Pathologic DSBs are arbitrarily defined as DSBs which serve no physiologic purpose and may lead to cell dysfunction.
Random DNA breaks due to ionizing radiation or oxidative free radicals
In many chromosomal rearrangements, the DSBs at one or both genes appear to be located randomly within large regions of many kilobases. Random positioning and the apparent lack of sequence propensity are suggestive of sequence-nonspecific DSB mechanisms such as oxidative free radicals, ionizing radiation, or less commonly, spontaneous DNA backbone hydrolysis.
About half of the natural ionizing radiation of the environment originates from natural heavy metals of the earth, such as uranium, thorium, and even potassium. The other half of the ionizing radiation emanates from cosmic radiation that is not entirely blocked by the atmosphere. In total, about 3 x 108 ionizing radiation particles pass through each of us every hour , producing hydroxyl free radicals from water in their wake. This tract of hydroxyl free radicals causes cluster damage on DNA, thereby breaking both DNA strands.
About 0.1% of the oxygen that we breathe is converted to free radicals . This generates 3 x1022 free radicals per hour within each of us, and these damaging free radicals are distributed across the 1014 cells of the human body. Free radicals cause predominantly single-strand DNA damage, but two nearby such events can result in a DSB.
RAG action at cryptic RSS sites at off-target locations in a sequence-specific manner: V(D)J-type breaks
The RSS heptamer/nonamer consensus sequence is by no means unique to the Ig and TCR loci, and the RAG complex can cut at sites which differ substantially from the 16 bp consensus . The minimal motif for RAG nicking is only CAC. Thus the RAG complex can act at RSS-like non-antigen receptor locus sites, termed cryptic RSS (cRSS). This occurs in many of the rearrangements observed in human T-cell acute lymphoblastic lymphoma [23–25]. In these cases, instead of the RAG complex pairing a 12-RSS with a 23-RSS, a 12-RSS pairs with a 23-cRSS or a 23-RSS pairs with a 12-cRSS. We call these breaks V(D)J-type breaks because they are occurring via the same mechanism as normal V(D)J recombination, regardless of the fact that one of the sites is outside of the usual antigen receptor loci (that is, it is off-target).
RAG action at DNA bubble structures and other regions of heterology in a structure-specific manner
In addition to its sequence-specific mode of cutting, the RAG complex can also nick in a structure-specific manner at sites of transition from dsDNA to ssDNA, such as occurs at the edges of bubble DNA structures or even single-base mismatches [26–28]. Such activity by the RAG complex may have arisen because the RAG complex is accustomed to creating hairpin structures, which involves substantial DNA distortion. Hence, any region of mismatch or slippage is a potential target for nicking by the RAG complex in lymphoid cells.
RAG-mediated transposition as a mechanism for chromosomal rearrangement
From 1998 to 2007, several laboratories proposed that the RAG complex might insert the blunt RSS-containing ends from V(D)J recombination, termed signal ends, into new locations in the genome. This is called RAG transposition, and occurs at a low level using a truncated form of the RAG proteins called core RAGs (reviewed in ). However, efforts to find RAG transposition events in vivo showed that these were much less common than random integration of DNA . Finally, there are no examples of human lymphoid malignancies (or any other type of malignancy) where the genome was altered by a RAG transpositional insertion of signal ends (or any other apparent variant of such a transposition).
AID action at off-target locations
As mentioned in the above discussion of class switch recombination, AID can convert C to U or methyl C or T at any region of ssDNA. This appears to occur not only at the switch sequences and variable domains of the Ig loci, but also at some pathologic locations, such as some oncogenes like c-myc [3, 4]. When targeted by AID, these regions may sustain point mutations or DSBs . AID action at the IgH switch region during CSR and independent AID action at the c-myc gene to create a DSB are thought to be the basis of the two initiating DSBs in both mouse and human c-myc translocations [1–4]. One could regard breaks of this type as CSR-type breaks (as mentioned above in the discussion of class switch recombination) or SHM-type breaks, where SHM refers to AID initiated events of the type similar to what normally occurs in somatic hypermutation.
Putative combined action of AID and RAGs at CpG sites: CpG-type breaks
Recently, we reported that DSBs at certain loci in pro-B/pre-B stage translocations – the bcl-2 from t(14;18), the bcl-1 from t(11;14), and E2A from t(1;19) – have a strong propensity to occur at the dinucleotide sequence CpG.
The bcl-2 translocation is the most common translocation in cancer, occurring in >90% of follicular lymphomas and a third of diffuse large cell lymphomas. Fifty percent of the breaks at the bcl-2 gene occur within the major breakpoint region (MBR), which is a 175 bp hotspot in the 3' most exon in the region encoding the 3'UTR. Two less-frequently used hotspots are located 18 and 29 kb further distal to the bcl-2 gene, the 105 bp bcl-2 intermediate cluster region (icr), and the 561 bp bcl-2 minor cluster region (mcr), respectively. Any of the CpG sites within any of these three bcl-2 translocation zones can be a target for a DSB . Thirteen percent of bcl-2 translocation breaks are located in the icr, and 5% in the mcr.
