Shared and Distinct Phenotypes and Functions of Human CD161++ Vα7.2+ T Cell Subsets

Human mucosal-associated invariant T (MAIT) cells are an important T cell subset that are enriched in tissues and possess potent effector functions. Typically such cells are marked by their expression of Vα7.2-Jα33/Jα20/Jα12 T cell receptors, and functionally they are major histocompatibility complex class I-related protein 1 (MR1)-restricted, responding to bacterially derived riboflavin synthesis intermediates. MAIT cells are contained within the CD161++ Vα7.2+ T cell population, the majority of which express the CD8 receptor (CD8+), while a smaller fraction expresses neither CD8 or CD4 coreceptor (double negative; DN) and a further minority are CD4+. Whether these cells have distinct homing patterns, phenotype and functions have not been examined in detail. We used a combination of phenotypic staining and functional assays to address the similarities and differences between these CD161++ Vα7.2+ T cell subsets. We find that most features are shared between CD8+ and DN CD161++ Vα7.2+ T cells, with a small but detectable role evident for CD8 binding in tuning functional responsiveness. By contrast, the CD4+ CD161++ Vα7.2+ T cell population, although showing MR1-dependent responsiveness to bacterial stimuli, display reduced T helper 1 effector functions, including cytolytic machinery, while retaining the capacity to secrete interleukin-4 (IL-4) and IL-13. This was consistent with underlying changes in transcription factor (TF) expression. Although we found that only a proportion of CD4+ CD161++ Vα7.2+ T cells stained for the MR1-tetramer, explaining some of the heterogeneity of CD4+ CD161++ Vα7.2+ T cells, these differences in TF expression were shared with CD4+ CD161++ MR1-tetramer+ cells. These data reveal the functional diversity of human CD161++ Vα7.2+ T cells and indicate potentially distinct roles for the different subsets in vivo.

The identification of MAIT cells using MR1-tetramers has clearly demonstrated that MAIT cells can be divided into CD8+, CD4+, and CD8− CD4− (double negative; DN) subsets (4,5,8). The frequency of each subset and their distribution varies between mammalian species and is also influenced by age and tissue location. In ruminants and Vα19-Jα33 TCR-transgenic mice, the majority of the MAIT cells are DN (9,10), whereas within humans they are predominantly CD8+ (4,11,12). Furthermore, while more than half of these CD8+ MAIT cells express the CD8αα homodimer in adults, MAIT cells in second trimester fetal thymi express the CD8αβ heterodimer, with CD8αα expression associated with the acquisition of a memory phenotype (12,13). The frequency of these CD8+ MAIT cells in the periphery falls with age (14)(15)(16). Furthermore, a recent study using MR1-tetramers in wild-type mice has demonstrated that, similar to humans, certain laboratory strains of mice have a larger fraction of MAIT cells that are CD8+ (8). Furthermore, the distribution of murine MAIT cell subsets differs between tissues, with an enrichment of CD4+ MAIT cells in lymph nodes (8). Altogether, these reports highlight the heterogeneity of the frequency and distribution of MAIT cell subsets. However, the functional relationship between the different subsets, especially in humans, is poorly understood.
Mucosal-associated invariant T cells share many developmental and functional features with invariant natural killer (NK) T (iNKT) cells. Similar to MAIT cells, CD1d-restricted iNKT cells consist of CD4+ and CD4− subsets (17)(18)(19), where CD4− iNKT cells secrete T helper 1 (Th1) cytokines, as well as interleukin-17 (IL-17), while CD4+ iNKT cells are the dominant producers of Th2 and immunoregulatory cytokines, such as interleukin-4 (IL-4), IL-10, and IL-13 (18,20,21). The balance between these subsets is thought to be key in determining the protective or pathological role of iNKT cells in disease (22). For example, while CD4+ iNKT cells protected non-obese diabetic mice from developing type 1 diabetes, CD4− iNKT cells secreting IL-17 exacerbated the disease (22). Furthermore, the cytolytic and Th1 cytokine-producing CD4− iNKT cells are the main effector population in tumor rejection (23) and control of microbial infections (24,25). Most of the previous MAIT cell studies have focused on the dominant population of MAIT cells in mice (DN) and humans (CD8+). Therefore, whether the different subsets have the potential to play a different function in disease remains unknown. Furthermore, whether differential expression of transcription factors (TFs) may account for the differences in MAIT cell subsets has not been fully investigated.
