Endometrial cancer is the most frequently diagnosed female genital tract malignancy in the western world.1 Endometrial adenocarcinoma (EAC) is the prevalent subtype, accounting for approximately 75% of the reported cases.2 It has been clearly demonstrated that an inherited genetic predisposition plays a critical role in the development of many cases of EAC, as the risk for a woman to develop EAC is tripled when there is an affected first-degree relative.3, 4 Molecular genetic analysis of uterine tumor biopsies have revealed alterations in a number of chromosomal regions harboring transforming genes, including tumor suppressor genes (e.g. TP53, PTEN and hMLH1) and oncogenes (e.g. K-RAS and c-ERBB2/neu).1, 5, 6, 7, 8, 9 However, the molecular genetic events underlying endometrial cancer tumorigenesis are still poorly understood.
Females of the inbred BDII rat strain are genetically prone to spontaneously occurring hormone-related endometrial carcinoma, providing a suitable experimental model system for genetic analysis of inherited EAC in humans.10, 11 Cytogenetic and comparative genome hybridization (CGH) analyses of the tumors pointed to common deletions in the proximal part of rat chromosome 10 (RNO10) in the tumor material.12 According to Knudson's two-hit theory, tumor suppressor genes are inactivated by a recessive mutation in one allele and the subsequent loss of the allele retaining wild-type function.13 Allelic loss is therefore considered to be a hallmark of chromosomal regions harboring tumor suppressor genes. Performing detailed analysis of allelic imbalance, we could narrow down and divide the region of common losses in the proximal part of RNO10 into 3 commonly deleted subregions.14, 15 One of the common deletion regions was located in the central part of the chromosome and the Tp53 gene (situated at the border between bands 10q24-q25) was singled out as a candidate gene to be affected by aberrations in this region.
In this work, we investigated whether the Tp53 gene was the molecular target for the frequent allelic losses at RNO10q24-q25. We performed fluorescent in situ hybridization (FISH) on the tumor materials using both dual-color gene-specific FISH and chromosome paint analysis to determine frequency and map position of chromosome breaks along RNO10 that have presumably resulted in allelic losses. Furthermore, we combined sequencing for gene mutations and analysis of allelic imbalance (AI) to investigate status of Tp53 in the tumor materials. The combined analysis suggested that Tp53 might not be the only molecular target in the region.
AI, allelic imbalance; CGH, comparative genome hybridization; EAC, endometrial adenocarcinoma; ESCC, endometrial squamous cell cancer; FISH, fluorescent in situ hybridization; F1, first intercross progeny; F2, second intercross progeny; HSA17, human chromosome 17; MPM, malignant peritoneal mesothelioma; N1, backcross progeny; RNO10, rat chromosome 10; Ttsg, target tumor suppressor gene at RNO10q24-q25.
Material and methods
Animals of the inbred BDII/Han rat strain are genetically predisposed to spontaneous EAC, with an incidence of more than 90% in virgin females starting as early as at 12 months of age.10, 11 BDII females were crossed to males from 2 nonsusceptible strains, BN/Han and SPRD-Cu3/Han. To generate F2 and backcross populations, the F1-animals were crossed to each other or crossed back to BDII females. In both strain crosses, spontaneously arising tumors developed in a proportion of F1, F2 and backcross (N1) animals.16, 17 These animals contain segments of heterozygosity making the tumors suitable for AI analysis.
At necropsy, tumor specimens and matched normal liver and/or spleen tissue samples were collected from animals for DNA extraction and cell culture establishment. Small pieces of fresh tumor tissue were used to set up primary cell cultures in DMEM. In the present study a total of 29 cultures were subjected to analysis, representing 27 EAC tumors, 1 endometrial squamous cell cancer (ESCC) and 1 malignant peritoneal mesothelioma (MPM). The 2 non-EAC tumors were included in the analysis for comparison. Using the GenElute (Mammalian Total RNA Kit (SIGMA) and SuperScript (First-Strand Synthesis System for RT-PCR (Invitrogen), total RNA and cDNA was prepared for all tumor cell cultures. Genomic DNA was extracted from paired tumor/normal tissues, using a standard phenol-based method in the Genepure™ 341 Nucleic Acid Purification System (PE Applied Biosystems, Foster City, CA).
