The role of G-protein signaling in hematopoietic stem/progenitor cell mobilization

Heterotrimeric guanosine triphosphate (GTP)–binding proteins (G proteins) participate in one of the most important signaling cascades used to relay extracellular signals in eukaryotic cells. Activation of G proteins and their effectors mediate a broad spectrum of cellular responses initiated by receptors coupled to G proteins (GPCRs). The nature of the response to a stimulus in any given cell is dictated by the receptor density, type of G protein activated, its duration of activation (also influenced by regulator proteins [RGS]), and the complement of effector molecules in each cell type.^1-4  G proteins have been functionally categorized by their susceptibility to bacterial toxins. The widely distributed G_i and G_s proteins are inhibited by pertussis toxin (G_i), or activated by cholera toxin (G_s). Pertussis toxin (Ptx) in particular has been instrumental in delineating signaling pathways dependent on G proteins. Much of the evidence favoring G-protein involvement in T-cell function is derived from studies using Ptx to block G_i function. In vivo, Ptx causes lymphocytosis by inhibiting lymphocyte migration from blood to tissues, or across lymph node endothelium.^5-8  In mature leukocytes, chemotactic receptors and their ligands signal through G proteins, and such signaling is essential for their adhesion and directed migration to inflammatory sites. This G protein–dependent effect is suppressed by Ptx.^9-11  Leukemic cell metastases also appear to be mediated through G-protein signaling,^12-17  although only certain G proteins are involved in some tissues.¹⁴  Signals from CXCR4R, a GPCR present in hematopoietic stem/progenitor cells (HPCs),^18-20  are important for their maintenance within bone marrow (BM).^21-23  Down-regulation of CXCR4 function leads to mobilization of stem/progenitor cells,^24-26  although conflicting views have been presented as to the nature of CXCR4/stomal-derived factor 1 (SDF-1) signaling required for mobilization.²⁷  The mechanistic effects of CXCR4/SDF-1 signaling on hematopoietic cells, either immature or mature, are dependent on integrin activation, and are largely delivered through Ptx-sensitive G proteins.^28-30 

To directly address the in vivo effects of G-protein signaling on the migration of HPCs from BM to blood or other tissues, we treated mice with Ptx and assessed its impact on mobilization of HPCs. We found that in addition to the expected leukocytosis, there was a protracted effect on HPC mobilization. Detailed in vitro studies using the plasma or cells from Ptx-treated animals have uncovered helpful insights for interpreting the longevity of the Ptx effect. The type of functional impairment of mobilized cells after Ptx treatment and the responses of several murine knockout models also provided a framework for approaching some of the molecular dynamics brought by Ptx treatment in vivo.

All mice used in this study were housed under specific pathogen-free (SPF) conditions and were free from known mouse pathogens. B6D2F₁ mice, WCB6F₁ Sl/Sl^d or Steel-Dickie and WCB6F₁ (+/+) littermates, WBB6F₁ (W/W^v) mice with WBB6F₁ (+/+) littermates were purchased from Jackson Laboratories (Bar Harbor, ME). The origin of mice deficient in granulocyte colony-stimulating factor receptor (G-CSFR^-/-), metalloproteinase 9 (MMP-9^-/-), monocyte chemoattractant protein 1 (MCP-1^-/-), or CD18 (CD18^-/-) and selectin (E+L+P^-/-) has been mentioned previously.^31,32  Both Tpo^-/- mice and double knockout Tpo^-/-, G-CSFR^-/- mice were kindly provided by Dr K. Kaushansky (University of California at San Diego). Nonobese/severe combined immunodeficient (NOD/SCID) mice were from Jackson Laboratories and Rag2^-/- from Taconic (Oxnard, CA). In addition to mice, one pigtailed monkey (Macaca nemestrina) was used and was housed in the facilities of the accredited University of Washington Regional Primate Center.

