Report

Lymphocyte Sequestration Through S1P Lyase Inhibition and Disruption of S1P Gradients

See allHide authors and affiliations

Science  09 Sep 2005:
Vol. 309, Issue 5741, pp. 1735-1739
DOI: 10.1126/science.1113640

Abstract

Lymphocyte egress from the thymus and from peripheral lymphoid organs depends on sphingosine 1-phosphate (S1P) receptor-1 and is thought to occur in response to circulatory S1P. However, the existence of an S1P gradient between lymphoid organs and blood or lymph has not been established. To further define egress requirements, we addressed why treatment with the food colorant 2-acetyl-4-tetrahydroxybutylimidazole (THI) induces lymphopenia. We found that S1P abundance in lymphoid tissues of mice is normally low but increases more than 100-fold after THI treatment and that this treatment inhibits the S1P-degrading enzyme S1P lyase. We conclude that lymphocyte egress is mediated by S1P gradients that are established by S1P lyase activity and that the lyase may represent a novel immunosuppressant drug target.

Lymphocyte egress from lymphoid organs is critical for immune surveillance and immune effector function and depends on intrinsic expression of sphingosine 1-phosphate receptor-1 (S1P1), a seven-transmembrane receptor that binds S1P (14). S1P is abundant in blood, and it has been suggested that lymphocytes exit from lymphoid tissues into blood or lymph in response to S1P gradients (5, 6). However, all cell types generate S1P intracellularly during sphingolipid metabolism, making it unclear which factors regulate the S1P levels in tissues versus in plasma and lymph (7, 8). To further define requirements for lymphocyte exit, we followed up on a finding made during toxicity testing of caramel food colorants. In these tests, a component of caramel color III, 2-acetyl-4-tetrahydroxybutylimidazole (THI), induced lymphopenia (911) and caused mature T cells to accumulate in the thymus, apparently as a result of reduced egress (12, 13).

We observed that in addition to causing an accumulation of mature T cells in the thymus, THI inhibited lymphocyte egress from lymph nodes (Fig. 1). Mice treated for three days with THI (14) were depleted of CD4 and CD8 T cells and, to a lesser degree, B cells in both lymph and blood, without losing cells in lymph nodes (Fig. 1, B to D). Because naïve lymphocytes enter lymph only by exiting lymph nodes and Peyer's patches, this result indicated that THI blocked exit. During toxicology studies on THI, an investigation of inconsistencies among data from different laboratories revealed that the sequestration was prevented by excess dietary vitamin B6 (11). We therefore also examined the effect of treating mice with 4′ deoxypyridoxine (DOP), a vitamin-B6 antagonist (15), and found that DOP mimicked THI in causing a build up of mature cells in the thymus, depleting T and B lymphocytes from lymph, and inducing lymphopenia (Fig. 1). Adoptively transferred cells disappeared from the blood and lymph during THI or DOP treatment but returned after treatment was stopped, indicating that the compounds were not causing a loss of cells due to toxicity (fig. S1) (16).

Fig. 1.

THI and DOP cause accumulation of mature cells in the thymus and loss of lymphocytes from the lymph and blood. (A to D) Lymphocyte numbers in the indicated tissues from mice given drinking water alone (control, white bars), given water containing 30 mg of DOP per liter (gray bars), or given water containing 50 mg of THI per liter (black bars) for three days. Mature thymocytes (A) were defined as CD4+ or CD8+ and CD69loCD62Lhi, and double-positive thymocytes were defined as CD4+ and CD8+. Points indicate values from individual mice and bars indicate the average. When data from three experiments were combined, the increase in mature thymocytes and decrease in blood and lymph lymphocytes were statistically significant (P < 0.05), with the exception of the decrease in B220+ cells in the lymph of DOP-treated mice.

