Involvement of Rb family proteins, focal adhesion proteins and protein synthesis in senescent morphogenesis induced by hydrogen peroxide

Morphological change plays an important role in many cellular processes such as migration, differentiation, apoptosis, necrosis and senescence. Cell migration in mammalian cells involves a series of morphological changes. Gaining function in differentiated cells often requires morphological changes. Altered morphology is an important criterion for apoptosis or necrosis. During cellular senescence, normal human diploid fibroblasts (HDFs) change their morphology from a spindle shape to an enlarged, flattened and irregular shape.

Normal animal cells from non-tumor origin undergo senescence as a result of serial passage in tissue culture. Senescent cells show a number of molecular changes that are observed during the process of aging in vivo (Campisi, 1996; Campisi, 1997; Campisi et al., 1996; Chen, 2000). Using senescence-associated β-galactosidase as a biomarker, senescent cells are found in skin biopsy samples from elderly persons (Dimri et al., 1995). Accelerated cellular senescence is associated with the diseases that occur at high rates in elderly persons such as atherosclerosis and osteoporosis (Bennett et al., 1998; Kassem et al., 1997). Senescence has been used as a model for aging studies since Hayflick and Moorhead first reported the limited replicative life span of HDFs in 1961 (Hayflick and Moorhead, 1961).

A distinct feature of replicatively senescent cells is the morphological change including cell enlargement. The cause of this morphological change has not been well studied. Much of the mechanistic studies on senescence have been focused on the irreversible growth arrest. Since these cells are not replicating but are enlarged, it is necessary to test whether growth arrest and cell enlargement share similar control mechanisms. Traditionally, senescence results from cell replication. Most strains of HDFs from fetal tissues can replicate 50-60 population doublings before reaching senescence. Several recent studies indicate that early passage HDFs can develop a phenotype resembling senescence in response to oxidants, inhibitors of histone deacetylase, hyperactivation of ras gene and overexpression of E2F1 transcription factor (Chen and Ames, 1994; von Zglinicki et al., 1995; Ogryzko et al., 1996; Serrano et al., 1997; Lee et al., 1999; Dimri et al., 2000). In the model of induction of premature senescence with oxidants a 2-hour pulse treatment with low or mild doses of H₂O₂ causes the cells to lose replicative potential immediately and to develop senescent morphology one week later (Chen and Ames, 1994; Chen et al., 1998; Chen et al., 2000b). H₂O₂ treatment causes transient elevation of p53 protein and sustained inhibition of Rb hyperphosphorylation (Chen et al., 1998). Expressing human papillomaviral (HPV) protein E6 or E7 results in inactivation of p53 or Rb and abrogation of G₁ cell cycle arrest (Chen et al., 1998). Since senescent morphogenesis is coupled to growth arrest, this model allows us to test whether or not cell cycle checkpoint proteins are involved in the control of the morphological change. Nevertheless, the long time course (i.e. one week) of senescent morphogenesis following H₂O₂ treatment indicates that the process may be a programmed event involving new protein synthesis.

A large number of morphological studies in the literature focus on cytoskeletal proteins. Actin and focal adhesion proteins build the framework of a cell (Gumbiner, 1996). Expression levels or spatial arrangement of these cytoskeletal proteins determine the morphology of a cell. Actin is a highly conserved protein with several isoforms existing in non-muscle cells such as fibroblasts (Herman, 1993). Actin monomers align to form a polymer string, which can bundle to form actin filaments. Actin filaments assemble to form stress fibers which transverse a cell and are the major component of the cytoskeletal framework (Mitchison and Cramer, 1996). The ends of actin stress fibers meet at focal adhesion plaques, where cells contact with the extracellular matrix (Hemmings et al., 1995; Mitchison and Cramer, 1996). Focal adhesion plaques are composed of a number of proteins including vinculin and paxillin (Craig and Johnson, 1996; Schaller and Parsons, 1994). Since fibroblasts undergo drastic morphological change after treatment with low or mild doses of H₂O₂, we test here whether senescent morphology is coupled to changes in actin and focal adhesion proteins and if so whether these changes are under the control of cell cycle checkpoints.

