Sex-specific effects of cocoa on hepatic redox regulation and lipid homeostasis in high-fat diet-fed mice

Abstract

Non-alcoholic fatty liver disease (NAFLD) is a comorbidity of obesity, and sex influences disease progression. Cocoa can mitigate NAFLD, but the impact of sex has never been investigated. Herein, we compared the hepatoprotective effects of cocoa with different processing histories (unfermented/unroasted, unfermented/roasted, fermented/unroasted, fermented/roasted) in obese male and female mice. C57BL/6J mice were fed high-fat diet (HFD) for 8 wks to induce obesity and then randomized to HFD or HFD supplemented with 80 mg cocoa/g diet for an additional 8 wks. Hepatic responses were analyzed at the mRNA and protein levels. Cocoa-treated male mice had decreased liver weight and enhanced expression of mitochondrial oxidative and endoplasmic reticulum (ER) stress response markers (p < 0.05). Cocoa-treated female mice had lower ER stress response markers and increased markers of very low-density lipoprotein biogenesis compared to HFD-fed controls (p < 0.05). Two-way multivariate analysis of variance revealed significant main effects of sex and diet, and a significant interaction. Canonical variate analysis demonstrated the efficacy of fermented/unroasted and fermented/roasted cocoas in both sexes, and unfermented/unroasted and unfermented/roasted cocoas in females. Cocoa mitigated obesity-related NAFLD by sexually dimorphic mechanisms. Further research is needed to develop personalized nutritional recommendations incorporating cocoa.

1. Introduction

By 2040, the global prevalence of non-alcoholic fatty liver disease (NAFLD), a co-morbidity of obesity, is projected to increase by 50% [1]. NAFLD is characterized by lipid accumulation in hepatocytes resulting from the imbalance of fatty acid influx from the diet and/or de novo lipogenesis and efflux through export or oxidation. This results in oxidative stress and inflammation [2,3]. Endoplasmic reticulum (ER) and mitochondrial dysfunction are pathophysiological factors in NAFLD progression [2]. The elevated flux of free fatty acids lowers the ratio of reduced to oxidized glutathione and causes protein misfolding in the ER, resulting in ER stress and activation of unfolded protein response (UPR) transducers linked to proinflammatory and apoptotic pathways [2,4]. Aberrant mitochondrial fatty acid oxidation and increased flux through the respiratory transport chain leads to increased mitochondrial oxidative stress, increased mitochondrial membrane permeability, and increased apoptotic cell death [2].

NAFLD pathogenesis can be influenced by sex, age, and endocrine status [3]. The prevalence of NAFLD is lower in premenopausal women compared to both men and postmenopausal women, which suggests a potential protective role for estrogen [5]. Still, a limited number of studies have directly examined the impact of biological sex on the development of NAFLD treatment [3].

Cocoa, derived from the seeds of Theobroma cacao L., is a popular food ingredient used in the manufacture of chocolate and is a rich source of polyphenols. Both laboratory and human intervention studies have been shown that cocoa and chocolate may mitigate cardiometabolic diseases [6,7]. A meta-analysis of randomized clinical trials showed that cocoa consumption was associated with reduced fasting blood glucose, total cholesterol, low-density lipoprotein-associated cholesterol, and blood pressure [7]. These effects were proportional to the polyphenol concentrations in the treatment. In another meta-analysis of randomized clinical trials, cocoa flavanols were shown to have beneficial effects against insulin resistance, dyslipidemia, and inflammation [8].

Cocoa and cocoa polyphenols have been reported to mitigate NAFLD progression by improving mitochondrial function, increasing fatty acid oxidation, and enhancing the expression of endogenous antioxidants [9–11]. In a single-blind trial involving subjects with non-alcoholic steatohepatitis, consumption of 40 g dark chocolate per day for 2 wks suppressed serum markers of oxidative stress and hepatic injury [10]. Dietary supplementation of high fat diet (HFD)-fed obese male C57BL/6J mice with 80 mg cocoa powder/g diet for 10 wks decreased hepatic lipid content and increased the expression of markers related to antioxidant response and mitochondrial biogenesis [9]. Only two studies have explored the sex-dependent health-beneficial effects of cocoa. Dietary supplementation of leptin-deficient (ob/ob) male and female mice with 80 mg cocoa powder/g diet for 10 wks was reported to reduce fasting insulin levels in males but to increase fasting insulin levels in female mice [12]. Interestingly, the same study found that cocoa treatment of females, but not males, reduced β cell death. In another study, dietary treatment of HFD-fed obese C57BL/6J mice with 80 mg cocoa powder/g diet for 8 wks was shown to reduce body weight gain and improve gut barrier function and inflammation in both male and female mice [13]. The reported magnitudes of effect, however, differed between male and female mice.

The aim of the present study was to characterize the effects of cocoa on NAFLD, to compare differences in those effects between male and female mice, and to quantify markers of hepatic oxidative and ER stress in HFD-fed obese mice. The hypothesis was that cocoa would be efficacious regardless of sex, but that the magnitude of the effects and the underlying protective mechanisms of action would be different. The insights obtained from our study could support a transition from a one-size-fits-all to an individualized dietary approach, that incorporates cocoa, to mitigate NAFLD.

