The Connexin 43 Regulator Rotigaptide Reduces Cytokine-Induced Cell Death in Human Islets

Background: Intercellular communication mediated by cationic fluxes through the Connexin family of gap junctions regulates glucose-stimulated insulin secretion and beta cell defense against inflammatory stress. Rotigaptide (RG, ZP123) is a peptide analog that increases intercellular conductance in cardiac muscle cells by the prevention of dephosphorylation and thereby uncoupling of Connexin-43 (Cx43), possibly via action on unidentified protein phosphatases. For this reason, it is being studied in human arrhythmias. It is unknown if RG protects islet cell function and viability against inflammatory or metabolic stress, a question of considerable translational interest for the treatment of diabetes. Methods: Apoptosis was measured in human islets shown to express Cx43, treated with RG or the control peptide ZP119 and exposed to glucolipotoxicity or IL-1β + IFNɣ. INS-1 cells shown to lack Cx43 were used to examine if RG protected human islet cells via Cx43 coupling. To study the mechanisms of action of Cx43-independent effects of RG, NO, IkBα degradation, mitochondrial activity, ROS, and insulin mRNA levels were determined. Results: RG reduced cytokine-induced apoptosis ~40% in human islets. In Cx43-deficient INS-1 cells, this protective effect was markedly blunted as expected, but unexpectedly, RG still modestly reduced apoptosis, and improved mitochondrial function, insulin-2 gene levels, and accumulated insulin release. RG reduced NO production in Cx43-deficient INS-1 cells associated with reduced iNOS expression, suggesting that RG blunts cytokine-induced NF-κB signaling in insulin-producing cells in a Cx43-independent manner. Conclusion: RG reduces cytokine-induced cell death in human islets. The protective action in Cx43-deficient INS-1 cells suggests a novel inhibitory mechanism of action of RG on NF-κB signaling.


51
Despite a strong preclinical rationale (4) anti-IL-1 monotherapy was ineffective overall in recent-onset 52 T1D. However, IL-1 antagonism did moderate inflammation and caused a 2,5-fold higher secretory 53 function in T1D patients with intermediary beta-cell function at baseline (5, 6). Since IL-1 induced 54 beta-cell apoptosis is potentiated by other proinflammatory cytokines such as TNF, IFNg and IL-6, and 55 since anti-TNF therapy improved beta-cell function in a small placebo controlled trial (7), it is likely 56 that a combination of treatments targeting various aspects of signaling caused by the cytokine network 57 is needed to improve efficacy of anti-cytokine strategies in T1D, as has indeed been demonstrated in 58 animal models (8). Thus, there is need for novel safe therapeutic approaches for such combination 59 therapies. 60 Appropriate pulsatile insulin secretion depends on islet intercellular communication and 61 synchronization (9). Accordingly, cytokine-mediated de-synchronization of intercellular oscillating 62 calcium fluxes alters the beta cell transcriptome and sensitizes it to stress-induced impaired secretory 4 63 function and apoptosis (10). Gap junctions are intercellular channels composed of two connexin (Cx) 64 hemi-channels, each comprised of six Cx subunits (11). Cx36 is a predominant Cx in beta cells and 65 regulates insulin secretion by calcium flux synchronization (12,13). Other Cxs are also expressed in 66 islets (Cx43 and Cx45 in mice; Cx30.3, Cx31, Cx31.1, Cx31.9, Cx43, and Cx45 in humans), although 67 the function of these Cxs is less characterized (11,14,15). Of note, several studies have failed to show 68 basal Cx43 expression in rat islets or insulin-producing cell lines (16)(17)(18). 69 Whole-body Cx36 deficient mice develop beta-cell destruction and hyperglycemia, whereas beta-cell 70 specific transgenic Cx36 over-expression protects against single high-dose streptozotocin-induced 71 diabetes and restores islet insulin contents in this model. In addition, proinflammatory cytokines 72 reduce beta-cell Cx36 expression, and Cx36 deficiency aggravates cytokine-induced beta-cell toxicity 73 (10). Further, beta-cell specific knockout of Cx43, but not of Cx36, reduces pancreatic insulin content 74 and islet size (19). Therefore, pharmacological targeting of gap junctions has been proposed as a novel 75 approach to rescue pancreatic beta cells from stress-induced apoptosis. An important knowledge gap to 76 guide clinical trials is the demonstration of the importance of the Cx family in human islet apoptosis 77 induced by inflammatory stress.

