|Home | About | Journals | Submit | Contact Us | Français|
γδT cells are prevalent in the mucosal epithelia and are postulated to act as sentries to maintain tissue integrity. What these γδT cells recognize is poorly defined, but based on the restricted T cell receptor (TCR) repertoire, the notion that they are selected by self-antigens of low complexity has been widely disseminated. We present data demonstrating that generation of the restricted TCR Vγ gene repertoire of intestinal intraepithelial lymphocytes is regulated by IL-15, which induces Vγ gene segment-specific local chromatin modifications and enhanced accessibility conducive for subsequent targeted gene rearrangement. This cytokine-directed tissue-specific TCR repertoire formation likely reflects distinct TCR repertoire selection criteria for γδ and αβT cell lineages adopted for different antigen recognition strategies.
Intestinal intraepithelial lymphocytes (i-IELs) found within the intestinal epithelial layer are an essential component of mucosal immunity. Similar to T cell subsets in other lymphoid tissues, i-IELs are composed of two T cell types, αβ and γδT cells. In contrast to other tissues, however, γδT cells are prevalent in the mucosal epithelia of most vertebrates and are thought to be a major source of secreted factors necessary for the maintenance of tissue integrity subsequent to infection, transformation, or cell injury1-3.
Despite the long history of i-IELs, their origin and ligand specificity remain uncertain. Earlier work has indicated that αβ i-IELs originate exclusively from the thymus in normal mice4, but whether γδ i-IELs are subject to similar developmental constraints is unclear. The majority of human and mouse γδ i-IELs express the homodimeric αα form of the CD8 coreceptor and exhibit restricted Variable (V)γ and Vδ gene usage. In C57BL/6 (B6) mice, for example, Vγ5 expressing cells constitute the predominant i-IEL population in the small intestine5-7. Mouse strain-specific preferential usage of Tcrg and Tcrd genes among i-IELs arises from complex molecular and cellular controls with evidence for genetic linkage to the Tcrg, Tcrd and H2 loci6,8,9, but the exact mechanism is unknown. Two non-mutually exclusive mechanistic processes can be considered to account for the restricted TCR gene usage: cellular selection and programmed TCR gene rearrangement. A hallmark of adaptive immunity is the cellular selection process geared toward forming a population of cells with antigen receptor repertoires that can recognize the entire external universe of antigens. This selection pressure acts on immature cell subsets expressing randomly generated clonal antigen receptors. Although the TCR-driven selection shaping the αβTCR repertoire of conventional αβT cells constitutes a fundamental theme in immunology, whether a similar process is needed for γδT cell development, and specifically for γδ i-IELs, remains a matter of debate10-12.
Alternatively, evidence exists for developmentally programmed antigen receptor gene rearrangements13, suggesting that the generation of the restricted TCR repertoire among i-IELs may also arise by a directed TCR gene rearrangement process. In this scenario, polymorphic cis-acting elements controlling Tcrg and Tcrd gene accessibility and transcription may be major contributors to the observed differences in Vγ and Vδ gene usage in different mouse strains. In strong support for this notion, it has been demonstrated that the relatively high prevalence of Vγ5+ cells in thymocytes of the 129 strain (~15% of γδTCR+ thymocytes versus <5% in B6 mice) can be attributed largely to specific genetic traits of the Tcrg locus itself14.
Generation of γδT cells is dependent on interleukin 7 receptor (IL-7R)-mediated signals, which are thought to temporally modulate chromatin accessibility at the Tcrg locus15,16. IL-7 belongs to the common γ chain cytokine superfamily (γc) that also includes IL-2, IL-4, IL-9, IL-15 and IL-21. One central biochemical pathway triggered by the γc cytokines to regulate gene transcription is the Janus kinase (JAK)-Signal Transducer and Activator of Transcription (STAT) cascade, and except for IL-4, activation of JAK3-STAT5 is primarily targeted by these cytokines. IL-15 exerts a strong influence on i-IELs, especially in the generation and maintenance of the dominant CD8ααThy1–Vγ5+ γδ i–IEL population7,17. The ligand for the CD8αα homodimer is the non-classical MHC class I (class Ib) molecule TL, which is expressed highly on epithelial cells of the small intestine18. The ligand(s) for the mouse γδTCRs on i-IELs is unknown, although in humans, the restricted i-IEL γδTCR repertoire has been suggested to focus on surveying for the stress induced ligands, MICA and MICB19. These molecules can also be recognized with high affinity by the natural killer (NK) cell receptor NKG2D20, which is expressed on human, but not mouse, γδ i-IELs, and whose expression is regulated by the cytokine IL-15 (ref. 21). A small fraction of mouse γδT cells (0.1-2%), including γδ i-IELs, show reactivity to the MHC class Ib molecule T22 (ref. 22). These limited documentations of γδT cell ligand specificity have been used to buttress the notion that γδT cells in general recognize a limited set of self-antigens using a restricted TCR repertoire, especially in the mucosal epithelia.
In this study we show that the preferential Vγ5 gene usage among γδ i-IELs is controlled by IL-15, which specifically modulates Vγ5 gene segment-associated histone acetylation. The chromatin domain(s) controlled by IL-15 is distinct from those under the control of IL-7, which eventually form the γδTCR repertoire found in the spleen and blood. Moreover, CD8αα+ Vγ5+ cells can be generated intrathymically by IL-15 and are especially dependent on IL-15-activated STAT5. This study constitutes the first demonstration of a cytokine-directed generation of tissue-specific TCR repertoire by selective control of local V gene chromatin accessibility.
