Non-random distribution of instability-associated chromosomal rearrangement breakpoints in human lymphoblastoid cells

Mutation Research 600 (2006) 113–124

Non-random distribution of instability-associated chromosomal rearrangement breakpoints in human lymphoblastoid cells Stephen R. Moore a,b , David Papworth b , Andrew J. Grosovsky a,∗ a

Environmental Toxicology Graduate Program, Department of Cell Biology and Neuroscience, University of California, Riverside, CA, USA b Radiation and Genome Stability Unit, Medical Research Council, Harwell, Oxfordshire, UK Received 30 July 2005; received in revised form 15 March 2006; accepted 24 March 2006 Available online 23 May 2006

Abstract Genomic instability is observed in tumors and in a large fraction of the progeny surviving irradiation. One of the best-characterized phenotypic manifestations of genomic instability is delayed chromosome aberrations. Our working hypothesis for the current study was that if genomic instability is in part attributable to cis mechanisms, we should observe a non-random distribution of chromosomes or sites involved in instability-associated rearrangements, regardless of radiation quality, dose, or trans factor expression. We report here the karyotypic examination of 296 instability-associated chromosomal rearrangement breaksites (IACRB) from 118 unstable TK6 human B lymphoblast, and isogenic derivative, clones. When we tested whether IACRB were distributed across the chromosomes based on target size, a significant non-random distribution was evident (p < 0.00001), and three IACRB hotspots (chromosomes 11, 12, and 22) and one IACRB coldspot (chromosome 2) were identified. Statistical analysis at the chromosomal band-level identified four IACRB hotspots accounting for 20% of all instability-associated breaks, two of which account for over 14% of all IACRB. Further, analysis of independent clones provided evidence within 14 individual clones of IACRB clustering at the chromosomal band level, suggesting a predisposition for further breaks after an initial break at some chromosomal bands. All of these events, independently, or when taken together, were highly unlikely to have occurred by chance (p < 0.000001). These IACRB band-level cluster hotspots were observed independent of radiation quality, dose, or cellular p53 status. The non-random distribution of instability-associated chromosomal rearrangements described here significantly differs from the distribution that was observed in a first-division post-irradiation metaphase analysis (p = 0.0004). Taken together, these results suggest that genomic instability may be in part driven by chromosomal cis mechanisms. © 2006 Published by Elsevier B.V. Keywords: Genomic instability; Chromosomal rearrangements; Distribution; TK6

1. Introduction Environmental exposure to ionizing radiation is a human health hazard and concern. Recent evidence has

∗ Corresponding author. Tel.: +1 951-827-7750; fax: +1 951-827-7745. E-mail address: [email protected] (A.J. Grosovsky).

0027-5107/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.mrfmmm.2006.03.006

demonstrated that even very low doses of radiation are capable of resulting in an instability phenotype resembling that observed in tumors [1–3]. Genomic instability is a pleiotropic hallmark of tumorigenic progression, marked by increased mutation rates, delayed reproductive cell death, and karyotypic instability [1,4]. Genomic instability is also observed in a fraction of the progeny of irradiated cells, which is not only important for human exposure studies, but also provides a powerful labora-

114

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

tory paradigm with which to study tumorigenic progression. Because mutations or loss of critical proteins are observed in tumor cells, and cell lines deficient in critical proteins are unstable [5–7], instability is often attributed to alterations in trans factors, such as p53 [8]. There is increasing evidence suggesting that chromosomal, or cis, mechanisms may play an important role in the initiation or promulgation of instability [9–15]. Some recent in vitro studies have demonstrated that persistent chromosomal rearrangements may be due to chromosomal mechanisms, such as: telomere dysfunction and the formation of dicentric chromosomes leading to fusion–bridge–breakage cycles [7,16–19]; unstable rearrangement junctions, including repeat sequences [20,21]; spontaneously expressed fragile sites [22]; jumping translocations [23]; global changes, such as altered replication timing, which affects metaphase chromosomal condensation [12], and chromatin compaction as a result of a rearrangement [24]. In vivo mouse studies have also demonstrated that global hypomethylation can result in chromosomal fragility and tumorigenesis [25,26]. We have recently demonstrated that some chromosomal regions, might serve as focal centers for instability associated breaks and rearrangements in TK6 human lymphoblasts [9,10]; however, these studies were carried out on a limited number of clones. Chromosomal rearrangements in cancer have been extensively described (Mitelman 20001 ). Often, these exhibit a non-random pattern of distribution within the karyotype, with some chromosomes involved in rearrangements more often than others; this is largely attributable to the formation of fusion peptides via rearrangements that are known to confer a growth advantage to the cell, loss or gain of tumor suppressor protein sequence or oncogenes sequences, respectively, through chromosomal breaks or anueploidy (Mitelman 20001 ), or expression of chromosomal fragile sites [27]. Chromosomal rearrangements that occur during tumorigenic progression can also be highly complex [28], with increasing complexity often correlating with poor prognosis [17,18,23]. Karyotypic analysis of human tumors suggests that chromosomal rearrangements in tumors are non-randomly distributed within the karyotype [29–33] and may focus on particular chromosomes or bands [29,33]. Although some groups have been able to temporally order the chromosomal events observed in tumors in retrospective analyses [34,35], experimental initiation of chromosomal rearrangements using ionizing radiation in

1

Updated at http://www.cgap.nci.gov/Chromosomes/Mitelman.

