• ovarian cancer;
  • tumor suppressor gene;
  • quantitative real-time RT-PCR


  1. Top of page
  2. Abstract
  6. Acknowledgements


Loss of heterozygosity on chromosomal band 8p22 is a common event in several epithelial tumors including ovarian carcinoma. So far, no clear evidence for a tumor suppressor gene (TSG) in this region has been found.


On the basis of publicly available expression data in ovarian tissues, the authors selected the eight most noteworthy genes from 8p22 (DLC1, N33, ZDHHC2, FLJ32642, PDGFRL, MTSG1, PCM1, and EFA6R) for a detailed expression analysis in 58 primary ovarian carcinoma tissues and in 38 ovarian cancer cell lines by using quantitative real-time reverse transcriptase-polymerase chain reaction (qRT-PCR). Expression data were correlated to various clinicopathologic characteristics and survival.


Two genes showed a significantly (P< 0.05) lower expression in grade 3 tumors compared with tumors of lower grade (N33) or compared with normal controls and tumors with lower grade (EFA6R). Expression of N33 and EFA6R seems to have an impact on survival, in particular when the combined expression of both genes was used as predictive factor (P< 0.003). In addition, N33 and EFA6R showed a complete loss of expression in several ovarian cancer cell lines. Three genes (FLJ32642, MTSG1, and PCM1) had a significantly (P< 0.001, P< 0.004, and P< 0.001) lower expression in primary ovarian carcinoma compared with controls (ovarian tissues and cysts).


Two to five new potential tumor suppressor or antagonizing gene candidates (N33 and EFA6R with impact on survival, and potentially FLJ32642, MTSG1, and PCM1) for ovarian carcinoma, were identified from the chromosomal band 8p22 and are promising candidates for further functional analysis in ovarian carcinoma. Cancer 2005. © 2005 American Cancer Society.

Ovarian carcinoma is the most lethal gynecologic malignancy and the fourth most frequent cause of carcinoma-related death of women in western countries. The majority of evidence suggests that reproductive factors and heredity may play roles in the origin of ovarian carcinoma. About 10–15% of cases are based on a highly penetrant autosomal dominant genetic predisposition. In this hereditary form of ovarian carcinoma, the cloning of BRCA1,1BRCA2,2 and the human DNA mismatch repair genes like MLH1,3MSH2,4 and, consequently, the identification of germ line mutations in disease-prone family members has brought some insight into the early genetic deregulations of ovarian carcinogenesis. However, the further progression to overt cancer in hereditary epithelial ovarian cancer and the genetic basis of the much more common so called “sporadic epithelial ovarian cancer” is less understood.

Studies of human ovarian carcinoma specimens have revealed several types of genetic alterations. Mutation of the p53 tumor suppressor gene (TSG) is the most frequently identified alteration in serous and poorly differentiated epithelial ovarian carcinoma, affecting greater than 50% of advanced and early stage carcinomas. Protooncogenes such as c-myc, K-ras, AKT, and members of the EGF/ErbB family of receptor tyrosine kinases, whose products are all involved in the control of growth stimulatory and cell death pathways, were found amplified or mutated in ovarian carcinoma.5 Most of the evidence that these oncogenes and TSGs are involved in ovarian carcinoma carcinogenesis is based on immunohistochemical examination of tumors with candidate oncogenes originally identified in other tumor entities. In particular, overexpression of the epidermal growth factor (EGF) receptor has been detected by immunohistochemistry in a high percentage of ovarian carcinoma specimens and has been correlated with poor prognosis.6HER-2/neu is overexpressed in about 30% of ovarian malignancies and also appears to indicate poor clinical prognosis and poor survival.7

Consequent workup of genetic abnormalities has been rare and, if performed, challenging in ovarian carcinoma. Loss of heterozygosity (LOH) on chromosomal band 8p22 in ovarian tumors is relatively high, as shown in several studies8–13 including a metaanalysis by us (unpublished data). Taken together, these observations indicate the presence of one or more tumor suppressor or antagonizing gene(s) in this region. Despite great efforts, no clear evidence for a TSG in this region could be worked out so far for ovarian carcinoma.11, 12, 14

In a previous study, we evaluated all 22 known genes on this chromosomal band using all available public gene profiling data and other published information to select the most probable TSG candidates. On the basis of these previous investigations, we identified eight genes worthy of further examination, namely DLC1, N33, ZDHHC2, FLJ32642, PDGFRL, MTSG1, PCM1, and EFA6R. All other genes were either not expressed in ovarian tissues, stably expressed in all ovarian carcinoma samples profiled (ASAH1), and/or had (proposed) functions making them very unlikely TSG candidates (ASAH1). In Figure 1, a scheme of the region under examination is shown outlining the relevant information for the individual genes.

thumbnail image

Figure 1. Genetic map of the chromosomal region on 8p under investigation with information about expression in ovarian tissue.

