Introduction

Cellular senescence is a permanent cell cycle arrest that limits the uncontrolled proliferation and tumorigenesis of cells (Greider, 2000; Campisi, 2003). Senescent cells, while remaining metabolically active, display a number of characteristic changes. Typically, these changes include an enlarged and flattened morphology, increased senescence-associated -galactosidase (SA--gal) activity and a distinct chromatin structure known as senescence-associated heterochromatic foci (SAHF) (Dimri et al., 1995; Campisi, 2001; Narita et al., 2003). Additionally, senescent human fibroblasts exhibit the activation of a cell cycle arrest enforced by the p53 and p16/Rb tumor suppressor pathways and is accompanied by the increased expression of the cyclin-dependent kinase (CDK) inhibitors p21^CIP1/WAF1 (p21) and p16^INK4a (p16) (Rangarajan and Weinberg, 2003). Moreover, inactivation of p53 and Rb can prevent senescence (Cong et al., 2002).

The first description of senescence was replicative senescence in which cultured human fibroblasts had exhausted their growth potential after a number of cell divisions (Hayflick, 1965). One well-studied mechanism responsible for replicative senescence in human cells is the shortening and dysfunction of telomeres (Harley et al., 1990; Bodnar et al., 1998; Karlseder et al., 2002). Recently, numerous studies have shown that senescence can be triggered independently of the number of cell divisions or telomere length (Shay and Roninson, 2004). This type of senescence, often referred to as 'premature senescence', can be induced by various types of cellular stress, including activated oncogenes, DNA damage, reactive oxygen species, etc. (Serrano et al., 1997; Chen et al., 1998; Lin et al., 1998; Robles and Adami, 1998; Zhu et al., 1998; Kramer et al., 2001; Suzuki et al., 2001).

Ectopic expression of oncogenic H-ras (ras) in normal fibroblasts results in a permanent cell cycle arrest that is phenotypically indistinguishable from replicative senescence (Serrano et al., 1997). Oncogenic ras-induced senescence depends on signal intensity, although activation of the Raf-1/MEK/p38MAPK pathway is essential for oncogenic ras-induced senescence (Lin et al., 1998; Zhu et al., 1998; Wang et al., 2002). Additionally, inactivation of the ARF/p53 tumor suppressor pathway in mouse fibroblasts and skin keratinocytes, or inactivation of the p16/Rb tumor suppressor pathway in human fibroblasts, can bypass ras-induced senescence, suggesting that the outcome in response to oncogenic ras signaling strictly depends on the cellular context (Serrano et al., 1997; Lin and Lowe, 2001; Brookes et al., 2002; Huot et al., 2002; Mallette et al., 2004). However, the specific molecular mechanism by which oncogenic ras triggers a senescence-like permanent growth arrest in normal cells remains poorly understood. The objectives of this study were to characterize molecular and cellular changes in oncogenic ras-induced senescent human IMR90 fibroblasts to gain insight into the senescence mechanism and to identify candidate senescence biomarkers.

Results

Gene expression profiling of oncogenic ras-induced senescent cells

To further delineate the molecular mechanism and identify important genes and biomarkers involved in the senescence program in normal human fibroblasts, we carried out DNA microarray analyses of oncogenic ras-induced senescent IMR90 fibroblasts as a function of time. The Affymetrix GeneChip^® HG-133A microarrays, which contain sets of probes for at least 22 000 genes, were used in this analysis.

Oncogenic ras was introduced into normal human diploid IMR90 fibroblasts via retroviral gene transfer as described previously (Lin et al., 1998). Immediately after selection, designated as day 0 (Figure 1a), IMR90 cells stably expressing oncogenic ras continued to proliferate in culture and did not display a senescent phenotype (Figure 1a; data not shown) (Narita et al., 2003). At day 3 after selection, oncogenic ras-expressing cells began to exhibit a significant reduction in their ability to incorporate BrdU, although only a small percentage of cells showed elevated SA -gal activity, suggesting that these cells were undergoing cell cycle arrest but had not reached senescence. In contrast, oncogenic ras expressing cells by day 6 postselection displayed a flat and enlarged cell morphology, dramatic reduction of BrdU incorporation and a significant increase of SA -gal activity (Figure 1a; data not shown) (Serrano et al., 1997). Therefore, gene expression profiles were analysed over time at day 0 (R0), day 3 (R3) and day 6 (R6) after selection (Figure 1a). As a control, IMR90 cells transduced with an empty vector were also analysed at day 6 after selection (V6). Gene expression profiling was obtained from two independent analyses using cells derived from two independent gene transfers.

Figure 1.

Altered expression of cell cycle regulators in oncogenic ras-induced senescent IMR90 fibroblasts. (a) Upper panel: time scheme of retroviral gene transfer. Lower panel: representative photomicrographs of IMR90 cells transduced with an empty vector (Vector) or oncogenic ras (Ras) at days 0, 3 and 6 postselection. (b) RT–PCR analysis of the indicated genes encoding cell cycle regulators in IMR90 cells transduced with an empty vector or oncogenic ras at days 0, 3 and 6 postselection (R0, R3 and R6). Expression of 2-microglobulin (2M) was used as a loading control. (c) Western blot analysis of the indicated cell cycle regulators in IMR90 cells transduced with an empty vector or oncogenic ras at days 0, 3 and 6 postselection (R0, R3 and R6). Expression of -tubulin (Tubulin) was used as a protein loading control

Full figure and legend (153K)

Significantly, the majority of the gene expression changes were detected between R0 and R3 (Table 1 and data not shown). Consistent with a cell cycle block, the downregulated genes included many genes required for cell division, such as cyclins, and DNA replication, DNA repair and mitotic genes (Table 1). Additionally, genes that were significantly upregulated included cell cycle inhibitors, degradative enzymes, cytokines and growth factors (Table 1 and Supplemental Table). Of note, some of these gene changes have been observed in replicative senescent cells (Shelton et al., 1999; Ly et al., 2000; Campisi, 2003).

