Molecular Analysis of the FHIT Gene in Human Prostate Cancer

Alain Latil1, Ivan BieÁ che1, Georges Fournier2, Olivier Cussenot3, Sandrine Pesche1 and Rosette Lidereau1

1Laboratoire d'OncogeÂneÂtique, Centre ReneÂ Huguenin, F-92211 St-Cloud; 2Departement d'Urologie, HoÃpital de la Cavale Blanche, F-29609 Brest; 3Departement d'Urologie, CHU Saint-Louis F-75010 Paris, France

A variety of studies suggest that the FHIT gene, which 1993; Ittman, 1996; Cooney et al., 1996; Carter et al., encompasses the fragile site at 3p14.2, is a candidate 1990; Gao et al., 1995; Latil et al., 1994). The tumor suppressor gene in several forms of human cancer. demonstration of LOH in certain chromosomes has To determine whether the FHIT gene is altered in potentially implicated previously identi®ed TSGs, prostate carcinomas, we examined 15 prostate tumors, incuding CDH1 on 16q (Umbas et al., 1992), and four normal prostate tissue specimens and RNA from a DCC on 18q (Gao et al., 1993), even though the pool of 62 normal human prostate tissues (Clontech) for precise role of these TSGs in prostate cancer is the integrity of FHIT transcripts, using a robust single- controversial. Nevertheless, inactivation of TSGs has stage PCR and a nested PCR method. In each case a been shown to play an important role in the major FHIT-speci®c full-length product was observed. development of a variety of human cancers, and the Additional aberrantly sized products, which were more search is underway for other TSGs involved in prostate numerous in the nested PCR strategy, were present at a cancer. Although a variety of studies in various cancers far lower level than the full-length transcripts in 14 of 15 have suggested that tumor suppressor loci on tumors, three of four normal human prostate tissues and chromosome 3p play an important role in human in the pooled normal prostate RNA. Sequence analysis neoplasia, this chromosome arm has rarely been revealed that these aberrant products corresponded to investigated in prostate cancer; the few relevant alternatively spliced FHIT transcripts, which were studies have focused on the VHL locus at 3p25 neither more numerous nor more prominent in the (Cunningham et al., 1996; Dahiya et al., 1997). tumors than in the normal prostate specimens. Deletion Among the chromosome 3p regions thought to harbor at the FHIT locus was also evaluated by using three tumor suppressor loci, chromosome 3p14.2 has recently intragenic polymorphic markers (D13S1481, D3S1300, been found to be prominent. Ohta et al. (1996), using a and D3S1234). Allelic loss was observed in two tumors, positional cloning approach, identi®ed a novel gene in but these genomic alterations did not correspond to the this region, which they named FHIT (fragile histidine aberrant FHIT transcripts. DNA analysis, furthermore, triad gene) for its homology with the histidine triad suggested that the tumor heterogeneity was not a likely gene family. FHIT which encompasses approximately explanation for presence of normal and alternatively 1 Mb of human genomic DNA, including the FRA3B spliced FHIT transcripts in the prostate tumors. In common fragile region, encodes a typical dinucleoside conclusion, we detected neither frequent loss of hetero- 5',5''-P1, P3-triphosphate (Ap3A) hydrolase (Barnes et zygosity nor tumor speci®c transcripts of the FHIT gene al., 1996). In addition to its disruption in a kindred in human prostate cancer. with a reciprocal t(3;8) chromosomal translocation, abnormal transcripts containing deletions of one or Keywords: FHIT; tumor suppressor gene; prostate more coding exons, together with homozygous deletion cancer; reverse transcription; single stage PCR; nested of the region containing the gene and genomic DNA PCR rearrangements have been found in many established cancer cell lines and solid tumors (Kastury et al., 1996; Mao et al., 1996; Negrini et al., 1996; Panagopoulos et al., 1996; Sozzi et al., 1996a,b; Thiagalingam et al., Introduction 1996; Virgilio et al., 1996; Yanagisawa et al., 1996; Druck et al., 1997; Fong et al., 1997; Hayashi et al., Prostate cancer is one of the most common malig- 1997; Hendricks et al., 1997; Luan et al., 1997). These