In preliminary experiments, we performed fetal thymic organ cultures (FTOCs) to determine whether other cytokines could substitute for IL-7 in regulating the Tcrg locus. When Il7r-/- fetal thymi were cultured with different combinations of cytokines, only IL-15 was able to rescue γδT cell development in the absence of IL-7R signaling (Fig. 1a and data not shown). IL-2, the cytokine most closely related to IL-15 in function that signals via shared β and γ chains, was not effective in promoting γδT cell development in FTOCs (data not shown). The extent of γδTCR+ thymocyte generation in Il7r-/- FTOCs by the addition of IL-15 was variable (2-25% of the total thymocytes). Unexpectedly, in cultures where γδTCR+ thymocytes developed, all uniformly expressed the Vγ5 TCR chain, the predominant γTCR found among γδ i-IELs (Fig. 1a).
To verify whether IL-15 can promote the development of Vγ5 expressing cells independent of IL-7 in vivo, we generated Il15 transgenic (TgN) Il7r-/- mice. Expression of IL-15 under the control of the Major Histocompatibility Complex (MHC) class I gene promoter23 led to the exclusive development of Vγ5+ γδ cells in the thymus, spleen and intestine in the absence of IL-7R-mediated signals (Fig. 1b), confirming and extending the FTOC results. This effect was not simply the result of IL-15 acting as a redundant factor for IL-7, since the low thymic cellularity and the deficit in Vγ5−γδT cell subsets, characteristic of Il7r-/- mice, were not corrected in Il15TgNIl7r-/- mice (data not shown). Remarkably, all Vγ5+ γδT cells found in Il15TgNIl7r-/- mice were CD8αα+ (Fig. 1c), whereas the majority of normal intrathymic γδTCR+ thymocytes from B6 mice and the few CD8-lineage αβTCR+ T cells found in Il7r-/- mice expressed the conventional CD8αβ heterodimer (Fig. 1c and data not shown). In addition, γδTCR+ thymocytes of Il15TgNIl7r-/- mice expressed relatively low amounts of the chemokine receptor CCR9 (data not shown), much like γδi-IELs24. As CD8αα+ Vγ5+ γδT cells are only found in the gut in normal mice and are specifically lost when IL-15R signaling is defective7,17,25, these results suggest that IL-15 is a major, if not the central, cytokine responsible for Vγ5+ γδ i-IEL development and/or expansion. Consistent with this interpretation, the loss of γδ i-IELs in IL-15-deficient mice predominantly involves the Vγ5+ subset7. Conversely, the frequency of Vγ5+ cells was increased among i-IELs of Il15TgN (IL-7R-sufficient) mice (Supplementary Fig. 1a). In addition, the effects of IL-15 were also evident in the thymus, as the frequency of Vγ5+ thymocytes was selectively and significantly (P<0.01) reduced in newborn Il15-/- mice as well as in mice lacking full-length (FL) STAT5 (Fig. 1d), a key intracellular signal transducer of IL-15 activated by JAK3 (ref. 25).
In support of the flow cytometry analysis, thymocytes from Il15TgNIl7r-/- mice did not appreciably express rearranged genes of the TcrCγ1 or TcrCγ4 loci other than Vγ5 (Fig. 1e, a schematic drawing of the ~60kb TcrCγ1 cluster depicted in Fig. 1f). Interestingly, while Il7r-/- thymocytes did not show expression of any rearranged Tcrg genes, as expected, transcripts corresponding to the rearranged Vγ5 gene, but not other Vγ genes, were detectable at low levels in intestinal intraepithelial cells of Il7r-/-, but not Jak3-/- mice (Fig. 1e). Since IL-7 and IL-15 are the only known JAK3-dependent cytokines to affect the frequency of Vγ5+ cells, these results suggest that the production of cells with a rearranged Vγ5 TCR gene can be regulated by IL-15, even when IL-7 function is absent.
How might IL-15 allow selective development of the Vγ5+ γδ i-IEL subset? One possibility was that IL-15 is a specific growth factor for Vγ5+ cells. In the absence of IL-7R function, high concentrations of IL-15 might be capable of expanding a few residual cells in the thymus that had rearranged the Vγ5 gene segment. However, in FTOC and suspension cultures of total i-IELs, IL-15 expanded all γδTCR+ cells, irrespective of TCR V gene usage (data not shown), ruling out the selective expansion of Vγ5+ cells by IL-15 as a likely explanation26,27. Alternatively, it was possible that IL-15, like IL-7, can regulate Tcrg locus chromatin accessibility, but that IL-15 specifically targets Vγ5 gene rearrangement. Data presented here demonstrate that IL-15 stimulates selective Vγ5 gene segment-associated histone acetylation in vivo, which correlates with enhanced chromatin accessibility conducive for V(D)J recombination28.
