Simvastatin treatment aggravates the glucocorticoid insufficiency associated with hypocholesterolemia in mice.

BACKGROUND AND AIMS
Statin treatment disrupts HMG-CoA reductase-mediated endogenous cholesterol synthesis and lowers plasma LDL-cholesterol levels. Although statin treatment can theoretically impair adrenal steroid hormone synthesis, thus far, no effect on glucocorticoid output has been described, as LDL-cholesterol levels usually remain within the physiological range. However, novel statin-based treatment regimens that dramatically decrease LDL-cholesterol levels are currently employed. Here, we assessed whether inhibition of cholesterol synthesis under these relatively hypocholesterolemic conditions may alter adrenal glucocorticoid output.


METHODS
Hypocholesterolemic apolipoprotein A1 (apoA1) knockout mice were administered high dose simvastatin twice daily for 3 days.


RESULTS
Simvastatin treatment did not change plasma cholesterol levels or modify the adrenal expression levels of genes involved in cholesterol metabolism. However, simvastatin treatment lowered basal plasma levels of the primary glucocorticoid corticosterone (-62%; p < 0.05). Upon injection with adrenocorticotropic hormone, control-treated apoA1 knockout mice already showed only a mild increase in plasma corticosterone levels, indicative of relative glucocorticoid insufficiency. Importantly, simvastatin treatment further diminished the adrenal glucocorticoid response to adrenocorticotropic hormone exposure (two-way ANOVA p < 0.05 for treatment). Peak corticosterone levels were 49% lower (p < 0.01) upon simvastatin treatment.


CONCLUSIONS
We have shown that simvastatin treatment aggravates the glucocorticoid insufficiency associated with hypocholesterolemia in mice. Our data suggest that (1) HMG-CoA reductase activity controls the adrenal steroidogenic capacity under hypocholesterolemic conditions and (2) imply that it might be important to monitor adrenal function in humans subjected to statin-based treatments aimed at achieving sub-physiological LDL-cholesterol levels, as these may potentially execute a negative impact on the glucocorticoid function in humans.


