<<

(1997) 15, 2773 ± 2781  1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

Human 7 carries a putative tumor suppressor (s) involved in choriocarcinoma

Takao Matsuda1,5,6, Masahiro Sasaki2,6, Hidenori Kato1, Hideto Yamada2, Maurice Cohen3, J Carl Barrett4, Mitsuo Oshimura2 and Norio Wake1

1Department of Reproductive Physiology and Endocrinology, Medical Institute of Bioregulation, Kyushu Univ., Beppu, Oita 874, Japan; 2Department of Molecular and , School of Science, Faculty of , Tottori University, Yonago, Tottori 683, Japan; 3Department of Corporate Molecular , Abbott Laboratories AP 20/5, Abbott Park, IL. 60064, USA; 4Laboratory of Molecular Carcinogenesis, National Institute of Environmental Health Sciences, National Institute of Health, P.O. Box 12233, Research Triangle Park, 27709, USA

Choriocarcinoma developed from a complete hydatidi- birth (Park, 1971). A complete hydatidiform mole, form mole has an unique genetic feature that involves which is characterized by grossly swollen villi in the monoallelic contribution from the paternal . To absence of a fetus, is androgenetic in origin. The entire determine the chromosome carrying putative tumor genome of the molar conceptus is paternally derived suppressor gene(s), microcell-hybrids were isolated (Kajii and Ohama, 1977; Wake et al., 1978a, b; Jacobs et following fusion of choriocarcinoma cells with microcells al., 1981). The majority of moles results from from A9 cells containing a single fertilization of an devoid of nuclei (an empty egg) chromosome (1, 2, 6, 7, 9 or 11). Microcell-hybrids with by a haploid . The paternally derived haploid set the introduction of were suppressed or then duplicates without cytokinesis and restores modulated for tumorigenicity and exhibited altered in diploidy (Jacobs et al., 1981; Yamashita et al., 1979). vitro growth properties. Introduction of 1, Invariably, this class of moles has a 46, XX 2, 6, 9 or 11 had no e€ect. Tumorigenic revertants and is completely homozygous for genetic markers. isolated from microcell-hybrids with the introduced Fertilization of an empty egg by two spermatozoa is chromosome 7 contains reduced numbers of chromosome responsible for the remaining cases (Ohama et al., 1981; 7. These ®ndings suggest that chromosome 7 contains a Pattillo et al., 1981; Surti et al., 1979). This mole is putative (s) for choriocarcinoma. heterozygous at some genetic loci but homozygous at Alterations in tumorigenic seen in microcell- others. It is assumed that genetic features seen in hybrids were not associated with the presence of either complete moles associate with their potential to progress ERV3 or H-plk located on the introduced to choriocarcinomas. The monoallelic contribution chromosome 7, indicating the putative tumor suppressor would render a certain gene susceptible to functional gene(s) is outside of ERV3 and H-plk gene loci. inactivation by `one hit' kinetics. Alternatively, unipar- Furthermore, we obtained evidence to de®ne a critical ental transmission of that are subject to parental region on chromosome 7 (7p12-7q11.23) that was imprinting in human would impair their regulation frequently lost in surgically removed choriocarcinoma (Scable et al., 1989). Thus, some tumor suppressor genes tissues and cell lines. Using a panel of microsatellite that are inactivated by one of the two previously markers, biallelic deletions were observed, which strongly mentioned mechanisms would play an important role suggests the presence of a tumor suppressor gene(s) for the high propensity to malignancy seen in complete within this critical region. moles. However, a unique characteristic of the mole has prevented exploration of genetic mechanisms associated Keywords: tumor suppressor gene; choriocarcinoma; with choriocarcinoma development. Many tumor chromosome 7; homozygous suppressor genes are inactivated by intragenic muta- tions in one , accompanied by the loss of a chromosomal region containing the remaining allele, termed loss of heterozygosity (Cavanee et al., 1983; Introduction Vogelstein et al., 1989; Yokota et al., 1987). Mapping such deleted regions has been done to identify sequences Choriocarcinoma is a highly invasive tumor consisting involved in malignancies. However, the method is not of markedly anaplastic trophoblasts totally lacking suitable for disclosing the genetic alternations in residual villous structures. However, little is known of choriocarcinomas because monoallelic contribution in the molecular mechanisms underlying the development the complete mole and its putative transformant should of this tumor. The tumor may originate from the mask the allelic loss. To resolve this diculty, we ®rst trophoblasts of any conception. Although malignant identi®ed a speci®c human chromosome on which complications infrequently occur after abortion, or after putative tumor suppressor gene is located, using an apparently normal live birth, the risk of choriocarci- microcell-mediated chromosome transfer. A transfer of noma associated with a complete mole is 2000 to 4000 chromosome 7 via microcell fusion clearly suppressed times greater than that of an apparently normal live tumorigenicity and in vitro growth properties of choriocarcinoma cells. Then, using a panel of micro- markers we found a homozygous deletion in the Correspondence: N Wake 5Research Fellow of the Japan Society for the Promotion of Science chromosome 7p12-7q11.23 region that was frequent in 6T Matsuda and M Sasaki contributed equally to this manuscript surgically removed choriocarcinoma tissues and cell Received 10 June 1997; revised 25 July 1997; accepted 28 July 1997 lines. These ®ndings should help positional of Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2774 the target tumor suppressor gene(s) involved in cutaneous inoculation of 16107 cells into nude mice. choriocarcinoma development. An average number of days until the long axis of the tumor to reach 10 mm (PMD) was 26 days. All of the microcell- clones containing a single chromo- some 1, 2, 6, 9 and 11 were also highly tumorigenic; the Results tumor-take incidence and the PMD ranged from 83 to 100%, and from 16 to 34 days in these clones. The A Single chromosome transfer into choriocarcinoma cells di€erence was not statistically signi®cant between the We introduced human chromosomes into CC1 cells and the microcell-hybrids. CC1 choriocarcinoma cells via microcell fusion to This was in sharp contrast with the suppression of identify the chromosome carrying putative tumor tumorigenic potentials in microcell-hybrid clones with suppressor gene(s). The karyotype of parent CC1 cells a transferred chromosome 7. We obtained 4 microcell- was unstable and the total number of chromosomes hybrid clones that contained the transferred chromo- ranged from 80 to 95 with 25 ± 30 rearranged marker some 7. Complete suppression of tumorigenicity was chromosomes. In two to three successive fusion observed in one clone (#7-1), from which no tumor was experiments with microcells prepared from mouse produced for up to 100 days following inoculation of cells with di€erent human chromosome (eq A9 16107 cells. Tumorigenic potentials were partially (), A9 (neo2), A9 (neo6), A9 (neo7), A9 (neo9), suppressed in the remaining 3 clones (#7-2, #7-3 and and A9 (neo11) cells), 2 to 8 G418 resistant CC1 #7-4). Reduced tumor-take incidence (18 to 59%) and microcell-hybrids were randomly isolated, propagated prolonged PMD (47 to more than 80 days) were in the presence of G418 for further analyses, and demonstrated in these clones. We obtained cells karyotyped. These clones were then examined for an (tumorigenic revertants) from tumors produced by the excess number of transferred chromosomes. As the inoculation of these clones to determine factors that parental CC1 cells contained complicated , may explain the variation in tumorigenic potential we could not con®rm by karyotypic analysis either the among the chromosome 7 transferred microcell-hybrid presence or absence of the transferred chromosome. clones. These revertants were highly tumorigenic, Thus, we carried out chromosomal ¯uorescence in situ producing progressively growing tumors with 75 ± hybridization (FISH), using the pSV2-neo as a probe 100% tumor-take incidence. The microcell-hybrid to demonstrate the successful transfer of the donor clones contained three to four copies of chromosome chromosome into CC1 cells. The introduced chromo- 7, whereas two or three copies of chromosome 7 were somes tagged with the pSV2 neo gene were identi®ed in observed in the parent CC1 cells (Figure 1). However, each microcell-hybrid clone (Figure 1). The chromoso- reduction of a single copy of chromosome 7 seemed location of signals corresponded to the integration apparent in three tumorigenic revertant cells, compared site of the pSV2-neo gene observed (Koi et al., 1989). to the copy number of chromosome 7 in the microcell- We evaluated tumorigenicity and in vitro trans- hybrid clones (date not shown). formed phenotypes in these microcell-hybrids and MspI-digested DNA from various cells were made comparisons with the parental CC1 cells (Table hybridized with a 32P-labeled ERV3 subclone 1). CC1 cells formed progressively growing tumors (pHP1.7) which showed a 3.3 kb restriction fragment with 100% tumor-take incidence following the sub- in the parent CC1 cells and 2.8 kb restriction fragment

