Fibronectin receptor modulates cyclin-dependent kinase activity.

The high affinity fibronectin receptor (FNR) is expressed by hematopoietic cells, fibroblasts, and proliferating epidermal cells. Expression of this integrin is altered by chemical and viral transformation, suggesting that FNR dysfunction may play a role in growth control. This study demonstrates that exposing FA-K562 cells to glycine-arginine-glycine-aspartate-serine (GRGDS), a peptide ligand of the FNR, specifically stimulates p34/cdc2- and cyclin A-associated kinase activities. This occurs within 2 h of peptide addition. The 110-kDa form of the retinoblastoma protein appears within 3 h of GRGDS addition, consistent with activation of a G1/S kinase. DNA staining profiles demonstrate that GRGDS induces cell cycle progression within 24 h. Increased anchorage-independent growth is subsequently observed in GRGDS-treated FA-K562 cells. The control peptide, GRGES, which cannot bind the FNR, has none of these effects. This demonstrates that an extracellular integrin ligand can regulate cell proliferation. Furthermore, these results suggest that integrins link the extracellular environment and intracellular growth regulators.

Although initially thought to mediate cell-substrate adhesion to fibronectin only, recent studies have demonstrated a role for the FNR in growth control. For example, FNR overexpression has been associated with loss of anchorage-independent growth of human erythroleukemia cells and hamster fibroblasts ( 5 , 6). T cell proliferation can be induced by ligation of CD3 and either the a& or a& FNR ( 7 , 8). This study examines whether the a& FNR alone plays a role in cell growth regulation. The FA-K562 cell line was used for most of these studies, This stable human cell line expresses 5-fold more cell-surface FNR than the parental K562 cell line ( 5 ) . The FNR is the only RGD-binding receptor expressed by both K562 and FA-K562 cell lines. Thus, FA-K562 cells provide a model for studying potential effects of FNR/peptide ligand interactions on cell growth.

EXPERIMENTAL PROCEDURES
Cell Culture"K562 cells were obtained from the American Type Culture Collection. The FA-K562 subline, selected for its ability to * This work was supported by National Heart, Lung, and Blood Institute Grant K11 HL02216. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. I The abbreviations used are: FNR, fibronectin receptor; PAGE, polyacrylamide gel electrophoresis.
adhere to fibronectin, has been previously described (5). Both cell lines were cultured in RPMI 1640 medium containing 10% fetal bovine serum at 37 "C in a humidified 5% Con, 95% air incubator.
Soft Agar Colony Formation-Cells were washed and plated a t a concentration of lo3 cells/ml in RPMI 1640 medium containing 20% fetal bovine serum, 1% Difco Bacto-agar, and 10 pg/ml GRGDS or GRGES (Peninsula Laboratories, Inc.). Colonies were counted 14-21 days later. Cloning efficiencies were calculated as the percent of plated cells that formed macroscopic colonies. Cell Cycle Analysis-Cells were incubated with or without GRGDS or GRGES peptide (10 pg/ml) in RPMI 1640 medium, 10% fetal calf serum (5 X lo5 cells/ml). Cell aliquots (lo5) were removed at 4 or 24 h, washed, and sequentially incubated in lysis buffer (10 mM Tris, 30 mM NaC1, 20 m M MgCl,, pH 7.4), followed by lysis buffer containing 1% Nonidet P-40 (5 min each on ice). RNase (Boehringer Mannheim) was then added at 50 units/ml, and the cells were incubated for 30 min at 37 "C. Propidium iodide was added to a final concentration of 40 pg/ml, and the cells were incubated for 30 min on ice prior to analysis using an EPICS V flow cytometer interfaced to an MDADS computer. Data were analyzed using REPROMAN software (True-Facts Inc., Seattle, WA). Cell doublets and debris were excluded from analysis by gating on integrated propidium iodide to peak propidium iodide.
Cell lysates were reacted with anti-cdc2, anti-cdk2, anti-cyclin E (9), or normal rabbit serum (1:50 final dilution), followed by protein A-agarose beads. The first two antisera were raised against the unique carboxyl terminus of cdc2 or cdk2 and can recognize both free and cyclin-bound cdk2 (9). In three experiments, cyclin A or E was immunoprecipitated using a mouse monoclonal or rabbit antiserum (9), respectively. Beads were washed, resuspended in 50 pl of kinase buffer (20 mM Tris, 10 mM MgC12, 1 mM dithiothreitol, 30 p~ ATP, 10 pCi of [Y-~'P]ATP (3000 Ci/mmol; Amersham Corp.) with or without 1 pg of histone Hl/reaction (Sigma)), and incubated for 30 min at 37 "C prior to solubilization in SDS-PAGE sample buffer and electrophoresis on 12% polyacrylamide slab gels. Gels were fixed in 10% methanol, 10% acetic acid dried; and exposed to Kodak X-Omat film for 30-120 min. Kinase activity was quantitated by two methods (densitometry of developed films and Cerenkov counting of excised gel bands), both of which gave equivalent results.

