Oncogene (1997) 14, 369 ± 373  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Deletion mapping of chromosome 4 in head and neck squamous cell carcinoma

Mark A Pershouse1,3, Adel K El-Naggar2, Kenneth Hurr2, Huai Lin1,3, WK Alfred Yung1,3 and Peter A Steck1,3

Departments of 1Neuro-Oncology and 2Pathology and 3The Brain Tumor Center, The University of Texas MD Anderson Cancer Center, Houston, Texas 77030, USA

Genomic deletions involving chromosome 4 have recently Cytogenetic studies have identi®ed recurring, but been implicated in several human cancers. To identify widely varied alterations of chromosomes 1, 3, 4, 5, 7, and characterize genetic events associated with the 8, 9, 11, 14, 15 and 17 (Jin et al., 1993; Sreekantaiah et development of head and neck squamous cell carinoma al., 1994). Although several molecular studies have (HNSCC), a ®ne mapping of allelic losses associated shown that mutation of p53 and ampli®cation of with chromosome 4 was performed on DNA isolated epidermal growth factor receptor are relatively from 27 matched primary tumor specimens and normal common events. However, the exact genes that are tissues. Loss of heterozygosity (LOH) of at least one targeted in the majority of the observed chromosomal chromosome 4 polymorphic allele was seen in the alterations are unknown (Brachman et al., 1992; majority of tumors (92%). Allelic deletions were con®ned Grandis et al., 1993; Shin et al., 1994). Recently, to short arm loci in four tumors and to the long arm loci several groups have performed allelotyping studies on in 12 tumors, suggesting the presence of two regions of HNSCC specimens to further de®ne regions of common deletion. One region of frequent deletion was consistent loss. Frequent loss of heterozygosity centered at D4S405 on 4p and included the loci D4S1546 (LOH) was observed on chromosome arms 3p, 5q, to D4S428 in *41% of the tumors. The common region 9p, 11q and 17p, with less frequent losses on 4, 6p, 8, of deletion on 4q was more complex and extended from 14q, 18 and 19q (Nawroz et al., 1994; Ah-See et al., D4S1571 to D4S1573. Frequent genetic alterations were 1994; El-Naggar et al., 1995). The combination of these observed within this region (4q25) and one marker, observations suggest that multiple molecular altera- D4S407, exhibited a high frequency of LOH (475%). tions are responsible for the oncogenesis of HNSCC. These results indicate that alterations of chromosome 4 Our interest in chromosome 4 results from an regions are associated with HNSCC tumorigenesis and observation that the microcell-mediated chromosomal further localizes the regions that may harbor tumor transfer of chromosome 4 in human gliomas resulted in suppressor genes. the suppression of the tumorigenicity of glioma cells (Pershouse et al., submitted). The putative tumor Keywords: chromosome 4; loss of heterozygosity; suppressor locus associated with this observed pheno- carcinoma typic suppression was further localized to a region (47 cM) near the epidermal growth factor gene on 4q. This study was undertaken to further examine the occurrence and location of regions of common deletion Introduction involving chromosome 4, particularly associated with the long arm in HNSCC. Approximately 50 000 North Americans will be diagnosed with head and neck squamous cell carcinoma (HNSCC) this year and over 14 000 will Results succumb to the disease (DeVesa et al., 1995). Etiological factors for HNSCC include tobacco and Tumor specimens from 27 patients with various grades alcohol use, along with occupational exposure to of HNSCC were screened for LOH with a panel of 27 carcinogens, implying that damage to DNA plays a microsatellite markers speci®c for chromosome 4 loci. major role in the oncogenesis of this cancer (Vokes et Typical autoradiographs of the markers D4S408 and al., 1993). This is further supported by studies D4S1546 illustrating LOH and retention of alleles are demonstrating mutagen-induced chromosomal fragi- shown in Figure 1. The overall frequency of deletion at lity as an independent risk factor (Spitz et al., 1993). each locus and their ordering is depicted in Figure 2. However, the molecular events in HNSCC are still Two tumors showed no loss of chromosome 4 loci poorly understood (Hong et al., 1995; Sidransky, (tumors #36 and 48). Alternatively, two other tumors 1995). By better de®ning the molecular genetic basis (#4 and 8) displayed extensive losses, suggesting the of tumor initiation and progression in HNSCC, new deletion of all or most of chromosome 4 in these insights into the biological behavior, diagnosis, and neoplasms. Allelic alterations (alleles not present in the possible new therapeutic approaches may be forth- matched normal specimens) were observed in six of 27 coming. specimens (22%), with three tumors (#41, 44, and 46) showing alterations at several markers. Correspondence: PA Steck The majority of tumors (22 of 27, 81%) had Received 28 June 1996; revised 17 September 1996; accepted 17 deletions and/or alterations associated with the long September 1996 arm of chromosome 4. Twelve tumor specimens Chromosome 4 loss in HNSCC MA Pershouse et al 370