Download figure to PowerPoint

The aim of this study was to narrow down the list of TSG candidates for ovarian carcinoma on chromosomal band 8p22. We quantified the expression of eight selected genes in a considerable number of primary ovarian carcinoma cell lines and tumor specimens and correlated it to various clinicopathologic characteristics and survival.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Normal Ovarian Tissues, Tumor Samples, and Tumor Cell Lines

For the expression analysis, we used 39 tumor samples obtained from patients undergoing initial staging or debulking laparotomy at several major teaching hospitals in Sydney and Brisbane, Australia, and 17 tumor samples obtained from patients undergoing surgery at the Medical University of Vienna, Austria. The estimated percentage of tumor tissue in the clinical samples was above 80%. In Table 1 a comprehensive list of all 58 tumor samples and their clinicopathologic characteristics is shown. As reference samples we used six normal ovaries (the mean age of the patients was 56 yrs, standard deviation [SD] 14.6 yrs) and two ovarian cysts (patients aged 38 and 84 yrs) from patients without history of malignant disease at the Medical University of Vienna. Normal ovaries were dissected for enrichment of epithelial tissue. In addition, 32 ovarian cancer cell lines from 28 individual patients established at the University of Ulm, Germany, were assayed (Table 2). These cancer cell lines were established from ovarian carcinoma tissue specimens and cultured as described.15 Six commercially available ovarian cancer cell lines were obtained either from the European Collection of Cell Cultures (ECACC, Salisbury, Wiltshire, UK), A2780 and A2780 ADR, or from the American Type Culture Collection (ATCC, Manassas, VA), MDAH-2774, OVCAR3, ES2, and HTB-77 (SKOV3).

Table 1. Clinicopathologic Characteristics of 58 Primary Ovarian Tumor Samples
Age in yrs5861.914.0
Age group   
 ≤ 50 yrs1039.69.5
 > 50 yrs4866.69.5
 w/o relapse28  
 Clear cell2  
Stage (FIGO)   
Table 2. Clinicopathologic Characteristics of 32 Ovarian Cancer Cell Lines from 28 Patients
  • a

    Numbers in parenthesis indicate duplicate cell lines from one patient, isolated at different time points or from a different source.

  • b

    FIGO stage was III or IV, except for one.

Age in yrs2359.810.9
Age group   
 ≤ 50 yrs545.22.6
 > 50 yrs1863.98.6
 Serous20 (4)a  
 Clear cell1  
 214 (2)  
 34 (2)  
 Ascites17 (2)  

Patient material used in this study comes from the University Hospital, Vienna, from the University of Ulm, Germany, and from collaborating hospitals in Australia. Informed consent for the scientific use of biologic material was obtained from all patients in accordance with the requirements of the ethics committee of the individual institutions.

RNA Isolation and cDNA Synthesis

RNA from primary ovarian tumors was isolated with TRIzol Reagent (GibcoBRL, Life Technologies, Rockville, MD) and from the six commercially obtained cell lines with the RNeasy Mini Kit (QIAGEN, Hilden, Germany) as described by the provider. To avoid the influence of tumor inherent stromal tissue on gene expression, all samples were dissected carefully by a pathologist before RNA extraction. RNA from the cell lines established at the University of Ulm was prepared using isopycnic centrifugation in a CsTFA solution.16 The cDNA synthesis was performed after a DNase I treatment from 0.5 μg total RNA (primary material) or 1 μg total RNA (cell lines) with the Enhanced Avian HS RT-PCR kit (Sigma-Aldrich, St. Louis, MO) with an 1:1 mixture of a random nonamer and an anchored oligo-(dT)23 primer provided in the kit according to the technical bulletin.

Quantitative Real-Time Polymerase Chain Reaction (RT-PCR)

Assay-on-Demand probes for the TaqMan real-time PCR system from Applied Biosystems (Foster City, CA) for the eight genes and the internal house-keeping control-gene, β-2–microglobulin (B2M), were as follows: DLC1, Hs00183436_m1; N33, Hs00185147; ZDHHC2, Hs-00275319_m1; FLJ32642, Hs00329751_m1; PDGFRL, Hs00185122_m1; MTSG1, Hs00368183_m1; PCM1, Hs00196390_m1; EFA6R, Hs00209633_m1; and B2M, Hs99999907_m1. Real-time PCR of 10 ng of the cDNA mixture of all probes in a volume of 20 μl were obtained in duplicates with the GeneAmp 5700 Sequence Detection System (Applied Biosystems) with cycle conditions as follows: Initially 50 °C for 2 minutes and 95 °C for 10 minutes followed by 40 cycles of 95 °C for 15 seconds and 60 °C for 1 minute. Relative expression (compared with a calibrator, a cDNA mixture of several cell lines) of all probes was calculated from the threshold cycles (Ct) obtained with the GeneAmp 5700 SDS Software v1.3 (Applied Biosystems) as follows: 2–[(Ctgene mean of duplicated probes – Ctgene mean of duplicated calibrator) – (CtB2M mean of duplicated probes – CtB2M mean of duplicated calibrator)].

