Loss of Heterozygosity at 3p14.2 in Clear Cell ... - Cancer Research

[CANCER RESEARCH 59, 1323–1326, March 15, 1999]

Loss of Heterozygosity at 3p14.2 in Clear Cell Renal Cell Carcinoma Is an Early Event and Is Highly Localized to the FHIT Gene Locus1 Marija Velickovic, Brett Delahunt, and Stefan K. G. Grebe2 Department of Pathology, Wellington School of Medicine, Wellington, New Zealand

ABSTRACT The VHL tumor suppressor gene (TSG) at 3p25–26 is strongly implicated in the pathogenesis of clear cell renal cell carcinoma (cRCC). In addition, 3p14.2 and 3p21 are suspected of harboring additional TSGs in cRCC, with FHIT being a candidate TSG at 3p14.2. We examined 87 microdissected, histologically well-defined cRCCs classified according to tumor-node-metastasis (TNM) stage (stage 1, 23 cases; stage 2, 14 cases; stage 3, 24 cases; stage 4, 26 cases) and Fuhrman grade (grade 1, 24 cases; grade 2, 19 cases; grade 3, 19 cases; grade 4, 8 cases; sarcomatoid cRCC, 17 cases) for loss of heterozygosity (LOH) at 3p14.2 and 3p25–26 using a series of precisely mapped microsatellite probes. We found that LOH at 3p14.2 exceeded LOH at 3p25–26 in frequency (69% versus 48.3%; P < 0.03) and was highly localized to markers within the FHIT gene locus (D3S1300 and D3S4260), with the majority of chromosomal breakpoints also mapping to this region. In addition, 3p14.2 LOH (P < 0.03), but not 3p25–26 LOH (P 5 nonsignificant), was associated with lower tumor grades (grades 1–3). These findings suggest that 3p14.2 genomic deletions may be among the earliest events in cRCC pathogenesis, preceding genomic deletions at the VHL locus. FHIT, or an as yet undiscovered TSG mapping to the D3S4103–D3S4260 interval, could be the molecular target of the 3p14.2 deletions.

INTRODUCTION Genomic deletions involving 3p are a typical feature of cRCC.3 It is believed that the VHL TSG is the molecular target of these deletions (1). However, based on LOH studies, it is suspected that additional cRCC TSGs may reside within 3p21 or 3p14.2 (2– 4). 3p14.2 contains FRA3B, the most common inducible fragile site in the human genome (5). Chromosomal breakage at FRA3B may represent a mechanism for loss of more telomeric portions of 3p, including VHL (1). Alternatively, 3p14.2 LOH may signify the presence of a TSG in this region. This latter possibility is suggested by the fact that FRA3B lies just telomeric of the constitutive familial RCC-associated translocation breakpoint t(3;8) (p14.2;q24.1) (6 –9) and within the recently cloned putative TSG FHIT (10). If FHIT is involved in cRCC pathogenesis, one would expect the majority of 3p14.2 genomic deletions and chromosomal breakpoints to be localized to markers within the FHIT gene. Furthermore, if FHIT inactivation is pathogenetically important in cRCC, a multistep carcinogenic model would predict LOH in this region to occur either before or after VHL inactivation but rarely simultaneously with VHL inactivation. Few previous LOH studies of RCC have included markers in the 3p14.2 region. In one of these, a high rate of LOH in the FRA3B and t(3;8) region was noted (9), whereas another study failed to confirm these findings (11). It is likely that the disparity in the results from

these studies was due to marker selection and the density of mapping. This apparent contradiction of results may be resolved by higher resolution LOH studies of 3p14.2. To date, there have been only two high-resolution LOH mapping studies of 3p14.2 in RCC (12, 13). These studies appeared to show that the majority of breakpoints lay outside of FHIT. However, in one of these studies, papillary RCCs and oncocytomas as well as cRCC were included, raising the possibility that the observations may not be representative for cRCC (12). In the second study, DNA had been extracted from short-term cultured tumor cells (13), and although this approach results in high-purity tumor DNA, it can also introduce tissue culture selection artifacts. In addition, neither study made attempts to correlate the molecular findings with clinical data or tumor histopathology. We therefore decided to examine LOH patterns at 3p14.2 in a series of carefully microdissected, histologically well-characterized, archival, paraffin-embedded cRCCs. This also gave us the opportunity to compare LOH rates at 3p14.2 with LOH rates near VHL and to correlate LOH rates for these regions with tumor differentiation as measured by tumor grade. We show that LOH at 3p14.2 in cRCC is highly localized to markers within FHIT, exceeds LOH near VHL, and occurs more commonly in lower grade tumors than LOH involving VHL. MATERIALS AND METHODS

