Homozygous deletion of the p16INK4a and the p15INK4b tumour suppressor genes in a subset of human...

Homozygous deletion of the p16INK4a and the p15INK4b tumoursuppressor genes in a subset of human sporadic cutaneousmalignant melanoma

S.N.WAGNER, C.WAGNER, L.BRIEDIGKEIT AND M.GOOSDepartment of Dermatology, School of Medicine, University of Essen, Hufelandstr. 55, D-45122 Essen, Germany

Accepted for publication 28 August 1997

Summary Chromosome 9p21 is frequently deleted in malignant melanoma, and one familial malignantmelanoma gene has been linked to 9p21–22. Recently, the cyclin D-dependent kinase inhibitors(CDKIs) p16INK4a and p15INK4b have been localized within chromosome 9p21, and the presence ofp16INK4a point mutations has been demonstrated in familial melanoma and melanoma cell lines invitro. To analyse the role of these CDKIs in sporadic human cutaneous non-metastatic malignantmelanoma, we examined 36 primary tumour specimens representing different stages of melanomaprogression and their corresponding normal skin samples for the three mechanisms of CDKIinactivation described so far. Homozygous codeletion of the p16INK4a and the p15INK4b gene wasdetected by Southern blot analysis in two tumour samples. By direct sequencing of polymerase chainreaction (PCR)-amplified microdissected genomic DNA, no somatic or germline p16INK4a pointmutations or small deletions were detected in the remaining 34 tumour samples; one individualexhibited the previously described germline codon 148 (Ala→Thr) polymorphism. In these tumourspecimens, comparative semiquantitative reverse PCR analysis of p16INK4a transcript levels revealedno evidence for repression of p16INK4a gene transcription and thus for p16INK4a promoter inactiva-tion by DNA methylation. These results indicate homozygous p16INK4a and p15INK4b loss to occur ina subset of cutaneous melanomas and suggest, in view of the frequent loss of heterozygosity onchromosome 9p21, the presence of another tumour suppressor gene within this chromosomalregion.

The cyclin D-dependent kinase inhibitors (CDKIs)p16INK4a and its relative p15INK4b negatively regulatecell growth by specifically inhibiting cyclin D-dependentkinase 4 and 6 complexes, whose activities enable cellsto enter the S phase by phosphorylation of their criticalsubstrate, the retinoblastoma-susceptibility tumoursuppressor protein (Rb). As inhibition of Rb phosphor-ylation results in growth arrest of cells at the G1 phase,p16INK4a and p15INK4b may act as tumour suppressorgenes, and their inactivation may contribute to uncon-trolled growth in human cancer (reviewed by Hiramaand Koeffler1). In line with this assumption, loss ofp16INK4a and p15INK4b expression has been demon-strated to be associated with extended proliferativepotential.2 Targeted deletion of the INK4a locus resultsin spontaneous tumour development at an early ageand in high sensitivity to carcinogenic treatments invivo,3 and the presence of a germline non-functionalp16INK4a allele has been reported to be associated with afamilial tumour syndrome.4 Reconstitution of their

respective expression by transfection into tumour cellscould revert the malignant phenotype by induction of aG1–S block, thereby suppressing uncontrolled cellgrowth and tumorigenicity of several tumour celltypes in vitro and in vivo, and forced expression ofCDKIs could block apoptosis during normal cell differ-entiation.5–10

The CDKN2A and CDKN2B genes, encoding p16INK4a

and p15INK4b, respectively, are located on chromosome9p21, a locus frequently deleted in many primaryhuman tumours,11,12 and which has previously beenlinked to at least 50% of all melanoma families analysedso far.13 In cutaneous malignant melanoma, cytoge-netic studies demonstrating frequent rearrangementsand deletions involving the short arm of chromosome 9(9p, reviewed by Dracopoli and Fountain13) and studieson tumour and cell line DNAs of sporadic malignantmelanoma implicating loss of heterozygosity (LOH) onchromosome 9p14 within region 9p21 (reviewed byDracopoli and Fountain13) imply chromosome region

British Journal of Dermatology 1998; 138: 13–21.

