Flt3 ligand and the Flt3 receptor regulate hematopoietic cell migration by modulating the SDF-1α(CXCL12)/CXCR4 axis

Homing and mobilization of hematopoietic stem and progenitor cells (HSPCs) are regulated by hematopoietic cytokines,^1-3  chemokines,^4-6  and adhesion molecules;^7,8  however, their mechanisms are poorly understood. Stromal-derived factor 1α (SDF-1α, CXC ligand 12 [CXCL12]) and its receptor CXCR4 play important roles in regulating migration, homing, and mobilization of hematopoietic cells.^2,6,9-11  CXCL12 attracts primitive hematopoietic cells expressing CXCR4 to bone marrow,^2,11  while disruption of CXCL12/CXCR4 interaction within marrow may facilitate their mobilization to the peripheral circulation.^6,12-14  Several studies demonstrate that CXCL12/CXCR4 function is influenced by hematopoietic cytokines. Stem cell factor (SCF) stimulates migration of hematopoietic progenitor cells and enhances migration induced by CXCL12.^11,15  Thrombopoietin, granulocyte-macrophage colony-stimulating factor (GM-CSF), and SCF synergistically enhance intracellular signals generated through the CXCL12/CXCR4 axis,¹⁶  and treatment of CD34⁺ cells with SCF plus interleukin-6 (IL-6) enhances their in vivo homing by up-regulating CXCR4 expression.²  These findings suggest that hematopoietic growth factors functionally interact with CXCL12/CXCR4 signaling pathways implicated in cell migration.

Flt3 is a type III tyrosine kinase receptor expressed mainly by primitive hematopoietic cells.^17-19  Flt3 ligand (FL) has little direct effect on progenitor cell proliferation but synergizes with other cytokines to enhance proliferation.^20,21  FL induces HSPC mobilization when administered in mice and demonstrates synergy when used in combination with granulocyte colony-stimulating factor (G-CSF) or GM-CSF,^3,22-24  suggesting that it may enhance HSPC motility. In addition, HSPC mobilization by G-CSF is associated with elevated serum FL, suggesting that FL may be involved in G-CSF–mediated mobilization.²⁵ 

Flt3 is overexpressed on hematopoietic cells from the majority of patients with acute leukemia,²⁶  and both Flt3 and FL are expressed in most acute myeloid leukemia (AML) cells, suggesting autocrine activation and signaling.^26-28  Internal tandem duplication (ITD) mutations of the Flt3 receptor gene that activate the Flt3 tyrosine kinase are found in 25% to 30% of patients with AML and are associated with poor prognosis.^27,29,30  Furthermore, their ectopic expression in mice results in myeloproliferative disease with extramedullary hematopoiesis.^31,32  AML cells may also have enhanced migratory potential compared with normal cells, since they are often found in extramedullary sites, which is an adverse prognostic factor.^33,34  The majority of acute leukemic cells express CXCR4 and migrate in response to CXCL12,^9,35,36  suggesting that the CXCL12/CXCR4 axis may be involved in extramedullary disease.

In this report, we investigated whether FL modulates migration of hematopoietic cells and if there are functional associations linking the FL/Flt3 and CXCL12/CXCR4 signaling pathways in migration. Our findings provide an in vitro model for FL-mediated migration of normal and transformed hematopoietic cells and suggest that the FL/Flt3 signaling pathway may regulate hematopoietic cell trafficking through positive and negative effects on the CXCL12/CXCR4 axis.

Phycoerythrin (PE)–conjugated antihuman Flt3 and mouse immunoglobulin G₁ (IgG₁) were purchased from Immunotech (Westbrook, ME). Fluorescein isothiocyanate (FITC)– or cytochrome (Cy)–conjugated anti-CD34 (BIRMA-K3) and mouse IgG₁ were from Dako (Carpinteria, CA). PE-Cy5 antihuman CXCR4 (Clone 12G5) was from eBioscience (San Diego, CA). Biotinylated antimouse CXCR4 (Clone 2B11/CXCR4), biotinylated rat IgG_2b, anti-FcγIII/II, FITC antihuman Ki-67 (clone MOPC-21), FITC mouse IgG₁ and PE-Cy5 mouse IgG_2a were obtained from BD Biosciences (San Diego, CA). Recombinant human (rh) FL was provided by Immunex (Seattle, WA). Recombinant murine (rm) and rhCXCL12 (SDF-1α) and rmIL-3 were purchased from R&D Systems (Minneapolis, MN). Recombinant murine stem cell factor (rmSCF) was a gift from Dr Karl Nocka (UCB Research, Cambridge, MA). Recombinant murine GM-CSF was purchased from BioVision (Palo Alto, CA). The selective pathway inhibitors PD98059 (mitogen-activated protein kinase p42/p44 [MAPK^p42/p44]), LY294002 (phosphatidylinositol 3–kinase [PI3-kinase]), H89 (protein kinase A [PKA]), and AG1296 (tyrosine kinase) were purchased from BioMol Research Laboratories (Plymouth Meeting, PA).

