Loss of heterozygosity and reduced expression of the CUTL1 ... - Nature

Oncogene (1997) 14, 2355 ± 2365  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Loss of heterozygosity and reduced expression of the CUTL1 gene in uterine leiomyomas Wendy Rong Zeng1, Stephen W Scherer2, Michael Koutsilieris3, Jack J Huizenga2, Frederic Filteau3, Lap-Chee Tsui2,4 and Alain Nepveu1,5 1

Departments of Medicine, Oncology, and 5Biochemistry (AN), McGill University, Montreal, Quebec; 2Department of Genetics, The Hospital for Sick Children, Toronto; 3Laboratoire d'Endocrinologie MoleÂculaire, Centre de Recherches du Centre Hospitalier Universitaire de l'UniversiteÂ Laval, Sainte-Foy, Quebec and 4The Department of Molecular and Medical Genetics, The University of Toronto, Toronto, Ontario, Canada

Cytogenetic analyses has revealed deletions and/or rearrangments at several chromosomal positions in approximately half of uterine leiomyomas. The most frequent genetic alteration, deletion of 7q22, was found in approximately 35% of studied cases with cytogenetic abnormalities (128/366=35%). The same chromosomal band was also found to be deleted in a fraction of acute myeloid leukemias and myelodysplastic syndromes. The frequent deletion of 7q22 in some tumors suggest that a tumor suppressor gene may be located in this region. The human Cut-like homeobox gene, CUTL1, is one of the genes localized to 7q22 and it was shown previously to encode a transcriptional repressor that down-modulates the expression of c-Myc. Activation of the c-Myc oncogenic potential has been shown in many cancers to result from alterations in one or the other of its several mechanisms of regulation. These observations led us to hypothesize that CUTL1 could act as a tumor suppressor gene. In the present study, we have identi®ed polymorphic markers within and directly adjacent to CUTL1 at 7q22 and demonstrated that these markers are present in a commonly deleted region in seven out of 50 uterine leiomyomas samples examined. Furthermore, Northern blot analysis revealed that CUTL1 mRNA levels were reduced in eight tumors out of 13. These results suggest that CUTL1 may act as a tumor suppressor gene whose inactivation could be of pathological importance in the etiology of uterine leiomyomas. Keywords: uterine leiomyoma; tumor suppressor gene; chromosome 7q22; CUTL1; human Cut homeobox gene; loss of heterozygosity

Introduction Leiomyomas are benign tumors of smooth muscle origin that arise at low frequency in a large number of tissues but are found most often in uterus (Ozisik et al., 1993b; Scherer and Tsui, 1991). Uterine leiomyomas arise more speci®cally from smooth muscle cells of the myometrium. Leiomyomas rarely evolve to become malignant (leiomyosarcomas), but they are extremely frequent. It is estimated that between 20 to 30% of Correspondence: A Nepveu, Molecular Oncology Group, Royal Victoria Hospital, H5-08, 687 Pine Ave West, Montreal, Quebec, Canada H3A 1A1 Received 30 September 1996; revised 31 January 1997; accepted 3 February 1997

women over the age of 30 will develop a leiomyoma (Ozisik et al., 1993b; Wilcox et al., 1994). Leiomyomas therefore represent the most frequent benign tumor in women. Between 1988 and 1990 in United States, 1 700 000 (1.7 million) women underwent hysterectomy and of that number approximately one-third, half a million, were diagnosed to have a leiomyoma (Wilcox et al., 1994). From the point of view of the experimentalist, uterine leiomyomas represent an excellent tumor system for molecular investigation. These tumors occur very frequently and they are always resected together with abundant adjacent normal myometrium which can be used as control tissues in molecular studies. Moreover, leiomyomas present themselves as homogenous tumor tissue, with little or no in®ltration by normal cells such as macrophages. Finally, leiomyomas are sometimes very large, a feature that makes it possible to isolate DNA and RNA from the same tumor. The expression pattern of candidate tumor suppressor genes can therefore be investigated directly in the tumor. Cytogenetic analysis of uterine leiomyomas has revealed that gross chromosomal rearrangements occur in approximately half of the cases. Five main cytogenetic subgroups have been de®ned: del(7q), 6p rearrangements, del(13q), t(12,14), trisomy 12 (Dal Cin et al., 1995a, b; Nilbert et al., 1988, 1989, 1990; Ozisik et al., 1993a; Pandis et al., 1991; Rein et al., 1991; Sait et al., 1989; Sargent et al., 1994; Vanni et al., 1991) (reviewed in Ozisik et al., 1993b). The most frequent genetic alteration, del(7q), was found in approximately 35% of studied cases with cytogenetic abnormalities (128/366), and the smallest commonly deleted region of 7q was mapped to band 7q22 (Ozisik et al., 1993b; Sargent et al., 1994). Deletion of 7q22 was also found in a small proportion of primary AML (7.6%) and MDS (19%), however, its incidence increased to 26.8 and 41% respectively in secondary AML and MDS (Litt et al., 1993; Pandis et al., 1990, 1991). The high proportion of cytogenetically detectable deletions of 7q22 in dierent cancers suggests that a tumor suppressor gene may be located within this chromosomal region (Ozisik et al., 1993b; Yunis et al., 1988a). Three homeobox genes (CUTL1, DLX5 and DLX6) have been mapped previously to 7q22 (Lemieux et al., 1994; Scherer et al., 1993a; Simeone et al., 1994). Since homeobox genes act as master genes that control cell fate, it is possible that the homeodomain proteins could act as tumor suppressors. Interestingly, the product of the CUTL1 (Cut-like 1) gene was shown to bind to the

