Role of ceramide synthase 2 in G-CSF signaling and G-CSF-R translocation into detergent-resistant membranes

Ceramides are sphingolipids with defined acyl chain lengths, which are produced by corresponding ceramide synthases (CerS1-6). In experimental autoimmune encephalomyelitis (EAE), an animal model of multiple sclerosis (MS), the ablation of CerS2 suppresses EAE-pathology by reducing neutrophil migration into the central nervous system. This migration is induced by granulocyte-colony stimulating factor (G-CSF) signaling. G-CSF signaling leads to a signal cascade including the phosphorylation of Lyn kinase and STAT3. This in turn regulates expression of the neutrophil surface receptor chemokine receptor 2 (CXCR2) and causes translocation of the receptor into detergent-resistant membranes (DRMs). In this study we investigated the role of ceramides in G-CSF signaling. We found, that G-CSF treatment of wild type bone marrow cells (BMCs) leads to translocation of G-CSF-receptor (G-CSF-R) into DRMs. G-CSF also induces downregulation of ceramides in WT and CerS2 null BMCs, as well as upregulation of very long chain lactosylceramides. However, in CerS2 null BMCs, G-CSF failed to induce translocation of G-CSF-R into DRMs, leading to reduced phosphorylation of Lyn and reduced CXCR2 expression. Interestingly, G-CSF signaling in CerS6 null BMCs was not affected. In conclusion, very long chain ceramides are important for G-CSF signaling and translocation of G-CSF-R into DRMs.

Multiple sclerosis (MS) is a neurodegenerative autoimmune disease characterized by the infiltration of immune cells into the central nervous system (CNS). The infiltrating immune cells release chemokines to recruit further immune cells and cytokines/inflammatory mediators to promote the inflammatory process, leading to the death of oligodendrocytes and subsequently to demyelination and neurodegeneration. A commonly used animal model for MS is experimental autoimmune encephalomyelitis (EAE), which has been used successfully for translation of drug candidates, including mitoxantrone, glatiramer acetate and natalizumab, into human therapeutic approaches 1 . Among other immune cells involved in the pathogenesis of EAE, neutrophils are the first to infiltrate into the CNS [2][3][4] . Migration into the CNS occurs after neutrophil expansion in the bone marrow and accumulation in the peripheral vascular system 3 .
Granulocyte colony-stimulating factor (G-CSF) induces migration of neutrophils by upregulation of the chemokine receptor CXCR2. The binding of G-CSF to its receptor (G-CSF-R) leads to the dimerization of the extracellular domain of the receptor that in turn activates Lyn kinase and Janus kinases (Jak). These kinases phosphorylate one or more tyrosine residues in the C-terminal region of the G-CSF-R leading to the activation of multiple intracellular signaling proteins, including the signal transducers and activators of transcription (STAT) and mitogen-activated protein (MAP) kinases [5][6][7] . This signaling pathway is also regulated by suppressor of cytokine signaling 3 (SOCS3) which is induced by STAT3 and inhibits the catalytic activity of Jak and thereby, the expression of STAT3. This results in negative feedback regulation 8,9 . Activated STAT3 translocates into the nucleus and induces the expression of several genes such as CXCR2 10 .  14). G-CSF protein levels were determined by ELISA. Pooled data are shown from three mice for each time point. The mRNA expression levels were normalized to peptidyl propyl isomerase A (PPIA) and were calculated using the mRNA level of untreated WT mice of the same age. Pooled data are shown from two (day 0/10/12/14) or three mice (day 4/6). The mRNA measurements were carried out in triplicate. Data are means ± SEM. *(p < 0.05) indicates significant differences within a group and # (p < 0.05) between the two groups. Statistical analysis was performed by Two-way ANOVA.
ScIEnTIFIc RepoRTS | (2019) 9:747 | DOI: 10.1038/s41598-018-37342-8 The expression of the G-CSF-R mRNA transiently increased in WBCs during the preclinical phase (day 0-day 6) (Fig. 1B). Also, G-CSF-R mRNA expression was similar in CerS2 null EAE mice and in WT EAE mice during the preclinical phase (Fig. 1B). After disease onset (day 10-day 14), G-CSF-R expression increased steadily in WBCs, whereas the G-CSF-R expression was significantly lower in CerS2 null cells compared to WT cells. However, these data indicate that neither the G-CSF nor the G-CSF-R levels are regulated by CerS2 in the preclinical phase of disease (day 0-day 6), which is critical for neutrophil migration into the CNS.