nancies among men in western countries (Parker et al., results make FHIT a candidate TSG for several human 1997). Little is known of the acquired genetic changes neoplasms. that drive these tumors through the various stages of With the aim of elucidating the role of FHIT in carcinogenesis. The most consistent genetic alterations prostate tumorigenesis, we examined 15 malignant in adenocarcinoma of the prostate are loss of prostate tumors for LOH status at 3p14.2, and the heterozygosity (LOH) involving chromosome arms transcription pattern of the FHIT gene. To investigate 7q, 8p, 10q, 13q, 16q, 17q and 18q, pointing to the the molecular basis of the potential involvement of the presence of tumor suppressor genes (TSGs) in these FHIT gene, we compared tumor DNAs and auto- deleted regions and their involvement in prostate logous normal DNAs, by using three polymorphic carcinogenesis (Zenklusen et al., 1994; Bova et al., markers withing the gene. After reverse transcription (RT), we examined the entire cDNA sequence of the Correspondence: R Lidereau gene (exon 1 ± 10) using both a robust single-stage PCR Received 1 September 1997; revised 3 November 1997; accepted method (Thiagalingam et al., 1996) and a nested PCR 4 November 1997 assay (Ohta et al., 1996). FHIT gene in prostate cancer A Latil et al 1864 junctions coincided with splice sites in all abnormal Results transcripts. The level of spliced FHIT variants was always below 1% and 10% of the FHIT-speci®c full- FHIT transcripts in primary prostate tumors length product in single-stage and nested PCR, Single-stage PCR yielded a FHIT-speci®c product of respectively. The alternatively spliced FHIT tran- the full length (756 bp) in each sample. In addition, scripts were detected due to the sensitive Applied seven tumors, two normal specimens and the pool of Biosystems model 373A DNA sequencing system; normal RNA yielded a faint band about 70 bp shorter indeed, most of them were not observed on agarose than the full-length product. Sequence analysis of this gel (Figure 2a and b). 687-bp product revealed a loss of exon 8 in all ten To con®rm the presence of spliced FHIT variants, cases. A 703-bp fragment generated by the loss of single-stage PCR was carried out using primer pairs exon 3 was also observed in four tumors and one 182U/864L, 195U/864L and 109U/726L, which re- normal prostate sample. Faint bands of smaller size vealed FHIT variants generated by the fusion of were also seen in one normal prostate sample (391 bp, exons 2 and 4, 2 and 7, and 7 and 9, respectively. A N111) and in one tumor (490 bp, T71). The 391-bp single band of variable intensity, corresponding to the product was generated by the fusion of exons 4 and 9, expected spliced FHIT transcripts (630, 258 and and the 490-bp transcript exhibited a loss of exons 3 ± 549 bp, respectively), was observed in 14 of the 15 5. tumors, three of the four normal prostate samples, and Nested PCR yielded a FHIT-speci®c product of the full-length (683 bp), in each sample. In addition, aberrantly sized products were observed in most of a

the samples; indeed only one normal prostate specimen T56 T59 T60 T62 T70 T72 T74 Blank N1 CLO N159 M (N88) and one tumor (T69) yielded a single major FHIT-speci®c full-length product. Aberrant bands observed in single-stage PCR were present in nested 756 bp PCR in the corresponding samples but novel aberrant bands were also seen after nested PCR. When another independent assay (repeat RT, and ®rst-round PCR) was performed, a major FHIT-speci®c full-length product was observed in each sample, but the size of the smaller band occasionally seen sometimes varied from one experiment to another. Abnormally sized transcripts were sequenced and were found to lack various regions between exons 2 and 9 both in normal and tumor tissues. The fusion

b T56 T59 T60 T62 T70 T72 T74 Blank N1 CLO N159 M

683 bp

