Brain mineralocorticoid receptors in cognition and cardiovascular homeostasis

Abstract

Mineralocorticoid receptors (MR) mediate diverse functions supporting osmotic and hemodynamic homeostasis, response to injury and inflammation, and neuronal changes required for learning and memory. Inappropriate MR activation in kidneys, heart, vessels, and brain hemodynamic control centers results in cardiovascular and renal pathology and hypertension. MR binds aldosterone, cortisol and corticosterone with similar affinity, while the glucocorticoid receptor (GR) has less affinity for cortisol and corticosterone. As glucocorticoids are more abundant than aldosterone, aldosterone activates MR in cells co-expressing enzymes with 11β-hydroxydehydrogenase activity to inactivate them. MR and GR co-expressed in the same cell interact at the molecular and functional level and these functions may be complementary or opposing depending on the cell type. Thus the balance between MR and GR expression and activation is crucial for normal function. Where 11β-hydroxydehydrogenase 2 (11β-HSD2) that inactivates cortisol and corticosterone in aldosterone target cells of the kidney and nucleus tractus solitarius (NTS) is not expressed, as in most neurons, MR are activated at basal glucocorticoid concentrations, GR at stress concentrations. An exception may be pre-autonomic neurons of the PVN which express MR and 11β-HSD1 in the absence of hexose-6-phosphate dehydrogenase required to generate the requisite cofactor for reductase activity, thus acts as a dehydrogenase. MR antagonists, valuable adjuncts to the treatment of cardiovascular disease, also inhibit MR in the brain that are crucial for memory formation and exacerbate detrimental effects of excessive GR activation on cognition and mood. 11β-HSD1 inhibitors combat metabolic and cognitive diseases related to glucocorticoid excess, but may exacerbate MR action where 11β-HSD1 acts as a dehydrogenase, while non-selective 11β-HSD1&2 inhibitors cause injurious disruption of MR hemodynamic control. MR functions in the brain are multifaceted and optimal MR:GR activity is crucial. Therefore selectively targeting down-stream effectors of MR specific actions may be a better therapeutic goal.

Introduction

Because of their complexity, understanding the functions of the mineralocorticoid receptor (MR) lagged that of the other receptors. Like other steroid hormone receptors, MR are ligand-activated nuclear transcription factors for protein effectors of MR action that also initiate rapid non-genomic effects through cell signaling pathways presumably through MR associated with the plasma membrane. A full discussion of the rapid non-transcriptional signaling by the MR is not attempted herein[1–9]. MR and glucocorticoid receptors (GR) are expressed in many tissues types, often in the same cells, where they interact at molecular and functional levels, at times in synergy, others in opposition. Ligand activated MR and GR in the cytosol enter the cell nucleus, form dimers that bind to hormone response elements on the DNA as a complex with co-transcription factors[10–13]. These transcription factors can be repressors or activators and are cell-type and context specific. MR and GR form homodimers and heterodimers with different transcriptional efficacy depending on the cell type both in vitro and in vivo[14–18], thus the appropriate balance of MR and GR activation is crucial for homeostasis and a discussion of one is often not adequate without consideration of the other, especially in the CNS. This is especially important in analyzing the function of neurons expressing both MR and GR where MR activation tends to be excitatory and GR mediated events tend to mitigate excitation[19–21].

Also unlike the other steroid receptors with relatively narrow ligand specificities, the MR has similar intrinsic affinity for aldosterone, the major glucocorticoids cortisol and corticosterone, deoxycorticosterone, and progesterone. Progesterone is a competitive MR antagonist that is inactivated by several enzymes in aldosterone target renal epithelial cells [22], but attains sufficient concentrations during pregnancy and the luteal phase of the estrus cycle in species with distinct luteal and follicular phases to inhibit MR, leading to renin-angiotensin-aldosterone system (RAAS) activation and doubling of aldosterone levels in women during the luteal phase[23–26]. The presence of MR in ovaries, testes, uterus and placenta[27–29] has led to the proposal that progesterone binding to the MR may serve a reproductive function[30]. Despite higher levels of aldosterone, premenopausal women have a reduced risk of cardiovascular disease due to as yet unclear interactions with the estrogen receptor and estrogen mediated events, in part through yet incompletely-defined, MR-independent effects of aldosterone[6, 31–35].

