Extracellular Matrix and Radiation G₁ Cell Cycle Arrest in Human Fibroblasts¹

Two temporally discrete cell cycle delays occur in cells in response to radiation: (a) one is late in G₁ phase, before DNA replication; and (b) the other is in G₂ phase, before mitosis. It is commonly proposed that these cell cycle delays provide cells with additional time to repair DNA damage and thereby promote genomic stability (1). A cell must have an intact p53 pathway to delay in G₁ phase after irradiation (2). Irradiation of p53-positive cells results in increased p53 protein amounts, which subsequently cause an increase in p21^Cip1 mRNA and protein. The p21^Cip1 protein then mediates the G₁ arrest by inhibiting CDK2³/cyclin E kinase activity, which retains the retinoblastoma protein in the hypophosphorylated form (3). p53 is required to sustain a G₂ arrest after high doses of radiation but not to initiate a G₂ delay (4). Rather, the predominate mechanism of G₂ delay is by inhibiting cdc2/cyclin B kinase activity because of phosphorylation of tyrosine 15 on cdc2. This occurs through activation of wee1 kinase and inhibition of cdc25C phosphatase by chk1 kinase and 14-3-3 sequestering protein (reviewed in Ref. 5).

Until recently, it was thought that sublethal doses of radiation cause cells to pause in either G₁ or G₂ phase, and that then cells with repaired DNA damage reenter the cell cycle. This is largely true with the G₂ delay. However, a variation of this theme has emerged when considering G₁ delays in anchorage-dependent, primary human fibroblasts. It has now been observed that human fibroblasts that delay in G₁ phase immediately after irradiation do not reenter the cell cycle (6, 7). Also, human fibroblasts, which are beyond the G₁ checkpoint, delay in G₂ phase but then divide and rearrest indefinitely in the next G₁ phase. The prolonged G₁ arrest occurs even at doses of 1 Gy, at which there is very little residual DNA damage. This arrest was called senescence-like because it was thought to be irreversible, because of morphological similarities to senescent cells, and because cellular p21^Cip1 amounts remain elevated (6). More recently, we have shown that the “permanent” radiation G₁ arrest in human fibroblasts is reversible in many of the cells, if they are subcultured (8). These results indicate that cell adhesion plays a role in maintaining the arrest.

Cell adhesion and the ECM are involved in signal transduction during a variety of cellular functions such as cell motility, morphogenesis, differentiation, and proliferation (reviewed in Refs. 9, 10, 11). G₁ phase CDKs are regulated by cell adhesion in primary, anchorage-dependent fibroblasts (reviewed in Ref. 11). Mitogenic stimulation of suspended, quiescent fibroblasts causes G₀ to G₁ progression; however, the cells arrest in G₁ phase because both CDK4/cyclin D and CDK2/cyclin E kinases remain inactive. Fibroblasts that are put into suspension in late G₁ phase contain hyperphosphorylated retinoblastoma protein and are resistant to cell cycle inhibition (12). These cells progress through the cell cycle and arrest in the next G₁ phase after one cell division.

Radiation and suspension of anchorage-dependent fibroblasts cause G₁ arrest because of the inhibition of G₁ CDKs. Indeed, our data indicate that cell adhesion affects the ability of fibroblasts to undergo and maintain the radiation G₁ arrest (8). A major mode of cell adhesion is via the interaction of integrins with ECM ligands, e.g., fibronectins, collagens, and laminins. Integrins form a family of heterodimeric cell surface receptors composed of α/β subunits (reviewed in Ref. 13). The identity of the α/β subunits determines the ECM ligand specificity. Particular integrin/ECM interactions lead to distinct cellular responses, such as cell proliferation and differentiation (Refs. 14 and 15; reviewed in Ref. 16).

