Flow Cytometric Staining Using CD1d Tetramers.
To investigate the functions of fresh peripheral blood CD1d-restricted T cells, we used tetrameric complexes of a human CD1d-Fc fusion protein, or the isotype-matched negative control UPC10 mAb, and fluorescently tagged protein A in flow cytometric analyses. To verify its specificity, the CD1d tetramer was tested for staining of two human CD1d-restricted NKT cell clones, DN2.B9 and DN1.10B3, that respond functionally to α-GalCer (21
). The CD1d tetramer was preincubated with α-GalCer dissolved in DMSO, or with an equivalent volume of DMSO alone. Clearly positive staining of NKT clones DN1.10B3 and DN2.B9 was observed using the CD1d tetramer preincubated with α-GalCer (MFI = 67.3 and 86.2, respectively), while staining by the CD1d tetramer that was mock-treated with DMSO (MFI = 3.2 and 3.03), was equivalent to that of the UPC10 negative control complex (MFI = 3.03 and 3.04). Thus, addition of a lipid antigen was required for CD1d tetramer binding to the NKT cell clones. Three αβ T cell clones and one γδ T cell clone that were not CD1d-restricted were stained by neither the α-GalCer antigen treated, nor the DMSO mock-treated CD1d tetramer (data not shown), suggesting the CD1d tetramer staining was specific for CD1d-restricted T cells.
We next used two-color flow cytometric analysis on PBMC samples purified from 20 healthy donors to assess the staining of peripheral blood T cells with the α-GalCer–loaded CD1d tetramer. The samples were stained with an anti-CD3 mAb and the CD1d tetramer treated with α-GalCer, or DMSO, or with the UPC10 negative control complex. A small population of T cells that stained positively with the α-GalCer–treated tetramer could be detected for 15 out of 20 donors, whereas staining with the DMSO mock-treated CD1d tetramer was equivalent to the UPC10 negative control complex ( a and b, and data not shown). For 5 out of 20 donors, the percentage of T cells stained by the α-GalCer–treated CD1d tetramer was not greater than the negative control staining (data not shown). The percentage of the total T cells that were specifically stained with the tetramer ranged from undetectable (<0.01% of the CD3+
lymphocytes) to 2.34%, with a median of 0.034% and a mean of 0.194%. Thus, in most healthy donors a small subpopulation of peripheral blood T cells could be detected using α-GalCer–loaded CD1d tetramers. The frequencies of α-GalCer reactive CD1d-restricted T cells observed in our analysis are similar to those obtained by Karadimitris et al. who used human CD1d tetramers loaded with α-GalCer to analyze PBMC samples from seven hepatitis C virus–infected patients and three healthy donors (39
). Two reports have suggested that some NK cells may recognize CD1 molecules (40
). However, we observed no CD1d-specific staining of CD3−
lymphocytes (see a and b).
Figure 1. Flow cytometric staining of fresh human PBMCs using CD1d-Fc tetramers and antibodies against cell surface markers. a, b, and d are composite contour/dot plots, in which areas of infrequent events are shown as individual dots and higher density areas are (more ...)
We confirmed that the CD1d tetramer staining of human peripheral blood lymphocytes was specific for CD1d-restricted T cells by deriving T cell clones from the stained population. 14 T cell clones were established from four donors by flow cytometric sorting of tetramer positive cells. Flow cytometric analysis of the clones revealed uniform positive staining using the α-GalCer–loaded CD1d tetramer and no staining using the DMSO-treated CD1d tetramer or the negative control UPC10 complex (data not shown). The clones also secreted cytokines (including IFN-γ, IL-4, and GM-CSF) in response to CD1d-transfectants in the presence of α-GalCer, but did not respond to α-GalCer–treated untransfected parent cells (unpublished data). Thus, human peripheral blood T cells stained by the α-GalCer loaded tetramer are CD1d specific and recognize the α-GalCer lipid antigen.
Phenotypic Characterization of CD1d Tetramer Positive Cells.
