|Home | About | Journals | Submit | Contact Us | Français|
T-bet is essential for nTreg to regulate Th1 inflammation, but whether T-bet controls other Treg functions after entering the inflammatory site, is unknown. In an islet allograft model, T-bet−/−nTreg but not iTreg failed to prolong graft survival as effectively as wild type Treg. T-bet−/− nTreg had no functional deficiency in vitro but failed to home from the graft to dLN as efficiently as wild type. T-bet regulated expression of adhesion and migration related molecules, influencing nTreg distribution in tissues, so that T-bet−/− nTreg remained in the grafts rather than migrating to lymphatics and dLN. In contrast, both wild type and T-bet−/− CD4+ Tconv and iTreg migrated normally toward afferent lymphatics. T-bet−/− nTreg displayed instability in the graft, failing to suppress antigen-specific CD4+ T cells and prevent their infiltration into the graft and dLN. Thus, T-bet regulates nTreg migration into afferent lymphatics and dLN and consequently their suppressive stability in vivo.
Differentiated effector CD4+ T cells are categorized into several major subsets, which rely on expression of specific transcription factors. T-bet (Tbx21) and its paralogue Eomesodermin (Eomes) are T-box derived transcription factors essential for mounting a Th1 pro-inflammatory response(1). T-bet acts as a master regulator that is both necessary and sufficient for the development of Th1 cells. Sustained expression of T-bet in Th2 cells not only elaborates Th1-specific cytokines, but also suppresses Th2-specific signaling(2). Gene regulation analysis shows that T-bet binds to the promoters of genes encoding migration related molecules(3, 4). Accordingly, T-bet is essential for selective Th1 cell trafficking to inflamed sites via P-selectin binding and CXCR3 expression(5–7).
Regulatory T cells (Treg) are important for tolerance and for limiting inflammatory responses(8). Specialized Treg populations have been identified whose transcriptional profiles are similar to their effector T cell counterparts and critical for their suppressive functions. For example, STAT-3 is required for Treg expression of CCR6, which directs Treg migration to areas of Th17 inflammation to enable suppression(9, 10); GATA-3 and IRF4 expression by Treg help to regulate Th2 cells by promoting expression of Foxp3(11, 12); and Bcl6 is required by follicular Treg to induce expression of CXCR5 and suppress Tfh cells in germinal centers(13, 14). T-bet expression in Treg is necessary for regulation of Th1-inflammation in various disease models(15, 16). In Foxp3-deficient scurfy mice, adoptively transferred T-bet−/− Treg do not prevent wasting disease, and their cytokine expression profile shows decreased levels of the immunosuppressive cytokines TGFβ and IL10, suggesting that T-bet is important for Treg inhibitory effector function(15). Further, T-bet regulates CXCR3 expression, and is required for effective Treg migration to and suppression of Th1 inflammatory sites that are rich in the ligands CXCL9 and CXCL10(15). In disease models dependent on Th2 or not dependent on CXCR3, T-bet−/− nTreg remain functionally intact(17–19). These observations demonstrate that T-bet is required to regulate various Treg functions to suppressTh1 immune responses.
Treg employ a variety of migration related molecules to temporally and geographically regulate their full immunosuppressive function. During cutaneous inflammation, activated Treg that migrate from skin to dLN and re-circulate back to periphery express higher levels of inhibitory molecules such as IL10, TGFβ and CTLA-4 compared to non-circulating LN-resident Treg(20). This suggests a relationship between Treg migration, trafficking, and their immunosuppressive capacity. We previously showed that to prolong islet allograft survival Treg must migrate first from blood to sites of tissue inflammation in a process dependent on CCR2, CCR4, CCR5, P- and E-selectin; and then sequentially migrate from the tissues via afferent lymphatics to the draining lymph nodes (dLN) dependent on CCR2, CCR5 and CCR7(21). Sequential blood to tissue to dLN migration was also accompanied by activation of Treg and increased expression of effector suppressor molecules. T-bet−/− Treg express many of the same homing receptors as wild type Treg, such as CD103, CCR6, and P-selectin ligands, but do not express CXCR3(15). These observations support a role for T-bet in the regulation of Treg migration from blood to tissue inflammatory sites. However, the relationship among T-bet and Eomes, trafficking related receptors, and effector molecules for the regulation of Treg suppression and migration has not been fully explored. We hypothesized that T-bet regulated Treg trafficking and suppression by affecting additional aspects of migration and functions distinct from Treg entry into the inflammatory site. Using murine models of tissue to afferent lymphatic migration and islet allografting, we first demonstrated that Treg migration to afferent lymphatics and dLN required T-bet expression, and T-bet regulated several receptors critical for chemotaxis and adhesion; and this regulation was specific for T-bet but not Eomes, and specific for nTreg but not iTreg or Tconv; further, retention of Treg within the allograft impaired suppressive stability and efficacy, highlighting a new site for regulation of Treg suppressive function.
C57BL/6 (wild type and T-bet−/−, H-2b) and BALB/c mice (H-2d) aged 10-12 weeks were purchased from The Jackson Laboratory (Bar Harbor, ME). Foxp3-GFP(22) and TEa T cell receptor (TCR) transgenic(23) mice on a C57BL/6 background were from A. Rudensky (Memorial Sloan Kettering Cancer Center, NY, NY). The TEa TCR transgene recognizes I-Ed peptide presented by I-Ab. Eomes−/− (24) mice on a C57BL/6 background were from A. Banerjee (University of Maryland, Baltimore, MD). T-bet−/− mice were crossed with Foxp3-GFP mice to generate T-bet−/− Foxp3-GFP transgenic mice. All experiments were performed with age- and sex-matched mice in accordance with Institutional Animal Care and Utilization Committee approved protocols.
