Sphingosine-1-phosphate Lyase Deficiency Produces a Pro-inflammatory Response While Impairing Neutrophil Trafficking*

Sphingosine-1-phosphate (S1P) lyase catalyzes the degradation of S1P, a potent signaling lysosphingolipid. Mice with an inactive S1P lyase gene are impaired in the capacity to degrade S1P, resulting in highly elevated S1P levels. These S1P lyase-deficient mice have low numbers of lymphocytes and high numbers of neutrophils in their blood. We found that the S1P lyase-deficient mice exhibited features of an inflammatory response including elevated levels of pro-inflammatory cytokines and an increased expression of genes in liver associated with an acute-phase response. However, the recruitment of their neutrophils into inflamed tissues was impaired and their neutrophils were defective in migration to chemotactic stimulus. The IL-23/IL-17/granulocyte-colony stimulating factor (G-CSF) cytokine-controlled loop regulating neutrophil homeostasis, which is dependent on neutrophil trafficking to tissues, was disturbed in S1P lyase-deficient mice. Deletion of the S1P4 receptor partially decreased the neutrophilia and inflammation in S1P lyase-deficient mice, implicating S1P receptor signaling in the phenotype. Thus, a genetic block in S1P degradation elicits a pro-inflammatory response but impairs neutrophil migration from blood into tissues.

Sphingosine 1-phosphate (S1P) 3 is a sphingolipid signaling molecule that exerts important physiologic functions through its interaction with a family of G protein-coupled receptors (S1P1-5) (1)(2)(3)(4). S1P is synthesized by the phosphorylation of sphingosine by either of two sphingosine kinases, Sphk1 and Sphk2 (5,6). After its formation, S1P may be either dephosphorylated back to sphingosine by the action of two specific S1P phosphatases, Sgpp1 and Sgpp2, or permanently degraded by the S1P lyase Sgpl1 to the nonsphingolipid substrates hexadecenal and phosphoethanolamine (7). Alternatively, S1P can be exported out of the cell where it is able to interact with the S1P receptors (5,8). S1P is found highly enriched in the circulation-in both blood and lymph-while its concentration in tissues remains very low by comparison, as the result of the combined synthetic and degradative activities involved in the S1P biosynthetic pathway (9). Blood S1P is produced by erythrocytes (10 -12), but other cells such as mast cells and platelets can also secrete S1P (13,14). Endothelial cells can contribute to the circulating S1P pool (15,16) and are likely responsible for the S1P that is present in the lymph (12).
The proper compartmentalization of S1P in circulation and in tissues is important for the trafficking and positioning of lymphocytes. The egress of lymphocytes out of primary and secondary lymphoid organs is dependent on S1P receptors on lymphocytes (17)(18)(19)(20)(21)(22)(23), which recognize the higher concentrations of S1P around exit points leading to the blood and lymph (12,24,25). When the activity of S1P lyase is inhibited or deleted as in the S1P lyase-knock-out (Sgpl1 Ϫ/Ϫ ) mice, compartmental S1P concentrations are altered, blocking lymphocyte egress and resulting in lymphophenia (24,26,27).
In S1P lyase-deficient mice, neutrophils are highly elevated in blood, in contrast to their low lymphocyte numbers (26). Here we report that the S1P lyase-null mice have an elevated pro-inflammatory response with impaired migration of neutrophils into tissues resulting in an abnormal neutrophil homeostatic regulatory loop. These results implicate S1P lyase activity as a regulator of inflammatory responses and neutrophil trafficking.
Bone Marrow Transplantation-Total bone marrow cells were isolated from Sgpl1 ϩ/ϩ and Sgpl1 Ϫ/Ϫ mice by flushing the femur and tibia from both legs two times with 1 ml of PBS. Cells were injected i.v. into lethally irradiated Rag2 Ϫ/Ϫ mice (Taconic, Germantown, NY). Transplanted mice were analyzed 8 weeks after the procedure.
