Semi-invariant natural killer T (NKT) cells are innate-like lymphocytes with immunoregulatory properties. NKT cell survival during development requires signal processing by activated RelA/NF-κB. Nonetheless, the upstream signal(s) integrated by NF-κB in developing NKT cells remains incompletely defined. We show that the introgression of Bcl-xL-coding Bcl2l1 transgene into NF-κB signalling-deficient IκBΔN transgenic mouse rescues NKT cell development and differentiation in this mouse model. We reasoned that NF-κB activation was protecting developing NKT cells from death signals emanating either from high affinity agonist recognition by the T cell receptor (TCR) or from a death receptor, such as tumor necrosis factor receptor 1 (TNFR1) or Fas. Surprisingly, the single and combined deficiency in PKC-θ or CARMA-1—the two signal transducers at the NKT TCR proximal signalling node—only partially recapitulated the NKT cell deficiency observed in IκBΔN tg mouse. Accordingly, introgression of the Bcl2l1 transgene into PKC-θ null mouse failed to rescue NKT cell development. Instead, TNFR1-deficiency, but not the Fas-deficiency, rescued NKT cell development in IκBΔN tg mice. Consistent with this finding, treatment of thymocytes with an antagonist of the inhibitor of κB kinase —which blocks downstream NF-κB activation— sensitized NKT cells to TNF-α-induced cell death in vitro. Hence, we conclude that signal integration by NF-κB protects developing NKT cells from death signals emanating from TNFR1, but not from the NKT TCR or Fas.
Natural killer T (NKT) cells are thymus-derived, innate-like lymphocytes that resemble both natural killer and T cells in phenotype and function. Based on the T cell receptor (TCR) α-chain expressed, NKT cells are broadly divided into two types: Type I NKT cells, the majority group, expresses an invariant TCR α-chain encoded by a conserved rearrangement between mouse Vα14 and Jα18 gene segments. This invariant α-chain predominantly pairs with mouse Vβ8.2 (and with Vβ7, 6, 2, and 14) TCR β-chain to express a semi-invariant NKT TCR. Type II NKT cells, the minority group, expresses more diverse TCR α- and β-chains. This report focuses on the semi-invariant NKT (hereinafter NKT) cells. In this regard, it is noteworthy that mouse, rat, human, and other mammalian species also develop functional NKT cells that express the orthologous semi-invariant TCR. Hence, findings in mice will have implications for understanding NKT cell development and function in other mammalian species including humans.
NKT cells control immune responses to microbial, allograft and self antigens. NKT cell functions are controlled by lipid ligand-presenting CD1d molecules. Consequently, NKT cells are capable of sensing cellular lipid homeostasis to maintain cardiovascular and metabolic health1. These seemingly divergent functions are mediated by the ability of NKT cells to rapidly produce pro-inflammatory and immunoregulatory cytokines, which permits these cells to interact with and transactivate cells of the innate and adaptive immune systems2. These properties of NKT cells are acquired during thymic development. Hence, insights into the mechanisms by which NKT cells develop, attain functional competence and are maintained can help understand the role these cells play in health and disease.
The unique phenotypic and functional features of NKT cells —especially their capacity to secrete cytokines promptly after thymic emigration3 —suggested that these cells might follow a distinct ontogenetic path. Commitment to the NKT cell lineage occurs at the CD4- and CD8-positive (DP) thymocyte stage4,5,6,7 whence self-lipid(s) bound CD1d/NKT TCR8,9,10,11 and homotypic/heterotypic SLAM-SLAM12,13 interactions lead to downstream PKCθ-NFκB14, NFAT-Egr215,16,17, and SAP-Fyn12,13,18 activation, which turns-on a unique transcriptional programming mediated by the Zbtb16-encoded master transcription factor PLZF (pro-myelocytic leukaemia zinc finger). PLZF controls multiple molecular events distinguishing NKT cells from conventional T cells19,20. NK1.1-negative NKT cell precursors undergo several rounds of cell division21,22, upon which some remain within the thymus whereas the majority emigrate and populate the peripheral lymphoid organs23. Both the thymic and peripheral NK1.1-negative NKT cells undergo further maturation and become NK1.1-positive NKT cells. During this maturation process, NKT cells differentiate into subsets and acquire distinct effector behaviours24,25,26,27,28,29,23.
