| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Biology and Genetics |
Departments of Biochemistry and Molecular Biology [I. D. C. M., J. C., P. W. L., P. A. J.], Urology [G. L., P. A. J.], Pathology [D. S.], and Surgery [P. W. L.], Urologic Cancer Research Laboratory, Norris Comprehensive Cancer Center, Keck School of Medicine, University of Southern California, Los Angeles, California 90089
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
7000 CpG islands altered in the genome. Surprisingly, few additional changes in methylation patterns were observed in cell lines derived from the tumors; however, all of the cell lines showed altered methylation in a common set of 3% of evaluable sequences. Three genes known to be aberrantly methylated in bladder cancer (p16, p15, and PAX6) were studied in detail by methylation-sensitive single nucleotide primer extension and showed increased methylation in culture at preexisting methylated sites for all of the exons but no de novo methylation in culture for the promoters in any cell line. Therefore, our investigation provides the first serial as well as parallel quantitation of the global epigenetic stability in two independent bladder cancer genomes over the course of progression and in culture. In addition, our investigation also provides the first direct comparison of the epigenetic and genetic patterns on the global scale, showing the epigenetic pattern to be relatively stable in vivo and in vitro over time within an individual. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Whereas there is a significant amount of information on the genetic status of several cancers including that of the bladder (6, 7, 8, 9)
, much less is known about the epigenetic status or, more specifically, the DNA methylation pattern. Modification of CpG dinucleotides in gene promoters by methylation has been shown to inhibit transcription initiation (10, 11, 12)
and lead to chromatin condensation (13)
. About 1% of the genome has the expected frequency of the CpG dinucleotides (14)
, and these "islands" are defined by having a GC content >0.5, an O/E for the frequency of occurrence of CpG sites >0.6, and a minimum length of 200 bp (15)
. There are 45,000 CpG islands in the human genome (14)
, which are typically unmethylated in somatic cells (10)
and frequently but not always associated with the promoters of genes (10
, 15, 16, 17)
.
Costello et al. (3) recently showed that alterations in methylation at CpG islands, as detected by restriction landmark genomic scanning (RLGS) (18) in 98 primary human tumors, were nonrandom and tumor-type specific. Muto et al. (19) demonstrated that 77% of normal urothelium samples from bladder cancer patients had either loss of heterozygosity at 1 of 10 microsatellite sites or methylation of the p16 5' region. This indicated that alterations had occurred in these patients early in tumorigenesis and before pathological detection. Even so, all of the data published to date represent static comparisons, and there have been no longitudinal studies on the global level that have looked at epigenetic variability inter- and intraindividually.
In this study, we investigated the epigenetic stability of genomes from two patients with transitional cell carcinoma of the bladder for which genetic information was also available from a previous study (20)
. Two metachronous tumors were obtained from each patient, in one case separated by 6 months and in the other case by 11 months, and cell lines were derived from each tumor for which analysis was performed at low and high passage (see Fig. 6
for the temporal and spatial relationships of the samples). These were compared with a panel of normal urothelium samples derived from patients free of bladder disease using global MS.AP-PCR3
and gene-specific MS.SNuPE approaches. We were able to distinguish between alterations in methylation that were shared by all of the tumors and were therefore possibly common in bladder cancer (cancer-specific), those that were shared by only one tumor of each patient (tumor-specific), those that were unique to the tumor of a given patient (patient-specific), and those that occurred in culture. The data allowed us to determine DNA methylation patterns in vivo, for alterations in the two tumors, and in vitro, for progression from low- to high-passage cell lines.
|
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Genomic DNA was isolated using a simplified procedure developed previously (21 , 22) . After spectophometric determination of the concentration, 2 µg of DNA was digested separately in a total volume of 40 µl with either 20 units of RsaI, 20 units of RsaI and HpaII, or 10 units each of RsaI and MspI (Roche Molecular Biochemicals, Indianapolis, IN) at 37°C for 16 h. Another 2 units of each enzyme were added for an additional 1 h at 37°C to ensure complete digestion with the methylation-sensitive restriction endonucleases.