G-CSF-R translocates into DRMs after activation.
Some receptors such as the insulin receptor translocate into DRMs following activation. However, in CerS2 null hepatocytes, the insulin receptor is not able to translocate into the DRMs. This in turn impairs its phosphorylation 21 . Therefore, we investigated whether the activation of the G-CSF-R by G-CSF induces translocation into DRMs and whether the translocation depends on the presence of very long chain ceramides produced by CerS2. Because the lack of CerS6 increases the expression of the G-CSF-R and worsens EAE pathology 14 , we also investigated whether the activation of the G-CSF-R is CerS6 dependent. First, we identified the fractions that contain the DRMs in CerS2/6 null and WT bone marrow cells (BMCs) by determination of sphingolipid and cholesterol content. The DRMs from BMCs were isolated using sucrose gradient centrifugation. The DRMs were located predominantly in fraction two, as determined by the cholesterol content in BMCs from WT mice (Supplemental 2A,B). This was confirmed by the finding that sphingolipids such as C16-Cer, C18-Cer, C20-Cer and C24-Cer were also located mainly in fraction two ( Fig. 2A,B,E,F (0 h), Supplemental 3A-D, Supplemental 4) of CerS2 WT DRMs without G-CSF stimulation (0 h). In CerS2 null cells, the DRMs were also predominantly located in fraction two, as shown by a significantly higher level of C16-Cer, C18-Cer, C20-Cer and LacCer16 in fraction 2 in comparison to fraction 4 ( Fig To investigate whether the G-CSF-R translocates into the DRMs after activation by G-CSF, BMCs were incubated with G-CSF for 5 min and the protein levels of G-CSF-R were determined in each fraction by ELISA. The G-CSF-R protein levels were increased in fraction three in WT BMCs after activation by G-CSF, indicating the translocation of the G-CSF receptor into the DRM fraction (Fig. 3A). Interestingly, the G-CSF-R protein levels were increased in nearly all fractions in CerS2 null BMCs (Fig. 3B) and in fractions 1-3 in CerS6 null cells (Fig. 3C). These data imply that the G-CSF-R translocates into the DRM in WT and CerS6 null BMCs but not in CerS2 null cells.
Phosphorylation of Lyn kinase is regulated in a CerS2 dependent manner. We investigated whether the activation of the G-CSF-R is dependent on CerS2/6. The activation of the G-CSF-R leads to phosphorylation of the Lyn kinase and finally to STAT3 activation 22 . Therefore, we determined the phosphorylation status of the Lyn kinase and STAT3 in CerS2/6 null and WT BMCs upon G-CSF stimulation. G-CSF led to a transient phosphorylation of the Lyn kinase in WT and CerS6 null BMCs, whereas in CerS2 null BMCs the phosphorylation of the Lyn kinase was not induced by G-CSF (Fig. 4A). Furthermore, G-CSF led to a phosphorylation of STAT3 in WT BMCs. Unexpectedly, in CerS2 null BMCs, the phosphorylation of STAT3 was not altered, whereas in CerS6 null BMCs the phosphorylation of STAT3 was slightly, but not significantly reduced, compared to WT cells (Fig. 4B). These data indicate that the phosphorylation of Lyn is CerS2-dependent, whereas STAT3 seems to be independent of regulation by CerS2/6. Expression of CXCR2 is regulated by CerS2. It is known that G-CSF stimulates the expression of CXCR2 via STAT3 in neutrophils 10 . However, to our knowledge it is unknown if Lyn kinase signaling plays a role in CXCR2 expression in neutrophils. Therefore, we investigated whether Lyn kinase also influences CXCR2 expression. We also checked whether an inhibitor of the Lyn kinase (PP1) can prevent the upregulation of CXCR2 on neutrophils isolated from BMCs. The cells were treated with G-CSF for 24 h or 46 h in the presence or absence of the inhibitor. G-CSF induced upregulation of CXCR2, whereas PP1 prevents the increase of CXCR2 expression after 24 h (Fig. 5A). To exclude that the observed effect is due to apoptosis induction, we investigated whether PP1 induces apoptosis in BMCs. However, PP1 did not increase apoptosis rate in BMCs (Supplemental 3E). These data indicate that CXCR2 expression is regulated via Lyn signaling. As our results demonstrate, ablation of CerS2 prevents phosphorylation of Lyn kinase after stimulation with G-CSF. Thus, we expected that the CXCR2 expression would be reduced in CerS2 null BMCs. As expected, G-CSF induced a time dependent expression of CXCR2 mRNA in WT BMCs, predominantly after 6 h (Fig. 5B). In CerS2 null BMCs, G-CSF-induced CXCR2 mRNA expression was significantly reduced at this time point (Fig. 5B). Protein expression was significantly reduced after 24 h and 46 h, a finding we observed previously 2 .

