Abstract The syndrome of apparent mineralocorticoid excess is a form of hypertension inherited in an autosomal recessive manner. This disorder results from mutations in the HSD11K (HSD11B2) gene, which encodes the kidney isozyme of 11β-hydroxysteroid dehydrogenase. This enzyme converts active glucocorticoids such as cortisol and corticosterone to their inactive metabolites cortisone and 11-dehydrocorticosterone. An elevated ratio of cortisol to cortisone metabolites in the urine (tetrahydrocortisol plus allotetrahydrocortisol to tetrahydrocortisone [(THF+aTHF)/THE]) is considered pathognomic for this disorder. To determine whether the biochemical phenotype of this disorder is correlated with genotype, we expressed enzymes carrying each of the six known missense mutations in cultured cells. Only one mutant, R337C, had detectable activity in cell lysates, but five of six mutants were partially active in whole cells. Apparent kinetic constants for cortisol and corticosterone were determined in whole cells, and the apparent first-order rate constant, Vmax/Km, was used as a measure of enzymatic activity. The urinary (THF+aTHF)/THE ratio in patients carrying each mutation was strongly correlated with in vitro enzymatic activity of the corresponding mutant (r=.839, P=.001 with cortisol as the substrate). We conclude that the biochemical phenotype of the syndrome of apparent mineralocorticoid excess is largely determined by genotype.
- hypertension, genetic
- hypertension, endocrine
- molecular biology
Apparent mineralocorticoid excess (AME) is a form of hypertension inherited in an autosomal recessive manner. It is characterized by signs of mineralocorticoid excess such as hypokalemia and suppressed renin activity although levels of aldosterone and other mineralocorticoids are very low. Biochemical and molecular genetic studies have demonstrated that this disorder results from deficiency of renal 11-HSD, an enzyme that converts active glucocorticoids such as cortisol and corticosterone to their inactive metabolites cortisone and 11-dehydrocorticosterone. Patients with AME excrete reduced amounts of cortisone metabolites such as THE in the urine compared with excretion of cortisol metabolites such as THF and aTHF. An elevated (THF+aTHF)/THE ratio is considered pathognomic for AME.1
There are two known isozymes of 11-HSD. The L or type 1 isozyme is expressed at high levels in the liver, uses NADP+ as an electron acceptor, and is reversible, catalyzing both dehydrogenation and reduction (conversion of cortisone to cortisol).2 3 The corresponding gene is unaffected in patients with AME.4 The K or type 2 isozyme is expressed at high levels in the kidney and placenta, uses NAD+, and catalyzes only dehydrogenation.5 6 The gene encoding this isozyme, HSD11K (or HSD11B2), is located on chromosome 16q22.7 Mutations in this gene have been documented in patients with AME8 9 10 ; most of these mutations have significant effects on enzymatic activity or mRNA splicing.8
Deficiency of 11-HSD causes hypertension because the mineralocorticoid receptor itself has identical affinities for glucocorticoids and mineralocorticoids.11 Under normal circumstances, 11-HSD expressed in renal distal tubules and cortical collecting ducts oxidizes cortisol to cortisone, which is not a ligand for the receptor. Aldosterone is not a substrate of this enzyme and is thus a much stronger mineralocorticoid agonist in vivo than cortisol.12 13 This protective mechanism is lost in patients with AME, so that cortisol, which normally circulates in amounts 100 to 1000 times higher than aldosterone, itself becomes a strong mineralocorticoid agonist. Thus, this disorder is best treated by measures such as a low sodium diet and blockade of the mineralocorticoid receptor with spironolactone.
Most inherited diseases vary in severity in a large part because of allelism at the affected locus. Although our previous study suggested that this might be the case in AME,8 it was difficult to make statistically significant correlations of genotype with phenotype because of the small number of patients and low levels of in vitro enzymatic expression. Therefore, we have now examined all known missense mutations using a cell line that permits higher levels of expression and have found that the degree of enzymatic compromise is indeed correlated with the biochemical severity of the disorder as measured by ratios of precursor to product.
Full-length human HSD11K cDNA was cloned in the expression vector pCMV514 (pCMV5-HSD) and mutagenized [R208C, R213C, R337Δ3nt (R337H/ΔY338), and L250P/L251S] as previously described.8 Additional mutations (R186C and R337C) were created by PCR15 with primers containing each desired mutation (GGCCACTTTCTGTAGCTGCAT, nucleotides 654-674 for R186C; and GCCCGCCGCTGCTATTACCCC, nucleotides 1107-1128 for R337C). After subcloning into pCR II vector (Invitrogen Co), each mutant insert was introduced into the Sfi I site of pCMV5-HSD. The complete coding sequence of each construct was checked to ensure that no undesirable mutations had been introduced by the PCR process.
