Amiodarone disrupts cholesterol biosynthesis pathway and causes accumulation of circulating desmosterol by inhibiting 24‐dehydrocholesterol reductase

We have earlier reported that amiodarone, a potent and commonly used antiarrhythmic drug increases serum desmosterol, the last precursor of cholesterol, in 20 cardiac patients by an unknown mechanism.


Introduction
Amiodarone has been used clinically for more than half a century and is the most commonly used potent antiarrhythmic drug to suppress atrial and ventricular tachyarrhythmias [1]. It acts through multiple mechanisms, including prolongation of repolarization, reduction of excitability and slowing of conduction [2]. It is an iodinated derivative of benzofuran and is metabolized in the liver to an active metabolite desethylamiodarone (DEA). Amiodarone has a very long half-life of approximately 60 days and has several adverse effects, adding up to 15% during the first year of treatment and up to 50% during long-term use [1]. The most serious adverse effects are pulmonary and hepatic toxicity and abnormalities in thyroid function. The benzofuran ring is highly lipophilic, resulting in effective distribution of the drug into different organs and cells, where the drug is assumed to cause structural and functional disturbances, a 'lipid traffic jam' [3,4]. It is not clear whether the amiodarone-induced lipid accumulation induces the organ toxicities described during amiodarone treatment.
We have earlier described in a limited number of cardiac patients with myocardial inflammatory disease (i.e. 20 patients with cardiac sarcoidosis or giant cell myocarditis) that amiodarone affected cholesterol biosynthesis [5]. An unexpected finding in amiodarone-treated patients was that the concentration of serum desmosterol, the last precursor of cholesterol in the Bloch unsaturated sterol side chain pathway, was increased on average 12-fold compared with the control subjects [5]. The mechanism by which amiodarone increases circulating desmosterol is not known [6]. However, desmosterol can regulate inflammatory responses [7], raising the possibility that the observed desmosterol accumulation upon amiodarone treatment in this study was somehow related to inflammatory cardiac diseases.
The aim of the present study was to evaluate in a clinical case-control setting if the amiodaroneinduced increase in serum desmosterol levels can be reproduced in a large and heterogeneous group of cardiac patients excluding inflammatory myocardial disease, and whether it depends on gender, age, body mass index (BMI) or different diseases. In addition, we evaluated whether the inhibition of hydroxy-methyl-glutaryl CoA reductase (HMGCR) by statins interfered with the amiodarone-induced effects on cholesterol precursors and whole-body cholesterol metabolism. Finally, we investigated the detailed mechanism(s) by which amiodarone interfered with the metabolism of desmosterol using in vitro studies.

Clinical study population
The study population consisted of 236 consecutive cardiac patients, 174 men and 62 women with a mean age of 61 AE 1 (SE) years, who were referred to the Department of Cardiology, Helsinki University Hospital in the years 2018-2019 because of various cardiac problems. Of these patients, we selected 126 consecutive subjects who were on long-term amiodarone treatment for severe arrhythmias (amiodarone group), and 110 ageand sex-matched controls without amiodarone treatment (control group). The primary cardiac diagnoses in the control group were matched as much as possible with those of the amiodarone group. Amiodarone and lipid-lowering treatment had to be unchanged for at least eight weeks before the study. In-hospital patients were studied when they had reached a stable state of health. All subjects gave their written informed consent. The study was performed according to the principles of the Declaration of Helsinki. The Ethics Committee for the Department of Medicine, Hospital District of Helsinki and Uusimaa approved the study protocol.

Clinical study design
Blood samples were drawn after a 12-h fasting. The subjects were weighed and their height was measured, BMI was calculated as weight (kg)/ height (m) 2 , and clinical data including medical history and symptoms, laboratory tests (especially alanine aminotransferase and total or free T4), and current drug treatments were recorded. The in-hospital subjects were studied on the standard hospital diet and outpatients on their normal habitual diet.

Clinical study methods
Serum cholesterol and noncholesterol sterols zymostenol (called cholestenol in our previous study) [5], desmosterol, lathosterol and squalene (i.e. cholesterol precursors), and campesterol, sitosterol, stigmasterol and avenasterol (i.e. plant sterols), and cholestanol (5a-saturated derivative of cholesterol) were analysed using gas-liquid chromatography (GLC) with a 50-m capillary column (Ultra 2, Agilent Technologies, Wilmington, DE) and flame ionization detection with 5a-cholestane as internal standard [8]. The identities of the sterol peaks were further verified by mass spectrometry [5]. Serum concentrations of the noncholesterol sterols and squalene were adjusted to cholesterol in the same GLC run and expressed as ratios to cholesterol (10 2 µmol mmol À1 cholesterol) in order to enable their comparison between subjects and populations with different serum cholesterol levels. Therefore, serum noncholesterol sterols are given as ratios to cholesterol herein. Serum triglyceride concentrations were analysed enzymatically using an automated analyser (Thermo Fisher Scientific Indiko TM ).

