The Role of LAT–PLCγ1 Interaction in γδ T Cell Development and Homeostasis

The majority of T cells that traffic through lymphoid tissues are αβ T cells, whereas γδ T cells represent only 1–5% of circulating lymphocytes. Rather, γδ T cells populate the gastrointestinal tract, skin, and other epithelial-rich tissues (1, 2). LAT, an important protein in the TCR-signaling pathway, is required for the development of both αβ and γδ T cells. LAT-deficient mice show an early block at the double-negative (DN)3 stage during thymic development and lack mature αβ and γδ T cells (3). After T cell activation, LAT is phosphorylated on multiple tyrosine residues. Among these tyrosine motifs, the four distal tyrosines, 136, 175, 195, and 235 in murine LAT, are responsible for binding PLCγ1, Gads, and Grb2 and are required for TCR-mediated calcium mobilization, Erk activation, and cytokine production (4).

Published studies (5–7) using mice with these residues mutated, LATY136F and LAT3YF (Y175, Y195, and Y235 are all mutated), suggest a differential requirement for LAT signaling during development and activation of αβ and γδ T cells. In LATY136F mice, in which the PLCγ1 binding site is mutated, αβ T cell development is partially blocked at DN3. CD8 and γδ T cells are nearly absent from secondary lymphoid organs. In contrast, CD4 T cells hyperproliferate, resulting in a severe autoimmune syndrome (5, 6). These CD4 T cells have an effector/memory-like phenotype and produce large amounts of Th2 cytokines, such as IL-4. Correspondingly, serum IgE and IgG1 are elevated, and autoantibodies are detectable (5, 6). These data demonstrate that the LAT–PLCγ1 interaction plays important roles in the development and homeostasis of αβ T cells. The effect of this mutation on γδ T cell development and homeostasis has not been determined.

In contrast, LAT3YF mice, in which the Gads and Grb2 binding sites are mutated, display a strikingly different phenotype (7). They have a complete block in αβ T cell development at DN3, indicating the importance of signaling through Gads and Grb2 at the pre-TCR checkpoint. Surprisingly, γδ T cells, which rearrange both γ and δ genes in the DN compartment, only have a partial defect in their development in these mice. These γδ T cells, ∼20% of which express CD4, are able to populate peripheral lymphoid tissues and instigate an autoimmune syndrome in 6-mo-old mice (7). Interestingly, mice devoid of Itk, a tyrosine kinase that phosphorylates PLCγ1, display a similar γδ T cell phenotype. Itk^−/− mice have a markedly increased number of CD4 γδ T cells in secondary lymphoid tissues (8). These γδ T cells are able to produce Th2 cytokines and were identified as innate-like γδ NKT cells, because they expressed the transcription factor PLZF (9, 10). Taken ogether, these data suggest that αβ and γδ T cells have different signaling requirements during development and that dampened TCR signaling in both αβ and γδ T cells can lead to a Th2-skewed phenotype.

Because CD4⁺ αβ T cells dominate the periphery of LATY136F mice, it is difficult to study the effect of this mutation on γδ T cells. To specifically investigate the importance of LAT–PLCγ1 signaling in γδ T cells, we crossed LATY136F mice, designated as LAT^m/m, with TCRβ^−/− mice to generate TCRβ^−/−LAT^m/m mice. These mice revealed that abolishing the LAT–PLCγ1 interaction resulted in a partial block in γδ T cell development. γδ T cells could be found in the skin and small intestine and had no obvious activated phenotypes. Interestingly, a population of CD4⁺ γδ T cells in the spleen and lymph nodes underwent rapid proliferation and caused a severe autoimmune syndrome. Our data suggested that LAT functions differently in maintaining the homeostasis of different populations of γδ T cells.

LAT^m/m and LAT^−/− mice were described previously (3, 6), and TCRβ^−/−, MHCII^−/−, and CD4^−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were housed in a specific pathogen–free facility. All experiments with mice were conducted in accordance with National Institutes of Health guidelines and were approved by the Duke University Institutional Animal Care and Use Committee.

