Regulation of Replicative and Stress-Induced Senescence by RSK4, which is Down-regulated in Human Tumors

Cellular senescence can be defined as an irreversible arrest of cell proliferation. It is activated in response to various types of stress, including oxidative stress (e.g. hydrogen peroxide treatment), oncogene activities, DNA damage, and treatment with DNA-damaging agents (e.g. doxorubicin or cisplatin), among others. This is called accelerated or stress-induced senescence. Moreover, senescence can occur as a result of telomere shortening after multiple cell divisions, a process known as replicative senescence (1, 2). Once cells enter senescence, they stop growing and develop dramatic morphologic changes, such as a flat, enlarged morphology, as well as metabolic changes (3–7).

Two main pathways are reported to be involved in senescence, p16^INK4a/Rb and p19^ARF/p53, which are considered to be the main activators of senescence (7, 8). P16 activates Rb by inhibiting CycD/Cdk4,6. P19 activates p53 by inhibiting MDM2; p53 can be also activated by phosphorylation done by the ATM/ATR and/or Chk1/Chk2 proteins. P53 and Rb can be connected through p21, activated by p53, which, in turn, can activate Rb by inhibiting CycE/Cdk2. Once Rb is activated, it is able to shut down transcription of the E2F target genes, inducing cell growth arrest. In addition, it seems that p53 can activate senescence in human cells independently of Rb (8, 9). Whether both these senescence pathways or just one of them is necessary for inducing senescence remains a matter of controversy (10–16).

It has been shown that normal cells in culture can undergo oncogene-induced senescence in response to overexpression of the RAS oncogene (3, 17, 18) or its effectors, including activated mutants of RAF, MEK, and BRAF (8, 15, 19) acting as tumor suppressor events by preventing aberrant cell growth (20). Moreover, other oncogenes, such as cyclin E, STAT5, and CDC6, have been shown to induce senescence through a DNA damage response (see ref. 19 for review). Recent reports have established that oncogene-induced senescence occurs during the early stages of tumorigenesis (19, 21–23). It has been suggested that premalignant neoplastic lesions are rich in senescent tumor cells, restricting the growth of these lesions, whereas in most malignant tumors the majority of cells are proliferating (24, 25). Therefore, clinical use of senescence markers could be useful for detecting cancer at premalignant stages, thereby helping in the diagnosis and prognosis (19, 24).

In a large RNA interference (RNAi) screen, Berns et al. (26) identified five new modulators of p53-dependent proliferation arrest in human cells, raising the possibility that these genes are tumor suppressor genes. One of them was RPS6KA6 [ribosomal S6 kinase 4 (RSK4)]. RSK4 belongs to the RSK family [90-KDa ribosomal S6 kinase (p90RSK)], which, along with mitogen- and stress-activated protein kinase, constitutes a family of protein kinases that mediate signal transduction downstream of the mitogen-activated protein kinase cascades (27–29). RSK4 is activated in cells under serum-starved conditions for unknown reasons and has very low expression levels, detectable in colon, kidney, heart, cerebellum, and skeletal muscle (28, 30, 31). The biological function of RSK4 is still poorly understood. An expression screen with mouse mRNA in Xenopus laevis embryos showed that RSK4 can disrupt mesoderm formation induced by the RAS-ERK pathway and was therefore proposed as an inhibitor of growth factor signal transduction (32). Depletion of RSK4 bypassed p53-dependent G₁ cell cycle arrest and suppressed mRNA expression of cyclin-dependent kinase inhibitor p21^cip1 (26). Recently, it has been described that overexpression of RSK4 resulted in reduced colony formation in soft agar and suppressed invasive and migratory activities of the breast carcinoma cell line MDA-MB-231 both in vitro and in vivo (33). Together, these studies suggest that RSK4 is distinct from RSK1-3 and seems to have growth inhibitory function.

