Report

GRK2-Dependent S1PR1 Desensitization Is Required for Lymphocytes to Overcome Their Attraction to Blood

See allHide authors and affiliations

Science  30 Sep 2011:
Vol. 333, Issue 6051, pp. 1898-1903
DOI: 10.1126/science.1208248

Abstract

Lymphocytes egress from lymphoid organs in response to sphingosine-1-phosphate (S1P); minutes later they migrate from blood into tissue against the S1P gradient. The mechanisms facilitating cell movement against the gradient have not been defined. Here, we show that heterotrimeric guanine nucleotide–binding protein–coupled receptor kinase-2 (GRK2) functions in down-regulation of S1P receptor-1 (S1PR1) on blood-exposed lymphocytes. T and B cell movement from blood into lymph nodes is reduced in the absence of GRK2 but is restored in S1P-deficient mice. In the spleen, B cell movement between the blood-rich marginal zone and follicles is disrupted by GRK2 deficiency and by mutation of an S1PR1 desensitization motif. Moreover, delivery of systemic antigen into follicles is impaired. Thus, GRK2-dependent S1PR1 desensitization allows lymphocytes to escape circulatory fluids and migrate into lymphoid tissues.

Blood and lymph contain high nM amounts of sphingosine-1-phosphate (S1P), and lymphocyte egress from lymphoid tissues is dependent on S1P triggering of S1P receptor-1 (S1PR1) on the lymphocyte (1). Within minutes of arriving in blood, lymphocytes are able to leave again (2). In lymph nodes (LNs), this occurs via high endothelial venules (HEVs) (3, 4). Localized chemokine-mediated Gi [the family of heterotrimeric guanine nucleotide–binding protein (G protein) α subunits that can inhibit adenylyl cyclases] signals activate integrins and are important for cell adhesion and subsequent transendothelial migration (5, 6). Yet S1PR1 is also a Gi-coupled receptor, and its ligand is uniformly abundant in blood. How does a cell avoid being distracted by S1PR1 engagement and achieve the temporally restricted and correctly orientated Gi-signaling needed for LN entry? S1PR1 is down-modulated by ligand exposure in lymph and blood, but the mechanism for and importance of this down-modulation has not been determined (7). We tested the hypothesis that S1PR1 desensitization is required for T and B cells to overcome their attraction to blood and enter lymphoid tissue.

G protein–coupled receptor kinase-2 (GRK2) is essential for mouse development, but the key in vivo targets of this kinase are unknown (810). GRK2 is one of several GRKs expressed in T and B cells (11, 12), and in vitro studies indicate that it can phosphorylate S1PR1 (13, 14). We asked whether GRK2 has a nonredundant role in S1PR1 desensitization within T cells. Thymocyte devlopment in GRK2f/– CD4-cre mice appeared normal, and T cell numbers in the spleen were in the wild-type (WT) range, whereas numbers in LNs were reduced (fig. S1A). GRK2-deficient blood T cells had high surface S1PR1 amounts compared with undetectable amounts on control blood T cells (Fig. 1A). S1PR1 on T cells from LNs was only slightly higher than control levels (Fig. 1A), consistent with low extracellular S1P concentrations in lymphoid organs (15). In contrast, surface expression of CCR7 and CXCR4 on T cells from LNs, sites of chemokine exposure, were similar between GRK2-deficient and WT cells (Fig. 1A). In vitro, GRK2-deficient T cells were resistant to S1P-mediated S1PR1 down-modulation even at μM amounts, in contrast to the low nM sensitivity of WT cells (fig. S1B). GRK2-deficient T cells showed elevated chemotactic responses to S1P (Fig. 1B and fig. S1C), whereas response to CXCL12 was similar to controls and the response to CCL21 was reduced (Fig. 1B). This may indicate that transwell migration assays to S1P are usually limited in magnitude because of receptor desensitization occurring in many cells before they reach the lower chamber.

