The Molecular Basis of 5α-Reductase Type 2 Deficiency

Back in 1961, a disorder of sex development affecting individuals with a 46,XY karyotype was named pseudovaginal perineoescrotal hypospadias [Nowakowski and Lenz, 1961]. The phenotype of this disorder included female-like external genitalia, bilateral testes, and ejaculatory ducts terminating in a blind-ending vagina. Subsequently, animal studies showed that male external genitalia virilization resulted from the conversion of testosterone (T) into dihydrotestosterone (DHT), a reaction catalyzed by the 5α-reductase enzyme [Imperato-McGinley and Zhu, 2002]. The 5α-reductase type 2 deficiency syndrome was biochemically and clinically characterized later in 24 individuals from the Dominican Republic and 2 siblings from North America concomitantly [Imperato-McGinley et al., 1974; Walsh et al., 1974]. Typically, affected individuals are born with female-like external genitalia since DHT is necessary for the prenatal external genitalia virilization, but present virilization (clinical and psychological) at puberty since it relies on T levels rather than DHT, with no evidence of gynecomastia. Both studies characterized this syndrome as a genetic condition with an autosomal recessive inheritance pattern, resulting from the inability to convert T into the most potent androgen DHT. By the 1990s, 2 different genes encoding 5α-reductase isoenzymes were isolated by cloning technology: the 5α-reductase type 1 and 2 (SRD5A1 and SRD5A2) [Andersson et al., 1991; Jenkins et al., 1992]. Mutations in the SRD5A2 gene were found in 2 individuals from Papua New Guinea with clinical features of this DSD. In contrast, normal controls presented no mutations in this gene [Imperato-McGinley et al., 1991]. By studying 25 individuals affected by this condition, Thigpen et al. [1992] discovered that the SRD5A2 gene is mapped within chromosome 2 (2p23) and comprises 5 exons and 4 introns.

Since then, several allelic variants in the SRD5A2 gene have been reported across the whole gene in individuals presenting this particular kind of 46,XY DSD. The impairment in the 5α-reductase type 2 enzymatic activity results from either homozygous or compound heterozygous allelic variants [Thigpen et al., 1992; Mendonca et al., 1996]. Initially, this disorder was reported in clusters worldwide in individuals from specific ethnic groups. There is growing evidence reporting affected individuals with a variety of ethnic backgrounds and coming from several geographical areas, suggesting that 5α-reductase type 2 deficiency has a worldwide distribution [Batista and Mendonca, 2020]. At the hormonal level, the basal hormonal measurements (at minipuberty or post puberty) demonstrated normal male serum T level, low or normal/low DHT level, and an elevated T to DHT (T/DHT) ratio [Mendonca et al., 2016]. This enzymatic deficiency has a wide phenotype variability, ranging from almost typical female external genitalia to hypospadias or isolated micropenis [Gad et al., 1997; Zhu et al., 2014; Cheng et al., 2015; Mendonca et al., 2016; Avendaño et al., 2018; Alswailem et al., 2019; Gui et al., 2019]. This phenotype variability is observed even in affected individuals carrying the same SRD5A2 mutation [Maimoun et al., 2011; Deeb et al., 2016; Andonova et al., 2017], which suggests that other factors than residual 5α-reductase type 2 activity may play a role in the 5α-reductase type 2 deficiency phenotype.

Testosterone is the most abundant androgen in the serum. It is synthesized by the Leydig cells of the testes under the control of pituitary gonadotrophins in postnatal life [Atta et al., 2014; Achermann et al., 2015]. In fetal life, hCG also regulates T synthesis. In male fetuses, T binds to the androgen receptor (AR) to differentiate the wolffian duct into the male internal genitalia, which includes seminal vesicles, vas deferens, and epididymis [Hiort, 2013]. Intracellularly, T is converted into DHT, a more potent androgen, by the 5α-reductase type 2 enzyme. DHT is more biologically active than T, binding to the AR with a 2-fold higher affinity and a decreased dissociation rate of 5-fold compared to T [Grino et al., 1990; Davey and Grossmann, 2016]. In utero, DHT is crucial for prostate gland growth and differentiation of the undifferentiated genitalia into the male genitalia [Imperato-McGinley, 2002; Banerjee et al., 2018]. Therefore, the fetal external genitalia virilization depends on DHT which, in turn, depends on T as a substrate for conversion. Ultimately, the impairment of DHT production will result in undervirilization of a male fetus, resulting in atypical genitalia in 46,XY newborns.