6 7 8 N T N T N T

D4S1546

41 42 46 N T N T N T

D4S408 Figure 1 Illustration of microsatellite polymorphism analysis in head and neck squamous cell carcinoma specimens. Paired normal lymphocyte (N) and tumor (T) DNA samples for patients 6, 7 and 8 using microsatellite markers D4S1546 demonstrating non-informative, retention of heterozygosity, and loss of the upper allele, respectively. For patients 41, 42 and 43 D4S408 marker shows retention of heterozygosity, non-informa- tive, and loss of the lower allele, respectively

revealed deletions con®ned only to 4q (Table 1; Figure 3, groups A and B). All of the tumors with deletions on 4q involved at least an area at 4q25 from the markers D4S1571 to D4S1573 which represents a region of about 7 cM (Gyapay et al., 1994). Two patterns of allelic losses within this region were de®ned. One series of tumor specimens exhibited a continuous region of possible deletions within this critical area at 4q25 (boxed markers Figure 3a). A common region of loss was centered at D4S1616 and involved the Figure 2 Illustration of the frequency and approximate location of LOH at loci on chromosome 4 in HNSCC and several other adjacent markers D4S1553 to D4S407 (*1.5 mb). human cancers. Genetic markers are listed in their most probable Breakpoints involving LOH and retention of alleles in order and the approximate physical map of several markers are tumors #6 and 41 would indicate a consistent region of shown deletion surrounding D4S1616. A similar pattern of continuous deletions was de®ned for tumors with allelic losses involving both the long and short arm of the retention. A similar pattern was seen for tumors with chromosome (Figure 3, group C1). However, for this and without allelic losses on the p arm (Figure 3, series of tumors, only the region surrounding D4S407 groups C2 and B, respectively). The most centromeric was consistently involved. The combination of these region of consistent deletion was centered at D4S1553, two series of tumors would suggest a minimal region of while the more telomeric region was centered at involvement between D4S1616 and D4S407, as D4S1616 and extended from D4S1651 to D4S1611. illustrated by breakpoints in tumors (#6 and 43). Four tumors (#5, #43, #44, and #45) displayed losses Two tumors (#11 and 53) revealed continuous deletion associated only with the short arm of the chromosome of the area D4S1616-D4S407, but had an additional (Table 1). Overall, LOH on 4p was shown in 11 of the 27 area of deletion telomeric to 4q25. (41%) of the tumors examined. Of these, four (14%) had A second more complex pattern of allelic losses for potentially lost the entire arm (Figure 4). A minimum 4q25 was also observed involving the same region. area of deletion involved the markers D4S405 and However, this series of tumors displayed two areas of D4S428. Three tumors (#44, 45, and 52) have single p- deletions accompanied by two markers within this arm, deletions at D4S405. One tumor (#53) displayed a region, EGF and D4S193, that showed frequent deletion towards the centromeric side suggesting a Chromosome 4 loss in HNSCC MA Pershouse et al 371 Table 1 Diagnosis, grading and site of HNSCC tissue samples mas including hepatocellular, bladder, ovarian and Tumor Allelic cervical (Buetow et al., 1989; Sato et al., 1991; Mitra et no. Gradea Stageb Site deletionc Diagnosisd al., 1994; Fujimoto et al., 1994; Polascik et al., 1995). 2 2 IV Tongue 4q MDSqCa For HNSCC the allelic deletions associated with 3 2 III Tongue 4q PDSqCa chromosome 4 were considered relatively infrequent 4 2 IV Tongue 4p+qe InvSqCa compared to other chromosomal alterations and 5 2 II H & Neck 4p MDMetSqCa `unexpected' as previous reports failed to implicate 6 2 IV Larynx 4q MDSqCa 7 3 IV Tongue 4q PDSqCa this chromosome (Narwoz et al., 1994). Although only 8 2 IV Maxilla 4p+qe InvSqCa one marker per arm was used in the latter study. The 9 2 III Tongue 4q MDSqCa results from our study illustrate a higher frequency of 10 3 IV Tongue 4q PDSqCa loss associated with chromosome 4 with 92% of the 11 2 IV Tongue 4q MDSqCa 13 1 IV Tongue 4p+q WDSqCa specimens showing allelic deletion involving chromo- 36 2 IV Larynx ±f MDSqCa some 4. Furthermore, two regions of common deletion 37 3 IV Neck 4q PDSqCa were observed: (1) a large area involving the 41 2 IV Tongue 4q MDSqCa centromeric region of the short arm centered at 42 2 III Tongue 4q MDSqCa D4S405 and (2) a smaller region showing complex 43 2 II Larynx 4p+q MDSqCa 44 1 I Tongue 4p+q InvSqCa alterations that involved at least 4q25. Alterations to 45 2 IV Tongue 4p MDMetSqCa to LN each region appear to be independent as individual 46 3 III Tongue 4p+q MDMetSqCa to LN tumors displayed only p- or q-arm deletions. 