Introduction
Glucocorticoids, through an interaction with their cognate nuclear glucocorticoid receptor, modulate a great variety of physiological processes, including glucose metabolism and inflammation. High levels of glucocorticoids are secreted by adrenals in response to activation of the hypothalamus-pituitary-adrenal axis to facilitate the physiological response to stress [1]. In line with the importance of glucocorticoids in normal physiology, subjects suffering from glucocorticoid insufficiency, i.e. Addison's disease patients, display an overall increased risk for mortality [2].
Glucocorticoids, such as cortisol in humans and corticosterone in mice, are synthesized from their common precursor cholesterol. Several lines of evidence have suggested that adrenal de novo cholesterol synthesis does not contribute to the generation of the steroidogenic cholesterol pool under normolipidemic conditions. More specifically, studies in rats have shown that activity of the HMG-CoA reductase, the rate-limiting enzyme in de novo cholesterol synthesis, within the adrenals is relatively low as long as cholesterol can be obtained from lipoproteins in the plasma compartment [4]. In addition, little to no activity of HMG-CoA reductase can be detected in adrenals of wild-type mice [3]. As such, lipoproteins should be regarded as the primary source of cholesterol used for glucocorticoid production by the adrenals.
From in vivo studies, it has become apparent that low-density lipoprotein (LDL) as compared to high-density lipoprotein (HDL) is the preferred lipoprotein substrate for the adrenals in humans. Human HDL deficiency or lack of a functional HDL receptor is associated with a decreased urinary cortisol concentration [7,8]. In addition, plasma HDL-cholesterol levels correlate with the synacthen-induced cortisol response in critically ill patients [9]. HDL may therefore also contribute, at least a bit, to adrenal glucocorticoid synthesis in humans. However, HDL-cholesterol is not able to compensate for a lack of cholesterol associated with LDL particles in humans. As a result, LDL deficient abetalipoproteinemic patients, that contain relatively normal amounts of HDLcholesterol, suffer from glucocorticoid insufficiency [11e13]. In further support of a limited relevance of HDL in the supply of cholesterol to the adrenal steroidogenic machinery, HDL is redundant for adrenal steroidogenesis in mice with a human-like, LDLrich, lipoprotein phenotype [10].
LDL-cholesterol lowering, i.e. through statin therapy, is the first treatment goal in high-risk cardiovascular disease patients, given that the presence of relatively high plasma levels of cholesterol associated with LDL is an established risk factor for the development of atherosclerotic lesions [14]. A~30% cardiovascular disease risk reduction is generally achieved through statin treatment [15], which shows the success of the application of statins as LDLcholesterol lowering drugs in clinical practice. Despite the fact that the adrenals acquire the majority of the cholesterol used for glucocorticoid synthesis from LDL particles, thus far, statin treatment has not shown to execute a major effect on the adrenal steroid output in humans [16e22]. It should, however, be acknowledged that LDL-cholesterol levels usually remain within the normal physiological range in response to classical statin therapies, as they mostly have been applied to reverse (excessive) hypercholesterolemia. In contrast, novel cardiovascular therapeutic approaches, e.g., anti-proprotein convertase subtilisin/kexin type 9 (PCSK9) antibody/statin co-treatment and ezetimibe/statin co-treatment, aim to create a treatment condition of relative hypocholesterolemia, i.e., in which the amount of LDL-cholesterol in the blood circulation is actually lower than that of normolipidemic subjects [23,24].
Hypocholesterolemic mice as compared to normolipidemic wild-type mice display a marked increase in adrenal relative mRNA expression levels of HMG-CoA reductase, which translates into a significant rise in adrenal HMG-CoA reductase activity [3,5]. It thus appears that when relatively low amounts of cholesterol can be acquired from lipoproteins, adrenal de novo cholesterol synthesis is activated in an attempt to overcome cholesterol insufficiency. Based upon this notion, it can be anticipated that de novo cholesterol synthesis also makes a significant contribution to the generation of the steroidogenic cholesterol pool in humans subjected to the aggressive LDL-cholesterol lowering therapies, such as high dose statin/PCSK9 inhibitor (combination) treatment. Statins execute their cholesterol-lowering effect by inhibiting HMG-CoA reductase activity, which leads to an upregulation of LDL receptors and a concomitant increase in the clearance of LDL particles from the blood circulation by the liver. Importantly, studies in mice have suggested that statin treatment not only diminishes hepatic cholesterol synthesis, but also significantly inhibits the incorporation of acetate into lipids within the adrenals [28]. When applied in aggressive LDL-cholesterol lowering combination therapies, statins may, therefore, also disrupt the (compensatory) increase in adrenal cholesterol synthesis induced by relative hypocholesterolemia, theoretically resulting in an impairment of the adrenal glucocorticoid function.
Here we aimed to provide experimental proof for our working hypothesis that statin-mediated suppression of adrenal de novo cholesterol synthesis will affect glucocorticoid output under hypocholesterolemic conditions. To this purpose, we subjected genetically hypocholesterolemic mice to simvastatin treatment and investigated the adrenal glucocorticoid function.

Experimental mice and treatment
Apolipoprotein A1 (apoA1) x LDL receptor (ldlr) double knockout mice were obtained from Dr. Jan Albert Kuivenhoven from the Department of Experimental Vascular Medicine, Academic Medical Centre, Amsterdam. Male double knockout mice were mated with female C57BL/6 mice to generate double heterozygous offspring, which was intercrossed to ultimately derive apoA1 single knockout mice that were inbred for further expansion of the colony. Animal experiments were performed in a temperature and light cycle (12 h light/12 h dark) controlled room at the Gorlaeus Laboratories of the Leiden Academic Centre for Drug Research, in accordance with the National Laws. All experimental protocols were approved by the Ethics Committee for Animal Experiments of Leiden University.
Male apoA1 knockout mice (13e18 weeks old;~30 g) were single housed for 1.5 weeks preluding as well as during the experiment. All mice had continuous free access to food and water. Simvastatin was obtained in tablets from Pharmachemie (Haarlem, The Netherlands). Tablets were crushed with a mortar and pestle and subsequently dissolved in tap water. Age-matched groups of apoA1 knockout mice (N ¼ 9 per group) were orally administered twice a day 250 ml of water either with or without 100 mg/kg body weight simvastatin supplementation for 3 days, with the final dose given at 9:00 AM [25]. Two hours after the final dosing, all mice were bled via tail tip cut for plasma cholesterol and basal corticosterone measurements. Subsequently, mice were administered 200 mg of adrenocorticotropic hormone (ACTH; MT-Diagnostics, Etten-Leur, The Netherlands) intraperitoneally to stimulate the adrenal glucocorticoid output [26]. One and two hours after ACTH injection, tail blood samples were obtained for corticosterone measurement and/or hematological analysis. During blood draws, mice were handled identically and restrained for a maximum of 30 s to exclude an impact on blood corticosterone levels. Mice were sacrificed through cervical dislocation and organs were harvested. Adrenals and livers of all mice were isolated within 30 min, i.e., maximally 2.5 h and 4.5 h after the administration of ACTH and the final dose of simvastatin, respectively, and stored at À20 C until further use.

