Congenital vs. Adult-Onset Estrogen Deficiency in Female Mice: Physiological Effects and Implications for Therapy

Abstract

This study compared the behavioral, metabolic, and molecular consequences of congenital complete estrogen deficiency (aromatase knockout, AROM KO) and adult-onset partial deficiency (ovariectomy, OVX) in female mice to identify shared phenotypes relevant for model selection and therapeutic targeting. Female AROM KO, OVX, and wild-type C57BL/6 littermates were fed a low-fat or high-fat diet for 17 weeks. Body composition, energy expenditure, physical activity, respiratory exchange ratio (RER), glucose metabolism, and gene expression (microarray and qRT-PCR) in adipose tissue and skeletal muscle were assessed. Both AROM KO and OVX mice exhibited increased adiposity, reduced physical activity (>40% reduction in ambulatory movement, ≥70% reduction in wheel running), and elevated RER. AROM KO mice displayed higher baseline body weight as well as hyperglycemia and hyperinsulinemia compared to OVX mice, reflecting more severe effects of complete estrogen loss. Transcriptomic analyses revealed downregulated metabolic pathways (e.g., TCA cycle, fatty acid metabolism) and upregulated inflammatory pathways in adipose tissue of AROM KO mice, with similar but less pronounced changes in OVX mice confirmed by qRT-PCR. Skeletal muscle showed downregulation of exercise-responsive (Nr4a3), insulin-signaling (Irs1), metabolic (Pcx), and antioxidant (Gpx3) genes with E2 deficiency, implicating impaired energy metabolism and increased oxidative stress in metabolic dysfunction. Unexpectedly, total energy expenditure was comparable across groups despite reduced activity. Overall, congenital and adult-onset E2 deficiency share common phenotypes, but congenital deficiency induces more severe metabolic impairments. These findings validate both models for studying E2 deficiency and highlight potential therapeutic targets (Nr4a3, Gpx3, Pcx, and Irs1) for mitigating E2-deficient-related metabolic dysfunction.

INTRODUCTION

All women experience menopause in which production of the most abundant and potent pre-menopausal estrogen, 17β-estradiol (E2) (1, 2), is significantly reduced. This decline increases the risk of obesity, which is associated with chronic low-grade inflammation, metabolic problems, and a higher risk of cancer, heart disease, and early death (2–4). With women now living longer, many will spend over half their lives in the menopausal state. As a result, developing treatments to address the health issues caused by estrogen loss is critically important.

To develop effective therapies, it is crucial to understand the various behavioral, metabolic, and molecular changes regulated by estrogen. Behaviorally, estrogen plays a role in regulating both energy intake and physical activity(5–7). For example, in rodents, food consumption fluctuates during the estrus cycle, with reduced intake observed during proestrus when E2 levels are highest(6, 8). Similarly, estrogen supplementation has been shown to decrease food intake in rodents(9–11). This effect is also seen in humans, where a reduction of 250–600 kcal/day occurs during the periovulatory phase of the menstrual cycle(6).

Decreased physical activity is also characteristic of E2-deficient rodents(5, 12, 13). Furthermore, metabolically, estrogen has been shown to regulate energy balance, fatty acid and glucose metabolism, and insulin sensitivity through its actions in a multitude of metabolic tissues(7). Molecularly, estrogen can regulate gene expression effecting physiological processes through several mechanisms, including activation of estrogen response elements (EREs) found in the host’s DNA(14).

Pre-clinical researchers have predominantly used two models to study the effects of estrogen deficiency on various physiological outcomes: the aromatase knockout (AROM KO) mouse and the ovariectomy (OVX) model. The AROM KO mouse is a genetically engineered model that is congenitally deficient in E2 due to its inability to convert testosterone into estrogen, which results in elevated circulating testosterone levels(15). In contrast, the OVX model achieves E2 deficiency through the surgical removal of the ovaries. Despite their frequent use, no studies have directly compared these models or thoroughly analyzed their behavioral, metabolic, and molecular differences and similarities(13, 16–22). To fill this gap, we conducted a comprehensive study employing a variety of techniques, including indirect calorimetry, behavioral assessments, glucose metabolism evaluation, high-fat diet (HFD) feeding, and multi-tissue microarray analysis. Our study aims to provide valuable insights for the scientific community by: 1) validating common E2-deficient phenotypes across both models, 2) comparing the effects of congenital E2 deficiency (AROM KO) with adult-onset deficiency (OVX) to assist researchers in choosing the most appropriate model for their studies, and 3) offering a broad range of data that may help identify potential therapeutic targets.

MATERIAL & METHODS

Animals

Female Aromatase knockout mice (F AROM KO) which lack exons 1 and 2 of the Cyp19 gene were generated by breeding male and female heterozygote mice on a C57Bl/6 background for the respective deletions(23). Mice were genotyped as previously described(24). For all experiments, either F AROM KO or WT littermate controls were used. A subset of WT littermates underwent ovariectomy (OVX) at 10 weeks of age when female mice were sexually mature. Sham-surgery controls were not included in the long-term dietary feeding study due primarily to logistical considerations. Inclusion of sham WT mice would have required performing sham surgeries across all experimental groups, including AROM KO mice, substantially increasing surgical burden.

Diets

At 11 weeks of age (one week after OVX), mice were assigned to receive either a purified low-fat diet (LFD) (open-source, AIN-76A diet (3.77 kcal/g)) or a custom high-fat diet (HFD) for a duration of 17 weeks (n = 12–26/group). The HFD (4.57 kcal/g) was a purified diet comprised of 47%, 40%, and 13% of total calories from carbohydrate, fat, and protein, respectively, with saturated fat making up 12% of total calories to mimic the standard American diet (BioServ, Frenchtown, NJ) (25–33). The inclusion of WT OVX LFD mice was omitted in the long-term feeding study because the clinically relevant HFD better mimics the dietary conditions associated with obesity and metabolic disturbances, which are critical outcomes of E2 deficiency in both clinical and preclinical settings. Mice were housed, 4–5/cage, maintained on a 12:12-h light-dark cycle in a low stress environment (22.5°C, 50% humidity, low noise) and given food and water ad libitum. All methods were in accordance with the American Association for Laboratory Animal Science, and the Institutional Animal Care and Usage Committee of the University of South Carolina approved all experiments.

Body Weight & Body Composition

Body weight was monitored weekly. Body composition was assessed after 15 weeks of diet to use lean mass as the basis for the dose of glucose administration for the glucose tolerance test. For this procedure, mice were briefly anesthetized via isoflurane inhalation, and lean mass, fat mass, bone mineral density, and percent body fat were assessed by dual-energy X-ray absorptiometry (Lunar PIXImus, Fitchburg WI).

Metabolism

A glucose tolerance test (GTT) was performed after 16 weeks of dietary treatment. For this procedure, mice were fasted for 5 h, and glucose was administered intraperitoneally at 2 g/kg lean mass. A glucometer (Bayer Contour) was used to measure blood glucose concentrations (tail sampling) intermittently over a 2-h period (0, 15, 30, 60, 90, and 120 min). Area of the Curve (AOC) after subtracting baseline blood glucose levels was calculated using the trapezoidal rule based off of the insights by Virtue and Vial-Puig (34). Blood was collected from the tip of the tail during the GTT (0, 30 and 60 min) to assess blood insulin levels. Plasma insulin concentrations were analyzed according to the manufacturer’s instructions using a mouse insulin ELISA kit (Mercodia, Winston Salem, NC). HOMA-IR was calculated from fasting glucose (mg/dL) and insulin (μU/mL) as (glucose × insulin) / 405.

Tissue Collection

After 17 weeks of dietary treatment, mice were euthanized via isoflurane inhalation for tissue collection. Visceral fat (gonadal, mesenteric, and perirenal fat pads), as well as the liver, and skeletal muscle (gastrocnemius) were removed, weighed, and immediately snap-frozen in liquid nitrogen and stored at −80°C.

Hepatic Lipid Content

For total hepatic lipid assessment, lipids were isolated from the liver utilizing a modified folch extraction method and quantified gravimetrically, as previously described(33, 35, 36).