systems with a stable karyotype has provided important information relating to the progression of karyotypic instability. Kano and Little [36] reported a non-random distribution of chromosomal rearrangement breakpoints in a population analysis of human diploid fibroblasts, after several doses of X-irradiation and up to 59 days post-irradiation. Chromosome 1 was a statistically significant hotspot, whereas chromosomes 8 and 13 were coldspots, and several chromosomal bands were identified where clustering of breakage was postulated to have occurred. However, an analysis of induced, delayed rearrangements on an individual clone basis would avoid the interpretative complications that often accompany bulk-population karyotypic analyses, such as questions concerning independence of rearrangement events, and evidence of breakpoint clustering. Some such clonal studies have been performed using conventional cytogenetics, but no evidence for preferential involvement of particular chromosomes was observed [15,37,38]; although it is likely that in these studies too few instability-associated rearrangements were examined to facilitate a complete distribution analysis. Exposure to low LET radiation causes random ionizations [39,40] and consequently, breaks associated with low LET exposure should be random and determined by target size (i.e. chromosome size) considerations. This is not always the case, and appears to depend on the cell system used and endpoints examined [41]. Firstdivision metaphase analyses in TK6 cultures that have been exposed to low LET radiation suggest that chromosomal damage is random according to chromosome size or DNA content [42]. Exposure to high LET radiation results in clustered damage, and results in patterns of breaks and rearrangements at first division that may reflect chromosomal target size or potential juxtaposition of chromosome territories, as demonstrated by a recent study [43]. These rearrangements tend to be complex, i.e. involving several chromosomes [43]. Although the damage is clustered and rearrangements may reflect interphase chromosome domains, the chance of a particular target being ‘hit’ is still dependent on target size, as with low LET exposure. Consequently, with both high and low LET exposures, initial chromosomal damage should reflect target size and be randomly distributed within karyotype. However, instability-associated chromosomal rearrangements are heterogeneously expressed, occurring in a delayed and stochastic manner during the outgrowth of individual clones surviving radiation exposure [9,10,15]. As these are non-clonal events, they are thought not to occur during the clonal outgrowth of individual clones, and imply that instability is in some part driven in cis. In this study, we tested the hypothe-

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

sis that genomic instability may be partially due to the generation of breakage prone sites or regions at chromosomal rearrangement junctions. Two predictions follow this hypothesis: (i) the composite karyotypic distribution of a large set of instability-associated chromosomal rearrangement breakpoints should significantly differ from the null hypothesis of random distribution according to chromosomal size and (ii) the number of unstable clones that exhibit chromosomal rearrangement breakpoint clustering should significantly exceed the number expected if delayed breaks occur randomly and independently of prior chromosomal rearrangements within a persistently unstable clone.

2. Materials and methods 2.1. Cell lines The current analysis includes karyotypic data from TK6 human B lymphoblastoid cells and isogenic derivatives differing in cellular p53 status and apoptotic capacity; TK6E6 (functionally devoid of p53 via cellular expression of the HPV 16 E6 oncoprotein; [44]), NH32 (p53 null via targeted disruption of both alleles; [45]) and BclXL (over-express the anti-apoptotic factor BCLXL and deficient in p53-mediated apoptosis; [46]). TK6E6 and NH32 cell lines were kind gifts of Dr. John Little (Harvard University) and Dr. Howard Liber (Colorado State University). Data presented on BclXL cells was collected in collaboration with Dr. Amy Kronenberg, Lawrence Berkeley National Laboratory. Both spontaneously arising clones (plated at very low density for collection of independently derived clones) and clones arising after treatment (plated at a treatment-specific density that allows collection of independently derived clones) are included in the current study. Also, some daughter clones generated by lowdensity plating of independent clones surviving treatment are included; these daughter clones are called subclones throughout this paper. Treatment details for each group of clones used the current analysis are given in Table 1. Stable clones (Group A) are only included in Table 1 for comparison; they did not contribute to the statistical analysis presented herein as they are uninformative in terms of instability-associated breaks. All cultures were maintained at 37 ◦ C and 5% CO2 in supplemented RPMI 1640 media, 10% bovine calf serum, 1% Pen-Strep and 1% l-glutamine, and were maintained in log phase growth (2 × 105 cells/ml). After treatment (Table 1), cells were plated at a very low density in duplicate 96 well Costar dishes (Corning, USA). Cells were plated in each well at densities that were determined by existing survival data at each dose. Plating at very low density assured that each positive well represented the clonal outgrowth of a single surviving cell [15]. At 14 days post-irradiation, independent clones were expanded for cytogenetic analysis.

115

2.2. Cytogenetic analysis Cells were harvested according to established protocols [9,10,15]. Briefly, 1–5 × 106 cells were blocked at metaphase by 2 h 37 ◦ C incubation with colcemid (10 (g/ml; Gibco BRL). Cells were resuspended in 10 ml pre-warmed 0.075 M KCl, and incubated for 12 min at 37 ◦ C. After hypotonic treatment, cells were pre-fixed with 1 ml of ice-cold Carnoy’s fixative (3:1, v/v, methanol:acetic acid). Cells were centrifuged and the pellet was washed three times in ice-cold fixative. Clean slides were prepared and aged for 2 days on a 60 ◦ C slide warmer prior to GTG banding. Slides were examined on an Olympus BH2 light microscope. Complete karyotypic analysis was performed on ten to thirty metaphases per clone by one or more of three trained cytogeneticists in our lab. For the current analysis, only structural chromosomal rearrangements were sufficient to classify a particular clone as unstable [9,10,15]. Clones were classified as unstable (Table 1) if one or more of the following criteria were met: multiple, distinct subclonal populations (metaphases) in a single clone were evident; an individual metaphase contained a multiply rearranged chromosome (e.g. a marker) or dicentric; rearrangements involving centromeric breakpoints were observed. Rearrangements involving centromeric breakpoints have been postulated to be important breakage prone focal centers for persistent genomic instability [9,10,14,15]. Herein, no further distinction will be made as to parameters by which any particular clone was classified as unstable except in the cases where individual chromosomal breaksites are discussed. 2.3. First division metaphase analysis Exponentially growing TK6 cells were divided into three flasks at 75,000 cells/ml 20 h prior to irradiation with 4 Gy 137 Cs ␥-rays. Cultures were incubated at 37 ◦ C in a humidified incubator for 20 and 30 h to enrich for the population irradiated in G1 [47,48]. Metaphases were collected for karyotypic analysis as above. Parallel flasks were treated with cytochalasin B (final concentration 3 ␮g/ml) to measure the fraction irradiated entering first division at time of analysis; induction of micronuclei was not examined in this group and it was only used as a control to indicate cell cycle progression after irradiation. One thousand micronuclei were examined per group, with approximately 90% of cells in first division at 20 h compared to 12% in the sham-irradiated controls (where most cells had undergone at least one division) (data not shown). After establishing the timing of the radiation-induced cell cycle delay, we examined chromosomal aberrations in 50 metaphase spreads from the 20 h group (irradiated and sham-irradiated). No breaks or rearrangements were observed in the sham-irradiated group. In the irradiated sample 129 breaks were observed and classified according to the chromosomal localization of the breaksite. First division breaks as a function of chromosome size were compared to instability-associated rearrangement breaks using a paired, two-tailed Student’s t-test. Descriptions of the relationship between target size and either instabilityassociated or first-division breaks were generated using