2. Materials and Methods

2.1. Materials

A single lot of unfermented, unroasted (NONO) cocoa beans were obtained from a commercial vendor (Cargill, Inc., Minnetonka, MN, USA). Beans were fermented and/or roasted to produce fermented/unroasted (LONO), unfermented/roasted (NOLO), and fermented/roasted (LOLO) beans as previously described [14]. Cocoa beans were then winnowed, ground, and pressed to produce cocoa powder [14]. The total polyphenol content, mean degree of polymerization of polyphenols, and the concentrations of individual catechins and proanthocyanidins in each unprocessed or processed cocoa were previously reported and are presented in Table S1 [14]. These cocoa powders were selected for study because they showed significant differences in polyphenol profile and represent the spectrum of operations involved in cocoa processing.

2.2. Animals and treatment

The present study is an analysis of archived plasma and liver samples derived from a previously reported study investigating the effect of fermentation and roasting parameters on the anti-inflammatory and gastrointestinal protective effects of cocoa in the HFD-fed mouse model of obesity [13]. The following animal and treatment information is from that previously reported study. All animal experiments were approved by the Pennsylvania State University Institutional Animal Care and Use Committee (University Park, PA, USA, Protocol No. 45380). Cocoa powders were incorporated into the semi-purified HFD containing 60% kcal from fat (Research Diets Inc., New Brunswick, NJ, USA) at a final concentration of 80 mg cocoa powder/g diet. Based on allometric scaling, this dose is equivalent to human dose of approximately 50 g cocoa powder/d [15]. Diets were matched for energy content and macro- and micronutrient content. The composition of the experimental diets were previously reported and are included in Table S2 [15]. Male and female C57BL/6J mice (4 wks old, Jackson Laboratories, Bar Harbor, ME, USA) were acclimated for 1 wk prior to starting the experiments and were then fed HFD for 8 wks to induce an obese phenotype. Based on previously established criteria, mice in the bottom 15^th percentile for body weight were excluded as non-responders to the HFD [13]. Mice (n = 11 to 13 per sex per treatment group) were randomized based on bodyweight to receive HFD or one of the cocoa-supplemented HFD (NONO, LONO, LONO, and LOLO) for another 8 wks. Since the obese mice underwent dietary cocoa intervention between 3 and 5 months of age, which corresponds to the peak fecundability and regular ovarian cyclicity in female rodents, this model corresponds to pre-menopausal period in humans [16]. Mice were maintained on a 12 h light-dark cycle and given ad libitum access to food and water throughout the experiment. Body weight and food intake were recorded weekly. At the end of the study, mice were euthanized by CO₂ asphyxiation, blood was collected via cardiac puncture, and tissue samples were collected at necropsy. The effects of cocoa treatment on body weight gain, body fat, markers of systemic inflammation, and the gastrointestinal microbiome for these mice have been previously reported [13]. It was also previously reported that food intake did not differ across treatment groups.

2.4. Quantitative reverse transcriptase polymerase chain reaction (PCR)

RNA was extracted from 100 mg of liver tissues with the PureLink RNA extraction Mini Kit (Thermo Fisher, Waltham, MA, USA) using a slight modification of the manufacturer’s protocol. Total RNA was quantified with a Nanodrop 2000 spectrophotometer, and 1 μg of RNA was reverse-transcribed to cDNA using an RT² HT First Strand Kit (SA Biosciences, Valencia, CA, USA). Quantitative real-time PCR analysis was performed using a QuantStudio^™ 3 Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). Consisting of 12.5 μL RT² SYBR Green ROX qPCR Mastermix, 0.2 μL per forward and reverse primers (100 μM in IDTE buffer, pH = 8.0), 4 μL cDNA, and DEPC-treated water, PCR reactions were incubated in a 96-well plate at 95 °C for 10 min, followed by 40 cycles of denaturation at 95 °C for 15 s and annealing/extension at 60 °C for 1 min. Melting curves of PCR products were analyzed after the reaction. Data were recorded and analyzed with Sequence Detector Software (Applied Biosystems). Relative fold change in gene expression was determined using the 2^−ΔΔCT method, where ΔΔC_T = (C_{T, target} – average C_{T, reference}), with β-actin as the reference gene. Expression data for both sexes were normalized to the expression in HFD-fed obese male mice to demonstrate both cocoa treatment effects and sex differences [17]. Primer sequences are shown in Table S3.

2.5. Mitochondrial copy number and mitochondrial DNA (mtDNA) oxidative damage

Mitochondrial DNA (mtDNA) copy number and mtDNA damage were quantified by quantitative PCR. mtDNA was isolated from liver tissues using a DNeasy Blood and Tissue Kit (Qiagen Inc, Germantown, MD, USA). mtDNA copy number was estimated by quantifying the mitochondria-encoded cytochrome c-oxidase subunit 2 (MT-Co2) copy number and normalizing it to nuclear ribosomal protein s18 (Rps18) copy number as previously described [9]. To quantify mtDNA damage, primers for HVII-FOR260 and HVII-L421 were used to detect large mtDNA fragments and primers for HVII-FOR260 and HVII-C339 were used to detect short mtDNA fragments as previously described [9]. Each 25 μL PCR reaction included 15 ng of mtDNA and 12.5 μL of SYBR ROX qPCR Green Mix in addition to the primers (Table S3). The PCR conditions for both assays were 95 °C for 10 min, 40 cycles of 95 °C for 15 s, and 60 °C for 1 min, followed by a melt curve analysis consisting of 95 °C for 15 s and from 60–95 °C at a rate of 0.1 °C/s.

