Identification of endogenous NKT cells in situ with α-GalCer loaded CD1d tetramers.
We have used CD1d tetramer for immunohistological analysis of NKT cells in a range of mouse tissues. To validate the specificity of our approach, fresh frozen spleen sections from BALB/c wild type (WT) or BALB/c.CD1d/− mice were stained with CD1d tetramers loaded with α-GalCer or left unloaded (presumably carrying endogenous lipids) (Fig. 1a and Supplementary Fig S1). Within the WT spleen, cells could be identified that had clearly bound the α-GalCer-loaded CD1d tetramer with limited staining observed on WT sections stained with the unloaded CD1d tetramer or CD1d−/− mouse spleen sections stained with CD1d-α-GalCer tetramer. Though, in all three groups, we observed rare examples of CD1d tetramer staining that appeared not to associate with the TCRβ stain (Supplementary Fig S1b), thus the co-localisation of CD1d tetramer and TCRβ was investigated. Voxel plots of the three groups indicated a clear association within BALB/c WT sections of voxels stained with CD1d-α-GalCer and TCRβ (Supplementary Fig S1c). Analysis of Pearson’s correlation and the area co-stained by CD1d-α-GalCer tetramer and TCRβ both indicated a positive correlation between CD1d-α-GalCer and TCRβ staining on WT sections compared to negative control stained sections (Supplementary Fig S1d and e). Taken together, these data highlight the ability of the CD1d-α-GalCer tetramers to specifically stain endogenous NKT cells in situ. Importantly, they also demonstrate the need to co-stain with T cell specific markers such as TCRβ or CD3 and include negative controls for stain and tissue to best eliminate, albeit rare, non-NKT cell staining from analysis.
Having established the technique to reliably identify NKT cells in situ, we next examined the location of NKT cells in primary and secondary lymphoid organs. On average, roughly 100 NKT cells per field of view could be observed within BALB/c spleen (Fig. 1b), the majority (~ 70%) of which were found within the TCZ rather than the BCZ or RP (Fig. 1c). Within these regions, NKT cells were scattered either alone or in small clusters. We also examined spleens from C57BL/6 (B6) mice and detected roughly 40 NKT cells per field of view (Supplementary Fig S2a and b). There was also a clear difference in the location of NKT cells between the two stains of mice. Where in the spleens of B6 mice, more NKT cells were found in the RP with the remaining NKT cells in roughly equal proportions in the TCZ and BCZ (Supplementary Fig S2c). In spleens from both BALB/c and B6 mice, we detected some CD1d tetramer+ cells that are not likely to be NKT cells because they did not co-stain for CD3/TCR. Similar infrequent staining was observed in CD1d−/− spleens and WT spleens stained with CD1d tetramer loaded with an irrelevant glycolipid antigen, disialoganglioside (GD3) (Fig. 1a, Supplementary Fig S2d).
The distribution of NKT cell subsets in BALB/c mice was also investigated based on the expression or absence of CD4, which defines many, but not all NKT cells, as shown by flow cytometry (Fig. 1d). In general, CD4 expression in the spleen was observed throughout the organ, though as expected the greatest concentration of expression was associated with CD3 within the TCZ (Fig. 1e), representing conventional CD4 T cells. Using colocation analysis between CD3+CD1d-α-GalCer+ and CD4+ stains to separated NKT cells into CD4 expressor and non-expressor channels, the distribution of CD4+ and CD4− NKT cells could be determined (Fig. 1f). Interestingly, while the majority (~ 75%) of NKT cells within the TCZ expressed CD4, there were roughly equal proportions of CD4+ and CD4− NKT cells within the RP or BCZ expressed this marker (Fig. 1g). These data suggest that there is differential distribution of NKT CD4+ and CD4− subsets may reflect differential expression of chemokine receptors or adhesion molecules 35,36.
Within peripheral (brachial) lymph nodes (LN) there were far fewer NKT cells observed per field of view than in spleen sections, nonetheless, compared to negative controls NKT cells could still be observed (Fig. 2a and b). By scanning multiple sections, we were able to determine that the majority of NKT cells were localised within the paracortex rather than the medulla or B cell follicles of lymph nodes (Fig. 2c).
