11β-Hydroxysteroid Dehydrogenases: Intracellular Gate-Keepers of Tissue Glucocorticoid Action

Abstract

Glucocorticoid action on target tissues is determined by the density of “nuclear” receptors and intracellular metabolism by the two isozymes of 11β-hydroxysteroid dehydrogenase (11β-HSD) which catalyze interconversion of active cortisol and corticosterone with inert cortisone and 11-dehydrocorticosterone. 11β-HSD type 1, a predominant reductase in most intact cells, catalyzes the regeneration of active glucocorticoids, thus amplifying cellular action. 11β-HSD1 is widely expressed in liver, adipose tissue, muscle, pancreatic islets, adult brain, inflammatory cells, and gonads. 11β-HSD1 is selectively elevated in adipose tissue in obesity where it contributes to metabolic complications. Similarly, 11β-HSD1 is elevated in the ageing brain where it exacerbates glucocorticoid-associated cognitive decline. Deficiency or selective inhibition of 11β-HSD1 improves multiple metabolic syndrome parameters in rodent models and human clinical trials and similarly improves cognitive function with ageing. The efficacy of inhibitors in human therapy remains unclear. 11β-HSD2 is a high-affinity dehydrogenase that inactivates glucocorticoids. In the distal nephron, 11β-HSD2 ensures that only aldosterone is an agonist at mineralocorticoid receptors (MR). 11β-HSD2 inhibition or genetic deficiency causes apparent mineralocorticoid excess and hypertension due to inappropriate glucocorticoid activation of renal MR. The placenta and fetus also highly express 11β-HSD2 which, by inactivating glucocorticoids, prevents premature maturation of fetal tissues and consequent developmental “programming.” The role of 11β-HSD2 as a marker of programming is being explored. The 11β-HSDs thus illuminate the emerging biology of intracrine control, afford important insights into human pathogenesis, and offer new tissue-restricted therapeutic avenues.

I. INTRODUCTION

A. The Historical Context

The story of 11β-hydroxysteroid dehydrogenase (11β-HSD) begins with the isolation, synthesis, and therapeutic exploitation of adrenal corticosteroids. In the late 1930s, the laboratories of Kendall and Reichstein independently isolated and characterized the structures of steroids from the adrenal cortex, a triumph of 20th century chemistry (444, 445, 578, 579, 665). A number of preclinical studies were undertaken with the tiny amounts of these compounds that could be obtained by tissue extraction methods, but it was only in 1949 that Kendall's compound E, soon known as cortisone, was synthesized in sufficient quantities for its detailed investigation to begin. Hench used this first in patients with rheumatoid arthritis and then in rheumatic fever and showed remarkable efficacy in what had hitherto been inexorable inflammatory disorders (273–275). Immediately there followed a plethora of studies trying out the new “miracle cure” on almost every known disorder. Cortisone effected dramatic improvements in inflammatory and adrenal deficiency disorders (57); produced beneficial responses in certain malignancies, such as lymphoma, though not in many solid tumours; but often caused deterioration in infections like syphilis and tuberculosis. In 1950 alone, over 300 papers were published on the use of cortisone, a remarkable level of activity, mirroring some of the initial hope and hype surrounding gene and cell therapies in the current era. This early enthusiasm is now tempered in the light of contemporary understanding of the extraordinary range of benefits and harms attributable to glucocorticoids. Nonetheless, glucocorticoids remain among the most prescribed medicines (758). In recognition of their scientific tour-de-force, the 1950 Nobel Prize for Physiology or Medicine was awarded to Kendall, Reichstein, and Hench. However, as early as 1951 when Kendall's compound F or cortisol (hydrocortisone) became available, it was increasingly clear that this was more active than cortisone (286). It is salutary to reflect that the remarkable therapeutic responses Hench observed involved cortisone, a steroid which is intrinsically inert. Explaining this paradox reflects one major role for 11β-HSD.

In all the kerfuffle over clinical efficacy and side effects of cortisone (Hench carefully reported Cushing's disease-like features in his initial patient population; Ref. 273), a few groups began to address metabolism of the new glucocorticoids using chromatographic techniques to separate steroid moieties. First, Burton et al. (106) demonstrated cortisone metabolism to other steroids including cortisol in humans. Then in 1953, Amelung et al. (21) in Frankfurt administered cortisone to rats in vivo and incubated cortisone with homogenates of various tissues in vitro and found conversion to cortisol. The activity, localized to microsomes, was highest in liver with lower amounts in kidney and muscle. This was 11β-HSD (FIGURE 1).

Figure 1.

Reactions catalyzed by 11β-hydroxysteroid dehydrogenase (11β-HSD) isozymes. A: interconversion of cortisol and cortisone by 11β-HSD1 and -2. In intact cells and in vivo, 11β-HSD1 is predominantly a reductase, catalyzing NADPH-dependent reduction of cortisone to cortisol, predominantly in the liver. Under some circumstances and in some cells, it may act as an NADP-dependent dehydrogenase, inactivating cortisol. 11β-HSD2, in contrast, catalyzes the NAD⁺-dependent inactivation of cortisol, converting it to cortisone, predominantly in the kidney. B: conversion of 7-ketocholesterol to 7β-hydroxycholesterol by 11β-HSD1. Other reactions catalyzed by 11β-HSD1, including oxysterol metabolism, are probably of physiological importance. The 11β- and 7α-positions of the steroid nucleus show rotational symmetry, probably explaining the metabolism of 7-ketocholesterol to 7β-hydroxycholesterol by 11β-HSD1 as well as its metabolism of other 7-oxygenated sterols and steroids.

In the 1960s, 11β-HSD activity was reported in a variety of organs, predominantly detected using histochemical techniques of poor specificity (47, 102, 353). For more than 30 years whilst the biology and pharmacology of various glucocorticoids was intensively explored and their intracellular receptors characterized, 11β-HSD remained an arcane biochemical detail, a bidirectional shuttle catalyzing interconversion of cortisol and corticosterone (a minor glucocorticoid in most mammals, but the only form in rats and mice) to cortisone and 11-dehydrocorticosterone, respectively. This was considered of little interest to physiologists or clinicians, merely one of several pathways of metabolism of glucocorticoids. Nonetheless, this otherwise obscure enzyme reaction has added a novel strand to the 60-year-old story of glucocorticoids, revealing that prereceptor glucocorticoid metabolism is a critical control of physiological steroid action, underpins pathogenesis of rare and common disorders, and affords the possibility of tissue-targeted therapeutic manipulation of glucocorticoid action.

B. Glucocorticoid and Mineralocorticoid Receptors

In mammals, the adrenal cortex synthesizes aldosterone, the major physiological mineralocorticoid, from the zona glomerulosa and glucocorticoids (cortisol, corticosterone) from the zona fasciculata-reticularis. Circulating levels of cortisol and corticosterone vary widely from low nanomolar during sleep, to low micromolar during severe stress/illness. They are tightly bound to a high-affinity, limited capacity protein carrier, corticosteroid-binding globulin (CBG), and more loosely to albumin such that around 5% is “free” basally, although stress levels may exceed CBG's capacity. Inert 11-keto corticoids bind poorly to plasma proteins, as does aldosterone, which circulates at picomolar concentrations, perhaps 2–3 logs lower than glucocorticoids.

Corticosteroids are highly lipophilic and are thought readily to diffuse across biological membranes to access their intracellular receptors. Inward glucocorticoid carriers analogous to the recently discovered monocarboxylate transporter-8 that facilitates thyroid hormone access to cells (284) are, as yet, unreported, although some glucocorticoids are partly excluded from tissues such as the CNS by members of the ABC transporter family, notably ABCB1 (Mdr1) (460).

At the cellular level, the myriad effects of corticosteroids are largely a consequence of transcriptional actions mediated via binding to two types of intracellular receptor: the high-affinity mineralocorticoid receptor (MR) and the lower affinity glucocorticoid receptor (GR) (208, 452). 11β-Hydroxycorticosteroids are ligands for these receptors, but binding of their 11-keto forms is negligible. On binding ligand, GR and MR dissociate from complexes with chaperone proteins, translocate to the nucleus, and bind directly or indirectly to the regulatory regions of target genes: ∼2% of the human genome is regulated by glucocorticoids (577), although few if any genes are exclusively controlled by corticosteroids. In addition, a role for membrane receptors is emerging at least in the brain (143). Mice lacking MR lose such membrane effects, suggesting that products of the same gene underpin both nuclear/transcriptional and surface/rapid actions, although GPR30 may also be involved (249). MR show a rather restricted tissue distribution, with high expression confined to classical aldosterone target organs such as the kidney, colon, and salivary glands, as well as specific brain regions (notably the hippocampus) and more modest expression in vascular tissues, adipocytes (803), and specific immune cell populations (722). In contrast, GR are widely expressed, at higher levels than MR at most sites.

In vitro, MR have a high and very similar affinity (K_d ∼0.5–1 nM) for corticosterone, cortisol, and aldosterone (37, 370, 637), whereas potent synthetic glucocorticoids such as dexamethasone dissociate rapidly and only activate MR at high concentrations (416). In contrast, GR have a lower K_d for physiological glucocorticoids (typically ∼10–25 nM in cytosolic extracts), although a much higher affinity for synthetic glucocorticoids such as dexamethasone (∼1–5 nM), but barely bind mineralocorticoids (287). GR are therefore little occupied by basal levels of glucocorticoids (free tissue levels are likely to be subnanomolar), but become progressively activated as glucocorticoid levels rise during ultradian pulses, the diurnal maximum, a stress response or pharmacotherapy (144, 406, 580).

C. 11β-HSD Renaissance

In the mid 1980s, an 11β-HSD activity was characterized biochemically in the laboratory of Carl Monder, who subsequently purified the enzyme from rat liver and isolated the encoding cDNA (4, 377–379). The enzyme activity was bidirectional in tissue homogenates, containing both 11β-dehydrogenase (glucocorticoid inactivating) and 11β-reductase (glucocorticoid regenerating) activities, with NADP(H) as cosubstrate (FIGURE 1). This rat liver enzyme had a modest affinity (micromolar K_m) for glucocorticoids (379).

A number of clinical case reports described deficiency in the interconversion of cortisol to cortisone (as evidenced by an elevated urinary ratio of cortisol to cortisone metabolites; Ref. 720) in association with a very rare disease, the syndrome of “apparent mineralocorticoid excess” (AME). AME presents in childhood with severe hypertension, sodium retention, potassium loss, metabolic alkalosis, and suppressed plasma renin activity, findings compatible with mineralocorticoid excess. However, it is accompanied by undetectable levels of all known mineralocorticoids, notably aldosterone and deoxycorticosterone. The condition is normally fatal in childhood (483, 631, 632, 719). In the late 1980s, Stewart, Edwards, and colleagues in Edinburgh investigated a unique adult patient with AME (668). Their elegant clinical experiments showed that the mineralocorticoid excess was due to cortisol. Suppression of endogenous cortisol with dexamethasone reversed mineralocorticoid excess, whilst concurrent “replacement” with physiological doses of cortisol recapitulated mineralocorticoid excess, an effect not seen in healthy controls. These investigators also recognized that the syndrome was analogous to the effects of licorice, long known to cause hypertension and hypokalemia (128), and showed that ingestion of licorice in humans produced AME only in the presence of cortisol (672).

Meanwhile, Evans and his colleagues at the Salk Institute had just cloned a cDNA encoding human MR. In vitro the recombinant MR bound and was activated by cortisol, corticosterone, and aldosterone with similar affinity (37), confirming data from earlier biochemical studies (637). Indeed, it seemed remarkable that MR in the distal nephron are activated in vivo selectively by picomolar levels of aldosterone but not the higher nanomolar levels of cortisol, whereas structurally identical MR in the hippocampus are occupied by glucocorticoids in vivo (124, 473, 580). Thus the paradox was that renal MR are selectively activated by aldosterone in vivo, but both physiological glucocorticoids and mineralocorticoids in vitro.

The solution was provided in the late 1980s by the Edinburgh group (181, 672), as well as Funder and colleagues in Melbourne (210), who recognized that selectivity of MR in the kidney in vivo was not due to any intrinsic specificity for aldosterone but to the activity of 11β-HSD. In the kidney, 11β-HSD catalyzes the rapid inactivation of cortisol and corticosterone to inert 11-keto forms (cortisone, 11-dehydrocorticosterone) which do not bind MR. Only aldosterone, which is not a substrate for 11β-HSD, was able to activate renal MR. 11β-HSD inhibition by licorice or deficiency in AME allowed cortisol to bypass the enzymic “barrier” and bind to and activate MR causing sodium retention, potassium loss, and hypertension. For the glucocorticoid system this was the first example of prereceptor metabolism apparently gating steroid access to receptors. Analogous biology had been described for thyroid hormone receptors with mono-deiodinase isozymes inactivating or activating thyroid hormones in a cell-specific manner (228).

D. One Enzyme or Two (or Three)?

The rat liver 11β-HSD isolated and characterized and subsequently cloned by Monder and colleagues was found to be widely expressed, including in rat kidney, consistent with the idea that it underpinned MR selectivity and AME. However, several lines of evidence militated against the “one enzyme” hypothesis. Notably, 1) there are few MR in liver, the highest site of expression of this enzyme; 2) no mutations in the encoding gene, hsd11b1, were found in AME patients (695); 3) the enzyme is in the proximal tubule in the rat kidney and thus does not colocalise with MR in the distal nephron (181); 4) the enzyme is bidirectional in liver yet apparently a unidirectional dehydrogenase in kidney (321); and 5) there are marked discrepancies between “liver” 11β-HSD mRNA levels and enzyme activity in kidney (but not liver) (418), suggesting the existence of a distinct isozyme in kidney.

In 1993, a novel enzyme with 11β-hydroxysteroid dehydrogenase activity was isolated and characterized from human placenta (90) by Seckl, Chapman and colleagues and from rat kidney (602) by Naray-Fejes-Toth et al. This 41-kDa protein was distinct in physicochemical characteristics and apparent molecular weight from Monder's 34 kDa enzyme, and acted as a high affinity (low nM K_m) exclusive 11β-dehydrogenase which used NAD⁺ as cosubstrate rather than NADP(H). The next year, Krozowski's group (15) isolated a cDNA encoding the “renal” 11β-HSD isozyme from human kidney and White and colleagues cloned the analogous gene product from sheep kidney (6). The enzyme was purified from human placenta, and the encoding cDNA was found to be identical to the kidney version (91, 92). The rodent homologs were also cloned (127, 583). The new enzyme was called 11β-HSD type 2 to distinguish it from Monder's 11β-HSD type 1 (FIGURE 1). Mutations in hsd11b2 encoding 11β-HSD2 were soon identified in patients with AME (141, 498, 777). Stewart and Edward's original adult patient is a compound heterozygote with modest activity encoded by one allele, presumably responsible for his survival (388). Mice homozygous for targeted disruption of the Hsd11b2 gene faithfully recapitulate AME (363) (FIGURE 2).

Figure 2.

Diagrammatic representation of the reactions catalyzed by 11β-HSDs. The adrenal cortex secretes nanomolar concentrations of cortisol (F) and picomolar concentrations of aldosterone (Aldo) into the circulation. While mineralocorticoid receptors (MR) in the kidney only bind Aldo in vivo, identical MR in the hippocampus are occupied by F in vivo and MR bind F and Aldo with similar affinity in vitro. The solution to this conundrum lies in the collocation of renal MR with 11β-HSD2, which catalyzes the rapid inactivation of cortisol to inert cortisone (E) thus only allowing the nonsubstrate Aldo to access MR. 11β-HSD2 is absent from hippocampus so MR bind F. In tissues such as the liver, adipose, and adult brain, 11β-HSD2 is absent, but there is abundant 11β-HSD1. This catalyzes the reverse reaction in intact cells and organs and thus regenerates active F from inert E, amplifying the local glucocorticoid signal particularly at glucocorticoid receptors (GR). GR have 10-fold lower affinity for F than MR and are thus partially unoccupied by F at physiological concentrations allowing a dynamic range for 11β-HSD1 amplification inside cells to impact on signaling. In contrast, MR are largely occupied by physiological F concentrations where 11β-HSD2 is absent so 11β-HSD1 may make less impact on signaling via MR.

This left open the question of the function of Monder's original enzyme, 11β-HSD1, which is widespread but most highly expressed in liver. Some suggested it might be a lower affinity 11β-dehydrogenase. However, studies in clonal amphibian and mammalian cells transfected with rat 11β-HSD1 cDNA showed that, whilst bidirectional in homogenates, surprisingly it acts as a predominant 11β-reductase in most intact cells and thus functionally amplifies glucocorticoid action (174, 420). Indeed, in cells expressing 11β-HSD1, cortisone (or 11dehydrocorticosterone) is equipotent with or even more potent than cortisol (or corticosterone) (206, 231, 646). 11β-Reductase predominance was also indicated in humans by in vivo observations of a high cortisol-to-cortisone ratio across the hepatic circulation (liver only expresses 11β-HSD1) (743). Thereafter, 11β-HSD1 action in a host of primary cell types was shown to be predominantly glucocorticoid regeneration and dependent on NADP(H) cosubstrate (FIGURE 2). Located inside the inner leaflet of the endoplasmic reticulum, 11β-HSD1 interacts with hexose-6-phosphate dehydrogenase (H6PDH), the major generator of endoplasmic reticulum NADPH, which in turn drives 11β-reduction (40, 41, 54, 97, 177) (FIGURE 3, and see below). Loss of H6PDH leads to reaction reversal of 11β-HSD1, although the physiological relevance remains uncertain (389, 391).

Figure 3.

Cartoon of the possible intracellular relationships of 11β-HSDs. Despite having a lower affinity for F (or corticosterone) than MR, 11β-HSD2 is able to successfully exclude glucocorticoids from MR. While the basis for this is unknown, a possible scenario invokes lipophilic steroids preferentially localizing to the membranes inside cells including those of the endoplasmic reticulum (ER). 11β-HSD2 is located on the cytosolic surface of the ER and is in close association with the MR complex. Therefore, F may perhaps have to pass via 11β-HSD2 before it can gain access to MR. If there is sufficient enzyme and its turnover is suitably fast, it may successfully form the biochemical equivalent of an anatomical moat around MR. In addition, some data suggest that the 11-dehydrocorticosteroid products of 11β-HSD2 (cortisone/E; 11-dehydrocorticosterone) functionally antagonize aldosterone activation of MR. 11β-HSD1 is bidirectional in homogenates and microsomal preparations, but a predominant reductase in intact cells and in vivo. 11β-HSD1 is located inside the inner leaflet of the ER in close association with hectose-6-phosphate dehydrogenase (H6PDH), a powerful generator of NADP(H). NADP(H) drives the 11β-reductase direction of 11β-HSD1 and maintains this in many cell types. Other redox active processes may be important in organs such as brain where H6PDH may be at low levels.

More recently, the existence of a third 11β-HSD isozyme has been suggested. An NADP⁺-dependent dehydrogenase activity in sheep kidney was first dubbed 11β-HSD3 (238), but this has not been further characterized and may represent 11β-HSD1 or -2. It is unlikely that this is the same as the discovered evolutionary ancestor of 11β-HSD1, confusingly called 11β-HSD3 (51), but also SDR26C2 according to accepted nomenclature (557), HSD11B1L and SCDR10. 11β-HSD3/SDR26C2 is present in humans, with orthologs in other species including fish, but is absent from rats and mice (299). 11β-HSD3/SDR26C2 only poorly catalyzes conversion of cortisol to cortisone (in an NADP⁺-dependent manner) (299). Indeed, it predates the glucocorticoid receptor in evolution (49), implying a primary function distinct from glucocorticoid biology.

E. Intracrine Versus Endocrine Effects

It is helpful to consider the effects of enzymes such as 11β-HSDs in two contexts: intracrine and endocrine. Intracrine effects are exemplified by the intracellular “gating” of glucocorticoid action by 11β-HSD2 within cells of the distal nephron. 11β-HSD2 modulates the activation of intracellular receptors by a ubiquitous ligand but only where receptors and enzyme are colocalized (395). This occurs without altering circulating levels of cortisol, itself largely determined by activation of, and glucocorticoid feedback upon, the hypothalamic-pituitary-adrenal (HPA) axis. Thus patients with AME have intense MR activation by cortisol within specific cells of the kidney but without change in systemic cortisol levels, an intracrine effect. Similarly, 11β-HSD1 acting as a reductase appears to amplify glucocorticoid levels inside expressing cells in adipose tissue and brain (494, 791), increasing local glucocorticoid levels and thus the signal via GR and, if present, MR.

However, these enzymes also contribute to the bulk turnover of glucocorticoids, with 11β-HSD1 in the splanchnic bed generating ∼30–40% of the total daily production of cortisol in humans (26, 62) and 11β-HSD2 in the kidney inactivating a similar proportion. These actions affect the turnover of glucocorticoids and may per se alter HPA axis function under some circumstances. Moreover, expression of 11β-HSDs in cells involved in glucocorticoid feedback control of the HPA axis could also alter circulating hormones (179, 622, 624). These are endocrine effects.

F. Measurement of 11β-HSDs

Differentiating 11β-HSD isozymes at the level of mRNA is straightforward, given their distinct sequences. In contrast, estimating 11β-HSD1 protein and activity can be more challenging yet is crucial to explore their biology, especially in humans. As discussed below, antisera for 11β-HSD1 and -2 are often species-specific (not surprising given the sequence and substrate differences) and probably cross-react with related members of the extensive short-chain dehydrogenase/reductase (SDR) protein family, although some sera perform well in practice. Ex vivo bioassays are well-established; both isozymes can readily be measured in tissue homogenates or microsomal preparations (418, 479). These are uncomplicated for measuring 11β-HSD in relevant samples and should be even more specific when coupled with selective 11β-HSD1 inhibition to distinguish isozyme activities in dehydrogenase assays. However, because 11β-HSD1 is bidirectional in tissue homogenates and microsomes, assays in semi-purified preparations will determine only the amount of active protein and not its reaction direction, unlike 11β-HSD2 which is essentially a unidirectional dehydrogenase with cortisol or corticosterone as substrates. Assays in intact cells can distinguish isozymes (91), but do not necessarily determine 11β-HSD1 reaction direction in vivo if this is dynamically regulated by intracellular energy levels, as detailed below.

In humans, some tissues (leukocytes, adipose, skeletal muscle, sometimes liver) can be directly assayed ex vivo from blood samples/biopsies, but for most organs indirect techniques have been needed to determine 11β-HSD activity and the reaction catalyzed by 11β-HSD1 in vivo. Initially, radioimmunoassay and then mass spectrometry approaches have been developed to estimate corticosteroid metabolites in urine (typically over 24 h to remove any distortion from circadian rhythms, etc.) (719). Cortisol (Kendal's compound F) is metabolized by 11β-HSD2 to cortisone (E). Both cortisol and cortisone are then subject to 5β-reduction (by 5β-reductase largely in liver) to form dihydrocortisol (DHF) and dihydrocortisone (DHE). Cortisol is also metabolized by 5α-reductase type 1 to allo-dihydrocortisol (aDHF), again largely but not exclusively in liver. The dihydro-metabolites are then further reduced by near ubiquitous 3α-hydroxysteroid dehydrogenases to form the tetrahydro metabolites (THF, aTHF, and THE) (745). Urinary ratios of THF+aTHF/THE give a convenient measure of combined 11β-HSD1 and -2 activity. A normal ratio of near unity becomes markedly elevated in AME patients with some correlation between the ratio and phenotype (491).

There are some caveats to the widespread use of urinary steroid ratios: 1) it measures both 11β-HSDs; 2) the ratio also has components of 5α- and 5β-reductase activities which, if altered, will impact upon the ratio; and 3) the ratio does not address tissue-specific impacts. In a refinement, urinary free cortisol (UFF):urinary free cortisone (UFE) have been advocated as selective for 11β-HSD2 (542), leaving the THF+aTHF/THE to reflect 11β-HSD1 when UFF:UFE is unaltered, but this assumption requires comprehensive validation. Whilst selective 11β-HSD1 inhibitors reduce the THF+aTHF/THE ratio within subjects, using this as an absolute measure of whole body 11β-HSD1 activity in population samples is a blunt and probably ineffective tool. Thus the ratio correlates positively (24), negatively (574), and not at all (23, 762) with obesity. Obesity and ageing also alter 5α- and 5β-reductases adding to the complexity and well-illustrating the limitations of this approach (745).

These indirect estimates have long been supplemented with administration of radio-labeled cortisol tracer to estimate bodily conversion to cortisone (668, 719). Again this works well for 11β-HSD2 mutations, but is otherwise complicated when trying to take into account recycling of cortisone back to cortisol by 11β-reductase.

To directly estimate 11β-reductase in the splanchnic bed (gut, mesentery, liver), oral cortisone (315, 574, 671) and prednisone (712) tests have been developed. These use first pass hepatic metabolism to measure the rate of appearance of cortisol and prednisolone in peripheral blood. This appears linear over 15–30 min, although with longer sampling, cycling through both 11β-HSDs and downstream metabolism complicate interpretation. Rigorous determination of the relationship between the output of some of these bioassays and the actual level of 11β-HSD1 in the splanchnic bed has not been reported as yet and may not be linear, a problem when using such data to assess the degree of inhibition by novel agents.

To estimate 11β-reductase and 11β-dehydrogenase more directly, quadruply deuterated (D4)-cortisol (9,11,12,12-[²H₄]cortisol) has been administered. 11β-HSD2 removes the deuterium on C11 creating D3-cortisone. This is a substrate for 11β-HSD1 which reduces D3-cortisone to D3-cortisol which can be distinguished from D4-cortisol by mass spectrometry (25, 62, 612). In addition to whole-body 11β-HSD activity, this can be used, if somewhat more invasively, to determine turnover of cortisol and cortisone in individual organs (59, 303, 674).

Finally, 11β-reductase can also be assessed in human subcutaneous adipose tissue by microdialysis, infusing [³H]cortisol and measuring conversion to [³H]cortisone in the dialysate. While attractive conceptually, and responsive to acute regulation (459, 612, 712, 740, 741), this is a challenging technique.

Overall, with care and attention to detail and controls, 11β-HSD activities can be measured with some accuracy and consistency in humans in vivo.