Transient Transfection and Assays of Enzymatic Activity
CHOP cells, a derivative of Chinese hamster ovary (CHO) cells that has been transformed with polyoma virus (kindly provided by Dr James Dennis),16 were grown in Ham’s F-12 medium (GIBCO) supplemented with 10% fetal bovine serum. Five micrograms of each construct was transfected into 80% to 90% confluent CHOP cells in 60-mm dishes by the modified bovine serum system (Stratagene), and enzymatic activity was assayed in whole cells or cell lysates after 20 hours.
For determination of apparent kinetics in intact mammalian cells, 0.9 to 1.2 nmol/L [3H]corticosterone with 3 to 1000 nmol/L unlabeled corticosterone or 4.4 to 7 nmol/L [3H]cortisol with 15.6 to 2000 nmol/L unlabeled cortisol was added. After 0.5 to 42 hours of incubation, culture medium was removed and extracted with methylene chloride. The conversion from corticosterone or cortisol to 11-dehydrocorticosterone or cortisone, respectively, was detected with a System 200 Imaging Scanner (Bioscan) after extracts had been subjected to silica thin-layer chromatography with a solvent mixture of chloroform/ethanol (92:8). Three independent experiments on different days were performed for the wild-type enzyme and R337C with cortisol as the substrate, and two such experiments for the other mutants. Measurements with corticosterone as a substrate were obtained in an independent transfection.
For assay of activity in cell lysates, the transfected cells were washed twice with phosphate-buffered saline, harvested, and kept frozen at −80°C for later processing. Cells were thawed and homogenized in 800 μL per dish (for wild-type) or 400 μL per dish (for R337C) of homogenizing buffer (50 mmol/L Tris-HCl, 0.25 mol/L sucrose, 2 mmol/L EDTA, 1 mmol/L MgCl2, 1 mmol/L dithiothreitol, pH 7.4) with a Dounce homogenizer. After brief sonication, lysates were centrifuged at 800g for 10 minutes to separate nuclei. Kinetic constants were determined by incubation of 50 μL supernatant for 1 hour at 37°C in 500 μL assay buffer (50 mmol/L Tris-HCl, 2 mmol/L EDTA, 1 mmol/L MgCl2, 1 mmol/L dithiothreitol, 20% glycerol, 0.4 mmol/L NAD+, pH 7.4) with 2.4 or 5.3 nmol/L 3H-labeled steroid and 1 to 100 nmol/L or 10 to 200 nmol/L unlabeled steroid (corticosterone or cortisol, respectively). Two independent experiments were performed for each substrate.
In all cases, incubations were carried out for different periods of time to ensure that the measurements represented initial rates.
Detection of Transcripts by Reverse Transcription–PCR
Total RNA was prepared from transfected CHOP cells with RNAzol B (Biotecx). One microgram total RNA was treated with DNase I (GIBCO-BRL) and reverse transcribed with 50 U Superscript II reverse transcriptase (GIBCO-BRL) in a 10-μL reaction volume containing 2.5 mmol/L hexamers, 1 mmol/L of each dNTP, 4 U placental RNase inhibitor, 2.5 mmol/L of a 3′-primer (GCAGCCAGGCTGGATGATG, complementary to nucleotides 882-900), and the manufacturer’s buffer. Each reaction was allowed to proceed at room temperature for 10 minutes followed by incubation at 37°C for 1.5 hour. PCR was performed by addition of 0.5 mmol/L of a 5′-primer (ACCAAGGCCCACACCACCAGC, nucleotides 562-582) to 2 μL of reverse-transcribed samples in 67 mmol/L Tris-HCl (pH 8.5), 16 mmol/L (NH4)2SO4, 2 mmol/L MgCl2, 17 μg/mL bovine serum albumin, and 5% glycerol. Samples were subjected to initial denaturation at 96°C for 1 minute and 40 cycles of 96°C for 10 seconds, and 59°C for 30 seconds.
