Amino acid homeostasis is a target of metformin therapy

Abstract

Objective: Unexplained changes in regulation of branched chain amino acids (BCAA) during diabetes therapy with metformin have been known for years. Here we have investigated mechanisms underlying this effect.

Methods: We used cellular approaches, including single gene/protein measurements, as well as systems-level proteomics. Findings were then cross-validated with electronic health records and other data from human material.

Results: In cell studies, we observed diminished uptake/incorporation of amino acids following metformin treatment of liver cells and cardiac myocytes. Supplementation of media with amino acids attenuated known effects of the drug, including on glucose production, providing a possible explanation for discrepancies between effective doses in vivo and in vitro observed in most studies. Data-Independent Acquisition proteomics identified that SNAT2, which mediates tertiary control of BCAA uptake, was the most strongly suppressed amino acid transporter in liver cells following metformin treatment. Other transporters were affected to a lesser extent. In humans, metformin attenuated increased risk of left ventricular hypertrophy due to the AA allele of KLF15, which is an inducer of BCAA catabolism. In plasma from a double-blind placebo-controlled trial in nondiabetic heart failure (trial registration: NCT00473876), metformin caused selective accumulation of plasma BCAA and glutamine, consistent with the effects in cells.

Conclusions: Metformin restricts tertiary control of BCAA cellular uptake. We conclude that modulation of amino acid homeostasis contributes to therapeutic actions of the drug.

Keywords: AMPK; Branched chain amino acids; Glutamine; Metformin; Rapamycin; SNAT2; mTOR.

Figure 1

Opposing effects of metformin and leucine on cell signaling.(A) Primary hepatocytes were starved of amino acids (2 h) followed by refeeding with/without 1X MEM amino acids supplemented with 4 mM l-glutamine for 60 min as shown, with or without the drugs shown present throughout the experiment. Cells were lysed and the lysates prepared for SDS-PAGE and immunoblotting. Effects on AMPK and mTOR signaling were measured using the antibodies shown, to study phosphorylation of p70S6K, S6, AMPK and ACC. (B) Wild-type (WT) and AMPK double knockout (AMPK KO) MEFs were treated as already described previously for hepatocytes, except that the MEFs had amino acids removed with/without drug for 15 h. N = 3 for each experiment.

Figure 2

Metformin suppresses leucine uptake and leucine selectively reverses effects of metformin but not rapamycin on mTOR signaling.(A) Primary hepatocytes were treated as in Figure 1A., except that they were supplemented with increasing concentrations of leucine 400 μM, 1 mM or 4 mM, with/without 2 mM metformin or 100 nM rapamycin (‘RAPA’) throughout the whole experiment. (B-E) Primary hepatocytes were pretreated for 150 min with metformin (B), A-769662 (C), 2-Aminobicyclo [2,2,1]heptane-2-carboxylic acid (BCH, D) and saturable leucine uptake was measured as described in the Methods. (E) Saturable metformin uptake was determined as described in the Methods, in the presence or absence of 50 μM added leucine. ∗∗∗ denotes p < 0.001 and ∗∗ denotes p < 0.01 with respect to basal. Each experiment was performed at least 3 times. Error bars are SEM. ∗ denotes p < 0.05 significant change between the columns marked, ‘ns’ denotes ‘not significant.’ N = 3 for each experiment.

Figure 3

Metformin suppresses function and expression of the glutamine transporter SNAT2.(A) Effect of adding metformin treatment in addition to BCH treatment (150 min for both agents). (B) Leucine uptake was determined as in Figure 3, except acute (10 min) effects of metformin and BCH were compared. (C-E) Cells were starved of amino acids for 3 h in the presence or absence of metformin as shown and gene expression of LAT1 (C), CAT1 (D) and SNAT2 (E) was measured by RTPCR as described in the Methods. (F) Primary hepatocytes were treated as in Figure 1, except that effects of single amino acids leucine, isoleucine, valine and glutamine on mTOR signaling were analysed. (G) Primary hepatocytes were pretreated for 150 min with metformin, and saturable MeAIB uptake was measured as described in the Methods. (H) Schematic of metformin and KLF15 actions on BCAA homeostasis. Metformin suppresses BCAA uptake by reducing functional SNAT2, which selectively inhibits mTOR activation by BCAA, in contrast rapamycin inhibits mTOR activation from all stimuli. These effects of metformin may limit the supply of amino acids for gluconeogenesis, or the supply of energy for gluconeogenesis in the liver. KLF15 also suppresses BCAA signaling, by inducing BCAA catabolism. (I) Effect of amino acid supplementation on glucose production with/without glucagon and metformin. The inset shows how fold change due to metformin is attenuated by adding amino acids Between columns, ∗∗∗ denotes p < 0.001, ∗∗ denotes p < 0.01, ∗ denotes p < 0.05 significant change. Each experiment was performed at least 3 times. Error bars are SEM.

Figure 4

Metformin selectively suppresses SNAT2, LAT1 and other amino acid trans-porters.(A-E) Liver cells were treated with/without 250 μM metformin for 24 h and DIA proteomics was carried out as described in the Methods. Amino acid transporters changed significantly by metformin are shown. Between columns, ∗∗ denotes p < 0.01, ∗ denotes p < 0.05 significant change. Each experiment was performed at least 3 times. Error bars are SEM.

Figure 5

Effects of metformin in cardiac myocytes.(A) Primary cardiac ventricular myocytes were starved of amino acids (2 h) followed by refeeding with/without 1X MEM amino acids supplemented with 4 mM l-glutamine for 60 min as shown, with or without the drugs shown present throughout the experiment. Cells were lysed and the lysates prepared for SDS-PAGE and immunoblotting. Effects on AMPK and mTOR signaling were measured using the antibodies shown, to study phosphorylation of p70S6K, S6, AMPK and ACC. (B, C) Primary cardiac ventricular myocytes were starved of amino acids for 2 h with or without metformin and refed with AA for 1 h prior to puromycin treatment for 10 min. Angiotensin II was used as a hypertrophic stimulus overnight (C). (D) Association of KLF15 genotype with LVH stratified by use of metformin. The significant association of the KLF15 AA genotype with LVH was completely attenuated in metformin users.

Figure 6

Metformin causes plasma build-up of branched chain amino acids and glutamine in nondiabetic aged humans.(A-F) Plasma was obtained from non-diabetic aged humans before and after 4 months of treatment with metformin (n = 23, 2 g/day) or placebo (n = 15) as described previous-ly [37]. Plasma amino acid levels determined as described in the Methods are presented for total ami-no acids (A), leucine (B), isoleucine (C), valine (D), glutamine (E) and tyrosine (F). Other amino ac-ids are presented in supplementary material. ∗∗∗ denotes p < 0.001, ∗∗ denotes p < 0.01, ∗ denotes p < 0.05 between first and second sample.