https://orcid.org/0000-0003-3858-7516
Abstract
Maternal obesity programs the risk for development of nonalcoholic fatty liver disease (NAFLD) in offspring. Maternal exercise is a potential intervention to prevent developmentally programmed phenotypes. We hypothesized that maternal exercise would protect from progression of NAFLD in offspring previously exposed to a maternal obesogenic diet. Female mice were fed chow (CON) or high fat, fructose, cholesterol (HFFC) and bred with lean males. A subset had an exercise wheel introduced 4 weeks after starting the diet to allow for voluntary exercise. The offspring were weaned to the HFFC diet for 7 weeks to induce NAFLD. Serum, adipose, and liver tissue were collected for metabolic, histologic, and gene expression analyses. Cecal contents were collected for 16S sequencing. Global metabolomics was performed on liver. Female mice fed the HFFC diet had increased body weight prior to adding an exercise wheel. Female mice fed the HFFC diet had an increase in exercise distance relative to CON during the preconception period. Exercise distance was similar between groups during pregnancy and lactation. CON-active and HFFC-active offspring exhibited decreased inflammation compared with offspring from sedentary dams. Fibrosis increased in offspring from HFFC-sedentary dams compared with CON-sedentary. Offspring from exercised HFFC dams exhibited less fibrosis than offspring from sedentary HFFC dams. While maternal diet significantly affected the microbiome of offspring, the effect of maternal exercise was minimal. Metabolomics analysis revealed shifts in multiple metabolites including several involved in bile acid, 1-carbon, histidine, and acylcarnitine metabolism. This study provides preclinical evidence that maternal exercise is a potential approach to prevent developmentally programmed liver disease progression in offspring.
Maternal obesity or obesogenic diet exposure is associated with development of metabolic syndrome and complications in offspring (1-4). Given that approximately 55% of women are overweight or obese at conception (5), understanding the effects of maternal obesity on offspring health is critical. A clear association between maternal prepregnancy body mass index and offspring nonalcoholic fatty liver disease (NAFLD) has been shown (6-8). Animal models (rodent and nonhuman primate) demonstrate that exposure to a maternal high-fat diet in utero and postnatally increase the propensity to develop NAFLD in offspring (9-12). The mechanisms behind this developmental programming event are still being defined, but 1 potential mode for transmission is through vertical transmission of an altered microbiome at birth. In models of maternal obesogenic diet exposure, the microbiome of offspring is altered (11, 12). There are also shifts in the microbiome of human offspring of mothers with body mass index ≥30 (13). As the underlying mechanisms continue to be defined, development of approaches to prevent programming of offspring liver disease are essential.
One potential approach to limit the effects of maternal obesity is through maternal exercise. Exercise during pregnancy in humans alters the maternal microbiome (13). It is also clear from rodent models that maternal exercise affects both the maternal and the offspring microbiome (14). The metabolic benefit of maternal exercise on mouse offspring was previously reported (15-18). Specifically, maternal exercise improves glucose tolerance in offspring through superoxide dismutase 3 signaling in the fetal liver (15). These effects are mediated in an epigenetic manner through histone modifications (19).
Recent studies have identified that maternal exercise protects offspring from the development of steatosis (20, 21). Whether maternal exercise protects offspring from NAFLD progression to fibrosis has not been defined. In the current study, we sought to identify whether maternal exercise, especially in the setting of maternal obesity, protects offspring from the development of steatosis, inflammation and fibrosis in a mouse model of nonalcoholic steatohepatitis (NASH). We utilized our established model of maternal obesogenic diet exposure to evaluate the effect of maternal exercise on offspring NAFLD development.
Materials and Methods
Mouse Breeding, Feeding and Exercise Paradigm, and Sample Collection
All procedures in this study were approved by the Animal Studies Committee at Washington University School of Medicine and conformed to National Institutes of Health guidelines and reporting remained consistent with ARRIVE guidelines. We fed 4-week-old female C57Bl/6J mice either a high-fat, fructose, cholesterol (HFFC) diet (Research Diets Inc. D09100310i; 40 kcal% fat [mostly palm oil], 20 kcal% fructose, and 2% cholesterol) or standard chow (CON) (Pico Lab rodent diet 20; 13% fat, 62% carbohydrates [3.2% sucrose], and 25% protein) for 6 weeks (Fig. 1) (22). A subset of female mice within each group had a low-profile wireless exercise wheel (Med Associates Inc.) added to the cage at 8 weeks of age to allow for voluntary exercise. Exercising mice were single housed to allow for monitoring of exercise distance. Female mice from all 4 groups (CON-sedentary, CON-active, HFFC-sedentary, and HFFC-active) were mated with chow-fed male mice at 10 weeks of age. Active dams continued to exercise voluntarily through pregnancy and lactation. Offspring were weaned onto the HFFC diet for 7 weeks to induce NASH. Serum and relevant tissues were collected at sacrifice. For these studies, at least 4 litters were represented in each group in each experiment. Body and liver weights were measured for all offspring at the time of weaning and at tissue collection. For all breeding, potential dams were staged to identify the most likely period for successful mating. The sire was placed in the cage for only 24 hours to limit cohousing effects. Females that exercised less than 2 km per day during the preconception period were excluded.