Figure 2 Expression of the FHIT gene in prostate tissues. RNAs from prostate tumors (T), normal prostate samples (N), and a pool of normal human prostate tissues (CLO) were reverse transcribed and analysed for FHIT expression by using two dierent PCR strategies: (a) robust single-stage PCR and (b) nested PCR. (a) All samples showed a single normal-sized product (756 bp) using primers 109U and 864L located in exon 1 and exon 10, respectively. (b) Nested PCR was performed Figure 1 (A) Scheme of the wild-type FHIT transcript. Robust using the primer pair 109U/864L for the ®rst round of PCR. The single stage PCR was performed using primer pair 109U/864L, ampli®cation products were then subjected to a second round of whereas inner primers 155U and 837L were used for nested PCR PCR using the inner primers 155U and 837L in exons 2 and 10. assay. (B) Schematic representation of the most frequent In addition to the major band of 683 bp, faint PCR products of alternatively spliced FHIT transcripts detected in prostate smaller size appeared in some samples. M, lane containing specimens. Primer pairs 182U/864L, 109U/726L and 195U/864L commercial molecular weight marker (éX174 DNA digested with revealed FHIT variants generated by the fusion of exons 2 and 4, HaeIII). Blank, control run in which reverse transcriptase was 7 and 9, and 2 and 7, respectively omitted FHIT gene in prostate cancer A Latil et al 1865 in the pool of normal prostate RNA. The results Sequence analysis of the abnormally sized transcripts matched those obtained by means of nested PCR: all revealed a lack of various regions between exons 2 and FHIT-speci®c variants observed in a given specimen by 9 both in normal prostate tissues and tumors. The use nested PCR were found using the appropriate primer of primer pairs encompassing the entire cDNA pair. N88 and T69 were the only specimens not to yield sequence of full-length FHIT allowed us to determine a signal with primer pairs 182U/864L, 195U/864L and that the three most frequent spliced FHIT variants 109U/726L. were FHIT-speci®c transcripts which had lost exon 3, exon 8, and exons 3 ± 6 (Figure 1b). To con®rm the presence of these RNA splice variants, we analysed all Analysis of genomic DNA from prostate carcinomas the samples using primer pairs 182U/864L, 195U/864L To investigate the molecular basis of the observed and 109U/726L (see experimental procedures). The abnormalities in FHIT mRNA, we analysed the 15 630, 258 and 549-bp PCR products, corresponding to prostate tumor DNAs and autologous normal DNA, the transcripts generated by the fusion of exons 2 and using the polymorphic markers D13S1481, D3S1300, 4, 2 and 7, and 7 and 9, respectively, were seen in most and D3S1234, which lie within the FHIT gene. of the tumors, three of the four normal prostate Fourteen of the 15 prostate tumors were informative samples, and in the pooled RNA from the 62 normal for at least two microsatellite markers, whereas the last prostate tissues. Figure 3 shows examples of the (T60) was only heterozygous for the D3S1481 locus. patterns obtained with primer pairs 195U/864L. All Appepic Imbalance (AI) on at least one locus was FHIT-speci®c variants observed in a given specimen by found in two (T60 and T62) of the 15 tumor DNAs nested PCR were also found using the appropriate (13%). Results obtained by means of comparitive primer pair. These alternatively spliced transcripts were multiplex PCR suggested that the AI in these two clearly no more prominent in the tumors than in the tumors corresponded to LOH. T60 was deleted for normal prostate specimens in keeping with results from D3S1481 and was homozygous for D3S1300 and Boldog et al. (1997). The present result con®rms that D3S1234; T62 exhibited LOH at D3S1481 and FHIT variants are present at a far lower level (51%) D3S1234, and was homozygous for D3S1300. than the full-length transcript in both normal and neoplastic prostate tissue. As the FHIT open reading frame is contained in exons 5 ± 9 (Figure 1a), it is Discussion unlikely that most of these alternatively spliced transcripts encode functional proteins. The absence of To study abnormalities in FHIT transcripts, we reverse exon 8, which contains the histidine triad domain, is transcribed mRNA and ampli®ed the cDNAs by using also one of the most frequent variation observed in both a robust single-stage PCR method, and a nested breast cancer (Negrini et al., 1996), squamous cell PCR assay. In each of the 20 cases (15 prostate