The arbitrarily primed PCR was conducted following the basic protocol described by Gonzalgo et al. (23)
. For each sample, 2 µl of restriction enzyme-digested DNA equivalent to 100 ng was amplified using a combination of two or three randomly designed GC-rich primers (see Table 1
) based on the principles of Welsh and McClelland (24)
. Conditions for amplification were: a primary denaturation step at 95°C for 5 min; then 95°C for 2 min, 40°C for 1 min, and 72°C for 2 min x 35 cycles; and a final extension at 72°C for 8 min. The PCR mixture contained 1 x buffer [10 mM Tris-HCl (pH 8.3), 1.5 mM MgCl2, 50 mM KCl, and 0.1% gelatin/ml], 200 µM each of the four deoxynucleotide triphosphates (Roche Molecular Biochemicals), 0.5 µM of each primer, 5% DMSO (Sigma-Aldrich), 1 unit of Taq polymerase (Roche Molecular Biochemicals), and 2 µCi of 
33P-labeled dATP (NEN Life Sciences Products Inc., Boston, MA) in a final volume of 25 µl.
|
55°C. The gel was then dried at 80°C for 1 h and exposed to autoradiographic film overnight with an intensifying screen at -70°C. Bands were scored visually by comparing the pattern among the three sets of digestions. Candidate bands were excised from the gel using the film as a marker and placed in a microfuge tube containing 50 µl of sterile H2O. The sample was then heated at 80°C for 10 min and vortexed to ensure dissolution of the DNA from the gel/paper slice. Two µl of eluate was then used in a PCR with the same primers used in the original arbitrarily primed PCR, to amplify the given band and generate a sufficient template for plasmid cloning and sequencing. Conditions for amplification were: a primary denaturation step at 95°C for 2 min; then 95°C for 1 min, 49°C for 45 s, and 72°C for 1 min x 35 cycles; and a final extension at 72°C for 10 min. The PCR mixture was the same as before except that no radiolabeled nucleotide was used, and the enzyme was a 1:1 combination of Taq polymerase (1 unit) and Taq Antibody (0.275 µg; Clontech Laboratories Inc., Palo Alto, CA).
The PCR products were resolved on a 2% (w/v) agarose gel (Life Technologies, Inc.) with 0.5 µg/ml ethidium bromide (Sigma-Aldrich) in 1 x TAE buffer [0.04 M Tris-base, 0.004 M acetic acid, and 0.001 M EDTA (pH 8.0)] at 100 V to ensure appropriate size and then isolated using the Gel Extraction kit (Qiagen Inc., Valencia, CA) with a final volume of 30 µl in sterile H2O. This product was immediately cloned into a plasmid (pCR) vector using the TA cloning kit (Invitrogen Corporation, Carlsbad, CA). Five colonies for each sample were selected and dissolved in 500 µl of sterile H2O assisted with vortexing. Ten µl of each suspension were then amplified by another round of PCR under these conditions: a primary denaturation step at 95°C for 2 min; then 95°C for 1 min, 58°C for 45 s, and 72°C for 1 min x 35 cycles; and a final extension at 72°C for 4 min. The PCR mixture contained 1 x buffer [10 mM Tris-HCl (pH 8.3), 1.5 mM MgCl2, 50 mM KCl, and 0.1% gelatin/ml], 200 µM each of the four deoxynucleotide triphosphates, 0.5 µM of each primer, 5% DMSO, and a 1:1 combination of Taq polymerase (1 unit) and Taq Antibody (0.275 µg) in a final volume of 25 µl.
The amplified product was divided into two aliquots, only one of which was further digested with 5 units of HpaII at 37°C for 1 h, and then both were run on a 2% agarose gel with 0.5 µg/ml ethidium bromide in 1 x TAE buffer at 100 V. This ensured that the cloned sequences contained the methylation-sensitive site, which was an internal experimental control. The sequence was isolated using the Gel Extraction kit and resuspended in 30 µl of sterile H2O.