SOCS3 expression is not
Translocation of CXCR2 is CerS2 dependently regulated. Since CXCR2 is a membrane receptor, we examined whether CXCR2 translocates into the DRMs and whether this depends on ceramide status. BMCs isolated from C57BL/6 mice were treated with G-CSF for different time periods (6 h, 16 h, 24 h, 30 h and 46 h). In unstimulated BMCs, the CXCR2 was primarily localized in fraction 10, whereas the activation with G-CSF led to an increase of CXCR2 in fraction 6 after 24 h. This was followed by an increase of CXCR2 in the next 22 h, in fractions 2 and 3, accompanied by a decrease in CXCR2 in fraction 6, indicating that G-CSF induces the expression of CXCR2 and leads to its translocation into DRM fractions 2 and 3 ( Fig. 5C). Therefore, CerS2/6 WT and null BMCs were investigated upon CXCR2 translocation into the DRMs after 46 h of G-CSF stimulation. After stimulation, WT BMCs exhibited an increased level of CXCR2 in DRM fractions two and three (Fig. 5D,E). In G-CSF activated CerS2 null cells, CXCR2 was not increased (Fig. 5F), whereas CXCR2 in G-CSF activated CerS6 null cells was localized in fraction two and three (Fig. 5G). These data indicate that G-CSF induced expression and translocation of CXCR2 is regulated in a CerS2 dependent manner, whereas CerS6 seems to have no effect on CXCR2 expression and translocation into DRMs.

G-CSF induces alteration of DRM composition. Recently, we observed that G-CSF treatment induces
CerS2 and CerS6 expression 2,14 . Therefore, we investigated whether this translates into an altered ceramide or cholesterol content of the DRMs. After 46 h G-CSF treatment cholesterol was found in fraction 2 or 3 in WT and CerS6 null cells, whereas in CerS2 null cells cholesterol was found in each fraction (Supplemental 2A,B). Surprisingly, C16-Cer and C24-Cer were reduced and C24-lactosylceramide (LacCer) was increased in DRMs isolated from G-CSF stimulated CerS2 WT BMCs after 46 h ( Fig. 2A-D). In DRMs isolated from CerS2 null cells, C16-Cer was reduced. In addition C24-LacCer and C16-LacCer were increased after G-CSF stimulation ( Fig   observation that CerS2 null cells lacking these sphingolipids are characterized by a loss of DRMs and of impaired G-CSF-R and CXCR2 translocation ( Fig. 3 and 5D-G).

G-CSF regulates CerS2 expression transcriptionally via Lyn.
To gain further insight into the transcriptional expression of CerS2 in response to G-CSF stimulation, a promoter luciferase reporter gene assay was performed. Five genomic DNA fragments, located upstream of the transcription start site of the murine CerS2 gene were cloned into the pGL3-plasmid (detailed promotor construct composition see Supplemental 1A). Transfection with the putative promoter constructs prom2_I/prom2_IV failed to induce luciferase activity in RAW247 cells (Fig. 6A). Prom2_II/prom2_V lead to moderate induction of luciferase activity. However, the transfection with the putative promotor prom2_III and subsequent stimulation of the cells with G-CSF resulted in the most distinct G-CSF-dependent induction of luciferase activity (Fig. 6A). These data indicate that the murine promotor of CerS2 is located on exon 1 and intron 1. Moreover, these data suggest that G-CSF inducible transcription factors also bind on exon 1 and/or intron 1. Next, we sought to establish whether Lyn signaling contributes to the activation of the CerS2 promotor. Therefore, RAW247 cells were transfected with pGL3-prom2_III and subsequently treated for 46 h with G-CSF in the presence or absence of PP1. Interestingly, PP1 prevented the G-CSF induced luciferase activity indicating that Lyn kinase signaling is involved in the induction of CerS2 expression (Fig. 6B). To exclude the possibility that reduced cell viability was responsible for the reduced luciferase activity, the viability of transfected cells treated with PP1 was assessed. PP1 only slightly reduced cell viability after 46 h and is therefore unlikely to be responsible for the observed reduction of CerS2 expression (Supplemental 3F).

