p.H62L, a Rare Mutation of the CYP21 Gene Identified in Two Forms of 21-Hydroxylase Deficiency

Abstract

Context: Steroid 21-hydroxylase deficiency is the most common enzymatic defect causing congenital adrenal hyperplasia with good genotype/phenotype relationships for common mutations. To determine the severity of rare mutations is essential for genetic counseling and better understanding of the structure-function of the cytochrome P450c21.

Objective: The p.H62L mutation was the most frequent of 60 new mutations detected in 2900 steroid 21-hydroxylase deficiency patients, either isolated or associated on the same allele with a mild mutation (p.P453S, p.P30L, or partial promoter). Because phenotypes seemed to differ between patients with isolated or associated p.H62L, a detailed phenotype description and functional studies were performed.

Results: Regarding phenotype, patients with isolated p.H62L had a nonclassical form, whereas patients with the association p.H62L + mild mutation had a simple virilizing form. Functional studies showed that p.H62L reduced the conversion of the two substrates, progesterone and 17-hydroxyprogesterone, in the same way as the mild p.P453S; the association p.H62L + p.P453S decreased enzymatic activity more strongly while conserving residual activity at a level intermediate between p.P453S and p.I172N. This suggested that p.H62L was a mild mutation, whereas a synergistic effect occurred when it was associated. Analysis of p.H62L in a three-dimensional model structure of the CYP21 protein explained the observed in vitro effects, the H62 being located in a domain implied in membrane anchoring.

Conclusion: According to phenotype and functional studies, p.H62L is a mild mutation, responsible for a more severe phenotype when associated with another mild mutation. These data are important for patient management and genetic counseling.

Congenital adrenal hyperplasia (CAH) is an inborn autosomal recessive adrenocortical steroidogenesis error (1). Steroid 21-hydroxylase deficiency (21OHD) represents about 95% of CAH, varying widely depending on the genetic lesion, and is traditionally divided into two groups, a severe classical and a milder nonclassical form (1, 2). The classical form comprises female external genitalia virilization by fetal adrenal androgen hyperproduction and severely impaired glucocorticoid and mineralocorticoid secretion; in salt-wasting (SW) 21OHD, life-threatening neonatal salt loss occurs during the first month of life, and in simple virilizing (SV) 21OHD, the mineralocorticoid disorder is less severe, with clinical salt loss only in case of stress. In both SW and SV, the mineralocorticoid defect is attested by increased plasma renin. Before neonatal screening, SV in males was often diagnosed only in childhood (3–7 yr) by precocious pseudopuberty leading to true precocious puberty and decreased adult height. In the nonclassical (NC) form, mineralocorticoid secretion is conserved, female virilization absent, and phenotypic manifestations milder. Children of both sexes present milder hyperandrogenism than in SV; women refer for hirsutism, menstrual irregularity, and decreased fertility, and men are often asymptomatic.

Two 21-hydroxylase loci are located on chromosome 6p21.3 within the HLA class III gene region: functional CYP21A2 and its nonfunctional pseudogene CYP21A1P, duplicated in tandem with C4 genes (3). The high homology of these duplicated regions causes misalignment during meiosis, resulting in genetic lesions (2, 4–6), large (deletions, gene conversions) in 20% and point mutations in 80% of cases, most of the latter being small conversions. About 114 deleterious mutations were published (http://www.imm.ki.se/CYPalleles/cyp21.htm), including IVS2-13A/C>G, p.I172N and p.V281L, the most common in SW (30–35%), SV (14–18%), and NC (>50%), respectively (5, 7, 8).

