bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 Hepatocyte androgen in females mediates androgen-induced hepatocellular mishandling and 2 systemic resistance 3 4 Stanley Andrisse1,13, Mingxiao Feng1, Zhiqiang Wang1, Olubusayo Awe1,4, Lexiang Yu1, Haiying Zhang5,

5 Sheng Bi5, Hongbing Wang6, Linhao Li6, Serene Joseph2, Nicola Heller7, Franck Mauvais-Jarvis8,9,10, Guang

6 William Wong4, James Segars3, Andrew Wolfe1, Sara Divall11, Rexford Ahima12, Sheng Wu1,2,3,4*

7 1Department of Pediatrics, Johns Hopkins University School of Medicine, Baltimore, MD, 21087

8 2Department of Physiology/Center for Metabolic Disease Research, Temple University School of Medicine,

9 Philadelphia, PA, 19140

10 3Department of Gynecology and Obstetrics, 4Department of Cellular and Molecular Physiology, 5Department of

11 Psychiatry and Behavioral Sciences, 7Department of Anesthesiology and Critical Care Medicine, 12Department

12 of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, 21287

13 6Department of Pharmaceutical Sciences, University of Maryland, Baltimore, MD, 21201

14 8Department of Medicine, Tulane University Health Sciences Center, New Orleans, LA 70112,

15 9Tulane Center of Excellence in Sex-Based Biology & Medicine, New Orleans, LA 70112,

16 10Southeast Louisiana, VA Medical Center, New Orleans, LA 70119,

17 11Department of Pediatrics, University of Washington, Seattle’s Children’s Hospital, Seattle, WA 98145-5005

18 13Department of Physiology and Biophysics, Howard University, Washington, D.C. 20059

19 * Correspondence and name of person to whom reprint requests should be addressed:

20 Sheng Wu

21 3500 North Broad Street / MERB 456

22 Philadelphia, PA, 19140

23 215-7073704

24 Email : [email protected]

25 Keywords: Androgen, , PCOS, insulin signaling, liver, glucose

26 Abbreviations: AR, androgen receptor; PCOS: polycystic ovary syndrome; DHT, dihydrotestosterone; CAH,

27 congenital adrenal hyperplasia; HA, hyperandrogenemia; Con, control; Veh, vehicle; LH, luteinizing hormone;

28 Foxo1, forkhead box O1; PEPCK, phosphoenolpyruvate carboxykinase; G6Pase, glucose 6-phosphatase; bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

29 I.P., intraperitoneal; DPM, determine disintegrations per minute; BMI, body mass index (BMI); GTT, glucose

30 tolerance test; ITT, insulin tolerance test; PTT, pyruvate tolerance test; GSIS, glucose stimulated insulin secretion;

31 HGP, hepatic glucose production; GIR, glucose infusion rate.

32

33 Abstract

34 Androgen excess is one of the most common endocrine disorders of reproductive-aged women, affecting up to

35 20% of this population. Women with elevated androgens often exhibit hyperinsulinemia and insulin resistance.

36 The mechanisms of how elevated androgens affect metabolic function are not clear. Hyperandrogenemia in a

37 dihydrotestosterone (DHT)-treated female mouse model induces whole body insulin resistance possibly through

38 activation of the hepatic androgen receptor (AR). We investigated the role of hepatocyte AR in

39 hyperandrogenemia-induced metabolic dysfunction by using several approaches to delete hepatic AR via animal-

40 , cell-, and clinical-based methodologies. We conditionally disrupted hepatocyte AR in female mice

41 developmentally (LivARKO) or acutely by tail vein injection of an adeno-associated virus with a liver-specific

42 for Cre expression in ARfl/fl mice (adLivARKO). We observed normal metabolic function in littermate

43 female Control (ARfl/fl) and LivARKO (ARfl/fl; Cre+/-) mice. Following chronic DHT treatment, female Control

44 mice treated with DHT (Con-DHT) developed impaired glucose tolerance, pyruvate tolerance, and insulin

45 tolerance, not observed in LivARKO mice treated with DHT (LivARKO-DHT). Further, during an euglycemic

46 hyperinsulinemic clamp, the glucose infusion rate was improved in LivARKO-DHT mice compared to Con-DHT

47 mice. Liver from LivARKO, and primary hepatocytes derived from LivARKO, and adLivARKO mice were

48 protected from DHT-induced insulin resistance and increased . These data support a paradigm in

49 which elevated androgens in females disrupt metabolic function via hepatic AR and insulin sensitivity was

50 restored by of hepatic AR.

51

52

53

54 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

55 1. Introduction

56 Androgen excess is one of the most common endocrine disorders of reproductive-aged women, affecting

57 up to 20% of this population[1]. Disorders of hyperandrogenemia in women, such as polycystic ovary syndrome

58 (PCOS) and congenital adrenal hyperplasia are associated with metabolic dysfunction and infertility[2-4]. Women

59 with androgen excess are at a much higher risk than weight-matched women without androgen excess of

60 developing dysglycemia, and the majority (70%) manifest insulin resistance whether lean or obese[5], with 10%

61 developing type-2 diabetes mellitus (T2D)[6].

62 Similar to women, animal models of female hyperandrogenemia (HA) display metabolic and reproductive

63 dysfunction[2,7-13]. Despite the importance of the liver in systemic insulin action, little is known about the role

64 of hepatocytes in androgen-induced insulin resistance in vivo in females. Using a model of low-dose

65 dihydrotestosterone (DHT) to achieve serum levels of DHT 2-fold higher than controls, similar to those of women

66 with PCOS or corrected late-onset congenital adrenal hyperplasia (CAH)[14,15], we showed that androgens

67 modulate liver glucose metabolism to mediate hyperandrogenemia-induced metabolic dysfunction in female

68 mice [16]. This 2XDHT model reflects the androgen level elevation encountered in female human disease and is

69 distinct from other animal models of female hyperandrogenemia[9,17-19] in that body composition (lean mass,

70 fat mass, total weight), serum levels of estradiol, testosterone, and luteinizing hormone (LH) [16,20] during two

71 to three months treatment are similar between treated and untreated mice. This model allows us to parse out the

72 weight independent effects of hyperandrogenemia on glucose metabolism.

73 While hepatic AR is not required for metabolic function in female mice with normal androgen levels[21],

74 the role of hepatic AR in hyperandrogenic environments has not been explored. The liver plays a major role in

75 determining systemic insulin resistance and metabolism[22] and its function is altered in women[23] and mice[24]

76 with hyperandrogenemia. Here, we investigate the direct role of hepatocyte AR in hyperandrogenemia-induced

77 metabolic dysfunction by using several approaches to delete hepatic AR via animal-, cell-, and clinical-based

78 methodologies. We hypothesize that hepatic AR mediates hyperandrogenemia-induced metabolic abnormalities

79 in female mice and that its deletion will ameliorate hepatic and peripheral metabolic dysfunction. This knowledge bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

80 may contribute to development of novel and targeted therapeutics to limit the metabolic sequelae of

81 hyperandrogenism experienced by females.

82 2. Materials and Methods

83 2.1. Generation of liver specific AR knockout females.

84 ARfl mice (exon 2) on a C57BL/6 background were generated by Karel De Gendt [25]. Frozen embryos

85 obtained from the European Mutant Mouse Archive (Rome, Italy), were re-derived by the Johns Hopkins

86 Transgenic Core. To produce developmental hepatic AR knockout mice (LivARKO, ARfl/fl; albumin (Alb)-Cre+/-

87 ), we crossed an exon 2-floxed AR [25,26] female (ARfl/fl; Cre-/-) mouse, with a male albumin-Cre+/- mouse,

88 (strain number: B6N.Cg-Speer6-ps1Tg(Alb-Cre)21Mgn/J) purchased from Jackson Laboratories, Bar Harbor,

89 ME. The Alb-Cre mouse produces liver-specific expression of Cre driven by the albumin promoter in a C57/BL6

90 background. Litter mates (ARfl/fl; Alb-Cre-/-) were referred to as Control (Con) mice. Alb-Cre mice are not known

91 to display any metabolic phenotype related to the transgenic allele [27]. Once female mice reached 2-months old,

92 4 mm-DHT or vehicle pellets were inserted to the mice under the skin [16,20,28-31]. To acutely knockout AR,

93 we used an adeno-associated virus expressing the Cre bacteriophage recombinase, AAV8.TBG.PI.-Cre.rBG. The

94 thyroid binding globulin (TBG) promoter is a specifically expressed in the liver [32]. One-hundred

95 microliters of AAV8.TBG.PI.-Cre.rBG [32] (referred to as adLivARKO mice) or pAAV.TBG.PI.Null.bGH

96 (referred to as adCon mice) were injected into ARfl/fl female mice. The liver has been proven as the only targeted

97 tissue by this vector [32]. The two-month-old female mice were inserted with 4 mm-DHT or vehicle pellets

98 simultaneously with tail vein virus injections. Metabolic tests were performed between 1-3-months post insertion.

