Ad libitum consumption of the HF diet for 10 wk. produced obesity characterized by increased body weight, % body fat, and plasma leptin levels as compared with the LF control group, as well as hepatic steatosis. Continued ad libitum consumption of the HF diet for another 8 weeks in Phase 2 did not change any of these measures of obesity. Eight weeks of energy restriction of obese HF2 rats on VLC, LC or HC diets produced similar reductions in body weights and plasma leptin levels.
In rats on the LC and HC, but not VLC, diets visceral fat, % body fat, and intramuscular TG were significantly reduced from that at wk. 10 (end of Phase 1), in agreement with our previous findings using VLC and HC diets [18]. In contrast, hepatic lipid levels decreased from HF1 values in the VLC and HC but not the LC group; this finding suggests that high dietary fat, coupled with an adequate amount of carbohydrate, promoted hepatic fat storage even during weight loss. These results show that the fat and carbohydrate compositions of the diets consumed in hypocaloric amounts during Phase 2 had differential effects on adipose tissue and hepatic fat loss during energy restriction.
Evidence for whole body insulin resistance in the VLC and HF 2 rats was observed during Phase 2 of the study. Higher HOMA-IR values in unanesthetized VLC and HF2 rats were due to elevated feed-deprived plasma insulin levels vs. those of HC rats. In addition, insulin levels were higher in VLC than HC rats at 30 min after the ip glucose load, despite similar plasma glucose AUC values. These results suggest that insulin release may have increased as an adaptation to tissue insulin resistance in VLC rats during Phase 2. In our previous study, the elevated plasma glucose levels after an ip glucose load in VLC vs. HC rats, but similar insulin levels, indicated insulin resistance [18]. The higher levels of plasma triglyceride in LF and HC vs. VLC groups are consistent with results reported in humans and rodents on high carbohydrate diets [34,35,36], and have been associated with hepatic lipogenesis [34].
The present study was designed to permit interpretation of differences in hepatic gene expression among diet groups to represent chronic adjustments, and not acute responses to the nutrient content of the current meal. To this end, all rats were feed-deprived for > 16 h before sampling, the same glucose stimulus was used for all groups in a given Phase, and each diet group’s gene expression responses to the oral glucose load were compared with that in the feed-deprived state.
The effects of diet composition on hepatic insulin resistance were evaluated by examining responses of insulin-regulated genes, many of which have roles in hepatic lipogenesis (Srebf1, Insig2, Acaca, Fasn, Scd1, and Gck). Although the lipogenic pathway (including de novo synthesis of fatty acids, DNL) is stimulated by insulin, there is disagreement concerning whether regulation of DNL is impaired by hepatic insulin resistance. In hepatic insulin resistance, insulin-regulated pathways that involve glucose metabolism are dysregulated (e.g., insulin fails to inhibit gluconeogenesis), while insulin-regulated triglyceride synthesis appears still to be appropriately increased [37]; this apparent paradox has been termed “selective insulin resistance” [38]. However, work using genetic modifications of the insulin-signaling pathway in mice [35, 39] has shown that hepatic DNL, including expression of Srebf1, is regulated by insulin; hepatic insulin resistance, as defined by defects in insulin signaling, therefore would prevent insulin from increasing Srebf1 expression. Furthermore, the hepatic lipid storage and export observed in hepatic insulin resistance may not involve stimulation of DNL by insulin, but may instead reflect packaging of fatty acids coming to the liver from insulin-resistant adipose tissue [40]. In light of these considerations, we utilized insulin’s effect on expression of genes in lipogenic pathways to assess hepatic insulin resistance.
Ad libitum intake of the HF diet suppressed the effects of an oral glucose load, used to raise plasma insulin level, on expression of some of the genes studied (Srebf1, Fasn, Scd1, Gck, Pck1, and Cpt1a). This result may be due, in part, to the high plasma levels of leptin in both HF1 and HF2 rats; leptin has been reported to decrease expression of genes involved in the lipogenic pathway, including Srebf1 [41] and Scd1 [42]. In contrast, only the LF, HF1 and HF2 groups appropriately decreased Insig2 mRNA after the glucose load. Given that Insig2 plays a role in controlling diurnal patterns of metabolism, the disturbance of ad libitum feeding in the VLC, LC and HC, but not LF, HF1 or HF2 groups by feed restriction may have affected regulation of Insig2 transcription [43].