Nuclear and 800g supernatant fractions from transfected CHOP cells were subjected to immunoblot analysis. Nuclei were washed twice with homogenizing buffer before being dissolved in sample buffer containing 1% sodium dodecyl sulfate. Aliquots of lysates (protein amounts are indicated in figure legends) were subjected to electrophoresis in a sodium dodecyl sulfate–10% polyacrylamide gel. An anti-peptide serum highly specific for human HSD11K residues 370-383 (HCLPRALQPGQPGT) was kindly provided by Drs Jonathan R. Seckl and Roger Brown. Immunoblots were performed at a 1/8000 dilution of HSD11K anti-peptide serum as primary antibody with the Amersham ECL system and methods recommended by the manufacturer (including use of donkey anti-rabbit serum at 1/4000 dilution as the secondary antibody).
Kinetic constants were derived from two (whole cells) or three (lysates) measurements of enzymatic activity at each of five to eight different substrate concentrations. The StatView 4.0 program was used for statistical analysis including regression coefficients.
Enzymatic Activity in Whole Cells
In the present study, we made use of CHOP cells, a derivative of CHO cells transformed with polyoma virus. When these cells are transfected with vectors such as pCMV5 that contain cytomegalovirus promoters, expression levels are several times higher than can be achieved in CHO cells (Reference 1616 and unpublished observations, 1995). We assayed enzymatic activity of the wild-type enzyme and of six previously reported mutants8 9 10 (Table 1⇓). We used reverse transcription–PCR to ensure that approximately equal levels of mRNA were expressed in the various transfectants. Four missense mutations (R186C, R208C, R213C, R337C) retained measurable activity with cortisol used as a substrate. Both a double missense mutation (L250P/L251S) and a three-nucleotide deletion in codons 337-338 that maintained the normal reading frame (R337Δ3nt) abolished activity. Corticosterone, although not the most physiologically important glucocorticoid in humans, is a better substrate for the enzyme, and R337Δ3nt retained measurable activity for this substrate.
By incubating cells with various concentrations of precursor steroids, we could derive apparent kinetic constants; one set of determinations is presented in Figs 1⇓ and 2⇓ and Table 1⇑. With cortisol as a substrate, no mutation affected apparent Km by more than approximately twofold; when corticosterone was the substrate, two mutants, R208C and R213C, had sixfold higher apparent Km values than the wild-type enzyme. Comparing the enzymes by their apparent first-order rate constants (Vmax/Km), mutant R337C retained 82% of normal activity for corticosterone and 53% of normal activity for cortisol. R186C retained 9% and 17% of normal activity for these substrates, respectively. The other mutants all had less than 4% of normal activity.
Enzymatic Activity in Cell Lysates
We attempted to determine kinetic constants in cell lysates to minimize problems inherent with kinetic measurements in whole cells such as uncertain transport of substrates and products. To maximize the stability of normal and mutant enzymes, we conducted assays in buffer containing 20% glycerol and for an incubation time of less than 1 hour. Under these conditions, kinetic constants could be obtained for the wild-type enzyme and for R337C (one set of data is presented in Table 2⇓ and Fig 3⇓); other mutants were essentially inactive. The apparent Km values obtained for the wild-type enzyme were much lower than observed in whole cells but were similar to those previously reported in cell lysates under comparable conditions.17 R337C had considerably lower affinity for both substrates (higher apparent Km values) than the wild-type enzyme but had very similar apparent Vmax values, so apparent first-order rate constants were 37% (corticosterone) and 8% (cortisol) of those of the wild-type enzyme.
Protein Immunoreactivity in Cellular Fractions
To determine whether any of the mutations affected subcellular localization or stability of the enzyme, we measured immunoreactive protein in nuclei and 800g supernatant fractions using an anti-peptide serum generated against the carboxyl terminus of HSD11K (Fig 4⇓). The wild-type enzyme was found in both the nuclear and supernatant fractions (in the latter fraction, presumably in the endoplasmic reticulum), but it was preferentially concentrated in nuclei. Less immunoreactivity was noted in lysates from cells transfected with mutant enzymes in both the nuclear and supernatant fractions. The decreases in immunoreactivity and enzymatic activity were correlated; significant amounts of immunoreactivity could be detected only in cells transfected with the most active mutants, R337C and R186C.
Correlation of Genotype With Biochemical Phenotype
We used the ratio of cortisol to cortisone metabolites in the urine, (THF+aTHF)/THE, as a measure of the biochemical severity of AME (Table 3⇓). These values were all reported previously.8 9 10 This ratio is normally 1.07±0.30 in children aged 4 to 8 years.18 In making these comparisons, we included an additional patient who carried other mutations (a deletion and a frameshift) that were presumed to destroy enzymatic activity.8 In patients with AME, the (THF+aTHF)/THE ratio in each patient was correlated with the degree of impairment of activity observed with the mutation or mutations carried by that patient (Fig 5⇓). The correlation was slightly better when activity was assayed with cortisol as the substrate (r=.839, P=.001) than with corticosterone as the substrate (r=.669, P=.01).
