Mutants of 11β-Hydroxysteroid Dehydrogenase (11-HSD2) With Partial Activity
Improved Correlations Between Genotype and Biochemical Phenotype in Apparent Mineralocorticoid Excess
Abstract—Mutations in the kidney isozyme of human 11-hydroxysteroid dehydrogenase (11-HSD2) cause apparent mineralocorticoid excess, an autosomal recessive form of familial hypertension. We studied 4 patients with AME, identifying 4 novel and 3 previously reported mutations in the HSD11B2 (HSD11K) gene. Point mutations causing amino acid substitutions were introduced into a pCMV5/11HSD2 expression construct and expressed in mammalian CHOP cells. Mutations L179R and R208H abolished activity in whole cells. Mutants S180F, A237V, and A328V had 19%, 72%, and 25%, respectively, of the activity of the wild-type enzyme in whole cells when cortisol was used as the substrate and 80%, 140%, and 55%, respectively, of wild-type activity when corticosterone was used as the substrate. However, these mutant proteins were only 0.6% to 5.7% as active as the wild-type enzyme in cell lysates, suggesting that these mutations alter stability of the enzyme. In regression analyses of all AME patients with published genotypes, several biochemical and clinical parameters were highly correlated with mutant enzymatic activity, demonstrated in whole cells, when cortisol was used as the substrate. These included the ratio of urinary cortisone to cortisol metabolites (R2=0.648, P<0.0001), age at presentation (R2=0.614, P<0.0001), and birth weight (R2=0.576, P=0.0004). Approximately 5% conversion of cortisol to cortisone is predicted in subjects with mutations that completely inactivate HSD11B2, suggesting that a low level of enzymatic activity is mediated by another enzyme, possibly 11-HSD1.
Mineralocorticoid receptor specificity is determined by the activity of the kidney (or type 2) isozyme of 11β-hydroxysteroid dehydrogenase (11-HSD2). This enzyme oxidizes cortisol, a potential ligand for the mineralocorticoid receptor, to cortisone, an inactive steroid. Deficiency of 11-HSD2 causes the syndrome of apparent mineralocorticoid excess (AME), which is characterized by hypertension, hypokalemia, and suppressed plasma renin and aldosterone levels (reviewed in White et al1 ). Patients with AME excrete reduced amounts of cortisone metabolites, ie, tetrahydrocortisone (THE), as compared with cortisol metabolites, ie, tetrahydrocortisol (THF) and allotetrahydrocortisol (aTHF). An elevated (THF+aTHF)/THE ratio is considered pathonomic for AME. This potentially fatal disease can be treated with mineralocorticoid receptor antagonists, such as spironolactone.
The HSD11B2 (HSD11K) gene encoding this enzyme is located on chromosome 16q22,2 and mutations in this gene have been identified in many patients with AME. Most mutations characterized thus far severely impair enzymatic activity. In the current study, we identified several novel 11-HSD2 mutations in 4 patients with AME; 3 of the mutants retain significant enzymatic activity in whole cells, permitting more robust correlations between genotype and phenotype than previously possible.
Detection of HSD11B2 Mutations in Patients With AME
DNA samples were obtained from 4 patients with AME and at least 1 parent of each patient. Data on patient 1 were reported previously.3 Urinary steroid levels were measured in patients 2-4 as previously described.4 5 Exons were amplified using polymerase chain reaction (PCR) as described elsewhere.6 Amplified products were sequenced with a radiolabeled terminator cycle sequencing kit (Amersham) following the manufacturer’s recommended procedure.
