The molecular targets of metformin have been expanded upon in the recent decades (Foretz et al., 2019; Lu et al., 2021; Wu et al., 2023) but the primary mechanism of this biguanide remains unclear. The activation of adenosine 5’-monophosphate activated protein kinase (AMPK) is considered a core mechanism of metformin; AMPK phosphorylates several key enzymes that regulate the activity of transcription factors, coactivators, and corepressors, thus regulating lipid metabolism, inflammation, autophagy, and gluconeogenesis (Wang et al., 2019a; Yang et al., 2019; Zhang et al., 2022a; Li et al., 2022; Yang et al., 2022) (Figure 1). For example, AMPK activation suppresses the expression of a key lipogenic transcription factor, SREBP-1 in the liver, and activates AMPK phosphorylated acetyl-CoA carboxylase (ACC), which is an inhibitory phenotype and suppresses malonyl-CoA biosynthesis, thus inhibiting fatty acid synthesis and increasing fatty acid oxidation (Zhou et al., 2001). Furthermore, activated AMPK inhibits the activity of the mammalian target of rapamycin complex (mTORC) to induce cellular autophagy (Wang et al., 2021a) and decrease protein synthesis (Knudsen et al., 2020). The inhibition of fatty acids and protein synthesis can promote cellular glucose uptake and metabolism to lower glucose levels (Lundsgaard et al., 2020). On the other hand, metformin-induced activation of AMPK has been shown to phosphorylate CREB-regulated transcription coactivator 2 (CRTC2, also known as TORC2), which prevents its translocation to the nucleus, thus inhibiting the transcriptional activation of gluconeogenic genes (Koo et al., 2005). Moreover, metformin inhibits Th1 and Th17 cells differentiation while promoting Tregs development through activation of AMPK (Duan et al., 2019), which may contribute to its anti-inflammatory effects.

However, many studies showed that the beneficial effects of metformin remained in a transgenic mouse model with lacked AMPK or its upstream activator LKB1 (Foretz et al., 2010), indicating that there are other critical targets other than AMPK for metformin action. In fact, increasing evidence points to other actions of metformin, for example, metformin has been shown to inhibit mitochondrial complex Ⅰ, which is independent of AMPK activity in cells (El-Mir et al., 2000; Owen et al., 2000); the inhibition results in a decrease in ATP production, thus suppressing gluconeogenesis in the liver. Additionally, metformin at low doses (<130 µM in plasma) was reported to inhibit mitochondrial glycerol-3-phosphate dehydrogenase (mGPDH) activity, resulting in an increased hepatic cytosolic NADH to NAD⁺ ratio (lactate to pyruvate ratio) independently of changes in intracellular levels of ATP and inhibition of glucose production from reduced gluconeogenic substrates (lactate and glycerol) (Madiraju et al., 2014; Madiraju et al., 2018). In addition to these, a study found that glucose tolerance improved in germ-free mice that received metformin-altered microbiota via fecal samples transfer, indicating that the altered gut microbiota mediates some of the antidiabetic effects of metformin (Wu et al., 2017) (Figure 1).

Inhibiting hepatic gluconeogenesis is the main mechanism of the blood glucose lowering action of metformin. However, the mechanisms underlying these effects are unclear. Studies have showed that the overexpression of the peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α) preserved metformin’s ability to reduce glucose output regardless of AMPK expression (Foretz et al., 2010), suggesting that PGC-1α is a key factor for metformin’s gluconeogenic inhibitory effects. Concomitantly, another study found that cyclic adenosine monophosphate (cAMP), a gluconeogenic stimulus, induces changes in the expression of several lncRNAs in primary mouse hepatocytes. Furthermore, metformin treatment preserved 189 of the 456 upregulated lncRNAs and 167 of the 409 downregulated lncRNAs. Among those changed lncRNAs, NR_027710 and ENSMUST00000138573 were identified as having an association with the gene expression of PGC-1α (Wang et al., 2018), suggesting that metformin inhibiting gluconeogenesis could be mediated by the regulation of NR_027710 and ENSMUST00000138573. Consistent with these results, another study showed that 3,076 lncRNAs (1,823 upregulated and 1,253 downregulated) are altered in HFD-induced insulin resistant mice compared with the normal control group, and metformin treatment reversed 588 lncRNAs. Among those lncRNAs downregulated by metformin, NONMMUT031874.2 was highly expressed and found to target miR-7054-5p. Downregulating the expression of NONMMUT031874.2 by metformin could increase the expression of miR-7054-5p and then downregulate the expression of Socs3, which inhibits PEPCK to inhibit gluconeogenesis (Zhang et al., 2022b).