Chromosome painting and dual-color FISH studies
Slides for cytogenetic analysis were prepared using standard procedures.18 Briefly, tumor cell cultures were treated with 0.2 mg/ml of 5′-bromo-2′-deoxyuridine (BrdU) for 17 h. Subsequently, the cells were washed 3 times and cultured for 6 h in medium supplemented with 0.05 μg/ml thymidine. Mitotic figures were accumulated by adding 0.05 μg/ml Colcemid (Sigma) during the final 30 min and metaphase cells were harvested by mitotic shake-off. The cells were resuspended in 0.07 M KCl at room temperature for 10 min, and then washed and fixed in 3 different dilutions of methanol:acetic acid (9:1, 5:1 and 3:1). Dual-color FISH analysis on RNO10 was performed using a biotinylated (Nick Translation Systems, GibcoBRL) or digoxigenine-11-dUTP labeled (DIG-NICK Translation Mix, Roche Diagnostics GmbH, Mannheim, Germany) PAC or BAC DNA probes for the genes (Table I).
Table I. Probes used in Dual-Color Fish Experiments
Location (#start, Mb)
Pfn1, Eno3, Spag7, Camta2, Kif1c and two predicted genes (LOC360555 and LOC497944)
Pafah1b1, Mnt, Rutbc1, Srr and two predicted genes (LOC360568 and LOC287522)
Hic1, Rtn4rl1, Ovca2 and two predicted genes (LOC287522 and LOC287523)
Slc43a2, Pitpn, Skip, Myo1c, Crk
Two predicted genes (LOC360228 and LOC497972) and Tcf2
Spag9, Tob1 and two predicted genes (LOC287631 and LOC360602)
Erbb2, Zpbp2, Znfn1a3 and two predicted genes (LOC287668 and LOC360618)
Roughly 500 ng of the coprecipitated probes along with about 15-fold excess of total sonicated rat genomic DNA were cohybridized to each slide. Detection of the dual-color labeling was performed using a mixture of Rhodamine-conjugated antidigoxigenin and FITC-conjugated avidin (GibcoBRL). For each chromosome paint experiment, 8–10 μl of the RNO10-specific probe solution (a biotin-labelled PCR amplification product of flow-sorted RNO10; Cambridge Resource Centre for Comparative Genomics) was applied per slide. To detect hybridization of the RNO10 paint probe, FITC-conjugated avidin was used. The chromosome preparations were washed and counter-stained and the fluorescence signals were visualized as described previously.19
DNA sequencing and mutation analysis
PCR primers corresponding to the genomic and cDNA sequence from codon 70 till 362 of the rat Tp53 gene (corresponding to codon 72–364 in human TP53 gene encoding the p53 core DNA-binding domain), were designed based on available sequences in GenBank. This region is highly homologous to the corresponding human gene, but there is no intron between exons 6 and 7 in the rat Tp53 gene.20 Primers for amplification and sequencing of the gene at genomic DNA were intron-4aF 5′-AACCAGAGGGAGGAAAACAAA-3′, intron-4bF 5′-CGCTGACCTTTGATTCTTTCTC-3′, intron-7R 5′-CCTGGCACACAGCTTCCTAC-3′, intron-7F 5′-CTTACTGCCTTGTGCTGTGC-3′, intron-9R 5′-CCTTGGTACCTTAAGGGTGAA-3′ and at cDNA level were cDNA-ex4F 5′-CAGCACAGGAACCTGGAACT-3′, cDNA-ex5F 5′-AAGACATGCCCTGTGCAGTT-3′, cDNA-ex8F 5′-TTTGAGGTTCGTGTTTGTGC-3′, cDNA-ex9R 5′-TTTTTCTTTTGCTGGGGAGA-3′ and cDNA-ex10R 5′-CCTGCTGTCTCCTGACTCCT-3′. PCR amplification products were purified using GFX™ PCR DNA and Gel Band Purification Kit (Amersham Pharmacia Biotech, Piscataway, NJ). Using ABI PRISM® BigDye® Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems), the purified DNA fragments were subjected to sequencing according to the protocol provided by the manufacturer. Sequencing products were then denatured, cooled on ice and 1.5–1.8 μl of each sample was electrophorezed in a prewarmed 5% denaturing polyacrylamide gel (Long Ranger® Single pack gel, in vitro) in 1× TBE buffer (89 mM Tris/89 mM boric acid/2 mM EDTA, pH 8.3) using an automated ABI Prism™ 377 Genescan Analyzer (Applied Biosystems). The sequence data obtained were subjected to BLAST searches (http://www.ncbi.nlm.nih.gov/BLAST/) for comparison to published sequences. Sequencing was performed on templates derived from solid tumor material as well as on the tissue culture samples. Observed Tp53 gene mutations were verified by repeating sequencing at the cDNA level.
