Age, Sex, and Depot-Specific Differences in Adipose Tissue Estrogen Receptors in Individuals with Obesity

Abstract

Objectives:

Examine the effects of sex and menopausal status on depot-specific white adipose tissue estrogen signaling in age-matched men and women with morbid obesity.

Methods:

Twenty-eight premenopausal females, 16 postmenopausal females, and 27 age-matched men undergoing bariatric surgery were compared for omental (OM) and abdominal subcutaneous (SQ) adipose tissue genes and proteins.

Results:

With the exception of fasting non-esterified fatty acids (NEFA) being higher in females (p<0.01), no differences were found in other indicators of glucose and lipid metabolism. In OM, estrogen receptor (ER) β levels were higher in older females than younger females and older males (sex*age interaction, p<0.01), and aromatase expression was higher in older males than older females (p<0.05). In SQ, women had lower expression of ERβ than men (p<0.05). Protein content of ERα and ERβ were highly correlated with the mitochondrial protein, uncoupling protein-1 across sexes and ages (p<0.001). Age increased SQ inflammatory gene expression in both sexes.

Conclusion:

In morbid obesity, sex and age affect adipose tissue estrogen receptors, lipid metabolism, mitochondrial UCP1, and inflammatory expression in an adipose tissue depot-dependent manner. SQ adipose tissue immunometabolic profile is heavily influenced by age and menopause status more so than OM adipose tissue.

Introduction

Excess adiposity and adipose tissue (AT) dysfunction are strongly linked to the development and progression of metabolic diseases, including cardiovascular disease and diabetes (1, 2). White AT is primarily stored viscerally and subcutaneously. Sex-specific patterns of AT distribution exist such that males tend to store more fat viscerally (android body shape) while females tend to store more fat subcutaneously (gynoid body shape). The traditional female pattern is thought to be associated with protection against obesity-induced metabolic dysfunction. The sexual dimorphic deposition of AT highlights its sex hormone influence. Estrogen receptors (ER) are expressed in AT, and estrogen has been shown to influence cardiometabolic health, with many studies illuminating that loss of estrogen with menopause is associated with increases in central adiposity, insulin resistance, and cardiovascular disease (3–5). In the post-menopausal state, however, estrogen production is not completely eliminated as estrogens are produced extragonadal in AT (6).

Estrogen receptor-mediated signaling affects AT physiological function, including regulation of glucose and lipid metabolism (7, 8). Estrogen receptor alpha (ERα) and beta (ERβ), are both expressed in AT, yet have distinct signaling pathways. Adipose tissue ERα knockout mice have increased adiposity, insulin resistance, and inflammation, suggesting that ERα signaling positively influences insulin sensitivity and cardiometabolic health (9–11). The role of ERβ on AT and systemic metabolism is less clear, but recent findings suggest anti-lipogenic/anti-obesity effects (12). Some rodent studies show that ERβ deletion protects female mice against diet-induced insulin resistance and glucose intolerance (11, 12), while other studies show that ERβ null mice are more susceptible to obesity, yet still protected against insulin resistance (12).

Adipose tissue is an estrogen sensitive tissue which plays a critical role in glucose and lipid homeostasis. The purpose of this study was to determine the effects of age, sex, menopausal status and depot location on AT isoform-specific ER expression in age-matched men and women with morbid obesity undergoing bariatric surgery. Additionally, we examined how differences in AT ER expression may be related to sex, age, and depot-specific differences in AT immunometabolic characteristics among this cohort of individuals with morbid obesity.

Methods

Participants

Individuals undergoing bariatric surgery were recruited from the University of Missouri Hospital and affiliated Columbia Bariatric Associates. Written informed consent was obtained from all participants during a pre-operative education class. Seventy-one patients (44 females, 27 males) agreed to blood and AT collection from the visceral cavity, specifically the omentum (OMAT), during surgery; 30 patients (17 females, 13 males) agreed to an additional subcutaneous AT (SQAT) sample collection during surgery. All patients followed a liquid diet for 1 week prior to surgery, as prescribed by Columbia Surgical Associates as part of surgical mandate.