78
The peptide Rotigaptide (RG) increases intercellular conductance in cardiac muscle cells and restores 79 gap junction intercellular communication (GJIC) in atrial cardiomyocytes of metabolically stressed 80 rats (20). RG prevents dephosphorylation and thereby uncoupling of Cx43, possibly by acting on yet 81 unidentified protein phosphatases (21,22). In addition, RG dose-dependently increases the Cx43 82 expression level in rat cardiomyocytes (23). Here we had two aims: 1) to investigate if Cx43 plays a 83 role in inflammatory-stress induced human islet cell apoptosis using RG as a Cx43 coupler 2) to verify 84 that RG exerted an anti-apoptotic effect via Cx43 in human islets by demonstrating loss-of-function of 85 RG on cytokine-induced apoptosis in Cx43 deficient INS-1 cells.   penicillin/streptomycin (P/S) and 5.6mM glucose at 5% CO 2 and 37°C as described previously (27).

108
There were no apparent differences in results obtained with islets from male or female donors, and 109 data were therefore combined.

115
Apoptosis and cell viability assays 116 For apoptosis assay carried out in duplicate independent cultures, DNA/histone complexes released 117 from the nucleus to the cytosol were measured using Roche cell death assay kit (Roche,Mannheim,118 Germany) according to the manufacturer's protocol. As a surrogate of cell viability, mitochondrial 119 function was measured in duplicate by MTT assay in which water soluble MTT (3-(4,5-120 dimethylthiazol-2-yl)-2,5-diphenyltetrazolium-bromide) is converted by normal cells to an insoluble 121 formazan salt with an optical density (OD) read at 570nm.

122
Nitric oxide (NO) assay 123 As a surrogate of nitric oxide production, accumulated nitrite was measured in duplicate samples from 124 two independent parallel cultures. Supernatants (100ul) from the wells used for INS-1 cell apoptosis 125 assay were mixed with an equal volume of the Griess reagent (one part 0.1% naphtylethylene diamine 126 dihydrochloride and one part 1% sulfanilamide in 5% H3PO4 (Merck, Darmstadt, Germany), and read 127 at 550nm in a plate reader (Thermo Scientific, Naerum, Denmark). The nitrite concentration was 128 calculated using a standard curve of 0.5-40uM concentrations of NaNO 2 (Merck) as described (28). for measurement of accumulated insulin using insulin competitive ELISA assay in duplicate samples 154 from two independent parallel cultures as described (31), except that the enzyme substrate 1-step 155 Ultra TMB (3,3', 5,5;-tetramethylbenzidine) (Life Technologies) was used here. In contrast to intact rat islets, the rat insulin-producing INS-1 cells did not express Cx43, and Cx43 apoptosis to the same extent as that observed in human islets. Unexpectedly, RG but not CP modestly 179 but significantly reduced apoptosis in cytokine-exposed cells ~10% at IL-1b concentrations above 180 15pg/ml. Exposure of INS-1 cells to glucolipotoxic conditions significantly increased apoptosis by 3.8 181 fold, but this was counteracted neither by RG nor CP.

182
Since cytokines induce mitochondrial stress in pancreatic beta cells, we then investigated if the Cx43 183 independent action of RG was associated with prevention of cytokine-induced mitochondrial 184 dysfunction. IL-1b dose-dependently reduced mitochondrial function by 64% at 150pg/ml IL-1b. RG, 185 but not CP, slightly but significantly improved mitochondrial function in cytokine-exposed cells at 37, 186 50 and 75pg/ml IL-1 ( fig. 2b).

187
Rotigaptide reduces neither cytokine nor glucolipotoxicity-induced ROS production in INS-1 188 cells. 189 We next asked if improved mitochondrial function caused by RG was associated with reduced ROS  (32), we explored if RG would prevent cytokine-induced mitochondrial dysfunction. We therefore measured the mRNA levels of two genes 195 encoding key subunits of the mitochondrial complex I and IV: NADH-dehydrogenase subunit 2 (ND2) 196 and Cytochrome C Oxidase II (Sco2), respectively. Cytokines down-regulated ND2 and there was a 197 trend for down-regulation of Sco2 (p=0.09), but RG did not restore these changes ( fig. 3b). 203 We next investigated if Rotigaptide mitigated nitroxidative stress by impeding IkBa degradation.