Northern blot analysis
Total RNA from tumor cell cultures was extracted using the GenElute™ Mammalian Total RNA Kit as described by the manufacturer (Sigma). Sufficient amounts of good quality RNA, between 2 and 5 μg, were obtained in 19 cases. Total RNA (about 3 μg) was separated by agarose gel electrophoresis using the NorthernMax-Gly kit and transferred to BrightStar-Plus nylon membrane according to the manufacturer's instructions (Ambion, Austin, TX). Northern blots were probed with [32P]UTP-labeled antisense RNA specific for rat Tp53 (NM_030989; Tp53-n210F 5′-CATCGAGCTCCCTCTGAGTC-3′ and SP6+Tp53-n1267R 5′-ATTTAGGTGACACTATAGAAGTGcctgctgtctcc-3′). As control, blots were probed with antisense RNA of the β-actin. Labeling, hybridization and probe removal were performed with “NorthernMax” and “Strip-EZ™ RNA” reagents (Ambion, Austin, TX) according to the manufacturer's instructions.
RNO10 aberrations revealed by chromosome paint and dual-color FISH
A total of 29 rat tumor cell cultures, including 27 EAC and 2 non-EAC tumors were subjected to karyotype analysis. Average chromosome numbers started in the near-diploid/hyperdiploid range through to near-triploid (modes at 2n = 39–71). The modal numbers are shown in Table II, but there was always considerable variation around the mode. Two of the lines with diploid modes (NUT39, NUT52) had a salient secondary mode in the tetraploid region. One line, RUT5, the ESCC tumor, was in the near-tetraploid range (mode at 2n = 77).
Table II. Molecular Cytogenetic Analysis of RNO10 Aberrations in 27 EAC and 2 Non-EAC Tumors
The RNO10-specific chromosome paint made it possible to identify all chromosomes containing RNO10 chromosomal material in each tumor (Table II). Our analysis verified that chromosomal aberrations affecting RNO10 were quite common among these tumors; 20 of the EAC tumors (74%) contained RNO10 paint-positive chromosomes that were obviously rearranged. The expected number of RNO10 copies per cell could be calculated from the modal chromosome number of each tumor line. After identifying all of the RNO10-derived chromosomes typical for each tumor, the number of RNO10 paint-positive elements in the 29 tumors could be counted at a total of 93 chromosomes and compared to the 75 RNO10 chromosomes expected. Thus, the number of RNO10 paint-positive chromosomes was in excess of the number expected in 15 of the 27 EAC tumors (Table II). On the other hand, many of the supernumerary RNO10 elements were smaller than normal RNO10 chromosomes, so there was not necessarily any corresponding excess in total amount of RNO10 material. It was noted that both non-EAC tumors exhibited the expected number of RNO10 chromosomes. In summary, it appeared that 9 of the 29 tumors (including the two non-EAC tumors) all paint-positive chromosomes appeared to be essentially normal RNO10 chromosomes (among them one case involved in a Robertsonian translocation). Among the EAC tumors, 20 of 27 (74%) displayed aberrant RNO10-derived chromosomes: deletion chromosomes, small but completely RNO10 paint-positive, or translocation chromosomes, RNO10 paint-positive segments translocated onto other chromosomes (Table II; Fig. 1). In addition, all of these tumors (except RUT2) displayed 1–3 essentially normal-looking RNO10 chromosomes (including two cases of Robersonian translocations of the entire chromosome). The total number of RNO10 paint-positive elements among the EAC tumors was 88 (average 3.26 per tumor), 48 of which looked as RNO10 chromosomes of essentially normal configuration, whereas 40 were clearly rearranged (deleted or translocated).
