Cross-talk between α₄β₁/α₅β₁ and c-Kit results in opposing effect on growth and survival of hematopoietic cells via the activation of focal adhesion kinase, mitogen-activated protein kinase, and Akt signaling pathways

Adhesive interactions between hematopoietic progenitor and stem cells and the hematopoietic microenvironment play a critical role in maintaining hematopoiesis.1-5Hematopoietic growth factors are potent regulators of hematopoiesis. In addition, these proteins have been implicated in modulating adhesion between hematopoietic progenitor cells and extracellular matrix proteins via changes in integrin receptor activation.6-9The role of adhesion molecules alone or with growth factors in maintaining proliferation, differentiation, and survival of hematopoietic cells is less understood,10,11 but collaboration between growth factor receptors and integrins has been hypothesized to be necessary for normal hematopoietic development.12 Specifically, integrin receptors such as alpha 4 beta 1 (α₄β₁) and/or alpha 5 beta 1 (α₅β₁), may collaborate in unique ways with receptor tyrosine kinases, to influence cellular events. In this regard, mice deficient in the receptor tyrosine kinase, c-Kit, its ligand stem cell factor (SCF), or β₁ integrins demonstrate hematopoietic defects of varying severity, suggesting critical roles for these proteins in normal blood development.1,13 14 

Our laboratory and other investigators have shown that receptors of the extracellular matrix protein, fibronectin (FN), are involved in the adhesion of hematopoietic cells, including stem and progenitor cells in the hematopoietic microenvironment.15-20 FN is expressed at high levels throughout the hematopoietic microenvironment.21,22 The FN molecule contains binding sites for heparin, collagen, fibrin, and gelatin, suggesting that it plays an important role in regulating the architecture of the hematopoietic microenvironment. The binding of hematopoietic cells to FN is mediated by at least 2 integrin receptors. The α₅β₁ receptor recognizes the minimal binding sequence Arg-Gly-Asp (single-letter amino acid code: RGD), as well as 2 other synergistic binding sites, all of which are located within the cell-binding domain of the FN molecule,23,24and α₄β₁ binds sequences within the alternatively spliced IIICS region of FN defined by the synthetic peptides CS-1 and CS-5.25,26 These receptors play a critical role in normal hematopoietic development.1,27 28 

Mutant mice homozygous for null mutations of c-Kit, or its ligand SCF, die in embryonic development or shortly after birth due to severe anemia.14,29-31 Viable homozygous mutants of c-Kit also demonstrate severe anemia and a marked reduction in both immature and mature erythroid progenitors. Data from these mutant mice show a critical role for c-Kit–mediated signaling in normal erythroid development.29,30 Interactions of erythroid cells with FN are also believed to be essential for erythropoiesis, particularly for terminal stages of erythroid differentiation.32-36Erythroid progenitors express both α₄β₁ and α₅β₁.37,38 Efficient production of mature cells in vitro requires adhesion to FN in some systems.39-42 In addition, treatment of normal mice with anti-α₄β₁ antibody completely blocks erythropoiesis.32 Some evidence of collaboration between c-Kit and integrins also exists. Exposure to SCF increases adherence of hematopoietic cells to FN by “inside-out signaling.”6,9 43-47 Together, these data suggest a role for both c-Kit and integrin receptors in normal erythroid development.

In some cell systems, signaling downstream of receptor tyrosine kinases and integrins appear to comodulate cellular events.48 In fibroblasts, autophosphorylation of receptor tyrosine kinases can be enhanced by adhesion to matrix proteins, such as FN.49,50In addition, platelet-derived growth factor receptor and epidermal growth factor receptor phosphorylation following growth factor treatment is greater in adherent cells compared to suspended cells.51-53 Synergy between growth factors and cell adhesion in activation of the mitogen-activated protein kinase (MAPK) and phosphoinositide-3 kinase (PI-3K) cascade has also been observed. Akt has been shown to be synergistically activated by adhesion and epidermal growth factor treatment in fibroblasts. These signaling effects correlated with increased cell survival, consistent with an important role for the PI-3K/Akt pathway in cell survival.54 55 

