CD28 controls the development of innate-like CD8+ T cells by promoting the functional maturation of NKT cells

Authors

  • Mitra Yousefi,

    1. Institut National de la Recherche Scientifique-Institut Armand-Frappier, Université du Québec, Laval, Canada
    Search for more papers by this author
  • Pascale Duplay

    Corresponding author
    1. Institut National de la Recherche Scientifique-Institut Armand-Frappier, Université du Québec, Laval, Canada
    • Full correspondence: Dr. Pascale Duplay, Institut National de la Recherche Scientifique-Institut Armand-Frappier, 531 Boulevard des Prairies, Laval, QC H7V 1B7, Canada

      Fax: +1-450-686-5301

      e-mail: pascale.duplay@iaf.inrs.ca

    Search for more papers by this author

Abstract

NK T cells(NKT cells) share functional characteristics and homing properties that are distinct from conventional T cells. In this study, we investigated the contribution of CD28 in the functional development of γδ NKT and αβ NKT cells in mice. We show that CD28 promotes the thymic maturation of promyelocytic leukemia zinc finger+ IL-4+ NKT cells and upregulation of LFA-1 expression on NKT cells. We demonstrate that the developmental defect of γδ NKT cells in CD28-deficient mice is cell autonomous. Moreover, we show in both wild-type C57BL/6 mice and in downstream of tyrosine kinase-1 transgenic mice, a mouse model with increased numbers of γδ NKT cells, that CD28-mediated regulation of thymic IL-4+ NKT cells promotes the differentiation of eomesodermin+ CD44high innate-like CD8+ T cells. These findings reveal a previously unappreciated mechanism by which CD28 controls NKT-cell homeostasis and the size of the innate-like CD8+ T-cell pool.

Introduction

NK T cells (NKT cells) have unique phenotypic and functional characteristics when compared to conventional T cells. Within the NKT-cell population, there are distinct subsets including CD1d-restricted TCRαβ NKT cells (hereafter referred to as αβ NKT cells) and TCRγδ NKT cells (hereafter referred to as γδ NKT cells). These two subsets share common properties. They exhibit an “activated” phenotype, secrete cytokines upon primary stimulation, and most of them express highly restricted repertoires. The development of αβ NKT cells has been extensively studied and is well characterized (reviewed in [1]). In contrast, the development of γδ NKT cells is poorly defined. αβ NKT cells develop from CD4+CD8+ double-positive thymocytes and type I αβ NKT cells use a TCR composed of an invariant Vα14-Jα18 TCR-α chain that recognizes glycolipids presented by CD1d [2, 3]. Although not formally demonstrated by fate-mapping experiments, γδ NKT cells likely arise from thymo-cytes at the CD4CD8 double-négative stage similar to conventional γδ T cells. In C57BL/6 mice, most of the γδ NKT cells express Vγ1.1 and Vδ6.3 [4]. The selecting ligands of γδ NKT cells are unknown, but are likely expressed on other thymocytes rather than stromal cells since γδ NKT-cell development is dependent on the signaling lymphocytic activation molecule-associated protein pathway [4]. Moreover, our previous results [5] together with work from other groups [reviewed in [6]], have clearly established that attenuated TCR signaling promotes the development of γδ NKT-cell subsets.

Maturation of αβ NKT cells progresses through well-defined stages [7]. Recently, the phenotypic, functional, and transcriptional changes occurring during γδ NKT-cell maturation were characterized [8-10]. The functional developmental intermediates of αβ and γδ NKT cells in the thymus can be distinguished by expression of promyelocytic leukemia zinc finger (PLZF) and NK1.1 [8]. PLZF is a transcriptional factor that controls the development and the acquisition of effector properties of αβ and γδ NKT cells and is absent in conventional T cells [4, 6, 11, 12]. More importantly, high expression of PLZF confers the capacity of NKT cells to produce IL-4 [8].

In different mouse models, high level of IL-4 production by PLZF+ thymic NKT cells has been shown to drive innate-like CD8+ development through induction of Eomesodemin (Eomes) expression (reviewed in [13]). These Ag-naive CD8+ T cells have a memory-like phenotype and by their capacity to rapidly produce IFN-γ may act as first responders during an immune response. Therefore, regulation of the size of PLZF-expressing cell population might have important functional consequences. The level of PLZF expression is linked to the maturation stage of NKT cells [8, 9, 11, 12]. Although, high levels of Egr2 induced by TCR- and SLAMF6-mediated signaling is clearly important for PLZF induction [14, 15], the signals controlling the development of PLZF-expressing cells are incompletely defined [6]. Several factors and receptors selectively control the early maturation stages of postselected NKT cells [7, 16, 17]. In particular, the CD28/B7 interaction was shown to be essential for the development of αβ NKT cells at an early stage of maturation [18-20].

In this study, we analyze whether the absence of CD28 correlates with defects in the generation of both PLZF–expressing αβ and γδ NKT cells. Our results demonstrate that CD28 controls the number of PLZF+ cells and thereby modulates the functional capacity of NKT cells to produce IL-4 during thymic development. These findings describe a mechanism by which CD28 regulates the generation of Eomes+ memory-like CD8+ T cells.

