|Home | About | Journals | Submit | Contact Us | Français|
The importance of regulatory T cells (Treg) for immune tolerance is well recognized, yet the signaling molecules influencing their suppressive activity are relatively poorly understood. Here, through in vivo studies and complementary ex vivo studies, we make several important observations. First, we identify the cytoplasmic tyrosine phosphatase SHP-1 as a novel ‘endogenous brake’ and modifier of the suppressive ability of Treg cells; consistent with this notion, loss of SHP-1 expression strongly augments the ability of Treg cells to suppress inflammation in a mouse model. Second, specific pharmacological inhibition of SHP-1 enzymatic activity via the cancer drug sodium stibogluconate (SSG) potently augmented Treg cell suppressor activity both in vivo and ex vivo. Finally, through a quantitative imaging approach, we directly demonstrate that Treg cells prevent the activation of conventional T cells, and that SHP-1-deficient Treg cells are more efficient suppressors. Collectively, our data reveal SHP-1 as a critical modifier of Treg cell function, and a potential therapeutic target for augmenting Treg cell-mediated suppression in certain disease states.
T regulatory cells (Treg) have been recently shown by a number of laboratories as important players in maintaining immune tolerance. However, the mechanism of Treg cell-mediated suppression is still poorly understood. While there are unanswered questions at many levels about Treg cell-mediated immune suppression [reviewed in (1, 2)], our knowledge is particularly limited with respect to intracellular signaling molecules that regulate/modify Treg cell function. Previously, several molecules essential for Treg cell development and/or function such as STAT5 (3), Foxp3 (4, 5), or CTLA-4 (6, 7) have been identified; while other molecules have been linked to some of the proposed mechanisms of Treg cell-mediated suppression either in vitro or in vivo, including cAMP (8), Galectin-1 (9), CD39/CD73 (10), LAG-3 (11), IL-35 (12), IL-10, and TGFβ [reviewed in (1)]. However, our knowledge on intracellular pathways that can modulate the suppressive activity of Treg cells is rather limited. To date, S1P1 is the only molecule that has been identified as being important for both the development and regulatory function of Treg cells via activation of the Akt-mTOR pathway (13).
SHP-1 is a non-transmembrane protein tyrosine phosphatase expressed in hematopoietic cells of all lineages including Tcon cells and Treg cells. SHP-1 is now widely accepted as an important negative regulator of T cell receptor (TCR)-mediated signaling in Tcon cells [reviewed in (14, 15)]. Since Treg cells have been shown to require stimulation via the TCR for their full function (16-19), we asked whether SHP-1 might regulate Treg cell activity through its effects on TCR mediated signaling. The existence of a murine genetic model for SHP-1 deficiency has significantly aided our understanding of the biological function of SHP-1. A splicing mutation within the Ptpn6 (Shp1) locus leads to no detectable SHP-1 protein, and causes the motheaten (me/me) mouse phenotype (20, 21). We have previously shown that me/me mice have increased numbers of naturally occurring Treg cells, and that these Treg cells can mediate suppression of Tcon cell responses (22). However, whether SHP-1 also influences the ‘suppressive potential’ of Treg cells (i.e. magnitude of suppression), and how loss of SHP-1 expression might affect Treg cell function are not known.
In this study, we have attempted to address how stimulation via the TCR affects Treg cell function using a combination of complementary genetic and pharmacological approaches, specifically targeting the function of the tyrosine phosphatase SHP-1. Our data presented here suggest that the strength of TCR-initiated signaling within the Treg cells directly affects their level of suppressive activity, and that SHP-1 functions downstream of the TCR in Treg cells and thereby directly modulate their suppressive potential. Our data using sodium stibogluconate (SSG), a specific inhibitor targeting SHP-1 activity (23-25), further support an important role for SHP-1 as modifier of the strength of suppression. Interestingly, SSG has been previously approved for treatment of leishmaniasis (26), and is currently tested in three phase 1 clinical trials for patients with advanced solid tumors, lymphoma, or myeloma (27-29). We also addressed mechanistically how Treg cells from motheaten mice are more capable of suppression. Using a quantitative single cell-based imaging approach; we show that Treg cells can suppress the activation of conventional T cells (Tcon) via at least two levels, both of which are regulated by SHP-1. During these studies, we also clarify a mechanism by which Treg cells by being part of the same complex with Tcon and APC can directly suppress the activation of Tcon cells. Thus, the data presented in this work linking SHP-1 to the strength of Treg cell-mediated immune suppression may also have clinical relevance in providing a therapeutic target to enhance Treg cell function in certain disease states.
Mice used for this study were bred in our colony. All mice are on the BALB/c background. TCR-transgenic (Tg) +/+: DO11.10 and me/+: DO11.10 and non-TCR-Tg me/+ were used to generate +/+, me/+ and me/me mice on TCR-Tg and non-Tg backgrounds (30). TS1-HA TCR-Tg mice (31) were generously provided by Dr. Kenneth Tung (University of Virginia). Genotyping of all mice was done by PCR as described previously for the me allele and DO11.10 TCR (30), and for TS1-HA TCR (32). Unless mentioned otherwise, 17-19 day-old mice were used for this study. All mice were bred and maintained in accordance with the policies of the Institutional Animal Care and Use Committee (IACUC) at the University of Virginia. All experiments involving mice were conducted with the approval of IACUC.
CD4+CD25- and CD4+CD25+ T cells were isolated from lymph nodes (combined inguinal, axillary, brachial, cervical, lumbar, sacral, renal and pancreatic nodes unless otherwise indicated) using the Regulatory T cell Isolation kit (Miltenyi Biotec; Auburn, CA) according to the manufacturer's protocol.
