The underlying mechanisms behind the observed effects of metformin treatment on GLUT4-mediated glucose transport can be divided into changes in insulin signaling pathway mediators, AMPK activation, epigenetic modifications, and enhancements in GLUT4 trafficking and translocation to the plasma membrane. The specific mechanisms are explained in detail in the following sections, and the schematic presentation can be seen in Figure 1.
Figure 1. The potential mechanisms of metformin’s effect on increased GLUT4-mediated glucose transport include changes in insulin signaling pathway mediators, AMPK activation, epigenetic modifications, and enhancements in GLUT4 trafficking and translocation to the plasma membrane. Legend: INRS—insulin receptor; IRS—insulin receptor substrate; PI3K—phosphatidylinositol 3-kinase; Akt—protein kinase B; AS160—Akt substrate of 160 kDa; SHIP2—inositol-5-phosphatase 2 with Src homologous domain 2; PIP2—phosphatidylinositol-3,4-diphosphate; PIP3—phosphatidylinositol-3,4,5-triphosphate; AMPD—AMP deaminase; AMPK—adenosine 5′-monophosphate-activated protein kinase; MEF2A—myocyte enhancer 2A; lncRNAs—long noncoding RNAs; miRNAs—microRNAs; TBC1D1—TBC domain family, member 1; CAP—Cbl-associated protein.
3.1. Insulin Signaling Pathway
The insulin signaling pathway needs to be explored when analyzing the effects of metformin treatment on glucose uptake. The binding of the insulin molecule to the α subunit of the insulin receptor (INSR) causes its dimerization and the activation of the intrinsic kinase activity of the β subunits, leading to their autophosphorylation. Activated INSR acts on adaptor proteins from the insulin receptor substrate family (IRS1, IRS2, and IRS3) in a further cascade. Of particular importance are IRS1 and IRS2, which bind to the p85 regulatory subunit of phosphatidylinositol 3-kinase (PI3K) and activate it through phosphorylation. Lipid products of PI3K catalyze the formation of phosphatidylinositol-3,4,5-triphosphate (PIP3). PIP3 then binds to phosphatidylinositol-3,4,5-phosphate kinase 1 (PDK1) and protein kinase B (Akt) and attracts Akt to the cell membrane, where PDK1 phosphorylates Akt at Thr
308. Additional phosphorylation at Ser
473/474 is required for the complete activation of Akt. Activated Akt, through its mediator Akt substrate of 160 kDa (AS160), represents a crucial step in the regulation of the kinase cascade involved in the translocation of GLUT4 from intracellular transport vesicles to the plasma membrane
[23,31,32,62][23][31][32][62]. By studying the effects of their inhibition and overexpression, a lot of evidence has been obtained suggesting that PI3K and Akt activation are essential for insulin-stimulated GLUT4 translocation
[63,64][63][64]. The initial stages of the signaling pathway, especially signaling via IRS, PI3K, and Akt, have been described as crucial for the development of IR, and decreased expression of PI3K, Akt, and GLUT4 has been reported in different tissues of IR patients
[32,55,65,66,67,68,69][32][55][65][66][67][68][69].
One of the first studies of metformin’s effect on insulin signaling mediators demonstrated that chronic insulin treatment reduced INRS tyrosine phosphorylation, PI3K activity, and Akt activity by 60–70% in cultured rat adipocytes, and these effects were prevented by the inclusion of metformin in the culture medium
[49]. A later study in C2C12 cells reached similar conclusions and suggested that chronic insulin treatment drastically reduced insulin-stimulated tyrosine phosphorylation of INRS and IRS1 as well as PI3K activity, and treatment with metformin was able to reverse these changes
[70]. Rice et al. demonstrated that incubation with metformin increased IRS1 mRNA and protein expression in human granulosa cell culture, as well as IRS2 protein expression
[71]. In this study, as well as in a similar study by Sonntag et al., metformin incubation contributed to the additional activation of Akt in the presence of insulin via PI3K activation
[71,72][71][72]. Furthermore, in ovarian tissue of rats exposed to a hyperandrogenic environment in fetal development with consequent ovulation disorders, markedly lower levels of INSR, IRS1, IRS2, and GLUT4 mRNA and proteins were found, and incubation with metformin increased mRNA and protein expression and the activation of most of these mediators
[73]. Another study using the rat PCOS model demonstrated a significant reduction in IRS2 and PI3K expression, and 4-week in vivo metformin treatment significantly increased their expression
[74]. Furthermore, in an additional study, reduced PI3K and Akt expression observed in ovarian and hepatic tissues of the rat PCOS model increased significantly after metformin treatment
[75]. In a study by Ferreira et al., both metformin and insulin increased PI3K and GLUT4 expression in endometrial cell culture, and their combined effect was even more pronounced
[67]. In contrast to insulin, metformin reduced the rate of Akt phosphorylation at Ser
473 in this study
[67].