RESULTS AND DISCUSSION
A key difference between FA-K562 and the parental cell line is that K562 cells grow well in soft agar, whereas FA-K562 cells do not (cloning efficiencies of >95 and <lo%, respectively) ( 5 ) . Thus, FA-K562 cells appear to be growth-inhibited compared to parental K562 cells. GRGDS and GRGES have no effect on the soft agar growth of K562 cells (96 versus 99% cloning efficiency, respectively) ( Fig. 1). In contrast, GRGDS, the FNR-binding peptide, stimulates the soft agar growth of FA-K562 cells (from 3 to 98% cloning efficiency) (Fig. 1). GRGES has no effect on FA-K562 soft agar colony formation. In previous studies, anti-FNR antibodies stimulated the soft agar growth of FA-K562 cells in a specific and concentration-dependent manner (5). Taken together, these results suggest that FNR perturbation (or occupancy) stimulates the growth of FA-K562 cells.
The effects of GRGDS peptide on cell cycle progression of FA-K562 and K562 cells was examined to evaluate the mechanism for colony growth in GRGDS-treated FA-K562 cells. FA-K562 and K562 cells were incubated for 4 or 24 h with GRGDS or control GRGES or with no peptide. The DNA staining profiles of cells treated for 24 h with GRGDS or GRGES peptide are shown in Fig. 2. The DNA profiles of untreated or GRGES-treated FA-K562 and K562 cells were identical (data not shown). No differences in the DNA staining profiles of any treatment group were observed after only 4 h. Incubation with GRGDS decreased the proportion of G, phase FA-K562 cells (from 52 to 36%) and increased the proportion of S phase FA-K562 cells compared to untreated or GRGES-treated cells (Fig. 2, left). In contrast, incubation with GRGDS had no effect on cell cycle progression of K562 cells (Fig. 2, right). Thus, peptide effects on both the cell cycle and anchorage-independent growth were limited to GRGDS peptide and the FNR-overexpressing FA-K562 line. These results indicate that increased DNA synthesis occurred within 24 h of exposure to GRGDS. This interval was found to be even shorter when [3H]thymidine uptake assays were performed. Five hours after the addition of peptides, [3H]thymidine uptake was 11-fold higher for GRGDS-treated than for GRGES-treated FA-K562 cells. This is consistent with the hypothesis that GRGDS rapidly stimulates cell proliferation. Furthermore, these results suggest that the increased soft agar growth of FA-K562 cells observed days after the addition of GRGDS peptide reflects an early commitment to cell proliferation induced by GRGDS.
The two major control points in the cell cycle occur at the G,/S transition and at the G2/M transition (10, 11). Activation of one or more cell cycle-regulated, cyclin-dependent, serine/threonine kinases regulates these transition points (10, 11). Mammalian p34 activation requires its association with cyclins and its dephosphorylation (12)(13)(14). The kinase activity of p34/cdc2 or cdk2 immunoprecipitated from cells incubated with GRGDS or GRGES peptide was measured. Antibodies to the unique carboxyl terminus of cdc2 or cdk2 were used for these experiments (9). p34/cdc2 kinase activity was specifically increased in GRGDS-treated FA-K562 cells by 2 h after the addition of peptide, as shown in Fig. 3. Little or no kinase activity was detected in immunoprecipitates using nonimmune rabbit serum (Fig. 3, NRS). No histone phosphorylation was observed in assays performed in the absence of exogenous histone. Histone phosphorylation was 15-fold higher using p34 immunoprecipitated from GRGDS-treated compared with GRGES-treated FA-K562 cells. Preliminary studies in normal human keratinocytes also demonstrated a specific increase in p34 activity induced by GRGDS treatment (data not shown).

