PK11007

SIRT1 inhibition restores apoptotic sensitivity in p53-mutated human keratinocytes

Abstract

Mutations to the p53 gene are common in UV-exposed keratinocytes and contribute to apoptotic resistance in skin cancer. P53-dependent activity is modulated, in part, by a complex, self-limiting feedback loop imposed by miR-34a-mediated regulation of the lysine deacetylase, SIRT1. Expression of numerous microRNAs is dysreg- ulated in squamous and basal cell carcinomas; however the contribution of specific microRNAs to the pathogen- esis of skin cancer remains untested. Through use of RNAi, miRNA target site blocking oligonucleotides and small molecule inhibitors, this study explored the influence of p53 mutational status, SIRT1 activity and miR-34a levels on apoptotic sensitivity in primary (NHEK) and p53-mutated (HaCaT) keratinocyte cell lines. SIRT1 and p53 are overexpressed in p53-mutated keratinocytes, whilst miR-34a levels are 90% less in HaCaT cells. HaCaTs have impaired responses to p53/SIRT1/miR-34a axis manipulation which enhanced survival during exposure to the chemotherapeutic agent, camptothecin. Inhibition of SIRT1 activity in this cell line increased p53 acetylation and doubled camptothecin-induced cell death. Our results demonstrate that p53 mutations increase apoptotic resistance in keratinocytes by interfering with miR-34a-mediated regulation of SIRT1 expression. Thus, SIRT1 inhibitors may have a therapeutic potential for overcoming apoptotic resistance during skin cancer treatment.

Introduction

The transcription factor p53 regulates the expression of proteins involved in the DNA damage response, cell cycle control, senescence, differentiation and apoptosis (Vousden and Lu, 2002). As such, p53 maintains the functional integrity of the genome and acts as a tumour suppressor during exposure to carcinogenic stressors. Altered genotoxic stress responses occurring as a consequence of p53 mutations promote apoptotic resistance and unregulated proliferation (Vogelstein and Kinzler, 2004), which are well established hallmarks of cancer (Hanahan and Weinberg, 2000). Consequently, p53 mutations are found in almost every cancer type, including those affecting the skin (Olivier et al., 2010; Rivlin et al., 2011). Keratinocytes are particularly prone to UV-induced DNA damage (Ziegler et al., 1993), and the signa- ture mutations caused by UV exposure affect p53 activity in ~ 50% of basal cell and N 90% of squamous cell carcinomas (Rass and Reichrath, 2008). Targeting p53 gain-of-function mutations, therefore, may be the key to overcoming pre-neoplastic proliferation and resistance to cell death signalling in non-melanoma skin cancer (Rodust et al., 2009). Silent information regulator 2 (Sir2) proteins (sirtuins) are a conserved family of NAD-dependent deacetylases which couple the enzymatic cleavage of NAD to the deacetylation of protein substrates. Sirtuins regulate numerous cellular processes, including cell cycle pro- gression, nutrient metabolism, and cellular ageing (Haigis and Sinclair, 2010). The most extensively investigated sirtuin, SIRT1, has both cytoplasmic and nuclear substrates, the latter including p53 and the DNA repair proteins APE/Ref1 and PARP1 (Luo et al., 2001; Rajamohan et al., 2009; Yamamori et al., 2010). SIRT1 has characteristics in common with both tumour suppressors and promoters (Fang and Nicholl, 2011; Yi and Luo, 2010); however the regulatory control of SIRT1 in carcino- genesis is poorly defined.