However, some studies reported opposing results. For example, in a study by Ma et al., 24 h incubation of human granulosa cell culture with metformin resulted in reduced IRS1 mRNA and protein expression
[76]. Moreover, in a study by Ciaraldi et al., unchanged levels of IRS1, GLUT1, GLUT4, p85, and Akt proteins were observed in adipocytes of patients with T2DM after 3-4 months of metformin treatment
[55]. Similarly, insulin signaling parameters were unchanged after 26 weeks of metformin treatment in newly diagnosed T2DM patients
[56]. Furthermore, Kim et al. studied the effect of 3-4 months of metformin treatment in T2DM subjects who failed glyburide treatment and noted no effect on basal or insulin-stimulated IRS1-associated PI3K or Akt activity in muscle cells. Additionally, protein expression of IRS1, the p85 subunit of PI3K, and Akt was unaltered after treatment
[77].
Additional perspective can be gained from an in vitro study of the effects of metformin on human podocytes and a mouse skeletal muscle cell line, which demonstrated an essential role for inositol-5-phosphatase 2 with Src homologous domain 2 (SHIP2)
[78]. SHIP2 hydrolyzes PIP3 back to phosphatidylinositol-3,4-diphosphate (PIP2), which is the opposite of the action of PI3K in the insulin signaling pathway. SHIP2 is overexpressed in IR tissues, and in this study, direct binding of metformin to SHIP2 and decreased activity of the phosphatase domain of SHIP2 were found
[78]. At the same time, no effect of metformin was observed on the level of Akt phosphorylation
[78].
3.2. AMPK Activation
AMPK is a serine/threonine-specific protein kinase, acting as a highly conserved master regulator of metabolism
[79]. It is regulated by multiple upstream signals and affects numerous downstream substrates
[80]. It exists as a trimeric complex consisting of a catalytic subunit (α-subunit) and two regulatory subunits (β- and γ-subunits)
[80]. The N-terminus of the α-subunit comprises the kinase domain, and the phosphorylation of a conserved threonine (referred to as Thr
172) in the kinase domain is required for the full activation of AMPK
[80]. An increase in AMPK activity was associated with quickly increased glucose uptake and translocation of GLUT4 to the plasma membrane
[52,63,81,82,83,84,85,86,87,88][52][63][81][82][83][84][85][86][87][88]. Zhou et al. were one of the first groups to propose AMPK phosphorylation and activation as a critical pathway for metformin’s pleiotropic effects
[63]. The hypothesis of the metformin–AMPK–GLUT4 pathway was established; however, the exact mediators remain the subject of debate. Although metformin treatment has been coupled to AMPK activation and upregulation of glucose uptake separately, only a few studies have measured both AMPK activation and glucose uptake
[63,89,90,91,92,93][63][89][90][91][92][93].
