DNA methyltransferase 1 (DNMT1)–deficient mice are tumor-prone, and this has been proposed to result from the induction of genomic instability. To address whether loss of DNMT1, or the related protein DNMT3b, results in genomic instability in human cancer cells, we used a near-diploid human colorectal cancer cell line, HCT116, in which one or both DNMT genes were disrupted by homologous recombination. Array-based comparative genomic hybridization analyses indicated that double, but not single, DNMT knock-out cells display two specific alterations in regional DNA copy number, suggesting that DNMT deficiency and genomic DNA hypomethylation are not associated with widespread genomic amplifications or deletions in human cancer cells. However, spectral karyotype analyses revealed that DNMT-deficient HCT116 cells are highly unstable with respect to large-scale chromosomal alterations; furthermore, this effect is characterized by a high degree of individual cell heterogeneity. The induction of chromosomal alterations in DNMT-deficient cells was evidenced both by aneuploidy and by large increases in the number of novel chromosomal translocations. Studies of double knock-out cells indicated that the generation of chromosomal alterations is spontaneous and persistent in vitro, meeting the formal definition of genomic instability. In summary, we show that DNMT deficiency in human cancer cells results in constitutive genomic instability manifested by chromosomal translocations.

The best-established epigenetic alterations in cancer are changes in the level of cytosine DNA methylation. In many human neoplasms, the methylation of specific genetic loci, methylation of repetitive sequences, or genomic 5-methyl-2′-deoxycytidine levels are reduced (1, 2). To date, the contribution of genomic DNA hypomethylation to human oncogenesis remains obscure. Originally, it was hypothesized that DNA hypomethylation in cancer is linked to oncogene activation (3). Although this may be the case in select instances, it does not seem to account for the global loss of DNA methylation from genomic DNA found in many human tumors (1, 2). A second model invoked to explain the oncogenic role of genomic DNA hypomethylation is genomic instability (1, 2), and investigations of the effect of genetic disruption of DNA methylation in murine systems provide evidence for this model (4, 5). Genomic instability is a cellular state characterized by an increased rate of genetic changes, including DNA sequence changes, aneuploidy, chromosome translocations, and/or gene amplifications (6). Despite the data from murine model systems, it remains plausible that human cancer cells (adult, somatic) respond differently to DNA hypomethylation and/or DNMT loss than do murine embryonic cells. Recently, Vogelstein and coworkers developed human cancer cell lines with varying levels of genomic DNA hypomethylation resulting from genetic knock-out of DNMT1 and/or DNMT3b (7, 8). Critically, the parental cell type, HCT116, a human colorectal carcinoma cell line, has a near-diploid karyotype (9). Thus, the HCT116 DNMT knock-out model is ideally suited for assessing the role of DNA methylation, DNMT1, and DNMT3b, in maintaining genomic stability in human cancer cells.

Cell lines. HCT116 parental and knock-out colon cancer cell lines [p21−/−; DNMT1−/−; DNMT3b−/−; and DNMT1−/−, DNMT 3b−/− (dual knock-out cells, DKO)] were a generous gift from Dr. Bert Vogelstein (Sidney Kimmel Cancer Center, Johns Hopkins University Medical Institutions, Baltimore, MD). DKO passage number indicates the number of cellular passages following the dual genetic knock-out. HCT116 Utah cells were kindly provided by Dr. David Jones (Huntsman Cancer Institute, University of Utah, Salt Lake City, UT). Cells were cultured as described previously (10).

BAC array comparative genomic hybridization. Genomic DNA from control and DNMT knock-out cell lines was isolated by standard methods, and 1 μg of DNA, which corresponds to ∼100,000 cells, was used for BAC array comparative genomic hybridization (aCGH) using the RPCI-11 BAC array, as described previously (11). DNA labeling, array hybridization, and digital data acquisition were done as described previously (11). Male and female DNA pools, each of which contained DNA from 20 cytogenetically normal individuals, were used as controls.

Spectral karyotype analysis. Spectral karyotype analysis (SKY) was done essentially as described elsewhere (12). Fluorescence color images of chromosomes developed by Rhodamine, Texas Red, Cy5, FITC, and Cy5.5 were captured under a Nikon microscope, equipped with a Spectral cube and Interferometer module. SKY View software (version 1.62), was used to sort numerical changes and structural alterations of chromosomes, including simple balanced translocations, unbalanced (or nonreciprocal) translocations, deletions, and duplications.