In a previous article, we reported that RSK4 mRNA was down-regulated in colon carcinomas. In the present study, we analyzed a larger series of colon carcinomas as well as renal cell carcinomas, and again observed significant down-regulation, which was already present in colon adenomas. Based on the idea that RSK4 down-regulation could be an early step in cell immortalization in cancer, we studied the effects of RSK4 in several cell lines. We observed that RSK4 overexpression is able to induce a senescence-like phenotype in normal and tumor cells in culture, regardless of p53 and p16 status. Moreover, its levels are increased in cells undergoing replicative senescence and stress-induced senescence; after blocking RSK4 mRNA with short hairpin RNA (shRNA), IMR90 cells became resistant to replicative senescence and more resistant to H₂O₂ and cisplatin treatment. The colon carcinoma cell line HCT116 also became resistant to cisplatin treatment. This converts this protein into a potentially crucial factor for controlling senescence in some human cells.

Patients. Normal tissue and tumor tissue from 30 patients with colon carcinoma, and 20 patients with renal carcinoma were randomly chosen from the tumor bank at the Pathology Department of Vall d'Hebron Hospital (Barcelona, Spain). Adenoma tissue samples from seven patients with colon carcinoma were also available. Fifteen of the carcinoma samples have been previously described (31). Biopsied samples were quick-frozen and stored at −80°C immediately after surgery. All tumors were histologically examined to confirm the diagnosis of carcinoma. All the procedures used in the study were approved by the Ethics Committee of Vall d'Hebron Hospital.

RNA extraction and quantitative RT-PCR. Total RNA was isolated from normal and tumor tissue with the RNeasy Mini Kit (Qiagen) following the manufacturer's instructions. Random primers and SuperScript II reverse transcriptase (Invitrogen) were used to carry out cDNA synthesis from 1.5 μg of total RNA. RSK4 expression was detected using the Taqman Gene Expression Assay (Hs00179523_m1; Applied Biosystems). An ABI PRISM 7000 instrument (Applied Biosystems) was used to do the relative quantification analysis, and data were analyzed with the 7000 Sequence Detection Software, v.1.2.3 (Applied Biosystems). The PCR cycling program consisted of denaturing at 95°C for 10 min and 40 cycles at 95°C for 15 s, and annealing and elongation at 60°C for 1 min. The reactions were done in triplicate.

Previously, a Taqman Human Endogenous Control Plate (Applied Biosystems) was done to determine which endogenous controls showed less variation between normal and tumoral tissue. POLR2A (Hs00172187_m1; Applied Biosystems) was chosen for colon samples and PPIA (4326316E; Applied Biosystems) for renal samples. Target and reference genes showed similar, nearly 100% amplification efficiencies (data not shown). Therefore, the ΔΔCT method was appropriate for relative gene expression analysis.

Statistical analysis. Statistical comparisons were made with the Statistical Package for Social Science, version 11.5 (SPSS, Inc.). The Wilcoxon test was used to compare differences between normal and tumor tissue.

Plasmid construction. pLPCX-RSK4 was constructed from pCDNA3.1-Higro-RSK4, kindly donated by R. Bernards (The Netherlands Cancer Institute, Amsterdam, Netherlands); pCR2.1 (Invitrogen); and pLPCX (Clontech), a retroviral vector with puromycin resistance. First, the RSK4 gene was PCR-amplified from pCDNA3.1-Hygro-RSK4 using T7 and BGH primers (Invitrogen). The PCR product was ligated to the linear vector pCR2.1 to obtain pCR2.1-RSK4. Then, pLPCX-RSK4 was constructed by cloning the respective cDNA fragment from pCR2.1-RSK4 into the pLPCX vector. pLPCX-GFP, kindly donated by S. Gutkind (NIH), was used as a control for virus infections.

The inducible retroviral vector RT-E1A-ER was kindly provided by A. Serrano (Hospital 12 de Octubre, Madrid, Spain). E1A was induced with 1 μmol/L of tamoxifen treatment.

