Mice with floxed GCS (Ugcg, EC 2.4.1.80) alleles were described previously (48). TCRVα14-Jα281 transgenic mice were kindly provided by Agnes Lehuen (49). CD1d-deficient (50) and Vav^Cre-transgenic (51) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). All strains were backcrossed for more than 10 generations to the C57BL/6 genetic background (Charles River Wiga, Sulzfeld, Germany) and housed under specific pathogen-free conditions. Vav^Cre-negative littermates were used as wild-type (WT) controls. Animal experiments were performed in compliance with the German guidelines on animal protection.

Single cell preparations from organs were prepared as described previously (52). Flow cytometry was performed as described in Ref. (53). The following monoclonal antibodies were used: anti-CD1d (clone: 1B1); anti-CD3ε (145-2C11), anti-CD4 (GK1.5), anti-CD8 (53-6.7), anti-CD11c (HL3), anti-CD19 (MB19-1), anti-CD25 (PC61.5), anti-CD44 (IM7), anti-CD150/SLAM (9D1), anti-Ly108 (13G3), anti-MHCII (M5/144.15.2), anti-NK1.1 (PK136), anti-TCR-Vβ2 (B20.6), anti-TCR-Vβ7 (TR310), and anti-TCR-Vβ8.1 and 8.2 (MR5-2) from BD, Heidelberg, Germany, Biolegend, San Diego, CA, USA, and eBioscience, San Diego, CA, USA. PBS57-loaded PE-labeled CD1d tetramers were kindly provided by NIH Tetramer Core Facility at Emory University (Atlanta, GA, USA). BrdU and Annexin V experiments were performed according to the manufacturer’s protocol (both BD). Analysis of flow cytometry data was performed using Cell Quest Pro software (BD) and FlowJo (Tree Star, Flow Cytometry Analysis Software) by gating on lymphocytes in the forward and side scatter. Double-positive thymocytes were sorted using FACSAria™ (BD) by gating stringently on CD4⁺/CD8⁺ DP lymphocytes and excluding 7AAD-positive dead cells.

RNA was extracted from cell pellets using the phenol/chloroform extraction method (54) followed by digestion by RNase-free DNaseI (turbo DNA free, Ambion, Huntingdon, UK). A total of 3 µg of total RNA were reverse transcribed in 20-µl total volume using SuperscriptII (Invitrogen, Karlsruhe, Germany) according to the manufacturers’ instructions. RT-PCR was performed with 1 µl cDNA and GCS primers: forward 5′—gat cta aga ggg tga agg cgc a—3′ and reverse 5′—ctg cct tgc aat cct gtc tgt c—3′.

Glycosphingolipids were extracted from lyophilized cell pellets as described in detail in Ref. (55, 56). For thin layer chromatography (TLC) analysis, an amount corresponding to 0.2 mg protein was loaded on a TLC plate (Merck, Darmstadt, Germany). Running solvent was CHCl₃/CH₃OH/H₂O (62.5:30:6, v/v/v) for neutral GSL and CHCl₃/CH₃OH/0.2% CaCl₂ in H₂O (60:35:8, v/v/v) for acidic GSL, respectively. Sialidase treatment was performed as described in Ref. (57). 0.05 U Vibrio cholerae sialidase in 0.2 M Na-acetate buffer, 2 mM CaCl₂, pH 5.2, was used to digest acidic GSL on a polyisobutylmethacrylate-fixed TLC plate at room temperature for 8 h.

Sphingolipids from DP thymocytes were extracted as previously described with slight modifications (58). Briefly, sorted thymocytes (~5 × 10⁶) were dried with 1-propanol and extracted twice at 37°C for 15 min with a chloroform/methanol/water mixture of 10/10/1 (v/v/v) and once with 30/60/8. The residual cell pellets were used for protein determination according to the Lowry method. The combined lipid extracts were dried under air flow and subsequently subjected to mild alkaline hydrolysis with 0.1 M potassium hydroxide in methanol for 2 h at 37°C. Saponified extracts were finally desalted by reverse-phase (C18) column chromatography. Aliquots corresponding to 30 µg of protein were dissolved in 1 ml 95% methanol containing the following internal standard mixture: Cer (d18:1;14:0), Cer (d18:1;19:0), Cer (d18:1;25:0), Cer (d18:1;31:0) each 4 pmol; GlcCer (d18:1;14:0), GlcCer (d18:1;19:0), GlcCer (d18:1;25:0), and GlcCer (d18:1;31:0) each 2 pmol.

