OX40L/Jagged1 co-signaling by GM-CSF induced bone-marrow dendritic cells is required for the expansion of functional Tregs

Abstract

Earlier, we had demonstrated that treatment with low dose of GM-CSF can prevent the development of experimental autoimmune thyroiditis (EAT), myasthenia gravis (EAMG) and type-1 diabetes; and could also reverse ongoing EAT and EAMG. The protective effect was mediated through the induction of tolerogenic CD11C+CD8α⁻ DCs and consequent expansion of Foxp3⁺ T-regulatory cells (Tregs). Subsequently, we showed that GM-CSF acted specifically on bone marrow precursors and facilitated their differentiation into tolerogenic DCs (GM-BMDCs), which directed Treg expansion in a contact dependent manner. This novel mechanism of Treg expansion was independent of TCR mediated signalling but required exogenous IL-2 and co-signalling from DC bound OX40L. In the present study, we observed that OX40L mediated signalling by GM-BMDCs although necessary was not sufficient for Treg expansion and required signalling by Jagged1. Concurrent signalling induced by OX40L and Jagged1 via OX40 and Notch3 receptors expressed on Tregs was essential for the Treg expansion with sustained FoxP3 expression. Adoptive transfer of only OX40L⁺Jagged1⁺ BMDCs led to Treg expansion, increased production of IL-4 and IL-10, and suppression of EAT in the recipient mice. These results showed a critical role for OX40L and Jagged1 induced co-signalling in GM-BMDC-induced Treg expansion.

Introduction

In our earlier studies, we have demonstrated that treatment with low dose GM-CSF was sufficient to prevent the development of Experimental Autoimmune Thyroiditis (EAT) in CBA mice, Experimental Autoimmune Myasthenia Gravis (EAMG) in C57BL mice and Type 1 Diabetes (T1D) in NOD mice (1–3). Moreover, our studies also showed that such a treatment could reverse ongoing EAT and EAMG, and restore normal thyroid and neuromuscular conduction respectively (1, 2). Others have shown similar protective effect of GM-CSF in T1D and Irritable Bowel Disease (IBD) (4, 5). Additionally, we have demonstrated that the therapeutic effect of GM-CSF was primarily mediated through the mobilization of CD11c⁺CD8α⁻ DCs (6), which caused expansion of regulatory T-cells (Tregs). These expanded Tregs suppressed the disease through increased IL-10 production (7).

Subsequently, we showed that GM-CSF could differentiate bone marrow derived DC precursors ex vivo and cause a selective expansion of CD11c⁺CD11b⁺ CD8α⁻DCs (GM-BMDCs) (8). Remarkably, unlike DCs isolated from the spleen (SpDCs), these ex vivo developed GM-BMDCs were able to directly and specifically expand Tregs upon co-culture with CD4⁺ T-cells. Furthermore, treatment of mice with GM-CSF led to an increase in CD11c⁺CD11b⁺CD8α⁻ DCs in vivo with concomitant increase in Foxp3⁺ Tregs, suggesting a parallel mechanism of CD11c⁺CD11b⁺CD8α⁻ DC mediated Treg expansion ex vivo and in vivo.

Using GM-BMDCs from MHC class-II^−/− mice, we were able to show that Treg expansion by these DCs did not require canonical antigen presentation to TCR but required addition of exogenous IL-2 (8). Using blocking antibodies to co-stimulatory molecules expressed on the surface of GM-BMDCs, we showed that the GM-BMDC mediated Treg proliferation was dependent upon GM-BMDC bound OX40L (8), a member of the tumor necrosis factor super family with co-stimulatory function (9). Studies by other groups have also suggested a novel role for OX40L-OX40 interaction in the expansion of Tregs (10, 11).

In the current study, we found that only OX40L⁺, and not OX40L⁻, GM-BMDCs can dramatically expand Tregs. Surprisingly, an OX40 agonist could not expand Tregs when added to cultures either alone or in the presence of SpDCs indicating that other co-stimulatory molecules expressed on GM-BMDC may be involved in GM-BMDC mediated Treg expansion. In this context, the Jagged members (Jagged1 and Jagged2) of Notch family ligands have been shown to play important role in Treg expansion (12, 13). Expression of specific Notch ligands on DCs is known to activate specific T-cell responses (14). The Notch family has 4 known receptors, Notch-1, -2, -3 and -4, and five known Notch ligands namely, DLL1, DLL3 and DLL4, and Jagged1 and Jagged2. Upon ligand binding, Notch receptors undergo two proteolytic cleavages. The first cleavage is catalysed by ADAM-family metalloproteases and is followed by the gamma-secretase mediated release of Notch intracellular domain (NICD). The NICD translocates to the nucleus where it forms a heterodimeric complex with various co-activator molecules and acts as a transcriptional activator (15). While Jagged ligands have been shown to direct the naive T-cells toward a Th2 and/or Treg type of responses, Delta like ligands (DLL) have been shown to skew them towards a Th1 response (16). Of particular relevance to the current study are earlier reports of Treg expansion by hematopoietic progenitors expressing Jagged2 and APCs over-expressing Jagged1 (12, 13, 17, 18). Interestingly, administration of Jagged1-Fc into mice with experimental allergic encephalomyelitis (EAE) caused an increase in Ag specific IL-10 producing cells, while DLL1-Fc exacerbated the disease (19). Similarly, DLL4 blockade ameliorated experimental EAE (20). Therefore, we investigated the potential role of Notch signalling in GM-BMDC mediated Treg proliferation.

Therefore, we looked for the expression of Notch ligands on OX40L⁺ GM-BMDC and investigated their role in Treg expansion. Our results strongly indicate that co-signaling induced by OX40L and Jagged1, expressed on GM-BMDC, through interaction with their cognate receptors OX40 and Notch3, expressed on Tregs, are essential for GM-BMDC mediated Treg expansion with sustained Foxp3 expression+ that results in the suppression of EAT.