carcinomas of the head and neck (Virgilio et al., 1996), tumors, four normal prostate tissue specimens and renal cell carcinomas (Luan et al., 1997), and also in total RNA from a pool of 62 normal human prostate normal fetal brain cDNA (Boldog et al., 1997), tissues) a major FHIT-speci®c full-length product, whereas loss of exon 3, that ¯anks the t(3;8) containing the complete FHIT coding region, was translocation, has been observed in the prostate observed. These results matched those obtained on cancer-derived cell line LNCaP (Ohta et al., 1996). colorectal and esophageal cancers (Thiagalingam et al., Tumor heterogeneity might explain the presence of 1996; Zou et al., 1997). both normal and abnormal FHIT RT ± PCR products Additional bands of smaller size, representing alternatively spliced transcripts, were also seen. All the aberrant bands obtained in single-stage PCR were also found in nested PCR in the corresponding sample, but some additonal bands appeared in nested PCR. The discrepancies between the two PCR T56 T57 T58 T59 T60 T62 Blank N1 CLO N111 N159 M strategies may be because the nested PCR over- expressed the shorter PCR products, as suggested by the observation of a major full-length FHIT band accompanied by a complex pattern of abnormal transcripts when we used robust nested PCR (two 258 bp consecutive rounds of 35 cycles) (data not shown). Because FHIT variants are present at very low level, nested RT ± PCR is often necessary to detect them. However, when two independent assays starting with dierent cDNAs generated from a same RNA sample were performed, the size of the aberrant bands was not always identical in the two experiments. The phenom- Figure 3 Representative results obtained using primer pair enon have been observed in another study (Druck et 195U/864L (see Materials and methods), which revealed al., 1996), and might be due to the yield of the reverse alternative FHIT transcripts generated by the fusion of exon 2 transcription. Alternatively, spliced transcripts are and exon 7. The 258 bp FHIT variant was detected in tumors present at a very lower level, and thus a robust (T56, T57, T58, T59, T60, T62), normal prostate tissues (N1, ampli®cation may strongly favor ampli®cation of one N111, N159), and pooled RNA from normal human prostate tissues (CLO). M, lane containing commercial molecular weight among the dierent FHIT variants in accordance with marker (éX174 DNA digested with HaeIII). Blank, control run the RT-yield. in which reverse transcriptase was omitted FHIT gene in prostate cancer A Latil et al 1866 in individual tumors. Indeed, our LOH study on presence of tumor cells. A sample was considered suitable prostate tumors using three polymorphic markers for experiments if the proportion of tumor cells was more which lie within the FHIT gene (D3S1481, D3S1300, than 60%. Peripheral blood leukocytes were used as a and D3S1234) showed that two tumors (T60 and T62) source of normal DNA for each patient, and four normal were deleted at the FHIT locus. The relative intensity prostate tissue specimens and RNA from a pool of 62 normal human prostate tissues (Clontech) were used as of the two alleles in these tumor DNAs diered from sources of normal RNA. the relative intensity in lymphocyte DNA by a factor of 1.9. The proportion of tumor cells in T60 and T62 was 90% and 80%, respectively, and the value of 1.9 RNA extraction and RT ± PCR indicates that LOH at the FHIT locus occurred in 52% Total RNA was extracted from prostate samples by using of tumor cells in T60 and 59% in T62, suggesting the acid-phenol guanidium method (Chomczynski and intratumoral heterogeneity. However, the spliced FHIT Sacchi, 1987). The quality of the RNA samples was variants in these two tumors were present at a very low determined by electrophoresis through denaturing agarose level. Comparison of the results on FHIT variants and gels and staining with ethidium bromide, and the 18S and those of genomic DNA analysis argues against a 28S RNA bands were visualized under ultraviolet light. correlation between LOH at 3p14.2 and the aberrant The yield was quanti®ed spectrophotometrically in a 50 ml transcripts in these two tumors. DNA