Anti-CD49d (very late activation antigen 4 [VLA-4] or α₄) was kindly provided by Dr Roy Lobb (Biogen, Cambridge, MA) and anti-CD106 (vascular cell adhesion molecule 1 [VCAM-1]) was purchased from Southern Biotech (Birmingham, AL). These contained no azide and low endotoxin (NA/LE). The following antibodies, either purified or directly conjugated, were purchased from BD Pharmingen (San Diego, CA): anti-CD3, anti-CD11a (leukocyte function associated antigen 1 [LFA-1]), anti-CD31 (platelet-endothelial cell adhesion molecule 1 [PECAM-1]), anti-CD41, anti-CD117 (c-kit), Sca-1, GR-1, TER-119, CD45R/B220, anti-CXCR4, and anti-CD29. Directly conjugated anti-VLA-4 was purchased from Southern Biotech. Ptx, Ptx B oligomer, and cholera toxin were purchased from List Biological Laboratories (Campbell, CA). Granulocyte colony-stimulating factor (G-CSF; filgrastim) was purchased from Amgen (Thousand Oaks, CA). Fucoidan was from Sigma (St Louis, MO), and murine SDF-1 was from Peprotech (Rocky Hill, NJ). Both the annexin V detection kit and the bromodeoxyuridine (BrDU) flow kit were from BD Pharmingen.

Levels of cytokines and chemokines in murine or primate plasma from peripheral blood (PB) or BM were analyzed by the Cytokine Analysis Laboratory, Fred Hutchinson Cancer Research Center (FHCRC; Seattle, WA). Endotoxin levels in plasmas from treated or control mice were analyzed by the Biologics Production Facility at FHCRC.

For cell cycle analysis, 10⁶ BM or spleen cells were permeabilized with PET buffer (phosphate-buffered saline plus [PBS] 1.0 mM EDTA [ethylenediaminetetraacetic acid] plus 0.1% Triton X-100) for 30 minutes on ice and labeled with propidium iodide (2 μg/mL), and were then analyzed on a FACS Calibur (BD Immunocytometry Systems, San Jose, CA) using ModFit-LT software (Verity Software House, Topsham, ME). To study actively proliferating cells, BrDU was injected at 167 mg/kg body weight intraperitoneally twice during the 24 hours before Ptx injection. An aliquot of white blood cells (WBCs) from lysed blood and BM 3 days after Ptx injection was stained with kit-APC at 10⁶ cells/mL for 30 minutes at 4°C. Control aliquots were stained with isotype control antibody. Cells were processed and stained with anti-BrDU-fluorescein isothiocyanate (FITC), as per BD Pharmingen protocol, then analyzed by flow cytometry.

Colony-forming unit-culture (CFU-C) assays were performed using a methylcellulose mixture and cytokine cocktail and were counted as described previously.³⁰  Colonies were scored as granulocyte-macrophage colony-forming units (CFU-GM), mixed colony-forming units (CFU-Mix), erythroid burst-forming units (BFU-E), and on some experiments as megakaryocyte colony-forming units (CFU-Meg). All types of colonies are reported as CFU-Cs.

Primary recipient mice were irradiated with 1200 cGy from a cesium source (at the dose rate of > 100 cGy/min) and then injected intravenously with either 0.1 mL or 0.2 mL blood from Ptx-treated or normal mice. The survival of mice for the next 30 days was recorded and moribund animals were humanely killed.

PB collected in heparin was layered over Histopaque 1.083 (Sigma Aldrich, St Louis, MO), centrifuged, and interface cells washed. To obtain lineage-negative (Lin^-) cells, we incubated them with lineage-specific antibodies (anti-CD3, -CD45R/B220, -CD11b, GR-1, and TER-119) followed by secondary antibody treatment and negative selection using a Midi-MACS system (Miltenyi Biotec, Auburn, CA). Lin^- PB cells were adjusted to 2 × 10⁶/mL and migration assays through 5-μm Transwells (Corning Costar, Cambridge, MA) were performed as described.³³  Migrated cells were collected from the bottom wells and total cells counted. The percentage of migrated cells was calculated relative to the total number of cells inoculated in the upper chamber.

Lin-depleted cells isolated as described were adjusted to 1 × 10⁶ cells/mL in Iscove modified Dulbecco medium (IMDM) plus 0.5% bovine serum albumin (BSA) and stimulated with SDF-1 at 100 ng/mL at 37°C.³⁴  At indicated time points, 100-μL aliquots were fixed in 100 μL 3.7% formaldehyde in PBS. The cells were then permeabilized and stained with Phalloidin Alexa 568 (Molecular Probes, Eugene, OR), as described³⁴  and analyzed on a FACSCalibur. To test the effect of red blood cells (RBC) or plasma from Ptx-treated mice on SDF-1–induced actin polymerization, 1 × 10⁶ Jurkat (human lymphoid cell line) cells were incubated overnight with either plasma or washed RBCs from both untreated mice and Ptx-treated mice, then assayed as described (see “SDF-1–dependent chemotaxis”).

