CD1d-dependent NKT cells represent a heterogeneous family of effector T cells including CD4+CD8− and CD4−CD8− subsets that respond to glycolipid Ags with rapid and potent cytokine production. NKT cell development is regulated by a unique combination of factors, however very little is known about factors that control the development of NKT subsets. In this study, we analyze a novel mouse strain (helpless) with a mis-sense mutation in the BTB-POZ domain of ZBTB7B and demonstrate that this mutation has dramatic, intrinsic effects on development of NKT cell subsets. Although NKT cell numbers are similar in Zbtb7b mutant mice, these cells are hyperproliferative and most lack CD4 and instead express CD8. Moreover, the majority of ZBTB7B mutant NKT cells in the thymus are retinoic acid–related orphan receptor γt positive, and a high frequency produce IL-17 while very few produce IFN-γ or other cytokines, sharply contrasting the profile of normal NKT cells. Mice heterozygous for the helpless mutation also have reduced numbers of CD4+ NKT cells and increased production of IL-17 without an increase in CD8+ cells, suggesting that ZBTB7B acts at multiple stages of NKT cell development. These results reveal ZBTB7B as a critical factor genetically predetermining the balance of effector subsets within the NKT cell population.
The factors that regulate formation of distinct subsets of effector T cells are not well understood. While these responses are clearly influenced by the nature and route of exposure of an encountered Ag, genetic wiring also influences the kinds of effector T cell responses. Understanding these genetic factors is important to explain individual variability in physiological or pathological immune reactions to common Ags.
NKT cells are CD1d-restricted, glycolipid Ag-reactive T cells that represent a unique population of effector T cells in mice and humans. These cells express a heavily biased TCR repertoire, composed of an invariant TCR α-chain (Vα14Jα18 in mice, Vα24Jα18 in humans) paired with a limited array of TCR β-chains (Vβ8.2, Vβ7, or Vβ2 in mice, Vβ11 in humans) (1, 2). NKT cells can influence a broad spectrum of diseases, ranging from suppression of autoimmune diseases like type 1 diabetes, to promotion of immunity, to cancer and infection (3). This paradoxical ability to promote or suppress immune responses is associated with the profound ability of NKT cells to produce a spectrum of cytokines within hours of stimulation. At the population level, NKT cells produce seemingly antagonistic cytokines including IFN-γ, IL-4, IL-10, IL-13, and IL-17, although NKT cells can be divided into functionally distinct subsets that are capable of preferentially producing only some of these cytokines (4–7), which may partly explain the diverse functional outcomes associated with these cells.
Human NKT cells vary widely in frequency between individuals, yet are stable within individuals (8, 9). Human NKT cells include CD4+, CD4−CD8− (double negative [DN]), and CD8+ subsets, and each of these exhibit distinct cytokine profiles, which again suggests that they have distinct functions in vivo. The ratio of CD4/CD8 defined subsets of NKT cells also varies widely between individuals (10). Given this variability, combined with their powerful immunoregulatory potential, it is very important to decipher the factors that regulate NKT cell development and homeostasis, including factors that determine the balance of functionally distinct NKT cell subsets.
Many molecules, including cell surface receptors, signal transduction and transcription factors, have been identified that selectively regulate NKT cell numbers independently from conventional T cells (11). For example, the SLAM/SAP/fyn signaling pathway is selectively important for NKT cell development while dispensable for T cell development in the thymus (11). However, little is known about what regulates the differentiation of NKT cell subsets. One study, an investigation of NKT cell development in GATA-3 knockout mice, provided data showing that CD4+ NKT cells were preferentially inhibited in the absence of this factor (12). What factors regulate the appropriate expression of the CD4 or CD8 coreceptors in MHC class II– or MHC class I–restricted thymocytes has itself been a long-standing issue. In 2005, two studies showed that the transcription factor ZBTB7B (previously called Th-POK and cKrox) plays a key role in maintaining CD4 expression in MHC class II–restricted thymocytes (13, 14). ZBTB7B promotes CD4 expression indirectly by preventing Runx1- and Runx3-mediated downregulation of CD4 in conventional TCRαβ T cells (15). It has been shown subsequently that CD4 expression by NKT cells is also dependent on ZBTB7B (16, 17). Furthermore, in the absence of ZBTB7B, NKT cell cytokine production was impaired, which led to the suggestion that ZBTB7B is required for full NKT cell maturation and activation (16).