Bone marrow derived dendritic cells (BMDC) were generated as previously described (33). Briefly, bone marrow cells isolated from mouse femurs and tibias were plated at 3×106 cells/well in 6-well plates and cultured for 5 or 6 days in RPMI medium (with 10% FCS, 5 × 10−5 M 2-ME, 2 mM L-glutamine, and antibiotics) supplemented with GM-CSF (1000U/ml) and IL-4 (100U/ml) (PeproTech, Rocky Hill, NJ) followed by positive selection using CD11c-microbeads (N418) on a MACS column (Miltenyi Biotec; Auburn, CA). Obtained BMDC were > 95% positive for CD11c as assessed by flow cytometric analysis.
Cells were stained with antibodies recognizing the indicated surface markers in PBS supplemented with 1% BSA and 0.1% sodium azide. Intracellular staining was done following fixation and permeabilization of the cells using the Fixation/Permeabilization kit from eBioscience (San Diego, CA). CD4-PE/perCP, CD25-APC/PE, CD62L-PE, CD38-PE, CTLA-4-PE, CD103-PE/FITC, ICAM-1-PE/FITC, LAG-3-PE, CD62P-FITC, CD3-perCP, CD279 (PD-1)-PE, CD45-RB-PE, IgG-1-PE, IgG-APC, CD95-PE.Cy7, CD127 (IL-7Rα)-PE, CD102 (ICAM-2)-FITC, CD184 (CXCR4)-FITC, CD28-PE, FR4-PE, ICOS-PE and CD122 (IL-2Rβ)-PE were purchased from BD Pharmingen (San Jose, CA). CD29-PE.Cy7, CD49c (integrin α3)-FITC, CD49d (integrin α4)-FITC, CD49e (integrin α5)-PE, CD44-Alexa 488, CD40L-PE, LFA-1-Alexa 488, GITR-Alexa 488 and IgG-Alexa 488 were purchased from BioLegend (San Diego, CA). TLR4-PE, CD80-PE, CD86-PE, 4-1BB-PE, CD69-PE.Cy5, CD4-PE.Cy7, IFN-g-PE, IgG-FITC, CD11c-FITC/APC, OX-40-PE, CCR7-APC and FoxP3-PE/FITC antibody and staining kit were from eBioscience (San Diego, CA). KJ1-26-PE and KJ1-26-FITC were purchased from Caltag (Burlingame, CA). CD104-PE, CD-2(LFA-2)-FITC, CD27 (TNFR)-FITC and CD49f (integrin α6)-FITC were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Stained cells were collected on a FACS Calibur instrument using CellQuest software (BD Biosciences, San Jose, CA) and subsequent analyses were done using FlowJo software (Tree Star Inc., Ashland, OR). Analyses were conducted on live cells (>95%) as defined by forward- and side-angle scatter. Gates were set using isotype-matched control antibodies.
Proliferation and suppression assays were performed as described previously (22). Briefly, to assess proliferation, 2.5 × 104 CD4+CD25- T cells were plated in triplicates in 200 μl of RPMI-1640 medium (supplemented with 10% FCS, 5 × 10−5 M 2-ME, 2 mM L-glutamine, and antibiotics) in round bottom 96-well plates. To measure suppression, CD4+CD25+ T cells were added at indicated ratios to the CD4+CD25- T cells. Irradiated (2000 rad) total (after RBC lysis) or T cell-depleted splenocytes were added at 5 × 104 cells per well together with either anti-CD3 Ab (145-2c11; Cedarlane Laboratories; Burlington, NC) at 6 μg/ml or OVA 323-339 peptide (ISQAVHAAHAEINEAGR) at 125ng/ml unless otherwise indicated. TS1-HA-TCR-Tg+ cells were stimulated using influenza hemagglutinin 110-119 peptide (SFERFEIFPK) at 500ng/ml. Peptides were synthesized at the Biomolecular Research Facility, University of Virginia. Cells were cultured for 72 h before they were pulsed with 1 μCi of [3H] Thymidine for 18 h. [3H] Thymidine incorporation was measured using a Tomec cell harvester and Betaplate counter (PerkinElmer, Waltham, MA).
For measuring proliferation and suppression CD4+CD25-T cells were stained with CFSE (Carboxyfluorescein succinimidyl ester) (CellTrace™ CFSE cell proliferation kit, Molecular Probes) at a final concentration of 10μM for 15 min at 37°C (34). The cells were washed and 2.5 × 104 cells were plated onto 96 well plates either alone or with different ratios of CD4+CD25+ regulatory T cells. Irradiated (2000 rad) T-cell depleted splenocytes (5 × 104 cells per well) or day 5- BMDCs (5 × 103 cells per well) were added along with OVA peptide at 125ng/ml. Cells were cultured for 4 days followed by flow cytometric analyses.
SHP-1 protein levels were assessed as described previously (35). Briefly, SHP-1 was immunoprecipitated from lysates of each 2.5 × 105 purified CD4+CD25+ regulatory T cells using 2 μg of rabbit anti-SHP-1 Abs (Santa Cruz Biotechnology, Santa Cruz, CA) followed by immunoblotting for SHP-1 with monoclonal anti-SHP-1 (clone 1SH01; NeoMarkers; Fremont, CA) at 1.35 μg/ml. Relative levels of SHP-1 were calculated based on densitometry measurements of immunoblots at linear range using the ImageQuant TL 2005 program.
