β-Cell-Specific Mafk Overexpression Impairs Pancreatic Endocrine Cell Development

The MAF family transcription factors are homologs of v-Maf, the oncogenic component of the avian retrovirus AS42. They are subdivided into 2 groups, small and large MAF proteins, according to their structure, function, and molecular size. MAFK is a member of the small MAF family and acts as a dominant negative form of large MAFs. In previous research we generated transgenic mice that overexpress MAFK in order to suppress the function of large MAF proteins in pancreatic β-cells. These mice developed hyperglycemia in adulthood due to impairment of glucose-stimulated insulin secretion. The aim of the current study is to examine the effects of β-cell-specific Mafk overexpression in endocrine cell development. The developing islets of Mafk-transgenic embryos appeared to be disorganized with an inversion of total numbers of insulin+ and glucagon+ cells due to reduced β-cell proliferation. Gene expression analysis by quantitative RT-PCR revealed decreased levels of β-cell-related genes whose expressions are known to be controlled by large MAF proteins. Additionally, these changes were accompanied with a significant increase in key β-cell transcription factors likely due to compensatory mechanisms that might have been activated in response to the β-cell loss. Finally, microarray comparison of gene expression profiles between wild-type and transgenic pancreata revealed alteration of some uncharacterized genes including Pcbd1, Fam132a, Cryba2, and Npy, which might play important roles during pancreatic endocrine development. Taken together, these results suggest that Mafk overexpression impairs endocrine development through a regulation of numerous β-cell-related genes. The microarray analysis provided a unique data set of differentially expressed genes that might contribute to a better understanding of the molecular basis that governs the development and function of endocrine pancreas.


Introduction
Diabetes mellitus is a group of metabolic diseases characterized by hyperglycemia resulting from defects in insulin secretion, insulin action, or both, which eventually leads to a series of complications in various organs [1]. Type 1 diabetes (T1D) results from the destruction of βcells by a β-cell-specific autoimmune reaction. In type 2 diabetes (T2D), the peripheral tissues are resistant to insulin action and the disease is often accompanied by obesity and hyperlipidemia. For decades, several approaches have been developed for the treatment of diabetes including insulin-secretion stimulants, improving insulin preparations, and islet transplantation, yet many unexamined avenues of research remain [2].
During pancreatic development, a subset of the pancreatic epithelial cells starts to express the proendocrine factor Ngn3 and gives rise to all types of endocrine cells [3][4][5][6]. The hormoneexpressing cells are produced during 2 sequential stages, the primary and secondary transitions. The primary transition begins before E13.5, and is characterized by an appearance of hormone+ cells that are not fully functional. During the secondary transition starting from around E13.5, the differentiating endocrine cells expand markedly, then migrate into mesenchyme, and eventually aggregate to form Langerhans islets [7]. The expression of a cascade of different transcription factors stimulates the differentiation into distinct endocrine lineages. αcell-related transcription factors include Mafb, Nkx2.2, Pax6, Foxa2, Pou3f4, and Arx, whereas β-cell differentiation is controlled by Pdx1, Mafb, Pax4, Pax6, Nkx2.2, and Nkx6.1 [8].
The MAF family transcription factors belong to the activator protein 1 (AP1) superfamily of basic leucine zipper (bZIP) proteins. It derives its name from v-Maf-the oncogenic component of the avian retrovirus AS42 that was originally isolated from chicken musculoaponeurotic fibrosarcoma [9]. The MAF family is subdivided into 2 groups according to their molecular size: the small MAF proteins comprising MAFG, MAFF, and MAFK; and large MAF proteins, including MAFA, MAFB, c-MAF, and NRL. All MAF proteins contain basic leucine zipper domain that allows DNA binding. Compared to the other bZIP proteins, MAF proteins can recognize a longer palindromic sequence of DNA (Maf-recognition element, MARE) [10][11][12]. Increasing numbers of studies on endocrine development reveal that the expression of large MAF proteins is tightly regulated in a spatiotemporal manner [13][14][15]. Mafb gene knockout (Mafb −/− ) mice show around a 50% reduction of αand β-cell numbers at E18. 5. In contrast, no developmental defects were observed in Mafa gene knockout (Mafa −/− ) mice [16,17]. Small MAF proteins are also found to display a complex expression pattern during embryogenesis [18]. They are able to form a homodimer or a heterodimer with other bZIP factors such as the cap'n'collar (CNC) family and play a role in many biological processes like hematopoiesis, neuronal function, and oxidative stress response [19][20][21][22]. Unlike the large MAF proteins, the small MAF proteins lack a transactivation domain and when they are expressed in large amounts, the homodimeric proteins compete with the binding of large MAF proteins to the cis-element of target genes at MARE sites, resulting in a dominant-negative effect [23].
Our previous studies demonstrated that β-cell-specific Mafk transgenic (Mafk-Tg) mice exhibited hyperglycemia due to an impaired insulin secretion during early postnatal life [24]. When these mice are crossed with Mafa −/− mice, the double mutants display destructive β-cell development and an overt diabetic phenotype with typical characteristics of human diabetic nephropathy [25]. The aim of the current study was to characterize and evaluate the impact of Mafk overexpression on the genetic pathways governing β-cell development using Mafk-Tg and Mafa −/− ;Mafk-Tg mice embryos. In the Mafk-Tg mutants we observed abnormalities in βcell development and islet morphogenesis along with a reduction of β-cell proliferation. In addition, we also performed a microarray analysis in order to investigate the factors that contribute to the phenotypic alteration in the Mafk-Tg mice at E15.5, to gain insights into the mechanisms controlling endocrine cell development and function.