p53-mediated stress responses are facilitated by microRNAs, which form a complex regulatory network controlling cell survival, senes- cence, growth and death (Shi et al., 2010). MicroRNAs (miRNAs) are small non-coding RNAs which regulate gene translation by complemen- tary base pairing with target mRNA sequences. Primary nuclear transcripts (pri-miRNAs) are processed by the nuclear RNase III enzyme Drosha into 70-nucleotide precursor sequences (pre-miRNAs) for cytoplasmic translocation and cleavage into ~ 22-nucleotide fragments (mature miRNAs) by the cytoplasmic RNase II enzyme, Dicer-1, prior to incorporation into RNA-induced silencing complexes (Iorio and Croce, 2012). The miR-34 family of miRNAs has recently emerged as important components of p53-mediated responses (Hermeking, 2010) by targeting substrates involved in cell cycle regulation and apoptosis (He et al., 2007). In particular miR-34a, whose expression is induced by p53-mediated transcriptional activation (Hermeking, 2012), regu- lates SIRT1 expression by base pairing with its 3′UTR (Yamakuchi et al., 2008). SIRT1 represses functional activity of p53 in the basal cell by maintaining its lysine residues in a deacetylated state, and thus sup- pression of SIRT1 expression by miR-34a reinforces p53 activation (Luo et al., 2001; Vaziri et al., 2001). In this manner, SIRT1, p53 and miR-34a form a coherent feed-forward loop, referred to here as the p53/SIRT1/ miR-34a axis, which influences cellular differentiation, senescence and apoptotic signalling (Aranha et al., 2011; Ito et al., 2010; Yamakuchi and Lowenstein, 2009; Zhao et al., 2010). Despite the positive correla- tion between gain of function p53 mutations and chemotherapeutic resistance in squamous cell carcinoma (Cabelguenne et al., 2000), manipulation of the p53/SIRT1/miR-34a axis in order to enhance the ef- ficacy of chemotherapy during skin cancer treatment has not previously been studied in keratinocyte cells.

In this paper, we demonstrate that regulation of SIRT1 expression in human keratinocytes is, in part, related to p53 activity and miR-34a ex- pression; and inhibition of SIRT1 activity sensitises p53-mutated keratinocytes to undergo apoptosis independently of miR-34a.

Materials and methods

Cell culture. Neonatal keratinocytes (NHEK) were cultured in Epilife medium supplemented with human keratinocyte growth supplement and 0.1 mg/ml kanamycin sulphate [Gibco-Invitrogen (Carlsbad, CA)]. Human transformed keratinocyte (HaCaT) cells (a gift from Dr Gary Halliday, University of Sydney) were adapted from DMEM + 10% FBS into Epilife medium over a period of 4 weeks to standardize growth conditions between the cell lines. Cells were incubated at 37 °C and 5% CO2 and passaged at 60–70% confluence (~every 3–4 days). Fifty thousand cells were seeded into each well of a 6-well plate, and allowed to recover overnight before transfection/treatment.

Reagents. EX-527 [6-chloro-2,3,4,9-tetrahydro-1H-carbazole-1- carboxamide] [Sigma-Aldrich (St Louis, MO)], (±)-Nutlin-3 [Cayman Chemical (Ann Arbor, MI)] and camptothecin (CPT) [Sigma-Aldrich (St Louis, MO)] were dissolved in dimethyl sulfoxide (DMSO) at a con- centration of 100 mM, and stored in aliquots away from light at −20 °C.

Further dilutions were made in culture media immediately prior to cell exposure. The final fraction of DMSO in culture media remained b 0.01% in all experiments.

RNAi. NHEK or HaCaT keratinocytes were transiently transfected in 6 well plates using RNAiMAX transfection reagent [Invitrogen (Carlsbad, CA)] with 10 nM ON-TARGET plus siRNA pools [Dharmacon (Lafayette, CO)] according to the manufacturer’s instructions. After overnight incubation, the transfection complexes were replaced with fresh keratinocyte growth medium containing experimental compounds and incubated for the required time period.

Immunoblotting. Cells were lysed in RIPA lysis buffer (50 mM Tris– HCl [pH 8.0], 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 10% glycerol) or chilled PARP sample buffer (62.5 mM Tris–HCl [pH 6.8], 6 M urea, 10 mM β-mercaptoethanol, 2% SDS, 0.001% Bromophenol Blue, 10% glycerol) to which protease inhibitors [Cell Sig- nalling Technology, Beverly, MA] and 1 mM phosphatase inhibitors had been added immediately before use. Immunoblots were prepared from SDS-PAGE separated cell lysates on PVDF membranes using antibodies raised against p53, SIRT1, PARP1 and p53 acetylated at lysine 382 [p53(AcK382)] [Cell Signalling Technology, Beverly, MA]. Densitometric analysis of the chemiluminescent signal was performed using Fujifilm Las-3000 Luminescent Image Analyzer Version 2.2 and Multi-Gauge Version 3.0 software [FujiFilm Co. Ltd, Tokyo, JPN] using Actin as a reference. PARP cleavage was determined from the cleaved fraction in relation to total PARP expression (full length + cleaved PARP).

