The No-Go and Nonsense-Mediated RNA Decay Pathways Are Regulated by Inflammatory Cytokines in Insulin-Producing Cells and Human Islets and Determine β-Cell Insulin Biosynthesis and Survival
Abstract
Stress-related changes in β-cell mRNA levels result from a balance between gene transcription and mRNA decay. The regulation of RNA decay pathways has not been investigated in pancreatic β-cells. We found that no-go and nonsense-mediated RNA decay pathway components (RDPCs) and exoribonuclease complexes were expressed in INS-1 cells and human islets. Pelo, Dcp2, Dis3L2, Upf2, and Smg1/5/6/7 were upregulated by inflammatory cytokines in INS-1 cells under conditions where central β-cell mRNAs were downregulated. These changes in RDPC mRNA or corresponding protein levels were largely confirmed in INS-1 cells and rat/human islets. Cytokine-induced upregulation of Pelo, Xrn1, Dis3L2, Upf2, and Smg1/6 was reduced by inducible nitric oxide synthase inhibition, as were endoplasmic reticulum (ER) stress, inhibition of Ins1/2 mRNA, and accumulated insulin secretion. Reactive oxygen species inhibition or iron chelation did not affect RDPC expression. Pelo or Xrn1 knockdown (KD) aggravated, whereas Smg6 KD ameliorated, cytokine-induced INS-1 cell death without affecting ER stress; both increased insulin biosynthesis and medium accumulation but not glucose-stimulated insulin secretion in cytokine-exposed INS-1 cells. In conclusion, RDPCs are regulated by inflammatory stress in β-cells. RDPC KD improved insulin biosynthesis, likely by preventing Ins1/2 mRNA clearance. Pelo/Xrn1 KD aggravated, but Smg6 KD ameliorated, cytokine-mediated β-cell death, possibly through prevention of proapoptotic and antiapoptotic mRNA degradation, respectively.
Introduction
Proinflammatory cytokines have been implied in β-cell failure and apoptosis in type 1 and type 2 diabetes (T1D and T2D) in part by regulation of the β-cell transcriptome (1). The levels of ∼20% of the >29,000 transcripts in human islets are altered by cytokines, including apoptosis- and inflammation-related genes (2). Interestingly, 35% of the genes expressed in human islets undergo alternative splicing, and cytokines cause substantial changes in the number of spliced transcripts (2). Most human genes exhibit alternative splicing, but not all alternatively spliced transcripts are translated into functional proteins. This varies in a cell-specific manner, contributing to tissue specificity of gene expression (3), and can be modulated by cellular signals such as those provided by proinflammatory cytokines (4). In pancreatic β-cells, cytokines upregulate the expression of >20 genes involved in RNA splicing, and cytokines induce changes in alternative splicing of >50% of the cytokine-modified genes as detected by exon array analysis (4).
mRNA molecules are prone to damage and fidelity errors, which can cause translation of aberrant mRNA into dysfunctional proteins (5), and so far, an RNA repair machinery analogous to that operating to restore damaged DNA has not been discovered. Rather, cellular RNA surveillance mechanisms assure the quality and fidelity of mRNA molecules (6), resulting in breakdown of damaged RNA. This is generally achieved through targeting aberrant mRNA molecules for degradation by various endogenous nucleases.
At least four forms of cotranslational mRNA quality control pathways have been described: nonsense-mediated decay (NMD), Staufen1 (STAU1)-mediated mRNA decay (SMD), no-go decay (NGD), and nonstop decay (NSD) (7–9). Both NSD and NGD pathways share the ribosomal dissociation factor Pelo, which is necessary for degradation of “non-stop codon” mRNAs and damaged mRNAs that cause translational stalling, respectively (10). In SMD, STAU1 recognizes either stem-loop intramolecular base pairing of 3′-UTR sequences or intermolecular base pairing of 3′-UTR sequences with long-noncoding RNA via partially complementary Alu elements (8). NMD engaging Upf1, 2, 3A, and 3B and Smg1, 5, 6, and 7 degrades mRNAs with premature termination codons (PTCs). PTCs can arise in cells through various mechanisms: germline or somatic mutations in DNA, errors in transcription, posttranscriptional mRNA damage, or notably errors in mRNA processing, such as alternative splicing (9). Failure to recognize and eliminate these mRNA transcripts can result in production of truncated and dysfunctional proteins that directly perturb cell function or accumulate in the endoplasmic reticulum (ER) and thus cause ER stress.