Quantitative measurement of chromosomal instability. We calculated two independent measures of chromosomal instability. The first, Instability index, was defined as the mean number of chromosomal rearrangements per mitotic cell, after subtraction of the rearrangements (i.e., stably present translocations) found in the HCT116 parental cell line. The second measure, Rearrangement diversity, was defined as the total number of distinct chromosomal rearrangements across a cell population, divided by the number of mitotic cells analyzed. Rearrangements scored included translocations, insertions, deletions, tandem duplications, and iso-chromosomes. Reciprocal and balanced translocations involving two chromosomes and/or iso-chromosomes were scored as two rearrangements.

DNMT knock-out has subtle effects on genomic DNA loci copy number in HCT116 cells. Genetic knock-out of one or both key DNA methyltransferase genes, DNMT1 and DNMT3b, were previously generated in the HCT116 colorectal cancer cell line (7, 8). We verified the expected genotype of the DNMT knock-out cell lines by measuring the expression of DNMT1 and DNMT3b by Western blot or reverse transcription-PCR analyses, respectively (data not shown). In addition, we confirmed the DNA methylation phenotype of the knock-out cells, previously reported by Rhee et al. (8), using a novel mass spectrometry assay (ref. 10; data not shown). Specifically, we found that compared with HCT116 parental cells, DNMT1−/− cells had a 20% reduction in genomic DNA methylation, DNMT3b−/− showed no reduction, and DKO had a reduction of ∼80%.

To assess the inherent stability of the HCT116 genotype, we compared two independently maintained lines of HCT116 cells, HCT116 Utah and HCT116 parental. The HCT116 parental cells are the line from which the DNMT knock-out cell lines were constructed. aCGH analysis indicated that the two HCT116 cell lines are nearly identical, with the exception of two specific genomic losses in the HCT116 parental cells, an interstitial Ch.13 deletion and a telomeric Ch.7p deletion (data not shown). This result suggests that whereas HCT116 cells may occasionally acquire mutations in vitro, they likely have no underlying genomic instability, a notion that is also supported by the results of SKY analyses (see below). Next, we used aCGH to compare the genome of DNMT knock-out cells to HCT116 parental cells. As a negative control, we analyzed p21−/− HCT116 cells (13). We did not detect any DNA copy number alterations in either of the individual DNMT knock-out lines, or in the p21−/− cells (Fig. 1A; data not shown). However, DKO cells contained two specific amplifications, one at 11q23 and the other at 18p (Fig. 1A -C, arrows).

Figure 1.

Genome copy number alterations in DNMT knock-out HCT116 cell lines. Genomic DNA was isolated from the indicated cell lines and used for aCGH analysis, as described in Materials and Methods. A, genomic copy number analysis of HCT116 parental and DNMT knock-out cell lines. Chromosome numbers are indicated below the graphs. No differences between the HCT116 parental cell line and the DNMT1−/− or DNMT3b−/− knock-out cells were observed. In contrast, two specific amplifications (arrows) were present in DKO cells at both passages 8 and 25. B, high-resolution graphs of the DKO cell-specific chromosome 11q23 amplification shown in (A). C, high-resolution graphs of the DKO cell-specific chromosome 18p amplification shown in (A).

Figure 1.

Genome copy number alterations in DNMT knock-out HCT116 cell lines. Genomic DNA was isolated from the indicated cell lines and used for aCGH analysis, as described in Materials and Methods. A, genomic copy number analysis of HCT116 parental and DNMT knock-out cell lines. Chromosome numbers are indicated below the graphs. No differences between the HCT116 parental cell line and the DNMT1−/− or DNMT3b−/− knock-out cells were observed. In contrast, two specific amplifications (arrows) were present in DKO cells at both passages 8 and 25. B, high-resolution graphs of the DKO cell-specific chromosome 11q23 amplification shown in (A). C, high-resolution graphs of the DKO cell-specific chromosome 18p amplification shown in (A).

