Critical Role for Very-Long Chain Sphingolipids in Invariant Natural Killer T Cell Development and Homeostasis

The role of sphingolipids (SLs) in the immune system has come under increasing scrutiny recently due to the emerging contributions that these important membrane components play in regulating a variety of immunological processes. The acyl chain length of SLs appears particularly critical in determining SL function. Here, we show a role for very-long acyl chain SLs (VLC-SLs) in invariant natural killer T (iNKT) cell maturation in the thymus and homeostasis in the liver. Ceramide synthase 2-null mice, which lack VLC-SLs, were susceptible to a hepatotropic strain of lymphocytic choriomeningitis virus, which is due to a reduction in the number of iNKT cells. Bone marrow chimera experiments indicated that hematopoietic-derived VLC-SLs are essential for maturation of iNKT cells in the thymus, whereas parenchymal-derived VLC-SLs are crucial for iNKT cell survival and maintenance in the liver. Our findings suggest a critical role for VLC-SL in iNKT cell physiology.

inTrODUcTiOn Natural killer T (NKT) cells are a subset of T lymphocytes that act at the interface between the innate and adaptive immune systems (1). A subset of NKT cells, termed invariant NKT (iNKT) cells, is found at low levels (<1% of hematopoietic cells) in many organs but is highly enriched in the liver, where it represents up to 30% of resident T lymphocytes (2). iNKT cells express a restricted T cell receptor (TCR) repertoire composed of an invariant TCR α chain (Vα14-Jα18 in mice and Vα24-Jα18 in humans) paired with a limited number of β chains (Vβ8.2, Vβ7, or Vβ2 in mice and Vβ11 in humans). iNKT TCRs recognize CD1d-restricted glycolipid ligands, among them a glycosphingolipid (GSL), α-galactosylceramide (α-GalCer), which was originally isolated from a marine sea sponge. Reactivity to α-GalCer is used to define iNKT cells and distinguishes them from other NKT subsets (3). Most mammalian GSLs exist as β anomers although there is some evidence that mammalian α-linked glycosylceramides are also found (4,5).

Mice
Ceramide synthase 2 null and WT littermate control mice were generated by crossing C57BL/6 CerS2 +/− mice with 129S2v/Jae CerS2 +/− mice (16). Mice were housed in a specific pathogenfree barrier facility at the Weizmann Institute of Science. Experimental procedures were performed in accordance with guidelines approved by the Institutional Animal Care and Use Committee.

lcMV-We Virus infection
LCMV-WE was propagated in baby hamster kidney 21 cells. The viral titer was determined by a plaque assay (31). Two million plaque-forming units (PFU) of LCMV-WE were injected intravenously through the retro-orbital plexus into CerS2-null mice. Mice were sacrificed 2 and 6 days later and their livers and spleens removed. The big lobe of the liver was fixed in optimal cutting temperature freezing medium and the rest of the liver frozen at −80°C. Liver sections (5 µm) were cut using a cryostat (Leica Systems, Nussloch, Germany) and dried overnight at room temperature. Sections were used for anti-LCMV staining or stored at −20°C. Sections were fixed with acetone for 10 min and blocked with 10% fetal bovine serum (FBS) for 1 h. Sections were stained with rat anti-LCMV serum for 1 h followed by three washings with PBS. Sections were incubated with a Cy3conjugated donkey anti-rat antibody (1:700 in 1% FBS) and counterstained with Hoechst 33342 (Molecular Probes, OR, USA) for 5 min and washed three times with PBS. Sections were mounted using Fluoromount G mounting medium (Southern Biotech, AL, USA). Slides were dried overnight and images were captured using a Nikon eclipse Ti-s inverted microscope. Images were processed using NIS elements 3.0 imaging software (Nikon Instruments Inc., NY, USA).

