γδ T Cells Acquire Effector Fates in the Thymus and Differentiate into Cytokine-Producing Effectors in a Listeria Model of Infection Independently of CD28 Costimulation

Both antigen recognition and CD28 costimulation are required for the activation of naïve αβ T cells and their subsequent differentiation into cytokine-producing or cytotoxic effectors. Notably, this two-signal paradigm holds true for all αβ T cell subsets, regardless of whether they acquire their effector function in the periphery or the thymus. Because of contradictory results, however, it remains unresolved as to whether CD28 costimulation is necessary for γδ T cell activation and differentiation. Given that γδ T cells have been recently shown to acquire their effector fates in the thymus, it is conceivable that the contradictory results may be explained, in part, by a differential requirement for CD28 costimulation in the development or differentiation of each γδ T cell effector subset. To test this, we examined the role of CD28 in γδ T cell effector fate determination and function. We report that, although IFNγ-producing γδ T (γδ-IFNγ) cells express higher levels of CD28 than IL-17-producing γδ T (γδ-17) cells, CD28-deficiency had no effect on the thymic development of either subset. Also, following Listeria infection, we found that the expansion and differentiation of γδ-17 and γδ-IFNγ effectors were comparable between CD28+/+ and CD28−/− mice. To understand why CD28 costimulation is dispensable for γδ T cell activation and differentiation, we assessed glucose uptake and utilization by γδ T cells, as CD28 costimulation is known to promote glycolysis in αβ T cells. Importantly, we found that γδ T cells express higher surface levels of glucose transporters than αβ T cells and, when activated, exhibit effector functions over a broader range of glucose concentrations than activated αβ T cells. Together, these data not only demonstrate an enhanced glucose metabolism in γδ T cells but also provide an explanation for why γδ T cells are less dependent on CD28 costimulation than αβ T cells.


Introduction
The current paradigm for the activation of naïve ab T cells and their subsequent differentiation into cytokine-producing or cytotoxic effectors is that two signals are required: one through the T cell antigen receptor (TCR) and the other through the costimulatory molecule, CD28. These two signals act together not only to prevent anergy [1][2][3], but also to promote cell survival [4], to activate the switch to glycolysis [5,6], to stabilize cytokine gene transcripts [7,8], and to regulate alternative splicing [9].
While most ab T cells differentiate into effectors in the periphery, some ab T cells subsets, such as Natural Killer T (NKT) cells and regulatory T (T reg ) cells, acquire their effector functions in the thymus [10][11][12][13][14]. Despite the change in their site of differentiation, NKT and T reg cells require CD28 costimulatory signals during their development in the thymus. Specifically, NKT cells require CD28 costimulation, following their selection, to expand and mature [15,16], whereas T reg cells require CD28 costimulation to activate the T reg genetic program, which includes the expression of genes encoding Foxp3, GITR and CTLA-4 [17].
Due to conflicting results, it is unclear whether CD28 costimulation is also required for the activation and differentiation of cd T cells. However, as the vast majority of these studies were conducted at a time when it was not known that cd T cells have distinct effector fates and that acquisition of these fates occurs in the thymus [18,19], it is possible that the conflicting results may be explained, in part, by each cd T cell effector subset having a different requirement for CD28 costimulation, either during their development in the thymus or during their differentiation into effectors in the periphery. For this reason, we decided to reevaluate the role of CD28 costimulation in the generation of cd T cell effectors.
Here, we report that CD28 is differentially expressed between IFNc-producing cd T (cd-IFNc) cells and IL-17-producing cd T (cd-17) cells, with cd-IFNc expressing 2 to 3-fold more CD28 than cd-17 cells. Despite this difference in expression, CD28 costimulation was not required to generate thymic and peripheral cd-IFNc and cd-17 cells. Surprisingly, CD28 signaling was required to generate wild-type numbers of cd thymocytes and cd T cells. The reduction in the number of cd lineage cells in CD28 2/2 mice was not due to defects in either cd lineage commitment or cd thymocyte proliferation and survival, but instead was due to impaired proliferation of thymic progenitors. We also assessed the ability of CD28-deficient cd T cells to differentiate into cytokineproducing effectors during infection with Listeria monocytogenes (Lm), and observed no difference in cd T cell expansion and differentiation between infected CD28 +/+ and CD28 2/2 mice. Thus, these data not only indicate that CD28 is dispensable for cd effector T cell development and differentiation but also highlight significant differences in the molecular requirements for the generation of ab and cd T cell effectors.

Ethics Statement
All research involving animals has been conducted according to the relevant national and international guidelines with respect to husbandry, experimentation and welfare. Mouse protocols were approved by the State University of New York (SUNY) Upstate Medical University Committee on the Humane Use of Animals (CHUA protocol numbers 262 and 281).