Each sequence was then sent to the University of Southern California/Norris Comprehensive Cancer Center Microchemical Core Facility for automated DNA sequencing by a dye-based Applied Biosystems procedure where 3'-fluorescence-labeled dideoxynucleotide triphosphates are incorporated into DNA extension products (asymmetric PCR) and then excited during electrophoresis by laser light. Each sample contained 0.15 µg of PCR product and 4.8 pmol of one vector-flanking primer (5' or 3') in a final volume of 18 µl.
The profile of the sequence was then determined by two calculated values, GC content and O/E (15)
, defined in Fig. 3
. After this, the sequence was inputted to the standard nucleotide-nucleotide BLAST databases [Nonredundant Sequence = GenBank + EMBL + DDBJ + PDB + Phase-3 finished human genomic sequences; Expressed Sequence Tag = GenBank + EMBL + DDBJ from EST Divisions; and High Throughput Genome Sequence = Phase 0/1/2 unfinished human genomic sequences databases]4
(25)
, the Human Genome Database (working draft human genomic sequences),5
and the RepeatMasker Web Server Sequencing6
to determine whether the sequences were part of known regions. In some cases, the sequences were identified as having putative promoter elements7
and/or CpG island DNA genomic MseI sequences (as originally submitted by A.P. Bird et al. c/o United Kingdom MRC Human Genome Mapping Project Resource Center).
|
. But as stated already, 16/60 (27%) of the sequenced set showed an altered methylation pattern. This discrepancy in the percentages of altered sequences (17% versus 27%) allowed us to realize that there was indeed a bias in the subset of bands selected for sequencing, although it was toward those with an altered methylation pattern instead of those that were CpG islands.
|
94 CpG islands from the total 214 sequences analyzed and 14 of these that had an altered methylation pattern. Applying this proportion, 14/94, to the 45,000 CpG islands in the genome yields 6,750 or 7,000 (because there is only one significant digit in the approximation) CpG islands that are altered. Only a fraction of these sequences (7/14 = 50% of 7,000 = 3,500), therefore, are CpG islands altered in a cancer-specific manner in the genome.
Gene-specific Methylation Analysis: MS.SNuPE
Three genes were analyzed by MS.SNuPE at their promoter regions and internal exons: p16, p15, and PAX6 as described previously (26)
. Genomic DNA (2 µg) was first treated with sodium bisulfite based on a protocol outlined by Frommer et al. (27)
and resuspended in a total of 40 µl of sterile H2O.
p15 Promoter.
Converted DNA (4 µl) was used to amplify the promoter region in a PCR as described previously (28)
.
p15 Exon 2.
Converted DNA (4 µl) was used to amplify the exon 2 region in a PCR with these conditions: hotstart with a primary denaturation step at 95°C for 2 min; then 95°C for 1 min, 58°C for 45 s, and 72°C for 1 min 15 s x 40 cycles; and a final extension at 72°C for 4 min. The PCR mixture contained 1 x buffer [20 mM Tris-HCl (pH 7.5), 100 mM KCl, 1 mM DTT, 0.1 M EDTA, 0.5% Tween 20 (v/v), 0.5% NP40 (v/v), and 50% gelatin/ml], 160 µM each of the four deoxynucleotide triphosphates, 0.5 µM of each primer, 2% DMSO, and 1.25 units Taq polymerase with 0.275 µg Taq Antibody in a total volume of 25 µl. The primers used were sense 5'-GGT-GAG-GGG-GTT-TTA-TAT-AAG-3' and antisense 5'-GAA-GGT-TTT-TTA-GGT-GAG-G-3'.
For the SNuPE reaction, 4 µl of PCR product was subject to a single nucleotide primer extension through one cycle of 95°C for 2 min, 50°C for 2 min, and 72°C for 1 min. The SNuPE mixture contained 1 x buffer [10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, and 0.001% gelatin], 1 µM primers, 0.4 units Taq polymerase and 0.088 µg Taq antibody (1:1), and 1 µCi of either 32P-dCTP or 32P-TTP in a total volume of 10 µl. The primers used were (22mer) 5'-TAT-TAG-AGG-TAG-TAA-TTA-TGT-T-3', (17mer) 5'-GTT-ATG-ATG-ATG-GGT-AG-3', and (14mer) 5'-TGG-TGG-TGT-TGT-AT-3'.
p16 Promoter.