99 At 2-3-months post insertion, tissues were extracted for further analysis as described below. All mice were housed

100 in the Johns Hopkins University mouse facility, and all experiments were conducted under a protocol approved

101 by the Johns Hopkins Animal Care and Use Committee.

102 2.2. Genotyping and DNA extraction.

103 Genomic DNA was isolated from ear as previously described [16]. Primers used for ARflox genotyping

104 were AR-R AGCCTGTATACTCAGTTGGGG [25], AR-F AATGCATCACATTAAGTTGATACC, Alb-Cre bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

105 Mutant were Cre-F (CGACCAAGTGACAGCAATGCT), Cre-R (GGTGCTAACCAGCGTTTTCGT). PCR

106 products were resolved on a 1.5 % agarose gel, and DNA bands were visualized by ethidium bromide staining.

107 2.3. Metabolic Testing: GTT, ITT, PTT and GSIS.

108 Mice metabolic testing was performed as described previously [16]. Briefly, for glucose tolerance tests

109 (GTT) and glucose stimulated insulin secretion (GSIS), mice were fasted overnight (16 h) and received

110 intraperitoneal (i.p.) injections of 2 g/kg body weight (BW) glucose. For insulin tolerance tests, mice were

111 fasted for 7 hours ad received i.p. injections of 0.3 units/kg BW insulin. For pyruvate tolerance tests (PTT),

112 mice were fasted for 12 hours and received i.p. injections of 1 g/kg BW pyruvate (PTT). For the GSIS, blood

113 samples were obtained at time zero and at 30 minutes. Serum insulin and c-peptide levels were measured by

114 insulin or C-peptide ELISA respectively following the manufacture’s protocol (Sigma Aldrich, St. Louis, MO,

115 USA). Mouse metabolic testing and fasting methods followed established protocols [16,33,34].

116 2.4. Glucose infusion rate.

117 The euglycemic-hyperinsulinemic glucose clamp is the gold-standard for glucose homeostasis testing

118 (54). The procedure was performed as previously described [33,35,36]. Euglycemia (100-120 mg/dL) was

119 maintained by a variable continuous infusion of a 25% glucose solution adjusted every 10 min. The steady-state

120 glucose infusion rate (GIR) was determined in the last 30 minutes of the 120-minute study.

121 2.5. Metabolic hormones measurement.

122 Six to eight weeks after DHT or empty pellet insertion, serum was obtained from mice fasted for 7 hour

123 and analyzed for insulin, leptin, adiponectin, IL-6 and TNFα using Milliplex Map Mouse Serum Adipokine

124 Panel (Millipore catalogue # MADKMAG-71K) on a Luminex 200IS platform.

125 2.6. Echo MRI: Body Composition Analysis.

126 Eight to ten weeks post pellet insertion, non-fasted Con and LivARKO mice were placed in an Echo

127 MRI apparatus (EchoMRI LLC, Houston, TX) without being restrained or anesthetized to determine body

128 composition, measuring whole body fat, lean, free water, and total water masses in live mice.

129 2.7. qRT-PCR. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

130 RNA isolation was performed on collected tissues from WT-Con, WT-DHT, LivARKO-Con, and

131 LivARKO-DHT mice using Trizol (BioRad). Reverse was performed using iScript cDNA

132 synthesis kit (BioRad) and real time quantitative polymerase chain reaction (qRT-PCR) was performed using iQ

133 SYBR Green reagent (BioRad) and an iCycler iQ5 Q-PCR machine (BioRad). Primers were used as previously

134 reported [16].

135 2.8. Western Blot.

136 As described previously [16], liver tissues from mice fasted for 16 h were extracted, quantified via BCA

137 assay, and equivalent quantities of protein were separated via SDS-PAGE (Thermo Scientific, Waltham, MA)

138 and transferred to nitrocellulose membranes. The primary antibodies used were p-AKT, AKT, p-Foxo1 S256,

139 Foxo1, and actin from Cell Signaling Technology (Danvers, Massachusetts) and PI3K-p85, AR (N20) from

140 Santa Cruz Biotechnology (Dallas, Texas), Glucose 6 phosphatase catalytic subunit (G6PC) from Novus

141 Biotechnologicals (Centennial, CO), and phosphoenolpyruvate carboxykinase (PEPCK) from Abcam

142 (Cambridge, MA) all 1:1000 dilutions. The blots were then placed in secondary antibodies (goat anti mouse or

143 goat anti-rabbit, BioRad), and detected using enhanced chemiluminescence (Perkin Elmer Life Sciences,

144 Boston, MA) or Odyssey CLx Imaging System (LI-COR Biosciences, Lincoln, NE).

145 2.9. (IP) and PI3K Activity Assay.

146 Lysates from liver tissues were placed in solution with AR (Santa Cruz) antibodies overnight at 4 oC.

147 Protein G sepharose beads (Sigma) were then placed in the mixture for 4 hours at 4 oC. Samples were

148 immunoprecipitated as done previously [16]. Immunoprecipitates from pY (Cell Signaling Technology) were

149 subjected to a PI3K activity assay (catalog no. 17-493; Millipore), as detailed in the manufacturer’s protocol

150 manual.

151 2.10. Glucose Uptake Assay.

152 Glucose uptake assays were performed as described previously [28,37] with slight alterations. Mice

153 fasted overnight were euthanized and the tissues were harvested, incubated in Hepes buffered saline (HBS)

154 (plus 32 mM mannitol, 8 mM glucose, and 0.1% radio-immunoassay grade bovine serum albumin (BSA)) for

155 one hour. Then, the glucose uptake assays were performed by incubating the tissues in 12 nM insulin for 20 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

156 minutes at 30oC prior to incubation in 2DG transport media (HBS with 4 mM 2DG and 2 µCi/ml 3H-2DG) for

157 10 minutes at 30oC. The tissues were then washed twice with a 4 mM 2DG rinse (HBS with no radiolabel) for 5

158 minutes on ice (to remove extracellular 3H-2DG) and stored at -80oC for further analysis. Lysates were counted

159 on a LS6500 liquid scintillation counter (Beckman Coulter, Fullerton, CA) to determine disintegrations per

160 minute (dpm). The DPM were normalized to total protein content.

161 2.11. Female Mouse Primary Hepatocytes.

162 Primary hepatocytes were isolated from livers of 8-10 weeks old Control and LivARKO female mice

163 without DHT treatment. The reagents and materials used for female mouse hepatocyte isolation and culture

164 were described previously [32]. Isolated primary hepatocytes were plated at 0.8 × 106 cells per well of six-well

165 dish in Williams E supplemented with 10% FBS (Gibco). Four to six hours after plating, Fresh media with or

166 without 1 nM DHT was subsequently added for 24 h. Serum starvation was performed for the final 3 h of DHT

167 treatment. Media with or without 100 nM insulin were added for the final half hour of DHT treatment. BioRad

168 cell lysis was used to harvest the cells and the lysate was subsequently processed for western blotting.

169 2.12. Female Human Primary Hepatocytes. University of Maryland Institutional Review Board approved the

170 obtainment, use and distribution of human tissues at the University of Maryland Brain and Tissue Bank. After

171 an approved application for use of the hepatocytes, human hepatocytes (three females, detailed in Suppl. Table

172 1) were isolated from resection of liver specimens from the Brain and Tissue Bank by a modification of the

173 two-step collagenase digestion method [38,39]. The primary hepatocytes from each individual were treated in

174 two to four separate groups. Group 1-DHT: incubated for 24 hours with media containing vehicle, 0.6 nM, 1

175 nM, or 6 nM DHT and for the last 4 h of the 24 h incubation, medium was replaced by serum free media. Group

176 2- DHT+flutamide: the same treatments as in Group1, with 10 nM flutamide (AR inhibitor) added at

177 experiment start. Group 3-DHT+insulin: the same treatments as in Group 1, plus 100nM insulin added in the

178 last 15 min of the 24h. Group 4-DHT+flutamide+insulin: the same treatments as in Group 2, plus 100nM

179 insulin inthe last 30 min of the 24h incubation period. At the end of the experiment, all cells of each well were

180 harvested using cell lysis buffer containing protease and phosphatase inhibitors, and stored at -80oC until further

181 analysis. 0.6-1 nM DHT is equivalent to roughly 2-3 fold of women with PCOS compared that of unaffected bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

182 women as calculated in [14]. Thus, 6 nM is representative of a supra-pathophysiological dose, and 1nM would

183 be pathophysiological.

184 2.13. Statistical analysis.

