Main

Aldosterone, the most important mineralocorticoid, regulates electrolyte excretion and intravascular volume mainly through its effects on renal distal tubules and cortical collecting ducts, where it acts to increase sodium resorption from and potassium excretion into the urine (Fig. 1) (reviewed inRef. 1). Excess secretion of aldosterone or other mineralocorticoids, or abnormal sensitivity to mineralocorticoids, may result in hypokalemia, suppressed plasma renin activity, and hypertension. Such conditions often have a genetic basis. Glucocorticoid-suppressible hyperaldosteronism and congenital adrenal hyperplasia due to 11β-hydroxylase deficiency are examples of autosomal dominant and recessive forms of mineralocorticoid excess, respectively(2). This article reviews studies of another such syndrome, AME, that provide valuable insights into normal and abnormal physiology of mineralocorticoid action.

Figure 1
figure 1

Schematic of mineralocorticoid action. Top, a normal mineralocorticoid target cell in a renal cortical collecting duct. Aldosterone occupies nuclear receptors (MR) that bind to hormone response elements, increasing transcription of genes and directly or indirectly increasing activities of apical sodium (Na) channels and the basolateral sodium-potassium (Na/K) ATPase. This increases resorption of sodium from and excretion of potassium into the tubular lumen. Cortisol, which circulates at higher levels than aldosterone, cannot occupy the receptor because it is oxidized to cortisone by 11β-HSD.Bottom, a cell from a patient with the syndrome of apparent mineralocorticoid excess. Because 11β-HSD is absent, cortisol inappropriately occupies mineralocorticoid receptors, leading to increased gene transcription, increased activity of sodium channels and the Na/K ATPase, increased resorption of sodium and excretion of potassium, and hypertension.

Full size image

Clinical features of the syndrome of apparent mineralocorticoid excess. AME is an inherited syndrome in which children present with hypertension, hypokalemia, and low plasma renin activity. Other clinical features include moderate intrauterine growth retardation and postnatal failure to thrive. Consequences of the often severe hypokalemia include nephrocalcinosis, nephrogenic diabetes insipidus, and rhabdomyolysis. Complications of hypertension have included cerebrovascular accidents, and several patients have died during infancy or adolescence. Several affected sibling pairs have been reported, but parents have usually been asymptomatic, suggesting that AME is a genetic disorder with an autosomal recessive mode of inheritance.

A low salt diet or blockade of mineralocorticoid receptors with spironolactone ameliorate the hypertension, whereas ACTH and hydrocortisone exacerbate it. Levels of all known mineralocorticoids are low(3, 4). These findings suggest that cortisol(i.e., hydrocortisone) acts as a stronger mineralocorticoid than is normally the case. Indeed, patients with AME have abnormal cortisol metabolism. Cortisol half-life in plasma is prolonged from approximately 80 min to 120-190 min(3). Very low levels of cortisone metabolites are excreted in the urine as compared with cortisol metabolites, indicating a marked deficiency in 11β-HSD, the enzyme catalyzing the conversion of cortisol to cortisone (Fig. 2). This is most often measured as an increase in the sum of the urinary concentrations of THF and allo-THF, divided by the concentration of THE [(THF + allo-THF)/THE]. However, 11-reduction is unimpaired; labeled cortisone administered to patients is excreted entirely as cortisol and other 11β-reduced metabolites(5). There is also an increase in the ratio of 5α- to 5β-reduced cortisol metabolites, allo-THF/THF(6), suggesting that 5β reduction is also impaired. The (THF + allo-THF)/THE is usually much higher than the allo-THF/THF ratio, and the two ratios are linearly related, implying that the primary defect in this disorder is indeed one of 11β-dehydrogenation. This has been assayed directly by administering 11α-[3H]cortisol to subjects and measuring the appearance of tritiated water.

Figure 2
figure 2

Positions at which enzymes metabolize cortisol.