One of the most frequently invoked mechanisms of metformin action is the inhibition of mitochondrial complex I and the consequent reduction in the [ATP]:[ADP] and [ATP]:[AMP] ratios
[80,82][80][82]. Due to the transient changes in cellular energy status, ATP is replaced with either ADP or AMP, allosterically activating AMPK
[82,94][82][94]. Metformin has also been shown to activate AMPK through phosphorylation at Thr
172 of the α-subunit independently of changes in adenine nucleotides
[82,95,96,97][82][95][96][97]. In addition, LKB1 was identified as an upstream kinase responsible for phosphorylating and activating AMPK and was implicated as a major target of metformin
[82,98,99][82][98][99]. However, metformin probably does not directly activate either LKB1 or AMPK since it did not influence the phosphorylation of AMPK by LKB1 in a cell-free assay
[63,100][63][100]. Metformin may also interact directly with the γ-subunit of the AMPK complex, producing a structural change that promotes activation
[101]. In a more novel approach, inhibition of AMP deaminase (AMPD) has been proposed as alternative mechanism to AMPK activation
[102], and one report suggested that metformin may control AMP levels via effects on AMPD
[103].
A study on mouse soleus muscle showed that chronic, but not acute, in vivo metformin treatment enhanced insulin-stimulated glucose uptake through activation of AMPK without notable changes in protein expression of insulin signaling mediators or GLUT4
[89]. An in vitro study in human adipocytes obtained from surgical biopsies showed that metformin incubation for 24 h increased glucose uptake, GLUT4 mRNA expression, and cellular protein level; however, suppression of metformin-induced AMPK activity by AMPKα1 silencing reduced the observed effects
[52]. An approximately twofold increase in AMPK activity, independent of the AMPK level or glycemia, was also shown in adipose tissue of patients with T2DM treated with metformin for 10 weeks compared with gliclazide therapy
[104]. The amount of GLUT4 protein was unaltered, and metformin also did not significantly alter the levels of insulin signaling pathway mediators
[104]. These results were additionally confirmed in 3T3-L1 adipocytes
[104]. However, at the time of recruitment, 12 out of 20 participants were already taking metformin, and the study had a six-week run-in period when medication was discontinued. The duration of this run-in period is important in light of a recent study that concluded that six months after metformin withdrawal, GLUT4 mRNA expression in subcutaneous adipose tissue of PCOS patients remained stable
[105]. Another study in patients with T2DM showed that metformin treatment for 10 weeks significantly increased AMPK activity in skeletal muscle, and the observed effect was associated with phosphorylation of AMPK at Thr
172 and resulted in enhanced peripheral glucose uptake
[92]. Furthermore, in the endometria of PCOS patients, metformin increased GLUT4 mRNA and protein levels
[58]. It significantly increased the phosphorylation and, therefore, the activity of both AMPKα and myocyte enhancer 2A (MEF2A), a transcription factor with a known binding site on the promoter region of the GLUT4 gene
[58,106][58][106].
Some studies have also demonstrated the importance of metformin’s action through AMPK activation and its downstream effect on GLUT4 translocation. Activated AMPK can phosphorylate TBC1D1 (TBC domain family, member 1), which controls the translocation and plasma membrane levels of GLUT4
[80,107][80][107]. Lee et al. showed the critical role of AMPK in GLUT4 translocation in the C2C12 skeletal muscle cell line
[108]. Two years later, Lee and colleagues published another study demonstrating the importance of AMPK in GLUT4 translocation in 3T3-L1 preadipocyte cells, and this effect was attenuated by AMPK knockdown
[109].
3.3. Epigenetic Modifications
In an effort to better understand the etiology of IR conditions, the focus has now shifted from genome-wide and candidate gene association studies to the epigenetic mechanisms that could explain the dynamic and well-observed interplay between known genetic and environmental factors
[110,111,112][110][111][112]. Therefore, a hypothesis that at least some of metformin’s effects are mediated through different epigenetic modifications was proposed, and the evidence supporting it has grown in recent years.
Treatment with metformin might influence the activity of numerous epigenetic modifying enzymes, mainly through the promotion of phosphorylation and hence activation of AMPK
[113,114][113][114]. Activated AMPK can phosphorylate multiple epigenetic enzymes, such as histone acetyltransferases, class II histone deacetylases, and DNA methyltransferases
[113]. For example, a study by McGee et al. reported that AMPK activation decreased the transcriptional repressor histone deacetylase 5 (HDAC5), with a known association with the GLUT4 gene, resulting in increased GLUT4 expression in human myotubes
[115]. In addition, metformin has also been reported to decrease the expression of multiple histone methyltransferases and to increase the activity of the class III histone deacetylase SIRT1
[113]. The importance of such mechanisms can be seen in a study in which 3T3-L1 adipocytes with knocked down SIRT1 showed inhibited insulin-stimulated glucose uptake and impaired GLUT4 translocation
[116].