Single-cell suspensions were stained with fluorescently conjugated Abs (BioLegend, San Diego, CA) in the presence of 2.4G2 (anti-FcγII/III receptor). For intracellular staining, splenocytes were stained for surface markers, fixed, permeabilized, and stained for IL-17, IL-4, IL-2, IFN-γ, T-bet, EOMES, GATA3, and PLZF. For cytokine production, cells were stimulated with PMA (20 ng/ml) and ionomycin (0.5 μg/ml) for 4 h in the presence of monensin. 7-aminoactinomycin D or a Live/Dead Staining kit (Invitrogen) was used to distinguish live cells from dead ones. Data were acquired on a FACSCanto II (BD Biosciences) and analyzed using FlowJo software (Tree Star, Ashland, OR). For the anti-dsDNA ELISAs, plates were coated with 5 μg/ml calf-thymus DNA in Reacti-Bind DNA Coating Solution (Pierce). ELISA on serum Abs was done as previously described (11).

Small intestines were cut longitudinally and incubated in HBSS containing 5% FBS and 5 mM EDTA. Cells were strained and resuspended in 45% Percoll (Amersham Biosciences) and isolated using a 45/70% Percoll gradient. Ears were separated into ventral and dorsal halves and incubated in 20 mM EDTA for 3 h at 37°C. Epidermal sheets were removed, incubated in trypsin, homogenized, and analyzed by FACS.

T cells from spleen and lymph nodes were isolated by negative depletion using EasySep Biotin Purification kits (STEMCELL Technologies). Regulatory T cells (Tregs) (CD4⁺CD25⁺), non-Tregs (CD4⁺CD25⁻), and γδ T cells were sorted on a FACSVantage flow cytometer (BD Biosciences). A total of 1 × 10⁶ T cells was transferred into recipient mice via i.v. injection.

γδ T cells from TCRβ^−/−LAT^m/m mice were isolated from the spleen, stimulated for 24 h with PMA and ionomycin, and fused with TCR-deficient BW5147 cells to generate T cell hybridomas. PCR primers for Vγ1.1, Cγ4, Vδ2.2, Vδ4, Vδ5, Vδ6.1, Vδ6.2, Vδ7, and Cδ were used to amplify cDNAs from these hybridomas (12, 13). PCR products were purified and sequenced. Sequences were analyzed on the international ImMunoGeneTics information system (http://www.imgt.org).

As previously reported, ablation of TCRβ expression caused a block in αβ T cell development at the DN3 stage, but a small number of double-positive thymocytes was still present in TCRβ^−/− mice (14). Similar to TCRβ^−/− mice, TCRβ^−/−LAT^m/m mice had ∼10% double-positive thymocytes (Fig. 1A). Analysis of the DN compartment in TCRβ^−/−LAT^m/m mice revealed that, although the percentages of DN1–DN4 subsets were similar to those in TCRβ^−/− mice, the percentage of CD3⁺TCRγδ⁺ DN thymocytes was moderately reduced, from 15.6% in TCRβ^−/− mice to 11.2% in TCRβ^−/−LAT^m/m mice (Fig. 1A). In addition, the numbers of total, DN, and CD3⁺TCRγδ⁺ thymocytes were reduced in TCRβ^−/−LAT^m/m mice (Fig. 1B).

We subsequently analyzed whether LAT^m/m γδ thymocytes were properly selected. CD5 is upregulated upon selection of γδ cells, whereas CD25 is downregulated prior to their exit from the thymus (15–17). As shown in Fig. 1C, TCRγδ⁺ thymocytes from TCRβ^−/−LAT^m/m mice had impaired CD5 upregulation and CD25 downregulation. The numbers of CD3⁺TCRγδ⁺CD25⁻ thymocytes in the DN compartment of TCRβ^−/−LAT^m/m mice were reduced significantly (Fig. 1B). These data indicated that γδ thymocyte selection is impaired by the LATY136F mutation. In addition, overall surface expression of γδ TCR was lower, and there were more γδ T cells that expressed an intermediate level of the TCR. γδ T cells in adult thymuses and spleens express Vγ1 and Vγ2. The percentages of γδ thymocytes expressing Vγ1⁺ and Vγ2⁺ were similar in TCRβ^−/−LAT^m/m and TCRβ^−/− mice; however, the percentage of Vδ6.3⁺ γδ thymocytes was reduced in TCRβ^−/−LAT^m/m mice (Fig. 1D).

