CARF Is a Vital Dual Regulator of Cellular Senescence and Apoptosis*

The tumor suppressor protein, p53, is central to the pathways that monitor the stress, DNA damage repair, cell cycle, aging, and cancer. Highly complex p53 networks involving its upstream sensors and regulators, downstream effectors and regulatory feedback loops have been identified. CARF (Collaborator of ARF) was shown to enhance ARF-dependent and -independent wild-type p53 function. Here we report that (i) CARF overexpression causes premature senescence of human fibroblasts, (ii) it is vital for replicative and stress-induced senescence, and (iii) the lack of CARF function causes aneuploidy and apoptosis. We provide evidence that CARF plays a dual role in regulating p53-mediated senescence and apoptosis, the two major tumor suppressor mechanisms.

The tumor suppressor protein, p53, is the most frequently inactivated protein in human cancers that symbolizes deregulation of genomic stability, cell cycle, senescence, and stress damage-repair response of cells (1)(2)(3)(4)(5). p53 signaling has been established as a complex network comprised of (i) upstream components consisting of stress signals and sensor proteins including kinases, transferases, methyalses, ligases, and others that regulate its activity either by post-translational modifications or by subcellular localization, (ii) core regulators including an upstream positive regulator (alternative reading frame, ARF) 4 protein that blocks its downstream effector and antago-nist human double minute-2 (HDM2) protein, and (iii) the downstream effectors that determine the fate of cells by instigation of growth arrest, senescence, or apoptosis, the three potent tumor suppressor mechanisms. In addition to the fact that the initiation of DNA damage-induced senescence and establishment of growth arrest require p53 activation (4), a large number of studies have shown that the persistent inactivation of p53 is required for tumor maintenance. Cancer cells undergo either growth arrest or apoptosis with restoration of wild-type p53 function in vitro and in vivo (6 -8). These studies have prioritized further understanding of p53 signaling and regulation that would have major impact in cancer drug development. It is yet to be resolved how functional restoration of p53 culminates to growth arrest/senescence in some cells and apoptosis in others. Is it driven by the level of p53 expression, modulating partner proteins or its upstream regulators?
Among the large number of p53-binding proteins that influence its activities, ARF and HDM2 have been demonstrated as its major regulators. HDM2 (an E3 ubiquitin ligase) is transcriptionally activated by p53 and acts to degrade p53 in turn; thus, executing a negative feedback loop on p53 activity. ARF has been shown to bind and inhibit HDM2 activity resulting in the activation of p53 pathway (9). CARF (Collaborator of ARF) protein was initially cloned as an ARF-binding protein by yeast two-hybrid screening and was shown to activate ARF-dependent and -independent p53 functions (10 -13). CARF interacts not only with ARF but also with p53 and HDM2, and thereby executes a new feedback loop in p53 pathway: CARF activates p53 3 activates HDM2 3 degrades CARF and p53. CARF also regulates HDM2 by its transcriptional repressor function (10 -12, 14). In the present study, we demonstrate that CARF is vital for senescence of normal human fibroblasts. Whereas the overexpression of CARF induces premature senescence, its suppression leads to genomic instability and apoptosis.

Construction of Retroviral Expression
Vectors-For construction of retrovirus-driven expression of CARF-Myc, pCX4neo (provided by Dr. Tsuyoshi Akagi, Osaka, Japan) was used.
cDNA-encoding Myc-tagged CARF protein was cloned into the BamHI and EcoRI sites of the vector. For pCX4neo-CARF-GFP construction, CARF cDNA was first cloned into the EcoRI and BamHI site of the pEGFP-N1 vector (Clontech), and then CARF-GFP fragment was isolated by EcoRI-NotI digestion and subcloned into the EcoRI-NotI site of the pCX4neo vector.