2.6. Protein extraction and Western blot analysis

Liver tissues (50 mg) were homogenized with 0.5 mm zirconium oxide beads in 500 μL RIPA buffer (Sigma-Aldrich, St. Louis, MO, USA) containing protease and phosphatase inhibitor cocktail (Thermo Fisher, Waltham, MA, USA) using a Bullet Blender Storm (Next Advance Inc., Averill Park, NY, USA). The homogenate was centrifuged at 12,000 × g for 10 min at 4 °C. Protein concentration was quantified using the Bradford Assay (Sigma-Aldrich, St. Louis, MO, USA). Protein extracts (37 μg) were resolved using 10% Tris-glycine gels for all target proteins except apolipoprotein B (APOB), which was resolved with 3 to 8% Tris-acetate gels (NuPAGE, Thermo-Fisher Scientific, Waltham, MA, USA). In all cases, proteins were resolved at 120 V. Proteins were transferred to polyvinylidene fluoride membranes (Millipore Corp, Bedford, MA) at 100 V for 2 h for all proteins except APOB, which was transferred at 20 V for 40 h. After blocking with 5% nonfat dried milk in tris-buffered saline containing 0.1% tween 20 (TBST) for 1 h at room temperature, the membrane was incubated with the appropriate primary antibodies (Table S4) overnight at 4 °C with rocking. After washing three times with TBST, the membrane was then incubated with a secondary antibody (m-IgG1 BP-HRP [sc-525408] or m-IgG Fc BP-HRP [sc-525409] at 1:2000, Santa Cruz Biotechnology, Inc, Santa Cruz, CA, USA) for 1 h at room temperature. The protein bands were visualized using SuperSignal^™ West Pico plus chemiluminescent substrate (Thermo Fisher, Waltham, MA, USA).

2.7. Plasma biomarkers of liver injury

Plasma alanine transferase (ALT) and aspartate aminotransferase (AST) concentrations were determined by the Animal Resource Program at The Pennsylvania State University (University Park, PA, USA) using a Catalyst One^® chemistry analyzer (IDEXX Laboratories, Inc., Westbrook, ME, USA) with IDEXX dry-slide technology.

2.8. Statistical analysis

All data are expressed as mean ± standard deviation. Grubbs' test was used to identify outliers in each data set and extreme outliers were removed [18]. Two-way analysis of variance (ANOVA) was conducted using the SPSS statistical software version 30.0.0.0 (SPSS Inc., Chicago, IL, USA) to evaluate the main effects of sex and diet, and the sex-diet interaction. Because the original study lacked sufficient statistical power to make all the pair-wise comparisons necessary following two-way ANOVA, exploratory analysis was performed within each sex using one-way ANOVA [13]. If variances were equal across treatment groups, then one-way ANOVA followed by Duncan's multiple range test was employed within each sex to compare the effects of cocoa treatment. If the variances were not the same, as determined by Levene's test, then Welch’s ANOVA was employed [19]. Two-way multivariate ANOVA (MANOVA) was performed using R Studio version 4.3.1 (Boston, MA, USA) to evaluate overall group differences. Canonical variate analysis (CVA) was then performed with the candis package to visualize the linear interrelation among multiple sets of variables by maximizing the ratio of the differences in inter-group to intra-group variances [20]. The significance of canonical variates was evaluated by Bartlett’s test [21]. If p < 0.05, then differences were considered statistically significant.

3. Results

3.1. Liver weight and plasma markers of liver injury

A significant main effect of both sex and diet was observed on final body weight (Table 1). In the present study, a significant main effect of sex on plasma ALT, plasma AST, and relative liver weight was observed, with male mice having higher values for all markers compared to females. There was also a significant sex-diet interaction (p = 0.004) for liver weight. Exploratory analysis showed that obese male mice treated with NONO cocoa-supplemented diets had 20% lower relative liver weights than HFD-fed male controls. No significant effect of cocoa on relative liver weight in female mice was observed. Moreover, no differences in plasma ALT or AST levels were observed across treatment groups within each sex. This lack of significant effect might be due to the small number of samples and the relatively large within group variance.

Table 1.

Effects of dietary cocoa on plasma levels of liver injury and liver weight in HFD-fed obese male and female C57BL/6J mice.

