The Common P450 Oxidoreductase Variant A503V Is Not a Modifier Gene for 21-Hydroxylase Deficiency

Abstract

Context: 21-hydroxylase deficiency (21OHD) is a common genetic disorder caused by mutations in the CYP21A2 gene, which encodes the adrenal 21-hydroxylase, microsomal P450c21. CYP21A2 gene mutations generally correlate well with impaired P450c21 enzymatic activity and the clinical findings in 21OHD, but occasional discrepancies between genotype and phenotype suggest the effects of modifier genes. Mutations in P450 oxidoreductase (POR), the protein that transfers electrons from reduced nicotinamide adenine dinucleotide phosphate to all microsomal P450s, can ameliorate the 21OHD phenotype and, therefore, could be a modifier gene.

Objectives: We sought to identify POR variants in patients with 21OHD having discordant phenotype and genotype, and to evaluate their effect on 21-hydroxylase activity.

Patients and Methods: We determined the CYP21A2 genotypes of 313 Brazilian patients with 21OHD and correlated the genotype and phenotype. The POR gene was sequenced in 17 patients with discordant genotype and phenotype. Wild-type and A503V POR, and P450c21 were expressed in bacteria and reconstituted in vitro. Activities were assayed by conversion of [¹⁴C]progesterone to deoxycorticosterone and [³H]17-hydroxyprogesterone to 11-deoxycortisol, and assessed by thin layer chromatography and phosphorimaging.

Results: The A503V POR variant was found in 10 of 30 alleles, the same ratio as in the normal population. There were no significant differences in Michaelis constant, maximum velocity and maximum velocity/Michaelis constant of 21-hydroxylase activity supported by wild-type and A503V POR.

Conclusion: The only POR missense polymorphism found in atypical 21OHD patients was A503V. Although A503V reduces P450c17 enzymatic activity, it does not influence P450c21 activity, indicating that POR A503V does not modify the 21OHD phenotype.

Although there is generally a good correlation between 21-hydroxylase mutations and the congenital adrenal hyperplasia (CAH) phenotype, some patients have phenotype/genotype discordance. The common P450 oxidoreductase allelic variant A503V reduces 17-hydroxylase activity but does not affect 21- hydroxylase activity, hence it is not a modifier gene for CAH.

21-hydroxylase deficiency (21OHD) is caused by mutations in the CYP21A2 gene encoding microsomal P450c21, which converts progesterone to deoxycorticosterone (DOC) and 17-hydroxyprogesterone (17OHP) to 11-deoxycortisol (1,2). The resulting low concentrations of cortisol in 21OHD stimulates the pituitary to increase ACTH secretion, which stimulates synthesis of adrenal androgen precursors. Thus, the clinical manifestations in 21OHD are related to the inability to synthesize cortisol and aldosterone efficiently, and the excessive androgen synthesis.

The degree to which CYP21A2 gene mutations compromise enzymatic activity correlates well with clinical manifestations (3,4,5). In salt wasting (SW) 21OHD, the impairment in 21-hydroxylation is severe, affecting both cortisol and aldosterone synthesis. In simple virilizing (SV) 21OHD, aldosterone is minimally affected. In both of these forms, the increased production of adrenal androgen precursors results in genital virilization in newborn females. In nonclassical (NC) 21OHD, patients can be asymptomatic or have mild virilization in adolescence (3,4,5). However, discrepancies between genotype and phenotype have been described (3,4,5,6). In some patients with 21OHD, mutations were not found in one or both alleles after complete CYP21A2 sequencing (3,4,5,7). These apparent discrepancies between genotype and phenotype suggest the presence of other genetic factors.

P450 oxidoreductase (POR) can modify the phenotype of 21OHD (8). POR is a flavoprotein that transfers electrons from reduced nicotinamide adenine dinucleotide phosphate (NADPH) to all microsomal P450s, including P450c21 (9). Congenital adrenal hyperplasia due to POR gene mutations can cause apparent deficiencies of both P450c21 and P450c17. However, since the first description of POR deficiency (10), a wide spectrum of phenotypes has been reported. Most reported cases have a skeletal dysplasia called Antley-Bixler syndrome and also have ambiguous genitalia in both sexes (11); some less severely affected patients only have mildly disordered steroidogenesis (10,12). Because POR is the only protein that can donate electrons to P450c21, we sought to determine whether POR sequence variants were found in 21OHD patients with discordant genotype/phenotype correlations, and to test the activity of such POR variants in vitro.