All studies were approved and monitored by the Wellington Ethics Committee. Tumor Samples. We studied 91 archival paraffin-embedded cRCC specimens from patients treated with radical nephrectomy at Wellington Hospital (Wellington, New Zealand) for whom a minimum 5-year postoperative follow-up was available. All patients had complete tumor staging according to the tumor-node-metastasis (TNM) classification (14) and were assigned a Fuhrman grade (15). DNA Extraction. Histologically representative unstained 10-mm sections were microdissected into neoplastic compartments containing at least 80% tumor cells (typically .90%) and control compartments containing normal renal tissue from the same specimens without any microscopic evidence of neoplasia. Following microdissection, the specimens were deparaffinized by successive xylene and ethanol washes and incubated in 150 ml of DNA extraction buffer [2.7 mg/ml proteinase K, 100 mM Tris, and 2 mM EDTA (pH 8)] at 55°C for 48 h [additional proteinase K (2.7 mg/ml ) was added after 12 h]. After heat inactivation of proteinase K, aliquots of this digest were directly used for PCR. Microsatellite LOH Analysis. LOH analysis was performed by paired normal-tumor microsatellite PCR. Fig. 1 shows the localization of the microsatellite markers used in our study in linkage order according to the Genome Database (http://gdbwww.gdb.org/gdb/gdbtop.html), Genetic Location Database (http://cedar.genetics.soton.ac.uk/public_html/), and Cooperative Human Linkage Center (http://www.chlc.org/) databases. Received 11/2/98; accepted 1/14/99. PCRs contained 5 ml of template DNA, 200 mM deoxynucleotide triphosThe costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with phates, 500 nM primers, 0.02 unit/ml Taq polymerase (Qiagen), 13 PCR buffer 18 U.S.C. Section 1734 solely to indicate this fact. (Qiagen), 1.5–3 mM Mg21, and 2 mCi of [a-32P]dCTP (3000 Ci/mmol; DuPont 1 Supported by a Wellington Medical Research Foundation grant and by a Health New England Nuclear) in total reaction volumes of 50 ml. PCR cycle condiResearch Council of New Zealand Repatriation Fellowship (to S. K. G. G.). 2 To whom requests for reprints should be addressed, at Department of Pathology, tions were optimized as necessary. After PCR, denatured samples underwent Wellington School of Medicine, P. O. Box 7343, Wellington South, Wellington, New electrophoresis on 6% polyacrylamide gels. The gels were dried and autoraZealand. Phone: 64-4-385-5578; Fax: 64-4-389-5725; E-mail: [email protected]. diographed. The autoradiographs were scored for LOH and MIN. An allelic 3 The abbreviations used are: cRCC, clear cell renal cell carcinoma; TSG, tumor imbalance ratio (the intensity ratio of control alleles divided by the intensity suppressor gene; LOH, loss of heterozygosity; FRA3B, fragile chromosome 3 site B; ratio of tumor alleles) of ,0.6 or .1.67 was considered indicative of LOH. RCC, renal cell carcinoma; MIN, microsatellite instability. 1323

FHIT LOCUS LOH IN RCC

Fig. 2. LOH rate by marker. p, rates that are significantly different from the average LOH rate per marker of 19.8%.

MIN was considered to be present if tumor alleles differed in size from their matched control alleles or if tumor samples contained additional alleles that were not present in the matched control samples. Breakpoints were defined as regions between immediately adjacent markers with and without LOH.