In this study, we describe a novel mouse strain, termed “helpless,” carrying a point mutation in Zbtb7b creating a single amino acid substitution in the BTB-POZ domain. Using this mouse model, we show that ZBTB7B plays an essential, cell intrinsic, and dose-dependent role in establishing the balance of different NKT cell subsets, including maintenance of CD4, inhibition of CD8 expression, and development of the retinoic acid–related orphan receptor γt (RORγt)-positive IL-17+ population of NKT cells. ZBTB7B thereby establishes a genetically predetermined profile of CD1d-restricted NKT cells.
Materials and Methods
The Zbtb7bhpls/hpls strain derived from a C57BL/6 male treated three times i.p. with 100 mg/kg N-ethyl-N-nitrosourea at weekly intervals. Mice were maintained on a pure C57BL/6 background or on a mixed CBA × C57BL/6 background. All mice were housed in specific pathogen–free conditions at the Australian Phenomics Facility. All experimental procedures were approved by the Australian National University Animal Ethics and Experimentation Committee.
Sequencing and genotyping
All exons and splice sites of Zbtb7b were amplified, and primers for sequencing were designed using Australian Phenomics Facility software to amplify all exons and splice sites for Zbtb7b. The amplification and dual sequence run were performed at the Brisbane node of the Australian Genome Research Facility. Sequence analysis was conducted at the Australian Phenomics Facility using Lasergene software (DNAStar). A T to G substitution of bp 480 was identified in exon 2 (ensmuse000001765423) of Zbtb7b, resulting in a CTG (Leu) to a CGG (Arg) amino acid change. Mice were genotyped using an Amplifluor assay (Chemicon). All primer sequences are available on request.
Cell suspensions from thymus and spleen were prepared by passing the cells through a cell strainer (BD Biosciences) or stainless steel sieve, followed by lysis of RBCs for spleen and liver samples. Liver lymphocytes were isolated by centrifugation over a Percoll gradient. Cell suspensions were labeled with fluorochrome-coupled Abs according to standard protocols and run on an LSR II or Canto flow cytometer (BD Biosciences) followed by analysis with FlowJo (Tree Star). α-GalCer–loaded CD1d tetramers were produced in house, using a mouse CD1d baculovirus construct originally provided by Prof. Mitchell Kronenberg, as previously described (18). For some experiments, α-GalCer (PBS57)-loaded CD1d tetramers provided by the National Institutes of Health Tetramer Facility were used.
In vitro stimulation and cytokine analysis
For the intracellular cytokine staining assay, cells were stimulated with 50 ng/ml phorbol ester and 500 ng/ml ionomycin for 2.5–3.5 h at 37°C in the presence of monensin in RPMI 1640 culture media supplemented with 10% heat-inactivated FCS, glutamine (Life Technologies), 10 mM sodium pyruvate (Life Technologies), 10 mM HEPES (Life Technologies), 10 mM MEM nonessential amino acids (Life Technologies), and 5.5 μM 2-mercaptoethanol. Stimulated cells were then washed, stained for surface markers, and stained intracellularly using FITC-conjugated anti-mouse IL-17A (BioLegend) or isotype-matched control Abs (BD Pharmingen), RORγt–PE, or allophycocyanin (eBioscience) using the eBioscience fixation/permeabilization kit.
For cytometric bead array, thymocytes were pooled from several mice per group and enriched by either staining with anti-CD24 (J11D) followed by depletion using rabbit complement (C-SIX Diagnostics) in the presence of DNase (Roche Diagnostics) or by staining cells with PE-conjugated CD1d–α-GalCer tetramer and subsequent incubation with anti-PE microbeads (Miltenyi Biotech). Labeled cells were then enriched by passing them through a magnetic column and were further stained for flow cytometric purification. This second method was also used to enrich splenic dendritic cells (based on CD11c expression). Enriched cells were sorted using a FACSAria (BD Biosciences) in the Department of Microbiology and Immunology Flow Cytometry Facility (University of Melbourne) to obtain highly purified populations. Sorted NKT cells were stimulated by placing them in 96-well plates (2 × 104 cells/well), coated with anti-CD3 and anti-CD28, or with soluble CD1d loaded with α-GalCer, or by coculture with sorted splenic dendritic cells (1 × 104 dendritic cells/well) loaded with α-GalCer. After 24 h, the supernatant was collected and the concentration of cytokines secreted into the medium determined by cytometric bead array (BD Pharmingen).
