p16 INK4a and p14 ARF Tumor Suppressor Genes Are Commonly Inactivated in Cutaneous Squamous Cell Carcinoma

Transcription

1 p16 INK4a and p14 ARF Tumor Suppressor Genes Are Commonly Inativated in Cutaneous Squamous Cell Carinoma Vitoria L. Brown, Catherine A. Harwood, Tim Crook,w James G. Cronin, David P. Kelsell, and Charlotte M. Proby Centre for Cutaneous Researh, St Bartholomew s and the Royal London Shool of Mediine and Dentistry, University of London, London, UK; wludwig Institute for Caner Researh, Imperial College Faulty of Mediine, London, UK The p16 INK4a and p14 ARF tumor suppressor genes (TSGs) are enoded within the CDKN2A lous on hromosome 9p21 and funtion as ell yle regulatory proteins in the p53 and RB pathways. Inativation of these genes by geneti and epigeneti hanges has been desribed in some human aners, but their importane in utaneous squamous ell arinoma (SCC) has not been established. Our detailed examination of 40 utaneous SCC revealed loss of heterozygosity of 9p21 markers in 32.5% of ases. Mutational analysis onfirmed five point mutations in four of 40 SCCs. These mutations hanged the amino aid sequene of p16 INK4a in four tumors and p14 ARF in three tumors. Promoter methylation of p16 INK4a and p14 ARF was deteted in 13 of 36 (36%) and 16 of 38 (42%) ases, respetively. Absent protein expression was onfirmed by immunohistohemistry in 13 of 16 (82%) of the tumors with bialleli inativating events. Overall, the frequeny of 9p21 alterations was 76% and for both p16 INK4a and p14 ARF, promoter methylation is the ommonest mehanism of gene inativation. Alterations at this lous were signifiantly more ommon in tumors from immunoompetent ompared with immunosuppressed individuals. These data onfirm the importane of inativation of p16 INK4a and p14 ARF TSGs in the pathogenesis of utaneous SCCs. Key words: p16 INK4a /p14 ARF /tumor supressor genes/cdkn2a/utaneous squamous ell arinoma/gene hyper methylation J Invest Dermatol 122: , 2004 Cutaneous squamous ell arinoma (SCC) is the seond most ommon aner in Cauasians (Johnson et al, 1992). The inidene of the disease is rising rapidly in Europe, North Ameria, and Australia (Holme et al, 2000; Alam and Ratner, 2001; Ramsay et al, 2002). Almost all ases of this arinoma arise on sun-exposed areas of skin, impliating hroni exposure to ultraviolet radiation as the major etiologial fator (Grossman and Leffell, 1997). Organ transplant reipients have a 100-fold inreased risk of developing SCC, implying that immunosuppression may signifiantly alter individual suseptibility to the disease (Lindelof et al, 2000). In these patients utaneous SCC has a more aggressive linial ourse, a greater tendeny to metastasize and a higher mortality rate when ompared with the general population (Martinez et al, 2003). Cutaneous SCC an also arise from within hroni wounds, sars, burns, or ulers (Alam and Ratner, 2001). The moleular events assoiated with the initiation and progression of utaneous SCC remain poorly understood. The ylin-dependent kinase inhibitor 2A (CDKN2A) lous on human hromosome 9p21 is the seond most ommonly altered gene lous in human aner after p53 (Liggett and Sidransky, 1998). Abbreviations: Bp, base pairs; CDKN2A, ylin-dependent kinase inhibitor 2A; HPV, human papillomavirus; LOH, loss of heterozygosity; MSP, methylation speifi PCR; SCC, squamous ell arinoma; TSG, tumour suppressor gene CDKN2A uniquely enodes for two andidate tumor suppressor genes (TSG): p16 INK4a and p14 ARF. They share ommon exons 2 and 3 but have alternatively splied first exons (exon 1a for p16 INK4a ; exon 1b for p14 ARF ). These first exons are under the ontrol of distint promoters and uniquely reate two proteins that have no sequene homology at the amino aid level (Stone et al, 1995). Both genes funtion as ruial ell yle regulators. p16 INK4a binds to CDK4 and CDK6, whih prevent prb phosphorylation and G1 S phase progression (Serrano et al, 1993). By binding to MDM2, p14 ARF prevents both MDM2- mediated p53 degradation and MDM2-mediated Rb inativation, ausing arrest of the G1 and G2 phases of the ell yle (Bates et al, 1998; Stott et al, 1998). Therefore loss of funtion of either p16 INK4a or p14 ARF may lead to unrestrained ell yling and unontrolled ell growth, a ommon alteration in the arinogeni proess. Inativation of the two TSG at the CDKN2A lous an our by a variety of geneti mehanisms inluding mutations and deletions (Liggett and Sidransky, 1998). Due to the unique arrangement of the exons, a mutation in exon 2 has the potential to alter the amino aid sequene of either or both of the genes. In addition to geneti mehanisms, hypermethylation of the CpG islands of ertain gene promoters, inluding p16 INK4a and p14 ARF, is a signifiant means of transriptional silening in a variety of tumor types (Esteller et al, 2001b). The importane of this epigeneti mehanism of gene silening in almost every stage of the Copyright r 2004 by The Soiety for Investigative Dermatology, In. 1284

2 122 : 5 MAY 2004 INACTIVATION OF p16 INK4a AND p14 ARF IN CUTANEOUS SCC 1285 initiation and progression of human aner has only reently begun to be fully appreiated (Jones and Baylin, 2002). Of the two genes, p16 INK4a has been more intensively studied and has been proven onlusively to play an important role in arinogenesis, partiularly head and nek SCC and familial melanoma (Reed et al, 1996; Liggett and Sidransky, 1998). The role of p14 ARF inativation in human aners is less well established, but is supported by studies of transgeni mie arrying a disrupted exon 1b whih are aner-prone at an early age (Kamijo et al, 1997). In addition, both p14 ARF mutations and methylation an have a deleterious funtional effet (Zhang and Xiong, 1999; Rizos et al, 2001; Esteller et al, 2001a). In utaneous SCC there has been relatively little published onerning the inativation of p16 INK4a and p14 ARF genes. Knokout mie experiments suggest a possible role for p14 ARF in the initiation of utaneous SCC development (Serrano et al, 1996; Kamijo et al, 1997) and for p16 INK4a in determining the differentiation status of skin squamous tumors (Sharpless et al, 2001). In primary utaneous SCC, loss of heterozygosity (LOH) at the 9p21 lous and CDKN2A mutations have been reported but with disparate results (Quinn et al, 1994a, b; Kubo et al, 1997; Kushida