The CGH analysis had indicated that in these tumors there were deletions/reduced copy numbers of RNO10 proximal segments, whereas RNO10 distal segments occurred in increased number of copies.12, 21 To get more specific knowledge of what parts of RNO10 were present in the various rearranged chromosomes in the tumor cells, the chromosome paint analysis was combined with dual-color FISH experiments. By using gene-specific probes it was possible to determine approximately where the breaks occurred when these RNO10-derived marker chromosomes had been generated. We used PAC clones representing Tp53 (red) and Thra (green) as probes (loci situated at RNO10q24-q25 and RNO10q32.1, respectively) in order to be able to identify segments both from the proximal-central and the distal part of the chromosome. The analysis verified that RNO10 chromosomes classified as essentially normal exhibited signals in the normal configuration with Tp53 in the central and Thra in the distal part of the chromosome, respectively (Fig. 1), supporting the notion that they represented RNO10 chromosomes without rearrangements. The FISH analysis showed that there was no evidence of breakage between the two genes in 9 EAC tumors. The two non-EAC tumors could be added to this group for a total of 11 out of 29. In the remaining 18 EAC tumors the RNO10-derived marker chromosomes gave evidence of breakage between the Tp53 and Thra genes (Table II; Fig. 1). The expected total number of normal RNO10 chromosomes in these 18 tumors can be calculated at 49, but only 29 of totally 67 RNO10 paint-positive chromosomes showed the normal configuration of two red and two green spots (in one tumor, RUT30, the order was reversed by an inversion), whereas 38 RNO10 paint-positive chromosomes were either deletion or translocation chromosomes. Thirty-six of these rearranged chromosomes (20 deletion and 16 translocation chromosomes; Table II) provided evidence of chromosome breakage between the two markers. Thus, among the deletion chromosomes, 17 contained the Thra but not the Tp53 gene, whereas three contained the Tp53 but not the Thra gene. Similarly, among the translocation chromosomes, 14 contained the Thra but not the Tp53 gene, and the remaining two exhibited one or two Tp53 signals but none from Thra. The conclusion is that these RNO10-derived marker chromosomes occurring in 18 EAC tumors had resulted from chromosome breakage in the chromosome segment between the two markers analyzed. By counting the number of signals in the FISH analysis it was shown that these rearrangements lead to a net reduction of the number of Tp53 copies in five EAC tumors, to a net gain of Thra copies in three EAC tumors, and to both reduction in Tp53 and gain in Thra in eight EAC tumors (Table II; Fig. 1).
To determine whether the location of the breaks between the Tp53 and Thra genes were random or not, we extended our dual-color FISH analysis by adding another eight locus-specific probes spaced out between Tp53 and Thra on RNO10 in 17 tumors (Table I; because of lack of chromosome preparations, the 18th tumor, RUT12, had to be excluded from this analysis). From the analysis, chromosome deletions could be defined in all 17 tumors, including eight of the tumors uninformative at RNO10q24-q25 (Table III). The most proximal break was between Tp53 (56.4 Mb) and Pfn1 (located proximally in BAC CH230-364I24, 57.5 Mb). The second most proximal break was between BACs CH230-8N1 (62.3 Mb) and CH230-19F2 (62.8 Mb) and was observed in three tumors (RUT2, RUT7 and RUT127). This break resulted in deletions in both directions in the tumors (proximal to the break point in RUT7 and RUT127 but distal to the break point in RUT2, Table III). The most distal break was between Znfn1a3 (located distally in BAC CH230-16I9, 86.8 Mb) and Thra (87.5 Mb). In RUT2 two breaks were detected. In addition to the proximal break between BACs CH230-8N1 and CH230-19F2, there was a distal break between BAC CH230-16I9 and Thra, resulting in loss of the whole segment in between. This tumor did not show deletion at Tp53. The remaining 16 tumors displayed Tp53 deletions either as large deletions with variable sizes of 6–31 Mb (in 13 tumors) or as losses at the Tp53 locus only (in 3 tumors).
Table III. RNO10 Deletion Map in 17 EAC Tumors using Fish Data
Two PACs and eight BACs were used as probes in dual-color FISH analysis to investigate approximate position of chromosome breaks and deletions along RNO10. Chromosome deletions are marked in gray. For reference, data on chromosome number, expected number of RNO10 and RNO10 aberrations from Table II are repeated. D – deletion, * – no deletion, */D – deletion break point inside the BAC probe, nd – not done.