Materials and Methods

Animals

Six to eight week old CBA/J mice were purchased from the Jackson laboratory. Mice were housed in the Biological Resources laboratory facility at the University of Illinois (Chicago, IL) and provided food and water ad libitum. CD80−/−, CD86−/−, CD80−/− CD86−/−, Foxp3GFP and WT (C57B6/j background) mice were kindly provided by Dr. Chenthamarakshan Vasu (Department of Surgery, Medical University of South Carolina). All animal experiments were approved by the University of Illinois at Chicago animal care and use committee.

GM-CSF, antibodies and thyroglobulin

Recombinant GM-CSF and Carboxy Fluorescein Succinimidyl Ester (CFSE) were purchased from Invitrogen (Carlsbad, USA). PE conjugated anti-H-2K^d (MHC class II), anti- Jagged1, anti-DLL1, anti-DLL3, anti-DLL4, anti-Notch 1; Pacific blue conjugated anti-CD4, APC conjugated anti-CD11c, anti-CD11b, anti-Foxp3, anti-CD3, PE conjugated anti-IL-4, and IFN-γ antibodies, and OX40 agonist (OX86) were purchased from eBioscience (San Diego, California). APC conjugated anti-OX40L antibody was purchased from Biolegend. Blocking antibodies to OX40L (AF1236), Jagged1 (AF599), Notch1 (AF1057) and Notch3 (AF1308) and normal goat IgG control (AB-108-C) were purchased from R&D systems (Minneapolis, MN). Primary and secondary antibodies for staining intracellular notch receptors (NICD) against Notch1 (sc-23307) and Notch3 (sc-5593) (12, 21) were purchased from Santa Cruz biotechnology. Mouse thyroids were purchased from Pel-Freeze (Rogers, AR) and thyroglobulin was prepared as described earlier (22). In brief, thyroids were homogenized in 2.5 ml PBS with pestle-homogenizer (Wheaton, Millville, NJ) with overnight extraction at 4°C. The extract was clarified by centrifugation (15000g) and fractionated on a Sephadex G-200 column (2.5 × 90 cm) which had been equilibrated with 0.1 M phosphate buffer, pH 7.2. The concentration and purity of mTg was determined spectrophotometrically at 280 nm and by resolving on 7% SDS-PAGE followed by coomassie blue staining. Gamma-secretase inhibitors (GSI) S-2188 and RO4929097 were purchased from Sigma-Aldrich (St. Louis, MO) and Selleck Chemicals (Houston, TX) respectively.

Isolation of DC and T-cell subpopulations

Bone marrow cells were cultured in complete RPMI medium containing 10% heat-inactivated FBS in the presence of 20ng/ml GM-CSF for 3 days. On days 4 and 6, fresh medium containing 20ng/ml GM-CSF was added. Non-adherent CD11c⁺ DCs from eight day old cultures were enriched using anti-CD11c coated magnetic beads according to the manufacturer's directions (Miltenyi Biotec). Specific sub population of GM-BMDCs and CD4+CD25+ T-cells were sorted using a MoFlo flow cytometer (Beckman/Coulter) following staining with appropriate antibodies (OX40L, jagged1, CD4, CD25). To obtain GFP+ and GFP- cells, total CD4+ cells were first separated using CD4 microbeads (Miltenyi) and then the cells were sorted based on GFP expression using a MoFlo flow cytometer (Beckman/Coulter).

In- vitro co-cultures of DCs and T cells

Each in vitro experiment was conducted in triplicate with T-cells, SpDCs and GM-BMDCs pooled from 3 mice. GM-BMDCs (5 × 10⁴) and CD11c⁺ SpDCs were cultured with CD4⁺, CD4⁺CD25⁻ and CD4⁺CD25⁺ T-cells at a ratio of 1:2 for 5 days. For proliferation assays, T-cell subpopulations were labelled with CFSE at 10μM according to manufacturer's instruction (Invitrogen, Carlsbad, CA) before co-culturing them with DCs. Some cultures were supplemented with IL-2 (10U/ml) (R&D Systems), anti-OX40L (up to10 μg/ml) antibody, OX40 agonist (OX-86, 5–10 μg/ml), anti- Jagged1 (10–20μg/ml) antibody or anti-Notch3 (10–20μg/ml) antibody. For blocking experiments with anti-OX40L or anti-Jagged1 antibodies, GM-BMDCs were pre-treated with the indicated antibodies for 30min at 37°c and then used in co-culture with naive CD4⁺ T-cells. For blocking experiments with anti-Notch3 antibody, CD4+ T-cells isolated from mouse splenocytes were first treated with anti-notch3 antibody at two different concentrations (i.e.10 and 20 μg/ml) or with 20μg/ml of an anti-notch1 antibody, incubated at 37°C for 30 minutes and then co-cultured with GM-BMDCs/SpDCs for 5 days. Some co-cultures were supplemented with different concentrations of gamma-secretase inhibitors (GSI) S-2188 (5 and 10 μM) or RO4929097 (200 nM-5μM).

Suppression assay

CD4⁺CD25⁻ effector T-cells were isolated from spleens, stained with CFSE and plated into flat bottom 96 well plates at 0.5×10⁶ cells/well in the presence of either OVA or mTg (100 μg/ml) and splenic APCs. Sorted CD4⁺CD25⁺ Tregs from ex vivo co-cultures of naïve CD4⁺ T-cells and GM-BMDC were added at different ratios to the co-culture containing CD4⁺CD25⁻ T-cells from primed mice.

Propidium iodide (PI) and Intracellular Staining

Briefly, at the end of co-culture experiments, T-cells were stained with Pacific blue labelled anti-mouse CD4 antibody and labelled with propidium iodide and subjected to FACS analysis to assess cell viability. For intracellular staining, surface stained cells were fixed and permeabilized using a commercial kit and according to the manufacturer's instructions (eBioscience) and incubated with specified antibodies.