Good genotype-phenotype correlations were demonstrated for common mutations (2, 4, 5, 9–13). The compound heterozygous phenotype is determined by the mildly affected CAH allele. Rare mutation severity is predicted for nonsense and frameshift but not missense mutations, which are the most frequent and where it should be deduced from the phenotype only if the cohort is large enough. In vitro enzyme function assay after site-directed mutagenesis is often the only way to determine impaired hormonal secretion, especially in males detected by neonatal screening and compound heterozygotes with mild mutation on the other allele. Moreover, the evaluation of the severity’s degree of the association of two mild mutations on the same allele, although more difficult to estimate, is essential for genetic counseling. Such associations, with more severe phenotype than either isolated mutation, have been reported: p.P453S with p.R339H (14) and p.P453S with p.P105L (15).

Our screening of more than 5000 21OHD alleles identified 60 new mutations, with p.H62L (c.185A>T), in the first exon, being the commonest (16); it was identified in 13 patients from 12 unrelated families, either isolated or associated on the same allele with a mild mutation: p.P453S, p.P30L, or partial promoter conversion of CYP21A2 (−113G>A, −110T>C, and −103A>G). The phenotype for p.H62L associated with mild mutation was more severe than with isolated p.H62L. We here report good phenotype-genotype correlation using detailed phenotypes and functional studies. This evaluation was essential for genetic counseling, because prenatal diagnosis and dexamethasone treatment have to be proposed only for fetuses at risk of classical 21OHD (17).

Patients and Methods

Genotyping of CYP21A2 was performed in 2900 patients, 1570 with classical and 1330 with NC form; new mutations were detected in 130 patients (4%), most frequently p.H62L (13 patients, 0.4%).

Patients

Data of the 13 patients are summarized in Table 1.

TABLE 1.

Clinical data of the patients

Case	Sex	Age of symptoms/age of diagnosis (yr)	Clinical data	17OHP (nmol/liter) basal/post-ACTH	Renin (ng/liter)	ACTH (pg/ml) cortisol	Androgens (nmol/liter)	Genotype
1	F	15/26	Dysmenorrhea, amenorrhea after pill	69/108			17.1 (Δ4), 4.8 (T)	H62L/del
2	F	NT/18	Primary amenorrhea, moderate hirsutism	99.4/ND			20 (Δ4), 3.4 (T)	H62L+partial promoter/IVS2–13A/C>G
3	F	NT/7	Prader 1, pubic hair, primary amenorrhea	115/345		67	4.76 (T)	H62L+P453S/large lesion
4	F	1/7.6	Prader 1, BA 11.6/CA 7.6	780/ND				H62L+P453S/Q318X
5	F	Birth/11	Prader 3, raucous voice, hirsutism	159/ND	Normal		39.6 (Δ4), 15 (T)	H62L+P453S/IVS2–13A/C>G
6	F	1.6/6	BA 16/CA 11, Prader 1, acne, pubic hair, BA 10/CA 6	360/ND			10.5 (Δ4), 5.4 (T)	H62L+P453S/I172N
7	F	Birth/0.3	Prader 2, no surgery	224/ND	Normal		25.75 (Δ4), 3.27 (T)	H62L+P453S/I172N
8	F	Birth /0.2	Prader 1, regression under treatment	112/ND	800		7.8 (Δ4), 0.71 (T)	H62L+P453S/I172N
9	M	6/0.1 (related to a SV)	BA 7.6/CA 6	48/ND	Normal		3.9 (Δ4), 4.18 (T)	H62L/I172N
10	M	5/7	BA 11.6/CA 7 Pubic hair acne	350/471		16.9	19.6 (Δ4), 2.6 (T)	H62L+P453S/Q318X+R356W
11	M	1/10.6	BA 14/CA10.6	456/ND		Increased		H62L+P453S/Q318X
12	M	NT/4.3	BA 9–10/CA 4.3, pubic hair, length of penis 8 cm	348/ND	45	Decreased	20 (Δ4), 2.75 (T)	H62L+P453S/IVS2–13A/C>G
13	M	4.6/6.10	BA 13/CA 6.10, pubic hair	1678/1809	46	114		H62L+P30L/H62L+P30L

Female 3 and male 11 are siblings. Hormonal data were measured at the diagnosis. Δ 4, Androstenedione; BA, bone age; CA, chronological age; F, female, M, male; ND, not done; NT, not transmitted; T, testosterone.

