| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Oncology, Markers, Clinical Correlates |
1 Westmead Institute for Cancer Research, University of Sydney at Westmead Millennium Institute, Westmead, New South Wales, Australia, and
2 Department of Cell and Molecular Biology, University of Technology, Sydney, New South Wales, Australia
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
We previously examined X inactivation status in a cohort of common benign nevi but found loss of heterozygosity (LOH), targeting the inactive X chromosome at ARTR, in 4 of 12 metastatic melanomas that had been included as positive controls for neoplasia (3) . We therefore undertook to construct a deletion map of the X chromosome to locate putative TSG(s) or oncogene(s) relevant to melanoma and explain the paradoxical deletion events involving the AR region.
In recent years, numerous reports have implicated X chromosome deletions and amplifications in breast (4 , 5) , ovarian (6) , prostate (7) , and testicular cancer, with a putative tumor suppressor defined to Xq27 (8) . The AR, as well as being a useful marker for X inactivation and clonality assays, has been directly implicated in a number of hormone-dependent neoplasms. Specific high-level Xq11–12 gains, which include AR, have been reported in metastatic prostate cancer (7) . However, aberrant methylation of CpG islands in prostate cancer cell lines has also been reported to be associated with loss of AR expression (9) . Loss of the active allele at ARTR, with, by implication, loss of AR expression, has been reported in female breast cancer (5) . Studies of ovarian carcinomas report exclusive loss of the inactive chromosome. Cheng et al. (6) report frequent interstitial loss of Xq12, including AR, in low metastatic potential tumors. Choi et al. (10) , in a study of advanced ovarian carcinomas, report frequent Xq25–26.1 deletions involving the inactive allele. In a study of borderline and invasive epithelial ovarian tumors, Edelson et al. (11) report a 1-cM region flanking AR as the most frequently deleted, whereas in borderline tumors, the nearby proximal locus DXS1194 was more frequently deleted.
In neoplasia showing LOH at ARTR, it is possible, by applying X inactivation analysis, to further determine whether the active or inactive gene copy is targeted by LOH. In the case of the active gene copy being targeted by LOH, digestion of the remaining tumor DNA with a methylation-sensitive restriction enzyme such as HpaII is expected to cause little further change in observed allelic ratios after PCR amplification of ARTR. This is because the remaining tumor DNA of such a neoplasm with LOH will be hypermethylated and will not be susceptible to enzyme digestion. In the case of LOH targeting the inactive gene copy, a different pattern is expected to result. In this case, the remaining tumor DNA will consist of predominantly unmethylated DNA, and digestion with HpaII should result in degradation of all remaining tumor DNA. Any signal obtained from ARTR will be of markedly lower intensity, being due only to residual methylated stromal DNA which will generally show random X inactivation status. This, however, should not be assumed because stromal tissues may also show skewed X inactivation, with blood lymphocytes from elderly women being a particularly dramatic example (12) .
In this study, we also carried out analysis of copy number at ARTR to shed light on the mechanism by which the inactive chromosome had been targeted. LOH data alone are insufficient to do this because a number of molecular mechanisms may give rise to a similar observed allelic shift (13) . Chromosomal loss in tumors is frequently followed by replacement with duplication of the remaining allele. Alternatively, gene amplification events may produce an allelic imbalance that mimics LOH. Because both mechanisms are possible explanations of our findings, we undertook gene dosage analysis using a multiplex PCR method based on that of Poropat et al. (14) .
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
LOH Analysis.
Twenty-three microsatellite markers spanning the whole X chromosome were studied with markers designed to provide a comprehensive deletion map (see Fig. 1
). Marker locations were checked using the National Center for Biotechnology Information and the University of California at Santa Cruz Biotechnology human genome portals. Primer sequences were obtained from the Genome Database.3
PCR primers were purchased from Sigma-Genosys, (Australia), fluorescent forward primers were 5'-HEX-labeled.
|
Scoring of LOH.