Results

γδ NKT-cell development depends on CD28

CD28/B7 interaction regulates intrathymic expansion and differentiation of postselected αβ NKT cells [18, 19]. We predicted that the development of γδ NKT cells would also be affected by CD28 deficiency given the shared properties and developmental program of αβ and γδ NKT cells. To investigate whether CD28 was involved in γδ NKT-cell development, we compared the percentage and absolute numbers of γδ NKT cells in WT and Cd28−/− C57BL/6 mice using the expression of Vδ6.3 as a marker for γδ NKT cells. In the thymus, liver, and spleen, there was a significant decrease in both the number and frequency of Vδ6.3+ cells in Cd28−/− mice (Fig. 1A and B). As expected, CD28 deficiency did not affect the numbers of Vδ6.3 γδ thymocytes (Fig. 1A). PLZF, which is mainly expressed on NKT cells in mice, is a key factor for the development and effector function of these cells (reviewed in [6]). In the γδ lineage, PLZF is mainly expressed in Vδ6.3+ cells but is also detected in a small percentage of Vδ6.3 T cells [4, 21]. We compared the number of PLZF+ bona fide γδ NKT cells in WT and Cd28−/− mice. As shown in Fig. 2A, the number of PLZF+ γδ NKT cells is dramatically reduced in Cd28−/− mice. Moreover, as previously reported [4, 21], the vast majority of Vδ6.3+ cells expressed PLZF in WT C57BL/6 (Fig 2B). In contrast, in absence of CD28, only 52% of Vδ6.3+ cells are PLZF+ (Fig. 2B). Importantly, the number of PLZF+ Vδ6.3+ NKT cells is dramatically reduced in Cd28−/− mice whereas the number of PLZF Vδ6.3+ cells is less affected (Fig. 2C). Altogether these results demonstrate that CD28 specifically regulates the development of the PLZF-expressing γδ T-cell lineage.

Figure 1.

CD28 controls the number of γδ NKT cells in thymus, spleen, and liver. (A) CD8+-depleted thymocytes from 4–5-week-old age-matched or littermate WT and Cd28−/−(KO) mice were stained with antibodies against TCRγδ and Vδ6.3. Representative dot plots of eight pairs of mice are shown (left). Absolute numbers of Vδ6.3+ NKT cells and Vδ6.3 γδ T cells in individual thymi are indicated (right). (B) Splenocytes and liver cells from 4–5-week-old age-matched or littermate WT and CD28 KO were stained with antibodies against TCRγδ and Vδ6.3. The percentage of Vδ6.3+ T cells on gated TCRγδ T cells is indicated for six and seven pairs of mice. Each pair corresponds to an independent experiment. Statistical significance was determined with a two-tailed paired Student's t-test. ns: not significant.

Figure 2.

CD28 controls the pool size of PLZF+ γδNKT cells. (A) Thymocytes from 4–6-week-old C57BL/6 WT and Cd28−/− (KO) mice were depleted of CD8+ cells and stained for cell surface expression of TCRγδ and Vδ6.3 followed by intracellular staining for PLZF expression. Quantification of absolute numbers of PLZF+ γδ T cells is indicated. (B) Representative histograms show the PLZF expression of Vδ6.3+ γδ T cells in WT (thin line histogram), in Cd28−/− (thick line histogram) and as a negative control in WT Vδ6.3 γδ T cells (filled histogram). Bracketed lines next to the graphs indicate the percentages of PLZF+ within Vδ6.3+ thymocytes. The percentages of PLZF+ in Vδ6.3+ WT and Cd28−/ (KO) thymocytes for eight pairs of mice are shown (right). (C) Absolute numbers of PLZF+ and PLZF Vδ6.3+ WT and Cd28−/ (KO) thymocytes are indicated. Data are representative of eight pairs of mice pooled from five independent experiments. The data are shown as the mean + SD of all mice tested. Statistical significance was determined with a two-tailed paired Student's t-test.

CD28 controls the maturation of γδ NKT cells after CD24 downregulation

CD44, NK1.1, and CD24 are useful cell surface markers to distinguish the maturation steps of thymic αβ NKT cells, from stage 0 (CD24high CD44lowNK1.1), to stage 1 (CD24lowCD44lowNK1.1), stage 2 (CD24lowCD44high NK1.1), and stage 3 (CD24lowCD44high NK1.1+) [16]. Recent studies indicate that γδ NKT cells likely progress through similar developmental stages as αβ NKT cells in the thymus [8-10].

We performed CD24, CD44, and NK1.1 cell surface staining on Vδ6.3+ cells. Vδ6.3+ cells are largely CD24low in the thymus of WT mice (Fig. 3A). There was a significant increase in the fraction of CD24high γδ NKT cells in Cd28−/− mice compared with that in WT mice, mainly in the CD44low NK1.1 cell population (stage 0) (Fig. 3A and Supporting Information Fig. 1). The increased proportion of CD24high CD44lowNK1.1 Vδ6.3+ cells (stage 0) most likely reflected a partial block in the transition to the later maturation stages of Cd28−/− thymic γδ NKT cells. Analysis of CD24low Vδ6.3+ cells revealed an increase in the percentage of the CD44lowNK1.1 cell subset (stage 1) with a concomitant decrease in the percentage of the CD44high cell subsets (stages 2 and 3) in Cd28−/− mice compared to WT mice (Fig. 3B). These results indicate that the upregulation of CD44 in thymic Vδ6.3+cells is greatly dependent on CD28.