Flow Cytometric Analysis
Flow cytometric analysis for surface antigen expression was performed by pre-incubating cells with the anti-CD16/CD32 mAb for at least 10 minutes to block non-specific binding of immunoglobulins to Fc receptors, followed by staining with fluorochrome-conjugated mAbs against various surface antigens. For CD28 surface staining, PE-conjugated anti-CD28 mAb was used, and samples were run on a Becton Dickinson LSR II fitted with a 568 nm laser. For GLUT1 and GLUT3 surface staining, nonspecific binding was blocked by incubating cells with a mixture of donkey serum and the anti-CD16/CD32 mAb for at least 10 minutes. Cells were then stained with either the anti-GLUT1 or anti-GLUT3 Ab for 20 minutes, followed by staining with a fluorochrome-conjugated donkey anti-goat secondary antibody (Invitrogen, Carlsbad, CA, USA) for another 20 minutes.
Ki-67 and Annexin V staining was performed according to the manufacturer's instructions (BD Biosciences). Intracellular staining for IL-17A and IFNc was performed by first fixing cells in a final concentration of 1.5% formaldehyde for 10 minutes at 37uC. Fixed cells were then stained for surface antigens, permeabilized with Perm/Wash Buffer (BD Biosciences) for 20 minutes at 4uC, and then stained with mAbs against IL-17A and IFNc.
For all experiments, 0.5 to 2.5 6 10 6 cells were acquired on a BD LSR II or a BD LSRFortessa using FACSDiva software (BD Immunocytometry Systems, San Jose, CA USA). Data analysis was performed using FlowJo software (Tree Star, Inc., San Carlos, CA, USA). Dead cells were excluded from analysis based on forward and side scatter profiles.

Cell Separation
cd T cells were purified by negative selection from the peripheral lymph nodes (pLNs; inguinal, axillary, brachial and cervical) of CD28 +/+ cdTCR Tg, CD28 2/2 cdTCR Tg and TCRa 2/2 mice using the MACSH magnetic bead separation system (Miltenyi Biotech, Auburn, CA, USA). First, cells were stained for 10 minutes at 4uC with a panel of FITC-labeled mAbs containing anti-CD19, anti-TCRb, anti-CD4, anti-CD8, anti-CD11b, anti-IA b and anti-DX5 mAbs. Next, cells were washed, incubated with anti-FITC MACS beads for 15 minutes at 4uC, and then separated on an autoMACS cell separator, according to manufacturer's directions. Typically, the purity of the resulting DN cd T cell populations from cdTCR Tg mice was $99%, and that of DN cd T cells from TCRa 2/2 mice was $85%.
ab T cells were purified from the pLNs of CD28 +/+ mice using the MACSH magnetic bead separation system described above, except that the following panel of FITC-conjugated antibodies was used: anti-CD19, anti-TCRcd, anti-CD44, anti-CD11b, anti-IA b and anti-DX5 mAbs. The purity of the resulting cell ab T cell population was typically $98%.

In Vitro Stimulation of cd T Cells
Purified cd T cells were resuspended in RPMI 1640 supplemented with non-essential amino acids, L-glutamine, HEPES, sodium pyruvate, and penicillin/streptomycin (all from Invitrogen) in addition to 10% FBS (Mediatech, Inc., Manassas, VA, USA), plated onto 5 mg/ml of immobilized hamster isotype control, 5 mg/ml of immobilized anti-CD3 mAb or 5 mg/ml each of immobilized anti-CD3 and anti-CD28 mAbs, and then cultured for 16 hours at 37uC. Cells were treated with Brefeldin A and Monensin (eBioscience) 5 hours prior to fixation, permeabilization, and intracellular staining with mAbs against IL-17A and IFNc.

Bacteria and Bacterial Infection of Mice
Listeria monocytogenes (Lm; strain 10403S) and Lm expressing the stable recombinant protein Lm ActA-Ub-acidic polymerase (PA)-SIINFEKL-FLAG [22] were grown as previously described [23]. Briefly, bacteria were grown overnight in 1 mL Brain Heart Infusion (BHI) broth (Teknova, Hollister, CA, USA) supplemented with 200 mg/mL streptomycin (Fisher Scientific, Pittsburgh, PA, USA). Overnight cultures were used at a 1:10 (vol/vol) dilution to inoculate fresh BHI broth supplemented with 200 mg/mL streptomycin. Bacteria were incubated in an orbital shaker for 1-2 hours at 37uC to mid-log phase, and the bacterial concentration was determined by measuring absorbance at 600 nm. Mice were infected with 3610 4 CFU Lm by i.p. inoculation.
To assess cd T cell effector function following Lm infection, spleens were harvested on days 1 and 5 post infection and processed into single cell suspensions in RPMI 1640, supplemented with non-essential amino acids, L-glutamine, HEPES, sodium pyruvate, penicillin/streptomycin, and 10% FBS. 3610 6 splenocytes were then plated onto 1 or 5 mg/ml of immobilized anti-TCRcd mAb (UC7-13D5) or 5 mg/ml of immobilized hamster isotype control and cultured at 37uC for 6 hours in the presence of Brefeldin A and Monensin before intracellular staining with mAbs against IL-17A and IFNc.
To assess CD8 + ab T cell effector function following Lm infection, spleens were harvested on day 7 and processed into single cell suspensions in supplemented RPMI 1640 as above. 2.5610 6 splenocytes were cultured with 1 mM SIINFEKL peptide or 1 mM of an irrelevant peptide (both from Genscript, Piscataway, NJ, USA) at 37uC for 5 hours in the presence of Brefeldin A before intracellular staining with a mAb against IFNc.
For bacterial colony counts, livers were harvested at day 5 post infection and then processed into a single cell suspension using the gentleMACS Dissociator (Miltenyi). The manufacturer-provided protocol for preparation of single-cell suspensions from mouse liver (gentleMACS-Liver) was followed to step 11. Bacteria were liberated from eukaryotic cells by the addition of 0.1% Triton-X 100 at a 1:1 (vol/vol) dilution. Ten-fold serial dilutions of the lysates were plated onto BHI plates supplemented with 200 mg/ mL streptomycin. Bacteria colonies were enumerated the following day.