For quantification of lipid extracts, UPLC–ESI–MS/MS analyses were performed as described in Ref. (59) with following modifications: lipid extracts were separated in a reverse-phase (C18) column, which was kept at 45°C, while the autosampler was maintained at 15°C. After equilibration with buffer A (95% methanol, 0.05% formic acid, and 1 mM ammonium formate), lipids were eluted with increasing percent of buffer B (99% 2-propanol, 1% methanol, 0.05% formic acid, and 1 mM ammonium formate) up to 85%. Ceramide and hexosylceramide species were detected with a precursor ion scan of m/z +264 corresponding to sphingosine (d18:1) while keeping the cone voltage at 50 V and the collision energy at 44 eV.

Double-positive thymocytes were isolated from WT, Vav^CreGCS^f/f, and CD1d^–/– thymi after depleting cells reactive with PBS57-loaded CD1d tetramers. 0.5 × 10⁶ DP thymocytes per well were placed in 96-well-plate and incubated with αGalCer (Avanti Polar Lipids, Alabaster, AL, USA) at indicated concentrations. iNKT cells were enriched from livers of TCRVα14-Jα281 transgenic mice using anti-CD5 micro beads (Miltenyi Biotec) and applied at 50,000/well. Activation of T cells in vitro was performed as described in Ref. (60). Briefly, splenic T-cells were enriched by anti-CD90.2 micro beads (Miltenyi Biotec) and incubated with 0.5 mg/ml calcium ionophore A23187 and 10 ng/mL phorbol 12-myristate 13-acetate (PMA, both Sigma). Supernatants were collected after 18 h and analyzed for IFNγ and IL4 concentrations by cytometric bead array technique (BD). For the in vivo testing of iNKT cells function, mice were injected i.p. with 0.2 or 3 µg αGalCer and sacrificed 8 h later.

Thymocytes of WT and Vav^CreGCS^f/f mice were enriched by magnetic separation using CD5 beads (Miltenyi), spinned down using cytospin system (4 × 10⁵ cells/slide) and fixed in 1% paraformaldehyde in PBS for 15 min at room temperature. Cells were then incubated with antibodies against CD1d-FITC (BD), early endosome antigen 1 (EEA1) (Cell Signaling), Rab7 (Santa Cruz Biotechnology), and lysosome-associated membrane protein 1 (LAMP1) (eBioscience). After washing, corresponding Alexa-Fluor 546-conjugated secondary antibodies were added and the slides were incubated for 1 h at room temperature in the dark. DAPI was used for nuclear visualization. Negative controls contained DAPI staining and Alexa-Fluor 546-conjugated secondary antibodies (for EEA1, Rab7, and LAMP1), or DAPI staining only (for CD1d). Images were acquired using the Olympus IX81 motorized microscope equipped with the MT20 illumination system; the Cy3, GFP, and DAPI HC-Filter sets; and Hamamatsu Orca-ER CCD camera. Two hundred images from each channel were acquired for each region using the 100×/1.4 PlanApo objective, and five regions were analyzed from each sample. Images were post-processed with ImageJ (http://rsbweb.nih.gov/ij) to obtain super-resolution optical fluctuation images—www.ncbi.nlm.nih.gov/pubmed/20018714. On average, 20 cells were analyzed for co-localization between red and green using the ImageJ’s co-localization plugin, and the ratio of co-localized and total green area was plotted and statistically analyzed. The images of lysosomes were further analyzed automatically with the same parameters using ImageJ macro developed at DKFZ Light Microscopy Core Facility (Heidelberg, Germany). Shortly, images of lysosomes were thresholded and segmented using the Find Maxima tool with the Segmented Particles above lower threshold option activated. The segmented particles above the minimum area limit of 10 pixels (0.022 µm²) were further counted for each cell using ImageJ’s Analyze Particles tool.

