Sphingosine 1-Phosphate (S1P) Receptors 1 and 2 Coordinately Induce Mesenchymal Cell Migration through S1P Activation of Complementary Kinase Pathways

Background: Coupling of bone degradation to subsequent bone formation requires recruitment of osteoblast precursors.

Results: Osteoclasts produce sphingosine 1-phosphate (S1P), which stimulates mesenchymal (skeletal) stem cell migration by activating kinase signaling pathways.

Conclusion: Coupling of bone resorption to bone formation involves S1P-mediated recruitment of osteoblastic cells.

Significance: Enhancing kinase signaling in osteoblastic cells may be a novel approach to enhance bone formation.

Abstract

Normal bone turnover requires tight coupling of bone resorption and bone formation to preserve bone quantity and structure. With aging and during several pathological conditions, this coupling breaks down, leading to either net bone loss or excess bone formation. To preserve or restore normal bone metabolism, it is crucial to determine the mechanisms by which osteoclasts and osteoblast precursors interact and contribute to coupling. We showed that osteoclasts produce the chemokine sphingosine 1-phosphate (S1P), which stimulates osteoblast migration. Thus, osteoclast-derived S1P may recruit osteoblasts to sites of bone resorption as an initial step in replacing lost bone. In this study we investigated the mechanisms by which S1P stimulates mesenchymal (skeletal) cell chemotaxis. S1P treatment of mesenchymal (skeletal) cells activated RhoA GTPase, but this small G protein did not contribute to migration. Rather, two S1P receptors, S1PR1 and S1PR2, coordinately promoted migration through activation of the JAK/STAT3 and FAK/PI3K/AKT signaling pathways, respectively. These data demonstrate that the chemokine S1P couples bone formation to bone resorption through activation of kinase signaling pathways.

Introduction

During longitudinal growth, osteoclast-mediated bone resorption is required to expand the marrow cavity whereas one formation by osteoblasts creates larger bones. In adults, the cycle of resorption and formation repairs damaged bones and modulates systemic and local calcium needs. In young adults, the rate of bone resorption and subsequent bone formation is tightly controlled in that resorbed bone is precisely replaced in both location and amount. This concept has been termed coupling (1, 2). With aging, coupling becomes unbalanced such that the increase in osteoclast numbers and activity cannot be met with equal bone formation, resulting in a net loss of bone. Because age-related bone loss is such a prevalent occurrence that encompasses nearly half of the human population worldwide, it is crucial to determine the mechanisms by which bone formation is linked to bone resorption to design the most effective therapies to prevent uncoupling.

A necessary early step in coupling is the recruitment of osteoblast progenitors to the bone surface through stimulating their migration to bone. We documented that osteoclasts secrete sphingosine 1-phosphate (S1P)² (3). S1P present in conditioned medium from cultured osteoclasts stimulates random cell movement (chemokinesis) in mesenchymal cells (3). S1P was also found to stimulate directed migration of cancer cells, including prostate, myeloma, thyroid, and breast (4–8). Many nontransformed cells in the body such as endothelial cells, stem cells, B cells, T cells, muscle cells, dendritic cells, and osteoclast precursors respond to S1P with a chemotactic response (9–14). We therefore investigated whether osteoclast S1P is chemotactic for mesenchymal (skeletal) stem cells (MSCs) and the mechanisms by which S1P stimulates MSC migration.

EXPERIMENTAL PROCEDURES

Unless otherwise indicated, all chemicals were from Sigma.

Osteoclast Culture and Conditioned Media Preparation

Six- to 8-week-old C57BL/6 mice (Jackson Laboratories) were killed, and bone marrow was harvested as we have reported previously (3). Red blood cells were lysed, and the remaining bone marrow cells were cultured in α-Minimal Essential Medium (αMEM) supplemented with 10% (v/v) fetal bovine serum (FBS) and 25 ng/ml macrophage colony-stimulating factor (M-CSF) and incubated at 37 °C, in 5% CO₂. One day after isolation, adherent mesenchymal cells were discarded, and the nonadherent cells were cultured in 100 ng/ml receptor activator of NFκB (RANKL) and 25 ng/ml M-CSF for 4 days with a refeeding on day 3. For migration assays, mature osteoclast-conditioned media were harvested and centrifuged to remove cell debris. The conditioned media were stored at −80 °C until assayed. All protocols were approved by the Mayo Clinic IACUC prior to the start of the studies.