Fig. 1

GRK2-deficient T cells are resistant to S1PR1 desensitization and have a reduced ability to enter LNs. (A) Flow-cytometric analysis of blood and peripheral LN (pLN) CD4 T cells from WT and GRK2f/– CD4-cre (knock-out, KO) mice for S1PR1, CCR7, and CXCR4. Shaded histrogram shows background (bg) staining of negative control. (B) Transwell migration of WT and KO CD4 T cells to S1P, CXCL12, or CCL21. Three replicates were run for each group. Bars indicate mean (±SEM). (C) Short-term in vivo homing efficiency of GRK2 KO T cells relative to cotransferred WT T cells. Transferred B cells were tracked as an additional control. bld, blood; spln, spleen; mLN, mesenteric LN. Data are from three experiments (n = 8 mice). Bars indicate mean (±SEM). (D) Short-term in vivo homing efficiency of KO T cells relative to cotransferred WT T cells in WT and S1P-deficient recipients. Data are from three experiments (n = 8 mice). Bars indicate mean (±SEM). (E) Rolling-to-sticking transition frequency and rolling velocity (Vroll) of transferred WT and KO T cells visualized by intravital microscopy in pLN HEVs. Summary of three experiments (n = 3 mice). Bars indicate mean (±SEM). *P < 0.05, **P < 0.005; ***P < 0.0005; ns, not significant (P > 0.05) by unpaired Student’s t test.

In 20-min and 20-hour cotransfer experiments, GRK2-deficient T cells showed a defect in LN entry while being represented similarly to control T cells in blood and spleen (Fig. 1C). Transfer into mice lacking circulatory S1P (15) overcame the entry defect (Fig. 1D), establishing that it was a consequence of S1P receptor signaling and was not due to a role of GRK2 in CCR7 signaling or in other pathways. Deficient and control T cells in the blood of recipient mice had comparable integrin levels (fig. S1D). To test for evidence of constitutive S1PR1 signaling by GRK2-deficient T cells in blood, we examined phosphorylation of ezrin/radixin/moesin (pERM) proteins because Gi signaling is sufficient to decrease pERM levels (16). A small but significant reduction in pERM was detected in GRK2-deficient blood T cells (fig. S1E). Reduced pERM was not observed in GRK2-deficient LN T cells (fig. S1E) or in blood cells after transfer into S1P-deficient recipients (fig. S1E). Despite the reduced LN entry efficiency, GRK2f/– CD4-cre mice showed reduced rather than increased T cell numbers in blood, indicating that GRK2 is needed for T cell homeostasis (fig. S1A), perhaps reflecting a need to access LN-restricted trophic signals (17) or other undefined functions of GRK2 in T cells (10). Video-rate microscopy of LN HEVs revealed a three- to fourfold reduction in GRK2-deficient T cells undergoing the rolling-to-sticking transition compared with controls (Fig. 1E). Rolling velocity was also reduced (Fig. 1E), perhaps reflecting alterations in membrane surface properties because of constitutive S1P signaling (fig. S1E) or the slightly elevated CD62L expression (fig. S1D).

Flow-cytometric analysis of GRK2f/– mice carrying a B cell–specific (Mb1-) cre demonstrated that B cells had a similar GRK2-dependence as T cells for ligand-mediated S1PR1 down-modulation in blood (Fig. 2A). The specificity of the antibody was confirmed by the lack of staining on S1PR1-deficient B cells (fig. S2A). Again the effect of GRK2 deficiency on surface GPCR expression showed selectivity for S1PR1, because expression of CXCR4, CXCR5, and CCR7 was similar to that in controls (Fig. 2A). GRK2-deficient B cells showed augmented chemotactic responses to S1P, whereas responses to chemokines were in the normal range (Fig. 2B and fig. S2B). GRK2f/– Mb1-cre mice had elevated splenic and blood B cell numbers but reduced numbers in LNs (fig. S2C). Unexpectedly, in short-term (30-min) transfer experiments GRK2-deficient B cells accumulated in recipient LNs similarly to controls (Fig. 2C); however, after 20 hours fewer GRK2-deficient than control B cells had accumulated (Fig. 2C). When the B cells were resident for 20 hours in S1P-deficient hosts (15), GRK2-deficient and control cells entered LNs in similar numbers (Fig. 2D). The GRK2-deficient cells expressed similar amounts of integrins and L-selectin as control cells (fig. S2D). We then asked whether an early entry defect in GRK2-deficient B cells might be obscured by the attachment of large numbers of transferred B cells to HEVs. To measure the fraction of B cells that had entered the tissue versus remained within the HEVs, we treated mice for a few minutes with a phycoerythrin (PE)-conjugated antibody, a procedure that allows selective labeling of blood-exposed cells (fig. S2E) (18). This analysis revealed that a greater fraction of GRK2-deficient than WT LN B cells remained blood-exposed 30 min after transfer (Fig. 2, E and F). This effect was S1P-dependent because it was lost in mice lacking circulatory S1P (Fig. 2F). A similar experiment with GRK2-deficient T cells did not reveal enrichment for blood-exposed cells (fig. S2F). Quantitative real-time measurement of HEV attachment is more difficult for B cells than T cells because of a ~10-fold lower rate of LN entry (19, 20). Although we have not excluded a possible effect of GRK2 deficiency on B cell rolling-to-sticking transitions, it appears that postadhesion transmigration may be more dependent on S1PR1 desensitization in B cells than in T cells.

