Type-I interferons inhibit interleukin-10 signaling and favor type 1 diabetes development in NOD mice

Destruction of insulin-producing β-cells by autoreactive T lymphocytes leads to the development of type 1 diabetes. Type I interferons (TI-IFN) and interleukin-10 (IL-10) have been connected with the pathophysiology of this disease; however, their interplay in the modulation of diabetogenic T cells remains unknown. We have discovered that TI-IFN cause a selective inhibition of IL-10 signaling in effector and regulatory T cells, altering their responses. This correlates with diabetes development in NOD mice, where the inhibition is also spatially localized to T cells of pancreatic and mesenteric lymph nodes. IL-10 signaling inhibition is reversible and can be restored via blockade of TI-IFN/IFN-R interaction, paralleling with the resulting delay in diabetes onset and reduced severity. Overall, we propose a novel molecular link between TI-IFN and IL-10 signaling that helps better understand the complex dynamics of autoimmune diabetes development and reveals new strategies of intervention. Abbreviations ALN axillary lymph nodes IL-10 interleukin-10 MFI mean fluorescence intensity MLN mesentheric lymph nodes NOD nonobese diabetic mice PLN pancreatic lymph nodes TI-IFN type-1 Interferons Tmem memory T cells Treg regulatory T cells


Introduction
Type 1 Diabetes is a complex autoimmune disease characterized by the progressive 2 destruction of the insulin-producing β-cells in the pancreas by autoreactive T lymphocytes [1]. It protected from the disease onset when deficient in CD4 T cells [5,6] and enriched CD4 + cells 10 from diabetic donors are able to transfer the disease when administered into NOD-scid/scid 11 recipients [7]. However, the connection between environment and the activity of diabetogenic T 12 cells remains elusive. IFN-α in the subset of AIRE-deficient (APS1) patients that developed diabetes [10], induction of 18 diabetes in non-autoimmune prone C57BL/6 mice by overexpression of IFN-α in β-cells [11], 19 accumulation of high levels of TI-IFN in NOD mice [12], and delay of disease onset (and 20 decreased incidence) with early blockade of TI-IFN receptor signaling [13]. More recently, 21 Ferreira and colleagues reported that an IFN signature in PBMC of genetically predisposed 22 children was detectable before the appearance of islet-specific autoantibodies [14]. Despite these 23 observations, the mechanism(s) through which TI-IFN promotes T1D remains poorly 1 understood.

3
The cytokine IL-10 has an essential role in the development of autoimmune pathologies. [15] 4 Previous studies suggested that the low expression of this cytokine in the pancreas mediates the 5 occurrence of diabetes [16] and decreased IL-10 levels in serum of newly diagnosed children 6 with type 1 diabetes has been observed [17]. Monocytes/macrophages have been historically 7 investigated as the main target of this cytokine [18]; however, IL-10 acts also directly on T cells. 8 This has been shown in the context of naïve T cells activation and differentiation [19,20], in the 9 regulation of effector and memory T cells [21][22][23] and in the preservation of regulatory T cell 10 function [24,25]. 11 Here we report a novel effect of TI-IFN that causes a selective inhibition of IL-10 12 signaling in T cells thereby reducing their capacity to be regulated. This loss of signaling 13 correlates with the development of the disease in NOD mice. This effect is sustained but  disease development, we tested if these TI-IFN-exposed T cells would show any alteration in 7 their response to IL-10. To evaluate signal integrity, we quantified the accumulation of the 8 phosphorylated (active) form of the transcription factor STAT3 (P-STAT3, a key molecule in the 9 IL-10 signaling pathway) in response to ex vivo stimulation with IL-10. The response to the pro- 10 inflammatory cytokine IL-6 (that also induces phosphorylation of STAT3) was measured to 11 distinguish between cytokine-specific Vs nonspecific effects of TI-IFN exposure. We compared 12 multiple CD4 T cell subsets: naïve (CD4 + CD44 low Foxp3 -), memory (CD4 + Foxp3 -CD44 hi ) and IFNα; [13]). Independent repeats of these measurements indicated a statistically significant 16 reduction in IL-10 signaling in both Tmem and Treg from PLN and MLN compared to the 17 response of the same T cell subsets in the spleen (Fig. 1A&B) and ALN (not shown). In 4-week-18 old non-diabetes prone B6 mice, which do not accumulate TI-IFN in pancreatic and mesenteric 19 lymph nodes, Tmem and Treg preserved their ability to fully respond to IL-10 in all lymphoid 20 tissues (Fig. 1A&C). Importantly, this decrement in STAT3 phosphorylation was specific to IL-21 10 signaling, as the response to IL-6 was unaltered in the T cells of NOD (and B6) mice from all 22 the lymphoid tissues tested (Fig.1C). Together, these results suggested that in NOD mice there is 23 a selective reprogramming of the signaling for IL-10, actuated specifically in lymphoid tissues 1 shown to accumulate TI-IFN [13].