Full-length wild-type human Flt3 cDNA was a gift from Immunex.¹⁸  The empty pLXSN or pLXSN harboring full-length Flt3 was transfected into GP+envAm12 cells, and the virus supernatant was added to GP+E-86 cells.³⁷  IL-3–dependent mouse Ba/F3 cells were incubated with fresh virus supernatant, and G418-resistant Flt3^positive cells (Ba/F3-Flt3) were sorted by fluorescence-activated cell sorter (FACS). G418-resistant vector control–transduced Ba/F3 and Ba/F3-Flt3 cells were maintained in RPMI plus 10% heat-inactivated fetal bovine serum (HI-FBS; Hyclone Sterile Systems, Logan, UT) supplemented with 0.1 ng/mL rmIL-3. Ba/F3 cells expressing internal tandem duplication (ITD) Flt3 mutants cloned from patients with AML (W51, W73, and W78) as well as wild-type Flt3 in MSCV vector were kindly provided by Dr D. Gary Gilliland, Harvard Medical School.³²  Mouse 32D cells expressing ectopic wild-type Flt3 or ITD-Flt3 were generous gifts from Drs Joachim Schwäble and Hubert Serve, University of Münster.³¹  Human acute leukemia cell line RS4;11 was maintained in RPMI plus 10% HI-FBS.³⁸ 

Normal umbilical cord blood (UCB) was obtained with institutional review board approval. Isolation of CD34⁺ cells^39,40  and transwell migration assays were performed as described.¹¹  CD34⁺ cells (100 000, purity > 95%) were loaded onto the top chamber of transwells (5.0-μm pore size; Corning, Corning, NY) with rhFL and/or rhCXCL12 in the bottom chamber and incubated for 4 to 72 hours at 37°C in 0.5% bovine serum albumin (BSA)/RPMI. Cells completely migrating to the bottom chamber were enumerated by FACS. In some experiments, partially purified CD34⁺ cells were stained with anti-CXCR4, anti-Flt3, and anti-CD34 antibodies, and Flt3⁺-CXCR4⁺-CD34⁺ cells were isolated by FACS and used in migration assays. The chemokinetic or chemotactic activity of FL on Ba/F3-Flt3 cells was determined by the checkerboard assay.¹¹  The percentage cell migration was calculated by dividing the number of cells migrated to the lower well by the total cell input multiplied by 100.

Actin polymerization was analyzed as described,¹¹  with minor modifications. Ba/F3-Flt3 cells were resuspended at 1.25 × 10⁶/mL in RPMI-1640/0.5% BSA and 0.2-mL aliquots were stimulated with phosphate-buffered saline (PBS), FL, or CXCL12 for up to 5 minutes. Cells were fixed and stained with FITC-labeled phalloidin (4 × 10^–7 M; Sigma Aldrich, St Louis, MO) in 0.5 mg/mL 1-α-lysophosphatidylcholine, 18% formaldehyde in PBS. Cells were incubated for 15 minutes at 37°C, washed, and resuspended in 1% paraformaldehyde. Mean channel fluorescence of phalloidin was analyzed by FACSCalibur (BD Biosciences).

Cell surface expression of Flt3 or CXCR4 on Ba/F3-Flt3 or CD34⁺ cells was analyzed by flow cytometry. Cells were washed with 0.1% BSA/PBS, blocked with 10 μg/mL human IgG or anti-FcγIII/II for 10 minutes, and stained with isotype control or antibodies against CXCR4 or Flt3 for 30 minutes in 0.1% BSA/PBS on ice. Expression of cell surface CXCR4 was compared by normalizing the mean fluorescence intensity (MFI) of CXCR4 with the isotype control. In some experiments, percent change of CXCR4 was determined using the following formula: % change of CXCR4 = 100 × [1 – (MFI of CXCR4 for cells cultured with FL ÷ MFI of isotype for cells cultured with FL) ÷ (MFI of CXCR4 for cells cultured without FL ÷ MFI of isotype for cells cultured without FL)].

