SEL1L Regulates Adhesion, Proliferation and Secretion of Insulin by Affecting Integrin Signaling

SEL1L, a component of the endoplasmic reticulum associated degradation (ERAD) pathway, has been reported to regulate the (i) differentiation of the pancreatic endocrine and exocrine tissue during the second transition of mouse embryonic development, (ii) neural stem cell self-renewal and lineage commitment and (iii) cell cycle progression through regulation of genes related to cell-matrix interaction. Here we show that in the pancreas the expression of SEL1L is developmentally regulated, such that it is readily detected in developing islet cells and in nascent acinar clusters adjacent to basement membranes, and becomes progressively restricted to the islets of Langherans in post-natal life. This peculiar expression pattern and the presence of two inverse RGD motifs in the fibronectin type II domain of SEL1L protein indicate a possible interaction with cell adhesion molecules to regulate islets architecture. Co-immunoprecipitation studies revealed SEL1L and ß1-integrin interaction and, down-modulation of SEL1L in pancreatic ß-cells, negatively influences both cell adhesion on selected matrix components and cell proliferation likely due to altered ERK signaling. Furthermore, the absence of SEL1L protein strongly inhibits glucose-stimulated insulin secretion in isolated mouse pancreatic islets unveiling an important role of SEL1L in insulin trafficking. This phenotype can be rescued by the ectopic expression of the ß1-integrin subunit confirming the close interaction of these two proteins in regulating the cross-talk between extracellular matrix and insulin signalling to create a favourable micro-environment for ß-cell development and function.


Introduction
SEL1L encodes an endoplasmic reticulum transmembrane protein with a complex structure implicated in a number of cellular functions [1][2][3][4] mostly associated with the endoplasmic reticulum associated degradation (ERAD) and unfolded protein response (UPR) pathways [5][6][7]. SEL1L established function is to complex with the E3 ligase HRD1 to regulate degradation and turnover of luminal and membrane proteins [8]. SEL1L is located close to the D14S67 locus on the chromosome 14q24 [9,10] hypothesized to be a candidate region for the type I diabetes mellitus (T1DM) [11]. However no evidence for SEL1L as candidate gene for IDDM11 was found [12,13]. Interestingly, it was suggested that mutations in SEL1L could influence MODY onset and/or progression [14]. To date, six MODY genes have been identified (glucokinase, hepatocyte nuclear factors HNF-1a, HNF-4a and HNF-1b, insulin promoter factors IPF1 and NEUROD1), all associated with islet cells development, maintenance of differentiation, and endocrine secretory function [15][16][17][18]. Interestingly, HNF-1a and HNF-4a have been reported to bind the SEL1L promoter, supporting its involvement in pancreas development [19]. It was reported that mice homozygous for a gene trap mutation in Sel1l developed systemic ER stress and died during mid-gestation [20] like the Hrd1 knock-out mouse model [21] but, in addition, Sel1l mutants displayed severe growth retardation and impaired differentiation of pancreatic and neural epithelial cells, suggesting an HRD1-indipendent function(s). Mice carrying one functional allele, revealed an increased susceptibility to diet-induced hyperglycemia and reduced b-cell mass [22,23], and its depletion in bTC3 cells resulted in vitro growth arrest and cell death [24]. All together these results suggest that SEL1L could play a significant role in regulating ß-cell function and growth. To date, a number of mechanisms have being proposed to explain the progressive loss of b-cell function that eventually leads to T2DM. Among them, ER-stress responses induced by chronically elevated circulating levels of glucose and lipids, collectively known as glucolipotoxicity [25], are centain to have a detrimental impact on b-cell function, and possible b-cell death [26,27]. More recently, evidence has been provided in support of more complex mechanisms of progressive impairement of b-cell function that involves a loss of b-cell identity rather than death by apoptosis, which leads to b-cell dedifferentiation into embryonic-like endocrine progenitors and interconversion into a-cell [28].
Causative mutations in SEL1L are very rare, however polymorphic variants have been reported: one associated with pancreatic cancer [29], a second with persistent hyperinsulinemic hypoglycemia of infancy [30] and a third in progressive childhood ataxia [31]. Of particular interest is the late evolutionary addition of the Fibronectin type II domain to the gene, increasing the protein functional complexity by contributing to cell-matrix interactions [32]. This domain is usually found in extracellular matrix fibronectin and in extra cytoplasmic regions of membrane associated-proteins and are thought to be involved in protein cell surface localization and activation through collagen-b1 integrin binding [33,34]. Integrin engagement is a key regulator of pancreatic b-cell function, induces ERK-dependent insulin secretion and promotes epithelial to mesenchymal transition (EMT) by regulating the WNT/SMAD pathway [35][36][37]. More recently, b1 integrin-dependent signaling has also been implicated in the regulation of embryonic and perinatal ß-cell expansion [38]. Moreover, SEL1L has been reported to play a key role in the improvement of pancreatic plasticity being involved in the combined action of several pathways such as WNT, TGF-b, NOTCH and MAPK [39].
Here we show that SEL1L down-modulation in pancreatic bcells negatively impacts on cell adhesion and proliferation, and inhibits glucose-stimulated insulin secretion by affecting ERK signaling. We also show that this phenotype can be rescued by overexpressing b1 integrin subunit and restoring ERK activation level.
Collectively, our results support a possible function of SEL1L in regulating the cross-talk between integrin signaling and insulin secretion.

