Islet neogenesis associated protein (INGAP) protects pancreatic β cells from IL-1β and IFNγ-induced apoptosis

The goal of this study was to determine whether recombinant Islet NeoGenesis Associated Protein (rINGAP) and its active core, a pentadecapeptide INGAP104–118 (Ingap-p), protect β cells against cytokine-induced death. INGAP has been shown to induce islet neogenesis in diabetic animals, to stimulate β-cell proliferation and differentiation, and to improve islet survival and function. Importantly, Ingap-p has shown promising results in clinical trials for diabetes (phase I/II). However, the full potential of INGAP and its mechanisms of action remain poorly understood. Using rat insulinoma cells RINm5F and INS-1 treated with interleukin-1β (IL-1β) and interferon‐gamma (IFN‐γ), we demonstrate here that both rINGAP and Ingap-p inhibit apoptosis, Caspase-3 activation, inducible nitric oxide synthase (iNOS) expression and nitric oxide (NO) production, and explore the related signaling pathways. As expected, IL-1β induced nuclear factor kappa B (NF-κB), p38, and JNK signaling, whereas interferon‐gamma (IFN‐γ) activated the JAK2/STAT1 pathway and potentiated the IL-1β effects. Both rINGAP and Ingap-p decreased phosphorylation of IKKα/β, IkBα, and p65, although p65 nuclear translocation was not inhibited. rINGAP, used for further analysis, also inhibited STAT3, p38, and JNK activation. Interestingly, all inhibitory effects of rINGAP were observed for the cytokine cocktail, not IL-1β alone, and were roughly equal to reversing the potentiating effects of INFγ. Furthermore, rINGAP had no effect on IL-1β/NF-κB-induced gene expression (e.g., Ccl2, Sod2) but downregulated several IFNγ-stimulated (Irf1, Socs1, Socs3) or IFNγ-potentiated (Nos2) genes. This, however, was observed again only for the cytokine cocktail, not IFNγ alone, and rINGAP did not inhibit the IFNγ-induced JAK2/STAT1 activation. Together, these intriguing results suggest that INGAP does not target either IL-1β or IFNγ individually but rather inhibits the signaling crosstalk between the two, the exact mechanism of which remains to be investigated. In summary, our study characterizes the anti-inflammatory effects of INGAP, both protein and peptide, and suggests a new therapeutic utility for INGAP in the treatment of diabetes.


Introduction
Diabetes mellitus is a metabolic disorder characterized by β-cell deficiency and the resulting hyperglycemia. In Type 1 diabetes (T1DM), β cells are destroyed by autoimmunity 1 , whereas in Type 2 diabetes (T2DM), β-cell failure and death come from progressive insulin resistance, metabolic and oxidative stress, and persistent hyperglycemia 2 .
Both T1DM and T2DM are characterized by islet inflammation, hyperglycemia, and β-cell apoptosis 3,4 . Studies in rodents and humans pinpoint proinflammatory cytokines and chemokines as key mediators of T1DM pathophysiology and as contributing factors to insulin resistance and β-cell dysfunction in T2DM 5,6 . Pro-inflammatory cytokines, interleukin-1β (IL-1β), tumor necrosis factor alpha (TNFα), and interferongamma (IFN-γ), have been reported to trigger destructive signaling pathways within β cells, leading to elevation of reactive oxygen species, β-cell dysfunction, and eventually to apoptosis [5][6][7] . Inhibition of these signaling pathways is therefore an important therapeutic objective aimed at preventing the progression of diabetes and preserving β-cell mass. Protection of β cells is also essential for regenerative therapies to enable survival and function of transplanted islets or stem cell-generated β cells. In this context, identification of factors with anti-inflammatory, anti-apoptotic and regenerative effects, is of interest.
The full-length INGAP protein expressed in the acinar tissue rendered mice resistant to streptozotocin (STZ)induced diabetes 34 . The protective effects of another INGAP transgene expressed in β cells correlated with downregulation of NADPH oxidase NOX1, one of the main mediators of the STZ-induced oxidative damage and DNA fragmentation in islets, thus suggesting that INGAP possesses anti-oxidative properties 35 .