116

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

Table 1 Description of clones used in the current project Projecta

Treatmentb

A

2 Gy 137 Cs

75

75

0

0.00

B

Benzo[a]pyrene diol epoxide Spontaneous 2 Gy 137 Cs

23 11 6

12 5 4

11 6 2

47.83 54.55 33.33

C

1 GeV Fe nuclei 55 MeV Protons

27 51

20 41

7 10

25.93 19.61

D

1 GeV Fe nuclei 55 MeV Protons

65 65

50 52

15 13

23.08 20.00

E

15 × 0.2 Gy 137 Cs 0.2 Gy 137 Cs

40 30

28 23

12 7

30.00 23.33

F G H I J

2 Gy 137 Cs 2 Gy 137 Cs 2 Gy 137 Cs 15 × 0.2 Gy 137 Cs Sub-clones 2 Gy 137 Cs Sub-clones

25 19 20 24 24

23 12 12 16 14

2 7 8 8 10

8.00 36.84 40.00 33.33 41.67

505

387

118

27.88

Total Clones

Total clones

Stable clones

Unstable clones

Percent unstable

a

The projects were labelled alphabetically to indicate that they each represent distinct experiments. (A) APRT mutant TK6 clones selected after ␥-irradiation. The 75 stable clones from project A are not included in the statistical analysis presented here as they are uninformative in terms of instability-associated chromosomal breaksites and are only included in the current table for comparison. (B) Clones selected for conversion at the TK6 locus [59], via the treatments listed in the second column, final concentration of BPDE was 0.3 ␮M. (C) BCL-XL clones, surviving heavy iron or proton exposure (data was pooled from two separate exposures, 0.63 Gy and 1.89 Gy, for both iron and protons). BCL-XL is a TK6 derivative that expresses the anti-apoptotic BCL-XL protein (Dr. A. Kronenberg). (D) TK6 clones surviving exposure to heavy iron or protons as in group C. (E) HPRT mutant TK6 clones selected after 137 Cs ␥-irradiation. (F) TK6 clones surviving 137 Cs ␥-radiation exposure. (G) TK6E6 clones surviving 137 Cs ␥-radiation exposure. TK6E6 is a TK6 derivative that expresses the HPV 16 E6 oncoprotein, which tags P53 for ubiquitin-mediated proteosomal degradation (gift of Dr. J. Little). (H) NH32 clones surviving 137 Cs ␥-radiation exposure. NH32 is a TK6 derivative wherein both copies of p53 have been inactivated by targeted allelic disruption (gift of Dr. H. Liber). (I and J) These clones are derived from unstable clones that survived either fractionated low-dose (I), or a single dose (J) of 137 Cs. Unstable clones collected after irradiation were plated at 0.1 cells/well in 96 well dishes to collect independent daughter clones. For clarity, these daughter clones are referred to as sub-clones in this table, to distinguish them from the parental clones that were collected after irradiation. b Treatments between groups varied. For projects C and D, half of the cells were exposed to Fe nuclei were exposed to low doses (63 cCy) and half were exposed to high doses (189 cGy). Identical low and high doses were used for exposure to protons. Group E received a graded dose of irradiation, 1 exposure per day over 15 days, to give a total dose of 200 cGy. 137 Cs

regression and linear correlation analysis on both data sets independently. 2.4. Statistical analysis of instability-associated chromosomal rearrangement breaksites Chromosomal size estimates were based on Morton [49]. For all statistical tests, a two-sided p ≤ 0.05 was considered significant. 2.4.1. Overall distribution Instability associated chromosomal rearrangements across all cell lines and treatments were compared using the χ2 analysis. The TK6 karyotype is trisomic for chromosome 13 and hyperdiploid for chromosomes 3 and 20 in ref. [9], which was accounted for in the statistical analysis. Based on the general karyotypic resolution in our lab, we used 391 band resolution in all statistical analyses (ISCN).

If instability-associated breaks are distributed randomly between whole chromosomes or chromosome bands, then the number of breaks in each site will follow a multinomial distribution with class probabilities proportional to chromosome length or band length. A test for an overall departure from randomness was obtained by comparing observed and expected class frequencies (Pearson χ2 ). Because of the generally small numbers of breaks at each site, exact significance levels were estimated by Monte Carlo simulation [50]. Using pseudo-random numbers from a random number generator, the observed total number of breaks was distributed randomly between the sites. This operation was repeated 100,000 times, and the Pearson χ2 statistic was calculated for each distribution. The proportion of occasions on which the observed Pearson χ2 statistic was equaled or exceeded by the calculated value then provided an empirical estimate of exact significance. If the overall departure from randomness is significant, then it is of interest to identify the sites (if any) which are chiefly responsible for this.

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

Hotspots and coldspots (whole chromosomes or bands receiving significantly greater or fewer breaks than expected on the null hypothesis of randomness) were identified by the binomial test [51]. Because of the multiplicity of comparisons, the overall risk of false positives exceeds 0.05, so we used a nominal significance level of considerably less than 0.05, which was estimated by Monte Carlo simulation [50]. A site was identified as a hotspot or coldspot only if the result of the binomial test for that particular site was significant at this reduced significance level. 2.4.2. Clonal analysis In this analysis, unstable clones were each considered independently. Clones with two or more breaks at a particular band were investigated further for evidence of breakpoint clustering, which was defined as two or more independent breaks at a particular chromosomal band within a particular clone. The breaks were considered independent if they were distinct by chromosomal analysis. In each clone, we determined the probability of observing the clustering event in that particular clone, and all less likely clustering events, given the total breaks in that clone. The multinomial distribution was used to evaluate the probabilities of the events concerned. Thus, in a clone with r breaks distributed randomly between N bands all at equal risk of receiving a breaks, the probability of getting r0 bands with no break, r1 particular bands with 1 break, r2 particular bands with 2 breaks, etc. (r1 + r2 + r3 + ··· = r) is r! (0!)r0 (1!)r1 (2!)r2 ...