It was suggested that γδ T cells acquire their effector fates in the thymus (18). CD27 is thought to be high on IFN-γ–producing γδ progenitors and low on IL-17–producing progenitors (19). Further studies indicated that TCRγδ⁺CCR6⁺ thymocytes produce IL-17, whereas TCRγδ⁺NK1.1⁺ thymocytes do not (20). The majority of TCRγδ⁺ in both TCRβ^−/− and TCRβ^−/−LAT^m/m mice were CD27^hi and CCR6⁻, indicating that they were not IL-17 producers (Fig. 1E). NK1.1 and the transcription factor PLZF identify a population of innate-like γδ NKT cells (21). Analysis of γδ⁺ thymocytes from TCRβ^−/− and TCRβ^−/−LAT^m/m mice revealed that the NK1.1⁺ population was reduced from 10.3% to 2.8%, and the PLZF⁺ population was reduced from 17.1% to 2.6% (Fig. 1E), indicating that the development of innate-like γδ NKT cells was impaired in the absence of LAT–PLCγ1. We also examined the development of γδ⁺ thymocytes that populate epithelial tissues. γδ thymocytes that traffic to the epidermis tend to be T-bet⁺, whereas those expressing Vδ4 go to the intestinal epithelial (22). Both the T-bet– and Vδ4-expressing γδ thymocyte populations were reduced in TCRβ^−/−LAT^m/m mice compared with those in TCRβ^−/− mice (Fig. 1F). Together, these data indicated that, in the absence of the LAT–PLCγ1 interaction, the development and selection of different γδ thymocyte subsets were partially blocked.

γδ T cells are highly enriched in epithelial, mucosal, and barrier tissues, including the skin, lungs, and small intestine, where they are thought to function in initial host defense (23). Epithelial γδ T cells, which develop in the fetal and perinatal thymus, have a restricted repertoire, using Vγ3 and Vγ4 (12). We wanted to determine whether γδ T cells are present in epithelial-rich compartments of TCRβ^−/−LAT^m/m mice.

Immunofluorescent staining of ear sections from 3-mo-old mice showed similar numbers and morphology of dendritic epidermal T cells (DETCs) present in TCRβ^−/− and TCRβ^−/−LAT^m/m mice (Supplemental Fig. 1A). FACS analysis of cells isolated from ear epidermal sheets confirmed similar percentages of DETCs in these mice. All of these DETCs expressed Vγ3 (Fig. 2A). Interestingly, although the surface expression of TCRβ in LAT^m/m mice was greatly reduced (6), expression of TCRγδ remained high on these DETCs.

Among the small intestine intraepithelial lymphocytes (IELs), there are two subsets of γδ T cells, derived either from the thymus (Thy1⁺) or extrathymically (Thy1⁻) (24). In TCRβ^−/−LAT^m/m mice, the percentages of both small intestine γδ subsets were reduced (Fig. 2B), which could be attributed to a partial block in development. In the Thy1.2⁻ subset, the majority of T cells in both TCRβ^−/− and TCRβ^−/−LAT^m/m mice were CD8α⁺; however, in the Thy1.2⁺ subset, there was an increased percentage of CD4⁺ T cells, even though the number of these cells was drastically reduced. The surface expression of TCRγδ for both subsets in TCRβ^−/−LAT^m/m mice was similar to that in TCRβ^−/− mice (Fig. 2B). Because the surface level of TCRγδ was high, we were able to analyze Vγ and Vδ usages by flow cytometry. The Thy1.2⁻ subset in TCRβ^−/−LAT^m/m mice had similar usages of Vγ1.1, Vγ2, Vδ6.3, and Vδ4 to TCRβ^−/− mice (Fig. 2C). The Thy1.2⁺ subset in TCRβ^−/−LAT^m/m mice seemed to be skewed, with increased Vγ1.1⁺Vδ6.3⁺ cells and reduced Vγ2⁺ or Vδ4⁺ cells compared with control mice (Fig. 2C).

