The mechanism of action of MET (MET) remains controversial. MET induces an iron starvation response in yeast, so we hypothesized that tissue iron might influence MET responsiveness in higher organisms. We reviewed electronic health records to correlate MET responsiveness in diabetic humans with serum ferritin, a measure of iron stores. Both higher and subnormal serum ferritin were associated with decreased HbA1c response to MET (N=80, p<0.001) , and MET responsiveness was lost entirely at higher than normal ferritin. Mice were placed on diabetogenic diets with different iron contents, and treated or not with MET. MET responses of fasting glucose values were superior in mice on low-normal and normal iron diets compared to those on high iron: Fasting glucose decreased 60.4 mg/dL on low-normal iron (N=9, p<0.001) , 40.5 mg/dL on normal iron (N=23, p<0.01) and 2.6 mg/dL on high-normal iron (N=27, p=NS) , with significant interaction between iron and MET (p=0.0015, ANOVA) . Other effects of iron (e.g. leptin expression) are known to be mediated by protein modification by O-linked N-acetylglucosamine (O-GlcNAc) . MET increased O-GlcNAc levels 200% in mice on normal iron (p=0.04) but not high-normal iron diets (-8%, p=NS) . In mice with elevated O-GlcNAc resulting from heterozygous deletion of the O-GlcNAcase gene, MET’s ability to decrease fasting glucose levels in mice on high iron was restored, with differences between MET and placebo being 30.3, 61.3, and 56.5 mg/dL on low-normal, normal, and high-normal iron diets (p<0.0001 by ANOVA) ; in contrast to wild type mice, the significant effects of iron and its interaction with MET were lost (p=0.13 and 0.47) . We conclude that tissue iron affects glycemic responses to MET, supporting the hypothesis that MET induces an iron starvation response but that high iron overwhelms that signal. The results on protein O-GlcNAc modification suggest that it mediates the interaction of iron and MET. MET responsiveness in patients with diabetes could be optimized by modulating iron stores.


D. A. Mcclain: None. C. O. Usoh: None. F. Lorenzo: None. A. V. Harrison: None.


VA Research Service 2I01 BX001140; NIH 1P30DK12472

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at