To develop a model system in which we could induce GCRs, we used the rare restriction enzyme I-SceI, whose 18 bp recognition sequence is not normally present in the human or mouse genome, to produce a single DNA DSB within a mammalian cell, based on the hypothesis that improper repair of these breaks could lead to GCRs. This enzyme has been used in a series of elegant studies to produce specific, non-random GCRs mediated by homologous recombination in mammalian cells. We generated a construct that expressed the Herpes simplex virus type I thymidine kinase (TK) gene under the control of the constitutive EF1a promoter, with the recognition sequence for the I-SceI restriction enzyme placed between the EF1a promoter and the TK gene. This pEF1aTK vector was introduced into the U937 cell line, and verified that expression of the TK gene conferred sensitivity to ganciclovir (GCV). We then carried out a series of experiments that utilized the negative selection provided by the expression of TK. Cells were transfected with an I-SceI expression vector and selected with GCV (to select for cells that had lost TK expression). All 156 of the clones had small deletions and showed evidence consistent with non-homologous end joining (NHEJ), such as microhomology or local sequence inversions. No clones showed evidence of having a GCR. 8 clones showed small DNA segments (50-800 bp) that were derived from distant regions of the genome inserted at the DNA DSB site. Additional experiments recovered over 50 clones with insertions. Most of these insertions were either genic (transcribed) regions or repeat elements (LINE/SINE/LTR). These findings led to the hypothesis that these DNA DSB may have been repaired by "patches" generated from reverse transcribed RNA. To determine whether this form of DNA repair was restricted to experimentally induced DNA DSB, we analyzed whole genome sequence data from two myeloma cell lines, and identified 15 instances of insertions derived from distant genomic regions (which we have termed template sequence insertions, or TSIs). Most of these TSIs were seen in normal individuals as well as the myeloma cell lines, indicating that most TSIs were polymorphic in the human germline. Six examples showed the hallmarks of LINE-1 retrotransposon-mediated insertions, with a 5'-TTTT/A-3'integration site, a target site duplication (TSD), a polyA tract at the insertion site, and a polyadenylation signal. However, the inserted sequence was not a LINE-1 nor a SINE, but instead was a transcribed, genic region that mapped to a distant genomic region. The presence of these L1 ORF1 hallmarks strongly suggests that these TSIs were caused by the LINE-1 integrase and reverse-transcriptase acting on nuclear pre-mRNA. A second class of TSIs showed no preference for a 5'-TTTT/A-3'integration site, no TSD, no poly A tract, but instead showed several bp of microhomology at the insertion junction. This second class is remarkably similar to insertions at I-SceI induced DNA DSB, and we predict that these insertions were also caused by reverse transcription of pre-mRNA creating a "patch" for a spontaneous DNA DSB that occurred in a germ cell. We suspected that these TSIs could be templated by RNA intermediates, as the genomic regions were not lost from their wild-type chromosomal location. We repeated the I-SceI experiment several times, modified by co-transfection of mouse RNA with the I-SceI expression vector. Approximately 10% of the TSIs recovered from these experiments contained mouse sequences, indicating that they were derived from mouse RNA. In addition, a subset of TSIs were due to telomere repeat sequences, lending further support for the hypothesis that RNA could serve as a template for the repair of DNA DSB. It may not be surprising that we were unable to generate GCRs by inducing a single DNA DSB, as other investigators have concluded that two induced breaks are required to produce a chromosomal translocation, and that the frequency of chromosomal translocations induced by a single DNA DSB in mouse embryonic stem (ES) cells is extraordinarily rare We then repeated these experiments using cells that were deficient for H2AX, a histone variant that becomes gamma-phosphorylated in response to DNA DSB, and "coats" the region of the DNA DSB. We inserted the EF1aTK vector into H2AX knockout (KO) cells, and then transfected an I-SceI expression vector. Although the majority of clones resulting from this experiment were vector capture events, in which the DNA DSB had become repaired by transfected plasmid DNA, approximately 15% of the clones had undergone a GCR (balanced translocation or megabase inversion), demonstrating that a GCR can be caused by a single, induced DNA DSB. A manuscript describing these findings is currently being prepared. A number of chromosomal translocation breakpoints in patients with leukemia have been mapped to regions of alternating purine and pyrimidne residues (Pu/Py repeats). To determine if these Pu/Py repeat regions were susceptible to GCR, we cloned an extended (200bp) Pu/Py repeat region between the EF1a promoter and the TK cDNA, and selected GCV resistant clones. There was a 2-fold greater frequency of spontaneous GCV resistance in the Pu/Py clones compared to the F5 clone described above, but we recovered no GCR. However, 19 of the 25 Pu/Py clones evaluated showed spontaneous mutations (single nucleotide replacements or frameshifts) of the Tk gene, compared to only 2 of 20 F5 clones evaluated, suggesting that Pu/Py sequence predisposed the Tk sequence to spontaneous mutation. A manuscript describing these findings is currently in preparation. We are also attempting to move this system from a cell-culture based system to an in vivo system. To do so, we take advantage of the observation that the MLL gene is known to have at least 100 oncogenic partners, and is even weakly oncogenic when fused to a LacZ reporter gene. These findings suggest that MLL fused to many other genes may be oncogenic. We used gene targeting to insert an I-SceI site in the MLL (MLLKI) locus. These mice have been crossed to mice that express the ISceI (vav-I) protein in hematopoietic cells, to test the possibility that a GCR involving MLL will be generated in vivo. We reasoned that even if this is a very rare event, it may be amplified in vivo, as cells that undergo an MLL fusion may be oncogenic. Unfortunately, none of over 40 MLLKI/vav-I mice developed leukemia. Analysis of bone marrow cells from the MLLKI/vav-I mice indicated that the I-SceI enzyme was active, as there were cells with small (1-10 bp) deletions or insertions at the I-SceI site. We reasoned that no leukemias may have been generated because repair of the I-SceI induced breaks is very efficient. We have now begun to cross these transgenes onto an H2AX deficient background, to determine if a more error-prone DNA repair pathway will result in MLL translocations in vivo. Unfortunately MLL and H2AX are closely linked on mouse chromosome 7, and we have been unable to generate MLLKI/H2AX KO mice. We have begun backcrossing these 2 alleles (MLLKI and vav-I) onto a scid background.
|Onozawa, Masahiro; Zhang, Zhenhua; Kim, Yoo Jung et al. (2014) Repair of DNA double-strand breaks by templated nucleotide sequence insertions derived from distant regions of the genome. Proc Natl Acad Sci U S A 111:7729-34|
|Cheng, Yue; Zhang, Zhenhua; Keenan, Bridget et al. (2010) Efficient repair of DNA double-strand breaks in malignant cells with structural instability. Mutat Res 683:115-22|