MS4A4B Is a GITR-Associated Membrane Adapter, Expressed by Regulatory T Cells, Which Modulates T Cell Activation¹

Immune tolerance is an essential aspect of the immune system and has evolved multiple mechanisms to ensure that host tissues are spared damage while pathogens are eradicated efficiently. Its major mechanisms are deletion of autoreactive cells and the generation of regulatory T cells (Tregs)³. Tregs are functionally characterized as T lymphocytes that have acquired the ability to inhibit both T cell-mediated and innate immune responses (1). Multiple routes by which Tregs can acquire this capacity have been described, including central education in the thymus, giving rise to the CD4⁺CD25⁺Foxp3⁺ natural Tregs (2, 3, 4), and peripheral induction through exposure to chronic, incomplete signals, along with the immunomodulatory cytokines TGFβ or IL-10 (5, 6, 7).

TGFβ has an essential role in immune homeostasis and is capable, depending on the tissue-specific context, of inducing anti-inflammatory peripheral Foxp3-expressing Tregs, as demonstrated in TGFβ1^−/− and dominant negative TGFβR mice, which develop a fatal early-onset autoimmune disease characterized by loss of Foxp3-positive Tregs. TGFβ is, however, a pleiotropic cytokine and, in concert with IL-6, can steer T cells toward a proinflammatory Th17 phenotype. Therefore, the physiological context in which TGFβ is experienced by T cells likely governs the outcome for the cell.

Using serial analysis of gene expression (SAGE) and DNA microarrays, we investigated whether there were any genes shared by Tregs induced by over-expression of Foxp3 and by TGFβ. We identified a family of four-pass membrane receptors, the MS4A family, with transcription up-regulated by both Foxp3 and TGFβ. Using a split-ubiquitin genetic screen, we showed interactions between MS4A4B and MS4A6B, the most Treg cell-specific MS4A members with multiple surface receptors, including the costimulatory receptor GITR (TNFRSF18). One of these tetraspan family members has already been described in Th1 cells but their functional relevance to T cells has not been resolved. We show that transduction of MS4A4B into naive T cells can heighten their sensitivity to Ag. MS4A4B overexpression in EL4 cells results in augmented signaling and IL-2 production in response to GITR triggering. We propose that Th1 cells (36) and Tregs use these adaptors to increase their ability to perceive Ag immunity and for ensuring adequate regulation of self-reactivity.

A1(M).RAG1^null TCR-transgenic mice and CBA/Ca mice were bred and maintained in specific pathogen-free conditions at the Sir William Dunn School of Pathology. All procedures were conducted in accordance with the Home Office Animals (Scientific Procedures) Act of 1986.

Four long SAGE libraries were generated from A1.RAG^null CD4⁺ T cells. SAGE libraries were generated as previously described (33). Two libraries were constructed from cells stimulated for 7 days with (DBYT) or without (DBY) exogenous human rTGFβ added at 2 ng/ml. The other two libraries were identical to DBY and DBYT apart from isolation of CD4 T cells from the bone marrow-derived dendritic cell (BMDCs) by Ficoll centrifugation after 7 days and reculture in IL-2 at 2000 U/ml for a further week. These libraries were named DBY.IL2 and DBYT.IL2. Data analysis was performed using the software package !SAGEClus (Ref. 33 ; www.molbiol.ox.ac.uk/pathology/tig).

Custom microarrays with immune system-directed bias were prepared using 70mer probes specific for 768 genes. Each probe was printed with four replicates on nonadjacent areas onto amine slides (Genetix). Details of the microarray are available from ArrayExpress (Accession Number: A-MEXP-239; www.ebi.ac.uk/microarray-as/ae/). RNA integrity was confirmed using a 2100 BioAnalyser (Agilent). Total RNA was reverse transcribed using Superscript III reverse transcriptase and labeled as directed using the Cy3 and Cy5 3 DNA 900 dendrimer labeling kit (Genisphere) using a two-step hybridization protocol with on-slide mixing on a Slide Booster (Advalytix). Arrays were scanned using a ScanArray Express HT scanner (PerkinElmer). Image analysis was performed using BlueFuse v 2 (BlueGnome Ltd).

DNase I-treated RNA was reverse transcribed using StrataScript first strand synthesis kits (Stratagene) and random hexamer primers. For each experimental sample, a minimum of two biological replicates was performed. Individual gene assays were performed in triplicate in a 96-well format using purpose-designed primers and fluorogenic probes (Eurogentec and supplementary Table I).⁴ Data was normalized to an endogenous control, such as hypoxanthine phosphoribosyltransferase or CD3γ-chain, and expressed relative to a reference population using the comparative cycle threshold method for relative quantification.