G-CSF did not regulate CerS6 expression transcriptionally.
A genomic DNA fragment (prom6), located upstream of the start codon of the murine CerS6 (Supplemental 1B) gene, was cloned into a pGL3 vector. Transfection with the putative CerS6 promotor prom6 increased luciferase activity in RAW247 cells indicating that the DNA fragment represents at least a part of the CerS6 promotor (Fig. 6C). However, G-CSF was not able to further increase the luciferase activity in prom6 transfected RAW247 cells. These data indicate that G-CSF did not induce CerS6 expression transcriptionally (Fig. 6C).

Discussion
Ceramides have been shown not only to provide structural integrity to cell membranes but to play an important role as second messengers in cell signaling. They are involved in crucial physiological processes such as cell differentiation, growth and apoptosis. Ceramides also regulate cell signaling processes in an acyl chain dependent manner. Our data revealed that the presence of very long chain ceramides is essential for G-CSF signaling. In detail, the lack of very long chain ceramides prevents G-CSF-induced translocation of the G-CSF-R into DRMs, phosphorylation of Lyn kinase and CXCR2 expression leading to reduced potential for neutrophil migration (Fig. 7). Accordingly, transcriptional CerS2 expression was induced by G-CSF. In contrast, long chain ceramides do not seem to influence G-CSF signaling in our study. However, we are unable to absolutely exclude the involvement of long chain ceramides in this study, since C16 ceramides have been shown to be increased in CerS2 null mice 18 .
Lipid rafts are defined as microdomains within the lipid bilayer of cellular membranes that assemble proteins and lipids (cholesterol and sphingolipids such as ceramides) and experimentally resist extraction in cold detergent (detergent-resistant membrane) 25 . In WT BMCs, DRMs consist of a mixture of ceramides and ceramides with sugar side chains (i. e. LacCer) of different chain lengths. In CerS2 null BMCs, DRMs also consist of ceramides and glucosylceramides but with almost no contribution of sphingolipids with a C24-and C24:1-Cer backbone. The lack of very long chain ceramides does not prevent the formation of DRMs but alters the composition of the DRM and the fraction location (at least in hepatocytes 21 ) and thereby the biophysical properties of the membrane. The lack of very long chain ceramides in CerS2 null cells leads to a more fluid 26 membrane and might influence cell signaling. For example the incorporation of 2-hydroxylated fatty acids (2OHFAs) into the cell membrane results in decreased lipid order and dipole potential as well as less rigid packing of acyl chains and hence, modulates membrane proteins by microdomain reorganization and membrane fluidity 27 . Furthermore, activation with G-CSF and thereby the switch to neutrophils leads to an increase in C24-LacCer in WT cells, whereas in CerS2 null cells, C16-LacCer and C24-LacCer were increased. This unexpected C24-LacCer increase in CerS2 null BMCs might be produced by CerS3 or CerS4, which are also capable of very long chain ceramide synthesis 17,28 . Also, in the dextran sodium salt (DSS) evoked model of ulcerative colitis, an increase in very long chain ceramides in CerS2 null mice was shown, accompanied by an increase in CerS3 mRNA expression 29 . The observed decrease in ceramides after G-CSF induction in CerS2 WT and null BMCs is possibly due to the fact that ceramides are precursors for the glucosylceramides. Additionally, the G-CSF induced altered membrane composition prevented the formation of DRMs in CerS2 null cells, as indicated by the lack of cholesterol accumulation in DRM-fractions (Supplemental 2A,B). However, this is in contrast to findings from Park and co-workers, who observed a shift of DRMs from fraction 1-3 in WT hepatocytes to fraction 4-6 in CerS2 null hepatocytes 21 .