One microgramme of total RNA extracted with a miRCURY RNA isolation kit [Exiqon (Vedbaek, Denmark)] was reverse transcribed to cDNA using a DyNAmo cDNA Synthesis Kit [Thermo Fisher Scientific Inc. (Lafayette, CO)], and five microlitres of a 1:10 dilution of template cDNA was amplified on an iCycler MYiQ™ [Bio-Rad Laboratories, CA, USA] using Bio-Rad IQ™ Multiplex Powermix (Bio-Rad Laboratories, Hercules, CA). Primer oligonucleotides [Geneworks (Adelaide, SA)] were used at a concentration of 100 nM, and all assays were carried out using the following cycling conditions: 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s & 60 °C for 60 s. Primer sequences were as follows: SIRT1 (forward — gca gat tag tag gcg gct tg, reverse — tct ggc atg tcc cac tat ca); p53 (forward — cca tgt gct caa gac tgg cgc t, re- verse — agt ctg gct gcc aat cca ggg aa); Pri-miR-34a (forward — cca gct gtg agt gtt tct ttg gca g, reverse — ccc aca acg tgc agc act tct ag); 5S rRNA (forward — gta acc cgt tga acc cca tt, reverse — cca tcc aat cgg tag tag cg); and U6 snoRNA (forward — cgc ttc ggc agc aca tat act aa, reverse — tat gga acg ctt cac gaa ttt gc). For BIRC5/Survivin and BCL2; 1:10 dilutions of cDNA extracts were amplified using Taqman Gene ex- pression assays [Applied Biosystems (Foster City, CA)] using RN18S1 as a reference gene.

MicroRNA expression. MicroRNAs from RNA extracts were polyadenylated and converted to cDNA using a miRCURY LNA Universal cDNA synthesis kit [Exiqon (Vedbaek, Denmark)]. Relative miR-34a was analysed by qRT-PCR using LNA PCR primer sets [Exiqon (Vedbaek, Denmark)], using U6 snoRNA and 5S rRNA as reference genes.

Taqman pri-miRNA assays [Applied Biosystems (Foster City, CA)]; and Taqman MicroRNA assays [Applied Biosystems (Foster City, CA)] for hsa-mir-145 and hsa-mir-16-1/miR-15a were performed according to the manufacturer’s instructions. RN18S1 and U6 snoRNA were used as reference genes for pri-miRNA and microRNA assays, respectively.

miR-34a target site analysis. LNA oligonucleotides were designed to block miR-34a-mediated regulation of SIRT1 by targeting the SIRT1 mRNA binding site [Exiqon (Vedbaek, Denmark) Oligo sequence (5′–3′): ggcagtaatggtcctagct]. miR-34a target site blocking and mis- matched control oligonucleotides were transfected into NHEK and HaCaT keratinocytes in 6 well plates using RNAiMAX transfection reagent [Invitrogen (Carlsbad, CA)]. After 48 h, the culture plates were treated with experimental compounds and incubated for the required time period.

Microplate assays. Cells were seeded in opaque black 96 well micro- plates at a density of 5 × 104 cells per well and allowed to recover over- night. Seeding media was replaced with treatments and the plates incubated for 24 h. Cell viability and apoptotic activity were measured using ATPLite (Perkin Elmer, Boston, MA) and Apo-ONE Homogeneous Caspase 3/7 assays (Promega, Madison, WI) by following the manufacturer’s instructions. Chemiluminescence and fluorescence in- tensity (excitation 485 nm/emission 530 nm) were recorded using an Infinite M200-Pro plate reader [Tecan, GmbH, Grödig, Austria].

Statistical analysis. Statistical analysis was performed using GraphPad Prism software [version 5; San Diego, CA]. Results were analysed using 2-way ANOVA with Bonferroni multiple comparison post-tests. Individ- ual treatments were compared to their untreated control sample by 2-tailed pairwise t test. Data are presented as the mean and standard deviation (SD) with n ≥ 3 for all experiments.