Several research groups30,31,32,33,34,35,36, including ours14,37, have reported that signal integration by NF-κB was essential for NKT cell development, survival, and function. RelA and RelB were shown to confer cell autonomous and cell extrinsic activities of NF-κB, respectively30,31,32. The cell-extrinsic NF-κB activity was required for IL-15 induction32, a cytokine essential for survival as well as for terminal NK1.1-negative to NK1.1+ NKT cell differentiation and effector maturation38,39,40,41,42,43,44,45,46,47. Likewise, the cell-intrinsic NF-κB activity was essential for the survival of NK1.1-negative NKT cells, as introgression of Bcl2l1 transgene under the control of the Lck proximal promotor, which thereby encodes Bcl-xL in thymocytes, rescued NKT cell numbers but not function14. Despite this understanding of the role of NF-κB in NKT cell development, a few key questions have remained unanswered: (a) Is signal integration by NF-κB essential for the survival of both immature NK1.1-negative and mature NK1.1+ NKT cells?; (b) Does NF-κB activation protect against TCR-induced NKT cell death?; and (c) What other death receptor(s) trigger NKT cell death during ontogeny? In seeking answers to these questions, we report herein that signal integration by NF-κB is essential for the survival of both immature NK1.1-negative and mature NK1.1+ NKT cells and to protect these cells from death signals induced by tumor necrosis factor (TNF)-α and emanating from TNF receptor superfamily 1a (TNFR1), but not the NKT TCR or Fas.
Bcl-xL overexpression rescues immature and mature NKT cell development in NF-κB signalling deficient mice
We previously showed that Bcl-xL overexpression rescued NKT cell development in mice expressing IκBΔN, a dominant-negative inhibitor of NF-κB activation37. Nonetheless, NKT cells that developed in these animals failed to respond to stimulation with αGalCer, a potent CD1d-restricted NKT cell agonist14. This lack of response to the agonist could be due to immaturity of the NKT cells that developed in Iκ BΔN tg;Bcl2l1 tg mice, their inability to process the TCR activation signals, or both. Because the genetic rescue experiment was performed in BALB/c mice, which do not express the Klrb1c allele that codes for NK1.1, maturation of NKT cells from NK1.1-negative to NK1.1+ NKT cells could not be determined37. Hence, we generated Iκ BΔN tg;Bcl2l1 tg mice in the C57BL/6 background (B6), which express the NK1.1-coding Klrb1c allele detectable with the PK136 mAb. Phenotyping of NKT cells from C57BL/6 and B6-Iκ BΔN tg;Bcl2l1 tg mice revealed that both NK1.1-negative and NK1.1+ NKT cells developed in B6-Iκ BΔN tg;Bcl2l1 tg mice (Fig. 1A–C). Further, the NKT cells of these mice showed phenotypic maturation as judged by expression of CD4, CD69, CD122 (IL-2/15 Rβ chain), Ly49 and Ly6C as well as the loss of CD62L expression similar to NKT cells that develop in C57BL/6 mice (Fig. 1D). Hence, both NK1.1-negative immature and NK1.1+ mature NKT cells escape death by thymocyte-specific Bcl-xL overexpression in NF-κB signalling-deficient mice.
PKC-θ and CARMA1-deficient mice poorly develop NKT cells
NF-κB activation integrates upstream signals from cell surface receptors, including both antigen receptors and death receptors. We initially focused on the antigen receptor because NKT TCRs are selected by agonistic self ligand(s). As well, we previously reported that a fraction of NK1.1-negative NKT cells that developed in Iκ BΔN tg mice expressed Nr4a1-encoded Nur7737. Nur77 is a TCR activation-induced pro-apoptotic transcription factor48, which was shown to be expressed in stage 0 precursor NKT cells49 and upon NKT TCR stimulation by αGalCer or bacterial glycosphingolipids50. Therefore, we determined whether the NKT TCR was the source of the death signal in developing NKT cells. Toward this goal, we determined the role of two NKT TCR proximal signal nodes, PKC-θ (encoded by Prkcq) and CARMA1. Additionally, the SLAM-SAP-FynT signalling axis partially integrates with PKC-θ51. As reported by us previously14, PKCθ-deficiency only partially impacts NKT cell development. This partial impact on NKT cell development was also reflected by CARMA1-deficiency and by combined PKC-θ and CARMA1 deficiencies (Fig. 2). Because NKT cell phenotype in PKCθ-deficient mice were reported previously (see Fig. 3 as well) and because NKT cell development in CARMA1-deficient mice reflected that of PKCθ-deficiency, NKT cell phenotype in neither type of mouse was further investigated. The results so far suggest that the membrane proximal NKT TCR-PKCθ-CARMA1 and/or SLAM-SAP-FynT-PKCθ-CARMA1 axes are unlikely required for transducing the death signal in developing NKT cells.