et al, 1999; Soufir et al, 1999; Saridaki et al, 2003). Only one study has addressed the important mehanism of gene promoter hypermethylation in this tumor type, looking at p16 INK4a alone (Soufir et al, 1999) and no previous work has examined p14 ARF promoter methylation. We have performed a omprehensive study of the CDKN2A lous in utaneous SCC to establish the importane of p16 INK4a and p14 ARF inativation in the pathogenesis of this tumor. Using highly speifi and sensitive tehniques, a variety of possible mehanisms for gene silening have been examined, inluding orrelation of the results with protein expression. Results LOH analysis A total of 13 out of the 40 tumors (32.5%) exhibited alleli loss/loh at one or more markers for 9p21. Eleven samples showed loss at all informative markers. This data is summarized in Table I and representative examples are given in Fig 1. No samples showed mirosatellite instability at this lous. Mutational analysis Five different point mutations were found in four of 40 SCC (Table II and Fig 2). All mutations were in exon 2 and onsisted of four single C to T nuleotide hanges and one double substitution CC to TT. Sine these all ourred in dipyrimidine sites, they are onsistent with a UVB fingerprint type of mutation. Corresponding genomi DNA did not show these mutations onfirming tumor speifiity. Sample 31T had two different independent point mutations in exon 2. LOH analysis had previously shown loss of one allele in this tumor speimen implying that both mutations must have ourred in the same allele. This was onfirmed by subsequent sub-loning (data not shown). Sample 49T had a double CC to TT mutation, whih would alter the amino aid sequene in two odons of p16 INK4a but only one of p14 ARF. Table I. LOH results Case no. D9S171 D9S1748 D9S N LOH % LOH N , loss of heterozygosity (LOH);, retention of heterozygosity;, homozygous/not informative;, no result/not done.

3 1286 BROWN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY Figure 1 Loss of heterozygosity (LOH) analysis. Units on the y-axis represent intensity of fluoresent signal; on the x-axis is allele length in base pairs. The y-axis saling differs between samples beause quantities of PCR produts vary in intensity. Examples show retention of heterozygosity (LOH index 0.64) for sample 14 at marker D9S171 and omplete loss of the upper allele (LOH) for sample 31T ompared to mathed blood, 31B (arrows) at marker D9S1748. Note stutter bands are evident prior to the allele (highest peak) signal. Two other samples 45T and 48T (and their orresponding genomi DNA) showed a G to A nuleotide polymorphism that is seen in 4% 6% of the general population (Bertram et al, 2002). Methylation analysis Using MSP, evidene of methylation was observed for the p16 INK4a promoter in 13 of 36 (36%) samples and for the p14 ARF promoter in 16 of 38 (42%) samples (Fig 3). When initial bisulfite modifiation of DNA was suessful, all samples produed a produt with the primers speifi for unmethylated DNA. This is due to amplifiation of the normal stromal tissue present in the initial biopsy speimen. In addition, in those samples with positive methylation, it is possible that the unmethylated band ould also represent the partial methylation status of the tumor ells or geneti diversity of the tumor ell population. Four samples were methylated for both the p16 INK4a and p14 ARF promoters: 32T, 35T, 44T, and 53T. Bialleli events Combining all the results from LOH studies, mutation sreening and methylation analysis, bialleli hanges (i.e. two of these three possible events) were present in ten of 38 samples for p16 INK4a and six of 39 samples for p14 ARF (Table III). Immunohistohemial results p16 INK4a expression was deteted in 20 of 40 utaneous SCC, of whih 75% were graded þ (Fig 4). Neither the amount nor the intensity of staining orrelated with tumor differentiation status. In 23 of 36 (64%) ases there was onordane between the presene/absene of protein expression and geneti/ epigeneti results. In seven of ten (70%) of tumors with bialleli geneti/epigeneti events, there was no protein expression evident. p14 ARF expression was seen in 17 of 40 (42.5%) of the samples; 14 of these being lassified as þ.in23of37 (62%) ases there was onordane between the presene/absene of protein expression and geneti/epigeneti results. In all of the six tumors with bialleli geneti/ epigeneti results, there was no protein expression evident. Comparison of immune status with mehanisti data For p16 INK4a, eight of ten (80%) immunoompetent patients had geneti/epigeneti events ompared with 12 of 28 (43%) immunosuppressed individuals (Table IV). Results were similar for p14 ARF with 9 of 10 (90%) vs 16 of 29 (55%) respetively (p ¼ 0.05 for both p16 INK4a and p14 ARF, Fisher s exat probability test). Table V summarizes the data on immune status, p14 ARF and p16 INK4a events (mutation, methylation, and immunohistohemial expression) and highlights those tumors in whih bilalleli events were identified. Disussion In this study we report the systemati analysis of the CDKN2A lous in a well-haraterized series of utaneous SCC. We demonstrate that alterations at the CDKN2A lous are ommon in these aners and we show inativation of the two TSG at this lous via both geneti and epigeneti mehanisms. Our results show that LOH at 9p21 is frequent in utaneous SCC. The frequeny of 9p21 alleli loss observed in this study (32.5%) is onsistent with the largest previous study by Quinn et al (1994b) who found LOH in 31% of samples. But in two smaller series of aners, Kushida et al (1999) and Soufir et al (1999) were unable to detet LOH. These three studies all used fewer markers at 9p21 in mostly non-mirodisseted tumor DNA, with a non-quantitative method of assessing LOH. Saridaki et al Table II. Mutations found in CDKN2A exon 2 and the onsequent hanges in amino aid sequene for p16 INK4a and p14 ARF proteins Sample p16 INK4a odon hange p16 INK4a amino aid hange p14 ARF odon hange p14 ARF amino aid hange 31T ) t Pro81Leu a ) at Thr142Thr 31T tt ) ttt Phe 90 Phe t ) tt Pro152Ser 41T ) t Pro114Leu g ) gt Ala175Ala 49T g ) gt Ala57Ala g ) ttg Pro118Leu ga ) tga Arg58STOP g ) ttg Pro118Leu 54T ga ) tga Arg80STOP g ) tg Pro141Leu All nuleotide hanges C! T substitutions exept for 49T that had a tandem CC! TT mutation.