Close modal

DNMT knock-out induces aneuploidy in HCT116 cells. We next did SKY, an individual cell-based analysis, to examine DNMT-deficient HCT116 cell cultures for evidence of chromosomal instability. Whereas the mean chromosome number of the DNMT knock-out lines was not significantly altered from the HCT116 parental cells (data not shown), many of the DNMT-deficient cells displayed a loss or gain of one or a few chromosomes, i.e., aneuploidy (Fig. 2A). Accordingly, statistical analysis revealed that chromosome number variance was sharply increased in each of the DNMT knock-out lines, as compared with either HCT116 parental or HCT116 Utah cells (Fig. 2B; data not shown). This effect was most pronounced in the DKO cell line, and was approximately additive with respect to the contribution of the individual DNMT knock-outs (Fig. 2B).

Figure 2.

Aneuploidy in DNMT-deficient HCT116 cell lines. Chromosome number was determined by SKY analysis of metaphase spreads from the indicated cell lines. A, chromosome number of DNMT-deficient cell lines. Each point represents an individually analyzed cell. B, chromosome number variance (i.e., the degree of aneuploidy) of each cell line shown in (A) was calculated by ANOVA single factor testing, using Microsoft Excel.

Figure 2.

Aneuploidy in DNMT-deficient HCT116 cell lines. Chromosome number was determined by SKY analysis of metaphase spreads from the indicated cell lines. A, chromosome number of DNMT-deficient cell lines. Each point represents an individually analyzed cell. B, chromosome number variance (i.e., the degree of aneuploidy) of each cell line shown in (A) was calculated by ANOVA single factor testing, using Microsoft Excel.

Close modal

Karyotype analysis of DNMT knock-out HCT116 cells. Detailed analysis of mitotic spreads of the HCT116 parental cell line by SKY revealed a few highly conserved chromosomal translocations (Fig. 3A,, white bars; Supplementary Table S1). The majority of these common alterations were also present in each of the DNMT knock-out lines (Fig. 3B,-D, white bars; Supplementary Table S1). Notably, we also observed a number of novel chromosomal rearrangements in the DNMT knock-out cell lines (Fig. 3B,-D, yellow bars; Supplementary Table S1). The novel chromosomal alterations present in DNMT-deficient cells were primarily comprised of reciprocal and nonreciprocal translocations. DKO cells in particular displayed a disproportionately high number of novel chromosomal rearrangements, involving the formation of bipartite and tripartite chromosomes (Fig. 3D; Supplementary Fig. S1).

Figure 3.

SKY analysis of DNMT-deficient HCT116 cell lines. Cells were processed for analysis by SKY as described in Materials and Methods, and a representative metaphase spread of each cell line is shown. A, HCT116 parental cells. B, DNMT1−/− cells. C, DNMT3b−/− cells. D, DKO p25 cells. White and yellow bars, chromosomal translocations present or not present in the HCT116 parental cell line, respectively.

Figure 3.

SKY analysis of DNMT-deficient HCT116 cell lines. Cells were processed for analysis by SKY as described in Materials and Methods, and a representative metaphase spread of each cell line is shown. A, HCT116 parental cells. B, DNMT1−/− cells. C, DNMT3b−/− cells. D, DKO p25 cells. White and yellow bars, chromosomal translocations present or not present in the HCT116 parental cell line, respectively.