2 3 2.2 The impact of TI-IFN on IL-10 signaling is not a genetic characteristic of NOD T cells. 4 We then tested if this effect was unique to T cells of NOD background, or if bystander exposure 5 of any T cells to unusual levels of TI-IFN could affect their ability to be controlled by IL-10. 6 Bulk T cells from wt B6 mice were exposed to IFN-β (or IFN-α) for 48 hours, and then the 7 levels of P-STAT3 induced by stimulation with IL-10 or IL-6 were quantified via Phospho-flow.

8
Exposure to IFN-β (and similarly to IFN-α, not shown) induced a statistically significant 9 reduction of STAT3 phosphorylation after IL-10 stimulation in Tmem and Treg cells when 10 compared to the response in fresh or mock-treated cells (cultured without IFN-β, to exclude any 11 impact from the culturing conditions) (Fig. 2A&B). Reduction of IL-10 signaling responses were 12 dose-dependent upon IFN-β levels, reaching maximum inhibition at 5 ng/ml of IFN-13 β ( Supplementary Fig. 1). As observed in T cells from NOD mice, the levels of P-STAT3 in 14 response to IL-6 stimulation remained unaltered under all conditions (Fig. 2C), confirming that 15 this effect was not a generalized saturation of the Jak/STAT signaling pathway. the modulation of T cells can be done via assessment of its transcriptional impact. However, the 20 transcriptional impact of IL-10 on T cells is unknown. We therefore harnessed the vast 21 knowledge about IL-10 signaling in antigen presenting cells. Taking advantage of data from the 22 most recent publicly available RNAseq analysis of mouse macrophages exposed to IL-10 [26], 23 we selected a pool of 29 genes highly upregulated (> 6σ) (Supplementary Fig. 2A) as initial lead 1 for genes that could be also induced in T cells. Ten of these upregulated genes had known 2 functions in T cells ( Supplementary Fig. 2B). We then tested the expression of these genes as a Sphk1, Tarm1 and 2B4 -to be significantly upregulated in Tmem by in vitro treatment with IL-9 10 ( Fig. 3A), and two genes, Sphk1 and 2B4, were upregulated in Treg (Fig. 3B). We then 10 analyzed if the induction of these genes was affected by pre-exposure of these cells in vitro to 11 IFN-β. In Tmem, the increased expression of LIGHT, Sphk1, Tarm1 and 2B4 was completely 12 abrogated ( Fig. 3A). mRNA levels of Sphk1, LIGHT and Tarm1 also showed an important 13 decrease in IFN-β exposed Treg, while the expression of 2B4 was not affected (Fig. 3B). These  To explore what length of exposure to IFN-β is required to impact IL-10 signaling in T cells, wt 19 B6 bulk T cells were exposed to IFN-β for different lengths of time and the response to IL-10 20 and IL-6 in different subsets assessed by Phospho-Flow. In Tmem cells, a 24-hour exposure 21 significantly reduced the levels of IL-10-induced P-STAT3, but a 48-hour exposure was 22 necessary to achieve maximal inhibition (Fig. 4A). In Treg cells, a 24-hour exposure was 1 sufficient to achieve the maximal inhibition of IL-10-induced P-STAT3 signaling (Fig. 4A). 2 We also tested the reversibility of this inhibition. To address this question, after 48 hours of 3 exposure to IFN-β, T cells were washed, rested in cytokine-free media for 24 or 48 hours and the 4 P-STAT3 response to IL-10 or IL-6 was then measured. Within 24 hours of removing IFN-β, 5 Treg recovered their normal P-STAT3 response to IL-10 (Fig. 4B). The recovery of Tmem was 6 slower, showing only partial restoration of the IL-10 signaling even at 48 hours after removing 7 the IFNβ (Fig. 4B). These results indicate that the bystander effect of IFN-β requires a prolonged 8 exposure to instigate inhibition of IL-10 signaling and, with some kinetic differences between 9 Treg and Tmem, a normal P-STAT3 response to IL-10 in T cells can be restored following 10 removal of IFN-β. hours, IFN-β was added to the cultures for 48 hours. Following incubation and washing, an 20 additional 6-8 hours resting phase allowed the cells to recover their signaling after removal of the 21 Jak inhibitor ( Supplementary Fig. 3). The impact on IL-10 or IL-6 signaling was then quantified 22 via phospho-flow. In Tmem, Tofacitinib treatment resulted in a statistically significant 23 preservation of IL-10 signaling and Ruxolitinib had an even stronger effect (Fig. 5A). In Treg, 1 both inhibitors restored IL-10 signaling to the same extent (Fig. 5A), though the cells had to be 2 rested for 8 hours (instead of 6 hours as in the case of Tmem), as their recovery of cytokines 3 signaling after Jak inhibition was slower than in Tmem. Collectively, these results suggest that 4 Jak1, and possibly Jak2, are essential mediators for IFN-β−mediated alterations of IL-10-induced 5 P-STAT3 signaling in T cells. 