MSC Culture and Migration Assay

Human bone marrow-derived MSCs overexpressing human telomerase reverse transcriptase (hMSC-TERT) were employed as a model for hMSC. Development and characterization of this cell line have been described previously (15, 16).

Cultures were maintained in αMEM supplemented with 10% FBS (base medium). For migration experiments, 95% confluent cells were harvested and assayed using the QCM 24-well Colorimetric Cell Migration Assay (Millipore). hMSC-TERT cells in base medium with or without pretreatment (see below) were added to the inserts. Base or conditioned media with or without additives (as detailed below and in the figure legends) were added to the lower chamber. The assembled assay was incubated at 37 °C for 6 h. Cell migration through the insert membrane was quantitated by staining cells with the provided reagent, cells on the upper side of the membrane were removed, and the cell stain was extracted and quantitated using absorbance at 560 nm. The following additions were made for various experiments: S1P receptor agonist VPC 24191 (5 μm, Avanti Polar Lipids); for hMSC-TERT cell treatments: 1 μl/ml vehicle (dimethyl sulfoxide (DMSO)) or pharmacologic inhibitors of AKT (phosphatidylinositol ether analog; Calbiochem; 5 μm), focal adhesion kinase (FAK) (PF-573228; Tocris Bioscience; 10 nm), PI3K (LY294002; 50 μm), JAK (AG490; 100 nm), STAT3 (WP1066; Millipore; 5 μm), or RhoA (CCG-1423; Cayman Chemical; 1 μm) were added to the cells at 37 °C for 15 min prior to assay assembly. In some experiments vehicle (5% acidified DMSO in H₂O), combined S1PR1,3 (VPC23019, 100 nm, Avanti Polar Lipids), and S1PR2 (JTE-013, 20 nm, Tocris Bioscience) antagonism or selective receptor antagonists were employed. The selective antagonists were S1PR1 antagonist W123, 1 μm, Cayman Chemical; S1PR2 antagonist JTE-013, 20 nm; S1PR3 antagonist BML-241, 10 μm, Cayman Chemical. Receptor antagonists were added to the cells just prior to assay assembly. For adenoviral treatment, 18 h prior to harvesting, hMSC-TERT cells were infected with constitutively active AKT (Vector Biolaboratories) at a multiplicity of infection of 10. For lower chamber additions, vehicle (1 μl/ml 5% acidified DMSO in H₂O) or S1P receptor agonist (3) was added.

Quantitative Real Time Polymerase Chain Reaction

Cells were rinsed with PBS and RNA harvested at times indicated in the figure legends using the RNeasy micropurification kit (Qiagen) according to the product literature. Following quantitation, cDNA was synthesized, and real time PCR analysis was carried out as we have reported (3). Primers were: S1PR1 forward, 5′-GTGTAGACCCAGAGTCCTGCG-3′ and reverse, 5′-AGCTTTTCCTTGGCTGGAGAG-3′; S1PR2 forward, 5′-GGCCTAGCCAGTGCTCAGC-3′ and reverse, 5′-CCTTGGTGTAATTG-3′; S1PR3 forward, ′-GGAGCCCCTAGACGGGAGT-3′ and reverse, 5′-CCGACTGCGGGAAGAGTGT-3′; tubulin A1A forward, 5′-GAGTGCATCTCCATCCACGTT-3′ and reverse, 5′-TAGAGCTCCCAGCAGGCATT-3′. Messenger RNA levels were calculated using the ΔΔ-Ct method (3).