Thioglycollate-induced Peritonitis-Sgpl1 ϩ/ϩ and Sgpl1 Ϫ/Ϫ mice were injected intraperitoneal with 300 l of 4% thioglycollate (Sigma-Aldrich). After 4 h, mice were euthanized and blood collected by heart puncture. Cells that were recruited into the peritoneal cavity were collected by lavage using 1 ml of ice-cold PBS three times.
Leukocyte Preparation-Total bone marrow cells were isolated from mice by flushing the femur and tibia from both legs two times with 1 ml of PBS. To obtain total leukocytes, spleen and mesenteric lymph nodes were dissected and mechanically disaggregated. Single-cell suspensions were obtained using a 40-m cell strainer. Peripheral blood was obtained by cardiac puncture. Red blood cells were removed from blood samples and splenic single-cell suspensions by ammonium chloride lysis. For some experiments, neutrophils were sorted from total bone marrow cells using anti-Ly-6G magnetic microbeads (Miltenyi Biotec, Auburn, CA). The absolute number of each cell subpopulation was determined by flow cytometry using CALTAG counting beads (Invitrogen, Carlsbad, CA). For bone marrow and spleen, the absolute cell counts were normalized by the individual body weight expressed in grams. Alternatively, blood cell counts were determined in the Department of Laboratory Medicine at the National Institutes of Health.
Histological Analysis-Tissues from Sgpl1 ϩ/ϩ and Sgpl1 Ϫ/Ϫ mice were fixed and embedded in paraffin. Sections were stained with H&E. A comprehensive histologic evaluation of the sections was performed by the National Institutes of Health Division of Veterinary Resources Pathology Service.
BrdU Labeling-For bromodeoxyuridine (BrdU) labeling, mice were injected intraperitoneally once with 100 l of a 10 mg/ml BrdU solution (BD Biosciences) and analyzed 48 h later. BrdU-positive cells were detected using a FITC-BrdU Flow kit (BD Biosciences) by flow cytometry.
Chemotaxis Assay-The response of neutrophils toward formyl-methionyl-leucyl-phenylalanine (fMLP) was studied using 6.5 mm Transwell inserts with a 5-m pore size (Corning, Cambridge, MA). A splenocyte suspension in RPMI 1640 plus 0.4 mg/ml fatty acid-free (FAF)-BSA (Sigma-Aldrich) (medium ϩ FAF-BSA) was added to each insert in a well containing 600 l of a solution of 1 M fMLP (SigmaAldrich) prepared in medium ϩ FAF-BSA. Wells containing medium ϩ FAF-BSA without fMLP were used as controls. After 3 h at 37°C, cells in the bottom of the wells were harvested, counted, and analyzed by flow cytometry.
Gene Expression-For RT-qPCR, total RNA was purified from bone marrow neutrophils and from liver using TRIZOL (Invitrogen). The mRNA expression levels of , S1pr1 (Mm00514644_m1), S1pr2 (Mm01177794), S1pr3 (Mm00515669_m1), S1pr4 (Mm00468695_s1), and S1pr5 (Mm00474763_m1) genes were determined by RT-qPCR using Assay-on-Demand probes and primers (Applied Biosystems, Foster City, CA) on an ABI Prism 7700 Sequence Detection System (Applied Biosystems). Glyceraldehyde-3phosphate dehydrogenase (Mm99999915_g1) mRNA level was used as an internal control. For microarray analysis, RNA purified from Sgpl1 ϩ/ϩ and Sgpl1 Ϫ/Ϫ livers was prepared and analyzed on Affymetrix GeneChip Mouse Genome 430 2.0 arrays as described (29). The National Center for Biotechnology Information Gene Expression Omnibus accession number for the microarray data is GSE18745. Genes that belong to the Gene Ontology (GO) category "acute inflammation" were compared between Sgpl1 ϩ/ϩ and Sgpl1 Ϫ/Ϫ mRNA samples using a heat map generated using the Partek Genomic Suite 6.5 software (Partek Inc., St. Louis, MO).
