Rare CD1d-α-galactosylceramide–specific T cells that do not express the invariant Vα24 chain of human NKT cells were recently identified after expansion in vitro with the lipid Ag, but their phenotype and frequency in vivo and lineage relationship with NKT cells could not be elucidated. By using a CD1d tetramer-based method to enrich these cells from fresh peripheral blood, we demonstrated their naive-like CD62LhighCD45ROCD4+ phenotype and relatively high frequency of ∼10−5 in several healthy individuals. Notably, these cells expressed the NKT lineage-specific transcription promyelocytic leukemia zinc finger (PLZF), indicating a developmental relationship with NKT cells and ruling out the possibility that they were conventional MHC-restricted T cells cross-reacting against CD1d-α-galactosylceramide. Although PLZF is known to direct the effector program of NKT cells, we show in this study that the naive-like cells expressed it at a significantly lower amount than NKT cells. Further, we present mouse studies demonstrating a sharp PLZF expression threshold requirement for induction of the effector phenotype. These findings directly demonstrate in vivo the existence of naive-like CD1d-restricted human T cells marked by intermediate levels of PLZF.

T cell lines recognizing various self- and foreign lipid Ags presented by glycoproteins of the CD1 family, CD1a, -b, -c, and -d in vitro have been described, but their lineages remain poorly defined in vivo (13). One exception is the prominent and well-characterized NKT cell population of CD1d-restricted T cells, which express conserved semi-invariant TCRs, mostly mVα14-Jα18/Vβ2, -7, and -8 in mice and hVα24-Jα18/Vβ11 in humans (h). These cells are readily identified in vivo using CD1d tetramers loaded with the synthetic Ag α-galactosylceramide (αGalCer), a mimetic of α-branched microbial glycolipids. They constitutively express innate-like effector properties that are naturally induced during thymic development, in the absence of foreign Ags, by the signature transcription factor promyelocytic leukemia zinc finger (PLZF) (4, 5).

Attempts to directly identify other CD1-restricted T cells, particularly naive unexpanded T cells, using tetramers loaded with foreign lipids have been unsuccessful to date. There are inherent difficulties with this approach. Estimates of the frequency of naive T cells specific for any given Ag are on the order of 10−5 for human MHC-restricted T cells, a frequency at which FACS analysis alone is not considered reliable. MACS-enrichment methods using beads coated with Abs against the fluorochrome attached to the tetramers have dramatically improved the ability to identify such rare cells, because their frequency can be enriched up to 1000-fold, reaching percentages of 0.1–10% that are more suitable for FACS studies. Nevertheless, it often remains difficult to provide appropriate positive and negative controls to validate these analyses. The T cells obtained after tetramer staining and MACS sorting tend to rapidly die in culture, likely because of extensive TCR cross-linking, and cannot be tested functionally or expanded efficiently to confirm their reactivity to Ag. As the cells represent only a few percent of the cell population recovered after tetramer-based enrichment in vitro, it is also important to have independent markers to support their bona fide representation of the rare in vivo precursors.

In this study, we have sought to overcome these difficulties by examining a rare population of CD1d-restricted T cells specific for the same αGalCer Ag as NKT cells but lacking the canonical invariant Vα24 chain. Previous studies have shown that these cells readily expanded in cultures of human PBMC stimulated with αGalCer and could generate αGalCer-responsive clones. Because these lines exhibited a diverse TCR repertoire and lacked expression of CD161, one of the NKT lineage markers, their precursors might resemble more naive MHC-restricted T cells than innate-like effector NKT cells. Of interest, a majority of these T cell clones expressed the same Vβ11 gene as NKT cells. As only a few percent of mainstream T cells express Vβ11, we reasoned that this singular feature would provide a robust independent method to confirm the tetramer-based identification of the fresh in vivo precursors of Vα24-negative αGalCer-specific T cells.

Using these methods, our studies of healthy human PBMC revealed a sizeable frequency, ∼10−5, and a naive-like CD62Lhigh CD45ROCD45RA+CCR7+CD161CD4+ cell-surface phenotype for these Vα24-negative CD1d-αGalCer–specific T cells, apparently undistinguishable from conventional naive MHC class II-restricted T cells. Surprisingly, however, these cells expressed PLZF, the NKT lineage-specific transcription factor, indicating a lineage relationship with NKT cells and ruling out the possibility that they were merely conventional MHC-restricted T cells with a CD1d-αGalCer cross-reactive TCR. Interestingly, these naive-like CD1d-restricted T cells expressed significantly lower amounts of PLZF than NKT cells. Further studies in the mouse system established that a high threshold of PLZF expression was required for effector differentiation.

Thus, our studies establish the existence in human peripheral blood of naive-like CD1d-restricted T cells expressing intermediate levels of PLZF and provide strong support to the notion that CD1 molecules can select both naive and effector T cells during thymic development.

Blood was drawn from 22–54-y-old healthy volunteers in accordance with a protocol approved by the University of Chicago’s Institutional Review Board.

PLZF−/− mice carrying an induced deletion of exon 2 of the Zbtb16 gene (6) also in the B6 background were a gift from Dr. P.P. Pandolfi (Beth Israel Deaconess Cancer Center, Boston, MA). The previously described B6.PLZF-transgenic line #1797 (driven by the CD4 promoter) (4) was produced in the same injection series as the other founder lines #1960 and #1963. Animals were 4–8 wk of age when studied and compared with littermate controls. Mice were housed in a specific pathogen-free environment at the University of Chicago, and experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee.

CD1d-αGalCer tetramers were prepared as described previously (7, 8) using the αGalCer analog PBS-57 (8). Fluorochrome-labeled mAbs (clone indicated in parentheses) against human CD3 (HIT3A or UCHT1), CD4 (RPA-T4 or OKT4), CD8α (RPA-T8), CD45RA (HI100), CD62L (DREG-56), CD161 (DX12 or HP-3G10), CD197 (TG8/CCR7), Vβ2 (MPB2D5), Vβ11 (C21), and Vα24 (C15) were purchased from BD Biosciences, Beckman Coulter, BioLegend, eBioscience, or Pierce Thermo Scientific. Fluorochrome-labeled mAbs against mouse B220 (RA3-6B2), CD4 (GK1.5 or RM4-5), CD8α (53-6.7), CD24 (M1/69), CD44 (IM7), CD62L (MEL-14), and TCRβ (H57-597) were purchased from BD Biosciences, BioLegend, or Invitrogen.