Rho GTPase Family Activation Assay

hTERT cells were plated in 6-well plates. When 95% confluent, wells containing base medium were treated with vehicle (1 μl/ml 5% acidified DMSO in H₂O) or the S1P agonist. Other wells were switched to osteoclast-conditioned medium and treated with vehicle or combined S1PR1 and S1PR2 receptor antagonist for 10 min at 37 °C. Cells were processed and assayed for RhoA, Rac1, and Cdc42 activity using a kit from Cell Biolabs according to the instructions. Briefly, cell lysates were incubated with agarose beads coupled to either Rhotekin Rho binding domain peptide (RhoA assay) or p21 binding domain of p21-activated protein kinase (Rac1 and Cdc42 assays). Separate aliquots of lysate were incubated for 30 min at 30 °C with nonhydrolyzable GTPγS to activate all GTPases prior to mixing with the agarose assay beads. After incubation on ice for 60 min, the beads were spun down and repeatedly washed. The agarose beads were suspended in Western blot sample buffer, and the presence of bound activated GTPases was evaluated by Western blotting with specific antibodies to the GTPases.

Western Blot Analysis

hMSC-TERT cells were plated in 100-mm tissue culture dishes. When 95% confluent, cells were rinsed three times with phosphate-buffered saline and serum-starved for 6 h in αMEM with 1% bovine serum albumin. Cells were treated with the indicated substances for the indicated time as detailed in the figure legends using reagents detailed above for migration assays. Cell extracts were harvested and protein concentrations determined with the Bio-Rad DC Protein Assay kit as instructed. Proteins (40 μg) were separated using 10% SDS-PAGE followed by electroblotting to Immobilon-P membranes (Millipore). Membranes were probed as described with antibodies to mouse phosphorylated signaling antibodies (3). All antibodies were from Cell Signaling. Lane loading was monitored by reprobing blots for tubulin (University of Iowa Hybridoma Bank). Signals were visualized using the ECL Plus detection system (Amersham Biosciences) according to the manufacturer's instructions.

Statistics

Each experiment had at least three replicates and was repeated at least three times. These results are representative of the repeats. Data were analyzed using a one-way analysis of variance compared with controls as indicated in each figure legend and are presented as mean ± S.E. Significance was determined at p < 0.05 using KaleidaGraph software (Synergy Software, Reading PA).

RESULTS

Osteoclasts Secrete S1P to Promote Chemotaxis of Mesenchymal Cells

Coupling requires recruitment of osteoprogenitors to the location of bone resorption through chemotaxis, or directed migration. Previously, we showed that osteoclasts promote MSC chemokinesis and that movement was reduced with an antagonist the blocks S1P-receptor interactions (3). Here we investigated whether secreted S1P induces MSC chemotaxis. Osteoclast-conditioned medium induced MSC chemotaxis and S1P-receptor antagonists blocked this response (Fig. 1A). MSC responded to a S1P agonist added to the base medium with increased migration, confirming that S1P is a chemotactic agent for mesenchymal cells (Fig. 1B). We examined expression of the three S1P receptors S1PR1, S1PR2, and S1PR3 in MSC cultures (Fig. 2). As cultures reach confluence, S1PR1 and S1PR2 expression increased significantly (Fig. 2A). S1PR3 expression was highly variable. We employed receptor-selective antagonists to resolve which of these was involved in the mesenchymal response (Fig. 2B). Blocking either S1PR1 or S1PR2 reduced S1P migration stimulation, and combining S1PR1 and S1PR2 antagonists further reduced migration to base media levels. We did not detect any role for S1P3 in mediating S1P influences on migration (data not shown).

FIGURE 1.

Osteoclast S1P stimulates osteoblast migration. A, base medium (BASE) or osteoclast conditioned medium (OC CM) was placed in the bottom of the migration chamber. hMSC-TERT cells in base medium were treated with either vehicle (VEH; 5% DMSO) or combined S1P receptors 1, 2, and 3 antagonist (VPC 23019 and JTE-013) and placed in the migration inserts as detailed under “Experimental Procedures.” The chambers were incubated and analyzed by staining cells that migrated through the membrane within 6 h as detailed under “Experimental Procedures.” *, p < 0.05 compared with Base + vehicle; **, p < 0.05 compared with OC CM + vehicle. B, base media alone, with vehicle (5% DMSO), or a S1P receptor agonist (VPC 24191) was placed in the bottom of the migration chamber. hMSC-TERT cells in base medium was placed in the migration inserts as detailed under “Experimental Procedures.” The chambers were incubated for 6 h, and cells that migrated through the membrane were analyzed as described under “Experimental Procedures.” *, p < 0.05 compared with vehicle or no treatment.

