| Nucleic Acids Research | Article |
©1999 Oxford University Press |
Detection of homozygous deletions in tumors by hybridization of representational difference analysis (RDA) products to chromosome-specific YAC clone arrays
Introduction
Materials And Methods
Representational difference analysis
RDA oligonucleotides
Chromosome 13 YAC clone filters
Filter hybridization
Filter analysis
Tumor DNA sample
Comparative multiplex STS analysis
Results
RDA
Filter hybridization
Discussion
Acknowledgements
References
Detection of homozygous deletions in tumors by hybridization of representational difference analysis (RDA) products to chromosome-specific YAC clone arrays
Received July 23, 1999; Revised and Accepted September 17, 1999
ABSTRACT Representational difference analysis (RDA), a subtractive hybridization method that enriches differences between complex genomes, can be used to isolate fragments deleted in tumor genomes. Usually, most of the clones obtained by this approach result from polymorphic fragments. Therefore, identification of homozygously deleted fragments, which can indicate the presence of tumor suppressor loci, is often tedious. To overcome this limitation, we devised a novel strategy in which labeled RDA products are hybridized in toto against membranes spotted with YAC clones covering a region of interest. In such a way, identified YAC clones provide positional information on homozygous deletions and loss of heterozygosity (LOH) regions. We have tested this approach with a tumor known to have a homozygous deletion within a region of LOH on chromosome 13. RDA was performed using representations generated with restriction enzymes BglII, NcoI and XbaI, and the difference products of each experiment were separately hybridized to chromosome 13 YAC filters. When collating the map positions of positive YACs from three different RDA experiments a cluster of hits clearly identified the region on chromosome 13 which comprised the homozygous deletion. This shows that our novel approach can be effective.
INTRODUCTION
The localization and isolation of genes that are mutated in cancer is important for the understanding of tumorigenesis. Tumor suppressor genes are one class of genes causally involved in this process (1). Biallelic inactivation of these genes is frequently accompanied by loss of genetic material (2). Therefore, identification of regions deleted in tumor genomes is an important step towards the isolation of tumor suppressor genes (3). Methods based on subtractive hybridization have been devised in order to enrich and isolate DNA fragments representing the genomic difference of similar genomes (4-7). These methods can help to identify homozygously deleted regions if tumor DNA is subtracted from constitutional DNA of the same individual. A more advanced scheme of subtraction, named representational difference analysis (RDA), was introduced by Lisitsyn et al. (8). Starting with relatively small amounts of DNA, this PCR based technique takes advantage of reiterated subtractive and kinetic enrichment to isolate restriction endonuclease fragments present in one population of DNA fragments (tester) but not in another (driver). In a first step, representations of the genomes that are to be compared are produced by adaptor-mediated PCR of restriction endonuclease fragments. Published protocols make use of the restriction enzymes BglII, BamHI and HindIII. Because of preferential amplification of smaller fragments, the resulting amplicons represent only a subset of the genome they originate from. The reduced complexity of the amplicons facilitates subtractive hybridization thus resulting in a more efficient enrichment of differences. The final difference products are cloned into plasmid vectors. Individual clones have to be analyzed in detail, because in addition to homozygously deleted fragments, RDA also enriches polymorphic fragments that are lost in driver representation due to chromosomal rearrangements that result in loss of heterozygosity (LOH). The clone by clone analysis is tedious, because typically lost polymorphic fragments prevail among cloned RDA products. RDA has successfully been used to isolate polymorphic loci (9), to identify sequences from the genomes of unknown pathogens (10), to isolate differentially expressed genes (11) and to detect homozygous deletions in tumor cells (12-14).
