The origin of B cell recurrent chromosomal translocations: proximity versus DNA damage.

摘要: To test the contribution of these parameters Alt and Nussenzweig laboratories developed genome-wide techniques to map chromosomal rearrangements (Chiarle et al., 2011; Klein 2011), Casellas laboratory a method measure AID-mediated damage (Hakim 2012; Yamane 2013). Genomic interactions were computed either by 4C or Hi-C. These studies uncovered two kinds translocations: AID RAG dependent independent. Translocations that occur in absence recurrent DNA (e.g. AID−/−) are widespread, join interacting loci epigenetically accessible. The location frequency events correlate with nuclear 2012). Furthermore, because randomly broadly distributed across genome (Figure S1A), they cannot be subtracted from sample sample. In presence AID, ~90% translocations trans recapitulate those obtained AID−/−, both their distribution frequency. contrast, remaining 10% AID-dependent, i.e. can between samples reproducibly accumulate near transcription start sites (TSSs) Ig subset non-Ig genes S1A, 2012)). Unlike AID-independent events, at hotspots does not target proximity but amount inflicted measured RPA Rad51 accumulation during DNA-end resection homologous recombination 5A 2012) Figure 3C (Yamane 2013), Spearman’s ρ > 0.6). ideas independently confirmed germinal center cells 3D FISH (Gramlich contrast studies, Rocha al. reanalyzed our published TC-Seq only AID+/+ concluded correlates (Rocha discrepancy is explained application different statistical analyses as claim authors’ failure include essential controls selective data analysis. Rocha failed analyze AID−/− translocation profiles; consequently, could distinguish AID-dependent rearrangements. This serious error stated above small fraction (~10%) outside Igh recurrent, overlap hypermutation hotspots, thus credited (Klein 2011). Consequently, “genome-wide correlation” reflects profiles Hakim S1B illustrates this point comparing Myc cis AID. cases density decreases increasing distance engineered I-SceI site, consistent folding preferences mammalian chromosomes (Lieberman-Aiden 2009). Conversely, same chromosome predicted based on breakpoint S1B). Ascribing non-transcribed (including intergenic domains, e.g. S1A) activity, imply, goes against well-established mechanism deamination. Rocha interpretation differs previous calculated using genomic windows centered TSSs while we used fixed non-overlapping windows. factual inaccuracy. Our analysis vs. 5 2012)) makes use 200kb islands, which turn targets. Regardless, argument invalid direct comparison TSS-centered shows highly correlated results S1C). so long range interchromosomal large domains rather than individual genes, resolution experiments no better (Simonis 2007). regard, selectively 20kb window one (IL4rα) an attempt improve correlation translocations. However, found such improvement when all targets included S1D). Instead, lack 4C-Seq reads altogether due size S1D), confirmation approach incompatible 4C. relatively infrequent, 200Kb some instances translocations. Rocha study further confounded separate treatment trans. For example, chr.12 plot values linear scale set arbitrary cut off 60Mb Igh. They indicate within window, determines translocations, feature explains beyond 60Mb. however contradicts main conclusion criterion applied trans, essentially excluded S1E). side-by-side dismisses alleged hotpots. instance, Gpr132, Klhdc2, Satb1 located 0.4Mb, 44Mb, relative Despite fact contact Klhdc2 10-fold 10,000-fold lower three translocate comparable frequencies S1E). As discussed 2012), notice AID-targets tend interact more frequently would expected random model interpret observation evidence close predisposes damage. over minority bona fide nearly 1,500 outrank interaction frequency, few translocated manner fail consider tendency comingle most transcribed S1F), Igh, other constitutively active preferentially interacts euchromatin S1G). In summary, do contend interactions, since must rearrange. Rather, clearly demonstrate Myc) neither determined nor phenomenon likely limits incidence (Robbiani