Microarray Experiments

An E.Z.N.A. Total RNA Kit (Omega Bio-Tek, Norcross, GA) was used to isolate RNA from a subset of F WT and F AROM KO LFD mice (n=11–13/group and n=5/group for gonadal adipose tissue and skeletal muscle (gastrocnemius), respectively). RNA quantity was assessed using an Agilent 2100 Bioanalyzer and RNA Integrity Numbers (RIN) ranged from 7–9. For the gene expression experiments, total RNA samples were amplified and biotinylated using GeneChip WT PLUS Reagent Kit (Thermo Fisher Scientific, Cat. No. 902930), according to the manufacturer’s recommendations and were analyzed as previously described(33, 37). Differentially expressed genes with p-values < 0.05 and fold change > 1.5 and < −1.5 were used for further bioinformatics analysis(33, 37). Microarray analyses were performed in AROM KO and WT mice under LFD conditions to define transcriptional signatures associated with complete estrogen deficiency independent of dietary lipid excess. Targeted qRT-PCR analyses were then used to assess whether estrogen-responsive genes identified in the microarray dataset were similarly altered following adult-onset estrogen loss (OVX) and HFD feeding.

Quantitative Real-Time PCR

An E.Z.N.A. Total RNA Kit was used to isolate RNA. Taqman reverse transcription reagents and probe assays (Applied Biosystems, Waltham, MA) were used to analyze expression of the following genes in gonadal adipose tissue: Emr1, Cd11c, Mcp1, Pdha1, Sfrp5, Spon1, and Irs1 and skeletal muscle (gastrocnemius): Nr4a3, Pcx, Irs1, and Gpx3. Potential reference genes (Hprt, 18s, Gapdh, β-Actin, Hmbs, Tbp, H2afv, Nono, Rplpo and B2m) were analyzed for stability using Qbase+ software (Biogazelle, Ghent, Belgium). The optimal number of reference genes was determined by Qbase+ (adipose tissue: H2afv & Nono; skeletal muscle: H2afv & Rplpo), and the geometric mean of these genes was used as the normalization factor for each analysis. Gene expression was quantified using the ΔΔC_T method and Qbase+ software(38).

Indirect Calorimetry and Behavioral Phenotyping

Indirect calorimetry and behavioral phenotyping were conducted across three separate experiments. In the first experiment, 12-week-old F AROM KO mice and their F WT littermates (n = 7–8 per group) were housed in a 16-cage Promethion indirect calorimetry system (Sable Systems International, Las Vegas, NV, USA) under a 12-hour light/dark cycle at 24.7 °C for 11 days. In the second experiment, F WT and WT OVX mice—three weeks post-sham or OVX surgery—were placed in the same system for 11 days at 24.5 °C (n = 7–8 per group). A third follow-up experiment was performed using F WT and WT OVX mice under thermoneutral conditions (30 °C). Sham-surgery controls were included for all metabolic cage experiments, as these studies involved short postoperative recovery periods during which surgical stress could influence behavioral and metabolic outcomes.

All mice were fed the AIN-76A purified LFD and underwent body composition analysis via DEXA prior to calorimetry. Indirect calorimetry and behavioral phenotyping were conducted under LFD conditions to isolate the effects of estrogen deficiency on energy expenditure, substrate utilization, and physical activity, independent of the confounding metabolic effects of HFD feeding, and to capture baseline phenotypes given that female AROM KO and OVX mice exhibit significant body weight gain and increased adiposity even under LFD conditions as we have previously shown (39). For each experiment, after a 4-day acclimation period, measurements were collected over the subsequent 7 days. The respiratory exchange ratio (RER) and locomotor data were analyzed using a two-tailed Student’s t-test, while all other variables were analyzed using ANCOVA with lean mass as a covariate via the MMPC Statistical Analysis Page (https://www.mmpc.org/shared/regression.aspx) (36, 40, 41).

For voluntary wheel running assessment, a separate cohort of singly housed F AROM KO, F WT, and WT OVX mice were placed in cages equipped with running wheels (Sable Systems International). Following a one-week acclimation period, average daily running distance was recorded over 10 days. These mice were also maintained on the AIN-76A LFD throughout the experiment.

Fecal Energy Content Determination

Approximately 75–100 mg of fecal pellets were collected from WT OVX and F WT mice (n = 5–/group). Fecal energy content was measured using a 6200 Isoperibol Bomb Calorimeter (Parr Instruments, Moline, Illinois), following the manufacturer’s instructions.

Statistical Analysis

Data were analyzed using Prism 10 (GraphPad Software, La Jolla, CA). A two-way ANOVA (genotype × diet) with Newman-Keuls post-hoc testing was used to compare F WT LFD, F AROM KO LFD, F WT HFD, and F AROM KO HFD mice. Student’s two-tailed t-tests assessed differences between OVX HFD and both F AROM KO HFD and F WT HFD groups. One-way ANOVA with Newman-Keuls post-hoc was used for wheel running data. Datasets failing Bartlett’s test for equal variances were log-transformed before reanalysis. Results are presented as mean ± SE, with significance set at p < 0.05.

RESULTS

Congenital estrogen deficiency leads to a higher baseline body weight, but both congenital and adult-onset E2 deficiency leads to a similar net body weight gain

Body weight:

At the initiation of dietary treatment, it was evident that the F AROM KO mice weighed significantly more than F WT gonadally intact and OVX mice (Figure 1A; p<0.05). Terminally, F AROM KO HFD mice exhibited the highest body weight, followed by OVX HFD, F WT HFD, F AROM KO LFD (similar to F WT HFD), and F WT LFD mice (p<0.05). However, when examining net body weight gain, the F AROM KO HFD and WT OVX HFD mice both increased body weight similarly, followed by the F WT HFD, F AROM KO LFD, and F WT LFD mice (Figure 1B; p<0.05).

Figure 1. Impact of E2 deficiency on body composition.

Female (F) wildtype (WT), F Aromatase (AROM) knockout (KO) mice, and WT ovariectomized (OVX) mice were fed a low-fat diet (LFD) or high-fat-diet (HFD) for 17 weeks (n=12–26/group). A) Body weight over the course of the study, B) Δ body weight (grams) from baseline, C) uterus weight, D) bone mineral density, E) total body fat, F) visceral fat, G) liver weight, H) liver weight as a percentage of body weight, I) and hepatic lipid accumulation. Data is presented as mean ± SE. Bar graphs not sharing a common letter are significantly different from one another (P<0.05) as assessed by a Two-Way ANOVA. # = p<0.05 for the F WT HFD vs. WT OVX HFD comparison; * = p<0.05 for the WT OVX HFD vs. F AROM KO HFD comparison; ns = not statistically significant as assessed by a Student’s Two-Tailed T-Tests. ME = Main Effect.

Uterine weight and bone mineral density (BMD):

As expected, uterine weights were significantly lower in the F AROM KO groups compared to F WT groups, regardless of diet (Figure 1C; p<0.05). Likewise, WT OVX HFD mice exhibited reduced uterine weights relative to F WT HFD mice, with values comparable to those of F AROM KO HFD mice (p<0.05). Consistently, the AROM KO genotype had a significant main effect in reducing BMD (Figure 1D). WT OVX HFD mice also showed a similar decrease in BMD, matching that of the F AROM KO HFD group and falling below that of the F WT HFD mice (p<0.05).

Adiposity:

F AROM KO HFD mice had the greatest total and visceral fat mass, followed by F AROM KO LFD and F WT HFD mice, while F WT LFD mice had the lowest fat mass (Figure 1E & 1F; p<0.05). WT OVX HFD mice showed increased fat mass compared to female WT HFD mice, but lower fat mass than F AROM KO HFD mice (p<0.05).

