The importance of MafA for glucose homeostasis was shown in MafA-deficient mice, which displayed glucose intolerance after weaning due to impaired glucose-stimulated insulin secretion (GSIS) and age-dependent diabetes progression ^[11][12]. MafA-deficient mice had decreased mRNA levels of Ins1, Ins2, Pdx1, Neurod1 and Slc2a2, suggesting that MafA is essential for the transcriptional identity of β-cells. MafA works synergistically with Neurod1 and Pdx1 to transactivate the insulin promoter, which is an effect that could not be fully replicated with MafB nor Maf ^[13]. Along these lines, MafA promotes insulin transcription when overexpressed in rat islets ^[14] and when ectopically expressed in chick embryonic endoderm ^[15] or in other non-β-cell lines of endodermal origin ^[16][17].

Several studies suggest that MafA is dispensable for β-cell development but plays an essential role in β-cell maturation and glucose responsiveness of adult β-cells. In rodents, Mafa expression is low at birth, increasing during the postnatal period ^[18][19]. Interestingly, premature expression of MafA in pancreatic endocrine progenitors prevents differentiation into hormone-expressing cells ^[20][21], suggesting that the induction of MafA expression in β-cells is time sensitive. Consistently, in mice with whole pancreas MafA knockout (Mafa^∆panc), the effects of MafA loss are not apparent until three weeks of age, when the KO mice showed lower β-cell mass, decreased insulin expression, and impaired glucose tolerance compared to controls ^[22]. Similarly, at twelve weeks of age in whole-body MafA knockout mice, there is a reduced β-cell:α-cell ratio, decreased islet insulin content and β-cell dedifferentiation into MafB-expressing, progenitor-like cells ^[12]. These results may recapitulate some aspects of normal islet development. MafB is actually the predominant large MAF protein expressed during pancreatic development in both α- and β-cells ^[4]. MafB can upregulate Ins1 and Ins2 transcription ^[8], but in adult mice, MafB is mainly expressed in α-cells and upregulates Gcg gene expression ^[4][23]. The decrease in β-cell MafB expression after birth is at least partly due to methylation, specifically by DNA methyltransferase 3a (Dnmt3a) which binds and represses at a region −1032 to −838 upstream of the Mafb transcriptional start site ^[24]. Mafb^∆panc mice have higher blood glucose levels at P1, but two weeks after birth, blood glucose levels normalize. Likewise, at E15.5, Mafb^∆panc mice have a lower number of insulin⁺ and glucagon⁺ cells compared to controls, but two weeks after birth, cell numbers became roughly the same ^[25]. MafB does seem to have a role in enhancing postnatal β-cell function under metabolic stress (pregnancy, high-fat diet) ^[24][26], but MafA remains critical for proper β-cell function and glucose homeostasis in mature β-cells, with forced MafB expression unable to compensate for MafA loss in adult mice ^[24].

Reduced MafA expression is associated with diabetes progression in mice and human patients. In islets isolated from diabetic db/db mice, MafA is decreased due to hyperglycemia-associated oxidative stress and c-Jun activity ^[9]. Similarly, islets from subjects with type 2 diabetes (T2D) display a marked decrease in MAFA mRNA levels and protein expression ^[10][27][28]. Several studies have shown possible mechanisms for this phenomenon. In β-cells from subjects with T2D ^[29] and in in vitro studies mimicking oxidative stress ^[10], MafA was found primarily in the cytoplasm rather than properly localized in the nucleus. However, when MafA levels were “rescued” in db/db mice, GSIS and β-cell mass improved, suggesting that preserving MafA expression in β-cells can still mitigate diabetes progression ^[30]. scRNA-seq analysis of human β-cells supports this concept as well, wherein metabolically inflexible β-cells show lower MAFA activity as compared to healthy β-cells ^[31]. This finding is thought to align with previous patterns of T2D-induced oxidative stress, in which case MafA activity is high under healthy islet conditions but can also increase as a protective mechanism against acute oxidative stress ^[32]. In a different scRNA-seq analysis of human islets, β-cells co-expressing MAFA/MAFB had increased expression of genes related to β-cell identity, glucose metabolism and exocytosis compared to β-cells that only expressed one or neither TF ^[33]. Since MAFA and MAFB are capable of forming heterodimers ^[8][34], it is possible that the co-expression of both can enhance the β-cell maturity transcriptional program. These and other studies have documented the functional and transcriptional heterogeneity of β-cells ^[31][33][35].

MafA transcriptional activity is necessary for proper postnatal β-cell function. MafA was first identified as a TF of insulin that binds to the C1 element of the insulin gene promoter. MAF proteins, including MafA, bind to the MAF-recognition elements (MAREs), through a DNA consensus sequence TGCTGAC(G)TCAGCA ^[36]. However, in β-cells, the MARE sequence within the C1 element deviates slightly from the consensus sequence. The MARE sequence in C1 element of the rat and mouse Ins2 promoter is TGCAGCTTCAGCC whereas the equivalent in the human INS promoter is TGCAGCCTCAGCC (the underlined nucleotides indicate deviations from the MARE consensus sequence) ^[7][16][37]. Interestingly, in addition to the C1 element, three other MARE sites (MARE1, MARE2 and MARE3) were also identified in the rat Ins2 and human INS promoter. MafA can bind to MARE2 and MARE3 in the rat Ins2 promoter while MafA can bind to MARE1 in the human INS promoter ^[16].

Although MafA can activate the insulin promoter alone, the addition of Neurod1 and Pdx1, which bind the E box and A box of the insulin promoter, respectively, synergistically promote insulin transcription ^[13]. Consistently, mice with homozygous CRISPR-Cas9-introduced mutations in the C1 box of the Ins1 and Ins2 gene promoters are glucose intolerant. These C1 box mutations also prevented MafA from activating the promoter in luciferase assays, further supporting the importance of MafA in upregulating the promoter of its target genes ^[38]. Human mutations characterized by a deletion of the C1 and E1 elements of the INS promoter have been shown to cause permanent neonatal diabetes, which may be attributed to the disruption of MAFA and NEUROD1 binding sites ^[39][40]. Other factors are also involved in this regulation. For example, Gli-similar 3 (Glis3), a Krüppel-like zinc finger transcription factor, activates insulin transcription by recruiting histone acetyltransferase Creb-binding protein (CBP) and by acting as a scaffold for the MafA, NeuroD1 and Pdx1 complex ^[41].

Beyond insulin, MafA increases the expression of a wide range of genes involved in β-cell function, including genes implicated in glucose sensing, insulin processing, Ca²⁺ influx and oxidative phosphorylation. Many studies have identified a plethora of MafA target genes, all having important roles in proper β-cell function, further supporting MafA’s critical role in β-cells. While some studies have found direct MafA target genes (i.e., MafA directly binds to the promoter of the target genes), other studies describe genes that are differentially expressed when MafA protein is overexpressed or silenced, without showing direct MafA binding to the target gene promoters (i.e., MafA knockout or overexpression causes a decrease or increase in the target gene expression, respectively). Some of these MafA-target genes are summarized in Table 1. The lack of promoter binding evidence does not necessarily mean MafA cannot bind to the promoter of these genes, but rather that MafA may regulate these genes through intermediaries rather than direct promoter binding.