Statistical Analysis

For all analyses described below, the logarithm of gene expression has been used, so that the use of parametric models is justified. For comparison of gene expression between normal and cell line, a one-way ANOVA was performed; P values were corrected for multiplicity (eight candidate genes) by the method of Bonferroni–Holm. For comparison between normal and tumor tissue, a nested variant of the indicator showing the origin of the tumor sample (Vienna or Australia) was included for adjustment in the ANOVA model.

To compare two groups of different grading levels in tumor samples and the group of normal tissue, the following two-step procedure has been applied. First, the indicator distinguishing among these three groups and the nested variant of the origin indicator were used in an ANOVA for each gene. For those genes, which, after correction for multiplicity, showed a significant overall difference between the three groups, in a second step all three pair-wise comparisons between the groups were performed and judged without further correction at the 5% level (according to the closed testing principle within each gene). To investigate the association between expression levels of the eight candidate genes, Pearson correlation coefficients of the logarithms were computed for the tumor samples.

Comparison of the logarithm of gene expression among various groups was performed using one-way ANOVA and two-sample t-tests, respectively.

To investigate the influence of the logarithm of gene expression on survival, Cox regression has been applied for each gene first with a univariate model and then adjusted for grading (two groups) and stratified for sample origin. None of the interactions of any pair of gene expression levels reached statistical significance in a model that also contained the corresponding main effects. For a graphic representation by means of Kaplan–Meier estimation of survival curves, gene expression levels have been dichotomized for each gene at the median and, alternatively, at the 33.33% quantile. P values of the corresponding log-rank tests are given. The survival analyses have not been adjusted for multiplicity because of their exploratory character.

P values ≤ 0.05 were considered to be statistically significant. All computations have been performed using SAS software Version 9.1 (SAS Institute Inc., Cary, NC, 2001). Boxplots and Kaplan–Meier plots were made with the SPSS software Version 12 (SPSS Inc., Chicago, IL, 2003).


  1. Top of page
  2. Abstract
  6. Acknowledgements

Expression of Candidate mRNAs in Primary Ovarian Tumor Samples

Relative expression levels of eight potential tumor suppressor gene candidates from chromosomal band 8p22, DLC1, N33, ZDHHC2, FLJ32642, PDGFRL, MTSG1, PCM1, and EFA6R, were measured in up to 58 primary ovarian tumor samples and 8 epithelial enriched normal ovarian tissues (6 ovaries and 2 cysts) by using quantitative real-time RT-PCR. One gene (ASAH1) was excluded from our analysis because of its relative stable expression in the gene profiling metaanalysis and its presumed function.17

By comparing expression profiles from tumor samples with expression profiles from corresponding so called “normal” tissues, it has to be considered that tumors consist and derive mainly from one cell type (e.g., epithelium), but normal tissue consists of several different cell types, often with an uneven quantitative distribution (e.g., between epithelium and stroma). To address this problem of comparing expression data from primary tumor material (of epithelial origin) and the normal ovary (mixture of stroma and epithelium), two groups (Su et al.18 and Welsh et al.19) compared gene expression of epithelial-enriched with stromal-enriched normal ovarian tissue by using Affymetrix GeneChips (Santa Clara, CA). Table 3 summarizes results after analysis of raw data with GeneSpring software (Agilent, Palo Alto, CA). About 10% of the (reliable) expressed genes are differentially expressed in the ovarian stromal-enriched compared with the epithelial-enriched tissue. From the eight analyzed genes, two have comparable high expression in epithelial- and stromal-enriched ovarian tissue, MTSG1 and EFA6R. Two genes, N33 and PCM1, show higher expression in the epithelial-enriched tissue, about 2.8- and 1.7-fold, respectively, and two genes, DLC1 and PDGFRL, show a slightly higher expression in the stromal-enriched tissue. For FLJ32642 and ZDHHC2 no data were available.

Table 3. Different Expression of Genes in Stromal and Epithelial Enriched Normal Ovarian Tissues
GeneChipProbe setsDifferentially expressed (threefold)
Stroma > EpitheliumStroma < EpitheliumTotal
  • a

    Reliably expressed probe sets (7810 of 25,276; 30.9%) filtered according to GeneSpring v6.0 software using base/proportional from the cross-error model.