The effect of Ptx on leukocyte redistribution was described more than 3 decades ago. To test whether Ptx could also increase circulating HPCs and to find the effective doses required, we treated groups of mice with a single injection of Ptx at 3 different doses (100, 200, 400 ng) and quantitated WBCs and circulating HPCs 3 days later. After Ptx treatment, a significant lymphocytosis and granulocytosis and a significant increase in circulating HPCs (from 10- to 20-fold above baseline) was documented. Because no significant differences were seen among the 3 doses tested, we adopted a dose of 100 ng/mouse in all subsequent experiments. To define the kinetics of CFU-C response relative to WBC increase, several groups of mice were treated with a single injection of 100 ng Ptx and analyzed at 1, 3, and 24 hours and daily thereafter for 8 days.

Figure 1 shows a peak in leukocytosis at about 72 hours (45 035 ± 3248 WBC/μL blood), and a peak in CFU-C levels at about 96 hours (4403 ± 1610 CFU-C/mL blood). An absolute increase in all lymphocyte subsets (data not shown) and in granulocytes was seen, as previously noted.³⁵  After Ptx treatment all classes of committed progenitors (BFU-E, CFU-GM, CFU-Meg, and CFU-Mix) were mobilized and found in proportions similar to ones in control mice (data not shown). In splenectomized animals, higher CFU-C mobilization was noted peaking at later times (day 7, CFU-Cs: 6237 ± 911 and WBCs: 42 916 ± 2436, n = 5). To test whether earlier progenitors like CFU-Ss and radioprotective cells were mobilized, we carried out short-term transplantation experiments using blood from Ptx-treated animals. For these experiments, an aliquot of blood (0.1 or 0.2 mL) from control or treated animals (a single injection of Ptx 72 hours previously) was given to each of the lethally irradiated recipients, and their survival was followed. Eleven recipients received blood from Ptx-treated animals and 8 control recipients were given blood from nontreated animals. Two moribund controls along with 5 recipients of blood from Ptx-treated mice were evaluated at day 12 for the presence of CFU-S12 in their spleens (Figure 1B inset). The estimated number of CFU-S12/mL blood in Ptx-treated animals was 78 ± 5.3 (n = 5) and 15 ± 2.2 in the 2 control mice. (The CFU-S levels in control mice were small, barely meeting the criteria of macroscopic colonies.) Furthermore, within the BM of the 5 mice that received blood from Ptx-treated animals, 62.0 ± 14.6 CFU-C/femur were recovered at day 12, and none from the BM of the 2 mice treated with control blood, suggesting the presence of higher numbers of short-term reconstituting cells in Ptx blood. Notably, mice that received blood from Ptx-treated mice survived beyond 5 weeks, whereas all 6 remaining control animals died before day 18 (Figure 2).

Ptx is a typical A-B toxin readily dissociated to S1 or A protomer (the subunit with the catalytic activity) and the pentamer or B-oligomer (responsible for binding to the cell surface). Most of the effects of Ptx on various cell types are attributed to its enzymatic activity,^36,37  despite the failure of the A protomer itself to act on intact cells, because the B-oligomer enables the protomer to traverse the plasma membrane to reach the site of action inside the cell. This binding is the first step required for the A protomer adenosine diphosphate (ADP)–ribosyltransferase to enter the cell. To test whether the mobilization of HPCs requires the enzymatic activity of Ptx, or whether the B-oligomer alone can elicit this effect, we treated animals with the B-oligomer and followed changes in WBCs and circulating progenitor cells for several days after this treatment. Two doses were used, 200 and 400 ng/mouse. At 3 hours after injection (Figure 1), a mild depression of cell numbers was seen, and for the next 3 days, the number of either total nucleated WBCs or circulating progenitors never exceeded the average numbers observed in the control animals at baseline. Neither the B-oligomer nor the holotoxin had a direct effect on the proliferation of CFU-Cs when added directly to cultures of BM or PB CFU-Cs at 10 ng/mL (data not shown). Treatment of mice with cholera toxin, which, in contrast to Ptx, activates G_s proteins and stimulates cyclic adenosine monophosphate (cAMP), elicited a significant depression of circulating WBCs and progenitors, lasting more than a week (Figure 1C-D).