Steroids are a particular type of lipids that are 5α-reduced into more potent steroids by 5α-reductases [Azzouni et al., 2012]. There are three 5α-reductases: 5α-R1, 5α-R2, and 5α-R3. The 5α-R1 and 5α-R2 isoenzymes are encoded by the SRD5A1 and SRD5A2 genes, respectively, sharing approximately 60% of sequence identity, suggesting a common precursor gene during evolution [Azzouni et al., 2012]. The SRD5A1 gene (5p15) encodes a 259 amino acid protein (5α-R1) whereas the SRD5A2 gene (2p23) encodes a 254 amino acid protein (5α-R2) [Russell and Wilson, 1994; Imperato-McGinley and Zhu, 2002]. More recently, a third 5α-R gene (SRD5A3) was identified (4q12), encoding a 318 amino acids protein (5α-R3), sharing around 20% of homology with 5α-R1 and 5α-R2 [Taylor et al., 2017].

All 5α-reductase isoenzymes irreversibly catalyze A-ring reduction of pregnene-based steroids (C19/C21 steroids), which include testosterone, progesterone, cortisol, and aldosterone [Schiffer et al., 2019]. The reaction involves a stereospecific, irreversible breakage of the double bond between carbons 4 and 5 (Δ 4,5). This reduction uses NADPH as a cofactor and leads to the insertion of a hydride anion at carbon C-5 (that is the 5α reduction) [Russell and Wilson, 1994; Azzouni et al., 2012]. Except for DHT, much of the physiological role of 5α-reduced steroids is unknown. The 5α-reductase isoenzymes are expressed in different tissues and are involved in different metabolic processes [Azzouni et al., 2012]. The 5α-R1 is highly expressed in the liver, and its disruption impacts steroid metabolism [Upreti et al., 2014]. The 5α-R2 is expressed in the prostate, seminal vesicles, epididymis, and liver and plays a crucial role in male development [Wilson et al., 1993].

The 5α-R3 is a polyprenol reductase and is required for dolichol synthesis, a final product of the mevalonate pathway [Taylor et al., 2017]. Still, no impact on the functioning or development of reproductive organs has been evidenced by SRD5A3 mutations so far.

We identified 451 cases of 5α-reductase type 2 deficiency in the literature from 48 countries (Fig. 1) carrying 121 different allelic variants in the SRD5A2 gene (Human Gene Mutation Database at the Institute of Medical Genetics in Cardiff, Wales, UK: SRD5A2 gene: http://www.hgmd.cf.ac.uk, Clinvar, and Pubmed). Most are missense mutations (n = 84), but splicing mutations (n = 6), premature stop codons (n = 4), small indels (n = 20), and gross deletions (n = 7) have also been described (Fig. 2).

These variants have been reported in all exons of this gene, mainly at exon 1 (33%) and exon 4 (25%) (Fig. 2). The SRD5A2 variants are located at 76 out of 254 amino acids that make up the 5α-R2 protein.

The 5α-reductase type 2 deficiency is an autosomal recessive inherited condition, resulting from either homozygous or compound heterozygous variants in the SRD5A2 gene [Thigpen et al., 1992; Mendonca et al., 2016]. Like other recessivly inherited conditions, consanguineous populations present a higher frequency of this disease [al-Attia, 1997; Alswailem et al., 2019]. Homozygous allelic variants are more frequent than compound heterozygous among affected individuals with 5α-reductase type 2 deficiency [Thigpen et al., 1992; Mendonca et al., 1996; Alswailem et al., 2019]. Overall, about 70% of allelic variants in the SRD5A2 gene leading to 5α-reductase type 2 deficiency are in a homozygous state, whereas the remaining 30% are compound heterozygous [Maimoun et al., 2011; Avendaño et al., 2018; Batista and Mendonca, 2020].