Table 1 Tumorigenicity and in vitro properties of parental CC1 cells and microcell hybrids with an introduced chromosome Tumorigenicity The average Eciency of Cell No. of day when *soft agar tumors/No. of the long axis growth (%) doubling inculation sites reached 10 mm (SAE/PE) time (h) Parental cell CC1 10/10 (100%) 26 38.8** 21.5** Microcell hybrid clones (introduction of Ch.1) #1 ± 1 6/7 (86%) 16 26.6 26.5 #1 ± 2 6/6 (100%) 23 79 26.2 Microcell hybrid clones (introduction of Ch.2) #2 ± 1 6/6 (100%) 16 67.6 26.1 #2 ± 2 8/8 (100%) 31 110 22.3 Microcell hybrid clones (introduction of Ch.6) #6 ± 1 6/6 (100%) 19 67.0 34.9 Microcell hybrid clones (introduction of Ch.7) #7 ± 1 0/9 (0%) 480 8.5 59.5 #7 ± 2 2/11 (18%) 70 4.9 80.2 #7 ± 3 3/10 (30%) 47 2.2 52.0 #7 ± 4 7/12 (59%) 49 14.1 46.3 Microcell hybrid clones (introduction of Ch.9) #9 ± 1 5/6 34 67.0 29.9 #9 ± 2 5/5 (100%) 24 67.0 29.1 Microcell hybrid clones (introduction of Ch.11) #11 ± 1 2/2 (100%) 32 32.7 28.9 *The eciency was calculated as the member of colonies in soft agar divided by eciency. **The means of triplicate experiments Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2775 in the donor A9 (neo7) cells (Figure 2). Both bands chromosome 7. The growth of these chromosome 7 were observed in MspI-digested cellular transferred microcell-hybrids with JEG3 was markedly obtained from 4 microcell-hybrids with the transferred reduced as was observed with CC1 cells (data not chromosome 7 as well as from their tumorigenic shown). These results indicated that a single human revertants. These ®ndings suggest that the tumorigenic chromosome 7 derived from normal ®broblasts had the revertants retained the donor chromosome's ERV3 potential to suppress or to modulate the tumorigenicity locus and that this region is not involved in the and the in vitro growth properties of human tumorigenic suppression. choriocarcinoma cells. This potential was clearly not In vitro growth properties of the microcell-hybrids shared by the human chromosome examined here. that contained chromosomes 1, 2, 6, 9 or 11 were Thus the observation is compatible with the idea that similar to those of the parent CC1 cells (Table 1). chromosome 7 carries putative tumor suppressor However, signi®cant prolongation of population gene(s) that negatively regulate choriocarcinoma cell doubling time and suppression of soft agar growth growth. eciency were demonstrated in the microcell-hybrid clones that contained a transferred chromosome 7. H-plk sequences not involved in tumorigenic suppression Morphology of the microcell-hybrid cells was remark- ably altered (Figure 3). A human ERV3 that locates on human In order to test whether chromosome 7 a€ected the chromosome 7 is abundantly expressed in the tumorigenicity of other choriocarcinoma cells, we placental chorionic villi throughout gestation. Cohen transferred chromosome 7 into another choriocarcino- et al., (1988) and Kato et al., (1988) demonstrated the ma cell line (JEG-3). All clones were a€ected by this complete absence of ERV3 in choriocar- chromosome. One clone showed a complete loss of cinomas or invasive moles. There are three env- tumorigenic potential, and three other clones showed a containing ERV3 mRNAs of 9, 7.3 and 3.5 kb. reduction in the tumor-take incidence and an increased Although the 3.5 kb mRNA contains only proviral latency for tumor growth from 23 days for the parental sequences, the 9 and 7.3 kb mRNAs extend to a splice cells to 58 ± 217 days for clones with an introduced donor site downstream from the provirus (Kato et al., 1987). As a result, these involve nonviral genomic sequences at their 3'-ends due to . The transcripts a Kruppel-related zinc ®nger (H-plk). The structural with Drosophila Kruppel that is essential for normal embryonic development suggests a putative function as a (Kato et al., 1990; Bellefroid et al., 1989). The failure to detect 9 and 7.3 kb transcripts in addition to 3.5 kb isolated from choriocarcinomas led to the speculation that the absence of H-plk protein may be the primary defect associated with choriocarcinoma development (Kato et al., 1990). This hypothesis is compatible with our results that the chromosome 7 introduced into choriocarcinoma cells can negatively regulate cell growth. We then asked whether the H-plk encoded from the transferred chromosome 7 is responsible for modulation of tumorigenicity and in vitro growth Figure 1 Fluorescence in situ hybridization (FISH) of CC1#7 properties of choriocarcinoma cells. As a control showing the presence of two copies of transferred chromosome 7, experiment, Northern blots of 8 placental using tagged with pSV2neo DNA, in the microcell hybrid clone (CC1#7-3). An arrow indicates 2 introduced chromosome 7 with the ERV3 probe detected 9, 7.3 and 3.5 kb mRNAs ¯uorescence signals from the integrated pSV2neo