Figure 3.

CD28 regulates thymic maturation of γδ NKT cells. CD8+-depleted thymocytes from 4-week-old age-matched WT and Cd28−/− (KO) mice were stained for cell surface expression of TCRγδ, Vδ6.3, CD24, CD44, and NK1.1 followed by intracellular staining for PLZF expression. (A) Representative histograms show CD24 expression of Vδ6.3+ γδ T cells in WT (black line) and Cd28−/− (red line). Bracketed lines next to the graphs indicate the percentages of CD24high in WT and CD28KO Vδ6.3+cells. (B) Representative contour plots of CD44 and NK1.1 expression on gated CD24lowTCRγδ+Vδ6.3+ thymocytes are shown for WT and Cd28−/− (KO) mice. Percentages of cells in each quadrant are indicated. (C) Representative histograms show PLZF expression of WT (black line) and Cd28−/− (red line) Vδ6.3+ γδ T cells at stage 0 (CD24high CD44lowNK1.1), stage 1 (CD24lowCD44lowNK1.1), stage 2 (CD24lowCD44high NK1.1), and stage 3 (CD24lowCD44high NK1.1+). Bracketed lines next to the graphs indicate the percentages of PLZF+Vδ6.3+cells. (D) Quantification of the absolute number of WT and Cd28−/− (KO) PLZF+ Vδ6.3+cells in stages 0–3 of maturation. Data are shown as mean + SD of three pairs of mice, each pair examined in an independent experiment. Statistical significance was determined with an unpaired Student's t-test. (E) Cells from 3-week-old littermate WT and Cd28−/−(KO) mice were labeled with BrdU. Enriched-TCRγδ thymocytes were stained for TCRγδ, followed by PLZF and BrdU staining. BrdU incorporation is shown for gated PLZF+ TCRγδ+ thymocytes. Data are representative of two pairs of mice analyzed.

In αβ NKT cells, PLZF expression increases after positive selection, is highest in NK1.1CD44low cells, and decreases as differentiation proceeds to stage 3 [11, 12]. We analyzed the expression of PLZF at different stages of γδ NKT cell maturation (Fig. 3C). PLZF is induced at stage 0, is maximum at stages 1 and 2, and decreased at stage 3 (Fig. 3C). The developmental regulation of PLZF expression of WT γδ NKT cells is therefore similar to that of αβ NKT cells. In the absence of CD28, the percentage of Vδ6.3+ cells that are PLZF+ decreased at stages 1 and 2. Similar to WT cells, all the NK1.1+ Vδ6.3+ cells are PLZFdim at stage 3 (Fig. 3C). The absolute number of stage 0 (CD24high CD44lowNK1.1) PLZF+ Vδ6.3+ cells was comparable in Cd28−/− and Cd28+/+ mice (Fig. 3C and D). Because PLZF+ CD24high CD44low thymocytes mainly correspond to cells that have just undergone positive selection, we can conclude that positive selection of γδ NKT cells seems to occur normally in Cd28−/− mice. In the absence of CD28, there was a clear decrease in the number of Vδ6.3+ PLZF+ cells after stage 0 (Fig. 3D). Altogether, our data suggest that CD28 controls the maturation of γδ NKT cells.

PLZFhigh γδ NKT cells do not express NK1.1 and constitute the actively proliferating population [8]. The reduced number of PLZF+ γδ NKT cells in Cd28−/− mice was not a direct consequence of altered proliferative capacity in the thymus as a comparable percentage of PLZF+ γδ NKT cells incorporated BrdU after a 4 h pulse in vivo in WT and Cd28−/− mice (Fig. 3E).

Altogether, these results clearly indicate that CD28 is critical for further maturation and/or survival of CD24low positively selected γδ NKT cells and further substantiate the important role of CD28 in the control of NKT-cell thymic maturation. The high level of CD28 expression in PLZF+ γδ NKT cells is consistent with this hypothesis (Supporting Information Fig. 2).

CD28 controls LFA-1 upregulation in NKT cells

Recently, PLZF was shown to control the induction of high levels of LFA-1 on tissue-resident αβ NKT cells [22]. We, therefore, analyzed whether CD28 regulates LFA-1 cell surface expression on NKT cells. Remarkably, in the thymus and liver, CD28-deficient γδ NKT cells exhibited reduced levels of LFA-1 cell surface expression (Fig. 4). This reduction was specific to NKT cells as similar levels of LFA-1 were detected on CD28-deficient and WT conventional γδ T cells (Fig. 4). Similarly, we observed a reduced LFA-1 expression on CD1d tetramer+ αβ NKT cells in the liver of CD28-deficient compared to WT mice (Supporting Information Fig. 3). Thus, CD28 deficiency results in impaired upregulation of PLZF and LFA-1 expression in NKT cells.