Lipid Analysis of Serum-Sphingolipids in serum were measured by high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) by the Lipidomics Core at the Medical University of South Carolina on a Thermo Finnigan (Waltham, MA) TSQ 7000 triple quadrupole mass spectrometer, operating in a multiple reaction monitoring-positive ionization mode as described (32).
Statistical Analysis-Statistical significance was determined using the Mann-Whitney or Student's t test. In all cases, values of p Ͻ 0.05 were considered statistically significant.

Profound Neutrophilia in Sgpl1
Ϫ/Ϫ Mice-Sgpl1 Ϫ/Ϫ mice generally die soon after weaning (26,28,29), necessitating studies on young mice, at P18 unless otherwise indicated. The Sgpl1 Ϫ/Ϫ mice were severely lymphopenic, with deficiencies of both T (CD4 ϩ , CD8 ϩ ) cells and B (B220 ϩ ) cells in blood (supplemental Fig. S1, A and C). Low numbers of T and B cells were also apparent in the spleen of the Sgpl1 Ϫ/Ϫ mice (supplemental Fig. S1, B and D). The lymphocyte deficiency in the Sgpl1 Ϫ/Ϫ mice has been attributed to a number of possible sources, including a block in the egress of recirculating lymphocytes from secondary lymphoid organs and to defects in the development of T cells and B cells (27). In contrast to the severe lymphopenia, blood levels of neutrophils, and monocytes were highly elevated in the Sgpl1 Ϫ/Ϫ mice ( Fig. 1A) (26). Sgpl1 ϩ/Ϫ mice had normal blood cell counts indicating that both Sgpl1 alleles must to be deleted to produce this immune phenotype (supplemental Fig. S2A).
We next determined the numbers and percentages of neutrophils and monocytes in the spleen and their precursors in the bone marrow. Compared with the WT mice, the percentages of splenic neutrophils and monocytes were dramatically elevated in Sgpl1 Ϫ/Ϫ compared with WT mice (Fig. 1C). The total numbers of neutrophils and monocytes in spleen, when normalized by body weight, were also significantly increased in the Sgpl1 Ϫ/Ϫ mice compared with WT mice (Fig. 1D). We enumerated myeloid cells, the precursors of neutrophils and monocytes, in bone marrow and found a substantial increase of both immature and mature myeloid cells in the Sgpl1 Ϫ/Ϫ mice when expressed as either percentage of total cells or absolute cell numbers normalized by body weight (Fig. 1, E and F). The ratio of immature to mature myeloid cells was similar between Sgpl1 Ϫ/Ϫ and WT mice suggesting differentiation of neutrophil precursors in the bone marrow of Sgpl1 Ϫ/Ϫ mice was normal.
To establish whether increased granulopoiesis was responsible for the elevated neutrophil numbers in Sgpl1 Ϫ/Ϫ mice, we labeled proliferating precursor cells in bone marrow by BrdU incorporation and determined the appearance of BrdU ϩ neutrophils in blood after 48 h. The numbers of BrdU ϩ neutrophils generated in Sgpl1 Ϫ/Ϫ mice were significantly higher than in WT mice (Fig. 1B). These data indicate that increased granulopoiesis was occurring in the Sgpl1 Ϫ/Ϫ mice.
We also determined the circulating ceramide, sphingosine, and S1P levels at P6 and P18 by HPLC-MS/MS analysis (supplemental Fig. S3). Both ceramide and S1P levels were substantially elevated at P6 and P18 in the Sgpl1 Ϫ/Ϫ mice compared with WT mice. With increasing age, ceramide and S1P levels increased in the Sgpl1 Ϫ/Ϫ mice. Sphingosine remained at similar low levels at both ages tested. Sgpl1 ϩ/Ϫ mice had serum sphingolipid levels similar to WT mice (supplemental Fig. S2B).
To rule out stress-induced changes on peripheral lymphocytes and neutrophils, we measured circulating corticosterone levels and found that there was no significant difference between the values of Sgpl1 Ϫ/Ϫ and WT mice (supplemental Fig. S4).