	Two-way ANOVA			Sex	Diet
	Sex	Diet	Interaction		HFD	NONO	NOLO	LONO	LOLO
Final body weight (g)	< 0.001	< 0.001	0.843	M	53.9a (1.56)	46.8b (2.41)	47.7b (2.69)	48.1b (1.91)	47.7b (3.38)
				F	42.9a (3.16)	34.9b (2.68)	35.8b (5.47)	37.5b (5.40)	37.7b (3.42)
ALT (U/L)	< 0.001	0.565	0.578	M	228.4 (81.26)	181.2 (114.82)	135.6 (39.21)	179.4 (111)	152.8 (98.78)
				F	52.8 (25.02)	33.2 (3.11)	48.2 (19.82)	37.8 (4.97)	51.7 (8.45)
AST (U/L)	< 0.001	0.755	0.419	M	274 (146.15)	239.2 (85.07)	187.4 (50.5)	222.6 (129.04)	179.8 (101.25)
				F	140 (43.99)	117.6 (39.43)	162.6 (74.67)	94.2 (29.3)	146.3 (87.63)
Relative liver weight (g/g body weight)	< 0.001	0.144	0.004	M	0.055a (0.0086)	0.044b (0.0094)	0.054a (0.0075)	0.047ab (0.011)	0.052ab (0.011)
				F	0.029 (0.0028)	0.031 (0.0032)	0.031 (0.0038)	0.032 (0.0053)	0.030 (0.0019)

Data were expressed as mean ± standard deviation (ALT: N=5/group/males, N=5/group/females, AST: N=5/group/males, N=5/group/females, relative liver weight: N=12–13/group/males, N=11–12/group/females, final body weight: N=12–13/group/males, N=11–12/group/females). Two-way ANOVA was used to test for main effects of sex and diet, and their interaction. One-way ANOVA followed by Duncan's multiple range test was used to test for treatment effects within each sex. Values showing a different superscript indicate a statistically significant difference at p < 0.05.

3.2. Markers of hepatic mitochondrial oxidative stress response

SOD2 is a key mitochondrial antioxidant enzyme [9]. A significant main effect of sex and diet, as well as a significant sex-diet interaction, was observed for hepatic Sod2 mRNA concentrations (Fig. 1A). Sod2 expression was higher in male mice compared to female mice and in cocoa-treated mice compared to HFD-fed controls. In males, exploratory analysis showed those treated with NOLO cocoa had significantly higher Sod2 expression compared to the HFD-fed controls. In females, exploratory analysis showed those treated with NONO, LONO, and LOLO cocoas had reduced expression of Sod2 mRNA. No significant main effect of either sex or diet was observed for hepatic SOD2 protein concentration, but a significant sex-diet interaction was observed (Fig. 1B). Within male treatment groups, exploratory analysis showed SOD2 protein concentrations were higher in NOLO (2.4-fold), LONO (2.4-fold), and LOLO (2.0-fold)-treated mice compared to HFD-fed control male mice (Fig. 1B). No statistically significant effect of cocoa on SOD2 protein expressions was observed in female mice (Fig. 1B). A significant main effect of sex, but not diet, and a significant sex-diet interaction for mtDNA damage was observed (Fig. 1C). Within each sex, exploratory analysis showed no significant effect of cocoa on mtDNA damage was observed compared to HFD-fed controls (Fig. 1C).

Figure 1.

Effects of dietary cocoa on hepatic superoxide dismutase expression and mitochondrial oxidative injury in HFD-fed obese male and female C57BL/6J mice. Hepatic levels of (A) Sod2 mRNA (N=7–10/group/males, N=6–9/group/females) and (B) SOD2 protein (N=5/group/males, N=7/group/females) expression was determined by quantitative reverse-transcriptase PCR and western blot, respectively. (C) Hepatic mitochondrial DNA (mtDNA) damage (N=5–7/group/males, N=3–4/group/females) was determined using a quantitative PCR-based approach and normalized to mtDNA copy number. All markers were normalized to the levels in male HFD-fed obese control mice and shown as mean ± standard deviation. Two-way ANOVA was used to test for main effects of sex and diet, and their interaction. One-way ANOVA followed by Duncan's multiple range test was used to test for treatment effects within each sex. Values showing a different superscript indicate a statistically significant difference at p < 0.05.

3.3. Markers of hepatic mitochondrial biogenesis

Elevated mitochondrial biogenesis has been associated with improvements in NAFLD [9]. In the present study, a significant main effect of sex, but not diet, on mtDNA copy number was observed (p = 0.046), whereas no significant sex-diet interaction was observed (Fig. 2A). Mitochondrial transcription factor A (TFAM) is responsible for maintaining, expressing, and organizing mtDNA [22]. A significant main effect of sex but not diet, as well as a significant sex-diet interaction, were observed for Tfam mRNA expression. Tfam mRNA expression was higher in females compared to males across treatment groups (p < 0.001). Within male treatment groups, exploratory analysis showed those treated with LOLO cocoa had 4.5-fold higher Tfam mRNA expression compared with HFD-fed controls, but no significant differences were observed among the female treatment groups (Fig. 2B).

Figure 2.