analysis also microcuve. Reverse transcription was performed in a total volume of 20 ml. Total RNA (1 mg) was reverse-transcribed suggested that genomic alteration of FHIT does not using random hexamers and MuLV reverse transcriptase play a causative role in the aberrant transcripts according to the protocol supplies with the GeneAmp1 observed in prostate cancer. Moreover, these alterna- RNA ± PCR Core kit (Perkin Elmer). One microliter of the tively spliced transcripts generated by fusion of reaction mix was used for PCR. dierent exons of the normal FHIT gene sequence were no more prominent in the tumors than in the normal prostate specimens, indicating that tumor Primers and PCR conditions heterogeneity is not a likely explanation for presence Single-stage and nested PCR were carried out in a ®nal of normal and abnormal FHIT RT ± PCR products in volume of 50 ml containing 1 ml of the RT reaction mix, individual tumors. The use of tumor samples contain- 20 pmoles of each primer, 200 mM each dNTP, 1.5 mM MgCl ,10mMTris-HCl pH 8.3, 50 mM KCland1unitof ing more than 60% of tumor cells made it impossible 2 that the normal products derived from contaminating Taq DNA polymerase (Perkin Elmer). Primers are designated by nucleotide position (relative to FHIT non neoplastic cells, contrary to lung cancer (Sozzi et Genbank #U49922) corresponding to the 5' position, al., 1996a), Merkel cell carcinoma (Sozzi et al., 1996b) followed by the letter U for upper (i.e. sense strand) or L and head and neck cancer (Virgilio et al., 1996). for lower (i.e. antisense strand). Single-stage PCR steps, In conclusion, our studies clearly show that an intact using the primer pair 109U(exon 1)/864L(exon 10) (Figure wild-type transcript, containing the complete FHIT 1), ampli®ed a 756-bp FHIT cDNA product. PCR coding region was observed in all normal and consisted of an initial denaturation at 948Cfor5min malignant prostate specimens studied here. This and 30 cycles of 1 min at 948C, 1 min at 628C, and 1.5 min normal product did not derive from contaminating at 728C, and a ®nal extension of 10 min at 728C, using a non neoplastic cells, and was always the major Perkin-Elmer 9600 DNA thermocycler. product. Alternatively spliced transcripts, which did Nested PCR was performed using 109U/864L for the ®rst round of ampli®cation (25 cycles). The ampli®ed products not appear to be due to DNA lesions, in the FHIT were then diluted 20-fold and 1 ml of the diluted product was locus, were coexpressed at a far lower level than the then subjected to 30 cycles in the above conditions, using the full-length transcript (51%). These aberrant tran- primer pair 155U(exon 2)/837L(exon 10), nested inside the scripts were also present in normal prostate tissue, original primers (Figure 1). The expected cDNA product is and were no more abundant in tumors than in normal 683 bp. Single-stage and nested PCR products from each prostate. Taken together, these results suggest that mRNA sample were then separated on 2% agarose gels, and aberrant FHIT transcripts were not related to prostate stained with ethidium bromide. To con®rm the presence of tumorigenesis, but simply re¯ected the plasticity of alternatively spliced speci®c FHIT transcripts, single-stage alternatively splicing. The observed FHIT variants may PCR ampli®cation of the RT products was carried out for 30 be due to the fact that the FHIT gene lies within the cycles using primer pairs 182U/864L, 195U/864L and 109U/ 726L. The oligonucleotide 182U spans the end of exon 2 and FRA3B common fragile site, as we had never the beginning of exon 4, 195U spans the end of exon 2 and previously observed such variants in other genes the beginning of exon 7, and 726L spans the beginning of (data not shown). exon 9 and the end of exon 7 (Figure 1). However, as alterations in translational control, or The sequences of the oligonucleotides were as follows: even altered patterns of Fhit protein expression, may 109U: 5'-CCTCCCTCTGCCTTTCATTC-3'; 155U: 5'-GAA also play a part in tumorigenesis, additional functional GCTCAGGAAAGAAGAGAAATC-3'; 182U: 5'-TGAGAA- studies are required to de®ne the potential role of CAGTCTGTAAAGGTATCCTA-3'; 195U: 