Our studies show that disrupting cell adhesion affects the longevity of the radiation G₁ arrest in human fibroblasts; hence, we hypothesize that changing the type of adhesion interactions will alter or alleviate the arrest. To address this proposal, we measured cell cycle delays after γ-irradiation of human fibroblasts on the substrates polystyrene, Mylar, collagen IV, fibronectin, and laminin. We found dramatic differences in the percentage of cells that undergo a sustained G₁ arrest after irradiation on the different substrates. Furthermore, substrates that promote a prolonged radiation G₁ arrest also cause an increase in SM α-actin expression in the arrested cells, which is characteristic of the myofibroblast phenotype (17). These results indicate that specific types of adhesion interactions contribute to the maintenance of the G₁ arrest observed after irradiation of human fibroblasts and that the arrest culminates in the differentiation of fibroblasts to myofibroblasts, rather than senescence.

HSF cells were derived from neonatal foreskin and cultured at 37°C under 5% CO₂:air in α-MEM supplemented with 10% fetal bovine serum, 100 μg/ml streptomycin sulfate, and 100 units/ml penicillin. Mycoplasma-free cells from passage 8–12 were used. Polystyrene culture flasks were purchased from Costar (Cambridge, MA). Mylar culture dishes were custom-fabricated as described in previous studies (7). Human collagen IV, murine laminin, and human fibronectin culture flasks were obtained from BioCoat Cellware (Becton Dickinson Labware, Bedford, MA).

For γ-ray exposures, cells were seeded at an initial density of 5.3–6.7 × 10³ cells/cm² 24 h before irradiation. A Mark I model 68A high-dose rate ¹³⁷Cs source chamber irradiator (J. L. Shepherd and Associates) was used to irradiate cells with 2 Gy of γ-rays at a dose rate of 2.1 Gy/min at room temperature. BrdUrd (Sigma, St. Louis, MO) was added directly to the culture medium at a final concentration of 10 μm at the indicated times after irradiation.

Flow cytometric analysis was performed as described previously (7). Cells were harvested using 0.25% trypsin/1 mm EDTA, fixed in 70% ethanol for at least 1 h, and then stained overnight at 4°C in PBS containing 0.5 μg/ml HO (Calbiochem, San Diego, CA), 5 μg/ml MI (Pfizer Corp., Groton, CT), and 5 mm MgCl₂ at a cell concentration of 7.5 × 10⁵ cells/ml. A two-laser flow cytometer was used for cell cycle analysis, with one laser operating in the UV region (333.6–363.8 nm), and one tuned to 547.9 nm. The lasers were separated by 250 μm to provide sequential excitation and analysis of each fluorochrome. HO emissions were measured over a 400–495-nm range, and MI fluorescence was measured above 495 nm. The electronic gains were adjusted so that the BrdUrd-negative G₁ peaks of the HO and MI DNA content histograms were initially in the same channel number, and the fluorescence signals were then compared electronically on a cell-cell basis. Such comparisons reflect the quenching of the HO fluorescence by BrdUrd, which is proportional to cellular BrdUrd content in cells that had synthesized DNA during the labeling period.

Cells were harvested using 0.25% trypsin/1 mm EDTA and washed with PBS, and cell pellets were stored at −70°C. Protein lysates were prepared as described previously (7). Protein concentration was determined using the D_C protein assay (Bio-Rad, Hercules, CA), and equal amounts of protein (8–10 μg) were resolved on 10% or 12% SDS polyacrylamide gels and transferred to nitrocellulose membranes. Equal loading and transfer were ensured by Ponceau S staining of the membranes. Specific proteins were detected using a mouse monoclonal antibody to p53 (DO-1); rabbit polyclonal antibodies to p21^Cip1 (C-19), CDK2 (M2), or CDK4 (C-22; Santa Cruz Biotechnology, Inc., Santa Cruz, CA); or a mouse monoclonal antibody to SM α-actin (Klon 1A4; DAKO, Carpinteria, CA). Horseradish peroxidase-linked secondary antibodies (Amersham) and Renaissance chemiluminescence reagents (DuPont New England Nuclear, Boston, MA) were used to visualize proteins.