Refined mapping of a uterine leiomyoma tumor suppressor gene at 7q22 WR Zeng et al

promoter of the c-Myc proto-oncogene and represses its expression (Dufort and Nepveu, 1994). Several studies have demonstrated that the c-Myc protooncogene is frequently deregulated in tumors originating from dierent tissues (reviewed in Marcu et al., 1992). In fact, c-Myc represents the proto-oncogene most frequently activated in tumors. Indeed, several mechanisms of oncogenic activation have been revealed through studies of a c-Myc: chromosomal translocation, retroviral insertion, retroviral transduction, gene ampli®cation and loss of transcription elongation control (reviewed in Marcu et al., 1992). The chromosomal position of CUTL1 together with the function of the human Cut protein as a transcriptional repressor of the c-Myc proto-oncogene suggests that CUTL1 may act as a tumor suppressor gene. To explore this hypothesis, we have identi®ed polymorphic markers near CUTL1 and assessed whether the gene is situated in a region of 7q22 that is deleted in uterine leiomyomas. In addition, we compared the expression of CUTL1 in leiomyomas and matched control tissues. The results indicate that CUTL1 is located in the smallest deleted region in our collection of uterine leiomyomas and that expression of CUTL1 is reduced in eight out of 13 leiomyomas studied.

microsatellite markers near to the gene needed to be identi®ed. Our mapping studies demonstrated that three (CA)n repeats, D7S515, D7S518, D7S666, were present in the same yeast arti®cial chromosome (YAC) clone (HSC7E404) that was shown previously to contain the complete CUTL1 gene (Semenza et al., 1991a ; Semenza and Wang, 1992). The YAC clone was approximately 440 kb in size, so in order to further de®ne the positions of the markers relative to CUTL1, a higher resolution map consisting of bacteriophage, cosmids and P1-arti®cial chromosome (PAC) clones was assembled.

a #7 #8 #12 #13 #14 #16 #17 #21 #22

#23 #25.1 #25.2 #26 #27 #27s #28 #29 #31

#7 #8 #12 #13 #14 #16 #17 #21 #22

#23 #25.1 #25.2 #26 #27 #27s #28 #29 #31

Probe: CUTL1 coiled-coil

Probe: D7S518

Results Identi®cation of polymorphic markers close to CUTL1

M.W. (Kbp)

Sma1I

NcoI

AccI

EcoRI

Apol

Sma1I

NcoI

AccI

b

EcoRI

To assess whether CUTL1 was located in a region of 7q22 that was frequently delted in uterine leiomyomas,

Apol

2356

— 11

➝

—5 —4

— 0.8 Probe: CUTL1 cDNA nt 23–997

Figure 1 A cosmid and PAC map spanning D7S518, D7S666, D7S515 and the CUTL1 gene. DNA marker content map of PACs (designated PAC-plate-row-column; 384-well microtiter plate format) and cosmids (plate-row-column; 96-well format) spanning the CUTL1 gene. The presence of each DNA marker within the clone is indicated by a black circle. The orientation of the contig along the chromosome is not known but based on YAC contig analysis and patient deletion analysis we have tentatively placed it as 7cen-3'CUTL1-5'-7qter. The average size inserts in PAC and cosmid vectors are 130 kb and 35 kb, respectively

Probe: D7S666

Figure 2 Mapping of markers D7S518 and D7S666 relative to CUTL1. (a) DNA from bacteriophage Lambda clones were slotblotted on a nitrocellulose membrane and hybridized with a CUTL1 cDNA fragment corresponding to coiled-coil region (left panel) and the D7S518 marker (right panel). Two clones, #8 and #16 gave a positive signal with both probes. A representation of the human Cut protein (CUTL1) is displayed at the bottom. The evolutionarily conserved domains are depicted as boxes: the coiled-coil (CC), the Cut repeats (CR) and the homeodomain (HD). (b) DNA from cosmid clone 22-1 was digested with a panel of restriction enzymes, separated by electrophoresis on an agarose gel, transferred to a nitrocellulose membrane and hybridized with a CUTL1 cDNA fragment corresponding to nucleotides 23 to 997 (left panel) and with marker D7S666. The arrow indicates the smallest fragment that hybridizes to both probes


D7S518 was located in intronic DNA close to exon sequences that encode the coiled-coil domain of the protein (Figure 2).

To complete CUTL1 gene was found to be contained within PAC clone 123e15 indicating the maximum size of the gene is approximately 100 kbp (Figure 1). The three microsatellite markers were found to be present in at least one PAC clone that contained the CUTL1 sequences. D7S515 was located immediately 5'-end of CUTL1 (Figure 1). D7S666 was positioned within 4 kb of the 5'-end of CUTL1 while

66 N

Microsatellite analysis of 7q in uterine leiomyomas To determine whether the genomic region encompassing the CUTL1 gene was commonly deleted in our collection

68 T

N

78 T

N

98

82 T

N

T

N

T

527

66 N

78 T

N

82 T

N

97 T

N

T

98 N T

515

66 N

68 T

N

T

N

T

82 N T

T

N

97 N T

98 N T

97

98 N T

666

67

66 N

68 T

N

T

518

Figure 3 Representative PCR ampli®cations of (CA)n microsatellite repeats. Representative PCR ampli®cations of (CA)n microsatellite repeats D7S527, D7S515, D7S666, D7S518, and D7S496 are shown. Oligonucleotide primers were used to PCR amplify the regions of DNA containing these markers, in the presence of radiolabeled dCTP. Products were denatured and separated on a standard 6% sequencing gel or a 6% sequencing gel containing formamide. DNAs from tumors (T) and matched normal myometrium tissues (N) were ampli®ed from 50 patients with leiomyomas. A patient is considered to be informative if there are two bands (corresponding to two alleles) in the normal DNA lane. A patient shows LOH if, in the tumor DNA lane, one of the alleles is absent or shows diminished intensity. For example, patient 66 is informative for the marker D7S527, and shows LOH at that locus, while patient 98 is informative with this marker but shows no LOH, and patient 97 is uninformative for marker D7S666