Hepatic Steatosis:

Liver weights were highest in the F AROM KO HFD mice, followed by the F WT HFD and F AROM KO LFD groups (Figure 1G; p<0.05). The F WT LFD mice had the smallest livers, significantly smaller than all groups except F AROM KO LFD. WT OVX HFD mice had greater liver weights than F WT HFD mice but did not differ from F AROM KO HFD mice (p<0.05). When liver weight was normalized to total body weight, F WT LFD mice displayed a higher liver-to-body weight ratio compared with all other groups (Figure 1H; p < 0.05), consistent with lower overall body mass rather than increased liver size. Liver-to-body weight ratios were comparable among F WT HFD, WT OVX HFD, and F AROM KO HFD mice, indicating similar relative liver size across HFD groups. Hepatic lipid content was highest in the F AROM KO HFD group, with WT OVX HFD and F WT HFD mice showing similar levels. Both the F WT LFD and F AROM KO LFD groups had significantly lower hepatic lipid accumulation (Figure 1G; p<0.05).

Congenital E2 deficiency impairs glucose metabolism to a greater extent than adult-onset E2 deficiency

Hyperglycemia, Hyperinsulinemia, and HOMA-IR:

Fasting blood glucose, fasting insulin, and HOMA-IR were significantly elevated in F AROM KO HFD mice compared to all other groups, including WT OVX HFD mice (Figure 2A–C; p<0.05). Additionally, F AROM KO LFD and F WT HFD mice presented hyperinsulinemia and a higher HOMA-IR but not hyperglycemia relative to F WT LFD mice (Figure 2A–C; p<0.05).

Figure 2. Congenital E2 deficiency impairs glucose metabolism more than adult-onset estrogen deficiency.

Mice were assessed metabolically by examining A) fasting blood glucose, B) fasting insulin, C) HOMA-IR, D) a glucose tolerance test (GTT), and the E) insulin response to the GTT (n=12–26/group). Data is presented as mean ± SE. Bar graphs not sharing a common letter are significantly different from one another (p<0.05) as assessed by a Two-Way ANOVA. # = p<0.05 for the F WT HFD vs. WT OVX HFD comparison; * = p<0.05 for the WT OVX HFD vs. F AROM KO HFD comparison; ns = not statistically significant as assessed by a Student’s Two-Tailed T-Tests. ME = Main Effect.

Glucose Tolerance Test:

All groups presented impaired glucose tolerance relative to the F WT LFD mice as assessed by AOC (Figure 2D; p<0.05). Furthermore, the F AROM KO HFD mice presented a greater degree of impaired glucose metabolism relative to the F AROM KO LFD mice (p<0.05). The pattern of the circulating insulin concentrations over the course of the GTT mimicked the fasting insulin values: the F AROM KO HFD mice presented the highest insulin concentrations compared to all other groups and the F AROM KO LFD and F WT HFD mice presented higher insulin concentrations relative to the F WT LFD mice (Figure 2E; p<0.05).

Microarray analysis shows E2 deficiency modulates pathways associated with the TCA cycle, fatty acid metabolism, insulin resistance, and inflammation in the adipose tissue

Adipose Tissue Microarray Findings:

We performed a microarray analysis comparing the adipose tissue of F WT LFD and F AROM KO LFD mice (Supplementary Dataset 1). We selected these groups for microarray analysis because they were on a low-fat diet (LFD), allowing us to eliminate the confounding effects of high-fat diet exposure. Additionally, we aimed to examine the impact of complete E2 deficiency using the AROM KO model. Genes that showed the most statistically significant differential expression or piqued our interest were subsequently analyzed by qRT-PCR across all experimental groups. KEGG pathway analysis showed a multitude of genes and associated pathways significantly impacted by E2 deficiency (Table 1 & Table 2) (p<0.05). The majority of the genes found to be significantly changed corresponding to pathways involving the TCA cycle, carbon metabolism, fatty acid metabolism, fatty acid degradation, AMPK signaling, pyruvate metabolism, glycolysis, thermogenesis, and glucagon signaling were found to be downregulated in the F AROM LFD mice relative to the F WT LFD mice. On the other hand, genes associated with inflammatory pathways (intestinal immune network for IgA production, inflammatory bowel disease, complement and coagulation cascades, and leukocyte transendothelial migration) were found to be upregulated in the F AROM KO LFD mice.

Table 1.

Kegg pathway analysis of adipose tissue alterations in congenital E2 deficiency (Aromatase KO) in mice; WT = wildtype.