185 Statistical analyses used were described in each individual figure legend. Some data were analyzed by

186 student t-tests. Some data were assessed by 2-way ANOVA with main effects of DHT treatment and genotype

187 assessed, or in some experiments DHT exposure and insulin treatment effects. All analyses were performed

188 using Prism software (GraphPad, Inc.). All results were expressed as means ± SEM. A value of p<0.05 was

189 defined as statistically significant.

190 3. Results

191 3.1. Deletion of AR in liver by conditional or adeno-associated virus knockout

192 To examine the role of hepatocyte AR in the pathophysiology of androgen excess in female mice, we

193 generated a hepatocyte-specific AR knockout mouse. To validate the absence of AR in liver, we measured AR

194 mRNA expression via qRT-PCR and protein expression by western blot. The AR mRNA levels in livers of mice

195 with developmental deletion of hepatic AR (ARfl/fl; Cre+/- or LivARKO-veh; Veh refers to vehicle-treated mice)

196 were significantly reduced, yet were not altered in skeletal muscles and gonadal adipose tissues compared with

197 control littermates (ARfl/fl or Con-veh; Figure 1 A-C). AR protein levels of LivARKO were barely detectable in

198 liver (Suppl. Figure 1A-B) compared to Con-veh mice. The AR protein levels were not altered in skeletal muscles

199 or gonadal adipose tissues (Suppl. Figure 1 A, C-D). Furthermore, the liver AR mRNA levels of adult ARfl/fl

200 mice injected with a liver-specific Cre driven adeno-associated virus (AAV8) (adLivARKO or adLivARKO-veh

201 mice) were significantly lower compared with liver AR mRNA levels of adult ARfl/fl mice injected with AAV8-

202 GFP (ad-Con or adCon-veh mice). adLivARKO AR mRNA levels were not altered in skeletal muscles and WAT

203 compared to ad-Con mice (Figure 1 D-F). Mice with genotyping of ARfl/fl; ARwt/wt and Alb-Cre+/- do not exhibit

204 different metabolic phenotype, thus many studies have ARfl/fl mice as control (wild type) comparisons[40]. Thus

205 we also used ARfl/fl littermates as controls in this study. Figure 1G details the naming convention we used for the

206 ARKO models in this study.

207 3.2. Deletion of AR in liver prevents DHT-induced insulin resistance and impaired glucose homeostasis bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

208 As previously reported[21], in basal conditions without DHT and fed normal chow, LivARKO-veh female

209 mice show similar metabolic function as Con-veh mice including normal glucose, insulin, and pyruvate tolerance

210 tests (Figure 2).

211 To determine the effect of hyperandrogenemia on female LivARKO mice, we implanted the mice with

212 slow releasing DHT pellets which were replaced every four weeks. Serum DHT levels were around two-fold

213 higher in mice with DHT pellets compared to mice with empty pellets as previously reported (Suppl. Figure 2A).

214 In the presence of hyperandrogenemia, LivARKO-DHT mice exhibited improved glucose tolerance, insulin

215 sensitivity, and pyruvate tolerance compared to Con-DHT mice (Figure 2). Lean mass, fat mass and body weight

216 (Figure 2G and H) were not different among groups.

217 LivARKO-veh mice had similar insulin concentrations as Con-veh mice at basal (0 min) and at 30 min

218 post glucose stimulation (Figure 3A); Insulin and c-peptide levels in LivARKO-DHT mice were significantly

219 lower than those of Con-DHT mice (Figure 3B and Suppl. Figure 2B). Improved glucose metabolism was further

220 confirmed by the improved glucose infusion rate and reduced hepatic glucose production (HPG) during

221 euglycemic hyperinsulinemic clamp (GIR, Suppl. Figure 2C-D) in LivARKO-DHT compared to Con-DHT mice.

222 HGP is not uncommon to have it negative[41,42] as far as insulin suppression of HGP is significantly greater in

223 LivARKO-DHT compared to Con-DHT. Hormones and metabolites commonly associated with insulin resistance

224 and obesity (i.e., leptin, IL-6, and -) showed similar patterns among Con-veh, LivARKO-

225 veh, Con-DHT, and LivARKO-DHT mice (Figure 3C-E).

226 We examined the consequences of adult-onset deletion of hepatic AR by tail vein injection of AAV8-Cre

227 concurrent with DHT insertion in ARfl/fl mice. Adult female mice injected with AAV8-GFP and treated with DHT

228 (adCon-DHT) showed impaired glucose tolerance (IGT) and impaired pyruvate tolerance (IPT), impaired insulin

229 tolerance (IIT) (Figure 4A-F) compared to adCon-veh mice. Concordantly, like LivARKO-DHT mice,

230 adLivARKO-DHT mice exhibited improved GTT and PTT compared to GFP-DHT mice. Additionally, adCon-

231 DHT mice had reduced glucose uptake compared to adCon-veh mice, but this DHT-induced reduction was

232 prevented in the adLivARKO-DHT mice (Figure 4G). bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

233 3.3. Deletion of hepatic AR ameliorates DHT-induced increase of gluconeogenesis and impaired insulin

234 signaling

235 AR is a , and the impaired glucose production may be due to the dysregulation of

236 gluconeogenic . Previously, we observed that female Con-DHT mice had higher expression of

237 gluconeogenic genes (Pck1, G6pc, and Foxo1) and of rate-limiting enzymes (PEPCK and G6Pase) for

238 glucose production [43] in fasted liver compared to Con-Veh female mice[16]. LivARKO-veh mice displayed

239 similar levels of gluconeogenic gene mRNA and PEPCK and G6Pase protein levels as Con-veh mice (Figure 5A,

240 C-E). However, LivARKO-DHT mice showed lower mRNA expression in gluconeogenic genes (Pck1, G6pc,

241 and Foxo1) (Figure 5B), and protein levels of PEPCK (relative fold change to Con-veh) compared to Con-DHT

242 (Figure 5F-G). We next sought to determine the impact of AR on hepatic insulin signaling in hyperandrogenemia

243 mice. Similar to previous studies[16,28], Con-DHT mice exhibit blunted insulin-stimulated of

244 AKT (Figure 6A-B; noted as insulin resistant) compared to Con-Veh mice. However, this DHT-induced insulin

245 resistance at the level of p-AKT was absent in the LivARKO-DHT mice. We previously observed that AR binds

246 the p85 regulatory subunit of PI3K in liver from Con-DHT mice and impairs insulin signaling at the PI3K

247 level[16]. AR-p85 binding was not observed in livers from LivARKO-DHT mice (Figure 6A, C), and liver PI3K

248 activity was maintained in female LivARKO-DHT mice (Figure 6D). The same phenomenon occurred in the

249 adLivARKO mouse model; DHT disruption of hepatic insulin signaling as indicated by lower insulin stimulated

250 p-AKT was prevented in adLivARKO-DHT mice (Suppl. Figure 4A-B).

251 Although low dose DHT lowered liver mRNA expression of insulin signaling genes (Pi3kca) in the fed

252 state[28], livers from LivARKO-DHT mice did not exhibit the DHT-induced decrease in Pi3kca (the regulatory

253 p85 subunit of PI3K), compared to livers from WT-DHT (Figure 6E).

254 3.4. Disruption of hepatic AR prevents DHT-induced insulin resistance in primary culture of female mouse

255 and or human hepatocytes

256 To evaluate if AR directly affects hepatocyte function to induce hepatic insulin resistance rather than

257 indirectly via systemic signaling molecules or tissues, we applied DHT to isolated primary hepatocytes from

258 Control and LivARKO mice ex vivo. As observed in vivo liver, insulin increased p-AKT and p-Foxo1 in bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

259 primary hepatocytes. DHT blunted this insulin-stimulated phosphorylation (Figure 7A), thus indicating DHT-

260 induced insulin resistance (marked red in the graphical figure). In primary hepatocytes derived from LivARKO

261 mice, DHT-induced blunting of insulin-stimulated p-AKT and p-Foxo1 was not observed, thus the cells

262 remained insulin-sensitive (marked green in the graphical figure; Figure 7B).

263 In human tissue, we examined the effect of DHT in human primary hepatocytes derived from three

264 individual female donors (Suppl. Figure 5). Insulin increased p-AKT and p-FOXO1 in the primary hepatocytes

265 derived from individual 1 (Suppl. Figure 4A-C) and individual 2 (Suppl. Figure 4D-E); and this insulin

266 stimulated increase in p-AKT and p-FOXO1 was significantly reduced in the presence of DHT (Suppl. Figure.

267 4A-E, marked insulin resistant in the graphical figures). This DHT-induced insulin resistance was prevented in

268 the presence of AR inhibitor, flutamide (Suppl. Figure. 4A-E, marked insulin sensitive in the graphical figure),

269 suggesting that AR-specific androgen induced insulin resistance is occurring in these human hepatocytes as

270 well. Human primary hepatocytes derived from individual 3 female did not exhibit this pattern (Suppl. Figure

271 4F-G). The n-value for primary, immortalized, and human derived cell culture is a topic of debate[44-47].