Figure 2 compares gene expression in primary tumor samples (PT), tumor cell lines (CL), and epithelial-enriched nonmalignant ovarian tissues (N). FLJ32642, MTSG1, and PCM1 showed significantly lower expression in primary tumor samples compared with normal controls (corrected for multiplicity, see Fig. 2). EFA6R is also expressed at a lower level but does not reach statistical significance in our data. Results of expression levels in ovarian cancer cell lines were less clear and only PCM1 had a significantly lower expression in the cell lines compared with normal controls (corrected P= 0.028).

thumbnail image

Figure 2. Boxplots of the expression of eight genes in normal ovarian tissue samples (N), primary ovarian tumor samples (PT), and ovarian cancer cell lines (CL). Expression is given in arbitrary units based on the expression of β-2–microglobulin as internal house-keeping gene control. P values are corrected for multiplicity. *Significant after correction for multiple testing.

Download figure to PowerPoint

Grading by morphologic appearance is a widely used clinical parameter in ovarian carcinoma. Tumors range from well differentiated (grade 1) to moderately differentiated and predominantly glandular (grade 2) to poorly differentiated and predominantly solid (grade 3) subtypes.20 One well supported possibility to interpret the relevance of grading for tumor biology is that its features reflect the progressive development of well differentiated to poorly differentiated cells. We subdivided the samples in three groups, namely normal tissues, tumors of grade 1 or 2, and tumors of grade 3. An ANOVA test among these three groups found five genes, N33, FLJ32642, MTSG1, PCM1, and EFA6R differentially expressed. For details see Figure 3A. These five genes can be subdivided into two groups, one whose expression is significantly lower in both grading groups of tumor samples compared with normal samples—early affected genes (FLJ32642, MTSG1, and PCM1)—and one whose expression is significantly lower only in tumors of grade 3 compared with normal and tumors with grade 1 or 2 (EFA6R) or compared only with tumors of grade 1 or 2 (N33)—late affected genes.

thumbnail image

Figure 3. Boxplots of the expression of (A) the five genes with significant different expression in tumors of different grades and/or normal ovarian tissue samples (two-step ANOVA test, corrected for multiple testing) and (B) different expression of N33 in tumors of different FIGO stages.

Download figure to PowerPoint

A one-way ANOVA test found no significant differences in expression for histologic types or FIGO stages (classified in FIGO I, II, III, and IV). However, a boxplot showing the expression of N33 in tumors from patients of different FIGO stages indicates an expression change between FIGO stages I–II and III–IV (Fig. 3B, uncorrected P= 0.017). FIGO stages beyond II are characterized by the occurrence of (micro)metastases. There was no association between age of patients at the time of surgery and the logarithm of gene expression. Pearson correlation coefficients are below ± 0.22 for all genes. Two-sample t-tests comparing the logarithm of expression between patients ≤ 50 years of age (presumed premenopausal) and patients > 50 years (presumed postmenopausal) were not significant (data not shown). In addition, the association between the expression levels of the eight genes was analyzed. About half of the correlation coefficients between the expression levels of the eight candidate genes for the tumor samples showed values higher than 0.50 and EFA6R, MTSG1, FLJ32642, and PCM1 showed coefficients around 0.75 between them (data not shown).

Expression of Candidate mRNAs in Ovarian Cancer Cell Lines

In Figure 4 the expression patterns of the 8 TSG candidates in 38 ovarian cancer cell lines are shown. For all graphs, the logarithmic scale for the y-axis is the same, allowing comparison of the high variability of expression between the individual genes, ranging from three decimal powers for the genes FLJ32642 and PCM1 to seven decimal powers for the gene DLC1. For five genes, we found cell lines with no measurable expression (i.e., no PCR product after 40 cycles in the GeneAmp Sequence Detection System), namely DLC1 (1 cell line), N33 (5 cell lines), ZDHHC2 (1 cell line), MTSG1 (1 cell line), and EFA6R (4 cell lines), marked with arrows in Figure 4. This complete loss of expression can be explained by genetic and epigenetic mechanisms like homozygote deletions or hypermethylation of CpG-islands of promoter regions of genes, a frequent event in down-regulation of TSGs during tumorigenesis.21 There were no significant differences in expression for cell lines derived from ovarian tumors of different grades (grades 1 or 2 compared with 3), different FIGO stages (classified in FIGO III and IV), or source of tumor tissue (Table 2) (one-way ANOVA tests). As for the primary tumor samples, no correlation between age at the time of surgery and gene expression of the tumor cell lines could be found.

thumbnail image

Figure 4. Expression of eight genes in ovarian cancer cell lines. Expression is given in arbitrary units (logarithmic scale) based on the expression of β-2–microglobulin as internal house-keeping gene control. Cell lines with no detectable expression are marked with an arrow.