Menopausal Status

Young females (28–55 years old) and older female (45–73 years old) groups were established via FSH concentrations (young female: 7.0±0.8 mIU/mL; older female: 44.8±5.1 mIU/mL), age, and a questionnaire inquiring about menstruation, irregularity, peri-menopausal symptoms, age at menopause, and medically-induced menopause. Thus, older females were postmenopausal. The menopause questionnaire was made available online through Research Electronic Data Capture (RedCap) (13, 14) and also mailed to patients following surgery. Females between the ages of 45–55 years were excluded until FSH levels and/or questionnaire confirmation of menopausal status were obtained.

Sample Collection

Blood samples were collected in 10 mL EDTA-coated tubes at pre-op or during surgery. Blood samples were spun at 4°C for 15 minutes and then plasma was aliquoted and stored at −80°C until further processing. AT samples were collected at time of surgery. Once the sample was received from surgical staff, AT was aliquoted for formalin fixed for histological analysis and 1–3 samples were flash frozen with liquid nitrogen.

Blood Chemistry

Circulating fasting plasma glucose levels (Infinity Glucose Hexokinase, Thermo Fisher), total cholesterol (Infinity Cholesterol, Thermo Fisher), triglycerides (Infinity Triglycerides, Thermo Fisher), HDL-C (Infinity Cholesterol, Thermo Fisher), and non-esterified free fatty acids (Wako NEFA-HR(2), Wako Diagnostics) were quantified with in vitro enzymatic colorimetric method assays. FSH, insulin, TNFα, and IL6 concentrations were determined using a MILLIPLEX magnetic bead-based quantitative multiplex immunoassay with the MAGPIX instrumentation (Millipore, Billerica, MA).

Western Blotting

Omental and subcutaneous AT samples were homogenized in a Triton X-100 buffer containing protease and phosphatase inhibitors (TissueLyser LT, Qiagen). Triton X-100 tissue lysates were used to produce Western blot-ready Laemmli samples. Protein samples (10 μg/lane) were separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and probed with the following primary antibodies: Oxidative phosphorylation (OxPhos) complexes I through V of the electron transport chain (1:2000, MitoProfile Total OxPhos Rodent WB Antibody Cocktail, Abcam, ab110413); ERα (1:1000, Abcam, ab75635); ER β(1:1000, Abcam, ab3577); GLUT4 (1:1000, Cell Signaling, #2213); HSL (1:1000, Cell Signaling, #4107); Phospho-HSL (1:1000, Cell Signaling, #4126); Acetyl CoA Carboxylase (1:1000, Cell Signaling, #3662); Uncoupling protein 1 (1:1000, Sigma, U6328); Nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB) (1:1000, Cell Signaling, #2390); and protein kinase B (Akt) (1:500, Cell Signaling, #4691).

Intensity of individual protein bands was quantified using FluoroChem HD2 (AlphaView, version 3.4.0.0) and were expressed as a ratio to total protein.

RNA Isolation, q-PCR

Omental AT and SQ AT samples were homogenized in TRIzol solution using a tissue homogenizer (TissueLyser LT, Qiagen, Valencia, CA). Total RNA was isolated with Qiagen’s RNeasy Lipid Tissue Mini Kit (Qiagen, Germany) and analyzed using a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE) to assess purity and concentration. First-strand cDNA was synthesized from total RNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Carlsbad, CA). Quantitative real-time PCR was performed as previously described (15) using the ABI StepOne Plus sequence detection system (Applied Biosystems). Primer sequences were designed using NCBI Primer Design tool and purchased from IDT (Coralville, IA). Beta Actin was used as the house-keeping control gene and mRNA expression values are presented relative to the young females. Forward and reverse primer sequences are provided in Table 1.

Table 1.