57 tumors (Table 1). These were clustered primarily between D3S4103 and D3S1300 (30 breakpoints) and between D3S1300 and D3S4260 (24 breakpoints), all within the FHIT gene. There was no correlation between 3p LOH and tumor stage or patient survival; however, 3p LOH was correlated with tumor grade (Fig. 3). Compared with tumors without 3p LOH, more tumors with 3p LOH were of grade 1, 2, or 3 and fewer tumors with 3p LOH were of grade 4 or of the sarcomatoid type (P , 0.009). Division of tumors according to the site of 3p LOH showed a similar LOH-grade relationship for tumors with 3p14.2 LOH (P , 0.009). There was no correlation between LOH at 3p25–26 and tumor grade (Fig. 3).

RESULTS

DISCUSSION

Of the 91 tumor samples that were originally selected, 87 were successfully amplified by PCR at one or more of the studied 3p loci. Four specimens failed PCR amplification at all examined loci and were excluded from the study. For the 87 cases in this series, the average patient age at diagnosis was 59.6 years; 26 patients were female, and 61 patients were male. The overall 5-year mortality rate was 51.7%, with a median survival interval of 59 months. There were 23 stage 1 tumors, 14 stage 2 tumors , 24 stage 3 tumors, and 26 stage 4 tumors. Division of tumors according to Fuhrman grade resulted in 24 grade 1 tumors, 19 grade 2 tumors, 19 grade 3 tumors, 8 grade 4 tumors, and 17 sarcomatoid cRCCs. The markers used in our study were informative in a mean of 91% of cases, with a range of 82% (D3S4260) – 95% (D3S1335). 3p LOH was detected in 66 of 87 samples (75.8%). LOH at 3p14.2 occurred in 60 of 87 cases (69%), significantly more frequently (P , 0.03) than LOH at 3p25–26, which was seen in 48.3% of the samples. A total of 31.3% of tumors showed LOH at 3p14.2 without 3p25–26 LOH. In contrast, a significantly smaller percentage of tumors (6.9%) displayed exclusive 3p25–26 LOH (P , 0.01). The average rate of LOH per marker was 19.8%; it was 17.34% for 3p25–26 markers and 21.1% for 3p14.2 markers (P 5 nonsignificant). The most frequent site of LOH, and the only site with statistically significantly higher rates of LOH than the average rate for all markers studied (P , 0.01), was D3S1300 within 3p14.2 (43.1%; Fig. 2). A total of 10 samples showed MIN: 1 sample showed MIN at D3S1335 and D3S4260, 1 sample showed MIN at D3S1560, 3 samples showed MIN at D3S1481, 1 sample showed MIN at D3S1313, and 4 samples showed MIN at D3S1300. Within the 3p14.2 region, we observed a total of 131 breakpoints in

In our large series of histologically and clinically well-characterized cRCCs, the frequency of LOH at 3p14.2 significantly exceeded the frequency of LOH at 3p25–26. In both regions, LOH was more commonly interstitial, rather than involving large continuous chromosomal regions. At 3p14.2, LOH was highly localized to markers mapping around the FRA3B site, especially to intronic regions of the FHIT gene. Finally, LOH at 3p14.2 was correlated with lower tumor grades, whereas no such correlation was observed for LOH in the 3p25–26 region. These findings suggest that genomic deletion and chromosomal breakage in the FRA3B region may be among the earliest events in cRCC pathogenesis, possibly preceding VHL TSG loss or inactivation. The mapping density of the probes we used to study LOH at the VHL locus was less than that used for the 3p14.2 region, and none of our markers were located within the VHL gene. However, D3S1317 maps very closely to VHL, with no evidence of any meiotic crossover between D3S1317 and VHL (16). In terms of physical distance, D3S1317 lies just centromeric to the boundary of the smallest constitutional deletion (of less than 100 kb), which was observed in those von Hippel-Lindau families who were analyzed during the positional

Fig. 1. Chromosomal map positions and linkage order of 3p25–26 and 3p14.2 microsatellite markers used in this study, with marker spacing within the 3p14.2 and 3p25–25 regions proportional to published genetic positions and distances. A representative LOH example is included for each marker. N, normal; T, tumor. The positions of the VHL gene in relation to the 3p25–26 markers and the positions of the FHIT gene in relation to the 3p14.2 markers are indicated (FHIT untranslated exons, M; translated exons, f).