Furthermore, due to the consistent findings of the importance of posttranscriptional modifications of GLUT4 mRNA and the previously established association of altered microRNA (miRNA) levels with diabetes, IR, and inflammation, the effect of metformin on various miRNAs has also been studied in recent years
[117,118][117][118]. miRNAs are small noncoding regulatory RNAs that act as negative posttranscriptional regulators of gene expression
[119,120][119][120]. Alterations of miRNA expression by metformin may be partially explained by an increase in DICER, one of the critical enzymes in miRNA processing
[113]. Increases in DICER protein levels have been reported in metformin-treated diabetic humans and mice and also in cells treated with the direct AMPK activator 5-aminoimidazole-4-carboxamide riboside (AICAR), suggesting that AMPK activation is the primary mechanism
[113,121,122,123][113][121][122][123].
Numerous miRNAs have been reported to directly or indirectly regulate insulin sensitivity and GLUT4 expression, and their levels were shown to be altered in IR conditions
[101,119,124,125,126,127,128,129][101][119][124][125][126][127][128][129]. Treatment with metformin has been associated with changes in several miRNAs, with many of them having a known effect on glucose metabolism. In a model of high-fat diet rats and IR skeletal muscle L6 cells, metformin dose-dependently decreased miR-21 expression, consequently improving skeletal muscle IR
[130]. A recent study examined the effect of metformin on miR-223 expression and the amount of Akt and GLUT4 proteins in IR 3T3-L1 adipocyte cells and adipocytes of diabetic patients
[101]. MiR-223 was overexpressed in both, and incubation of the cell line and 3-month treatment of patients with metformin decreased its expression while increasing Akt and GLUT4 expression
[101]. Mensà et al. found that miR-146a levels show a significant age-related decline that is even more notable in T2DM patients
[131]. They also observed that miR-146a was significantly overexpressed in T2DM patients treated with metformin
[131]. Although miR-146a is more commonly associated with inflammation
[132], it has also been associated with IR and poor glycemic control in Asian Indian T2DM patients
[133]. Furthermore, a randomized, double-blinded, and placebo-controlled three-month trial of metformin treatment in T2DM patients found that metformin (but not placebo) led to significant changes in circulating miR-192, miR-140-5p, and miR-222, in parallel to decreased fasting glucose and HbA1c
[134]. More recently, the same concept of metformin’s effect on extracellular miRNAs was demonstrated by Ghai et al.
[135]. They observed that the concentrations of several miRNAs were increased in T2DM, but they decreased to the normal range after metformin treatment. However, at present, there is little evidence that these circulating miRNAs are involved in GLUT4 expression or IR
[135]. In a study in PCOS patients, 12-month metformin treatment reduced serum levels of miR-122, miR-223, and miR-29a, all of which have previously been reported to influence glucose metabolism
[136].
A new area of research of metformin’s mode of action is likely to expand to other epigenetic mechanisms. In 2020, García-Calzón et al. published a paper where they discovered that by measuring blood-based epigenetic markers in the form of DNA methylation of 11 specific loci in drug-naive patients with T2DM, they could discriminate between glycemic responders and nonresponders to metformin, providing some further evidence into the role of DNA methylation
[137]. Additionally, there has been limited research covering the role of long noncoding RNAs (lncRNAs) to date. In C2C12 myotubes, metformin reduced medium glucose concentration in the culture medium, increased levels of GLUT4 in the plasma membrane, and repressed lncRNA Dreh expression
[138]. Interestingly, knockdown of Dreh had even more profound results, while its overexpression attenuated the glucose-lowering effect of metformin, suggesting that glucoregulatory actions of metformin are mediated in part by the lncRNA Dreh
[138]. In gastric cancer cells, metformin treatment significantly inhibited the cellular functions of cancer cells, and lncRNA H19 was a crucial component in that process
[139]. Interestingly, the same lncRNA H19 was previously shown to improve IR in skeletal muscle, although the proposed mechanism involves reducing ectopic lipid accumulation, not GLUT4 expression
[140].