Mice
The mice were maintained in specific pathogen-free conditions, in the Laboratory Animal Resource Center at the University of Tsukuba. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Committee on the Ethics of Animal Experiments of the University of Tsukuba (Permit Number: 14-049). All mice were euthanized with carbon dioxide gas, and all efforts were made to minimize suffering. The generations of Mafa knockout (Mafa −/− ) mice, transgenic mice expressing Mafk under the control of rat Insulin 1 promoter (Mafk-Tg), R26GRR mice, and Ins1-Cre25 mice have previously been described [16,24,26,27]. Double mutant Mafa −/− ;Mafk-Tg mice were generated by mating Mafa +/− females with Mafa +/− ;Mafk-Tg males. DNA extraction and embryo genotyping were performed by NaOH extraction methods from tails as previously described [16,24].

Measurement of total insulin contents
The whole pancreas was collected from embryos at both E15.5 and E18.5. The total insulin content was determined after extraction with acid-ethanol. Insulin levels were detected using a mouse insulin ELISA kit (Morinaga, Yokohama, Japan).

Luciferase assay
The Mafk-expression plasmid and RIPII-251 reporter plasmid have been previously described [24,29]. The Mafb cDNA was subcloned into the pcDNA3.1-FLAG expression vector, and these plasmids were transfected into NIH3T3 cells using FuGENE 6 transfection reagent (Roche, Indianapolis, IN, USA). The total amount of DNA was adjusted by cotransfection of pcDNA3.1-FLAG plasmid with Mafk-expression plasmids. Luciferase activities were determined by the Dual Luciferase Reporter Assay System (Promega) 48 hours after transfection.

Quantitative RT-PCR
The whole pancreas of E15.5 embryos were homogenized and total RNA isolated using the NucleoSpin RNA kit (Macherey-Nagel, Düeren, Germany). The cDNA was synthesized using a QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany). Quantitative PCR reactions were carried out using a Thermal Cycler Dice Real Time System (Takara) with a SYBR Green PCR Master Mix (Takara). Expression of Hprt was utilized to analyze the relative gene expression of other genes. All primer sequences are listed in Table 1.