1324

Table 1 3p14.2 breakpointsa

a

3p14.2 marker intervals in linkage order

Breakpoints (n)

D3S1313–D3S2977 D3S2977–D3S1234 D3S1234–D3S4103 D3S4103–D3S1300 D3S1300–D3S4260 D3S4260–D3S1480 D3S1480–D3S1481

15 14 16 30 24 17 15

Regions between immediately adjacent markers with and without LOH.


Fig. 3. Grade distribution of tumors according to the presence or absence of LOH anywhere on 3p (left panel), at 3p14.2 (middle panel), and at 3p25–26 (right panel). SRC, sarcomatoid tumors. Tumors with LOH on 3p or at 3p14.2 were more likely to be of a lower grade (grades 1–3) than tumors without LOH in these regions (P , 0.009). There was no significant relationship between tumor grade and 3p25–26 LOH.

cloning of the VHL gene (16). Similarly, according to the Genome Database and the Genetic Location Database, D3S1560 lies approximately 700 kb telomeric to VHL, whereas D3S1335 and D3S1038 are located within 1.4 and 2.5 Mb centromeric to VHL, respectively. Therefore, it seems unlikely that we have significantly underestimated the LOH rate at the VHL locus, and the demonstrated higher rate of LOH at 3p14.2 represents a valid observation. Chromosomal breakage at 3p14.2 may be merely a marker for early tumor genetic instability; however, if this was the case, then one might expect 3p14.2 LOH to occur with equal frequency in tumors of all grades and degrees of differentiation. The fact that we observed 3p14.2 LOH predominately in lower grade tumors is more consistent with a true pathogenetic role for FRA3B breakage in early cRCC tumorigenesis. It is possible that chromosomal breakage at FRA3B leads to loss of the more telomeric regions of 3p, thereby deleting one allele of a putative upstream TSG, and this has been proposed as a mechanism for VHL inactivation (1). The predominantly interstitial pattern of LOH that we observed in our series of tumors argues against this scenario because contiguous loss of the entire portion of 3p telomeric to 3p14.2 was rarely observed. Nevertheless, we cannot completely discard this possibility because apparent interstitial LOH patterns may be generated by amplification of contaminating normal DNA in regions of homozygous tumor genetic deletions flanked by LOH regions (17, 18). However, because we used carefully microdissected specimens of high tumor purity, this occurrence is unlikely. Moreover, if VHL was the target of chromosomal breakage at FRA3B, then the relationship of LOH to tumor grade for the 3p14.2 region should mirror that of the 3p25–26 region. Our data show markedly different grade-LOH associations for LOH at 3p14.2 and LOH at 3p25–26, again suggesting that VHL is not a remote target of FRA3B chromosomal breakage. It thus appears that 3p14.2 LOH is probably not related to VHL inactivation, but it may be related to other cRCC pathogenetic events. Based on our data, the most likely target seems to be the FHIT gene. FHIT was cloned in 1996 from a region of homozygous deletion at 3p14.2 in gastrointestinal tumors (10). FHIT maps to 3p14.2 around the FRA3B site, and abnormal FHIT transcripts have been identified in several human tumor types (10, 19 –22). FHIT inactivation has also been proposed to be an early tumorigenic event (23, 24), consistent with our data on early 3p14.2 LOH.