FIGURE 2.

S1P receptor involvement in hMSC-TERT migration response. A, hTERT cells were plated at 10% confluence cultured for 1–4 days in base medium as detailed under “Experimental Procedures.” Cells were confluent on day 4. Cells were harvested, RNA was isolated, and mRNA levels of the S1P receptors were quantitated by quantitative real time PCR as described under “Experimental Procedures.” *, p < 0.05 compared with day 1. B, base medium or osteoclast conditioned medium (OC CM) was placed in the bottom of the migration chamber. hMSC-TERT cells in base medium were treated with either vehicle (5% DMSO) or a selective S1P receptor antagonist to S1PR1, S1PR2, or S1PR1 and S1PR2 (W123, JTE-013, and BML-241) and placed in the migration inserts as detailed under “Experimental Procedures.” The chambers were incubated for 6 h, and cell migration through the membrane was analyzed as detailed under “Experimental Procedures.” *, p < 0.05 compared with BASE; **, p < 0.05 compared with VEH; ***, p < 0.05 compared with vehicle or single inhibitors.

Rho GTPase and Kinase Signaling Involvement in S1P-induced Migration of Mesenchymal Cells

S1P receptors are G protein-coupled receptors that activate several GTPases (for review, see Ref. 17). To determine how S1P promoted MSC chemotaxis, the Rho GTPase family was evaluated (Fig. 3). RhoA was rapidly activated in MSC cultured in base medium containing the S1P agonist or cultured with osteoclast-conditioned media (Fig. 3A, second and third lanes). Antagonists of S1PR1 and S1PR2 combined reduced conditioned media mediated RhoA activation (Fig. 3A, fourth lane). We were unable to detect any S1P activation of Rac1 or Cdc42 (data not shown). To determine the role of RhoA GTPase in S1P stimulation of migration, mesenchymal cells were treated with a selective RhoA inhibitor prior to assessing mesenchymal cell migration in response to the S1P agonist (Fig. 3B). Rho inhibition did not alter mesenchymal cell migration in response to S1P.

FIGURE 3.

Rho GTPase activation. A, hMSC-TERT cells were treated with either BASE with vehicle or S1P agonist or osteoclast conditioned medium (OC CM) combined with either vehicle or a S1P receptor antagonists (S1PR1 and S1PR2 antagonists together) (W123 and JTE-013) for 10 min. Cells were rinsed, harvested, and lysed. The lysate was split in half and one aliquot incubated with GTPγS at 30 °C for 30 min to activate all GTPases. GTPγS and untreated aliquots were individually incubated with agarose beads coupled to selective GTPase-binding peptides. The beads were washed and analyzed for Western blotting for bound Rho family GTPase by Western blotting as described under “Experimental Procedures.” The top set of lanes are RhoA activation by the indicated treatment, and the bottom set of lanes are the aliquots of the respective lysates incubated with GTPγS to activate all RhoA present in the samples. B, base medium with either vehicle (5% DMSO) or a S1P agonist (VPC 24191) was placed in the bottom of the migration chamber. hMSC-TERT cells were lifted and pretreated with either DMSO (vehicle) or Rho GTPase inhibitor as described under “Experimental Procedures.” Cells were placed in the migration inserts, and the chambers were incubated 6 h prior to analysis for cell migration through the membrane as detailed under “Experimental Procedures.” *, p < 0.05 compared with vehicle treatment.