rna-seq Processing and analysis
Liver was homogenized in 750 µl Trizol reagent (Invitrogen) and vortexed after adding 150 µl chloroform and incubated for 5 min at room temperature, followed by centrifugation for 15 min at 20,000 g. 300 µl from the upper phase was transferred to a new microtube and after the addition of 70% ethanol (1:1 vol), transferred to an RNeasy column (RNeasy mini kit, Qiagen). RNA was purified following DNase I digestion on the column according to manufacturer's instructions. For library preparation, 6.1 µl of total RNA (5 ng/µl) was used for the MARS-seq protocol (33). RNA-seq libraries were sequenced using Illumina NextSeq-500. For analysis, adapters and low quality bases were removed from the raw reads using cutadapt (34). The remaining reads were mapped to the Mus musculus genome (mm10) using STAR v2.4.2a (35) with the option alignEndsType EndToEnd. Only reads with unique mapping were considered for further analysis. Gene expression levels were calculated using htseq-count (36) with option intersection-strict and mm10 Refseq 3′UTR GTF annotations. Duplicate reads were filtered if they mapped to the same gene and had identical UMIs. Normalization and differential expression analysis was performed using the DESeq2 R-package (Bioconductor, https:// bioconductor.org/packages/release/bioc/html/DESeq2.html).

Quantitative real-time Pcr
Total RNA was isolated using an RNeasy mini kit according to manufacturer's instructions. cDNA synthesis was performed using a QScript™ C-DNA synthesis kit and qPCR performed using the Perfecta SYBR Green fastMix and an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Life Technologies).

Flow cytometry
Approximately 10 6 cells were blocked with an anti-CD16/32 FcγR antibody for 10 min and subsequently stained with fluorescently labeled antibodies for 10 min. Flow cytometry was performed on an LSR II flow cytometer (BD Biosciences, San Jose, CA, USA) and analyzed using Flow Jo V10 software (Flow Jo, LLC, Ashland, OR, USA).

hepatocytes and Mononuclear cell (Mnc) isolation
Hepatocytes were isolated as described (37). For isolation of MNCs, the supernatant obtained after separating hepatocytes was centrifuged at 300 g to pellet MNCs. Erythrocytes were lysed with ammonium chloride, potassium (ACK) buffer (150 mM NH4Cl, 10 mM KHCO3, 0.1 mM EDTA, pH 7.2), and dead cells separated on a 40% Percoll gradient by centrifugation (30 min, 300 g). The pellet containing MNCs was resuspended in PBS containing 1% BSA and cells counted using a hemocytometer. Splenocytes and thymocytes were isolated after mechanical disruption in PBS containing 1% BSA followed by filtration through a 70-µm cell strainer. Splenocytes were incubated in ACK buffer for 2 min to lyse erythrocytes. Live cells were counted in a hemocytometer using trypan blue. Blood was withdrawn retro-orbitally using capillaries containing anti-coagulating agents. Erythrocytes were lysed using ACK lysis buffer for 2 min. The cells were washed, suspended in PBS, and counted in a hemocytometer.

generation of Bone Marrow (BM)-Derived cells and In Vitro inKT cell activation assay
Bone marrow-derived cells were cultured in Iscove's Modified Dulbecco's Medium supplemented with 10% FCS with 1% (of total volume) of a supernatant from a granulocyte-macrophage colony-stimulating factor (GM-CSF)-expressing cell line (J558-GM-CSF) (38). Dendritic cells (DCs) were used on day 14 of culture. This protocol yielded >90% cell purity as evaluated by FACS (DCs were identified by cell surface expression of CD11c + MHC II + ). DCs (1 × 10 5 cells/ml) were cocultured with 1 × 10 5 cells of a mouse type I NKT cell hybridoma (DN32.D3) in the presence of vehicle or various concentrations of α-GalCer (10-100 ng/ml) in complete Roswell Park Memorial Institute media supplemented with 5% FCS for 24 h. Supernatants were collected and the IL-2 concentration measured by ELISA (R&D systems, MN, USA).
inKT cell enrichment and adoptive Transfer Invariant NKT cells from liver MNCs were enriched using anti-PE microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany).
Briefly, MNCs from livers obtained from at least seven mice were isolated, pooled, and blocked with an anti-CD16/32 antibody for 10 min at 4°C in PBS containing 1% BSA. Ten microliters each of a PE-conjugated CD1d-PBS57 tetramer were added per 10 7 cells in 100 µl of magnetic-activated cell sorting (MACS) buffer (PBS containing 0.5% BSA and 2 mM EDTA) and incubated for 10 min in the dark at 4°C. The cells were washed and resuspended in 80 µl of PBS to which 20 µl of anti-PE microbeads per 10 7 cells were added and kept in the dark for 15 min at 4°C. Cells were washed and resuspended in 500 µl of MACS buffer. Cells were passed through a magnetic separation column followed by three washings to remove unbound cells. The magnetic labeled-bound cells were released from the column by removing the magnet and flushing the column with a syringe plunger. The purity of enriched CD3 + and CD1d/PBS-57 tetramer + cells was determined by flow cytometry. One million iNKT cells were washed and resuspended in 200 µl of PBS. Cells were injected i.v. through the retro-orbital plexus into CerS2-null mice. As a control, an equal amount of PBS was injected into CerS2-null mice.