Cells (5 million) were lysed in 100 μL sodium dodecyl sulfate (SDS) lysis buffer (20 mM dithiothreitol, 6% SDS, 0.25M Tris [tris(hydroxymethyl)aminomethane, pH 6.8], 10% glycerol, and bromophenol blue), sonicated, and boiled, and the lysates from 1 × 10⁶ cells were subjected to Western analysis. The blot was blocked with 2% nonfat dry milk in PBST (0.05% Tween 20 in PBS) at room temperature for one hour and probed with total or phospho-specific antibodies against MAPK^p42/p44 (Thr²⁰²/Tyr²⁰⁴), Akt (Ser⁴⁷³), and cyclic adenosine monophosphate response element binding protein, Ser¹³³ (CREB) (Cell Signaling Technology, Beverly, MA) in blocking buffer (1:1000 dilution) overnight at 4°C. The filters were washed 5 times with PBST and incubated with horseradish peroxidase (HRP)–conjugated anti–rabbit or –mouse IgG (Amersham Biosciences, Piscataway, NJ) in blocking buffer (1:2500 dilution) at room temperature for one hour. The blots were washed 5 times with PBST and developed with enhanced chemiluminescence (ECL) reagent (Amersham Biosciences).

Data are expressed as mean ± standard error (SEM). Statistical differences between single agents and control were determined using the 2-tailed Student t test in Microsoft Excel (Microsoft, Seattle, WA). Synergy in response to FL plus CXCL12 was compared with FL and CXCL12 used alone by analysis of variance (ANOVA), and the Bonferroni method was used for multiple comparisons. This analysis was performed by the Biostatistics Core Facility of the Indiana University Cancer Center.

Since several hematopoietic cytokines, notably SCF and IL-3, which, like FL, support proliferation and/or survival of early hematopoietic cells, have been shown to enhance chemotaxis,^11,15,41  we examined whether FL affects hematopoietic cell migration/motility. In an in vitro transwell migration assay, both positive (FL in bottom chamber only: 0/+FL) and zero gradients (FL in upper and bottom chamber: +FL/+FL) of rhFL induced migration of UCB CD34⁺ cells over a 48-hour period (Figure 1A). Despite increased migration to FL at 48 hours (28% ± 2% and 25% ± 2% migration in positive and zero gradients, respectively), FL did not increase total CD34⁺ cells during this time frame (0.5% ± 0.1% increase in viable cells), suggesting that enhanced migration to FL was not due to an effect on cell proliferation. Similar to CD34⁺ cells, rhFL induced modest migration of Ba/F3-Flt3 cells both in a positive (0/+) and a zero (+/+) gradient after 4 hours (Figure 1B). Maximum response to FL was observed at 10 ng/mL using a positive gradient and declined with increasing dose. The response was specific to the FL/Flt3 pathway, since no migration was observed in Ba/F3 cells, which do not express Flt3, transduced with empty vector (Figure 1C). Migration of Ba/F3-Flt3 cells to FL was associated with significantly increased actin polymerization (Figure 1D). The maximum migration of Ba/F3-Flt3 cells in a positive (2.2%) and in a negative (1.9%) gradient (Table 1) suggests that FL primarily induces chemokinetic activity.

