Estradiol treatment or modest exercise improves hepatic health and mitochondrial outcomes in female mice following ovariectomy

Abstract

We recently reported that compared with males, female mice have increased hepatic mitochondrial respiratory capacity and are protected against high-fat diet-induced steatosis. Here, we sought to determine the role of estrogen in hepatic mitochondrial function, steatosis, and bile acid metabolism in female mice and investigate potential benefits of exercise in the absence or presence of estrogen via ovariectomy (OVX). Female C57BL mice (n = 6 per group) were randomly assigned to sham surgery (sham), ovariectomy (OVX), or OVX plus estradiol replacement therapy (OVX + Est). Half of the mice in each treatment group were sedentary (SED) or had access to voluntary wheel running (VWR). All mice were fed a high-fat diet (HFD) and were housed at thermoneutral temperatures. We assessed isolated hepatic mitochondrial respiratory capacity using the Oroboros O2k with both pyruvate and palmitoylcarnitine as substrates. As expected, OVX mice presented with greater hepatic steatosis, weight gain, and fat mass gain compared with sham and OVX + Est animals. Hepatic mitochondrial coupling (basal/state 3 respiration) with pyruvate was impaired following OVX, but both VWR and estradiol treatment rescued coupling to levels greater than or equal to sham animals. Estradiol and exercise also had different effects on liver electron transport chain protein expression depending on OVX status. Markers of bile acid metabolism and excretion were also impaired by ovariectomy but rescued with estradiol add-back. Together our data suggest that estrogen depletion impairs hepatic mitochondrial function and liver health, and that estradiol replacement and modest exercise can aid in rescuing this phenotype.

NEW & NOTEWORTHY OVX induces hepatic steatosis in sedentary mice which can be prevented by modest physical activity (VWR) and/or estradiol treatment. Estrogen impacts hepatic mitochondrial coupling in a substrate-specific manner. OVX mice have impaired fecal bile acid excretion, which was rescued with estradiol treatment.

INTRODUCTION

Epidemiological data in women show strong associations between estrogen action and liver metabolic health. Compared with age-matched men, premenopausal women have lower rates of high cholesterol (1), diabetes (2), and hypertension (3)—all risk factors that contribute to nonalcoholic fatty liver disease (NAFLD). However, following menopause, the rate of NAFLD in women increases to levels equal to or greater than that of men (4, 5). These data broadly point to female sex hormones as key mediators in the sex differences seen in the susceptibility to liver metabolic disease. However, many of the underlying mechanisms remain incompletely understood.

We have recently reported that compared with males, female mice have increased hepatic mitochondrial respiratory capacity (6). These inherent sex differences were more significant than the impact of diet content (low fat vs. high fat), high-fat diet duration (acute vs. chronic), or exercise, on hepatic mitochondrial respiration, coupling, and H₂O₂ emission (6–8). Additionally, the effects of sex on hepatic mitochondrial respiratory capacity were even stronger than the effect induced by heterozygosity for PGC1α (a driver of mitochondrial biogenesis) and complete knockout of BNIP3 (an important mitophagy protein) (6). It is well accepted that mitochondria play an integral role in hepatic health, regulating oxidative nutrient flux and oxidative stress, while contributing to numerous other normal functions of the liver (9). Also appreciated is that poor mitochondrial function and oxidative stress have been detected in liver tissues from patients with an array of metabolic disorders, ranging from steatosis and insulin resistance to nonalcoholic steatohepatitis (10). Given these data, it is not surprising that female mice with normal estrogen function appear protected from high-fat diet-induced steatosis (11).

Ovariectomy (the surgical removal of the ovaries; OVX) data in rats suggest that female protection against steatosis is dependent, at least in part, on estrogen. OVX has been shown to increase hepatic triglycerides but normalized with estradiol replacement treatment (12). Importantly, these OVX rodent studies mirror clinical observations of greater steatosis following estrogen withdrawal via menopause (13). Furthermore, this and other studies in mice have shown that OVX animals have impaired hepatic mitochondrial function, evidenced by decreased complex II specific activity (14), lower mRNA and protein levels of peroxisome proliferator-activated receptor gamma coactivator-1 beta (PGC1B), peroxisome proliferator-activated receptor gamma coactivator-1 beta (NRF1), and cyclooxygenase 1 (COX1), and reduced superoxide dismutase activity (12). Thus, it is possible that estrogen reduces the susceptibility of hepatic steatosis via modulation of liver mitochondrial function. One previous study found that compared with sham controls, OVX tended to increase hepatic triglycerides in young mice without impacting mitochondrial function, whereas OVX in older mice lowered mitochondrial respiration but did not have a significant impact on hepatic triglycerides (15). However, this study did not include an estradiol replacement group, making the independent effects of estradiol difficult to establish.

A commonly prescribed therapy for NAFLD is increased physical activity and/or exercise, and the reported reductions in hepatic lipid content are thought to arise from an upregulation in liver mitochondrial function (measured by fat oxidation or respiratory capacity) and decreased de novo lipogenesis (8, 16–18). However, the role of estrogen in the female hepatic response to exercise remains poorly characterized. Specifically, it is unknown whether exercise in female animals can lower/alter hepatic lipid content and improve liver mitochondrial function in the absence of estrogen (OVX). Moreover, it remains unknown if there is an additional metabolic benefit of exercise with estrogen replacement therapy or if estrogen replacement therapy alone is effective in reducing hepatic steatosis.