Description	Upregulated (↑) Genes Relative to WT	Downregulated (↓) Genes Relative to WT	Total ↑ vs ↓	p-value
Citrate cycle (TCA cycle)	Ogdhl, Idh2	Pdha1, Pcx, Pdhb, Aco1, Sdha, Idh3b, Aco2, Idh3g, Cs, Acly, Sdhb	2↑, 11↓	4.56E-11
Carbon metabolism	Phgdh, Ogdhl, Idh2, Psat1	Adpgk, Pdha1, Pcx, Pfkl, Me1, Acat3, Acat2, Gpi1, Pdhb, Aco1, Acss2, Sdha, Idh3b, Aco2, Idh3, Cs, G6pd2, Echs1, Pgd, Sdhb	4↑, 20↓	2.58E-10
Fatty acid metabolism	Cpt1a	Acadsb, Cpt2, Elovl6, Acaca, Fasn, Acat3, Acat2, Elovl3, Fads1, Acadvl, Tecr, Acsl1, Echs1	1↑, 13↓	2.06E-09
Fatty acid degradation	Cpt1a	Acat3, Acat2, Acadsb, Cpt2, Eci3, Acadvl, Acsl1, Echs1	1↑, 8↓	3.73E-06
Diabetic cardiomyopathy	Pdk3, Cyba, Atpa3, Ncf4, Rac2	Pdha1, Irs1, Slc25a5, Plcb1, Pdhb, Mmp2, Uqcrc2, Cpt2, Mpc1, Sdha, G6pd2, Sdhb, Cox7b	5↑, 13↓	5.98E-06
2-Oxocarboxylic acid metabolism	Bcat1, Idh2	Aco1, Idh3b, Aco2, Idh3g, Cs	2↑, 5↓	2.21E-05
Pyruvate metabolism	Acyp2	Acaca, Pdha1, Pcx, Me1, Acat3, Acat2, Acacb, Pdhb, Acss2	1↑, 9↓	2.80E-05
Butanoate metabolism	Abat	Aacs, Acat3, Acat2, Acsm3, Echs1	1↑, 5↓	3.04E-05
Propanoate metabolism	Abat	Acaca, Acacb, Acss2, Echdc1, Echs1	1↑, 5↓	6.80E-05
PPAR signaling pathway	Aqp7, Angptl4, Cpt1a	Me1, Cpt2, Slc27a1, Fabp3, Acls1	3↑, 5↓	0.000247
Valine, leucine and isoleucine degradation	Bcat1, Abat	Aacs, Acat3, Acat2, Acadsb, Mccc1, Echs1	2↑, 6↓	0.00028
Biosynthesis of amino acids	Phgdh, Bcat1, Idh2, Psat1	Pcx, Pfkl, Aco1, Idh3b, Cs, Asns	4↑, 6↓	0.00029
Glucagon signaling pathway	Cpt1a, Sik1	Acaca, Pfkl, Pygl, Pdha1, Acacb, Plcb1, Pdhb, Phkg1, Sik2	2↑, 9↓	0.00034
Steroid biosynthesis		Dhcr7, Cyp51	0↑, 2↓	0.00193
Fatty acid elongation		Elovl6, Elovl3, Tecr, Echs1	0↑, 4↓	0.00197
Proximal tubule bicarbonate reclamation	Aqp1, Fxyd2	Atp1a2, Atp1a3	2↑, 2↓	0.00304
Glyoxylate and dicarboxylate metabolism		Acat3, Acat2, Aco1, Acss2, Aco2, Cs	0↑, 6↓	0.00334
Hematopoietic cell lineage	Il7r, H2-DMb2, Cd44, H2-DMb1, Cd59a, H2-Ob, Ms4a1, Cd9, Il2ra, Cd37	Il4a	10↑, 1↓	0.00447
Insulin resistance	Cpt1a	Pygl, Mlxipl, Ppp1r3b, Acacb, Irs1, Slc27a1	1↑, 6↓	0.00541
Adipocytokine signaling pathway	Lep, Cpt1a	Acacb, Irs1, Adipor2, Acsl1	2↑, 4↓	0.00579
Glycosphingolipid biosynthesis - lacto and neolacto series	Gcnt2	St3gal6, St3gal4, B3galt1, B4galt1	1↑, 4↓	0.00652
Non-alcoholic fatty liver disease	Lep	Irs1, Adipor2, Sdha, Sdhb, Mlxipl, Uqcrc2, Cyp2e1	1↑, 7↓	0.00658
Endocrine and other factor-regulated calcium reabsorption	Cltb, Fxyd2	Esr1, Plcb1, Atp1a2, Atp1a3, Pth1r, Klk1b11	2↑, 7↓	0.00740
Cardiac muscle contraction	Atp2a3, Myl4, Tpm3, Tpm4, Fxyd2	Atp1a2, Atp1a3	5↑, 2↓	0.00746
Inflammatory bowel disease	H2-DMb1, Stat6, Stat4, H2-DMb2, H2-Ob	Rorc, Il4ra	5↑, 2↓	0.00816
Leukocyte transendothelial migration	Myl12a, Actn1, Cyba, Cldn5, Cxcr4, Itgb2, Ncf4, Rac2, Ezr, Actn4	Mmp2, Cldn1	10↑, 2↓	0.00938
Fatty acid biosynthesis		Acaca, Fasn, Acacb, Acsl1	0↑, 4↓	0.01065
Complement and coagulation cascades	Itgax, F2r, Plaur, Itgb2, Cd59a, Serping1	F3, C2	6↑, 2↓	0.01176
Central carbon metabolism in cancer	Ret, Idh2	Pdha1, Pdk1, Pfkl, Pdhb, G6pd2, Met	2↑, 6↓	0.01519
AMPK signaling pathway	Lep, Cpt1a, Ppp2r1b	Acaca, Ppp2r5b, Pfkl, Fasn, Acacb, Irs1, Adipor2	3↑, 7↓	0.01532
beta-Alanine metabolism	Abat, Aldh3b1	Csad, Echs1	2↑, 2↓	0.01585
Glycerophospholipid metabolism	Pla2g5, Lpgat1	Gpam, Gpd2, Crls1, Agpat2, Cept1, Ptdss2	2↑, 6↓	0.01658
Peroxisome	Pex5l, Prdx5, Idh2	Eci3, Crat, Hacl1, Acsl1	3↑, 4↓	0.01933
Metabolism of xenobiotics by cytochrome P450	Ephx1, Aldh3b1	Cyp2s1, Dhdh, Gsta3, Cyp2e1	2↑, 4↓	0.02078
Terpenoid backbone biosynthesis		Acat3, Acat2	0↑, 2↓	0.02100
Thermogenesis	Cpt1a	Uqcrc2, Cpt2, Sdha, Acsl1, Adcy10, Sdhb, Adrb3	1↑, 7↓	0.02435
Asthma	H2-DMb1, Ccl11, H2-DMb2, H2-Ob		4↑, 0↓	0.02787
Regulation of actin cytoskeleton	Itgax, F2r, Myl12a, Actn1, Pfn1, Cxcr4, Igb2, Nckap1l, Itga7, Arhgef6, Rac2, Ezr, Iqgap2, Actn4	Fgf10	14↑, 1↓	0.03018
Alcoholic liver disease	Lbp, Cpt1a	Acaca, Fasn, Acacb, Adipor2, Acadvl, C2	2↑, 6↓	0.03357
Calcium signaling pathway	F2r, Atp2a3, Ret, Cxcr4, Cacna1a, Cacna1e, Stim2	Ntrk2, Plcb1, Slc25a5, Phkg1, Met, Ptgfr, Fgf10, Adrb3, Pde1a	7↑, 9↓	0.03390
Alanine, aspartate and glutamate metabolism	Abat, Folh1	Nat8l, Asns	2↑, 2↓	0.03472
Cell adhesion molecules	H2-DMb1, H2-DMb2, H2-Ob, Cldn5, Itgb2, Ptprc, Vsir, Sdc1	Nrxn1, Cntnap1, Vtcn1, Siglec1, Cldn1	8↑, 5↓	0.03475
Glycolysis / Gluconeogenesis	Aldh3b1	Adpgk, Pdha1, Pfkl, Gpi1, Pdhb, Acss2	1↑, 6↓	0.03721
Arachidonic acid metabolism	Pla2g5, Ltc4s	Gpx3, Gpx8, Cyp2e1	2↑, 3↓	0.04870
Intestinal immune network for IgA production	H2-DMb1, H2-DMb2, H2-Ob, Cxcr4		4↑, 0↓	0.04991
Fat digestion and absorption	Pla2g5, Slc17a1	Acat3, Acat2, Agpat2	2↑, 3↓	0.04991

Table 2.

Microarray analysis of adipose tissue genes significantly (p<9.5E-05) downregulated or upregulated (±2.0 fold) in response to congenital E2 deficiency in mice

ADIPOSE TISSUE
DOWNREGULATED GENES
Gene	Fold Change	P-Value
Spon1	−33.42	1.70E-09
Elovl6	−21.3	2.28E-06
Fam13a	−8.19	9.16E-05
Gpd2	−7.71	5.22E-06
Slc2a5	−6.74	1.36E-06
Parm1	−5.97	1.29E-05
Acaca	−5.95	5.60E-06
Cybrd1	−5.45	4.84E-05
Bnc2	−4.25	2.26E-07
Gpam	−4.23	1.50E-07
Pdha1	−3.57	1.71E-05
Nrxn1	−3.55	8.64E-08
Me1	−3.49	5.30E-05
Aacs	−3.35	7.93E-06
Ntrk2	−3.01	2.30E-07
Tlcd2	−2.89	7.02E-06
Egln3	−2.8	4.60E-06
Ppp1r3b	−2.71	6.49E-05
Ddhd2	−2.69	1.13E-06
Trp63	−2.69	5.42E-05
Slc25a1	−2.67	6.96E-06
Pth1r	−2.66	5.39E-07
Spon2	−2.44	1.35E-05
Letmd1	−2.38	8.00E-05
Vnn3	−2.35	3.44E-05
Cped1	−2.34	4.93E-05
Gm8113	−2.33	1.37E-05
Pygl	−2.31	1.38E-05
Dhrs7	−2.26	5.95E-05
Mettl7a1	−2.25	2.12E-05
Pcx	−2.24	1.74E-05
Pdk1	−2.18	2.04E-05
Pfkl	−2.18	2.59E-05
Csrp2	−2.17	7.03E-05
Mlxipl	−2.16	5.24E-05
Tmem79	−2.12	7.11E-07
Pomt1	−2.11	6.94E-06
Adpgk	−2.09	8.94E-08
Gpx8	−2.05	3.00E-05
Ppp2r5b	−2.03	1.69E-05
Mapk8ip1	−2.02	1.43E-05
UPREGULATED GENES
Gene	Fold Change	P-Value
Mmp12	137.46	1.16E-05
Sfrp5	28.34	5.11E-07
Atp6v0d2	19.51	1.48E-05
Slc5a7	13.67	3.88E-05
Mest	11.28	6.06E-06
Ubd	7.14	2.91E-05
Gpnmb	7.07	2.58E-05
Syp	6.04	8.70E-07
Abat	4.55	9.72E-06
Map1a	4.32	2.52E-05
Serpine2	4.21	4.45E-06
Itgax	3.89	8.17E-06
Prelp	3.5	1.08E-05
Synpo2	3.32	2.14E-05
Anxa1	3.3	1.49E-06
Il7r	3.27	4.89E-05
Gltp	3.15	6.73E-08
Pla2g5	3.15	2.46E-05
Nos1ap	3.04	8.26E-05
Mfge8	2.91	1.03E-06
Crip2	2.87	2.61E-07
H2-DMb1	2.7	7.41E-05
Jade3	2.69	1.09E-05
Adam23	2.67	9.22E-05
Trp53i11	2.57	2.81E-05
Pde2a	2.38	2.29E-05
Arhgap25	2.37	2.03E-05
Slco3a1	2.35	6.47E-05
Blnk	2.31	5.68E-06
2810032G03Rik	2.28	8.97E-05
Myo1e	2.23	8.06E-06
Dynlt1c; Dynlt1f	2.22	1.77E-05
Pea15a	2.18	8.38E-05
Gm7694	2.13	8.75E-06
Phgdh	2.1	6.28E-06
Pdk3	2.1	2.76E-05
Prkch	2.09	2.39E-05
Btg2	2.06	2.22E-06
Sod3	2.01	2.88E-05
Anxa2	2	3.50E-05
Mmp12	137.46	1.16E-05
Sfrp5	28.34	5.11E-07
Atp6v0d2	19.51	1.48E-05
Slc5a7	13.67	3.88E-05
Mest	11.28	6.06E-06
Ubd	7.14	2.91E-05
Gpnmb	7.07	2.58E-05
Syp	6.04	8.70E-07
Abat	4.55	9.72E-06
Map1a	4.32	2.52E-05
Serpine2	4.21	4.45E-06
Itgax	3.89	8.17E-06
Prelp	3.5	1.08E-05
Synpo2	3.32	2.14E-05
Anxa1	3.3	1.49E-06
Il7r	3.27	4.89E-05
Gltp	3.15	6.73E-08
Pla2g5	3.15	2.46E-05
Nos1ap	3.04	8.26E-05
Mfge8	2.91	1.03E-06
Crip2	2.87	2.61E-07
H2-DMb1	2.7	7.41E-05
Jade3	2.69	1.09E-05
Adam23	2.67	9.22E-05
Trp53i11	2.57	2.81E-05
Pde2a	2.38	2.29E-05
Arhgap25	2.37	2.03E-05
Slco3a1	2.35	6.47E-05
Blnk	2.31	5.68E-06
2810032G03Rik	2.28	8.97E-05
Myo1e	2.23	8.06E-06
Dynlt1c; Dynlt1f	2.22	1.77E-05
Pea15a	2.18	8.38E-05
Gm7694	2.13	8.75E-06
Phgdh	2.1	6.28E-06
Pdk3	2.1	2.76E-05
Prkch	2.09	2.39E-05
Btg2	2.06	2.22E-06
Sod3	2.01	2.88E-05
Anxa2	2	3.50E-05