In addition, hepatic gluconeogenesis is catalyzed by multiple enzymes, including G6Pase, PEPCK, and FBP1 (Mues et al., 2009). Suppressing hepatic gluconeogenesis is usually accompanied by the activation of glycolysis enzymes, including GCK, PFKL, and PK (Figure 2) (Al-Oanzi et al., 2017). Metformin treatment decreased the expression of lncRNA-NONMMUT034936.2 in liver tissue from HFD mice and exhibited similar expression patterns with G6P target genes, indicating that metformin may affect gluconeogenesis by regulating NONMMUT034936.2 (Shu et al., 2021). Furthermore, transcription factor hepatocyte nuclear factor 4α (HNF-4α) binding to promoters of G6Pase or GCK is required for GCK gene activation and insulin-mediated G6Pase gene repression (Hirota et al., 2008). Metformin-exposed mouse fetuses were shown to elevate H19 expression (Deng et al., 2017), a lncRNA reported to be related to T2DM and influence DNA methylation (Ding et al., 2012), which induces hypomethylation and the overexpression of HNF-4α, thus efficiently repressing hepatic gluconeogenesis (Figure 2).

On the other hand, the increased glucose uptake effects by metformin are mediated by lncRNAs. LncRNA-Dreh is reduced by metformin in C2C12 myotubes. Furthermore, knockdown of Dreh resulted in reduced glucose concentrations in the culture medium and increased levels of GLUT4 protein in the plasma membrane. The overexpression of Dreh attenuated the glucose-lowering effect of metformin in myotubes (Takahashi et al., 2019), suggesting the blood glucose lowering actions of metformin are mediated by Dreh in the skeletal muscle cells. Moreover, metformin treatment is reported to increase GLUT4 expression by inducing expression of MEG3 and SNHG20, thus improving insulin resistance in the endometrium of PCOS rats (Liu et al., 2022b), indicating the metformin regulation of lncRNAs in glucose uptake is not specific to a type of lncRNA.

Dyslipidemia is common among patients with T2DM (prevalence > 75%) (Athyros et al., 2018). Aberrations in lipid metabolism contribute to the progression of T2DM and the risk of diabetic complications. For example, excessive dietary lipid intake and enhanced lipogenesis can cause ectopic lipids to accumulate in adipose, pancreatic, vascular, and cardiac cells (Yang et al., 2018b), resulting in atherosclerosis and cardiomyopathy due to insulin resistance (Wolf et al., 2014). Metformin has been shown to inhibit lipogenesis by suppressing sterol regulatory element binding protein 1 (SREBP-1) (Madsen et al., 2015), which is a transcriptional factor for lipogenesis. In particular, the inhibition of SREBP-1 by metformin is reported to be mediated by lncRNA; a study found that lncRNA-Gm15622 was upregulated in the liver of obese mice fed a HFD and in liver cells treated with saturated fatty acid. Metformin suppresses lncRNA-Gm15622 both in vivo and in vitro, thus decreasing the expression of SREBP-1c and inhibiting lipid accumulation by sponge for SREBP-1 targeted microRNA (Ma et al., 2020). Additionally, MEG3 and SNHG20 have been shown to be downregulated in HFD-induced fatty liver mice, and metformin improves lipid accumulation by reversing the expressions of MEG3 and SNHG20 (Guo et al., 2018). As stated above, MEG3 and SNHG20 were shown to regulate glucose uptake, indicating that these two lncRNAs may mediate the cross-talk of lipid and glucose metabolism via metformin.

