Second MAFA Variant Causing a Phosphorylation Defect in the Transactivation Domain and Familial Insulinomatosis

Simple Summary Adult-onset familial insulinomatosis is a rare disorder with recurrent, severe hypoglycemia caused by multiple insulin-secreting pancreatic tumors. The etiology was unclear until the genetic variant p.Ser64Phe in the MAFA protein, a key coordinator of insulin secretion in pancreatic cells, was defined as the cause in two families. Based on the cases of two sisters with insulinomatosis, we aimed to identify further disease causes. The sequencing of the complete coding regions of the patients’ genomes revealed a second genetic MAFA variant, p.Thr57Arg, as the cause of familial insulinomatosis, linking genetic, clinical, and biochemical analyses from the patients’ family to the pre-described cell culture data. Thus, we confirm a defect in a crucial regulatory region of the MAFA protein as an important cause of a specific hereditary syndrome, which is characterized by insulinomatosis and/or mild hyperglycemia. This study extends the pathophysiological and diagnostic disease concept and verifies the inheritance pattern of familial insulinomatosis. Abstract Adult-onset familial insulinomatosis is a rare disorder with recurrent, severe hypoglycemia caused by multiple insulin-secreting pancreatic tumors. The etiology was unclear until the variant p.Ser64Phe in the transcription factor MAFA, a key coordinator of β-cell insulin secretion, was defined as the cause in two families. We here describe detailed genetic, clinical, and family analyses of two sisters with insulinomatosis, aiming to identify further disease causes. Using exome sequencing, we detected a novel, heterozygous missense variant, p.Thr57Arg, in MAFA’s highly conserved transactivation domain. The impact of the affected region is so crucial that in vitro expression studies replacing Thr57 have already been performed, demonstrating a phosphorylation defect with the impairment of transactivation activity and degradation. However, prior to our study, the link to human disease was missing. Furthermore, mild hyperglycemia was observed in six additional, heterozygote family members, indicating that not only insulinomatosis but also MODY-like symptoms co-segregate with p.Thr57Arg. The pre-described MAFA variant, p.Ser64Phe, is located in the same domain, impairs the same phosphorylation cascade, and results in the same symptoms. We confirm MAFA phosphorylation defects are important causes of a characteristic syndrome, thus complementing the pathophysiological and diagnostic disease concept. Additionally, we verify the high penetrance and autosomal dominant inheritance pattern.


Introduction
Hypoglycemia due to endogenous hyperinsulinism is a rare clinical condition. In newborn and infants, rare monogenetic diseases caused by variants in genes involved in the regulation of insulin secretion (ABCC8, KCNJ11, GCK, HADHSC, INSR, GLUD1, SLC16A1, HNF1A, HNF4A, and UCP2) present with severe persistent hypoglycemia [1]. In adults, however, the most common cause of this pathology is either sporadic insulinoma (80%) or insulinoma associated with multiple endocrine neoplasia type I (MEN1) (9-10%) [2,3]. Although insulinomas can be small and sometimes challenging to detect, modern diagnostic and surgical techniques usually allow localization and successful resection today. In the remaining 10% of cases, endogenous pancreatic hyperinsulinism is either due to insulinomatosis or functional β-cell disorders, clinically termed as noninsulinoma pancreatogenous hypoglycemia syndrome (NIPS) and adult nesidioblastosis [4,5]. They are histopathologically characterized by β-cell hypertrophy, islet hyperplasia, and an increase in β-cell mass. Disease pathogenesis is largely unknown and may be different from congenital hyperinsulinemic hypoglycemia, given that so far no distinct genetic causes have been identified in adult patients with NIPS/nesidioblastosis [1,6,7]. However, the observed association with gastric bypass surgery in obese patients suggests that a reactive process possibly unmasks or induces a defect in β-cells, resulting in hyperfunction [8,9].