Cell sorting, RNA extraction, cDNA synthesis, and quantitative PCR
Single-cell suspensions from thymocytes were prepared and stained as for flow cytometric analysis. Samples were sorted on a FACSAria sorter (BD Biosciences) at the Flow Cytometry Facility of the John Curtin School of Medical Research (Australian National University). Total RNA was extracted using RNA TRIzol reagent (Molecular Research Centre) and reverse transcribed using random oligonucleotide primers and 50 U Superscript II reverse transcriptase (Invitrogen) as detailed in the manufacturer’s guidelines. SYBR Green real-time PCR reactions were performed in 96-well plates (PerkinElmer) with an ABI PRISM 7900 Real-Time System (PerkinElmer/PE Biosystems) at the Biomolecular Resource Facility (John Curtin School of Medical Research, Australian National University). To correlate the threshold (Ct) values from the cDNA amplification plots to fold differences between samples, the ΔΔCt method was applied using the housekeeping gene GAPDH.
Generation of bone marrow chimeras
B6.SJL CD45.1 mice were irradiated with 10 Gy and injected with 2 × 106 bone marrow cells consisting of a 50:50 mix of wild-type (WT) B6.SJL CD45.1+ and either WT or mutant C57BL/6 (CD45.2+) cells. They were allowed to reconstitute for 8 wk before analysis.
Identification of the Zbtb7b mutant strain helpless (hpls)
In a genome-wide screen for N-ethyl-N-nitrosurea–induced point mutations affecting the development of the immune system (19), we identified a strain that was deficient in CD4+ T cells in peripheral blood and spleen. Further analysis revealed a block in CD4 development at the CD4+CD8dim stage in the thymus (Fig. 1A). This phenotype is identical to the phenotype described for the HelperDeficient (HD) strain caused by an amino acid substitution in the DNA-binding zinc finger domain of the transcription factor ZBTB7B, previously called Th-POK or cKrox (13, 20) or knockout mice with a Zbtb7b null allele (21). Because of these similarities, we sequenced Zbtb7b and identified a mutation changing a conserved leucine to arginine in the BTB-POZ domain (Fig. 1B). BTB-POZ domains mediate homodimerization and heterodimerization, association with nuclear corepressors, and ubiquitination (22). Deletion of the BTB-POZ domain from a Zbtb7b transgene inactivated its ability to deviate MHC class I–restricted T cells into the CD4 lineage (14). The leucine that is mutated in helpless mice is buried within the homologous PZLF BTB-POZ domain (Supplemental Fig. 1 and Ref. 23). The ZBTB7BL102R disrupts CD4 cell differentiation as completely as the null mutation, but whether this reflects misfolding of the BTB domain or destabilization of the ZBTB7B protein as a whole or loss of particular protein–protein interactions is unclear. Genotyping of affected and unaffected progeny from numerous helpless carriers confirmed that the failure of CD4 cell differentiation was inherited in complete concordance with the Zbtb7b mutation in a recessive fashion. As noted previously for the HD strain (24), homozygous affected mice on the parental C57BL/6 background were born at around half the expected frequency, and affected mice showed poor breeding efficiency. This was rescued by keeping the mice on a mixed C57BL/6 × CBA background, where homozygotes were obtained at Mendelian ratios. The embryonic lethality on the C57BL/6 background may indicate essential functions for ZBTB7B in other processes such as collagen gene regulation (25).
It was previously reported that CD4 expression by NKT cells is disrupted in Th-POK–deficient mice (16, 17), so we first investigated whether the Zbtb7b point mutation in hpls/hpls mice had a similar impact on these cells. Normally, around 70% of spleen NKT cells express CD4, and none are CD8+. In the helpless strain, this was completely reversed with ∼70–80% of NKT cells in the spleen expressing CD8, and none were CD4+ (Fig. 1C).
Examination of NKT cells in different tissues revealed that the percentage in thymus, spleen, and liver was comparable between WT and hpls/hpls mice, whereas the percentages and absolute cell numbers of these cells in blood, bone marrow, and lymph nodes (LNs) were clearly higher (Fig. 1D, 1E). Notably, Zbtb7b heterozygous mice also showed an increased percentage of NKT cells in thymus, LN, and bone marrow, but not in spleen and liver (Fig. 1D). The CD8+ NKT cell phenotype was detectable in thymus of hpls/hpls mice (∼20% CD8+) but far more pronounced in the spleen and liver where ∼80% were CD8+ (Fig. 1F). Further analysis of the CD8+ NKT cells in thymus, spleen, and liver showed that they included some cells that were CD8α+β− and some that were CD8α+β+ (Fig. 1G). Furthermore, heterozygous hpls/+ mice exhibited an intermediate phenotype, where CD4+ NKT cells were reduced, yet there was no sign of CD8 upregulation, in both thymus and periphery, thus hpls/+ mice were highly enriched for CD4−CD8− NKT cells (Fig. 1F). This contrasts with the development of conventional CD4+ T cells that is seemingly unaffected in heterozygous mice (Fig. 1A) strongly suggesting that the effect on NKT surface marker expression is not due to a dominant-negative effect of the hpls mutation.
