Maternal high-fat feeding primes steatohepatitis in adult mice offspring, involving mitochondrial dysfunction and altered lipogenesis gene expression


  • Potential conflict of interest: Nothing to report.


Nonalcoholic fatty liver disease (NAFLD) describes an increasingly prevalent spectrum of liver disorders associated with obesity and metabolic syndrome. It is uncertain why steatosis occurs in some individuals, whereas nonalcoholic steatohepatitis (NASH) occurs in others. We have generated a novel mouse model to test our hypothesis: that maternal fat intake contributes to the development of NAFLD in adult offspring. Female mice were fed either a high-fat (HF) or control chow (C) diet before and during gestation and lactation. Resulting offspring were fed either a C or a HF diet after weaning, to generate four offspring groups; HF/HF, HF/C, C/HF, C/C. At 15 weeks of age, liver histology was normal in both the C/C and HF/C offspring. Kleiner scoring showed that although the C/HF offspring developed nonalcoholic fatty liver, the HF/HF offspring developed NASH. At 30 weeks, histological analysis and Kleiner scoring showed that both the HF/C and C/HF groups had NAFLD, whereas the HF/HF had a more severe form of NASH. Therefore, exposure to a HF diet in utero and during lactation contributes toward NAFLD progression. We investigated the mechanisms by which this developmental priming is mediated. At 15 weeks of age, hepatic mitochondrial electron transport chain (ETC) enzyme complex activity (I, II/III, and IV) was reduced in both groups of offspring from HF-fed mothers (HF/C and HF/HF). In addition, measurement of hepatic gene expression indicated that lipogenesis, oxidative stress, and inflammatory pathways were up-regulated in the 15-week-old HF/C and HF/HF offspring. Conclusion: Maternal fat intake contributes toward the NAFLD progression in adult offspring, which is mediated through impaired hepatic mitochondrial metabolism and up-regulated hepatic lipogenesis. (HEPATOLOGY 2009.)

Nonalcoholic fatty liver disease (NAFLD) is one of the most common causes of chronic liver disease in Western countries.1 The pathogenesis of NAFLD is the subject of substantial research interest, because its incidence in adults and children is rising rapidly because of ongoing epidemics of obesity and type 2 diabetes.2, 3 NAFLD occurs commonly with metabolic syndrome (central obesity, type 2 diabetes, dyslipidemia, and cardiovascular disease) and is considered the hepatic manifestation of this condition.4–6 Recent estimates of prevalence in the United States are 20% to 30% for hepatic steatosis and 3.5% to 5% for nonalcoholic steatohepatitis (NASH), and up to 80% of people with type 2 diabetes may have some form of NAFLD.7, 8 NAFLD is linked to excessive triglyceride (TG) accumulation and encompasses a broad spectrum of liver disease, ranging from simple fatty liver (steatosis) to nonalcoholic steatohepatitis (NASH) with fibrosis that may progress to liver cirrhosis, portal hypertension, and even hepatocellular carcinoma.5, 9

It is unlikely that such a high prevalence of NAFLD can be explained by obesity alone, because not all obese individuals develop NAFLD, and not all individuals with simple steatosis progress to NASH. In fact, the pathogenic mechanisms involved in the disease progression from simple steatosis to NASH are uncertain. One candidate factor for disease progression is maternal obesity, since one third of women at childbearing age in the United States are currently obese.10 Maternal obesity at conception has been shown to alter gestational metabolism and affects placental, embryonic, and fetal growth and development.10 In addition, there is increasing evidence to show that exposure to poor nutrition in the developmental environment increases the risk of developing noncommunicable disease, specifically features of the metabolic syndrome such as dysglycemia, type 2 diabetes, and obesity later in life.11–18 Our group has recently presented preliminary data to suggest that maternal high fat feeding can influence the development of hepatic steatosis in adult (36-week-old) mice.19 However, the role of maternal nutrition on the development of more severe NASH onset in adulthood is unknown. More importantly, the biochemical and molecular mechanisms underlying this increased susceptibility are yet to be determined.

Because mitochondria are a major site of fatty acid oxidation and are maternally inherited, they provide an ideal candidate mechanism for the inheritance of diet-induced maternal stress. Mitochondrial dysfunction has previously been described in animal models and in human patients with diet-induced NASH.20–26 Specifically, adult rats fed a high-fat diet for several weeks have been shown to exhibit hepatic steatosis, reduced electron transport chain (ETC) capacity, and increased oxidative stress in liver mitochondria.27 However, hepatic mitochondrial impairment and hepatic steatosis resulting from a high-fat diet fed early in development has not previously been described.

NAFLD is primarily a result of inappropriate fat storage, an issue that is emerging as central to the pathogenesis of metabolic syndrome components such as insulin resistance and cardiovascular disease. Any mechanism leading to “ectopic” fat accumulation, influenced by nutrition in the developmental period, must involve persistent alterations in lipid homeostasis. Recently, several mouse models have helped to clarify the molecular mechanisms leading to the development of hepatic steatosis in the pathogenesis of NAFLD. These are multiple and include enhanced nonesterified fatty acid release from adipose tissue (lipolysis), increased de novo fatty acid and TG synthesis (lipogenesis), and decreased β-oxidation.28 However, similar to mitochondrial function, the influence of maternal high fat feeding on these pathways remains unknown.

In response to these observations, the aim of this study was to test the hypothesis that offspring of dams fed a high-fat (HF) diet during gestation and lactation are predisposed to develop a liver condition similar to human NASH in adulthood, because of persistent alterations in mitochondrial metabolism and lipogenesis. We also aimed to test whether offspring of dams fed a HF diet that also consumed a HF diet after weaning (in other words, suffering both a developmental and adult nutritional insult HF/HF) would exhibit a exacerbated form of the disease, in keeping with human NASH. We now report that offspring of dams fed an HF diet before or during pregnancy and lactation exhibit impaired mitochondrial ETC complex activity and up-regulated lipogenesis gene expression, contributing to the development and disease progression of a NASH-like phenotype.