Fig. 2

GRK2-deficient follicular B cells are resistant to S1PR1 desensitization and have a reduced ability to enter LNs. (A) Flow-cytometric analysis of blood and peripheral LN follicular B cells from WT and GRK2f/– Mb1-cre mice (KO) for S1PR1, CXCR5, CXCR4, and CCR7. (B) Transwell migration of WT and KO B cells to S1P, CXCL12, and CXCL13. Three replicates were run for each group. Bars indicate mean (±SEM). (C) Short-term homing efficiency of KO B cells relative to cotransferred WT B cells. Transferred T cells are tracked as an additional control. Summary of three experiments (n = 8 mice). Bars indicate mean (±SEM). (D) Short-term homing of GRK2-deficient B cells relative to cotransferred WT B cells in WT and S1P-deficient recipients. Summary of four experiments (n = 11 mice). Bars indicate mean (±SEM). (E) Flow-cytometric detection of 30-min transferred B220+ B cells and CD3+ T cells labeled by 2 min in vivo anti–CD19-PE and anti–CD4/8-PE treatment. (F) Short-term homing efficiency of KO B cells relative to cotransferred WT B cells in WT (n = 10) and S1P-deficient (n = 6) recipients, showing cells that were protected from anti–CD19/4/8-PE exposure (PE-) or blood exposed (PE+), identified as in (E). Summary of three experiments. Bars indicate mean (±SEM).

The spleen is unusual in having an open circulation, with blood being released in an area that surrounds the follicles called the marginal zone (MZ) (21). Specialized B cells reside there, termed MZ B cells, and they help deliver antigens into the follicles to promote humoral immunity (2123). MZ B cell positioning in the MZ requires S1PR1 in the B cell to overcome attraction to follicular CXCL13 (24). We next asked whether MZ B cell movement from MZ into follicle requires GRK2-dependent S1PR1 desensitization.

MZ B cells from Mb1-cre GRK2-deficient mice had elevated surface S1PR1, whereas CXCR4, CXCR5 (Fig. 3A), and other markers were expressed normally (fig. S3A). In transwell migration assays, GRK2-deficient MZ B cells showed increased migration to S1P (Fig. 3B). CCL21 responses were reduced, like in T cells, suggesting GRK2 may augment CCR7 function in some cell types (fig. S2B). MZ B cells also express S1PR3 (23, 24). GRK2 deficiency augmented S1PR1 function, however, because the MZ B cells showed elevated responses to SEW2871 (Fig. 3B), an S1PR1-selective agonist (25).

Fig. 3

MZ B cell migration into follicles requires GRK2-dependent S1PR1 desensitization. (A) Flow-cytometric analysis of splenic MZ B cells from WT and GRK2f/– Mb1-cre (KO) mice for S1PR1, CXCR5, and CXCR4. (B) Transwell migration of WT and KO MZ B cells to S1P, CXCL12, CXCL13, and S1PR1 agonist SEW2871. Three replicates were run for each group. Bars indicate mean (±SEM). (C) Flow-cytometric detection of 5 min in vivo anti–CD19-PE labeled MZ B cells from WT, GRK2f/– Mb1-cre (KO), and GRK2+/− (Het) mice, shown as a histogram overlay (left) and a summary of data from four experiments (WT n = 28, KO n = 15, Het n = 13 mice) (center). Right graph shows a summary of data from three experiments for WT and GRK2f/– Mb1-cre (KO) cells present in mixed BM chimeras (n = 8). Bars indicate mean (±SEM). (D) Flow-cytometric detection of 5 min in vivo anti–CD19-PE labeled MZ B cells from WT, GRK2+/− (GRK2 het), S1PR1+/− (S1PR1 het), and GRK2+/− S1PR1+/− (GRK2/S1PR1 het/het) mice shown as a histogram overlay (left) and a summary of data from five mice (right). Bars indicate mean (±SEM). (E) Immunohistochemical localization of CD1dhi MZ B cells (blue) with respect to the MAdCAM1+ marginal sinus (brown) in spleen sections from GRK2+/+ Mb1-cre (WT) and GRK2f/–Mb1-cre (KO) mice. FO, follicle.

