Constitutive Activation of β-Catenin in Conventional Dendritic Cells Increases the Insulin Reserve to Ameliorate the Development of Type 2 Diabetes in Mice

Obesity is now a current global epidemic, considered the most serious public health problem worldwide with a huge economic impact on health care providers (1). An important connection between obesity and inflammation has been recognized in recent decades, with several proinflammatory cytokines, chemokines, and immune cells implicated in the pathogenesis of obesity (2,3). Chronic overnutrition leads to excessive adiposity, characterized by adipocyte hypertrophy and dysfunction, which results in adipocyte cell death, dysregulated adipokine profile, proinflammatory immune cell infiltration, and subsequent tissue inflammation. The inability of adipocytes to store excess lipids in obesity causes the dysregulation of lipid and glucose metabolism leading to ectopic fat deposition, hyperglycemia, insulin resistance, and dyslipidemia (4).

Obesity is a high risk factor for the development of type 2 diabetes mellitus (T2DM), which arises due to the inability of pancreatic islets to produce sufficient insulin to compensate for the increased insulin resistance of the peripheral tissues (5). Metabolic insults, e.g., glucotoxicity, lipotoxicity, and endoplasmic reticulum stress, which are associated with the pathology of obesity, are widely thought to be the main cause of damage to insulin-secreting β-cells in the islets (6). However, inflammation initiated in the visceral adipose tissue (VAT) and later expanded locally to the pancreas is also considered a contributory factor in β-cell dysfunction (7). The deleterious effect of inflammation on β-cell function highlights the potential of immune modulation in the treatment of T2DM. Indeed, clinical studies using IL-1 antagonism have shown beneficial effects on blood glucose levels, β-cell secretory function, and insulin sensitivity (8,9). Thus, it may be possible to envisage that manipulation of immune responses in the VAT and/or the pancreas could prevent the disruption of β-cell function and ameliorate T2DM progression. We have recently reported that β-catenin signaling in conventional dendritic cells (cDCs) residing in VAT stimulates the production of IL-10 and contributes to the maintenance of an anti-inflammatory milieu in steady state (10). However, during obesity, β-catenin signaling in VAT cDCs diminishes partially, contributing to the development of tissue inflammation. Consequently, we hypothesized that sustained activation of β-catenin signaling in cDCs in obese VAT could revert the development of tissue inflammation and insulin resistance.

In this work, we demonstrate that constitutive activation of β-catenin in cDCs improves glucose homeostasis. This was caused not by an improvement in peripheral insulin sensitivity but, rather, by increased insulin production. In particular, we found that constitutive activation of β-catenin in cDCs enables the expansion of islet size in the pancreas due to increased proliferation while, in addition, generating an immunosuppressive environment for increased insulin release. This improved insulin reserve suggests a possible compensatory mechanism to prevent hyperglycemia in the development of T2DM.

C57BL/6 and BALB/c mice were purchased from Charles River laboratories, and Zbtb46^tm1Kmm/J and B6.Cg-Tg(TcraTcrb)425Cbn/J (OT-II) mice were purchased from The Jackson Laboratory. Zbtb46-Cre mice were kindly provided by M. Nussenzweig (The Rockefeller University) (11), and Ctnnb1^lox(ex3) mice were kindly provided by M. Taketo (Kyoto University) (12). Mice were housed at 22°C and 55% humidity, with a 12-h light/12-h dark cycle. Mice were fed a regular chow diet or AIN-76A Western diet (WD) (TestDiet) and given water ad libitum; animals were rehoused in clean cages weekly. Z-NEST (15 g; IPS) was used as nesting material. For chimeras, C57BL/6 male mice were γ-irradiated twice with 5 Gγ three hours apart. Mice were then reconstituted intravenously with bone marrow cells from Ctnnb1^lox(ex3)/+Zbtb46-Cre^/+ (gain-of-function [GOF]) and control Ctnnb1^lox(ex3)/+Zbtb46^+/+ and Ctnnb1^+/+Zbtb46-Cre^/+ sex-matched wild-type (WT) littermates. We found no differences in immune responses between Ctnnb1^lox(ex3)/+Zbtb46^+/+ and Ctnnb1^+/+Zbtb46-Cre^/+ mice and, therefore, both littermates were used as controls. All animal work was carried out in accordance with the U.K. government Home Office licensing procedures.