C, control; C/C, control fed offspring from control fed dams; C/HF, high fat fed offspring from control fed dams; ETC, electron transport chain; HF, high fat; HF/C, control fed offspring from high fat fed dams; HF/HF, high fat fed offspring from high fat fed dams; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; PCR, polymerase chain reaction; TAG, triacylglycerol; TG, triglyceride.

Materials and Methods

Animal Model.

All studies were conducted under UK Home Office License. Female C57 BL6J mice (n = 20), were maintained under controlled conditions and randomly assigned to either a HF diet (HF; 45% kcal fat, 20% kcal protein, 35% kcal carbohydrate; Special Diet Services, UK) or standard chow diet (C; 21% kcal fat, 17% kcal protein, 63% kcal carbohydrate; RM1—Special Diet Services, UK) (for more detailed dietary constituents, see Supporting Table 1). Dams were fed 4 weeks before conception and during gestation and lactation. Weight gain and food intake were measured at these periods. Litter size was standardized to six pups, to ensure no litter was nutritionally biased. At weaning, the offspring were assigned either HF or C diet, generating four experimental groups: HF/HF (n = 8), HF/C (n = 7), C/HF (n = 8), C/C (n = 6), which represents prenatal, pregnancy, and lactation versus the postweaning diet, respectively (Supporting Fig. 1). Offspring body weights were recorded at weekly intervals. Offspring were killed at 15 weeks of age, and a subgroup were allowed to grow until 30 weeks old. Fat depots (in other words, gonadal, retroperitoneal, intrascapular, inguinal, and perirenal) were dissected and weighed. Cumulative weights of the fat depots for each animal and body fat as a percentage of total body weight were calculated. The livers were dissected, fixed in 10% neutral buffered formaldehyde, and stored for further histological analyses. Data from males and females showed the same pattern in the variables measured; therefore, for the purpose of this manuscript, we only present female data.


Fixed liver sections were stained with hematoxylin-eosin for visual assessment of steatosis and cellular infiltrate. Alternate sections were stained with Oil-Red-O to visualize lipid accumulation. CD44 protein expression was visualized using an anti-CD44 antibody (Abcam ab41478). Antigen retrieval was performed by microwaving for 25 minutes in 0.01 M citrate buffer pH6.0. Sections were then washed in Tris-buffered saline (3 × 2 minutes), and avidin solution was applied for 20 minutes. Slides were repeatedly washed with Tris-buffered saline (3 × 2 minutes), followed by incubation in biotin solution for 20 minutes. The sections were washed once more, followed by application of culture medium for 20 minutes, and incubated in a 1/1000 dilution with anti-CD44 antibody overnight at 4°C. Sections were then washed in Tris-buffered saline (3 × 5 minutes) and were incubated with biotinylated swine anti-rabbit secondary antibody (1/400 dilution) for 30 minutes, followed by addition of strepdavidin biotin-peroxidase vectors (1/75 dilution) for 30 minutes. CD44 protein was visualized after addition of 3,3′-diaminobenzidine substrate for 5 minutes, sections were counterstained with Mayer's hematoxylin, and photomicrographs were taken.

Kleiner Scores.

The Kleiner scoring system was used to assess the severity of NAFLD.29 An activity score was generated by adding the individual scores for the following features; steatosis (<5% = 0, 5%–33% = 1, 33%–66% = 2, >66% = 3); ballooning (none = 0, few = 1, prominent = 2); and lobular inflammation (none = 0, <2 foci = 1, 2–4 foci = 2, >4 foci = 3). A score of less than 3 correlates with mild nonalcoholic fatty liver, a score of 3 to 4 correlates with moderate nonalcoholic fatty liver, and a score of 5 or more correlates with NASH. The average (mode) score for each histological characteristic in each group was used.

Plasma Analysis.

Plasma insulin concentrations were determined using a mouse insulin enzyme-linked immunosorbent assay kit. Glucose and cholesterol concentrations were measured. β-hydroxybutyrate concentrations were determined using 150 μL plasma from each animal, using a Ranbut test (Randox, according to manufacturer's instructions) on an autoanalyzer (RX series, Daytona).

Mitochondrial Complex Activity.

Liver was homogenized on ice in 1:9 (wt/vol) buffer; 320 mM sucrose, 1 mM ethylenediaminetetra-acetic acid dipotassium salt, 10 mM Trizma-base, using a mechanical homogenizer (Kinematica, Switzerland). Samples were snap-frozen in liquid nitrogen in 50-μL aliquots and stored at −80°C. The activities of the electron transport chain enzymes from the offspring's liver tissue were determined using a spectrophotomic activity assay as previously described.30 For assay of nicotinamide adenine dinucleotide phosphate, reduced form, coenzyme Q1 (complex I), 20 μL sample was added to a cuvette and 1 mM rotenone added for assessment of rotenone-dependent complex I enzyme by spectrophotometry. For assay of succinate cytochrome c reductase (complex II + III), antimycin was added for assessment of the antimycin-dependent complex II + III enzyme rate. For assay of cytochrome c oxidase (complex IV), 20 μL sample was added to a cuvette containing 980 μL 100 mM potassium phosphate buffer, pH 7.5, 50 μM reduced cytochrome c. The reaction was run for 5 minutes, and k was calculated from the curve. For assay of citrate synthase, 20 μL sample was added to a cuvette containing 980 μL 100 mM Tris/0.1% (vol/vol) Triton, pH 8.0, 10 mM acetyl CoA, 20 mM oxaloacetate, and 20 mM 5,5′-dithio-bis (2-nitrobenzoic acid). The reaction was run for 5 minutes, and the rate was calculated from the linear part of the curve in nanomoles per minute per milliliter. Protein content was determined by a commercial modified Lowry method (Bio-Rad, UK).