Real-time RT-PCR primer sequences

Gene	Forward	Reverse
ADRB3	5′-CCAGGTGATTTGGGAGACCC-3′	5′-CACGTTGGTCATGGTCTGGA-3′
Aromatase	5′-GGTGAGAGAGACATAAAGATTG-3′	5′-TTCAGGATAATGTTTGTCCC-3′
ATGL	5′-TATCCCACTTCAACTCCAAG-3′	5′-GTGTTCTTAAGCTCATAGAGTG-3′
β-actin	5′-AGACCTGTACGCCAACACAG-3′	5′-TTCTGCATCCTGTCGGCAAT-3′
ERα	5′-CTGGGACTGCACTTGCTCC-3′	5′-CAGGGCAGAAGGCTCAGAAA-3′
ERβ	5′-AAATCTTTGACATGCTCCTG	5′-AGGGTACATACTGGAATTGAG-3′
FGF21	5′-TGATGCCCAGCAGACAGAAG-3′	5′-GTGGGCTTCGGACTGGTAAA-3′
GLUT4	5′-TAGGCTCCGAAGATGGGGAA-3′	5′-AAAAGATGGCCACGGAGAGG-3′
HSL	5′-CTATGCTGGTGCAAAGAC-3′	5′-CTCCAGGAAGGAGTTGAG-3′
IL1β	5′-TCGCCAGTGAAATGATGGCT-3′	5′-GGTCGGAGATTCGTAGCTGG-3′
IL6	5′-GCAGAAAAAGGCAAAGAATC-3′	5′-CTACATTTGCCGAAGAGC-3′
Leptin	5′-AACCCTGTGCGGATTCTTGT-3′	5′-GGAGACTGACTGCGTGTGTG-3′
LPL	5′-AGTAGCAGAGTCCGTGGCTA-3′	5′-GGGACCCTCTGGTGAATGTG-3′
NFκB	5′GACAACTATGAGGTCTCTGG-3′	5′-ATCACTTCAATTGCTTCGG-3′
PGC1α	5′-GTTGCCTGCATGAGTGTGTG-3′	5′-TAGAGACGGCTCTTCTGCCT-3′
TNFα	5′-GACAGGCCTGTAGCCCATGT-3′	5′-GGAGGTTGACCTTGGTCTGG-3′
UCP1	5′-ACAGCACCTAGTTTAGGAAG-3′	5′-CTGTACGCATTATAAGTCCC-3′

Histology

Formalin-fixed OM and SQ samples were processed through paraffin embedment, sectioned at 4 μm, and stained with hematoxylin and eosin (H&E) for morphometric assessment. Sections were evaluated using an Olympus BX60 photomicroscope (Olympus, Melville, NY) and photographs were taken at 20x magnification via Spot Insight digital camera (Diagnostic Instruments, Inc., Sterling Heights, MI). Adipocyte size was calculated based on 50 adipocytes/depot obtained from 1–4 images acquired at 20x magnification. Cross-sectional areas of the adipocytes were obtained from perimeter tracings using Image J software.

Statistical Analysis

The statistical analyses were performed using SPSS statistical software, version 21 (IBM, Inc.). To establish sample size for our primary outcome variable, ER alpha protein expression in white AT, we used data from a rodent study because prior human studies were not available. We determined a valid minimum sample of n=8/group. This was based on a male/female comparison: male: 0.46±0.40 v. female: 1.24±0.4 standard deviation; p=0.004; alpha set at 0.05 and power set at 0.8. The number of subjects/group was increased s we recognize there is greater heterogeneity in humans than in animal models. Two-way analysis of variance (ANOVA) was performed to determine differences by age, sex, and age*sex interactions. Tukey post hoc tests were conducted following significant interactions to determine specific between-group differences while controlling for multiple comparisons. Statistical significance was p<0.05.

Results

Subject characteristics and adipose tissue histology

Seventy-one adults undergoing bariatric surgery consented to AT collection while undergoing surgery; detailed anthropometrics and baseline characteristics between age-matched men and women are presented in Table 2. The majority of patients were self-classified as White (Caucasian) with one older female responding as “some other race”, one young female as Black/African American, and one older male as “American Indian, Eskimo or Aleut.” Male patients were taller, weighed more, and had greater a BMI than female counterparts (sex: p<0.05). Mean age for the younger and older groups was 38.9±1.1 and 55.5±1.3 years, respectively. FSH concentrations for young and old female groups were 7.0±0.8 and 44.9±5.1 mIU/mL (P<0.01), respectively, validating group differences in menopause status. Circulating FFA levels were higher in females than males (sex: p<0.01). Circulating levels of glucose, insulin, cholesterol, triglycerides, and HDL-c concentrations were similar between groups (Table 2). Male OM AT mean cell size (6382±511 μm²) was greater than that from females (4868±473 μm², p=0.05) (Table 2). No sex or age differences were observed in mean SQ AT cell size. Paired samples between depots showed that OM cell size was smaller than SQ, 5540.6±363 vs. 6875.5±345 μm², respectively, p=0.01).