Table 1 - Gene expression comparisons of cell cycle regulators.

Full table

Altered expression of cell cycle regulators in oncogenic ras-induced senescent cells

The p53 and pRB tumor-suppressor pathways play essential roles in mediating the growth arrest in senescence (Campisi, 2003). CDK inhibitors, which act on the Rb pathway, are often upregulated in senescent cells (Campisi, 2003). The DNA microarray analyses indicated that the transcripts of the cell cycle inhibitors p16, p15^INK4B (p15), and p21 were all significantly upregulated in oncogenic ras-induced senescent cells as compared to the vector control (Table 1, comparing R6 and V6). In contrast, the expression of the genes encoding p18^INK4C, p27^KIP1 and p57^KIP2 were decreased, while the gene encoding p19^INK4D remained unchanged (Table 1).

We next examined the Rb family member proteins, Rb, p107 and p130, which play an important role in mediating growth suppression signals in nonproliferating cells (Classon and Harlow, 2002). In oncogenic ras-induced senescent cells, Rb and p130 gene expression exhibited insignificant changes, while p107 gene expression was significantly reduced (Table 1). Rb family members are known to regulate the cell cycle through their interactions with E2F transcription factors. There are at least six E2F family members identified in mammalian cells (Cam and Dynlacht, 2003). Interestingly, only the expression of E2F1 was significantly reduced in oncogenic ras-induced senescent cells, while the other E2Fs remained unchanged (Table 1). E2F1 is a transcription factor that can activate the transcription of many genes required for the progression through G₁ and S phase (Trimarchi and Lees, 2002). Consistent with this notion, many of the E2F target genes, including cyclin E, PCNA and MCM6, were downregulated in oncogenic ras-induced senescent cells (Table 1). Altered expression of these cell cycle regulators was confirmed by RT–PCR (Figure 1b) and/or Western blot analysis (Figure 1c), and the results were highly consistent with the microarray data. The slight reduction of Rb protein and upregulation of E2F4 protein in ras-induced senescent cells suggests post-translation mechanisms may be involved. In summary, the cell cycle arrest in oncogenic ras-induced senescence involved a specific downregulation of E2F1 and upregulation of p16, p15 and p21.

Oncogenic ras-induced senescent cells exhibit significant downregulation of G₂/M genes and contain G₁ tetraploidy

The altered expression of cell cycle regulators in cells expressing oncogenic ras is highly indicative of a growth arrest. Unlike quiescent cells, forced expression of oncogenic ras or Raf-1 in normal fibroblasts leads to an irreversible growth arrest (Lin et al., 1998; Zhu et al., 1998). Indeed, the fundamental difference between senescent cells and quiescent cells is that the latter are able to re-enter the cell cycle once the trigger is removed (Campisi, 1992). Additionally, quiescent cells, generated by serum starvation, predominantly accumulated a cell population with 2 N DNA content, indicating a cell cycle arrest in G₁ (Figure 2a). In contrast, oncogenic ras-induced senescent cells reproducibly showed an accumulation of cells with both 2 and 4 N DNA content (Figure 2a).

Figure 2.

Oncogenic ras-induced senescence is accompanied by downregulation of G₂/M checkpoint genes and accumulation of G₁ tetraploid cells. (a) DNA content analysis of IMR90 cells transduced with an empty vector (Vector) or oncogenic ras (Ras) at day 6 postselection, or serum starvation-induced quiescent IMR90 cells. Each histogram is equal in scale with DNA content indicated below. (b) RT–PCR analysis of the indicated G₂/M genes in IMR90 cells transduced with an empty vector at day 6 postselection (V6), oncogenic ras at days 0, 3 and 6 postselection (R0, R3 and R6), replicative senescent cells (RS), serum starvation-induced quiescent cells (Q) and normal proliferating cells (N). Expression of 2-microglobulin (2M) serves as a loading control. (c) Representative photomicrographs of IMR90 cells expressing oncogenic ras at day 6 postselection. Cells with two nuclei are indicated by the white arrowheads (top panel), and are shown in higher magnification (middle and bottom panel). (d) Representative immunofluorescent images for BrdU incorporation (BrdU) and p21 expression (p21) in IMR90 cells transduced with an empty vector (Vector) or oncogenic ras (Ras) at day 6 postselection. DAPI was used to counterstain nuclei

Full figure and legend (141K)