All chemicals were purchased from Sigma Chemical Inc. (St Louis, MO) unless otherwise indicated.

IMR-90 cells (obtained from the Coriell Institute for Medical Research at the population doubling level (PDL)10.85, Camden, NJ) were subcultured weekly in 10 ml of Dulbecco’s modified Eagle medium (DMEM; Life Science Technologies, Grand Island, NY) containing 10% (v/v) fetal bovine serum (FBS, Life Science Technologies, Grand Island, NY) at a seeding density of 0.5×10⁶ cells per 100 mm Corning dish. The cells reach confluence at 6-7 days after subculture. For treatment with H₂O₂, cells were seeded at the density of 2×10⁶ per 100 mm dish 20-24 hours before the treatment. After incubating for 2 hours in the presence of H₂O₂, the cells were placed in fresh DMEM containing 10% (v/v) FBS.

The retroviral producing cells of human papillomavirus (HPV) type 16 E6, E7 and E6E7 genes were obtained from the American Type Culture Collection (Rockville, MD) (Demers et al., 1994). The cells producing E7 mutant d21-24, C24G or E26G virus were provided by Dr Denise Galloway (Demers et al., 1996). Exponentially growing IMR-90 cells (PDL ≤17) were infected with retroviruses carrying E6, E7, E6E7 or a mutant E7 gene and were selected by culturing in 500 μg/ml of G418 as described (Chen et al., 1998).

Cells in a 100 mm dish were lysed by scraping in Laemmli buffer (0.5 M Tris, pH 6.8, 2.4% (w/v) SDS, 50% (v/v) glycerol with protease inhibitors). Protein concentration was determined by the bicinchoninic acid (BCA) method (Pierce Inc, Rockford, IL). Proteins were separated by SDS-polyacrylamide gel electrophoresis using a mini-Protean II electrophoresis apparatus (Bio-Rad, Richmond, CA) run at 100 volts. The separating gel contained 8% acrylamide for detecting p53, actin or paxillin and 6% acrylamide for detecting Rb or vinculin. The separated proteins were transferred to immobilon-P membranes (Millipore, Bedford, MA) by electrophoresis as previously described (Chen et al., 1998). The membrane was incubated with anti-p53 (Ab-6, 1:100, Oncogene Science, NY), anti-Rb (polyclonal, 1:100, Santa Cruz Biotechnology, CA), anti-actin (clone AC-40, 1:1,000, Sigma, St Louis, MO), anti-vinculin (1:1,000, Sigma, St Louis, MO), or anti-paxillin (1:1,000, Zymed, South San Francisco, CA) antibody as described (Chen et al., 1998). The bound antibodies were detected by secondary antibodies conjugated with horseradish peroxidase and enhanced chemiluminescence reaction. The densities of the bands were quantified using an Eagle Eye II Image System (Stratagene, La Jolla, CA).

At 10 days after H₂O₂ treatment, adherent cells were detached by trypsinization. Rounded cells were loaded onto a microslide field finder (Fisher Scientific, Pittsburgh, PA). The diameters of the cells that randomly landed on the grids of the microslide field finder were viewed with a microscope and recorded. For each group, at least 99 cells were measured for diameters within one experiment. The cell volume was calculated based on the diameter using the equation for a sphere 1.33 × π × radius³. The number of cells showing diameters greater than 0.04 millimeter (mm) was counted for calculating the percentage of enlarged cells within a group of 33 cells.