To quantitate MZ B cell distribution, we labeled blood-exposed B cells by injection of CD19-PE antibody. After 5 min, ~55% of WT MZ B cells were labeled (Fig. 3C); the remaining cells were protected from rapid antibody exposure because of their follicular localization (23). By contrast, ~80% of the GRK2-deficient MZ B cells became antibody-labeled, indicating that a significantly greater fraction of these cells were situated in the MZ (Fig. 3C). An intermediate effect was observed in GRK2+/− B cells (Fig. 3C). In S1PR1/GRK2 compound heterozygous mice, MZ B cell distribution returned to a state similar to WT (Fig. 3D). Accumulation of MZ B cells in the MZ was not due to indirect effects on other cell types because there was a similar increase in antibody labeling of GRK2-deficient cells in mixed WT:GRK2f/–Mb1-cre BM chimeras (Fig. 3C). FTY720 treatment caused rapid down-modulation of S1PR1 and relocalization of WT MZ B cells to the follicle (fig. S3B) (24). These events were delayed in GRK2-deficient MZ B cells (fig. S3B), consistent with a role for GRK2 in FTY720-mediated S1PR1 down-modulation (26) and suggesting that S1PR1 function is needed in GRK2-deficient cells for retention in the MZ. In sections, GRK2-deficient MZ B cells were restricted to the MZ, whereas control cells were distributed in both MZ and follicle (Fig. 3E).

Previous studies have identified several regions in the S1PR1 C terminus that contribute to desensitization (13, 2628). We established a role for the last 12 amino acids in supporting desensitization in a B cell line, and mutating the Thr-Ser-Ser (TSS) motif in this region to Ala-Ala-Ala (AAA) (fig. S4A) led to a loss in sensitivity for S1P-mediated receptor down-modulation (fig. S4B). Mice carrying the TSS-to-AAA mutation in the endogenous S1PR1 locus (S1PR1TSS; fig. S4C) had normal follicular and MZ B cells numbers (fig. S4D). By flow cytometry, MZ B cells from S1PR1TSS mice showed higher S1PR1 expression than control cells (Fig. 4A). In vitro, S1PR1TSS MZ B cells were resistant to S1P-mediated S1PR1 internalization (fig. S4E). In chemotaxis assays, MZ B cells showed augmented responses to S1P while retaining normal responses to CXCL12 and CXCL13 (Fig. 4B). In vivo CD19-PE labeling revealed an accumulation of MZ B cells in the MZ (Fig. 4C). In sections, MZ B cells in S1PR1TSS mice were restricted to the MZ (Fig. 4D). Transcriptional down-modulation of S1PR1 by lipopolysaccharide treatment (24) caused a similar follicular repositioning of WT and S1PR1TSS MZ B cells (fig. S4F). These observations establish a requirement for S1PR1 desensitization in MZ B cell movement between MZ and follicle. Recirculating S1PR1TSS T and B cells were not as desensitization-resistant as GRK2-deficient T and B cells, because they showed almost complete S1PR1 down-modulation in blood (fig. S4G) and entered LNs with normal efficiency (fig. S4H), consistent with the presence of additional desensitization motifs in S1PR1 (13, 2628). The different sensitivities of MZ and recirculating lymphocytes to the TSS mutation may reflect partially distinct modes of desensitization or be a consequence of lower S1P concentrations in MZ than blood.