Several studies show that ionizing radiation causes normal human fibroblasts to accumulate in the G₁ phase of the cell cycle in an apparently permanent arrest (6, 7). However, we have shown that the arrest is reversible in many cells if cell adhesion interactions are interrupted by subculturing the radiation-arrested cells (8). If cell adhesion influences radiation arrest in human fibroblasts, then particular integrin interactions with certain ECM ligands might lead to varied cell cycle responses to radiation. In an initial attempt to test this hypothesis, we compared the radiation arrest we have observed on tissue culture polystyrene to radiation arrest in fibroblasts grown on Mylar. We chose to test Mylar because it is a commonly used culture substrate when irradiating cells with α-particles. Also, it has been shown that the surface chemistry of the plastic polymer dictates the amounts and conformation of adsorbed ECM molecules (18, 19, 20, 21).

HSF cells were irradiated with 2 Gy of γ-rays on either polystyrene or Mylar, and cells were labeled with BrdUrd for 24 h before the given harvest times (Fig. 1). Even at 24 h after irradiation, there is an increase in BrdUrd-labeled/proliferating cells in Mylar cultures (78%) compared to polystyrene (65%) cultures. However, the effect of substrate on G₁ arrest is dramatically illustrated in the 48 h time point. A total of 74% of HSFs in Mylar cultures but only 28% of cells in polystyrene cultures become BrdUrd-labeled 24–48 h after irradiation. Also, many proliferating cells in the Mylar cultures are in the S phase and G₂ phase of the second cell cycle (Fig. 1, arrows, 48 h), whereas cells that divide in the polystyrene cultures remain in G₁ phase.

At 48–72 h after irradiation, 81% of the cells on Mylar are BrdUrd-labeled compared to 34% of cells in the polystyrene cultures. Mylar cultures contain many cells that are present in the second S phase and G₂ phase, whereas such cells are absent in the polystyrene cultures (Fig. 1, arrows, 72 h). In control experiments, HSF cells that are cultured on polystyrene or Mylar grow at similar rates, so the difference in proliferation after irradiation is not due to a general difference in cell kinetics or cell cycle distribution at the time of irradiation.

Human fibroblasts that undergo a late G₁ arrest after γ-irradiation have a senescence-like phenotype in that the cells become very large and flat on the culture dish. We observed that cells irradiated on polystyrene are very large and flat after irradiation (data not shown). In contrast, cells that were irradiated on Mylar retained a smaller, more normal appearance, which correlates with our observation that HSF cells that are irradiated on Mylar do not remain arrested in G₁. Together, the data described in this section indicate that fewer fibroblasts that are irradiated on Mylar arrest permanently in G₁, and they also do not take on a senescence-like phenotype. Although the expression pattern of adhesion molecules on polystyrene and Mylar is not known, these data provide proof-of-principle support for our hypothesis that changing the type of adhesion interactions alters the percentage of HSF cells that arrest in G₁ after irradiation.

We next determined the effect of naturally occurring ECM molecules on radiation G₁ arrest of HSF cells. HSF cells were irradiated with 2 Gy of γ-rays on polystyrene, collagen IV, fibronectin, or laminin, and proliferating cells were labeled with BrdUrd for 24 h before the given harvest times. The total percentage of BrdUrd-labeled cells (proliferating cells) was determined using flow cytometry as described in Fig. 1, and a graphic representation of this data is shown in Fig. 2. The percentage of BrdUrd-labeled cells ranges from about 58% on polystyrene to about 70% on laminin during the first 24 h after irradiation. The difference in proliferation is much more dramatic at the 48 h time point, with only about 22% BrdUrd-labeled cells on polystyrene, 40–43% BrdUrd-labeled cells on fibronectin or collagen IV, and over 60% BrdUrd-labeled cells on laminin. At 72 h, there are only about 20% BrdUrd-labeled cells on polystyrene, 40% cycling cells on collagen IV or laminin, or about 60% BrdUrd-labeled cells on laminin. In Fig. 2, the percentage of cycling, irradiated cells on fibronectin and collagen is nearly identical. However, we sometimes find that the percentage of cycling cells on fibronectin is 5–10% higher than on collagen IV. Therefore, the trend that we observe for the percentage of cycling, irradiated HSF cells on different substrata is as follows: Mylar > laminin > fibronectin ≥ collagen IV > polystyrene.