NKT cell development is abnormal in Zbtb7b mutant mice
The developmental events giving rise to the unusual CD8+ NKT cell phenotype in Zbtb7b mutant mice have not been determined; however, the accumulation of these cells in peripheral tissues at much higher levels than in thymus suggested this occurs as a late event in NKT cell development. Therefore, we more carefully examined NKT cells in the thymus to determine the origins of this defect. Analysis of Zbtb7b mRNA expression by real-time PCR revealed that both immature NK1.1− and mature NK1.1+ NKT cells expressed Zbtb7b (data not shown). Zbtb7b expression levels were similar to CD4+CD8dim thymocytes, lower than total CD4+ single-positive thymocytes, but clearly above double-positive and CD8 single-positive thymocytes that have expression at or just above background (13). Because NKT cell numbers in the thymus of Zbtb7b mutant mice were comparable to those in WT mice, this suggested there was no major problem with the selection and expansion of these cells as a total population. NKT cell development can be divided into three developmentally distinct stages: stage 1 (CD44loNK1.1−); stage 2 (CD44+NK1.1−); stage 3 (CD44+NK1.1+) (26, 27); and these stages can be further divided into CD4+ and CD4−, a split that occurs at approximately stage 2 (4). Although hpls/hpls NKT cells were mostly mature, as defined by their NK1.1+CD44hi phenotype, they expressed slightly lower levels of NK1.1 (Fig. 2A). Although CD8+ NKT cells were detected in the hpls/hpls thymus, the high-intensity CD8 expression observed with hpls/hpls peripheral NKT cells (Fig. 1C) was not reflected in the thymus where they were mostly CD8lo. This suggested that the emergence of CD8+ NKT cells begins in the thymus but is not fully manifested until these cells are in the periphery. Examination of increasingly mature NKT cell subsets revealed that low CD8 expression was detectable from the earliest CD44loNK1.1− stage, but the percentage of CD8+ cells increased as the NKT cells matured through CD44+NK1.1− and CD44+NK1.1+ stages (Fig. 2A, 2B).
To determine whether the defects in NKT cell development in hpls/hpls mice were cell intrinsic, we performed mixed bone marrow chimera experiments to compare WT and hpls/hpls NKT cells developing in the same environment. These results demonstrated that the CD4−CD8+ phenotype was indeed cell intrinsic (Fig. 2C) and also indicated that hpls/hpls NKT cells had a competitive developmental advantage as can be seen by a higher ratio of these cells as a total population (Fig. 2D). Analysis of the maturational stages revealed that this bias was not detected at stage 1 of NKT development (CD44−NK1.1−) but was first apparent at stage 2 (CD44+NK1.1−) and maintained at stage 3 (CD44+NK1.1+) of NKT cell development (Fig. 2D). This bias toward hpls/hpls NKT cells with maturation appeared to be at least partly due to hyperproliferation of stage 3 cells, as indicated by high-frequency staining with Ki67 compared with the resting state of the corresponding WT cells in the same thymus (Fig. 2E). The highest level of Ki67 staining was associated with the CD8+ NKT cell fraction (Fig. 2F). Thus, these data demonstrate that ZBTB7B plays an intrinsic role in the regulation of NKT cell development that seems to be first manifest after positive selection in the thymus as these cells begin to mature (stage 2). Analysis of NKT cells in the peripheral tissues of mixed bone marrow chimeras showed that the hpls/hpls NKT cells had a cell-intrinsic competitive advantage in all analyzed organs including thymus, spleen, liver, and LN (Fig. 2G).
