Identification of the matricellular protein Fibulin-5 as a target molecule of glucokinase-mediated calcineurin/NFAT signaling in pancreatic islets

Glucokinase-mediated glucose signaling induces insulin secretion, proliferation, and apoptosis in pancreatic β-cells. However, the precise molecular mechanisms underlying these processes are not clearly understood. Here, we demonstrated that glucokinase activation using a glucokinase activator (GKA) significantly upregulated the expression of Fibulin-5 (Fbln5), a matricellular protein involved in matrix-cell signaling, in isolated mouse islets. The islet Fbln5 expression was induced by ambient glucose in a time- and dose-dependent manner and further enhanced by high-fat diet or the deletion of insulin receptor substrate 2 (IRS-2), whereas the GKA-induced increase in Fbln5 expression was diminished in Irs-2-deficient islets. GKA-induced Fbln5 upregulation in the islets was blunted by a glucokinase inhibitor, KATP channel opener, Ca2+ channel blocker and calcineurin inhibitor, while it was augmented by harmine, a dual-specificity tyrosine phosphorylation-regulated kinase (DYRK) 1 A inhibitor. Although deletion of Fbln5 in mice had no significant effects on the glucose tolerance or β-cell functions, adenovirus-mediated Fbln5 overexpression increased glucose-stimulated insulin secretion in INS-1 rat insulinoma cells. Since the islet Fbln5 expression is regulated through a glucokinase/KATP channel/calcineurin/nuclear factor of activated T cells (NFAT) pathway crucial for the maintenance of β-cell functions, further investigation of Fbln5 functions in the islets is warranted.


Results
Glucokinase activation induced Fbln5 expression in the pancreatic islets. At first, we identified by gene expression microarray analysis (GSE41248), that stimulation of mouse pancreatic islets with a GKA for 24 hours induced Fbln5 expression in the islets (12.6-fold enhanced expression as compared to that in the vehicle control; p = 0.0043) 12 . To validate this upregulation of Fbln5 expression by treatment with a GKA in mouse pancreatic islets, we investigated Fbln5 mRNA expression in isolated islets from C57BL/6 J mice. Consistent with the results of the microarray analysis, the Fbln5 mRNA expression in the isolated islets was significantly increased, in a time-dependent manner, by treatment with a GKA (Fig. 1a). Ambient glucose also induced Fbln5 expression in the islets in a concentration-dependent manner (Fig. 1b). We detected FBLN5 protein expression in the wild-type mouse islets, as well as in INS-1 rat insulinoma cell line ( Fig. 1c and d) but not in the Fbln5-deficient (Fbln5 −/− ) islets (Fig. 1c). The treatment with a GKA also increased FBLN5 protein expression levels in INS-1 cells (Fig. 1d). Moreover, in glucokinase hetero-deficient (Gck +/− ) mouse islets, GKA-stimulated Fbln5 mRNA expression levels were reduced as compared to those in the islets from wild-type mice (Fig. 1e). No difference was detected in Fbln5 mRNA expression levels between vehicle-treated Gck +/− islets and the wild-type islets (p = 0.357) (Fig. 1e). These results suggest that Fbln5 expression is induced by glucokinase activation in the pancreatic islets. Furthermore, the GKA-induced increase in Fbln5 expression was more pronounced in the islets of mice reared on a high-fat diet for 20 weeks than in the islets of standard chow-fed mice (Fig. 1f), although there were no significant differences between the vehicle-treated islets from standard chow-fed and high-fat diet-fed mice (p = 0.24), consistent with the report that glucokinase-mediated signaling in the β-cells is activated by a high-fat diet 8,29 . In contrast, in insulin receptor substrate 2 (IRS-2)-deficient (Irs-2 −/− ) mouse islets, basal Fbln5 expression was significantly increased compared with those of wild-type mice (Fig. 1g). However, the response of Fbln5 induction to GKA was almost abolished in Irs-2 −/− mouse islets (Fig. 1g). It may also explain the more pronounced upregulation of islet Fbln5 expression in high-fat diet-fed mice than in normal chow-fed mice, as GKA is known to induce IRS-2 expression in the β-cells of mice reared on a high-fat diet 8 . The lack of Fbln5 induction in Irs-2 −/− islets suggests that IRS-2 is involved in the GKA-induced upregulation of islet Fbln5 expression. Moreover, we found that Fbln5 was strongly expressed in the islets of 2-week-old pre-weaning mice, the expression level decreasing by 6 or 12 weeks of age (Fig. 1h). This expression pattern of Fbln5 is consistent with the expression of the proliferation marker Ki67 in the islets (Fig. 1i).