Statistical analyses
The statistical analyses were performed with SPSS for Windows 22.0 (SPSS, Chicago, IL). Sample size calculation was based on significance levels (a = 0.05 and b = 0.20) and essential information provided from a previous study [5]. Using these estimates, the size of the required population was appropriate. Normality and homogeneity of variance assumptions were checked before further analyses, and variables not normally distributed were transformed logarithmically. Continuous variables in the groups were compared by analysis of variance, and the noncontinuous variables were tested by using the chisquare or Fisher's exact test. Variables not normally distributed even after logarithmic transformation or nonhomogenous in variance were tested with Kruskal-Wallis and Mann-Whitney U-tests. Spearman's correlation coefficients were calculated. A P-value of <0.05 was considered statistically significant, whereas a P-value of >0.05 was denoted as nonsignificant (NS). The results are expressed as mean AE SE.

Study population and clinical characteristics
In the amiodarone (n = 126) and control (n = 110) groups, age and BMI were similar, and 74% of the subjects were men in both groups (Table 1). Age and BMI were also similar between gender in the groups. In the amiodarone group, ventricular tachyarrhythmias were more frequent than in the control group (P < 0.05), whereas atrial tachyarrhythmias were similar between the groups. The number of cases with diagnosis of cardiomyopathy (CMP), either hypertrophic or dilated, or coronary artery disease (CAD) was comparable between the study groups as were congenital anomalies or valvular defects (9% in the amiodarone vs 5% in the control group (NS)) and the frequency of type 2 diabetes (13% in the amiodarone and 11% in the control group (NS)). In the amiodarone and control groups, 48% and 42% of the subjects were on statin treatment (NS), and in both groups, 80% of the subjects on statin were males. In both groups, one subject was on ezetimibe treatment and two subjects had a combination therapy of statin + ezetimibe. In the control group, one and in the amiodarone group two subjects were on low-dose corticosteroid therapy. In the amiodarone group, the median daily dose of amiodarone was 200 mg with a range of 100-400 mg. No signs of thyroid, pulmonary or liver toxicity were observed in the patients at the time of our study.

Serum lipids
We measured cholesterol, noncholesterol sterols and triglycerides in the patients' sera. The ratios to   [14,15]. Statin treatment was taken as a covariate in all serum lipid analyses. The concentrations of serum total cholesterol and triglycerides did not differ significantly between the study groups (Table 1), reversing our previous finding of elevated serum triglyceride concentrations by amiodarone in a pilot study (5). Neither gender nor the low-dose corticosteroid therapy had any effect on serum cholesterol, triglycerides or noncholesterol sterols in the groups.
Of the serum cholesterol precursors, desmosterol and zymostenol were~7-and~2-fold higher and lathosterol 50% lower, on average, in the amiodarone than in the control group (P < 0.001 for all; Table 1). Serum desmosterol correlated with the daily amiodarone dose (r = 0.365, P < 0.001), but was independent of age, gender, BMI, lipid-lowering treatment or the type of cardiac disease in the amiodarone-treated patients.
The moderate increase in serum zymostenol, already observed in our previous study (5), indicates that amiodarone also interfered with the Kandutsch-Russell saturated sterol side chain pathway. This can be explained by the in vitro observation that amiodarone inhibits the activity of D8-D7 isomerase (also known as EBP), blocking the conversion of zymostenol to lathosterol 16 and, in fact, serum lathosterol was lowered by amiodarone (Table 1). Of the cholesterol absorption markers, serum stigmasterol and cholestanol were elevated in the amiodarone compared with the control group. In spite of these interferences, serum cholesterol precursors correlated significantly inversely with serum absorption markers in the amiodarone and control groups (e.g. lathosterol vs. campesterol r = À0.455, P < 0.001 in the amiodarone and r = À0.436, P < 0.001 in the control group), as well as in both groups without and with statin treatment, suggesting intact cholesterol homeostatic regulation in all groups.
Statin treatment affected serum cholesterol and the noncholesterol sterols alike in the amiodarone and the control groups. Thus, serum cholesterol and the cholesterol precursors were lower and the plant sterols and cholestanol higher in statin + than statin-subjects in both groups, as shown for serum cholesterol, desmosterol and campesterol in Fig. 1. In fact, the plant sterols and cholestanol were even higher in the statin + amiodarone than statin + control group, as shown for campesterol in Fig. 1.
In the amiodarone group, we had access to serum samples taken before and during amiodarone In each case, serum desmosterol levels were elevated during amiodarone treatment, with the increase of desmosterol ranging from 297 to 1071 10 2 µmol mmol À1 cholesterol (Fig. 2).