Mitochondrial copy number was quantified by determining the ratio of mitochondrial 16S to genomic18S. Total DNA was extracted using a Wizard genomic DNA purification kit (Promega) according to the manufacturer's instructions. Mitochondrial 16S and genomic 18S copy number was determined using Sybr green real-time polymerase chain reaction (PCR).

Microarray Analysis.

Because of financial constraints, we could not perform a whole genome array for each biological replicate. To obtain a reliable estimate of the mean gene expression for each of the four offspring groups, we performed microarray with pooled RNA samples, a method that has been shown to be appropriate and statistically valid for efficient microarray experiments.31 Total RNA was extracted from liver tissue using TRIzol reagent (Invitrogen, UK) according to the manufacturer's instructions. RNA was further purified using an RNA clean up kit (Zymo Research). Total RNA from each offspring group was pooled (1 μg total RNA from each sample, n = 6 each group) and sent for whole genome gene expression analysis (Nimblegen, Iceland). Each pooled sample was hybridized to a 4-plex whole genome mouse array (44k oligonucleotides), and initial analysis (normalization) was performed. ArrayStar (DNASTAR) software was used to compare arbitrary expression values from each group exposed to an HF diet (C/HF, HF/C, and HF/HF) against those from the control group (C/C), generating a fold difference value for each gene. Expression changes were noted for genes with greater than 1.7-fold change relative to the C/C group.

Hepatic Gene Expression for Validation of Microarray Analysis.

Complementary DNA was synthesized from total RNA extracts. Specific primer and probe sequences were designed using primer express, or Universal ProbeLibrary online assay design center (Roche). Oligonucleotide sequences were synthesized by Eurogentec Ltd (Romsey, UK), and real-time PCR was performed to measure the messenger RNA levels of each gene (see Supporting Table 2 for primer and probe sequences). Each assay was performed in a single 96-well plate, with each sample measured in duplicate. A mean quantity value for each sample was then extrapolated from a standard curve generated with known concentration standards to gain measurement of expression levels. Levels of expression for each gene of interest were presented as percentage expression of appropriate housekeeping gene (beta-actin). Reaction conditions were 95°C for 15 minutes (hotstart), and 95°C for 15 seconds, 60°C for 30 seconds, and 72°C for 30 seconds.

Statistical Analysis.

The data were normally distributed and were expressed as mean ± standard error of the mean. Differences between the groups were determined by analysis of variance with Bonferroni post hoc tests, to correct for type 1 statistical error. Statistical significance was accepted when P < 0.05, and changes were considered as trends when 0.05 < P < 0.1. To determine the independent effects of maternal and postweaning diets on each parameter, the data were also analyzed using a two-way analysis of variance (Supporting Tables 3, 4).


Offspring of Dams Fed an HF Diet Have an Exaggerated Diet-Induced Metabolic Syndrome Phenotype.

Because NAFLD is thought to be the hepatic expression of metabolic syndrome,23 we attempted to generate a mouse model with a metabolic syndrome–like phenotype by HF exposure during development and postnatal life. Dams fed a HF diet were heavier and had higher total fat mass compared with dams fed a C diet (P < 0.001; see Supporting Fig. 2). From 7 weeks of age, offspring from these dams, which were also fed a HF diet postweaning (HF/HF) were heavier (P < 0.05) than those from all other groups (Fig. 1A). During this period, the calorie intake was not significantly different between offspring groups. At 15 weeks of age, the HF/HF offspring had greater total fat mass compared with all other groups (P < 0.001 versus C/C, Fig. 1B). Both the maternal (P < 0.01) and postweaning (P < 0.0001) diets have a significant effect on fat accumulation (Supporting Table 3). However, the combined effect of the maternal and postweaning diets (that is, HF/HF) has a greater effect (P < 0.0001) on fat accumulation. We also determined abdominal fat deposition as a marker of central obesity. Our observations (Fig. 1C) show the HF/HF exposure resulted in the biggest increase (P < 0.0001, Supporting Table 1) in abdominal fat accumulation. Although exposure to a HF diet postweaning (C/HF) had a significant effect on serum cholesterol concentrations (P < 0.05 versus C/C), once again this effect was exacerbated in the HF/HF animals (P < 0.01 versus C/C, Fig. 1D). Insulin-to-glucose (I/G) ratios (absolute values for which are provided in Supporting Fig. 3), which is a proxy measure of insulin sensitivity (Fig. 1E), showed a trend toward higher ratios and lower insulin sensitivity in the HF/HF group versus C/C.

Figure 1.

Phenotypic characterization of 15-week-old female offspring HF/HF (n = 8), HF/C (n = 7), C/HF (n = 8), C/C (n = 6). (A) Increase in body weight over time where *P < 0.05 versus C/C. (B) Total fat mass (gonadal + retroperitoneal + renal + inguinal + intrascapular fat mass divided by body weight), where *P < 0.05 versus C/C, ***P < 0.001 versus C/C, #P < 0.001 versus HF/C. (C) Abdominal (gonadal) fat mass, where *P < 0.05 versus C/C, ***P < 0.001 versus C/C, #P < 0.001 versus HF/C. (D) Total plasma cholesterol levels, where **P < 0.01 versus C/C and *P < 0.05 versus C/C. (E) Plasma insulin/glucose ratios as a proxy measure of glucose homeostasis (HF/HF P < 0.1 versus C/C). (F) Plasma β-hydroxybutyrate levels, where *P < 0.05 versus C/C, **P < 0.01 versus C/C, and ***P < 0.001 versus C/C.