Effects of dietary cocoa on hepatic mitochondrial biogenesis in HFD-fed obese male and female C57BL/6J mice. (A) Hepatic mtDNA copy number (N=5–7/group/males, N=3–4/group/females) was determined using a quantitative PCR-based approach to measure the ratio of mitochondrial cytochrome oxidase subunit 2 (mitochondrial DNA) and 40S ribosomal protein s18 (nuclear DNA). (B) Hepatic Tfam mRNA (N=4–6/group/males, N=4–7/group/females), (C) Pgc1a mRNA (N=6–8/group/males, N=6–8/group/females), (D) Nrf2 mRNA (N=3–5/group/males, N=3–6/group/females), and (E) Sirt3 mRNA (N=4–8/group/males, N=3/group/females) expression were determined using qRT-PCR. All markers were normalized to the levels in male HFD-fed obese control mice and shown as mean ± standard deviation. Two-way ANOVA was used to test for main effects of sex and diet, and their interaction. One-way ANOVA followed by Duncan's multiple range test was used to test for treatment effects within each sex. Values showing a different superscript indicate a statistically significant difference at p < 0.05.

Proliferator-activated receptor γ coactivator-1α (PGC1α) and nuclear respiratory factor 2 (NRF2) are upstream regulators of TFAM [22]. A significant main effect of both sex and diet, as well as a significant sex-diet interaction, for both Pgc1a and Nrf2 mRNA expression was found. Expression of both genes was higher in females than in males (Fig. 2C, D). Within each sex, exploratory analysis showed no significant differences in Pgc1a expression were observed between the cocoa-treated groups and HFD-fed controls (Fig. 2C). In females, but not males, 12.8-fold higher expression of Nrf2 mRNA expression was observed in those treated with NONO cocoa compared to HF-fed controls (Fig. 2D).

Sirtuin (SIRT) 3 has been reported to coordinate mitochondrial antioxidant response and mitochondrial biogenesis by deacetylating lysine in SOD2 and TFAM [23]. A significant main effect of both sex and diet on hepatic Sirt3 mRNA concentration was observed (Fig. 2E). exploratory analysis showed that in male mice, those treated with NONO cocoa had 2.3-fold higher Sirt3 mRNA expression compared to HFD-fed controls (Fig. 2E). No significant difference among the treatment groups was observed in females (p = 0.056).

3.4. Hepatic ER stress response

NAFLD is associated with increased misfolded proteins and ER stress due to excess energy intake [2]. The impact of dietary cocoa on markers of hepatic UPR and ER stress was examined. A significant main effect of sex and diet, and significant sex-diet interaction, was observed for Ire1a mRNA concentrations. Female mice had a 3.4-fold higher Ire1a mRNA expression than males (Fig. 3). Within the female treatment groups, exploratory analysis showed those treated with NOLO or LONO cocoa had Ire1a mRNA levels that were 68% and 63% lower than the HFD-fed controls (Fig. 3A). By contrast, cocoa had no significant effect on hepatic Ire1a mRNA concentrations in male mice. For IRE1α protein concentration, a significant main effect of sex, but not diet, was observed (Fig. 3B). No significant sex-diet interaction was observed. Female mice had 50% lower IRE1α protein levels than males. Exploratory analysis showed that in females, but not males, mice treated with NONO, LONO, and LOLO cocoas had 55 – 70% lower IRE1α expression compared to HFD-fed controls (Fig. 3B).

Figure 3.

Effects of dietary cocoa on hepatic ER stress response in HFD-fed obese male and female C57BL/6J mice. Hepatic (A) Ire1a mRNA (N=5–7/group/males, N=3–5/group/females), (B) IRE1α protein (N=4/group/males, N=4/group/females), (C) Ero1a mRNA (N=6–7/group/males, N=3–8/group/females), (D) ERO1α protein (N=4/group/males, N=7–9/group/females), (E) Pdia mRNA (N=5–7/group/males, N=6–11/group/females), (F) PDIA1 protein (N=3/group/males, N=4/group/females), and (G) Gpx7 mRNA (N=4–9/group/males, N=6–11/group/females) expression were determined using qRT-PCR or western blot. All markers were normalized to the levels in male HFD-fed obese control mice and shown as mean ± standard deviation. Two-way ANOVA was used to test for main effects of sex and diet, and their interaction. One-way ANOVA followed by Duncan's multiple range test was used to test for treatment effects within each sex. Values showing a different superscript indicate a statistically significant difference at p < 0.05.

The effects of cocoa treatment on hepatic expression of ERO1α, PDIA1, and glutathione peroxidase 7 (GPX7), which work in concert to refold misfolded proteins in response to ER stress, were analyzed [4]. No significant main effects of sex or diet and no significant sex-diet interaction were observed for Pdia1 mRNA concentration (Fig. 3C). In female mice, exploratory analysis showed those treated with LONO cocoa had a 3.4-fold higher Pdia1 mRNA expression than the HFD-fed controls. No significant differences among male mice were found. Similarly, no significant main effects of sex or diet and no significant sex-diet interaction for PDIA1 protein concentration were observed (Fig. 3D). Within female treatment groups, exploratory analysis showed those treated with NONO cocoa had higher PDIA1 protein concentrations compared to those treated with LONO or LOLO cocoas but not compared to HFD-fed controls. No significant differences were observed among the male treatment groups. A significant main effect of diet, but not sex, was observed for Ero1a mRNA concentration (Fig. 3E). No significant sex-diet interaction was observed. For female mice, exploratory analysis showed those treated with NOLO cocoa had higher Ero1a mRNA concentrations compared to those treated with NONO, LONO, or LOLO cocoas, but not compared to HFD-fed controls. No significant differences were found among the male treatment groups. A significant main effect of sex and diet, as well as a significant sex-diet interaction was observed for ERO1α protein concentration (Fig. 3F). In male mice, exploratory analysis showed ERO1α protein levels were higher in those treated with LONO or LOLO cocoa compared to HFD-fed controls. No difference in ERO1α protein concentrations was observed between treatment groups within females. A main effect of sex, but not diet, as well as a significant sex-diet interaction was observed for hepatic Gpx7 mRNA concentration (Fig. 3G). In males, those treated with LONO or LOLO cocoas had higher Gpx7 mRNA concentrations compared to HFD-fed controls; however, no significant difference was found among treatment groups for females.