Cell Culture-Cells (HeLa, U2OS, TIG-1, MRC5 and the multicolor MDA-MB-435 cells stably expressing GFP-Aurora A kinase, DsRed-Dimportin ␣2, and CFP-histone H3 as markers for the centrosomes/mitotic spindles, nuclear membrane, and the nucleus/chromosomes, respectively) were grown at 37°C in a 5% CO 2 incubator in low glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 50 units/ml penicillin, and 5 mg/ml streptomycin. U2OS cells with Zn-inducible CARF expression were prepared as described earlier (11). p53 Ϫ/Ϫ and MDM2 Ϫ/Ϫ fibroblasts were a kind gift from Dr. Lozano (15) For the production of retroviruses, Plat-E, an ecotropic murine leukemia virus (MuLV)-packaging cell line, was transfected with the pVPack-GP (expression gal and pol) and pVPack VSVG (expressing env) vectors (Stratagene, CA), with CARF expressing pCXneo retroviral vector either as pCX4neo/CARF-Myc or pCX4neo-CARF-GFP using FuGENE6 (Roche Applied Science). After 48 h, culture supernatants were collected, passed through 0.45-m filters and used as viral stocks for infection. TIG-1 and MRC5 cells in 10-cm dishes were treated with 8 g/ml polybrene at 37°C for 1 h, following which cells were infected with 1 ml of filtered viral stock for 1 h, then 2 ml of Dulbecco's modified Eagle's medium was added, and the plates were incubated at 37°C for another 48 h. Cells were selected in medium containing G418 (200 g/ml) until stable cell lines were obtained. The expression of CARF was examined by Western blotting and immunostaining using anti-Myc or anti-GFP antibody as indicated. For growth rate analysis, 4 ϫ 10 4 cells were seeded in 35-mm dish in duplicates. The cells were trypsinized and counted everyday for next 5 days. For induction of oncogene-induced SIPS, U2OS cells were transfected with expression plasmids encoding either the wild-type Ras (Wt Ras) or mutant oncogenic Ras (Mut Ras, G12V). Cells were selected in medium supplemented with G418 (500 g/ml) for 15 days and were then examined for Ras and CARF levels by Western blotting as described below.
CARF-compromised cells were generated by small interference RNA (siRNA) treatment. CARF-targeting 21-nucleotide RNAs were chemically synthesized using Expedite RNA phosphoramidites and thymidine phosphoramidite. Synthetic oligonucleotides were deprotected and gel-purified. Sequences of two control and two target oligos for CARF were 5Ј-AAGAC-CGAGUCCAUGAGGCUT-3Ј, 5Ј-GCCUCAUGGACUCGG-UCUUUT-3Ј and 5Ј-CGGAGUACCUGAGCCAGAAUT-3Ј, 5Ј-UUCUGGCUCAGGUACUCCGUT-3Ј, respectively. For annealing of siRNAs, 20 mM of two control or target single strands were incubated in annealing buffer (100 mM potassium acetate, 30 mM HEPES-KOH at pH 7.4, 2 mM magnesium acetate) for 1 min at 90°C followed by 1 h at 37°C. Transfections of siRNA duplexes were carried out using Oligofectamine reagent (Invitrogen). Typically, 1-5 l of the 20 mM duplexes were used per 12-well dish and were assayed after 24 -48 h by immuno-staining and Western blotting with anti-CARF antibody as described below.