We next determined the progression of inflammatory changes in Sgpl1 Ϫ/Ϫ mice by examining the mRNA expression of some genes related to inflammation in liver at P6 and P18. An increase in Tnf, Saa1, and Saa3 was detected by RT-qPCR at both ages in the Sgpl1 Ϫ/Ϫ livers compared with those of WT mice (Fig. 2B). We also detected an increase in the expression level of the endothelial adhesion molecule Vcam-1 in the P18 Sgpl1 Ϫ/Ϫ livers when compared with controls (Fig.  2B). Vcam-1 is elevated during the endothelium response to inflammatory stimuli (33). Together, these results indicate the presence of a pro-inflammatory process in the liver of Sgpl1 Ϫ/Ϫ mice that intensifies with age.
In the serum of Sgpl1 Ϫ/Ϫ mice, we detected a significant increase in the levels of the pro-inflammatory cytokines TNF, IFN-␥, MCP-1, and IL-6 compared with levels observed in WT mice (Fig. 2C), demonstrating that the inflammation was systemic. Sgpl1 ϩ/Ϫ mice did not display any elevated serum cytokine levels (supplemental Fig. S2C). Injection of Sgpl1 Ϫ/Ϫ mice with a dose of LPS, which was not lethal to WT mice, caused their death by 8 h (supplemental Fig. S5A). Within 120 min after the LPS injection, the Sgpl1 Ϫ/Ϫ mice expressed substantially higher levels of TNF, IL-6, and MCP-1 in serum compared with the WT mice injected with LPS (supplemental Fig. S5B). Serum ceramide, sphingosine, and S1P concentrations were not significantly elevated over the respective baseline values in either Sgpl1 Ϫ/Ϫ or control mice after LPS treatment during the same time period (supplemental Fig. S5C).
The pro-inflammatory state of the Sgpl1 Ϫ/Ϫ mice raised the possibility that stimulation of the innate immune system by opportunistic bacterial infection, perhaps as a result of their severe lymphocyte deficiency, might be inducing systemic inflammation. We found no histologic evidence for the presence of infection in the tissue of Sgpl1 Ϫ/Ϫ mice compared with WT mice (supplemental Fig. S6). We therefore studied the effect of deleting the Toll-like receptor-4 (Tlr4) in Sgpl1 Ϫ/Ϫ mice (Fig. 3). Tlr4 is a receptor that detects the LPS of Gram-negative bacterial cell walls, and triggers a cascade of events that activates the transcription of many pro-inflammatory genes involved in the innate immune response (34). If the inflammatory response in Sgpl1 Ϫ/Ϫ mice was the result of a Gram-negative bacterial infection, the inflammation in the Sgpl1 Ϫ/Ϫ Tlr4 Ϫ/Ϫ double knock-out (DKO) mice should be suppressed relative to the Sgpl1 Ϫ/Ϫ mice. We found that the Sgpl1 Ϫ/Ϫ Tlr4 Ϫ/Ϫ DKO and Sgpl1 Ϫ/Ϫ mice had similarly depressed lymphocyte and elevated neutrophil blood counts (Fig. 3A) and similar elevations of pro-inflammatory cytokines in serum (Fig. 3B) and of mRNA levels for Tnf and Vcam-1 in liver (Fig. 3C). Whereas the Sgpl1 Ϫ/Ϫ mice were hypersensitive to LPS, dying by 8 h after being challenged with a nonlethal dose of LPS, the Sgpl1 Ϫ/Ϫ Tlr4 Ϫ/Ϫ DKO mice were resistant (supplemental Fig. S5A). These results indicate that the pro-inflammatory condition of the Sgpl1 Ϫ/Ϫ mice was unlikely to be the result of an activation of the innate immune system by an opportunistic Gram-negative bacterial infection.