Since the CXCL12/CXCR4 axis regulates migration and homing of HSPCs,^2,11  we examined whether FL could enhance migration of CD34⁺ cells induced by CXCL12. Analysis of CXCR4 and Flt3 expression on freshly isolated UCB CD34⁺ cells indicated that 36% ± 2% (n = 3) of cells coexpress Flt3 and CXCR4. Transmigration of CD34⁺ cells toward a positive gradient of 50 or 100 ng/mL CXCL12 for 4 hours was significantly enhanced in the presence of 10 ng/mL FL (Figure 2A). For 50 and 100 ng/mL CXCL12, the presence of 10 ng/mL FL produced significant synergistic migration at both dosage levels (P < .0001). The combination of 10 ng/mL FL plus 50 ng/mL or 100 ng/mL CXCL12 resulted in a 2.6 ± 0.1-fold and 1.7 ± 0.2-fold increase, respectively, compared with CXCL12 alone, although 100 ng/mL CXCL12 resulted in a higher absolute percentage migration than 50 ng/mL CXCL12. FL had no significant effect on CXCR4 expression (1.02 ± 0.2-fold) during this time period; therefore the enhancing effect of FL on CXCL12-mediated migration cannot be explained by alteration in CXCR4 expression. The dramatic synergistic effect of FL on CXCL12-mediated migration was also observed using isolated CXCR4⁺-Flt3⁺-CD34⁺ cells (P < .0001; Figure 2B), likely due to enrichment of double-positive cells that can respond to both FL and CXCL12. However, migration of sorted cells to CXCL12 was lower than unsorted cells (Figure 2A-B), and the enhancing effect of FL on CXCL12-mediated migration of unsorted CD34⁺ cells was more obvious at a lower dose (50 ng/mL) than a higher dose (100 ng/mL) of CXCL12, suggesting that enhanced migration may also be a consequence of inhibition of CXCR4 by the 12G5 anti-CXCR4 antibody used for cell sorting, which is known to block CXCL12 binding to CXCR4.

Similar synergistic migration to CXCL12 was observed for Ki-67^negative CD34⁺ cells (P < .0001), that is, quiescent cells,^42,43  and for the Ki-67^positive population (P < .0001; Figure 2C), although migration to FL and/or CXCL12 was significantly higher in Ki-67^negative cells (P < .05). FL also synergistically enhanced CXCL12-induced migration of Ba/F3-Flt3 cells (Figure 2D) after 4 hours of incubation (P < .0001). Although the effect was weak, migration of Ba/F3-Flt3 cells toward CXCL12 in the face of a negative gradient of FL (FL in top chamber and CXCL12 in bottom chamber) was also synergistically increased (P = .01; Figure 2E). Similarly, FL synergistically enhanced migration of human RS4;11 acute leukemia cells, which express wild-type Flt3 and CXCR4 (not shown) to CXCL12 (P < .0001; Figure 2F).

Since the stimulatory effect of FL on CD34⁺ cell migration was time dependent (Figure 1A), we compared the long-term versus short-term effects of FL on CXCL12-mediated migration. Migration of CD34⁺ cells to a positive gradient of FL and/or CXCL12 was evaluated for up to 72 hours. Time-dependent increased migration was observed in all groups (Figure 3A) but did not result from an increase in cell proliferation (not shown). At 48 and 72 hours, migration to FL alone became greater than migration to CXCL12 alone and was higher than migration induced by the combination of FL plus CXCL12 at 4 hours. Migration to the combination of FL plus CXCL12 was the highest at all time points. The effect of FL plus CXCL12 was additive beyond 24 hours. We also compared migration of CD34⁺ cells induced by a zero gradient of FL plus CXCL12 (FL+CXCL12 in upper and lower chambers) with a positive gradient of FL plus CXCL12 (FL+CXCL12 in lower chamber only) (Figure 3B). The zero gradient of FL plus CXCL12 had no effect on migration at 4 hours but significantly induced migration after 12 hours, although the response was lower than the positive gradient of FL plus CXCL12, suggesting that FL plus CXCL12 can induce random cell migration over time.

Our results demonstrating synergistically enhanced migration to CXCL12 by FL suggested the possibility of cross talk between FL/Flt3 and CXCL12/CXCR4. Ligand binding to Flt3²⁷  or chemokine receptors activates MAPK^p42/p44 and PI3-kinase.^16,44  In order to investigate pathways involved in synergistic migration to FL plus CXCL12, we examined phosphorylation of MAPK^p42/p44, Akt, CREB (cAMP response element binding protein), a downstream target of PKA, and its family member activating transcriptional factor-1 (ATF).⁴⁵  In Ba/F3-Flt3 cells, FL induced barely detectable MAPK^p42/p44, Akt, and CREB phosphorylation, while CXCL12 stimulation resulted in significant phosphorylation (Figure 4A). The combination of FL plus CXCL12 synergistically enhanced phosphorylation of MAPK^p42/p44, Akt, and CREB (Figure 4A), consistent with synergistically enhanced migration. To confirm the involvement of these pathways in migration toward FL plus CXCL12, we evaluated migration of Ba/F3-Flt3 cells pretreated with PD98059, LY294002, and H89, selective inhibitors for MAPK^p42/p44, PI3-kinase, and PKA, respectively (Figure 4B). Although synergistic migration was not completely blocked and remained statistically significant, pretreatment with PD98059, LY294002, or H89 significantly inhibited migration induced by the combination of FL plus CXCL12 (P < .05). The inhibitors had little effect on migration to CXCL12 alone but reduced migration in response to FL alone (Figure 4B; Table 2), suggesting that suppression of synergistic migration was a consequence of inhibition of the FL/Flt3 signaling pathway rather than the CXCL12/CXCR4 pathway. Consistent with this finding, pretreatment of cells with AG1296, which inhibits Flt3 tyrosine kinase,^46,47  blocked synergistic migration to FL plus CXCL12 (73% ± 10% reduction in synergistic response) (Figure 4C).