G₁ phase radiation arrest is initiated by an increase in p53 protein amounts, followed by the induction of p21^Cip1 mRNA and protein (2, 3). This results in an inhibition of G₁ phase CDKs and G₁ phase arrest (3). The percentage of HSF cells that remain arrested in G₁ phase after irradiation differs, depending on the ECM substrate. Therefore, we determined whether amounts of p53, p21^Cip1, CDK2, and CDK4 correlate with the percentage of cells that arrest in G₁ after irradiation.

Our data indicate that the amount of induction of p53 is not a controlling factor in the length of the substrate-dependent radiation arrest. In all cases, p53 is induced at early times after irradiation and remains elevated at late times after irradiation when there are large differences in the percentage of proliferating cells (Fig. 3).

There are subtle but reproducible differences in the expression of p21^Cip1, CDK2, and CDK4 (Fig. 3). Irradiation of HSF cells on polystyrene, collagen IV, or fibronectin (where a high percentage of cells arrest) causes an increase in p21^Cip1 by 4 h after irradiation, which then decreases to a constant but elevated level by 48 h after irradiation. Many cells escape the late arrest when irradiated on laminin, and p21^Cip1 is induced at a later time compared to polystyrene, collagen IV, and fibronectin. On Mylar (where few cells arrest), p21^Cip1 is only modestly induced but remains slightly elevated throughout the experiment.

The expression patterns of CDK2 and CDK4 are more complex. For example, the amount of CDK2 decreases dramatically in polystyrene cultures, whereas there is only a modest decrease in CDK2 amounts of HSF cells irradiated on Mylar, where many cells are proliferating. Lower amounts of p21^Cip1 and higher amounts of CDK2 in Mylar cultures might correlate with higher CDK2 activity and an increase in proliferation. However, the amounts of CDK2 also decrease in laminin cultures, and many of the cells escape the late arrest. CDK4 amounts increase at late times after irradiation on laminin and Mylar. Increased proliferation after irradiation on these substrates might be supported by increased CDK4 activity or a titration of p21^Cip1 by increased CDK4, which could allow lower amounts of CDK2 to remain active.

Radiation causes the differentiation of fibroblasts to myofibroblasts (22, 23). A molecular marker characteristic of myofibroblasts is SM α-actin (17). Therefore, we determined whether radiation cell cycle arrest of HSF cells on different substrates is accompanied by an increase in SM α-actin expression. As seen in Fig. 4, SM α-actin is induced during the late arrest on polystyrene, collagen IV, and fibronectin, but not when HSF cells remain proliferating on laminin or Mylar. These data show that fibroblasts take on a differentiated myofibroblast phenotype after irradiation that is ECM dependent.

The dogma that radiation cell cycle delays occur primarily to repair DNA damage is well accepted but has not been rigorously proven. There is no doubt that DNA is the target of radiation in terms of cellular survival (see Ref. 24). It is also clear that DNA damage is a signal for increased p53 expression, which then results in cell cycle delay (25). The purpose of cell cycle delay is less obvious. One of the few studies that directly tests the idea that cell cycle delay results in increased repair of DNA damage shows that there is increased survival if cells are held in confluence-induced G₀ phase for several hours after irradiation (26). This indicates that, indeed, cell cycle delay after irradiation allows time to repair DNA damage. However, this is a very simplistic viewpoint that does not consider the potential function of individual cell types during radiation stress response in a tissue environment. For example, epithelial cells, which are a primary target of carcinogenesis, may apoptose after irradiation to subtract damaged cells from the reproductive population. Fibroblasts provide signals to epithelial cells during growth, development, differentiation, and apoptosis (reviewed in Refs. 27 and 28). Therefore, radiation may cause fibroblasts to undergo cell cycle arrest and then differentiation to provide specific stress signals to neighboring cells. In this study, we demonstrate that fibroblasts arrest in G₁ phase indefinitely and then differentiate to myofibroblasts, depending on the type of ECM environment. By inducing differentiation, the ECM may influence the type of signals that irradiated fibroblasts provide to surrounding cells. These results indicate that the interpretation of radiation cell cycle responses must be expanded to include the extracellular environment, in particular, the ECM.