Th1 cells were generated and adoptively transferred as previously described (36) with the following modifications. CD4+CD25- T cells from BALB/c: +/+: DO11.10 mice (1× 106cells/ml) were cultured with T-cell depleted irradiated splenocytes (2 × 106cells/ml), 1μg/ml of OVA peptide, 10μg/ml of anti-mouse IL-4 neutralization antibody (R&D systems, Minneapolis, MN) and 10ng/ml of recombinant mIL-12 (R&D systems, Minneapolis, MN) in 24-well plates. The cells were supplemented with fresh media on day 3 and day 5 and were harvested on day 7. 75-80% of the cells, which were adoptively transferred, expressed IFN-γ as assessed by flow cytometric analysis.
CD4+CD25+ cells were isolated from BALB/c: DO11: +/+ and me/+ mice as described above. For adoptive transfer, purified cells were resuspended in sterile PBS (4 × 105 cells in a max. volume of 25μl).
BALB/c mice (2-3 months old) were divided into three experimental groups as indicated in the table below: (I) Th1 only, (II) Th1 plus +/+ Treg cells and (III) Th1 plus me/+ Treg cells. For group I, equal numbers of Th1 cells and APCs (1.6 × 106) were mixed and injected subcutaneously in the left footpad along with 10μg OVA peptide in a volume of 25μl. For the two groups (II and III) that received Treg cells, mice were injected with either +/+ or me/+ Treg cells (at a Th1:Treg ratio of 4:1) along with Th1 cells, APCs and OVA peptide in the left footpad in a volume of 25μl. For all mice, the right footpad received the same number of Th1 cells and APCs without any peptide and served as the control. Footpad measurements were taken before and 24hrs after the injections, using a micrometer (Mitutoyo, Japan). All footpad measurements were taken in a manner blinded to the experimental conditions. Footpad swelling is expressed as the percentage increase of the left footpad over the control right footpad.
|Group||Left footpad||Right footpad|
|I - Th1 only||Th1+ APC+ peptide||Th1+APC|
|II - Th1-Treg (+/+)||Th1+ APC + peptide + Treg (+/+)||Th1+APC|
|III - Th1- Treg (me/+)||Th1+ APC + peptide + Treg (me/+)||Th1+APC|
Cells were purified and plated into the wells of a 96-well plate as described above for the standard suppression assay using CFSE dilution as readout for proliferation. SSG [10μg/ml] (Calbiochem, SanDiego, CA) was added to the culture. 10μg/ml SSG have been determined to inhibit 99% of SHP-1 activity, while minimally affecting the phosphatase activity of SHP-2 or PTP1B (23).
+/+ DO11.10 mice (2 month old) were injected with SSG subcutaneously at a concentration of 10mg/mouse on days 1 and 5. Control mice of the same age were treated with sterile PBS. On day 7, Treg cells were isolated from the lymph nodes of these mice and used for proliferation/suppression assays.
Conjugation assays were performed as described previously (37) with the following modifications. CD4+CD25- and CD4+CD25+ T cells were isolated, resuspended in PBS to a concentration of 2 × 107 cells/ml and labeled with an equal volume of succinimidyl esters Alexa 488 (diluted 1:100) or Alexa 633 (diluted 1:50) (Molecular Probes, Invitrogen, Carlsbad, CA) for 15 min at room temperature. The cells were washed three times and resuspended in complete RPMI media. Bone marrow-derived dendritic cells, which had been pulsed with the indicated concentrations of OVA peptide for 1 h, were used as APCs and added to T cells (CD4+CD25- or CD4+CD25+ or a combination of both) in a final volume of 200μl at a T cell: APC ratio of 1:1, 1:2 or 1:4 in a 96-well round bottom plate along with respective concentrations of OVA peptide. Settling of the cells were initiated by spinning the plate at 500rpm for 30 sec followed by incubation at 37°C for indicated time periods indicated. To analyze cell conjugates, cell mixtures were vortexed to disrupt the non-specific conjugates, washed with PBS (1% BSA) and fixed with 4% paraformaldehyde for 15 min at room temperature. Cells were then stained with anti-CD4 and anti-CD11c for additional identification of CD4 T cells and CD11c+ BMDCs and collected on a FACS Calibur and analyzed using FlowJo software. For analysis of conjugates on an ImageStream100 instrument (Amnis Corporation; Seattle, WA), cells were pre-stained (1:100 dilution) with succinimidyl esters as follows: Tcon- Alexa 647, Treg- Alexa 488 and BMDC-Alexa 405 followed by incubation for 4 h at a ratio of 1:1:1. Samples were collected on an ImageStream100 instrument using the EDF (Extended Depth of Field Technology) program and analyzed by IDEAS analytical software.
To detect the upregulation of CD25 by Tcon cells when conjugated to BMDCs, experiments were set up as described above for conjugate assays. Prestained cells were incubated for 8 h, stained for the surface expression of CD25 and acquired using the EDF program on an ImageStream100 instrument. The data were analyzed using the IDEAS analytical software. To quantitate CD25 expression on the surface of Tcon cells in a triple conjugate with Treg cells and BMDCs, CD25 expression by Treg cells was masked using the masking function provided by the IDEAS applications. CD25 expression of Tcon cells was quantitated using the similarity feature, which is the log transformed Pearson's Correlation Coefficient and a measure of the degree to which two images are linearly correlated within a masked region (per manufacturers instructions). Similarity is expressed as the median. For the detection of intracellular IL-2 production by Tcon cells, a conjugation assay was performed, and cells were fixed and permeabilized with the reagents from the Fixation-permeabilization kit from BD Biosciences following the manufacturer's protocol. The cells were not treated with any protein transport inhibitor. Anti-IL-2-PE antibody from eBioscience was used for staining.
p values were calculated with unpaired Student's t-test. p values of less than 0.05 were considered significant.
