To the editor

The article by Lin et al.1 in the September 2000 issue of Nature Medicine proposed that metformin treatment reverses fatty liver disease (FLD) in obese, leptin-deficient (ob/ob) mice by decreasing hepatic TNF-α mRNA expression. In that study, metformin reversed hepatomegaly, lowered elevated serum transaminase levels, and decreased TNF-α, uncoupling protein 2 (UCP2) mRNA and fatty acid synthase protein in the liver in ob/ob mice. Because earlier studies have shown that TNF-α is elevated in ob/ob mice, stimulates hepatic fatty acid synthesis and induces UCP2 expression in the liver, the authors postulated that TNF-α and TNF-inducible factors that promote hepatic lipid accumulation contribute to FLD in obese mice.

While the hypothesis that increased hepatic TNF-α expression caused FLD is attractive, it is also conceivable that the increase in hepatic TNF-α expression is a consequence of lipid accumulation rather than the cause. One approach to obtain definitive evidence for or against the role of TNF-α as a mechanism of FLD is to use knockout mice. Previous studies have shown that ob/ob mice that lack both TNF receptors (ob/ob–p55/p75-null) and therefore lack functional TNF-α activity, have improved insulin sensitivity2, but no change in body weight or fat mass per lean body mass3. We have reported that ob/ob–p55/p75-null mice have twice as much UCP2 expression in the liver and adipose tissue compared with ob/ob mice4 suggesting that endogenous TNF-α has an inhibitory effect on UCP2 and therefore a lack of TNF-α further upregulates UCP2 expression in obese mice. Since these data were in contrast to the proposed hypothesis of Lin et al.1 that TNF-α may be the cause of FLD and subsequent susceptibility to liver toxicity in ob/ob mice, we have measured hepatic lipid content and serum transaminase levels in control, ob/ob, p55/p75-null and ob/ob–p55/p75-null mice.

We show that hepatic lipid content is similar in ob/ob and ob/ob–p55/p75-null mice (Table 1). Moreover, serum transaminase levels are also not significantly different between ob/ob mice with or without TNF receptors. Thus, these data demonstrate that lack of TNF-α action neither reverses hepatic lipid accumulation nor lowers elevated serum transaminase levels in obese mice, indicating that TNF-α is not the mechanism of FLD in obese diabetic mice. Our data indicate that FLD occurs in ob/ob mice, even in the absence of TNF action and TNF-induced insulin resistance.

Table 1 Total hepatic lipid content and serum transaminase levels in ob/ob and ob/ob–p55/p75-null mice.

See “Reply to 'TNF-α is not the cause of fatty liver disease in obese diabetic mice'“ by Anna Mae Diehl.