|Home | About | Journals | Submit | Contact Us | Français|
Reciprocal chromosomal translocations are early and essential events in the malignant transformation of several tumor types, yet the precise mechanisms that mediate translocation formation are poorly understood. We review here the development of approaches to induce and recover translocations between two targeted DNA double-strand breaks (DSBs) in mammalian chromosomes. Using mouse cells, we find that nonhomologous end-joining readily mediates translocation formation between two DSBs generated by site-specific endonucleases. Translocations occur much less frequently, however, than intrachromosomal repair of a single DSB. Translocation junctions obtained with this approach have similar end modifications to translocation junctions found in human tumors, including deletions, insertions, and repair at short stretches of homology. These modifications are more extensive than repair junctions at a single DSB, suggesting that different factors may be involved in translocation formation and repair of a single DSB. Finally, we describe a novel approach to induce translocations in human cells. Translocation model systems provide an opportunity to study the involvement of mammalian DNA repair and signaling factors in the etiology of chromosomal rearrangements.
Many hematologic and mesenchymal malignancies, as well as some solid tumors, harbor recurrent reciprocal chromosomal translocations (1–3). Reciprocal translocations require the breakage and reunion of DNA in nonparental chromosome configurations, suggesting that misrepair of DNA double-strand breaks (DSBs) is required for translocation formation. Pathways for the faithful repair of DSBs have been characterized in mammalian cells and are broadly defined as nonhomologous end-joining (NHEJ) and homologous recombination (HR) (4,5). In NHEJ, DNA ends are rejoined after processing without the use of extensive sequence homology (6,7). Although NHEJ plays an important role in maintaining the overall integrity of chromosomes, it is potentially mutagenic at the DSB site because DNA ends may undergo modifications before ligation. Several NHEJ factors have been identified, although cells deficient for these factors are still capable of orchestrating NHEJ through alternate, as yet poorly defined, pathways, and often show more extensive end modifications than wild-type cells (8). HR differs from NHEJ in that it utilizes a homologous sequence, typically the sister chromatid, to template repair, leading to precise repair at the DSB site (9). A third DSB repair pathway has been identified that specifically involves sequence repeats, termed single-strand annealing (SSA) (10). In SSA, single strands that are formed at repeats after DNA-end processing anneal to each other, leading to a deletion of one copy of the repeat, as well as the sequence between the repeats.
The repair factors that mediate oncogenic chromosomal rearrangements have not been elucidated, although an NHEJ pathway is implicated because breakpoint junctions almost invariably occur at sites of little or no sequence homology. To examine the mechanisms of chromosomal translocation formation in mammalian cells, we established a murine model system for inducing and recovering translocations between nonhomologous sequences (11). This approach has led to several conclusions: 1) some type of NHEJ is highly proficient at mediating translocation formation in mouse cells; 2) the translocation junctions recapitulate repair characteristics from oncogenic rearrangements in human leukemias, lymphomas, and sarcomas; and 3) translocation junctions contain more extensive-end modifications than repair of a single DSB. Moreover, homology-based mechanisms are rarely used to generate translocations between diverged repetitive elements found near DSBs (12). In this report, we briefly review the murine system and describe a novel human cell system for inducing chromosomal translocations. A more extensive review of the murine translocation approach was previously published (13).
To establish a model for oncogenic translocations, we developed an approach in which DSBs are introduced into nonhomologous sequences on murine chromosomes 14 and 17 (Figure 1) (11), using modifications of our earlier models (12,14). DSBs occur in these chromosomes through the expression of the rare-cutting endonuclease I-SceI, which efficiently cleaves an 18-bp recognition sequence that has been introduced into mammalian cells (15,16). (Mis)repair of the two DSBs by NHEJ involving the two chromosomes leads to a translocation that can reconstruct a functional gene that confers resistance to a neomycin analogue, allowing for selection.
More precisely, the translocation selection is based on a neo gene that has been split by an intron to form neoSD (containing a splice donor) and SAneo (containing a splice acceptor) (Figure 1) (12). The neoSD and SAneo gene fragments each contain an I-SceI–recognitio1n site that defines the border of their intronic sequences. The gene fragments are targeted to loci on chr.17 and chr.14, respectively, in mouse embryonic stem (ES) cells, given the ease of targeting these cells (17,18). Cleavage by I-SceI, followed by NHEJ to join neoSD and SAneo, results in a translocation that forms a neo+ gene on der17 (Figure 1). Because the DSBs occur within intronic sequences, a variety of breakpoint junctions are recoverable with neo+ selection, although the maximal deletion that is allowable to recover a neo+ gene is the size of the intron, which here is 2.7 kb. Formation of der(14) is not under constraint.