RNA from 100 mg VAT or from 20–30 islets was isolated using RNeasy Lipid Tissue Mini Kits (Qiagen) following the manufacturer’s instructions. RNA was quantified with SPECTROstar Omega (BMG LABTECH). Reverse transcription to cDNA was performed using High-Capacity RNA-to-cDNA kits (Applied Biosystems). Primer sequences can be found in Supplementary Table 1. Relative gene expression was calculated using the ΔΔCT method and normalized to a reference control (GAPDH) with control sample set as 1. The geometric mean was used as a more appropriate representation of ΔΔCT, which does not have a normal data distribution.

Immune cells from spleen, mesenteric lymph nodes (LN), and inguinal LN were isolated after digestion with Collagenase D (Roche) or Collagenase II (Sigma-Aldrich) and DNase (Sigma-Aldrich) for VAT-immune cells (gonadal and epididymal fat depot). All samples were stained with fixable Aqua Dead Cell Stain (Invitrogen) to exclude dead cells from analysis. Cells were stained for surface markers using the following antibodies: CD16/32-FITC, F4/80-PerCP, CD206-BV421, CD11c-BV605, CD103- allophycocyanin (APC), MHCII-AF700, CD11b-AF780, MertK-phycoerythrin (PE), CD45-PE-CF594, CD64-PE-Cy7, CD8-FITC, Ly6G-BV421, NK1.1-BV605, CD3-AF700, SiglecF-PE-CF594, CD4-PE-Cy7, B220- PerCP, CD44-e450, CD62L-BV605, TCRVα2-PE, FOXP3-PE, CCR4-PE, IFNγ-APC, IL-17–PE, TNFα-PE-Cy7, Nkx6.1-AF647, and Ki67-AF700 (see Supplementary Table 1). Samples were the analyzed by flow cytometry using an LSR Fortessa (BD Biosciences) and FlowJo, version 10, software.

Fasting adiponectin and insulin levels were measured by ELISA (Merck Millipore) and analyzed with a four-parameter logistic equation following the manufacturer’s instructions. For glucose (GTT) and insulin (ITT) tolerance tests, mice were fasted for 5–6 h and administered with 1.5 mg D-glucose/g body wt (Sigma-Aldrich) or 0.5 units of insulin/kg body wt i.p. (Actrapid), and blood glucose measurements were taken from the tail bleeds at 0,15, 30, 60, 90, 120, and 180 min after injection.

DC were isolated by CD11c⁺ bead positive selection (MACS; Miltenyi Biotec). Cells were plated at 1 × 10⁵ cells/well and cultured overnight, supernatant was collected, and levels of IL-10, IL-6, and CCL17 were determined by ELISA (Invitrogen). For ex vivo mixed-leukocyte reactions (MLRs), CD11c⁺ dendritic cells (DCs) were mixed at a ratio of 1:5 (5 × 10⁴ DCs: 2.5 × 10⁵ T cells per well) with allogeneic CD3⁺ T cells from BALB/c mice. Cells were incubated for 5 days, and proliferation was analyzed by carboxyfluorescein diacetate succinimidyl ester (CFSE) dilution.

For measurement of DC antigen presentation in vivo, mice were injected with 1 × 10⁷ i.v. CFSE-labeled ovalbumin (OVA)-specific OT-II cells and immunized with 200 μg OVA (Sigma-Aldrich) the following day. Three days later, T-cell proliferation was assessed by CFSE dilution.