47 1 III Mandible 4q WDSqCa Functional evidence for tumor suppressor genes on 48 2 IV Tongue ±f InvSqCa 49 2 II Larynx 4p+q MDSqCa chromosome 4 has been provided by several studies. 50 3 IV Larynx 4p+q PDSqCa Phenotypic tumor suppression has been observed with 52 2 III Neck 4p+q MDSqCa introduction of chromosome 4 into teratocarcinoma 53 3 I Tongue 4p+q PDSqCa cells (McGowan-Jordan et al., 1994). The presence of a 55 3 III Neck 4p+q PDSqCa senescence gene on chromosome 4 was also shown by 56 3 III Neck 4p+q Anaplastic Thyroid Ca a b chromosomal transfer (Ning et al., 1991). Recently, the Histological grade of tumor. Clinical stage of tumor, including transfer of an intact copy of chromosome 4 or a size, grade, and metastasis status. c Chromosomal location of observed allelic deletions on chromosome 4 arms. d The diagnosis fragment of the chromosome retaining the region of the HNSCC specimens are MDSqCa, moderately di€erentiated surrounding EGF gene in human glioma cells was squamous carcinoma; PDSqCa, poorly di€erentiated squamous demonstrated to suppress the tumorigenic phenotype of carcinoma, InvSqCa, invasive squamous carcinoma; MDMetSqCa, the parental cells (Pershouse et al., submitted). This moderately di€erentiated metastatic squamous carcinoma; WDSqCa, well di€erentiated squamous cell carcinoma; LN, metastatic to lymph latter study focused our e€orts at de®ning deletions node. eAll or most of chromosome 4 deleted. fNo observed allelic associated with 4q in additional human cancers. deletions associated with chromosome 4 The highest frequency of deletion in our study was found at 4q25 surrounding the EGF locus. This locus was previously implicated in an allelic deletion study of hepatocellular carcinomas (Buetow et al., 1989). Also, an potential critical region between D4S405 and D4S428, investigation of cervical carcinomas showed the region although this tumor displayed multiple regions of 4q23-25 displayed frequent LOH (Mitra et al., 1994). deletion, indicating a complex cytogenetic rearrange- However, in our localization analysis of LOH directed at ment. An additional tumor (#43) revealed a deletion only 4q, two patterns of deletion were observed in the region at the short arm telomeric marker D4S412. 4q23-25. The independent deletion patterns do not Comparison of the allelic deletions on either the appear to be due to artifactual factors as the incorrect short or long arm of chromosome 4 to tumor grade, ordering of the allelic markers was excluded from YAC location or histological di€erentiation suggested no contig data and identical results were obtained from signi®cant correlations. Furthermore, we failed to duplicate experiments. This raises several possible observe any relationship between the number of alternative explanations. First, the observed retention deletions per specimen and the histological differentia- of heterozygosity within the region EGF to D4S193 tion. Since the allelic deletions were found in all grades could be due to ampli®cation of alleles in contaminating and stages of tumors, this would suggest that deletion normal cells in the sections, thus masking an area of involving chromosome 4 may represent an early event homozygous deletion (Cairns et al., 1995). This in HNSCC oncogenesis. possibility is well illustrated by tumor #46 where the vast majority of allelic markers of chromosome 4 exhibit LOH except for the area EGF to D4S407. Unfortu- Discussion nately, sucient DNA from the microdissected speci- mens was not available to test this possibility by The functional loss of tumor suppressor genes is Southern blotting. An alternative possibility is that the intimately associated with the initiation and/or region (EGF ± D4S193) may contain a critical gene(s) progression of human cancers (Weinberg, 1991). LOH that would have a negative impact on cell viability, even in HNSCC has been described for a number of if only one copy of the gene is lost, thus the area is chromosomes (Nawroz et al., 1994; Ah-See et al., preferentially retained. The retention of these loci in our 1994), however, ®ne mapping of common deletions has suppressed somatic cell hybrids containing fragments of mainly focused on chromosomes 3p, 11p, 13q and 17p chromosome 4 would favor this possibility (Pershouse et exhibiting frequent deletions (450%). The involvement al., submitted). A third possibility is that the region of LOH on chromosome 4 alleles has only recently D4S423 to D4S400 could represent a third area of been observed in HNSCC and several other carcino- deletion in HNSCC in a subset of the tumors. Support Chromosome 4 loss in HNSCC MA Pershouse et al 372