To induce translocations, I-SceI is transiently expressed in ES cells containing the targeted gene fragments and then the cells are plated. After 48 hours the cells are exposed to the neomycin analogue G418. Within 6 days individual, G418-resistant clones are visible as colonies; these colonies are isolated and expanded. Translocations are evaluated by polymerase chain reaction (PCR), sequencing, Southern blot, and/or fluorescence in situ hybridization (FISH) with chromosome-specific paints. The frequency of translocation formation is calculated as the number of G418-resistant colonies divided by the number of transfected cells.
After expression of I-SceI, neo+ clones were recovered at a frequency of ~3 × 10−5 (11). Recombinants were confirmed to contain the reciprocal t(14;17) using FISH. The high frequency of translocations suggests that endonuclease expression is particularly favorable for generating concurrent DSBs. I-SceI cleaves mammalian chromosomes very efficiently (16). Both sister chromatids may be cleaved by I-SceI concurrently, precluding repair by HR between sister chromatids and permitting DNA ends to participate in other repair events. Moreover, if a single DSB is repaired by precise ligation, the I-SceI site is restored, offering the potential for multiple cycles of cleavage that could result in translocation formation. The frequency of precise ligation after I-SceI expression is not quantifiable in these experiments; only events that disrupt the I-SceI site can be scored.
Although extensive sequence homology is very infrequent at translocation junctions, repetitive sequences are commonly present within the vicinity of putative chromosomal breakpoints (19,20). Repetitive elements comprise at least 45% of the human genome (21). Therefore, the presence of repetitive elements near breakpoints could be purely coincidental. On the other hand, several studies have postulated that homologous repetitive sequences may favor interactions between heterologous chromosomes, increasing the likelihood of translocation formation (20).
To begin to address whether homologous sequences in the vicinity of breakpoints affect translocation formation, we modified the neo introns to include an identical 210-bp repeat on each of the two chromosomes (11). After expression of I-SceI, neo+ recombinants were recovered at similar frequencies whether the repeat was present or not. Thus, the repeat did not affect the recovery of translocations. We also examined the breakpoint junctions by PCR and sequencing. Repair by SSA between the 210-bp repeats results in a 0.9-kb deletion on der(14), whereas NHEJ results in a variety of junctions. Only 35% of the der(14) junctions occurred by NHEJ when the repeat was present. The remaining 65% repaired by SSA, demonstrating the high efficiency of SSA between identical repeats. Interestingly, HR, which would have left both copies of the repeat intact, was not involved in any of the translocations. These results also indicate that mammalian cells can orchestrate different pathways of repair (ie, NHEJ and SSA) on contemporaneous DSBs (11). Yet, oncogenic translocations with breakpoint junctions within homology are exceedingly rare. This seeming paradox is resolved by other experiments in which we have found that sequence divergence between repetitive elements strongly skews translocation formation to NHEJ (12). In these latter experiments, the repetitive elements were two Alu elements with ~20% sequence divergence. Because most repetitive elements in mammalian genomes are diverged from each other, NHEJ is expected to be favored as a repair pathway for translocation formation, even when translocations occur within the vicinity of these elements.
We evaluated 172 breakpoint junctions derived from our recovered translocations (11). Approximately 80% of the NHEJ junctions contained simple deletions, 12% contained insertions, and 2% were complex, the latter containing deletions, duplications, and inversions of nearby sequences, and other insertions. The remaining 6% of junctions were not sequenced or proved difficult to sequence. The majority of deletions were less than 100 bp for both der(17) and der(14). No correlation was observed between deletion lengths on der(17) and der(14) for an individual neo+ clone or between deletion lengths from the centromeric and telomeric ends for an individual derivative chromosome. Most junctions (85%) occurred at one or more base pairs of sequence identity between the chromosomes (ie, microhomology). Insertions were derived by several mechanisms and included nontemplated sequences, duplications, and locally derived inverted sequences. Most insertions (60%) were less than 10 bp, whereas a few were more than 100 bp and were derived from a variety of sources (nearby genomic DNA, mitochondrial DNA, and I-SceI expression vector).
We compared end modifications from the murine translocations to end modifications of parental chromosomes from 142 patients with reciprocal translocations (74 leukemias, 63 lymphomas, and 5 sarcomas) (11). There were no statistically significant differences between the two groups in the percent of chromosomes that underwent deletions or duplications or in the mean deletion length. Microhomology use was common in both patient-derived (64.4%) and murine cell (79.8%) translocations. Duplication of sequences adjacent to the breakpoints is a frequent finding at translocation junctions from human cancer cells. One parameter that distinguished the two groups was the mean duplication length (ie, average length of parental chromosome sequence which was duplicated among the junctions that contain duplications), which was significantly longer in the patient-derived cells (75 ± 119 bp vs 3.1 ± 0.96 bp). The long duplications, which were observed primarily in leukemias, have been proposed to arise from staggered nicks that are distant from each other (more than 100 bp) and which are processed to long single-strand overhangs before duplication (22, 23). In contrast, cleavage by I-SceI results in four nucleotide overhangs.
