-mediated effects. Importantly, we found that the effects of this drug were equivalent in males and females (Bortolato et al., 2013), and, more importantly, were not modified by orchiectomy in adult males, strongly suggesting that the antidopaminergic effects of 5αR inhibitors are not mediated by changes in circulating male sex hormones (Devoto et al., 2012). We then determined that the systemic effects of finasteride were replicated by i.c.v. administration of this drug, as well as by local infusions in the nucleus accumbens, and, to a lesser extent, medial prefrontal cortex (Devoto et al., 2012). The effects of finasteride, however, were not associated to variations in dopamine efflux in these regions (Devoto et al., 2012), suggesting that the antidopaminergic effects of finasteride are due to postsynaptic receptors. GABAergic medium spiny neurons of the nucleus accumbens and the glutamatergic pyramidal cells of the medial prefrontal cortex express D1-type dopamine receptors and contain high levels of 5αR (Castelli et al., 2013). In this cell, 5αR substrates and products, such as progesterone, pregnenolone and DHEA are known to modulate several receptors (NMDA, σ1 receptors) that, in turn, may interact with D1 receptors and modify the downstream cascade of dopamine receptors via direct or indirect mechanisms (Hernandez et al., 2005, Navarro et al., 2010, Pei et al., 2004). The effect of finasteride on the dopaminergic modulation of sensorimotor gating appears to be region specific, as the same dose of finasteride injected in other forebrain regions implicated in gating regulation, such as ventral hippocampus, basolateral amygdala and dorsal caudate, failed to revert PPI disruption induced by systemic apomorphine (Devoto et al., 2012).

We then investigated the antidopaminergic effects of finasteride on the PPI deficits produced by the selective D₁, D₂, and D₃ dopamine receptor agonists. In two subsequent studies conducted in rats and mice with different susceptibilities to the PPI-disruptive effects of selective dopamine compounds, we demonstrated that the antipsychotic-like properties of finasteride are mediated by the modulation of D₁ and D₃, but not D₂, receptors (Frau et al., 2016, Frau et al., 2013). These results may account for the lack of extrapyramidal complications in response to 5αR inhibitors (Bortolato et al., 2008, Paba et al., 2011).

The effects of finasteride and dutasteride in dopaminergic regulation likely reflect a complex set of alterations in steroid profile. For example, our studies have recently shown that the deficits in PPI induced by sleep deprivation are contributed by imbalances between AP and progesterone (Frau et al., 2017); on the other hand, studies on the effects of stressful manipulations, such as isolation rearing, on the same paradigms, point to an implications of neurosteroid precursors, such as pregnenolone (Frau et al., 2015), in the ameliorative effects of 5αR inhibitors. Finally, we ascertained that both systemic and intracerebral injections of the CYP17A1 inhibitor abiraterone, elicit antidopaminergic effects in PPI akin to those observed after 5αR inhibitors (Frau et al., 2014). Unlike finasteride, abiraterone primarily reduces the synthesis of androgenic neurosteroids, but not AP. Our analyses indicate that, although the reduction in AP and other 3α,5α-reduced neurosteroids participates to the general mechanism of action of finasteride, additional mechanisms are certainly involved to account for the antidopaminergic properties of this drug.

Our first attempt to translate these preclinical results into clinical leads was focused on Tourette syndrome, a neuropsychiatric disorder characterized by the presence of physical and vocal tics, as well as sensorimotor gating deficits, which are contributed by overactive dopaminergic neurotransmission in the striatum (Denys et al., 2013, Sutherland et al., 2011, Swerdlow and Sutherland, 2006). Of note, while the traditional vistas on this syndrome always focused on D2 receptors, recent clinical research indicates that D1 receptors plays a key role in tic ontogeny, given that its specific antagonist ecopipam significantly reduces the severity of tics in Tourette syndrome patients (Gilbert et al., 2014).