Figure 4.

CD28 controls LFA-1 upregulation in γδ NKT cells. Quantification of LFA-1 median fluorescence intensity (MFI) of Vδ6.3+ γδ T cells (γδ NKT), Vδ6.3 γδ T cells (γδ) from liver and thymus of WT and Cd28−/− (KO) mice is shown. Data are representative of four to eight pairs of mice pooled from three independent experiments. Horizontal bars indicate the mean. Representative LFA-1 histograms on gated Vδ6.3+ γδ T cells from liver or thymus of WT (thin line histogram) and CD28 KO (thick line histogram) mice are shown (bottom). Statistical significance was determined with a two-tailed paired Student's t-test. ns: not significant.

CD28 is required for NKT-cell maturation by cell-intrinsic mechanisms

To determine whether the development of γδ NKT cells is dependent on CD28 in a cell autonomous manner, we generated mixed BM chimeras. After lethal irradiation, CD45.1/2 WT mice were reconstituted with a 1:1 ratio of CD45.1 BM cells from WT mice and CD45.2 BM cells from WT or Cd28−/− mice. After 6 weeks, γδ NKT thymocytes from the mixed BM chimeras were analyzed for expression of cell surface markers (Fig. 5). We observed a slight increase in the percentage of immature CD44lowNK1.1 (stages 0–1) γδ NKT-cell subset, while the percentage of CD44highNK1.1 and CD44highNK1.1+ stages 2–3, respectively, γδ NKT-cell subsets were significantly reduced in the Cd28−/− mice (Fig. 5A and B). More importantly, the proportion of PLZF+ αβ and γδ NKT thymocytes was significantly reduced in the absence of CD28 (Fig. 5C). These findings demonstrate that the impaired development of NKT cells observed in Cd28−/− mice could not be rescued with WT BM cells. Thus, the defect in the development of Cd28−/− PLZF+ NKT cells is cell-autonomous.

Figure 5.

Cell-intrinsic NKT-cell defects in CD28 KO mice. CD45.1+ WT (WT.1) and CD45.2+ WT or Cd28−/− (WT.2 or KO.2) bone marrow cells were mixed at a 1:1 ratio and transferred into lethally irradiated CD45.1/2 hosts. (A) CD8+-depleted thymocytes were stained for CD45.1, CD45.2, TCRγδ, Vδ6.3, CD44, and NK1.1. The relative proportions of CD45.1 and CD45.2 thymocytes in chimeras reconstituted with a mix of WT CD45.1 and WT CD45.2 BM (WT.1/WT.2; left) and a mix of WT CD45.1 and Cd28−/− CD45.2 BM are shown (WT.1/KO.2; right). Representative contour plots show CD44 and NK1.1 expression on gated Vδ6.3+ CD45.1 and CD45.2 thymocytes. Data are representative of six individual chimeras. (B) The percentage of thymic Vδ6.3+cells at stages 0–1 (CD44lowNK1.1), stage 2 (CD44highNK1.1), and stage 3 (CD44highNK1.1+) is indicated for six chimeras WT.1/WT.2 and six chimeras WT.1/KO.2. Data are shown as mean + SD of six individual chimeras from two independent experiments. (C) Quantifications of the proportion of PLZF+ γδ and αβ T cells are indicated for six chimeras. Data are shown as mean + SD. Statistical significance was determined with a paired Student's t-test.

CD28-mediated regulation of IL-4+ NKT cells promotes the differentiation of innate-like CD8+ T cells