Treatment of cells with the combinations of PD98059 plus LY294002, PD98059 plus H89, or LY294002 plus H89 resulted in a greater inhibitory effect on migration to CXCL12 or FL plus CXCL12 compared with each inhibitor used alone (Table 2). The inhibitory effects of PD98059 plus LY294002 or PD98059 plus H89 on migration to FL plus CXCL12 were the same as CXCL12 alone, whereas pretreatment with LY294002 plus H89 resulted in greater inhibition of migration to FL plus CXCL12 than migration to CXCL12 alone. Inhibition of migration was not due to compound toxicity since forward and side scatter analysis demonstrated that more than 95% of cells remained viable following treatment with inhibitors and during the migration assay.

Disruption of CXCL12/CXCR4 interaction^13,14  or multiday administration of FL to mice induces HSPC mobilization,^3,22,23  and elevated serum FL is associated with mobilization induced by a multiday regimen of G-CSF.²⁵  This finding suggests that prolonged exposure of CD34⁺ cells to FL might antagonize CXCL12/CXCR4 function and facilitate mobilization. Incubation of CD34⁺ cells with FL for 24 hours or Ba/F3-Flt3 cells for 8 hours had no effect on CXCR4 expression; however, culture of CD34⁺ cells with FL for 48 hours or Ba/F3-Flt3 cells for 24 hours significantly decreased CXCR4 expression (Figure 5A,C), coincident with significantly reduced migration toward CXCL12 (Figure 5B,D) compared with cells incubated without FL. CD34⁺ cell CXCR4 expression and migration to CXCL12 were reduced by 31% ± 5% (P < .001, n = 6) and 23% ± 6% (P < .05, n = 3), respectively, while Ba/F3-Flt3 cell CXCR4 expression and migration to CXCL12 were reduced by 52% ± 8% (P < .005, n = 3) and 64% ± 7% (P < .005, n = 3), respectively. FL had no effect on CXCR4 expression (Figure 5C) and migration (not shown) in vector-transduced Ba/F3 cells that do not express Flt3. Moreover, pretreatment of Ba/F3-Flt3 cells with FL for 24 hours inhibited phosphorylation of MAPK^p42/p44, Akt, CREB, and ATF-1 induced by CXCL12 (Figure 5E). In control cells, MAPK^p42/p44 phosphorylation peaked between 5 and 10 minutes after CXCL12 incubation and remained elevated but started to decline by 15 minutes. In contrast, FL pretreatment resulted in significantly weaker and delayed phosphorylation. CXCL12-induced Akt phosphorylation also peaked at 5 minutes and declined slightly thereafter in control cells, whereas in cells pretreated with FL, Akt phosphorylation reached maximum at 10 minutes and diminished rapidly at 15 minutes. Maximum CREB phosphorylation in response to CXCL12 was observed at 10 minutes and was sustained until 15 minutes in control cells but was weaker and reduced at 15 minutes in FL-pretreated cells. Similarly, ATF-1 phosphorylation was weaker in FL-treated cells than cells without FL treatment at 10 and 15 minutes (Figure 5E). These results suggest that with time, FL can downmodulate CXCR4 expression, inhibit CXCL12/CXCR4 signaling, and negatively regulate migration to CXCL12.