While cells arrested in G₂/M normally exhibit 4 N DNA content, cells with 4 N DNA content can also be arrested in G₁ (G₁ tetraploid cells) (Margolis et al., 2003). Tetraploid cells can arise from checkpoint failures in mitosis or cytokinesis causing cells to exit mitosis without chromosome segregation and eventually arrest in G₁ (Margolis et al., 2003). Interestingly, gene expression analysis revealed that cdc2, cyclin B1, cdc25c and PLK-1, were among several G₂/M genes that were significantly downregulated in oncogenic ras-induced senescent cells (Table 1). Additionally, mitotic genes such as DDX11 (CHL-1like), BUB1B, CENPA, CENPF and topoisomerase II (topo II) were also dramatically downregulated (Table 1). We next compared gene expression changes in replicative senescent and quiescent cells relative to oncogenic ras-induced senescent cells by RT–PCR. For all the G₂/M genes examined, gene expression was downregulated in both oncogenic ras-induced and replicative senescent cells (Figure 2b). In contrast, quiescent cells consistently showed a small reduction in gene expression of G₂/M genes, perhaps due to the diminished number of G₂/M cells (Figure 2a and b). Moreover, genes encoding protein regulator of cytokinesis (PRC1) and survivin were also significantly reduced in oncogenic ras-induced and replicative senescent cells (Table 1, Figure 2b). PRC1 and survivin are required for the later stages of mitosis and cytokinesis and their inhibition can lead to tetraploid cells with two nuclei (Andreassen et al., 2001; Mollinari et al., 2002; Altieri, 2003). Consistent with these findings, approximately 10% of oncogenic ras-induced senescent cells reproducibly exhibit two nuclei, suggesting that the cells with 4 N DNA content may reflect, at least in part, binucleated cells (Figure 2c). Taken together, these data suggest that the dramatic downregulation of G₂/M genes is rather significant in senescent cells, and that oncogenic ras-induced senescent cells contain tetraploid cells that might have arisen as mitotic errors due to altered expression of G₂/M genes.

To test the hypothesis that the tetraploid cells might have entered G₁ phase and undergone G₁ arrest, we examined the expression of p21 in these cells by performing immunostaining using an antibody specific for p21. Consistent with the results from Western blot analysis, cells containing an empty vector exhibited low expression of p21 (Figure 2d). In contrast, the majority of cells expressing oncogenic ras displayed elevated expression of p21. Significantly, binucleated cells also exhibited high levels of p21 expression, suggesting that these cells might have undergone a G₁ arrest (Figure 2d).

Mitotic errors resulting from DNA damage have been shown to cause G₁ tetraploidy, and certain tumor cell lines can undergo senescence in response to treatment with DNA damage-inducing drugs (Roninson, 2003). Moreover, oncogenic ras signaling has been associated with genomic instability (Saavedra et al., 1999), raising the possibility that oncogenic ras-induced senescence may be mediated through DNA damage. To examine if oncogenic ras-induced senescence involves DNA damage, comet assays were carried out to detect double-strand breaks in oncogenic ras-expressing cells. If significant DNA damage has occurred in cells, the ethidium bromide-stained nuclei will display a comet-like tail emanating from the cell nucleus corresponding to fragmented DNA (Singh et al., 1988). As controls, Phoenix cells were treated with the DNA-damaging agent, etoposide, or DMSO solvent (Figure 3). In contrast to the etoposide-treated Phoenix cells, neither the vector control nor the oncogenic ras-induced senescent IMR90 cells showed comet tails (Figure 3). Thus, oncogenic ras-induced senescence did not involve significant DNA damage that could be detected by the comet assay.

Figure 3.

Oncogenic ras-induced senescent cells do not exhibit detectable DNA breaks in a comet assay. Representative images of comet assays on IMR90 cells transduced with an empty vector (Vector) or oncogenic ras (Ras) at day 6 postselection. Phoenix cells treated with 100 M etoposide (Etoposide) or DMSO (DMSO) for 3 h were used as a positive control and a negative control, respectively. Senescent cells have enlarged nuclei and DAPI staining is not uniform likely due to heterochromatic foci (SAHF)

Full figure and legend (52K)

Specific downregulation of topoisomerase II and upregulation of HDAC9 in senescent cells

We anticipated that the gene expression analysis of oncogenic ras-induced senescent cells would help to identify novel senescence biomarkers. Many genes essential for executing critical functions throughout the cell cycle have altered expression in senescent cells. One particular gene was topo II, which is required for catalysing topological changes in DNA and plays an essential role in chromosome condensation and segregation in mitosis (DiNardo et al., 1984; Uemura et al., 1987; Nitiss, 1998; Larsen et al., 2003). Significantly, among the five topoisomerases represented in the microarray, only topo II was downregulated in oncogenic ras-induced senescent cells (Table 2). Importantly, downregulation of topo II, but not topoisomerase II (topo II), was detected in both oncogenic ras-induced and replicative senescent cells by RT–PCR analysis (Figure 2b and 4b). In contrast to the dramatic reduction of topo II expression in senescent cells, only a slight reduction of the topo II transcript was detected in serum starvation-induced quiescent cells (Figure 2b). It has been shown that the expression of topo II generally peaks in the G₂/M phase of the cell cycle (Whitfield et al., 2002). Therefore, it is likely that the slightly reduced expression of topo II in serum starvation-induced quiescent cells is due to the lack of G₂/M cells in quiescence.

Figure 4.