Given the significance of c-Kit and β₁-integrin signals in the development of erythroid cells, and the known interaction between receptor tyrosine kinases and integrins in cells of nonhematopoietic origin, we hypothesized that concurrent ligation of integrins and c-Kit has significant effects on the activation of intracellular signaling pathways and subsequent behavior of cells in the erythroid lineage. Because adhesion of hematopoietic cells to FN is a dynamic process, involving coordinated, successive attachment and detachment,12,56 57 we have examined the effect of this dynamic process over time on erythroid progenitor cell (EPC) survival/apoptosis and proliferation using cell cultures grown on FN peptides that specifically bind α₄β₁(FN-H296) or α₅β₁ (FN-CH271) or both α₄β₁ and α₅β₁(FN-CH296) integrin.

The G1E-ER2 cells have been previously described and were obtained from Dr Mitch Weiss (Ontogeny, Boston, MA). Unless otherwise specified, G1E-ER2 cells were grown in Iscove modified Dulbecco medium (IMDM; GIBCO/BRL, Gaithersburg, MD) with 15% heat-inactivated embryonic stem cell (ES) serum (Hyclone, Logan, UT), recombinant erythropoietin (Epo; 2 U/mL; Amgen, Thousand Oaks, CA), and recombinant rat (rr) SCF (50 ng/mL; Amgen). For starvation experiments, cells were washed 3 times in IMDM, then resuspended in the same medium without serum and growth factors for 6 to 8 hours. Primary EPCs were derived from fetal livers of 12.5-day-old wild-type embryos. Briefly, single-cell suspensions were prepared and fetal liver cells were cultured in IMDM with 10% fetal calf serum (FCS; Hyclone), recombinant Epo (2 U/mL; Amgen), and rrSCF (50 ng/mL; Amgen) for 3 to 4 weeks.

Phycoerythrin (PE)–conjugated monoclonal antibodies (mAbs) were directed against c-Kit and α₅β₁. Fluorescence isothyocyanate (FITC)–conjugated antibodies were directed against α₄β₁. All the PE- and FITC-conjugated mAbs, including the isotype control antibodies, were purchased from Pharmingen (San Diego, CA). G1E-ER2 cells (1 × 10⁶) were incubated at 4°C for 30 minutes with 1 μg of the primary mAb. Cells were washed 3 times with phosphate-buffered saline (PBS) containing 0.1% bovine serum albumin (BSA; Sigma, St Louis, MO), and analyzed by fluorescence-activated cell sorter (FACS; Becton Dickinson, San Jose, CA).

Recombinant human FN peptides H296 and CH271 (Figure1A) were obtained from Takara Shuza (Otsu, Japan). Nontissue culture 6-well plates were coated with FN fragments diluted in PBS at 100 nmol/cm² overnight as described previously.18 We have also previously demonstrated that adhesion to these various recombinant FNs was mediated specifically on hematopoietic cells, through their integrin receptors α₄β₁ and α₅β₁.18 To block nonspecific binding sites, plates were incubated for 30 minutes with 2% BSA in PBS. Wells were then washed 3 times with PBS. Factor-starved G1E-ER2 cells (2 × 10⁶) were allowed to adhere to FN peptides for various time periods at 37°C in the presence of 10 ng/mL rrSCF. After incubation, nonadherent cells were collected by carefully rinsing the plates with medium and cell counts were performed.