Before the shRSK4-pLKO.1 construction, three different validated short interfering RNAs (siRNA) for RSK4 were tested (AM16708A; Ambion) following the manufacturer's instructions; siRNA 945 (GGUAAAUGGUCUUAAAAUG) was selected as it presented the highest RSK4 down-regulation (data not shown). Using this siRNA sequence, the shRNA for RSK4 GATCCGGTAAATGGTCTTAAAATGTTCAAGAGACATTTTAAGACCATTTACCTTTTTTACGCGTG was designed and cloned into the pLKO.1 puro vector (Sigma-Aldrich) and a Non-Target shRNA vector (Sigma-Aldrich) was used as control for lentivirus infections.

Cell culture. The following cell lines were used in this study: human primary fibroblasts, IMR90 (American Type Culture Collection), and colon carcinoma cells HCT116 p53 wt and HCT116 p53 null (kindly donated by B. Vogelstein, The John Hopkins Oncology Center, Baltimore, MD). All cells were grown in DMEM supplemented with 10% FCS and antibiotics. HCT116 p53 wt and HCT116 p53 null cell lines were maintained in culture supplemented with 200 μg/mL of gentamicin.

The IMR90-Tx-E1A cell line was generated by retrovirus infection with the E1A tamoxifen induction vector RT-E1A-ER.

Retrovirus-based gene transduction. Retrovirus-based gene transduction was carried out using a packaging cell line (GP-293; Clontech) according to the manufacturer's instructions. After two consecutive virus infections, cells were selected with puromycin (0.7 μg/mL for HCT116, 1.25 μg/mL for IMR90) for 3 d to eliminate uninfected cells.

Lentivirus-based shRNA transduction. Lentivirus-based transduction was carried out using a packaging cell line, HEK293T, by cotransfection of pCMV-dR8.91 dvpr (kindly donated by D. Trono, Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland) and VSV-G (Clontech). After two consecutive virus infections, cells were selected with puromycin for 3 d.

siRNA transfection. Validated siRNA for Rb (s523; Ambion), Silencer negative control siRNA (AM4611; Ambion), and the positive transfection control Silencer Cy3-labeled GADPH siRNA (AM4623; Ambion) were transfected into HCT116 p53 wt cells, previously transduced with RSK4 or GFP. After selection, 7.5 × 10⁴ cells were seeded in a 12-well plate and 15 nmol/L of each siRNA were transfected using Lipofectamine 2000 Reagent (Invitrogen), following the manufacturer's instructions.

Growth curves. Two days after selection, cells were counted and seeded in duplicate every 4 d. Population doublings (PD) were determined by the following formula: PD = Log (N_f/N_i)/Log2, where N_f is the number of cells counted and N_i is the number of cells seeded. Cumulative PD numbers (PDL) represent the sum of PDs from previous passages. For HCT116 p53 wt and HEK293T, 1.8 × 10⁵ cells were seeded. For HCT116 p53 null and MDA-MB-468, 2.5 × 10⁵ cells were seeded. Each curve was done at least twice with similar results, and each time point was determined in duplicate.

For cisplatin (CDDP) and hydrogen peroxide (H₂O₂) resistance growth curves, cells were plated into 24-well plates (2 × 10⁴ cells for IMR90; 1.5 × 10⁴ cells for HCT116) and treated for 2 h with H₂O₂ or for 3 d with CDDP with the concentrations indicated below. After treatment, fresh medium was added for 3 d. Then cells were washed with PBS, fixed in 10% formalin, and rinsed with distilled water. Cells were stained with 0.1% crystal violet (Sigma) for 30 min, rinsed thoroughly, and dried. Cell-associated dye was extracted with 2 mL 10% acetic acid and the optical density was determined at 590 nm. A value of 100% was assigned to untreated control cultures. All points within each experiment were determined in duplicate, and each experiment was done at least twice.

Replicative senescence induction and treatment with oxidative and DNA-damaging agents. Replicative senescence in IMR90 was attained after 40 passages in culture. Oxidative and DNA-damage stress induction in HCT116 cell line was carried out by 2 h of treatment with H₂O₂ 500 μmol/L or 6 h of treatment with 20 μg/mL of CDDP.