To determine whether IL-15 has a role in specifically directing Vγ5 gene rearrangement we first re-examined the Tcrg locus activity in the absence of IL-7 function. In Il7r-/- mice IL-15 signaling should operate in the thymus, and if IL-15 can affect Tcrg gene rearrangement, it was possible that Vγ5 gene rearrangements are present in Il7r-/- thymocytes, but perhaps not transcribed optimally. Vγ5-Jγ1 gene rearrangements were rare, but detectable, in thymocytes of adult Il7r-/- mice, whereas the normally dominant Vγ2 gene rearrangement was not (Fig. 2a). It has been previously shown that the Vγ1.1-Jγ4 gene rearrangements are not strictly dependent on IL-7 signals transmitted via the γc chain29, and this finding was corroborated in Il7r-/- and Jak3-/- thymocytes, where reduced, but detectable, Vγ1.1 gene rearrangements were observed. Notably, only the Vγ5 gene rearrangements were decreased (by approximately one-fifth) in the absence of IL-15 or FL-STAT5, the predominant signal transducer of the IL-15R, and Vγ5 rearrangements were absent in Jak3-/- cells (Fig. 2a). These results, in conjunction with the fact that IL-7 and IL-15 are the only known JAK3-dependent cytokines to affect the development of Vγ5+ cells, suggest that the residual Vγ5 gene rearrangements in Il7r-/- thymocytes are dependent on IL-15-JAK3 signals. The absence of rearranged Tcrg gene transcripts (other than Vγ5+ transcripts in the intestine, Fig. 1e) in Il7r-/- mice despite the presence of cells with rearranged Tcrg genes, and the corresponding lack of γδT cells, suggest that IL-7 is necessary for optimal transcription of rearranged Tcrg genes30, and/or the maintenance of these cells29.
To define potential IL-15-regulated molecular events contributing to the modulation of Vγ5 gene rearrangements, we examined the relative efficiency of recombination activating gene (RAG)-mediated double-strand signal sequence breaks at the Vγ5 and Vγ2 genes in adult thymocytes using ligation mediated-PCR (LM-PCR)31. This assay serves as an indirect measure of the frequency of cells undergoing active rearrangement at their Tcr loci. DNA breaks at recombination signal sequences (RSSs) were not observed in Rag1-/- thymocytes, as expected (Fig. 2b). In normal adult B6 thymocytes, Vγ2 and Vγ5 breaks were detectable, indicative of ongoing Tcrg locus gene rearrangements. Strikingly, Vγ5 breaks were substantially reduced, being barely detectable in Il15-/- thymocytes compared to control Il15+/+ thymocytes, whereas the amounts of Vγ2 breaks were unaltered. Hence, the noted amount of reduction in Vγ5-Jγ1 gene rearrangements (Fig. 2a) was an underestimate of the inefficiency of Vγ5 gene recombination in the absence of IL-15. In Il7r-/- thymocytes no Vγ2 breaks were seen, but rare Vγ5 breaks were detectable, supporting the earlier conclusion that some Vγ5 gene rearrangements are occurring in the absence of IL-7-mediated signals in vivo. In Il15TgNIl7r-/- thymocytes, Vγ5 breaks were enhanced when compared to Il7r-/- thymocytes, indicating that IL-15 was increasing the frequency of Vγ5 gene rearrangements. These results demonstrate that IL-15 concentrations correlate with the incidence of ongoing Vγ5 gene-specific DNA cleavage, strongly suggesting that IL-15 controls Vγ5+ thymocyte production by directly affecting the rearrangement process.
We next determined the histone3 (H3) acetylation status of Il15TgNIl7r-/- thymocytes. Rearrangement- and transcription-permissive chromatin is usually associated with acetylated histones 3 or 4 (AcH3 or AcH4)32. It has been shown previously by semi-quantitative chromatin immunoprecipitation (ChIP) assays that the TcrCγ1 locus in Il7r-/- thymocytes is relatively inaccessible16,33. This conclusion was based on the fact that the Jγ gene segment, the 3′ enhancer (ECγ1), as well as the locus control region-like element (DNaseI hypersensistive site A, HsA14,34), of the TcrCγ1 locus are hypoacetylated. The latter two cis-acting elements are redundantly required for transcription of rearranged Vγ2, Vγ4, and Vγ3 genes, but substantially less so for the Vγ5 gene14. The histone acetylation pattern at the Vγ genes has not been analyzed previously. In Il7r-/- thymocytes, AcH3 was nearly undetectable in all regions of the TcrCγ1 gene cluster, consistent with an inactive, inaccessible locus (Fig. 2c, middle). However, low amounts of AcH3 were detected exclusively at the Vγ5 and Jγ1 gene segments (Fig. 2c). In comparison, quantitative ChIP assay (Fig. 2d and illustrated by semi-quantitative gel assays, Fig. 2c) for AcH3 in Il15TgNIl7r-/- thymocytes showed that the Vγ5 (and Jγ1) gene segment chromatin was selectively hyperacetylated, while other regions of the Cγ1 locus remained hypoacetylated. In the negative control, the fetal-specific Vγ3 gene was not associated with detectable amounts of AcH3 in adult thymocytes, regardless of their genotype. These data suggest that IL-15 specifically regulates the local chromatin domain(s) modification or transcription in the vicinity of the Vγ5 gene. The lack of detectable acetylated histones at ECγ1 and HsA in Il15TgNIl7r-/- thymocytes, correlative to an inaccessible chromatin state, further solidifies previous findings14 that the rearrangement of the Vγ5 gene is largely independent of these regulatory sequences (see below).