Females 1 and 2 were diagnosed in adulthood by mild signs of hyperandrogenism (hirsutism and dysmenorrhea). 21OHD was confirmed by elevated 17-hydroxyprogesterone (17OHP), from 69 to 108 nmol/liter after ACTH test for female 1 and basal 17OHP at 99.4 nmol/liter for female 2. Final heights were as expected (respectively, 1.75 and 1.63 m).

Female 3 was diagnosed at 7 yr by isolated clitoromegaly and pubic hair. 21OHD was confirmed by elevated 17OHP, from 115 to 345 nmol/liter after ACTH test. Not treated, she presented with primary amenorrhea.

Female 4 was symptomatic since 1 yr of age, presenting with clitoromegaly and advanced stature and bone age. 21OHD was confirmed at 7 yr by elevated 17OHP at 780 nmol/liter; stature was advanced at +5 sd and bone age at 11 yr, with decreased final height of 1.57 m.

Female 5 was not diagnosed at birth despite considerable clitoromegaly (Prader stage 3), noted by her mother. Pubic hair appeared at 8 yr and axillary hair at 9, with advanced stature (+2.5 sd); she was referred to endocrinology only at 11 yr, for increased virilization, deep voice, hirsutism, and abnormal muscular mass. Bone age was advanced at 16 yr, leading to decreased final height (1.53 m vs. a target of 1.60 m). Basal 17OHP was elevated (159 nmol/liter) as were androgens, although mineralocorticoid function was unaltered (data not transmitted).

Female 6 was seen at 18 months for advanced stature and clitoromegaly; acne appeared at 2 yr, and pubic hair at 6. 21OHD was diagnosed by elevated 17OHP at 360 nmol/liter, reaching 540 nmol/liter at the age of 8 yr, despite hydrocortisone treatment. Androgen levels were high. Final height was decreased (1.63 m) compared with her parents (1.90 and 1.60 m).

Female 7 was presented with external genitalia virilization (Prader 2); neonatal screening was negative (48 nmol/liter, cutoff at 60 nmol/liter), but elevated 17OHP (224 nmol/liter) and androgens at 3 months confirmed 21OHD.

Female 8 was born at 33 wk gestation, presented with clitoromegaly at birth; neonatal screening was positive (121 nmol/liter, cutoff at 60 nmol/liter), but because she was born preterm, successive high controls (112 nmol/liter on d 10 to 297 nmol/liter on d 66) were essential to confirm diagnosis. Elevated renin suggested mineralocorticoid deficiency.

Male 9 was a nephew of a SV patient (p.I172N/deletion); basal 17OHP at 1 month (48 nmol/liter) and ACTH test at 3 months (17OHP from 14 to 130 nmol/liter) confirmed 21OHD. He presented with moderately advanced bone age at 6 yr.

Male 10 was symptomatic since 5 yr of age, presenting with precocious pseudopuberty and acne. 21OHD was confirmed at 7 yr by elevated 17OHP (from 350 to 471 nmol/liter after ACTH) and androgens.

Male 11 presented with advanced stature since 1 yr of age but was diagnosed at only 10.6 yr by precocious pseudopuberty. Basal 17OHP levels were elevated at 456 nmol/liter.

Male 12 was undetected by neonatal screening and referred to pediatrics at 4.3 yr with considerable phallic growth; he presented with greatly advanced stature (+4.2 sd) and bone age estimated at 9–10 yr. Elevated basal 17OHP (350 nmol/liter) and androgens confirmed 21OHD.

Male 13 presented with precocious pseudopuberty at 4.6 yr and was diagnosed at 6.10 yr; 17OHP was elevated, from 1678 to 1809 nmol/liter after ACTH.

