Altered connexin 43 expression underlies age dependent decrease of T_reg cell suppressor function in NOD mice

Abstract

Type I diabetes (T1D) is one of the most extensively studied autoimmune diseases but the cellular and molecular mechanisms leading to T cell-mediated destruction of insulin-producing β-cells are still not well understood. Here we show that T_reg cells in NOD mice undergo age-dependent loss of suppressor functions exacerbated by the decreased ability of activated effector T cells to upregulate Foxp3 and generate T_reg cells in the peripheral organs. This age-dependent loss is associated with reduced intercellular communication mediated by gap junctions, which is caused by impaired upregulation and decreased expression of connexin 43. Regulatory functions can be corrected, even in T cells isolated from aged, diabetic mice, by a synergistic activity of retinoic acid, TGF-β, and IL-2, which enhance connexin 43 and Foxp3 expression in T_reg cells and restore the ability of conventional CD4⁺ T cells to upregulate Foxp3 and generate peripherally derived T_reg cells. Moreover, we demonstrate that suppression mediated by T_reg cells from diabetic mice is enhanced by a novel reagent, which facilitates gap junction aggregation. In summary, our report identifies gap junction-mediated intercellular communication as an important component of the T_reg cell suppression mechanism compromised in NOD mice and suggests how T_reg mediated immune regulation can be improved.

Introduction

Type 1 diabetes (T1D) is a complex autoimmune disease resulting from destruction of pancreatic β-islets by infiltrating immune cells (1). Both environmental and genetic factors contribute to the development of the disease in humans and in NOD mice, which constitute an animal model of diabetes (2). Cellular mechanisms implicated in the progression of T1D include aberrant activation of autoreactive T cells, inefficient peripheral tolerance, and impaired function of immunoregulatory cells, especially CD4⁺ regulatory (T_reg) cells expressing the transcription factor Foxp3 (3). (1)

T_reg cells regulate homeostasis of the immune system and a variety of immune responses, including autoimmunity, and rely on numerous mechanisms that involve soluble, immunomodulatory cytokines, IL-10 and TGF-β, as well as membrane molecules and cell contact-dependent interactions (4, 5). T_reg cells may also sustain tolerance by converting effector CD4⁺ T cells into T_reg cells. This mechanism of “infectious” tolerance relies on genetic reprogramming of effector cells to become T_reg cells and involves intercellular transport of cAMP (6, 7). The determinants of which suppression mechanisms are used in particular immune contexts remain poorly understood.

Two major subsets of T_reg cells, thymus-derived (tT_reg) and peripherally-derived (pT_reg), were defined based on whether their suppressor function is acquired during normal thymic T cell development or following antigenic stimulation in peripheral lymphoid organs (8). In vitro-induced T_reg (iT_reg) cells may be generated by stimulation of CD4⁺ T cells in the presence of TGF-β and IL-2 (5). In vivo pT_reg cells are induced by a specialized population of dendritic cells in a process dependent on TGF-β and retinoic acid (RA) (9). Treatment of NOD mice with RA delayed the development of diabetes by inducing and expanding T_reg cells and by protecting islets from immune system-mediated destruction (10, 11).

Several lines of evidence directly showed that T_reg cells regulate autoimmunity in diabetes. Transfer of pT_reg or iT_reg cells into NOD mice, or in vivo induction of T_reg cells, can protect NOD mice from diabetes (12–14). Conversely, compromised function of T_reg cells was found to induce or exacerbate diabetes (15, 16). A number of genes associated with diabetes susceptibility loci regulate the survival and/or functions of T_reg cells (e.g. CTLA4, IL-2, STAT5) (17–19). Despite clear evidence of T_reg influence on T1D development, it remains controversial as to what the changes are in the T_reg population that actually contribute to the natural pathogenesis of diabetes in NOD mice. While some studies suggested a primary defect in the number and/or suppressor function of T_reg cells, other studies pointed to the resistance of effector T cells to T_reg-mediated suppression as a possible mechanism of autoimmune diabetes (20–25). Some of the discrepancies in the experimental results may stem from the use of different markers, (e.g. CD25 or Foxp3), to identify and isolate the T_reg population.

To better define the cellular and molecular basis of impaired T_reg function in diabetes we examined populations of these cells in young, prediabetic and aged, diabetic NOD mice expressing a Foxp3^GFP reporter that allows for unambiguous identification of T_reg cells. We have found that compromised suppression mediated by T_reg cells was associated with decreased ability of conventional T cells to upregulate Foxp3 and convert into iT_reg cells in aging NOD mice. We show that expression of connexin 43 (Cx43), a gap junction protein and one of the TGF-β-inducible genes, progressively declined in NOD mice progressing to diabetes. Gap junctions are essential for transporting cAMP from T_reg cells into target T cells, which initiates the genetic program of inhibiting T cell activation (7, 26). Here we find that dysregulated expression of Cx43 and alleviated cAMP signaling underlie progressive loss of T_reg suppressor function in NOD mice. This signaling defect and impaired iT_reg cell generation can be corrected by treatment of effector T cells with TGF-β, which promotes upregulation of Cx43, and RA, which regulates phosphorylation of connexin molecules and intercellular communication through gap junctions. Our data suggest that interactions requiring cell contact and intercellular communication are compromised in aged T cells in NOD mice. Finally, using a novel reagent that inhibits a PDZ-based interaction of Cx43 with the scaffolding protein zona occludens-1 (ZO-1), we demonstrate that suppressor function could be augmented even in T_reg cells isolated from NOD mice with diabetes.

MATERIALS AND METHODS

Mice

NOD mice expressing Foxp3^GFP reporter (NOD^GFP mice) were constructed as reported previously (27). A fragment of Foxp3 locus (located on BAC clone RP23-446O15) was modified to express GFP controlled by the Foxp3 regulatory sequences. Transgenic mice were produced in Joslin Diabetes Center at Harvard University by injecting NOD oocytes. Founders were identified by PCR of tail DNA.

All control mice were healthy, 2–4 week old NOD^GFP prediabetic females referred to in the text as young mice and diseased animals, referred to as diabetic, were 20-week-old or older females with diabetes (mice with blood glucose levels less than 120 mg/dL were considered healthy and those with levels higher than 300 mg/dL were considered diabetic). In some experiments, age-matched Foxp3^GFP reporter mice on the C57BL/6 (C57BL/6-Tg (Foxp3-GFP)90Pkraj/J; Jackson Labs) genetic background (B6^GFP mice) were used as additional controls. The incidence of diabetes in our colony was observed to be 85–90% for females and 15–20% for males.