For the isolation of islets, the pancreas was inflated with Collagenase-V (Sigma-Aldrich) and then incubated with Collagenase-V at 37°C for 10 min and separated by sucrose gradient. Three sized-matched handpicked islets were cultured overnight per well, and the supernatant was harvested the next day to measure insulin release. For cDC islet co- cultures, CD11c⁺MHCII⁺ cDCs were purified from Flt3L bone marrow cultures by cell sorting based on CD11c⁺ MHCII⁺ and B220⁻ expression using FACSAria, and 5 × 10⁴ cells/well were cultured with islets overnight in full medium with 13 mmol/L glucose according to nonfasting blood glucose concentration in mice with diet-induced obesity and treated or not with 1 µg/mL anti–IL-10 (BioLegend). For measurement of β-cell proliferation, eight handpicked islets were seeded per well with 5 × 10⁴ sorted cDCs and cocultured for 5 days. The islets were harvested, digested with Collagenase IV, and analyzed by flow cytometry.

VAT and pancreas were fixed in 4% paraformaldehyde and embedded in paraffin by BCI histopathology core services (Queen Mary University of London [QMUL]). Pancreas sections were immunostained with rabbit-polyclonal anti-CD3 (Dako), insulin (ICN Pharmaceuticals), Ki67 (Abcam), or cleaved caspase-3 (Cell Signaling Technologies) using a Ventana OmniMap DAB-HRP staining system (Discovery XT). Stained slides were scanned using a Panoramic 250 High Throughput scanner. Adipocytes and islet size were calculated using ImageJ Adiposoft software (13). All analysis was performed blinded. For immunofluorescence, sections were mounted on slides, and 10 mmol/L sodium citrate buffer, pH 6, at 100°C was used for antigen retrieval for 30 min. Tissue was stained with anti-mouse Ki67 antibody (Abcam) overnight at 4°C followed by secondary AF555 anti-rabbit IgG (Life Technologies) and for 1 h with AF647 anti-mouse Nkx6.1 (BD Bioscience). Slides were analyzed on an Axio Observer Z1 microscope (Zeiss) using AxioVision software (Zeiss).

In situ hybridization was performed as previously described (14). The mouse Wnt3a probe was kindly provided by Ybot-González and colleagues (15). The digoxigenin-labeled antisense probe Wnt3a was generated by in vitro transcription using RNA polymerase (Roche). Hybridization with 100 ng of the digoxigenin-labeled probe was carried out overnight at 65°C. Detection was achieved by colorimetric reaction using 4-Nitro blue tetrazolium chloride solution (Sigma-Aldrich) and 5-Bromo-4-chloro-3-indolyl phosphate disodium salt (Sigma-Aldrich).

Protein lysates were prepared from islets or VAT cDCs using CelLytic M (Sigma-Aldrich) and combined phosphatase and proteinase inhibitors (Pierce). Proteins were separated with SDS-PAGE and transferred to nitrocellulose membrane (GE Healthcare). Membranes were incubated with anti-mouse β-catenin antibody, anti-mouse β-actin antibody (Cell Signaling Technologies), or anti-mouse p65 antibody (Santa Cruz Biotechnology) overnight at 4°C. Subsequently, membranes were incubated with horseradish peroxidase secondary antibody (GE Healthcare). Density of the β-catenin or p65 bands was calculated relative to β-actin by ImageJ software.

Data reported in the figures represent the average of at least three independent experiments (Student two-tailed t test or one-way ANOVA as specified). For GTT and ITT, statistical significance was evaluated with two-way ANOVA followed by Bonferroni posttest. Data were analyzed using Prism 5 (GraphPad Software). Significant differences were denoted as *P < 0.05, **P < 0.01, ***P < 0.001.

β-Catenin signaling regulates preadipocyte differentiation (16). In mice, decreased Wnt10b levels in VAT from obese mice have been associated with tissue hyperplasia and decreased activation of the Wnt/β-catenin pathway in VAT cDCs (10). Thus, we investigated whether restoring β-catenin activation in cDCs during obesity could restrain VAT inflammation. To do so, we crossed Zbtb46-Cre^/+ with Ctnnb1^lox(ex3)/+ mice, where Lox/Cre-dependent deletion of exon 3 prevents the phosphorylation and degradation of β-catenin, rendering the β-catenin constitutively active. Due to expression of Zbtb46 in endothelial cells, mice died between 3 and 4 weeks of birth. Thus, we generated bone marrow chimeras where GOF or control bone marrow cells were transferred into irradiated C57BL/6 recipient mice (Fig. 1A). In this model, only cDCs express Zbtb46 and therefore have constitutively active β-catenin, which was confirmed by Western blotting (Fig. 1B).