Another key mediator of migration that is activated by S1P is FAK) (for review, see Ref. 18), which is an upstream activator of the PI3K/AKT signaling pathway (for review, see Ref. 19). We therefore examined S1P influences on FAK/AKT activation and observed rapid activation of both FAK and AKT (Fig. 4A). Because the JAK/STAT pathway has been implicated in migration, we also examined mesenchymal S1P responses for evidence of activation of this pathway (20). We observed rapid phosphorylation of JAK1 and STAT3 in response to S1P treatment (Fig. 4A). We did not detect phosphorylation of JAK2, JAK3, or STAT5 (data not shown). Inhibition of AKT, FAK, PI3K, JAK, or STAT3 blocked S1P-induced migration of mesenchymal cells (Fig. 4B).

FIGURE 4.

Kinase pathway activation and roles in migration. A, hMSC-TERT cells were treated for the indicated time with S1P agonist (VPC 24191) prior to harvesting as described. Forty μg of cell extract was analyzed for the indicated phospho protein or total protein as indicated. B, base medium with either vehicle (5% DMSO) or the S1P agonist (VPC 24191) was placed in the bottom of the migration chamber. hMSC-TERT cells were lifted and pretreated with either DMSO (vehicle) or the indicated kinase inhibitor as described under “Experimental Procedures.” Cells were placed in the migration inserts, and the chambers were incubated for 6 h prior to analysis of cell migration through the membrane as detailed under “Experimental Procedures.” *, p < 0.05 compared with vehicle treatment.

S1PR1 and S1PR2 Coordinately Activate Kinase Signaling Pathways (Summarized in Fig. 8)

FIGURE 8.

Schematic of coupling and S1P signaling in mesenchymal cells. Osteoclast SPHK generates S1P, which activates receptors S1PR1 and S1PR2 on mesenchymal cells. S1PR1 activated the JAK/STAT pathway, and S1PR2 activates the FAK/PI3K/AKT pathway to stimulate MSC migration.

To investigate the mechanisms of pathway activation, we co-treated mesenchymal cells with the S1P agonist and receptor-selective antagonists (Fig. 5). Based on our results documenting that S1P activated S1PR1 and S1PR2, but not S1PR3, we surmised that co-treatment with S1P and blocking S1PR2 would allow activation of only S1PR1 whereas blocking S1PR1 would allow activation of only S1PR2. S1PR2 antagonists blocked phosphorylation of FAK and AKT, indicating that S1PR1 activated JAK/STAT signaling (Fig. 5A). In contrast, S1PR1 antagonists blocked phosphorylation of JAK1 and STAT3, supporting S1PR2 activation of FAK/PI3K/AKT signaling (Fig. 5B). Inhibiting FAK, PI3K, or AKT had no impact on migration when S1PR1 was activated by S1P, but blocking JAK or STAT3 inhibited the migratory response (Fig. 6A). Blocking AKT, FAK, or PI3K inhibited migration stimulated by S1PR2 whereas blocking either JAK or STAT3 had no impact on the migratory response to S1P (Fig. 6B). JAK has been reported to activate the PI3K/AKT pathway in cancer cells (21). To determine whether FAK was upstream or downstream of PI3K and AKT signaling, the ability of constitutively active PI3K to overcome targeted pathway deletion was analyzed (Fig. 7A). Activated PI3K overcame FAK inhibition but was unable to overcome inhibition of AKT, JAK, or STAT3. To evaluate whether there was cross-talk between signaling pathways in mesenchymal cells in response to S1P, the ability of constitutively active AKT to overcome targeted pathway activation was evaluated (Fig. 7B). Constitutively active AKT overcame FAK or PI3K inhibition, but was unable to overcome inhibition of either JAK or STAT3. These data demonstrate that JAK/STAT and FAK/PI3K/AKT signaling independently coordinate to promote mesenchymal cell migration in response to S1P.

FIGURE 5.

Receptor-selective kinase pathway activation. hMSC-TERT cells in base media were pretreated for 15 min with either DMSO (vehicle) or the indicated inhibitor and a S1P receptor antagonist (receptor antagonist to S1PR2 (JTE-013), thus measuring the impacts of S1PR1 activation (A), and receptor antagonist to S1PR1 (W123), thus measuring the impacts of S1PR2 activation (B)). Cells were treated for 10 min with vehicle (5% DMSO) or the S1P agonist prior to harvesting. Forty μg of cell extract was analyzed for the indicated phospho protein or total protein as indicated.

