A. BACKGROUND Accumulation of fat in the liver due to non-alcoholic causes (NAFLD) is associated with hepatic insulin resistance, which impairs the ability of insulin to inhibit the production of glucose and VLDL . This leads to increases in serum glucose, insulin and triglyceride concentrations as well as hyperinsulinemia. Recent epidemiologic studies have shown that a major reason for the metabolic syndrome as well as the accompanying increased risk of cardiovascular disease and type 2 diabetes is overconsumption of simple sugars. The investigators have recently shown that overeating simple sugars (1000 extra calories/day, "CANDY" diet) increases liver fat content by 30% within 3 weeks (4), and recapitulates features of the metabolic syndrome such as hypertriglyceridemia and a low HDL cholesterol concentration.

The fatty acids in intrahepatocellular triglycerides may originate from peripheral lipolysis, de novo lipogenesis, uptake of chylomicron remnants by the liver and from hepatic uptake of fatty acids released during intravascular hydrolysis of triglyceride-rich lipoproteins (the spillover pathway). A classic study using stable isotope methodology by the group of Elisabeth Parks showed that in subjects with NAFLD, the excess intrahepatocellular triglycerides originate from peripheral lipolysis and de novo lipogenesis.

It is well-established that ingestion of a high carbohydrate as compared to high fat diet stimulates de novo lipogenesis in humans. Meta-analyses comparing isocaloric high fat and high carbohydrate diets have shown that high carbohydrate but not high fat diets increase increase serum triglycerides and lower HDL cholesterol. Since hypertriglyceridemia results from overproduction of VLDL from the liver, these data suggest the composition of the diet influences hepatic lipid metabolism. Whether this is because overfeeding fat leads to preferential deposition of fat in adipose tissue while high carbohydrate diets induce a relative greater increase in liver fat is unknown. There are no previous studies comparing effects of chronic overfeeding of fat as compared to carbohydrate on liver fat or and the sources of intrahepatic fatty acids.

A common polymorphism in PNPLA3 at rs738409 (adiponutrin) gene is associated with a markedly increase liver fat content. This finding has been replicated in at least 20 studies across the world. The investigators have shown that PNPLA3 is regulated by the carbohydrate response element binding protein 1. Mice overexpressing the human I148M PNPLA3 variant in the liver exhibit an increase in liver triglycerides and cholesteryl esters on a high sucrose but not high fat diet. These data suggest that overfeeding a high carbohydrate as compared to a high fat diet may increase liver fat more in subjects carrying the I148M allele than in non-carriers.

B. HYPOTHESIS The investigators hypothesize that overfeeding a high fat as compared to an isocaloric high carbohydrate diet influences the source of intrahepatocellular triglycerides. During a high fat diet, relatively more of intrahepatocellular triglycerides originate from peripheral lipolysis and less from DNL than during a high carbohydrate diet in the face of a similar increase in liver fat. It is also possible given the lack of previous overfeeding data comparing 2 different overfeeding diets that the high fat diet induces a smaller increase in liver fat than a high carbohydrate diet even in the face of an identical increase in caloric intake because a greater fraction of ingested fat is channeled to adipose tissue than the liver. The investigators also hypothesize that liver fat may increase more in carriers than non-carriers of the I148M variant in PNPLA3 during a high carbohydrate than a high fat diet.

C. SPECIFIC AIMS The investigators wish to randomize, using the method of minimization (considers baseline age, BMI, gender, liver fat, PNPLA3 genotype) 40 non-diabetic subjects with NAFLD as determined by the non-invasive score developed in our laboratory or previous knowledge of liver fat content based on MRS to overeat either a high carbohydrate or high fat diet (1000 extra calories per day) for 3 weeks. Before and after the overfeeding diets, will measure liver fat content by 1H-MRS and the rate of adipose tissue lipolysis using doubly labeled water (DDW) and [1,1,2,3,3-2H5] glycerol as described in detail below. The investigators also wish to characterize glucose, insulin, fatty acid and triacylglyceride profiles before and while on the experimental diet. An adipose tissue biopsy is taken to determine whether expression of genes involved in lipogenesis or lipolysis, or those involved in adipose tissue inflammation change in response to overfeeding, and for measurement of LPL activity. After overfeeding, both groups will undergo weight loss to restore normal weight as described in our recent study. The metabolic study is repeated after weight loss.

Inclusion Criteria:

• BMI 27-35 kg/m2
• Age 18-65 yrs

Exclusion Criteria:

• type 1 or 2 diabetes
• renal insufficiency
• pre-existing liver or significant other disease other than NAFLD (i.e. autoimmune, viral or drug-induced liver disease)
• excessive use of alcohol (over 20g/day)
• pregnancy or lactation