The altered sphingolipid composition, the more fluid membrane and the lack of DRM in G-CSF-activated cells are possibly responsible for the lack of G-CSF-R translocation and the impaired G-CSF signaling, such as reduced Lyn phosphorylation and reduced CXCR2 expression. The importance of sphingolipid composition and altered Dashed bars indicate DRM fractions. *(p < 0.05), **(p < 0.01) and ***(p < 0.001) indicate significant differences between DMSO-treated and PP1 treated WT cells (D,E), between WT cells and CerS2 null (F) or CerS6 null cells (G) or differences between time points in C57BL/6 DRMs analyzed by Two-way ANOVA (A-C) and t-test (D-G).
ScIEnTIFIc RepoRTS | (2019) 9:747 | DOI:10.1038/s41598-018-37342-8 membrane properties for the function of membrane receptors has already been observed by others. CD36 which is responsible for the uptake of triglycerides and the insulin receptor did not translocate into DRMs in CerS2 null hepatocytes and consequently loose their functions 21,30 . The gap junction protein connexin 32 displays a mislocalization in CerS2 null hepatocytes due to disruption of DRMs 31 . For FcγRIIB, it was shown that exclusion from DRMs prevents interaction with signaling molecules 32,33 and therefore, possibly their inhibitory potential in immune regulation. In CerS2 null hepatocytes, TNFR1 is not internalized and inhibits selective pro-apoptotic downstream signaling for apoptosis 34 . We cannot exclude that ceramides synthesized by sphingomyelinase (SMase) also contribute to the observed effects, since ceramides generated by SMase enhance Tyr-phosphorylation of Jak2 and subsequently STAT3 phosphorylation in fibroblasts 35 . The existence of cholesterol and sphingolipid rich microdomains is a prerequisite for this activation. It is also possible that ceramides directly interact with proteins and thereby, induce dephosphorylation of Jak2 or Lyn kinase. For example, C18 ceramide activates protein phosphatase-2A 36 which in turn, can dephosphorylate Jak2 37 .
As it has already been shown by others, our data confirm that G-CSF induces the Lyn signaling cascade and leads to activation of STAT3 38,39 . G-CSF also induced CerS2 promotor activity. To identify which transcription factor, activated by G-CSF signaling, was responsible for CerS2 promotor activity, we analyzed five different upstream regions of the start codon of CerS2. Promotor construct prom2_III was able to function as a promotor. Prom2_III comprises exon 1 and intron 1. To narrow down the putative binding site, we took into account that construct prom2_IV lacks the first 643 Bp of intron 1 and has no promotor activity. These data suggest that the transcription factor binds to intron 1 at the first 643 Bp and/or to exon 1. This is in line with the findings of Gong et al. who identified exon 1 as a regulatory region for gene transcription in human CerS2 and as binding site of the transcription factor SP1, which interacts with KLF6 40 . Analysis of the putative promotor region DNA fragment of CerS2 including exon 1 and 700 Bp from intron 1 by the "Promo program" (maximum matrix dissimilarity rate: 5%) 41,42 suggests several putative transcription factor binding sites. This includes CCAAT/enhancer-binding protein C/EBPα, C/EBPβ, IRF-1 and STAT5 which are induced by G-CSF or belong to G-CSF-responsive genes 22 , but no STAT3 binding site. These data indicate that CerS2 mRNA expression is not induced via STAT3 but via a yet unknown transcription factor. Previously, we demonstrated that the genetic deletion of CerS2 ameliorates clinical symptoms in EAE mice. We showed that the ameliorated clinical symptoms in CerS2 null mice were due to a reduced potential for migration of neutrophils that express reduced CXCR2 protein levels. In this study, we showed that in CerS2 null cells, translocation of the G-CSF-R is prevented leading to impaired receptor function and subsequently, to reduced expression of CXCR2. We assume that besides the G-CSF-R, other DRM-located proteins that regulate the immune response or inflammatory processes can be influenced by the alteration of DRM properties in CerS2 null mice. Further insight into the influence of chain-length specific ceramides on cell signaling is therefore important to better understand their involvement in diseases such as cancer and autoimmune diseases and to pinpoint possible therapeutic targets.