r 1 N

For example, in a clone with 4 breaks total, the probability that 2 or more of these breaks will occur in a single band is the sum of the probabilities of event A (below) and all less likely probabilities (events B–D): Event A 4 total breaks; 1 band with 2 breaks; 2 bands with 1 break; 388 bands with 0 breaks.The probability of this particular distribution when the breaks occur in particular bands is
4 4! 12 1 = [(0!)388 (1!)2 (2!)1 ] 391 (391)4 The band that receives 2 breaks may be chosen in 391 different ways, after which, the bands that receive 1 break may be chosen in (390/2) different ways. The overall probability of event A is therefore 12 390 × 391 × × 389 = 0.015227731 2 (391)4 The three other possibilities (below) are calculated in the same way, are less likely than event A, and only contribute minimally to the overall probability: Event B

117

4 total breaks; 1 band with 3 breaks; 1 band with 1 break; 389 bands with 0 breaks. 4! [(0!)389 (1!)1 (3!)1 ] =




1 391

4 × 391 × 390

4 × 391 × 390 = 0.000026097 (391)4

Event C 4 total breaks; 1 band with 4 breaks; 390 bands with
0 breaks. 4 4! 1 × 391 = 0.000000017 [(0!)390 (4!)1 ] 391 Event D 4 total breaks; 2 bands with 2 breaks; 389 bands with 0 breaks. 4! 389 [(0!) (2!)2 ] =




1 391

4




×

391 2



6 391 × 390 × = 0.000019573 2 (391)4

The sum of the probabilities of events A–D (0.0153) is the overall probability that 2 or more breaks will occur at a single chromosomal band, given 4 breaks in total, and assuming 391 bands. In the collection of 118 unstable clones, 70 had only 1 break and were thus uninformative with respect to IACRB clustering at a band or generating expectations of clustering. In the 48 that had two or more breaks, including those 14 with IACRB clustering at a band, we determined the likelihood of observing 14 clones with IACRB clustering if the distribution of breaks between bands were random. Using a Monte Carlo simulation, the observed number of breaks in each clone was distributed amongst the 391 bands at random, and the number of times two or more breaks occurred at the same band was counted. This simulation was repeated 1,000,000 times.

3. Results One of the most completely characterized manifestations of genomic instability, delayed chromosomal aberrations [1], are expressed in a heterogeneous manner in individual clones surviving radiation exposure [9,10,15,38,52], suggesting ongoing genomic destabilization. The current results represent the karyotypic analysis of over 500 independent TK6 and TK6derivative clones (Table 1), 118 of which were classified as unstable (Table 1). The minimum threshold for classification of a clone as unstable was the demon-

118

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

stration of at least one non-clonal structural chromosomal aberration in the metaphases examined for that clone [15]. Structural chromosomal aberrations that we generally observe include chromosome deletions and chromosome breaks as well as balanced (reciprocal) and unbalanced (non-reciprocal) chromosomal translocations which may result in the formation of dicentric or multiply rearranged chromosomes; in the current study, the types of chromosomal aberrations identified was independent of treatment and genetic background (data not shown). Furthermore, the number of chromosomal aberrations in any particular clone varied and was independent of treatment or genetic background (Table 1). There was also variation in the fraction of unstable clones collected between projects (Table 1), which is in line with the heterogeneous expression of genomic instability in the literature [1]. No stable clones were included in the analyses herein and are only included in Table 1 for comparison (Table 1; Project ‘A’). Within the available statistics, there was no evidence that any treatment or genetic background contributed unequally to the pool of unstable clones. All unstable clones, and instability-associated chromosomal rearrangement breakpoints (IACRB), were compiled and used for further statistical analyses.

3.1. Overall distribution analysis To test the hypothesis that IACRB are randomly distributed among the chromosomes, according to chromosome (target) size, we pooled all 296 IACRB and tested the null hypothesis that all chromosomes and all chromosomal bands are equal targets based strictly on size considerations. Overall, when the data were pooled and 296 IACRB were distributed across all chromosomes, the distribution was significantly non-random (p < 0.00001; Fig. 1). Therefore, we rejected the null hypothesis that IACRB are randomly distributed according to target size. A similar overall non-random distribution of chromosomal rearrangements has been observed in human tumor cells [28–33] and in X-irradiated human fibroblasts [36]. Importantly, however, the current study is not an examination of chromosomal rearrangements derived from a population of cells [28,34,36], but rather a compilation of IACRB from independently derived clones. In addition to the overall non-random distribution of IACRB described above, the binomial statistical analysis [51] of the pooled data identified three chromosomes, 11, 12, and 22, as chromosomal hotspots for IACRB (p = 0.003, Fig. 1), and one chromosome, 2, a chromosomal coldspot for IACRB (p = 0.004, Fig. 1). The

Fig. 1. Instability-associated breaks by chromosome. Two hundred ninety-six breaks were compiled from 101 independently derived, unstable TK6 clones, and TK6 derivative clones (see Section 2). Expected breaks were assigned according to chromosome size and concessions were made for the constitutional trisomy (+13) and derivative chromosomes (der(14) t(14;20) (q32;q11.2), and der(21) t(3;21) (q22;p12)) defining the TK6 karyotype (ref. [9]). With 100,000 Monte Carlo simulations of Pearson’s χ2 test, the overall distribution was significantly non-random (p < 0.00001), and three chromosomal hotspots and one coldspot were uncovered, which are depicted by solid black bars in the figure. See text for further discussion of these chromosomes where instability-associated breaks were significantly elevated (hotspots) or reduced (coldspots).