Figure 3 Deletion and alteration map of chromosome 4q for head and neck squamous cell carcinoma. Regions of possible loss are denoted by the boxed areas. The four groups of tumors represent those that have deletions only associated with 4q and are continuous (a), non-continuous within D4S1571 to D4S1573 (b); or tumors that display deletions on both arms and are continuous (C1) or non continuous (C2). The diagnosis of the tumors is presented in Table 1

the resulting genetic recombinations within the region may e€ect a distal tumor suppressor locus. This is supported by the observation that IF and EGF genes are genetically 3 cM apart, yet physically map to the same YAC 416e1 at 330 kb (Fan et al., 1994), suggesting frequent recombination within the region. Distinguish- ing among these various possibilities will require further extensive analyses. The observed frequency of LOH on the p-arm in this study is similar to that previously reported for HNSCC with approximately 40% of the specimens exhibiting losses (Nawroz et al., 1994). A recent report in bladder carcinomas has identi®ed two critical regions of LOH on chromosome 4 (Polascik et al., 1995); one region maps to the p-arm between markers D4S1608 and D4S404. This area is smaller and more telomeric than the region of most frequent deletion Figure 4 Summary of allelic loss at 8 loci on 4p in 27 HNSCC observed in our specimens, which is centered at tumors. Tumors correspond to those in Table 1. Boxed areas represent regions of potential loss, losses in conjunction with D4S405. However, additional ®ne mapping of these adjacent non-informative markers regions would be required to determine if they overlap. The other region of common deletion in bladder carcinoma mapped near the telomere on the long for this possibility is that the common regions of deletion arm of chromosome 4 between markers D4S408 and de®ned by groups A and C1 and B and C2 do not D4S426. In our cohort, infrequent deletions were overlap. A fourth possibility is that the region is observed in this region in our HNSCC specimens, genetically unstable and retention or deletion of the although three tumors (#11, #46 and #53) may harbor loci surrounding EGF and D4S193 may both occur, but independent chromosomal aberrations in this region. Chromosome 4 loss in HNSCC MA Pershouse et al 373 The combination of allelic deletions and chromosomal tion of 200 mg/ml. After 24 h at 508C, the DNA was transfer studies strongly suggest the presence of a tumor puri®ed by extraction with phenol/chloroform and pre- suppressor gene with 4q24-26. This region was involved cipitated with ethanol. DNA was also extracted from in 480% of the tumors examined, suggesting that a