Bcl-xL overexpression does not rescue NKT cell development in PKCθ-deficient mice
To provide further evidence that PKCθ-CARMA1 activation does not induce a key death signal in NKT cell development, we tested whether Bcl-x L transgene introgression into PKCθ mutant mice rescues NKT cell development. Hence, we generated Prkcq −/−;Bcl2l1 tg mice and analysed NKT cell development in them. Surprisingly, Prkcq −/−;Bcl2l1 tg mice poorly developed NKT cells akin to Prkcq −/− and Carma1 −/− mice (Figs 2 and 3). Together, the data indicated that the NKT TCR and its downstream signalling modules PKC-θ and CARMA1 unlikely transduce death signals within developing NKT cells. Or alternatively, it is quite possible that Bcl-xL overexpression was unable to rescue the loss of NKT cells from the extrinsic apoptotic pathway that may be induced by the NKT TCR.
NKT cells constitutively express several TNF receptor superfamily members
We next turned to death domain-containing cell surface receptors as the source of death signalling in developing NKT cells. To lay focus on a particular TNFR superfamily (TNFRSF) member, we first defined TNFR1 (TNFRSF1A/CD120a), TNFR2 (TNFRSF1B/CD120b), Fas (TNFRSF6/CD95), OX40 (TNFRSF4/CD134), and GITR (TNFRSF18/CD357) expression by thymic, splenic and hepatic NKT cells. Thymic and splenic NKT cells constitutively expressed TNFR1, TNFR2, Fas, and GITR but not OX40. Of these, TNFR1 expression by NKT cells was highest amongst thymocyte subsets tested (Fig. 4A). We then focused on TNFR1 and Fas expression on the three developmental stages (stages 0 + 1, 2 and 3) of thymic NKT cells because these two TNFRSF members contain a cytoplasmic death domain. Both TNFR1 and Fas were equally expressed by all three thymic NKT cell developmental stages (Fig. 4B). Thus, NKT cells indiscriminately express the two tested TNFRSF members that contain death domains.
Fas-deficiency fails to recue NKT cell development in NF-κB signalling-deficient mice
We next determined whether the expression of TNFRSF members containing a death domain altered NKT cell development. Because Fas expression was high, the development of NKT cells in B6-lpr/lpr mice, which express a nonfunctional mutant form of Fas molecule, was determined first. Thymic, splenic and hepatic NKT cells developed in the lpr/lpr mutant mice in frequencies similar to wt/lpr mice (Fig. 5). Further, the introgression of the lpr/lpr mutation into NF-κB signalling deficient IκBΔN tg mice (B6-lpr/lpr;Iκ BΔN tg) mice did not rescue NKT cell development in these mice because very few if any NKT cells were observed in the thymus, spleen or liver of B6-lpr/lpr;Iκ BΔN tg mice (Fig. 5). From these data, we concluded that Fas unlikely transmits the death signal counteracted by NF-κB activation.
Thymic NKT cells develop unhindered in TNFR1-deficient mice
In a similar experiment, we next determined whether TNFR1-deficiency impacted NKT cell development. The data revealed that NKT cell development proceeded unhindered in the thymus of TNFR1-deficient mice and these cells emigrated to the spleen and liver as in wild type mice (Fig. 6A,B). We noticed that the absolute NKT cell number was increased by 25–30% in the thymus, which was statistically significant, but only modestly, albeit not significantly, reduced in the spleen (Fig. 6B). Further, NKT cells proceeded to phenotypic maturity in TNFR1-deficient mice with regards to CD44 and NK1.1 expression as well as CD4, CD69, CD122 (IL-2/15 Rβ chain), Ly6C, Ly49 and CD62L (L-selectin) expression (Fig. 6C–E). Notably, the number of splenic NK1.1+ NKT cells was reduced in a statistically significant manner (Fig. 6D). Thus, TNFR1-deficiency marginally impacted thymic NKT cell development, whilst TNFR1 signalling was partially required for splenic NK1.1+ NKT cell maturation.