Cholesterol and bile acid synthesis are also central to the liver’s role of lipid flux and nutrient partitioning, and liver disease is associated with perturbations in these signaling pathways (19). Important for this present study, there are several reports suggesting that cholesterol and bile acid metabolism are sexually dimorphic in nature and tied to hepatic mitochondrial function (20, 21). When challenged with a high-fat diet, excess acetyl-CoA resulting from glycolysis and β-oxidation of fatty acids leads to upregulated citrate export from the mitochondria. Citrate then serves as fuel for de novo lipogenesis or cholesterol biosynthesis (and a portion of cholesterol is subsequently converted to bile acids). Beyond bile acid’s primary role of aiding in absorption of dietary fat, recent data have demonstrated that bile acids can also act as signaling molecules to regulate energy metabolism (19). Generally, greater bile acid production is beneficial to whole body metabolic health as bile acids prevent cholesterol and triglyceride accumulation and signaling through G protein-coupled receptors to regulate hepatic lipid and glucose metabolism and inhibit proinflammatory cytokine production by macrophages (22). Although directionality is not always in agreement, human trials and rodent models both report sex differences in bile acid levels (23, 24), suggesting bile acid metabolism may be a contributing mechanism to the sexual dimorphism seen in hepatic mitochondrial function and liver health. In particular, female mice tend to have higher circulating bile acids during their lifespan than males (25). Also of interest is that exercise upregulates the rate-limiting enzyme in bile acid synthesis cytochrome P450 family 7 subfamily A member 1 (CYP7a1) in OVX rats (26) and increases bile acid fecal loss in mice (27). Importantly, an upregulation of hepatic bile acid synthesis and fecal loss induced by CYP7a1 overexpression (28) or bile acid sequestrants protects against high-fat diet-induced hepatic steatosis (29). Still, the relationship between estrogen and exercise on hepatic bile acid production remains incompletely understood.

The main goal of this study was to investigate the effect of estrogen withdrawal via OVX on liver mitochondrial function, hepatic steatosis, and bile acid metabolism in female mice. Moreover, we aimed to determine whether estradiol replacement or modest exercise via voluntary wheel running could modulate this relationship. Our hypothesis was that OVX would lower hepatic mitochondrial respiration and coupling, increase hepatic steatosis, and impair markers of bile acid metabolism, and that exercise would have a more potent impact in the absence of estrogen (OVX vs. sham and OVX + Est).

MATERIALS AND METHODS

Animals

Female C57BL/6J mice (n = 6 per treatment group) were individually housed at thermoneutral temperatures (30°C) on a reverse light cycle (dark: 10:00–22:00), with ad libitum access to water and high-fat diet (research diets D12451: 45% kcal fat, 17% kcal sucrose, 1% cholesterol wt./wt.; 4.68 kcal/g). All animal protocols were approved by the Institutional Animal Care and Use Committee at the University of Kansas Medical Center (2018-2456).

Experimental Treatment

Experimental treatment groups (n = 6) were randomly assigned, with groups matched by initial body weight. As outlined in Fig. 1, 6-wk-old mice were single-housed and started the high-fat diet (HFD) on experimental day 1. Within a week (5–7 days) after initiation of the HFD, baseline body composition was determined via MRI (Model 900, EchoMRI). At 8 wk of age, animals underwent either sham surgery (sham) or bilateral ovariectomy (OVX). One week after surgery, half of the OVX animals started estradiol replacement (OVX + Est) via the implantation of a 36 µg/mL 17β-estradiol silastic capsule (20-mm tubing with 3-mm wood plug) as previously described (30). Estradiol replacement therapy continued for the remainder of the study (4 wk), and estradiol capsules were removed and replaced every other week. Throughout the duration of the experiment, an equal number of mice from each group (sham, OVX, OVX + Est) were assigned to a sedentary condition (SED) or were given access to a running wheel [voluntary wheel running (VWR); ENV-047V Med Associates Inc.]. Wheel running distances were recorded in real time and presented as average daily running distance. To control for our previously observed lower VWR distance in the OVX treatment groups (unpublished data), sham animals had restricted access to their wheels (6 h during the dark cycle), whereas OVX animals were unrestricted.

Figure 1.

Experimental timeline. OVX, ovariectomy; VWR, voluntary wheel running.

Anthropometrics

Body weight was collected once a week, whereas body composition was measured 1 wk before surgery (pre) and at the end of the study (post) via MRI (EchoMRI). At sacrifice, whole blood was collected via cardiac puncture, left to clot at room temperature, placed on ice for 10 min, and then centrifuged for serum separation. Endpoint serum nonesterified fatty acids (NEFAs) were determined via a commercially available assay (Wako Diagnostics).

Liver Steatosis

As done previously (31), hepatic triglyceride (TG) content was measured from flash-frozen liver tissue. Briefly, tissue was homogenized (TissueLyser II, Qiagen) in a lipid extraction solution (1:2 vol/vol methanol:chloroform) and then rotated continuously overnight. Centrifugation separated the organic phase, which was then removed and reconstituted with a butanol-Triton X solution following evaporation. Finally, TG content was measured with a commercially available kit (Sigma) and normalized to hepatic tissue wet weight. Fresh liver tissue was fixed in 10% formalin and paraffin embedded for histological analysis. Liver sections were routinely stained with hematoxylin and eosin (H&E) and imaged for examination of fat deposition (steatosis).