Adherent cells were detached by trypsinization and collected by centrifugation at 3,300 rpm before fixation with 25% (v/v) ethanol containing 15 mM MgCl₂. After RNase digestion (0.1 μg RNase A and 2 units RNase T1/ml at 37°C for 1 hour), the cells were stained with 50 μg/ml propidium iodide for at least 30 minutes before analysis by flow cytometry (Becton Dickinson FACSorter) using CELLQuest software. The machine was set to collect 20,000 events. Forward light scatter, side light scatter, and DNA content were recorded simultaneously. During data analysis, the cells distributed in G₁, S, or G₂/M phase were gated for light scatters. Since the cells distributed in the S or G₂/M phase showed a plateau in the parameters of forward and side light scatters before H₂O₂ treatment, the data are presented from the cells distributed in G₁ phase of the cell cycle.

Control or H₂O₂ treated cells were seeded 7 days after H₂O₂ treatment onto 12 mm round coverglass (Carolina Biological Supply, NC) in 24-well plates at a density of 1.25×10⁴ cells per well. After culturing for an additional 48 hours, cells that were attached and spread on the coverglass were fixed by a 15-minute incubation in 5% formalin. After 70% (v/v) ethanol and PBS washes, the cells were incubated with 1 μg/ml tetramethylrhodamine B isothiocyanate (TRITC)-conjugated phalloidin for 1 hour with gentle shaking.

For indirect immunocytochemical staining, the cells grown on coverglass were fixed with 100% methanol for actin staining or 5% formalin for vinculin or paxillin staining. The coverglass was washed in sequence with PBS twice, PBS containing 0.5% NP-40 once, PBS three times, and PBS containing 1% bovine serum albumin (BSA) once. The coverglass was incubated for 2 hours with an antibody specific to actin (clone AC-40, 1:200, Sigma, St Louis, MO), vinculin (1:500, Sigma, St Louis, MO) or paxillin (1:500, Zymed, South San Francisco, CA). A secondary antibody conjugated with fluorescein isothiocyanate (FITC) was added to bind to the primary antibody for 1 hour. The cells on the coverglass were washed with PBS and mounted with 50% glycerol/PBS for fluorescent microscopy. Fluorescence images were obtained with a digital camera attached to a fluorescent microscope using IPlab spectrum software.

Cells were seeded at a density of 5×10⁴ per well in 24-well plates and treated with 150 μM H₂O₂ for 2 hours at 20-24 hours after seeding. Untreated cells or treated cells were incubated in complete medium containing 0.5 μCi/well [³H]leucine (ICN Pharmaceuticals, Costa Mesa, CA) for additional 24 hours before fixation with 10% trichloroacetic acid for determination of incorporated [³H]leucine as described (Chen and Ames, 1994).

A pulse treatment with low or mild doses of oxidants causes HDFs to develop features of senescence including growth arrest governed by cell cycle checkpoint proteins and development of senescent morphology. Since growth arrest is coupled to senescent morphogenesis, we test whether cell cycle checkpoint proteins play a role in senescent morphogenesis by expressing HPV E6 and E7 genes. These genes were introduced into HDFs by retrovirus-mediated gene transfer (Crystal, 1995; Sokol and Gewirtz, 1996), which gives nearly 100% transduction efficiency. The infection does not interfere with cell proliferation or produce cytotoxicity. To ensure that the E6, E7 or E6E7 genes were indeed expressed, we selected the cells for at least two weeks by G418 resistance conferred from a neo resistant gene under the control of the same promoter that drives the expression of E6, E7 or E6E7 genes in the pLXSN plasmid. Since almost all cells were resistant to G418, this method allowed us to study the consequence of E6, E7 or E6E7 gene expression in a population of cells without clonal selection.