Expression of 11-HSD2 Mutants in Mammalian Cells
Mutations were inserted into a pCMV-5/11-HSD2 construct using 2 rounds of PCR as previously described.6 Mutant constructs were sequenced completely to ensure that no undesired mutations were introduced into the 11-HSD2 coding sequence. Cells were transfected as described,7 and transfection efficiencies were determined by reverse transcription (RT)–PCR within the exponential phase of the reaction. Metabolism of cortisol and corticosterone to their respective 11-keto derivatives was determined in whole cells and in cell lysates using thin-layer chromatography as previously described.7 For determination of kinetic parameters, CHOP cells transfected with either pCMV5/11-HSD wild-type, S180F, or G237A constructs were grown for 24 hours, collected in 1.5 mL of buffer A (50 mmol/L Tris, 2 mmol/L EDTA, 1 mmol/L MgCl2, 20% glycerol, 2 mg/mL aprotinin, 0.5 mg/mL leupeptin, 100 mg/mL PMSF, and 1 mg/mL pepstatin; pH 8.0) and briefly sonicated. Aliquots (100 μL) of the resulting cell lysate were incubated with an equal volume of buffer A plus 2 mmol/L NAD+ and 20 to 2560 nmol/L tritiated cortisol. Cells were incubated for times that resulted in <20% conversion of substrate to product to ensure first-order kinetics. Western blot analysis was performed as described previously.5
Mutations in HSD11B2 Leading to AME
Seven mutations in the 11-HSD2 gene were identified by sequencing genomic DNA from 4 patients with abnormally high urinary (THF+aTHF)/THE ratios (Table⇓). Patient 1, a Japanese girl,3 was found to be homozygous for a mutation in exon 3 (S180F, TCT to TTT). Patient 2, a white Australian boy, had a missense mutation in exon 3 (L179R, CTG to CGG; also present in the mother) and a single nucleotide insertion in codon 246 in exon 4, leading to a frame shift. A missense mutation in exon 3 (R208H, CGC to CAC) was identified in patient 3, a Mexican American girl (Table⇓). R208H had previously been identified in a Japanese kindred.8 Patient 3 also had a mutation in intron 3 (C to T) that was previously described in another Mexican American patient.6 This mutation leads to incorrect pre-mRNA splicing (skipping of exon 4). The father of this patient was heterozygous for this mutation but did not carry R208H. A mutation in exon 4 (A237V, GCG to GTG) was identified in a patient 4, a white American boy. This patient harbored a second mutation in exon 5 (A328V, GCG to GTG) that was also carried by the mother. This latter mutation was previously identified in a Brazilian kindred.9
Enzymatic Activity in Whole Cells
CHOP cells transfected with wild-type 11-HSD2 were able to convert both cortisol and corticosterone into cortisone and 11-dehydrocorticosterone, respectively (Figure 1⇓). Neither sham-transfected CHOP cells nor CHOP cells transfected with an empty pCMV-5 vector metabolized cortisol or corticosterone to an appreciable extent (data not shown). Metabolism of cortisol and corticosterone by cells transfected with mutants L179R and R208H was not significantly different from sham-transfected cells or cells transfected with pCMV5 only. Mutants S180F, A237V, and A328V retained significant activity, having 19%, 72%, and 25% of wild-type activity, respectively, when cortisol was used as the substrate and 80%, 140%, and 55% of wild-type activity, respectively when corticosterone was used as the substrate. CHOP cells cotransfected with equal amounts of A237V and A328V metabolized both cortisol and corticosterone at rates intermediate of those transfected with either A237V or A328V alone (not shown).
Enzymatic Activity in Cell Lysates
The enzymatic activities of wild-type 11-HSD2 and mutants were also assessed in lysed CHOP cells (Figure 1⇑). The wild-type protein retained high activity after lysis. Because mutants L179R and R208H were inactive in whole cells, their activity in lysed cells was not assayed. Mutants S180F, A237V, and A328V were 1.8%, 3.1%, and 0.6% as active as the wild-type enzyme, respectively, when cortisol was used as the substrate and 5.7%, 5.3%, and 3.9% as active, respectively, when corticosterone was used as the substrate. Cotransfection of the A237V mutant with A328V did not significantly alter the stability of either enzyme in cell lysates (not shown).
The wild-type protein was found to have characteristics consistent with those reported for the same enzyme in other studies7 10 (ie, Km=59 nmol/L and Vmax=175 pmol/min · mg−1 protein with cortisol as the substrate). The mutations S180F and A237V generated proteins with decreased affinity for cortisol (Km=631 and 868 nmol/L, respectively) and decreased maximum velocity (Vmax=8 and 16 pmol/min · mg−1 protein, respectively).
Western blot analysis of lysates from transfected CHOP cells indicated that all pCMV5/11-HSD2 constructs were translated into protein, but less immunoreactivity was observed in lysates from cells transfected with mutant constructs as compared with wild-type constructs (not shown). RT-PCR analysis of total RNA from transfected cells indicated comparable production of 11-HSD2 RNA from all constructs (not shown).