Fecal Bile Acids

For quantification of fecal bile acids, feces were first collected from the bedding of single-housed mouse cages at sacrifice and fecal matter was subsequently ground using an electric grinder. Roughly, 40–50 mg of ground feces were measured out and combined with 1 mL extraction buffer (7:11:0.1 vol/vol chloroform:isopropanol:Nonidit P-40). Samples were homogenized using a bead homogenizer (TissueLyser II, Qiagen) and rotated overnight at 4°C. Following centrifugation, the resulting supernatant was dried down at room temperature overnight. Extracted bile acids were dissolved in kit-provided Reagent 1, and quantification of total bile acids was performed via a commercially available colorimetric assay (Diazyme). Fecal bile acids were normalized to weight of ground fecal matter used in the extraction step.

Mitochondrial Isolation, Respiration, and Coupling

As previously described (6, 7), mitochondria were isolated from fresh liver tissue via several rounds of glass-on-glass homogenization in cold mitochondria isolation buffer (220 mM mannitol, 70 mM sucrose, 10 mM Tris, 1 mM EDTA, with pH adjusted to 7.4) and centrifugation. Resulting mitochondria were resuspended in 300–500 µL modified MiR05 mitochondrial respiration buffer (0.5 mM EGTA, 3 mM MgCl₂, 60 mM MES potassium salt, 20 mM glucose,10 mM KH₂PO₄, 20 mM HEPES, 110 mM sucrose, and 0.1% BSA, pH adjusted to 7.1). Mitochondria respiration was measured using high-resolution respirometry (Oroboros O2k-FluoRespirometer), as previously reported. Briefly, 25–50 µL hepatic mitochondrial isolate were added to chamber, and oxygen consumption was measured using starting substrates of l-palmitoylcarnitine (PC; 10 µM) or potassium-pyruvate (PYR; 5 mM) at basal (2 mM malate, 10 µM coenzyme A, and 2.5 mM l-carnitine), ADP-stimulated state 3 (2.5 mM adenosine 5′-disphosphate), ADP-stimulated plus succinate state 3S (10 mM succinate), and uncoupled (0.005 µM carbonyl cyanide-p-trifluoromethoxyphenylhydrazone injections until max) respiratory states. Mitochondrial coupling was estimated via the coupling control ratio [basal respiration divided by state 3 respiration (32)]. Hepatic mitochondrial respiration was analyzed with DatLab7 and normalized for volume and total protein in the mitochondrial isolate determined via bicinchoninic acid protein assay kit (Thermo Scientific).

Western Blotting

Flash frozen liver tissue was powdered and subsequently homogenized in homogenization buffer with phosphatase (phosphatase inhibitor cocktail 2, 3, Sigma) and protease (protease inhibitor cocktail tablet, Roche Diagnostics) inhibitors. Western blot Laemmli samples were separated via SDS-PAGE and transferred to a polyvinylidene difluoride membrane. Membranes were incubated with primary antibody solution for total OXPHOS (Abcam ab110413) and 4‐hydroxynonenal (4HNE; Abcam ab110413) at a concentration of 1:2,000, followed by secondary antibody solution at 1:10,000 (Cell Signaling). Membranes were imaged and quantified with Image Lab software (BioRad) and normalized to 0.1% amido-black total protein stain.

mRNA Expression

As done previously (33), RNA was isolated from tissue using the RNeasy Mini Kit (Qiagen), and cDNA was prepared using the ImProm-II RT system (Promega). cDNA concentration was read on a NanoDrop spectrophotometer (Thermo Scientific) and diluted to 10 ng/µL. Real-time quantitative PCR was determined using a QuantStudio 3 Real-Time PCR System (Thermo Scientific) and SYBR green mouse primers (Sigma; Table 1). All mRNA gene expression was normalized to the housekeeping gene Cyclophilin B (PPIB).

Table 1.

RT-PCR primer sequences

Gene	Forward Primer	Reverse primer
ACLY	CTCCAAGAAGCCAAATCTTATC	ATATTCATCAGCTTCCTCCC
CAT	CTCCATCAGGTTTCTTTCTTG	CAACAGGCAAGTTTTTGATG
COL1A1	AGCACGTCTGGTTTGGAGAG	ACATTAGGCGCAGGAAGGTC
CYP7A1	CTTGAGGATGGTTCCTATAAC	TTAAAAGTCAAAGGGTCTGG
EMR1	AGCTCCCATTCCCAGACTTC	TGCCATCAACTCATGATACCCT
HMGCR	GATAGCTGATCCTTCTCCTC	ATGCTGATCATCTTGGAGAG
NR1H4	GAATTCGAAAGAGTGGTATCTC	TCCTCCTGAGTCATTTTGAG
NRF1	TTATCCCAGAGATGCTCAAG	CATGAAACCCTTTGCTTTTG
PPARGC1A	TCACCATATTCCAGGTCAAG	TCATAGGCTTCATAGCTGTC
PPIB	TGGAGATGAATCTGTAGGAC	CAAATCCTTTCTCTCCTGTAG
SOD2	CCATTTTCTGGACAAACCTG	GACCTTGCTCCTTATTGAAG
TFAM	GACCTCGTTCAGCATATAAC	ACAAGCTTCAATTTTCCCTG

ACLY, ATP citrate lyase; COL1A1, collagen type 1 Alpha 1 chain; CYP7A1, cytochrome P450 family 7 subfamily A member 1; EMR1, adhesion G protein-coupled receptor E1; HMGCR, 3-hydroxy-3-methylglutaryl-CoA reductase; NR1H4, nuclear receptor subfamily 1 group H member 4 (FXR, farnesoid X receptor); NRF1, nuclear respiratory factor 1; PPARGC1A, peroxisome proliferative activated receptor, gamma, coactivator 1 alpha; TFAM, transcription factor A, mitochondrial. CAT, catalase; PPIB, cyclophilin B; SOD2, superoxide dismutase 2.