272 4. Discussion

273 Our study shows that hepatocyte AR signaling mediates hyperandrogenemia- induced hepatic insulin

274 resistance in female mice by altering hepatic insulin signaling and phosphorylation cascades to increase

275 gluconeogenesis and systemic glucose intolerance. Hepatocyte AR did not play a role in the developmental

276 formation of insulin signaling in the hepatocyte, as LivARKO-veh female had no metabolic phenotype under

277 physiological androgen levels. In hyperandrogenic states, female mice with developmental or postnatal deletion

278 of AR did not exhibit impaired metabolic function compared to female mice with intact AR.

279 AR signaling in multiple cells including the theca cell, gonadotroph, , and neurons have been

280 found to be involved in the pathophysiology of metabolic and ovarian dysfunction in mouse models of female

281 hyperandrogenemia[20,30,48]. That multiple cell types are involved is not surprising considering the

282 multiorgan dysfunction associated with hyperandrogenemia. In these murine models of hyperandrogenemia,

283 various degrees of hyperandrogenemia are achieved experimentally, which may account for the various degrees

284 of weight gain, body composition, and metabolic dysfunction observed among models. Our low dose DHT bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

285 model offers an opportunity to investigate metabolic effects of hyperandrogenemia in a lean state as body

286 weight, composition, and systemic markers of insulin resistance are similar among the groups (Figure 2 and 3)

287 during two to three months of DHT treatment.

288 Low dose DHT blunts insulin action and alters glucose uptake in liver and adipose tissues without

289 affecting insulin action in skeletal muscle [16,28]. The impaired glucose tolerance is not likely due to pancreatic

290 β cell insufficiency since there is higher basal insulin and C-peptide levels in low dose DHT versus vehicle

291 treated mice (Figure 3B and Suppl. Figure 2B). In contrast to our findings, female mice exposed to a higher

292 dose of DHT (4-fold higher than untreated) on a western (high carbohydrate and high fat) diet exhibit pancreatic

293 β cell dysfunction driven by β cell AR and central neuronal insulin resistance with resultant failure of insulin to

294 suppress hepatic glucose production [49,50]. In another study, female mice exposed to higher dose DHT (8-

295 fold increase in serum DHT levels)[51] than our study exhibit weight gain, adipocyte hypertrophy, and hepatic

296 steatosis that is mostly ameliorated by loss of AR (AR floxed on exon 3) in the adipocyte and neuron [48].

297 Unlike other models of DHT treatment, we do not observe weight gain, increased basal blood glucose levels or

298 hepatic steatosis [16,28]. This may be due to dose of DHT or exposure time to DHT. Combined, these studies

299 suggest that there is differential cellular sensitivity to AR receptor occupation, with the hepatocyte sensitive to a

300 lower amount of AR receptor signaling to invoke glucose mishandling while the beta cell require higher dose of

301 androgen to trigger cellular dysfunction via AR signaling.

302 These results are in agreement with data from lean women with hyperandrogenism due to PCOS [52].

303 Dunaif et al., found that baseline hepatic glucose production is much higher, and that hepatic glucose

304 production was less suppressed upon insulin infusion in lean PCOS women compared to lean non-PCOS

305 women, suggesting that hepatic glucose production in women with PCOS is influenced by other factors besides

306 insulin. Insulin levels of PCOS women are typically more than two-fold higher than non PCOS women [52].

307 Thus, lean PCOS women have greater hepatic glucose production upon fasting even with a higher circulating

308 insulin level. Like what is seen in lean PCOS women, the liver was insulin-resistant in our low dose DHT

309 mouse model. Impaired hepatic insulin signaling and glucose homeostasis contributions to the metabolic bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

310 phenotype of lean hyperandrogenic females is confirmed with the studies herein, as LivARKO-DHT mice do

311 not experiences DHT-disrupted glucose metabolism.

312 Androgen excess impairs glucose handling via direct AR actions in the hepatocyte (Figure 2). Hepatic

313 glucose production occurs via and gluconeogenic enzyme activity [53]. LivARKO mice did not

314 exhibit DHT-induced increase in mRNA and protein expression of the gluconeogenic genes Pck1 orG6pc

315 (Figure 5). Foxo1 is a gluconeogenic transcription factor and reducing Foxo1 expression will dampen

316 gluconeogenesis by reducing Pck1 and G6pc. AR inhibition of insulin-stimulated Foxo1 degradation results in

317 increased Foxo1 protein levels and consequent increased PEPCK and or G6Pase protein levels in liver of

318 control but not LivARKO mice (Figure 5 and [16]). The increased gluconeogenesis in control DHT treated

319 mice was not observed with an increase in higher fasting glucose levels (Figure 2) likely because the insulin

320 levels trended higher in control-DHT treated animals (Figure 3).

321 AR also directly modulates the hepatocellular insulin signaling cascade. Insulin-induced

322 phosphorylation of AKT is reduced in control mice pretreated with DHT (Figure 6A-B) while p-AKT protein

323 levels are preserved in LivARKO-DHT mice indicating that DHT modulates downstream insulin signaling

324 (Figure 6A-B). Immunoprecipitation of p85 and AR occurred in the presence of DHT with or without insulin

325 and did not occur in LivARKO mice, (Figure 6C), indicating that AR directly binds p85. Insulin induced

326 hepatocyte PI3K activity in LivARKO mice was unaltered by the presence of DHT (Figure 6D) in contrast to

327 control mice [16]. DHT alteration of insulin-induced phosphorylation of AKT and Foxo1 also occur in primary

328 hepatocytes from control but not LivARKO mice (Figure 7) indicating that AR action on insulin signaling is not

329 dependent upon other paracrine or endocrine signaling. Hyperandrogenism also disrupts insulin signaling in

330 female human liver as 2 of 3 human samples displayed differential insulin-induced phosphorylation of AKT and

331 FOXO1 in the presence of DHT. Differences in body mass index (BMI was only 19.5), race, and medical

332 history may account for the different pattern in the 3rd human sample. Our previous studies indicate that AR

333 binds directly to p85 in vitro, which in turn may blunt cytosolic signaling activity of PI3K and insulin induced

334 phosphorylation (16); disruption of p85α / p85β binding to p110 blocks docking of p110 to IRS1/2 thus

335 decreases PI3K activity [16,54]. Via modulating insulin induced phosphorylation cascades critical for bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

336 regulating gluconeogenesis and glycogenolysis, androgen action via the androgen receptor modulates the

337 hepatocyte response to insulin, leading to hepatocellular insulin resistance.

338 Azziz et al., [1] indicated that the mechanism of insulin resistance in obese PCOS women was increased

339 serine phosphorylation of insulin receptor and IRS1 in myocytes, which inhibits metabolic insulin action. Our

340 DHT mice showed similar body mass and body composition and suggest a different or additional mechanism in

341 the non-obese state involving the interaction of hepatic AR at the level of PI3K leading to hepatocyte insulin

342 resistance. Hyperandrogenic-induced alteration in hepatocyte lipid metabolism may also affect hepatocyte

343 cellular function and insulin signaling to account for some of the results observed. However, we have not noted

344 a difference in fat accumulation as assessed by oil-red-O staining in our low dose DHT model [16] suggesting

345 that hepatic steatosis is not a factor in the hepatocyte insulin resistance. It is possible that with longer DHT

346 treatment at this low dose, hepatic lipid metabolism may be altered further affecting hepatocellular function and

347 glucose handling. Subtle effects on lipogenesis and lipid handling may occur with the doses and duration of

348 DHT used in these experiments but are beyond the scope of the current manuscript. Nuclear steroid receptors

349 may regulate mitochondrial function and thus cellular ATP[55] to affect cellular function and/or enhance gene

350 transcription. We cannot exclude the possibility that the observed effect of AR on gluconeogenesis is in part

351 due to effects on mitochondrial function. Our in vitro studies indicate that AR directly binds enzymes in the

352 insulin signaling cascade, a task that would not be affected by mitochondrial function. Hyperandrogenemia-

353 induced inhibition of insulin signaling could also be regulated by other novel pathways such as the intracellular

354 tyrosine kinase C (Src) [56] and extracellular signal-regulated kinase (Erk1/2) [54,57], a focus of future study.

355 In conclusion, we show that androgen action via hepatocyte AR mediates hepatic insulin resistance and

356 glucose metabolism upon hyperandrogenemia in female mice (Figure 8) and murine and human primary

357 hepatocytes. Hepatic AR could thus be a target to direct therapies specifically for women with

358 hyperandrogenemia- associated altered glucose homeostasis.

