CXCL12-induced chemotaxis and adhesion to VCAM-1 decrease as B cells differentiate in the bone marrow. However, the mechanisms that regulate CXCL12/CXCR4-mediated signaling are poorly understood. We report that after CXCL12 stimulation of progenitor B cells, focal adhesion kinase (FAK) and PI3K are inducibly recruited to raft-associated membrane domains. After CXCL12 stimulation, phosphorylated FAK is also localized in membrane domains. The CXCL12/CXCR4-FAK pathway is membrane cholesterol dependent and impaired by metabolic inhibitors of Gi, Src family, and the GTPase-activating protein, regulator of G protein signaling 1 (RGS1). In the bone marrow, RGS1 mRNA expression is low in progenitor B cells and high in mature B cells, implying developmental regulation of CXCL12/CXCR4 signaling by RGS1. CXCL12-induced chemotaxis and adhesion are impaired when FAK recruitment and phosphorylation are inhibited by either membrane cholesterol depletion or overexpression of RGS1 in progenitor B cells. We conclude that the recruitment of signaling molecules to specific membrane domains plays an important role in CXCL12/CXCR4-induced cellular responses.
Bcell development is dependent on the nonlymphoid stromal cells found in the bone marrow microenvironment. In addition to making specific adhesion contacts with developing B-lineage cells, bone marrow stromal cells also secrete a number of factors that are indispensable in B cell development (1). One of these molecules is the chemokine CXCL12, also known as stromal cell-derived factor-1. Targeted disruption of the CXCL12 gene is lethal in mice and is accompanied by defects in B cell lymphopoiesis (2). Our recent studies have shown that although developing bone marrow B cells maintain high levels of CXCR4 surface expression, their responsiveness to CXCL12 diminishes with B cell maturation (3, 4). Notably, CXCL12 triggers increased chemotactic and adhesive responses in progenitor B cells compared with mature B cells in bone marrow. Moreover, the increased responsiveness correlates with prolonged activation of focal adhesion kinase (FAK),3 observed only in progenitor B cells. However, the mechanisms that regulate CXCL12-induced cellular responses and the CXCL12/CXCR4-FAK pathway during B cell development are not clear. We hypothesized that these CXCL12/CXCR4-mediated responses might be dependent on raft-associated membrane domains, because these membrane regions play a central role in regulating B cell Ag, e.g. immune receptor, responses (5). Because raft domains are not readily visible by light microscopy and are heterogeneous with respect to cholesterol and glycosphingolipid content, a combination of approaches is used to localize receptors, including CXCR4, and signaling proteins in or near raft-associated membrane domains, e.g. lipid rafts (6, 7, 8, 9). Besides lipid rafts, we hypothesized that intracellular proteins known as regulators of G protein-signaling (RGS) also might play a role in the modulation of CXCL12/CXCR4-induced responses in developing B cells. RGS has been demonstrated to modulate G protein responses by accelerating the GTPase activity of Giα proteins (10). Overexpression of RGS1, RGS3, and RGS13 in transfected lymphoid cells has an inhibitory effect on CXCL12-mediated chemotaxis (10, 11, 12, 13). Moreover, germinal center B lymphocytes distinctly express high levels of RGS1 and RGS13, correlating with their refractoriness to CXCL12-induced chemotaxis (12, 13). The latter suggests that RGS proteins might developmentally modulate Gi signaling responses. In the present study we demonstrate that lipid rafts and RGS1 play key roles in CXCL12/CXCR4-FAK signaling and in CXCL12-induced cellular responses of progenitor B cells. Moreover, we find that these roles are interrelated, suggesting that CXCL12-induced FAK activation and its recruitment to lipid rafts are critical parameters of CXCL12-induced chemotactic and adhesive responses.
Materials and Methods
mAbs against phosphotyrosine (4G10) were provided by Dr. T. Roberts (Dana-Farber Cancer Institute, Boston, MA). Antisera against FAK, p130Cas, and Lyn were purchased from Santa Cruz Biotechnology. Antisera against PI3K(p85) were obtained from Upstate Biotechnology. Abs against MAPK and phospho-MAPK were obtained from New England Biolabs. HRP-conjugated goat anti-mouse and goat anti-rabbit secondary Abs were purchased from Caltag Laboratories and Bio-Rad, respectively. Cholera toxin conjugated to HRP (CTX-HRP) was purchased from Sigma-Aldrich. Immunofluorescence experiments were performed using rabbit anti-FAK (PY397) Abs (BioSource International), Alexa 568-conjugated goat anti-rabbit IgG, Alexa 488-conjugated goat anti-mouse IgM, and Alexa 488- or Alexa 555-conjugated cholera toxin subunit B (Molecular Probes).