PTCs have been implicated in ∼30% of all inherited diseases, indicating that the NMD pathway plays a vital role in survival and health. Expressed sequence tag analysis of RNA-sequencing data of the human transcriptome identified 35% of 5,693 alternatively spliced isoforms to be NMD targets, and at least 12% of human genes have a PTC-containing splice isoform (11), thus necessitating regulated unproductive splicing and translation (RUST) to link alternative splicing and NMD and regulate the abundance of mRNA transcripts (12). Accordingly, NMD inhibition leads to an ∼15% increase in PTC-containing mRNA isoforms (13).
Very little is known about the role of RNA decay in stress regulation of the transcriptome. We therefore hypothesized that cytokines regulate RNA decay with consequences for β-cell survival in response to inflammatory stress. We report that RNA decay pathway components (RDPCs) are regulated by proinflammatory cytokines in β-cells in a nitric oxide (NO), but not reactive oxygen species (ROS), dependent manner. RDPC knockdown (KD) partially restored insulin transcripts, content and medium accumulation, but not glucose-stimulated insulin secretion (GSIS), and modulated cytokine-mediated death in cytokine-exposed INS-1 cells.
Research Design and Methods
Cell Culture and Cytokine Exposures
The rat INS-1 cell line provided by Claes Wollheim (University of Geneva, Geneva, Switzerland) tested negative for Mycoplasma and was maintained as previously described (14). INS-1 cells were seeded in 6-well plates (1 × 106 cells/well for RNA isolation), 48-well plates (50,000 cells/well for cell death assay), or 96-well plates (30,000 cells/well for MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium-bromide] and Alamarblue assays) (NUNC, Roskilde, Denmark). After 48 h of preincubation, cells were exposed to cytokine mixture at the concentrations indicated in figures or figure legends for the indicated time points, glucolipotoxic (GLT) conditions (0.5 mmol/L palmitate conjugated with 0.1% albumin + 25 mmol/L glucose), or control medium (RPMI-glutamax, 10% FBS, 1% penicillin/streptomycin, 50 μmol/L 2-mercaptoethanol).
Human and Rat Islet Culture and Cytokine Exposures
Islets from six human heart-beating organ donors (>80% purity, donor characteristics listed in Supplementary Table 1A) were isolated by the European Consortium for Islet Transplantation (ECIT) under local approval in Milan, Italy, received in fully anonymous form and precultured as described previously (15). There were no apparent differences in results obtained with islets from male or female donors, and data were therefore combined. For experiments, 50 human islets were cultured in RPMI supplemented with 2% human serum (Life Technologies, Naerum, Denmark), 1% penicillin/streptomycin, and 5.6 mmol/L glucose for 24 h exposed with or without cytokine mixture (300 pg/mL rrIL-1β + 10 ng/mL rhIFN-γ + 10 ng/mL rhTNF-α) or control medium for 4 days.
Rat islets were isolated from 1-week-old rat pups and precultured for 72 h before exposure to cytokines for 12, 18, and 24 h as previously described (16).
Single-Cell RNA Sequencing of Pancreatic Islets
The expression of genes in islet cell types was determined by reanalyzing published human islet single-cell sequencing data (donor information in Supplementary Table 1B; EBI accession number MTAB-5061) (17).
Apoptosis and Cell Viability Assays
Apoptosis assay was performed in duplicate through detection of either caspase-3 activity using the fluorimetric caspase-3 activity assay kit (Sigma-Aldrich, Copenhagen, Denmark) or DNA/histone complexes released from the nucleus to the cytosol using the Roche cell death assay kit (Roche, Mannheim, Germany) according to the manufacturers’ protocols. Cell viability was measured by MTT and Alamarblue assays (Life Technologies), as previously described (18,19).
NO Assay
As a surrogate of NO production, accumulated nitrite was measured in supernatants (100 µL) from the wells used for INS-1 cell apoptosis assay by the Griess reagent, as previously described (20).
Real-time Reverse Transcriptase Quantitative PCR
Total RNA was extracted using the NucleoSpin kit (Macheray-Nagel, Bethlehem, PA) according to the manufacturer’s instructions. Quality and quantity of the extracted RNA were assessed using a NanoDrop-1000 (Thermo Fisher Scientific). Five-hundred nanograms total RNA was used for cDNA synthesis with the iScript-cDNA kit (Bio-Rad, Copenhagen, Denmark). Real-time reverse transcriptase quantitative PCR (RT-qPCR) was performed on 12 ng cDNA with SYBR Green PCR Master Mix (Life Technologies) and specific primers (Supplementary Table 2) and run in a real-time PCR machine (Applied Biosystems, Naerum, Denmark). NormFinder software (21) was used to select the most stable reference gene among hypoxanthine-guanine phosphor-ribosyl-transferase 1 (Hprt1), actin, and 18S rRNA. Statistical analysis was performed on the gene expression levels normalized to the chosen reference gene (see figure legends) through −∆Ct analysis, and for simplicity some figures were presented as fold change versus control, which was calculated by the −∆∆Ct method.