3.5. Markers of hepatic very-low-density lipoprotein (VLDL) biosynthesis and assembly

Increased lipid export from the liver to peripheral tissues has been shown to improve NAFLD in a sexually dimorphic manner [24]. APOB determines the rate of VLDL synthesis [25]. A main effect of sex, but not diet, was found for Apob mRNA concentration (Fig. 4A). A significant sex-diet interaction was not observed. Female mice had 2.5-fold higher hepatic Apob mRNA concentrations than males. Within female treatment groups, exploratory analysis showed that those treated with NONO, LONO, and LOLO cocoas had Apob mRNA concentrations that were 1.9 – 2.6-fold higher than HFD-fed controls. No significant effect of cocoa on Apob mRNA concentrations was observed in male mice. For APOB protein concentration, no significant main effects for sex or diet, and no significant sex-diet interaction was found. Within female treatment groups, exploratory analysis showed that those treated with NONO, LONO, or LOLO cocoa had higher hepatic APOB protein concentrations compared to HFD-fed control (Fig. 4B). A significant treatment effect was not observed in males. MTP (gene abbreviation: Mttp) is responsible for the transfer of triacylglycerols to VLDL particles [26]. No significant main effect of sex or diet on Mttp mRNA concentration was observed, but a significant sex-diet interaction was found. Examining each sex separately, exploratory analysis showed that Mttp mRNA concentrations were higher in NOLO cocoa-treated female mice than in HFD-fed controls, whereas no differences among treatment groups were observed in males. No significant main effect of sex or diet and no significant sex-diet interaction were observed for MTP protein (Fig. 4D). Within female treatment groups, exploratory analysis showed that MTP concentration was higher in those treated with LONO cocoa compared to the HFD-fed controls. No differences were observed among male treatment groups. Estrogen-related receptor α (ERRα) controls hepatic VLDL biosynthesis by binding to the promoters of APOB and MTP [24]. A significant main effect of sex, but not diet, was observed for hepatic Erra mRNA concentration (Fig. 4E). A significant sex-diet interaction was not observed. Among female treatment groups, exploratory analysis showed that those treated with NOLO or LONO cocoas had 1.8-fold higher Erra mRNA concentrations than HFD-fed controls. No significant effect of cocoa on Erra mRNA levels was observed in males.

Figure 4.

Effects of dietary cocoa on hepatic biosynthesis of VLDL components in HFD-fed obese male and female C57BL/6J mice. (A) Apob mRNA (N=5–7/group/males, N=5–8/group/females), (B) APOB protein (N=3/group/males, N=5/group/females), (C) Mttp mRNA (N=5–7/group/males, N=4–7/group/females), (D) MTP protein (N=4/group/males, N=4/group/females), and (E) Erra mRNA (N=5–6/group/males, N=6–9/group/females) expression were determined using qRT-PCR or western blot. All markers were normalized to the levels in male HFD-fed obese control mice and shown as mean ± standard deviation. Two-way ANOVA was used to test for main effects of sex and diet, and their interaction. One-way ANOVA followed by Duncan's multiple range test was used to test for treatment effects within each sex. Values showing a different superscript indicate a statistically significant difference at p < 0.05.

3.6. Multivariate statistical analysis of the effect of dietary cocoa and sex

Two-way MANOVA incorporating all data collected in this study was used to understand the overall main effects of sex and cocoa, and the sex-diet interaction on NAFLD in HFD-fed obese mice. Significant main effects of sex (p < 0.001) and diet (p < 0.001), as well as a significant sex-diet interaction (p < 0.001), were observed. CVA analysis was used to examine how treatment groups within sex separated (Fig. 5). Non-overlapping 95% confidence ellipses were used as an indicator of statistically significant differences. Can1 and Can2 collectively explained 70.5% of the variance in the data. Among male treatment groups, the LONO and LOLO treatment groups were also separated from the HFD-fed control along Can1. ERO1α protein, IRE1α protein, Sod2 mRNA, and Gpx7 mRNA concentrations were the most important drivers of separation. Among female treatment groups, the NONO, LONO, and LOLO treatment groups were separated from the HFD-fed controls along both Can1 and Can2, whereas the NOLO treatment group was separated only along Can1. Separation along Can1 was driven mainly by Apob, Erra, Pgc1a, Nrf2, and Sirt3 mRNA levels. APOB protein, PDIA1 mRNA and protein, and Ire1a mRNA levels were the major drivers of separation in Can2.

Figure 5.