In diverse tumors, genome-scanning methods such as allelotyping or comparative genomic hybridization (CGH) have identified regions frequently showing loss of genetic material. At some of these sites tumor suppressor genes have been identified, e.g. the RB1 gene on chromosome 13 q (15), the p16 (INK4a) gene on chromosome 9 (16) and PTEN/MAC on chromosome 10 (17,18). In other regions, novel tumor suppressor loci are still to be found. For example, loss of one chromosome 3 is observed in ~50% of uveal melanomas (19). Intriguingly, monosomy 3 is a highly significant predictor for poor prognosis in uveal melanoma patients, supporting the idea that chromosome 3 contains genes which are involved in development and progression of metastatic disease (19). Supposing that loss of one chromosome 3 in these tumors is one step along a two-hit inactivation path, the identification of genes deleted or inactivated on the remaining homologue is of high interest. As the spectrum of inactivating mutations is diverse, homozygous deletions that affect a given tumor suppressor locus are to be expected in some tumors only. Therefore, several tumors have to be scanned in order to identify a presumed tumor suppressor locus via a homozygous deletion.
Here we describe a novel strategy in which difference products generated by RDA are labeled and hybridized in toto against an array of YAC clones covering a region of interest (Fig. 1). The YAC clones identified by the RDA products (hits) provide positional information on homozygous deletions and LOH regions. To distinguish these two classes of hits, different restriction enzymes are used to generate distinct representations of the same set of tumor and constitutional DNA. Ideally, RDA products originating from a homozygous deletion will highlight a region irrespective of the enzyme that was used to generate the representation, whereas polymorphic fragments enriched by RDA should result in an accumulation of hits for one representation only. If a region shows hits in different RDAs, comparative multiplex PCR with STSs mapped to this region can be used to confirm the presence of a homozygous deletion in the tumor.
Figure 1. Schematic protocol of the approach that combines RDA with hybridization of the RDA products to chromosome-specific YAC clone arrays to identify homozygous deletions in a region of interest. One set of chromosomes with a homozygously deleted region within a region of LOH is shown. (A) RDA is performed with tumor DNA as driver and the corresponding constitutional DNA as tester using representations generated with restriction enzymes BglII, NcoI and XbaI. RDA products enriched for DNA fragments deleted in tumor DNA are obtained after three rounds of hybridization and amplification. (B) Difference products of each RDA experiment are hybridized in toto against filters spotted with YAC clones covering the region of interest. Positive YAC clones become visible as black dots on the autoradiograph arranged in a definite spotting pattern. The coordinates of positive signals are determined and used to obtain the corresponding YAC clone names and map positions. (C) Map positions of positive YACs are entered into a chromosome map. The candidate region for a homozygous deletion is characterized by an accumulation of overlapping YACs highlighting a region irrespective of the enzyme that was used to generate the representation.
We have tested this strategy by analyzing a tumor known to carry a homozygous deletion affecting the RB1 gene and LOH at neighboring polymorphic loci. RDA was performed using blood DNA as tester and tumor DNA as driver, and representations were generated with restriction enzymes BglII, NcoI and XbaI. For the latter two enzymes, new linker oligonucleotides had to be devised. Difference products of each RDA experiment were hybridized to filters spotted with an array of YACs covering almost all of chromosome 13. Map positions of YACs identified by RDA products enriched from each representation were scattered over the whole chromosome with a slight accumulation in the RB1 and flanking LOH region. However, when collating position information from all three RDA experiments only one region showed a cluster of hits. This candidate region was located at 47/48 cM and comprised the homozygously deleted region on chromosome 13. Our new strategy facilitates scanning of multiple tumors for homozygously deleted regions in the presence of LOH regions.
MATERIALS AND METHODS
Representational difference analysis
RDA was performed as described by Lisitsyn et al. (20). For preparation of amplicons, 5 µg of both tumor (driver) and blood (tester) DNAs were cleaved with a restriction enzyme (BglII, XbaI or NcoI). For ligation of adaptors, 2 µg of digested DNA was mixed with 12mer and 24mer primer R, annealed to genomic DNA and T4 DNA ligase was added. PCR for generation of amplicons was performed in a Perkin Elmer thermal cycler 480 under the following cycle conditions: 95°C for 1 min, 72°C for 3 min for 20 cycles, the last cycle was followed by extension at 72°C for 7 min. Amplicons were cleaved with the chosen restriction enzyme. Tester amplicons were loaded on 1.6% NuSieve agarose gels, and DNA fragments from 200 to 1000 bp were cut out of the gel [for XbaI amplicon, the 340 bp band containing [alpha]-satellite DNA repeat sequences (21) was omitted]. DNA fragments were recovered by Qiagen tip 20 chromatography. An aliquot of tester amplicon (1 µg) was ligated to J (12mer and 24mer) oligonucleotides. In first round of hybridization, 400 ng of tester ligate J was mixed with 40 µg of driver amplicon and denatured at 97°C, followed by an incubation at 67°C for 20 h. Selective amplification of difference products was performed for 10 cycles in the standard PCR mixture, treated with mung bean nuclease and amplified under the same conditions as before for 20 cycles. The first round RDA product was digested with the original restriction enzyme, and 50 ng of the digest ligated to N oligonucleotides was subjected to the second round of selective hybridization/amplification. Annealing and elongation temperature was adjusted to 72°C. For third round of hybridization/amplification, 100 pg difference product ligated to adaptor J was used. Difference products were analyzed on 2% agarose gels.