Concentrations of endogenous glucocorticoids exceed those of aldosterone by 2–3 orders of magnitude, thus they are the physiological ligand for MR as well as for GR, except in aldosterone target cells in which pre-receptor inactivation of glucocorticoids by enzymes with 11β-steroid dehydrogenase (11β-HSD) activity occurs. Description of the 11β-steroid dehydrogenase enzymes, 11β-HSD types 1 & 2 will be addressed below. As a gross generalization, activation of MR in the higher centers of the brain by glucocorticoids is crucial for neuronal plasticity, learning and memory, while activation of MR in the hypothalamus and brain stem modulates electrolyte and hemodynamic homeostasis[8, 9]. At a time in which the use of MR antagonists is being resurrected to protect from the ravages of heart failure and the development and therapeutic use of MR agonists and 11β–HSD1 antagonists are being considered to stop the progression of cardiometabolic disorders and age-related cognitive decline, there is a pressing need to widen the focus of studies of MR activities in the CNS.

Mineralocorticoid receptors function is crucial in epithelial and non-epithelial cells

Several decades after the adrenal gland was reported to be critically important for life by Thomas Addison in the mid-1800s, it was recognized that the adrenal cortex produced different factors classified by their functions. Mineralocorticoids increased vectorial transport of Na⁺, K⁺ and H⁺, followed by water, across epithelia of the nephron, colon and amphibian urinary bladder; glucocorticoids had gluconeogenesis/glycogen depletion and anti-inflammatory properties [36–43]. The first mineralocorticoid to be isolated, deoxycorticosterone (deoxycorticosterone; DOC)[36] was life-saving for patients in adrenal failure, however excessive amounts were soon found to cause hypertension and pathological remodeling of the heart and kidneys [44, 45] (reviewed in [8]. The selective effect of aldosterone on fluid and electrolyte homeostasis is reflected in its first name upon its isolation, ‘electrocortin’[46, 47]. Subsequent studies over the last 50 years have demonstrated that MR regulates fluid and electrolyte homeostasis by multiple complex events far beyond the scope of this review [48, 49]. Three years later, Jerome Conn described Primary Aldosteronism in patients with persistent hypertension and hypokalemia [50]. Primary Aldosteronism is now recognized as a major cause of secondary hypertension associated with more severe cardiovascular and renal complications than essential hypertension of similar severity and duration[51–55]. Primary aldosteronism is also associated with depression and cognitive decline[56–59].

Mineralocorticoids exert extra-renal effects on blood pressure in addition to increasing Na⁺ and water retention by kidneys[44, 47], as discovered in early studies showing that they increased vascular tone directly by increasing ion, particularly calcium, within vascular smooth muscle cells, effects that occurred before an increase in blood pressure was evident, as well as indirectly through activation of the sympathetic nervous system (SNS)[60–66]. The massive literature addressing MR-mediated regulation of ion channel function of vascular endothelial and smooth muscle cells, as well as vessel remodeling is beyond the scope of this review [5, 67–72]. Similarly, MR-mediated changes in ion transport in neurons increases their excitability[19–21]. Inappropriate MR activation in cerebral vessels impacts neurological function, thus must be considered in discussions of the clinical effects of altered MR actions in the brain[73, 74].

The role of the brain in mineralocorticoid hypertension was first shown by ablation studies. The paraventricular nucleus (PVN) of the hypothalamus, circumventricular organs, brain stem, central sympathetic neurons, and interconnecting tracts were shown to also contribute to or be essential for mineralocorticoid-salt excess and renovascular hypertension [75–79], reviewed in[8]. The PVN comprises many types of neurons that receive and integrate information from diverse areas of the brain, including neuroendocrine neurons that secrete peptides that control pituitary function, pre-sympathetic neurons projecting to the medulla and spinal cord, parasympathetic neurons projecting to the dorsal motor nucleus of the vagus, and interneurons[80–82]. The r o le of brain MR in mineralocorticoid-salt hypertension was confirmed by the lateral intracerebroventricular (icv) infusion of MR agonists and antagonists at doses too small to cause an effect when diluted by the blood volume upon peripheral infusion[83–87], reviewed in [8]. Similar studies in several labs confirmed that MR-regulated increases in ion channel activity previously documented in the kidney and vessels were also instrumental in centrally mediated mineralocorticoid-salt hypertension, [88–93] and that activation of the SNS was a primary effector of mineralocorticoid hypertension [85, 94, 95]. The selectivity of the MR antagonists in these studies was confirmed by the demonstration that the intracerebroventricular infusion of siRNA for MR abrogated the increased SNS activation and hypertension induced by either aldosterone-salt or AngII-salt excess [96]. More recently MR were visualized within pre-sympathetic neurons of the rat PVN projecting to the spinal cord by a combination of retrograde tracer and fluoroimmunohistochemistry[82]. Chronically increasing or decreasing endogenous plasma aldosterone concentrations with a low or high sodium diet, respectively, significantly increased total MR expression and translocation of MR from the cytoplasm to the nuclei of pre-sympathetic neurons in rats adapted to the low, compared to high salt diet[82]. This is consistent with the activation of the SNS to support the circulation during sodium, thus water and volume, depletion [97, 98]. This adaptive mechanism for terrestrial animals, most of which do not have ready access to large amounts of sodium, is quite different from the experimental model of mineralocorticoid-salt excess hypertension in which mineralocorticoids are elevated in the context of a high sodium intake.