Insulinomatosis is another rare cause of hyperinsulinemic hypoglycemia in adults and is characterized by the synchronous and metachronous occurrence of small insulinomas, multiple insulinoma precursor lesions, and small proliferative insulin-expressing monohormonal endocrine cell clusters (IMECCs) [10]. Clinically, in parallel to few macrotumors (usually 0.5-1 cm), multiple microtumors are found throughout the entire pancreas, expressing exclusively insulin. Due to the small size and the multicentric occurrence of the tumors, both diagnosis and therapy are clinically challenging. The clinical course of the disease is characterized by early recurrent hypoglycemia requiring repeated surgical intervention. Already in 1977, Tragl et al. were the first to describe a family with multiple adenomas restricted to β-cells and the occurrence of diabetes mellitus in different members of the same family, raising the possibility of a common genetic origin [11]. However, only recently, a disease-causing variant in the V-MAF avian musculoaponeurotic fibrosarcoma oncogene homolog A (MAFA) basic leucine zipper-containing protein has been identified causing insulinomatosis [12]. MAFA belongs to the family of large MAF transcription factors and is expressed in islet β-cells. It is required for postnatal β-cell function and acts as a transactivator of insulin and several genes involved in glucose-stimulated insulin secretion [13][14][15][16][17]. The only clinically relevant MAFA variant known, Ser64Phe, is associated with the autosomal dominant inheritance of either insulinomatosis or diabetes mellitus, documenting the physiological properties of MAFA, both as an oncogene and key islet β-cell transcription factor [12]. In this study, we investigated two sisters with recurrent hyperinsulinemic hypoglycemia for genetic causes. Using whole exome sequencing, bioinformatics, and family analysis, we identified a second, hitherto unreported MAFA variant (NM_201589.4: c.170C>G, p.(Thr57Arg)) as a cause of insulinomatosis and mild hyperglycemia in other family members. The exchange of Thr57 results in a phosphorylation defect as it is located in the functionally crucial part of the transactivator domain, and its relevance has already been demonstrated by others. Figure 1. Pedigree, presenting genotype, phenotype and age of the family members. Four generations of the family underwent further molecular genetic (MAFA Thr57Arg and PROX1 Ala394Thr; genotyped individuals marked by the bold border of the symbols) and laboratory testing as well as an evaluation of the medical history. In individuals with the symbol filled in dark blue, only a buccal swab was available for laboratory analysis (III2, V1, and VI1-5), suitable for genotyping but not for metabolic analyses, e.g., the determination of HbA1c. MAFA Thr57Arg was detected (symbols with red bold border) in 12 family members (III1-2, IV3-4, V1-3, V5, and VI1-4), 2 with insulinomatosis (IV3-4, arrow points on index case), 4 with an overt diabetes (III1 (based on patient medical history), V2-3 and V5 (HbA1c > 5.6 or impaired fasting glucose (IFG)), and 6 members (4 children below 10 years (VI1-4) and 2 adults (III2 and V1)) whose glycemic status was unclear, because blood samples were not available. Furthermore, diabetes mellitus was known in 2 deceased individuals (III3 and IV2). In 7 family members who did not carry Thr57Arg (symbols with green bold border), neither diabetes nor insulinomatosis has been detected (III4-5, IV5, and V4 + 6-7) or they had an unclear glycemic status (VI5). Subject numbers of the six PROX1 Ala394Thr carriers are underlined (III4, IV3-4, V4-5, and VI5). The MAFA mutation is inherited through the left wing of the pedigree (starting with the left couple in generation I). The PROX1 variant, on the other hand, was introduced into the pedigree by person III4 and is, thus, inherited only throughout one branch. CUP = cancer of unknown primary; y. = age in years; † = deceased. Pedigree, presenting genotype, phenotype and age of the family members. Four generations of the family underwent further molecular genetic (MAFA Thr57Arg and PROX1 Ala394Thr; genotyped individuals marked by the bold border of the symbols) and laboratory testing as well as an evaluation of the medical history. In individuals with the symbol filled in dark blue, only a buccal swab was available for laboratory analysis (III2, V1, and VI1-5), suitable for genotyping but not for metabolic analyses, e.g., the determination of HbA1c. MAFA Thr57Arg was detected (symbols with red bold border) in 12 family members (III1-2, IV3-4, V1-3, V5, and VI1-4), 2 with insulinomatosis (IV3-4, arrow points on index case), 4 with an overt diabetes (III1 (based on patient medical history), V2-3 and V5 (HbA1c > 5.6 or impaired fasting glucose (IFG)), and 6 members (4 children below 10 years (VI1-4) and 2 adults (III2 and V1)) whose glycemic status was unclear, because blood samples were not available. Furthermore, diabetes mellitus was known in 2 deceased individuals (III3 and IV2). In 7 family members who did not carry Thr57Arg (symbols with green bold border), neither diabetes nor insulinomatosis has been detected (III4-5, IV5, and V4 + 6-7) or they had an unclear glycemic status (VI5). Subject numbers of the six PROX1 Ala394Thr carriers are underlined (III4, IV3-4, V4-5, and VI5). The MAFA mutation is inherited through the left wing of the pedigree (starting with the left couple in generation I). The PROX1 variant, on the other hand, was introduced into the pedigree by person III4 and is, thus, inherited only throughout one branch. CUP = cancer of unknown primary; y. = age in years; † = deceased. Ten samples from German healthy volunteers were analyzed in parallel with the genomic DNA (gDNA) of the two sisters by Next Generation Sequencing to eliminate common local variants and sequencing artifacts.