Cell culture and isolation
Human REH pro-B cells (American Type Culture Collection) were maintained in RPMI 1640 supplemented with 10% FBS, 1% penicillin and streptomycin, and 2 mM glutamine (all from Invitrogen Life Technologies). Heparinized bone marrow was obtained by iliac crest aspiration from healthy adult volunteers in accordance with guidelines approved by the institutional review committees of the Dana-Farber Cancer Institute. Peripheral blood cells were isolated from the buffy coats of donated whole blood. Mononuclear cells were isolated by Ficoll-Hypaque (Amersham Biosciences) gradient centrifugation (density, 1.077 g/ml). Pelleted cells were collected and washed three times in PBS. Bone marrow cells were stained and sorted into two populations, designated early lineage and late lineage B cells, at >98.5% purity. The early lineage B cell population included both pro-B and pre-B cell subsets (CD19+, κ−/λ−), whereas the late lineage B cell population included immature and mature B cells (CD19+, κ+/λ+) (3, 14). Peripheral B cells were stained and sorted with allophycocyanin-labeled anti-CD19 (Caltag Laboratories) and were regarded as mature B cells. Isolated B cells were stored at 37°C in StemSpan H2000 serum-free medium (StemCell Technologies) for 16 h before experimentation.
Gene mutation and transfection
The human RGS1 cDNA, a gift from Dr. J. Kehrl (National Institutes of Health, Bethesda, MD), was cloned into the GFP-C1 vector (Clontech Laboratories). An alanine substitution at Cys105 of RGS1 was generated by QuikChange site-directed mutagenesis following the manufacturer’s protocols (BD Clontech). Vectors were amplified in Escherichia coli strain JM109 and then sequenced for verification. The GFP-RGS1, GFP-RGS1(C105A), and GFP vectors were linearized by digestion with SalI (New England Biolabs) and transfected into REH cells by electroporation. Neomycin-resistant clones were analyzed for GFP expression by FACS. For analyses, a minimum of 10 positive clones from each stably transfected cell line were pooled and sorted by GFP expression on a MoFlo cytometer (DakoCytomation) to establish comparable expression levels (15).
Cholesterol extraction and use of metabolic inhibitors
Cultured cells were resuspended in RPMI 1640 medium lacking FBS, then stimulated with 100 nM recombinant human CXCL12 (R&D Systems) at 37°C for the indicated times before being stopped by addition of ice-cold PBS. Pretreatment of cells with inhibitors was performed by incubating cells in RPMI 1640 supplemented with 100 ng/ml pertussis toxin (PTX; Invitrogen Life Technologies), 10 μM 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo(3,4-d)pyrimidine (PP2), or 100 nM wortmannin (all from Calbiochem) for 30 min at 37°C. To deplete cholesterol from the lipid rafts, cells were incubated with 10 mM methyl-β-cyclodextrin (MCD; Sigma-Aldrich) for 30 min at 37°C. After washing with RPMI 1640 medium to remove pretreatment chemicals, the cells were stimulated with CXCL12 or 10 ng/ml PMA (Sigma-Aldrich). To replenish cholesterol, MCD-treated cells were incubated in RPMI 1640 medium containing 2 μg/ml free cholesterol (Sigma-Aldrich) for 30 min at 37°C. Cell viability was not affected by any of these treatments, as determined by trypan blue exclusion.
After CXCL12 stimulation, cells were washed with ice-cold PBS and lysed for 30 min at 4°C in buffer A (150 mM NaCl, 50 mM Tris-HCl (pH 7.6), 1 mM PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 mM sodium orthovanadate, and 10 mM sodium fluoride) containing 1% Triton X-100. Cell lysates were clarified by centrifugation at 13,000 × g for 10 min at 4°C, incubated with 2 μg of anti-FAK or anti-p130Cas Abs for 1 h, then incubated with 20 μl of protein A-Sepharose 4B beads (Amersham Biosciences) for 2 h at 4°C. Immunoprecipitates were washed three times with wash buffer (150 mM NaCl, 50 mM Tris-HCl (pH 7.6), and 0.1% Triton X-100) to remove nonspecifically bound proteins. Bound proteins were eluted by boiling in SDS-PAGE loading buffer, then analyzed by Western blotting. The relative intensities of bands detected by ECL Western Blotting Detection Reagents (Amersham Biosciences) were quantified by densitometry using ImageQuant software (Molecular Dynamics).