Plasma cholesterol analysis
Concentrations of free and total cholesterol were determined in plasma obtained from tail blood using enzymatic colorimetric assays (Roche Diagnostics).

Plasma corticosterone analysis
Corticosterone levels were determined in tail blood plasma using the [ 3 H] radioactive kit from MP Biomedicals (Irvine, CA).

Analysis of gene expression by real-time quantitative PCR
Quantitative gene expression analysis on livers and adrenals was performed as described [27]. In short, total RNA was isolated using a standard phenol/chloroform extraction method and reverse transcribed with RevertAid™ reverse transcriptase. Gene expression analysis was performed using real-time SYBR Green technology (Eurogentec). Primers were validated for identical efficiencies. Primer sequences are available upon request. Beta-actin, glyceraldehyde-3-phosphate dehydrogenase, ribosomal protein L27, and peptidylprolyl isomerase A were used as standard housekeeping genes in both liver and adrenal gene expression analyses. Relative expression levels were calculated by subtracting the cycle threshold number (Ct) from the gene of interest from the average housekeeping Ct and raising 2 to the power of this difference. No significant difference in the housekeeping Ct values was observed between the two experimental groups.

Data analysis
Statistical analysis was performed using Graphpad Instat software (San Diego, USA, http://www.graphpad.com). Normality testing of the experimental groups was performed using the method of Kolmogorov and Smirnov. Significance was calculated using a two-tailed Student's t-test or two-way analysis of variance (ANOVA) with Bonferroni post-test where appropriate. Probability values less than 0.05 were considered significant.