The effect of FN peptides H296 and CH271 and SCF on proliferation of G1E-ER2 cells and primary EPC was assayed using thymidine incorporation. The 96-well nontissue culture plates were coated with FN peptides as described above. Growth factor-starved G1E-ER2 cells and primary EPC were plated at 5 × 10⁴cells/well for 48 hours, either in the presence or absence of 100 ng/mL rrSCF. Subsequently, 1.0 μCi of [³H]-thymidine (Amersham) was added to each well for 6 to 8 hours at 37°C. Cells were then harvested using an automated cell harvester (96-well harvester, Brandel, Gaithersburg, MD) and thymidine incorporation was determined in a scintillation counter. The effect of FN peptides and SCF on cell death (apoptosis and necrosis) of G1E-ER2 cells was assayed by staining the cells with annexin-FITC and propidium iodide (PI) according to the manufacturer's instructions (Pharmingen, San Diego, CA). The 24-well nontissue culture plates were coated with FN peptides CH296, CH271, and H296 as described above. Growth factor-starved G1E-ER2 cells were plated at 5 × 10⁵ cells/well for 48 hours, either in the presence or absence of 100 ng/mL rrSCF. Subsequently, cells were harvested and stained with annexin-FITC and PI and analyzed by flow cytometry.

Activation of MAPK (ERK-1 and ERK-2) was determined by using phospho-specific ERK antibody (New England Biolabs, Beverly, MA). This antibody detects ERKs only when they are catalytically activated by phosphorylation. Activation of Akt was determined by using a phospho-specific Akt (S473) antibody (New England Biolabs). Activation and expression of FAK was determined by using an anti-FAK antibody (Upstate Biotechnology, Lake Placid, NY). Expression of Bcl-2 and Bcl-x_L was determined by using anti–Bcl-2 and anti–Bcl-x_L antibodies (Pharmingen). All antibodies were used at 1:2000 dilution. Briefly, nontissue culture 6-well plates were coated with FN fragments as described above. Factor-starved 5 to 8 × 10⁶ G1E-ER2 cells were loaded onto FN-coated wells and cultured for various time points at 37°C in the presence or absence of SCF. Thereafter, cells were harvested and lysed in lysis buffer at 4°C for 30 minutes. Cell lysates were clarified by centrifugation for 30 minutes at 10 000g at 4°C. Equal amount of protein was fractionated on 12% polyacrylamide/sodium dodecyl sulfate (SDS) gel and electrophoretically transferred to nitrocellulose membrane. Western blot analysis was performed according to the manufacturer's instructions (New England Biolabs).

To study potential interactions between c-Kit and integrins α₄β₁ and α₅β₁, we confirmed c-Kit, α₄β₁, and α₅β₁ expression on the erythroid cell surface by flow cytometric analysis using anti–c-Kit, anti-α₄β₁, and anti-α₅β₁ mAbs coupled to either FITC or PE. G1E-ER2 cells uniformly express c-Kit (Figure 1B), α₄β₁ (Figure 1C), and α₅β₁ (Figure 1D). One hundred percent of G1E-ER2 cells express integrins α₄β₁ and α₅β₁ (Figure 1C,D). To confirm that α₄β₁ or α₅β₁ or both mediate the adhesion of G1E-ER2 cells to FN, we used 2 recombinant peptides containing the single binding domain for α₄β₁ (H296) or α₅β₁ (CH271) (Figure 1A).11,18G1E-ER2 cells were plated on FN-H296 or FN-CH271 and adhesion measured over 2 hours. Significant adhesion to FN fragments was observed by either α₄β₁ or α₅β₁. 71% ± 3.1% G1E-ER2 cells were adherent to FN-CH271 via α₅β₁ and 77% ± 5.8% via α₄β₁ to FN-H296 (data not shown). Previous studies using primary EPCs have shown similar levels of adhesion to FN as well as to FN peptides.58 In contrast, the majority of the cells (90%) were in suspension in BSA-coated dishes used as control cultures. In previous studies using these same fragments and other hematopoietic cell lines or primary cells, we have shown specificity of adhesion on these fragments with blocking antibodies.18 Because the majority of EPCs were adherent to FN peptides and because previous studies have shown that adhesion of hematopoietic cells to FN is a dynamic process, in subsequent studies we have examined the impact of this process on cell cultures incubated in dishes coated with FN-H296, FN-CH271, or BSA.