Northern Blotting Analysis
Seven micrograms total RNA was blotted to a positive charged nylon membrane (Ambion, Naerum, Denmark) and probed against insulin-2 (Ins2) mRNA and 18S rRNA using radiolabeled-specific probes as described in detail in the Supplementary Data.
Western Blotting Analysis
Western blotting (WB) was performed as described in the Supplementary Data (22) using antibodies against Pelo (Novus Biologicals, Littleton, CO), Xrn1, Dis3L1, Dis3L2, α-tubulin (Sigma-Aldrich), Smg1, Smg5, Smg6, Smg7, Upf1, or Upf2 (Santa Cruz Biotechnology, Heidelberg, Germany).
GSIS
INS-1 cells (300,000) were cultured in 12-well plates (Nunc) and incubated for 2 days before being exposed to cytokine mixture (150 pg/mL IL-1β + 0.1 ng/mL IFN-γ) or control medium. GSIS was performed using Krebs-Ringer buffer containing 2 or 17 mmol/L glucose as previously described (23).
Insulin Assay
Insulin concentration (pmol/L) was measured using insulin-competitive ELISA assay as previously described (24), with the modification that the enzyme substrate 1-Step Ultra TMB (3,3′,5,5-tetramethylbenzidine) (Life Technologies) was used.
Lentiviral Short Hairpin RNA Gene KD
GIPZ lentiviral short hairpin RNA (shRNA) particles (Dharmacon, Søborg, Denmark) against Pelo, Xrn1, Smg6, and Upf1 genes along with a nonsilencing shRNA (NS) vector as negative control were produced using the Trans-Lentiviral shRNA Packaging System (Dharmacon) in HEK293 cells according to the manufacturer’s instructions and as described in detail in the Supplementary Data.
Ins2 mRNA Decay Kinetics in CHO Cells
The Ins2 gene was cloned into CHO cells (CHO-Ins), and its mRNA half-life and decay kinetics were determined using the Lenti-X Tet-One Inducible Expression System as fully described in the Supplementary Data.
Statistical Analysis
Data are presented as means ± SEM. Statistical analysis was performed on raw data even in cases where normalized data are shown. Group comparisons were conducted as two- or one-way ANOVA as appropriate. Two-way ANOVA P values are shown in Supplementary Tables 3 and 4. One-way ANOVA P values are given next to the related figures. Significant ANOVAs were followed by post hoc paired Student t test with Bonferroni correction using GraphPad Prism version 6 (GraphPad Software, La Jolla, CA). The paired Student t test was chosen to normalize for interpassage variability in outcome parameters. Since the experimental conditions did not allow sequential sampling from the same cell culture, parallel control and interventional plate wells were considered to be paired observations and analyzed accordingly statistically. If the post hoc paired Student t test did not reveal the comparison carrying the ANOVA statistical difference, individual paired Student t tests were performed and corrected for multiple comparisons. Bonferroni-corrected P values ≤0.05 were considered significant and ≤0.10 a trend.
Results
Cytokines and GLT Differentially Regulate the Expression of RNA Decay Components in INS-1 Cells and Human Islets
Since cytokines and, to a lesser extent, GLT downregulated mRNA levels of insulin and β-cell–specific transcription factors (TFs) (Supplementary Fig. 1A and B) and inhibited accumulated insulin release (Supplementary Fig. 1C), we examined whether cytokines or GLT regulate proteins belonging to either NGD or NMD pathways in INS-1 cells or rat/human islets, respectively. RT-qPCR analysis showed that cytokines at one or more time points significantly upregulated the mRNA level of the ribosomal dissociation factor of the NGD pathway (Pelo), NMD components (Upf2, Smg1, Smg5, Smg6, and Smg7), and the exosomal system (mRNA decapping [Dcp2] and exoribonuclease [Dis3L2]) but did not change Hbsl1, Abce1, Upf1, DcpS, Xrn1, Dis3L1, or Dis3 mRNA levels in INS-1 cells (Fig. 1A). The NMD component Upf3B was downregulated.