peripheral blood lymphocytes from the appropriate putative chromosome 4q tumor suppressor gene may patients (El-Naggar et al., 1995). Microsatellite repeat DNA polymorphisms are examined play an important role in the evolution of HNSCC. This by PCR with slight modi®cations to previous methods is underscored by the lack of associations between allelic (Weber and May, 1989; Pershouse et al., 1993). PCR loss and clinopathologic characteristics in our cases. primers were obtained from Research Genetics (Huntsville, E€orts are currently underway to further molecularly AL) or were synthesized from published sequences (Murray de®ne the alterations a€ecting this region in order to et al., 1994). PCR (25 ml) reactions were mixed as follows, search for candidate tumor suppressor genes. 60 ng of DNA template, 12 pMoles of each primer, 6 mlof premixed PCR mix (0.833 mM dATP, 0.833 mM dCTP, 0.833 mM dTTP, 0.833 mM dGTP, 4.17 mM Tris-HCl,

Materials and methods 41.67 mM NaCl, 2.08 mM MgCl2 and 0.417 mM 2-mercap- toethanol), 0.16 ml[a32P]dCTP, 0.625 U Taq polymerase. The Tissue samples PCR products were generated during standard PCR reactions consisting of an initial 5 min, 948C denaturation step Twenty-seven HNSCC tumors were prospectively acces- followed by 25 cycles of 948C, 45 s; 558C, 45 s; 728C, 45 s, sioned in the Frozen Section Unit of the Department of followed by ®nal 3 min at 728C. Resolution of the Pathology, MD Anderson Cancer Center from 1991 to polymorphic bands was accomplished by denaturing poly- 1993. Peripheral blood lymphocytes from each patient were acrylamide gel electrophoresis (6% acrylamide, 8% urea, collected after Ficoll-Hypaque centrifugation. Tumor Sequegel; run at 80 watts for 2 ± 3 h) utilizing a Bio Rad samples were carefully trimmed to remove normal ®brous 38650 cm sequencing apparatus. The gels were vacuum material and veri®ed by frozen section and cytospin dried and exposed to X-ray ®lm with intensifying screen at preparation to contain 515% non-tumorous elements. 7708C for 4 ± 48 h. Physical mapping data was based upon Tissue samples from each specimen were immediately snap the Genethon map of chromosome 4 (Gyapay et al., 1994). frozen and kept at 7808C until used.

Acknowledgements Microsatellite polymorphism analysis This work was supported by grants from the National For preparation of high molecular weight DNA, frozen Cancer Institute RO1 CA 56041 and CA 51148, The tissues were ground in a lysing bu€er containing 0.2 M Gilland Foundation, and The MD Anderson PRS. MAP Tris-Cl, 1% SDS, 0.25 M NaCl, 25 mM EDTA (pH 8.5) is a postdoctoral Fellow of the American Brain Tumor followed by addition of Proteinase-K to a ®nal concentra- Association.