Experimental design and exercise data. (A) Breeding scheme, diet feeding, and maternal exercise. Created with BioRender.com. (B) Exercise distance in km/day during preconception period. (C) Exercise distance in km/day during pregnancy. (D) Exercise distance in km/day during lactation period. (E) Weights of female mice prior to addition of exercise wheel. (F) Weights of female mice at time of mating. (G) Dam weights at the time of offspring weaning. (H) Average litter size for each group. (I) Offspring weights at the time of weaning. Quantitative data presented as mean (±SD). For dam data, n ≥ 7 dams represented in each group. For offspring data, n ≥ 5 in each group and ≥5 separate litters represented in each group. P-values indicated on graph.
Serum Analysis
Blood was collected at sacrifice with serum isolated and frozen at −80 °C. Measurements of triglycerides (TGs) was undertaken by the Division of Comparative Medicine Research Animal Diagnostic Laboratory at Washington University.
Histology, Stains, and Immunohistochemistry
Tissues fixed in 10% formalin were embedded in paraffin, and 5-μm sections were stained with hematoxylin and eosin (H&E) and picrosirius red (PSR), as previously reported (23). For immunohistochemistry, sections were rehydrated by passing through xylene, graded alcohol, and distilled water and then stained with primary antibodies (Mac-2 AB_837133 and alpha smooth muscle actin AB_11129103) and the Vectastain Elite ABC-AP kit (Vector Laboratories, Burlingame, CA, USA) protocol. Antigen retrieval was performed by heating the sections for 15 minutes in citrate buffer (pH 6.0) in a pressure cooker. Endogenous peroxide activity was then blocked and tissues were incubated with primary antibody (∼20 hours, 4 °C). Sections were then washed and incubated with biotin-conjugated secondary antibody (30 minutes) (Vector Laboratories), washed, incubated with ABC-AP reagent, washed, and incubated with alkaline phosphatase red substrate kit (Vector Laboratories). Sections were counterstained with hematoxylin and cover-slipped using Cytoseal XYL (Richard Allen Scientific, Kalamazoo, MI, USA). PSR staining, alpha-smooth muscle actin, and Mac-2 immunohistochemistry were quantified using ImageJ to calculate the percent area of the section that was positive. For liver immunohistochemistry quantitation, 3 representative images were taken (with 10 × objective) for each mouse with 1 image coming from each of 3 different liver lobes. For all quantitative graphs for staining, each data point represents a single mouse.
Quantitative Polymerase Chain Reaction Analysis
Total hepatic and adipose RNA were extracted and cDNA prepared using an ABI high-capacity cDNA reverse transcription kit with 1 µg of total RNA. Real-time quantitative polymerase chain reaction (PCR) used cDNAs from at least 6 animals per group and was performed in duplicate on an ABI 7500 sequence detection system using SYBR Green PCR Master Mix (Applied Biosystems) and appropriate primer pairs (provided on request). Relative mRNA abundance is expressed as fold change to CON group after normalization to Actin.
16S Sequencing for Gut Microbiome Analysis
Cecal stool samples were collected and sent to MRDNA (http://www.mrdnalab.com) for 16S rRNA gene sequencing and bioinformatics analysis. DNA was isolated from cecal contents using the PowerSoil DNA Isolation Kit (Qiagen) per the manufacturer’s instructions. The samples were lysed with beads and spin filtered to elute purified DNA, which was stored at −20 °C until PCR amplification. A re-engineered version of bTEFAP, a form of amplicon sequencing utilizing next-generation sequencing, was used to evaluate the microbiota. The 16S rRNA primer pair, 515F GTGYCAGCMGCCGCGGTAA/806R GGACTACNVGGGTWTCTAAT, was utilized to evaluate the microbial ecology of each sample on the Illumina NovaSeq via the bTEFAP DNA analysis service. Each sample underwent a single-step 35 cycle PCR using HotStarTaq Plus Master Mix Kit (Qiagen, Valencia, CA) were used under the following conditions: 95 °C for 5 minutes, followed by 35 cycles of 95 °C for 30 seconds; 53 °C for 40 seconds, and 72 °C for 1 minute; after which a final elongation step at 72 °C for 10 minutes was performed. The amplification products from the different samples were mixed in equal concentrations, purified, and sequenced (Illumina NovaSeq) sequencing.
The Q25 sequence data derived from the sequencing process was processed using the MR DNA ribosomal and functional gene analysis pipeline (www.mrdnalab.com, MR DNA, Shallowater, TX). Sequences are depleted of primers, short sequences <150 bp are removed, and sequences with ambiguous base calls removed. Sequences are quality filtered using a maximum expected error threshold of 1.0 and dereplicated. The dereplicated or unique sequences are denoised; unique sequences identified with sequencing or PCR point errors are removed, followed by chimera removal, thereby providing a denoised sequence or zero-radius operational taxonomic unit (zOTU). Final zOTUs were taxonomically classified using BLASTn against a curated database derived from NCBI (www.ncbi.nlm.nih.gov) and compiled into each taxonomic level into both “counts” and “percentage” files. Alpha and beta diversity analysis were viewed through Qiime 2. Taxonomic data are presented as abundance relative to all counts.