13q 1998 British Association of Dermatologists

9p21 to be a putative tumour suppressor gene locus.Furthermore, p16INK4a is located in a small commonregion of homozygous deletion in multiple melanomacell lines,11,12 and p16INK4a germline mutations havebeen described in a proportion of familial melanoma(reviewed by Dracopoli and Fountain13), indicating anassociation of mutant p16INK4a inheritance with melan-oma predisposition. Because of its 9p21 location, itsfrequent homozygous deletion in unrelated melanomacell lines, and its sequence similarity to p16INK4a,p15INK4b seems to represent another reasonablecandidate for melanoma predisposition.11

Several mechanisms of CDKI inactivation have beencharacterized so far. These initially included pointmutations, small deletions and large homozygous dele-tions, and many different types of tumour cell linescarrying such abnormalities in the CDKIs have beenreported.1,11 However, primary tumour tissue samplesincluding those with loss of 9p heterozygosity, i.e. non-small cell lung cancer, transitional cell carcinoma ofthe bladder, renal clear cell carcinoma, squamous cellcarcinoma of the head and neck, and gliomas havebeen reported to carry CDKI gene deletions or muta-tions only at a low frequency compared with those oftheir respective cell lines.1,15,16 Thus, it was assumedthat these alterations may represent cell culture arte-facts. Recently, another inactivating mechanism hasbeen demonstrated, DNA methylation of the p16INK4a

promoter region with subsequent repression oftissue-specific p16INK4a gene transcription.17,18

Thus, the high prevalence of p16INK4a and p15INK4b

gene deletions and mutations reported for melanomacell lines may not necessarily reflect a relevant mechan-ism in the development or progression of primaryhuman sporadic cutaneous melanoma in vivo. To deter-mine the role of CDKI inactivation in the pathogenesis ofprimary human sporadic cutaneous melanoma, weexamined these CDKIs in 36 tumour specimens repre-senting different stages of melanoma progression forhomozygous gene deletion, the presence of somatic orgermline gene mutations and small deletions, andtranscriptional silencing.

Materials and methods

Tissue specimens

Tissue samples obtained intraoperatively were halved,one part being quick frozen in liquid nitrogen, the otherpart being fixed in 10% buffered formalin (pH 7.0) andembedded in paraffin for histological examination.

Histological diagnosis and classification of all specimenswas performed independently by two of the authors.These data are summarized in Table 1. Melanomapatients with a family history of malignant melanomaand evidence of atypical mole phenotype were excludedfrom this study.

Direct sequencing of polymerase chain reaction products

Microdissection was performed essentially as describedby Wagner et al.19 to minimize surrounding non-lesional tissue material. Briefly, between one and fiveserial cryostat sections were cut from each specimendepending on the total area of the section processed.Histological examination, before and after microdissec-tion, confirmed the analysis of representative areas ofthe respective lesion. Only tissue sections with Clark’slevel and Breslow thickness corresponding to thoseobtained by routine histological analysis were includedin this study. Microdissected tumour tissue was trans-ferred into a reaction tube containing 100 mL distilledwater, heated at 98 8C for 7 min, and subsequentlycooled on ice.

Twenty microlitres of this DNA preparation weresubjected to PCR direct sequencing analysis. Two geno-mic regions of p16INK4a (exons 1 and 2) were amplifiedby use of PCR intronic primers. The PCR reaction mixconsisted of 1 × PCR buffer (Perkin Elmer Cetus, Nor-walk, CT, U.S.A.), 20 pmol primers, 100 mmol dNTPs, 5units Taq polymerase, and 5% dimethyl sulphoxide.Primers for exon 1 were 2F, GAAGAAAGAG-GAGGGGCTG (sense), and 1108R, GCGCTACCTGATTC-CAATTC (antisense), according to Kamb et al.11 PCRconditions for amplification of exon 1 consisted of a5 min denaturation at 94 8C, four cycles of 60 s at 94 8C,60 s at 64 8C, 60 s at 72 8C, followed by a decrease of the

14 S.N.WAGNER et al.

q 1998 British Association of Dermatologists, British Journal of Dermatology, 138, 13–21

Table 1. Tissue samples subjected to mutation analysis

Parameter n Parameter n Parameter n

Clark’s level Breslow thickness (mm) Typea

I 4 <0.75 9 in situ 4II 7 0.75–1.5 8 LMM 3III 6 1.51–3.0 8 SSM 14IV 9 >3.0–7.5 9 ALM 2V 8 Total 34 NM 11Total 34 Total 34

a LMM, lentigo maligna melanoma; SSM, superficial spreading melan-oma; ALM, acrolentiginous melanoma; NM, nodular melanoma.

annealing temperature by 2 8C every four cycles, and 20cycles of 60 s at 94 8C, 60 s at 54 8C, 60 s at 72 8C, then afinal elongation for 10 min at 72 8C. Primers for exon 2were 63F, CTCTACACACAAGCTTCCTTTCC (sense), and64R, CATCAGTCCTCACCTGAGG (antisense). PCR con-ditions for exon 2 consisted of a 5 min denaturation at94 8C, 40 cycles of 60 s at 94 8C, 60 s at 60 8C, 60 s at72 8C, and a final elongation of 10 min at 72 8C.