FACS

Freshly isolated and ex vivo cultured cells were washed with PBS-BSA-EDTA. For surface staining, the cells were labelled with specified FITC, PE, APC conjugated antibodies for 30 min. For cell proliferation assays, the cells were labelled with CFSE, fixed, permeabilized and incubated with fluorescent coupled antibodies for intracellular staining. Stained cells were washed three times and analysed by Cyan flow cytometer (Beckman/Coulter).

SiRNA transfection into GM-BMDC

A 21bp siRNA sequence (Dharmacon) specific to Jagged1 (5'-CTCGTAATCCTTAATGGTT-3') was used at a final concentration of 120 nM as previously described (23). Briefly, 3 μl of 20 μM annealed siRNA was incubated with 3μl of GenePorter (Gene Therapy Systems) in a volume of 94μl of serum-free RPMI 1640 at room temperature for 30 min. This mixture was added to each well containing GM-BMDC in a volume of 500 μl and incubated for 4 h at 37°C. 3μl of GenePorter alone was used for mock transfection as a negative control. After incubation, 500μl/well of RPMI 1640 supplemented with 20% FCS was added and twenty-four hours later, GM-BMDCs were washed and used.

RT-PCR

Total RNA was extracted using TRIzol reagent (Invitrogen) and the first strand cDNA was synthesized using Superscript 2 (Invitrogen). Gene specific primers were used for semi quantitative PCR amplification (0.5 min at 94°C, 0.5min at 55°C, and 0.5 min at 72°C for 33 cycles) to detect relative amounts of different transcripts. The following primer sets were used to amplify the indicated products:

• HPRT-F, GTTGGATACAGGCCAGACTTTGTTG

• HPRT-R, TACTAGGCAGATGGCCAGGACTA

• Notch1-F, TGTTAATGAGTGCATCTCCAA

• Notch1-R, CATTCGTAGCCATCAATCTTGTC

• Notch2-F, TGGAGGTAAATGAATGCCAGAGC

• Notch2-R, TGTAGCGATTGATGCCGTC

• Notch3-F, ACACTGGGAGTTCTCTGT

• Notch3-R, GTCTGCTGGCATGGGATA

• Notch4-F, CACCTCCTGCCATAACACCTTG

• Notch4-R, ACACAGTCATCTGGGTTCATCTCAC

Priming mice with mTg and OVA

Groups of CBA/J mice were immunized (3 mice per group for each experiment) subcutaneously with OVA (100μg/mouse) or mTg (100μg/mouse) emulsified in CFA on day 1 and day 10. Various subsets of T cells from these mice were used in Treg expansion and proliferation assays

Adoptive transfer

Three groups of 3 mice each were immunized twice, 10 days apart, with mTg (100μg/ml) emulsified in CFA. Ten days after the 2^nd immunization, mice received i.v. injection of either i) PBS, ii) 2 × 10⁶ purified CD11c⁺ DCs from untreated CBA/J mice or iii) 2 × 10⁶ CD11c⁺ GM-BMDC purified and sorted from BM cultures. Two identical adoptive transfers were done for each group at 5 day intervals. Five-days after the 2^nd transfer, mice were sacrificed and spleen and thyroid draining lymph node cells were analysed for Treg percentages.

Statistical analysis

Mean, standard deviation, and statistical significance were calculated using the Graph pad software and MS-Excel application software. Statistical significance was determined using the one tailed Students t-test. A P value of ≤0.05 was considered significant.

Results

OX40L is necessary but not sufficient for the expansion of Tregs mediated by GM-BMDCs

In our earlier study, we used a blocking antibody to OX40L to demonstrate a dose dependent abrogation of Treg proliferation by GM-BMDC (8), which was restored upon addition of a soluble OX40 agonist. In a typical 7-day bone marrow culture with GM-CSF, ~30% CD11c⁺ GM-BMDCs are OX40L⁺ (Fig 1A). To determine if OX40L induced signalling is sufficient for the expansion of Tregs, we set up co-cultures with sorted populations of OX40L⁺ and OX40L⁻ GM-BMDCs with naive CD4⁺ T-cells. As expected, only OX40L⁺ GM-BMDC drove the proliferation of Foxp3⁺ Tregs (10.1±0.6%) relative to OX40L⁻ GM-BMDC (0.5±0.1%, p=0.002) (Fig 1B).

FIGURE 1. OX40L is necessary but not sufficient for GM-BMDC directed ex vivo expansion of Tregs.

(A). Percentage of OX40L⁺ GM-BMDCs gated on CD11c⁺ cells. (B) GM-BMDCs derived ex vivo from bone marrow cells of WT C57B6/j mice were sorted after 7 days of differentiation with GM-CSF. Naïve CFSE labelled CD4⁺ T-cells were co-cultured with either splenic dendritic cells (SpDCs), or total, OX40L⁺ or OX40L⁻ enriched GM-BMDCS for 5 days without exogenous antigen and analysed by FACS. (C). Total, OX40L⁺ or OX40L⁻ GM-BMDCs were co-cultured with Cell-Trace violet labelled sorted GFP+ and GFP− T-cells from Foxp3-GFP mice after CD4⁺ based enrichment for 5 days without exogenous antigen and analysed by FACS. Co-cultures of GFP⁺ (Foxp3⁺) cells without (upper panel and with IL-2 (lower panel). (D). Co-cultures of GFP⁻ (Foxp3⁻) cells without (upper panel and with IL-2 (lower panel). Each scatter plot is representative of five independent experiments, gated over 3500 live CD4⁺ T-cells. Each in vitro experiment was conducted with T-cells, SpDCs and GM-BMDCs pooled from 3 mice. (E) Sorted OX40L⁺ or OX40L⁻ GM-BMDCs were co-cultured with CFSE labelled CD4⁺ T-cells and supplemented with OX40 agonist. Co-cultures were analysed by FACS on day 5 to determine T-cell proliferation.