Cytokines or GLT increase the expression of genes regulating RNA decay components in INS-1 cells and human islets. mRNA levels of RDPCs in INS-1 cells exposed to cytokine combination (Cyt;150 pg/mL IL-1β + 0.1 ng/mL IFN-γ) (A), GLT (0.5 mmol/L palmitate conjugated with 0.1% albumin + 25 mmol/L glucose) (B), or control (Ctl) medium for 12, 18, and 24 h were quantified by RT-qPCR and normalized to 18S rRNA (n = 6). The basal relative mRNA level is shown in Supplementary Fig. 1F. C: Expression level of key components of RNA decay pathways in dispersed donor human islet β-cells from healthy subjects (n = 6) and patients with T2D (n = 4) was determined by single-cell RNA sequencing and quantified by reads per kilobase per million mapped reads (RPKM) per cell. The symbol t indicates ANOVA with Bonferroni-corrected post hoc Student t test P value of T2D vs. healthy donors. D: The protein levels of key candidates of RDPCs in INS-1 cells exposed to the cytokine combination for the given times were determined by WB and quantified normalized to tubulin (n = 3, but n = 2 for Smg6 in INS-1). Blots are shown in Supplementary Fig. 1G. E: mRNA levels of RDPCs and UPR markers in human islets exposed to Cyt (300 pg/mL IL-1β + 10 ng/mL IFN-γ + 10 ng/mL TNF-α) for 12, 18, and 24 h were quantified by RT-qPCR and normalized to Hprt1 (n = 6). The basal relative mRNA level is shown in Supplementary Fig. 1H. F: The protein levels of key candidates of RDPCs in human islets exposed to the cytokine combination for the given times were determined by WB and quantified normalized to tubulin (n = 3). Blots are shown in Supplementary Fig. 1I. G–J: INS-1 cells or human islets were exposed to Cyt (for INS-1 cells: 150 pg/mL IL-1β + 0.1 ng/mL IFN-γ; for human islets: 300 pg/mL rrIL-1β + 10 ng/mL rhIFN-γ + 10 ng/mL rhTNF-α) with or without coincubation with 10 or 100 µg/mL PTC124, respectively. Apoptosis was measured in human islets (n = 3) (G) and INS-1 cells (n = 3) (H). NO was measured in the collected supernatant from INS-1 cells (n = 3) (I), and mRNA levels of UPR markers were measured in INS-1 cells by RT-qPCR and normalized to 18S rRNA (n = 6) (J). Cyt 15, 15 pg/mL IL-1β + 0.1 ng/mL IFN-γ; Cyt 37, 37 pg/mL IL-1β + 0.1 ng/mL IFN-γ; Cyt 150, 150 pg/mL IL-1β + 0.1 ng/mL IFN-γ. Data are means ± SEM. The symbols t and * indicate the Bonferroni-corrected paired Student t test P values of treatments vs. Ctl and Cyt conditions, respectively, unless otherwise stated. * or t, ≤0.05; ** or tt, ≤0.01; *** or ttt, ≤0.001; **** or tttt, ≤0.0001.
Since glucotoxicity at least in part involves activation of β-cell–derived IL-1β production and secretion (25), we investigated if exposure of INS-1 cells to GLT conditions induced alteration in the RDPC expression similar to those increased by cytokines. The mRNA levels of Pelo/Smg1/Smg5 were significantly increased in GLT-exposed INS-1 cells at one or more time points (Fig. 1B). Hbsl1 and Abce1 were unaffected. Single-cell RNA sequencing of β-cells (Fig. 1C), but not of the other endocrine cell subtypes (Supplementary Fig. 1D), derived from four deceased patients with T2D and six control subjects (Supplementary Table 1B) confirmed the significant upregulation of Pelo observed in the GLT-exposed INS-1 cells (Fig. 1B).
WB analysis confirmed significant upregulation (or trends for upregulation) of Pelo, Xrn1, Dis3L2, Upf1 (P = 0.08), Upf2, Smg5, Smg6 (no P value, n = 2), and Smg7 protein (P = 0.059) but downregulation of Smg1 protein in cytokine-exposed INS-1 cells (Fig. 1D). WB analysis of two independent experiments also confirmed the increase in protein levels of the key RNA decay proteins (Pelo, Xrn1, Dis3L2, Upf1, and Smg6) in rat islets at one or more time points (Supplementary Fig. 1E).
In human islets exposed to cytokines, mRNA levels of Xrn1, Upf2, Upf3B, Smg1, and Smg7, but not Pelo, Dis3L1/2, Upf3A, or Smg5, were significantly upregulated at one or more time points with a trend for upregulation of Smg6 (Fig. 1E). The protein levels of Xrn1, Upf1, and Smg6 largely reflected changes observed for the corresponding mRNA levels (Fig. 1F).