Chronic inflammation is found in adipose tissues, pancreatic islets, and the liver (Lontchi-Yimagou et al., 2013), which contributes to insulin resistance and progression to T2DM and the long-term complications of diabetes (Berbudi et al., 2020; Gasmi et al., 2021). The nucleotide-binding oligomerization domain leucine-rich repeat and pyrin domain-containing protein 3 (NLRP3) inflammasome has emerged as a key mediator of pathological inflammation in diabetes (Esser et al., 2014). Firstly, inflammatory stimuli (such as TNFα, SFA, or LPS) induce NF-κB expression and result in the expression of pro-IL-1β and pro-IL-18; secondly, NLRP3 and GSDMD mediated the maturation and secretion of IL-1β and IL-18 after activation. Subsequently, cells secrete a large number of pro-inflammatory cytokines (e.g., IL-1β and IL-18) and induce a type of cell death called pyroptosis (Ding et al., 2019). The specific regulatory mechanisms of NLRP3 inflammasome activation remain unclear, and lncRNAs were shown to modulate NLRP3 activation in diabetes. For example, growth arrest-specific 5 (GAS5) overexpression suppressed NLRP3 inflammasome activation-mediated pyroptosis by regulating the miR-34b-3p/AHR axis in diabetic cardiomyocytes (Xu et al., 2020a). Although there is no direct evidence showing that metformin could regulate NLRP3 by regulating GAS5, clinical trials showed that taking metformin for 2 months was shown to increase GAS5 in patients with T2D (Parvar et al., 2023), indicating its anti-inflammatory role in the diabetic heart. Furthermore, metformin significantly alleviated NLRP3 and GSDMD expressions in diabetic periodontitis mice with a severe inflammation state, and in this study, lncRNA_1810058I24Rik, lncRNA_Gm12474, and lncRNA_Gm41514 were found to be co-expressed with NLRP3 and GSDMD (Zhou et al., 2020), which may potentially be regulated by metformin. However, a study conducted on the mouse glomerular membrane epithelial cell line SV40-MES-13 showed that metformin treatment efficiently inhibits elevation of H19 via high glucose in cells, with decreased levels of TNF-a, IL-6, and TGFβ-1, and overexpressed H19 abolishes the effects of metformin in SV40-MES-13 cells (Xu et al., 2020b), suggesting that the anti-inflammation effects of metformin are related to the regulation of lncRNAs.

Mitochondrial dysfunction is a key pathological factor for diabetes. Reduced oxidative phosphorylation and excessive ROS production have been found in patients with obesity and type 2 diabetes (Wada and Nakatsuka, 2016; Luc et al., 2019). In addition, mitochondrial dynamic changes, such as mitochondrial fusion and fission, have been considered to be closely related to insulin sensitivity (Watanabe et al., 2014; Rovira-Llopis et al., 2017). For example, mice with knockdown of liver-specific mitofusin 2 (Mfn-2, a key protein for mitochondrial fusion) exhibit hepatic insulin resistance and glucose intolerant (Sebastian et al., 2012). For a decade, mitochondria have been considered a classic target for the blood glucose lowering actions of metformin. A most recent study identified the possible binding sites for biguanides on mitochondrial complex I protein subunits (Bridges et al., 2023). Additionally, mGPDH and complex IV have also been considered targets of metformin in animal experiments (Madiraju et al., 2014; LaMoia et al., 2022), and more targets have been found in the research into the actions of metformin in mitochondria (Pecinova et al., 2019; MacDonald et al., 2021). LncRNAs regulated by metformin potentially affect mitochondrial function; clinical research on DNA methylation in plasma from metformin-treated patients with T2DM showed that 57 lncRNA-associated DNA methylation regions were found, which included the mitochondrial ATP synthase-coupling factor 6 (ATP5J) (Solomon et al., 2020), suggesting that metformin may regulate ATP5J through lncRNAs. In addition, metformin increases GAS5 expression in HepG2 cells and the plasma of patients with T2DM (Parvar et al., 2023), and elevated GAS5 decreases mitochondrial tricarboxylic acid flux by disrupting the metabolic enzyme tandem of the tricarboxylic acid cycle (Sang et al., 2021), providing another explanation for the mitochondrial inhibition effects of metformin. Moreover, metformin decreases the expression of H19, while H19 has been shown to downregulate the expression of Mfn-2, a key protein for mitochondrial fusion, in type 1 diabetic renal tissues (Sultan et al., 2021), and a decline in Mfn2 was observed in both type 1 and type 2 diabetic rodents (Gao et al., 2012; Veeranki et al., 2016), suggesting that metformin may regulate Mfn2 by decreasing H19.