Diabetes was induced in 5–6 month old female B6^GFP mice by streptozotocin injections. Streptozotocin (Sigma) was dissolved in 0.1 M citrate buffer (pH 4.5) and injected i.p. at a dose of 50 mg/kg/day for 5 days (28). Mice were sacrificed at day 14 after initial injection when blood glucose levels, measured for 3 consecutive days, were more than 350 mg/dL.

Full details of the study and all procedures performed on animals were approved by the Institutional Animal Care and Use Committee of the Georgia Regents University (approval #09-06-213) and complied with all state, federal, and NIH regulations.

Cell purification, flow cytometry analysis, and cell sorting

Cells from peripheral (axillary, brachial and inguinal), mesenteric, and pancreatic LN, spleens, and thymi were stained with antibodies specific for: CD4, CD8, CD25, CD44, CD62L, CTLA4, GITR, pSTAT5, Helios, and anti-rat CD2 (rCD2) (BD, eBioscience, BioLegend), and analyzed on a BD FACS-Canto instrument. Cells were sorted on a MoFlo cell sorter (Beckman Coulter)(purity of all sorts exceeded 98%). Data were analyzed using FlowJo (Treestar).

Detection of phosphorylated STAT5

pSTAT5 was detected in CD4⁺ T cells isolated from pancreatic LNs from young and old NOD^GFP mice and from B6^GFP mice by intracellular staining as described (29). Lymph node cells were cultured in complete media in the presence of IL-2 (10 ng/ml), stained with antibodies specific for CD4, CD8, CD25, washed, fixed, and stained with pSTAT5 antibody (pY694, BD).

Induction of Foxp3 expression

Sorted CD4⁺Foxp3^GFP− cells were activated with plate-bound anti-CD3 (10 μg/ml) and anti-CD28 (1 μg/ml) antibodies (BioLegend) in the presence of recombinant murine IL-2 (50 units/ml) and TGF-β (3 ng/ml) (both from Peprotech) and monitored daily for GFP and Foxp3 upregulation by flow cytometry and RT-PCR. Some wells contained RA (3 nM, Sigma). This experiment was done at least 5 times.

RT-PCR

Total RNA was isolated using the RNeasy Mini Kit (Qiagen) and converted into cDNA using the Superscript III kit (Invitrogen). Gene expression studies were performed using quantitative real-time PCR and TaqMan probes (Invitrogen). All data were normalized for β-actin expression and expression differences were shown as relative expression. Data were acquired with ABI7900HT instrument. All primers had been used previously (27, 30).

Analysis of Foxp3 expression in single cells was done as previously. Briefly, single cells were sorted from populations of CD4⁺Foxp3^GFP− and T_reg cells expressing low and high levels of Foxp3, directly into 96-well PCR plates containing reverse transcription buffer and cDNA was synthesized. Foxp3 and β-actin transcripts were amplifed by two rounds of PCR using nested primers: Foxp3 first round 5′GCCTGCCACCTGGGATCAAT3′ and 5′CACAGATGGAGCCTTGGCCA primers, second round 5′CACCCAGGAAAGACAGCAACC3′ and 5′CTTCTCCTTTTCCAGCTCCAGC3′, β-actin first round 5′GACATGGAGAAGATCTGGCACCA3′ and 5′CTCTTTGATGTCACGCACGATTTC3′ and second round 5′CCTTCTACAATGAGCTGCGTGTGGC3′ and 5′CATGAGGTAGTCTGTCAGGTCC3′ primers (30) (31).

Western blot

Sorted CD4⁺GFP⁻, CD4⁺GFP⁺, or CD4⁺CD25⁻ and CD4⁺CD25⁺ cells were lysed directly in SDS-loading buffer and resolved on 12% polyacrylamide gel (5×10⁴ cells/lane) and electrotransferred onto a PVDF membrane (Millipore). Membranes were probed with antibody specific for Foxp3 (1 μg/ml, eBio7979, eBioscience) as described (27). For Smad2 and phospho-Smad2 detection membranes were probed overnight (4°C) with the corresponding antibodies (Cell Signaling Technology). Images were scanned and analyzed using ImageQuant 5.2 software (GE).

Proliferation inhibition assay

Sorted CD4⁺Foxp3^GFP+ cells were used as T_reg cells and CD4⁺Foxp3^GFP− cells as responders as described (27). Responder cells (5×10⁴/well) were incubated on a 96-well plate with irradiated splenocytes (5×10⁴/well, 3000 rad) and soluble anti-CD3ε antibody (5 μg/ml). Various numbers of sorted CD4⁺Foxp3^GFP+ cells (0.25–5×10⁴/well) were added. Cells were sorted using MoFlo sorter. Proliferation was assessed by measurement of incorporated ³H thymidine added (1μCi/well) on the third day of a four-day culture. For some assays, Gap26 (100 μM, Anaspec) or αCT-1 (75 μM, American Peptide Company) peptides were added to the wells.

Measurement of IFN-γ

Flow cytometry sorted CD4⁺Foxp3^GFP− effector cells were mixed with increasing proportions of sorted CD4⁺Foxp3^GFP+ T_reg cells and irradiated splenocytes and stimulated with soluble anti-CD3ε antibody as described above. Levels of IFN-γ in the culture supernatant were measured in triplicates with commercial ELISA kit (eBioscience).

STAT5b and Cx43 overexpression

Full-length cDNA encoding a constitutively active form of STAT5b (STAT5b-CA)(gift of Dr. A. Farrar) or Cx43 (Open Biosystems) were cloned into retroviral vector pMx-IRES-rCD2 expressing extracellular rat CD2 domain (gift of Dr. M. Iwashima). Retroviral particles were prepared in Phoenix-Eco packaging cell line (gift of Dr. G. Nolan) and culture media were used to infect target cells. Sorted CD4⁺Foxp3^GFP− cells (5–10×10⁴) were activated with anti-CD3 and anti-CD28 antibodies and spin-infected with viral supernatant. Transduction efficiency was assessed by staining for rat CD2 extracellular domain. Transduced cells in some wells were incubated with IL-2/TGF-β or RA.