The ability of NKT cells to rapidly secrete large amounts of cytokines is dependent on their stage of maturation [23, 24]. In particular, high PLZF expression confers the capacity of NKT cells to produce high levels of IL-4 [8]. Since CD28 regulates the number of PLZF-expressing cells, we predicted that CD28 deficiency would interfere with the production of IL-4 by NKT cells. As previously reported [8], the majority of γδ NKT cells that produced high levels of IL-4 after stimulation with PMA and ionomycin were in the PLZFhigh cell population (Supporting Information Fig. 4A). In Cd28−/− mice, we observed a 3- and 12-fold decrease in the proportion of αβ and γδ PLZF+ IL-4+ cells, respectively (Fig. 6A). The negative effect of CD28 on IL-4 production is due to the decreased frequency of PLZF+ NKT cells (CD8-depleted thymocytes; 0.74%, Cd28+/+ and 0.29%, Cd28−/− for αβ NKT cells; 0.10%, Cd28+/+ and 0.02%, Cd28−/− for γδ NKT cells; Fig 6B) and to the decreased percentage of IL-4-producing cells among PLZF+ NKT cells, especially for γδ NKT cells (35%, Cd28+/+ and 15% Cd28−/−, Fig. 6B). CD28 deficiency affected more IFN-γ/IL-4 double-producing than IL-4 single-producing PLZF γδ NKT cells (Supporting Information Fig. 4A). Given that CD24high produces less IL-4 than CD24low PLZF+ NKT cells (Supporting Information Fig. 4B), the reduced production of IL-4 among PLZF+ γδ NKT cells in Cd28−/− C56BL/6 mice is due to the higher proportion of functionally immature CD24high PLZF+ γδ NKT cells compared to Cd28+/+ mice (12% in Cd28+/+ versus 31% in Cd28−/− mice, Fig. 3D and Supporting Information Fig. 4B). To further substantiate the CD28-mediated regulation of IL-4 production in NKT cells, we used downstream of tyrosine kinase-1 (Dok-1) transgenic mice. In these mice, there was an expanded γδ NKT-cell population due to Dok-1 overproduction in thymocytes [5]. In Dok-1 transgenic mice, the majority of PLZF+ thymocytes were γδ T cells, whereas in C57BL/6 WT mice, the majority of thymic PLZF+ NKT cells corresponded to αβ NKT cells, (Fig. 6B and D). In addition, there was a 79-fold increase in the percentage of PLZF+ IL-4+ γδ NKT cells compared with that in C57BL/6 WT mice, while the percentage of PLZF+ IL-4+ αβ NKT cells was comparable with that in C57BL/6 mice (Fig. 6A and C). Maturation of PLZF+ NKT cells in Dok-1 transgenic mice occurred normally, except for the lower percentage of stage 0 (CD24high) PLZF+ Vδ6.3+ cells compared to C57BL/6 mice (Supporting Information Fig. 4C and D). Moreover, CD28 regulated the same maturation steps in C57BL/6 and Dok-1 transgenic mice although with quantitative differences (Supporting Information Fig. 4C–E). Importantly, as shown for C57BL/6 mice, in the absence of CD28, there was a significant reduction in the frequency of PLZF+ IL-4+ NKT cells in Dok-1 transgenic mice which was mainly due to the decreased proportion of αβ and γδ PLZF+ NKT cells (0.12%, Cd28+/+ and 0.06%, Cd28−/ for αβ NKT cells; 0.65%, Cd28+/+ and 0.18%, Cd28−/− for γδ NKT cells, Fig. 6 C and D). CD28 deficiency in Dok-1 transgenic mice did not alter significantly the ratio of immature CD24high/mature CD24low PLZF+ γδ NKT cells and therefore did not change the overall capacity of PLZF+ γδ NKT cells to produce IL-4 or IFN-γ (Fig. 6D and Supporting Information Fig. 4F).

Figure 6.

CD28 controls the number of PLZF+IL-4+ NKT cells. CD8+-depleted thymocytes from WT and Cd28−/ (KO) C56BL/six mice or total thymocytes from WT and Cd28−/− (KO) Dok-1 transgenic mice were stimulated with PMA and ionomycin. After stimulation, cells were stained for surface expression of TCRγδ and stained for PLZF and IL-4 intracellular expression. (A) Quantification of the proportion of PLZF+ IL-4+ TCRγδ (γδ) or TCRαβ (αβ) thymocytes from WT and Cd28−/ (KO) C56BL/six mice is shown. Values were obtained by multiplying the ratio of IL-4+/PLZF+ by the percentage of PLZF+ αβ or γδ thymocytes divided by the cell-enrichment factor after CD8 depletion and are shown as mean + SD of five pairs of mice pooled from three independent experiments. (B) Representative dot plots show TCRγδ and PLZF expression of CD8+ cell-depleted thymocytes in WT and Cd28−/−(KO) C56BL/six mice. A minimum of 3 × 106 events were acquired. Representative histograms below show the proportion of IL-4+cells in PLZF+ αβ or γδ thymocytes (black line histogram) and as a negative control in WT PLZF αβ or γδ thymocytes (filled histogram). Data shown are representative of five pairs of mice pooled from three independent experiments. (C) Quantification of the proportion of PLZF+ IL-4+ γδ or αβ thymocytes from WT and Cd28−/− (KO) Dok-1 transgenic mice is shown. Values were obtained by multiplying the ratio of IL-4+/PLZF+ by the percentage of PLZF+ αβ or γδ thymocytes and are shown as mean + SD of seven pairs of mice pooled from four independent experiments. (D) Representative dot plots show TCRγδ and PLZF expression of total thymocytes in WT and Cd28−/−(KO) Dok-1 Tg mice. Representative histograms below show the proportion of IL-4+cells in PLZF+ αβ or γδ thymocytes (black line histogram) and as a negative control in WT PLZFαβ or γδ thymocytes (filled histogram). Data are representative of seven pairs of mice pooled from four independent experiments. Statistical significance was determined with a two-tailed paired Student's t-test. ns: not significant.