Flt3 is constitutively activated by autocrine mechanisms or ITD-Flt3 mutations in the majority of AML cells.^26,28,29  Since AML cells can migrate to CXCL12,³⁵  we investigated whether ITD-Flt3 affects CXCL12-induced migration. Forced expression of the ITD-Flt3 mutants W51, W73, and W78 in Ba/F3 cells resulted in respective 18 ± 1-, 17 ± 2-, and 31 ± 4-fold increases in spontaneous migration compared with cells expressing wild-type Flt3 (Figure 6A; 0/0) and were associated with increased baseline actin polymerization (46% ± 3%, 33% ± 3%, and 41% ± 1% increases in W51, W73, and W78, respectively, P < .001, compared with wild-type Flt3) (Figure 6B). In addition, migration of ITD-Flt3 Ba/F3 cells to CXCL12 was significantly increased compared with cells expressing wild-type Flt3 (0/CXCL12; Figure 6C). Identical results were observed in myeloid 32D cells expressing ectopic ITD-Flt3 compared with cells expressing wild-type Flt3 (not shown). No further enhancement in migration to FL alone (not shown) or FL plus CXCL12 (0/FL+CXCL12; Figure 6C) was observed in Ba/F3 cells expressing ITD-Flt3.

In a negative gradient of CXCL12 (CXCL12/0; Figure 6A), spontaneous migration of BaF3 cells expressing either wild-type or ITD-Flt3 was repressed; however, migration of ITD-Flt3 cells was significantly higher than cells expressing wild-type Flt3, indicating that ITD-Flt3 facilitates migration even in a negative gradient of CXCL12. Despite enhanced migration to CXCL12, cell surface CXCR4 was down-regulated by 53% to 69% in all ITD-Flt3–expressing Ba/F3 cells (Figure 6D) and by 39% ± 2% in 32D cells expressing ITD-Flt3 compared with cells expressing wild-type Flt3 (P < .001).

It has been suggested that hematopoietic cells that migrate in vitro represent a quiescent cell population.^48,49  We evaluated the cell cycle distribution of migrated Ba/F3 cells expressing wild-type or ITD-Flt3 (Figure 7) and found that less than 10% of spontaneously migrating Ba/F3 cells expressing wild-type Flt3 were in S+G₂/M phase, while more than 36% to 44% of spontaneously migrating Ba/F3 cells expressing ITD-Flt3 were in S+G₂/M phase. Similarly, a significantly higher proportion of Ba/F3 cells expressing ITD-Flt3 and migrating to CXCL12 were in S+G₂/M phase compared with cells expressing wild-type Flt3. The proportion of Ba/F3 cells in S+G₂/M phase was comparable (50%-60%) between Ba/F3 cells expressing wild-type or ITD-Flt3. Incubation of Ba/F3 cells expressing wild-type or ITD-Flt3 with CXCL12 for 4 hours had no effect on cell cycle distribution (not shown), indicating that the larger S+G₂/M fraction of migrating cells expressing ITD-Flt3 did not result from a larger number of S+G₂/M input cells.

Flt3 ligand and its receptor Flt3 have been implicated in survival, proliferation, and adhesion of HSPCs.^27,50,51  We now provide compelling evidence that the FL/Flt3 axis also regulates migration of normal and transformed hematopoietic cells and that this effect is mediated through the CXCL12/CXCR4 axis. The combination of FL and CXCL12 synergistically enhances migration of CD34⁺ cells, Ba/F3-Flt3, and human acute leukemia RS4;11 cells. Furthermore, ITD-Flt3 derived from AML patients significantly enhanced spontaneous and CXCL12-induced migration when overexpressed in Ba/F3 and 32D cells. In contrast to enhanced migration in short-term assays, prolonged exposure to FL can down-modulate CXCR4 expression and impair migration of CD34⁺ cells and Ba/F3-Flt3 cells toward CXCL12. These findings indicate that FL can provide both positive and negative effects on CXCL12/CXCR4-mediated migration and suggest that the FL/Flt3 axis participates in the trafficking of normal and transformed hematopoietic cells.

The facts that FL had an effect only on Ba/F3 cells expressing ectopic Flt3 but not control vector–transduced Ba/F3 cells, which do not express Flt3; that the effect of FL on CD34⁺ cell migration increased with time; and that actin polymerization induced by FL directly paralleled cell migration all support a direct effect of FL on cell migration. The fact that positive or zero gradients of FL resulted in similar effects suggests that the predominant effect of FL is to enhance random cell migration or cell motility (ie, chemokinesis).