The location of NKT cells in thymus was examined by costaining for thymic cortical and medullary regions using CD205, cytokeratin 5 and CD3 and CD1d-α-GalCer (Fig. 3a and b). This demonstrated that, consistent with previous reports 18 the vast majority of NKT cells were located within the medulla. Furthermore, the medullary NKT cells tended to localise close to, or within, the corticomedullary junction (CMJ) as a higher density of these cells were observed within 100 µm of the CMJ (Fig. 3c). Thymic NKT cells can also be subdivided into CD4+ and CD4− fractions, as shown by flow cytometry (Fig. 3d) and immunohistology (Fig. 3e). Following co-localisation analysis (Fig. 3f and 3g), we found that within both thymic regions most NKT cells expressed CD4 (~ 75% within the cortex and ~ 60% within the medulla).
These results demonstrate that NKT cells occupy specific locations within the primary and secondary lymphoid organs and, for the most part, they are located in similar regions to conventional T cells in BALB/c mice. We have also shown that while NKT can be divided into CD4 + and CD4- subsets, for the most part both populations were intermingled, although there were some differences in the ratio of these subsets in different locations.
Thymic medullary structure in mice lacking NKT cells.
A recent study showed that emigration of mature thymocytes from the thymus was dependent on the type-2 cytokine receptor IL-4Rα chain 37. Lack of IL-4Rα and impaired thymic emigration led to accumulation of mature thymocytes in the medulla around perivascular spaces, leading to the formation of large medullary areas devoid of epithelial stromal cells. Moreover, this study also showed the appearance of these medullary epithelial cell-free areas in CD1d-deficient mice, suggesting a key role for NKT cells as a source of IL-4 that regulates thymic emigration37. CD1d-deficient mice lack both type I and type II NKT cells, both of which can produce IL-4 and IL-13. Therefore, we sought to explore if these epithelial cell-free areas (that we refer to as voids) were specifically due to the lack of type I NKT cells by comparing medullary thymic epithelial cell staining in WT, CD1d−/− and Jα18−/− thymuses (the latter deficient in type I but not type II NKT cells). While we were able to detect some areas in the medulla that were devoid of thymic epithelial cells (as outlined with green line) (Fig. 4a), these did not resemble the clear ring like structures that were previously reported37. Moreover, we found no significant difference in the number of voids in CD1d−/− (Fig. 4b) and while the average size of these was marginally increased in CD1d−/− mice, this was also not significant (Fig. 4c). We also failed to detect any increase in the size of medullary holes in type I NKT cell deficient TCRJα18−/− mice. As these data were acquired in B6 strain background, we also carried out similar studies in BALB/c mice where NKT cells produce higher levels of IL-4, but again, we were unable to detect a clear increase in the appearance of medullary voids (Supplementary Fig S3). As a positive control for detection of medullary epithelial cell voids, we also tested thymuses from NOD mice, as these are known to contain large numbers of B cells in the thymic medulla that lead to enlarged perivascular spaces that disrupt thymic epithelial structure 38. Clear evidence of medullary voids were detected in NOD mouse thymi (Fig. 4a). Another consideration was that CD1d-deficient mice have increased MAIT cells, particularly on the BALB/c background 39. Therefore, we also examined Vα19 TCR transgenic.Cα−/− mice (Vα19Tg) and MR1−/− mice, which have increased, and decreased, MAIT cells respectively 40,41 for medullary voids, but these were also not significantly different from control B6 mice (Supplementary Fig S3). Taken together, while we have confirmed the presence of NKT cells in the thymic medulla, we have been unable to verify that the absence of any thymic NKT cell population has an impact on thymic medullary architecture.
In situ observation of NKT cells in peripheral organs.