Adipose Tissue Confirmatory PCR:

Confirmatory qRT-PCR generally corroborated the microarray data – the F AROM KO mice exhibited increased gene expression of the pro-inflammatory macrophage marker, Cd11c (Itgax), and the pro-inflammatory cytokine, Mcp1, irrespective of diet (Figure 3B & 3C; p<0.05). In the case of Emr1, a pan macrophage marker, the F AROM KO HFD mice displayed increase mRNA expression relative to the F WT HFD mice (Figure 3A) (p<0.05). Although there was no statistically significant increase in Emr1 expression between the F WT LFD and F AROM KO mice, the F AROM KO LFD mice did display similar levels of Emr1 expression as the F WT HFD mice (p<0.05). With respect to the WT OVX HFD mice, compared to the F WT HFD mice, the WT OVX HFD mice exhibited significantly elevated Emr1, Cd11c, and Mcp1 mRNA expression (p<0.05). Compared to the F AROM KO HFD mice, the WT OVX HFD mice presented a similar content of Emr1 and Mcp1, with Cd11c being slightly lower (p<0.05). There was a main effect for the AROM KO to downregulate Pdha1, a gene encoding a protein that forms part of the PDH complex which catalyzes the conversion of pyruvate to acetyl CoA (Figure 3D; p<0.05) (42), Spon1, a protein involved in extracellular matrix remodeling(43) (Figure 3E; p<0.05), and Irs1, a key receptor in the insulin signaling pathway (Figure 3F; p<0.05)(44), and upregulate Sfrp5, an anti-inflammatory adipokine(45) (Figure 3G; p<0.05). The WT OVX HFD mice also exhibited decreased expression of Pdha1, Spon1, and Irs1 relative to F WT HFD mice (p<0.05).

Figure 3. Confirmatory qRT-PCR of genes impacted by E2 deficiency in adipose tissue.

qRT-PCR was used to assess adipose tissue gene expression of A) Emr1, B) Cd11c, C) Mcp1, D) Pdha1, E) Sfrp5, F) spon1, and G) Irs1 (n=12–26/group). Data is presented as mean ± SE Bar graphs not sharing a common letter are significantly different from one another (p<0.05) as assessed by a Two-Way ANOVA. # = p<0.05 for the F WT HFD vs. WT OVX HFD comparison; * = p<0.05 for the WT OVX HFD vs. F AROM KO HFD comparison; ns = not statistically significant as assessed by a Student’s Two-Tailed T-Tests. ME = Main Effect.

E2 deficiency induces changes in genes associated with physical activity, cellular energy metabolism, insulin signaling, and glutathione metabolism in skeletal muscle

Skeletal Muscle Microarray Findings:

Microarray analysis of the skeletal muscle of F AROM KO LFD and F WT LFD revealed far less significantly changed genes than in adipose tissue (Table 3; complete data found in Supplementary Data Set 1). Some of the pathways that were affected as determined by KEGG analysis included protein processing in the endoplasmic reticulum, glutathione metabolism, as well as linoleic acid and arachidonic acid metabolism (Table 4) (p<0.05).

Table 3.

Kegg pathway analysis of skeletal muscle alterations in congenital E2 deficiency (Aromatase KO) in mice; WT = wildtype.

Description	Upregulated (↑) Genes Relative to WT	Downregulated (↓) Genes Relative to WT	Total ↑ vs ↓	p-value
Protein processing in endoplasmic reticulum	Sel212, Casp12	Nfe2l2, Xbp1, Ube2d1, Ssr3, Cryab, Dnajc5, Ubxn1, Txndc5, Pdia3, Sec61b	2↑, 10↓	2.34E-04
Glutathione metabolism	Idh2, Chac1, Mgst3, Gsta1	Lancl1, Gpx3	4↑, 2↓	3.70E-03
Proximal tubule bicarbonate reclamation	Atp1b1, Aqp1	Slc4a4	2↑, 1↓	1.01E-02
Carbohydrate digestion and absorption	Atp1b1, Plcb4, Gnat3, Amy1		4↑, 0↓	1.72E-02
Linoleic acid metabolism		Cyp2e1, Pla2g5, Cyp3a44, Cyp2c37	0↑, 4↓	1.97E-02
Chemical carcinogenesis - DNA adducts	Mgst3, Gsta1	Cyp2e1, Cyp3a44, Cyp2c37	2↑, 3↓	3.02E-02
Pancreatic secretion	Atp1b1, Plcb4, Amy1	Slc4a4, Pla2g5	3↑, 2↓	3.11E-02
Arachidonic acid metabolism		Cyp2e1, Pla2g5, Gpx3, Cyc2c37, Cyp2b10	0↑, 5↓	3.29E-02
Salivary secretion	Atp1b1, Plcb4, Gucy1a2, Amy1	Nos1	4↑, 1↓	3.29E-02
Systemic lupus erythematosus		C3, C7, Trim21	0↑, 3↓	0.034
Aldosterone-regulated sodium reabsorption	Atp1b1, Sgk1	Irs1	2↑, 1↓	0.044
Nitrogen metabolism		Car3, Car14	0↑, 2↓	0.047

Table 4.