Together, these data demonstrated that, in TCRβ^−/−LAT^m/m mice, the development of IEL γδ T cells was partially blocked; however, DETCs in the skin were not affected. Interestingly, unlike αβ T cells in LAT^m/m mice, these γδ T cells in the epithelial tissues maintained normal levels of surface TCR and were not hyperproliferative, suggesting that LAT functions differently in αβ and γδ T cells.

LAT^m/m mice have splenomegaly and lymphadenopathy at 4–6 wk of age, whereas LAT3YF mice show a similar syndrome at 6 mo. Comparatively, 4–6-wk-old TCRβ^−/−LAT^m/m mice had no apparent splenomegaly. However, at 8–12 wk, their spleens and lymph nodes appeared enlarged, and the spleen weight and number of splenocytes were increased dramatically (Fig. 3A, 3B). Because CD4 T cells in LAT^m/m mice downregulate the expression of surface TCR (5, 6), we used the T cell markers Thy1.2 and CD5 to identify γδ T cells in 4-, 6-, and 8–12-wk-old mice. In agreement with a partial block in γδ thymocyte development, 4-wk-old TCRβ^−/−LAT^m/m mice had decreased percentages of splenic γδ T cells (1.0% compared with 1.8% in control mice) (Fig. 3C). Interestingly, at 6 wk of age, a second population of γδ T cells appeared in the spleens and lymph nodes of TCRβ^−/−LAT^m/m mice. This new population expressed higher levels of CD5 than did γδ T cells in TCRβ^−/− mice or those in 4-wk-old TCRβ^−/−LAT^m/m mice. By 8–12 wk, these CD5^hi T cells expanded tremendously and composed ∼30% of total splenocytes (Fig. 3C).

Further analysis of CD5^hi γδ T cells revealed that the majority of them (∼70%) expressed CD4. In contrast, very few γδ T cells in TCRβ^−/− mice were CD4⁺. These CD5^hi cells in TCRβ^−/−LAT^m/m mice had an effector/memory-like phenotype (CD44^hiCD62L^lo) (Fig. 3D). Reminiscent of hyperproliferative CD4⁺ αβ T cells in LAT^m/m mice, CD5^hi γδ T cells in 12-wk-old TCRβ^−/−LAT^m/m mice downregulated surface TCR expression; however, γδ TCR surface expression was only slightly reduced in 4-wk-old mice (Fig. 3E). Neither TCRβ^−/− nor TCRβ^−/−LAT^m/m γδ T cells from aged mice expressed NK1.1, a marker for innate γδ NKT cells (Fig. 3F).

To investigate whether CD5^hi γδ T cells were derived from CD5^int γδ T cells in the periphery, T cell–transfer experiments were performed. Thy1.2⁺CD5^int γδ T cells purified from 4-wk-old TCRβ^−/− and TCRβ^−/−LAT^m/m mice and Thy1.2⁺CD5^hi γδ T cells from 12-wk-old TCRβ^−/−LAT^m/m mice were adoptively transferred into Thy1.1⁺LAT^−/− mice, which lack T cells. Twelve weeks posttransfer, there was no conversion from Thy1.2⁺CD5^int T cells to Thy1.2⁺CD5^hi T cells (Fig. 3G). CD5^hi γδ T cells remained CD4⁺, and CD5^int γδ T cells remained CD4⁻. Moreover, CD5^int γδ T cells from TCRβ^−/−LAT^m/m mice did not expand as much as those from TCRβ^−/− mice, suggesting that their homeostatic proliferation is impaired. In contrast, CD5^hi γδ T cells from TCRβ^−/−LAT^m/m mice expanded much more than did those from TCRβ^−/− mice (6.9% versus 0.14%). These data suggested that these two γδ subsets in TCRβ^−/−LAT^m/m spleens are likely derived from two different thymic outputs: a conventional pool that expresses relatively normal surface TCR and are defective in homeostatic proliferation and a distinct subset of CD4⁺ γδ T cells that express low levels of TCR and hyperproliferate.