359

360 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

361 Author Contributions

362 SA and SW contributed to the conceptual design, performance of experiments, interpretation, and

363 analysis of data, and writing and editing the manuscript. All other authors contributed to performing some of the

364 experiments, analyzing the corresponding data, and reviewing and editing the manuscript.

365 Acknowledgements

366 This work was supported by the National Institutes of Health (Grants R00-HD068130 04 and R01-

367 HD09551201A1 to S.W.). FMJ was supported by R01 grants DK074970 and DK107444, a Department of

368 Veterans Affairs Merit Review Award (#BX003725) and the Tulane Center of Excellence in Sex-Based

369 Biology & Medicine. NMH was supported by R01 HL124477 and a DOD Award (W81XWH-16-I-0509). SB

370 was supported by DK103710. Technical support was provided by the Integrated Physiology Core of the

371 Baltimore DRTC (P30DK079637). We thank Dr. Jun Luo (Johns Hopkins University School of Medicine) and

372 Dr. William Henry Walker (University of Pittsburg) for offering us the control and DNA binding mutant AR

373 plasmids in this experiment.

374 Conflict of Interest: The authors have nothing to disclose.

375

376 References

377 [1] Azziz, R., Carmina, E., Chen, Z., Dunaif, A., Laven, J.S., Legro, R.S., et al., 2016. Polycystic ovary 378 syndrome. Nature reviews.Disease primers 2:16057.

379 [2] Padmanabhan, V., Veiga-Lopez, A., 2013. Sheep models of polycystic ovary syndrome phenotype. 380 Molecular and cellular endocrinology 373:8-20.

381 [3] Navarro, G., Allard, C., Xu, W., Mauvais-Jarvis, F., 2015. The role of androgens in metabolism, obesity, 382 and diabetes in males and females. Obesity (Silver Spring, Md.) 23:713-719.

383 [4] Escobar-Morreale, H.F., Alvarez-Blasco, F., Botella-Carretero, J.I., Luque-Ramirez, M., 2014. The striking 384 similarities in the metabolic associations of female androgen excess and male androgen deficiency. Human 385 reproduction (Oxford, England) 29:2083-2091.

386 [5] Barber, T.M., Dimitriadis, G.K., Andreou, A., Franks, S., 2015. Polycystic ovary syndrome: insight into 387 pathogenesis and a common association with insulin resistance. Clinical medicine (London, England) 15 Suppl 388 6:s72-6.

389 [6] Ehrmann, D.A., 2012. Metabolic dysfunction in pcos: Relationship to obstructive sleep apnea. Steroids 390 77:290-294. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

391 [7] Stener-Victorin, E., Padmanabhan, V., Walters, K.A., Campbell, R.E., Benrick, A., Giacobini, P., et al., 392 2020. Animal Models to Understand the Etiology and Pathophysiology of Polycystic Ovary Syndrome. 393 Endocrine reviews 41:10.1210/endrev/bnaa010.

394 [8] Roland, A.V., Nunemaker, C.S., Keller, S.R., Moenter, S.M., 2010. Prenatal androgen exposure programs 395 metabolic dysfunction in female mice. The Journal of endocrinology 207:213-223.

396 [9] van Houten, E.L., Kramer, P., McLuskey, A., Karels, B., Themmen, A.P., Visser, J.A., 2012. Reproductive 397 and metabolic phenotype of a mouse model of PCOS. Endocrinology 153:2861-2869.

398 [10] Walters, K.A., Allan, C.M., Handelsman, D.J., 2012. Rodent models for human polycystic ovary 399 syndrome. Biology of reproduction 86:149, 1-12.

400 [11] Abbott, D.H., Bruns, C.R., Barnett, D.K., Dunaif, A., Goodfriend, T.L., Dumesic, D.A., et al., 2010. 401 Experimentally induced gestational androgen excess disrupts glucoregulation in rhesus monkey dams and their 402 female offspring. American journal of physiology.Endocrinology and metabolism 299:E741-51.

403 [12] van Houten, E.L., Visser, J.A., 2014. Mouse models to study polycystic ovary syndrome: a possible link 404 between metabolism and ovarian function?. Reproductive biology 14:32-43.

405 [13] Cardoso, R.C., Puttabyatappa, M., Padmanabhan, V., 2015. Steroidogenic versus Metabolic Programming 406 of Reproductive Neuroendocrine, Ovarian and Metabolic Dysfunctions. Neuroendocrinology 102:226-237.

407 [14] Silfen, M.E., Denburg, M.R., Manibo, A.M., Lobo, R.A., Jaffe, R., Ferin, M., et al., 2003. Early endocrine, 408 metabolic, and sonographic characteristics of polycystic ovary syndrome (PCOS): comparison between 409 nonobese and obese adolescents. The Journal of clinical endocrinology and metabolism 88:4682-4688.

410 [15] Ambroziak, U., Kepczynska-Nyk, A., Kurylowicz, A., Malunowicz, E.M., Wojcicka, A., Miskiewicz, P., et 411 al., 2016. The diagnosis of nonclassic congenital adrenal hyperplasia due to 21-hydroxylase deficiency, based 412 on serum basal or post-ACTH stimulation 17-hydroxyprogesterone, can lead to false-positive diagnosis. 413 Clinical endocrinology 84:23-29.

414 [16] Andrisse, S., Childress, S., Ma, Y., Billings, K., Chen, Y., Xue, P., et al., 2016. Low Dose 415 Dihydrotestosterone Drives Metabolic Dysfunction via Cytosolic and Nuclear Hepatic Androgen Receptor 416 Mechanisms. Endocrinology:en20161553.

417 [17] Manneras, L., Cajander, S., Holmang, A., Seleskovic, Z., Lystig, T., Lonn, M., et al., 2007. A new rat 418 model exhibiting both ovarian and metabolic characteristics of polycystic ovary syndrome. Endocrinology 419 148:3781-3791.

420 [18] Lai, H., Jia, X., Yu, Q., Zhang, C., Qiao, J., Guan, Y., et al., 2014. High-fat diet induces significant 421 metabolic disorders in a mouse model of polycystic ovary syndrome. Biology of reproduction 91:127.

422 [19] Marino, J.S., Iler, J., Dowling, A.R., Chua, S., Bruning, J.C., Coppari, R., et al., 2012. Adipocyte 423 dysfunction in a mouse model of polycystic ovary syndrome (PCOS): evidence of adipocyte hypertrophy and 424 tissue-specific . PloS one 7:e48643.

425 [20] Ma, Y., Andrisse, S., Chen, Y., Childress, S., Xue, P., Wang, Z., et al., 2016. Androgen Receptor in the 426 Ovary Theca Cells Plays a Critical Role in Androgen-Induced Reproductive Dysfunction. 427 Endocrinology:en20161608. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

428 [21] Lin, H.Y., Yu, I.C., Wang, R.S., Chen, Y.T., Liu, N.C., Altuwaijri, S., et al., 2008. Increased hepatic 429 steatosis and insulin resistance in mice lacking hepatic androgen receptor. Hepatology (Baltimore, Md.) 430 47:1924-1935.

431 [22] Santoleri, D., Titchenell, P.M., 2019. Resolving the Paradox of Hepatic Insulin Resistance. Cellular and 432 molecular gastroenterology and hepatology 7:447-456.

433 [23] Vassilatou, E., 2014. Nonalcoholic fatty liver disease and polycystic ovary syndrome. World journal of 434 gastroenterology 20:8351-8363.

435 [24] Sanchez-Garrido, M.A., Tena-Sempere, M., 2020. Metabolic dysfunction in polycystic ovary syndrome: 436 Pathogenic role of androgen excess and potential therapeutic strategies. Molecular metabolism 35:100937.

437 [25] De Gendt, K., Swinnen, J.V., Saunders, P.T., Schoonjans, L., Dewerchin, M., Devos, A., et al., 2004. A 438 Sertoli cell-selective knockout of the androgen receptor causes spermatogenic arrest in meiosis. Proceedings of 439 the National Academy of Sciences of the United States of America 101:1327-1332.

440 [26] Yeh, S., Tsai, M.Y., Xu, Q., Mu, X.M., Lardy, H., Huang, K.E., et al., 2002. Generation and 441 characterization of androgen receptor knockout (ARKO) mice: an in vivo model for the study of androgen 442 functions in selective tissues. Proceedings of the National Academy of Sciences of the United States of America 443 99:13498-13503.

444 [27] Postic, C., Dentin, R., Girard, J., 2004. Role of the liver in the control of carbohydrate and lipid 445 homeostasis. Diabetes & metabolism 30:398-408.

446 [28] Andrisse, S., Billings, K., Xue, P., Wu, S., 2018. Insulin signaling displayed a differential tissue-specific 447 response to low-dose dihydrotestosterone in female mice. American journal of physiology.Endocrinology and 448 metabolism 314:E353-E365.