GRGDS
A number of different cyclins can associate with the p34/ cdc2 kinase (9,13). Cyclin B-cdc2 complexes have been implicated in the control of mitosis, whereas complexes of cyclin E or A with cdc2 act at the GI/S transition (9, 13). K' mase activities associated with cyclins A and E were measured in control and RGD-treated FA-K562 cells. As shown in Fig. 3, little kinase activity was found associated with cyclin E isolated from control or RGD-treated cells. Longer exposures of autoradiographs showed that cyclin E-associated kinase activity was unchanged by RGD treatment. Kinase activity was detected in cyclin A immunoprecipitates from control cells. Modest but consistent increases in cyclin A-associated kinase activity were observed in RGD-treated compared with control-treated cells in each of three experiments. These findings are compatible with a number of interpretations. Immunoprecipitation followed by Western blotting was performed to determine whether RGD treatment caused an association between cyclin A and cdc2 kinase. Cyclin A or control immunoprecipitates from control or RGD-treated cells were probed with antisera to cdc2 or cdk2. The reverse of this experiment, where cdc2 and irrelevant antibody immunoprecipitates were probed with anti-cyclin A, was also performed. These experiments demonstrated that the RGD-induced increase in cdc2 kinase activity was not due to cyclin A-associated cdc2 (data not shown).
These results clearly demonstrate that an FNR ligand activates cdc2. p34/cdc2 controls both the GI/S and G2/M transitions in yeast (11), although the situation in mammalian cells is more complicated. Since increased DNA synthesis was observed in less time than would be expected for cells to complete mitosis and enter another G,/S transition, it appeared that GRGDS was affecting a kinase acting at the GI/ S transition. To investigate this further, the SDS-PAGE mobility of the retinoblastoma protein immunoprecipitated from control and GRGDS-treated cells was examined. I reasoned that if retinoblastoma protein mobility was altered coincident with cdc2 kinase activation, then GRGDS was indeed affecting a Gl/S kinase. The retinoblastoma protein is an endogenous cyclin-dependent kinase substrate whose growth-suppressing function is inhibited by phosphorylation (15). Phosphorylation inhibits the ability of the retinoblastoma protein to bind cyclins and triggers the GI/S transition (16)(17)(18). The retinoblastoma protein does not appear to participate in the G2/M transition. Several studies have shown that retinoblastoma protein phosphorylation reduces its SDS-PAGE mobility (18)(19)(20). Cells were biosynthetically labeled with [RsS]methionine so that both phosphorylated and nonphosphorylated retinoblastoma protein isoforms would be de- These data provide additional indirect evidence that cdc2 activated by GRGDS treatment is acting at the G1/S transition.
In summary, FNR perturbation by GRGDS leads to early activation of a normal cdc2-dependent proliferation pathway in FA-K562 cells. Increased DNA synthesis and soft agar colony formation are subsequently observed. The commitment of FA-K562 cells to proliferate occurs within 2-3 h of the addition of peptide, arguing against a requirement for de novo production of growth factors. Peptide effects on cdc2 activation, retinoblastoma protein phosphorylation, cell cycle progression, and cell proliferation are limited to the FNRbinding peptide, GRGDS. This suggests that FNR perturbation is involved in each of these events. Furthermore, the close temporal association of FNR perturbation, cdc2 activation, and retinoblastoma protein phosphorylation suggests that some type of interaction occurs. Future studies will investigate more proximal signals transduced by FNR occupancy and seek to link these signals to retinoblastoma protein activity.
The data presented here demonstrate that integrin perturbation alone can affect the cell cycle. This is interesting in light of a recent report that cyclin-cyclin-dependent kinase interactions may be affected by suspension versus adherent culture of HeLa cells (21). In suspension cells, cyclin A was associated with cdc2, whereas in adherent HeLa cells, cyclin A was associated with cdk2. This "switching" of cyclin-cyclindependent kinase partners could result from perturbation of integrins that are known to mediate cell-substrate adhesion. Ongoing studies using normal human keratinocytes suggest that these effects are not restricted to cultured or transformed cell lines. It is unclear why only the FNR-overexpressing hematopoietic cell line was growth-stimulated by GRGDS. One potential explanation is that GRGDS displaces a fibro-nectin-associated growth inhibitor, such as fibronectin-transforming growth factor+ (22). In this model, the more FNR expressed by the cell, the higher the local concentration of growth inhibitor. Another possibility is that the parental K562 cell line expresses a mutant form of the retinoblastoma protein whose function is corrected by FNR overexpression. The latter explanation is more likely in view of results obtained using normal human keratinocytes, which express low levels of the a& FNR. It is intriguing, in this context, that FNR overexpression and retinoblastoma protein re-expression similarly inhibit tumor formation in nude mice, colony formation in soft agar, and growth in culture (5, 6, 23). Regardless of the mode of action of GRGDS, this study demonstrates that FNR perturbation can result in mitogenesis. This may explain earlier observations that neoplastic transformation of cells affects cell-surface fibronectin and FNR expression or function (1, 3, 4, 24). It also suggests that one way to control tumor cell growth may be to manipulate FNR ligands presented to these cells.