4 122 : 5 MAY 2004 INACTIVATION OF p16 INK4a AND p14 ARF IN CUTANEOUS SCC 1287 Figure 3 Representative examples of 10% arylamide gels of methylation speifi PCR produts. The presene of a visible band in U lanes indiates unmethylated genes (normal stromal tissue); in M lanes indiates methylated genes. (A) p16 INK4a is methylated in sample 21. (B) p14 ARF is methylated in samples 3, 8, 27. Positive (Pos) ontrols were universally methylated or unmethylated DNA for the M and U PCR, respetively. H 2 O served as a negative (Neg) ontrol for both reations. Controls were inluded in every run. L ¼ marker ladder. Table III. Summary of monoalleli and bialleli moleular abnormalities in utaneous SCC Figure 2 Examples of mutations found in two utaneous SCCs. Denaturing high-performane liquid hromatography traes (left) and orresponding sequening results (right) for samples 31 (upper panel) and 54 (lower panel). Sequening traes show heterozygous C! T base hanges (arrowed) in exon 2 at nuleotide (nt) positions 242 and 270 for 31T; 238 for 54T. Moleular abnormalities No. of SCC with p16 INK4a alterations No. of SCC with p14 ARF alterations None LOH alone 4 8 Mutation alone 1 0 Methylation alone 5 11 LOH þ mutation 2 1 LOH þ methylation 7 3 Mutation þ 1 1 methylation LOH þ mutation þ methylation 0 1 SCC, squamous ell arinoma; LOH, loss of heterozygosity. (2003) has reently shown a higher rate of LOH of 52% in a smaller series, whih ombined Bowen s disease (arinoma in situ) and SCC. The samples in the present work were all histologially onfirmed invasive SCC. Furthermore, mirodisseted DNA was tested and a quantitative method was used to alulate LOH. As suh, our results are likely to be a more aurate estimate of the true prevalene of LOH at this lous in utaneous SCC. Using DHPLC, point mutations were deteted in only four of 40 samples (10%). The sensitivity of our methodology is validated by detetion of the Ala148Thr polymorphism that is found in 4% 6% of the general population (Bertram et al, 2002) and was deteted in two of 40 of our samples (both tumor and genomi DNA from samples 45 and 48). All the mutations arried a UV signature (Brash et al, 1991), refleting the importane of UV in the etiology of these tumors, whih all arose on sun-exposed sites. This finding also provides evidene of a role for mutagenesis of the p16 INK4a and p14 ARF genes in the pathogenesis of UV-indued SCC. Eah of the mutations ourred in exon 2, the exon most ommonly mutated in other human aners. For p16 INK4a, two of the mutations result in amino aid substitutions (Pro81Leu and Pro114Leu), whereas the other two mutations result in premature termination of the protein (Arg58- STOP and Arg80STOP). Eah of these mutations has been previously desribed in other human aner types (Pollok et al, 1996; Kubo et al, 1997; Fargnoli et al, 1998; Soufir et al, 1999; Bazan et al, 2002). The two non-trunating mutations, Pro81Leu and Pro114Leu, whih both our in the key ankyrin repeat domains, result in funtional impairment of the p16 INK4a protein, (Koh et al, 1995; Ranade et al, 1995). Both of the premature termination p16 INK4a mutations result in expression of trunated proteins laking CDK binding ativity (Parry and Peters, 1996). Three of the p16 INK4a mutations also affet p14 ARF. Of these, one (Pro118Leu) has previously been desribed in a utaneous SCC (Soufir et al, 1999). The