6 Our data indicate that TI-IFN causes inhibition of IL-10 signaling through a process that requires 7 24/48 hours of exposure. This suggests that multiple intracellular molecular modifications are 8 needed to achieve this phenotype and the process could require the synthesis and activity of 9 additional extracellular mediators. To test this hypothesis, we employed a co-culture system with 10 T cells deficient for the receptor of TI-IFN (IFN-AR1 -/-; unable to respond to IFN-α or -β). 11 These cells have a response to IL-10 comparable to that displayed by wild type cells 12 ( Supplementary Fig. 4). We then exposed B6 wild type congenic T cells (CD45.1 + from B6/SJL 13 mice) mixed at 1:1 ratio with B6-IFN-AR1 -/-T cells (expressing the CD45.2 isoform) to IFN-β.  Downregulation of IL-10R surface expression would be a plausible mechanism to account for 1 the inhibition of IL-10 signaling following exposure to TI-IFN. However, flow cytometric 2 analysis of IL-10R surface expression did not support this hypothesis. IFN-β exposed T cells 3 (both Tmem and Treg), expressed levels of the receptor comparable to that of non-exposed cells 4 (Fig. 6A). In line with this, the comparison of IL-10R expression between NOD T cells isolated 5 from the PLN, MLN, ALN, and spleen showed no differences in the mean fluorescence intensity 6 ( Fig. 6B). This result suggested the involvement of an IL-10R specific regulator acting between 7 the receptor and STAT3 (as STAT3 remains available for the IL-6 receptor to be was significantly lower than that exhibited by B6 mice (Fig. 7A). Interestingly, in 2 week old B6 7 mice the response to IL-10 of Tmem and Treg cells in PLN and MLN was slightly lower than 8 that observed in the spleen (<100%), but it quickly recovered and stabilized (Fig. 7A).   Our results provide some important clues on the molecular mechanism behind the inhibition 20 of IL-10 signaling. The normal level of phosphorylation of STAT3 we measured in response to 21 IL-6 in T cells pre-exposed to IFN-β suggested that the impairment observed in IL-10 signaling Jak1/Jak2 inhibition in vivo is effective at preventing, and reverting, established insulitis in NOD mice [36,37]. Improving the efficacy and safety of this type of intervention would be a major 1 advancement for type 1 diabetes patients. is not properly regulated. The initial impairment in IL-10 signaling we observed in two-week-old 9 B6 pups (similar to, but not to the same extent as, that of NOD mice, Figure 7), could indicate an 10 initial adaptation phase of the newly generated pool of T cells (exposed to TI-IFN in the thymus) 11 [39] to variations in the intestinal flora during the breast-feeding phase. This would suggest that 12 the genetic predisposition of NOD mice encompass a defect in establishing the proper balance in 13 the response to microbiota derivatives in the gut-draining lymph nodes that ultimately affects the 14 regulation of diabetogenic T cells by IL-10. 15 We observed that the impact of TI-IFN is not specific to the NOD genetic background, 16 suggesting that this novel mechanism of alteration of immunoregulation could also contribute to 17 the development of other disorders with a TI-IFN signature. [40] This is a significant finding as timed and localized intervention) would achieve more successful therapeutic outcomes. To this 7 end, understanding how IL-10 modulates T cell function is necessary to identify the best strategy 8 to recover an appropriate level of regulation but, to date, very little is known. We report here for 9 the first time several genes (Sphk1, LIGHT, Tarm1 and 2B4) that IL-10 induces in T cells. We 10 used their expression as readout of IL-10 function, demonstrating the impact of pre-exposure to 11 TI-IFN. Future studies will be needed to understand their (and others') involvement in the 12 modulation of T cell functions. Moreover, their differential expression profile between Tmem 13 and Treg populations, suggests a distinct role in each population: a property that could also 14 reveal strategies to selectively impact these two subsets.