Close modal

DNMT knock-out induces chromosomal instability in HCT116 cells. To quantify the chromosomal alterations in DNMT knock-out cells, we first calculated an instability index, which is a measure of the total number of rearrangements observed in each cell type (see Materials and Methods). The HCT116 Utah line displayed the lowest instability index (raw instability value of 1.25; Table 1), suggesting again that HCT116 cells have a relatively stable genotype in vitro. p21−/− cells had a 1.2-fold increase in the instability index relative to HCT116 Utah, indicating that the conditions used to drive homologous recombination in the HCT116 system does not by itself cause sizable increases in instability (data not shown). In contrast, DNMT1−/−, and DNMT3b−/− cells showed approximately 2- and 4-fold increases in the instability index, respectively (Table 1). In DKO cells, the effect was far more dramatic, suggesting synergy (Table 1). The instability index values in both the single and dual DNMT knock-out cell lines were significantly different than that found in HCT116 Utah (Supplementary Table S2). We also examined DKO cells harvested at different passage numbers, and these all showed high levels of chromosomal instability (Table 1). In addition to the instability index, we calculated rearrangement diversity, a measure of the number of distinct rearrangements observed in each cell type (see Materials and Methods). The chief distinction between the instability index and rearrangement diversity is that the instability index value is increased by conserved rearrangements in the cell population, whereas rearrangement diversity is not. Similar to the instability index, HCT116 Utah cells showed a low value for rearrangement diversity (Table 1). However, unlike the instability index, the rearrangement diversity of DNMT3b−/− cells was only slightly increased (1.5-fold) over HCT116 Utah, and was not statistically significant (Supplementary Table S2). In contrast, DNMT1−/− cells showed a profound enhancement in rearrangement diversity (4.01-fold), and this was highly significant (Supplementary Table S2). The rearrangement diversity of the DKO cell line was substantially greater than either of the individual DNMT knock-out lines, again pointing to a synergistic effect (Table 1). The extraordinary phenotype of DKO cells was also evidenced by the fact that ∼90% of DKO p25 cells analyzed contained at least one unique chromosomal rearrangement (data not shown).

Table 1.

Quantification of genomic instability in DNMT knock-out cell lines

Cell lineCells analyzedChromosomal rearrangements*Instability indexFold increaseRearrangement diversity§Fold increase
HCT116 Utah 32 40 1.25 1.0 0.343 1.0 
DNMT1−/− 40 93 2.33 1.86 1.375 4.01 
DNMT3b−/− 35 186 5.31 4.25 0.514 1.50 
DKO p8 38 456 12.00 9.60 2.89 8.43 
DKO p25 54 529 9.80 7.84 2.61 7.61 
DKO p46 70 842 12.03 9.62 2.46 7.17 
Cell lineCells analyzedChromosomal rearrangements*Instability indexFold increaseRearrangement diversity§Fold increase
HCT116 Utah 32 40 1.25 1.0 0.343 1.0 
DNMT1−/− 40 93 2.33 1.86 1.375 4.01 
DNMT3b−/− 35 186 5.31 4.25 0.514 1.50 
DKO p8 38 456 12.00 9.60 2.89 8.43 
DKO p25 54 529 9.80 7.84 2.61 7.61 
DKO p46 70 842 12.03 9.62 2.46 7.17 
*

Not observed in HCT116 parental cells.

Mean number of chromosomal rearrangements per cell, following subtraction of the conserved rearrangements present in the HCT116 parental cell line.

Relative to the HCT116 Utah cell line.

§

Total number of distinct chromosomal rearrangements observed across the cell population, divided by the number of cells analyzed.

These striking results encouraged us to directly examine whether the DKO cell line generates novel chromosomal rearrangements in vitro or, alternatively, whether the diversity of rearrangements observed in this cell type are generated only during the initial establishment of the line. To this end, we examined the specific chromosomal rearrangements found in the DKO cell line at the three different passage numbers indicated in Table 1. Remarkably, >60% of the rearrangements observed in DKO p46 cells were unique to DKO cells of that passage number. In addition, there was a similar predominance of unique rearrangements in DKO cells at the earlier passage numbers (data not shown).

We have investigated the potential link between DNA hypomethylation and/or DNMT loss and genomic instability in human cancer cells. Herein, we present a number of novel findings: (a) DNMT knock-out cells display aneuploidy, (b) DNMT knock-out cells show robust increases in multiple types of chromosomal rearrangements, and (c) DNMT deficiency is characterized by the spontaneous and persistent generation of chromosomal rearrangements. Taken together, our data reveal that DNMT loss, and the associated DNA hypomethylation, results in bona fide chromosomal instability (change/time) in human cancer cells. The extent of chromosomal instability in the DKO cell line, in which ∼90% of cells show at least one chromosomal rearrangement, seems to be comparable to, or possibly exceeding, that observed in other well-characterized genetically unstable human cancer cell lines (9, 14, 15).