Computer software provided with the Gel-Scan 2000 gives real-time plots of signal voltage versus gel retention time. The package allows analysis to determine product size and the area under the curve, which corresponds to product yield for LOH calculations. We used the formula of Cawkwell (18)
with a change in allelic ratios of >30% in duplicate samples scored as LOH. This criterion is designed to permit LOH detection in tumor samples with up to 50% stromal contamination. Additionally, we have distinguished samples that showed LOH with an allelic change of >50% from those that showed a less dramatic allelic shift in the 30–50% range. Weaker LOH values may have a number of possible causes, such as stromal contamination and genetic heterogeneity of the tumor tissue sampled. Flagging of such samples with weaker LOH values (represented in lowercase) gives a measure of robustness of the deletion map.
X Inactivation Analysis.
For X inactivation analysis, the test DNA from tumors with LOH at AR was split into an untreated aliquot and another aliquot subjected to digestion with the methylation-sensitive restriction enzyme HpaII (Promega Corp.). Digests were conducted at 37°C overnight, and the enzyme was heat inactivated (95°C, 10 min) before PCR of the ARTR locus. Relative allelic signal ratio before and after digestion of both normal and lesion tissue was compared. The X inactivation assay algorithm is analogous to that used for LOH analysis, and a >30% change in allelic ratio in at least duplicate samples of undigested and HpaII-digested tissue DNA was predetermined to indicate that the remaining chromosome was in fact the active X.
Gene Dosage Determination.
In this assay, we sought to study whether in female patients with metastatic melanoma, LOH at the ARTR locus occurred via a mechanism that resulted in a change in gene dosage of the retained chromosomal copy. The protocol is based on the published method of Poropat et al. (14)
. Two single copy genes (target and reference) were amplified by PCR: a 463-bp product comprising exon 4 of the AR gene was coamplified with a reference sequence [283 bp comprising exon 2 of the ß-globin (BG) gene located on 11p and unlikely to be deleted in melanoma]. The BG exon 2 forward primer used was 5'-CTCTGCCTATTGGTCTATTTTCCC-3', and the reverse primer used was 5'-GAAAACATCAAGGGTCCCATAGAC-3'. The PCR was initiated by denaturation (95°C, 3 min) followed by 28 cycles of denaturation (95°C, 30 s), annealing (60°C, 45 s), and extension (72°C, 45 s), followed by a final extension of 72°C for 10 min. PCR reactions were performed as quintuple replicates in a 10-µl final volume and contained 20 pmol of each primer, 0.2 mM each deoxynucleotide triphosphate, 1 unit of Taq DNA polymerase (Amersham), proprietary buffer adjusted to a final MgCl2 concentration of 2.5 mM, 1 µg/µl BSA (17)
, and 5 ng of DNA sample. PCR reactions included healthy male and female control samples, and each tumor sample was matched against corresponding constitutional DNA. Male and female healthy control reactions were included on each gel. A one third volume of nondenaturing loading dye containing bromphenol blue was mixed into each sample, and 2 µl were loaded onto a 5% (29:1) nondenaturing polyacrylamide gel and electrophoresed in 0.5x Tris-borate EDTA (700 V, 35°C) on a Corbett Gel-Scan 2000 DNA Analyzer, which allows quantitative analysis of PCR product yields in a manner analogous to that used in LOH analysis.
Statistical Analysis.