Mutations in the Tp53 tumor suppressor gene
Using primers designed for the Tp53 gene exons 4–10, including the core DNA-binding domain of the p53 protein, genomic DNA from all 29 tumor samples (27 EAC and the 2 non-EAC tumors) as well as the three parental inbred strains (BN, BDII and SPRD) was PCR-amplified and sequenced (Table IV). The sequences obtained from normal DNA were identical in the three inbred strains (with the exception for an intronic SNP, see later), but differed in one nucleotide when compared with the available GenBank reference sequence for rat p53 mRNA (RefSeq NM_030989). The difference was at nucleotide 705, where we found a C rather than a G in the coding sequence, which will cause codon 174 to be changed from TGG (Trp) to TGC (Cys). Since we found the same sequence in all three rat strains, we believe that this sequence must be representative for an important subset of inbred rat strains. This conclusion is supported by the fact that there is a Cys, rather than a Trp, at the corresponding position in the homologous human and mouse proteins. In fact, there may be a distinct possibility that the RefSeq sequence is in error, since the available GenBank sequence for RNO10 WGS supercontig (NW_047334) is in agreement with our sequence (the Rat Genome Project sequence has been derived from a BN inbred rat). As mentioned already, an SNP was detected at nucleotide 43246020, which is situated at nucleotide 194 in the 311-bp long intron 7. At this position the BN strain exhibited an A (again in agreement with the Rat Genome Project sequence), whereas both BDII and SPRD DNA displayed a G.
Table IV. TP53 Mutations and SNP Data at TP53 Intron 7 in 27 EAC and two Non-EAC Tumor Cell Cultures
AI data, I/U
Tp53 FISH data from Table II is repeated for convenience. Tumor type: EAC – endometrial adenocarcinoma; MPM – malignant peritoneal mesothelioma; ESCC – endometrial squamous cell carcinoma. Cross: SP, BN – crossed in susceptible inbred strain was SPRD or BN, respectively. Generation: F1 – first generation intercross offspring; F2 – second generation intercross offspring; N1 – first back-cross generation offspring. Tp53 mutation mode: H – homozygous/hemizygous mutation; T – heterozygous mutation. SNP data: allele retained: G, A – guanine (BDII allele) or adenine (BN allele), respectively, at nucleotide 194 in intron 7; na – not applicable because both parental strains have the same allele at SNP locus (SPRD cross). AI data: I, U – informative or uninformative at RNO10q24-q25, respectively, as determined by genotyping liver DNA with polymorphic microsatellite markers.
The Tp53 mutation analysis was performed for codon 70 till 362 on both genomic and cDNA templates from both solid tumor and tissue culture cell material (material was not available from a few solid tumors). Mutations in the Tp53 gene were recorded for 18 EAC tumors, whereas no mutations could be detected in 9 EAC tumors (Table IV). For samples where data were available, the same mutations were seen in both solid tumor and tissue culture material from the same tumor, suggesting that the mutations originated in vivo. The part of the Tp53 gene that was analyzed in this study includes amino acids (aa) 72–364 in human p53 (covering the core DNA-binding segment of the protein, corresponding aa in rat are 70–362), in which about 98.9% of TP53 mutations are found in human tumors.22, 23 The human and rat genes are highly homologous in this segment and the genomic structure is also highly conserved. However, there is an intron of 576 bp between human exons 6 (113 bp) and 7 (110 bp), which does not occur in the rat.20 Because of the structural homology between the rat and human genes we have preferred to keep the human exon numbering, which means that we designated the single corresponding rat exon 6/7 (213 bp). Eight of the mutations were detected in exon 5, three in exon 6/7, six in exon 8 and one in exon 9. In addition, both of the non-EAC tumors displayed Tp53 mutations (in exon 7 and 8). Among the 18 EAC mutations there were 12 transitions (the most common was 7 G-A replacements), three transversions and three deletions. Fourteen of them will lead to single amino acid substitutions, whereas four would give truncated or highly abnormal products. Characteristically, the solid tumor material often yielded both mutant and normal Tp53 sequence, where the cultured tumor cells contained only the mutated Tp53 allele. This was the case even though our previous FISH analysis had shown that in most cases there was more than 1 copy of Tp53 present, often dwelling in normal-looking RNO10 chromosomes (Table II; Tp53 FISH data repeated for convenience in Table IV). The exception was NUT16, in which both the normal and the mutated version of Tp53 were present also in the cultured cells. In this tumor there were also 2 normal-looking RNO10 copies (Table II). It cannot be excluded that in some of the tumors there had been an evolution from mutant/wildtype Tp53 heterozygosity in the solid tumor material to Tp53 mutant homozygosity/hemizygosity in the corresponding tissue culture.