To specifically address the role of OX40L⁺ GM-BMDCs on Foxp3⁺ Tregs, we made use of the Foxp3-GFP transgenic mice. We set up co-cultures of sorted OX40L⁺ and OX40L⁻ GM-BMDCs (Supplementary Fig-1) with sorted and Cell-Trace Violet labelled CD4⁺GFP⁺ (Fig 1C) or CD4⁺ GFP⁻ (Fig 1D) T cells isolated from Foxp3-GFP mice (Supplementary Fig-1), in the presence or absence of IL-2. The extent of Cell-Trace Violet dilution revealed that in the absence of IL-2, a very small fraction of GFP⁺ T-cells proliferated after 5-days of co-culture with either total, OX40L⁺ or OX40L⁻ GM-BMDCs. However, in the presence of IL-2, Foxp3⁺ T-cells proliferated efficiently only when co-cultured with either total (25.0±1.7%) or OX40L⁺ (34±3.2%), and not with OX40L⁻, GM-BMDCs (7.4±1.0%). In contrast, GFP⁻ T-cells (Foxp3⁻) showed either modest or robust proliferation based on absence or presence of IL-2 irrespective of whether they were co-cultured in the presence of total, OX40L⁺ or OX40L⁻ GM-BMDCs. Most notably, we failed to see any adaptive conversion of Teff into Tregs in any cultures involving GFP⁻ cells. It is important to note that none of these co-cultures are stimulated with anti-CD3 or any exogenous antigen. Thus, our data strongly suggest that only OX40L⁺ GM-BMDCs, a subset of the CD11c⁺CD11b⁺B220⁻CD8a⁻ GM-BMDCs (Supplementary Fig 2), can cause efficient proliferation Foxp3⁺ Tregs; consequently we focused our studies on understanding the mechanism of GM-BMDC mediated Treg proliferation.

To see if signalling by OX40L alone is sufficient to expand Tregs, we supplemented CD4⁺ T cells co-cultured with either OX40L⁻ GM-BMDC or splenic DCs with a functional OX40 agonist. Such a treatment failed to cause significant proliferation of Foxp3⁺ Tregs (0.8±0.1%) when compared to the Treg proliferation noted in the presence of OX40L⁺ GM-BMDC (13.5±0.7%, p<0.001) (Fig 1E). These results suggested that OX40L, although required, may not be sufficient for the GM-BMDC mediated ex vivo expansion of Tregs.

Surface bound ligand/s other than the B7 family co-stimulatory molecules are involved in GM-BMDC induced Treg expansion

To determine if in addition to OX40L expressed on GM-BMDC, co-signalling by a soluble factor from, or a surface bound molecule on, GM-BMDC is required for Treg expansion, we set up co-cultures of CD4⁺ T-cells and DCs in trans-well plates. Splenic APCs and CD4⁺ T-cells along with an OX40 agonist were physically separated from GM-BMDC cultured in trans-wells, which allowed for free exchange of soluble factors in culture medium. If soluble factors from GM-BMDC were contributing to Treg expansion, we expected those factors to cross the trans-well barrier and aid in Treg expansion in the presence of OX40 agonist and splenic APCs. However, we noted little or no proliferation of Tregs (0.2±0.1%) in the trans-well when compared to CD4+T-cell-GM-BMDC co-cultures (10.3±0.7%) (Fig 2A). These results suggested that in addition to OX40L, co-signalling by another GM-BMDC surface bound molecule/s was essential for GM-BMDC mediated Treg expansion.

FIGURE 2. OX40L is necessary but not sufficient in GM-BMDC mediated Treg expansion.

(A) CD4⁺ cells from naïve mice were placed in the lower wells and co-cultured with wild type GM-BMDCs either together (upper right panel) or in transwells in which the T-cells were exposed to only the soluble factors from the BMDCs (upper 2^nd and 3^rd panels); in some cases the T-cells were supplemented with SpDCs and an OX40 agonist (upper 3^rd panel). Data were analysed by FACS. (B) GM-BMDCs from CD80, CD86 and CD80/86 deficient mice were co-cultured with naïve CFSE labelled CD4⁺ T-cells without exogenous antigen and analysed by FACS (lower panel). Experiments A and B were repeated three times with similar results.

To identify another cell surface molecule involved in GM-BMDC mediated Treg proliferation, we co-cultured naive CD4+ T-cells with GM-BMDC derived from CD80 and CD86 knockout mice. Lack of expression of either CD80 or CD86 on GM-BMDC had little or no effect on their ability to induce Treg proliferation (7.6±1.0% and 7.2±0.8%) relative to GM-BMDC derived from WT mice (7.9±0.6%) (Fig 2B). In fact, GM-BMDC developed ex vivo from CD80/CD86 double knock-out mice could cause robust Treg proliferation in co-cultures (8.1±0.9%). These data strongly suggested that a molecule/s other than CD80 or CD86 was involved in signaling required for the GM-BMDC induced Treg expansion.

Jagged1 mediated Notch signaling is involved in the GM-BMDC induced proliferation of Foxp3⁺ Tregs

To test if Notch signaling was involved in our observed ex vivo Treg proliferation, we added S-2188, a γ-secretase inhibitor (GSI), which blocks Notch signaling, to the GM-BMDC/T-cell co-cultures. Interestingly, blocking Notch signaling with S-2188 completely abrogated Treg proliferation (2.1±0.5%-0.4±0.1%) in a dose dependent manner (5–10 μM) compared to the proliferation of Tregs in untreated cultures (11.2±1.0%, p<0.001) (Fig 3A). To assess if this difference was attributable to a difference in cell viability, we stained the co-cultures with propidium iodide (PI) and analyzed for cell death by FACS; S-2188 treatment did not affect cell survival (Fig 3B). We also tested the effects of treating the cells with RO4929097, another GSI known to be effective at lower doses, at different concentrations (250nM-5μM) (Supplementary Fig 3). While co-cultures of CD4+ T-cells with GM-BMDCs alone led to robust proliferation (~13.2±0.4%), treatment with GSI severely restricted proliferation in a dose dependent manner (e.g. 1.7±0.3% at 5 μM; and 5.2±0.4% at 250nM GSI) These results suggested that Notch signalling was important for GM-BMDC mediated Treg proliferation.