Lineage tracing experiments
To test the possibility that β-cell specific Mafk overexpression induced β-cell transdifferentiation into α-cells during early stages of development, we crossed Mafk-Tg females with Ins1-Cre;R26GRR males, which express GFP ubiquitously before and tdsRed after Cre excision; thus we could label around 90% of the β-cell lineage [26]. The pancreata of Mafk-Tg;Ins1-Cre; R26GRR and control WT;Ins1-Cre;R26GRR mice were collected at both P0 and 4 weeks of age, fixed overnight in 4% PFA, dehydrated in sucrose and embedded in OCT compound (Sakura, Tokyo, Japan). Frozen sections were stained using anti-glucagon antibody in order to determine the cell fraction coexpressing both tdsRed and glucagon.

Microarray experiment
Total RNA was isolated from 3 pairs of WT and Mafk-Tg pancreata at E15.5 using the NucleoSpin RNA kit (Macherey-Nagel). The total RNA was used to synthesize cRNA using the Ambion WT Expression Kit (Life Technologies). Fragmentation and labeling of cDNA were performed using the GeneChip WT Terminal Labeling and Control Kit (Affymetrix, Santa Clara, CA, USA). The hybridization cocktail containing fragmented and biotin-labeled cDNAs was transferred into GeneChip MoGene-1_0-st-v1 cassettes (Affymetrix), which were incubated at 45°C inside a hybridization oven by rotating them at 60 rpm for 17 hours. The Gene-Chip arrays were then washed and developed using the Hybridization Wash and Stain Kit (Affymetrix) in a Fluidics Station 450 (Affymetrix). The GeneChip arrays were read using the GeneChip Scanner 3000 (Affymetrix) and the image files were generated using the GeneChip Command Console (Affymetrix). Normalization and probe set summarization were performed using the Affymetrix Expression Console software. CEL files and normalized data were deposited into the NCBI GEO repository under the accession number GSE62834.

Statistical analysis
Data were expressed as the means ± standard errors of the means and compared using an unpaired t test. Probability values of less than 0.05 were considered significant.

β-cell-specific Mafk overexpression resulted in impaired endocrine cell development and an abnormal islet structure
In order to study the impact of β-cell-specific Mafk overexpression on endocrine development, especially during the primary and secondary transitions, we performed immunohistochemical  We also stained pancreatic sections at postnatal day (P) 4, P21, and 5 weeks of age (5W). The islets from WT mice at all periods appeared with a clear central core of insulin+ cells surrounded by a peripheral layer of glucagon+ cells, whereas Mafa −/− mice developed an abnormal structure with many glucagon+ cells in the center of the islets in the early neonatal period [16]. The islets of Mafk-Tg mice were still displaying changes similar to their embryonic abnormal phenotype at P4 and P21, however at 5W the islet structure was apparently reverted to normal. As shown previously, in the islet of Mafa −/− ;Mafk-Tg mice, the destructive changes were more severe compared to Mafa −/− mice [25]. These results suggest that the expression of Mafa during the neonatal period compensates for the effect of Mafk overexpression in embryos, which is consistent with a previous report showing the functional significances of Mafa after birth [33]. Hereafter, our experiment was focused on the comparison of Mafk-Tg and control WT mice in each experiment.

MAFK inhibited the activation of insulin promoter
The ability of MAFK to suppress the activation of the insulin promoter was examined using luciferase assay. A reporter plasmid containing the luciferase gene under the control of the rat insulin 2 promoter (pGL2/RIPII-251) was used, as previously described [24] (Fig 4A). NIH3T3 cells were transfected with MAFK and MAFB expression plasmids with a reporter plasmid, and the luciferase activity was monitored 48 hours after transfection (Fig 4B). MAFB activated the insulin promoter more than 100-fold. In the presence of increased amounts of MAFK, the enhanced activity of insulin promoter by MAFB showed a dose-dependent reduction. These findings indicate that MAFK inhibited MAFB binding to the C-box of the insulin promoter. Chromatin immunoprecipitation (ChIP) using FLAG antibody against NMuMG cells treated with either pCAGIP-FLAG-MAFK or mock plasmids further confirmed these observations   Endocrine Development in β-Cell-Specific Mafk Transgenic Mouse (Fig 4C). The binding of MAFK directly to the insulin promoter was evident upon analysis of the immunoprecipitated protein-DNA complex with specific PCR primers.

