Received: 25 September 2020 | Revised: 26 November 2020 | Accepted: 7 December 2020 DOI: 10.1096/fj.202002212R

RESEARCH ARTICLE

Maternal prevents visceral adiposity and adipocyte hypertrophy in prenatal androgenized female mice

Yanling Wu1 | Belén Chanclón1 | Peter Micallef1 | qElisabet Stener-Victorin2 | Ingrid Wernstedt Asterholm1 | Anna Benrick1,3

1Department of Physiology, Institute of Neuroscience and Physiology, Sahlgrenska Abstract Academy, University of Gothenburg, Hyperandrogenism is the main characteristic of polycystic syndrome, which Gothenburg, Sweden affects placental function and fetal growth, and leads to reproductive and metabolic 2 Department of Physiology and dysfunction in female offspring. Adiponectin acts on the and may exert Pharmacology, Karolinska Institute, Stockholm, Sweden endocrine effects on the developing fetus. This study aims to investigate if mater- 3School of Health Sciences, University of nal and/or fetal adiponectin can prevent metabolic and reproductive dysfunction in Skövde, Skövde, Sweden prenatal androgenized (PNA) female offspring. Adiponectin transgenic (APNtg) and wild-type dams received dihydrotestosterone/vehicle injections between gesta- Correspondence Anna Benrick, Department of Physiology, tional days 16.5-18.5 to induce PNA offspring, which were followed for 4 months. University of Gothenburg, Institute of Offspring from APNtg dams were smaller than offspring from wild-type dams, in- Neuroscience and Physiology, Box 423, 405 30 Gothenburg, Sweden. dependent of genotype. sensitivity was higher in wild-type mice from APNtg Email: [email protected] dams compared to wild-types from wild-type dams, and insulin sensitivity correlated with fat mass and adipocyte size. PNA increased visceral fat% and adipocyte size in Funding information Novo Nordisk Fonden (NNF), Grant/ wild-type offspring from wild-type dams, while wild-type and APNtg offspring from Award Number: NNF19OC0056601; APNtg dams were protected against this effect. APNtg mice had smaller adipocytes Diabetesfonden (Stiftelsen Svenska than wild-types and this morphology was associated with an increased expression Diabetesförbundets Forskningsfond), Grant/Award Number: DIA2019-419; of regulating adipogenesis (Ppard, Pparg, Cebpa, and Cebpb) and metabo- Magnus Bergvalls Stiftelse (Magnus lism (Chrebp and Lpl). Anogenital distance was increased in all PNA-exposed wild- Bergvall Foundation), Grant/Award Number: 2017-02069; Adlerbertska type offspring, but there was no increase in PNA APNtg offspring, suggesting that Stiftelserna (Adlerbertska Foundations), adiponectin overexpression protects against this effect. In conclusion, elevated adi- Grant/Award Number: GU2019/86; ponectin levels in utero improve insulin sensitivity, reduce body weight and fat mass Stiftelsen Handlanden Hjalmar Svenssons (Hjalmar Svensson Foundation), Grant/ gain in the adult offspring and protect against PNA-induced visceral adiposity. In Award Number: HJSV2019056; Kungl. conclusion, these data suggest that PNA offspring benefit from prenatal adiponectin Vetenskaps- och Vitterhets-Samhället i supplementation. Göteborg (KVVS), Grant/Award Number: 2018-217; Vetenskapsrådet (VR), Grant/ Award Number: 2017-00792, 2018-02435, KEYWORDS 2013-7107 and 2020-02435 adiponectin, adiposity, imprinting, PCOS, PNA

Abbreviations: AGD, anogenital distance; APNtg, adiponectin transgenic; DEXA, dual-energy x-ray absorptiometry; DHT, dihydrotestosterone; GD, gestational day; ITT, insulin tolerance test; PCOS, polycystic ovary syndrome; PNA, prenatal androgenized; wt, wild-type. Anna Benrick and Ingrid Wernstedt Asterholm jointly supervised this work.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2020 The Authors. The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology.

wileyonlinelibrary.com/journal/fsb2 | 1 of 13  The FASEB Journal. 2021;35:e21299. 2 of 13 | WU et al. 1 | INTRODUCTION chronically androgen exposed PCOS model.19 Interestingly, female adiponectin transgenic mice of this model have lower The prevalence of polycystic ovary syndrome (PCOS) ranges and higher levels compared to controls, from 4% to 21% depending on the diagnostic criteria used, supporting a direct effect of adiponectin on sex ste- and the syndrome causes reproductive and metabolic disease roidogenesis in the .19,20 This is also supported by in in women of reproductive age.1 Both clinical and preclinical vitro studies.21,22 Prenatally androgenized (PNA) mice are evidence pinpoint excessive androgen secretion as the under- another well-described PCOS-like mouse model in which lying cause of PCOS,2 and we have previously shown that the mice are exposed to androgens in utero, during late ges- daughters of women with PCOS are more likely to develop tation.23 PNA mice develop PCOS-like symptoms including the condition.3 Women with PCOS are also hyperandrogenic polycystic ovaries, irregular cycles, increased adipocyte size, during pregnancy,4-6 and have an increased risk of compli- and impaired glucose tolerance.24-26 Anogenital distance is cations during pregnancy and/or delivery, for example, pre- increased in PNA offspring and can be used as an sign of eclampsia and gestational diabetes.7 PCOS is associated with prenatal androgenization in utero in mice, and correlates with placental dysfunction5 that likely impacts fetal development androgen-dependent physiological effects in adulthood.27 and programing, and women with PCOS often deliver babies Adiponectin receptors are expressed in the placenta and adi- with a large or small birth weight for their gestational age.8 ponectin supplementation to pregnant dams during the rapid Birth weight is a sensitive marker of the overall impact of the fetal growth phase decreases glucose transport to the fetus intrauterine environment on fetal development and reduced or by inhibiting mTOR and insulin signaling in the placenta.8,28 excessive birth weight indicates an increased risk of the baby Lack of adiponectin, on the contrary, increases fetal growth developing obesity, diabetes, cardiovascular disease, and/or in mice.29 As adiponectin affects placental nutrient transport PCOS in adult life 9. The mechanisms underlying PCOS are during pregnancy, this allows for speculation that adiponectin poorly understood. However, it has been proposed that the may exert endocrine effects on the developing fetus during disease may be caused by an elevated exposure