As alluded to above, γδ thymocytes of the 129 strain contain a ~3 fold higher frequency of Vγ5+ cells in comparison to B6 mice (14 and data not shown). We examined whether this difference in Vγ gene usage was independent of a TCR driven cellular selection process by performing a ChIP assay for acetylated histones in thymocytes of 129Rag1-/- and B6Rag1-/- mice. Owing to the absence of TCRs on the cells of Rag1-/- mice, no thymic cellular selection can take place, and T cell development is blocked at the precursor stages, providing a relatively uniform population of cells with their TCR loci poised for the actions of the V(D)J recombinase. While the amounts of AcH3 associated with the Vγ2 gene segment were similar between these two precursor thymocyte preparations, a higher amount (~3 fold) of AcH3 was found in the Vγ5 locus of 129Rag1-/- precursor thymocytes (Fig. 2e). This result strongly indicates that the strain-dependent enhancement of Vγ5 usage in thymocytes is correlated at the level of chromatin accessibility, prior to Tcrg gene rearrangement.
To more strictly demonstrate a link between IL-15 and the TCR V gene-specific epigenetic modifications, we next performed the sterile transcription and ChIP assays using immature lymphoid lineage cells from Rag1-/- mice expressing varying levels of IL-15. The amount of germline sterile transcripts emanating from the unrearranged TcrVγ gene segments is an indicator of relative gene activity, and is correlative and predicative of the developmentally-regulated Tcrg gene rearrangement pattern35. In adult Rag1-/- thymocytes, unrearranged Vγ2 and Vγ5 gene segment-specific transcripts were detectable, whereas Vγ3 gene-specific sterile transcripts were absent, reflecting the developmental stage-specific rearrangement pattern of these V genes (Fig. 3a). In addition, the Vγ2 sterile transcripts were relatively more abundant than those of Vγ5 in adult Rag1-/- thymocytes, correlating with the TCR repertoire of γδ thymocytes in normal B6 mice (among adult γδ thymocytes ~50% are Vγ2+, while ~5% express Vγ5). Significantly, while the Vγ2 (and Vγ1.1 or Vγ4, data not shown) gene segment-specific transcript abundance was not altered by varying concentrations of IL-15, those of Vγ5 were decreased by ~one-fifth when IL-15 was absent, and increased dramatically (>10 fold) when IL-15 was over-expressed.
The histone acetylation pattern correlated strongly with the sterile transcription assay. Unlike the situation in Il7r-/- mice, thymocytes of Rag1-/- mice displayed a histone acetylation pattern consistent with the adult Vγ gene usage: ECγ1 and HsA were active, as shown by the relatively high amounts of co-localized AcH3 (Fig. 3b and quantitative real-time PCR in Fig. 3c). In addition, among the Vγ gene segments, the Vγ2 gene was associated with the highest amounts of acetylated histones, whereas the fetal-specific Vγ3 gene was hypoacetylated, consistent with published results36. The Vγ5 gene had low, but significant, levels of acetylated histones bound to it (Fig. 3b), again correlating with the low frequency of Vγ5+ γδ thymocytes in normal adult B6 mice. Similar results have been interpreted to support the chromatin accessibility model of developmental regulation of V(D)J recombination at the Tcrg locus36. Importantly, in the absence of IL-15, consistently less acetylated histones were found on Vγ5 gene segments (Fig. 3b, c). This pattern was observed in all experiments using thymocytes and intestinal intraepithelial cells (much more pronounced in the latter, see below), and was consistent with results from the sterile transcription assay (Fig. 3a). However, the difference was difficult to quantify due to the low basal amount of acetylated histones detectable at the Vγ5 gene of normal thymic precursors using this assay. On the other hand, in thymocytes of Il15TgNIl7r-/- mice, greatly increased amounts of AcH3 were detected at the Vγ5 gene, again corroborating the sterile transcription pattern. The H3 acetylation profile in other regions of the Cγ1 locus was not altered by varying levels of IL-15. In conjunction with the observation that mice lacking IL-15 function have a pronounced deficit in Vγ5+ i-IELs7 (Fig. 1d), these results clearly demonstrate that IL-15 controls the generation of Vγ5+ cells by modulating histone acetylation, and thereby chromatin accessibility, at the Vγ5 locus prior to V(D)J recombination.
To verify that the histone acetylation patterns observed in thymocytes from Il7r-/- and Rag1-/- mice are also found in normal thymocytes, we sorted the triple negative thymic precursor cell subset (CD3−CD4−CD8−, TN) from B6 mice and repeated the ChIP assay. Increased expression of IL-15 led to an enhanced association of AcH3 specifically with the Vγ5 gene segment in normal precursor cells (Fig. 3d). Collectively, these results demonstrate that the basis for IL-15-mediated generation of Vγ5+ γδT cells is the selective enhancement of Vγ5 gene segment activity prior to the RAG-mediated Tcrg gene rearrangement.