Although signaling pathways involved in hematopoietic cell migration are not completely understood, we found that synergistic migration to FL plus CXCL12 was associated with synergistic phosphorylation of MAPK^p42/p44, Akt, and CREB, implicating these pathways in hematopoietic cell migration. The effects of inhibitors for MAPK^p42/p44, PI3-kinase, or PKA were not selective for synergistic migration and did not completely block synergistic migration. Rather, the inhibitory effects of PD98059, LY294002, or H89 on migration by FL plus CXCL12 were almost identical with reduction in migration to FL alone (Table 2), suggesting that the inhibition of synergistic migration is likely due to antagonizing Flt3 signaling pathways. This is supported by the reduction in synergistic migration by AG1296, which blocks Flt3 activity.^46,47  However, the failure of inhibitors for MAPK^p42/p44, PI3-kinase, or PKA pathways to inhibit migration to CXCL12 under the conditions tested does not rule out involvement of these pathways in CXCL12-mediated migration. A role for PKA in CD14⁺ peripheral blood mononuclear cell migration or cytokine-mediated cancer cell migration has been demonstrated,^52,53  and PI3-kinase has been implicated in CXCL12-mediated migration.^54-56  A role for MAPK^p42/p44 in CXCL12-mediated migration appears to be cell-type specific.^54,57-59  Phosphorylation of MAPK^p42/p44, Akt, and CREB by 100 ng/mL CXCL12 was consistently more than 10 ng/mL FL (Figure 4A), suggesting that the inhibitors may be less effective on CXCL12/CXCR4 than FL/Flt3 at the doses and time frame tested. Indeed, migration to CXCL12 was dramatically inhibited by combinations of inhibitors, compared with any of them used alone, coincident with further reduction in migration to FL plus CXCL12 (Table 2). This suggests that MAPK^p42/p44, Akt, or PKA are involved in migration to CXCL12 and that inhibition of these pathways downstream of CXCL12/CXCR4 may also account for reduction of migration to FL plus CXCL12. Overall, these data suggest involvement of MAPK^p42/p44, PI3-kinase, or PKA pathways in migration to FL plus CXCL12; however, our data also suggest the presence of pathways other than MAPK^p42/p44, PI3-kinase/Akt, or PKA/CREB that regulate synergistic migration, since even the combination of all 3 inhibitors failed to completely inhibit migration to FL plus CXCL12 (not shown).

Our results suggest that positive and negative cross talk between the FL/Flt3 and CXCL12/CXCR4 pathways may differentially regulate HSPC homing and mobilization. The synergistic migration to FL plus CXCL12 observed in primary hematopoietic cells suggests that FL and CXCL12 may cooperatively enhance homing of HSPCs to the bone marrow micronenvironment, which is mediated by short-term activation of key intracellular signaling cascades. In contrast, multiday administration of FL can induce HSPC mobilization in mice.^3,22-24  It has been suggested that marrow stroma–derived CXCL12 forms a positive gradient that maintains HSPCs within the marrow microenvironment,^1,11  and disruption of CXCL12/CXCR4 interactions has been implicated in stem cell mobilization.^6,12-14  Our findings demonstrate that prolonged incubation of CD34⁺ cells or Ba/F3-Flt3 cells with FL down-modulates migration to CXCL12, blocks CXCR4 signaling, and reduces cell surface CXCR4 expression, and therefore may be one mechanism that facilitates HSPC mobilization in response to FL. Consistent with the negative effect of FL on in vitro migration to CXCL12, preliminary studies indicate that preincubation of mouse bone marrow cells with FL significantly reduces homing of colony-forming unit–granulocyte-macrophage (CFU-GM) to recipient bone marrow in vivo (S.F. and L.M.P., unpublished observation, July 2004), suggesting that long-term exposure to FL can negatively regulate in vivo homing of progenitor cells.

FL increases CD34⁺ cell migration in a time-dependent fashion in vitro, suggesting that in vivo, the effects of FL on CXCL12-induced migration may also be time dependent. When short-term versus long-term migration of CD34⁺ cells was compared, the effect of FL alone after 48 hours became greater than the effect of FL plus CXCL12 at 4 hours. However, in contrast to FL, which induced random migration over the long term, FL plus CXCL12 induced directed migration to a positive gradient in short-term (4 hours) assays; although it also stimulated moderate random migration after 12 hours, indicating that the nature of long-term migration by FL and the early effect of FL plus CXCL12 are different. Synergistic migration by FL plus CXCL12 was observed within only 24 hours, suggesting that FL and CXCL12 synergy is an early effect. This may be due to desensitization of the CXCR4 receptor and/or disruption of a CXCL12 gradient. Although our data are limited, it would suggest that an early effect of FL in vivo may be to favor recruitment of cells to bone marrow through synergistic activation of CXCL12/CXCR4 pathways, whereas in the long term, FL or FL plus CXCL12 would increase random cell migration, and eventually, antagonize CXCL12/CXCR4 function, which would contribute to cell peripheralization.