G. Investigation of the Physiological Functions of 11β-HSDs

Study of the physiological function of 11β-HSDs began with nonselective, licorice-based inhibitors such as glycyrrhetinic acid and its hemisuccinate carbenoxolone (FIGURE 7), a drug historically used to treat peptic ulcers (79). These compounds inhibit both isozymes with a low nanomolar K_i in vitro. Potency is lower in vivo, and the drugs also inhibit related enzymes such as 15-hydroxyprostaglandin dehydrogenase (697) and gap junctions (235, 252, 254), although with lower potency. Why the root of the licorice plant, Glycyrrhiza glabra and related species, produces a triterpenoid which so potently inhibits 11β-HSDs is unclear, but its production is sensitive to the environment (356), so perhaps these substances promote plant survival and parasite resistance.

Figure 7.

Structures of nonselective licorice-based 11β-HSD inhibitors, glycyrrhetinic acid and carbenoxolone, BVT2733, a synthetic selective 11β-HSD1 inhibitor, and the natural cortisone substrate for 11β-HSD1.

Dissection of 11β-HSD function came of age with the development of knockout mouse models, 11β-HSD1 in 1997 and 11β-HSD2 in 1999 (363, 364). Subsequently, a variety of transgenic lines and more recently selective 11β-HSD1 inhibitors and tissue-specific genetic manipulations have expanded the weaponry to dissect function and pathophysiology. These studies are described below in an organ and system approach.

Here we review the intriguing biology of 11β-HSDs and highlight their role in pathogenesis through the lifespan.

II. 11β-HSD1 GENE STRUCTURE, ENZYMOLOGY, AND REGULATION

A. Gene Structure and Control

11β-HSD1 is found in land vertebrates and sharks, but is absent from fish, likely arising from an ancestral 11β-HSD3/SDR26C2 gene (51). The encoding gene, HSD11B1, is located close to the end of the long arm of chromosome 1 in mice and humans and chromosome 13 in rats. It comprises seven exons; exons 2–7 encode the full-length protein (FIGURE 4). The gene is transcribed from three promoters: P1, P2, and P3 (95, 475); P2-initiated transcripts predominate in most tissues including liver, brain, and adipose (95, 663). P1-initiated transcripts are the majority in mouse lung, but the minority in human lung (663) and kidney (our unpublished data). The P3 transcript, discovered in rat kidney (and negligible in other tissues), initiates within the intron between exons 2 and 3 (475). It encodes an expressed NH₂-terminally truncated 11β-HSD1 protein that lacks 11β-HSD activity (462). Its biological relevance (if any) is unclear.

Figure 4.

Structure of the human HSD11B1 gene, associated promoters and relevant transcription factor binding sites. Schematic representation of the HSD11B1 gene (not to scale). Exonic sequences are shown as boxes; white boxes encode the open reading frame with red boxes and a blue box indicating the 5′ leader (dependent on the promoter used) and the 3′ untranslated sequence, respectively. Arrows indicate the positions of the 3 promoters. The gene spans 30 kb with a large intron (∼25 kb) between exons 5 and 6. The position of 2 conserved C/EBP binding sites, located between −112 and −160 in the promoter (244, 775), are shown. These sites are bound by C/EBPα and/or C/EBPβ in hepatocytes and adipocytes and are implicated in HSD11B1 regulation by glucocorticoids, cAMP, ceramide, AMPK, and DHEA. See text for details.

The predominant P2 promoter of hsd11b1 contains several binding sites for the CCAAT/enhancer binding protein (C/EBP) family of transcription factors (775). Two C/EBP sites close to the transcription start and highly conserved between human, rat, and mouse (244, 775) are critical for basal and regulated expression (775), either alone or in concert with other transcription factors including GR (789).

In vivo, transcription from P2, at least in liver and brown adipose tissue, requires C/EBPα (95, 775), whereas transcription from P1 in lung is independent of C/EBPα (95). While C/EBPβ is an activator of Hsd11b1 expression in adipocytes and C/EBPβ knockout mice have reduced adipose 11β-HSD1 expression (552), C/EBPβ is only a weak activator in hepatoma cells and acts as a relative repressor of 11β-HSD1 expression in liver, possibly because of heterodimerization with the strong activator C/EBPα (775). A decreased C/EBPα-β ratio is implicated in dehydroepiandrosterone (DHEA) repression of Hsd11b1 in liver and adipose tissue (30). Involvement of other C/EBP family members has not been found as yet (190, 604).

Importantly, C/EBPβ mediates 11β-HSD1 (P2) regulation by proinflammatory cytokines, glucocorticoids, diet, and other regulators (cAMP, ceramide, AMP-activated protein kinase) in a variety of cell types including adipocytes, fibroblasts, and lung epithelial cells (31, 244, 312, 604, 790), although C/EBPα mediates glucocorticoid effects in human amnion fibroblasts (789). C/EBPβ is a central regulator of inflammation and metabolism (35), suggesting that HSD11B1 might play an important downstream role in these pleiotropic effects. C/EBPβ is itself glucocorticoid induced, including during inflammation (64, 807) and adipocyte differentiation (708). This indicates a possible feed-forward loop-inflammation stimulating the HPA axis to secrete glucocorticoids, both of which induce C/EBPβ in turn increasing 11β-HSD1, further amplifying local glucocorticoid signaling and shaping the trajectory of an inflammatory response via pro-resolution macrophage polarization (600). A recent intriguing twist is added by the discovery that the C/EBPβ posttranslational isoforms, liver inhibitory protein (LIP) and liver activator protein (LAP), play distinct physiological roles (647, 763) and oppose each other in mediating 11β-HSD1 regulation in adipose tissue by high-fat diet (190), itself linked to adipose inflammation (750).

Beyond C/EBPs, indirect regulatory effects appear to predominate (see regulation below). Thus HNF1α occupies the HSD11B1 promoter in human hepatocytes (532) and is functionally relevant since mice lacking HNF1α have negligible hepatic 11β-HSD1 mRNA (639). However, this could be an indirect effect mediated via alterations in bile acids; bile duct ligation reduces hepatic 11β-HSD1 activity and mRNA levels in rats (2, 189), and HNF1α is a key regulator of bile acid and HDL-cholesterol metabolism. In contrast, although HNF4α occupies the HSD11B1 promoter in human hepatocytes (532), any function is unclear. The wealth of data emerging from chromatin immunoprecipitation experiments, coupled to microarray or RNA sequencing, will clarify the molecular mechanisms regulating HSD11B1.

B. 11β-HSD1: Protein Structure

In most species, including humans and mice, 11β-HSD1 (also called SDR26C1; Ref. 557) comprises 292 amino acids with a predicted molecular mass of 34 kDa. 11β-HSD1, like 11β-HSD2, is a member of the large family of SDR (557, 778). These show relatively low overall identity (18% sequence identity between 11β-HSD1 and 11β-HSD2) but share a structurally conserved nucleotide cofactor-binding Rossmann-fold in the NH₂-terminal region (Thr-Gly-[Xaa]₃-Gly-Xaa-Gly) and an invariant Tyr-[Xaa]₃-Lys motif in the active site (Tyr183 and Lys187 in human 11β-HSD1 and most other species). Tyr183 and Lys187 sit with the conserved Asn143 and Ser170 in the catalytic site and are essential for the proton transfer between substrate and NAD(P)H cofactor: mutation of either inactivates 11β-HSD1 (526), whereas mutation of nearby serines has little effect on K_m but markedly reduces the rate of catalysis (528).

Within cells, tagged fusion proteins have localized 11β-HSD1 to the inner leaflet of the endoplasmic reticulum (ER) (507), confirming earlier structural analyses showing 11β-HSD1 is a high mannose glycoprotein typical of the ER lumen (541). A single transmembrane helix in the NH₂ terminus of 11β-HSD1 spans the ER membrane and dictates the orientation of the protein within the membrane, with the NH₂-terminal end in the cytoplasm and the bulk of the protein within the ER lumen (204, 530). The luminal orientation is maintained by Lys5 and/or Glu25/Glu26 (204, 530). Surprisingly, mutant 11β-HSD1 with cytoplasmic orientation retained reductase activity with cortisone, but not with 7-ketocholesterol (204, 530), yet cortisol oxidation (low in most intact cells) was abolished. The source of cytoplasmic cofactor for reductase activity was not clear, although it should be noted that these experiments were performed in intact HEK293 cells which, when transfected with 11β-HSD1, show both reductase and dehydrogenase activity due to low levels of endogenous H6PDH (41).

Functional 11β-HSD1 most probably comprises a homodimer (442, 533, 804), although tetrameric structures are possible (295). The two subunits may be linked by a disulfide bond (541), although this is not supported by the crystal structures (533, 804). The human enzyme is glycosylated at three asparagine residues (Asn123, Asn162, and Asn207) (81). Asn207 is invariant in 11β-HSD1 sequences, Asn162 is highly conserved, and Asn123 shows little conservation. Thus there are only two N-linked glycosylation sites in rat and mouse 11β-HSD1 (Asn158 and Asn203 in rat, equivalent to Asn162 and Asn207, respectively, in the human enzyme) (9) and just one in guinea pig. Despite an early study suggesting glycosylation is important for activity (5), the current consensus is that it plays little or no role in activity or reaction direction: enzymatic removal of glycan chains has no effect (9, 541) and human 11β-HSD1 expressed in yeast retained full activity despite lacking glycosylation (81). Similarly, our own experiments have shown no effect of mutation of Asn162 or Asn207 to glutamine on either activity or reaction direction of human 11β-HSD1 in intact cells (unpublished data).

Two human patients with cortisone reductase deficiency have been reported. These subjects are heterozygous for mutations in the coding sequence of HSD11B1. One mutation (Arg137Cys) disrupts salt bridges at the dimerization interface and reduces enzyme activity, and the other (Lys187Asn) interrupts the active site and not surprisingly abolishes 11β-HSD1 activity altogether (393). The mutants appear to exert dominant negative effects on the normal allelic product.

C. Crystal Structure of 11β-HSD1

To date, 26 11β-HSD1 crystal structures have been deposited in the Protein Data Bank (see Ref. 704 for a recent review). Of these, three are mouse 11β-HSD1 (without substrate, with corticosterone bound, and with selective inhibitor bound), three are guinea pig 11β-HSD1 (one without substrate and two with inhibitor bound), and the remainder are human 11β-HSD1 with a variety of inhibitors bound, reflecting the strong pharmaceutical interest in 11β-HSD1 inhibition. All have NADP(H) bound in the cofactor binding site, consistent with the ordered bi-bi sequential enzymatic mechanism proposed by Monder (481), in which NADPH cofactor binds first, followed by substrate binding and catalysis, with cofactor leaving last, confirmed in more recent studies (603).

The 11β-HSD1 structures are similar overall, although the hydrophobic substrate binding pocket, predominantly lined by nonpolar residues and with lower sequence conservation than other regions, shows some variability between species, accounting for the species specificity of inhibitors (32, 626, 704). The two subunits of the 11β-HSD1 homodimer are related by a pseudo-twofold axis. All structures show a central seven-stranded parallel β-sheet sandwiched between six parallel α-helices (the Rossmann-fold; FIGURE 5). The NH₂ termini of the two subunits are on the same face of the dimer and point in the same direction, anchoring the protein within the membrane.

Figure 5.

Structure of human 11β-hydroxysteroid dehydrogenase in complex with NADP and carbenoxolone. Structure of a dimer of human 11β-HSD1 with NADP cosubstrate (bottom middle) and the inhibitor carbenoxolone (top right) bound. [Image from the RCSB PDB (www.pdb.org) of PDB ID 2BEL (339).]

The COOH-terminal part of 11β-HSD1, unique among SDR proteins, is involved in dimer interactions, contributes to substrate binding and active site architecture, and is important for membrane lipid bilayer interactions through a large hydrophobic surface (533, 804). The crystal structures suggest it funnels substrate from the lipid bilayer into the active site cleft, driving catalysis. This facilitated entry into the active site may explain the discrepancy between the relatively high K_m values measured in vitro with hydrophobic glucocorticoid or oxysterol substrates and the efficient substrate conversion at low concentrations by intact cells (140).

Both ends of the substrate binding pocket are open to solvent, allowing ligands that exceed the 12 Å length of the pocket to extend out of it (704). Interactions with Ser170 and Tyr183 stabilize the 11-keto group of the substrate, anchoring it in the active site (331, 533). Protonation of the reactive keto oxygen of the substrate is catalyzed by Tyr183, which functions as a catalytic acid, with Lys187 lowering the pK_a of the Tyr183 hydroxyl. Concomitantly, hydride transfer takes place from the nicotinamide ring to the C11 position of steroid substrate. Hydrogen exchange with bulk solvent reprotonates Tyr183 via a conserved relay system which includes Asn143, the 2'-OH of the nicotinamide ribose, and a water molecule (533).

D. 11β-HSD1: Reaction Direction and Cosubstrate

11β-HSD1 is predominantly an oxo-reductase in most intact cells and in vivo (FIGURE 1) and is the only enzyme able to reduce 11-keto-glucocorticoids, at least in mice (364). The reported K_m of 11β-HSD1 for its glucocorticoid substrates ranges from ∼150 nM to almost 20 μM (379, 442), higher than the low nanomolar free circulating substrate levels. However, in intact cells, 11β-HSD1 is clearly functionally relevant even at low (e.g., 2 nM) substrate levels (231). In mice, there is a circadian rhythm in plasma 11-dehydrocorticosterone levels that mirrors corticosterone, with basal levels 2–7 nM, but rising markedly (>25 nM) following acute stress (265). Humans show a similar diurnal rhythm in cortisone levels, varying from a nadir of 10–20 nM to a peak of ∼60 nM in men (519). As cortisol, cortisone levels are increased and the diurnal rhythm flattened in patients with depression (757); such elevated cortisone levels imply increased glucocorticoid action in brain cells expressing 11β-HSD1, potentially contributing to pathogenesis.

In tissue homogenates, 11β-HSD1 is bidirectional. Initially it was postulated that liver 11β-HSD (11β-HSD1) comprised two distinct enzymes (377, 379). The literature was further confused by the inability adequately to distinguish 11β-HSD1 and 11β-HSD2 activities; 11β-HSD1 can use either NAD(H) or NADP(H) as cofactor (36, 437). This was resolved by two discoveries: 1) there are two 11β-HSD isozymes, and 2) the predominant oxo-reductase activity of 11β-HSD1 in intact cells is due to its colocalization with H6PDH (41, 97).

1. H6PDH determines 11β-HSD1 reaction direction

H6PDH physically and functionally interacts with 11β-HSD1 within the ER lumen (54), and the two enzymes copurify (541) (FIGURE 3). H6PDH catalyzes the first two steps of the ER pentose-phosphate pathway (distinct from the cytosolic pathway), converting glucose-6-phosphate to 6-phosphogluconolactone and concomitantly reducing NADP⁺ to NADPH (693). H6PDH thus generates the high NADPH/NADP⁺ ratio inside the ER lumen required for 11β-HSD1 to function efficiently as a reductase (177). In human omental preadipocytes which lack H6PDH, 11β-HSD1 is a dehydrogenase (101), whereas in mouse visceral preadipocytes which express some H6PDH, reductase activity predominates (146).

In mice and humans, inactivating mutations in H6PDH switch 11β-HSD1 activity from oxo-reduction to dehydrogenation (389, 390) and are causal in the human disorder “apparent cortisone reductase deficiency” (390). Interestingly, mutations which inactivate 11β-HSD1 itself, cortisone reductase deficiency (393), produce a milder phenotype because, although 11β-reductase activity is lost, there is no “gain” of dehydrogenase activity which occurs with H6PDH deficiency.

The ER is impermeable to NADP(H). Thus its generation within the ER by H6PDH is dependent on the intraluminal concentration of glucose-6-phosphate which readily enters the ER via the glucose-6-phosphate transporter (693). 11β-HSD1 reaction direction is therefore coupled to cellular energy (glucose) levels. Indeed, 11β-HSD1 reductase activity is decreased in humans and mice with mutations in the glucose-6-phosphate transporter (glycogen storage disease, type 1b) in whom ER glucose-6-phosphate levels are reduced (748). Conversely, 11β-reductase activity is markedly increased in patients with glycogen storage disease type 1a caused by deficiency of glucose-6-phosphatase-α that, in liver, competes with H6PDH for glucose-6-phosphate (748). Fructose-6-phosphate converted to glucose-6-phosphate by an ER hexose-6-phosphate isomerase (628), also increases 11β-reductase activity in liver microsomes (449, 628).

Glucose levels probably regulate 11β-HSD1 activity, but whether they alter reaction direction is uncertain. Lowering extracellular glucose levels (from 25 mM to a more physiological 6 mM) switches 11β-HSD1 activity from reductase to dehydrogenase in transfected cells. However, in hepatoma and adipocyte cell lines which express endogenous H6PDH, 11β-reductase activity predominates irrespective of extracellular glucose levels (177). Conversely, rat Leydig cells, which express ample H6PDH (241), switch 11β-HSD1 activity from reductase to dehydrogenase when glucose concentrations fall (196). The extent to which this happens in vivo is unclear. In the perfused rat liver (with 11 mM glucose), 11β-reductase activity predominates (319). Even in starvation, which reduces intraluminal NADPH/NADP⁺ in rat liver, reductase activity still predominates though 11β-dehydrogenase activity is increased (341). Clearly understanding of the coupling between glucose, ER luminal redox, and 11β-HSD1 reaction direction is incomplete.

2. But there may be more than H6PDH?

Although H6PDH is often coexpressed with 11β-HSD1 (241), there is an apparent lack of complete congruity even in cells clearly having 11β-reductase. In kidney, 11β-HSD1 but not H6PDH is expressed in medullary interstitial cells. In the CNS, the expression of H6PDH and 11β-HSD1 generally differs (241), contrasting with the exclusive 11β-reductase activity in brain (572) that is indeed unusually stable (377, 379, 380). Finally, as mentioned above, an 11β-HSD1 mutant directed to the cytoplasm retains 11β-reductase activity with cortisone and loses cortisol oxidation altogether (204, 530). Thus whether or not H6PDH is the exclusive supplier of reducing equivalents to 11β-HSD1 in all tissues is currently unclear; it is possible that ER-localized isocitrate dehydrogenase may also provide NADPH to 11β-HSD1 (430). Dissecting the relative importance of this and other possible pathways generating luminal NADPH in brain, adipose, and other tissues is of key interest, especially as drugs inhibiting 11β-reductase are in clinical trials.

3. 11β-HSD1: nonglucocorticoid substrates

11β-HSD1 is a promiscuous enzyme. It has broad substrate (and inhibitor) specificity (TABLES 1 AND 2). Moreover, differences in the substrate-binding pocket result in considerable species differences in 11β-HSD1 substrate specificity and reaction kinetics. In addition to its role in glucocorticoid metabolism, which includes the synthetic glucocorticoids prednisolone and betamethasone, but not dexamethasone (154), 11β-HSD1 catalyzes other steroid and sterol interconversions. It also detoxifies xenobiotics (440).

Table 1.

Selected substrates of 11β-HSD1

Substrate	Apparent Km, μM	Species	Source	Reference Nos.
Cortisol	10–50	Human	Purified or expressed in yeast	305
Cortisol	300	Mouse	Purified liver protein	437
Corticosterone	6–40	Human	Purified liver protein	305
Corticosterone	2–7	Mouse	Liver or lung microsomes	127, 435
Cortisone	2–40	Human	Purified, liver microsomes or expressed in yeast	305
Cortisone	10	Mouse	Purified liver protein	438, 443
11-Dehydrocorticosterone	0.3–20	Human	Purified liver protein, transfected cell lysate, or expressed in yeast	305
11-Dehydrocorticosterone	1–40	Mouse	Purified liver protein or liver/lung microsomes	127, 435, 438, 443
NNK	630	Mouse	Lung microsomes	435
7-Ketocholesterol	0.4–50	Human, mouse	Transfected cell lysates, purified recombinant proteins from yeast	304, 621
7-KetoDHEA	1	Human	Expressed in yeast	497
7-Keto-5α-androstane-3β,17β-diol	7	Human	Expressed in yeast	277
7-Ketoepiandrosterone	0.5	Human	Expressed in yeast	278
Metyrapone	370	Human	Expressed in yeast	306
Ketoprofen	20	Human	Expressed in yeast	306
Prednisone	21	Human	Expressed in yeast	305

For additional substrates, see Reference 107. Also see Reference 107 for ability of 11β-HSD1 to metabolize clinically relevant glucocorticoids. Values for transfected cells are ascertained in lysates.

Table 2.

Selected inhibitors of 11β-HSD1

Inhibitor (of Glucocorticoid Metabolism)	Ki, μM	Species	Source	Reference Nos.
Carbenoxolone	0.02–0.3	Human	Liver microsomes or expressed in yeast	305
Carbenoxolone	0.1	Mouse	Expressed in yeast	55
Glycyrrhetinic acid	0.04–2	Human	Liver microsomes, transfected cells, or expressed in yeast	305
Progesterone	2	Human	Liver microsomes	155
11β-Hydroxyprogesterone	0.4	Human	Liver microsomes	155
7-Ketocholesterol	8
7-Ketopregnenolone	0.7	Human	Transfected cells	515
Deoxycorticosterone	4	Human	Liver microsomes	155
Dexamethasone	8	Human	Expressed in yeast	305
Budesonide	60	Human	Expressed in yeast	305
7-Keto-DHEA	1	Human	Human skin microsomes	276, 515
CDCA	4	Human	Liver microsomes	155
Lithocholic acid	3	Human	Liver microsomes	155
Metyrapone*	3,000*	Human	Liver microsomes	155
Ketoconazole*	>10*	Human	Liver microsomes	155
Flavanone	18–21	Human	Transfected cells	620
Abietic acid	5–27	Human	Transfected cells	620
BVT14255	0.052	Human	Transfected cells	55
BVT14255	0.28	Mouse	Transfected cells	55
BVT2733	3.34	Human	Transfected cells	55
BVT2733	0.096	Mouse	Transfected cells	55
Merck-544/T0504	0.008+–0.015	Human	Transfected cells	282
Merck-544/T0504	0.08–0.097+	Mouse	Transfected cells	282

For additional inhibitors, see Reference 811. K_i values are in μM and normally measured for inhibition of cortisone reduction. Values for transfected cells are ascertained in lysates.

^*Inhibits reductase only. ⁺IC₅₀ values. Merck-544 is the same compound as T0504.

A) 7-STEROLS.

The metabolism of 7-oxygenated sterols and steroids by 11β-HSD1 probably reflects the rotational symmetry of the 11β- and 7α- positions of the steroid nucleus (386) (FIGURE 1). 11β-HSD1 interconverts 7-keto- and 7-hydroxy-DHEA (both bind with higher affinity to 11β-HSD1 than do glucocorticoid substrates; Refs. 497, 515), 7-keto- and 7-hydroxy-pregnenolone (515), and 5α-androstane-3β-ol-7,17-dione and 5α-androstane-3β-ol-7-hydroxy,17-one (515). Additionally, 11β-HSD1 may act as an epimerase, interconverting 7α- and 7β-hydroxy-steroids via a 7-keto intermediate (277, 278, 497). Epimerase activity requires oxidative and reductive activities and therefore may be determined by H6PDH levels and glucose availability (515).

In humans, rats and mice (though not dog or guinea pig; Ref. 32), 11β-HSD1 interconverts 7-ketocholesterol and 7β-hydroxycholesterol (304, 621) (hamster 11β-HSD1 produces both 7β- and 7α-hydroxycholesterol; Ref. 32). The 7-keto-sterol substrates competitively inhibit glucocorticoid metabolism, and vice versa, with glucocorticoid substrates inhibiting oxysterol reduction (53, 515, 582, 749), suggesting the oxysterols could be important modulators of intracellular glucocorticoid levels in tissues where they accumulate to significant levels, with liver, brain and adipose tissue being the most obvious sites. This also has the potential to impact 11β-HSD1 reaction direction when H6PDH is limiting; 7-ketocholesterol represses whereas 7β-hydroxycholesterol enhances cellular glucocorticoid activity at GR, an effect alleviated by increasing NADPH supply to 11β-HSD1 through over-expression of H6PDH (749).

B) BILE ACIDS.

Other substrates are emerging. Human 11β-HSD1 unidirectionally converts the secondary bile acid 7-oxo-lithocholic acid (generated by gut micro-organisms) to chenodeoxycholic acid and, to a lesser extent, ursodeoxycholic acid (531). Whether this reaction is relevant to bile acid hepatotoxicity and cholestasis merits attention, but the potential for ursodeoxycholic acid in therapy of primary biliary cirrhosis and as an anti-inflammatory (543) supports some involvement.

C) CARBONYL REDUCTION.

Carbonyl reductases play key roles in the detoxification of reactive intermediary metabolites and xenobiotics (e.g., drugs, insecticides, carcinogens) containing carbonyl (aldehyde or ketone) groups (537). In mouse liver, 11β-HSD1 shows carbonyl reductase activity with metyrapone, an inhibitor of 11β-hydroxylase and thus adrenocortical glucocorticoid synthesis (436, 439). As other nonglucocorticoid substrates, metyrapone competes with glucocorticoids for metabolism by 11β-HSD1 (155, 609). Other xenobiotics metabolized by 11β-HSD1 include, in lung, nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (441), a potent nitrosamine carcinogen in tobacco. 11β-HSD1 also metabolizes triadimefon, a broad-spectrum conazole fungicide used extensively in agriculture (340). In liver, 11β-HSD1 metabolizes the anti-inflammatory drug ketoprofen and activates the pro-drug DFU-lactol (306). Activity on other drugs and xenobiotics is probable.

This broad substrate specificity of 11β-HSD1, with its role in detoxification reactions as well as potential activation of neurosteroids such as DHEA and pregnenolone, has implications for the use of inhibitors. Indeed, some of the beneficial cognitive (613, 660, 791, 794) and atheroprotective (282, 493) effects of 11β-HSD1 deficiency or inhibition may, in part, reflect decreased metabolism of nonglucocorticoid substrates (792).

E. 11β-HSD1: Distribution

In adult mammals, 11β-HSD1 is widely distributed. Roughly similar distributions are reported in rodents, non-human primates, and humans (4, 478, 695). Expression is highest in the liver (478, 695). In addition, 11β-HSD1 mRNA and protein/enzyme activity occur in adipose tissues (100), vasculature (747), ovary (68), testis (in some species including human and rat; Refs. 558, 695), but not mouse or squirrel monkey (486, 571), brain (380, 477, 478), uterus (103), placenta (notably in the decidua) (734), immune and inflammatory cells (231), skeletal muscle (767), and heart (747). 11β-HSD1 is not expressed in fetal life until late in gestation when levels are greatest in organs where glucocorticoid activity is required for late maturation prior to birth, notably lung and liver (662).