Patients
All laboratory analyses in the patients were performed for diagnostic purposes at the Institute of Clinical Chemistry and Laboratory Medicine and the Institute of Pathology, University Medical Center Mainz. Patients and healthy controls provided explicit consent to the use of their pseudonymized data for research purposes. The studies were designed and executed in accordance with all local legal and regulatory requirements, notably the General Data Protection Regulation (EU 2016/679) and the Declaration of Helsinki, 7th revision.

Histopathology and Immunohistochemistry
Serial sections from archival pancreatic tissue were subjected to MAFA immunohistochemistry as well as insulin and glucagon staining. The result was evaluated and compared to that of normal human pancreas by an experienced pathologist. To achieve comparability, MAFA immunohistochemistry was performed using the same MAFA antibody (Anti-MafA antibody ab26405, Abcam, Cambridge, UK, RRID: AB_776146; dilution: 1:2000) as Iacovazzo et al. [12] following the manufacturer's instructions. After deparaffinization, hydration, and heat-induced epitope retrieval, incubation steps were carried out in an Autostainer 480-A (Thermo Fisher Scientific, Dublin, Ireland) using the Dako EnVision™ FLEX HRP/Dab (Agilent, Santa Clara, CA, USA) detection system with signal amplification by EnVision™ FLEX + Rabbit (LINKER) (Agilent, Santa Clara, CA, USA).
Standard staining/immunohistochemisty protocols of the Institute of Pathology (University Medical Center Mainz) were used to detect the neuroendocrine/proliferation markers chromogranin, synaptophysin, insulin, glucagon, and KI-67 in formalin-fixed, paraffinembedded pancreatic tissue sections from patients IV3 and IV4 following resection.

Bioinformatics Analyses
The quality of WES was checked with fastQC (Babraham Bioinformatics) and the sequencing analysis viewer (Illumina, San Diego, CA, USA). The format conversion of the resulting fastq-files, whole-genome alignment (GRCh37/hg19), variant calling, and copy number variation (CNV) detection was performed using NextGENe, version 2.4.1.1 (Softgenetics, State College, PA, USA). Variants were filtered by an in-house Perl-pipeline: All variants present in the ten controls were excluded, as well as variants with a minor allele frequency (MAF) above 0.01 in dbSNP. The remaining variants were reduced by two gene lists: one targeting genes known to cause endocrine tumors ( PRC1, PRKAA2, PROX1, PTF1A, RBMS1, SIRT1, SLC16A1, SLC2A2, SLC30A8, TCF7L2,  THADA, TLE4, TMEM195, TP53INP1, TSPAN8, UCP2, VDR, WFS1, and ZBED3). Variants detected in both sisters, were considered as potential causes of the disease. Starting at the age of 34, the patient experienced continuous weight gain (>20 kg) and recurrent episodes of severe hunger. An initial medical and psychiatric examination was unremarkable. It has been performed due to a three-hour incident in which the patient had damaged other vehicles with her car but could not remember it afterwards. At the age of 38, a "low blood glucose level" was incidentally detected. Consequently, a 72-h fasting test was initiated and had to be stopped after 4 h due to hypoglycemia (16 mg/dL) with increased insulin and C-peptide levels (data not shown). An endoscopic ultrasound (EUS) of the pancreas showed a 7 mm lesion in the tail of the pancreas (data not shown). The patient received left-sided pancreatic resection with the histological diagnosis of a well-differentiated neuroendocrine tumor (NET) of 0.6 cm that was immunohistochemically positive for chromogranin, synaptophysin, and insulin (NET G1, Ki-67 index 1%, pT1, pN0, L0, V0, Pn0, R0, G1). After surgery, she still complained about hypoglycemia symptoms. Diagnostic work-up revealed persistent hyperinsulinemic hypoglycemia with another solitary lesion of about 5 mm in the pancreatic body (in EUS) (Figure 2a-c). Functional imaging with 68Ga-DOTATOC-PET/CT was negative (Figure 2d). Biochemical and genetic analyses (Sanger sequencing and MLPA of the MEN1 gene) provided no evidence for MEN1 variants. Diagnostic laparotomy with intraoperative ultrasound identified two focal lesions that were approximately 5 mm in diameter. Histological examination after partial resection of the pancreatic head confirmed the diagnosis of insulinomatosis with multifocal microadenomas that were monohormonally positive for insulin and mostly <5 mm in size, except for two larger adenomas (NET G1) of 6 and 7 mm in diameter. Since the second surgical intervention, the patient was asymptomatic. After the detection of the germline MAFA variant Thr57Arg, MAFA immunohistochemistry was performed on residual archival tissue of the resected pancreatic specimen. MAFA protein expression was found in the nuclei of islet cells as well as in the nuclei of an insulinoma. Although the data should be considered preliminary, at least no obvious difference in MAFA expression was found between patient islets and insulinoma and wild-type islets from non-insulinoma patients ( Figure 3).

