Sexual Dimorphism in Testosterone Programming of Cardiomyocyte Development in Sheep

: 26 Perturbed in-utero hormone milieu leads to intrauterine growth retardation (IUGR), a 27 known risk factor for left ventricular (LV) dysfunction later in life. Gestational 28 testosterone (T) excess predisposes offspring to IUGR and leads to LV myocardial 29 disarray and hypertension in adult females. However, the early impact of T excess on 30 LV programming and if it is female-specific is unknown. LV tissues were obtained at 31 day 90 gestation from days 30-90 T-treated or control fetuses (n=6/group/sex) and 32 morphometric and molecular analyses were conducted . Gestational T treatment 33 increased cardiomyocyte number only in female fetuses. T excess up-regulated 34 receptor expression of insulin and insulin-like growth factor. Furthermore, in a sex- 35 specific manner, T increased expression of Phosphatidylinositol 3-kinase (PI3K) while 36 down regulating phosphorylated mammalian target of rapamycin (pmTOR) /mTOR 37 ratio suggestive of compensatory response. T excess 1) upregulated atrial natriuretic 38 peptide (ANP) and brain natriuretic peptide (BNP), markers of stress and cardiac 39 hypertrophy 2) upregulated estrogen receptors1 (ESR1) and 2 (ESR2) but not in 40 androgen receptor (AR). Thus, gestational T excess upregulated markers of cardiac 41 stress and hypertrophy in both sexes while inducing cardiomyocyte hyperplasia only in 42 females, likely mediated via insulin and estrogenic programming.


46
CVD remains the leading cause of death worldwide, with a prevalence rate of 49.2% 47 increasing with age in males and females [1]. Pathological cardiac remodeling and left 48 ventricle hypertrophy (LVH), typically preceding progression to heart failure, account for 49 a substantial portion of CVD morbidity and mortality [2][3][4]. Despite significant advances 50 in identifying CVD risk factors and their therapeutics, the morbidity and mortality from 51 CVD remain high [5,6]. Of note, sex differences exist in both prevalence and burden of 52 CVD. Despite it being the leading cause of death in both men and women, men have a 53 higher age-adjusted rate of CVD mortality [7]. Men are more prone to pathological LV 54 remodeling, and heart failure outcomes are worse in men compared to women [8]. 55 Furthermore, with aging, there is a sex-specific decrease in cardiomyocyte number and 56 an increase in myocytes size, more pronounced in males than in females [9]. Of 57 relevance, alteration in sex steroid levels impacts cardiovascular function in a sex-58 specific manner [10]. While estrogen is considered cardioprotective in women [10], 59 androgens are linked to CVD risk in both sexes [11,12]. However, the underlying 60 molecular mechanisms and sex-specific impact of CVD-related pathological cardiac 61 remodeling and associated morbidity and mortality are not well understood. 62 Epidemiological data points to obesity [13], hyperglycemia [14], hypertension 63 [15], and physical inactivity [16] as major CVD risk factors that contribute to morbidity 64 and mortality [17]. Substantial evidence from human, animal, and epidemiological 65 studies indicate that early life insults in-utero lead to IUGR, adversely program the 66 cardiovascular system, thereby predisposing to CVD later in life [18][19][20], including LVH 67 [21,22]. Interestingly, many of these insults are associated with androgen excess [23, 68 24], a well-known risk factor for the development of pathological LVH and CVD in 69 offspring [25][26][27]. For instance, a recent epidemiological study noted that an elevated 70 third-trimester androgen level was positively correlated with a 4.84-fold increased risk of 71 hypertension in female offspring [28]. Moreover, higher fetal testosterone (T) levels in 72 late pregnancy were also associated with higher blood pressure in young adults [29]. 73 Similarly, offspring of hyperandrogenic women with PCOS were reported to have 74 alteration in blood pressure, left ventricular dilation, and increased carotid intima 75 thickness [30]. 76 Animal studies have corroborated the adverse impact of excess prenatal T 77 exposure on cardiovascular health. Using our well-characterized sheep model of 78 prenatal T excess [31], we previously found exposure to excess T during fetal life 79 results in mild hypertension and pathological cardiac remodeling in adult female 80 offspring [32,33] (the impact in male offspring remain to be determined). In rodent 81 studies, where both sexes were investigated, prenatal T excess induced hypertension in 82 both males and females with greater effect seen in males [34]. 83 While the detrimental effects of exposure to excess T during fetal life on 84 cardiometabolic phenotype are well-validated in several species [35][36][37], the early 85 perturbations leading to pathological cardiac remodeling and LVH are not well 86 understood. Relative to mediators of cardiovascular programming, estrogen and T play 87 a key role in cardiac growth [38,39] Considering CVD outcomes differ between men and women and the limited 110 information available on the sexually dimorphic impact of prenatal T excess, especially 111 the early cardiac perturbations that contribute to adverse programming of LVH, the 112 objective of this study was to investigate the impact of exposure to excess T from days 113 30-90 of gestation on sex-specific early cardiac effects using the sheep model. Because 114 T fetuses are exposed to high androgen, estrogen, insulin, and altered IGF during fetal 115 life, we hypothesized that the impact of prenatal T excess on LV remodeling involved 116 early changes in androgen/estrogen/insulin/IGF signaling pathways leading to 117 alterations in cardiomyocyte morphology. 118