Unpaired two-tailed Student’s t-test was performed to compare data sets. Differences were considered significant if p < 0.05. Numbers of independent observations per group are indicated for each result.

Because the development of iNKT cells depends on presentation of lipid antigens on DP thymocytes, we analyzed the latter cell population for its GSL composition in WT mice. In the neutral GSL fraction, hexosylceramides represented the major fraction (Figure 2A). In the acidic fraction, two compounds running at the height of GM1 and GD1, respectively, emerged (Figure 2A). In order to further characterize these substances, the acidic fraction was digested by neuraminidase and the products were subsequently separated into acidic and neutral fractions (Figure 2B). The hereby obtained acidic compound ran at the height of the GM1 standard suggesting that this band probably consisted of a mixture of non-digestible GM1 or was derived from GD1b after the release of the terminal sialic acid. By contrast, the neutral compound was not represented in the standards; however, due to its running properties, it likely corresponded to neutral ganglioside GA1 (Figure 2B). To identify this product, the original acidic fraction was on-plate digested with neuraminidase and subsequently immunostained using anti-GA1 antibodies (Figure 2C). Both the upper and the lower compounds had a neutral backbone of GA1 (Figure 2C, left panel). Based on the running properties and the comparison with the standards, these compounds likely correspond to GM1a, GM1b, GD1b, and GD1c.

Although several lines of evidence have implicated that GlcCer-based GSL might belong to the lipid antigens relevant for the iNKT cell development, a direct proof of this hypothesis was precluded by the early embryonal lethality of GCS-deficient mice (47). To overcome this problem, we implemented a tissue-specific deletion of this gene. To this end, mice with floxed alleles of the GCS gene (GCS^f/f) were crossed with mice expressing iCre under the control of the Vav-promoter (VavCre) that activates the recombinase activity very early in T-cell development with virtually 100% of DN1 (CD25⁻/CD44⁺) thymocytes being already positive (51).

In terms of TLC analysis, no GSL could be detected in DP thymocytes from Vav^CreGCS^f/f mice (Figure 3A). This was in line with extensive and significant reduction of GCS mRNA in these cells (Figure 3B). Mass spectrometry performed on FACS-sorted DP thymocytes revealed a 99.6% reduction of the GlcCer content in Vav^CreGCS^f/f mice as compared to WT (Figure 3C). This reduction occurred independently of the analyzed acyl moiety (Figure 3D). By contrast, the ceramide content of Vav^CreGCS^f/f DP thymocytes was indistinguishable from WT (Figures 3E,F).

Vav^CreGCS^f/f mice reproduced normally and progeny were born at expected Mendelian ratios (data not shown). Newborn and adult Vav^CreGCS^f/f mice did not exhibit any overt growth, developmental or behavioral defects. Body weight, the weight, and cellularity of thymus and spleen were indistinguishable from Vav^Cre-negative littermates (Figure 4A). Similarly, no aberration in the maturation of the conventional thymocytes could be revealed by flow cytometry in Vav^CreGCS^f/f mice (Figure 4B). In spleens, the amount of CD3- and CD19-positive T- and B-lymphocytes, respectively, was unaffected by the deletion of the GCS gene (Figure 4C). The expression levels of CD1d on Vav^CreGCS^f/f DP thymocytes and splenic CD11c⁺/MHCII⁺ dendritic cells were indistinguishable from WT mice (Figures 4D,E). The expression of SLAM (CD150) and Ly108 molecules, which provide important signals on DP thymocytes during the thymic iNKT cell development (61), did not significantly differ between Vav^CreGCS^f/f and WT mice (Figures 4F,G).

Furthermore, CD1d trafficking was analyzed using super-resolution microscopy and EEA1, Rab7, and LAMP1 as markers of early endosomes, late endosomes and lysosomes, respectively (Figures 5A–C). The quantification of signal co-localization revealed a statistically significant shift of CD1d from late to early endosomes in Vav^CreGCS^f/f mice. In contrast, the CD1d amount in lysosomes remained unaffected (Figure 5D). Although a tendency toward less but larger lysosomes could be seen in DP thymocytes of Vav^CreGCS^f/f mice, the difference was not statistically significant (Figure 5E).