Glucose Uptake
Purified ab and cd T cells were cultured at 37uC in glucose-free RPMI (Invitrogen) supplemented with 30 mM 2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl) Amino)-2-Deoxyglucose (2-NBDG) (Invitrogen) at 5610 5 cells per well, in a 48-well plate, in the presence or absence of 1 mg/mL plate-bound anti-CD3 mAb. T cells were harvested from the plate at various time points, and the amount of 2-NBDG taken up by the cells was measured by flow cytometry.

Measuring Glucose-dependent Proliferation and Cytokine Production
Purified cd T cells were labeled with 5-carboxyfluorescein diacetate succinimidl ester (CFSE), according to the manufacturer's instructions (eBioscience). Cells were cultured in glucose-free RPMI supplemented with non-essential amino acids, L-glutamine, HEPES, sodium pyruvate, penicillin/streptomycin, 10% dialyzed FBS (Invitrogen) and various concentrations of D-glucose at 3610 5 cells per well, in a 48-well plate, in the presence of 0.2 or 1 mg/mL plate-bound anti-CD3 mAb. After 48 hours at 37uC, cells were harvested from the plate and cellular proliferation was analyzed by flow cytometry. Supernatants were also collected and analyzed by ELISA for the presence of IL-2, IFNc, and IL-17A, according to the manufacturer's (eBioscience) instructions.

Statistical Analysis
Data are presented as mean 6 SEM. Student's t-test was used for all statistical comparisons (Graph Pad Prism or Microsoft Excel software). Only p values less than or equal to 0.05 (i.e., statistically significant) are denoted.

CD28 is Differentially Expressed on cd T Cell Effector Subsets
Our re-evaluation of the role of CD28 in cd T cell development and differentiation began with comparing CD28 expression levels on the recently identified cd-IFNc and cd-17 effector subsets. To optimize the detection of CD28 expression on the cell surface, we stained cells with a PE-conjugated anti-CD28 mAb and ran them on a flow cytometer fitted with a 568 nm laser, which excites PE significantly better than the conventional 488 nm laser [24]. cd-IFNc and cd-17 effector subsets can be identified in the thymus and periphery by their expression of different surface markers. All cd-IFNc cells express the TNFR family member CD27, with a subset of them expressing CD122 [18,19]. cd-17 cells express neither CD27 nor CD122 [18,19] but instead express IL-23R and the chemokine receptor CCR6 [20,[25][26][27]. Using CD27 and IL-23R to detect cd-IFNc and cd-17 cells, respectively, we found that CD28 expression levels were 2 to 3-fold higher on cd-IFNc cells than on cd-17 cells, regardless of whether the cd T cells were from C57BL/6 or Vc6/Vd1 cdTCR Tg mice ( Figure 1 and Figure S1). Since the CD27 + IFNc-producing cd T cell effector subset consists of both CD122and CD122 + cells, we analyzed each subpopulation and found that their CD28 expression levels were comparable (data not shown). In addition, we noted that cd-IFNc and cd-17 cells in the thymus expressed CD28 at considerably higher levels than their peripheral counterparts. This is in contrast to CD4 + thymocytes and CD4 + LN cells, which were found to express equivalent levels of CD28 ( Figure 1A and B). Taken together, these data demonstrate that CD28 is differentially expressed between cd-17 and cd-IFNc cells and, as these cells leave the thymus and enter the periphery, they downregulate their expression of CD28.