A problem with tissue localization of 11β-HSD1 reflects the difficulties in generating specific antisera. Early efforts gave a major 34-kDa band on western blots, compatible with the cDNA-predicted molecular weight plus known glycosylation (9, 541). A 68-kDa band is reasonably assumed to be a dimer. However, other higher and lower molecular weight bands were prominent and were not unequivocally 11β-HSD1 (380). More recent antisera also yield additional bands (738). Their use to localize 11β-HSD1 by immunocytochemistry remains complicated by issues of specificity, although preabsorption with purified antigen suggests this may be feasible in specific tissues (477). Similar caveats apply to commercially available antisera, generally raised to peptides rather than the purified or recombinant proteins. Allowing this caveat, 11β-HSD1 mRNA and immunoreactivity have been localized in a variety of cell types. Usefully, such localization can be coupled with 11β-HSD1 activity assays to show functional protein, crucial given the presence of alternative transcripts encoding inactive protein in some organs (kidney) and the sensitivity of PCR which may detect low levels of transcripts in almost any cell, whether or not this “expression” is of biological significance.

F. 11β-HSD1 Regulation

Many factors, endocrine, inflammatory, and metabolic, regulate 11β-HSD1. This is often in a temporal and tissue-specific manner. For example, proinflammatory cytokines, notably interleukin (IL)-1 and tumor necrosis factor (TNF)-α, potently upregulate 11β-HSD1 mRNA in a variety of cells, although interestingly not in immune cells (discussed in sect. V).

Glucocorticoids themselves increase 11β-HSD1 expression in many cells (99, 130, 183, 259, 316, 572, 680, 681, 768, 789), probably via C/EBPβ (604). In vivo, adrenalectomy lowers liver and hippocampal 11β-HSD1 expression in the short-term, but enzyme activity subsequently recovers (317), much like GR itself (202, 292). Conversely, elevated glucocorticoids in intact animals increase 11β-HSD1 in some tissues (adipose tissue, perinatal lung) but not in others (liver) (52, 465, 650).

The effects of insulin on 11β-HSD1 are less clear. In vivo, insulin-induced perturbations of metabolism complicate interpretation. Acutely, in human adipose tissue insulin increases 11β-HSD1 activity, perhaps via changes in glucose availability (194), without changes in 11β-HSD1 mRNA (355, 612, 740). Insulin has no effect on renal 11β-HSD1 in diabetic rats (411). In vitro, high levels of insulin downregulate 11β-HSD1 expression (259, 316, 410, 731) probably through insulin-like growth factor I (IGF-1) receptors, whereas lower concentrations of insulin (too low to bind IGF-1 receptors) have no effect or modestly increase 11β-HSD1 expression in adipocytes (52, 510). Interpretation is confounded by the use of high levels of glucose, which mimic an insulin-resistant state, and the mitogenic effects of insulin in culture which may decrease 11β-HSD1: there is negligible 11β-HSD1 in most proliferating cell lines.

Hepatic 11β-HSD1 is sexually dimorphic in rodents, with lower levels in females. In rats, this is underpinned by sex-specific patterns of pituitary growth hormone (GH) secretion (419). How GH secretion patterns alter 11β-HSD1 remains unclear. It is probably not a direct effect of GH on hepatocytes (316), but could be mediated by IGF-I (402, 410, 510, 711, 731, 768). Repressive effects of GH on 11β-HSD1 in humans have been inferred from patients with hypopituitarism or acromegaly (225, 487) but may be mediated via other hormones or intermediary metabolism.

Testosterone has no effect on 11β-HSD1, at least in liver and kidney (239, 418). However, estradiol, at high (pharmacological) levels, strongly represses 11β-HSD1 in rodent liver and kidney (though not in hippocampus) (239, 318, 394, 418), probably indirectly (316). In rats, repression of 11β-HSD1 by thyroid hormone is similarly tissue specific (354, 771), and disturbances in glucocorticoid metabolism in patients with thyroid dysfunction are also consistent with thyroid hormone repression of 11β-HSD1 (272, 813). The pleiotropic effects of thyroid hormone on metabolism are likely to underlie this tissue-specific regulation, rather than direct effects on the 11β-HSD1 gene.

Like GR, the peroxisome proliferator-activated receptors (PPARs) and the liver X receptors (LXRs), intersect metabolism and inflammation. Agonists of both classes of nuclear receptor are reported to downregulate 11β-HSD1 in liver and brown adipose tissue (69, 283, 679). However, others report no effect of LXRα agonism on adipocyte 11β-HSD1 (749). The time course and sensitivity to cycloheximide suggest that any changes following PPAR or LXR activation is unlikely to involve binding to the Hsd11b1 gene (and hepatic 11β-HSD1 is unaltered in PPARα or LXRα -β knockout mice) (283, 679). In vitro, a PPARγ binding site in the human HSD11B1 gene promoter mediates induction in alternatively activated macrophages (differentiated with IL-4/IL-13) (117), but this site is not conserved in rodents.

To date, the C/EBPs, critical integrators of inflammatory and metabolic signaling, remain the only factors directly implicated in binding to the gene promoter to effect alterations in 11β-HSD1 transcription following endocrine, inflammatory, and metabolic stimuli, possibly also mediating at least some of the effects of nuclear receptors, as recently suggested in rat vascular smooth muscle cells (725).

III. 11β-HSD2 GENE STRUCTURE, ENZYMOLOGY, AND REGULATION

A. The HSD11B2 Gene and Its Transcriptional Control

11β-HSD2 is present in sharks and land vertebrates, and an ancestral gene has been identified in the nematode worm Caenorhabditis elegans (50). This may catalyze retinol interconversions, thus potentially controlling ligand access to nematode nuclear receptors (471) in analogy to its mammalian functions. Orthologs of HSD11B2 have been found in 36 species to date (http://www.ensembl.org). Unlike 11β-HSD1, 11β-HSD2 is present in fish (51) where it may confer selective access of the potent mineralocorticoid deoxycorticosterone to MR (324, 373). In fish, 11β-HSD2 is also important in testicular androgen activation and sex determination (197) by catalyzing conversion of 11β-hydroxy-testosterone to active 11-keto-testosterone (470).

11β-HSD2 cDNA was first cloned from sheep kidney (6), although a sequencing error wrongly predicted the COOH terminus of the protein sequence (111). This was followed by 11β-HSD2 cDNAs from human, rat, rabbit, mouse (15, 91, 127, 512, 812), and other species (591).

The human HSD11B2 gene is located on chromosome 16q22 and comprises five exons spanning just 6 kb (8) (FIGURE 6). Transcription starts at multiple sites (8), typical of its TATA-less, CpG island-containing promoter. The 5′-flanking region and exon 1 of HSD11B2 are encompassed within a CpG island (8), conserved in baboon (556) and rodents (17). This contains binding sites for members of the Sp1 transcription factor family (10), NF1 (17), NF-κB, and Egr-1 (362) (FIGURE 6). Occupancy of the Sp1 sites correlates with expression levels of 11β-HSD2, although their relative importance differs between cell lines (10, 516). Sp1 also plays a key role in the induction of 11β-HSD2 that occurs with syncytialization of human placental trophoblasts (400). In colon carcinoma cells, HSD11B2 repression by TNF-α or phorbol ester (mimicking the effects of inflammation upon colon; see below) is associated with Egr-1 and NF-κB (p50 homodimers) binding to the HSD11B2 promoter, with Sp1/Sp3 constitutively bound (362). Selective PPARδ activation in primary human placental trophoblast cells also decreases HSD11B2 promoter activity (332), although whether this is direct or via Sp1 is unknown. A polymorphism in the proximal HSD11B2 promoter that decreases NF1 and Sp1 binding and reduces promoter activity associates with salt-sensitive blood pressure excursions in normotensive subjects (18). The molecular basis of tissue and disease-associated regulation of HSD11B2 is an important goal and has potential therapeutic implications.

Figure 6.

Structure of the human HSD11B2 gene showing positions of relevant transcription factor binding sites. Schematic representation of the human HSD11B2 gene (not to scale) showing positions of relevant transcription factor binding sites. The gene spans 6.4 kb. Exonic sequences are shown as boxes; white boxes encode the open reading frame, with a red box and a blue box indicating the 5′ leader and the 3′ untranslated sequence, respectively. An arrow indicates the position of the promoter, although it should be noted that transcription starts at a number of sites clustered around the site shown (see text for details). Sp1/Sp3 sites are shown as green ovals (these overlap with Egr1 sites), an NF1 site is shown as a pink hexagon, and an NF-κB binding site is shown as a purple diamond. The positions of 2 polymorphisms that reduce transcription factor binding and promoter activity (18) are indicated by asterisks. See text for details.

The HSD11B2 CpG island is subject to epigenetic regulation. Greater cytosine methylation within the CpG island associates with lower 11β-HSD2 expression across a range of human cells and tissues (17). Intrauterine growth retardation in the rat is linked to increased Hsd11b2 CpG island methylation in kidney and with reduced adult renal 11β-HSD2 expression (58), although overall methylation levels are low. However, fetal exposure to glucocorticoids in rats, which also reduces adult renal 11β-HSD2 levels, had no effect on methylation of the Hsd11b2 CpG island (782). Increased methylation of the HSD11B2 promoter in peripheral blood mononuclear cells has been reported in subjects with essential hypertension (205), although 11β-HSD2 is not normally expressed in leukocytes (compatible with the relatively high levels of methylation observed); any relevance to kidney 11β-HSD2 is moot. Nevertheless, maternal consumption of a highly unbalanced diet in pregnancy, which associates with hypertension and hyperglycaemia in her adult children, has also been linked to increased methylation of the HSD11B2 promoter in leukocytes (164). Thus leukocyte methylation may represent an echo of changes in kidney and, if so, implies that the early life environment impacts on this promoter, as others including GR (755) in the HPA axis system.

B. 11β-HSD2 Protein

11β-HSD2 (SDR9C3; Ref. 557) is a member of the SDR family (557, 778). Its nearest relative is 17β-HSD2 (35% identical) (15, 48). Translation starts in the first exon (8). Human 11β-HSD2 comprises 405 amino acids, with a predicted molecular mass of 44 kDa (15, 91), whereas mouse is shorter, with 386 amino acids and a predicted mass of 42 kDa (http://www.ensembl.org) due to a truncated COOH terminus. Like all other members of the SDR family, 11β-HSD2 contains a nucleotide cofactor-binding Rossmann-fold in the NH₂-terminal region (Thr88 to Gly95) and a Tyr-[Xaa]₃-Lys motif in the active site (Tyr232 to Lys236 in human 11β-HSD2).

C. Structure and Mutations

11β-HSD2 is a very basic (pI 9.9) and stable intrinsic membrane protein (91) that loses activity once extracted from membranes (91). Within cells, it is anchored in the ER membrane, probably via three transmembrane helices predicted in the NH₂ terminus (91). Although the NH₂-terminal five amino acids are likely to reside in the ER lumen, the COOH-terminal bulk of the protein following Ala73 sits within the cytoplasm (an orientation opposite to 11β-HSD1) (511, 530). Intriguingly, 11β-HSD2 may physically interact with cytoplasmic MR (529). Like 11β-HSD1, 11β-HSD2 is probably active as a dimer (527).

There are no three-dimensional crystal structures of 11β-HSD2 yet available. Homology models (715) put the conserved active site Tyr232 and Lys236 close to the nicotinamide group of NAD⁺, with the adenosine moiety of NAD⁺ bound in the Rossmann fold in the NH₂ terminus (36). The preference of 11β-HSD2 for NAD⁺ over NADP⁺ depends primarily on Glu115, which sterically and electrically repulses the phosphate on NADP⁺ (36) and is conserved among SDR enzymes using NAD⁺.

The three-dimensional models have offered structural insights into AME (36). Most AME-causing mutations lie in the COOH-terminal half of 11β-HSD2 and affect conserved regions (491). Mutations that affect catalysis, cofactor binding, or dimerization produce the most severe phenotypes. These include mutation of the active site Tyr232 in a compound heterozygote (388) and deletion of Leu114 and Glu115, predicted to constrain the cofactor binding site (36). Mutation of Tyr226 in another compound heterozygote AME patient (388) is predicted to disrupt the dimer interface (714, 715) and may also affect substrate binding that occurs close to the side chain of Tyr226 (429). A mutation in the adjacent Pro227, which may normally position Tyr226 close to substrate (429), produces a mild form of AME with ∼50% reduced 11β-HSD2 activity and low-renin hypertension but without hypokalaemia or metabolic alkalosis (776). AME mutations at Tyr338 or Arg337 in the COOH terminus cause thermodynamic instability and misfolding, reducing protein half-life (39), consistent with reduced stability following deletion of the nonconserved COOH terminus of 11β-HSD2 (527). Mutations at Arg359 and Leu376 in the COOH terminus may similarly affect protein folding and stability (388, 429). Likewise, heterologous expression systems suggest that AME mutations at Ser180, Ala237, and Ala 328, which partially reduce 11β-HSD2 activity, do so by reducing stability (520). A homozygous mutation in Arg279 (399), predicted to lie on the surface of 11β-HSD2 (429), reduces V_max but has little effect on substrate affinity, causing the milder “type II” AME, probably due to minor structural disruption. Overall, AME patients show clear relationships between loss of 11β-HSD2 activity and phenotype (491, 765).

D. 11β-HSD2 Enzymology

In all species examined, 11β-HSD2 acts exclusively as an NAD⁺-dependent, high-affinity but low turnover dehydrogenase with physiological glucocorticoid substrates (K_m corticosterone ∼10 nM; cortisol ∼50 nM), showing little activity with NADP⁺ (15, 90–92, 511–513, 527, 812). It utilizes an ordered bi-bi sequential enzymatic mechanism in which cofactor binds prior to and leaves after the steroid (92).

1. Substrates and inhibitors

In addition to cortisol and corticosterone, dexamethasone is metabolized to some extent by 11β-HSD2 (91), with a K_m of ∼100 nM (91). Moreover, while 11β-HSD2 shows exclusively oxidative activity with physiological glucocorticoid substrates, in microsomes and whole cell assays it reduces 9-fluorinated glucocorticoids, thus converting 11-dehydrodexamethasone to dexamethasone. This regeneration contributes to the long half-life of dexamethasone in vivo (154, 156, 157, 527) (TABLE 3).

Table 3.

Selected substrates of 11β-HSD2

Substrate	Apparent Km, μM	Species	Source	Reference Nos.
Cortisol	0.025–0.055	Human	Purified placental protein, kidney microsomes, transfected cells	90, 91
Cortisol		Mouse
Corticosterone	0.010–0.012	Human	Purified placental protein, transfected cells	90, 91
Corticosterone	0.01	Rat	Transfected cells	812
Corticosterone	100	Mouse	Mouse kidney microsomes	127
Dexamethasone	0.119	Human	Transfected cells	91
11-Dehydrodexamethasone	0.068	Human	Kidney microsomes	157

Values for transfected cells are ascertained in lysates.

The classical licorice-derived 11β-HSD inhibitors glycyrrhetinic acid and carbenoxolone (see FIGURE 7) are more potent inhibitors of 11β-HSD2 than 11β-HSD1, with K_i or IC₅₀ typically severalfold lower for 11β-HSD2 than 11β-HSD1 (658) (TABLE 4). Modification of glycyrrhetinic acid at the carboxyl and/or 3-hydroxyl positions creates more selective potent 11β-HSD2 inhibitors (367). Other 11β-HSD2 inhibitors, with varying degrees of selectivity, have been identified. Gossypol, a natural phenol isolated from cotton seeds and originally proposed as a male contraceptive, inhibits 11β-HSD2 (116), conceivably accounting for the hypokalemic paralysis it can cause (739). Synthetic endocrine disrupters, including some phthalate derivatives widely used in plasticisers and solvents, organotins used in agriculture, and alkylphenols used as industrial surfactants, also inhibit 11β-HSD2 (424), although the importance of this is uncertain.

Table 4.

Selected inhibitors of 11β-HSD2

Inhibitor (of Glucocorticoid Metabolism)	Ki, μM	Species	Source	Reference Nos.
Carbenoxolone	0.010–0.083	Human	Kidney microsomes, transfected cells	91
Glycyrrhetinic acid	0.006–0.028	Human	Kidney microsomes, transfected cells	91
Progesterone	0.048	Human	Kidney microsomes	155
11β-Hydroxyprogesterone	0.007	Human	Kidney microsomes	155
Deoxycorticosterone	0.1	Human	Kidney microsomes	155
CDCA	20/none	Human	Kidney microsomes, transfected cells	155
Lithocholic acid	>10	Human	Kidney microsomes	155
Metyrapone	>1,000	Human	Kidney microsomes	155
Ketoconazole	10	Human	Kidney microsomes	155
Flavanone	200	Human	Transfected cells	32
Abietic acid	11–12	Human	Transfected cells	620
BVT14225, BVT2733	>10	Human	Transfected cells	55
Merck-544/T0504	1.8–>3.3	Human	Transfected cells	282

For additional inhibitors, see Reference 620. Values for transfected cells are ascertained in lysates. Merck-544 is the same compound as T0504.

E. 11β-HSD2: Tissue Distribution

In adults, 11β-HSD2 is largely restricted to the classical aldosterone (mineralocorticoid)-target tissues including distal nephron, sweat and salivary glands, and colonic epithelium (6, 15, 285, 368, 512, 588, 589, 652, 654, 770). It is also expressed in exocrine pancreas (15) and adrenal cortex (6, 588) and at more modest levels in the ileum, female and male reproductive systems (15, 588), and other epithelia including skin, lung (7, 654), and vascular endothelium (84). 11β-HSD2 expression in kidney and colon is sexually dimorphic, at least in mice (127). There are species differences in adult 11β-HSD2 expression. 11β-HSD2 is highly expressed in adrenal in sheep and rat, but less in mouse (127). 11β-HSD2 is expressed in testis in humans (15, 695), but not mice (127, 488) or rats (4, 588). Expression in some sites, including vascular smooth muscle cells, remains controversial (84, 120, 653), although a functional role in the nongenomic effects of aldosterone in the vasculature has been suggested (19).

During early to mid-gestation, 11β-HSD2 is widely expressed in the fetus and placenta (673), localized to syncytial trophoblasts (91, 368) (see developmental programming below). Many neoplastic cells express 11β-HSD2 perhaps reflecting their less differentiated state.

F. Regulation of 11β-HSD2

Estrogen upregulates 11β-HSD2 in kidney (239, 267, 418), the opposite of its effect on 11β-HSD1. This reciprocal control is generally true of other factors. Thus, whereas 11β-HSD1 is increased by inflammation and proinflammatory cytokines, such as TNF-α (129, 271, 362, 692), expression of 11β-HSD2 is usually reduced (discussed below). Hypoxia downregulates 11β-HSD2 in rat kidney, probably through Egr-1 (270). Beyond cytokines, protein kinases A and C stimulate and repress 11β-HSD2 mRNA and protein, respectively, in renal cells. The former is a pathway plausibly activated by vasopressin (599), affording a route to minimize renal retention of sodium alongside water, thus attenuating the risk of hypertension. Finally glucocorticoids upregulate 11β-HSD2 in lung cells (685), but apparently not in fetal kidney, and downregulate placental 11β-HSD2 (122, 343). More understanding is needed of this complex age- and organ-specific regulation. Even the widespread silencing of 11β-HSD2 in fetal tissues at the point of entry into terminal maturation is poorly understood. Such questions are of substantial importance, notably in dissecting mechanisms of developmental programming of offspring physiology and pathology.

IV. 11β-HSDs, METABOLISM, BLOOD PRESSURE, AND CARDIOVASCULAR FUNCTION

11β-HSD1 is highly expressed in the key metabolic organs liver, adipose tissue, skeletal muscle, and the islets of Langerhans, while 11β-HSD2 is crucial to blood pressure regulation. Both isozymes are expressed in the vasculature, and 11β-HSD1 at least is present in the heart and inflammatory cells (see below). Given this distribution and the striking metabolic disease seen in Cushing's syndrome of circulating glucocorticoid excess, it is not surprising that the role of 11β-HSDs in cardiometabolic function in health and disease has been studied in some detail. Although this is discussed below at the level of individual organs, of course in vivo these are highly integrated physiological systems.

A. Liver

11β-HSD1 was initially discovered, characterized, purified, cloned, and most highly expressed in liver (4, 21, 378, 478, 695). Despite this head start, and the absence of 11β-HSD2, the function of 11β-HSD1 in hepatocytes and especially in liver in vivo is far from completely understood.

In primary rat hepatocytes, 11β-HSD1 acts predominantly as a reductase (316). The in situ perfused rat liver also shows substantial and predominant 11β-reductase activity ex vivo (319), as does the dog liver in vivo (60, 178). In humans, the liver exhibits net regeneration of cortisol from cortisone (743). This activity is lower in simple obesity but unaltered in obesity complicated by type 2 diabetes (59, 61, 675), an observation suggesting maintenance of this “target” for selective 11β-HSD1 inhibitors in a patient population.

1. Endocrine effects of hepatic 11β-HSD1

Overall, the human liver/splanchnic bed contributes between 20 and 40% of daily cortisol production (25, 62), thus making a major impact on the half-life of cortisol. This affects the HPA axis. On some genetic backgrounds 11β-HSD1 null mice show elevated basal corticosterone levels and exaggerated responses to stress (265). This was originally attributed to deficient glucocorticoid regeneration in critical central sites of glucocorticoid negative feedback, such as the pituitary, paraventricular nucleus of the hypothalamus (PVN), and hippocampus, all of which express 11β-HSD1 (380, 478, 624). However, transgenic “rescue” of 11β-HSD1 solely in liver of null mice restores HPA axis function to normal (547). Conversely, liver-specific deletion of 11β-HSD1 increases adrenal size, suggesting HPA axis activation, consistent with a role for peripheral 11β-HSD1 in control of the HPA axis (392). While the underlying basis for this is not fully established, it is known that sugars and perhaps other fuel substrates impact on the HPA axis (139, 374, 387), and an afferent hepatic pathway via the vagus nerve also modulates HPA axis function (375). The notion that liver metabolism of glucocorticoids controls the stress axis is intriguing, plausibly representing a key node in the intimate association between the key organ determining metabolic fuel availability and the major neuroendocrine control of fuel homeostasis. This concept merits dissection.

2. Intracrine effects of hepatic 11β-HSD1

In general, 11β-HSD1 deficiency has rather little effect on basal hepatic function in the fed state. Thus 11β-HSD1 null mice fed ad libitum have modestly lowered triglyceride levels associated with less triglyceride-associated apoCIII in the HDL fraction (493). This associates with a modest increase in overall HDL cholesterol. The changes in triglyceride appear driven by increased hepatic expression of key enzymes of fat metabolism such as carnitine-palmitoyl transferase 1 (CPT-1), acyl-CoA oxidase, and uncoupling protein 2 (UCP2), themselves plausibly induced by their controlling transcription factor, PPARα, which is elevated in liver in 11β-HSD1 deficiency (493). Similar effects to increase β-oxidation of fats are seen with short-term selective 11β-HSD1 inhibition, which lowers hepatic very-low-density lipoprotein-triglyceride secretion (74).

On fasting, the normal elevation of PPARα is lost in mice deficient in 11β-HSD1, consistent with reduced glucocorticoid action (493). However, the oxidative response to fasting (induction of CPT-1) is maintained at wild-type values. Refeeding 11β-HSD1 knockout mice reveals exaggerated induction of enzymes of lipogenesis and suppression of enzymes of fat metabolism, effects again consistent with reduced glucocorticoid action and hence increased hepatic insulin sensitivity. Indeed, glucose and triglyceride levels on refeeding are lower in 11β-HSD1 null mice or with selective inhibitor treatment (73, 493).

Beyond the liver, 11β-HSD1 inhibition shifts the uptake of derived fatty acids to oxidative tissues, notably muscle and heart where increased lipid oxidation minimizes deleterious fat accumulation (74). These tisue-specific complex effects underline potential multi-organ benefits of 11β-HSD1 inhibition/deficiency.

In obesity or diabetes, fasting glucose levels are lower with 11β-HSD1 deficiency, consistent with reduced glucocorticoid action in the liver, particularly reduced gluconeogenesis. Specifically hepatic phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6Pase) are lower with 11β-HSD1 deficiency or inhibition (12–14, 364, 494). Interestingly, in lean animals, there is no fasting hypoglycemia (364), despite reduced G6Pase, presumably reflecting intact counterregulatory responses. In dogs, selective 11β-HSD1 inhibition attenuates liver glucose production in the face of insulinopenia (178), underpinned by reduced hepatic PEPCK and G6Pase mRNA expression. Gluconeogenic flux is unaffected, implying an effect on glycogenolysis, mirroring similar findings in humans given carbenoxolone (27), which inhibits 11β-HSD1 in liver but poorly in adipose tissue in humans (27) and rodents (413), and improves hepatic insulin sensitivity in humans in vivo, albeit modestly (744).

Overall, the data suggest that 11β-HSD1 increases glucocorticoid action within hepatocytes, stimulating gluconeogenesis and inhibiting β-oxidation of fats. Inhibition of 11β-HSD1 in the liver is predicted to be therapeutically beneficial in obesity and its associated metabolic disease, although the magnitude of the hepatic contribution remains uncertain. Indeed, the metabolic phenotype in liver-specific knockout of 11β-HSD1 in mice is relatively mild with modestly improved glucose tolerance (392). Perhaps the interplay between organs is of greater importance.

Hepatic 11β-HSD1 is regulated by diet, gender, and hormones. In most cases of metabolic syndrome, 11β-HSD1 expression in liver is either maintained or modestly lowered (in contrast to the consistent finding of elevated adipose tissue 11β-HSD1 in obesity). This perhaps represents a homeostatic attempt to minimize glucocorticoid-mediated insulin resistance. However, 11β-HSD1 is increased in liver in myotonic dystrophy (327), which is characterized by marked insulin resistance and dyslipidemia. Modeling hepatic overexpression in transgenic mice creates an insulin-resistant phenotype without obesity (548), but with increased hepatic PPARα and LXRα and reduced lipid clearance. Transgenic overexpression of hepatic 11β-HSD1 also causes hypertension, plausibly driven by increased angiotensinogen expression in liver.