Since NMD may lead to synthesis of mutated proteins that elicit an unfolded protein response (UPR) (9,26), we examined UPR mRNA levels in cytokine-exposed human islets by RT-qPCR. The Bip mRNA level was significantly upregulated at 18 and 24 h (Fig. 1E). However, likely due to high baseline levels of Chop and ATF4 due to handling stress of the human islets, we were unable to detect significant cytokine-induced changes in the expression of these ER stress markers. Because cytokines induce alternatively spliced β-cell mRNA isoforms potentially containing PTCs, the translation of which is expected to trigger apoptosis through ER stress, we next asked if forcing ribosomal readthrough of such PTC-containing mRNAs with the drug PTC124 would aggravate cytokine-induced apoptosis in human islets and INS-1 cells, which was the case (Fig. 1G and H). PTC124 also potentiated INS-1 cell NO production (Fig. 1I) and expression of the ER stress markers Bip (P = 0.0005) and Chop (P = 0.06) (Fig. 1J). Taken together, these results show that cytokines and GLT conditions differentially regulate the expression of RDPCs in INS-1 cells and human islets.
Pelo and Xrn1 KD Aggravates, Whereas Smg6 KD Reduces, Cytokine-Induced INS-1 Cell Viability
We next investigated the impact of Pelo, Upf1, Smg6, and Xrn1 stable KD (Supplementary Fig. 2) on cytokine-induced cell death. Pelo (Fig. 2A) or Xrn1 KD (Fig. 2B) modestly, but significantly, aggravated, whereas Smg6 (Fig. 2C) modestly, but significantly, ameliorated cytokine-induced cell toxicity measured either by the caspase-3 activity assay kit or by MTT and Alamarblue assays. In addition, the effect of the KD of RDPCs on cytokine-induced toxicity was confirmed by measuring cell death using the Roche apoptosis detection kit (Supplementary Fig. 3). Smg6 KD also improved viability measured by MTT in the absence of cytokines (Fig. 2C).
KD of Pelo and Xrn1 reduces β-cell viability, whereas KD of Smg6 rescues INS-1 cell viability when exposed to cytokines. The INS-1 cell lines with the most efficient stable KD of Pelo (three shRNAs) (A), Xrn1 (three shRNAs) (B), Smg6 (two shRNAs) (C), and NS were exposed to cytokine combination (Cyt; 150 pg/mL IL-1β + 0.1 ng/mL IFN-γ) or control (Ctl) medium for 24 h. Cell viability was measured by fluorometric caspase-3 activity assay kit or Alamarblue and MTT assays. The data are means ± SEM of n = 6. The symbols t and * indicate the Bonferroni-corrected paired Student t test P values of treatments vs. Ctl and Cyt conditions, respectively. * or t, ≤0.05; ** or tt, ≤0.01; *** or ttt, ≤0.001; tttt, ≤0.0001.
These data indicate that in cytokine-induced β-cell apoptosis, the NGD and exoribonuclease pathways serve protective functions, whereas the NMD pathway potentiates toxicity.
KD of the RNA Decay Effectors Pelo, Xrn1, and Smg6 Restores Insulin mRNA, Content and Accumulated, but Not GSIS in Cytokine-Exposed INS-1 Cells
In non–cytokine-exposed cells, KD of Pelo and Xrn1, but not Smg6, increased Ins1/2 mRNAs (Fig. 3A–C) and insulin content (Fig. 3D–F). KD of all three RDPC effectors increased accumulated insulin secretion (Fig. 3G). Xrn1 and Smg6 KD also increased GSIS (Fig. 3E and F). Because only one shRNA against Pelo had the same effect (Fig. 3D), we cannot exclude an off-target effect of this particular shRNA. As expected, cytokines dramatically reduced accumulated insulin secretion (Fig. 3G) and GSIS (Fig. 3D–F) from nonsense (NS) cells. Pelo, Xrn1, or Smg6 KD partially restored cytokine-inhibited insulin content (Fig. 3D–F) and accumulated insulin secretion (Fig. 3G) but not GSIS (Fig. 3D–F). Taken together, these results suggest that KD of all three RDPC effectors improves insulin biosynthesis, but not GSIS, in cytokine-exposed INS-1 cells.
KD of the RNA decay effectors Pelo, Xrn1, and Smg6 restores insulin mRNA, content and accumulated, but not GSIS in cytokine-exposed INS-1 cells. A–G: The INS-1 cell lines with the most efficient KD of Pelo and Xrn1 (three shRNAs of each), Smg6 (two shRNAs), and NS were exposed to cytokine combination (Cyt; 150 pg/mL IL-1β + 0.1 ng/mL IFN-γ) or without as control (Ctl) for 18 h. The mRNA levels of Ins1/2, central β-cell–specific TFs, UPR markers, Ica512, and Gck genes were determined in Pelo (A), Xrn1 (B), and Smg6 (C) KD INS-1 cell lines by RT-qPCR and normalized to 18S rRNA. GSIS and insulin contents were investigated in Pelo (D), Xrn1 (E), and Smg6 (F) KD INS-1 cell lines. Insulin concentrations (pmol/L) were measured by ELISA (n = 6). Accumulated insulin (pmol/L) over 18 h was measured by insulin ELISA in the supernatants of Pelo, Xrn1, and Smg6 (G) KD INS-1 cell lines. H: Pelo or Xrn1 was transiently knocked down in CHO-Ins cells using NS as Ctl. The Pelo or Xrn1 KD or NS CHO-Ins cell lines were treated with 100 ng/mL Dox for 24 h, and the Ins2 mRNA decay was studied using qPCR at 128, 256, 512, and 1,024 min after Dox removal. The RT-qPCR data were normalized to 18S rRNA. The area under the curve of Ins2 mRNA decay was calculated for Pelo, Xrn1, and NS KD cell lines (n = 6). Data are means ± SEM. The symbols t and * indicate the Bonferroni-corrected paired Student t test P values of treatments vs. Ctl and Cyt conditions, respectively. * or t, ≤0.05; ** or tt, ≤0.01; *** or ttt, ≤0.001; **** or tttt, ≤0.0001. ns, not significant.