Subconfluent cultures were used in this study because cells require space to develop senescent morphology after H₂O₂ treatment. This condition differs from our previous work in which cells were treated with H₂O₂ after reaching confluence and becoming quiescent (Chen and Ames, 1994; Chen et al., 1998). Since the response to H₂O₂ is determined by pmol per cell concentration rather than absolute micromolar (μM) concentration, addition of 150 μM (or 0.85 pmol/cell) H₂O₂ into the culture medium is sufficient for inducing senescent morphology in subconfluent cultures. Wild-type cells, cells expressing HPV E6 gene (E6 cells), cells expressing HPV E7 gene (E7 cells), or cells expressing both HPV E6 and E7 genes (E6E7 cells) at an equal density (2×10⁶ cells per 100 mm dish) were treated with 150 μM H₂O₂ for 2 hours. The cells were collected 20 hours later for measurement of p53 and Rb proteins using western blot analysis. An equal amount of proteins from different cell samples was loaded to each lane. Equal loading was verified by staining the gel with Coomassie blue after electrophoresis or by blotting the membrane with an antibody against actin. The results showed that E6 expression reduced the level of p53 in untreated and H₂O₂ treated cells (Fig. 1A). HPV E7 reduced levels of underphosphorylated Rb and phosphorylated Rb in control or H₂O₂ treated cells (Fig. 1A,B). In addition to enhancing the proteolytic degradation of Rb, HPV E7 presumably bound to Rb family proteins and prevented them from interacting with their cellular partners (Farthing and Vousden, 1994; Tommasino and Crawford, 1995). E6E7 cells failed to elevate p53 in response to H₂O₂ and showed some reduction of overall Rb protein levels (Fig. 1A,B). Rb was mainly presented in a hyperphosphorylated state after H₂O₂ treatment in E6E7 cells (Fig. 1A,B). An absence of underphosphorylated (activated) Rb was observed in E6E7 cells regardless of H₂O₂ treatment (Fig. 1A,B).

The cells expressing E6, E7 or E6E7 do not appear to be different from wild-type cells morphologically (Fig. 2A). E6, E7 or E6E7 cells showed similar proliferation doubling time with wild-type cells. E7 and E6E7 cells can achieve a higher saturation density at confluence than wild-type or E6 cells. Wild-type, E6, E7 or E6E7 cells at similar PDLs were seeded at an equal density for H₂O₂ treatment (Fig. 2A for the morphology of wild-type cells before H₂O₂ treatment. The morphology of E6, E7, or E6E7 cells before H₂O₂ treatment was not different from that of wild-type cells). After H₂O₂ treatment, wild-type cells or E6 cells developed senescent morphology while E7 cells could not do so (Fig. 2A). E6E7 cells were also reluctant to develop senescent morphology (Fig. 2A). Flow cytometry allows semiquantitative analysis of cell volume (reflected by X-axis forward light scatter) and cell surface irregularity (reflected by Y-axis side light scatter) from a large population of cells. The analysis confirmed the reluctance of E7 and E6E7 cells to increase cell volume or cell surface irregularity in response to H₂O₂ treatment (Fig. 2B). Since senescent morphology is coupled to cell volume increase, we quantified the average cell volume and scored the percentage of cells showing increased diameters after rounding up the cells by trypsin treatment. At 10 days after H₂O₂ treatment, wild-type or E6 cells showed about 10-fold increase in cell volume (Fig. 2C, top panel). Similar increase in cell volume was observed in replicatively senescent cells (Fig. 2C, top panel). In contrast, E7 cells and E6E7 cells failed to increase cell volume following H₂O₂ treatment (Fig. 2C, top panel). Most early passage cells showed diameters of 0.02 to 0.03 millimeter (mm). Replicatively senescent cells or H₂O₂ treated early passage wild-type cells show diameters varying from 0.02 to 0.07 mm. We scored the percentage of cells showing diameters greater than 0.04 mm as a way to quantify cell enlargement. Using this method, cell enlargement was not observed in E7 cells (Fig. 2C, bottom panel). E6E7 cells showed only a small proportion of enlarged cells in response to H₂O₂ treatment (Fig. 2C, bottom panel). The data point to the possibility that E7 interacting proteins namely Rb family proteins, but not E6 interacting protein such as p53, play a role in senescent morphogenesis induced by H₂O₂.

