Soluble γc cytokine receptor suppresses IL-15 signaling and impairs iNKT cell development in the thymus

The soluble γc protein (sγc) is a naturally occurring splice isoform of the γc cytokine receptor that is produced by activated T cells and inhibits γc cytokine signaling. Here we show that sγc expression is also highly upregulated in immature CD4+CD8+ thymocytes but then downregulated in mature thymocytes. These results indicate a developmentally controlled mechanism for sγc expression and suggest a potential role for sγc in regulating T cell development in the thymus. Indeed, sγc overexpression resulted in significantly reduced thymocyte numbers and diminished expansion of immature thymocytes, concordant to its role in suppressing signaling by IL-7, a critical γc cytokine in early thymopoiesis. Notably, sγc overexpression also impaired generation of iNKT cells, resulting in reduced iNKT cell percentages and numbers in the thymus. iNKT cell development requires IL-15, and we found that sγc interfered with IL-15 signaling to suppress iNKT cell generation in the thymus. Thus, sγc represents a new mechanism to control cytokine availability during T cell development that constrains mature T cell production and specifically iNKT cell generation in the thymus.

γ c is expressed on immature CD4, CD8 double-negative (DN) thymocytes for survival and proliferation, and γ c expression is also upregulated upon positive selection to mediate lineage choice and effector cell differentiation 19 . Importantly, γ c expression is downregulated on immature DP cells, presumably to suppress aberrant γ c cytokine signaling that could provide pro-survival effects on pre-selection thymocytes 19 . However, the molecular pathway that suppresses γ c expression on DP cells remains still veiled.
We have previously identified alternative splicing of γ c pre-mRNA as a new mechanism to reduce surface γ c protein expression 24 . The γ c gene is encoded in 8 exons, and exon 6 encodes the entire transmembrane domain 25 . While the full-length γ c protein is a transmembrane protein, the new splice isoform lacks exon 6 and thus the transmembrane region, making it a soluble secreted protein. Because the soluble form of γ c (sγ c) is generated at the expense of membrane γ c protein expression, sγ c expression inversely correlates with the amount of surface γc expression. Therefore, sγ c expression represents a novel mechanism to suppress γ c expression on cell surface.
In the present study, we now identify DP thymocytes as a major source of sγ c and we propose that alternative splicing into sγ c could promote establishing the low level of surface γ c on pre-selection DP thymocytes. Moreover, because sγ c proteins suppress signaling of γ c cytokines, such as IL-2 and IL-7 24 , sγ c production by DP thymocytes would create an overall suppressive milieu for γ c cytokine signaling in the thymus. In fact, we found that sγ c overexpression resulted in significantly diminished percentages and numbers of thymic iNKT cells, which are critically dependent on IL-15 signaling for their development and differentiation 6,7,20,21 . Specifically, increased sγ c expression resulted in the loss of HSA lo mature iNKT cells, as it interfered with upregulation of anti-apoptotic Bcl-2 expression and induced increased cell death. Collectively, these data demonstrate a previously unappreciated role for sγ c in downregulating surface γ c expression and also in dampening γ c cytokine signaling in thymocytes, which can inhibit the generation and differentiation of specific T cell subsets in the thymus.

Results
γc family cytokine receptor expression on thymocytes. Surface staining for γ c family cytokine receptors revealed distinct and stage-specific expression of individual cytokine receptors (Fig. 1). Most γ c family cytokine receptors were found on both CD4 and CD8 single positive (SP) thymocytes but absent on immature DP thymocytes. IL-4Rα , IL-21R and γ c differed as they were also expressed on DP cells. Consequently, DP thymocytes would be unable to respond to IL-7, but they are equipped with IL-4 and IL-21 responsiveness. Importantly, while DP cells did express γ c, the amount of surface γ c was markedly lower compared to that on immature DN or mature SP thymocytes (Fig. 1). These results indicated and confirmed that γ c expression is a developmentally controlled event that is specifically suppressed on pre-selection DP cells 19 . Reduced γ c expression presumably helps avoiding signaling by pro-survival γ c cytokines which could interfere with TCR-induced positive selection as previously suggested 16 .