The reported protective effects of INGAP in isolated islets are particularly interesting, suggesting its potential utility for islet transplantation but the mechanisms underlying these effects are unknown. The isolation process causes enzymatic, ischemic, and mechanical stresses that negatively impact islet survival 36,37 , induce cytokines and chemokines in the islets and activate nuclear factor kappa B (NF-κB), resulting in a strong inflammatory response 38,39 .
The anti-inflammatory effects of INGAP have not been investigated. We hypothesized that INGAP may offset the cytotoxicity of pro-inflammatory cytokines in β cells and tested our hypothesis using two β-cell lines RINm5F and INS-1 treated with IL-1β and IFNγ, commonly used in the models of cytokine-induced β-cell death 40 . IL-1β exerts its effects predominantly through the NF-κB pathway 41 , whereas IFNγ acts mostly through the JAK/ STAT1 pathway 42 but may also activate STAT3 43,44 and was shown to potentiate the effects of other cytokines [45][46][47] . NF-κB activation by IL-1β and IFNγ leads to stimulation of inducible nitric oxide synthase (iNOS) and production of nitric oxide (NO) directly implicated in β-cell apoptosis 46 . To characterize the anti-cytokine effects of INGAP and to investigate the underlying signaling events we used, as previously 17,18 , both INGAPpeptide and the full-length recombinant protein (rINGAP).
As summarized in Supplementary Fig. 1, Nos2 expression in RINm5F cells (3 h and 6 h) and NO levels in both cell lines (24 h and 48 h) were slightly induced by IL-1β and significantly potentiated by IFNγ, which alone was ineffective (A-C). Vice versa, IL-1β did not reduce viability, which was affected mostly by IFNγ (E). NO levels were significantly increased at 10× and 100× cytokine concentrations, whereas viability varied much less (D, F, G). We concluded that a cocktail of 100 pg/mL IL-1β+1 ng/mL IFNγ was sufficient to induce cytotoxicity in both cell lines.

Assessment of viability and measurement of NO production
Viability (cellular metabolic activity) was measured by MTT assay. Following treatment in 96-well plates, 20 μL/ well of 5 mg/mL MTT (Sigma-Aldrich) was added for 4 h at 37°C, 5% CO 2 . Formazan pellets were solubilized overnight at 37°C with 0.01 M HCl+10% SDS (100 μL/ well). Absorbance was measured at 570 nm using microplate spectrophotometer (Bio-Rad Laboratories, Mississauga, ON, Canada). All samples were tested in triplicates and each experiment was repeated 4-5 times. NO production was measured in 50 μL of cell culture media (duplicate samples) using the Griess assay kit (Promega, Madison, WI, USA), according to the manufacturer's instructions. Experiments were repeated three times.

Assessment of apoptosis by Annexin-V/PI FACS analysis
Following treatment, cells were trypsinized, washed with 1× phosphate-buffered saline (PBS) and stained using the FITC Annexin-V Apoptosis Detection Kit with PI (BioLegend, San Diego, CA, USA), following the manufacturer's instructions. Stained cells and controls (unstained cells, FITC Annexin-V only, and PI only) were then processed by fluorescence-activated cell sorting (FACS) (BD LSR Fortessa, LDI Flow Cytometry Services) and analyzed by FlowJo software. Each experiment was repeated three times.

Real-time quantitative RT-PCR
Cells were lysed in RLT buffer for RNA extraction with RNeasy or RNeasy Plus Kits in QIAcube (Qiagen, Toronto, ON, Canada). One microgram of RNA was reverse-transcribed using the Omniscript kit (Qiagen). Quantitative PCR was performed using custom-made primers (Supplementary Table 1), iQ TM SYBR ® Green master mix (Bio-Rad) in CFX96 TM Real-Time System (Bio-Rad). Normalized gene expression was calculated using the ΔΔC T method and Gapdh or multiple reference genes (Gapdh, α-tubulin, and β-actin) by the CFX Manager TM software (Bio-Rad). All experiments were repeated at least three to five times.

Protein extraction and Western blot analysis
Following the treatments with cytokines, INGAP, and/ or pharmacological inhibitors of signaling pathways (Supplementary Table 2 Table 3), secondary anti-mouse or anti-rabbit HRP-conjugated antibodies (Bio-Rad or Cell Signaling Technology), and developed using Clarity Western ECL (Bio-Rad). Chemiluminescence was acquired by ChemiDoc MP (Bio-Rad) and quantified by Image Lab software. Quantification of stain-free blots utilized total protein for normalization. Otherwise, α-tubulin was used as a loading control. Membranes were first incubated with phospho-antibodies, followed by mild stripping (0.2 M glycine, 0.1% SDS, and 0.05% Tween-20, pH 2.2) and re-probing with the corresponding non-phospho-antibodies.