1 2 3 4 5 6 7

3.3 —

2.8 — (kb)

Figure 2 Southern blot analysis showing the introduction of chromosome 7. DNAs were digested with Msp-I and hybridized with the pHP 1.7 (ERV3 env fragment) located on human chromosome 7. Lane 1, A9 (neo7), Lane 2, parental CC1, Lanes 3 and 4, CC1#7-2 and CC1#7-3, Lanes 5 and 6, tumorigenic Figure 3 Phase contrast photomicrographs of cells in culture: revertant cells from CC1#7-2 and CC1#7-3, Lane 7, parental CC1 (right) CC1; (left) CC1#7-3 (64X) Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2776 that were characteristic of ERV3 transcription. As ing the homozygous deletion. Homozygous deletions expected, the probe failed to detect these transcripts in occurred in at least one locus on chromosome 7, in 10 RNAs obtained from each 4 choriocarcinoma tissues of 14 tumors (71%). The incidence of homozygous and cell lines (Figure 4). Recovery of 3.5 kb mRNA deletion was unexpectedly high and thus eight expression was demonstrated in the microcell-hybrid choriocarcinoma cell lines were further examined to containing chromosome 7. This contrasted with the con®rm the involvement of a homozygous deletion on complete abrogation of 9 and 7.3 kb mRNAs chromosome 7 in choriocarcinomas. CC2 choriocarci- transcription in the hybrids. These results were noma cells developed from a complete mole whereas con®rmed using RT-PCR (Figure 4b). Primer pairs the remaining 7 cells were derived from a normal were designed for the ERV3 env region and the H-plk fertilization product. We did not obtain DNAs from (ORF), respectively (Figure 4c). the patients or their partners. All eight cell lines lost a The PCR of cDNAs obtained from 8 placentae particular region on chromosome 7 biallelically, which ampli®ed 420 and 400 bp fragments that were is signi®cantly higher than the incidence of homo- compatible with the size predicted from the ERV3 env and H-plk ORF gene sequences. Either trace levels or no PCR products was detectable in 8 choriocarci- nomas. The ampli®ed cDNAs obtained from the a n

microcell-hybrids containing chromosome 7 showed o i s

no 400 bp fragment encoded from H-plk gene u f

l l sequences, whereas the 420 bp from ERV3 env region a t e c n

choriocarcinoma i was readily detected. As a result, transfer of a e c n i a

chromosome 7 into CC1 choriocarcinoma cells did l tissue cell line m p

not lead to the recovery of H-plk transcription. It is case 7 o # G 3 1 1 unlikely that the deletion of a transferred chromosome w J C e C C

7 from the hybrids corresponded to the absence of H- 9 10 11 12 N C B C C plk transcription because the ERV3 cDNA was readily 9.0 kb ampli®ed. Thus, H-plk transcription apparently did not 7.3 kb contribute to the potential of a chromosome 7 to suppress tumorigenicity and in vitro growth properties of CC1 cells. 3.5 kb

Homozygous deletion of chromosome 7

Because DNA samples from the preceding complete b mole were not available, we used constitutional DNAs (RT-PCR) from the patient and her sexual partner as paired ERV3-1.2 420 bp (RT-PCR) 400 bp samples of choriocarcinoma. Despite the rarity of this H-plk tumor, we were able to collect 14 surgically resected choriocarcinoma samples that were sequel to the c complete mole. We used 25 PCR-ampli®ed microsa- tellite repeats to screen the 14 choriocarcinomas for rearrangements or deletions on chromosome 7 (Figure 5). Choriocarcinomas, like many solid tumors, contain expanded necrotic tissues and various numbers of nonneoplastic cells that could mask genomic altera- tions occurring within the neoplastic tumor cell population. A cryostat sectioning method for physi- cally fractioning such tumors to enrich for neoplastic cells was used. DNAs from 8 control placentae revealed intact Figure 4 (a) Northern blots using ERV3-env as a probe. Placenta microsatellite loci by PCR ampli®cation of the marker expressed abundant 9, 7.3 and 3.5 kb mRNAs. The probe failed to fragments using published (Tsui, 1995; Weissenbach et detect these transcripts in 8 choriocarcinomas. However, 3.5 kb al., 1992) forward and reverse primers (data not transcripts were detectable in the nontumorigenic microcell hybrid shown). In turn, electrophoresis of PCR products run containing a chromosome 7 (CC1#7). (b) Expression analyses of on a 6% denaturing polyacrylamide gel demonstrated ERV3 and H-plk using RT-PCR. The PCR of cDNA from placenta ampli®ed both 420 bp ERV3 env and the 400 bp H-plk biallelic loss of part of chromosome 7 in choriocarci- fragments, whereas any fragments were undetectable in 8 noma samples. Con®rmation of homozygous deletions choriocarcinomas. The 420 bp fragment that was compatible with was made by multiplex PCR ampli®cation of two ERV3 env was ampli®ed, but the 400 bp H-plk fragment was independent STS markers. One STS content could not undetectable in the microcell hybrid containing a chromosome 7(CC1#7) that had lost tumorigenicity. Samples loaded on the gel be ampli®ed while ampli®cation of the other did occur are indicated in the above of the ®gure. Arrows on the Northern in these choriocarcinoma specimens. The results were blot indicate transcript size. Arrows in the lower ®gure indicates further con®rmed in parallel ampli®cations of DNAs 420 bp (ERV3 env) and 400 bp (H-plk) fragments ampli®ed by derived from leukocytes for the patients and partners. RT-PCR. Control GAPDH cDNAs were ampli®ed in all samples The STS contents were absent from tumor DNA (data not shown). (c) A map of ERV3 and H-plk gene. Position of the pHP 1.7 probe is indicated. The primer pairs for ERV3 env samples but present in constitutional DNA samples and H-plk ORF are indicated by arrows. The primer pairs were from the patients and their partners, thereby confirm- designed for the speci®cally ampli®ed region of H-plk ORF Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2777

