Metformin targets mitochondrial complex I to lower blood glucose levels

Abstract

Metformin is among the most prescribed antidiabetic drugs, but the primary molecular mechanism by which metformin lowers blood glucose levels is unknown. Previous studies have proposed numerous mechanisms by which acute metformin lowers blood glucose, including the inhibition of mitochondrial complex I of the electron transport chain (ETC). Here, we used transgenic mice that globally express the Saccharomyces cerevisiae internal alternative NADH dehydrogenase (NDI1) protein to determine whether the glucose-lowering effect of acute oral administration of metformin requires inhibition of mitochondrial complex I of the ETC in vivo. NDI1 is a yeast NADH dehydrogenase enzyme that complements the loss of mammalian mitochondrial complex I electron transport function and is insensitive to pharmacologic mitochondrial complex I inhibitors including metformin. We demonstrate that NDI1 expression attenuates metformin's ability to lower blood glucose levels under standard chow and high-fat diet conditions. Our results indicate that acute oral administration of metformin targets mitochondrial complex I to lower blood glucose.

Fig. 1.. NDI1 mice display normal glucose homeostasis.

(A) Schematic representation of the mammalian mitochondrial electron transport chain containing the yeast NDI1 protein. Created with BioRender.com. (B) NDI1 mRNA expression in the livers of control and NDI1 mice. n = 13 control mice and n = 8 NDI1 mice (q = 3.7 × 10⁻³⁹, Wald test). (C) MA plot of mRNA expression changes between control and NDI1 mice. n = 13 control mice and n = 8 NDI1 mice. Genes colored red are significantly up-regulated in NDI1 mice and genes colored blue are significantly down-regulated in NDI1 mice (q < 0.05, Wald test). (D and E) Multiple cohorts of control and NDI1 mice, 8 to 12 weeks of age and on a standard chow diet, were fasted overnight for 16 to 18 hours. Mice were then split into two groups, fasted and refed. Blood was taken either immediately (fasted) or 4 hours after standard chow diet food was provided (refed). Blood glucose (D) and plasma insulin (E) levels were assayed from these blood samples. n = 8 control fasted, n = 9 control refed, n = 6 NDI1 fasted, and n = 6 NDI1 refed in both (D) and (E). Results represent mean ± SEM. Two-way ANOVA (Bonferroni) shows no significance (n.s.) between genotypes within the fasted and refed groups. ^n.s.P > 0.9999 [(D); fasted], ^n.s.P = 0.5822 [(D); refed], ^n.s.P > 0.9999 [(E); fasted and refed]. (F) Glucose tolerance (2 g/kg, oral gavage) was assessed on multiple cohorts of 8- to 12-week-old standard chow diet–fed control and NDI1 mice fasted overnight for 16 to 18 hours. n = 17 control mice and n = 12 NDI1 mice. Results represent mean ± SEM. (G) Incremental area under the curve (iAUC) for the GTT in (F). Results represent mean + SEM. Statistical significance was determined using a two-tailed, unpaired t test and an α level of 0.05. n.s. means nonsignificant, ^n.s.P = 0.6853.

Fig. 2.. NDI1 expression attenuates the blood glucose–lowering effects of metformin in mice fed a standard chow diet.

(A and B) Metformin levels were measured in the livers (A) and intestines (B) of 12- to 16-week-old standard chow diet–fed control and NDI1 mice fasted overnight for 16 to 18 hours and then administered metformin (200 mg/kg, oral gavage) or vehicle (water, oral gavage) 1 hour before tissue harvest. n = 5 control vehicle, n = 4 control metformin, n = 5 NDI1 vehicle, and n = 5 NDI1 metformin. Results represent mean ± SEM. Two-way ANOVA (Bonferroni) shows no significance (n.s.) between genotypes when treated with metformin, ^n.s.P > 0.9999 [(A) and (B)]. (C) Multiple cohorts of control and NDI1 mice, 8 to 12 weeks old and on a standard chow diet, were fasted overnight for 16 to 18 hours and then administered metformin (200 mg/kg, oral gavage) or vehicle (water, oral gavage) 30 min before a glucose (2 g/kg, oral gavage) tolerance test. n = 57 control vehicle, n = 64 control metformin, n = 26 NDI1 vehicle, and n = 79 NDI1 metformin. Results represent mean ± SEM. (D) Incremental area under the curve (iAUC) for the GTT in the presence or absence of metformin shown in (C). Results represent mean + SEM. Two-way ANOVA (Bonferroni) shows no significance (n.s.) between vehicle-treated control and vehicle-treated NDI1 mice, ^n.s.P = 0.4522, but shows statistical significance between metformin-treated control and metformin-treated NDI1 mice, *P < 0.0001.

Fig. 3.. NDI1 expression attenuates the blood glucose–lowering effects of metformin in mice fed a high-fat diet.

(A) Weight gain of control and NDI1 mice fed a 60 kcal% HFD ad libitum over an 8-week period. Mice were switched from a standard chow diet to an HFD at 10 to 12 weeks of age. n = 61 control mice and n = 57 NDI1 mice. Results represent mean ± SEM. (B and C) Multiple cohorts of control and NDI1 mice, on an HFD for 8 to 12 weeks, were fasted overnight for 16 to 18 hours. Mice were then split into two groups, fasted and refed. Blood was taken either immediately (fasted) or 4 hours after HFD food was provided (refed). Blood glucose (B) and plasma insulin (C) levels were assayed from these blood samples. n = 23 control fasted, n = 17 control refed, n = 17 NDI1 fasted, and n = 16 NDI1 refed in both (B) and (C). Results represent mean ± SEM. Two-way ANOVA (Bonferroni) shows no significance (n.s.) between genotypes within the fasted and refed groups. ^n.s.P > 0.9999 [(B); fasted and refed], ^n.s.P > 0.9999 [(C); fasted], ^n.s.P = 0.0946 [(C); refed]. (D) Multiple cohorts of control and NDI1 mice, on an HFD for 8 to 10 weeks, were fasted overnight for 16 to 18 hours and then administered metformin (200 mg/kg, oral gavage) or vehicle (water, oral gavage) 30 min before a glucose (2 g/kg, oral gavage) tolerance test. n = 31 control vehicle, n = 30 control metformin, n = 22 NDI1 vehicle, and n = 35 NDI1 metformin. Results represent mean ± SEM. (E) Incremental area under the curve (iAUC) for the GTT in the presence or absence of metformin shown in (D). Results represent mean + SEM. Two-way ANOVA (Bonferroni) shows no significance (n.s.) between vehicle-treated control and vehicle-treated NDI1 mice, ^n.s.P = 0.1808, but statistical significance between metformin-treated control and metformin-treated NDI1 mice, *P < 0.0001.