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

overall non-random distribution of IACRB reported here is significantly different from our analysis of 129 breaks in first-division metaphases (p = 0.0004, two tailed Student’s t test; Fig. 2a), which follow a pattern more closely associated with chromosomal target size (Fig. 2b). In our first division analysis, no chromosomal hotspots or coldspots were identified (data not shown). When each of 391 chromosomal bands was considered an equal target, and the likelihood of independent

119

events occurring at any particular band was evaluated using binomial statistics, four chromosomal bands were found to be significant IACRB hotspots, 1q11, 4p16, 11q21, and 12q24.3 (p = 0.000025; Fig. 3). Almost 20% of all breaks occurred at these four hotspot bands (Fig. 3), which represent in sum approximately 1% of the genomic target. With the exception of band 11q21, these sites map to centromeric (1q11) or telomeric (4p16 and 12q24.3) regions. None of these hotspots (1q11,

Fig. 2. First division breaks and instability-associated breaks by chromosome. (a) Chromosomes are plotted by size on the x-axis (cf. Fig. 1, where chromosomes are plotted by number). One hundred twenty-nine first-division breaks were collected; the overall distribution did not differ from the expectation of randomness, nor were any hotspots or coldspots observed. (b) Linear regression analysis of chromosomal breaks as a function of chromosome size. Left panel: Instability associated breaks as a function of chromosome size, although significantly deviating from the null hypothesis of a zero slope, this analysis showed a very weak correlation between number of breaks and chromosome size; however, if the outlier, chromosome 1 which contains a breakage hotspot is removed, the p-value becomes 0.353, R2 = 0.04. Right panel: Breaks in first division metaphases after 4 Gy ␥-irradiation, a very highly significant deviation from the null hypothesis of a zero slope is evident, which remains when chromosome 1 is removed from the analysis.

120

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

Fig. 3. Fraction of total instability-associated breaks that mapped to hotspot bands. Almost 20% of the 296 breaks were found at 4 chromosomal bands, 1q11, 4p16, 11q21, and 12q24.

4p16, 11q21, and 12q24.3) occur at known fragile sites [53]. 3.2. Clonal analysis We observed several independent clones in which a particular chromosome, arm or band was the site of two or more independent IACRB, suggesting that chromosomal breaks at some sites may predispose to further breaks at the same site. IACRB clustering was defined as two or more independent breaks on the same chromosome, arm or band in a particular clone. To prevent inclusion of clonally expanded IACRB in the current analysis, we have only included two or more breaks at a particular site if all were identifiable as distinct IACRB by karyotypic analysis, which is similar to, but more stringent than ref. [54]. To assess whole chromosome or arm IACRB clustering, we excluded band-level IACRB clustering. There was no evidence for chromosomal level clustering and only two arms, 11q and 12q, exhibited clustering that was highly significant by multinomial statistics (p = 0.032 and 0.0106, respectively). Of the 118 unstable clones (Table 1), 14 clones demonstrated clustering at an individual chromosomal band, and 1/14 displayed two distinct clustering events, at different chromosomes (Table 2). The IACRB breakpoint clustering in all of the 14 clones, taken individually, was extremely unlikely to have occurred by chance in all cases (Table 2; p < 0.01).

The IACRB in 11/14 clones were at centromeric or telomeric regions (Table 2), a large fraction of which (4/14 clones) occurred at centromeric regions of acrocentric chromosomes (13–15; Table 2). At the level of resolution of the current analysis it is difficult to estimate the exact contribution of centromere, telomere, or satellite DNA to the clustering in these four clones. The two IACRB focal centers identified in Section 3.1 and discussed above (11q21 and 12q24.3) were also IACRB cluster hotspots. Additionally, we observed one clone (HX316 Sc2, Table 2) where two band-level IACRB clustering events were observed (12q24.3 and 16q11.2). Overall these data suggest that such sites might be prone to further breakage once broken or rearranged [14]. None of these band-level IACRB cluster hotspots occurred at known fragile sites [53]. We extended the multinomial approach taken above to determine whether the number of clones from our pool of 118 that demonstrated band-level clustering was unlikely to have occurred by chance. Only the clones that were informative with respect to clustering (those with two or more band-level IACRB) were included. Breaks were distributed randomly amongst the 391 bands in each clone in 1,000,000 Monte Carlo simulations for each clone. There were no occasions in which 14 or more clones demonstrated clustering at a band, indicating that if breakage were random, the chances of obtaining two or more breaks at a particular band in as many as 14 of 48 clones is less than 1 in 1,000,000. This observation lends considerable support to our hypothesis

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

121

Table 2 Instability-associated breakpoint clustering at chromosome bands Clone designation

Chromosome with cluster band

Chromosome band with cluster

Total independent breaks at this site in this clone

Total breaks in this clone

Multinomial probabilitya

TK6 13Bpar. TK6 Hx316 par. TK6 CB11 TK6.43 TK6 CB22 TK6 B31 NH32.15 NH32.12 TK6 HX316 Sc2c TK6 H33B par TK6. P32 TK6. P3 TK6. P6 TK6 HX316 Sc2c TK6 HX329 par.

1 3 6 7 11 12 12 12 12 13 14 15 15 16 X

q43 p10 q22 p22 q21 q24.3 p13 q24.3 q24.3 p11.2 q10 q10 q26 q11.2 q25

2 2 2 3 21 2 2 6 4 2 2 4 3 2 2

2 4 5 4 30 4 4 8 6 3 8 5 4 6 2

2.43E−03 1.59E−02 2.58E−02 4.00E−05 <0.00001 1.50E−02 1.60E−02 <0.00001 <0.00001b 7.51E−03 7.07E−02 <0.00001 2.00E−05 <0.00001b 2.43E−03

a b c

The multinomial distribution is an approximation from 100,000 Monte Carlo simulations. This is the combined probability, since both clustering events occurred in the same clone. Clone TK6 HX316 Sc2 has two breakpoint clusters, one at 12q24.3 and one at 16q11.2.

that band-level IACRB are not independent of previous breaks. 4. Discussion In this report we tested the hypothesis that genomic instability may be partially due to the generation of breakage prone sites or regions at chromosomal rearrangement junctions. In our composite karyotypic analysis, we demonstrated that instability-associated chromosomal rearrangement breaksites (IACRB) are non-randomly distributed amongst the chromosomes in the karyotype (Fig. 1), and identified several chromosomes (Fig. 1) and chromosomal bands (Fig. 3) as hotspots for IACRB. The distribution of our collection of instability-associated breaks differs significantly from the expectation of random breaks according to size (Fig. 1) and our collection of first division radiationinduced breaks (Fig. 2). Furthermore, in our analysis of independent clones we observed a highly significant induction of IACRB clustering at chromosomal bands in 12% of the clones (Table 2). IACRB and IACRB associated with clustering within clones were observed independent of radiation quality, dose, or cellular p53 status (Table 2). For instance, the IACRB at 12q24.3 was observed in cells treated with several doses of ␥-rays, protons, Fe nuclei and BPDE; additionally, this IACRB was involved in IACRB clustering in p53 wild-type and P53 null TK6 clones (Table 2, and ref. [9]).