Statistics

Statistical analyses were performed using SPSS Statistics 27 (IBM). Data were first checked for normality by the Shapiro–Wilk test and for outliers via Grubbs with Prism 8 (GraphPad). Differences between treatment (sham vs. OVX vs. OVX + Est) and VWR (SED vs. VWR) were assessed via a two-way ANOVA and subsequent least significant difference (LSD) post hoc test. For VWR data, differences between treatment and time (pre- vs. postsurgery) were assessed via a repeated measures ANOVA with LSD post hoc test. Significance was set at P < 0.05 and data are presented as means ± SE.

RESULTS

Anthropometrics and Running Distance

As shown in Table 2, there was a significant main effect of treatment on change in body weight, fat mass change, fat-free mass change, and feeding efficiency. OVX animals gained significantly more body weight and fat mass throughout the study compared with sham controls, regardless of VWR. Importantly, the OVX + Est group had significantly lower body weight and fat mass change compared with the OVX treatment group and did not differ from the sham controls. This pattern was mirrored in feeding efficiency, where OVX animals gained ∼2-fold more weight per kilocalorie of energy intake than both sham and OVX + Est animals, which were statistically similar. VWR leads to significant fat-free mass gains in the sham and OVX + Est treatment groups, but this effect was ablated in the OVX group. All treatment groups averaged statistically similar daily run distances presurgery, despite the sham animals having restricted wheel access (6 h during dark cycle). There was, however, a main effect of surgery where running distance decreased across all groups following surgery, with the greatest decrease seen in the OVX treatment group.

Table 2.

Anthropometrics

		Sham		OVX		OVX + Est
Variable, units		SED	VWR	SED	VWR	SED	VWR
Body weight change, g	*T	4.8 ± 0.8	4.8 ± 0.4	10.8 ± 1.2*t	11.8 ± 1.0*t	3.8 ± 0.2	5.1 ± 0.3
Fat mass change, g	*T	1.7 ± 1.1	1.0 ± 0.3	7.0 ± 1.0*t	8.0 ± 1.0*t	−0.9 ± 0.3*t	−0.6 ± 0.1
Fat-free mass change, g	*T, V	2.7 ± 0.3	3.6 ± 0.2*v	3.5 ± 0.2*t	3.6 ± 0.3	4.2 ± 0.2*t	5.0 ± 0.3*t, v
Perigonadal fat, mg/g BW		25.9 ± 12.9	37.3 ± 14.7	21.4 ± 9.9	21.7 ± 10.0	55.2 ± 13.0	26.2 ± 9.8
Weekly energy intake, kcals	*V	55.2 ± 1.9	73.5 ± 1.4*v	60.1 ± 1.9	75.3 ± 2.3	68.5 ± 13.4	73.9 ± 2.0
Feeding efficiency, mg/kcal	*T	14.4 ± 2.1	10.8 ± 0.7	29.8 ± 2.8*t	25.8 ± 1.7*t	10.4 ± 1.3	11.5 ± 0.6
Serum NEFAs, ng/mL	*T	0.25 ± 0.3	0.24 ± 0.3	0.30 ± 0.3	0.30 ± 0.3	0.22 ± 0.2	0.19 ± 0.3
VWR presurgery, km/day		–	7.5 ± 1.2	–	10.8 ± 4.5	–	9.1 ± 3.3
VWR postsurgery, km/day	*S	–	6.2 ± 2.3	–	4.3 ± 1.6	–	6.0 ± 2.0

BW, body weight; Est, estradiol; LSD, least significant difference; NEFAs, nonesterified fatty acids; OVX, ovariectomy; SED, sedentary; VWR, voluntary wheel running. N = 6 mice per group, raw values presented as means ± SE. Uppercase *T, *V, and *S: main effects of treatment, VWR, or surgery, respectively from ANOVA. Lowercase *t and *v denote significant LSD post hoc for treatment and VWR, respectively. All P < 0.05.

Hepatic Steatosis and Markers of Fibrosis

In line with previous reports, we report main effects of treatment and VWR on liver triglycerides, where OVX resulted in greater steatosis which was rescued with estradiol replacement, and VWR was protective (Fig. 2A). More specifically, VWR significantly reduced liver TGs in the OVX group, and had a lowering effect that was trending in sham controls (P = 0.099). Importantly, the OVX + Est SED and VWR groups had lower steatosis than the respective sham controls. Though not quantitative, H&E-stained liver sections show similar patterns, with OVX animals having greater lipid deposition and VWR having a protective effect (Fig. 2B). To further assess liver health, we measured 4-Hydroxynonenal (4HNE) protein expression and superoxide dismutase 2 (SOD2), catalase (CAT), collagen type I alpha 1 chain (COL1A1), and Adhesion G Protein-Coupled Receptor E1 (EMR1) mRNA expression in liver whole homogenate. There was no effect of treatment or VWR on 4HNE, a marker of lipid peroxidation (Fig. 3A). We report a significant treatment by VWR interaction for SOD2 mRNA expression, where OVX VWR animals had greater expression of this potent mitochondrial antioxidant enzyme compared with both their sedentary counterparts and the VWR animals in the other treatment groups (Fig. 3B). Given these results, we investigated another important antioxidant enzyme, catalase, but saw no differences in CAT mRNA expression (Fig. 3C).