Analysis of senescence. Senescence-associated β-galactosidase (SA-β-gal) activity, a biomarker of cellular senescence (34), was determined with the Senescence β-galactosidase kit (Cell Signaling), following the manufacturer's instructions.

Western blot. Lysates were obtained from both cell lines and tissues. Tissues were ground and sonicated in a lysis buffer (HEPES 50 mmol/L, pH 7.5; NaCl 150 mmol/L; 1% Triton X; EDTA 1 mmol/L; 10% glycerol) in the presence of protease and phosphatase inhibitors, whereas subconfluent cells were lysed in the same buffer for 10 min at 4°C. After the lysates were cleared by centrifugation, protein concentrations were determined using the Bradford assay (Bio-Rad Protein Assay). About 50 to 100 μg of protein were denaturated and resolved on SDS-PAGE and transferred to polyvinylidene difluoride membranes (Bio-Rad). The primary antibodies used were anti-RSK4 (sc-17178; Santa Cruz Biotechnologies; diluted 1:100), anti-β-actin (CP01; Calbiochem; diluted 1:7,000), anti-p21^WAF1 (MS-891; Neomarkers; diluted 1:500), anti-p16 (DB018; Deltabiolabs; diluted 1:200), anti-Rb (G3-245; BD Biosciences; diluted 1:200), and anti-p53 clone DO-7 (M7001; Dako; diluted 1:500). The secondary antibodies used were donkey antigoat IgG-HRP (sc-2020; Santa Cruz Biotechnologies; diluted 1:5,000), sheep antimouse IgG-HRP (NA9310; Amersham Pharma-Biotech; diluted 1:3,000), and donkey antirabbit IgG-HRP (NA9340; Amersham Pharma-Biotech; diluted 1:2,000). Bound antibodies were visualized with an enhanced chemiluminescence detection kit (Amersham Pharma-Biotech).

RSK4 is down-regulated in human tumors. RSK4 was found to be down-regulated in colon and renal tumor tissues as compared with normal colon and renal tissues, respectively (Fig. 1A). Down-regulation was detected by quantitative real-time PCR (Wilcoxon test; P < 0.001 and P < 0.01, respectively), and confirmed by Western blot in a subset of tumors (Fig. 1B). RSK4 mRNA showed a 5-fold mean reduction in colon carcinomas and 10-fold reduction in renal carcinomas. Among the 30 patients with colon carcinoma studied and the 20 with renal carcinoma, RSK4 was down-regulated in 27 and 16 patients, respectively. No clinicopathologic correlation was observed in patients with mRNA down-regulation. The following parameters were considered: tumor size, infiltration, Duke stage, presence of lymph node metastases, survival, and tumor-node-metastasis staging.

In addition, we investigated at what stage of colon tumor progression RSK4 down-regulation started. We analyzed RSK4 mRNA expression in seven normal, seven adenoma, and seven carcinoma samples (Fig. 1C). Down-regulation was significant (P < 0.01) in adenomas, but expression differed between adenoma and carcinoma (P < 0.05). Moreover, we were able to analyze RSK4 expression in normal, adenoma, and carcinoma samples from the same individual for four patients, and progressive down-regulation of RSK4 was observed in all of them (Fig. 1D).

Transduction of RSK4 induces cell growth arrest in colon carcinoma cell lines. To determine the effect of RSK4 in tumor cell lines, we chose two colon carcinoma cell lines (HCT116 p53 wt and HCT116 p53 null) with no endogenous expression of RSK4. The RSK4 gene was transduced into these cell lines via retroviral infection. We observed dramatic morphologic changes characteristic of senescent (35) cells together with an increment in SA-β-gal activity 6 days postselection (Fig. 2A). This effect occurred in both the HCT116 p53 wt and p53 null cell lines. Growth curve analysis showed that cells expressing RSK4 did not accumulate in culture (Supplementary Fig. S1). Western blot analysis confirmed that these cells expressed RSK4, whereas cells infected with the GFP gene showed no RSK4 expression (Fig. 2B). To determine which of the senescence pathways (p16^INK4a/Rb or p19^ARF/p53) was activated, we analyzed the expression of senescence-related proteins, such as p21, p53, and Rb (Fig. 2B). Induction of the senescence-like phenotype by RSK4 correlated with an increase in p21 expression levels. However, this induction was independent of p53 because it was observed in both p53 wt and p53 null cells. P53 expression was not affected in HCT116 p53 wt by RSK4 overexpression. Both cell types showed an increment of the underphosphorylated form of Rb. Expression of p16 was not analyzed because HCT116 cells are characterized by mutations in this gene (a frameshift mutation in one allele and a promoter methylation in the other allele; ref. 36).

