Fibroblast quiescence in floating or released collagen matrices: contribution of the ERK signaling pathway and actin cytoskeletal organization.

Fibroblasts in attached collagen matrices proliferate, whereas cells in floating or released matrices become quiescent. Cells in attached matrices had prominent actin stress fibers, indicating that they were under isometric tension, whereas stress fibers were absent from fibroblasts in floating or released matrices. Compared with cells in attached matrices, cells in floating or released matrices showed down-regulation of cyclin D1 and up-regulation of p27(Kip1) cyclin-dependent kinase inhibitor, and similar changes occurred after the ERK signaling pathway was blocked by UO126 in cells in attached matrices. A different pattern of changes in cell cycle regulatory proteins occurred, however, after serum deprivation or actin cytoskeletal depolymerization by latrunculin B, which did not prevent signaling through the ERK pathway. Therefore, cell quiescence in floating or released collagen matrices could be explained by decreased signaling through the ERK pathway, but these changes were not accounted for by the absence of isometric tension in the cells.

Proliferation of normal (untransformed) cells typically requires a combination of signals generated by growth factor stimulation and cell adhesion (1)(2)(3)(4)(5), and isometric tension within cells has been implicated in the integration of these signals (1,5). The cellular response to growth factors and adhesion converges on the cyclins, cyclin-dependent kinases (CDKs), 1 and CDK inhibitors (6). D-type cyclins act as cell cycle sensors of growth factor stimulation, and traverse of G 1 depends on the activities of cyclin D-Cdk4/6 and cyclin E-Cdk2 complexes, which function in early and late G 1 , respectively (7,8). In general, growth factor stimulation of quiescent cells leading to proliferation results in up-regulation of cyclin D and down-regulation of the CDK inhibitor p27 Kip1 (4,5,9,10). In addition, quiescent cells usually contain low levels of the CDK inhibitor p21 Cip1 (11)(12)(13), which increase in response to growth factor stimulation and contribute to formation of cyclin-CDK complexes (8). Subsequently, p21 Cip1 levels decline in response to cell adhesion in which case proliferation occurs (11,13).
Signaling through the extracellular signal-regulated kinase (ERK) pathway plays a key role in control of the cell cycle regulatory proteins. Onset of cell proliferation and up-regulation of cyclin D1 both require sustained activation of ERK (14 -16). In addition, cells that are attached but lack isometric tension are unable to up-regulate cyclin D1 (17,18) and cannot translocate activated ERK to the nucleus (19). The interplay between isometric tension and the ERK signaling pathway remains unclear, however, since absence of isometric tension has been reported to block activation of ERK in some reports (17,20) but not in others (18,21).
Most studies on the regulation of cell proliferation have been carried out with cells in monolayer or suspension culture. We have been interested in cell proliferation in the three-dimensional extracellular matrix environment. Fibroblasts cultured within collagen matrices reorganize the matrix into a dense, tissue-like structure (22), a culture system that has been used as an in vitro model of wound contraction (23) and tissue morphogenesis (24). Cells proliferate in attached matrices but become quiescent and apoptotic in floating or released matrices (25)(26)(27). The switch from proliferation to quiescence and apoptosis is a characteristic feature of fibroblasts at the end of cutaneous repair (28).
Previously, we showed that fibroblasts in floating or released collagen matrices lose the ability to signal normally through the ERK pathway (27). We speculated that decreased ERK signaling might account for cell quiescence in floating or released matrices. In addition, we suggested that quiescence and decreased ERK pathway signaling occurred because of the absence of isometric tension in cells in floating or released matrices. In the current report, we have tested these possibilities. To determine if decreased ERK pathway signaling could account for quiescence, we measured cell cycle regulatory proteins in fibroblasts in collagen matrices under proliferating conditions in attached matrices and compared these levels to those found in cells under quiescent conditions in floating or released matrices or in cells in attached matrices in the presence of the ERK pathway inhibitor UO126. In addition, to determine the possible role of isometric tension in quiescence and decreased ERK pathway signaling, we analyzed cell cycle regulatory proteins in fibroblasts in attached matrices in the presence of the actin cytoskeletal depolymerizing agent latrunculin B. We also tested the interrelated effects of UO126 and latrunculin B on actin cytoskeleton organization and signaling through the ERK pathway. Our findings support the idea that decreased signaling through the ERK pathway is responsible for cell quiescence in floating or released collagen matrices, but this decrease could not be accounted for by the loss of isometric tension. Details are reported herein.