sl analysis
Sphingolipid levels were analyzed by liquid chromatographyelectrospray ionization-tandem mass spectrometry (LC ESI-MS/ MS) using an ABI4000 quadrupole-linear ion trap mass spectrometer as described previously (39).

statistical analysis
Data are expressed as means ± SEM. An unpaired student's t-test was used to determine statistically significant differences between groups; *p < 0.05, **p ≤ 0.01, and ***p ≤ 0.001.

cers2-null Mice are susceptible to lcMV-We infection
To determine the role of CerS2 in host defense against pathogens, we infected CerS2-null mice with a hepatotropic strain of LCMV, which activates pathways of both innate and adaptive immunity and causes acute hepatitis in mice (40). Two days postinfection (DPI) with 2 × 10 6 PFU of LCMV, increased levels of the virus were detected in the liver of CerS2-null mice compared to WT mice as ascertained by immunofluorescence ( Figure 1A) and by viral titers (Figure 1B). No differences in infection were observed 6 DPI in the liver, and no changes in infection were seen in the spleen at either time point (Figure 1B). Cultured hepatocytes from WT and CerS2-null mice did not display differences in infection excluding the possibility that altered LCMV internalization or proliferative capacity in hepatocytes is the primary cause for the enhanced susceptibility of CerS2-null mice to hepatic LCMV infection ( Figure S1A in Supplementary Material).
IFNα is critical for defense against virus infection (41) but, unexpectedly, levels of IFNα were higher in CerS2-null liver 2 DPI compared to WT mice ( Figure 1C) and downstream components of IFNAR1 signaling pathway were unaltered ( Figure 1D). Likewise, levels of plasmacytoid dendritic cells (pDCs), the major source of IFNα after viral infection (42), were increased in CerS2 null liver ( Figure S1B in Supplementary Material). Levels of monocytes were unaltered ( Figure S1C in Supplementary Material), whereas neutrophils ( Figure S1D in Supplementary Material) and CD8 + T cells ( Figure 1E) were increased in CerS2-null mouse liver prior to infection, probably due to increased hepatocyte turnover (17). The percent of CD4 + T cells was lower in CerS2-null mouse liver before and after viral infection ( Figure 1F). However, LCMV-specific CD8 + T cells were unaltered (Figure 1G), while LCMV-specific CD4 + T cells ( Figure 1H) were increased in CerS2 null mouse liver 6 DPI, indicating that impairment of conventional T cell responses cannot explain the enhanced susceptibility. In conclusion, the susceptibility of CerS2 null liver to LCMV infection after 2 days cannot be accounted for by increased viral internalization into hepatocytes, to a type I IFN response or to virus-specific T cell responses.

lack of inKT cells renders cers2-null Mice susceptible to lcMV infection
Since the number of CD4 + T cells was decreased in CerS2 null mouse liver (Figure 1F), we examined levels of iNKT cells, which form a major fraction of hepatic CD4 + T cells and are involved in protection against pathogens, including viral infection (43)(44)(45). Flow cytometry revealed a striking decrease in the percent and number of iNKT cells in CerS2 null liver (Figures 2A,B).
To determine if the reduction in the number of iNKT cells is responsible for viral infection, mice were engrafted with WT iNKT cells 1 day prior to LCMV infection. Adoptive transfer of an enriched iNKT fraction abolished the enhanced susceptibility of CerS2-null mice to LCMV infection (Figures 2C,D), while the iNKT-depleted fraction (containing high levels of conventional T cells and low levels of iNKT cells) did not ( Figure S3 in Supplementary Material).
To further delineate the role of iNKT cells in hepatic LCMV infection, we performed 3′UTR digital gene expression RNA-seq (Data Sheet S1 in Supplementary Material) and compared genes that were differentially expressed between LCMV-infected CerS2 null mouse liver before and after transfer of iNKT cells, with differentially expressed genes in WT mice before and after infection ( Figure 2E). Differentially expressed genes that were common in both comparisons, but were not common between CerS2-null mice with and without infection, were considered to be a downstream effect of  iNKT cells (Figure 2F). Transfer of iNKT cell into CerS2-null mice altered the expression of some genes similar to those of WT infected liver ( Figure 2E, mainly in cluster VII). Since iNKT cells contribute a small fraction of the total RNA (as only 10 6 iNKT cells were transferred), most of the genes whose expression was altered are hepatocyte derived (Figure 2F). Increased expression of several acute-phase reactants, such as amyloid A proteins (Saa1, Saa2, Saa3), amyloid P component serum (Apcs), haptoglobin (Hp), hemopexin (Hpx) and orosomucoid 2 (Orm2), was observed only in CerS2-null mice after transfer of iNKT cells and infection with LCMV ( Figure 2F). Changes in gene expression were validated for selected acutephase genes by qRT-PCR, and compared to uninfected mice ( Figure 2G). While expression of Hpx, Orm2, Saa2, and Saa3 were increased upon transfer of WT iNKT cells into LCMVinfected CerS2-null mice, Apcs, and Hp were not increased, indicating that these genes are not influenced directly by iNKT cells (Figure 2F).