5 1288 BROWN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY Figure 4 Immunohistohemistry slides showing predominant nulear staining for p16 INK4a antibody in tumor 44 and p14 ARF antibody in tumor 29. Absene of expression is noted for p16 INK4a in tumor 54 and p14 ARF in tumor 31. Both these tumors had bialleli inativating events detetable. Table IV. Comparison of geneti events in immunosuppressed versus immunoompetent patients p16 INK4a alterations p14 ARF alterations Present Absent Present Absent Immunosuppressed Immunoompetent p ¼ p ¼ effet of these mutations on the funtion of p14 ARF awaits formal haraterization; however, other human anerassoiated point mutations have been shown to funtionally impair p14 ARF by diminishing its ability to ativate and stabilize p53 (Zhang and Xiong, 1999; Rizos et al, 2001). As desribed above, the mutations in CDKN2A target both p16 INK4a and p14 ARF in three ases, implying seletive pressure to abrogate the funtion of both proteins during tumorigenesis in squamous epithelium. Unusually, 31T in whih there was unequivoal evidene of LOH, ontained two independent exon 2 missense mutations on the retained allele. One of these affets the amino aid sequene of p16 INK4a, the other p14 ARF. The presene of distint mutations targeting eah gene in 31T, 49T, and 54T, strongly argues that loss of funtion in both proteins is a key event in arinogenesis in utaneous SCC. In this study, we did not examine exon 3 (whih ontributes oding sequene to only the four most -terminal amino aids of p16 INK4a ) nor did we analyze homozygous deletions. Homozygous deletions, while relatively ommon in arinoma ell lines are rare in primary arinomas (e.g. primary gastri arinomas, Lee et al, 1997). Similarly, bialleli deletion is very rare in primary arinomas of prostate and liver (Herman et al, 1995; Peng et al, 2002). Our results show that methylation of the normally unmethylated promoter regions of both genes is the most ommon mehanism of inativation in utaneous SCC. p14 ARF promoter methylation has never previously been examined in utaneous SCC and was the ommonest abnormality found in our series. In addition, this is the first formal demonstration of p16 INK4a promoter methylation in utaneous SCC. In a previous study using Southern blotting, it was reported that the p16 INK4a promoter was not methylated in utaneous SCC (Soufir et al, 1999). It is well reognized that the sensitivity of this method is substantially lower than MSP for detetion of methylation (Herman et al, 1996). Moreover, our study analyzed 40 ases in ontrast to the eight SCC analyzed by Soufir and o-workers. These fators may well aount for the differenes between the two studies. Our results aord well with studies using MSP in other squamous aners (Reed et al, 1996; Gaso et al, 2002; Huang et al, 2002). Further analysis of our results reveals that p16 INK4a was hypermethylated in the presene of an unmethylated p14 ARF in nine of 13 (69%) of the ases; p16 INK4a was unmethylated with p14 ARF hypermethylation in 12 of 16 (75%) of tumors and four (10%) ases demonstrated methylation of both gene promoters. Thus, onsistent with previous studies of oloretal arinomas, methylation of the p16 INK4a and p14 ARF promoters are independent events (Esteller et al, 2000).

6 122 : 5 MAY 2004 INACTIVATION OF p16 INK4a AND p14 ARF IN CUTANEOUS SCC 1289 Table V. Summary table of all results for both p16 INK4a and p14 ARF genes p16 INK4a alterations p14 ARF alterations Sample no. IC/IS LOH a Mutation b Methylation Expression Mutation b Methylation Expression 1 IS Ret nr þ 3 IS Ret þ þ þ 5 IS Ret 8 IS Ret þ 11 IS Ret þ þ 14 IS Ret þ 17 IS Ret þ þ þ 18 IS LOH þ 19 IS Ret nr nr þ 20 IS LOH nr nr 21 IS LOH þ þ 24 IC LOH nr þ þ 25 IS Ret þ þ 26 IS Ret 27 IS Ret þ þ 28 IS Ret þ 29 IS Ret þ þ þ þ þ 31 IC LOH Y Y 32 IC LOH þ þ 33 IS Ret þ 34 IS Ret þ þ 35 IS LOH þ þ 36 IC LOH þ 37 IC Ret þ þ 38 IS LOH þ 39 IS LOH þ þ 40 IC LOH þ þ þ 41 IC Ret Y þ þ N 42 IS Ret þ þ 44 IS Ret þ þ þ þ þ þ 45 IS Ret þ þ þ 46 IS Ret þ þ þ 47 IS Ret 48 IC Ret þ þ þ 49 IC Ret Y Y þ 50 IS Ret þ þ þ þ 51 IC LOH þ þ 52 IS Ret þ 53 IS Ret þ þ þ 54 IS LOH Y Y þ Bialleli events, d, d, d d d, d IC, immunoompetent; IS, immunosuppressed; LOH, loss of heterozygosity; Ret, retention of heterozygosity; nr, no result. a See Table I for details. b See Table II for details. Bialleli events for p16 INK4a. d Bialleli events for p14 ARF.