Lipid raft isolation and immunoblotting
Lipid rafts were isolated by fractionation on sucrose gradients as previously described (16). Briefly, REH cells were stimulated by treatment with 100 mM CXCL12 at 37°C for 10 min and stopped by addition of ice-cold PBS. Cells were lysed in 1 ml of buffer A containing 0.2% Triton X-100. After 20-min incubation at 4°C, the cell lysates were homogenized with 10 strokes in a Dounce homogenizer (Wheaton) and adjusted to 40% sucrose by addition of an equal volume of ice-cold 80% (w/v) sucrose prepared in buffer A. The mixture was transferred to an SW55 ultracentrifuge tube (Beckman Coulter) and overlayered successively with 2 ml of 30% and 1 ml of 5% (w/v) sucrose, also prepared in buffer A. After centrifugation at 39,000 rpm in a Sorvall AH-650 rotor at 4°C for 16 h, 0.4-ml fractions were collected from the top of the sucrose gradient designated from 1 (top) to 12 (bottom). Equal aliquots of each fraction were loaded on SDS-PAGE and run at constant voltage. In some experiments, fractions 2–4 were pooled to concentrate the detergent-resistant membrane (DRM) fractions, and fractions 10–12 were pooled to represent soluble fractions. Proteins were transferred to nitrocellulose membranes (Bio-Rad) for immunoblotting, and bound HRP-conjugated secondary Abs were detected by ECL. For GM1 detection, aliquots of each fraction were spotted onto nitrocellulose membranes and blotted with CTX-HRP.
For surface staining of GM1, stimulated or unstimulated cells were washed with PBS containing 2% BSA, labeled with Alexa-conjugated CTX (10 μg/ml) for 30 min at 4°C, washed with three times PBS, and fixed for 30 min in 1% paraformaldehyde in PBS at 4°C. For intracellular staining of phospho-FAK(PY397), paraformaldehyde-fixed cells were permeabilized with 0.1% Triton X-100 in PBS for 1 min at 4°C, washed with PBS containing 2% BSA, and incubated with rabbit anti-FAK(PY397) Ab (1/100 dilution) or rabbit IgG isotype at 4°C for 30 min. Cells were subsequently washed and incubated with Alexa-conjugated secondary Ab (1/500 dilution) at 4°C for 1 h. Cells were washed and mounted onto glass slides, and data were collected from a Zeiss LSM510 confocal laser scanning microscope and analyzed by LSM510 software.
Chemotaxis and cell adhesion assays
Chemotaxis assays were performed using previously described methods (3). Migratory cells were collected and counted by timed acquisition (60 s/sample) on a FACSCalibur flow cytometer (BD Biosciences). Cell adhesion assays were performed as previously described (4). The number of bound cells was determined using a CyQuant Cell Proliferation Assay kit (Molecular Probes). Briefly, cells were stained with CyQuant DNA dye, and the percentage of bound cells was calculated using a microtiter plate fluorometer (Dynex Technologies). In some cases 100 ng/ml PMA was added to the cells instead of CXCL12 before the 30-min incubation in VCAM-coated wells. For Mn2+-induced adhesion, cells were washed with 2 mM EDTA in a modified adhesion medium (HBSS containing HEPES, but lacking Ca2+ and Mg2+). Cells were subsequently incubated with 2 mM MnCl2 and CaCl2 in the modified adhesion medium.
Total RNA was purified using TRIzol reagent following the manufacturer’s protocol (Invitrogen Life Technologies). Purified RNA from early lineage, late bone marrow, and peripheral B cells was diluted to 1 μg/μl and treated with RNase-free DNase I (Invitrogen Life Technologies). RT-PCR was performed on 1 ng of DNase I-treated RNA using an iCycler real-time detection system (Bio-Rad) and the SYBR Green RT-PCR kit (Qiagen). The following primer pairs (Invitrogen Life Technologies) were used to amplify the RGS1, RGS3, and GAPDH cDNAs, respectively: 5′-AGAAGGAATGTGCCAGTATG-3′ and 5′-TCTGCGCCTGGATAACTTTCA-3′, 5′-GTGAGGAGAATCTGGAGTT-3′ and 5′-CCATCTTGGACTGTGACTT-3′, and 5′-CAGAAGACTGTGGATGG-3′ and 5′-GCTTCACCACCTTCTTG-3′. The real-time thermocycler conditions used to amplify these cDNAs included 30-min incubation at 50°C, followed by 15 min at 95°C and 50 cycles of 95°C for 15 s, 50°C for 30 s, and 72°C for 30 s. Each sample was assayed in triplicate, and the number of copies of the gene of interest in each sample was extrapolated from a corresponding standard curve and normalized by the amount of GAPDH cDNA amplified from the same RNA sample. Primer specificity was confirmed by electrophoretic analysis of the RT-PCR products and by use of template- and reverse-transcriptase-free controls.
CXCL12 stimulation recruits FAK and PI3K to lipid raft membrane domains
One of the principal properties of lipid rafts is their ability to include or exclude certain signaling molecules. Because FAK and PI3K are thought to play important roles in chemotaxis and adhesion responses, we investigated their localization with respect to raft-associated membrane domains in B cells after CXCL12 stimulation. The Lyn kinase served as a raft marker and was predominantly localized in DRM sucrose gradient fractions 2, 3, and 4 (Fig. 1,A, bottom panels) (17). FAK was not localized in the DRM fractions in unstimulated cells, but was recruited to rafts after stimulation with CXCL12 (Fig. 1,A, top panels). Because PI3K signaling is associated with CXCL12-induced tyrosine FAK phosphorylation (18), we reprobed the same membrane with anti-PI3K(p85) Ab and found that substantial amounts of PI3K(p85) were translocated to raft fractions in CXCL12-stimulated cells (Fig. 1 A, center panels). Thus, both FAK and PI3K are recruited into DRM domains after CXCL12 stimulation.