Overall, therefore, despite high expression, the importance of hepatic 11β-HSD1 in human pathogenesis remains uncertain, and even its major physiological functions are essentially unresolved. Important questions for the future include the role of liver 11β-HSD1 in xenobiotic and oxysterol metabolism, determining its importance in starvation when glucocorticoids may be key to survival (although plasma glucocorticoids are not elevated in chronic malnutrition; Ref. 427), dissecting the relative importance of the various sources of 11β-HSD1 in the splanchnic bed in the overall metabolic and endocrine effects of 11β-HSD1, and resolving its function in the link between the liver and HPA axis activity. An interesting question is the importance of alternative substrates of 11β-HSD1 (7-ketocholesterol, bile acids) in liver. 7-Ketocholesterol in particular may regulate de novo cholesterol biosynthesis (89) which could also impact bile acid pathways. Bile acids are competitive inhibitors of 11β-HSD1 (2, 155, 189) (TABLE 2), but also potentially substrates (531). Glucocorticoids regulate bile acid homeostasis (593), so it is important to determine whether any metabolic effects of 11β-HSD1-deficiency are mediated through altered bile acid homeostasis.

B. Adipose Tissue

11β-HSD1 is expressed in all white and brown adipose tissue depots with higher levels in subcutaneous than visceral (intra-abdominal) fat in mice (75, 495), but either higher (234) or similar (550) levels in human visceral than subcutaneous fat. Interestingly, 11β-HSD1 mRNA and protein/activity can become dissociated in visceral but not subcutaneous adipose tissue in rodents (146) and humans (101, 397), implying posttranscriptional controls in this depot. This might explain discrepancies in the human visceral adipose tissue literature and calls into question studies based solely on mRNA estimations.

11β-HSD1 is found in both preadipocytes and adipocytes. It is a predominant reductase in adipocytes and murine preadipocytes (146, 494), but in human omental preadipocytes is a dehydrogenase as H6PDH is limiting (101). Glucocorticoids promote adipocyte differentiation. In murine and human cells, 11β-HSD1 activity is induced during adipogenesis, assisting the later stages of maturation (98, 510).

The detailed biology of glucocorticoids in adipocytes is imperfectly understood. Both GR and MR are expressed in adipocytes (434), with MR mediating proinflammatory effects (253) whereas GR are anti-inflammatory. This raises the possibility of inverted U-shaped relationships between glucocorticoid concentrations and function, analogous to the hippocampus, the archetypal tissue expressing both GR and MR as well as 11β-HSD1. Moreover, the adipose effects of glucocorticoids depend on context, differing in the fed (insulin high), fasted (insulin low), and obese/insulin-resistant states and differing by adipose depot.

With these caveats, glucocorticoids alone stimulate fatty acid uptake via lipoprotein lipase and lipolysis via induction of hormone-sensitive lipase and adipose triglyceride lipase (643, 730) and thus promote free fatty acid synthesis which increases metabolic fuel provision. However, at least in vitro, low doses of exogenous glucocorticoids suppress adipocyte lipogenesis (217). In line with this, 11β-HSD1 inhibition decreases lipolysis and increases lipogenesis in vitro (217). This conforms with increased subcutaneous adipose depots in high-fat-fed 11β-HSD1 knockout mice, but not with their attenuated visceral depots (494, 750).

In contrast, in the presence of insulin, glucocorticoids augment insulin-stimulated lipogenesis in vitro (217). Ex vivo, adipocytes taken from 11β-HSD1-deficient mice show reduced intra-adipose glucocorticoid levels and greater insulin-mediated glucose uptake and triglyceride hydrolysis (494). 11β-HSD1 inhibition decreases lipid synthesis and fatty acid cycling gene expression and increases fatty acid oxidation via CPT-1 in mesenteric adipose, but increases lipid synthesis and fatty acid cycling and decreases CPT-1 in epididymal fat (72). Interestingly, in humans, access of circulating glucocorticoids to adipose tissue is restricted and slow. This implies that 11β-HSD1 exerts a more substantial impact on local adipocyte steroid levels (303), and hence biology in vivo, than in other tissues where plasma glucocorticoids gain more rapid access. Overall, the effects of 11β-HSD1 on adipose tissues are depot and dietary state dependent. In general, the effects are slight in lean, insulin-sensitive states but, as discussed below, become marked with obesity-insulin resistance.

1. Adipose tissue 11β-HSD1 in obesity

A striking and consistent finding is a tissue-selective two- to threefold increase in 11β-HSD1 activity in subcutaneous adipose tissue in obese men and women (234, 472, 549, 550, 574, 575, 612). The effect is found both ex vivo in enzyme activity assays in adipose biopsies or by in vivo microdialysis in abdominal adipose (612) and is accompanied by lowering of hepatic 11β-HSD1 activity in simple obesity (574), but maintenance of hepatic activity if obesity is accompanied by type 2 diabetes (675). These effects are probably driven by increased transcription since 11β-HSD1 mRNA levels are also elevated (549). Such data have been reported in distinct human populations and appear to be a robust measure of an expanded adipose depot. Similar elevations of 11β-HSD1 in adipose tissue are found in obese rodents with deficits in leptin signaling, sometimes with a predominant effect in visceral adipose depots (413).

In humans, the literature on the effect of obesity on visceral adipose tissue 11β-HSD1 is discordant, with reports of both elevation (150) and no change (234). A possible confounding factor is the upregulation of 11β-HSD1 by cytokines (711), which are elevated in obesity, infection, and perhaps following surgery. Of course, human visceral adipose tissue can only be obtained during operation and if the indication was an inflammatory condition or perhaps a long bariatric procedure then measured 11β-HSD1 may not reflect the underlying state. Two studies of 11β-HSD1 in omental adipose tissue in healthy women undergoing routine minimally invasive surgery show 11β-HSD1 activity and mRNA levels correlate directly with adiposity, its metabolic complications (472) and adipocyte size (larger adipocytes are less favorable metabolically) (466), but in two other studies visceral 11β-HSD1 did not correlate with obesity (234). Thus, unlike the clear consensus in subcutaneous adipose tissue, the jury remains out on whether or not visceral adipose 11β-HSD1 is increased in human obesity.

2. Manipulation of 11β-HSD1 in adipose tissue

Modeling the increased 11β-HSD1 in adipose tissue in obesity, Masuzaki et al. (446) generated a mouse overexpressing 11β-HSD1 in adipocytes under the fatty acid binding protein-4 (aP2) promoter. This produced an ∼2.5-fold overexpression of 11β-HSD1 in adipose depots, mirroring changes in human obesity. This increased intra-adipose corticosterone levels twofold without changing circulating glucocorticoid levels. There was, however, an increase in corticosterone concentration in the portal vein and thus glucocorticoid supply to the liver. Adult aP2-HSD1 transgenic mice show predominantly visceral obesity, attributed to the higher GR expression in visceral than subcutaneous adipose (446). The mice faithfully replicate metabolic syndrome with insulin resistance/impaired glucose tolerance (exacerbated by high fat feeding), dyslipidemia, hypertension (plausibly caused by increased adipose expression of angiotensinogen and reversed by angiotensin 2 receptor antagonism), and hyperphagia (446, 447). These data suggest that elevated 11β-HSD1 in adipose tissue may be pathogenic in metabolic disease associated with obesity. Since plasma cortisol levels are normal or modestly reduced in simple obesity without metabolic disease (745), fat-specific 11β-HSD1 excess has been dubbed “Cushing's disease of the omentum” (100). The prominent involvement of subcutaneous fat suggests this is more accurately “Cushing's Syndrome of adipose tissue.”

Conversely, knockout or inhibition of 11β-HSD1 produces a “favorable” metabolic state, particularly in the presence of obesity. 11β-HSD1 gene disruption in mice produces insulin sensitization and lowers fasting blood glucose without hypoglycemia (364). Importantly, insulin sensitivity and cardiometabolic “health” are improved in diet-induced and in genetic models of obesity (493, 494). Analogous effects are seen with short- to medium-term administration of selective 11β-HSD1 inhibitors in a variety of rodent models of obesity/type 2 diabetes (12, 13, 282, 751, 752). The current consensus, that these beneficial effects are due to lowered intra-adipose glucocorticoid action, is supported by mice with ectopic transgenic expression of 11β-HSD2 in adipose tissue (decreasing intra-adipose levels of corticosterone) which show insulin sensitization and resistance to metabolic disease with obesity (342).

Intriguingly, on a high-fat diet, 11β-HSD1 deficiency leads to preferential fat loss from metabolically disadvantageous visceral and accumulation in “safer” subcutaneous depots (494). Early in the adaption to high-fat diet, subcutaneous adipose tissue from 11β-HSD1-deficient mice shows enhanced insulin and beta-adrenergic signaling associated with the expansion of smaller, more metabolically active adipocytes, rather than the large adipocytes seen in similarly fed controls (750). In contrast, 11β-HSD1 deficiency does not produce insulin sensitization in visceral adipose tissue, but associates instead with maintained “lean mouse” levels of AMPK, although these are reduced in controls. 11β-HSD1-deficient visceral adipose tissue also exhibits marked suppression of inflammatory signaling with reduced expression of proinflammatory chemokines and cytotoxic T-cell and macrophage infiltration (750). Both fat depots show higher levels of PPARγ mRNA and evidence of enhanced PPARγ activation. The altered adipose distribution seen in rodents lacking 11β-HSD1 resembles to some extent the effects of PPARγ agonists (788), but whether PPARγ activation is critical to any of the effects of 11β-HSD1 deficiency/inhibition remains to be clarified. Overall, 11β-HSD1 deficiency engenders anti-inflammatory, AMPK-activated visceral adipose tissue and insulin-sensitized, “safer” metabolic storage subcutaneous depots. This subtle partitioning of metabolism and inflammation between the different adipose depots underlies a potential for beneficial therapeutic effect if it applies to humans.

Of course, 11β-HSD1 is not expressed in adipose tissues merely for contemporary therapeutic benefit. Its evolutionary importance has been little addressed but may reflect needs for rapid storage of excess calories when conditions allow in an environment of alternating feast and famine. Higher adipose 11β-HSD1 would perhaps support expanded visceral adipose depots to allow rapid delivery of calories to the liver. This may have optimized local storage and then release during sleep, dormancy, and the next period of starvation.

3. Regulation of 11β-HSD1 in adipose tissue

Mice or rats fed a high-fat diet show unexpected downregulation of 11β-HSD1 in adipose tissue (165). This is greatest in obesity and metabolic disease-resistant strains and has been hypothesized to reflect an attempt to attenuate some of the adverse metabolic consequences of chronic dietary excess (495).

So how is 11β-HSD1 regulated in adipose tissue? In vivo and in vitro evidence in rodents suggests upregulation by glucocorticoids (604), proinflammatory cytokines (711) and weight loss/stearate feeding (426), and downregulation by high-fat diet/weight gain (495), PPARγ (283), and perhaps LXR agonists (679), although the latter is contested (749). Humans differ in some aspects (740).

The main factors directly regulating 11β-HSD1 transcription in adipose tissues appear to be C/EBPs, notably C/EBPβ (244) and its processed isoforms LIP (inhibitor) and LAP (activator). Indeed, the LIP-to-LAP ratio is key to downregulation of 11β-HSD1 in adipose with high-fat feeding (191). Whether or not this also underpins the upregulation in human obesity remains uncertain, yet crucial.

4. 11β-HSD1, adipose tissue hypoxia, and fibrosis

The expansion in adipose tissue accompanying dietary obesity is associated with a poor angiogenic response that contributes to adipose tissue hypoxia (296). Consequently, adipose tissue synthesizes hypoxia-inducible factor-1α (HIF-1α), potentially precipitating inflammation and fibrosis (258). 11β-HSD1 deficiency attenuates adipose tissue hypoxia, with reduced induction of HIF-1α and less fibrogenesis, associated with reduced activation of the transforming growth factor (TGF)-β/Smad3/α-smooth muscle actin signaling pathway (467). Underlying this is greater angiogenesis in 11β-HSD1-deficient adipose tissue, plausibly driven by PPARγ-stimulated expression of the proangiogenic factors VEGF-A, apelin, and angiopoietin-like protein 4. Thus, with 11β-HDSD1 deficiency, increased angiogenesis may underpin the nonpathological expansion of fat depots allowing safe weight gain. Increased angiogenesis is also seen in the ischemic 11β-HSD1-deficient myocardium and aorta (646) and indeed may contribute to other cardiometabolic benefits of 11β-HSD1 deficiency.

5. 11β-HSD1 deficiency/inhibition and human adipose tissue biology

Does adipose 11β-HSD1 matter in human obesity? Whilst splanchnic 11β-HSD1 contributes substantially (20–40%) to total daily glucocorticoid production rates, this is largely ascribed to the liver, so the adipose effects are likely to be mainly intracrine (674). The nonselective 11β-HSD inhibitor carbenoxolone increases insulin sensitivity (744) by an action largely mediated on the liver (27, 413), although some adipose tissue effects on glycerol production occur (712). Acutely (over hours), enzyme activity may be regulated: lipid infusion increases adipose 11β-HSD1 activity (740). However, over the longer term (days or weeks), PPARα and PPARγ agonists have no impact (741), unlike their repressive effects in rodents.

Sadly, no data on the metabolic phenotype, if any, of subjects with HSD11B1 mutations in cortisone reductase deficiency are available (393). However, a unique patient with “apparent cortisone reductase deficiency” who developed pituitary ACTH-dependent Cushing's disease presenting with secondary amenorrhea illustrates well the importance of 11β-reductase. Despite unequivocally elevated plasma and urinary free cortisol levels (2- to 3-fold above the reference range), the patient showed a normal habitus with a normal fat distribution, no obesity, no hirsutism, myopathy, bruising, striae, or hypertension (709). She was deficient in regeneration of cortisol from oral cortisone and her urinary THF+aTHF/THE ratio was approximately twofold lower than normal and approximately threefold lower than usual for Cushing's patients. Although no mutation was found in HSD11B1, so molecular proof of this effect is missing, the case supports the critical importance of the intracrine amplification of glucocorticoids in their effects on tissues. It is interesting to speculate whether the corticotroph adenoma was coincidental or a tertiary hyperplastic process in response to cortisone reductase deficiency causing chronic HPA axis activation. If the latter, this is likely a rare event given the high frequency of chronic HPA activation in affective disorders apparently without corticotroph adenoma formation.

Selective 11β-HSD1 inhibitors have entered Phase II clinical trials in patients with type 2 diabetes. One agent, ICNB13739, administered in a double-blind and placebo controlled study to patients taking but suboptimally controlled by metformin, significantly lowered fasting plasma glucose and haemoglobin A1c levels, plausibly via increased insulin sensitivity (594). There was a small but significant fall in body weight. The drug also reduced plasma cholesterol and triglyceride levels in hyperlipidaemic subjects. Therapeutic effects were greatest in those with the poorest metabolic control. In a similar trial, another drug, MK-0916, modestly reduced haemoglobin A1c, LDL-cholesterol, blood pressure, and body weight, without altering fasting glucose levels (195). A related agent, MK-0736, reduced ambulatory blood pressure, weight and LDL-cholesterol in nondiabetic obese hypertensive subjects (633).

Thus, in humans, as in mice and rats, 11β-HSD1 deficiency/inhibition improves multiple metabolic disease risk factors. To date, the effects on individual indices such as HbA1c are relatively modest, although probably in keeping with other insulin-sensitizing agents. Perhaps coadministration of 11β-HSD1 inhibitors with metformin has masked their impact, since the metformin target AMPK appears to be a core pathway downstream of 11β-HSD1 in visceral fat, at least in mice (750). While one study was in patients the majority of whom had been withdrawn from metformin (195), the drug-free duration was quite short. Studies in unmedicated (perhaps in metformin-intolerant) subjects may be revealing.

More importantly, 11β-HSD1 inhibition is a multifaceted therapeutic target and probably changes individual components of metabolic disease risk modestly; after all, glucocorticoid levels are being reduced, not obliterated, in target cells. Indeed, the advantage over GR antagonists is that receptors remain functional so any life-threatening stress can induce the HPA axis to provide glucocorticoids for survival. Crucial will be the impact of 11β-HSD1 inhibitors on cardiovascular events, which are the main cause of mortality in the obesity-type 2 diabetes-metabolic syndrome continuum (88).

6. Steatohepatitis: adipose-liver interactions

A potential pathogenic process involving 11β-HSD1 in liver is the deposition of ectopic lipid that underlies the spectrum of nonalcoholic fatty liver disease/nonalcoholic steatohepatitis. This affects around one-third of Americans (690) and is increasingly prevalent globally. Transgenic overexpression of 11β-HSD1 in liver causes fatty liver and insulin resistance without obesity (548), suggesting a potential pathogenic role for the enzyme. While most obese individuals have reduced 11β-HSD1 levels in liver (574, 575), liver 11β-HSD1 is increased in metabolic syndrome associated with myotonic dystrophy (327). Moreover, short-term 11β-HSD1 inhibition reduces liver triglyceride levels without altering food intake or weight (73). However, hepatic lipid is also increased with adipose-specific 11β-HSD1 overexpression in mice, plausibly reflecting increased glucocorticoid and fatty acid delivery to liver via the portal blood (446). Clinical studies show no association between liver 11β-reductase and liver fat content (357, 762), whereas visceral adipose tissue 11β-HSD1 is associated with hepatic steatosis (112). Thus there is, as yet, no strong evidence that hepatic 11β-HSD1 contributes to fatty liver disease, although 11β-HSD1 in visceral adipose tissue may well play a role.

7. Polycystic ovary syndrome

This syndromic disorder incorporates the common coincidence of reproductive (secondary oligomenorrhea/amenorrhea with reduced or absent ovulation, polycystic or enlarged ovaries on ultrasonography, reduced fertility), hyperandrogenism (hirsutism, male-pattern hair loss, acne, sebaceous hyperactivity), and sometimes metabolic (obesity with or without insulin resistance) features in women. There are many hypotheses of causation, but genetics and aberrant metabolism are commonly implicated. Lower 11β-HSD1 activity has been advocated based on the prominent hyperandrogenism seen in apparent cortisone reductase (H6PDH) deficiency (390) and cortisone reductase (HSD11B1) deficiency (393).

A functional HSD11B1 SNP (rs12086634 encoding a third intron variant that links with reduced 11β-HSD1 activity; Ref. 170) has been associated with hyperandrogenism in 102 patients with polycystic ovary syndrome (PCOS), a relationship seen in both obese and lean patients (suggesting the link is not merely secondary to obesity) (213). Indeed, lean PCOS patients show enhanced cortisol clearance and compensatory adrenal hyperandrogenism and interestingly appear partly to resist obesity and metabolic disease. These findings build on data showing increased adrenal cortisol and androgen production in lean PCOS with a lower ratio of cortisone to cortisol metabolites compatible with less 11β-HSD1 reductase and/or more 11β-HSD2 dehydrogenase (584). However, a recent larger study in 300 PCOS patients failed to replicate any link between HSD11B1 SNPs and PCOS features, although one combination of rs846910 and rs12086634 associated with increased enzyme activity and a substantially elevated risk of metabolic syndrome (odds ratio 2.77) (212).

Polymorphisms in H6PDH may also affect risk of PCOS and cardiometabolic disease (611), plausibly through effects on 11β-HSD1 activity (e.g., via increased adrenal androgen production secondary to altered hepatic 11β-HSD1 activity; Ref. 390). However, another study found no association of the same polymorphism (R453Q) in H6PD with metabolic parameters or adrenal androgen production (648) and a somewhat larger study of ∼450 patients also showed no linkage between SNPs in HSD11B1 or H6PD and PCOS (169). These discordances may reflect the uncertainties of genetic studies in relatively small populations. Overall, while an attractive hypothesis, a causative role for lower 11β-HSD1 activity in PCOS is far from established.

C. Skeletal Muscle

Skeletal muscle is a major target for insulin action and glucose storage/catabolism. 11β-HSD1 is expressed in skeletal muscle, albeit at lower levels than liver (767, 768). 11β-HSD2 is not expressed in muscle. In humans, 11β-HSD1 levels ex vivo do not appear to correlate with body weight or adiposity, at least under standard culture conditions (767, 768), although other evidence suggests that skeletal muscle 11β-HSD1 mRNA and protein levels are increased in obesity with diabetes in rodents (805) and in myotubes grown from humans with type 2 diabetes (1). These data require extension. Selective 11β-HSD1 inhibition increases biochemical markers of muscle insulin sensitivity in diabetic rodents in vivo (490). However, the contribution of muscle 11β-HSD1 to overall glucose-insulin homeostasis remains uncertain. Although effects of 11β-HSD1 on metabolism and glucocorticoid-induced muscle atrophy have been suggested (80), any biological importance, especially in the sarcopenia of ageing, remains undetermined.

Another potential approach to reduce intracellular glucocorticoid action is to reverse the reaction direction of 11β-HSD1 (389). Mice lacking H6PDH are relatively insensitive to glucocorticoids, exhibiting increased insulin sensitivity presumably due, in part, to the reversed reaction direction of 11β-HSD1 reducing glucocorticoid levels in key metabolic tissues (391). These mice also have a striking skeletal myopathy and show fasting hypoglycemia not seen in 11β-HSD1-deficient mice, but these features are independent of 11β-HSD1 and probably relate to loss of other NADPH-dependent processes (627).

D. Pancreas

11β-HSD1 mRNA and activity are present in mouse and human pancreatic islets (140). It remains contentious whether this localizes to alpha cells (689) or beta cells (140, 716) with evidence in support of both based on immunohistochemistry and effects of inhibitors on insulin or glucagon secretion. Probably therefore both alpha and beta cells have 11β-HSD1.

11β-HSD1 activity is increased in islets from diabetic rodents (175). Since glucocorticoids, albeit in pharmacological concentrations, directly inhibit beta-cell insulin secretion (248, 381), it has been proposed that increased islet 11β-HSD1 is part of the pathogenesis of beta cell failure (140). However, modest transgenic overexpression of 11β-HSD1 in beta cells alone, driven by the insulin promoter, to a level comparable to that observed in some type 2 diabetes models, is associated with enhanced glucose-stimulated insulin secretion both in vivo and in isolated islet cells in vitro (716). On the C57BL/KsJ mouse strain background, which develops beta-cell failure on high-fat diet, elevated 11β-HSD1 in beta cells maintains their survival and function. The underlying mechanisms appear to reflect protection from apoptosis and cellular stress, as well as facilitated secretion. Biochemically, the beta cells show induction of heat shock proteins, protein kinase A, ERK, and p21 signaling pathways. In contrast, greater beta-cell overexpression of 11β-HSD1 or lifelong deletion of 11β-HSD1 associates with poorer glucose-stimulated insulin secretion. Although these may be developmental effects, as glucocorticoids play critical roles in pancreatic islet formation and perinatal maturation of endocrine function, they also raise the possibility of an inverted U-shaped relationship between 11β-HSD1 in beta cells and insulin secretion with the modest elevation seen in early diabetes associated with improved insulin secretory function. A similar relationship between glucocorticoid dose and beta-cell function is also observed in vitro. Intriguingly, high-fat diet induces a similar modest elevation of 11β-HSD1 in islets of mouse strains that can compensate for the caloric challenge, while strains that are susceptible to beta-cell failure downregulate islet 11β-HSD1 and those susceptible to obesity-driven diabetes (Lep^db/db) have very high islet 11β-HSD1 levels, both of which cause failure of insulin secretion.

This is perhaps the first evidence of a physiological function for 11β-HSD1 in metabolic control. Normal mice given a high-fat diet exhibit a robust compensatory insulin secretion plausibly driven by the modest upregulation of beta-cell 11β-HSD1. At the same time, healthy mice substantially downregulate adipose tissue 11β-HSD1, driving subcutaneous fat cell insulin sensitization and “safe” calorie storage while AMPK-mediated beneficial metabolic and noninflammatory processes occur in visceral adipose (495, 750). In contrast, in severe obesity, excessive pancreatic 11β-HSD1 and hence glucocorticoid regeneration contributes to the failure of insulin secretion and perhaps beta cell failure, and increased adipose tissue 11β-HSD1 causes visceral obesity, insulin resistance, and metabolic syndrome (FIGURE 8).

Figure 8.

Cartoon of potential role of 11β-HSD1 in metabolic organ interrelationships in obesity. A: in normal weight health, 11β-HSD1 acts largely as a reductase. This performs an important endocrine role in the splanchnic bed by regenerating glucocorticoids, thus contributing ∼40% of daily glucocorticoid production. In addition, it has intracrine actions, amplifying the glucocorticoid signal inside hepatocytes, pancreatic islets (beta and alpha cells), and adipocytes. B: with modest obesity, 11β-HSD1 is elevated in adipocytes and beta cells but not hepatocytes. This increases insulin resistance in adipose tissues, increases release of proinflammatory/antimetabolic adipokines, and portal blood glucocorticoid and fatty acid deliver to the liver, but also increases islet insulin release to glucose, plausibly without changing hepatic insulin sensitivity. This may support storage of calories in adipose tissue in preparation for less bountiful nutritional circumstances. C: with severe obesity, greater rises in adipose and beta cell 11β-HSD1 lead to failure of pancreatic insulin release, worsening peripheral insulin resistance and metabolic disease despite potentially “compensatory” declines in hepatic glucocorticoid regeneration, which might even contribute to HPA axis activation due to loss of bulk glucocorticoid regeneration. D: 11β-HSD1 inhibition maximizes subcutaneous adipose and liver insulin sensitivity and visceral adipose AMPK signaling and reduces its inflammation and portal glucocorticoid and fatty acid release. Despite reduced beta cell insulin release and HPA axis activation (without elevation of glucocorticoid levels), the balance of metabolism favors reduced hepatic gluconeogenesis, increase beta-oxidation of fats, “safe” calorie storage in subcutaneous adipose depots, reduction of visceral adipose mass, and metabolic health.

The partitioning of metabolism effected by alterations in 11β-HSD1 in beta cells and adipocytes in response to high-fat diet appears advantageous in an ancestral environment where feast and famine alternated. Increased insulin secretion during early calorie excess appears to facilitate fat depot-specific storage that might aid survival in the occurrence of famine. With marked calorie excess, increased adipose tissue 11β-HSD1 may have allowed even more fat storage in rapidly accessible visceral depots. Of course in the modern continuous calorie-dense environment, elevated adipose tissue 11β-HSD1 causes metabolic disease, but this unusual environment is unlikely to have driven the evolution of this system. These data, while fascinating, also illustrate our paucity of understanding of the physiological (as opposed to pathological) roles of 11β-HSD1 in metabolic tissues.