To uncover the identity of mobilized cell populations along with alterations in their surface markers, we tested the expression of several antigens characterizing either progenitor-type cells (ie, kit or Sca-1), or cells of specific lineages (GR-1, MAC-1, TER-119, CD41, CD31, and B220 or CD3). In addition, the abundance of CXCR4 and of certain adhesion molecules, like α₄ (CD49d), β₁ integrin (CD29), LFA-1 (CD11a), and L-selectin (CD62L), was tested. Analysis was done 3 days after Ptx injection to match the CFU-C mobilization peak, and the data were confirmed in 3 independent experiments. After comparing mobilized cells to those remaining in the BM, we compared BM cells from normal and Ptx-treated animals.

The results of these experiments are presented in Figure 3 and Table 1 and can be briefly summarized as follows. There was a reduction in B220⁺ lymphoid cells within the BM as expected,^7-9  with a concomitant increase in the PB of treated animals (Figure 3); a reduction in CD31⁺ population in the BM; a significant increase in c-kit⁺ population in PB; a down-regulation of α₄-integrin expression in a subset of

To investigate whether similar changes can be seen in the primate model, we treated one monkey (M nemestrina) with 25 μg/kg Ptx and recorded the changes in BM and PB 3 days after treatment. A spectacular 45-fold increase in CFU-C levels in PB was noted in this animal (from 118.6 ± 9 to 5359 ± 705 CFU-C/mL of blood). The expression of α₄ and CXCR4 in PB and BM (Figure 3) followed the same trends as in mice, but there were some differences (ie, CD31, CD11a; Table 1).

To probe for alterations in the functional properties of Ptx-mobilized cells, we used 2 assays proven useful in previously published studies, that is, a migration assay in response to SDF-1 and actin polymerization in the presence or absence of SDF-1 using phalloidin staining. These showed that Lin-depleted murine cells from Ptx-treated BM or PB exhibited a very blunted response in SDF-1–induced actin polymerization compared with control cells and little or no migration toward SDF-1 showing that both chemokinesis and chemotaxis were reduced (Figure 5). Similar results were obtained when mononuclear or kit⁺ cells were used instead of Lin^- cells (data not shown). These data support a direct functional modification of both mature and immature hematopoietic cells by in vivo administration of Ptx.

To explore the protracted response of Ptx-induced mobilization, the following studies were done. Plasma from controls or Ptx-treated animals at several time points after treatment (3-5 days and 10 days) were tested using sensitive in vitro assays. To probe for persistence of Ptx-like activity, Jurkat cells were used as target cells and their response to SDF-1–induced actin polymerization was recorded (Figure 6).

Plasma collected between days 3 and 5 after treatment exerted a Ptx-like effect on these cells, whereas plasma at day 10 after treatment behaved as control plasma. We speculated that the enzymatic activity detected in plasma was likely released from cellular elements, that is, RBCs, that serve as depots of Ptx toxin⁶  and could explain its protracted effect on cells. This turned out to be true (data not shown). To inquire whether proliferative effects were present after Ptx treatment, we evaluated the proportion of cycling cells in BM and spleen (by propidium iodide or BrDU labeling of kit⁺ cells on day 3). Although there were small, nonstatistically significant increases in the cellularity of BM and spleen, there were significant elevations in the cycling of cells (control BM, 19.9% ± 1.3% in S phase versus Ptx BM, 24.8% ± 1.3%, P < 0.02; control spleen cells, 1.35% ± 0.2% versus Ptx spleen cells, 8.75% ± 1.5%, P < 0.002; Figure 7). Further evidence was provided by BrDU experiments documenting an increase in BrDU⁺ cells at day 3 after Ptx in BM and also a higher proportion of kit⁺ cells that were BrDU⁺ (25% in Ptx treated versus 20% in controls). To test whether cytokine/chemokine levels changed after Ptx treatment and mediated the effects on proliferation, we analyzed plasma samples from the Ptx-treated monkey before and during the 3 days after treatment by enzyme-linked immunosorbent assays (ELISAs; with the exception of SDF-1α and β, the assays available to us were for human rather than murine sources). In the PB, only interleukin 6 (IL-6) and granulocyte colony-stimulating factor (G-CSF) showed significant changes (Table 2). The levels of G-CSF were modestly increased and unlikely to be responsible for mobilization, although such levels could provide a proliferative stimulus if sustained. The IL-6 levels by day 3 were high, but its significance in mobilization is unclear in view of the general anesthesia required for each blood sampling in primates, the effects of Ptx on liver,³⁸  and the fact that similar changes have been seen previously with other treatments in monkeys albeit with different kinetics.³²  Its importance, however, should be further evaluated in IL-6^-/- mice. SDF-1 levels in PB remain unchanged, but there is a trend of SDF-1 reduction in BM samples (Table 2). Finally, it is of note that annexin V labeling tested in control and Ptx-treated BM and PB showed no significant differences between control and Ptx-treated cells (data not shown).