The generation of Eomes+ CD44high memory-like CD8+ T cells has been shown to be dependent on IL-4 produced by PLZF+ NKT cells in several mouse models [13, 25-28]. The percentage of CD8+ thymocytes that were Eomes+ CD44high was reproducibly lower in Cd28−/− than in Cd28+/+; C57BL/6 mice (1.75%, Cd28+/+; 0.46%, Cd28−/−, Fig. 7A). Moreover, the percentage of CD44high that produced IFN-γ following PMA plus ionomycin was reduced in Cd28−/− mice (Fig. 7A). However, the difference did not reach statistical significance most likely because of the small number of innate-like CD8+ T cells in C57BL/6 mice. To confirm that CD28-mediated regulation of PLZF+IL-4+ NKT cells participates in the generation of innate-like CD8+ T cells, we analyzed Dok-1 transgenic mice, a mouse strain that generates larger numbers of PLZF+ NKT cells than C57BL/6 mice. In these mice, a high percentage of the CD8+ thymocytes have an innate-like phenotype induced by an increased frequency of IL-4-producing γδ NKT cells (Fig. 7B and [5]). As shown in Figure 7B, 50% of the CD8+ T cells from 4-week-old mice, expressed Eomes, a transcription factor upregulated in innate-like CD8+ T cells [13]. A high percentage of these Eomes+ CD8+ cells expressed markers associated with activated/memory phenotype such as CD44, CXCR3, and CD122 (Fig. 7C). In addition, there was an increase in the expression levels of CD124 in the CD8+ T cells from Dok-1 transgenic mice compared with those in C57BL/6 mice that corresponded to the specific upregulation of CD124 in Eomes+ cells compared with that in Eomes cells (Fig. 7D). Since IL-4 signaling leads to increased surface expression of CD124 [29], this result suggests that CD8+ T cells have been stimulated with IL-4 during their development in the thymus of Dok-1 transgenic mice. Finally, the increased expression of Eomes correlated with enhanced production of IFN-γ after ex vivo stimulation (Fig. 7E). Remarkably, the reduced number of PLZF+ IL-4+ NKT cells in Cd28−/− Dok-1 transgenic mice compared to WT Dok-1 transgenic mice correlated with the reduced number of innate-like CD8+ thymocytes (Fig. 7B–E). It should be noted that the remaining higher frequency of IL-4+ NKT cells in Cd28−/ Dok-1 transgenic mice compared to C57BL/6 mice (8.9% Cd28−/− Dok-1 Tg and 4.1% WT, Fig. 6A and C) led to the development of a higher percentage of innate-like CD8+ T cells. Collectively, these results suggest that CD28 positively contributes to the generation of innate-like CD8+ T cells through the control of IL-4+ PLZF+ NKT-cell development.

Figure 7.

CD28 controls the number innate-like CD8+ T cells in the thymus. (A) Thymocytes from WT and CD28KO C56BL/6 mice were depleted for CD4+ cells. Representative contour plots of the expression of CD44 and Eomes in CD8+ thymocytes (left) and the expression of CD44 and IFN-γ in CD8+ thymocytes after 4 h of stimulation with PMA and ionomycin (right) are shown. (B) Representative histograms show the expression of Eomes in CD8+ thymocytes from Dok-1 Tg CD28WT (thin line histogram), Dok-1 Tg CD28KO (thick line histogram), and in CD28WT C56BL/6 mice (filled histogram). The percentages of Eomes+ fraction of CD8+ cells in Dok-1 Tg, Dok-1 Tg CD28KO, and C56BL/6 mice are shown as mean ± SD of three pairs of mice. (C) Representative contour plots of the expression of Eomes and CD44/CXCR3/CD122 in CD8+ Dok-1 Tg, Dok-1 Tg CD28KO, and C56BL/6 thymocytes are shown. Percentages of Eomes+ CD44+ (top), Eomes+ CXCR3+ (middle), and Eomes+ CD122+ (bottom) in CD8+ Dok-1 Tg, Dok-1 Tg CD28KO, and C56BL/6 thymocytes are shown. Data are also shown as mean ± SD of three pairs of mice. (D) The CD124 median fluorescence intensity (MFI) of total CD8+, Eomes+ and Eomes Dok-1 Tg, Dok-1 Tg CD28KO, and C56BL/6 thymocytes is shown as mean ± SD of three pairs of mice. (E) Contour plots represent the expression of Eomes and IFN-γ in CD8+ Dok-1 Tg, Dok-1 Tg CD28KO thymocytes after 4 h of stimulation with PMA and ionomycin. Percentages of IFN-γ+ CD8+ Dok-1 Tg and Dok-1 Tg CD28KO thymocytes are also depicted as mean ± SD of three pairs of mice. All data shown are from one experiment. Statistical significance was determined with an unpaired Student's t-test. ns: not significant.

Discussion

This study investigates the role of CD28 in the regulation of γδ NKT-cell development. We demonstrate for the first time a critical role of CD28 in governing the generation of PLZF- and LFA-1high-expressing NKT cells. These findings have important consequences on the homeostasis and function of NKT cells.

Our results show that upregulation of LFA-1 within NKT cells in the thymus is dependent on CD28. LFA-1 is believed to play an essential role in tissue-specific cell migration and retention of NKT cells in the liver [30, 31]. Thus, it is possible that impaired LFA-1 expression contributes in part to the reduced NKT-cell number in the liver of Cd28−/− mice. Additional experiments will be required to directly test this hypothesis. The signal transduction pathway involved in the regulation of LFA-1 expression initiated after CD28 stimulation may be mediated in part by WASp. Indeed, WASp was shown to promote LFA-1 high expression on αβ NKT cells [32] and to regulate actin remodeling required for CD28 signaling [33].