Western Blotting-Cells treated with CARF and control siRNAs were harvested at 72-h post-transfection by resuspending cell pellets with lysis buffer containing 50 mM HEPES pH 7.5, 150 mM NaCl, 100 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 0.5% Triton X-100, 0.5% Nonidet P-40, 1 mM dithiothreitol, 10% glycerol, and supplemented with a protease inhibitor mixture (Roche). The whole cell lysate was centrifuged at 13,000 rpm for 10 min at 4°C to remove cell debris. The protein concentrations were determined by the Bradford protein assay following the manufacturer's instructions. Lysates containing 20 -30 g of protein were boiled in sodium dodecyl sulfate (SDS) sample buffer for 5 min, subjected to SDS-polyacrylamide gel electrophoresis under reducing conditions and then transferred onto Immobilon-P membranes (Millipore) using a semi-dry transfer apparatus (ATTO, Japan). After washing with TBS-T, the membrane was incubated with horseradish peroxidase-conjugated secondary antibody (goat anti-mouse or antirabbit from Santa Cruz Biotechnology) for 30 min. The membrane was washed three times in TBS-T and once in TBS, then subjected to enhanced chemiluminescence (ECL)-mediated visualization (Amersham Biosciences) using Lumino Image Analyzer equipped with CCD camera (LAS3000-mini, Fuji Fluorescence and Confocal Microscopy-Fluorescence and laser confocal microscopy were used to observe cell phenotypes, cell cycle stages, and immunofluorescence. Cells grown on coverslips were fixed in 4% formaldehyde, permeabilized with 0.2% Triton X-100 and blocked with 5% bovine serum albumin in PBS. Cells were probed with primary antibodies against ␤-tubulin (monoclonal, Sigma), CARF (FL-A10, rabbit polyclonal), and phospho-H3 (monoclonal, Santa Cruz Biotechnology). Staining was visualized by Alexa 488 (mouse)-or Alexa 594 (rabbit)-conjugated secondary antibodies. Cells were also stained for nucleus with Hoechst dye (H-33258, Sigma) for 5 min and then mounted using FA mounting fluid (Difco). Images were taken using an Olympus (IX7151F-2, Japan) microscope equipped with confocal scanner unit (CSU-20, Yokogawa Inc., Japan) and EMCCD camera (Andor, UK), operating with the 488 nm (Argon), 351 nm and 364 nm (Argon and UV) laser, and a Carl Zeiss Axiovert microscope attached with Photometrics Synsys monochrome CCD.
Senescence-associated ␤-Gal Assay-Senescent cells were detected using the standard protocol for senescence associated ␤-gal staining assay. In brief, cells were washed with PBS, fixed with 2% formaldehyde/0.2% glutaraldehyde in PBS for 10 min, washed again with PBS, followed by overnight incubation in staining solution (citric acid/phosphate buffer pH 6.0, 5 mM K 3 Fe (CN) 6 , 5 mM K 4 Fe(CN) 6 , 2 mM MgCl 2 , 150 mM NaCl, supplemented with 1 mg/ml of X-gal) at 37°C. Stained cells were observed under the microscope and photodocumented with a NIKON camera.
BrdU Incorporation Assay-TIG-1 cells were transfected with pEGFPN1 (Clontech) and pEGFPN1-CARF plasmids. After 48 h of transfection, 10 ml of BrdU labeling reagent (Amersham Biosciences) was added per ml of culture medium for 4 h. Cells were fixed and immunostained with anti-BrdU antibody following manufacturer's protocol. BrdU was visualized with Alexa 594-conjugated secondary antibody. About 400 GFP-positive cells were counted for BrdU incorporation in both CARF and vector-transfected culture.
Apoptosis Assay-Cells grown on coverslips in 12-well plates were transfected with CARF and control siRNA for 72 h, then subjected to TUNEL assay (Promega), following the manufacturer's instructions provided with the kit. In addition, to verify apoptosis, cells were stained with Hoechst dye (333242) to determine the nuclear condensation and fragmentation.
Live Cell Imaging-Cells were grown in glass bottom dish and transfected with CARF and control siRNA as described above. Between 48 -60-h post-transfection, live cell microscopy was performed with a heated (37°C) objective and stage incubator (Tokai Hit) using a laser spinning disc confocal microscope (Yokogawa electric Corp.) equipped with EMCCD camera (Andor, UK). Multicolor live cell imaging system was performed as previously described in detail (16). Images were recorded at intervals of 3 min and each z-series typically con-tained 5-10 slices of up to 20 m in thickness. Image stacks were converted to maximum intensity projections using Lumina Vision software (Mitani Corp., Fukui, Japan).