Offspring of Dams Fed a HF Diet Have an Exaggerated Form of Diet-Induced Fatty Liver.

We assessed liver morphology in 15-week-old offspring livers. No lipid accumulation and a normal hepatic architecture were observed in livers from the C/C group (Fig. 2A). In livers from HF/C offspring, few lipid droplets were observed (Fig. 2B). C/HF offspring livers showed mild steatosis (Fig. 2C). However, in HF/HF livers, extensive fat accumulation was observed (Figs. 2D-F), with evidence of severe inflammation and a marked mixed lympho-monocytic perivenular infiltrate (Fig. 2E). The severity of the NAFLD in the offspring livers was assessed using the Kleiner scoring system,29 which allows scoring of individual features, including steatosis, ballooned hepatocytes, and inflammation. Whereas both the C/C and HF/C offspring received no scores, the C/HF offspring achieved a score of 2, indicative in human histopathology Kleiner scoring as mild NAFLD (Table 1). In contrast, the HF/HF liver generated a necro-inflammatory score of 6, which is compatible with a diagnosis of NASH (Table 1). These findings demonstrate that exposure to a postweaning HF diet causes hepatic steatosis and that this effect is markedly exaggerated when the offspring also have been exposed to an HF diet in the early developmental period leading to development of NASH in the offspring.

Figure 2.

Histological analysis (A-J), in 15-week-old (unless stated) offspring livers HF/HF (n = 8), HF/C (n = 7), C/HF (n = 8), C/C (n = 6). (A) Hematoxylin-eosin–stained C/C liver. (B) Hematoxylin-eosin–stained HF/C liver. (C) Hematoxylin-eosin–stained C/HF liver showing mild steatosis (arrow). (D) Hematoxylin-eosin–stained HF/HF liver showing steatosis (arrow). (E) Hematoxylin-eosin–stained HF/HF mouse showing a mixed perivenular lymphomonocytic infiltrate (arrow). (F). Oil Red O–stained HF/HF liver showing extensive fat accumulation (arrow). (G) Hematoxylin-eosin–stained 30-week C/C liver. (H) Hematoxylin-eosin–stained 30-week C/C liver showing mild steatosis (arrow). (I) Hematoxylin-eosin–stained 30-week C/HF liver showing steatosis (arrow). (J) Hematoxylin-eosin–stained 30-week HF/HF liver showing microvesicular (large arrow) and macrovesicular (small arrow) steatosis.

Table 1. Assessment of NAFLD Severity in Offspring's Liver Using the Kleiner Scoring System
 SteatosisBallooningLobular InflammationActivity ScoreIndication
  1. Kleiner scoring system described in Kleiner et al.29 For each group (n = 3), the mean score for each characteristic is given.

C/C 15 week0000Normal
C/HF 15 week1001NAFLD
HF/C 15 week0000Normal
HF/HF 15 week2125NASH
C/C 30 week0000Normal
C/HF 30 week2115NAFLD
HF/C 30 week2013NAFLD
HF/HF 30 week3227NASH

Maternal Fat Intake Contributes Toward NAFLD Progression in Adult Offspring.

We also determined the NAFLD severity in a subgroup of older (30-week-old) offspring to determine whether the liver condition progressed as the animal aged. The phenotype of these animals is very similar to that observed for the 15-week-old animals (Fig. 5), and corresponds to histological observations. Similarly to the earlier time point, both histology (Fig. 2G) and Kleiner scoring (Table 1) revealed normal hepatic architecture in the C/C group. In contrast, the HF/C group showed moderate levels of steatosis at 30 weeks (Fig. 2H) that had progressed from the 15-week point. At 30 weeks, both the C/HF and HF/HF offspring livers showed a NASH-like histology, the severity of which was worse in the HF/HF group. This was of a similar pattern to the earlier time point of 15 weeks. In addition, the 30-week HF/HF livers exhibited a distinct microanatomy with both microvesicular and macrovesicular steatosis within the distinct acinar zones (Fig. 2J).

Figure 5.

Phenotypical characterization of 30-week-old female offspring HF/HF (n = 5), HF/C (n = 7), C/HF (n = 6), C/C (n = 8). (A) Total body weight ***P < 0.001 versus C/C, and #P < 0.001 versus C/HF. (B) Total fat mass (gonadal + retroperitoneal + renal + inguinal + intrascapular fat mass divided by body weight), where **P < 0.01 versus C/C, ***P < 0.001 versus C/C, (C) Abdominal (gonadal) fat mass, where ***P < 0.001 versus C/C, #P < 0.05 versus C/HF. (D) Total plasma cholesterol levels, where *P < 0.05 versus C/C and ***P < 0.001 versus C/C. (E) Plasma insulin/glucose ratios as a proxy measure of glucose homeostasis, where *P < 0.05 versus C/C and #P < 0.05 versus HF/C. (F) Plasma β-hydroxybutyrate levels, where **P < 0.01 versus C/C.

Offspring of Dams Fed an HF Diet Exhibit Up-regulated CD44 Protein Expression.

To further explore the inflammatory status in these offspring, liver sections were stained for CD44. CD44 is a type I transmembrane protein, which acts as a cell surface receptor for hyaluronan (a component of the extracellular matrix overproduced during fibrosis) and is commonly expressed on hepatic Kupffer cells and infiltrating lymphocytes and therefore can be considered as an indirect indicator of both inflammation and early fibrosis/cirrhosis.32, 33 CD44 staining was undetectable in liver from C/C and C/HF offspring (Fig. 3A, B). However, sections from both the HF/C and HF/HF groups show staining around the portal vein regions (Fig. 3C, D). These data suggest that offspring from dams fed an HF diet may be predisposed to early onset of inflammation and fibrosis.