FIGURE 6.

Role of kinase pathways in S1P receptor stimulation of migration. Base medium with either vehicle (5% DMSO) or a S1P agonist (VPC 24191) was placed in the bottom of the migration chamber. hMSC-TERT cells were lifted and pretreated with either vehicle (DMSO) or the indicated kinase inhibitor and S1P receptor antagonist (receptor antagonist to S1PR2 (JTE-013), thus measuring the impacts of S1PR1 activation (A), and receptor antagonist to S1PR1 (W123), thus measuring the impacts of S1PR2 activation (B)) as indicated. Cells were placed in the migration inserts, and the chambers were incubated for 6 h prior to analysis of cell migration through the membrane as detailed under “Experimental Procedures.” *, p < 0.05 compared with combined agonist, S1PR inhibitor, and vehicle treatment (the fourth column from the left).

FIGURE 7.

Roles of FAK and AKT in receptor signaling. hMSC-TERT cells were infected with vector or constitutively active PI3K (A) or AKT (B) adenovirus for 24 h. Base medium with either vehicle (5% DMSO) or a S1P agonist (VPC 24191) was placed in the bottom of the migration chamber. Infected hTERT cells were lifted and pretreated with either DMSO (vehicle) or the indicated kinase inhibitor as described. Cells were placed in the migration inserts and incubated and analyzed as detailed under “Experimental Procedures.” *, p < 0.05 compared with agonist plus vehicle treatment. Error bars indicate other significant differences.

DISCUSSION

Sphingosine kinases (SPHKs) are lipid kinases related to diacylglyceraol kinases or ceramide kinases and are evolutionarily conserved from yeast to mammals (22). SPHK1 and SPHK2 generate S1P in cells by the transfer of a phosphate group from ATP to sphingosine. Functionally, these enzymes seemed to be interchangeable in S1P production because mice lacking either of them appear normal and breed normally whereas double knock-out mice die embryonically (23). The enzymes do have unique tissue-specific functions, however, as mice lacking SPHK1, but not mice lacking SPHK2, are more resistant to LPS-induced inflammation and are resistant to the progressive neurodegeneration seen in genetically induced Sandhoff disease (24, 25). At the amino acid level, SPHK1 and SPHK2 are ∼50% homologous. Although they both generate S1P from the same substrates, ATP and sphingosine, they exhibit distinct functional differences (26). For example, SPHK1 is more selective in its substrate, and SPHK2 phosphorylates a broader spectrum of sphingoid-like compounds (27). Our studies demonstrate that osteoclast precursors express higher levels of SPHK1 as they mature, supporting a possible role for SPHK1 in osteoclast-mediated coupling (3).

The SPHKs are G protein-coupled receptors that activate Rho family GTPases, but reports have also documented that they also signaling through other pathways such as the JAK/STAT and PI3K pathways (28, 29). In the studies reported here, mesenchymal cell pathways activated by S1P include RhoA GTPase, FAK/PI3K/AKT, and JAK/STAT. RhoA mediates migration responses downstream of G proteins in many cell types in response to multiple stimuli (30–32). In lymphocytes, S1PR1 activation of both Rac1 and Cdc42 is required for S1P migratory stimulation (33). We examined S1P action of three members of the Rho GTPase family in mesenchymal cells. We found that RhoA but not Rac1 or Cdc42 was rapidly activated by S1P treatment. Given the well documented roles of Rho family members in migration, it was unexpected that RhoA does not mediate S1P migration stimulation. Instead, we uncovered distinct roles for S1PR1 and S1PR2 in activating JAK/STAT3 and FAK/PI3K/AKT pathways, respectively. Constitutively active AKT was able to overcome pharmacological inhibition of FAK and PI3K, but not inhibition of JAK or STAT3. These data confirm that these pathways are activated in parallel, and each of these pathways contributes to S1P activation of mesenchymal cell migration (Fig. 8). Kinase-driven signaling regulates migration of many types of cells. The PI3K/AKT pathway controls migration of both normal cells and tumor cells, and deregulation of this pathway has been implicated in driving tumor cell progression (34–38). Components of JAK/STAT signaling including JAK1 and STAT3 are also well documented to modulate cell migration (39–42).