We found, using aCGH analyses, that DNMT deficiency and DNA hypomethylation in human cancer cells does not result in widespread conserved genomic amplifications or deletions. Because this assay requires thousands of cells for the input DNA, cellular heterogeneity can mask copy number alterations present in subsets of the cell population. aCGH did not reveal copy number alterations in single DNMT knock-out cells, however, in DKO, two specific amplifications were observed, at 11q23 and 18p. 11q23 alterations are relatively common in leukemias, and have also been observed in solid tumors; 18p alterations, although far less frequent in human cancer, have also been noted (16). These genomic regions contain a number of established and candidate oncogenes, including MLL and Ets1 on 11q23, and Yes1 on 18p (16). We hypothesize that the two amplifications observed in DKO cells do not indicate genomic instability but instead reflect selectable genetic events that promote DKO cell growth or survival.

SKY analyses, an individual cell-based assay, revealed that DNMT-deficient cell lines each display varying degrees of chromosomal instability. A striking effect was seen in DKO cells, which showed approximately an order of magnitude increase in chromosomal instability relative to HCT116 parental cells. Experiments comparing DKO cells of different passage numbers revealed that chromosomal instability is a constitutively expressed genetic property of these cells. Moreover, there seemed to be no statistical bias towards the involvement of any particular chromosome members in the rearrangements (data not shown). The increased structural rearrangements in DNMT knock-out cells include both balanced and unbalanced translocations, and likely result from increased rates of DNA recombination. This hypothesis is consistent with earlier work, which found that DNA methylation suppresses crossing-over in Ascobolus, and that DNMT1 knock-out mouse ES cells show enhanced rates of gene targeting (17, 18).

One of the key questions arising from our study is whether chromosomal instability in the DNMT knock-out lines is a direct consequence of genomic DNA hypomethylation, or alternatively, results from the loss of methylation-independent protein functions. In support of the first model, dual DNMT loss has a synergistic, rather than an additive, effect on chromosomal instability, which is analogous to its effect on genomic DNA hypomethylation. Also consistent with this model, DNMT1−/− cells, which show moderate levels of genomic DNA hypomethylation, have significantly higher rearrangement diversity than DNMT3b−/− cells, which do not have detectable genomic DNA hypomethylation. On the other hand, DNMT3b−/− cells, which do not show global DNA hypomethylation, display moderate increases in chromosomal instability, suggesting a role for DNMT3b in maintaining genomic stability independent of its enzymatic activity. This could potentially be explained by the participation of DNMT3b in the condensin complex, which is involved in proper segregation of sister chromatids during mitosis (19). However, even in the case of DNMT3b−/− cells, we cannot exclude the possibility that loss of methylation at specific loci is critical for the instability phenotype. For example, enzymatic region mutations of DNMT3b, which are associated with the human disease ICF syndrome (immunodeficiency, centromere instability, and facial anomalies), lead to hypomethylation of and chromosomal abnormalities involving the pericentromeric regions of chromosomes 1, 9, and 16 (20). In the current study, DNMT3b−/− cells showed two conserved rearrangements involving chromosome 1 [t(1;7), der(3)t(1;3)], and one involving chromosome 16 [der(16)t(8;16)] (Supplementary Table S1). Thus, it will be relevant to determine whether the chromosomal rearrangements observed in DNMT3b−/− cells are related to hypomethylation of specific DNA sequences.

Cytosine DNA methyltransferases have been proposed as suitable targets for pharmacologic intervention in cancer based on the fact that tumor suppressor genes are silenced by DNA hypermethylation (21). Our data, which indicate that DNA hypomethylation leads to chromosomal instability in human cancer cells, could be interpreted either positively or negatively from the standpoint of epigenetic therapy. The interpretation is dependent on whether instability shows specificity towards tumor cells, and to what extent (and in what direction) instability affects tumor cell fitness. The potentially adverse effects of DNA hypomethylation have led us to previously suggest that the most effective clinical use of DNMT inhibitors may be to combine these drugs, in a temporary and acute fashion, with secondary agents whose efficacy is enhanced by DNA hypomethylation (21). Our current findings seem to lend further support to this idea.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

Grant support: Ralph Wilson Medical Research Foundation, Phi Beta Psi, and the Roswell Park Alliance Foundation (to A.R. Karpf), and CA16056 (to Roswell Park Cancer Institute). The RPCI SKY/FISH Facility was established by a donation from the J.H. Cummings Foundation.

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.