Cell Lines, Culture Conditions and Transfections
CFPAC-1 human ductal adenocarcinoma cells (ATCC) were grown in Iscove's modified Dulbecco's medium (Life Technologies) supplemented with 10% fetal bovine serum and 2 mM L-Glutamine. MIN6 cells (obtained from Prof. Paolo Meda, University of Geneva, Switzerland [40] originally from Dr Miyazaki [41]) were grown in DMEM-high glucose medium with 2 g/L sodium bicarbonate, supplemented with 10% FBS and 70 mM of b-Mercaptoethanol.
Islets were isolated by intraductal injection of 0.5 mg/ml liberase and purified on a Ficoll gradient [42]. Islets were cultured overnight in RPMI-10% FCS and handpicked before being further processed.

Imunofluorescence and Immunohistochemistry
Dual immuno fluorescent labeling was performed on 4-mm paraffin sections prepared from adult (3-month-old) and fetal (E16.5) of C57BL/6J mice pancreata. Use of animal subjects was carried out in strict accordance with the recommendations for the Care and Use of Laboratory Animals of the National Institute of Health (Italy). The protocol was approved by the Committee on the ethics of Animal Experiments of the National Research Council (Protocol Number Biunno-2/2011). All surgery was performed under Avertin anesthesia, and all efforts were made to minimize suffering. Human fetal pancreata (18-21 weeks gestational age) were obtained from Advanced BioResources (Alameda, CA) and samples of human adult pancreas were prepared at The Diabetes Research Institute (University of Florida, Miami, FL) as previously described [46]. Human pancreatic tissue specimens were provided as ''preexisting pathological specimens'' (i.e., not through the recruitment of living human subjects), with written consent for tissue donation obtained by the procurement entity. All studies described in this study were reviewed and approved by the Human research Protections Programs of the University of California San Diego (la Jolla, CA) and the University of Washington (Seattle, WA).

Cell Adhesion, Proliferation and Secretion Assay
For adhesion assay, MIN6 trasfected cells and untreated control were plated onto different ECMs as previously described [48]. After 1 hour, cells were fixed in 4%PFA, stained with 1% Tolouidine and counted under the microscope.
For in vitro proliferation assessment, MIN6 trasfected cells and untreated control were pulsed with 10 mM BrdU (Sigma-Aldrich) and cultured for 1 hour. Cells were then fixed and processed for immunofluorescence analysis as described below. BrdU-positive cells were counted in 5 random fields under a fluorescence microscope and results expressed as percentage of total counterstained nuclei. For in vitro insulin secretion measurements, MIN6 transfected and mouse islet nucleofected cells were incubated in HEPES-balanced Krebs-Ringer buffer (KRBH), pH 7.4 (10 mM HEPES, 120 mM NaCl, 4.7 mM KCl, 1.2 mM MgCl2, 1.2 mM NaH2PO4, 25 mM NaHCO3, 2.5 mM CaCl2) supplemented with 0.25% bovine serum albumin and deprived of glucose for 1 hour and then 2.8 mM glucose was added for a further hour. Cells were washed and the medium was collected after 30-min incubation with fresh KRBH solution containing 2.8 mM glucose followed by 1-hour stimulation with 22.8 mM glucose. Insulin secreted in the medium was measured by ultrasensitive EIA (Alpco Diagnostic); results were normalized by DNA content determined with Quant-IT PicoGreen reagent and Qubit fluorimeter (Life Technologies).

Q-PCR
Total RNA was purified from MIN6 and mous islet cells using Tri-Reagent (Applied BioSystems) according to manufacturer's instructions and retro-transcribed with RevertAid cDNA synthesis kit (Thermo Scientific). SYBR green qPCR was performed on RotorGene Q (QIAGEN) machine, using Maxima SYBR green master mix (Thermo Scientific) and the following primers:

Statistics
Statistical significance of differences in data values was validated by two tailed student's t test with significance limit set at p,0.05.