Microarray analysis of skeletal muscle genes significantly (p<0.001) downregulated or upregulated (±1.5 fold) in response to congenital E2 deficiency in mice

SKELETAL MUSCLE
DOWNREGULATED GENES
Gene	Fold Change		P-Value
Nr4a3	−6.74		3.37E-07
Stc2	−2.97		5.22E-07
Rhobtb1	−3.48		1.31E-06
Nxpe4	−3.38		2.69E-06
Kcnq5	−3.47		5.81E-06
Pcx	−5.76		2.15E-05
Tbc1d1	−3.08		3.80E-05
Irs1	−2.34		6.00E-05
Setd8	−2.16		6.65E-05
Sntb1	−2.35		0.0002
Ube4a	−1.88		0.0002
Dnaja4	−1.63		0.0002
Best3	−2.49		0.0003
Slc4a4	−2.2		0.0003
Nfe2l2	−1.57		0.0003
Fbp2	−3.75		0.0004
Xbp1	−2.58		0.0004
Rhou	−1.57		0.0004
Il15	−1.94		0.0005
Tfdp1	−1.66		0.0005
Rab3ip	−2.36		0.0006
Ube2r2	−1.68		0.0006
Mup8; Mup9; Mup2; Mup1	−17.43		0.0007
Cyp2e1	−6.2		0.0007
Naa20	−1.54		0.0007
Socs2	−3.35		0.0009
Car3	−1.59		0.0009
Ehd4	−1.55		0.0009
UPREGULATED GENES
Gene	Fold Change	P-Value
Amd2; Amd1	5.33	2.50E-07
Actr3b	2.95	8.83E-07
Amd2	7.23	6.69E-06
Cdk19	3.05	1.80E-05
Musk	2.65	2.21E-05
Fdx1	2.3	5.25E-05
Dkk3	2.06	5.39E-05
Sgk1	1.86	0.0001
Lonrf3	2.41	0.0001
Gabrg2	1.76	0.0003
Spry1	1.92	0.0003
Slc15a5	2.36	0.0003
Pim1	2.59	0.0003
Apol6	1.59	0.0004
Arrdc3	1.92	0.0004
Irx3	2.12	0.0005
Mnd1	1.58	0.0006
Tfcp2l1	1.88	0.0006
Cpvl	1.71	0.0008
Krt27	1.7	0.0009
Depdc7	1.72	0.0009
Ddit4l	2.38	0.0009

Skeletal Muscle Confirmatory PCR:

Confirmatory qRT-PCR of selected genes of interest corroborated the microarray data. There was a main effect of the F AROM KO genotype to downregulate nuclear receptor subfamily 4 group A member 3 (Nr4a3), a gene previously identified as one of the most exercise- and inactivity-responsive genes(46) (Figure 4A; p<0.05), pyruvate carboxylase (Pcx), a gene that encodes a protein responsible for converting pyruvate to oxaloacetate and replenishing TCA cycle intermediates(47) (Figure 4B; p<0.05), Irs1 (Figure 4C; p<0.05), and glutathione peroxidase 3 (Gpx3) (Figure 4D; p<0.05), a gene involved in antioxidant defense, previously shown to be regulated by E2 in skeletal muscle(48).

Figure 4. Confirmatory qRT-PCR of genes impacted by E2 deficiency in skeletal muscle.

qRT-PCR was used to assess skeletal muscle gene expression of A) Nr4a3, B) Pcx, C) Irs1, and D) Gpx3 (n=12–26/group). Data is presented as mean ± SE. Bar graphs not sharing a common letter are significantly different from one another (p<0.05) as assessed by a Two-Way ANOVA. # = p<0.05 for the F WT HFD vs. WT OVX HFD comparison; * = p<0.05 for the WT OVX HFD vs. F AROM KO HFD comparison; ns = not statistically significant as assessed by a Student’s Two-Tailed T-Tests. ME = Main Effect.

Both congenital and adult-onset E2 deficiency results in decreased fat oxidation, ambulatory movement, and voluntary wheel running

Metabolic and Behavioral Phenotyping:

Relative to their F WT controls, both E2-deficient models showed similar lean mass ANCOVA-adjusted outcomes, including a >40% reduction in ambulatory movement and a 24-hour elevated respiratory exchange ratio (RER), indicating decreased fat oxidation (Table 5 and 6; p<0.05). Although there were no statistically significant differences in lean mass–adjusted total energy intake, there was a trend toward increased intake in the F AROM KO mice (p = 0.054). Resting energy expenditure was also unaffected. Most unexpectedly, total energy expenditure in both E2-deficient models was not significantly different from their F WT controls, despite moving > 40% less. We found similar results even at thermoneutrality (Supplementary Table 1).

Table 5.

Indirect calorimetry and behavioral phenotyping of female (F) wildtype (WT) and F aromatase (AROM) KO mice (n=7–8); RER = Respiratory exchange ratio.

Average Temperature: 24.7 (0.2) °C	F WT LFD	F AROM KO LFD	Unadjusted p-value	F WT LFD	F AROM KO LFD	ANCOVA-adjusted p-value
Δ Body Weight (grams)	0.42 (0.30)	1.2 (0.30)	0.08
Lean Mass (g)	15.0 (0.42)	16.8 (0.91)	0.1
Total Energy Expenditure (kcal/day)	8.41 (0.13)	8.60 (0.14)	0.4	8.87 (0.19)	8.49 (0.17)	0.2
Energy Intake (kcal/day)	8.84 (0.49)	10.67 (0.52)	0.03	8.79 (0.56)	10.73 (0.60)	0.05
Resting Energy Expenditure (kcal/day)	5.63 (0.13)	6.23 (0.14)	0.01	5.96 (0.19)	6.26 (0.17)	0.3
Non-Resting Energy Expenditure (kcal/day)	2.78 (0.08)	2.36 (0.08)	0.002	2.9 (0.08)	2.23 (0.09)	0.0004
O2 Consumption Light Cycle Avg (mL/min)	1.06 (0.02)	1.14 (0.03)	0.01	1.13 (0.03)	1.13 (0.02)	0.9
O2 Consumption Dark Cycle avg (mL/min)	1.27 (0.03)	1.24 (0.03)	0.4	1.3 (0.03)	1.21 (0.03)	0.08
O2 Consumption 24-hour Avg (mL/min)	1.17 (0.02)	1.19 (0.02)	0.4	1.23 (0.03)	1.18 (0.02)	0.1
CO2 Production Light Cycle Avg (mL/min)	0.92 (0.02)	1.02 (0.02)	0.001	0.95 (0.02)	0.99 (0.022)	0.3
CO2 Production Dark Cycle avg (mL/min)	1.23 (0.03)	1.24 (0.03)	0.8	1.25 (0.03)	1.22 (0.03)	0.6
CO2 24 Hour Avg (mL/min)	1.08 (0.02)	1.13 (0.02)	0.09	1.10 (0.02)	1.10 (0.023)	0.9
All Meters (Per Day)	325 (15.9)	177 (17.0)	5.43E-05
RER (Light Cycle Avg)	0.86 (.01)	0.89 (.01)	0.047
RER Dark Cycle Avg)	0.96 (.00)	0.99 (.01)	0.01
RER (24 Hour Avg)	0.91 (.01)	0.94 (.01)	0.001

Table 6.

Indirect calorimetry and behavioral phenotyping of female (F) wildtype (WT) and WT ovariectomized (OVX) mice (n=7–8); RER = Respiratory exchange ratio.

Average Temperature: 24.5 (0.0) °C	F WT LFD	F OVX LFD	Unadjusted p-value	F WT LFD	F OVX LFD	ANCOVA-adjusted p-value
Lean Mass (g)	15.9 (0.8)	16.6 (0.4)	0.5
Δ Body Weight (grams)	0.2 (0.2)	1.3 (0.2)	0.001
Total Energy Expenditure (kcal/day)	8.56 (0.52)	8.62 (0.49)	0.9	8.67 (0.53)	8.53 (0.50)	0.8
Energy Intake (kcal/day)	11.08 (0.92)	10.67 (0.86)	0.8	11.12 (0.93)	10.64 (0.87)	0.7
Resting Energy Expenditure (kcal/day)	5.32 (0.44)	6.02 (0.41)	0.3	5.40 (0.4)	5.95 (0.4)	0.4
Non-Resting Energy Expenditure (kcal/day)	3.25 (0.13)	2.60 (0.13)	0.003	3.25 (0.13)	2.60 (0.13)	0.004
O2 Consumption Light Cycle Avg (mL/min)	1.03 (0.07)	1.08 (0.7)	0.5	1.04 (0.07)	1.07 (0.07)	0.7
O2 Consumption Dark Cycle avg (mL/min)	1.39 (0.08)	1.33 (0.08)	0.6	1.40 (0.08)	1.32 (0.08)	0.5
O2 Consumption 24-hour Avg (mL/min)	1.21 (0.07)	1.21 (0.07)	1.0	1.22 (0.08)	1.19 (0.07)	0.8
CO2 Production Light Cycle Avg (mL/min)	0.83 (0.05)	0.91 (0.05)	0.3	0.84 (0.05)	0.90 (0.05)	0.4
CO2 Production Dark Cycle avg (mL/min)	1.33 (0.08)	1.31 (0.08)	0.9	1.34 (0.08)	1.29 (0.08)	0.7
CO2 24 Hour Avg (mL/min)	1.08 (0.06)	1.11 (0.06)	0.7	1.09 (0.06)	1.10 (0.06)	1.0
All Meters (Per Day)	244 (11.1)	147 (10.4)	3.31E-05
RER (Light Cycle Avg)	0.80 (0.01)	0.83 (0.01)	0.06
RER (Dark Cycle Avg)	0.95 (0.01)	0.98 (0.01)	0.05
RER (24 Hour Avg)	0.87 (0.01)	0.90 (0.01)	0.01