The ratio of target:reference (AR/BG) PCR product yield for five replicates of male and female control samples and five replicates of malignant melanoma samples (along with constitutional DNA pairs) was studied. For each set of five replicate samples, the following statistics were determined: mean; SD; and 95% confidence interval (95% confidence interval, 1.96 SD/
). The population mean and 95% confidence bandwidth were also determined for all female control samples.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
shows a deletion map of the X chromosome obtained using 23 polymorphic microsatellite markers in 29 female subjects with metastatic melanoma. One (tumor 1082) was homozygous for all markers examined, presumably due to Turner syndrome, and is not counted among informative tumors below. Ten (10 of 28, 36%) tumors showed LOH of at least one marker. Six of these showed LOH in more than one region, whereas LOH was found at all informative loci in two tumors. The most frequent losses were found at DXS1061 at Xp21–22 (7 of 20 tumors, 35%), DXS991 and DXS1190 at Xp11 [6 of 19 tumors (31.6%) and 6 of 18 tumors (33.3%)], and DXS1053 at Xp22 (4 of 17 tumors, 23.5%).
|
Four smallest regions of overlap of deletions could be resolved on the X chromosome in females with metastatic melanoma. Three tumors (tumors 1079, 1380, and 1915) define an 18.6-Mb region between the Xp markers DXS999 and DXS1068 with 9 of 28 (32%) of samples implicated. An equally frequent smallest region of overlap straddled the centromere, being bounded by DXS1204 on the p arm and DXS983, 14.6 Mb telomeric in Xq11–12. In tumor 1079, the total signal from marker DXS1190 was reduced to a level 10–20% of that observed in the flanking markers DXS991 and DXS1213, both of which showed LOH. DXS1190 is therefore very likely to be homozygously deleted in that tumor. One tumor (tumor 1942), bearing a single deletion in this region, potentially further defines the smallest region of overlap to a 10.6-Mb interval between DXS1190 and DXS981. An additional Xp smallest region of overlap involving 8 of 28 (28%) tumors can be defined between markers DXS8051 and DXS9902. This region is defined by the tumors 1079 distally and 1380 proximally. Tumor 1088 contains an Xp deletion that straddles these smallest regions of overlap. Inclusion of this information potentially further reduces the Xp smallest region of overlap to either the interval between DXS987 and DXS9902 distally or between DXS999 and DXS1061 proximally. Finally, 5 of 28 tumors define a 25.7-Mb region on distal Xq defined by the markers DXS1212 proximally and DXS1193 distally.
Deletion and X Inactivation at AR.
The AR gene lies within the pericentromeric smallest region of overlap defined by DXS1190 and DXS983, though it was not included in the homozygous deletion bounded by DXS991 and DXS1213.
Six of 22 informative samples studied for deletions at ARTR showed LOH (22.7%), and X inactivation analysis was carried out on five of these samples. The remaining sample (tumor 1088) showed marginal LOH with only 35% average allele reduction compared with blood DNA. Samples of HpaII-digested tumor DNA showed reduced ARTR PCR product yield relative to undigested DNA. Additionally, where one allele was reduced by the LOH event, the remaining allele was always reduced in the X inactivation assay. This indicates that the hypermethylated, inactive allele had been deleted by the LOH event in each case, a result with only a 3% probability of occurring by chance.
One of these samples (tumor 1163) is illustrated in Fig. 3
; it shows a shift in allelic ratios of 
80% due to LOH and a similarly large reduction of the remaining active, unmethylated allele in X inactivation analysis. This outcome is compared schematically with the expected profile that would be obtained for a sample with LOH of the active allele. In such a case, tumor DNA, contributed by the active, unmethylated allele would already be reduced by LOH and would no longer be available for enzyme digestion in the X inactivation assay. As a consequence, little change would be expected in allelic ratios in tumor DNA both before and after the X inactivation assay.
|
. AR/BG ratios were standardized against a male control sample and set arbitrarily at 1.0. Healthy female control samples with theoretical X chromosome gene dosage of 2 had a mean AR/BG of 1.84, approximately as expected.