Collagen Matrix, Monolayer, and Suspension Culture-Fibroblasts from human foreskin specimens (Ͻ10th passages) were maintained in Falcon 75-cm 2 tissue culture flasks in DMEM supplemented with 10% fetal bovine serum. Collagen matrix cultures were prepared using Vitrogen 100 collagen as described previously (26,27). The cell/collagen mixture containing 10 6 fibroblasts/ml and 1.5 mg/ml collagen in DMEM without serum was prewarmed to 37°C for 3-4 min, after which aliquots (0.2 ml) were polymerized in Corning 24-well culture plates for 60 min at 37°C in a humidified incubator with 5% CO 2 . After polymerization, 1.0 ml of DMEM, 10% fetal bovine serum, and 50 g/ml ascorbic acid was added to each well. Floating matrices were gently released from the underlying culture dish with a spatula immediately after polymerization. Attached matrices were released at the times indicated in the figure legends. Culture medium was changed at 24-h intervals.
DNA Synthesis-Cells in collagen matrices were incubated in 10% fetal bovine serum DMEM containing 5 Ci/ml [ 3 H]thymidine (specific activity 5 Ci/mmol) for 1 h at 37°C in a humidified incubator with 5% CO 2 and then washed three times for 5 min with phosphate-buffered saline (PBS) (150 mM NaCl, 3 mM KCl, 6 mM Na 2 HPO 4 , 1 mM KH 2 PO 4 , pH 7.2). Subsequently, the cells were harvested from matrices (26), and DNA was precipitated by incubating the cells with 10% trichloroacetic acid in PBS containing 125 g/ml bovine serum albumin for 20 min at 4°C. Precipitates were collected on Whatman 2.5-cm glass microfiber filters, washed with 10% trichloroacetic acid and 5% trichloroacetic acid, and transferred to scintillation vials containing 10 ml of Budget Solve. Radioactivity was counted using a Beckman scintillation counter (LS 6000 SC). Radioactive counts were adjusted for equal cell numbers by measuring lactate dehydrogenase activity of an aliquot of the harvested cells with the lactate dehydrogenase diagnostic kit (Sigma). All experiments were carried out in duplicate, and each experiment was carried out at least two times. Data points and error bars in the figures represent averages and S.D. values.
Actin Distribution-Collagen matrices were fixed for 30 min at 22°C with 3% paraformaldehyde in DPBS (150 mM NaCl, 3 mM KCl, 6 mM Na 2 HPO 4 , 1 mM KH 2 PO 4, 1 mM CaCl 2 , 0.5 mM MgCl 2 , pH 7.2), washed for 2 ϫ 10 min with DPBS, blocked for 30 min with DPBS containing 2% bovine serum albumin and 1% glycine, and then permeabilized for 10 min with DPBS containing 0.5% Triton X-100. Subsequently, the samples were incubated with RITC-conjugated phalloidin (8 units/ml) for 30 min at room temperature followed by six washes with DPBS. After mounting the samples on glass slides with Fluoromount G, observations were made using a Nikon Elipse 400 Fluorescent Microscope, and digital images were collected using a Photometrics SenSys camera and MetaView. Fig. 1 shows that cellular DNA synthesis (measured by [ 3 H]thymidine incorporation) was low in fibroblasts in floating (F) compared with attached (A) collagen matrices and declined markedly when attached collagen matrices (A) were released (R). Fig. 2A shows that fibroblasts in attached collagen matrices developed bipolar morphology and contained prominent actin stress fibers, indicating the presence of isometric tension. Fig. 2B, on the other hand, shows that cells in floating collagen matrices had rounded cell bodies with numerous fine processes and diffuse staining of the actin cytoskeleton, indicating an absence of isometric tension. After the release of attached collagen matrices, fibroblasts lost their actin stress fibers (not shown).

