p16INK4a-induced senescence is disabled by melanoma-associated mutations

The p16INK4a-Rb tumour suppressor pathway is required for the initiation and maintenance of cellular senescence, a state of permanent growth arrest that acts as a natural barrier against cancer progression. Senescence can be overcome if the pathway is not fully engaged, and this may occur when p16INK4a is inactivated. p16INK4a is frequently altered in human cancer and germline mutations affecting p16INK4a have been linked to melanoma susceptibility. To characterize the functions of melanoma-associated p16INK4a mutations, in terms of promoting proliferative arrest and initiating senescence, we utilized an inducible expression system in a melanoma cell model. We show that wild-type p16INK4a promotes rapid cell cycle arrest that leads to a senescence programme characterized by the appearance of chromatin foci, activation of acidic β-galactosidase activity, p53 independence and Rb dependence. Accumulation of wild-type p16INK4a also promoted cell enlargement and extensive vacuolization independent of Rb status. In contrast, the highly penetrant p16INK4a variants, R24P and A36P failed to arrest cell proliferation and did not initiate senescence. We also show that overexpression of CDK4, or its homologue CDK6, but not the downstream kinase, CDK2, inhibited the ability of wild-type p16INK4a to promote cell cycle arrest and senescence. Our data provide the first evidence that p16INK4a can initiate a CDK4/6-dependent autonomous senescence programme that is disabled by inherited melanoma-associated mutations.


Introduction
The INK4a/ARF locus, situated on chromosome band 9p21, is one of the most frequently altered sequences in human cancer and germline mutations affecting this locus have been linked to melanoma incidence in approximately 39% of melanomaprone families (Goldstein et al ., 2006b). The lifetime risk of melanoma in p16 INK4a germline mutation carriers ranges from 58% in Europe to 91% in Australia by the age of 80 . This locus encodes two potent, but distinct tumour suppressor proteins; the cyclin-dependent kinase inhibitor, p16 INK4a (Serrano et al ., 1993) and the p53 activator p14ARF (Quelle et al ., 1995). Both proteins are critically important in the regulation of cell cycle progression and senescence (reviewed in Sharpless, 2005;Collado et al ., 2007). p14ARF blocks proliferation by inhibiting the p53 ubiquitin ligase hdm2, to stabilize and activate p53 (Pomerantz et al ., 1998;Stott et al ., 1998;Zhang et al ., 1998) and ARF-null mouse embryonic fibroblasts do not senesce (Kamijo et al ., 1997). p16 INK4a promotes cell cycle arrest by inhibiting the kinase activities of the cyclin D-dependent kinases, CDK4 and CDK6, to maintain the retinoblastoma protein, Rb in its hypophosphorylated, antiproliferative state (Serrano et al ., 1993). The progressive accumulation of p16 INK4a is associated with the onset of replicative senescence in primary human epithelial cells (Alcorta et al ., 1996;Brenner et al ., 1998) and ectopic p16 INK4a expression induces growth arrest that phenotypically resembles cellular senescence in human diploid fibroblasts (Zhu et al ., 1998;McConnell et al ., 1999) and in INK4a/ARF -deficient murine melanocytes (Sviderskaya et al ., 2002). Furthermore, p16 INK4a -deficient human diploid fibroblasts and melanocytes, isolated from melanoma-prone individuals with inactivating mutations affecting both INK4a alleles, undergo delayed senescence (Sviderskaya et al ., 2003;Brookes et al ., 2004;Jones et al ., 2007) and are readily immortalized by the introduction of the telomerase reverse transcriptase (Sviderskaya et al ., 2003).
Cellular senescence can be triggered by multiple mechanisms including induction of the INK4a/ARF locus, telomere attrition, DNA damage, oxidative damage and the aberrant proliferative signals of oncogenes (reviewed in Collado & Serrano, 2006). Once established, senescence permanently limits cellular proliferation and protects against the development of malignant cancer. Accordingly, senescent cells are abundant in premalignant lesions of the skin, the lung and the pancreas whereas they are almost completely absent in malignant tumours (Collado et al ., 2005). Senescent cells have been identified, both in vitro and in vivo , using a series of markers (reviewed in Collado & Serrano, 2006;Campisi & d'Adda di Fagagna, 2007). Increased activity of acidic β -galactosidase, termed senescence-associated βgalactosidase (SA-β -gal) is the most widely accepted marker of senescence cells (Dimri et al ., 1995). More recently, the appearance of DAPI-stained heterochromatic regions, known as senescence-associated heterochromatic foci, which result in the stable repression of some E2F target genes are involved in the irreversible growth arrest associated with senescence . These foci are enriched for histone H3 modified at lysine 9 as well as its binding partner heterochromatin protein-1 γ (HP-1 γ ) . Several other markers of senescence have also been described and validated, including in the WMM1175_p16 INK4a cells (Fig. 1E). These p16 INK4ainduced morphological changes were confirmed using microscopy. As shown in Fig. 2A We observed a dramatic increase in the appearance of nuclear foci at 3 and 5 days post-IPTG treatment; 49 ± 5% and 75 ± 5% of p16 INK4a -induced cells stained positive for nuclear foci at 3 and 5 days post-induction, respectively, with the appearance of large, prominent, often irregular shaped nuclei ( Fig. 2A). The accumulation the HP-1γ, within these nuclear foci confirmed that they are senescence-associated heterochromatic foci (Fig. 2B). Taken together our results confirm that wild-type p16 INK4a can induce senescence in WMM1175 melanoma cells in a p53-and p21 Waf1 -independent manner.