Since E7 interacts with Rb family proteins as well as a few other cellular proteins, the E7 mutants that are defective in binding to Rb family proteins allow us to further test the role of Rb family proteins in senescent morphogenesis. E7 protein interacts with Rb family proteins through LxCxE consensus sequence at the amino acid position 21-25 (Demers et al., 1996; Farthing and Vousden, 1994; Tommasino and Crawford, 1995). Deletion of amino acids 21 to 24 or mutation at the position 24 or 26 reduces the Rb binding activity of E7 (Demers et al., 1996). The d21-24 (deletion of amino acids at the position 21 to 24), C24G (replacing cysteine at the position 24 to glycine), and E26G (replacing glutamate at the position 26 to glycine) mutants of E7 gene were introduced into IMR-90 cells by the retroviral technique described above. Expression of d21-24, C24G or E26G did not result in changes in growth rate or cell morphology. Western blot analysis showed that cells expressing d21-24, C24G or E26G mutant E7 gene contained Rb protein at the level similar to that of wild-type cells (Fig. 3A,B). Like wild-type cells, cells expressing d21-24, C24G or E26G mutant E7 gene contained only underphosphorylated Rb protein 20 hours after H₂O₂ treatment (Fig. 3A,B). Unlike E7 cells, the cells expressing mutant E7 genes appeared to develop senescent morphology and showed a degree of cell volume increase or percentage of enlarged cells similar to that of wild-type cells in response to H₂O₂ (Fig. 3C,D). These data further point to the possibility that HPV E7 prohibits senescent morphogenesis through interaction with Rb family proteins.

The drastic change in morphology following H₂O₂ treatment suggests the involvement of cytoskeletal proteins, since actin, focal adhesion proteins and other cytoskeletal proteins build the framework of mammalian cells. Phalloidin, a toxin originally isolated from the poisonous fungus Amanita phalloides, binds to polymeric actin. Fluorescence labeled phalloidin allows the determination of intensity and arrangement of actin filaments. Fluorescent phalloidin was added to formalin-fixed cells collected at 9-10 days after H₂O₂ treatment. The results indicated that most untreated wild-type cells contained limited actin stress fibers visible at the magnification of 66 times (Fig. 4A). In contrast, H₂O₂ treated wild-type cells showing senescent morphology contained extensive actin stress fibers transversing the entire cells at the same magnification (Fig. 4A). Indirect immunocytochemical staining with the antibody developed against the conserved C-terminal peptide sequence (Ser-Gly-Pro-Ser-Ile-Val-His-Arg-Lys-Cys-Phe) of all actin isoforms generated the same results (data not shown). Enhanced actin filaments were also observed in replicatively senescent cells (Fig. 4B).

Since H₂O₂ treatment results in an enhanced actin filament formation and HPV E7 can prevent H₂O₂ from inducing senescent morphology, we test whether E7 expression can abolish changes in actin following H₂O₂ treatment. Phalloidin staining of formalin-fixed cells revealed that E7 cells did not contain detectable changes in actin filament density, strength or length after H₂O₂ treatment (Fig. 4A).

The enhancement of actin filaments in wild-type cells following H₂O₂ treatment suggests a possible increase in the level of actin protein. Using an antibody that recognizes various actin isoforms (e.g. α, β and γ actin), we determined the level of actin protein 7-10 days after H₂O₂ treatment by western blot. The results indicate no changes in the level of actin protein in wild-type, pLXSN or E7 cells treated with various concentrations of H₂O₂ (Fig. 5).