The first patient who gave informed consent for experimental treatment with finasteride as an adjunctive therapy was a severe case of Tourette syndrome with explosive vocalizations, stereotyped coprolalic utterances, ritual behaviors, self-injuring motor tics and excessive sex drive (Bortolato et al., 2007). Previous therapeutic attempts with typical antipsychotics had resulted in transient improvements, but the high rate of extrapyramidal and cognitive side effects had led him to repeated withdrawals. Finasteride (5 mg/day) led to a gradual improvement of his motor and vocal tics, as assessed by the Yale Tic Severity Scale with no reported side effects. The discontinuation of the regimen after 18 weeks, however, resulted in an abrupt, dramatic exacerbation of the symptoms, which was countered by reinstatement of the 5αR inhibitor. Over the last 2 years of follow-up treatment, the patient has shown a stable improvement of his tics, with a marked reduction in severity of vocal and motor tics (about 30% and 50% of his pre-treatment scores, respectively). The treatment has not led to any overt sign of toxicity or side effects; while a reduction in sex drive was reported, this effect was described by the patient as “beneficial”, in consideration of his excessive libido before initiation of the therapy. The therapeutic effects of finasteride as an adjunctive treatment in Tourette syndrome have been confirmed in a first open-label study with treatment-refractory adult male patients (Muroni et al., 2011). Again, finasteride was found to elicit significant reduction of the severity of tics and associated compulsive (but not obsessive) manifestations within six weeks from initiation of the therapy (Muroni et al., 2011).

To understand the mechanisms of these findings, we studied one of the best-characterized animal models of Tourette syndrome, the D1CT-7 mouse. This transgenic mutant, generated by the attachment of the D1 dopamine receptor promoter to the gene encoding the neuro-potentiating cholera toxin, manifests sporadic “tic-like” clonic axial jerks, which can be reduced by standard anti-tic medications (Campbell et al., 1999, Nordstrom and Burton, 2002). In line with the clinical phenomenology of tics, these manifestations are increased by stress and accompanied by PPI deficits (Godar et al., 2016). Recent studies from our group identified that the stress-induced exacerbation of tics in this model is due to the synthesis of AP (Mosher et al., 2017). Accordingly, finasteride extinguished the effects of stress on tic-like responses and PPI reductions, while AP administration potentiated them (Mosher et al., 2017).

Several epidemiological studies have demonstrated that Tourette syndrome is commonly associated with other psychiatric and emotional co-morbidities, such as attention-deficit hyperactivity disorder (ADHD), obsessive–compulsive disorder (OCD), poor impulse-control, and aggressive/antisocial behaviors (Kumar et al., 2016). Consistently, Tourette syndrome patients have significant difficulties in controlling impulsivity, which generally increases with the presence of other comorbidities, such as OCD (Buzan et al., 2000). Although it is uncommon that children with Tourette syndrome meet full criteria for an OCD diagnosis, a wide array of OCD-like behaviors may overlap or parallel with those of impulse-control disorders (Palumbo and Kurlan, 2007). Androgens (including testosterone and DHT) have been shown to modulate impulse control and play a critical role in the pathogenesis of compulsiveness (Potenza et al., 2009). In addition, both OCD and ICDs are disorders markedly linked to high dopaminergic states (Weintraub, 2008), OCD symptoms are markedly exacerbated by stress and associated with sensorimotor gating dysfunctions (Morgado et al., 2013, Stojanov et al., 2003).

Therefore, apart from the tics, since most of the patients with Tourette syndrome exhibit these neuropsychiatric comorbidities and these conditions seem to be particularly susceptible to sex hormones and stress, we also studied the effects of finasteride on a well-validated murine model of OCD, i.e. the marble-burying test in mice. In this test, mice are placed in a standard cage filled with wood-chip bedding in order to cover 10 marbles. After thirty minutes, the amount of marbles buried (at least 2/3 of the marble are considered covered) is measured. The procedure takes advantage of the proclivity of spontaneous behavior of mice in digging in natural settings (e.g. burrows, escape tunnels) and the repetitive, unwanted behaviors in presence of marbles, provides a subtype of compulsive behavior frequently encountered in OCD patients. As shown in Fig. 2, finasteride elicited a significant attenuation in digging and marble-burying compared to vehicle-treated controls. Furthermore, the effective dose of finasteride in reducing marble-burying behaviors did not induce significant changes in locomotor activity.

Fig. 2.

Effects of finasteride (6,25-12,5–25 mg/kg, IP) on marble burying (left) and motility (right) in adult CD1 mice. VEH, vehicle of finasteride; FIN, finasteride. Values are expressed as mean ± S.E.M. *p < 0.05; **p < 0.01 vs VEH treated mice.