Impaired Neutrophil Migration into Sites of Inflammation in Sgpl1 Ϫ/Ϫ Mice-We next sought to determine if the neutrophilia was correlated with large numbers of neutrophils infiltrating into the liver of Sgpl1 Ϫ/Ϫ mice, as would be expected based on the acute inflammation exhibited by this tissue (Fig. 2, A and B). Surprisingly, we found that the liver parenchyma in Sgpl1 Ϫ/Ϫ mice appeared similar to Sgpl1 ϩ/ϩ mice and did not exhibit extensive neutrophil infiltration (Fig.  4, A and B); however, we did observe abnormally high accumulations of neutrophils that appeared to be confined at the hepatic sinusoids in Sgpl1 Ϫ/Ϫ mice compared with Sgpl1 ϩ/ϩ mice (Fig. 4, C and D). These results suggested a possible impairment of neutrophil recruitment from the blood into the  inflamed tissue in the Sgpl1 Ϫ/Ϫ mice. A survey of other Sgpl1 Ϫ/Ϫ tissues showed that they were spared from substantial neutrophil infiltration (supplemental Fig. S6, C-L). As reported previously (26), the Spgl1 Ϫ/Ϫ lungs contained a slight increase in macrophages within alveoli (supplemental Fig. S6, A and B).
To establish whether Sgpl1 deficiency altered the ability of neutrophils to migrate to sites of inflammation, we injected mice intraperitoneal with thioglycollate and examined the influx of neutrophils into the peritoneal cavity after 4 h. We calculated the number of neutrophils recruited into the peritoneum relative to the available neutrophils in the blood (Fig.  5, A and B) and found that Sgpl1 Ϫ/Ϫ mice had a significantly lower efficiency of neutrophil recruitment into thioglycollatetreated peritoneum than did the WT mice (Fig. 5B).
We tested the ability of neutrophils from Sgpl1 Ϫ/Ϫ mice to migrate to a chemotaxic stimulus. As shown in Fig. 5C, neutrophils from Sgpl1 Ϫ/Ϫ mice were significantly impaired relative to WT neutrophils in their ability to migrate toward fMLP peptide.
The lower levels of CD62L on the Sgpl1 Ϫ/Ϫ neutrophils did not appear to be the result of enhanced surface shedding of this molecule, because the serum levels of soluble CD62L were similar in Sgpl1 Ϫ/Ϫ and Sgpl1 ϩ/ϩ mice (supplemental Fig. S7A). Instead, the lower expression levels correlated with a significantly lower expression of CD62L (Sell) mRNA in Sgpl1 Ϫ/Ϫ compared with Sgpl1 ϩ/ϩ mice (supplemental Fig. S7B).

S1P Lyase, Inflammation, and Neutrophil Trafficking
Sgpl1 Ϫ/Ϫ Gr-S1pr1KO mice did not significantly differ from the single mutant Sgpl1 Ϫ/Ϫ mice in blood levels of lymphocytes and neutrophils and in serum concentrations of pro-inflammatory cytokines (Fig. 8, A and B). In contrast, Sgpl1 Ϫ/Ϫ S1pr4 Ϫ/Ϫ mice had significantly lower blood neutrophil counts and serum pro-inflammatory cytokines than the single-mutant Sgpl1 Ϫ/Ϫ mice, although both Sgpl1 Ϫ/Ϫ  S1pr4 Ϫ/Ϫ , and Sgpl1 Ϫ/Ϫ mice were similarly lymphopenic (Fig. 8, D and E). The expression of CD62L on the Sgpl1 Ϫ/Ϫ neutrophils was not normalized by either the deletion of S1pr1 or S1pr4 Ϫ/Ϫ (Fig. 8, E and F). The results indicate that the S1P4 receptor plays a role in the extent of pro-inflammatory responses caused by the S1P lyase deficiency. However, the S1P1 receptor, when specifically deleted in the granulocyte lineage, does not alter the phenotype.
Homeostatic Regulation of Neutrophil Production Is Abnormal in Sgpl1 Ϫ/Ϫ Mice-Neutrophils normally traffic to tissues where they undergo apoptosis and are phagocytosed by macrophages and dendritic cells, down-regulating cytokine production and leading to increased granulocyte production. When transmigration of neutrophils to tissues is blocked, as is the case in mice deficient in leukocyte or endothelial adhesion molecules, IL-23 production by macrophages and dendritic cells increases, stimulating IL-17 generation by T cells. IL-17 in turn induces G-CSF production, leading to higher granulopoiesis in the bone marrow and elevated circulating neutrophils (Fig. 9E) (47,48).