Global Metabolomics
Liver samples from baseline and HFFC-fed offspring from each group (CON-sedentary, CON-active, HFFC-sedentary, HFFC-active) were collected for untargeted metabolomics by ultraperformance liquid chromatography tandem mass spectrometry–based methods at Metabolon (Morrisville, NC, USA). A full description of methods can be found elsewhere (24).
Statistical Analysis
Two-way analysis of variance with post hoc comparison or unpaired Student’s t-tests were used when appropriate, using GraphPad prism software and sample size noted in the figure legend of each figure. Data are presented as means (±SD) with 2-tailed P < .05 representing statistical significance and 2-tailed P > .05 and <.10 representing a nonstatistically significant trend.
Results
Female Mice Exposed to Maternal Obesogenic Diet Exercise More in Preconception Period and Similar During Pregnancy and Lactation Compared with Controls With no Effect on Offspring Weaning Weight
To evaluate the impact of maternal exercise on development of NAFLD in offspring, we utilized a model of maternal exercise previously reported (Fig. 1A) (17). Briefly, female mice were started on HFFC or CON diet for 4 weeks prior to adding a low-profile exercise wheel to the cage. The mice were continued on the same diet with or without an exercise wheel for an additional 2 weeks prior to mating. Exercising females were kept with their wheel during pregnancy and lactation. During the preconception period, HFFC females had an increased daily exercise distance compared with CON females (Fig. 1B). HFFC and CON females had a similar daily exercise distance during pregnancy and lactation periods (Fig. 1C and 1D). Prior to adding an exercise wheel, female mice fed the HFFC diet had a significantly higher body weight than CON females (Fig. 1E). HFFC-sedentary females continued to have higher weight at 6 weeks after starting the diet and HFFC-active females showed a trend toward a decrease in body weight compared with HFFC-sedentary females (P = .0682, Fig. 1F). Dam weights at weaning were similar across all 4 groups (Fig. 1G). Maternal HFFC decreased the average litter size significantly (Fig. 1H). HFFC-active average litter size increased relative to HFFC-sedentary but did not reach statistical significance (P = .0917). Offspring weights at weaning were also similar across all 4 groups (Fig. 1I).
Maternal Exercise Shift Pattern of Steatosis in Offspring
To evaluate whether maternal exercise had an effect on the development of NAFLD in offspring, male offspring from each group were weaned on to the HFFC diet for 7 weeks to induce hepatic steatosis, inflammation, and fibrosis. Body weight, liver weight, and liver weight to body weight ratio were similar across all 4 groups after HFFC feeding (Fig. 2A-2C). Serum TG levels were not different between groups (Fig. 2D). Histology showed shifts in the appearance of steatosis between groups. HFFC-sedentary offspring exhibited more severe steatosis than CON-sedentary offspring (Fig. 2E and 2H). This was associated with increased levels of hepatic TG in HFFC-sedentary compared with CON-sedentary (Fig. 2F). Levels of hepatic TG were not decreased in HFFC-active offspring compared with HFFC-sedentary offspring, but the pattern of steatosis was shifted. HFFC-sedentary offspring develop large droplet macrovesicular steatosis, while HFFC-active offspring developed either small droplet macrovesicular or microvesicular steatosis (Fig. 2E, microvesicular shown in inset). Gene expression analysis for genes involved in lipid metabolism identified several that were affected by maternal diet and exercise. Hepatic expression of Srebf1 was significantly increased in CON-active offspring compared with CON-sedentary and HFFC-sedentary offspring and significantly increased in HFFC-active offspring compared with CON-sedentary (Fig. 2G). There was also a trend toward an increase in HFFC-active (P = .0646). Expression of Pdk4 was increased in HFFC-active offspring compared with HFFC-sedentary offspring (Fig. 2G). To support histology findings, NAFLD scoring for steatosis was performed by a pathologist and showed increased frequency of higher steatosis scores in HFFC-sedentary offspring with mostly steatosis scores of 0 and 1 in each of the other groups (Fig. 2H).
Maternal exercise shifts hepatic lipid metabolism in male offspring. (A) Body weights (BW) of male offspring weaned to HFFC diet. (B) Liver weights of male offspring weaned to HFFC diet. (C) LW/BW ratio of offspring weaned to HFFC diet. (D) Levels of serum TGs from male offspring weaned to HFFC. (E) Representative photomicrographs of H&E staining for offspring weaned to HFFC diet (10 × objective on top, 20 × objective on bottom, inset image is at 40 × objective). (F) Hepatic triglyceride levels in offspring weaned to HFFC diet. (G) Relative expression of genes involved in lipid metabolism in liver of offspring weaned to HFFC. (H) Distribution of steatosis score among offspring in each group. (I) Distribution of inflammation score in each group. (J) Distribution of total NAS in each group. (K) Distribution of fibrosis score in each group. Quantitative data presented as mean (±SD) with n ≥ 6 in each group and ≥5 separate litters represented in each group. P values indicated on graph.