References

Ah-See KW, Cooke TG, Pickford IR, Soutar D and Balmain McGowan-Jordan IJ, Speevak MD, Blakey D and Chevrette A. (1994). Cancer Res., 54, 1617 ± 1621. M. (1994). Cancer Res., 54, 2568 ± 2572. Brachman DG, Graves D, Vokes E, Beckett M, Haraf D, Mitra AB, Murty VV, Li RG, Pratap M, Luthra UK and Montag A, Dumphy E, Mick R, Yandell D and Chaganti RS. (1994). Cancer Res., 54, 4481 ± 4487. Weichselbaum RR. (1992). Cancer Res., 52, 4832 ± 4836. Murray JC, Bennett SR, Kwitek AE, Small KW, Schinzel A, Buetow KH, Murray JC, Israel JL, London WT, Smith M, Alward WLM, Weber JL, Bell GI and Buetow KH. (1994). Kew M, Blanquet V, Brechot C, Redeker A and Govindar- Nature Genet., 2, 46 ± 49. ajah S. (1989). Proc. Natl. Acad. Sci. USA, 86, 8852 ± 8856. Nawroz H, van der Riet P, Hruban RH, Kock W, Ruppert Cairns P, Polascik TJ, Eby Y, Tokino K, Califano J, Merlo JM and Sidransky D. (1994). Cancer Res., 54, 1152 ± 1155. A, Mao L, Herath J, Jenkins R, Westra W, Rutter JI, Ning Y, Weber JL, Killary AM, Ledbetter DH, Smith JR Buckler A, Gabrielson E, Tockman M, Cho KR, Hedrick and Pereira-Smith OM. (1991). Proc. Natl. Acad. Sci. L, Bova GS, Isaacs W, Koch W, Schwab D and Sidransky USA, 88, 5635 ± 5639. D. (1995). Nature Genet., 11, 210 ± 212. Pershouse M, Stubble®eld E, Hadi A, Killary A, Yung WKA DeVesa SS, Blot WJ, Stone BJ, Miller BA, Tarone RE and and Steck P. (1993). Cancer Res., 53, 5043 ± 5050. Fraumeni JF. (1995). J. Natl. Cancer Inst., 87, 175 ± 182. Polascik TJ, Cairns P, Chang YH, Schoenberg MP and El-Naggar AK, Hurr K, Batsakis JG, Luna MA, Goepfert H Sidransky D. (1995). Cancer Res., 55, 5396 ± 5399. and Hu€ V. (1995). Cancer Res., 55, 2656 ± 2659. Sato T, Saito H, Morita R, Kol S, Lee JH and Nakamura Y. Fan J-B, DeYoung J, Lagace R, Lina RA, Xu Z, Murray JC, (1991). Cancer Res., 54, 5118 ± 5122. Buetow KH, Weissenbach J, Goold RD, Cox DR and Shin DM, Ro JY, Hong WK and Hittelman WN, (1994). Myers RM. (1994). Hum. Mol. Genet., 3, 243 ± 246. Cancer Res., 54, 321 ± 326. Fujimoto Y, Hampton LL, Wirth PJ, Wang NJ, Xie JP and Sidransky D. (1995). Curr. Op. Onc., 7, 229 ± 233. Thorgeirsson SS. (1994). Cancer Res., 54, 281 ± 285. Spitz MR, Fueger JJ, Halabi S, Schantz SP, Sample D and Grandis JR and Tweardy DJ. (1993). Cancer Res., 53, 3579 ± Hsu TC. (1993). Cancer Epidemiol. Biomarkers Prev., 2, 3584. 329 ± 333. Gyapay G, Morisette J, Vignal A, Dib C, Fizames C, Sreekantaiah C, Rao PH, Xu L, Sacks PG, Schantz SP and Millasseau P, Marc S, Bernardi G, Lathrop M and Chaganti RSK. (1994). Genes Chromosom. Cancer, 11, Weissenbach J. (1994). Nature Genet., 7, 268 ± 271. 29 ± 39. Hong WK, Lippman SM, Hittelman WN and Lotan R. Vokes EE, Weichselbaum RR, Lippman SM and Hong WK. (1995). Clinic Cancer Res., 1, 677 ± 686. (1993). N. Engl. J. Med., 328, 184 ± 194. Jin Y, Mertens F, Mandahl N, Heim S, Olegard C, Weber JL and May PE. (1989). Am.J.Genet.,44, 388 ± 396. Wennerberg J, Biorklund A and Mitelman F. (1993). Weinberg RA. (1991). Science, 254, 706 ± 709. Cancer Res., 53, 2140 ± 2146.