RDA oligonucleotides
| R Bgl24 | 5[prime]-AGCACTCTCCAGCCTCTCACCGCA-3[prime]; |
| R Bgl12 | 5[prime]-GATCTGCGGTGA-3[prime]; |
| J Bgl24 | 5[prime]-ACCGACGTCGACTATCCATGAACA-3[prime]; |
| J Bgl12 | 5[prime]-GATCTGTTCATG-3[prime]; |
| N Bgl24 | 5[prime]-AGGCAACTGTGCTATCCGAGGGAA-3[prime]; |
| N Bgl12 | 5[prime]-GATCTTCCCTCG-3[prime]; |
| R Xba24 | 5[prime]-AGCACTCTCCAGCCTCTCACCGCT-3[prime]; |
| R Xba12 | 5[prime]-CTAGAGCGGTGA-3[prime]; |
| J Xba24 | 5[prime]-ACCGACGTCGACTATCCATGAACT-3[prime]; |
| J Xba12 | 5[prime]-CTAGAGTTCATG-3[prime]; |
| N Xba24 | 5[prime]-AGGCAACTGTGCTATCCGAGGGAT-3[prime]; |
| N Xba12 | 5[prime]-CTAGATCCCTCG-3[prime]; |
| R Nco24 | 5[prime]-AGCACTCTCCAGCCTCTCACCGCC-3[prime]; |
| R Nco12 | 5[prime]-CATGGGCGGTGA-3[prime]; |
| J Nco24 | 5[prime]-ACCGACGTCGACTATCCATGAACC-3[prime]; |
| J Nco12 | 5[prime]-CATGGGTTCATG-3[prime]; |
| N Nco24 | 5[prime]-AGGCAACTGTGCTATCCGAGGGAC-3[prime]; |
| N Nco12 | 5[prime]-CATGGTCCCTCG-3[prime]. |
Chromosome 13 YAC clone filters
YAC filters were provided by the Recource Center/Primary Database of the German Human Genome Project (RZPD; http://www.rzpd.de ). They were spotted with 1536 YAC clones from the CEPH Chromosome 13 Subset (library no. 920), including a few more clones of the pooling procedure with ambiguous links to chromosome 13. The spot positions are arranged in 384 squares each containing four clones spotted in duplicate. The duplicate clones are located at defined positions within the square and, therefore, a specific hybridization signal shows up as two autoradiography signals that conform to one of four defined patterns.
Filter hybridization
Adaptors of the difference products were removed by restriction enzyme digestion, and 25 ng of the RDA products were labeled by random priming procedure in the presence of 25 µCi of [32P]dCTP (22). After labeling and removal of unincorporated nucleotides by dialysis, the probes were suppressed with 50 µg Human COT-1 DNA (Gibco BRL) as described by the manufacturer. The YAC clone filters (Hybond N+; Amersham) were pre-hybridized at 67°C for 2 h in hybridization buffer composed of 7% SDS, 0.5 M sodium phosphate (pH 7.2), 1% BSA and 1 mM EDTA. COT-1 suppressed probe was mixed with hybridization fluid at a concentration of 3-5 × 106 c.p.m./ml (Cerenkov). Hybridization was carried out for 16 h at 67°C. Washing conditions were as follows: twice in 40 mM sodium phosphate (pH 7.2), 0.1% SDS at room temperature, and twice in 40 mM sodium phosphate (pH 7.2), 0.1% SDS at 65°C for 15 min each wash. Exposure was overnight at -70°C.