PCR products were run by agarose gel electrophoresison 4% NuSieve GTG (FMC, Rockland, ME, U.S.A.) andrecovered from agarose with glass powder (Mermaid Kit,Bio101, La Jolla, CA, U.S.A.). Purified PCR products weresequenced according to the dideoxy chain terminationmethod,20 following the Sequenase 2.0 protocol (USB,Cleveland, OH, U.S.A.) with PCR primers used as sequen-cing primers. Sequencing products were analysed on a8% denaturing polyacrylamide gel. This PCR directsequencing approach allows the analysis of 96.2% ofthe entire open reading frame of the p16INK4a gene.

Southern blot analysis

After removal of one to five serial cryostat sections, theremaining tissue of each specimen was dissected accord-ing to histological evaluation of a representative tissuesection until the content of contaminating lymphocytesand stromal tissue accounted for maximally 20% ofcells. The dissected tissue specimen was subjected tonucleic acid extraction by guanidinium thiocyanateextraction and caesium chloride centrifugation.21

Genomic DNA was isolated directly from the guanidi-nium thiocyanate cell lysate according to standardprotocols.22 Ten microgram aliquots of DNA weredigested with EcoRI, separated on 0.8% agarose gelsand transferred by alkaline blotting on to nylon mem-branes. Membranes were UV cross-linked and pre-hybridized in the presence of 5× saline sodium citratebuffer (SSC), 1% sodium dodecyl sulphate (SDS),5× Denhardt’s solution, 25 mmol/L sodium phosphatebuffer, 1 mg/mL denatured salmon DNA, and 50%formamide. DNA was probed by hybridization at 42 8Cwith [32P]-labelled random primed cDNA under pre-hybridization conditions. The probe used was the PCRproduct of exon 2 of the p16INK4a gene (see above forconditions). After hybridization, membranes werewashed for 40 min in 2× SSC, 0.1% SDS and twice for20 min in 0.3× SSC, 0.5% SDS at 65 8C and subse-quently exposed to Kodak X-AR film (Eastman KodakCo., Rochester, NY, U.S.A.) at – 70 8C with intensifyingscreens. Hybridization signals were quantified by scan-ning autoradiographs (Ultroscan XL, Pharmacia-LKB,

Freiburg, Germany). Tumour samples with signal inten-sities weaker than 20% as compared with referenceDNA obtained from matched normal tissue were scoredas having homozygous gene deletions.

Semiquantitative p16INK4a RT–PCR assay

mRNA was subjected to reverse transcription usingoligo (dT) primers and Moloney murine leukaemiavirus reverse transcriptase (Gibco-BRL, Gaithersburg,MD, U.S.A.). First-strand cDNA synthesis was carriedout for 1 h at 37 8C in a total volume of 20 mL. Twomicrolitres of first-strand cDNA were amplified by Taqpolymerase (Stratagene, La Jolla, CA, U.S.A.). The rela-tive expression of b-actin was used for standardizing thereaction. The following primers were used: AAGGA-GAAGCTGTGCTACGTCG (sense; positions 678–700)and ATCCACACGGAGTACTTGCG (antisense; positions1063–1043), yielding a b-actin cDNA product of386 bp in length. Based on densitometric scanning ofb-actin cDNA products, the amount of cDNA subjectedto RT–PCR analysis was adjusted for each tissue speci-men. p16INK4a PCR was performed with oligonucleotideprimers CACTCTCACCCGACCCGT (sense) and AGG-ACCTTCGGTGACTGATGATC (antisense) correspondingto positions 222–239 and 526–504, respectively, of thehuman p16INK4a cDNA sequence. The cDNA product of305 bp in length comprised sequences of exons 2 and 3,and the 30-untranslated region. Ten microlitre aliquotsof the 50 mL reaction were separated on a 2% agarosegel and the amplification products were visualizedunder UV light after staining the gel with ethidiumbromide.

Semiquantitative tyrosinase RT–PCR assay

cDNA synthesis and standardizing the reaction for b-actin expression were performed essentially as describedabove. Primers for tyrosinase cDNA amplification wereTGGAGGAGTACAACAGCCATC (sense) and TTGA-GAGGCATCCGCTATC (antisense). PCR conditions con-sisted of a 5 min denaturation at 94 8C, 20 cycles of 60 sat 94 8C, 40 s at 56 8C, 60 s at 72 8C, and a finalelongation of 10 min at 72 8C.