Since the origin of i-IELs remains an unresolved issue, it was important to address the effects of IL-15 on immature lymphoid cells of the gut. As physiological IL-15 concentration is thought to be high in the intestine21,37,38, we predicted that the selective Vγ5 gene accessibility should be much more evident in lymphoid precursors of the intestine if these cells were subjected to a similar regulation of Tcrg gene usage as in thymocytes. In Rag1-/- intestinal intraepithelial cells, sterile transcription of the unrearranged Vγ5 gene segment was relatively prominent compared to the Vγ2 gene, whereas the reverse was the case in the thymus (Fig. 4a). Sterile transcription of the Vγ1.1 gene, utilized by both the splenic and intestinal γδT cells, was not substantially different in thymocytes and intestinal cells of Rag1-/- mice. It should be noted that the amounts of Vγ5 sterile transcript in the intestinal intraepithelial cells were very high in comparison to Rag1-/- thymocytes. In all experiments, the putative lymphoid lineage cells (identified by the expression of IL-7R and/or CD4, and only sample sets that contained similar proportions of lymphoid lineage cells were compared) among the Rag1-/- intestinal intraepithelial cell preparations constituted less than 15% of the total, making it highly probable that the difference in Vγ5 sterile transcript abundance between the intestine and thymus is an underestimate. Hence, the relative Vγ gene segment activity before V(D)J gene rearrangement correlates well with the observed tissue-specific Vγ repertoire, suggesting a spatial control of Tcrg gene rearrangement and/or expression.
We next examined the histone acetylation pattern of the Vγ gene segments of the TcrCγ1 locus in intestinal intraepithelial cells. In Rag1-/- intestinal intraepithelial cell preparations, AcH3 was found selectively at the Vγ5 gene segment (Fig. 4b). This result is consistent with the sterile transcription assay showing a biased Vγ5 gene activity (Fig. 4a), and it strongly supports the existence of a distinct pre-rearrangement Vγ segment accessibility in favor of the Vγ5 gene in immature intestinal intraepithelial cells compared to thymocyte precursors (Fig. 3a). Critically, this basal Vγ5 gene specific H3 acetylation was dependent on IL-15, as evidenced by the diminution of AcH3 detected in Il15-/-Rag1-/- intraepithelial cell preparations (Fig. 4b, c). Conversely, an increased level of IL-15 in the intestine of Rag1-/- (Fig. 4b, c) and Il7r-/- mice (Fig. 4d, e) resulted in greatly enhanced amounts of AcH3 specifically at the Vγ5 gene segment. These results demonstrate that IL-15 specifically induces local histone modifications consistent with enhanced chromatin accessibility at the Vγ5 gene segment in immature intestinal intraepithelial cells. Taken together, these data indicate that IL-15 establishes permissive Vγ5 gene segment accessibility for directed gene rearrangement in both the thymus and intestine. Whether sufficient functional recombinase activity exists in the putative intestinal lymphoid precursors of normal mice is controversial 39,40, but irrespective of anatomical sites, our results show that IL-15 programs targeted enhancement in Vγ5 gene accessibility conducive for selective gene rearrangement.
To determine whether the intracellular components of the IL-15 signaling pathway can directly modulate Tcrg V gene segment chromatin accessibility, we tested the contribution of STAT5, a central transcription factor activated by IL-7 and IL-15 signaling that has been proposed to be the key factor controlling general Tcrg locus accessibility41. Although adult mice that cannot express FL-STAT5 (hypomorphic Stat5-/- mice42,43) do not exhibit pronounced defects in Tcrg gene rearrangement, peripheral γδT cells are severely depleted in these mice25. Similar to defects observed in lymphoid lineage cells of Il15-/- mice, Fl-Stat5-/- mice have greatly reduced numbers of γδ i-IELs25, while neonatal Fl-Stat5-/- mice have a specific reduction in Vγ5+ γδ thymocytes (Fig. 1d), raising the possibility that IL-15-activated STAT5 is responsible for the induction of Vγ5 gene rearrangement. A unique function of STAT5 for IL-15 signaling in Tcrg gene regulation is further suggested by the presence of a STAT5 consensus binding site in the Vγ5 gene promoter. Other Vγ gene promoters of the TcrCγ1 locus do not possess this sequence motif (data not shown). Our ChIP analysis of ex vivo Rag1-/- thymocytes showed that STAT5 binds to the consensus site in the Vγ5 gene promoter, and that the level of association was substantially increased by IL15TgN expression (Supplementary Fig. 1b). In addition, a specific reduction in Vγ5 gene rearrangement was observed in Fl-Stat5-/- thymocytes (Fig. 2a), consistent with the role of STAT5 in Vγ5 gene rearrangement. Lastly, among the lymphocyte subsets regulated by IL-15, γδ i-IELs are the most dependent on FL-STAT5: While the paucity of NK and activated (CD44+) CD8+ T cells in Fl-Stat5-/- (and in IL15-/-) mice is corrected in Il15TgNFl-Stat5-/- mice, strikingly, γδ i-IEL defects persist (unpublished results).