Altered expression of topo II and HDAC9 is specifically associated with senescent fibroblasts. (a) Western blot analysis of the indicated proteins in IMR90 cells transduced with an empty vector at day 6 postselection (V6), oncogenic ras (R) at day 6 postselection, early passage normal IMR90 cells (N), replicative senescent IMR90 cells (RS), and IMR90 cells cultured in low serum for the indicated days. (b) RT–PCR analysis of the indicated genes in IMR90 cells transduced with empty vector (V6), oncogenic ras at day 0, 3, and 6 postselection (R0, R3, R6), normal early passage IMR90 cells (N), replicative senescent IMR90 cells (RS), serum starvation-induced quiescent cells (Q) and serum starvation-induced senescent cells (SSS). Expression of 2-microglobulin (2M) serves as a loading control. (c) SA--gal activity assay (SA--gal) and SAHF formation (SAHF) in MRC-5 cells transduced with an empty vector at day 6 postselection (Vector) or oncogenic ras (Ras) at day 6 postselection. SAHF formation is detected by staining nuclei with DAPI and is evident by a punctate nuclear pattern. (d) RT–PCR analysis of the indicated genes in MRC-5 cells transduced with an empty vector at day 6 postselection (V6), or oncogenic ras at days 0, 3, and 6 postselection (R0, R3 and R6). Expression of 2-microglobulin (2M) serves as a loading control

Full figure and legend (209K)

Table 2 - Gene expression comparisons of topoisomerases and HDACs.

Full table

While short-term serum starvation is known to be able to trigger quiescence, we have shown that prolonged serum starvation can also trigger a senescence-like irreversible growth arrest (Weebadda et al., 2004). Cells grown in low serum (0.1%) for 2 days were already growth arrested as indicated by the presence of only the hypophosphorylated form of Rb (Figure 4a). Consistent with the results obtained from RT–PCR analysis, topo II protein levels were only slightly reduced in quiescent cells generated by serum starvation for 2 days (Figure 4a). Interestingly, after 21 days of culturing in low serum, these cells exhibited irreversible growth arrest associated with a senescence phenotype (Weebadda et al., 2004). Moreover, serum starvation-induced senescence is associated with a dramatic reduction in topo II expression, in a manner similar to that seen in oncogenic ras-induced and replicative senescent cells (Figure 4a). Thus, the dramatic downregulation of topo II is highly specific to senescent cells.

The regulation of histone deacetylases (HDACs) and histone acetyl transferases (HATs) is an important component of transcriptional control (Legube and Trouche, 2003). There are at least 11 HDACs identified in mammalian cells (Gray and Ekstrom, 2001; Zhou et al., 2001; Gao et al., 2002; Kao et al., 2002). The gene expression analysis indicated that HDAC9 was specifically and significantly upregulated in oncogenic ras-induced senescent cells (Table 2). Most other genes encoding HDACs remained unchanged with the exception of HDAC5, which was slightly downregulated in oncogenic ras-expressing cells (Table 2). RT–PCR analysis confirmed that HDAC9, but not HDAC1 or HDAC5, was upregulated in oncogenic ras-induced senescent cells (Figure 4b). HDAC9 is alternatively spliced to generate multiple isoforms that may have different biological activities (Petrie et al., 2003). One of these splice variants lacks the histone deacetylase catalytic domain. RT–PCR analysis showed isoforms with and without the catalytic domain were upregulated in oncogenic ras-induced senescent cells (data not shown). Significantly, HDAC9 expression levels were also increased in replicative senescent cells and serum starvation-induced senescent cells, but not in quiescent cells (Figure 4b). Real-time RT–PCR also validated the upregulation of HDAC9 in these senescent cells (data not shown).

To investigate whether the altered expression of topo II and HDAC9 was also seen in other senescent fibroblasts, oncogenic H-ras or an empty vector was transduced into early passage human MRC-5 fibroblasts via retroviral gene transfer. Significantly, MRC-5 cells stably transduced with oncogenic ras exhibited senescence phenotypes that were indistinguishable from oncogenic ras-induced senescent IMR90 cells, including the induction of SA--gal activity and SAHF formation (Figure 4c). Moreover, RT–PCR analysis showed that topo II, but not topo II, was significantly downregulated and HDAC9, but not HDAC1, was significantly upregulated in oncogenic ras-induced senescent MRC-5 cells (Figure 4d and data not shown). These data suggest that the gene expression changes of topo II and HDAC9 were highly specific in senescent fibroblasts.

In summary, we show for the first time that changes in topo II and HDAC9 expression was specifically associated with senescent human fibroblasts, suggesting that these two molecules may be novel biomarkers for senescence.

Discussion

Gene expression in oncogenic H-ras-induced senescence

The characteristics of replicative senescence and oncogenic ras-induced senescence are nearly indistinguishable in normal human IMR90 fibroblasts (Serrano et al., 1997). This current study further identified several molecular and cellular changes in oncogenic ras-induced senescent cells that are similar to replicative senescent cells, but not quiescent cells. These results suggest that oncogenic ras-induced senescence in normal human fibroblasts may provide a useful model system for studying the senescence mechanism.

Senescence is accompanied by the upregulation of p21 and p16 (Robles and Adami, 1998; Chang et al., 1999; Campisi, 2003). Overexpression of p21 can result in the downregulation of many genes involved in cell-cycle progression and DNA repair (Chang et al., 2000). However, abolishing p21 in fibroblasts is not sufficient to bypass senescence induced by oncogenic ras (Wei et al., 2003; Castro et al., 2004). In contrast, fibroblasts null for p16 or exhibiting low expression of p16 do not undergo senescence in response to oncogenic ras (Brookes et al., 2002; Huot et al., 2002; Narita et al., 2003; Benanti and Galloway, 2004), suggesting that the upregulation of p16 is necessary to induce senescence in human fibroblasts. Some cell types can undergo senescence independent of p16 or p21 (Shelton et al., 1999; Olsen et al., 2002), indicating that the molecules involved in mediating growth arrest vary depending on the cellular context. Our study shows that p15 is also induced in oncogenic ras-induced senescent cells. Consistent with these data, overexpression of p15 can induce senescence in fibroblasts (McConnell et al., 1998). Thus, p16, p21 and p15 may all contribute to the cell cycle arrest in oncogenic ras-induced senescence, although p15 is not upregulated in replicative senescent cells (data not shown) (McConnell et al., 1998).