5'-TAAAGGAT FHIT in prostate cancer. GGCCCCGAAG-3'; 726L: 5'-TGTCATGTTTCTGGAGC TTCAC-3'; 837L: 5'-GCTGGAATTCAGGATCTGAAA-3'; 864L: 5'-TTCAAACTGGTTGGCAATAGC-3'. Materials and methods Quantitative FHIT transcript analysis Patients and samples The ratio of DNA fragments generated by FHIT Fifteen primary prostate tumor specimens were obtained transcripts (aberrant/wild-type) was independent of the from patients undergoing surgery at St Louis Hospital in cycle number. Thus, coampli®cation of all transcripts Paris, and La Cavale Blanche Hospital in Brest (France). provided a semiquantitative RT ± PCR method, in which The samples were examined histologically to evaluate the the internal control was the coampli®ed wild-type FHIT FHIT gene in prostate cancer A Latil et al 1867 mRNA, and comparative expression of abnormal mRNA sequence of the primer pairs were obtained from the and wild-type FHIT mRNA could be determined for each Genome DataBase: D13S1481 (GDB ID: GOO-197-992), sample. Oligonucleotides 109U and 837L were ¯uorescein- D3S1300 (GDB ID: GOO-188-420), and D3S1234 labeled at their 5' end by adding Fluorprime during (GDB ID: GOO-186-387). The PCR reactions were oligonucleotide synthesis, and the sensitive Applied carried out in a total volume of 15 ml containing 50 ng of Biosystems model 373A DNA sequencing system (Perkin genomic DNA, 3 pmol of each primer (one ¯uorescently Elmer) was used to detect and quantify the dierent labeled), 1 mM each deoxynucleotide triphosphate, 1.5 mM signals. Single-stage and nested PCR products were MgCl2,10mMTris-HCl pH 8.3, 50 mM KCl and 0.6 units diluted tenfold and 2.5 ml-aliquots of each were added to of Taq DNA polymerase. Conditions for ampli®cations 2.5 ml of deionized formamide containing 0.3 mlofa were: 948C for 5 min, 12 cycles of 948Cfor15s,558Cfor molecular size marker (Genescan 2500 ROX, Perkin 15 s 728C for 30 s followed by 22 cycles of 898Cfor15s, Elmer). The mixtures were heat-denatured and 2.5 ml 558C for 15 s and 728C for 30 s. The ®nal extension step at aliquots of each were loaded on 6% polyacrylamide 8-M 728C was extended to 10 min. Products were diluted 20- urea gel and run for 6 h at 1200 V in the sequencer. The fold. The procedure which followed is described above (see dierent fragments were quanti®ed with the Genescan1 672 Quantitative FHIT transcript analysis). Allelic imbalance Fragment Analysis software (Perkin Elmer), which was considered to be present when the relative intensity of calculated the size and area of the peaks. The area of the two alleles in tumor DNA diered from the relative each peak for a given cDNA sample was then normalized intensity in lymphocyte DNA by a factor of at least 1.5; to the sum of the areas of all the peaks obtained with the this is a validated cut-o value to determine AI of same sample. microsatellite loci in tumor samples containing 460% cancer cells (McGrogan et al., 1994). Each analysis was performed at least twice to ensure reproducible detection of Sequencing of PCR products AI (repeat independent PCR ampli®cation, gel separation, PCR products were resolved in 1.5% low-melting-point and quanti®cation). In an attempt to distinguish allelic agarose gel (BRL) stained with etidium bromide. DNA gain from LOH, comparative multiplex PCR (Cairns et al., fragments were cut from gels and puri®ed using the 1994) was performed with the microsatellite marker QIAquick PCR puri®cation kit (Qiagen). The puri®ed D4S244 or D21S222 (Hudson et al., 1992) located in fragments were sequenced using the Applied Biosystems unaltered chromosomal regions (Watanabe et al., 1995), as (ABI) PRISM Dye Terminator Cycle Sequencing kit internal controls. (Perkin Elmer) on the Applied Biosystems model 373A DNA sequencing system (Perkin Elmer). Acknowledgements This work was supported by the Ligue Nationale de Lutte DNA analysis Contre le Cancer (LNCC); the ComiteÂ sReÂgionaux des We used the polymorphic markers D13S1481, D3S1300, Hauts de Seine, du Val d'Oise et des Yvelines; and the and D3S1234 which lie within the FHIT gene (intron 3, Association pour la Recherche sur les Tumeurs Prosta- intron 5 and intron 8, respectively), for LOH analysis. The tiques (ARTP).