Effects of Mutations on Enzymatic Activity
It was of interest to see whether any relationships between structure and function could be identified by examining the effects of each mutation on enzymatic activity. The K isozyme of 11-HSD, like the L isozyme, is a short-chain alcohol dehydrogenase. A crystal structure has been elucidated for a related enzyme, 3α,20β-hydroxysteroid dehydrogenase19 ; on the basis of these data and mutagenesis of other short-chain dehydrogenases,20 regions have been defined that bind the nucleotide cofactor or are active in catalysis. Thus far, no mutations in HSD11K have been identified in the nucleotide binding domain. A deletion of the catalytic domain was previously reported but it was presumed to yield an inactive enzyme and was not analyzed further.8
Two of the mutations in the present study, R208C and R213C, lie within a region homologous to a probable steroid binding domain in 3α,20β-hydroxysteroid dehydrogenase,19 and it was therefore previously speculated that these mutations might affect affinity of the enzyme for steroids.8 Consistent with this hypothesis, enzymes with these mutations indeed have high apparent Km values for corticosterone when expressed in whole cells. However, apparent Vmax is also affected, and moreover, these mutations do not have a strong effect on the apparent Km for cortisol. Thus, these mutations do not affect steroid substrate binding in a straightforward way.
It is notable that both the normal enzyme and several of the mutants are preferentially localized in the nucleus. This makes teleological sense if the enzyme is to protect the mineralocorticoid receptor from excessive concentrations of cortisol, given that this receptor acts in the nucleus. We imagine that the presence of the enzyme in multiple cellular compartments allows it to act as a multitiered defense system, with most cortisol being oxidized in the endoplasmic reticulum and any cortisol that does get through to the nucleus being destroyed by the nuclear enzyme. It is not known what structural signal directs this enzyme to the nucleus. Nuclear localization signals often consist of a cluster of basic residues (arginine or lysine) in close proximity to a helix-breaking residue such as proline or glycine.21 Such a region in HSD11K is residues 333-337 (RPRRR), and it was therefore reasonable to speculate that the two mutations R337C and R337Δ3nt (R337H/ΔY338) might affect nuclear localization. Such an effect was not observed, and the mechanisms by which these mutations affect activity is not known.
All of the mutations in the present study are associated with decreased intensity on Western blots. The anti-peptide serum used to probe these blots recognizes an epitope that should not be affected by any of the mutations, and any changes in conformation induced by the mutations should be irrelevant in any case, considering that the proteins have been denatured in Western blots. Thus, the difficulty in detecting several of these mutants on Western blots might mean that the mutations alter the stability or cellular localization of the enzyme. Altered stability could occur in intact cells and be at least partly responsible for the decreased enzymatic activity observed for these mutants. Alternatively, it could be an artifact of cellular homogenization. The latter possibility is supported by the decreased enzymatic activity of all of the mutants (relative to the wild-type) in cell lysates compared with their activity in whole cells.
Comparisons With Previous Reports
We previously reported that several of these mutations affected enzymatic activity in whole cells.8 Expression at higher levels, permitting more quantitative analysis, has yielded similar results in terms of the degree to which each mutation impairs activity. Two of the mutations in the present study were first reported by other researchers. One, R186C, was found in a kindred in which we were previously unable to identify a mutation.10 We have reanalyzed this kindred using new aliquots of DNA and have confirmed the presence of this mutation (unpublished observations, 1995). We suspect that the previous sample we analyzed was contaminated at some point during the PCR process despite our routine use of measures to prevent such a problem. All positive identifications of mutations in our previous study were replicated on fresh aliquots of DNA before publication, so we therefore have no reason to suspect any other inaccuracies in our previous report. In any event, we now confirm that R186C does have significant effects on enzymatic activity. Thus, mutations have been detected in all AME patients analyzed thus far, and currently there is no evidence for genetic heterogeneity of this syndrome.