Glucokinase/K ATP channel/calcineurin/NFAT signaling is required for glucose-mediated Fbln5 expression in islets. We next assessed the signaling pathways underlying the GKA-induced upregulation of Fbln5 in the pancreatic islets. Treatment with D-mannoheptulose, a specific inhibitor of glucokinase, completely abolished the GKA-induced upregulation of Fbln5 in the pancreatic islets (Fig. 2a). In addition, treatment with diazoxide, a K ATP channel (ATP-sensitive potassium channel) opener, also suppressed the GKA-induced elevation of Fbln5 expression in the islets (Fig. 2b). Treatment with OSI-906, a dual insulin and IGF-1 receptor inhibitor, did not reduce the Fbln5 induction by GKA, but enhanced it (Fig. 2c). These results imply that an influx of Ca 2+ into the β-cells via depolarization of the plasma membrane accompanied by the closure of K ATP channel, and not the autocrine action of insulin, is involved in the GKA-induced upregulation of Fbln5 in the pancreatic islets.
Calcineurin is activated in an intracellular Ca 2+ -dependent manner 30 , leading to NFAT activation by dephosphorylation and subsequent translocation of NFAT from the cytosol to the nucleus 31 . Glucose-induced regulation of Irs-2 expression has been reported to be mediated via this Ca 2+ /calcineurin/NFAT signaling in the pancreatic β-cells 32 . Hence, we evaluated the effects of a Ca 2+ channel blocker, a calcineurin inhibitor, and a DYRK1A inhibitor on the upregulation of Fbln5 in the islets treated with a GKA. Blockade of the L-type voltage-dependent Ca 2+ channels (L-type VDCCs) with nifedipine in isolated mouse islets abrogated the GKA-induced increase in Fbln5 expression in the islets (Fig. 2d). Moreover, treatment with FK506, which specifically inhibits calcineurin activity, also almost completely abolished the GKA-induced increase in Fbln5 expression in the pancreatic islets (Fig. 2d). Dual-specificity tyrosine phosphorylation-regulated kinases (DYRKs), including DYRK1A, inactivate the NFAT1 proteins by phosphorylating its SP-3 motif 33 . Notably, harmine, a DYRK1A inhibitor, enhanced the Fbln5 expression induced by treatment with a GKA (Fig. 2e). The effect of harmine on the increment in Fbln5 expression in islets was blunted in the presence of FK506 (Fig. 2f). These results suggest that the transcriptional regulation of Fbln5 in the islets is mediated by glucose signaling and downstream Ca 2+ /calcineurin/NFAT signaling.
Fbln5 −/− mice exhibited normal glucose tolerance and normal glucose-stimulated insulin secretion and β-cell proliferation evoked by GKA. To investigate the role of Fbln5 in glucose metabolism and insulin secretion, we used 8-to 12-week-old Fbln5 knockout (Fbln5 −/− ) mice 16 to evaluate whether Fbln5 deletion may influence glucose homeostasis in vivo. Fbln5 −/− mice showed normal glucose tolerance and comparable insulin secretion during an oral glucose tolerance test ( Fig. 3a and b). No significant difference in glucose-stimulated insulin secretion was observed between islets isolated from Fbln5 −/− mice and wild-type mice (Fig. 3c). These results imply that Fbln5 has no effect on insulin secretion in healthy young adult mice.
We next assessed the β-cell mass in the 8-to 12-week-old Fbln5 −/− mice. No significant differences in the islet morphology and the β-cells area relative to the total pancreatic area were observed between the wild-type mice and Fbln5 −/− mice ( Fig. 3d and e). Furthermore, we evaluated the GKA-induced β-cell proliferation activity in Fbln5 −/− and wild-type islets. Treatment with GKA for 48 hours markedly increased the EdU-incorporated proliferating insulin-positive β-cells to a similar extent in the islets isolated from both genotypes of mice ( Fig. 3f and g). On the other hand, the fluorescent intensity of insulin was significantly increased in the GKA-treated islets compared with the vehicle-treated islets in wild-type mice, but not in Fbln5 −/− mice (see Supplementary  Fig. S1a). This result in islets from wild-type mice is consistent with the observation that glucokinase activation enhances insulin gene expression and insulin secretion in β-cells 12, 34 . However, GKA-induced insulin secretion was not decreased in Fbln5 −/− islets compared with wild-type islets (see Supplementary Fig. S1b). Insulin content in GKA-treated Fbln5 −/− islets showed a tendency to be decreased compared to wild-type islets, but it did not reach statistical significance (see Supplementary Fig. S1b). Thus, Fbln5 is not required for the development and maintenance of β-cell function or proliferation.