Molecular study of CYP21A2

Peripheral blood was sampled from patients and parents (if available) after informed consent. Molecular study of CYP21A2 used specific amplification by PCR followed by entire gene sequencing, from the 5′ regulatory region (c.−400) to the end of translation. The PCR program was: 95 C for 10 min; 30 cycles of 20 sec at 95 C, 20 sec at 60 C, and 20 sec at 72 C; and a final 10-min 72 C extension. PCR and sequencing products were purified using the Millipore (Billerica, MA) system. Sequencing products were run on ABI Prism Genetic Analyzer 3130.xl (Applied Biosystems, Foster City, CA). Primers sequences are available on request.

Construction of plasmids and mutagenesis

In vitro studies were validated using three frequent mutations in addition to the new mutated alleles: p.I172N (c.515T>A), with about 2% residual 21-hydroxylase activity and responsible for SV (18), and two mutations associated with the NC form: p.V281L (c.841G>T), with about 50% residual 21-hydroxylase activity for 17OHP and 20% for progesterone (18), and p.P453S (c.1357C>T), with 68% residual 21-hydroxylase activity for 17OHP and 46% for progesterone (15).

Human full-length CYP21A2 cDNA cloned into the pCMV4 vector was kindly provided by Dr. Bon Chu Chung (19). Mutations were introduced using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA).

Control plasmid construction used the following primers (with mutation introduced in the middle of the primer and indicated in bold lowercase letter): for p.I172N, I172N forward 5′-CAC CTG CAG CAT CAa CTG TTA CCT CAC CTT CGG-3′ and I172N reverse 5′-CCG AAG GTG AGG TAA CAG tTG ATG CTG CAG GTG-3′; for p.V281L, V281L forward 5′-GGA AGG GCA CtT GCA CAT GGC TGC AGT GGA CC-3′ and V281L reverse 5′-GGT CCA CTG CAG CCA TGT GCA aGT GCC CTT CC-3′; and for p.P453S, P453S forward 5′-CCT TCA CGC TGC TGt CCT CCG GGG ACG CCC TGC CCT CC-3′ and P453S reverse 5′-GGA GGG CAG GGC GTC CCC GGA GGa CAG CAG CGT GAA GG-3′.

Mutated plasmid construction used the following primers: for p.H62L, H62L forward 5′-CCA TCT ACA GGC TCC tCC TTG GGC TGC AAG-3′ and H62L reverse 5′-CTT GCA GCC CAA GGa GGA GCC TGT AGA TGG-3′. For p.H62L+p.P453S, the plasmid containing p.H62L was used to introduce the p.P453S mutation with the primers described above.

XL1-Blue supercompetent cells were transformed with pCMV4 vectors by heat pulse, and bacteria were grown in ampicillin-containing medium. The complete CYP21A2 cDNA was sequenced to check for the expected mutation and exclude additional sequence aberrations.

In vitro P450c21 expression in COS-1

Liposomes (Lipofectamine reagent) were used to transfect the COS-1 cells; transfection used mutant pCMV4-CYP21A2 constructs, wild-type (WT) pCMV4-CYP21A2 as reference, and native pCMV4 without CYP21A2 cDNA as negative control. According to the manufacturer’s instructions (Invitrogen, Karlsruhe, Germany), 3 × 10⁵ cells (50% confluence) were cotransfected per well with 1 μg pCMV4 constructs and 1 μg growth factor (GH) as transfection control. Cell recovery and protein expression were achieved by incubating cells at 37 C, 5% CO₂ for 48 h.

Mutation severity assessment

Each mutant was compared with WT and known mutant CYP21A2. Transfected cells were incubated for 2 and 8 h at 37 C, 5% CO₂ with 1 ml DMEM/F12 medium (supplemented with l-glutamine, antibiotics, and 5% fetal bovine serum) containing 1 μCi ³H-labeled substrate 17OHP or progesterone. Supernatant (500 μl) was extracted twice with 5 ml (isooctane/ethyl acetate 1:1, vol/vol), evaporated, and dissolved in ethanol [containing size markers 17OHP plus 11-deoxycortisol/progesterone plus 11-deoxycorticosterone (DOC)]. Steroids were separated by thin-layer chromatography with chloroform/ethyl acetate (80:20, vol/vol) as mobile phase.