Figure 3.

CD44 Immunohistochemistry (solid bar = 40 μm) (A-D), and mitochondrial electron transport chain complex activity (E-G) in 15-week-old offspring livers HF/HF (n = 8), HF/C (n = 7), C/HF (n = 8), C/C (n = 6). (A) CD44 C/C–stained liver. (B) CD44-stained HF/C liver. (C) CD44-stained HF/C liver showing intense staining around intrahepatic bile duct (arrow). (D) CD44 HF/HF–stained liver showing intense staining around portal vein (arrow). (E) Complex I activity expressed relative to citrate synthase. *P < 0.05 versus C/C. #P < 0.05 versus C/HF. (F) Complex II and III activity expressed relative to citrate synthase. *P < 0.05 versus C/C. (G) Complex IV activity expressed relative to citrate synthase, where *P < 0.05 versus C/C and #P < 0.05 versus C/HF.

Offspring of Dams Fed an HF Diet Have Reduced Mitochondrial Electron Transport Chain Complex Activity.

Recent animal models have demonstrated the importance of defective hepatic mitochondria in the development of insulin resistance27 and hepatic fat accumulation.23, 25 Therefore, we determined mitochondrial function in offspring liver by measuring the activity of ETC enzyme complexes. Complex I activity was reduced in livers from HF/C and HF/HF offspring compared with the C/C group (P < 0.05) (Fig. 3E). This pattern was also observed for complexes II/III and IV activity, with significant reductions in the HF/HF and HF/C groups compared with the C/C offspring (P < 0.05). Because exposure to the HF diet during the developmental period is the common denominator for both groups presenting ETC impairment, these data suggest that complex activity is impaired in offspring from dams fed an HF diet. Our data show that maternal diet has a significant effect on ETC enzyme activity (P < 0.0001, Supporting Table 3), whereas both the postweaning diet and the combined diets have no significant effects. Plasma β-hydroxybutyrate was assessed as a proxy measure of mitochondrial flux of acetyl-CoA to keto-acid production (Fig. 1F). The highest concentrations of β-hydroxybutyrate are seen in the C/C group, whereas animals from all other groups have significantly lower levels of β-hydroxybutyrate, the lowest being the HF/HF group (P < 0.001).

Offspring of Dams Fed an HF Diet Exhibit “Primed” Expression of Genes Involved in Lipogenesis, Oxidative Stress, and Inflammation.

Because the development and disease progression of human NAFLD is thought to involve inflammation, oxidative stress, and disturbances in lipogenesis and β-oxidation, we measured the hepatic messenger RNA expression of key genes within these pathways primarily by microarray analysis. Our data (Table 2) indicate that both HF/C and HF/HF offspring have elevated levels of genes involved in oxidative stress (Nos3, Nos2, Gstm6, and Lcn2) versus C/C. These data suggest that offspring from dams exposed to an HF diet exhibit markers of oxidative stress. A similar pattern was observed in the expression of genes associated with inflammation: Crp, Mmd2, Tnfsf1, and Il-12b were all increased in offspring of dams exposed to an HF diet. Moreover, these genes were further up-regulated when offspring had also been exposed to an HF diet from weaning.

Table 2. Summary of Microarray-Generated Relative Expression of Genes Involved in Inflammation, Oxidative Stress, Lipogenesis, and Beta-Oxidation
GeneFull NameActivityHF/CC/HFHF/HF
  1. In each of the four offspring groups, 1 μg RNA from six animals per group was pooled and the mean mRNA level was determined. The fold difference in mRNA level represents differences between mean mRNA levels for each group versus the C/C group.

Nox2NADPH oxidase 2Oxidative stress+1.353+1.157−1.017
Nox3NADPH oxidase 3Oxidative stress−1.402−1.151−1.812
Nox4NADPH oxidase 4Oxidative stress−2.701−1.110−1.017
Nos3Endothelial nitric oxide synthaseOxidative stress+1.939+1.010+2.095
Nos2Inducible nitric oxide synthaseOxidative stress/Inflammation+3.210+1.585+1.813
Gstm6Glutathione S-transferase, mu 6Oxidative stress+2.678+2.678+2.028
Lcn2Lipocalin 2Fatty acid transport Oxidative stress+1.01+1.48+2.08
CrpC-reactive proteinInflammation+1.67+2.382+2.768
Mmd2Monocyte to macrophage differentiation factorInflammation+2.956+2.370+9.563
Tnfsf1Tumour necrosis factor ligand 1Inflammation+1.657+1.040+2.701
Il-12bInterleukin 12bInflammation+1.883+1.267+2.551
Acox1Acyl-coenzyme A oxidase 1, palmitoylFatty acid transport (peroxisomal)−1.072−1.023−1.026
Cpt-1Carnitine palmitoyl transferaseFatty acid transport+1.655+1.068+1.038
AcadsAcyl-coenzyme A dehydrogenaseβ-Oxidation (mitochondrial)−1.022−1.062−1.091
MtpaHydroxyacyl-coenzyme A dehydrogenase alpha subunitβ-Oxidation (mitochondrial)−1.025−1.102−1.128
MtpbAs above beta subunitβ-Oxidation (mitochondrial)−1.012−1.135−1.056
PgpPhosphoglycolate phosphataseCardiolipin+1.919+2.292+2.824
GpamGlycerol-3-phosphate acyltransferaseTAG synthesis ER/Cardiolipin+1.217+1.226+2.266
Agpat1-acylglycerol-3-phosphate O-acyltransferase 1TAG synthesis ER/Cardiolipin+1.025+1.448+1.819
Lpp2Phosphatidic acid phosphatase type 2cTAG synthesis ER+1.410+1.089+2.410
Dgat1Diacylglycerol O-acyltransferase 1TAG synthesis ER+1.244+1.715+1.715
AclATP citrate lyaseFatty acid synthesis+1.8389+1.005+2.395
AcacbAcetyl CoA carboxylaseFatty acid synthesis+1.544+1.063+3.000
FasFatty acid synthaseTAG synthesis+2.613+1.797+6.845
Srebp1cSterol regulatory element binding transcription factor 1FAS gene activation+1.031+1.103+2.681

