Mapping common deleted regions on 5p15 in cervical carcinoma and their occurrence in precancerous lesions

Background Previous studies have shown that the short arm of chromosome 5 (5p) exhibit frequent genetic changes in invasive cervical carcinoma (CC), and that these changes arise early during the carcinogenesis, in precancerous lesions. These data therefore suggest that loss of candidate tumor suppressor genes located on 5p is associated with the development of CC. However, the precise location of 5p deletions is not known. Results We performed a detailed deletion mapping of 5p in 60 cases of invasive CC. We found that 60% of the tumors exhibit a 5p loss of heterozygosity (LOH). The patterns of LOH allowed us to identify two minimal regions of deletions, one at 5p15.3 spanning a 5.5 cM genetic distance and a second site of 7 cM at 5p15.2-15.3. In addition, we also identified 5p deletions in 16% lesions of high-grade cervical intraepithelial neoplasia (CIN). 5p LOH was found in 63% of HPV 16 positive tumors, while only 33% tumors with other HPV-types had 5p LOH. The differences in frequency of 5p LOH between tumors harboring HPV16 in combination with other HPV types and tumors harboring HPV16 DNA alone were significantly higher, suggesting a synergistic effect of high-risk types in causing genomic instability. Conclusion These findings implicate the presence of tumor suppressor gene(s) on 5p relevant to CC tumorigenesis.

mutations play a critical role in the initiation and progression of CC. Delineation of these genetic changes is crucial to understanding the molecular basis of CC. A number of genetic changes have been identified so far in invasive CC. However, the precise mechanism of each of these genetic alterations in the multi-step pathway of cervical tumor development is largely unknown. The development of CC is preceded by distinct morphological changes from normal epithelium to carcinoma through low-grade and highgrade squamous intraepithelial lesions (SILs). The genetic basis of this progression is also poorly understood.
Cytogenetic and molecular genetic analyses have identified the short arm of chromosome 5, 5p, as frequently affected by duplications and deletions in a high proportion of CCs [3][4][5][6]. These findings suggest the presence of positively regulated amplified oncogenes and negatively regulated tumor suppressor genes (TSGs) on this chromosomal arm. Comparative genomic hybridization (CGH), a molecular cytogenetic method that identifies chromosomal gains and losses, and loss of heterozygosity (LOH) studies have also demonstrated recurrent deletions on the 5p in precancerous lesions [7][8][9][10]. Although, these studies have identified 5p as a commonly affected region, the precise site of deletion that harbor putative TSG is unknown because of the lack of systematic and detailed deletion mapping studies.
To further characterize the 5p region in CC, we performed a high-resolution deletion mapping in invasive cancer and found that 60% of patients exhibit LOH. We identified two sites of minimal deletions at 5p15.3 and 5p15.2-15.3 spanning 5.5 cM and 7 cM genetic distance, respectively. We have also shown that the 5p15 region deletions occur in 16% of high-grade CIN lesions. Here, we have identified two sites of candidate tumor suppressor genes and confirmed the previous reports that the 5p deletions occur in a small proportion of precancerous lesions.

Identification of two regions of minimal deletions at 5p15.2-p15.3
Analysis of LOH in 60 invasive CCs using 20 STRP markers mapped to 5p showed deletions in 36 of 60 (60%) tumors in at least one marker (Table 1). Of the 36 tumors with LOH, 17 (47.2%) had LOH at all the informative markers suggesting genetic monosomy of 5p (Fig. 1). The patterns of LOH in the remaining 19 tumors that exhibited regional losses on 5p were used to map the common deleted regions (Fig. 1). This has identified two minimal deletions at 5p15.3 and 5p15.2-15.3 ( Fig. 1 & Fig. 2). All the tumors that defined the minimal deletions showed more than 60% reduction in signal intensity in the deleted alleles.
The distal 5p15.3 deletion spans the telomeric region encompassing D5S392, D5S678, and D5S417 markers in 32 tumors. Although two tumors (T-13 and T-49) showed deletions of terminal marker D5S392 alone at 0 cM genetic distance, which is the distal most marker on 5p, we considered the deletion boundaries defined by the proximal marker D5S417 in tumors T-6 and T-105 ( Fig. 1 & Fig. 2). The second and proximal region of deletion spans 6 markers (D5S406, GATA82H02, D5S1455, D5S2054, D5S635 and D5S676), and was deleted in 28 tumors. The deletion boundaries were defined proximally by the tumors T-59 and T-66, while the distal boundary was identified by the tumors T-59 and T-114 ( Fig. 1 & Fig. 2). All tumors that showed 5p deletions fell within the two defined regions of deletions, while 25 tumors exhibited deletions at both re-gions. Thus, this analysis identified two discrete and closely mapped sites of minimal deletions at 5p15.3 and 5p15.2-15.3, which were restricted to 5.5 cM and 7 cM genetic distance, respectively.