In immunohistochemical analysis of paraffin-embedded endocrine pancreatic tissue from 8-week-old wild-type mice and Fbln5 −/− mice, FBLN5 was seemed to be around CD34 (endothelial marker) -positive interstitial tissue, but not in β-cells, α-cells, or δ-cells in the islets (see Supplementary Fig. S2). Next, fetal pancreatic tissue paraffin sections from wild-type mice at the age of embryonic day 15 were immunostained for FBLN5. The area and intensity of the FBLN5 signal in the islets seemed to be more abundant compared with those in adult mice (see Supplementary Fig. S3a). Then, we used non-paraffinized cultured islets from 8-week-old wild-type mice for immunostaining. Notably, FBLN5-positive β-cells were detectable in non-paraffinized adult wild-type islets (see Supplementary Fig. S3b). In addition, FBLN5 is observed at cytoplasmic granular structures in INS-1 cells, as shown in Supplementary Fig. S3c, suggesting that FBLN5 is also expressed in β-cells.  (Fig. 4a and b). The cells overexpressing Fbln5 showed enhanced insulin secretion in the presence of 11.1 mmol/L glucose as compared to the control cells (1.6-fold, p = 0.017), although basal insulin secretion was not significantly different between the Ad-Fbln5-and Ad-GFP-infected INS-1 cells (Fig. 4c). The effects of Fbln5 overexpression on the cell proliferation activity was evaluated by measuring the EdU incorporation and Ki67 expression in Ad-Fbln5-infected INS-1 cells. We were almost able to ignore the GFP-signals when we adjusted the gain of signals according to the fluorescent intensity of EdU. (Fig. 4d left panel). The ratio of EdU-incorporated proliferating INS-1 cells to the total count of INS-1 cells tended to be decreased in the Ad-Fbln5-infected cells as compared with that in the control cells (Fig. 4d). In addition, Ad-Fbln5-infected INS-1 cells showed significant reduction in the Ki67 expression (Fig. 4e). These results indicate that overexpression of Fbln5 enhances insulin secretion whereas decreases cell proliferation in β-cells.

Discussion
In this study, we showed the glucose signaling-induced transcriptional regulation of Fbln5 in pancreatic islets, which is mediated by glucose metabolism via glucokinase and downstream Ca 2+ /calcineurin/NFAT signaling pathway (Fig. 5).
Because β-cells are exposed to high ambient glucose concentrations under the diabetic condition, glucokinase, which acts as a glucose sensor, transmits the impact of the hyperglycemia to the β-cells. In the pancreatic islets, glucokinase is mainly expressed in the β-cells, with very low levels of expression observed in the α-cells (unpublished data). We confirmed GKA-induced increase in FBLN5 expression in INS-1 cells. Immunohistochemical analysis of INS-1 cells also supported that FBLN5 is expressed in the β-cells. However, further investigation using more specific antibody is needed to clarify the localization of FBLN5 since we observed FBLN5 signal not only in cytoplasm but also in the nucleus in INS-1 cells. FBLN5 immunofluorescence from paraffin-embedded tissue was mainly detected in non-β-cells tissue in the islets. FBLN5 is reported to be deposited on microfibrils during the development of mature elastic fiber 26 . Co-staining FBLN5 with CD34 in islets from paraffin-embedded specimens indicated that FBLN5 is strongly expressed in endothelial cells or small vessels in islets, consistent with the previous reports, which showed FBLN5 secretion from vascular smooth muscle cells or endothelial cells 35,36 . FBLN5 was also detectable in β-cells in the non-paraffinized cultured islets. It is therefore possible that FBLN5 in β-cells was lost or masked in the process of paraffin embedding or deparaffinization. Since dual inhibition of insulin and IGF-1 receptor with OSI-906 did not abrogate GKA-induced Fbln5 upregulation in the islets, it is unlikely that this upregulation is mediated by an autocrine action of insulin on the insulin receptor. Because FBLN5 is a secreted protein, islet-derived FBLN5 might be deposited outside of the islets and play a role in non-islet tissue functions.