Because S1P1 is essential for lymphocyte egress from the thymus and secondary lymphoid organs, we tested whether THI and DOP were acting by altering S1P1 function. Flow cytometric analysis revealed that S1P1 expression was almost undetectable on mature CD4 or CD8 single-positive thymocytes (CD4+ or CD8+) from THI- and DOP-treated mice but was readily detected on control thymocytes (Fig. 2A). Similarly, THI and DOP treatment caused a loss of surface S1P1 on lymph node CD4 and CD8 T cells (Fig. 2B). S1P1 mediates chemotaxis of lymphocytes to S1P. Consistent with the loss of surface expression, neither CD4 nor CD8 splenic T cells from THI- and DOP-treated mice could migrate in response to S1P, and a similar, although less severe, impairment was seen in follicular B cells (Fig. 2, C and D).

Fig. 2.

THI and DOP treatment lead to reduced S1P1 expression and impaired ability to migrate to S1P. (A) Mature single-positive thymocytes and (B) peripheral lymph node cells were stained for S1P1 expression. Green lines show staining of cells from three control-treated mice, blue from three DOP-treated mice, and red from three THI-treated mice. Gray lines show staining of all nine cell types with a control antibody. Staining is representative of at least three experiments. (C and D) Splenocytes from THI- (red), DOP- (blue), or control- (green) treated mice were tested for their ability to migrate to the indicated concentrations of S1P or to 0.3 μg of stromal-cell derived factor (CXCL12) per ml of medium. Individual points show migration in replicate wells. When data from three experiments were compiled, the impaired migration of CD4+ and CD8+ T cells from DOP- and THI-treated mice to 10 nM S1P was statistically significant (P < 0.05). The impaired migration of B220+ cells was significant for THI (P = 0.05) but not for DOP.

Cells internalize S1P1 upon exposure to S1P (17), and incubation of freshly isolated thymocytes with as little as 0.5 nM S1P down-modulated surface S1P1 expression (Fig. 3A). We therefore considered the possibility that THI and DOP were causing S1P concentrations within lymphoid tissues to increase. The extraction of lipids from plasma and whole tissue and the quantification of total S1P by LC/MS/MS (liquid chromatography followed by tandem mass spectrometry) (14) revealed that lipid levels in plasma were approximately 330 ng/g (860 nM) and were not changed by THI or DOP treatment (Fig. 3B and fig. S2). In contrast, tissue S1P concentrations rose markedly from ∼20, 40, and 150 ng/g of tissue in the thymus, lymph nodes, and spleen, respectively, to 5,000 to 10,000 ng/g in each of these tissues after treatment (Fig. 3B). Total tissue analysis measures intracellular and extracellular S1P. To selectively determine the receptor-accessible extracellular S1P concentration, we took two steps. First, we developed a bioassay in which S1P-induced down-modulation of FLAG-tagged S1P1 expression in WEHI231 cells could be measured. When S1P was titrated onto the cells, detectable receptor down-regulation at 1 nM was observed (Fig. 3C). Second, we prepared tissue extracts enriched in interstitial fluid by disaggregating lymphoid organs to a single cell suspension in isotonic medium and removing the cells by centrifugation. Because our goal was to gain an indication of the bioavailable S1P, we did not further purify the lipids, and it is possible that our measurements underestimate total extracellular S1P, because some may be sequestered. When we titrated these extracts onto the WEHI231 cells, we found that S1P was present in low to undetectable amounts in extracts from control thymus tissue, lymph nodes, and Peyer's patches (Fig. 3, D to F). The detection of S1P in the control spleen extract is consistent with this organ being partly composed of plasma-rich red pulp (Fig. 3G). S1P was abundant in plasma, and amounts in lymph were lower but within sixfold of those in plasma (Fig. 3, H and I). Analysis of extracts from THI- and DOP-treated mice revealed that S1P abundance was increased more than 1000-fold in the thymus, lymph node, and Peyer's patch extracts and more than 100-fold in spleen extracts (Fig. 3, D to G). In contrast, there was no difference in plasma and lymph S1P concentrations between the treated and untreated mice (Fig. 3, H and I).

Fig. 3.