The Development of cers2 null inKT cells is arrested in the Thymus
Lower levels of iNKT cells were also detected in the thymus of CerS2-null mice, while levels of conventional T cells were unaltered (Figures 3A,B). iNKT cells were arrested in their development at the immature DP stage since CerS2-null mice accumulated immature CD24 + CD44 − iNKT cells (stage 0) and had less mature CD24 − CD44 high iNKT cells (stage 2/3) (Figures 3C,D). Further analysis revealed that most of the cells which were CD24 − CD44 + were at stage 3 (NK1.1 + cells) in both WT and CerS2-null mice (Figures 3E,F; Figure S4 in Supplementary Material). However, in CerS2-null mice, there was a small but non-significant increase in stage 2 iNKT cells (Figures 3E,F). The arrest of CerS2 null iNKT cell maturation could reflect a cell-intrinsic defect in the survival or proliferation of iNKT cells as they mature in CerS2-null mice, or alternatively could be due to extrinsic defects such as reduced surface expression of CD1d or lack of lipid self-antigens allowing thymic positive selection of iNKT cells. Unlike conventional T cells, iNKT cells are selected via the presentation of lipid self-antigen(s) by CD1d on DP thymocytes (46,47). CerS2 null DP thymocytes exhibited a 34 ± 1.5% reduction in CD1d surface expression (Figures 4A,B). Our previous studies demonstrated that surface expression of a number of receptors is reduced in CerS2-null mice (18,25,26). To directly test the impact of reduced levels of CD1d on iNKT cells, we generated BM culture-derived dendritic cells (BMDCs). In response to α-GalCer, CerS2-null BMDCs, although they displayed a 48 ± 4% reduction in surface expression of CD1d (Figures 4C,D), activated an iNKT cell hybridoma cell line (DN32.D3) to a similar extent as WT controls ( Figure 4E). Moreover, CD1d +/− mice, which have a 50% reduction in surface CD1d expression on DP thymocytes, contain a normal iNKT cell compartment (46). We conclude that the reduced expression of CD1d in CerS2 null DP thymocytes is unlikely to account for the reduced number of iNKT cells.