Our data showed that GKA-induced upregulation of Fbln5 was more pronounced in islets isolated from obese mice reared on a high-fat diet than in the islets of control mice fed normal chow. This could be explained by the involvement of glucokinase in the compensatory β-cell hyperplasia induced by a high-fat diet 8,29 . Consistent with this notion, we found that Irs-2 deletion increased basal Fbln5 expression and attenuated the GKA-induced upregulation of Fbln5 in the isolated islets. Chronic hyperglycemia in Irs-2 −/− mouse may cause the elevation in Fbln5 in the islets at the basal state. Other factors that are related to insulin resistance with Irs-2 deletion in mice can possibly be involved in this basal elevation. Glucose-induced transcriptional regulation of Irs-2 gene expression in the β-cells is mediated by the Ca 2+ /calcineurin/NFAT pathway 32 , which is involved in β-cell proliferation in mice and humans 37,38 . In addition, the DYRK1A inhibitor has been demonstrated to enhance β-cell proliferation in mice 39,40 . It is also reported that GKA-induced increase of the mRNA expressions of Nfatc1 and its downstream genes are involved in β-cell maturation and β-cell proliferation in neonatal islets 38 . A recent study showed that glucose-induced mouse pancreatic β-cell proliferation is mediated via IRS-2, MTOR and cyclin D2, but not by the insulin receptor 41 . We also found that the Fbln5 expression was higher in the islets harvested from pre-weaning mice, which showed robust β-cell proliferation as confirmed by the high Ki67 expression. Fbln5 is strongly expressed during embryogenesis and plays a role in tissue remodeling 25 . Therefore, Fbln5 could be a predictor for compensatory β-cell proliferation and remodeling of β-cell mass induced by activation of IRS-2 expression.
How does NFAT signaling regulate the transcriptional activity of Fbln5 ? Fbln5 expression is positively regulated via transforming growth factor β1 (TGF-β1) in fibroblasts or epithelial cells 25 . Calcineurin inhibitors induce the TGF-β receptor-triggered signaling cascade in the mesangial cells or kidney 42 . Hypoxia-inducible factor-1α (HIF-1α) is also a Fbln5-inducible factor in the endothelial cells 43 . HIF-1α expression is also reportedly regulated through calcineurin activity or dephosphorylation of RACK1 in mast cells 44,45 . We have identified NFAT consensus sequences in the 5-upstream region of the mouse Fbln5 gene at: −698 to −693 (AGGAAA), +386 to +391 (TGGAAA), +428 to +433 (TGGAAA), +591 to +596 (TGGAAA), and 4 other sites from the first transcription initiation site. Further analysis, including of the TGF-β and HIF-1α pathways, are needed to clarify the precise mechanism of Fbln5 transcription via NFAT in pancreatic islets.
Loss of systemic Fbln5 expression had no significant effects on the insulin secretion from the pancreatic islets or β-cell proliferation/expansion in young adult mice, suggesting that Fbln5 does not seem to be involved in β-cell Thus, further investigation of the pathway that mediate Fbln5 action on β-cell proliferation is required. These effects of Fbln5 on β-cell functions and β-cell proliferation might be explained by the distinct proliferative and functional state of the β-cells. A previous study showed that a high rate of insulin production suppressed β-cell proliferation because of increased ER stress, in a cell-autonomous manner 51 . On the other hand, genes involved in β-cell functions were suppressed when proliferation-related genes were upregulated in replicating β-cells 52 .
There is a report in the literature which suggests that another matricellular protein, SPARC, which is expressed in stromal cells within the islets, can regulate β-cell growth and survival by inhibiting growth factor responses 53 . Thus, the interactions between Fbln5 and pancreatic β-cell functions, which are still poorly understood, may represent novel molecular mechanisms involved in glucose metabolism and provide new insights for the treatment in diabetes. In summary, we demonstrated that expression of the matricellular protein Fbln5 is upregulated by high ambient glucose concentrations in the pancreatic islets though glucokinase-dependent glucose and downstream Ca 2+ / calcineurin/NFAT signaling. Further study of the regulation of islet Fbln5 expression is warranted, especially in relation to glucose signaling and proliferation of β-cells.