Human PBMC were isolated from the blood of healthy donors after centrifugation over Ficoll-Paque Plus (GE Healthcare). Mouse lymphocytes were obtained from thymus, spleen, and peripheral lymph nodes (inguinal and axillary) by passing organs through a nylon mesh. Prior to Ab staining, human and mouse samples were blocked with human serum and purified anti-mouse CD16/32 (clone 93; BioLegend), respectively. For enrichment, PBMC were labeled with allophycocyanin-conjugated hCD1d-αGalCer tetramers, bound by anti-allophycocyanin magnetic MicroBeads (Miltenyi Biotec), and enriched on an autoMACS cell separator (Miltenyi Biotec), as previously described (9). Samples were analyzed on an LSRII (BD Biosciences) or sorted on an FACSAria (BD Biosciences) or MoFlo (DakoCytomation). Sorted samples were mixed with 20,000 unstained mouse Jα18−/− thymocytes to aid cell pelleting and reduce cell losses during the intracellular staining procedure. Most experiments used doublet exclusion and DAPI (Invitrogen) staining of dead cells. Data were analyzed with FlowJo (Tree Star).

For intracellular PLZF staining, the D-9 anti-PLZF mouse mAb (Santa Cruz Biotechnology) was conjugated to Alexa Fluor 546 using a labeling kit (Invitrogen). Following extracellular labeling, samples were fixed and permeabilized using the Foxp3 Staining Buffer Set (eBioscience), blocked with purified mouse IgG1 (clone P3; eBioscience), and stained with AF546-conjugated anti-PLZF.

All in vitro experiments were performed using RPMI 1640 medium containing 2 mM l-glutamine (Cellgro), 1 mM sodium pyruvate (Invitrogen), 1 mM nonessential amino acids (Invitrogen), 55 μM 2-ME, 8% FCS (Biowest), 2% AB human serum (Atlanta Biologicals), and penicillin-streptomycin (Sigma-Aldrich). A total of 2 × 106 freshly isolated PBMC were plated in 24-well plates at a density of 1 × 106 cells/ml and cultured in the presence of 100 ng/ml αGalCer analog PBS-57 (8) and 100 U/ml human IL-2. Cultures were split when necessary and fed every 3 to 4 d with fresh medium containing IL-2.

CD3+CD1d-αGC+Vα24+, CD3+CD1d-αGC+Vα24, and CD4+CD3+CD1d-αGC cells were single-cell sorted from PBMC of three healthy donors using an FACSAria (BD Biosciences) or MoFlo (DakoCytomation) and plated in 96-well plates. Sorted single cells were stimulated with feeder cells consisting of irradiated (2500 rad) allogenic PBMC (50,000/well) and irradiated (5000 rad) EBV-transformed cells (5000/well) in medium containing 1 μg/ml PHA and 100 U/ml human IL-2. Proliferating clones were maintained by restimulation every 3 wk, and reactivity to CD1d-αGC was verified by tetramer staining.

Total RNA was extracted from PBMC clones with a combination of TRIzol (Invitrogen) and the RNeasy Micro Kit (Qiagen) and reverse-transcribed with SuperScript III reverse transcriptase (Invitrogen) and oligo(dT). TCRα sequences were PCR amplified from cDNA using Platinum Taq DNA polymerase (Invitrogen) and the oTCA4 C region primer paired with a V region primer other than oVA24, as described previously (10). TCRβ sequences of Vβ11 clones were amplified with the CB C region primer and a V region primer other than BV11, whereas the TCRβ sequences of Vβ11+ clones were amplified using the C region primer and BV11 (11). Following amplification, PCR products were gel purified and subcloned using the TOPO TA Cloning Kit (Invitrogen). Single colonies were inoculated, DNA-prepared, and sequenced by the University of Chicago DNA Sequencing Facility using a 3730xl DNA Analyzer (Applied Biosystems). Sequencing results were aligned to known V and J genes using the ImMunoGeneTics V-QUEST program (http://www.imgt.org/IMGT_vquest/share/textes/) (12).

Total RNA was isolated from resting PBMC clones using a combination of TRIzol (Invitrogen) and the RNeasy Mini Kit (Qiagen) and reverse-transcribed with random primers using the AffinityScript qPCR cDNA Synthesis Kit (Stratagene). Transcripts for human PLZF were quantified with primers spanning intron 3 (forward primer: 5′-TAGTTTGCGGCTGAGAATGC-3′; reverse primer: 5′-ACCGCACTGATCACAGACAAAG-3′) and normalized to hypoxanthine phosphoribosyltransferase (forward primer: 5′-GAAAGGGTGTTTATTCCTCATGG-3′; reverse primer: 5′-CTCCCATCTCCTTCATCACATC-3′). Quantitative PCR was performed on an Mx3005p system (Stratagene) using Brilliant SYBR Green qPCR Master Mix (Stratagene).

Total RNA was isolated from either resting (18 d following stimulation) or stimulated (3 d following stimulation) PBMC clones using a combination of TRIzol (Invitrogen) and the RNeasy Mini Kit (Qiagen). The resulting RNA was processed by the University of Chicago Functional Genomics Facility and hybridized to GeneChip Human Genome U133 Plus 2.0 arrays (Affymetrix). Arrays were scanned by a GeneChip Scanner 3000 (Affymetrix), and intensity values were generated by MicroArray Suite 5.0 software (Affymetrix) and scaled to a median intensity of 500 prior to analysis. Gene expression data are available at the Gene Expression Omnibus under the accession number GSE28726 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE28726).

The unpaired, two-tailed t test was used for all statistical calculations and performed with Prism (GraphPad).

Fig. 1 shows that, as previously reported (13, 14), a rare population of Vα24CD1d-αGalCer+ T cells, which is barely seen in whole PBMC, can preferentially expand upon stimulation with αGalCer in vitro and typically express a weaker tetramer staining than canonical Vα24+ NKT cells. Studies of 18 clones derived from three healthy individuals (#1, #2, and #4) are presented in Table I as specific background for the fresh precursor characterization presented below. The findings confirmed the expression of a diverse set of Vα-chains, the overrepresentation of other components of the canonical NKT TCR, including the Jα18 segment and the Vβ11 chain, and the predominant CD4 phenotype.

