Integrin inhibition through Lyn-dependent cross talk from CXCR4 chemokine receptors in normal human CD34⁺ marrow cells

The interaction of stem/progenitor cells with stromal marrow cells is critical for normal hematopoiesis. Stromal cells regulate hematopoiesis by binding directly to hematopoietic precursors and providing numerous secreted factors. Stromal-derived factor (SDF-1) and its G protein–coupled chemokine receptor, CXCR4, are required for hematopoietic cell migration, adhesion, and bone marrow retention.^1-4  Mice lacking the chemokine SDF-1 or its receptor CXCR4 are unable to carry out normal hematopoiesis.^2,3  SDF-1–induced signaling pathways are defective in BCR-ABL–positive human leukemia cells and these disruptions contribute to the migration, adhesion, and retention of abnormalities characteristic of leukemia.^4,5  Intracellular signaling pathways that mediate CXCR4-dependent adhesion of hematopoietic precursors to stromal cells are poorly understood.

The movement and adhesion of hematopoietic cells is regulated at several levels. One level is the regulation of cell adhesion through integrins.⁶  Integrins can be regulated by intracellular signaling mechanisms through a process called inside-out signal transduction.⁷  This type of regulation provides different adherence responses to various intracellular stimuli, which is important in coordinating such cellular events as migration and adhesion. Chemokine receptors, including CXCR4, generate intracellular signals that supposedly lead to regulation of integrin-mediated cell adhesion. Studies with different proadhesive chemokines indicate that they increase phosphatidylinositol-3 kinase (PI3-K) activity to trigger rapid β₂-integrin–dependent lymphocyte arrest to vascular endothelium, suggesting that PI3-K kinase is a positive regulator of integrin function in lymphocytes.⁸  In platelets, studies with compounds that induce aggregation indicate that heterotrimeric G-proteins, phospholipid metabolism, and tyrosine kinases are implicated in the activation of the integrin αIIbβ₃.⁹  Recently, it was shown that SHIP1 and Lyn tyrosine kinase can negatively regulate αIIbβ₃ integrin function in platelets¹⁰  and neutrophil integrin signaling in mice during stimulation with TNF-α.¹¹  The β₂ integrins, leukocyte function antigen-1 (LFA-1, CD11a) and macrophage antigen-1 (Mac-1, CD11b), have been reported to play an important role in the attachment of CD34⁺ hematopoietic stem/progenitor cells to bone marrow stromal cells through their ligand, intracellular adhesion molecule-1 (ICAM-1).^12,13  While the process of integrin activation has been extensively studied, no specific inside-out signaling pathway, or proteins directly involved in the inhibition of integrin function by CXCR4, has been identified. We previously showed that the migratory response of marrow progenitor cells to SDF-1 is impaired in the absence of Lyn.⁴  We now report that Lyn is the mediator that relays suppressing signals from the chemokine receptor CXCR4 to β₂ integrins in hematopoietic precursors. Our present results provide new insights into the involvement of Lyn and the chemokine receptor CXCR4 in hematopoietic cell movement, arrest, and mobilization within the stromal marrow microenvironment.

Human ICAM-1 (CD54), VCAM-1 (CD106), and integrin β₁, β₂, αL, αM antibodies were purchased from R&D Systems (Minneapolis, MN) and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. Human CD11b (integrin Mac-1)–activation epitope antibodies were purchased from eBioscience. Monoclonal antibody (mAb) 24 for the active conformation of CD11/CD18 was kindly provided by Dr Nancy Hoggs. Human Lyn and Hck antibodies were purchased from Upstate Biotechnology (Lake Placid, NY). We used 4 short interfering RNAs targeting human Lyn, and 3′ UTR (for the functional rescue experiment) described in detail in Ptasznik et al¹⁴  and 4 nonspecific siRNAs combined into one pool, as a negative control. All small siRNA duplexes (19 bp double-stranded RNAs with additional 3′-UU overhangs on both sense and antisense strands) were designed and synthetized by Dharmacon (Lafayette, CO; catalog nos. M-003 153-00-05, M-003 153-00-50, and D-001 206-13-05). Lyn siRNA sequences are described in detail in the online Supplemental Methods of Ptasznik et al.¹⁴  Human Lyn complementary DNA (cDNA) in the pSVL vector (Pharmacia Biotech, Piscataway, NJ) for functional rescue experiment was kindly provided by Dr Diana Linnekin (NCI, Friederick, MD).