Results

SIRT1/miR-34a feedback regulation in human keratinocytes

The impact of p53 mutations on SIRT1/miR-34a feedback regulation in keratinocytes was determined by comparing the expression of total p53, p53 acetylated at K382, SIRT1, and miR-34a in NHEK and HaCaT cell lines. These experiments revealed that expression of p53 protein is 3.5-fold higher in HaCaT keratinocytes however its relative acetylation is 70% that measured in NHEKs (Figs. 1a & b; p b 0.01). Additionally, the level of miR-34a in unstimulated HaCaTs is suppressed, contributing to a 2.5-fold higher expression of SIRT1 protein when compared to NHEKs (Figs. 1a & b; p b 0.001).

We next used EX-527 to investigate the effect of SIRT1 inhibition on p53, acetylated p53, SIRT and miR-34a levels in the HaCaT cell line. An increase in relative p53 acetylation was achieved when SIRT1 was inhibited (Figs. 1c & d; p b 0.01) in a manner consistent with EX-527- treated NHEKs (Herbert et al., 2014), however miR-34a expression halved and SIRT1 protein doubled in SIRT1-inhibited HaCaTs (Figs. 1c & d; p b 0.001). Although p53 mutations in HaCaT cells have not altered the capacity for SIRT1 to influence p53 acetylation, these results indicate that regulatory control over miR-34a and SIRT1 expression is impaired in this cell line.

Apoptotic responses are impaired in p53 mutated keratinocytes

In order to determine the impact of the p53/miR-34a/SIRT1 axis on programmed cell death in keratinocytes, DNA damage was induced by treating NHEK and HaCaT cell lines with the topoisomerase I inhibitor, camptothecin (CPT). Camptothecin-induced apoptosis is considered to be predominantly p53-dependent and has been extensively utilised in chemotherapy (Garcia-Carbonero and Supko, 2002). As such, CPT- induced apoptosis provides a model for studying the impact of p53 mu- tations on cell death and DNA damage signalling. PARP activity is p53 dependent and linked to DNA damage-induced cell death (Beneke et al., 2000; Luo and Kraus, 2012). During apoptosis, PARP is inactivated
by caspase-mediated cleavage into ~25 kDa and ~85 kDa fragments and is considered a characteristic marker for p53-dependent apoptotic signalling (Fischer et al., 2003; Kaufmann et al., 1993).

Initially, we used p53 induction and PARP cleavage to determine a time course for apoptotic induction in wild-type and p53-mutated keratinocytes by exposing NHEK and HaCaT cells to 5 μM CPT for 6, 12, 18 and 24 h periods. As anticipated, primary keratinocytes were significantly more sensitive than p53-mutated keratinocytes to CPT- induced apoptosis. CPT-mediated PARP cleavage in NHEKs was asso- ciated with DNA damage-induced induction of p53, which doubled cumulatively with each 6 h period (Figs. 2a, b, c) and preceded PARP cleavage by approximately 6 h. Conversely, p53 did not accu- mulate in CPT-exposed HaCaT keratinocytes (Figs. 2b, c) and PARP cleavage was 3-fold lower than cleavage measured in NHEKs after 24 h (Figs. 2a, c). These observations indicated that the mutations to p53 are impeding programmed cell death in HaCaT cells.