FIGURE 1.

In vitro expansion of CD1d-αGC–specific PBMC. Fresh human PBMC before and after stimulation (day 18) with 100 ng/ml αGalCer analog PBS-57. Plots are pregated on CD3+ cells, and the percentages of hCD1d-αGC+ Vα24+ and hCD1d-αGC+Vα24 cells are indicated. Data are representative of four individuals in six independent experiments.

FIGURE 1.

In vitro expansion of CD1d-αGC–specific PBMC. Fresh human PBMC before and after stimulation (day 18) with 100 ng/ml αGalCer analog PBS-57. Plots are pregated on CD3+ cells, and the percentages of hCD1d-αGC+ Vα24+ and hCD1d-αGC+Vα24 cells are indicated. Data are representative of four individuals in six independent experiments.

Close modal
Table I.
TCR chain usage and coreceptor expression by CD1d-αGC+Vα24 cells
CloneCD4Vα24Vβ11TRAVTRAJTRBV
1-1 − 12-3 18 25-1 
1-2 − 18 25-1 
1-3 − − 1-1 27 
1-4 − 27 22 25-1 
1-5 − NI NI 25-1 
1-6 − 12-3 18 25-1 
1-7 − 12-3 18 25-1 
1-8 − − 27 18 25-1 
1-9 − 35 18 25-1 
2-1 − − NI NI 6-1 
2-2 − − 12-3 18 25-1 
2-3 − 17 18 25-1 
4-1 − − NI NI NI 
4-2 − − 17 7-2 
4-3 − 17 18 25-1 
4-4 − 8-1 23 25-1 
4-5 − 35 18 25-1 
4-6 − − 1-1 30 20-1 
CloneCD4Vα24Vβ11TRAVTRAJTRBV
1-1 − 12-3 18 25-1 
1-2 − 18 25-1 
1-3 − − 1-1 27 
1-4 − 27 22 25-1 
1-5 − NI NI 25-1 
1-6 − 12-3 18 25-1 
1-7 − 12-3 18 25-1 
1-8 − − 27 18 25-1 
1-9 − 35 18 25-1 
2-1 − − NI NI 6-1 
2-2 − − 12-3 18 25-1 
2-3 − 17 18 25-1 
4-1 − − NI NI NI 
4-2 − − 17 7-2 
4-3 − 17 18 25-1 
4-4 − 8-1 23 25-1 
4-5 − 35 18 25-1 
4-6 − − 1-1 30 20-1 

V and J regions of TCRα-chains and V regions of TCRβ-chains of 18 CD1d-αGC+Vα24 clones obtained from three individuals were determined by sequencing. CD4, Vα24, and Vβ11 expression were assessed by flow cytometry. Note that Vβ11 corresponds to TRBV 25-1 in the gene nomenclature.

NI, not identified.

To directly examine the fresh precursors of these CD1d-αGalCer–specific cells in vivo, we used a method of MACS enrichment based on tetramer staining to identify the rare cells that bound the tetramers, yet lacked Vα24 (Fig. 2A). In four individuals (#1, #2, #3, and #4) examined, these cells could be sufficiently enriched from PBMC for meaningful analyses. They represented a variable proportion of the total tetramer-positive population, but were present at a relatively stable frequency between 4 × 10−6 and 1 × 10−5 among total CD3+ PBMC in repeat examinations over a time period ranging from 1 mo to 2 y (Table II). In contrast, as previously reported (15), the frequency of Vα24+ CD1d-αGalCer+ NKT cells in circulating blood varied widely among these individuals, ranging between 1 × 10−4 and 5 × 10−3, although it remained stable in each individual over time.

FIGURE 2.

Phenotype of CD1d-αGC+Vα24 PBMC after MACS enrichment. A, Fresh human PBMC were MACS-enriched with CD1d-αGC tetramers prior to FACS analysis. The middle and bottom panels present the phenotype of the CD1d-αGC–positive and –negative populations, respectively, with numbers representing frequencies among upper quadrants (Vα24+) and lower quadrants (Vα24). Summary data of four individuals are presented in Table II. B, Fresh human PBMC were MACS-enriched with CD1d-αGC tetramers and then FACS-sorted into tetramer+Vα24 and tetramer+ Vα24+ populations, as indicated, prior to intranuclear staining for PLZF. The CD4 population represents tetramerCD4+ cells from the same individual. The MFI of PLZF is indicated for each population. C, Summary data as in B for three different individuals.

FIGURE 2.

Phenotype of CD1d-αGC+Vα24 PBMC after MACS enrichment. A, Fresh human PBMC were MACS-enriched with CD1d-αGC tetramers prior to FACS analysis. The middle and bottom panels present the phenotype of the CD1d-αGC–positive and –negative populations, respectively, with numbers representing frequencies among upper quadrants (Vα24+) and lower quadrants (Vα24). Summary data of four individuals are presented in Table II. B, Fresh human PBMC were MACS-enriched with CD1d-αGC tetramers and then FACS-sorted into tetramer+Vα24 and tetramer+ Vα24+ populations, as indicated, prior to intranuclear staining for PLZF. The CD4 population represents tetramerCD4+ cells from the same individual. The MFI of PLZF is indicated for each population. C, Summary data as in B for three different individuals.