Maternal Exercise Attenuates Programming of Worse Hepatic Inflammation and Fibrosis in Offspring by Maternal Obesogenic Diet Exposure
To assess NAFLD progression in offspring exposed to maternal exercise, staining for Mac-2 (macrophages) and picrosirius red (collagen) were performed. As previously reported, staining for both Mac-2 and PSR were increased in HFFC-sedentary offspring compared with CON-sedentary offspring, reaffirming worse NASH in maternal HFFC diet–exposed (MHDE) offspring (Fig. 3A, 3B, and 3D). Both CON-active and HFFC-active offspring exhibited less Mac-2 staining than CON-sedentary and HFFC-sedentary, respectively (Fig. 3A and 3B). No difference was observed in expression of Lect2, Mcp1, and Il1b (Fig. 3C). HFFC-active offspring exhibited less hepatic PSR staining than HFFC-sedentary offspring (Fig. 3A and 3D). Alpha-smooth muscle actin positive stellate cells were present in HFFC-sedentary liver, but were less abundant in HFFC-active offspring (Fig. 3A and 3E). HFFC-sedentary offspring had increased expression of Col1a1 and Col3a1, which was attenuated in HFFC-active offspring (Fig. 3E). Likewise, Mmp12 and Timp1 expression was increased in HFFC-sedentary liver and decreased in HFFC-active liver (Fig. 3F and 3G). NAFLD scoring showed higher inflammatory and total NAS scores in HFFC-sedentary offspring than in CON-sedentary, CON-active, and HFFC-active offspring (Fig. 2I and J). HFFC-sedentary offspring also had higher fibrosis scores (Fig. 2K).
Maternal exercise protects male offspring from development of inflammation and fibrosis during NAFLD development. (A) Representative photomicrographs of Mac-2, Picrosirius Red (PSR), and alpha-smooth muscle actin (SMA) staining of liver from male offspring weaned to HFFC diet. (B) Quantification of Mac-2 positive area in liver of offspring weaned to HFFC. (C) Relative expression of Lect2, Mcp1, and Il1b in liver of offspring weaned to HFFC. (D) Quantification of PSR positive area in liver of offspring weaned to HFFC. (E) Relative expression of collagen genes (Col1a1, Col3a1, Col4a1) in liver of offspring weaned to HFFC. (F) Relative expression of MMP genes (Mmp12, Mmp13) in liver of offspring weaned to HFFC. (G) Relative expression of TIMP genes (Timp1, Timp2) in liver of offspring weaned to HFFC. Quantitative data presented as mean (±SD) with n ≥ 7 in each group and ≥5 separate litters represented in each group. P values indicated on graph.
No Difference in Female Offspring During NAFLD Development With Maternal Exercise
The effect of maternal high-fat diet and exercise on female offspring is not clearly defined. To assess NAFLD progression in female offspring from our model, female offspring from each group were fed the HFFC diet for 7 weeks. No difference in body weight, liver weight, and liver weight/body weight were observed between groups (Fig. 4A-4C). There was also no difference observed in serum TG levels (Fig. 4D). On H&E staining, no clear differences in steatosis were present (Fig. 4E). Histology showed minimal inflammation and fibrosis in all 4 groups (Fig. 4E-4G). Alpha-smooth muscle actin staining did not show the presence of activated stellate cells in any of the groups (Fig. 4E and 4H). Gene expression analysis for genes involved in lipogenesis (Fig. 4I) and fatty acid oxidation/transport (Fig. 4J) showed a significant decrease in Cpt1 expression in HFFC-sedentary offspring compared with CON-sedentary. No additional changes in gene expression were observed in the genes evaluated.
Maternal exercise protects female offspring from development of steatosis. (A) Body weights (BW) of female offspring weaned to HFFC diet. (B) Liver weights of female offspring weaned to HFFC diet. (C) LW/BW ratio of female offspring weaned to HFFC diet. (D) Levels of serum triglycerides in female offspring weaned to HFFC diet. (E) Representative photomicrographs of H&E, Mac-2, PSR, and alpha SMA staining for female offspring weaned to HFFC diet. (F) Quantification of Mac-2 positive area in liver of female offspring weaned to HFFC. (G) Quantification of PSR positive area in liver of female offspring weaned to HFFC. (H) Quantification of alpha-SMA positive area in liver of female offspring weaned to HFFC diet. (I) Relative expression of genes involved in lipogenesis. (J) Relative expression of genes involved in fatty acid oxidation/transport. Quantitative data presented as mean (±SD) with n ≥ 5 in each group and ≥5 separate litters represented in each group. P-values indicated on graph.