Filter analysis
Hybridization signals that conformed to the expected spotting pattern and that were clearly distinct from background signals of the same array and neighboring arrays were scored as positives. As the YAC clones were not spotted according to their chromosomal position, the coordinates of positive signals had to be determined and a database at the RZPD (http://www.rzpd.de ) was required to obtain the RZPD clone names and the CEPH YAC clone names. YAC map and contig information was obtained from the database at the Whitehead Institute/MIT Genome Center (http://www-genome.wi.mit.edu ).
Tumor DNA sample
Tumor and blood DNA from a retinoblastoma patient was prepared and analyzed as described (22).
Comparative multiplex STS analysis
Comparative multiplex PCR of STSs was performed in 20 µl reaction mix containing 80 ng of tumor DNA and 10-20 pmol of each primer under the following conditions: an initial denaturation step at 95°C for 5 min was followed by 35 cycles of denaturation at 95°C, annealing at 55°C and elongation at 72°C. Primer set 1 consists of markers WI-1939, WI-3641, WI-3715, WI-5710, WI-6795, NIB1753 and AFMA310WB5, primer set 2 consists of AFMA310WB5, WI-1939, WI-6319, WI-5710, WI-2379 and WI-3715. Primer sequences for amplification of the STSs were obtained from GDB.
RESULTS
We have investigated if combining RDA with direct hybridization of the resulting products against YAC clone arrays is a feasible approach to localize homozygously deleted regions in tumors. We have tested this strategy by analyzing a retinoblastoma sample known to carry a homozygous deletion of the RB1 gene flanked by LOH.
RDA
Initially, we followed the original RDA protocol using BglII, BamHI and HindIII as reported by Lisitsyn et al. (8,20). To establish the RDA protocol, adenovirus type II or bacterio-phage [lambda] DNA were added to human placental DNA as target in equimolar amounts to generate tester. While RDA performed with BglII representations enriched the expected target fragment, representations generated with BamHI and HindIII repeatedly showed poor efficiency in our hands. Therefore, we have adapted RDA for the use of representations generated with XbaI and NcoI. When performing RDA on NcoI representations with adenovirus type II DNA as target, we observed that of the two expected difference products of 486 and 420 bp only the 486 bp product was amplified after three rounds of selective hybridization and amplification (data not shown). This may indicate that not all target sequences in a chosen representation are enriched efficiently.
DNA samples isolated from a retinoblastoma with the known RB1 gene deletion and corresponding peripheral blood were used as driver and tester to enrich DNA fragments not represented in the tumor. Representations of both genomes were prepared with the restriction enzymes BglII, XbaI and NcoI, and three rounds of subtractive hybridization and amplification were performed (Fig. 2).
Figure 2. Representational difference analysis with retinoblastoma DNA as driver and constitutional DNA as tester. Agarose gel electrophoresis of difference products of BglII, XbaI and NcoI representations, obtained after three rounds of hybridization and amplification. Adaptors were removed from difference products by cleavage prior to gel analysis. RDA products were separated on 2% agarose gel in 1× TAE and visualized by ethidium bromide staining. Marker (M), 1 kb DNA ladder (Gibco BRL).
Filter hybridization
Each of the filters used for hybridization of RDA products was spotted with 1536 YAC clones in duplicate. Identical clones are arranged in a definite spotting pattern and, therefore, a specific hybridization signal is characterized by two autoradio-graphy dots. To check the quality of these filters, several STSs mapped to YACs of the CEPH chromosome 13 library were used as probes. Between four and seven YACs were hit by a single STS probe. This is in concordance with the high degree of overlap of YAC clones within chromosome 13 contigs spotted on the filters. However, only 40-60% of all spotted YACs linked to the contig by the STS used as probe were detected in this experiment. This indicates that some YACs, although spotted, are not detected by the STS probe. This might be due to partial deletion of YACs obtained during yeast propagation, weak hybridization signals indistinguishable from background noise and incomplete spotting or poor growth of yeast cells. The latter point is also likely to be responsible for the occasional observation that the two autoradio-graphy dots indicating a specific clone show unequal intensity. In this case, hybridization of the same difference products to another filter and collating the results of both filters leads to additional double dots and thus to additional positive signals.