Converted DNA (4 µl) was used to amplify the promoter region in a PCR as described previously (29
, 30)
.
p16 Exon 2.
Converted DNA (4 µl) was used to amplify the exon 2 region in a PCR as described previously (31)
.
PAX6 Promoter.
Converted DNA (2 µl) was used to amplify the promoter region in a PCR as described previously (32)
.
PAX6 Exon 5.
Converted DNA (2 µl) was used to amplify the exon 5 region in a PCR as described previously (33)
.
For all of the SNuPE samples, 6 µl of formamide dye-loading buffer was added to each, and 2.5 µl was loaded onto a high-resolution 15% polyacrylamide gel under denaturing conditions (7 M urea), which was run at 121 W for 1 h to maintain a running temperature of 55°C. The gel was then dried at 80°C for 1 h and exposed to a phosphorimager screen overnight. The exposure was scanned (PhosphorImager 445-SI Version 4.0; Molecular Dynamics, Sunnyvale, CA) and the degree of methylation quantitated using the ImageQuant version 5.0 software (Molecular Dynamics), correcting each data point for background lane noise. The following calculation yielded percentage of methylation: % C = volume C/(volume C + volume T) x 100.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). The method also allowed us to determine how frequently methylation changes occur in a cancer-specific way (occurring in all four tumors), a tumor-specific way (occurring in only the first or second tumor of each patient), or a patient-specific way (occurring in one or both tumors of one patient but not in those of the other).
An example of an MS.AP-PCR gel is given in Fig. 1
. The method relies on the predigestion of DNA with an enzyme to reduce the size of genomic sequences (in this case RsaI) before double digestion with either an enzyme sensitive to the methylation of the internal cytosine in the context of the sequence CCGG (HpaII) or digestion with an enzyme insensitive to the methylation of the internal cytosine residue in the same sequence (MspI). The three different digestions are then subjected to an arbitrarily primed PCR reaction with two or three GC-rich primers at a time for a total of 16 unique primer sets. The amplified regions are resolved on a 15% denaturing polyacrylamide gel and scored in one of four ways: (a) unevaluable, where a band is present in the RsaI, the RsaI + HpaII, and the RsaI + MspI digestions indicating the lack of an internal CCGG site; (b) methylated, where a band is present in the RsaI and RsaI + HpaII but absent or with much reduced intensity in the RsaI + MspI digestions; (c) unmethylated, where a band is present in the RsaI digestion only; and (d) altered, where a band is present in the RsaI, variably present in the RsaI + HpaII, and not present in the MspI digestion. Previous work has shown that if the bands are excised from the gel, cloned, sequenced, and subjected to quantitative methylation assays (MS.SNuPE), the resulting level of methylation at the individual CCGG sites corresponds to the visual pattern seen by MS.AP-PCR (33)
.
|
A summary of the methylation patterns is provided in Fig. 2
. Results obtained from patient A showed that 89 sequences that were unmethylated in normal urothelium were unchanged in the tumors, whereas 106 of the total 214 evaluable sequences remained methylated in all of the samples (Fig. 2A)
. Thus, in patient A, 91% of the sequences did not change their methylation status as assessed by MS.AP-PCR. Hypermethylation events relative to normal urothelium occurred in 8 sequences and were observed in both tumors suggesting that the changes were either cancer- or patient-specific, whereas a total of 11 sequences were found to be either hyper- or hypomethylated in a tumor-specific way. Results obtained for patient B using the identical primer sets showed 83 sequences unmethylated in all of the samples and 101 sequences methylated in all of the samples, representing 86% of the total sequences scored. Of the 30 altered sequences, 19 sequences were methylated and 5 were unmethylated in both tumors, again representing potential cancer- or patient-specific changes. Of the remaining 6 sequences that were altered in a tumor-specific way, only 1 sequence represented a hypermethylation event.