References

Barnes LD, Garrison PN, Siprashvili Z, Guranowski A, Fong KM, Biesterveld EJ, Virmani A, Wistuba I, Sekido Y, Robinson AK, Ingram SW, Croce CM, Ohta M and Ader SA, Ahmadian M, Ong ST, Rassool FV, Zimmer- Huebner K. (1996). Biochemistry, 35, 11529 ± 11535. man PV, Giaccone G, Gazdar AF and Minna JD. (1997). Boldog F, Gemmill RM, West J, Robinson M, Robinson L, Cancer Res., 57, 2256 ± 2267. Li E, Roche J, Todd S, Waggoner B, Lundstrom R, Gao X, Honn KV, Grignon D, Sakr W and Chen YQ. (1993). Jacobson J, Mullokandov MR, Klinger H and Drabkin Cancer Res., 53, 2723 ± 2727. HA. (1997). Human Mol. Genet., 2, 193 ± 203. Gao X, Zacharek A, Salkowski A, Grignon DJ, Sakr W, Bova GS, Carter BS, Bussemakers MJ, Emi M, Fujiwara Y, Porter AT and Honn KV. (1995). Cancer Res., 55, 1002 ± Kiprianou N, Jacobs SC, Robinson JC, Epstein JI, Walsh 1005. PC and Isaacs WB. (1993). Cancer Res., 53, 3869 ± 3873. Hayashi S-I, Tanimoto K, Hajiro-Nakanishi K, Tsuchiya E, Cairns P, Tokino K, Eby Y and Sidransky D. (1994). Cancer Kurosumi M, Higashi Y, Imai K, Suga K and Nakachi K. Res., 54, 1422 ± 1424. (1997). Cancer Res., 57, 1981 ± 1985. Carter BS, Ewing CM, Ward WS, Treiger BF, Aalders TW, Hendricks DT, Taylor R, Reed M and Birrer MJ. (1997). Schalken JA, Epstein JI and Isaacs WB. (1990). Proc. Cancer Res., 57, 2112 ± 2115. Natl. Acad. Sci. USA, 87, 8751 ± 8755. Hudson TJ, Engelstein M, Lee MK, Ho EC, Ruben®eld MJ, Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162, Adams CP, Housman DE and Dracopoli NC. (1992). 156 ± 159. Genomics, 13, 622 ± 629. Cooney KA, Wetzel JC, Merajver SD, Macoska JA, Ittman M. (1996). Cancer Res., 56, 2143 ± 2147. Singleton TP and Wojno KJ. (1996). Cancer Res., 56, Kastury K, Baa R, Druck T, Ohta M, Coticelli MG, Inoue 1142 ± 1145. H, Negrini M, Rugge M, Huang D, Croce CM, Palazzo J Cunningham JM, Shan A, Wick MJ, McDonnell SK, Schaid and Huebner K. (1996). Cancer Res., 56, 978 ± 983. DJ, Tester DJ, Qian J, Takahashi S, Jenkings RB, Latil A, Baron J-C, Cussenot O, Fournier G, Soussi T, Bostwick DG and Thibodeau SN. (1996). Cancer Res., Boccon-Gibod L, Leduc A, RoueÈ sseÂ J and Lidereau R. 56, 4475 ± 4482. (1994). Genes Chrom. Cancer, 11, 119 ± 125. Dahiya R, McCarville J, Hu W, Lee C, Chui RM, Kaur G Luan X, Shi G, Zohouri M, Paradee W, Smith DI, Decker