of circulat- pregnancy, and protect against the effects of prenatal andro- ing androgens during pregnancy, causing an unfavorable in- gen exposure, on the offspring. This study aims to investigate trauterine environment that can impair embryo development if maternal and/or fetal adiponectin can prevent metabolic and tissue patterning.10 and reproductive dysfunction in PNA female offspring. Adiponectin is a hormone that is secreted from that acts to improve insulin sensitivity and whole-body metabolism. Women with PCOS have decreased circulating 2 | MATERIAL AND METHODS adiponectin which, together with increased adipocyte size, represent the strongest contributing factors to the high-risk 2.1 | Animals PCOS women of developing insulin resistance.11 Jin and colleagues suggest that maternal adiponectin levels affect Wild-type (wt) and adiponectin transgenic (APNtg) mice, neonatal inflammation through effects on the milk compo- on a C57BL/6J background,30 were maintained under stand- sition,12 but its role during pregnancy, and in reproductive ard housing conditions at the animal facility, Sahlgrenska function, is underexplored. Adiponectin receptors are widely Academy, University of Gothenburg. They had ad libitum ac- distributed throughout the hypothalamic-pituitary-ovarian cess to food and water under controlled conditions (12-hour axis and reproductive organs, including the placenta and light/dark cycle environment). All experiments were per- endometrium, enabling adiponectin to affect metabolism in formed with permission from the Animal Ethics Committee these tissues,13-15 and potentially affect steroidogenesis.16 of the University of Gothenburg, and were performed in ac- Adiponectin is decreased before and throughout pregnancies cordance with EU guidelines for the care and use of labora- in women with PCOS, and women with PCOS who develop tory animals (2010/63/EU). gestational diabetes have lower adiponectin levels during pregnancy than euglycemic women with PCOS17 suggest- ing that low adiponectin levels lead to an impaired capacity 2.2 | Study design to handle metabolic changes during pregnancy. This is sup- ported by animal studies showing that pregnant adiponectin APNtg (n = 14) and wt (n = 11) female mice were mated with knockout mice are hyperlipidemic, with abnormal glucose wt males. Females in estrus phase, as determined by vaginal tolerance in late pregnancy.18 smears,31 were mated with a wt male overnight. Pregnancy We have shown that adiponectin overexpression pre- was confirmed by the presence of a post-copulation plug the vents the development of metabolic disorders in prepuber- following morning, and was this day was defined as gesta- tally androgenized PCOS-like mice.19 However, only minor tional day (GD) 0.5. On GD 16.5 the groups were subdivided improvement was seen on reproductive function in this and assigned to receive a 50 μL subcutaneous injection in WU et al. | 3 of 13 the interscapular area of 250 μg of 5α-Androstan-17β-ol-3- 2.4 | Dual-energy x-ray absorptiometry one (dihydrotestosterone (DHT), A8380, Sigma-Aldrich, St. Louis, USA), dissolved in a mixture of 2.5 μL benzyl benzo- Bone, fat, and lean body mass were determined using Lunar ate (B6630, Sigma-Aldrich) and 47.5 μL Sesame oil (S3547, PIXImus Mouse Densitometer (Wipro GE Healthcare, Sigma-Aldrich), or vehicle, for 3 days (until GD 18.5), to in- Madison, USA). Mice were anesthetized during the proce- duce a PCOS-like phenotype in the offspring.25 Adiponectin dure using isoflurane inhalation (2%; Isoba vet, Schering- transgenic dams gave birth to wt and APNtg offspring Plough, Uxbridge, UK).32 (Figure 1). Body weight and anogenital distance (AGD) were measured at 22 days of age27 (Figure 2A). Male offspring were killed and female offspring were weaned before geno- 2.5 | Tissue collection typing was performed as prevously described,19 generating six groups: wild-type offspring from wt dams with (wt-wt One week after the ITT, mice in diestrus were fasted for 3 PNA, n = 25) and without (wt-wt Veh, n = 14) DHT-induced hours before sedation with isoflurane and collection of serum PNA, wild-type offspring from APNtg dams with (APNtg-wt samples. Animals were decapitated and the pituitary, ovaries, PNA, n = 12) and without (APNtg-wt Veh, n = 11) DHT- uterus, white inguinal (subcutaneous) and gonadal (visceral) induced PNA, and APNtg offspring from APNtg dams with and brown adipose tissue, , and were dissected. (APNtg- APNtg PNA, n = 10) and without (APNtg- APNtg Veh, n = 10) DHT-induced PNA. The breeding scheme and study design are shown in Figure 1. The estrous cyclicity was 2.6 | Histological methods determined via daily vaginal smears in 6-week-old female mice.31 One ovary, and visceral (parametrial) fat, were fixed in Histofix (Histolab, Västra Frölunda, Sweden) for 48 hours, placed in 70% of ethanol, dehydrated, and embedded in par- 2.3 | Insulin tolerance test affin. The ovaries were sectioned at 3.5-4 µm. A 40 µm of tissue was discarded between each section. Four sections Intraperitoneal insulin tolerance test (ITT) was carried out in from each ovary was collected, mounted on a glass slide, 17-wk-old mice, fasted for 3 hours. Baseline blood glucose stained with hematoxylin and eosin, and analyzed using a and fasting insulin at 0 minute was taken from the tail before conventional light microscope. Definition of corpus luteum, the mice received 0.5 U/kg insulin (Actrapid, Novo Nordisk, atretic/cystic follicles, and antral follicles was done accord- Kalundborg, Denmark) dissolved in saline. Blood glucose ing to Kauffman et al.31 (n = 4/group). Visceral adipose was measured at 15, 30, and 45 minutes, using a glucose tissue was embedded in paraffin, sectioned at 7 μm, and analyzer (Contour Next, Bayer Health Care, Mishawaka, stained with hematoxylin and eosin. Six representative mi- USA). crographs were taken per sample at a 20x magnification with

FIGURE 1 Study design for the development of the prenatal androgenized offspring. A, Breeding scheme of wild-type (wt) and adiponectin transgenic (APNtg) dams mated with wt males. Females are shown in orange and males in blue. The different groups show the genotype of the dam, followed by the offspring genotype. B, Study design. Veh; vehicle, PNA; prenatal androgenization, GD; gestational day, DHT; dihydrotestosterone, ITT; insulin tolerance test, DEXA; dual-energy x-ray absorptiometry 4 of 13 | WU et al.