Controls for semiquantitative RT–PCR assays

For every primer pair, the PCR conditions were chosen togive linear and template-input dependent amplification.The oligonucleotide primer pairs and conditions used forPCR resulted in amplification of a single product of the

TUMOUR SUPPRESSOR GENE DELETIONS AND MALIGNANT MELANOMA 15


expected size. Thereby, serial dilutions of individualcDNAs resulted in a concomitant decrease in amplifiedproducts. An extensive series of controls was performedto ensure accurate comparison of the different samplesstudied. PCR analysis with serially diluted plasmidscontaining the respective cDNAs revealed a sensitivityof an approach to the order of 102–103 copies. Thequantity of PCR products correlated with the number ofstarting template copies to about 108 plasmid copies in alog-linear fashion, and serial dilutions of sample cDNAsrevealed a concomitant decrease in PCR products. Vary-ing the number of PCR cycles within the linear phase ofamplification did not change the relative differencesbetween the different samples, and mixing of cDNAsfrom different samples did not alter PCR amplification.Furthermore, titration experiments were performed todetermine the range of linear correlation between thedensity of signals and the amount of PCR product onethidium bromide-stained agarose gels. The amount ofloaded PCR product was adjusted accordingly for eachPCR. To distinguish amplification of contaminant geno-mic DNA, all primer pairs amplified sequences spanningthrough different exons, and the amplification of cDNAwas confirmed additionally by the absence of any PCRproduct when RNase-treated genomic DNA was used astemplate. Further controls included the omission ofreverse transcriptase and the replacement of cDNA bywater in the reverse transcription and the amplificationmixtures. Furthermore, cross-contamination wasavoided by the use of reagents in small aliquots, separatepipettes for PCR reagents and template cDNA, andseparate rooms for pre- and post-PCR handling.

Results

Semiquantitative tyrosinase RT–PCR assay

Tumour and matched normal skin tissue specimenswere analysed for tyrosinase mRNA levels by semiquan-titative RT–PCR analysis. Tyrosinase mRNA was detect-able by RT–PCR in every tissue sample. Semiquantitativeanalysis revealed that every tumour specimen includedin this study contained significantly higher tyrosinasetranscript levels than its corresponding normal skinsample (Fig. 1).

Southern blot analysis for homozygous p16INK4a andp15INK4b gene deletions

Among the 36 matched normal and tumour tissuesamples included in our study, 17 matched pairs repre-senting various histological subtypes (nine nodular

melanoma (NM) and eight superficial spreading melan-oma (SSM)) and different stages of tumour progression(Clark’s level III–V, Breslow thickness 0.75 to >3 mm)revealed enough genomic DNA to be subjected to South-ern blot analysis. As exon 1 of the p16INK4a gene hadbeen reported not always to be included into thehomozygously deleted area, we used an exon 2 probe.This probe also hybridizes with exon 2 of the p15INK4b

gene due to its high sequence homology to exon 2 of thep16INK4a gene. In 15 of 17 matched pairs, tumourtissues exhibited a comparable signal intensity withthat obtained in the corresponding normal tissue speci-men. Loss of signal indicative of homozygous deletion ofthe p16INK4a gene was found in two tumour tissuespecimens (NM, Breslow thickness 6.0 mm, Clark’slevel V; SSM, Breslow thickness 3.3 mm, Clark’s levelIV), whereas the corresponding normal tissue samplesexhibited no evidence for homozygous deletion ofp16INK4a (Fig. 2). Both tumour specimens also exhibitedhomozygous deletion of the p15INK4b gene. These resultsindicate homozygous codeletion of both the p16INK4a

and the p15INK4b genes in two of 17 melanoma tissuespecimens. No rearranged fragments were observedafter EcoRI-digestion in any of the samples tested.

Analysis of somatic and germline p16INK4a (MTS1) genemutations

In 34 of 36 microdissected tumour samples, exons 1and 2 of the p16INK4a gene could be amplified by PCRand subsequently subjected to dideoxy sequencing. Inboth tumour specimens exhibiting homozygous deletionof the p16INK4a and the p15INK4b gene, no specific PCRproducts could be amplified. All tumour specimensanalysed contained no sequence alterations in exon 1,



Figure 1. Tyrosinase expression profile in paired normal skin tissuespecimens (n) and human cutaneous non-metastatic malignantmelanoma samples (t) as obtained by semiquantitative tyrosinaseRT–PCR (upper panel). Note the preferential tyrosinase transcriptexpression in tumour samples. The amount of cDNA template sub-jected to PCR was equilibrated for b-actin expression for each tissuespecimen (lower panel).

including point mutations or short deletions that havebeen described previously in a several melanoma celllines in vitro.11,23 As expected from these results, noexon 1 sequence alterations were detected in the corre-sponding normal skin tissue specimens. When exon 2was analysed, one tumour sample (no. 25) of 34exhibited the presence of a heterozygous G→A substitu-tion at nt 460 (Fig. 3). This G→A transition results in asubstitution of alanine with threonine at codon 148 ofthe p16INK4a gene. This kind of nucleic acid substitutionhas been described in a variety of different tumoursamples in vivo and in vitro (previously designated ascodon 140) and is considered most likely to represent apolymorphism rather than a point mutation of thisgene.15,24 This heterozygous G→A transition was alsopresent in the corresponding normal tissue specimenindicating a germline codon 148 (Ala→Thr) poly-morphism in this individual (Fig. 3). No exon 2sequence alterations were detected in the remaining33 corresponding normal skin tissue specimens.