For retroviral transduction of T_reg cells, sorted Foxp3^GFP+ cells were activated overnight as above in the presence of IL-2 (50 units/ml). Cells were then infected with Cx43-encoding or control virus and further expanded on αCD3/αCD28-coated plates in the presence of IL-2. Two days later rCD2 expression was assessed by flow cytometry. T_reg cells expressing high level of rCD2 (>90%) were used as suppressors and freshly sorted CD4⁺Foxp3^GFP− cells were used as effector cells in proliferation inhibition assay.

Calcein loading and dye transfer

Induced T_reg and activated T cells were prepared from CD4⁺ T cells expressing trangenic TCR specific for pigeon cytochrome C antigenic peptide presented by I-A^b (32). Induced T_reg cells were loaded with 0.5 μM calcein violet, AM (Invitrogen) according to manufacturer’s instructions and stained with anti-CD4-APC. Responder cells stained with anti-CD4-PE were mixed with induced T_reg cells and bone marrow derived dendritic cells at 1:1:0.1 ratio in the presence (0.25 μM) or absence of antigenic peptide. Cells were seeded in round-bottom 96 well plate and calcein transfer into responder CD4-PE cells was recorded after 5 hours using LSRII flow cytometer (BD) equipped with violet laser.

Statistical analysis

All data are presented as the mean values ± s.d.. The significance of differences between samples or groups of mice was determined using paired, one-tailed Student t test. Differences between samples with p values ≤ 0.05 were considered significant. Statistical analsysis was done using Origin 9.0 software (OriginLab).

Results

The suppressor function but not cell numbers of T_reg cells decline in aged NOD mice

To uncover what cellular and molecular characteristics are associated with progression to disease we have studied phenotypes and an abundance of CD4⁺ T cell subsets in peripheral and pancreatic lymph nodes of young (2–4 week) NOD mice with normal glucose levels (120 mg/dL) and diabetic NOD mice (≥20 week) with overt diabetes (>300 mg/dL glucose). To facilitate identification of T_reg cells we have produced Foxp3^GFP reporter mice (NOD^GFP). A modified BAC clone previously used to generate Foxp3^GFP reporter mice in the C57BL6 strain (B6^GFP) was introduced into NOD oocytes (Fig. 1A–D) (27). Thus, the generation of our reporter mice did not involve backcrossing of transgenic founder animals onto a NOD genetic background. Transgenic reporter and wild type NOD mice had equivalent numbers, percentages and cell surface phenotypes of all T cell subsets, including CD4⁺CD25⁺ T cells, in thymus, lymph nodes, and spleen. Consistent with earlier reports, a normal, or even increased, proportion of Foxp3^GFP+ T cells was observed in NOD^GFP mice of every age, including young and diabetic mice (Fig. 1E, F) (24). Moreover, we observed a moderate increase of T_reg cells in peripheral and pancreatic lymph nodes. Despite stable proportions of T_reg cells, we observed phenotypic changes of T_reg populations: cells from diabetic mice expressed lower levels of GITR, CTLA4, and Helios, but higher levels of CD39 and PD-1 (Fig. 1G and not shown). Aged NOD^GFP mice had expanded populations of activated CD44⁺CD62L⁻ conventional CD4⁺ T cells in peripheral lymph nodes including pancreatic lymph nodes (Fig. 1H, I). We have also observed enhanced proliferative responses of Foxp3^GFP− effector cells stimulated in vitro, which correlated with elevated blood glucose levels (Fig. 1J, K).

FIGURE 1.

Analysis of T cell subsets in young and diabetic NOD mice expressing Foxp3^GFP reporter transgene. (A, B) Flow cytometry analysis of cells from thymus (A) and peripheral lymph node (LN) (B) of young and diabetic mice. Dot plots show expression of CD4 and CD8 on the populations of total thymic and lymph node cells (upper panels) and expression of CD25 and Foxp3^GFP on gated CD4⁺ cells (lower panels). (C) Analysis of Foxp3 expression in single cells expressing low and high levels of Foxp3 (left panel shows gating strategy used for cell sorting). Single cells were sorted onto 96-well plates and β-actin amplification was used to identify wells containing cells. (D) The Foxp3^GFP transgene is expressed exclusively in cells expressing endogenous Foxp3. Western blot (upper panels) and RT-PCR (lower panels) analysis of Foxp3 expression in total CD4⁺ T cells and CD4⁺Foxp3^GFP−, CD4⁺Foxp3^GFP+ subsets sorted from NOD^GFP mice, and in total CD4⁺ T cells and CD4⁺CD25⁻ and CD4⁺CD25⁺ subsets sorted from wild-type NOD mice. β-actin was used to normalize samples. (E) Expression of CD4 and Foxp3 in peripheral and pancreatic lymph nodes in young and diabetic NOD mice. (F) Proportions of Foxp3^GFP+ in populations of gated CD4⁺ T cells isolated from peripheral and pancreatic lymph nodes and pancreas of young and diabetic NOD mice. The absolute numbers of cells isolated from peripheral, pancreatic lymph nodes and from pancreas are shown as plot inserts. Cells isolated from inguinal, brachial and axillary lymph nodes were combined for peripheral lymph node analysis. Individual young mice are represented by squares (□) and diabetic mice are represented by triangles (△). Mean values and standard deviations are shown by bars.

(G) Expression of endogenous Foxp3, CTLA4, GITR and Helios in sorted populations of Foxp3^GFP− (solid line), Foxp3^GFPlo (broken line), and Foxp3^GFPhi (dotted line) cells. (H) Percentage of activated CD44⁺CD62L⁻ cells in the gated population of CD4⁺Foxp3^GFP− T cells isolated from peripheral (upper plots) and pancreatic (lower plots) lymph nodes of young (left panels) and diabetic (right panels) NOD^GFP mice. Flow cytometry plots show expression of CD44 and CD62L. (I) Percentage of activated cells in the gated population of CD4⁺Foxp3^GFP− cells in the peripheral and pancreatic lymph nodes. Individual young mice are represented by squares (□) and triangles (△) represent individual aged mice. Mean ± s.d. are shown. Asterisks (*) denote statistical significance (p<0.05). (J) CD4⁺ cells from aged NOD mice have higher proliferative potential. [³H]-thymidine uptake of anti-CD3 and anti-CD28 stimulated cells from young (light bar) and diabetic (dark bar) mice is shown. Cells were pulsed at day 3 and proliferation (CPM, count per minute) was measured after 18 hours. Experiment was repeated at least three times. (K) Blood glucose level measured in individual young (□) (3–4 week) and diabetic NOD^GFP mice (△) (13-week or older). Glucose levels in aged B6^GFP mice (◇) are shown for comparison.