To study the role of SHP-1 in Treg cell function, we used a murine model for SHP-1 deficiency, the so-called motheaten mouse, that in previous studies has significantly aided the understanding of the biological function of SHP-1. We had recently shown that motheaten mice have an increased percentage of functionally active CD4+CD25+ Treg cells (22); however, whether and if so, how the loss of SHP-1 influences Treg suppressive activity is not known. To specifically assess antigen/peptide specific signaling/activation downstream of the TCR expressed on Treg cells, mice carrying the motheaten allele were crossed with mice transgenic for the DO11.10 TCR that recognizes the OVA323-339 peptide (30). Although there are multiple subsets of T cells exhibiting suppressive activity, in this study we focused on CD4+CD25+Foxp3+ Treg cells. Treg cells purified from control and mutant SHP-1 background were assessed for in vitro suppression using varying ratios of Treg vs. Tcon cells. Treg cells from homozygous me/me and heterozygous me/+ mice were more suppressive than control +/+ littermates in response to OVA peptide stimulation (Fig. 1A). Monitoring the proliferative response of Tcon cells (by CFSE labeling and assessing the dilution of the CFSE signal using flow cytometry) confirmed that the Treg cell effect seen is specifically due to decreased proliferation of Tcon cells under these assay conditions (Fig. 1B). In addition to the OVA peptide-induced responses, the increased suppressive potential of Treg cells lacking SHP-1 was also observed after more generic anti-CD3 mediated TCR stimulation (Fig. 1C). Notably, SHP-1-deficient Treg cells showed a strong suppressive activity even at a Treg:Tcon ratio of 1:32, while control Treg cells had no suppressive activity at this ratio (Figs. 1 A, B, and C). It should also be noted that the purified Treg cell populations from me/+, and me/me cells displayed almost identical profiles for CD4, CD25, Foxp3, CD3 and the transgenic TCR DO11.10 compared to Treg cells from +/+ mice (Fig. 1D), ruling out another not previously identified population of Treg cells as being responsible for the observed effects.
Notably, Treg cells with a heterozygous me/+ genotype displayed an intermediate phenotype, suggesting that ~50% reduction in SHP-1 protein can still have a detectable effect on the suppressive activity of Treg cells (Fig. 1D). The me/me mice have a short life span and die at about 3 weeks of age, requiring that the above set of experiments be performed using mice <20 days old. To rule out the possibility that the observed phenotype might have arisen due to the young mice used, we repeated the suppression assays comparing +/+ and me/+ Treg cells from adult mice. These assays using me/+ Treg cells confirmed that the increased suppressive activity of Treg cells from the motheaten background is independent of their age (Supplementary Fig. S1).
To better understand the mechanism, by which SHP-1 may influence Treg cell activity, we asked whether SHP-1 affects the strength of TCR-mediated signaling. We designed suppression assays such that the Treg cells can be stimulated with increasing concentrations of antigenic stimulation, while the Tcon cells are stimulated at a constant dose of antigen peptide. This was achieved through the use of two different TCR transgenic mice as sources for Tcon and Treg cells, respectively; HA-specific TCR expressing Tcon cells were used as responders, while the OVA-specific TCR expressing Treg cells were used as suppressors. Both TCRs recognize their respective peptides presented by class II MHC H-2d, allowing peptide presentation to both Treg and Tcon cells by the same APCs. These ‘mixed suppression’ assays revealed several important observations (Fig. 2). (i) The Treg cell-mediated suppression was absolutely dependent on TCR-dependent activation of Treg cells, since no suppression was detectable in the absence of OVA peptide that triggers the Treg cells. Moreover, Treg cells were able to suppress proliferation of Tcon cells even if the Tcon and Treg cells recognize different MHC-peptide complexes [as it has been reported previously (16)]. (ii) me/me Treg cells were more potent suppressors than +/+ Treg cells under these ‘mixed suppression’ conditions. (iii) Strikingly, upon increased peptide concentrations both +/+ and me/+ Treg cells showed enhanced suppressive activity indicating a direct correlation between that the strength of the TCR-mediated signaling and the suppressive activity of Treg cells. In contrast, me/me Treg cells lacking SHP-1 were insensitive to increased concentrations of OVA peptide; this indicated that SHP-1 is a critical, non-redundant, and perhaps the most significant ‘brake’ on TCR signal strength in Treg cells such that in the absence of SHP-1 Treg cell- mediated suppression reaches a maximum even at lower peptide concentrations. Lastly (iv), even though increased TCR stimulation on Treg cells (via increasing OVA peptide concentrations) caused a more potent suppression by +/+ Treg cells, their suppressive activities never reached the level observed by me/me Treg cells, suggesting that the SHP-1 mediated regulation of suppression may involve events beyond the recognition of peptide+MHC by TCR on Treg cells (addressed further below).
To assess whether the increased suppressive activity of SHP-1-deficient Treg cells seen ex vivo would be relevant under in vivo conditions, we tested Treg cell-mediated suppression in a mouse model of delayed type hypersensitivity (DTH) (36). Limiting numbers of Treg cells were injected and the efficiency of suppression was assessed. Under conditions when +/+ Treg cells failed to show a statistically significant suppression, the me/+ Treg cells were able to potently suppress OVA-induced footpad swelling (Fig. 3). Thus, SHP-1 deficiency causes Treg cells to become more potent suppressor cells in the context of a whole animal. Collectively, these data identified SHP-1 as a key regulator of Treg cell suppressive activity in vitro and in vivo.