Adipocyte Size and Inflammation in Male Offspring After HFFC Diet Feeding
The benefit of maternal exercise on offspring adiposity was previously reported (18). We evaluated adipocyte size and inflammation in offspring that were fed HFFC diet to identify any changes induced by maternal diet and/or exercise. No difference in adipocyte size was observed between CON-sedentary and HFFC-sedentary offspring (Fig. 5A and 5B). However, HFFC-active offspring had an increase in average adipocyte size (Fig. 5A and 5B). Expression Srebf1 was increased in HFFC-active offspring compared with CON-sedentary and CON-active (Fig. 5C). A similar increase in Elovl6 was observed in HFFC-active offspring (Fig. 5D). No difference was observed in Pparg expression (Fig. 5E).
Effect of maternal exercise on male offspring adipocyte size and inflammation during HFFC feeding. (A) Representative photomicrographs of H&E staining of eWAT from offspring after 7 weeks of HFFC feeding. (B) Quantitation of adipocyte area from offspring after 7 weeks of HFFC feeding. (C) Relative expression of Srebf1 in eWAT. (D) Relative expression of Elovl6 in eWAT. (E) Relative expression of Pparg in eWAT. (F) Representative photomicrographs of Mac-2 staining of eWAT from offspring after 7 weeks of HFFC feeding. (G) Quantitation of crown-like structures (CLS) in eWAT from offspring after 7 weeks of HFFC feeding. (H) Relative expression of Tnfa in eWAT. I. Relative expression of Mcp1 in eWAT. (J) Relative expression of Adipoq in eWAT. Quantitative data presented as mean (±SD) with n ≥ 5 in each group and ≥5 separate litters represented in each group. P values indicated on graph. eWAT-epididymal white adipose tissue.
To evaluate adipose inflammation, staining for Mac-2 was performed to quantify crown-like structures. No difference in the number of crown-like structures was identified between groups (Fig. 5F and 5G). However, expression of the proinflammatory cytokine Tnfa was decreased in HFFC-active offspring compared with CON-sedentary and CON-active (Fig. 5H). We also observed a decrease in Mcp1 expression in HFFC-sedentary and HFFC-active offspring compared with CON-sedentary (Fig. 5I).
Crosstalk between adipose tissue and liver plays an important role in the development of NAFLD (reviewed in (25)). One mechanism of crosstalk is through production of the adipokine adiponectin. Decreased serum levels of adiponectin are associated with NAFLD (26). We measured adiponectin expression in eWAT of offspring. Expression of Adipoq was increased in HFFC-active offspring (Fig. 5J). These data suggest adaptations in the HFFC-active offspring adipose tissue that could contribute to protection from development of NAFLD.
Microbiome Analysis of Male Offspring After HFFC Diet Feeding
We previously reported shifts in the microbiome after maternal obesogenic diet exposure and a role in driving both steatosis and cholestatic liver disease in the offspring (27, 28). Maternal exercise also shifts the offspring microbiome (14). We performed 16S sequencing on cecal contents of offspring fed the HFFC diet from each group. Principal component analysis identified a clear effect of maternal diet but no statistical difference in beta-diversity between offspring from sedentary and active dams (Fig. 6A). Measures of alpha-diversity (observed OTUs and Faith PD) were decreased in HFFC offspring, but no change with maternal exercise (Fig. 6B). At the phylum level, HFFC offspring exhibited decreased abundance of Firmicutes and increased abundance of Verrucomicrobia (Fig. 6C and 6E). Maternal exercise did not affect the abundance of different phyla. The increase in Verrucomicrobia was primarily Akkermansia (Fig. 6D and 6F). HFFC offspring had decreased abundance of Lachnoclostridium. Maternal exercise was associated with an increase in Clostridium in HFFC offspring (Fig. 6D and 6F).
Microbiome analysis of cecal contents from male offspring after HFFC feeding. (A) Bray–Curtis plot for beta-diversity of cecal microbiome from offspring after 7 weeks of HFFC feeding with statistical analysis below. (B) Measures of alpha-diversity (observed OTUs and faith PD) in cecal microbiome of offspring after 7 weeks of HFFC feeding. (C) Relative abundance of each bacterial phyla in cecal microbiome of offspring after 7 weeks of HFFC feeding. (D) Relative abundance of the top genera in cecal microbiome of offspring after 7 weeks of HFFC feeding. (E) Phyla with statistically significant differences between groups. (F) Genera with statistically significant differences between groups. Quantitative data presented as mean ±SD with n = 6 in each group and 6 separate litters represented in each group. P values as indicated on graph.