Clinical Case 2; Sister of Clinical Case 1 (IV3, 57 Years Old at Time of Study)
Since the age of 45, the patient had suffered from recurrent episodes of fatigue, confusion, blurred vision, and tremors, which improved with food intake. The consequence was continual weight gain (11 kg). At the age of 53, a low blood glucose level (60 mg/dL) was detected when the patient was found to be unconscious after physical stress. Persistent symptoms and the sister's medical history finally prompted a diagnostic work-up. A prolonged oral glucose tolerance test (oGTT) was unremarkable. A 72 h fasting test was stopped after 11 h with a blood glucose level of 35 mg/dl and clinical signs of hypoglycemia, immediately reversible after intravenous glucose infusion ( Figure 4a). The corresponding insulin-, C-peptide-and pro-insulin levels were inappropriately elevated (14 mU/mL, 2.97 ng/mL, and 20.3 pmol/L). Molecular genetic testing for MEN1 was negative, as were imaging procedures (including magnetic resonance imaging (MRI), EUS, and 68Ga-DOTATOC-PET/CT). Functional imaging with 18F-DOPA-PET/CT (GLP-1-receptorimaging was not available at that time) showed physiological radionuclide accumulation in the common bile duct and a diffuse, homogeneous uptake throughout the pancreas, although this was judged to be nonspecific (Figure 4b). A 2-fold to 3-fold increase in insulin levels was observed in all three arteries (superior mesenteric, gastroduodenal, and splenic artery) in response to selective arterial calcium infusion (SACI) (Figure 4c). Histological evaluation after a duodenum-preserving resection of the pancreatic head showed multiple small insulin-positive tumorlets with 2-4 mm in diameter, consistent with the diagnosis of insulinomatosis. Since the operation, the patient has been asymptomatic.   Since the age of 45, the patient had suffered from recurrent episodes of fatigue, confusion, blurred vision, and tremors, which improved with food intake. The consequence was continual weight gain (11 kg). At the age of 53, a low blood glucose level (60 mg/dL) was detected when the patient was found to be unconscious after physical stress. Persistent symptoms and the sister's medical history finally prompted a diagnostic work-up. A prolonged oral glucose tolerance test (oGTT) was unremarkable. A 72 h fasting test was stopped after 11 h with a blood glucose level of 35 mg/dl and clinical signs of hypoglycemia, immediately reversible after intravenous glucose infusion (Figure 4a). The corresponding insulin-, C-peptide-and pro-insulin levels were inappropriately elevated (14 mU/mL, 2.97 ng/mL, and 20.3 pmol/L). Molecular genetic testing for MEN1 was negative, as were imaging procedures (including magnetic resonance imaging (MRI), EUS, and 68Ga-DOTATOC-PET/CT). Functional imaging with 18F-DOPA-PET/CT (GLP- increase in insulin levels was observed in all three arteries (superior mesenteric, gastroduodenal, and splenic artery) in response to selective arterial calcium infusion (SACI) (Figure 4c). Histological evaluation after a duodenum-preserving resection of the pancreatic head showed multiple small insulin-positive tumorlets with 2-4 mm in diameter, consistent with the diagnosis of insulinomatosis. Since the operation, the patient has been asymptomatic.    (Table 1), which were confirmed by Sanger Sequencing (Figure 5). PROX1 and MAFA are both transcription factors involved in β-cell differentiation, although unlike PROX1, MAFA is expressed in mature β-cells and plays a major role in the regulation of insulin secretion [21]. The PROX1 variant is very rare (Table 1), and the MAFA variant has not been described in the literature or databases yet. Both variants cause amino acid exchanges. Bioinformatics tools consistently predict pathogenic significance for the MAFA variant Thr57Arg, while prediction for the PROX1 variant c.1180G>A, p.Ala394Thr, is predominantly benign ( Table 1). A further classification of the variants according to the ACMG (American College of Medical Genetics) [22] criteria was based on multiple sequence alignments and extensive functional data from the literature ( Table 1). The MAFA variant is located in a region of the transactivation domain that is highly conserved between species (Figure 6a) and among the members of the large MAF transcription factor family (Figure 6b). The phosphorylation of this D/E/S/T/P-rich region is known to be crucial for binding MAFA to its target DNA sequences [23][24][25][26], regulation of β-cell insulin secretion [13,27], transforming activity of MAFA [28], and MAFA degradation by the proteasome [23,25,[29][30][31]. Thr57 is one of the four serine and threonine residues phosphorylated in a series of phosphorylation steps mediated by GSK3 (Glycogen synthase kinase 3) and initiated primarily by phosphorylation of Ser65 by a priming kinase (Figure 6b). A set of evidence from in vitro and in vivo studies has confirmed that this basal mechanism is very similar among human large MAF proteins [21]. Transfection studies in which the five crucial serine and threonine residues of MAFA were systematically replaced by alanine show a significant phosphorylation defect for all constructs, including the Thr57Ala construct [23]. Apart from MAFA translocations in patients with multiple myeloma, only one pathogenic variant in the MAFA gene is known, Ser64Phe. It prevents phosphorylation of Ser65 by the priming kinase. Ser64Phe leads to a characteristic syn-drome in which some family members, mainly women, are affected by insulinomatosis and other family members, mainly men, by mild diabetes [12], consistent with the observations in the family described here. No benign variants have been reported within this part of the transactivation domain. In the large databases (e.g., dbSNP and GnomAD), single, extremely rare variants of uncertain