Survivin and BCL-2 expression is dysregulated in p53 mutated keratinocytes

We next considered how p53 mutations may contribute to apoptotic resistance and so quantified the expression of the known p53-regulated anti-apoptotic genes Survivin/BIRC5 and BCL2. In apoptotic NHEKs, both Survivin and BCL2 were suppressed to 20% and 10% when compared to untreated cells (Fig. 3a, p b 0.001). In contrast, Survivin and BCL2 were significantly up-regulated in CPT-treated HaCaTs [2-fold in the case of Survivin (p b 0.05) and 4-fold for BCL2 (p b 0.001; Fig. 3a), respectively]. Survivin and BCL2 are both down-regulated by miR-34a (Bommer et al., 2007; Shen et al., 2012), therefore we tested if resistance to p53-dependent apoptosis in HaCaT cells was related to their altered miR-34a expression and biosynthesis. Although pri-miR-34a was induced by CPT in both NHEK and HaCaT cell lines (Fig. 3b), only NHEKs showed an in- crease in the mature miR-34a sequence (Fig. 3b, p b 0.05). Because BCL2 is regulated by a group of microRNAs in addition to miR-34a – notably the miR-15a/miR-16-1 cluster (Cimmino et al., 2005) – we asked whether p53 mutations in HaCaTs affect expression of tumour suppres- sor microRNAs other than miR-34a. The miR-15a/miR-16-1 cluster is located within a 0.5 kilobase region of 13q14.3, and the microRNA products, miR-15a and miR-16-1, are processed from a single primary transcript during p53-mediated signalling (Fabbri et al., 2011). Here, pri-miR-16-1 provides a measure of primary transcript levels, and miR-15a has been assayed to represent expression of the mature microRNA. CPT significantly up-regulated the primary transcript (pri- miR-16-1) in both cell lines (Fig. 3c, p b 0.05). Although the mature sequence (miR-15a) was up-regulated nearly 10-fold in NHEKs (Fig. 3c, p b 0.001), miR-15a expression was suppressed in CPT- treated HaCaTs (Fig. 3c, p b 0.001). Additionally, expression of the c-Myc targeting microRNA, miR-145, was up-regulated 3.5-fold by CPT in NHEKs (Fig. 3d, p b 0.001), and was halved in apoptotic HaCaTs (Fig. 3d, p b 0.05). Results from these experiments confirm that apoptotic resistance in p53-mutated keratinocytes is associated with impaired microRNA-mediated regulation of genes involved in the apoptotic signalling programme.

Manipulation of SIRT1 and p53 expression alters apoptotic sensitivity in normal, but not p53-mutated keratinocytes

To further explore the relationship between p53 levels with respect to apoptotic sensitivity in these cell lines, Nutlin-3a was used to block MDM2-mediated proteasomal targeting, and hence degradation of the p53 protein. p53 accumulated 4-fold in NHEKs treated with 10 μM Nutlin-3a for 24 h (Figs. 4a & g; p b 0.001; n = 3), but no change in p53 content was achieved by Nutlin-3a treatment in the HaCaT cell line (Figs. 4b & g). Notably, the 3.5-fold increase in p53 expression with Nutlin-3a treatment in NHEKs brings cellular p53 content to a level which correlates more closely with the cellular content of p53 protein measured in untreated HaCaT cells (Fig. 1b). This Nutlin-3a- induced increase in p53 protein was not associated with a change in acetylation of lysine 382 relative to total p53 protein content in NHEKs (Figs. 4c & g) or HaCaT cells (Figs. 4d & g). Next, apoptosis was induced in NHEK and HaCaT cells following 30 min pretreatment with Nutlin-3a using CPT. Although p53 protein accumulated 5-fold in CPT-treated NHEKs (Figs. 4a & g, p b 0.001), no significant change in p53 protein or acetylation was measured in CPT-treated HaCaTs (Figs. 4b, d & g). Moreover, Nutlin-3a treatment did not significantly alter CPT-induced PARP cleavage for either cell line (Figs. 4e–g). So whilst CPT treatment triggers p53-dependent apoptosis for keratinocytes bearing either wild-type or p53 mutations, improving sensitivity of p53- mediated functional responses to DNA damage requires more than a stabilisation of p53 protein. Furthermore, mutated p53 protein in HaCaT keratinocytes is resistant to MDM2-mediated proteasomal targeting, and hence Nutlin-3a is ineffective for recovering p53 functional responses in this cell line.