We demonstrate that the ECM affects the percentage of HSF cells that remain arrested in G₁ phase after irradiation. Dramatic differences in proliferation after irradiation on polystyrene or Mylar are observed. The significance of this is not immediately apparent because plastic is not a biological substrate. However, ECM molecules in the serum or that are synthesized by the cells rapidly coat the plastic. It is this layer of conditioning protein that controls cell adhesion and subsequent cellular processes. The amount and orientation of protein adsorption to a substrate is controlled by the chemistry of the polymer surface (Ref. 18 and references therein). For example, the amount of laminin 5 that coats polystyrene depends on the type of functional groups that are created on the plastic during manufacturer’s plasma treatment of culture dishes (19). Surface chemistry also affects the orientation of bound laminin 5, resulting in differences in the availability of integrin-binding sites (28). Substrata chemistry also determines the conformation of bound fibronectin and, in turn, RGDS-dependent adhesion of fibroblasts and differentiation of neuronal cells (20, 21). The functional groups in plasma-treated polystyrene and Mylar are vastly different. Perhaps this influences the composition and/or orientation of adsorbed ECM and results in an altered cell cycle response to radiation. This is supported by our observation that radiation causes a different percentage of HSF cells to arrest on the naturally occurring ECM substrates, collagen IV, fibronectin, and laminin.

It is interesting that irradiated fibroblasts do not arrest well or differentiate on laminin because laminin is not an ECM protein that is commonly thought to be associated with the stroma. Rather, laminin is associated with the basement membrane and interacts more with epithelial cells. However, it has been found that transformed fibroblasts exhibit increased expression of integrin α₆, a component of the laminin receptor (29). Also, the expression of the fibronectin receptor α₅β₁ is lower in transformed fibroblasts than in normal fibroblasts (30). Another indication that integrin expression is involved in tumorigenesis comes from studies that showed that treatment of malignant breast cells with an inhibitory antibody to integrin β₁ results in differentiation/reversion of the malignant phenotype (31). Perhaps the irregular expression pattern of integrins in transformed fibroblasts or an aberrant ECM environment around nontransformed fibroblasts would result in an abnormal cell cycle response to radiation or other stress insults.

We demonstrate that the prolonged G₁ arrest results in SM α-actin expression, a hallmark of the myofibroblast phenotype. Several studies have reported that radiation causes fibroblast differentiation (22, 23), and ECM control of differentiation is well documented (reviewed in Ref. 16). However, to our knowledge, this is the first report of an ECM dependence of radiation-induced differentiation of fibroblasts. The SM α-actin promoter is p53 responsive (32). It is unlikely that p53 induction is solely responsible for increased SM α-actin expression because p53 amounts increase within the first few hours after irradiation, whereas increases in SM α-actin begin much later. Also, p53 amounts remain elevated on all substrates, and SM α-actin expression increases only on substrates where there is a late G₁ arrest. Therefore, some other molecular change, in addition to p53 expression, probably occurs in order for SM α-actin to be induced.