Overexpression of CARF Caused Premature Senescence of Human
Fibroblasts-To resolve the function of CARF, it was overexpressed in human fetal lung fibroblasts (MRC5) that harbor a complete set of its downstream effector components (p53, HDM2 and p21 WAF1 ). MRC5 cells were infected with CARF-Myc expressing retroviral vector at PD 30; the expression was confirmed by Western blotting and immunostaining with anti-Myc antibody for more than three subsequent population doublings (Fig. 1, A and B; data not shown). Control and CARF-overexpressing cells were serially passaged. Whereas CARF-overexpressing cells showed senescence at PD 40, the control cells senesced at about 60 PD. We examined senescence specific markers in control and CARF-overexpressing cells and found that consistent with senescence-like morphology and loss of proliferation, the CARF-overexpressing cells showed autofluorescence, senescence-specific ␤-gal staining and enhanced levels of p53 protein and its downstream effector, p21 WAF1 (Fig. 1C). Similar results were obtained when CARF was overexpressed in human fetal skin fibroblasts (TIG-1) using GFP-CARF fusion protein (Fig. 1, A and D). CARF-overexpressing TIG-1 cells showed premature senescence (45 PD in contrast to 67 PD in control cells), also signified by altered cell morphology (enlarged, flattened, multinucleated cells and lack of spiral arrangement), enhanced ␤-gal staining, autofluorescence, and inhibition of DNA synthesis ( Fig. 1D and data not shown) and slower growth rate as compared with the vector-infected control cells (Fig.  1E). Of note, CARF overexpression in cells that lack p53, MDM2 and p21 WAF1 , did not induce senescence suggesting that the effect is predominantly mediated by p53 pathway. This was consistent with our earlier data in which overexpression of CARF induced strong growth arrest in U2OS and HCT116 (wild-type p53 function) but was not effective in Saos-2 that lack functional p53 (10).
CARF Is Up-regulated during Replicative and Stress-induced Senescence-Induction of premature senescence with overexpression of CARF led us to examine its role during replicative senescence of normal cells. We found that CARF expression was distinctly increased at later passages of human normal cells and most interestingly, it correlated with enhanced levels of p21 WAF1 and reduction in HDM2 (Fig. 2A). These data were consistent with the previously demonstrated functions of CARF as (i) collaborator of ARF (16) and p53 (10) and (ii) transcriptional repressor of HDM2 (11). Of note, we found a strong correlation between up-regulation of CARF with increase in p53 and p21 WAF1 (Fig. 2, B and C). Furthermore, it also correlated with presence of endogenous ␤-gal activity in senescent fibroblasts (Fig. 2D). In senescent cultures, more than 75% cells were enriched with CARF and p21 WAF1 expression in concert. Endogenous ␤-gal staining revealed equivalent percentage of positive cells (Fig. 2D). We next investigated if CARF was also involved in stress-induced premature senescence (SIPS) of human cells. As shown in Fig. 3, exposure of TIG-1 (normal human fibroblasts) and U2OS (osteosarcoma) cells to an oxidative stress agent (H 2 O 2 ) caused senescence as detected by ␤-gal staining (Fig. 3A). Significantly, CARF was up-regulated in both the cell types, and paralleled the increase in p53 and p21 WAF1 proteins (Fig. 3, B and C). Of note, premature senescence induced by oncogenic ras also showed clear up-regulation of CARF (Fig. 3D).
CARF-compromised Cells Undergo Apoptosis-Based on the above data, on the role of CARF in cellular senescence, we predicted that CARF-compromised cells might be able to escape senescence and progress to immortalization. We tested this by employing CARF siRNA in normal (TIG-1 and MRC5) cells (Fig. 4A). Contrary to our prediction, we found that CARF siRNA caused cell death resembling apoptosis. Similar results were obtained in cancer (HeLa and U2OS) cells (Fig. 4A), suggesting that CARF may have some essential function in both normal and cancer cells. In CARF-compromised cells, CARF siRNA (which caused 60 -80% reduction in CARF expression) ( Fig. 4B and data not shown) not only reduced the levels of p53 and p21 WAF1 (assessed as transcriptional activation function of p53) (Fig. 4, B and C), but also caused induction of cleavage of caspase-8, caspase-3, caspase-7, and caspase-9 (Fig. 4, B-D). These data firmly suggested that the CARF-compromised cells undergo cell death by caspase-mediated apoptosis. Furthermore caspase inhibitor z-VAD-fmk caused inhibition of cell death caused by CARF knockdown (Fig. 4E). These data confirmed the role of CARF in apoptosis that was mediated by caspase pathways. We also tested if CARF was suppressed during drug-induced apoptosis by using staurosporine as a model reagent.