Fig. 2B
shows the methylation changes common to both patients, an analysis which was possible because of the highly reproducible banding pattern by each primer set in the MS.AP-PCR assay. The methylation patterns were compared on a band-by-band basis so that the 214 sequences analyzed were identical for both patients, allowing for a more accurate estimation than simply totaling or averaging the individual results. In this case, the number of unchanged regions in all of the tumors was 83% (77 of 214 sequences). There was a considerable overlap of 3% (7 of 214 sequences) of evaluable sequences that underwent methylation changes in all four tumors, which represented cancer-specific changes. This suggested that alterations in these sequences may play a key role in the genesis of bladder cancer. Also of importance is that the changes that were specific to an individual patient could be identified in this analysis and represented 13% (29 of 214) of evaluable sequences.
We next examined the sequence characteristics of 60 of the 214 evaluated sequences and divided them into three categories depending on whether they were located within (or contained) CpG islands, non-CpG islands, or repetitive elements [which includes short interspersed nuclear elements (SINEs) (Alu and mammalian-wide interspersed repeats (MIR), long interspersed nuclear elements (LINEs) (LINE 1 and LINE 2), and long terminal repeats (LTR) but not simple repeats or microsatellite repeats; Fig. 3
]. The size range of the sequences was 136–618 bp with only 2 sequences falling below 200 bp (at 179 and 136 bp) that were included in the analysis. Sequences are available on request.
Sequences with an altered methylation pattern were almost evenly distributed between CpG islands and non-CpG islands with fewer changes occurring in sequences located within repetitive elements (6, 7, and 3 sequences, respectively). The sequences that were methylated in all of the samples were relatively evenly distributed between the three sequence classes (7, 8, and 6 sequences), whereas a higher percentage of sequences that were unmethylated in all of the samples were located in non-CpG islands with considerably fewer falling into the repetitive element category (13, 8, and 2 sequences). A preponderance of unmethylated samples was therefore located in CpG islands, which was expected because these regions are known to be free of methylation in somatic tissues (16) . They also showed that alterations in methylation occurred in all sequence classes of DNA and that, on average, the sequences identified were relatively GC-rich, with a GC content of 0.51 and an O/E of 0.64.
Table 2
shows the details of 32 sequences that matched to a chromosome in the genome database. The sequences are listed in three groups based on whether the methylation pattern was methylated in all of the samples, unmethylated in all of the samples, or altered in the tumors and/or cell lines. The details included are GC content, O/E, chromosome match, ID or Accession No., clone name, sequence status, and region of base matches (with the percentage match). These sequences were part of the set of 60 analyzed in Fig. 3
.
|
7000 CpG islands in the genome are altered in bladder cancer with 3500 altered in a cancer-specific manner (See "Materials and Methods" for an explanation of the derivation of this estimate).
Few Additional Methylation Changes Occur in Cell Culture.
It is well known that normal tissues undergo considerable changes in methylation patterns in processes associated with adaptation to culture (13
, 35)
. The four tumors studied in this report easily adapted to culture so that we were able to determine how methylation patterns of bladder cancers as opposed to normal tissues evolved in vitro. Low and high passages were analyzed for each cell line and compared with the respective tumor by MS.AP-PCR. The results of these studies (data not shown) indicated that an additional 10% (21 of 214 sequences) of methylation changes occurred with the culturing of tumors 1 and 2 from patient A, whereas 6% (13 of 214 sequences) of methylation changes occurred in the case of patient B. The details of this analysis (the distribution of hyper- versus hypomethylated sequences) are shown in Fig. 6
, which also contains data published previously on the genetic changes present in the cell lines (20)
.