To directly test the proposal that STAT5 controls Vγ5 gene segment accessibility, we generated Il15TgNFl-Stat5-/-Rag1-/- mice and examined histone modification at the Tcrg locus. Conspicuously, the IL-15-dependent hyperacetylation at the Vγ5 gene segment in STAT5-sufficient cells was completely abrogated when FL-STAT5 was absent (Fig. 5a, comparing the amounts of AcH3 in the first and last sample sets). The quantitative real-time ChIP data for AcH3 (Fig. 5b) and AcH4 (data not shown) showed the same trend. In comparison, the amounts of AcH3 and AcH4 associated with the Vγ2 gene segment in adult immature thymocytes were not dependent on FL-STAT5. The effects of FL-STAT5 were difficult to discern in precursor thymocytes of Rag1-/- mice since the basal amount of AcH3 at the Vγ5 gene was very low and the assay may not be sufficiently sensitive to discriminate potential changes. However, in Rag1-/- intestinal intraepithelial cells, the basal AcH3 detectable at the Vγ5 gene is relatively high and the absence of FL-STAT5 led to a reproducible decrease of AcH3 specifically at the Vγ5 gene, showing that the endogenous IL-15-mediated epigenetic modification of the Tcrg locus is also dependent on FL-STAT5 (Fig. 5c). Taken together, these results demonstrate a distinct requirement for FL-STAT5 in IL-7 versus IL-15-mediated signals. While IL-7-activated FL-STAT5 is not absolutely essential to the overall Tcrg locus regulation, including chromatin modification, gene rearrangement and transcription in adult γδT cell development25, FL-STAT5 activated in the context of IL-15 signaling is indispensable for induction of local TcrVγ5 gene specific chromatin modification.
Despite considerable efforts, events leading to the establishment of restricted TCR gene usage among i-IELs in the mucosal epithelia are largely unknown. This is an important issue since it has implications for deducing the nature of γδTCR-ligand interactions. Here, we demonstrate that IL-15 programs targeted Vγ5 gene rearrangement in immature precursor cells by modulation of Tcrg subdomain chromatin accessibility. This molecular connection between IL-15 and Tcrg repertoire formation provides one explanation for the unusual dependence of Vγ5+ TCR expressing cells on IL-15, as evidenced by the selective loss of Vγ5+ i-IELs in IL-15 signaling-defective mice7. It should be stressed that the primary molecular programming of Tcrg gene rearrangement by IL-15 does not necessarily preclude a subsequent additional checkpoint, perhaps governed by cell surface TCR expression, to generate a stable, functional γδ IEL repertoire. A precedent for a cellular selection process acting on a uniform population of fetal Vγ3+ γδ thymocytes generated by targeted TCR gene rearrangement has been recently described44. However, the major ligand(s) for Vγ5+ γδ i-IELs is currently unknown, and the nature of potential secondary cellular selection remains speculative.
This study does not directly address the anatomical origin of γδ IELs. Recent findings have indicated the exclusive origin αβTCR+ i-IELs from CD4+CD8+ thymocytes in euthymic mice4, but the situation remains murky for γδ i-IELs. The developmental requirements for γδ i-IELs are distinct from those of αβ i-IELs4,10 and suggestive evidence for the extrathymic development of γδ i-IELs exists40,45. However, in unperturbed normal mice, robust RAG1 and RAG2 expression is difficult to detect in the gut39. Consistent with this finding, we have not been able to detect RAG-mediated DNA breaks at Vγ gene segments among i-IELs of normal or Il15TgN mice using LM-PCR assays (data not shown), whereas they are easily detectable in thymocytes. Accordingly, IL-15-driven intrathymic generation of CD8αα+Vγ5+ cells with gut-homing properties, perhaps from a distinct precursor subset than for the peripheral γδT cells, may be the primary source of γδ i-IELs that are then expanded and maintained in the gut by the combination of IL-15 and IL-7. This scenario is reminiscent of the Vγ3+ γδ dendritic epidermal T cells that arise in the thymus via a programmed gene rearrangement exclusively during the fetal stage, which are positively selected in the thymus to acquire the skin homing ability46. Hence, despite the fact that immature lymphoid cells in both the thymus and intestine are governed similarly by IL-15 to become preferentially poised to generate Vγ5-expressing cells, as assessed by histone acetylation patterns, and γδ i-IEL development is less compromised than αβ i-IELs in athymic mice39,40, the apparent paucity of RAG proteins in the gut of euthymic mice may prohibit substantial generation of γδ i-IELs in the intestine. Caveats to this interpretation are that there might be as yet unidentified intestinal niches where active Tcrg and Tcrd recombination does occur, but currently employed experimental assays are insensitive to detect such events and/or RAG1 and RAG2 are transiently expressed in the intestine in certain conditions, such as in neonatal stages or during acute infection of the gut to permit extrathymic γδ T cell development.
Although our data indicate that IL-15 was sufficient to program Vγ5 gene rearrangement intrathymically, the absence of Vγ5+ (or Vγ1.1+) γδT cells in Il7r-/- mice, despite the evidence for gene rearrangements involving these V genes, indicate that IL-7 is necessary for full maturation of γδi-IELs in normal physiological condition. It has been shown that IL-7 is required for normal transcription of rearranged Tcrg genes30,47. The 3′ECγ1 and the HsA, although not essential, are cooperatively required for optimal transcription of the rearranged Vγ5 genes in thymocytes14. The greatly impeded accessibility of these two regulatory elements in Il7r-/- cells33 may lead to suboptimal expression of the TCR and potentially contributes to the loss of cells that had rearranged the Vγ5 gene. In the presence of high amounts of IL-15 and STAT5, Vγ5 gene expression appears less dependent on the regulatory elements and Vγ5+ cells are more likely to develop even when IL-7 activity is absent. This interpretation is consistent with the detectable expression of the rearranged Vγ5 gene in intestinal intraepithelial cells of Il7r-/- mice, which are presumably exposed to high levels of IL-15, but not in Il7r-/- thymocytes. Definitively establishing the quantitative aspect of complex IL-15 signaling7 in distinct tissues in vivo will require the development of reliable reagents to detect IL-15 and the IL-15Rα chain. It is also possible that IL-7 signaling is required for the maintenance and survival of i-IELs and that in its absence, some cells with productive Vγ5 gene rearrangements directed by IL-15 may arise, but these cells cannot persist. Although mature i-IELs, unlike resting peripheral T cells, do not express high amounts of IL-7R, the requirement for IL-7 during certain periods of i-IEL maintenance cannot be ruled out.