As shown in Fig. 4F, the expression level of CARF was found to be decreased during staurosporineinduced apoptosis, also marked by an increase in amount of cleaved caspase-8. To investigate the role of CARF in both growth arrest and apoptosis, we next used a doxorubicin model of cell death. Low doses of doxorubicin are known to cause growth arrest of cells that resembles senescence, and high doses cause apoptosis (Fig. 4G). Of note, consistent with the above data on the induction of senescence with overexpression of CARF, its expression was increased with low doses of doxorubicin (0.25-0.50 g/ml) in both U2OS and HeLa cells (Fig. 4G). An increase in the level of CARF expression was paralleled with increase in p53 and p21 WAF1 proteins. High doses of doxorubicin  (1-3 g/ml) that resulted in cell death (Fig. 4G), caused reduction in CARF, p53 level and activity (observed by p21 WAF1 level), and induced cleavage of caspase-8 (Fig.  4G), signifying apoptosis. Similar results were obtained with doxorubicin, etoposide, and camptothecin treatments in U2OS, HeLa, and HCT116 cells (data not shown), suggesting that the involvement of CARF in senescence and apoptosis is not a cell type or drug-specific response. In view of the role of CARF both in growth arrest and apoptosis as described above, we also performed reconstitution of CARF in CARF-compromised cells by Zn-inducible CARF expression system (11,16). CARF expression was induced at 3-h post-transfection of CARF-siRNA. The results revealed that (i) CARF siRNA was specific and highly efficient; whereas CARF was strongly induced by ZnSO 4 in control-siRNA-transfected cells, the presence of CARF-siRNA strongly compromised the reconstitution, (ii) consistent with the overexpression of CARF and p21 WAF1 , the control-siRNA transfected cells showed strong growth arrest, and (iii) CARF-siRNA compromised the CARF and p21 WAF1 levels and resulted in decline in viability (Fig.  4H) that represented combined effect of growth arrest and apoptosis mediated by CARF overexpression and silencing, respectively as also supported by the data in Figs. 2A, 3,  A-D, and 4, A-G). Similar effects of CARF-siRNA and reconstitution of CARF expression were recorded by using HDM2 reporter assay (11). Whereas CARF overexpression caused decline in HDM-2, increase in p53 and growth arrest in U2OS cells, CARF siRNA abrogated these effects (11,16).
To get further insights into the mechanism of CARF-silencing induced cell death, we observed that a high number of CARF-compromised cells were TUNEL positive and had condensed chromatin (Hoechst staining), suggesting that CARF silencing caused apoptosis  1-3) and induction of apoptosis and downregulation of CARF were marked by decrease in p53, p21 WAF1 , and cleavage of caspase-8 (lanes 4 -6). Actin was used as a loading control. H, overexpression of CARF resulted in growth arrest in control siRNA-transfected cells; CARF-siRNA compromised the reconstitution and the cell viability (Trypan Blue exclusion assay) declined from the combined growth arrest and apoptosis effects. and abnormal cell division (Fig. 5A). Consistent with the presence of condensed chromatin, CARF-suppressed cells were enriched with phosphorylated H3 histone (Fig. 5B), an early marker of mitosis (18). The result indicated that the mitotic progression was inhibited in CARF knockdown cells and that may have been arrested in pro-metaphase stage. Cell cycle analysis of control and CARF siRNA transfected also revealed a 2.6-fold increase in cells at G2/M phase and decrease in cells at G1 suggesting CARF is required for mitotic progression (Fig.  5C). We examined ␥-tubulin immunostaining of spindle fibers in CARF-compromised cells and found high proportion of multi-polar, aneuploid, and enlarged cells (Fig. 5, D and E).