Collagenase P was from Roche Diagnostics (Mannheim, Germany). FITC-anti-CD25 (clone 7D4), PE-Cy7-anti-Foxp3 (clone 3G3), PE-Cy7-anti-CD62L (clone MEL-14), anti-LFA-1 (clones 2D7 and M17/4), anti-PSGL-1 (clone 4RA10), P-selectin-Ig, E-selectin-Ig, APC-anti-CD103 (clone 2E7), anti-integrin av (clone RMV-7), PE-Cy7-anti-CD44 (clone IM7), FITC-anti-CD69 (clone H1.2F3), APC-anti-CCR2 (clone 475301), APC-anti-CCR6 (clone 11A9), PE-anti-CCR5 (clone C34 3448), anti-granzyme B (clone 16G6), anti-CTLA-4 (clone UC10-4F10-11), anti-IL10 (JES5-2A5), and anti-TGFβ (clone 1D11) were from BD PharMingen (San Diego, CA). APC-eFluor780-anti-CD4 (clone RM4-5), PerCP-Cy5.5-anti-CD8 (clone 53-6.7), APC-anti-CD8 (clone 53-6.7), PE-anti-CD25 (clone PC61.5), FITC-anti-Foxp3 (FJK-16s), PE-anti-CCR7 (clone 4B12), anti-CD3 (clone HIT3a), APC-anti-CXCR3 (clone CXCE3-173), rat monoclonal IgG, anti-CD62L (MEL-14), control rat IgG and recombinant protein IL2 and TGFβ were from eBioscience (San Diego, CA). Brilliant violet-anti-CCR4 (clone 2G12) and anti-CXCL9 (MIG-2F5.5) were from Biolegend (San Diego, CA). Anti-CXCL10, recombinant murine CCL19, CCL5, CCL22, CXCL10 and E-cadherin-Fc were from R&D Systems (Minneapolis, MN). Guinea pig anti-swan insulin was from Dako Cytomation, Inc. (Carpinteria, CA). RPMI-1640, DMEM and fetal bovine serum (FBS) were purchased from Mediatech, Inc. (Manassas, VA). Sodium pyruvate, L-glutamine, penicillin/streptomycin, nonessential amino acids, and 2-Mercaptoethanol (2-ME) were purchased from Lonza Group Ltd. (Visp, Switzerland).
BALB/c islets were isolated as previously reported(21). In brief, pancreata were perfused with collagenase P, harvested, digested, and purified over a discontinuous Ficoll (Sigma-Aldrich, St. Louis, MO) gradient. 400 freshly isolated islets were handpicked and transplanted beneath the renal capsule of C57BL/6 recipients made diabetic (blood glucose >300mg/dl) by intraperitoneal injection of 160 mg/kg streptozotocin 1 week prior to transplant. Blood glucose <150mg/dl after transplantation was considered engraftment, and >300mg/dl was considered graft rejection. The advantages of our islet model are that the recipient is exposed to antigen at a defined time and place. There is limited afferent lymphatic drainage with only a single dLN to the kidney, so that migration from blood to tissue to afferent lymphatics to dLN is precisely defined. The contralateral kidney and non-dLN serve as important internal controls for specificity; or if more cells and tissues are required, bilateral transplantation to both kidney capsules is performed. Adoptive transfer of cells differentiates place (iv vs. intra-islet), migration patterns (blood, tissue, lymph, LN), cell types (Treg, Tconv), and molecules (gene sufficient or deficient). This model precisely answers questions about Treg and Tconv in tissues and LN at specified times and places.
iTreg were generated as previously reported(21). In brief, mice were euthanized, spleens were removed and gently dissociated into single cell suspensions, CD4+ T cells were purified using Easysep mouse CD4+ enrichment kit (STEMCELL Technologies, Inc., Vancouver, Canada). CD4+ T cells were then stained with APC-efluor780-anti-CD4, APC-anti-CD8, PE-anti-CD25 and PE-Cy7-anti-CD44. CD4+CD25−CD44low (Tconv cells) and CD4+CD25+ (nTreg) were sorted with a FACSAria flow cytometer (BD Biosciences, Mountain View, CA). In a 96 well plate, each well contained 5×104 Tconv cells suspended in 200 μl RPMI1640 supplemented with 10% FBS, 1mM sodium pyruvate, 2mM L-glutamine, 100 IU/ml penicillin, 100ug/ml streptomycin, 1× nonessential amino acids, and 20μM 2-ME (termed complete RPMI1640 medium), stimulated with 20ng/ml IL2, 1μg/ml anti-CD3 mAb, and 10ng/ml TGFβ in the presence of 5×104 T cell-depleted splenocytes(21) isolated according to manufacturer’s protocol (STEMCELL Technologies, Inc.). 5 days later, iTreg were harvested, stained with APC-eluor780-anti-CD4, APC-anti-CD8 and PE-anti-CD25, and re-sorted by gating on CD4+CD25+ T cells (iTreg). Foxp3 positivity was greater than 98% in CD4+CD25+ cells (iTreg). In a 96 well plate, 5×104 nTreg were stimulated with 20ng/ml IL2, 1μg/ml anti-CD3 mAb, and 10ng/ml TGFβ in the presence of 5×104 T cell-depleted splenocytes in 200μl complete medium for 3 days. Our previous report clearly showed nTreg and iTreg obtained by these protocols were stable and clearly differentiated by methylation patterns of the upstream of Foxp3 enhancer(25).
1×106 freshly isolated nTreg, or iTreg generated from C57BL/6 wild type or T-bet−/− mice (see above), were transferred either i.v. in 300ml of PBS, or locally together with islets in a 20μl pellet, at the time of transplantation. Purity of CD4+CD25+ T cells was >90% and Foxp3 positivity was 90-95% in CD4+CD25+ cells.
Recipient mice were euthanized, islet grafts were exposed by partially peeling the capsule from the kidney surface, then harvested, digested with collagenase P (1.5mg/ml) in a 37°C water bath for 5 minutes, and made into a single cell suspension through a nylon mesh.