In newborn and adult mice, the iNKT cell populations were characterized by flow cytometry using PBS57-loaded CD1d tetramers. Adult Vav^CreGCS^f/f mice showed a significant reduction of the iNKT population in terms of absolute numbers and percentages in thymus, spleen, and liver as compared to WT littermates (Figure 6A). In all three organs, a reduction of the iNKT cell population by approximately 50% could be observed. In Vav^CreGCS^f/f mice, the remaining iNKT cells could be clearly identified and discerned from any unspecific staining as visualized by comparison with CD1d-deficient mice that do not produce iNKT cells (Figure 6A). Newborn mice showed identical results (data not shown). To test for possible unspecific effects of the Vav^Cre transgene, iNKT cell frequencies and absolute numbers were compared between Vav^Cre-positive and Vav^Cre-negative GCS^+/+ mice showing no statistically significant differences (Table S1 in Supplementary Material).

In course of thymic maturation, iNKT cells upregulate expression of NK1.1 and CD44 allowing the identification of three developmental stages: immature, CD44⁻/NK1.1⁻; semi-mature, CD44⁺/NK1.1⁻; and mature, CD44⁺/NK1.1⁺ (62). In terms of absolute numbers, iNKT cells were significantly reduced in all three developmental stages in Vav^CreGCS^f/fmice. However, no significant difference was observed in the percentual distribution among these three stages (Figure 6B).

In iNKT cells, the invariant Vα14-chain pairs almost exclusively with Vβ2, 7, or 8.2 (63). We tested whether the depletion of GCS-derived GSL would lead to a shift of the Vβ-chain repertoire in Vav^CreGCS^f/f mice. However, no statistically significant difference in the percentage distribution of the Vβ-chains could be found between Vav^CreGCS^f/f mice and WT littermates. In terms of absolute numbers, a decrease corresponding to the diminished iNKT cell population could be observed (Figure 6C).

Measurements of proliferation and apoptosis rate by BrdU incorporation and Annexin V staining, respectively, revealed that in Vav^CreGCS^f/f mice, thymic iNKT cells showed a significantly reduced proliferation and an increased apoptosis as compared to WT mice. By contrast, conventional thymocytes were unaffected (Figures 6D,E).

Depletion of GCS-derived GSL in DP thymocytes might not only alter the repertoire of lipid antigens but also impact the processes of their presentation. Thus, in order to test the antigen presenting capacity of Vav^CreGCS^f/f DP thymocytes, these cells were exposed to increasing concentrations of the exogenous antigen αGalCer and co-incubated with WT responder iNKT cells enriched from livers of TCRVα14-Jα281 transgenic mice. As measured by secretion of IFNγ and IL4, no statistically significant difference could be observed between the antigen presentation on DP thymocytes from Vav^CreGCS^f/f and WT mice (Figures 7A,B).

Furthermore, we have subjected Vav^CreGCS^f/f iNKT cells to functional tests in vivo and in vitro. Upon injection of αGalCer, upregulation of CD69 on iNKT cells was unaltered in Vav^CreGCS^f/f mice (Figure 7C). IFNγ levels were significantly lower in Vav^CreGCS^f/f mice injected with 3 µg αGalCer. Similarly, IL4 levels tended to be lower in the Vav^CreGCS^f/f mice although a statistical significance was not reached (Figure 7C). Therefore, we tested the reactivity of Vav^CreGCS^f/f iNKT cells in vitro with equalized cell numbers. To this end, WT DP thymocytes were loaded with αGalCer and co-incubated with iNKT cells enriched from spleens and livers of Vav^CreGCS^f/f or WT mice. No functional deficiency could be observed between iNKT cells from Vav^CreGCS^f/f and WT mice as measured by IFNγ secretion (Figure 7D).

In line with the latter result, the general T cell population of Vav^CreGCS^f/f mice was unaffected and showed an unaltered production of IFNγ in response to TCR-independent (PMA/calcium ionophore) or TCR-dependent (CD3/CD28) stimulation (Figure 7E).