Table 2.

Subject Characteristics

	Female		Male
	Younger	Older	Younger	Older
N	28	16	14	13
SQAT samples	8	9	7	6
Age (years) **	40.0 ± 1.3	56.4 ± 2.2	37.9 ± 1.3	54.6 ± 1.6
BMI (kg/m2) ‡	46.5 ± 1.1	43.6 ± 1.6	49.5 ± 1.8	51.9 ± 2.2
Circulating Markers
FSH (mIU/mL) **	7.0 ± 0.8	44.8 ± 5.1	NA	NA
Glucose (mmol/L)	5.8 ± 0.2	6.2 ± 0.4	6.1 ± 0.4	6.5 ± 0.5
Insulin (μU/L)	22.6 ± 2.8	23.3 ± 4.8	27.3 ± 4.1	23.7 ± 1.8
Cholesterol (mg/dL)	160.4 ± 4.7	173.6 ± 15.0	160.2 ± 9.1	162.8 ± 9.4
Triglycerides (mg/dL)	87.9 ± 6.9	102.3 ± 12.9	104.4 ± 21.2	108.6 ± 8.7
HDL-c (mg/dL)	58.7 ± 2.1	62.8 ± 4.3	54.4 ± 2.7	56.5 ± 3.4
FFA (mmol/L) †	1.08 ± 0.05	1.11 ± 0.07	0.92 ± 0.06	0.97 ± 0.05
TNFα (pg/mL)	4.8 ± 0.6	5.5 ±0.8	6.2 ± 0.7	6.2 ±0.7
Insulin Resistance Indexes
HOMA-IR	6.3 ± 1.1	7.1 ± 1.8	8.1 ± 1.7	7.0 ± 0.8
Adipo-IR	164 ± 22	198 ± 56	178 ± 35	162 ±16
Adipose Tissue Cell Size (μm2)
OMAT †	4693 ± 590	5044 ± 1003	6605 ± 695	6159 ± 362
SQAT	7064 ± 400	7307 ± 1048	7047 ± 732	6009 ± 737

Values are means ± SEM. BMI, body mass index; OMAT, omental adipose tissue; SQAT, subcutaneous adipose tissue; HOMA-IR, homeostatic model assessment of insulin resistance; HDL-c, high density lipoprotein cholesterol; FFA, free fatty acids; Adipo-IR, adipose tissue insulin resistance (FFA (mmol/L)

^*Insulin (pmol/L)); FSH, follicle-stimulating hormone.

^**P<0.01 by age;

^†P<0.05 by sex;

^‡P<0.01 by sex

Omental adipose tissue estrogen signaling

In the omental AT (in humans, the visceral depot), no differences were seen in ERα gene expression between ages or sexes, but ERβ gene expression was greater in female postmenopausal women compared to young females and old males (sex*age: p=0.01, Figure 1A). No significant protein content differences were observed for either ERα or ERβ (Figure 1C). Aromatase (i.e., the enzyme responsible for estrogen production in AT) gene expression was lower in older females compared to older males, (sex*age interaction, p<0.01). There was a trend for a sex*age interaction (p=0.06) for the ratio of relative expression ratio of ERα:ERβ, however, such that older females tended to have a lower ratio than older males (Figure 1E).

Figure 1.

ER gene expression and protein content.

OMAT gene expression (A), SQ adipose tissue gene expression (B), OM adipose tissue protein content (C), and SQ adipose tissue protein content (D); OM ratio of ER subtype expression (E); SQ ratio of ER subtype expression (F). ERα - estrogen receptor alpha; ERβ - estrogen receptor beta. Values are means ± SEM.; ^SP<0.05 main effect of sex; ^AP<0.05 main effect of age; ^S*^AP<0.05 sex by age interaction; ^&p<0.05 Tukey post hoc different than Female-Young; and ^#p<0.05 Tukey post hoc different than Male-Old.