Figure 2.

Hepatic steatosis. To determine steatosis, hepatic triglyceride (TAG) content was measured with a biochemical assay and normalized to liver tissue weight (A) and liver sections were stained with H&E (B). Liver TAG data are presented as means ± SE (n = 5–6 mice per group). T, V: main effect of treatment or VWR; T × V: interaction of treatment and VWR from ANOVA. Lowercase t and v denote significant LSD post hoc for treatment and VWR, respectively. All P < 0.05. H&E, hematoxylin and eosin; LSD, least significant difference; OVX, ovariectomy; SED, sedentary; VWR, voluntary wheel running.

Figure 3.

Markers of liver health. Markers of oxidative stress, 4HNE (A), SOD2 (B), and CAT (C), were measured in liver whole homogenate via Western blotting and RT-PCR, respectively. Other indicators of liver health, COL1A1 (D) and EMR1 (E), were evaluated via RT-PCR. Data are presented as means ± SE (n = 5–6 mice per group). T: main effect of treatment; T × V: interaction of treatment and VWR. Lowercase t and v denote significant LSD post hoc for treatment and VWR, respectively from ANOVA. All P < 0.05. CAT, catalase; COL1A1, collagen type I alpha 1 chain; EMR1, adhesion G protein-coupled receptor E1; Est, estradiol; LSD, least significant difference; OVX, ovariectomy; PPIB, cyclophilin B; RT-PCR, reverse transcription-polymerase chain reaction; SOD2, superoxide dismutase 2; VWR, voluntary wheel running; 4HNE, 4‐hydroxynonenal.

In the disease progression of NAFLD, nonalcoholic steatohepatitis with fibrosis follows steatosis and is characterized by the presence of inflammation and scar tissue. To investigate these measures, we assessed COL1A1 and EMR1 (F4/80), key players in collagen formation and macrophage infiltration, respectively. There was a treatment effect for both of these outcomes (Fig. 3, D and E), which was largely driven by higher hepatic mRNA expression in the OVX + Est group. Though not statistically significant, there was a trend of lower hepatic COL1A1 mRNA expression with VWR across all treatment groups.

Hepatic Mitochondria Respiration

As shown in Figs. 4 and 5, experimental treatment modulated hepatic mitochondrial respiration and coupling in a substrate-specific manner, showing differences under pyruvate-based substrates but no changes under PC conditions. When looking at maximum ADP-stimulated state 3 respiration, we report a main effect of treatment, driven by increased oxygen flux in the OVX + Est group (Fig. 4B). Given that our measure of mitochondria coupling (coupling control ratio) is calculated as basal respiration divided by ADP-stimulated respiration, it was no surprise that we also saw a treatment effect here (Fig. 4E). Of note, there was also a significant interaction between treatment and VWR, whereby VWR significantly improved coupling only in the OVX group, and OVX sedentary animals were poorly coupled compared with their sham counterparts. In line with our previous data, under fat-based (PC) starting substrates, basal oxygen flux was lower in VWR animals, possibly indicative of greater efficiency (Fig. 5A).

Figure 4.

Mitochondrial respiration and coupling with pyruvate. Mitochondrial respiratory capacity was measured in isolated hepatic mitochondria using pyruvate as the main starting substrate at basal (A), ADP-stimulated (state 3; B), ADP-stimulated plus succinate (state 3S; C), and uncoupled (D) respiratory states. All respiration was normalized to mitochondrial protein in the O2k chamber measured via BCA assay. Coupling control ratio was calculated as basal respiration divided by ADP-stimulated respiration (E). Data are presented as means ± SE (n = 6 mice per group). T and V: main effect of treatment or VWR; T × V: interaction of treatment and VWR from ANOVA. Lowercase t and v denote significant LSD post hoc for treatment and VWR, respectively. All P < 0.05. BCA, bicinchoninic acid; Est, estradiol; LSD, least significant difference; OVX, ovariectomy; VWR, voluntary wheel running.

Figure 5.

Mitochondrial respiration and coupling with PC. Isolated hepatic mitochondrial respiratory capacity was measured with an Oroboros O2k-Fluo-Respirometer using l-palmitoylcarnitine as the main starting substrate at basal (A), ADP-stimulated (state 3; B), ADP-stimulated plus succinate (state 3S; C), and uncoupled (D) respiratory states. All respiration was normalized to mitochondrial protein via BCA assay. Coupling control ratio was calculated as basal respiration divided by ADP-stimulated respiration (E) to infer coupling. Data are presented as means ± SE (n = 6 mice per group). V: main effect of VWR from ANOVA when averaged across all treatment groups. All P < 0.05. BCA, bicinchoninic acid; Est, estradiol; OVX, ovariectomy; PC, l-palmitoylcarnitine; VWR, voluntary wheel running.