Other cell surface markers including CD62-L and NK cell receptors Ly6C, NKG2A/C/E, Ly49C/I, and NKG2D were also differently expressed by the NKT cells in hpls/hpls mice (Fig. 3). The lower levels of NK1.1 observed on thymic hpls/hpls NKT cells was not observed on hpls/hpls NKT cells from spleen or liver. Similarly, hpls/hpls NKT cells also had lower levels of Ly6C and the NKG2 receptors in the thymus but higher levels in spleen and liver compared with WT NKT cells. Mutant NKT cells also tended toward higher expression of CD62-L, especially in spleen, possibly explaining the increased percentage of NKT cells observed in LN of hpls/hpls mutant mice.
Zbtb7b mutant NKT cells preferentially develop into an IL-17–producing subset
A recent study (16) suggested that NKT cells from ZBTB7B-deficient mice were functionally impaired with much lower cytokine production compared with their WT counterparts, which was surprising given their otherwise mature phenotype. Our findings were consistent with this for the cytokines previously tested (IFN-γ, IL-4). Analysis of cytokine production by these cells, using intracellular cytokine staining following in vitro stimulation, revealed that very few mutant NKT cells produced IFN-γ or TNF, whereas a high proportion of NKT cells produced IL-17, which was the opposite of WT NKT cells (Fig. 4A). Given that hpls/hpls NKT cells clearly were functional and capable of cytokine production, we used cytometric bead array to test cytokine production by these cells more comprehensively. Purified NKT cells were stimulated in three different ways: plate-bound CD3 and CD28, plate-bound CD1d loaded with α-GalCer, or spleen-derived dendritic cells loaded with α-GalCer, and supernatants were harvested after 24 h. This revealed that cytokine production by hpls/hpls NKT cells, with the exception of IL-17, was drastically reduced, to the extent that many cytokines were near or below the detection limit (Fig. 4B). In some experiments, we also compared DN and CD8+ NKT cells from hpls/hpls mice and observed similar cytokine production regardless of the expression of CD8 (data not shown) suggesting that ZBTB7B acts at multiple levels on NKT cell development. This suggests that ZBTB7B is important in regulating the developmental balance of IL-17–producing NKT cells, which have recently been identified as a distinct subset of NKT cells (known as NKT-17 cells) that make lower amounts of other cytokines and have unique functions in vivo (4, 7).
IL-17–producing hpls/hpls NKT cells are RORγt+
The production of IL-17 is usually dependent on the transcription factor RORγt, (28) and previous studies showed that IL-17–producing NKT cells express RORγt and the receptor for IL-23 (4, 7, 29). To test whether hpls/hpls NKT cells overexpress either of these molecules, we performed real-time PCR for IL-23R and RORγt and found both to be increased in hpls/hpls NKT cells (Fig. 5A). The hpls/+ heterozygous mice showed a small increase in expression compared with the WT NKT cells. We also tested for the transcription factors RUNX1 and RUNX3, which have been shown to play an important role in the regulation of CD4 and CD8 coreceptor expression in conventional T cells (30), and the transcription factor T-bet, which is essential for progression from the NK1.1− to the NK1.1+ stage of NKT development (31). No clear difference was observed for Runx1 and Runx3 expression, but expression of T-bet was ∼3-fold reduced in the hpls/hpls mutant NKT cells. RT-PCR is unable to distinguish between an increased frequency of positive cells and increased expression per cell. To test for this directly and simultaneously determine if the increased expression of IL-17 in Zbtb7b mutant NKT cells coincides with RORγt expression, NKT cells from +/+, hpls/+, and hpls/hpls mice were stimulated with PMA and ionomycin and intracellular staining used to assess RORγt and IL-17 expression by flow cytometry. This revealed that ∼80% of the mutant NKT cells in the thymus were positive for RORγt and about two-thirds of those produced IL-17 (Fig. 5B–D). For spleen-derived NKT cells, the percentage of RORγt+ NKT cells was 20% in the hpls/hpls animals, and again, approximately two-thirds of these produced IL-17. In contrast, WT NKT cells were only 3–8% RORγt+, whereas heterozygous mice showed an intermediate phenotype with around 20% of NKT cells in thymus expressing RORγt (Fig. 5C, 5D). Analysis of NKT cells in the thymus and liver of mixed bone marrow chimeras showed that the increased expression of RORγt and production of IL-17 in hpls/hpls NKT cells was cell intrinsic (Supplemental Fig. 3). Notably, the RORγt+ NKT cells had a lower expression of NK1.1 than RORγt− NKT cells (Fig. 5B). Because this is similar to IL-17–producing NKT cells (NKT-17 cells) in WT mice (4, 7), this most likely does not reflect a specific effect of the mutant ZBTB7B protein on NK1.1 expression, but rather the predominance of an NK1.1low IL-17–producing NKT subset with a specific phenotype. This interpretation is also supported by the mutually exclusive expression of the transcription factors T-bet and RORγt as observed by dual labeling of NKT cells (Fig. 5E) and the expression of IL-23R by hpls/hpls NKT cells (Fig. 5A), which is also a marker of NKT-17 cells in WT mice (7, 29). To explore this possibility further, we examined hpls/hpls NKT cells for CD103 and CCR6 expression because high expression of these markers is associated with NKT-17 cells in LN (32). Indeed, we found that the RORγt+ LN NKT subset from both WT and hpls/hpls mice was CD103hi and CCR6hi, whereas the RORγt− NKT subsets were heterogeneous for these markers (Supplemental Fig. 2). This is consistent with the previous study where IL-17 was only produced by a subset of CD103hi, CCR6hi, and RORγthi LN NKT cells (32).
