Complement C3a promotes CXCL12-induced migration and engraftment of human and murine hemopoietic progenitor cells, suggesting a cross-influence between anaphylatoxin and chemokine axes. Here we have explored the underlying mechanism(s) of complement anaphylatoxin and chemokine cooperation. In addition to C3a, C3a-desArg and C4a but not C5a, are potent enhancers of CXCL12-induced chemotaxis of human and murine bone marrow (BM) stem/progenitor cells and B lineage cells. C3a enhancement of chemotaxis is chemokine specific because it is also observed for chemotaxis to CCL19 but not to CXCL13. The potentiating effect of C3a on CXCL12 is independent of the classical C3a receptor (C3aR). First, human BM CD34+ and B lineage cells do not express C3aR by flow cytometry. Second, the competitive C3aR inhibitor SB290157 does not affect C3a-mediated enhancement of CXCL12-induced chemotaxis. Third, enhancement of chemotaxis of hemopoietic cells is also mediated by C3a-desArg, which does not bind to C3aR. Finally, C3a enhances CXCL12-induced chemotaxis of BM cells from C3aR knockout mice similar to BM cells from wild-type mice. Subsequent studies revealed that C3a increased the binding affinity of CXCL12 to human CXCR4+/C3aR−, REH pro-B cells, which is compatible with a direct interaction between C3a and CXCL12. BM stromal cells were able to generate C3a, C3a-desArg, C4a, as well as CXCL12, suggesting that this pathway could function in vivo. Taken together, we demonstrate a C3a-CXCL12 interaction independent of the C3aR, which may provide a mechanism to modulate the function of CXCL12 in the BM microenvironment.
As part of innate immunity, the complement system has roles in host defense as well as in shaping the adaptive immune response (1). Complement anaphylatoxins, C4a, C3a, and C5a, are generated during cleavage activation of C4, C3, and C5, respectively. C3a-desArg is the carboxypeptidase cleaved form of C3a. C3a and C5a, which have specific receptors, are effective activators of granulocytes and mast cells and induce a wide range of cell responses, including chemotaxis, adhesion to endothelial cells and transendothelial migration (2, 3, 4, 5, 6). Several lines of evidence suggest that C3a and C5a also may affect lymphocytes (7, 8, 9, 10). A characteristic of complement anaphylatoxins is their rapid generation and inactivation (reviewed in Ref.11).
Chemokines and complement anaphylatoxins have many features in common. Both chemokine and C3a/C5a receptors share structural and functional motifs, belonging to the subfamily of seven transmembrane-spanning, G-protein-coupled receptors (12, 13, 14). Chemokines, similar to anaphylatoxins, also affect the migration and adhesion of cells. CXCL12 signaling through its unique receptor, CXCR4, is critical for hematopoiesis, particularly myelopoiesis and lymphopoiesis, and influences stem/progenitor cell trafficking to and from the bone marrow (BM)3 (15, 16, 17, 18, 19). Unlike many chemokines that are up-regulated by inflammatory stimuli, CXCL12 belongs to the group of constitutively expressed chemokines (20). Down-regulation of the tonic stimulation by CXCL12 through its unique receptor, CXCR4, has been correlated with the release of progenitors from the BM (reviewed in Ref.21). Conversely, it has been proposed that enhanced CXCL12 function relates to the ability of C3a to accelerate the engraftment of transplanted murine Sca-1+ stem cells (22). Evidence that C3a potentiates CXCL12 comes from the chemotactic responses of human CD34+ stem cells (22). The mechanism for this cross-talk between the chemokine and complement pathways was thought to rely on the expression of the C3a receptor (C3aR) on the responding cells (22).
We have further investigated the biology and mechanism of the C3a-CXCL12 interaction. Our data show that C3a enhances chemotaxis of human BM progenitor and B cells in response to CXCL12. Furthermore, the responding cells lack the classical C3aR (12, 23) by antigenic and functional criteria. Confirmation of the C3aR independence of this phenomenon came from the ability of C3a to enhance the CXCL12-mediated chemotaxis of C3aR knockout mouse-derived BM cells. We also find that BM stromal cells can generate complement anaphylatoxins, suggesting that they may modulate the function of constitutively expressed CXCL12 in the BM.