It has been suggested that FHIT may not be important in RCC pathogenesis. In 1995, van den Berg et al. (11) screened 44 RCCs, 34 of which were cRCCs, for LOH on 3p and found high rates of LOH at 3p25, 3p21, and 3p14. When they investigated a subgroup of 31 of these tumors with additional 3p12–14 probes, 3p14 LOH was found to be uncommon. However, only two of their probes (D3S1480 and D3S1481) mapped to 3p14.2, and these two probes are usually not among the most commonly lost 3p14.2 markers. This suggests that probe selection bias may have led the authors to erroneously conclude that 3p14.2 was not a region of frequent LOH in RCC. More recently, Bugert et al. (13) examined the FRA3B region in 100 sporadic RCCs for LOH and concluded that, in the majority of cases, the breakpoints were proximal to FRA3B. Unfortunately, they did not provide any detail on actual LOH rates in the region, and all their tumor DNA specimens were derived from short-term cell cultures subject to possible culture selection effects. This confounding influence may also explain why they observed mainly contiguous stretches of LOH rather than interstitial LOH patterns in their material. In contrast, Shridhar et al. (12) analyzed a series of 94 archival RCCs using five 3p14.2 probes and showed a high 3p14.2 LOH rate of interstitial 3p deletions predominating over continuous losses of all or part of 3p. However, they also found that the majority of 3p14.2 breakpoints and the markers with the highest LOH rates mapped outside the FHIT locus. This latter study included an unselected mixture of RCC morphotypes derived from various sources with incomplete and inconsistent staging and grading, whereas the tumors in the study by Bugert et al. (13) were similarly unselected with samples identified only as nonpapillary RCCs, without any further histopathological classification, staging, or grading information. In our selected and graded tumor group consisting exclusively of cRCCs, we found that, contrary to the observations of the previous studies mentioned above, LOH was sharply confined to markers mapping to intronic regions of FHIT (D3S4103, D3S1300, and D3S4260), with two clear breakpoint clusters occurring between these three markers. These results are in agreement with earlier observations in RCC by Druck et al. (9) and are also consistent with the findings from other studies involving various nonrenal tumors (10, 19, 21, 22) and studies mapping the 3p14.2 breakpoint positions in lymphocytes (25, 26). The majority of FRA3B LOH and chromosome gaps occur within FHIT intron 5 and, less commonly, between exons 3 and 4 (intron 4), the two exons flanking the familial RCC t(3;8) breakpoint. FHIT exon 5 is the FHIT exon most frequently lost in cancer cells with a homozygote deletion at the FHIT locus (22). In this context, the observed association between lower grade tumors and 3p14.2 LOH, coupled with a LOH rate at the FHIT locus that exceeded that for 3p25–26 markers at the VHL locus, is consistent with the hypothesis that FHIT may be a very early cRCC TSG. Recent studies that have shown retention of wild-type FHIT RNA in short-term cultured and immortalized RCC cell lines (13, 27) do not necessarily contradict this hypothesis. Not all tumors may progress via the same pathways, and most tumors have significant clonal heterogeneity. If FHIT inactivation and subsequent VHL inactivation are early events, it is likely that they will occur, as observed, in relatively well-differentiated tumors, which may not be well suited for successful survival or growth in short-term or long-term tissue culture. Minor, less well-differentiated tumor subclones, which progress via alternative pathways that do not involve FHIT, may rapidly gain selective growth advantages under cell culture conditions. They might then contribute significant proportions of FHIT wild-type transcripts to RNA isolated from short- or long-term cultures or form the basis of a tumor cell line, thus leading to the impression that biallelic FHIT inactivation has not occurred in these tumors. Furthermore, a domi-

1325


nant negative effect of monoallelic abnormal FHIT transcripts remains a possibility. The results of a recent immunochemical study on 41 cRCC tumors provide evidence that FHIT protein is indeed underexpressed in cRCCs (28). These observations and our own data strongly support a role for FHIT in early cRCC tumorigenesis. We suggest that deletions at 3p14.2 within the FHIT locus are the earliest steps in the development of cRCC. We also suggest that these deletions precede loss of genetic material at the VHL locus and that FHIT inactivation is a crucial early tumorigenic event.

13.

14.

15. 16.

17.