Multivariate statistical analysis of the effect of sex and cocoa on NAFLD in HFD-fed obese C57BL/6J mice. Canonical variate analysis was performed using all data collected in this study. The score and structure plots are shown. The “+” symbol represents the group mean and the ellipses represent the 95% confidence interval.

4. Discussion

Consumption of cocoa and chocolate represents a potential dietary approach to mitigate NAFLD [10]. Previous studies have reported that dietary cocoa can reduce hepatic steatosis, reduce hepatic and systematic inflammation, improve markers of insulin resistance, and increase makers of hepatic mitochondrial biogenesis and antioxidant defense [9,15]. In vitro experiments have shown that cocoa-derived proanthocyanidins can inhibit pancreatic α-amylase, lipase, and phospholipase A₂, as well as small intestinal α-glucosidase, which is expected to reduce the digestion of dietary carbohydrates and triglycerides and lead to reduced energy absorption [14,27]. Consistent with this, treatment of HFD-fed C57BL/6J mice with dietary cocoa has been show to increase fecal lipid content by more than 50% [15]. It has been previously reported that the magnitude of the effects of dietary cocoa on body weight and markers of systemic and gastrointestinal inflammation differed between male and female mice [13]. To date, the impact of biological sex on the hepatoprotective effects of cocoa has not been studied. In the present study, the hepatoprotective effects of differently processed cocoas were compared in male and female HFD-fed obese C57BL/6J mice using liver and plasma samples derived from a previous study [13]. Cocoa powders with different processing histories were selected to capture differences in polyphenol profile and increase the generalizability of the results across different cocoa samples. It was hypothesized that cocoa would have hepatoprotective effects in both sexes, but that the magnitude of the effects and the underlying mechanisms of action would show sexual dimorphism.

Accumulating evidence indicates that males are at greater for NAFLD development and are more likely to develop severe disease than premenopausal females due to the sex differences in lipid handling [28]. In accordance with this, the present study found a significant main effect of sex on markers of liver injury, with male obese mice having higher relative liver weight, plasma ALT, and AST levels compared to female obese mice. Although there was large within group variance in the data for both sexes, hepatomegaly was mitigated by some cocoa treatments in males, but not females. No treatment effects were seen for the other markers in either sex. The results in males are consistent with a previous study of cocoa in the same model [9].

Increased free fatty acid flux in the context of NAFLD leads to elevated leakage of reactive oxygen species from the mitochondrial respiratory chain complexes, oxidative damage, and expedited NAFLD progression [2]. Dietary flavonoids, including those in cocoa, have been shown to restore mitochondrial redox homeostasis in part by increasing the expression of antioxidant response proteins [29]. In the current study, it was found that some cocoa treatments increased the mRNA and protein expression of SOD2. This is consistent with previous work in high fat-fed obese male mice by Sun et al. [9]. By contrast, in female mice, most cocoa treatments were found to reduce Sod2 mRNA concentrations and were found to have no effect on SOD2 protein expression. The observed treatment effects on SOD2 expression did not translate to changes in mtDNA damage or mtDNA copy number in either males or females. This is in contrast to the work by Sun et al., who found that cocoa treatment reduced mtDNA damage and increased mtDNA copy number [9]. The inconsistent effects of dietary cocoa on mtDNA damage and copy number may be due to difference in the chemical composition of the cocoa powders used in the different studies or the relatively large variance in the data and the small sample size available in the current study.

SIRT3, PGC1α, and NRF2 coordinate mitochondrial biogenesis and mitochondrial antioxidant response, and TFAM is a key regulator of mtDNA replication [22]. In male mice, but not female mice, it was found that some cocoa powders increased the expression of Sirt3 and Tfam but had no effect on the other transcriptional regulators examined. Although this seems to contradict the observed lack of effect of cocoa on mtDNA copy number in the present study, it is in agreement with previous in vivo experiments which demonstrated no correlation between the TFAM upregulation and mtDNA content and replication [30]. Similarly, an in vitro study found that resveratrol and quercetin enhanced pre-existing mitochondrial differentiation in palmitic acid-treated HepG2 hepatocytes without causing changes in mitochondrial replication [31]. In female mice treated with some cocoas in the present study, increased hepatic Pgc1a and Nrf2 mRNA expression was observed. Previous studies have shown that both transcriptional regulators are responsive to estrogen [32,33]. This estrogen responsiveness likely explains the much higher expression of Pgc1a and Nrf2 that was observed in all female treatment groups compared to males in the present study.

ER stress induced by obesity-related increases in hepatic protein synthesis and accompanying increases in protein misfolding is another etiological mechanism associated with NAFLD [4]. Cocoa was found to have no effect on the mRNA or protein expression IRE1α, a sensor of ER stress, in male mice [4]. By contrast, most cocoa powders were observed to cause a decrease in the expression of both IRE1α mRNA and protein in female mice. These results indicate that cocoa reduces ER stress in females. Consistent with this hypothesis is the observation that cocoa treatments reduced the expression of PDIA1 protein and ERO1α mRNA in female mice. In males, however, treatment with some cocoas was found to increase expression of ERO1α protein and Gpx7 mRNA, indicating that cocoa enhanced protein folding cycles in male mice [4]. In addition, the increased expression of ERO1α could contribute to restoring redox balance in the ER by causing increased glutathione transfer from the cytosol into the ER [28]. The sex differences in UPR/ER stress-related markers in this study may be due to the impact of estrogen and/or testosterone. Estrogen has been shown to help maintain calcium concentration in the ER and to support proper protein folding by enhancing the expression of heat shock protein 70 [34]. This hypothesis was not directly tested in the present study as differences in sex hormones were not quantified. Future studies using pharmacological/genetic modulation of sex hormone signaling are necessary to directly interrogate the role of sex hormones in driving differences in the effects of cocoa.