Regardless of the mediator of growth arrest, to establish a permanent cell cycle arrest likely engages a common pathway. Oncogenic ras-induced and replicative senescence involves the recruitment of a repressor complex to the promoters of various E2F target genes, leading to their transcriptional repression and, hence, a halt in S phase progression (Narita et al., 2003). Among the E2Fs identified in mammalian cells, E2Fs 1–3 function primarily as transcriptional activators, while E2F4 and 5 act mainly as transcriptional repressors (Trimarchi and Lees, 2002). Interestingly, expression of E2F1, but not E2F4 or E2F5, was dramatically downregulated in senescent cells, raising the hypothesis that E2F4/5 may be involved in repressing the transcription of E2F1 and other E2F targets in senescence. Indeed, E2F4 is responsible for mediating the transcriptional repression of E2F1 and various E2F targets in nondividing cells (Trimarchi and Lees, 2002). The E2F proteins are known to interact with the pocket protein family members, Rb, p107 and p130 (Cam and Dynlacht, 2003). Expression of p107 was dramatically downregulated in senescent cells, perhaps due to the downregulation of E2F1 (Ren et al., 2002), suggesting that p107 may not be involved in mediating transcriptional repression.

Senescence and G₁ tetraploidy

In contrast to quiescence, oncogenic ras-induced senescence is accompanied by the accumulation of cells with 2 and 4 N DNA content and the appearance of binucleated cells. Interestingly, replicative senescent cells also accumulate with both 2 and 4 N DNA content with a significant percentage of cells with two nuclei (Sherwood et al., 1988; Ly et al., 2000), and that most of the replicative senescent cells with 4 N DNA content are likely G₁ tetraploid cells (Sherwood et al., 1988). The formation of G₁ tetraploidy is mainly triggered by aberrant mitosis or failure of cytokinesis, followed by the mitotic exit of tetraploid cells that undergo a G₁ arrest (Margolis et al., 2003). Consistent with this notion, our study and other studies show that senescent cells exhibit a dramatic downregulation of many G₂/M genes, including genes associated with centromere assembly and genes essential for cytokinesis (Shelton et al., 1999; Ly et al., 2000; Wells et al., 2003; Larsson et al., 2004). Hence, the dramatic downregulation of G₂/M genes in senescent cells may play an important role in triggering G₁ tetraploidy. While the specific mechanisms involved remains to be further elucidated, the existence of senescent G₁ tetraploid cells suggests that senescence is guarded by a multifaceted mechanism to ensure a tight growth arrest.

Senescence biomarkers

Currently, there are a limited number of markers available to determine whether a population of arrested cells may be senescent. The gene expression analysis from oncogenic ras-induced senescent IMR90 fibroblasts has allowed for the identification of genes whose altered expression may be specific to senescent cells. Here, we report the identification of two candidate molecules, topoisomerase II and HDAC9.

DNA topoisomerases play an important role in resolving topological problems of DNA in replication, transcription and chromatin structure (Wang, 2002). The type II class of enzymes, topo II and topo II, play very similar roles during interphase. In contrast, topo II is essential for condensation and separation of mitotic chromosomes, while topo II is dispensable in proliferating cells (Larsen et al., 2003). Moreover, topo II is upregulated in many cancer cells (Wang, 2002). We show that senescent cells exhibit a specific and dramatic reduction in topo II expression. Considering its essential role in cell proliferation, topo II may be a critical repression target by the senescence mechanism.

Chromatin structure can interfere with the essential processes of transcription and DNA replication. HAT and HDACs play essential roles in transcription by regulating chromatin structure (Legube and Trouche, 2003). The specific upregulation of HDAC9 in senescent human fibroblasts suggests that it may be a novel senescence biomarker and may play an important role in senescence. Indeed, senescent IMR90 fibroblast exhibited transcription repression of numerous genes and altered chromatin structure (SAHF), raising the hypothesis that HDAC may play a role in mediating transcription repression and/or SAHF formation (Narita et al., 2003). HDAC9 is a newly identified molecule and, therefore, its function is relatively less understood. However, it has been shown that HDAC9 can repress the transcription of the myocyte enhancer factor 2D (Zhou et al., 2001; Petrie et al., 2003), suggesting that HDAC9 can impact transcriptional regulation. Further investigation will be required to fully understand the role of HDAC9 in human cells and in senescence.

In summary, we report the study of the molecular signature of oncogenic ras-induced senescent human fibroblasts. We have identified several molecular and cellular changes in ras-induced senescent cells that are similar to replicative senescent cells, but not quiescent cells. Our study suggests that senescence is mediated by a multifaceted dynamic mechanism. Both activation and inactivation of cell cycle regulators occur in concert to cause a permanent cell cycle arrest accompanied by G₁ tetraploidy. Moreover, we show that the altered expression of topo II and HDAC9 occur specifically in senescent human fibroblasts and, therefore, may be novel senescence biomarkers. Further studies on the roles of these two molecules in senescence may shed light into the senescence mechanism.