MS4A4B was cloned into the retrovirus vector MIGR1, which is bicistronic and uses GFP as a reporter (a gift from Shohei Hori, RIKEN Research Centre for Allergy and Immunology, Yokohama City, Japan) using PCR. MIGR1 and MS4A4B-MIGR1 were packaged in Plat-E 293 cells (a gift from Dr. Siamon Gordon, Oxford University, Oxford, United Kingdom) by transfection using Fugene-6 reagent (Roche). Following 3 days of culture at 32°C, polybrene was added at 8 μg/ml. CD4 T cells from A1.RAG^null mice were stimulated with peptide-pulsed dendritic cells (DCs) for 3 days before transduction with MIGR1 or MS4A4B-MIGR1 constructs. Plat-E supernatants containing the packaged MIGR1 or MS4A4B-MIGR1 virions were added to BMDC/T cell cultures and the culture plates were centrifuged at 3000 rpm for one hour at 37°C followed by 72 h of culture at 37°C.

The coding sequence of mouse MS4A4B was cloned into the BglII and BamH1 sites of the expression vector pMTF, which contains an EF1α promoter and neomycin resistance gene. EL4 cells were transfected by electroporation with the pMTF-MS4A4B construct and cloned by limiting dilution. MS4A4B-expressing cells were selected by neomycin resistance using G418 at 1 mg per ml. Positive clones were confirmed using flow cytometry for MS4A4B.

CD4 T cells were stimulated with peptide-pulsed BMDCs or plate-bound anti-CD3 (145.2C11) in 96-well culture plates for 48 h. Tritiated thymidine (Amersham Biosciences) at 0.018 MBq per well was added 16 hours before harvesting the cultures onto glass-fiber filters and scintillation counting using standard techniques.

IFN-γ and IL-2 were measured by ELISA using paired Ab sets from BD Biosciences. IFN-γ was assayed using R46A2 and biotinylated XMG1.2 for detection. IL-2 was measured using 4 μg/ml JES6–1A12 and biotinylated JES6–5H4 for detection.

Cell lysis was conducted with ice-cold lysis buffer (Tris 20 mM pH 7.6, NaCl 140 mM, EDTA 2 mM, NaF 0.5 mM, sodium orthovanadate 1 mM, β-glycerophosphate 25 mM, sodium pyrophosphate 2 mM, benzamidine 2 mM, DTT 0.5 mM) and complete EDTA-free protease inhibitors containing 0.5% Triton X-100 (Roche). Cell lysates were clarified by centrifugation at 14,000g for 15 min at 4°C. Lysates were subjected to SDS-PAGE, using ∼10 μg protein per lane, and transferred onto polyvinylidene difluoride filters (Invitrogen). Filters were blocked for 1 h with 5% skim milk or 5% BSA and then probed with the indicated Abs. Bound Ab was revealed using HRP-conjugated secondary Abs using ECL (SuperSignal; Pierce Biotechnology). Abs used were anti-β-actin (Santa Cruz Biotechnology), rabbit anti-MS4A4B (34), GITR (YGITR 765), Thy1.2 (30H12), p42/44 MAPK (3A7), and phospho p42/44 MAPK (Thr202/Tyr204) (both from Cell Signaling Technology).

Cells were stained with the following Abs: anti-GITR YGITR 765.4.2 (produced in house), fluorochrome-conjugated anti-CD25, anti-CD4, and anti-CD8 (BD Biosciences), PE- or allophycocyanin-conjugated anti-Foxp3 (eBioscience), and rabbit polyclonal anti-MS4A4B (34). Surface staining for CD4 and CD25 was performed on ice for 30 min, followed by washing in PBS and permeabilization with 0.5% saponin before staining for MS4A4B or rabbit Ig as a control for 30 min and detection with FITC-conjugated goat anti-rabbit Ig. Blocking of MS4A4B staining with immunizing peptide (HQGTNVPGNVYKNHPGEIV) was performed in some cases for 15 min before staining. Foxp3 was detected according to the protocol provided by eBioscience. For detection of IL-2 production by EL4 cells, cells cultured for 24 h with 10 μg/ml brefeldin A for the final 4 h were fixed in 2% formalin followed by permeabilization with 0.5% saponin. APC-conjugated anti-IL2 (BD Biosciences) was used for IL-2 detection.