Because disease progression was not associated with declining proportions of T_reg cells, we checked the ability of these cells to suppress the immune response. For that purpose, we examined T_reg cells expressing low and high levels of Foxp3 from young and diabetic NOD^GFP mice in a standard proliferation inhibition assay and compared these mice to diabetes resistant B6^GFP mice. We consistently observed decreased capacity of T_reg cells isolated from diabetic animals to inhibit proliferation of activated, conventional (CD4⁺Foxp3^GFP−) T helper cells from young NOD^GFP mice (Fig. 2A). However, T_reg cells from both young and diabetic NOD^GFP mice had lower suppressor capacity than corresponding cells from B6^GFP mice. In a reciprocal experiment, T_reg cells from young NOD^GFP mice were able to significantly inhibit proliferation of effector CD4⁺ T cells from older, diabetic animals. Finally, proliferation inhibition was smallest when T_reg and responder cells were both isolated from aged, diabetic mice (Fig. 2B). To examine if hyperglycemia could alter T_reg cell functions we have induced diabetes in 4–5 month old B6 mice by streptozotocin injection. This experiment shows comparable inhibition of proliferation and IFN-γ production by cells isolated from normal and hyperglycemic mice demonstrating that observed differences in T_reg suppression are likely not caused by differences in blood glucose levels (Fig. S1A, B). Together, our data demonstrate that functional, not quantitative, changes of T_reg cells are associated with disease progression in NOD mice. Higher proportion of T_reg cells from old NOD mice than from young and aged B6 mice is needed to achieve the same level of suppression, suggesting that immune environment permissive for autoimmune disease already exists before overt diabetes.

FIGURE 2.

T_reg cells from aged NOD^GFP mice lose suppressor function. (A) T cell proliferation inhibition assay with T_reg cells sorted from young and diabetic NOD^GFP mice. Circles, triangles and diamonds represent T_reg cells from young and diabetic NOD^GFP mice and B6^GFP mice, respectively, expressing low (● ▲ ◆) and high (○ △ ◇) levels of Foxp3. Gating strategy to isolate T_reg cells expressing low and high levels of Foxp3 is shown on Fig. 1C. CD4⁺Foxp3^GFP− responder cells were sorted from young NOD^GFP mice. Asterisks (*) denote data points statistically different (p<0.05) between young and aged NOD^GFP mice. The figure represents one of at least three independent experiments. (B) Conventional CD4⁺Foxp3^GFP− cells from diabetic NOD^GFP mice remain sensitive to suppression by T_reg cells from young NOD^GFP mice and to much lower extent by T_reg cells from diabetic mice. Proliferation of CD4⁺Foxp3^GFP− responder cells sorted from diabetic NOD^GFP mice is inhibited by T_reg cells expressing low (● ▲) or high levels of Foxp3 (○ △) sorted from young (abbreviated to “y”)(left panel) and diabetic (abbreviated to “d”)(right plot) NOD^GFP mice. The experiment was repeated two times. Error bars denote mean ± s.d.

Effector CD4⁺ T cells and Foxp3^GFPlo T_reg cells from diabetic mice have decreased ability to upregulate Foxp3 expression when activated in the presence of IL-2 and TGF-β

We have routinely observed that a small, but consistent proportion of sorted naive CD4⁺ T cells from B6^GFP mice upregulated Foxp3 when activated in vitro without exogenous IL-2 and TGF-β (30). In NOD mice we noticed that a larger fraction of effector cells from young NOD^GFP mice than from older mice upregulated Foxp3 when cells were activated in vitro (Fig. 3A). This age-dependent decline in the ability of effector T cells to upregulate Foxp3, not seen in aging B6^GFP mice, suggests reduced and delayed capacity to generate iT_reg cells in aged NOD mice (Fig. 3B).

FIGURE 3.

Stability and kinetics of Foxp3 upregulation in in vitro activated CD4⁺ T cells isolated from young and diabetic NOD mice and aged B6 mice. (A) The percentage of CD4⁺ T cells isolated from peripheral lymph nodes of young (□) and diabetic (△) NOD^GFP mice and B6^GFP mice (◇) that upregulate Foxp3 upon activation. Sorted CD4⁺Foxp3^GFP− cells were activated with plate-bound anti-CD3 and anti-CD28 antibodies and analyzed after 3 days. Individual mice are represented by squares, triangles and diamonds and mean ± s.d. are shown for each group of mice. (B) Kinetics of Foxp3 upregulation in CD4⁺ T cells isolated from young and diabetic NOD^GFP mice and B6^GFP mice. CD4⁺Foxp3^GFP cells sorted from young (light grey) and diabetic (grey) NOD^GFP and B6^GFP (dark grey) mice were activated with plate-bound anti-CD3 and anti-CD28 antibodies in the presence of IL-2 and TGF-β and expression of Foxp3 was quantitated with TaqMan probes after 1, 2 or 3 days of a culture. Experiment was repeated two times. (C, D) Stability of Foxp3 expression in T_reg cells from young (C) and diabetic (D) NOD^GFP mice. Sorted CD4⁺Foxp3^GFP− (4×10⁵ cells, upper row), Foxp3^GFPlo (2.5×10⁵ cells, middle row), and Foxp3^GFPhi (2.5×10⁵ cells, lower row) cells from young and diabetic mice were stimulated in vitro with plate-bound anti-CD3 and anti-CD28 antibodies in the absence (columns one and three) or in the presence of IL-2 (50 u/ml) and TGF-β (3 ng/ml)(columns two and four) and analyzed after 3 days. In vitro stimulated cells proliferated and their numbers increased 1.6±0.4, 2.3±0.4, 1.7±0.4 fold for CD4⁺Foxp3^GFP−, Foxp3^GFPlo and Foxp3^GFPhi cells stimulated with anti-CD3 and anti-CD28 in the absence of IL-2 and TGF-β and increased 1.9±0.3, 3.1±0.2, 1.8±0.1 fold for the respective populations stimulated in the presence of cytokines. Gating strategy to isolate Foxp3^GFP−, Foxp3^GFPlo and Foxp3^GFPhi cells is shown on Fig. 1C. Rectangles show gates used for sorting (solid lines) and cell populations that up- or downregulated Foxp3 expression (broken lines). Numbers denote percentages of cells. Data from one experiment of three is shown.