Because hepatic lipid accumulation is central to the onset of human NAFLD, we investigated genes involved in lipid homeostasis (Table 2). We anticipated that increased supply of fatty acids from the diet would increase expression of genes required for mitochondrial and peroxisomal β-oxidation and fatty acid transport (Cpt-1 and Acox, respectively). However, we observed that Cpt-1 and Acox expression remain unchanged in all groups. Acads, Mtpa, and Mtpb, which are successive enzymes in the β-oxidation pathway, also showed no increase in gene expression. In contrast, genes involved in cardiolipin (Pgp), fatty acid (Acl, Acacb, Fas, Srebp1c), and triacylglycerol (TAG) synthesis (Gpam, Agpat, Lpp2, Dgat1) were up-regulated in the offspring of dams fed a HF diet, and this was further increased when these offspring were exposed to the HF diet postweaning.

Real-Time PCR Validation of “Primed” Lipogenesis Gene Expression.

To confirm these observations, we validated the expression of the rate-limiting enzymes in mitochondrial and peroxisomal β-oxidation (Cpt1 and Acox1, respectively) and TAG and fatty acid synthesis (Dgat and Fas, respectively) by reverse transcription PCR (Fig. 4). Our observations largely confirmed the gene expression profile generated by the microarray. For example, we demonstrate that, at a transcriptional level, the mitochondrial (and peroxisomal) β-oxidation enzymes remain unchanged in response to a postweaning HF diet, and, in contrast, the rate-limiting enzymes of the de novo lipogenesis (Fas) and TG synthesis (Dgat) pathways are up-regulated in the HF/HF offspring and contribute to the severe hepatic steatosis observed in these animals.

Figure 4.

Hepatic gene expression of rate-limiting lipogenesis and fat oxidation enzymes in 15-week-old offspring HF/HF (n = 8), HF/C (n = 7), C/HF (n = 8), C/C (n = 6) (mean ± standard error of the mean). **P < 0.01 versus C/C. ***P < 0.001 versus C/C. #P < 0.05 versus C/HF. (A) Aox1 (acyl-coenzyme A oxidase 1). (B) Cpt1 (carnitine palmitoyltransferase 1). (C) Dgat (diacylglycerol O-acyltransferase 1). (D) Fas (fatty acid synthase).

ETC Complex Activity and Gene Expression in the 30-Week-Old Offspring.

To see how these biochemical processes change over time, we measured both ETC enzyme activity (Fig. 6) and gene expression (Table 3 and Fig. 6) in the livers of the 30-week-old offspring subgroup. Some differences were observed between the 15-week and 30-week offspring subgroups; however, the pattern or trend for each group was maintained. For example, the ETC complex activities (Fig. 6A-C), are generally lower in the HF/C and HF/HF groups compared with the C/C and C/HF groups. However, this is only significant (P < 0.05 versus C/C) in the 30-week HF/C offspring for complexes II/III and IV. The apparent upturn in complex activity in the HF/HF group could be attributable to compensatory mechanisms that have been previously reported in middle-age and old-age mouse kidney and may allow the cell to transiently balance the decrease in enzymatic activity with age, or in this case severe liver pathological conditions.34 Interestingly, at 30 weeks, the C/HF animals show a significantly reduced complex IV activity, which may correspond to a progressing liver condition. The results of the microarray data also show a similar pattern of expression of genes involved in inflammation, oxidative stress, and lipid metabolism, as observed in the earlier time point (Table 3). Once again, we validated the rate-limiting genes in oxidation and lipogenesis by real-time PCR (Fig. 6D-F). The results show that although at this time the differences between the offspring groups were less marked for Cpt-1 and Fas, the HF/HF offspring group still had elevated Dgat1 gene expression (P < 0.05 versus C/C).

Figure 6.

Hepatic mitochondrial electron transport chain complex activity (A-C) and hepatic gene expression of rate-limiting lipogenesis (D-F) enzymes in 30-week-old offspring HF/HF (n = 5), HF/C (n = 7), C/HF (n = 6), C/C (n = 8) (mean ± standard error of the mean. (A) Complex I activity expressed relative to citrate synthase. (B) Complex II and III activity expressed relative to citrate synthase. *P < 0.05 versus C/C. (C) Complex IV activity expressed relative to citrate synthase, where *P < 0.05 versus C/C. (D) Cpt1 (carnitine palmitoyltransferase 1). (E) Dgat (diacylglycerol O-acyltransferase 1), where *P < 0.05 versus C/C. (F) Fas (fatty acid synthase).

Table 3. Summary of Microarray-Generated Relative Expression of Genes Involved in Inflammation, Oxidative Stress, Lipogenesis and Beta-Oxidation in 30 Week Female Offspring.
GeneFull nameActivity   
  1. In each of the 4 offspring groups 1 μg of RNA from 6 animals per group was pooled and the mean mRNA level determined. The fold difference in mRNA level represents differences between mean mRNA levels for each group versus the C/C group.