Split-ubiquitin libraries were constructed by Dualsystems Biotech. The libraries DBYT.IL2/x-NubG and DBYT.IL2/NubG-x were directionally cloned into pPR3-C and pPR3-N, respectively. MS4A4B and MS4A6B cDNAs were cloned into the bait construct pBT3-STE. The bait constructs were transformed into the yeast strain NMY51 (MATa his3Δ200 trp1–901 leu2–3,112 ade2 LYS2::(lexAop)₄-HIS3 ura3::(lexAop)₈-LacZ ade2:: (lexAop)₈-ADE2 GAL4) using standard procedures (35). For two-hybrid screens, transformants were grown on selective medium lacking leucine, tryptophan, histidine, and adenine, with addition of 20 mM 3-AT. Positive clones were sequenced by colony PCR using the primer set pPR3NFOR 5′-GTCGAAAATTCAAGACAAGG-3′ and pPR3NREV 5′-AAGCGTGACATAACTAATTAC-3′ for the NubGx library and pPRCFOR 5′- TTTCTGCACAATATTTCAAGC-3′ AND pPRCREV 5′-CTTGACGAAAATCTGCATGG-3′ for the xNubG library. Library plasmids were isolated from positive clones and retransformed into NMY51 to test bait-dependency. Only preys which activated the histidine and adenine reporters in the presence of MS4A4B or MS4A6B and not pCCW-Alg5 or pBT3STE were considered true interactors.

Mouse GITRL (41 to 174 aa) was cloned into the expression plasmid pMF-neo, which encodes EF1α promoter, neomycin resistance, murine IgG2a signal sequence, and a human aglycosyl IgG1 Fc segment. The resulting plasmid was transfected into CHO-S cells (Invitrogen) and stable transfectants selected using G418. Fc-GITRL was purified using protein A agarose (Invitrogen).

TGFβ has an important role in the peripheral induction and maintenance of Foxp3 expression by Tregs (5, 8, 9, 10). SAGE library analysis and custom immune-biased DNA microarrays were used to gain insight into the molecular mechanisms underlying the TGFβ-induced conversion process. In the first approach, four new SAGE libraries were constructed from activated TCR-transgenic CD4⁺ T cells from mice mono-specific for the male Ag H-Y in the context of H2-E^k (A1.RAG1^null), with or without added TGFβ according to Materials and Methods. A1.RAG^null mice contain no Foxp3-positive natural Tregs (nTregs), so any experimentally induced emergence of Foxp3-positive CD4⁺ cells is due to de novo induction. The suppressive properties of TGFβ-induced Tregs (iTregs), Foxp3-transduced Tregs, and CD4⁺CD25⁺ nTregs are equivalent (data not shown). The data set of 48 libraries (described in detail in Ref. 11) was interrogated with a seed tag representing nTregs and iTregs; those tags which best fit the cell distribution of the seed tag libraries indicated by bold typeface are listed in Fig. 1. Previously described Treg or anergy-associated transcripts were abundantly expressed by these cells, CD27 (12), GITR (13), Tgtp (14), and Nkd2 (15); however, the highest ranked transcript which was expressed preferentially in the Treg libraries was MS4A4B. MS4A4B is a four-pass membrane protein of the MS4A family (Fig. 1,A). The MS4A family has at least 13 members on chromosome 19 in the mouse (16). Within the 48 SAGE libraries, four additional MS4A members were identified with widely different expression patterns (Fig. 1,B). Notably, MS4A6B had a broadly similar expression pattern to MS4A4B, albeit with lower tag numbers. MS4A6D was found to be expressed at the highest tag frequency in libraries derived from DCs with a tolerizing bias i.e., immature DCs differentiated from embryonic stem cells or murine bone marrow in the presence of TGFβ or IL10 (Fig. 1 B).

The SAGE data were validated using a custom DNA microarray approach. This data is available from Array Express under accession number E-MEXP-1157, (www.ebi.ac.uk/microarray-as/ae/). cDNA from A1.RAG^null T cells transduced with Foxp3 or treated with or without TGFβ were analyzed. These results are presented in Tables I and II. Fourteen transcripts were significantly down-regulated vs nine up-regulated by Foxp3 transduction and 26 down-regulated vs 11 up-regulated in response to TGFβ. There was also overlap in the identity of transcripts that were down-regulated by both Foxp3 and TGFβ. Five transcripts significantly down-regulated following Foxp3 transduction were also decreased by TGFβ treatment (indicated in Tables I and II). The Th2-related transcripts IL-4 and IL-10 were significantly down-regulated in response to both Foxp3 transduction and TGFβ treatment. MS4A4B and MS4A6C transcripts were significantly up-regulated following expression of Foxp3.