We also examined whether lipid rafts are important components of the FAK and PI3K signal transduction pathways by using a raft-disrupting agent, MCD, which disrupts rafts by cholesterol extraction from the plasma membrane. FAK, PI3K(p85), and Lyn were no longer present in lipid raft fractions after CXCL12 stimulation in MCD-treated REH cells (Fig. 1,B, left panels). However, replenishment of MCD-treated cells with cholesterol allowed partial reassociation of FAK, PI3K(p85), and Lyn with lipid rafts after CXCL12 stimulation (Fig. 1 B, right panels). These results suggested that FAK and PI3K(p85) are recruited to cholesterol-enriched membrane domains. We were concerned that MCD treatment could induce conformational changes in CXCR4. To this end, we stained untreated and 10 mM MCD-treated REH cells with four different commercially available monoclonal anti-CXCR4 Abs (R&D Systems) and found no difference in immunofluorescence staining. Moreover, binding of biotinylated CXCL12 to untreated vs MCD-treated cells was unchanged. These data (not shown) thus argue that MCD treatment of REH pro-B cells did not induce conformational changes in CXCR4.
Recruitment of FAK into lipid rafts is dependent on the Gi protein and Src family signaling pathways
To further characterize the CXCR4-FAK pathway, we used various kinase inhibitors to determine whether any of these might affect recruitment of FAK into lipid rafts. The inhibitors PTX, PP2, and wortmannin block the Gi protein, Src family, and PI3K signal transduction pathways, respectively. After CXCL12 stimulation, FAK was detected in the pooled lipid raft fractions (Fig. 1,C, lane 2). In contrast, no FAK localized to the lipid rafts in the absence of CXCL12 stimulation (Fig. 1,C, lane 1). Inhibition of Gi protein signaling by PTX significantly reduced CXCL12-induced recruitment of FAK into lipid rafts (Fig. 1 C, lane 3), indicating that association of FAK with lipid rafts is dependent on Gi protein signaling.
Pretreatment of cells with PP2, a strong inhibitor of Src family proteins, also abolished CXCL12-induced recruitment of FAK to lipid rafts (Fig. 1,C, lane 4). In contrast, pretreatment of cells with wortmannin, an inhibitor of PI3K, did not significantly alter FAK recruitment (Fig. 1,C, lane 2 vs lane 5). Interestingly, only wortmannin, not PTX or PP2, blocked the CXCL12-induced recruitment of PI3K(p85) into lipid rafts (Fig. 1,C, middle panel), suggesting that recruitment of PI3K(p85) is only dependent on its kinase activation, not on the Gi protein and Src family pathways. Lyn, whose recruitment is not affected by these inhibitors, served as a loading control (Fig. 1 C, lower panel).
CXCL12 stimulation induces tyrosine phosphorylation of FAK, which colocalizes to raft-associated membrane domains
The results presented in Fig. 1 demonstrate that CXCL12 stimulation induces the translocation of FAK into the lipid rafts in a signaling-dependent manner. We next investigated whether lipid rafts play a role in FAK phosphorylation. To this end, REH pro-B cells were treated with or without MCD, then lysed and immunoprecipitated with anti-FAK Abs, followed by Western blot with an anti-phosphotyrosine Ab 4G10 to examine total FAK phosphorylation. CXCL12 stimulation induced strong tyrosine phosphorylation of FAK (5.9-fold increase) in REH cells (Fig. 2,A, lane 2). However, tyrosine phosphorylation of FAK was significantly reduced in MCD-treated cells (Fig. 2,A, lane 3) and was partially restored after replenishment of MCD-treated cells with cholesterol (Fig. 2,A, lane 4). No change in CXCL12-induced FAK phosphorylation was observed in REH cells that were treated only with cholesterol (Fig. 2,A, lane 5). We also found that PMA induced tyrosine phosphorylation of FAK in REH cells (Fig. 2 A, lane 7). FAK phosphorylation was abolished in MCD-treated cells (lane 8) and was restored after replenishment of MCD-treated cells with cholesterol (lane 9). The blots were subsequently reprobed with anti-FAK Abs to demonstrate that equivalent amounts of FAK were loaded onto each lane.