NF-κB activation protects developing NKT cells from TNFR1-induced death
Deficiency in the inhibitor of κB kinase (IKK)-β, which is essential for NF-κB activation, results in poor lymphopoiesis due to enhanced sensitivity to TNF-α52. Consequently, normal thymopoiesis and conventional CD4 and CD8 T cell development ensues when mice are doubly deficient in TNFR1 and IKK-β52 suggesting that TNF-α might be a thymic inducer of NKT cell death. Therefore, we determined whether TNFR1-null mutation introgressed into IκBαΔN tg mice rescues NKT cells from death. For this purpose, we bred Tnfrsf1a −/− mice with IκBαΔNtg mice and the resulting progenies were analysed for NKT cell development.
The data revealed that NKT cell development was rescued in Tnfrsf1a −/−;IκBαΔN tg mice when compared with IκBαΔN tg mice (Fig. 7A,B). Even though Tnfrsf1a −/−;IκBαΔN tg NKT cells matured to the CD44hiNK1.1+ stage 3 (Fig. 7C,D), they expressed lower levels of Ly49C/I (Fig. 7E) while expressing normal levels of other developmentally regulated markers (e.g., CD4, CD62L, CD122 and Ly6C) when compared to thymic Tnfrsf1a −/−; IκBαΔN tg NKT cells (Fig. 7E). Because introgression of the lpr/lpr mutation into IκBαΔN tg mice did not rescue NKT cells from death (Fig. 5), we conclude that the effect of TNFR1-deficiency on NKT cell development in Tnfrsf1a −/−;IκBαΔN tg mice is specific. The data also demonstrate that NF-κB activation controls NKT cell homeostasis by overcoming TNFR1-induced death.
NF-κB signalling deficiency renders NKT cells sensitive to TNF-α-induced apoptosis
The data so far indicate that NF-κB signalling protects NKT cells from TNFR1-induced cell death. Further, in a previous study, we reported that NKT cells in Iκ BΔN tg mice, in which thymocytes are unable to activate NF-κB, expressed higher frequency of caspase 8 and caspase 9 positive cells. Forced expression of Bcl-xL in Iκ BΔN tg NKT cells lowered the frequency of caspase 8 and caspase 9 positive cells to levels observed in wild type non-transgenic mice37. Guided by that finding, we designed the following experiment: C57BL/6 mouse thymocytes were cultivated in the presence of 3–24 μM BMS345541 —an inhibitor of IKK, a key node upstream of NF-κB activation53— or DMSO —the vehicle used to dissolve BMS345541— in the presence or absence of 10 or 100 ng/ml mouse TNF-α. After 14 hours, cells were labelled with fluorescently-tagged inhibitors for active caspase 3, caspase 8, and caspase 9. The binding of fluorescently-tagged inhibitors to active caspases allowed their detection within NKT cells by flow-cytometry. Caspase substrate tagged cells were co-stained with 7-aminoactinomycin D (7-AAD) to reveal dying/dead cells. The data presented herein are those performed with 6 μM BMS345541 treatment. This is because 12 μM BMS345541 was partially and 24 μM BMS345541 was fully toxic to thymocytes in vitro even in the absence of TNF-α stimulation (data not shown). Further, 3 μM BMS345541-treated thymocytes showed suboptimal caspase activation upon stimulation with 10 ng/ml and even 100 ng/ml TNF-α (data not shown).
We found that 10 ng/ml TNF-α activated caspase 3, caspase 8, and caspase 9 within thymic NKT cells in which IKK activation was inhibited with 6 μM BMS345541. In contrast, the same amount of TNF-α poorly induced cell death in NKT cells in which IKK signalling was intact (Fig. 8). This finding is consistent with the notion that NF-κB signalling is critical for protecting developing NKT cells from TNFR1-induced cell death. Furthermore, the finding that TNF-α induces both caspase 8 and caspase 9 suggested that, under the conditions tested, thymic NKT cell death is mediated by both extrinsic and intrinsic apoptotic pathways.