Materials and methods

Cell culture and retroviral gene transfer

Human diploid IMR90 fibroblasts (ATCC), human diploid MRC-5 fibroblasts (ATCC), and the Phoenix retrovirus packaging cell line (ATCC) were cultured in DMEM (Mediatech) supplemented with 10% FBS (HyClone) and antibiotics, as previously described (Serrano et al., 1997). Oncogenic ras (H-RasV12) or an empty vector (pBabe-puro) was introduced into IMR90 (passage 5–7) or MRC-5 (passage 5–7) cells by retroviral gene transfer, as described previously (Lin et al., 1998). Replicative senescent cells were established by passaging IMR90 cells up to approximately 65 population doublings. Quiescent cells were generated by plating cells at low density and maintaining them in DMEM containing 0.1% FBS for 2 days. Serum starvation-induced senescent cells were generated by culturing IMR90 cells in DMEM containing 0.1% FBS for 21 days with media change twice a week (Weebadda et al., 2004).

Gene expression analysis

Total RNA was isolated from cell pellets and converted to cDNA, followed by in vitro transcription to cRNA. The biotinylated cRNAs were then hybridized to the Affymetrix HG133A Array Chip (Affymetrix, Santa Clara, CA, USA) following the prescribed Affymetrix protocols. The gene expression analysis was performed at the Gene Expression Facility at Roswell Park Cancer Institute. Expression value (signal) was calculated using Affymetrix Genechip software MAS 5.0 (for full description of the statistical algorithms see http://affymetrix.com/support/
technical/whitepapers/sadd_whi
tepaper.pdf).

Cell cycle analysis

Cell cycle analysis was performed as previously described with some modifications (Lin et al., 1998). Subconfluent cells were labeled with bromodeoxyuridine (BrdU; Sigma) and fluorodeoxyuridine (FdU; Sigma). Cells were then incubated with anti-BrdU antibody (Pharmingen). Alexa Fluoro 488 anti-mouse antibody (Molecular Probes) was used as the secondary antibody. Cell cycle distribution was analysed by two-dimensional flow cytometry (Cellquest, Becton Dickinson). Cell cycle analysis was performed using the WinList and ModFit LT analysis programs (Verity).

RT–PCR analysis

RNA was isolated from frozen cell pellets using the RNeasy RNA isolation kit (Qiagen). A measure of 1 g of total RNA was reverse transcribed with M-MLV-reverse transcriptase following the manufacturer's instruction (Invitrogen). PCR reactions were performed using the HotMaster Taq DNA Polymerase (Eppendorf), as described by the manufacturer. PCR products were resolved on agarose gels followed by ethidium bromide staining. Real-time PCR was performed with the iCycler (Bio-Rad) using the QuantiTect SYBR green PCR mix as described by the manufacturer (Qiagen). Sequences of PCR primers and reaction parameters are available upon request.

Western blotting

Whole cell lysates were generated by resuspending cell pellets in Laemmli sample buffer (Laemmli, 1970) and subjected to immunoblotting as described previously (Serrano et al., 1997). The following primary antibodies were used: p16 (C20; Delta Biolabs); p21 (C-19; Santa Cruz); RB (PharMingen); E2F1 (C20X; Santa Cruz); E2F4 (C20; Santa Cruz); p107 (C18; Santa Cruz); p130 (C20; Santa Cruz); topoisomerase II (Ab-1; Oncogene); c-H-ras (Ab-1; Oncogene); -tubulin (Sigma). HRP-conjugated goat anti-rabbit or mouse antibodies (Bio-Rad) were used as secondary antibodies.

Immunofluorescence

Immunofluorescence was performed as previously described with some modifications (Lin and Lowe, 2001). Cells were immunostained with anti-BrdU (PharMingen) and anti-p21 (C19; Santa Cruz) antibodies. Alexa Fluoro 488 anti-mouse antibody and Alexa Fluoro 594 anti-rabbit antibody (Molecular Probes) were used as secondary antibodies. DAPI (Sigma) was used to stain nuclei and stained cells were visualized using an immunofluorescent microscope (Zeiss).

SA--gal assay

Senescence was monitored by SA--gal staining as previously described in Dimri et al. (1995).

Comet assay

To assess DNA damage, the comet assay was performed with minor modifications as previously described in Singh et al. (1988). Cells were gently resuspended in PBS at 5.7 10⁶ cells/ml. Cells (2 10⁴) were plated on microscope slides. Slides were equilibrated in 0.4 M Tris, pH 7.5, immersed in methanol for 5 min, and immersed in ethanol for 5 min.