E. Kidney

11β-HSD2 is highly expressed in the distal nephron that allows aldosterone selectivity in vivo. 11β-HSD1 is also expressed in the kidney, primarily in proximal tubule cells (478, 562, 571, 589, 617, 695), although any role remains largely unexplored.

In the distal nephron, 11β-HSD2 is concentrated in cortical collecting duct cells (91), largely overlapping with MR in principal cells (512, 640) which mediate aldosterone actions on salt-water homeostasis. An unresolved paradox lies in the relative affinities of MR and 11β-HSD2 for glucocorticoids. Human MR binds cortisol, corticosterone, and aldosterone with a K_d of ∼0.5 nM (37, 637), whereas the affinity of human 11β-HSD2 for cortisol (∼50 nM) and corticosterone (∼5 nM) is more than an order of magnitude lower (15, 92). Even allowing for molar excess of 11β-HSD2 over MR in distal nephron cells (514), these affinities suggest 11β-HSD2 is unlikely to completely exclude glucocorticoids from MR. However, modeling the cellular context by transfecting distal nephron cells with MR and an MR reporter gene with or without 11β-HSD2 showed that coexpression of 11β-HSD2 is sufficient to prevent glucocorticoid activation of MR, abrogating target gene activation, an effect reversed by carbenoxolone inhibition of 11β-HSD2 (395). Transfection with 11β-HSD2 cDNA carrying mutations that cause SAME fails to protect MR from activation by cortisol (529). Perhaps the cellular microanatomy involved may be sufficient to overcome the differences in affinity: lipophilic glucocorticoids may preferentially track through intracellular membranes such as the ER and exit past cytoplasmically directed 11β-HSD2 (530), with inactivation before the cytoplasmic MR can “compete” for ligand. Indeed, MR and 11β-HSD2 coassociate, at least in transfected cells (529), and it has been suggested that this is crucial to prevent activation of MR by bound cortisol through an, as yet unknown, mechanism (209), perhaps involving the apparent functional antagonism by physiologically relevant concentrations of cortisone and 11-dehydrocorticosterone of aldosterone activation and nuclear translocation of MR (529) (FIGURE 3).

Patients with AME are homozygous or compound heterozygous for mutations in the HSD11B2 gene (141, 498, 777). 11β-HSD2 knockout mice also show apparent mineralocorticoid excess (363). Interestingly, the mouse model suggests that the high early mortality of children with AME may relate to irreversible developmental effects on renal structure (distal nephron hyperplasia and hypertrophy) and the aorta, which shows a propensity to dissection associated with disorganised elastic laminae. These effects are partly responsive to lowering of blood pressure (363). The background genotype impacts substantially on lethality (288); how is uncertain.

Mice heterozygous for a gene disruption in Hsd11b2 have normal basal blood pressure which, in contrast to control C57BL/6 mice, is elevated on sodium loading. Unexpectedly this is reversed by GR rather than MR antagonism (46). The mechanism is unknown. It may reflect developmental programming of the kidney following fetal exposure to excess glucocorticoids due to feto-placental 11β-HDSD2 deficiency in utero (see below).

Such data suggest HSD11B2 as a candidate gene/gene product for hypertension beyond rare AME (198). Indeed, a subset of subjects with essential hypertension show a prolonged half-life of [³H]cortisol, implying 11β-HSD2 deficiency (661, 746), although not all studies find this (313). Measures of 11β-HSD2 activity (UFF-to-UFE ratio) show a decline in enzyme activity with ageing, perhaps correlating with the rise in blood pressure with advancing years (281). More directly, HSD11B2 polymorphisms have been associated with blood pressure in black men with renal failure (754). However, other studies in black subjects have been negative (764), and the gene has not emerged as a general candidate for hypertension from genome-wide association studies (182). In white males HSD11B2 polymorphisms associate with sodium-sensitive changes in blood pressure (417), reminiscent of the phenotype of Hsd11b2^+/− heterozygous mice. Specifically, shorter alleles of a CA repeat polymorphism in the first intron of the HSD11B2 gene that associate with reduced 11β-HSD2 activity measured in vivo also associate with greater rises in blood pressure between sodium-depleted and loaded states (3). A molecular mechanism linking this intronic CA repeat length to enzyme function remains questionable. Perhaps the CA repeat is in linkage disequilibrium with a functional polymorphism.

A more functional connection is afforded by the G209A polymorphism affecting an NF-1 binding site in the HSD11B2 gene promoter, which reduces association of both NF-1 and GR with the promoter (18). The G209A polymorphism, which appears in linkage disequilibrium with shorter CA repeat alleles, is more common in salt-sensitive subjects (18). Intriguingly, methylation of a CpG island in the HSD11B2 promoter (see above), an epigenetic event that in cancer is associated with gene silencing (330), is associated with lower 11β-HSD2 expression and higher blood pressure (205) in keeping with epigenetic control and a key role for this gene in developmental programming of blood pressure. However, given the absolute difference in promoter methylation between “low enzyme function” hypertensive and controls was only ∼8%, any causal role remains uncertain.

11β-HSD1 is also expressed in kidney, predominantly in the proximal tubule, interstitial cells, and medulla (601). Interestingly, HSD11B1 polymorphisms associate with blood pressure in some populations (172, 201), and even with pressure changes to potassium challenge (266), although separating effects on blood pressure from metabolic syndrome as a whole is challenging and the low levels of 11β-HSD1 mRNA in human kidney mitigate against a major role in this tissue. Studies with selective 11β-HSD1 inhibitors in animals and humans suggest these reduce blood pressure, at least in subjects with metabolic syndrome (195, 633), although there is no primary reason to invoke a renal mechanism. Intriguingly, in Dahl salt-sensitive rats, renal medullary 11β-HSD1 levels fail to decrease normally on salt loading, although forced reduction (siRNA-mediated knock-down) keeps blood pressure low during salt loading (409). Importantly, the siRNA studies suggest 11β-HSD1 acts as a reductase in the renal medulla. The molecular mechanisms by which lowering active glucocorticoids in the renal medulla attenuate blood pressure responses to dietary sodium merit attention. In addition, the proximal nephron is an important site for gluconeogenesis, and indeed, renal PEPCK appears predominantly sensitive to glucocorticoids rather than insulin (114). Any importance of 11β-HSD1 in this key metabolic function is unexplored.

F. The Vasculature and Heart

Both 11β-HSD1 and 11β-HSD2 are expressed in the vasculature (747). In the vascular wall, 11β-HSD2 is mainly located in the endothelium (257), whereas 11β-HSD1 is highest in the vascular smooth muscle in both rodents and humans where it catalyzes 11β-reduction (747). 11β-HSD1 is expressed in neonatal heart (638), perhaps contributing to the perinatal maturational effects of glucocorticoids in this tissue (586). Expression of 11β-HSD1 in adult heart is also very modest and confined to cardiac fibroblasts (86). Any expression of 11β-HSD2 in heart is low and probably restricted to the coronary vasculature.

1. Vasculature

Expression of 11β-HSD1 in vascular smooth muscle suggests actions on vasoconstrictor function, but such effects have not been found in ex vivo preparations (257). However, 11β-HSD1 does play a key role in angiogenesis and vascular remodeling. Glucocorticoids inhibit angiogenesis in animals and humans. In line with this, 11β-HSD1-deficient mice show increased angiogenesis from aortic rings in vitro and into implanted sponges and in the infarcted myocardium following coronary artery ligation in vivo (646). The latter associates with improved ventricular ejection fraction despite a similar initial infarct size (454), implying facilitated recovery. The greater angiogenic response in the infarcted Hsd11b1^−/− myocardium is perhaps driven by angiogenic factors such as IL-8, which is concomitantly increased. The Hsd11b1^−/− infarct also shows enhanced early infiltration of neutrophils, followed by increased numbers of macrophages, plausibly attracted by locally increased MCP-1 expression. These are predominantly YM1⁺ (M2) macrophages, implying a reparative, anti-inflammatory phenotype. M2-like polarization is not a general feature of 11β-HSD1^−/− macrophages (750), implying a vascular-specific effect.

2. Atherosclerosis

11β-HSD1 deficiency/inhibition improves multiple cardiovascular risk factors, anticipated to be atheroprotective. However, atherosclerosis has an important inflammatory component. 11β-HSD1 is expressed in differentiated/activated macrophages, and its deficiency exacerbates some models of acute inflammation (231). Nevertheless, 11β-HSD1 deficiency/inhibition indeed appears atheroprotective. A selective 11β-HSD1 inhibitor strikingly reduces circulating and intra-aortic cholesterol levels in atherosclerosis-prone Apoe^−/− mice (282), although another selective inhibitor has no effect on atherosclerotic lesions in Ldlr^−/− mice (415). Carbenoxolone, which inhibits both 11β-HSD1 and 11β-HSD2, reduces atherosclerosis in susceptible Ldlr^−/− mice (521). Moreover, Apoe^−/− mice crossed with Hsd11b1^−/− mice show a substantial reduction of atherosclerosis on cholesterol-rich “western diet,” in association with reduced lesional inflammation perhaps due to reduced vascular cell adhesion molecule (VCAM)-1-mediated recruitment of inflammatory cells to the plaque (347). Such findings suggest direct atheroprotective effects as a result of 11β-HSD1 inhibition, over and above reduction of cardiovascular risk factors. Indeed, recent data show atheroprotection can be conferred by transplanting 11β-HSD1-deficient bone marrow cells into irradiated Apoe^−/− hosts (346), implying a key role for 11β-HSD1 in macrophages and/or related leukocytes. If pertinent to humans such effects could be of substantial therapeutic importance (FIGURE 9).

Figure 9.

Possible impacts of 11β-HDSD1 inhibition/deficiency on atherosclerosis. Known effects include reductions of multiple risk factors and reduced atherosclerotic lesion size. If a myocardial infarction occurs, the angiogenic effects of 11β-HSD1 deficiency/inhibition promote angiogenesis and recovery of myocardial function for the same initial infarct size.

11β-HSD2 is also expressed in the vasculature, notably in the endothelium (84). Hsd11b2^−/− mice show endothelial dysfunction with enhanced norepinephrine-mediated contraction and reduced relaxation associated with impaired NO-related function (120, 257). Mice lacking 11β-HSD2 crossed with Apoe^−/− mice develop accelerated atherosclerosis even on chow diet, with increased lesional inflammation and reduced lesion-stabilising collagen (151). The accelerated atherogenesis is attenuated by the MR antagonist epleronone at a dose without effects on blood pressure implying, but not proving, actions via MR in the vasculature. A possible mechanism is via increased chemotaxis mediated by the observed increase in endothelial VCAM-1. The latter results from enhanced glucocorticoid activation of MR as it is mimicked by co-incubation of corticosterone with carbenoxolone (inhibiting 11β-HSD2) in mouse aortic endothelial cells (151). Thus loss of 11β-HSD2 leads to striking atherogenesis associated with activation of MR stimulating proinflammatory processes in the vessel wall. Tissue-specific manipulations are needed to dissect the cells involved in this process with potential implications for therapy.

3. Vascular injury

Glucocorticoids inhibit neointimal proliferation (729). 11β-HSD1 is expressed in the crucial vascular smooth muscle cells, and 11β-HSD1 deletion or inhibition reduces neointimal formation to wire injury. However, this is only observed in western diet-fed, atherosclerosis-prone Apoe^−/− mice, suggesting effects mediated primarily via impacts on metabolic risk factors. 11β-HSD2 deficiency had no effect (314). Such data do not support local vascular effects, at least in this model, on injury responses.

4. The heart

A single nucleotide polymorphism (SNP) in the HSD11B1 locus associates with left ventricular mass (568), apparently independent of any effect on blood pressure or body weight. Moreover, heart size of Hsd11b1^−/− mice is reduced (453). However, any function of the relatively low levels of either isozyme in the heart is unclear, as is the mechanism underlying the influence on heart mass. Cardiac MR appear to be occupied and tonically inhibited by physiological glucocorticoids (468). Transgenic overexpression of 11β-HSD2 in cardiac muscle induces fibrosis, supporting the notion that aldosterone-selective access to cardiac MR activates pathways leading to pathology, a concept supported by its improvement with the selective MR antagonist epleronone (561). Such molecular pharmacology is intriguing but hardly relevant to physiology. However, 11β-HSD2 mRNA levels selectively increase in hypoxic cardiomyocytes in association with inflammation, perhaps in part a consequence of reduced local glucocorticoids (352). Whether the benefits of 11β-HSD1 deficiency on recovery after experimental myocardial infarction reflect actions within the heart, over angiogenic and inflammatory actions, remains unexplored.

5. Genetics of HSD11B1 and cardiometabolic disease

Strong supporting evidence that 11β-HSD1 is causative in cardiometabolic disease (this is of course a distinct argument from its role as a drug target) could come from genetics. One study reported a weak association between HSD11B1 microsatellite sequence variation in intron 4 and central obesity (168) and another linked an intron 3 polymorphism to obesity, insulin resistance, and body composition in children (226), data echoed by the reported linkage of rs3753519 to obesity in children (536). The same intron 3 variant is linked to protection against type 2 diabetes, lower plasma insulin and insulin sensitivity independent of obesity in more than 700 Pima Native Americans (508). In the same population, the rs846910 SNP in the HSD11B1 promoter associates with blood pressure and hypertension, again independent of obesity (201), findings also reported in other populations (489). Also, combination of rs846910 GA heterozygosity and rs12086634 TT homozygosity that links to increased enzyme activity, associates with metabolic syndrome risk in southern European women (212). However, others (371) found effects on promoter activity without corresponding effects on metabolic phenotype in Koreans, and HSD11B1 has also not emerged from relevant genome-wide studies, although representation of this gene on many GWAS chips may be inadequate (212). The positive reports therefore either represent isolated examples in particular populations or may perhaps be false positives. Given the importance of the 11β-HSD1 hypothesis of metabolic disease, the genetics merit further consideration. Overall, however, the balance of current evidence suggests it is 11β-HSD1 dysregulation rather than HSD11B1 genetic variation that underlie the preponderance of its link to metabolic and other disorders.

V. 11β-HSD1 AND 11β-HSD2 IN IMMUNITY AND INFLAMMATION

Exogenous glucocorticoids in pharmacological doses are generally immunosuppressive. Endogenous glucocorticoids have a more nuanced effect, exerting permissive, suppressive, or stimulatory effects depending on concentration and the cellular and environmental context, thus modulating and shaping inflammatory and immune responses (reviewed in Refs. 135, 407, 452, 615, 655, 773, 774, 796). Perhaps unsurprisingly, a primary anti-inflammatory role for 11β-HSD1 has been demonstrated (136, 231, 808). However, the effects are complex and evidence, reviewed above, suggests that 11β-HSD1 deficiency/inhibition attenuates “metabolic inflammation” (obesity, atherosclerosis). Additionally, under some circumstances, 11β-HSD2 has an anti-inflammatory role, perhaps through its ability to control glucocorticoid activation of MR (151).

A. Expression of 11β-HSD1 and 11β-HSD2 in Immune Cells

Early studies suggesting immune effects of 11β-HSD largely failed to distinguish 11β-HSD1 and -2 or used nonselective licorice-based 11β-HSD inhibitors (199). The observed antiviral and anti-inflammatory effects could reflect renal 11β-HSD2 inhibition, delaying clearance of glucocorticoids (294), rather than inhibition of 11β-HSDs or effects on other targets. Recent studies unambiguously show isozyme expression in immune cells.

1. 11β-HSD1

A) MYELOID CELLS.

Glucocorticoids affect macrophage differentiation and function (135) which, in vivo, is highly diverse. During inflammation, macrophages predominantly derive from blood monocytes, which differentiate under instruction from the cytokine and chemokine microenvironment. 11β-HSD1 is not expressed in nonstimulated human monocytes but is induced, albeit to relatively low levels, following activation with the “Th2” cytokines IL-4, IL-13, or if monocytes are differentiated to macrophages in vitro (703). 11β-HSD1 expression is similarly modest in tissue-resident or bone marrow-derived macrophages of mice, the latter resembling human macrophages differentiated from monocytes in vitro (146, 231).

In macrophages, 11β-HSD1 expression is dynamically regulated depending on stimulus and context. Whereas incubation of human monocytes with bacterial lipopolysaccharide (LPS) has no effect on 11β-HSD1, in macrophages 11β-HSD1 is increased following LPS activation (termed “classical activation”) (326, 433, 703) (FIGURE 10), which polarizes macrophages to a pro-inflammatory M1 phenotype. Similarly, LPS induces 11β-HSD1 mRNA in microglia, the brain's resident macrophages, although curiously this is not accompanied by changes in protein (243).

Figure 10.

11β-HSD1 in macrophages reflects activation state and shapes the subsequent response towards resolution. Monocytes have negligible levels of 11β-HSD1 (703). After differentiation of human peripheral blood monocytes to macrophages, 11β-HSD1 expression is increased (703). Polarization to M1 with LPS and IFN-γ further increases 11β-HSD1 mRNA levels, whereas polarization to M2 with IL-4 and/or IL-13 has no further effect (326). Treatment of monocytes with IL-4 or IL-13 also increases 11β-HSD1 mRNA and activity levels (703), which are further increased by PPARγ activation (117). Differentiation of monocytes in the presence of glucocorticoid produces a highly phagocytic macrophage phenotype (230). Similarly, glucocorticoid treatment of macrophages promotes a phagocytic phenotype (412). High expression of 11β-HSD1 in M1 macrophages may promote the subsequent transition to a phagocytic phenotype, thus promoting resolution of inflammation. 11β-HSD1 is markedly downregulated following phagocytosis of apoptotic leukocytes (115). Dashed lines indicate speculative connections, whereas solid lines have been demonstrated.

“Alternative activation” of unstimulated macrophages with IL-4 or IL-13, which induces a pro-resolution M2 phenotype, has no effect on 11β-HSD1 (326, 433). This contrasts with the situation in monocytes (above). However, if monocytes are differentiated into macrophages in the presence of IL-4/IL-13, then levels of 11β-HSD1 are even higher than in M1-polarized macrophages and are further induced by PPARγ agonists, probably via a PPAR binding site exclusive to the human promoter (117). Chronic cold stress, which induces a regulatory (immunosuppressive) macrophage phenotype, also increases macrophage 11β-HSD1 (630). The relevance of these in vitro defined phenotypes to the situation in vivo is unclear (171), although the dynamic regulation of 11β-HSD1 by various stimuli suggests it shapes macrophage phenotype during the inflammatory response (231, 454). Phagocytosis of apoptotic neutrophils by macrophages which promotes resolution of inflammation, decreases 11β-HSD1, suggesting an active process (115).

Glucocorticoids suppress dendritic cell maturation and induce a tolerogenic phenotype eliciting T-cell hyporesponsiveness and regulatory T-cell formation (reviewed in Ref. 135). 11β-HSD1 induction accompanies dendritic cell differentiation. It is maintained or increased following maturation by innate immune activating signals (TNF-α, LPS) but decreased by CD40 ligation mimicking T_h cell engagement, an adaptive immune activation signal (203). The latter occurs without alteration in 11β-HSD1 mRNA or protein levels. These data suggest 11β-HSD1 inhibition might enhance immunity in some circumstances, a notion requiring exploration.

Other myeloid cell types express 11β-HSD1. Again levels depend on cell state. Neutrophil expression of 11β-HSD1 (336) is increased during inflammation (137) but is absent in apoptosis (115). Neutrophil number and apoptosis depend on glucose shuttling between the cytoplasm and ER (119), and perhaps 11β-HSD1 alters because of coupling to H6PDH. 11β-HSD1 is expressed in mast cells, where it limits degranulation (134), with implications for inflammation, infection and allergy.

B) LYMPHOCYTES.

Although 11β-HSD1 activity and protein have been demonstrated in lymphoid organs, most is localized to stromal cells (280), consistent with low 11β-HSD1 activity in mouse splenic lymphocytes and thymocytes (809) and negligible 11β-HSD1 mRNA in human blood leukocytes (703). Activation of lymphocytes, by T-cell receptor activation or cytokine-induced polarization, increases 11β-HSD1 activity (809). The biological role in lymphocytes remains unclear, although it appears to dampen both pro- and anti-inflammatory cytokine release (809).

11β-HSD1 (but not 11β-HSD2) mRNA has also been reported in uterine natural killer cells (451), important in preparing the uterus for pregnancy and in placental development (162). Expression may not be high in other NK cells (http://biogps.gnf.org), again emphasizing dependence on differentiation and/or activation state of the cell.

2. 11β-HSD2

Except in some human diseases, immune cells do not express 11β-HSD2 (134, 203, 703). 11β-HSD2 appears transiently expressed in peripheral blood mononuclear cells of patients with early rheumatoid arthritis (535), and 11β-HSD2 expression is higher in transformed B cell lines established from rheumatoid arthritis compared with those of nonarthritic twins (256) (levels in nonarthritic controls are probably extremely low). Immunohistochemistry suggests 11β-HSD2, as well as 11β-HSD1, in synovial macrophages in both osteoarthritis and rheumatoid arthritis (256, 264, 618). 11β-HSD2-immunoreactive macrophages are found in lungs of patients dying with acute respiratory distress syndrome (686). The significance and generality of these findings remain to be established, but it is noteworthy that 11β-HSD2, in contrast to 11β-HSD1, is widely expressed during early embryonic development and is frequently expressed in transformed cell lines (see below), but shows a highly restricted distribution in mature tissues and primary cells. Macrophage 11β-HSD2 in the rheumatoid arthritis synovium may reflect an adaptation to allow increased cell proliferation or an altered differentiation programme in the chronically inflamed microenvironment. These findings raise the possibility that intracellular glucocorticoid action may be very different in chronic compared with acute inflammatory responses.

B. 11β-HSDs in Acute Inflammation and Infection

1. Acute inflammation and 11β-HSD1

Neutrophil recruitment is a critical early determinant of inflammation and tightly coupled to monocyte recruitment (656). In Hsd11b1^−/− mice, more neutrophils and monocytes-macrophages are present early in acute inflammation in the peritoneal or pleural cavities (136). In contrast, 11β-HSD2 deficiency has no effect (136). Neutrophil density is also increased in the injured myocardium of Hsd11b1^−/− mice early following myocardial infarction (454), reflecting greater recruitment and/or delayed clearance. In support of the latter, Hsd11b1^−/− mice show delayed macrophage acquisition of in vivo phagocytic capacity for apoptotic neutrophils during sterile peritonitis, although surprisingly, the peritonitis resolves at the same time (231). 11β-HSD1 activity may promote neutrophil survival in vitro, although this is dependent on coupling of cortisol oxidation to glucose metabolism/NADPH generation in the ER rather than glucocorticoid metabolism per se (336). The significance of these findings merits further investigation.

Hsd11b1^−/− mice are also more susceptible to endotoxemia (808), classically restrained by glucocorticoids (76, 78), with higher serum pro-inflammatory cytokines TNF-α, IL-6, and IL-12p40 due to increased activation of NF-κB- and MAPK-signaling pathways in macrophages (808). This was attributed to attenuated PI3K-dependent Akt activation in 11β-HSD1-deficient macrophages, secondary to elevated levels of Src homology 2-containing-inositol 5′-phosphatase (SHIP)-1 (808), which hydrolyzes the PI3K-generated second messenger PIP₃ to PIP₂. Importantly, there are no differences between the glucocorticoid responsiveness and functional properties of Hsd11b1^−/− and wild-type macrophages differentiated in vitro, but Hsd11b1^−/− macrophages ex vivo are hyperresponsive to LPS (808), suggesting altered myeloid cell differentiation in vivo in Hsd11b1^−/− mice. Thus 11β-HSD1 plays an early constraining role in the inflammatory response, shaping the subsequent trajectory, probably through regulation of cytokine and chemokine production (231, 454, 808) or local glucocorticoid-mediated changes in cellular differentiation in inflamed tissues (135).

2. 11β-HSDs in the response to infection

There are few direct data on regulation of 11β-HSD1 and/or -2 during infection. What exists suggests the overall balance favors increased glucocorticoid generation. Whether 11β-HSD1 (or -2) affects resistance to pathogens will be of great interest.

Downregulation of 11β-HSD2 without alteration of 11β-HSD1 mRNA occurs in skin lesions of leprosy patients compared with unaffected skin (22), although this could be a response to damage rather than infection (707). In placenta, 11β-HSD1 increases and 11β-HSD2 decreases in chorioamnionitis (328). Carbenoxolone increases susceptibility to progressive disease in Listeria monocytogenes infection; as dexamethasone has a similar effect (279), this could reflect decreased glucocorticoid clearance by renal 11β-HSD2. A role for 11β-HSDs in shaping the course of tuberculosis has also been suggested (592), but is based largely on changes in urinary steroid profiles and remains speculative.

3. Regulation of 11β-HSDs in inflammation

Increased whole body conversion of cortisone to cortisol suggests 11β-HSD balance is altered in favor of 11β-HSD1 in patients with inflammatory disease (311). Increasing evidence suggests reciprocal regulation of 11β-HSD1 and 11β-HSD2 by pro-inflammatory cytokines at sites of inflammation, with upregulation of 11β-HSD1 and downregulation of 11β-HSD2. This regulation, particularly of 11β-HSD2, which is not expressed under most circumstances in leukocytes, is probably attributable primarily to 11β-HSD1 and -2 in nonimmune cells. This balance is predicted to increase local glucocorticoid concentrations, but direct evidence of this in inflamed tissues is lacking. There are also important exceptions to this generalization.

4. Reciprocal regulation of 11β-HSD1 and 2 by pro-inflammatory mediators

Although coexpressed within tissues, 11β-HSD1 and -2 are not usually colocalized within a cell. If this occurs, reciprocal regulation appears to minimize simultaneous expression. In vitro, in most cell types the pro-inflammatory cytokines TNF-α and IL-1 up regulate 11β-HSD1 (109, 188, 260, 263, 403, 567, 681, 700, 711, 798) and downregulate 11β-HSD2 (271), the latter mediated through a switch from active p65/p50 NF-κB heterodimers bound to the HSD11B2 promoter to inactive p50/p50 homodimers which coordinately bind with Egr-1 (362). This affords a plausible basis for the generalization that in vivo 11β-HSD1 is upregulated at sites of inflammation and 11β-HSD2 downregulated. Indeed, TNF-α-expressing transgenic mice have elevated hepatic 11β-HSD1 (312) and reduced renal 11β-HSD2 (362). However, 11β-HSD1 was not among mRNAs increased in liver, muscle, or adipose tissue following 1- or 4-day TNF-α infusion in rats (598), nor was it altered in murine vasculature following LPS (163). Importantly, TNF-α and LPS effects on 11β-HSD1 in immune cells themselves are highly context dependent and probably require a costimulus (203).