FIGURE 3. Jagged1 mediated Notch signalling is required for Treg expansion by GM-BMDCs.

(A) Co-cultures of GM-BMDCs with CFSE labelled CD4⁺ T-cells were supplemented with Gamma-secretase-inhibitor (GSI), an inhibitor of Notch signalling, and analysed by FACS. (B) Summary of FACS data from Propidium Iodide staining of co-cultures from GSI experiment showing little or no cell necrosis in all co-cultures. (C) Phenotypic characterization of CD11c⁺ SpDCs and GM-BMDCs comparing the levels of expression of different Notch ligands. Cells were gated on the CD11c+ populations. (D) Co-cultures of GM-BMDCs with CFSE labelled CD4⁺ T-cells were supplemented with two concentrations of a Jagged1 neutralizing antibody and analysed by FACS. Experiments A through D were repeated three times with similar results.

Subsequently, we stained GM-BMDC and SpDCs to analyze for the expression of different Notch ligands. We observed that a much higher proportion of GM-BMDC expressed Jagged1 (18.1±2.8%, p<0.01) relative to SpDCs (1.8±0.5%) (Fig 3C). In contrast, all other Notch ligands were expressed on a higher percentage of SpDCs than on GM-BMDC. This unique expression pattern of ligands on GM-BMDC led us to investigate the potential role of Jagged1 in GM-BMDC mediated Treg expansion. Addition of a blocking antibody against Jagged1 (lo=10μg/ml; hi=20μg/ml) suppressed Treg expansion in a dose dependent manner (reduced from 13.4%±1% to 9.9%±0.5% with low dose and to 6.9±0.2% with high dose (p<0.01 in all instances) (Fig 3D). Interestingly, Jagged1 blocking antibody had little or no effect on the percentages of non-dividing Tregs (~7–8%) and indicated that the effect of Jagged1 blockage primarily affected Treg proliferation without affecting their survival.

Jagged1 and OX40L are critical for GM-BMDC mediated Treg-expansion

We used specific antibodies to block OX40L and Jagged1 to determine if concurrent signaling by both ligands was essential for Treg expansion. Blocking either OX40L or Jagged1, using specific antibodies, reduced Treg proliferation from 13.0% in the absence of antibody to 5.3±0.3% and 3.7±0.2% in the presence of anti-Jagged1-Hi and anti-OX40L-Hi respectively. However, simultaneous blockade of both molecules completely prevented the GM-BMDC mediated Treg proliferation (13.0% v/s 0.5% ± 0.1%, p<0.01) (Fig 4A). These data suggested that Notch signaling, likely induced by Jagged1, along with OX4O signaling induced by OX40L were essential for GM-BMDC mediated Treg proliferation.

FIGURE 4. OX40L and Jagged1 function are critical for GM-BMDC mediated expansion of Tregs.

(A) Co-cultures of GM-BMDCs with CFSE labelled CD4⁺ T-cells were supplemented with neutralizing antibodies to Jagged1 and OX40L, either alone or in combination and analysed by FACS. (B) FACS analysis of CD25⁺Foxp3⁺ T-cells from the co-cultures of APCs and MHCII−/− GM-BMDCs with CD25⁺ T-cells in the presence and absence of IL-2 and neutralizing antibodies to OX40L and Jagged1. (C) GM-BMDCs were treated with control or Jagged1 specific siRNAs. FACS analyses of cell surface expression of Jagged1 and OX40L shows specific knockdown of Jagged1, but not OX40L, after jagged1 specific SiRNA treatment. (D) CFSE labelled CD4⁺ T-cells were cultured with control or jagged1 specific SiRNA treated GM-BMDC in the presence or absence of anti-OX40L antibodies. Results A through D are representative of 3 independent experiments.

We had earlier demonstrated that the OX40L mediated Treg expansion by GM-BMDC did not require TCR stimulation (8). To determine if the Jagged1 mediated signaling was also independent of TCR signaling, we co-cultured CD25⁺ T cells with GM-BMDCs derived from MHC class-II^−/− mice in the presence of IL-2. As expected, MHC class-II^−/− GMBMDCs were able to expand Tregs (78.0±1.4%). However, blocking either OX40L or Jagged1 significantly reduced Treg proliferation from 78.0±1.4% to 31.7 ± 0.5% in the presence of anti-OX40L and to 27.0±1.1% in the presence of anti-Jagged1. Blocking both ligands almost completely prevented Treg proliferation (p<0.01 in all instances) (Fig 4B).

To further substantiate the relative importance of these two ligands, we used specific siRNA to knock down Jagged1 (Fig 4C) on GM-BMDC and co-cultured them with CD4+T-cells. The siRNA treatment (120 nM) significantly reduced the expression of Jagged1 in GM-BMDC (1.5±0.4%; p<0.01) relative to its expression on either untreated (18.0±2.1%) or control siRNA treated (17.7±2.4%) GM-BMDC, without altering the expression of OX40L (approximately 27% in both Jagged1 siRNA treated and control siRNA treated cells) (Fig 4C, right panels). These GM-BMDCs were used in co-culture with CFSE labelled naive CD4+ T-cells. The Treg proliferation was significantly reduced from 8.1±1.0% in the presence of control GM-BMDC to 1.6±0.5% in the presence of Jagged1 knocked down GM-BMDC (Fig 4D). Combined treatment of GM-BMDC with an OX40L blocking antibody (hi=10μg/ml) along with Jagged1 knockdown almost completely abrogated their ability to expand Tregs (0.2±0.1%). These results clearly showed that both OX40L and Jagged1 expressed on GM-BMDC are required for efficient Treg expansion.

OX40L and Jagged1 mediated co-signalling is required for GM-BMDC mediated Treg expansion

Since we had noted that OX40L⁺ GM-BMDC had a superior ability to cause Treg proliferation relative to OX40L⁻ GM-BMDC, we determined if Jagged1 was co-expressed on OX40L⁺ GM-BMDC. Interestingly, GM-BMDCs that were OX40L⁻ were also Jagged1⁻ (Fig 5A). On the other hand, about half of OX40L⁺ GM-BMDCs were Jagged1+ (50.3±0.5%, p<0.02) (Fig 5A).