Maternal Diet and Exercise Affect the Male Offspring Hepatic Metabolome
Maternal diet and exercise have the potential to induce global changes in the offspring metabolome, which could have implications on development of metabolic liver disease. Unbiased global metabolomics analysis was performed on liver from all groups at baseline (3 weeks of age) and after 7 weeks of HFFC diet feeding. In total, 1149 metabolites were analyzed and a summary of the numbers of metabolites that achieved statistical significance (P ≤ .05), as well as those approaching significance (.05 < P < .10), is shown in Fig. 4A and 4B. At baseline, 24 metabolites were significantly different between CON-active and CON-sedentary offspring while 124 metabolites were significantly different between HFFC-active and HFFC-sedentary offspring (Fig. 7A). In the fatty liver disease groups, only 39 metabolites were significantly different between CON-active and CON-sedentary, but 170 metabolites were significantly different between HFFC-active and HFFC-sedentary (Fig. 7B). Several metabolites in bile acid (BA) metabolism are affected by both maternal diet and exercise, consistent with our prior reports (28, 29). Cholesterol was unchanged in all groups, but the BA precursor 7 alpha-hydroxy-3-oxo-4-cholestenoate (7-Hoca) is increased in active groups compared with sedentary, an effect that is exaggerated in offspring fed the HFFC (Fig. 7C). Conjugated and modified primary BAs including taurocholate, glycochenodeoxycholate, and cholate sulfate are similarly elevated in the active compared with sedentary particularly in the livers from offspring of mothers fed a chow diet. Notably, ursodeoxycholate is increased in offspring exposed to maternal exercise during HFFC feeding (Fig. 7C).
Maternal diet and exercise shift the hepatic metabolome in male offspring. (A) Number of metabolites different between each group in liver of 3 week old offspring. (B) Number of metabolites different between each group in liver of offspring after 7 weeks of HFFC diet. (C) Representative metabolites in bile acid metabolism that are different among the groups. (D) Representative metabolites in 1 carbon metabolism that are different among the groups. (E) Representative acylcarnitines that are different among groups. Quantitative data presented as mean ±SD with n ≥ 6 in each group and ≥4 separate litters represented in each group. P-values as indicated on graph or *−P < .05, **−P < .01.
Another area where several metabolites are affected is 1-carbon metabolism. Sarcosine and S-adnosylhomocysteine are increased in offspring previously exposed to maternal exercise during HFFC feeding (Fig. 7D). Choline is also increased in HFFC-active offspring during HFFC diet feeding.
A prominent effect is observed on monoacyglycerides in the HFFC-active offspring at baseline as most monoacyglycerides are significantly increased compared with HFFC-sedentary offspring (Table S1 (24)). A similar effect was not observed in CON-active offspring. In the setting of offspring fed HFFC diet, acylcarnitines were significantly decreased in HFFC-active offspring compared with HFFC-sedentary offspring. In particular, 4 acylcarnitines were increased in the HFFC-sedentary offspring compared with CON-sedentary and were normalized by prior maternal exercise (Fig. 7E; Table S1 (24)).
Pathway analysis was performed on significantly different metabolites between groups utilizing MetaboAnalyst 5.0. Analysis of differential metabolites between CON-sedentary and HFFC-sedentary identified sphingolipid metabolism as a differentially represented pathway (Fig. 8A). Differential analysis between HFFC-sedentary and HFFC-active metabolites also identified sphingolipid metabolism (Fig 8B). Multiple other pathways were highlighted between HFFC-sedentary and HFFC-active, including nicotinamide metabolism, histidine metabolism, purine metabolism, and arginine biosynthesis. On metabolite set enrichment analysis, histidine metabolism was the top differential pathway between HFFC-sedentary and HFFC-active (Fig. 8C). Levels of histidine, formiminoglutamate, glutamate, and alpha-ketoglutarate were all increased in HFFC-active offspring compared with HFFC-sedentary offspring (Fig. 8D). Levels of glutamine were significantly decreased in HFFC-active offspring (Fig. 8D). These findings suggest increase in histidine metabolism with flux directed toward the tricarboxylic acid cycle rather than glutamine production.
Pathway analysis of metabolomics data. (A) Pathway analysis with MetaboAnalayst 5.0 of metabolites significantly different between CON-sedentary and HFFC-sedentary offspring after 7 weeks of HFFC diet. (B) Pathway analysis of metabolites significantly different between HFFC-sedentary and HFFC-active offspring after 7 weeks of HFFC diet. (C) Metabolite set enrichment analysis of differential metabolites between HFFC-sedentary and HFFC-active offspring liver. (D) Histidine metabolism with notation of fold change of metabolites and associated P value. Created with BioRender.com.
Discussion
The association between maternal obesity/obesogenic diet exposure and offspring susceptibility to NAFLD is evident from both the birth cohort and the animal model (rodent and nonhuman primates) studies (6, 10, 12). The mechanisms that drive risk in the offspring are still being defined. We present here preclinical evidence that maternal exercise attenuates MHDE-induced worsening of NAFLD. MHDE offspring of exercised dams developed less inflammation and fibrosis when fed a NASH-inducing diet compared with MHDE offspring of sedentary dams. It was previously reported that maternal exercise protects control offspring from development of NAFLD (30, 31). A recent study showed maternal exercise protects offspring exposed to maternal high fat/high sugar diet from development of NAFLD (20). To our knowledge, this is the first demonstration of protection from NAFLD progression to fibrosis in mice after maternal exercise exposure in the setting of maternal obesity. Several aspects of this work warrant further discussion.