Because the in vivo effects of Ptx on mobilization are not immediate and show a delayed and progressive pattern, the sequence of molecular events leading to this process are difficult to unravel. To approach this issue in an indirect manner, we treated transgenic animal models deficient in MCP-1, MMP-9, G-CSFR, and CD18 integrins or in selectins, to explore the putative or indirect participation of any of these molecules in Ptx-induced mobilization. Rag 2^-/- mice were also treated to test whether mobilization of a lymphoid population was in some indirect way contributing to the maintenance of mobilization. Three to 5 mice were used per group and evaluated 3 days after a single injection of 100 ng Ptx per mouse. The results of this study are depicted in Figure 8.

Taking all data from these animals together, several conclusions can be drawn. The fact that Rag-2^-/- mice respond excessively with granulocytosis rather than lymphocytosis with a great increase in HPCs suggested that peripheral migration of lymphocytes per se does not influence the granulocytic or the progenitor cell response. The magnitude of progenitor mobilization cannot be predicted by the magnitude of WBC response. W/W^V mice responded well in WBCs, but less well in CFU-Cs, despite the fact that their BM CFU-C content is similar to controls.³⁹  Mice with a decrease in stem/progenitor pools (G-CSFR/Tpo^-/- mice,⁴⁰  or the Sl/Sl^d mice) responded less than their normal counterparts. In Sl/Sl^d mice, it is also possible that the decreased response is dictated by a decreased contribution of splenic responses, because the documented increased proliferation and cellularity in spleen may not be present in these mice. Mice with already increased baseline levels appear to reach much higher response levels than their normal controls. This is compatible with the expectation that a heightened response may be present in inflammatory situations, such as in CD18 or selectin-deficient animals. Because G-CSFR^-/- mice responded similarly or better than their controls, but do not respond to G-CSF– or IL-8–induced mobilization,⁴¹  one surmises that G-CSFR or IL-8 signals are not necessary for Ptx-induced mobilization. Likewise, MMP-9^-/- mice responded well and although a small increase in MMP-9 protein was seen in normal mice after Ptx treatment (data not shown), this was found after day 3 and is an unlikely contributor to Ptx-induced mobilization.

Coadministration of Fuc-S and Ptx. Because Fuc-S (fucoidan) elicits an immediate chemokine-like response (in contrast to Ptx), 2 treatments were designed incorporating either simultaneous or sequential administration of both agents. In the first experiment, both Fuc-S (100 mg/kg) and Ptx (100 ng/mouse) were simultaneously administered intravenously in 6 BDF1 mice. Mice were bled 1.5 hours later and WBC and circulating CFU-C levels were assessed. The same was repeated at 72 hours. The results show that there was leukocytosis and an increase in progenitors (Figure 9A) both at 1.5 hours (reflecting the effect of Fuc-S) and at 72 hours in the combined treatment (reflecting the Ptx effect) with no additive or synergistic effect compared with Ptx alone. Alternatively, when Ptx was given at day 1 and Fuc-S 72 hours later with the animals being bled 1.5 hours after Fuc-S, a very large response was seen in both the WBCs (data not shown) and the CFU-Cs (Figure 9B), above the levels achieved by either Ptx or Fuc-S alone. The heightened response seen when Fuc-S was administered at the peak of the expected Ptx response could denote an expanded pool of cells on which the Fuc-S could exert its mobilizing effect rather than cooperativity.