Subcutaneous adipose tissue estrogen signaling

Transcript levels of ERα trended toward being affected by age, with greater expression among older subjects in both sexes (age: p=0.06); whereas, ERβ transcript levels were not affected by age, yet higher levels were observed among females than males (sex: p<0.05, Figure 1B). Similar to OM AT, protein content of SQ AT ER subtypes did not differ significantly between sexes or ages (Figure 1D). Aromatase mRNA expression did not differ between groups, but tended to follow the same pattern of expression observed in OM AT (Figure 1B). On the other hand, the gene expression ratio of ERα to ERβ was different from that observed in OM AT, in that it was higher in males compared to females (p<0.05) (Figure 1F).

Omental adipose tissue immunometabolic phenotype

Proteins related to insulin-stimulated glucose uptake (i.e., Akt, GLUT4) were not significantly affected by age or sex in the visceral depot, although a sex*age interaction was observed for GLUT4, which tended to be reduced with age in males but not females (Figure 2A). Leptin mRNA was higher in males than females, and lower in older versus younger individuals of both sexes (Figure 2B). Acetyl CoA Carboxylase (ACC) protein content, the rate limiting enzyme of fat synthesis, followed the same pattern as GLUT4 (sex*age: p<0.05) such that ACC tended to be reduced with age in males but not females (no post hoc differences) (Figure 2A). Total HSL protein content, rate limiting enzyme of adipocyte lipolysis, was not different between young males and females, but was higher in older females compared to older males (sex*age: p<0.05) (Figure 2A). LPL (encoding lipoprotein lipase, the enzyme responsible for fatty acid uptake) increased with age in both sexes (age: p<0.05) but genes associated with lipolysis were not different (Figure 2B). UCP1 protein (mitochondrial protein) was unchanged by age and sex (Figure 2A). To assess inflammation, NFκB protein content and gene expression were measured, but no group differences were observed (Figure 2A/B, respectively). Sex*age interactions (p<0.05) were observed for gene expression of both inflammatory cytokines, TNFα and IL1β (Figure 2B). IL1β tended to be elevated in older males compared to older females (p=0.052), whereas levels of TNFα and IL6 were relatively consistent among the groups.

Figure 2.

Omental immunometabolic characteristics.

OM adipose tissue protein content (A), and gene expression (mRNA) (B). Akt - protein kinase B; GLUT4, glucose transporter 4; NFkB - nuclear factor kappa-light-chain-enhancer of activated B cells; ACC - acetyl CoA carboxylase; HSL - hormone sensitive lipase; UCP1 - uncoupling protein 1; TNFα - tumor necrosis factor alpha; IL6 - interleukin 6; IL1β - interleukin 1 beta; LPL - lipoprotein lipase; ATGL - adipose tissue triglyceride lipase. Values are means ± SEM; ^SP<0.05 main effect of sex; ^AP<0.055 main effect of age; ^S*^AP<0.05 sex by age interaction; ^#p<0.05 Tukey post hoc different than Male-Old.

Subcutaneous adipose tissue immunometabolic phenotype

As was the case in the visceral depot, the insulin signaling proteins, Akt and GLUT4 were similar in SQ AT among the four groups (Figure 3A). Leptin mRNA levels were higher in females compared to males (sex: p<0.05; Figure 3B). The pattern of ACC and HSL protein content between groups mirrored those observed in the OM AT depot with a sex*age interaction (p<0.05) such that protein content tended to increase in females with age but decrease with age in males (no post-hoc differences observed, Figure 3A). Gene expression of HSL followed a similar pattern as HSL protein content across age and sex with a significant sex*age interaction (p<0.05; Figure 3B). ATGL (a key lipolytic enzyme) mRNA tended to be lower in males (sex: p=0.06). There was a sex*age interaction (p<0.05) for UCP1 protein content, such that older females having a greater content than older males (sex*age: p<0.05, Figure 3A). Surprisingly, the pattern of UCP1 gene expression across age and sex was opposite to that found in UCP1 protein content (sex*age interaction, p<0.05). NFκB protein content tended to increase with age in females and decrease with age in males (sex*age: p<0.05), but gene expression did not differ. Inflammatory gene expression increased with age in both sexes, with IL6 and IL1β significantly increased (age: p<0.05; Figure 3B) and there was a trend for TNFα to increase (p=0.08).