449 [29] Wang, Z., Shen, M., Xue, P., DiVall, S.A., Segars, J., Wu, S., 2018. Female Offspring From Chronic 450 Hyperandrogenemic Dams Exhibit Delayed Puberty and Impaired Ovarian Reserve. Endocrinology 159:1242- 451 1252.

452 [30] Wang, Z., Feng, M., Awe, O., Ma, Y., Shen, M., Xue, P., et al., 2019. Gonadotrope androgen receptor 453 mediates pituitary responsiveness to hormones and androgen-induced subfertility. JCI insight 454 5:10.1172/jci.insight.127817.

455 [31] Xue, P., Wang, Z., Fu, X., Wang, J., Punchhi, G., Wolfe, A., et al., 2018. A Hyperandrogenic Mouse 456 Model to Study Polycystic Ovary Syndrome. Journal of visualized experiments : JoVE (140). 457 doi:10.3791/58379.

458 [32] Qiu, S., Vazquez, J.T., Boulger, E., Liu, H., Xue, P., Hussain, M.A., et al., 2017. Hepatic 459 alpha is critical for regulation of gluconeogenesis and lipid metabolism in males. Scientific reports 7:1661-017- 460 01937-4.

461 [33] Ayala, J.E., Samuel, V.T., Morton, G.J., Obici, S., Croniger, C.M., Shulman, G.I., et al., 2010. Standard 462 operating procedures for describing and performing metabolic tests of glucose homeostasis in mice. Disease 463 models & mechanisms 3:525-534.

464 [34] Jensen, T.L., Kiersgaard, M.K., Sorensen, D.B., Mikkelsen, L.F., 2013. Fasting of mice: a review. 465 Laboratory animals 47:225-240. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

466 [35] Li, L., de La Serre, C.B., Zhang, N., Yang, L., Li, H., Bi, S., 2016. Knockdown of Neuropeptide Y in the 467 Dorsomedial Hypothalamus Promotes Hepatic Insulin Sensitivity in Male Rats. Endocrinology 157:4842-4852.

468 [36] Lee, B., Qiao, L., Lu, M., Yoo, H.S., Cheung, W., Mak, R., et al., 2014. C/EBPalpha regulates macrophage 469 activation and systemic metabolism. American journal of physiology.Endocrinology and metabolism 470 306:E1144-54.

471 [37] Andrisse, S., Patel, G.D., Chen, J.E., Webber, A.M., Spears, L.D., Koehler, R.M., et al., 2013. ATM and 472 GLUT1-S490 phosphorylation regulate GLUT1 mediated transport in skeletal muscle. PloS one 8:e66027.

473 [38] LeCluyse, E.L., Alexandre, E., Hamilton, G.A., Viollon-Abadie, C., Coon, D.J., Jolley, S., et al., 2005. 474 Isolation and culture of primary human hepatocytes. Methods in molecular biology (Clifton, N.J.) 290:207-229.

475 [39] Lecluyse, E.L., Alexandre, E., 2010. Isolation and culture of primary hepatocytes from resected human 476 liver tissue. Methods in molecular biology (Clifton, N.J.) 640:57-82.

477 [40] Chang, C., Yeh, S., Lee, S.O., Chang, T.M., 2013. Androgen receptor (AR) pathophysiological roles in 478 androgen-related diseases in skin, bone/muscle, metabolic syndrome and neuron/immune systems: lessons 479 learned from mice lacking AR in specific cells. signaling 11:e001.

480 [41] Allard, C., Morford, J.J., Xu, B., Salwen, B., Xu, W., Desmoulins, L., et al., 2019. Loss of Nuclear and 481 Membrane Estrogen Receptor-alpha Differentially Impairs Insulin Secretion and Action in Male and Female 482 Mice. Diabetes 68:490-501.

483 [42] Batista, T.M., Dagdeviren, S., Carroll, S.H., Cai, W., Melnik, V.Y., Noh, H.L., et al., 2020. Arrestin 484 domain-containing 3 (Arrdc3) modulates insulin action and glucose metabolism in liver. Proceedings of the 485 National Academy of Sciences of the United States of America 117:6733-6740.

486 [43] Jitrapakdee, S., 2012. Transcription factors and coactivators controlling nutrient and hormonal regulation 487 of hepatic gluconeogenesis. The international journal of biochemistry & cell biology 44:33-45.

488 [44] Lazic, S.E., Clarke-Williams, C.J., Munafo, M.R., 2018. What exactly is 'N' in cell culture and animal 489 experiments?. PLoS biology 16:e2005282.

490 [45] Cumming, G., Fidler, F., Vaux, D.L., 2007. Error bars in experimental biology. The Journal of cell biology 491 177:7-11.

492 [46] Geraghty, R.J., Capes-Davis, A., Davis, J.M., Downward, J., Freshney, R.I., Knezevic, I., et al., 2014. 493 Guidelines for the use of cell lines in biomedical research. British journal of 111:1021-1046.

494 [47] Naegle, K., Gough, N.R., Yaffe, M.B., 2015. Criteria for biological reproducibility: what does "n" mean?. 495 Science signaling 8:fs7.

496 [48] Cox, M.J., Edwards, M.C., Rodriguez Paris, V., Aflatounian, A., Ledger, W.L., Gilchrist, R.B., et al., 2020. 497 Androgen Action in Adipose Tissue and the Brain are Key Mediators in the Development of PCOS Traits in a 498 Mouse Model. Endocrinology 161:10.1210/endocr/bqaa061.

499 [49] Xu, W., Morford, J., Mauvais-Jarvis, F., 2019. Emerging role of testosterone in pancreatic beta-cell 500 function and insulin secretion. The Journal of endocrinology. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

501 [50] Navarro, G., Allard, C., Morford, J.J., Xu, W., Liu, S., Molinas, A.J., et al., 2018. Androgen excess in 502 pancreatic beta cells and neurons predisposes female mice to . JCI insight 503 3:10.1172/jci.insight.98607. eCollection 2018 Jun 21.

504 [51] Caldwell, A.S., Middleton, L.J., Jimenez, M., Desai, R., McMahon, A.C., Allan, C.M., et al., 2014. 505 Characterization of reproductive, metabolic, and endocrine features of polycystic ovary syndrome in female 506 hyperandrogenic mouse models. Endocrinology 155:3146-3159.

507 [52] Dunaif, A., Segal, K.R., Shelley, D.R., Green, G., Dobrjansky, A., Licholai, T., 1992. Evidence for 508 distinctive and intrinsic defects in insulin action in polycystic ovary syndrome. Diabetes 41:1257-1266.

509 [53] van Schaftingen, E., Gerin, I., 2002. The glucose-6-phosphatase system. The Biochemical journal 362:513- 510 532.

511 [54] Taniguchi, C.M., Emanuelli, B., Kahn, C.R., 2006. Critical nodes in signalling pathways: insights into 512 insulin action. Nature reviews.Molecular cell biology 7:85-96.

513 [55] Kobayashi, A., Azuma, K., Ikeda, K., Inoue, S., 2020. Mechanisms Underlying the Regulation of 514 Mitochondrial Respiratory Chain Complexes by Nuclear Steroid Receptors. International journal of molecular 515 sciences 21:10.3390/ijms21186683.

516 [56] Levin, E.R., Hammes, S.R., 2016. Nuclear receptors outside the nucleus: extranuclear signalling by steroid 517 receptors. Nature reviews.Molecular cell biology 17:783-797.

518 [57] Arkun, Y., 2016. Dynamic Modeling and Analysis of the Cross-Talk between Insulin/AKT and 519 MAPK/ERK Signaling Pathways. PloS one 11:e0149684.

520 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

521 Figure Legends

522 Figure 1 Androgen receptor (AR) is specifically deleted in liver. (A-C) AR mRNA levels of LivARKO

523 compared to control littermates, measured by qRT-PCR, in livers (A), gonadal adipose tissues (WAT, B) and

524 skeletal muscles (C). (D-F) AR mRNA levels in livers (D), WAT (E) and skeletal muscles (F) of adLivARKO

525 mice compared to GFP mice. Two-tailed Students t-tests were applied. P-values were stated in the graphs.

526 Values were mean±S.E.M.

527 (G) Naming and symbol convention used to describe and represent the animals in this study.

528 • Con-veh = Control mice implanted with an empty pellet represented by an empty circle in the data

529 figures.

530 • Con-DHT = Control mice implanted with a 4 mm DHT pellet represented by a filled circle.

531 • LivARKO-veh = mice with liver AR knocked out (KO) conventionally and implanted with an empty

532 pellet represented by an open square.

533 • LivARKO-DHT = mice with liver ARKO conventionally and implanted with a 4 mm DHT pellet

534 represented by a filled square.