The plate was autoradiographed after addition of Enhance Spray for 24 h, and the autoradiographs were scanned. Mutated plasmid enzyme activity was determined by comparing spot intensity with that of the known mild and severe mutations.

GH/total protein ratios were calculated to check reproducibility of transfection efficiency, using the hGH-RIACT kit (CIS Bio International, Gif-Sur-Yvette, France).

Determination of kinetic constants

The COS-1 cells were incubated for 1 h at 37 C with 1 ml DMEM/F12 medium containing 0.5 μCi ³H-labeled substrate and six unlabeled steroid concentrations: 0.5, 1, 2, 3, 4, or 6.0 μmol/liter. Postincubation treatment was as described above. To quantify residual 21-hydroxylase activity, radioactivity was counted using Scion image software (Scion, Frederick, MD). Substrate conversion per steroid concentration was calculated after correction of total protein content. Apparent kinetic constants were then calculated, using GraphPad Prism software (version 5.0; GraphPad Software Inc., San Diego, CA).

Western blot

To determine whether decreased enzyme activity was due to reduced enzyme expression or function, protein analysis of WT and the various mutants was performed by Western blot. Antihuman CYP21A2 rabbit polyclonal antiserum was provided by Dr. W. L. Miller. This technique was performed as recommended in the standard protocol (Invitrogen kit).

Molecular modeling

Multiple sequence alignment

Many P450 cytochrome sequences were used to analyze structurally conserved regions by multiple-sequence alignment. Sequences homologous to human CYP21A2 were found using BLAST searches in the UNIPROT database (20) and aligned with CLUSTAL W (http://www.ebi.ac.uk/clustalw/) (21) using default parameters, displayed, and then edited using Genedoc (http://www.psc.edu/biomed/genedoc).

Human CYP21A2 (P08686) was aligned with three cytochrome groups: 1) P450c21s of mouse (P03940), rat (Q64562), bovine (P00191), sheep (Q7M366), pig (P15540), and dog (Q8WNWO); 2) other human steroidogenic P450 cytochromes, CYP11B1 (P15538), CYP11B2 (P19099), CYP11A1 (P05108), CYP17A1 (P05093), and CYP19A1 (P11511); and 3) mammalian crystallized cytochromes, rabbit CYP2C5 (P00179), rabbit 2B4 (P00178), human CYP2C8 (P10632), human CYP2C9 (P11712), and human CYP3A4 (P08684).

Homology modeling

Our model was human CYP21A2, previously generated (22) and available at the Protein Data Bank with ID 2GEG (23). It is based on the mammalian x-ray crystallographic cytochrome CYP2C5 (PDB code 1N6B), which shares 32% sequence identity and 50% sequence similarity. We determined the protein domains (heme group binding, steroid binding, and redox partner interaction) and studied new mutation effects in silico, with the help of the DeepView program (http://www.expasy.org/spdbv) (24).

Results

Molecular analysis

Whole CYP21A2 gene sequencing identified the new mutation p.H62L, either isolated (female 1 and male 9) or associated on the same allele with a mild mutation (11 of 13 patients), usually p.P453S (females 3–8 and males 10–12) but also p.P30L (male 13) and partial promoter conversion of CYP21A2 (−113G>A, −110T>C, −103A>G) (female 2), the usually associated −126C>T being absent in this case.

Phenotype severity

Enzymatic impairment was assessed by neonatal screening detection, basal and post-ACTH 17OHP levels, mineralocorticoid defect attested by renin level, and androgen levels.