Close modal
Table II.
Frequency and phenotype of CD1d-αGC+Vα24 cells in vivo
Individual No.PopulationCD3+ (%)CD4+ (%)CD8α+ (%)CD45RA+ (%)CD62L+ (%)CCR7+ (%)CD161+ (%)Vβ11+ (%)Vβ2+ (%)
Tetr+Vα24+ 0.012 ± 0.0038 ND 43 13 ± 3.4 43 ± 7.2 ND 60 ± 15 97 ± 2 0.093 ± 0.13 
 Tetr+Vα24 0.0011 ± 0.00051 ND 26 73 ± 20 86 ± 5.2 ND 5.1 ± 2.6 63 ± 18 7.4 ± 3.7 
 Tetr 100 ± 0.0041 ND 41 72 ± 9.2 65 ± 11 ND 9.7 ± 6.6 5.1 ± 5.6 8.3 ± 0.082 
Tetr+Vα24+ 0.46 ± 0.018 ND 12 4.1 ± 4.5 6.8 ± 4.5 13 91 ± 3.1 97 ± 2.8 0.044 ± 0.068 
 Tetr+Vα24 0.001 ± 0.00011 71 ± 5.9 16 ± 6.7 62 ± 16 61 ± 13 69 28 ± 3.4 59 ± 23 3.6 ± 1.5 
 Tetr 100 ± 0.018 72 ± 0.35 20 ± 7.1 59 ± 7.1 73 ± 4.2 63 9.6 ± 0.62 6.2 ± 3.5 4.9 ± 3.7 
Tetr+Vα24+ 0.016 ± 0.0032 58 ± 3.3 31 14 35 ± 17 21 73 ± 2 100 ± 0.085 0.78 
 Tetr+Vα24 0.00043 ± 0.00009 72 ± 4.8 28 64 79 ± 11 67 18 ± 17 63 ± 9.7 
 Tetr 100 ± 0.0033 66 ± 1.5 32 58 66 ± 17 80 6.2 ± 3.4 0.97 ± 0.044 4.1 
Tetr+Vα24+ ND ND ND 10 16 ND ND 100 0.016 
 Tetr+Vα24 ND ND ND 86 78 ND ND 47 1.2 
 Tetr ND ND ND 67 82 ND ND 7.6 6.2 
Individual No.PopulationCD3+ (%)CD4+ (%)CD8α+ (%)CD45RA+ (%)CD62L+ (%)CCR7+ (%)CD161+ (%)Vβ11+ (%)Vβ2+ (%)
Tetr+Vα24+ 0.012 ± 0.0038 ND 43 13 ± 3.4 43 ± 7.2 ND 60 ± 15 97 ± 2 0.093 ± 0.13 
 Tetr+Vα24 0.0011 ± 0.00051 ND 26 73 ± 20 86 ± 5.2 ND 5.1 ± 2.6 63 ± 18 7.4 ± 3.7 
 Tetr 100 ± 0.0041 ND 41 72 ± 9.2 65 ± 11 ND 9.7 ± 6.6 5.1 ± 5.6 8.3 ± 0.082 
Tetr+Vα24+ 0.46 ± 0.018 ND 12 4.1 ± 4.5 6.8 ± 4.5 13 91 ± 3.1 97 ± 2.8 0.044 ± 0.068 
 Tetr+Vα24 0.001 ± 0.00011 71 ± 5.9 16 ± 6.7 62 ± 16 61 ± 13 69 28 ± 3.4 59 ± 23 3.6 ± 1.5 
 Tetr 100 ± 0.018 72 ± 0.35 20 ± 7.1 59 ± 7.1 73 ± 4.2 63 9.6 ± 0.62 6.2 ± 3.5 4.9 ± 3.7 
Tetr+Vα24+ 0.016 ± 0.0032 58 ± 3.3 31 14 35 ± 17 21 73 ± 2 100 ± 0.085 0.78 
 Tetr+Vα24 0.00043 ± 0.00009 72 ± 4.8 28 64 79 ± 11 67 18 ± 17 63 ± 9.7 
 Tetr 100 ± 0.0033 66 ± 1.5 32 58 66 ± 17 80 6.2 ± 3.4 0.97 ± 0.044 4.1 
Tetr+Vα24+ ND ND ND 10 16 ND ND 100 0.016 
 Tetr+Vα24 ND ND ND 86 78 ND ND 47 1.2 
 Tetr ND ND ND 67 82 ND ND 7.6 6.2 

The frequency of CD1d-αGC+Vα24 cells within CD3+ PBMC is shown for different individuals. Their phenotype is compared with those of CD1d-αGC+Vα24+ and CD1d-αGC cells. Data shown as mean ± SD of two to four independent blood samples for each individual, except individual 4, who was examined only once.

ND, not determined; Tetr, tetramer.

In agreement with studies of cultured cells (Table I) (13, 14), the fresh Vα24CD1d-αGalCer+ cells exhibited an overrepresentation of Vβ11, reaching 47–63% on average, unambiguously confirming that they were the bona fide counterparts of the Vα24-negative clones (Fig. 2A, Table II). In marked contrast with the Vα24+CD1d-αGalCer+ NKT cells, the majority of Vα24CD1d-αGalCer+ expressed a CD45RA+CCR7+CD62Lhigh phenotype characteristic of naive lymph node-seeking T cells, and they conspicuously lacked the CD161 (NKR-P1A) receptor. Thus, although only a limited number of key surface markers could be examined due to the rarity of these cells, their predominantly naive phenotype stood in stark contrast with the effector phenotype of NKT cells. The Vα24-negative cells mainly expressed CD4, as previously reported for their in vitro-derived clones, although in some individuals, a substantial fraction was found to express CD8α (Table II).

The expression of the transcription factor PLZF in Vα24CD1d-αGalCer+ cells was directly examined by intranuclear FACS staining. Because of the cell losses associated with the permeabilization/fixation procedure in samples containing few cells, the human Vα24CD1d-αGalCer+ cells were FACS-sorted after MACS enrichment and mixed with Jα18−/− mouse filler cells prior to intranuclear staining of PLZF. In three out of three individuals examined, PLZF expression was detected well above the background level of control CD4 cells (Fig. 2B, 2C). Notably, however, the amount of PLZF was significantly reduced by ∼50% when compared side by side with the Vα24-positive NKT cells from the same individuals in the same experiments.

PLZF mRNA was quantitated in clones simultaneously derived from Vα24CD1d-αGalCer+ (n = 16), Vα24+CD1d-αGalCer+ (n = 12), and control CD4 (n = 15) PBMC from the three individuals shown in Table I. Fig. 3A shows PLZF levels measured by quantitative RT-PCR in resting clones harvested 18 d after stimulation with PHA and IL-2. All CD1d-αGalCer+ cell clones, whether Vα24 or Vα24+, expressed high PLZF mRNA, whereas 13 out of 15 control CD4 cell clones derived from the same three healthy donors did not. As in the fresh cell experiments, the Vα24CD1d-αGalCer+ clones tended to express less PLZF than the Vα24+ CD1d-αGalCer+, although the 37% average difference did not quite reach statistical significance (p = 0.084). As shown below, this may be due to the natural variations of PLZF levels in cultured cells over time.