Immunofluorescence (IF) Analysis of SEL1L Expression in Fetal and Adult Pancreas
IF analysis of murine fetal (E16.5) and adult pancreas identifies SEL1L-specific immunoreactivity in developing acinar and in aand b-cells (Fig. 1A-F). Interestingly, as development proceeds, SEL1L expression becomes restricted to the aand bcells ( Fig. 1 G-L), virtually absent in the acini, with only occasional single cells dispersed throughout the entire organ. Among the different types of islet cells, SEL1L-specific immunoreactivity is significantly stronger in the cytoplasm of a-cells, compared to b-cells (Fig. 1G-L). This expression pattern appears conserved in the human pancreas ( Fig. 2A-L), with strong immunoreactivity detected in developing islets, nascent acinar clusters ( Fig. 2A-F) and in differentiating islet aand bcells (Fig. 2G-L). Interestingly, both in mouse and human pancreas, SEL1L immunoreactivity localizes preferentially to cells adjacent to the basement membranes and vascular/ductal structures, suggesting a possible interaction with the extracellular matrix (ECM) ( Figure S1). Collectively, these results indicate the existence of a spatiotemporal regulation of SEL1L expression during pancreas development, and suggest important functions in the islet cell compartment, possibly involving the regulation of hormone secretory function.

SEL1L Co-immunoprecipates with b1 Integrin
The presence of the Fibronectin type II domain (a collagen binding domain) in SEL1L exon 4 [1,30] led us to postulate the possible interaction of SEL1L with integrin receptors. To investigate this possibility, we used the human pancreatic ductal cell line CFPAC-1, a valuable model for assessing pancreatic cell adhesion and migration on selected ECMs [51,52], and tested SEL1L ability to interact with integrin receptor subunits by classical co-immunoprecipitation experiments. Due to lack of reliable antibodies to mouse SEL1L for IP assays, and lack of human pancreatic b-cell line, we decided to use the human CFPAC-1 cell line only for co-IP experiments with integrins, and the mouse b-cell line MIN6 for all of the functional studies. Therefore, CFPAC-1 lysate were immunoprecipitated with antiintegrin antibodies, resolved by SDS-PAGE, and probed with SEL1L antibody. As shown in Figure 3A, a5-, b1-, b4integrinspecific antibodies co-immunoprecipitated SEL1L. Interestingly among the a subunits only a5 integrin is able to co-immunoprecipate with SEL1L, while among the b subunits, b1-integrin is able to pull-down SEL1L more efficiently than b4, suggesting that the fibronectin receptor a5b1 is the main integrin heterodimer able to interact with SEL1L. We therefore focused on the b1-integrin subunit to test the reverse approach. Interestingly, SEL1L-specific antibody co-immunoprecipitated the 110-kDa b1 precursor, but not the mature 130-kDa form, suggesting that SEL1L may interact with a protein complex prior to the post-translational modification of the b1 integrin-subunit to mediate the recruitment and assembly of integrin ab heterodimer receptors. However, it remains to be determined whether this interaction occurs directly, i.e. b1 integrin/SEL1L, or indirectly through an alternative binding mechanism such as recruitment into a b1 integrin/ Collagen/SEL1L.

SEL1L Downmodulation Interferes with Integrinmediated Adhesion and ERK Activation in MIN6 Cells
The functional relationship between SEL1L and b1 integrin was analyzed in MIN6 cells. These cells have extensively been used as a reliable model system to study the biology of pancreatic b-cells since they retain a high degree of endocrine differentiation (i.e. insulin expression) and glucose-stimulated insulin secretion [41]. Hence, we first confirmed the colocalization of SEL1L and b1integrin in MIN6 cells. Immunofluorescence analysis revealed SEL1L and b1-integrin specific immunoreactivity targeted to plasma membrane, suggesting a functional activity in ''inside-out'' integrin signaling and/or trafficking of the integrin receptors (Fig. 3B). MIN6 cells were transfected with either siRNA specific for Sel1l transcripts, or with a scrambled control, and then used in cell adhesion short-term assays to study their ability to adhere to a panel of ECM proteins (Collagen-IV, Fibronectin, Laminin and Vitronectin) known to promote b-cell attachment and function. Sel1l-interfered cells adhered less efficiently to the matrixes tested when compared to the mock-transfected ones (Fig. 4A). Interestingly, overexpression of the b1 integrin subunit in Sel1l-interfered cells rescued their adhesion defect to all matrixes analyzed, suggesting that constitutive b1 integrin overexpression is able to override its dependence on SEL1L. In fact, SEL1L-knockdown appears to efficiently down modulate b1 integrin protein levels resulting in impaired utilization of endogenous levels of b1 integrin and reduced activation of ERK1/2 (phospho ERK1/2) (Fig. 4B-C). Furthermore, interfered cells lost the usual polygonal shape and assumed a round-shape morphology (Fig. 4D), suggesting a switch from cell-matrix to cell-cell adhesion mechanisms as a consequence of a reduced adherence to ECMs. Overexpression of b1 integrin construct restored ERK activation (Fig. 4B) and the cells re-acquired their normal morphological appearance (Fig. 4D). Cells transfected with control siRNA-scrambled alone or cotransfected with b1 integrin overexpressing construct did not significantly alter ERK activation levels, nor cell morphology, further attesting a significant functional interdependence between SEL1L and b1 integrin.