Effect of amiodarone on sterol levels in human cells
To analyse if amiodarone or its major metabolite in vivo DEA affects cellular sterol synthesis, we cultivated human A431 cells in lipoprotein-deficient serum (to increase endogenous cellular cholesterol synthesis) in the presence of amiodarone or DEA. For comparison, we generated A431 cells from which the enzyme converting desmosterol to cholesterol, DHCR24, was knocked out using CRISPR/Cas9 gene editing (DHCR24-KO cells). We found that amiodarone and DEA effectively increased cellular desmosterol levels and concomitantly reduced cholesterol levels (Fig. 3), suggesting that both compounds inhibit DHCR24 in cells. In addition, zymosterol and zymostenol accumulated in amiodarone or DEA treated cells. DHCR24-KO cells exhibited a pronounced accumulation of desmosterol but not zymosterol or zymostenol (Fig. 3), indicating that accumulation of the latter two sterols is not secondary to DHCR24 inhibition. Rather, amiodarone and DEA also partially blocked EBP (D8-D7 isomerase) activity, in agreement with previous findings [16].

Effect of amiodarone on DHCR24 enzyme activity
To test if amiodarone directly inhibits DHCR24, we set up an in vitro assay. As a proof-of-principle, we first tested amiodarone versus an established inhibitor of DHCR24, triparanol, in cell culture using CHO-7 stable cells that express human DHCR24 (CHO-DHCR24-V5) [11]. Cells cultivated in lipoprotein-deficient serum were incubated with deuterated desmosterol for 4 h. We found that both amiodarone and triparanol inhibited DHCR24 activity in cell culture ( Fig. 4a and b), demonstrated by the higher ratio of [ 2 H 6 ]-desmosterol to [ 2 H 6 ]-cholesterol in triparanol and amiodaronetreated cells (Fig. 4a, bottom panels), compared to DMSO control (Fig. 4a, top right). Unlabelled cells were included to distinguish peaks from cellular endogenous cholesterol and desmosterol from their deuterated counterparts (Fig. 4a, top left). To test if amiodarone directly inhibits DHCR24, we isolated microsomes from these cells and performed an in vitro enzyme activity assay. This demonstrated that amiodarone directly inhibits the activity of DHCR24 (Fig. 4c) across two concentrations.