Subjects and Methods

Subjects

A group of 313 Brazilian patients with 21OHD, 130 of whom were previously reported (4), was diagnosed by clinical history, hormonal data, and genetic analysis. The study protocol was approved by the Ethical Committee of São Paulo University and informed consent obtained from the patients or caretakers. SV 21OHD was characterized by ambiguous genitalia in girls and signs of early postnatal virilization in both sexes. SW 21OHD was characterized by adrenal crisis or hyponatremia with high renin activities in the first month of life. NC 21OHD was diagnosed in adolescents and adults with signs of hyperandrogenism, and 17OHP levels greater than 51 nmol/liter after an ACTH test. This cutoff for 17OHP level is higher than that found in obligate heterozygotes for severe 21OHD (13).

Hormonal assays

RIA for 17OHP was performed using commercial reagents (Diagnostic Systems Laboratories, Inc., Webster, TX); the intraassay and interassay coefficients of variation were 12 and 18%, respectively.

Genetic analysis

The CYP21A2 gene was characterized by Southern blotting studies to determine large rearrangements (14) and by allele-specific PCR to determine 10 common micro-conversions (4) in all 313 patients. Direct sequencing of the entire CYP21A2 gene was performed in patients with discordant genotype/phenotype or in whom PCR detected no mutations, or mutations on only one allele. The 16 exons of the POR gene were sequenced as described (15).

Expression of P450c21

To facilitate expression in bacteria, the N-terminal region of human P450c21 cDNA was changed to MALLLAVFL by site-direct mutagenesis, and a 6-His tag was added to the C terminus to facilitate purification. This construct (built by Dr. Christa E. Flück) was subcloned in pCWori and expressed in Escherichia coli C41(DE3)pLysS as described (16), except that 34 μg/ml chloramphenicol and 1 mm thiamine were added to the media. The culture was shaken at 28 C for 36 h, bacterial membranes were prepared as described (16), resuspended in 4 ml 20 mm potassium phosphate (pH 7.4), 20% glycerol, 0.25 mm EDTA, 1% sodium cholate and 1% Tween 20, shaken for 2 h, and centrifuged at 150000 g for 90 min. The supernatant was applied to a Ni-NTA agarose column (Sigma-Aldrich, St. Louis, MO) equilibrated with 0.3 m NaCl, 20% glycerol, 0.1 mm dithiothreitol, 0.1 mm EDTA, 1% sodium cholate, and 1% Tween 20 in 50 mm potassium phosphate buffer (pH 8.0). The column was washed with the same buffer containing 5 mm imidazole and eluted with the same buffer containing 250 mm imidazole. The eluate was sequentially applied to DEAE-Sepharose and to SP-Sepharose columns (Sigma-Aldrich). The purified protein was concentrated by Amicon Ultra-15 (Millipore, Billerica, MA). Purity was assessed by Coomassie blue staining of an SDS-PAGE gel and confirmed by immunoblotting.

Expression of POR

Wild-type and A503V human POR cDNA lacking codons for 27 N-terminal residues were subcloned in pET22b, expressed in E. coli C41(DE3)pLysS, and membranes were prepared, as described (11). POR proteins were quantified by Western blotting in comparison to a standard curve of purified wild-type POR (17). POR proteins were separated by SDS-PAGE, transferred to polyvinyldifluoride membrane, incubated with rabbit polyclonal antibody against rat POR (Stressgen, Ann Arbor, MI), and assessed with the Odyssey Infrared Imaging System (LI-COR, Lincoln, NE).