Sensitivity of the p16INK4a (MTS1) PCR direct sequencingapproach

Contaminating normal tissue and/or infiltrating mono-nuclear cells that were not removed by microdissectionmay diminish the detection of tumour gene mutations

in tissue samples. To determine the sensitivity of oursequencing approach, we performed serial dilutionexperiments using codon 148-Ala (wild type) tumourDNA and heterozygous codon 148-Ala/Thr (normalskin specimen no. 25, Fig. 3) DNA. In a mixture of80% wild type DNA with 20% of the heterozygouscodon 148-Ala/Thr polymorphic DNA, the G→A tran-sition at codon 148 was still detectable (Fig. 4). Theseresults indicate that 10% of homozygously mutated/polymorphic DNA in our tumour samples would havebeen detected by our PCR direct sequencing approach—at least at defined positions.

Semiquantitative RT–PCR assay of p16INK4a (MTS1)mRNA expression

Decreased or totally lost expression of p16INK4a mRNAmay indicate transcriptional silencing by DNAmethylation of the p16INK4a promoter region17,18 orallelic loss or homozygous deletion of the p16INK4a genein tumour tissue samples.12,25 Thus, all tumour tissuespecimens were analysed for p16INK4a mRNA expres-sion by semiquantitative RT–PCR analysis, and mRNAlevels were compared with those detected in the corre-sponding normal skin tissue specimens. Semiquantita-tive analysis revealed the presence of p16INK4a

transcripts in 34 of the 36 tumour tissue samplesanalysed at levels comparable with those observed incorresponding normal skin (Fig. 5). Furthermore, sig-nificant differences in p16INK4a transcript levels



Figure 2. Example of Southern blot analysis of EcoRI-digested corre-sponding normal and tumour DNA samples from sporadic cutaneousmelanoma patients. A labelled exon 2 probe of p16INK4a was generatedby PCR and used for hybridizing (see Materials and methods). TwoEcoRI fragments were detected. The upper fragment represents exon 2of the p15INK4b gene, the lower fragment exon 2 of p16INK4a. Upper leftpanel: lanes 1 and 2: control DNA (placenta, normal peripheral bloodmononuclear cells). Note the homozygous loss of exon 2 of thep16INK4a and the p15INK4b genes in lanes 3 and 5 representingmelanoma samples, but not in lanes 4 and 6 representing therespective corresponding normal skin specimens. Upper right panel:representative pairs of tumour (lanes 1, 3, 5 and 7) and correspondingnormal skin (lanes 2, 4, 6 and 8) samples. Molecular sizes are given inkb. The lanes are numbered from left to right in both panels. Lowerpanels: respective ethidium bromide-stained agarose gels for compar-ison of DNA loading.

Figure 3. Codon 148-Ala/Thr polymorphism in individual no. 25.Genomic DNA sequence analyses of exon 2 of the p16INK4a gene of anormal skin tissue sample, of melanoma sample no. 25, and itscorresponding normal skin specimen are shown (from left to right).Codon 148 is indicated. The normal cell sample has a wild-typesequence; note the heterozygous G →A transition in the tumoursample and in the corresponding normal skin tissue sample. Antisensesequences are shown by direct sequencing with primer 64R.

between different tumour stages were not observed.Despite the presence of b-actin mRNA, no p16INK4a

transcripts could be detected in the two tissue samplesexhibiting homozygous deletion of the p16INK4a and thep15INK4b genes.

Discussion

We have investigated a series of primary melanomatissue specimens for CDKI alterations occurring invivo. The tissue samples analysed represented sporadiccutaneous malignant melanoma samples from almostall clinically and histopathologically defined stages ofdisease progression. These included initial stages asmalignant melanoma in situ and the radial growthphase melanoma with low capacity for metastasis, aswell as stages as advanced as vertical growth phasemelanoma with high capacity for metastasis. Thus,tumour suppressor genes that may play a part eitherin the development, the progression, or the acquisitionof metastatic capacity of malignant melanoma shouldexhibit alterations in the tissue specimens selected forthis study.