Consistent with our previous analysis of B6^GFP reporter mice, we examined whether the population of T_reg cells in NOD^GFP mice is heterogeneous, with cells being differentially sensitive to antigen/cytokine stimulation. To examine how Foxp3 is regulated in T cell subsets in young and aged NOD^GFP mice, we activated conventional CD4⁺Foxp3⁻ cells and Foxp3^lo and Foxp3^hi T_reg cells with plate bound anti-CD3/anti-CD28 antibodies without and in the presence of TGF-β and IL-2 (33). A large proportion of Foxp3^hi T_reg cells, which constituted a majority of T_reg population, preserved a stable T_reg phenotype in young and old NOD^GFP mice when activated (Fig. 3C, D). In contrast, conventional Foxp3⁻ cells and Foxp3^lo T_reg cells from aged NOD^GFP mice have a decreased ability to upregulate and/or maintain Foxp3 expression compared to the respective populations from young mice. In particular, the Foxp3^lo subset in aged mice had an unstable T_reg phenotype, and only a small proportion of these cells was able to sustain Foxp3 expression when stimulated with anti-CD3/anti-CD28 antibodies even in the presence of TGF-β (Fig. 3D). Altogether, our results suggest that CD4⁺ T cell populations, identified in previous reports as precursors of pT_reg cells, are most affected by the disregulated expression of Foxp3 in aging NOD mice.

Disregulated upregulation/expression of the Foxp3 in aged NOD mice is associated with decreased signaling through STAT5 and Smad2/3

To delineate molecular mechanisms of age-related weakening of T_reg suppression, we examined regulation of Foxp3 expression by IL-2, TGF-β. IL-2 and TGF-β are cytokines critical for generation and maintenance of peripheral T_reg cells (34). Previous studies have shown that components of the IL-2 signaling pathway map to diabetes susceptibility loci (19, 35, 36). Mutations of STAT5, a component of the IL-2 receptor signaling pathway, were associated with impaired function of T_reg cells in NOD mice (18). Similarly, TGF-β, acting through Smad2/3, was found to regulate genes controlling T_reg functions and insulin-specific T_reg cells (37, 38). Decreased expression and/or phosphorylation of STAT5 or Smad2/3 could suggest impaired signaling by IL-2 and TGF-β receptors and explain decreased T_reg suppression in aging NOD mice. To test this hypothesis we studied protein expression and signaling induced phosphorylation of STAT5 or Smad2/3 in young and diabetic NOD mice.

As shown on Fig. 4A and 4C T cells from peripheral and pancreatic lymph nodes of young and diabetic NOD^GFP mice express similar levels of STAT5, but much lower levels of Smad2/3. When activated in the presence of IL-2 and TGF-β decreased levels of STAT5 and Smad2/3 phosphorylation were found in T cells from diabetic mice (Fig. 4B, D). When compared to B6^GFP mice, decreased STAT5 phosphorylation was found even in CD4⁺ T cells from young mice (Fig. 4B). A recent report demonstrated that mice with a constitutively active form of STAT5 have more T_reg cells (39). To determine if increased STAT5 signaling promotes Foxp3 upregulation, CD4⁺ T cells from NOD^GFP and B6^GFP mice were transduced with a retroviral vector expressing a constitutively active STAT5 mutant (gift of Dr. M. Farrar) and activated in the presence of IL-2 and TGF-β (40). Enhanced signaling mediated by STAT5 resulted in an increased proportion of cells upregulating Foxp3 only in effector cells isolated from B6^GFP mice (Fig. S2). Collectively, examining STAT5 and Smad2/3 shows that impaired signaling by IL-2 and TGF-β likely contributes to age-dependent decrease of T_reg functions in diabetic NOD mice.

FIGURE 4.

IL-2 and TGF-β signaling pathways are altered in diabetic NOD^GFP mice. (A) Total STAT5 protein level in total CD4⁺ cells isolated from young and diabetic NOD^GFP mice. (B) Phosphorylation of STAT5 was assessed by flow cytometry in total CD4⁺ T cells isolated from pancreatic LN of young (solid line) and diabetic (broken line) NOD^GFP mice and control B6^GFP mice (dotted line) stimulated for 15 minutes in the presence of IL-2 (10 ng/ml). Filled histogram shows staining with isotype matched antibody. Percentages of positive cells from each mouse are shown. (C, D) Expression of Smad2 (C) and its phosphorylated form (D) in young and diabetic NOD^GFP mice. CD4⁺Foxp3^GFP− cells sorted from pancreatic lymph nodes were lysed directly or stimulated with TGF-β (1 ng/ml) for 30 minutes and then lysed in SDS gel-loading buffer. β-actin was used as a control for equal loading. Bars represent comparison of protein quantities. Experiment was repeated twice.

Decreased expression/upregulation of Cx43 in old NOD mice underlies impaired suppressor function of T_reg cells

In an attempt to restore Foxp3 upregulation in T cells isolated from diabetic NOD^GFP mice to the levels observed in young NOD^GFP mice, we treated naive cells isolated from diabetic NOD^GFP mice with RA. As shown in Fig. 5A, addition of RA to T cells activated in the presence of IL-2 and TGF-β resulted in a significant increase in the proportion of iT_reg cells. Effector cells from both young and diabetic NOD^GFP mice upregulated Foxp3 with the same kinetics and produced the same percentage of iT_reg cells at the conclusion of culture.

FIGURE 5.