γδ T cells that populate lymphoid tissues develop in late fetal and early postnatal life. These cells typically display high levels of junctional diversity and commonly rearrange using Vγ1.1, Vγ1.2, and Vγ2 (12). We investigated Vγ and Vδ usage and clonality of CD5^hi γδ T cells in TCRβ^−/−LAT^m/m mice. Because TCRγδ was downregulated from the cell surface, it was difficult to analyze Vγ and Vδ usage by flow cytometry. Instead, we generated hybridomas by fusing CD5^hi γδ T cells with TCR⁻ BW5147 cells. FACS analysis of 40 hybridoma clones showed that they all were Vγ1.1⁺ (Supplemental Fig. 1B).

To further analyze Vγ and Vδ usage, total RNAs were isolated and used in PCR amplification of Vγ and Vδ regions. Sequencing of the PCR fragments indicated that these T cells were polyclonal and, in agreement with the FACS data, they indeed used Vγ1.1. Their γ-chain CDR3 lengths varied from 11 to 15 residues (Table I). Sequencing δ fragments showed that they primarily used Vδ5, Vδ6.1, and Vδ6.2. Interestingly, their δ chain CDR3 lengths varied greatly, from 12 to 22 residues. In addition, three clones (B14, D4, and D19) expressed two rearranged and in-frame δ alleles. These data indicated that the hyperproliferative CD5^hi γδ T cells are polyclonal. They use Vγ1.1 and different Vδ segments with diverse junctional regions and CDR3 lengths.

As shown in Fig. 3, γδ T cells in the spleens and lymph nodes of TCRβ^−/−LAT^m/m mice underwent a large expansion, similar to αβ T cells in LAT^m/m mice. We next examined whether these T cells were Th2 skewed and caused autoimmunity. Splenocytes from 4- or 12-wk-old mice were stimulated in vitro prior to intracellular staining. Similar to TCRβ^−/− splenic γδ T cells, ∼30% of CD5^int γδ T cells from 4-wk-old TCRβ^−/−LAT^m/m mice produced IFN-γ, and a small percentage of them produced IL-17 or IL-4. In contrast, ∼90% of CD5^hi γδ T cells in 12-wk-old TCRβ^−/−LAT^m/m mice produced IL-4 (Fig. 4A). Further analysis revealed that these CD5^hi γδ T cells downregulated T-bet and EOMES and upregulated GATA3, the master regulator of Th2 differentiation (Fig. 4B, 4C). Itk-deficient mice have increased γδ T cells, which express Vγ1.1 and Vδ6.3 and produce IL-4. These γδ cells express PLZF and are γδ NKT cells (9, 10). Although TCRβ^−/−γδ T cells had a small population of cells expressing PLZF, TCRβ^−/−LAT^m/m CD5^hi γδ T cells did not express PLZF, indicating that they were not γδ NKT cells (Fig. 4B).

We next wanted to determine the effect of the hyperproliferative γδ T cells on B cell activation and maturation. Although the numbers of B cells were not significantly elevated in TCRβ^−/−LAT^m/m mice (data not shown), they did have an activated phenotype, with upregulated expression of MHC class II and CD86 (Fig. 4D). We also assessed serum Ab levels by ELISA. Our data showed that the concentrations of IgG1 and IgE were significantly elevated in aged TCRβ^−/−LAT^m/m mice, which also had enhanced levels of anti-dsDNA Abs (Fig. 4E). Taken together, these data suggested that hyperproliferative γδ T cells in TCRβ^−/−LAT^m/m mice secrete Th2 cytokines, resulting in B cell activation, class switching, and autoantibody production.

Further evaluation of other organs showed the ability of CD5^hi γδ T cells to infiltrate. In the livers of 4-wk-old TCRβ^−/−LAT^m/m mice, the number of γδ T cells was much reduced compared with TCRβ^−/− mice (0.3% versus 4.3%), and most of them were CD5^int (Fig. 5A). However, in the livers of 12-wk-old mice, most γδ T cells were TCRγδ^loCD5^hiCD4⁺ (Fig. 5A), and their numbers were drastically increased (Fig. 5B). These data indicated that, in addition to excessive proliferation in the spleen and lymph nodes, CD5^hi γδ T cells also infiltrated into the liver.