CD28 Deficiency Results in Reduced Numbers of cd Lineage Cells
Given the relatively high expression levels of CD28 on DN cd thymocytes, we next sought to determine whether CD28 is required for cd T cell development by comparing the ability of CD28 +/+ and CD28 2/2 mice to generate cd lineage cells. Significant decreases were observed in both the percentage and number of TCRcd + cells in the thymus, spleen, and pLNs of CD28 2/2 mice compared to CD28 +/+ mice (Figure 2A and B). The reduction in cd T cell numbers was not due to a partial block in early cd T cell development, as the percentage of CD25 + CD27 + cd thymocytes in CD28 +/+ and CD28 2/2 mice was equivalent ( Figure 2C). Furthermore, phenotypic analysis of the cd lineage cells from CD28 2/2 mice revealed no appreciable differences in CD5 and cdTCR surface levels, or in Vc usage, relative to the cd lineage cells from CD28 +/+ mice ( Figure 2D), demonstrating that cdTCR signal strength and cdTCR repertoire selection were not altered by the lack of CD28 costimulation.
There are several possible explanations for the loss of cd T cells in CD28 2/2 mice. The first is that CD28 signaling is required for the optimal proliferation and/or survival of DN cd thymocytes. To test this possibility, we used flow cytometric analysis to compare the percentages of CD28 +/+ and CD28 2/2 DN cd thymocytes expressing the Ki-67 antigen, a marker for actively cycling cells, and staining positive for Annexin-V, a marker of apoptosis. There were no differences in the proportions of Ki-67 + or Annexin-V + DN cd thymocytes between CD28 +/+ and CD28 2/2 mice ( Figure 3A and B), indicating that neither diminished cellular proliferation nor increased cell death can account for the reduced cd T cell numbers in CD28 2/2 mice.
Another possible explanation for the loss of cd T cells in CD28 2/2 mice is that CD28 signaling plays a role in commitment to the cd lineage. To investigate this possibility, we used a cdTCR Tg mouse model in which the ab/cd lineage decision is mediated by the cdTCR and in which alterations in ab/cd lineage choice can be detected by enumerating the number of cd lineage (DN TCRcd + ) and ab lineage (CD4 + CD8 + ; DP) thymocytes [28]. In the thymus of cdTCR Tg CD28 +/+ and cdTCR Tg CD28 2/2 mice, we found no significant differences in the numbers of ab and cd lineage cells ( Figure 3C and D), demonstrating that, in the absence of CD28, there is no inherent defect in cd lineage commitment.

CD28 Deficiency Affects the Early Stages of T Cell Development
Since CD28 signaling has no apparent role in cd lineage commitment or in cd thymocyte proliferation and survival, we next examined the possibility that CD28 signaling is required prior to the expression of the cdTCR, for the proliferation and/or survival of the thymic progenitors that give rise to cd T cells. Consistent with a previous report [29], we detected CD28 expression on the surface of all immature DN thymocytes, with CD28 surface levels increasing as thymocytes transition through the DN1 (lineage -CD44 + CD25 -), DN2 (lineage -CD44 + CD25 + ) and DN3 (lineage -CD44 -CD25 + ) stages to the DN4 stage (lineage -CD44 -CD25 -) ( Figure 4A). To determine whether CD28 expression is required for the progression of thymocytes through the DN stages, we compared the distribution, proliferative status and viability of DN subsets in CD28 +/+ and CD28 2/2 mice. Notably, we found that the percentages and numbers of DN1 [including the early thymic progenitor (ETP) subset], DN2 and DN3 thymocytes were decreased, while those of DN4 thymocytes were equivalent, in CD28 2/2 mice compared to CD28 +/+ mice ( Figure 4B and C). Comparison of the proliferative status of the DN subsets in CD28 +/+ and CD28 2/2 thymi revealed that the percentages of proliferating thymocytes in the DN1 and DN4 subsets, but not the DN2 and DN3 subsets, were significantly reduced in the absence of CD28 ( Figure 4D). Moreover, only the DN4 subset in CD28 2/2 mice exhibited increased cell death, as evidenced by the higher percentage of Annexin V + DN4 thymocytes in CD28-deficient mice than in CD28-sufficient mice ( Table 1). Taken together, these data indicate that proliferation of DN1 thymocytes is impaired in CD28 2/2 mice and suggest that CD28 signaling regulates the size of the thymic progenitor pool. Importantly, considering that thymocytes committing to the cd lineage undergo limited proliferation [30], it follows that CD28 2/2 mice, which have fewer thymic progenitors than CD28 +/+ mice, generate fewer cd T cells.  . Effect of CD28 deficiency on cd T cell development. (A) Phenotypic analysis of CD28 +/+ and CD28 2/2 mice. Dot plots show representative TCRcd versus CD3 staining profiles on gated DN thymocytes, DN pLN cells and DN splenocytes. Numbers within the two-color plots represent the percentage of TCRcd + CD3 + cells in the gate. The mean cell number 6 SEM for each tissue and genotype are displayed above the twocolor plots. (B) Bars represent the mean number 6 SEM of DN TCRcd + cells in the thymus, pLNs and spleen of CD28 +/+ (n = 4 to 8) and CD28 2/2 (n = 4 to 8) mice. *p#0.05, **p#0.01 and #p#0.001. (C) Representative histograms comparing CD3 and CD5 levels on cd (TCRcd + ) thymocytes from CD28 +/+ (n = 8) and CD28 2/2 (n = 8) mice. (D) Representative histograms comparing the percentages of Vc1 + and Vc4 + cd thymocytes from CD28 +/+ (n = 8) and CD28 2/2 (n = 8) mice. doi:10.1371/journal.pone.0063178.g002