FIGURE 3.

CD1d-αGC+Vα24 T cell clones express intermediate amounts of PLZF. PLZF mRNA expression was quantitated by quantitative RT-PCR (A) and microarray (B) experiments. A, Each data point represents the mean of triplicate PCR values of an individual resting clone harvested 18 d after stimulation with PHA and IL-2. Control (tetramer) CD4+ clones, tetramer+Vα24 clones, and tetramer+Vα24+ clones were derived from each of three individuals, and a total of 15, 16 and 12 clones of each respective phenotype are represented in the figure. B, Four to five randomly selected clones of each category were further analyzed by microarray 18 d poststimulation (resting) or 3 d after PHA and IL-2 (stimulated). The normalized expression levels of the 205883_at PLZF probe set from individual Affymetrix GeneChip Human Genome U133 Plus 2.0 arrays (Affymetrix) are represented in a two-dimensional plot showing the PLZF level in stimulated (ordinate) and resting (abscissa) clones. The differences in PLZF expression between tetramer+Vα24 clones and tetramer+Vα24+ clones were statistically significant (p < 0.05) whether examined at the resting or activated state. Note that most of the dots fell below the diagonal: a linear regression of all data points had a slope of 0.503, significantly different from 1 (p < 0.0001), indicating downregulation of PLZF upon activation.

FIGURE 3.

CD1d-αGC+Vα24 T cell clones express intermediate amounts of PLZF. PLZF mRNA expression was quantitated by quantitative RT-PCR (A) and microarray (B) experiments. A, Each data point represents the mean of triplicate PCR values of an individual resting clone harvested 18 d after stimulation with PHA and IL-2. Control (tetramer) CD4+ clones, tetramer+Vα24 clones, and tetramer+Vα24+ clones were derived from each of three individuals, and a total of 15, 16 and 12 clones of each respective phenotype are represented in the figure. B, Four to five randomly selected clones of each category were further analyzed by microarray 18 d poststimulation (resting) or 3 d after PHA and IL-2 (stimulated). The normalized expression levels of the 205883_at PLZF probe set from individual Affymetrix GeneChip Human Genome U133 Plus 2.0 arrays (Affymetrix) are represented in a two-dimensional plot showing the PLZF level in stimulated (ordinate) and resting (abscissa) clones. The differences in PLZF expression between tetramer+Vα24 clones and tetramer+Vα24+ clones were statistically significant (p < 0.05) whether examined at the resting or activated state. Note that most of the dots fell below the diagonal: a linear regression of all data points had a slope of 0.503, significantly different from 1 (p < 0.0001), indicating downregulation of PLZF upon activation.

Close modal

To test the stability of PLZF expression in culture, we examined a randomly selected set of clones 3 d after PHA and IL-2 stimulation when the cells were blasting or 18 d after when they were resting. In these experiments, the differences between the Vα24CD1d-αGalCer+ and the Vα24+CD1d-αGalCer+ clones reached statistical significance in both the resting and stimulated states. Note, however, that the clones showed a uniform tendency to decrease their levels of PLZF by 2-fold on average after stimulation, as shown by the significant deviation from the diagonal in Fig. 3B, indicating that PLZF expression fluctuated according to the rested or stimulated status of cells in culture. These experiments were performed on clones derived immediately after the first round of PBMC stimulation with αGalCer in vitro, and it is not known whether the cultured cells would maintain PLZF expression after multiple rounds of stimulation.

Altogether, these findings establish the existence of a T cell population with a naive-like phenotype and a diverse TCR repertoire specific for CD1d associated with the foreign synthetic Ag αGalCer. These cells also expressed PLZF, the master regulator of the effector phenotype in the innate-like NKT cell lineage, albeit at a reduced level.

Although seemingly modest, the 2-fold reduction of PLZF protein might explain the lack of effector phenotype in the Vα24CD1d-αGalCer+ cells by comparison with their Vα24+ counterparts. Indeed, in the mouse system, we observed haploinsufficiency for PLZF as mice heterozygous for exon 2 deletion (+/−) exhibited a ∼5-fold reduction of NKT cell frequency compared with +/+ littermates (Fig. 4A, 4B). Approximately half of the residual CD1d-αGalCer+ cells reverted to a naive CD62Lhigh phenotype, a portion of which also downregulated CD44 (Fig. 4C).

FIGURE 4.

PLZF heterozygous mice show NKT cell defects. The frequency of thymic (A) and splenic (B) NKT cells among all lymphocytes was assessed in PLZF−/−, PLZF+/−, and WT littermates. NKT cells were identified as CD1d-αGalCer+CD24low cells in the thymus and as CD1d-αGalCer+TCRβ+ cells in the spleen. C, Splenic NKT cells stained for CD44 and CD62L. Data are representative of four to five mice of each genotype analyzed in three independent experiments.

FIGURE 4.

PLZF heterozygous mice show NKT cell defects. The frequency of thymic (A) and splenic (B) NKT cells among all lymphocytes was assessed in PLZF−/−, PLZF+/−, and WT littermates. NKT cells were identified as CD1d-αGalCer+CD24low cells in the thymus and as CD1d-αGalCer+TCRβ+ cells in the spleen. C, Splenic NKT cells stained for CD44 and CD62L. Data are representative of four to five mice of each genotype analyzed in three independent experiments.