We quantified the observed increase in MS4A4B expression in response to Foxp3 using real-time quantitative RT-PCR. As shown in Fig. 2,A, there was a 3-fold increase in MS4A4B transcripts in peptide-activated A1.RAG^null T cells following the addition of TGFβ to the culture medium. If the cells were subsequently expanded in high-dose IL-2 (2000U/ml), the MS4A4B transcripts increased 6-fold compared with the equivalent cells expanded in the absence of TGFβ. A 12-fold increase in MS4A4B transcripts was observed following retroviral transduction of A1.RAG^null T cells with Foxp3 in comparison with those transduced with empty retrovirus. Using flow cytometry on permeabilized A1.RAG^null T cells, the protein was increased by one order of magnitude in response to TGFβ (Fig. 2 B). Western blotting with this Ab revealed a protein band of the expected 22kD size, which increased in A1.RAG^null T cells in response to TGFβ.

Flow cytometry was performed to test whether up-regulation of MS4A4B protein observed in response to TGFβ was secondary to induction of Foxp3 in these cells. A1.RAG^null T cells, which had been retrovirally transduced with MIGR1-Foxp3 or empty MIGR1 (Fig. 2,C) and FACS sorted for the top 20% brightest GFP expressing cells, showed a substantial increase in MS4A4B expression compared with Foxp3 negative cells. To determine whether a positive correlation exists between cellular Foxp3 protein expression and MS4A4B, a further FACS sort experiment was performed where four fractions based on GFP brightness were sorted from A1.RAG^null T cells retrovirally transduced with MIGR-Foxp3 (Fig. 2,D). As shown in Fig. 2, D and E, there was correlation between the level of expression of Foxp3 as shown by FACS and the intensity of MS4A4B by Western blot. Staining of CBA/Ca splenocytes for CD4, CD25, Foxp3, and MS4A4B shows that around 99% of CD4⁺CD25⁺Foxp3⁺ nTregs stain positively for MS4A4B (Fig. 2 F). Thus it would appear that MS4A4B expression is strongly influenced by Foxp3 expression and is induced in cells expressing Foxp3.

At least thirteen MS4A genes are located in close vicinity on mouse chromosome 19. Their relative locations and transcriptional orientation are shown in Fig. 2,G. It has been proposed that the different MS4A genes arose during evolution by gene duplication (17). It is possible therefore that transcription of the different MS4A members is driven by similar promoter sequences. We used real-time quantitative RT-PCR to quantify all 13 members of the MS4A family (Fig. 2 G) and found that all of the MS4A members, with the exception of MS4A2, 3, 6D, and 8, were strongly up-regulated (5- to 10-fold increase) by TGFβ or Foxp3 transduction.

We next tested whether Foxp3 was an absolute requirement for T cell MS4A4B expression during development by staining thymocytes and splenocytes from the CBA/Ca mouse strain and A1.RAG^null mice (the latter having no detectable Foxp3 expression at any stage of T cell development). MS4A4B was detectable at equivalent low levels on double-negative thymocytes in both strains of mice (supplemental Fig. 1A). MS4A4B expression transiently dropped in wild-type mice at the double-positive stage but not in A1.RAG^null double positives, implying that MS4A4B may be required for positive selection in A1.RAG thymocytes but not in conventional mice. CD4 single-positive thymocytes in both strains expressed bimodal expression of MS4A4B at equivalent intensities (supplemental Fig. 1A). MS4A4B was expressed by Foxp3-positive thymocytes with a single peak of staining (supplemental Fig. 1B). On analysis of splenic CD4⁺ and CD8⁺ T cells, it was clear that peripheral CD4⁺ T cells exhibit a nearly two-log range in expression of MS4A4B in both A1.RAG and CBA/Ca mice (supplemental Fig. 1C). Thus, MS4A4B is expressed early in T cell development in a Foxp3-independent manner, although in the absence of Foxp3 there is a greater range of expression than in nTregs, which express MS4A4B strongly.