In addition to actin filaments, focal adhesion plaques are also important components of cell morphology. The distribution of focal adhesion protein vinculin or paxillin was determined using an immunocytochemical technique with cells fixed at 9-10 days after H₂O₂ treatment. The results showed that vinculin and paxillin preferentially distributed at the edge of untreated cells (Fig. 4C,D), while many H₂O₂ treated wild-type cells showed a sporadic and random distribution of vinculin and paxillin (Fig. 4C,D). Changes in vinculin distribution were also observed in replicatively senescent cells (Fig. 4B). The number of vinculin or paxillin foci was quantified to further document changes in the distribution of focal adhesion plaques. Table 1 shows an overall 7-fold increase in the number of vinculin or paxillin foci per senescent-like cell compared to untreated control. While 87% vinculin foci were distributed along the edge of the untreated cells, 93% vinculin foci were not associated with edge distribution in senescent-like cells (Table 1). Similar results were obtained with paxillin (Table 1). In comparison, E7 cells did not change vinculin or paxillin distribution significantly after H₂O₂ treatment (Fig. 4C,D; Table 1).

The apparent increase in the number of vinculin or paxillin foci per H₂O₂ treated wild-type cell indicated a possible increase in the level of vinculin or paxillin protein. When the level of vinculin or paxillin was measured by western blot, we did not observe any significant increase resulting from H₂O₂ treatment in wild-type, pLXSN or E7 cells (Fig. 5).

H₂O₂ is a toxic agent. If senescent morphogenesis is a result of degeneration after oxidative injury, the process may not require de novo protein synthesis. The long time course required for the development of senescent morphology and the correlation of senescent morphogenesis with the presence of underphosphorylated (activated) Rb protein plus changes in the cytoskeletal proteins indicate the process may be a programmed event involving de novo protein synthesis. H₂O₂ at the dose range capable of inducing premature senescence does not appear to abolish protein synthesis (Chen and Ames, 1994). An increase in the number of vinculin or paxillin foci per H₂O₂-treated wild-type cell in the absence of an increase in the protein level indicates a possible increase in the overall protein content per H₂O₂ treated wild-type cell. We tested whether de novo protein synthesis was required for the morphological change by placing the cells in methionine-free medium following a 2-hour treatment with 150 μM H₂O₂. [³H]Leucine incorporation was reduced to 40% in untreated or H₂O₂ treated cells by methionine depletion when measured during a 24-hour time period. When morphology was observed 7 days after H₂O₂ treatment, the cells incubated in methionine-free medium did not appear to enlarge (Fig. 6A,B). These data suggest a role of de novo protein synthesis in the induction of senescent morphology by H₂O₂ treatment.

To determine the relationship between de novo protein synthesis and underphosphorylated (activated) Rb in senescent morphogenesis, we measured protein synthesis rate after introducing the E7 gene or the status of Rb phosphorylation by methionine depletion. Introducing the E7 gene into IMR-90 cells did not affect the rate of protein synthesis (data not shown). In contrast, H₂O₂ treated cells showed mainly underphosphorylated Rb when kept in complete medium or methionine-free medium for 20 hours (Fig. 7). The data indicate that H₂O₂ treated cells were capable of dephosphorylating Rb in methionine-free medium.

The present study supports that senescent morphogenesis is controlled by specific molecular changes. Senescent morphology develops when H₂O₂ treated cells are kept in culture for 7 or more days. Among a few early molecular changes induced by H₂O₂ treatment are transient elevation of p53 and prolonged inhibition of Rb phosphorylation (Chen et al., 1998). HPV E6 reduces p53 protein level while HPV E7 reduces Rb protein level in HDFs. We found that HPV E7 but not E6 can abolish senescent morphogenesis.