TEa TCR transgenic mice do not express the homologous antigen and thus do not normally have nTreg or iTreg. TEa mice were tolerized with an intravenous (i.v.) injection of 1×107 BALB/c spleen cells and 0.25 mg anti-CD40L mAb (clone MR1; Bio X Cell, West Lebanon, NH) on day 0, and another 0.25 mg anti-CD40L mAb injection on day 3. On day 7, mice were euthanized and spleen and LN harvested and analyzed for iTreg generation by flow cytometry.
Staining was performed with the indicated antibody (1mg/106 cells) at 4°C for 30 minutes, except CCR7 at 37°C for 30 minutes. Cells were analyzed with an LSRFortessa cell analyzer (BD Biosciences). Positive staining was defined with FMO or isotype controls and we observed no meaningful differences between those controls. Data were analyzed with FlowJo software (Tree Star, Ashland, OR).
nTreg were CSFE-labeled; stimulated in round-bottom 96-well plates with 20 ng/ml IL2 with or without 2μg/ml anti-CD28 and plate-bound (5 μg/ml) anti-CD3 mAb in 200μl complete RPMI1640 medium for 3 days; and then stained with viability dye fluor450 (eBioscience) according to the manufacturer’s protocol, anti-CD4-APC-Cy7, anti-CD25-PE and anti-Foxp3-APC; and then analyzed by flow cytometry.
mRNA expression levels for chemokine receptors and adhesion molecules were quantified by real-time PCR (qRT-PCR) using the SYBR Green PCR kit (QIAGEN, Valencia, CA) and Applied Biosystems 7900HT Fast Real-time PCR System (Life Technologies, Grand Island, NY). RNA isolation, reverse transcription, PCR conditions, and primer sequences were previously published(21, 26). Additional primers included: CD73 forward: 5’-GCTTCAGGGAATGCAACATG-3’ reverse: T5’-GCCACCTCCGTTTACAATG-3’; and CD39 forward: 5’-CGAGAAGGAGAATGACACAGG-3’ reverse: 5’-ATGTTGGTATCAGTTCGGTGG-3’. PCR consisted of a 15 minute 95°C denaturation step followed by 45 cycles of 15s at 94°C, 20s at 56°C, and 20s at 72°C. Values for mRNA expression of specific genes were presented as a % of cyclophilin A and were calculated as: 2(Ct of Cyclophilin A — Ct specific gene) × 100. Each RNA sample was run in triplicate, and each experimental group consisted of three individual samples.
Subsets of CD4+CD25+ Treg cells and CD4+CD25− T cells were sorted from indicated strains and used in a suppression assay. Freshly sorted responder eFluo670 (Molecular Probes, Eugene, OR) labeled CD4+CD25− T cells (5×104 cells/well) were cultured with or without addition of Treg cells at responder:Treg ratios of 1:0, 1:1, 2:1, 4:1, and 8:1 with irradiated (800 Rad) syngeneic T cell-depleted splenocytes (5×104 cells/well) and anti-CD3 mAb (1mg/ml) for 3 days. Cells were harvested, and cell division was measured by assessing relative eFluo670 dilution by an LSRFortessa cell analyzer.
1×106 freshly isolated, efluor670-labeled nTreg were transferred locally with islets beneath the renal capsule. At the same time, 1×106 CFSE-labeled TEa CD4+ T cells were injected i.v. into the same recipients. Four days later, islet grafts, dLN, ndLN and spleen were collected, stained with the indicated antibodies, and analyzed by flow cytometry.
A total of 2×105 sorted CD4+CD25+ or CD4+CD25− cells of each strain were added in a volume of 100ml to the upper chambers of a 24-well transwell plate with a 5mm insert (Corning International, Corning, NY)(27). Lower wells contained various concentrations of chemokines: 500ng/ml CCL19, 200ng/ml CCL5, 100nM S1P or 100ng/ml CCL22 in 600ml of RPMI 1640/0.5% fatty acid-free BSA (Sigma-Aldrich). The number of T cells that migrated to the lower well after 4 hours was counted with a hemocytometer. The percentage of migrated cells was calculated as number of transmigrated cells over total cell input. Each experiment was performed in triplicate at least three times.
Fresh islets were purified as described above and previously(21). 50 islets were added to the upper chamber of a 96-well insert along with 1×105 naïve T cells or nTreg. The lower chamber contained 250ng/ml CCL19. To block CD103 with E-cadherin, nTreg were re-suspended with RPMI 1640/0.5% fatty acid-free BSA supplemented with or without 2mg/ml E-cadherin-Fc. Three hours later, the number of T cells that migrated to the lower well was counted with a hemocytometer.
5×106 CFSE-labeled wild type CD4+ and efluor670-labeled T-bet−/−CD4+ T cells were injected intravenously into wild type C57BL/6 mice. 24 hours later, recipient mice received 100μg rat monoclonal IgG or anti-CD62L (MEL-14) mAb (eBioscience) to block further T cell entry into LN(28). 18 hours after anti-CD62L or control mAb treatment, LN and spleens were collected and stained with APC-efluor780-anti-CD4, PE-anti-CD25 and PE-Cy7-Foxp3; and the samples analyzed with flow cytometry.
1×106 Treg were labeled with 5uM carboxyfluorescein diacetate succinimidyl ester (CFSE) (Molecular Probes, Eugene, OR) for 10 min at 37°C and injected into mouse ears intradermally. 12 hours later, ears were collected and peeled into two halves, fixed with 3% paraformaldehyde (PFA) in PBS for 5 minutes at 4°C, permeabilized with 1% Triton X in PBS for 30 minutes at 4°C, incubated with 5% donkey serum for 30 minutes at 4°C, and incubated with the primary antibody at 4°C overnight. The ears were then washed with PBS, incubated with secondary antibody for 4 hours at 4°C, washed with PBS, and fixed with 3% PFA at 4°C for 10 minutes. The stained ears were analyzed by fluorescence microscopy (Nikon Eclipse E800, Nikon Co., Tokyo, Japan) or confocal microscopy (Zeiss LSM510, Jena, Germany).