Difference products obtained after three rounds of selective hybridization/amplification on the retinoblastoma (Fig. 2) were cleaved to remove the adaptors and labeled in the presence of [[alpha]-32P]dCTP. As labeled RDA probes have to be considered as a probe pool, they were used at an activity of 3-5 × 106 c.p.m./ml (Cerenkov) hybridization solution. Human COT-1 DNA was used to suppress repetitive DNA sequences reported to be present in RDA products (20). The RDA products obtained with each of the three representations (BglII, XbaI and NcoI) were hybridized to separate YAC filters. Analysis of the autoradiographs (Fig. 3) resulted in the identification of a total of 92 YACs. Seventy-four of these YACs were part of chromosome 13 YAC contigs (Whitehead Institute for Biomedical Research/MIT Center for Genome Research) (Fig. 4). Eighteen YACs are linked to chromosome 13 by ambiguous hits only and were thus omitted from further analysis (not listed in Fig. 4). Thirty-five (47%) of chromosome 13 YACs do not show any overlap with other YAC hits. Because of the overlap of several YAC clones within chromosome 13 contigs (Fig. 4B), RDA products originating from chromosome 13 are expected to highlight multiple YAC clones. Non-overlapping hits, therefore, might be caused by hybridization of non-chromosome-13 RDA products to chimeric or contaminated YACs. In addition, non-specific hybridization signals might contribute to these hits. Hits of overlapping YACs (Fig. 4A), are found in nine regions (including the 47/48cM region) along the map of chromosome 13. The RDA fragments causing these signals are likely to originate from chromosome 13.
Figure 3. Autoradiograph of third round XbaI difference product hybridized to chromosome 13 YAC clone filters.
Figure 4. Positional information of positive YACs obtained by hybridization of retinoblastoma RDA products to chromosome 13 filters. Representations of both tumor and constitutional DNA were prepared with the restriction enzymes BglII, XbaI and NcoI. (A) Schematic diagram of a pair of chromosomes 13 as present in the tumor investigated here. Polymorphic loci that had been analyzed to determine LOH are indicated by arrows. A region flanking the RB1 locus including markers D13S291 and D13S166 is affected by LOH. However, the LOH-region does not extend beyond D13S219 and D13S762 because at these loci heterozygosity is retained. The LOH status of the region between D13S219 and D13S291 as well as D13S166 and D13S762 was not analyzed (hatched boxes). The homozygous deletion including the RB1 gene is indicated as a gap at 48 cM. Symbols (filled circles) indicate the map positions of positive YACs obtained by third round RDA products hybridized to chromosome 13 YAC filters. Positive YACs (hits) were entered into the chromosome 13 map according to their position in cM. Symbols grouped in a vertical line indicate overlapping or closely neighbored YACs. (B) Part of the YAC contig WC13.2 covering the 47/48 cM region on chromosome 13 (according to the information obtained by the Whitehead Institute for Biomedical Research/MIT Center for Genome Research). YACs identified by hybridization of RDA products against chromosome 13 filters are in bold. Characters in front of these YACs indicate the enzymes used to generate the representations (B, BglII; N, NcoI; X, XbaI). Open triangles indicate markers (D13S153, WI-6319 and D13S1307) which are homozygously deleted in the tumor DNA. The deletion of RB1 was confirmed by Southern blot analysis (data not shown). Filled triangles indicate loci which are not homozygously deleted. This was either shown by LOH analysis (D13S273, D13S262 and D13S270) or comparative multiplex PCR (WI-3641, NIB1753, WI-6795, WI-2379, WI-5710 and AFMA301WB5). The shaded bars indicate the borders of the homozygous deletion.
A few of these overlapping YAC hits are located in regions that, according to the results of genotyping of polymorphic loci, are disomic in the tumor. Possibly, these difference products are generated by incomplete subtractive hybridization in the RDA. This means that there is a background of RDA products that does not result from a difference between tester and driver genomes (8). Repetitive sequences, which are frequently present in RDA products (20) are one example for this category. As some of these RDA products can also be derived from chromosome 13 disomic regions, they will recognize overlapping YACs when hybridized to chromosome 13 filters (e.g. 105, 96, 82 and 78 cM).