535 Figure 2: Conditional KO of Liver AR prevents DHT-induced impaired glucose homeostasis. Control and

536 LivARKO mice, with (6-8 weeks after DHT insertion) and without DHT treatment were subjected to (A-B)

537 glucose tolerance test (GTT) (N = 4-6/group), (E-F) insulin tolerance test (ITT) (N = 4-14 /group), or (I-J)

538 Pyruvate tolerance test (PTT) (N = 6-13/group). Con-veh and LivARKO-veh (groups without DHT) are shown

539 on the left. Con-DHT and LivARKO-DHT (groups with DHT) are on the right. The area under the curve (AUC)

540 was determined for the (C-D) GTT, (G-H) ITT, and (K-L) PTT. Statistical analysis was performed for the line

541 graph at the same time point between groups and *=P<0.05, **=P<0.01. (M-N) Fat mass and lean mass were

542 determined via echo magnetic resonance imaging 8 weeks after insertion. (O-P) body weight among groups was

543 recorded (N= 5-10/group). Data were analyzed using two-tailed student’s t-tests. Values are means ±S.E.M.

544 Figure 3. Insulin and hormone levels

545 Insulin levels were measured from control and LivARKO mice without DHT (A) or at 6-8 weeks after DHT

546 insertion (B) after 16-hour fasted conditions at basal (0 min) and upon GSIS (glucose stimulated insulin bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

547 secretion at 30min after glucose injection). N = 5-24/group. Hormone levels (C) leptin, (D) IL-6, (E) TNF,

548 were measured after 7-hour fasted conditions. N=9-12/group. Data were compared by two-tailed student’s t-

549 tests. Values are means ±S.E.M. IL-6 = interleukin 6; TNF–tumor necrosis factor alpha.

550 Figure 4. Adult-onset KO of Liver AR prevents DHT-induced impaired glucose homeostasis. DHT and empty

551 pellets were inserted in conjunction with tail vein AAV injections. After 6 weeks post insertion, adCon-veh,

552 adCon-DHT and adLivARKO-DHT were subjected to (A) GTT (N = 3-4/group), (C) PTT (N = 4-6/group), and

553 (D) ITT (N=3-4/group); AUC was determined for the (B) GTT, (D) PTT and (F) ITT. (G). Livers were

554 subjected to ex vivo 3H-2-deoxy-glucose glucose transport assays (N=3-4/group). Statistical analysis was

555 performed for the line graph at the same time point using two-way ANOVA followed by Tukey’s multiple tests.

556 *=P<0.05, **=P<0.01, ***=P<0.001 (A,C: adCon-DHT vs adLivARKO-DHT and adCon-veh groups; E

557 adCon-DHT vs adCon-veh groups). Data (B,D,F,G) were analyzed using two-tailed student’s t-tests (adCon-

558 DHT vs adCon-veh or adLivARKO-DHT). Values are mean±S.E.M.

559 Figure 5. Conditional KO of Liver AR prevents DHT-induced increased expression of gluconeogenic

560 mRNA and proteins. At 10-12 weeks post insertion of pellet, livers of fasted control and LivARKO, with and

561 without DHT mice were taken and processed. (A-B) qRT-PCR analysis to determine mRNA expression of

562 Pck1, G6pc, Foxo1 was performed, where (A) shows Con-veh vs. LivARKO-veh and (B) shows Con-DHT vs.

563 LivARKO-DHT. Western blot analysis probing for antibodies against PEPCK and G6Pase and graphical

564 representations of the densitometry, where (C-E) show Con-veh vs. LivARKO-veh and (F-H) show Con-DHT

565 vs. LivARKO-DHT (data are relative fold change to Con-veh). Two-tailed student’s t-tests were used to analyze

566 the data. Values are means ±S.E.M. Pck1 = PEPCK, phosphoenolpyruvate carboxykinase; G6pc, glucose 6

567 phosphatase catalytic subunit; Foxo1, .

568 Figure 6. Conditional KO of Liver AR prevents DHT-induced impaired insulin signaling in liver. After 10

569 weeks post insertion, control and LivARKO mice with or without DHT were fasted for 16 hours and then

570 injected with saline or 0.5 U/kg insulin. After 10 minutes injection, liver samples were collected and subjected

571 to (A) immunoprecipitation and/or western blot analysis. Graphical representations of densitometry for control

572 and LivARKO are shown: (B) p-AKT (S473)/AKT. (C) IP: AR, IB: p85 (N=3-4/group). (D) Liver lysates from bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

573 LivARKO mice were immunoprecipitated using p-Y antibodies and the IP-pY was subjected to a PI3K activity

574 assay. (N=6/group). A different subset of liver tissues from fasted mice were subjected to (E) qRT-PCR analysis

575 for Pi3kca mRNA expression, n = 3-4 per group. Data in this figure were assessed by two-way ANOVA with

576 Tukey’s multiple comparisons tests. Values are means ±S.E.M. Different letters represent statistical difference,

577 P < 0.05. Veh: vehicle; D: DHT; I: insulin; IP: immunoprecipitation; IB: immunoblot.

578 Figure 7. Prevention of DHT-induced insulin resistance was recapitulated in primary hepatocytes from

579 Control and LivARKO female mice. Adult primary hepatocytes were extracted from (A) Control and (B)

580 LivARKO female mice, cultured for 24 hours, and subjected to a cell culture as detailed in the methods.

581 Western blot analysis was performed using antibodies against p-AKT, AKT, p-Foxo1, Foxo1, and Actin.

582 Graphical figures for the densitometry of the blot images are shown below the images. Data were assessed by

583 two-way ANOVA followed by Tukey’s multiple comparisons test. Values are means ±S.E.M. Different letter

584 represents statistical difference, P<0.05.

585 Figure 8. The effects of physiological and pathophysiological levels of androgen in female liver. Under

586 physiological levels of DHT (black line), insulin signaling and gluconeogenesis proceed normally. Under

587 pathophysiological levels of DHT (green line), cytosolic liver AR impairs insulin signaling and nuclear AR

588 directly regulates gluconeogenesis resulting in enhanced gluconeogenesis. This is associated with

589 hyperinsulinemia and systemic insulin resistance.

590 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 1

A B C Gonadal fat Liver Skeletal Muscle

) 2.5 NS h p<0.001

) 4

e l ) NS

v 4

o

h

-

r e

2.0 t

n

v

n

- o

o 3

n

A

C

C A

o 3

N

A

N o

1.5 C

o

t

t

R

N

R

o

e t

e 2

R

m

m

g

g

e 2 m

1.0 n

R

g

R

n

a

n

A

a

R A

Nearly Zero h

a

h c

1 A

Expression h c

d 1 c

0.5

l

d

o

d

l

f

l

(

o o

f 0

f

(

0.0 ( 0 Con-veh LivARKO-veh Con-veh LivARKO-veh Con-veh LivARKO-veh D E F Liver Gonadal fat Skeletal muscle 2.5

1.5 ) NS 4 )

) p<0.0001 NS

n

n

n

o

o

o

C C

C 2.0

d d

d 3

a

a

a

A

A

A

1.0

o

N

o

o

N

N

t t

t 1.5

R

R

R

e e

e 2

m

g

m

m

g

g

n n

n 1.0

R

R

R

a

a a

0.5 A

h

A A

h

h

c

c c

1

0.5

d

d

d

l

l

l

o

o

o

f

f

f ( ( 0.0 0.0 ( 0 adCon adLivARKO adCon adLivARKO adCon adLivARKO

G Naming Convention Used Genotype Control LivARKO

Vehicle Con-veh LivARKO-veh

DHT Con-DHT LivARKO-DHT Treatment bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 2 A Con-veh GTT livARKO-veh P<0.01 P<0.05 800 GTT B 60000 Con-DHT

LivARKO-DHT ) l 600 40000 NS

d *

/ C

g *

U

m

A

(

e 400 20000

s

o

c

u l

G 200 0

h h T T e e H H -v v D D n - - - o O n O C K o K 0 R C R A A iv iv 0 30 60 90 120 L L Min ITT Con-veh P<0.0001 P<0.0001 C 200 ITT D 8000 livARKO-veh Con-DHT

6000 )

l 160 LivARKO-DHT NS

d

/

C g

U 4000

m

A

(

e 120

s 2000

o

c

u l

G 80 0

h h T T e e H H -v v D D n - - - o O n O C K o K 40 R C R A A iv iv 0 30 60 90 120 L L Min PTT Con-veh F 25000 livARKO-veh P<0.001 P<0.001 250 PTT 20000 E Con-DHT LivARKO-DHT 15000 NS

200 C )

* U

l

A d

/ 10000 g ** **

m 150 (

5000

e s

o 100 0 c

u

l h T T e h v e H H

G - v D D n - - - o O n O 50 C K o K R C R A A iv iv L L 0 0 30 60 90 120 Body mass Min H 30 Body Composition Con-no DHT

G ) g

LivARKO-no DHT ( 20

100 Con-DHT s

s a

s LivARKO-DHT

s

m a

80

y

m

d

l 10

o a

t 60

B

o

t

f

o

t 40

n 0

e c

r 20 h h T T

e e e v v H H P - - -D -D n O n o K o O 0 C R C K A R v A Fat Mass Lean Mass i iv L L bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 3.