tissues cell line

case 8 4 5 6 7 1013 1 - o C 1 G 3 R w J U e C C A

F M T B C N C J N lane 1 2 3 4 5 6 7 8 9 101112131415

D7S663

D7S520

PCR –Southern D7S663 D7S520

lane 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 tissues cell line Figure 5 Biallelic losses at the D7S663 and D7S520 loci on chromosome 7. Representative autoradiographs at the D7S663 and D7S520 loci indicate homozygous deletions in the choriocarcinomas. Sources of DNA are shown in the following; 1-DNA obtained from peripheral blood of the case 8 husband, 2- DNA obtained from peripheral blood of the case 8 patient, 3- DNA obtained from case 8 choriocarcinoma tissues, 4-9-DNAs from cases 4, 5, 6, 7, 10 and 13 choriocarcinoma tissues, 10-15- Figure 6 Map of homozygous deletions on chromosome 7. DNA DNAs from cell lines of Bewo, CC1, NJG, CC3, JAR and NUC- samples from 14 choriocarcinoma tissues and 8 choriocarcinoma 1. The absence of any PCR products at D7S663 locus was evident cell lines were examined for biallelic losses of STS markers located in case 8 choriocarcinomas. However, ampli®ed fragments were on chromosome 7. Closed squares represent homozygous evident in constitutive DNAs obtained from the case 8 husband deletions but blanked squares mean the retention of a single or and the patient. Similarly, the PCR product was undetactable in both . Hatched squares mean that the locus indicated was DNAs from case 4, 6, 7, choriocarcinomas and Bewo, CC1 and not tested. The physical locations of STS markers are based on JAR cells. DNA fragments at D7S520 failed to be ampli®ed in data described tentatively by Tsui, 1995; Ohnishi et al., 1996). The case 4 and Bewo and CC1 cells. Both fragments at D7S663 and smallest common deleted region is enclosed between D7S520 and D7S520 were readily ampli®ed in DNAs from cases 5, 10 and 13 D7S663 (7p12-7q11.23) choriocarcinoma tissues, and NJG, CC3 and NUC-1 cells. The homozygous deletion was con®rmed in repeated multiplex PCR. The homozygous deletion shown by electrophoresis on 7M denaturing acrylamide gel (upper) were further con®rmed by electrophoresis on 2% agarose X gel (lower) Discussion