Download figure to PowerPoint

Loss of Expression of the Five Genes and Survival

Survival analyses by Cox regression model based on loss of expression of the individual genes in tumor samples was done. The influence of none of the candidate genes on survival reached statistical significance neither in a univariate nor in a multiple model adjusting for grading and stratifying for tumor sample origin. For graphic presentation, the expression values of all five down-regulated genes were dichotomized at the median and, alternatively, at the 33.33% quantile, and Kaplan–Meier plots with log-rank tests were calculated. In Table 4 the cut point (median or 33.33% quantile) showing the lower P value (uncorrected) for the log-rank test is given for each of the five genes together with this lower P value. Only for EFA6R did the 33.33% quantile result in a lower P value compared with the median as a cut point. In Figures 5A and 5B, the Kaplan–Meier plots of the two genes that showed some apparent impact on survival are given. As a next step, we analyzed the combined effect of two genes on survival. Therefore, we subdivided the samples (dichotomized at the median) into four groups representing the following expression patterns, both genes low expressed, either one gene low and the other one high expressed, and both genes high expressed. We calculated Kaplan–Meier plots with log-rank tests for all 10 possible gene combinations. Only one pair of genes showed an uncorrected P value below 0.05. The Kaplan–Maier plot of this gene combination, N33 and EFA6R (uncorrected P= 0.003), is shown in Figure 5C, indicating a negative impact on survival in the group where both genes are low expressed compared with the other three groups. For comparison, the impact of grading on survival is shown by means of a Kaplan–Meier plot (Fig. 5D). The Pearson correlation coefficient for the logarithm of the expression of N33 and EFA6R is 0.42.

Table 4. Summary of Tumor Suppressor Gene Candidates on 8p22
GeneExpression differenceMissing expression in cancer cell linesPutative function
Normal vs. primary tumoraANOVA test (grading-groups)abEarly or late affectedacImpact on survivald
  • a

    Corrected for multiple testing.

  • b

    Groups: (1) normal ovary, (2) primary ovarian tumors with grade 1 or 2, and (3) primary ovarian tumors with grade 3.

  • c

    Early, significantly lower expression in tumors of grade 1 or 2 compared with normal controls; late, significantly lower expressed in tumor of grade 3 compared with tumors of grade 1 or 2 and/or normal samples. For details see Fig. 3.

  • d

    Uncorrected P values for the log-rank test. The samples were dichotomized by using either the median or the 33% quantile of the expression and the lower of the two corresponding P values is given. For the Cox regression analysis see Results.

  • e

    Uncorrected P value for the log-rank test. For details see Fig. 5.

  • f

    Significant at the 5% level after correction for multiple testing.

N33P = 1.000sign.flatefP = 0.0643 (median)13.2%Oligosaccharyl-transferase complex subunit32
FLJ32642P < 0.001fsign.fearlyfP = 0.2920 (median)0.0%Subunit of the endosomal sorting complex required for transport, mediates EGF receptor degradation33, 34
MTSG1P = 0.004fsign.fbothfP = 0.4143 (median)2.6%Mediates the inhibition of ERK2 activation via the angiotensin II AT2 receptor (mouse),35, 36
PCM1P < 0.001fsign.fearlyfP = 0.1009 (median)0.0%Cell cycle-dependent association with the centrosome complex, role in centriolar replication37, 38
EFA6RP = 0.118sign.flatefP = 0.0166 (33% quantile)10.5%Guanine nucleotide exchange factor, GTPase activating protein?
N33 and EFA6R   P = 0.0027e23.7% 
thumbnail image

Figure 5. Kaplan–Meier plots and log-rank tests based on the dichotomized expression of the genes (A) N33 (median), (B) EFA6R (33.33% quantile), (C) the combined expression of N33 and EFA6R, and (D) based on the grading of the tumor samples. P values for the log-rank test are not corrected for multiple testing.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. Acknowledgements

The frequency of LOH on chromosomal band 8p22 in ovarian carcinoma is high, ranging from 38% to 55% as recently reanalyzed and reviewed in our metaanalysis (unpublished). This observation points toward the possibility that the region harbors potential TSGs. As with other LOH regions in various cancer genomes, elucidating the molecular mechanisms further by identifying the responsible genes has been challenging. However, the development of the Human Genome Project and the ample availability of gene expression data in public databases have made a more systematic approach reasonable and possible. Identification of all expressed sequences in the region can be followed by an in silico expression analysis in regular tissues (present or not) as a basis for further quantitative expression analysis and correlation with clinical parameters in tumor tissues.