Both IL-7 and IL-15 predominantly activate JAK3-STAT5, but the unique dependence of IL-15 on FL-STAT5 (ref. 25) further accentuates the differing signaling properties of these two γc cytokines. Lymphocytes of Fl-Stat5-/- mice still express partly functional NH2-terminal truncated proteins (H. N. and J. K., unpublished results) and these STAT5 isoforms, which are also found in normal cells, are sufficient to ensure normal intrathymic T cell development regulated by IL-7, but they are insufficient for IL-15 signaling25,43. The N-terminal region of STAT5 has been implicated in both positive and negative regulatory functions of nuclear STAT5, including nuclear retention, protein dephosphorylation, oligomerization, and co-factor recruitment48. Among these properties, STAT5 tetramerization has emerged as a central feature of optimal STAT5 activity, especially critical for stimulating the expression of genes with suboptimal STAT5 consensus binding motifs49. The underlying basis for the relatively normal IL-7-regulated adult intrathymic T cell development mediated by N-terminal truncated STAT5 protein versus aberrant development of lymphoid lineages controlled by IL-15 in Fl-Stat5-/- mice is likely to entail cell-type specific differences in both the quantitative and qualitative aspects of IL-7 and IL-15-mediated signal transduction.
A highly related issue to the unique signaling properties of IL-7 and IL-15 is why the increased IL-15 signaling via STAT5 is incapable of establishing an accessible chromatin across the entire Tcrg locus since it has been shown that a constitutively active form of STAT5 (CA-STAT5) can bypass the requirement for IL-7R signaling in TCRγ gene rearrangement in FTOC16,41. One possibility is that the Tcrg locus of the cells capable of responding to IL-15 is irreversibly maintained in a relatively inaccessible chromatin configuration. An attendant prediction of this model is that all i-IELs with Tcrg gene rearrangements involving the Cγ1 cluster on both chromosomes exclusively exhibit Vγ5 gene rearrangements. However, analysis of Tcrg gene rearrangement pattern in Vγ5+ γδ thymocyte clones does not support this model: In four Vγ5+ T cell clones with gene rearrangements in the Cγ1 locus in both alleles, the non-productive rearrangement involves Vγ2, indicating that Vγ5+ cells are not inherently restricted to rearranging Vγ5 genes50. An alternative explanation is that IL-7 activates an unknown cofactor(s) that cooperate with STAT5 in making the locus globally accessible and/or IL-15 activates factors that inhibit accessibility of other Vγ genes and/or preferentially target active STAT5 to the Vγ5 gene segment. Implicit in this scheme is that cells triggered by IL-15 do not receive sufficient IL-7 signals and/or the outcome of IL-7 signaling is qualitatively different. Indirect evidence indicating that the history of γc cytokine exposure can alter a cell's response to other γc cytokines exists51,52. Our study clearly underscores exquisitely distinct functions of IL-7 and IL-15 on the Tcrg locus, which cannot be accounted for simply by activation of STAT5 alone. We have reported previously that FL-STAT5's function in T cell development is developmentally regulated, being more critical in fetal stages25. The reported ability of CA-STAT5 to restore γδT cell development in the absence of IL-7R41 may also be restricted to the fetal stage (or FTOC) since in adult CA-Stat5TgNIl7r-/- mice, γδT cell development is not restored to the numbers observed in WT mice (M. Farrar, Univ. of Minnesota, personal communication).
The γc cytokines IL-7 and IL-4 can globally and regionally permit increased chromatin accessibility for Tcrg and Igh chain gene rearrangement and immunoglobulin class switching, respectively. The results from this study establish another level of regulation of chromatin accessibility by a cytokine in vivo that displays an exquisite selectivity: STAT5-dependent IL-15 regulated V gene segment-specific chromatin modification leading to the generation of a tissue-restricted γδTCR repertoire. Hence, cellular selection is unlikely to be the initial driving force behind the formation of the gut-restricted γδTCR repertoire. To date, no firm evidence for the requirement of MHC molecules in the overall γδTCR repertoire formation has been obtained10,12 and the cytokine-directed TCR repertoire shaping most likely represents a substitute mechanism for the MHC selection processes. Whether development of other lymphocyte subsets with a restricted TCR repertoire, such as NKT53 and mucosal-associated invariant T cells54, is similarly influenced by cytokines will require further investigation.