Our data that mesenchymal cells express both S1PR1 and S1PR2 mirror the finding that osteoclast precursors express both receptors (9). Both S1P receptors are essential for the recruitment of osteoclast precursors from circulation, but they function in diametrically opposed ways (9). A subset of circulating osteoclast precursors express high levels of S1PR2. Binding of S1P to S1PR2 causes a chemorepellant response to S1P, unlike the chemoattactive response in mesenchymal cells reported here. Because blood has much higher levels of S1P than the bone marrow, high S1PR2 expression in circulating osteoclast precursors causes the cells to leave the circulation and enter the bone marrow environment. The interactions between S1P and S1PR2 lead to reduced S1PR2 expression, decreasing the chemorepulsive response. Once S1P binds to S1PR2, S1PR1 expression increases, stimulating a chemoattactive response to S1P, leading to osteoclast precursor migration to the bone surface. The chemorepulsion functions of osteoclast precursor S1P/SIPR2 responses is supported by the correlation of high circulating S1P levels with low bone density recently revealed (43). Our studies document that S1PR1 and S1PR2 are also essential for the S1P chemoattractive response in mesenchymal cells. Thus, the chemoattractive influences of local S1P in the bone marrow microenvironment would promote migration of both osteoclast and osteoblast precursors. This may contribute to the increases in both osteoclasts and osteoblasts during high bone turnover.

The mesenchymal cells used in this study are resident bone marrow stromal cells (15). These cells exhibit extensive random movement, or chemokinesis, as documented by the presence of cells that pass through the membrane in the absence of any stimulus. Directed movement stimulation in vivo by S1P would bring the cells close to higher concentrations of differentiation factors such as BMPs and Wnts that we have shown to be secreted by osteoclasts and TGF-β released from the bone matrix by bone resorption (3, 44). Thus, the in vivo significance of stimulating movement toward bone resorbing osteoclasts is likely to be significant. It could be interesting to determine whether S1P influence on circulating osteoprogenitors differs. It seems unlikely because our data support that both S1PR1 and S1PR2 are involved in pro-migration responses, unlike osteoclast precursors. S1P stimulates RANKL production by osteoblasts; thus, there is also an indirect influence of S1P to promote osteoclast differentiation, suggesting a positive feedback in which osteoclast S1P production may enhance osteoclast differentiation and survival as well as osteoblast precursor recruitment (25).

Our studies support a role for S1P in recruiting osteoblast precursors and Sato et al. (45) recently documented that S1P enhances osteoblast differentiation responses to BMP2, indicating that S1P promotes anabolic responses through multiple mechanisms. Evaluations of clinical samples, in vitro studies, and in vivo animal models have supported the hypothesis that targeting SPHK/S1P may be beneficial therapeutically. Blocking S1P production is the focus of intense interest due to the link between SPHK1 and cancer, fibrosis, rheumatoid arthritis, and inflammation development (46–58). In contrast to potential benefits of targeting S1P, the work reported here reveals a positive S1P influence in stimulating mesenchymal cell recruitment, which would enhance bone formation. Evidence supports several additional positive S1P roles. SPHK1^−/− mice have defects in endothelial barrier functions and are more sensitive to heart injury (59, 60). Furthermore, the SIPR agonist FTY720 blocks lymphocyte trafficking, prevents allograft rejection in renal transplants, and reduces multiple sclerosis burden in patients (46). S1P also regulates endothelial cell functions, induces angiogenesis, and regulates lymphocyte trafficking (61–66). Moreover, S1P is required for full mast cell activation, cytokine and PGE₂ production by epithelial and endothelial cells, and promotes immune cell survival (63–66). Because of these beneficial influences, it is imperative to understand the mechanisms by which S1P exerts these positive influences to preserve these aspects of its functions in the development of any therapies targeting S1P. Our studies of coordinate activation of JAK/STAT and FAK/PI3K/AKT signaling to stimulate migration of mesenchymal cells contributes to this needed insight.