Unlike conventional T cells that are selected by antagonistic ligands54, several thymic T cell lineages —such as NKT cells and T regulatory cells— are selected by agonistic ligands. Agonistic ligands that positively select NKT cells in the thymus are similar or identical to the ligands that activate these cells in the periphery55,56,57. In addition to NKT TCR agonistic interactions, NKT cells persistently interact with the selecting DP cells through SLAM-SLAM interactions, which activates PKC-θ via the SAP-FynT signalling module12,13,18,51,58,59,60. Hence, we reasoned that developing NKT cells need a survival signal to protect them from death induced by high affinity NKT TCR interaction with cognate agonistic ligand and potentially SLAM-SLAM molecules as well. Consistent with this line of thinking, several laboratories30,31,32,33,34,35,61, including ours37, have found a critical requirement for NF-κB in providing a survival signal during NKT cell development. Similarly, NKT cell development partially depended on signals relayed through the TCR proximal signalling nodes, PKC-θ14 (this report) and CARMA1 (this report). Nonetheless, even though forced overexpression of Bcl-xL partially rescued NKT development in NF-κB signalling deficient mice, the introgression of the same transgene into PKCθ-deficient mice failed to phenocopy the rescue seen in B6-IκBαΔN tg;Bcl2l1 tg mice. Hence, NF-κB provides a survival signal to NKT cells, in part through the TCR-PKCθ-CARMA1-axis, although other receptor(s) may play a role as well. This conclusion is consistent with other reports61,62,63,64 including the role of IL-15 in controlling NKT cell homeostasis41,42,43,46,47. Together these findings suggest that NKT cells narrowly escape death during thymic development.
The finding that TNF-α induces both caspase 8 and caspase 9 suggested that, under the conditions tested, thymic NKT cell death is mediated by both extrinsic and intrinsic apoptotic pathways. This result is consistent with our previous observation in IκBΔN tg mice wherein active caspase 8 and caspase 9 expression was increased within residual NKT cells37. How TNF-α induces both the intrinsic and extrinsic apoptotic pathways in NKT cells is unknown. Based on studies in hepatocytes and cancer cells, we speculate that TNFR1 ligation in NKT cells deficient in NF-κB signalling activates caspase 8 within complex II death-inducing signalling complex. Activated caspase 8 in addition to activating the executioner caspases 3 and 7 then activates caspase 9 perhaps by cleaving BH3-interacting domain death agonist (BID) to truncated BID. Truncated BID formation induces mitochondrial outer membrane permeabilization (MOMP) via B cell lymphoma 2 antagonist or killer activation. Proteins released by MOMP triggers death receptor-induced apoptosis mediated by caspase 9 apoptosome and the activation of the executioner caspases 3 and 765,66,67. Our finding that forced BCL-xL overexpression partially rescues NKT cell development in NF-κB signalling deficient IκBΔN tg mice (this report and ref.37) lends credence to this hypothesis.
NF-κB activation is tightly controlled in NKT cells and too little or too much signal is detrimental. Hence, the absence of RelA31 or inability to activate NF-κB as in the IκBΔN tg mice prevents NKT cell development14,30. In contrast, when the NF-κB pathway is overactive, as in mice expressing the constitutively activated IκB kinase in hematopoietic cells or mice deficient in the negative regulator of NF-κB signalling CYLD, NKT cells develop but these cells fail to mature and to populate the periphery61. These findings suggest that NF-κB functions as a rheostat to control the development and survival of NKT cells. In doing so, NF-κB activation perhaps sets the threshold for peripheral NKT cell activation. Setting such a threshold may be critical for agonistic ligand(s)-dependent NKT cell selection in the thymus and function in the periphery so as not to set off an autoreactive response should the agonistic ligand(s) be displayed by CD1d-positive cells in peripheral tissues such as hepatocytes and intestinal epithelial cells68,69,70. The exact mechanism by with NF-κB functions as a rheostat and sets the activation threshold remains to be elucidated.