Taken together, the findings in this study strongly suggest that NKT cell lineage decisions, and specifically the ratio of NKT-17 cells to other NKT cells, is intrinsically regulated by the transcription factor ZBTB7B.
Although several factors that control NKT cell numbers have been identified (33), there is little understanding of what controls the development of distinct subsets of NKT cells. In this study, we demonstrate that although ZBTB7B is not required for the development of normal numbers of NKT cells, it plays a critical role in genetically predetermining their differentiation into different subsets defined by patterns of cell surface markers and cytokine production.
It has been appreciated, almost since their discovery, that NKT cells in mice and humans can be divided into subsets based on CD4 and CD8 expression. Furthermore, there are clear differences in cytokine production by CD4+ and CD4− subsets of human NKT cells, where CD4+ cells produced both Th1 and Th2 type cytokines while CD4− NKT cells produced predominantly Th1 type cytokines (5, 6). Moreover, human CD8+ NKT cells may also be functionally distinct from CD4+ and DN NKT cells (34). Thus, the cell surface phenotype of NKT cells appears to correlate with important functional diversity. While it is also clear that mouse NKT cells include diverse subsets defined by cell surface markers including CD4, there are some important distinctions with human NKT cells. There is no clear difference in cytokine production between mature mouse CD4+ and CD4− NKT cells; both can make Th1 and Th2 type cytokines (4). Also, mouse NKT cells do not normally include a population of CD8+ cells. However, mouse NKT cells can be subdivided into functionally distinct subsets based on the expression of NK1.1, and NK1.1− mouse NKT cells produce a distinct array of cytokines compared with NK1.1+ NKT cells, including higher IL-4 and lower IFN-γ (4). Of particular interest, a subset of CD4−NK1.1− mouse NKT cells predominantly produce the proinflammatory cytokine IL-17 but not other cytokines (4, 7). This subset is thought to represent a distinct lineage of NKT cells, whose development depends on the transcription factor RORγt (4, 7, 35). The production of IL-17, a proinflammatory cytokine, imbues this subset of NKT cells with markedly different functional potential. IL-17–producing NKT cells have been associated with induction of airway neutrophilia (7), ozone-induced airway hyperreactivity (36), and collagen-induced arthritis (a model for rheumatoid arthritis) in mice (37). It is also likely, given the unique functions of IL-17 in autoimmunity and anti-microbial immunity (38), that IL-17–producing NKT cells will have different functions from non–IL-17–producing NKT cells in other disease settings.
In contrast to normal WT mice, Zbtb7b mutant mice lack CD4+ NKT cells and instead harbor a population of CD8+ NKT cells (16, 17). Furthermore, antigenic stimulation of these cells revealed a major defect in IL-4 and IFN-γ production, which led to the suggestion that these cells were functionally hyporesponsive to TCR stimulation (16). However, IL-17 was not tested in that study, and our findings present a very different interpretation for the role of ZBTB7B in NKT cell development, showing that although these cells are deficient in IFN-γ and IL-4 as published (16), they are clearly capable of producing high levels of IL-17. Moreover, IL-17–producing Zbtb7b mutant NKT have lower levels of NK1.1, and they express high levels of RORγt and IL-23R, which also closely aligns them with IL-17+ NKT cells in WT mice. Furthermore, both CD8− and CD8+ NKT cells in Zbtb7b mutant mice produced IL-17, indicating that the CD8 phenotype is not directly related to the altered cytokine profile and may reflect multiple developmental checkpoints that are controlled by ZBTB7B. Our data suggest that the NKT-17 lineage may be a default pathway for NKT cell development and that ZBTB7B is a key transcription factor that drives the development of other phenotypically and functionally distinct NKT cell subsets. In support of this, T-bet, a transcription factor that is known to be critical for maturation of IFN-γ–producing NK1.1hi NKT cells (31), was present at much reduced levels in Zbtb7b mutant NKT cells. At present, it is unclear if ZBTB7B regulates the balance of development of these NKT cell subsets by directly binding to the Rorc gene to influence expression of RORγt or if ZBTB7B somehow affects signaling through STAT3, another transcription factor required for the development of IL-17–producing T cells (39). Further studies including chromatin immunoprecipitation assays are required to differentiate between these possibilities, but given the dramatic functional difference between NKT-17 cells and other NKT cells, identification of ZBTB7B as a major switch factor represents an important step toward being able to control NKT cell function.