Cytoplasmic and nuclear fractions were isolated using the Nuclear Protein Isolation-Translocation Assay Kit (FIVEphoton Biochemicals, San-Diego, CA, USA), following the manufacturer's instructions, and were analyzed as above. In addition to total lane protein (stain-free gels), α-tubulin and nuclear matrix protein p84 were used as respective loading controls.

Statistical analysis
Experiments were repeated three to eight times. The sample size (indicated in Figure legends for each set of data) was guided by the variation of effects and the statistical significance between groups. Results are expressed as means ± SEM of biological replicates. The data were normalized to control values (PBS or cytokines) to determine fold change. To determine data normality for each experiment, residual QQ plots were generated on R (Foundation for Statistical Computing). Data were considered normally distributed when points aligned strongly with the QQ line. Data were analyzed either by one-way analysis of variance (ANOVA) followed by Bonferroni post hoc test, two-way ANOVA followed by Bonferroni post hoc test or multiple t test comparisons corrected by Sidak-Bonferroni method. All analyses were performed using Prism 6.0 (GraphPad) software or Microsoft Excel. A P value < 0.05 was considered statistically significant for all comparisons.

rINGAP and Ingap-p protect β cells against cytokinemediated cell death
We first pre-treated cells with 1 nM rINGAP or 1.67 μM Ingap-p for 2 h prior to cytokines and measured their viability, apoptosis, and NO production after 24, 48, or 72 h. Most data were obtained in RINm5F cells, using INS-1 cells to confirm some findings. rINGAP significantly improved viability in RINm5F cells (MTT assay,  Fig. 3A, B and E, F).
Both rINGAP and Ingap-p downregulated the cytokineinduced expression of iNOS mRNA (Nos2) and protein in RINm5F (Fig. 2a-c) and INS-1 cells (Supplementary Fig.  4A, B) and reduced NO production (Fig. 2d), although rINGAP had a longer-lasting effect (up to 48 h) than Ingap-p. Together, these data show that both rINGAP and Ingap-p partially inhibit cytokine-induced apoptosis in β cells and that this correlates with inhibition of Caspase-3, iNOS expression, and NO production.
We determined next whether the timing of INGAP administration relative to cytokine treatment (±2 h) had an impact on its anti-apoptotic effect. rINGAP reduced cleaved caspase-3 when administered either pre-, co-, or post-cytokine treatment (Fig. 3a, c), whereas Ingap-p was effective in pre-and co-but not post-cytokine treatment (Fig. 3b, d). NO production after 48 h was significantly reduced by rINGAP and Ingap-p in all groups, however, rINGAP was more effective as co-and pre-treatment (Fig.  3e, f). We chose the pre-treatment mode to characterize the signaling mechanism by which rINGAP and Ingap-p elicit anti-apoptotic effects.
rINGAP and Ingap-p decreased the cytokine-induced phosphorylation of IKK and IkBα (Fig. 4f) but did not inhibit p65 nuclear translocation, which is necessary for exerting the effects of NF-κB signaling on the target genes. In contrary, nuclear p65 was increased in INGAP pre-treated cells (Fig. 4b, c). The cause of this is unclear since rINGAP itself does not induce p65 translocation (data not shown). However, rINGAP significantly decreased p65 phosphorylation at Ser 536 between 30 min and 6 h of cytokine treatment (Fig. 4d, e). Same effect was observed for Ingap-p at 6 h ( Supplementary Fig. 7A, B). These results suggest that INGAP may affect the NF-κB signaling through p65 phosphorylation, potentially modulating p65 transcriptional activity or stability. Interestingly, inhibition of P-p65 (Ser 536 ) was further increased when rINGAP was used together with IKKVII inhibitor ( Supplementary Fig. 7C, D), suggesting that besides IKK, INGAP may target other kinases implicated in p65 phosphorylation at Ser 536 53 .