CD28 Deficiency Does Not Affect cd T Cell Effector Fate Commitment or Function
Because of the difference in CD28 expression levels on cd-17 and cd-IFNc cells, we sought to determine whether these effector lineages had a differential requirement for CD28 in their development or differentiation. To test whether CD28 signaling is required to generate cd-17 or cd-IFNc cells, we measured the frequency of CCR6 + CD27 -(cd-17) and CCR6 -CD27 + (cd-IFNc) DN TCRcd + cells in CD28-sufficient and CD28-deficient mice. Although there were fewer cd lineage cells in CD28 2/2 mice ( Figure 2B), we observed no differences in the distribution of CCR6 + CD27and CCR6 -CD27 + cells in the thymus, spleen and pLNs between CD28 +/+ and CD28 2/2 mice ( Figure 5A and B).  These findings indicate that CD28 signaling does not play a role in commitment to either effector fate. We next investigated whether CD28 costimulation is required for the differentiation of cd-17 and cd-IFNc cells into cytokineproducing effectors. We and others have previously demonstrated that in vitro activation of cd T cells through TCR stimulation alone is sufficient to induce not only robust proliferation but also IL-17 and IFNc production [18,[31][32][33][34][35]. To determine whether cytokine production by cd T cells is enhanced by cdTCR and CD28 coengagement, we stimulated purified cd T cells, from both cdTCR Tg and TCRa 2/2 mice, with anti-CD3 mAb in the presence or absence of anti-CD28 mAb. Although IFNc production was modestly enhanced, no change in IL-17A production was observed following TCR/CD28 co-engagement ( Figure S2A). To determine whether CD28 costimulation is required for in vivo cd T cell effector functions, we infected CD28 +/+ and CD28 2/2 CD28 and the Generation of cd T Cell Effectors PLOS ONE | www.plosone.org mice with Listeria monocytogenes, a model pathogen that not only induces IL-17A-and IFNc-producing cd T cells [36,37], but also upregulates the expression of the CD28 ligands, B7.1 and B7.2, on antigen presenting cells early in the course of infection [38-40; data not shown]. Leading up to day 5 post infection, the peak of the cd T cell response [36,37,40,41], we noted no difference in the total number of splenocytes in CD28 +/+ and CD28 2/2 mice ( Figure 6A). Furthermore, on days 1 and 5 post infection, equivalent fold increases in the percentage and number of splenic cd T cells were observed in both genotypes ( Figure 6B and C), indicating that CD28 deficiency has no effect on the expansion and/or recruitment of cd T cells in response to Lm. When we compared the ability of CD28 +/+ and CD28 2/2 cd T cells to differentiate into cytokine-producing cells, we observed similar percentages and numbers of IL-17A + and IFNc + cd T cells in both genotypes, regardless of the duration of infection ( Figure S2B) or the dose of anti-TCRcd mAb used to re-stimulate cd T cell effectors in vitro ( Figure 6D and E). Because cd-17 and cd-IFNc cells play a critical role in mediating bacterial clearance [36,37,40,41], we also enumerated bacterial CFU in the liver on day 5 post infection. Consistent with equivalent numbers of IL-17A + and IFNc + cd T cells in infected CD28 +/+ and CD28 2/2 mice, we observed no difference in the bacterial burden of CD28 +/+ and CD28 2/2 livers ( Figure 6F). Together, these findings demonstrate that neither cd-17 nor cd-IFNc cells require in vivo CD28 costimulation during Lm infection to differentiate into cytokine-producing effectors.
It has been previously reported that Lm-specific CD8 + ab T cell effectors require CD28 costimulation [38,42]. To confirm this finding in our system, we infected CD28 +/+ and CD28 2/2 mice with recombinant Lm expressing a protein containing SIINFEKL, an ovalbumin-derived MHC Class I-restricted determinant. On day 7 post infection, we quantified SIINFEKL-specific CD8 + ab T cells by intracellular IFNc staining. In agreement with the previous reports, we observed dramatically fewer IFNc + CD8 + ab T cells in CD28 2/2 mice than in CD28 +/+ mice ( Figure 6G and H). Collectively, these findings highlight major differences in the molecular requirements for the generation of ab and cd T cell effectors during the course of Lm infection.