RDA products derived from lost polymorphic restriction fragments in LOH regions will also generate hits of overlapping YACs. The hit cluster in the 58 cM region generated by BglII and NcoI representations might fall into this category. As polymorphic loci flanking this region showed LOH in the tumor (23) the RDA products that identified this hit cluster are likely to be generated by lost polymorphic BglII and NcoI fragments. The LOH status of the 33 cM region is not precisely defined; therefore, this hit cluster cannot be assigned to disomic or LOH regions.
In one region only, all three difference products recognized overlapping YACs thus conforming to the hypothesized signal pattern caused by a homozygous deletion. This cluster, which is localized at 47/48 cM within YAC contig WC13.2, comprised 17 (23%) of all 74 chromosome 13 linked YAC hits (Fig. 4A). The difference products identified 13 different YACs located in this region, with four YACs identified by difference products derived from two different representations.
The results obtained by RDA-YAC filter hybridization indicated the presence of a homozygous deletion spanning several Mb (Fig. 4B). To determine which of the YACs lie within the region homozygously deleted in the retinoblastma, comparative STS multiplex PCR (WI-3641, NIB1753, WI-6795, WI-6319, D13S1307, WI-2379, WI-5710, AFMA301WB5 and D13S270; Fig. 4B) was used to map the extent of the homozygous deletion. In a previous analysis, the polymorphic loci D13S273, D13S262 but not D13S153 had shown LOH (23).
This way, the deletion was narrowed down to ~1 Mb and was identified by six hits (887H7, 830C8, 762D6, 969H9 and 2× 928E4) (Fig. 4B). Twelve hits in the candidate region were located in the RB1 flanking LOH regions. Due to its long extension into the LOH region, YAC 969H9 is assigned to both the homozygously and hemizygously deleted regions.
In summary, the homozygous deletion including the flanking region was highlighted by 23% of all hits. As the extent of the LOH region is difficult to define, the number of YAC hits located in this region lies between 29 (for 15 cM of LOH) and 43 (for 44 cM of LOH). Thirty-five percent of all hits are scattered over the remaining 76 cM disomic region of chromosome 13. When neglecting non-overlapping YACs, the homozygous deletion would become even more prominent.
DISCUSSION
The identification of novel tumor suppressor loci is often led by the detection of regions recurrently deleted in tumors. Various methods are currently available for the identification of genomic deletions. CGH is frequently used for a genome wide analysis of chromosomal aberrations (24). However, deletions have to exceed 10 Mb in size in order to be detectable. The application of the original RDA protocol is limited when large regions of LOH are present in the genome, because polymorphic fragments result in a prominent background of clones. Probably, that is why there are only few reports on the detection of homozygous deletions in tumors by RDA (13,15). Screening for homozygous deletions in several tumors by STS-PCR analysis (25) is an expensive and time consuming procedure if a large region of interest has to be investigated.
To facilitate scanning of multiple tumors for homozygous deletions, we have devised a new procedure which combines the RDA with hybridization of the difference products to filters spotted with a clone contig that covers a region of interest. We postulated that if subtracting tumor DNA from corresponding constitutional DNA by RDA, deletions will result in RDA products that hybridize to those clones that cover the deleted region. Furthermore, a homozygous deletion but not simple loss of a polymorphic fragment will result in RDA products that consistently highlight a region irrespective of the restriction enzyme that was used to create the representations.
To test this approach, we analyzed a tumor known to carry a homozygous deletion of the RB1 gene. The whole of chromosome 13 was specified as the region of interest. Representations of driver (tumor DNA) and tester (corresponding constitutional DNA) genomes were generated with restriction enzymes BglII, XbaI or NcoI, and each pair of driver and tester amplicons was subjected to RDA. The resulting products were labeled and hybridized to arrays of chromosome 13 YACs. A total of 74 chromosome 13 YAC clones showed hybridization signals. When analyzing these hits within the framework of the chromosome 13 YAC map, positive YACs fell into one of the following three categories: (i) 35 YAC hits (47%), not overlapping with any other positive YAC. (ii) In eight regions (not including the candidate region at 47/48 cM), RDA products enriched from one or two representations hybridized to overlapping YACs. These regions contained 22 (30%) of all YAC hits. (iii) In only one region, 17 (23%) overlapping YACs were highlighted by difference products from all three representations. Only these YAC hits conform to the signal pattern predicted to be caused by a homozygous deletion and thus define a candidate region.