We have adopted a new strategy to search for the locus of tumor suppressor gene(s) associated with chorio- zygous deletions in various malignancies (Ohnishi et carcinoma. This was necessary because tumor samples al., 1996; Kastury et al., 1996). are scarce and the contribution that is characteristic of In summary, 18 choriocarcinomas had interstitial the preceding complete mole is monoallelic. To deletions on chromosome 7. Overlapping deletions of maximize validity of the genetic investigation, we the samples showing interstitial losses indicated that directed attention to modulation of tumorigenicity the smallest common deleted region was enclosed and in vitro growth properties of choriocarcinoma between D7S520 and D7S663 (7p12-7q11.23). The cells by introducing a normal chromosome via most frequent deletion among 25 markers was microcell fusion. The expression of genes locating on observed with D7S502 and D7S482. In addition, the the transferred chromosome is regulated by the ERV3 proviral genome that was also mapped to the physiological control of chromosomal sequences. 7p12-7q11.23 was lost in 6 or 14 tumor samples, Thus, microcell-mediated chromosome transfer proce- whereas all 8 choriocarcinoma cell lines were retained dures have been used successfully to introduce speci®c (Figure 6). None of the samples studied showed chromosomes to test for the suppression of tumori- deletions at any other microsatellite loci with even genic phenotypes of cell lines derived from various the smallest common deleted region retaining intact. human malignancies (Stanbridge, 1988; Yamada et al., These results were con®rmed by repeated multiplex 1990; Zenklusen et al., 1994a,b,c; Oshimura et al., PCR reactions followed by electrophoresis on a 2% 1990; Saxon et al., 1986; Tanaka et al., 1991; Porteous agarose X gel. By analogy to deletions found in various et al., 1989; Weissman et al., 1987; Conway et al., malignancies, the homozygous deletions observed here 1992). We generated and characterized microcell- might possibly include a tumor suppressor gene, the hybrids containing di€erent individual chromosomes. loss or inactivation of which would play a role in The tumorigenicity and in vitro growth of the initiation or progression of choriocarcinoma. microcell-hybrids with transferred chromosomes 1, 2, Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2778 6, 9 or 11 showed that the introduction of a human 1996; Boyd et al., 1993; Schutte et al., 1995; Roche et chromosome into a choriocarcinoma cell line did not al., 1996; Decker et al., 1996). The monoallelic a€ect its tumorigenicity or proliferative potential to contribution could contribute to the high incidence of any signi®cant extent. However, all 4 microcell-hybrid homozygous deletions in choriocarcinomas. However, clones that contained a transferred chromosome 7 were all of the seven choriocarcinoma cell lines that clearly suppressed both in tumorigenicity and parti- developed from normal placentae also contained a cular properties of in vitro growth. In addition, deletion of biparental alleles in this critical region. A tumorigenic revertants obtained from these microcell- possible analogous situation exists with the tumor hybrid clones lost major parts of transferred chromo- suppressor gene, wherein a variety of human malig- some 7 even when the donor chromosome's ERV3 nancies have a homozygous deletion (Ohnishi et al., locus was retained. The consistency between these two 1996). Interstitial deletions may represent the main sets of data led to the notion that a putative tumor mechanism to inactivate the putative tumor suppressor suppressor gene(s) probably locates on chromosome 7. gene on chromosome 7. If such is indeed the case, Some gene products of murine endogenous retro- additional cases of choriocarcinoma harboring intra- , but not intact viruses, are involved in the genic in this a€ected region may have gone pathogenesis of malignant diseases (Martin et al., undetected. 1981). Among them, a human ERV3 provirus and its The homozygous deletions identi®ed here maps to related H-plk sequences are localized on chromosome 7 chromosome 7p12-7q11.23. The region that is proximal (O'Connell et al., 1984; Bellefroid et al., 1992). The to the is bordered by markers D7S520 and alternative splicing produces the transcripts that encode D7S663. Although a number of microsatellite markers cellular H-plk sequences (Kato et al., 1990; Bellefroid have been mapped to the region, their precise location et al., 1989). remains to be determined. Thus, determination of the Extinction of H-plk transcription has been suggested physical order of markers is necessary to accurately as the primary defect involved in choriocarcinoma determine the deletion pattern. The presence of tumor development. However, the absence of H-plk transcrip- suppressor gene on chromosome 7 has been suggested tion in nontumorigenic microcell-hybrids containing an for various . Deletions in 7q with breakpoints introduced chromosome 7 excluded the H-plk in varying from q11 to q34 have been described in choriocarcinoma suppression. ERV3 itself has a long esophageal and gastric cancers (Kuniyasu et al., 1994; ORF in the env region (O'Connell et al., 1984). Thus, Atkin and Baker, 1993). Recent investigations using the present result does not completely exclude the VNTR or microsatellite markers described high involvement of ERV3 env region in choriocarcinoma, frequencies of allelic deletion in 7q in colon as the possibility exists, although unlikely, that (Zenklusen et al., 1995), head and neck (Speicher et transcript from the donor chromosome's ERV3 gene al., 1995), gastric (Kuniyasu et al., 1994), pancreatic was present in the nontumorigenic microcell-hybrids (Achille et al., 1996), prostate (Atkin and Baker, 1993; but absent in their tumorigenic revertants. Preferential Zenklusen et al., 1994c), ovary (Atkin and Baker, localization of ERV3 mRNAs in syncytiotrophoblasts 1993), breast cancers (Zenklusen et al., 1994a; Bieche et contributes to the observed high expression in placenta al., 1992) leiomyosarcoma (Sreekantia, 1993), and non (Boyd et al., 1993; Venables et al., 1995). This tumorigenic immortal human cell line SUSM-1 (Ogata expression correlates with the fusion process itself. et al., 1993). However, the commonly deleted region in The lack of ERV3 transcription in choriocarcinoma these cancers is localized at 7q31.1-32, which is more most probably corresponds to the scarcity of such distal than the a€ected region detected here. Ogata et fused cells because choriocarcinoma is predominantly al. (1993) showed that cellular senescence of 2 non- composed of intermediate trophoblasts and cytotro- tumorigenic human immortalized-®broblast cell lines phoblasts. This is compatible with the possibility that, was overcome by transfer of chromosome 7. However, outside of ERV3 and its related H-plk gene loci, at it remains to be determined whether the cellular least one target tumor suppressor gene is localized on senescence gene and the tumor suppressive gene in chromosome 7. choriocarcinomas are causally related to the same gene We have in fact obtained evidence to de®ne the on chromosome 7. Homozygous deletion detected in critical region that is lost biallelically in choriocarci- the present study should provide an invaluable tool for nomas by using a panel of microsatellite markers positional cloning of the target tumor suppressor gene located on chromosome 7. It is probable that some at 7p12-7q11.23. A YAC library construction is now in genetic alterations that occur do not a€ect cell progress for the cloning of this putative tumor growth but represent errors that become ®xed during suppressor gene. clonal expansion. However, the unexpectedly high incidence and the involvement of more than one contiguous marker in the deleted region strongly Materials and methods suggest that the homozygous deletion detected here is important and de®ne locus of the putative tumor Specimens and cells suppressor gene. Although complete data are not available, the occurrence of deletions involving both Specimens: Eight normal placental tissues were obtained alleles is infrequent. Because of the total loss of at delivery. Fourteen choriocarcinoma tissues, all of which particular genetic information, the cellular e€ect of were sequel to the complete mole, were removed surgically. Each specimen was immediately frozen in most homozygous deletions is assumed to be deleter- liquid nitrogen and stored at 7708C. A section made ious. Indeed, the homozygous deletions reported to from each carcinoma sample was stained with hematox- date are relatively small and have been noted only in a ylin/eosin, following microscopic con®rmation by a few malignancies (Ohnishi et al., 1996; Kastury et al., registered pathologist. Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2779 Cells lines: The choriocarcinoma cell lines, Bewo, CC1, After washing, another layer of FITC-avidin was added for CC2, CC3, NUC-1, JEG3, JAR and NJG were kindly ampli®cation. The slides were washed and mounted in provided by Drs S Sekiya, (Chiba Univ.), K Tanaka, (Niigata antifade solution [1% diazabicyclo-octane in glycerol with Univ.), S Nozawa, (Keio Univ.), Y Tomoda, (Nagoya Univ.), 10% -bu€ered saline] containing 0.75 mg/ml of and K Suzumori (Nagoya City Univ.). A CC2 cell line was propidium iodide. Signals were observed and photographed derived from a complete mole whereas the other choriocarci- under a ¯uorescence microscope (Nikon) with Ektachrome noma cell lines were established from choriocarcinomas ®lm (Kodak, ASA100). developed from an apparently normal live birth (Bewo, NUC1, JEG3, JAR, NJG), stillbirth (CC3) or abortion Evaluation of tumorigenicity and in vitro transformed properties (CC1). Growth properties and tumorigenicity of these cells have been documented (Pattillo and Gey, 1968; Pattillo et al., Tumorigenicity in vivo was evaluated following the dorsal 1971; Suzumori et al., 1983; Kohler and Bridson, 1971; subcutaneous inoculation of 1610 7 cells per site in 4- to 9- Sugiura et al., 1988; Nozawa et al., 1987). These cell lines week-old athymic ICR nu/nu mice. The were were maintained in Dulbecco's modi®ed Eagle's medium observed for 100 days. Both tumor-take incidence and the (DMEM) or F12 supplemented with 10% fetal bovine serum that required for a tumor to grow to a maximum