Correlations Between Genotype and Biochemical Phenotype
In many inherited diseases, the nature of the particular mutations causing the disease is the best predictor of disease severity. For example, in the most common disorder of steroid metabolism, congenital adrenal hyperplasia due to steroid 21-hydroxylase deficiency, allelic variation accounts for 80% of interindividual differences in disease severity when severity is assessed with use of a consistent scoring system.11 We previously attempted to correlate the HSD11B2 genotype with its phenotype in AME and found that enzymatic activity in whole cells was correlated with urinary precursor/product ratios ([THF+aTHF]/THE).7 However, the analysis was limited by the small number of mutants retaining partial activity that had been identified at that time. This problem was resolved in the present study, in which 3 mutants retained >15% of wild-type activity when cortisol (the major glucocorticoid in humans) was used as the substrate. In addition, 2 other kindreds carrying mutations retaining significant activity were recently reported.12 13 These additional data allow more informative correlations to be made between enzymatic activity of the recombinant enzyme expressed in whole cells and the product/precursor ratio (THE/[THF+aTHF]). Although the inverse of this ratio has generally been used to describe patients, product/precursor ratios are expected to be directly correlated with enzymatic activity, whereas precursor/product ratios should be inversely correlated, making it more difficult to fit the data to a linear relationship.
Indeed, THE/(THF+aTHF) is highly correlated (R2=0.648, P<0.0001) with the percentage of wild-type enzymatic activity when cortisol is used as the substrate (Figure 2⇓). This ratio is also correlated with enzymatic activity in whole cells when corticosterone is the substrate (R2=0.413, P<0.0005) and weakly correlated with enzymatic activity in cell lysates when cortisol is the substrate (R2=0.275, P=0.07).
The free urinary cortisone/free urinary cortisol ratio (E/F)5 was also correlated with enzymatic activity when cortisol was used as the substrate (R2=0.597, P<0.0001). Although the slightly lower correlation, as compared with THE/(THF+aTHF), was apparently due to the smaller number of subjects for whom free cortisone and cortisol excretion data were available, the data suggest that these 2 ratios have nearly equivalent validity for evaluation of AME.
Several assumptions were made in this analysis. In the entire analysis, patient 4 was the only compound heterozygote with 2 missense mutations associated with different levels of activity. The analysis assumed the patient’s actual enzymatic activity to be the mean of the enzymatic activities of each of the 2 mutations, which is consistent with the results obtained when the 2 corresponding expression constructs were transfected together into CHOP cells (not shown). However, the results of the correlation between THE/(THF+aTHF) and whole-cell metabolism of cortisol changed minimally when the activity of either the less (R2=0.627) or more (R2=0.651) severely affected mutant was assumed instead. For mutations R337C and A328V, we observed greater enzymatic activity of the recombinant enzyme than has been reported by others, and we used our own measurements in the analysis. In the case of R337C, the discrepancy applied only to activity in cell lysates,7 14 but A328V9 was previously reported as being inactive in whole cells. Possible explanations for this difference include higher overall levels of expression in our study or perhaps an additional unsuspected mutation in the expression construct used by other investigators. For R279C, activity in whole cells was not reported, and we assumed it to be 70% of wild-type activity, the percentage observed in cell lysates,12 with cortisol as the substrate. Because most mutants have less activity relative to the wild-type enzyme in cell lysates than in whole cells, this is probably an underestimate. The correlation was slightly improved (R2=0.719) when 100% of wild-type activity was assumed. We did not include an average “normal” individual (ratio of 0.9 at 100% activity) in the analysis. The correlation became extremely strong (R2=0.935, P<0.0001) when the regression line was forced through this point.
Finally, patient 3 wasn’t included in the analysis because it is difficult to estimate the expression of 11-HSD2 due to normally spliced mRNA in the patient. However, both this patient and a previously reported patient who was homozygous for the same mutation6 have only moderate elevations in (THF+aTHF)/THE ratio (14.6 and 7.9, respectively), suggesting that at least a small proportion of mRNA transcripts of the mutant gene are properly spliced.
It should be noted that the intercept of the regression line (0.045±0.015) differed significantly (P=0.0053) from 0. In other words, an average of 5% of the normal amount of cortisone metabolites is produced in patients who are predicted to completely lack 11-HSD2 activity. In such patients, cortisone must presumably be produced by an enzyme other than 11-HSD2. The most likely candidate is the liver isozyme of 11-HSD, 11-HSD1, which functions predominantly as a reductase in vivo but catalyzes both oxidation and reduction when expressed in cultured cells.15
Correlations Between Genotype and Clinical Phenotype
The increased number of patients with relatively mild disease prompted us to attempt to correlate various parameters of clinical severity with whole-cell enzymatic activity using cortisol as the substrate (Figure 2⇑). Birth weight (R2=0.576, P=0.0004) and age at diagnosis (R2=0.614, P<0.0001) were both strongly correlated with enzymatic activity, and serum potassium levels were weakly but significantly correlated (R2=0.133, P<0.05). Because blood pressure varies with age, Z scores (number of standard deviations above the mean for that age) were calculated for blood pressures at presentation by using published normative data.16 No correlation with either systolic or diastolic blood pressure could be established regardless of whether all patients or only the first diagnosed patient in each kindred was included. This may reflect the inherent “noisiness” of random blood pressure measurements compared with many other clinical parameters, or the pathophysiology of AME may result in maximal stimulation of mineralocorticoid receptors with even moderate compromise of 11-HSD2 enzymatic activity.