Normal human CD34⁺ cells were obtained from healthy consenting donors using MACS cell isolation kits (Miltenyi Biotec, Bergisch Gladbach, Germany), using guidelines approved by the institutional review board of the University of Pennsylvania. Primary cells and cell lines were maintained in Isocov modified Dulbecco media (IMDM) or RPMI 1640 supplemented with 10% to 20% fetal calf serum (FCS), 100 μg/mL streptomycin, 100 U/mL penicillin, and 2 mM GlutaMAX (Invitrogen/Life Technologies, Carlsbad, CA). For fresh primary human CD34⁺ cells, and HL-60 and Nalm-6 cell lines, no stimulation was required prior to nucleoporation. Mo7e cells were supplemented before and after nucleofection with granulocyte macrophage–colony-stimulating factor (GM-CSF; 10 ng/mL). Preparation of all types of cells before nucleoporation and nucleoporation with siRNAs were performed according to the manufacturer's recommendations (for detailed protocols please see General Protocol for Nucleofection of Suspension Cells; catalog no.VPA-1003 and DLA-1002, respectively (AMAXA Biosystems). Functional rescue experiment: 4 individual SMART selection-designed siRNAs silencing endogenous Lyn through the 3′ UTR (Dharmacon, Lafayette, CO) were introduced into Mo7e cells by nucleoporation. These cells were then cotransfected with a Lyn expression construct that was lacking the 3′ UTR.

We measured Lyn, Hck, and β-actin protein expression and phosphorylation in Lyn siRNA–treated versus control siRNA–treated cells. Cell lysates, immunoprecipitations, Western blots, and in vitro kinase assays were performed as previously described.^15,16 

For chemotaxis and cell adhesion assays, cells were stained with 2 μM calcein am (Molecular Probes, Eugene, OR) and measurements were performed using FluoroCount (Packard BioScience, Meriden, CT). For chemotaxis assays, 1 × 10⁵ cells in IMDM (CD34⁺, Mo7e) or RPMI-1640 (HL-60, Nalm-6) with 2% FBS were placed in the upper compartment of a Transwell (Corning, Corning, NY; catalog no. 3421, 6.5 mm diameter, 5.0 μm pore size). SDF-1α (Upstate Biotechnology) was added to the lower compartment at a final concentration of 100 ng/mL. The cells were incubated for 3 hours at 37°C. Trypsinization was used to detach the cells from the membrane. The chemotactic index, a measure of the specificity of migration, was determined as follows: (number of cells migrating to SDF-1 chemokine) / (number of cells that migrated to medium alone). In certain experiments the spontaneous movement was also tested in the absence of SDF-1 over surfaces coated with ICAM-1. For cell adhesion studies, 5 × 10⁴ cells/well were added to recombinant human ICAM-1/Fc– or VCAM-1–coated (R&D Systems) plates (10 μg/mL with 100 μL/well). Human bone marrow stromal cells from healthy donors were put in 96-well plates 24 hours before the assays. The cells (CD34⁺ primary cells or cell lines) were cultured for 1 hour at 37°C with SDF-1α (100 ng/mL). After incubation, each well was washed with the exact same procedure. Antihuman ICAM-1 antibody (12.5 μg/mL; clone BBIG-I1, R&D Systems) was used for adhesion blockade against rhICAM-1/Fc. For assessing proliferation and viability in some control experiments, measurements were carried out by trypan blue exclusion and MTT assay, according to the manufacturer's recommendations (R&D Systems; catalog no. TA5355).

The Lyn-specific inhibitory peptide KRX-123.302 (sequence N to C terminus, where lowercase signifies D-isomer: myristyl-G-L-V-T-Y(di-iodo)-k-K-I-K-(amino-benzoyl)-NH2]) targets a unique interaction site within Lyn and has been described previously.¹⁷  Peptide was synthesized by Bachem (King of Prussia, PA), high-performance liquid chromatography (HPLC)–purified and dissolved in DMSO (1% maximum final concentration). HL-60 cells were pretreated for 2 hours with a Lyn-specific peptide (1 μM) or control buffer and then stimulated in the presence or absence of peptide with SDF-1 (100 ng/mL) or fMLP (100 nM), and a chemotaxis (transwell migration assays) or adhesion assay on ICAM-1–coated plates was performed.

Cells were incubated 48 hours after electroporation with Lyn siRNA or control siRNA, with mAb 24, mAbCD11b, or IgG1 for 30 minutes at 37°C in 200 μL buffer with SDF-1α (100 ng/mL). Cells were washed in ice-cold PBS, 1% BSA, 0.1% sodium azide, incubated with FITC-F(ab′)₂ of goat anti–mouse IgG at 4°C for 30 minutes, washed again, fixed with 1% formaldehyde, and then analyzed on a fluorescence-activated cell sorter (FACS).