Transduction of RSK4 induces cell growth arrest in IMR90 cells. To assay the effect of RSK4 in normal cells, the RSK4 gene was transduced in IMR90 primary human diploid fibroblasts. Six days after selection, RSK4 expression induced morphologic changes and cells began to display a flat and enlarged morphology similar to the senescence morphology described previously for this cell line (3). Cells expressing RSK4 also had SA-β-gal activity (Fig. 2A), and senescence-associated heterochromatin foci (SAHF) formation was observed with 4’, 6-diamidino-2-phenylindole staining (data not shown). These effects indicated that RSK4 induced a senescence-like phenotype. The senescence-related proteins p53, p21, and p16 were analyzed and an increment of p21 and p16 was observed (Fig. 2B).

RSK4 is up-regulated after induction of senescence in a normal fibroblast cell line. To investigate whether RSK4 is a senescence modulator, we tested RSK4 expression levels in IMR90 after inducing senescence by cumulative passages. Cells stopped proliferating, acquired the enlarged, flattened morphology typical of senescent cells, and were positive for SA-β-gal staining (data not shown). The senescent cells showed increased RSK4 expression on quantitative real-time PCR and Western blot (Fig. 3A).

RSK4 is up-regulated after treatment with H₂O₂ or CDDP in colon carcinoma cell lines. After observing the up-regulation of RSK4 in normal senescent cells, we wanted to determine whether this occurs in tumor cell lines and whether the event is p53-dependent. Interestingly, the same up-regulation effect was observed in carcinoma cell lines (HCT116 p53 wt and p53 null) after H₂O₂ or CDDP treatment (Fig. 3B). Moreover, the effect was p53-independent, as both cell lines showed the same results.

RSK4 inhibition induces senescence resistance. Because RSK4 was increased in replicative senescence and stress-induced senescence, we then investigated whether RSK4 inhibition could confer resistance to senescence induced through different pathways. To assess this objective we infected normal IMR90 cells and colon carcinoma cells with shRSK4-pLKO.1 or shNonTarget-pLKO.1 vector and checked the RSK4 inhibition by quantitative real-time PCR (Supplementary Fig. S3). We observed that IMR90 cells with inhibited RSK4 became immortalized and did not show senescence features at passage 40 of lifespan, whereas control IMR90 cells (IMR90-shNonTarget) were senescent (Fig. 4A). Moreover, IMR90 transduced with shRSK4-pLKO vector were more resistant to H₂O₂ and CDDP treatment (Fig. 4B and C) and showed a smaller number of SA-β-gal-positive cells (data not shown). Interestingly, HCT116 cells lacking RSK4 expression showed a clear increase in survival after treatment with DNA-damaging agents (Fig. 4D), but not with H₂O₂ (data not shown).