From all 22 RefSeq genes (RefSeq Release 7, mapped to this relatively gene-poor chromosomal band 8p22, localized between positions 12,700,001 and 19,100,000 in the May 2004 (hg17) assembly (NCBI Build 35) of the UCSC Human Genome Browser (, 9 genes are reliably expressed in regular ovarian tissue (Fig. 1). From these nine genes, five genes FLJ32642, MTSG1, PCM1, N33, and EFA6R showed a significantly decreased expression in tumors of higher grade versus normal ovarian samples or tumors of low grade (Figs. 2 and 3A). In Table 4, the relevant information for these five genes are summarized. Expression of N33 and EFA6R was particularly low in tumors of higher grade, potentially reflecting the loss of these genes as a relatively late event in ovarian carcinogenesis. The expression of these two genes, N33 and EFA6R, had also an apparent impact on survival. The predictive value of the combined expression of these two genes on survival was higher than the predictive value of tumor grading (Fig. 5). Unfortunately, only for one gene (N33) hypermethylation data are available for one tumor entity (colorectal carcinoma), showing that N33 is not hypermethylated in colorectal tumor samples.22

Primary tumor tissues include also cells of the immediate tumor environment, which makes it hard to draw conclusions from expression data for genetic analysis. Cancer cell lines allow investigation of the expression of a clonal cancer cell population without the adjoining normal tissue. Whereas for all genes investigated, expression in primary ovarian cancer cell lines and primary tumors was comparable, the expression of N33 was, in general, higher in cell lines than in primary tumors. N33 could have an antitumor effect, which is not relevant for proliferation in vitro. In combination with the finding that expression of N33 is lower only in tumors of advanced grade (significant) and FIGO stages and in patients with relapse (trends not significant), this may indicate that N33 is involved in later events like metastasis rather than in tumorigenesis.

Expression analysis in ovarian cancer cell lines divided the five down-regulated genes in two groups, one with a rather stable expression (FLJ32642 and PCM1) —difference below three decimal powers—and one with a higher difference in expression (N33, MTSG1, and EFA6R)—above five decimal powers (Fig. 4). In the latter group, at least one cell line had a complete loss of expression with no detectable PCR product (marked with arrows in Fig. 4), indicating a genetic or epigenetic base (silencing by hypermethylation of CpG islands in the promoter region) for the regulation. DNA studies did not show a homozygous deletion (data not shown); however, epigenetic studies and mutation analysis have not yet been done.

From what is known from the putative functions of the gene products, all are promising tumor suppressor gene candidates.

N33 (TUSC3) was first cloned from a homozygous deletion in a metastatic prostate carcinoma.23 In our metaanalysis of gene profiling data, we found supporting information on N33 expression in a gene-profiling study of primary ovarian tumors.19 The authors clustered the 27 tumors by using the gene profiling data in two groups, one more similar to normal tissue (7 tumors) and one more similar to ovarian cancer cell lines (20 tumors). N33 was down-regulated more often (but not significantly, P= 0.204) in the latter cluster (9/20, 45.0%) than in the former (1/7, 14.3%), indicating that N33 expression is more often affected in tumors that are more undifferentiated, which is in accordance with our finding of a grade-dependent expression of N33 (Fig. 3A). Kelleher et al. showed that N33 is a subunit of the oligosaccharyltransferase (OST) complex but is not essential for in vitro function.24

Xu et al. recently described the gene product of FLJ32642 as HCRP1, hepatocellular carcinoma-related protein 1.25 Overexpression of HCRP1 in a hepatocellular carcinoma cell line significantly suppressed proliferation and malignant transformation.26

MTSG1, mitochondrial tumor suppressor gene, formerly called ATIP1 (AT2 receptor-interacting protein 1) and now renamed to mitochondrial tumor suppressor 1 (MTUS1), was recently described as a tumor suppressor gene in the pancreatic carcinoma cell line MIA-PaCa-2 with a suggested function as a regulator of cellular proliferation.27

PCM1, pericentriol material 1, was isolated by exon trapping in the course of searching for TSGs on 8p22.28 A distinct cell cycle-dependent association of the 228 kDa PCM1 protein with the centrosome complex was shown.29 Injection of PCM1 into mouse fertilized eggs can induce cell cycle arrest.30 Armes et al. identified PCM1 as a potential candidate tumor suppressor gene in breast carcinoma.31

EFA6R contains a PH domain (pleckstrin homology), a motif which is located in many proteins involved in intracellular signaling or constituents of the cytoskeleton. The function of this domain is not clear. Several putative functions have been suggested, one of which is the binding to the gamma subunit of heterotrimeric G proteins. Moreover, EFA6R contains a Sec7 domain. Sec7 family proteins are guanine nucleotide exchange factors (GEF), specific for the ADP-ribosylation factors (ARF), a Ras-like GTPase, which is important for vesicular protein trafficking.