|
2 in the deleted regions accompanied by preservation of the activated methylation status of the duplicated/amplified alleles. |
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The Xq11–12 region containing the AR gene is frequently amplified or deleted in a number of cancer types (5, 6, 7) , and this has also been found to be true in this study of female melanoma patients. In this case, LOH analysis has defined a 14.9-Mb region straddling the centromere, between the microsatellites DXS1204 and DXS983. Two tumors with restricted deletions (tumors 1088 and 1603) narrow the p arm region to DXS1204 with a q arm boundary at DXS983, and this region is supported by retention of heterozygosity in two tumors (tumors 1079 and 1603). Several other tumors suggest the region can be further narrowed, perhaps as far as the DXS1190 to DXS981 interval defined by tumor 1942. The importance of this interval is highlighted by a probable homozygous deletion of DXS1190. The DXS1204 to DXS983 region is relatively gene poor and of course includes the centromeric gap in which completeness of the physical map and genome sequence cannot be assumed. For reasons explained below, TSGs and/or oncogenes may have a role to play with the following known genes of particular relevance: cyclin B3 (CCNB3) and the melanoma antigen MAGED2 between DXS1039 and DXS1991; a ras-like GTPase (RAGB) and ubiquitin ligase (UBQLN2) between DXS991 and DXS1190; the Cdc42-interacting guanine nucleotide exchange factor intersectin-1 (ARHGEF9) between DXS1190 and DXS1194; and, finally, the AR between DXS1194 and DXS983.
In this Xq11–12 region, we have characterized the changes in copy number and parental chromosome associated with these deletion events to interpret the likely mechanisms of selection for them. The assay used (14) was previously validated in the context of copy number changes in the germ line, successfully demonstrating integer differences in gene dosage. The tumor samples here contain variable amounts of stromal cells, and the tumor cells themselves potentially show different changes in copy number in different subpopulations. We obtained a level of signal:noise similar to that of Poropat et al. (14) in our modified version of their assay. Analysis of copy number at the ARTR locus indicated that the tumors with LOH show either no change or an increase, in one case, to three copies. The signal in most of the tumors was lower than an exact multiple of half the (diploid signal) value obtained from constitutional DNA at the same time. This suggests that these tumors were composed of a mixture of cells in which a minority had copy number reduction of ARTR.
The AR gene itself may confer a proliferative advantage to melanoma tumor cells in female patients and may be the target of selection in LOH events. However, it need not be the only candidate because cyclin B3 is also in the region affected. Malignant melanoma is generally considered to be a hormone-independent tumor, and clinical trials of endocrine treatment against estrogens, androgens, and progestins have shown little response in malignant melanoma (20) . However, an increased mole count often observed during puberty and hypermelanization in pregnant women suggest that melanocytes may respond to steroid hormones, including androgens (21) . In prostate cancer, tumors treated with antiandrogen therapy often show high-level amplification of AR gene copy number (7) , and low level increases in AR gene copy number have been observed in untreated advanced tumors (22) .
An 18.6-Mb proximal smallest region of overlap on the p arm may be resolved with a deletion frequency equal to that in the centromeric region between DXS999 and DXS1068. A number of genes in this region show potential for a role in tumorigenesis. These include PPEF1, a serine-threonine protein phosphatase with an EF-hand calcium-binding domain involved in sensory neuron differentiation; RUK, regulator of ubiquitin kinase; and PRDX4, a peroxidase with high activation of nuclear factor-
B protease.
A 6.2-Mb smallest region of overlap can be defined at Xp22 between the markers DXS8051 distally and DXS9902 proximally. Genes with possible involvement in cancer in this region include ocular albinism 1 (OA1), which regulates pigment production in skin and eyes (23) , Ras-associated protein (RAB9), and epidermal growth factor-like 6 (EGFL6), although meningiomas are the only cancers known to express this gene.
Finally, a 23.9-Mb smallest region of overlap with a lower frequency of deletion (5 of 28 tumors, 18%) was found in distal Xq. This region is defined by the markers DXS1212 proximally and DXS1193 distally. Deletions in this region have been reported previously in prostate cancer (8) , and a number of genes located in this area show potential for a role in cancer. SMARCA1 is a SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin structure. MTS4 is an Ste20-related kinase that activates the MEK/extracellular signal-regulated kinase pathway, and FGF13 is a member of the fibroblast growth factor family expressed in the central nervous system.
