The microRNAs miR-211-5p and miR-204-5p modulate ER stress in human beta cells.

Abstract

MicroRNAs (miRNAs) are a class of small non-coding RNAs that regulate gene expression. Type 1 diabetes is an autoimmune disease characterized by insulits (islets inflammation) and pancreatic beta cell destruction. The pro-inflammatory cytokines interleukin-1β (IL-1β) and interferon-γ (IFN-γ) are released during insulitis and trigger endoplasmic reticulum (ER) stress and expression of pro-apoptotic Bcl-2 proteins in beta cells, thus contributing to their death. The nature of miRNAs that regulate ER stress and beta cell apoptosis remains to be elucidated. We have performed a global miRNA expression profile on cytokine-treated human islets and observed a marked down-regulation of miR-211-5p. By real-time PCR and western blot analysis, we confirmed cytokine-induced changes in the expression of miR-211-5p and the closely related miR-204-5p and downstream ER-stress related genes in human beta cells. Blocking of endogenous miRNA-211-5p and miR-204-5p by the same inhibitor (it is not possible to block separately these two miRs) increased human beta cell apoptosis, as measured by Hoechst/Propidium Iodide staining and by determination of cleaved caspase-3 activation. Interestingly, miRs-211-5p and 204-5p regulate the expression of several ER stress markers downstream of PERK, particularly the pro-apoptotic transcription factor CHOP. Blocking CHOP expression by a specific siRNA partially prevented the increased apoptosis observed following miR-211-5p/miR-204-5p inhibition. These observations identify a novel crosstalk between miRNAs, ER stress and beta cell apoptosis in early type 1 diabetes.

INTRODUCTION

Selective beta cell death and the consequent progressive decrease in functional beta cell mass are hallmarks of type 1 Diabetes. Immune-mediated apoptosis is probably the major form of beta cell death in type 1 diabetes (Cnop et al. 2005). This selective beta cell destruction occurs in the course of a specific islets inflammatory process (insulitis). Candidate genes for type 1 diabetes regulate several crucial steps in the complex dialog between beta cells and immune systems taking place during insulitis (Eizirik et al. 2009, Eizirik et al. 2012, Santin & Eizirik 2013, Floyel et al. 2015). Islets inflammation triggers endoplasmic reticulum (ER) stress in beta cells exacerbating inflammation and contributing to beta cell death (Brozzi et al. 2015, Brozzi & Eizirik 2016, Eizirik et al. 2013, Eizirik et al. 2008, Cnop et al. 2017). ER stress results from the accumulation of misfolded proteins inside the lumen of the ER, leading to the activation of the unfolded protein response (UPR) (Walter & Ron 2011). Inositol requiring enzyme 1 (IRE-1), activating transcription factor 6 (ATF-6) and double-stranded RNA-activated protein kinase (PKR)-like ER kinase (PERK) are the three signalling proteins activated during UPR. Activated PERK induces the phosphorylation of eIF2α, which causes a general decrease in protein translation while inducing a paradoxical increase in the translation of activating transcription factor 4 (ATF4) and other regulators of the UPR. ATF4 increases the expression of two important transcription factors, namely ATF3 and C/EBP homologous protein (CHOP) that contribute to beta cell apoptosis(Cnop et al. 2017, Walter & Ron 2011). The UPR aims to restore ER homeostasis and prevent beta cell death by decreasing global protein synthesis though up-regulating expression of key ER chaperones (Ron & Walter 2007). In case of severe and/or prolonged ER stress, apoptosis is eventually triggered (Eizirik & Cnop 2010).

Importantly, markers of ER stress are present in islets from type 1 diabetes patients (Marhfour et al. 2012) and diabetes-prone NOD mice (Tersey et al. 2012), and restoring ER stress function decreases the incidence of diabetes in NOD mice (Engin et al. 2013). MiRNAs are short (~ 22 nucleotides in length) non-coding single stranded RNAs that act at a post-transcriptional level controlling gene expression (Bartel 2004, Friedman et al. 2009). MiRNAs function either as single molecules or as clusters of related families of miRNAs that regulate key gene networks (Haier et al. 2016, Dambal et al. 2015). MiRNAs participate in beta cell differentiation, endocrine cell identity and preservation of beta cell phenotype during adult life (Martinez-Sanchez et al. 2016). MiRNAs may also contribute to autoimmunity, beta cells apoptosis and diabetes development (Sayed & Abdellatif 2011, Isaacs et al. 2016, Sebastiani et al. 2011, Ventriglia et al. 2015). In line with this, we have recently described a family of miRNAs, namely miR-23a-3p, miR-23b-3p and miR-149-5p, that regulate the expression of the pro-apoptotic BH3-only proteins DP5 and PUMA in human pancreatic beta cells (Grieco et al. 2017).