(FBS) at 378 C in a humidi®ed 5% CO2 atmosphere. The diameter of 10 mm (PMD) were recorded. For calculation absence of micoplasma was con®rmed in the cell lines before of the number of cell population doublings, cells from the they were subjected to the following experiments. parent CC1 and microcell-hybrids were plated in triplicate in DMEM containing 10% FBS on 24-well dishes at a density of 56104 cells. Media were changed at 3-day Chromosome transfer via microcell fusion intervals, and the numbers of cells was counted for the Mouse A9 cells carrying a signal copy of normal human following 1 ± 10 days. For evaluation of anchorage chromosome tagged with the neomycin phosphotransferase independent growth ability, a single cell suspension of gene (pSV2-neo) were used as donor line for microcell parent CC1 cells and microcell hybrids (16103) in 0.3% mediated chromosome transfer. A9 (neo 1), A9 (neo 2), A9 agar was overlaid on 0.5% basal agar, followed by 4 weeks (neo 6), A9 (neo 7), A9 (neo 9) and A9 (neo 11) retained of observation. Colonies with a diameter of 4100 mm were normal human , 2, 6, 7, 9 and 11, scored. respectively. The details of the procedure have been described previously (Koi et al., 1989). A9 cell clones DNA and RNA isolation containing a human chromosome were incubated with 0.05 mg/ml colcemid in DMEM supplemented with 20% Genomic DNA was isolated from placentae and chor- FBS for 48 h. The culture ¯asks were then ®lled with iocarcinoma tissues, and choriocarcinoma cell lines by medium containing 10 mg/ml cytochalasin B and centri- proteinase K and SDS treatment followed by phenol and fuged in a Sorvall GSA ®xed-angle rotor at 10 000 g for chloroform extraction (Sambrook et al., 1982). Tumor 60 min at 348C. The obtained microcells were suspended in samples were puri®ed by microdissection of cryostat serum-free medium and puri®ed by sequential polycarbo- sections of tissues to minimize contamination of normal nate ®ltration (Nucleopore). The puri®ed microcells were tissues. RNA was isolated by acid-guanidine method overlaid and attached to recipient CC1 cells with 100 mg/ml (Sambrook et al., 1982) and was treated with RNase-free phytohemagglutinin. Microcell fusion were done by placing DNase, extracted with phenol/chloroform and precipitated 50% polyethylene glycol 1500 (Boehringer Mannheim, with ethanol. Germany) solution over the cell layer for 1 min. After 24 h incubation in DMEM containing 10% FBS, G418 at 400 mg/ml was added. G418 resistant colonies were isolated Southern hybridization after the 21 ± 60 days of selection. Genomic DNA from the parent CC1, microcell-hybrids containing chromosome 7 and their tumorigenic revertants were isolated by lysis in 10 mM Tris-HCl (pH 7.5)/10 mM Chromosomal ¯uorescence in situ hybridization EDTA/2% SDS with proteinase K at 200 mg/ml, incubated The hybridization protocol followed that of Cherif et al. for 3 h at 508C, puri®ed by phenol and chloroform (1989), with slight modi®cations. cells on glass extraction, and precipitated with ethanol. Southern blot slides were prepared according to standard procedures, analysis was carried out using restriction fragments labeled denatured in 70% formamide/26SSC at 708C for 2 min, to high speci®c activity (Rediprime Kit, Amersham, UK) then dehydrated through cold 70, 90, and 100% ethanol as a hybridization probe. The pHP 1.7 plasmid DNA for 5 min each. The probe pSV2neo was labeled by nick which is located in chromosome 7 was used for the with biotin-16-dUTP for 1.5 h at 158C. The detection of the transferred chromosome 7 (O'Connell et labeled probe was puri®ed by ethanol precipitation, al., 1984). dissolved in 20 ml of 100% formamide, and denatured at 758C for 10 min. Twenty microliters of hybridization Northern blot and polymerase chain solution (bovine serum albumin (BSA; 20 mg/ml), reaction (RT-PCR) 106SSC, and 50% dextran sulfate, 1 : 2 : 2) and 20 ml of labeled probe DNA were placed in an Eppendorf tube. The Northern blot hybridization analysis was performed with solution (20 ml/slide) was placed drop-wise on the poly(A)+ RNA. Five micrograms of poly(A)+RNA were denatured chromosomes, covered with Para®lm, and electrophoretically separated on 1% agarose containing incubated overnight at 378C in a humidi®ed chamber. 