Gene expression analysis of candidate tumor suppressor genes in CC cell lines
In an effort to identify the genes inactivated in the recurrent 5p deletions in CC, we identified candidate genes with tumor suppressor function. Three genes were identified at the terminal 5p15.3 deletion. These include, PDCD6, a protein that plays a role in T cell receptor-, FAS-, and glucocorticoid-induced cell death [11]; TERT, a gene that encodes a reverse transcriptase component of telomerase, dysfunction of which promotes chromosomal instability [12]; and TRIP13, a thyroid hormone receptor interacting protein that binds with the human papillomavirus type 16 (HPV16) E1 protein [13]. We have identified POLS gene that encodes a DNA polymerase sigma, and plays a role in DNA replication and sister chromatid cohesion at the proximal minimal deleted region [14]. We performed a semi-quantitative RT-PCR analysis of all four genes (PDCD6, TERT, TRIP13, and POLS) in eight CC cell lines and did not find evidence for lack of or down regulated expression compared to normal cervix (Fig. 3).
5p deletions in relation to HPV 5p deletions did not show any significant differences with age at diagnosis and stage of the tumor (data not shown). Fifty-eight (96.7%) of the 60 tumors studied were positive for HPV DNA. The remaining two tumors were HPV-neg-

5p deletions occur at low frequency in precancerous lesions
To understand the role of 5p deletions in cervical cancer progression, we evaluated LOH in 26 microdissected CIN specimens (25 high-grade CINs and one low-grade CIN) using four STRP markers ( Table 2). Four of the 25 informative tumors had LOH at one or more markers. One lowgrade CIN did not show LOH. Thus, deletions at 5p identified in high-grade CINs suggest that these genetic alterations occur prior to development of invasive carcinoma.

Discussion
Extensive genetic studies have documented frequent alterations affecting various chromosomal regions in CC. LOH and CGH analyses have consistently shown frequent loss of genomic material in a number of chromosomal regions  suggesting the presence of potential tumor suppressor genes at these sites [4,6,15,16]. Of these, chromosome 5p deletions have been noted to be very frequent in invasive CC [5][6][7]. However, the locations of exact regions of LOH are not known because of a lack of systematic and highresolution deletion mapping of the 5p region in CC. We have previously identified that markers mapped to 5p15 regions are more frequently deleted [4,7]. To further characterize the 5p deletions in CC, we performed a high-density LOH mapping in the present study. We found high frequency of LOH and identified two minimal regions of deletions at 5p15.3 and 5p15.2-15.3.
The 5p15.3 minimal deletion spans between 0-5.5 cM genetic distance. Consistent with this pattern, the markers D5S392 and D5S417 exhibited the highest frequency (53.8% and 53.6%, respectively) of LOH among all the tested loci. The genomic size of the 5p15.3 deletion interval is approximately 3.2 Mb. A database search has identified three candidate genes, PDCD6, TERT, and TRIP13, with suggestive tumor suppressor gene function. The programmed cell death gene, PDCD6, is a member of the family of intracellular Ca2 + -binding proteins and a part of the apoptotic machinery controlled by T-cell receptor (TCR), Fas, and glucocorticoid signals [11,17]. TERT encodes a reverse transcriptase required for the replication of chromosome termini and plays a role in telomere elongation. Although telomerase expression is a hallmark of cancer, the mice lacking the RNA component of telomerase (mTERC) exhibit progressive telomere shortening and chromosomal instability associated with epithelial tumors [12]. Consistent with the established role of this enzyme in cell immortalization and cancer pathogenesis [18], we did not find expression in normal cervix while it was abundantly present in all eight cell lines studied (Fig.  3). Recently, it has been shown that the hTERT amplification and over expression is common in CC [19]. TRIP13, thyroid receptor interacting protein 13, is a transcription factor that regulates expression of a variety of specific target genes including the human papillomavirus type 16 (HPV16) E1 protein [13]. Similar to TERT, the TRIP13 also showed very low expression in normal cervix and highly elevated levels of expression in CC cell lines (Fig.  3). Presence of these genes at the 5p15.3 interval suggests that they may be target TSGs of these deletions in CC. However, expression analysis of these genes did not provide evidence of down-regulation in CC cell lines.
The second site of minimal deletion at 5p15.2-15.3 between 10.7 and 17.7 cM genetic distance spans approximately 2.5 Mb genomic region. A database search has identified POLS, a topoisomerase-related function protein-4-1, gene that encodes polymerase (DNA directed) sigma within the deleted region. The yeast homolog TRF4 plays a critical role in chromosome segregation by coordi-nating between DNA replication and sister chromatid cohesion [14]. Analysis of expression of this gene in CC cell lines did not show evidence of down regulation. Although the genes tested so far showed no deregulated expression, it remains to be seen whether any of these genes or the others mapped to the minimal deleted regions are targets of mutations in CC. However, the present study identified two discrete sites of deletions at 5p15.3 and 5p15.2-15.3 that harbor putative TSGs important in CC development providing a basis for their identification.
We have previously shown, in a different geographical population, that 21% of CINs had 5p deletions. Based on the follow-up studies, we suggested that 5p genetic deletions might identify high-risk lesions for progression to invasive cancer [7]. In a more recent study it was also shown that the 5p deletions were significantly more frequent in dysplasias that were recurrent compared to nonrecurrent dysplasias [9]. These data therefore suggest that 5p genetic alterations occur early in cervical carcinogenesis and they may be critical to the development of invasive cancer. Supporting previous reports, in the present study, we detected LOH at 5p15 in 16% of the high-grade CINs ( Table 2).
The role and mechanisms involved in high-risk HPV induced genomic instability has been of a considerable speculation [reviewed in ref. [20]]. High-risk HPV E6 and E7 gene expression can independently immortalize various human cell types in vitro. The E6 gene interacts with p53 and an apoptotic protein BAK, followed by its degradation, which results in chromosomal instability and resistance to apoptosis [21,22]. The E7 protein interacts with pRB followed by its degradation [23]. The p53 and pRB proteins have been shown to play multiple roles in cell growth and in genomic integrity. There have been contradictory reports on targeted 5p deletions in relation to HPV infection [5,24]. Ku and co-workers have shown that 5p deletions occur in significantly high proportion of CC infected with HPV 16 compared to low-risk HPVtypes [5]. Contradictory to this, another study has shown that 5p deletions are independent of HPV16 and HPV18 infec- tion [24]. In the present study, we did not find any significant differences in 5p LOH between patients harboring HPV16 and other HPV types. However, we found that patients harboring HPV16 in combination with other HPV types showed significant increase in 5p loss compared to patients harboring HPV16 alone. These data, therefore, suggest that multiple high-risk HPV infections may have synergistic effect in causing genomic instability. Thus, currently it is unclear the role of high-risk HPV types in causing 5p genomic instability. Further studies are needed to address this question.