Altogether, residues 196, 227, 235, and 246 are hotspots of the SRD5A2 gene, making up 25% of all variants reported as causative of 5α-reductase type 2 deficiency. Mutations at these residues have been reported worldwide in individuals from several ethnicities. For example, the p.Arg246Gln variant was reported in India [Shabir et al., 2015], China [Cheng et al., 2019], Pakistan [Berra et al., 2011], Italy [Bertelloni et al., 2016], Turkey [Abacı et al., 2019], and Saudi Arabia [Alswailem et al., 2019], while the p.Gly196Ser variant was reported in Brazil [Hackel et al., 2005], Bulgaria [Andonova et al., 2017], China [Cheng et al., 2019], Italy [Baldinotti et al., 2008], Iran [Hashemi-Gorji et al., 2021], United Kingdom [Berra et al., 2011], and Turkey [Abacı et al., 2019]. However, many other SRD5A2 mutations remain from specific ethnicities, such as the p.Pro59Arg from Algeria [Deeb et al., 2016], the p.Gly183Ser from Brazil [Thigpen et al., 1992], the p.Asn160Asp from Egypt [Maimoun et al., 2011], c.188_189insTA from India [Shabir et al., 2013], c.453delC from Italy [Di Marco et al., 2013], and p.Ala65Pro from Turkey [Abacı et al., 2019].

Most previous reports on 5α-reductase type 2 deficiency patients mainly focused on phenotype, hormonal data, and SRD5A2 sequencing, followed by in silico analysis for pathogenicity prediction. However, 46 variants in SRD5A2 have been functionally investigated (Table 1). Most are non-synonymous allelic variants (42 out 46) within exons 1–5. Since the 5α-R2 enzyme uses NADPH as a cofactor to catalyze the conversion of T into DHT, most functional studies have focused on enzymatic kinetics studies [Wigley et al., 1994; Can et al., 1998; Makridakis et al., 2000; Vilchis et al., 2008, 2010; Zhang et al., 2011; Cheng et al., 2019; Ramos et al., 2020]. In these studies, the enzymatic activity was estimated by V_max, and most of the non-synonymous 5α-R2 allelic variants tested affect the V_max of the 5α-R2 enzyme. Among the SRD5A2 variants that impair enzyme activity, 9 resulted in a protein with no detectable enzyme activity, whereas the remaining led to proteins with severely decreased enzymatic activity (Table 1). The SRD5A2 variants that cause impairment in the 5α-R2 activity are divided among those that affect the affinity for NADPH and those that affect the ligand-binding domain (testosterone) [Thigpen et al., 1992; Russell and Wilson, 1994; Makridakis et al., 2000; Vilchis et al., 2008; Zhu et al., 2014]. Intriguingly, SRD5A2 variants showing no residual enzymatic function, such as p.Leu55Gly and p.Arg246Gln, resulted in a wide range of external genitalia undervirilization [Berra et al., 2011; Maimoun et al., 2011; Zhu et al., 2014], suggesting that some degree of external genitalia virilization is possible even in those carrying SRD5A2 variants with no residual function. An alternative pathway for androgen biosynthesis could explain it since androstenediol can generate DHT through the androgen backdoor pathway [Reisch et al., 2019].

A 3D protein model for SRD5A2 was recently constructed [Katharopoulos et al., 2019]. This model suggests that the SRD5A2 residues Y98, N102, Y107, L167, R171, H231, and Y235 are in direct contact with NADPH at the binding cavity. Since NADPH is the cofactor for 5α-R2 enzyme activity, exploring these residues is interesting. Three (R171, H231, and Y235) have been recurrently reported in individuals with 5α-reductase type 2 deficiency. Mutations in 2 out of these 3 (p.Arg171Ser and p.His231Arg) were functionally evaluated [Wigley et al., 1994]. The p.His231Arg primarily affected the T binding, while the p.HisR171Ser severely decreased the enzyme’s affinity for NADPH.

The conservation of the SRD5A2 residues was analyzed by bioinformatic tools considering evolutionary conservation across species [Katharopoulos et al., 2019]. We previously compared all variants in the SRD5A2 gene in the literature based on amino acid conservation [Batista and Mendonca, 2020] and identified that 76% of the SRD5A2 variants are located at conserved amino acids. Even more intriguing, allelic variants at non-conserved amino acids were more frequently indels (28 vs. 6%) than those at conserved amino acids, which were mostly missense mutations. Surprisingly, allelic variants at non-conserved amino acids were related to less external genitalia virilization due to the predominance of indel mutations [de la Chaux et al., 2007].