Next, we determined whether hyperproliferation of CD5^hi γδ T cells could be suppressed by natural Tregs (nTregs). A total of 1 × 10⁶ CD4⁺CD25⁺ Tregs or CD4⁺CD25⁻ conventional T cells (Tcons) from congenic Thy1.1⁺ mice was adoptively transferred into 4-wk-old TCRβ^−/− and TCRβ^−/−LAT^m/m mice. Twelve weeks after transfer, these mice were analyzed for development of the autoimmune syndrome. Donor cells (Thy1.1⁺) were clearly detected in these mice and had no apparent effect on γδ T cells in TCRβ^−/− mice (Fig. 6A). Conversely, TCRβ^−/−LAT^m/m mice that received Tregs had reduced percentages of CD5^hi γδ T cells (Fig. 6A) and much smaller spleens (Fig. 6B) compared with both uninjected controls and mice that received Tcons. Interestingly, TCRβ^−/−LAT^m/m mice that received Tcons displayed an intermediate phenotype. They had slightly larger spleens than did mice injected with Tregs, yet they had similar percentages of γδ T cells as uninjected mice (Fig. 6A, 6B). It is also interesting to note that the addition of Tregs allowed the persistence of CD5^int γδ T cells. Both Tregs and Tcons had no effect on the phenotype of γδ T cells because T cells still displayed high surface CD5 and CD4 and low TCR expression (Supplemental Fig. 2A).

We also examined the ability of Tregs to inhibit proliferation of CD5^hi γδ T cells by transferring CFSE-labeled γδ T cells, with or without Tregs, into LAT^−/− hosts. As shown in Fig. 6C, Tregs slowed the proliferation of CD5^hi γδ T cells from TCRβ^−/−LAT^m/m mice. These results suggested that CD4⁺ Tregs and, to some extent, Tcons are able to inhibit the proliferation of CD5^hi γδ T cells.

Published studies (25–27) demonstrated that the absence of classical MHC molecules has no effect on the development of γδ T cells; however, because most of the γδ T cells in TCRβ^−/−LAT^m/m mice are CD4⁺, we investigated whether these unconventional CD4⁺ γδ T cells are MHC class II restricted or require CD4 signaling by crossing them with MHCII^−/− and CD4^−/− mice. MHCII^−/−TCRβ^−/−LAT^m/m mice had enlarged spleens (Supplemental Fig. 2B). Their γδ T cells expressed CD4, downregulated surface TCR (Fig. 7A), and overproduced IL-4 (Fig. 7B). Interestingly, MHCII^−/−TCRβ^−/−LAT^m/m mice actually had a lower percentage of γδ T cells than did TCRβ^−/−LAT^m/m mice, 12.9% versus 25.8% (Fig. 7A), suggesting that a proportion of the polyclonal γδ T cells might be selected on MHC class II. Still, in the absence of MHC class II molecules, these mutant γδ T cells hyperproliferated and induced splenomegaly. Interestingly, CD4^−/−TCRβ^−/−LAT^m/m mice also developed splenomegaly and lymphadenopathy (Supplemental Fig. 2B). CD5^hi γδ T cells from these mice expanded to make up almost 50% of the splenocytes (Fig. 7A) and overproduced IL-4 (Fig. 7B). These data indicated that CD4 is not required for the development, proliferation, or cytokine production of LAT^m/m γδ T cells. In contrast, because CD4 deficiency enhanced their proliferation, CD4 may actually function to negatively regulate the development or proliferation of these cells.

Mutation of Y136, the PLCγ1 binding site in LAT, results in a severe block in thymocyte development of αβ T cells. This tyrosine residue is also indispensable for controlling CD4 T cell homeostasis, because LATY136F mice bear hyperproliferative CD4 T cells that are Th2 skewed (5, 6). Our data demonstrated that, without the LAT–PLCγ1 interaction, γδ T cells had a partial block in their development and selection. Consequently, the amount of γδ T cells in the thymus, intestine, and liver in 4-wk-old mice was reduced, although the percentage of γδ T cells in the skin remained similar. Despite the developmental defect, γδ T cells in the skin and intestine were not hyperactivated and had normal levels of the TCR at their surface. However, a population of CD4 γδ T cells in the secondary lymphoid organs of TCRβ^−/−LAT^m/m mice had a very similar phenotype to αβ T cells in LAT^m/m mice. These γδ T cells, most of which were CD4⁺, underwent uncontrolled expansion and caused an autoimmune syndrome. These data indicated that the LAT–PLCγ1 interaction may function differently in different subsets of γδ T cells.