THI and DOP increase lymphoid tissue S1P concentrations. (A) Mature CD4+ thymocytes from an untreated mouse were incubated in the indicated concentration of S1P, and S1P1 mean fluorescence intensity (MFI) levels were measured. (B) S1P levels in the indicated tissues from control- (open bars), DOP- (gray), or THI- (black) treated mice, measured by LC/MS/MS. Points represent values from individual mice, and bars represent the average. In some cases, the mice also received vitamin B6 (pyridoxine HCl, 450 mg per liter of water), as indicated. The thymic, lymph node, and spleen S1P concentrations in THI- and DOP-treated mice were significantly different from the control (P < 0.05). (C) FLAG-S1P1-expressing WEHI231 cells were incubated in the indicated concentration of S1P, and FLAG-S1P1 MFI levels were measured. (D to G) Extracts enriched in interstitial fluid from the indicated tissues were prepared (14) and titrated onto FLAG-S1P1 WEHI231 cells. The x axis shows dilution of total tissue from which the extract was made, assuming that the tissue density is 1 g/ml. (H and I) As in (C), but x axis shows dilution of plasma or lymph. Maximum FLAG-S1P1 levels vary among experiments because they were performed on different days. All data are representative of at least three experiments, with the exception of MS analysis of vitamin-B6 treated mice and S1P measurement in Peyer's patches.

Of the enzymes involved in determining S1P abundance, S1P lyase, a cytoplasmic enzyme that degrades S1P into phosphoethanolamine and 2-hexadecanal, represented an appealing candidate as a potential target for THI and DOP, because it is dependent on vitamin B6 (8, 18, 19). Mice that had received THI or DOP combined with a 10-fold molar excess of vitamin B6 in their drinking water showed no accumulation of mature thymocytes, induction of lymphopenia, down-modulation of S1P1, or increase in lymphoid tissue S1P levels (Fig. 3B and fig. S3). We therefore tested whether S1P lyase activity in the thymus was affected by THI or DOP treatment. Thymic lysates from treated and untreated mice were incubated with tritiated dhS1P (dihydrosphingosine 1-phosphate), and the products were separated by thin-layer chromatography (14). Lyase activity, measured by the generation of hexadecanal, was severely inhibited in the thymus of DOP- and THI-treated mice (Fig. 4, A and B). Excess vitamin B6 overcame the effect of THI or DOP and restored lyase activity (Fig. 4, A and B). These findings suggest that THI and DOP cause increased lymphoid tissue S1P abundance because of the inhibition of S1P lyase.

Fig. 4.

THI and DOP inhibit S1P lyase, and RNAi-mediated S1P lyase knockdown recapitulates the THI- and DOP-induced phenotype. (A and B) Thymic lysates from mice treated as indicated (B6, vitamin B6) were incubated with tritiated dhS1P. The products were separated by thin-layer chromatography and visualized by autoradiography (A). The lyase product hexadecanal and its metabolites hexadecanol and palmitic acid have a similar Rf (where Rf is the distance moved by the lipid divided by the distance from the origin to the solvent front) (19) and were not distinguished in our assay. dh-Sph, dihydrosphingosine. In (B), quantitative analysis by PhosphorImager (GE Healthcare, Chalfont St. Giles, UK) of thymic lyase activity in three experiments is plotted as percent of control activity. The activity reduction in lysates from THI- and DOP-treated mice was statistically significant (P < 0.05). (C) Quantitative real-time PCR analysis of lyase relative to hypoxanthine-guanine phosphoribosyltransferase (HPRT) transcript abundance in the indicated tissues from mice reconstituted with HSC transduced with a retrovirus encoding the shRNA and a GFP reporter (black bars), or the GFP reporter alone (open bars). GFPhi CD4SP indicates GFP-high, CD4+ CD8 thymocytes; thy. susp., thymic suspension; thymus, total thymus; spl. susp., splenic suspension; spleen, total spleen. Bars show average, points show values from individual animals. The RNA levels in GFPhi thymocytes, thymic suspension, splenic suspension, and total spleen of mice receiving shRNA were significantly different from the control (P < 0.05). (D) Flow cytometric analysis of representative thymuses, spleens, and blood, gated by size on lymphocytes. Numbers in the upper corners show the total number of cells in the tissue or cells per milliliter in blood, and the numbers next to gates show the percent of total cells in the gate. (E) Representative flow cytometric analysis of S1P1 expression on mature CD4+ thymocytes. Shaded histograms show staining of both cell types with a control antibody. (F) S1P levels in total thymus, spleen, and plasma measured by LC/MS/MS. The thymic and spleen S1P concentrations in mice that received shRNA were significantly different from the control (P < 0.05). (G to I) Representative bioassay measurement of S1P levels in the indicated tissue extracts, as in Fig. 3.