Effects of rINGAP on NF-κB signaling is observed when cytokines are used together but not separately To better understand the impact of rINGAP on NF-κB signaling, we repeated the experiment with each cytokine separately. As expected, NF-κB pathway was induced by IL-1β but not by IFNγ, which, however, potentiated the effects of IL-1β on phosphorylation of IKK and IκBα (Fig.  5a-c), and on iNOS expression (Fig. 5e). p65 phosphorylation was not enhanced by IFNγ (Fig. 5d), indicating that it was under control of IL-1β only. Interestingly, all inhibitory effects of rINGAP were only observed for the cytokine cocktail but not IL-1β alone, even for p65. In the case of IKK, IκBα, and iNOS, the decreases were roughly equal to reversing the potentiation by IFNγ (Fig. 5a-c, e). This suggests that INGAP targeted the IFNγ signaling rather than IL-1β.
INGAP is more effective against the cytokine cocktail than individual cytokines in inhibiting STAT3, p38, and JNK JAK2/STAT1 signaling was activated by IFNγ, resulting in phosphorylation of JAK2 (Tyr 1007/1008 ), and STAT1 (Tyr 701 ) at 1 h and 6 h, whereas IL-1β had no effect. In contrast, STAT3 phosphorylation (Tyr 705 ), although induced mostly by IFNγ at 1 h, was driven mostly by IL-1β at 6 h (Fig. 6a). Used in combination, cytokines increased P-STAT3 up to 12 h of treatment ( Supplementary Fig.  5B). rINGAP had no effect on STAT1 phosphorylation (Fig. 6a, c) but slightly decreased P-JAK2 induced by IFNγ or both cytokines at 6 h (Fig. 6a, b). It also inhibited P-STAT3 at 1 h and 6 h (all combinations, Fig. 6a, d), but the differences were statistically significant only for both cytokines at 6 h, corroborating that rINGAP was more potent against the cytokine cocktail.
Phosphorylation of JNK (Thr 183 /Tyr 185 ) and p38 (Thr 180 /Tyr 182 ), most prominent at 30 min ( Supplementary Fig. 5B), was induced by IL-1β but not by IFNγ, which, however, potentiated the IL-1β effects (Fig.  7). rINGAP pre-treatment inhibited both kinases and, in keeping with earlier results, this was observed when IL-1β was used with IFNγ but not alone and was roughly equal to eliminating the effect of IFNγ.

Discussion
This is the first study to demonstrate the anti-apoptotic effects of recombinant INGAP and INGAP-peptide in the cytokine-treated β cells. Although pro-survival effects of (see figure on previous page) Fig. 1 INGAP pre-treatment partially inhibits cytokine cytotoxicity in RINm5F β cells. Cells were pre-treated for 2 h with 1 nM rINGAP or 1.67 μM Ingap-p and exposed to cytokines 100 pg/mL IL-1β and 1 ng/mL IFNγ. A Metabolic activity was assessed by MTT assay after 24, 48, and 72 h of cytokine treatment. Values were normalized to vehicle control (PBS) and expressed as fold change. All data represent the mean of at least three independent experiments ± SEM *p < 0.05 and **p < 0.01 vs. time-matched treatment with cytokine alone; multiple t test comparisons corrected by the Sidak-Bonferroni method. Here and thereafter rINGAP and Ingap-p are shortened to rING and Ing-p, respectively. INGAP have been reported 29,33 , the impact of proinflammatory conditions was not evaluated. In this work, we used a common model of cytokine-induced death that employs IL-1β and IFNγ, two cytokines most implicated in the development of diabetes 5,6 . In contrast to other studies, we also used each cytokine separately to better understand the underlying signaling events. Our main findings are: (1) both rINGAP and Ingap-p reduced cytokine-induced apoptosis, which correlated with inhibition of cleaved Caspase-3 and of the iNOS-NO axis; (2) rINGAP inhibited phosphorylation of p38, JNK, STAT3, and the constituents of NF-κB pathway, whereas nuclear translocation of NF-κB and the downstream gene expression was not inhibited; (3) rINGAP was more effective against the cytokine cocktail than each cytokine separately and appeared to reverse the potentiating effects of IFNγ; (4) expression of IFNγ-stimulated genes was downregulated by rINGAP when cytokines were used in combination but not IFNγ alone; (5) taken together, our data suggest that rINGAP affects the crosstalk between IL-1β and IFNγ signaling.