15% formaldehyde, transferred to a nylon membrane, After hybridization, the slides were washed in 50% Hybond-N+ (Amersham, UK), and then hybridized with formamide/26SSC (378C), and 16SSC for 15 min each the 32P-labeled pHP 1.7 probe speci®c to the env region of and once in 46SSC for 5 min. The slides were immersed in ERV3. Autoradiographic evaluation of the 70 ml of ¯uorescein isothiocynate (FITC) - avidin (5 mg/ml was done by the BAS1000 imaging plate system (Fuji Film, in 46SSC with 1% BSA) and incubated at 378C for Japan). 45 min, then washed with 46SSC, 46SSC containing RT-PCR was performed using 4 mg of total RNA with 0.1%. Triton X-100, and 46SSC for 5 min each. The slides ~Tth polymerase (Perkin Elmer, USA) according to the were incubated with 70 ml of biotinylated anti-avidin manufacturer's protocol. ERV3 and H-plk cDNAs were (5 mg/ml in 46SSC with 1% BAS) at 378C for 60 min. ampli®ed with primers designed for the env-region of ERV3 Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2780 gene and for the speci®c region of open reading frame of to an imaging plate, and bands were detected using the H-plk gene (Figure 2c). The primer pairs were BAS 1000 system and an imaging plate (Fujix, Japan). 5'CTTGGTACCAACTTGAAGC-3' (ERV3-1) and 5'-CCC- When a homozygous deletion was suggested by the total CTGCAGTTTCATTGGTGGT-3' (ERV3-2), and 5'-GAGC- loss of alleles in the tumor DNA, further multiple PCR CTTTAACTGGTCCTC-3' (H-plk-1) and 5'-TTGTGGG- reaction was performed with two or three independent GATTCTCTCCAG-3' (H-plk-2), respectively. PCR condi- primer pairs in the same tube without radio-labeling. The tions involved an initial denaturation for 5 min at 948C, ampli®ed products were electrophoresed on a 2% agarose X followed by 30 cycles of 948C for 1 min, 558C for 1 min, gel, followed by Southern blotting using a-32P-dCTP labeled 728C for 2 min, with ®nal 7 min incubation at 728C. PCR probes that were extracted from the agarose containing the products were electrophoresed on 2% agarose X (Nippon- ampli®ed DNA fragment of STSs. gene, Japan) gels and stained with ethidium bromide. GAPDH (Arcari et al., 1984) cDNA was also ampli®ed Genbank Accession Numbers with the primer pairs as a control. ERV3 (Genbank Accession No. M12140) H-plk (Genbank Accession No. M55422). PCR ampli®cation of STS and ERV3 To detect deleted regions on chromosome 7, speci®c EMBL accession numbers sequenced tag site (STS) primers purchased from Research Genetics (USA) were used. Locus and marker name in AFM254yc9 (EMBL Accession No. Z17146) AFM238vbl2 parenthesis are listed below; D7S531 (AFM254yc9), (EMBL Accession No. Z17061) AFM165zf4 (EMBL D7S513 (AFM217yc5), IL-6, D7S484 (AFM087yd11), Accession No. Z16756) AFM191xh6 (EMBL Accession TCRG, GCPS, EGF-R (SAVH3), D7S519 (AFM- No. Z16826) AFM240ve9 (EMBL Accession No. Z17080) 238vb12), D7S494 (AFM165zf4), D7S499 (AFM19lxh6), AFM199vh8 (EMBL Accession No. Z16869) AFM070ye1 D7S520 (AFM240ve9), D7S502 (AFM199vh8), D7S482 (EMBL Accession No. Z16519) AFM280zc5 (EMBL (AFM070ye1), D7S663 (AFM280zc5), D7S639 (AFM- Accession No. Z23996) AFM220ya3 (EMBL Accession 220ya3), D7S645 (AFM238zc9), ELN, D7S669 (AFM- No. Z23668) AFM238zc9 (EMBL Accession No. Z23753) 286xf9), D7S489 (AFM136xe3), D7S524 (AFM248ta5), AFM286xf9 (EMBL Accession No. Z24049) AFM248ta5 D7S486 (AFM098xg9), D7S483 (AFM074xg5) and (EMBL Accession No. Z17104) AFM254yc9 (EMBL D7S594 (AFM254cy9). D7S482, 502, 520, 663 were Accession No. Z17146). reported to be assigned to the same chromosomal position (Tsui, 1995). ERV3 sequence was ampli®ed with the same primers Acknowledgements used in the RT-PCR. Simultaneous PCR ampli®cation of This work was supported in part by a Grants-in-Aid for 2 ± 3 STSs was performed as recommended by the Scienti®c Research, Grants-in-Aid for Special Cancer manufacturer (Research Genetics, USA) but with some Research from the Ministry of Education, Science, Sports modi®cations (Weissenbach et al., 1992). PCR products and Culture, Japan, and CREST (Core Research for were labeled by incorporation of [a-32P] dCTP during Evolutional Science and Technology) of Japan Science thermo- reactions. After denaturation by heating at Technology Cooperation (JST), Japan. We thank Ms M 958C, the products were run on 6% denaturing polyacry- Ohara for critical comments and assistance with the lamide gel containing 7 M urea. Then, the gel was exposed manuscript.