Full size image

Similar but milder abnormalities occur with licorice intoxication(7). The active component of licorice, glycyrrhetinic acid, inhibits 11β-HSD in isolated rat kidney microsomes(8). Thus, it appears that licorice intoxication is a reversible pharmacologic counterpart to the inherited syndrome of apparent mineralocorticoid excess.

WHY DOES 11β-HSD DEFICIENCY CAUSE HYPERTENSION?

Aldosterone acts through transcriptional effects mediated by a specific nuclear receptor referred to as the mineralocorticoid or “type 1 steroid” receptor. These receptors are expressed at high levels in renal distal tubules and cortical collecting ducts but also in other mineralocorticoid target tissues, including salivary glands and the colon. The mineralocorticoid receptor has a high degree of sequence identity with the glucocorticoid or “type 2” receptor(9), and it has very similar in vitro binding affinities for aldosterone and for glucocorticoids such as corticosterone and cortisol(9, 10).

It has been proposed(7, 11, 12) that oxidation by 11β-HSD of cortisol or corticosterone to cortisone or 11-dehydrocorticosterone, respectively, represents the physiologic mechanism conferring specificity for aldosterone upon the mineralocorticoid receptor (Fig. 1). Although cortisol and corticosterone bind the receptor well in vitro, cortisone and 11-dehydrocorticosterone are poor agonists for this receptor. Aldosterone is a poor substrate for 11β-HSD, because, in solution, its 11-hydroxyl group is normally in a hemiacetal conformation with the 18-aldehyde group. Thus, in AME or licorice intoxication, 11β-HSD deficiency permits cortisol to occupy the mineralocorticoid receptor. Because cortisol normally circulates at levels 100-1000 times those of aldosterone, this leads to signs of mineralocorticoid excess even when aldosterone secretion is suppressed. To confirm this hypothesis, we cloned cDNA and genes for two isozymes of 11β-HSD and tested their involvement in AME.

CLONING OF cDNA-ENCODING ISOZYMES OF 11B-HSD

There are two distinct isozymes of 11β-HSD (Table 1). The first, termed the liver (L) or type 1 isozyme, was originally isolated from rat liver microsomes(13). It is a glycoprotein with a molecular mass of 34 kD that requires NADP+ as a cofactor. A cDNA clone encoding this enzyme was isolated using an antiserum to the purified rat protein(14). The fulllength cDNA is 1.4 kb long including an open reading frame of 876 bp, predicting a protein of 292 amino acids. Although the enzyme purified from rat liver functions only as a dehydrogenase, the recombinant enzyme expressed from cloned cDNA exhibits both 11β-dehydrogenase and the reverse oxoreductase activity (conversion of 11-dehydrocorticosterone to corticosterone) when expressed in mammalian cells(14). At physiologic pH in cell lysates, the kinetic constants for dehydrogenation and reduction (Km of 1.1 and 1.4 μM, respectively) are almost identical(15). These findings imply that this isozyme actually catalyzes a fully reversible reaction and that reductase activity is destroyed during purification from the liver. The corresponding human gene for this isozyme, HSD11L(HSD11B1) is located on chromosome 1, and contains 6 exons with a total length of over 9 kb(16) (Fig. 3). No mutations were identified in this gene in patients with AME(17).

Table 1 Characteristics of 11β-HSD isozymes and genes
Full size table
Figure 3
figure 3

Genes encoding 11β-HSD isozymes. Numbered black boxes represent exons, and those exons that encode domains of functional importance are indicated. Highly conserved predicted amino acid sequences are shown in single-letter code; identical residues within these regions are shaded.

Full size image

Other lines of evidence also suggest that this isozyme does not play a significant role in conferring ligand specificity on the mineralocorticoid receptor. This isozyme is expressed at highest levels in the liver, which does not respond to mineralocorticoids, and although it is expressed at high levels in the rat kidney(14), it is expressed at much lower levels in human(16) and sheep(18) kidneys. Even in rat kidney, immunoreactivity to the protein is observed primarily in proximal tubules and not in distal tubules and collecting ducts, the sites of mineralocorticoid action(19).