In keeping with our previous research 16,17 , we compared the effects of rINGAP and Ingap-p. Ingap-p, shown to reproduce the effects of the INGAP protein 11 , has been used in multiple studies worldwide [22][23][24][25][26][27][28][29] including clinical trials 19 . The interest in Ingap-p is obvious because it is synthetic and well-tolerated by patients 13,19 . However, as we have previously shown, rINGAP is more stable and over 100× more potent on a molar basis than Ingap-p, and it interacts differently with the cell surface, which may potentially lead to different outcomes 11,16,17 . Here again, we demonstrate a stronger potency for rINGAP versus Ingap-p in inhibiting apoptosis, improving viability (72 h vs 24 h), and reducing NO production (48 h vs 24 h). Importantly, rINGAP demonstrated a "rescue" effect when administered 2 h post cytokines, whereas Ingap-p was effective only in pre-and co-treatment modes. Whereas a greater stability of rINGAP could account for a Fig. 3 Effects of INGAP administration timing on cytokine-induced β-cell death. 1 nM rINGAP or 1.67 μM Ingap-p were administered: 2 h pre-or post-cytokines, or simultaneously with cytokines (co-treatment); A-D levels of cytokine-induced cleaved caspase-3 as a measure of apoptosis were determined by Western blot. A, B Shown are representative stain-free blots of cell lysates treated with rINGAP and Ingap-p, respectively, and C, D densitometric analysis presented as relative quantities of cleaved caspase-3 normalized to total lane protein and expressed as a percentage. *P < 0.05, **p < 0.01, and ***p < 0.001 vs. treatment with cytokines only; one-way ANOVA with Bonferroni corrected multiple comparisons and multiple t test comparisons corrected by the Sidak-Bonferroni method. E, F Levels of nitric oxide production were measured by Griess assay in RINm5F cells after 48 h cytokine exposure when E timing of 1 nM rINGAP or F timing of Ingap-p was varied. Values were normalized to cytokines only and expressed as a percentage. **P < 0.01 and ***p < 0.001 vs. treatment with cytokine alone; & ♦p < 0.05 vs. treatment with rINGAP post cytokines; oneway ANOVA with Bonferroni corrected multiple comparisons and multiple t test comparisons corrected by the Sidak-Bonferroni method. All data represent the mean of at least three independent experiments ± SEM. stronger improvement in viability, the "rescue" effect clearly points to some differences in the mechanism of action. These differences would be interesting to explore to potentially design a better peptide for future clinical use.
Among signaling pathways involved in cytokineinduced β-cell death, NF-κB plays a major role 55 , exerted mainly through activation of iNOS expression and NO production. Although cytokines may act in a NOindependent fashion 56 , the critical role of NO in IL-1β/ NF-κB-induced β-cell dysfunction and apoptosis has been well established 46,[57][58][59] . Inhibition of NF-κB signaling leads to cytoplasmic retention of p65 that halts its transcriptional activity 60 . As we showed, although rINGAP and Ingap-p decreased IL-1β + IFNγ-induced phosphorylation of IKK, IκBα, and p65, p65 nuclear translocation was not inhibited and even slightly enhanced. This result was unexpected, as it somewhat contradicted the inhibitory effects of INGAP on iNOS expression/NO production and on the aforementioned phosphorylation events. We then investigated by qRT-PCR the expression of several NF-κB-regulated chemokine genes 40,61 to test whether rINGAP inhibited the p65 transcriptional activity. Expression of NF-κB2, Sod2, Cxcl1 shown in Fig. 8a,   Fig. 4 Effect of INGAP pre-treatment on cytokine-induced p65 nuclear translocation and phosphorylation. RINm5F cells were treated with cytokines ± INGAP in the time-course experiments followed by extraction of cytoplasmic (C) and nuclear (N) proteins A-C or total proteins D-F and analyzed by Western blot. Shown are representative stain-free blots of subcellular lysates from cells treated with cytokines compared with A PBS; B rINGAP; and C Ingap-p. p84 was included as a loading control for nuclear fractions and total protein was used as a loading control for all samples. D p65 phosphorylation (P-p65) in cell lysates during time-course treatment of 6 h. E Densitometric analysis of each time point; values presented as P-p65 normalized to total lane protein and expressed as fold change over cytokines only. Data expressed as means ± SEM *P < 0.05, **p < 0.01, ***p < 0.001 vs time-matched cytokine cocktail; multiple t test comparisons corrected by the Sidak-Bonferroni method. F Analysis of IKK and IκBα phosphorylation after 1 h exposure to cytokines. Tubulin was used as a loading control. All data are representations from three independent experiments. plus a few others, such as Ccl2, Lcn2, Ccl5, Ccl19 (data not shown) upregulated by IL-1β ± IFNγ, but not by IFNγ, was not reduced by rINGAP. We, therefore, concluded that rINGAP did not inhibit NF-κB signaling.