Impact of gestational T excess on fetal body and heart weights: 120
Analyses of body weights revealed no sex (P=0.52), treatment (P=0.93), or sex by 121 treatment interaction (P=0.18). Two-way ANOVA showed no significant sex (P=0.7) or 122 treatment (P=0.56) effect in heart weights, although a trend (P=0.07) for sex by 123 treatment interaction was evident. This trend in interaction was reflected as a large 124 magnitude decrease in heart weight (Cohen's d=0.85) in T females compared to CON 125 females as opposed to a medium magnitude increase (Cohen's d=0.65) in heart weight 126 in T male fetuses compared to CON fetuses (Fig. 1B). However, analysis of heart 127 weight to body weight ratio revealed no effect of sex (P=0.92), treatment (P=0.28), or 128 sex by treatment interaction (p=0.15) (Fig. 1C). 129 130 Impact of gestational T excess on cardiomyocyte number, diameter, and collagen 131 content: Two-way ANOVA showed significant sex, treatment, and sex by treatment 132 interaction in cardiomyocyte number. T-treated females exhibited a significant increase 133 in cardiomyocyte number compared to control females ( Fig.2A, 2B). Cohens effect size 134 analysis also revealed a large magnitude increase in cardiomyocyte number in T-135 treated female fetuses (Cohen's d =2.93) but not male (Cohen's d =0.02). In contrast, 136 two-way ANOVA showed a significant sex effect but no treatment (P =0.11) or sex by 137 treatment interaction (P=0.17) in cardiomyocyte diameter (Fig.2C). There was no 138 significant impact of sex, treatment, and sex by treatment interaction in collagen 139 accumulation ( Fig.3A and B). 140 141

Impact of gestational T excess on steroid receptors 142
Two-way ANOVA of ESR1 found no sex effect (P =0.46), a significant T treatment 143 effect, and no significant sex by treatment interaction (P =0.38). Cohen's effect size 144 analysis revealed a large magnitude increase in T males (Cohen's d =0.99) and T 145 females (Cohen's d =1.73) relative to their sex-matched controls (Fig.4A). Similarly, 146 while there was no significant sex difference in ESR2 expression (P= 0.28), there was a 147 significant T treatment effect. Consistent with the lack of sex by treatment interaction, 148 Cohen's effect size analysis revealed an increase in ESR2 expression in T female 149 fetuses (Cohen's d = 1.6) as well as the T-treated males (Cohen's d = 1.03 ( Fig. 4B) 150 relative to their sex-matched controls. In contrast to effects of T treatment on ESR1 and 151 2 expressions, there were no significant sex (P =0.8), treatment (P =0.6), or interaction 152 effects (P =0.8) relative to AR expression (Fig. 4C). 153 154

Impact of gestational T excess on markers of LV hyperplasia and stress. 155
Two-way ANOVA of α-MHC gene expression showed no sex effect (P =0.2), a 156 significant T treatment effect, but no significant sex by treatment interaction (P =0.31). 157 Relative to treatment effects, Cohens effect analyses also revealed a large increase in 158 T-treated males (Cohen's d =1.15) and T females (Cohen's d =1.01) compared to their 159 sex-matched controls (Fig.5A). Similarly, as opposed to a lack of sex effect but a 160 treatment effect with α-MHC, there was a significant sex effect but no treatment (P 161 =0.44) effect with β-MHC expression. There was also no significant sex by treatment 162 interaction with both α-MHC (P =0.31) and β-MHC (P =0.21) expression (Fig.5B). There 163 were no sex (P =0.23), treatment (P =0.54) or sex by treatment interaction (P =0.86) in 164 β-MHC/ α-MHC expression ratio (Fig.5C). 165 ANP gene expression showed significant sex and treatment effect, although sex by 166 treatment interaction was not significant (P =0.10). Cohens effect size analyses 167 revealed a large effect size difference in ANP expression between control and T-treated 168 females (Cohen's d =1.44) that reached significance, but only a medium magnitude 169 difference (Cohen's d =0.73) between control and T-treated males that did not achieve 170 significance (Fig.5D). In contrast, BNP gene expression showed no sex effect (P = 171 0.21) although there was a significant treatment effect with Cohens effect analysis 172 revealing a large effect size increase in T-treated males (Cohen's d =1.27) and a 173 medium effect size increase (Cohen's d =0.5) in T-treated females (Fig.5E). There was 174 no sex by treatment interaction (P =0.22). 175 176