Aldosterone increases sodium appetite, also an adaptive mechanism shaped by evolution, that is centrally regulated[99–101]. Inappropriate mineralocorticoid excess increases sodium appetite even when in the sodium replete state[102, 103]. While MR that mediate the increases in blood pressure appear to reside primarily in the hypothalamus and circumventricular organs of the third ventricle, those initiating sodium appetite are primarily in the medulla and amygdala [99, 104–107]. Small concentrations and volumes of aldosterone infused into the lateral cerebroventricles produce hypertension without altering sodium appetite[108], similar infusions into the fourth ventricle increase sodium appetite with no effect on the blood pressure[109], presumably by activating aldosterone sensitive neurons of the NTS[107, 110]. The capacity for common laboratory animals to regulation their sodium intake has often been underappreciated and over-ridden by the excessive amounts of sodium in standard maintenance chows. The voluntary sodium intake between Sprague-Dawley rats weaned to a sodium deficient diet and an al libitum choice between saline and water to drink was shown to be about 15% of the sodium consumed by cohorts provided a commonly used rodent chow containing 0.5% Na. At 10 weeks of age body weights were no different between groups, however blood pressures were significantly higher in those receiving the standard chow than those consuming sodium by free choice[111].

Pathological cardiac remodeling produced by inappropriate systemic MR activation is not entirely dependent upon hypertension as demonstrated by a combination of systemic and intracerebroventricular (icv) infusions of MR agonists and antagonists in rats, emphasizing the importance of local MR action in peripheral non-epithelial tissues [84, 112]. The extensive literature on the mechanisms of injury mediated by MR in the heart, vessels, and kidneys led to the first clinical trials, RALES and EPHESUS, that demonstrated that addition of mineralocorticoid antagonists to standard therapy for chronic heart failure at low doses that do not further lower the blood pressure, significantly benefited cardiovascular function, as well as prolonged and increased the quality of life[113, 114]. Benefits of MR antagonists occurred even in patients who did not have elevated aldosterone levels. Successful outcomes of similar subsequent trials have led to increased use of MR antagonists in treatment regimens for renal and cardiac failure, as well as milder dysfunction [113, 115–118].

Inappropriate MR activation leads to inflammation that precedes the cell death and fibrosis in the heart, vessels and kidneys in several experimental models including the L-NAME, Angiotensin II-salt excess, mineralocorticoid-salt excess, and several genetic models of hypertension [119–126] and involves classic mediators of inflammation, including NADPH oxidase, ROS and inflammatory cytokines[73, 127–130]. Activation of NADPH oxidase and production of ROS are among the rapid non-transcriptional actions of MR critical for its rapid signaling in the heart[131] and in neurons[132–135] required for normal function, but that become pathological when inappropriate[136–144, 145, 146]. Excessive neuronal NADPH oxidase expression and activity in the NTS, RVLM and PVN are found in animal models of hypertension including those produced by myocardial infarction and by the chronic administration of phenylephrine, AngII, mineralocorticoids, or lipopolysaccharides [96, 133, 147–149]. Excessive neuronal NADPH oxidase activation in the hippocampus correlates with chronic stress and depression[145, 146]. The physiological ligand for the rapid non-genomic actions mediated by MR associated with the plasma membrane, is not certain. Studies in peripheral cells assume that aldosterone is the ligand, however 11β-HSD1&2 enzymes are microsomal. In hippocampal neurons the affinity of the membrane associated MR for glucocorticoids was less than that of the MR mediating transcriptional effects[150]. A high salt diet, oxidative and nitrosative stresses may induce post-translational modifications and alter MR intracellular trafficking, including to the membrane[1, 2], that might alter its ligand affinity, as well as its activity. There is evidence that increased ROS activates Rac1 GTP in neurons, as well as the kidney and heart, to produce ligand-independent activation of the transcriptional MR [134, 151]. Ligand-independent activation of the transcriptional MR has been used to explain why MR antagonists are effective treatments for heart failure even when circulating aldosterone levels are normal[134, 151].