There is growing evidence that miRNAs regulated ER stress and UPR activation in other cell types (Byrd & Brewer 2013, Malhi 2014, Maurel & Chevet 2013, Bartoszewska et al. 2013, Urra et al. 2013). As mentioned above, ER stress plays an important role in the “dialog” between the immune system and beta cells during diabetes progression (Brozzi et al. 2015, Brozzi & Eizirik 2016, Eizirik et al. 2013, Eizirik et al. 2008, Cnop et al. 2017) and it has been recently shown that miR-204-5p regulates ER stress and the UPR response in pancreatic mouse and human beta cells (Xu et al. 2016), while miR-708 is induced by ER stress in mouse beta cells and leads to impaired beta cell function and growth (Rodriguez-Comas et al. 2017). Against this background, the aim of the present study was to further investigate a possible link between miRNAs and ER stress in human pancreatic beta cells, focusing on miR-211-5p. Of note, many of the presently described experiments were based on the use of a miR-211-5p inhibitor to evaluate the effects of mRNA-211-5p depletion in beta cell apoptosis and ER stress markers expression, but this inhibitor also decreases expression of the closely related family member, miR-204-5p, recently shown to target PERK and modulate rodent beta cells apoptosis (Xu et al. 2016). Due to the very high similarity between these two miRNAs, it is not possible to obtain specific inhibitors for miR-211-5p or miR-204-5p. Thus, most findings described in the present study in cytokine-treated beta cells are probably the result of parallel miR-211-5p and miR-204-5p inhibition.

These observations indicate that two closely related miRNAs, namely miR-211-5p and miR-204-5p, regulate ER stress in human pancreatic beta cells exposed to pro-inflammatory cytokines.

RESEARCH DESIGN AND METHODS

Ethic statements.

Pancreases not suitable for clinical purposes were obtained from the Endocrinology and Metabolism of Organ and Cellular Transplantation unit at Cisanello University Hospital at the University of Pisa, Italy with informed written consent. They were processed with the approval of the local Medical and Health Research Ethics Committees of the Pisa University, Italy. The donors were anonymized, and all experiments and methods using human pancreatic islets were approved by and performed in accordance with the guidelines and regulations made by the regional Medical and Health Research Ethics committees of the Pisa University, Italy. No organs/tissues were procured from prisoners.

Culture of human islet cells and the human beta cell line EndoC-βH1.

Human islet isolation was performed using collagenase digestion and density gradient purification from 16 heart-beating brain-dead organ donors with no medical history of diabetes or metabolic disorders (Marchetti et al. 2007). The donors (9 men and 7 women) were 72 ± 2 years old and had a BMI of 25 ± 0.6 [kg/m²] (Supplementary Table 1). The islets were cultured in M199 culture medium with 5.5 mmol/l glucose (Thermo Fisher Scientific, Waltham, Massachusetts, USA) in Pisa and sent to Brussels within 1 - 5 days following isolation. After arrival in Brussels and overnight recovery, the islets were dispersed and cultured in Ham’s F-10 medium containing 6.1 mmol/l glucose (Thermo Fisher Scientific) as previously described (Eizirik et al. 2012, Moore et al. 2009). Islet beta cell purity (52 ± 4%) was evaluated by immunocytochemistry in dispersed islet cells using a specific anti-insulin antibody (Supplementary Table 2). The human insulin-producing EndoC-βH1 cell line (kindly provided by Dr R. Scharfmann, University of Paris, France; RRID:CVCL_L909) (Ravassard et al. 2011) was cultured as described (Grieco et al. 2014, Brozzi et al. 2015).

Cell treatment.

Human islet cells and insulin-producing EndoC-βH1 cells were exposed to recombinant human IL-1β (R&D Systems, Minneapolis, Minnesota, USA), 50 U/ml and recombinant human IFN-γ (Peprotech, London, UK), 1000 U/ml; cytokine concentrations were selected based on our previous experiments (Brozzi et al. 2015, Moore et al. 2009, Eizirik et al. 1994). EndoC-βH1 cells were also exposed to 1 μM of the chemical ER stressor thapsigargin, (Sigma-Aldrich, St. Louis, Missouri, USA), based on previous experiments by our group (Cnop et al. 2014).

Transfection and RNA interference.