We reasoned that the increase in insulin secretion and content could be related to a reduced decay of mRNAs critical for β-cell secretory function. Indeed, Pelo KD markedly increased Ins1/2 mRNA levels in the absence of cytokines, restored cytokine-induced reductions in Ins1/2, and partially restored cytokine-induced reductions in Pdx-1, Isl-1, glucokinase (Gck), and islet cell autoantigen 512 (Ica512) (Fig. 3A). Xrn1 KD increased Ins1/2 mRNA levels in the absence of cytokines and partially restored cytokine-induced reductions in Ins1/2 and Ica512 mRNAs (Fig. 3B). In contrast, Smg6 KD did not affect Ins1/2 mRNA levels in the absence of cytokines but completely restored cytokine-induced reductions in Ins1 (P = 0.052), Ins2 (P = 0.008), and Ica512 (P = 0.0006) mRNAs and partially restored Pdx-1 and Isl-1 mRNAs (Fig. 3C). The relative changes in insulin mRNA levels observed in Pelo, Xrn1, and Smg6 KD were confirmed by Northern blotting analysis (Supplementary Fig. 4).
We next investigated if the increase in insulin RNA levels was associated with the prolonged half-life of insulin mRNA due to impaired RNA decay. We created CHO cells transfected with Ins2 mRNA under doxycycline (Dox) induction using a tetracycline-on/off expression system, termed CHO-Ins cells. Transient KD of Pelo, but not Xrn1, in CHO-Ins cells caused a significant increase in the Ins2 mRNA area under the curve level compared with NS control (Fig. 3H).
We then asked if RNA decay effector KD caused ER stress. In the absence of cytokines, Pelo KD modestly increased Atf4 (P = 0.06) and Bip (P = 0.031), and Xrn1 KD increased Bip (P = 0.029). Pelo, Xrn1, and Smg6 KD did not significantly alter cytokine-induced ER stress markers, although there was a trend for Pelo KD to increase (P = 0.059) and for Smg6 KD to reduce sXbp1 (P = 0.08) (Fig. 3A–C).
These results suggest that the NGD and NMD, but not the exoribonuclease, pathways mediate cytokine-induced reductions in mRNAs encoding proteins regulating glucose sensing and insulin gene transcription and exocytosis.
Cytokine-Induced Regulation of RNA Decay Effectors and ER Stress Marker Expression Is NO, but Not ROS, Dependent
Inflammatory cytokines increase the production of ROS in β-cells through iron catalysis of the Fenton reaction (27) and via activation of NF-κB–mediated inducible nitric oxide synthase (iNOS) gene expression and NO production (28). Further, cytokine-mediated gene regulation is in many cases NO dependent (29). We therefore investigated whether inhibition of nitroxidative or oxidative stress by the use of the iNOS inhibitor NG-methyl-l-arginine acetate (NMA), the ROS scavenger N-acetyl cysteine (NAC), and the iron chelator deferoxamine (DFO) alone or in combination alleviated cytokine-induced regulation of RNA decay components and ensuing ER stress. Several RNA decay components were regulated in cytokine-exposed INS-1 cells (as also found in Fig. 1A), and these changes were largely NO dependent in that they were reversed or partially reversed by NMA alone or in combination with NAC and DFO, but not by NAC or DFO alone (Fig. 4A). To verify that the effect of NMA was mediated via iNOS inhibition, we demonstrated in Supplementary Fig. 5A and B that NMA and the combination of NMA with NAC and DFO markedly reduced NO synthesis while not affecting iNOS mRNA expression. NAC and DFO modestly but significantly reduced NO synthesis but not iNOS mRNA expression, in agreement with previous reports demonstrating that ROS facilitates iNOS activity (30).