Differences in Glucose Uptake Between Resting ab and cd T Cells
CD28 costimulation supports ab T cell growth, proliferation and effector function by regulating glucose uptake and utilization [5,6]. Accordingly, differences between ab and cd T cells in their regulation of glucose metabolism may explain the CD28 independence of cd T cell activation, expansion and differentiation during Lm infection. To test this, we first assessed the ability of cd T cells from both CD28 +/+ and CD28 2/2 mice to take up 2-NBDG, a fluorescent glucose derivative, in the presence or absence of CD3 engagement. We used cdTCR Tg mice for this assay, as large numbers of DN cd T cells can be purified from these mice by negative selection [31,33,34], which eliminates any concern that purification with anti-TCRcd mAbs results in cd T cell activation. Remarkably, we observed no difference in the kinetics or magnitude of glucose uptake between resting and activated cd T cells from either cdTCR Tg CD28 +/+ or cdTCR Tg CD28 2/2 mice ( Figure 7A, D and data not shown), indicating that cdTCR signaling has minimum effect on glucose uptake. Because the ability of a cell to take up glucose is dependent on surface expression of glucose transporters (GLUTs), we next determined whether GLUT1 and GLUT3, the two isoforms expressed predominantly by leukocytes [43,44], were differentially expressed by ab and cd T cells. To accomplish this, we developed a flow cytometric assay to detect surface expression of GLUT1 and GLUT3, using cells that are known to express high levels (e.g., neutrophils) and low levels (e.g., DP thymocytes) of the GLUT isoforms ( Figure 7B) [43][44][45][46]. When we compared surface expression of these isoforms on peripheral ab T cells and cd T cells from CD28 +/+ and CD28 2/2 mice, we found that cd T cells and CD8 + ab T cells expressed higher surface levels of GLUT1 than CD4 + ab T cells ( Figure 7B and S3A). GLUT3 expression, on the other hand, was detected only on cd T cells ( Figure 7B and S3A). Importantly, both cd-17 and cd-IFNc cells were found within the GLUT1 hi and GLUT3 hi cd T cell subsets, but a higher percentage of cd-17 cells was contained within the GLUT3 hi subset than in the GLUT1 hi subset ( Figure 7C). Interestingly, GLUT1 and GLUT3 expression levels were also expressed at relatively high levels on cd thymocytes ( Figure S3B), suggesting that GLUT1 and GLUT3 expression levels are induced in the thymus as part of a developmental program. Together, these data indicate that GLUT isoforms are differentially expressed between ab and cd T cells, with cd T cells, on average, expressing higher surface levels of GLUT1 and GLUT3 than ab T cells. This differential expression suggests that cd T cells are better equipped to take up glucose than ab T cells.
Because GLUT3 has a higher affinity for glucose, and a greater glucose transport capacity, than GLUT1 [46], we sought to determine whether there was a difference in the kinetics of glucose uptake between resting ab and cd T cells under low glucose conditions. Indeed, we found that cd T cells take up 2-NBDG, when present at a low (0.03 mM) concentration, at a rate that is twice as fast as that of either CD4 + or CD8 + ab T cells ( Figure 7D), demonstrating that the GLUT3 on cd T cells is functional and that cd T cells have a higher basal rate of glucose uptake than ab T cells.

cd T Cell Effector Function is Optimal over a Wide Range of Glucose Concentrations
In addition to signals from the TCR and CD28, ab T cells require relatively high glucose concentrations to proliferate and to produce cytokines ($0.05 mM of glucose to produce IL-2 and $0.5 mM to proliferate and to produce IFNc [6]). The ability of cd T cells to produce cytokines during Lm infection independently of CD28 costimulation combined with their high GLUT1 and GLUT3 surface expression suggested that cd T cells, when solely stimulated through the cdTCR, would exhibit effector functions over a wide range of glucose concentrations. To test this, we in vitro stimulated purified cd T cells, from both cdTCR Tg CD28 +/+ and cdTCR Tg CD28 2/2 mice (Figure 8), in the presence of varying concentrations of glucose (0 to 5 mM), and then measured their ability to proliferate and produce various cytokines. When activated with a low but optimal dose of anti-CD3 mAb (1 mg/ ml), we noted that cd T cells from both genotypes proliferated in as little as 0.05 mM glucose and reached maximum proliferation at 0.5 mM glucose ( Figure 8A). In addition, stimulated cd T cells from both CD28 +/+ and CD28 2/2 mice produced low amounts of IL-17A, and high amounts of IL-2, in the absence of glucose ( Figure 8A). However, while production of IL-2 decreased as glucose concentrations increased, production of both IL-17A and IFNc improved under these same conditions, suggesting that ambient glucose concentrations regulate cd T cell effector potential ( Figure 8A). Surprisingly, we found that CD28-deficient cd T cells produced, on average, more IL-17A and IFNc than CD28-sufficient cd T cells, irrespective of glucose concentration. These data suggest that CD28-B7 interactions restrain the production of IL-17A and IFNc by cd T cells.
It has been shown that cd T cells require CD28 costimulation to proliferate when stimulated with a suboptimal dose of anti-CD3 mAb (,1 mg/ml) but not with an optimal dose of anti-CD3 mAb ($1 mg/ml) [47], which is consistent with high TCR occupancy or a ''strong'' signal 1 compensating for the lack of CD28 costimulation [48]. However, as the proliferation assays were performed in RPMI 1640 [47], which contains glucose at a concentration of 11 mM, twice the glucose concentration found in blood [49][50][51], we next sought to determine whether low glucose concentrations altered the ability of cd T cells to proliferate and produce cytokines following stimulation with a ''suboptimal'' dose (0.2 mg/ml) of anti-CD3 mAb. Strikingly, we found that the magnitude of the proliferative and cytokine responses of activated CD28-deficient and CD28-sufficient cd T cells was higher when glucose concentrations were #0.5 mM than when the concentration was 5 mM ( Figure 8B). These data indicate that ambient glucose concentrations play an important role in regulating cd T cell activation, proliferation, and effector function.