Materials and Methods
The REH pro-B cell line (ATCC CRL-8286), HS-Sultan Burkitt’s lymphoma cell line (ATCC CRL-1484), and 293 human embryonic kidney cell line (ATCC CRL-1573) were obtained from American Type Culture Collection (ATCC). The U937 cell line was kindly provided by Dr. N. P. Gerard (Harvard Medical School, Boston, MA). Heparinized BM was obtained by iliac crest aspiration from healthy adult volunteers after informed consent and in accordance with the guidelines approved by the Institutional Review Committees of the Dana-Farber Cancer Institute. Peripheral blood lymphocytes were isolated from human whole blood buffy coat preparations, in accordance with the guidelines approved by the Institutional Review Committee of Children’s Hospital Boston. Mononuclear cells were isolated from BM and peripheral blood by Ficoll-Hypaque (Pharmacia) gradient centrifugation (density 1.077 g/ml).
C3aR knockout mice, which were generated and bred on a BALB/c background as described (24), were provided by the laboratory of Dr. C. Gerard (Children’s Hospital, Boston, MA). All procedures performed on the animals were in accordance with the Animal Care and Use Committee of the Children’s Hospital Boston. BM was collected by flushing femurs and tibiae with cold PBS. Lymphocytes were separated through Lympholite-M (Cedarlane Laboratories) gradient centrifugation before use in the chemotaxis assays.
Abs and cell surface staining of human and murine lymphocytes
Surface staining of BM and peripheral blood human cells was performed with the following mAbs (all from BD Pharmingen, unless otherwise noted): FITC-anti-κ L chain, FITC-anti-λ L chain, PE-anti-CD19, Cy-Chrome-anti-CD34, PE-anti-IgD, ECD-anti-CD10 (Immunotech), allophycocyanin-anti-CD19 (Caltag Laboratories), allophycocyanin-Cy7-anti-CD19, anti-C3aR (clone 8H1), anti-C3aR (clone 218) (22), mouse IgG1 isotype control. Goat anti-mouse IgG-Cy5 (Jackson ImmunoResearch Laboratories) was used as the secondary Ab. Murine BM cells were stained with the following Abs (all from BD Pharmingen, unless otherwise stated): FITC-anti-BP-1, PE-anti-CD43, PE-Cy-5-anti-CD24 (eBioscience), allophycocyanin-anti-IgM, allophycocyanin-Cy7-anti-B220 (Cedarlane Laboratories), PE-anti-Sca-1, PE-Cy5-anti-CD3. Stages of B cell differentiation in human and murine BM were defined as described (25, 26). Six-color immunofluorescence analysis was used to examine the expression of C3aR on the surface of progenitor cells and different B cell populations. At least 7,500 events for CD34+ cells and early stages of B cell development and 10,000 events for later stages of B cell development were acquired using a MoFlo flow cytometer (DakoCytomation) and analyzed with Summit software (DakoCytomation).
Detection of complement anaphylatoxins and CXCL12 in cell supernatants
Human BM stromal cells were isolated as described before (27). Production of C3a, C4a, and C5a was detected by BD cytometric bead array human anaphylatoxin kit (BD Pharmingen), CXCL12 was detected by Quantikine human immunoassay (R&D Systems), according to the manufacturers’ protocol. Stromal cells, REH, HS Sultan, or 293 cell lines were washed three times with PBS and then cultured in serum-free medium (StemSpan H2000; StemCell Technologies) at 106 cells/ml medium. After 24 h, culture supernatants were harvested and analyzed. Detection threshold of the assays was >4 pg/ml for C3a, >10.5 pg/ml for C4a, >1.2 pg/ml for C5a, and >18 pg/ml for CXCL12.
Ficoll or Lympholite-M purified human or murine lymphocytes (106 cells in 0.5% BSA, RPMI 1640) were placed into the upper well of Transwell inserts (6.5 mm diameter, 5 μm filter pore size) (Costar), either in medium alone or with 1000 ng/ml C3a, C3a-desArg, C4a, or C5a (all from Calbiochem). The lower well contained either 600 μl of 10 ng/ml CXCL12 (human or murine, accordingly), 100 ng/ml CCL19 (PeproTech) or 2000 ng/ml CXCL13 (R&D Systems) or 600 μl of chemokines together with 1000 ng/ml C3a, C3a-desArg, C4a, or C5a in 1% BSA, RPMI 1640. Cells were allowed to migrate for 2 h at 37°C. Cells that passed through the membrane to the lower well were collected and stained with Abs to CD34 Ag or B cell determinants, as described above. Cells were counted by timed acquisition (120 s each sample, at 2 psi sample pressure differential) on a MoFlo flow cytometer (DakoCytomation).