To evaluate the possible role of Src family kinases in SDF-1 chemokine signaling, we initially screened SDF-1–stimulated cells for the activities of known Src kinases. We used the Mo7e cell line, as these primitive cells are highly responsive to SDF-1. We observed that the activity of Lyn tyrosine kinase was significantly increased in SDF-1–stimulated Mo7e cells, as we previously reported in HL-60 cells and primary CD34⁺ stem/progenitor cells.⁴  A representative time course for Lyn activation by SDF-1 in Mo7e cells is shown in Figure 1A. Lyn autophosphorylation was increased within 1 minute, with maximal activity detected between 1 minute and 3 minutes. Increased activity of the Src-family kinase, Lyn, indicates that Lyn is involved in SDF-1–induced activation of the CXCR4 signaling pathway.

An important question that has not yet been fully resolved is the contribution of a Lyn kinase–dependent pathway to the overall regulation of migration in SDF-1–stimulated hematopoietic cells. Previously we suggested that migratory responses to SDF-1 may be impaired in the absence of Lyn.⁴  To further evaluate the possible role of Lyn in the migration activation process, we silenced Lyn expression using siRNA^18,19  in primary CD34⁺ stem/progenitor cells, and Mo7e, HL-60, and Nalm-6 cell lines. Figure 1B shows that Lyn siRNA efficiently inhibited Lyn gene expression in these cells as reflected by an approximately 70% reduction in Lyn protein in CD34⁺ cells and an approximately 90% to 95% reduction in cell lines. We then evaluated the involvement of Lyn in SDF-1–mediated cell migration. We observed that selective ablation of Lyn by siRNA potently inhibited SDF-1–dependent cell migration (Figure 1C). The chemotactic index was reduced approximately 3- to 7-fold in Lyn siRNA–treated cells as compared with control siRNA–treated cells, thus providing support for the role of Lyn as a positive regulator of cell migration. These assays reflected changes in cell polarization directly induced by chemoattractant and thus, true migration. In order to address alterations in spontaneous cell movement, we also perfomed migration assays in the absence of SDF-1 over surfaces coated with an integrin ligand, ICAM-1. Our results with Mo7e, HL-60, and Nalm-6 cells clearly indicated that absence of Lyn had no effect on spontaneous cell movement (Figure 1D).

To be certain that Lyn siRNA did not exert toxic effects on cells that could account for the block of SDF-1–dependent migration, we also measured cell proliferation and viability by MTT assay. We did not detect any changes in cell proliferation or viability between Lyn siRNA–treated and control Mo7e cells (Figure 2A). The specificity of Lyn inhibition was demonstrated by the absence of changes in protein expression for the related Src family kinase Hck or in the “housekeeping” gene β-actin (Figure 1B). To be certain that unintended silencing of nontargeted genes was not responsible for the observed reduction in cell migration, we performed a functional rescue experiment (Figure 2B). We observed that cells treated with Lyn siRNA targeting the 3′ UTR underwent migration inhibition. In contrast, cells cotransfected with a Lyn expression construct that was lacking the 3′ UTR continued to migrate in response to SDF-1. These results unequivocally show that the SDF-1–induced migration of hematopoietic cells requires the Src family kinase Lyn.

Our results indicated that SDF-1–stimulated CD34⁺ stem/progenitor cells attached better to marrow stromal cells when transfected with Lyn siRNA (Figure 3A). Similarly, Lyn siRNA increased attachment of HL-60 and Nalm-6 cells to stromal cells in the presence of SDF-1 (Figure 3B). This initial observation suggested that when Lyn was activated by SDF-1 through CXCR4, the adherence of hematopoietic cells to stromal cells was inhibited. Possible reasons for enhanced cell attachment following Lyn depletion could include increased expression of cell surface integrins or alternatively, increased activity of pre-existing integrins. FACS analysis with antibodies directed against the integrins αL, αM, or β₁ and β₂ subunits showed that integrin expression levels in Lyn siRNA–transfected cells and control siRNA–transfected cells were almost identical (Figure 3C). These results indicate that changes in the cell-surface expression of integrins are not responsible for Lyn-mediated inhibition of cell adhesion. Furthermore, this suggests that Lyn decreases cell adherence by inhibiting the activity of integrins that are already present on the cell surface.