Discussion
To determine whether cd T cells require CD28 signaling in the thymus during effector fate acquisition, or in the periphery during effector cell differentiation, we re-evaluated the effects of CD28deficiency on cd T cell development and function. We report that CD28 costimulation is not required for commitment to the cd lineage, proliferation and survival of cd thymocytes, nor adoption of the cd-17 and cd-IFNc effector fates. Furthermore, in both in vitro and in vivo functional assays, we demonstrated that CD28 costimulation is dispensable for the generation of both IL-17 + and IFNc + cd T cell effectors. To understand why cd T cells are less dependent on CD28 costimulation than ab T cells for their activation and differentiation, we investigated whether cd T cells are endowed with properties that bypass the requirement for CD28 costimulation. Indeed, when we measured glucose uptake and utilization by cd T cells, we noted significant differences between ab and cd T cells in their surface levels of GLUT1 and GLUT3 and in their threshold concentration of glucose required for optimal effector cell function. Together, these data underscore significant differences in the molecular requirements for the generation of ab and cd T cell effectors.
The re-evaluation of the role of CD28 in cd T cell development revealed that CD28 signaling regulates the size of the thymocyte progenitor pool by controlling the proliferative ability of DN1 thymocytes. CD28 is expressed on the most immature thymocyte populations, including ETPs, whereas its ligands, B7.1 and B7.2, are expressed in the medulla and, to a lesser extent, the cortex [29,52]. Once progenitors colonize the thymus through the vasculature located at the cortico-medullary junction, they traverse the cortex, initiating a T cell developmental program as they migrate toward the subcapsule [53]. During their journey through the cortex, the immature DN thymocytes encounter and interact with B7-expressing cortical epithelial cells. According to our data, this interaction is necessary for optimal cellular proliferation by DN1 thymocytes. Although its mechanism of action is currently unknown, it is possible that CD28 signaling controls proliferation of DN1 thymocytes 1) by regulating expression of the components of the IL-7R, c-kit, Hedgehog and/or Notch signaling pathways, all of which play a role in DN thymocyte proliferation [54][55][56][57][58], or 2) by acting in concert with one or more of these signaling pathways to promote maximum proliferation. Regardless of the mechanism involved, it is important to note that because the CD28-B7 interaction occurs prior to the expression of a TCR Figure 8. Effect of glucose concentration on cd T cell proliferation and cytokine production. Purified cd T cells from CD28 +/+ cdTCR Tg and CD28 2/2 cdTCR Tg mice were labeled with CFSE and then cultured in glucose-free medium or glucose-free medium supplemented with increasing concentrations of glucose in the presence of 1 mg/mL (A) or 0.2 mg/ml (B) of plate-bound anti-CD3 mAb. 48 h later, cells were harvested and their proliferative response was measured by flow cytometric analysis. Supernatants were also collected and cytokine production was measured by ELISA. Effect of glucose concentration on cellular proliferation, IL-2 production, IL-17A production, and IFNc production. Data are representative of at least three mice per genotype. *p # 0.05, **p # 0.01, #p # 0.001. doi:10.1371/journal.pone.0063178.g008 isoform, these data provide evidence for CD28 delivering a unique signal during the early stages of T cell development.
Although CD28-B7 interactions play no appreciable role in cd T cell development, maturation or effector fate specification in the thymus, they do have a role in early ab T cell development, where they promote survival and cellular proliferation of DN4 thymocytes [52)]. Our data showing that CD28 2/2 DN4 thymocytes undergo more apoptosis and less proliferation than their wild-type counterparts are consistent with these findings. However, it is unclear how DN4 thymocyte numbers remain equivalent in CD28 +/+ and CD28 2/2 mice, in the face of the significant effects of CD28-deficiency on DN4 thymocyte survival and proliferation. Possible explanations include an accelerated transition time between the DN3 and DN4 stages [52] and a slower transition time between the DN4 and DP stages [59].
ab T cells are dependent upon CD28 costimulation to upregulate expression of GLUTs on their cell surface as well as the enzymes involved in the glycolytic pathway [5,6,60]. This is not the case for cd T cells, as both CD28 +/+ and CD28 2/2 cd T cells express relatively high levels of GLUT1 and GLUT3 and, when activated by anti-CD3 mAb alone, are able to proliferate and secrete IL-17 and IFNc in relatively low glucose concentrations (0.05 mM). In fact, CD28 2/2 cd T cells produced significantly more IL-17A and IFNc than CD28 +/+ cd T cells in vitro, suggesting that CD28 is a negative regulator of cd T cell effector cytokine production. Given that cd T cells and CD8 + ab T cells express equivalent levels of GLUT1 but have different basal rates of glucose uptake, we propose that the ability of cd T cell effectors to function in low glucose concentrations is due to their expression of GLUT3, which has a higher affinity for glucose (K m = 1.4 mM) than GLUT1 (K m = 6.9 mM) [46]. However, because GLUT3-deficiency is embryonic lethal [61], determining whether GLUT3 expression confers cd T cells with the ability to differentiate into cytokine-producing effectors independently of CD28 costimulation during Lm infection awaits the generation of mice in which Slc2a3, the gene that encodes GLUT3, can be conditionally deleted in T cells.