Decreased suppression of T_reg cells in NOD mice is associated with altered expression of Foxp3 and Cx43. (A) Coordinated signaling of retinoic acid and TGF-β enhances the ability of CD4⁺ T cells from diabtetic NOD mice to upregulate Foxp3 expression in vitro. CD4⁺Foxp3^GFP− cells sorted from young (solid line) and diabetic (broken line) NOD^GFP mice and control B6^GFP mice (dotted line) were stimulated for 3 days with anti-CD3/anti-CD28 antibodies in the presence of IL-2 (50 u/ml) and TGF-β (3 ng/ml)(left panel) or IL-2, TGF-β and RA (3 nM)(right panel). Representative plots from at least three experiments with 3–4 mice each are shown. (B) Expression levels of Cx43 regulate intercellular communication between iT_reg and target CD4⁺ T cells. Histograms show calcein fluorescence of target CD4⁺ T cells expressing two, one or no functional Cx43 gene alleles. Target CD4⁺ T cell were activated with PCC50V54A (0.25 μM) antigenic peptide presented by A^b expressing bone marrow derived dendritic cells in the presence of calcein loaded iT_reg cells. (C) Proliferation inhibition assay using Cx43-sufficient responder cells and Cx43 sufficient (diagonal lines) and deficient (horizontal lines) iT_reg cells and Cx43 sufficient natural T_reg cells (crossed lines). Solid bar represents proliferation of T cells (T_c) in the absence of T_reg cells. T_reg (abbreviated to T_r) and responder T cells (T_eff) were used in a ratio of 1:2 or 1:1. (D) Decreased expression of Cx43 in diabetic NOD^GFP mice. Cx43 expression levels were measured by quantitative PCR in freshly sorted naive (CD44⁻CD62L⁺Foxp3^GFP−) and activated (CD44⁺CD62L⁻Foxp3^GFP−, act) conventional CD4⁺ T cells and in T_reg cells expressing low (Fp3^lo) and high (Fp3^hi) levels of Foxp3. (E) Upregulation of Cx43 expression in sorted naive CD4⁺ T cells activated in vitro with plate-bound anti-CD3/anti-CD28 antibodies in the presence of IL-2, TGF-β and RA. (F) Connexin 43 expression in sorted CD4⁺Foxp3^GFP+ T_reg cells activated with anti-CD3/anti-CD28 in the presence of IL-2 for 3 days. Representative experiment of three is shown. Quantitative PCR data was obtained with TaqMan probes. Western blot analysis confirmed that Cx43 protein levels in CD4⁺ T cell subsets from B6 and NOD mice activated in vitro increased, especially in cells activated in the presence of TGF-β (data not shown). Light gray bars in plots D, E and F represent cells isolated from young NOD^GFP mice, gray bars – cells isolated from diabetic NOD^GFP mice and dark grey bars – cells from B6^GFP mice.

One of TGF-β and RA-induced genes essential for T_reg development, Foxp3 expression, and contact-dependent suppression is Cx43 (26, 41). This gene is not expressed in naive CD4⁺ T cells in steady state mice but it is expressed in cells with activated phenotype and in T_reg cells, especially T_reg subset expressing low levels of Foxp3.

Cx43 is a major gap junction protein in T lymphocytes and intercellular transport of cAMP between activated T_reg and target CD4⁺ T cells was shown to inhibit upregulation of CD69, IL-2 gene expression and cellular proliferation (26). Co-culture of target CD4⁺ T cells with T_reg cells loaded with a fluorescent dye, calcein, demonstrated that inhibition of activation was proportional to the magnitude of calcein transfer and was blocked by gap junction inhibitors. To show that the level of Cx43 expression correlates with gap junction activity we compared dye transfer between calcein loaded, induced T_reg cells, and responder CD4⁺ T cells isolated from B6^GFP mice harboring two, one or no functional alleles of Cx43 gene in CD4⁺ T cells (41). To mimic natural interactions between T cells and antigen presenting cells we used T_reg and responder cells expressing transgenic TCR and activated with antigenic peptide presented by bone marrow derived dendritic cells. Fluorescence of responder cells correlated with Cx43 expression, revealing that gap junction mediated transport depends on Cx43 expression (Fig. 5B). This result is consistent with a report demonstrating intercellular transport mediated by Cx43 recruited into immunological synapse (42). To further examine Cx43 involvement in T_reg suppression we have generated induced T_reg cells from Cx43-sufficient and deficient CD4⁺ T cells. Induced T_reg expressing natural levels of Cx43 proved to better inhibit activation of Cx43 sufficient T cells (Fig. 5C). In summary, previously published and current data show how levels of Cx43 regulate intercellular, gap junction mediated transport and impact T_reg cell mediated suppression.

To study how Cx43 could contribute to altered T_reg function we examined its expression pattern in CD4⁺ T cells in young and diabetic NOD mice. Analysis of CD4⁺ T cell subsets sorted directly from experimental mice established a baseline for Cx43 expression and revealed low expression in activated and Foxp3^lo T_reg cells in diabetic NOD^GFP mice (Fig. 5D). Moreover, Cx43 is upregulated to much higher levels only in activated and T_reg cells from young, but not diabetic NOD^GFP mice, especially in the presence of TGF-β and in the absence of RA (Fig. 5E, F).

Age related changes in the Cx43 expression coincide with the loss of T_reg function and suggest that the T_reg suppression mechanism that relies on generation and intercellular transfer of cAMP might be progressively compromised in aging NOD^GFP mice (7). To further elucidate this mechanism, we compared expression of adenylyl cyclases and phosphodiesterase, enzymes that regulate cAMP levels, in young and diabetic NOD^GFP mice (43). Fig. 6A shows that expression of adenylyl cyclases 4, 7, and 9, major enzymes that increase intracellular cAMP concentration in T lymphocytes, are low in resting T_reg cells and are upregulated upon activation in T_reg cells isolated from young and diabetic NOD^GFP mice. Despite lower expression of cyclases in activated T_reg cells from diabetic NOD^GFP mice, especially of adenylyl cyclase 7, all cyclases are significantly upregulated when compared to unstimulated T_reg cells. Expression of phosphodiesterase 3b, a major cAMP-degrading enzyme in T_reg cells does not increase in cells from diabetic NOD^GFP mice upon T_reg activation (Fig. 6A) (44). Since transcription of the phosphodiesterase gene is downregulated by Foxp3, our data suggest that Foxp3 is still able to control transcription of its target genes even in aging NOD^GFP mice. Altogether, we conclude that, despite finding dysregulated expression of one adenylyl cyclase, expression of enzymes that control balance between cAMP generation and hydrolysis does not considerably change between young and old NOD^GFP mice. Thus, low expression of Cx43 in diabetic NOD^GFP mice likely results in decreased intercellular transport of cAMP, which could contribute to decreased suppressor function of T_reg cells in these mice.

FIGURE 6.