To investigate the possibility that Lyn could inhibit the ligand-binding activity of integrins in chemokine-stimulated myeloid cells, we tested cell adherence to surfaces coated with the integrin ligand, cell adhesion molecule (CAM). On bone marrow stromal cells, intracellular adhesion molecule-1 (ICAM-1) is highly expressed and required for the adherence of hematopoietic cells to stroma.^12,13,20  We observed that Mo7e cells stimulated with SDF-1 attached poorly to surfaces coated with ICAM-1, as compared with unstimulated cells (Figure 3D). Attachment of chemokine-stimulated cells to surfaces coated with VCAM-1 or control BSA was not altered. Since these results suggested that activation of Lyn by SDF-1 inhibited integrin-binding activity to ICAM-1, it seemed plausible that inhibition of Lyn might reverse this effect. Indeed, Mo7e cells transfected with Lyn siRNA attached dramatically better to surfaces coated with ICAM-1 than cells transfected with control siRNA (Figure 3D). Again, attachment to VCAM-1 or BSA was not changed in Lyn siRNA–transfected cells, as compared with control cells. Since ICAM-1 binds exclusively to the β₂ integrins LFA-1 or Mac-1, and VCAM-1 binds to integrins VLA-4 or α4β₇, our results indicate that Lyn selectively down-regulates affinity of β₂ integrins, but not β₁ and β₇ integrins in hematopoietic cells. As was true for Mo7e cells, HL-60 and Nalm-6 cells transfected with Lyn siRNA attached better to surfaces coated with ICAM-1 in the presence of SDF-1 (Figure 3E).

To be certain that the effects of Lyn ablation on the attachment of hematopoietic cells to bone marrow stromal cells is mediated through the β₂ integrin and its ligand ICAM-1, we used antibodies capable of preventing ICAM-1–dependent cell adhesion. Knockdown of Lyn in Mo7e cells stimulated with SDF-1 increased their attachment to stromal cells (Figure 3F). Antihuman ICAM-1 antibody, which selectively inhibits cell adhesion mediated by ICAM-1, reduced attachment of Lyn-deficient cells to stromal cells (Figure 3F). Next we examined the expression of activation-dependent epitopes of β₂ integrins in Lyn-deficient and control MO7e and HL-60 cells stimulated with SDF-1. Toward this end, we employed FACS analysis using mAb 24, which recognizes activated β₂ integrins LFA-1 and Mac-1 (CD11/CD18),²¹  or antibodies that recognize only activated Mac-1 (CD11b; “Materials and methods”). Our results clearly show that activation state–dependent conformational epitopes for β₂ integrin expression are increased in Lyn siRNA–transfected cells as compared with control siRNA–transfected cells (Figure 3G). However, the expression of the activation-dependent epitopes for Mac-1 (CD11b) was unchanged in these cells. Taken together, these results indicate that in SDF-1–stimulated cells, Lyn kinase inhibits the affinity of LFA-1 to ICAM-1.

To determine whether other chemoattractants, in addition to SDF-1, require Lyn to induce or inhibit chemotactic migration and cell adhesion, respectively, we evaluated effects of the N-formyl peptide, fMLP (fMetLeuPhe) in HL-60 cells treated with the Lyn-specific inhibitory peptide KRX-123.302.¹⁷  Previous published reports have shown that the Lyn kinase is activated in fMLP-stimulated human primary neutrophils^15,16,22 ; however, the biologic relevance of this activation has yet to be determined. Exposure of HL-60 cells to the Lyn inhibitory peptide decreased both fMLP-induced and SDF-1–induced cell migration but increased cell adhesion on ICAM-1–coated plates (Figure 4). These results clearly indicate that the fMLP chemoattractant, similar to SDF-1, requires Lyn to induce migration and inhibits β₂ integrin–dependent cell adhesion.

The intracellular signaling pathways that are required for coordinating hematopoietic cell movement and adhesion in marrow are largely unknown. The present studies clearly indicate that the Src family kinase Lyn is an important part of the regulatory network that links chemotactic signals to adhesive responses on the hematopoietic cell surface. We demonstrate a previously unrecognized function for Lyn as a negative regulator of SDF-1 chemokine–mediated cell adhesion in marrow microenvironment. Blockade of Lyn expression with siRNA in CD34⁺ stem/progenitor cells and several cell lines resulted in an increase of attachment of these cells to bone marrow stromal cells and decreased their movement in response to SDF-1. Treatment of cells with Lyn-specific siRNA had no effect on cell proliferation or survival. Inhibition of migration in these cells was overcome by cotransfection with a Lyn expression construct. Therefore, the effects we observed are not secondary to blockade of unintended genes, but can be attributed directly to specific inhibition of Lyn.