15
In summary, our study unveils the existence of a new molecular mechanism through which            interpretation, troubleshooting, and provided essential manuscript feedback. G.R is the guarantor of this work and, as such, had full access to all the data in the study and take responsibility for the 1 integrity of the data and the accuracy of the data analysis.  ratio between IFN-β exposed and not exposed (mock) cultured Tmem and Treg cells after IL-10 (B) or IL-6 (C) stimulation. Ratio MFI was calculated comparing the coefficient index of P- cells from C57BL/6 mice were exposed to Tofacitinib (25µM) or Ruxolitinib (5µM) for 2 h before 4 addition of IFN-β and then cultured for 48h (followed by a 6 to 8 h resting phase in cytokine-free 5 media). Their ability to respond to IL-10 (40 ng/ml) or IL-6 (40 ng/ml) was then measured by This mix was then cultured for 48h with or without IFN-β (5ng/ml), and rested in cytokine-free   Table supplementary

Figure Supplementary 1. Dose-effect of IFN- inhibiting IL-10 signaling in T cells.
Purified T cells from C57BL/6 mice were cultured for 48h in RPMI complete media in the absence (mock culture) or presence of different concentrations of IFN- (0.2-25ng/ml), and rested in cytokinefree media for six more hours. Cells then were either left untreated or stimulated with IL-10 (40 ng/ml) for 20'. After a fixation step, the levels of P-STAT3 in CD4 T cell subpopulations (Tmem: CD4 + CD44 hi Foxp3 -, Treg: CD4 + Foxp3 + ) were measured by phosphor-flow. The graph bars compare the percentage of P-STAT3 MFI ratio in Tmem and Treg cells between the groups exposed to IFN- and mock condition (considered as 100% of response) after IL-10 stimulation. Ratio MFI was calculated comparing the coeficient index of P-STAT3 after stimulation between each of the IFN- exposed groups and mock. Data of n=3 individual experiments are shown and expressed as % of Ratio MFI±SEM, *p<0.05, paired Student's t test. A two-tailed one-way ANOVA was performed to determine IL-10-treated versus untreated cells' differential gene expression. This differential expression was then evaluated for the standard deviation, σ or SD, of each gene's log2 fold change from the mean, of zero or unchanged. The 29 genes highly upregulated (>6SD) are indicated. (b) The table shows the 10 genes among the 29 indicated in (a), that we found also described in T lymphocytes. Characteristics and function (if known) of the protein encoded are indicated.  Freshly purified T cells from C57BL/6 and IFN-AR1 -/mice were either left untreated or stimulated with IL10 (40 ng/ml) or IL-6 (40 ng/ml) for 20'. The levels of P-STAT3 after stimulation in CD4 T cell subpopulations were measured by phosphor-flow. Representative histograms show P-STAT3 levels in Tnaive: CD4 + CD44 low Foxp3 -, Tmem: CD4 + CD44 hi Foxp3and Treg: CD4 + Foxp3 + cells after stimulation in both cell types. Results are representative of n=3 independent experiments. Table Supplementary 1