To further test whether a reduction in S1P lyase activity could account for the increased tissue S1P abundance and disruption in egress, we reduced hematopoietic cell lyase expression by RNA interference (RNAi)–mediated knockdown (14). Hematopoietic stem cells (HSCs) were transduced with a retrovirus containing a lyase RNAi-inducing small hairpin RNA (shRNA) and a GFP reporter (ZsGreen green fluorescent protein), or the reporter alone. The transduced HSCs were used to reconstitute irradiated lymphocyte-deficient (RAG2–/–) mice. Quantitative polymerase chain reaction (PCR) analysis revealed that S1P lyase mRNA abundance was reduced by over 80% in single-positive thymocytes with high amounts of GFP (GFPhi) from mice reconstituted with HSCs expressing the shRNA (Fig. 4C and fig. S4). Lyase expression was reduced ∼threefold in a thymocyte suspension composed predominantly of hematopoetic cells, and by about 30% in total thymus tissue (Fig. 4C). Threefold reductions in lyase expression were observed in the spleen (Fig. 4C). Lyase knockdown led to an accumulation of mature single-positive cells in the thymus, mirrored by a T cell deficiency in the spleen and blood (Fig. 4D). Surface S1P1 was partially down-regulated on mature thymocytes, and S1P abundance in the thymus and spleen was elevated as determined by both LC/MS/MS analysis and bioassay (Fig. 4, F to I). The generally less severe phenotype that exists in shRNA-expressing mice, as opposed to that in THI- or DOP-treated mice, most likely occurs because THI and DOP reduce S1P lyase activity in both hematopoietic and nonhematopoietic cells, whereas the knockdown approach affects only a fraction of the hematopoietic cells. Nonetheless, selective RNAi-mediated knockdown of S1P lyase recapitulated key aspects of the THI- and DOP-induced phenotype.

Although there are several mechanisms by which S1P levels can be regulated (8), we provide evidence that within lymphoid tissues, S1P lyase expressed by hematopoetic cells has an essential role in degrading S1P and keeping tonic S1P concentrations low. THI, DOP, and shRNA expression may each have effects unrelated to S1P lyase, but the finding that three independent means of inhibiting the lyase all elevate lymphoid S1P concentrations firmly establishes a role for the lyase in controlling S1P levels. These findings provide evidence that there is normally a large differential in bioavailable S1P abundance between the circulatory fluids—plasma and lymph—and the lymphoid organs. The increased S1P concentration induced by lyase inhibition provides an explanation for the loss of surface S1P1 from mature lymphocytes, and reduced S1P1 function could account for the inhibition of egress from the thymus and peripheral lymphoid organs. It also is possible that exit could be inhibited by a loss of guidance information provided by the S1P gradient or by S1P-mediated changes in other cell types, such as endothelial cells (20). A loss of guidance information has been suggested as an explanation for the germ-cell migration defect in wunen mutant Drosophila embryos, which lack a lipid phosphatase (21). Chronic vitamin-B6 deficiency is associated with lymphopenia and immune deficiency (11, 22), effects that might in part be explained by reduced lyase activity and defective lymphocyte egress. Although our findings demonstrate that THI-ingestion leads to an inhibition of S1P lyase activity, further studies will be needed to define whether THI can bind the lyase directly or whether it first becomes modified in vivo or acts by an indirect mechanism. The disruption of S1P1 function by small molecules such as FTY720 is currently being tested as a mode of immunosuppression in humans (6, 20). Our studies suggest that S1P lyase represents a novel target for immunosuppressive drugs.

Supporting Online Material

www.sciencemag.org/cgi/content/full/309/5741/1735/DC1

Materials and Methods

Figs. S1 to S4

References

References and Notes

View Abstract

Navigate This Article