5mm frozen sections of LNs and islet grafts in the kidneys were fixed with acetone. After incubation with blocking buffer (PBS with 5% donkey serum) for 30 minutes, sections were incubated with primary antibody in PBS with 5% donkey serum for 1 hour, followed by secondary antibody in PBS with 5% donkey serum for 40 minutes at room temperature, and mounted with Vectashield with DAPI (Vector Laboratories Inc., Berlingame, CA). Quantitative analysis was performed with Volocity 3D Image Analysis Software (PerkinElmer, Waltham, MA). T cells were enumerated as either on the luminal side of lymphatic vessel, outside yet touching the lymphatic vessel edge, or not touching the lymphatic vessel. For each cell not touching the lympatic, the distance from center of cell to edge of nearest lymphatic was measured.
As previously described(26), Treg were labeled with CFSE, incubated with 1mg/ml control IgG (Life Technologies, Grand Island, NY) or E-cadherin-Fc for 15 min at 37° C, and 2×106 Treg in 20 ml PBS were injected into the footpads to assess LN migration. 12 hours after injection, popliteal LNs were harvested and single cell suspensions or tissue sections prepared for flow cytometric analysis.
CD4+CD25+Foxp3-GFP+ nTreg were sorted from Foxp3-GFP or T-bet−/− Foxp3-GFP mice, labeled with efluor670, and the labeled nTreg transferred together with islets underneath the renal capsules of diabetic mice. Four days later, dLN and islet grafts were collected from the recipient mice, stained with anti-CD4-APC-efluor780, CD8-Percp-efluor710 and CD25-PE, and sorted to calculate the percentage of CD4+Foxp3-GFP+efluor670+ from CD4+efluor670+ cells. Isolated CD4+Foxp3-GFP+efluor670+ nTreg were analyzed by qRT-PCR to measure effector molecule expression. CD4+CD25+Foxp3-GFP+ and CD4+Foxp3-GFP+efluor670+ purity were both greater than 99%. Between 1000-2000 nTreg cells (CD4+CD25+Foxp3-GFP+efluor670+) could routinely be recovered from grafts of the recipient mice to perform the qRT-PCR experiments.
In vivo survival and migration results represent pools of more than 4 mice per group per experiment. In vitro migration results represent mean values of triplicate samples. All experiments were performed at least 3 times. All results were analyzed by GraphPad Prism Software (version 5, GraphPad Software, Inc. La Jolla, CA) and presented as the mean±SD. Graft survival curves were generated and compared using Kaplan-Meier life-table analysis and log-rank test. Statistical analyses were performed using Student’s t test. One-way ANOVA was used for multiple comparisons. P<0.05 was considered statistically significant.
All mouse procedures were performed in accordance with protocols approved by the Institutional Animal Care and Utilization Committee at the University of Maryland School of Medicine.
Treg use T-bet to regulate Th1-inflammation(15). Since islet allograft rejection in C57BL/6 mice is primarily Th1 mediated(29, 30), we investigated whether T-bet expression by Treg also influenced graft survival. Freshly isolated wild type nTreg or T-bet−/− nTreg were adoptively transferred along with islet transplants into diabetic mice. As expected, wild type nTreg significantly prolonged graft survival compared to untreated mice when transferred i.v. (mean survival time (MST) = 30.5 vs 10.3 days (d), p < 0.0001). In contrast, T-bet−/− nTreg failed to prolong islet allograft survival after i.v. transfer (MST = 17d) (Figure 1A). To exclude the possibility that T-bet−/− nTreg failed to migrate from blood to islet graft, nTreg were transferred locally beneath the renal capsule with the islets. As expected, wild type nTreg prolonged graft survival even more effectively compared to i.v. administration (MST = 37d). However, locally transferred T-bet−/− nTreg still failed to prolong islet allograft survival effectively (MST = 17.7d) (Figure 1A). Unlike nTreg, there was no difference in graft survival between wild type and T-bet−/− iTreg after i.v. (MST=12d vs 14d) or local transfer (MST=32.4d vs 26d) (Figure 1A). Thus T-bet regulated the function of nTreg but not iTreg. It should be noted that in this model only the transferred Treg were T-bet−/−, all other endogenous effector cells and Treg were wild type, so that epiphenomena of the gene deficiency were minimized or obviated.
Since both locally and i.v. transferred T-bet−/− nTreg failed to prolong graft survival, these results suggested that T-bet−/− nTreg: 1) had a functional suppressive deficiency; 2) had other functional defects such as cell survival, activation, or proliferation; and/or 3) failed to migrate into LN from grafts.
Both wild type and T-bet−/− nTreg and iTreg similarly suppressed anti-CD3 mAb driven proliferation of Tconv cells (Figures 1B). Treg effector molecules Foxp3, CTLA-4, TGFβ, IL10, CD39, CD73, perforin, and granzymes A and B analyzed by qRT-PCR or flow cytometry were expressed similarly or differed by less than two fold in these cells, but T-bet−/− nTreg expressed less IFNγ than wild type nTreg, as expected, whether freshly isolated or stimulated in culture (Supplementary Fig. 1A-C). These cells expressed similar levels of Bcl-2 and Bim, (Supplementary Fig. 1A and 1B), and tissue analyses showed equivalent or even better survival or presence of T-bet−/− nTreg within allografts (vide infra) making it unlikely that decreased cell survival explained differences in immunosuppressive effects. T-bet−/− nTreg proliferated slightly better than wild type nTreg (Figure 2A, representative data on left, summary on right), although a higher percentage of dividing T-bet−/− nTreg underwent apoptosis (Figure 2B), and both wild type and T-bet−/− nTreg had very similar Foxp3 protein expression after stimulation (Figure 2C, left panel), so that the overall recovery of wild type and T-bet−/− nTreg as judged by total cell number was identical (Figure 2C, right panel). Since T-bet−/− mice behave normally in vivo without evidence of autoimmune or inflammatory pathologies(31, 32), and since T-bet−/− Treg suppressed Tconv proliferation as well as wild type in vitro(15) (figure 1B), these results suggested that the defect in the T-bet−/− was unlikely to be an epiphenomenon of the gene deficiency. These results suggested that T-bet must influence nTreg suppressive functions important for graft survival via mechanisms other than direct T cell suppression or survival.