No significant differences were observed in VAT or body weight between groups (Fig. 1C and D). Interestingly, systemic adiponectin concentration was significantly increased in WD-fed GOF compared with WT mice (Fig. 1E). Adiponectin is exclusively expressed by adipocytes and known to regulate a number of metabolic processes including glucose regulation and fatty acid oxidation (17). However, there was no change in adipocyte area between WD-fed GOF and WT mice, suggesting that adipocyte hypertrophy had not been reverted (Fig. 1F and G). Ectopic fat deposition in the liver was not altered between groups (Fig. 1H).

Next, we focused on the effect on local VAT inflammation. No significant changes in cell numbers or proportion of Tregs could be observed in spleen, LN, and VAT between groups when fed a chow diet (Supplementary Fig. 1B and C and data not shown). In analysis of WD mice, the total numbers of cDC1 in the spleen were significantly increased in the GOF mice compared with WT (Fig. 2A), in concordance with previous publications (18,19). Despite this effect, there was no change in numbers of cDC2 or macrophages in the spleen. Interestingly, there was a significant reduction in the number of splenic neutrophils, indicating a decrease in systemic inflammation in the GOF model (Fig. 2A).

This was also supported by the significant decrease in systemic levels of TNFα in the GOF mice (Fig. 2B). Similarly to the spleen, in VAT the cDC1 population displayed a trend toward an increase in these mice (Fig. 2C), although this was not observed in the LN (Supplementary Fig. 2A and B). Furthermore, a trend of increased numbers of Tregs of the CD4⁺ population was observed in VAT of GOF mice compared with WT (Fig. 2C), which was significantly increased in the mesenteric draining LN (Supplementary Fig. 2A). RT-PCR analysis from VAT showed a significant reduction in the expression of the proinflammatory cytokine IFNγ and chemokine CCL17 in VAT of GOF mice compared with WT (Fig. 2D). These results suggest that the constitutive activation of β-catenin in cDCs enables a T cell–driven immunosuppressive phenotype in local VAT diet–induced inflammation. The numbers of other immune cells were not significantly altered between the GOF and WT mice (Supplementary Fig. 2C and D).

The above data suggest that VAT cDCs in GOF mice exhibit a less activated phenotype compared with WT. cDCs play a central role in the initiation and shaping of T-cell immune responses. Therefore, we examined the antigen-presenting capacity of VAT cDCs ex vivo with MLRs and in vivo with OVA immunization. As expected, cDCs purified from obese GOF mice showed a decreased allostimulatory capacity compared with WT (Fig. 2E), indicated by a reduction in T-cell division. Furthermore, VAT cDCs displayed decreased antigen-specific T-cell activation, shown by decreased CFSE dilution of OT-II CD4⁺ TCRVα2⁺ cells (Fig. 2F). These differences were only observed in VAT but not in spleen (Fig. 2E and F). Confirming the previous findings of an altered T-cell phenotype in VAT, the IFNγ produced from CD4⁺ T-cells was significantly decreased in the MLR cultures of VAT cDC from the GOF model (Fig. 2G). Furthermore, expression of CCL17 from VAT cDCs cultured ex vivo was significantly decreased in the GOF model (Fig. 2H), while no significantly differences were detected in IL-6 and IL-10 production (Fig. 2I and J). CCL17 induces chemotaxis and cDC–T cell interactions, suggesting a reduction of inflammatory T-cell responses. No significant differences could be detected in steady state (Supplementary Fig. 1D).