Results
Previous studies using gene knockout mice have indicated that, probably since HDL is the predominant lipoprotein species circulating in murine plasma, HDL represents the major source of cholesterol utilized for glucocorticoid synthesis in mice [3,5,6]. Genetic deletion of apolipoprotein A1 (apoA1) expression in mice is associated with marked hypocholesterolemia due to HDL deficiency [3]. To identify a potential effect on steroidogenesis of statinmediated disruption of adrenal de novo cholesterol synthesis under hypocholesterolemic conditions, we determined the impact of simvastatin treatment on the glucocorticoid function of apoA1 deficient mice. Simvastatin was chosen for our studies, since this statin species appears to be most effective in targeting adrenals in vivo [28]. Koga et al. previously observed that treatment of mice with a single oral dose of 20 mg/kg simvastatin reduces adrenal HMG-CoA reductase activity by 40e50% [28]. With the aim to achieve a maximal extent of adrenal de novo cholesterol synthesis inhibition, we applied an oral dosing regimen of simvastatin 100 mg/kg twice daily for 3 days [25] in our apoA1 knockout mouse model.
Schonewelle et al. have previously shown that, in C57Bl/6 wildtype mice, statin-induced inhibition of HMG-CoA reductase activity is associated with a compensatory increase in the hepatic mRNA expression of genes involved in cholesterol acquisition and synthesis [29]. As can be appreciated from Fig. 1, we also observed a significant rise in LDL receptor mRNA expression (þ39%; p < 0.001) in livers of our apoA1 knockout mice upon simvastatin treatment. Furthermore, a 45% higher mRNA expression of HMG-CoA reductase was detected in livers from simvastatin-treated apoA1 knockout mice as compared to those from water devoid of simvastatin control-treated apoA1 knockout mice (Fig. 1). Although this latter effect did not reach statistical significance due to the large intra-group variation (p ¼ 0.07), it can be anticipated that in our current experimental setup simvastatin, indeed, effectively inhibited HMG-CoA reductase activity.
In contrast to the potent lipid-lowering effect of simvastatin observed in humans [30], simvastatin treatment did not impact on the extent of hypocholesterolemia. Simvastatin-treated apoA1 knockout mice exhibited free and total cholesterol levels of 11 ± 1 mg/dl and 28 ± 2 mg/dl, while these respective levels were 10 ± 1 mg/dl and 30 ± 3 mg/dl in control apoA1 knockout mice orally administered the solvent only. A potential difference in the steroidogenic capacity of the adrenals in response to simvastatin exposure cannot be attributed to a change in exogenous substrate availability.
Since an efficient (intra)cellular mobilization of cholesterol by the adrenals is essential for glucocorticoid synthesis [6,31,32], we verified whether simvastatin treatment affected the adrenal expression levels of genes involved in cholesterol homeostasis. As evident from Fig. 2, no significant change in the relative transcript abundance of the scavenger receptor BI (sr-bi), the LDL receptor (ldlr), or HMG-CoA reductase (hmgcr) was observed in response to simvastatin treatment. A direct effect of simvastatin on adrenal lipoprotein-cholesterol uptake and de novo cholesterol synthesis can thus be considered highly unlikely. Relative mRNA expression levels of the ATP-binding cassette transporter A1 (abca1) and apolipoprotein E (apoE) were equal within the two groups, suggesting a similar rate of cholesterol efflux from adrenocortical cells to apoE-containing HDL-like particles. Notably, as judged from the unchanged expression levels of steroidogenic acute regulatory protein (star), mitochondria localized glutamic acid rich protein (hummr/mgarp), and cytochrome P450 family 11 subfamily A member 1 (cyp11a1), simvastatin treatment did not impact on the intracellular trafficking of free cholesterol between the endoplasmatic reticulum and mitochondria or the actual conversion of cholesterol into steroids.
Plasma levels of the primary glucocorticoid species corticosterone are normally~100 ng/ml in the basal state and increase rapidly to levels >250 ng/ml in response to stress in normolipidemic wildtype mice [3,5,6,31]. In agreement with the findings of Plump et al. that apoA1 deficient mice as compared to wild-type mice show a reduced (basal) steroidogenesis rate [3], basal levels of corticosterone were 31 ± 9 ng/ml in our water-treated apoA1 knockout mice (Fig. 3A). Strikingly, despite the already low levels in the control condition, simvastatin treatment was able to further decrease baseline corticosterone values (À62% vs. control-treated mice; p < 0.05; Fig. 3A). Fig. 1. The effect of simvastatin treatment on hepatic relative expression levels of the LDL receptor (ldlr) and HMG-CoA reductase (hmgcr) in apoA1 knockout mice. Beta-actin, glyceraldehyde-3-phosphate dehydrogenase, ribosomal protein L27, and peptidylprolyl isomerase A (PPIA) were used as housekeeping genes for normalization. Data represent means ±SEM of 8e9 mice per group. ***p < 0.001 versus Control.
Pituitary-derived adrenocorticotropic hormone (ACTH) is a potent endogenous activator of adrenal glucocorticoid production and an ACTH challenge is, therefore, an appropriate method to measure the maximal adrenal glucocorticoid output in mice [5,26]. Control apoA1 knockout mice displayed a marked rise in plasma corticosterone levels as compared to baseline, at one hour after ACTH injection, which was, however, completely abrogated one hour later. In light of the fact that wild-type mice display a maximal plasma corticosterone level over a period of at least three hours in response to an ACTH challenge [5], this finding provides additional proof of a generally low maximal steroidogenic capacity of apoA1 knockout adrenals [3]. Importantly, as evident from Fig. 3B, the one-hour plasma peak value of corticosterone was 49% lower in simvastatin-treated mice as compared to controls (two-way ANOVA Bonferroni post-test: p < 0.01).