The occurrence of an SNP inside the Tp53 gene offered the opportunity to verify the homozygosity/hemizygosity status of Tp53 in a subset of the tumors. The SNP data could also be used to verify the previous AI results. Informative tumors had to be derived from the BN cross, since the BDII and SPRD alleles were identical. Among the 18 tumors derived from the BN cross, eight were informative in region III and among them four had been classified as AI with BN allele dominating, three as AI with BDII as dominating allele and one as retained heterozygosity (Table IV). In each case the SNP allele analysis fully confirmed the AI data, and, in fact, the SNP analysis supported the notion that AI in this region represented true loss of one allele (LOH). Furthermore, in six of the eight cases there was also a Tp53 mutation, three of them were homozygous/hemizygous for a mutation in the BN allele (RUT7, RUT25, NUT127), two of them were homozygous/hemizygous for a mutation in the BDII allele (NUT52 and the non-EAC RUT5 tumor) and one was heterozygous for a normal and a mutated allele (NUT16). In all cases the SNP analysis showed that only one allele was present (except in the case of NUT16, Table IV), this was in spite of the fact that the FISH analysis had shown that four of tumors contained two copies of the Tp53 gene and one of them (RUT5) actually had four copies.
Northern blot analysis of Tp53
Northern blot analyses of 19 EAC tumors revealed significant decrease in Tp53 transcript in two tumors, RUT12 and RUT30 (Fig. 2). Both tumors had frame-shift mutations in the Tp53 gene, resulting in production of truncated Tp53. No significant decrease in expression level of Tp53 was observed in the tumors with AI and/or deletion at RNO10q24 and no Tp53 mutation (RUT2, RUT4, NUT39, NUT42 and NUT128; Fig. 2).
Our analysis, using RNO10-specific chromosome paint combined with gene-specific dual-color FISH, suggested that the 27 EAC tumors could be subdivided into a group of 9, in which there was no chromosome break between the two mid-to-distal markers (Tp53 and Thra), and a group of 18, in which there was evidence of random chromosome breakage separating the middle and distal parts of RNO10 generating deletion and/or translocation chromosomes. These data were in line with our previous CGH findings12 in that there was a relative decrease in the number of signals from the proximal probe (Tp53, in 13 tumors) and an increase in signals from the distal one (Thra, in 11 tumors).
Allelic imbalance analysis of RNO10 led to the identification of four chromosomal regions that were commonly affected by AI.14, 15 The third of these regions was situated at RNO10q24-q25. Fifteen of the 27 EAC tumors were informative (tumor-bearing animal heterozygous) in this region and the AI analysis showed that there was AI in 12 tumors (Table V). Since the Tp53 gene (already shown by FISH to be commonly reduced in these tumors) is located in this chromosome region, Tp53 was selected as the candidate target gene for further analysis at molecular level. The Tp53 mutation analysis in the 27 EAC tumor samples showed that there were mutations for the Tp53 gene in 18 tumors, whereas no mutations could be detected in 9 tumors (Tables IV and V). Seven of the 9 tumors without Tp53 mutation had AI and/or deletion at RNO10q24-q25 (marked in gray in Table V).
Table V. TP53 Mutation, AI and Fish Data at RNO10Q24-Q25 in 27 EAC Tumors
The first 20 tumors in the Table displayed AI and/or deletion at RNO10q24-q25; seven of these tumors did not have Tp53 mutation (marked in gray). nd – not done; AI data at RNO10q24-q2514, 15: U – uninformative; BN, BDII, SP – the allete retained; HET – no AI.
In human malignancies the TP53 tumor suppressor gene is the most commonly altered gene. Frequency of TP53 mutation in human estrogen-dependent (type I) EACs ranges from 16 to 40%.24 We compared the mutations we had detected with those found in humans as reported in the IARC Tp53 Database (release 9) listing 19,809 somatic mutations from human tumors (http://www-p53.iarc.fr/Somatic.html). At the protein level there is a two aa difference between human and rat p53, the human protein being 393 aa and the rat protein 391 aa. Consequently, there is a shift by two aa between the two species in the core DNA binding sequence, which spans aa 102–292 in the human p53 protein and exhibits 90% aa sequence identity with rat. Allowing for this aa displacement, all of the mutations that we found in the EAC tumors have also been seen in human tumors. We noted that two individual mutations were seen twice each in our material, Arg-His at codon 173 (in NUT16 and NUT97) and Arg-Cys at codon 271 (in RUT6 and RUT29). The corresponding changes in human p53 (Arg-His at codon 175 and Arg-Cys at codon 273) are among the six most common mutations in human cancer.
The occurrence of an SNP inside the Tp53 gene offered the opportunity to verify the homozygosity/hemizygosity of Tp53 in the informative subset of tumors derived from the BN cross. The SNP data could also be used to verify the previous AI results and it was shown that in each case of AI, the SNP analysis confirmed that AI was present and that AI in these cases actually should be regarded as LOH (Table IV).