Previously we showed that Treg must sequentially migrate from blood to graft to dLN to optimally suppress the alloimmune response and prolong islet allograft survival(21). To compare in vivo migration of wild type and T-bet−/− nTreg, islets, draining LN (dLN), non-draining LN (non-dLN), and spleen were harvested 4-7 days after transplant and nTreg adoptive transfer. After i.v. transfer, both nTreg subsets equally migrated from blood into allografts (Figure 3 and Supplementary Fig. 1D). Both were also found similarly distributed in dLN, ndLN, and spleen (Figures 3 and Supplementary Fig. 1D). Therefore T-bet−/− nTreg were fully capable of traversing diverse microvascular endothelia into different tissues and secondary lymphoid organs, despite differences in expression in several receptors, including CXCR3(15). These data also further suggested that there was no difference in cell survival in vivo between the wild type and T-bet−/− cells.
When nTreg were co-transferred with the islet allografts beneath the renal capsule, a marked difference in cell distribution was observed. The T-bet−/− nTreg remained in the graft and failed to traffic into the dLN (p<.0001), while wild type nTreg successfully migrated to the dLN (Figures 3 and Supplementary Fig. 1D). There was a tendency (that did not reach significance) for both a greater percentage (Figure 3) and number (Supplementary Fig. 1D) of T-bet−/− nTreg to remain in the grafts compared to wild type, compatible with the lack of lymphatic migration and again suggesting that there was no survival difference in vivo between the two nTreg subsets within the grafts. Together, these data showed that T-bet−/− nTreg could traffic into dLN from blood through HEV, but failed to home to dLN from graft through afferent lymphatics.
To investigate the mechanisms for differential migration, T-bet−/− and wild type nTreg were analyzed by flow cytometry and qRT-PCR for several important trafficking receptors. While many adhesion and migration related molecules were expressed at identical or similar levels on both resting and activated nTreg (Supplementary Fig. 2A and 2B), there were a few notable differences. As expected(15, 33), resting Tbet−/− nTreg expressed less CXCR3 and slightly more CD103 than wild type nTreg (Figure 4A, upper panel). Since Treg become activated in vivo after exposure to alloantigen(21), and iTreg were generated by in vitro stimulation, nTreg were also stimulated in vitro to determine if this influenced receptor expression. Activated T-bet−/− nTreg expressed substantially less CXCR3 and more CD103 than activated wild type, further accentuating their differences (Figure 4A, lower panel). Activated T-bet−/− nTreg also expressed more CCR4 and significantly less S1P1 than wild type nTreg (Figure 4A and 4B). CCR4+CD103+ nTreg were also more frequent among T-bet−/− than wild type nTreg after stimulation (Figure 4C). Because Eomesodermin is an orthologue of T-bet, we hypothesized that these two transcription factors might have similar influence over Treg expression of the receptors above, however, Eomes−/− nTreg showed no differences in expression of these receptors when compared to wild type nTreg under resting or activated conditions (Supplementary Fig. 2C and 2D). These data implicated T-bet alone in influencing CXCR3, CCR4, CD103, and S1P1 expression and thus in the regulation of Treg afferent lymphatic migration.
Phenotypic differences were next investigated in nTreg recovered 4 days after local co-transfer of islets and Treg. Compared to wild type, CD103 expression was increased on T-bet−/− nTreg in the islet allograft and dLN by flow cytometry (Figure 4D). Expression of other receptors was measured by qRT-PCR due to collagenase treatment reducing surface receptor expression during tissue digestion for Treg recovery. T-bet−/− nTreg recovered from grafts expressed more CCR4 and less S1P1 transcripts than wild type nTreg (Figure 4D), confirming the in vitro findings.
In contrast to nTreg, both wild type and T-bet−/− iTreg had similar expression of these trafficking molecules, while the overall expression was lower in iTreg compared to nTreg (Supplementary Fig. 2E). Wild type and Eomes−/− iTreg also had similar receptor expression (Supplementary Fig. 2F).
T-bet was measured in the various T cell subsets. Wild type nTreg recovered from islets grafts expressed even more T-bet compared to pre-injection, which suggested that T-bet induction played an important role in their function in the allografts (Figure 4E). Wild type iTreg induced in vitro expressed very low levels of T-bet, measured by both qRT-PCR and flow cytometry, compared to wild type Tconv, resting nTreg, and activated nTreg (Figures 4F and G). In vivo generated iTreg also had minimal T-bet expression (Supplementary Fig. 3A). These data further supported the evidence that T-bet was important for regulation of receptor expression in Treg, differential expression between nTreg and iTreg, and influenced differences in nTreg migration.
Since resting and/or activated T-bet−/− nTreg expressed more CCR4 and CD103 and less S1P1 and CXCR3 than wild type nTreg, we hypothesized that their migration and adhesion characteristics would also differ. Transwell migration assays were performed using various chemokines, including the corresponding ligands for these receptors. As predicted, resting or activated T-bet−/− nTreg migrated more effectively toward the CCR4 ligand CCL22 (Figure 5A) and less toward the CXCR3 ligand CXCL10 (Supplementary Fig. 3B)(15). T-bet−/− nTreg also adhered better to the CD103 ligand E-cadherin compared to wild type nTreg (Figure 5B). In contrast, migration to CCL5 and CCL19 were similar between wild type and T-bet−/− nTreg (Supplementary Fig. 3B, Figure 5E), which was also predicted since both expressed similar levels of CCR5 and CCR7 (Supplementary Fig. 2). Resting wild type and T-bet−/− nTreg expressed approximately the same amount of S1P1 (Figure 4B), and migrated toward S1P with the same efficiency (Figure 5C, left). Activated T-bet−/− nTreg expressed markedly less S1P1 (Figure 4B), and migrated significantly less toward S1P compared to wild type nTreg (Figure 5C, right). In contrast, wild type and T-bet−/− iTreg expressed the same amount of S1P1, and migrated toward S1P with the same efficiency (Supplementary Fig. 3B). Consistent with their similar expression profiles, wild type and T-bet−/− iTreg migrated toward CCL22 and adhered to E-cadherin with the same efficiency (data not shown). Together these results show that T-bet regulated expression of several receptors, which resulted in commensurate and measurably significant differences in migration and adhesion.