FIGURE 5. OX40L/Jagged1 co-signalling is required for GM-BMDC mediated Treg expansion.

(A) GM-BMDCs were analysed for surface expression of OX40L and Jagged1. Cells were successively gated over the CD11c⁺ and OX40L⁺ populations and analysed for Jagged1 expression. (B) CFSE labelled CD4⁺ T-cells were co-cultured with either total or OX40L⁺ Jagged1⁺ or OX40L⁺ Jagged1⁻ GM-BMDCs. Some cultures were supplemented with anti-OX40L and/or anti-Jagged1 antibodies. Figure shows summary of cell proliferation data analysed by FACS. The experiment was repeated three times with similar results.

To determine if OX40L and Jagged1 co-expression was required for the OX40L⁺ GM-BMDC -induced expansion of Tregs, we sorted the GM-BMDC into OX40L⁺ Jagged1⁺ and OX40L⁺ Jagged1⁻ DCs and used them in co-culture with naive CD4⁺T-cells. While total GM-BMDC could induce Treg proliferation (i.e. 8.2%), the OX40L⁺ Jagged1⁺ GM-BMDCs were able to more efficiently expand Tregs (12.5±0.2%). In contrast, OX40L⁺ Jagged1⁻ failed to mediate significant expansion of Tregs (1.4±0.1%, p<0.001) (Fig 5B). Blocking either ligand with the corresponding blocking antibody caused significant reduction in Treg expansion. However, blocking both ligands (Anti-OX40L=10g/ml, Anti-Jagged1=20μg/ml) on OX40L⁺ Jagged1⁺ GM-BMDCs abrogated Treg expansion (reduced from 12.5±0.2% to 0.7±0.1%; p<0.01). These results clearly demonstrated that GM-BMDC mediated ex vivo Treg expansion required cell surface expression of both OX40L and Jagged1.

GM-BMDC associated Jagged1 can induce Treg proliferation by activating Treg associated Notch3

To determine the specific Notch receptor that was activated by Jagged1 to cause Treg proliferation, we analyzed the mRNA expression patterns of all four Notch receptors in Foxp3⁺ (i.e. GFP⁺) and Foxp3- (GFP⁻) cells from Foxp3-GFP mice. Semi-quantitative PCR indicated that transcripts for Notch1 and Notch4 were similarly expressed in Teffs and Tregs. However, expression of Notch3 transcript was significantly higher in Foxp3⁺ Tregs relative to Foxp3⁻ effector T cells, while the transcripts for Notch2 was predominantly expressed in Teff cells (Fig 6A). These findings suggested that Jagged1 expressed on GM-BMDC may be binding specifically to Notch3 expressed on Tregs to cause their expansion.

FIGURE 6. Jagged1 on GM-BMDCs transduces proliferation signals to Tregs through Notch3.

(A) GFP⁺ and GFP⁻ cells isolated from Foxp3-GFP mice were analysed for the expression of Notch receptor transcripts by RT-PCR. Notch3 transcript is detected specifically in Tregs. cDNAs from different T-cell populations were subjected to PCR using different Notch specific primers and analysed on 2% agarose gel. Parts of the gel relevant to the specific subpopulation were assembled together. (B) Co-culture of GM-BMDCs and CD4⁺ T-cells in the presence of neutralizing antibody to Notch3 or Notch1. Each scatter plot in B and C represents five separate experiments. (C) Shows Notch3 specific Notch Intracellular Domain (NICD) only in proliferating Foxp3+ T-cells in GM-BMDC/T-cell co-cultures analysed by FACS. CFSE dilution was used to measure cell-proliferation and cells were gated on CFSE diluted or undiluted populations and analysed for NICD.

The importance of Notch3 signalling was substantiated by a reduction in GM-BMDC induced Treg proliferation upon addition of a Notch3 blocking antibody to the GM-BMDC -T cell co-culture in a dose dependent manner. The proliferation was reduced from 8.5±0.3% in untreated culture to 5.1±0.4% and 2.6±0.2% in the presence of low (10μg) and high dose (20μg) of anti-Notch3 antibody respectively: p<0.02 (Fig 6B). In contrast, a blocking antibody to Notch1 did not have any apparent effect on Treg proliferation.

The detection of cytoplasmic Notch Intra-Cellular Domain (NICD) has been used as a marker for activated Notch3 (24). To confirm the role of Notch3 in mediating Jagged1 induced signaling, we used a Notch3 specific polyclonal antibody (12, 21) to detect the intracellular portion of Notch3 in the GM-BMDC/T-cell co-cultures. Analyses of proliferating and non-proliferating Foxp3⁺ and Foxp3⁻ cells showed that nearly 97% of the proliferating Foxp3⁺ T cells were positive for Notch3 NICD, while approximately 98% of non-proliferating Foxp3⁺ or Foxp3⁻ T cells were negative for Notch3 NICD (Fig 6C). Collectively, our data suggested that Notch3, expressed selectively on Tregs, is activated by Jagged1 expressed on GM-BMDCs and this interaction is essential for Treg proliferation.

OX40L⁺ Jagged 1⁺ GM-BMDCs can suppress ongoing EAT

We first tested the suppressive effect of ex vivo generated Tregs on antigen-induced T cell proliferation. Mice were immunized with 100 μg mTg or OVA to induce an antigen specific effector T cell response, which we monitored through the emergence of serum antibodies to mTg and OVA respectively. T cells from naïve mice were used to set up GM-BMDC/T-cell co-cultures to generate Tregs. In the absence of TCR stimulation, the expanded Tregs were a major fraction of the CD25⁺ T-cells and were therefore isolated on the basis of CD25 expression. We then isolated CD4⁺CD25⁻ T cells from the above mentioned immunized animals, stained them with CFSE and set up co-cultures with splenic APCs in the presence of mTg or OVA with or without sorted Tregs (CD4⁺CD25⁺). CD25⁻ cells from OVA immunized mice and mTg immunized mice proliferated in the presence of OVA and mTg respectively. As expected, exogenous antigen OVA-induced proliferation was much more robust as compared to the autoantigen mTg-induced proliferation. Both mTg- and OVA-induced proliferations were significantly suppressed when CD25⁺ Tregs were added at either 1:1, 1:2, 1:4 Tregs:Teffs ratios (Fig 7A). These results showed that ex vivo generated Tregs are functionally competent.