We are grateful to Dr. Bert Vogelstein for providing the DNMT and p21 knock-out cell lines used in this study, and to Dr. David Jones for providing the HCT116 Utah cell line. We thank Jeff Conroy, Devin McQuaid, and Jeff LaDuca of the RPCI Genomics Core Facility for excellent technical assistance. We thank Dr. Terry Beerman of RPCI for critical reading of the manuscript.

1
Ehrlich M. DNA methylation in cancer: too much, but also too little.
Oncogene
2002
;
21
:
5400
–13.
2
Feinberg AP, Tycko B. The history of cancer epigenetics.
Nat Rev Cancer
2004
;
4
:
143
–53.
3
Feinberg AP, Vogelstein B. Hypomethylation distinguishes genes of some human cancers from their normal counterparts.
Nature
1983
;
301
:
89
–92.
4
Eden A, Gaudet F, Waghmare A, Jaenisch R. Chromosomal instability and tumors promoted by DNA hypomethylation.
Science
2003
;
300
:
455
.
5
Dodge JE, Okano M, Dick F, et al. Inactivation of dnmt3b in mouse embryonic fibroblasts results in DNA hypomethylation, chromosomal instability, and spontaneous immortalization.
J Biol Chem
2005
;
280
:
17986
–91.
6
Lengauer C, Kinzler KW, Vogelstein B. Genetic instabilities in human cancers.
Nature
1998
;
396
:
643
–9.
7
Rhee I, Jair KW, Yen RW, et al. CpG methylation is maintained in human cancer cells lacking DNMT1.
Nature
2000
;
404
:
1003
–7.
8
Rhee I, Bachman KE, Park BH, et al. DNMT1 and DNMT3b cooperate to silence genes in human cancer cells.
Nature
2002
;
416
:
552
–6.
9
Lengauer C, Kinzler KW, Vogelstein B. Genetic instability in colorectal cancers.
Nature
1997
;
386
:
623
–7.
10
Song L, James SR, Kazim L, Karpf AR. Specific method for the determination of genomic DNA methylation by liquid chromatography-electrospray ionization tandem mass spectrometry.
Anal Chem
2005
;
77
:
504
–10.
11
Cowell JK, Nowak NJ. High-resolution analysis of genetic events in cancer cells using bacterial artificial chromosome arrays and comparative genome hybridization.
Adv Cancer Res
2003
;
90
:
91
–125.
12
Ried T, Koehler M, Padilla-Nash H, Schroek E. Chromosome analysis by spectral karyotyping. In: Spector DL, Goldman RD, Leinwand LA, editors. Cells: a laboratory manual. New York: CSHL Press; 1997.
13
Bunz F, Dutriaux A, Lengauer C, et al. Requirement for p53 and p21 to sustain G2 arrest after DNA damage.
Science
1998
;
282
:
1497
–501.
14
Rajagopalan H, Jallepalli PV, Rago C, et al. Inactivation of hCDC4 can cause chromosomal instability.
Nature
2004
;
428
:
77
–81.
15
Jallepalli PV, Waizenegger IC, Bunz F, et al. Securin is required for chromosomal stability in human cells.
Cell
2001
;
105
:
445
–57.
16
Huret JL, Dessen P, Bernheim A. Atlas of genetics and cytogenetics in oncology and haematology, year 2003.
Nucleic Acids Res
2003
;
31
:
272
–4.
17
Maloisel L, Rossignol JL. Suppression of crossing-over by DNA methylation in Ascobolus.
Genes Dev
1998
;
12
:
1381
–9.
18
Dominguez-Bendala J, McWhir J. Enhanced gene targeting frequency in ES cells with low genomic methylation levels.
Transgenic Res
2004
;
13
:
69
–74.
19
Geiman TM, Sankpal UT, Robertson AK, et al. Isolation and characterization of a novel DNA methyltransferase complex linking DNMT3B with components of the mitotic chromosome condensation machinery.
Nucleic Acids Res
2004
;
32
:
2716
–29.
20
Tuck-Muller CM, Narayan A, Tsien F, et al. DNA hypomethylation and unusual chromosome instability in cell lines from ICF syndrome patients.
Cytogenet Cell Genet
2000
;
89
:
121
–8.
21
Karpf AR, Jones DA. Reactivating the expression of methylation silenced genes in human cancer.
Oncogene
2002
;
21
:
5496
–503.

Supplementary data