Cytokine-induced expressional regulations of RNA decay effectors, Ins1/2, and ER stress markers are NO, but not ROS, dependent. INS-1 cells were exposed to cytokine combination (Cyt; 150 pg/mL IL-1β + 0.1 ng/mL IFN-γ) and cotreated with 1 mmol/L NAC, 1 mmol/L NMA, 0.1 mmol/L DFO, or a mixture (Mix) of the three for 12 h. The levels of mRNA encoding for proteins belonging to the RNA decay pathway (A) and UPR markers and Ins1/2 (B) were quantified by RT-qPCR and normalized to 18S rRNA (n = 6). Data are means ± SEM. The symbols t and * indicate the Bonferroni-corrected paired Student t test P values of treatments vs. control (Ctl) and Cyt conditions, respectively. * or t, ≤0.05; ** or tt, ≤0.01; *** or ttt, ≤0.001; ****, ≤0.0001.
The counterregulation of the RNA decay pathways by iNOS inhibition, but not by ROS inhibition, alone was associated with partial normalization of Ins1/2 mRNA levels and ER stress markers (Fig. 4B). In summary, these data show that cytokine-induced regulation of RDPCs and ER stress marker expression is NO, but not ROS, dependent.
Discussion
In this study, we found that a decrease of steady-state mRNA levels of Ins1/2 and central β-cell–specific TFs in cytokine-exposed INS-1 and rat or human islets was associated with upregulation of RDPCs, exoribonucleases, and key UPR components in an NO, but not ROS, dependent manner. When considering the low sensitivity of single-cell RNA sequencing of human islets ex vivo, it is remarkable that increased Pelo expression was detectable in T2D islets, compatible with GLT-mediated upregulation of Pelo in INS-1 cells. Interestingly, this response was β-cell specific. KD of Pelo or Xrn1 aggravated, whereas Smg6 KD ameliorated, cytokine-induced INS-1 cell death without affecting ER stress; both increased insulin biosynthesis and medium accumulation but not GSIS in cytokine-exposed INS-1 cells.
This study represents, to the best of our knowledge, the first comprehensive transcriptional and translational profiling of the β-cell mRNA surveillance and decay system and the first description in any cell of how this system is regulated by inflammatory and metabolic stress. Studies have so far focused on how the mRNA surveillance system regulates inflammatory cytokine transcripts with the implication that aberrant mRNA surveillance by NMD exacerbates inflammation (31); thus, Smg1 depletion results in chronic inflammation in the mouse model (32).
The cytokine-induced changes in RDPCs may seem modest but are similar to the compensatory effect sizes caused by mutations in UPF3B associated with mental retardation in humans (33).
iNOS inhibition reversed cytokine-mediated ER stress as expected, but also upregulation of RNA decay effector mRNAs and downregulation of Ins1/2 mRNA. NO-mediated nitration activates the NF-κB (34,35), p53 (36), AP-1 (37), and other signaling pathways (35). Alternatively, nucleotide nitration leading to 8-nitroguanosine modifications of cellular RNAs may cause ribosomal stalling (7,38) and mediate RDPC expressional upregulation as a protective response. Measurement of 8-nitroguanosine modifications in cytokine-exposed INS-1 cells failed due to a lack of specific and sensitive methods (data not shown). Finally, the possibility that induction of NMD effector expression is secondary to NO-mediated alternative splicing (2,29) and generation of PTC-containing transcripts remains to be investigated.
To mimic the cellular outcome of a mismatch between ribosomal load with PTC-containing RNAs and RNA decay, we forced ribosomal readthrough in INS-1 cells and human islets with PTC124 (39) in an inflammatory setting expected to enhance generation of alternatively spliced β-cell mRNA isoforms potentially containing PTCs. We anticipated, as shown, that the forced translation would trigger apoptosis associated with ER stress through the ATF6 and PERK pathways. We did not observe ER stress in response to RDPC KD likely due to qualitative differences in the ER load of unfolded protein, expected to be higher after forced readthrough, even though NO induction was comparable.
Pelo and Xrn1 KD increased INS-1 cell sensitivity to cytokine-induced cell death without affecting ER stress, and to the best of our knowledge, there is no evidence that Pelo or Xrn1 deletion induces ER stress in any cell. Possibly cytokine-induced nitration of mRNA causing ribosomal stalling or NO-induced eIF2a phosphorylation limiting translation initiation may limit protein influx into the ER, thereby not impeding ER folding/chaperoning capacity. Accordingly, deletion of Dom34, the yeast homolog of Pelo, causes ribosomal stalling, blocking ER protein influx (40,41).