FIGURE 7. OX40L+ Jagged1+ GM-BMDCs can induce Tregs in vivo and suppress EAT.

(A) Ex vivo expanded Tregs can suppress effector T-cell proliferation. CD4⁺CD25⁺ T- cells were sorted from the co-culture of OX40L⁺ Jagged1⁺ GM-BMDCs and T-cells from naive mice. The sorted Tregs were set up in co-culture with CFSE labelled effector T-cells isolated from OVA and mTg immunized mice at different ratios. After 5 days in culture, CD4⁺T-cells were analysed for CFSE dilution by FACS. (B) EAT was induced in mice as described before (1). Briefly, mice were immunized with mTg+CFA on days 1 and 10 to induce EAT. On days 17 and 22, mice were treated with mTg pulsed OX40L⁺ Jagged1⁺ or OX40L⁺ Jagged1⁻ GM-BMDCs. Mice were sacrificed on day 35 and analyzed for Foxp3⁺ Tregs in the spleen by FACS. (C) Bar graphs shows percentage of IFN-γ, IL-4 and IL-10 producing CD4⁺ cells in the spleen of treated mice analysed by FACS. (D) Bar graphs shows percentage of IFN-γ, IL-4 and IL-10 producing CD4⁺ cells in the thyroid draining lymph nodes of differently treated mice analysed by FACS. (E) H and E stained sections of thyroid tissue showing extent of tissue infiltration by lymphocytes. Note no infiltration in unimmunized mice. While significant infiltration is seen in thyroids from mice that were either treated with PBS or with OX40L+Jagged1− GM-BMDCs, there was minimal inflammation in mice treated with OX40L+Jagged1+ GM-BMDCs. Results shown are representative of three independent experiments.

Since only a small fraction of GM-BMDC, viz. the OX40L⁺ Jagged1⁺ fraction, could expand Tregs ex vivo, we tested to see if this subpopulation of DCs can also expand Tregs in vivo and confer protection against EAT. We immunized mice with mTg+CFA on days 1 and 10 to induce EAT. On days 17 and 22, these mice were adoptively transferred with different subsets of GM-BMDC. Mice were sacrificed on day 35 and analyzed for Foxp3⁺ Tregs. The OX40L⁺ Jagged1⁺ GM-BMDC recipient mice showed a significant increase in the percentage of Foxp3⁺ Tregs in the spleen (15.0±0.5%) compared to control mice that were treated with PBS (9.2±1.0%) or mice that received OX40L⁺Jagged1⁻ GM-BMDC (9.0±0.5%) (p<0.01 v/s OX40L⁺Jagged1⁺ GM-BMDC in both cases) (Fig 7B). CD4⁺ T-cells from these recipient mice were re-stimulated with mTg in the presence of APCs for 3 days and analyzed for cytokine production. Mice that received OX40L⁺ Jagged1⁺ GM-BMDC showed a significant decrease in IFNγ producing CD4⁺ T cells (p<0.01), while they showed a significant increase (p<0.01) in IL-4⁺ and IL-10⁺ CD4⁺T cells compared to the controls (Fig 7C). Similarly, the cytokine profile of T-cells from the thyroid draining lymph nodes of OX40L⁺Jagged1⁺ GM-BMDC recipient mice showed significantly lower percentages of IFN-γ⁺ cells, while the percentages of IL-4⁺ and IL-10⁺ CD4⁺T cells were significantly (p=0.001) higher (Fig 7D) relative to the controls.

Thyroid histopathology revealed reduced infiltration of lymphocytes into the thyroid of OX40L⁺ Jagged1⁺ GM-BMDC-recipient mice compared to the control groups either treated with OX40L⁺ Jagged1⁻ GM-BMDCs or left untreated (p=0.02 in both cases; Fig 7E). Our results showed that OX40L⁺ Jagged1⁺ GM-BMDC can increase the number of Tregs in vivo, with a concomitant decrease in Th1 cytokines and increase in suppressor cytokines, and suppress ongoing EAT.