We next explored possible signaling mechanisms for MS4A4B in T effectors and Tregs using a yeast split-ubiquitin screen. The advantage of this approach over conventional yeast two-hybrid screens is that protein interaction is measured at the plasma membrane, rather than as truncated fusions in the yeast nucleus (18). Two libraries were constructed from A1.RAG^null T cells (see Materials and Methods). RNA used for library construction was validated to test expression of Foxp3, MS4A4B, MS4A6B, and MS4A6C (supplemental Fig. 2A). Additionally, an in vitro suppression assay was performed using DBYT cells to confirm that they displayed suppressive activity on Dby Ag-activated A1.RAG^null T cells (supplemental Fig. 2B). MS4A4B and MS4A6B were expressed as C-terminal ubiquitin-LexA-VP16 fusions (Fig. 3,A) in S. cerevisiae strain NMY51 and tested for their ability to activate reporter transcription of the histidine and adenine reporters only in the presence of a strong ubiquitin reassociation (supplemental Fig. 2, C and D, middle and right panels). Having established that MS4A4B and MS4A6B activates transcription of the yeast nutritional reporters with a membrane-located positive control construct, we screened the two iTreg libraries with MS4A4B and MS4A6B as baits. Six integral membrane proteins, GITR, ORAI1, MS4A6B, Ly6E, GPR177 and H2-DMa, were shown to interact with MS4A4B in a bait-dependent fashion (Fig. 3,B). We confirmed the MS4A4B-MS4A6B interaction with our MS4A6B screen, obtaining two MS4A4B interacting clones in addition to 12 clones of GITR and two of MS4A6B, suggesting MS4A6B self-association. We then confirmed the GITR/MS4A4B interaction using GITRL-Fc to coimmunoprecipitate MS4A4B-GITR complexes from iTreg lysates (Fig. 3,C). GITR has been shown to enhance TCR-driven signaling via p42/44 MAPK (ERK) (19). We next tested whether, in the GITR-expressing EL4 cell line, over- expression of MS4A4B might costimulate ERK signaling. As shown in Fig. 3 D, following activation of EL4 cells with anti-CD3 Abs, a 3- to 8-fold increase in ERK phosphorylation was seen in EL4 cells expressing MS4A4B relative to those untransfected. Thus, the presence of MS4A4B enhances proximal TCR signaling.

The two canonical members of the MS4A family MS4A1 (CD20) and MS4A2 (FcεR1β) serve to amplify signals transduced through the BCR and high affinity IgE receptor FcεRI, respectively. As MS4A4B was observed to augment TCR signaling in EL4 cells, we pursued the idea that it may serve to lower TCR sensitivity to Ag in primary T cells. We transduced A1.RAG^null T cells with MS4A4B, Moflow sorted the 20% highest MS4A4B-expressing cells, then tested their activation to Ag or anti-CD3 stimulation. T cell IL-2 and IFN-γ production and proliferation were moderately but significantly increased in cells transduced with MS4A4B compared with those transduced with empty retrovirus (Fig. 4). Enhancement of the proliferative response was more apparent at lower concentrations of antigenic peptide (0.1 nM) than at higher amounts. These differences were not due to alterations in surface TCR or CD28 levels (data not shown). Thus, MS4A4B lowers the threshold for Ag-induced activation of primary T cells, resulting in increased IL-2 synthesis and proliferation.

We next asked whether the interaction of MS4A4B with GITR had functional consequences for GITR costimulation. To this end we stimulated MS4A4B-expressing EL4 cells (Fig. 5,A) with anti-CD3 and GITRL-Fc or anti-GITR. Upon stimulation of EL4–4B cells with anti-CD3 and anti-GITR, p44 ERK phosphorylation was increased at all time-points in comparison to EL4 cells lacking MS4A4B. To determine whether augmented functional responses to GITR triggering could be measured in MS4A4B-expressing cells, we analyzed IL-2 production by flow cytometry (Fig. 5 C). Measurement of IL-2 production by these cells revealed that expression of MS4A4B in the absence of GITR stimulation was sufficient to increase IL-2 production in response to anti-CD3 stimulation by 4- to 5-fold. Remarkably, whereas EL4 cells lacking MS4A4B responded with little increase of IL-2 production to GITRL or anti-GITR stimulation, 10% of those cells expressing MS4A4B consistently produced IL-2 in response to either GITRL or anti-GITR. Thus, MS4A4B appears to enhance the capacity of GITR to promote IL-2 production and T cell activation.