To determine if these Treg migration characteristics were relevant for their interaction with islets, we first determined islet expression of the various ligands. Confirming our earlier investigations(27), freshly isolated islets did not express significant transcripts for CXCL9, CXCL10, or CCL22 and only low levels of CCL17 (Supplementary Fig. 3C). Further, when wild type or T-bet−/− nTreg were added to the upper chamber and islet supernatant from freshly isolated islets to the lower chamber of a transwell plate, both cell subsets migrated poorly (Figure 5D and E), suggesting that there were few active chemotactic molecules produced. When CCL19 was added to the lower chamber and islet supernatant added to the upper chamber along with nTreg, the nTreg subsets migrated equally, suggesting that an islet-secreted factor was not responsible for inhibiting Treg migration (Figure 5E).
To determine if the islets themselves influenced migration, CCL19 was added to the lower chamber and fresh islets plus nTreg added to the upper chamber. This combination resulted in significant inhibition of migration, with T-bet−/− nTreg inhibited to a greater degree than wild type nTreg. In contrast, there was no migration inhibition and no difference observed between wild type and T-bet−/− Tconv or iTreg migration toward CCL19 (Figure 5F and data not shown). These results suggested that T-bet−/− nTreg preferentially remained with the islets, likely due to adhesion to the islets rather than preferential migration into the islets or an inhibitory factor secreted by the islets. Since the T-bet−/− nTreg expressed elevated CD103 and since the islets expressed E-cadherin(34), we tested if nTreg retention with the islets was due to this specific receptor-ligand interaction. Addition of soluble E-cadherin-Fc to the islets and Treg in the upper chamber restored T-bet−/− nTreg migration toward CCL19 to the same level as that of wild type nTreg (Figure 5G). This supported the notion that a mechanism contributing to islet retention and impaired lymphatic migration involved nTreg CD103 adhesion to islet E-cadherin.
Migration of Treg from the graft into the dLN requires trafficking from peripheral tissues toward afferent lymphatics, transmigration across the lymphatic endothelium, and exit from the lymphatics into the LN. There were at least two reasons that T-bet−/− nTreg did not migrate from grafts into dLN: 1) they failed to migrate from tissues toward lymphatics; or 2) they failed to cross the lymphatic endothelium. To examine migration from tissue to lymphatics, islet grafts were recovered from the recipient mice after 4 days. Immunohistochemistry showed that more T-bet−/− nTreg remained in the grafts (insulin+ area) and fewer T-bet−/− nTreg co-localized with the VEGFR-3+ lymphatics compared to wild type nTreg (Figure 6A and B), suggesting that T-bet−/− nTreg were not accessing the lymphatics. The co-localization of Treg and islets was commensurate with the results above of nTreg adherence via the interaction of CD103 and E-cadherin (Figure 5 F and G).
To exclude the possibility that T-bet−/− nTreg did migrate into dLN but then exited faster, trafficking was assessed with the LN equilibration and egress assay(28). Separately labeled wild type and T-bet−/− CD4+ Tconv and nTreg were injected i.v. into C57BL/6 mice treated with or without alloantigen, and allowed to equilibrate for 24 hours. Next anti-CD62L or control IgG mAb was administered i.v. to prevent further entry into LN, and 18 hours later the remaining numbers and ratios of wild type and T-bet−/− Tconv and nTreg in the LN were determined. The ratios of T-bet−/−: wild type nTreg in the LN from the control IgG or anti-CD62L mice were the same, showing that LN egress was the same for the two subsets under both steady state and inflammatory conditions (Supplementary Fig. 3D). Similar results were observed for wild type and T-bet−/− Tconv (Supplementary Fig. 3D). Therefore, T-bet−/− nTreg efferent lymphatic migration was not altered compared to wild type, and the deficiency of T-bet−/− nTreg in the dLN was due to altered afferent lymphatic migration but not faster transit through the LN. These results again demonstrated that there was no survival difference between wild type and gene deficient cells in the wild type environment.
To further confirm that the migration deficiency in vivo was due to the inability to access the lymphatic vasculature, and ensure our findings were not simply model specific, the ear pinna assay for lymphatic migration was used(25). CFSE-labeled wild type or T-bet−/− nTreg were transferred intradermally into ear pinnae. Twelve hours later whole mount staining of the pinnae was performed. Immunohistochemistry showed the position of nTreg in relation to Lyve-1+ afferent lymphatic vessels. More wild type nTreg were located near and within the lymphatic lumen than T-bet−/− nTreg (Figure 6C). Confocal microscopy also located more wild type than T-bet−/− nTreg within the lymphatic lumen (Figure 6D). In contrast, both wild type and T-bet−/− Tconv cells and iTreg migrated into tissue lymphatics with the same efficiency (Figure 6E and 6F). These data demonstrated that T-bet−/− nTreg preferentially remained in the peripheral tissue and did not migrate toward the lymphatic vessels and lumens.
To determine whether nTreg afferent lymphatic migration influenced by T-bet regulated receptor expression corresponded to the distribution of the relevant cognate ligands, immunohistochemistry of the pinnae for the CCR4 ligands CCL22 and CCL17 (Figure 7, Supplementary Fig. 3E) and the CD103 ligand E-cadherin was performed. These molecules were expressed near blood vessels or within the interstitial tissue, but not near the lymphatics (Figure 7A). In contrast, the CXCR3 ligands CXCL9 and CXCL10 co-localized with lymphatic vessels and blood vessels (Figure 7). This ligand expression pattern suggested that the T-bet−/− nTreg receptor profile resulted in a distribution within the tissues that preferentially allowed for tissue retention rather than afferent lymphatic migration.