Discussion

Earlier, using GM-BMDC from MHC class-II deficient mice, we had shown that OX40L mediated ex vivo expansion of Tregs did not require TCR stimulation per se although it was critically depended on exogenous IL-2 (8). Although much is known about TCR mediated T cell activation and proliferation (25), signalling required for Treg proliferation in the absence of TCR stimulation remains largely unknown. In the present study, we show that OX40L induced signalling was essential but not sufficient for GM-BMDC mediated Treg expansion (Fig 1). This indicated that the Treg expansion may require additional molecular interactions. However, neither signalling by soluble mediators nor by CD80/86 co-stimulatory molecules was essential for the GM-BMDC mediated Treg expansion (2A and 2B). Since the addition of a γ-secretase inhibitor (GSI) prevented GM-BMDC mediated Treg expansion (Fig 3A and and Fig-S3), we suspected a critical role for Notch signaling in this process. Interestingly, GM-BMDC had higher expression of Jagged1 and little or no expression of the other notch ligands Jagged2, DLL1, DLL3 and DLL4 (Fig 3C). Additionally, Jagged1 blockade using a blocking antibody abrogated Treg expansion, suggesting a role for Jagged1 (Fig3D). Simultaneous inhibition of OX40L and Jagged1 mediated signaling by combination of blocking antibodies and SiRNA completely abrogated GM-BMDC mediated Treg proliferation (Fig 4A–D). Further investigation into the relationship between the ability of OX40L⁺ GM-BMDC to mediate Treg proliferation and the requirement of Jagged1 expression showed that only the OX40L⁺ Jagged1⁺ GM-BMDC were capable of inducing significant Treg expansion (Fig 5, A–B). In trying to determine the most likely Notch receptor expressed on Tregs through which Jagged1 was mediating signaling, we detected differential higher levels of expression of Notch3 transcripts in Foxp3⁺ Tregs relative to Foxp3- effector T cells (Fig 6A). Remarkably, blockade of Notch3 mediated signalling, but not Notch1 mediated signalling, prevented Treg expansion (Fig 6B). Detection of Notch3 specific NICD in the proliferating Tregs (Fig 6C) supported the notion that Jagged1 was likely inducing signalling through Notch3. Collectively, our studies suggest that co-signaling by OX40L through OX40 and Jagged1 through Notch3, was essential for GM-BMDC mediated Treg expansion. Consistent with our ex vivo data, our in vivo studies showed that mice that were adoptively transferred with OX40L⁺ Jagged1⁺ GM-BMDCs had increased numbers of Foxp3⁺ Tregs and IL-10⁺CD4⁺ T-cells while they showed a decrease in IFN-γ⁺CD4⁺ T-cells compared to mice that received OX40L⁺ Jagged1⁻ GM-BMDCs (Fig 7B, 7C & 7D). Additionally, mice that received OX40L⁺ Jagged1⁺ GM-BMDCs showed a significant reduction in thyroid infiltration of lymphocytes relative to controls (Fig 7E). These results indicated that treatment of mice with OX40L⁺ Jagged1⁺ GM-BMDCs caused suppression of ongoing EAT likely through the expansion of CD4⁺IL-10⁺ Tregs. This observation is consistent with our earlier findings which showed that the protection conferred by the treatment with low dose GM-CSF was primarily mediated through increased production of IL-10 as a result of expansion of IL-10⁺CD4⁺Foxp3⁺T-regs in these mice (6).

Notch3 mediated signalling has been reported to sustain regulatory phenotype on Tregs (26). Furthermore, the thymocytes and T cells from transgenic mice expressing Notch3 NICD (N3-tg mice) in which Notch3 is constitutively active contain a significantly higher proportion of CD4⁺CD25⁺ cells (24). Our observations on the critical roles of OX40L and Jagged1 in GM-BMDC mediated Treg expansion stand in support of earlier studies that have implicated Notch3 mediated signalling in Treg expansion. Although other studies have shown a role for Jagged2 mediated activation of Notch3 in Treg expansion (12), such an interaction was not likely in the present study due to little or no expression of Jagged2 (less than 1% of the cells) on GM-BMDCs.

While OX40 is constitutively expressed on Tregs (27), the Notch3 is preferentially expressed on Tregs (24). However, it is not yet clear as to how signalling mediated through OX40 and Notch3 receptors co-operate to cause Treg expansion. In the context of TCR signalling, OX40 mediated signalling can increase T cell proliferation by activating PI3 kinase (PI3K) and Akt, which are upstream activators of mTOR (28). GM-BMDCs derived from MHC class-II knockout mice were also able to expand Tregs and indicated that TCR signalling was not necessary (8). In this context, it is interesting to note that OX40 activation can form a signalosome consisting of CARMA1, PKC-Q and TRAF2 and cause enhanced NF-κB activation and contribute to cell survival and expansion (29, 30). Notch3 has been reported to activate both the alternate and the canonical NF-κB pathways. It can activate the alternative (RelB) NF-κB pathway in murine thymocytes (31) via cytoplasmic IKKα and cooperate with canonical NF-κB in stimulating FoxP3 expression (32). Thus NF-κB may be an important point of convergence between OX40 and Notch3 signalling in Tregs.

Notch1 has been reported to maintain the expression of FoxP3 in peripheral Tregs in collaboration with TGFβ (33). Therefore, it is possible that different Notch paralogs can maintain FoxP3 expression depending on other signals and cellular context. It is well known that Foxp3⁺ Tregs are unable to proliferate or proliferate poorly when stimulated (34, 35) and upon proliferation they lose Foxp3 expression. Notch3 has been shown to co-operatively regulate Foxp3 expression through trans-activation of the Foxp3 promoter (32). Therefore, it is likely that the interaction of Jagged1 with Notch3 helps sustain Foxp3 transcription while OX40 signalosome formation, in the absence of TCR signalling, may drive Foxp3⁺ Treg cell-proliferation. Thus, concurrent signals from Notch3 and OX40 may allow Treg proliferation while sustaining Foxp3 expression.

Our observations have strong implications for the treatment of various autoimmune diseases. Deficiency of naturally occurring Tregs has been observed in a variety of autoimmune conditions (36, 37). Moreover, adoptive transfer of polyclonal or antigen selected nTregs can overcome autoimmune and allergic conditions (38–40). A limitation that prevents therapeutic utilization of Tregs in autoimmune diseases is the relative difficulty in obtaining large numbers of Tregs. Identifying key signaling molecules may help overcome this limitation and allow ready generation of large numbers of Tregs. Our data strongly suggest that co-culturing T cells with OX40L⁺ Jagged1⁺ GM-BMDCs may provide a simple and efficient method for selective expansion of polyclonal Tregs ex vivo, which can then be adoptively transferred as a potential therapy for human autoimmune conditions. If soluble OX40L and Jagged1 can similarly induce Treg expansion, it may be feasible to use them directly to treat various autoimmune conditions.

Abbreviations

Flt3L

fms-like tyrosine kinase 3-ligand

Treg

regulatory T-cells

nTreg

natural regulatory T cells

iTreg

inducible Treg

mTg

mouse thyroglobulin

EAT

experimental autoimmune thyroiditis

EAE

Experimental autoimmune encephalomyelitis

BM

bone marrow

DCs

dendritic cells

GM-BMDCs

bone marrow dendritic cells

SpDC

spleen derived dendritic cells

L+

OX40L+

L−

OX40L−

J1

Jagged 1

GSI

Gamma-secretase inhibitor

NICD

notch-intracellular-domain