Both EndoC-βH1 cells and human islet cells were transiently transfected with Lipofectamine RNAiMAX lipid reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. After 8h transfection EndoC-βH1 cells and human islets cells were cultured for an additional 24h or 48h recovery period respectively before exposure to cytokines or thapsigargin. Single-stranded miRCURY LNA inhibitor (Sequence 5’-3’: GGCGAAGGATGACAAAGGGA, Exiqon, Vedbaek, Denmark) that blocks endogenous miR-211-5p was used at a concentration of 120 nmol/l based our own dose response experiments (data not shown) and as described (Grieco et al. 2017, Roggli et al. 2012). The siRNA targeting CHOP (Sequence 5’-3: GUCCUGUCUUCAGAUGAAtt, siRNA#1 s3995, Ambion-Thermo Fisher Scientific) has been previously validated by our group; this included comparison against a second siRNA that induces similar biological effects (Allagnat et al. 2012). The optimal concentration of siRNAs used for transfections (30 nmol/l) was previously established by our group (Moore et al. 2009, Moore et al. 2012). Allstar Negative Control siRNA (siCTRL) (Qiagen, Hilden, Germany) was used as negative control. This siCTRL does not affect beta cell function, gene expression or viability as compared with non-transfected human islet cells (Moore et al. 2012) and EndoC-βH1 cells (data not shown).

Assessment of cell viability.

Inverted fluorescence microscopy was used to determine viable, necrotic, and apoptotic cells after 15 minutes incubation with the DNA dyes Hoechst 33342 (5 μg/ml) and propidium iodide (5 μg/ml) (Sigma-Aldrich). At least 600 cells were counted per experimental condition by two independent observers, with one of them unaware of the identity of the samples. This assay is quantitative and has been validated by systematic comparison against ladder formation, electron microscopy and caspase 3/9 activation (Moore et al. 2009, Hoorens et al. 1996).

Measurement of miRNA and mRNA expression.

The miRNeasy micro kit (Qiagen) was used for isolation of total RNA. DNase digestion was done using Rnase-Free Dnase kit (Qiagen) according to the manufacturer’s instructions. RNA quality and quantity was evaluated by a Bio Drop instrument (Isogen Life Science, Utrecht, Netherlands). cDNAs preparation and determination of miRNA and mRNA expression were performed as described (Grieco et al. 2017). MiRNA expression was evaluated using the Relative Quantification (RQ) method (RQ = 2^−Δct and 2^−ΔΔct). The concentration of the genes of interest was calculated as copies per μl using the standad curve method (Overbergh et al. 1999). Gene expression values were corrected for the housekeeping gene β-actin (actin expression in human islets is not affected by cytokine treatment (Moore et al. 2009, Cardozo et al. 2001)). miRNAs expression were normalized by small nucleolar RNAs miR-U6 and miR-RD61 in the cytokine-only experiments and by miR-375-3p and let-7a-5p for experiments with transfected cells (this is specified for each experiment in the appropriate figure legend). The list of primers used in this study are provided in Supplementary Table 3.

Western blot and Immunofluorescence.

Cells were washed with cold PBS and lysed with Laemmli Sample Buffer for Western blot. Total proteins were then extracted, resolved by 8-14% SDS-PAGE and transferred to a nitrocellulose membrane. Immunoblot analysis for the proteins of interest was done by overnight incubation with the antibodies listed in Supplementary Table 2. Membranes were then incubated for 1 h at room temperature with secondary peroxidase-conjugated antibody (anti-IgG (H+L)-HRP). The immunoreactive bands were visualized with the SuperSignal West Femto chemiluminescent substrate (Thermo Fisher Scientific), detected by ChemiDoc XRS+ (Bio-Rad, Hercules, California, USA) and quantified using the Image Laboratory software (Bio-Rad). Densitometric values were normalized by the housekeeping protein α-tubulin. Double immunofluorescence for insulin and cleaved caspase-3 was performed as previously described (Marroqui et al. 2015, Grieco et al. 2017). Immunofluorescence images were acquired on a Zeiss Imager A1 microscope via AxioVision software (Zeiss, Oberkochen, Germany).

Statistical analysis.

Data are presented as mean values ± SEM or plotted as box plots, indicating lower quartile, median, and higher quartile, with whiskers representing the range of the remaining data points when the number of experiments was ≥ 6 per condition. In some experiments data are represented as points indicating individual experiments. Statistical analysis was performed by two-tailed paired Student’s t-test or by ANOVA with post hoc Bonferroni, as appropriate, using GraphPad Prism 6 (GraphPad Software, San Diego, California, USA). A p value <0.05 was considered statistically significant.

RESULTS

Cytokines co-regulate miR-211-5p and CHOP expression in human beta cells.