In conclusion, we have identified five noteworthy TSG candidates on 8p22 in ovarian carcinoma (N33, FLJ32642, MTSG1, PCM1, and EFA6R) by expression analysis. The expression of N33 and EFA6R was significantly lower in tumors of higher grade (late affected), and their expression was often completely lost in cancer cell lines rather pointing toward a cell autonomous phenomenon on a genetic or epigenetic basis. Loss of expression of both of these genes had a significant impact on overall survival, which was independent from tumor grade and might be a useful prognostic marker in the future. Expression of FLJ32642 and PCM1 is down-regulated or lost, presumably earlier in the course of disease. As discussed above, FLJ32642 is intimately involved in the regulation of EGFR, which is overexpressed in a high percentage of ovarian carcinoma cases. To understand whether this overexpression may be based on the down-regulation of FLJ32642, and, consequently, impaired degradation of EGFR, will be important in the future, particularly because various therapeutic modalities to influence the EGFR are available and make it a promising therapeutic target for ovarian carcinoma. Furthermore, genetic and functional studies are indicated to understand the potential role of these genes in the pathogenesis of ovarian carcinoma and will, hopefully, lead to better diagnostic and therapeutic options for this patient group in the future.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The authors thank Neville F. Hacker (Gynecological Cancer Centre, Royal Hospital Women, Randwick, NSW 2031, Australia) and Michael L. Friedlander (Department of Medical Oncology, Prince of Wales Hospital, Randwick, NSW 2031, Australia) for providing tumor tissue specimens and clinicopathologic characteristics.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  • 1
    Miki Y, Swensen J, Shattuck-Eidens D, et al. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science. 1994; 266: 6671.
  • 2
    Wooster R, Bignell G, Lancaster J, et al. Identification of the breast cancer susceptibility gene BRCA2. Nature. 1995; 378: 789792.
  • 3
    Bronner CE, Baker SM, Morrison PT, et al. Mutation in the DNA mismatch repair gene homologue hMLH1 is associated with hereditary non-polyposis colon cancer. Nature. 1994; 368: 258261.
  • 4
    Fishel R, Lescoe MK, Rao MR, et al. The human mutator gene homolog MSH2 and its association with hereditary nonpolyposis colon cancer. Cell. 1993; 75: 10271038.
  • 5
    Aunoble B, Sanches R, Didier E, Bignon YJ. Major oncogenes and tumor suppressor genes involved in epithelial ovarian cancer. [Review]. Int J Oncol. 2000; 16: 567576.
  • 6
    Berek JS, Martinez-Maza O, Hamilton T, et al. Molecular and biological factors in the pathogenesis of ovarian cancer. Ann Oncol. 1993; (4 Suppl 4 ): 316.
  • 7
    Berchuck A, Kamel A, Whitaker R, et al. Overexpression of HER-2/neu is associated with poor survival in advanced epithelial ovarian cancer. Cancer Res. 1990; 50: 40874091.
  • 8
    Vocke CD, Pozzatti RO, Bostwick DG, et al. Analysis of 99 microdissected prostate carcinomas reveals a high frequency of allelic loss on chromosome 8p12–21. Cancer Res. 1996; 56: 24112416.
  • 9
    Brown MR, Chuaqui R, Vocke CD, et al. Allelic loss on chromosome arm 8p: Analysis of sporadic epithelial ovarian tumors. Gynecol Oncol. 1999; 74: 98102.
  • 10
    Lassus H, Laitinen MPE, Anttonen M, et al. Comparison of serous and mucinous ovarian carcinomas: Distinct pattern of allelic loss at distal 8p and expression of transcription factor GATA-4. Lab Invest. 2001; 81: 517526.
  • 11
    Pribill I, Speiser P, Leary J, et al. High frequency of allelic imbalance at regions of chromosome arm 8p in ovarian carcinoma. Cancer Genet Cytogenet. 2001; 129: 2329.
  • 12
    Wilson PJ, McGlinn E, Marsh A, et al. Sequence variants of DLC1 in colorectal and ovarian tumours. Hum Mutat. 2000; 15: 156165.
  • 13
    Wright K, Wilson PJ, Kerr J, et al. Frequent loss of heterozygosity and three critical regions on the short arm of chromosome 8 in ovarian adenocarcinomas. Oncogene. 1998; 17: 11851188.
  • 14