2357


2358

of 50 uterine leiomyomas, the D7S515, D7S518, and D7S666 markers were analysed for loss of heterozygosity (LOH). In addition, the extent of the deletion in these tumors was determined by examining 15 additional (CA)n markers than span 7q. The physical order of all of the microsatellite markers and CUTL1 with respect to each other have been established previously using YAC contig and somatic cell hybrid analysis (see Materials and methods). Representative LOH results obtained with four markers (D7S527, D7S515, D7S666 and D7S518) are shown in Figure 3 and the results are summarized in Table 1. LOH was scored with at least one marker in seven of 50 (14%) of the tumors examined (patients 66, 67, 68, 82, 97, 98, and 78, see Figure 4). Tumor #67 and #82 contained non-contiguous deletions but the superposition of the overlapping deletions in the seven tumors indicted a common region of deletion existed and that it encompassed the CUTL1 gene (Figure 4). The proximal (centromeric) and distal (telomeric) boundaries of the critical region were de®ned by breakpoints in tumors #67 and #98 that were ¯anked by the D7S518 and D7S496 markers, respectively (Figure 4). Through our physical mapping study we also determined that a gene homologous to the yeast PMS gene, which are members of the mismatch repair genes, was positioned within the cosmid and PAC contig covering CUTL1 (Figure 1).

Northern blot analysis of CUTL1 expression in leiomyomas RNAs from tumors and matched control tissues were ®rst analysed by agarose electrophoresis and ethidium bromide staining to ascertain their quality. Since CUTL1 mRNA is of 5.5 kb, samples that showed a lower ration of 28 s versus 18 s ribosomal RNAs were not further investigated (data not shown). In the end, samples from 13 patients were analysed by Northern blot hybridization, including three patients with LOH of CUTL1 (#67, 68 and 98). Nitrocellulose membranes were hybridized successively with radiolabeled probes for CUTL1, c-Myc and g-actin (Figure 5). Levels of CUTL1 mRNAs were found to be lower in tumors than in normal tissues in eight out of 13 (8/13) patients (Figure 5, tumors #86, 96, 102, 99, 46, 67, 68, 16), including two with LOH of CUTL1 (#67 and 68). CUTL1 mRNA levels were similar in tumors and normal tissues in four of the remaining patients (Figure 5, #70, 98, 10, 15), while it was higher in one patient (Figure 5, #73). There was no obvious correlation between the levels of CUTL1 and c-Myc mRNAs. Moreover, variations in the level of CUTL1 and c-Myc mRNAs could be seen between dierent normal tissues. These dierences may re¯ect changes in gene expression in dierent phases of the menstrual cycle or dierent positions within the tumors and organs (Jacobs et al., 1985; Lin et al., 1985; Scherer and Tsui, 1991). Indeed, it is possible that the relative postion of a sample within a tumor, whether it localized closer to the periphery or the center of the tumor, may aect the expression of these genes. In summary, CUTL1 mRNA expression was reduced in most leiomyomas studied, but there was no correlation between CUTL1 and c-Myc expression.

Table 1

LOH analysis of 50 leiomyomas using 18 polymorphic markers on chromosome 7

Markers

Patients tested

Informative patients

Patients with LOH

D7S524 D7S527 D7S479 D7S647 D7S477 D7S662 D7S518 D7S666 D7S515 D7S658 D7S501 D7S496 D7S471 D7S486 D7S480 D7S504 D7S500 D7S483

6 7 7 7 7 9 50 50 50 8 50 7 50 7 3 2 5 6

5 7 6 6 3 8 42 33 39 3 32 6 30 4 3 2 5 6

0 5 5 4 2 2 6 4 6 3 5 6 6 3 2 1 1 1

DNAs from tumors and matched normal myometrium tissues were ampli®ed from 50 patients with leiomyomas using oligonucleotide primers for 18 polymorphic markers on chromosome 7q. The list of markers is presented, together with the numbers of patients tested, informative patients and patients with LOH for each marker. The level of informativeness observed for each of the markers in our cohort of patients was consistent with the published values

Table 2 Polymorphic markers used for the study of microsatellite instability No.

Marker

1 2 3 4 5 6 7 8 9 10

D2S123 D10S197 D11S904 D13S175 SCA1 D12S755E DM AR VWF-DNR DXS981

Type Dinucleotide Dinucleotide Dinucleotide Dinucleotide Trinucleotide Trinucleotide Trinucleotide Trinucleotide Tetranucleotide Tetranucleotide

Chromosome Location 2 10 11 13 6 12 19 X X X

Patients with LOH for markers on 7q22 were analysed using 10 polymorphic markers previously shown to be susceptible to instability in tumor cells with defective mismatch repair. The name of each marker is given, together with the type of repeat, the sequence of repeat unit and the chromosome location

RNase mapping analysis of CUTL1 alternatively spliced products A Cut alternatively spliced product, called CASP, has recently been characterized (Simons et al., manuscript in preparation). CASP shares with CUTL1 the aminoterminal region including a coiled-coil domain, but none of the Cut DNA binding and repression domains (Figure 6). Although the function of the CASP protein yet remains to be de®ned, the presence of a putative dimerization domain in both CASP and CUTL1 proteins raises the possibility that CASP functions as an inhibitor of CUTL1 function in a manner similar to IC/EBP (Buck et al., 1994; Ron and Habener, 1992). Therefore, we hypothesized that genetic alterations in leiomyomas may lead to the aberrant expression of this short protein. To compare the ratio of the two alternatively spliced products of the CUTL1 gene in


leiomyomas and matched normal tissues, RNase mapping analysis was performed using a riboprobe derived from the CASP cDNA and encompasing sequences shared by the CUTL1 and CASP mRNAs (Figure 6). It should be noted that this assay cannot serve to compare the level of CUTL1 mRNAs in tumor and normal tissue samples, since there is no internal control to verify the amount of RNAs in each sample. However, within each sample, the amount of the two CUTL1 mRNAs can be compared. Five pairs of samples were analysed by RNase mapping and the relative levels of the two mRNAs was found to be identical in each paired sample. We conclude that the relative abundance of the two CUTL1 mRNAs is not changed in leiomyomas. Study of microsatellite instability in leiomyomas The deleted region in some of our leiomyomas includes a cluster of genes of the PMS family, that show