We have previously identified 57 cytokines (IL-1β+IFN-γ)-modified miRNAs in human islets using TaqMan Array Human MicroRNA Cards Panel A v2.1 (Grieco et al. 2017). Using miRNA-target prediction programs (TargetScan and TargetRank) and literature analysis (Chitnis et al. 2012, Bartoszewska et al. 2013, Malhi 2014) we have now searched for miRNAs that might modulate ER stress, particularly the pro-apoptotic transcription factor CHOP. Based on this analysis, we identified miR-211-5p as an interesting candidate, as it has been already shown to directly target CHOP (Chitnis et al. 2012). Among miRNAs down-regulated after cytokines treatment miR-211-5p was consistently modulated (i.e. changed in the same direction in each individual experiment) with a 74% decrease (i.e. fold value of 0.26 ± 0.02) versus controls (not treated with cytokines) whole human islet (Grieco et al. 2017). MiR-211-5p belongs to the same family as miR-204-5p, previously found to modulate ER stress-related genes (Xu et al. 2016), regulate apoptosis in the human tubercular meshwork cells of the eye (Li et al. 2011) and promote vascular ER stress and endothelial dysfunction (Kassan et al. 2017). MiR-204-5p, however, was not considered down-regulated after cytokine treatment in our analysis, since it didn’t reach the established cut-off value of fold change ≤ 0.75 versus untreated islets (the observed fold-change of cytokine-treated versus controls was 0.79) (Grieco et al. 2017). Against this background, we initially focused on miR-211-5p. Cytokine-induced down-regulation of miR-211-5p was first confirmed in the same set of samples used for the miRNA expression profiling (Grieco et al. 2017). MiR-211-5p was found down-regulated after cytokines treatment with a fold change of 0.65 ± 0.04 versus controls (not treated) human islets cells (n=3). We next investigated whether the target gene CHOP was regulated by cytokine treatment in an opposite manner as compared to miR-211-5p. Following cytokine treatment there was significant decrease of miR-211-5p expression in dispersed human islets (Fig. 1A) and in human insulin-producing EndoC-βH1 cells (Fig. 1C), which was paralleled by a significant increase in mRNA expression of CHOP (Fig. 1B and 1D). We also tested miR-211-5p and CHOP expression in whole human islets after cytokine treatment and found the same trend in decrease for the first and increase for the second (Supplementary Figure 1). MiR-211-5p was also down-regulated by the ER stressor thapsigargin (Supplementary Figure 2A) in EndoC-βH1 cells, again paralleled by increased expression of CHOP (Supplementary Figure 2B).

Figure 1. Cytokines co-regulate miR-211-5p and CHOP expression in human beta cells.

Dispersed human islets cells (A and B) and EndoC-βH1 cells (C and D) were left untreated or treated with IL-1β+IFN-γ for 48h. Expression of miR-211-5p (A and C) was assayed by RT-PCR and normalized by two different small nucleolar RNAs (miR-U6 and miR-RD61). Expression of CHOP (B and D) was evaluated by RT-PCR and normalized by the housekeeping gene β-actin. The results are shown as individual experiments, before and after cytokines treatment; n=5-6 independent experiments. * p<0.05 and ** p<0.01 versus untreated cells; paired Student’s t-test.

Knock-down of CHOP protects EndoC-βH1 cells against cytokine-induced apoptosis in the context of miR-211-5p inhibition.

To determine whether CHOP is regulated by miR-211-5p in beta cells, we next used a miR-211-5p inhibitor and observed a clear up-regulation of CHOP mRNA expression both in untreated and cytokine-treated EndoC-βH1 cells (Fig. 2A). This miR-211-5p inhibitor-dependent induction of CHOP was abrogated when we combined the miR-211-5p inhibitor with a siRNA targeting CHOP in both untreated and cytokine-treated human EndoC-βH1 cells (Fig. 2A). The inhibition of miR-211-5p, in the range of 60-70%, was confirmed by real-time qPCR (Fig. 2B). The strong CHOP induction after miR-211-5p inhibition led to increased apoptosis in both untreated and cytokine-treated human EndoC-βH1 cells (Fig. 2C); interestingly there seems to be a feedback mechanism between CHOP and miR-211, as miR-211-5p inhibition is less marked after blocking CHOP. Importantly, blocking CHOP induction in parallel to miR-211-5p inhibition partially prevented the increase in apoptosis observed in cytokine-treated cells (Fig. 2C), indicating that CHOP up-regulation is at least in part responsible for the cell death observed following miR-211-5p inhibition.

Figure 2. KD of CHOP protects EndoC-βH1 cells against cytokine-induced apoptosis in the context of miR-211-5p inhibition.