We next determined if this negative feedback loop for neutrophil homeostasis was disrupted in the Sgpl1 Ϫ/Ϫ mice. An increase in IL-23 mRNA expression was found in Sgpl1 Ϫ/Ϫ lung, liver, kidney, and small intestine compared with levels FIGURE 8. S1pr4 deletion lowers blood neutrophil levels and inflammatory markers in Sgpl1 ؊/؊ mice. S1Pr1 fl/fl , Gr-S1pr1KO, Sgpl1 Ϫ/Ϫ , and Sgpl1 Ϫ/Ϫ Gr-S1pr1KO DKO mice (A, C, E), and WT, S1pr4 Ϫ/Ϫ , Sgpl1 Ϫ/Ϫ , and Sgpl1 Ϫ/Ϫ S1pr4 Ϫ/Ϫ DKO mice (B, D, F) were analyzed for blood cell counts, pro-inflammatory cytokines, and CD62L expression on neutrophils. A and B, blood lymphocyte and neutrophil counts were determined by the Department of Laboratory Medicine at the NIH using an automated blood cell analyzer. Results are shown as number of cells per l of blood. The bars represent mean values, and the closed circles are individual mice. Results are shown as number of cells per l of blood. The bars represent mean values, and the closed circles are individual mice. C and D, concentration of pro-inflammatory cytokines in serum determined using the BD CBA Array. Data are shown as mean values Ϯ S.D., n ϭ 5. E and F, CD62L expression on the surface of blood neutrophils from DKO mice. E, expression levels of CD62L on S1Pr1 fl/fl , Gr-S1pr1KO, Sgpl1 Ϫ/Ϫ , and Sgpl1 Ϫ/Ϫ Gr-S1pr1 DKO neutrophils are shown as mean fluorescence intensity (MFI). F, expression levels of CD62L on WT, S1pr4 Ϫ/Ϫ , Sgpl1 Ϫ/Ϫ , and Sgpl1 Ϫ/Ϫ S1pr4 Ϫ/Ϫ DKO neutrophils are shown as the MFI. The bars represent mean values, and the closed circles are individual mice. All mice were analyzed at P18. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.005. ns, not significant. observed in Sgpl1 ϩ/ϩ mice (Fig. 9A). To investigate IL-17 generation, we analyzed the presence of IL-17-producing Th-17 cells in mesenteric lymph nodes and found a significant increase in the percentage of IL-17 ϩ Foxp3 Ϫ CD4 ϩ cells in the Sgpl1 Ϫ/Ϫ mice compared with percentages observed in Sgpl1 ϩ/ϩ mice (Fig. 9B). We also determined the serum concentration of IL-17 and found elevated levels of IL-17 in the serum of Sgpl1 Ϫ/Ϫ mice compared with those in the serum of Sgpl1 ϩ/ϩ mice (Fig. 9C). Finally, the G-CSF concentration in serum was significantly increased in the Sgpl1 Ϫ/Ϫ mice compared with the G-CSF concentration in Sgpl1 ϩ/ϩ mice (Fig. 9D).

DISCUSSION
In the absence of S1P lyase, Sgpl1 Ϫ/Ϫ mice lack the major degradation pathway for S1P and, as a result, express high levels of S1P in circulation and in tissues (26,29). These mice show diverse phenotypic abnormalities including a shortened lifespan, abnormal lipid metabolism and severe immune system disturbances affecting both the adaptive and innate components.
As shown here, the Sgpl1 Ϫ/Ϫ mice were deficient in B and T lymphocytes yet had high blood levels of neutrophils and monocytes along with elevated expression of pro-inflammatory cytokines. Surprisingly, their tissues were largely clear of infiltrating leukocytes.