    Arnold JM, Woollatt E, Chenevix-Trench G. Chromosome 8 genetic analysis and phenotypic characterization of 21 ovarian cancer cell lines. Cancer Genet Cytogenet. 2002; 139: 109114.
  • 15
    Mobus V, Gerharz CD, Press U, et al. Morphological, immunohistochemical and biochemical characterization of 6 newly established human ovarian carcinoma cell lines. Int J Cancer. 1992; 52: 7684.
  • 16
    Kury FD, Schneeberger C, Sliutz G, et al. Determination of HER-2/neu amplification and expression in tumor tissue and cultured cells using a simple, phenol free method for nucleic acid isolation. Oncogene. 1990; 5: 14031408.
  • 17
    Strelow A, Bernardo K, Adam-Klages S, et al. Overexpression of acid ceramidase protects from tumor necrosis factor-induced cell death. J Exper Med. 2000; 192: 601611.
  • 18
    Su AI, Cooke MP, Ching KA, et al. Large-scale analysis of the human and mouse transcriptomes. Proc Natl Acad Sci U S A. 2002; 99: 44654470.
  • 19
    Welsh JB, Zarrinkar PP, Sapinoso LM, et al. Analysis of gene expression profiles in normal and neoplastic ovarian tissue samples identifies candidate molecular markers of epithelial ovarian cancer. Proc Natl Acad Sci U S A. 2001; 98: 11761181.
  • 20
    Silverberg SG. Histopathologic grading of ovarian carcinoma: a review and proposal. Int J Gynecol Pathol. 2000; 19: 715.
  • 21
    Jones PA, Laird PW. Cancer epigenetics comes of age. Nat Genet. 1999; 21: 163167.
  • 22
    Xu XL, Yu J, Zhang HY, et al. Methylation profile of the promoter CpG islands of 31 genes that may contribute to colorectal carcinogenesis. World J Gastroenterol. 2004; 10: 34413454.
  • 23
    Bova GS, Carter BS, Bussemakers MJ, et al. Homozygous deletion and frequent allelic loss of chromosome 8p22 loci in human prostate cancer. Cancer Res. 1993; 53: 38693873.
  • 24
    Kelleher D, Levy A, Bova GS, Gilmore R. Oligosaccharyltransferase isoforms that contain different catalytic STT3 subunits have distinct enzymatic properties. Genomics. 1996; 35: 5565.
  • 25
    Xu ZH, Liang L, Wang HF, Gilmore R. HCRP1, a novel gene that contain different catalytic STT3 subunits have distinct enzymatic properties. Mol Cell. 2003; 12: 101111.
  • 26
    Xu ZH, Liang L, Suzuki S, Li TP, Zhao MJ. TMX, a novel gene that is downregulated in disulfide-linked protein. Arch Biochem Biophys. 2004; 423: 8187.
  • 27
    Momoi T. Caspases involved in ER stress-mediated cell death. FASEB J. 2004; 28.
  • 28
    Xu ZH, Liang L, Koyama K, Nakamura Y. HCRP1, a novel gene to chromosome 8p21.3-p22. Genomics. 2003; 24: 404406.
  • 29
    Xu ZH, Bao L, Zimmer WE. HCRP1, a distinct cell cycle distribution. J Cell Biol. 2003; 311: 783793.
  • 30
    Bache KG. Arrest of mammalian ESCRT-I and mediates receptor down-regulation. Cell. 2004; 15: 183192.
  • 31
    Seibold S, Hammet F, Weber M, Galle J, Wanner C, Marx M. Candidate tumor suppressor genes on chromosome 8p21.3-22. Oncogene. 2003; 17.
  • 32
    Matsuo Y, Nishinaka Y, Suzuki S, et al. TMX, a human transmembrane oxidoreductase of the thioredoxin family: the possible role in disulfide-linked protein folding in the endoplasmic reticulum. Arch Biochem Biophys. 2004; 423: 8187.
  • 33
    Bache KG, Slagsvold T, Cabezas A, Rosendal KR, Raiborg C, Stenmark H. The growth-regulatory protein HCRP1/hVps37A is a subunit of mammalian ESCRT-I and mediates receptor down-regulation. Mol Biol Cell. 2004; 15: 43374346.
  • 34
    Xu ZH, Liang L, Wang HF, Li TP, Zhao MJ. HCRP1, a novel gene that is downregulated in hepatocellular carcinoma, encodes a growth-inhibitory protein. Biochem Biophys Res Comm. 2003; 311: 10571066.
  • 35
    Seibold S, Rudroff C, Weber M, Galle J, Wanner C, Marx M. Identification of a new tumor suppressor gene located at chromosome 8p21.3-22. FASEB J. 2003; 17.
  • 36
    Nouet S, Amzallag N, Li JM, et al. Trans-inactivation of receptor tyrosine kinases by novel angiotensin II AT2 receptor-interacting protein, ATIP. J Biol Chem. 2004; 279: 2898928997.
  • 37
    Dammermann A, Merdes A. Assembly of centrosomal proteins and microtubule organization depends on PCM-1. J Cell Biol. 2002; 159: 255266.
  • 38
    Kubo A, Tsukita S. Non-membranous granular organelle consisting of PCM-1: subcellular distribution and cell-cycle-dependent assembly/disassembly. J Cell Sci. 2003; 116: 919928.