Fig. 4

MZ B cell shuttling and immune complex delivery into follicles is disrupted by mutation of an S1PR1 desensitization motif. (A) Flow-cytometric analysis of S1PR1 expression on MZ B cells from freshly isolated WT and S1PR1TSS (TSS) mice. (B) Transwell migration of WT and TSS MZ B cells to S1P, CXCL12, and CXCL13. Three replicates were run for each group. Bars indicate mean (±SEM). (C) Flow-cytometric detection of 5 min in vivo anti–CD19-PE labeled MZ B cells from the indicated mice shown as a histogram overlay (left) and a summary of data from nine mice. (D) Immunohistochemical localization of CD1dhi MZ B cells (blue) with respect to the MAdCAM1+ marginal sinus (brown) in spleen sections from WT and TSS mice. (E) Surface S1PR1 internalization and recovery on MZ B cells exposed to 1 μM S1P (internalization) or removed from S1P exposure (recovery) for 10, 30, or 60 min. Shaded histrogram shows background (bg) staining of negative control. (F) Flow-cytometric analysis of S1PR1 and CXCR5 expression on 5 min in vivo CD45.2 antibody-labeled (MZ) and unlabeled (FO) MZ B cells. Shaded histrogram shows background (bg) staining of negative control. Right graph shows ratio of S1PR1 or CXCR5 mean fluorescence intensities (MFI) of MZ B cells in the two compartments from three independent experiments. (G) Immunofluorescence microscopy of spleen sections taken from the indicated mice 18 hours after treatment to form systemic PE-ICs (red), stained to detect marginal sinus-associated macrophages (CD169+, green) and FDCs [complement receptor 1 (CR1)+, blue]. All data are representative of at least three mice of each type.

In vivo labeling studies suggested that many MZ B cells migrate back and forth between MZ and follicle in an hour (23). Incubation of MZ B cells in blood concentrations of S1P led to partial down-regulation of S1PR1 within 10 min and complete modulation by 30 min (Fig. 4E). Reciprocally, when MZ B cells were removed from exogenous S1P there was an increase in surface S1PR1 expression within 10 min, reaching maximum levels by ~30 min (Fig. 4E). In vivo, average S1PR1 surface levels were slightly lower on MZ B cells in the MZ compared with MZ B cells in the follicle, whereas CXCR5 levels were the same (Fig. 4F). These observations suggest MZ B cell movement between MZ and follicle because of S1PR1 desensitization in the S1P-high MZ and resensitization in the S1P-low follicle could occur in periods under an hour.

MZ B cells mediate delivery of immune complexes (ICs) to splenic follicular dendritic cells (FDCs) (2123). When control mice were treated to generate systemic PE-containing ICs, complexes could be visualized on MZ B cells within 1 hour (fig. S5), and by 18 hours they had been deposited on FDCs (Fig. 4G). In GRK2f/– Mb1-cre mice, although similar amounts of PE-ICs were initially found on MZ B cells (fig. S5A), deposition on FDCs was reduced (Fig. 4G). A similar defect in IC deposition on splenic FDCs was observed in S1PR1TSS mice (Fig. 4G). By contrast, B cell GRK2-deficient and S1PR1TSS mice showed a similar ability to their matched controls to deposit subcutaneously forming ICs on draining LN FDCs (fig. S5C), a process that does not involve MZ B cells (29).

We identify a central role for GRK2 in ligand-mediated desensitization of lymphocyte S1PR1 (fig. S6). In vitro studies indicate that Gi signals need to be received in a polarized fashion for T cells to undergo shear resistant adhesion (5, 6). By demonstrating that nonpolarized S1PR1-mediated Gi signaling disrupts the rolling-to-sticking transition in T cells, our data provide in vivo support for this model. Nonpolarized S1PR1 signaling in GRK2-deficient cells also likely distracts cells from undergoing the polarized migration needed to cross the endothelium. Our results are also consistent with a model where cycles of S1PR1 de- and resensitization form the basis of a migratory oscillator that supports movement of MZ B cells between MZ and follicle (fig. S6). A general implication of these findings is that GRK2 antagonists may suppress lymphocyte migration from blood into tissue and thus could have therapeutic potential as immunosuppressants.

Supporting Online Material

www.sciencemag.org/cgi/content/full/333/6051/1898/DC1

Materials and Methods

Figs. S1 to S6

References (3036)

References and Notes

  1. Acknowledgments: We thank M. Caron and M. von Zastrow for GRK2+/− mice; J. Green for helpful input; and O. Bannard, J. Green, A. Reboldi, and S. Rosen for comments on the manuscript. T.I.A. was a Jane Coffin Childs Memorial Fellow, and J.G.C. is an Investigator of the Howard Hughes Medical Institute. This work was supported in part by NIH grant AI74847 and by a Research Exchange Grant with Osaka University. The data reported in this paper are tabulated in the main text and in the supporting online material. Use of both the GRK2f/f mice and S1PR1TSS mice will require a material transfer agreement.

Stay Connected to Science

Navigate This Article