Peripheral blood numbers of neutrophils are tightly regulated by a homeostatic feedback loop that controls granulopoiesis (Fig. 9E) (47,48). Within the loop, IL-23 is expressed by tissue resident macrophages and dendritic cells to initiate a cytokine cascade that links IL-17 and G-CSF. First, IL-23 potently elevates IL-17 levels through increases in the numbers of IL-17-secreting T cells; then, IL-17 induces an increase in the levels of G-CSF, which promotes the proliferation of promyelocytes and the maturation of granulocytes in the bone marrow (47,48). Normally, after release from the bone marrow, neutrophils move from blood to tissues, where they undergo apoptosis and are phagocytosed by macrophages and dendritic cells (47). This process of phagocytosis supplies a potent signal to macrophages and dendritic cells that suppresses IL-23 synthesis, closing the feedback loop.
We found that the Sgpl1 Ϫ/Ϫ mice have disturbed neutrophil homeostasis. They have activation of the IL-23/IL-17/G-CSF cytokine axis leading to increased granulopoiesis. The increased neutrophil production in the bone marrow of the Sgpl1 Ϫ/Ϫ mice did not appear to be the result of a bacterial infection, because their tissues appeared clear of infectious agents upon histologic examination, and because neutrophil numbers did not decrease when the LPS sensor Tlr4 was deleted from the Sgpl1 Ϫ/Ϫ mice. Instead, our results point to direct effects by the Sgpl1 deficiency on neutrophil homostasis.
A key point in neutrophil homeostasis occurs with the transmigration of neutrophils across the endothelium and into tissues. When this step is blocked, as with mice with integrin and selectin deficiencies (35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46) or with humans with leukocyte adhesion deficiencies (49,50), neutrophilia is a prominent feature. Adhesion molecules, both selectins and integrins, on neutrophils are essential for their interaction with the endothelium and entry into tissues (51). The increase in blood neutrophils is not a passive phenomenon in the adhesion molecule deficiencies, but rather is due to activation of the IL-23/IL-17/G-CSF cytokine axis. This axis is activated by the absence of suppressive signals due to the inability of neutrophils to transmigrate into tissues where they are phagocytosed by macrophages and dendritic cells (47,48). As has been reported for several of the adhesion molecule knockout mice, the Sgpl1 Ϫ/Ϫ mice exhibited reduced efficiency of the recruitment of neutrophils to sites of inflammation (35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46). The reduced expression of multiple adhesion molecules on the neutrophils in Sgpl1 Ϫ/Ϫ mice is one mechanism to explain the reduced entry of blood neutrophils into tissues, thereby disrupting the homeostatic regulatory loop and activating the IL-23/IL-17/G-CSF pathway controlling granulopoiesis. The greatest adhesion molecule deficiency on the Sgpl1 Ϫ/Ϫ neutrophils was found for CD62L, with smaller de-  (47,48). Sgpl1 deletion impedes the migration of blood neutrophils into tissues. Tissue macrophages and dendritic cells respond to reduced phagocytosis of apoptotic neutrophils by increased Il-23 production leading to elevated 1L-17 and G-CSF synthesis. Sgpl1 ϩ/ϩ (open bars) and Sgpl1 Ϫ/Ϫ (gray bars). *, p Ͻ 0.05. creases in surface expression of CD11b, CD11a, CD18, and CD49d. Whereas a single deficiency of CD62L did not lead to neutrophilia in mice (37), compound deficiencies of CD62L with other adhesion molecules did result in elevated numbers of neutrophils in blood (42,44,45). The large decrease of surface CD62L expression on Sgpl1 Ϫ/Ϫ neutrophils did not appear to be the result of enhanced shedding, which occurs during neutrophil activation, because the levels of soluble CD62L in serum were similar in Sgpl1 Ϫ/Ϫ and Sgpl1 ϩ/ϩ mice. Furthermore, levels of CD11b, a marker of neutrophil activation (52), were reduced on the Sgpl1 Ϫ/Ϫ neutrophils. Instead, a transcriptional mechanism appeared to be responsible for the decreased expression of CD62L, because of significantly decreased levels of its mRNA in the Sgpl1 Ϫ/Ϫ neutrophils.