Materials and Methods
Age-matched, 8–12-week old mice were used in the experiments described herein. C57BL/6, B6-lpr/lpr and B6.129-Tnfrsf1 −/− mice were purchased from the Jackson Laboratory. B6.129-Prkcq −/− 71 and B6.129-Carma1 −/− 72 were generous gifts from Drs. D.R. Littman (NYU) and V.M. Dixit (Genentech). B6-Iκ BΔN tg;Bcl2l1 tg mice were generated by crossing B6-IκBΔN tg with C-Bcl2l1 tg, both previously described mice14,37, and backcrossed to the C57BL/6 background for eight generations. B6.129-Tnfrsf1 −/−;IκBΔN tg mice were generated by crossing B6.129-Tnfrsf1 −/− mice with B6-IκBΔN tg mice and inter-crossing the F1 B6.129-Tnfrsf1 +/−;IκBΔN tg heterozygotes to generate B6.129-Tnfrsf1 −/−;IκBΔN tg homozygotes with the Tnfrsf1 null mutation. A similar strategy was used to generate B6.129-Prkcq −/−;Bcl2l1 tg and B6-lpr/lpr;IκBΔN tg mice. All mouse crosses and experiments complied with the approved protocol M/13/193 in accordance with relevant guidelines and regulations of Vanderbilt University Institutional Animal Care and Use Committee.
Antibodies and Reagents
All antibodies (Abs) used for the identification of NKT cells and developmental stages are listed in Table 1 as described73. α-Galactosylceramide (αGalCer; KRN7000) was generously provided by Kirin Brewery Company (Gunma, Japan) or purchased (Funakoshi, Japan). Mouse CD1d monomer was obtained from the NIH Tetramer Facility (Emory University). Preparation of CD1d tetramers from monomers and their use are described elsewhere73. IKK inhibitor BMS 345541 was purchased from Sigma. Recombinant mouse TNF-α and 7AAD were purchased from BD Biosciences. CaspGLOW fluorescein Active Caspase 3 (FITC-DEVD-FMK), Caspase 8 (FITC-LETD-FMK) and Caspase 9 (FITC-LEHD-FMK) staining kit were purchased from Invitrogen.
Approximately 3 × 106 C57BL/6 mouse thymocytes were treated with 3–24 μM BMS345541 or DMSO control for 60 min. As 12 μM BMS345541 was partially and 24 μM BMS345541 fully toxic to thymocytes in vitro, and because 3 μM BMS345541 showed suboptimal caspase activation, all experiments were performed with 6 μM BMS345541 treatment. BMS345541-Treated cells were cultivated in the presence or absence of 10 ng/ml TNF-α for 14 hours. Cells were then incubated with fluorescently labelled caspase 3, caspase 8 or caspase 9 inhibitors for an additional 60 min. Cells were washed twice with wash buffer provided with CaspGLOW caspase assay kits according to the Manufacturer’s instructions. Washed cells were surface stained with CD1d tetramer and with antibodies against mouse CD3ε. Finally, cells were incubated with 7-AAD for 15 min and analyzed by FACS.
Comparisons between three or more groups were performed by one-way ANOVA with Tukey’s post test to determine significance. Comparison between two groups was performed by two-tailed, unpaired t test. All statistical analyses were performed using Prism (GraphPad software).
NKT cells were identified as CD3ε+tetramer+ cells among B220lo splenocytes and hepatic mononuclear cells or CD8lo thymocytes by using four-colour flow cytometry, which was performed with a FACSCalibur instrument (Becton-Dickinson). NKT cell developmental stages were phenotyped by using six-colour flow cytometry, which was performed with a LSR-II instrument (Becton-Dickinson). Data were analyzed with FlowJo software (Treestar Inc.). Absolute NKT cell numbers were calculated from % tetramer+ cells and total number of cells recovered from each organ.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank the NIH Tetramer Facility in Atlanta, GA, for α-galactosylceramide-loaded mouse CD1d tetramers and Kirin Brewery Company for KRN7000. This work was supported by VA Merit Award (BX001444) to SJ and by NIH training (HL069765 & AI007611), research (AI042284, AI061721, AI070305, HL089667, AI068149, AI074754 & AI064639) and core (CA068485 & DK058404) grants.