The significance of the INGAP effect on p65 phosphorylation is less clear, as the role of phosphorylation in the IκBα-regulated p65 translocation and transcriptional activity is poorly understood. Aside from the IKK complex (mostly IKKβ) 52,62 , several kinases such as RSK1, MSK1, TAK1 have been shown to phosphorylate p65 on Ser 536 52 . There is evidence that phosphorylation may promote p65 disassociation from IκBα and p50, which carries a dominant nuclear export sequence 63,64 , causing a delayed nuclear translocation 65 or an increase in p65 turnover 66 . It is possible, therefore, that inhibition of p65 phosphorylation by INGAP contributed to accumulation and/or a delayed degradation of nuclear p65 observed in this study.
We show that p65 phosphorylation is induced by IL-1β and is not affected by IFNγ. It is not clear, however, why rINGAP inhibited p65 phosphorylation only when IL-1β was used together with IFNγ but not alone. It is possible that rINGAP and IFNγ synergistically inhibit one of the kinases that phosphorylate p65, whereas such a synergy is absent between INGAP and IL-1β but to verify this possibility a more-detailed investigation into the signaling pathways of IFNγ and rINGAP is needed.
IFNγ is often used with IL-1β to model the diabetic milieu, although its role in β-cell apoptosis and the signaling pathways leading to it are less understood. We show that IFNγ had no effect on the NF-κB-iNOS-NO axis but potentiated the action of IL-1β on the phosphorylation of IKK, IκBα, and significantly increased iNos expression. As rINGAP eliminated the IFNγ-potentiating effects, we reasoned that it targeted IFNγ signaling and Fig. 5 rINGAP pre-treatment inhibits NF-κB signaling when IL-1β and IFNγ used together but not individually. RINm5F cells were pre-treated with 1 nM rINGAP for 2 h and then exposed to 100 pg/mL IL-1β, 1 ng/mL IFNγ, or a cocktail of both. Equal amounts of protein were resolved on stainfree SDS-PAGE and analyzed by Western blot to assess changes in: A phospho(P)-IKK, and phospho(P)-IκBα, with densitometric analysis shown in B and C, respectively; D, E phospo(P)-p65 and F, G iNOS. Shown are representative blots from three independent experiments. Densitometric data in B, C and E are presented as relative quantities of phosphorylated/total protein ratio normalized to total lane protein and expressed as fold change of cytokine cocktail. G Values for iNOS were normalized by total lane protein and are shown as fold change of cytokines. Data expressed as means ± SEM *P < 0.05, **p < 0.01, ***p < 0.001 vs time-matched cytokine cocktail; multiple t test comparisons corrected by the Sidak-Bonferroni method.
screened for known IFNγ-responsive genes 54 . Of note, out of several genes tried, only three-Irf1, Socs1, and Socs3, were induced by IFNγ more than by IL-1β. We show that rINGAP indeed reduced the expression of these genes but, once again, for the cocktail of both cytokines and not IFNγ alone. rINGAP did not inhibit STAT1 phosphorylation, which is required for IFNγ signaling 67 . This suggests that INGAP targets neither IFNγ nor IL-1β specifically but modulates the crosstalk between the two, the precise mechanisms of which are yet to be elucidated.