7 1290 BROWN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY Epigeneti hanges may predate geneti events in the natural history of some arinomas. For example, p14 ARF hypermethylation is present in 32% of oloretal adenomas (Esteller et al, 2000) and p16 INK4a hypermethylation ours in pre-malignant, dysplasti oral squamous lesions (Gaso et al, 2002). It would therefore be of interest to test the premalignant forms of utaneous SCC, namely Bowen s disease and atini keratoses, and hronially UV-exposed skin, for this epigeneti abnormality. Inativation of the p16 INK4a gene was assoiated with absent protein expression in almost all tumor ells in a study of head and nek SCC (Reed et al, 1996) and non-small ell lung aner (Sanhez-Cespedes et al, 1999). Absent mrna was also found to orrelate with lak of p16 INK4a protein expression in a study of myosis fungoides (Peris et al, 1999). In our study onordane with geneti events was only found in 64% and 62% of samples for p16 INK4a and p14 ARF, respetively. The most likely explanation for this is that methylation is not homogeneous in the SCC under study, but, rather, is restrited to a sub-population of tumor ells. This latter suggestion is supported by the immunohistohemial findings, whih in most positive ases was pathy and heterogeneous throughout the tumor areas. This resembles the expression pattern seen in p16 INK4a methylated oral and ervial SCC (Nuovo et al, 1999; Huang et al, 2002). Tumors with bialleli inativating events would be expeted to lak funtional protein. Immunohistohemistry onfirmed that this was indeed the ase in 13 of 16 of the studied ases (all six of the p14 ARF ases and seven of ten of the p16 INK4a ases). In eah of the three tumors that retained some p16 INK4a expression, one of the events was methylation of the p16 INK4a promoter. The likely explanation for these observations is that methylation-dependent silening ourred only in speifi parts of the tumor, and that residual areas of unmethylated p16 INK4a were present in eah ase in whih p16 INK4a was expressed. Use of in situ MSP has shown onsiderable heterogeneity in squamous arinomas methylation, onsistent with this explanation (Nuovo et al, 1999). Three tumors, 31T, 32T, and 54T, had bialleli events in both genes and all had no expression of either protein on immunohistohemistry. For both p16 INK4a and p14 ARF signifiantly more tumors from immunoompetent individuals had geneti/epigeneti inativating events ompared with utaneous SCC from immunosuppressed individuals. These results, however, are in keeping with a previous study whih reported less alleli loss at hromosome 9p in skin tumors from immunosuppressed ompared with immunoompetent patients (Rehman et al, 1997). The explanation for these findings remain elusive, but one ould speulate that the signifiantly higher prevalene of human papillomavirus found in immunosuppressed utaneous SCC obviates the need for diret geneti/epigeneti TSG alterations, similar to the mehanism proven in anogenital arinomas (Harwood and Proby, 2002). In onlusion, this is the largest and most omprehensive study of the CDKN2A lous in utaneous SCC reported to date. Overall, the total frequeny of 9p21 alterations was 76%, with abnormalities of p16 INK4a deteted in 53% of ases studied and p14 ARF hanges present in 64%. Promoter methylation in utaneous SCC has never previously been demonstrated at this lous, but was the predominant mehanism of inativation for both genes in our series. Bialleli events were ommon. These findings emphasize the importane of p16 INK4a and p14 ARF TSG in the pathogenesis of utaneous SCC. Materials and Methods Patient harateristis Forty utaneous, invasive SCC were obtained from 39 patients, following informed onsent. These omprised 30 samples from 29 iatrogenially immunosuppressed patients and 10 samples from 10 immunoompetent patients. Twenty-nine patients were male and 10 female. All speimens were primary lesions and arose on sun-exposed areas of skin; namely fae, salp, upper limbs, and upper trunk. Speimen preparation A small portion of tumor tissue was snap frozen immediately at operation and stored at 701C. The remainder was formalin-fixed, proessed routinely and embedded in paraffin. The diagnosis was onfirmed histologially by H&E staining of all samples. Venous blood was obtained from all but one patient. DNA extration Genomi DNA was extrated from the frozen tissue and venous blood using the Nuleon Genomi DNA Extration Kit (Tepnel Life Sienes, Manhester, UK) aording to the supplied protool. Mirodissetion and DNA isolation For the alleli imbalane studies tumor-enrihed DNA was required. This was obtained by mirodissetion of a majority tumor ell population from up to six 8 mm setions ut from the paraffin bloks from eah ase. The tumor area on eah slide was identified by referene to an adjaent hematoxylin and eosin-stained slide. The tumor tissue was obtained by sraping off unstained, waxed slides, with a singleedged razor blade. The tissue was plaed in an eppendorf mirofuge tube and dewaxed in 1 ml xylene on a rotating wheel for 1 h. After all the xylene had been removed, 4 ml Proteinase K solution (14 mg per ml; Rohe Diagnostis Ltd., East Sussex, UK) and 100 ml 10% Chelex (Bio-Rad Laboratories, Herules, CA) solution were added, and the mixture inubated in a rotating oven at 561C for 24 h. The tubes were then boiled at 1001C for 8 min to inativate the Proteinase K. After entrifugation, 5 ml aliquots of the supernatant were used in the subsequent