Rb protein can be inactivated by three mechanisms: hyperphosphorylation, proteolytic degradation and E7 binding. E7 expression in IMR-90 cells results in decreases of Rb protein levels. This observation is consistent with the report by Boyer et al. (Boyer et al., 1996) showing that E7 expression enhances proteolytic degradation of Rb protein. In addition to reducing the level of Rb protein, E7 binds to Rb family proteins such as Rb, p107 and p130 (Farthing and Vousden, 1994; Tommasino and Crawford, 1995). Upon binding to Rb, E7 prevents Rb from interacting with its cellular partners including E2F (Farthing and Vousden, 1994; Tommasino and Crawford, 1995). Although the decrease of Rb protein in E6E7 cells is not as dramatic as in E7 cells, the observed hyperphosphorylation and likely E7 binding can inactivate Rb in E6E7 cells. Both E7 cells and E6E7 cells show a shift in the dose response of DNA synthesis inhibition towards higher concentration compared to wild-type cells (Chen et al., 2000a). Other similarity between E7 and E6E7 cells include failures to undergo G₁ arrest immediately, to replicate and to develop senescent morphology after H₂O₂ treatment (Chen et al., 1998; Chen et al., 2000a). Deletion or mutation at LxCxE consensus sequence abolishes the interaction of E7 with Rb family proteins (Demers et al., 1996). These mutants failed to reduce the level of Rb protein (Fig. 3A,B) or prevent senescent morphogenesis, indicating a role of Rb family proteins in senescent morphogenesis. The argument that Rb family proteins play a critical role in senescent morphogenesis is supported by a recent report showing that overexpression of the Rb gene can induce senescent morphology in an immortalized cell line (Xu et al., 1997). However, HPV E7 has been reported to interact with a number of cellular proteins other than Rb family proteins, for example p27KIP1 (Zerfass-Thome et al., 1996), AP1 family transcription factors (Antinore et al., 1996), cyclin E (McIntyre et al., 1996), cyclin A and p33CDK2 (Tommasino et al., 1993). HPV E7 expression has been linked to elevated expression of cyclin A, cyclin E and p21 (Jian et al., 1999; Martin et al., 1998; Schulze et al., 1998; Vogt et al., 1999). Although some of these E7 functions appear to be mediated through its interaction with Rb family proteins (Martin et al., 1998), the mechanisms of how E7 interacting with these cellular proteins or how E7 affecting the expression of these cellular proteins are unknown. Because we have not examined the effect of the E7 mutants on each of these cellular proteins, we cannot exclude the possibility that some of these cellular proteins may participate in the control of senescent morphogenesis.

Underphosphorylated (activated) Rb is known to interact with a large number of cellular proteins (Taya, 1997). Induction of senescent morphology correlates with the presence of underphosphorylated (activated) Rb, indicating a possible role of Rb-interacting proteins in senescent morphogenesis. One candidate is the nuclear tyrosine kinase c-abl (Welch and Wang, 1993; Wen et al., 1996). c-Abl tyrosine kinase can be activated by DNA-damaging agents (Kharbanda et al., 1995; Liu et al., 1996; Welch and Wang, 1993). The activated c-abl can translocate to the cytoplasm and phosphorylate the Focal Adhesion Kinase and a number of focal adhesion proteins (Salgia et al., 1995; Welch and Wang, 1993). Phosphorylation of focal adhesion proteins is thought to regulate focal adhesion assembly and cell-matrix contact (Hanks and Polte, 1997; Schaller and Parsons, 1994). In addition, c-abl is found capable of binding to actin filaments and cooperating the actin filament bundling (Van et al., 1994). The presence of Rb appears to be critical for c-abl function (Wen et al., 1996). In parallel with this c-abl hypothesis, Rb may control senescent morphogenesis through interaction with the chromatin modulating factor hBRG1/hBRM (Strober et al., 1996). These proteins are the mammalian homologues of the yeast SNF2/SWI2 or Drosophila brm gene and form a large complex that serves to remodel chromatin and facilitates the function of specific transcription factors (Phelan et al., 1999; Wang et al., 1996). The interaction of hBRG1/hBRM and Rb is known to result in cell spreading and enlargement (Strober et al., 1996). Therefore c-abl and hBRG1/hBRM are two possible mediators between Rb and senescent morphogenesis.

