Mitochondrial dysfunction results in enhanced adrenal androgen production in H295R cells

The role of mitochondria in steroidogenesis is well established. However, the specific effects of mitochondrial dysfunction on androgen synthesis are not fully understood. In this study, we investigate the effects of various mitochondrial and metabolic inhibitors in H295R adrenal cells and perform a comprehensive analysis of steroid and metabolite profiling. We report that mitochondrial complex I inhibition by rotenone shifts cells toward anaerobic metabolism with a concomitant hyperandrogenic phenotype characterized by rapid stimulation of dehydroepiandrosterone (DHEA, 2 h) and slower accumulation of androstenedione and testosterone (24 h). Screening of metabolic inhibitors confirmed DHEA stimulation, which included mitochondrial complex III and mitochondrial pyruvate carrier inhibition. Metabolomic studies revealed truncated tricarboxylic acid cycle with an inverse correlation between citric acid and DHEA production as a common metabolic marker of hyper-androgenic inhibitors. The current study sheds light on a direct interplay between energy metabolism and androgen biosynthesis that could be further explored to identify novel molecular targets for efficient treatment of androgen excess disorders.


Introduction
Androgens are steroid hormones essential for reproduction and sexual development that are produced in the human fetal adrenals, adult adrenal zona reticularis (ZR), and gonads of both males and females [1].Steroidogenesis in the ZR requires a catalytic cascade of specific enzymes and cofactors that are located in the mitochondria and endoplasmic reticulum (ER).Mitochondria play an indispensable role in the initiation of steroid synthesis, as they contain the cholesterol side-chain cleavage enzyme (CYP11A1) on the inner mitochondrial membrane, which represents the enzymatic rate-limiting step in the cellular capacity for pregnenolone synthesis [2,3].The further conversion of pregnenolone to androgens requires two essential enzymes, namely cytochrome P450c17 (CYP17A1) and 3β-hydroxysteroid dehydrogenase type 2 (HSD3B2).CYP17A1 is a NADPH-dependent enzyme localized on the ER membranes that can catalyze 17α-hydroxylase and 17,20-lyase biosynthetic activities.HSD3B2 is a NAD+-dependent dehydrogenase with distinct subcellular localizations in the ER or mitochondria [3,4].
Reprogramming of the steroidogenic pathway in the ZR towards dehydroepiandrosterone (DHEA) synthesis is characterized by low expression of HSD3B2 and increased 17,20-lyase activity through increased expression of the allosteric activator cytochrome b5A (CYB5A) [5].Thus, androgen synthesis depends on a complex intraorganellar interplay between the mitochondrial and ER compartments.
Premature adrenarche (PA) and polycystic ovary syndrome (PCOS) are two common hyperandrogenic conditions in females [6,7].Adrenarche represents the postnatal development of the ZR in prepubertal children.PA is characterized by early maturation of the ZR and concomitant development of secondary sexual characteristics, typically before 8 years of age in girls, which may progress to PCOS [8,9].PCOS is a multisystem disorder with reproductive and metabolic perturbations characterized by hyperandrogenism with oligo-or anovulation and polycystic ovaries, and is often associated with insulin resistance, dyslipidemia and obesity.Recent studies have described mitochondrial gene mutations in patients with PCOS, suggesting that mitochondrial dysfunction may be involved in the development and progression of PCOS [10][11][12][13].Clinical and preclinical studies have reported that mitochondrial dysfunction indeed plays an essential role in insulin resistance, lipid metabolism, and follicular development [14,15].However, the driving factors of hyperandrogenic disorders and their possible links to energy metabolism are still poorly understood, making current treatment options scarce [6,16].
In the current study, the adrenal H295R cell model is used to investigate the effect of mitochondrial dysfunction on androgen production.We show that inhibition of mitochondrial complex I by rotenone (Rot) induces a rapid shift towards hyperandrogenic steroidogenesis.Furthermore, we have performed a comprehensive bioenergetic screening of metabolic and mitochondrial inhibitors, which demonstrated stimulated DHEA production upon inhibition of complex I, III or mitochondrial pyruvate carrier.Our metabolomic studies reveal a common pattern of truncated tricarboxylic acid cycle metabolism with an inverse correlation of citric acid and DHEA production upon hyperandrogenic inhibitors.Taken together, our study demonstrates that androgen biosynthesis is tightly coupled to a dysfunctional mitochondrial environment.Specific alterations reported here could be explored in the future to identify novel molecular targets for the treatment of androgen excess disorders.