Chemokine and adhesion molecule expression was next determined in the islet allograft 4 days after transplantation. Compared to freshly isolated islets and the surrounding kidney capsule, the transplanted islets expressed significantly higher levels of the CCR4 ligands CCL17 and CCL22, and the CXCR3 ligands CXCL9 and CXCL10 by qRT-PCR (Figure 7B). IHC of the grafts confirmed expression of CCL17, CCL22, CXCL9, CXCL10, and E-cadherin in and around the islets (Figure 7C). Since T-bet−/− nTreg expressed higher levels of functional CCR4 and CD103 (Figures 4A, 4C, 5A, 5B), expression of CCL17, CCL22, and E-cadherin by the graft was commensurate with the observed retention of T-bet−/− nTreg in the graft.
To analyze whether ligand expression and distribution influenced nTreg homing into the dLN, a third in vivo model was used. Wild type and T-bet−/− nTreg were injected into the footpad and migrated cells retrieved from the draining popliteal node(25). T-bet−/− nTreg failed to migrate into dLN as well as wild type nTreg. However, co-administration of soluble E-cadherin-Fc dramatically increased migration of T-bet−/− nTreg to the dLN (Figure 7D). These results confirm that the CD103-E-cadherin interaction played an important role in retaining T-bet−/− nTreg in tissues.
To determine if there was a difference for in vivo suppression, CFSE-labeled, alloantigen-specific, efluor670-labeled wild type or T-bet−/− nTreg were transferred with the islet allografts while naïve TEa CD4+CD25− T cells were co-transferred i.v. Locally transferred wild type nTreg reduced the number of TEa T cells in both allografts and dLN (Figure 8A), whereas T-bet−/− nTreg did not. T-bet−/− nTreg also failed to inhibit proliferation of TEa CD4+ T cells in allografts and dLN to the same extent as wild type nTreg (Figure 8B). Additionally, there were more macrophages and dendritic cells (DC) in the dLN after T-bet−/− nTreg transfer (Figure 8C). These dLN macrophages and DC expressed more mature markers (CD80 and CD86) (Figure 8D) (Supplementary Fig. 4). Together, these data showed that T-bet−/− nTreg lost the ability to reduce accumulation of specific effector T cells in islets and the dLN, and macrophages and DC in the dLN, likely through effects on migration, proliferation, and/or survival.
T-bet and GATA-3 expression by nTreg are actively regulated to permit Treg to respond to different inflammatory environments and optimize immune suppression(32). Since wild type and T-bet−/− nTreg behaved differently in suppression of CD4 T cell proliferation in vivo (Figure 8A) compared to in vitro (Figure 1B), and since Foxp3 nTreg are unstable under inflammatory conditions(35–37), we determined whether nTreg stability was regulated by T-bet. CD4+CD25+Foxp3-GFP nTreg were sorted from Foxp3-GFP or T-bet−/− X Foxp3-GFP mice, efluor670-labeled, and transferred together with islets into diabetic mice. Four days later, the proportion of Foxp3-GFP+ out of total efluor670+CD4+ in dLNs and grafts were calculated. T-bet−/− nTreg more readily lost Foxp3 expression than wild type within both islets and dLN (Figure 9A and B). Since some cells could have divided rapidly and lost efluor670, these data may represent an underestimate of the percentage of cells that lost Foxp3 expression. Compared to wild type, T-bet−/− also expressed lower IL-10, CD39, CD73 and CTLA4; and there was a nonsignificant trend for decreased TGFβ (Figure 9C). These data showed that T-bet−/− nTreg lost expression of many critical suppressive genes, commensurate with impaired in vivo suppressive function (Figure 8A) and graft protection (Figure 1A).
Here we used three different in vivo migration models, including one disease model, and two other in vivo models under steady state, to demonstrate that T-bet−/− nTreg failed to traffic into dLN, which suggests that: 1) nTreg homing into dLN was not antigen dependent; 2) Treg did not require inflammation to drive their trafficking, even though inflammation may promote Treg trafficking; and 3) T-bet controlled migration and adhesion related molecule expression to regulate Treg migration and distribution. Failure to prolong graft survival was not due to differences in expression of effector molecules, proliferation, or cell survival, but rather due to changes in migration and suppressive capacity. In comparison to wild type, T-bet−/− nTreg expressed more CCR4 and migrated better to the CCR4 ligand CCL22, which was located away from lymphatic vessels. T-bet−/− nTreg expressed more CD103, and adhered better to E-cadherin within the peripheral tissue and islets. In contrast, T-bet−/− nTreg had decreased CXCR3 and S1P1, migrated less toward CXCL10, which was located around the lymphatic vessels, and less toward S1P which is required to migrate across lymphatic endothelial cells (26). These results uncovered a T-bet specific regulation for the migration, adhesion and suppressive functions of nTreg. Further, the requirement for T-bet was specific for nTreg, but not iTreg or non-Treg CD4+ Tconv.
The physical interplay between Foxp3 and T-bet on a genetic or protein level has yet to be defined. The results here (Supplementary Fig. S1A and B) suggested that T-bet does not directly regulate Foxp3 expression; and genome wide analyses have shown that Foxp3 and T-bet do not bind each other’s genes directly(4, 38, 39). Although both bind Cmah, Lrig1, Irf4, Prdm1, Sosc2, and Trat1, they differently regulate their expression. T-bet up-regulates CMAH and LRIG1and down-regulates IRF4, PRDM1, SOSC2, and TRAT1. Foxp3 up-regulates LRIG1, IRF4, SOSC2, and TRAT1 in the thymus; up-regulates PRDM1 in thymus and periphery; and down-regulates CMAH in the periphery(38, 39). Further experiments are warranted to determine whether these shared genetic targets are linked to altered Treg function and migration, and if so, by what mechanisms Foxp3 and T-bet use them to regulate Th1 inflammation. Recent evidence also shows that both T-bet and GATA3 indirectly regulate Foxp3 and in turn Treg function in a manner that depends upon the cytokine environment(32).