We also demonstrate that CD28 controls the number of thymocytes that express high levels of PLZF and produce large amount of IL-4 following in vitro stimulation with PMA and ionomycin. Constitutive secretion of IL-4 by immature NK1.1 thymocytes has been recently demonstrated in vivo [34]. It is tempting to speculate that developing thymocytes producing IL-4 in the steady state correspond to the PLZF+ IL-4+ thymocytes described in this study. Importantly, previous reports demonstrated that PLZF+ NKT cells are the only cells in the thymus secreting IL-4 and have the capacity to promote the differentiation of innate-like CD8+ T cells [13]. It has been proposed that the low abundance of PLZF+ NKT-cell population in the thymus limits the production of IL-4 and thereby the number of innate-like CD8+ T cells generated in the thymus [35]. A prediction for this hypothesis is that CD28 controls the generation of innate-like CD8+ T cells by promoting the development of thymocytes that produce large amount of IL-4. In this study, we verified this prediction in C57BL/6 and Dok-1 transgenic mice where, in the absence of CD28, a decreased number of PLZF+ IL-4+ NKT cells correlates with a reduced number of innate-like CD8+ T cells generated in the thymus. It should be emphasized that in Dok-1 transgenic mice, innate-like CD8+ T-cell development is strictly dependent on the presence of PLZF+ NKT cells since we have previously shown that signaling lymphocytic activation molecule-associated protein deficiency in these mice led to a complete absence of innate-like CD8+ thymocytes [5]. CD28 requirement for the generation of innate-like CD8+ T cells was previously reported in Itk−/ mice (where Itk is IL-2 inducible T-cell kinase) [36]. Like Dok-1 transgenic mice, Itk-deficient mice have an expanded population of PLZF+ cells that are mostly γδ NKT cells [26]. Moreover, the development of innate-like CD8+ T cells in Itk−/− mice is attributable to a cell-extrinsic effect dependent on IL-4 [26]. Therefore, similar to our present findings in Dok-1 transgenic mice, we propose that the CD28-dependent development of innate-like CD8+ T cells in Itk−/− mice is an indirect effect of CD28 on the maturation of γδ NKT cells. The fact that in Itk−/− mice, the requirement of CD28 for full development of innate cell characteristics occurred upon selection on hematopoietic cells and not on the thymic stroma is in agreement with this interpretation. However, CD28 signaling may also synergize with IL-4 to induce the innate-like phenotype in CD8+ T cells by a cell-intrinsic mechanism.

Mice with different relative proportions of αβ and γδ NKT-cell population have been previously characterized [13]. In several C57BL/6 mutant mice including Dok-1 transgenic, Id3−/− and Itk−/− mice, it is likely the expanded PLZF+ γδ NKT-cell population that promotes the IL-4-mediated generation of innate-like CD8+ T cells [5, 21, 25, 26]. By contrast, in BALB/c mice, the PLZF+ NKT cells that mediate the memory conversion of thymic CD8+ T cells are αβ CD1-d restricted NKT cells [26]. Additional experiments would be required to determine the role of CD28 in the relative contribution of αβ and γδ NKT cells to the development of thymic innate-like CD8+ T cells.

Reduced TCR signaling strength such as in Itk−/− or Dok-1 transgenic mice resulted in a dramatic expansion of γδ NKT cell subset [5, 37]. In contrast, reduced CD28 signal led to a decreased number of γδ NKT cells. This suggests that CD28, in γδ NKT cells, initiates downstream signaling events different from amplifying TCR-initiated signals.

CD28 does not appear to play a role in positive selection of γδ NKT cells as normal numbers of CD24high PLZF+ γδ NKT cells were present in Cd28−/ mice. Although CD28 deficiency does not affect the proliferative capacity of NKT cells, it may lead to an increased susceptibility to apoptosis of immature NKT-cell populations in the thymus. Alternatively and not exclusively, CD28 may modulate the expression levels of selected genes that are important for further differentiation of immature NKT cells. Importantly, we clearly show that CD28 specifically affects the development of PLZF-expressing Vδ6.3+ T cells and seems to be dispensable to direct cells into the non-PLZF-expressing Vδ6.3+ T-cell lineage. Additional experiments will be required to determine the precise mechanisms by which CD28 regulates the survival and/or maturation of PLZF+ γδ NKT cells.

In summary, our findings highlight the critical role of CD28 in the maturation and function of NKT cells. We demonstrate that CD28 regulates the generation of Eomes+ CD44high innate-like CD8+ T cells in part by controlling the size of the PLZF+ IL-4+ pool in the thymus. Innate-like CD8+ T cells appear to contribute to early host defense against pathogens by rapidly producing IFN-γ in an Ag-independent [38-40] and Ag-dependent manner [41]. Hence, CD28 is among others an important regulator of the innate immune response by controlling the development of NKT cells and innate-like CD8+ T cells.

Materials and methods

Mice

C56BL/6 (CD45.2) WT and Cd28−/− mice were obtained from the Jackson Laboratories. C56BL6.SJL (CD45.1) mice were kindly provided by A. Lamarre (INRS-Institut Armand-Frappier, Laval, Canada). Dok-1 transgenic mice on a C57BL/6 genetic background (Tg82) were previously described [5]. All mice were maintained in our pathogen-free animal facilities (INRS-Institut Armand-Frappier, Laval, Canada). Mice were manipulated in strict accordance to protocols 0910–01 and 1201–02, approved by the institutional animal care committee of the INRS-Institut Armand-Frappier. This protocol respects guidelines on good animal practice provided by the Canadian Council on animal care.