Figure 3.

Subcutaneous immunometabolic characteristics.

SQAT protein content (A), and gene expression (mRNA) (B). Akt - protein kinase B; GLUT4 - glucose transporter 4; NFkB - nuclear factor kappa-light-chain-enhancer of activated B cells; ACC - acetyl CoA carboxylase; HSL - hormone sensitive lipase; UCP1 - uncoupling protein 1; TNFα - tumor necrosis factor alpha; IL6 - interleukin 6; IL1β - interleukin 1 beta; LPL - lipoprotein lipase; ATGL - adipose tissue triglyceride lipase. Values are means ± SEM; ^SP<0.05 main effect of sex; ^AP<0.055 main effect of age; ^S*^AP<0.05 sex by age interaction; ^#p<0.05 Tukey post hoc different than Male-Old.

Depot and sex-specific correlations among variables

Given the relatively homogeneous sample of subjects with obesity, we performed correlation analyses to determine potential relationships among the variables measured. AT protein content of the mitochondrial membrane protein, UCP1 correlated positively with ERα in both AT depots (SQ: p=0.001, r=0.578; OM: p=0.042, r=0.251) (Figure 4A) and with ERβ in both depots (OM: p<0.001, r=0.461; SQ: p<0.001, r=0.725; Figure 4B). Subcutaneous AT UCP1 protein content also strongly correlated with circulating FSH concentrations (r = 0.618, p=0.001). Besides UCP1, FSH correlated positivity with the circulating inflammatory cytokine, TNFα (r = 0.335, p=0.03).

Figure 4.

Pearson Correlations for omental and subcutaneous adipose tissue Protein content of UCP1 and protein content of ERα (A), protein content of UCP1 and ERβ (B), Leptin expression and the ratio of ER subtypes in subcutaneous (SQ) and omental (OM) adipose tissue (C). UCP1 - uncoupling protein 1; ERα - estrogen receptor alpha; ERβ - estrogen receptor beta; SQ-OM ratio, gene expression ratio of SQ(ERα:ERβ):OM(ERα:ERβ).

The relative ratio for gene expression of ERα to ERβ in SQ AT versus OM AT has been previously shown by Shin et al. (16) to correlate with anthropometric measures of obesity, so we explored similar correlations in our study population. The SQ-OM ratio of ER subtypes was not correlated with BMI (p=0.825), which contrasts the findings of Shin et al. who studied a population with less severe obesity. Interestingly, this ratio of ER subtypes reported by Shin et al. correlated strongly with OM leptin mRNA, and we also found a modest correlation (p=0.020, r=0.444) in our population (Figure 4C) in OM AT.

Discussion

This study in humans is the first to our knowledge to examine how morbid obesity affects ER expression in the two major white AT depots, and the potential impact that sex and age, including menopausal state, may have on such differences. The subjects in this study were all morbidly obese, and the groups did not differ in variables related to metabolic health (e.g., HOMA-IR, circulating TNFα). The major findings are: (1) Gene expressions of ER and aromatase showed robust differences by age and sex in OM and SQ AT, while neither depot had any changes in protein content for ER; (2) In SQ AT, aging in both sexes affected expression levels of the mitochondrial protein, UCP1, as well as markers of lipid metabolism with a fairly consistent sexually-dimorphic pattern of gene and protein expression; and (3) Remarkably strong relationships were observed in both AT depots, across ages, and in both sexes between expression levels of ER and UCP1. Furthermore, circulating FSH levels, recently shown to affect inflammation, fat accumulation and AT metabolism (17), correlated with circulating inflammation (TNFα) and AT UCP1. Although causal relationships cannot be determined by this cross-sectional study, these novel findings may shed light on mechanisms responsible for obesity-mediated changes in AT ER signaling, and the impact these changes may have on AT and systemic metabolic health among individuals with morbid obesity.