As well as primary and secondary lymphoid organs, NKT cells reside in a range of non-lymphoid organs, particularly in the liver, small intestines and lungs. Thus, we endeavoured to detect NKT cells in a range of peripheral organs using CD1d tetramer staining. While CD1d-α-GalCer tetramer+ cells were detected within the small intestine, lungs, kidneys and heart of BALB/c WT mice (Supplementary Fig S4 to S7), many of these stained brightly with the tetramer, but not with CD3. Furthermore, the importance of the unloaded CD1d tetramer control stain was highlighted by the existence of CD3− cells that bound to unloaded CD1d tetramer in these tissues. For example, in the small intestine non-specific (CD1d tetramer positive, but CD3 negative) staining could be observed prominently in the villi, regardless of whether CD1d was loaded with negative control antigen (Supplementary Fig S4a) or α-GalCer (Supplementary Fig S4b). Similarly, in the lung (Supplementary Fig S5), kidney (Supplementary Fig S6) and heart (Supplementary Fig S7), GD3 and α-GalCer loaded CD1d tetramer positive, but CD3 negative staining could be also observed. Importantly, despite these non-specifically stained cells, CD3+CD1d/α-GalCer+ cells were also detected, albeit infrequently, in Payers Patch (Supplementary Fig S4b) and the villi (Supplementary Fig S4c) of small intestine. Rare NKT cells were detected in the lungs within the alveolar ducts (Supplementary Fig S5b). In kidney, NKT cells could be observed between convoluted tubules (Supplementary Fig S6b) and in the heart within the myocardium between perimysical septa (Supplementary Fig S7b). Because NKT cells were so infrequent in lung, kidney and heart (1–4 per section), it was difficult to determine their preferred location.
Relocation of NKT in the spleen following α-GalCer stimulation.
Having demonstrated the in situ location of NKT cells in lymphoid tissues, we next examined what happened to the location of these cells following in vivo activation. Mice were injected intraperitoneally with 2µg α-GalCer and the expansion of NKT cells was determined on days 3 and 5 post injection. As expected, FACS analysis of spleen cell suspensions from these mice showed a rapid increase in NKT cell numbers in the spleen 3 days post injection, which decreased by day 5 post injection (Fig. 5a and b). This was also reflected by a clear increase in CD1d-α-GalCer tetramer staining of spleen tissue sections taken at day 3 after activation, and subsequent reduction in the extent of this staining at day 5 (Fig. 5c and d). As expected CD1d-α-GalCer staining was associated with CD3 staining (Supplementary Fig S8). As previously shown, CD3+CD1d-α-GalCer+ NKT cells in unstimulated mice were observed throughout the spleen with largely equal proportions of NKT cells in the BCZ and RP, while most (~ 10x as many) were located within the splenic TCZ. By day 3 following antigen stimulation, the proportion of NKT cells in all three regions increased, and some redistribution of NKT cells was apparent by day 5 (Fig. 5e). While NKT cells within the TCZ remained the largest population, increasing ~ 2.5-fold 3 days post stimulation, the BCZ saw the greatest increase in NKT cell density such that by day 3, the population of NKT cells in this region had increased ~ 7-fold. We also observed a ~ 3-fold increase in the frequency of NKT cells within the RP 3 days following stimulation. By day 5, the frequency of CD3+CD1d-α-GalCer+ NKT cells in the TCZ and RP had declined to levels that were similar to unstimulated mice. In contrast, the frequency of cells in the BCZ remained ~ 2.8-fold higher compared to unstimulated mice. The increased CD3+CD1d-α-GalCer+ NKT cells within the BCZ may reflect their role in modulating B cell responses following activation.
We next investigated NKT cells in the periphery following stimulation. The liver is a major non-lymphoid organ known to contain a large population of NKT cells. Following injection of α-GalCer, an expanded population of NKT cells was detected that followed a similar expansion pattern as in the spleen; a large expansion 3 days post stimulation, which had contracted by day 5 (Supplementary Fig S9a and b). Detecting NKT cells in situ in this organ proved challenging due to its large size, and low frequency of lymphocytes in general; nonetheless, NKT cells were detected, albeit infrequently, within the liver sinusoids. Consistent with the cytometry data, NKT cells were more frequent at day 3 where it was possible to observed 2–7 NKT cells per field of view. In livers of either unstimulated or day 5 post challenge mice, however, they remained difficult to locate (1–3 per section). Importantly, regardless of the time point, these cells co-stained with CD3 and this staining was not observed in negative control sections (Supplementary Fig S9c).