To correlate γ c expression with positive selection, next, we analyzed surface expression of γ c and IL-7Rα on HSA hi TCRβ lo pre-selection (gate I) and HSA lo TCRβ hi post-selection thymocytes (gate II) (Fig. 2a). Expression of IL-7Rα and γ c was low on gate I immature DP thymocytes but upregulated on gate II mature SP cells, which illustrated developmental control of cytokine receptor expression in thymocytes. The molecular mechanism that downregulates γ c expression on DP thymocytes is not known. However, we previously reported a post-transcriptional mechanism that can downregulate surface γ c expression 24 . Specifically, we found that alternative splicing of γ c transcripts produced a soluble form of γ c (sγ c) that was generated at the expense of membrane γ c (mγ c) protein expression 24 . Thus, increase in sγ c expression conversely results in reduced surface γ c expression. Interestingly, here we found that DP thymocytes expressed markedly higher levels of sγ c transcripts than mature SP thymocytes ( Fig. 2b right), and that increased sγ c expression inversely correlated with decreased mγ c protein expression in the same cells (Fig. 2b). These data suggest that alternative splicing of γ c mRNA might contribute to downregulation of surface γ c expression on DP thymocytes. Moreover, DP cells comprise up to 90% of total thymocytes so that they are a major source of sγ c proteins, and thus render the thymus into an sγ c-rich environment. However, if sγ c plays a role in thymocyte differentiation is not known. sγc overexpression impairs thymocyte development. To interrogate sγ c's effect on T cell development, we analyzed thymocytes in sγ c transgenic mice (sγ cTg) 24 . To generate sγ cTg mice, a murine sγ c cDNA was placed under the control of a human CD2 mini-cassette so that sγ c is overexpressed in all T lineage cells. Increased sγ c expression significantly reduced total thymocyte numbers, and we observed an inverse correlation of sγ c expression and total thymocyte numbers in WT, sγ c medium (M) and sγ c high (H) expresser transgenes (Fig. 3a). All further experiments in this study were done with the sγ c high expresser line. Assessing thymocyte profiles of sγ cTg mice did not reveal any significant changes in TCRβ hi mature T cell generation (Fig. 3b) or in CD4/CD8 lineage commitment (Fig. 3c left). However, we did find a significant increase in DN cell frequency ( Fig. 3c right), suggesting a developmental defect in DN to DP cell transition, which would also explain the reduction in thymocyte numbers in sγ cTg mice (Fig. 3d) 26 .
To directly address this point, we examined surface CD44 and CD25 expression in lineage marker negative DN thymocytes and determined DN1-DN4 differentiation in sγ cTg and WT thymocytes (Fig. 4a) 27 . Contrary to our expectation, however, we did not find any significant differences in DN1-4 subset frequencies between WT and sγ cTg mice. We also did not find any significant difference in Ki-67 expression in individual DN subsets (Fig. 4b), suggesting that the proliferative potential of sγ cTg DN cells did not differ from WT thymocytes. Finally, to examine the possibility that increased cell death of DN thymocytes would account for reduced cell numbers, we assessed caspase-3 activity and intracellular Bcl-2 contents in sγ cTg DN thymocytes (Fig. 4c). Decreased Bcl-2 expression is associated with increased susceptibility to apoptosis, and elevated caspase-3 activity is indicative of increased cell death 28,29 . However, we did not find any differences in their expression either between sγ cTg and WT DN thymocytes (Fig. 4c).
Scientific RepoRts | 6:36962 | DOI: 10.1038/srep36962 Notably, DP thymocytes in sγ cTg mice had been previously reported to contain increased percentages of CD25-positive cells 24 . Also, surface CD25 expression is diluted during the proliferative burst of DN to DP transition 30 . Thus, these results collectively suggested that reduced thymocyte numbers and increased DN cell percentages are results of reduced cell proliferation during DN to DP cell transition and not due to a developmental arrest at DN2/DN3 stage of T cell development.