Close modal

We next examined the degree of effector conversion imparted by various levels of PLZF in transgenic lines expressing PLZF under the control of the CD4 promoter. A previous report suggested a dose effect of a PLZF transgene but did not directly compare the level of PLZF protein expression achieved with that of fresh NKT cells (16). In line 1797, where heterozygous expression exactly matched the natural levels of PLZF expressed by NKT cells during their effector conversion at development stages 1 and 2, as previously reported (4), all splenic CD4 T cells had downregulated CD62L, and many also upregulated CD44 compared with wild-type (WT) (Fig. 5A). Thus, <1% of transgenic CD4 cells expressed the naive CD62LhighCD44low phenotype compared with 80% of WT. In line 1963, where heterozygous expression of PLZF reached 25% on average of the level expressed in the heterozygous 1797 line, the CD62L/CD44 phenotype was comparable to WT. Interestingly, homozygous expression of the PLZF transgene in the same line, which induced a mean fluorescence intensity (MFI) equivalent to 34% of the level expressed in heterozygous 1797 line on average, brought about a very modest downregulation of CD62L and upregulation of CD44. In heterozygous 1960 mice, expression showed a variegate phenotype. One fraction expressed 62% of the PLZF levels of the heterozygous 1797 on average and displayed full conversion to the CD62Llow/CD44high/int phenotype. The other fraction, which expressed only 34% on average of the PLZF amount found in heterozygous 1797, did not change its naive phenotype. Fig. 5B shows how CD4 T cells populated the peripheral lymph nodes of these mice. The frequency of CD4 T cells varied according to their level of PLZF expression and extent of CD62L downregulation. Thus, a partial drop of the frequency of CD4 T cells from 59 to 34% on average was observed in line 1960 compared with WT, whereas the highest PLZF-expressing line 1797 had only 3% CD4 cells on average among lymph node T cells. These data establish that, as judged by changes in surface expression of CD62L and CD44, a high threshold of PLZF is required for acquisition of the effector phenotype, with drastic functional consequences on the recirculation pattern of CD4 T cells.

FIGURE 5.

A sharp threshold for effector conversion in transgenic lines expressing different levels of PLZF. A, TCRβ+CD4+ splenocytes from three different lines of PLZF transgenic mice compared with WT littermates for expression PLZF versus CD62L or CD44 versus CD62L. MFI of PLZF staining in each mouse is shown, with split values shown in line 1960 because of variegated expression. Note that the intensity of CD62L in the PLZF stained samples (top panel) is decreased compared with the CD44 stained samples (bottom panel) because of the fixation/permeabilization procedure. B, Frequency of TCRβ+CD4+ cells in peripheral lymph nodes of WT and PLZF transgenic mice. Plots are pregated on B220 cells. On average, the 1960 heterozygotes (n = 3) and 1797 heterozygotes (n = 4) had significantly decreased frequencies of CD4 T cells at 34 and 3%, respectively, compared with WT littermates (n = 5) (p < 0.0005), whereas the heterozygous (n = 5) and homozygous 1963 (n = 5) mice showed no significant changes. Data are representative of three to six mice of each genotype analyzed in three or more independent experiments. The WT, 1963 heterozygous, and 1963 homozygous mice depicted are littermates.

FIGURE 5.

A sharp threshold for effector conversion in transgenic lines expressing different levels of PLZF. A, TCRβ+CD4+ splenocytes from three different lines of PLZF transgenic mice compared with WT littermates for expression PLZF versus CD62L or CD44 versus CD62L. MFI of PLZF staining in each mouse is shown, with split values shown in line 1960 because of variegated expression. Note that the intensity of CD62L in the PLZF stained samples (top panel) is decreased compared with the CD44 stained samples (bottom panel) because of the fixation/permeabilization procedure. B, Frequency of TCRβ+CD4+ cells in peripheral lymph nodes of WT and PLZF transgenic mice. Plots are pregated on B220 cells. On average, the 1960 heterozygotes (n = 3) and 1797 heterozygotes (n = 4) had significantly decreased frequencies of CD4 T cells at 34 and 3%, respectively, compared with WT littermates (n = 5) (p < 0.0005), whereas the heterozygous (n = 5) and homozygous 1963 (n = 5) mice showed no significant changes. Data are representative of three to six mice of each genotype analyzed in three or more independent experiments. The WT, 1963 heterozygous, and 1963 homozygous mice depicted are littermates.

Close modal

We have used a tetramer-based MACS enrichment strategy to identify, enumerate, and characterize the rare Vα24CD1d-αGC+ T cells circulating in human peripheral blood. The identification of these cells in fresh blood was unambiguously validated by their characteristic overuse of Vβ11, similar to their Ag-stimulated progeny recently described by others (13, 14).

Our study revealed novel features of profound significance for our understanding of immune responses to lipid Ags. First, these cells predominantly expressed a naive CD45RA+CCR7+CD62Lhigh phenotype that predicted recirculation properties of lymph node-seeking T cells. Such properties are a prerequisite for the initiation of adaptive-like immune responses in lymph node draining sites of microbial invasion and are consistent with studies documenting the adaptive-like dynamics of CD1-restricted mycobacterial lipid-specific responses in mice and guinea pigs (17, 18). The functional properties of these naive-like cells could not be assessed in this study because the tetramer-MACS method used for enrichment impaired the survival and expansion of cells in culture. Our Vα24-negative T cell clones expanded from cultures stimulated with αGalCer produced IL-4 and IFN-γ upon stimulation, similar to previous reports, but a similar profile was observed in control clones derived from CD1d-αGalCer–negative CD4 T cells or from Vα24 NKT cells (not shown), consistent with the ability of human T cells to acquire a broad range of cytokines after in vitro culture. Thus, the cytokine properties of the naive-like cells remain to be determined.

Secondly, the naive-like cells expressed PLZF, the signature transcription factor that induces NKT cells to developmentally acquire their innate-like effector program (4, 5). This observation is remarkably consistent with the predominant expression of CD1 isotypes by thymocytes and with the emerging understanding that PLZF is induced upon selection of thymocytes by bone marrow-derived rather than epithelial ligands. Indeed, when MHC class II expression was experimentally redirected from epithelial cells to thymocytes, CD4 thymocytes acquired a CD62LlowCD44high effector phenotype with dual production of IL-4 and IFN-γ, and they expressed PLZF (19, 20). As most MHC-restricted T cells do not express PLZF, the expression of PLZF by the entire population of naive-like CD1d-αGalCer–specific T cells also provides a major argument against the possibility that these CD1d-restricted T cells were primarily selected by MHC-peptide ligands and secondarily cross-reacted onto CD1d–lipid complexes.