Despite the fact that age affected the gene expression levels of ERs in visceral fat, protein content was not affected (Figure 1). Our data contrast that of McInnes et al. (18) who observed two-fold higher ERβ mRNA expression in healthy (BMI<25 kg/m²) premenopausal women compared to healthy postmenopausal women within SQ AT. In the present study, age did not influence SQ adipose tissue ERβ mRNA expression in men or women. These data, however, support the findings of Dieudonné et al. (19) who reported healthy premenopausal women expressed ERβ to a greater extent than that of healthy males (Figure 1A, B). Additionally, Park et al. (20) explored the importance of ERα:ERβ in AT distribution and function and reported that postmenopausal women had lower ratio of ERα:ERβ than premenopausal women within the SQ AT in both abdominal and femoral regions. Here we show that, in SQ AT of morbidly obese women, no changes in the gene expression ratio occurred with menopause (young vs old females). Additionally, we show the opposite relationship within OM AT, where menopausal status tended to decreased this ratio. These findings are discrepant with those reported by the McInnes and Park groups and may be due in part to the health status of the study populations; our study utilized women with morbid obesity undergoing bariatric surgery as opposed to healthy women in the previous reports, providing evidence that severity of obesity plays an important role in estrogen regulation in AT metabolism. Further, our data are the first to compare these variables in both men and women with morbid obesity; this is important because it suggests that these endocrine/immunometabolic changes in AT are greatly affected by obesity status, and perhaps differ by obesity classification.

A novel finding of the present study was that a significantly greater ERβ gene expression was observed in both OM and SQ AT depots in women compared to men, although these differences were not reflected by increased protein content. However, robust novel correlations were observed between ERβ (and ERα) protein and the mitochondrial protein, UCP1 across the entire cohort of individuals, suggesting a relationship in AT between changes in estrogen signaling and mitochondrial activity. The metabolic implications of this finding are not yet clear, but certainly warrant further investigation. Interestingly, protein levels of UCP1 also correlated with circulating levels of FSH, further suggesting a relationship between circulating sex hormones and AT mitochondrial metabolism.

Emerging animal research implicates the ERs as playing important roles in obesity and metabolic health of AT. Genetic knock out of ERα results in increased adiposity and metabolic dysfunction (21), while genetic knock out of ERβ does not, although these animals may be more susceptible to obesity. Recent studies using ERβ selective ligands have implicated ERβ as a potential therapeutic target for obesity and associated metabolic disorders (22, 23). Other reports demonstrate that AT-specific ERα deletion causes a similar adverse phenotype especially in males (11, 24). In rodent and pig models, females have greater AT expression of ERα (10); however, we found neither AT ERα content nor expression levels were different between men and women with morbid obesity.

Aromatase is expressed in AT and is thought to be activated by inflammation (25, 26). Thus, its levels increase with obesity, presumably increasing levels of estrogen in circulation. In terms of depot differences, aromatase expression has been shown to be elevated in SQ AT in both men and women compared to OM AT (27–29). Interestingly, increased aromatase activity within AT has been shown to improve insulin sensitivity and inflammatory profiles for male mice (30), likely due to the positive effects of estrogen signaling noted above. Alternatively, aromatase deficiency causes obesity and insulin resistance in mice and humans (31, 32). In the present study, OM AT aromatase expression was upregulated in males compared to females. In terms of a possible association with insulin sensitivity, we found no differences in insulin signaling proteins, Akt and/or GLUT4. Although we did not directly assess insulin sensitivity of AT, the relationship between insulin sensitivity and aromatase expression may be dependent on obesity state and/or aromatase activity. Additional studies are necessary to tease these things out.

White AT has been well regarded over the past few decades as a major contributor to the obesity-associated elevated inflammatory profile. Generally, visceral AT is considered the depot responsible for inflammation induced by obesity and the related health issues (33). Also, it is thought that inflammation plays an important role in the complex process of aging (34–36). With severe obesity, we observed an age dependent upregulation of expression for inflammatory markers primarily in the subcutaneous depot. Inflammatory gene expression in both OM AT and SQ AT depots were not strongly correlated with systemic insulin concentrations or indexes of resistance, which contrasts many clinical studies displaying clear associations (37, 38). We did find though, that circulating TNFα was correlated positively (albeit only moderately; data not shown) with HOMA-IR, which is consistent with the literature (36, 39). Also consistent with the emerging literature on this topic (40), circulating TNFα was positively correlated with FSH.