B6Il7r-/-, Fl-Stat5−/−42, B6Il15TgN23 and B6Il15−/−17 mice have been described. Fl-Stat5−/− mice were backcrossed to B6 mice five times. 129Rag1-/- mice were purchased from Taconic. Routinely, 4-6 weeks old mice were used for experiments. For fetuses, the day of vaginal plug was designated day 0. All mice used in this report were housed in a pathogen free rodent barrier facility. All animal experiments were approved by the University of Massachusetts Medical School Institutional Animal Care and Use Committee (Worcester, MA).
i-IELs, intestinal intraepithelial cell preparations and nylon wool column-passed splenic cell preparations enriched for T and NK lymphocytes were obtained using published protocols25. For comparison of intestinal intraepithelial cell subsets from Rag1-/- genetic background, preparations containing similar proportions (routinely 12-15% of the total) of lymphoid progenitors (IL-7Rα+ and/or CD4+) were used. For FTOC, Il7r-/- and littermate control embryonic day (E) 14 or E15 thymic lobes were cultured for 5-12 days on transwell plates (Corning). Standard culture media (RPMI 1640 with 10% FCS, 50μM 2-mercaptoethanol, 2mM L-glutamine, 20mM HEPES and antibiotics) supplemented with recombinant IL-15 (10-20ng/ml) or IL-7 (20ng/ml) (R&D) was used. In some experiments supernatants from Il7 or Il2 transfected cell lines were used at 2% v / v. After culture, single cell preparations were analyzed by FACS.
Anti-TCR Vγ5 mAb GL15 was generously provided by L Puddington (UConn, Farmington, CT) and anti-mouse CCR9 Ab24 was from Dr. P. Love (NIH, Bethesda, MD). The following Abs were purchased from EBiosciences or BDPharmingen: Abs specific for CD4 (Cy5 conjugated), CD8α (Cy5), CD8β (biotin), CD3ε (Cy5), TCRβ (fluorescein isothiocyanate, FITC), TCRγδ (biotin), Vγ2TCR (biotin and FITC), Vγ3TCR (FITC), CD122 (biotin), CD127 (biotin), and NKRP1.1 (FITC). Strepavidin-phycoerythrin was purchased from BDPharmingen. Flow cytometric analyses were performed on an EPICS XL cytometer (BD-Coulter). Data were analyzed using FlowJo software (Tree Star). Cells were sorted using MoFlo (Cytomation).
The ChIP assay was carried out according to manufacturer's instructions (Upstate Biotechnology) with minor modifications. Briefly, 1-2 × 106 cells (each pooled from 3-5 mice) were fixed with 1% formaldehyde. Cells were washed in ice-cold PBS containing 1 μg/ml pepstatin A (Sigma) and a complete protease inhibitor cocktail (Roche). Cell pellets were resuspended in SDS lysis buffer supplemented with the protease inhibitor mixture for 10 min on ice. Samples were sonicated to an average DNA length to 0.1-0.6 kb. Cellular debris was removed by centrifugation for 10 min at 16,000g at 4 °C. The supernatant was diluted 10-fold in buffer supplemented with the protease inhibitor mixture. An aliquot of the diluted supernatant was used as “input” DNA control after reverse cross-linking. The cell supernatant was pre-cleared with salmon sperm DNA/protein A agarose for 30 min at 4 °C with agitation. Immunoprecipitation was carried out (overnight at 4°C on a rotator) with a 1:200 dilution of the anti-AcH3, AcH4 (Upstate) or STAT5 Abs (Santa Cruz) along with no Ab controls. Immune complexes were mixed with salmon sperm DNA/protein A agarose followed by incubation for 1 h at 4 °C on a rotator. After extensive washing, the complexes were eluted by adding a 250-μl aliquot of a freshly prepared 1% SDS, 0.1 M NaHCO3 solution. The sample was briefly vortexed and incubated at room temperature for 15 min on a rotator. After microcentrifugation, the protein-DNA complex was washed with a second aliquot, and the resulting supernatant was pooled with the first aliquot. The protein-DNA cross-link was reversed at 65 °C for 4 h. Samples were digested with 2 μl of proteinase K (10 mg/ml) at 45 °C for 1 h, and the DNA was recovered by phenol and chloroform extractions. The precipitated DNA was dissolved in 20 μl of Tris-EDTA pH 7.4 buffer (designated the “bound DNA fraction”) and subjected to PCR or real-time PCR analysis. PCR was carried out for 28-35 cycles, and the products were visualized by ethidium bromide staining. The sequences of the primer pairs used are listed in Supplemental Table 1. SYBR green PCR core reagents (Applied Biosystems) were used for the real time PCR assay using the BioRad iCycler. The amount of the immunoprecipitated “bound” DNA was normalized against input DNA.
PCR and RT-PCR assays to detect Tcrg gene rearrangements and transcripts were performed as described14,34,35 using primers published and listed in Supplementary Table 1. LM-PCR to detect double-strand signal sequence breaks 3′ of the Vγ5 and Vγ2 gene segments were carried out as described31,44 using primers listed in Supplementary Table 1. Semi-quantitative radioactive PCR (40 or 41 cycles) was used to detect PCR products, using a 5% polyacrylamide gel. Genomic DNA was extracted using Wizard genomic DNA purification kit (Promega). RNAs were isolated using a micro-to-midi total RNA purification Kit (Invitrogen) or Trizol. RNA was treated with DNase I to digest contaminating genomic DNA. RNA samples were reverse transcribed (RT) with Superscript II (RNase H—) reverse transcriptase using oligo-dT(12-18) primers (Invitrogen).
We thank D. H. Raulet for initial support of this work; C. Chambers for discussions; M. Caligiuri, J. Ihle and J. Peschon for Il15TgN, Fl-Stat5-/- and Il15-/- mice, respectively; D. H. Raulet, W. Held, L. Berg and members of the lab for critical comments on the manuscript; and K. Pinault and T. Mascenik for assistance with animal husbandry.