Mode of action of RSK4-induced senescence-like phenotype. As shown in Fig. 2, RSK4 overexpression induced cell growth arrest in normal IMR90 fibroblasts and in malignant carcinoma cells, regardless of p53, p16, and K-Ras status (K-RAS is mutated in HCT116 cells). RSK4 overexpression was associated with an increase of p21 protein in all these cell lines, and with an increase of p16 in IMR90 cells. Note that HCT116 carcinoma cells do not express p16. Thus, these data seemed to suggest that RSK4-mediated senescence is not related to p53 or p16, and that other downstream factors may be involved. E1A efficiently negates the cell cycle arrest induced by oncogenic RAS (3). To test whether E1A can bypass the senescence-like phenotype induced by RSK4 overexpression, the RSK4 gene was transduced into the IMR90-Tx-E1A cell line, and E1A was then induced with tamoxifen. E1A was able to immortalize fibroblasts expressing RSK4 at the same level as cells with GFP (Fig. 5A). Western blot analysis confirmed expression of RSK4 and E1A in these cells (Fig. 5B). To test whether Rb could be a central target for mediating RSK4 senescence, we transduced RSK4 or GFP in HCT116 p53 wt. After selection, cells were transfected with Silencer Cy3-labeled GADPH siRNA, Silencer negative control siRNA, or Rb siRNA. Three days after transfection, SA-β-gal activity was determined. Silencer Cy3-labeled GAPDH siRNA was used for monitoring delivery of siRNAs (Fig. 6). Cells encoding RSK4 and Rb siRNA resumed growth, whereas cells harboring RSK4 and NonTarget siRNA were arrested and were positive for SA-β-gal staining. RSK4 and Rb expression was checked by Western blot (Supplementary Fig. S4).

RSK4, a member of the RSK family, is involved or expressed in specific clinical and physiologic situations, such as embryonal development, p53-dependent cell cycle arrest, and even in starving conditions (26, 28, 37). The observation that RSK4 becomes activated in situations of starving or a lack of growth factors is intriguing, as it suggests that RSK4 autophosphorylation could be a way to maintain cell life under conditions of stress.

In a preliminary study of the five genes associated with p53 senescence identified by Berns et al. (26), we found that RSK4 was significantly down-regulated in colon carcinomas as compared with normal tissue (31). In the present study, we have extended the number of colon tumors, and investigated RSK4 levels in renal cell tumors and colon adenomas, focusing on its role in in vitro senescence. Interestingly, RSK4 mRNA levels decreased significantly in colon adenomas with respect to normal mucosa. These results seem to indicate that decreased RSK4 expression may be an early event in colon tumorigenesis. As supported by in vitro studies with knocked-down RSK4 cells, which became immortalized, these results suggest that an absence of RSK4 expression could be relevant in benign adenomas harboring a large population of immortalized cells.

Abrogation of cellular senescence has been shown to be the first step in creating an in vitro model of human cancer (12, 19, 38). Moreover, in colon carcinomas where the number of senescent cells is supposed to be extremely low, the level of RSK4 was also minimal. Therefore, we propose RSK4 as a putative novel tumor suppressor and that its absence results in failure of senescence induction, which, combined with other alterations, possibly leads to tumor formation. It is well known that in colon adenomas, cells acquire cell growth autonomy by different oncogenic targets that can be activated, including growth factor receptors, K-RAS mutations, and others. In this context, RSK4 down-regulation could prevent oncogenic-induced senescence. Moreover, similar results were obtained in renal cell carcinomas, which showed a highly significant decrease of RSK4 mRNA levels with respect to normal kidney cells. It is noteworthy that colon and kidney are among the tissues that show the highest RSK4 expression (28, 30, 31). Hence, in these tissues, RSK4 down-regulation could be indicative of early steps in cell transformation.

Based on this evidence, we decided to work with normal and tumor cell lines to determine whether RSK4 can induce senescence when overexpressed, and to investigate which pathways might be involved. Interestingly, RSK4 overexpression induced marked cell growth arrest and a senescence-like phenotype in colon carcinoma cell lines (HCT116 p53 wt and p53 null) and in primary human fibroblasts (IMR90) confirming in vitro that high levels of RSK4 can stop immortalization, acting as a tumor suppressor mechanism.