FIGURE 2 Body weight and adiposity in female PNA-exposed wt and APNtg offspring. A, Body weight, B, subcutaneous and C, visceral fat mass (n = 11-24/group), D, visceral adipocyte size (n = 4-5/group) in wild-type (wt) and adiponectin transgenic (APNtg) mice with and without prenatal androgenization (PNA). E, Representative histological pictures of adipose tissue, 20x magnification, scale bar: 100 µm. Data were analyzed using two-way ANOVA followed by Tukey post hoc test, a; main effect of DHT-exposure (PNA) and b; main effect of offspring group (wt-wt, APNtg-wt, and APNtg-APNtg). There was no significant interaction effect. *P < .05, **P < .01 for the effect of DHT-exposure within groups as analyzed by unpaired t test. Data are presented as mean ± SEM. Veh; vehicle a light microscope (TrimScope, LaVision BioTec, Bielefeld, °C for 20 seconds, 95°C for 15 seconds (40 cycles), followed Germany). The adipocyte size was quantified using ImageJ by a melting curve (0.5°C/s) from 60°C to 95°C to confirm software as previously described (n = 4-5/group).19 one PCR product. Real-time PCR products were detected using Fast SYBR Green Master Mix (4385612, Thermo Fisher Scientific), forward and reverse primers are found in 2.7 | Quantitative real-time PCR (Supplementary Table S1). The norm finder algorithm was used to determine the most stable reference in each tis- RNA isolation was performed using RNeasy Lipid Tissue sue.33 Hprt1 (visceral fat), Gapdh (ovary), and Rplp0 (pitui- Mini Kit (74804, Qiagen, Sollentuna, Sweden) for visceral tary) had the lowest inter- and intragroup variability in each fat and pituitary, and, RNeasy Fibrous Tissue Mini Kit tissue type. levels were calculated using the (74704, Qiagen), for ovarian tissue. Isolation was conducted 2–∆Ct method. according to the manufacturer´s instructions (n = 8/group). cDNA was prepared using High-capacity RNA-to-DNA Kit (4387406, Thermo Fisher Scientific, Gothenburg, Sweden). 2.8 | Serum measurements Real-time PCR was performed using Quant studio 7 flex (Applied Biosystems, Thermo Fisher Scientific). Conditions Fasted blood samples were taken to measure serum insu- were: 95°C for 20 seconds (1 cycle), 95°C for 1 second, 60 lin using ELISA (Mouse insulin 10-1247-01; Mercodia, WU et al. | 5 of 13 = (2,76) = 7.813) P = .001 = 10.206) P < .000 = 102.263) (2,76) (2,76) (2,76) ( F ( F ( F P < .000 27.954) P < .000 ns Offspring group ( F ns Offspring group ns Two-way-ANOVA Offspring group ns ns ns Offspring group 0.33 ± 0.31 0.44 ± 0.03 0.53 ± 0.07 a, c 0.033 ± 0.004 0.048 ± 0.002 a, c 3.5 ± 0.3 0.49 ± 0.05 21.4 ± 1.7 a APNtg-APNtg PNA (n = 10) 76.2 ± 1.1 15.4 ± 1.4 a 0.298 ± 0.09 0.42 ± 0.04 0.47 ± 0.10 a, c 0.031 ± 0.004 0.047 ± 0.002 a, c 3.7 ± 0.5 0.52 ± 0.07 21.3 ± 2.2 a APNtg-APNtg Veh (n = 10) 73.0 ± 5.4 16.5 ± 1,6 a 0.28 ± 0.09 0.45 ± 0.04 0.24 ± 0.08 0.030 ± 0.005 0.049 ± 0.002 3.6 ± 0.3 0.53 ± 0.06 22.2 ± 3.2 a APNtg-wt PNA (n = 12) 73.3 ± 2.6 19.9 ± 4.2 b 0.26 ± 0.09 0.46 ± 0.04 0.26 ± 0.09 0.032 ± 0.006 0.050 ± 0.002 3.4 ± 0.5 0.55 ± 0.14 22.6 ± 2.8 a APNtg-wt Veh (n = 11) 74.3 ± 2.0 17.3 ± 2.6 b 0.29 ± 0.09 0.44 ± 0.04 0.23 ± 0.04 0.030 ± 0.009 0.049 ± 0.002 3.7 ± 0.4 0.54 ± 0.21 26.9 ± 3.3 wt-wt PNA (n = 25) 72.5 ± 2.9 20.8 ± 4.4 0.39 ± 0.53 0.45 ± 0.03 0.23 ± 0.06 0.032 ± 0.005 0.050 ± 0.001 3.8 ± 0.4 0.54 ± 0.06 26.5 ± 2.9 wt-wt Veh (n = 14) 74.2 ± 3.4 18.8 ± 4.2 Body composition measured by DEXA and dissected tissue weights in 4.5 months old PNA-exposed control offspring from wt APNtg dams ) 2 BW) BW) BW) BW) BW) BMD (g/cm Uterus (% of Body weight (g) BMC (g) BAT (% of BW) Ovaries (% of Tissue Lean mass (% of Liver (% of BW) Pancreas (% of Fat mass (% of TABLE 1 Note: Data are presented as mean ± SD and analyzed with two-way ANOVA for the effect of PNA offspring groups (wt-wt, APNtg-wt, a nd APNtg-APNtg), Tukey post hoc comparisons differences between offspring groups a; P < .001 vs wt-wt, b; .05 c; .01 APNtg-wt. There was no significant main effect of PNA Abbreviation: ns, not statistically significant. 6 of 13 | WU et al. Sweden). Serum triglyceride levels were analyzed using DHT-exposure did not alter body weight gain within the TR210 kit (Randox, Crumlin, UK) with the following modifi- groups at 4 months of age (Table 1). Dissected tissue weights cations to the manufacturer’s instructions: 200 µL of reagent of liver, pancreas, ovaries, and uterus did not differ between R1 was added to each well containing either 2 µL serum, or groups, and was not affected by DHT-exposure (Table 1). The standard, in duplicates. The absorbance was measured using brown adipose tissue depot was larger in APNtg offspring 19 a plate reader (SoftMax pro 7.0). (F(2,76) = 102.263, P < .000), as previously reported. Body composition, measured by DEXA, showed no difference in lean body mass, or bone mineral content, between groups; 2.9 | Triglyceride liver content however, APNtg offspring had lower bone density compared to wt offspring (F(2,76) = 10.206, P < .000) (Table 1). The