DNA Synthesis and Cytoskeletal Organization-
Cell Cycle Regulatory Proteins in Mechanically Loaded and Unloaded Fibroblasts-The above experiments confirmed that cell quiescence occurred in floating and released collagen matrices in which the cells lacked actin stress fibers although the cells were attached with projections extending into the matrix. Studies then were carried out to determine the levels of cell cycle regulatory proteins in fibroblasts in the collagen matrices. Fig. 3A shows typical results in which Western blotting was used to measure the levels of cyclin D1, p21 Cip1 , p27 Kip1 , and Cdk4. Fig. 3B shows data from three separate experiments averaged using Cdk4 levels to normalize the results. Cdk4 is typically used as a loading control in studies on cell cycle regulatory proteins because Cdk4 levels appear to be the same in proliferating and quiescent cells (29).
A similar pattern of differences was detected regardless whether the cells were switched from attached to released matrices or cultured in floating matrices. In either case, compared with proliferating cells, fibroblasts that became quiescent in the collagen matrices had lower levels of cyclin D1 and p21 Cip1 and increased levels of p27 Kip1 .
Cell Quiescence and Changes in Cell Cycle Regulatory Proteins in Response to Decreased Signaling through the ERK Pathway and Disruption of Actin Stress Fibers-The above results demonstrated that quiescence of fibroblasts in released or floating collagen matrices was accompanied by down-regulation of cyclin D1 and p21 Cip1 and up-regulation of p27 Kip1 . If cell quiescence in floating or released matrices depended upon decreased signaling through the ERK pathway, then we predicted that a pharmacologic block of signaling through the ERK pathway in cells in attached matrices should result in cell quiescence and changes in the levels of cell cycle regulatory proteins similar to those in released or floating matrices.
To test this prediction, we measured the effects on cell proliferation and cell cycle regulatory proteins following 24-h treatment with the mitogen-activated protein kinase/ERK kinase inhibitor U0126 to block the ERK pathway (30). Fig. 4 shows that proliferating fibroblasts in attached collagen matrices became quiescent after 24 h in the presence of U0126 (ϩUO). Inhibition was completely reversible, and DNA synthesis returned to normal levels after a 24-h wash-out period. Fig. 5A shows typical results of experiments to determine the levels of cyclin D1, p21 Cip1 , p27 Kip1 , and Cdk4 in control and treated cells. Fig. 5B shows data from four separate experiments averaged and normalized to Cdk4 levels. Treatment of cells with U0126 to block the ERK signaling pathway resulted in down-regulation of cyclin D1 and up-regulation of p27 Kip1 similar to fibroblasts in floating or released collagen matrices. These findings were consistent with the idea that cell quiescence in floating or released matrices depended upon decreased signaling through the ERK pathway.
If loss of signaling through the ERK pathway resulted from the absence of isometric tension, then disruption of stress fibers by depolymerizing the actin cytoskeleton should have similar consequences for cell cycle regulatory proteins as culturing cells in released or floating matrices. Therefore, we measured the effects on cell proliferation and cell cycle regulatory proteins following 24-h treatment with latrunculin B to depolymerize the actin cytoskeleton (31). As an additional control, we also tested the consequences of serum deprivation, which was expected to result in cell quiescence without affecting isometric tension. Fig. 4 shows that proliferating fibroblasts in attached collagen matrices also became quiescent after 24 h in serum-free medium (ϪSer) or in the presence of latrunculin B (ϩLB), and these effects also were reversible. Cytochalasin D treatment to disrupt actin stress fibers also was tested, since most research on the role of isometric tension in cell proliferation in monolayer culture has utilized cytochalasin D. While proliferation of cells in collagen matrices was blocked by cytochalasin D, the effect was only partly reversible (not shown) and therefore not studied further. Fig. 5, A and B, show that treatment of cells with latrunculin B caused down-regulation of cyclin D1, p27 Kip1 , and p21 Cip1 . Serum withdrawal, on the other hand, had no effect on cyclin D1 or p27 Kip1 , although a decrease in p21 Cip1 was observed. Given the different profiles of cell cycle regulatory proteins in cells in released or floating matrices or after inhibition of the ERK signaling pathway compared with depolymerization of the actin cytoskeleton or serum withdrawal, it appeared that fibroblast quiescence caused by the latter two conditions depended on different mechanisms than those involved in cells in released or floating matrices or cells in attached matrices with blocked ERK pathway signaling.