As a complementary approach, and to directly correlate the phenotype of the Treg cells from me/me mice to the enzymatic activity of SHP-1, we used SSG, a pharmacological inhibitor of SHP-1. The drug SSG has previously been shown to specifically inhibit 99% of SHP-1 activity at low concentrations (10 μg/ml), while the closely related phosphatase SHP-2 was fully active under these conditions (23). We next asked how SSG affected Treg cell activity. To focus on the effect of SSG on Treg cell activity, suppression assays were performed using peptide concentrations that provided optimal proliferative stimulation for Tcon cells. It is important to note that SSG treatment alone did not affect Tcon cell proliferation under these conditions (Fig. 4A, top panel - noTreg). This allowed us to directly address the effect of SSG on Treg cell-mediated suppression and to make the following key observations. First, SSG treatment of Treg cells from +/+ cells potently increased their suppressive activity. (Fig. 4A, left panel). Second, in the context of Treg cells from me/+ mice, SSG further enhanced the suppressive activity, likely due to inhibition of the residual SHP-1 activity in the heterozygous mice (Fig. 4A, middle panel). Third, SSG had no effect on the already enhanced suppressive activity of Treg cells from me/me mice (i.e. no effect in the absence of SHP-1 expression), which further supports the proposed specificity of SSG (Fig. 4A, right panel). Moreover, the short term SSG treatment of wild type Treg cells leading to the enhanced suppressive potential, essentially phenocopying the results with Treg cells from me/me mice, helped rule out any secondary effects due to the loss of SHP-1 in other tissues within the motheaten mice (i.e. cell autonomous). Collectively, these data demonstrate that the enzymatic activity of SHP-1 normally functions as ‘brake’ to control the suppressive potential of Treg cells, and that altering the function of SHP-1 could be a tool to modify the functional efficiency of Treg cells.
Two additional observations supported and extended our conclusion that SSG enhances the suppressive activity of Treg cells in vitro and in vivo. First, SSG due to its ability to inhibit SHP-1 has been shown to enhance Tcon effector function (24, 38). When we tested the effect of SSG under suboptimal peptide concentrations where SSG augments Tcon cell proliferation (Fig. 4B, compare rows 1 and 3), addition of Treg cells suppresses Tcon cell proliferation to the same extent or even further than in the absence of SSG (Fig. 4B, compare rows 2 and 4). Thus, SSG-mediated enhancement of Treg cell suppressive activity appears more ‘dominant’ than the SSG effect on Tcon cell proliferation. Secondly, when mice were injected with SSG, we found that the suppressive activity of Treg cells from SSG-treated mice was enhanced compared to the activity of PBS-injected control mice (Fig. 4C). Collectively, these observations suggested that SSG enhances Treg cell suppressive activity in vivo, and that the effect of SSG on Treg cells is greater than the effect on Tcon cells.
To mechanistically understand how SHP-1 could influence the suppressive activity of Treg cells, we compared the gene expression and protein profile of Treg cells from control and SHP-1 deficient mice. Using microarray analyses, the mRNA profiles of three genetically different Treg populations (+/+, me/+ and me/me) were compared. Based on our observation that me/+ Treg cells display an intermediate phenotype in the functional assays, a special emphasis was also placed on detecting mRNAs that show a consistent change from +/+ to me/+ and me/me. However, no obvious changes in the gene expression profile were identified in our microarray. Although small differences between the +/+, me/+ and me/me Treg cell gene expression profiles were noted, no statistically significant differences were observable among Treg signature genes (Supplementary Fig. S2).
We next explored the possibility that protein expression may vary between the different genotypes, and focused on proteins associated with Treg cell activity and/or function. Many surface molecules previously linked to Treg cell development/function were unchanged (including Foxp3, CD25, CD122, LAG-3, ICAM-2, CD28, CD86, CD69, TLR-4, CD95, CD38, CCR7, CXCR4, CD49c, CD49d, CD49e, CD49f, CD31) or even down-modulated (CD62L, FR4, CD40L). However, a few surface molecules showed notable differences in expression between +/+, me/+, and me/me Treg cells (Fig. 5A and Supplementary Fig. S3). The difference was manifested at two levels. First, at steady state, a higher fraction of me/me Treg cells expressed the following markers - CD103, 4-1BB, icCTLA-4, CD80, CD44, CD27, CD62P, LFA-1, ICOS, CD29; however, the expression level of the individual markers on a per cell basis was not changed. Second, Treg cells from SHP-1-deficient mice displayed an increase in surface expression (on a per cell basis) of specific molecules, such as ICAM-1, IL-7Rα, CD104, OX40, GITR, LFA-2. This difference in surface profile of Treg cells in the me background was not due to the source of the lymph node from which the cells were derived, since Treg cells from inguinal, axillary, brachial, cervical, lumbar, sacral, renal and pancreatic nodes as well as spleen were comparable (Supplementary Figs. S4A and B). Moreover, the difference seen in Treg cells in the me background was independent of whether the Treg cells were derived from non-TCR-Tg or the DO11.10 TCR-transgenic mice (Supplementary Fig. S4C). Taken together, these data showed that SHP-1 expression (or the lack thereof) affects the expression of a subset of proteins on Treg cells.
One of the main differences between +/+, me/+, and me/me Treg cells is the fraction of cells positive for CD103 (αE-integrin), and CD103 is thought to mark the in vivo activated effector/memory subpopulation of CD4+Foxp3+ Treg cells (39, 40). When we gated on this CD103+ population, many other markers linked to Treg-mediated suppression were also altered, including ICAM-1, CD80, 4-1BB (CD137), LAG-3 (CD223) and CD62L (Supplementary Fig. S5A). To correlate the altered marker expression to Treg function, we took advantage of the variability we see in the suppressive ability of Treg cells from me/+ mice. While heterozygous me/+ Treg cells consistently show an intermediate phenotype between +/+ and me/me mice, there are significant variations between individual me/+ mice with some more closely resembling the motheaten phenotype, while others are closer to a wild type phenotype. Interestingly, when we compared four randomly chosen me/+ littermates (denoted mouse A, B, C and D in Fig. 5B) for CD103 and ICAM-1 expression, we observed a close correlation between suppressive potential and the fraction of Treg cells that were CD103 positive (Fig. 5B). When we tested the hypothesis that the increase in the fraction of CD103 positive cells in the me background might be due to the heightened TCR response in the absence of SHP-1, this was not found to be the case (Supplementary Fig. S5B). This further supports our previous findings (Fig. 2) that in Treg cells, SHP-1 affects additional signaling pathways besides TCR-mediated signaling.