Accordingly, a second isozyme was sought in mineralocorticoid target tissues. Evidence for such an isozyme was obtained from biochemical studies of isolated rabbit kidney cortical collecting duct cells(20). Activity of 11β-HSD in the microsomal fraction was almost exclusively NAD+ dependent and had a Km for corticosterone of 26 nM. There was almost no reduction of 11-dehydrocorticosterone to corticosterone, suggesting that, unlike the liver(L) isozyme, the kidney (K) or type 2 isozyme catalyzed only dehydrogenation. The enzyme in the human placenta had similar characteristics(21); it was NAD+-dependent and had Km values for corticosterone and cortisol of 14 and 55 nM. Partial purification using AMP affinity chromatography suggested that this isozyme had a molecular weight of 40,000. Thus far, the K isozyme has not been purified to homogeneity in active form from any source. However, a homogenous preparation was recently obtained by a combination of affinity chromatography, affinity labeling, and preparative two-dimensional electrophoresis(22).

Cloning of cDNA encoding the K isozyme of 11β-HSD was rendered more difficult by the unavailability of purified enzyme that could be used to produce an antiserum or to obtain amino acid sequence data. However, because sheep and human kidneys predominantly expressed this type of enzyme, it was feasible to clone the corresponding cDNA by expression screening strategies in which pools of clones were assayed for their ability to confer NAD+-dependent 11β-HSD activity on Xenopus oocytes or cultured mammalian cells. Positive pools were divided into smaller pools and rescreened until a single positive clone was identified. Both sheep(23) and human(24) cDNA encoding this isoform were isolated in this manner. The recombinant K isozyme has properties that are virtually identical to the activity found in mineralocorticoid target tissues. The recombinant enzyme functions exclusively as a dehydrogenase; no reductase activity is detectable with either NADH or NADPH as a cofactor(23, 24). It has an almost exclusive preference for NAD+ as a cofactor and a very high affinity for glucocorticoids. Corticosterone is the preferred substrate, with first order rate constants 10 times higher than those for cortisol, even in mammalian species in which cortisol is the predominant glucocorticoid. Reported Km values for corticosterone are 0.7-10.1 nM and for cortisol, 14-47 nM.

The protein is predicted to contain 404 (sheep) or 405 (human) amino acid residues with a total Mr of 41,000 [the published sheep sequence(23) contains a frameshift error near the 3′ end of the coding sequence]. The human and sheep predicted peptide sequences are 83% identical. A search of sequence databases reveals sequence similarity to members of the short chain alcohol dehydrogenase superfamily. The 11β-HSD K isozyme is most similar (37% sequence identity) to the type 2 (placental, NAD+-dependent, microsomal) isozyme of 17β-HSD(25). It is only 21% identical to the L isozyme of 11β-HSD. The relatively high degree of similarity between the 11β-HSD K isozyme and placental 17β-HSD (comparable to the similarity between cytochrome P450 gene family members) suggests that these two enzymes may be in the same gene family within the short chain dehydrogenase superfamily.

Regions of sequence similarity between the two isozymes (Fig. 3) include part of the putative binding site for the nucleotide cofactor (residues 85-95 in the 11β-HSD K isozyme) and absolutely conserved tyrosine and lysine residues (Tyr-232 and Lys-236 in this enzyme) that function in catalysis(26, 27). The region immediately to the N-terminal side of the catalytic residues forms part of a putative steroid binding pocket in the short chain dehydrogenase that has been analyzed by x-ray crystallography, 3α,20β-HSD(28). This region is notably well conserved (10/18 identical residues) between the two isozymes of 11β-HSD, consistent with a role in binding the substrate.

The corresponding gene, termed HSD11K or HSD11B2, contains five exons spaced over approximately 6.2 kb(29)(Fig. 3). The putative binding site for the NAD+ cofactor (including the core sequence, G XXXG XG) is split between exons 1 and 2, whereas the putative catalytic residues, Tyr-232 and Lys-236, are encoded by exon 4. This structure is different from that of the HSD11L gene encoding the liver (type 1) isozyme of 11β-HSD(16), indicating that the two isozymes belong to different gene families.