Changes in Actin Stress Fibers and Cell Signaling in Response to Growth Factor Deprivation, Blocked ERK Pathway
Signaling, and Actin Cytoskeleton Depolymerization-The above findings were consistent with the possibility that decreased ERK pathway signaling was responsible for cell quies-cence in floating or released collagen matrices. On the other hand, our findings did not support the idea that decreased ERK pathway signaling depended on the absence of stress fibers. To examine further possible interrelationships between these conditions, we tested the effects of the conditions leading to cell quiescence (serum deprivation, blocking ERK pathway signaling, and actin cytoskeleton depolymerization) on actin cytoskeletal organization and ERK pathway signaling. ERK pathway signaling was stimulated using PDGF.   6 shows that neither serum deprivation (Fig. 6B) nor the addition of U0126 (Fig. 6C) had a detectable effect on cell shape or actin stress fibers, which were reversibly disrupted by latrunculin B (Fig. 6, D and E). Conversely, Fig. 7 shows that neither serum deprivation nor latrunculin B blocked ERK activation after PDGF stimulation, which was completely prevented by U0126. In addition, as shown in Fig. 8, the time course of ERK activation of fibroblasts in attached matrices following PDGF stimulation was similar in the presence or absence of latrunculin B but markedly decreased in cells in released matrices. Taken together, these findings provided additional evidence that cell quiescence caused by decreased ERK pathway signaling, actin stress fiber disruption, and serum deprivation occurred independently of each other. DISCUSSION Fibroblasts cultured in floating or released collagen matrices become quiescent and apoptotic (25,26), similar to fibroblastic cells at the end of cutaneous repair (28). Previously, we proposed that the mechanism accounting for cell quiescence in floating or released matrices depended upon decreased signaling through the ERK pathway (27). We also suggested that this decrease resulted from the absence of isometric tension in the cells, since cells in floating or released matrices lacked actin stress fibers, which were prominent in cells in attached matrices. The present findings support the idea that decreased signaling through the ERK pathway is responsible for cell quiescence in floating or released collagen matrices, but this decrease could not be accounted for by the loss of isometric tension.
Consistent with earlier studies, we found that fibroblasts in attached collagen matrices proliferated, whereas the cells in floating or released collagen matrices became quiescent. Also, fibroblasts in attached matrices had prominent actin stress fibers, indicating that they were under isometric tension, whereas stress fibers were absent from fibroblasts in floating or released matrices. Compared with proliferating fibroblasts in attached matrices, cells in floating matrices had lower levels of cyclin D1 and p21 Cip1 and elevated levels of p27 Kip1 . Similar changes in the levels of cell cycle regulatory proteins occurred after attached matrices were released. Down-regulation of cyclin D1 and up-regulation of p27 Kip1 probably contribute to cell cycle arrest and quiescence of fibroblasts in floating or released collagen matrices based on monolayer culture studies (4,5,9,10). While cell contact with fibrillar collagen itself was reported to cause an increase in p27 Kip1 (32,33), the cells in those previous studies were spread poorly. As a result, altered cell spreading rather than interaction with fibrillar collagen could have been responsible for the increased levels of p27 Kip1 .
The observation of higher levels of p21 Cip1 in fibroblasts in attached compared with floating or released collagen matrices suggests that p21 Cip1 does not contribute to cell cycle arrest under the latter conditions. In monolayer culture, the level of p21 Cip1 has been shown to be regulated by the timing of growth factor stimulation and cell adhesion; i.e. quiescent cells begin with low levels of p21 Cip1 , which increase in response to growth factor stimulation and then decrease in response to cell adhesion and integrin engagement through a Rho-dependent mechanism (10 -13). In the absence of integrin engagement, such as with cells in suspension, down-regulation of p21 does not occur, and the cells are unable to enter the cell cycle (11,13). If p21 Cip1 had contributed to cell cycle arrest in floating or released collagen matrices, then we would have expected to observe an increase in p21 Cip1 levels compared with cells in attached matrices. Why p21 Cip1 is actually higher in fibroblasts in attached compared with floating or released collagen matrices remains to be determined but might relate to the pattern of integrin occupancy in collagen matrices; i.e. integrin ␣ 2 ␤ 1 is required for matrix contraction by cells (34 -36) and is probably engaged in all types of matrices, whereas ␣ 5 ␤ 1 plays a role in fibronectin assembly within the collagen matrix and is engaged by cells only when the matrices are attached (37,38).