Murine α-GalCer–specific CD1d-restricted T cells have been shown to be CD4+
or CD4 and CD8 double negative, and to include CD161+
). We investigated CD4, CD8α, and CD8β staining for CD1d tetramer positive T cells from human peripheral blood of four healthy donors. An anti-CD4 mAb stained approximately half of the tetramer positive cells (mean = 51 ± 19%, and see c). Approximately half of the tetramer positive cells stained positively for CD8α (mean = 51 ± 15%), but little or no positive staining was observed for CD8β ( c). Therefore, CD1d tetramer positive T cells in human peripheral blood can be divided into two subsets: a CD4+
subset and a CD4−
subset that contains cells expressing CD8αα homodimers, but almost none expressing CD8αβ heterodimers.
We next examined expression of the NK complex receptors CD161, CD94, and CD69 by CD1d tetramer positive T cells. In C57Bl/6 mice, generally <5% of splenic T cells coexpress CD161, and 60–80% of these T cells are stained by α-GalCer loaded murine CD1d tetramers (13
). However, an analysis of CD161 staining of human peripheral blood lymphocytes found that ~25% of the T cells from healthy adult donors coexpressed CD161 (42
). We found that a mean of 17 ± 8.4% of the total peripheral blood T cells from five healthy donors expressed CD161, but only a small fraction stained positively with the α-GalCer loaded human CD1d tetramer ( d). Most tetramer positive T cells were positive for CD161 (mean = 72 ± 14%). In contrast, CD94 was expressed by a mean of 49 ± 22% of the CD1d tetramer positive cells, and CD69 was detected on very few tetramer positive cells (mean = 6.4 ± 2.8%). Hence, our results show that most CD1d tetramer positive T cells express CD161 but are not necessarily positive for other NK complex markers, and the great majority of human CD161+
T cells are not α-GalCer specific CD1d-restricted T cells.
Cytokine Production by Fresh CD1d-restricted T Cells.
CD1d-restricted T cells were first noted for their ability to rapidly secrete substantial amounts of the Th2 cytokine IL-4 upon anti-CD3 stimulation, but more recent investigations using α-GalCer to selectively stimulate murine CD1d-restricted T cells have observed Th1 biased cytokine responses immediately after stimulation (43
). Analysis of cytokine production by in vitro cultured human CD1d-restricted T cell clones has demonstrated that most clones from healthy donors produce both IFN-γ and IL-4, but it is unclear whether this is representative of the cytokine secretion of CD1d-restricted T cells in vivo (21
). Therefore, we performed intracellular cytokine staining to investigate cytokine production by CD1d-restricted T cells freshly isolated from human peripheral blood.
PBMC samples were treated for 6 h with PMA and ionomycin or incubated overnight with α-GalCer, or with no stimulus, in culture medium containing monensin to block exocytosis. The samples were then washed and stained with the α-GalCer–loaded CD1d tetramer and anti-CD3 or anti-CD4, then fixed and permeabilized and stained with antibodies to Th1 or Th2 cytokines, or isotype-matched negative control antibodies. This protocol resulted in clear positive staining by the anti-cytokine antibodies compared with the isotype matched negative control antibodies for a fraction of both the tetramer positive and tetramer negative T cells in the PMA/ionomycin treated samples ( and ). For the α-GalCer–treated samples, a fraction of the tetramer positive cells stained positively for cytokines, while the tetramer negative cells were equivalent to the unstimulated control (, and data not shown). Unstimulated samples gave little or no positive staining for any of the cytokines (, and data not shown).
Figure 2. Flow cytometric probability contour plots of CD1d tetramer positive lymphocytes stained for intracellular cytokines. The plots are gated on the α-GalCer–loaded CD1d tetramer positive lymphocytes within a PBMC sample, and show staining (more ...)
Cytokine Production by CD14 Tetramer-stained T Cell Subsets
Remarkably, there was a clear-cut difference in the cytokines produced by the CD4+ and CD4− tetramer positive subsets: the CD4+ subset made both Th1 and Th2 cytokines, whereas the CD4− subset overwhelmingly made IFN-γ and TNF-α ( and ). PMA/ionomycin treatment was more efficient at inducing cytokine production than was incubation with α-GalCer, but the difference in cytokine production between the CD4+ and CD4− subsets was observed for both types of stimulation (). Hence, CD4 expression distinguishes two subsets of CD1d-restricted T cells that have different patterns of cytokine production.
