Fibroblast Quiescence and the Disruption of ERK Signaling in Mechanically Unloaded Collagen Matrices*

Fibroblasts in mechanically unloaded collagen matrices had low levels of DNA synthesis compared with cells in mechanically loaded matrices. Under the former conditions, the cellular ERK signaling pathway appeared to be disrupted. Also, pharmacologic inhibition of ERK signaling blocked DNA synthesis by fibroblasts in mechanically loaded matrices. These results were consistent with the idea that mechanoregulation of fibroblast DNA synthesis in collagen matrices occurs at the level of the ERK signaling pathway. During wound repair, fibroblasts go though a sequence of cell proliferation, quiescence, and apoptosis (1–4). To study these processes in vitro , we have been studying growth regulation of fibroblasts cultured in three-dimensional collagen matrices (5). The collagen matrix model provides a unique opportunity to study the transition from cell proliferation to quiescence with-out pharmacologic intervention or removal of soluble growth factors. and then permeabilized for 10 min with DPBS containing 0.5% Triton X-100. Subsequently, the sam- ples were incubated with rhodamine isothiocyanate-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 Fluoro-mount G, observations and photographs were made under a Bio-Rad MRC 1024 Laser Sharp Confocal microscope.

Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) and Southern Blotting-RNA was isolated as described previously (21). Briefly, collagen matrices (one/sample) were dissolved using Solution D, and total RNA was extracted with phenol:chloroform:isoamyl alcohol and precipitated with isopropanol and 20 g of glycogen (as carrier) (20).
RNA precipitates were redissolved in 10 l of autoclaved water. First-strand cDNA synthesis reaction with Moloney murine leukemia virus reverse transcriptase (M-MLV RT, Life Technologies, Inc.) using oligo(dT) (15) primer was carried out according to the manufacturer's protocol in a final volume of 20 l. After incubation for 3 h at 37°C, M-MLV RT enzyme was heat-inactivated by incubation at 70°C for 10 min.
One l of each cDNA preparation was used in PCR reactions using c-fos and GAPDH primers (see below). DNA was denatured at 94°C for 4 min followed by 5 cycles of the following: 30 s denaturation at 94°C, 30 s annealing at 57°C, 60 s elongation at 72°C, and a final elongation phase for 10 min at 72°C. Reaction mixtures (12 l) contained 20 mM Tris-HCl (pH 8.4), 0.05 units/l Taq polymerase, 50 mM KC1, 1.5 mM MgCl 2 , 200 M dNTPs, and 1.0 M of each primer.
After PCR, 7.5 l was removed from each sample, and 1.5 l of 6ϫ loading buffer II (23) was added and the mixture loaded onto 2% agarose gels containing 1ϫ Tris-acetate-EDTA (23). Samples were subjected to electrophoresis for 30 min at 100 V. The gel was then soaked in 0.5 M NaOH, 1.5 M NaCl for 45 min to denature the DNA and then pH-neutralized in 20ϫ saline/sodium phosphate/EDTA. After transfer, DNA was cross-linked with a Stratalinker UV light source (Stratagene) at a setting of 1,200 J of radiation/cm 2 . c-fos and GAPDH PCR products were then detected under aqueous conditions at 65°C according to standard protocols (23). DNA probes for hybridization were generated with a Roche Molecular Biochemicals random-primed labeling kit. Gelpurified PCR products using c-fos and GAPDH primers served as templates. c-fos primers were synthesized by the University of Texas Southwestern Molecular Biology Core Facility as follows: 5Ј-atgatgttctcgggcttcaacgcagcag-3Ј and 5Ј-tctggagataactgttccaccttg-3Ј. GAPDH primers were purchased from CLONTECH.
Actin Distribution-Collagen matrices were fixed for 30 min at 22°C with 3% paraformaldehyde in DPBS, washed 2 ϫ 10 min with DPBS (1 mM CaCl 2 , 0.5 mM MgCl 2 , 150 mM NaCl, 3 mM KCl, 6 mM Na 2 HPO 4 , 1 mM KH 2 PO 4, pH 7.2), 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 rhodamine isothiocyanate-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 and photographs were made under a Bio-Rad MRC 1024 Laser Sharp Confocal microscope. Fig. 1A shows that in mechanically loaded (i.e. attached) collagen matrix cultures, the rate of total DNA synthesis increased over 72 h. When matrices that had been attached for 48 h were released for 4 h, DNA synthesis in the cultures was unchanged, as shown in Fig.  1B (A48,R4). By 10 h, however, DNA synthesis was reduced more than 50% compared with attached samples (A48,R10). Finally, Fig. 1C shows that total DNA synthesis in cultures that were floating for 48 h or attached for 24 h and then released for 24 h, amounted to less than ϳ10% of that observed in attached matrix cultures. It should be noted that the total number of cells that can be recovered from attached versus released collagen matrices remains constant for several days after release (10). The foregoing results confirmed that DNA synthesis is markedly reduced in mechanically unloaded colla-gen matrices compared with mechanically loaded matrices (see the Introduction).