The phenotypic and functional differences between me/me, me/+, and +/+ Treg cells indicate that SHP-1-deficient Treg cells are in a state of heightened activation. One of the functional requirements for successful suppression is the formation of a conjugate between the Treg cells and the antigen presenting cells (APCs), and many of the surface molecules up-regulated on me/+ and me/me Treg cells are adhesion molecules. This prompted us to test whether me/+ and me/me Treg cells might form conjugates with APCs more efficiently. To have a more homogenous antigen presenting population for these assays, we used primary bone marrow derived dendritic cells (BMDCs) from control mice (reviewed in (41)). We confirmed that differences in suppressive activities between me/+ and +/+ Treg cells can still be observed using BMDCs as APCs (data not shown). Using a flow cytometric approach, we examined the conjugate formation between Treg cells and BMDCs (Fig. 6A). Even in the absence of cognate peptide, 7-9% of the Treg cells formed conjugates with BMDCs, with me/me-derived Treg cells forming slightly more conjugates. Upon addition of peptide, an increased 16% of Treg cells from control mice formed conjugates, while Treg cells from me/me mice formed many more conjugates (25%). To better define the composition of the conjugates, such as the number of cells that form the conjugates, we used the Imagestream100 (AMNIS) instrument, which combines flow cytometry with fluorescence microscopy; this allowed imaging large numbers of individual conjugates while the cells are being analyzed by traditional parameters used in flow cytometry. We scored the individual BMDC:Treg conjugates as BMDC conjugated to 1, 2, or ≥3 Treg cells (Fig. 6B). While the overall fraction of Treg cells found in conjugate with APCs increases upon addition of peptide, the me/+ and me/me Treg cells consistently formed more conjugates than +/+ Treg cells (Fig. 6C and Supplementary Fig. S6). Taken together, these data support our hypothesis that SHP-1-deficient Treg cells are more efficient in forming conjugates with antigen presenting cells, which may contribute to their increased suppressive activity.
We next asked whether the increased conjugate formation due to SHP-1 deficiency correlates with a Treg cell-mediated inhibitory effect on the APCs. Treg cells have been shown to decrease the surface expression of the co-receptors CD80 and CD86 (18, 42) on APCs, and this is considered one of the mechanisms by which Treg cells mediate their suppressive function. We noted that day 5 BMDCs have very low levels of surface CD80/CD86, while their expression is clearly detectable at day 6. The addition of Treg cells for 24 h caused a down-modulation of CD80 surface expression (Figs. 7 A and B). Incubation with me/+ or me/me Treg cells resulted in an additional decrease of CD80 expression, which was small but consistent, when compared to BMDCs incubated with +/+ Treg cells. This effect of Treg cells requires antigen-mediated interaction with BMDCs, as there was no effect in the absence of OVA peptide. In comparison, conventional T cells showed no inhibitory effect on CD80, but rather an increase in CD80 expression (Figs. 7 A and B).
Since under physiological conditions, APCs can form conjugates containing both Tcon and Treg cells, we next asked how CD80 expression is influenced when both Treg and Tcon cells are added to BMDCs. While Tcon cells cause an increase in CD80 expression, addition of Treg cells still decreases CD80 expression (Fig. 7C). Under these conditions, Treg cells from me/+ and me/me mice were again slightly more effective in inhibiting CD80 expression. In fact, at high peptide concentrations (1μg) where Treg cells have been shown to be less efficient (11), Treg cells from the motheaten background were still capable of substantially decreasing CD80 expression (while the +/+ Treg cells showed a diminished effect) (Fig. 7C). These data further suggest that Treg cells with loss of SHP-1 expression are more effective suppressor cells and are capable of actively down-modulating CD80 expression to levels lower than initially expressed on the APCs.
We then tested whether Treg cells would affect the conjugate formation between Tcon cells and APCs, and whether Treg cells with or without SHP-1 differed in regulating Tcon/APC interaction and/or Tcon activation (Fig. 8A). Addition of Treg cells decreased the number of conjugates formed between Tcon and APCs, with Treg cells from me/me mice being slightly more efficient (Fig. 8B). We next assessed whether the decrease in Tcon/BMDC conjugate formation was due to a competition for binding on the BMDC or due to other mechanisms, such as the change in surface molecule expression on BMDCs induced by Treg cells (Fig. 7). To test the competition model, we added excess Tcon cells (with a second fluorescent label) to assess whether it would alter the Tcon/BMDC conjugate formation. Addition of extra Tcon cells caused only a minor decrease in the Tcon/BMDC conjugate formation; under the same conditions, addition of Treg cells caused a significant decrease in Tcon:BMDC conjugates (Fig. 8B). This suggested that inhibition likely occurred due to direct effects of Treg cells on BMDCs, which in turn result in lower conjugate formation between Tcon and BMDCs; while under our experimental conditions (BMDC:Tcon:Treg ratio – 1:1:1) direct competition for BMDC binding between Treg and Tcon cells contributes only minimally to the inhibitory effects.