Impact of melanoma-associated p16 INK4a mutations on melanoma cell senescence
Although it has been shown that wild-type p16 INK4a can promote an autonomous senescence programme (Dai & Enders, 2000), there has been no detailed analysis on the impact of melanoma-associated mutations on this programme. This is particularly relevant as there are significant variations in the penetrance of p16 INK4a mutations for melanoma (Berwick et al., 2006) and this may relate to loss of specific functions, including the induction of senescence. We analysed two well-defined and common p16 INK4a mutants that segregate with disease in highrisk melanoma families. These mutant proteins were also selected because they displayed expression levels comparable to the wild-type p16 INK4a protein in the inducible WMM1175 melanoma cell model (Fig. 3A). All clones expressing mutant p16 INK4a proteins were analysed as the wild-type WMM1175_p16 INK4a clone in cell proliferation and senescence assays. The R24P mutant, which retains CDK6 inhibitory activity, partially inhibited Rb phosphorylation and slightly diminished the levels of total Rb over the 5-day induction period (Fig. 3B). In contrast, the A36P mutant had no consistent impact on Rb levels or its phosphorylation status (Fig. 3B). Expression of the p16 INK4a mutations (R24P and A36P) had no long-term inhibitory effect on WMM1175 cell proliferation ( Fig. 3C) and this correlated closely with the consistently weaker inhibition of S-phase induced by the mutants when compared with the sustained S-phase inhibition and G1 arrest observed when the wild-type p16 INK4a protein was induced (Fig. 3D).
Furthermore, these mutant p16 INK4a proteins produced no detectable changes in cell size and morphology (data not shown), had no impact on heterochromatic foci and did not induce SA-β-gal activity (Fig. 3E).

p16 INK4a -induced arrest and senescence requires inhibition of CDK4 and CDK6
Given that the R24P variant, which binds and inhibits CDK6, but not CDK4 , was still incapable of promoting arrest, we hypothesized that inhibition of both CDK4 and CDK6 was required for p16 INK4a  showed all the characteristic markers of senescence (Fig. 5).
To ensure that p16 INK4a -mediated cell cycle arrest was specifically overcome by expression of its CDK4 and CDK6 binding partners, we also transiently introduced CDK2, a kinase that accelerates and augments CDK4/6-initiated Rb hyperphosphorylation (reviewed in Johnson & Walker, 1999;Sherr, 1993). Ectopically expressed CDK2 did not overcome the ability of p16 INK4a to induce cell cycle arrest or senescence (Fig. 5). These data confirm that the inhibition of both CDK4 and CDK6 kinase activity is required for p16 INK4a -mediated cell cycle arrest and senescence. More importantly, they suggest that all known functions of p16 INK4a , including the induction of chromatin condensation and p16 INK4a -mediated changes in cell morphology and size, depend on CDK4/6 binding and inhibition.