Obesity-induced inflammation is linked to increased levels of insulin resistance and the development of T2DM. Systemic glucose homeostasis was measured in mice by intraperitoneal GTTs and ITTs. No differences in response to insulin or glucose were observed in mice fed a normal chow diet (ND) or after 4 weeks of WD (Supplementary Fig. 3A and B). However, by 16 weeks of WD, GOF mice had significantly improved glucose clearance compared with WT (Fig. 3A). Despite this, we did not observe differences in peripheral insulin sensitivity between groups, as assessed by ITT (Fig. 3B). This suggested that the improved glucose tolerance of GOF mice was due to changes in the insulin secretion of the pancreatic β-cells. No differences were observed under ND. However, a significant increase in insulin concentration in the plasma of GOF mice could be detected under WD (Fig. 3C). In addition to the increased basal level of insulin in the plasma of the GOF mice fed WD, these levels remained significantly higher 15 min post–administration of intraperitoneal glucose (Fig. 3D).

Islets contain endocrine cells that secrete hormones, predominately comprised of insulin-secreting β-cells. Using hematoxylin-eosin (H-E)-stained and insulin-immunostained sections (Fig. 3E and F), we calculated the islet size in the pancreas of the GOF and WT mice. There was a trend toward an increase in the mean islet size in the GOF mice compared with WT fed ND; however, in WD-fed mice, there was a significant expansion of islet size (Fig. 3G and Supplementary Fig. 3C). In addition, the distribution of islet size was modulated, indicating a significant decrease of smaller islets in the GOF model (Fig. 3H). For confirmation of the increase in plasma insulin in the GOF mice, sized-matched islets from mice fed WD were cultured overnight. Islets isolated from GOF mice demonstrated a significant increase in insulin content at high glucose concentrations compared with WT (Fig. 3I and Supplementary Fig. 3D). This suggests that islets from GOF mice had an increased capacity to release insulin at the cellular level. Therefore, despite changes in VAT inflammation, the improved whole-body glucose metabolism observed in GOF mice is due to modulation of insulin secretion in the pancreas.

To understand the expansion of the islets in GOF mice, we investigated cell proliferation by measuring Ki67. Using a coculture setup with islets from donor B6 mice and cDCs cultured from bone marrow of GOF or WT mice (Supplementary Fig. 4A), we observed an increase, albeit nonsignificant, in Ki67⁺ cells in the nonimmune population of the islets incubated with GOF cDCs for 5 days (Fig. 4A). Proliferation markers at the RNA level showed a modest change of cMyc and cyclin D1 expression in the islets from WD-fed GOF mice (Fig. 4B). Furthermore, immunohistochemistry staining demonstrated a significant increase in Ki67⁺ cells in the islets of GOF mice fed WD but not ND (Fig. 4C and D). This change in cell proliferation was observed while no change in apoptosis was measured by cleaved caspase-3 staining (Supplementary Fig. 4B). We further investigated the proliferating cell type using immunofluorescence staining, where we observed a significant increase in the proliferation of Nkx6.1⁺ β-cells in the islets of GOF mice fed WD (Fig. 4E and F). This was supported by in vitro cocultures assays, with an increase in number of Ki67⁺Nkx6.1⁺ β-cells in GOF mice (Fig. 4G). Although we have focused on β-cells, Ki67 expression could also be observed in a minority of Nkx6.1⁻ cells, and therefore these results warrant further investigation into the proliferation of other islet cells such as α-cells. In addition, Nkx6.1 is expressed in most but not all β-cells.

Wnt proteins are known to affect cell growth, and the expression of these is tightly regulated during development. Studies have reported that Wnt signaling in the pancreas is necessary and sufficient for β-cell proliferation (20). cDCs have been shown to stimulate the expression of Wnt ligands in the gut (21), suggesting a possible mechanism in which cDCs promote β-cell proliferation in the islet. Interestingly, in the GOF model there is a minor increase, though not significant, in islet Wnt3a expression by RT-PCR analysis (Fig. 4H). Expression of Wnt3a per islet was variable; however, it was significantly increased in the islets from GOF mice (Fig. 4I and J). This finding supports previous observations of Wnt3a involvement in β-cell proliferation (20).

The link between the development of T2DM and chronic low-grade inflammation is well reported. However, it is yet to be understood whether and how cDCs can directly affect islet inflammation and β-cell insulin secretion in T2DM.