significance are listed, but none of them affects phosphorylated residues 49, 53, 57, 61, and 65. However, variants in homologous residues of other large MAF proteins have been described, and they are all pathogenic. In both, MAFB and c-MAF (alias MAF), a Thr62Arg variant (corresponding to Thr57Arg in MAFA) causes characteristic syndromes with autosomal dominant inheritance. MAFB Thr62Arg and several other amino acid exchanges at this position cause Multicentric Carpotarsal Osteolysis Syndrome (MCTO, OMIM #166300) [21,32], and Thr62Arg in c-MAF results in the Ayme-Gripp syndrome (OMIM #601088) [24] (Figure 6b). Interestingly, the clinical picture caused by MAFB and c-MAF variants depends on their localization. Variants that are not located in the highly phosphorylated part of the transactivation domain cause different syndromes, respectively. Based on this evidence and in accordance with the ACMG criteria, we classified the MAFA variant Thr57Arg as pathogenic even at this stage of investigation. In contrast to that, the detected PROX1 variant is not located in one of the essential domains. Several PROX1 variants have been associated with a risk for alterations in glucose and lipid metabolism [33], but not with monogenic diabetes or insulinomatosis. For these reasons and because we had already classified the MAFA variant as pathogenic, we assessed the PROX1 variant as likely benign, even though it cannot be ruled out as a modifier of the disease. The MAFA variant is located in a region of the transactivation domain that is highly conserved between species (Figure 6a) and among the members of the large MAF transcription factor family (Figure 6b). The phosphorylation of this D/E/S/T/P-rich region is known to be crucial for binding MAFA to its target DNA sequences [23][24][25][26], regulation of β-cell insulin secretion [13,27], transforming activity of MAFA [28], and MAFA degradation by the proteasome [23,25,[29][30][31]. Thr57 is one of the four serine and threonine residues phosphorylated in a series of phosphorylation steps mediated by GSK3 (Glycogen synthase kinase 3) and initiated primarily by phosphorylation of Ser65 by a priming kinase (Figure 6b). A set of evidence from in vitro and in vivo studies has confirmed that this basal mechanism is very similar among human large MAF proteins [21]. Transfection studies in which the five crucial serine and threonine residues of MAFA

Molecular Genetic Testing of Asymptomatic Family Members
Four generations underwent further molecular genetic and laboratory testing as well as an evaluation of their medical history ( Figure 1, Table 2). In addition to the index patient and her sister, a total of 17 clinically asymptomatic family members were investigated (eight female and nine male). Apart from the two index cases, MAFA variant Thr57Arg was detected in 10 other family members (five female and five male). None of them had been diagnosed with insulinomatosis so far. In the entire family, however, four members were already diagnosed with overt type 2 diabetes (three men and one female). Two of them (III3 and IV2; Figure 1) had died from other causes (pancreatic cancer and cancer of unknown primary): One is treated with insulin and metformin (III1), and one is treated with diet only (V3). Two further family members (one female and one male) showed borderline HbA1c and/or impaired fasting glucose (IFG) levels (V2, HbA1c 5.8% and V5, IFG, HbA1c 5.9% [ref.: 4.1-5.6%]; Figure 1, Table 2). , x FOR PEER REVIEW 13 of 22 classified the MAFA variant as pathogenic, we assessed the PROX1 variant as likely benign, even though it cannot be ruled out as a modifier of the disease.  (21,23). Amino acid residues known to be affected by pathogenic variants are provided in blue and underlined, and the MAFA variant Ser64Phe is among them (blue arrow, S64F), which has been shown to cause a phosphorylation defect and a syndrome characterized by insulinomatosis, mild diabetes, and cataract (12). The red arrow marks the amino acid affected by the MAFA variant Thr57Arg (T57R) reported in this manuscript. In vitro transfection studies (23)

Molecular Genetic Testing of Asymptomatic Family Members
Four generations underwent further molecular genetic and laboratory testing as well as an evaluation of their medical history (Figure 1, Table 2). In addition to the index patient and her sister, a total of 17 clinically asymptomatic family members were investigated (eight female and nine male). Apart from the two index cases, MAFA variant Thr57Arg  (21,23). Amino acid residues known to be affected by pathogenic variants are provided in blue and underlined, and the MAFA variant Ser64Phe is among them (blue arrow, S64F), which has been shown to cause a phosphorylation defect and a syndrome characterized by insulinomatosis, mild diabetes, and cataract (12). The red arrow marks the amino acid affected by the MAFA variant Thr57Arg (T57R) reported in this manuscript. In vitro transfection studies (23) have confirmed that the exchange of MAFA residues 65, 61, 57, 53, and/or 49 by alanine, an amino acid which cannot be phosphorylated, causes phosphorylation defects and impairs MAFA-binding to DNA. For two further large MAF proteins (MAFB and c-MAF), pathogenic missense variants have already been detected and among them include the MAFB variants Thr62Arg (21) and Thr62Ala (ClinVar VCV000807627.3, T62I/P/R/A) and the MAF (c-MAF) variant Thr62Arg (T62R) (24), variants with the same amino acid exchange as we now found in MAFA.  Four patients presented with a tumor disease (one with lung cancer (IV1), one with lung cancer and a brain tumor (III1), one with pancreatic cancer (IV2), and one with a cancer of unknown primary (III3)), and one female with advanced age has dementia. There was no history of congenital eye disorders in this family.