fat Liver triglycerides were extracted from ~50 mg of liver tis- mass % was lower in all offspring from APNtg dams compared sue after homogenization in a mixture of 5% IGEPAL (CA- to wt offspring from wt dams (F(2,76) = 7.813, P = .001), while 630, Sigma Aldrich) in distilled water. The samples were PNA had no effect (Table 1). Subcutaneous (inguinal) fat then heated to 90°C for 5 minutes, until the samples became showed was significantly different between all three offspring cloudy. Samples were then cooled to room temperature, groups (F(2,76) = 18.403, P < .000), where post hoc compari- heated for 5 minutes, cooled again, and then, centrifuged at sons showed that APNtg-APNtg mice had less subcutaneous 13,000 g for 2 minutes. The supernatant was removed, diluted fat compared to APNtg-wt mice (P = .03), and wt offspring x5, and analyzed using the TG kit TR210 (Randox, Crumlin, from APNtg dams had less subcutaneous fat compared to wt UK). A 200 µL of reagent R1 was then added to each well offspring from wt dams (P = .003) (Figure 2B). There was no containing either 2 µL supernatant, or standard, in duplicates. effect of PNA on subcutaneous fat mass. A similar pattern be- tween offspring groups was seen in the visceral (parametrial) depot (F(2,76) = 40.702, P < .000). However, in visceral fat 2.10 | Statistical analyses pads, PNA led to increased fat mass in wt mice from wt dams, whereas offspring from APNtg dams had no changes in fat Statistical analyses were performed with SPSS (version 19.0; mass weight between PNA and control (Figure 2C). The PNA- SPSS, Chicago, USA). The main effect of PNA-exposure induced increase in visceral fat mass was also associated with and offspring group (wt-wt, APNtg-wt, and APNtg-APNtg), an enlargement of the adipocytes in wt mice from wt dams as well as any interactions between PNA-exposure and off- (F(2,22) = 38.077, P < .000), while APNtg mice had signifi- spring groups, were measured using two-way ANOVA fol- cantly smaller adipocytes (P < .000 vs wt-wt and P = .003 lowed by Tukey post hoc test. To compare the effect of DHT vs APNtg-wt). Moreover, both wt and APNtg offspring from within genotypes, an independent samples t test was used. APNtg dams were protected against PNA-induced adipocyte Correlations between insulin sensitivity (AUC) and visceral hypertrophy (Figure 2D,E). There was no difference in serum fat mass, subcutaneous fat mass, and liver triglyceride levels triglyceride levels between groups, as expected in animals on were analyzed using Spearman correlations. Data were ex- normal chow (data not shown). pressed as mean ± SEM in figures, and mean ± SD in Tables, and P < .05 was considered significant. 3.2 | Insulin sensitivity 3 | RESULTS A two-way ANOVA was conducted to explore the impact of PNA and offspring group on glucose homeostasis. There was F P 3.1 | Body composition, tissue weights, and a significant main effect of group ( (2,76) = 6.150, = .003) adiposity and PNA (F(2,76) = 9.275, P = .003) on fasting glucose, and a significant main effect of offspring group (F(2,73) = 5.895, There was no difference in body weight between PNA and P = .004) and PNA (F(2,73) = 7.331, P = .008) on insulin levels control mice from wt dams at 3 weeks of age. Offspring from (Figure 3A,B). Post hoc comparisons indicated that glucose APNtg dams had a lower body weight, compared with off- and insulin were lower in the APNtg-APNtg group compared spring from wt dams, at 3 weeks of age, in agreement with Jin to wt-wt (P < .001, both). There was no effect of PNA on 12 et al, irrespective of offspring genotype (F(2,76) = 147.289, insulin sensitivity during ITT, but there was a significant P < .000) (Figure 2A). This reduction persisted after wean- main effect of group on insulin sensitivity (F(2,71) = 35.536, ing and was also present at 4 months of age (F(2,76) = 27.954, P < .000), as measured by AUC during the ITT (Figure 3C). P < .000) (Table 1). There was, however, no difference in APNtg offspring were the most sensitive to insulin (APNtg- body weight gain during this time (Supplemental Figure 1). APNtg vs APNtg-wt, P < .000), followed by wt offspring WU et al. | 7 of 13 Glucose homeostasis measures in female PNA-exposed wt and APNtg offspring. A, Fasted blood glucose (B) fasted insulin level s, wild-type (wt) adiponectin transgenic FIGURE 3 (APNtg) mice with and without prenatal androgenization (PNA). C, Area under the curve (AUC) for insulin tolerance test (ITT) a nd D, blood glucose during ITT (n = 11-20/group), E, correlations between insulin sensitivity and fat mass adipocyte size, F, liver triglyceride content. Data were analyzed using two-wa y ANOVA, a; main effect of DHT-exposure (PNA) b; of offspring group (wt-wt, APNtg-wt, and APNtg-APNtg). There was no interaction effect. ** P < .01 for the effect PNA-exposure within groups as analyzed by unpaired t test. Correlations were analyzed by Spearman correlations. Data are presented as mean ± SEM. Veh; vehicle 8 of 13 | WU et al.