Mutations Cluster in Exons 3 to 5
It is noteworthy that all reported mutations cluster in exons 3-5, a region representing 60% of the coding sequence.2 The likelihood of this occurring by chance is 1 in 7.1×10−5. The explanation for this is not immediately obvious; 50% of AME mutations are CpG to TpG changes, but exon 1 has a higher proportion of CpG dinucleotides (20% versus 4%) than the remainder of the gene. However, CpG to TpG mutations occur by deamination of methylcytosine, so hypomethylation of the 5′ end of this gene in germ cells is a possible partial explanation for the observed distribution of mutations. It cannot explain the distribution of other mutations. Moreover, several codons, including R208, L250, and R337, have undergone several independent mutations, as evidenced by both occurrence in unrelated ethnic groups and the presence of different mutations in each codon; the latter 2 codons have undergone both CpG to TpG and other types of mutations in different patients. Half of all kindreds carry mutations in at least 1 of these 3 codons, suggesting that these codons are actual mutational hot spots.
Functional Effects of Mutations
The missense mutations identified in the present study are all in rather highly conserved positions in the enzyme, although none affects residues known to be involved in binding the nucleotide cofactor or in catalysis. These mutations all decrease enzymatic activity in cell lysates more than activity in whole cells, suggesting that they affect stability of the enzyme. Consistent with this, decreased amounts of the mutant proteins are detected on Western blots. This has been previously noted with other mutations.7
Although at least 1 mutant has been previously reported to have full activity in whole cells,13 it is not surprising that altered stability of such mutant enzymes might nevertheless have adverse effects on activity in vivo. Presumably, 11-HSD2 is synthesized at lower levels in renal tubular epithelial cells than in cultured cells transfected with a high level expression vector, and steady-state levels of the enzyme may be more sensitive to its rate of degradation when levels of synthesis are relatively low.
How Common Is AME?
It is unlikely that mild forms of AME could be a significant cause of essential hypertension as previously suggested,12 13 except perhaps in the relatively inbred populations in which several of these mutations have been reported. AME clearly behaves as an autosomal recessive disease in the great majority of kindreds. Thus, it will be rare unless there are frequent unsuspected mutations in the population. This is also unlikely; most patients, including those who have mild disease, are homozygous for a single mutation and come from relatively inbred populations, suggesting that matings between 2 carriers are rare in the general population. Moreover, searches for coding sequence polymorphisms in normal individuals have been unrewarding (Smolenicka et al17 and unpublished observations). Thus, we cannot recommend routine genetic characterization of this locus in patients with low-renin essential hypertension, although it may be worthwhile in individuals with a suggestive history and altered cortisol/cortisone metabolite ratios. However, polymorphisms outside the coding sequence that affect gene expression have not been ruled out and could be a more important risk factor for hypertension, particularly that of the salt-sensitive type.
This work was supported by grant DK42169 from the National Institutes of Health. C.S. was supported by the Children’s Hospital Oakland Endowment.
Mario Palermo is currently with the Institute of Endocrinology, University of Sassari, Sassari, Italy.
- Received May 13, 1999.
- Revision received June 2, 1999.
- Accepted June 7, 1999.
Igarashi Y, Egi S, Takehiro A, Ohzeki T, Kawaguchi H. Studies on the metabolic abnormality of cortisol and corticosterone in a case of dexamethasone responsive mineralocorticoid excess. Folia Endocrinol Jpn. 1979;55:1341–1357.
Mune T, White PC. Apparent mineralocorticoid excess: genotype is correlated with biochemical phenotype. Hypertension. 1996;27:1193–1199.
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.
Wilson RC, Dave-Sharma S, Wei JQ, Obeyesekere VR, Li K, Ferrari P, Krozowski ZS, Shackleton CH, Bradlow L, Wiens T, New MI. A genetic defect resulting in mild low-renin hypertension. Proc Natl Acad Sci U S A. 1998;95:10200–10205.
Task Force on Blood Pressure Control in Children. Report of the Second Task Force on Blood Pressure Control in Children: 1987. Pediatrics. 1987;79:1–25.