Mitochondrial Electron Transport System and Biogenesis

Interestingly, we found that there was a significant main effect of experimental treatment or an interaction between treatment and VWR on all oxidative phosphorylation protein complex expression measured in liver whole homogenate (Fig. 6). These differences were mainly driven by greater expression in the OVX group (Fig. 6, A and B) and lower levels in the OVX + Est group (Fig. 6, C and D). VWR resulted in significant depressions of complexes III, IV, and V in sham animals, whereas impact in the OVX group was less uniform, with exercise leading to greater Complex I expression but lower expression of Complexes II and III (Fig. 6). These differences reported in oxidative phosphorylation (OXPHOS) protein levels lead us to further investigate markers of mitochondrial biogenesis in liver whole homogenates via reverse transcription-polymerase chain reaction (RT-PCR). First, we see no difference in PPARGC1A expression, a gene encoding for the primary transcriptional coactivator responsible for mitochondrial biogenesis (peroxisome proliferator-activated receptor gamma coactivator 1-alpha; PGC1α), due to treatment or VWR (Fig. 7A). One of the transcription factors under PGC1α regulation is NRF1, and we also report no differences in mRNA expression (Fig. 7B). Interestingly, (though not statistically significant; P = 0.05), there is a treatment effect on transcription factor A, mitochondrial (TFAM), whereby the OVX + Est treatment group has lower gene expression (Fig. 7C). Given that TFAM plays an important role in mitochondrial DNA transcription and replication, this may be one contributing factor to the lower OXPHOS protein levels in this group.

Figure 6.

Electron transport system protein complexes. Quantified protein expression of oxidative phosphorylation complexes in liver whole homogenate was determined via Western blotting and normalized to total protein (A–E). A representative blot is shown in F. Data are presented as means ± SE (n = 6 mice per group). T and V: main effect of treatment or VWR; T × V: interaction of treatment and VWR from ANOVA. Lowercase t and v denote significant LSD post hoc for treatment and VWR, respectively, when averaged across all other factors. All P < 0.05. Est, estradiol; LSD, least significant difference; OVX, ovariectomy; VWR, voluntary wheel running.

Figure 7.

Markers of mitochondrial biogenesis. Quantified mRNA expression of genes involved in mitochondrial biogenesis—peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PPARGC1a) (A), nuclear respiratory factor 1 (NRF1) (B), and transcription factor A, mitochondrial (TFAM) (C), normalized to the housekeeping gene Cyclophilin B (PPIB). Data are presented as means ± SE (n = 6 mice per group). T: main effect of treatment from ANOVA; lowercase t: significant LSD post hoc for treatment when averaged across all other factors. All P < 0.05 or equal to 0.05 where noted. Est, estradiol; LSD, least significant difference; mRNA, messenger RNA; OVX, ovariectomy; PPIB, cyclophilin B; VWR, voluntary wheel running.

Cholesterol and Bile Acid Metabolism

Given that citrate from the mitochondria can be shuttled toward cholesterol and bile acid synthesis in conditions of nutrient excess, we were interested in the potential impacts of estrogen status and/or VWR on these outcomes. In liver whole homogenates, we see an effect of treatment (P = 0.05) on ATP citrate lyase (ACLY) mRNA expression, the rate-limiting step of the mevalonate pathway that produces cholesterol, likely driven by a significant decrease in the OVX + Est treatment group (Fig. 8A). Neither treatment nor VWR changed liver 3-hydroxy-3-methylglutaryl-CoA reductase HMGCR mRNA expression, the rate limiting enzyme in the mevalonate pathway that produces cholesterol (Fig. 8B). When investigating bile acid metabolism, we report a main effect of treatment on both hepatic CYP7A1 mRNA expression and fecal bile acids, but no differences in nuclear receptor subfamily 1 Group H member 4 (NR1H4) (farnesoid X receptor; FXR) mRNA expression (Fig. 8, C–E). CYP7A1 is the rate-limiting enzyme in the classical bile acid synthesis pathway, and here we report mRNA expression is upregulated in the livers from OVX + Est mice, regardless of activity status. Comparatively, OVX mice without estradiol add-back have lower hepatic mRNA expression levels of CYP7A1, which likely relates to the significantly lower amounts of bile acid excretion.

Figure 8.

Cholesterol and bile acid metabolism. ACLY (A) and HMGCR (B), genes involved in cholesterol metabolism, and CPY7A1 (C) and NR1H4 (D), genes involved in bile acid metabolism, were measured with RT-PCR and normalized to the housekeeping gene PPIB. Bile acids were extracted from feces (E), quantified using a commercially available bile acid assay, and normalized to fecal weight used in the extraction. Data are presented as means ± SE (n = 6 mice per group). Capital T and lowercase t reflect a main effect from ANOVA or significant LSD post hoc for treatment, respectively. Significant values are P < 0.05 or equal to 0.05 where noted. ACLY, ATP ctrate lyase; CPY7A1, Cytochrome P450 Family 7 Subfamily A Member 1; Est, estradiol; HMGCR, 3-hydroxy-3-methylglutaryl-CoA reductase; LSD, least significant difference; NR1H4, nuclear receptor subfamily 1 group H member 4 (FXR, farnesoid X receptor); mOVX, ovariectomy; PPIB, cyclophilin B; RT-PCR, reverse transcription-polymerase chain reaction; VWR, voluntary wheel running.