Blockade of α₄integrin and Ptx treatment. The impact of combined treatments of Ptx and of antibody blockade of α₄ integrin was subsequently tested. Because the mobilization kinetics with Ptx treatment are much slower than with anti-α₄ treatment (72 hours versus 24 hours), 2 different schedules of combined anti-α₄ and Ptx treatments were evaluated. In the first experiment, a single injection of anti-α₄ antibody (PS/2 at 2 mg/kg) plus Ptx was given intravenously simultaneously. Animals given only anti-α₄ or only Ptx were used as controls (n = 3). The numbers of circulating leukocytes or progenitors were evaluated before and 24 and 48 hours after the injection. In agreement with previous studies⁴²  a modest increase in WBC levels was seen at 24 hours in the group given anti-α₄ (data not shown), whereas a more pronounced leukocytosis was seen in the Ptx group and the group with the combined anti-α₄ plus Ptx treatment. Circulating progenitors increased approximately 4-fold at 24 hours in the single treatments with anti-α₄ (1431 ± 570/mL) or with Ptx (1271 ± 321/mL). A much higher level was observed in the combined treatment (5090 ± 1536/mL at 24 hours, P < .05; Figure 9C). Because the effect of Ptx is not seen during the first 3 hours, then slowly peaks at days 3 or 4, whereas the effect of anti-α₄ peaks during the first 24 hours, to synchronize the maximal effect of each treatment, in the next experiment Ptx injection was given first, followed 24 hours later by one anti-α₄ injection. Two other groups of mice received Ptx alone or anti-α₄ alone. The mice were evaluated 24 hours after anti-α₄ (48 hours after Ptx). Again, the anti-α₄ group had the expected 24-hour level of response, but now the Ptx alone group or the group with the combined treatment responded similarly, in contrast to the enhancement seen in the first experiment (Figure 9D). To provide additional data and to maximize the anti-α₄ effect, in 2 additional experiments we treated mice with one injection of Ptx plus anti-α₄, followed by 2 additional anti-α₄ injections, one at 24 hours and another 48 hours later. The results of these experiments (Figure 9E) are basically no different from those of the experiments with sequential Ptx and anti-α₄ treatment and, in aggregate, suggest that if the effect of Ptx precedes that of anti-α₄ blockade, no additional benefit is derived by the anti-α₄ treatment. However, if the antibody treatment has a functional influence before that of Ptx, an enhancement of the mobilizing effect is seen only at early time points. The above data, together with those documenting a down-regulation of VLA-4 expression in BM and PB cells after Ptx (Figures 3 and 4) suggest that a down regulation of VLA-4 function may be a consequence of Ptx-induced mobilization, as seen with other mobilizing agents.³⁹ 

Coadministration of Ptx and G-CSF. G-CSF alone at 100 μg/kg twice daily was administered to one group of 3 mice for 3 days, and 2 other groups (3 mice/group) received either G-CSF (for 3 days) and Ptx (at day 1) or Ptx alone. Seventy-two hours after Ptx injections, a dramatic mobilization of progenitors was noted in the combination treatment compared with G-CSF or Ptx alone (Figure 10), which exceeded the 5-fold effect of G-CSF alone and the 15-fold increase of Ptx alone. Although both agents have a rather protracted response in mobilization, the synergistic rather than additive effect suggests enhanced cooperativity in mobilization between these 2 agents. Because increases in MMP-9 have been previously assigned a critical role in G-CSF–induced mobilization,⁴³  we tested MMP-9^-/- mice with either G-CSF or G-CSF plus Ptx (Figure 10). Although G-CSF response in the latter mice was in the expected range of normal (+/+) animals, the synergy of G-CSF with Ptx found in normal mice was largely eliminated, suggesting that it is MMP-9 dependent.