Using Zbtb46^GFP mice, we demonstrated the presence of cDCs in both the pancreas and islets (Supplementary Fig. 4C and D). Islet-cDC cocultures were generated to examine the direct effect of cDC on insulin secretion (Supplementary Fig. 4A). This acute overnight incubation was designed to analyze the effects of inflammation on insulin secretion independently of β-cell proliferation and was performed in glucose-supplemented medium, as cDC function is impaired at low glucose levels. Insulin measured from GOF cDC cocultures was increased compared with that in WT controls (Fig. 5A), suggesting that WT cDCs exhibit a more activated phenotype than GOF cDCs and subsequently inhibit insulin release. Furthermore, when neutralizing IL-10, insulin secretion was significantly decreased in the GOF cocultures (Fig. 5B). This effect on insulin release can in part be explained by the increase in IL-10 produced from GOF cDCs (Fig. 5C), which could contribute toward generating an immunosuppressive environment in the islets, preventing deleterious effects on β-cell insulin secretion.

The above results suggest that modulation of in vivo insulin secretion in GOF mice can be controlled by local inflammation. To investigate this possibility, we studied the immune cell responses in the islets of GOF and WT mice fed WD. Immune population analysis demonstrated a general increase in cell recruitment in the islets of WD- compared with ND-fed mice (Fig. 5D and Supplementary Fig. 5A). No significant changes in T-cell numbers or proportion could be detected under ND (Supplementary Fig. 5A). In addition, we could not detect any visible T-cell clusters in islets from WT and GOF mice fed an ND (Supplementary Fig. 5B). Under WD, a reduction, although not significant, in CD3⁺ T-cell recruitment and significant increase in Tregs were seen in the GOF model compared with WT (Fig. 5D). This was also confirmed by immunohistochemistry staining of CD3⁺ T cells in the islets (Fig. 5E). The percentage of T-cell memory subsets were unchanged; however, in the GOF mice there was a significant reduction of IFNγ-producing CD4⁺ T-cells, which, combined with increased frequency of Tregs, indicates a change of T-cell phenotype (Fig. 5F). In addition, we observed a significant reduction in the expression of transcription factor nuclear factor-κB (NF-κB) p65 subunit, an indicator of less inflammation (Fig. 5G). Overall, islets in the GOF model have a healthier immune phenotype, which preserves an immunosuppressive environment in the islet during diet-induced inflammation.

Metabolic control is tightly entwined with the immune system. Studies are now exploring the effect of immune cells during the development of T2DM; however, the role of cDCs is yet to be established. In this work, we have shown a novel role of β-catenin activation in cDCs to enhance β-cell insulin secretion and modulate islet inflammation, thus providing an increased ability to control hyperglycemia in response to diet-induced obesity.

The Wnt/β-catenin pathway plays a fundamental role in cell proliferation and differentiation regulating adult tissue homeostasis and remodelling (22). It has previously been reported that constitutive activation of Wnt/β-catenin pathway in DCs is important for immune tolerance (21,23). We found that improved tolerogenic responses of VAT cDCs in GOF mice alter T-cell recruitment and the induction of a T-cell immunosuppressive phenotype. During obesity, a shift in CD4⁺ T-cell subsets has been well described, with the expansion of activated Th1 and Th17 populations and a reduction of Th2 cells and Tregs (24). Recently, a study has shown that in homeostatic conditions, VAT harbors a large population of memory T cells with long-term protective functions (25). However, there were no significant changes in the abundance of CD4⁺ T cells in VAT in our model; yet, VAT cDCs provide immunological tolerance to influence the T-cell immune phenotype in diet-induced tissue inflammation. Although VAT inflammation in part was reverted, this was not sufficient to revert insulin resistance.