When the results for the cell lines were compared with the normal urothelium and tumors concurrently, there were only 3% (7 of 214 sequences) of cell line-specific methylation changes in common between the patients (Fig. 4)
. The majority of changes (28/35) involved alterations in the methylation pattern of sequences in one patient but not the other and were therefore labeled "discrepant." This result was surprising, because it had been anticipated that the methylation changes in culture would be much more extensive. However, the data clearly show that there is no convergence toward a relatively uniform methylation pattern in vitro because of a subset of sequences that is susceptible to proliferation-based hypermethylation.
|
shows that in patient B the levels of methylation in the exonic regions of all three genes increased as a function of time in culture for all of the specimens examined, although there were no discernable alterations in the level of methylation of the respective promoters, which remained unmethylated. We were unable to measure the methylation of the p15 and p16 genes in patient A, because this region had been homozygously deleted within the tumor (8)
. However, for patient A, the PAX6 gene showed no methylation of the promoter region, although methylation of the exonic region approached 100% in those lines that had been cultured for an extended period of time. Thus, we observed no de novo methylation of regions that were unmethylated in the tumor but rather an apparent enhancement of existing methylation patterns within these three loci.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
It has been shown previously that methylation patterns are tissue-specific and show little interindividual variation (34) . Because our study focused on a sample set derived from the bladder, we obtained a panel of five normal urothelium controls from individuals known to be pathologically free of bladder disease. Of the 214 examined regions, we found <1% variation between five individuals (data not shown). This was critical, because it substantiated earlier work that showed that variations in the methylation pattern between individuals are minimal (34) and allowed a more accurate quantitation of the differences in methylation pattern measured in this study by giving us a margin of error of <1% when comparing data between patients.
Our data indicate that there were 17% (37/214) differences in methylation pattern of the bladder cancer genomes of the two patients, of which 14 were located in CpG islands. Through the calculations explained in "Results," this corresponds with 7,000 of the estimated 45,000 CpG islands in the human genome (14)
. About 3,500 of these CpG islands represent bladder cancer-specific changes, because these sequences were altered in the genomes of both tumors from both patients. Costello et al. (3)
suggested that 0–4,500 of 45,000 CpG islands (with an average of 600 actually related to the cancer) in the genome were aberrantly methylated between the tumors and matched normal controls of 98 primary human tumors from six types of cancer not including those of the bladder. By comparison, we estimated that 7,000 versus 0–4,500 CpG islands are altered in the genome, and 3,500 versus 600 CpG islands are altered in a cancer-specific manner. Costello et al. (3)
compared separate tumors in independent patients from a variety of malignancies and then provided an average of all of them, which may explain why our values are somewhat different from theirs, as our study focused on a smaller sample set and only bladder cancer.
The nature of the sample set in this study allows us to also address the effect of chemotherapy on global methylation levels. Because most patients receive some treatment after removal of a tumor(s), it is a complication that is poorly understood but biologically relevant. The two patients analyzed here both received an adjuvant methotrexate vinblastine Adriamycin cisplatin (MVAC) cycle after surgical resection of the first tumor. For patient A, additional chemotherapy regimens were applied when the second tumor was noted. However, the patient failed to respond, and the second tumor was then surgically removed. For patient B, the recurrence of transitional cell carcinoma was followed by expiration shortly after the second tumor was removed but before follow-up chemotherapy was implemented. The data, then, supports the notion that for these patients chemotherapy did not seem to result in widespread epigenetic changes, although it remains impossible to distinguish between which methylation changes that did occur were attributable to the initial chemotherapy, disease progression, and/or aging.
By focusing on patients from which we had obtained metachronous tumors of one type of cancer, the estimation of our study allowed differentiation in the number of alterations between normal and tumor (4% and 11%, patient A and B), between patients (13%), and between tumors (5% and 3%, patient A and B). In addition, our data reflect possible alterations associated with progression, although the precise temporal order for the evolution of the metachronous tumors in patient B remains uncertain because they are both metastatic. To our knowledge, this has never been done before, and we conclude that tumor genomes within individual patients have more in common than those observed between different patients.
Methylation Patterns in Vitro.