Thymic development of γδ T cells and Foxp3 + Treg cells in sγcTg mice.
To further assess the impact of increased sγ c expression, next, we analyzed generation of individual thymic T cell subsets. We first assessed γ δ T cell generation in the thymus and found it unaffected in sγ cTg mice. Thymic γ δ T cell numbers did not differ between WT and sγ cTg mice, and because overall thymocyte numbers were decreased in sγ cTg mice, this translated into increased percentages of γ δ T cells in the thymus (Fig. 5a). Next, we examined generation of Foxp3 + T regulatory (Treg) cells in sγ cTg thymocytes, and found a significant decrease in Foxp3 + CD25 + CD4SP Treg cell numbers (Fig. 5b). However, we did not find a decrease in Foxp3 + CD25 + cell percentages among CD4SP thymocytes (Fig. 5b), which indicated that reduced Foxp3 + Treg cell number is due to an overall impairment in thymopoiesis and not because of a specific defect in thymic Treg cell generation. sγc overexpression impairs iNKT cell generation. iNKT cells are thymus-generated innate T lineage cells that depend on IL-15 for their development and differentiation 7,31,32 . iNKT cells can be identified by their TCR reactivity to lipid-loaded CD1d tetramers (CD1dTet) 33 , and here we found that both frequency and number of CD1dTet + iNKT cells were significantly reduced in sγ cTg thymocytes (Fig. 6a). Conventionally, iNKT cell development had been understood based on cell surface HSA (CD24), CD44 and NK1.1 expression [34][35][36] . The most immature CD1dTet + iNKT cells express high levels of HSA and are defined as stage 0 iNKT cells. Upon further maturation, iNKT cells lose HSA expression but start expressing CD44 and then NK1.1, so that CD44 -NK1.1 − cells are stage 1, CD44 + NK1.1 − cells are stage 2, and CD44 + NK1.1 + cells are referred to as stage 3 iNKT cells 35 . Assessing WT and sγ cTg thymic iNKT cells revealed no significant differences between WT and sγ cTg mice when comparing in individual stages ( Fig. 6b-d). However, there was a significant loss of sγ cTg iNKT cells when comparing the combined frequency of mature iNKT cells, i.e. stage 1-3 (Fig. 6c). Because the frequency of immature stage 0 iNKT cells did not differ between sγ cTg and WT control mice, these results suggest that sγc overexpression did not target a specific developmental stage but rather induces an overall reduction of thymic iNKT cells.
iNKT cells can be also categorized into discrete subsets based on their function and transcription factor expression 37 . PLZF lo T-bet + cells correspond to IFNγ -producing NKT1, PLZF hi RORγ t − cells are IL-4-producing NKT2, and PLZF int RORγ t + are IL-17-producing NKT17 cells 38 . In C57BL/6 (B6) WT mice, the majority of thymic iNKT cells are NKT1 cells with only few NKT2 and NKT17 cells. Such iNKT cell distribution is not developmentally fixed, and changes with mouse strains as illustrated by significantly increased NKT2 and NKT17 cell percentages in BALB/c mice (Fig. 6e) 37 . We found that sγ cTg mice, which were maintained on a B6 background, showed identical distribution of NKT subsets to control WT B6 cells (Fig. 6e). Additionally, when dividing iNKT cells into two major subsets of CD4 + and DN iNKT cells 32 , we also did not find any difference between sγ cTg and WT mice (Fig. 6f). Collectively, these results demonstrate that sγ c overexpression is detrimental for thymic iNKT cell generation, and that sγ c affected iNKT cell frequency and number without targeting a specific iNKT subset or specific developmental stage.