Impact of cAMP production and intercellular transport on the suppressor function of T_reg cells. (A) Expression of adenylyl cyclases (Adcy) 4, 7 and 9 (A) and phosphodiesterase 3b (Pde3b) was measured in freshly isolated conventional CD4⁺Foxp3^GFP− (T_C) or T_reg cells (T_reg) or in the same T cell populations (actT_c and actT_reg) activated in vitro. Light gray bars represent cells isolated from young NOD^GFP mice and gray bars – cells from diabetic NOD^GFP mice. Quantitative PCR was performed using TaqMan assay. Experiment was repeated twice. (B, C) Enhanced expression of Cx43 improved tT_reg cell suppression. Cx43 was overexpressed in T_reg cells sorted from young (○) and diabetic (△) NOD^GFP mice activated in vitro and used in the proliferation inhibition (B) and IFN-γ production assays (C). Cells transduced with empty vector served as a control (● ▲). Transduction efficiency exceeded 90% as monitored by the expression of rat CD2 extracellular domain which was part of the expression vector. (D, E) Cx43 mimetic peptide (Gap26) inhibited gap junction formation and decreased suppressor function of T_reg cells isolated from young (◆) and aged (▲) NOD^GFP mice. Proliferation of responder T cells (D) and IFN-γ production (E) was higher in the presence of gap junction inhibitor peptide (Gap26). All experiments were done twice with six repetitions for each measurement.

To test this hypothesis, we overexpressed Cx43 in expanded T_reg cells in vitro. T_reg cells expressing higher levels of Cx43 were more efficient in inhibiting proliferation and IFN-γ production than the same cells expressing only low levels of Cx43 (Fig. 6B, C). Increased suppression of T_reg cells overexpressing Cx43 closely matches and compensates for the reduction of suppression seen between T_reg cells isolated from young and diabetic mice. Both inhibition of proliferation and IFN-γ production, were abrogated by reagents that restrict intercellular communication through gap junctions, α-glycerrhetinic acid and connexin extracellular loop mimetic peptide (Gap26), further demonstrating that overexpression of Cx43 enhances T_reg cell function by facilitating contact dependent suppression (Fig. 6D, E) (45, 46). In summary, both cell proliferation and IFN-γ production are affected by decreased or increased function of Cx43 in T_reg cells prompting a possibility that Cx43-dependent mechanisms could be targeted to modulate T_reg cell suppression (3, 4).

Redistribution of Cx43 within plasma membrane which promotes gap junction formation and intercellular transport improves suppressor function of T_reg cells in aged NOD mice

Intercellular communication through gap junctions depends on interaction of two connexons (hemichannels), which are hexamers of Cx43 forming a transmembrane channel, each contributed by the contacting cell (47). Cx43 assembled in gap junctions and in the adjacent domain, called perinexus, is associated with the ZO-1 protein through its carboxy terminus. Disrupting interactions between Cx43 and ZO-1 increased the proportion of connexon hemichannels docked to gap junctions relative to connexons in the perinexus, increased size of gap junction plaques and facilitated intercellular communication (48). Since ZO-1, as well as Cx43, are expressed in T_reg cells upon activation, interactions between these two proteins may control the extent of gap junction formation and increase T_reg suppression. To test this hypothesis we disrupted interaction between ZO-1 and Cx43 using peptide reagent that mimics carboxy terminus of Cx43 (49). This reagent, called αCT-1, is engineered to enter cytoplasm and bind the PDZ2 domain of ZO-1, competitively inhibiting its interaction with the carboxy terminus of Cx43. We measured inhibition of proliferation and IFN-γ production in the absence and presence of αCT-1 using responder and T_reg cells isolated from normal B6 mice. As shown in Fig. 7A αCT-1 increased intercellular communication between T_reg and target cells, shown by calcein transport. Addition of αCT-1 which was associated with increased suppressor function of T_reg cells as demonstrated by proliferation inhibition assay and IFN-γ production (Fig. 7B). T_reg cells isolated from mice with Cx43 gene deleted in T cells were not affected by presence of αCT-1 (Fig. 7C). Considering that Cx43 expression is decreased in T_reg cells from diabetic NOD^GFP mice, we reasoned that suppressor function could be augmented in these cells by increasing proportion of Cx43 available to form gap junctions. In fact, we consistently noticed that αCT-1 improved suppressor function of T_reg cells isolated from peripheral and pancreatic lymph nodes of young and diabetic NOD^GFP mice and streptozotocin treated B6 mice, as demonstrated by inhibition of proliferation and IFN-γ production (Fig. 7D–G, Fig. S1C, D). Increased suppression of T_reg cells was most apparent when these cells constituted a small fraction of all T cells (5–10%) in the assay, close to the natural proportion of T_reg cells in the lymphocyte population. This new reagent enhances gap junction mediated suppression mechanism, and may allow for compensation of T_reg deficit without the need to correct the underlying molecular deficit e.g. decreased Cx43 upregulation in diabetic NOD mice. The data presented open a possibility that T_reg cells, even in diabetic mice, retain suppressive potential that could be unmasked by treatments that compensate for decreased expression and function of key molecules mediating T_reg functions.

FIGURE 7.

Carboxyl-terminal Cx43 mimetic peptide αCT-1 improves T_reg-mediated suppression. (A) αCT-1 increases intercellular communication through gap junctions. Histograms show calcein fluorescence in target Cx43-deficient or sufficient CD4⁺ T cells expressing transgenic receptor activated with antigenic peptide PCC50V54A (0.25 μM) in the presence of bone marrow-derived dendritic cells and calcein loaded iT_reg cells in the presence (dotted lines) or absence (continuous lines) αCT-1. (B) Proliferation inhibition and IFN-γ production production in the absence and presence of αCT-1. T_reg and responder cells were isolated from Cx43-sufficient B6 mice. (C) αCT-1 does not enhance suppression of T_reg cells isolated from B6 mice where Cx43 gene was deleted in T cells. Proliferation inhibition assay in the presence or absence (horizontal lines) of αCT-1. Cx43 sufficient (diagonal lines) and deficient (horizontal lines) T_reg cells were incubated with Cx43-sufficient responder cells in a ratio of 1:2. Due to low proportion of T_reg cells in mutant mice only one ratio of T_reg/responder cells was tested. Solid bar represents proliferation of responder cells in the absence of T_reg cells. (D-G) T_reg cells isolated from peripheral (D, F) and pancreatic (E, G) lymph nodes of young (○ ●) and aged (△ ▲) NOD^GFP mice were used in the proliferation inhibition assay (D, E) and to inhibit IFN-γ production (F, G) in the presence (○ △) or absence (● ▲) of αCT-1. One experiment of two is shown with six repetitions of each well. Statistical differences in cell proliferations between T_reg cells cultured in the presence or absence of αCT-1 peptide are shown by asterisks (*).