In humans, MAIT cells can be identified as cells expressing the C-type lectin-like receptor CD161 at a high level together with the Vα7.2 TCRα chain (4,11,26). These cells have previously been shown to overlap with cells stained with the MR1tetramer, particularly in the CD8+ and DN subsets (4,27). High expression of CD161 is a feature of innate-like T cells that have the ability to respond to innate cytokines and share a transcriptional signature, regardless of the specificity of their TCR (28), and CD161++ Vα7.2+ T cells-the majority of which are MAIT cells-are contained within this family of T cells. In this study, we have examined the frequency of CD161++ Vα7.2+ T cell subsets in human peripheral blood, liver, and bone marrow (BM), and performed a detailed analysis of the overlapping and distinct phenotypic and functional features of each.

Blood and Tissue samples
Whole blood was obtained from leukocyte cones (NHS Blood and Transplant); 2-year-old donors (obtained from a cohort of Swedish infants) (29); or umbilical cord blood samples (Stem Cell Services, NHS Blood and Transplant); or healthy laboratory volunteers. Intrahepatic lymphocytes were collected from donors after portal flush using cold preservation solution following removal of the right lobe of the donor's livers (non-pathological liver grafts preceding liver transplantation; Duke-NUS Graduate Medical School, Singapore) as previously described (30). The BM samples were obtained from routine hip joint operations (Newcastle University). Samples were filtered (40 µm), washed with phosphate buffered saline (PBS) and homogenized. Mononuclear cells from the above blood and tissues were isolated by standard density gradient centrifugation (Lymphoprep™ Axis Shield Diagnostics). Blood and tissue samples were cryopreserved and thawed before use. Adult and cord blood samples were collected after ethical approval by the Central Office for Research Ethics Committees (COREC, local research ethics committee Oxford), reference number COREC 04.OXA.010. Liver samples were collected after ethic approval of the Asian American Liver Center Ethic committee (Glean Eagle Hospital, Singapore), reference number PIEC/2012/037. For samples used for genomic DNA (gDNA) analysis, samples were collected after ethical approval by the University of Otago Human Ethics Committee (Health), reference number H14/046. All patients from the studies above provided their informed written consent. The collection of blood samples for the 2-year-old study cohort was approved by the Human Ethics Committee at Huddinge University Hospital, Stockholm, reference code 75/97, 331/02, and the parents provided their informed verbal consent. No written documentation of the participants informed approval was required, which was agreed to by the Human Ethics Committee and was according to the regulations at the time of the initiation of the study. (e) Frequencies of CD8+ (red), DN (blue) or CD4+ (green) cells within total CD161++ Vα7.2+ T cells in the indicated tissues. N = 32 (adult blood), 28 (2-year-old blood), 3 (cord blood), 3 (liver), and 4 (BM). ***P < 0.001 by two-way ANOVA with Dunnett's multiple comparisons test compared to adult peripheral blood. All other comparisons were non-significant. (F-i) The frequency of CD161++ Vα7.2+ T cells within (F) total T cells, (g) CD8+ T cells, (h) DN T cells, or (i) CD4+ T cells in the indicated tissues. N = 32 (adult blood), 28 (2-year-old blood), 3 (cord blood), 3 (liver), 4 (BM). Bars indicate mean ± SEM. Yr, year. ****P < 0.0001, ***P < 0.001, *P < 0.05 by two-way ANOVA with Dunnett's multiple comparisons test compared to adult peripheral blood. All other comparisons were non-significant.