No evidence of α-cell transdifferentiation in Mafk-Tg β-cells
The first observation of the reversion of the absolute number of insulin+ and glucagon+ cells in Mafk-Tg mice raised the possibility that β-cell-specific Mafk overexpression could induce their transdifferentiation to α-cells during endocrine development. To answer this question, we labeled the β-cell lineage using Ins1-Cre;R26GRR mice, which express GFP ubiquitously before and tdsRed exclusively in β cells after Cre recombination (Fig 5A). We collected 8 sections at 100-μm intervals apart from Mafk-Tg;Ins1-Cre;R26GRR and control WT;Ins1-Cre25;R26GRR (n = 3 per mouse) at P0 and 4 weeks of age and stained them with glucagon antibody in order to examine the adult α-cells that might be derived from the β-cell lineage. Although higher numbers of glucagon+ cells were detected in Mafk-Tg mice at 4 weeks of age, glucagon + tdsRed+ cells were observed in 0.52% of the total glucagon+ cells in Mafk-Tg mice (n = 763) and these cells were 0.66% of the total glucagon+ cells in the controls (n = 752) (Fig 5B). These observations suggest that the increased number of α-cells in Mafk-Tg mice might be due to other causes rather than β-cell transdifferentiation.

Mafk overexpression suppressed β-cell proliferation
To examine the reason for reduction in β-cell numbers during the prenatal and early postnatal periods, we performed immunohistochemical analysis using the cell cycle marker Ki-67 ( Fig  6A). Six different sections from WT and Mafk-Tg embryos (n = 3 per embryo) were collected at 100-μm intervals and stained using Ki-67 and insulin antibodies. The number of double positive cells was counted and divided by the total number of insulin+ cells to estimate the percentage of proliferating β-cells. The result showed that the percentage of proliferating β-cells was lower in Mafk-Tg mice (3.9 ± 0.31%) than in WT mice (10.3 ± 0.91%, P = 0.001) at E18.5 (Fig 6B). This finding was further examined using another cell cycle marker, pHH3 (Fig 6C). Quantitative comparison of the percentage of the cells that appeared as double positive for both pHH3 and insulin (2.2 ± 0.4% in Tg vs. 7.2 ± 0.6% in WT, P = 0.001) was in line with that obtained from Ki-67 staining, suggesting that β-cell proliferation is reduced or delayed in Mafk-Tg mice (Fig 6D). Impaired proliferation of blood cells induced by Mafk overexpression is also reported in vivo, suggesting that excess amounts of MAFK has an antiproliferative effect regardless of cell type [23,34].
We also examined the changes in α-cell proliferation using Ki-67 staining and found higher percentage of cells with double immunoreactivity for both glucagon and Ki-67 at E18.5 in Mafk-Tg (10.5 ± 0.9%) compared to WT embryos (6.7 ± 0.7%, P = 0.025)(n = 3 per embryo) (Fig 6E and 6F). This observation indicated that the increase in the α-cell numbers was mainly due to their proliferation rather than β-cell dedifferentiation.