Cell metabolic analysis
Cells were seeded in XFe96-well plates (7'500/well) and analyzed using the Seahorse XFe96 real-time cell metabolic analyzer (Agilent Technologies).Cellular oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were assessed in DMEM medium supplemented with 10 mM glucose, 2 mM glutamine, 2 mM pyruvate and 5 mM HEPES.After initial assessment of basal OCR and ECAR rates, specific inhibitors were added as indicated, followed by injection of oligomycin (1 μM) to determine the oxidative leak, and the combined injection of Rot/ antimycin A (1 μM/ 1 μM) to determine nonmitochondrial respiration.Each experiment was normalized to cell number by Hoechst staining (1 μg/mL).For the calculation of specific steady-states, OCR was corrected for non-mitochondrial oxygen consumption.ATP production rates were calculated using Agilent Wave software (Agilent Technologies).

Steroid profiling by LC-MS or DHEA ELISA assay
Cells were seeded in 12-well plates (1*10 6 /well) and exposed to specific inhibitors as indicated above.Quantification of complete steroid profiles in the culture supernatant was performed by liquid chromatography coupled to high-resolution mass spectrometry as previously described [17].For direct assessment of DHEA, cells were seeded in 96-well plates (50'000/well) and after exposure to specific inhibitors, the supernatant was assessed by a DHEA ELISA assay kit according to the manufacturer's instructions (Catalog ♯ DHA31-K01, Eagle Biosciences).Results from both assays were normalized to total amount of protein using the DC protein determination assay (BioRad).

Untargeted metabolomics and targeted quantification of TCA cycle metabolites
Cells were seeded in 12-well plates (1*10 6 /well), treated as indicated, washed with ammonium carbonate (75 mM, pH 7.4), and snap frozen in liquid nitrogen.For untargeted metabolomics, snap frozen cells were extracted with 70 % ethanol and analyzed by untargeted flow injection analysis on an Agilent 6550 Q-TOF instrument as previously described [18].Ions were putatively annotated based on accurate mass against the KEGG database with a tolerance of 0.001 Da.For quantification of TCA cycle metabolites including citrate and isocitrate, snap frozen cells were extracted with 50 % acetonitrile and analyzed using liquid chromatography tandem mass spectrometry on a Shimadzu Nexera X2 LCX30 coupled to AB Sciex 6500+ Qtrap instrument as previously described [19].Data was acquired with Analyst (version 1.7) and processed with SCIEX OS (version 3.1).Results were normalized to total amount of protein using the DC protein determination assay (BioRad).

Lipidomics
Cells were seeded in 12-well plates (1*10 6 /well), treated as indicated and pellets were harvested.Collected pellets were extracted using 50:50 (v/v) methanol/isopropanol for 1 h at − 20 • C. Untargeted lipidomics was performed by reversed-phase liquid chromatography coupled to high-resolution mass spectrometry on a Thermo Fisher Q-Exactive HF-X mass spectrometer as previously described [20].Data processing was performed using Compound Discoverer 3.1 (Thermo).Lipids were annotated with MS2 information and class-specific internal standards used for quantification.

Statistical analysis
All results are represented as the mean +/-standard error of the mean of at least three independent experiments.GraphPad Prism software (version 9) was used to plot graphs and perform statistical analysis, except for PCA analysis and heat map graphs, which were analyzed and plotted using MetaboAnalyst software (version 6.0).Data were analyzed by 2-tailed Student's t-test, unless otherwise indicated, and p-values of <0.05 (*) or <0.01 (**) were considered significant.