Discussion
In the current study, we evaluated the effect of statin treatment on the adrenal glucocorticoid output under hypocholesterolemic conditions in mice. Our data confirm previous observations that hypocholesterolemia, i.e., HDL deficiency, in mice is associated with an impaired adrenal glucocorticoid response to stress [3,5,33]. Given that ACTH treatment was able to induce only a short-term (<2 h) and rather limited rise in plasma corticosterone levels in our HDL deficient apoA1 knockout mice, lipoprotein-associated cholesterol should definitely be regarded as the primary steroidogenic cholesterol source.
Here we have shown that simvastatin-induced inhibition of HMG-CoA reductase activity in apoA1 knockout mice translates into a >60% decrease in basal plasma corticosterone levels and a marked impairment to respond to an established steroidogenic trigger, e.g., ACTH exposure. We have not been able to measure the actual extent of inhibition of adrenal HMG-CoA reductase activity in response to simvastatin treatment. Although, based on previously published studies [28], we are confident that we achieved at least a 50% reduction in adrenal HMG-CoA reductase activity through simvastatin treatment, it is fair to anticipate that we did not completely block enzyme activity in the adrenals. This can explain why our simvastatin-treated animals were still able to increase their plasma corticosterone level, albeit to a relatively very low extent, upon ACTH exposure. Nevertheless, our data clearly indicate that intra-adrenal cholesterol synthesis contributes significantly to the generation of the steroidogenic cholesterol pool in a lipoprotein deficiency context as simvastatin treatment markedly reduced both the basal and maximal adrenal glucocorticoid output in apoA1 knockout mice.
The decrease in glucocorticoid output upon simvastatin treatment was not due to a decrease in exogenous substrate availability as the extent of hypocholesterolemia was identical in the two groups of apoA1 knockout mice. The absence of an effect on plasma cholesterol levels was to be expected, since statin treatment generally does not impact on plasma total cholesterol levels in mice [34]. Since statin treatment does lower LDL-cholesterol levels in humans, one could argue that the absence of a lipid-lowering effect in response to simvastatin treatment makes it hard to translate our findings from mice to the human situation. However, the main conclusion of our study is that under conditions where the amount Fig. 2. The effect of simvastatin treatment on relative expression levels of genes involved in cholesterol uptake (sr-bi/ldlr) and synthesis (hmgcr), intracellular trafficking (star/hummr), and efflux (abca1/apoE) and metabolism of cholesterol to steroids (cyp11a1) in adrenals from apoA1 knockout mice. Beta-actin, glyceraldehyde-3-phosphate dehydrogenase, ribosomal protein L27, and peptidylprolyl isomerase A (PPIA) were used as housekeeping genes for normalization. Data represent means ±SEM of 8e9 mice per group. of cholesterol substrate in the plasma compartment is limited, e.g., under low HDL conditions in mice and low LDL conditions in humans, adrenal HMG-CoA reductase is of importance to generate sufficient cholesterol to maintain a proper steroidogenesis rate. Furthermore, given the high similarity in the glucocorticoid insufficiency phenotype of LDL deficient humans [11e13] and HDL deficient mice [3,5], we do value that our findings have a high translational power. As such, we regard it important to determine the adrenal function in patients treated with aggressive statin (combination) treatments aimed at inducing relatively low LDLcholesterol levels to identify a potential risk of developing glucocorticoid insufficiency.
In conclusion, we have shown that simvastatin treatment aggravates the glucocorticoid insufficiency associated with hypocholesterolemia. Our data (1) supply novel proof for the hypothesis, originally formulated by Plump et al. [3], that HMG-CoA reductase activity is increased under hypocholesterolemic conditions to compensate for the loss of lipoproteins as external cholesterol source and maintain the adrenal steroidogenic capacity and (2) imply that it might be important to monitor the adrenal function in humans subjected to high dose statins in (combination) therapies aimed at achieving sub-physiological LDL-cholesterol levels, as these may potentially execute a negative impact on the glucocorticoid function in humans.

Conflict of interest
The authors declared they do not have anything to disclose regarding conflict of interest with respect to this manuscript.

Financial support
These studies were supported by the Dutch Heart Foundation (Grants 2008T070 and 2012T080 awarded to Menno Hoekstra) and the Netherlands Organization for Scientific Research (VICI Grant 91813603 awarded to Miranda Van Eck).

Author contributions
Menno Hoekstra designed, executed and supervised the study and drafted the manuscript. Amber B. Ouweneel, Ronald J. van der Sluis and Joya E. Nahon were essential in the execution of the experiments and the analysis and interpretation of the data and critically read the manuscript. Miranda Van Eck is head of the lipid group within the Division of Biopharmaceutics, Cluster Bio-Therapeutics, of the Leiden Academic Centre for Drug Research and did critical revisions to the manuscript.