Discussion
This study shows that in cardiac patients amiodarone increased the levels of the circulating cholesterol precursors, desmosterol and zymostenol~7-and~2-fold and decreased serum lathosterol level to a half, on average, irrespective of gender, age, BMI, or primary cardiac or other diseases, or the use of lipid-lowering drugs. We have previously demonstrated that amiodarone markedly elevated serum levels of desmosterol and to a lesser extent zymostenol in a limited number of subjects with inflammatory cardiac disease [5]. These findings were confirmed and expanded in the present larger, controlled and well-matched study. Amiodarone seemed to interfere with both Bloch and Kandutsch-Russell cholesterol synthesis pathways.
Regarding the basic mechanisms, we revealed that amiodarone results in desmosterol accumulation in cultured human cells and that the compound directly inhibits the DHCR24 enzyme activity in microsomal membranes. Moreover, the increase in zymostenol was not secondary to DHCR24 inhibition, but rather amiodarone and DEA appeared to partially block D8-D7 isomerase activity, in keeping with previous findings [16].
Paradoxically, elevated serum desmosterol and zymostenol levels reflect upregulated whilst lowered serum lathosterol reflects downregulated whole-body cholesterol synthesis, as verified in different populations both at baseline and during lipid-lowering drug treatments [14,15,17]. It is possible that blocking the conversion of zymostenol to lathosterol by amiodarone resulted in low lathosterol levels in vivo. Another possibility is that high desmosterol per se interfered with cholesterol homeostasis. In fact, desmosterol is a ligand for liver X receptor (LXR), and inhibits the activation of sterol response element-binding protein-2 (SREBP-2), thereby downregulating cholesterol synthesis [18,19]. However, the effect of amiodarone on LXR and SREBP-2 would require formal investigation.
Due to the increase of some but decrease of other cholesterol synthesis markers, it was not straight forward to determine the effect of amiodarone on whole-body cholesterol synthesis. However, elevated stigmasterol and cholestanol levels in the amiodarone group suggested increased cholesterol absorption efficiency. According to the reciprocal regulation of cholesterol synthesis and absorption [20], which, importantly, was intact in the amiodarone group, amiodarone most likely upregulated intestinal cholesterol absorption whilst downregulating cholesterol synthesis. The tendency to steady state in human cholesterol metabolism explains the similar cholesterol concentrations in the amiodarone and control groups, unlike in the in vitro studies, where cellular cholesterol was decreased by amiodarone.
Regarding the combination of amiodarone with statins, several interesting observations were made. First, in the amiodarone group, the statin-induced inhibition of HMGCR was clearly only partial, since we still observed the desmosterol accumulation in this group. Metabolic studies have demonstrated that the statin-induced reduction in whole-body cholesterol synthesis is~40% [21], but this may vary between statins and dosage. Second, a more significant upregulation of the absorption markers with the combination of statin + amiodarone treatment compared to statin alone may result from the fact that amiodarone increases the circulating concentration of statins by inhibiting CYP3A4; consequently, cholesterol absorption increases [17]. In addition, it is possible that blocking cholesterol synthesis at two steps (early and late) might give an enhanced compensatory increase in cholesterol and sterol absorption in general, and this may contribute to the similar cholesterol reduction with both synthesis inhibitors vs. statin alone. During amiodarone + statin treatment, inhibiting cholesterol absorption with phytostanols or ezetimibe could still lower serum cholesterol concentration. In fact, some patients in the amiodarone group had amiodarone + statin + ezetimibe combination therapy, with serum cholesterol concentrations within the goal values and without side effects (data not shown).
As previously proposed [5], the simplest explanation for the increase in serum desmosterol observed in response to amiodarone treatment is that this drug inhibits DHCR24, the enzyme reducing the double bond in the side chain of desmosterol to make cholesterol. Supporting this idea, we observed in cell experiments that amiodarone increased desmosterol at the expense of cholesterol (Figs 3, 4a and b). Moreover, we found using an isolated microsome system that amiodarone directly inhibits DHCR24 activity to a comparable extent as the archetypal DHCR24 inhibitor, triparanol (Fig. 4c). Amiodarone and triparanol share structural similarities [6]. In particular, they both possess a diethylaminoethoxy side chain which corresponds in length and general conformation to the sterol side chain [22]. It has been proposed that the lone electron of the nitrogen atom, which is situated at a position analogous to C-24 of the sterol, plays the same binding role as the D24 double bond electrons of the desmosterol substrate [23]. Further work would be required to determine if amiodarone and triparanol share the same mechanism of inhibition, which for triparanol has been reported to be noncompetitive [24].
At therapeutic concentrations, amiodarone alters the functions of multiple ion channels controlling the cardiac action potential [25]. Importantly, these are integral membrane proteins that are affected by the surrounding lipid bilayer. Based on our results, amiodarone modifies the sterol composition of cell membranes, and it seems plausible that this may contribute to the multi-target effects of the drug. On the other hand, a major limitation of long-term amiodarone therapy is its broad and unusual side effect profile. The present data also open the possibility that some of the adverse effects might in fact result from DHCR24 inhibition. For instance, DHCR4 inhibition can lead to cataracts [26], a known side effect of amiodarone. However, the amiodarone-related transient cataract is less severe compared with that caused by triparanol treatment. The latter develops rapidly within 3 to 19 months from starting the therapy and leading to severe cataracts requiring operative treatment even in middle-aged subjects without any predisposing risk factors [27,28]. Also, the sensitivity of the thyroid to amiodarone-induced toxicity (hypo-or hyperthyroidism) might be linked to high levels of DHCR24 in the thyroid: DHCR24 is upregulated by thyroid hormones and thyroid hormone receptor [29,30]. Pulmonary side effects might, instead, be related to the phospholipidosis typically induced by cationic amphiphilic drugs, rather than direct enzyme inhibition. Regarding the liver toxicity, amiodarone can cause nonalcoholic steatohepatitis [31] and interestingly, preliminary results demonstrate that liver desmosterol accumulation was correlated with the severity of inflammation and steatosis [32].

Conclusions
We have demonstrated that amiodarone increased serum desmosterol level independently of gender, age, BMI, cardiac and other diseases, and lipidlowering medication. Lipid analyses in patient samples taken before and after initiation of amiodarone showed a systematic increase of desmosterol upon drug administration, strongly arguing for a direct causal link between amiodarone and desmosterol accumulation. Mechanistically, we found that amiodarone resulted in desmosterol accumulation in cultured human cells and that the compound directly inhibited the DHCR24 enzyme activity in microsomal membranes.