Flow cytometry
The gating strategy used in this study is detailed in Figure 1A.

Data acquisition and statistical analysis
Data were collected on the MACSQuant Analyzer (Miltenyi Biotech) and were analyzed using FlowJo v9.8 (TreeStar). All graphs and statistical analyses were completed using GraphPad Prism software Version 6. All data are presented as means with SEM, unless otherwise indicated.
Next, coreceptor usage of CD161++ Vα7.2+ T cells in BM and liver was compared to that of blood from cord blood, 2-year-old, and adult donors. This revealed that the distribution of CD161++ Vα7.2+ T cell subsets within intrahepatic lymphocytes [intrahepatic lymphocyte sample (IHL); Figure 1C] and in the memory T cell fraction of BM ( Figure 1D) is similar to peripheral blood, where the CD8+ CD161++ Vα7.2+ T cell population constitutes the majority of CD161++ Vα7.2+ T cells ( Figure 1E). Interestingly, the expression of the CD8 coreceptor by blood CD161++ Vα7.2+ T cells increased with age, when comparing cord, 2-year-old, and adult peripheral blood.
Looking at the frequency of these cells within T cell populations, CD161++ Vα7.2+ T cells accounted for a mean of 2% of adult circulating T cells (Figure 1F), comparable to previous reports (27,34). Within different T cell subsets in adult peripheral blood, CD161++ Vα7.2+ T cells were found at an average frequency of 7% within CD8+, 11% within DN, and 0.05% of CD4+ T cells (Figures 1G-I). The frequency of CD161++ Vα7.2+ cells within T cells increased with age, as previously described. Interestingly, this increase occurred within CD4+ T cells as well as within CD8+ and DN T cells, suggesting that CD161++Vα7.2+ T cells expand with age regardless of coreceptor expression.
cell surface Phenotype heterogeneity of human cD161++ Vα7.2+ T cells The previously reported enrichment of CD4+ MAIT cells in lymph nodes in mice (8) suggests that different MAIT cell subsets may traffic to different tissues and, therefore, the expression of chemokine receptors was compared. Blood CD161++ Vα7.2+ T cell subsets were uniformly high for CCR2, CCR5, and CCR6, and had low or little expression of CCR3, CXCR3, and CXCR4 (Figures 2A,B), as previously reported in CD8+ MAIT cells (11). Although some differences in expression were seen for CCR2, CCR5, and CCR6, the largest difference between the subsets was the expression of CCR4 and CCR7. A small proportion (25%) of CD4+ CD161++ Vα7.2+ T cells were found to express CCR4, which was significantly higher than the CD8+ and DN cells. CCR7 expression was similarly only found on a minority (27%) of CD4+ CD161++ Vα7.2+ T cells, while completely absent on CD8+ and DN subsets.

cD4+ cD161++ Vα7.2+ T cells have reduced eomes expression and low cytotoxic Potential
Next we looked at the differences in TF expression in CD161++ Vα7.2+ T cell subsets. All three subsets of blood CD161++ Vα7.2+ T cells expressed high levels of RORγt and were found to be T-betlow (T-box expressed in T cells; Figures 3A-C), as previously described (35), although CD4+ CD161++ Vα7.2+ T cells had a slightly lower frequency of cells expressing RORγt compared to CD4− CD161++ Vα7.2+ T cells. In addition, CD8+ MAIT cells have also been shown to express the master regulator of