and Deng G. (1997). Int. J. Cancer, 71, 20 ± 25. HJ and Cannizzaro LA. (1997). Oncogene, 15, 79 ± 86. Druck T, Hadaczek P, Fu T-B, Ohta M, Siprashvili Z, Baa Mao L, Fan Y-H, Lotan R and Hong WK. (1996). Cancer R, Negrini M, Kastury K, Veronese ML, Rosen D, Res., 56, 5128 ± 5131. Rothstein J, McCue P, Cotticelli G, Inoue H, Croce CM and Huebner K. (1997). Cancer Res., 57, 504 ± 512. FHIT gene in prostate cancer A Latil et al 1868 McGrogan D, Levy A, Bostwick D, Wagner M, Wells D and Thiagalingam S, Lisitsyn NA, Hamagushi M, Wigler MH, Bookstein R. (1994). Genes Chrom. Cancer, 10, 151 ± 159. Willson JKV, Markowitz SD, Leach FS, Kinzler KW and NegriniM,MonacoC,VorechovskyI,OhtaM,DruckT, Vogelstein B. (1996). Cancer Res., 56, 2936 ± 2939. Baa R, Huebner K and Croce CM. (1996). Cancer Res., Umbas R, Schalken JA, Aalders TW, Carter BS, Karthaus 56, 3173 ± 3178. HFM, Schaafsma HE, Debruyne FMJ and Isaacs WB. Ohta M, Inoue H, Coticelli MG, Kastury K, Baa R, (1992). Cancer Res., 52, 5104 ± 5109. Palazzo J, Siprashvili Z, Mori M, McCue P, Druck T, Virgilio L, Shuster M, Gollin SM, Veronese ML, Ohta M, Croce CM and Huebner K. (1996). Cell, 84, 587 ± 597. Huebner K and Croce CM. (1996). Proc. Natl. Acad. Sci. Panagopoulos I, Pandis N, Thelin S, Peterson C, Mertens F, USA, 93, 9770 ± 9775. Borg A, Kristoersson U, Mitelman F and Aman P. Watanabe M, Imai H, Shiraishi T, Shimazaki J, Kotate T (1996). Cancer Res., 56, 4871 ± 4875. and Yatani R. (1995). Br. J. Cancer, 72, 562 ± 564. Parker SL, Tong T, Bolden S and Wingo PA. (1997). CA. Yanagisawa K, Kondo M, Osada H, Uchida K, Takagi K, Cancer J. Clin., 47, 5±27. Masuda A, Takahashi T and Takahashi T. (1996). Cancer Sozzi G, Veronese ML, Negrini M, Baa R, Coticelli MG, Res., 56, 5579 ± 5582. Inoue H, Tornielli S, Pilotti S, De Gregorio L, Pastorino Zenklusen JC, Thompson JC, Troncoso P, Kagan J and U, Pierotti MA, Ohta M, Huebner K and Croce CM. Conti CJ. (1994). Cancer Res., 54, 6370 ± 6373. (1996a). Cell, 85, 17 ± 26. Zou T-T, Lei J, Shi Y-Q, Yin J, Wang S, Souza RF, Kong D, Sozzi G, Alder H, Tornielli S, Corletto V, Baa R, Veronese Shimada Y, Smolinski KN, Greenwald BD, Abraham JM, ML, Negrini M, Pilotti S, Pierotti MA, Huebner K and Harpaz N and Meltzer SJ. (1997). Oncogene, 15, 101 ± 105. Croce CM. (1996b). Cancer Res., 56, 2472 ± 2474.