The developmental and functional basis for CD4 and CD8 coreceptor expression by NKT cells has been a long-standing puzzle in the field. Because NKT cell TCRs are CD1d restricted, there is no obvious role for CD4 or CD8 in binding to MHC class II or MHC class I, respectively, nor is it clear how or why these molecules are modulated during NKT cell development. Our data using “helpless” Zbtb7b mutant mice sheds new light on this problem, demonstrating that ZBTB7B is critical for development and/or maintenance of CD4+ NKT cells, but it also inhibits the emergence of CD8+ NKT cells. This seems to be regulated at multiple levels in a dose-dependent manner, because hpls/+ mice have diminished CD4+ NKT cells but do not have increased CD8+ NKT cells. The altered ratio of CD4+ and CD8+ NKT cells may be at least partly related to proliferative differences in the thymus, where mutant NKT cells that are DN and CD8+ proliferate at a higher rate than WT CD4+ or DN NKT cells, and our analysis of mixed bone marrow chimeras shows that this effect is cell intrinsic. However, proliferation is unlikely to explain fully the differences because CD8+ NKT cells are simply nonexistent in the periphery of normal WT mice. Importantly, our study demonstrates that CD8 is gradually acquired by Zbtb7b mutant NKT cells, rather than by a failure to downregulate this surface marker after NKT cell selection from DP precursors in the thymus. This is further supported by the fact that many of these cells are CD8α+CD8β− in contrast to DP thymocytes that express both CD8α and CD8β. An early study demonstrated that forced (transgenic) expression of CD8 by all T cells resulted in depletion of NKT cells (40), which suggested that mouse NKT cells that express CD8 are deleted in the thymus, perhaps due to enhanced signaling via CD8 resulting in negative selection. However, a more recent publication showed that thymocytes from homozygous CD8 transgenic mice have a shorter life span, which affects their ability to undergo distal TCR Vα-Jα gene rearrangements (16). This makes secondary TCR rearrangements, required to incorporate distal TCR-Jα genes such as Jα18 into the TCR-α-chain (41–44), less likely. This study also demonstrated that CD8 does not detectably bind to CD1d (16). Thus, the conclusion most consistent with our data are that the expression of ZBTB7B is normally activated at a very early stage in NKT cell development, and in the absence of this transcription factor, CD8 expression is reacquired during NKT cell maturation. This is an intriguing consideration in light of the existence of human CD8+ NKT cells. In the future, it will be important to determine whether human CD8+ NKT cells have lower amounts of ZBTB7B and express RORγt.
The regulation of CD4 expression by conventional αβ T cells involves binding of RUNX1 and RUNX3 to the Cd4 silencer at the different stages during their development to suppress the expression of CD4 at the DN stage of thymic development or in CD8+ cytotoxic T cells (45). Furthermore, the expression of Zbtb7b in CD8+ cytotoxic T cells is silenced by Runx proteins (46), and RUNX1 has also been shown to enhance the expression of CD8 by cytotoxic T cells (41). But consistent with previous reports showing an essential role for RUNX1 in the development of NKT cells (41), we observed a comparable expression for both Runx1 and Runx3 mRNA in Zbtb7b mutant and WT NKT cells.