To assess whether C3a enhancement of chemotaxis to CXCL12 was dependent on the C3aR interaction, lymphocytes were first preincubated for 30 min with 10 μM or 100 μM of the C3aR antagonist SB290157 (Calbiochem) (28). Next, cells were placed into the upper well (106 cells/well) of Transwell inserts along with 1000 ng/ml C3a mixed with C3aR antagonist (10 or 100 μM SB290157). The lower well contained 600 μl of 10 ng/ml CXCL12 along with 1000 ng/ml C3a mixed with SB290157 (10 or 100 μM) in 1% BSA, RPMI 1640.
To determine the influence of C3a on CXCL12-mediated migration of REH cells, cells were put into upper well alone or together with 1000 ng/ml C3a (Calbiochem), whereas the lower well contained 600 μl of serially diluted (0.1–2000 ng/ml) CXCL12 alone, or mixed with 1000 ng/ml C3a in 1% BSA, RPMI 1640.
For titration of C3a, REH cells were put into upper well alone or together with increasing concentrations of C3a (1–10,000 ng/ml) (Calbiochem), whereas the lower well contained 600 μl of CXCL12 alone (10 ng/ml), or mixed with the same concentration of C3a as in the upper well (1–10,000 ng/ml), in 1% BSA, RPMI 1640.
C3a cell binding studies.
CD34+ progenitor cells and CD19+ B cells (>98% purity) were sorted from BM using a MoFlo cell sorter (DakoCytomation). As a positive control, U937 cells were incubated for 72 h with 1000 U/ml IFN-γ (R&D Systems) to induce a high expression of C3aR (29). Triplicate samples of CD34+ progenitor cells, BM B cells, REH cells, or U937 cells (106 cells/experimental point) were incubated at 22°C with 1 nM 125I-C3a (PerkinElmer Life Sciences) and increasing concentrations (0.5–300 nM) of unlabeled C3a (Calbiochem). Aliquots (50 μl) of samples were layered on 300 μl of a 3/4 dibutyl, 1/4 dinonyl (v/v) phthalate (ICN Biomedicals) mixture. Cells were sedimented by centrifugation, tubes cut, and the pellets and supernatants were assessed for cell-bound and free 125I-C3a, respectively, by gamma counting. Kd (affinity of binding) was determined by iterative curve fitting of the binding data using GraphPad Prism 4 software (GraphPad Software).
CXCL12 cell binding studies.
REH cells (106 cells/experimental point) were incubated at 40C with 0.25 nM of 125I-CXCL12 (PerkinElmer Life Sciences) and increasing concentrations (0.5–500 nM) of unlabeled CXCL12 (R&D Systems). Cell bound and free 125I-CXCL12 counts were obtained and analyzed as described above. To assess whether C3a affected CXCL12 binding, we used the following controls before proceeding with binding assay as described above: 1) C3a (1000 ng/ml) was coincubated for 30 min with 0.25 nM 125I-CXCL12; 2) cells were preincubated for 30 min with C3a (1000 ng/ml) at 37°C or 4°C; 3) cells were preincubated for 30 min with C3a (1000 ng/ml) at 37°C and then washed by centrifugation.
CXCL12-C3a binding studies.
Flat bottom Immulon 2HB wells (Thermo Labsystems) were coated with 1000 ng/ml C3a (Calbiochem) per well, or with BSA (Fisher Scientific) (background binding control). Coating was done at 4°C, over night in pH 9.0 bicarbonate buffer. Wells were washed three times in bicarbonate buffer and blocked for 1 h at 37°C with 1% FCS. Wells were incubated for 45 min with 0.25 nM 125I-CXCL12 (PerkinElmer), washed and plate-bound radioactivity analyzed by gamma counting. In competitive displacement experiments, C3a-coated wells were first preincubated with increasing concentrations (0.1–10,000 ng/ml) of unlabeled CXCL12 and then exposed to 0.25 nM of 125I-CXCL12. BSA-coated wells were used as control for background binding.