Intriguingly, we still observed many Tcon:BMDC conjugates under conditions where we do not observe robust proliferation of Tcon cells. When we separated the Tcon/BMDC conjugates into Treg-containing (Tcon/BMDC/Treg) and no-Treg containing conjugates (Tcon/BMDC), we uncovered that a substantial percentage of Tcon:BMDC conjugates also contained Treg cells (Fig. 8C and Supplementary Fig. S8). This tri-partite conjugate was more pronounced with me/+ or me/me Treg cells (e.g. 52% at 100ng OVA peptide) compared to +/+ Treg cells (30%).
Since ‘activated’ Tcon cells up-regulate CD25 and produce IL-2 in an antigen-dependent manner (Figs. 9A and B bottom panels and Supplementary Figs. S7A and B), we directly addressed how the presence of Treg cells would affect these parameters. Tcon cells that are part of the Tcon/BMDC/Treg conjugate failed to up-regulate CD25 or produce IL-2 (Figs. 9A and B). Interestingly in the same culture, Tcon cells in the bi-partite Tcon:BMDC conjugates (without Treg cells) were able to efficiently up-regulate CD25 (Fig. 9A). Previous studies have suggested that Treg cells can mediate bystander suppression wherein the presence of Treg cells in the same culture with Tcon and APC, even when they are not part of the same complex, is sufficient to induce suppression (reviewed in (1, 2)). However, the level and mechanism of such suppression remains poorly understood. Our system of being able to simultaneously quantitate Tcon cell responses at the single cell level in the context of conjugates with or without Treg cells, allowed us to visualize bystander suppression. Remarkably, we observe that IL-2 production by Tcon cells is partially inhibited by the simple presence of Treg cells in the culture, even when the Tcon:BMDC conjugates do not contain Treg cells (Fig. 9B). We were concerned that this inhibition of IL-2 production in the Treg-lacking Tcon:BMDC conjugates represent those that initially were tri-partite complexes, from which the Treg have dissociated during the experimental procedures; however based on time course experiments this was not found to be the case (data not shown). Moreover, full suppression of CD25 up-regulation on Tcon cells did not occur via this ‘bystander’ mechanism, but was only on those Tcon cells that were part of the Tcon:BMDC:Treg tri-partite conjugate. This suggests that different parameters of Tcon activation are differentially suppressed, perhaps indicating that Treg cells may use multiple mechanisms of suppression.
While there is some inhibition of Tcon activation even in the exclusive Tcon:BMDCs conjugates, these conjugates are also the only ones in these cultures that are capable of any activation. This prompted us to focus on the number of bi-partite Tcon:BMDC conjugates, and in turn ask how this would be influenced by Treg cells from control and SHP-1-deficient mice. Calculating the conjugates using IDEAS™ analytical software (Supplementary Fig. S8), we made two key observations: first, addition of +/+ Treg cells decreases the percentage of exclusive Tcon:BMDC conjugates by 50-65% (Fig. 9C); second, Treg cells from the motheaten background are more efficient and decrease the percentage of Tcon:BMDC conjugates by 75-80% (me/me) (Fig. 9D). Taken together, in the context of SHP-1 deficiency, the overall decrease in Tcon/BMDC conjugates with a concurrent increased fraction of Treg:BMDC:Tcon conjugates (which are non-productive for Tcon cell activation) results in fewer exclusive Tcon:BMDCs conjugates and translates to an enhanced suppressive activity of Treg cells from me/me mice readily seen ex vivo and in vivo. These studies identify SHP-1 as an essential and non-redundant intracellular signaling molecule that modulates the potency of Treg cell-mediated suppression.
Despite the significant knowledge acquired about Treg cells over the past few years, very little is known about signaling events within Treg cells that can alter the potency of their suppressive activity. In this report, we identify the intracellular signaling molecule SHP-1 as a non-redundant and cell autonomous modulator of Treg function both in vitro and in vivo. Using mice either homozygous or heterozygous for the motheaten allele, we found that SHP-1 is an endogenous ‘brake’ for the potency of Treg mediated suppression, and that loss of SHP-1 expression manifests as increased suppressive activity. Furthermore the SHP-1-specific pharmacological inhibitor SSG complemented the data from motheaten mice, and demonstrated that SHP-1 modulates Treg function in a cell-autonomous manner. It is interesting to note that SSG via its inhibitory effect on SHP-1 causes an activation of Treg function. SSG is a drug originally marketed to treat Leishmania infections (26). Interestingly, Treg cells have been shown to dampen the immune response thereby allowing low level persistence of the Leishmania parasite in the host, which confers long-term immunity and resistance to re-infection (43). In particular, CD103+ Treg cells, which we found to be enriched in SHP-1-deficient mice, have been shown to be the subpopulation accumulating at the site of Leishmania infection (44). In addition, more recent studies have proposed an anti-tumor effect for SSG via a T cell dependent mechanism (24, 25). In fact, SSG is currently tested in several phase 1 clinical trials for patients with advanced solid tumors, lymphoma, or myeloma (27-29). Since enhanced Treg suppressor activity could interfere with anti-tumor immunity, our findings that SSG can potentiate Treg suppressor activity suggests that these treatments should also be evaluated for concurrent effects on Treg cells in the patients.