innate-like T cells, PLZF, at high levels (11, 28). We found that all three subsets uniformly expressed PLZF, but the frequency of cells expressing PLZF was slightly, but significantly lower in the CD4+ CD161++ Vα7.2+ T cells compared to the CD8+ CD161++ Vα7.2+ T cells (Figures 3A,B). Next, the expression of ThPOK was examined, a TF associated with CD4 lineage commitment during thymic development (36), which also suppresses RORγt expression in iNKT cells (37). We found that all three subsets of CD161++ Vα7.2+ T cells expressed ThPOK (Figures 3A,B) Lastly, we found that CD4+ CD161++ Vα7.2+ T cells expressed significantly less Eomesodermin (Eomes) compared to CD8+ and DN CD161++ Vα7.2+ T cells in frequency ( Figure 3A), with an average of 50% of CD4+ CD161++ Vα7.2+ T cells expressing Eomes compared to CD8+ and DN CD161++ Vα7.2+ T cells, in which the majority of cells expressed Eomes. Previously, we have shown that CD8+ CD161++ Vα7.2+ T cells express granzyme A (GrA), GrK, and low levels of perforin at resting conditions (38). The greatly lower expression of Eomes by CD4+ CD161++ Vα7.2+ T cells suggests that these cells may have low cytotoxic potential. Indeed, we found that DN CD161++ Vα7.2+ T cells expressed GrA, GrK, and perforin at comparable frequencies to CD8+ cells, but significantly reduced frequencies of CD4+ CD161++ Vα7.2+ T cells expressed GrA, GrK, and perforin compared to CD8+ cells (Figures 3D-G). All subsets expressed little GrB at resting conditions. CD8+ MAIT cells can rapidly upregulate GrB and perforin following bacterial stimulation, which arms these cells with the ability to efficiently kill target cells (35,38). To investigate whether CD4+ CD161++ Vα7.2+ T cells can become killers, CD161++ Vα7.2+ T cells were stimulated with E. coli-treated THP1 cells for 24 h and GrB and perforin expression was analyzed. As shown in Figures 3H,I, the increase in the fraction of CD4+ CD161++ Vα7.2+ T cells expressing GrB was significantly lower compared to CD8+ CD161++ Vα7.2+ T cells. There was no significant difference between all subsets in their ability to upregulate perforin in response to bacterial stimulation. cD8+ and Dn cD161++ Vα7.2+ T cells have a higher capacity to secrete Th1 cytokines Next, to determine the functional differences between the three CD161++ Vα7.2+ T cell subsets, E. coli-treated THP1 cells were used to probe the MR1-dependent activation of CD161++ Vα7.2+ T cells. THP1 cells were cultured with PFA-fixed E. coli overnight before washing and co-culturing with PBMCs for 5 h. We did not observe a significant difference in the expression of the CD8 or CD4 coreceptors or proportions of CD8, DN, and CD4+ CD161++ Vα7.2+ T cells following E. coli stimulation due to change in coreceptor expression (Figures S2A-C in Supplementary Material) in control experiments. There was a clear production of interferon-γ (IFNγ) from all three subsets of CD161++ Vα7.2+ T cells after stimulation with E. coli-treated THP1 cells, which was completely blocked by the addition of an anti-MR1 blocking antibody (Figures 4A,B), consistent with previous work in this time frame (39). IL-17 was expressed by all CD161++ Vα7.2+ T cell subsets and its expression was also completely MR1 dependent (Figure 4C).