Circulating inflammatory cytokines due to peripheral injury and inflammation increase SNS drive through an MR-mediated mechanism and may also explain some of the therapeutic effects MR antagonists in heart failure patients who have no significant aldosterone elevation. Pioneering studies from Robert Felder’s lab in rats, confirmed and extended by others, demonstrate that cardiac injury produced by a myocardial infarction activates neural afferents to the brain and increases circulating inflammatory cytokines that activate MR in the PVN, thus increase SNS drive to the periphery, exacerbating cardiac injury[92, 144, 149, 152–155]. Inflammatory cytokines do not normally cross the blood brain barrier to directly influence neurons in the PVN, however they increase cyclooxygenase 2 (COX-2) activity in perivascular macrophages, resulting in increased prostaglandin E2, which does diffuse across the blood brain barrier, and activation of NADPH oxidase through MR. These events lead to inflammatory cytokine production within the brain and sympathetic nervous system excitation[156]. Antagonists of cytokines, COX-2, NADPH oxidase, and ROS scavengers, as well as of the MR, inhibit this cascade of events and decrease the maladaptive persistent increase in SNS drive that prevents healing and exacerbates damage and dysfunction of the infarcted heart[96, 149, 154]. Similarly, inhibitors of inflammation also prevent the excessive sustained sympathetic nervous system activation and hypertension in aldo-salt and AII-salt excess models[96].

Activation of the sympathetic nervous and RAAS systems is an adaptive response to a decrease in blood pressure, however the sustained inappropriate increase in sympathetic activity is maladaptive and associated with hypertension and chronic heart failure[157]. The standard practice of using thiazide and thiazide-like diuretics as first line treatment for hypertension lowers the blood pressure, but may cause a compensatory increase in SNS activation, resulting in unintended and too often unmeasured and underappreciated effects of sympathoexcitation, including increased blood glucose and insulin resistance[158, 159]. Untoward effects related to inappropriate SNS activation were avoided with the use of low doses of MR antagonists with or instead of the diuretic[158, 159]. These findings further support the increased use of MR antagonists for the treatment of cardiovascular and cardiometabolic disease[160–162], but ignore other crucial functions of MR, particularly those in the cerebral cortex and hippocampus.

The mineralocorticoid receptor is highly expressed in the whole brain in areas that do not have direct influence over the cardiovascular system, but are involved in memory, learning, affect, and the regulation of the HPA axis [6, 9, 150, 163, 164]. MR function is essential for long term potentiation (LTP) formation, the basis for memory and learning[132, 165–167]. Diabetes in the rat causes a reduction in in vitro LTP formation and levels of brain derived neurotrophic factor that is reversed by an MR agonist [168]. In contrast to magnocellular neurons and pre-sympathetic neurons of the PVN, that express MR, but not GR[169], most, if not all MR in neurons in the hippocampus and cortex are co-expressed with GR and, as explained more fully below, are occupied by endogenous glucocorticoids, with MR occupied at normal diurnal peak levels and GR activated during higher stress levels[164, 170, 171]. The MR and GR in these neurons interact functionally to ensure the appropriate level of vigilance and reaction to stress[6, 150, 163, 164]. In addition to transcriptional effects, rapid non-nuclear effects involving presynaptic ERK1/2 and/or NADPH oxidase are essential for MR-mediated learning and memory formation[21, 134, 163, 172–177].

Confusion over the natural ligand for the mineralocorticoid receptor (MR)

Binding and activity studies demonstrated that natural glucocorticoids, albeit at significantly higher concentrations than those required for DOCA or aldosterone, also produced mineralocorticoid effects on epithelial ion transport in vitro in toad bladder-Ussing chamber studies used to define mineralocorticoid action [178, 179]. Similarly, early studies showed that aldosterone and DOCA activate the glucocorticoid receptor (GR)[180], but at higher concentrations than those attained even in Primary Aldosteronism, a fact to be considered when interpreting experimental results involving supra-physiological amounts of mineralocorticoids, especially DOCA[181]. Intravenous infusion of tritiated aldosterone and corticosterone in adrenalectomized rats pretreated with unlabeled GR-specific ligands produced distinct overlapping autoradiograph patterns in the brain[182–185]. Binding of both steroids was highest in the hippocampus, followed by the choroid plexus, and select nuclei of the brain stem. Binding in the cortex, thalamus and hypothalamus was significantly greater for aldosterone than for corticosterone [183–185]. By this time it was recognized that aldosterone binding to its receptor caused it to move into the cell nucleus and initiate transcription [186]. To avoid the bias of using large amounts of a single steroid in adrenalectomized rats as required for autoradiography, Yongue and Roy measured endogenous aldosterone and corticosterone within cell nuclei isolated from different parts of the brains of intact rats under basal and stressed conditions that increased corticosterone levels, and after a chronically low salt diet to stimulate or high salt diet to suppress endogenous aldosterone production[187]. Under these conditions of physiological concentrations of both aldosterone and corticosterone, the highest concentrations of aldosterone were found in cell nuclei from the hypothalamus, not from the hippocampus where autoradiography indicated the highest concentrations of MR would be found. The highest concentrations of corticosterone were measured in the hippocampus even at basal corticosterone levels[187]. Aldosterone retention in the brain was not as greatly influenced by circulating endogenous levels of steroid as was that of corticosterone[187].