sγc interferes with IL-15 signaling in iNKT cells. To further understand the molecular basis of iNKT cell loss in sγ cTg mice, next we examined whether increased sγ c expression is a cell intrinsic requirement to suppress iNKT cell generation. We generated bone marrow (BM) chimeras where WT origin donor cells were used to reconstitute thymus development in RAG-deficient host mice, either alone or mixed at an unequal ratio (1:2) with sγ cTg origin bone marrow cells. When analyzing the frequency of WT donor origin (CD45.1) iNKT cells, we found that WT origin BM cells gave rise to significantly reduced frequencies of iNKT cells, if they developed in a mixed thymic environment with sγ cTg origin thymocytes. Thus, sγ cTg origin BM cells impaired the generation of iNKT cells not only for sγ cTg but also for WT iNKT cells (Fig. 7a). These results indicate that sγ c's effect to suppress iNKT cell development is mediated by a cell extrinsic mechanism.  40 . We did not find any significant difference in γ c and IL-2Rβ expression between WT and sγ cTg iNKT cells, which indicated that sγ cTg did not impair iNKT cell generation because of defects in cytokine receptor expression (Fig. 7b). To further examine if sγ c protein interferes with IL-15 signaling, we examined IL-15 downstream signaling in iNKT cells in the presence or absence of recombinant sγ c proteins. Recombinant sγ c proteins were produced in 293 T cells, and we confirmed successful formation of disulfide-linked sγ c homo-dimers which represent the bioactive form of sγ c protein (Fig. 7c) 24 .
IL-15 signaling is considered critical for in iNKT cells because it induces expression of anti-apoptotic proteins 6 . Bcl-2 is a pro-survival factor downstream of IL-15 signaling, and we found that IL-15-induced Bcl-2 expression was profoundly impaired in the presence sγ c proteins. In particular, recombinant sγ c interfered with the pro-survival effect of IL-15 during in vitro culture of thymic iNKT cells, as illustrated by significantly increased Annexin V binding (Fig. 7d) and diminished Bcl-2 expression (Fig. 7e). Thus, sγ c inhibits iNKT cell development in the thymus, presumably by inhibiting IL-15 signaling.

Discussion
Generation of soluble γ c cytokine receptors through alternative pre-mRNA splicing results in two distinct but interlaced events: production of sγ c proteins and diminished surface γ c protein expression 24 . Both events are detrimental for γ c cytokine signaling. Notably, the effect of alternative splicing is limited to sγ c producing cells themselves, but secretion of sγ c proteins can influence the function of other cells in trans. Thus, the physiological role of sγ c proteins can be wide-ranging and diverse. Here we assessed the effect of sγ c expression on thymic development, and we show that increased sγ c production results in impaired thymopoiesis, which is a process dependent on IL-7 signaling 3 , and also in diminished iNKT cell generation, which is an event dependent on IL-15 signaling 7,21,39 . Generation of IL-2-dependent Foxp3 + Treg cells or IL-7-dependent CD8SP thymocytes 5,10,41 , on the other hand, were not affected. These results propose a hierarchy in γ c cytokine responsiveness of post-selection thymocytes, with IL-15 being highly susceptible to increased concentrations of inhibitory sγ c proteins, and IL-2 and IL-7 signaling more resistant to sγ c-mediated inhibition. Collectively, this study reports a new role for sγ c in suppressing IL-15 signaling, and it demonstrates that sγ c can affect generation and differentiation of mature T cell subsets in the thymus.