The expression of PLZF stands in apparent contradiction with the naive phenotype of these Vα24CD1d-αGC+ T cells because PLZF is necessary and sufficient for effector differentiation during T cell development. Our study establishes, however, that the expression level of PLZF in these lipid-specific cells was significantly reduced compared with NKT cells in fresh PBMC samples and that, as experimentally assessed in the mouse system, a sharp threshold of expression was required for effector differentiation, as judged by the CD44/CD62L phenotype. Thus, a 2-fold reduction of PLZF protein in PLZF+/− mice was sufficient to drastically impair NKT cell development and prevent their acquisition of effector properties. In addition, transgenic studies demonstrated that substantial amounts of PLZF protein, estimated between 34 and 64% of the maximal levels observed in NKT cells, were required for effector conversion. Collectively, these findings suggest that, although PLZF may mark T cell lineages selected by ligands expressed on thymocytes, different levels can be induced with substantial consequences on the naive versus effector type of differentiation imparted.

The mechanisms governing the induction of different levels of PLZF remain unclear at present. It is conceivable that the levels may depend on the avidity of TCR–ligand interactions during thymic development. Although they lack Vα24 but often express Jα18 and Vβ11, the other components of the NKT TCR, the Vα24CD1d-αGC+ thymocytes may have lower avidity than NKT cells for a common set of self-lipid Ags involved in their thymic selection. For example, they did not respond to iGb3 (14), a self-Ag recognized as a weak agonist by most human NKT cells (21).

The frequency of the naive recirculating CD1d-αGalCer–specific T cells was on the order of 1 × 10−5 and was stable over repeated samplings in the individuals studied. This surprisingly high frequency, which is similar or superior to estimates for MHC-peptide–specific T cells (22), may be explained by the existence of two germline gene segments, Jα18 and Vβ11, with intrinsic recognition properties for αGalCer and CD1d, respectively (23). As cells expressing PLZF represent a very small percentage of the naive human T cell repertoire, it is possible that the size of the naive CD1-restricted T cell population is very small compared with that of the MHC-restricted T cells. Alternatively, many CD1-restricted cells may express little or no PLZF, perhaps because they follow distinct developmental rules or because they are primarily selected by MHC ligands and cross-react against CD1. Recent reports have begun to assess the frequency and phenotype of fresh CD1-restricted T cells based on their ability to exhibit self-reactivity against CD1-transfectants in vitro (24, 25), and one report further demonstrated their presence among PBMC with a naive CD45RA+RO phenotype (25). Although these studies did not attempt to assess the expression of PLZF in fresh cells, studies of T cell lines and clones, however, may help evaluate the frequency of PLZF expression among CD1-restricted T cells, although it is unknown at present whether PLZF expression is a stable marker in long-term cultures. Mouse models transgenically expressing the human CD1 isotypes that are missing in mice may also prove particularly valuable to elucidate this important issue.

In conclusion, this study provides direct physical evidence for the existence of a sizable population of naive-like CD1d-restricted T cells that express a diverse TCR repertoire and recirculate in human peripheral blood. It suggests that at least a fraction of these cells may express PLZF at low or intermediate levels and therefore belong to a lineage distinct from most MHC-restricted T cells but related to NKT cells. The results also reveal the critical importance of a high threshold of PLZF expression for induction of the effector phenotype in T cells and raise the fundamental issue of the mechanisms regulating PLZF expression during T cell development.

We thank the members of the Bendelac laboratory for discussions, advice, and help at various stages of the study, the National Institute of Allergy and Infectious Diseases tetramer facility for CD1d tetramers, the University of Chicago Animal Resource Center, Core Flow Cytometry Facility, and DNA Sequencing Facility.

This work was supported by National Institutes of Health Grants AI038339 and AI053725. A.K.S. is the recipient of the Harper Fellowship of the University of Chicago. A.B. is a Howard Hughes Medical Institute Investigator.

The sequences presented in this article have been submitted to the Gene Expression Omnibus under accession number GSE28726.

Abbreviations used in this article:

αGalCer

α-galactosylceramide

h

human

MFI

mean fluorescence intensity

PLZF

promyelocytic leukemia zinc finger

WT

wild-type.