We subsequently used sucrose gradient centrifugation and confocal microscopy to examine the cellular localization of phosphorylated FAK by using an Ab specific for the FAK autophosphorylation site on tyrosine 397 (PY397). In Fig. 2,B, we show that, similar to total FAK, phosphorylated FAK cosegregates with Lyn and GM1 to the DRM fractions. For confocal microscopic analysis (Fig. 2,C), REH cells were stained for phosphorylated FAK and for the lipid raft component GM1, which, like other raft markers, may not always be associated with cholesterol; they may be resistant to MCD treatment (8). As reported for other lymphoblastoid cells (6), we noted that GM1 membrane domains were detectable by fluorescence microscopy in unstimulated REH pro-B cells. In contrast, primary PBL required stimulation with CXCL12 to allow GM1 visualization (data not shown). In unstimulated REH cells, phospho-FAK(PY397) was not detectable (Fig. 2,C). After CXCL12 stimulation (3 min), we noted significant phospho-FAK(PY397) staining distributed primarily near the cell membrane. The yellow margin observed after merging the two images indicates that a significant portion of phospho-FAK(PY397) colocalizes with lipid raft marker GM1 after CXCL12 stimulation (Fig. 2,C). After CXCL12 stimulation for 10 min, we observed a clustering of lipid rafts and colocalization of phospho-FAK(PY397) with GM1. Thus, these fluorescence microscopy studies are in agreement with our biochemical studies illustrating cosegregation of FAK(PY397) to DRM fractions after CXCL12 stimulation (Fig. 2 B).
Mutation of Cys105 reduces the association of RGS1 with lipid rafts
The cellular localization of RGS proteins is diverse. Certain RGS proteins are tightly membrane bound and behave as hydrophobic molecules, whereas others are soluble and are found in the cytoplasm (19, 20, 21, 22). Another subset of RGS proteins displays intermediate behavior and is found in both membrane fractions and the cytoplasm (19). The mechanism by which RGS proteins associate with lipid membranes is not clear. As with other membrane-associated signaling proteins, it has been proposed that palmitoylation of cysteine residues helps anchor the RGS proteins to lipid bilayers. Moreover, palmitoylation of conserved cysteine residues in the RGS domain modulates the GTPase-activating protein (GAP) activities of RGS4, RGS10, and RGS16 (19, 23, 24). In RGS1, the conserved cysteine residue in the RGS domain is positioned at residue 105 (Cys105; Fig. 3,A). To determine whether this conserved cysteine is important for the function and/or membrane localization of RGS1, we generated a mutant form of RGS1 in which Cys105 was substituted with alanine, RGS1(C105A). We transfected GFP-RGS1 and GFP-RGS1(C105A) expression vectors as well as a GFP control into REH cells. A minimum of 10 GFP-positive clones were pooled and sorted for similar levels of GFP expression. Three representative pooled cell lines containing 95–97% GFP-positive cells are shown in Fig. 3,B. We subsequently examined the association of RGS1 and lipid rafts by sucrose gradient fractionation, followed by Western blot analysis with anti-GFP Abs. The majority of GFP-RGS1 fusion protein was detectable in the DRM fractions, whereas there was very little detected in the soluble fractions. In contrast, the mutant GFP-RGS1(C105A) protein exhibited significantly reduced association with lipid rafts, and relatively more mutant protein was present in the soluble fractions (Fig. 3,C). In Fig. 3,D, we compared the relative amounts of GFP-RGS1 vs GFP-RGS1(C105A) in the pooled DRM and pooled soluble fractions. In this experiment the pooled DRM fractions of each of the transfectants were run on one gel, and the same was done for the pooled soluble fractions. In the DRM fractions, with and without CXCL12 stimulation, the relative amount of GFP-RGS1(C105A) was one-third that of GFP-RGS1. By contrast, the pooled soluble fractions showed almost twice as much GFP-RGS1(C105A) compared with GFP-RGS1. Moreover, CXCL12 stimulation did not alter the distribution of GFP-RGS1 or the GFP-RGS1(C105A) mutant protein in the sucrose gradient fractions (Fig. 3 D).
The cellular localization of GFP-RGS1 and GFP-RGS1(C105) was similar when investigated by confocal microscopy (Fig. 3,E). GM1 staining is indicated in red, whereas GFP-RGS1 and GFP-RGS1(C105) staining are indicated in green in Fig. 3. The yellow margin observed after merging the images indicates that the majority of GFP-RGS1 colocalizes with GM1. In contrast, mutant GFP-RGS1(C105A) was localized to a large extent in the cytoplasm (Fig. 3,E, middle panel). As expected, the GFP control did not colocalize with GM1 (Fig. 3 E, bottom panel). Taken together, these findings from biochemical and fluorescence microscopy studies suggested that Cys105 is important in mediating the association of RGS1 with lipid rafts.