5. Glucocorticoids amplify 11β-HSD1 induction by inflammation

The actions of proinflammatory cytokines and glucocorticoids are usually mutually antagonistic. However, 11β-HSD1 expression is synergistically induced by TNF-α or IL-1 and glucocorticoids in a range of cell types (338, 403, 567, 681). The purpose might be to amplify intracellular glucocorticoid-mediated attenuation of proinflammatory cytokine action in preparation for recovery in key tissues.

6. Molecular mechanisms

IL-1β and TNF-α induce 11β-HSD1 in some human-derived cells (fetal lung fibroblasts, hepatoma) via C/EBPβ binding to the proximal HSD11B1 P2 promoter (790). However, these cytokines induce 11β-HSD1 via NF-κB in mouse mesenchymal cells (11). Curiously, in adipocytes, TNF-α decreases 11β-HSD1 expression through NF-κB, although this may have been related to decreased C/EBPα (597). This complexity probably underpins the cell specific effects of pro-inflammatory cytokines on 11β-HSD1 expression. Induction of 11β-HSD1 by TNF-α in adipose stromal cells is inhibited by insulin (260), suggesting insulin counters the effect of TNF-α and that TNF-α induction of 11β-HSD1 predominates in insulin resistance (e.g., obesity).

Interestingly, hypoxia downregulates 11β-HSD2 in rat kidney (270) and in placenta (262). Translational and transcriptional mechanisms (293), the latter perhaps via Egr-1 (270), have been invoked, although hypoxia causes inflammation (499), suggesting proinflammatory cytokine actions. Reduced glucocorticoid catabolism by 11β-HSD2, and hence increased local steroid action in placenta, may engender adaptation to hypoxia and its associated inflammation (300) by increasing fetoplacental maturational signals and suppressing inflammation. However, in adipose tissue, higher 11β-HSD1 (and hence increased local glucocorticoid levels) maximizes susceptibility to hypoxic damage via inhibition of compensatory angiogenesis (467). Clearly the balance of isozymes and tissues is complex.

C. 11β-HSDs in Disorders of Immunity, Inflammation, and Neoplasia

1. Bone and joint: inflammatory arthritis

The first clinical glucocorticoid therapy used cortisone to elicit a remarkable improvement in rheumatoid arthritis (275). While inert cortisone is largely activated to cortisol by the liver, 11β-HSD1 also plays a local role in controlling inflammation during rheumatoid arthritis. In the K/BxN serum transfer model of self-resolving experimental arthritis, Hsd11b1^−/− mice show an earlier onset and delayed resolution of inflammation (136). They also exhibit greater reactive bone growth (exostosis), which could result from greater inflammation within the joint or a greater reaction of the tissue to inflammatory mediators. Hsd11b2^−/− mice show a “wild-type” response (136). A role for endogenous glucocorticoids in bone turnover in arthritis has been suggested in transgenic mice expressing 11β-HSD2 specifically in osteoblasts and osteocytes (108). In these, the initial phase of K/BxN serum transfer arthritis is identical to wild-type, so presumably does not involve glucocorticoid action in osteoblasts. However, in the subsequent subacute phase, inflammation and cartilage destruction are attenuated indicating these processes depend on glucocorticoid action in osteoblasts (108). This contrasts with the phenotype of the Hsd11b1^−/− mice in the same experimental model and suggests that glucocorticoids exert opposing effects in different cells/tissues during inflammation.

Patients with rheumatoid arthritis have increased 11β-HSD1 activity in synovial tissues (618) which correlates with the severity of inflammation (618). Within the rheumatic joint, 11β-HSD1 is mainly expressed in synovial fibroblasts where it is highly inducible by IL-1 or TNF-α (263, 264). In a rat adjuvant arthritis model, levels of 11β-HSD1 mRNA in the inflamed synovium and in immune cells of the draining lymph nodes increase markedly and are reduced by anti-inflammatory therapy by IL-1 receptor or TNF-α antagonism (185). 11β-HSD2 is restricted to the nonimmune cells of lymph nodes and is decreased in arthritis (185).

Dysregulation of 11β-HSD isozymes may contribute to the failure of resolution of chronic inflammation. 11β-HSD2 is expressed in synovial macrophages in rheumatoid arthritis (618) and, transiently, in peripheral blood mononuclear cells (535). Moreover, whereas IL-10 rapidly increases 11β-HSD1 mRNA in normal human macrophages, this fails in synovial macrophages in rheumatoid arthritis (29). This suggests that during chronic inflammation, a maladaptive mechanism may be engaged in which the balance of 11β-HSD activities is unfavorably altered, resulting in reduced intracellular glucocorticoid action in synovial macrophages and possibly local resistance to glucocorticoids that are substrates for 11β-HSD (e.g., prednisolone).

A) BONE.

In bone, 11β-HSD1 is mainly expressed in osteoblasts (131). Young and middle-aged Hsd11b1^−/− mice have normal bone mass, bone formation, and bone resorption (333). However, 11β-HSD1 levels in bone increase with age or glucocorticoid exposure (130, 759), suggesting that endogenous glucocorticoids may contribute to age-related osteoporosis. Consistent with this notion, transgenic expression of 11β-HSD2 in osteoblasts and osteocytes protects against the age-related decline in bone strength and mass (759). The underlying mechanism may be alleviation of glucocorticoid inhibition of angiogenesis, mediated through suppression of HIF1α and VEGF-α production by osteoblasts and osteocytes (759). The effects of 11β-HSD1 inhibitors on bone structure and function with ageing remain to be established.

2. Lung

A) 11β-HSDS ARE EXPRESSED IN LUNG.

Both 11β-HSD isozymes are expressed in human lung. 11β-HSD2 is present in airway epithelia in adults (bronchiole to trachea) and the vascular endothelium (538, 687). 11β-HSD1 is found in alveoli. Levels vary greatly between individuals (687), possibly as a consequence of disease or glucocorticoid therapy (686). The adult rat lung barely expresses 11β-HSD2 (17). 11β-HSD1 is expressed (369), although more in interstitial fibroblasts than alveolar cells (86). Any function in adult lung remains uncertain, although the presence of fibrous adhesions in inflamed lungs of Hsd11b1^−/−, but not wild-type mice (136), suggests 11β-HSD1 limits fibrosis during lung inflammation.

Late fetal lung maturation is dependent on, and accelerated by glucocorticoids (200, 405), which induce the synthesis and release of lung surfactant (43) and regulate epithelial cell differentiation (125). This is exploited clinically in the administration of synthetic glucocorticoids to women suffering threatened preterm labor to improve neonatal survival, especially respiratory distress syndrome (28). Human and rodent fetal lung have a developmental switch from 11β-HSD2 in early and mid-gestation (93, 126, 502) to 11β-HSD1 activity in late gestation (502, 662, 705). The timing of this “switch” is advanced by glucocorticoid administration in rodents (717) and in humans, by a pituitary factor (502), probably ACTH, driving fetal glucocorticoid secretion. Increased oxoreductase (11β-HSD1) activity at term parallels increased phospholipid biosynthesis (518) consistent with glucocorticoid stimulation of surfactant production (43). Deficiency or inhibition of 11β-HSD1 reduces fetal lung phospholipid and surfactant synthesis (309, 310, 649) and increases lung weight (649), the latter consistent with a permanent reduction in lung-to-body weight ratio following prenatal glucocorticoid exposure (667). Thus 11β-HSD1 inhibition in late pregnancy seems ill-advised. 11β-HSD1 induction in the late gestation lung may help its maturation without unwanted systemic glucocorticoid effects such as developmental programming (625).

Little definitive is known of any role for 11β-HSDs in asthma susceptibility or progression, although 11β-HSD2 immunoreactivity is inversely related to the dose of inhaled corticosteroid required for efficacy in asthma (538), perhaps reflecting intact downregulation of 11β-HSD2 by glucocorticoids in otherwise glucocorticoid-resistant airways. In human airway smooth muscle cells, the pro-asthmatic Th2 cytokine IL-13 increases 11β-HSD1 expression (via MAPK signaling through ERK1/2 and JNK pathways). This is anticipated to increase local glucocorticoid regeneration and may be a normal mechanism to enhance protection against the pro-asthmatic effects of IL-13 on bronchoconstrictor pathways (298). Not normally expressed in alveoli, in acute respiratory distress syndrome, 11β-HSD2 is induced in alveolar macrophages and type II cells (686), raising the possibility that dysregulated 11β-HSD2 contributes to the glucocorticoid resistance of this severe disorder (431). These interesting disparate observations emphasize the lack of any mechanistic understanding of 11β-HSDs in the lung, a topic worthy of considerable study.

B) LUNG CANCER.

Beyond glucocorticoid metabolism, 11β-HSD1 also has carbonyl reductase activity, metabolizing key carcinogenic nitrosamines in tobacco smoke (657). Levels of the 11β-HSD1 substrate 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone and its active products predict lung cancer risk (800). The striking variation in lung 11β-HSD1 may associate with individual susceptibility to tobacco-induced cancers. At least some of this variation is genetic and associates with SNPs in the hsd11b1 promoter that are in linkage disequilibrium with a functional variant, rs13306401 (698). This genetic variance of 11β-HSD1 may be compounded by inhibition of 11β-HSD1 by licorice added to smoking tobacco (438). However, there is some lack of clarity as to whether the enzyme activates or inactivates carcinogens. Again, the biological processes underlying this idea remain to be tested, including straightforward studies of the susceptibility to tobacco carcinogenesis with 11β-HSD1-deficiency or inhibition.

3. The gastrointestinal tract and inflammatory bowel disease

In humans, 11β-HSD1 is expressed in the normal colonic epithelium as well as in nonepithelial cells of the lamina propria (654, 770). In rats, 11β-HSD1 is exclusively localized in the lamina propria (766), whereas 11β-HSD2 is confined to surface and crypt epithelial cells (770).

In both ulcerative colitis and in Crohn's disease, 11β-HSD1 mRNA is increased in the inflamed colon, whereas 11β-HSD2 mRNA and protein are decreased (802). These effects are replicated by TNF-α in colonic tissue explants (186). Interestingly, the 11β-HSD1 increase in patients with inflammatory bowel disease is greater in men (664). The same reciprocal regulation of 11β-HSD1 (increased) and 11β-HSD2 (decreased) is seen in rat experimental models of colitis (96, 184), with the increase in 11β-HSD1 in colonic intraepithelial lymphocytes consistent with migration of activated lymphocytes from the inflamed colon to mesenteric lymph nodes and spleen (184, 186). Mouse models of inflammatory bowel disease also show increased 11β-HSD1 in intestinal intraepithelial lymphocytes and mesenteric lymph nodes, but 11β-HSD2 is unchanged (724). The reciprocal regulation of 11β-HSD1 and -2 expression in colon may be mediated by the pro-inflammatory cytokines TNF-α and IL-1, which decrease 11β-HSD2 mRNA levels and enzyme activity in human SW620 colon carcinoma cells (362). Again, beyond the observations, the biological importance of these changes is unclear and requires investigation, especially as 11β-HSD1 inhibitors enter human clinical trials. Any clinical hazards of these agents in inflammatory bowel disease and other inflammatory disorders are unknown.

4. The reproductive system

A) THE FEMALE REPRODUCTIVE SYSTEM.

In most species, ovary (15, 68, 127, 581, 588) and uterus (103, 451) express both 11β-HSD1 and -2. 11β-HSD1 is also expressed in oocytes (68, 581), although any function is uncertain. In ovarian granulosa cells, a switch from 11β-HSD2 to 11β-HSD1 expression occurs immediately prior to ovulation, with the induction of 11β-HSD1 probably mediated by luteinizing hormone and augmented by IL-1β (700, 701) or prostaglandins (329), although the latter may be mediated posttranslationally. Interestingly, granulosa cell 11β-HSD1 activity is decreased by depletion of the leukocytes in human follicular aspirates, and restored by addition of IL-4 and IFN-γ, suggesting the source of cytokines in this regulation may be leukocytes (192). The presence of 11β-HSD2 activity (conversion of cortisol to cortisone) in cultured granulosa cells predicted pregnancy failure in women undergoing in vitro fertilization, whereas its absence was associated with successful pregnancy, although there was no association with fertilization per se (464). The underlying reason for this has not been elucidated. In rats, 11β-HSD2 is induced in the corpus luteum, coincident with luteal regression (735).

Ovulation is a natural inflammatory process in which the ovarian surface is cyclically ruptured and repaired following ovulation. The induction of 11β-HSD1 by IL-1 and other pro-inflammatory mediators in ovarian surface epithelial cells (567, 798) may quell inflammatory tissue damage following ovulatory rupture, supporting the nonscarring of this tissue despite cyclical injury.

There are conflicting reports on the localization and regulation of 11β-HSD1 and 11β-HSD2 in uterus. Some report mutually exclusive distributions with 11β-HSD1 in epithelial cells and 11β-HSD2 in endometrial stroma and myometrium (103, 706). In contrast, in humans, 11β-HSD2 is localized to the endometrial glandular epithelium, with little 11β-HSD1 in endometrium until menstruation (451). In rats, expression of both 11β-HSD1 and -2 is cycle dependent, maximal at proestrus, and minimal at diestrus, dependent on estrogen (103). During pregnancy in rats, 11β-HSD1 is dramatically upregulated in the myometrium shortly before parturition (736), an effect dependent on the placenta (736). In human endometrial cells, 11β-HSD1 mRNA levels markedly increase with decidualization (33, 34), suggesting a possible role for the enzyme in implantation. This has yet to be tested.

The biological relevance of these 11β-HSD expression patterns in ovary and uterus is unknown. Neither 11β-HSD1- nor 11β-HSD2-deficient mice show gross fertility defects. However, both 11β-HSD1 and -2 may play a role in fine-tuning uterine growth and remodeling, during menstrual/estrus cycling and in pregnancy. A role for 11β-HSD2 in promoting endometrial angiogenesis may underlie excessive menstrual bleeding in some women (566). Moreover, and consistent with reciprocal regulation of 11β-HSD1 and -2 during inflammation, 11β-HSD1 expression is increased and 11β-HSD2 decreased in endometriosis tissue, mimicked in the induction of 11β-HSD1 but repression of 11β-HSD2 by TNF-α in primary endometriotic stromal cells (484), suggesting increased glucocorticoid action in endometriosis. Again, the mechanism and relevance of this observation needs to be dissected.

B) THE MALE REPRODUCTIVE SYSTEM.

Glucocorticoids suppress Leydig cell production of testosterone by inhibiting steroidogenic enzyme expression. Colocalized 11β-HSD might be expected to influence this. However, although first described in the testis over 40 years ago (321, 353), the role of testicular 11β-HSD and indeed even its reaction direction remain contentious. Both 11β-HSD1 and 11β-HSD2 are expressed in the adult male reproductive tract, although there are substantial species differences (4, 369). In fish testis, 11β-HSD2 plays a glucocorticoid-independent role, converting 11β-hydroxytestosterone to 11-ketotestosterone, a major fish androgen important in fertility and sex determination (197, 324, 573). 11β-HSD2 is absent from the chicken testis (351). In mammals, 11β-HSD2 is expressed in human but not rat testis (15, 401). Neither 11β-HSD isozyme is found in the mouse testis (127, 488), and both 11β-HSD1 and 11β-HSD2 knockout male mice are normally fertile (363, 364), establishing that they are not essential for testicular function in mice, at least.

11β-HSD1 is expressed in rat Leydig cells (558). Some report it exhibits 11β-dehydrogenase activity (196, 214, 482), others 11β-reductase (396). There may be even be a switch from reductase in immature to dehydrogenase in mature Leydig cells (218). H6PDH is highly expressed in rat testis (241), so any oxidative activity of 11β-HSD1 cannot be attributed to lack of H6PDH. Rat studies with selective inhibitors will be particularly informative in establishing a role, if any, in the testis.

11β-HSD1 is present in prostate, principal epithelial cells of the epididymis, the epithelium of the vas deferens, seminal vesicle, and penile urethra as well as smooth muscle cells of the vas deferens and penile blood vessels (480, 738). 11β-HSD2 shows a complementary distribution, with expression in the clear epithelial cells of the epididymis and the corpora cavernous of the penis but not other tissues (738). The differential expression patterns of 11β-HSD1 and 11β-HSD2 throughout the male reproductive tract suggest they may modulate glucocorticoid and mineralocorticoid actions, particularly in the penile vasculature (738), but any function remains to be established.

Whole body glucocorticoid metabolism by 11β-HSD1 has endocrine consequences on gonads. Patients with apparent cortisone reductase deficiency due to H6PDH mutations (390) or the milder cortisone reductase deficiency due to HSD11B1 mutations (393) have hirsutism and often precocious pseudopuberty. The phenotype is plausibly due to overproduction of adrenal androgens, mainly DHEA and its sulfate, secondary to HPA axis activation to compensate for the lack of glucocorticoid regeneration by hepatic 11β-reductase.

5. Cancer, neoplasia, and cell proliferation

In accordance with the largely antiproliferative, pro-differentiation effects of glucocorticoids and their maturational role in preparation for birth (200), 11β-HSD2 is expressed in immature cells, protecting tissues from inappropriate glucocorticoid exposure during development (290, 780), whereas 11β-HSD1 is expressed only from late in development, apparently amplifying glucocorticoid maturational effects on fetal tissues (309). This ontogenic pattern is recapitulated in tumor-derived clonal cell lines, very few of which express much 11β-HSD1. Cells that do highly express 11β-HSD1, such as 3T3-L1 adipocytes (derived from normal cells; Refs. 245, 246), only do so once they stop proliferating and terminally differentiate (510). In contrast, many cell lines express 11β-HSD2. This has led to the concept that 11β-HSDs influence cell proliferation and neoplasia (308, 564, 565). Not surprisingly, transfection with 11β-HSD2 renders cells insensitive to the antiproliferative effects of cortisol, whereas cortisone is antiproliferative only in cells transfected with 11β-HSD1 (565). Similarly, inhibition of endogenous 11β-HSD2 by glycyrrhetinic acid in breast cancer or endometrial cell lines decreases the antiproliferative effect of glucocorticoids (308, 365, 366).

Consistent with this, the expression of 11β-HSD isozymes is altered in malignancy. 11β-HSD1 is decreased and 11β-HSD2 increased in pituitary tumors compared with normal pituitary (359, 563). In breast cancer, 11β-HSD1 expression is reduced compared with adjacent nonmalignant tissue (421) and, while 11β-HSD2 is not normally expressed in breast, it is found in two-thirds of tumor samples (366). Similarly, although 11β-HSD2 is not expressed in normal adult human adrenal gland (but is in fetal), it is expressed in adrenal cortical carcinoma and adenoma (133) and in colonic adenoma (806). However, the opposite regulation has been reported when adenomas have progressed to adenocarcinomas, with 11β-HSD1 increased and 11β-HSD2 decreased (801). Whether this regulation is causal in neoplastic transformation or a consequence (bystander effect) is unknown. The former is suggested by the effects of the nonselective 11β-HSD inhibitor glycyrrhizic acid, which decreases adenoma number and size in a mouse model of colorectal cancer. However, the anti-cancer effect apparently of 11β-HSD2 inhibition was attributed to enhanced glucocorticoid-suppression of COX-2 and PGE₂ which drive neoplasia in this model (806) rather than a direct effect of glucocorticoids on cellular proliferation.

In childhood acute lymphoblastic leukemia (ALL) cells, 11β-HSD1 is upregulated by dexamethasone in cells sensitive to glucocorticoid-induced apoptosis in vitro, but downregulated in glucocorticoid-resistant cells (605). 11β-HSD2 is also expressed in leukemic cell lines, where inhibition increases the cytotoxic potency of cortisol (696) or prednisolone (606) (also readily catabolized by 11β-HSD2), to the equivalent of dexamethasone, poorly inactivated by 11β-HSD2, demonstrating that 11β-HSD2 contributes to prednisolone resistance in these cells, although it by no means accounts for it. Given that childhood ALL is treated primarily with dexamethasone in many countries, 11β-HSD2 is unlikely to contribute to resistance in a clinical setting. More probable, the dysregulation of 11β-HSDs reflects the hematopoietic differentiation stage (intrinsically sensitive or resistant to endogenous glucocorticoids; Ref. 87) at which ALL lymphoblasts are arrested. Overall, the data support the idea that 11β-HSD dysregulation is a consequence rather than cause of neoplastic transformation, but their expression pattern may be a useful biomarker of disease state and steroid sensitivity in relevant cancers such as ALL.

6. Skin

Glucocorticoids inhibit wound healing by reducing angiogenesis (via VEGF), cell motility, scar formation, and remodeling (via TGF-β and the matrix metalloproteinase system), suppressing differentiation and promoting terminal epidermal differentiation (677). 11β-HSD1 is widely expressed in human and mouse epidermis and dermis (276, 699), but is highest in keratinocytes in the suprabasal area of the epidermis and dermal fibroblasts where it is a reductase (121). Keratinocyte 11β-HSD1 increases with differentiation (699), parallelling the enzyme's pattern during differentiation of preadipocytes (510). 11β-HSD2 mRNA is also detectable by PCR in human skin (732), although whether this extends beyond known expression in the mineralocorticoid target cells of sweat glands (82) or if there is sufficient to influence glucocorticoid function elsewhere in the dermis/epidermis is unknown.

Skin 11β-HSD1 expression increases with aging and photodamage (707). Selective inhibition of 11β-HSD1 in mouse skin causes keratinocyte proliferation and enhances cutaneous wound healing. This effect was greatest in Lep^ob/ob obese and diabetic mice which have higher cutaneous 11β-HSD1 expression (although whether or not this reflects increased enzyme in subcutaneous adipose tissue is unclear). This suggests a mechanism by which increased 11β-HSD1 may contribute to loss of skin elasticity and poorer healing with age (699). The application of inhibitors in skin disease merits examination, especially to promote wound healing, notably in the obese, diabetics, and the elderly.

7. The eye

Both isozymes of 11β-HSD have been reported in the mammalian eye including in humans (678). 11β-HSD1 is most highly expressed in the nonpigmented epithelium of the ciliary body (576), but mRNA is also present in trabecular meshwork, corneal epithelium and endothelium, and anterior lens epithelium (678). Expression of 11β-HSD2 is less certain with both negative (576) and positive (678) reports of mRNA and activity in human and rodent ocular tissues, notably also in the ciliary body nonpigmented epithelium. In terms of function, location of 11β-HSD in the ciliary body implies a role in aqueous humor formation. Glucocorticoid excess/hypersensitivity is linked to open-angle glaucoma. In rabbits, 11β-HSD inhibition with the nonselective agent glycyrrhizin reduced intraocular pressure elevated by the synthetic glucocorticoid triamcinolone (659), although the mechanism(s) and isozyme(s) involved is uncertain. In humans, carbenoxolone reduces intraocular pressure, compatible with reduced glucocorticoid action via 11β-HSD1 inhibition (576). Such data require confirmation and dissection with isozyme selective drugs.

An additional role of 11β-HSD1 in the eye involves the substantial retro-orbital adipogenesis in thyroid-associated (Grave's disease) ophthalmopathy. Primary cultures of adipocytes from thyroid-associated ophthalmopathy patients have much higher expression of 11β-HSD1 than controls and produce a host of proinflammatory cytokines, which might underlie the increase in the enzyme, although this might merely reflect the more differentiated state of ophthalmopathy adipocytes. Cytokine production by these cells is attenuated by 11β-HSD inhibition (710), and the enzyme plays a role in adipogenesis in retro-ocular as other sites. Any role in therapy of this challenging disorder is uncertain, but merits exploration.

VI. 11β-HSDs AND THE BRAIN

While early studies reported 11β-dehydrogenase activity in glioma cell lines (250) and neonatal primate brain homogenates (251), the key reports of what turned out to be 11β-HSD2 in kidney failed to find 11β-HSD activity in the brain (181). This was fully in keeping with the presence of MR occupied by glucocorticoids in hippocampus (580), implying a lack of “protection” of such MR by 11β-dehydrogenase. However, the reason early studies failed to find ex vivo 11β-HSD activity in brain homogenates is probably due to lower levels of endogenous NADP(H) in brain than liver (215, 232). Exogenous cofactor reveals the activity (477, 478). 11β-HSD1 mRNA, immunoreactivity, and bioactivity have been repeatedly shown in hippocampus and myriad other CNS regions. 11β-HSD2 is highly expressed in the fetal CNS but is nearly absent from the adult brain. 11β-HSD1 is emerging as a key control of local glucocorticoid exposure with particular relevance to cognitive function in the ageing brain.

A. Distribution and Ontogeny

1. 11β-HSD1 in the adult CNS

11β-HSD1 is widely and unevenly distributed in the adult CNS. Although 11β-HSD1 is highly expressed in the cerebellum, hippocampus, and cortex, there is a discrete microdistribution within these areas. Highly expressing cells are seen in the Purkinje cell layer of the cerebellum, CA3 pyramidal cells of the hippocampus, and layer V neurons of the neocortex (478). There is also low expression of 11β-HSD1 generally throughout the brain and spinal cord, including the paraventricular nucleus of the hypothalamus, a key locus for glucocorticoid feedback control of the HPA axis (607). Intriguingly, 11β-HSD1 is found both in neurons and glia (607), notably in microglia (243) which express high levels of 11β-HSD1 particularly when activated, implicating this enzyme as a regulator of central inflammatory signals. Indeed, in humans, 11β-HSD1 levels in microglia appear increased in multiple sclerosis (268). 11β-HSD1 is also expressed in anterior pituitary cells including corticotrophs (359).

Rajan et al. (572) showed that primary cultures of rat hippocampal cells express 11β-HSD1 but not 11β-HSD2. Activity is exclusively 11β-reductase. In vitro, pretreatment with intrinsically inert 11-dehydrocorticosterone is equipotent with corticosterone in promoting death of hippocampal cells in the presence of high, but sublethal doses of the excitatory glutamatergic neurotransmitter kainic acid. Addition of carbenoxolone, itself without neurotoxic effects, attenuates the toxicity of 11-dehydrocorticosterone, but not corticosterone (572). These data support the 11β-reductase reaction direction of 11β-HSD1 in hippocampal cells and imply a role in amplifying intracellular glucocorticoid action.