1
Bendelac
A.
,
Savage
P. B.
,
Teyton
L.
.
2007
.
The biology of NKT cells.
Annu. Rev. Immunol.
25
:
297
336
.
2
Moody
D. B.
2006
.
The surprising diversity of lipid antigens for CD1-restricted T cells.
Adv. Immunol.
89
:
87
139
.
3
De Libero
G.
,
Mori
L.
.
2010
.
How the immune system detects lipid antigens.
Prog. Lipid Res.
49
:
120
127
.
4
Savage
A. K.
,
Constantinides
M. G.
,
Han
J.
,
Picard
D.
,
Martin
E.
,
Li
B.
,
Lantz
O.
,
Bendelac
A.
.
2008
.
The transcription factor PLZF directs the effector program of the NKT cell lineage.
Immunity
29
:
391
403
.
5
Kovalovsky
D.
,
Uche
O. U.
,
Eladad
S.
,
Hobbs
R. M.
,
Yi
W.
,
Alonzo
E.
,
Chua
K.
,
Eidson
M.
,
Kim
H. J.
,
Im
J. S.
, et al
.
2008
.
The BTB-zinc finger transcriptional regulator PLZF controls the development of invariant natural killer T cell effector functions.
Nat. Immunol.
9
:
1055
1064
.
6
Barna
M.
,
Hawe
N.
,
Niswander
L.
,
Pandolfi
P. P.
.
2000
.
Plzf regulates limb and axial skeletal patterning.
Nat. Genet.
25
:
166
172
.
7
Benlagha
K.
,
Weiss
A.
,
Beavis
A.
,
Teyton
L.
,
Bendelac
A.
.
2000
.
In vivo identification of glycolipid antigen-specific T cells using fluorescent CD1d tetramers.
J. Exp. Med.
191
:
1895
1903
.
8
Liu
Y.
,
Goff
R. D.
,
Zhou
D.
,
Mattner
J.
,
Sullivan
B. A.
,
Khurana
A.
,
Cantu
C.
 III
,
Ravkov
E. V.
,
Ibegbu
C. C.
,
Altman
J. D.
, et al
.
2006
.
A modified alpha-galactosyl ceramide for staining and stimulating natural killer T cells.
J. Immunol. Methods
312
:
34
39
.
9
Benlagha
K.
,
Wei
D. G.
,
Veiga
J.
,
Teyton
L.
,
Bendelac
A.
.
2005
.
Characterization of the early stages of thymic NKT cell development.
J. Exp. Med.
202
:
485
492
.
10
Han
M.
,
Harrison
L.
,
Kehn
P.
,
Stevenson
K.
,
Currier
J.
,
Robinson
M. A.
.
1999
.
Invariant or highly conserved TCR alpha are expressed on double-negative (CD3+CD4-CD8-) and CD8+ T cells.
J. Immunol.
163
:
301
311
.
11
Costello
P. J.
,
Winchester
R. J.
,
Curran
S. A.
,
Peterson
K. S.
,
Kane
D. J.
,
Bresnihan
B.
,
FitzGerald
O. M.
.
2001
.
Psoriatic arthritis joint fluids are characterized by CD8 and CD4 T cell clonal expansions appear antigen driven.
J. Immunol.
166
:
2878
2886
.
12
Brochet
X.
,
Lefranc
M. P.
,
Giudicelli
V.
.
2008
.
IMGT/V-QUEST: the highly customized and integrated system for IG and TR standardized V-J and V-D-J sequence analysis.
Nucleic Acids Res.
36
(
Web Server issue
):
W503
W508
.
13
Gadola
S. D.
,
Dulphy
N.
,
Salio
M.
,
Cerundolo
V.
.
2002
.
Valpha24-JalphaQ-independent, CD1d-restricted recognition of alpha-galactosylceramide by human CD4(+) and CD8alphabeta(+) T lymphocytes.
J. Immunol.
168
:
5514
5520
.
14
Brigl
M.
,
van den Elzen
P.
,
Chen
X.
,
Meyers
J. H.
,
Wu
D.
,
Wong
C. H.
,
Reddington
F.
,
Illarianov
P. A.
,
Besra
G. S.
,
Brenner
M. B.
,
Gumperz
J. E.
.
2006
.
Conserved and heterogeneous lipid antigen specificities of CD1d-restricted NKT cell receptors.
J. Immunol.
176
:
3625
3634
.
15
Lee
P. T.
,
Putnam
A.
,
Benlagha
K.
,
Teyton
L.
,
Gottlieb
P. A.
,
Bendelac
A.
.
2002
.
Testing the NKT cell hypothesis of human IDDM pathogenesis.
J. Clin. Invest.
110
:
793
800
.
16
Raberger
J.
,
Schebesta
A.
,
Sakaguchi
S.
,
Boucheron
N.
,
Blomberg
K. E.
,
Berglöf
A.
,
Kolbe
T.
,
Smith
C. I.
,
Rülicke
T.
,
Ellmeier
W.
.
2008
.
The transcriptional regulator PLZF induces the development of CD44 high memory phenotype T cells.
Proc. Natl. Acad. Sci. USA
105
:
17919
17924
.
17
Felio
K.
,
Nguyen
H.
,
Dascher
C. C.
,
Choi
H. J.
,
Li
S.
,
Zimmer
M. I.
,
Colmone
A.
,
Moody
D. B.
,
Brenner
M. B.
,
Wang
C. R.
.
2009
.
CD1-restricted adaptive immune responses to Mycobacteria in human group 1 CD1 transgenic mice.
J. Exp. Med.
206
:
2497
2509
.
18
Hiromatsu
K.
,
Dascher
C. C.
,
LeClair
K. P.
,
Sugita
M.
,
Furlong
S. T.
,
Brenner
M. B.
,
Porcelli
S. A.
.
2002
.
Induction of CD1-restricted immune responses in guinea pigs by immunization with mycobacterial lipid antigens.
J. Immunol.
169
:
330
339
.
19
Lee
Y. J.
,
Jeon
Y. K.
,
Kang
B. H.
,
Chung
D. H.
,
Park
C. G.
,
Shin
H. Y.
,
Jung
K. C.
,
Park
S. H.
.
2010
.
Generation of PLZF+ CD4+ T cells via MHC class II-dependent thymocyte-thymocyte interaction is a physiological process in humans.
J. Exp. Med.
207
:
237
246
.
20
Li
W.
,
Sofi
M. H.
,
Rietdijk
S.
,
Wang
N.
,
Terhorst
C.
,
Chang
C. H.
.
2007
.
The SLAM-associated protein signaling pathway is required for development of CD4+ T cells selected by homotypic thymocyte interaction.
Immunity
27
:
763
774
.
21
Zhou
D.
,
Mattner
J.
,
Cantu
C.
 III
,
Schrantz
N.
,
Yin
N.
,
Gao
Y.
,
Sagiv
Y.
,
Hudspeth
K.
,
Wu
Y. P.
,
Yamashita
T.
, et al
.
2004
.
Lysosomal glycosphingolipid recognition by NKT cells.
Science
306
:
1786
1789
.
22
Jenkins
M. K.
,
Chu
H. H.
,
McLachlan
J. B.
,
Moon
J. J.
.
2010
.
On the composition of the preimmune repertoire of T cells specific for Peptide-major histocompatibility complex ligands.
Annu. Rev. Immunol.
28
:
275
294
.
23
Borg
N. A.
,
Wun
K. S.
,
Kjer-Nielsen
L.
,
Wilce
M. C.
,
Pellicci
D. G.
,
Koh
R.
,
Besra
G. S.
,
Bharadwaj
M.
,
Godfrey
D. I.
,
McCluskey
J.
,
Rossjohn
J.
.
2007
.
CD1d-lipid-antigen recognition by the semi-invariant NKT T-cell receptor.
Nature
448
:
44
49
.
24
de Jong
A.
,
Peña-Cruz
V.
,
Cheng
T. Y.
,
Clark
R. A.
,
Van Rhijn
I.
,
Moody
D. B.
.
2010
.
CD1a-autoreactive T cells are a normal component of the human αβ T cell repertoire.
Nat. Immunol.
11
:
1102
1109
.
25
de Lalla
C.
,
Lepore
M.
,
Piccolo
F. M.
,
Rinaldi
A.
,
Scelfo
A.
,
Garavaglia
C.
,
Mori
L.
,
De Libero
G.
,
Dellabona
P.
,
Casorati
G.
.
2011
.
High-frequency and adaptive-like dynamics of human CD1 self-reactive T cells.
Eur. J. Immunol.
41
:
602
610
.

The authors have no financial conflicts of interest.