The role of integrins in erythroid cell proliferation has not been determined, although studies have suggested that FN is necessary both in vitro and in vivo to provide the appropriate niche for erythroid development.34-40 To determine if integrins are important mediators of mitogenesis in erythroid cells, we measured DNA synthesis in G1E-ER2 cells and primary fetal liver-derived EPCs cultured on FN peptides in the presence or absence of SCF. After 48 hours in culture on FN peptides, [³H]-thymidine was added for 6 hours and [³H]-thymidine incorporation was determined. In experiments in which no SCF was added to cultures, factor-starved G1E-ER2 cells cultured on BSA or FN-CH271 showed similar proliferative response (11 071 ± 1257 counts per minute [cpm] for BSA versus 10 085 ± 1108 cpm for CH271, mean ± SEM, respectively, from 6 different experiments, P > .05). In contrast, cells cultured on FN-H296 or FN-CH296 resulted in significantly less DNA synthesis (6958 ± 756 cpm for H296 and 5689 ± 614 cpm for CH296 versus 10 085 ± 1108 cpm for CH271, mean ± SEM, respectively, from 6 different experiments, P < .05).

In experiments in which SCF was added to cultures, G1E-ER2 cells cultured on FN-CH271 (mediating adhesion via α₅β₁) demonstrated significantly higher proliferation compared with cells stimulated with SCF in suspension (BSA) (231 301 ± 12 114 cpm CH271 versus 145 173 ± 10 832 cpm BSA, mean ± SEM, respectively, from 6 different experiments,P < .05). In contrast, even in the presence of SCF, culturing these cells on FN-H296 (mediating adhesion via α₄β₁) or FN-CH296 (mediating adhesion via both α₅β₁ and α₄β₁) was associated with reduced proliferation compared to cells cultured in suspension or FN-CH271 (Figure 2A). A similar decrease in proliferation was also noted in primary fetal liver-derived erythroid progenitors cultured on FN-H296 or FN-CH296 compared to FN-CH271 (Figure 2B). Thus, these studies demonstrate that different biologic outcomes are stimulated by adhesion of erythroid cells to FN via α₄β₁ compared with α₅β₁.

To test whether the increase in proliferation of erythroid progenitors that occurs after engagement of α₅β₁ was associated with activation of downstream kinase signaling pathways, we measured FAK and MAPK (ERK-1 and ERK-2) activation. These pathways have previously been implicated in integrin-mediated signal transduction in hematopoietic cells.59 G1E-ER2 cells were starved for 7 hours, allowed to adhere to FN via α₄β₁ or α₅β₁ for various time points, lysed, and subjected to Western blot analysis using an anti-FAK or antiphospho-MAPK antibody. In the presence of SCF, engagement of α₅β₁ strongly induced FAK activation as early as 12 hours after stimulation, reaching maximum levels at 48 hours (Figure 3A). Ligation induced by α₄β₁ also resulted in FAK activation, however, at significantly reduced levels in comparison to activation via α₅β₁ (Figure 3A). We next examined the activation of MAPK (ERKs), and first measured activation of ERKs in G1E-ER2 cells stimulated with SCF in suspension. Stimulation of these cells by SCF resulted in significant but only transient activation of ERK-1 and ERK-2 (Figure 3B). Activation peaked 10 minutes after SCF stimulation and dropped to baseline thereafter. In contrast, in the presence of engagement of α₅β₁, SCF strongly induced ERK-1 and ERK-2 activation in erythroid cells as early as 10 minutes after stimulation that persisted and reached maximum levels at 120 minutes (Figure 3C). Once again, engagement via α₄β₁ also resulted in ERK-1 and ERK-2 activation, however, at significantly reduced levels in comparison to α₅β₁ (Figure 3C). These data suggest that increased proliferation of erythroid cells associated with ligation via α₅β₁ may in part be due to sustained and enhanced activation of FAK/MAPK (ERK-1 and ERK-2) cascade in these cells.