Evidence indicates that Lyn ablation increases adhesion through the modulation of integrins already present on the cell surface. Ligand-binding experiments showed increased attachment of hematopoietic cells to surfaces coated with ICAM-1 and to bone marrow stromal cells, following treatment with Lyn siRNA. Binding specificity was confirmed by the use of antibodies that block ICAM-1 on stromal cells. Since ligand binding is the most direct measure of integrin affinity, we conclude that Lyn ablation increases the affinity of the β₂ integrins to their ligand ICAM-1 in SDF-1–stimulated cells. We further addressed this issue by examining expression of activation-dependent epitopes of β₂ integrins in Lyn-deprived and control cells upon stimulation with SDF-1. We show that absence of Lyn increases the expression of the 24 epitope specific for β₂ integrin active conformation (Figure 3G). This result indicates that Lyn is directly involved in inside-out signaling leading to β₂ integrin inhibition. Since the active conformation of CD11b (Mac-1) was unchanged in these cells we conclude that in SDF-1–stimulated cells, Lyn kinase selectively inhibits the affinity of LFA-1 to ICAM-1. Thus, Lyn is a negative regulator of LFA-1 activity in hematopoieitc cells stimulated through the chemokine receptor CXCR4.

Hematopoietic precursors need to be held in close contact to bone marrow stromal cells for the extensive interactions necessary for proper growth and maturation. We suggest that the CXCR4-induced down-regulation of β₂ integrin affinity, through Lyn activity, may represent one mechanism by which hematopoietic precursors are aided in their migration and mobilization within the stromal microenvironment. We and others have observed that Lyn becomes activated after cells are stimulated with SDF-1 or other chemokines and chemoattractants. Kinase autophosphorylation is increased within seconds, with maximal activity detected typically between 0.5 and 3 minutes (Figure 1).^15,16,22  It is known that Src-dependent outside-in signaling triggered by ligand binding is necessary to stabilize β₂-integrin– and β₃-integrin–mediated adhesion.^23-25  The results we report indicate that when Lyn is activated through the CXCR4 chemokine receptor, β₂-integrin–mediated adhesion is transiently destabilized through inside-out signaling, thereby loosening otherwise tight cell attachments and facilitating hematopoietic cell movement within stroma in response to the chemokine SDF-1 (Figure 5). Conversely, reduced Lyn activity can activate integrins and attach hematopoietic cells strongly to stromal cells. The data presented here imply a Lyn-mediated mechanism that explains the ability of SDF-1 to promote the migration and homing of normal stem/progenitor cells to different niches within bone marrow during differentiation and maturation. Interestingly, Lyn-deficient mice are prone to myeloid expansion, as they display an increase in extramedullary myelopoiesis and severe splenomegaly, which are associated with significant increases and accumulation of myeloid progenitors in the spleen.²⁶  This strongly suggests that myeloid progenitor cells are not able to move adequately toward the SDF-1 gradient in Lyn-deficient mice, and it is consistent with our present and previous results.⁴  It is tempting to speculate that prolonged up-regulation of Lyn activity, due to stimulation by various factors (ie, oncoproteins), might promote cellular detachment and release from bone marrow extravascular space to the peripheral circulation. Since myeloid and B-lymphoid human malignant cells often contain constitutively active Lyn tyrosine kinase,^4,14,27  it will be interesting to see whether Lyn-mediated interactions between CXCR4 and integrins are altered in human leukemias and lymphomas. Our results also suggest that Lyn-mediated positive regulation of cell migration, which is accompanied by the negative regulation of adhesion, plays a role in macrophage/neutrophil activation through the NFP chemoattractant receptor during inflammation (Figure 4). Thus, the signaling paradigm we have described for CXCR4 could also provide a mechanism to explain the ability of other G-protein–coupled chemoattractant receptors to inhibit integrin activity. Future studies investigating the mechanisms by which chemoattractant receptors are linked to Lyn activation should prove informative, as will further investigation of integrin and cytoskeleton signaling elements coupled to this pathway.

We thank Dr Diane Linnekin (National Institutes of Health/National Cancer Institute, Bethesda, MD) for kindly providing us with Lyn cDNA and Dr Nancy Hogg (London Research Institute, Lincoln's Inn Fields Laboratories, London, United Kingdom) for mAb 24 for the active conformation of CD11/CD18. We are grateful to Dr Vincenzo Cirulli (University of California, San Diego) for his valuable comments.