CCR4 is associated with protective effects of Treg. Thus CCR4 is highly expressed by Treg residing in peripheral tissue(40); CCR4 mediates protective Treg migration to inflamed tissues(41, 42); and anti-CCR4 mAb treatment depletes Treg and prevents their infiltration into tissues(43). Over-expression of CCL22 mediates nTreg recruitment to pancreatic islets to prevent autoimmune diabetes(44), and recruitment of Treg to an allograft to promote graft survival is dependent on CCR4(45). We previously showed that nTreg recovered from islet grafts expressed elevated CCR4, CCR4−/− nTreg migrate poorly to the inflamed grafts, and local transfer of CCR4−/− nTreg bypassed the need for migration to the graft and enhanced graft survival(21). We showed here that T-bet−/− nTreg expressed more CCR4 and migrated better to CCL22 and transplanted islet expressed elevated CCL17 and CCL22, suggesting these all contributed to T-bet−/− retention in the graft. Since CCR4 may increase Treg suppressive effects in vivo(44, 45), the lack of islet allograft protection in our study suggests that other mechanisms contributed to changes in suppression.
S1P1 regulates the suppressive function and differentiation of Treg(46, 47), so that genetic S1P1 deficiency in the thymus leads to enhanced precursor Treg maturation and suppression, while S1P1 gain of function slows Treg development(48). S1P1 is required for mature T cells to exit the thymus(49). Mature T cells circulate from the blood to the LN, during which S1P1 is down-regulated, and then S1P1 is again up-regulated to allow egress into efferent lymph(50, 51). Down modulation of S1P1 expression or function results in LN retention(51). Thus, S1P1 expression and function are required for T cells to egress the thymus, and LN. We previously demonstrated(26) that direct S1P1 stimulation impaired CD4 Tconv migration from tissues across afferent lymphatic endothelium in both homeostatic and inflammatory conditions. Here we observed Tbet−/− nTreg had decreased S1P1 expression and failed to enter afferent lymphatics, suggesting that S1P1 may regulate nTreg migration into the lymphatic lumen. Since wild type and T-bet−/− iTreg equally expressed S1P1 and migrated similarly to S1P, this suggests that iTreg used and regulated S1P1 in a different fashion compared to nTreg.
CD103 has been implicated in efficient Treg migration to and retention within inflamed sites(52, 53). Islets express high levels of the CD103 ligand E-cadherin(54). Here we showed that T-bet−/− nTreg expressed more CD103, adhered better to E-cadherin than wild type nTreg, and adhesion impaired nTreg migration to CCL19, the major chemokine responsible for afferent lymphatic migration(55). Blockade with soluble E-cadherin-Fc restored T-bet−/− nTreg migration toward CCL19 and homing into the dLN, strongly supporting the role of E-cadherin in tissue retention and deficient lymphatic migration. These observations not only confirmed our previous report that nTreg must migrate to lymphatics and dLN to optimally prolong allograft survival, but also demonstrated that Treg retention within the allograft was not sufficient for fully effective suppression. This conclusion suggests that therapeutic delivery of Treg designed only to migrate to areas of inflammation will be inadequate for sustained and effective clinical results.
The mechanism of why T-bet−/− nTreg islet retention failed to prolong graft survival is not fully solved. Presumably, LN homing is necessary to suppress Tconv in the LN. It is likely that Treg instability further contributed to loss of graft protection, as our data demonstrated that tissue retained T-bet−/− nTreg lost Foxp3 and effector suppressor molecule expression. While nTreg are generally committed to their cell fate, both nTreg and peripheral iTreg have been shown to down-regulate Foxp3 expression and convert to “ex-Treg” effector cells under certain conditions(36, 37, 56, 57). We previously demonstrated that deficiency of chemokine receptors resulted in Treg tissue retention and impaired in vivo suppressive function(21). Since T-bet does not directly regulate Foxp3 expressions (Supplementary Fig. S1A and B), it is likely that the tissue environment of the retained Treg influences Foxp3 and their fate. Future investigations will address if Treg instability in tissues is a generalized phenomenon, if tissue to lymphatic migration is a required process in Treg function, and what factors, such as cytokines or stromal elements, may contribute to reprogramming in the graft.
While epigenetically and developmentally different, nTreg and iTreg are often difficult to distinguish in vivo. Requiring self-antigen specific T-cell receptor and RAG-related cues to develop, nTreg are important in autoimmunity(58, 59). iTreg are generated after encountering foreign antigen in the periphery, and thus impact commensal microbiota and allergic inflammation at mucosal interfaces(60). Primary stimulation of nTreg may occur at inflamed sites, triggering suppression at that site or inducing subsequent homing to the dLN to suppress Tconv. iTreg, on the other hand, may be primarily stimulated in the dLN upon encountering antigen and either suppress Tconv in the LN, or must exit the LN and traffic to the site of inflammation to exert their newly acquired immunosuppressive phenotype(61). Recent studies identified IL27 and IFNγ as important environmental cues that promote both nTreg and iTreg expression of T-bet to control peripheral inflammation(62). In our model, T-bet was linked to several important migration and adhesion receptors and to afferent lymphatic migration, revealing a novel difference between the Treg subsets. Migration and suppressive function for nTreg was thus regulated differently than iTreg, and this difference was manifested in the critical afferent lymphatic step. These distinctions between Treg subsets and between Treg and Tconv Th1-nflammation provide novel loci for therapeutic manipulation of suppression in tolerance and immunity.
We are grateful to Usha Rai for mouse genotyping and technical assistance. Flow cytometry analyses were performed at the University of Maryland Marlene and Stewart Greenebaum Cancer Center Flow Cytometry Shared Service
The authors have declared that no conflict of interest exists.