In addition to his pioneering work with the role of mineralocorticoid excess in hypertension and its pathological effects in the heart and kidneys in the 1940’s, Hans Selye recognized the importance of steroid activity in the brain and that glucocorticoids acted both in the brain and periphery to produce the “general adaptation syndrome”, now known as the stress response [188, 189]. Demonstration of the selective retention of corticosterone in the hippocampus prompted studies on the role of glucocorticoids in cognition, learning, memory, and mood, as well as the response to stress by many labs[163, 190]. It became apparent that corticosterone was bound by two receptors in the hippocampus, a higher affinity, Type I corticosteroid receptor occupied at basal levels of corticosterone at the peak of its circadian rhythm, and Type II corticosteroid receptor that was occupied at stress levels [191] and that the balance of Type I and Type II corticosteroid receptor activation and function was crucial for normal adaptation to environmental stressors[189, 191]. Acute activation of the lower affinity Type II receptors dampened the responses to stress; severe or chronic over-activation of these was shown to be detrimental to neurons, reducing dendrites and synapses and neuronal death. Use of sophisticated methods for the time demonstrated that the hippocampal Type I and Type II corticosteroid receptors were the same as the MR and GR identified in peripheral tissues[192, 193]. Cloning of the human MR conclusively demonstrated that the MR and Type I corticosteroid receptor were the same[194]. The cloned MR was also shown to have similar affinities for aldosterone, cortisol, corticosterone, DOC, and progesterone in vitro[194], though it clearly had tissue-specific ligand preference in vivo described years before.

The glucocorticoids cortisol and corticosterone circulate at 1000 times the concentration of aldosterone in rats and humans with about 80–90% bound to corticosteroid binding globulin (CBG; transcortin) and albumin, providing concentrations of free glucocorticoids that are 100 times those of aldosterone. Stanley Ulick had demonstrated that patients with Apparent Mineralocorticoid Excess, a pseudohyperaldosteronism characterized by hypertension and hypokalemia, low aldosterone and normal cortisol levels, had impaired conversion of cortisol to its inactive metabolite cortisone due to a deficiency in 11β-hydroxysteroid dehydrogenase (11β-HSD) activity [195–197]. Paul Stewart showed that licorice inhibited 11β-HSD dehydrogenase activity and induced a similar syndrome[198]. Very soon thereafter the role of 11β-HSD dehydrogenase activity in providing pre-receptor ligand selectivity to the MR for aldosterone in aldosterone target cells was demonstrated[199–206].

There are at least two 11β-hydroxysteroid dehydrogenases, 11β-HSD1 and 11β-HSD2[204, 207–209];Gomez-Sanchez, 1997 #3110;Gomez-Sanchez, 1997 #2519;Krozowski, 1999 #7650}, located primarily in endoplasmic reticulum (ER) [210–212]. The existence of other 11β-HSDs has been postulated[213–215] and an 11β-HSD3 has been described in porcine testes[216]. Cloning and characterization of 11β-HSD2 [217, 218], clarified the confusion caused by experiments using tissue homogenates that tore the ER and juxtaposed enzymes and co-factors that normally reside in separate cellular compartments. 11β-HSD1 is bi-directional, but in the liver where it is most abundant it is responsible for the oxido-reduction of inactive cortisone and 11-dehydrocorticosterone, converting them to cortisol and corticosterone, thus increasing the intracellular availability of activating ligand for both the GR and MR, particularly in the liver, adipose tissue and hippocampus, where the MR is normally bound by glucocorticoids[219, 220]. The obligate cofactor for 11β-HSD1 reductase activity is NADPH which is regenerated from NADP⁺ within microsomes by hexose-6-phosphate dehydrogenase (H6PDH). (NADPH formed by glucose-6-phosphate dehydrogenase in the cytosol does not readily cross the ER membrane). Pre-adipocytes, do not express H6PD, thus 11β-HSD1 functions primarily as a dehydrogenase using NADP⁺ as cofactor in these cells. Upon adipocyte maturation and expression of H6PD, 11β-HSD1 becomes a reductase[221–223].