DISCUSSION

Long-standing epidemiological data agree that females are protected from NAFLD development during their reproductive years, with risk rapidly increasing following menopause (34). Our research group and others have shown that this same sexual dimorphism exists in rodents, where females are protected against high-fat diet-induced steatosis (7). Together, these data clearly suggest that estrogen is protective against hepatic lipid deposition (steatosis); however, the underlying mechanisms are incompletely understood. In this current report, we sought to investigate the impact of estrogen status (via OVX and OVX + Est treatment groups) and exercise (via VWR) on known modulators of steatosis risk—hepatic mitochondrial function, cholesterol/bile acid metabolism, and other markers of liver health. As summarized in Fig. 9, we show that OVX induces a robust phenotype of metabolic dysfunction, marked by steatosis, increased body weight and fat mass gain, greater feeding efficiency, with lower mitochondrial coupling and fecal bile acid excretion. Modest voluntary wheel running protected against steatosis and maintained good coupling of hepatic mitochondrial oxygen flux to ATP production in the OVX treatment group. Not surprisingly, estradiol replacement was also effective in counteracting many of the OVX-induced phenotypes. Notably, OVX + Est animals gained less fat mass, had lower hepatic lipid deposition, improved mitochondrial coupling, and maintained bile acid excretion.

Figure 9.

Summary of main findings. Up and down arrows designate the directionality of the statistically significant difference (all P < 0.05) between groups due to either treatment (sham, OVX, OVX + Est) or VWR. A horizontal line designates no significant differences. BAs, bile acids; BW, body weight; Est, estradiol; OVX, ovariectomy; PYR, potassium-pyruvate; TAGs, triglycerides; VWR, voluntary wheel running.

Our data on weight gain and obesity align with both previously published OVX rodent models (35) and clinical/epidemiological findings in postmenopausal women (36). Energy intake was similar across all treatment groups with greater feeding efficiency in the OVX animals, suggesting energy expenditure as a likely underlying mechanism to the increased weight gain in the OVX group. Indeed, previous findings suggest lower total energy expenditure in OVX mice compared with both sham and OVX + Est-treated animals fed a high-fat diet for 4 wk (37). However, the impact of OVX on energy expenditure may be transient, as Zengin et al. (38) report reduced energy expenditure in OVX mice 2 wk after surgery, which was absent when measured again 4–5 wk later, despite increased obesity. When assessing the impact of OVX on specific components of energy expenditure—resting metabolic rate, the thermic effect of food, spontaneous physical activity, and substrate utilization, the results are inconsistent (38–40). Thus, the exact mechanisms and potential adaptations that occur related to energy expenditure, obesity, and metabolic disease risk with OVX warrant further investigation. Clear data may help optimize treatment intervention timing for peri- and postmenopausal women.

We are certainly not the first to show that estrogen depletion via OVX increases steatosis, and that treatment with 17β-estradiol and/or exercise can help protect against lipid deposition (41, 42). Compared with wild-type female mice, constitutive ERα knockout mice show elevated levels of steatosis, but there were no differences in liver-specific ERα knockout (LERKO) mice (11). Although these data may suggest that estrogen’s protective effect against liver steatosis is not mediated through hepatic ERα signaling, investigation with an inducible LERKO model is warranted given the potent effects of estrogen in fetal development (43). Despite similar results on liver triglyceride content, our study and others vary in the dose and delivery method of estradiol replacement therapy (37, 42). This raises further questions about the necessary features of estradiol replacement, and whether there is a critical concentration of circulating estrogen needed to confer hepatic benefits.

As far as the effects of exercise, it is important to note that the physical activity intervention in this study was quite modest compared with typical mouse studies employing VWR. The OVX treatment groups only performed voluntary running for 4.3 and 6.0 km per day compared with the 8–10 km/day in our previous studies (6, 8). This agrees with recent clinical data showing that small changes in sedentary time or aerobic capacity result in meaningful changes in NAFLD risk (44, 45), an important and promising result as lifestyle modifications is the only approved treatment for NAFLD. Interestingly, both OVX treatment groups (with and without 17β-estradiol) ran significantly shorter distances after surgery, and sham animals required restricted wheel access to match the reduced exercise levels. There is also a sex difference in human physical activity, whereby males are more active than females (46, 47). Though exact mechanisms remain incompletely understood, the regulation of physical activity in rodents is thought to partially occur via estrogen and estrogen receptor alpha (ERα) signaling and its effect on dopaminergic systems, among others (48).