Genetic testing revealed that the six symptomatic (insulinomatosis or mild diabetes; classification here after laboratory analyses) individuals studied from generations III-V (Figure 1, III1; IV3, 4; V2, 3, 5) all carried the MAFA variant, but only three of them had the PROX1 variant (IV3, 4; V5). Generation VI and individuals III2 and V1 could not be analyzed in this regard because only a buccal swab was available for genotyping, but no EDTA blood was available for determining HbA1c. In the six healthy subjects studied, the MAFA variant was absent (III4, 5; IV5; V4, 6, 7), whereas the PROX1 variant was detected in two of them (III4, V4). Thus, the segregation of these variants in the family clearly suggests the MAFA variant as the cause of the disease and further autosomal dominant inheritance.

Discussion
Using whole exome sequencing, we identified a so far unreported missense MAFA variant (c.170C>G, p.Thr57Arg) as the cause of insulinomatosis in two sisters with recurrent hyperinsulinemic hypoglycemia. Thr57Arg is located in MAFA's highly conserved transactivation domain, which is so crucial that extensive functional analyses have already been performed. These include expression studies confirming that the replacement of threonine at position 57 with an amino acid that cannot be phosphorylated by GSK3 results in a defect in the stepwise phosphorylation of MAFA, impairing the binding of MAFA to its DNA target and preventing proteasome-mediated degradation [23,25,[29][30][31]. So far, only one other pathogenic MAFA variant (c.191C>T, p.Ser64Phe) has been described. It is located in the same MAFA domain, impairs the same phosphorylation cascade, and results in the same characteristic symptoms in the two affected families as observed in the family described here [12]. In summary, we describe a novel pathogenic MAFA missense variant that is highly likely to cause impaired phosphorylation and, therefore, functional defects, cosegregating with both phenotypes, insulinomatosis and MODY-like diabetes ("MODYlike": autosomal dominantly inherited, rather mild hyperglycemia (diabetes/IFG/HbA1c > 5.6) caused by a pathogenic variant in a transcription factor regulating insulin secretion). The detection of this second pathogenic MAFA variant confirms that a phosphorylation defect in the transactivation domain is accompanied by a characteristic set of symptoms in the affected family and confirms the autosomal dominant inheritance of the disease.
Iacovazzo et al. recently demonstrated for the first time that the Ser64Phe mutation of the MAFA gene was linked to insulinoma or mild, non-insulin dependent diabetes [12]. Ser64Phe was discovered to impair phosphorylation within MAFA's transactivation domain and to profoundly increase MAFA protein stability in β-cell-lines at high and low glucose concentrations. There was no significant change in the amount of wild-type and mutant MAFA mRNA in transfected cells, confirming that the effect on protein turnover was posttranscriptional. These findings imply that the activity of Ser64Phe-MAFA is enhanced in β-cells/pancreas islets due to an increased transactivation capacity that is at least partly related to decreased MAFA degradation [12,23,25,[29][30][31]. However, the transactivation activity of MAFA can be both decreased and increased depending on the cell type studied and the expression of other transcription factors [12,23,25,[28][29][30][31]. The Thr57Arg MAFAvariant found in our family, similar to that reported by Iacovazzo et al. [12], is located within the highly conserved D/E/S/T/P-rich domain, impairing phosphorylation within MAFA's transactivation domain, thus enhancing the protein stability and probably the total transactivation capacity of MAFA. Whether a phosphorylation defect in the D/E/S/T/Prich domain of MAFA increases or decreases, net MAFA protein expression in β-cells cannot be conclusively answered at present. Both Iacovazzo et al. [12] and our group found no obvious difference by immunohistochemistry between MAFA expression in islets and insulinomas in insulinomatosis patients and islets from patients not affected by insulinomatosis, but the data are based on very few cases so far.