11β-HSD2 is a unidirectional NAD⁺-dependent dehydrogenase that converts cortisol and corticosterone to the inactive cortisone and 11-dehydrocorticosterone[210, 215, 224]. Aldosterone is not a substrate. In the absence of ligand, MR in aldosterone target cells are thought to associate with 11β-HSD2 within the endoplasmic reticulum creating a micro-environment in which aldosterone can attain sufficient concentrations compared to those of cortisol or corticosterone to bind and activate the MR[210, 211, 225]. An alternative hypothesis is that the change in redox potential produced by the dehydrogenase activity alters the structure, thus activity, of the MR bound by the more abundant glucocorticoid[226]. The metabolite of 11β-HSD2, 11-dehydrocorticosterone, was found to dampen MR-mediated increase in sodium transport, presumably acting as a check on aldosterone stimulated activity[227, 228]. Estrogens were found to significantly increase 11β-HSD2 expression in the rat kidney, however the formation of inactive dimers was also increased and the increase in 11β-HSD2 expression was not associated with an increase dehydrogenase activity[229, 230]. The mechanism for control of 11β-HSD2 dimerization has not been clarified.

11β-HSD2 is highly expressed in the normal placenta and the fetal brain where it mitigates exposure of the fetus to high glucocorticoid levels required for the mother’s metabolism during gestation[231–233]. Inhibition of 11β-HSD1 & 2 by relatively non-selective antagonists results intrauterine growth retardation and alterations of the HPA axis, increased anxiety, impaired ability to learn under stressful conditions, as well as hypertension[234–240]. The effects of 11β-HSD2 inhibition during gestation on the adult progeny are similar to those produced by fetal malnutrition, relatively benign perinatal stress, and iatrogenic glucocorticoid treatment during ontogeny, especially with synthetic steroids that are not inactivated by 11β-HSD2 that are also associated with low for term birth weight or intrauterine growth restriction (IUGR). IUGR is associated with epigenetic changes in the corticoid receptors that alter functional MR:GR ratios in tissue-specific ways and have significant negative implications on the HPA axis, energy metabolism, cardiovascular health, as well as the ability to cope with psychological stress as an adult[165, 234, 241–243].

As a reductase of cortisone/11-dehydrocorticosterone, 11β-HSD1 augments glucocorticoid action by increasing intracellular concentrations of active steroid, thus activation of either the MR or GR or both, depending on the cell type[220, 234, 243–246]. 11β-HSD1&2 may be found in different cell types in the same tissue. 11β-HSD1 in proximal cells and 11β-HSD2 in collecting duct cells of the kidney where both appear to function as a dehydrogenase[247]. Both are expressed and tightly regulated throughout gestation in the placenta, 11 beta-HSD1 in decidualized stromal cells on the maternal side of the placenta, 11 beta-HSD2 in villous cytotrophoblast, syncytiotrophoblasts and trophoblast cells from the fetus that invade the placental bed and maternal vasculature[248]. While most have found 11β-HSD1 expression in smooth muscle cells and 11β-HSD2 in endothelial cells of vessels[249], both enzymes were reported to be expressed in human aortic vascular and bronchial smooth muscle cells[250] 11β-HSD1, but not 11β-HSD2, in these smooth muscle cells was increased in by inflammatory cytokines, resulting in an increased net conversion of ³H-cortisone to ³H-cortisol [250]. Both 11β-HSD1 & 2, along with GR, and MR mRNAs, were reported to be expressed in isolated rat and bovine granulosa cells. Net glucocorticoid inactivation by 11 beta HSD2 in immature cells changed to activation by 11β-HSD1 in granulosa cells stimulated by gonadotropins [251–254]. Imunohistochemistry of the rat ovary detected strong 11β-HSD2, but no 11β-HSD1 in granulosa cells of immature follicles and both in the granulosa cells of the corpora luteum [27]. An increase in conversion of the inactive 11-dehydro metabolites to active cortisol and corticosterone after ovulation would mitigate local inflammation in acute injury, including rupture of the ovarian follicle or injured vessel[250, 253, 254], however chronically it could lead to inappropriate MR and GR activation. Another example of 11β-HSD dehydrogenase/reductase activity switching during differentiation is the adipocyte[221–223]. The requirement for pre-receptor modulation of glucocorticoid concentrations is clearly dynamic.

DNA methylation and epigenetic repression of 11β-HSD2 transcription was shown to cause of the increase in the ratio of tissue 11β-HSD1:11β-HSD2, thus glucocorticoid action, and to be implicated in the cardiometabolic syndrome[255]. 11beta-HSD2-deficiency or inhibition causes hypertension, while 11beta-HSD1 deficiency or inhibition is atheroprotective, improves glycemic control and reduces the ravages of aging on vessels, memory and cognition in animals [220, 249, 256–260] and humans[261]. Therefore 11β-HSD1 has become a target for the development of selective antagonists to mitigate diseases associated with excessive glucocorticoid action, including obesity, type II diabetes, cardiometabolic syndrome, and age-associated dementia in which excessive cortisol action on the GR is thought to have a role[257, 258, 262–267]. Selective 11β-HSD1 antagonists have been synthesized and tested [263, 264, 268–271], though not much information on their clinical efficacy is found in the literature as yet. They have shown promise in accelerating wound healing in stressed patients[272] and in ameliorating glycemic control in obese, but not simply overweight, type II diabetic humans[264].