Three mechanisms of CDKI inactivation have been

described so far. These include large homozygous dele-tions, point mutations and small deletions, and DNAmethylation of the promoter region. The prevalence ofthe various types of CDKI inactivation seems to varywith the type of tumour. Large homozygous deletionsare common in non-small cell lung cancers, malignantgliomas, renal cell carcinomas, head and neck tumours,prostate tumours, bladder carcinomas, and acute lym-phoblastic leukaemias, point mutations and small dele-tions are common in pancreatic adenocarcinomas,oesophageal carcinomas, biliary tract cancers, and infamilies with susceptibility to melanoma and pancreaticcarcinoma (reviewed by Hirama and Koeffler1 and



Figure 4. Sensitivity of the PCR-direct sequencing approach. Serialdilution experiments using nt 460 wild-type tumour DNA and hetero-zygous DNA (normal skin specimen no. 25, Fig. 2) exhibiting an nt460 G →A transition. Left, DNA mixture of 50% wild-type DNA with50% heterozygously mutated DNA. Right, detection of the G →Atransition in a mixture of 80% wild-type DNA with 20% heterozy-gously mutated DNA, indicating a detection level of at least 10% ofhomozygously mutated DNA.

Figure 5. RT–PCR analysis reveals comparable amounts of p16INK4a

transcripts in tissue specimens of sporadic cutaneous non-metastaticmalignant melanoma (t) and normal skin (n). Quantification ofp16INK4a cDNA was performed by standardizing for equal amountsof b-actin cDNA (bottom panel). p16INK4a cDNA amplification pro-ducts obtained after 25, 35 and 45 PCR cycles are shown. Notecomparable amounts of cDNAs from five representative pairs ofsamples from tumour tissue and corresponding normal tissue duringlinear amplification of PCR products. c, negative control.

Pollock et al.26), and methylation of the p16INK4a pro-moter region is common in breast and colon cancer,nasopharyngeal carcinomas, and gliomas.18,27–29

In the present study, homozygous deletion of thep16INK4a gene was detectable in two primary humansporadic cutaneous non-metastatic malignant melan-oma samples. Both cases represented sporadic malig-nant melanoma of the vertical growth phase with highinvasion levels but without signs of metastasis within2 years after primary diagnosis. This result comple-ments previous reports on a low frequency of homo-zygous deletion of the p16INK4a gene in metastaticmalignant melanoma.30–33 For p15INK4b, the predomi-nant mechanism of inactivation in melanoma cell linesand in metastatic malignant melanoma has beenreported to constitute homozygous gene deletion.6,30

Our results indicate the presence of homozygousp15INK4b deletion in a subset of primary human spora-dic cutaneous non-metastatic malignant melanomasamples. Homozygous deletion could be observed inthe two cases also exhibiting homozygous p16INK4a

gene deletion, indicating codeletion of both CDKIs, aphenomenon already described in or deduced fromstudies in melanoma cell lines.11,34 The low prevalenceof these alterations in vivo, however, did not enableassociation of this gene inactivation with prognosis.

Whereas homozygous deletion of CDKIs p16INK4a andp15INK4b in sporadic malignant melanoma obviouslyrepresents a rare event, loss of heterozygosity at 9p hasbeen reported at a high frequency.14 These results andthe presence of p16INK4a germline mutations in familialmelanoma (reviewed by Dracopoli and Fountain13)underline the potential role of inactivating point muta-tions and small deletions of tumour suppressor genes inthe tumorigenesis of malignant melanoma. In ourstudy, neither somatic nor germline p16INK4a genemutations could be detected in primary sporadic malig-nant melanoma. These results are not due to possibletissue sampling artefacts, as demonstrated by semi-quantitative tyrosinase RT–PCR, or to a low sensitivityof the chosen PCR direct sequencing approach. Thus,our study complements previous reports of a low fre-quency of p16INK4a gene mutations in metastatic malig-nant melanoma,23,30–32 and supports the results ofHealy et al. obtained from paraffin-embedded primarytumours.35 Taken together, these results indicate that incontrast to reports on cultured melanoma cells,11,33

primary sporadic malignant melanoma cells containp16INK4a gene mutations only at low frequency invivo. From studies on melanoma cell lines, thep16INK4a tumour suppressor gene has been suggested

to be a target of UV radiation-induced DNA mutagen-esis.36 Even if CC→TT mutations may sporadically bepresent in primary malignant melanoma,35 the resultsof our study showing a lack of p16INK4a gene mutationsin melanoma samples from constantly sun-exposedbody sites and in the lentigo maligna melanoma sub-type, which is associated with long-term constant UVexposure, do not support UV radiation-inducedp16INK4a gene mutations as a relevant pathogeneticfactor in malignant melanoma in vivo.