Wheel Running:

With respect to voluntary wheel running, both E2-deficient groups exhibited ≥ 70% reduction in physical activity (Figure 5; p < 0.05).

Figure 5. Both congenital and adult-onset E2 deficiency similarly reduce voluntary wheel running activity.

Mice were singly housed and given access to running wheels (n=8–10/group). Data is presented as mean ± SE. Bar graphs not sharing a common letter are significantly different from one another (p<0.05) as assessed by a One-Way ANOVA.

Fecal energy content is not affected by E2 deficiency

Given the surprising finding that differences in energy expenditure or food intake did not account for the body weight gain in E2-deficient OVX mice, we analyzed fecal energy content in both WT OVX and F WT mice to determine whether the OVX mice were more efficient at extracting energy from their food. We found no difference with respect to fecal energy content (Supplementary Figure 1), which suggests that E2 deficiency does not seem to impact energy extraction efficiency from food.

DISCUSSION

Our comprehensive analysis of congenital (AROM KO) and adult-onset (OVX) E2 deficiency models reveal shared and distinct physiological impacts, providing insights into the behavioral, metabolic, and molecular consequences of E2 loss. By employing indirect calorimetry, behavioral phenotyping, glucose metabolism assessments, and multi-tissue transcriptomic analyses, we validated common E2-deficient phenotypes and highlighted differences to guide model selection and therapeutic target identification.

Adiposity

Increased adiposity is a well-documented consequence of E2 deficiency, both in clinical settings and in rodent models(36, 49–51). This finding is further supported by our current study. Notably, F AROM KO mice exhibited significantly higher body weights compared to their WT littermates even before the onset of dietary treatment. This early divergence suggests that E2 deficiency exerts a substantial impact on body weight regulation even at a young age. Interestingly, following OVX, WT mice gained an equivalent amount of body weight as the F AROM KO mice, suggesting that the effects of E2 loss on body weight gain (primarily in the form of excess adipose tissue), are robust and converge across both congenital and adult-onset models.

The absence of increased hepatic lipid accumulation in AROM KO mice under LFD conditions, contrasted with the exaggerated steatotic response observed during HFD feeding, indicates that aromatase deficiency alone is insufficient to drive hepatic steatosis in the absence of dietary lipid excess, at least over the duration of feeding examined in this study. Rather, these findings support a model in which estrogen deficiency sensitizes the liver to diet-induced lipid overload rather than constitutively promoting hepatic lipid accumulation. Estrogen signaling is known to regulate hepatic lipid uptake, de novo lipogenesis, and fatty acid oxidation; under low-fat conditions, these pathways may remain sufficiently balanced to limit lipid accumulation, whereas high-fat feeding may unmask impairments in hepatic lipid handling in the absence of whole body estrogen (52). Accordingly, the exacerbated steatosis observed in AROM KO mice during high-fat feeding likely reflects a gene–diet interaction rather than a baseline hepatic phenotype.

Physical Activity & Energy Expenditure

Changes in body weight are driven by alternations in energy expenditure, energy intake, or a combination of both. Regarding energy expenditure, previous research in rodents has shown that both congenital E2 deficiency and OVX-induced E2 deficiency is associated with reduced physical activity(5, 12, 13). Notably, research has shown that E2 activates melanocortin-4 receptor (MC4R) signaling in the brain to promote physical activity in female mice(53). Similarly, in humans, E2 deficiency due to menopause has been linked to reduced physical activity levels(54). We confirm that both congenital and adult-onset E2 deficiency leads to significantly reduced ambulatory movement and voluntary wheel running.

In addition to influencing physical activity, E2 also regulates energy intake(6). In general, elevated E2 levels are associated with reduced food intake (a hypophagic response), while E2 deficiency leads to increased food intake (a hyperphagic response)(55). Our behavioral and metabolic cage experiments revealed that, when not normalized for lean mass, F AROM KO mice consumed more energy than their F WT littermates—a pattern not observed in OVX mice. Even when energy intake was normalized to lean mass, there was a trend toward higher food consumption in F AROM KO mice. Previous studies in OVX rats have consistently shown that E2 deficiency leads to increased food intake(6, 56–59). However, similar effects have not been consistently observed in mice. Some studies have reported no change or a decrease in food intake following OVX in mice, suggesting possible species differences(12, 60–64). One hypothesis is that OVX may increase brain aromatase expression, which could suppress the typical hyperphagic response associated with E2 deficiency(65). This is supported by evidence showing significant aromatase expression in the female mouse brain unlike in peripheral tissues such as skeletal muscle and adipose tissue(66). Unlike OVX mice, F AROM KO mice are completely E2-deficient due to their inability to convert testosterone into E2, eliminating any potential appetite-suppressing effects of E2. This is supported by Hayashi et. al. who reported that F AROM KO mice are hyperphagic, while OVX mice are not(65). However, in this investigation, food intake was not normalized to lean mass or body weight, making it difficult to determine whether the increased intake was a direct result of E2 deficiency or secondary to increased weight gain.(65). Further complicating the picture, other studies have found no difference in food intake between F AROM KO and control mice(13). It should be noted that discrepancies between studies may be attributed to differences in housing temperature or social housing conditions, both of which are often underreported, and have been shown to significantly influence food intake(41, 67).

Given the significant decline in physical activity observed—despite only modest or negligible changes in energy intake—we initially hypothesized that the resulting weight gain was primarily driven by reduced energy expenditure from decreased physical activity, rather than by increased food consumption. However, one of the most puzzling findings of our study was that, even with a >40% reduction in ambulatory movement and the use of state-of-the-art indirect calorimetry over 7-day trials (following a 3-day acclimation period), both AROM KO and OVX mice exhibited total energy expenditure comparable to control animals. This unexpected result aligns with prior findings by Witte et al., who reported that OVX mice gained significant body weight without changes in energy expenditure, despite maintaining normal food intake and showing markedly reduced physical activity over a 20-day period.(12). Since our initial metabolic cage experiments were conducted at 24.5°C—a temperature that could introduce mild cold stress—we considered whether this stress might account for the unexpectedly similar energy expenditure across groups. Yet, even under thermoneutral conditions, our results remained consistent. To explore further, we assessed fecal energy content and found no significant differences between OVX and control mice, suggesting that food energy extraction was equivalent between groups. This suggests that there is no difference in food energy extraction. A known limitation of indirect calorimetry is its inability to capture anaerobic metabolism, which can be quantified through direct calorimetry(68, 69). Indeed, studies have shown that indirect calorimetry may underestimate resting metabolic rate by 10–12%(68). Additionally, recent research has highlighted the role of anaerobic gut metabolism in overall energy expenditure—an aspect not captured by indirect calorimetry(70). Therefore, it is plausible that changes in gut microbial composition in OVX mice may reduce anaerobic energy expenditure, leading to a decrease in total energy expenditure that remains undetected by indirect calorimetry.