FigUre 4 | Continued
case, there was a significant difference between the frequency of marker expression in CD4+ CD161++ Vα7.2+ T cells compared to sorted CD4+ MR1-tetramer+ T cells due to some of the CD4+ CD161++ Vα7.2+ T cell population being non-MAIT, MR1-tetramer-negative cells (Figures 6C-F). However, we found that the frequency of Eomes+ cells as well as PLZF+ cells was significantly lower in CD4+ MR1-tetramer+ cells compared to CD8+ and DN MR1-tetramer+ cells, as found in CD4+ CD161++ Vα7.2+ T cells (Figures 6B-D). A small proportion of CD4+ MR1-tetramer+ cells also expressed CCR4, similar to CD4+ CD161++ Vα7.2+ T cells, although at a lower frequency compared to the frequency of CCR4+ cells within CD4+ CD161++ Vα7.2+ T cells and the expression was heterogeneous ( Figure 6E). CCR7 expression, by contrast, was significantly different between CD161++ Vα7.2+ T cell subsets but not between MR1-tetramer sorted cells (Figure 6F). Of note, in both CD161++ MR1-tetramer+ T cells and CD161++ Vα7.2+ T cells, Eomes and PLZF were coexpressed, while Eomes and PLZF were expressed in a mutually exclusive manner with CCR4 ( Figure 6G). This suggests that although some CD4+ CD161++ Vα7.2+ T cells did not stain for the MR1-tetramer, the pattern of expression of Eomes, PLZF, and CCR4 is shared between CD4+ CD161++ Vα7.2+ T cells and CD4+ MR1-tetramer+ cells. Further analysis of the coexpression of Eomes and PLZF with  Thus far, no major differences could be observed between the CD8+ and DN MAIT/CD161++Vα7.2+ T cells in phenotype and function in our model. Given that the ratio of DN MAIT cells to CD8+ MAIT cells has been reported to increase with age (15,16), we asked whether there may be difference in survival after antigen-dependent activation. Therefore, PBMCs were added to E. coli-treated THP1 cells and expression of the apoptosis marker Annexin V on CD161++ Vα7.2+ T cells was assessed ( Figure 7A). This in fact showed that the increase in Annexin V-expressing cells following stimulation was significantly greater in DN CD161++ Vα7.2+ T cells, compared to CD8+ cells ( Figure 7B). Thus, although both CD8+ and DN CD161++ Vα7.2+ T cells are similarly activated in an MR1dependent manner in our model, the CD8+ CD161++ Vα7.2+ T cell population may be protected from activation-induced cell death.
To further investigate the role of the CD8 coreceptor in the function of MAIT/CD161++ Vα7.2+ T cells, PBMCs were incubated with E. coli-treated THP1 cells over 5 h in the presence or absence of an anti-CD8α blocking antibody (clone LT8). Due to the residual loss of CD8 staining from the presence of the anti-CD8 antibody, CD4−CD161++ Vα7.2+ T cells were gated in the following experiments. We found a dose-dependent reduction in the frequency of IFNγ+ CD161++ Vα7.2+ T cells with CD8 coreceptor blockade, compared to the isotype control (Figures 7C,D). This effect was also seen when measuring different outputs of TCR signaling, namely TNFα, CD107a, and macrophage inflammatory protein-1β (Figures 7E-H). Interestingly, the degree to which CD8-coreceptor blockade affected the response of CD161++ Vα7.2+ T cells differed depending on the response

DiscUssiOn
In this study, we have performed a detailed comparison of the CD4+, CD8+, and DN CD161++ Vα7.2+ T cell subsets in parallel, in terms of their phenotype, cytokine secretion, cytotoxicity, and TF expression, which has highlighted their distinct and overlapping characteristics. First, we found that CD4+ CD161++ Vα7.2+ T cells have a lower frequency of cells expressing Eomes and PLZF compared Human CD161++ Vα7.2+ T Cell Subsets Frontiers in Immunology | www.frontiersin.org August 2017 | Volume 8 | Article 1031     (42). Interestingly, all subsets of intrahepatic CD161++ Vα7.2+ T cells expressed CD56 at high levels, which was associated with a higher effector function, especially in the CD4+ subset, secreting abundant IFNγ in response to MR1-presented antigen. As CD56 expression has been previously associated with increased cytotoxic effector function of T cells (43,44), CD4+ CD161++ Vα7.2+ T cells may also have heterogeneous cytotoxic capacities depending on the tissue they reside in. Increased CD56 expression in T cells and NK cells have been reported in in vitro cultures of cells with common γ-chain cytokines (43,45). It is, therefore, possible that the intrahepatic cytokine milieu upregulates CD56 expression on all MAIT cell subsets and lowers their activation threshold and/or skews them toward a Th1 response. Indeed, intrahepatic lymphocytes are dominated by rapidly acting innate cells, including MAIT cells, γδ T cells, NK cells, and T cells expressing NK receptors, e.g., CD56, and constitutive expression of cytokines, such as IL-15 (46) and IL-7 (30), may activate and induce CD56 upregulation in MAIT cells.