These results support the hypothesis that CD1d-restricted T cells overall are a potent cytokine producing subpopulation, since compared with tetramer negative T cells, a large percentage of the CD1d tetramer positive T cells stained positively for each of the cytokines tested (). Surprisingly, we find that in healthy donors most CD1d-restricted T cells produced Th1 cytokines, and only a minority produced each of the Th2 cytokines analyzed. Because each cytokine was tested separately in this analysis, it is unclear whether the Th2 cytokines were all produced by the same CD4+ tetramer positive cells or by different subsets. However, because only ~8% of the CD4+ tetramer positive cells did not produce IFN-γ, and Th2 cytokines were generally produced by at least 20% of the CD4+ tetramer positive cells, most of the CD4+ tetramer positive cells that stained positively for Th2 cytokines in our analysis probably also produce IFN-γ (see ). Thus, many Th2 cytokine producing tetramer positive cells probably have a Th0 cytokine production phenotype, rather than a traditional Th2 phenotype.
Nevertheless, compared with tetramer negative T cells, Th2 cytokine producing cells were unusually common in the CD4+ tetramer positive population (see ). Particularly for IL-5 and IL-6, tetramer negative T cells that stained positively were extremely rare, and CD1d tetramer positive cells made up a significant fraction of the total IL-5– and IL-6–producing PBMC T cells (mean = 13 ± 7.1% and 19 ± 14%, respectively). Thus, while the tetramer positive population as a whole appeared biased toward Th1 cytokine production, CD1d-restricted T cells may be a very important source of certain Th2 cytokines. Moreover, if the CD4− and CD4+ subsets can be differentially activated in vivo, the segregation of CD1d-restricted T cells into these two subsets that have different cytokine secretion profiles could explain how some CD1d-restricted T cell–mediated responses appear strongly biased toward Th1 cytokines, while in other cases Th2 responses are prominent.
Cytolytic Functions of CD1d-restricted T Cells.
To investigate the cytotoxic potential of fresh peripheral blood CD1d-restricted T cells, we performed intracellular staining for perforin on unstimulated PBMC samples from five healthy donors. There was generally little or no positive perforin staining of the CD4+ tetramer positive subset (mean = 8.5% positive and a, top quadrants). In contrast, the percentage of the CD4− tetramer positive subset that stained positively for perforin varied substantially from donor to donor (range = 0–55%, mean = 23%, and a, bottom quadrants). Hence, most CD1d-restricted T cells from unstimulated peripheral blood did not express perforin, but cells of the CD4− subset more frequently contained perforin than the CD4+ subset.
Figure 3. Flow cytometric probability contour plots showing intracellular perforin and IFN-γ staining of CD1d tetramer positive lymphocytes after stimulation. The plots are gated on α-GalCer–loaded CD1d tetramer positive lymphocytes within (more ...)
Several reports have indicated that exposure to IL-2, IL-12, or α-GalCer may enhance cytotoxicity by cultured CD1d-restricted NKT cells (47
). To investigate the effects of stimulation on effector functions of CD1d-restricted T cells, we incubated PBMC samples overnight with α-GalCer, IL-2, IL-12, or LPS which potently induces secretion of inflammatory cytokines, or treated them for 6 h with PMA and ionomycin. Treatment with α-GalCer, IL-2, IL-12, and PMA/ionomycin was performed in the presence of monensin to prevent secretion of cytokines or other factors that could secondarily affect the CD1d-restricted T cells. LPS treatment was performed in the presence or absence of monensin, to compare effects resulting from the inflammatory response induced by LPS, with direct effects of LPS. After stimulation the PBMC samples were stained with the α-GalCer–treated CD1d tetramer and anti-CD4, then fixed, permeabilized, and stained for perforin or IFN-γ expression.
IL-2, IL-12, and PMA/Ionomycin Stimulation.
After IL-2 or IL-12 treatment, a large fraction of the CD4− CD1d tetramer positive cells expressed perforin, but the CD4+ subset was still almost completely perforin negative ( b and a). Hence, exposure to IL-2 or IL-12 alone is sufficient to upregulate perforin for much of the CD4−, but not the CD4+ CD1d-restricted subset. In contrast, PMA and ionomycin stimulation reproducibly induced a fraction of the CD4+ tetramer positive cells to express perforin, but did not enhance perforin staining in the CD4− subset ( c and a). Thus, perforin expression is induced by different stimuli in CD1d-restricted T cells of the CD4+ and CD4− subsets. CD1d-restricted T cells of the two subsets may therefore carry out cytolytic functions in response to different signals in vivo.