C3a, C3a-desArg, and C4a enhance human BM CD34+ progenitor and B cell migration to CXCL12
Previous reports indicated that C3a can increase chemotaxis of stem/progenitor cells to CXCL12 (22). We questioned whether complement anaphylatoxins, including C3a, C3a-desArg, C4a, and C5a might also influence CXCL12-mediated migration of BM B lineage cells, which highly express CXCR4 (25). Given that C3a is a chemotactic factor for some human leukocytes, we first assessed whether C3a by itself could induce migration of BM stem/progenitor and B cells (2, 5). Depending on the cell type, the optimal chemotactic dose of C3a varies from nanomolar to micromolar concentration, therefore, we tested cell migration to a wide range of C3a concentrations (2). C3a, C3a-desArg or C5a alone, at concentrations ranging from 1 to 20,000 ng/ml, failed to stimulate migration of CD34+ progenitors and B cells (data not shown).
Next, we assessed the influence of anaphylatoxins on the migration of human BM cells to a suboptimal concentration of CXCL12 (10 ng/ml), which was added to the lower well of the migration chamber. Anaphylatoxins (1000 ng/ml, determined in titration experiments to be the dose causing the highest enhancement of migration, see Fig. 5,b) were added to both upper and lower wells of the chamber to ensure that they did not form any chemotactic gradient. The fixed dose of C3a, C3a-desArg, and C4a significantly enhanced the chemotactic response of BM CD34+ progenitor cells and all BM B cell subsets to CXCL12 (pre-pro, pro, pre, immature, and mature) (Fig. 1,a). Enhancement of migration was highest for C3a, lower for C3a-desArg, and C4a. Interestingly, preincubation of progenitor and B cells with C5a had no effect on their chemotaxis toward CXCL12 (Fig. 1 a).
C3a enhancement of B cell migration is chemokine specific
In prior experiments, we found that only late stage human BM B cells, i.e., immature and mature, migrated to CCL19 and CXCL13 chemokines, ligands for CCR7 and CXCR5 receptors, respectively. Human CD34+ progenitor cells and early stages of B cell development (pro-B and pre-B) did not migrate to CCL19 or CXCL13 (manuscript in preparation). Therefore, it was relevant to assess whether C3a (1000 ng/ml) could augment the chemotactic responses of late stage BM B cells to suboptimal concentrations of these chemokines. Based on bell-shaped dose-response curves, 100 ng/ml for CCL19 and 2000 ng/ml for CXCL13 were suboptimal chemotactic concentrations. These concentrations were different than the concentration used for CXCL12, as chemokines differ in their potency to induce migration. C3a significantly enhanced CCL19-mediated chemotaxis, but had no effect on B cell chemotaxis to CXCL13 (Fig. 1 b) and CCL20 (data not shown). Thus, C3a enhancement of cell migration is chemokine specific.
Human BM CD34+ progenitor and B lineage cells do not express surface C3aR
As a first measure of investigating the mechanism of C3a interaction with BM cells, we assessed whether CD34+ progenitors and developing B cells expressed C3aR on their surface. BM cells were analyzed by flow cytometry with anti-C3aR mAb 8H1. Because different anti-C3aR Abs have produced conflicting flow cytometric results in the literature (7, 30), we confirmed the specificity of anti-C3aR mAb 8H1 using undifferentiated U937 cell line as a negative control (data not shown) and IFN-γ-differentiated monocyte/macrophage U937 cells as a positive control (Fig. 2) (29, 31). Neither CD34+ progenitors nor BM B cells demonstrated any C3aR-specific fluorescence as compared with the IgG1 isotype control, whereas the differentiated U937 cells showed high C3aR expression (Fig. 2). To rule out the possibility that lack of detectable C3aR expression was due to variations in antigenic epitopes recognized by different mAbs, we stained freshly isolated BM cells with an additional anti-C3aR Ab (22), which also did not detect C3aR expression on CD34+ progenitor or B cells (data not shown).