To further examine the extent of involvement of the ERK pathway in c-Kit and/or integrin-mediated proliferation of erythroid cells, we used a specific pharmacologic inhibitor (PD98059) of the MAPK (ERK) cascade. Factor-starved G1E-ER2 cells were cultured on FN-CH271 or H296 or in suspension in the presence or absence of SCF and PD98059. After 48 hours of coculture, [³H]-thymidine was added for 6 hours and [³H]-thymidine incorporation was determined. Despite evidence of transient ERK activation (Figure 3B), PD98059 had minimal effect on the proliferation of G1E-ER2 cells grown in suspension in presence of SCF (Figure 3D). In contrast, proliferation of G1E-ER2 cells cultured on FN-H296 and FN-CH271 was inhibited by 65% and 30%, respectively, in the presence of PD98059. These data suggest that integrin- and c-Kit-stimulated growth of G1E-ER2 cells is dependent on activation of the MAPK (ERK) pathway, although the extent of use of this pathway differs significantly after ligation of α₄β₁ and α₅β₁.

Stem cell factor is an important survival factor for c-Kit⁺ cells and protects some cells from growth factor withdrawal-induced apoptosis.29 Integrin-mediated interactions with FN also play an important role in regulating cell survival and apoptosis in some cells. To determine if the reduction in proliferation of G1E-ER2 cells noted in cells cultured on FN–H296 or FN-CH296 was in part due to increased cell death, we compared apoptosis in G1E-ER2 cells cultured on different FN peptides in the presence or absence of SCF using a combination of PI and annexin V staining. G1E-ER2 cells were factor starved for 6 to 8 hours and then cultured for 48 hours in suspension or on FN-H296 or FN-CH296 or FN-CH271 in medium with or without SCF. Cells were harvested, stained with annexin V and PI, and scored for the percentage of apoptotic (annexin V⁺/PI⁻) and necrotic (annexin V⁺/PI⁺) cells using FACS analysis. In experiments performed in the absence of c-Kit stimulation, G1E-ER2 cells cultured on FN-H296 demonstrated significantly more cell death compared to cells cultured on FN-CH271 (52% ± 1 [H296] versus 33.4% ± 1 [CH271], mean ± SD, respectively,P < .05; Figure 4A). A similar increase in cell death was noted in cells cultured on FN-CH296 (mediating adhesion via both α₄β₁ and α₅β₁) in comparison to FN-CH271 (55% ± 3 [CH296] versus 33.4% ± 1 [CH271], mean ± SD, respectively, P < .05). The survival of cells grown in suspension was similar to that observed in cultures grown on FN-CH271 (29.3% ± 2 [BSA] versus 33.4% ± 1 [CH271], mean ± SD, respectively P > .05). These data suggest that adhesion to FN via α₄β₁ stimulates apoptosis of G1E-ER2 cells in the absence of growth factor. Further, these data demonstrate that adhesion to FN via α₄β₁has a dominant effect on the survival of G1E-ER2 cells.

To investigate the effect of c-Kit activation on reversal of apoptosis, G1E-ER2 cells were cultured on FN peptides in the presence of SCF and apoptosis measured. G1E-ER2 cells cultured on FN-H296 in the presence of SCF demonstrated significant improvement in survival in comparison with cells grown in absence of SCF. However, the rescue was only partial and significantly less in comparison to cells cultured on FN-CH271 and SCF (24.8% ± 5.3% α₄β₁versus 8.6% ± 0.2% α₅β₁, mean ± SD, P < .05; Figure 4B). Similar results were seen when these cells were cultured on FN-CH296 and SCF (27.2% ± 2.2% α₄β₁ and α₅β₁versus 8.6% ± 0.2% α₅β₁, mean ± SD, P < .05). The percentage of surviving cells in suspension culture in the presence of SCF was similar to that seen in cells cultured on FN-CH271 (7.45% ± 1.45% BSA versus 8.6% ± 0.2% CH271, mean ± SD, respectively). These results suggest that the reduced proliferation seen in erythroid cells following adhesion to FN via α₄β₁ is due in part due to an increase in the number of erythroid cells undergoing apoptosis.