In Phase 1, even though the HF1 group failed to increase Srebf1 mRNA in response to the glucose load, expression of Acaca, a target of SREBP1c, did increase in both LF and HF1rats. It is possible that SREBP1c may have mediated this response in both groups, since the reduction in Insig2 mRNA after the glucose load in LF, HF1, and HF2 groups may have led to a decrease in Insig2 protein, thereby increasing conversion of SREBP1c to its mature form. It is also possible that expression of Acaca was stimulated by other regulators, such as ChREBP, which is activated by a metabolite of glucose [44]. However, there were no detectable changes in expression of the ChREBP gene (Mlxipl) between feed-deprived and 3 h post-glucose states in either LF or HF1 rats.
The observed reduction in lipogenic gene response to the glucose load in the HF1 and HF2 groups is consistent with the decrease in DNL seen in mice on a high-fat diet [45]. In contrast, increased expression of lipogenic genes has been reported in rats [46] and mice [35, 47, 48] consuming high-fat, or high-fat, sucrose-containing diets. However, those findings pertain to gene expression in the feed-deprived state [46, 47] or in a single fed condition, and not a comparison of pre- vs. post-glucose gene expression as in the present study, and so do not assess stimulation of expression by glucose or insulin. Although high cholesterol intakes (4% of diet weight) have been associated with insulin resistance [49] such an effect would not be probable for rats on the HF1, HF2, VLC or LC diets, which contained less than 0.5% cholesterol. Furthermore, mice on 1% cholesterol high-fat diets showed increased expression of SREBP1c [50], not the decrease that we observed.
Weight loss and reduction in plasma leptin levels, regardless of diet, were associated with improved effects of the glucose load on Fasn, Gck, and Pck1 expression in VLC, LC and HC rats (Fig. 4), as compared with the HF1 rats before weight reduction (Fig. 3). However, like the HF2 group, VLC rats did not show significant changes in expression of Srebf1, Scd, or Cpt1a in response to the glucose load at the end of Phase 2. This result supports hepatic insulin resistance in VLC rats, despite reduction in hepatic lipid levels and body weights from HF1 levels. The stimulation of Srebf1 expression by the glucose load in rats on LC or HC diets, which differed in fat and cholesterol levels, suggests that their higher carbohydrate intake, vs. that of VLC rats, had improved their hepatic insulin sensitivity. Regulation of Srebf1 transcription involves not only insulin, but includes the liver X receptor (LXR), mechanistic target of rapamycin complex (mTorc), and positive feedback from the mature form of SREBP1c [51]. Since dietary protein levels in VLC, LC and HC diets were similar, differences in amino acid intake seem unlikely to account for lower mTorc activation by insulin in the VLC group; similarly high fat intakes of VLC and LC rats also make it unlikely that those groups differed in their intake of dietary regulators of LXR.
Failure of a carbohydrate stimulus to increase Srebf1 mRNA by rodents consuming high-fat diets has been attributed to their polyunsaturated fatty acid (PUFA) intake [51,52,53] . This possibility is unlikely to explain the differences between VLC and LC or HC response for the following reasons: 1) PUFA (92% n-6) intakes of VLC (~ 1.57 g/d) and LC (~ 1.25) groups were similar; 2) in other studies, PUFA were consumed near the time of liver sampling, [52,53,54,55], whereas rats in our study were feed-deprived for at least 16 h before dissection, precluding an acute effect of dietary PUFA; and 3) if chronically higher PUFA intake were responsible for reducing levels of Srebf1 mRNA, then lower feed-deprived expression in HF1, HF2, VLC or LC vs. HC groups would be expected, but this was not observed.