A Fasted Insulin Levels B Fasted Insulin Levels p<0.05 1.0 1.0 Con-veh Con-DHT LivARKO-veh 0.8 LivARKO-DHT

0.8 )

l p<0.01

)

l m

/ 0.6 m g NS

/ 0.6

n

g

(

n

(

n

i

l NS

n

i

u l

s 0.4

u

n s

I 0.4

n I

0.2 0.2

0.0 0.0 Fasted Fasted Fasted Fasted 0 min GSIS 30 min 0 min GSIS 30 min

D C Leptin IL-6 E TNFα 20 1200 NS 25 NS NS

1000 15 )

) 20

l

l

)

l

m

m /

/ 800

m

/

g

g n g 15

n 10

(

(

n

600 (

6

n

-

i

α

t

L I

F 10 p 400

e 5

N

L T 200 5 0 0 h h T T e e 0 h h T T v v H H e e - - D D v v H H n O T h h T T - - D D o K O e e H H n O T C W K -v -v D D o K O R R n C W K A A o O T O R R iv v K K A A L i C R W iv v L A R L i v A L i iv L L bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 4.

A a d C o n -D H T B GTT p<0.001 p<0.001 6 0 0 G T T a d L iv A R K O -D H T 60000 a d C o n - v e h

**

)

l d

/ 4 0 0 * 40000

g C

m *

(

U

e

A s

o 20000 c

u 2 0 0

l G

0 h T T 0 e v H H - -D -D 0 3 0 6 0 9 0 1 2 0 n n o o O C C K d d R M in a a A iv L d a

C D P T T a d C o n - D H T 2 5 0 P T T a d L iv A R K O -D H T 1 5 0 0 0 p < 0 .0 1 p < 0 .0 5 a d C o n - v e h 2 0 0

) *

l *

d /

g 1 0 0 0 0 1 5 0

m * **

(

C

e

U

s o

1 0 0 A

c u

l 5 0 0 0 G 5 0

0 0 0 3 0 6 0 9 0 1 2 0 h T T e H H v M in - D -D n - n O o o C K C R d d a a A iv L d E F a IT T 1 5 0 a d C o n - D H T I T T a d L iv A R K O -D H T 1 5 0 0 0

) a d C o n - v e h l

d P < 0 .0 1 N S /

g 1 0 0

m ** 1 0 0 0 0

(

e **

C

s U

o * A

c 5 0 u

l 5 0 0 0 G

0 0 0 3 0 6 0 9 0 1 2 0 h T T e H H v M in - D -D n - o n O o K C C d d R a a A iv L d G a L iv e r G lu c o s e U p ta k e

p = 0 .0 5

0 .2 5 p < 0 .0 5

e 0 .2 0

k

a

)

t l

p 0 .1 5

m

/

U

l

e

o

s m

o 0 .1 0

p

c

(

u l

G 0 .0 5

0 .0 0

h T T e H H v D - -D - n n K o o C R C A d d v a a i L d a bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 5.

A L iv e r m R N A E x p re s s io n B L iv e r m R N A E x p re s s io n 5 p < 0 .0 5 3 N S C o n -D H T C o n -v e h L ivA R K O -D H T L ivA R K O -ve h

l 4

N S

o l

N S r t o p < 0 .0 5

r p < 0 .0 5

n

t

o n

2 c

o

c 3

o

t

o

t

d

l

d

l

o

f

o

f

e 2

e

v

i

v

t

i t

1 a

l

a

l

e

e R

R 1

0 0

P c k 1 G 6 p c F o x o 1 P c k 1 G 6 p c F o x o 1 G lu c o n e o g e n ic G lu c o n e o g e n ic m R N A E x p re s s io n m R N A E x p re s s io n

D E ) G6Pase h C ) PEPCK e 1.5

h NS

v e 1.5 - Con-veh v

LivARKO-veh NS n

n

-

i

l

o

n

u

n

o

i

C

l

PEPCK b 1.0

C

u

o

u t

1.0 t

b

o

/

t

u

e

d

t

l

/

d

s

l

o

K a

G6Pase f

o 0.5

f

P

C

0.5 e

6

P

e

v

i

v

E

G

i

t

t P

β tubulin a

a l

t 0.0

e

a r

l 0.0

( h h e e e

r h v v e h - - ( e -v v n O n - o K o O C R C K A R iv A L iv L

F G H

PEPCK ) G6Pase

)

h h

Con-DHT LivARKO-DHT e 2.5

e 2.0

P<0.001 v -

V NS

-

n

n

i n

l 2.0

o

n o PEPCK i

1.5 u

t

C

C

b

c

o

u 1.5

A

o

t

t

/

t

/

K 1.0

e

d

d l

G6Pase l s

C 1.0

o

o

a

P

f

f

P

E e

0.5 e

6

P v

v 0.5

i

i

G

t t

Actin a

a l 0.0 l

e 0.0

e r

r ( T T ( T T H H H H -D D n - -D -D o O n O C K o R C K A R iv A L iv L bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 6.

A Con LivARKO B p -A K T /A K T In s u lin veh I D D+I veh I D D+I S e n s itive

) 1 0 0 In s u lin n T R e sista n t b P-AKT o

K 8 0

C

A -

/ b b

) T

AKT 3 6 0

W

7

4 o

Lysate

t

S

(

Actin 4 0

e T

v c

i K

t 2 0

a

A

l

- e

AR P a a a a IP AR: r

( 0

l

l

l

l

+

+

+

+

D H T

p85

IP AR:

l

l

l

l

+

+ + Ins +

C o n L iv A R K O

C IP :A R L iv A R K O IB :p 8 5 D P I3 K A c tiv ity A s s a y

) 1 2

) 4

l

n

y

e

i

o l

c a

e r

c b

s

t t 1 0 c i

s b

o h

n 3

r

A

e

o v

p 8

y

-

C

t

i

o P

t 2

v

I

o

i

t 6

t

e

l

c a

v a

a

e

i

a

t

t

v 4 1

i

a o

b K

l

t

T

3

e

a

/ I

l a N o E x p re s s io n R

5 2

P (

e 0

8

R p ( h T in in

0 e H l l

l

l l

l V u u

+

+ D +

+ s s

D H T

n

I In

+

T

l

l

l

l

+

+ + + H Ins D IP : p Y C o n L iv A R K O

L iv e r E P i3 k c a m R N A E x p re s s io n

5 c

l 4

o

r

t

n

o c

3

o

t

d

l

o f

a

e 2

v

i t

a a

l

e R 1 b

0

h n T T e o H H v - -C D D n - - o O T O C K W K R R A A v i iv L L bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 7.

Mouse Primary Hepatocytes A Mouse Primary Hepatocytes from Con B Mouse Primary Hepatocytes from LivARKO Veh Ins DHT DHT+Ins Veh Ins DHT DHT+Ins

P-AKT P-AKT

AKT AKT

P-Foxo1 P-FOXO1

Foxo1 FOXO1

Actin Actin

P h o s p h o /T o ta l P h o s p h o /T o ta l C o n L iv A R K O M o u s e P rim a ry H e p a to c y te s M o u s e P rim a ry H e p a to c y te s In s u lin In s u lin In s u lin R e sista n t R e sista n t S e n s itive 2 0

) 6 0

) l

b l

b b

o

l

o

l

r

r

t

a

t a

t In s u lin

t b

n 1 5

n

o o

o c S e n s itive o

T 4 0

T

/

C

/

C

o

o b

o 1 0 o

h c

t

h

t

p

p b

e

e

s

s v

v 2 0

i

o

i

o t

5 t

h

h

a

a

l P

a a l

P e a e a r a

r a a a (

0 ( 0

l l

l l

l

l

l

l

+ +

+ +

+

+

+

+

D H T

D H T

l l

l l

l

l

l

l

+ +

+ +

+

+ + Ins Ins +

P -A K T P -F O X O 1 P -A K T P -F O X O 1 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.09.447759; this version posted June 9, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 8.

Pancreas insulin Physiological levels of DHT Insulin Pathophysiological DHT DHT levels of DHT

DHT

DHT DHT DHT DHT PI3K DHT

AKT

DHT Glucose PEPCK1 Glucose uptake G6Pase production

Systematic metabolic dysfunction