reduced levels of inKT cells in cers2null Mice likely results from lack of self-antigens
We next generated BM chimeras by transfer of WT BM into irradiated CerS2 null hosts (WT > KO) and vice versa (KO > WT). WT > KO and WT > WT chimeras had a similar percent of iNKT cells in the thymus (Figures 5A,B). In contrast, levels of iNKT cells in KO > WT chimeras were reduced and similar to levels found in KO > KO chimeras (Figures 5A,B). These results demonstrate that CerS2 expression by hematopoietic cells is required for thymic iNKT cell development.
Consistent with changes in the SL composition in other organs in CerS2-null mice (16,21,23), levels of VLC-ceramide and VLC-GlcCer were reduced in CerS2 null thymocytes ( Figure 5C). However, GalCer levels were below the limit of detection (0.05 pmol/ mg protein) in WT and CerS2-null thymocytes.
While thymic iNKT cells in WT > KO chimeras were unaffected (Figures 5A,B), levels of iNKT cells in the liver of WT > KO chimeras were reduced to similar levels as found in KO > KO chimeras (Figures 5D,E). Moreover, KO > WT chimeras displayed normal iNKT cell levels in the liver, similar to WT > WT chimeras (Figures 5D,E), indicating that radiationresistant parenchymal cells are essential for iNKT cell survival and maintenance in the liver.
Further analysis of BM chimeras using CD45.1 + WT donors revealed that in KO > WT chimeras, ~90% of the iNKT cells in the liver were WT-derived ( Figure 5F). This is likely due to incomplete depletion of host T cells during irradiation (48,49), which also occurs for conventional T cells but not for B cells ( Figure S2A in Supplementary Material), perhaps indicating that CerS2 null T lymphocytes (including iNKT cells) have a reduced capacity to compete with residual WT cells (Figure 5F). In the thymus, although very few CD4 + iNKT cells were recovered in KO > WT chimeras (Figure 5A), ~50% were KO-derived (CD45.1cells; Figure 5F), probably due to antigen presentation by residual WT DP thymocytes, allowing limited CerS2 null iNKT cell maturation. However, these cells were nearly completely absent in the liver of KO > WT chimeras (<10%; Figure 5F). In conclusion, the chimera experiments reveal that CerS2-deficient conventional T and iNKT cells have an intrinsic defect when competing with WT cells. However, in CerS2-null mice, where no competition with WT cells occurs, this is unlikely to be the reason for reduced levels of iNKT in the thymus, which are arrested at the DP stage when positive selection takes place.

DiscUssiOn
In the current study we provide evidence that generation and maintenance of iNKT cells depends on levels of VLC-SLs. The reversibility of the susceptibility to LCMV infection upon adoptive transfer of WT iNKT cells into CerS2-null mice strongly supports this contention. These results are in agreement with reports showing that increased levels of iNKT cell in the liver are protective against early stages of LCMV infection (50,51), consistent with a critical role for iNKT cells in the early stages of hepatic viral infection (52). The redundant role  We demonstrate that iNKT cells promote the secretion of Saa1 and other acute-phase proteins in the early stage of viral infection in the liver. Acute-phase proteins are secreted from hepatocytes upon infection to control infection and restore homeostasis (54). Saa1 switches neutrophils from an anti-to a pro-inflammatory state and promotes their interaction with iNKT cells (55). Saa1 and Saa3 are increased upon HCV infection, while Acot1, which hydrolyzes long chain fatty acyl CoAs was downregulated (56), similar to our results upon iNKT cell transfer into CerS2 null infected mice (Figure 2F). Together our results suggest that CerS2-null mice lacking iNKT cells cannot control viral propagation in part due to lack of specific acutephase proteins and possibly also due to other pathways such as those involved in lipid metabolism.
When considering the role of GSLs in the generation and activation of iNKT cells, three issues need to be addressed, namely, the acyl chain length of the GSL, the type of the glycosyl head group, and the anomeric configuration of the glycoside bond. For the first issue, previous studies suggested a role for the SL acyl chain length in binding to CD1d and in the affinity of this complex for the iNKT TCR, with both the long chain base and the acyl chain length determining the stability of lipids bound to CD1d molecules and iNKT activation (50,57,58). Our study supports the idea that VLC-GSLs are major endogenous ligands necessary for iNKT cell maturation in the thymus (59,60) and maintenance in the liver where iNKT cells seek foreign lipids antigens within hepatic sinusoids. In respect to the headgroup of the glycolipid, mice deficient in β-GalCer synthase have normal iNKT cell levels and their antigen-presenting cells (APC) can activate an iNKT hybridoma cell line, while cell lines deficient in β-GlcCer synthase cannot (61). Moreover, depletion of GlcCer-derived GSLs in DP thymocytes resulted in ~50% reduction of the iNKT cell population in the thymus and periphery (62). This supports the notion that C24:1-GlcCer, or more complex GSLs, are the major self-antigens required for iNKT cell development in the thymus and that lack of VLC-SLs in CerS2 null antigen-presenting DP thymocytes leads to reduced levels of iNKT cells. Concerning the anomeric configuration, previous studies have suggested that an α-linked monoglycosylceramide with a C24:1 acyl chain might be an endogenous ligand of iNKT cells (4), although the existence of α-GlcCer in mammals is still somewhat controversial (5,14) whereas C24:1-β-GlcCer is found at high levels in mouse thymus (63). Our data showing lower levels of VLC-GlcCer, concomitant with lower levels of iNKT cells, suggest that VLC-GlcCer, or complex VLC-GSLs, act as the endogenous self-antigens presented by CD1d molecules.
The precise mechanism by which iNKT cell maturation is affected in the thymus of CerS2-null mice is currently unknown. However, even though a number of cell surface receptors are mislocalized in CerS2-null mice (18,25,26), we can exclude the possibility that changes in cell surface expression of CD1d are responsible for the lower levels of iNKT cells in CerS2-null mice. However, we cannot rule out other intrinsic defects in iNKT cells. For example, the increased levels of C16-SLs in CerS2 null thymocytes ( Figure 5C) might affect lysosomal loading of other self-antigens onto CD1d (64), as occurs in some mouse models of lysosomal storage diseases (59), where massive accumulation of SLs and GSLs occurs (65). However, given the relatively low level of C16-SL accumulation in CerS2 null thymocytes (<2-fold; Figure 5C), we consider this unlikely. Not only are VLC-GSLs required for the maturation of iNKT cells in the thymus, our data show that they are also appear to be essential for iNKT cell survival and maintenance in the liver. However, previous studies have shown that in the absence of CD1d, iNKT cells can still proliferate in the liver although to a lower extent (66)(67)(68). Thus, it seems that survival of iNKT cells in the liver is a combined effect of antigen presentation by CD1d, as well as other cofactors that may also be deficient in CerS2-null mice. However, reduced expression of known factors required for iNKT cell recruitment and survival, such as CXCL16 (69), IL-15 (67), IL-7 (70), and ICOSL (71) was not observed in CerS2 null mouse liver (17). Recently, CerS2 was shown to regulate sphingosine 1-phosphate (S1P) levels and S1P-dependent egress of mature thymocytes from the thymus to the periphery (72), with increased levels of mature thymocytes found in the thymus of CerS2-null mice, which could not exit due to the lack of a S1P gradient. Our study did not show significant changes in levels of mature thymocytes in CerS2-null mice (Figure 3A), which could be due to the different genetic backgrounds of the CerS2null mice used in the two studies. Irrespective, this is unrelated to the reduced levels of iNKT thymocytes in CerS2-null mice, which results from an arrest in maturation arrest at the DP stage (Figures 3C,D).
In conclusion, our data suggest that VLC-SLs (such as C24:1-GSLs) are endogenous self-antigens recognized by developing iNKT cells in the thymus and highlight the importance of parenchymal VLC-GSLs, which might also be required for iNKT cell survival in the liver.