Overexpression of RGS1 inhibits CXCL12-induced recruitment of FAK and p130Cas into lipid rafts
We next determined whether RGS1 could influence the recruitment of FAK to lipid rafts after CXCL12 stimulation. After CXCL12 stimulation, the recruitment of FAK and its downstream signaling molecule p130Cas to lipid rafts was impaired in GFP-RGS1-transfected cells compared with cells transfected with the GFP control (Fig. 4,A, top panel). In contrast, recruitment of FAK and p130Cas to raft domains was retained in cells expressing GFP-RGS1(C105A) (Fig. 4 A). Taken together, these findings suggested that the ability of RGS1 to inhibit the recruitment of FAK and P130Cas was linked to the residence of RGS1 in or near raft-associated membrane domains. However, we cannot exclude the possibility that mutation of RGS1 Cys105 to alanine could influence GAP activity independently of RGS1 cellular location.
Interestingly, the recruitment of PI3K was not affected by the expression of RGS1 (Fig. 4,A, third panel). The membrane was reprobed with anti-Lyn Abs to control for DRM loading (Fig. 4,A, fourth panel). Moreover, levels of FAK, p130Cas, and PI3K were unchanged in the soluble fractions of CXCL12-stimulated cell lines transfected with GFP, GFP-RGS1, or GFP-RGS1(C105A) (Fig. 4,A, right panels). Densitometric histograms of relative band intensities from Western blot experiments of DRM fractions are illustrated in Fig. 4 B.
RGS1 down-regulates CXCL12-induced tyrosine phosphorylation of FAK and p130Cas
As previously reported for primary pro-B cells (4), CXCL12 quickly induced sustained tyrosine phosphorylation of both FAK and p130Cas in GFP-transfected REH pro-B cells (Fig. 4, C and D), lasting up to 30 min. Overexpressed GFP-RGS1 markedly impaired the initiation and duration of FAK phosphorylation, whereas GFP-RGS1(C105A) did not have a striking effect on FAK phosphorylation. Interestingly, GFP-RGS1 similarly affected the initiation and duration of p130CAS phosphorylation, whereas GFP-RGS1(C105) had minimal or no effect. Thus, the effects of RGS1 on FAK and p130Cas phosphorylation are also correlated with the localization of RGS1 in or near raft membrane domains.
CXCL12-induced MAPK activation is dependent on lipid rafts and is modulated by RGS1
The MAPK pathway is activated by CXCL12 and plays an important role in cell growth and survival (25). In Fig. 5, A and B, we found that MCD-treated cells significantly reduced CXCL12-induced ERK and p38 phosphorylation. We also investigated the kinetics of ERK and p38 activation in GFP-, GFP-RGS1-, and GFP-RGS1(C105A)-transfected REH cells. In GFP-RGS1-transfected REH cells, CXCL12 stimulation resulted in a significant reduction in the level of activation of both ERK and p38 compared with that in GFP-transfected cells (Fig. 5, C and D). The duration of MAPK activation in GFP-RGS1-transfected cells was also shorter than that observed in GFP-transfected cells. In contrast, overexpression of GFP-RGS1(C105A) restored the activation of both ERK and p38 and extended the duration of MAPK activation compared with cells expressing GFP-RGS1.
Increased expression of RGS1 during B cell development
RGS proteins deactivate G protein signaling pathways by accelerating the GTPase activity of the G protein α subunit (26). Because CXCL12-induced responses are down-regulated during B cell maturation, we examined the expression of RGS1 during B cell development. We isolated CD34+CD19+ early (pro- and pre-) and late (immature and mature) B cells from bone marrow. We also isolated CD19+ B cells from peripheral blood. Because insufficient primary cells could be isolated from bone marrow to detect RGS protein expression by immunoblot analysis, we performed quantitative RT-PCR. RGS1 expression, which was normalized to the housekeeping gene GAPDH, was low in early lineage B cells, and was dramatically increased in late B cells (Fig. 6,A). Normalized RGS1 was most highly expressed in peripheral B cells (∼60- to 70-fold more than in early bone marrow B cells; Fig. 6,B), indicating that increased RGS1 levels correlate with decreased CXCL12-mediated adhesion to VCAM-1 of mature B cells. In contrast, there was no difference in the expression of GAPDH-normalized RGS3 in any of the three populations of B cells we tested. For comparison, we performed quantitative RT-PCR on the transfected REH pro-B cells. Both GFP-RGS1 and GFP-RGS1 (C105A) expressed comparable levels of RGS1 as peripheral blood B cells (Fig. 6, A and C).