2. 11β-HSD2 in the adult brain

11β-HSD2 expression in the adult nervous system is highly spatially restricted, much more so than MR. In the rat brain, in situ hybridization localized 11β-HSD2 mRNA to scattered cells of the ventromedial and paraventricular (PVN) nuclei of the hypothalamus, amygdala, locus ceruleus, subcommissural organ, and nucleus tractus solitarius (NTS) (583, 590, 810), areas associated with salt appetite and blood pressure regulation which are selectively affected by aldosterone rather than glucocorticoids. Distribution of 11β-HSD2 within the mouse brain is even more limited, confined to the NTS (290, 291), consistent with lesser aldosterone dependence on salt regulation in mice (596). Any 11β-HSD2 mRNA outside the NTS-sodium appetite/central cardiovascular control circuitry appears to reflect low-level expression without clear functional importance. A “knock-in” mouse in which Cre recombinase was driven by the endogenous Hsd11b2 gene promoter most likely reflects widespread fetal brain Hsd11b2-driven Cre-mediated activation of the ROSA26 reporter used, which remains expressed thereafter (221).

With the use of highly sensitive quantitative RT-PCR, some reports detect 11β-HSD2 mRNA in the adult human brain (799) but have yet to establish any functional significance of the low copy number transcripts. Within the PVN, 11β-HSD2 mRNA has been detected using RT-PCR (810), inhibition of 11β-HSD within the PVN increased sympathetic outflow and PVN activity (810), and the sympatho-excitory effects of carbenoxolone were blocked by intracerebroventricular spironolactone, an MR antagonist (810). However, the lack of specificity of the inhibitors used in this study [carbenoxolone and glycyrrhetinic acid also potently inhibit 11β-HSD1 which is expressed in the hypothalamus (478, 607, 624)] and the lack of spatial selectivity with intracerebroventricular injections, means any functional role for 11β-HSD2 mRNA in the PVN presently remains uncertain. In human brain, exploiting samples with very short post-mortem delays (<4 h), no 11β-HSD2 mRNA or 11β-dehydrogenase activity was found in cortex, hippocampus, or cerebellum (613).

In contrast, there is extensive 11β-HSD2 mRNA and activity in the fetal brain with complex age- and locus-specific patterns of silencing (discussed below).

B. The HPA Axis

Expression of 11β-HSD1 in brain regions responsible for the negative-feedback actions of glucocorticoids upon the HPA axis (cerebral cortex, hippocampus, hypothalamic PVN, and pituitary) suggests the enzyme might modulate feedback. Indeed, the original mice lacking 11β-HSD1 on a 129/MF1 strain background exhibit elevated basal and poststress corticosterone levels consistent with reduced glucocorticoid feedback (265). Interestingly, moving the disrupted 11β-HSD allele onto the C57BL/6J background normalized basal plasma corticosterone, with an efficient negative-feedback signal. The difference between mouse strains appears due to variability in plasticity in the key feedback sites, with most strains, including C57BL/6, showing a compensatory rise in GR expression in the hippocampus and PVN of the hypothalamus while the opposite occurs in the 129 strain of mice (113). However, all strains of 11β-HSD1-deficient mice have larger adrenals than their wild-type littermates with an exaggerated corticosteroid response to stress (113). Thus the genetic background is a crucial factor governing the consequences of 11β-HSD1 loss on the HPA axis.

Interindividual differences in vulnerability to disorders associated with HPA axis dysregulation raise the possibility that despite the fact that 11β-HSD1 inhibition does not inevitably activate the HPA axis and glucocorticoid levels, it may in certain susceptible individuals. However, to date, trials of 11β-HSD1 inhibitors in rodent models or clinical trials have failed to show elevation of basal glucocorticoid levels (594). Although adrenal enlargement is observed with 11β-HSD1 inhibition, this is thought to result from compensatory increased drive to the adrenals to replace the contribution of cortisol that normally derives from 11β-HSD1 regeneration. Peripheral 11β-HSD1 in the liver and splanchnic bed contributes substantially (20–40%) to total daily glucocorticoid production by regenerating cortisol from inert cortisone (62, 742). 11β-HSD1 inhibition in humans increases serum levels of ACTH and the ACTH-induced adrenal sex steroid precursor dehydroepiandrosterone, without changes in cortisol, presumably reflecting these compensatory processes (195, 594). The increased adrenal androgen production with short-term inhibition of 11β-HSD1 remains within the normal range (195). Whether or not longer term inhibition produces hirsutism, as reported in rare congenital HSD11B1 mutations in humans (393), remains to be seen, but dehydroepiandrosterone levels fall dramatically with ageing (71) so the side effect may be less in middle-aged and elderly subjects.

Intriguingly, on the mixed C57BL6/CBA/129 strain background in which Hsd11b1 gene disruption causes basal HPA axis activation, transgenic replacement of 11β-HSD1 selectively in the liver normalizes the basal and stress-induced elevations of plasma corticosterone levels and adrenal weight (547). This implies that peripheral (liver) 11β-HSD1 is key to the HPA abnormalities, a suggestion supported by the increased adrenal size of mice conditionally lacking hepatic 11β-HSD1 (392). No effects on circulating levels of corticosterone are observed in mice with overexpression of 11β-HSD1 in the liver (548), in fat (446), or in the forebrain (the hypothalamus does not express the transgene) (289). Furthermore, ectopic expression of 11β-HSD2 in fat reduces adipose corticosterone levels, but circulating glucocorticoid levels remain normal (342). It remains unknown whether the consequences of global deletion of 11β-HSD1 on HPA activity can be mimicked by loss of 11β-HSD1 solely in the brain. Currently, the data support the interpretation that bulk changes in glucocorticoid flux (largely mediated in liver) cause the observed adrenocortical hyperplasia with 11β-HSD1 deficiency. This also squares with emerging data on the role of carbohydrates and other energy substrates produced by the liver and of hepatic vagal afferents in HPA regulation (375, 387).

1. 11β-HSD1 and appetite regulation

Some findings suggest that glucocorticoids act within adipose tissues to control appetite. Mice overexpressing 11β-HSD1 under the aP2 promoter in adipose tissue show hyperphagia, whereas ectopic expression of 11β-HSD2 in adipose tissue induces hypophagia (342, 446) (a caveat is possible ectopic transgene expression). Paradoxically 11β-HSD1 knockout mice show increased transiently appetite for high-fat diet (493, 494), indicating a possible distinct central effect. Indeed, 11β-HSD1 mRNA and enzyme activity are expressed in the hypothalamic arcuate nucleus (478), a key locus for appetite control. Arcuate 11β-HSD1 is acutely induced by high-fat feeding (149), with the timing suggesting this may be part of the acute inflammatory hypothalamic response to high-fat diet (702). Furthermore, arcuate expression of appetite control neuropeptides is altered in 11β-HSD1-deficient mice with reduced anorexigenic cocaine and amphetamine-regulated transcript and melanocortin-4 receptor mRNAs, implying increased appetitive drive. Moreover, on high-fat diet, arcuate mRNA for orexigenic agouti-related peptide increases in Hsd11b1^−/− mice, whereas it decreases in wild-type (149). This regulation appears mediated through μ-opioid receptor tone, a key control of appetitive reward. Hence, arcuate 11β-HSD1 appears to contribute to central adaption to dietary challenge, at least in rodents.

C. Cognition and Ageing

1. Cognition in young animals

Although 11β-HSD1 is widely expressed in areas of the brain involved in learning and memory (hippocampus and neocortex), young adult 11β-HSD1 knockout show normal acquisition and retention of spatial memory in tests of cognitive function such as the watermaze and Y-maze (791, 794). However, young mice are adept at these tasks and require fairly major cognitive deficits to alter behavior, an impact not anticipated with 11β-HSD1 deficiency. It is only as the animals age that major cognitive phenotypes associated with 11β-HSD1 and its manipulation emerge.

2. 11β-HSD1 and the ageing brain

People exposed to chronically high glucocorticoids (endogenous or exogenous) are susceptible to cognitive, affective and, rarely, psychotic disorders (688). Individual differences in cognitive function with aging associate with stressful events or rising cortisol levels across a lifespan in rodents and humans (423, 458). If, however, glucocorticoid levels are kept low throughout life, either via neonatal “programming” or adult treatment with antidepressant drugs, both of which increase GR and MR in hippocampus and provide tighter HPA axis feedback control (793), or by midlife adrenalectomy with low-dose glucocorticoid replacement (382, 456), the subsequent emergence of cognitive deficits with age is prevented. Why the brain becomes sensitive to what is probably relatively subtle chronic glucocorticoid excess with ageing is a key question to be addressed, but measurement of plasma cortisone, the substrate for 11β-HSD1, may be informative in this respect, as elevated substrate availability coupled with increased levels of enzyme (see below) could turn a subtle circulating glucocorticoid excess into a marked excess within sensitive cells.

The identification of 11β-reductase activity in hippocampus and other cognitive circuitry in rodents (478, 607) and humans (613) seeded the notion that it might amplify glucocorticoid action and, with ageing, adversely impact on cognitive function and the underlying biology (622). In this regard, Hsd11b1^−/− mice (on two different genetic backgrounds) resist the expected high prevalence of age-associated cognitive decline measured in tests of hippocampus-associated spatial learning and memory such as the watermaze and Y-maze (791, 794). This is associated with reduced intrahippocampal glucocorticoid levels in the face of maintained or modestly elevated plasma glucocorticoid levels (791, 794). These data emphasize the importance of 11β-HSD1 in determining glucocorticoid action in the brain over and above any effect of circulating steroid levels. Importantly, short-term treatments with either nonselective (licorice derivatives) in young mice (152) or, crucially, with a selective 11β-HSD1 inhibitor (660) in already aged mice improves memory retention, suggesting that the action is on the adult brain rather than any notional developmental effect. These data endorse 11β-HSD1 inhibition as a potentially tractable approach to cognitive decline in aged humans. Indeed, treatment of elderly healthy men (aged 65–69 years) with carbenoxolone for 4 wk, plus amiloride to block the adverse effects of renal 11β-HSD2 inhibition, in a double-blind, placebo controlled, randomized study, led to cognitive improvements in verbal memory, a hippocampus-associated function (613). Similar effects were found in middle-aged subjects with type 2 diabetes. Moreover, a very short-term (1 h) effect of 11β-HSD1 inhibition on cognition in young rats has been reported (474); while the mechanisms are unclear, this raises again the intriguing possibility of rapid cognitive actions possibly mediated via membrane-associated corticosteroid receptors (337).

Cognitive alterations may be confounded by changes in mood, but, despite the adverse impacts of chronic glucocorticoid excess on affective function (123), 11β-HSD1 gene disruption does not noticeably alter anxiety-related behaviors (791, 794). Of course, gene disruption is life-long and affective behaviors are susceptible to developmental programming by glucocorticoids (623), potentially confounding these findings. Whether or not shorter-term 11β-HSD1 inhibition will alter mood, notably amygdala-associated fear and anxiety behaviors which are often sensitive to glucocorticoid manipulations (585), remains to be determined. While mood did not alter in healthy subjects and patients with type 2 diabetes treatment with carbenoxolone (613), it will be more pertinent to determine if 11β-HSD1 manipulations impact in individuals susceptible to depressive or anxiety-related illness.

3. Mechanisms of cognitive actions of 11β-HSD1

In terms of mechanism, improvement in peripheral metabolic factors, notably glucose homeostasis, might underlie the improvement in cognition (727). However, 11β-HSD1 deficiency did not alter glucose or lipid levels in lean, aged rodents, suggesting direct CNS effects (791). Indeed, long-term potentiation (LTP), a putative electrophysiological basis for synaptic “learning,” is increased in hippocampal slices from old 11β-HSD1-deficient mice compared with age-matched controls (791). Ectopic overexpression of 11β-HSD2 in the hippocampus improves learning, memory, and LTP in young animals (173), emphasizing the benefits of specifically reducing intrahippocampal glucocorticoid levels.

Neurogenesis is important in the hippocampus and its cognitive functions and is suppressed by glucocorticoids (110). However, while 11β-HSD1 deficiency increases hippocampal neurogenesis in young mice (as predicted), the negligible neurogenesis in old mice is unaltered by 11β-HSD1 deficiency (791).

The higher affinity of MR for glucocorticoids ensures it is largely occupied by physiological glucocorticoids even at the diurnal nadir, while GR are only activated as glucocorticoids rise to diurnal peak levels or during stress. The inverted-U-shaped relationship between glucocorticoid concentrations and both LTP in vitro and cognitive function in vivo (with both very low and high levels of steroids adversely impacting upon function) is underpinned by the procognitive effects of MR agonism and anticognitive effects of higher concentrations of glucocorticoids acting through GR (325). In line with this concept, intracerebroventricular infusion of the GR antagonist RU38486 (mifepristone) in old mice improves memory in the Y-maze, whereas the MR antagonist spironolactone has no effect (795). In contrast, despite similar age-associated elevation in plasma corticosterone, the maintained memory in elderly 11β-HSD1-deficient mice is attenuated by spironolactone, while RU38486 is without effect. These data suggest 11β-HSD1 deficiency sufficiently reduces intrahippocampal glucocorticoids to minimize GR activation by elevated plasma corticosterone with ageing, whilst maintaining optimal MR-mediated procognitive effects.

4. Does 11β-HSD1 cause age-related cognitive disorders?

A crucial issue beyond 11β-HSD1 as a therapeutic target in age-related cognitive dysfunction is whether or not it contributes to fundamental pathogenesis. Genetic linkage would suggest causation, and an intriguing report associates a rare (1:200) polymorphism in the promoter of the HSD11B1 gene with a sixfold increased risk of Alzheimer's disease (145), although the low frequency of the allele implies it is unlikely to make more than an occasional impact. More common variants of the gene do not associate with cognition in a healthy elderly population (148). However, in a prospective human study, an index of total body 11β-HSD1 activity “predicted” subsequent cerebral ventricular expansion (brain volume loss) and especially cognitive decline over 6 years (425). Whilst the locus of this effect is uncertain, the findings imply possible causal roles of elevated 11β-HSD1 in adverse cognitive and brain structural effects with ageing.

In the mouse, ageing associates with increased 11β-HSD1 mRNA in specific subfields of the hippocampus (CA3) and neocortex (layer V) (289) (FIGURE 11). Modelling this by transgenic overexpression of 11β-HSD1 in the forebrain causes cognitive impairments that only emerge with ageing (289). Such data suggest the enzyme may indeed be on the causal pathway, although the downstream processes are undetermined. Plausible mechanisms by which glucocorticoid excess might produce age-related cognitive decline are myriad and include changes in electrophysiological processes such as LTP (551), probably via changes in key target genes such as subunits of the NMDA and other glutamate receptors (509); second messenger systems such as intracellular calcium release (383); neuronal integrity and cytoskeleton; cellular energetics; cellular cholesterol dynamics and thus membrane function; amyloid precursor protein synthesis, processing, and degradation (247), the last notably via glucocorticoid regulation of insulin degrading enzyme expression (372); tau assembly and degradation (247); glial function, including microglia, the brain's phagocytes, and antigen presenting cells which express 11β-HSD1 that is increased in inflammation in human CNS (268); cerebral vasculature; and neurotransmitter synthesis, release, and breakdown (461, 645). Alternatively, the potential neuroprotective effects of 11β-HSD1 deficiency might reflect its contribution to other steroid or oxysterol metabolic pathways, 7-hydroxylated steroids/oxysterols being one obvious possibility (see below).

Figure 11.

11β-HSD1 in the ageing brain. A: 11β-HSD1 is elevated in the ageing (27 mo old) mouse brain, specifically in regions of the hippocampus and cortex underpinning learning and memory. B: with ageing there is considerable interindividual variation in 11β-HSD1 that correlates with cognitive function (seconds to find a hidden platform after training in the watermaze). C: modeling this by transgenic overexpression of 11β-HSD1 in the forebrain from the age of weaning causes cognitive impairments that only emerge with ageing. [Adapted from Holmes et al. (289).]

Whether or not 11β-HSD1 inhibition will impact upon pathological brain aging, for example, in Alzheimer's disease, remains uncertain, but some preliminary data suggest that there may even be some reduction in beta-amyloid lesions in animal models of Alzheimer's disease (728). This “disease-modifying” effect is unexpected and, if confirmed, then dissection of the mechanism will be of great importance. In this regard, exogenous glucocorticoids increase Aβ and tau deposition in mouse models of Alzheimer's disease (247), raising the intriguing possibility that the elevated cortisol levels seen in patients (688) may be more than merely a reflection of neuronal loss and the stress of ill health.

Overall, accepting that translation from preclinical to clinical circumstances is tortuous, 11β-HSD1 inhibition in the brain is emerging as a target in the treatment for age-related memory and other cognitive disorders with functional and perhaps even disease-modifying effects.

5. Oxysterols

As discussed above, 7-ketocholesterol is a substrate for 11β-HSD1 (582, 621). Oxysterols such as 7-ketocholesterol rise with excitotoxicity (158, 345), suggesting it may perhaps be neurotoxic, although it is unknown whether 7-keto and 7-hydroxy cholesterol differ. Additionally, 7-keto- and 7β-hydroxy derivatives of the neurosteroids dehydroepiandrosterone (DHEA) and pregnenolone can be metabolized by 11β-HSD1 (515). Intriguingly, DHEA and pregnenolone modified in the 7-position potentiate neurosteroid activity, producing cognitive enhancement with ageing (792). Any involvement of 11β-HSD1 in these reactions is unaddressed.

D. Central Regulation of Blood Pressure and Salt Appetite

While most MR in the CNS are not linked to 11β-HSD2 and thus bind glucocorticoids under physiological conditions (144), some central effects appear aldosterone-sensitive but corticosterone-insensitive (219). Intracerebroventricular (icv) administration of aldosterone causes arterial hypertension (220, 223, 242). Moreover, salt-sensitive hypertension is mitigated by intracerebroventricular administration of MR antagonists (236, 237). Analogous studies suggest similar aldosterone-specific effects on salt appetite (534). It has therefore been postulated that aldosterone-selective central MR modulate blood pressure and salt appetite. In support of this hypothesis, in the rat brain, MR are expressed in loci that regulate blood pressure and salt appetite, namely, the nucleus of the tractus solitarius (NTS) and circumventricular organs such as subfornical organ and hypothalamic regions (153, 320). However, intracerebroventricular corticosterone antagonizes aldosterone-mediated hypertension (242), implying either distinct MR, some “protected” from corticosterone elevating blood pressure, whereas other MR bind corticosterone and exert opposing actions, or GR (locus unclear) may counter MR effects.

Although 11β-HSD2 has very limited expression within the adult brain, in the rat this is congruent with MR expression in loci relevant to blood pressure and salt appetite control (583, 590). Furthermore, intracerebroventricular administration of 11β-HSD inhibitors increases blood pressure, although no effect has been observed on sodium appetite (240). In the adult mouse brain, 11β-HSD2 colocalization with MR is only evident in the NTS (222, 290). Whether or not this locus impacts on salt appetite has been unclear.

Addressing this issue, recent data show that mice lacking 11β-HSD2 solely in the brain (exploiting the Cre-lox system) exhibit a dramatic immediate drive to ingest proffered saline, even when the animals are sodium replete. This salt appetite is reversed by the MR antagonist spironolactone, showing unequivocally that 11β-HSD2-protected MR, presumably in the NTS, control salt appetite in mice (193). Indeed, even though mice lacking 11β-HSD2 in the brain ingest large quantities of sodium, they maintain normal salt balance by increasing natriuresis and polyuria (193).

Curiously, dietary sodium deprivation still causes activation of 11β-HSD2 neurons within the NTS in adrenalectomized animals (219), which suggests that these 11β-HSD2 positive neurons are activated by other factors or pathways in addition to adrenal aldosterone and other corticosteroids. Angiotensin II has emerged as one potential regulator (222). It is also possible that corticosteroids locally produced within the brain could be responsible. Indeed, 11β-hydroxylase and aldosterone synthase activity (and their encoding mRNAs, Cyp11b1 and Cyp11b2) have been found within the brain (799). In addition, 11β-HSD2 positive neurons within the NTS are innervated by the amygdala, PVN, dorsomedial NTS, and components of the vagus (222, 223, 629, 641), which could provide a neural rather than hormonal regulation of the system. It is likely that all of these inputs may act in conjunction with circulating aldosterone to modulate NTS control of sodium appetite. Intriguingly, the 11β-HSD2 neurons in the NTS innervate both relay nuclei within the brain stem that project to the forebrain as well as directly to the forebrain itself (224), areas associated with behavioral changes related to sodium appetite, reward, arousal, and mood (222). Indeed, salt appetite is gratified very rapidly by drinking saline, implicating a direct association of salt appetite with the reward pathways (358), a hypothesis supported by pharmacological modulation of the reward pathways (404) and MR antagonism (492).

The importance of this 11β-HSD2-mediated aldosterone-sensitive blood pressure and salt-appetite control circuitry in the brain merits detailed investigation, especially in the light of the importance of the CNS in pressor control including in human hypertension (691).

VII. FETO-PLACENTAL 11β-HSD AND DEVELOPMENTAL “PROGRAMMING”

In the early 1950s it was noted that human placenta contained considerable amounts of cortisone (70, 142). High 11β-HSD activity was subsequently found in placenta (539). This was suggested to catalyze conversion of maternal cortisol to cortisone before the active steroid could reach the umbilical vein (544). Indeed, in placental mammals, cortisol (corticosterone) levels in the maternal circulation are 5- to 10-fold higher than in the fetus (65, 138, 485), a gradient thought to be maintained by high placental 11β-HSD2 (90, 93, 368). Indeed, placental 11β-HSD2 inactivates the majority of maternal glucocorticoids passing to the fetus in rodents (132) and humans (66). Prior to mid-gestation, there is also widespread expression of 11β-HSD2 in fetal tissues in rodents and humans (93, 153, 670). Importantly, the fetus, unlike the adult, has little or no 11β-HSD1 in its tissues to regenerate cortisol from cortisone, at least until near term when 11β-HSD1 becomes expressed in organs, such as lung and liver, sensitive to prenatal maturation by glucocorticoids (662).

Glucocorticoids are potent regulators of fetal growth; high levels restrict fetal growth and prime tissues for maturation in anticipation of extrauterine life. Placental 11β-HSD2 is thus thought to form a “protective barrier” to the much higher glucocorticoid levels in the maternal circulation. A longer-term importance of this barrier function has more recently emerged.

A. Glucocorticoid Programming

During development, the environment can change a growing organism's structure and function. Such effects may exert persisting influences for the lifespan, a phenomenon called “developmental programming.” Programming stimuli can impact from before conception to adolescence, acting during critical developmental “windows” of sensitivity to alter cell number and/or cellular gene expression and thus tissue function (625). In humans, a plethora of recent epidemiological studies have associated lower birth weight, albeit in the normal range, with a substantial increase in the prevalence of cardiometabolic and neuropsychiatric disorders in later life (56). These data imply programming of disease risk by factors acting on growth before birth. Such developmental programming phenomena appear near ubiquitous in the animal kingdom, and the contemporary linkage with adult disease may mask the biological importance to optimize an offspring's chances to survive long enough to reproduce in adverse environments (the thrifty phenotype hypothesis) and hence its evolutionary conservation. It has been suggested that when there is a mismatch between prenatal “predictions” of the postnatal environment and its reality that the offspring is placed at a disadvantage, largely of disorders of the postreproductive age (233).

The two major etiological hypotheses of developmental programming invoke early life exposure to malnutrition or stress/glucocorticoids. Indeed, in many mammalian species, maternal undernutrition (total calorie or protein deficiency) (56) or exposure to stress/glucocorticoid excess (625) reduce birth weight and cause persisting hypertension, hyperglycemia, affective dysfunction, and neuroendocrine abnormalities in the offspring.

1. Role of 11β-HSD2 in developmental programming

A) ANIMAL MODELS OF PLACENTAL 11β-HSD2 IN PROGRAMMING.

Prenatal exposure to endogenous or exogenous glucocorticoids accelerates fetal organ maturation in a trade-off against overall growth. Physiological glucocorticoids are catabolized by fetal and placental 11β-HSD2 (FIGURE 12). From this emerged the hypothesis that variations in feto-placental 11β-HSD2 may underlie prenatal glucocorticoid programming (180). In rodents, 11β-HSD2 activity in the placenta correlates with pup birth weight (67), a phenomenon reproduced in most (673), but not all (587), human studies, suggesting that normal variation in 11β-HSD2 activity and the consequent degree of exposure of the fetus to maternal glucocorticoids impacts on fetal growth.

Figure 12.

Glucocorticoid programming and placental 11β-HSD2. Placental 11β-HSD2 debulks the much higher levels of active glucocorticoids in the maternal blood. A: normally the enzyme oxidizes cortisol to inert cortisone that cannot be regenerated in the fetus which lacks 11β-HSD1 until near term. Thus the major source of active cortisol in the fetus is its own adrenal glands. B: maternal treatment with dexamethasone, which is a poor substrate for 11β-HSD2 and thus passes the placenta intact, increases glucocorticoid action on the fetus and placenta reducing growth and altering the developmental trajectory of specific tissues. C: similarly inhibition or relative deficiency of placental 11β-HSD2 allows increased passage of active maternal glucocorticoids to the fetus and placental receptors. Lowering of placental 11β-HSD2 occurs with genetic mutations, consumption of licorice, maternal malnutrition, infection, or stress.

Beyond such observational studies, inhibition, genetic deficiency, or bypass (using poor substrate steroids such as dexamethasone or betamethasone) of placental 11β-HSD2 in model species and humans associates with reduced pregnancy duration, lower birth weight, and programmed outcomes in the offspring (67, 141, 288, 408, 498, 517, 522, 525, 651, 760, 761, 781, 782). Thus administration of dexamethasone or the 11β-HSD inhibitor carbenoxolone to pregnant rats reduces birth weight and programs hypertension, hyperglycemia, HPA axis hyperactivity, and anxiety behaviors in the offspring (67, 105). Similar glucocorticoid/stress programming effects occur in the guinea pig (335), sheep (159–161, 216, 323), and non-human primates (522), including singleton-bearing old world species with physiology and placental structure similar to humans (147, 560).