Inhibition of mitochondrial complex I stimulates cellular androgen content
To test whether inhibition of aerobic respiration affects cellular androgen levels, we treated H295R cells with the mitochondrial complex I inhibitor Rot and conducted cell metabolic analysis and profiled steroidogenesis.As expected, Rot exposure caused an immediate and dose-dependent inhibition of mitochondrial respiration with a subsequent increase in anaerobic metabolism leading to a profound adaptation in the bioenergetic profile and a shift of calculated ATP production rates from aerobic to glycolytic metabolism (Fig. 1A-C, Suppl.Fig. 1 A-D).Exposure time up to 24 h at 200 nM Rot did not induce cellular cytotoxicity and resulted only in a mild accumulation of mitochondrial superoxide (Suppl.Fig. 1E-I).Time-dependent analysis of steroidogenic profiles (Fig. 2 A-M) revealed a rapid increase in DHEA after 2 h exposure to Rot, which further accumulated after 6 and 24 h (Fig. 2K).Further androgens, such as DHEA-sulfate (DHEAS), androstenedione (Adione), and testosterone, were stimulated only after 24 h (Fig. 2J, L,  M).Other intermediates, such as pregnenolone (Preg), progesterone (Prog), 17αOH-pregnenolone (17αOH-Preg) and 17αOH-progesterone, were also increased after exposure to Rot (Fig. 2B, C, F, G).However, no significant stimulation of corticosterone or cortisol levels was observed, indicating no specific effect on mineralocorticoid or glucocorticoid accumulation (Fig. 2E, I).Based on the ratios of individual steroids to corresponding precursors, we implied related enzymatic activities in a subsidiary analysis (Suppl.Table 1).After 2 h of Rot exposure, we found decreased metabolic ratios of HSD3B2 (Prog/Preg and Adione/DHEA) and a slightly stimulated CYP17A1+CYB5A ratio (DHEA/17αOHPreg), which were not persistent after 6 or 24 h.Consistent with increased testosterone levels at 24 h, we found an increased 17β-hydroxysteroid dehydrogenase (HSD17B) enzymatic metabolic ratio at 24 h Rot exposure (Suppl.Table 1).Taken together, these results indicate that inhibition of aerobic respiration by Rot causes a rapid and profound shift in steroidogenesis towards androgen excess.

Specific mitochondrial dysfunction environments induce cellular DHEA excess
Based on the rapid onset of DHEA excess, we hypothesized that acute post-transcriptional metabolic adaptation is a major driver of the observed phenotype and characterized the metabolic adaptation in more detail.Knowing that Rot causes broad metabolic adaptations, we wanted to analyze whether other mitochondrial or metabolic disruptors have a similar ability to induce acute hyperandrogenic states.We therefore screened DHEA secretion in parallel with bioenergetic analysis after 6 h drug exposure using compound concentrations with an optimal metabolic output, but without a relevant cell proliferative effect (Fig. 3A-C).Exposure to the mitochondrial complex III inhibitor antimycin A (AA) caused a profound shift from aerobic to glycolytic metabolism and simultaneously stimulated DHEA accumulation.In contrast, inhibition of mitochondrial complex V by oligomycin (Oligo) also shifted metabolism to glycolysis but decreased DHEA secretion.Dissipation of mitochondrial membrane potential by the uncoupler FCCP (carbonylcyanide-p-trifluoromethoxyphenylhydrazone) shifted metabolism to a stressed metabolic phenotype with stimulated aerobic and anerobic metabolism, but was unable to affect DHEA secretion.Competitive inhibition of glycolysis by 2DG shifted metabolism toward higher aerobic metabolism and did not affect DHEA secretion.Interestingly, direct inhibition of the mitochondrial pyruvate carrier (MPC) by UK5099 (UK), which imports pyruvate into the mitochondrial matrix and thus promotes the direct flow of glycolytic flux into the mitochondria, increased DHEA levels and shifted the metabolic profile towards a similar glycolytic state as Rot.Other metabolic perturbations, such as inhibition of mitochondrial fatty acid uptake by Etomoxir (Eto) or inhibition of mitochondrial glutamine uptake by BPTES (bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl) ethyl sulfide 3), did not affect DHEA levels.Taken together, the screening suggests that acute inhibition of CI, CIII, or MPC stimulates DHEA accumulation in adrenal H295R cells.