We further analyzed this region in detail by comparative multiplex STS-PCR. We found that homozygously deleted STSs (D13S153, WI-6319, D13S164) are mapped to only six of the 17 YAC hits identified by all three representations. The remaining 12 YAC hits (YAC 969H9 is assigned to both the homozygously and hemizygously deleted regions), however, are mapped to regions that showed LOH in the tumor. The LOH region, which extends at least 15 cM, was identified by a total of 30 YAC hits. It is surprising that 12 of the 30 YAC hits within the LOH region are located in the immediate vicinity of the homozygous deletion rather than showing a random distribution. Some of the YAC hits mapped to the LOH region may extend into the deleted region (e.g. 900G6 and 889H6) and thus may be recognized by RDA products derived from the homozygous deletion. In addition, while the presence of a homozyous deletion including the RB1 locus in this tumor was confirmed by several independent analyses, we cannot exclude rearrangements or additional homozygous deletions in these breakpoint regions that result in RDA products.
In our test system, the only region that fulfilled the proposed criteria for a homozygous deletion, i.e. accumulated YACs identified by RDA products from all three representations, in fact spanned the site of homozygous deletion in the tumor investigated here. However, for smaller deletions the likelihood of being represented in three different representations may be reduced. Therefore, regions identified by overlapping hits of two different representations as seen for the 58 cM region may also be regarded as candidate regions and thus further tested for deletions. In addition, in establishing NcoI based RDA we have observed that DNA endonuclease fragments might get lost during three rounds of RDA. This effect might also cause absence of an expected hybridization signal in some representations.
The size of a homozygous deletion that is detected by RDA-array hybridization depends on several factors. In our test system we were able to detect a deletion of ~1 Mb on the background of at least 17 Mb of LOH (Fig. 4A). However, it is obvious that the smaller a deletion the smaller the chance to enrich RDA fragments representing this deletion. On the other hand, larger LOH regions are expected to generate more RDA products, making signals derived from homozygous deletions less prominent. Consequently, the number of candidate regions and in parallel the work which has to be done for their verification might increase with larger LOH regions. So far, we do not have data that enable us to predict the minimum deletion size and the ratio of LOH to homozygous deletions that is required for successful application of the RDA with filter hybridization.
We are well aware that our approach is sensitive to some pitfalls such as chimeric YACs, incomplete subtraction efficiency in the RDA, non-specific hybridization signals and some variability in filter spotting and autoradiograph interpretation. Most probably these problems will lead to additional background signals which are expected to be equally distributed over the whole region of interest and which will therefore slightly increase the number of candidate regions. Other problems like incomplete coverage of chromosomes with YACs, inaccurate mapping of YACs and markers can lead to the loss of positive signals and thus reduce the sensitivity of the deletion detection. RDA products derived from contaminating DNA fragments present in tester (e.g. bacterial or viral DNA in cultured cell lines), are not expected to identify any YAC covering the region of interest.
For smaller regions of interest, the resolution of the method may be increased significantly. Small regions of interest facilitate the use of filters spotted with BAC or cosmid clones instead of the YAC clones. Fine mapping of deletions should be possible using these filters. The approach can be further improved when microarrays are used instead of filters which would in addition facilitate automated analysis of the hybridization signals.
ACKNOWLEDGEMENTS
We thank W. Berger for technical advice and the RZPD for providing the chromosome 13 YAC clone filters. This research was supported by the Deutsche Forschungsgemeinschaft (DFGLO530/3-1).
REFERENCES
*To whom correspondence should be addressed. Tel: +49 201 723 4556; Fax: +49 201 723 5900; Email: michael.zeschnigk{at}uni-essen.de
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||