EndoC-βH1 cells were transfected with siCTRL, siCHOP, an anti-miRNA targeting miR-211-5p (anti-miR-211) or co-transfected with both siCHOP+anti-miR-211 (A-C) for 8h. After 24h of recovery, cells were left untreated or treated with IL-1β+IFN-γ for 48h as indicated. The KD of CHOP (A) was confirmed by RT-PCR, normalized by the housekeeping gene β-actin. Expression of miR-211-5p (B) was assayed by RT-PCR and normalized by miR-375-3p and let-7a-5p expression. Apoptosis was evaluated by propidium iodide / Hoechst staining after 48h of cytokine treatment (C). In (A) results were normalized against the highest value in each independent experiment, considered as 1. Data shown are mean ± SEM of 6-7 independent experiments (A-C). * p<0.05, ** p<0.01 and *** p<0.001 versus siCTRL untreated cells; $ p<0.05, $ $ p<0.01 and $ $ $ p<0.001 versus siCTRL cytokine-treated cells; §§ p<0.01 and §§§ p<0.001 as indicated by bars; ANOVA followed by Bonferroni’s correction.

Inhibition of miR-211-5p exacerbates apoptosis of human beta cells.

In a subsequent set of experiments, the single stranded miRNA inhibitor was used to block endogenous miR-211-5p expression in both human EndoC-βH1 cells and human dispersed islets cells, leading to a 60% and 80% decrease in miR-211-5p expression in human EndoC-βH1 cells and human islets respectively (Fig. 3A and 3F). The miR-211-5p inhibitor exacerbated both basal and cytokine-induced apoptosis, as evaluated by nuclear dyes, in both human EndoC-βH1 cells and human islets (Fig. 3B and 3G, respectively). In line with these observations, the inhibition of miR-211-5p increased expression of cleaved caspase-3 in human EndoC-βH1 cells (Fig. 3C and 3D) as evaluated by western blot, confirming cell death activation. Similar findings were observed by double immunofluorescence for insulin and cleaved caspase-3 in human EndoC-βH1 cells (Supplementary Figure 3). For each experimental condition the percentage of cleaved caspase-3 positive cells is shown in Figure 3E. Similar observations were obtained in human dispersed islets cells (Supplementary Figure 4), indicating that at least part of the observed cell death in human islet cells takes place at the beta cell level.

Figure 3. Inhibition of miR-211-5p exacerbates apoptosis of human beta cells.

EndoC-βH1 cells (A-E) were transfected with siCTRL or with an anti-miRNA targeting miR-211-5p (anti-miR-211) for 8h. After 24h of recovery, the cells were left untreated or treated with IL-1β+IFN-γ for 48h as indicated. The expression of miR-211-5p (A) was assayed by RT-PCR and normalized by miR-375-3p expression. Apoptosis was evaluated by propidium iodide/Hoechst staining after 48h of cytokine treatment (B). Cleaved caspase-3 and tubulin protein expression were evaluated by western blot after 48h of cytokine treatment; one representative blot of 6 independent experiments is shown (C). The optical density of cleaved caspase-3 (example shown in C) was quantified and corrected by tubulin expression (D). After cytokine treatment, cells were fixed and used for histological studies (as indicated in Supplementary Figure 3). The quantification of fluorescent cleaved caspase-3 positive cells is shown (E). Dispersed human islets cells (F and G) were transfected with siCTRL or with an anti-miRNA targeting miR-211-5p (anti-miR-211) for 8h. After 48h of recovery, the cells were left untreated or treated with IL-1β+IFN-γ for 48h as indicated. The expression of miR-211-5p (F) was assayed by RT-PCR and normalized by miR-375-3p and let-7a-5p expression. Apoptosis was evaluated by propidium iodide/Hoechst staining after 48h of cytokine treatment (G). The results are represented as box plots, indicating lower quartile, median, and higher quartile, with whiskers representing the range of the remaining data points (B and E); n= 4-6 independent experiments. In (D) results were normalized against the highest value in each independent experiment, considered as 1. Data shown are mean ± SEM of 6 independent experiments (A and D) or 3 independent experiments (F). In (G) the results are represented as box plots, indicating lower value, mean and higher value; n=3 independent experiments. * p<0.05, ** p<0.01 and *** p<0.001 versus siCTRL untreated cells; $ p<0.05, $ $ p<0.01 and $ $ $ p<0.001 versus siCTRL cytokine-treated cells; § p<0.05 as indicated by bars; ANOVA followed by Bonferroni’s correction (A, B, D and E); paired Student’s t-test (F and G). The order of the bands in the blot image (C) is the same as in the quantification histogram (D): from left to right are indicated respectively untreated cells transfected with a control siRNA, untreated cells transfected with the miR-211-5p inhibitor, cytokine-treated cells transfected with a control siRNA and last cytokine-treated cells transfected with the miR-211-5p inhibitor. Full-length blots are presented in Supplementary Figure 8.

MiR-211-5p inhibition increases expression of ER stress markers in EndoC-βH1 cells in a PERK-dependent manner.