Figure 4. Mean percentages of CD1d tetramer positive cells staining positively for perforin and IFN-γ after stimulation. Plot a shows perforin staining, plot b shows IFN-γ staining, after stimulation as shown on the y-axis at the left. Gray bars (more ...)
Interestingly, treatment of the PBMC samples with LPS in the absence of monensin provided sufficient stimulation for CD1d-restricted T cells of both the CD4− and CD4+ subsets to up-regulate perforin expression ( a). However, perforin staining was not clearly enhanced by incubation with LPS in the presence of monensin ( a). Therefore, the effect of LPS may be due to expression of soluble and/or cell surface molecules that depend on intracellular transport. These results show that inflammatory conditions, including exposure to LPS which is an early indicator of gram negative bacterial infection, may prime a fraction of CD1d-restricted T cells for cytolytic function.
Unexpectedly, TCR stimulation by antigen led to a qualitatively different outcome than stimulation by cytokines or LPS. Treatment of PBMC samples with α-GalCer in the presence of monensin did not result in significantly increased numbers of perforin-positive CD1d tetramer positive cells ( d and 4 a). However, such α-GalCer treatment did induce IFN-γ production in both CD4− and CD4+ tetramer positive cells ( h and b). Remarkably, in contrast to their effects on perforin, IL-2, IL-12, and LPS completely failed to induce IFN-γ production in either subset of CD1d tetramer positive cells ( f and b). Thus, exposure to a lipid antigen resulted in IFN-γ production, but did not lead to up-regulation of intracellular perforin, whereas exposure to IL-2, IL-12, or LPS enhanced perforin staining, but did not induce IFN-γ production.
This result contrasts with two reports that found increased cytotoxicity or granzyme B expression in Vα24+
NKT cells cultured with α-GalCer (47
). However, in these studies the Vα24+
NKT cells were cultured with monocytes or monocyte-derived dendritic cells and treated with α-GalCer in the absence of monensin. In a murine model, NKT cells induced monocyte-derived cells to secrete IL-12 after addition of α-GalCer (49
). Hence, culture of CD1d-restricted T cells with α-GalCer and myeloid cells could result in secondary stimulation of cytotoxicity by IL-12 or other factors.
Overnight incubation of the PBMC samples with α-GalCer in the absence of monensin resulted in a dramatic (~80%) reduction in the number of CD1d tetramer positive cells detected compared with unstimulated samples (data not shown). This effect resembles the disappearance of CD1d-restricted T cells in vivo upon administration of α-GalCer that has been observed in a murine system (14
). Incubation of the PBMC samples with IL-2, IL-12, LPS, or PMA and ionomycin did not lead to a significant reduction in the number of CD1d tetramer positive cells detected. Moreover, when incubation of PBMC samples with α-GalCer was performed in the presence of monensin the number of CD1d tetramer positive cells was generally 80–100% of the number detected in unstimulated control samples. Hence, the disappearance of CD1d tetramer positive cells was specific to α-GalCer stimulation and depended on intracellular transport, suggesting it may result from secretion or cell surface expression of as yet unidentified factors.
Cytokine Receptor Expression.