FIGURE 4 Visceral fat gene expression in PNA-exposed wt and APNtg offspring. Genes involved in (A) adipogenesis and (B) metabolism, in wild-type (wt) and adiponectin transgenic (APNtg) mice with and without prenatal androgenization (PNA). Data were analyzed using two-way ANOVA, a; main effect of PNA exposure and b; main effect of offspring group (wt-wt, APNtg-wt, and APNtg-APNtg). *P < .05, **P < .01 for the effect of PNA-exposure within groups as analyzed by unpaired t test. Data are presented as mean ± SEM, n = 8/group. Veh; vehicle from APNtg dams (APNtg-wt vs wt-wt, P < .000) (Figure to metabolic functions and adipogenesis was investigated 3D). Insulin sensitivity correlated with adipocyte size, subcu- in parametrial fat. A two-way ANOVA was conducted to taneous, and visceral fat mass (P< .000) (Figure 3E), but not explore the impact of PNA and group (wt-wt, APNtg-wt, with liver triglyceride levels. There was no significant dif- and APNtg-APNtg) on visceral fat gene expression. There ference in liver triglyceride levels between offspring groups, was a significant main effect of group for genes involved and liver triglycerides were not altered by PNA exposure in in adipogenesis: Ppard (F(2,41) = 7.052, P = .002), Pparg wt offspring, regardless of the dams’ genotype. There was, (F(2,41) = 30.322, P < .000), Cebpa (F(2,40) = 98.976, however, a small main stimulating effect of PNA exposure P < .000), and Cebpb (F(2,41) = 17.379, P < .000), where on liver triglyceride levels (F(2,57) = 6.822, P = .011), and APNtg mice had higher expression (Figure 4A), with no higher liver triglyceride levels in APNtg-APNtg PNA com- difference in Bmp4 and Zfp423 expression (Supplemental pared to APNtg-APNtg Veh (Figure 3F). Figure S2 ). Ppard and Cebpa were increased in PNA wt mice from wt dams while PNA did not alter the expression of these genes in wt and APNtg offspring from APNtg dams. All 3.3 | Visceral adipose tissue gene expression investigated genes involved in metabolic functions showed a significant main effect of group, but not PNA exposure, when The largest effect on fat mass by PNA exposure was seen analyzed by two-way ANOVA, with increased expression in in visceral fat, therefore, gene expression of genes related APNtg offspring; Lpl (F(2,40) = 8.156, P = .001), Chrebp WU et al. | 9 of 13

FIGURE 5 Reproductive measures in female PNA-exposed wt and APNtg offspring. A, Anogenital distance (AGD) was measured at 22 days of age) in wild-type (wt) and adiponectin transgenic (APNtg) mice with and without prenatal androgenization (PNA) (n = 11-24/group). B, Estrous cyclicity determined by vaginal smears was classified as cyclic or acyclic. Ovarian histology was used to quantify the number of (C) corpus luteum, (D) antral follicles, and (E) cystic follicles (n =4 /group). ADG and histology measures were analyzed using two-way ANOVA, a; main effect of DHT-exposure (PNA) and b; main effect of offspring group (wt-wt, APNtg-wt, and APNtg-APNtg. *P < .05 for the effect of PNA-exposure within groups as analyzed by unpaired t test. Data are presented as mean ± SEM. Veh; vehicle, ns; not statistically significant

27 (F(2,41) = 19.882, P = .000), Irs1 (F(2,40) = 16.360, P < .000), measured as an indicator of androgen exposure in utero. and Insr (F(2,41) = 8.457, P = .001) (Figure 4B). The expres- As expected, female wt offspring exposed to PNA displayed sion of the pro-inflammatory gene Mcp1 was similar between larger anogenital distance compared with vehicle-exposed wt groups (Supplemental Figure 2). offspring (P = .023), with no impact of maternal genotype in APNtg-wt offspring (P = .034) (Figure 5A). There was also a significant main effect of group (F(2,57) = 7.648, P = .001), 3.4 | AGD and estrous cyclicity where APNtg offspring hade increased AGD (P = .002 vs wt-wt). However, this effect was primarily due to differences Number of mating attempts before pregnancy was compara- in body weight (Figure 2A), and without further effect of ble in wt and APNtg mice (2.0 ± 0.3 vs 1.9 ± 0.3, P = .86). DHT-exposure. We next assessed whether maternal andro- Wt and APNtg dams gave birth to a similar number of pups gen excess affected estrous cyclicity in 6-week-old female (wt-Veh 7.0 ± 0.4, APNtg-Veh 7.2 ± 0.9, P = .89), and offspring (Figure 5B). More than 90% of female wt-wt off- DHT-exposure did not affect the number of pups (wt-Veh spring exposed to PNA were acyclic in a chronic metestrus or 7.7 ± 1.2, APNtg-Veh 7.9 ± 0.8, P = .90). At 22 days of diestrus phase, while 50% of PNA-exposed wt offspring from age, anogenital distance (adjusted for body weight) was APNtg dams maintained estrous cyclicity. The majority of 10 of 13 | WU et al. APNtg-APNtg PNA offspring were acyclic but 30% ovulated neuroendocrine pathology of PCOS, with less pronounced (Figure 5B). metabolic effects.24,34-36 This mouse model was selected to specifically study the imprinting effect of adiponectin and androgens on reproductive and metabolic functions. PNA 3.5 | Ovary weight and morphology in late gestation enables understanding of androgen recep- tor-mediated imprinting mechanisms involved in PCOS The uterus and ovaries were collected in order to investi- pathophysiology. gate the effect of adiponectin on the reproductive organs in The reproductive phenotype in adult female PNA off- this PCOS-like mouse model induced by PNA exposure. spring includes impaired estrous cyclicity, and PCO- A two-way ANOVA showed no significant effect of PNA- morphology, exhibiting a decreased granulosa cell layer exposure or offspring group on ovary or uterus weight and increased theca cell layer thickness.24,34-36 In this study, (Table 1). As expected, there was a main effect of PNA impaired estrous cyclicity was manifested as acyclicity, as a exposure, with a lower number of corpus luteum in PNA- chronic pseudo-diestrus phase. PCO-morphology was man- exposed offspring (Figure 5C). PNA wt mice, irrespective ifested as increased number of cystic follicles and reduced of the dam’s genotype, had a significantly reduced number corpus luteum. Our findings showed less corpus luteum in of corpus luteum compared to vehicle-exposed mice, in- PNA-exposed female offspring, across all genotypes, sug- dicating