Metabolic markers of truncated TCA cycle are associated with hyperandrogenic mitochondrial dysfunction
Next, we addressed the question of whether the hyperandrogenic inhibitors Rot, AA and UK share common metabolic drivers and examined the adaptive effects on the cellular metabolome and lipidome after 6 h drug exposure.We additionally compared the generated profiles with the non-androgenic OXPHOS inhibitor Oligo to exclude general metabolic effects upon mitochondrial dysfunction.Principal component analysis (PCA) of the assessed untargeted metabolome revealed clustering of biological replicates with preferential grouping of the OXPHOS inhibitors Rot, AA and Oligo versus DMSO and UK, thus not separating hyperandrogenic conditions (Fig. 4A).Analysis of the most altered metabolite ions (log2FC >0.5, adj.p-value <0.01) identified decreased intracellular (iso)citrate as a common marker in all hyperandrogenic treatments (Fig. 4B, Suppl.Table 2), while no common stimulated metabolic markers could be identified.Next, we analyzed tricarboxylic  acid (TCA) cycle-related metabolite ions in more detail and additionally found that aconitate, pyruvate, and glutamate were decreased by all androgenic inhibitors (Fig. 4C-D).Targeted metabolite analysis confirmed decreased intracellular citrate and isocitrate levels in all hyperandrogenic treatments (Rot, AA, and UK), which was unaffected by the non-androgenic inhibitor Oligo (Fig. 4E-F).Finally, we tested the effect of direct supplementation of the cell culture medium with citrate, which did not affect DHEA levels after 6 h of drug exposure (Fig. 4G).
Knowing that hyperandrogenic states are associated with alterations in lipid metabolism [21,22], we additionally performed a lipidomic investigation to search for specific lipid markers.PCA again revealed a grouping of the OXPHOS inhibitors Rot, AA and Oligo versus DMSO and UK, thus not separating hyperandrogenic conditions (Fig. 5A).Analysis by total lipid class showed triglyceride accumulation in all OXPHOS inhibitors (Rot, AA, and Oligo) (Fig. 5B).Analysis of top altered lipid ions (log2FC >0.5, adj.p-value <0.01) did not identify a common hyperandrogenic marker, but additionally revealed an accumulation of unsaturated triglycerides upon exposure to OXPHOS inhibitors (Fig. 5C, Suppl.Table 3).Assuming that these lipids may be stored in intracellular lipid droplets, we performed additional studies using confocal imaging, but did not find any relevant reorganization of intracellular lipid droplet formation upon 24 h Rot exposure (Suppl.Fig. 2A-B).Taken together, these results demonstrate that the acute adaptive effect on the lipid desaturation response is associated with mitochondrial OXPHOS dysfunction induced by Rot, AA, and Oligo, whereas the common metabolic pattern upon hyperandrogenic condition induced by Rot, AA, and UK represents a dysfunctional TCA cycle.