Differences in DNA Synthesis by Fibroblasts in Attached, Floating, and Released Collagen Matrices-
Inhibition of the ERK Signaling Pathway in Collagen Matrices Switched from Mechanically Loaded to Unloaded Conditions-Cessation of DNA synthesis after loss of cell anchorage in monolayer culture has been suggested to result, at least in part, from the inability of cells to phosphorylate ERK in response to growth factor stimulation (24 -26). Consequently, studies were carried out to examine whether fibroblasts in released collagen matrices were altered in the ERK signaling pathway similar to fibroblasts lacking suitable anchorage. Fig. 2 shows that cells cultured in attached collagen matrices for 52 h had low levels of active ERK. The addition of serum to these cells for 15 or 40 min resulted in an acute ERK phosphorylation response detected by immunoblotting with antibodies specific for the phosphorylated form of ERK. If, however, attached matrices were released 4 h before serum was added, then the ability of serum to stimulate ERK phosphorylation was reduced. Therefore, after mechanical unloading, the sig- Fibroblast Quiescence and ERK Signaling naling pathway leading from serum to ERK phosphorylation lost the ability to be acutely activated.
Some cell surface growth factor receptors show decreased signaling capacity in mechanically unloaded collagen matrices (27,28). Therefore, further experiments were carried out to examine the ERK signaling pathway downstream of cell surface receptors by using phorbol ester (TPA) to activate the ERK pathway pharmacologically (29,30). Fig. 3 shows that the addition of phorbol ester to fibroblasts cultured in attached matrices for 52 h resulted in acute ERK phosphorylation detected after 15 or 45 min. The ERK response to phorbol ester was decreased, however, if the attached matrices had been released for 4 h before the addition of TPA.
Stimulation of ERK by phorbol ester causes downstream activation of c-fos transcription (31)(32)(33). Therefore, additional experiments were carried out to learn whether there was any difference in TPA stimulation of c-fos transcription in cells in attached versus released matrices. Fig. 4 shows that an acute c-fos transcriptional response could be detected by RT-PCR when fibroblasts in attached matrices were treated with phorbol ester. In marked contrast, cells in matrices that had been released for 4 h lost their c-fos response. Cells in released matrices retained the ability to up-regulate c-fos mRNA transcription, however, in response to microwounding the cells by tearing the matrices in half.
The foregoing findings demonstrated that mechanical unloading of collagen matrices resulted in disruption of the ERK signaling pathway. In other experiments, we measured the ability of TPA to acutely stimulate MEK, the protein kinase that activates ERK (34). Fig. 5 shows that MEK was activated when cells in attached matrices were stimulated with phorbol ester but that the extent of activation was reduced after the matrices were released, which suggested that in released matrices the ERK signaling pathway was blocked at some point upstream of MEK.
Inhibition of the ERK Signaling Pathway in Floating Collagen Matrices-In the preceding section, ERK phosphorylation by serum and phorbol ester was examined in cells before and after mechanical unloading (i.e. attached and released matrices). In related experiments, we studied stimulation of ERK phosphorylation in fibroblasts in floating collagen matrices, which also have low levels of DNA synthesis (Fig. 1C). Fig. 6 shows that phorbol treatment of cells cultured for 24 h in attached collagen matrices resulted in acute ERK phosphorylation (see also Fig. 2). If, however, the cells had been cultured overnight in floating collagen matrices, then little ERK phosphorylation occurred in response to phorbol ester. Therefore, the ERK signaling pathway appeared to be disrupted in fibroblasts in floating collagen matrices as well as in matrices that were attached and released.  2. ERK phosphorylation response to serum in fibroblasts cultured in attached and attached-released matrices. Cells were cultured in attached matrices for 24 h followed by culture in serum-free media for 24 h, and then matrices were attached (A) or released (R) for 4 h. Subsequently, medium containing 10% FBS was added to the cultures for the indicated times, after which cell extracts were prepared and analyzed to detect phospho-ERK or total ERK.

Fibroblast Quiescence and ERK Signaling
fibroblasts in collagen matrices, additional experiments were carried out using the ERK signaling pathway inhibitor, PD98059 (35). Fig. 7 shows that the addition of PD98059 for 1 h resulted in marked inhibition of DNA synthesis, measured in 48-and 72-h attached collagen matrix cultures. The effect of the inhibitor was completely reversible after a 24-h washout (W 24, Fig. 7), however. It could be concluded, therefore, that blocking the ERK signaling pathway in these cells was sufficient to block DNA synthesis.