The extent of methylation changes attributable to the process of culturing tumor cells as a function of time was also of interest to us, because many studies use cell lines to investigate the range of epigenetic changes present in cancers. It was shown previously that normal cells undergo a large number of methylation changes in response to being cultured (13
, 35)
, and this has brought into question the validity of using tumor cell lines as models of methylation pattern alterations in cancer. Relative to the tumors, patient A showed 10% additional differences in methylation pattern in the respective cell lines, and patient B showed 6%, whereas only 3% of these changes were in common between the patients (data not shown). This supports the notion that the majority of the methylation changes occurred in the tumor, with few additional alterations accumulating in the cell line as culture artifacts. The observation is also interesting, because patient A had less methylation changes than patient B when comparing the normal urothelium panel and the tumors, specifically, 4% for patient A and 11% for patient B (Fig. 6)
. The overall variation between patients therefore was 2-fold larger than the overall variation attributable to the culture process, 7% (11–4%) versus 4% (10–6%), respectively. These data provide for the first time a quantitation of the level of methylation in culture of human bladder tumors and argues that cell lines can be used with caution as relevant models for cancer epigenetic studies.
A recent report by Ross et al. (39)
supports this finding in concept through a study of the gene expression of 8,000 unique genes in 60 cancer cell lines using cDNA microarrays. Their results showed a consistent relationship between the gene expression patterns in the cell lines and the tissue of origin. The authors were able to determine specific features as "related to physiological properties of the cell lines" and concluded that for the most part "neither physiological nor experimental adaptation for growth in culture was sufficient to overwrite the gene expression patterns established during differentiation in vivo."
Another observation related to this issue is that increased methylation was observed in the cell lines at sites with preexisting methylation in the genes examined in the tumors. The sites that became methylated correspond to the internal exons and not to the respective promoters of genes shown to become methylated with an increased frequency in tumors and in culture (p16, p15, and PAX6; Refs. 28, 29, 30, 31, 32, 33 ). Methylation patterns seen in cultured tumors may, therefore, be a result of an enhancement of existing patterns as opposed to extensive de novo methylation of previously unmethylated CpG islands.
Several observations can be made when comparing these data to that of a previous study (20)
in regard to the epigenetic versus genetic stability (Fig. 6)
. The majority of epigenetic alterations that occurred in the tumors (in vivo) were hypermethylation events (a total of 14 hyper- versus 5 hypo- and 20 versus 10 in patient A and B, respectively), whereas the majority of epigenetic alterations that occurred in the cell lines (in vitro) were hypomethylation events (a total of 22 hypo- versus 16 hyper- and 13 versus 6 in patient A and B, respectively). For all of the four tumors, the extent of additional epigenetic changes was greater when putting the tumor in culture than with subsequent passage. Interestingly, the extent of genetic alterations (measured as fractional allelic loss) was of the same order of magnitude in both patients as the epigenetic alterations (measured as total epigenetic changes), although patient A showed a 2–3-fold difference between these values. The fractional allelic loss between the patients ranged from approximately 15–53%, whereas the total epigenetic changes between the patients ranged from only approximately 9–18%.
We are currently attempting to additionally describe the regions where differences in methylation pattern were observed in a cancer-specific manner and for which no known matches were found in the sequence databases. In addition, we are conducting studies to determine the role of known genes that became altered in our samples. Whereas our investigation was limited to only two individuals with transitional cell carcinoma of the bladder, our analysis of the epigenetic pattern provides a quantitative profile on the global scale of cancer genomes in vivo and in vitro over time as well as a context for relative stability compared with the genetic pattern.
|
FOOTNOTES |
|---|
1 Supported by NIH Grant 1R01 CA 83867-01 from the National Cancer Institute. 
2 To whom requests for reprints should be addressed, at USC/Norris Comprehensive Cancer Center, 1441 Eastlake Avenue, Mail Stop #83, Los Angeles, CA 90089. Phone: (323) 865-0816; Fax: (323) 865-0102; E-mail: jones_p{at}ccnt.hsc.usc.edu
3 The abbreviations used are: MS.AP-PCR, methylation-sensitive arbitrarily primed PCR; MS.SNuPE, methylation-sensitive single nucleotide primer extension; O/E, observed:expected ratio.
4 http://www.ncbi.nlm.nih.gov/blast.
5 http://www.ncbi.nlm.nih.gov/genome/seq.
6 http://ftp.genome.washington.edu/cgi-bin/RepeatMasker.