Discussion
In this study, we describe a close interaction between mitochondrial bioenergetics and androgen biosynthesis in a steroid cell model.Our study shows that mitochondrial CI, CIII or MPC inhibition leads to a rapid DHEA accumulation.The three identified hyperandrogenic mitochondrial inhibitors have different molecular mechanisms: Rot affects mitochondrial NADH oxidation by inhibiting the transfer of electrons from CI to ubiquinone and thus still allowing convergent electron flow toward ubiquinone (e.g. through CII or alternative dehydrogenases).In contrast, AA inhibits the electron transfer of CIII from ubiquinol to cytochrome c, which effectively inhibits electron flow through the OXPHOS system [23].Finally, inhibition of MPC by UK reduces the mitochondrial pyruvate uptake as a substrate for the TCA cycle and indirectly affects the respiratory capacity from pyruvate oxidation, thus stimulating the entry of alternative metabolites into the TCA cycle (e.g. through glutamine anaplerosis) [24].
We showed that the increase in DHEA is associated with a shift in cellular metabolism from aerobic to anaerobic glycolysis with a concomitant dysfunctional TCA cycle leading to decreased cellular citrate.Stimulation of anaerobic glycolysis has been described in patients with PCOS, with a strong positive correlation between lactate levels and insulin resistance [25].Our study indicates that stimulated anaerobic glycolysis alone is not specific for hyperandrogenic condition, as shown by metabolic screening of various metabolic or mitochondrial inhibitors.In agreement with our in vitro study, metabolomic analysis performed in plasma of women with PCOS reported an inhibited TCA cycle with reduced citrate levels [25,26], suggesting similar systemic changes in whole-body metabolism.Citrate plays several important roles in cellular energy production and biosynthetic processes: It is a key intermediate in Indicated is (iso)citrate as a common metabolite ion found in all hyperandrogenic treatments (highlighted with a green line).C-D.Schematic representation of TCA-associated metabolism and corresponding heatmap of detected metabolites using untargeted metabolomics.Hyperandrogenic treatments (highlighted with a green line) are characterized by decreased glutamate, pyruvate, (iso)citrate, and aconitate.E-F.Quantitative validation of intracellular citrate and isocitrate levels by targeted LC-MS analysis.G. Accumulation of DHEA in the supernatant after 6 h drug exposure with or without 0.5 mM citrate supplementation in the culture medium.DHEA levels were significantly changed compared to the respective control (*), while there was no significant (ns) change with additional citrate supplementation.
the mitochondrial TCA cycle, where citrate synthase catalyzes the condensation of acetyl-CoA and oxaloacetate as a first step, and citrate is then further oxidized to produce reduction equivalents (NADH and FADH 2 ) for mitochondrial OXPHOS.It can also be transported to the cytoplasm via the mitochondrial citrate transporter, where ATP-citrate lyase generates acetyl-CoA for metabolic processes such as lipid biosynthesis.It is also a known inhibitor of phosphofructokinase and thus plays a regulatory role in glycolysis [27].Consequently, low levels of citrate in the cytoplasm can stimulate glycolysis and are key features of the Warburg effect in tumor cells [28].Since mitochondrial citrate production is the primary source for most cells, exogenous citrate transport is usually not of functional importance.Interestingly, RNA sequencing of microdissections of the human ZR identified SLC13A5, a plasma membrane sodium/citrate-cotransporter, as a highly expressed transporter [29,30].This is a surprising finding that may indicate a nutrient-limited state in human ZR that has not been addressed so far.In our study, direct supplementation of citrate in the culture medium did not affect DHEA levels after hyperandrogenic treatment, suggesting that citrate itself may not be the main player, but rather a marker of an underlying metabolic adaptation leading to hyperandrogenic steroidogenesis.
Another hallmark of human hyperandrogenic conditions is an altered lipid status [21,22].Our study found a decrease in cholesterolesters (CE) and an accumulation of total triglycerides (TGs) with a concomitant desaturation response to CI, CIII and CV inhibition, which was thus specific to OXPHOS inhibition but not to DHEA excess.The decreased cellular CE were concomitant with lower cholesterol levels, which might represent the previously described coregulation of mitochondrial function and cholesterol biosynthesis [31,32].While the increased total TG may be the consequence of decreased energy-dependent mitochondrial fatty acid oxidation [33,34], the observed lipid desaturation response is related to stimulated ER metabolism, known as a mechanism for glycolytic NAD+ recycling [35] and highlights the close adaptive interplay between mitochondria and ER during mitochondrial dysfunction.Indeed, a detailed analysis of the ER-mitochondrial interplay may be important to gain further insight into the role of mitochondrial metabolism in the regulation of androgen synthesis.Steroidogenesis is performed in compartmentalized fashion between mitochondrial and ER subcellular compartments and is dependent on different subcellular redox pools and pairs, e.g.NADP(H) and NAD(H).ER-localized CYP17A1 is NADPH-dependent, which can be generated by anabolic reactions such as the cytosolic pentose phosphate pathway.HSD3B2, localized in the mitochondrial matrix but also in extramitochondrial compartments [36,37], depends on NAD+ and is therefore mainly associated with catabolic reactions.The relative lack of HSD3B2 expression/activity required for androgen production could thus be influenced by the pathways regulating mitochondrial NADH production and/or regulation of the subcellular distribution of HSD3B2.
The identification of several mitochondrial disruptors with hyperandrogenic effects supports the hypothesis that androgen excess is influenced by specific environments of mitochondrial dysfunction.Of course, the drugs reviewed in this study do not represent a complete survey of disruptors of mitochondrial metabolism.Therefore, it is highly likely that additional hyperandrogenic stimulators will be identified in subsequent studies.Several low-molecular weight drugs with off-target mitochondrial dysfunction have reached the market.One example is the anticonvulsant valproic acid, which has hyperandrogenic side effects [38,39] and is a well-known mitochondrial toxicant with diverse pathomechanisms on fatty acid oxidation and TCA cycle dysfunction [40].Metformin is an insulin-sensitizing drug used as a treatment option for PCOS with a mechanism of action that is not fully understood [6,41].It has been associated with CI inhibition in vitro, but this finding is related to supraphysiologic concentrations (1 mM) [42], which are not reached in humans and therefore do not contradict our results.Interestingly, metformin is not an effective treatment option for all PCOS patients, suggesting that hyperandrogenic clinical phenotypes may be associated with multiple pathomechanisms.
The direct role of androgens and the androgen receptor (AR) in the regulation of glucose and lipid metabolism is well established.There is increasing evidence for a direct link between androgens and AR in the control of mitochondrial function [43,44].For example, AR has been shown to be imported into and localized to mitochondria, where it has an inverse relationship with mitochondrial DNA content and thus negatively affects OXPHOS [45,46].Furthermore, in prostate cancer cells, AR directly regulates the transcription of MPC2, a co-stabilizing protein of MPC, thus maintaining metabolic homeostasis by controlling the import of pyruvate into the mitochondrial matrix [47].Taken together, these results provide further evidence to support the hypothesis of mitochondrial dysfunction associated with androgen excess.
Our study is limited in several ways.We used the H295R adrenal cancer cell line, which is a simplified system compared to a human, but is well established to study human steroidogenesis and is known to respond with a hyperandrogenic steroid profile under different cell culture conditions [48,49].Consistent with our results, we have discussed confirmatory in vitro and in vivo data in the literature, so we are confident in the validity of our findings.To date, we have measured total levels of steroids and have not performed precursor labelled experiments to support where and how steroidogenesis is affected and whether other compensatory mechanisms occur.We also measured the total intracellular metabolomic content and have not provided a comprehensive picture of subcellular metabolites.Further resolution of metabolic adaptation is needed to specify differences in metabolic flux, e.g. through 13 C tracing and subcellular metabolite measurements.We have shown that different inhibitors of mitochondrial metabolism are fast drivers of DHEA production within 2-6 h, thus making strict genomic regulation unlikely, but gene expression profiling could add information to a potentially complex metabolic interplay.
In conclusion, our functional and metabolomic screening of mitochondrial dysfunction in adrenal H295R cells allowed us to identify a critical role for mitochondrial metabolism in androgen biosynthesis.The current data provide a starting point for further research to explore key metabolic drivers to provide specific pathomechanisms, identify predictive biomarkers, and advance the development of specific treatment options for patients with hyperandrogenic symptoms.