CXCL12-induced chemotaxis and adhesion to VCAM-1 are impaired by membrane cholesterol depletion and overexpression of RGS1
MCD-treated cells had impaired B cell chemotaxis and adhesion to VCAM-1 after CXCL12 stimulation (Fig. 7, A and B), and replenishing MCD-treated cells with cholesterol restored chemotaxis and cell adhesion to VCAM-1 in response to CXCL12. Cells treated only with cholesterol exhibited no difference in adhesion to VCAM-1 compared with untreated cells (data not shown). In the presence of EDTA or EGTA, the divalent cation Mn2+ can activate cell adhesion by binding to integrin ectodomains, e.g., through outside-in signaling (27). To confirm that MCD itself does not affect the binding capacity of integrins, we investigated the effect of Mn2+ on cell adhesion to VCAM-1. There was no statistical difference in Mn2+-induced adhesion between MCD-treated and untreated cells, indicating that MCD does not alter the binding capacity of integrins (Fig. 7 C). Only background levels of cell adhesion were observed after treatment of cells with medium.
CXCL12-mediated chemotaxis and adhesion were also impaired by overexpression of GFP-RGS1 (Fig. 7, D and E). Furthermore, this reduction was abrogated in GFP-RGS1(C105A)-transfected cells, which showed equivalent CXCL12-mediated chemotaxis and adhesion as GFP-transfected cells. Transfection alone did not alter the adhesion of REH cells, because stimulation of cells with PMA (which triggers adhesion through G protein-coupled receptor (GPCR)-independent mechanism) yielded equivalent levels of cell adhesion in all transfected cell lines (Fig. 7 F). Interestingly, CXCL12-induced adhesion was less sensitive to RGS1 action than chemotaxis. A similar observation was previously noted in experiments using a murine cell line stimulated with the synthetic peptide FMLP (28). The importance of RGS proteins in the desensitization of CXCR4 signaling in rgs1−/− mice was discussed in a recent report (29).
In the present study we have explored the importance of lipid raft membrane domains and RGS proteins in the CXCR4-FAK signaling pathway. We present the novel finding that FAK is recruited to lipid rafts upon CXCL12 stimulation in REH pro-B cells (Fig. 1,A). Additionally, disruption of lipid rafts by MCD inhibits both FAK recruitment and FAK activation in response to CXCL12 stimulation (Figs. 1,B and 2,A). We also found that phosphorylated FAK colocalizes with lipid rafts after CXCL12 stimulation (Figs. 1,B and 2 C). These observations suggest a significant role for lipid rafts as a platform for signaling molecules to facilitate CXCL12/CXCR4 signal transduction.
In general, there are two mechanisms by which signaling molecules can associate with lipid rafts (30). Molecules such as Lyn, Lck, Fyn, LAT, and the CXCR4-associated signaling units, the G proteins, may constitutively associate with membrane rafts (31). Other molecules, such as Zap-70, PKC, PLCγ, or Vav, inducibly associate with lipid rafts in response to receptor cross-linking or other stimuli. Our results show that FAK inducibly associates with lipid rafts upon CXCL12 stimulation and that this translocation is dependent on Gi and Src family protein-mediated signaling (Fig. 1). The precise mechanism by which FAK translocation occurs is not clear. Based on our findings, however, we postulate the following sequence of signaling events. As previously suggested by other groups, CXCR4-mediated signaling occurs in or near lipid rafts (6, 32, 33). The initial signaling event involves the association of the CXCR4 receptor with Gi signaling units, which, in turn, leads to the activation of several effector pathways (18, 34). We hypothesize that the activation of Src family proteins, e.g., Lyn, which constitutively reside in lipid rafts, is a proximal event in CXCR4 signaling and occurs upstream of FAK (35). The latter is in agreement with our observation that inhibition of Src proteins by PP2 abrogates FAK recruitment to lipid rafts after CXCL12 stimulation (Fig. 1,C). The phosphorylation of FAK is sensitive to membrane cholesterol depletion (Fig. 2,A) and thus may require the close association of FAK and activated Src proteins in lipid rafts. Alternatively, however, after CXCL12 stimulation, FAK may first be phosphorylated in nonraft domains and subsequently recruited to the lipid raft signaling complex. To explore this latter possibility, we stimulated REH pro-B cells while they were in suspension for 3 min with CXCL12 or with the phorbol ester PMA, which, unlike CXCL12/CXCR4, does not require lipid rafts for the initiation of signaling. In both instances we found that FAK phosphorylation was inhibited by cholesterol depletion (Fig. 2 A), suggesting that FAK phosphorylation, induced by either CXCL12 or PMA, is dependent on intact lipid raft domains. Thus, FAK is probably first recruited to lipid raft domains and subsequently phosphorylated. It is important to note that the REH pro-B cells were stimulated in suspension for only 3 min. Under these conditions, integrin engagement is considered to be minimal and thus represent inside-out, e.g., PMA-FAK or CXCL12/CXCR4-FAK, signaling. In contrast, in adherent cells, where integrins are activated through outside-in signaling, FAK phosphorylation appears independent of cholesterol-enriched raft domains, suggesting that FAK is not recruited to lipid raft membrane domains (36).