To assess whether the lack of perforin up-regulation in response to IL-12 and IL-2 in the CD4+ CD1d-restricted T cell subset could be due to the absence of appropriate cytokine receptors, we evaluated expression of the IL-12 and IL-2 receptors by tetramer positive cells. The IL-12 receptor β1 chain was detected on cells of both the CD4− and CD4+ CD1d tetramer positive subsets (mean = 61 ± 6.8% and 57 ± 27% positive, respectively). Similarly, IL-2 receptor expression was detected on cells of both CD1d tetramer positive subsets, but expression of the intermediate- and high-affinity forms differed between the CD4+ and CD4− subsets. The IL-2 receptor α chain (CD25) was expressed by 26 ± 1.3% of the CD4+ tetramer positive cells, but by very few of the CD4− cells (mean = 7.0 ± 7.3%), while the IL-2 receptor β chain (CD122) was detected on a mean of 74 ± 8.2% of the CD4- tetramer positive cells, and on 59 ± 28% of the CD4+ cells. Hence, a fraction of the CD4+ CD1d tetramer positive cells expressed the high affinity IL-2 receptor (CD25+CD122), whereas most CD4− CD1d-restricted T cells expressed the intermediate affinity IL-2 receptor (CD122). Nonetheless, the lack of perforin expression in response to IL-12 and IL-2 by CD4+ tetramer positive cells was not simply due to an absence of the cytokine receptors, underscoring the finding that expression of intracellular perforin appears to be regulated differently in the CD4+ and CD4− CD1d-restricted T cell subsets.
Expression of Receptors Associated with Cell Killing.
NKG2d is a lectin encoded in the NK complex that is expressed by NK cells, γδ T cells, and CD8+
αβ T cells, that mediates or costimulates cytolysis of virally and bacterially infected or neoplastic cells that express certain stress-induced antigens (51
). We investigated CD1d-restricted T cell expression of cell surface NKG2d by two color flow cytometric analysis. As a directly conjugated anti-NKG2d antibody was not available, we compared PBMC samples depleted of CD4+
cells to CD4 undepleted samples to evaluate whether NKG2d expression was biased toward CD4+
CD1d-restricted T cells. In PBMC samples that were not depleted of CD4+
cells, approximately half of the CD1d tetramer positive cells stained positively for NKG2d ( a, top panel). In PBMC samples that were CD4 depleted, the fraction of CD1d tetramer-positive cells that were NKG2d-positive was increased ( a, bottom panel), suggesting that the CD4−
CD1d-restricted T cell subset is enriched for NKG2d expression compared with the CD4+
Figure 5. Flow cytometric probability contour plots showing NKG2d and CD95L staining of CD1d tetramer positive cells. The plots are gated on CD1d tetramer positive lymphocytes. a shows NKG2d staining of a sample before depletion of CD4+ cells (top panel), compared (more ...)
Taken together with our finding that exposure to IL-2, IL-12, or LPS results in enhanced perforin expression, this observation suggests CD4−
CD1d-restricted T cells could play a previously unrecognized role in microbial infections. Exposure to inflammatory conditions may serve to activate the cytotoxic functions of CD4−
CD1d-restricted T cells, and NKG2d expression could permit cytolysis of a broad range of virally or bacterially infected cells. Moreover, this observation could provide insight into the role of CD1d-restricted T cells in tumor rejection. CD1d-restricted T cells are required for the rejection of murine metastatic tumors induced by pharmacological administration of IL-12 or α-GalCer, and also appear to be involved in elimination of tumors mediated by endogenous IL-12 secretion (25
). Paradoxically, however, tumor rejection is not blocked by antibodies to CD1d (55
). Our results indicate IL-12–activated CD4−
CD1d-restricted T cells could recognize tumor cells via engagement of NKG2d rather than the TCR, and that cytolysis could thus be CD1d-independent.
Cytotoxicity mediated by NKT cells has also been associated with Fas/Fas ligand interactions (57
). We investigated FasL (CD95L) expression by CD1d-restricted T cells after stimulation of PBMC samples with IL-2, IL-12, α-GalCer, or PMA/ionomycin. Because CD95L is rapidly cleaved from the cell surface by serum metalloproteinases (58
), we used intracellular staining to detect expression. In unstimulated samples, there was little or no positive staining for CD95L among CD4−
tetramer positive cells (mean = 1.9 ± 3.4%), but slightly more positive staining of the CD4+
subset (mean = 5.6 ± 5.2%). (, top panel). Treatment with α-GalCer, IL-2, or IL-12 did not enhance CD95L staining for either subset (data not shown). In contrast, after PMA/ionomycin stimulation a mean of 28 ± 19% of the CD4+
subset, and 7.5 ± 8.1% of the CD4−
subset, were positive for CD95L (see , bottom panel). Hence, mainly CD4+
CD1d-restricted T cells could be induced to express CD95L, suggesting that regulatory functions mediated by Fas/FasL interactions may be performed mostly by the CD4+
CD1d-restricted subset in vivo.