Enzyme assays

Purified P450c21 (5 pmol) was incubated with bacterial membranes containing 10 pmol POR, 10 μg 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC), 10 μg 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), 50 mm potassium phosphate (pH 7.4), 10 mm MgCl₂, 6 mm KOAc, 1 mm reduced glutathione, and radiolabeled substrates ([¹⁴C]progesterone or [³H]17OHP), in a total volume of 200 μl. The reactions, performed with substrates at concentrations of 0.3, 1, 3, and 5 μm, were started by adding 2 mm reduced nicotinamide adenine dinucleotide phosphate (NADPH). After 1 h at 37 C, reactions were stopped with 1:1 ethyl-acetate/isooctane. Steroids were analyzed by thin layer chromatography using two different solvent systems (18). The conversion of [¹⁴C]progesterone to DOC and [³H]17OHP to 11-deoxycortisol was quantified by phosphorimaging. Maximum velocity (V_max), Michaelis constant (Km), and the V_max/Km (an estimative of enzymatic efficiency) were calculated by Lineweaver-Burk plots using GraphPad Prism 3 software (GraphPad Software Inc., San Diego, CA).

Results

Patients

From 313 Brazilian patients with 21OHD, we selected 17 who had discordances between genotype and phenotype (Table 1). Patients 1–6 had a phenotype milder than predicted by the genotype. We found no CYP21A2 mutations on either allele in patients 8 and 12, and we found mutations on only one allele in the other nine patients. Patients 1 and 3–6 had CYP21A2 mutations that completely abolish 21-hydroxylase activity; all had high plasma rennin activities but were not salt wasters as newborns and did not require mineralocorticoid replacement therapy. Patient 2, an asymptomatic woman without virilization, was investigated because she is a member of a consanguineous family with SV 21OHD; she had P30L on one allele and IVS2 -2 A>G on the other allele. Patients 9, 13, 14, and 17 had precocious pubarche and very high levels of 17OHP after ACTH stimulation. Patients 7, 8–12, 15, and 16 are women with hirsutism, menstrual irregularities, and grossly elevated 17OHP after ACTH stimulation. Patients 13/14 and 15/16 are pairs of siblings who had mutations on only one allele. In calculating allele frequencies, each sibling pair is counted as two alleles.

Table 1.

Clinical, hormonal, and genetic data in 21OHD patients studied

Patient No.	Sex	Age at diagnosis (yr)	Basal/stimulated 17OHP (nmol/liter)	21OHD genotypec	21OHD phenotype predicted/observed	POR genotyped
1	Fa	4.5	915.1	R356W/R356W, I2	SW/SV	AV
2	F	35	248.5	Conv + P30L/IVS2 − 2 A>G	SV/NC	AA
3	M	4.5	163.6	Del/Q318X	SW/SV	AV
4	M	5.3	581.8	Del/Del	SW/SV	AV
5	M	1	721.1	R356W/Ins T, V281L	SW/SV	AV
6	M	5	>61	Q318X/R408C	SW/SV	AV
7	F	34	9.9/96	I172N/−		AA
8	F	12	9.3/78	−/−		AV
9	M	8	17.1/108	V281L/−		AA
10	F	20	10.5/132	V281L/−		AA
11	F	20	33.3/445	I2/−		AV
12	F	63	3.3/90	−/−		AA
13b	F	8	151.2/251	V281L/−		VV
14b	F	8	176.7/249	V281L/−		VV
15b	F	25	51/81	V281L/−		AV
16b	F	22	54/66	V281L/−		AV
17	F	6	20.7/195	V281L/−		AA

F, female; M, male. 

^aRaised as a male. 

^bSiblings (13 and 14; 15 and 16). 

^cDel, CYP21A2 deletion; I2, IV2-13 A/C>G. 

^dAA, Wild type; AV heterozygous for A503V; VV homozygous for A503V. 

Genetic variations in POR

Sequencing the POR gene in the 17 patients revealed one polymorphism in exon 1U at position −47A>C and four polymorphisms that do not change the POR amino acid sequence: 15 A>G (G5G) (rs10262966), 387 A>G (P129P) (rs1135612), 1455 C>T (A485A) (rs2228104), and 1716 G>A (S572S) (rs1057870). These polymorphisms are found in the Caucasian and African-American populations (15), and have no known impact in steroidogenesis. The only POR amino acid sequence variant found was A503V, present in 10 of 30 alleles analyzed. There were no clinical, hormonal, or CYP21A2 genetic data that differentiate the patients with the wild-type or A503V POR.