Conclusions
In summary, the present study confirms our previous observations that genetic loss of 5p is a common feature of invasive CC. Here we identified two common sites of deletions that harbor candidate tumor suppressor genes on this chromosomal arm and showed that multiple infection of high-risk HPV types may cause targeted genomic instability at 5p. In addition, we have shown that the genetic deletions on 5p occur early during the development of CC in high-grade CIN lesions. Our identification of minimal deletions on 5p and LOH in CINs has several potential implications. First, it should facilitate the identification of the target genes of importance in cervical tumorigenesis. Second, it may provide critical genetic markers for identifying cervical intraepithelial neoplasias at high-risk to progression.

Tumor and normal tissues
We analyzed 60 tumor biopsies consisting of previously untreated primary invasive CCs with the corresponding peripheral blood samples, and 26 formalin-fixed and paraffin-embedded CIN specimens. The tissues were obtained from patients treated at the Instituto Nacional de Cancerologia, Colombia after appropriate informed consent and approval of the protocols by the institutional review board. The invasive tumors were clinically classified as FIGO stage IB (5 tumors), IIB (14 tumors), IIIB (37 tumors), and IV (4 tumors). Histologically, 57 tumors were classified as squamous cell carcinomas and 3 as adenocarcinomas. Twenty-five CIN specimens were classified as high-grade and one as low-grade lesions. Eight CC cell lines SiHa, SW756, C-4I, Ca Ski, C-33A, HT-3, MS751 and ME-180 were obtained from the American Type Culture Collection (ATCC, Manassas, VA), and were grown in culture according to the supplier's recommendations.

Microdissection, DNA and RNA isolation and LOH analysis
High-molecular weight DNA from frozen tumor and the corresponding peripheral blood specimens, and frozen cells from cell lines was isolated as previously described [25]. Tumor cells from paraffin-embedded CIN tissue specimens were isolated by laser capture microdissection (Arcturus, Mountain View, CA) after methyl green-staining and the extraction of DNA as described earlier [26]. Total RNA was isolated from semi-confluent cell cultures using the TRIzol reagent (Life Technologies, Gaithersburg, MD) according to the manufacturer's protocol. A panel of 20 sequence tagged repeat polymorphic (STRP) (18 diand two tetra-nucleotide) markers was chosen on the basis of their map position and heterozygosity ( Fig. 1) [25,26]. All autoradiograms were visually analyzed. Reduction of signal intensity by 50% or more of one allele over the other allele in tumor DNA compared with the intensity of constitutional alleles was considered LOH. The definition of minimal region of deletion was defined earlier [25]. The LOH analysis was performed at least twice on all tumors with the corresponding markers that define the minimal deletions.

Detection of HPV types
The PGMY09/11 L1 consensus HPV L1 primers and reverse line blot hybridization system was used to detect 38 genital HPV types as described earlier [27,28]. Appropriate control experiments were set up using bulk master mix components to eliminate potential contamination. Individual assay sensitivity was analyzed by the use of serial dilutions of SiHa cell line crude cell digest, targeting 10 4 , 10 3 , 10 2 , 10 1 , and 10 0 input copies of HPV-16. A no template negative control was set up in all experiments. The PCR products were run on 1.5-% agarose gels, visualized by ethidium bromide staining and quantitated using the Kodak Digital Image Analysis System (Kodak, New Haven, CT).