SEL1L Downregulation Impairs Glucose-stimulated Insulin Secretion
Integrin heterodimers containing the b1 subunit have been implicated in the regulation of insulin secretion through ERK signaling, thus supporting an important function for this class of integrins in the secretory function of pancreatic b-cells [36,37]. Therefore, we set out to determine if SEL1L down modulation interfered with the b-cells response to glucose. For these studies, MIN6 cells were first transfected with siRNA against Sel1l and, 48 hours later, used in assays of glucose-mediated insulin secretion. As shown in Figure 5A, Sel1l-interfered cells failed to respond to glucose stimulation at the maximal concentration of 22.8 mml/l, whereas a 2-fold increase over basal was observed in scrambled and un-treated control. Moreover, as predicted, overexpression of b1 integrin subunit restored the glucose responsiveness in anti-Sel1l-siRNA transfected cells.
Significantly, as shown in Figure 5B, these results were validated using isolated mouse islet cells nucleofected with Sel1l siRNA and grown for 48 hours onto HTB9 matrix coated dish [45]. Quantitative PCR performed on all samples confirmed the specific down regulation of Sel1l transcript over the control cells (Fig. 5C-D).

SEL1L Donwmodulation Affects Proliferation of MIN6 Cells
SEL1L has been previously described to regulate cell cycle progression [53,54] and b-cell proliferation [24] but little is known about how it can exert this function. This prompted us to investigate if this mechanism could be mediated by b1 integrin. MIN6 cells were transfected with siRNA against Sel1l, and proliferating cells were pulsed for 1-hour with BrdU. As shown in Figure 5E, Sel1l-interfered cells displayed about 50% reduction in the percentage of BrdU positive cells compared to the scrambletransfected control, or untreated cells. Overexpression of b1 integrin in Sel1l-interfered cells restored DNA synthesis in MIN6 cells. To confirm the involvement of SEL1L in the b1 integrin signaling, changes in levels of expression of the key players of the cell cycle progression mediated by integrins were analyzed by qPCR. As shown in Figure 5F, Sel1l-interfered cells displayed a significant down-regulation of the cyclin D1 (Ccnd1), and an upregulation of cyclin inhibitor p21 (Cdkn1a), while overexpression of . Dual-color immunoflurescence showed SEL1L specific immunoreactivity (green, C and F) in the nascent acinar cells adjacent to basement membrane (arrowheads) and in few interspersed cells (arrows); a strong cytoplasmatic staining is also observed in the developing islets (asterisks) stained for glucagon (red, C) and insulin (red, F); in adult human pancreas, the exocrine tissue didn't show any SEL1L immunoreactivity (green, G and J), while endocrine cells revealed a marked expression of SEL1L protein with a strong cytoplasmic immunoreactivity in a-cells (stained for glucagon in red, I) and b-cells (stained for insulin in red, L). Scale bar = 50 mm. doi:10.1371/journal.pone.0079458.g002 b1 integrin in Sel1l-interfered cells restored cyclin D1 level and strongly inhibit p21 expression. Collectively, these results demonstrate that SEL1L plays multiple functions in pancreatic b-cells, spanning from control of cell-matrix adhesion, to cell proliferation and insulin secretion.

