F0 dams gained approximately 40 g more weight at term compared to F1 and F2 dams (generation* gestational age F(184,50) 3.7, P<0.0001) (). However, there was no difference in food intake during pregnancy between generations () which suggests that the increase in dietary energy provision altered nutrient partitioning. Pregnant mammals produce an exaggerated metabolic response to the additional stress of fasting, characterised by a faster induction and greater level of gluconeogenic and ketogenic activity than in non-pregnant females 
. We therefore used the metabolic challenge of fasting to assess the metabolic phenotype of the dams on day 8 pregnancy in each generation. Plasma glucose concentration was lower (generation F(1856,5.3) 25.6, P<0.0001) and β-hydroxybutyrate (βHB) (generation F(1934712,164232) 11.8, P<0.0001) concentration was higher in F2 dams than F0 dams during this test suggesting that there was a transition towards greater glucose utilisation and increased glucose sparing by ketogenesis in F2 dams (). These altered responses to fasting were associated with changes in mRNA expression of hepatic genes involved in gluconeogenesis and ketogenesis (). GR (generation F(3.9,0.2) 16.5, P<0.0001), PPARα (F(5.9,0.1) 43.8, P<0.0001), carnitine palmitoy1 transferase (CPT)-1 (F(15.1,0.2) 65.1, P<0.0001), glucose-6-phosphatase (G6Pase) (F(12.1,0.1) 97.4, P<0.0001) and phosphoenolpyruvate carboxykinase (PEPCK) (F(1.3,0.3) 4.3, P
0.001) mRNA expression was increased in F2 compared to F0 and F1 dams. These observations suggest that altered regulation of specific genes underlies phenotypic change between generations.
Change in maternal body weight from conception and energy intake during pregnancy and lactation.
Maternal blood metabolite concentrations and mRNA expression of genes involved in hepatic gluconeogenesis and ketogenesis.
Weight gain post-partum (post-partum age* generation F(852,427) 2.0, P<0.0001) and food intake (post-partum age* generation F(87,44) 2.0, P
0.009) was greater in F1 and F2 dams than F0 (). There was no significant difference between groups in length of gestation, litter size or litter weight during suckling (data not shown). These observations show that transition between two levels of dietary energy induced changes in the phenotype of pregnant and lactating dams across generations.
The effect of increased energy intake on the phenotype of the adult offspring was assessed by comparison with the offspring of dams fed a lower energy chow diet and which were themselves fed chow from weaning (CF group). Weight gain (F(8274,1426) 5.8, P
0.006) on postnatal day 70 was significantly greater in, but did not differ between, F1, F2 and F3 offspring of dams fed the higher energy diet than CF offspring (). Energy intake did not differ between groups (). Plasma glucose concentration during fasting was higher in F1 and F2 offspring, but was not significantly different from CF offspring in the F3 generation (F(24,10) 2.4, P<0.0001), while βHB concentration was higher in F1, F2 and F3 compared to CF offspring (F(2944,2220) 1.3, P
0.027) (). PEPCK mRNA expression was higher (F(102,3.1) 33.2, P
0.001) and G6Pase lower (F(166,47) 3.5, P
0.001) in all three generations compared to CF offspring ( E to I). GR expression was lower in F1 and F2 offspring compared to CF and F3 offspring (P<0.0001). There were no significant differences between generations in PPARα or CPT-1 mRNA expression. Thus the shift in energy intake induced adjustments in the phenotype of the adult offspring which, at least in part, was reflected in altered gene expression. However, such effects were more modest than observed in pregnancy which suggests the effects of the developmental environment on the female offspring were cryptic unless challenged by the metabolic demands of pregnancy.
Offspring phenotype and mRNA expression of genes involved in hepatic gluconeogenesis and ketogenesis.
Because PEPCK is rate limiting in gluconeogenesis and hence is critical to fasting glucose metabolism, the mechanism underlying the change in gene expression between generations was investigated by measuring the methylation of nine individual CpGs in the PEPCK promoter (). Compared to CF offspring, CpGs -606, -440 were hypomethylated and CpGs -248 and -218 were hypermethylated in all three generations (F(661,11) 63.7; F(1424,18) 78; F(212,7) 32.4; F(44,2) 25, respectively, all P<0.001) (). CpGs −508, −100, and −90 were hypomethylated in F1 only (F(313,21) 15; F(1151,18) 62.5; F976,8) 9.7; F(30,7) 4.4, respectively, all P<0.05) (). CpG -129 was hypermethylated in F1, but was hypomethylated in F2 offspring (F(458,28) 16.3, P
0.0007) (). CpG -81 was only hypomethylated in F3 offspring (F(257,8) 32.1, P
0.0004) (). The methylation status of CpGs -508 (r
0.02), -129 (r
0.001) and CpG -100 (r
0.002) was significantly associated with PEPCK mRNA expression. These findings show that persistent change in dietary energy induces adjustment of the level of methylation of specific CpGs over three generations, providing a mechanism by which the effects of the developmental environment induce changes in the offspring phenotype, even though these may not become apparent until a further challenge such as pregnancy.
Structure of the phosphoenolpyruvate carboxykinase and DNA methyltransferase 3a promoters.
Methylation of individual CpGs in the PEPCK promoter in the liver of the adult offspring.
The mRNA expression of Dnmt1, 3a and 3b was measured in the liver of both adult non-pregnant and pregnant offspring (). Dnmt1 expression did not differ significantly between groups. However, expression of Dnmt3a was decreased in F1 non-pregnant offspring, but increased in F2 and F3 offspring (F(726,43) 40.1, P<0.0001) compared to CF offspring. Dnmt3b mRNA expression was increased compared to CF offspring in all three generations (F(562,46) 12.2,P
0.0001). These findings indicate that overall capacity to induce DNA methylation de novo
differed between generations and thus suggests a mechanism for altered epigenetic regulation. Because Dnmt3a showed marked variation in expression between generations, we investigated the mechanism underlying changes in Dnmt3a expression we measured the methylation status of four CpGs in the Dnmt3a2 promoter which accounts of approximately 50% of the Dnmt3a expression in adult liver and is the predominant isoform in developing tissues 
. The methylation of CpGs -207 and -190 was not altered by generation or pregnancy (). However, the methylation of CpGs -56 and -39 was increased in F1 non-pregnant offspring, but decreased in F2 and F3 offspring (). Both of these CpGs were hypermethylated in F1 and F2 pregnant offspring compared to their non-pregnant siblings. These observations suggest that altered epigenetic regulation of Dnmts is involved in phenotypic variation between generations.
Hepatic DNA methyltransferase expression and Dnmt3a2 promoter methylation, and embryo heat shock protein 90 expression.
Finally, because HSP90 has been implicated in regulating developmental plasticity via a mechanism involving epigenetic change 
, we measured the mRNA expression of HSP90β in day 8 gastrulating embryos. HSP90 mRNA expression differed significantly between generations (F(770,13) 59.2, P
<0.0001). There was a non-significant trend (P<0.1) towards lower HSP90 expression in F1 than CF embryos. HSP90 expression was significantly lower in F2 and F3 embryos than in CF and F1 embryos () which suggests less stringent regulation of development and thus increased plasticity