The C3aR antagonist (SB290157) does not influence C3a-mediated enhancement of hemopoietic cell chemotaxis to CXCL12
Next, we assessed the ability of the C3aR antagonist SB290157 to influence C3a-mediated enhancement of BM progenitor and B cell chemotactic response to a suboptimal concentration of CXCL12. SB290157 inhibits, at nanomolar concentrations, C3a-C3aR binding and subsequent calcium mobilization and receptor internalization (28). To ensure saturation of the C3a binding sites on BM cells, we used 100 times higher, micromolar concentrations (10 or 100 μM) of the inhibitor in our chemotaxis assays. Cells were first preincubated for 30 min with SB290157 before the migration assay, and subsequently, the same concentration of inhibitor was added to both wells of the migration chamber to provide constant exposure of the cells to the C3aR inhibitor. We also confirmed that the inhibitor did not have an enhancing effect on cell migration to CXCL12, as cell treatment with 100 μM of SB290157 alone did not change cell chemotaxis to chemokine (Fig. 3). Exposure of the cells to a high concentration (100 μM) of C3aR antagonist did not inhibit C3a augmentation of CXCL12-mediated progenitor and B cell migration (Fig. 3). These data indicated that both CD34+ progenitor and B cells lacked functional expression of C3aR.
125I-C3a binding studies of human BM CD34+ progenitor cells and B lineage cells
To evaluate the possibility that B cells might express an alternative receptor for C3a, which was not recognized by the two monoclonal anti-C3aR (clones 8H1 and 218), we performed competitive 125I-C3a binding studies. Highly purified, freshly isolated human BM CD34+ progenitor cells and B cells were incubated with 1 nM 125I-C3a and with additions of increasing concentrations of unlabeled C3a (up to 300 nM). Binding data were analyzed by iterative curve fitting to one site model. Differentiated U937 cells, which express high levels of C3aR, were used as a positive control (29, 31). Differentiated U937 cells showed a high affinity binding of 125I-C3a, with one binding site of apparent Kd equal to 1.1 ± 0.2 nM (Fig. 4, inset). BM B cells bound low levels of 125I-C3a only, and this was clearly nonspecific binding because it was not inhibited by increasing doses of native ligand (Fig. 4). Unexpectedly, CD34+ progenitor cells demonstrated a high affinity binding of 125I-C3a, as indicated by inhibition of radiolabeled C3a binding by increasing doses of competing ligand and the characteristic shape of the binding curve. C3a bound to CD34+ progenitor cells with apparent Kd of 5.22 ± 0.43 nM. The estimated average number of C3a binding sites per cell was ∼12 times lower for CD34+ cells (3,200 C3a binding sites/cell) than for IFN-γ-stimulated U937 cells (36,000 C3a binding sites/cell) (Fig. 4).
C3a enhances chemotaxis to CXCL12 of C3aR-negative REH pro-B cells in a dose-dependent manner
To confirm our findings that C3a enhancement of CXCL12 activity was C3aR-independent, and gain insight as to how C3a might function, we studied a C3aR-negative CXCR4-positive B cell line. The CXCR4+ pro-B cell line REH (32) was negative for C3aR Ag surface expression by flow cytometry (data not shown). Functional assays for the C3aR on REH cells were also negative. These assays included radiolabeled ligand binding and anaphylatoxin-induced activation of calcium flux (data not shown).
We investigated the influence of C3a on REH pro-B cell migration to a wide range of CXCL12 concentrations (0.01–2000 ng/ml). Exposure of REH cells to C3a (1000 ng/ml) dramatically increased their chemotactic responses to suboptimal concentrations of CXCL12 (0.1–10 ng/ml). Addition of C3a to suboptimal concentrations (1 and 10 ng/ml) of CXCL12 increased REH cell migration but did not further increase chemotaxis to optimal concentrations of CXCL12 (100–300 ng/ml). C3a decreased chemotaxis to higher concentrations of CXCL12 (1000–2000 ng/ml). Thus, C3a caused a shift in the CXCL12 chemotactic response curve of REH cells (Fig. 5 a), providing further evidence that all the signaling is going through CXCR4.
The influence of varying concentrations of C3a (1–10,000 ng/ml) on REH cell migration toward a fixed dose of 10 ng/ml CXCL12 was assessed. A significant increase of REH cell migration was already noted at C3a concentration as low as 1 ng/ml. The maximal enhancing effect was observed at a C3a concentration between 1,000 and 5,000 ng/ml, and was decreased when C3a was added at 10,000 ng/ml (Fig. 5 b). Thus, we found that C3a increased CXCL12 chemotaxis even at low physiological concentrations (33). The highest enhancement of migration was observed when the CXCL12:C3a molar ratio was between 1:100 and 1:500.