It is interesting to note that the induction of GLUT1 and GLUT3 expression by cd lineage cells occurs in the thymus, presumably as part of their developmental program. The relatively high expression of GLUT isoforms at an early stage in cd T cell development raises the question as to how the expression of GLUT1 and GLUT3 is induced in cd lineage cells. In light of previous studies showing that TCR stimulation, IL-7 signaling, and insulin signaling induce GLUT1 and GLUT3 expression on the surface of ab T cells [43,44,62,63], it is tempting to speculate that similar mechanisms act in both cd thymocytes to induce GLUT expression and in peripheral cd T cells to maintain GLUT expression.
While we have found that cd T cells do not require CD28 costimulation during Lm infection to generate cd-17 and cd-IFNc effectors, a recent study reported that CD28 costimulation was required for the expansion of cd-17 and cd-IFNc effectors during blood-stage Plasmodium infection [47]. One possible explanation for the discrepancy is that cdTCR occupancy is significantly higher during Lm infection than during Plasmodium infection, with the strong signal 1 compensating for the lack of CD28 costimulation. The idea that cdTCR occupancy is high during Lm infection is supported by the study of O'Brien and colleagues [64], which showed that the expression levels of the ligand for the Vc6/ Vd1 TCR, the TCR expressed by one of the major cd T cell subsets responding to Lm [36], are significantly increased on macrophages following Lm infection. Another possible explanation is that Lm infection, but not blood-stage Plasmodium infection, lowers glucose levels in the spleen to levels that permit the activation, expansion and differentiation of cd T cell effectors in the absence of CD28 signaling. Interestingly, although both pathogens induce hypoglycemia in the host, significant reductions in blood glucose levels are observed as early as day 1 post Lm infection [49] but not until days 6 to 8 post blood-stage Plasmodium infection [50,51]. Considering that cd T cell effector function peaks on day 5 or earlier during the immune response [36,37,40,41,47] and that cd T cell proliferation and cytokine production are measurably better in low glucose concentrations than in high glucose concentrations following stimulation with suboptimal doses of anti-CD3 mAb, this 5 to 7 day difference in the onset of hypoglycemia could have a great impact on the cd T cell response.
In summary, we have shown that cd T cells, unlike NKT and T reg cells, do not require costimulatory signals from CD28 to acquire their effector fates in the thymus. Likewise, we have shown that, in an Lm infection model, cd T cells differentiate into IL-17and IFNc-producing effectors, which contribute to bacterial clearance, independently of CD28 costimulation. This independence may be explained by enhanced glucose metabolism, a strong signal through the cdTCR, or both. Figure S1 Comparison of CD28 expression levels on cd T cell subsets in the thymus and periphery of cdTCR Tg mice. Analysis of CD28 expression on various gated subsets in the thymus (A) and pLNs (B) of CD28 +/+ (i.e., IL-23R gfp/+ ) cdTCR Tg mice, Black histograms show representative staining of CD28 on total, IL-23R + (cd-17) and CD27 + (cd-IFNc) DN TCRcd + subsets. Staining of total thymocytes (A) and pLN cells (B) from CD28 2/2 mice are shown as negative controls (shaded histograms). Numbers in the plots represent the mean fluorescent intensity (MFI) of CD28 expression. Data are representative of nine mice in four independent experiments. (TIF) Figure S2 Effect of CD28 deficiency on cd T cell cytokine production, in vitro and in vivo. (A) Comparison of IL-17A and IFNc production by anti-CD3 stimulated DN cdTCR + cells from cdTCR Tg and TCRa 2/2 mice in the presence or absence of anti-CD28 mAb. LN cells were in vitro stimulated with 5 mg/ ml of hamster IgG, 5 mg/ml of anti-CD3 mAb or 5 mg/ml each of anti-CD3 and anti-CD28 mAbs. 16 h later, cells were harvested and cytokine production was assayed by intracellular (i.c.) flow cytometric analysis. Dot plots show representative i.c. staining for IFNc versus IL-17A in gated DN cdTCR + cells. Numbers in quadrants represent percentage of cells in that quadrant. Data shown are representative of at least 3 mice per genotype. (B) CD28 +/+ and CD28 2/2 mice were infected i.p. with 3610 4 CFU Lm and sacrificed on day 1 and 5 post infection to examine the cd T cell response. Dot plots show representative i.c. staining for IFNc and IL-17A on gated TCRcd + cells of CD28 +/+ and CD28 2/2 mice, in vitro re-stimulated with 5 mg/ml anti-TCRcd mAb, at day 1 (n = 5 mice per genotype) and day 5 (n = 13 mice per genotype). Numbers in gate represent percentage of cells in that gate. (TIF) Figure S3 Comparison of GLUT1 and GLUT3 expression levels on cd lineage cells. (A) GLUT1 (top panel) and GLUT3 (bottom panel) expression levels on DN cd T cells from the pLNs of cdTCR Tg CD28 +/+ (left panels) and cdTCR Tg CD28 2/2 (right panels) mice. Staining of DP thymocytes from each genotype is also shown as a negative control. Data are representative of 3 to 6 mice per genotype. (B) GLUT1 (left panel) and GLUT3 (right panel) expression levels on CD4 + CD3 + thymocytes, CD8 + CD3 + thymocytes, and cd thymocytes. Staining on DP thymocytes (shaded histogram) is also shown as a negative control. Data are representative of three independent experiments and 7 CD28 +/+ mice. (TIF)