Differences in Cell Spreading and Actin Cytoskeletal Organization of Fibroblasts in Attached, Floating, and Released Collagen
Matrices-Most studies on the anchorage dependence of fibroblasts in monolayer culture have utilized poorly adhesive substrata or suspension cultures, resulting in poor cell adhesion and spreading, from which comes the original idea of anchorage dependence (36 -38). By contrast, fibroblasts in collagen matrices are surrounded by the substrata. Consequently, it was of interest to compare cell shape and actin cytoskeletal organization of fibroblasts under proliferative and quiescent conditions. Fig. 8a shows by confocal microscopy that fibroblasts cultured for 48 h in attached collagen matrices were elongated and under a mechanical load as indicated by the prominent stress fibers. Cells that were attached for 24 h and then released for an additional 24 h (Fig. 8b) lost their stress fibers but remained spread in an elongated shape. Fibroblasts cultured for 48 h in floating matrices (Fig. 8c) also lacked actin stress fibers but had stellate morphology with round cell bodies and long pseudopodia. These results show that fibroblasts in collagen matrices are attached and well spread under all conditions. There was no correlation between DNA synthesis and cell spreading per se, but only cells under a mechanical load had actin stress fibers.

DISCUSSION
The collagen matrix model provides a unique opportunity to study the transition from cell proliferation to quiescence without pharmacologic intervention or removal of soluble growth factors. In the current studies, we analyzed ERK signaling after switching fibroblasts from mechanically loaded to unloaded conditions. Based on the decrease in acute ERK phosphorylation in response to serum or phorbol ester and on the inability of phorbol ester to acutely stimulate c-fos transcription, it could be concluded that the ERK signaling pathway was disrupted in fibroblasts after mechanical unloading.
In collagen matrices switched from mechanically loaded to unloaded conditions, ERK signaling was disrupted after 4 h. DNA synthesis was reduced more than 50% after 10 h and 90% after 24 h. ERK signaling also was disrupted in fibroblasts in floating collagen matrices where only low levels of DNA synthesis occurred. Moreover, inhibition of the ERK signaling pathway by the inhibitor PD98059 reversibly inhibited DNA synthesis in fibroblasts in mechanically loaded collagen matrices (see also Ref. 39). Taken together, these studies are consistent with the idea that disruption of the ERK signaling pathway contributes to the low levels of DNA synthesis by fibroblasts in mechanically unloaded matrices.
The mechanism responsible for loss of ERK signaling in mechanically unloaded matrices remains to be determined but probably occurs upstream of the ERK activator MEK because MEK responded to mechanical unloading similarly to ERK. Receptor tyrosine kinases could be involved because plateletderived growth factor receptors have been reported activated by increased mechanical stress (40) but desensitized after mechanical unloading (27,28). In the current experiments, however, the ERK signaling pathway in mechanically unloaded fibroblasts did not function normally even when tested with phorbol ester, which stimulates the signaling pathway downstream of cell surface receptors (29,30,41). Therefore, in addition to receptor activity, loss of signaling likely depends on downstream factors. Physical organization of the signaling pathway itself also could be important. Scaffolding proteins are required for proper functioning of mitogen-activated protein kinase signaling modules (42). In endothelial cells, caveolae have been reported to play a role in organizing the signaling molecules leading to ERK activation, and mechanoregulated signaling can be blocked by disrupting caveolae (43). Finally, mechanostimulation has been shown to cause Shc association with integrins, which is followed by recruitment of the Grb2-Sos complex leading to ERK signaling (44). Therefore, mechanical unloading could alter the association of Shc or other adapter proteins with the cell surface matrix receptors.
Whatever the precise mechanism accounting for the loss of ERK signaling, it is interesting that this loss and concomitant cell quiescence in mechanically unloaded collagen matrices resembles the regulation of cells in monolayer cultures by anchorage dependence. Although it has been possible through the use of micropatterned substrata to make a distinction between the effects of cell tension versus shape on cell proliferation (45), such studies cannot replicate the situation of cells engaged in three-dimensional adhesive interactions over their entire surfaces such as occurs in tissue. Use of a three-dimensional culture system such as the collagen matrix model provides such an opportunity. In mechanically unloaded collagen matrices, fibroblasts were attached and well spread but lacked stressed fibers, which indicated an absence of isometric tension. Moreover, fibroblasts in collagen matrices could not organize and maintain a fibronectin matrix in the absence of mechanical loading (7,46). Therefore, it appears that the interplay between the features of cytoskeletal tension and growth in the three-dimensional environment is similar to that reported for cells in monolayer culture (14 -17).
Finally, the current studies offer a novel mechanism to account for cell growth regulation at the end of wound repair. Before wounding, tissue fibroblasts are stellate, quiescent, and sessile. After wounding, a highly cellular wound tissue develops through the migration and proliferation of fibroblastic cells from the wound margins (1,2,4). These cells differentiate into myofibroblasts that are under a mechanical load (isometric tension) as shown by their elongated shape, prominent stress fibers, and fibronexus junctions (3,47). After the wound defect has been replaced through a combination of cellular contractile activity and extracellular matrix biosynthesis, the tissue returns to a mechanically unloaded state, and myofibroblast regression occurs through apoptosis (3). Based on our experiments with fibroblasts in collagen matrices, we speculate that quiescence at the end of wound repair depends on mechanically regulated disruption of the ERK signaling pathway, which leads to cessation of DNA synthesis even in the continued presence of soluble growth factors in the wound environment.