It is possible that the chromosomal rearrangements that we report here as IACRB are simply the most compatible with cellular growth, thus facilitating their collection and observation. According to this interpretation, we would anticipate that at least some of the IACRB reported here would be observed as clonal alterations in stable clones or in the background stable karyotype of unstable clones. Additionally, we would not expect persistent breakage and rearrangements at individual IACRB, as growth advantage should be a relatively stable phenotypic change. Alternatively, some IACRB may be capable of initiating or driving genomic instability in individual clones in cis; that is, the IACRB themselves are unstable. A prediction of this model is that within individual clones, these IACRB sites should serve as clusters for persistent chromosomal breakage and rearrangement. The strongest support for this model in the current study comes from the IACRB clustering within individual clones (Table 2; Fig. 3), but recent reports from our group are also compatible with this model of cis-driven genomic instability [9,10]. Four IACRB focal regions emerged from our analysis, 1q11, 4p16, 11q21, and 12q24.3 (Fig. 3), none of which mapped to known fragile sites. The chromosomal region 1q11 contains a large block of paracentromeric heterochromatin and has been shown to be sensitive to localized decondensation and induction of genomic instability after exposure to the demethylating chemical 2,6-diaminopurine (DAP) [14]. Global hypomethy-

122

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

lation, which is observed during tumorigenesis, is also associated with chromosomal instability [25,26]. We did not observe any instances of chromosomal decondensation in any of the metaphases examined in the current study, at 1q11 or on any other chromosomes. However, all of the IACRB at this site were whole-arm chromosomal deletions, similar to the breakage at this chromosome previously reported [14], and similar to 1q whole-arm deletions described in some human malignancies [55]. It is possible that these deletion events are associated with fragility consequent to decondensation at this site, which remains to be tested. The IACRB on chromosome 4p16 were generally non-reciprocal terminal additions of material unidentifiable by karyotypic examination. Importantly, in most cases the background karyotype of metaphases containing this IACRB was unchanged, suggesting that the additional chromosomal material was a partially trisomic or hyperdiploid region from another chromosome. The IACRB at chromosome 11q21 (Fig. 3) were generally additions of unknown material with extensive size heterogeneity within and between clones. The MRE11 locus has been mapped to 11q21 [56], and is mutated in patients with the chromosomal instability disorder Nijmegen Breakage Syndrome, which makes it tempting to speculate that disruption of this locus via translocation might lead to an instability phenotype. However, this IACRB was never observed as a clonal, or stable, alteration, and since disruption of MRE11 is expected to confer a selective advantage [57], it is unlikely that disruption of MRE11 is a major contributing factor to instability in this case; however, we have not determined the status of the MRE11 locus in any of the clones described in this study. We have recently used multicolour FISH to completely characterize one of the clones demonstrating frequent IACRB at this site (CB22, Table 2). The rearrangement partner for the derivative 11 chromosome in this clone was hyperdiploid material from either chromosome 3 or 14 [58]. Chromosome 3 is hyperdiploid in the parental TK6 karyotype [9], and chromosome 14 is involved in an unbalanced rearrangement with a hyperdiploid segment of chromosome 20 [9]. The clone HX316 Sc2 and the associated IACRB hotspot at chromosome 12q24.3 (Fig. 3) were described in detail in Pongsaensook et al. [10], who reported an overall disproportionate involvement of centromeric and hyperdiploid chromosomal regions, including 12q24.3, in IACRB and suggested that these regions may be prone to further rearrangements. The authors [10] suggest that the propagation of rearrangements involving hyperdiploid segments may be due to relaxed copy number constraints, which would be expected to limit such extensive and sequential rearrangements in

diploid regions. Our data support this interpretation as three of the four IACRB focal centers described in the current study (4p16, 11q21, and 12q24.3, Fig. 3) were associated with hyperdiploid segments in a large fraction of metaphases in which they were observed. The comprehensive karyotypic and statistical evaluation of delayed chromosomal rearrangement breaksites in independent clones described in the current study supports a major role for cis mechanisms in radiationinduced genomic instability. The persistent destabilization at particular chromosomal bands identified in the current study may have some predictive clinical value. Acknowledgements The authors thank C.F. Gibbons, L.E. Ritter and K.K. Parks for analysis of some of the clones presented here. We also thank Dr. M.A. Kadhim for helpful discussions. This work was supported by a grant to AJG (RO1CA75129) from the National Institutes of Health. References [1] W.F. Morgan, Non-targeted and delayed effects of exposure to ionizing radiation: I. Radiation-induced genomic instability and bystander effects in vitro, Radiat. Res. 159 (2003) 567–580. [2] W.F. Morgan, J.P. Day, M.I. Kaplan, E.M. McGhee, C.L. Limoli, Genomic instability induced by ionizing radiation, Radiat. Res. 146 (1996) 247–258. [3] J.B. Little, Radiation carcinogenesis, Carcinogenesis 21 (2000) 397–404. [4] C.L. Limoli, B. Ponnaiya, J.J. Corcoran, E. Giedzinski, M.I. Kaplan, A. Hartmann, W.F. Morgan, Genomic instability induced by high and low LET ionizing radiation, Adv. Space Res. 25 (2000) 2107–2117. [5] J.B. Little, H. Nagasawa, W.K. Dahlberg, M.Z. Zdzienicka, S. Burma, D.J. Chen, Differing responses of Nijmejen breakage syndrome and ataxia telangiectasia cells to ionizing radiation, Radiat. Res. 158 (2002) 319–326. [6] S. Dhar, J.A. Squire, M.P. Hande, R.J. Wellinger, T.K. Pandita, Inactivation of 14-3-3 sigma influences telomere behavior and ionizing radiation-induced chromosomal instability, Mol. Cell. Biol. 20 (2000) 7764–7772. [7] S.M. Bailey, J. Meyne, D.J. Chen, A. Kurimasa, G.C. Li, B.E. Lehnert, E.H. Goodwin, DNA double-strand break repair proteins are required to cap the ends of mammalian chromosomes, Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 14899–14904. [8] M.L. Smith, A.J. Fornace Jr., Genomic instability and the role of p53 mutations in cancer cells, Curr. Opin. Oncol. 7 (1995) 69–75. [9] S.R. Moore, L.E. Ritter, C.F. Gibbons, A.J. Grosovsky, Spontaneous and radiation-induced genomic instability in human cell lines differing in cellular p53 status, Radiat. Res. 164 (2005) 357–368. [10] P. Pongsaensook, L.E. Ritter, K.K. Parks, A.J. Grosovsky, Cisacting transmission of genomic instability, Mutat. Res. 568 (2004) 49–68.