Microarray analysis of pancreas from Mafk-Tg embryos revealed differential expression of islet-related genes
To identify genes potentially involved in endocrine development and function, we conducted unsupervised microarray analysis using total RNA from pancreata of WT and Mafk-Tg embryos collected at E15.5. qRT-PCR analyses, as shown in Fig 3, indicated that novel genes involved in β-cell functions are expected to show differential expression between genotypes. After data normalization and probe summarization, a list of 554 upregulated and 548 downregulated probe sets was generated (S1 Table). Due to the limited sample size (n = 3 per genotype) and the fact that there is a low abundance of endocrine cells within the pancreas during  Endocrine Development in β-Cell-Specific Mafk Transgenic Mouse this stage of development, the cutoff was set at 1.2-fold change [35,36]. Sixteen genes (namely Ins1, Ins2, Slc30a8, G6pc2, Sytl4, Gcg, Mafb, Pax4, Pax6, Rfx6, Pdx1, Foxa2, Ngn3, Neurod1, Nkx2.2 and Pou3f4) from a total of 22 genes (72.7%) that show significant difference (Fig 3) were found to be included in the up-and down-regulated gene list, suggesting that microarray can be used for further analysis. In order to identify any uncharacterized genes that might be involved in endocrine development and function, we chose several candidate genes for qRT-PCR analyses. For that experiment we used individual biological samples (n = 4-8 per experiment) to confirm the changes observed with the technical replicates that were used for microarray analysis. For qRT-PCR, we chose several transcripts that we thought to be potential factors involved in endocrine development and function, some of which had not been fully characterized (Fig 7). In addition to Ins1, we identified 3 novel candidate genes that showed a significant reduction of gene expression in Mafk-Tg, namely Npy, Cryba2, and Fam132a ( Fig  7A). Additionally, from the upregulated gene group, the expression of Tmem27, Chgb, Rbp4, Hnf1a and Pcbd1 was confirmed (Fig 7B). These results suggest that the gene expression profile of Mafk-Tg pancreata can provide a unique set of novel genes that possibly play various roles in endocrine development and function.