CD1d-restricted T Cell Homing Receptors.
Elevated numbers of T cells expressing Vα24/JαQ TCRs have recently been demonstrated in two inflammatory sites, chronic inflammatory demyelinating polyneuropathy lesions and periodontitis lesions, suggesting CD1d-restricted T cells may migrate preferentially to areas of inflammation (59
). To investigate the possible destinations of CD1d-restricted T cells in peripheral blood, we stained for a series of homing and chemokine receptors. In contrast to cytokine production and perforin upregulation, chemokine and homing receptor expression did not correlate clearly with CD4 expression by CD1d tetramer positive cells (). Most tetramer positive cells were negative for CD62L and CCR7, which are receptors involved in trafficking to lymph nodes (). A fraction of the CD1d tetramer positive cells stained positively for the integrin α4β7, a molecule that is expressed on effector cells that home to the gut and associated lymphoid sites, and a fraction was positive for the CLA, which is associated with homing to skin (). The tetramer positive cells were almost completely negative for the integrin αEβ7, which is expressed by intraepithelial lymphocytes and a small percentage of peripheral blood T cells (). Two chemokine receptors that are associated with Th1 responses and migration to sites of inflammation, CCR5 and CXCR3, were expressed by large percentages of CD1d tetramer positive cells (). Almost all tetramer positive cells stained positively for CXCR4, a chemokine receptor that recognizes a broadly distributed ligand, stromal cell–derived factor 1α. Hence, most CD1d tetramer positive cells in peripheral blood had a phenotype consistent with homing to peripheral tissues and recruitment to sites of inflammation.
Chemokine and Homing Receptor Expression by CD1d Tetramer-stained T Cells
Our results emphasize that CD1d-restricted T cells may be important cytolytic as well as cytokine-producing effector cells that migrate to peripheral sites of inflammation or immunological activity. A large percentage of the CD4− CD1d tetramer positive cells appeared to become primed for cytolytic function by exposure to IL-2 and IL-12, and this subset was also primarily oriented toward secretion of Th1 cytokines. Hence, CD4− CD1d-restricted T cells had characteristics associated with activation of cell-mediated effector functions and cytolysis, a profile resembling NK cells. Their up-regulation of perforin in response to inflammatory stimuli and expression of NKG2d suggests CD4− CD1d-restricted T cell may also resemble NK cells by performing cytolytic effector functions in antibacterial, antiviral, and antitumor immune responses.
In contrast, CD4+ CD1d-restricted T cells may be more oriented toward providing B cell help or immunoregulatory functions, as this subset accounted for almost all Th2 cytokine production by CD1d-restricted T cells, and could be induced to express CD95L. Based on their potent Th2 cytokine production, CD4+ CD1d-restricted T cells might be predicted to be responsible for the protective effect of NKT cells observed in autoimmune diabetes. An intriguing further possibility, however, is that CD4− CD1d-restricted T cells could play a pathogenic role in autoimmune disease by mediating cytotoxicity in response to inflammatory conditions, and/or by secreting Th1 cytokines in response to self antigen recognition.
We also show that antigenic- and cytokine-mediated stimulation can have profoundly different effects on CD1d-restricted T cells, and that the CD4−
CD1d-restricted T cell populations respond differently to these stimuli. The principal APCs that have been found to express CD1d in humans are B cells and myeloid cells, in particular monocytes, macrophages, and dendritic cells (61
). Fresh B cells do not secrete IL-2 or IL-12, whereas monocytes can produce IL-12, and activated T cells may secrete IL-2 locally. Thus, our results suggest three potential functional outcomes for CD4−
CD1d-restricted T cell activation: (i) those that recognize antigens presented by CD1d+
B cells may become activated for cytokine secretion but not cytotoxicity; (ii) those that receive antigenic stimulation from monocytes could be primed for cytolysis in addition to cytokine secretion; and (iii) exposure to IL-2, IL-12, or inflammatory agents such as LPS, in the absence of antigenic stimulation could lead to cytolytic activation without cytokine production. Hence, in addition to recognition of specific antigens, the type of APC and the local cytokine environment are likely to be critical factors that regulate the effector functions of CD1d-restricted T cells.