Similarly, Smg6 KD ameliorated INS-1 cell sensitivity to cytokine-induced cell death without affecting ER stress, which would have been expected due to cytokine-induced alternative splicing with the generation of PTC-containing mRNAs, leading to the formation of “mutated” proteins in turn unfolded or misfolded in the ER (2). This expectation was supported by our observation of Bip and a trend for Chop induction in INS-1 cells exposed to cytokines in the presence of PTC124, forcing readthrough of PTC-containing mRNAs. Either the incomplete (∼50%) KD or redundancy provided by the several Smg sub- and isoforms may explain the absence of ER stress in Smg6-deficient cytokine-exposed INS-1 cells.
The balance of pro- and antiapoptotic transcripts is regulated by alternative splicing and thereby defines apoptosis (42). Thus, Pelo or Xrn1 KD may prevent clearance of proapoptotic mRNAs, and Smg6 KD may prevent clearance of antiapoptotic mRNAs. To support this speculation, overexpression of the apoptotic inhibitor growth arrest and DNA damage–inducible 45β (Gadd45b) in INS-1E cells decreases IL-1β–induced apoptosis (43), and shRNA-mediated KD of Gadd45b aggravates neuron cell apoptosis (44). Gadd45b mRNA is an NMD target (45) and would therefore be expected to be more stable in Smg6 KD cells. We found that Upf1 KD did not affect cytokine-induced apoptosis, in line with the observation that Upf1 or Upf2 depletion does not significantly influence sensitivity to staurosporine-induced apoptosis (46).
We demonstrate that KD of all three RDPC effectors partially restores insulin mRNA levels and contents but not GSIS in cytokine-exposed INS-1 cells. These observations indicate that the RDPC effectors regulate insulin biosynthesis but not directly the stimulus-secretion coupling, leading to insulin retention, which drives insulin release via the constitutive pathway, thereby explaining why cytokine-mediated reductions in accumulated insulin secretion are modestly improved by RDPC KD. Using a tetracycline-on/off inducible expression system of the Ins2 gene in CHO cells to avoid confounding by endogenous Ins gene expression in INS-1 cells, we confirm restoration of Ins2 mRNA in Pelo but not in Xrn1 KD CHO cells. There are several Xrn1 isomers but only one Pelo gene, and the lack of effect of Xrn1 KD on Ins2 mRNA levels in Ins-CHO cells may be explained by species differences in Xrn1 isomer redundancy (47) or failure of Xrn1 capacity to deal with forced overexpression of Ins2.
In conclusion, RDPCs are regulated by inflammatory stress in β-cells in an NO, but not ROS, dependent manner. RDPC KD partially restored insulin transcript levels, content and medium accumulation, but not GSIS in cytokine-exposed INS-1 cells, likely explained by the effect of RDPC KD on insulin mRNA stability leading to increased insulin biosynthesis but not directly affecting stimulus-secretion coupling. Pelo/Xrn1 KD aggravated, but Smg6 KD ameliorated, cytokine-mediated β-cell death, possibly through preventing proapoptotic and antiapoptotic mRNA degradation, respectively.
Article Information
Acknowledgments. The authors thank Claes Wollheim for generously providing INS-1 cells, Dr. J.B. Hansen and Dr. Z. Gerhart-Hines (both from Section for Metabolic Receptology, Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, Copenhagen, Denmark) for providing DMT-1 KO mice, Henrik E. Poulsen (Clinical Pharmacology, State University Hospital, Copenhagen, Denmark) and Peter E. Nielsen (Department of Cellular and Molecular Medicine, University of Copenhagen, Copenhagen, Denmark) for helpful discussions and comments, N. Billestrup (Department of Biomedical Sciences, University of Copenhagen, Copenhagen, Denmark) for providing neonatal rat islets, Henrik Nielsen (Department of Cellular and Molecular Medicine, University of Copenhagen, Copenhagen, Denmark) for providing materials and laboratory facilities for Northern blotting analyses, and the ECIT (Milan, Italy) for providing human islets.
Funding. This project was funded by the Danish Diabetes Academy (DDA), the Department of Biomedical Sciences (BMI) at the University of Copenhagen, the Augustinus Foundation, and the Bjarne Jensen Foundation.
Duality of Interest. This project was funded by Zealand Pharma A/S. B.T. is employed by AstraZeneca. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. S.M.G. initiated the study, developed the protocols for the experiments, conducted experiments, performed statistical analysis, constructed figures and tables, and wrote the first draft of the manuscript. N.K. performed the Northern blotting and its data analysis and edited the manuscript. B.T. performed the bioinformatics analysis of single-cell RNA sequencing data and edited the manuscript. T.M.-P. initiated the study, developed the protocols for the experiments, and edited the manuscript. S.M.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Footnotes
This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db18-0073/-/DC1.
- Received January 17, 2018.
- Accepted July 23, 2018.
- © 2018 by the American Diabetes Association.
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.
References
In this Issue