There are some similarities between our findings and those previously reported in GATA-3–deficient mice (12), primarily a lack of CD4+ NKT cells and reduced IFN-γ production by NKT cells after TCR ligation. GATA-3 is known to bind the Zbtb7b locus and is thought to promote Zbtb7b expression in NKT cells (17). However, GATA-3–deficient mice lack stage 2 (CD44hiNK1.1low) NKT cells in the thymus (12) and exhibit a deficiency in peripheral NKT cells, which clearly distinguishes this defect from that observed in Zbtb7b mutant NKT cells. Furthermore, NK1.1 was not lower in GATA-3–deficient NKT cells, they were able to produce IFN-γ, and they do not express CD8 (17). Thus, there appears to be at least two factors (GATA-3 and ZBTB7B) that are important for controlling CD4 expression by NKT cells, and they work in a nonredundant manner.
The precise mechanisms by which ZBTB7B regulates the development of NKT cell subsets and how the Leu102Arg mutation affects the function of the protein remain to be determined. It is currently unclear if the L102R mutation is a null allele or has a dominant-negative effect, but as the number of RORγt+, IL-17–producing cells is intermediate in the heterozygous mice (Figs. 4A, 5B), this appears to support the concept that the mutation is having a null effect. This also fits well with the fact that the Leu102Arg mutation is in the dimerization domain (Supplemental Fig. 1) and very likely interferes with the dimerization of ZBTB7B required for transcriptional activation, and this likely results in a null allele in the homozygous state. Furthermore, based on our observations from hpls/+ and hpls/hpls mice, the data suggest that full expression of this protein is required to maintain normal numbers of CD4+ NKT cells because these are diminished (but not absent) in the heterozygous mice. In contrast, heterozygous mice showed no apparent defect in the percentage or number of conventional CD4+ T cells in the periphery. It was only in the homozygous mutant mice that we found an absence of CD4+ NKT cells and an abundance of CD8+ NKT cells, suggesting that partial expression of ZBTB7B is still sufficient to prevent CD8+ NKT cells. Whether this simply represents a dose effect acting at the same point in development or independent effects at different stages in development is less clear. However, the fact that CD8+ NKT cells were far more abundant in spleen and liver compared with thymus supports the second scenario, where an absence of functional ZBTB7B allows a delayed increase in CD8+ NKT cells, which does not occur in the presence of suboptimal ZBTB7B expression. Given the large variability in the population size of preexisting NKT cells especially in humans, variability in this process may contribute to individual variability in the types of immune responses made to common Ags. Future functional studies comparing WT and hpls/hpls NKT cells, using adoptive transfer into NKT cell–deficient mice (containing normal numbers of conventional CD4 T cells), are required to study the influence of the hpls mutation on NKT cells in disease models such as infection, cancer, and autoimmunity.
In summary, we have determined that ZBTB7B is a critical factor controlling the development of NKT cell subsets defined by cell surface CD4 and CD8 expression, where CD4 expression is tightly regulated by ZBTB7B such that even partial reduction of this factor diminishes CD4+ NKT cells, while the emergence of CD8+ NKT cells only occurs in the absence of ZBTB7B. ZBTB7B also determines the functional potential of NKT cells, such that in its absence, an RORγt+ IL-23R+ IL-17+ (NKT-17) phenotype predominates. This study demonstrates that the ZBTB7B transcription factor has a profound impact on the development and functional potential of NKT cells. Further studies into the ZBTB7B-mediated molecular pathways that culminate in the altered phenotypes are required.
We thank Owen Siggs and Lina Tze for suggesting Zbtb7b as a candidate gene and Belinda Whittle and the genotyping team of the Australian Phenomics Facility for genotyping, Ken Field (University of Melbourne) and Natalie Saunders (St. Vincent’s Institute) for assistance with flow cytometric cell sorting, and we acknowledge the support of the National Institutes of Health Tetramer Facility for providing the PBS57-loaded CD1d tetramer for some of the experiments.
This work was supported by research grants from the Wellcome Trust, the National Institutes of Health (AI054523), the National Health and Medical Research Council of Australia (NHMRC), the CASS Foundation, and the Ramaciotti Foundations. A.E. was supported by a Deutsche Forschungsgemeinschaft Research Fellowship (EN 790/1-1) and an NHMRC Project Grant (APP1009190) and Career Development Fellowship (APP1035858). D.I.G. was supported by an NHMRC Senior Principal Research Fellowship (1020770). C.C.G. was supported by an NHMRC Australia Fellowship (585490) and by an Australian Research Council Federation Fellowship.
The funders of the study had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The online version of this article contains supplemental material.
Abbreviations used in this article:
retinoic acid–related orphan receptor γt
The authors have no financial conflicts of interest.