The Sgpl1 Ϫ/Ϫ neutrophils also displayed a reduced capacity in vitro to navigate toward a strong chemotaxic stimulus suggesting intrinsic defects in their migration response, which may also contribute to their reduced recruitment to imflamed tissues. Inhibition of chemotaxis by extracellular S1P have been reported previously for other cell types (53)(54)(55).
Heightened S1P levels in the Sgpl1 Ϫ/Ϫ mice may also intersect at another point on the neutrophil homeostatic loop to enhance pro-inflammatory signaling. Acting through the S1P1 receptor on T cells, S1P has the same capacity as IL-23 to augment the number of Th-17 cells and the secretory activity of IL-17 by Th-17 cells (56). In the Sgpl1 Ϫ/Ϫ mice, although the total numbers of T cells were reduced, the percentage of Th-17 cells was significantly elevated, demonstrating a specific enrichment of the Th-17 lineage. Th-17 cells are highly proinflammatory (57), and may contribute to the systemic inflammation observed in the Sgpl1 Ϫ/Ϫ mice. IL-17 acts on many cell types to stimulate the production of pro-inflammatory cytokines, including TNF, IL-6, and MCP-1, which were all found to be elevated in the Sgpl1 Ϫ/Ϫ mice. Sphingosine kinase and its product, S1P, have been shown to have pro-inflammatory effects by acting, in some cases, through cell-surface receptors (58 -64). Also, intracellular S1P has recently been shown to act as a cofactor for the E3 ubiquitin-ligase TRAF2, which stimulates NF-B activation (65).
The deletion of the S1P4 receptor in the Sgpl1 Ϫ/Ϫ background caused a decrease in the pro-inflammatory responses, implying that signaling through the S1P4 receptor is contributing to this phenotype. Because S1P binds and activates the S1P4 receptor at concentrations that are substantially higher than those for the other S1P receptors (66), it is possible that the S1P4 receptor becomes engaged in the Sgpl1 Ϫ/Ϫ mice due to high concentrations of S1P. The S1P4 receptor is widely expressed on the cells of the immune and hematopoietic systems (67,68), suggesting several possible cellular sites of action (including on neutrophils, T cells, and macrophages). In T cells, the S1P4 receptor has been reported to regulate proliferation and cytokine secretion (69) and could potentially control the levels of cytokines expressed in the Sgpl1 Ϫ/Ϫ mice. Excessive signaling through the S1P4 receptor does not seem to be the primary cause of the early demise of the Sgpl1 Ϫ/Ϫ mice because the Sgpl1 Ϫ/Ϫ S1pr4 Ϫ/Ϫ DKO mice do not live longer than single mutant Sgpl1 Ϫ/Ϫ mice (data not shown).
The partial normalization of the inflammatory phenotype in the Sgpl1 Ϫ/Ϫ mice by the S1pr4 deletion indicates that other receptor signaling pathways may be operating. The S1P1 receptor, when deleted within granulocytes and monocytes in the Sgpl1 Ϫ/Ϫ mice, failed to significantly reduce the levels of pro-inflammatory markers observed in Sgpl1 Ϫ/Ϫ mice; however, it is possible that the S1P1 receptor might function in other tissues or cells to regulate inflammatory pathways in the Sgpl1 Ϫ/Ϫ mice. The S1P1 receptor has an important role in controlling endothelial barrier integrity (70), and its activation in endothelial cells in the Sgpl1 Ϫ/Ϫ mice may impede neutrophil migration into tissues (Fig. 9E). The S1P1 receptor also activates the Stat3 pathway, which regulates the acute-phase response in liver and the generation of Th-17 cells, both features of the Sgpl1 Ϫ/Ϫ mice (71)(72)(73).
Our results show that in the absence of S1P lyase activity pro-inflammatory responses are elevated while neutrophil migration into inflamed tissues is impaired, thereby disrupting homeostatic regulation of neutrophils. S1P lyase activity and the relevant S1P receptor signaling pathways represent new potential targets for manipulating neutrophil levels, their function, and inflammation.