R337C was first reported in a rapid publication in which analysis of only one exon of the gene was presented.9 Thus, the possibility of other mutations in the gene has not been eliminated. An expression study of this mutation using the identical system to that employed in the present article was recently presented in a second rapid publication.22 In whole cells, this mutation was reported to increase apparent Km for cortisol more than 10-fold, and it was also reported to abolish enzymatic activity in cell lysates. These results are obviously somewhat different from those in the present study. However, examination of the data in this previous publication suggests some possible explanations for the discrepancies. First, although R337C seemed to have a higher apparent Km for cortisol when analyzed in whole cells, it also had a much higher apparent Vmax, so the apparent first-order rate constant, Vmax/Km, of the mutant for cortisol was approximately 50% that of the wild-type enzyme. This value is in excellent agreement with that obtained in the present study. Second, the expression in cell lysates in the previous study suggested that the wild-type enzyme had an extremely high apparent Km value for cortisol; when we examined these data (Fig 2⇑ in Reference 2222 ) in a double reciprocal plot, they yielded an apparent Km for cortisol of 20 μmol/L, which is several orders of magnitude higher than that found in the present study. This suggests that there were significant problems with the stability of even the normal enzyme in these experiments. The procedure in the present study differs from that used previously in its inclusion of glycerol in the lysis buffer; this agent has been shown to stabilize partially active mutants both for cytochrome P450 enzymes such as steroid 21-hydroxylase23 and for another short-chain dehydrogenase, 3β-hydroxysteroid dehydrogenase.24
We feel that it is difficult to reliably measure apparent Km values in whole cells, as illustrated by the differences observed in the present study between the apparent Km values of the wild-type enzyme in whole cells and in cell lysates. However, measurements of relative enzymatic activities in whole cells as measured by apparent first-order rate constants may be of some value in drawing phenotype-genotype correlations, as discussed below. It should be kept in mind that such comparisons involve apparent kinetic constants; true kinetic constants can be determined only on purified enzymes. Altered stability of mutant enzymes may affect their apparent kinetic constants, although this should not decrease the validity of comparisons between the various mutants as long as their expression levels are similar.
Genotype Is Correlated With Phenotype
Most inherited disorders exhibit allelic variation. The best-studied disorder of steroid metabolism is congenital adrenal hyperplasia due to 21-hydroxylase deficiency, in which allelism accounts for approximately 80% of observed individual variations in both biochemical and clinical parameters of severity.25
Although the number of patients with AME is small, sufficient data now exist to demonstrate a statistically significant correlation between the degree of enzymatic impairment and biochemical severity as measured by the ratio of precursor to product, (THF+aTHF)/THE. This correlation is most obvious for the partially active mutants. We assume in this analysis that R337C is the only significant mutation in the patients who carry it, even though only one exon of the gene was sequenced. If so, a 50% impairment of enzymatic activity is apparently sufficient to compromise cortisol metabolism in the kidney, suggesting that this organ has very little excess capacity to metabolize cortisol. This seems to raise a paradox, because AME is a recessive disorder, and heterozygous carriers, who would be expected to have 50% of normal activity, are asymptomatic. The same apparent paradox exists for 21-hydroxylase deficiency, which is also a recessive disorder; a clinically mild or nonclassic form of 21-hydroxylase deficiency is seen in individuals who are homozygous for mutations that are approximately 50% active when expressed in cultured cells.23 26 27 Altered stability or kinetic properties of the R337C mutant may be important, including alterations in enzyme inhibition by end product (ie, cortisone or corticosterone) or by other circulating steroids.
The widest range of (THF+aTHF)/THE ratios is seen in patients who carry mutations that apparently destroy enzymatic activity. This is not very surprising; small variations in the very low levels of cortisone metabolites (ie, THE) excreted by these patients will obviously lead to quite large differences in ratios in which this value is the denominator. It is possible that the normally low renal expression of the L isozyme of 11-HSD28 is sufficient to metabolize small amounts of cortisol when the K isozyme is absent. Thus, the relatively wide variations in the ratios of precursor to product seen in patients carrying inactive K isozymes may reflect differences in expression or activity of the L isozyme in the kidney. Alternatively, a small amount of dehydrogenation may be catalyzed by the L isozyme in the liver, even though the reaction in this organ is predominantly in the reductive direction.
Because of the small numbers of patients, and the possible confounding effects of prior antihypertensive therapy, it is difficult to correlate biochemical severity with measures of clinical severity. With the elucidation of the molecular genetic basis of this disorder, ascertainment of additional cases may permit these questions to be answered.
Selected Abbreviations and Acronyms
|AME||=||apparent mineralocorticoid excess|
|CHOP||=||modified Chinese hamster ovary|
|PCR||=||polymerase chain reaction|
This work was supported by grant DK42169 from the National Institutes of Health, Bethesda, Md. We respectfully dedicate this article to the memory of our late colleague, Carl Monder. We thank James Dennis for CHOP cells, and Jonathan Seckl and Roger Brown for the antiserum to HSD11K.