The Rb protein is the critical downstream target of p16 INK4a
It is well established that the downstream impact of p16 INK4amediated inhibition of CDK4 and CDK6 activity is the hypophosphorylation and activation of Rb, as shown in Fig. 1A. It has also been recognized that p16 INK4a accumulation promotes the rapid disappearance of Rb (Serrano et al., 1997;Fang et al., 1998;Ausserlechner et al., 2005), and we observed Rb loss in both the WMM1175 (Fig. 1A) (Fig. 6A), did not in itself promote cell cycle arrest or senescence as judged by the continued proliferation of Rb shRNA-transduced WMM1175 (data not shown) and WMM1175_p16 INK4a cells (Fig. 6B). These Rb-null cells did not stain positive for SA-β-gal activity, did not form DNA foci (Fig. 6B) and did not enlarge (Fig. 6C). Thus, down-regulation of Rb expression alone does  (Fig. 7A). Intriguingly, the ability of p16 INK4a to increase cell size and granularity did not require Rb, and p16 INK4a induced these distinctive changes in cell morphology regardless of Rb status (Fig. 7B).

Considering that depletion of Rb occurs after the onset of p16
INK4a -induced arrest but prior to the onset of p16 INK4ainduced senescence (see Fig. 1), it was possible that reinstating the expression of Rb could influence p16 INK4a -induced senescence. As shown in Fig. 7

p16
INK4a is a highly penetrant melanoma tumour suppressor that regulates cell cycle progression by inhibiting the kinase activities of cyclin D-associated CDK4 and CDK6 (Serrano et al., 1993).    (Huot et al., 2002;Jones et al., 2007) and had an extended, but finite lifespan that terminated with senescence (Sviderskaya et al., 2003;Brookes et al., 2004). Interpretation of these results is complicated, however, by the possible contribution of defective p14ARF (Huot et al., 2002;Sviderskaya et al., 2003;Brookes et al., 2004), which is known to confer a growth advantage when silenced (Voorhoeve & Agami, 2003). Moreover, there is considerable variability in the different cell strains with regard to their lifespan (Brookes et al., 2004;Jones et al., 2007), p16 INK4a expression (Beausejour et al., 2003), chromosomal stability (Sviderskaya et al., 2003) and inducibility of p16 INK4a by the RAS oncogene . To avoid some of these confounding effects, silencing molecules have been applied to deliberately and specifically ablate p16 INK4a . Nevertheless, the data remain inconclusive; in most, but not all reports, p16 INK4a deficiency modestly extended the replicative lifespan of cells but did not impair senescence (Bond et al., 1999;Voorhoeve & Agami, 2003;Denoyelle et al., 2006). p16 INK4a was also not essential for H-RAS-induced melanocyte senescence (Denoyelle et al., 2006), although it was required for RAS-induced fibroblast senescence (Bond et al., 1999;Huot et al., 2002). transiently transfected with an Rb or empty expression plasmid, and approximately 6 h post-transfection the cells were treated for three days with IPTG (+) or PBS (-) and stained for Rb, markers of senescence (SA-β-gal, DAPI) and proliferation (Ki67), as indicated. Cell counts for each of these markers are shown as histograms, which correspond to the average ± standard deviation of at least two independent induction experiments from a total of at least 300 cells.
As an alternative strategy, we applied an inducible melanoma cell model to thoroughly characterize the p16 INK4a senescence pathway, with a particular emphasis on the analysis of wellestablished markers of senescence. We then examined the impact of two melanoma-associated mis-sense p16 INK4a mutations on the senescence of this melanoma cell model. By utilizing an inducible cell clone we eliminated cell-related variations and manipulated the induction of p16 INK4a and melanoma-associated p16 INK4a variants, to obtain near-physiological expression levels. As expected, the wild-type p16 INK4a protein promoted rapid cell cycle arrest that was associated at later time points with the onset of senescence and the appearance of classic senescence markers, including enlarged cells with heterochromatic foci and SA-β-gal activity. As expected, these markers proved useful in combination, as none are specific or persistent in all senescence cells (Collado & Serrano, 2006). In fact, U20S cells induced to express wild-type p16 INK4a acquired SA-β-gal activity, showed a large increase in cellular size but did not feature condensed chromatin. Although senescenceassociated heterochromatic foci are associated with the silencing of E2F-1 genes , it is evident that they are late markers of senescence, occur later than E2F-1 target gene silencing (data not shown) and can be absent in highly vacuolized and arrested cells (Denoyelle et al., 2006). The mis-sense p16 INK4a variants, R24P and A36P, failed to inhibit proliferation and to initiate senescence. The R24P mutation was able to slightly reduce the hyperphopshorylation of Rb, presumably because it retains CDK6 inhibitory activity but this was not sufficient to maintain a G1 arrest. This is consistent with data indicating that this is a highly penetrant melanomasusceptibility mutant that has been identified in at least eight melanoma-prone families worldwide (Goldstein et al., 2006b). Furthermore, it reinforces that CDK4, rather than CDK6 (Shennan et al., 2000), is the critical kinase in melanoma. CDK4 germline mutations have been identified in eight melanoma-prone families worldwide (Zuo et al., 1996;Soufir et al., 1998;Molven et al., 2005;Pjanova et al., 2007;Soufir et al., 2007) and these disrupt the interaction between p16 INK4a and CDK4 (Zuo et al., 1996). Mouse embryonic fibroblasts derived from CDK4 R24C/R24C mice (CDK4 R24C is resistant to p16 INK4a inhibition) (Rane et al., 2002) and human diploid fibroblasts overexpressing CDK4 have an extended lifespan (Morris et al., 2002;Ramirez et al., 2003) and carcinogen-treated mice carrying oncogenic CDK4 are highly susceptible to melanoma development (Sotillo et al., 2001