Comparable expression profiles between SHP-1-sufficient and SHP-1-deficient Treg cells obtained from the microarray data further confirmed that SHP-1 affects Treg function at the level of signaling, and likely not at the developmental level. Interestingly, despite the lack of differences at the mRNA level (Supplementary Fig. S2), we detected significant changes in the surface expression profiles of several molecules associated with Treg function between +/+, me/+, and me/me Treg cells indicating post-transcriptional regulatory mechanisms. This finding emphasizes the need to complement microarray analyses with studies assessing protein levels. At this point, it is unknown at what level the expression of the individual proteins is regulated, such as whether protein synthesis or degradation is affected by SHP-1 deficiency, whether this is directly linked to SHP-1, or whether it is a secondary effect reflecting a difference in activation status. While earlier studies had supported the notion that TCR-mediated activation is absolutely required for Treg cells to mediate their suppression (16-18), a recent study by Szymczak-Workman et al. (45) suggested that Treg cells may not require activation by the cognate peptide, although activation increased the suppressive activity. In our study, we found that antigen-mediated activation is required for suppression (Fig. 2). However, we used a very low Treg:Tcon cell ratio of 1:16. At this ratio, Szymczak-Workman et al. also did not detect any suppression in the absence of TCR-mediated activation. Interestingly, we found that me/me Treg cells demonstrated some suppression even in the absence of stimulation. Similarly, our conjugate assays showed a basal conjugate formation between Treg cells and APCs even in the absence of peptide (Fig. 6), which was not observed when Tcon cells were conjugated with APCs (data not shown). It is therefore possible that a subpopulation (5-10% of total) of freshly isolated Treg cells is already activated, either in the animal or during the isolation procedure, and can confer suppression without further stimulation. Consistent with the data presented by Szymczak-Workman et al., this phenotype will be more obvious at higher ratios of Treg cells. Our studies extend previous findings by demonstrating that the strength of signaling downstream of the TCR directly affects the suppressive activity of the Treg cells. Moreover, our data indicate that at least one mechanism how SHP-1 modulates Treg cell activity is by targeting signaling pathways downstream of the TCR.
Although the mRNA expression profiles are similar between +/+, me/+ and me/me Treg cells, there are substantial differences at the protein expression levels. While SHP-1-deficient Treg cells showed selective up-regulation of a number of proteins, the expression of other molecules associated with the status of Treg cell activation, such as LAG-3 (46) and CD69, was not affected by the presence or absence of SHP-1. Most prominently, SHP-1 deficiency caused an increase in Treg cells expressing adhesion molecules associated with an activated phenotype. This, in turn correlated with SHP-1-deficient Treg cells being more efficient in conjugate formation than Treg cells of control mice. However, conjugate formation of me/me or me/+ Treg cells is still antigen-dependent indicating that loss of SHP-1 heightens the activity but does not uncouple activity from other regulatory mechanisms (such as TCR dependency). Correlating with the increased ability to form conjugates, SHP-1-deficient Treg cells are more efficient in inhibiting the up-regulation of the co-stimulatory molecules CD80/CD86 on APCs. This further supports the hypothesis that APCs are a target of Treg-mediated suppression, which in turn could affect how well these APCs can activate conventional T cells.
While several previous studies have documented that Treg cells inhibit the overall proliferation and IL-2 production of Tcon cells in vitro, many of these studies were based on analyses of whole populations and not at the level of single cells. Using a combined flow cytometry/microscopy approach that allows visualization of individual Tcon/APC/Treg conjugates, we show that Treg cells inhibit the activation of Tcon cells at two levels. First, there is a modest inhibition of conjugate formation between Tcon cells and APCs. This is more evident at suboptimal levels of antigen (100 ng OVA peptide) than at optimal levels (500 ng OVA peptide), consistent with previous data that Treg cells are more efficient suppressors at low antigen dose (11). Second, the Tcon/APC conjugates can be divided into two groups depending on whether or not Treg cells are co-conjugated to the complexes. Our data show that Tcon cells conjugated to APCs with Treg cells are not activated as evidenced by an almost complete failure to up-regulate CD25 or express IL-2. Remarkably in the same culture, a fraction of Tcon cells conjugated to APCs without Treg cells can be fully activated, while partial inhibition is also detectable in other conjugates; this suggests that the most efficient and complete Treg-mediated suppression is limited to the APCs that are in direct contact with the Treg cells. Interestingly, any so-called bystander suppression that we observed in our culture was limited to IL-2 production by Tcon cells with relatively little apparent effect on CD25 up-regulation. This suggests that suppression by Treg cells may occur via different mechanisms and thereby affect or interfere with Tcon activation at various levels. Furthermore, our dose response experiments demonstrate that at low antigen concentration (100 ng/ml OVA peptide), direct inhibition of Tcon/APC conjugate formation is a major contributor to Treg-mediated suppression. In contrast, at high concentrations (500 ng/ml OVA peptide) inhibition of conjugate formation plays a relatively minor role, but suppression is instead mediated by the formation of tri-partite conjugates (Tcon/APC/Treg). Quite interestingly, SHP-1-deficient Treg cells are more effective at both levels of suppression. me/me and me/+ Treg cells are more efficient than +/+ Treg cells at inhibiting conjugate formation between Tcon cells and BMDCs. In addition, SHP-1-deficient Treg cells are also more effective in forming tri-partite conjugates (Tcon/APC/Treg) and thereby limiting the number of bi-partite Tcon/APC conjugates. Based on our quantitative analysis of these assays, loss of SHP-1 results in about 2-3-fold increased suppressive activity compared to control +/+ Treg cells. Finally, the data presented in this work linking SHP-1 to the strength of Treg-mediated immune suppression, along with the effect of the drug SSG, could prove therapeutically useful in disease states where enhancing Treg cell function could be beneficial.
We are grateful to Dr. Kodi Ravichandran for critical reading of the manuscript, comments and suggestions. We thank Ms. Joanne Lannigan and Mr. Michael Solga (Flow Cytometry Core Facility at UVA) for their outstanding technical and intellectual help with flow cytometric and Imagestream100 analyses. We very much appreciate the help Dr. Irene Mullins provided with the RNA preparation and amplification for the micro array analyses.
This work was supported by National Institutes of Health Grant RO1 AI48672 to U.L.