HSD11K is expressed in placenta and mineralocorticoid target tissues, particularly the kidney, whereas it is not detected in the liver. Whereas human fetal and adult tissues contain transcripts of 1.9-2.0 kb(24, 29), fetal tissues also express transcripts of 5 and 7 kb(29). These may represent utilization of alternative polyadenylation sites or partially processed transcripts.

DETECTION OF MUTATIONS IN HSD11K IN PATIENTS WITH AME

We identified seven different mutations the HSD11K gene in eight kindreds with AME (Fig. 4). These mutations all affect enzymatic activity or pre-mRNA splicing, thus confirming in its entirety the hypothesis that 11β-HSD protects the mineralocorticoid receptor from high concentrations of cortisol(30). Seven other kindreds have been studied and an additional four mutations detected by others(31–33). Only one patient has been a compound heterozygote for two different mutations, whereas all other patients have carried homozygous mutations. This suggests that the prevalence of AME mutations in the general population is low, so that the disease is found mostly in limited populations in which inbreeding is relatively high. Six kindreds are of Native American origin. Three from Minnesota or Canada carry the same mutation (L250S,L251P), consistent with a founder effect, but the others are each homozygous for a different mutation. The reason for the relatively high prevalence of this very rare disease among Native Americans is not immediately apparent.

Figure 4
figure 4

Locations of mutations causing apparent mineralocorticoid excess. Intron 1 is not drawn to scale. The predicted percent of wild type enzymatic activity associated with each mutation is listed.

Full size image

Of the mutations identified thus far, two shift the reading frame of translation, a third deletes three amino acids including a crucial catalytic residue (Tyr-232), and one is a nonsense mutation. These mutations are all presumed to completely destroy enzymatic activity. One mutation in the third intron leads to skipping of the fourth exon during processing of pre-mRNA(30). As the fourth exon encodes the catalytic site, the resulting enzyme is again presumably inactive. The other six mutations have been introduced into cDNA and expressed in cultured cells to determine their effects. One (L250P,L251S) is completely inactive and one (R337Δ3nt) has only a trace of activity. The others are all partially active in cultured cells with one, R337C, having greater than 50% of normal activity(34). Only R337C is partially active in lysed cells[although one group reported this mutation to be inactive in cell lysates(35), they did not use appropriate conditions to maximize enzyme stability(36)]. We believe that comparisons of activity are best made using the apparent first order rate constant,Vmax/Km, which predicts reaction velocity at low substrate concentrations. Valid comparisons in whole cells require controls (Western blots or determinations of mRNA levels) for transfection efficiency. However, determinations of apparent Km in whole cells need to be interpreted cautiously, particularly when high concentrations of substrate are used, because many substrates including steroids are subject to active transport into or out of cells(37). Such mechanisms, which have their own kinetics, can confound kinetic measurements of enzymes.

Both the wild type enzyme and most mutants are concentrated in the nucleus as determined by Western blots of cell fractions. All six mutants that we examined are expressed in decreased amounts suggesting that most of the mutations adversely affect protein stability once cells are lysed(34).

Although the number of patients with AME is small, sufficient data now exist to demonstrate a statistically significant correlation between degree of enzymatic impairment and biochemical severity as measured by the precursor:product ratio (THF + allo-THF)/THE(34). This correlation is most obvious for the partially active mutants. We assume that R337C is the only significant mutation in the patients who carry it, even though only one exon of the gene was sequenced(31). If so, a 50% impairment of enzymatic activity is apparently sufficient to compromise metabolism of cortisol in the kidney, suggesting that there is very little excess capacity to metabolize cortisol in this organ. 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. Altered stability or kinetic properties of the R337C mutant may be important, including alterations in enzyme inhibition by end product(i.e., cortisone or corticosterone) or by other circulating steroids.

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, although anecdotal reports suggest that mutations that do not destroy activity may be associated with milder disease(30, 31). With the elucidation of the molecular genetic basis of this disorder, ascertainment of additional cases may permit these questions to be answered.