Discussion

To better understand molecular processes associated with diabetes development we examined CD4⁺ T cell subsets using a new Foxp3^GFP reporter mouse, produced on a pure NOD genetic background. This analysis suggests that, similar to C57BL6 mice, the population of T_reg cells in NOD^GFP mice is heterogeneous with regard to the level and stability of Foxp3 expression (30, 50). Our study extends earlier reports that T_reg cell function deteriorates in aged NOD mice and identifies T_reg cells expressing low levels of Foxp3 and conventional CD4⁺ T cells as cell populations that are mostly affected by age-related dysregulation of Foxp3 expression (22, 51). This is consistent with the recent report that a significant fraction (about 20%) of all T_reg cells may lose Foxp3 expression (16). Our data suggest that loss and/or downregulation of Foxp3 expression might be caused by impaired signaling by IL-2 and TGF-β as demonstrated by decreased phosphorylation of STAT5 and Smad2 in aged mice, even in optimal concentration of both cytokines. IL-2 and TGF-β are major cytokines that support peripheral maintenance of tT_reg cells and generation of pT_reg cells upon antigenic stimulation (34). pT_reg cells were specifically identified as suppressor cells protecting NOD mice subject to immunotherapy with anti-CD3 antibody and their decreased generation in older NOD mice could be regarded as a contributing factor for diabetes development (14). We have previously suggested that continuous up and downregulation of Foxp3 in conventional CD4⁺ T cells in response to self-antigens could be an important mechanism of immune tolerance and, based on current data, one could speculate that altered kinetics of this process may impact immunoregulation in NOD mice (30).

The finding that Cx43, a molecule essential for T_reg suppression and regulated by TGF-β, is only weakly upregulated in activated CD4⁺ T cells in NOD^GFP mice progressing to diabetes complements earlier studies that dysregulation of TGF-β-dependent genes might be an important molecular feature of progression to diabetes (38). Loss and gain of function studies undertaken here identify Cx43 as a molecule which directly mediates T_reg suppression. Moreover, this Cx43 dependence is not regulated by Foxp3, which is consistent with previous reports that functional features and transcriptional signature of T_reg cells are controlled by Foxp3-dependent and independent genes(27, 52). The outcome of decreasing expression of TGF-β dependent genes in the course of progression to diabetes might be magnified in case of Cx43 by developing hyperglycemia. Both, expression of Cx43 and transport through gap junctions are decreased by hyperglycemia (53, 54). Thus, progression to clinical disease likely combines immunological and metabolic factors that form a positive feedback loop and amplify each other.

Expression of Foxp3 and Cx43 is greatly enhanced in activated CD4⁺ T cells from aged NOD^GFP mice by RA, and in vitro generation of iT_reg cells is restored to the level seen in young mice. This outcome shows that coordinate signaling, known to utilize different pathways, is able to compensate for the signal transduction deficit observed in aged cells. In recent reports RA was shown to inhibit T1D progression and our data demonstrate that one of the underlying molecular mechanisms of its effect could be to compensate for the age-dependent decline in the ability to upregulate and maintain Foxp3 and Cx43 expression in T_reg cells (10, 11). A recent report, however, shows that RA may also decrease specific phosphorylations of Cx43 that lead to increases in intercellular communication through gap junctions revealing complexity of regulatory mechanisms acting directly at the protein level or indirectly through regulating gene expression (55). This activity of RA is most likely due to directly enhancing association between Cx43 and protein phosphatase 2A, since it does not depend on transcription or new protein synthesis. Altogether, the current and previous reports extend our understanding of how RA regulates T_reg cell suppression, especially in the context of diabetes development. It also highlights how interactions between immune system and environmental factors (e.g. diet) may promote or prevent disease progression.

Intercellular communication mediated by gap junctions has been proposed as an important mechanism of T_reg suppression (7, 26). Induced T_reg cells generated in vitro from CD4⁺ T cells isolated from B6 mice that have two, one or no functional Cx43 gene alleles have progressively lower suppression (41). Here, by examining Cx43 gene expression pattern in effector and T_reg cells, directly isolated from young (prediabetic) and diabetic mice, or activated in vitro, we attempted to explain how intercellular communication, mediated by Cx43, may mechanistically contribute to immune regulation of diabetes in NOD mice. We have introduced a novel reagent, the Cx43 mimetic peptide αCT-1, to promote T_reg mediated suppression. Distinct from Gap26, the other mimetic peptide used in this study which inhibits both hemichannels and gap junction channels, αCT-1 selectively reduces hemichannel activity, while promoting gap junction plaque formation (48). This is achieved by a novel mechanism in which αCT-1 releases undocked hemichannels from ZO-1 PDZ2 tethering at the plaque edge (i.e., the perinexus), prompting them to dock into gap junctions – thereby resulting in a loss of hemichannel activity and a complementary gain of gap junction function. Our results show that αCT-1 serves as a new experimental tool demonstrating that T_reg cells, even in diabetic, hyperglycemic mice, preserve T_reg suppressor capacity, which can be, at least partially, restored. It is also notable that Cx43 hemichannels have key assignments in the inflammatory response – acting as conduits for the release of proinflammatory mediators such as ATP (56). Inhibition of hemichannel activity by promoting of gap junction formation thus may represent a further factor to consider in the maintenance of beta-cells. Dysregulated expression of Cx43 in smooth muscle cells and altered communication through gap junctions were reported in diabetes patients and were considered as a contributing factor in the development of diabetes complications including vascular retinopathy (54, 57). In Phase II clinical testing, αCT-1 improves healing of diabetic foot ulcers and venous leg ulcers - pathologic, slow healing wounds that are caught in chronic inflammatory states (58). This report shows that Cx43 functions are compromised in T_reg cells, directly involved in controlling immunological mechanisms leading to the destruction of pancreatic islets. Exploiting these mechanisms further may lead to a better understanding of immune regulation in diabetes and to the development of new therapies for diabetes and other autoimmune diseases. In conclusion, we hope our report contributes to better understanding of signaling pathways and individual molecules in regulating immune suppression mediated by T_reg cells in NOD mice.

Abbreviations

T_reg

regulatory T cell

RA

retinoic acid

T1D

type 1 diabetes mellitus

GFP^lo/GFP^hi

T_reg cells with low (lo) and high (hi) expression of Foxp3 (GFP)