Although readily obtained with our approach, translocations are estimated to occur more than three orders of magnitude less frequently than simple intrachromosomal repair of a single DSB (24). Translocations may either be a rare outcome of “normal” NHEJ or they may have a distinct etiology. To approach this question, translocation junctions were compared with junctions derived from repair of a single DSB by intrachromosomal NHEJ ( Figure 2) (11). For repair of a single DSB, we used cell lines that had reestablished an I-SceI cleavage site on der17 upon translocation formation (Figure 2, A). Therefore, cleavage of the single I-SceI site on der(17) in these cell lines results in exactly the same end sequences that are found after cleavage of the two I-SceI sites in the parental cell lines before der(17) formation. Cleavage and precise repair of a single I-SceI site results in reestablishment of the site, leaving no genetic evidence that repair occurred, and complicating any calculation of cleavage frequency. Thus, we compared only the single DSB repair and translocation events that resulted in modification of the ends.
Although the spectrum of end modifications overlapped between the two groups of junctions, significant differences were observed (Figure 2, B). For example, only 13% of single DSB repair events had deletions more than 30 bp compared to 52% of translocation junctions. Microhomology of more than 1 bp was present in 35% of the single DSB junctions compared to 74% of translocation junctions. Thus, translocation formation may be different from repair of a single DSB, eg, by involving long-lived DSBs. Alternatively, distinct factors may contribute to translocation formation, such as proteins involved in poorly defined, non-classic NHEJ pathways. We recently demonstrated using the murine system that the classic NHEJ protein Ku70 is not required for, but rather suppresses, translocations (29). We are actively investigating the genetic dependencies of translocation events to better define the role of nonclassic NHEJ pathways.
More recently, we have developed an approach to induce translocations in human cells. For this, we utilized the novel technology of zinc finger nucleases (ZFNs) to cleave an endogenous chromosomal sequence. ZFNs are hybrid proteins comprising a nonspecific nuclease domain derived from the FokI restriction enzyme linked to sequence-specific DNA-binding domains of the zinc finger type (25). Each ZFN is composed of two subunits—ZFN-L and ZFN-R—which dimerize such that each subunit is directed to a pair of binding sites that are in opposite orientation (Figure 3, A and B). Cleavage typically results in a DSB with 5-base 3′ overhangs, although it may also occur within a few base pairs of the expected site (26). In our experiments, we used a ZFN directed toward a site in exon 5 of the IL2Rγ gene on the X chromosome (27). To create a second DSB, we utilized the I-SceI endonuclease, making use of a human embryonic kidney cell line that contains an I-SceI cleavage site randomly integrated in the genome in the context of a GFP gene (TOS4A cells; Figure 3, A) (28). Concurrent cleavage by I-SceI and the ZFN results in two DSBs that have the potential to join to form a translocation.
ZFN-L and ZFN-R were transiently coexpressed with I-SceI in TOS4A cells in 24-well plates (5 × 104 cells per well). After 48 hours, pooled DNA from each well was subjected to nested PCR (Figure 3, C) and the PCR products were sequenced (Figure 3, D). We designed primers to recover PCR products in both orientations. Presumably, reciprocal translocation formation in one orientation results in monocentric derivative chromosomes, whereas translocation formation in the opposite orientation results in an acentric and dicentric pair of derivative chromosomes that would be expected to undergo further rearrangement or eventually be eliminated from the population. No PCR products were recovered after expression of either I-SceI or a single ZFN subunit (Figure 3, C). After expression of both ZFN subunits and I-SceI, however, products consistent with translocation formation were recovered in both orientations at frequencies estimated between 10−4 and 10−5, similar to our murine cell lines. Sequencing of the PCR products identified deletions, insertions, and microhomology usage (Figure 3, C).
We are currently modifying this system to allow for precise quantification of translocation frequency and repair characteristics in primary and immortalized cells. These studies highlight the exciting potential for the use of ZFNs in studies of DSB repair in human cells. As more ZFNs become available, it will be possible to perform translocation studies with ZFNs directed to endogenous sites on two different chromosomes, rather than relying on an introduced I-SceI site for one chromosome.
Translocations can be recovered at high frequencies from mammalian cells after concurrent cleavage of heterologous chromosomes by site-specific endonucleases. Translocation junctions recapitulate many of the features of oncogenic translocations. Translocation formation is several orders of magnitude less frequent than the simple rejoining of ends at a single DSB, and translocation junctions formed by NHEJ have more extensive deletions and micro-homology use than repair of a single DSB, arguing that distinct factors may influence translocation repair. It remains unclear whether these factors represent alternative NHEJ pathways or simply the sequela of long-lived DSBs.
We thank Fyodor Urnov, Michael Holmes, and Sangamo BioSciences for providing the IL2Rγ ZFN. D.M.W. was supported by the Leukemia and Lymphoma Society, Byrne Fund and a Career Award in the Biomedical Sciences from the Burroughs Wellcome Fund. This work was also supported by grants to M.J. from the Lehman Brothers Foundation, NSF (0346354), and the NIH (GM54688).