homology to the mutL gene of E. coli (Horii et al., 1994; Nicolaides et al., 1995). (J Huizenga and S Scherer, unpublished observations). It is not known yet whether the PMS genes on 7q22 are functional, however it is formally possible that inactivation of these genes would cause a defect in mismatch repair. To test this hypothesis, we investigated microsatellite instability in the seven tumors with LOH on 7q22. Our analysis included ten polymorphic markers, situated on chromosomes other than seven, and previously shown to be susceptible to length variation in tumors with defects in mismatch repair (Table 2) (Aaltonen et al., 1993; Wooster et al., 1994). No change in allele length was observed (Figure 7). Since microsatellite instability in other studies were readily revealed by the analysis of a smaller number of markers, we conclude that leiomyomas are not defective mismatch repair and that the PMS genes on 7q22 do not represent the tumor suppressor gene that is selectively inactivated in this type of tumor.

Figure 4 Mapping of LOH within 7q. Left: representation of human chromosome 7 including band assignments; right: names of polymorphic markers used are given, along with chromosome band assignments and the position of several genes. Seven columns representing seven patients with a loss of heterozygosity on chromosome 7q are shown. Explanation of the symbols used is given. The smallest common deletion in these seven patients is indicated by the vertical line to the right. All 43 tumors not shown in this ®gure were informative for at least one marker within the critical region but failed to show LOH

2359


2360

muscle (leiomyomas) and myeloid cells (AML and MDS). In the present study, we have identi®ed polymorphic markers within or close to the CUTL1 gene and shown that these markers are consistently deleted in uterine leiomyomas with LOH at 7q22. These results are in agreement with the study of Ishwad et al. (1995), in which the smallest common deleted region was mapped between markers D7S471 and D7S518 (Ishwad et al., 1995). Moreover, the LOH frequency of polymorphic markers as determined in these two studies agreed well with the frequency of chromosomal deletion in cytogenetic studies (17%). However, Ishwad et al. noted that approximately 40% (17 of 40) of the tumors with cytogenetically de®ned 7q deletions had no LOH for markers in this chromosomal region (Ishwad et al., 1995). Two explanations were provided by the authors to explain the lack of allelic loss in many tumors with cytogenetically visible 7q deletion. First, many leiomyomas were mosaic for the 7q deletion and included many karyotipically normal cells. Secondly,

Discussion Cytogenic analyses indicated that the chromosomal region 7q22 was deleted in approximately 17% of uterine leiomyomas (reviewed in Ozisik et al., 1993b). This chromosomal region has also been found to be rearranged in acute myeloid leukemia (AML) and myelodisplastic syndromes (MDS) (Fenaux et al., 1989; Heim, 1992; Swansbury et al., 1994; Yunis et al., 1988b). Frequent deletion or rearrangements of a speci®c chromosomal region in cancers is generally taken to suggest that a tumor suppressor gene resides in this region and that deletion or gene conversion of one allele occurs during the development of a tumor, while the other allele suers mutations. These two events, mutation of one allele and deletion or gene conversion of the other allele, contribute to cause loss-of-function of that particular tumor suppressor gene. It should be noted that in the case of 7q22, it is not clear whether the same tumor suppressor gene is inactivated in tumors originating from smooth

96

86

70 N

N

T

T

N

T

N

N

T

10

99

102

98

N

T

N

T

T

c-Myc

Cut

Actin

46

73 N

T

N

67 T

N

68 T

N

15 T

N

16 T

N

T

c-Myc

Cut

Actin

Figure 5 Northern blot analysis of RNAs from leiomyomas and matched normal tissues. RNAs were prepared from seven leiomyomas and matched normal tissues, separated by electrophoresis on a denaturing agarose gel, transferred to a nitrocellulose membrane and hybridized with probes for the c-Myc, CUTL1 and actin genes


4

N

T

296 nt

➝ 3

N

256 nt

5

6

96 T

7

N

8

98 T

9

N

10

T

296 nt

➝

2

T

➝ 1

N

67

70

➝

Probe

2361

86

256 nt

11

Figure 6 RNase mapping analysis of RNAs from leiomyomas and matched normal tissues. RNAs from ®ve leiomyomas and matched normal tissues were submitted to RNase mapping analysis using a riboprobe derived from the CUTL1 alternatively spliced product (CASP) and encompassing sequences shared by the CUTL1 and CASP mRNAs. Maps of the CUTL1 and CASP mRNAs are shown at the bottom along the riboprobe and the fragments protected by each mRNAs. Note that that the fragment protected by the CASP mRNA is larger, since the riboprobe is derived from its cDNA

some tumors had complex chromosomal rearrangements in addition to the 7q deletion, suggesting that some of the genetic information deleted from 7q had probably integrated elsewhere in the genome (Ishwad et al., 1995). These results indicate that complex genomic rearrangements occur frequently in uterine leiomyomas and that LOH analysis is probably more accurate than cytogentic analysis in assessing allelic loss. As shown in Figure 4, the smallest common region of LOH in uterine leiomyomas remains large and includes several other genes in addition to CUTL1 : DRA (down-regulated in adenomas), LAMB1 (laminin B1), DLD (dihydrolipoamide dehydrogenase gene), PRKAR2B (the gene encoding the regulatory subunit RII beta of human cAMPdependent protein kinase) and PMS (members of the mismatch repair gene family) (Figure 4) (Horii et al., 1994; Nicolaides et al., 1995; Pikkarainen et al., 1987; Scherer et al., 1993b; Solberg et al., 1992; Taguchi et al., 1994). A complementary approach for the identi®cation of a tumor suppressor gene consists in investigating the function and expression pattern of genes that reside in the interval de®ned by genetic analyses. First, a putative tumor suppressor gene for a particular type of cancer should code for a product whose function is compatible with the notion that it can serve as a growth-suppressor for that tissue-type.