Multiparameter histo-cytometry of splenic NKT cells, T cells, B cells and antigen-presenting cells.
The cellular environment within lymphoid organs is complex and involves many different subsets of immune cells each influencing the actions of the others. Indeed, NKT cells interact with and modulate the responses of a range of innate and adaptive immune cells 42. These varying cell types are difficult to investigate together with typical histological techniques due to lack of methods to stain for the multitude of cell surface receptors required to fully identify the diverse array of immune cells in situ. Therefore, we used multiparameter histo-cytometry 43 to visualise NKT cells in the context of other immune cells. To this end, spleen sections were stained with a range of cell surface receptors (CD11c, CD11b, CD3, CD4, CD1d/α-GalCer, B220 and MHC-II) and cell nuclei with DAPI and the separate fluorescence of each stain was determined by spectral unmixing and compensation (Fig. 6a and Supplementary Fig S10a). Cells were segmented based on the DAPI strain (Supplementary Fig S10b) and separate cell populations determined by conventional cytometry analysis (Fig. 6b). Clear populations of B cells and T cells as well as NKT cells and various APC populations could be identified and were plotted back to cells within the image and the original stains (Fig. 6c). CD4 positive and negative T and NKT cells proved difficult to separate using this technique, likely owning to their being closely packed and intermingled in the similar spatial location, namely the TCZ. Overall, however, the various cell populations could be mapped within the same histological regions within the spleen as they were observed in the original stained sample (Supplementary Fig S10d), but with greater clarity, providing a better way to analyse these cells in association with other diverse cell types. This approach supports our previous observations that NKT cells were largely located within the TCZ, of which most were CD4+ (Fig. 6d, zoom panels III, VI and VII), while a small number of CD4− NKT cells were found in the BCZ (Fig. 6d, zoom panels V and IX) and RP (Fig. 6d, zoom panels IV and VIII). Some NKT cells were also found in marginal zone (Fig. 6d, zoom panels V and VII). As expected, a majority of the APCs were observed outside the white pulp with macrophages prominent throughout the RP, while CD11b− and CD11b+ DCs mostly associated with the marginal zone (Fig. 6d, zoom panels IV, V, VII and VIII). In particular, CD4− NKT cells were observed in amongst the CD11b− DCs (Fig. 6d zoom panel IV and V).
In situ identification of MAIT cells with MR1-5-OP-RU tetramers.
Very little is known about the location of MAIT cells within tissues. Similar to the technique to stain for NKT cells, MR1 tetramers loaded with MAIT cell-specific antigen 5-OP-RU were used to identify MAIT cells. MAIT cells are much rarer than NKT cells in mice 41 so to begin with, we used the Vα19Tg mice, which contain a much larger proportion of MAIT cells than WT mice. A clear population of brightly stained MR1-5-OP-RU tetramer+ CD3+ T cells was observed in spleen, lymph nodes and thymus of Vα19Tg mice. Importantly, this staining was not evident on sections stained with control acetyl-6-formylpterin (Ac-6-FP)-loaded MR1 tetramer (Fig. 7), supporting the specificity of this staining for MAIT cells. In all organs, the pattern of MR1-5-OP-RU tetramer staining in the Vα19Tg mice was similar to that observed for NKT in WT mice, with the major difference being the greater number of MAIT cells observed in the Vα19Tg mice. In spleen, MAIT cells were mainly detected in the TCZs and less frequent cells were detected in the BCZs and RP (Fig. 7a). Similarly, MAIT cells were prominent in the LN paracortex, though unlike NKT cells, MAIT cells were also observed in the Vα19Tg LN medulla (Fig. 7b). Curiously, non-specific staining in the region of the glass slide where OCT was located was consistently observed with the 5-OP-RU loaded MR1 tetramer, but not MR1-Ac-6-FP (Fig. 7b and S13a). The reason for this effect is unknown but may be a result of the natural fluorescence of 5-OP-RU 44. In the thymus, the vast majority of MAIT cells were detected in the thymic medulla (Fig. 7c). These results show that, similar to the identification of NKT cells with CD1d-α-GalCer tetramers, MAIT cells can also be observed in situ with the use of MR1-5-OP-RU tetramers.