Obesity-induced inflammation plays a crucial role in the development of insulin resistance and consequently T2DM (26). A decrease in insulin sensitivity initiates in VAT, where modulation of adipokines promotes chronic systemic exposure of proinflammatory mediators. These mediators disrupt insulin signaling in peripheral tissues, promoting insulin resistance and resulting in reduced glucose uptake from the blood. In this context, β-cells increase the production of insulin in an attempt to counteract high glucose levels. Thus, elevated insulin levels are ordinarily associated with the development of insulin resistance. Interestingly, in our model, improved whole-body glucose homeostasis was accompanied, paradoxically, by increased levels of circulating insulin. In this case, enhanced insulin levels were not an indication of increased insulin resistance but were a result of improved β-cell function and an enhanced ability to control hyperglycemia.

The metabolic alterations in GOF mice indicated changes in the insulin secretion of the β-cells in the pancreas. The number of functional β-cells is a main factor that contributes to the amount of insulin secreted in response to metabolic demand. In obesity, there is a dramatic increase in compensatory β-cell mass, with an expansion in response to the onset of diet-induced obesity (27). Understanding what regulates β-cell proliferation in obesity is a complex picture, and many factors have potential roles including nutrients, insulin, incretins, and growth factors (28). Interesting, we found that the control of hyperglycemia by GOF mice was due to an increase in islet size in these mice. Although the mechanism is not clear, it is possible that modulation in Wnt ligands, specifically Wnt3a, in the islets of GOF mice fed WD could enable the expansion of islet size by increasing β-cell proliferation.

Characterization of islet inflammation in T2DM has begun to reveal how immune cells affect β-cell insulin secretion. Increased numbers of leukocyte and proinflammatory mediators have been identified in islets of individuals with T2DM (29). Although transcriptomic analysis of islets has shown heterogeneity in the expression of cytokines and chemokines in T2DM individuals, it is widely accepted that they play a role in exacerbating islet inflammation (30). However, there remains a poor understanding of the role of cDCs in islets in T2DM. We have shown a size- independent effect where activation of β-catenin generates an immunosuppressive environment in the islet through IL-10 production and increased infiltration of Tregs. IL-10 has well-characterized anti-inflammatory properties, and emerging evidence suggests the cytokine can exert direct effects on β-cell function and viability (31,32). In addition, Tregs are known to promote systemic insulin sensitivity (33). At an islet-specific level the action of Tregs has yet to be explored in T2DM, but it is known that their numbers and function are decreased in the inflamed islets of patients with type 1 diabetes mellitus (34). The use of tolerogenic DCs for the treatment of type 1 diabetes mellitus has been explored with promising results where ex vivo functional manipulation of cDC is possible (35). This study opens up the possibility of manipulating DC function for the treatment of T2DM.

In summary, we have demonstrated a novel mechanism by which cDCs promote a healthier immune phenotype in the islets, which, in combination with an improved metabolic phenotype, enables the islets to provide an increased insulin reserve in response to diet-induced inflammation. Furthermore, this work suggests a possible new compensatory mechanism to prevent hyperglycemia in the development of T2DM.

Acknowledgments. The authors thank M. Nussenzweig (The Rockefeller University) for providing the Zbtb46-Cre mice. The authors thank George Elia at QMUL for all histology assistance and members of the Marelli-Berg laboratory at QMUL for technical support and scientific discussions.

Funding. This work was supported by the British Heart Foundation (FS/14/66/31293 to C.E.M., FS/13/49/30421 to M.P.L., and PG/16/79/32419 to M.P.L. and E.G.W.). Funding for M.C. and A.S. was from the Medical Research Council (MR/L002345/1 and MR/R022836/1). Funding for V.S. was from Action Medical Research (GN2272) and Barts Charity (417/2238).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. C.E.M., E.G.W., and C.G.-M. performed and analyzed experiments. C.E.M., C.G.-M., M.C., and M.P.L. designed experiments. C.E.M. and M.P.L. conceived the study. C.E.M. and M.P.L. wrote the manuscript. A.S. and V.S. provided technical support. M.M.T. provided the Ctnnb1^lox(ex3) strain. F.M.M.-B. contributed intellectually. C.E.M. and M.P.L. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 15th International Symposium on Dendritic Cells, 10–14 June 2018, Aachen, Germany.