A single injection of Ptx elicits a long-lasting lymphocytosis, granulocytosis, and, as shown in the present studies, a sustained increase in circulating hematopoietic progenitor cells. All 3 components of this response are dependent on the presence of the S1 enzymatic activity of the pertussis holotoxin documented here by comparing the effects of B-oligomer that lacks ADP ribosyltransferase activity to that of the holotoxin (Figure 1). Even if clearance between the 2 proteins is different, the B-oligomer was unable to elicit a response during the first 24 hours. What is the long-lasting effect of a single injection of Ptx treatment due to? To provide insight we tested the plasma from Ptx-treated animals several days after treatment for the detection of persisting Ptx-like activity in vitro. The amount of Ptx protein present in plasma at 3 days after treatment or later was capable of inhibiting the SDF-1–induced actin polymerization in vitro in contrast to normal plasma (Figure 6). Such a Ptx-like activity in circulation for several days after a single Ptx injection originates from cellular depots from which it is slowly and continuously being released into the circulation. In addition, an increase in proliferation of HPCs was documented in BM and spleen (Figure 7) that could enhance the sustained increase in circulation of both polymorphonuclear cells (PMNs) and progenitor cells. We speculate that the effects on proliferation represent a secondary, indirect effect elicited by Ptx treatment. A survey of cytokines up to 4 days after Ptx treatment did not reveal any increase in IL-1α, GM-CSF, macrophage colony-stimulating factor (M-CSF), or stem cell factor (SCF). However, a significant increase in IL-6 and a modest increase in G-CSF were noted at day 3 (in the monkey), which, if persisted, could contribute to proliferative responses. Elaboration of additional (not identified) diffusible molecules or cytokines responsible for the increased proliferation cannot be excluded. The possibility that a Ptx-induced intrinsic alteration in cell signaling changing not only the adhesive properties, but the cycling of hematopoietic cells was also considered,^44,45  but thought to be unlikely, because Ptx added in vitro influenced neither the number nor the size of hematopoietic colonies. Furthermore the effects of Ptx-induced perturbations of CXCR4/SDF-1 (CD184/CXCL12) signaling on proliferation, both in vitro and in vivo, are controversial,^46-50  but recent studies suggest that injection of an antagonist (SDF-1-G2) reactivates the cycling of primitive stem cells in vivo.⁵⁰ 

The mobilization of stem/progenitor cells observed in our studies with Ptx treatment invite comparisons with previously described mobilization of CFU-S by endotoxin from Escherichia coli or Salmonella typhi.⁵¹  High doses of injected endotoxin (500 μg) caused an immediate but transient increase (peak at 1-2 hours) in circulating CFU-S levels, without any increase in total granulocytes or total WBCs, and a later peak several days after treatment. Such a response was not seen in our Ptx-treated animals, in which no responses were documented from 1 to 3 hours. There was also a gradual increase in granulocytes, to heights never seen with the endotoxin, and in progenitor cells for the next 10 days. At no point was there any decrease in bone marrow cellularity, or anemia and thrombocytopenia, as observed in endotoxin-treated animals.⁵¹  Our results were induced by the injection of only 100 ng Ptx per mouse, but endotoxin injection even at 100 μg/mouse produced a small or an erratic response.⁵¹  Furthermore, endotoxin levels in plasma of our Ptx-treated mice were very low and similar to levels of untreated animals (data not shown). Finally, circulating cells after Ptx treatment showed specific responses in vitro (Figure 5), which are only compatible with the inhibition of G_i-protein signaling.

Responses to Ptx treatment of several murine models with targeted adhesion molecules or cytokines/chemokines led to several indirect conclusions relevant to the mechanism of Ptx-induced mobilization, and support the independence of responses between mature lymphogranulocytic cells and immature HPCs. Of particular note is the response of MMP-9^-/- mice to Ptx-induced mobilization. If MMP-9^-/- cells fail to respond to SDF-1 as recently suggested,⁵²  one would not expect them to respond to inhibition of its downstream signaling by Ptx, unless the responses seen here are independent of SDF-1. Moreover, the response of the MMP-9^-/- mice to G-CSF–induced mobilization was in agreement with previous data,⁵³  but in contrast to other data.⁵²  Nevertheless, the synergy in mobilization seen with G-CSF and Ptx in normal mice was not observed in these mice, possibly implicating the presence of MMP-9 in the synergistic effect.