Earlier work (28, 41) has shown that men and women with obesity have greater expression of HSL in the subcutaneous depot compared the visceral depot, while we observed that OM AT HSL expression was upregulated in females with morbid obesity. This parallels the elevated circulating FFA levels, which are well known to associate with inflammatory status. Despite females having elevated OM AT HSL content and increased circulating FFAs, females did not exhibit greater inflammatory phenotype within the OMAT.

Study Limitations:

Since subjects were patients for bariatric surgery, no healthy controls were available for comparison. Also, weight loss can substantially impact metabolism, thus it is important to acknowledge that patients were prescribed a liquid diet for the week prior to surgery and may have had some weight loss. Inconsistencies in the adherence to this diet protocol are possible which has potential for impacting the study’s outcomes. Comorbidities often seen with severe obesity were also seen within this study’s population. Covariate analysis of prior diagnosis of insulin dependent and independent diabetes, as well as other diseases and complications, did not statistically influence gene expression or protein content of the majority of variables measured, thus was not included within this discussion. A strength of this study was the separation of women by menopausal status as well as exclusion of patients with questionable or perimenopausal status through the use of a detailed questionnaire about menstrual events, age, and assessment of FSH for menopause classification. It is important to note that contraceptive use is common in young females, yet was difficult for us to find in the medical records, so it is unknown if any of the young females were currently using contraceptives during the time of surgery or leading up to surgery. Additionally, one postmenopausal subject was using hormone replacement therapy, estradiol, during the time of surgery. This subject was not considered a statistical outlier in any primary outcome variable and was therefore retained within subject population. Use of AT homogenates as opposed to adipocyte specific measures were explored. Previous work (42, 43) has suggested that the stromal vascular fraction is a major contributor to the differences between visceral and subcutaneous AT inflammatory status, lipid metabolism, and adipogenesis. Further in animal models, bone-marrow derived progenitor cells can be a potential source of adipocytes. In individuals with morbid obesity this contribution could be significant (44), and may display higher expression of proinflammatory genes and lower expression of mitochondrial biogenesis and lipid oxidation (44). Further mice studies have shown that bone-marrow derived adipocytes can accumulate with age, be higher in female vs. male mice and be more prevalent in visceral than in subcutaneous fat (45). Thus, discrepancies between our findings and previous reports may be due to a greater accumulation of BMP-derived adipocytes in our morbidly obese subjects, resulting in resulting in greater heterogeneity in ER expression and characterization of this tissue between young and older individuals. Future studies may benefit from the exploring region-specific deposition of subcutaneous AT and considering the stromal vascular fraction composition of visceral and subcutaneous AT.

The current study demonstrated the effects of age, sex, menopausal status, and depot location on AT isoform-specific ER expression in age-matched men and women with morbid obesityIn the setting of morbid obesity, this study demonstrates for the first time that regional AT depot differences exist. In particular, the SQ AT immunometabolic profile is heavily influenced by age and menopause status. Further research is needed to examine mechanism by which sex, age, and menopausal state may affect weight loss-induced changes in these AT immunometabolic characteristics, and the effects of such AT-specific effects on systemic metabolic health and disease susceptibility.

What is already known about this subject?

• Sex-specific patterns of AT deposition exist.

• A bidirectional relationship exists, such that obesity affects AT ER expression, and ER expression affects obesity (as demonstrated in ERa/ERb KO mouse models).

• Aging influences hormonal responses.

What are the new findings in your manuscript?

• Robust age, sex and depot-specific differences were observed in genes associated with AT estrogen signaling.

• Especially in subcutaneous AT, aging affected expression levels of the mitochondrial protein, UCP1, and markers of lipid metabolism in a sexually dimorphic manner.

• Remarkably strong relationships were observed in both AT depots, across ages, and in both sexes between expression levels of ER and UCP1.

How might your results change the direction of research or the focus of clinical practice?

• In obesity, sex, age, and menopausal state differentially affect AT immunometabolic profiles in a depot-dependent manner. Future work should identify how such AT changes relate to sex and age-specific obesity-associated comorbidities.

Study Importance questions.

• Sex-specific patterns of AT deposition exist, indicating that sex hormones directly influence AT.

• Sex and age affect AT immunometabolic profiles in a depot-dependent manner, and obesity status may affect how estrogen regulates AT.