Next, we examined the location of MAIT cells in non-TCR transgenic C57BL/6 (B6), BALB/c and B6-MAITCAST (CAST) mice (Fig. 8 and Supplementary Fig S11). The latter were also tested because this congenic strain is reported to have an increased number of MAIT cells, due to increased intrathymic selection of these cells 45. While infrequent, nonetheless, MAIT cells could be readily observed in spleens stained with MR1/5-OP-RU tetramer in all three mouse strains. Importantly, these cells were not detected in the spleens from B6, BALB/c or CAST mice when stained with MR1/Ac-6-FP, nor were they detected in B6.MR1-/- or CAST.MR1-/- mice stained with MR1-5-OP-RU tetramer (Fig. 8b, S11b and e). Furthermore, MR1/5-OP-RU staining associated with CD3 staining. Similar numbers of MAIT cells (~ 10–20 cells) were observed in the spleens of B6 and CAST mice, while in BALB/c spleen only 5 cells on average could be observed. Interestingly, and in contrast to MAIT cells in the Vα19Tg mice, for each of the non-transgenic strains tested, most of the MAIT cells were detected outside the TCZ; collectively within the BCZ and RP of the spleens (Fig. 8c and S11c and f). We also attempted to investigate the locational differences between the NKT cells and MAIT cells by co-staining spleen sections of B6 and BALB/c mice with both CD1d-α-GalCer and MR1-5-OP-RU tetramers (Supplementary Fig S12a and b). A challenge with this approach is that both tetramers work optimally using the same fluorochrome, so a suboptimal fluorochrome was used for CD1d-α-GalCer tetramer staining. Furthermore, MAIT cells were more readily detectable in B6 mice while NKT cells were more detectable in BALB/c mice, so both were tested. Nonetheless, this co-staining supported the concept that NKT cells and MAIT cells occupy different locations in spleen. The thymuses of all three strains of mice were also stained with MR1-5-OP-RU tetramer (Supplementary Fig S13a to c). While a very small number of MR1/5-OP-RU + CD3+ cells could be seen in the cortex of the thymus of B6 and CAST mice, they were undetectable in BALB/c mice. This is not surprising because we have previously published MAIT cells are exceedingly rare in mouse thymus 41. Furthermore, a similar number of MR1 tetramer+ but CD3- cells were also observed in negative controls, which again highlights the importance of appropriate controls in attempting to detect MAIT cells in situ. Nevertheless, with appropriate caution based on detection of very few cells, this suggests that MAIT cells may preferentially reside in the cortex of the thymus of non-Vα19 TCR transgenic mice, in contrast to their medullary location in Vα19 TCR transgenic mice (Fig. 7c).
The distribution of MAIT cells following in vivo activation was investigated by intranasal (i.n.) administration of 5-OP-RU in CAST mice and MR1-5-OP-RU tetramer staining of mediastinal LN (Supplementary Fig S14). Similar to the MR1 tetramer staining in the spleen, while only a small number of MAIT cells could be observed in these LNs, these were not detected than when the same tissue was stained with negative control Ac-6-FP loaded MR1 tetramer. Not-withstanding the small numbers of MAIT cells detected, there appeared to be an increase in their abundance following 5-OP-RU challenge (Supplementary Fig S14c) and most were located within or on the edge of the paracortex (Supplementary Fig S14b and d).
While interpretation of these data is limited by the scarcity of cells detected in non-TCR transgenic mice, these results suggest that MAIT cells in WT mice occupy specific locations within lymphoid tissues, primarily located in regions of the spleen that differ from NKT cells and conventional T cells and that this is not reflected by the location of MAIT cells in TCR transgenic mice.