We next considered whether manipulating SIRT1 expression levels could alter apoptotic sensitivity. Initially, RNAi was used to determine the relationship between SIRT1 overexpression and im- paired apoptotic signalling in the HaCaT cell line. p53 acetylation in- creased 1.5-fold (p b 0.05) when SIRT1 was ablated in HaCaTs prior to CPT treatment (Figs. 5a, b), but no change in PARP cleavage was measured (Figs. 5a, c). We then designed oligonucleotides to specif- ically block the miR-34a targeting site within the SIRT1 3′UTR to in- vestigate the relationship between miR-34a-mediated regulation of SIRT1 expression and apoptotic resistance in keratinocytes. Cellular SIRT1 protein accumulated following a miR-34a target site block for both NHEK and HaCaT cell lines (Figs. 6a, b, g; p b 0.05), confirming that miR-34a post-transcriptionally regulates SIRT1 ex- pression in wild-type and p53-mutated keratinocytes. SIRT1 was suppressed in CPT-treated NHEKs and HaCaTs, however SIRT1 did not accumulate when the miR-34a targeting site was blocked prior to CPT treatment (Figs. 6a, b, g). Although relative p53 acetylation was suppressed after 24 h of CPT treatment in NHEKs (Figs. 6c & g; p b 0.01), no change in acetylation was measured in CPT-treated HaCaTs (Figs. 6d & g). Additionally, the target site block did not produce a significant change in acetylation in CPT-induced NHEKs, however PARP cleavage decreased (Figs. 6e & g; p b 0.001). No change in PARP cleavage was measured in the HaCaT cell line after a target site block (Figs. 6f & g).

SIRT1 inhibition overcomes apoptotic resistance in p53 mutated keratinocytes

Because neither depletion of SIRT1 by RNAi nor the miR-34a/SIRT1 mRNA target site block was sufficient to influence p53-mediated apopto- tic resistance in the HaCaT cell line, we lastly asked if inhibition of SIRT1 activity with EX-527 would increase apoptotic sensitivity. p53 acetylation was increased by 24 h EX-527 treatment in both cell lines (Figs. 7a, b, g; p b 0.01). After 24 h of CPT treatment, SIRT1 expression was suppressed in both cell lines (Figs. 7c, d, g). When SIRT1 was inhibited, CPT-induced PARP cleavage decreased in NHEK cells (Figs. 7e, g; p b 0.01), and dou- bled in HaCaT cells (Figs. 7f, g; p b 0.001). Consistent with these results, we found that enhanced PARP cleavage was associated with a slight but statistically significant increase in relative p53 acetylation for CPT- treated HaCaTs when SIRT1 was inhibited (Figs. 7b, f & g; p b 0.01). Note that in the analysis, values for p53 acetylation are calculated rela- tive to a measurement of total p53 in cells with each treatment. Thus, whilst p53 protein accumulated in CPT-treated NHEKs, no change was detected for HaCaTs during CPT treatment (Fig. 7g) and EX-527 has increased the fraction of acetylated p53 in CPT treated HaCaT cells with- out increasing total protein. Conversely, EX-527 also increases p53 acet- ylation in NHEKs, however because total p53 protein is up-regulated during CPT treatment, the acetylated fraction (as calculated) remains unchanged.

To expand on evidence from our PARP cleavage data, we assessed CPT- and EX-527-treated NHEK and HaCaT cells using markers of apo- ptosis (caspase 3/7 activity) and cell viability (ATP content). Although cell death was not altered by SIRT1 inhibition alone, caspase 3/7 activity doubled in NHEKs after CPT treatment (Figs. 8a, c; p b 0.001). For SIRT1- inhibited, CPT-treated HaCaT cells, these assays support results obtained from analysis of PARP cleavage, demonstrating that caspase 3/7 activity had increased and cell viability had decreased after 24 h of treatment (Figs. 8b, d; p b 0.001). Collectively, these data indicate that inhibition of SIRT1 modulates cell death signalling in a p53-dependent manner, and preferentially promotes apoptosis in keratinocytes with p53 mutations.

Discussion

HaCaT keratinocytes bear compound heterozygous mutations in each p53 allele which alters the activity and expression of p53 protein without complete loss of function (Lehman et al., 1993; Martynova et al., 2012). By manipulating the activity of p53 in the HaCaT cell line, we were able to uncover conditions by which this phenotype can be targeted to promote cell death in a p53-dependent manner. Apoptotic sensitivity was enhanced in HaCaT keratinocytes by increasing relative p53 acetylation through inhibition of SIRT1 activity (EX-527); but not by p53 or SIRT1 ablation (RNAi), blocking miR-34a-mediated regulation of SIRT1 (miR-34a TSB), or by inhibiting MDM2-mediated proteolytic degradation of p53 protein (Nutlin-3a).