Nox2NADPH oxidase 2Oxidative stress+2.102+1.028+2.2
Nox3NADPH oxidase 3Oxidative stress+1.143+2.640−1.151
Nox4NADPH oxidase 4Oxidative stress+1.039+1.422+1.431
Nos3Endothelial nitric oxide synthaseOxidative stress+1.258+1.039+1.040
Nos2Inducible nitric oxide synthaseOxidative stress/Inflammation+1.003−1.15+1.075
Gstm6Glutathione S-transferase, mu 6Oxidative stress+1.345+1.036+1.244
Lcn2Lipocalin 2Fatty acid transport, Oxidative stress−1.318+2.091+6.833
CrpC-reactive proteinInflammation+1.106+1.43+1.842
Mmd2Monocyte to macrophage differentiation factorInflammation+1.126+3.340+5.171
Tnfsf1Tumour necrosis factor ligand 1Inflammation+1.128+1.468+1.568
Il-12bInterleukin 12bInflammation+1.128+1.245+1.707
Acox1Acyl-Coenzyme A oxidase 1, palmitoylFatty acid transport (peroxisomal)−1.01−1.278−1.325
Cpt-1Carnitine Palmitoyl TransferaseFatty acid transport−1.206−1.401−1.596
AcadsAcyl-Coenzyme A dehydrogenaseβ-oxidation (mitochondrial)−1.005−1.05−1.05
Mtpahydroxyacyl-Coenzyme A dehydrogenase alpha subunitβ-oxidation (mitochondrial)−1.203+1.442+1.363
MtpbAs above beta subunitβ-oxidation (mitochondrial)+1.186+1.047+1.164
PgpPhosphoglycolate phosphataseCardiolipin+1.027+1.092+1.279
GpamGlycerol-3-phosphate acyltransferaseTAG synthesis ER/Cardiolipin+1.383+1.361+1.287
Agpat1-acylglycerol-3-phosphate O-acyltransferase 1TAG synthesis ER/Cardiolipin−1.1+1.4+1.3
Lpp2Phosphatidic acid phosphatase type 2cTAG synthesis ER+1.154+1.048+1.314
Dgat1diacylglycerol O-acyltransferase 1TAG synthesis ER−1.05+1.618+2.982
AclATP citrate lyaseFatty acid synthesis+1.2+1.3+1.7
AcacbAcetyl CoA CarboxylaseFatty acid synthesis+1.440+1.246+1.757
FasFatty acid synthaseTAG synthesis+1.599+1.365+1.900
Srebp1cSterol regulatory element binding transcription factor 1FAS gene activation+1.084+2.365+3.035


An understanding of the mechanisms contributing to disease progression from simple hepatic steatosis to NASH is crucial to developing strategies to prevent chronic irreversible liver disease. We reasoned that exposure to a high-fat diet in the developmental period could be an important factor that influences the offspring's susceptibility to the development of NASH in later life. Our study demonstrates that 15-week-old offspring of dams fed an HF diet, which were also fed an HF diet after weaning (HF/HF), are predisposed to develop steatohepatitis rather than simple steatosis. This phenotype is similar to human NASH (Fig. 2D), a progressive liver condition that may result in cirrhosis and end-stage liver failure. In contrast, offspring that were only fed an HF diet after weaning exhibited mild steatosis without evidence of inflammation, a phenotype more akin to simple nonprogressive human fatty liver (Fig. 2C). Therefore, exposure to a HF diet during both developmental and postweaning periods is worse than HF exposure in the postweaning period alone. Our observations from older (30 weeks old) animals show that a maternal HF diet can lead to development of NAFLD later in life, even if a control diet has been given in the postweaning period (Fig. 2H), evidence that changes occurring during development are persistent and affect disease outcome in later life. Importantly, analysis of the 30-week HF/HF offsprings' livers show a progressive form of steatohepatitis (Fig. 2J), which is (again) more severe than their C/HF counterparts (Fig. 2I). This confirms our observations from the 15-week HF/HF offspring. Taken together, we have demonstrated that HF exposure in early development exacerbates the effect of an HF diet in later life, leading to progressive fatty liver disease and highlighting the critical nature of maternal nutrition.

It is widely accepted that a second “hit” in addition to baseline steatosis is required for NAFLD disease progression to occur.26, 35 To our knowledge, these findings are the first to show that the HF exposure received in early development mediates an additional pathogenic mechanism that promotes the liver disease progression beyond simple fatty liver and toward steatohepatitis (NASH). In response, we then investigated the mechanism by which a HF maternal diet may manifest this second hit and lead to increased susceptibility to NASH development in adult offspring. Mitochondria are central to lipid homeostasis and are maternally inherited; therefore, they provide a good candidate vector for the inheritance of developmentally primed NASH. In addition, alterations in mitochondrial form and function have been shown to constitute a central element of NAFLD pathophysiology.20, 22–26, 29, 36 In this study, we have shown that the activities of the mitochondrial ETC enzymes are reduced in the livers of offspring exposed to an HF diet in utero and during lactation. A decrease in ETC activity is likely to lead to increased generation of reactive oxygen species,37 which have previously been shown to initiate lipid peroxidation and trigger the release of inflammatory cytokines contributing to the development of NASH.25 Therefore, we hypothesize that this developmentally induced mitochondrial dysfunction may contribute to NASH development in adult offspring. In support of this hypothesis, we show that the expression of several genes involved in the oxidative stress and inflammatory pathways are elevated in offspring from dams fed an HF diet (Table 2). For example, we observed up-regulated Nos3, Nos2, and Gstm6 gene expression, known to be induced by increased reactive oxygen species,38, 39 in both the HF/C and HF/HF offspring (Table 2). Interestingly, Gstm6, an enzyme involved in the detoxification of oxidative stress products, is considered a candidate gene in the disease progression of human NASH.38 In the HF/HF group, we also observed an increase in Lcn2 expression, which has previously been implicated in oxidative stress40 and human NASH development38 (Tables 2, 3). We also showed that genes associated with inflammation (Crp, Mmd2, Tnfsf1, and Il-12b; Tables 2 and 3) are up-regulated in offspring of dams fed a HF diet.