Secondly, a candidate gene should be expressed in the normal tissue in which this cancer arises. Thirdly, its expression should be reduced and/or altered in a fraction of tumors. In the later case, unless the gene product is of a dierent size than normal, altered expression is usually revealed by the presence of mutations in the expressed mRNA or in the remaining allele. Using the above criteria, certain predictions can be made concerning the involvement of genes present in this interval. Both DRA and PRKAR2B were found to be expressed in a strictly tissue-speci®c manner and in tissues other than the uterus (Levy et al., 1988; Schweinfest et al., 1993). On the other hand, DLD exhibited all the hallmarks of a house-keeping gene (Johanning et al., 1992). Thus, the expression pattern of these three genes would make their involvement in uterine leiomyomas very unlikely. The various laminins represent the major glycoprotein component of the basement membrane and as such, are believed to play a critical role in the process of tumor invasion and metastasis (Timpl et al., 1979). Indeed, an abundant literature exists on the interaction between cancer cells and the extracellular matrix, and the expression of some laminin receptors and laminin-binding proteins was reported to be increased in a variety of cancer cells (reviewed in Castronovo, 1993; Honn and Tang, 1992). In the context of a benign tumor like the uterine


2362

However, using ten polymorphic markers, situated elsewhere than on chromosome 7, we have shown that there was no length variation within these markers in the seven tumors with LOH of 7q22. Previous studies on tumors with a defect in DNA repair showed that marker length variation could be detected using a smaller number of markers. Therefore we conclude that uterine leiomyomas do not exhibit defects in DNA repair and that the PMS genes must not be the aected tumor suppressor genes. The smallest commonly deleted region, as de®ned in this study, does not include the PAI1 gene (Figure 4). Indeed, in a recent study PAI1 mRNA levels were found to be higher in uterine leiomyomas than in

leiomyoma, it is not immediately apparent how loss-offunction of LAMB1 could contribute to cellular transformation, however, we cannot exclude that alterations in the extracellular matrix could result in the production of a proliferation stimulatory signal. Therefore, it would seem appropriate to maintain LAMB1 in the list of candidate tumor suppressor gene for uterine leiomyomas. The PMS genes on 7q22, if expressed and functional, would likely code for proteins involved in DNA repair and as such could function as tumor suppressor genes (Horii et al., 1994; Nicolaides et al., 1995). Inactivation of DNA repair genes in tumors has been shown to cause the phenomenon of microsatellite instability. D2S123 66 67 68 82 97 N T N T N T N T N T

D10S197 66 67 68 82 N T N T N T N T

D11S904 66 67 N T N T

D13S175 66 67 N T N T

VWF-DNR 67 68 N T N T

68 N T

68 N T

78 N T

98 N T

97 NT

98 N T

82 97 N T N T

82 N T

97 N T

82 N T

66 N T

78 N T

T

97 N T

D12S755E 66 67 68 N T N TN T

78 N

78 N

67 N T

78 N

98 N T

98 NT

DM 66 67 68 78 82 97 98 NTNT NT NT NT NT NT AR 66 68 78 82 97 98 NT NT NT NT NT NT

T

98 N T

T

78 82 97 N T N TN T

66 68 67 78 N T N T N T N T

SCA1 66 N T

67 N T

68 N T

78 N T

98 N T

82 97 N T N T

82 N T

97 N T

98 N T

98 N T

Figure 7 Study of microsatellite instability. Patients with LOH on 7q22 were analysed for microsatellite instability. Presented are PCR ampli®cations of 10 microsatellite repeats previously shown to be susceptible to length instability in tumor cells with defective mismatch repair. Note that the markers are localized to chromosomes other than 7


normal myometrium in the majority of the cases (Sourla et al, 1996, #115). Thus, genetic data and expression studies both indicate that, at least in uterine leiomyomas, PAI1 does not function as a tumor suppressor gene. We have demonstrated that CUTL1 is expressed in uterus. Moreover, in addition to the large CUTL1 mRNA cording for the human Cut protein, we have shown that an alternatively spliced product is also present in uterus. We have investigated CUTL1 mRNA expression in 13 tumors and found reduced level of expression in eight cases, including tumors with and without LOH of 7q22. These results and the role of the CUTL1 transcription factor as a repressor of cMyc, a proto-oncogene that is frequently activated in human cancers, are consistent with the notion that CUTL1 could be a tumor suppressor for leiomyomas. Howver, c-Myc mRNA levels did not increase proportionally to the decrease in CUTL1 expression, suggesting that regulation of c-Myc expression is dependent on additional factors including the time within the menstrual cycle, hormonal stimulation or the position within a tumor (Jacobs et al., 1985; Lin et al., 1985; Scherer and Tsui, 1991). Further studies should aim to investigate the factors that aect c-Myc expression in uterine myometrium and leiomyomas. In conclusion, CUTL1 is the ®rst gene mapped in the interval of LOH at 7q22 to be studied in leiomyomas. The decrease in CUTL1 expression observed in several tumors con®rms its status as a candidate tumor suppressor gene. A ®nal assessment of the involvement of CUTL1 in the development of leiomyomas will require the determination of whether the remaining allele is mutated. Materials and methods Tissue specimens and genomic DNA extraction Tumor and matched normal tissue samples were obtained from 43 patients ± at the time of hysterectomy at the Centre Hospitalier Universitaire de l'UniversiteÂ Laval and HoÃpital Saint-FrancËois d'Assise, Quebec, and from seven patients at the Royal Victoria Hospital, Montreal, Canada. All patients have signed an inform consent approved by the local Ethics Committees. All tumors were classi®ed as benign leiomyomas. Tissue samples were snap frozen in liquid nitrogen immediately after surgery and stored at 7808C until DNA or RNA extraction. DNA was extracted from tumor and normal tissue using a standard proteinase K digestion and phenol-chloroform extraction protocol. RNA was prepared using the procedure described by Chomczynski and Sacchi (1987). DNA probes The complete human CUTL1 cDNA probe was used for the cosmid and PAC library screening (Neufeld et al., 1992). Probes speci®c for the human PMS3-8 genes were also used (Horii et al., 1994). The 1102-2, 1102-12, 1102-15, and 1067-12 probes were sub-clones of PAC76h2 (1102) and 82p8 (1067), respectively. Microsatellite repeat analysis Eighteen (CA)n repeat microsatellite markers on chromosome 7 were used to identify the region of loss of chromosome 7 in uterine leiomyomas: D7S524, D7S527,