less recent ovulations (Figure 5C). There was a gesting that adiponectin overexpression cannot restore ovu- small main effect of group on the number of antral follicles lation. But, APNtg-wt PNA mice had an increased number (F(2,57) = 3.781, P = .043), but no main effect of PNA ex- of corpus luteum compared to wt-wt PNA mice, which is in posure (Figure 5D). PNA increased the number of cystic line with an improved estrous cyclicity. And, interestingly, follicles in wt-wt mice, however, PNA exposure did not PNA-exposed offspring from APNtg dams had no, or very lead to the development of cystic follicles in offspring from few, cystic follicles. Adiponectin downregulates Bmp2, an APN-tg dams (Figure 5E), suggesting that maternal adi- apoptosis-related gene, in granulosa cells, suggesting that ponectin has a protective effect. high levels of adiponectin may help to protect granulosa cells from undergoing apoptosis, while it increases the ex- pression of ovulation-related genes in cumulus cell-oocyte 3.6 | Ovarian and pituitary gene expression complexes,37 supporting the hypothesis that adiponectin can modulate follicle growth. The presence of adiponectin recep- To determine if ovarian gene expression of enzymes in- tors in the ovaries during all periods of the estrous cycle indi- volved in sex hormone steroidogenesis were altered in PNA- cates the important role of this adipokine in the regulation of exposed offspring, we measured expression levels of Hsd3b, oocyte maturation, corpus luteum formation and activity, and Cyp17a1, Cyp11a1, Cyp19a1, and Pgr. There was no main a proper course of the estrous cycle via its influence on the effect of PNA exposure, or offspring group, on any of the in- sex hormone production. We speculate that the more regular vestigated genes (Table S1). To determine if altered pituitary ovulations in PNA wt offspring from APNtg dams has to do gene expression could help explain the reproductive effects with imprinting effects on the oocytes during the embryonic of PNA offspring in control and APNtg mice, we meas- development. ured expression levels of Lhb, Fshb, Gnrhr, and Pgr. There Functional ovarian hyperandrogenism has been identified was a main effect of offspring group on Gnrhr expression in some,25,35,38,39 but not all,24,25,40,41 publications utilizing (F(2,41) = 3.393, P = .043), where APNtg-APNtg offspring PNA-exposed offspring, an effect that seems to be age depen- had lower expression than wt-wt mice (P = .038). In addi- dent. Ovarian steroidogenesis has, to our knowledge not been tion, PNA in wt-wt offspring led to lower Gnrhr expression published on PNA offspring. We did not observe any changes compared with wt-wt Veh. There was no main effect of PNA- in Cyp11a1, Cyp17a1, Cyp19a1, or Hsd3b gene expression, exposure or offspring group on any of the other investigated suggesting a lack of ovarian hyperandrogenism. genes (Table S1). Previous findings strongly indicate that PNA exposure change the reproductive neuroendocrine system.35,36,41 Adult PNA wt mice exhibit increased GnRH neuron firing rate41 4 | DISCUSSION and increased LH pulsatility,36 but often unaltered LH lev- els.24,36,40,41 Hypothalamic adiponectin signaling, on the Animal models of relevance to PCOS must exhibit two or contrary, has been shown to decrease the GnRH neuronal more PCOS-like characteristics of the Rotterdam criteria to activity and to inhibit GnRH secretion, leading to impaired have comparability to women with PCOS.23 In mice, PNA- basal and GnRH-induced release of sex from the exposed female offspring represent a lean PCOS-like phe- ovaries.42 Pituitary gene expression was measured to investi- notype in adulthood by recapitulating the reproductive and gate if this could be associated with neuroendocrine defects WU et al. | 11 of 13 in PNA-exposed offspring. Unfortunately, the present gene PPARγ, in cooperation with adiponectin, induces adipo- expression studies did not add to our understanding of how genic programing during development, promotes adipose the components of the hypothalamic-pituitary-ovarian axis in remodeling in response to environmental changes, and reg- PNA offspring differ. In line with a recent publication, gene ulates adipocyte function, for example, lipid storage, adipo- expression levels of Lhb, Fshb, Egr1, Fst, Cga, Esr1, Ar, and kine secretion, and energy homeostasis.45 PNA increased Pgr were unaltered in the pituitary of control and PNA off- the expression of peroxisome proliferator-activated recep- spring.41 However, in the present study, PNA wt-wt offspring tor delta (Ppard) and CCAAT enhancer-binding had significantly lower levels of Gnrhr, possibly reflecting a alpha (Cebpa) in wt-wt PNA offspring, but had no effect on desensitization to an increased GnRH neuron firing rate. Pparg expression. These alterations are likely involved in PNA-exposed wt offspring present with a mild metabolic the mechanism leading to adipocyte hypertrophy in wt-wt phenotype, shown in this paper and by others, including in- PNA offspring. However, wt offspring from APNtg dams creased fasting glucose and insulin levels, but normal insulin (APNtg-wt PNA) had a gene expression pattern similar to sensitivity.24,25 APNtg mice on normal chow have previously wt-wt Veh offspring, suggesting a protective effect of ma- been shown to have an insulin sensitivity comparable to wild- ternal adiponectin levels during gestation. types,19,30 while in this study APNtg offspring were more Adiponectin receptors are expressed in the placenta and insulin sensitive than wt. Interestingly, wt offspring from adiponectin is known to affect nutrient transport and fetal APNtg dams had an improved insulin sensitivity compared to growth,8 allowing for developmental effects on the offspring. wt offspring from wt dams, suggesting an imprinting effect of Our work shows that elevated maternal adiponectin levels the elevated adiponectin levels in utero. The PNA-exposure both alter energy balance, as judged by lower body weight model has been shown to display hepatic steatosis with in- and fat mass in the adult female offspring, and protect the creased hepatic triglyceride content.24 Hepatic steatosis is a offspring against