Glucose Metabolism & Insulin Resistance

Both preclinical and clinical research has found that E2 is protective against insulin resistance(71). Consistent with these findings, we observed that female AROM KO mice exhibited hyperinsulinemia and impaired glucose metabolism compared to their WT counterparts, regardless of diet. Notably, we found that complete E2 deficiency in AROM KO mice resulted in significantly more pronounced hyperinsulinemia—an indicator of insulin resistance—than what was observed in OVX mice. In fact, compared to F WT HFD mice, OVX HFD mice did not exhibit increased insulin levels. OVX mice are not completely E2 deficient as they retain some E2 production, primarily from the adrenal glands(8). Unlike humans, female mice do not express aromatase in peripheral tissues such as adipose tissue and skeletal muscle, which limits extra-ovarian E2 synthesis in these sites(66, 72). This low E2 tone exhibited by OVX mice may be sufficient to mitigate the severe hyperinsulinemia observed in AROM KO mice, though it appears inadequate to prevent the associated increase in adiposity. It is important to note that our evaluation of glucose metabolism was based solely on the intraperitoneal glucose tolerance test (IP GTT). While this test provides useful insights into glucose handling, it may not fully reflect the physiological responses captured by the more relevant oral glucose tolerance test (OGTT)(73).

Transcriptomic Changes in Adipose Tissue and Skeletal Muscle

Our transcriptomic analyses revealed that complete E2 deficiency has profound effects on metabolic and inflammatory gene expression in both adipose tissue and skeletal muscle. Microarray and confirmatory qRT-PCR data from adipose tissue of AROM KO mice demonstrated marked downregulation of key metabolic pathways. These findings suggest a shift toward impaired mitochondrial oxidative metabolism and insulin resistance with E2 deficiency, consistent with previous reports linking E2 to enhanced metabolic flexibility and insulin sensitivity(71, 74). The confirmatory qRT-PCR results showed similar gene expression changes in OVX mice, indicating that even partial E2 loss promotes inflammation and induces metabolic changes, though complete deficiency appears to exacerbate these effects.

In skeletal muscle, E2 deficiency induced fewer transcriptomic changes, but key alterations were still evident. One of the most significantly downregulated genes was Nr4a3, an exercise-responsive transcription factor known to be downregulated because of inactivity in humans(46). Recent studies have shown that Nr4a3 downregulation impairs glucose metabolism and protein synthesis in skeletal muscle, further implicating this gene in metabolic regulation(75). We also observed reduced expression of Pcx, a mitochondrial enzyme that catalyzes the conversion of pyruvate to oxaloacetate, supporting anaplerotic replenishment of the TCA cycle (76, 77). In the liver, Pcx serves to provide oxaloacetate for gluconeogenesis and has recently been shown to play a fundamental role in promoting hepatic antioxidant capacity and redox metabolism (77). While Pcx is well-established in hepatic gluconeogenesis and antioxidant regulation, its role in skeletal muscle metabolism remains understudied. Nonetheless, previous reports show that Pcx is present in skeletal muscle (78), responsive to exercise (79), and upregulated by E2 supplementation in mice (80, 81), suggesting it may be under estrogenic control.

Expression of Gpx3 was also significantly reduced in both adipose tissue and skeletal muscle under E2-deficient conditions. As a key antioxidant enzyme, Gpx3 protects against oxidative stress by reducing hydrogen peroxide and lipid hydroperoxides, thus preserving redox balance(82). Its downregulation is consistent with increased oxidative stress and mitochondrial dysfunction observed in E2-deficient states(74). Supporting these findings, prior studies have shown Gpx3 is E2-responsive in skeletal muscle(48), and our previous work demonstrated that ERα overexpression in adipose tissue upregulates Gpx3 (37). These observations and those of others further highlight Gpx3 as a potential therapeutic target for mitigating E2 deficiency–related metabolic dysfunction(83–85).

Finally, the consistent downregulation of Irs1 across both adipose tissue and skeletal muscle underscores a systemic impairment in insulin signaling associated with E2 loss. While skeletal muscle appears less transcriptionally sensitive to E2 deficiency than adipose tissue, it still undergoes key gene expression changes affecting metabolism and redox regulation. Altogether, our findings support a model in which E2 deficiency impairs glucose metabolism, disrupts redox and mitochondrial function, and promotes inflammation across multiple tissues, thereby contributing to insulin resistance and broader metabolic dysfunction.

Inflammation and Metabolic Dysfunction: Inactivity vs. Direct E2 Effects

A critical question is whether the observed inflammation and impaired metabolism result indirectly from reduced physical activity, which drives adiposity and subsequent metabolic dysfunction, or from direct effects of E2 loss. Reduced activity, a known driver of adiposity and inflammation, may contribute, as both models exhibited significant declines in movement. However, E2 directly regulates gene expression via E2 response elements, influencing metabolic and inflammatory pathways(86). In support of this, we have previously shown that manipulation of adipose tissue ERα can incur significant gene expression changes, particularly with respect to inflammation, without necessarily impacting body weight(37) This suggests that both reduced activity and direct E2 loss likely contribute to the observed inflammation and metabolic dysfunction in E2 deficiency, with their relative roles requiring further exploration.

Study Limitations

Several limitations of the present study should be acknowledged. First, sham-surgery controls were not included in the long-term dietary feeding experiments. The extended duration of dietary feeding (17 weeks) makes it unlikely that transient surgical effects would meaningfully contribute to long-term metabolic phenotypes. This interpretation is consistent with prior long-term OVX studies demonstrating that sustained metabolic dysfunction arises from estrogen deficiency rather than prolonged effects of surgery (12, 60, 87), which we have also shown (39). Nonetheless, the absence of sham-surgery controls limits the ability to completely exclude contributions of surgical stress to OVX-related outcomes.

Second, not all experimental groups were included in every assay, reflecting the distinct aims of individual experiments. Specifically, some assays were conducted under LFD conditions to assess baseline effects of estrogen deficiency (indirect calorimetry, behavioral phenotyping, and voluntary wheel running), whereas longitudinal experiments focused on HFD–induced metabolic dysfunction. While this design limits certain direct comparisons across assays, the convergence of physiological, behavioral, and molecular findings across complementary approaches supports the overall conclusions. Future studies incorporating sham controls and expanded metabolic and molecular analyses across dietary conditions will further strengthen these findings.

Third, histological analyses of liver and adipose tissue were not performed in the present study. While biochemical and physiological measures provide strong evidence of diet- and genotype-dependent metabolic changes, histological validation would further strengthen interpretation of tissue-specific phenotypes.

Conclusion and Future Directions

This study provides a comprehensive comparison of congenital/complete and adult-onset/partial E2 deficiency models, revealing shared and distinct behavioral, metabolic, and molecular consequences. Notably, congenital E2 deficiency led to more severe impairments to insulin sensitivity compared to adult-onset deficiency, likely due to complete E2 loss in AROM KO mice versus partial retention in OVX mice. The unexpected finding of comparable total energy expenditure despite reduced activity suggests compensatory mechanisms and limitations of indirect calorimetry, warranting further investigation. While both reduced physical activity and direct E2 loss likely contribute to inflammation and metabolic dysfunction, their relative roles remain inconclusive. These findings validate AROM KO and OVX as robust models for studying E2 deficiency, with AROM KO better suited for severe metabolic impairments and OVX better mimicking adult-onset E2 deficiency (menopause). Future research should incorporate direct calorimetry and gut microbiome analyses to better resolve potential differences in energy expenditure. Additionally, exploring the interaction between activity-driven and E2-mediated molecular pathways may help refine therapeutic strategies for mitigating health consequences associated with E2 deficiency.

Supplementary Material

02

Supplementary Table 1. Indirect calorimetry and behavioral phenotyping of female (F) wildtype (WT) and WT ovariectomized (OVX) mice performed at thermoneutrality (n=7–8); RER = Respiratory exchange ratio.

03

Supplementary Figure 1. No difference in fecal energy content with E2 deficiency. Fecal pellets were collected from individually housed F WT and WT OVX mice (n = 5–6), and energy content was measured using bomb calorimetry.