Perhaps the production of an extracellular molecule is involved in myofibroblast differentiation because radiation causes both an increase in TGF-β1 and fibroblast differentiation (22, 23, 33, 34). These two events might be inter-related because direct treatment of human and rat fibroblast cultures with activated TGF-β1 causes differentiation to myofibroblasts (35, 36, 37). In fact, TGF-β1 induction of specific ECM proteins might evoke myofibroblast differentiation. TGF-β1 expression leads to a large increase in the amounts of the ED-A splice form of fibronectin in cultured fibroblasts (38). The increase in ED-A fibronectin expression is followed by SM α-actin expression (39). Plating fibroblasts in the presence of neutralizing antibodies to ED-A fibronectin blocks the induction of SM α-actin by TGF-β1. It has been postulated that TGF-β1 induction in fibroblasts causes a cell cycle arrest through the inhibition of G₁ phase CDKs, which then predisposes the cells to differentiation (36). The basal amounts of TGF-β1 produced by myofibroblasts are 10–15 times higher than those produced by fibroblasts; however, radiation induces TGF-β1 only in fibroblasts (36). Thus, it was further suggested that the high basal amounts of TGF-β1 in myofibroblasts maintain the differentiated phenotype. This scenario would agree with our data that show that cell cycle arrest occurs early after irradiation, whereas differentiation is a later event.

The function of myofibroblast differentiation is not known, but it may be involved in radiation fibrosis, a dose-limiting and debilitating side effect of radiation cancer therapy in which fibroblasts deposit copious amounts of collagen and other matrix proteins in the irradiated tissue. This results in the progressive formation of aberrant ECM and localized regions of fibrosis. It is thought that myofibroblasts support ECM remodeling in fibrotic tissue (see Ref. 36) and that changes in the differentiation pattern of fibroblasts reflect the risk of patients to develop fibrosis (36, 40). Radiation fibrosis is often compared to wound healing. In both, there is a remodeling of the ECM, an increase in TGF-β1, and the differentiation of fibroblasts to myofibroblasts. The difference is that these events are reversed when the wound heals but are permanent in fibrotic tissue (41).

Another possible function for myofibroblasts is in cell-cell signalling, as suggested in prostate tissue (42). In the prostate gland, the epithelial cell mass is organized into distinct regions of viable or apoptotic epithelial cells. The surrounding stroma is also spatially organized: fibroblasts neighbor the regions of viable epithelial cells; and myofibroblasts neighbor areas of apoptotic epithelial cells. Androgen ablation causes an increase in TGF-β1 and mass apoptosis of epithelial cells throughout the prostate gland. Concurrently, the number of fibroblasts in the stroma decreases, and myofibroblasts surround the epithelial mass. The accumulation of myofibroblasts is thought to be caused by the differentiation of preexisting fibroblasts, and it is proposed that the myofibroblasts are responsible for producing TGF-β1 that leads to epithelial cell apoptosis. This is intriguing because human lung and prostate fibroblasts undergo a late radiation G₁ arrest along with an induction of p53 and p21^Cip1 proteins. However, cultured bronchial and prostate epithelial cells do not arrest and do not display an increased expression of p53 and p21^Cip1 proteins (43, 44). Perhaps in vivo or in fibroblast-epithelial cell coculture, TGF-β1 produced by myofibroblasts would cause cell cycle arrest or apoptosis of epithelial cells after irradiation.

The p53 control of radiation arrest has been well studied. However, there is a huge gap in our knowledge concerning the effects of the extracellular environment on radiation arrest. Also, there are several demonstrations that radiation causes fibroblast differentiation, but little is known about the significance of myofibroblast function or how the ECM affects myofibroblast differentiation. Obviously, the extracellular environment is of extreme importance because it more closely approaches the in vivo situation. Perhaps the phenomenon of permanent arrest of irradiated fibroblasts observed in vitro reflects the differentiation of fibroblasts in fibrosis and provides an opportunity to investigate whether cellular interactions with the ECM play an active role in this process.

We thank Robb Habbersett for expertise in flow cytometry and Jim Freyer and Karen LaRue for constructive criticism of the manuscript and experimental design.