Like other inherited disorders, allelic variants in SRD5A2 at canonical splicing sites are deleterious [Wang and Cooper, 2007; Abramowicz and Gos, 2018]. Variants at the canonical splicing sites usually result in exon skipping, leading to an aberrant protein and causing more severe phenotypes [Abramowicz and Gos, 2018]. There are 7 different mutations at canonical splicing sites: c.698+1G>T [Berra et al., 2011], IVS4+2T>C [Abacı et al., 2019], c.725+1G>T [Thigpen et al., 1992], IVS1–2A>G [Alswailem et al., 2019], IVS1–2T>C [Shabir et al., 2015], c.699–1G>A [Berra et al., 2011], and c.282–2A>G [Maimoun et al., 2011]. Although mutations at the splicing region are not that rare among individuals with 5α-reductase type 2 deficiency, only 1 variant was functionally investigated (IVS4+2T>C). This variant results in exon 4 skipping, leading to a short 205 amino acid protein that lacks 5α-R2 residual activity [Zhu et al., 2014]. However, the real impact of non-synonymous variants of SRD5A2 in splicing has not been explored yet [Abramowicz and Gos, 2018].

We detected 21 small indels in the SRD5A2 gene reported in individuals with 5α-reductase type 2 deficiency, ranging from 1 to 11 nucleotides. As expected, all small indels (not 3 nucleotides insertion or deletion included), led to frameshift mutations. In addition, most of them (11 out 21) presented as part of compound heterozygous variants.

Gross deletions are relatively rare among 5α-reductase type 2 deficiency individuals. Deletion of the almost entire SRD5A2 gene, as well as exons 1 and 2 deletions, have been reported, either in homozygosity or as a component of compound heterozygosity [Andersson et al., 1991; Fénichel et al., 2013; Deeb et al., 2016] among 5α-reductase type 2 deficiency individuals from Arab Emirates [Deeb et al., 2016], Bulgary [Andonova et al., 2017], France [Fénichel et al., 2013], Iran [Hashemi-Gorji et al., 2021], Papua New Guinea [Andersson et al., 1991], and Macedonia [Kocova et al., 2019]. Curiously, most SRD5A2 large deletions involve exon 1 or exon 2 deletion. Most large deletions are presented in homozygosity, but large deletions as part of compound heterozygosity have also been reported [Andonova et al., 2017; Kocova et al., 2019]. The first large deletion of the SRD5A2 gene (comprising almost the entire gene) was reported in 1991 in 2 related individuals from Papua New Guinea [Andersson et al., 1991]. Later, a homozygous deletion of the SRD5A2 exon 1 was reported in a young elite athlete, whose parents were first cousins, through hormonal screening for hyperandrogenism [Fénichel et al., 2013]. Large deletions including the SRD5A2 exon 1 were also reported in two 5α-reductase type 2 deficiency individuals from Bulgaria, one carrying this deletion in homozygosity and the other as part of compound heterozygosity [Andonova et al., 2017]. In 2016, an Emirati family with 11 affected individuals carrying a homozygous deletion of the SRD5A2 exon 2 was reported. Intriguingly, this family presented a range of phenotypes that led to different sex of rearing even with the same genetic background [Deeb et al., 2016]. Another SRD5A2 deletion was reported in 2 siblings from Macedonia as part of compound heterozygosity (along with the p.Ala49Asp variant). To define the deletion more precisely, the authors performed aCGH and real-time PCR analysis. They were able to establish that the entire exon 1 and 5′ untranslated region (5′-UTR) of the SRD5A2 gene were deleted within the region at least approximately 8 kb upstream and approximately 2 kb downstream of exon 1. A more precise determination of the deletion breakpoints was hampered by the presence of 64 kb of interspersed repeats and low complexity DNA sequences upstream of the SRD5A2 gene [Kocova et al., 2019]. Recently, a 13-kb SRD5A2 deletion was reported from Iran (g.40936_53878del12943insTG). In this case, the SRD5A2 deletion occurred between introns 1 and 2, leading to exon 2 deletion [Hashemi-Gorji et al., 2021].