allelotyping PCR. The samples were stored at 41C until used. From the one patient for whom venous blood was not obtained, normal skin was mirodisseted and genomi DNA extrated as desribed above. Allelotype/LOH analysis Mirodisseted tumor DNA and mathed genomi DNA from eah patient was analyzed for alleli loss using three highly polymorphi mirosatellite (dinuleotide repeat) markers, D9S171, D9S1748, and D9S974, whih span the CDKN2A lous at 9p21. These markers were hosen beause of their loation with respet to the CDKN2A exons (Fig 5). In addition they all have a high informativity rate (heterozygosity sore) and produe a PCR produt of a small enough size for suessful amplifiation of DNA from paraffin-embedded samples. All primer oligonuleotide sequenes are available through The Genome Database ( For eah primer pair, the 5 0 end of the forward primer was synthesized with a fluoresent label (6FAM or HEX) using amidite-labelling hemistry. Amplifiation was arried out using a standard PCR protool (PCR reagents inluding Taq polymerase supplied by Bioline, London, UK) on a Hybaid Cyler (Hybaid Ltd, Middlesex, UK) using a 601C annealing temperature for D9S171 and 651C for D9S1748 and D9S974. The PCR produts produed for eah sample were pooled in a 1:1 ratio in the presene of Hi-Di formamide (Applied Biosystems Warrington, UK) and a 500-ROX internal size standard (Applied Biosys-

8 122 : 5 MAY 2004 INACTIVATION OF p16 INK4a AND p14 ARF IN CUTANEOUS SCC 1291 tems). Genotyping was performed on an ABI 3100 automated DNA Analyzer (Applied Biosystems), whih is apillary based and laser ativated to detet fluoresent PCR produts. The resultant allele profiles for eah sample were analyzed using ABI Genesan 3.1 software (Applied Biosystems), whih yields quantifiation of results in terms of trae peak size and height. Assessment of allele loss Different marker traes were distinguishable by size and fluoresent tag/olor. Heterozygous individuals were identified i.e. those with two PCR produts of different size in normal DNA. The sizes of the two alleles were assigned to the peaks of the greatest height; smaller peaks were interpreted as stutter bands aused by slippage of the polymerase. LOH was assessed for eah blood/tumor pair by a quantitative method alulating the LOH index (Okabe et al, 2000). The LOH index is defined as the allele ratio in normal DNA divided by the allele ratio in the orresponding tumor DNA. The allele ratio is the peak height of the smaller length allele divided by the peak height of the larger length allele. Alleli loss was defined by an LOH index of less than 0.5 or more than 2.0 i.e. at least a 50% signal hange in tumor allele profile ompared to normal allele profile. Mutational analysis by denaturing high-performane liquid hromatography (DHPLC) and diret sequening Exons 1a, 1b, and 2 of CDKN2A were amplified for eah sample using primers that spanned the entire exon and intron exon boundary. The PCR produts for exons 1b and 2 were split to provide two overlapping PCR produts for ease of subsequent analysis. The primer sequenes used were as follows (PCR produt sizes in base pairs (bp) in brakets): exon 1a forward: GAAGAAAGAGGA-GGGGCTG; reverse: GCGCTACCTGATTCCAATTC (340 bp); exons 1b-1 forward: TCAGGGAAGGCGGGTGCG; reverse: GCCGCGGGATGT- GAACCA (245 bp); Exons 1b-2 forward: GCCGCGAGTGAGGGTTTT; reverse: CACCGGCGGTTATCTCCTC (263 bp); Exon 2A forward: AGCTTCC-TTTCCGTCATGC; reverse: GCAGCACCACCAGCGTG (203 bp); Exon 2B forward: GACCCCGCCACTCTCACC; reverse: GTGCTGGAAAATGAATGC-TCTG (310 bp). PCR amplifiation was arried out using ng of the frozen tissue genomi DNA in a standard PCR protool (PCR reagents inluding Taq polymerase supplied by Bioline). Five perent dimethyl sulfoxide (BDH Laboratories, Poole, UK) was added to the reation mixture for optimal PCR amplifiation of GC-rih regions. The annealing temperature for exons 1b and 2-2 was 601C and for exon 1a and 2A was 551C. The PCR produts obtained were heated to 941C PCR for 5 min followed by ooling to 401C at a rate of 0.031C per s to ensure heteroduplex formation. Mutational analysis was then arried out using DHPLC with the Transgenomi WAVE System (Transgenomi, UK). The optimum temperatures for the analysis of eah PCR fragment were alulated using the WAVE-Maker software (Transgenomi). Mutations in samples showing DHPLC traes that varied from the wild-type profile were identified by sequening an independent PCR produt. These PCR produts were first purified using a PCR purifiation kit (QIAquik; Qiagen, Crawley, UK), followed by diret sequening using ABI Prism Big- Figure 5 Loation of mirosatellite markers used in allelotyping, relative to CDKN2A exons. Dye terminator hemistry (Applied Biosystems) in aordane with the manufaturer s instrutions. Eah sample was sequened in both forward and reverse diretions in independent PCR reations and the produts were analyzed on an ABI 377 automated sequener. Sequene eletropherograms were aligned using Sequene Navigator software (Applied Biosystems). Methylation analysis For methylation speifi PCR (MSP), genomi DNA was bisulfite modified using the reagents supplied in the CpGenome Modifiation Kit (Intergen Company, Oxford, UK), aording to the manufaturer s instrutions. To analyze the methylation status of both the p16 INK4a and the p14 ARF gene promoter regions, MSP was arried out with separate reation mixtures using unmethylated (U) and methylated primers (M). Control samples were inluded in every run: positive ontrols were bisulfite