Impact of gestational T excess on insulin signaling mediators. 177
Two-way ANOVA showed a trend in sex (P= 0.058), a significant treatment, and a trend 178 in sex by treatment interaction (P= 0.067) in IR gene expression. Cohens effect size 179 analyses revealed a large magnitude increase (Cohen's d =1.40) in IR expression in T-180 treated female fetuses that achieved statistical significance by post hoc analyses and 181 only a medium magnitude increase (Cohen's d =0.63) in T-treated males relative to 182 corresponding controls that did not achieve significance (Fig.6A). IRS1 gene expression 183 showed no significant sex (P =0.65), treatment (P =0.45) or sex by treatment (P =0.69) 184 effects (Fig.6B). In contrast, IRS 2 showed no significant sex (P =0.11) or treatment (P 185 =0.29) effect, although a trend in sex by treatment effect was evident (Fig.6C). Cohens 186 effect analyses revealed a large effect size increase (Cohen's d =1.07) only in T-treated 187 females compared to controls that did not reach statistical significance due to the small 188 sample size. Similar to IRS2, PI3K expression also showed no sex (P =0.22) or 189 treatment (P =0.11) effect but revealed significant sex by treatment interaction. The 190 directionality of T treatment effects differed between sexes, with Cohens effect size 191 analysis revealing a large magnitude increase (Cohen's d =1.11) in T females as 192 opposed to a large magnitude decrease (Cohen's d =1.11) in T-treated males compared 193 to their respective controls (Fig.6D). There were no significant sex (P =0.91), treatment 194 (P =0.54) or sex by treatment interaction (P =0.88) with AKT gene expression (Fig.6E). Cohen's effect size analyses revealing a large magnitude increase in T-treated females 225 (Cohen's d =0.98) and a medium magnitude increase (Cohen's d =0.73) in T-treated 226 males compared to their sex-matched controls (Fig.7B-A). There were no significant 227 sex (P =0.9), treatment (P =0.9) or sex by treatment interaction (P =0.53) with total 228 GSK-3β protein expression (Fig.7B-B). Ratio of pGSK-3β/ GSK-3β showed a significant 229 sex effect but no treatment (P =0.19) or sex by treatment interaction P =0.52) (  Female-specific upregulation in markers of insulin signaling, namely PI3K and trend in 393 IRS2 as well as higher magnitude increase in IR and IGFR suggests that insulin/IGF 394 signaling pathway may be major contributors in the sexually dimorphic cardiomyocyte 395 hyperplasia. To what extent the sex-specific differences in LV morphological and 396 molecular phenotype seen in the heart of T fetuses would translate to adult 397 cardiovascular differences is unknown. While information is available on the impact of 398 fetal exposure to T excess on cardiac morphology and molecular markers in adult 399 females following gestational day 30-90 T treatment [33], data is lacking in males. 400 Nonetheless, a large body of evidence prevails that points to sex differences in 401 cardiovascular physiology and pathophysiology [71][72][73] Determining the early histological and molecular alteration in the LV of both males and 421 females from prenatal T excess allows designing sex-specific early interventions to 422 prevent CVD onset. 423 In conclusion, findings of the present study demonstrate sex-specific effects of 424 gestational T excess between days 30-90 of gestation on the cardiac phenotype. 425 Furthermore, the sex-specific programming is likely secondary to perturbation in both 426 estrogen and insulin signaling pathways collectively. These findings are supportive of 427 the role of androgen excess to serve as early biomarkers of CVD and could be critical in 428 identifying therapeutic targets for LV hypertrophy and predict long-term CVD. 429