Discussion
Despite significant efforts devoted to the identification of genes regulating b-cell ontogeny in either animal models of pancreas development or in population of putative stem/ progenitor cells, our knowledge of mechanisms regulating bcell replication and/or regeneration remains relatively incom-plete. In this report we present data supporting an important involvement of SEL1L in islet cell adhesion through modulation of b1 integrin signaling, insulin secretion, and replication. Hence, we find that SEL1L is highly expressed in fetal and adult islet clusters, and in developing acinar cells, with a subcellular localization in the basal pole of islet cells adjacent to the basement membrane and vascular/ductal structures, where b1 integrin-specific immunoreactivity is preferentially identified ( Figure S1) [38], that is suggestive for a functional involvement of SEL1L in integrin signaling and/or trafficking of the integrin receptors. It has been reported that SEL1L is highly and predominantly involved in the quality control system managing UPR-inducing stresses [2,55] and studies on mouse models have shown that haplo-insufficiency of SEL1L predisposes mice to high fat-induced hyperglycemia probably due to elevated ER stress and up-regulation of the unfolded protein response pathway. Loss of b-cell function in T2DM has been associated to endoplasmic reticulum stress responses induced by chronically elevated circulating levels of glucose and lipids, collectively referred to as glucolipotoxicity [25,27]. However, how ER stress mechanistically impacts on b-cell replication and insulin secretion remains unclear [22].
Our data demonstrate that SEL1L is able to regulate cell adhesion, proliferation and secretion of b-cells through b1 integrin signaling. Both integrins and the extracellular matrixes are known to play an essential role in promoting b-cell migration, proliferation and differentiation through the activation of the ERK pathway. Our experiments of SEL1L knockdown uncovered a mechanistic link with b1 integrin-dependent adhesion and proliferation, possibly by a decreased activation of ERK signaling. Thus, SEL1L, as a key regulator of the ERAD pathway, may play an important role in the trafficking/folding of b1 integrin, such that its downregulation may lead to improper degradation of this adhesion receptor and a defective activation of downstream ERK signaling that can only be rescued by the ectopic overexpression of b1 integrin.
Although the direct interaction between SEL1L and b1 integrin remains to be demonstrated, it is likely to occur through the fibronectin type II domain within SEL1L. This domain is missing from the invertebrate orthologs of SEL1L, and arises only in the chordate lineage, indicating the acquisition of new functions in higher organisms. It is worth mentioning that among the chordates, only the mouse and rat SEL1L proteins show a variant form consisting in the absence of the FNII domain, likely due to alternative splicing (NP_035474 and NP_808794, respectively).
Interestingly, the SEL1L-FNII domain contains two inverse RGD motifs (DGR), a non-canonical adhesion sequence capable of binding to various RGD-dependent integrins, albeit with a low affinity [56]. DGR-containing peptides have been shown to compete with adhesive proteins for integrin interaction [56][57][58][59], and found to be important for the cell adhesion and biological activities mediated by basic fibroblast growth factor (FGF2) and secreted frizzled-related protein (sFRP) through modulation of Wnt signaling [60,61]. Based on the well-known role of these pathways in the control of cell growth and differentiation, DGR containing proteins such as SEL1L may contribute to regulate the crosstalk between Wnt, integrin, and growth factor signaling. Further studies will be required to identify protein domains mediating SEL1L-/b1-integrin interaction.
SEL1L may also be involved in the regulation of glucoseinduced insulin secretion. Thus, this is a complex function that requires a multifactorial regulations: (i) insulin needs to be properly folded and packaged in secretory granules before being shuttled to the plasma membrane, (ii) the right outside-in integrin-signaling cascade must be properly enacted, and (iii) glucose sensing mechanisms efficiently activated. In this complex cascade of events, SEL1L may control both the ''inside-out'' and ''outside-in'' trafficking of molecules important in islet b-cell function such as folding of insulin, integrin-transduction cascades from the microenviroment, and turnover of glucose transporters. Matrix interaction can also influence insulin secretion via engagement of b1 integrin, and alteration in the glycosylation levels of the glucose transporter GLUT2 can impair glucose-stimulated insulin release [62]. Our findings on SEL1L-dependent insulin secretion are consistent with prior reports linking SEL1L deficiency to severe impairment of protein secretory pathways [23]. In this context, our demonstration, that overexpression of b1 integrin rescues the insulin secretory defect resulting from SEL1L knockdown, further support an important role of SEL1L in the integration of extracellular cues with the intracellular machinery regulating endocrine secretory functions.
Collectively, our studies provide new insights on the implication of SEL1L in pancreatic islet cell development and function that involve the recruitment and interaction with cell adhesion receptors of the integrin family. Further exploration of SEL1L function in the integration of mechanism governing b-cell replication and function may have significant implications for the development of cell-based replacement therapies for diabetes. Figure S1.