modified universally methylated or unmethylated DNA (Intergen) in the M and U PCR, respetively. Negative ontrols substituted distilled water for DNA in the PCR. A hot-start PCR was employed using AmpliTaq Gold polymerase (Rohe). For p16 INK4a MSP was arried out using the CpG WIZ Amplifiation Kit (Intergen). For p14 ARF MSP, primers were synthesized as previously desribed (Esteller et al, 2000) and used in a 25 ml reation mixture omposed of: 2.5 ml 10 GeneAmp PCR Buffer II, 2.0 ml MgCl 2 solution 25 mm, 4.0 ml 5 mm dntp mix, U or M primer 20 pmol, 1 U AmpliTaq Gold polymerase and 5 ml (50 ng) bisulfite modified DNA. The annealing temperature for both the U and M reations was 601C. PCR produts were resolved on 10% non-denaturing arylamide gels. The sizes of the PCR fragments for the p16 INK4a U and M forms are 154 and 145 bp, respetively, and for the p14 ARF U and M forms are 132 and 122 bp, respetively. Immunohistohemial analysis Serial 4 mm setions of formalinfixed, paraffin-embedded biopsy samples were ut and one stained with H&E to onfirm the histologial diagnosis. Further setions were stained for both p16 INK4a and p14 ARF proteins using ommerially available primary antibodies: p16 INK4a /MTS1 Ab-7 (16P07) mouse monolonal antibody (NeoMarkers In., Fremont, California) and p14 ARF (C-18) (s-8613) goat polylonal antibody (Santa Cruz Biotehnology, In.) Setions were dewaxed in xylene and rehydrated through graded alohol. Endogenous peroxidase ativity was inhibited by inubating in 3% hydrogen peroxide in methanol for 10 min. Antigen retrieval was performed by boiling in 10 mm itrate buffer ph 6.0, for 4 min in a mirowave. After bloking of non-speifi binding by pre-inubation with donor horse/rabbit serum for 30 min, the primary antibody at a onentration of 1:100 was added for overnight inubation at 41C. For detetion, biotinylated horse anti-mouse (for p16 INK4a ) or biotinylated rabbit anti-goat (for p14 ARF ) sera (Vetastain ABC Kit, Vetor Laboratories, Burlingame, California) were applied as seondary antibodies for 30 min at room temperature, followed by inubation with the avidin biotin omplex hromogen (Vetastain) for 20 min at room temperature. The reation was developed using DAB (3,3 0 -diaminobenzidine) hromogen solution (Biogenex Liquid DAB, Biogenex, California) and ounterstained with hematoxylin. Appropriate positive and negative ontrol slides were inluded in every run. Interpretation of immunostaining Slides were assessed by two independent observers (VB and CP). CP had no prior knowledge of the linial details or moleular studies. p16 INK4a and p14 ARF are both nulear proteins and only nulear reativity was onsidered positive for protein expression. Internal positive ontrols inluded infiltrating inflammatory ells and basal keratinoytes for p16 INK4a and appendageal strutures, suh as hair folliles and glandular strutures; for p14 ARF. Staining was graded as negative ( ), less than 5% of tumor nulei positive or ytoplasmi staining only ( ), less than 25% of tumor ells positive ( þ ), 25% 50% of tumor ells positive ( þþ) or more than 50% of tumor ells positive ( þþþ). Ethial approval This study was approved by the East London and City Health Authority Researh Ethis Committee.

9 1292 BROWN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY Vitoria Brown is a Medial Researh Counil Clinial Training Fellow. The authors would like to thank Luy Ghali, Sott Edmunds, David Ballard, and Dirk Brinkmann for tehnial assistane. DOI: /j X x Manusript reeived July 10, 2003; revised Deember 3, 2003; aepted for publiation Deember 5, 2003 Address orrespondene to: Charlotte M. Proby, Centre for Cutaneous Researh 2, Newark Street, Whitehapel, London E1 2AT. .proby@aner.org.uk Referenes Alam M, Ratner D: Cutaneous squamous-ell arinoma. N Engl J Med 344: , 2001 Bates S, Phillips AC, Clark PA, Stott F, Peters G, Ludwig RL, Vousden KH: p14arf links the tumour suppressors RB and p53. Nature 395: , 1998 Bazan V, Zanna I, Migliavaa M, et al: Prognosti signifiane of p16ink4a alterations and 9p21 loss of heterozygosity in loally advaned laryngeal squamous ell arinoma. J Cell Physiol 192: , 2002 Bertram CG, Gaut RM, Barrett JH, et al: An assessment of the CDKN2A variant Ala148Thr as a nevus/melanoma suseptibility allele. J Invest Dermatol 119: , 2002 Brash DE, Rudolph JA, Simon JA, et al: A role for sunlight in skin aner: UVindued p53 mutations in squamous ell arinoma. Pro Natl Aad Si USA 88: , 1991 Esteller M, Cordon-Cardo C, Corn PG, et al: p14arf silening by promoter hypermethylation mediates abnormal intraellular loalization of MDM2. Caner Res 61: , 2001a Esteller M, Corn PG, Baylin SB, Herman JG: A gene hypermethylation profile of human aner. Caner Res 61: , 2001b Esteller M, Tortola S, Toyota M, Capella G, Peinado MA, Baylin SB, Herman JG: Hypermethylation-assoiated inativation of p14(arf) is independent of p16(ink4a) methylation and p53 mutational status. Caner Res 60: , 2000 Fargnoli MC, Chimenti S, Keller G, Soyer HP, Dal Pozzo V, Hofler H, Peris K: CDKN2a/p16INK4a mutations and lak of p19arf involvement in familial melanoma kindreds. J Invest Dermatol 111: , 1998 Gaso M, Bell AK, Heath V, et al: Epigeneti inativation of sigma in oral arinoma: Assoiation with p16(ink4a) silening and human papillomavirus negativity. Caner Res 62: , 2002 Grossman D, Leffell DJ: The moleular basis of nonmelanoma skin aner: New understanding. Arh Dermatol 133: , 1997 Harwood CA, Proby CM: Human papillomaviruses and non-melanoma skin aner. Curr Opin