Going further, we investigated whether cells in replicative or stress-induced senescence had high levels of RSK4 mRNA. In IMR90 normal fibroblasts exposed to replicative senescence induction, we observed an increase of RSK4, with levels up to 70% higher. In the HCT116 colon carcinoma cell line exposed to H₂O₂ treatment, RSK4 mRNA levels increased up to 6-fold in HCT116 wt p53 cells and 4-fold in HCT116 p53 null cells. After treatment with CDDP, the increase was 4.5-fold and 6.5-fold in HCT116 wt and null cell lines, respectively. Various experiments have indicated that inactivation of p53 generally delays the onset of senescence (5, 11, 13, 39, 40). This could explain why the two types of cells collected on the same day progressed to senescence at different rates and showed different RSK4 levels. It has been suggested that oxidative stress due to H₂O₂ treatment induces senescence through p53-p21-Rb (38), although other studies have indicated that p16 can also be activated, possibly through the action of p38-MAPK protein (1, 7). Taking into account these data, we hypothesized that RSK4 could be a mediator of stress-induced and replicative senescence. To investigate this idea, we introduced RSK4 shRNA into HCT116 p53 wt and null cells. After treatment with CDDP, cells lacking RSK4 became more resistant, whereas no clear differences were noted after H₂O₂ treatment, although we had observed that RSK4 expression increased in H₂O₂-induced senescence. These data may indicate that RSK4 is necessary but not indispensable to achieve growth arrest in this stress situation. Interestingly, p53 wt and null HCT116 cells showed a similar increase in resistance to CDDP when RSK4 was knocked out. These results could indicate that apoptosis was not the only factor involved in the cell number decrease. When RSK4 was blocked in IMR90 fibroblasts, the cells became more resistant to both H₂O₂ and CDDP, indicating that different cell types may utilize different pathways in the presence of stress signals.

Due to the fact that a senescence-like phenotype was induced in cells with p53 null mutations, K-RAS mutations, and lack of p16, among other genetic alterations, it seems that RSK4-induced senescence is not dependent on these specific factors. Because it is well known that the p53 and Rb proteins play the most critical roles in induction of senescence (7), we thought that Rb might be the key factor in this context. We studied the effect of RSK4 on E1A-expressing IMR90 cells and found that they were resistant to RSK4-induced senescence. Moreover, senescence was not induced when Rb was knocked down in HCT116 p53 wt, suggesting that Rb could be an essential pathway to drive cells to senescence via activation of RSK4.

These results raise the point that RSK4 may be an important mediator of several pathways driving cells to senescence. The biochemical pathways controlled or regulated by RSK4 are not well defined or understood. As occurs in p16 and p21, RSK4 levels increase in replicative senescence. In our preliminary approach, we observed that RSK4 overexpression induced cell growth arrest regardless of p53 status in malignant cells carrying many oncogenic alterations, including K-RAS mutations and lack of p16. Senescence-like phenotypes were not observed after RSK4 overexpression in cells carrying adenovirus E1A.

Taking our clinical data and data from basic studies together, we propose that RSK4 may play a relevant role in senescence and immortalization control in at least some cell types that display a relevant mRNA baseline RSK4 level, such as normal colon and kidney cells. RSK4 mRNA down-regulation at the early stages of tumorigenesis could enable cells to become immortalized and avoid senescence after accumulation of genetic alterations, such as K-RAS mutations. Recent work on RSK4 in the MDA-MB-431 breast cancer cell line has indicated that RSK4 expression exerts a tumor suppressor effect by anti-invasive and antimetastatic activities through regulation of claudin-2 (33). Our results agree with the findings of this study and support the concept of RSK4 as a new tumor suppressor gene whose function deserves to be completely explored, with investigation into the biochemical pathways and factors through which RSK4 can mediate senescence.

We propose that several, still poorly understood biochemical pathways can control senescence in addition to the p53 and p16 proteins. Factors such as RSK4, with as yet not completely defined biochemical substrates, may act in a way similar to the so-called funnel factors (41) and drive senescence signals, becoming a potential hallmark factor in senescence-mediating pathways.

No potential conflicts of interest were disclosed.

We are grateful to the tumor banks of Hospital Vall d'Hebron and Xarxa del Banc de Tumors de Catalunya for providing the tumor samples, and thank A. Carnero, R. Sánchez-Prieto, and M. Serrano for their careful reading and suggestions, and C. Cavallo for English editing.