D7S479, D7S477, D7S662, D7S647, D7S515, D7S666, D7S518, D7S658, D7S501, D7S471, D7S496, D7S486, D7S480, D7S504, D7S500 and D7S483. Information of the primer sequences and allele lengths are available in the Genome DataBase (GDB). These (CA) n repeats were ampli®ed by PCR in a ®nal volume of 50 ml, containing 50 ng DNA, 1.5 mM MgCl2, 5 ml standard 106 PCR buer (200 mM Tris-HCl pH 8.4, 500 mM KCl), 0.4 mM of each primer, 0.125 mM dNTPs, 1 mCi [a-32P] dCTP, and 1 units of Taq polymerase (GibcoBRL, Burlington, Ontario). An initial step of 5 min at 958C was followed by 35 cycles of 1 min of denaturation at 948C, 30 s of annealing at 598C, and 1.5 min of extension at 728C, followed by a ®nal extension step of 7 min at 728C. PCR products were separated in a 6% sequencing gel containing formamide (Litt et al., 1993) or a conventional 6% sequencing gel. For the study of microsatellite instability, we chose ten polymorphic markers, situated on chromosomes other than 7, and previously shown to be susceptible to length variation in tumors with defects in mismatch repair, including four (CA)n dinucleotide repeats (D2S123, D10S197, D11S904, D13S175), four trinucleotide repeats (SCA1, D12S755E, DM, AR) and two tetranucleotide repeats (VWF-DNR, DXS981) (Aaltonen et al., 1993; Wooster et al., 1994). Ordering of the microsatellite markers The order of the 18 microsatellite markers shown in Figure 3 was determined in the following way. The order 7q21.3D7S524 - D7S527 - D7S479 - D7S477 - D7S662-D7S647-7q22 was determined by YAC contig and somatic cell hybrid analysis in a previous study (Pugh et al., 1991). The order 7q22-D7S658-D7S501-D7S496-D7S471-D7S486-7q31 was determined using YAC contig analysis, pulsed-®eld gel electrophoresis, and radiation hybrid mapping (Goldberg et al., 1988; Schuster et al., 1989). D7S518-D7S666-D7S515 were placed between the two previously mentioned sets of markers using somatic cell hybrid analysis and genetic linkage mapping (Beru et al., 1986; Semenza et al., 1991b). Although a YAC contig has been established surrounding these markers these three markers were suciently close that their order along the chromosome could not be discerned. D7S504 and D7S483 were previously mapped to 7q32 and 7q36, respectively (Fu et al., 1993; Maxwell et al., 1993; Shoemaker and Mitsock, 1986). Probes 2071, 2091 and CDP-3' correspond to the following sequences of the HSCDP cDNA (accession #M74099): 1 to 997, 1968 to 2505 and 3936 to 5376. Finally D7S480 and D7S500 were mapped unequivocally to 7q31 and 7q33-q35, respectively, in our unpublished work. The published and unpublished mapping studies are available at: http://www.genet. sickkids.on.ca/chromosome7/. Determination of LOH Allelic loss was scored only on informative patients whose normal DNA samples were polymorphic at a given locus. Patients who were uninformative were not considered. LOH was identi®ed, visually or following phosphoimager densitometric analysis, as a loss in intensity (490%) or complete loss of one allele in the tumor DNA when compared with the normal DNA from the same patient. All cases of LOH were con®rmed by three separate experiments with two dierent reviewers. Southern blot analysis DNA preparation of the genomic clones was carried out using the alkaline lysis method. DNA was digested with restriction endonucleases, separated by gel electrophoresis, and blotted to nylon membranes (Hybond N, Amersham)

2363


2364

according to standard procedures. The blots were u.v. cross-linked, prehybridized for 2 h at 428C, and hybridized overnight at 428C with 16106 c.p.m./ml 32P-labeled DNA probes. Filters were washed four times in 0.16SSC, 0.1% SDS, at 658C for 20 min each. They were exposed to X-ray ®lm with intensifying screens at 7808C overnight. Northern blot analysis Northern blot analysis was performed essentially as described previously (Thomas, 1980). Total RNA was size fractionated on a denaturing 6% formaldehyde ± 1% agarose gel and transferred to Nitrocellulose (Schleicher & Schuell). Hybridization was carried out overnight at 428C with a 32P-dCTP-labeled probe, 50 mM HEPES (pH 7), 0.75 M NaCl, 50% formamide, 3.5% SDS, 56Denharts', 2 mM EDTA, 0.1% SDS, and 200 mg/ml salmon sperm DNA. Filters were washed four times in the same buer, 0.16SSC, 0.1% SDS, at 658C for 20 min each. They were exposed to X-ray ®lm with intensifying screens at 7808C overnight. RNase T2 protection analysis The riboprobe was prepared by combining 1 mg of template DNA, transcription buer (200 mM PIPES, 2 M NaCl, 5 mM EDTA), 10 m M DTT, RNasin (40 units), 500 mM ATP, CTP, GTP, 12 mM UTP, 50 mCi a-32P-UTP and T7 RNA polymerase (69 units) (Pharmacia) and then incubating for 1 h at 378C. After 1 h, 500 mM UTP was added and further incubated for 5 min. Next the riboprobes were treated with RNase free DNase at 378C for 15 min and then extracted with chloropane and run through a Sephadex G50 spun column. Forty mg of total