References

1. Altieri DC. (2003). Oncogene, 22, 8581–8589. | Article | PubMed | ISI | ChemPort |
2. Andreassen PR, Lohez OD, Lacroix FB and Margolis RL. (2001). Mol. Biol. Cell, 12, 1315–1328. | PubMed | ISI | ChemPort |
3. Benanti JA and Galloway DA. (2004). Cell Cycle, 3, 715–717. | PubMed |
4. Bodnar AG, Ouellette M, Frolkis M, Holt SE, Chiu CP, Morin GB, Harley CB, Shay JW, Lichtsteiner S and Wright WE. (1998). Science, 279, 349–352. | Article | PubMed | ISI | ChemPort |
5. Brookes S, Rowe J, Ruas M, Llanos S, Clark PA, Lomax M, James MC, Vatcheva R, Bates S, Vousden KH, Parry D, Gruis N, Smit N, Bergman W and Peters G. (2002). EMBO J., 21, 2936–2945. | Article | PubMed | ISI | ChemPort |
6. Cam H and Dynlacht BD. (2003). Cancer Cell, 3, 311–316. | Article | PubMed | ISI | ChemPort |
7. Campisi J. (1992). Ann. NY Acad. Sci., 663, 195–201. | PubMed |
8. Campisi J. (2001). Sci. World J., 1, 65.
9. Campisi J. (2003). Nat. Rev. Cancer, 3, 339–349. | Article | PubMed | ISI | ChemPort |
10. Castro ME, Guijarro Md Mdel V, Moneo V and Carnero A. (2004). J. Cell Biochem., 92, 514–524. | Article | PubMed |
11. Chang BD, Watanabe K, Broude EV, Fang J, Poole JC, Kalinichenko TV and Roninson IB. (2000). Proc. Natl. Acad. Sci. USA, 97, 4291–4296. | Article | PubMed | ChemPort |
12. Chang BD, Xuan Y, Broude EV, Zhu H, Schott B, Fang J and Roninson IB. (1999). Oncogene, 18, 4808–4818. | Article | PubMed | ISI | ChemPort |
13. Chen QM, Bartholomew JC, Campisi J, Acosta M, Reagan JD and Ames BN. (1998). Biochem. J., 332 (Part 1), 43–50. | PubMed | ISI | ChemPort |
14. Classon M and Harlow E. (2002). Nat. Rev. Cancer, 2, 910–917. | Article | PubMed | ISI | ChemPort |
15. Cong YS, Wright WE and Shay JW. (2002). Microbiol. Mol. Biol. Rev., 66, 407–425 (table of contents). | Article | PubMed | ISI | ChemPort |
16. Dimri GP, Lee X, Basile G, Acosta M, Scott G, Roskelley C, Medrano EE, Linskens M, Rubelj I, Pereira-Smith O, Peacocke M and Campisi J. (1995). Proc. Natl. Acad. Sci. USA, 92, 9363–9367. | Article | PubMed | ChemPort |
17. DiNardo S, Voelkel K and Sternglanz R. (1984). Proc. Natl. Acad. Sci. USA, 81, 2616–2620. | PubMed | ChemPort |
18. Gao L, Cueto MA, Asselbergs F and Atadja P. (2002). J. Biol. Chem., 277, 25748–25755. | Article | PubMed | ISI | ChemPort |
19. Gray SG and Ekstrom TJ. (2001). Exp. Cell Res., 262, 75–83. | Article | PubMed | ISI | ChemPort |
20. Greider CW. (2000). Harvey Lect., 96, 33–50. | PubMed | ChemPort |
21. Harley CB, Futcher AB and Greider CW. (1990). Nature, 345, 458–460. | Article | PubMed | ISI | ChemPort |
22. Hayflick L. (1965). Exp. Cell Res., 37, 614–636. | Article | PubMed | ISI | ChemPort |
23. Huot TJ, Rowe J, Harland M, Drayton S, Brookes S, Gooptu C, Purkis P, Fried M, Bataille V, Hara E, Newton-Bishop J and Peters G. (2002). Mol. Cell Biol., 22, 8135–8143. | Article | PubMed | ISI | ChemPort |
24. Kao HY, Lee CH, Komarov A, Han CC and Evans RM. (2002). J. Biol. Chem., 277, 187–193. | Article | PubMed | ISI | ChemPort |
25. Karlseder J, Smogorzewska A and de Lange T. (2002). Science, 295, 2446–2449. | Article | PubMed | ISI | ChemPort |
26. Kramer DL, Chang BD, Chen Y, Diegelman P, Alm K, Black AR, Roninson IB and Porter CW. (2001). Cancer Res., 61, 7754–7762. | PubMed | ISI | ChemPort |
27. Laemmli UK. (1970). Nature, 227, 680–685. | Article | PubMed | ISI | ChemPort |
28. Larsen AK, Escargueil AE and Skladanowski A. (2003). Prog. Cell Cycle Res., 5, 295–300. | PubMed |
29. Larsson O, Scheele C, Liang Z, Moll J, Karlsson C and Wahlestedt C. (2004). Cancer Res., 64, 482–489. | PubMed |
30. Legube G and Trouche D. (2003). EMBO Rep., 4, 944–947. | Article | PubMed | ISI | ChemPort |
31. Lin AW and Lowe SW. (2001). Proc. Natl. Acad. Sci. USA, 98, 5025–5030. | Article | PubMed | ChemPort |
32. Lin AW, Barradas M, Stone JC, van Aelst L, Serrano M and Lowe SW. (1998). Genes Dev., 12, 3008–3019. | PubMed | ISI | ChemPort |
33. Ly DH, Lockhart DJ, Lerner RA and Schultz PG. (2000). Science, 287, 2486–2492. | Article | PubMed | ISI | ChemPort |
34. Mallette FA, Goumard S, Gaumont-Leclerc MF, Moiseeva O and Ferbeyre G. (2004). Oncogene, 23, 91–99. | Article | PubMed | ISI | ChemPort |
35. Margolis RL, Lohez OD and Andreassen PR. (2003). J. Cell Biochem., 88, 673–683. | Article | PubMed | ISI | ChemPort |