204
Cytokines caused IkBa degradation in a time-dependent manner ( fig. 4c). There was a strong trend 205 (p=0.06) for RG but not CP diminishing the AUC for IkBa degradation. Since cytokine-induced c-Src 206 levels might be higher in Cx43 deficient cells and since c-Src binds and thereby acts as a sink for IkBa 207 (33, 34), we investigated c-Src expression in INS-1 cells exposed to cytokines in the presence or 208 absence of RG or CP. As shown in fig. 4d c-Src mRNA levels were unaffected by cytokines and RG.

209
Taken together, the data suggest that Rotigaptide reduces nitroxidative stress independently of its 210 action on Cx43 activity. To explore if RG restores cytokine-mediated inhibition of insulin biosynthesis and secretion, we  An important question is how RG intercepts NF-kB activation in the absence of Cx43. In Cx43 235 deficient cells, lack of Cx43/c-Src interaction would be expected to lead to higher c-Src availability for 236 IkBa binding, thereby priming the cells for cytokine-triggered NF-kB activation. We investigated c-Src 237 expression in INS-1 cells and were unable to show altered c-Src mRNA levels caused by either 238 cytokines or RG. In Cx43 competent cells however, this mechanism would be expected only to play a 239 role under conditions of inhibited Cx43 expression, such as those induced by high glucose (33, 34).

240
The lack of protection of RG on high glucose-high lipid toxicity in INS-1 cells may be explained by 241 absence of Cx43, and these experiments should therefore be repeated in human islets, in which RG 242 would be expected to counteract glucose-mediated reduction in Cx43 expression.

243
Strengths of this study were the use of human islets to demonstrate the protective effect of RG against 244 cytokine-induced apoptosis, and the inclusion of a scrambled peptide as a control for RG. We also took 245 advantage of INS-1 cells known to lack Cx43 expression as a 'natural' Cx43 knock-out model ( fig.1a)

248
A limitation was the restricted amounts of human islets available to us for these studies, precluding 249 more mechanistic investigations such as Cx43 activity, Ca 2+ flux synchronization and studies of 250 glucose-stimulated insulin secretion. Electromobility shift assay, iNOS promotor chromatin 251 immunoprecipitation, or luciferase-based NF-kB reporter assay would help define if NF-kB promoter 252 binding or NF-KB transcriptional activity is reduced in Rotigaptide-treated islets exposed to cytokines.

253
Rotigaptide is also known to increase Cx43 half-life in cardiomyocytes by slowing its trafficking and 254 proteasomal degradation (23,35,36). However, proof of Cx43 dependence of such effects of RG in 255 human islets would require knock-down or knock-out of Cx43, which is not effective in intact islets 256 but requires islet dispersion into monolayers, thereby disrupting the normal cell-to-cell contact and communication. Further research is needed to clarify if these effects contribute to the protective action 258 of Rotigaptide in human islets.

259
When added to the fact that beta-cell specific overexpression of Cx43, but not of Cx36, increases 260 pancreatic insulin content and islet size 19, our observations highlight the translational potential RG as 261 a novel approach to prevent inflammatory beta-cell failure and apoptosis in diabetes and warrant 262 further preclinical studies. The notion that RG may have a dual protective action related to its known 263 activity as a Cx43 activator and to a novel Cx43 independent inhibiting action on NF-kB activation 264 makes RG an attractive candidate for mono-or combination therapy in diabetes.

265
In conclusion, RG reduces cytokine-induced cell death in human islets, likely by preventing Cx43 266 uncoupling. RG conferred protection against inflammatory assault even in Cx43 deficient INS-1 cells, 267 suggesting a novel inhibitory mechanism of action of RG on NF-kB signaling. These observations 268 support further development of RG as a novel therapy to protect beta-cell functional mass in diabetes 269 due to its dual protective action on key beta-cell pro-apoptotic pathways. in untreated human and rat islets, and in INS-1 cells exposed to cytokine combination (150 pg/ml IL-398 1b+0.1 ng/ml IFNg) using specific primers with qPCR. The expression of the genes normalized to 399 HPRT1 was calculated by -∆Ct. B) Apoptosis was measured by Roche cell death assay according to 400 the manufacturer's protocol. Data are means ± SEM of N=5 donor human islets. * ≤ 0.05, tt ≤ 0.01.

401
The symbols t and star indicate the Bonferroni-corrected paired t-test values of treatments versus 402 control (CTL) and cytokine (Cyt) conditions, respectively. NS: not significant.