maternal androgen excess effects on the vis- factor that could contribute to insulin resistance. However, ceral adipose tissue. This suggests that offspring may benefit PNA exposure did not increase the liver triglyceride content from prenatal adiponectin supplementation. in wt offspring and, unlike fat mass, the liver triglyceride lev- els did not correlate with insulin sensitivity in this study. ACKNOWLEDGMENTS In accordance with previous studies, we found that We thank Philipp E. Scherer at the University of Texas wt-wt PNA offspring display normal body weight gain and Southwestern Medical Center, Dallas, Texas, for the kind body composition, but have increased visceral fat mass supply of APNtg mice. This work was supported by the with enlarged adipocytes.24,25,40 Maternal adiponectin Swedish Research Council (2017-00792, 2018-02435, overexpression led to decreased subcutaneous and visceral 2020-02485, and 2013-7107), the Novo Nordisk Foundation fat mass and protected against PNA-induced visceral fat (NNF19OC0056601), the Swedish Diabetes Foundation accumulation and adipocyte hypertrophy in wt offspring. (DIA2019-419), the Magnus Bergvall Foundation (2017- APNtg offspring were also protected against these effects, 02069), the Adlerbertska Foundation (GU2019/86), and the decrease in adipose tissue mass was even more Hjalmar Svensson Foundation (HJSV2019056), and Royal pronounced in APNtg-APNtg offspring. Our data suggest Society of Arts and Sciences in Gothenburg (2018-217). that these effects are associated with maternal imprinting The funders had no role in the study design, data collec- effects in all offspring from APNtg dams, with an addi- tion and analysis, decision to publish, or preparation of the tional effect of elevated adiponectin levels in the APNtg manuscript. offspring. Moreover, the body weight and inguinal fat mass 30 are unchanged in female APNtg offspring from wt dams CONFLICT OF INTERESTS indicating that the reduction in weight gain and subcutane- The authors declare that there are no conflicts of interest in ous adiposity in this study are primarily driven by maternal connection with the research described in this article. adiponectin. APNtg offspring have very small visceral ad- ipocytes compared to wt offspring and this striking mor- AUTHOR CONTRIBUTIONS phology was associated with an increased expression of A. Benrick designed the study, performed research, ana- genes involved in adipogenesis and metabolism in visceral lyzed and interpreted data, and wrote as well as edited the fat, for example, the peroxisome proliferator-activated re- manuscript; Y. Wu, B. Chanclón, and P. Micallef performed ceptors (PPARs) δ and γ. PPARδ plays an important role research and reviewed/edited the manuscript; E. Stener- in lipid metabolism and stimulates adipocyte prolifera- Victorin designed the study and reviewed/edited the manu- tion by controlling the induction of PPARγ, together with script; I W Asterholm analyzed and interpreted data and C/EBPβ.43 Moreover, adiponectin is believed to have direct reviewed/edited the manuscript. All authors approved the stimulating effects on PPARγ and C/EBPα expression.44 final version of this manuscript. 12 of 13 | WU et al. ORCID 16. Barbe A, Bongrani A, Mellouk N, et al. Mechanisms of adiponec- Elisabet Stener-Victorin https://orcid. tin action in fertility: an overview from gametogenesis to gestation in humans and animal models in normal and pathological condi- org/0000-0002-3424-1502 tions. Int J Mol Sci. 2019;20:1526. Ingrid Wernstedt Asterholm https://orcid. 17. Sir-Petermann T, Echiburu B, Maliqueo MM, et al. Serum adi- org/0000-0002-0755-5784 ponectin and lipid concentrations in pregnant women with poly- Anna Benrick https://orcid.org/0000-0003-4616-6789 cystic ovary syndrome. Hum Reprod. 2007;22:1830-1836. 18. Qiao L, Wattez JS, Lee S, et al. Adiponectin deficiency impairs REFERENCES maternal metabolic adaptation to pregnancy in mice. Diabetes. 1. Lizneva D, Suturina L, Walker W, Brakta S, Gavrilova-Jordan L, 2017;66:1126-1135. Azziz R. Criteria, prevalence, and phenotypes of polycystic ovary 19. Benrick A, Chanclon B, Micallef P, et al. Adiponectin protects syndrome. Fertil Steril. 2016;106:6-15. against development of metabolic disturbances in a PCOS mouse Proc Natl Acad Sci USA 2. Abbott DH, Dumesic DA, Levine JE. Hyperandrogenic origins of model. . 2017;114:E7187-E7196. polycystic ovary syndrome—implications for pathophysiology and 20. Maleszka A, Smolinska N, Nitkiewicz A, et al. Adiponectin expres- therapy. Expert Rev Endocrinol Metab. 2019;14:131-143. sion in the porcine ovary during the oestrous cycle and its effect on Int J Endocrinol 3. Risal S, Pei Y, Lu H, et al. Prenatal androgen exposure and trans- ovarian steroidogenesis. . 2014;2014:957076. generational susceptibility to polycystic ovary syndrome. Nat Med. 21. Chabrolle C, Tosca L, Rame C, Lecomte P, Royere D, Dupont J. 2019;25:1894-1904. Adiponectin increases insulin-like growth factor I-induced pro- Fertil 4. Maliqueo M, Lara HE, Sanchez F, Echiburu B, Crisosto N, Sir- gesterone and estradiol secretion in human granulosa cells. Steril Petermann T. Placental steroidogenesis in pregnant women with . 2009;92:1988-1996. polycystic ovary syndrome. Eur J Obstet Gynecol Reprod Biol. 22. Lagaly DV, Aad PY, Grado-Ahuir JA, Hulsey LB, Spicer LJ. Role 2013;166:151-155. of adiponectin in regulating ovarian theca and granulosa cell func- Mol Cell Endocrinol 5. Maliqueo M, Sundstrom Poromaa I, Vanky E, et al. Placental tion. . 2008;284:38-45. STAT3 signaling is activated in women with polycystic ovary syn- 23. Stener-Victorin E, Padmanabhan V, Walters KA, et al. Animal drome. Hum Reprod. 2015;30:692-700. models to understand the etiology and pathophysiology of polycys- Endocr Rev 6. Sir-Petermann T, Maliqueo M, Angel B, Lara HE, Perez-Bravo tic ovary syndrome. . 2020;41:538–576. F, Recabarren SE. Maternal serum androgens in pregnant women 24. Caldwell AS, Middleton LJ, Jimenez M, et al. Characterization with polycystic ovarian syndrome: possible implications in prena- of reproductive, metabolic, and endocrine features of polycys- tal androgenization. Hum Reprod. 