In addition, we found that all three CD161++ Vα7.2+ T cell subsets expressed ThPOK, the master regulator of the CD4 lineage (47), at an intermediate level (ThPOKlow). Whether this TF may be expressed at a higher level in the CD4+ subset of CD161++ Vα7.2+ T cells during their development is unknown. Interestingly, a recent report showed that developing MAIT cells in the thymus transition from mostly CD4+ or CD4+ CD8+ to DN or CD8+ cells, which does not occur in PLZF-deficient mice (48). This suggests that the expression of coreceptors during development may be determined by the maturation status of MAIT cells, rather than CD4/CD8 lineage commitment signals regulated by TFs such as ThPOK. Of note, ThPOK also negatively regulates Th17 differentiation and, thus, only cells that express low levels of ThPOK are permissive for the differentiation of type-17 iNKT cells (20,37,49). As all subsets of CD161++ Vα7.2+ T cells expressed IL-17 to a similar level, CD161++ Vα7.2+ T cells may also express a reduced amount of ThPOK that is permissive for the expression of RORγt. Human CD161++ Vα7.2+ T Cell Subsets Frontiers in Immunology | www.frontiersin.org August 2017 | Volume 8 | Article 1031

FigUre 7 | Continued
The most accurate method of defining a MAIT cell currently available is by identification of cells that bind the MR1-tetramer. The MR1-tetramer has been developed (4, 5) but was not widely available until recently, and so the most commonly used method for identifying MAIT cells has previously been to look at CD161++ Vα7.2+ T cells. These cells have been shown to overlap with CD161++ MR1-tetramer stained cells (4), particularly in CD8+ and DN cells (27). In this study, we show that although the MAIT cell TCR Vα7.2-Jα33/12/20 is enriched within the CD4+ CD161++ Vα7.2+ T cell population, not all CD4+ CD161++ Vα7.2+ T cells could be detected by the 5-OP-RUloaded MR1-tetramer. This suggests that some CD4+ CD161++ Vα7.2+ T cells may be conventional T cells expressing the Vα7.2 chain, consistent with a previous report (50). This heterogeneity of the CD4+ CD161++ Vα7.2+ T cell population can explain some of the differences observed between CD4+ and CD8+/DN CD161++ Vα7.2+ T cell populations, as the latter subsets mostly overlap with CD161++ MR1-tetramer+ cells. This shows that particularly when looking at CD4 populations, the MR1-tetramer should be used to accurately identify MAIT cells. Thus, when we compared the phenotypic differences between CD161++ Vα7.2+ T cell subsets with MR1-tetramer defined MAIT cell subsets, there was a significant difference between the level of marker expression in CD4+ CD161++ Vα7.2+ cells and CD4+ MR1-tetramer+ cells. Interestingly, however, CD4+ MR1-tetramer+ cells sorted using the MR1-tetramer were found to have significantly lower levels of Eomes and PLZF, and a significantly higher level of CCR4 compared to their CD4− counterparts, as found in CD161++ Vα7.2+ T cell subsets. The pattern of coexpression of these markers was also found to be similar between CD4+ MR1-tetramer+ cells and CD4+ CD161++ Vα7.2+ T cells. Thus, although there is less heterogeneity between MAIT cell subsets as defined by MR1-tetramers compared to CD161++ Vα7.2+ T cell subsets, the CD4 subset seems to be inherently more heterogeneous compared to the CD8/DN MAIT cell subset. Nevertheless, it will be important for future studies to further confirm any differences between MAIT cell subsets using the MR1-tetramer. CD8+ and DN CD161++ Vα7.2+ T cells were functionally and phenotypically similar in this study and largely overlapped with 5-OP-RU-loaded MR1-tetramer stained cells. Thus, whether the CD8 coreceptor may affect the activation of CD8+ CD161++ Vα7.2+/MAIT cells was investigated. Although CD8α was not necessary for the activation of MAIT cells in our model, CD8+ MAIT cell responses were reduced in a dose-dependent manner by the addition of an anti-CD8α blocking antibody. As the structure and residues of the CD8-binding domain of MHC class I is conserved between MR1 and MHC class I molecules, it has been suggested that MR1 may bind CD8αα (4,33). It is well attested that the binding of CD8 to MHC class I stabilizes