Infet Dis 15: , 2002 Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB: Methylation-speifi PCR: A novel PCR assay for methylation status of CpG islands. Pro Natl Aad Si USA 93: , 1996 Herman JG, Merlo A, Mao L, et al: Inativation of the CDKN2/p16/MTS1 gene is frequently assoiated with aberrant DNA methylation in all ommon human aners. Caner Res 55: , 1995 Holme SA, Malinovszky K, Roberts DL: Changing trends in non-melanoma skin aner in South Wales, Br J Dermatol 143: , 2000 Huang MJ, Yeh KT, Shih HC, et al: The orrelation between CpG methylation and protein expression of P16 in oral squamous ell arinomas. Int J Mol Med 10: , 2002 Johnson TM, Rowe DE, Nelson BR, Swanson NA: Squamous ell arinoma of the skin (exluding lip and oral muosa). J Am Aad Dermatol 26: , 1992 Jones PA, Baylin SB: The fundamental role of epigeneti events in aner. Nat Rev Genet 3: , 2002 Kamijo T, Zindy F, Roussel MF, et al: Tumor suppression at the mouse INK4a lous mediated by the alternative reading frame produt p19arf. Cell 91: , 1997 Koh J, Enders GH, Dynlaht BD, Harlow E: Tumour-derived p16 alleles enoding proteins defetive in ell-yle inhibition. Nature 375: , 1995 Kubo Y, Urano Y, Matsumoto K, Ahsan K, Arase S: Mutations of the INK4a lous in squamous ell arinomas of human skin. Biohem Biophys Res Commun 232:38 41, 1997 Kushida Y, Miki H, Ohmori M: Loss of heterozygosity in atini keratosis, squamous ell arinoma and sun-exposed normal-appearing skin in Japanese: Differene between Japanese and Cauasians. Caner Lett 140: , 1999 Lee YY, Kang SH, Seo JY, et al: Alterations of p16ink4a and p15ink4b genes in gastri arinomas. Caner 80: , 1997 Liggett WH Jr, Sidransky D: Role of the p16 tumor suppressor gene in aner. J Clin Onol 16: , 1998 Lindelof B, Sigurgeirsson B, Gabel H, Stern RS: Inidene of skin aner in 5356 patients following organ transplantation. Br J Dermatol 143: , 2000 Martinez JC, Otley CC, Stasko T, Euvrard S, Brown C, Shanbaher CF, Weaver AL: Defining the linial ourse of metastati skin aner in organ transplant reipients: A multienter ollaborative study. Arh Dermatol 139: , 2003 Nuovo GJ, Plaia TW, Belinsky SA, Baylin SB, Herman JG: In situ detetion of the hypermethylation-indued inativation of the p16 gene as an early event in onogenesis. Pro Natl Aad Si USA 96: , 1999 Okabe H, Ikai I, Matsuo K, et al: Comprehensive allelotype study of hepatoellular arinoma: Potential differenes in pathways to hepatoellular arinoma between hepatitis B virus-positive and -negative tumors. Hepatology 31: , 2000 Parry D, Peters G: Temperature-sensitive mutants of p16cdkn2 assoiated with familial melanoma. Mol Cell Biol 16: , 1996 Peng CY, Chen TC, Hung SP, et al: Geneti alterations of INK4alpha/ARF lous and p53 in human hepatoellular arinoma. Antianer Res 22: , 2002 Peris K, Stanta G, Fargnoli MC, Bonin S, Felli A, Amantea A, Chimenti S: Redued expression of CDKN2a/P16INK4a in myosis fungoides. Arh Dermatol Res 291: , 1999 Pollok PM, Pearson JV, Hayward NK: Compilation of somati mutations of the CDKN2 gene in human aners: Non-random distribution of base substitutions. Genes Chromosomes Caner 15:77 88, 1996 Quinn AG, Sikkink S, Rees JL: Basal ell arinomas and squamous ell arinomas of human skin show distint patterns of hromosome loss. Caner Res 54: , 1994a Quinn AG, Sikkink S, Rees JL: Delineation of two distint deleted regions on hromosome 9 in human non-melanoma skin aners. Genes Chromosomes Caner 11: , 1994b Ramsay HM, Fryer AA, Hawley CM, Smith AG, Harden PN: Non-melanoma skin aner risk in the Queensland renal transplant population. Br J Dermatol 147: , 2002 Ranade K, Hussussian CJ, Sikorski RS, et al: Mutations assoiated with familial melanoma impair p16ink4 funtion. Nat Genet 10: , 1995 Reed AL, Califano J, Cairns P, et al: High frequeny of p16 (CDKN2/MTS-1/ INK4A) inativation in head and nek squamous ell arinoma. Caner Res 56: , 1996 Rehman I, Quinn AG, Takata M, Taylor AEM, Rees JL: Low frequeny of alleli loss in skin tumors from immunosupressed individuals. Br J Caner 76: , 1997 Rizos H, Darmanian AP, Holland EA, Mann GJ, Kefford RF: Mutations in the INK4a/ARF melanoma suseptibility lous funtionally impair p14arf. J Biol Chem 276: , 2001 Sanhez-Cespedes M, Reed AL, Buta M, et al: Inativation of the INK4A/ARF lous frequently oexists with TP53 mutations in non-small ell lung aner. Onogene 18: , 1999 Saridaki Z, Liloglou T, Zafiropoulos A, Koumantaki E, Zoras O, Spandidos DA: Mutational analysis of CDKN2A genes in patients with squamous ell arinoma of the skin. Br J Dermatol 148: , 2003 Serrano M, Hannon GJ, Beah D: A new regulatory motif in ell-yle ontrol ausing speifi inhibition of ylin D/CDK4. Nature 366: , 1993 Serrano M, Lee H, Chin L, Cordon-Cardo C, Beah D, DePinho RA: Role of the INK4a lous in tumor suppression and ell mortality. Cell 85:27 37, 1996 Sharpless NE, Bardeesy N, Lee KH, et al: Loss of p16ink4a with retention of p19arf predisposes mie to tumorigenesis. Nature 413:86 91, 2001 Soufir N, Moles JP, Vilmer C, et al: P16 UV mutations in human skin epithelial tumors. Onogene 18: , 1999 Stone S, Jiang P, Dayananth P, et al: Complex struture and regulation of the P16 (MTS1) lous. Caner Res 55: , 1995 Stott FJ, Bates S, James MC, et al: The alternative produt from the human CDKN2A lous, p14(arf), partiipates in a regulatory feedbak loop with p53 and MDM2. EMBO J 17: , 1998 Zhang Y, Xiong Y: Mutations in human ARF exon 2 disrupt its nuleolar loalization and impair its ability to blok nulear export of MDM2 and p53. Mol Cell 3: , 1999