C3a enhances CXCL12-induced chemotaxis of BM progenitor, B and T cells from C3aR knockout mice
To substantiate our findings that C3a-enhanced chemotaxis occurred independently of the C3aR in human BM cells, we assessed whether C3a-enhancing effect also was present in murine cells lacking the C3aR due to targeted disruption of the C3aR gene. First, in a control experiment, we determined that 1000 ng/ml C3a significantly enhanced migration to CXCL12 (10 ng/ml) of murine BM progenitor cells, T cells and developing B cells from wild-type BALB/c mice (Fig. 6). Next, we tested C3a-mediated enhancement of BM cell migration to CXCL12 in C3aR knockout mice: C3a significantly (p < 0.001) enhanced migration to CXCL12 by all subsets of BM B cells (pre-pro-B, pro-B, pre-B, immature, and mature B) isolated from the C3aR knockout mouse (Fig. 6,a). C3a enhancement of CXCL12-mediated migration was also found for Sca-1+ progenitor cells as well as BM T cells derived from C3aR knockout animals (Fig. 6 b).
C3a increases CXCL12 binding affinity for CXCR4
At this juncture, we theorized that C3a might affect CXCL12 binding to CXCR4+ cells. We used the C3aR-negative REH cells to carry out CXCL12/CXCR4 binding studies in the presence of C3a. Baseline CXCL12 binding to REH cells was characterized by competitive binding studies, using a fixed dose of 125I-CXCL12 (0.25 nM) and increasing concentrations of unlabeled CXCL12 (1–300 nM). 125I-CXCL12 bound to REH cells with one apparent binding site and with a Kd of 25.15 ± 1.9 nM (Fig. 7).
When 125I-CXCL12 was preincubated with C3a in solution, iterative curve fitting showed an apparent Kd of CXCL12 binding as 15.45 ± 1.4 nM. Thus, C3a caused an increase in the affinity of CXCL12 binding to CXCR4, possibly through direct interaction with CXCL12 in solution (Fig. 7).
To reproduce conditions of C3a-CXCL12 interaction, which might better model the chemotaxis assay, REH cells were first preincubated with 1000 ng/ml C3a, either at 37° or 4°C before addition of 125I-CXCL12. Preincubation at 37°C decreased the Kd of CXCL12/CXCR4 binding to 10.75 ± 0.1 nM, whereas preincubation at 4°C decreased the Kd further to 6.75 ± 0.2 nM (Fig. 7). Thus, preincubation of cells with C3a caused an even greater increase in CXCL12-CXCR4 binding affinity.
To evaluate whether C3a induced a change in the cells or alternatively affected the chemokine itself, we conducted the following experiment. REH cells were preincubated with C3a as described above and then washed to remove any free or low affinity bound C3a. Removal of the C3a abrogated its enhancing effect on 125I-CXCL12 binding to REH cells, indicating that C3a had to be present to modify CXCL12 cell binding (Fig. 7).
CXCL12 chemokine directly binds to C3a
To test the hypothesis that C3a directly interacted with CXCL12, we examined binding of soluble 125I-CXCL12 to C3a (1000 ng/ml) that was immobilized in plastic wells. C3a-bound CXCL12 was detected by gamma counting. 125I-CXCL12 binding to BSA-coated wells was used as a background control. 125I-CXCL12-specific binding to C3a was high and statistically significant, p = 0.00001 (Fig. 8,a). Next, to evaluate the specificity of CXCL12-C3a binding, we performed the experiment in which direct C3a-CXCL12 interaction could be demonstrated under equilibrium binding conditions. C3a was immobilized in plastic microtiter wells and a variable dose of “cold” CXCL12 ligand (from 0.1 to 10,000 ng/ml) was added, followed by a fixed dose of 125I-CXCL12. Specificity of the C3a-CXCL12 interaction was indicated by the ability of increasing doses of “cold” CXCL12 to block radiolabeled CXCL12 binding to immobilized C3a (Fig. 8 b).