Cytokine signaling involves multiple pathways. IL-1β has been shown to activate several signaling pathways including NF-κB, JAK/STAT, PI3K, and MAPK 40,41,48 . Likewise, IFNγ-induced signaling, previously thought to be limited to JAK/STAT1 pathway, has been expanded to include STAT3, STAT5, Src family of non-receptor tyrosine kinases, PI3K, NF-κB, MAP kinases that are recruited by IFNγR or JAK 42,68 . These pathways constitute a complex network that may account for multiple effects elicited by IFNγ alone, and in combination with IL-1β 45,46,67 . This complexity makes it difficult for us to pinpoint the possible target of INGAP, as we addressed only the most common pathways, such as NF-κB, JAK/ STAT, and MAPK. However, this is the first step that builds the base for further investigation.
In addition to the complexity of cytokine crosstalk, INGAP signaling is not well understood either. No specific receptor has been identified and the existing studies (see figure on previous page) Fig. 6 Effects of rINGAP on Jak2/Stat1/Stat3 phosphorylation in RINm5F treated with cytokines IL-1β, IFNγ individually, and in combination. RINm5F cells were pre-treated with 1 nM rINGAP for 2 h and then exposed to 100 pg/mL IL-1β, 1 ng/mL IFNγ, or a cocktail of both for 1 or 6 h. Equal amounts of protein were resolved on stain-free SDS-PAGE and Western blot analysis was used to assess changes in phosphorylation of Jak2, Stat1, and Stat3. A Shown are representative blots from three independent experiments and B-D densitometric analysis presented as relative quantities of phosphorylated/total protein ratio normalized to total lane protein and expressed as fold change of cytokines only. Values expressed as means ± SEM **P < 0.01 vs time-matched cytokine cocktail; multiple t test comparisons corrected by the Sidak-Bonferroni method. Fig. 7 rINGAP pre-treatment inhibits p38 and JNK activation when IL-1β used with IFNγ but not alone. RINm5F cells were pre-treated with 1 nM rINGAP for 2 h and exposed to cytokines for 30 min. Lysates were resolved on stain-free SDS-PAGE and probed for: A, B P-(phospho)-p38 and C, D P-JNK followed by pan-p38 and JNK antibodies. A, C Shown are representative blots of eight experiments and B, D the results of densitometric analysis shown as a ratio of phospho/total protein relative to the cytokine cocktail. *P < 0.05, **p < 0.005, ***p < 0.001 (t test). report a wide variety of INGAP-activated pathways, such as PI3K/Akt 15,17,26 , cholinergic pathway 26 , NF-κB 69 , or cAMP-PKA 70 . We have previously shown that rINGAP and Ingap-p stimulate proliferation in RINm5F cells by activating Ras-Raf-Erk1/2 pathway, which may also involve Src 17 . However, we did not observe activation of NF-κB, PKA, p38 MAPK, or PKC in RINm5F cells by either rINGAP or Ingap-p, which is perhaps cell type specific. The search for possible mechanisms by which INGAP exerts its anti-cytokine effects might be perhaps narrowed to the Ras-Raf-Erk1/2 and PI3K/Akt pathways, which should be investigated in more detail.
In summary, INGAP has been known for its regenerative properties and the ability to reverse diabetes in animal models 10,13,15,18,20,71 . Our work shows that INGAP also protects β cells from cytokine-induced apoptosis, thus providing new insights into the anti-diabetic potential of INGAP. The significance of these findings is determined by the need to develop β-cell survival therapies for the protection of the existing or transplanted β cells in the pro-inflammatory diabetic milieu. Although this study did not propose a definite mechanism of antiapoptotic action of INGAP, it built the foundation for further research that will also include human islets and other cytokines, e.g., TNFα, to better approximate the diabetic milieu and to better evaluate the therapeutic potential of INGAP.

Funding
This study was funded by the Jewish General Hospital Foundation (Montreal, QC, Canada) and by the Liana's Dream Foundation (to L.R.).

Ethics statement
Our study did not use animal or human samples or data, therefore it did not require ethical approval. Fig. 8 Effect of rINGAP on the expression of cytokine-stimulated genes. Rinm5F cells were pre-treated with 1 nM rINGAP for 2 h and treated with cytokines, individually or in combination for 3 h and processed for RNA extraction and qRT-PCR. Experiments were repeated five times. Fold change in gene expression over control (PBS) was calculated with the CFX Manager Software 3.1 using three reference genes (Gapdh, β-actin, and αtubulin). Target stability was within the range M < 0.5, CV < 0.25. *P < 0.05 vs PBS control: t test one-tailed, two-sample equal variance.