REFERENCES 1. Gnarra, J. R., Tory, K., Weng, Y., Schmidt, L., and Wei, M. H. Mutations of the VHL tumour suppressor gene in renal carcinoma. Nat. Genet., 7: 85–90, 1994. 2. Lubinski, J., Hadaczek, P., Podolski, J., Toloczko, A., Sikorski, A., McCue, P., Druck, T., and Huebner, K. Common regions of deletion in chromosome regions 3p12 and 3p14.2 in primary clear cell renal carcinoma. Cancer Res., 55: 5348 –5353, 1995. 3. Willers, C. P., Siebert, R., Bardenheuer, W., Lux, A., Michaelis, S., Seeber, S., Opalka, B., and Schutte, J. Genetic instability of 3p12–p21-specific microsatellite sequences in renal cell carcinoma. Br. J. Urol., 77: 524 –529, 1996. 4. Wilhelm, M., Bugert, P., Kenck, C., Staehler, G., and Kovacs, G. Terminal deletion of chromosome 3p sequences in nonpapillary renal cell carcinomas: a breakpoint cluster between loci D3S1285 and D3S1603. Cancer Res., 55: 5383–5385, 1995. 5. Smeets, D. F. C. M., Scheres, J. M. J. C., and Husinx, T. W. J. The most common fragile site in men is 3p14. Hum. Genet., 72: 215–220, 1986. 6. Cohen, A. J., Li, F. P., Berg, S., Marchetto, D. J., Tsai, S., Jacobs, S. C., and Brown, R. S. Hereditary renal-cell carcinoma associated with a chromosomal translocation. N. Engl. J. Med., 301: 592–595, 1979. 7. Boldog, F. L., Gemmill, R. M., Wilke, C. M., Glover, T. W., Nilsson, A. S., Chandrasekharappa, S. C., Brown, R. S., Li, F. P., and Drabkin, H. A. Positional cloning of the hereditary renal carcinoma 3;8 chromosome translocation breakpoint. Proc. Natl. Acad. Sci. USA, 90: 8509 – 8513, 1993. 8. Paradee, W., Mullins, C., Zhanquan, H., Glover, T., Wilke, C., Opalka, B., Schutte, J., and Smith, D. I. Precise localization of aphidicolin-induced breakpoints on the short arm of human chromosome 3. Genomics, 27: 358 –361, 1995. 9. Druck, T., Kastury, K., Hadaczek, P., Podolski, J., Toloczko, A., Sikorski, A., Ohta, M., LaForgia, S., Lasota, J., McCue, P., Lubinski, J., and Huebner, K. Loss of heterozygosity at the familial RCC t(3;8) locus in most clear cell renal carcinomas. Cancer Res., 55: 5348 –5353, 1995. 10. Ohta, M., Inoue, H., Cotticelli, M. G., Kastury, K., Baffa, R., Palazzo, J., Siprashvili, Z., Mori, M., McCue, P., Druck, T., Croce, C. M., and Huebner, K. The FHIT gene, spanning the chromosome 3p14.2 fragile site and renal carcinoma-associated t(3;8) breakpoint, is abnormal in digestive tract cancers. Cell, 84: 587–597, 1996. 11. van den Berg, A., Hulsbeek, M. M. F., de Jong, D., Kok, K., Veldhuis, P. M. M. F., Roche, J., and Buys, C. H. C. M. Major role for a 3p21 region and lack of involvement of the t(3;8) breakpoint region in the development of renal cell carcinoma suggested by loss of heterozygosity analysis. Genes Chromosomes Cancer, 15: 64 –72, 1996. 12. Shridhar, V., Wang, L., Rosati, R., Paradee, W., Shridhar, R., Mullins, C., Sakr, W., Grignon, D., Miller, O. J., Sun, Q. C., Petros, J., and Smith, D. I. Frequent breakpoints