Only three patients were born after neonatal screening was introduced; all bore the p.H62L+p.P453S allele, and the only one detected on screening was born preterm. Basal 17OHP was significantly higher for the 10 patients with p.H62L+p.P453S or p.P30L (from 115-1678 nmol/liter; mean = 494.7 nmol/liter) than for the female with p.H62L+partial promoter (99 nmol/liter) or for the two isolated p.H62L patients (from 13–69 nmol/liter; mean = 41 nmol/liter). Post-ACTH levels were also higher, from 345-1809 nmol/liter (mean = 875 nmol/liter) for patients with p.H62L+p.P453S or p.P30L and from 108–116 nmol/liter (mean = 112 nmol/liter) with isolated p.H62L. Mineralocorticoid function remained difficult to assess; renin levels were not available for patients born 20 yr ago (females 1–4 and males 10–11) or measured under hydrocortisone (female 6). Levels were elevated only in female 8, bearing p.H62L+p.P453S. Androgen levels were higher for the 10 patients with p.H62L+p.P453S than for the other three, taking into account age and sex (25).

Hyperandrogenism was assessed by external genitalia virilization, age at symptom onset and diagnosis, bone vs. chronological age at diagnosis, and final height (indicators of disease evolution).

All six females bearing p.H62L+p.P453S presented with external genitalia virilization, in contrast to the two with isolated p.H62L or p.H62L+promoter. Two had presented with considerable virilization at birth: Prader 2 for female 7 (not requiring surgery) and Prader 3 for female 5 diagnosed at puberty and corrected by surgery. Virilization was less severe in four females (isolated clitoromegaly), although female 6 asked for repair surgery.

First symptoms appeared during childhood for the six females with p.H62L+p.P453S and after puberty for the other two (p.H62L and p.H62L+partial promoter). Advanced childhood bone age was significant for the six females, leading to final height below target; the two diagnosed in adulthood presented with NC symptoms (hirsutism and dysmenorrhea), and their final height was more reasonable. All four males with associated p.H62L presented with first signs early in childhood (1–5 yr), whereas male 9, bearing isolated p.H62L, became symptomatic after age 5 yr; likewise, bone age was more advanced in the four males, leading to decreased final height, whereas growth was well controlled by hydrocortisone for male 9. Male 12 presented with very severe hyperandrogenism, with 8 cm phallic growth at age 4.3 yr.

Patients with isolated p.H62L were under hydrocortisone, and the female with p.H62L+partial promoter under Androcur; fludrocortisone was added for four of the 10 with associated p.H62L. Treatment was initiated as of diagnosis for all patients except female 6, treated 7 yr later.

In vitro determination of mutation severity

Control transfections using CYP21A2 cDNA sequence showed 100% conversion of progesterone and 17-OHP to DOC and 11-deoxycortisol, respectively, demonstrating expected WT enzyme 21-hydroxylase activity. Transfection with vector alone did not convert the substrate (Fig. 1).

Fig. 1.

Functional study of the normal protein (WT) and the different mutant human P450c21 (P453S, H62L, and H62L+P453S). Autoradiography of thin layer chromatography showing conversion of 17-OHP to 11-deoxycortisol (S) (A and C) and progesterone (P) to DOC (D and F). Cells were transfected with the different vectors pCMV4 (vector without insert), WT (wild-type), and mutated CYP21A2 cDNAs [P453S, H62L, H+P (H62L+P453S), I172N, and V281L] and incubated with 1 μCi substrates for 2 h (A and D) and 8 h (C and F). CYP 21 in vitro activity columns represent the relative activities of CYP21 mutants after 2 h of incubation, WT being defined as 100% (B and E).