B) HUMAN DATA ON THE ROLE OF 11β-HSD2 IN PROGRAMMING.

Human with homozygous/compound heterozygous mutations in HSD11B2 have decreased birth weight compared with their siblings, who are mostly heterozygous (141, 348, 498). Similarly, 11β-HSD2 knockout mice have decreased weight at birth (288). Furthermore, Finnish women who voluntarily eat larger amounts of licorice-containing foodstuffs in pregnancy have shorter gestations and their children show poorer cognitive function and behavioral disturbances (increases in attention seeking, rule breaking, and aggression) coupled with hyperactivity of their HPA axis (569, 570). Licorice directly inhibits human placental 11β-HSD2 immediately ex vivo (66), increasing passage of cortisol to the fetal circulation. The relationships between maternal licorice intake and offspring cognitive function and behavior are continuous, suggesting no threshold is completely safe.

2. Mechanisms: epigenetics

The mechanisms of these programmed effects are tissue-specific and involve glucocorticoid-driven changes in target organ structure, gene expression, and function. Epigenetic mechanisms to maintain such effects have been advocated (167, 523, 524, 623, 755, 760, 761, 781, 782). Indeed, the key HPA axis drivers, corticotropin releasing hormone (CRH) and arginine vasopressin (AVP), and feedback receptor, glucocorticoid receptor (GR), show altered expression in models of programming and the promoters of all three encoding genes show persisting changes in methylation in response to early life environmental challenges (496, 500, 756). Methylation of DNA during gestation is in part determined by the availability of methyl donors (753), and placental 11β-HSD2 impacts on placental transport of key methyl donors (779), implying a possible additional role in determining fetal epigenetic status.

Intriguingly, 11β-HSD2 itself appears to be a target for fetal programming with persisting reductions in renal 11β-HSD2 mRNA levels and increased mineralocorticoid activity of glucocorticoids in rats exposed prenatally to glucocorticoids (694) and sheep exposed to undernutrition (769). The HSD11B2 gene promoter is subject to methylation (17). Children exposed to the horrors of the Nazi Holocaust have reduced renal 11β-HSD2 activity in old age with the greatest effect seen in those youngest at traumatization (797). These data suggest 11β-HSD2 is a programming target in humans, possibly to maximize renal sodium retention in what may be anticipated to be a starvation-associated sodium-poor environment. This adaptation does not necessarily optimize physiology for the modern sodium-replete world.

11β-HSD1 is also programmed. Prenatal glucocorticoids (522) and malnutrition (769) permanently increase the isozyme in adipose tissue, liver, pancreas (522), and hippocampus (642). As outlined above, this may contribute to the risk of metabolic and cognitive disorders with ageing.

B. 11β-HSDs in the Placenta

1. Localization and ontogeny of placental 11β-HSD2

Placental 11β-HSD2 has been reported in many mammalian species including human (90, 92, 503, 553), non-human primates (44, 45), pig (350), sheep (785, 787), guinea pig (608), rat (67, 104, 105, 588), and mouse (93, 127). Placental 11β-HSD2 in the rat (588, 734) and human (368, 682) is localized in the syncytiotrophoblast, precisely where maternal and fetal circulations are in close proximity for transfer of nutrients and other substances. Thus 11β-HSD2 is strategically positioned to function as a “barrier” to glucocorticoid access not only to placental cells but also to the fetus.

The ontogeny of placental 11β-HSD2 is species-specific. In humans, there is a steady rise in placental 11β-HSD2 activity through most of gestation (229, 455, 634). A similar gestational rise occurs in the baboon placenta (554). In humans, 11β-HSD2 activity (but not mRNA) decreases in the last 2 wk prior to parturition (504). Fetal membranes also express 11β-HSD. In the human amnion, 11β-HSD1 is concentrated in fibroblasts where expression is increased by glucocorticoids and cytokines (681), providing a feed-forward loop (506). Here, glucocorticoids increase prostaglandin E₂ production, which is central to the induction of parturition. 11β-HSD1 is also expressed in human chorion cells where glucocorticoids inhibit prostaglandin catabolism by repressing the gene encoding 15-hydroxy-PG dehydrogenase (546). Thus 11β-HSD1 amplification of glucocorticoid action reduces prostaglandin breakdown, further increasing the parturition signal. This fetal membrane system affords a possible link between maternal infection, cytokine production, local glucocorticoid production, and the induction of preterm labor. Moreover, increased amniotic fluid cortisol levels amplified by local 11β-HSD1 will be swallowed by the fetus and thus may contribute to accelerated fetal organ maturation at labor (681).

C. 11β-HSDs in Fetal Tissues

The 11β-HSDs are not only expressed in placenta and fetal membranes but also in the fetus itself where they shows complex stage- and organ-specific patterns implying exquisite control of cellular exposure to glucocorticoids as each tissue matures (93, 662). Most work on this topic addresses the developing brain.

1. 11β-HSDs in the fetal brain

A) 11β-HSD2.

The developing brain is extremely sensitive to glucocorticoids, which regulate cellular and biochemical development (360, 457) by initiating terminal maturation and remodeling of axons and dendrites, as well as inducing programmed cell death (463). The high expression of 11β-HSD2 in the fetal brain may be crucial to protect immature neurons from maturation through untimely exposure to fetal and/or maternal physiological glucocorticoids. A reduction in brain weight is observed in sheep at birth following antenatal glucocorticoid treatment (301), associated with delayed maturation of neurons, glia, and vasculature (302). In rhesus monkeys, antenatal glucocorticoid exposure causes degeneration of hippocampal neurons and reduces subsequent hippocampal volume (721). The hippocampus is particularly sensitive to perinatal glucocorticoid exposure, which can cause life-long changes in memory and behavior (85, 636, 666). In the rat, prenatal stress (repeated maternal restraint) reduces the number of proliferating hippocampal cells and feminizes the adult hippocampus (428). Prenatal stress similarly reduces hippocampal neurogenesis, with reduced placental 11β-HSD2 implicated in the mechanism (422).

GR, MR, and 11β-HSDs show intricate temporal and regional patterns in the developing brain (153, 211, 349). GR is highly expressed in the neuroepithelium of the fetal rat brain, whereas MR is restricted the epithelium of the septal-hippocampal system, areas of the anterior hypothalamus, pituitary, deep layers of the superior colliculus, piriform cortex, and lateral septum (153). Interestingly, neuronal 11β-HSD2 does not coincide with MR in the fetal brain but with GR, suggesting the enzyme protects immature neurons from premature exposure to glucocorticoids via GR. Throughout midgestation, 11β-HSD2 is abundant in neuroepithelium, but its expression rapidly declines as each brain area approaches the terminal stage of neurogenesis (93, 153). Perhaps analogously, 11β-HSD2 expression in the human fetal brain is silenced between gestational weeks 19–26 (669). This postulated protective role for 11β-HSD2 in fetal brain is akin to the proposed role of placental 11β-HSD2 in protecting the fetus from overexposure to maternal glucocorticoids (153, 625, 734) and may provide an additional barrier to protect the developing CNS from excess glucocorticoids derived from the mother or the fetal adrenal itself.

11β-HSD2 expression is restricted after birth to the proliferating external granular layer (EGL) of the cerebellum and several central, late-maturing nuclei of the thalamus (583, 589). Glucocorticoids and stress are known to inhibit the production of cells from the EGL resulting in decreased size and cell content of the molecular and/or granular layers of the mature cerebellum, depending on the timing of exposure (414, 469, 772). As anticipated, cerebellar size is also reduced in Hsd11b2^−/− mice early in postnatal life with a decrease in the molecular and internal granule layers (290), associated with a delay in attainment of cerebellum-associated neurodevelopmental landmarks such as negative geotaxis and eye opening (290). Hence, the temporal and spatial expression of 11β-HSD2 in the developing brain tightly regulates the timing of its exposure to glucocorticoids which, in turn, dictate the timing of terminal maturation.

But does this relate to fetal programming of persisting effects? To address this, mice with brain-specific deletion of 11β-HSD2 have been made. These exhibit depressive-like behavior as adults, even though placental and fetal growth and circulating corticosterone levels are normal (Wyrwoll, Seckl, and Holmes, unpublished data). Thus 11β-HSD2 in the developing brain appears crucial to protect against glucocorticoid programming of behavior, even when there is normal placental expression of 11β-HSD2.

11β-HSD2 is significantly induced in cerebellar granular neuron precursor cells by the developmental patterning gene sonic hedgehog (269). Decreased 11β-HSD2 expression in frontal cortex and hippocampus of piglets follows early weaning or social isolation (559). In rodents, 11β-HSD2 is rapidly downregulated in the fetal brain at midgestation (94), perhaps due to epigenetic silencing of the promoter. The notion that bypass or mistiming of the shut-off of fetal brain 11β-HSD2 engenders developmental abnormalities or programming merits investigation.

B) 11β-HSD1.

In the hippocampus of the fetal sheep, 11β-HSD1 mRNA rises from mid to late gestation but rapidly falls prior to parturition (644). In the rat (153, 476) and mouse (662) fetal brain, 11β-HSD1 mRNA is not observed until the last third of gestation, a time when 11β-HSD2 is declining. Thereafter, 11β-HSD1 increases in the fetal brain with age. Expression is first seen in the hippocampus, precerebellar area, and medulla at E16.5. At term, highest expression is observed in the thalamus, neocortex, hippocampus, and hindbrain, and levels continue to rise through the postnatal period (153, 476). No neurodevelopmental abnormalities or related functional phenotypes have been associated with 11β-HSD1 deletion. However, the observation that 11β-HSD1-deficient mice are heavier at birth (780) suggests that these might have been overlooked.

2. Fetal kidney

The developing mammalian kidney highly expresses 11β-HSD2. In early to mid gestation, 11β-HSD2 is found throughout the kidney in the mouse (93), sheep (385), and human (285, 673), notably in Bowman's capsule and the vascular tufts of the developing glomeruli as they migrate from the surface to the inner cortex (126). In late gestation, the expression pattern becomes more discrete with very high levels in the distal convoluted tubules and collecting ducts (93, 126, 127, 307), resembling the adult pattern. Indeed, in the later stages of gestation, MR becomes colocalized with 11β-HSD2, suggesting 11β-HSD2 is protecting inappropriate activation of MR at this time, whereas at earlier time points 11β-HSD2 may modulate glucocorticoid signaling through GR (93, 126). Interestingly, 11β-HSD2 levels are low at birth, despite maintained MR expression (432). Plausibly, this could be a mechanism for glucocorticoid maturation of renal structure and function, although this remains to be tested. Conversely, high glucocorticoid levels in the perinatal period cause renal cysts (450).

Fetal kidney 11β-HSD2 is itself a target of programming. In the ovine fetus, renal 11β-HSD2 is strikingly downregulated by hypoxia (38, 501, 786) or maternal undernutrition (769). In rat offspring, maternal undernutrition, diabetes, or placental vasculature insufficiency (uterine artery ligation) all reduce 11β-HSD2 in the kidney (58, 207, 540). Such changes could, if persistent, contribute to the programmed hypertension seen in these models. Indeed, maternal glucocorticoid treatment in pregnancy in the rat reduces offspring renal 11β-HSD2 that persists from neonatal to mid-adult life and associates with increased mineralocorticoid actions of corticosterone compatible with mild AME (694), in essence a form of low renin hypertension. In terms of molecular mechanisms, intrauterine growth retardation exerts sex-specific effects on transcription factor binding to the Hsd11b2 gene promoter in the kidneys of offspring (58). These changes are linked to increased CpG methylation of the Hsd11b2 promoter compatible with reduced transcription of 11β-HSD2 mRNA.

Such effects may be hardwired. Thus sodium retention in 11β-HSD2 knockout mice can be readily reversed by suppression of corticosterone with dexamethasone, but hypertension and renal structural abnormalities persist (363), emphasizing the key importance of 11β-HSD2 in renal development.

D. Feto-placental 11β-HSD Regulation and Function

1. Placental functions of 11β-HSD2

A core issue in fetal programming is to determine the locus of action of environmental challenges, specifically distinguishing indirect effects on the mother from more direct impacts on the placenta and/or the fetus. Progress on this issue has been achieved by crossing male and female mice heterozygous for a null allele of Hsd11b2. This gives wild-type, heterozygous, and homozygous null Hsd11b2 offspring in the same dam. Since placental 11β-HSD2 in the labyrinth zone is derived from the fetal genotype, this design affords a clear differentiation of feto-placental from maternal effects. In this model, birth weight and offspring affective behavior follow the feto-placental genotype (288), excluding maternal effects from this manipulation of 11β-HSD2. Of course, the Hsd11b2^−/− offspring also lack the enzyme in kidney and are thus hypertensive and hypokalemic (363), which complicates assessment of cardiometabolic programming. Studies of tissue-specific knockout of 11β-HSD2 in fetal organs will allow distinction of the relative importance of the placental and individual fetal tissue “barriers” to glucocorticoids.

An additional and important level of impact of 11β-HSD2 is on placental function per se. In the heterozygous cross model, growth of placentas completely lacking 11β-HSD2 is restricted in late gestation, even before fetal growth is retarded (783). In mid- to late-gestation, placental amino acid transport to Hsd11b2^−/− offspring is increased, suggesting compensation for placental dysfunction. However, near term the Hsd11b2^−/− placenta is failing with markedly reduced placental glucose transport and decreased glucose transporter 3 expression. The late-gestation Hsd11b2^−/− placenta lacks the normal increase in density of fetally derived blood vessels and shows reduced expression of the key angiogenic factor VEGF (normally downregulated by glucocorticoids; Ref. 322), and a key angiogenic transcription factor, PPARγ. Glucocorticoids and 11β-HSD inhibitors advance apoptosis in the term placenta (737). Such underweight, undervascularized, and underfunctioning placentas are poorly equipped to supply the rapidly growing and maturing near-term fetus. In terms of molecular mechanisms, placenta expresses both GR and MR. Some data suggest that mineralocorticoids acting via MR promote trophoblast division and placental growth in vitro and in vivo, while glucocorticoid inhibit these effects (227). Moreover, glucocorticoids inhibit 15-hydroxyprostaglandin dehydrogenase, a key degradative enzyme. In human trophoblast cells which express 11β-HSD2, inhibition is greater with cortisol plus carbenoxolone than the GR-selective agonist dexamethasone, implying MR are involved in this control (546). The impact of the placental balance of 11β-HSD2 and 11β-HSD1 on local GR and MR remains unclear, although glycyrrhetinic acid (inhibiting both isozymes) enhances glucocorticoid-mediated placental cell hypoplasia in vitro (227), suggesting 11β-HSD2 predominates.

2. Pathogenesis and placental 11β-HSD2

Placental 11β-HSD2 is modified in disease. Activity is reduced in human pregnancies complicated by preeclampsia (42, 448), in association with increased cortisol in cord blood, suggesting increased fetal exposure to maternal (and/or fetal) glucocorticoids. Similarly, placental 11β-HSD2 is decreased in fetuses with intrauterine growth retardation (176, 455, 619, 634, 718, 733), perhaps a mechanism to increase fetal glucocorticoid exposure to optimize maturation in preparation for birth, trading off against growth (334). Low placental 11β-HSD2 in IUGR correlates inversely with catch-up growth (718), a marker of programming risk (187). Maternal inflammation (asthma) associates with reduced placental 11β-HSD2 and birth weight and increased fetal cortisol levels (505). Indeed proinflammatory cytokines, notably IL-1β, IL-6, and TNF-α, downregulate 11β-HSD2 activity and mRNA levels in primary cultures of human trophoblasts (118, 361).

Other regulators of placental 11β-HSD2 include nitric oxide, progesterone, estrogen, protein kinase A, retinoic acid, prostaglandins, catecholamines, oxygen, glucocorticoids, PPAR agonists, proinflammatory cytokines, and toxic heavy metals (16, 261, 262, 332, 545, 555, 616, 683, 684, 713, 726, 784). Furthermore, 11β-HSD2 expression is upregulated by the action of p38 MAPK on 11β-HSD2 mRNA stability in primary trophoblast cells (635). Thus a host of factors, many determined by the maternal environment and health, impact on placental 11β-HSD2, which may represent a key node between the maternal and feto-placental environments.

3. Is placental 11β-HSD2 the common link between glucocorticoid and nutritional programming?

Interestingly, low-protein diet in pregnancy causes an increase in maternal and fetal glucocorticoid levels (255, 398), together with a decrease in placental 11β-HSD2 activity (77, 384, 676). Conversely, exposure to dexamethasone during pregnancy decreases maternal food intake (376) and reduces maternal weight gain (525). These results suggest that there are common mechanisms underpinning both nutritional and glucocorticoid developmental programming. Feeding low-protein diet to the pregnant dams of heterozygous Hsd11b2 crosses shows that although both maternal undernutrition and 11β-HSD2 deficiency reduce fetal growth and both involve feto-placental overexposure to glucocorticoids, the mechanisms are distinct. Thus 11β-HSD2 insufficiency increases feto-placental exposure to maternal glucocorticoids, whereas low-protein diet activates the fetal HPA axis to produce increased fetal glucocorticoid levels (132). Consistent with this, prenatal undernutrition increases levels of adrenal steroidogenic enzymes at birth (344). With the obesity epidemic producing more babies born to obese mothers, it is a concern that that there is also adverse programming of cardiometabolic outcomes in these offspring, but at present, any role for 11β-HSD2 is unknown (166, 610).

VIII. CONCLUSIONS AND FUTURE PERSPECTIVES

In the six decades since the discovery of 11β-HSD, only in the last 25 years has interest burgeoned and understanding emerged. Summarizing the current state of play, there are two isozymes. 11β-HSD2 is a potent dehydrogenase that acts to “protect” MR from activation by glucocorticoids, thus engendering aldosterone specificity in the distal nephron and other classical mineralocorticoid target sites in vivo. Deficiency or inhibition in the kidney causes AME. This enzyme also protects GR and MR in the placenta and fetal tissues from premature activation by glucocorticoids. Deficiency or inhibition of feto-placental 11β-HSD2 (or its bypass with dexamethasone) causes developmental programming, increasing the risk of subsequent cardiometabolic and neuropsychiatric disorders. The biology is conserved from rodents to humans. In contrast, 11β-HSD1 is predominantly but not exclusively an 11β-reductase in intact cells, regenerating and thus amplifying glucocorticoid action. Enzyme levels are elevated in adipose tissue in obesity and appear to play a causal role in metabolic syndrome. Similarly, elevated 11β-HSD1 in the ageing CNS contributes to cognitive decline. 11β-HSD1 inhibition is an interesting drug target for metabolic syndrome and its complications, age-related cognitive disorders, and other less well validated indications, typically diseases associated with age-related physical decline.

A. Ageing

11β-HSD1 expression appears elevated in key glucocorticoid target organs with ageing or age-associated pathologies. Thus the aged hippocampus and neocortex overexpress 11β-HSD1 in mice (289). Similarly old bones in mice (759) and osteoblasts obtained from aged humans (130) have higher 11β-HSD1 levels. 11β-HSD1 is elevated in adipose tissue of obese individuals. Perhaps obesity mimics an age-related process in adipose tissues as it associates with hypoxia, fibrosis, and degenerative metabolic disease. Similarly, 11β-HSD1 is increased in skeletal muscle in type 2 diabetes (1), again a pathology associated with accelerated senescence of multiple organs. 11β-HSD1 and corticosterone levels are increased in the ageing obese mouse ovary (83), another organ that fails with senescence.

With aging, plasma glucocorticoids become elevated in a subset of individuals in rodents and humans. This coincides with and marks “unsuccessful” aging and its associated pathologies. Glucocorticoids may be causal, and attempts to keep circulating glucocorticoids low or reduce their tissue impacts can ameliorate age-related pathologies. Such findings have spawned the glucocorticoid hypothesis of age-related pathogenesis in the brain (614) as well as immune (63) and cardiovascular (595) systems.

The increase in expression of 11β-HSD1 in key tissues with aging and its amplification of glucocorticoid concentrations in organs sensitive to glucocorticoid pathogenesis suggests a component acting well beyond its physiological purpose. Perhaps the glucocorticoid-11β-HSD1 system is intended to configure medium-term tissue glucocorticoid responses, amplifying steroid action in some targets (after all 11β-HSD1 expression is generally increased by glucocorticoids) but not others, optimally to configure tissue effects to promote survival (increasing glucocorticoid-driven storage of excess calories in adipose tissue in preparation for the next period of starvation, facilitating fear conditioning to survive the inevitable succeeding mortal challenge, preventing overshoot of inflammatory responses, etc.). A lifetime of such stress hormone amplification and its trade-off consequences (reduced hippocampus-associated learning, insulin resistance, expanded and inflamed adipose tissue), beyond the age of peak reproductive fitness and thus Darwinian selection, or in the exceptional circumstances of chronic calorie excess beyond the range biology can be expected readily to handle, puts such subtle tuning of tissue steroid exposure beyond its purpose.

The mechanisms by which 11β-HSD1 becomes elevated in specific cells in the aging brain, bone, ovary, immune system, and perhaps adipose tissue is not fully understood, but proinflammatory cytokines and glucocorticoids themselves may be key. Indeed, proinflammatory cytokines also stimulate the HPA axis. Thus inflammation may drive both systemic (HPA axis) and local (11β-HSD1) glucocorticoid excess which, with chronicity and age-related vulnerability, mediates organ pathology. Whatever the mechanism and its biological importance, 11β-HSD1 and its elevation in specific tissues in association with pathogenesis offers an opportunity for therapy.

B. 11β-HSD1 as a Therapeutic Target

Inhibition of 11β-HSD1 is a plausible target for treating metabolic disease, atherosclerosis, and disorders of cognition with aging. The efficacy of this approach has been proven in rodents and other preclinical models of obesity-metabolic disease and age-related memory loss. The effects of selective 11β-HSD1 inhibitors in humans with diabetes and metabolic disease have been modest to date, at least for the chosen therapeutic target of lowering markers of glycemic control or blood pressure, and the utility of such inhibitors in clinical therapy remains to be established. Perhaps the main indication might be in early metabolic disease to ameliorate multiple risk factors, a contention indicated by the clinical trials published to date. Crucial is the effect on atherosclerosis, the main final cause of mortality in the obesity-diabetes-metabolic syndrome continuum. Whilst preclinical studies are rather encouraging, such end-points are long term and thus expensive for clinical trials and require a degree of confidence in the target to persuade funding.

Trials in other indications such as hepatic steatosis, wound healing, mild cognitive impairment, Alzheimer's disease, other cognitive disorders, glaucoma, and thyroid eye disease are being explored but are unreported as yet. In the era of stratified medicine, subgroups of subjects with elevated 11β-HSD1 in key target organs may represent the main beneficiaries. Indeed, the greatest glucose- and lipid-lowering actions in one phase II trial were seen in subjects with the highest plasma glucose and lipid levels (594).

However, the efficacy and indeed even the most promising therapeutic target of 11β-HSD1 inhibition are not yet clear. The preclinical data suggest that adipose tissue is the most important metabolic organ target, yet some agents access this poorly, as exemplified by carbenoxolone. A subtlety is that an inhibitor should probably preferentially inhibit 11β-reductase without impact on the reverse 11β-dehydrogenase direction. Such considerations militate for direction-specific screening assays in intact cells and tissues.

C. Side Effects of 11β-HSD1 Inhibition

An inevitable effect of 11β-HSDS1 inhibition is activation of the HPA axis merely to maintain circulating cortisol levels against a reduced glucocorticoid half-life because of the lack of regeneration in splanchnic tissues, etc. The elevation of ACTH drives cortisol to normal levels, but elevates DHEA and other adrenal androgens potentially causing hirsutism in sensitive women. This is an unwanted, but not usually a serious, side effect and may be readily reversible as ACTH and DHEA levels fall to normal levels on cessation of 11β-HSD1 inhibition (594). However, adrenal hypertrophy/hyperplasia are also likely. While this is a physiological response and is observed in other causes of HPA axis activation such as depression, there is a theoretical concern about persistence and even the possibility of tertiary tumor formation. No data support this eventuality which is not seen in other HPA activation states. Hirsutism is the main adult phenotype of cortisone reductase deficiency, also supporting the general safety of this approach.

Other adverse effects include delayed maturation of the fetal lung at term, and such agents should probably be avoided in pregnancy. Effects on the immune/inflammatory systems are complex, but it will be important to know the impact on clearance of acute and chronic infections and inflammatory states such as rheumatoid arthritis and connective tissue disorders. An important caveat to the inflammation studies in preclinical models is that most studies have been conducted in young, lean mice. Obesity, type 2 diabetes, and metabolic syndrome associate with dysregulated immune responses (297) and impaired resistance to bacterial infection, with increased mortality due to sepsis (20, 723). It will be important to dissect the complex relationships between obesity-metabolic syndrome, 11β-HSD1, and the immune system in human trials. To date, no excess of side effects over placebo have been reported in short-term (months) clinical trials with selective 11β-HSD1 inhibitors. This is somewhat reassuring.

D. 11β-HSD2 as a Biomarker and Therapeutic Target

11β-HSD2 deficiency is pathogenic, but the isozyme may be a biomarker for developmental programming of the fetus, both in placental biopsies and in leukocytes in cord blood and in the adult. Epigenetic changes in this promoter, even in leukocytes, may be a persisting reflection of the quality of the prenatal environment marking those at risk of sequelae.

There is much about these enzymes still to be understood from fundamental biology, for example, simply understanding the major physiological roles of 11β-HSD1, its reaction direction in all tissues, the balance of intracrine and endocrine impacts, and its true value as a therapeutic target. For 11β-HSD2, it remains unclear exactly how biochemically it generates MR selectivity, its importance in developmental programming beyond licorice eaters, and its role as a biomarker of this process in placenta or offspring tissues. Much has been achieved since the biology of 11β-HSD2 was revealed 25 years ago; the next 25 years should see equal if not greater progress.

GRANTS

Work in the authors' laboratories is funded by a Wellcome Trust Programme Grant (to J. Seckl and K. Chapman), Wellcome Trust project grants (to K. Chapman, J. Seckl, and M. Holmes), as well as project grants from the MRC (to K. Chapman, J. Seckl, and M. Holmes), the British Heart Foundation (to K. Chapman and J. Seckl), and the European Union (to J. Seckl).

DISCLOSURES

J. Seckl holds patents on the use of inhibitors of 11β-HSD1.