36. McConnell BB, Starborg M, Brookes S and Peters G. (1998). Curr. Biol., 8, 351–354. | Article | PubMed | ISI | ChemPort |
37. Mollinari C, Kleman JP, Jiang W, Schoehn G, Hunter T and Margolis RL. (2002). J. Cell Biol., 157, 1175–1186. | Article | PubMed | ISI | ChemPort |
38. Narita M, Nunez S, Heard E, Lin AW, Hearn SA, Spector DL, Hannon GJ and Lowe SW. (2003). Cell, 113, 703–716. | Article | PubMed | ISI | ChemPort |
39. Nitiss JL. (1998). Biochim. Biophys. Acta, 1400, 63–81. | Article | PubMed | ISI | ChemPort |
40. Olsen CL, Gardie B, Yaswen P and Stampfer MR. (2002). Oncogene, 21, 6328–6339. | Article | PubMed | ISI | ChemPort |
41. Petrie K, Guidez F, Howell L, Healy L, Waxman S, Greaves M and Zelent A. (2003). J. Biol. Chem., 278, 16059–16072. | Article | PubMed | ISI | ChemPort |
42. Rangarajan A and Weinberg RA. (2003). Nat. Rev. Cancer, 3, 952–959. | Article | PubMed | ISI | ChemPort |
43. Ren B, Cam H, Takahashi Y, Volkert T, Terragni J, Young RA and Dynlacht BD. (2002). Genes Dev., 16, 245–256. | Article | PubMed | ISI | ChemPort |
44. Robles SJ and Adami GR. (1998). Oncogene, 16, 1113–1123. | Article | PubMed | ISI | ChemPort |
45. Roninson IB. (2003). Cancer Res., 63, 2705–2715. | PubMed | ISI | ChemPort |
46. Saavedra HI, Fukasawa K, Conn CW and Stambrook PJ. (1999). J. Biol. Chem., 274, 38083–38090. | Article | PubMed |
47. Serrano M, Lin AW, McCurrach ME, Beach D and Lowe SW. (1997). Cell, 88, 593–602. | Article | PubMed | ISI | ChemPort |
48. Shay JW and Roninson IB. (2004). Oncogene, 23, 2919–2933. | Article | PubMed | ISI | ChemPort |
49. Shelton DN, Chang E, Whittier PS, Choi D and Funk WD. (1999). Curr. Biol., 9, 939–945. | Article | PubMed | ISI | ChemPort |
50. Sherwood SW, Rush D, Ellsworth JL and Schimke RT. (1988). Proc. Natl. Acad. Sci. USA, 85, 9086–9090. | Article | PubMed | ChemPort |
51. Singh NP, McCoy MT, Tice RR and Schneider EL. (1988). Exp. Cell Res., 175, 184–191. | Article | PubMed | ISI | ChemPort |
52. Suzuki K, Mori I, Nakayama Y, Miyakoda M, Kodama S and Watanabe M. (2001). Radiat. Res., 155, 248–253. | Article | PubMed | ISI | ChemPort |
53. Trimarchi JM and Lees JA. (2002). Nat. Rev. Mol. Cell Biol., 3, 11–20. | Article | PubMed | ISI | ChemPort |
54. Uemura T, Ohkura H, Adachi Y, Morino K, Shiozaki K and Yanagida M. (1987). Cell, 50, 917–925. | Article | PubMed | ISI | ChemPort |
55. Wang JC. (2002). Nat. Rev. Mol. Cell Biol., 3, 430–440. | Article | PubMed | ISI | ChemPort |
56. Wang W, Chen JX, Liao R, Deng Q, Zhou JJ, Huang S and Sun P. (2002). Mol. Cell Biol., 22, 3389–3403. | Article | PubMed | ISI | ChemPort |
57. Weebadda WKC, Jackson T and Lin AW. (2004) in press.
58. Wei W, Jobling WA, Chen W, Hahn WC and Sedivy JM. (2003). Mol. Cell Biol., 23, 2859–2870. | Article | PubMed | ISI | ChemPort |
59. Wells SI, Aronow BJ, Wise TM, Williams SS, Couget JA and Howley PM. (2003). Proc. Natl. Acad. Sci. USA, 100, 7093–7098. | Article | PubMed | ChemPort |
60. Whitfield ML, Sherlock G, Saldanha AJ, Murray JI, Ball CA, Alexander KE, Matese JC, Perou CM, Hurt MM, Brown PO and Botstein D. (2002). Mol. Biol. Cell, 13, 1977–2000. | Article | PubMed | ISI | ChemPort |
61. Zhou X, Marks PA, Rifkind RA and Richon VM. (2001). Proc. Natl. Acad. Sci. USA, 98, 10572–10577. | Article | PubMed | ChemPort |
62. Zhu J, Woods D, McMahon M and Bishop JM. (1998). Genes Dev., 12, 2997–3007. | PubMed | ISI | ChemPort |

Acknowledgements

We thank J Platt and L Stein of the Gene Expression Core Facility at Roswell Park Cancer Institute for their assistance with the microarray analysis. We are also grateful to Dr J Black and Dr A Black for generously providing antibodies, Dr M Kimura and Dr H Nagase for assistance with real-time PCR, Dr M McHugh and Dr T Beerman for reagents and assistance with the comet assay, Dr F Li and Dr I Roninson (Ordway Research Institute, Inc.) for PCR primer sequences. We also thank Dr M Brattain, Dr C Porter, Dr J Black, Dr T Beerman, Dr W Weebadda and T Bihani for critical reading and helpful comments of this manuscript. This work was supported by the Roswell Park Cancer Center Core Grant CA16056 and the Greater Buffalo Community Foundation contract Grant number 6234401.

Supplementary Information accompanies the paper on Oncogene website (http://www.nature.com/onc).