S.R. Moore et al. / Mutation Research 600 (2006) 113–124 [11] T. Suzuki, M. Fujii, D. Ayusawa, Demethylation of classical satellite 2 and 3 DNA with chromosomal instability in human fibroblasts, Exp. Gerontol. 37 (2002) 1005–1014. [12] L. Smith, A. Plug, M. Thayer, Delayed replication timing leads to delayed mitotic chromosome condensation and chromosomal instability of chromosome translocations, Proc. Natl. Acad. Sci. U.S.A. 98 (2001) 13300–13305. [13] B. Ponnaiya, C.L. Limoli, J. Corcoran, M.I. Kaplan, A. Hartmann, W.F. Morgan, The evolution of chromosomal instability in Chinese hamster cells: a changing picture? Int. J. Radiat. Biol. 74 (1998) 765–770. [14] L.E. Smith, K.K. Parks, L.S. Hasegawa, D.A. Eastmond, A.J. Grosovsky, Targeted breakage of paracentromeric heterochromatin induces chromosomal instability, Mutagenesis 13 (1998) 435–443. [15] A.J. Grosovsky, K.K. Parks, C.R. Giver, S.L. Nelson, Clonal analysis of delayed karyotypic abnormalities and gene mutations in radiation-induced genetic instability, Mol. Cell. Biol. 16 (1996) 6252–6262. [16] A.W. Lo, C.N. Sprung, B. Fouladi, M. Pedram, L. Sabatier, M. Ricoul, G.E. Reynolds, J.P. Murnane, Chromosome instability as a result of double-strand breaks near telomeres in mouse embryonic stem cells, Mol. Cell Biol. 22 (2002) 4836–4850. [17] D. Gisselsson, L. Pettersson, M. Hoglund, M. Heidenblad, L. Gorunova, J. Wiegant, F. Mertens, P. Cin, F. Mitelman, N. Mandahl, Chromosomal breakage–fusion–bridge events cause genetic intratumor heterogeneity, Proc. Natl. Acad. Sci. U.S.A. 97 (2000) 5357–5362. [18] D. Gisselsson, T. Jonson, A. Petersen, B. Strombeck, P. Cin, M. Hoglund, F. Mertens, N. Mandahl, Telomere dysfunction triggers extensive DNA fragmentation and evolution of complex chromosome abnormalities in human malignant tumors, Proc. Natl. Acad. Sci. U.S.A. 98 (2001) 12683–12688. [19] C. Ducray, J.P. Pommier, L. Martins, D.F. Boussin, L. Sabatier, Telomere dynamics, end-to-end fusions and telomerase activation during the human fibroblast immortalization process, Oncogene 18 (1999) 4211–4223. [20] E. Kolomietz, M.S. Meyn, A. Pandita, J.A. Squire, The role of Alu repeat clusters as mediators of recurrent chromosomal aberrations in tumors, Genes Chromosomes Cancer 35 (2002) 97–112. [21] B.K. Singleton, C.S. Griffin, J. Thacker, Clustered DNA damage leads to complex genetic changes in irradiated human cells, Cancer Res. 62 (2002) 6263–6269. [22] I. Matzner, I. Savelyeva, M. Schwab, Preferential integration of a transfected marker gene into spontaneously expressed fragile sites of a breast cancer cell line, Cancer Lett. 189 (2003) 207–219. [23] H.M. Padilla-Nash, K. Heselmeyer-Haddad, D. Wangsa, H. Zhang, M.B. Ghadimi, M. Macville, M. Augustus, E. Schrock, E. Hilgenfeld, T. Ried, Jumping translocations are common in solid tumor cell lines and result in recurrent fusions of whole chromosome arms, Genes Chromosomes Cancer 30 (2001) 349–363. [24] D.F. Callen, H. Eyre, S. McDonnell, S. Schuffenhauer, K. Bhalla, A complex rearrangement involving simultaneous translocation and inversion is associated with a change in chromatin compaction, Chromosoma 111 (2002) 170–175. [25] A. Eden, F. Gaudet, A. Waghmare, R. Jaenisch, Chromosomal instability and tumors promoted by DNA hypomethylation, Science 300 (2003) 455. [26] F. Gaudet, J.G. Hodgson, A. Eden, L. Jackson-Grusby, J. Dausman, J.W. Gray, H. Leonhardt, R. Jaenisch, Induction of tumors in mice by genomic hypomethylation, Science 300 (2003) 489– 492.