Reprint requests to Dr Perrin C. White, UT Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75235-9063. E-mail firstname.lastname@example.org.
- Received February 8, 1996.
- Revision received March 1, 1996.
- Accepted March 11, 1996.
Agarwal AK, Monder C, Eckstein B, White PC. Cloning and expression of rat cDNA encoding corticosteroid 11 beta-dehydrogenase. J Biol Chem. 1989;264:18939-18943.
Nikkila H, Tannin GM, New MI, Taylor NF, Kalaitzoglou G, Monder C, White PC. Defects in the HSD11 gene encoding 11 beta-hydroxysteroid dehydrogenase are not found in patients with apparent mineralocorticoid excess or 11-oxoreductase deficiency. J Clin Endocrinol Metab. 1993;77:687-691.
Agarwal AK, Mune T, Monder C, White PC. NAD+-dependent isoform of 11 beta hydroxysteroid dehydrogenase: cloning and characterization of cDNA from sheep kidney. J Biol Chem. 1994;269:25959-25962.
Wilson RC, Harbison MD, Krozowski ZS, Funder JW, Shackleton CH, Hanauske-Abel HM, Wei JQ, Hertecant J, Moran A, Neiberger RE, Balfe JW, Fattah A, Daneman D, Licholai T, New MI. Several homozygous mutations in the gene for 11β-hydroxysteroid dehydrogenase type 2 in patients with apparent mineralocorticoid excess. J Clin Endocrinol Metab. 1995;80:3145-3150.
Arriza JL, Weinberger C, Cerelli G, Glaser TM, Handelin BL, Housman DE, Evans RM. Cloning of human mineralocorticoid receptor complementary DNA: structural and functional kinship with the glucocorticoid receptor. Science. 1987;237:268-275.
Funder JW, Pearce PT, Smith R, Smith AI. Mineralocorticoid action: target tissue specificity is enzyme, not receptor, mediated. Science. 1988;242:583-585.
Andersson S, Bishop RW, Russell DW. Expression cloning and regulation of steroid 5 alpha-reductase, an enzyme essential for male sexual differentiation. J Biol Chem. 1989;264:16249-16255.
Higuchi R. Recombinant PCR. In: Innis MA, Gelfand DH, Sninsky JJ, White TJ, eds. PCR Protocols. San Diego, Calif: Academic Press; 1990:177-183.
Heffernan M, Dennis JW. Polyoma and hamster papovavirus large T antigen-mediated replication of expression shuttle vectors in Chinese hamster ovary cells. Nucleic Acids Res. 1991;19:85-92.
Albiston AL, Obeyesekere VR, Smith RE, Krozowski ZS. Cloning and tissue distribution of the human 11 beta-hydroxysteroid dehydrogenase type 2 enzyme. Mol Cell Endocrinol. 1994;105:R11-R17.
Ghosh D, Weeks CM, Grochulski P, Duax WL, Erman M, Rimsay RL, Orr JC. Three-dimensional structure of holo 3 alpha,20 beta-hydroxysteroid dehydrogenase: a member of a short-chain dehydrogenase family. Proc Natl Acad Sci U S A. 1991;88:10064-10068.
Tusie-Luna MT, Traktman P, White PC. Determination of functional effects of mutations in the steroid 21-hydroxylase gene (CYP21) using recombinant vaccinia virus. J Biol Chem. 1990;265:20916-20922.
Simard J, Rheaume E, Sanchez R, Laflamme N, de Launoit Y, Luu-The V, van Seters AP, Gordon RD, Bettendorf M, Heinrich U, Moshang T, New MI, Labrie F. Molecular basis of congenital adrenal hyperplasia due to 3 beta-hydroxysteroid dehydrogenase deficiency. Mol Endocrinol. 1993;7:716-728.
Speiser PW, Dupont J, Zhu D, Serrat J, Buegeleisen M, Tusie-Luna MT, Lesser M, New MI, White PC. Disease expression and molecular genotype in congenital adrenal hyperplasia due to 21-hydroxylase deficiency. J Clin Invest. 1992;90:584-595.
Tannin GM, Agarwal AK, Monder C, New MI, White PC. The human gene for 11 beta-hydroxysteroid dehydrogenase: structure, tissue distribution, and chromosomal localization. J Biol Chem. 1991;266:16653-16658.