The majority of AML cells or cell lines express Flt3 as well as CXCR4^{9,26-28,35,36}  and migrate in response to CXCL12. Our data indicate that RS4;11 acute leukemic cells, which coexpress wild-type Flt3 and CXCR4, demonstrated synergistic migration to FL plus CXCL12. Our results demonstrating that expression of ITD-Flt3 significantly accelerates migration of Ba/F3 and 32D cells to CXCL12 support a positive role of Flt3 in regulating migration to CXCL12, and are in good agreement with superior engraftment of AML cells expressing ITD-Flt3 in nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice compared with cells expressing wild-type Flt3.⁶⁰  This suggests that activation of Flt3 signaling either by ITD-Flt3 or autocrine mechanisms may enhance migration/homing of AML cells to secondary organs that express CXCL12, a notion supported by recent reports demonstrating a positive role of the CXCL12/CXCR4 axis in metastasis of breast and prostate cancers.^61-63  Ba/F3 cells expressing ITD-Flt3 demonstrate significantly elevated migration even in the presence of a negative gradient of CXCL12, providing further evidence that Flt3 activation enhances cytokinesis and motility. Enhanced cell motility/migration resulting from activation of Flt3 signaling may allow transformed hematopoietic cells to migrate out of bone marrow despite existence of a positive marrow CXCL12 gradient. In addition, ITD-Flt3 facilitates migration of cells in S+G₂/M phase, which may accelerate homing or peripheralization of actively dividing cells.

The percentage of Ba/F3 cells harboring ITD-Flt3 and migrating to CXCL12 was significantly higher than wild-type Flt3–expressing Ba/F3 cells migrating to FL plus CXCL12, suggesting that signals generated by ITD-Flt3 are not identical to those induced by FL and wild-type Flt3. ITD-Flt3–expressing Ba/F3 cells do not directly migrate to FL and show no enhanced migration to FL plus CXCL12, likely because autoactivation of ITD-Flt3 already provides a maximal response. Despite enhanced migration to CXCL12, CXCR4 expression on ITD-Flt3–expressing cells was down-regulated compared with cells expressing wild-type Flt3. This is consistent with down-regulation of CXCR4 in CD34⁺ cells and Ba/F3-Flt3 cells following prolonged culture with FL and suggests that activation of intracellular signaling pathways is likely responsible for enhanced migration. On the other hand, a recent report suggests a positive correlation between ITD-Flt3 expression and higher CXCR4 expression in AML cells, although it lacks direct evidence that higher CXCR4 levels are a consequence of ITD-Flt3.⁶⁴ 

In summary, our data provide evidence that CXCL12-mediated hematopoietic cell migration can be modulated in positive and negative fashion by FL/Flt3, which may be important for trafficking of normal or malignant hematopoietic cells. Manipulation of their interaction could be clinically beneficial for hematopoietic cell transplantation and for treatment of hematopoietic malignancies in which both functional Flt3 and CXCR4 are expressed. Activation of FL/Flt3 and CXCL12/CXCR4 signaling pathways may enhance homing of normal HSPCs to the bone marrow microenvironment. In this regard, enhanced migration of quiescent Ki-67^negative CD34⁺ cells^42,43  to CXCL12 by FL suggest that FL may facilitate homing of CD34⁺ cells with long-term in vivo engraftment ability.^65,66  Stimulating the FL/Flt3 axis could potentially facilitate collection of HSPCs in patients and donors undergoing peripheral blood stem cell mobilization for transplantation by attenuating CXCL12/CXCR4 function while blocking interaction of the FL/Flt3 and CXCL12/CXCR4 signaling pathways by Flt3 inhibitors currently used in clinical trails,^67,68  and AMD3100, which can block cell migration to CXCL12,^69,70  may help to reduce extramedullary leukemia infiltration in patients with AML.

The authors thank Chang H. Kim, PhD, Department of Pathobiology, Purdue University for critical reading of the manuscript; Huimin Bian for technical assistance; Susan Rice and Denessa Luckett for cell sorting; and Constantin T. Yiannoutsos, PhD, and Qianqian Zhao, Biostatistics Core Facility, Indiana University Cancer Center for statistical analysis.