C3a and C4a anaphylatoxin proteins are generated by BM stromal cells
Previous reports indicated that the complement component C3, is a physiologic constituent of the BM microenvironment (22, 34). We were interested in whether active complement fragments derived from the C3, C4, and C5 cleavage, anaphylatoxins C3a, C4a, and C5a could be generated by human BM stromal cells (35). As controls, we examined whether human B cell lines REH and HS Sultan or human kidney epithelial cell line 293 could generate anaphylatoxins. C3a, C4a, or C5a were not detected after 24 h in the serum-free culture supernatants of these cells (data not shown). In contrast, “second passage” primary human BM stromal cells, which were cultured for 24 h in serum-free conditions, generated high amounts of C3a and C4a but not C5a (Fig. 9) in the culture supernatants. In addition, we confirmed that the same stromal cells secreted CXCL12. The amounts of C3a and C4a detected in stromal cell supernatant were significant and comparable with levels found in normal human plasma (33).
CXCL12 is constitutively expressed by BM stromal cells in the BM microenvironment. The CXCL12/CXCR4 axis appears not only important for the retention and maturation of stem/progenitor cells in the BM but also for the homing and engraftment of stem/progenitor cells to the BM (15, 17). It was recently shown that C3a accelerated homing of murine progenitor cells to the BM and in vitro experiments with human stem cells documented that C3a could potentiate CXCL12-induced migration (22). We further investigated the mechanism by which complement anaphylatoxins modulate chemokine-induced cell responses in well defined BM cell populations expressing high levels of CXCR4 (25). Our initial hypothesis was that C3aR was present on the surface of BM CD34+ progenitor and B cells, and that signaling through C3aR (e.g., calcium flux) could modify CXCL12-mediated chemotaxis. However, subsequent experiments indicated that C3a mediated enhancement of CXCL12-induced chemotaxis of BM hemopoietic cells is independent of C3aR. First, human BM CD34+ and B lineage cells do not express C3aR by flow cytometry (Fig. 2 and data not shown). Second, the competitive C3aR inhibitor SB290157 does not affect C3a-mediated enhancement of CXCL12-induced chemotaxis (Fig. 3). Third, enhancement of chemotaxis of hemopoietic cells is also mediated by C3a-desArg, which does not bind to C3aR (Fig. 1,a). Finally, C3a-enhanced CXCL12 induces chemotaxis of BM cells from C3aR knockout mice similar to BM cells from wild-type mice (Fig. 6).
To rule out the possibility of an antigenically distinct receptor for C3a, we assessed 125I-C3a binding to BM CD34+ and B cells. In agreement with FACS data, we did not observe specific binding of 125I-C3a to the surface of primary BM B cells (Fig. 4). However, we did detect a high affinity (Kd equal to 5.22 nM) binding of radiolabeled C3a to CD34+ progenitor cells (Fig. 4). The estimated average number of C3a binding sites/cell was ∼12 times lower for CD34+ cells (3,200 C3a binding sites/cell) than for the positive control, i.e., IFN-γ-stimulated U937 cells (36,000 C3a binding sites/cell). These data could be interpreted in two ways: each CD34+ cell has a similar low number of C3a binding sites, or alternatively the number of C3a binding sites/cell differs among the CD34+ cells, because CD34+ cells represent a heterogeneous population of both stem and committed progenitor cells. It is thus possible that a minor subpopulation of CD34+ cells possesses a relatively high number of C3a binding sites.
These findings were indeed unexpected in view of the lack of C3a binding to B lineage cells. Because we were unable to detect C3aR using two different mAb reagents (Fig. 2 and data not shown), it is possible that C3a binding to freshly isolated CD34+ progenitor cells occurs via an alternative receptor for C3a. Nevertheless as summarized above, similar to B lineage cells, several lines of evidence indicate that C3a-mediated enhancement of chemotaxis of CD34+ cells also is independent of C3aR (Figs. 1–3). Thus, further studies are needed to determine whether C3a binding to CD34+ cells possibly via an alternative receptor for C3a may contribute to C3a-mediated enhancement of chemotaxis of CD34+ cells to CXCL12.
C5L2 is a potential alternative receptor for C3a and C3a-desArg (36) and thus could play a role in C3a-mediated enhancement of CXCL12-induced chemotaxis of hemopoietic cells. To explore this possibility, we first stained human BM progenitor and B cells with a commercially available anti-C5L2 Ab (Cell Sciences). Staining with this Ab did not reveal the presence of C5L2 Ag on human BM CD34+ and CD19+ cells (data not shown). Moreover, through collaboration with Dr. C. Gerard (Children’s Hospital Boston, Boston, MA) we found that both C3a and C3a-desArg are potent enhancers of CXCL12-mediated migration of BM C5L2-deficient progenitor cells, B and T cells isolated from C5L2 knockout mice (44).