RNA was annealed to 86105 c.p.m. of labeled riboprobe at 548C for 16 h in 80% formamide 70.4 M NaCl 70.4 M piperazine-N,N-bis (2-ethanesulfonic acid) (PIPES) (pH 6.4)- 1 mM EDTA. RNA-RNA hybrids were digested with 30 U of RNase T2 (Gibco) per ml at 308C for 1 h. After digestion hybrids were precipitated with 20 mg of tRNA, 295 ml 4 M guanidine thiocyanate and 590 ml of isopropanol. Pellets were resuspended in 80% formamide, 16TBE and 0.1% XC+BPB, denatured and electrophoresed on 4% acrylamide-8M urea gel. Gels were dried and exposed to X-ray ®lm with intensifying screens at 7808C for appropriate time.

Acknowledgements We are grateful to Dr Peter Watson for introducing us to the problem of uterine leiomyomas and to Dr Mark Boyd and Ms Lise Potvin from the Department of Obstetrics and Gynecology of the Royal Victoria Hospital for helping us in obtaining uterine leiomyoma samples. We thank Dr Morag Park for critical reading of this manuscript, Dr France Maily for some precious advices on PCR assays; Ginette BeÂrubeÂ, Jenny Lin and Robert Sladek for helpful technical advices; Jonathan Simons for providing the plasmid for riboprobe for RNase mapping. AN is the recipient of a scholarship from the Fonds de la Recherche en SanteÂ du QueÂbec. WZ is the recipient of a fellowship from The Royal Victoria Hospital Research Institute. This research was funded by operating grants to AN from the Cancer Research Society, and to SWS from the Canadian Genome Analysis and Technology Program.

References Aaltonen LA, Peltomaki P, Leach FS, Sistonen P, Pylkkanen L, Mecklin JP, Jarvinen H, Powell SM, Jen J, Hamilton SR et al. (1993). Science, 260, 812 ± 816. Beru N, McDonald J, Lacombe C and Goldwasser E. (1986). Molecular & Cellular Biology, 6, 2571 ± 2575. Buck M, Turler H and Chojkier M. (1994). EMBO Journal, 13, 851 ± 860. Castronovo V. (1993). Invasion & Metastasis, 13, 1 ± 30. Chomczynski P and Sacchi N. (1987). Analyt Biochem., 162, 156 ± 159. Dal Cin P, Moerman P, Deprest J, Brosens I and Ven den Berghe H. (1995a). Genes, Chromosomes & Cancer, 13, 219 ± 220. Dal Cin P, Vanni R, Marras S, Moerman P, Kools P, Andria M, Valdes E, Deprest J, Van de Ven W and Van den Berghe H. (1995b). Cancer Research, 55, 1565 ± 1568. Dufort D and Nepveu A. (1994). Molecular & Cellular Biology, 14, 4251 ± 4257. Fenaux P, Preudhomme C, Lai JL, Morel P, Beuscart R and Bauters F. (1989). British Journal of Haematology, 73, 61 ± 67. Fu P, Evans B, Lim GB, Moritz K and Wintour EM. (1993). Molecular & Cellular Endocrinology, 93, 107 ± 116. Goldberg MA, Dunning SP and Bunn HF. (1988). Science, 242, 1412 ± 1415. Heim S. (1992). Leukemia Research, 16, 43 ± 46. Honn KV and Tang DG. (1992). Cancer & Metastasis Reviews, 11, 353 ± 375. Horii A, Han JH, Sasaki S, Shimada M and Nakamura Y. (1994). Biochemical & Biophysical Research Communications, 204, 1257 ± 1264. Ishwad CS, Ferrell RE, Davare J, Meloni AM, Sandberg AA and Surti U. (1995). Genes, Chromosomes & Cancer, 14, 51 ± 55.

Jacobs K, Shoemaker C, Rudersdorf R, Neill SD, Kaufman RJ, Mufson A, Seehra J, Jones SS, Hewick R, Fritsch EF et al. (1985). Nature, 313, 806 ± 810. Johanning GL, Morris JI, Madhusudhan KT, Samols D and Patel MS. (1992). Proceedings of the National Academy of Sciences of the United States of America 89, 10964 ± 10968. Lemieux N, Zhang XX, Dufort D and Nepveu A. (1994). Genomics, 24, 191 ± 193. Levy FO, Oyen O, Sandberg M, Tasken K, Eskild W, Hansson V and Jahnsen T. (1988). Molecular Endocrinolgy, 2, 1364 ± 1373. Lin FK, Suggs S. Lin CH, Browne JK, Smalling R, Egrie JC, Chen KK, Fox GM, Martin F, Stabinsky Z et al. (1985). Proceedings of the National Academy of Sciences of the United States of America, 82, 7580 ± 7584. Litt M, Hauge X and Sharma V. (1993). Biotechniques, 15, 280 ± 284. Marcu KB, Bossone SA and Patel AJ. (1992). Annu Rev Biochem., 61, 809 ± 860. Maxwell PH, Pugh CW and Ratclie PJ. (1993). Proceedings of the National Academy of Sciences of the United States of America, 90, 2423 ± 2427. Neufeld EJ, Skalnik DG, Lievens P-J and Orkin SH. (1992). Nature Genet., 1, 50 ± 55. Nicolaides NC, Carter KC, Shell BK, Papadopoulos N, Vogelstein B and Kinzler KW. (1995). Genomics, 30, 195 ± 206. Nilbert M, Heim S, Mandahl N, Floderus UM, Willen H and Mitelman F. (1988). Cytogenetics & Cell Genetics, 49, 300 ± 304. Nilbert M, Heim S, Mandahl N, Floderus UM, Willen H and Mitelman F. (1989). Cancer Genetics & Cytogenetics, 42, 51 ± 53.