Recently, another inactivating mechanism ofp16INK4a has been demonstrated: methylation of thep16INK4a gene locus.17,18,27–29 The locus encodingp16INK4a, named INK4a, has the capacity to generatetwo distinct transcripts from different promoters. Eachtranscript has a specific 50 exon, E1a or E1b, which arespliced into common exon E2. The E1a-containingtranscript encodes p16INK4a, and the E1b transcriptencodes p19ARF from a different reading frame(reviewed by Larsen37). Whereas methylation atINK4a results in silencing of the p16INK4a promoterand subsequent repression of p16INK4a gene transcrip-tion,17,18,27 the p19ARF promoter remains unaf-fected.5,38 Thus, analysis of p16INK4a transcript levelsmay give relevant information on p16INK4a inactivationby DNA methylation. Even if offering no truly quanti-tative values, a semiquantitative reverse PCR approachis sufficient enough for performing relative comparisonsof gene expressions between tissue samples. None of theprimary malignant melanoma specimens withouthomozygous p16INK4a gene deletion contained signifi-cantly decreased or lost expression of p16INK4a tran-scripts as compared with its corresponding normal skintissue specimen. Furthermore, no significant differencesin p16INK4a mRNA expression could be observedbetween different tumour stages. Thus, our resultsobtained in primary human cutaneous malignant mel-anomas provide no evidence for repression of p16INK4a

gene transcription that may be indicative of p16INK4a

promoter inactivation by DNA methylation.From these results, we conclude that homozygous

p16INK4a and p15INK4b gene loss may be involved in thedevelopment of a subset of sporadic cutaneous non-metastatic malignant melanoma tissues in vivo. How-ever, genetic alterations of these CDKIs seem not to beinvolved in the development or progression of themajority of sporadic cutaneous non-metastatic malig-nant melanoma cells in vivo. Cytogenetic studies, LOHanalyses, and the assignment of a familial cutaneousmalignant melanoma gene to 9p21–22 strongly implychromosome region 9p21 as a putative tumour



suppressor gene locus. Evidence from our study onsporadic cutaneous non-metastatic melanoma andfrom studies on metastatic malignant melanoma invivo support the suggestion that homozygous p16INK4a

and p15INK4b loss may occur in a subset of cutaneousmelanoma samples. These results also argue for thepresence of another tumour suppressor gene withinchromosomal region 9p21 that may be involved in thedevelopment and progression at least of the majority ofsporadic cutaneous malignant melanomas.

Acknowledgments

We are grateful to Dr D.Lohmann and Dr B.Opalka(University of Essen) for helpful discussion and toH.Apel for photographic work.

References1 Hirama T, Koeffler HP. Role of cyclin-dependent kinase inhibitors

in the development of cancer. Blood 1995; 86: 841–54.2 Noble JR, Rogan EM, Neumann AA et al. Association of extended

in vitro proliferative potential with loss of p16INK4 expression.Oncogene 1996; 13: 1259–68.

3 Serrano M, Lee H, Chin L et al. Role of the INK4a locus in tumorsuppression and cell mortality. Cell 1996; 85: 27–37.

4 Yarbrough WG, Aprelikova O, Pei H et al. Familial tumor syndromeassociated with a germline nonfunctional p16INK4a allele. J NatlCancer Inst 1996; 88: 1489–91.

5 Shapiro GI, Park JE, Edwards CD et al. Multiple mechanisms ofp16INK4A inactivation in non-small cell lung cancer cell lines.Cancer Res 1995; 55: 6200–9.

6 Stone S, Dayananth P, Jiang P et al. Genomic structure, expression,and mutational analysis of the p15 (MTS2) gene. Oncogene 1995;11: 987–91.

7 Tenan M, Benedetti S, Finocchiaro G. Deletion and transfectionanalysis of the p15/MTS2 gene in malignant gliomas. BiochemBiophys Res Commun 1995; 217: 195–202.

8 Liggett WH Jr, Sewell DA, Rocco J et al. p16 and p16 beta arepotent growth suppressors of head and neck squamous carcinomacells in vitro. Cancer Res 1996; 56: 4119–23.

9 Spillare EA, Okamoto A, Hagiwara K et al. Suppression of growthin vitro and tumorigenicity in vivo of human carcinoma cell linesby transfected p16INK4. Mol Carcinog 1996; 16: 53–60.

10 Wang J, Walsh K. Resistance to apoptosis conferred by Cdkinhibitors during myocyte differentiation. Science 1996; 273:359–61.

11 Kamb A, Gruis NA, Weaver-Feldhaus J et al. A cell cycle regulatorpotentially involved in genesis of many tumor types. Science 1994;264: 436–40.

12 Nobori T, Miura K, Wu DJ et al. Deletions of the cyclin-dependentkinase-4 inhibitor gene in multiple human cancers. Nature 1994;368: 753–6.

13 Dracopoli NC, Fountain JW. CDKN2 mutations in melanoma.Cancer Surv 1996; 26: 115–32.

14 Healy E, Belgaid CE, Takata M et al. Allelotypes of primarycutaneous melanoma and benign melanocytic nevi. Cancer Res1996; 56: 589–93.