When fibroblasts in attached collagen matrices were serumdeprived or treated to block signaling through the ERK pathway or depolymerize the actin cytoskeleton, cellular DNA synthesis was reversibly inhibited. Blocking ERK pathway signaling led to a decrease in cyclin D1 and increase in p27 Kip1 similar to that which occurred in floating or released matrices. An increase in p27 Kip1 caused by blocked ERK pathway signaling is consistent with previous studies implicating ERK in the regulation of p27 Kip1 (39).
Depolymerizing the actin cytoskeleton, on the other hand, resulted in decreased levels of both p27 Kip1 and cyclin D1. Our results differ from those reported for endothelial cells in monolayer culture in which treatment with cytochalasin resulted in an increase in p27 Kip1 (18). The explanation for this is unknown and could relate to the culture conditions, cell type, or actin cytoskeleton disrupting agents used.
Taken together, the foregoing findings support the idea that cell quiescence in floating or released matrices depended upon decreased signaling through the ERK pathway but argued against our previous idea that the absence of isometric tension in cells in floating or released matrices accounted for the changes in the ERK signaling pathway and cell proliferation. The latter point was tested directly in the experiments showing the independence of the ERK signaling pathway in relation to FIG. 8. Time course of ERK stimulation in fibroblasts in attached collagen matrices subjected to treatment with latrunculin B and in released matrices. Fibroblasts were cultured in attached collagen gels for 24 h and then for 24 h in control medium (Con) or control medium containing 1 M latrunculin B or released as indicated. At the end of the culture period, 50 ng/ml PDGF was added for the time periods designated. Subsequently, extracts of samples were subjected to SDS-PAGE and immunoblotted with antibodies against ERK and phospho-ERK.

FIG. 7. Levels of ERK and phospho-ERK in fibroblasts in at-
tached collagen matrices subjected to serum deprivation or treatment with UO126 or latrunculin B. Fibroblasts were cultured in attached collagen gels for 24 h and then for 24 h in control medium (Con), serum-free medium (ϪSer), or control medium containing 100 M UO126 (UO) or 1 M latrunculin B as indicated. At the end of the culture period, 50 ng/ml PDGF was added for 15 min. Subsequently, extracts of samples were subjected to SDS-PAGE and immunoblotted with antibodies against ERK and phospho-ERK. actin stress fibers in fibroblasts in attached collagen matrices; i.e. blocking ERK pathway signaling had no effect on the actin cytoskeleton, and depolymerizing the actin cytoskeleton had no effect on ERK pathway signaling. Down-regulation of cyclin D by depolymerizing the actin cytoskeleton may depend on the ERK-independent mechanism that utilizes the phosphatidylinositol 3-kinase/AKT pathway (40,41).
The observation that loss of actin stress fibers could not account for differences in ERK signaling and cell cycle regulatory proteins in fibroblasts in released or floating versus attached collagen matrices was unexpected given the importance of isometric tension for regulation of proliferation of cells in monolayer culture (42,43). Tension often is a reflection of cell geometry, and tension and geometry varied differently in cells in floating or released collagen matrices compared with cells in attached matrices after actin cytoskeletal depolymerization. In either case, actin stress fibers were absent, but cells in floating or released matrices were rounded with thin extensions, whereas cells in attached matrices treated with latrunculin B retained their elongated, bipolar shape.
The mechanism that accounts for decreased ERK pathway signaling in released or floating matrices presents an important problem for future investigation. A variety of mechanisms are possible including physical disorganization of signaling scaffolds (44,45) or changes in cell surface signaling centers such as caveolae (46 -49).
Finally, it is important to note the novel observation that serum deprivation caused proliferating cells to become quiescent without affecting cellular levels of cyclin D1. This was surprising, since cells in monolayer culture show a close linkage between the presence of serum and cyclin D expression (7). Presumably, quiescence here is regulated by signaling pathways parallel to cyclin D1 (4). The difference in response to serum deprivation emphasizes that cell regulation in monolayer culture and the three-dimensional matrix environment differ in some respects. An attractive hypothesis for future study is that growth factor binding to the matrix occurs (50) and serves as a reservoir to stimulate cells even after removal of serum, which would not be possible in monolayer culture.
In summary, our studies suggest that down-regulation of cyclin D1 and up-regulation of p27 Kip1 are key regulatory events linked to mechanical unloading of fibroblasts in released or floating collagen matrices. The findings support the idea that decreased signaling through the ERK pathway is responsible for cell quiescence in floating or released collagen matrices, but this decrease could not be accounted for by the loss of isometric tension.