Effect of POR A503V on P450c21 activity

To determine whether A503V POR affects 21-hydroxylation, we assayed its ability to support catalysis by P450c21 in vitro. Bacterially expressed wild-type and A503V POR were combined with purified, bacterially-expressed P450c21. The assays measured conversion of [¹⁴C]progesterone to DOC and [H³]17OHP to 11-deoxycortisol (Table 2). The enzymatic parameters with A503V and wild-type POR were essentially the same: V_max/Km for 21-hydroxylation of progesterone and 17OHP supported by A503V POR was 80 and 95%, respectively, of control POR. There was no difference in the efficiency of 21-hydroxylase supported by wild-type or A503V POR (P = 0.873 for progesterone assays; P = 0.836 for 17OHP assays).

Table 2.

Kinetics of 21-hydroxylation supported by wild-type and A503V POR

	Wild type	A503V	P value
Substrate: progesterone
Km (μm)	2.65 ± 0.37	2.47 ± 0.49	0.814
Vmax (pmol/min/pmol P450)	0.26 ± 0.06	0.21 ± 0.07	0.330
Vmax/Km	0.1 ± 0.03	0.08 ± 0.01	0.873
Vmax/Km (% wild type)	100	80
Substrate: 17OHP
Km (μm)	1.43 ± 0.4	1.5 ± 0.17	0.836
Vmax (pmol/min/pmol P450)	0.28 ± 0.08	0.28 ± 0.09	0.965
Vmax/Km	0.2 ± 0.06	0.19 ± 0.05	0.836
Vmax/Km (% wild type)	100	95

Unless stated, data are mean ± sd of four independent experiments, each performed in duplicate. 

Discussion

We found discordance between genotype and phenotype in 17 of 313 Brazilian patients, suggesting the action of other factors modifying the 21OHD phenotype. Because severe POR mutations may ameliorate the phenotype of 21OHD, we examined POR as a potential modifier gene in our 17 patients who had a discordance between their 21OHD genotype and phenotype. The only POR amino acid sequence variant found was A503V, present in 10 of 30 alleles. A503V is common, found in 26.4% of Caucasian alleles, 19.1% of African-American alleles, 36.7% of Asian alleles, and 31% of Mexican-American alleles (15). A503V POR diminishes the reduction of cytochrome c to 67% of the wild-type POR (11,15), and reduces the 17α-hydroxylase and 17,20 lyase activities of P450c17 to 68 and 58%, respectively (11). However, we found that wild-type and A503V POR have equivalent abilities to support the 21-hydroxylation of progesterone and 17OHP by P450c21. Therefore, A503V POR does not appear to be a factor that modulates the phenotype on 21OHD.

We have shown that the activity of a POR variant with one electron recipient may not predict its activity with another recipient (10,11,15,17), therefore, the normal activity of A503V POR with P450c21, whereas it has reduced activity with P450c17, is consistent with other studies. The disease-causing POR mutant A287P has different activities to support the activities of P450c17 and P450c21 (19), and an extensive analysis of 35 POR sequence variants revealed that some mutants had dramatic differences in their abilities to support catalysis by P450c17 and by hepatic, drug-metabolizing CYP1A2 and CYP2C19 (17). Modeling studies indicate that such variation primarily occurs with POR mutants located in the domain that interacts with the P450 enzyme (20). These observations help to explain the phenotypic variability in patients with POR deficiency and indicate that each particular POR mutation must be tested separately with each particular P450s of interest.

Although our study did not find POR variations that could explain discordances between the 21OHD phenotype and genotype, our study consisted of a small, specific population; we cannot exclude POR variants as a factor modulating the 21OHD phenotype in other populations. The POR gene is substantially more polymorphic than most human genes, and some variants are more common in particular ethnic groups (15). Other genes involved in androgen sensitivity, salt balance, or extra-adrenal 21-hydroxylase activity may account for variations in the 21OHD phenotype.