These combined data clearly resolve four independent X chromosome smallest regions of overlap. Selective pressures should favor deletions that confer a growth advantage to tumor cell lineages. However, it may be going too far to suggest that all deletions in a tumor, such as tumor 1079 in this study, with three apparently independent regions of deletion, can necessarily be explained in terms of target genes alone. Some deletions arising through chromosomal instability, whether on the X chromosome or elsewhere in the genome, that may confer no growth advantage may be propagated by association with selectively advantageous regions in a growing tumor cell population.
Of particular interest was the observation that collectively, the more broadly defined smallest regions of overlap include most members of the MAGE gene family with the exception of the MAGE-A members on Xq28, a region distal to the study region. Most MAGE genes, with the exception of MAGE-D, are expressed exclusively in tumors and in the germ line (24) . It is therefore possible that these deletions are selected during the tumorigenic process and lead to reduced MAGE gene expression with consequent reduction in immunogenic exposure of melanoma cells during tumor progression.
Knudson’s two-hit hypothesis (25) concerning the inactivation of TSGs states that both parental copies must be inactivated for a tumor cell to be formed or progress. However, LOH (DNA deletion) events are seldom simply reduction to monosomy at the gene or marker concerned. In the study by de Nooij-van Dalen et al. (13) , LOH events were mainly found to be due to somatic recombination or chromosome loss followed by duplication of the remaining chromosome. Other possible mechanisms include simple deletion of a chromosomal segment, nondisjunction (monosomy), loss of a chromosomal region with duplication from the remaining homologous chromosome (gene conversion), or translocation.
We found that every case with LOH at ARTR showed only a small residual signal after X inactivation analysis, in which the allelic ratio suggested random (polyclonal) X inactivation, presumably from contaminating adjacent stromal tissue. The fact that all samples showed the same pattern of LOH and X inactivation is evidence that LOH occurred via a similar mechanism in all these samples. This leads us to the conclusion that only one basic mechanism can explain these results: an inactive allele of the AR region was targeted by LOH and subsequently replaced by the remaining active allele. A mechanism involving tandem duplication may account for further gene amplification. We found that the AR gene copies present in the tumor after the LOH event(s) were unmethylated and therefore active, at least doubling the number of active copies. These findings suggest that one of the selective pressures favoring the LOH event is, paradoxically, selection for increased activity of a dominantly acting putative melanoma oncogene. Could observed deletions target a TSG? The Knudson two-hit hypothesis seems incompatible with targeting of a chromosome that expresses few, if any, genes. Thus a further hypothesis could be that a putative TSG is expressed from the "inactive" chromosome at the stage of tumorigenesis at which these deletions occur. Furthermore, they may be preceded by breakdown of X inactivation brought about by prior oncogenic mutations elsewhere in the genome or in X itself. The absence of LOH events targeting the same putative TSG on the active chromosome could be explained in three ways. The first is that the initial preferred mode of activation of the TSG may be point mutation or that there may be a closely neighboring essential gene that leads to death of clones with deletions of the active chromosome in this region. Alternatively, the initial event may be an activating oncogenic point mutation on the active chromosome. We have shown that one outcome of these events may be increased copy number of the active chromosome, with potential for increased dosage of dominantly acting oncogenes. Further mapping of these complex events and identification of the genes responsible will permit direct testing of these hypotheses.
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: James Indsto, Westmead Institute for Cancer Research, University of Sydney at Westmead Millennium Institute, Darcy Road, Westmead, New South Wales 2145, Australia.
Received 6/ 4/03; revised 9/ 2/03; accepted 9/ 3/03.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  I. Ganmore, G. Smooha, and S. Izraeli Constitutional aneuploidy and cancer predisposition Hum. Mol. Genet., April 15, 2009; 18(R1): R84 - R93. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