7 http://dot.imgen.bcm.tmc.edu:9331/seq-search/gene-search.html.
Received 11/ 8/00. Accepted 5/29/01.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. N. Gal-Yam, G. Egger, L. Iniguez, H. Holster, S. Einarsson, X. Zhang, J. C. Lin, G. Liang, P. A. Jones, and A. Tanay Frequent switching of Polycomb repressive marks and DNA hypermethylation in the PC3 prostate cancer cell line PNAS, September 2, 2008; 105(35): 12979 - 12984. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. W. Laird Cancer epigenetics Hum. Mol. Genet., April 15, 2005; 14(suppl_1): R65 - R76. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Zou, N. K. Osborn, J. J. Harrington, K. K. Klatt, J. R. Molina, L. J. Burgart, and D. A. Ahlquist Frequent Methylation of Eyes Absent 4 Gene in Barrett's Esophagus and Esophageal Adenocarcinoma Cancer Epidemiol. Biomarkers Prev., April 1, 2005; 14(4): 830 - 834. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Chen, C. Rocken, C. Lofton-Day, H.-U. Schulz, O. Muller, N. Kutzner, P. Malfertheiner, and M. P.A. Ebert Molecular analysis of APC promoter methylation and protein expression in colorectal cancer metastasis Carcinogenesis, January 1, 2005; 26(1): 37 - 43. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. F. V. Woude, G. J. Kelloff, R. W. Ruddon, H.-M. Koo, C. C. Sigman, J. C. Barrett, R. W. Day, A. P. Dicker, R. S. Kerbel, D. R. Parkinson, et al. Reanalysis of Cancer Drugs: Old Drugs, New Tricks Clin. Cancer Res., June 1, 2004; 10(11): 3897 - 3907. [Full Text] [PDF]
|
|
|
|
 |
|
 D. Machover, J. Zittoun, R. Saffroy, P. Broet, S. Giraudier, T. Magnaldo, E. Goldschmidt, B. Debuire, M. Orrico, Y. Tan, et al. Treatment of Cancer Cells with Methioninase Produces DNA Hypomethylation and Increases DNA Synthesis Cancer Res., August 15, 2002; 62(16): 4685 - 4689. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. K. Virmani, J. A. Tsou, K. D. Siegmund, L. Y. C. Shen, T. I. Long, P. W. Laird, A. F. Gazdar, and I. A. Laird-Offringa Hierarchical Clustering of Lung Cancer Cell Lines Using DNA Methylation Markers Cancer Epidemiol. Biomarkers Prev., March 1, 2002; 11(3): 291 - 297. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Widschwendter and P. A. Jones The Potential Prognostic, Predictive, and Therapeutic Values of DNA Methylation in Cancer : Commentary re: J. Kwong et al., Promoter Hypermethylation of Multiple Genes in Nasopharyngeal Carcinoma. Clin. Cancer Res., 8: 131-137, 2002, and H-Z. Zou et al., Detection of Aberrant p16 Methylation in the Serum of Colorectal Cancer Patients. Clin. Cancer Res., 8: 188-191, 2002. Clin. Cancer Res., January 1, 2002; 8(1): 17 - 21. [Full Text] [PDF]
|
|
|
|
 |
|
 S. Toyooka, K. O. Toyooka, R. Maruyama, A. K. Virmani, L. Girard, K. Miyajima, K. Harada, Y. Ariyoshi, T. Takahashi, K. Sugio, et al. DNA Methylation Profiles of Lung Tumors Mol. Cancer Ther., November 1, 2001; 1(1): 61 - 67. [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 |
) alterations compared with normal urothelium, and the percentage these sequences represent of the total sequences scored (%). The values below the boxes represent total epigenetic changes (TEC: sum of methylation alterations from the Normal Urothelium panel through the respective point) and fractional allelic loss (FAL: percentage of microsatellite loss of heterozygosity; Ref. 
, values for promoter; 
, values for exons. There is no data for patient A with respect to p16 and p15 because there is a homozygous deletion (HD) in the region of these genes