Nilbert M, Heim S, Mandahl N, Floderus UM, Willen H and Mitelman F. (1990). Human Genetics, 85, 605 ± 611. Ozisik YY, Meloni AM, Powell M, Surti U and Sandberg AA. (1993a). Cancer Genetics & Cytogenetics, 67, 59 ± 64. Ozisik YY, Meloni AM, Surti U and Sandberg AA. (1993b). Cancer Genetics & Cytogenetics, 71, 1 ± 6. Pandis N, Heim S, Bardi G, Floderus UM, Willen H, Mandahl N and Mitelman F. (1990). Genes, Chromosomes & Cancer, 2, 311 ± 317. Pandis N, Heim S, Bardi G, Floderus UM, Willen H, Mandahl N and Mitelman F. (1991). Cancer Genetics & Cytogenetics, 55, 11 ± 18. Pikkarainen T, Eddy R, Fukushima Y, Byers M, Shows T, Pihlajaniemi T, Saraste M and Tryggvason K. (1987). Journal of Biological Chemistry, 262, 10454 ± 10462. Pugh CW, Tan CC, Jones RW and Ratclie PJ. (1991). Proceedings of the National Academy of Sciences of the United States of America, 88, 10553 ± 10557. Rein MS, Friedman AJ, Barbieri RL, Pavelka K, Fletcher JA and Morton CC. (1991). Obstetrics & Gynecology, 77, 923 ± 926. Ron D and Habener JF. (1992). Genes & Development, 6, 439 ± 453. Sait SN, Dal Cin P, Ovanesso S and Sandberg AA. (1989). Cancer Genetics & Cytogenetics, 37, 157 ± 161. Sargent MS, Weremowicz S, Rein MS and Morton CC. (1994). Cancer Genetics & Cytogenetics, 77, 65 ± 68. Scherer SW, Neufeld EJ, Lievens PM, Orkin SH, Kim J and Tsui LC. (1993a). Genomics, 15, 695 ± 696. Scherer SW, Rommens JM, Soder S, Wong E, Plavsic N, Tompkins BJ, Beattie A, Kim J and Tsui LC. (1993b). Human Molecular Genetics, 2, 751 ± 760. Scherer SW and Tsui, L-C. (1991). Advanced Techniques in Chromosome Research. Adolph K (ed). Marcel Dekkar Inc: New York, pp. 33 ± 72. Schuster SJ, Badiavas EV, Costa-Giomi P, Weinmann R, Erslev AJ and Caro J. (1989). Blood, 73, 13 ± 16. Schweinfest CW, Henderson KW, Suster S, Kondoh N and Papas TS. (1993). Proceedings of the National Academy of Sciences of the United States of America, 90, 4166 ± 4170. Semenza GL, Koury ST, Nejfelt MK, Gearhart JD and Antonarakis SE. (1991a). Proceedings of the National Academy of Sciences of the United States of America, 88, 8725 ± 8729. Semenza GL, Nejfelt MK, Chi SM and Antonarakis SE. (1991b). Proceedings of the National Academy of Sciences of the United States of America, 88, 5680 ± 5684.

Semenza GL and Wang GL. (1992). Molecular & Cellular Biology, 12, 5447 ± 5454. Shoemaker CB and Mitsock LD. (1986). Molecular & Cellular Biology, 6, 849 ± 858. Simeone A, Acampora D, Pannese M, D'Esposito M, Stornaiuolo A, Gulisano M, Mallamaci A, Kastury K, Druck T, Huebner K et al. (1994). Proceedings of the National Academy of Sciences of the United States of America, 91, 2250 ± 2254. Solberg R, Sistonen P, Traskelin AL, Berube D, Simard J, Krajci P, Jahnsen T and de la Chapelle A. (1992). Genomics, 14, 63 ± 69. Sourla A, Polychronakos C, Zeng WR, Nepveu A, Kukuvitis A, Naud F and Koutsilieris M. (1996). Cancer Research, 56, 3123 ± 3128. Swansbury GJ, Lawler SD, Alimena G, Arthur D, Berger R, Van Den Berghe H, Bloom®eld CD, de la Chappelle A, Dewald G, Garson OM, Hagemeijer A, Mitelman F, Rowley JD and Sakurai M. (1994). Cancer Genet. Cytogenet, 73, 1 ± 7. Taguchi T, Testa JR, Papas TS and Schweinfest C. (1994). Genomics, 20, 146 ± 147. Timpl R, Rohde H, Robey PG, Rennard SI, Foidart JM and Martin GR. (1979). Journal of Biological Chemistry 254, 9933 ± 9937. Thomas PS. (1980). Proc. Natl. Acad. Sci. USA.,77, 5201 ± 5205. Vanni R, Lecca U and Faa G. (1991). Cancer Genetics & Cytogenetics, 53, 247 ± 256. Wilcox LS, Koonin LM, Pokras R, Strauss LT, Xia Z and Peterson HB. (1994). Obstetrics & Gynecology, 83, 549 ± 555. Wooster R, Cleton-Jansen AM, Collins N, Mangion J, Cornelis RS, Cooper CS, Gusterson BA, Ponder BA, von Deimling A, Wiestler OD et al. (1994). Nature Genetics, 6, 152 ± 156. Yunis JJ, Lobell M, Arnesen MA, Oken MM, Mayer MG, Rydell RE and Brunning RD. (1988a). British Journal of Haematology, 68, 189 ± 194. Yunis JJ, Lobell M, Arnesen MA, Oken MM, Mayer MG, Rydell RE and Brunning RD. (1988b). British Journal of Haematology, 68, 189 ± 194.

2365