The downstream effector of PI-3K, Akt, has been implicated as an important antagonist of apoptosis.55 Previous studies have linked reduced activation of Akt to enhanced apoptosis.55Further, in some cells expression of constitutively activated forms of Akt has been shown to block apoptosis, whereas use of an Akt inhibitor, wortmannin, augments apoptosis.60 61 Therefore, we measured Akt activation in the cell culture conditions described above. Ligation of either α₄β₁ or α₅β₁ induced Akt activation in erythroid cells, but to different degrees (Figure5A). Akt activation induced by α₄β₁ was significantly less throughout the incubation period compared with Akt activation mediated by α₅β₁ in G1E-ER2 cells. This 3- to 4-fold reduction in Akt activation via α₄β₁attachment was seen at all time points examined (Figure 5A). To further investigate the importance of Akt activation in preventing apoptotsis of these cells, we cultured G1E-ER2 cells on FN-CH271 or in suspension (BSA control) in the presence of SCF and wortmannin. The presence of wortmannin under these culture conditions resulted in a significant increase in apoptosis of G1E-ER2 cells grown on FN-CH271 (Figure 5B). A similar increase in apoptosis was also observed in G1E-ER2 cells grown on FN-H296 (Figure 5B). In addition to reduced Akt activation, adhesion of G1E-ER2 cells to FN-H296 also resulted in reduced Bcl-2 and Bcl-x_L expression (Figure 5C,D). Together, these data suggest that the PI-3K/Akt signaling cascade plays an important role in mediating erythroid cell apoptosis downstream from c-Kit and integrins. Adhesion of G1E-ER2 cells to FN via α₄β₁significantly impairs Akt activation and the expression of downstream antiapoptotic proteins; this impairment may be one of the mechanism(s) of enhanced apoptosis of these cells in comparison to cells grown in suspension or on FN via α₅β₁.

In the developing embryo and in adult animals c-Kit and integrin-mediated signaling are necessary for erythroid cell survival, proliferation and differentiation.1,14,29,30 Therefore, erythroid progenitors provide a unique model to study the mechanism of c-Kit- and integrin-mediated signaling in a physiologically relevant context. To facilitate studies on the mechanisms governing the pleiotropic responses of c-Kit and integrins we have used an EPC line (G1E-ER2) that resembles primary cells at the progenitor stage of development. These cells are similar to primary erythroid progenitors with respect to globin- and erythroid-specific transcription factor expression and differentiation.62 63 These cells were derived from ES cells of mice deficient in GATA-1 expression. Instead of undergoing apoptosis, these cells grow continuously in culture as developmentally arrested precursors. We demonstrate that G1E-ER2 cells express c-Kit, α₄β₁, and α₅β₁ and adhere to FN peptides at levels similar to primary EPCs.

The role of cell adhesion to FN during erythroid cell development per se has not been completely determined, although several studies have suggested that FN is necessary both in vitro and in vivo to provide an appropriate niche for erythroid development and also to provide proliferative stimulus for erythroid cells.32-34 36 The studies presented here demonstrate significantly greater proliferation and enhanced activation of ERKs in cells grown in cultures containing α₅β₁ adhesion sites compared with cells grown in suspension or in cultures containing the α₄β₁ binding site. In addition, our results show significant quantitative differences in the activation of both FAK and MAPK (ERK-1 and ERK-2) in cells grown on FN mediating adhesion via these 2 different β₁-integrin receptors. Specifically, engagement of α₄β₁ on G1E-ER2 cells results in significant reduction of FAK, ERK-1, and ERK-2 activation in comparison with cells grown in cultures containing α₅β₁ binding sites. The decrease in FAK, ERK-1, and ERK-2 activation correlated with reduced proliferation of these cells. In addition, the differences in activation appear to be important in cell proliferation because cells cultured on FN containing α₄β₁ binding sites were more resistant to the growth-inhibitory effects of the specific MEK inhibitor PD98059.