Our data from methionine depletion experiments indicate that the presence of underphosphorylated (activated) Rb may be necessary but not sufficient for senescent morphogenesis. Methionine is an essential amino acid for human cells and depletion of methionine ultimately results in inhibition of cell proliferation. Depletion of methionine alone cannot lead to senescent morphology but can prevent H₂O₂ from inducing senescent morphology (Fig. 6), supporting that senescent morphogenesis is not simply a consequence of inhibition of cell proliferation. Overall protein synthesis is likely important in driving the complex process of morphological change. The dramatic increases in cell surface area and cell volume indicate an increase in organic cellular content. Since H₂O₂ treated cells cannot synthesize DNA or replicate but remain capable of synthesizing protein, the protein content likely accumulates and contributes to cell enlargement. Nevertheless, H₂O₂ has been shown to activate mitogen activated protein (MAP) kinases (Aikawa et al., 1997; Guyton et al., 1996) and protein kinase C (Konishi et al., 1997), which can lead to changes in gene expression. A number of genes increase their expression as a result of oxidative stress (Sen and Packer, 1996). These suggest the possibility that synthesis of a particular factor is equally possible as the overall protein synthesis for senescent morphogenesis following H₂O₂ treatment.

Senescent morphology correlates with enhanced actin stress fibers and redistribution of focal adhesion plaques in senescent-like cells. Similar changes of actin have been reported (Wang and Gundersen, 1984) and observed here with replicatively senescent cells (Fig. 4B). Our data further indicate the homology between H₂O₂ treated cells and replicatively senescent cells. Since the protein level of actin does not appear to change, a likely explanation of observed enhancement of actin stress fibers is an increased rate of actin polymerization or a decreased rate of actin filament depolymerization in H₂O₂ treated cells. On the other hand, the distribution pattern of focal adhesion plaques explains the observed increases in adhesion and spreading of senescent-like cells or replicatively senescent cells. Rodriguez et al. (Rodriguez et al., 1992; Rodriguez et al., 1993) report that the level of vinculin protein influences the number of focal adhesion plaques and cell surface area. Although we did not observe an increase in the overall protein level of vinculin or paxillin by western blot analysis, the immunocytochemical studies indicated an increase in the number of focal adhesion plaques per senescent-like cell. As discussed in the above section, H₂O₂ treated cells appear to be bigger and may contain more protein per cell than untreated early passage cells. This phenomenon may explain the difference between western blot versus immunocytochemistry, since western blot measures the concentration of a particular protein in the pool of total cellular proteins while immunocytochemistry measures a particular protein in each individual cell. Functionally while enhanced actin stress fibers may construct a frame structure necessary to support the enlarged cells, the increased number of focal adhesion plaques per cell and their random distribution may be necessary for the enlarged cells to adhere and to spread.

Senescence is thought to be controlled by mechanisms overlapping with tumor suppression (Campisi et al., 1996; Smith and Pereira, 1996). Changes in cell morphology and adhesion are important parameters of cancer metastasis and invasion (Button et al., 1995; Gumbiner, 1996). Tumor cells in general are often smaller and less adhesive than their parental cells. Many tumor cells appear to have less actin stress fibers and often reduce the level of vinculin or the number of focal adhesion plaques compared to their normal counterparts (Button et al., 1995; Otto, 1990; Zigmond, 1996). In contrast, senescent or senescent-like cells are enlarged and contain enhanced actin stress fibers and an increased number of focal adhesion plaques. From the morphological point of view, senescent or senescent-like cells may contribute to the mechanism of tumor suppression simply by cell enlargement and changes in cytoskeletal proteins.

This work was supported by the Burroughs Wellcome New Investigator Award (Q.M.C.), the start-up fund from the Department of Pharmacology and the Dean’s Research Award from College of Medicine, University of Arizona (Q.M.C.), Flinn predoctoral fellowship (V.C.T.), Fulbright, CIES, and NATO fellowships (M.B. and O.T.). We are greatly indebted to Dr Denise Galloway for mutant E7 producing cells, Dr John Regan for access to the microscopic imaging system and Juping Liu for technical assistance.