MAFA, MAFB, and MAF are large MAF proteins that have been shown to be oncogenes in tissue cultures, animal models, and human cancer. MAFA's ability to induce proliferation of quail neuroretina cells [34] and chicken embryo fibroblasts [35] when overexpressed demonstrated its transformation potential. Large MAF proteins are overexpressed in 50% of multiple myelomas and 60% of angioimmunoblastic T-cell lymphomas in humans. In particular, in 8-10% of multiple myelomas the MAF, MAFB, and MAFA genes are translocated to the immunoglobulin heavy chain locus [36]. When expressed alone in β-cells, MAFA is only a weak transactivator of the insulin promoter, but when co-expressed with Pdx1 and Beta2/NEUROD1, these three factors synergistically and strongly activate the promoter [27]. The simultaneous expression of these three factors in non-β-cells (e.g., liver cells) induced the expression of the endogenous insulin gene, as well as other important β-cell genes (GCK, SLC2A2, G6PC2, KCNJ11, and SUR1) [23,29,30]. Thus, these three key β-cell transcription factors can be used for trans-differentiating non-β-cells into insulinproducing cells [37]. Immunohistochemical examinations in patients with MEN1 associated insulinomas demonstrated a linked decrease in both menin and MAFA. Moreover, in mouse insulinomas, decreased MAFA expression resulting from targeted MEN1 ablation was consistently observed [38]. Furthermore, in vitro analysis using insulinoma-derived cell lines revealed that menin regulates MAFA on both protein and mRNA levels and binds to MAFA promoter sequences. Menin knockdown concomitantly decreased the mRNA expression of both MAFA and β-cell differentiation markers (INS, GCK, SLC2A2, and PDX1) while increasing tumor proliferation, indicating that altered menin expression disrupts the MAFA differentiation pathway in insulinoma.
In conclusion, there is growing evidence that mutations of MAFA, which impair phosphorylation within the transactivation domain, result in increased MAFA protein stability and probably transactivation capacity. The transactivation activity exerted by MAFA in this process is presumably dependent on the maturity and degree of differentiation of the β-cell or its progenitor, as well as on the expression of further transcription factors and possibly also on the current metabolic conditions. This mechanism could explain the physiological properties of MAFA acting as an oncogene as well as a key transcription factor of islet β-cells, which regulates insulin secretion and the function of differentiated β-cells.
However, the mechanisms explaining how the same gene variant causes either diabetes or insulinomatosis are still unclear. Iacovazzo et al. [12] discussed a similarly paradoxical phenotype described for pathogenic variants of the transcription factor HNF4A and the potassium channel gene ABCC8, in which diabetes is preceded by transient congenital hyperinsulinism in some patients [39]. In accordance with the results of Iacovazzo et al. [12], the two phenotypes appeared to be mutually exclusive in the family described here, even though we cannot comment on the glycemic status of the two patients prior to developing insulinomatosis. All mutation-positive patients with available glycemic status had insulinomatosis or diabetes/impaired glucose tolerance. The pedigree (Figure 1) clearly confirms the previously reported [14] autosomal dominant mode of inheritance as well as the high penetrance of the disease (insulinomatosis or mild impairment of glucose metabolism). Family members with hyperglycemia had a mean BMI of 27.2 kg/m 2 (excluding one outlier (V2) with BMI of 38), with a slight preponderance of the male gender. Diabetes was managed through diet and in one case by insulin administration (III1). There were no significant diabetic complications reported by family members. In contrast to the family members with hyperglycemia, both patients with insulinomatosis were females, which is consistent with the findings of Iacovazzo et al. [12] demonstrating that insulinomatosis is predominant in females. Additionally, sporadic insulinomas are more common in females too (male-to-female ratio of 1:1.4).