3.4. GLUT4 Trafficking and Translocation
Many studies have suggested that defects in GLUT4 translocation are closely related to IR
[141,142,143,144,145][141][142][143][144][145]. The stimulation of glucose uptake requires the translocation of GLUT4 from tubulovesicular structures, named GLUT4 storage vesicles (GSV), to the cell surface, rapidly increasing the GLUT4 density at the plasma membrane and consequently glucose uptake
[119]. It is well documented that insulin enhances GLUT4 translocation through incompletely defined intracellular signaling pathways involved in releasing GSVs from intracellular retention, their trafficking, tethering, and, finally, docking and fusion to the plasma membrane
[7,52,119,146,147,148,149,150][7][52][119][146][147][148][149][150]. Furthermore, studies from the 1990s onwards suggested that metformin ameliorates IR independently of GLUT4 protein synthesis, mainly through subcellular redistribution of GLUT4
[49,147,151,152,153,154,155,156][49][147][151][152][153][154][155][156]. However, the mechanisms by which metformin acts on GLUT4 translocation are still largely unknown.
The understanding of glucose uptake involves the interplay between the insulin signaling pathway and GLUT4 membrane trafficking at the cellular level. Not surprisingly, identifying the molecules that link them has been a significant research focus. Since metformin has a known effect on the insulin signaling pathway, this could also explain the noted effects of metformin treatment on the GLUT4 translocation process
[148]. The lipid products of PI3K contribute to the activation of Akt and atypical protein kinases C zeta and lambda (PKCζ/λ)
[55,119][55][119]. The discovery in 2003 that a RabGAP, AS160, is a highly insulin-responsive Akt target provided one of the first links between insulin signaling and GLUT4 translocation, given that a significant function of Rab GTPases is to regulate vesicle traffic
[1,119,148,157][1][119][148][157]. AS160 has emerged as a negative regulator in the insulin transduction relay since overexpressing a phosphorylation-defective mutant reduced insulin-dependent GLUT4 translocation, and, conversely, deletion of AS160 elevated GLUT4 levels in the plasma membrane in the absence of insulin stimulation
[1,158][1][158]. Moreover, Rice et al. demonstrated that metformin increased insulin-stimulated translocation of GLUT4 to the plasma membrane in human ovarian granulosa cells via a mechanism involving PI3K activation of Akt
[71]. Additionally, Polianskyte-Prause et al. demonstrated a novel mechanism of metformin action through a reduction in SHIP2 activity, since in SHIP2-overexpressing myotubes, metformin ameliorated reduced glucose uptake by slowing down GLUT4 endocytosis
[78]. In addition to this, a step-by-step review of the protein related to the GLUT4 translocation machinery in skeletal muscle reported that several miRNAs have been potentially correlated with some target genes and could present additional targets of metformin action
[119].
Some authors also demonstrated the crucial role of AMPK activation and its downstream effect on GLUT4 translocation mediators after metformin treatment. For example, Lee et al. noted metformin’s activity through the phosphorylation of Cbl and stimulated expression of Cbl-associated protein (CAP) in an AMPK-dependent manner, and Cbl/CAP-associated multicomplex formation modulated GLUT4 translocation in 3T3-L1 cells
[109]. In another paper, Lee et al. demonstrated that metformin induced Rab4 expression via the AMPK pathway and demonstrated that the activities of AS160 and PKCζ are involved in metformin-induced Rab4 regulation in C2C12 cells
[108]. Together, these studies suggest that metformin also induces Rab4 expression via AMPK–AS160–PKCζ and modulates insulin-mediated GLUT4 translocation
[108]. On the other hand, Kristensen et al. did not note any changes in insulin signaling upon Akt and AS160 protein expression or their phosphorylation or changes in protein expression of Rab4 after two weeks of metformin treatment in mouse soleus muscle
[89].