the TCR/peptide-MHC class I interaction (51), and the results here support the idea that the CD8 coreceptor similarly stabilizes the MAIT TCR/MR1 interaction. We found that there was a hierarchy in cellular responses of MAIT cells that are affected by CD8 coreceptor blockade-the order was as follows, from least affected to most affected: MIP-1β < CD107a < TNFα ≤ IFNγ. This series is similar to the hierarchy in cellular responses elicited by peptide-MHC class I stimulation (52). Interestingly, studies using other activation models, such as Helicobacter pylori, have found a higher capacity of CD8+ MAIT cells to secrete cytokines and degranulate, suggesting that the CD8 coreceptor may play a more significant role in MR1-dependent activation in response to certain pathogens, or in response to low doses of antigen (53). Importantly, however, we cannot rule out that, in our assays, CD8 is acting as a simple adhesion molecule, or that the antibody is hindering the interaction between MR1 and the MAIT cell TCR. Furthermore, the CD8 coreceptor blockade may only be modifying the activation of CD8αβ+ MAIT cells. This is because the CD8β chain is required for efficient coreceptor function (54). A large fraction of the CD8+ MAIT cells express the CD8αα homodyne, which sequesters p56lck away from the TCR due to its exclusion from lipid rafts and, therefore, are thought to be inhibitory (55,56). In order to confirm the function of the CD8 coreceptor on MAIT cell activation, further studies disrupting the potential CD8-binding site of MR1 will be necessary (57).
In conclusion, we have explored the heterogeneous as well as homogeneous phenotypes and functions of the three defined subsets of CD161++ Vα7.2+ T cells. Differences in TF expression, chemokine receptor expression, and capacity to secrete Th1 cytokines support the notion that different CD161++ Vα7.2+ T cell subsets, and the MAIT cells contained within this population, may play distinct roles in health and disease, and future studies will be warranted to further investigate the development and function of these cells.

eThics sTaTeMenT
The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institutions' human research committees. Adult and cord blood samples were collected after ethical approval by the Central Office for Research Ethics Committees (COREC, local research ethics committee Oxford), reference number COREC 04.OXA.010. Liver samples were collected after ethic approval of the Asian American Liver Center Ethic committee (Glean Eagle Hospital, Singapore), reference number PIEC/2012/037. For samples used for gDNA analysis, samples were collected after ethical approval by the University of Otago Human Ethics Committee (Health), reference number H14/046. All participants from the studies above provided their informed written consent. The collection of blood samples for the 2-year-old study cohort was approved by the Human Ethics Committee at Huddinge University Hospital, Stockholm, reference code 75/97, 331/02, and the parents provided their informed verbal consent. No written documentation of the participants informed approval was required, which was agreed to by the Human Ethics Committee and was according to the regulations at the time of the initiation of the study. aUThOr cOnTriBUTiOns AK designed and performed experiments, and wrote the manuscript. AJ, RH, JF, and LW performed experiments, and ES-E provided samples. AC provided the MR1-teramer and valuable advice, JU and CW provided advice and support, and PK supervised research work and data analysis.

acKnOWleDgMenTs
We thank Professor Antonio Bertoletti for providing the intrahepatic lymphocyte samples and Professor Ted H. Hansen for providing the anti-MR1 blocking antibody. The MR1-tetramer technology was developed jointly by Professor James McCluskey, Professor Jamie Rossjohn, and Professor David Fairlie (5), and we thank Professor James McCluskey for the donation of the MR1tetramer. Further material was produced by the NIH Tetramer Core Facility as permitted to be distributed by the University of Melbourne. We also thank Dr. Bonnie van Wilgenburg for comments and critical reading of this manuscript.