In order to elucidate the molecular mechanisms by which miR-211-5p regulates ER stress we used two different miRNA-target prediction programs, namely TargetScan and TargetRank (Agarwal et al. 2015, Nielsen et al. 2007) to determine whether other ER stress genes are potential targets of miR-211-5p. Both programs found that EIF2AK3 (also known as PERK), one of the three key ER transmembrane sensors (Ron & Walter 2007, Eizirik et al. 2008), is a predicted target of two miRNAs belonging to the same family, namely miR-211-5p and miR-204-5p (Fig. 4A). Exposure of EndoC-βH1 cells to the miR-211-5p inhibitor increased mRNA expression of PERK both in untreated and cytokine-treated human EndoC-βH1 cells (Fig. 4B). These results suggested that miR-211-5p targets and inhibits PERK in beta cells. In line with this finding, up-regulation at the mRNA level was observed for the transcription factors downstream of PERK, ATF4, ATF3, and CHOP (Fig. 4D, 4E and 4F, respectively), as well as for eIF2α (Fig. 4C). In agreement with mRNA expression, the miR-211-5p inhibitor increased basal protein expression of PERK (Fig. 5A and 5B), ATF4 (Fig. 5D and 5F) and ATF3 (Fig. 5D and 5H) and, to a less extent, of p-eIF2α and eIF2-α (Fig. 5C, 5E and 5G) in EndoC-βH1 cells. These effects of the miR-211-5p inhibitor were less marked following cytokine-treatment (Fig. 5 A-H). On the other hand, miR-211-5p inhibition had not effect on the two other branches of the ER response pathways, namely ATF6 and IRE1α, as evaluated by mRNA and protein expression for BIP (downstream of ATF6; Supplementary Figure 5A, 5C and 5D) and mRNA expression of XBP-1s (downstream of IRE1-α; Supplementary Figure 5B), respectively. Taken together, these results indicate that miR-211-5p has a preferential effect on the PERK pathway, consistent with its predicted binging site on the 3’UTR of PERK (Fig. 4A).

Figure 4. miR-211-5p inhibition increases mRNA expression of ER stress markers in EndoC-βH1 cells in a PERK-dependent manner.

(A) Alignment of miR-211-5p and miR-204-5p, miRNAs belonging to the same family; seed sequence (grey shading) and predicted binding site for the miRNAs (grey shading) of human EIF2AK3 (also known as PERK) 3’UTR using TargetScan software version 7.1. EndoC-βH1 cells (B-F) were transfected with siCTRL or with an anti-miRNA targeting miR-211-5p (anti-miR-211) for 8h. After 24h of recovery, the cells were left untreated or treated with IL-1β+IFN-γ for 48h as indicated. Expression of PERK (B), eIF2α (C), ATF4 (D), ATF3 (E) and CHOP (F) were assayed by RT-PCR and normalized by the housekeeping gene β-actin. In (B-F) results were normalized against the highest value in each independent experiment, considered as 1. Data shown are mean ± SEM of 6 independent experiments (B-F). * p<0.05, ** p<0.01 and *** p<0.001 versus siCTRL untreated cells; $ p<0.05, $ $ p<0.01 and $ $ $ p<0.001 versus siCTRL cytokine-treated cells; § p<0.05 and §§§ p<0.001 as indicated by bars; ANOVA followed by Bonferroni’s correction.

Figure 5. miR-211-5p inhibition increases protein expression of ER stress markers in EndoC-βH1 cells in a PERK-dependent manner.

Human insulin-producing EndoC-βH1 cells (A-H) were transfected with siCTRL or with an anti-miRNA targeting miR-211-5p (anti-miR-211) for 8h. After 24h of recovery, the cells were left untreated or treated with IL-1β+IFN-γ for 48h as indicated. PERK and tubulin expression were evaluated by western blot after 48h of cytokines treatment; one representative blot of 8 independent experiments is shown; the order of the lanes is similar to the ones shown in (B). (A). The optical density quantification of PERK (example shown in A) was quantified and corrected by tubulin expression (B). p-eIF2α, eIF2α and tubulin expression were evaluated by western blot after 48h of cytokine treatment; one representative blot of 9-11 independent experiments is shown (C). The optical density of p-eIF2α (E) and eIF2α (G) bands (shown in C) was quantified and corrected by tubulin expression. ATF4, ATF3 and tubulin expression were evaluated by western blot after 48h of cytokine treatment; one representative blot of 8-9 independent experiments is shown (D). The optical density of ATF4 (F) and ATF3 (H) bands (shown in D) was quantified and corrected by tubulin expression. In (B and E-H) results were normalized against the highest value in each independent experiment, considered as 1. Data shown are mean ± SEM of 8-11 independent experiments. * p<0.05, ** p<0.01 and *** p<0.001 versus untreated cells; $ p<0.05 versus siCTRL cytokine-treated cells; ANOVA followed by Bonferroni’s correction. The order of the bands in the blot images (A, C and D) is the same as in the quantification histograms (B and E-H, respectively): from left to right are indicated respectively untreated cells transfected with a control siRNA, untreated cells transfected with the miR-211 inhibitor, cytokine-treated cells transfected with a control siRNA and last cytokine-treated cells transfected with the miR-211-5p inhibitor. Full-length blots are presented in Supplementary Figure 9.