Maternal Exercise Limits Liver disease Progression in MHDE Offspring
Prior studies show that MODE promotes worse hepatic inflammation and fibrosis in offspring during the development of NAFLD (32, 33). We show here similar findings that hepatic inflammation is worse in HFFC offspring fed the HFFC diet than CON offspring fed the HFFC diet. Given that degree of fibrosis is a primary predictor of outcome in patient with NAFLD, approaches to prevent disease progression are important (34). We show in this study that maternal exercise prevented disease progression to fibrosis, and notably did so in MHDE offspring. While CON-sedentary offspring did not develop fibrosis, maternal exercise did prevent development of hepatic inflammation in CON offspring, suggesting that protection from fibrosis is likely in offspring of lean dams as well. No study to date has described protection from liver fibrosis in offspring after maternal exercise and thus no mechanistic leads have been identified. We observed a decrease in expression of both Col1a1 and Col3a1, suggesting that multiple components of collagen fibril formation are decreased. It is unclear whether these shifts in expression are due to epigenetic modifications or simply a result of less steatosis and necroinflammation leading to less stellate cell activation. Our findings will provide a model to begin evaluating mechanisms for protection from hepatic fibrosis which will be important to support maternal exercise as a preventive approach in the setting of maternal obesity. Mechanistic studies will also guide development of therapeutic approaches to provide similar benefit as observed with maternal exercise. Our metabolomics data set may provide insight into potential metabolic pathways involved. Our metabolite set enrichment analysis identified histidine metabolism as the most overrepresented among the set of significantly altered metabolites with several of the metabolites in the pathway increased by maternal exercise. Notably, histidine was protective in an experimental model of hepatic fibrosis (35). Knockout of histidine decarboxylase was protective in models of NAFLD and primary sclerosing cholangitis (36, 37). While not evaluated in these studies, knockout of histidine decarboxylase would likely increase levels of histidine which could be mechanistically involved in the protection observed. Further evaluation of this metabolic pathway is warranted in liver disease as well as how this pathway is affected by exercise.
Maternal Exercise Shifts Bile Acid Levels in Offspring
Given the previously described benefits of maternal exercise on offspring metabolism, we hypothesized that maternal exercise would decrease development of NAFLD in offspring. Prior studies showed a benefit of this intervention on development of steatosis in the offspring (30, 31). Part of the shift in hepatic metabolism may be due to shifts in energy sensing pathways and glucose metabolism (30). These studies only evaluated maternal exercise with the dam fed a chow diet. We show here that maternal exercise did not affect TG accumulation but did shift the steatosis pattern in MHDE offspring also protects from steatosis in offspring with developmental programming induced by MODE. The mechanism and significance of this shift are not yet clear of protection is not yet clear, but changes in BA metabolism could affect steatosis metabolomics analysis showed shifts in several BAs. Our metabolomics analysis did identify changes in several BAs in the liver. Specifically, ursodeoxycholic acid (UDCA) was increased in offspring previously exposed to maternal exercise during HFFC diet feeding. The benefits of UDCA in reversing NAFLD have been reported in rodents (38, 39). However, 1 clinical study showed that short-term high-dose UDCA resulted in increased neutral lipid accumulation in liver (40). We have previously reported shifts in BA metabolism in maternal high fat/high sucrose diet–exposed offspring (28, 29). A more thorough analysis of the effect of maternal exercise on offspring BA homeostasis will be essential to delineate the role of such changes in protecting offspring from NAFLD.
Changes in BA levels could be an indicator of maternal exercise induced shifts in the offspring microbiome. Indeed, several studies have shown that maternal exercise can affect the composition of the offspring microbiome in mice with associated improvement in metabolism (14, 41). MODE also alters the offspring microbiome and drives changes in offspring BA metabolism (28). We confirm in this study that MHDE significantly alters the offspring microbiome, and changes are still present in offspring that have been fed a NASH-inducing diet for 7 weeks. However, we only observed minimal effect of maternal exercise and no difference in alpha- and beta-diversity. One limitation of this analysis is the performance of 16S sequencing after 7 weeks of HFFC diet. It is possible that maternal exercise shifts the microbiome in younger offspring and that this could have an effect on developmental programming and the initial response to HFFC feeding. Future studies will be important to evaluate how maternal exercise affects the MHDE-induced changes in the early offspring microbiome and if maternal exercise induced shifts are beneficial to the metabolic health of the offspring.