The connection between impaired hepatic mitochondrial function and both broad metabolic disease and NAFLD specifically is well documented (49, 50). Also appreciated is that through a multitude of signaling pathways (via nuclear ERα and ERβ, membrane-bound estrogen receptors (ERs), and/or mitochondrial localized of ERα and ERβ), 17β-estradiol plays a direct role in the preservation and regulation of mitochondrial structure and function (51). Previous research suggests that the effects of 17β-estradiol on mitochondria are organ specific (52), with some reporting only modest impact on hepatic mitochondria measures as compared with the effects in skeletal muscle (14). Though it may have been a supraphysiologic dose of estradiol, here we show a potent effect of 17β-estradiol in the OVX + Est group to both lower levels of steatosis and increased state 3 hepatic mitochondrial respiration compared with sham animals. We also show for the first time that estrogen impacts hepatic mitochondrial function in a substrate-specific manner, with OVX-induced impairments in mitochondrial coupling control under carbohydrate conditions (pyruvate) but maintained function with lipid-based substrate (PC) in sedentary animals. We also show that VWR and estrogen replacement restore mitochondrial coupling control in the OVX groups, suggesting that VWR and estrogen provide similar metabolic benefit at the mitochondrial level. Without having used specific complex inhibitors to isolate activity of each OXPHOS complex or measuring reactive oxygen species production, we cannot determine what may be causing the diminished levels of ATP-coupled respiration with carbohydrate-based substrates in the OVX group. Furthermore, it is unclear if this impaired coupling is pathologic (contributes to the NAFLD phenotype in OVX animals) or protective (via increased electron flux). If controlled, mitochondrial uncoupling and subsequent reactive oxygen species (ROS) generation may play important signaling roles in substrate utilization switch from glucose to fatty acids and protection against oxidative damage (53). However, the lipid peroxidation product 4HNE is known to activate uncoupling protein UCPs as part of the negative feedback mechanism to diminish ROS production (54) and this is reported unchanged here with OVX or estradiol add-back. Future studies should investigate whether estrogen modulates mitochondrial pyruvate transporters voltage-dependent anion channel (VDAC) and mitochondrial pyruvate carrier (MPC) in the liver, as studies show that 17β-estradiol promotes VDAC phosphorylation and channel inactivation in the brain (55), and VDAC deficiency alters pyruvate oxidation and ATP production in humans (56).

Especially during nutrient excess (high-fat feeding), citrate is exported from the mitochondria for either de novo lipogenesis or cholesterol and bile acid synthesis. Given that bile acid synthesis and excretion has been tied to hepatic steatosis susceptibility (28) and that female mice have higher bile acid levels throughout the lifespan (25), we sought to determine if OVX would influence markers of bile acid metabolism and fecal bile acid excretion. Our data nicely support this, with estradiol-deficient OVX animals showing lower fecal bile acid excretion and greater steatosis compared with sham controls, an effect that was ablated with estrogen replacement. Given the known negative side effects of estrogen replacement therapy in women, targeting bile acid and cholesterol pathways may serve as a safe and effective means of rescuing metabolic disturbances in menopause and warrants further investigation.

This present study is not without limitations. First, though statistically significant, many of the observed changes by treatment group or VWR are low in magnitude. This begs the question of whether the observed changes (i.e., bile acid excretion, OXPHOS protein content, mitochondrial coupling) are large enough to induce and sustain physiologically significant differences in disease prevention (steatosis) in more complex systems such as in human health. Second, although it is the predominant approach used in the field, it is well recognized that mitochondrial isolation does not produce pure isolates. Research groups with similar isolation protocols to ours suggest that isolates are roughly 65% pure mitochondria and include peroxisomes, which also consume oxygen (57). Importantly, the substrates and agents used in the present respirometry studies specifically target mitochondria, making any impacts of oxygen consumption by organelles other than the mitochondria only a concern in the basal state. Third, we used ovariectomy in the presence or absence of estradiol treatment to investigate the impact of estrogen on sex differences in liver mitochondria and hepatic health. Despite this technique being a strong initial approach, it ignores the limitation that OVX induces physiological changes that are mediated through estrogen-independent pathways as well as those dominated by progesterone action (58). Lastly, many of our suggested mechanistic insights are phenotypic in nature and lack direct evidence for a causative role. Future work should directly manipulate mitochondrial function and/or bile acid metabolism to further probe any relation to liver steatosis.

Despite these limitations, these data suggest that ovariectomy leads to hepatic steatosis, impaired hepatic mitochondrial function under carbohydrate-derived substrates, and altered bile acid metabolism, all of which can be rescued by 17β-estradiol treatment and/or modest voluntary physical activity. These novel investigations of the underlying mechanisms of NAFLD susceptibility by estrogen status are especially important given the current rise in hepatic steatosis and life expectancy in women—who are now spending a large proportion of their lives in a low estradiol (postmenopausal) state.

GRANTS

This work was supported by Veterans Affairs Merit Review Grant (1I01BX002567-01; J. P. Thyfault), NIH Grants (R01KD121497; J. P. Thyfault), Institutional Development Award (NIGMS P20 GM103418; J. P. Thyfault), Clinical and Translational Science Awards TL1 Postdoctoral Training Grant (TL1TR002368; C. S. McCoin), American Heart Association Grants (20PRE35120098; K. N. Z. Fuller), and Center for Children’s Healthy Lifestyles and Nutrition Pilot Award (C. S. McCoin). Technical expertise for OVX was supplied by the Smith Intellectual and Developmental Disabilities Research Center (NIH U54 HD 090216) at the University of Kansas Medical Center, Kansas City, KS 66160.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

K.N.Z.F. and J.P.T. conceived and designed research; K.N.Z.F., C.S.M., A.T.V., C.J.H., and M.A.C. performed experiments; K.N.Z.F. and C.S.M. analyzed data; K.N.Z.F. and J.P.T. interpreted results of experiments; K.N.Z.F. prepared figures; K.N.Z.F. drafted manuscript; K.N.Z.F., C.S.M., A.T.V., and J.P.T. edited and revised manuscript; K.N.Z.F., C.S.M., A.T.V., C.J.H., M.A.C., and J.P.T. approved final version of manuscript.