Material and Methods
Cells and reagents. BMCs were cultured and stimulated in RPMI 1640 GlutaMAX medium containing 2.5% hormone-free fetal calf serum (FCS) and 1% penicillin/streptomycin. RAW247 macrophages were cultured in RPMI 1640 medium -GlutaMAX containing 10% FCS. For the reporter gene assay, RAW247 macrophages were incubated in RPMI 1640 medium -GlutaMAX containing 2.5% FCS (charcoal stripped or normal). Cells were cultured at 37 °C in an atmosphere containing 5% CO 2 . G-CSF was purchased from BioLegend (Fell, Germany).

Animal models.
In all experiments, the ethics guidelines for investigations in conscious animals were followed and the experiments were approved by the local Ethics Committee for Animal Research (Regierungspräsidium Darmstadt, F152/03). Unless otherwise stated, mice for all experiments were female, between 8-14 weeks of age. EAE mouse model. 10-to 13-week-old female 129S4/SvJae × C57BL/6 mice (F1 WT or F1 CerS2 null) were used for induction of EAE. The induction was achieved as recommended by the supplier using an EAE kit of Hooke Laboratories (Lawrence, USA). The induction of the EAE model has been described previously 2 . Untreated mice were used as the control group.

FACS analysis.
BMCs were treated with 10 ng/ml recombinant mouse G-CSF for 24 h and 46 h to check for CXCR2 surface expression on neutrophils. One hour before G-CSF treatment, the cells were incubated with 10 µM Lyn inhibitor PP1 (Tocris Bioscience, Bristol, UK) and then with an antibody cocktail consisting of CD11b-eFluor450 (eBioscience, Frankfurt, Germany), Ly6G-APC-Cy7 (BD, Heidelberg, Germany) and CXCR2-APC (BioLegend, Fell, Germany) for 15 min at room temperature. Samples were acquired with a MACSQuant Analyzer 10 flow cytometer (Miltenyi Biotec, Bergisch Gladbach, Germany) and analyzed using FlowJo software v10 (Treestar, Ashland, USA). All antibodies were previously titrated to determine optimal concentrations. Antibody-capturing CompBeads (BD, Heidelberg, Germany) were used for single-color compensation to create multi-color compensation matrices.
Reporter gene assay-cloning of putative CerS2 constructs. The putative promoter region of murine CerS2 was assumed to lie upstream of exon 1, in exon 1 or in intron 1, because the ATG lies in exon 2 (Supplemental 1A). Various promotor constructs (prom2_I-V) from the putative promoter regions (Supplemental 1A) of CerS2 gene were amplified by PCR using genomic DNA from murine BMCs. Primers, used for the PCR, contained an identification restriction site ( Reporter gene assay-cloning of putative CerS6 constructs. The putative promoter region of the murine CerS6 was assumed to lie upstream of exon 1 or in exon 1, because the ATG lies in exon 1 (Supplemental 1B). One promotor construct (prom6) from the putative promoter region (Supplemental 1B) of the CerS6 gene was amplified by PCR using genomic DNA from murine BMCs. Primers used for the PCR contained an identification restriction site (  Table 1. Primer for cloning strategy of CerS2-promoter constructs (prom2_I-V) and CerS6 promotor construct (prom6).
constructs and 2.5% normal FCS for CerS6 promotor constructs. The cells were transfected with 125 ng of the distinct firefly luciferase reporter vector (pGL3 Basic vector, pGL3 prom2_I-V or prom6 vector) and 12.5 ng of the Renilla luciferase control reporter vector (pRL-TK, Promega, Fitchburg, USA) using 0.5 μl MACSfectin reagent. One day after transfection, the cells were incubated for 46 h with or without 10 ng/ml G-CSF. Statistics. Results are presented as means ± SEM (standard error of the mean) or ± SD (standard deviation).
Significant differences in multiple data sets were analyzed by Two-way ANOVA with Bonferroni post hoc-test or alternatively with Tukey's or Sidak's multiple comparisons test. Significant differences between two groups were analyzed by t-test (Graphpad Prism 6 software). The level of significance was set at p < 0.05.

Data Availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files).