Unfortunately, the mechanisms underlying SRD5A2 deletions are still unexplored. Recombination among repetitive elements in the DNA has increasingly been identified as a mechanism for genomic deletions through non-allelic homologous recombination [Parks et al., 2015; Kazazian and Moran, 2017; Lisch and Burns, 2018; Szafranski et al., 2018]. This possibility was suggested by the interesting case from Macedonia in which the large SRD5A2 deletion happened in a genomic region rich in repetitive elements [Kocova et al., 2019].

Based on the genome browser, the chromosomal region covering the SRD5A2 gene (chr2:31,522,480–31,580,938) is, indeed, enriched by mobile DNA elements, especially LINE-1 (long interspersed nuclear elements) sequences, including L1Hs (chr2:31,575,999–31,577,158), L1PA15 (chr2:31,569,929–31,572,454), L1M1 (chr2:31,540,540–31,541,521), L1MB3(chr2:31,540,540–31,541,521), and L1MDa (chr2:31,539,048–31,539,981) and also by SINE (short interspersed nuclear elements), such as Alu elements (AluSx1, AluSg4, AluSp, and AluSx3). Due to the ability of some repetitive DNA elements to mobilize across the genome, as well as their similarity, they are a target for non-allelic homologous recombination. As reported for other inherited conditions [Ule, 2013; Hancks and Kazazian, 2016; Batista et al., 2019], the LINE-1 and Alu sequences within the SRD5A2 region might be evolved in the molecular etiology of 5α-reductase type 2 deficiency.

In an interesting case of uniparental disomy, a patient born to non-consanguineous parents carried 2 distinct variants (p.Glu197Asp and p.Pro212Arg). The patient was homozygous for the p.Glu197Asp, indicating an alternative mechanism whereby 5α-reductase type 2 deficiency can derive from a single parent [Chávez et al., 2000].

In a case series with 14 individuals with 5α-reductase type 2 deficiency from China, 2 patients carried 3 mutations in the SRD5A2 gene. The first one presented the p.Gln6Term, p.Lys35Asn, and p.Phe234Leu, and the second one the p.Gly34Arg, p.Gly203Ser, and p.Arg227Gln[Cheng et al., 2019]. Patient 1 was born with female-like external genitalia, was assigned as female, but presented virilization at 13 years of age with gender change from female to male later, whereas patient 2 was born with microphallus and perineoscrotal hypospadias. It is worth to note that some degree of virilization occurred in patient 2 even in the presence of 3 pathogenic variants in the SRD5A2 gene that were functionally evaluated: the p.Gly34Arg affects T affinity [Wigley et al., 1994], whereas both p.Gly203Ser and p.Arg227Gln affect 5α-R2 enzymatic activity [Makridakis et al., 2000; Zhang et al., 2011].

The role of polymorphisms as causative of human diseases or influencers of the phenotype has been debatable in many human conditions. In the SRD5A2 gene, the most frequent polymorphism is located at exon 1 (p.Val89Leu), in which the substitution from valine to leucine at position 89 proved to impair the 5α-R2 activity by 30% compared with the wild type [Makridakis et al., 2000]. This polymorphism has an allele frequency of 0.6648 based on GnomeAD (www.gnomad.broadinstitute.org; rs523349). Despite being debatable, this SRD5A2 polymorphism is worthy of attention because there is evidence suggesting its role in the 5α-reductase type 2 deficiency. Firstly, it was reported as causative of isolated hypospadias among Indian children [Samtani et al., 2011]. That associated risk is reinforced by a meta-analysis of 6 studies focused on exploring the relationship between SRD5A2 polymorphisms and hypospadias [Sun et al., 2019]. The authors concluded that the SRD5A2 polymorphisms might be one of the risk factors for isolated hypospadias development. Secondly, this polymorphism has been reported as a compound heterozygous mutation in patients with 5α-reductase type 2 deficiency [Maimoun et al., 2010; Shabir et al., 2015]. Third, the p.Val89Leu may influence the phenotype by itself. In that last sense, a recent study showed affected individuals whose fathers were healthy carriers of the p.Arg246Gln variant in homozygosity. Interestingly, the affected patients carried homozygous p.Val89Leu variant whereas the fathers were heterozygous for polymorphism, suggesting that the 5α-reductase type 2 deficiency was caused by the presence of the polymorphism instead of the p.Arg246Gln variant [Arya et al., 2020].