2002;17:2573-2579. tic ovary syndrome in female hyperandrogenic mouse models. Endocrinology 7. Mills G, Badeghiesh A, Suarthana E, Baghlaf H, Dahan MH. Polycystic . 2014;155:3146-3159. ovary syndrome as an independent risk factor for gestational diabetes 25. Roland AV, Nunemaker CS, Keller SR, Moenter SM. Prenatal an- J and hypertensive disorders of pregnancy: a population-based study on drogen exposure programs metabolic dysfunction in female mice. Endocrinol 9.1 million pregnancies. Hum Reprod. 2020;35:1666-1674. . 2010;207:213-223. 8. Aye IL, Powell TL, Jansson T. Review: adiponectin—the miss- 26. Manti M, Fornes R, Qi XJ, et al. Maternal androgen excess and ing link between maternal adiposity, placental transport and fetal obesity induce sexually dimorphic anxiety-like behavior in the off- FASEB J growth? Placenta. 2013;34(Suppl):S40-S45. spring. . 2018;32:4158-4171. 9. Gluckman PD, Hanson MA, Cooper C, Thornburg KL. Effect of in 27. Manno FA 3rd. Measurement of the digit lengths and the anogeni- Physiol Behav utero and early-life conditions on adult health and disease. N Engl tal distance in mice. . 2008;93:364-368. J Med. 2008;359:61-73. 28. Aye IL, Rosario FJ, Powell TL, Jansson T. Adiponectin supplemen- 10. Abbott DH, Tarantal AF, Dumesic DA. Fetal, infant, adoles- tation in pregnant mice prevents the adverse effects of maternal Proc Natl Acad Sci cent and adult phenotypes of polycystic ovary syndrome in pre- obesity on placental function and fetal growth. USA natally androgenized female rhesus monkeys. Am J Primatol. . 2015;112:12858-12863. 2009;71:776-784. 29. Qiao L, Wattez JS, Lee S, et al. Knockout maternal adiponectin 11. Manneras-Holm L, Leonhardt H, Kullberg J, et al. Adipose tissue increases fetal growth in mice: potential role for trophoblast Diabetologia has aberrant morphology and function in PCOS: enlarged adipo- IGFBP-1. . 2016;59:2417–2425. cytes and low serum adiponectin, but not circulating sex steroids, 30. Asterholm IW, Scherer PE. Enhanced metabolic flexibil- Am J Pathol are strongly associated with insulin resistance. J Clin Endocrinol ity associated with elevated adiponectin levels. . Metab. 2011;96:E304-E311. 2010;176:1364-1376. 12. Jin Z, Du Y, Schwaid AG, et al. Maternal adiponectin controls milk 31. Kauffman AS, Thackray VG, Ryan GE, et al. A novel letrozole composition to prevent neonatal inflammation. Endocrinology. model recapitulates both the reproductive and metabolic pheno- Biol Reprod 2015;156:1504-1513. types of polycystic ovary syndrome in female mice. . 13. Palin MF, Bordignon VV, Murphy BD. Adiponectin and the control 2015;93:69. of female reproductive functions. Vitam Horm. 2012;90:239-287. 32. Nagy TR, Clair AL. Precision and accuracy of dual-energy X-ray 14. Takemura Y, Osuga Y, Yamauchi T, et al. Expression of adiponec- absorptiometry for determining in vivo body composition of mice. Obes Res tin receptors and its possible implication in the human endome- . 2000;8:392-398. trium. Endocrinology. 2006;147:3203-3210. 33. Andersen CL, Jensen JL, Orntoft TF. Normalization of real-time quan- 15. Dobrzyn K, Smolinska N, Kiezun M, et al. Adiponectin: a new titative reverse transcription-PCR data: a model-based variance esti- regulator of female reproductive system. Int J Endocrinol. mation approach to identify genes suited for normalization, applied to Cancer Res 2018;2018:7965071. bladder and colon cancer data sets. . 2004;64:5245-5250. WU et al. | 13 of 13 34. Sullivan SD, Moenter SM. Prenatal androgens alter GABAergic drive 42. Wen JP, Lv WS, Yang J, et al. Globular adiponectin inhibits GnRH to gonadotropin-releasing hormone neurons: implications for a com- secretion from GT1-7 hypothalamic GnRH neurons by induction mon fertility disorder. Proc Natl Acad Sci USA. 2004;101:7129-7134. of hyperpolarization of membrane potential. Biochem Biophys Res 35. Moore AM, Prescott M, Campbell RE. Estradiol negative and posi- Comm. 2008;371:756-761. tive feedback in a prenatal androgen-induced mouse model of poly- 43. Luquet S, Gaudel C, Holst D, et al. Roles of PPAR delta in lipid cystic ovarian syndrome. Endocrinology. 2013;154:796-806. absorption and metabolism: a new target for the treatment of type 2 36. Moore AM, Prescott M, Marshall CJ, Yip SH, Campbell RE. diabetes. Biochim Biophys Acta. 2005;1740:313-317. Enhancement of a robust arcuate GABAergic input to gonadotro- 44. Yang W, Yang C, Luo J, Wei Y, Wang W, Zhong Y. Adiponectin pin-releasing hormone neurons in a model of polycystic ovarian promotes preadipocyte differentiation via the PPARgamma path- syndrome. Proc Natl Acad Sci USA. 2015;112:596-601. way. Mol Med Rep. 2018;17:428-435. 37. Richards JS, Liu Z, Kawai T, et al. Adiponectin and its receptors 45. Astapova O, Leff T. Adiponectin and PPARgamma: cooperative modulate granulosa cell and cumulus cell functions, fertility, and and interdependent actions of two key regulators of metabolism. early embryo development in the mouse and human. Fertil Steril. Vitam Horm. 2012;90:143-162. 2012;98:471-479.e1. 38. Sullivan SD, Moenter SM. Prenatal androgens alter GABAergic drive to gonadotropin-releasing hormone neurons: implications for SUPPORTING INFORMATION a common fertility disorder. Proc Natl Acad Sci USA. 2004;101: Additional Supporting Information may be found online in 7129-7134. the Supporting Information section. 39. Witham EA, Meadows JD, Shojaei S, Kauffman AS, Mellon PL. Prenatal exposure to low levels of androgen accelerates female pu- Endocrinology berty onset and reproductive senescence in mice. . How to cite this article: Wu Y, Chanclón B, Micallef P, 2012;153:4522-4532. Stener-Victorin E, Wernstedt Asterholm I, Benrick A. 40. Caldwell AS, Eid S, Kay CR, et al. Haplosufficient genomic an- Maternal adiponectin prevents visceral adiposity and drogen receptor signaling is adequate to protect female mice from induction of polycystic ovary syndrome features by prenatal hyper- adipocyte hypertrophy in prenatal androgenized female androgenization. Endocrinology. 2015;156:1441-1452. mice. The FASEB Journal. 2021;35:e21299. https://doi. 41. Dulka EA, Burger LL, Moenter SM. Ovarian androgens maintain org/10.1096/fj.20200​2212R high GnRH neuron firing rate in adult prenatally-androgenized female mice. Endocrinology. 2020;161.