D
.Mathis et al.

Fig. 1 .
Fig. 1.Mitochondrial Complex I Inhibition shifts Metabolic Flux to Anaerobic Metabolism. A. Schematic representation of mitochondrial complex I inhibition by Rot, resulting in decreased oxidative phosphorylation (OXPHOS) and compensatory stimulation of glycolysis.Arrow thickness indicates stimulation or inhibition of the respective pathway.B. Bioenergetic plot showing shift towards anaerobic metabolism.Decreased oxygen consumption rate (OCR) and stimulated extracellular acidification rate (ECAR) were assessed by extracellular flux analysis after 6 h exposure to Rot (200 nM).C. Calculated ATP production rates from bioenergetic flux analysis.

Fig. 4 .
Fig. 4. Impaired TCA Cycle is Associated with Mitochondrial Dysfunction Causing DHEA Excess. A. Principal Component Analysis (PCA) with two principal components after 6 h of Rot, AA, UK, Oligo exposure versus DMSO control.B. Volcano plots showing statistically significant differences for the four treatments (closed blue dots log2FC >0.5, adj.p-value <0.01).Indicated is (iso)citrate as a common metabolite ion found in all hyperandrogenic treatments (highlighted with a green line).C-D.Schematic representation of TCA-associated metabolism and corresponding heatmap of detected metabolites using untargeted metabolomics.Hyperandrogenic treatments (highlighted with a green line) are characterized by decreased glutamate, pyruvate, (iso)citrate, and aconitate.E-F.Quantitative validation of intracellular citrate and isocitrate levels by targeted LC-MS analysis.G. Accumulation of DHEA in the supernatant after 6 h drug exposure with or without 0.5 mM citrate supplementation in the culture medium.DHEA levels were significantly changed compared to the respective control (*), while there was no significant (ns) change with additional citrate supplementation.