The precise sequence of molecular interactions leading to mobilization after Ptx treatment is unclear and warrants further exploration with innovative reagents and using mice with altered migratory responses. If in vivo Ptx treatment interrupts signals emanating from G protein–coupled receptors on hematopoietic cells, then one must assume that a major relevant pathway involved is the CXCR4/SDF-1 pathway. There is mounting evidence about the significance of this pathway in homing/engraftment as well as in mobilization of HPCs.^54,55  Several experimental approaches, including antibodies, receptor antagonists, or modified chemokine ligands have been used to study its influence and several scenarios have been suggested to explain its effects on hematopoietic homeostasis. Experiments with CXCR4 or SDF-1 knockout mice emphasized the importance of this pathway in the maintenance of adult hematopoiesis within BM.^21-23  Furthermore, treatment of normal mice or human volunteers with small molecule antagonists of CXCR4,^25,26  or modified forms of SDF-1²⁴  leads to mobilization of stem/progenitor cells presumably through the disruption of signals from the CXCR4/SDF-1 pathway. In fact, it has been advocated that such a mechanism is mainly responsible for the G-CSF–induced mobilization.^27,56-58  Following G-CSF treatment, an increased release of leukocyte proteinases, especially neutrophil elastase, cathepsin G, or proteinase-3 from PMNs was documented within BM and in circulation.^56,57  Such increased levels of serine proteases within BM can attack several target proteins, including CXCR4, SDF-1,^56,57  or VCAM-1,⁵⁸  leading to inactivation of CXCR4/SDF1- or VCAM-1/VLA-4–dependent signals and cell migration out of BM. Although additional mechanisms could be operable in G-CSF–induced mobilization, the concomitant degradation of a receptor and of its ligand,⁵⁶  or of other endothelial and extracellular matrix (ECM) molecules could magnify perturbation of several adhesion mechanisms maintaining developing hematopoietic cells within BM and would lead to mobilization of stem/progenitor cells. If one assumes that Ptx disrupts downstream signaling through the CXCR4/SDF-1 pathway, as it occurs in vitro, our data are in concert with the studies mentioned above, implicating a disruption of CXCR4/SDF1 signaling in mobilization. Ptx-mobilized cells were more severely crippled in their SDF-1 response than the G-CSF–mobilized cells, although conflicting data are available for the response of G-CSF–mobilized cells to SDF-1.^33,59,60  Moreover, such an exacerbated effect on abrogation of SDF-1 signaling together with a possible reduction of SDF-1 within BM (Table 2) and/or an increase in MMP-9 on day 3 after Ptx treatment could explain the synergy of Ptx with G-CSF that we observed. Nevertheless, the picture is far from clear, and other contradictory interpretations have also surfaced. It was recently suggested that activation rather than abrogation of CXCR4/SDF-1 signaling is required for mobilization, because antibodies to CXCR4, or small molecule antagonist (TC14012) for this receptor suppressed the G-CSF–induced mobilization.²⁷  These data are in direct contradiction to other studies showing that CXCR4 antagonists promote mobilization and enhance the G-CSF–induced mobilization.^24-26  Whether in vivo treatments with polyclonal antibodies versus different CXCR4 antagonistic molecules with potentially different modes of action could account for these differences is difficult to discern at this point and further experimentation is warranted. On a relevant vein, data with in vitro anti-CXCR4 antibody treatments are inconsistent^61-64  and at odds with studies using Ptx-treated cells in homing experiments of murine BM,⁶⁰  murine fetal liver cells,⁶⁵  or human CD34⁺ cells.⁶²  Collectively, the weight of the evidence suggests that CXCR4/SDF-1 signaling may not be crucial for the initial anchoring of transplanted cells to endothelial cells within BM, but their subsequent retention within BM is greatly dependent on CXCR4/SDF-1 signaling. Whether cells from Ptx-treated animals or hematopoietic cells with genetically modified G_i protein signaling³⁵  display a different behavior in homing assays will be addressed in future experiments.

We are grateful to Drs Daniel C. Link and Robert M. Senior, both of Washington University School of Medicine for providing the G-CSFR^-/- and MMP-9^-/- mice, respectively, and Dr Barrett Rollins, Harvard Medical School for MCP-1^-/- mice. We are also grateful to Dr A. Beaudet (Baylor College, Houston, TX) for the CD18^-/- and EPL^-/- mice. We thank Marta Hill for her expert secretarial assistance, Linda M. Scott for the genotyping of CD18^-/- mice and Vivian Zafiropoulos for skillful technical assistance.