To our knowledge, this is the first report investigating the expression of C3aR on freshly isolated BM stem/progenitor and B lineage cells. Our data are in agreement with a report showing that freshly isolated peripheral blood B cells also lack C3aR expression (30). In contrast, Reca et al. (22) presented evidence for C3aR expression on CD34+ progenitor cells. This discrepancy might be explained by different treatment of CD34+ cells before experiments. Whereas our data were generated using freshly isolated, un-manipulated BM CD34+ cells, the studies by Reca et al. (22) used immunomagnetic bead-isolated CD34+ cells sorted from cord blood or from mobilized peripheral blood. In this regard, staining of freshly isolated BM cells with the same clone of anti-C3aR Ab as used in studies by Reca et al. (22) also did not detect C3aR expression on BM progenitor cells and B cells (data not shown).
We do not know the mechanism by which C3a modulates CXCL12-induced chemotaxis in a C3aR-independent manner. We found, however, that C3a can directly interact with CXCL12 and that C3a also affects CXCL12 binding to cells (Figs. 7 and 8). The presence of C3a shifts the binding curve of radio-iodinated CXCL12 in a CXCR4+/C3aR− progenitor B cell line (Fig. 7). The fact that C3a as well as C3a-desArg and C4a share similar activity and all are strongly positively charged is intriguing (37). Importantly, C5a, which failed to enhance CXCL12 migration, lacks this strong charge (37, 38). CXCL12 has a heparin-binding domain, and heparin, like the active complement anaphylatoxins, has a strong positive charge (39). Anaphylatoxins might aggregate CXCL12 and thereby make the chemokine more active or protect CXCL12 from the cleavage by peptidases (40, 41). The exact molecular mechanism(s) by which C3a interacts with CXCL12, especially specificity and affinity of this interaction, remains the subject of our ongoing studies using the Biacore system. We found that the increase in CXCL12 receptor binding affinity is higher when cells are first exposed to C3a followed by the addition of CXCL12, compared with the simultaneous exposure to C3a and CXCL12 (Fig. 7). This could argue in favor of C3a binding to the cells, but if such binding occurs, it must be of low avidity because it was removed by washing the cells (Fig. 7, lowest bar).
The importance of our findings is that complement anaphylatoxin generation provides a means to quickly and transiently modulate CXCL12 potency for what is otherwise a constitutively active chemokine. Of the active anaphylatoxins, C3a and C3a-desArg may be more important than C4a because C3 is present, at least in plasma, at a 2.5-fold higher concentration than C4 (reviewed in Refs.42 and 43). Although we do not know the mechanism by which stromal cells generate complement anaphylatoxins, the fact that C4a is also generated is consistent with activation of the lectin and/or classical pathways. The capacity of stromal cells to generate complement anaphylatoxins (Fig. 9) potentially provides local control of CXCL12-mediated cell migration in BM microenvironments. Moreover, the observation that in vitro exposure of stem/progenitor cells to C3a enhances their BM homing and engraftment is of clinical importance to BM transplantation, where in some instances the number of available stem/progenitor cells is limited resulting in delayed or non-engraftment. Thus, it may be feasible to enhance transplant efficiency by use of molecules with physicochemical properties similar to C3a.
We thank Drs. Norma P. Gerard, Craig Gerard, and Alison Humbles (Children’s Hospital Boston, Boston, MA) for help with experiments, valuable discussion of the results, and for providing C3aR knockout mice; Dr. David E. Isenman (University of Toronto, Toronto, Canada) for critical reading of the manuscript and helpful suggestions, and Dr. James J. Campbell (Children’s Hospital Boston, Boston, MA) for discussion. We thank Dr. David Miklos (Dana-Farber Cancer Institute, Boston, MA) for help with human BM aspirates.
The authors have no financial conflict of interest.
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 the following grants from the National Institutes of Health: P0I HL56949 (to L.E.S.), R01 AI42987 (to A.N.-W.), and R01 HL61796 (to M.Z.R.).
Abbreviation used in this paper: BM, bone marrow.