431
Experimental animals, prenatal T-treatment, and tissue collection 432 gestational day 30 to 90; term: 147 days (Fig.1A). Controls received 2 ml of corn oil 444 vehicle. At the end of T treatment (mean ± SEM: 91.9 ± 0.2) of gestation, ewes were 445 euthanized as previously described [46]. Briefly, sedation was induced with 20-30 ml of 446 pentobarbital i.v. (Nembutol Na solution, 50 mg/ml; Abbott Laboratories, Chicago, IL), 447 and anesthesia was maintained with 1-2% halothane (Halocarbon Laboratories, River 448 Edge, NJ). The gravid uterus was exposed through a midline incision, and male and 449 female fetuses were collected, and the hearts removed. The dam was the experimental 450 unit with only one randomly selected male or female fetus used from each dam if there 451 were more than one fetus. Fetal body and heart weights were recorded. LV tissues 452 were separated, snap-frozen, and stored at -80°C until utilized for mRNA and western 453 blot analysis or fixed in formaldehyde in PBS pH 7.4 and paraffin-embedded for 454 molecular and histological analyses. 455

RNA extraction and quantitative RT PCR analysis 457
Real-time PCR (RT-PCR) was used to examine gene expression in the myocardial 458 tissue (n= 6 Controls and 6 T-Treated males and females). LV tissue was used to 459 extract total RNA using TRIzol reagent (Invitrogen, Carlsbad, CA). Following isolation, 460 RNA was treated with DNAse to digest DNA and purified using the RNAeasy kit 461 (Qiagen, Germantown, MD) to obtain high-quality RNA according to the manufacturers' 462 instructions. RNA quality was determined spectrophotometrically using NanoDrop 463 Inc., Hercules, CA). All primer sequences for genes investigated were previously 478 published and are shown in Table 1. Each gene was tested in triplicate, averaged, and 479 the expression was determined after normalization with an expression of the 480 housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The 481 relative amount of each transcript (in fold) was measured by using the ΔΔCT method 482 [85]. 483 484

Protein extraction and western blot analysis 485
Frozen LV tissues were homogenized in radioimmunoprecipitation assay buffer (Pierce 486 RIPA buffer; Thermo Scientific) containing Protease inhibitors (Thermo Scientific) and 487 phosphatase inhibitor (Thermo Scientific). The homogenized extracts were centrifuged 488 at 10000g for 15 minutes at 4°C, and the whole-cell protein extract was used for the 489 analysis. Equal amounts of protein (15µg) were resolved on SDS-PAGE and transferred 490 into a nitrocellulose membrane (Bio-Rad). Membranes were incubated in blocking buffer 491 (5% nonfat milk diluted in Tris-buffered saline) for 1 hr at room temperature and 492 incubated overnight at 4°C with primary antibodies (all previously published and are 493 shown in Table 2). Primary antibodies including mTOR, P-mTOR, glycogen synthase diameter of cardiomyocytes were quantified by measuring the short axis of nucleated 511 transverse sections from six randomly chosen images per animal as previously 512 described [33]. For each image three counting frames (50.10 x 50.10µm) were 513 quantified by an experimenter blinded to the treatment groups with ~300 514 cardiomyocytes per animal measured. Images were acquired by Leica DM 1000 LED 515 microscope at 63X magnification and quantified using Image J software. To assess 516 development of fibrosis in LV tissues, 3 sections 100 µm apart were stained with 517 Masson's trichrome staining (Abcam, ab150686) according to the manufacturer's 518 instructions. Fibrotic tissue quantification was performed as previously described [86]. 519 They were imaged on Zeisss microscope and analyzed using Image J software (NIH, 520 Bethesda, MD). 521 522

Statistical Analysis 523
After testing for homogeneity of variance, data were log-transformed when needed and 524 analyzed using two-way ANOVA with fetal sex (males; female), treatment (T; CON), and 525 their interaction as the main effects. Statistical outliers were excluded from the analysis 526 using Grubbs' test (https://www.graphpad.com/quickcalcs/grubbs1/). Analyses were 527 performed using GraphPad Prism (Prism 9.0, GraphPad Software, San Diego, CA). 528 Data are presented as mean ± SEM and differences were considered significant at P 529 <0.05, tendencies at P ≤ 0.10. A post hoc test using Bonferroni's multiple comparisons 530 analysis was performed when there was either a significant (P <0.05) effect or a trend 531 towards significance (P ≤ 0.10) in sex by treatment interaction was evident. Additionally, 532 due to the small sample size, data were also analyzed by Cohen's effect size analyses 533 [87]. Cohen's d value of 0.5-0.8 and 0.8 represent medium and large effect size 534 differences between the T-treated and control groups. 535 536

Data availability 537
All data from the current study are available from the corresponding author upon 538 reasonable request. 539 540 References 541