In the present study, most dietary cocoa treatments were found to enhance the protein expression of APOB and MTP in female mice compared to HFD-fed controls, but cocoa treatments had no effects in male mice. Previous studies which have reported that estrogen can reduce NAFLD risk in part by increasing hepatic VLDL production and lipid export [28]. Unfortunately, because archived samples from a previous study were used in the present analysis, sufficient plasma samples were not available for direct measurement of VLDL secretion rate, plasma VLDL concentrations, or plasma sex hormone concentrations. The sex differences observed here suggest a potential role for estrogen, but future studies are needed to directly test this hypothesis.

Taken together, the present data reveal significant differences in the levels of NAFLD-related markers between HFD-fed obese male and female C57BL/6J mice and their response to dietary cocoa. These results expand previous findings on sex-related differences in the effects of cocoa intervention on body weight gain, systemic inflammation, and gastrointestinal health in this mouse model [13]. In that previous study, cocoa powders with different fermentation and roasting histories had significantly different overall anti-inflammatory efficacy. The results of the present study appear consistent with idea that cocoa powders of differing processing history have differential impacts health-related biomarkers, in this case, biochemical and molecular markers related to NAFLD. The CVA analysis performed as part of the present study showed that males treated with LONO or LOLO cocoa were significantly different from HFD-fed controls. In females, all cocoa-treated groups were significantly different from HFD-fed controls. Given the significant within group variation in the data used for the CVA and the lack of a clear phytochemical-base mechanism to explain the differences in the apparent hepatoprotective effects of the cocoa powders, the results should be interpreted with caution and require additional study to confirm.

The present study has some weaknesses. Due to limitations in sample availability (e.g., limited volumes of plasma), some key physiological parameters (e.g., plasma VLDL and hormone levels) could not be measured. In addition, it was not possible to measure changes in both mRNA and protein levels of some markers. To address this shortcoming, molecular markers at multiple levels within a pathway, either at the mRNA or protein level, were measured to demonstrate the effect of cocoa on that pathway. Additionally, due to the limited number of mice per treatment group, the relatively large number of pairwise comparisons, and the significant interindividual variation, some markers which showed nominal apparent treatment effects were not found to be statistically significantly different. The use of CVA to provide an overall picture of the treatment effects partially offsets this problem with statistical power. However, the results of the present study, in many cases represent exploratory findings that need to be confirmed using an appropriately powered study.

Despite these limitations, the present study has several strengths. The cocoa powders were derived from a single lot of cocoa beans processed under defined parameters and extensively chemically characterized [14]. This reduces potential confounding effects due to unknown processing history. Similarly, mice were treated with cocoa powders of different processing histories, which increases the potential impact of our results by making them more generalizable to the variety of cocoa powders on the market. Finally, multivariate statistical analysis was used to provide a more holistic view of the data and to better leverage all the data generated in the study to develop conclusions about the potential hepatoprotective effects of.

In conclusion, the present study demonstrated that dietary supplementation with cocoa is associated improvements in markers of NAFLD and that these effects are related to different mechanisms based on biological sex. Our findings indicate that cocoa may be useful as part of a personalized dietary approach to mitigate NAFLD. Further studies are needed to directly test the impact of specific sex hormones on the hepatoprotective effects of cocoa; to demonstrate the functional significance of the molecular changes observed in protein folding and lipoprotein biosynthetic processes, and to determine what chemical differences in cocoas of different processing are critical for explaining the differences in efficacy observed in this study.

Abbreviations:

ALT

alanine transferase

AST

aspartate aminotransferase

ANOVA

analysis of variance

APOB

apolipoprotein B

CVA

canonical variate analysis

ER

endoplasmic reticulum

ERO1α

endoplasmic reticulum oxidoreductin-1-α

ERRα

estrogen-related receptor α

ERRE

estrogen-related response elements

GPX7

glutathione peroxidase 7

HFD

high fat diet

IRE1α

inositol-requiring enzyme type 1

LOLO

fermented/roasted

LONO

fermented/unroasted

MANOVA

multivariate ANOVA

mtDNA

mitochondrial DNA

MTP

microsomal triglyceride transfer protein

NAFLD

non-alcoholic fatty liver disease

UPR

unfolded protein response

NOLO

unfermented/roasted

NONO

unfermented/unroasted

NRF2

nuclear respiratory factor 2

PDIA1

protein disulfide isomerase

PGC1α

proliferator-activated receptor γ coactivator-1α

SIRT

sirtuin

SOD2

superoxide dismutase2

TBST

tris-buffered saline containing 0.1% tween 20

TFAM

mitochondrial transcription factor A

VLDL

very-low-density lipoprotein