Because sγ c is highly expressed by DP thymocytes and because DP thymocytes comprise the vast majority (~85%) of thymocytes 42 , these results further suggest a role for DP thymocytes as a major source of sγ c protein that dampens γ c cytokine signaling in the thymus. Consequently, secretion of sγ c proteins represents a new function for DP cells, and it suggests that DP thymocytes play an active role in thymic T cell differentiation by modulating γc cytokine signaling. Conventionally, DP thymocytes have been considered as only a transient developmental stage that is short-lived and that serves no other purpose than providing a pool of random TCR repertoire to be positively selected by the thymic self-peptide/MHC complexes 43,44 . In fact, DP thymocytes do not produce cytokines, and they are not considered to participate in T cell selection or maturation. Moreover, DP thymocytes are metabolically inactive and do not consume nutrients or compete for pro-survival factors 41 . Along these lines, termination of IL-7Rα expression on DP thymocytes has been suggested to prevent DP cells from consuming IL-7 which would interfere with IL-7-dependent proliferation of DN thymocytes 18,23 . Thus, DP thymocytes are thought to be a developmentally inert population that do not affect or control differentiation or selection of T cells in the thymus. On the other hand, there is an increasing body of evidence that shows DP thymocytes actively participating in T cell development in a cell extrinsic fashion. Such an idea is illustrated by the requirement for DP thymocytes to promote γ δ T cell signature gene expression in immature DN thymocytes 45 , and also by a requirement for SLAM-SLAM homotypic interactions among DP thymocytes for positive selection of iNKT cells 46,47 . In the current study, we report a new mechanism of how DP cells affect thymic T cell differentiation, which is through the secretion of inhibitory sγ c proteins. We think that sγ c is the first of a class of soluble factors that are expressed by DP cells to interfere with thymic development. Sγ c differs from other factors expressed by DP thymocytes, such as lymphotoxin and SLAM 46,47 , because it is not expressed in a membrane-bound form and does not require cell-cell contact. Collectively, DP thymocytes are a major source of sγ c proteins, and sγ c sets the threshold for γ c cytokine signaling and tunes γ c cytokine responsiveness during T cell development in the thymus.
The inhibitory mechanism of sγ c proteins has been previously described 24 . In brief, sγ c proteins form homo-dimers that bind with high affinity to unliganded cytokine receptors, such as IL-7Rα and IL-2Rβ . Direct binding of sγ c to IL-7Rα or IL-2Rβ sequesters these receptors and can prevent them from binding to membrane γ c proteins, which is necessary for cytokine signaling. Because IL-2 and IL-15 share the same IL-2Rβ /γ c complex for ligand binding and signaling 1 , by implication, sγ c binding to IL-2Rβ should interfere with both IL-2 and IL-15 signaling. Interestingly, during sγ cTg T cell development, we found that IL-15 but not IL-2-dependent events were impaired.  iNKT cell development was significantly blunted but Foxp3 + Treg cell generation remained intact. These results suggested distinct susceptibility of IL-2 versus IL-15 signaling to sγ c-mediated inhibition. Why IL-15 signaling would be more perceptive to sγ c blockade than IL-2 signaling is not clear. As a potential explanation, we considered the fact that IL-15 signaling requires IL-15 trans-presentation by IL-15Rα 48 , and that iNKT cell development depends on IL-15Rα -mediated IL-15 trans-presentation by thymic stromal cells 49,50 . Formation of a quaternary complex of IL-2Rβ /γ c hetero-dimers on one cell with an IL-15/IL-15Rα complex on another cell could be more susceptible to steric hindrance by sγ c proteins than the assembly of a functional IL-2Rα , β /γ c signaling complex on the same cell. Altogether, the current results demonstrate an interference of sγ c with IL-15-dependent steps during T cell development, and confirm the in vivo significance of sγ c proteins in thymocyte differentiation.
The roles of γ c cytokines in thymocyte development are well appreciated. Positive selection and lineage choice are two distinct events 51 . While TCR signaling controls positive selection, γ c signaling plays a critical role in lineage fate decision and differentiation of post-selection thymocytes 52 . Following positive selection, IL-7 signaling induces Runx3 expression and imposes CD8 lineage choice 10,16,19 , whereas IL-2 signaling is necessary to upregulate Foxp3 and promote Treg cell differentiation in CD4SP cells 5 . For iNKT cells, IL-15 is a critical maturation and differentiation signal, and the absence of IL-15 results in paucity of iNKT cells in both the thymus and peripheral tissues 7,20,39 . Thus, the reduced thymic iNKT cell numbers in sγ cTg is in line with impaired IL-15 signaling by sγc and the requirement for IL-15 in iNKT cell generation.