15 Cairns P, Mao L, Merlo A et al. Rates of p16 (MTS1) mutations inprimary tumors with 9p loss. Science 1994; 265: 415–16.

16 Spruck CH, Gonzalez-Zulueta M, Shibata A et al. p16 gene inuncultured tumours. Nature 1994; 370: 183–4.

17 Merlo A, Herman JG, Mao L et al. 50 CpG island methylationis associated with transcriptional silencing of the tumor suppres-sor p16/CDKN2/MTS1 in human cancers. Nat Med 1995; 1:686–92.

18 Gonzalez-Zulueta M, Bender CM, Yang AS et al. Methylation of the50 CpG island of the p16/CDKN2 tumor suppressor gene in normaland transformed human tissues correlates with gene silencing.Cancer Res 1995; 55: 4531–5.

19 Wagner SN, Ockenfels HM, Wagner C et al. Ras gene mutations: arare event in nonmetastatic primary malignant melanoma. JInvest Dermatol 1995; 104: 868–71.

20 Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sci USA 1977; 74: 5463–7.

21 Chirgwin JM, Przybyla AE, McDonald RJ, Rutter WJ. Isolation ofbiologically active RNA from sources enriched in ribonuclease.Biochemistry 1979; 18: 5294–7.

22 Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning. A LaboratoryManual. Cold Spring Harbor: Cold Spring Harbor Laboratory Press,1989.

23 Ohta M, Nagai H, Shimizu M et al. Rarity of somatic and germlinemutations of the cyclin-dependent kinase 4 inhibitor gene, CDK4I,in melanoma. Cancer Res 1994; 54: 5269–72.

24 Sun Y, Hildesheim A, Lanier AEP et al. No point mutation butdecreased expression of the p16/MTS1 tumor suppressor gene innasopharyngeal carcinomas. Oncogene 1995; 10: 785–8.

25 Okamoto A, Demetrick DJ, Spillare EA et al. Mutations and alteredexpression of p16ink4 in human cancer. Proc Natl Acad Sci USA1994; 91: 11045–9.

26 Pollock PM, Pearson JV, Hayward NK. Compilation of somaticmutations of the CDKN2 gene in human cancers: non-randomdistribution of base substitutions. Genes Chromosom Cancer 1996;15: 77–88.

27 Herman JG, Merlo A, Mao L et al. Abnormal DNA methylationfrequently inactivates the putative tumor suppressor CDKN2/p16in many tumor types. Proc Am Assoc Cancer Res 1995; 36: 201–12.

28 Costello JF, Berger MS, Huang HS, Cavenee WK. Silencing of p16/CDKN2 expression in human gliomas by methylation and chro-matin condensation. Cancer Res 1996; 56: 2405–10.

29 Lo KW, Cheung ST, Leung SF et al. Hypermethylation of the p16gene in nasopharyngeal carcinoma. Cancer Res 1996; 56: 2721–5.

30 Glendening JM, Flores JF, Walker GJ et al. Homozygous loss of thep15/INK4B gene (and not of the p16/INK4A gene) during tumorprogression in a sporadic melanoma patient. Cancer Res 1995; 55:5531–5.

31 Maelandsmo GM, Florenes VA, Hovig E et al. Involvement of thepRb/p16/cdk4/cyclin D1 pathway in the tumorigenesis ofsporadic malignant melanoma. Br J Cancer 1996; 73: 909–16.

32 Platz A, Sevigny P, Norberg T et al. Genes involved in cell cycle G1checkpoint control are frequently mutated in human melanomametastases. Br J Cancer 1996; 74: 936–41.

33 Wang Y, Becker D. Differential expression of the cyclin-dependentkinase inhibitors p16 and p21 in the human melanocytic system.Oncogene 1996; 12: 1069–75.

34 Robinson WA, Elefanty AG, Hersey P. Expression of the tumoursuppressor genes p15 and p16 in malignant melanoma. MelanomaRes 1996; 6: 285–9.



35 Healy E, Sikkink S, Rees JL. Infrequent mutation of p16/INK4in sporadic melanoma. J Invest Dermatol 1996; 107: 318–21.

36 Pollock PM, Yu F, Qiu L et al. Evidence for u.v. induction ofCDKN2 mutations in melanoma cell lines. Oncogene 1995; 11:663–8.

37 Larsen C-J. p16/INK4a: a gene with a dual capacity to encodeunrelated proteins that inhibit cell cycle progression. Oncogene1996; 12: 2041–4.

38 Mao L, Merlo A, Bedi G et al. A novel p16/INK4a transcript. CancerRes 1995; 55: 2995–7.



Homozygous deletion of the p16INK4a and the p15INK4b tumour suppressor genes in a subset of human...

Documents

Transcript of Homozygous deletion of the p16INK4a and the p15INK4b tumour suppressor genes in a subset of human...