The association of this polymorphism with androgen-related cancer risk is even more controversial. No association between the p.Val89Leu polymorphism and an increased risk of developing prostate cancer was identified among 456 Slovak male patients [Dušenka et al., 2014]. On the contrary, the presence of p.Val89Leu polymorphism has been implicated as a risk for breast cancer development [Francis et al., 2014].

Another polymorphism of the SRD5A2 gene which deserves attention is the change from alanine to threonine at codon 49 (p.Arg49Thr). In vitro studies showed that this polymorphism increases the enzymatic activity of the 5α-R2 enzyme by 5 folds [Makridakis et al., 2000], which has been suggested as a risk for prostate cancer development [Paz-y-Miño et al., 2009; Li et al., 2013]. However, it has not been confirmed in other studies [Pearce et al., 2008; Li et al., 2010]. On the other hand, a modest association of this polymorphism has been reported in individuals with less severe hypospadias, suggesting a “protective role” for more severe phenotypes in those carrying this polymorphism [Silver and Russell, 1999].

A variable number of dinucleotide TA repeat lengths exists at the 3′ unstranslated region (3′-UTR) of the SRD5A2 gene, ranging from (TA)0 or (TA)9 to (TA)18 [Samtani et al., 2015]. However, it remains unclear if it can influence the enzymatic activity of the SRD5A2 gene. To investigate a risk associated with TA repeats, a study failed to prove that the TA repeats number favors breast or ovarian cancer, and another study failed to prove this polymorphism as causative of isolated hypospadias among children from North India [Spurdle et al., 2001; Samtani et al., 2015]. As far as prostate cancer is concerned, evidence suggested that long TA repeats might decrease the risk for prostate cancer [Li et al., 2011], whereas the (TA)9 polymorphism confers a prostate cancer risk among Lebanese men [El Ezzi et al., 2017].

The 5α-reductase type 2 deficiency is a condition frequently reported as having no genotype-phenotype correlation. This observation is based on many 5α-reductase type 2 deficiency families carrying the same genotype that present a broad range of phenotypes [Maimoun et al., 2011; Deeb et al., 2016; Alswailem et al., 2019].

Recently, we analyzed the range of external genitalia virilization reported among individuals with 5α-reductase type 2 deficiency carrying recurrent SRD5A2 homozygous variants in the literature to estimate the genotype-phenotype relationship [Batista and Mendonca, 2020]. External virilization, based on the Sinnecker’s scores (from 1 to 5, being 1 a typically male external genitalia) [Sinnecker et al., 1996] ranged from 2a to 4b in affected individuals from Turkey, Lebanon, and Iraq carrying the p.Leu55Gly variant [Hochberg et al., 1996; Ocala et al., 2002; Adiyaman et al., 2006; Walter et al., 2010; Abacı et al., 2019]. It was also evidenced by individuals carrying the p.Pro181Leu variant, which was reported in Saudi Arabia and Turkey, in whom the Sinnecker’s scores ranged from 2a to 4a [Parlak et al., 2014; Abacı et al., 2019; Alswailem et al., 2019]. Phenotypic variability is also observed in patients whose genotype is ethnic-exclusive, such as the p.Ala65Pro reported in individuals from Turkey (Sinnecker’s scores ranged from 2a to 5) [Abacı et al., 2019]. This phenotype variability is also observed in more complex mutations, such as the large deletion in the SRD5A2 gene.

We also analyzed the phenotype variability in mutations at hotspots of the SRD5A2 gene [Batista and Mendonca, 2020]. No genotype-phenotype correlation was observed in individuals carrying the p.Thy235Phe variant (Sinnecker’s scores ranged from 2a to 5) [Wigley et al., 1994; Mazen et al., 2003; Parlak et al., 2014; Bertelloni et al., 2016] and also in those carrying the p.Gly196Ser variant (Sinnecker’s score ranged from 1a to 4a) [Thigpen et al., 1992; Baldinotti et al., 2008; Bertelloni et al., 2016; Abacı et al., 2019].

However, some SRD5A2 variants show consistency in the way they affect the phenotype. The p.Arg246Gln variant is associated with more virilization (2a–3a) than both p.Gly183Ser and p.Gln126Arg variants, that consistently lead to more severe external genitalia undervirilization (3b–4b and 4a–4b, respectively) [Thigpen et al., 1992; Mendonca et al., 1996; Vilchis et al., 2000; Hackel et al., 2005; Baldinotti et al., 2008; Bertelloni et al., 2016; Abacı et al., 2019].