3.2. Neuropsychiatric side effects of 5αR inhibitors

Although the results of our preclinical and clinical studies hold promise for potential applications of 5αR inhibitors in psychiatry, recent studies have shown that these agents may lead, in a subset of susceptible individuals, to serious adverse neuropsychological effects (Traish et al., 2017). Finasteride or dutasteride treatment have been associated to erectile and ejaculatory dysfunctions, with loss of libido, depression with suicidal thoughts, anxiety, irritability and sleep problems (Ali et al., 2015, Altomare and Capella, 2002, Corona et al., 2012, Ganzer et al., 2015, Gur et al., 2013, Irwig, 2012, Traish et al., 2017). Of note, these side effects can persist after the discontinuation of the drug (Irwig, 2012), warranting the name of “post-finasteride syndrome” (PFS) to these side effects. Following the report of these side effects, regulatory agencies in Sweden, UK and the USA issued warnings against the risk of persistent adverse effects following finasteride use. Following initial anectodal reports of PFS, some studies have been devoted to these iatrogenic manifestations (Basaria et al., 2016, Melcangi et al., 2017). These studies have confirmed impaired sexual function and higher depression scores in individuals affected by PFS (Basaria et al., 2016, Melcangi et al., 2017). Importantly, these impairments were associated with functional abnormalities of the nucleus accumbens and prefrontal cortex, pointing to the possibility that the same antidopaminergic properties identified in our studies may also account for these side effects. This idea is in line with the well-established role of dopamine in motivational behaviors and the pathogenesis of anhedonia. On the other hand, the high variability of symptoms in PFS may suggest that multiple pathogenic mechanisms may participate in these complications. In this perspective, it is interesting to point out that, unlike finasteride, the CYP17A1 inhibitor abiraterone has not been associated with overt psychological symptoms or depression (Salem et al., 2017).

4. Conclusions

Abundant literature has shown that neurosteroids are very important in the enactment of stress response, as they enable the activation of neuroprotective and anxiolytic mechanisms to promote coping and resiliency. From this perspective, the idea of suppressing the synthesis of neurosteroids in the context of psychiatric disorders may seem counterintuitive at first. Nevertheless, it should be emphasized that the pathophysiology of several mental disorders is fueled by maladaptive mechanisms of coping, in which certain pathological behaviors are enacted to relieve the stressfulness of urges. For example, the execution of compulsive and impulsive behaviors (including problematic substance use and pathological gambling) is generally initiated as a response to a pervasive sense of anticipation and leads to a transient reduction of the intensity of these urges. It is certainly possible that, in pathological circumstances such as these, neurosteroids may promote stress coping by enabling these maladaptive behaviors.

In this framework, the link between neurosteroids and dopamine may be highly critical, given the importance of dopamine in motivational behaviors. At the same time, this idea may help explain our findings on the antidopaminergic effects of 5αR inhibitors in conditions characterized by “hyperdopaminergic states” and poor impulse control. Most of the conditions that may respond to the therapeutic effects of 5αR inhibitors, including Tourette syndrome, are indeed characterized by the presence of urges. By reducing the relieving effects of impulsive and compulsive actions, 5αR inhibitors may be helpful in countering the mechanisms that support maladaptive habits and may synergize with psychotherapy to improve impulse control and reduce the negative psychological impact of urges.

While these premises highlight the therapeutic potential of steroidogenic inhibitors to reduce the synthesis of neurosteroids in specific psychiatric conditions, this strategy may be quite problematic with respect to adverse effects, such as depression and anhedonia. Future avenues will have to include the possibility of targeting brain-specific isoenzymes to minimize certain peripheral effects (including dysmetabolic problems and sexual dysfunctions), as well as designing allosteric modulators of these enzymes, rather than inhibitors, to limit the liability for psychological side effects.

Even with these limitations in mind, however, repurposing steroidogenic drugs can be highly helpful to improve our understanding of the specific role of neuroactive steroids in the brain and capture areas of intervention where these compounds may play an important, albeit overlooked, pathophysiological role. Increased knowledge on the steroid-based mechanisms of psychopathology may be important to discover new mechanisms of action for novel therapies and understand the biological underpinnings of stress. This approach may lead to a better comprehension of the biological mechanisms of psychotherapy and environmental interventions.