This study provides a mechanistic link between characteristic UV- induced p53 mutations, miR-34a insufficiency, and impaired regulation of SIRT1 expression in the skin. Increased SIRT1 expression is common in non-melanoma skin cancer (Hida et al., 2007). Consistent with this observation, we have found that SIRT1 protein accumulates in HaCaT keratinocytes as a consequence of failed miR-34a-mediated repression of SIRT1 at the translational level, culminating in reduced p53 acetyla- tion and resistance to p53-dependent apoptosis. Thus inhibition of SIRT1 in this cell line restores apoptotic sensitivity by partially overcom- ing the effects of miR-34a insufficiency. Together, these data indicate that p53 mutations contribute to apoptotic resistance in HaCaT keratinocytes by altering the regulatory capacity of the p53/SIRT1/ miR-34a axis during apoptotic signalling. Whether this is a mechanism common to all keratinocytes with p53 mutations is yet to be deter- mined, but the frequent occurrence of UV-induced p53 mutations and increased SIRT1 expression in skin cancers suggests that it is likely to be widespread. Furthermore, low constitutive expression of miR-34a in p53-mutated keratinocytes supports the potential value of miR-34a as a prognostic marker for skin cancer diagnosis and therapeutic monitoring (Jansson and Lund, 2012).

Results from this study show that microRNA-mediated regulation of programmed cell death is altered in the HaCaT cell line. Induction of miR-34a, miR-16-1/15a and miR-145 during p53-dependent apoptotic signalling is suppressed in p53-mutated keratinocytes, and this has consequences for regulatory control of mRNA targets. Specifically, down-regulation of Survivin and BCL-2 during apoptosis is dependent on normal p53 activity. Aberrant microRNA expression is widespread in primary tumours, and is attributed to impaired post-transcriptional regulation of the primary sequence (Lujambio and Lowe, 2012; Thomson et al., 2006). We have previously found that p53 regulates bio- genesis and maturation of microRNAs which have key downstream functions in p53-mediated signalling, and p53 mutations in HaCaTs in- terfere with this process (Herbert et al., 2014). P53 modulates microRNA biogenesis by associating with the RNA helicase p68, which facilitates processing of the primary transcript by Drosha (Suzuki et al., 2009). In cells with mutant p53, formation of the Drosha complex is im- peded, blocking processing of the primary microRNA transcript (Chang et al., 2013). Additionally, mutant p53 directly binds to and inactivates p63, inhibiting Dicer-1 expression and microRNA maturation (Su et al., 2010). Basal and squamous cell carcinomas develop significant changes to the keratinocyte microRNA expression profile (Sand et al., 2012a, 2012b, 2013). Our results indicate that impairments to microRNA expression and target regulation in keratinocytes are driven by the dysregulating effect of p53 mutations on microRNA biogenesis and processing, and occur at the premalignant stage of carcinogenesis. Exploitation of the mutational status of p53 in cancer has been an ongoing chemotherapeutic strategy for many years. Loss of a single functional p53 allele reduces cellular tumour suppressor capacity (Lynch and Milner, 2006), and can generate a phenotype with unique stress responses which promote carcinogenesis (Olive et al., 2004). Nu- merous options, including gene-based therapy, manipulation of p53 ex- pression by disruption of the p53-MDM2 axis, and reactivation of mutant p53 have been proposed as a means for restoring apoptotic sen- sitivity in chemotherapy (Brown et al., 2009; Desilet et al., 2010). Nutlin-3a is most effective when used in combination with inhibitors of S and M phase entry for selective targeting of cells with insufficient p53. However, as our results have demonstrated (Figs. 1a & b), p53 is overexpressed and underacetylated in the HaCaT cell line, indicating that p53 activity is suppressed. Moreover, although induction of MDM2 in the HaCaT cell line is sufficient to decrease p53 expression (Kwon et al., 2004), our results show that HaCaT keratinocytes are resis- tant to MDM2 inhibition, and consequently Nutlin-3a does not promote cytotoxicity in these cells. Early results from clinical trials indicate that a combinatorial approach is required for chemotherapy to effectively overcome p53 inactivating mutations (Hoe et al., 2014; Wang and Sun, 2010). Aside from loss-of function effects, p53 mutations increase protein stability and cause gain-of-function effects which increase resis- tance to p53-based therapeutic agents (Oren and Rotter, 2010; van Oijen and Slootweg, 2000), and hence reactivation of mutant p53 func- tion has emerged as a strategy for improving the efficacy of traditional chemotherapy (Wiman, 2007). In cells with heterozygous p53 muta- tions such as the HaCaT cell line, mutant p53 can have a dominant- negative effect on any non-mutated functional p53 within the cell by forming functionally inactive aggregates which inhibit DNA binding (Goh et al., 2011; Rangel et al., 2014; Willis et al., 2004). Additionally, p53 mutations can produce a gain-of-function phenotype which results from the co-association of mutant p53 with other p53 regulatory pro- teins such as p63, p73 and the acetyltransferase p300 (Xu et al., 2011). This process is particularly problematic for cells exposed to genotoxic stress, during which p53 protein accumulates and tetramerisation in- creases exponentially (Gaglia et al., 2013). Current strategies developed to overcome dominant-negative and gain-of-function effects of p53 in- clude stimulation of p53 degradation pathways, siRNA-mediated abla- tion of mutant p53 protein, blocking co-association with effector proteins, and reactivation of mutant p53 by restoration of wild-type function (Hong et al., 2014; Li and Prives, 2007; Rangel et al., 2014).