Discussion
In this study, we were able to detect insulin/glucagon double positive cells, which are considered to be endocrine progenitors, in Mafa −/− , Mafk-Tg, and Mafa −/− ;Mafk-Tg embryos as well as WT embryos at E12.5. Our findings indicate that neither Mafa deficiency nor β-cell-specific Mafk overexpression have any deleterious effect on the development of the early pancreatic progenitors. The islets of the Mafa −/− mice also showed no marked structural changes when compared to WT mice during the whole prenatal life, consistent with a previous study using Mafa Δpanc mice [17]. In contrast, as observed in Mafk-Tg and Mafa −/− ;Mafk-Tg mice, overexpression of Mafk impaired the islet morphogenesis at the late gestational stage, with up to 50% reduction in the number of β-cells and total insulin content. This finding corresponds with that of systemic Mafb-deficient embryos in regard to β-cell development [15]. The phenotype observed in Mafk-Tg mice probably represents a part of conditional deletion of Mafb in β-cells. On this point, we could not rule out the direct role of overexpressed MAFK on other differentiation factors in a large MAF independent manner. However, the increasing expression of Mafa in neonatal β-cells probably restored the islet structure in Mafk-Tg mice by 5 weeks of age, suggesting that Mafk overexpression primarily blocked the large MAF function in embryos [37]. This idea is also supported by a recent report showing the opposing effect of MAFA and small MAFs on the insulin promoter [38].
Mafk-deficient mice do not exhibited any abnormalities [19]. β-cell specific transgenic mice, which express a dominant negative form of MAFK (DN-MAFK), show normal glucose tolerance under normal chow diet [38]. In contrast, under high fat diet conditions, DN-MAFK overexpression improves insulin secretion and glucose metabolism. Together with the observation that high fat diet feeding increases small MAF expression in β-cells, MAFK probably downregulates the expression of genes regulated by MAFA under stressed conditions partly due to the blocking of MAFA binding to MARE on the target [38].
Several lines of evidence suggest the capacity of terminally differentiated cells in the endocrine pancreas to lose their integrity and thereby adopt another cell identity [39]. Forced expression of Arx in mature β-cells can induce their conversion into a glucagon-positive α-cell phenotype [40]. Moreover β-cells convert into α-cells following deletion of Dnmt1 or FoxO1 [41] [42]. Based on these observations, we hypothesized that Mafk overexpression might trigger β-cell transdifferentiation/conversation into α-cells. However, as shown in Fig 5, we were Endocrine Development in β-Cell-Specific Mafk Transgenic Mouse not able to find any cells coexpressing tdsRED and glucagon using β-cell lineage tracing, indicating that the increase in α-cells of Mafk-Tg mice is likely not to be due to βto α-cell transdifferentiation, but rather due to the increased proliferation of α-cells. Recently, it is proposed that derepression of Mafb in β-cells activates βto α-cell reprogramming in the absence of Pdx1 or Mafa [43] [44]. Because overexpressed MAFK blocks MAFB function and probably does not initiate the reprogramming cascade in Mafk-Tg β-cells, the increase in α-cell numbers as well as α-cell transcription factors might be due to a loss of insulin action [45]. Alternatively, disruption of key β-cell transcription factors (Mafs or Pdx1) during the early postnatal period does not cause βto αtransdifferentiation (this manuscript and [46]), which occurs after 4 weeks of age [43] [44].
Finally, our analyses of microarray and subsequent qRT-PCR revealed that the downregulated gene group included a set of important transcripts for β-cell function. Npy, Fam132a, and Cryba2, as well as Ins1 were encountered as potential β-cell-related genes. NPY (Neuropeptide Y) has pleiotropic functions in various tissues including hypothalamus, autonomic nervous system, and adipose tissue. In pancreas, its expression is observed in β-cells during the secondary transition to neonatal period [47,48]. A previous report showing that NPY treatment of mouse islets significantly enhanced β-cell replication supports our finding of a reduction of βcell proliferation in Mafk-Tg mice [49]. Fam132a (also called Adipolin) is a novel adipokine associated with roles in glycemic control and insulin sensitization [50]. In addition to adipose tissue, Fam132a is thought to be secreted from mouse islets, according to the T1D base (https://www.t1dbase.org/page/Welcome/display). Notably, Cryba2 is identified as an enriched gene in developing and adult pancreas and its expression is affected by Ngn3-deficiency during development, although no clear phenotype in pancreas of ENU-induced Cryba2 mutants has been reported [51][52][53].
From the upregulated gene group, TMEM27 plays a role in controlling insulin exocytosis by regulating the soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complex assembly [54,55]. Retinol binding protein 4 (RBP4), which is a principle carrier of blood retinol, contributes insulin resistance in mice and humans [56]. Interestingly, both Tmem27 and Rbp4 is upregulated in pancreas from Mafb -/during late embryonic period, implying that Mafk-Tg mice and Mafb -/mice share common gene expression patterns as well as phenotypic similarities [17]. Chromogranin-B (CHGB) is a secretory glycoprotein co-stored with insulin and is found to control the rapid initial phase of insulin secretion [57]. Hepatocyte nuclear factor 1 alpha (HNF1A) controls many genes related to β-cell differentiation, and gene mutations are the most common cause of maturity-onset diabetes of the young [58]. Pterin-4-alpha-carbinolamine dehydratase (PCBD1) is a novel protein that acts as a cofactor for HNF1Adependent transcription protein, and it is reported that PCBD1 mutations cause early-onset nonautoimmune diabetes with features similar to dominantly inherited HNF1A-diabetes [59].
In conclusion, β-cell-specific Mafk overexpression resulted in impairment in endocrine development through alteration of the expression of many important genes for endocrine development and function. Since Mafk overexpression can mimic the targeted disruption of many genes containing MARE sites in their regulatory regions to which MAFK homodimers can bind, our microarray analysis of Mafk-Tg embryos provides a unique data set for investigating novel factors that might have possible roles in β-cell development, function, and survival.
Supporting Information S1 Table. List of the up-and down-regulated genes in the pancreas of Mafk-Tg embryos at E15.5. (XLSX)