Time-Lapse Sequence of G2/M Progression in CARF-compromised (with siRNA) Cells-Based on the above data, we performed live cell imaging (time-lapse microscopy) to observe cell division in control and CARF-siRNA-transfected cells as described under "Materials and Methods." As shown in Fig. 6A, mitotic cycle was visually determined by rounding of the parent cells with condensed chromatin and separation into two daughter cells by cytokinesis. In control cells, mitosis completed in ϳ40 min, while in CARF-siRNA-treated cells, although many cells rounded up as in normal mitosis, they did not complete cell division even over a period of 8 h. Instead, these cells were triggered to undergo apoptosis, as shown by their typical blebbing phenotype (Fig. 6, A and B; supplemental Movies S1A and S1B). Similar results were obtained in U2OS, HeLa, and MDA-MB-435 cells. To further prove this phenomenon, we used the multicolor MDA-MB-435 Auro/imp/H3 cell line stably expressing GFP-Aurora A kinase, DsRed-D importin ␣2 (DsRed-importin), and CFP-histone H3, simultaneously, as markers for the centrosomes/mitotic spindles, nuclear membrane and nucleus/chromosomes, respectively (17), and performed live cell imaging. As shown in Fig. 7A, untransfected and control siRNA-transfected cells clearly showed the different stages of mitosis, including disappearance of nuclear envelope (red), condensation of chromatin (blue), appearance of spindle poles (green), aligning of chromosomes on metaphase plates, separation of two chromosome sets and progression to anaphase, resulting in two daughter nuclei after the disappearance of the centrosome and reformation of the nuclear envelope (Fig. 7A, supplemental Movie S2). In sharp contrast, CARF-compromised cells showed very faint signal for the centrosome, extensive chromatin condensation ultimately forming a single large structure and an accumulation of importin in the nucleus (Fig. 7B and supplemental Movie S3). Taken together, these data demonstrated that CARF is essential for normal progression of cell division; its absence causes inadequate centrosome formation, excessive chromatin condensation leading to aneuploidy and apoptosis, as evidenced by data in Figs. 6 and 7.

DISCUSSION
p53 pathway is known to respond to a wide variety of stress signals, including: telomere shortening, hypoxia, mitotic spindle damage, heat or cold shock, unfolded proteins, improper ribosomal biogenesis, nutritional deprivation, and mutational activation of some oncogenes (19,20). It is a single core module that governs the three potent tumor suppression mechanisms: growth arrest, senescence, and apoptosis by its activation through a multitude of stress sensors, its post-translational modification and cross-talk with other signaling cascades. CARF was first cloned as an ARF-binding protein that activated ARF-dependent and -independent p53 functions (10, 11, and 16). It is negatively regulated by degradation by HDM2, a downstream effector and an antagonist of p53, and protects itself from such degradation by acting as a transcriptional repressor of HDM2 (11). The present study is the first to document CARF as a dual regulator of cellular senescence and apoptosis via the p53-dependent and -independent pathways.
Replicative senescence is characterized by progressive erosion of telomeres, leading to chromosomal instability and permanent growth arrest. Dysfunctional telomeres are recognized as DNA double-stranded breaks, thereby activating the DNA damage response program (21). Stress-induced senescence can occur in both mortal and immortal cells via exposure to DNA damage, chromatin remodeling, oxidative stress, oncogenic stress, and strong mitogenic responses (22)(23)(24). Although endogenous replicative senescence and exogenous stress-induced stem from completely different origins, both processes demonstrate strong similarities with regard to activation of DNA damage response and repair programs and thus inducing cell cycle arrest. Both of these involve up-regulation of p53 function and distinct activation of its downstream effector, cyclindependent kinase inhibitor p21 WAF1 , that causes permanent growth arrest by stopping cell cycle progression at both G1/S and G2/M checkpoints (25,26). We have found that CARF is up-regulated in senescent fibroblasts and showed a strong correlation with ␤-gal activity as well as increase in p53 activity and p21 WAF1 levels, established molecular markers of senescence (Fig. 2). Furthermore, an overexpression of CARF in normal human fibroblasts caused their premature senescence suggesting that CARF is a key regulator of senescence pathway (Fig. 1). Consistent with this, an induction of senescence by oxidative stress, DNA-damaging agents and oncogenic Ras involved CARF in both normal and cancer cells (Figs. 3 and 4, data not shown). In each case it was seen to activate p53 function.