Cytogenetic and FISH analysis of RNO10 showed that in 18 EAC tumors, including 10 tumors with AI at RNO10q24 and eight tumors uninformative at this region, there was evidence of breakage in between the Tp53 (10q24-q25) and Thra (10q32.1) genes. Detailed FISH analyses revealed that the chromosome breaks resulted in deletions in all 17 tumors tested (the 18th tumor was not included in the analysis, Table III). The observed patterns of deletions combined with Tp53 mutation data suggest that selection against Tp53 alone is not sufficient to explain the extent of loss of genetic material in all cases. This is particularly clear in RUT2, which exhibited no involvement of Tp53 at all. The data from this tumors point to a target tumor suppressor region (Ttsg) located distally of Tp53, near BAC CH230-19F2 (64.8 Mb). Three tumors exhibited mutation and deletion only in the Tp53 gene (bottom of Table III), making this gene the most likely target in these three tumors. The remaining 14 tumors, including five without Tp53 mutation, displayed deletions of different sizes, comprising either the candidate tumor suppressor region alone (in RUT2) or both Tp53 and the candidate tumor suppressor region (Table III).
In summary, in the present study the association between chromosomal aberrations in RNO10, AI and FISH analysis at RNO10q24, and Tp53 mutation was analyzed in 27 rat EAC tumors. AI analysis suggested that the 27 EAC tumors could be subdivided into a group of 15 that were informative at RNO10q24-q25 and a group of 12 uninformative tumors in this region. Among the 15 informative tumors, 12 showed AI at RNO10q24. In dual-color FISH experiments, eight of the 12 uninformative tumors displayed deletions at RNO10q24-q25. Thus, altogether, 20 of 27 EAC tumors (74%) showed AI and/or deletion at RNO10q24-q25. However, there were no Tp53 mutation in seven of these 20 tumors (35%, marked in gray in Table V), and among them there was no reduction of Tp53 expression, as determined by Northern blot analysis, suggesting that, at least in these seven tumors the target for AI was not Tp53.
Taken together, the occurrence of recurrent chromosome breaks and losses distal to Tp53 along with high frequency of allelic imbalance at RNO10q24-q25 and relatively lower frequency of Tp53 mutation suggests that RNO10q24-q25 may harbor another tumor suppressor gene (target tumor suppressor gene, Ttsg), which plays an important role in EAC development. Seven of 20 tumors with AI and/or deletions at RNO10q24-q25 do not have mutation in the Tp53 gene. Consequently, at least part of the observed Tp53 mutations may be events independent from those related to allelic imbalance and loss incidents during EAC development. The observed pattern indicates that Ttsg is located close to and distal of the Tp53 gene, somewhere in the chromosomal segment 62.3–63 Mb. The FISH data for 17 EAC tumors are indicative of deletion of Ttsg in one tumor, of deletion of Tp53 in three EAC tumors, and of deletion of both Tp53 and Ttsg in 13 EAC tumors (Table III).
High frequency of LOH at 17p13.3 (HSA17p13.3) accompanied by low frequency of Tp53 mutation has been reported to be common in a variety of human malignancies including breast, lung and hepatocellular carcinomas as well as neuronal tumors.25, 26, 27, 28, 29, 30, 31, 32 Several candidate genes with possible tumor suppressor activity, including CRK, YWHAE, TUSC5 by Konishi H. et al. 200328 and DPH2L1 (OVCA1), CRK, ABR, GEMIN4 and PRP8 by Zhao X. et al. 200332 are mentioned in these reports, all located telomeric to TP53, around the band 17p13.3. It is a distinct possibility that the candidate tumor suppressor region (Ttsg) implicated by the present study may be analogous to that discussed in these human cancer studies. From approximately the band 10q23 to its telomere, RNO10 corresponds to the entire HSA17, representing the longest region of conserved synteny between humans and rodents.33 However, based on the comparison of the rat and the human draft DNA sequence, it is clear that the conservation with respect to gene order is not so great and that several intrachromosomal rearrangements have taken place in this region during the divergence between human and rat (Table VI). Starting from HSA17p telomere (PRH3AL, 0.06 Mb) till position of TP53 in the human DNA sequence (7.51 Mb), it is necessary to subdivide HSA17 into at least 5 segments (Table VI, segments A–E), which have been involved in different intrachromosomal rearrangements to generate the present RNO10 configuration. Three genomic regions at HSA17p13.3 have been reported to harbor potential tumor suppressor genes in different types of human malignancies. A region around D17S379 and D17S22 (D17S5) had been shown to exhibit the highest frequency of LOH in ovarian, lung and astrocytic tumors.28, 29, 31 The second frequently affected region is a