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

1326

in the region surrounding FRA3B in sporadic renal cell carcinomas. Oncogene, 14: 1269 –1277, 1997. Bugert, P., Wilhelm, M., and Kovacs, G. FHIT gene and the FRA3B region are not involved in the genetics of renal cell carcinomas. Genes Chromosomes Cancer, 20: 9 –15, 1997. Sobin, L. H., and Wittekind, C. (eds.). International Union Against Cancer (UICC) TNM Classification of Malignant Tumors, pp. 180 –182. New York: Wiley-Liss, 1997. Fuhrman, S. A., Lasky, L. C., and Limas, C. Prognostic significance of morphologic parameters in renal cell carcinoma. Am. J. Surg. Pathol., 6: 655– 663, 1982. Gnarra, J. R., Duan, D. R., Weng, Y., Humphrey, J. S., Chen, D. Y. T., Le, S., Pause, A., Dudley, C. F., Latif, F., Kuzmin, I., Schmidt, L., Duh, F-M., Stackhouse, T., Chen, F., Kishida, T., Wei, M. H., Lerman, M. I., Zbar, B., Klausner, D. R., and Linehan, W. M. Molecular cloning of the von Hippel-Lindau tumor suppressor gene and its role in renal carcinoma. Biochim. Biophys. Acta, 1242: 201–210, 1996. Cairns, P., Tokino, K., Eby, Y., and Sidransky, D. Homozygous deletions of 9p21 in primary human bladder tumors detected by comparative multiplex polymerase chain reaction. Cancer Res., 54: 1422–1424, 1994. Wang, S. I., Parsons, R., and Ittmann, M. Homozygous deletion of the PTEN tumor suppressor gene in a subset of prostate adenocarcinomas. Clin. Cancer Res., 4: 811– 815, 1998. Sozzi, G., Veronese, M. L., Negrini, M., Baffa, R., Cotticelli, M. G., Inoue, H., Tornielli, S., Pilotti, M. A., De Gregorio, L., Pastorino, U., Pierotti, M. A., Ohta, M., Huebner, K., and Croce, C. M. The FHIT gene at 3p14.2 is abnormal in lung cancer. Cell, 85: 17–26, 1996. Negrini, M., Monaco, C., Vorechovsky, I., Ohta, M., Druck, T., Baffa, R., Huebner, K., and Croce, C. M. The FHIT gene is abnormal in breast carcinomas. Cancer Res., 56: 3173–3179, 1996. Mao, L., Fan, Y. H., Lotan, R., and Hong, W. K. Frequent abnormalities of FHIT, a candidate tumor suppressor gene, in head and neck cancer cell lines. Cancer Res., 56: 5128 –5131, 1996. Huebner, K., Hadaczek, P., Siprashvili, Z., Druck, T., and Croce, C. M. The FHIT gene, a multiple tumor suppressor gene encompassing the carcinogen sensitive fragile site, FRA3B. Biochim. Biophys. Acta, 1332: M65–M70, 1997. Man, S., Ellis, I. O., Sibbering, M., Blamey, R. W., and Brook, J. D. High levels of allele loss at the FHIT and ATM genes in non-comedo ductal carcinoma in situ and grade I tubular invasive breast cancers. Cancer Res., 56: 5484 –5489, 1996. Ahmadian, M., Wistuba, I. I., Fong, K. M., Behrens, C., Kodagoda, D. R., Saboorian, M. H., Shay, J., Tomlinson, G. E., Blum, J., Minna, J. D., and Gazdar, A. F. Analysis of the FHIT gene and FRA3B region in sporadic breast cancer, preneoplastic lesions, and familial breast cancer probands. Cancer Res., 57: 3664 –3668, 1997. Zimonjic, D. B., Druck, T., Ohta, M., Kastury, K., Croce, C. M., and Popescu, N. C. Positions of chromosome 3p14.2 fragile sites (FRA3B) within the FHIT gene. Cancer Res., 57: 1166 –1170, 1997. Wilke, M. C., Hall, B. K., Hoge, A., Paradee, W., Smith, D. I., and Glover, T. W. FRA3B extended over a broad region and contains a spontaneous HPV integration site: direct evidence for the coincidence of viral integration sites and fragile sites. Hum. Mol. Genet., 5: 187–195, 1996. Wang, L., Darling, J., Zhang, J. S., Qian, C. P., Hartmann, L., Conover, C., Jenkins, R., and Smith, D. I. Frequent homozygous deletions in the FRA3B region in tumor cell lines still leave the FHIT exons intact. Oncogene, 16: 635– 642, 1998. Hadaczek, P., Siprashvili, Z., Markiewski, M., Domagala, W., Druck, T., McCue, P., Pekarsky, Y., Ohta, M., Huebner, K., and Lubinski, J. Absence or reduction of FHIT expression in most clear cell renal carcinomas. Cancer Res., 58: 2946 –2951, 1998.