An alternative explanation for the ability of γδ T cells to maintain homeostasis in the epithelium but hyperproliferate in the spleen may be due to temporal differences in development or reactivity. Epithelial γδ T cells that develop in the fetus may not express TCRs that have high affinities for self-Ags. It is possible that the Ags that these T cells react to in the fetal thymus are not present in the skin and small intestine. Thus, these T cells may not be constantly stimulated by self-Ags. Conversely, CD5 and CD25 staining of TCRβ^−/−LAT^m/m thymocytes indicated that their postnatal selection is impaired. This may allow highly self-reactive T cells into the circulation, where they are activated by self-Ags.

Our results showed that secondary lymphoid tissues in TCRβ^−/−LAT^m/m mice contain two distinct populations of γδ T cells. The CD5^hi population expressed CD4, had low TCR levels, and rapidly expanded. Conversely, CD5^int cells did not undergo uncontrolled expansion. Over time, the CD5^int T cells population actually diminished, likely as a result of competition for cytokines. Transfer experiments also revealed that these CD5^int T cells may have a defect in homeostatic proliferation or cell survival. A similar phenomenon was seen with CD8 T cells in LAT^m/m mice (28). The disparity in proliferation potential of these γδ cells might be due to a difference in self-reactivity or dependence on normal LAT signaling to maintain homeostasis.

In LAT3YF mice, LAT interaction with Gads and Grb2 is disrupted. LAT3YF mice have a complete block in αβ T cell development, whereas LAT^m/m mice have a partial block. γδ T cells are able to develop in both of these mutant lines, suggesting that they may require less TCR-mediated signaling during development. In LAT3YF and TCRβ^−/−LAT^m/m mice, γδ T cells are able to hyperproliferate in secondary lymphoid organs (7), although the LAT3YF disease is much less severe; the autoimmune syndrome does not develop until 30 wk of age, and only ∼30% of T cells produce IL-4. LAT^m/m γδ T cells induced splenomegaly as early as 8 wk, and ∼90% of these T cells produced IL-4. The γδ thymocyte selection defect appears similar between these two knock-in lines, suggesting that the difference in autoimmune syndrome severity is attributed to signaling differences in the periphery.

Recruitment of Gads and Grb2 to LAT is important for PLCγ1 function. Grb2 stabilizes the interactions between LAT and PLCγ1 (29). Furthermore, Gads binds SLP-76, which recruits Itk to phosphorylate PLCγ1 (30, 31). The inability of LAT to bind Gads and Grb2 in LAT3YF mice also may have ramifications for LAT–PLCγ1 signaling. Therefore, the LAT3YF mutation may have more of an impact on LAT function in TCR-mediated signaling than does the LATY136F mutation. The residual function of these LAT mutants is likely important in T cell expansion, thus explaining the delayed and less severe phenotype in LAT3YF mice compared with TCRβ^−/−LAT^m/m mice.

Finally, it is known that, in the absence of Tregs, autoimmunity occurs because of the loss of tolerance in the periphery (32–34). nTregs are able to suppress the proliferation of conventional αβ T cells. Our data showed that nTregs were able to greatly inhibit γδ T cell expansion and prevent splenomegaly. Interestingly, Tcons also had some ability to dampen splenomegaly, perhaps by competing for limited cytokines. Although Treg-deficient mice have a defect in T cell tolerance, the absence of nTregs in TCRβ^−/− mice does not result in spontaneous autoimmunity. This indicates that the absence of Tregs alone is not responsible for T cell hyperproliferation in TCRβ^−/−LAT^m/m mice; rather it is caused by the absence of Tregs in combination with strong self-reactivity or abnormal TCR signaling. The cell-intrinsic mechanisms by which weakened LAT signaling in αβ or γδ T cells causes this type of autoimmunity remain to be fully explored.

We thank the Duke University Cancer Center Flow Cytometry and DNA Sequencing facilities.

The authors have no financial conflicts of interest.