CYP21A2 cDNA constructs containing either p.P453S or p.H62L encoded enzymes catalyzing synthesis of both 11-deoxycortisol and DOC, with moderate efficiency (81.5 and 29.2% for 17-OHP to 11-deoxycortisol, respectively, and 63 and 66.5% for conversion of progesterone to DOC); this activity is similar to that of p.V281L (30.4% for 17-OHP and 50.5% for progesterone). When both p.P453S and p.H62L were introduced in the same construct, the encoded enzyme displayed more severely decreased activity than with the two isolated mutants (11.1% for 17-OHP and 31.1% for progesterone) but less severely than the construct with the p.I172N mutant (2.3% for 17-OHP and 7.8% for progesterone).

Kinetic analysis of the mutants revealed K_m values in the same range, whereas maximal velocity was lower than WT (Fig. 2).

Fig. 2.

A–D, Kinetic analyses of normal and mutant vectors P453S, H62L, H+P (H62L+P453S). A and B, Plots of velocity (V) (picomoles per milligram total protein per minute) against substrate concentration (micromoles per liter): conversion of 17-OHP to 11-deoxycortisol (A) and of progesterone (Prog) to DOC (B). C and D, Corresponding Lineweaver-Burk plots of enzymatic activity (C and D). E, Table of K_m and V_max values of normal (WT) and mutant variants [P453S, H62L, and (H62L+P453S)].

Western blot analysis detected comparable amounts of P450c21 for the various constructions, suggesting that reduced enzyme activity was not due to a decreased amount of protein (Fig. 3).

Fig. 3.

Immunoblots of CYP21A2 protein expressed in COS-1 cells and collected after harvesting cells directly from the well with a lysis buffer.

Discussion

We reported a new mutation p.H62L in 13 patients, isolated or associated on the same chromosome with a mild mutation. Because 12 patients had a severe CYP21A2 lesion on their other allele (null lesion responsible for SW, IVS2–13A/C>G for SW or SV, and p.I172N for SV), the new mutated allele’s severity was predictable from phenotype.

Clinical and biological analysis indicated that isolated p.H62L was rather responsible for the NC form, whereas the phenotype was more severe, but never for SW, when associated.

The degree of virilization of our eight females bearing allele p.H62L+p.P453S was compared with that of two cohort subgroups: 1050 females with mild mutation (p.V281L, p.P453S) on at least one allele and 230 females either homozygous for p.I172N or compound heterozygous for p.I172N and IVS2–13A/C>G or a null mutation (data not published). Females of the first subgroup were not virilized, except for 4% with isolated clitoromegaly undetected at birth and not requiring surgery. Females of the second group were virilized, except for 6%; 14% of virilized females presented with isolated clitoromegaly, 50% with associated posterior labial fusion (Prader 2 or 3), and only 30% with more severe virilization (stage 4–5).

According to in vitro studies, isolated p.H62L is responsible for the NC form, and the association p.H62L+p.P453S for a more severe form, intermediate between NC represented by p.V281L and SV form represented by p.I172N (Fig. 1).

The p.H62L+p.P30L allele in patient 13 also seemed more severe than isolated p.P30L in terms of bone age advancement and very elevated 17-OHP, suggesting synergy. However, p.P30L phenotypic heterogeneity was reported in eight patients, five with the SV and three with the NC form (12). Nevertheless, according to in vitro studies, p.P30L led to the same residual activity as p.V281L and p.P453S (26). After this previous study, we reported that p.P30L severity depends on the presence or absence of a 5′ gene conversion of the CYP21A2 gene to CYP21A1P, not including the IVS2–13A/C>G (27).

In contrast, the allele p.H62L+promoter of CYP21A2 is less severe (female 2), probably because of the presence of only three of the four mutations usually described.