123

[27] G. Gumus, A. Sunguroglu, A. Tukun, B.D. Sayin, I. Bokesoy, Common fragile sites associated with the breakpoints of chromosomal aberrations in hematologic neoplasms, Cancer Genet. Cytogenet. 133 (2002) 168–171. [28] M. Hoglund, D. Gisselsson, T. Sall, F. Mitelman, Coping with complexity. Multivariate analysis of tumor karyotypes, Cancer Genet. Cytogenet. 135 (2002) 103–109. [29] T. Nilsson, M. Hoglund, S. Lenhoff, L. Rylander, I. Turesson, J. Westin, F. Mitelman, B. Johansson, A pooled analysis of karyotypic patterns, breakpoints and imbalances in 783 cytogenetically abnormal multiple myelomas reveals frequently involved chromosome segments as well as significant age- and sex-related differences, Br. J. Haematol. 120 (2003) 960–969. [30] Y. Jin, C. Martins, L. Salemark, B. Persson, C. Jin, J. Miranda, I. Fonseca, N. Jonsson, Nonrandom karyotypic features in basal cell carcinomas of the skin, Cancer Genet. Cytogenet. 131 (2001) 109–119. [31] C. Jin, Y. Jin, J. Wennerberg, M. Dictor, F. Mertens, Non-random pattern of cytogenetic abnormalities in squamous cell carcinoma of the larynx, Genes Chromosomes Cancer 28 (2000) 66–76. [32] J.R. Teyssier, D. Ferre, Identification of a clustering of chromosomal breakpoints in the analysis of 203 human primary solid tumor non specific karyotypic rearrangements, Anticancer Res. 12 (1992) 997–1004. [33] F. Mitelman, S. Heim, Consistent involvement of only 71 of the 329 chromosomal bands of the human genome in primary neoplasia-associated rearrangements, Cancer Res. 48 (1988) 7115–7119. [34] M. Hogland, D. Gisselsson, N. Mandahl, B. Johansson, F. Mertens, F. Mitelman, T. Sall, Multivariate analyses of genomic imbalances in solid tumors reveal distinct and converging pathways of karyotypic evolution, Genes Chromosomes Cancer 31 (2001) 156–171. [35] M. Hogland, D. Gisselsson, G.B. Hansen, T. Sall, F. Mitelman, M. Nilbert, Dissecting karyotypic patterns in colorectal tumors: two distinct but overlapping pathways in the adenoma-carcinoma transition, Cancer Res. 62 (2002) 5939–5946. [36] Y. Kano, J.B. Little, Site-specific chromosomal rearrangements induced in human diploid cells by x-irradiation, Cytogenet. Cell Genet. 41 (1986) 22–29. [37] K. Holmberg, A.E. Meijer, G. Auer, B.O. Lambert, Delayed chromosomal instability in human T-lymphocyte clones exposed to ionizing radiation, Int. J. Radiat. Biol. 68 (1995) 245–255. [38] K. Holmberg, S. Falt, A. Johansson, B. Lambert, Clonal chromosome aberrations and genomic instability in X-irradiated human T-lymphocyte cultures, Mutat. Res. 286 (1993) 321–330. [39] M.A. Hill, D.L. Stevens, S.J. Marsden, R. Allot, I.C. Turcu, D.T. Goodhead, Is the increased relative biological effectiveness of high LET particles due to spatial or temporal effects? Characterization and OER in V79-4 cells, Phys. Med. Biol. 47 (2002) 3543–3555. [40] D.T. Goodhead, H. Nikjoo, Track structure analysis of ultrasoft X-rays compared to high- and low-LET radiations, Int. J. Radiat. Biol. 55 (1989) 513–529. [41] M.S. Kovacs, J.W. Evans, I.M. Johnstone, J.M. Brown, Radiationinduced damage, repair and exchange formation in different chromosomes of human fibroblasts determined by fluorescence in situ hybridization, Radiat. Res. 137 (1994) 34–43. [42] S. Puerto, J. Surrall´es, M.J. Ram´ırez, A. Creus, R. Marcos, Equal induction and persistence of chromosome aberrations involving chromosomes with heterogeneous lengths and gene densities, Cytogenet. Cell Genet. 87 (1999) 62–68.

124

S.R. Moore et al. / Mutation Research 600 (2006) 113–124

[43] R.M. Anderson, D.L. Stevens, D.T. Goodhead, M-FISH analysis shows that complex chromosome aberrations induced by alphaparticle tracks are cumulative products of localized rearrangements, Proc. Natl. Acad. Sci. U.S.A. 99 (2002) 12167–12172. [44] Y. Yu, C.Y. Li, J.B. Little, Abrogation of p53 function by HPV 16 E6 gene delays apoptosis and enhances mutagenesis but does not alter radiosensitivity in TK6 human lymphoblast cells, Oncogene 14 (1997) 1661–1667. [45] Y.Y. Chuang, Q. Chen, H. Liber, Radiation-induced mutations at the autosomal thymidine kinase locus are not elevated in p53-null cells, Cancer Res. 59 (1999) 3073–3076. [46] C. Cherbonnel-Lasserre, S. Gauny, A. Kronenberg, Suppression of apoptosis by Bcl-2 or Bcl-xL promotes susceptibility to mutagenesis, Oncogene 13 (1996) 1489–1497. [47] R. Jordan, J.L. Schwartz, Noninvolvement of the X chromosome in radiation-induced chromosome translocations in the human lymphoblastoid cell line TK6, Radiat. Res. 137 (1994) 290–294. [48] J.L. Schwartz, R. Jordan, B.A. Sedita, M.J. Swenningson, J.P. Banath, P.L. Olive, Different sensitivity to cell killing and chromosome mutation induction by gamma rays in two human lymphoblastoid cell lines derived from a single donor: possible role of apoptosis, Mutagenesis 10 (1995) 227–233. [49] N.E. Morton, Parameters of the human genome, Proc. Natl. Acad. Sci. U.S.A. 88 (1991) 7474–7476. [50] R.E. Tarone, Testing for non-randomness of events in sparse data situations, Ann. Hum. Genet. 53 (1989) 381–387.

[51] M. De Braekeleer, B. Smith, Two methods for measuring the nonrandomness of chromosome abnormalities, Ann. Hum. Genet. 52 (1988) 63–67. [52] M.A. Kadhim, D.A. Macdonald, D.T. Goodhead, S.A. Lorimore, S.J. Marsden, E.G. Wright, Transmission of chromosomal instability after plutonium alpha-particle irradiation, Nature 355 (1992) 738–740. [53] N.C. Popescu, Genetic alterations in cancer as a result of breakage at fragile sites, Cancer Lett. 192 (2003) 1–17. [54] K.L. Johnson, J. Nath, D.J. Brenner, C.R. Geard, J.D. Tucker, Chromosome aberrations of clonal origin in irradiated and unexposed individuals: Assessment and implications, Radiat. Res. 152 (1999) 1–5. [55] F. Mitelman, Recurrent chromosome aberrations in cancer, Mutat. Res. 462 (2000) 247–253. [56] J.H. Petrini, M.E. Walsh, C. DiMare, X.N. Chen, J.J. Korenberg, D.T. Weaver, Isolation and characterization of the human MRE11 homologue, Genomics 29 (1995) 80–86. [57] L.H. Thompson, D. Schild, Recombinational DNA repair and human disease, Mutat. Res. 509 (2002) 49–78. [58] S.R. Moore, M.A. Kadhim, V. Buckle, A.J. Grosovsky, Trisomic chromosomal regions drive karyotypic instability in a single unstable TK6 clone, submitted for publication. [59] P.J. Quintana, E.A. Neuwirth, A.J. Grosovsky, Interchromosomal gene conversion at an endogenous human cell locus, Genetics 158 (2001) 757–767.