Strongly autoreactive T cells are eliminated in the thymus in the course of T cell development. To ensure self-tolerance, a further fraction of the positively selected T cells are directed toward Treg cell function and exported to the periphery as CD4⁺CD25⁺ nTregs (2, 3, 4). Emerging evidence in recent years also points to further routes of peripheral Treg induction (5, 6, 9, 20). The common theme arising from these peripheral Treg induction modes is one of repeated stimulation of naive T cells under suboptimal activation conditions where danger signals are absent. It is unknown what molecular mechanisms instruct some Ag-naive cells to preferentially expand to danger-free Ag with the outcome of a Treg phenotype whereas other naive cells remain quiescent to such stimuli.

In this study we found multiple significantly up- and down-regulated genes in Foxp3- and iTreg populations. These observations are in accordance with recent reports that Foxp3 acts as both a transcription repressor and activator of multiple genes, both directly and indirectly (21). MS4A4B was up-regulated most strongly and specifically by Treg populations in response to both TGFβ and transduction with Foxp3. A correlation was shown between Foxp3 expression and MS4A4B using retrovirally expressed Foxp3. CD4⁺CD25⁺Foxp3⁺ nTregs were uniformly shown to be strongly positive for MS4A4B expression.

MS4A4B is a member of a family of 13 highly homologous MS4A genes on mouse chromosome 19, which is syntenic with human chromosome 11 (17). Control of expression of the MS4A members is highly cell type specific (22, 23). Likewise, MS4A4B, MS4A4C, MS4A6B, MS4A6C, and MS4A6D have differential distributions in our SAGE libraries. Their transcription is coordinately up-regulated by TGFβ or expression of Foxp3. The archetypal MS4A members CD20 (MS4A1), FcεR1β (MS4A2), and HTm4 (MS4A3) are components of larger oligomeric complexes (22, 24, 25, 26) in which they transmit positive and negative signals (27, 28, 29, 30). The possibility that MS4A4B amplifies positive TCR signals via clustering other surface costimulatory receptors is supported by our data. This might occur in Tregs or cells destined to become such through interacting with multiple other cell surface receptors in a way analogous to that described for MS4A1–3 or as part of a structure similar to the tetraspanin web (reviewed in Ref. 31). The results of our split-ubiquitin screens show clearly that lateral associations within the MS4A family can occur, such as MS4A4B↔MS4A6B and MS4A6B↔MS4A6B. In addition, MS4A4B and MS4A6B have other interactors, both shared and unique, showing the potential for MS4A4B/6B-mediated recruitment of each other’s cis-ligands, such as GITR, into supramolecular signaling clusters on Tregs.

Interaction of GITR with MS4A4B and MS4A6B is of considerable interest as GITR is highly expressed on CD4⁺CD25⁺Foxp3⁺ nTregs and was originally described in the context of Tregs as a negative regulator of Treg activity (13). Subsequently, a positive role in T cell activation has been demonstrated for GITR on non-Tregs when triggered by agonist anti-GITR mAbs or the natural ligand GITRL (32). Our data show that cells expressing MS4A4B have increased sensitivity to GITRL stimulation, resulting in increased IL-2 production. This may be a consequence of GITR clustering in response to its cis-interaction with MS4A4B, resulting in signal amplification. It is conceivable that different signals may be generated by different GITR binding partners, such as MS4A4B/6B.

What might be the value of this to the immune system and why would such adapters be found in both Th1 cells and Tregs? We propose that expression of MS4A family members may serve to reduce the threshold for TCR-mediated signaling directed toward danger-associated extrinsic Ags. In this way, the system would stay responsive until the last traces of Ag had gone. In contrast, expression of such members in Tregs would enable them to remain reactive to danger-free self Ags, even if, as a result of negative selection, their TCR were of lower affinity.

In summary, Foxp3 and TGFβ signaling are linked to up-regulation of the MS4A gene family. All Foxp3-positive cell populations in our experiments expressed elevated levels of MS4A4B. Moreover, MS4A4B and MS4A6B form complexes with each other as well as GITR, in addition to multiple cell surface receptors. MS4A4B interaction with GITR results in augmented GITR signals and T cell IL-2 production. These associations may explain how MS4A4B over-expression was shown to increase activation of T cells.

We thank Dr. Nigel Rust for excellent FACS sorting assistance, the staff of the Dunn School PSB for careful animal husbandry, and Mark Frewin for expert technical assistance.

The authors have no financial conflict of interest.