Methods
Animals and Animal Care. All the animal procedures were performed in accordance with the guidelines of the Animal Care Committee of Yokohama City University. The protocol was approved by the Yokohama City University Institutional Animal Care and Use Committee (IACUC) (Permit Number: F-A-16-026). C57BL/6 J mice were purchased from Jackson. We backcrossed Fbln5 knockout (Fbln5 −/− ) mice 16,19 with C57BL/6 J mice more than 10 times. Both Fbln5 −/− mice and wild-type littermates were fed a standard chow (MF, Oriental Yeast, Tokyo, Japan) or a high-fat diet (Clea Japan, Tokyo, Japan). All the experiments were conducted on male littermates. Animal housing rooms were maintained at a constant room temperature (25 °C) and a 12-hour light (7:00 a.m.) /dark (7:00 p.m.) cycle.
Adenovirus. Fbln5-overexpressing recombinant adenovirus (Ad-Fbln5) 18 and GFP-expressing control adenovirus (Ad-GFP) were used for the experiments at a multiplicity of infection of 50 viruses per cell. In brief, the FLAG-tagged full-length rat Fbln5 was inserted in an adenoviral vector (pACCMVpLpA(−) loxP-SSP). Viruses were generated by transfection into the Human Embryo Kidney 293 (HEK293) cell line.
Islet isolation and culture. Isolation of islets from mice was conducted using collagenase, as described in a previous report 54  Oral glucose tolerance test. All the mice were denied access to food for 14-16 hours before the oral glucose tolerance test (OGTT) and then orally loaded with glucose at 1.5 mg/g body weight. Blood glucose levels and serum insulin levels were determined using Glutest Neo Super (Sanwa Chemical Co. Kanagawa, Japan) and an insulin ELISA kit (Morinaga Institute of Biological Science, Yokohama, Japan), respectively.

Glucose-stimulated insulin secretion in isolated islets and INS-1 cells. Ten islets isolated from
Fbln5 −/− mice and wild-type mice were incubated at 37 °C for 1.5 hours in Krebs-Ringer bicarbonate buffer containing 2.8, 11.1 or 22.2 mmol/L of glucose. When examining the effect of Fbln5 deficiency on GKA-induced insulin secretion, islets were incubated at 37 °C for 1.5 hours in Krebs-Ringer bicarbonate buffer containing 2.8 mmol/L glucose with or without 30 µmol/L of GKA CpdA, or 11.1 mmol/L glucose without GKA CpdA. For measuring insulin content, islets were extracted with acid ethanol. INS-1 cells were infected with adenovirus (Ad-GFP or Ad-Fbln5) and cultured for 48 hours. Subsequently, the cells were incubated at 37 °C for 2 hours in Krebs-Ringer bicarbonate buffer containing 2.8, 11.1 or 22.2 mmol/L of glucose. Then, the insulin concentration in the assay buffer and insulin content was measured with an insulin ELISA kit.
Histological analysis. Pancreatic tissue sections from embryonic day15 and 8-week-old Fbln5 −/− mice and wild-type mice were analyzed after formalin fixation and paraffin embedding. For non-paraffinized tissue staining, isolated islets from 8-week-old wild-type mice attached to 0.1%-gelatin-coated coverslips (Falcon) were analyzed after fixation with paraformaldehyde. Pancreatic islets isolated from 8-week-old wild-type mice were analyzed after fixation without paraffin embedding. The sections or attached islets on coverslips were immunostained with antibody directed against insulin (Santa Cruz Biotechnology), glucagon (Abcam), somatostatin (GeneTex), CD34 (Santa Cruz Biotechnology), or rabbit polyclonal anti-fibulin-5 (BSYN 1923; 1:100) 16 . FBLN5 signal was enhanced by tyramide signal amplification, using a TSA Fluorescein System (Perkin Elmer, NEL741001KT), in paraffin-embedding sections. Biotinylated secondary antibodies, a VECTASTAIN Elite ABC Kit, and a DAB Substrate Kit (Vector Laboratories) were used to examine the sections using bright-field microscopy to determine the β-cell mass, and Alexa Fluor 488-, 555-and 647-conjugated secondary antibodies (Invitrogen) were used for the fluorescence microscopy. Images were acquired using a BZ-9000 microscope (Keyence) or the FluoView FV1000-D confocal laser scanning microscope (Olympus). The proportion of the area of the pancreatic tissue occupied by β-cells was calculated using BIOREVO software (Keyence), as described previously 56 . The fluorescence levels of insulin in GKA-treated wild-type and Fbln5 −/− islets were determined using Image J software. All images, which were acquired under the same condition, were converted to gray scale. Then, we randomly selected 5 regions of separate islets in each group and measured fluorescence levels. The fluorescent intensity were normalized by the mean background fluorescence levels.