In contrast, adhesion of cells via α₄β₁ resulted in significant cell death. In addition to reduced activation of FAK and MAPK, inhibition of growth in response to ligation via α₄β₁ correlated with significant reduction in Akt activation in these cells. Although, SCF-induced activation of c-Kit partially prevented apoptosis in these cells, in comparison to cells grown on FN mediating adhesion via α₅β₁ or in suspension this reversal of apoptosis was incomplete. Studies in fibroblasts have shown that integrins and growth factors can jointly regulate Akt activation. In fibroblasts, activation of Akt leads to suppression of apoptosis, whereas in other cells activated Akt can protect cells from apoptosis in response to growth factor withdrawal, and apoptosis can be accelerated in these cells by dominant-negative Akt.61Recently, we have demonstrated that prolonged activation of Akt by stimulation of c-Kit via membrane-associated SCF is associated with erythroid cell survival/proliferation.64 Our finding that α₄β₁-mediated ligation results in reduced Akt activation and consequently greater apoptosis is consistent with these previous studies implicating Akt in hematopoietic cell survival. A key role for PI-3K-dependent Akt activation in erythroid cell survival via integrins and/or growth factor receptor stimulation was further demonstrated in the studies presented here by the use of the PI-3K inhibitor wortmannin. The results suggest that the PI-3K/Akt pathway is a key antagonist of cell death downstream of c-Kit and integrins in erythroid cells.

The mechanism(s) of the opposing effects of ligation of α₄β₁ and α₅β₁on growth and survival of erythroid cells is not clear. Although, G1E-ER2 cells express both α₄β₁ and α₅β₁ at comparable levels, we cannot rule out the possibility that the overall avidity of α₄β₁ and α₅β₁for FN might be different. These differences in overall avidity may result in different biologic and biochemical outcomes. Alternatively, the observed differences may be due to the generation of unique signals via the engagement of specific α₅ and α₄chains of α₅β₁ and α₄β₁ with FN, respectively. For instance, the studies presented here using EPCs demonstrate significantly greater activation of FAK in cells grown in the presence of FN containing α₅β₁ binding sites compared to α₄β₁. In this regard, it is possible that enhanced FAK activation via α₅β₁ and c-Kit could lead to more efficient recruitment of Ras, and hence to downstream kinase cascade of Raf-1, MEK, and MAPK. In fibroblasts evidence supports this model. Adhesion-mediated autophosphorylation of FAK leads to Src recruitment, increasing tyrosine phosphorylation of FAK, and the subsequent binding of SH2-domain proteins including Shc and the Grb2/Sos complex.65-67 The formation of a FAK/Src/Grb2 complex suggests the possibility of further signaling to MAPK. In addition, studies have shown that overexpression of FAK results in Ras-dependent activation of MAPK.68 Thus, enhanced phosphorylation of FAK via α₅β₁and c-Kit activation may explain the enhanced activation of ERKs in cells grown on FN mediating adhesion via α₅β₁. In contrast, recent studies using primary human periodontal ligament fibroblasts have shown that the adhesion to FN via α₄β₁ results in reduced FAK expression. Changes in FAK expression after adhesion via α₄β₁ in these cells were shown to be due to the activation of the caspase cascade.69 Our preliminary data, using the caspase-3 inhibitor, also demonstrate inhibition of apoptosis in cells cultured on FN-H296 (R.K. and D.A.W., unpublished observations, July 1999). Thus, our data in addition to these published results provide a mechanism by which opposing effects of integrin ligation could be mediated in erythroid cells.

Previous studies, using long-term human bone marrow cultures have demonstrated inhibition of hematopoietic cell proliferation in response to direct adhesion to bone marrow stroma via integrin receptors.17 The data presented here suggest that α₄β₁ compared to α₅β₁-mediated ligation to FN triggers unique signals in erythroid progenitors. The results of these signals are divergent, death versus proliferation. Taken together, these data support the notion that specific integrin-FN interactions in hematopoietic cells can regulate cell survival and proliferation and these effects can be modulated via ligand-induced stimulation of growth factor receptors.

We thank Eva Meunier and Sharon Smoot for assistance in preparation of this manuscript and expert administrative assistance. We thank Takara Shuzo, Biomedical Group (Otsu, Japan) for providing fibronectin peptides. We thank Drs Mervin Yoder and Don Durden for review of the manuscript and members of our laboratories for useful discussions.