eThics sTaTeMenT
This study was carried out in accordance with the recommendations of international guidelines. The protocol was approved by the Institutional Animal Care and Use Committee (IACUC).

aUThOr cOnTriBUTiOns
AS and YP-J performed and designed the experiments and wrote the manuscript. NSF performed the RNA seq and EF did the bioinformatics analysis. PS analyzed viral titers. AM and SK carried out ESI-MS/MS analyses of SLs. YJ performed in vitro iNKT cell activation experiments, and FT and CP helped write the manuscript. KL provided the LCMV-WE virus, infected mice, and helped write the manuscript. AHF supervised and funded the studies and wrote the manuscript.

acKnOWleDgMenTs
The authors thank Dr. Tomer Meir Salame from the FACS unit in the Weizmann Institute of Science for his excellent technical support. The authors also thank the NIH Tetramer Core Facility, Emory University, Georgia, USA for providing CD1d-PBS57 loaded tetramers. This study was supported by the Atheroflux consortium (EU grant FP7-602222-2), the Israel Science Foundation (grant 1728/15), and the National Institutes of Health (grant GM076217). AS was supported by the Planning and Budgeting Committee program for outstanding post-doctoral researchers. AHF is the Joseph Meyerhoff Professor of Biochemistry at the Weizmann Institute of Science.

sUPPleMenTarY MaTerial
The Supplementary Material for this article can be found online at http://www.frontiersin.org/article/10.3389/fimmu.2017.01386/ full#supplementary-material.   DaTa sheeT s1 | Raw RNAseq data and analysis of differentially expressed genes in livers isolated from wild-type (WT) and ceramide synthase 2 (CerS2)null mice with and without LCMV infection, and LCMV-infected CerS2-null mice after transfer of WT iNKT cells. The data have been uploaded to Gene Expression Omnibus (GEO), accession number GSE104205. reFerences