Potentially significant environmental 11β-HSD enzyme inhibitors have been identified as disruptors of the balance of MR:GR action with implications for human health[273]. Endogenous antagonists of the 11β-HSDs, glycyrrhetininc acid like factors (GALFs) were proposed as modulators of MR action in experimental models and humans and a cause of hypertension[274–276]. GALFs are thought to be produced from endogenous glucocorticoids and their metabolites excreted in the bile, converted by microbial action in the gut to 21-dehydroxylated products, 11β-OH-progesterone, or 11β-OH-(allo)-5a-preganolones, then reabsorbed, later to be excreted in the urine from whence they have been isolated[274–276]. While the significance of these compounds has been questioned [277, 278], recent recognition of the importance of intestinal microbial action in metabolism and risk for a surprising variety of diseases may make this issue worth revisiting.

The possibility that 11β-HSD1 can function as a dehydrogenase is often overlooked; as both reductase and dehydrogenase functions are blocked by 11β-HSD1 inhibitors and may cause unexpected results in clinical trials of selective 11β-HSD1 inhibitors. Non-selective11β-HSD inhibitors increase blood pressure and SNS activity through MR activation in the PVN [279–281], but where 11β-HSD2 is not found [82, 107, 110]. This does not occur in adrenalectomized rats, confirming the requirement for endogenous adrenal steroids including corticosterone[281]. H6PD is required in the microsome to generate NADPH, the obligatory cofactor for 11β-HSD1 reductase activity. Pre-sympathetic neurons of the PVN were found to co-express MR and 11β-HSD1, but not H6PD or GR, suggesting that 11β-HSD1 acts as a dehydrogenase, providing extrinsic selectivity for aldosterone[82] and explain how the non-selective 11β-HSD inhibitors increase SNS and blood pressure[281]. In addition to cardiovascular and renal effects, SNS activation also increases gluconeogenesis, thus exacerbating metabolic problems associated with obesity and cardiometabolic syndrome[159]. Unfortunately, selective 11β-HSD1 inhibitors may not be the panacea for aging and obesity.

Expression of 11β-HSD2 is very limited in the adult brain

It has been detected in epithelia cells of the subcommissural organ and a few neurons of the ventromediolateral hypothalamus, vestibular nucleus, and NTS [82, 282–287]. Sodium depletion increases endogenous aldosterone levels and activity of aldosterone target neurons that co-express MR and 11β-HSD2 in the NTS, resulting in an increase in sodium appetite[286, 288]. These neurons become quiescent and sodium appetite decreases upon oral sodium repletion[286, 288]. Despite the absence of 11β-HSD2, ample evidence accrued over several decades by many labs indicates that aldosterone activates MR in autonomic neurons of the PVN. In addition the hypertension produced by the oral, subcutaneous or icv administration of non-selective 11β-HSD1&2 inhibitors glycyrrhizic and glycyrrhetinic acid and carbenoxolone, and the more 11β-HSD2 selective 11α-hydroxyprogesterone [289] produce hypertension that is prevented by the concomitant icv infusion of MR antagonists[279, 280]. Moreover, treatment or rats with glycyrrhizic acid and carbenoxolone produced sympathoexcitation that was inhibited by an MR antagonist and adrenalectomy prevented the effect, suggesting that a pre-receptor dehydrogenase mechanism protecting the MR from endogenous glucocorticoid activation exists in the PVN[281]. Both 11β-HSD1 and H6PD mRNA and protein were reported in the hypothalamus and other areas of the rat brain, however these studies were of relatively large blocks of tissue; they did not address individual cells[290]. More recently MR and 11β-HSD1, but neither GR nor H6PD, were detected by immunofluorescence within pre-autonomic neurons in the PVN identified by a retrograde tracer from the intermediolateral cell column of the spinal cord. Plasma aldosterone, MR expression and translocation to the nucleus, and c-fos activity were increased in pre-sympathetic neurons in the PVN of rats adapted to a low, compared to high salt diet, while the plasma corticosterone levels were not different [82]. These data suggest that 11β-HSD1 acts as a dehydrogenase and provide aldosterone selectivity to the MR in these pre-sympathetic neurons[82] and that the physiological role for MR in these neurons is to activate the SNS and maintain normal blood pressure during sodium depletion.

The balance between MR and GR mediated functions in the brain is crucial, differs for each cell type and is dynamic[6, 164, 291–293]. In the hippocampus, where GR activation is crucial for the modulation of MR-mediated excitation during stress[164], epigenetic repression of GR is associated with increased anxiety and response to stress[241, 242, 294]. Imbalance between MR and GR expression and/or activation also contributes significantly to depression and the loss of cognitive function during depression [163, 164, 173, 292, 293, 295, 296]. Decreases in absolute MR and relative levels of MR:GR message in the prefrontal and anterior cingulate cortex and, conversely, the increase in the MR:GR ratio in the PVN were documented in patients with depression compared to controls with no history of depression[297].