Importantly, the efficacy of these approaches is highly variable because the dominant-negative effect of mutant p53 depends on its propensity to inactivate circulating wild-type p53, and this is directly related to the site of mutation (Chan et al., 2004).Histone deacetylase (HDAC) inhibitors have recently been suggested as adjuvants for p53-based chemotherapy (Dickinson et al., 2010). Of the numerous HDAC-targeting therapies, only HDAC Class III (SIRT1) in- hibitors, such as Tenovin-1 are exclusively 53-dependent (Sonnemann et al., 2014). EX-527 is a SIRT1 inhibitor which induces conformational remodelling in the SIRT1 catalytic domain, and which inhibits enzyme activity by blocking substrate release (Zhao et al., 2013). As a conse- quence of its p53 mutations, HaCaT cells express a mixture of p53 sub- types with functional anomalies affecting protein stability and transcriptional activation. However, some p53-dependent activity is still retained in this cell line, albeit at highly suppressed levels. Based on its inhibitory mechanism, we postulate that EX-527 may trap mutant p53 in HaCaT keratinocytes by preventing its release from the catalytic domain of SIRT1. Removing a proportion of mutant p53 from circulation may relieve its dominant negative effect, and enable p53 to become transcriptionally active. This mechanism may also explain the discrep- ancy between the ability of EX-527 to increase apoptotic sensitivity in HaCaT cells, compared to the inability of SIRT1 knockdown to improve apoptotic sensitivity for HaCaTs.

p53 mutations in keratinocytes attenuate p53-dependent stress responses by blunting functional control of the p53/SIRT1/miR-34a axis. Defective miR-34a processing has widespread implications for cellular processes regulated by SIRT1 activity, there is some evidence that inhibiting SIRT1 sensitises pre-cancerous cells to apoptosis (Kim et al., 2013). Although loss of miR-34a alone does not appear to be oncogenic (Concepcion et al., 2012), impaired miR-34a-mediated regulation of p53 responses appears to be a hurdle which must be overcome to pro- mote the efficacy of p53-based chemotherapy. Our results have wider chemotherapeutic potential as a treatment strategy in skin cancer. Up- regulation of cellular SIRT1 expression and inhibition of miR-34a processing in keratinocytes with p53 mutations causes apoptotic resis- tance and attenuation of p53-dependent stress responses when com- pared to wild-type cells. These changes may be overcome, in part, by synthetic SIRT1 inhibitors. By manipulating the activity of p53 with EX-527 in the HaCaT cell line, we were able to uncover conditions by which this phenotype can be targeted to promote cell death in a p53- dependent manner. Results from this investigation demonstrate the need for further investigation to explore the value of SIRT1 inhibition in the design of targeted chemotherapeutic regimens in tumours with p53-mutated PK11007 genotypes.