Isolated micropenis has been reported in individuals from Japan carrying the p.Arg227Gln variant in homozygosity [Sasaki et al., 2003]. Male-phenotype (Sinnecker’s score 1a) was reported in several patients from China carrying the same homozygous variant [Cheng et al., 2015]. Intriguingly, an in vitro study showed that the p.Arg227Gln variant reduces the maximum velocity of enzyme reaction (V_max) to near 3.2 of normal activity [Makridakis et al., 2000], suggesting that the external genitalia virilization is not only dependent on the 5α-R2 activity in these patients.

To explore the genotype-phenotype correlation in the 5α-reductase type 2 deficiency, an interesting review [Avendaño et al., 2018] collected reported 5α-reductase type 2 deficiency patients carrying only homozygous variants in the SRD5A2 gene, that were phenotypically classified using the external masculinization score, a 0–12 scale, being 12 a fully male phenotype [Ahmed et al., 2000]. The authors calculated the EMS average and standard deviation (SD) for some SRD5A2 variants presented in homozygosity among individuals with 5α-reductase type 2 deficiency. Some of them showed a very low SD, such as the p.Gly115Asp and the p.Gly196Ser variants, suggesting less phenotype variability in patients carrying those variants. However, other variants, such as p.Tyr235Phe, p.Leu55Gln, and p.Asn160Asp had high SD, indicating more phenotype variability. The authors also divided the SRD5A2 variants into 3 subgroups based on the functional consequence of the SRD5A2 variant. Among variants that diminish T affinity (such as p.Gly34Arg and p.His231Arg), the EMS average varied between 2.0 and 3.3, indicating a predominantly female phenotype [Avendaño et al., 2018]. Among SRD5A2 variants that interfere with the NADPH domain (p.Pro181Leu, p.Gly183Ser, p.Gly196Ser, p.Tyr235Phe, p.Arg246Gln, p.Arg246Trp), the EMS average varied between 2.67 and 4.17, indicating more virilization than SRD5A2 variants that diminish T affinity. In the subgroup with SRD5A2 variants decreasing enzymatic activities (p.Leu55Gln, p.Gln126Arg, p.Tyr91His, p.Asn126Asp, p.Arg227Gln, c.282–2A>G), the EMS average ranged from 3.0 to 8.0, suggesting that the residual enzymatic activities caused by these variants may play a role in the phenotype variability in this SRD5A2 variants subgroup [Avendaño et al., 2018].

Collectively, these data show that the 5α-reductase type 2 deficiency is a condition without genotype-phenotype correlation. The underlying reasons for this incongruence are still unclear. Since DHT acts trough the AR, it is possible that AR polymorphisms may play a role in the genotype-phenotype correlation.

Another possible explanation is the DHT generation by the androgen backdoor pathway. Since this pathway depends on enzymes encoded by specific genes, such as HSD17B6 and AKR1C2 [Fukami et al., 2013], allelic variants or polymorphisms in these genes may also be related to the genotype-phenotype incongruence observed in this 46,XY DSD condition.

However, some SRD5A2 variants present genotype-phenotype congruence, and SRD5A2 variants affecting residual enzymatic activity show less genotype-phenotype correlation than those affecting T or NADPH binding.

46,XX individuals carrying mutations in the SRD5A2 gene have been rarely reported. Three 46,XX females from a Dominican kindred with 5α-reductase type 2 deficiency were homozygous for the p.Arg246Trp variant in the SRD5A2 gene [Katz et al., 1995]. These individuals presented delayed menarche, but all were fertile, and 2 had twins. Plasma T was normal to elevate, with low DHT, resulting in an increased T/DHT ratio. Ovulatory gonadotropin peaks were evidenced during menstrual cycle profiling in 2 of them. The authors discussed that delayed puberty might suggest involvement of 5α-R2 in menarche at the hypothalamic/pituitary and/or ovarian level. They also indicated that DHT and/or the DHT/estradiol ratio might play a role in follicular development, with lower levels allowing more than one dominant follicle per cycle, since 2 individuals had nonidentical twins [Katz et al., 1995].