References

Achille A, Biasi MO, Zamboui G, Bogina G, Magalini AR, Jacobs PA, Wilson CM, Sprenkle JA, Rosenheim NB and Pederzoli P, Perucho M and Scarpa A. (1996). Cancer Migcon BR. (1981). , 286, 714 ± 716. Res., 56, 3808 ± 3813. Kajii T and Ohama K. (1977). Nature, 286, 633 ± 634. Arcari P, Martinelli R and Salvatore F. (1984). Nucleic Acids Kastury K, Ba€a R, Druck T, Ohata M, Cotticelli MG, Res., 12, 9179 ± 9189. Inoue M, Negrini M, Rugge M, Huang D, Croce CM, Atkin NB and Baker MC. (1993). Cancer Genet. Cytogenet., Palazzo J and Huebner K. (1996). Cancer Res., 56, 978 ± 67, 123 ± 125. 983. Bellefroid EJ, Lecocq PJ, Benhida A, Poncelet DA, Belayew Kato N, Larsson E and Cohen M. (1988). Int. J. Cancer, 41, A and Maritial JA. (1989). DNA, 8, 377 ± 387. 380 ± 385. Bellefroid EJ, Ried T, Riviere M, Marine JC, Levan G, Kato N, Pfeifer-Ohlsson S, Kato M, Larsson E, Rydnert J, Szpirer J, Szpirer C, Ward DC and Martial JA. (1992). Ohlsson R and Cohen M. (1987). J. , 61, 2182 ± , 14, 780 ± 781. 2191. Bieche I, Champeme MH, Matifas F, Hacene K, Callahan R Kato N, Shimotohno K, Vanleeuwen D and Cohen M. and Lidereau R. (1992). Lancet, 339, 139 ± 143. (1990). Mol. Cell. Biol., 10, 4401 ± 4405. Boyd MT, Bax CMR, Bax BE, Bloxam DL and Weiss RA. Kohler PO and Bridson WE. (1971). J. Clin. Endocrin., 32, (1993). Virology, 196, 905 ± 909. 683 ± 687. Cavanee WK, Dryja P, Phillips RA, Benedict WF, Godbout Koi M, Shimizu M, Morita H, Yamada H and Oshimura M. R, Gallie BL, Murphee AL, Strong LC and RL. (1989). Jpn. J. Cancer Res., 801, 413 ± 418. (1983). Nature, 305, 779 ± 784. Kuniyasu H, Yasui W, Yokozaki H, Akagi M, Akama Y, Cherif D, Benard O and Berger R. (1989). Hum. Genet., 81, Kitanara K, Fujii K and Tahara E. (1994). Int. J. Cancer, 358 ± 362. 59, 597 ± 600. Cohen M, Kato N and Larsson E. (1988). J. Cell Biochem., Martin MA, Bryan T, Rasheed S and Khan AS. (1981). Proc. 36, 121 ± 128. Natl. Acad. Sci. USA, 78, 4896 ± 4989. Conway K, Morgan D, Phillips KK, Yuspa SH and Nozawa S, Udagawa Y, Sakayori M, Tsukazaki K, Weissman BE. (1992). Cancer Res., 52, 6487 ± 6495. Takayama Y, Aoki D, Kubushiro K, Iizuka R and Inaba Decker HJ, Neuhans C, Janch A, Speicher M, Ried T, Bujard N. (1987). Acta. Obst. Gynaec. Jpn., 39, 889 ± 896. M, Branch H, Stockle S, Stockle M, Seliger B and Huber O'Connell CD, O'Brien S, Nash WG and Cohen M. (1984). C. (1996). Hum. Genet., 97, 770 ± 776. Virology, 138, 225 ± 235. Putative choriocarcinoma suppressor gene(s) on Chr.7 T Matsuda et al 2781 Ogata T, Ayusawa D, Namba M, Takahashi E, Oshimura M Stanbridge EJ. (1988). Cancer Surveys, 29, 317 ± 324. and Oishi M. (1993). Mol. Cell. Biol., 13, 6036 ± 6043. Sugiura M, Goto S, Saito M, Kato S, Hattori S and Tomoda Ohama K, Kajii T, Okamoto E, Fukuda Y, Imaizumi K and Y. (1988). Acta. Obst. Gynaec. Jpn., 40, 1447 ± 1454. Tsukahara M. (1981). Nature, 292, 551 ± 552. Surti U, Szuiman AE and O'Brien S. (1979). Hum. Genet., 51, Ohnishi H, Hanada R, Horibe K, Hongo T, Kawamura M, 153 ± 156. Naritaka S, Bessho F, Yanagisawa M, Nobori T, Suzumori K, Sugimoto Y, Suzumori K, Yagami Y and Yamamori S and Hayashi Y. (1996). Leukemia, 10, Takeda A. (1983). Asia-Oceania J. Obstet. Gynaecol., 9, 1104 ± 1110. 309 ± 315. Oshimura M, Kugoh H, Koi M, Shimizu M, Yamada H, Tanaka K, Oshimura M, Kikuchi R, Seki M, Hayashi I and Satoh H and Barrett JC. (1990). J. Cell Biochem., 42, 135 ± Miyaki M. (1991). Nature, 349, 340 ± 342. 142. Tsui LC. (1995). Cytogenet. Cell Genet., 71, 1 ± 31. Park WW. (1971). Choriocarcinoma: a study of its pathology. Venables PJ, Brookes SM, Griths D, Weiss RA and Boyd FA Davis Co: Philadelphia, pp.55 ± 138. MT. (1995). Virology, 211, 589 ± 592. Pattillo RA and Gey GO. (1968). Cancer Res., 28, 1231 ± Vogelstein B, Fearon ER, Kern SE, Hamilton SR, Preisinger 1236. AC, Nakamura Y and White R. (1989). Science, 244, 207 ± Pattillo RA, Ruckert A, Hussa PO, Bernsteiu R and Delft E. 211. (1971). In vitro., 6, 398 ± 399. Wake N, Shiina Y and Ichinoe K. (1978a). Proc. Jpn. Acad., Pattillo PA, Sasaki S, Katayama KP, Roesler M and 54, 533 ± 537. Mattingly RF. (1981). Am. J. Obstet. Gynecol., 141, Wake N, Takagi N and Sasaki M. (1978b). J. Natl. Cancer. 104 ± 105. Inst., 60, 51 ± 57. Porteous DJ, Saxou PJ, Pasquale SR, Jones GR, Geiser AG Weissenbach J, Gyapay G, Dib C, Vignal A, Morisssette J, and Stanbridge EJ. (1989). Genomics, 5, 680 ± 684. Millasseau P, Vaysseix G and Lathrop M. (1992). Nature, Roche J, Boldog F, Robinson M, Robinson L, Varella- 359, 794 ± 801. Gracia M, Swanton M, Waggoner B, Fishel R, Franklin Weissman BE, Saxon PJ, Pasquale SR, Jones GR, Geiser AG W, Gemmill R and Drabkin H. (1996). Oncogene, 12, and Stanbridge EJ. (1987). Science, 236, 175 ± 180. 1289 ± 1297. Yamada H, Wake N, Fujimoto S, Barrett JC and Oshimura Sambrook J, Fritsch EF and Maniatis T. (1982). Molecular M. (1990). Oncogene, 5, 1141 ± 1147. Cloning: A Laboratory Manual. Cold Spring Habor Yamashita K, Wake N, Araki T, Ichinoe K and Kuroda M. Laboratory: Cold Spring Harbor, NY. (1979). Am. J. Obstet. Gynecol., 135, 597 ± 600. Saxon PJ, Srivatsan ES and Stanbridge EJ. (1986). EMBO J., Yokota J, Wada M, Shimosato Y, Terada M and Sugiura T. 5, 3461 ± 3466. (1987). Proc. Natl. Acad. Sci. USA, 84, 9252 ± 9256. Scable H, Cavenee W, Ghavimi F, Lovell M, Morgan K and Zenklusen JC, Bieche I, Lidereau R and Conti CJ. (1994a). Sapienza C. (1989). Proc. Natl. Acad. Sci. USA, 86, 7480 ± Proc. Natl. Acad. Sci. USA, 91, 12155 ± 12158. 7484. Zenklusen JC, Oshimura M, Barrett JC and Conti CJ. Schutte M, daCosta LT, Hahn SA, Moskaluk C, Hoque (1994b). Oncogene, 9, 2817 ± 2825. ATMS, Rosenblum E, Weinstein CL, Bittner M, Meltzer Zenklusen JC, Thompson JC, Troncoso P, Kagan J and PS, Trent JM, Yeo CJ, Hruban RH and Kern SE. (1995). Conti CJ. (1994c). Cancer Res., 54, 6370 ± 6373. Proc. Natl. Acad. Sci. USA, 92, 5950 ± 5954. Zenklusen JC, Thompson JC, Szanto AJK and Conti CJ. Speicher MR, Howe C, Crotty P, du Manoir S, Costa J and (1995). Cancer Res., 55, 1347 ± 1350. Ward DC. (1995). Cancer Res., 55, 1010 ± 1013. Sreekantaiah C, Davis JR and Samdberg AA. (1993). Am. J. Pathol., 142, 293 ± 305.