To investigate whether miR-211-5p also regulates PERK and its downstream transcription factors in the context of a “pure” ER stress, Endo-βH1 cells transfected with the miR-211-5p inhibitor were exposed to the chemical ER stressor thapsigargin. In line with the cytokine data (see above), the miR-211-5p inhibitor exacerbated apoptosis in both untreated and thapsigargin-treated cells (Supplementary Figure 6A). There was also increased mRNA expression of PERK, ATF4, ATF3 and CHOP, but not in eIF2α (Supplementary Figure 6B, 6D, 6E, 6F and 6C, respectively) in miR-211-5p-depleted and thapsigargin-treated cells. The inhibition of miR-211-5p (30-40%) was confirmed by real-time qPCR (data not shown).

MiR-211-5p inhibition increases expression of pro-apoptotic Bcl-2 family members in human EndoC-βH1 cells.

ER stress-mediated-cell death is executed via the canonical mitochondrial apoptosis pathway, with BH3-only BCL-2 family members DP5 and PUMA playing a crucial role (Urra et al. 2013, Gurzov et al. 2009). MiR-211-5p inhibition in untreated and cytokine-treated Endo-βH1 cells increased mRNA expression of the transcription factor c-JUN (a key regulator of DP5 expression in beta cells (Gurzov et al. 2009), Supplementary Figure 7A), and both DP5 and PUMA (Supplementary Figure 7B and 7C, respectively). On the other hand, miR-211-5p inhibition did not modify BAX, BIM, NOXA, MCL-1 and BCL-XL mRNA expression (data not shown). The miR-211-5p-dependent increase of c-JUN and DP5 was not mediated by CHOP since it was not reverted by its parallel inhibition (Supplementary Figure 7A and 7B, respectively). On the other hand, CHOP KD partially reverted PUMA induction following miR-211-5p inhibition (Supplementary Figure 7C). Taken together, these results suggest that miR-211-5p may also contribute to beta cell apoptosis, besides its effects on ER stress, by modulating directly or indirectly the expression of key pro-apoptotic Bcl-2 protein family members. The mechanisms involved in this modulation remain to be clarified but they may be mediated at least in part via up-regulation of the transcription factor c-JUN (Supplementary Figure 7A).

Cytokines down-regulate miR-204-5p expression and miR-211-5p inhibition decreases miR-204-5p expression.

While the present work was in progress, a new and interesting study reported that miR-204-5p targets PERK and regulates UPR signalling and beta cells apoptosis (Xu et al. 2016). Moreover, other studies described the importance of miR-204-5p in beta cell function and endocrine phenotype in rodent and human pancreatic islets (Xu et al. 2013), with a controversial role of miR-204-5p in regulating insulin mRNA expression trough MAFA (Marzinotto et al. 2017). In light of all these observations, we re-evaluated the impact of cytokines on miR-204-5p expression, and confirmed that miR-204-5p is indeed down-regulated by cytokines in both dispersed human islets and in human EndoC-βH1 cells (Fig. 6A and 6B, respectively). Surprisingly, miR-204-5p expression was not down-regulated by thapsigargin exposure (Fig. 6C), suggesting that miR-204-5p inhibition is a specific effect of cytokines exposure.

Figure 6. Cytokines but not thapsigargin down-regulate miR-204-5p expression. The use of a miR-211-5p inhibitior also decreases miR-204-5p expression.

Dispersed human islets cells (A) and human insulin-producing EndoC-βH1 cells (B and C), were left untreated or treated with IL-1β+IFN-γ for 48h (A and B) and with the ER stressor thapsigargin for 24h (C) as indicated. Expression of miR-204-5p was assayed by RT-PCR and normalized by two different small nucleolar RNAs (miR-U6 and miR-RD61). The results are represented as individual experiments; n=5-6 independent experiments. ** p<0.01 versus untreated cells; paired Student’s t-test. Human insulin-producing EndoC-βH1 cells (D) and human dispersed islets cells (E) were transfected with siControl (CTRL) or with an anti-miRNA targeting miR-211-5p (anti-miR-211) for 8h. After 24h (D) and 48h (E) of recovery, cells were left untreated of treated with IL-1β+IFN-γ for 48h as indicated. Expression of miR-204-5p was assayed by RT-PCR and normalized by and miR-375-3p and let7a-5p expression. Data shown are mean ± SEM of 6 independent experiments (D) or 3 independent experiments (E). ** p<0.01 and *** p<0.001 versus siCTRL untreated cells; $ $ $ p<0.001 versus siCTRL cytokines-treated cells; ANOVA followed by Bonferroni’s correction.