Maternal Diet and Exercise Shift Acylcarnitines in the Offspring Liver
Maternal exercise in the setting of MHDE resulted in decreases in acylcarnitines in the liver of offspring during HFFC feeding compared with HFFC-sedentary offspring. Acylcarnitines have an important role in the development of NAFLD. Acylcarnitines are an intermediate of beta-oxidation and can accumulate in the setting of incomplete beta-oxidation of fatty acids (reviewed in (42)). We show in our metabolomic analysis that acylcarnitines are increased in HFFC-sedentary offspring compared with CON-sedentary offspring fed the HFFC diet. Notably maternal exercise in MHDE offspring decreased many of the measured acylcarnitine species. This is consistent with recent studies in mice showing that maternal exercise shifts hepatic acylcarnitine levels in offspring (20, 21). Further evaluation of acylcarnitines in this model may define a mechanistic role. Notably, treatment of NAFLD in mice using a methyl donor decreased steatosis and was associated with a decrease in acylcarnitine levels (43).
Adipose Adaptations Specific to HFFC-Active Offspring may Contribute to Protection From NAFLD
One mechanism by which maternal exercise may impact hepatic lipid metabolism and development of NAFLD is through adaptations in the white adipose tissue and ultimately adipose–liver crosstalk. We identified in our study that the average adipocyte size in offspring exposed to both maternal HFFC diet and exercise was larger and that this occurred without evidence of adipose inflammation. An increase in the adipocyte size could suggest an increased capacity for storage of fat in the adipose tissue and thus decrease the distribution to the liver. This idea is consistent with adipose tissue expandability hypothesis such that the once the limit is reached for storage of lipid in adipose tissue the net flux to nonadipose tissues will increase (44). We also identified an increase in adiponectin expression in the adipose tissue of HFFC-active offspring. Adiponectin is known to play a protective role in the development of NAFLD (45). While these findings are only associative at this point, they highlight an unexpected finding that point toward a potential mechanism by which maternal exercise may be beneficial specifically in the setting of maternal high-fat diet exposure. Future studies will be necessary to identify if these adipose adaptations are causal in the protection and how this may translate to a potential protective effect of maternal exercise in humans.
In addition to the already noted limitations, a few others warrant discussion. In our current study, exercise was maintained from before pregnancy until the time of weaning. There are likely windows during development where maternal exercise has the greatest affect and our study does not allow for delineation of the affect during and before pregnancy vs during the lactation period. It is already clear from prior studies that exercise during the lactation period has a critical impact on the composition of the breast milk (46). Maternal exercise led to increases in the oligosaccharide 3′-sialyllactose in breast milk of both humans and mice, which is critical for the observed metabolic benefits in the offspring. A critical analysis of maternal exercise during each window during development will provide important information for development of the best exercise based approaches during and around pregnancy. Another limitation of the current study is that the fiber content and micronutrient composition are not the same between the chow and HFFC diet. It is possible that these differences will contribute to some of the maternal diet driven changes that are observed. Specifically, fiber content affects the microbiome and susceptibility to inflammatory disease (47). Maternal fiber content also affects the offspring microbiome and development of obesity (48). Future studies evaluating maternal fiber content and offspring NAFLD progression will be important to delineate the contribution of individual dietary factors. While female offspring were included in this study, we were unable to define if similar protection from inflammation and fibrosis occurs as observed in male offspring. Neither CON-sedentary or HFFC-sedentary offspring developed significant inflammation or fibrosis suggesting that longer exposure to the HFFC diet than 7 weeks will be necessary. Future studies will identify if MHDE leads to worse inflammation and fibrosis in offspring after a longer period of HFFC diet feeding and subsequently if maternal exercise decrease inflammation/fibrosis in this experimental design.
Conclusions
In summary, maternal HFFC diet exposure leads to worse NASH in male mouse offspring. Maternal exercise protects offspring from development of inflammation and progression to fibrosis during NASH development in male offspring previously exposed to maternal HFFC diet. While it is clear there is a protective effect from maternal exercise on offspring NAFLD, the mechanism is not defined. Our metabolomics analysis provides direction for potential mechanisms including shifts in 1 carbon metabolism, which could impact epigenetics in the offspring. Histidine metabolism is also altered, which could affect development of hepatic fibrosis. Further evaluation of the early microbiome is also necessary to definitively evaluate the effect of maternal exercise on the offspring microbiome. Identification of protective mechanisms is a critical next step. We propose that maternal exercise is a potential approach to limit developmental programming of liver disease caused by maternal obesity.
Acknowledgments
We thank the Advanced Imaging and Tissue Analysis Core for processing of tissues for histology (DDRCC Grant P30 DK052574)
Funding
NIH DK-122018, NORC DK-56342, DRC DK-20579, DDRCC DK-52574 and the American Gastroenterological Association Research Scholar Award (M.D.T.).
Conflict of Interest
No conflicts of interest exist.
Data Availability
Original data generated and analyzed during this study are included in this published article or in the data repositories listed in References.
References
Abbreviations
- BA
bile acid
- CON
standard chow
- eWAT
epididymal white adipose tissue
- H&E
hematoxylin and eosin
- HFFC
high fat, fructose, cholesterol
- MHDE
maternal HFFC diet exposure
- NAFLD
nonalcoholic fatty liver disease
- NASH
nonalcoholic steatohepatitis
- PSR
picrosirius red
- UDCA
ursodeoxycholic acid
- TG
triglyceride