Only the HC group showed a significant stimulation of Scd1 expression in Phase 2; overall expression of Scd1 was also higher in LF vs. HF1 rats in Phase 1. High carbohydrate diets (e.g., LF and HC) stimulate Scd1 transcription via SREBP1c and ChREBP [56]. Normal and diabetic mice consuming a diet similar to the LC (18%C, 37%F) showed both reduced insulin sensitivity and lower hepatic expression of Scd1 than mice consuming a diet similar to the HC or LF diets (63% C, 22%F) [57]; the small difference in their fat intake suggests that the difference in carbohydrate intake was responsible for the effect on Scd1 expression.
No difference in feed-deprived levels of Cpt1a mRNA was observed between any high-fat group (HF2, VLC or LC) and the HC group, although Cpt1a mRNA has been reported to be higher in rats on high-fat vs. low-fat diets in the feed-deprived state [58]. However, the finding that only the HC group, but not the HF2, VLC or LC groups, significantly decreased Cpt1a mRNA after the glucose load is consistent with the lack of difference in Cpt1a expression reported between the fed and feed-deprived state in rats consuming high-fat diets [59]; this lack of suppression of Cpt1a expression should promote continued use of fat as a fuel, despite increased availability of glucose.
In Phase 3, the lower responsiveness of hepatic gene expression to the glucose load by the VC3 rats, vs. that of VLC, LC or HC rats at the end of Phase 2, may have been associated with weight loss during the first 3 days after the switch to the HC diet. The stabilization of body weight during days 4–7 on the HC diet by VC7 rats may have promoted better responses. However, VC7 rats still differed from HC rats in their lack of increase in Scd1 mRNA in response to glucose, and mRNA levels of many genes remained below that exhibited by either VLC or HC rats. These results indicate that full transition from the effects of the VLC to those of the HC diet did not occur within 1 week.
The HC diet promoted greater responses of the insulin-regulated genes, Srebf, Scd1 and Cpt1a, to the glucose load than the VLC diet. As discussed above, the higher carbohydrate intake of the HC vs. the VLC or LC groups is likely to have accounted for the greater Scd1 gene response, and the lower fat intake of the HC vs. the LC or VLC groups may have accounted for the greater Cpt1a gene response the glucose load in HC rats. Although the VLC and LC groups had similar fat intakes, only the LC group increased expression of Srebf in response to glucose; the severely restricted carbohydrate intake of VLC rats may account for their lack of response.. However, since it is not possible to vary only fat or only carbohydrate intake, while holding protein and energy intake essentially constant, it remains possible that differences in both fat and carbohydrate intake are responsible for the observed dietary effects.
VLC, but not LC rats had elevated HOMA-IR and higher insulin levels after the ip glucose load as compared with HC rats, indicating whole body insulin resistance in VLC rats; since LC and VLC groups had similar fat intakes, this finding suggests a mitigating effect of the higher carbohydrate intake of the LC rats. However, the combined effects of high dietary fat and stimulation of expression of lipogenic genes (perhaps including targets of SREBP1c that are involved in esterification) in the LC group may have promoted hepatic TG storage [60]. Nonetheless, hepatic steatosis was not associated with whole body insulin resistance in weight-reduced rats in the present study. In addition, loss of the response of hepatic gene expression to a glucose load was not associated with hepatic steatosis.; Although hepatic lipid levels were similar in HF1, HF2 and LC rats, LC rats showed greater response to the glucose load than HF1 or HF2 rats, or than the VLC group which had lower hepatic lipid levels.
In HC rats, the insulin-stimulated increases in hepatic SREBP1c and its targets would be expected to promote conversion of glucose to fatty acids and triglyceride; decreases in hepatic Cpt1a would be expected to suppress fat oxidation in favor of glucose oxidation. Both processes should support the ability of a given concentration of insulin to lower plasma glucose levels after a glucose load; this is consistent with lower whole body insulin resistance in HC rats. Lack of this hepatic response in VLC rats would be expected to diminish the glucose-lowering effect of insulin; this is consistent with the higher insulin levels observed for the same level of glucose in VLC vs. HC rats.