telomeric-subtelomeric region of HSA1713.3 encompassing D17S1866, D17S831 and D17S1574, which corresponds to the minimum deleted regions in hepatocellular carcinoma.26, 32 A homozygous deletion fine-mapping study has demonstrated a third region of LOH involving D17S1174 in lung cancer.28 The present study suggests that a potential tumor suppressor gene(s) with involvement in rat EAC development is located in a chromosome segment with approximate size of 0.5 Mb, including BACs CH230-8N1 and CH230-19F2 on RNO10. As shown in Table VI, by comparing our candidate region in RNO10 with the corresponding region in HSA17p13.3, candidate genes located distal to Crk may potentially be excluded from further analysis, provided that the rat and human candidate regions are analogous. This includes four of the candidate genes GEMIN4, TUSC5 and ABR as well as YWHAE suggested by Konishi H. et al. (2003) and Zhao X. et al. (2003)28, 32 (Table VI). In a reciprocal comparison approach, genes located in segments D, E as well as in the beginning of segment C may also be excluded, since they are located upstream of D17S379 and, thus, are outside of the suggested candidate region in the human studies. As a result, taking both the human and the rat data into consideration, only a small segment of HSA17p13.3 with approximate size of 0.64 Mb (the region between Crk at 1.27 Mb and marker EST1A at 1.91 Mb) would remain for further investigation. The equivalent rat chromosomal segment is about the same size. We have performed cDNA microarray-based RNA profiling (unpublished data) on 12 of the tissue cultures investigated in the present study, on six additional EAC tissue cultures, and on four tissue cultures from other cell lines from endometrial tissues pathologically classified as normal/pre-malignant lesions. The genes located within the candidate region were not all represented on the chips used in the microarray experiments (18K Rat 70mer oligonucleotide set). However the results from the preliminary analysis of the microarrays revealed that the endometrial tumor tissue cultures on average had 3.8 times lower expression of Crk in comparison to the normal/pre-malignant endometrial tissue cultures. This implies that this region might contain relevant candidate genes including Crk and other genes located close to it (such as Hic1, Ovca2, Prfp8, Rilp, Scarf1, Slc43a2, Pitpn, Skip and Myo1c) and, therefore, the region will be subjected to further investigations.
Table VI. RNO10 “Gene on Sequence” Data Compared to Corresponding Data for Human Chromosome Segments Counterparts
The data is extracted from NCBI Map Viewer and is completed using Ensembl v37 data. For all predicted genes in rat and human, HGNC Symbol ID is used (genes marked in italic), when the blast search for the human sequence resulted in one single hit in rat at expected position in the sequence data and the other way around. For segments A, B and C all genes are presented, while for other segments only the first and the last genes plus a few more are shown. The position of some human anonymous markers (boxed) have been included to facilitate comparison with data from human tumors.
In conclusion, together with similar observations in human, the present study provides further evidence for the presence of a putative tumor suppressor gene close to, but distal of the Tp53 gene. The association between observed patterns of chromosomal aberrations in RNO10, AI at RNO10q24 and Tp53 mutation was analyzed in 27 rat EAC tumors. In combination with the results from human cancer studies our findings were used in a genome sequence comparative approach to reduce the size of the suggested candidate tumor suppressor region. It is particularly noteworthy that our analysis in the BDII inbred rat model indicates the involvement of a segment of RNO10 homologous to HSA17p13.3, which has been implicated in different types of human tumors. The system in the BDII inbred rat model supposedly has quite limited genetic variability compared to natural mammalian populations, and one might expect that the set of genetic aberrations leading to a specific type of cancer should be likewise limited. The present results give an indication that the suggested tumor suppressor gene(s) might have a particular significance in rat EAC development and that further analysis of this locus might serve as an important foundation for future efforts towards the identification of the putative tumor suppressor gene(s) at 17p13.3 in human tumors.
The RNO10 paint probe was generously provided by Drs. Malcolm Ferguson-Smith and Fengtang Yang (Clinical Veterinary Medicine, Cambridge Resource Centre for Comparative Genomics, University of Cambridge). The authors thank Prof. Göran Levan for critical reading of this manuscript. We are grateful to Elisabet Jansson and Brita Bjönness for excellent technical assistance.