The H62 residue is located in the β1-sheet region, in a large hydrophobic area considered important for membrane anchoring. Three hydrophobic patches are important to cytochrome structure, including amino acids 60–69 of the CYP2C5 cytochrome corresponding to 54–63 in CYP21A2. The basic residues of this patch operate as a halt transfer signal, preventing the remainder of the protein translocating through the endoplasmic reticulum after the cotranslational insertion of the leader sequence into the membrane (28). Substituting basic histidine by hydrophobic leucine should increase protein immersion in the membrane and thus disturb enzymatic function. The P453 residue, located in the C-terminal part of the protein, distant from steroid and heme binding (Fig. 4B), is well conserved between 21-hydroxylases in various species but less in other human steroidogenic P450 cytochromes and other mammalian nonsteroidogenic cytochromes (Fig. 4A). Replacing proline by polar serine should increase the protein’s structural flexibility and thus decrease its stability. The association of the two mutations should have a synergistic effect, decreasing exposure to the active site of the endoplasmic reticulum and protein stability.

Fig. 4.

A, Multiple amino acids alignments of human CYP21A2 with other mammalian CYP21A2s, mammalian crystallized CYPs, or human steroidogenic cytochromes. Sequence alignment was performed with CLUSTAL W and edited with Genedoc. The dark-shaded regions represent the amino acids conserved across all P450s in alignment and lighter shades represent lesser levels of similarity according to Genedoc. The mutant residues are indicated with an arrow in human CYP21A2. B–D, Molecular modeling of the human P450c21; B, total view of the three-dimensional model structure of CYP21A2; localization of progesterone (green) and heme (red); C, localization of the H62 in the N-terminal part of the protein, in a hydrophobic patch probably involved in the anchoring of the cytochrome into the membrane; D, localization of the P453 in the C-terminal part of the protein distant from the substrate and the heme disturbing the protein function by another mechanism.

The association of p.P453S with another new mild mutation has already been described. Association with p.P105L was studied in vitro, revealing a synergistic inhibitory effect with enzyme activity reduced to 10% for 17OHP and 7% for progesterone (15). Association with mild mutation p.R339H was described in a woman presenting severe hirsutism and clitoromegaly (14), but the combined effect of the two mutations has never been studied, in vitro or in silico.

In conclusion, p.H62L is a mild new mutation responsible for the NC form when isolated or associated with partial mutation of the promoter. In contrast, its association with mild p.P453S or p.P30L is responsible for a more severe phenotype, intermediate between the NC form associated with mild p.V281L and the SV form associated with severe p.I172N. This study underlines the importance of complete CYP21A2 sequencing for accurate genotyping and genetic counseling. The risk of external genitalia virilization in females, due to fetal androgen excess, has to be considered; as with the SV form associated with p.I172N, genetic counseling should be proposed to parents at risk of having a child with such a genotype. Fetal sex should be determined in maternal serum around 4–5 wk gestation (17) and prenatal treatment by dexamethasone and prenatal diagnosis discussed only for female fetuses. This more severe phenotype, not always detected by neonatal screening, should be diagnosed as soon as the first signs appear, especially in males, to limit deleterious effects. In both sexes, adapted treatment should be proposed, based on complete hormonal balance of the gluco- and mineralocorticoid pathways.

Acknowledgments

We express our sincere gratitude to our clinician colleagues, who have sent samples with clinical and biological data of the patients. We are grateful to Dr. Bon Chu Chung for providing cDNA vector, to Dr. W. L. Miller for providing the antihuman CYP21A2 rabbit polyclonal antiserum. We also thank Maryline Murena, Muriel Manigand, and Audrey Favre for their technical assistance, Dr. Jacques Chomilier for bioinformatics assistance, and Iain McGill for linguistic help with the manuscript.

Disclosure Statement: The authors have nothing to disclose.

Abbreviations

 
• CAH,

  Congenital adrenal hyperplasia;

 
• DOC,

  11-deoxycorticosterone;

 
• NC,

  nonclassical;

 
• 21OHD,

  steroid 21-hydroxylase deficiency;

 
• 17OHP,

  17-hydroxyprogesterone;

 
• PDB,

  Protein Data Bank;

 
• SV,

  simple virilizing;

 
• SW,

  salt-wasting;

 
• WT,

  wild-type.

Author notes

^*

R.M. and V.T. contributed equally to this work and should be considered as equal first authors.