These data concur with recent findings showing that diet-induced obesity leads to mitochondrial impairment.27 However, these data link diet in early development with impaired mitochondrial function and the development of fatty liver disease in adulthood. We have previously demonstrated that renal mitochondrial copy number is influenced after maternal HF exposure.17 Although hepatic mitochondrial copy number is not altered in this model (Supporting Fig. 2), ETC activity is persistently altered by a high-fat maternal diet, and these mitochondrial abnormalities correspond to histological abnormalities associated with NASH in the 15-week HF/HF group and also may be causal to the steatosis observed in the 30-week-old HF/C offspring. In simple invertebrate model systems, it has been established that very early stresses during the initial stages of development cause persistent changes to mitochondrial activity,41 thus establishing a precedent for “priming” of the mitochondria. Specifically, this study provides the first evidence in a mammalian system showing that exposure to a maternal HF diet can lead to developmental induction of impaired ETC enzymes, which persists into adulthood and contributes to the adult onset of NAFLD and the progression to NASH when the HF exposure is continued into adulthood.

The role of oxidative stress and inflammation in NASH development are well documented. However, does this mitochondrial dysfunction also play a role in increased fat deposition? We reason that exposure to an HF diet in the developmental environment has compromised the liver's metabolic capacity (through mitochondrial impairment in the first instance) to deal with the high levels of dietary fat it receives in the adult environment, resulting in increased steatosis and progression toward a NASH-like phenotype. We measured serum concentrations of β-hydroxybutyrate as an indicator of hepatic capacity for mitochondrial ketogenesis via β-oxidation (Fig. 1E) and therefore a proxy measure for mitochondrial metabolism as a whole. As expected, the levels of plasma β-hydroxybutyrate were lowest in the HF/HF offspring (threefold lower versus C/C). Gene expression of the rate-limiting enzyme involved in mitochondrial β-oxidation (Cpt1) was unchanged in the HF/HF group compared with the control offspring. This implies that mitochondrial metabolism is unresponsive to the increased demand for β-oxidation coming from the postweaning HF diet (Fig. 2D). Subsequently, the reduction in mitochondrial function occurring in the developmental period steers the surplus flux of intracellular acyl-CoA toward TG synthesis pathways, rather than β-oxidation (Fig. 7). In response, a developmental “priming” occurs. Genes involved in TAG (Gpam, Agpat, Lpp2, Dgat1) synthesis pathways are up-regulated to cope with these excess nutrients. These elevated expression profiles then persist into early adulthood. Therefore, offspring that have received a HF diet in development and are also exposed to a HF diet in adulthood (HF/HF offspring) shuttle their cytosolic fatty acids toward lipogenesis (Fig. 7), resulting in severe steatosis and a NASH-like phenotype. Conversely, this hypothesis also explains why the offspring exposed to a HF diet in adult life do not exhibit such a severe phenotype, because their mitochondrial function is maintained and their lipogenesis pathways are not developmentally primed. This study shows for the first time that early life exposure to HF feeding developmentally and biochemically “primes” metabolic pathways associated with NAFLD disease onset and progression in the offspring. Specifically, in the case of the offspring exposed to a HF diet in both developmental and adult environments, these “primed” pathways are induced to a greater effect, resulting in a florid NASH phenotype.

Figure 7.

Hypothetical schematic representation of developmentally programmed NASH. During the developmental period, the high-fat maternal diet causes decreased function of the ETC in the offspring, resulting in decreased capacity to metabolize macronutrients. The increased fat flux increases cytosolic acyl-CoAs, which are channeled toward TAG and phospholipid synthesis. The liver responds by up-regulating the expression of glycerol-3-phosphate acyl tranferase, 1-acylglycerol-3-phosphate O-acyltransferase 1, and phosphatidyl glycerophosphate synthase, and diacylglycerol O-acyltransferase. Therefore, the high fat exposure in the developmental environment may impair mitochondrial function and “primed” TAG synthesis pathways. After high fat exposure in the adult environment, there is increased flux of acyl CoAs to TAG synthesis in the affected offspring. In addition, as a result of increased insulin concentrations, there is an increase in acetyl CoA carboxylase and sterol regulatory element binding transcription factor 1 (srebp1c), resulting in further up-regulation of FAS and TAG formation.

At this point, we are unable to define a mechanism that leads to this developmental priming of both impaired ETC function and gene expression. However, it is plausible to speculate that suboptimal nutrition during the developmental period may alter the epigenetic profile of key metabolic genes, subsequently leading to persistent modulations in gene transcription, and increasing the risk of developing NASH in adulthood.42 Further studies are needed to investigate these priming mechanisms and to fully elucidate the timing of events on the causal pathway to NAFLD onset and progression to NASH.

In conclusion, our data provide evidence that exposure to a HF diet during early periods of development primes an increased susceptibility to hepatic steatosis and inflammation in adult offspring. Specifically, if previously “primed” offspring (with increased maternal fat intake) are subsequently fed a HF diet, this results in severe disease similar to progressive human NASH. These data emphasize the importance of a balanced diet during pregnancy and lactation and the consequences for incidence of chronic disease if an unhealthy diet is consumed across generations.


The authors thank Debbie Smith, Dyan Sellayah, and Frederick Anthony for technical and analytical support during the project.