Based on the p53 and p21 WAF1 activation function of CARF, we speculated that similar to p21 WAF1compromised normal fibroblasts (27), CARF-compromised normal fibroblasts may bypass senescence and become immortalized. However, to our surprise we found that CARF silencing (and subsequent decrease in p53 and p21 WAF1 ) resulted in the loss of viability and apoptosis in normal fibroblasts (TIG-1) (Fig. 4). Senescent cells are known to maintain molecular machinery that makes them resistant to apoptotic death (28 -31). The data that CARF is up-regulated in senescent cells and its knockdown caused apoptosis suggested that CARF is involved in maintenance of senescent stage of normal cells that resists apoptosis. Indeed, we found that CARF silencing was accompanied by cleavage activation of caspase-3, -7, -8, and -9 (Fig. 4). Furthermore, doxorubin model of senescence and apoptosis clearly endorsed the involvement of CARF in induction of senescence (mediated by increase in p53 and p21 WAF1 ) and apoptosis (mediated by decrease in p53 and p21 WAF1 and induction of caspase-8 cleavage) at its low and high doses, respectively (Figs. 4 and 5). Real time visual examination of cell division in control and CARF-compromised cells revealed that the latter do not undergo normal cell division, instead gets triggered to apoptosis ( Fig. 6 and supplemental Movies S1A and 1B). Such CARF-silencing triggered apoptosis was observed in a variety of cells that varied in their p53 status (wild-type p53-U2OS and MCF-7, inactive p53-HeLa, p53 null-HCT116, and mutant p53-MDA-MB435), endorsing both p53-dependent and -independent activation of apoptosis. The molecular and visual data revealed two kinds of mutually nonexclusive mechanisms resulting in apoptotic fate of CARF-compromised cells. First, in wild-type p53 (U2OS) cells, multiple centrosomes (the microtubule-organizing center responsible for spindle fiber and microtubule formation during mitosis) (32) were largely observed that led to the generation of aneuploid cells entering apoptosis (Figs. 5 and 6). Second, in cells that lack wild-type p53 function (MDA-MB435), diminished formation of centrosomes, prominent excessive condensation of the chromosomes and abnormal accumulation of nuclear envelope protein importin was observed. These abnormalities resulted into apoptosis of cells both in interphase and pro-metaphase stages. Wild type p53 plays an essential role in control of centrosome duplication; its loss or mutational inactivation has been associated with a high rate of aneuploidy in cancers (25,(33)(34)(35). In light of these reports and our data on activation of p53 with CARF up-regulation (Figs. 2 and 3) (10, 16) and diminished p53 activity in its absence (Fig. 4), increased centrosome duplica- tion in CARF-compromised wild-type p53 cells could be explained by insufficient p53 function. In mutant p53 CARFcompromised cells, we observed apoptosis both in interphase and mitotic phase. Immunostaining and real time examinations in multi-labeled cells revealed diminished formation of centro-somes and excessive condensation of chromosomes as marked by phosphorylation of serine 10 of H3 histone (Figs. 5 and 7; supplemental Movies S2 and S3) that signified entrance to the mitotic phase (36,37). However, these cells did not exit from mitosis and terminated into apoptosis as evidenced by Hoechst, TUNEL assays and real time observations (Figs. 5 and 6). Interestingly, in control cultures also, the cells that showed spontaneous decrease in CARF were seen to undergo apoptosis and accumulation of phosphorylated histone H3 (Fig. 5B). These data strongly suggested that, in addition to its p53-activating function, CARF is required for normal progression and completion of cell division and its absence causes mitotic catastrophe/apoptosis.
The present data showing the involvement of CARF in replicative and stress-induced senescence as well as apoptosis highlight its importance as a crucial/core regulator of p53 activities. Our results suggest that CARF acts upstream of p53 as a key regulator of the bifurcation process; while high level of CARF signal the cell to undergo senescence, its low level trigger them to undergo apoptosis. Such dual control of p53 activity by CARF explains strict regulation of tumor suppression in normal human cells in which apoptosis works as a second checkpoint to immortalization. In addition, CARF is found to be essential for normal progression of cell division; its absence evoked a variety of cell division abnormalities culminating into mitotic catastrophe. Besides this, we found that CARF overexpression induced premature senescence also involved up-regulation of p16 INK4A and pRB proteins (Fig. 8, A and B) that have been shown to play a causal role in the intrinsic cellular senescence program (5,22,38). Further studies are warranted to unravel the CARF pathways and network in fine control of senescence and apoptosis in human cells.