Subsequently, we explored the role of RGS molecules in CXCL12–FAK signaling and CXCL12-induced cellular responses. Using REH pro-B cells expressing wild-type (GFP-RGS1) or mutant RGS1 (C105) proteins, our studies indicate that GFP-RGS1 protein constitutively associates with raft membrane domains, suggesting that RGS1 is not translocated to raft membrane domains after ligand engagement of GPCRs. Furthermore, the association with lipid rafts is affected by mutation of the conserved cysteine residue at position 105 (Fig. 3, C and E). Palmitoylation of additional cysteine residues in the RGS1 protein is needed to direct RGS1 to the plasma membrane microdomains, because the alanine substitution at Cys105 did not completely abolish the association of mutant RGS1(C105A) with lipid rafts (Fig. 3,C). Moreover, RGS1 function, as measured by its ability to impair CXCL12-mediated lipid raft recruitment and activation of FAK and p130Cas, correlates with RGS1 segregation to lipid raft domains (Fig. 4). However, we cannot exclude the possibility that the C105A mutation alters RGS1 GAP activity regardless of its localization in lipid rafts. Western blot analyses conducted at various time intervals of CXCL12 stimulation suggest that RGS action inhibits the initiation and duration of FAK and p130Cas signaling, whereas this effect is attenuated for RGS1(C105A) (Fig. 4, C and D). Thus, the presence of FAK in lipid rafts correlates with prolonged activation of FAK and its downstream signaling molecule p130Cas. This finding is intriguing in view of recent data proposing that FAK signaling prevents internalization of raft membrane domains (36, 37, 38). Taken together, the current data argue that both the initiation of CXR4-mediated signaling as well as desensitization by RGS1 occur in or near lipid raft membrane domains.
Interestingly, the recruitment of PI3K(p85) to lipid rafts is not affected by RGS1 (Fig. 4,A) or PTX (Fig. 1,C), suggesting that activation of the p85/110 isoform of PI3K may be due to binding of an adapter protein to the CXCR4 receptor and is Gi-protein-independent (39). The CXCL12-induced activation of PI3K (p85) appears distinct from other major CXCR4-mediated signaling pathways, including CXCL12-mediated phosphorylation of FAK (Fig. 1) and MAPK (Fig. 5), both of which depend on proximal coupling of the CXCR4 receptor to heterotrimeric G proteins and are desensitized by RGS1 (Figs. 4 and 5). Moreover, FAK recruitment to lipid rafts is not affected by the PI3K inhibitor, wortmannin (Fig. 1 C), further supporting the idea that CXCR-FAK and CXCR4-PI3K pathways may be unlinked.
Based on current data and previous studies (4), we propose the following model to explain the roles of the CXCL12/CXCR4 axis and RGS1 in B cell development. In the bone marrow, CXCL12, which is secreted by stromal cells, binds and activates CXCR4 on progenitor B cells. Activated CXCR4 receptors form clusters in lipid raft domains (6) and subsequently trigger signaling molecules, e.g., Gi protein, Src family proteins, and FAK. Progenitor B cell surface integrins, such as VLA-4, are activated via inside-out signaling, bind to VCAM-1, and form close contacts with stromal cells. This process retains B cells in the bone marrow microenvironment (40) and promotes their growth and differentiation. During B cell differentiation, however, CXCL12-induced signaling and cellular responses are down-regulated (4), which may be attributed in part to RGS1 action. In this regard, in primary B cells from bone marrow and peripheral blood, RGS1 expression is low in progenitor (pro- and pre-) B cells and highest in mature B cells (Fig. 6). Moreover, overexpression of RGS1 in progenitor pro-B cells (which have little endogenous RGS1) impairs CXCL12-induced FAK activation, chemotaxis, and adhesion to VCAM-1 (Fig. 7). The relatively increased RGS1 expression in mature B cells correlates with decreased CXCL12-induced adhesion to VCAM-1 (4), a process that might contribute to the release of mature B cells from the bone marrow into the peripheral circulation. Besides the action of RGS proteins, cells potentially use other mechanisms to desensitize signaling via GPCRs, including G protein degradation, GPCR sequestration, and phosphatases targeting phosphorylated signaling proteins (41, 42, 43). Collectively, these findings argue that in the bone marrow, RGS1 may developmentally regulate CXCL12-mediated responses, which play a critical role in progenitor B cell positioning and maturation.
The authors have no financial conflict of interest.
We gratefully acknowledge the technical assistance of Harry Leung and Tao Lu. We thank Dr. James J. Campbell for his helpful discussions.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health Grant POI84800.
Abbreviations used in this paper: FAK, focal adhesion kinase; CTX-HRP, cholera toxin conjugated to HRP; DRM, detergent-resistant membrane; GPCR, G protein-coupled receptor; MCD, methyl-β-cyclodextrin; PP2, 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo(3,4-d)pyrimidine; PTX, pertussis toxin; RGS, regulator of G protein signaling; GAP, GTPase-activating protein.