Antibodies and reagents

PE-anti-CD4 or allophycocyanin-anti-CD4 (clone GK1.5), FITC-anti-CD8α or PE-anti-CD8α (clone 53–6.7), PE-anti-CD44 or allophycocyanin-anti-CD44 (clone IM7), biotinylated anti-CD3ε or allophycocyanin-anti-CD3ε (clone 145–2C11), FITC-anti-TCRβ or allophycocyanin-anti-TCRβ (clone H57–597), biotinylated anti-TCRγδ or FITC-anti-TCRδγ (clone GL3), PerCp-Cy5.5-anti-CD45.1 (clone A20), PF-anti-CD28 (clone 37.51), Alexa647-anti-Eomesodermin (clone Dan11mag), PE-anti-IL-4 (clone 11B11), FITC-anti-CD62L (clone MEL-14), and FITC-anti-CD24 (clone 30-F1) were purchased from eBioscience. PerCP-anti-CD8 (clone 53–6.7) PE-Vδ6.3/2 (clone 8F4H7B7) and Streptavidin PerCP were purchased from BD Biosciences. Brilliant violet 421TM anti-CD3ε (clone 145–2C11), allophycocyanin-anti-CD62L (clone MEL-14), allophycocyanin-anti-NK1.1 or allophycocyanin/Cy7-anti-NK1.1 (clone PK136), Streptavidin allophycocyanin or Brilliant violet 421TM, PerCp-Cy5.5-anti-LFA-1 (clone H155–78) and Pacific BlueTM anti-CD45.2 (clone 104) were purchased from Biolegend. Alexa647-anti-PLZF (clone 9E12) Abs were generated at the Memorial Sloan-Kettering Cancer Center (MSKCC) antibody facility. Allophycocyanin-CD1d tetramers unloaded or loaded with PBS-57 were supplied by the National Institutes of Health Tetramer facility.

Cell preparations, NKT-cell enrichments, and flow cytometry

Cell suspensions from thymus, spleen, and liver were prepared as previously described [5]. Red blood cells were removed using ACK lysis buffer (150 mM NH4Cl, 1 mM KHCO3, and 0.1 mM Na2EDTA, pH 7.2). For analysis of γδ NKT thymocytes from C57BL/6 mice, thymocytes were depleted of CD8+ cells using anti-CD8α antibody-coated magnetic beads from Miltenyi Biotec (Bergish Gladbach, Germany) according to the manufacturer's protocols. Alternatively, cells were labeled with biotinylated anti-TCRγδantibodies and purified with anti-biotin-coated magnetic beads from Miltenyi according to the manufacturer's protocols.

Cells were stained as described before [5]. Viable cells were acquired on a Fortessa Flow cytometer (Becton Dickinson) and analysis was performed using FlowJo software (Tree Star).

Intracellular IL-4 detection

For intracellular IL-4 detection, cells were stimulated with PMA (100 ng/mL) and ionomycin (1 μg/mL) in the presence of GolgiPlug (BD Pharmingen) for 4 h at 37°C. Surface and intracellular antigens were stained sequentially using a fixation and permeabilization kit from BD Biosciences according to the manufacturer's instructions.

BrdU incorporation

Mice were injected i.p. with 1 mg BrdU (in PBS). After 4 h, thymocytes were stained for cell surface markers, followed by fixation, permeabilization, and intracellular staining for BrdU following the manufacturer's instructions (BD Biosciences).

Bone marrow chimeras

Chimeras were generated by reconstituting lethally irradiated (950 rad) C57BL/6 CD45.1/CD45.2 mice with 107 BM cells from CD45.2 WT or Cd28−/− mice mixed 1:1 with BM from CD45.1 WT mice. Mice were analyzed 6 weeks after reconstitution.

Statistical analysis

All statistical analyses were performed with prism software (Graphpad). Two groups were compared using unpaired or paired Student's t-test when appropriate. Differences were considered to be statistically significant when p < 0.05. Graph results are shown as the mean ± SD.

Acknowledgements

We thank Simona Stäger for expert assistance with the generation of chimeric mice, the National Institutes of Health tetramer facility for providing the CD1d tetramer, Isabelle Meunier for help with mouse work, and Claude Daniel, Sylvie Fournier, Krista Heinonen, Alain Lamarre, and Simona Stäger for critical reading of the manuscript. Gilles Besin did initial experiments with Cd28−/− mice. This work was supported by grants from the Canadian Institutes of Health Research (CIHR) to P.D. M.Y. was partly supported by a studentship from the Fondation Armand-Frappier.

Conflict of interest

The authors declare no financial or commercial conflict of interest.

Abbreviations
Dok-1

downstream of tyrosine kinase-1

Eomes

eomesodermin

NKT cell

NK T cell

PLZF

promyelocytic leukemia zinc finger

Ancillary