Importantly, thymic iNKT cells comprise a functionally and phenotypically heterogeneous population that contains distinct subsets of iNKT cells with differing degree of IL-15 dependency 6,37 . NKT1 cells, which correspond largely to stage 3 iNKT cells, express high levels of T-bet which in turn is critical for their maturation, survival and effector function 53,54 . Both NKT1 lineage choice and T-bet upregulation depend on IL-15 signaling 6,55 , so that impaired IL-15 signaling mostly affects NKT1 cells. NKT17 cells, on the other hand, depend exclusively on IL-7, but not IL-15, for their survival and homeostasis 56 . Thus, it was curious that sγ c overexpression not only reduced number and frequency of IL-15-dependent NKT1 cells, but also of NKT17 and even NKT2 cells. However, these results can be reconciled when taking into account that sγ c does only not inhibit IL-15 signaling, but also signaling by IL-2, IL-7, and presumably other γ c cytokines 24 . Accordingly, sγ c would not only block generation of IL-15-dependent NKT1 cells, but could also impair IL-7-dependent NKT17 cell development in the thymus. Because NKT2 cells were also reduced by sγ c overexpression, this scenario further suggests a role of γ c signaling in NKT2 lineage differentiation too.
Finally, the current results do not exclude the possibility that sγ cTg could have interfered with cell proliferation to diminish thymic iNKT cell numbers. Positively selected stage 0 iNKT cells undergo massive (~100 fold) expansion upon differentiation into stage 1 iNKT cells which is dependent on c-Myc 57 . What cellular signals drive the proliferation is not clear, and we cannot formally discard the possibility that IL-15 could be involved in c-Myc-dependent proliferation during stage 0/1 transition. Whether this is indeed the case still remains to be tested. In sum, the inhibitory effect of sγ c on IL-15 signaling in vivo and the impaired generation of thymic iNKT cells in sγ cTg mice put forward a model of cytokine regulatory mechanism that requires integration of a role of sγ c in controlling γ c cytokine signaling.

Materials and Methods
Mice. C57BL/6 (CD45.2), CD45.1 congenic mice, and RAG −/− mice were obtained from Charles River, Wilmington, MA, and from the Orient Bio, Korea. Soluble γ c-transgenic mice were described and maintained in our colony 24 . Animal experiments were approved by the Pusan National University Institutional Animal Care and Use Committee (PNU-2014-0620) and the NCI Animal Care and Use Committee. All mice were cared for in accordance with Pusan National University School of Medicine and NIH guidelines.
Quantitative Real-Time PCR. Total RNA was isolated from sorted thymocytes with the RNeasy Mini kit (Qiagen). RNA was reverse transcribed into cDNA by oligo (dT) priming with the QuantiTect Reverse transcription kit (Qiagen). Quantitative RT-PCR (qRT-PCR) was performed with an ABI PRISM 7900HT Sequence Detection System and the QuantiTect SYBR Green detection system (Qiagen). Primers sequences are as follows. sγ c (F: 5′ -CATGAACCTAGATTCTCCCTGCC-3′ ; R: 5′ -TGATGGGGGGAATTGGAGIIIIICCTCTACA-3′ ) and In vitro stimulation with recombinant IL-15. Thymocytes were incubated in vitro with 20 ng/ml recombinant human IL-15 (Peprotech) in the presence or absence of recombinant sγ c (500 ng/ml). Thymocytes were harvested 3 days after incubation, and stained for intracellular Bcl-2 expression. Annexin V staining was performed according to the manufacturer's instructions (BD Biosciences).

Statistical analysis.
Data are shown as mean ± SEM. Statistical differences were analyzed by unpaired two-tailed Student's t-test. P values of less than 0.05 were considered significant. *p < 0.05, **p < 0.01, ***p < 0.001. All statistical analysis was performed using GraphPad Prism.