Several mechanisms are proposed by Iacovazzo et al. [12] as possible explanations for this phenomenon. In vitro, the treatment of human β-cells and human insulinomas with estrogens has been shown to enhance proliferation and insulin secretion [40]. Furthermore, during pregnancy, an expansion of β-cell mass induced by prolactin and placental lactogen signaling mediated by the prolactin receptor (PRLR) is observed. PRLR was significantly downregulated and the PRLR promoter was shown to be directly activated by MAFA in luciferase reporter assays in both MAFA knockout islets and MIN6 β-cells with siRNA-mediated knockdown of MAFA [41]. Therefore, estrogens and prolactin could potentially promote β-cell proliferation, predisposing female carriers of MAFA mutations to insulinomatosis. Similarly to the family members with insulinomatosis reported by Iacovazzo et al. [12], one of our female insulinomatosis patients experienced the first hypoglycemic symptoms after pregnancy. She did not complain about any clinical problems during the pregnancy. The other affected sister, however, had never been pregnant and developed the first symptoms after puberty. In summary, it is striking that there is a female preponderance for developing insulinomatosis and a male preponderance for developing diabetes/impaired glucose tolerance in the three families with MAFA mutations reported so far. However, the precise mechanisms that determine the development of either insulinomatosis or diabetes remain unknown, and it is not ruled out that diabetes and insulinomatosis may develop consecutively. Furthermore, it is not excluded that the PROX1 variant detected in our family acts as a modifier in this regard.
In clinical practice, diagnosis and therapy of familial insulinomatosis remains challenging. In large studies, insulinomatosis is responsible for less than 5% of all patients with hyperinsulinemic hypoglycemia [2,10]. There are no distinct clinical or biochemical characteristics that distinguish solitary insulinoma from multifocal insulinomatosis. Multiple insulin-secreting microtumors are present in affected patients, distributed throughout the entire pancreas and only detectable by histological examination. Even when utilizing modern functional imaging modalities such as Glucagon-like peptide-1 receptor (GLP-1R) imaging, only few macrotumors with a mean size of 0.5-1 cm can be detected [42,43]. Thus, the clinical course of the disease is still characterized by early recurrent hypoglycemia requiring repeated surgical intervention. All documented clinical cases had persistent or recurrent disease after the initial surgical intervention. Only the histological examination of a larger, representative pancreatic specimen confirmed the diagnosis of insulinomatosis and distinguished it from multifocal NET with MEN1 mutations or nesidioblastosis. The SACI test should be considered in cases where noninvasive imaging modalities are negative or when there is evidence of a multifocal disease, as it can distinguish solitary insulinoma from multifocal/diffuse diseases with sensitivity and specificity of 95-100% and 90%, respectively, and determine the extent of surgical intervention [44,45].
The mean age at the diagnosis of insulinomatosis in our family was 39.5 years, being almost exactly the same as reported in the families by Iacovazzo et al. [12] (39.4 ± 13.1 years). Unfortunately, we were unable to determine the exact age at the diagnosis of diabetes. However, the recorded past medical history of the living family members indicated that the initial presentation of diabetes in our family occurred between 29 (V5) and 50 years of age (compared to 38.4 ± 16.4 years as reported by Iacovazzo et al. [12]). Therefore, most patients with MAFA mutations show clinical features of insulinomatosis or diabetes already in young adulthood. In contrast, the youngest patients with MAFA mutations and diagnosed diabetes or insulinomatosis were 11 and 18 years old, respectively [12]. We, therefore, recommend that asymptomatic mutation carriers should be monitored for diabetes/impaired glucose tolerance beginning in adolescence and that these patients should receive regular clinical screening as well as information/counselling, especially for signs and symptoms of neuroglycopenia.
The findings of this study have to be seen in light of some limitations. A more detailed characterization of the mild hyperglycemia observed in four variant carriers ( Figure 1) could provide further insights in the dynamics of the glucose metabolism in these patients. The affected family members consented to a one-time examination (including blood glucose and HbA1c), but not to an oral glucose tolerance test. Furthermore, it would be valuable to assess the glycemic status of the variant carriers in generation VI (genotyping by buccal swab). The parents were educated about potential symptoms of hypoglycemia and hyperglycemia, but they declined blood sampling and functional testing of the asymptomatic children at that time.
While it is clear that Thr57Arg causes a phosphorylation defect, which increases protein stability, and while Thr57Arg probably increases MAFA-mediated transactivation capacity of the beta cell, the consequences for MAFA's transactivation activity per molecule remain largely unknown. Using the MAFA Thr57Ala construct, transactivation activity has already been studied extensively by others. As the result was clearly dependent on cell type, differentiation, and embryonic or neoplastic origin of the cell model, it did not seem reasonable to repeat the expression studies on another (or the same) cell models only replacing Thr57Ala by Thr57Arg. Extensive further studies are needed to elucidate the complex regulation of MAFA's transactivation activity.

Conclusions
We define a second MAFA variant, p.Thr57Arg, as the cause of familial insulinomatosis by linking genetic, bioinformatic, clinical, biochemical, and family data to functional in vitro analyses from the literature. We, thus, confirm MAFA phosphorylation defects as an important cause of a hereditary syndrome, which is characterized by insulinomatosis and/or mild, MODY-like hyperglycemia. This study extends the pathophysiological and diagnostic disease concept and verifies the high penetrance and autosomal dominant inheritance pattern of familial insulinomatosis.