Lentivirus transductions
Lentiviruses were produced in HEK293T cells using the pSIH1-H1-copGFP (Copepod green fluorescent protein) shRNA expression vector (Systems Biosciences, Mountain View, CA, USA) encased in viral capsid encoded by three packaging plasmids as described previously (Dull et al., 1998). Viruses were concentrated as described previously (Reiser, 2000). Viral titres were determined using 1 × 10 5 U2OS cells/well in six-well plates, transduced with serial dilutions of the concentrated viral stocks in the presence of Polybrene (8 μg mL -1 ; Sigma, St. Louis, MO, USA). Cells were harvested 48 h post-transduction, analysed by flow cytometry for GFP expression and viral titre calculated. For Rb silencing experiments, cells were transduced at an MOI of 10 with either a virus encoding Rb shRNA or a control shRNA, with no homology to any human gene. Cells were incubated for 72-96 h prior to analysis to allow expression of shRNA constructs and efficient silence of Rb.

Proliferation assays
Cells were seeded at 1000 cells per well in a 96-well plate, with or without 4 mM IPTG. Number of viable cells was determined daily over 5-day induction period using the MTS assay (Promega, Madison, WI, USA) and analysed with the VICTOR 2 1420 Multilabel Counter (PerkinElmer, Waltham, MA, USA).

Flow cytometry
For cell cycle analysis, cells were fixed in 70% ethanol at 4 °C for at least 1 h, washed in PBS and stained with propidium iodide (50 ng μL -1 ) containing ribonuclease A (50 ng μL -1 ). DNA content from at least 6000 cells was analysed using ModFIT software (Verity Software House, Topsham, ME, USA). The percentage of S-phase inhibition was calculated using the following formula: [(percentage of S-phase cells in uninduced cells) -(percentage of S-phase cells in induced cells)/(percentage of S-phase cells in uninduced cells)] × 100. Cell size and granularity was determined using flow cytometry on unfixed cells or cells fixed in 1% paraformaldehyde/PBS and analysed with CellQuest Pro (BD Biosciences)