Due to the close similarity between miR-211-5p and miR-204-5p, we next evaluated whether the presently utilized miR-211-5p inhibitor also affects miR-204-5p expression. There was indeed an important cross-reaction, with the miR-211-5p inhibitor decreasing the expression of both miRNAs in human EndoC-βH1 cells (Fig. 3A and Fig. 6D) and in human dispersed islets cells (Fig. 3F and Fig. 6E). Thus, the above-described findings based on the use of this inhibitor should be interpreted as mediated by inhibition of a family of miRNAs comprising both miR-211-5p and miR-204-5p, and not miR-211-5p alone.

DISCUSSION

MicroRNAs are key regulators of gene expression and have been associated to autoimmunity progression and diabetes development (Ventriglia et al. 2015, Martinez-Sanchez et al. 2016). We have recently reported that the miRNAs miR-23a, miR-23b and miR-149-5p sensitize human beta cells to cytokine-induced apoptosis by modulating expression of the BH3-only proteins DP5 and PUMA in human pancreatic beta cells (Grieco et al. 2017). In the present study we show that the family of miRs-211-5p and −204-5p act as regulators of ER stress, particularly of the downstream pro-apoptotic transcription factor CHOP in a PERK-dependent manner. We first observed a parallel cytokine-induced down-regulation of miR-211-5p and up-regulation of CHOP in human islets ((Grieco et al. 2017) and present data), suggesting that this could be an important mechanism by which cytokines up-regulate CHOP. We subsequently used a miR-211-5p inhibitor to evaluate the effects of mRNA-211-5p depletion in beta cell apoptosis and ER stress markers expression. This inhibitor also decreases expression of miR-204-5p, a closely related family member. miR-204-5p was recently shown to target PERK and modulate rodent beta cells apoptosis (Xu et al. 2016) and due to the close similarity between these two miRNAs we could not design specific inhibitors for miR-211-5p or miR-204-5p. Taking this into account, the bulk of the present findings in cytokine-treated human beta cells are probably the result of parallel miR-211-5p and miR-204-5p inhibition. Expression of miR-204-5p is at least 50-fold higher in human beta cells than the expression of miR-211-5p (our own unpublished data; this is line with a previous study from Kameswaran and colleagues showing an even more marked different between miR-204-5p and miR-211-5p expression(Kameswaran et al. 2014)), suggesting that iR-204-5p may have a more relevant biologically function than miR-211-5p. On the other hand, the fact that miR-211-5p, but not miR-204-5p, is inhibited by the pure ER stressor thapsigargin suggests that under some experimental conditions miR-211-5p may play a more relevant role in cellular responses.

MiR-211-5p inhibition increased both basal and cytokine- or thapsigargin-induced apoptosis. This was paralleled by an increased expression of ER stress markers downstream of the PERK branch, namely eIF2α, ATF4, ATF3 and CHOP. On the other hand, there were no modifications in BIP and XBP1-s, which are regulated by the two other branches of UPR, namely ATF6 and IRE1α, reinforcing the idea that the miR-211-5p family targets specifically PERK and downstream genes.

Besides its effect on ER stress, miR-211-5p depletion up-regulates the expression of pro-apoptotic Bcl-2 proteins DP5 and PUMA, at least in part via the up-regulation of c-JUN, a transcription factor previously shown by us to modulate DP5 expression (Gurzov et al. 2009). KD of CHOP in parallel with miR-211-5p inhibition partially prevents the up-regulation of PUMA and the pro-apoptotic effects of miR-211-5p inhibition, suggesting that the miR-211-5p family may be a key component in the crosstalk between ER stress and pro-apoptotic Bcl-2 family member in human beta cells.

In conclusion, the cytokines (IL-1β+IFN-γ) inhibit expression of both miR-211-5p and miR-204-5p in human beta cells. Decreased expression of these inhibitory miRNAs enables – directly or indirectly - the up-regulation of PERK and c-JUN. PERK activation triggers eIF2α, ATF4, ATF3 and CHOP, while the transcription factor c-JUN contributes for the up-regulation of the pro-apoptotic BH3-only proteins DP5 and PUMA. CHOP probably also contributes to up-regulate PUMA. Activation of these two pathways culminates in caspase-3 activation and beta cell apoptosis. As a whole, the present and previous findings (Xu et al. 2016) indicates that the miR-211-5p and miR-204-5p family is a key hub in the regulation of cytokine-induced ER stress and pro-apoptotic pathways in human beta cells.