Transgenic mice with MR deleted only in the forebrain neurons have uncontrolled arousal and anxiety that impedes learning[173, 176], while over-expression of MR in the forebrain neurons reduces anxiety, and increases resilience to ischemia [298]. GR deletion decreases anxiety, but also impairs hippocampal-dependent explicit memory and increases HPA activity [296]. MR and GR neurotrophic effects are quite specific. Granule cells of the dentate gyrus of the hippocampus are lost in mice with targeted MR deletion in forebrain neurons; the morphology and function of hippocampal CA1 neurons are abnormal in those with GR deleted in forebrain neurons[296].

Similarly over-expression of the MR in the dentate gyrus granule cell layer in rats enhances the consolidation of non-spatial memory, augments short term memory, and protects against the effects of glucocorticoid excess in rats[174, 299], while selective over-expression of the MR in the basolateral amygdala of adult rats is anxiolytic and dampens to the response to acute stress [300]. Activation of GR decreases the excitability of neurons produced by a stressful event, thereby restoring normal function[171], however severe stress early or chronic stress later in life leads to epigenetic changes increasing GR expression and chronic suppression of neuronal excitation producing animal behaviors analogous to depression in humans[165, 301].

The chronic infusion of corticosterone into the lateral ventricle at the same or double the molar dose of aldosterone that produces hypertension by itself had no effect on blood pressure, however its co-infusion in equimolar amounts of aldosterone blocked the hypertension produced by the icv infusion of aldo alone [302]. Similarly, a series of experiments in which MR and GR agonists and antagonists were infused intracerebroventricularly confirmed that activation of central MR increases, and of GR decreases, the blood pressure[303]. While we detected no GR in the pre-synaptic neurons of the PVN, GR and MR are abundantly expressed in adjacent parvocellular neurons, some of which are interneurons that modulate the pre-sympathetic neurons[304]. The infused corticosterone in these experiments may have acted through GR or MR in neurons that moderate the activity of the pre-sympathetic neurons of the PVN. If this is so, while selective inhibitors of GR or 11β-HSD1 curb excessive GR activation within the hippocampus and slow the progression of the ravages of obesity and age experimentally, it may be at the cost of losing dampening effect on excessive sympathoexcitation. Similarly, though the benefits of MR antagonists in heart failure are indisputable, there is persuasive evidence based on animal and human studies that decreasing the functional balance between MR and GR in the cerebral cortex and hippocampus by inhibiting MR and/or increasing GR activities leads to deleterious changes in the HPA axis, depression, memory deficits and lower cognition[8, 163].

Fludrocortisone, an MR agonist, was shown to be an asset to the standard treatment for depression[305], however it would seem contraindicated for those at risk for hypertension or in heart failure. Depression, hypertension and cardiovascular disease are frequent co-morbidities and independent risk factors for each other[155, 306, 307], as are primary Aldosteronism and depression[59, 308]. All involve an imbalance of MR:GR activation in the brain.

In summary, as was shown for the stress response, memory, cognition, and mood[163, 188, 309, 310], the MR and GR work in concert in the brain to mediate essential and complementary actions for cardiovascular and osmotic homeostasis. In seeking to correct an imbalance between MR and GR, and aldosterone- and glucocorticoid-mediated actions in one target system, it is essential to consider the effects on others.

Figure.

Schematic of 3 types of MR ligand selectivity in neurons. Hp: Hippocampal neurons express both MR and GR; endogenous glucocorticoids are the physiological ligand for the MR and 11β-HSD1 is thought to act as a reductase. PVN: Pre-sympathetic neurons of the PVN express only MR, no GR; aldosterone appears to be the physiological ligand for the MR due to dehydrogenase activity of 11β-HSD1 in the absence of H6PD. NTS: Aldosterone target cells of the NTS express MR and 11β-HSD2, an obligate dehydrogenase. ?: preferred ligand for the membrane-associated MR is not certain. ??: whether GR is Expressed GR in aldo-target cells is not certain.

Highlights.

• Brain MR mediate diverse events including memory, learning & hemodynamic regulation

• The ratio of MR:GR is crucial for normal cell function and whole body homeostasis.

• Glucocorticoids occupy most neuron MR and GR at basal & stress levels, respectively.

• 11β-HSD1 dehydrogenase activity may confer aldosterone selectivity to MR in some neurons.

• MR & 11β-HSD1 antagonists may have good & bad consequences on brain MR function.