The evidence that SRD5A2 pathogenic mutations do not affect fertility among 46,XX carriers was reinforced by another study that included two 46,XX women carrying pathogenic SRD5A2 mutations (one in homozygosity and the other in compound heterozygosity) who had 7 children. The authors found a normal serum concentration of 5 alpha-dihydroprogesterone during the luteal phase, suggesting that the 5α-reduction of progesterone in women is mediated mainly by the 5α-R1 isoenzyme instead of by the 5α-R2 [Milewich et al., 1995]. However, the role of DHT in female physiological processes remains unrevealed.

As for other 46,XY DSD conditions, fertility remains a challenge for 46,XY individuals with 5α-reductase type 2 deficiency [Guercio et al., 2015; Wisniewski et al., 2019]. Infertility among these patients is due to several reasons: an impact of inguinal/abdominal temperature on spermatogenesis, impairment of germ cell maturation, low ejaculation, and viscous semen by underdeveloped seminal vesicles and prostate [Nordenskjöld and Ivarsson, 1998; Bertelloni et al., 2019]. Therefore, proven paternity has been rarely reported among 5α-reductase type 2 deficiency patients, either by natural conception [Nordenskjöld and Ivarsson, 1998; Bertelloni et al., 2019] or after assisted reproductive techniques [Katz et al., 1997; Matsubara et al., 2010; Kang et al., 2011; Costa et al., 2012]. Intriguingly, in both studies presenting natural conception, the patients carried 2 missense variants in the SRD5A2 gene in heterozygosity (p.Gly196Ser/p.His231Arg and p.Arg103Pro/p.His230Pro) [Nordenskjöld and Ivarsson, 1998; Bertelloni et al., 2019], while the other cases with assisted paternity showed SRD5A2 variants in homozygosity (p.Arg246Trp/Arg246Trp and Arg246Gln/Arg246Gln) [Katz et al., 1997; Matsubara et al., 2010; Kang et al., 2011]. The reasons why compound heterozygous individuals could reach paternity are discussed in the light of p.Gly196Ser/p.His231Arg cases. The authors suggested a preponderance of the less severe variant (p.Gly196Ser) over the p.His231Arg variant, which abolishes the 5α-R2 enzymatic activity. Among those with assisted paternity, 2 were submitted to intrauterine insemination (IUI) [Katz et al., 1997], whereas the remaining performed intracytoplasmatic sperm injection (ICSI) [Matsubara et al., 2010; Kang et al., 2011; Costa et al., 2012; Bertelloni et al., 2019]. All newborns were healthy. Twin pregnancy was reported in 3 cases, 1 after IUI [Katz et al., 1997] and 2 after ICSI [Kang et al., 2011; Costa et al., 2012]. The older brother could have natural conception from the Italian brothers, while his youngest brother needed assisted reproduction. The older brother received a trial of extractive gonadotropins during adolescence (the exact drug dosage is not available), whereas the youngest did not receive it. The authors highlighted it, discussing if gonadotropin administration before adolescence may have induced germline maturation, allowing further natural conception [Bertelloni et al., 2019]. Indeed, hormonal treatment with either human chorionic gonadotropin (hCG) or gonadotropin-releasing hormone (GnRH) stimulates testicular growth and spermatogenesis initiation in the cryptorchid testes, but it also can be harmful to germ cells [Virtanen and Toppari, 2015]. It has been suggested that hormonal treatment may benefit particular cryptorchidism subgroups [Hadziselimovic, 2017], but studies are necessary to identify whether this treatment could benefit 5α-reductase type 2 deficiency patients.

5α-reductase type 2 deficiency is a rare inherited condition with a worldwide distribution resulting from allelic variants in the SRD5A2 gene, leading to a broad spectrum of external genitalia development with no evident genotype-phenotype correlation. Although several allelic variants in the SRD5A2 gene have been reported as causative of 5α-reductase type 2 deficiency, the role of the SRD5A2 polymorphisms in male development and other human diseases remains debatable.

The authors have no conflicts of interest to declare.

This study is supported by FAPESP (Fundação de Amparo a Pesquisa do Estado de São Paulo), grant 2019/26780-9, and by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq 307571/2021-1) to B.B.M.

Rafael L. Batista and Berenice B. Mendonca performed the data and bibliographic search, discussed the contents, contributed to the manuscript writing, and approved the final version of this manuscript.