The use of CpGs applies also to the bcl-1 major translocation cluster, which is the location involved in the t(11;14) translocation. The bcl-1 translocation occurs in almost all mantle cell lymphomas, with 30% of the breaks occurring at the 150 bp bcl-1 major translocation cluster (MTC).
CpG-type breaks also occur in a third lymphoid malignancy, the t(1;19) in a small percentage of pre-B ALLs, a translocation which occurs between the Pbx1 gene and the E2A gene. The breaks at the E2A gene occur in a zone of only 23 bp, and these DSBs are also significantly clustered around CpG sites . All three translocations involving the bcl-2, bcl-1 and E2A occur at the pro-B/pre-B stage of B-cell development.
The bcl-2 MBR is reactive with a chemical probe for single-strandedness called bisulfite . Like the bcl-2 MBR, this bcl-1 MTC is relatively small (150 bp) and features a similar reactivity to bisulfite . These highly bisulfite reactive zones are rich in runs of Cs. Based on circular dichroism, X-ray crystallography, NMR, and chemical probing, such runs of Cs tend to adopt a DNA structure that is intermediate between B-form DNA and A-form DNA, termed B/A-intermediate . The B/A-intermediate structure has more rapid opening kinetics, perhaps accounting for part of the observed increase in bisulfite reactivity. Such unusual DNA regions may be more prone to slippage events, perhaps induced by DNA replication or transcription. This may then account for their vulnerability in minichromosomal recombination assays .
The Cs of the CpGs within or directly adjacent to these B/A-intermediate zones are at increased risk of undergoing deamination . This deamination does not apply to all Cs in the region, but only the Cs that are within CpG sites. The only distinctive feature about such Cs within CpGs is that they can be methylated by DNA methyltransferase. When regular Cs deaminate, they become U, resulting in a U:G mismatch. But when methyl Cs deaminate, they become T, resulting in a T:G mismatch. The repair of U:G mismatches is very efficient, but the repair of T:G mismatches is not efficient. In fact, T:G mismatch repair is so inefficient, it accounts for about half of the point mutations at the p53 gene across a wide range of human cancers. These T:G mismatch sites are always at CpG sites.
What causes the break at these T:G mismatch sites? Interestingly, this deamination at these lymphoid translocation hotspots appears to occur at the pre-B stage of differentiation. This is the stage of B cell development when D to J recombination is occurring most vigorously. Since the bcl-2 and bcl-1 translocations occur at this stage, this seems likely to be the stage of the translocation. We have shown that the RAG complex can cause a DSB at sites of small bubble structures, and even single base pair mismatches. (As mentioned above, this action by the RAG complex reflects its structure-specific nuclease activity, perhaps a feature that reflects the structure-specific actions by the RAG complex during the hairpin formation step of V(D)J recombination.) Therefore, we have proposed that the RAG complex makes the DSBs at the sites of T:G mismatch .
If the RAG complex causes the DSBs at CpG sites, then why do such CpG-type breaks not occur in pre-T cells, which also express the RAG enzyme complex? The B cell lineage expresses a cytidine deaminase for class switch recombination and somatic hypermutation. As mentioned above, this enzyme is called activation-induced deaminase (AID). AID is expressed in B cells but not other somatic cells. AID is most highly expressed in B cells when they are in the germinal centers. However, a low level of AID expression has been described in pre-B cells [32–34]. Moreover, B cells just leaving the bone marrow, called transitional B cells, also are thought to express AID . Therefore, there is a period of time when B cells are completing V(D)J recombination and beginning to express AID when both AID and the RAG complex are present in the B cells. AID has been shown to be capable of deaminating methyl C to T. Therefore, we propose that AID is likely responsible for the mutation of meC to T at CpG sites in early B cells. The resulting T:G mismatch is then cut by the RAG complex, resulting in a DSB. This model explains three peaks of translocation located within the bcl-2 MBR, all of which are centered at CpG sites .
Other causes of pathologic DSBs of unknown mechanism
Certain translocations are heavily associated with type II topoisiomerase inhibitor therapy . After such therapy, some patients develop secondary malignancies with these characteristic translocations. Topoisomerases in general make single- or double-strand breaks in order to wind or unwind DNA, thus they have a nuclease activity as part of their function. After winding or unwinding the DNA, they normally reseal the break(s). It has been proposed that interruption or prevention of resealing may result in stable breaks seen in chromosomal rearrangements [36, 37].
Some DSBs arise at sites nearby direct or inverted DNA repeats. Such repeats may give rise to slipped DNA structures containing regions of single-stranded DNA, which may be targets for cleavage. The best example of this is the constitutional translocation t(11;22)(q23;q11), which contains an AT-rich palindrome of several hundred bases, with potential for cruciform formation.
Combination of multiple DSB mechanisms within a rearrangement
Given that two DSBs are required to generate a translocation, the two breaks are often not related to one another. In the bcl-2 and bcl-1 translocations, for example, the break at the IgH locus is a V(D)J-type break generated by the sequence-specific action of the RAG complex during V(D)J recombination. (One could consider this to be a failure in the completion of the normal V(D)J recombination process [23–25].) The DSB at the bcl-2 or bcl-1 locus is a CpG-type break that has been proposed to be due to the sequential action of AID and the structure-specific nicking activity of the RAG complex .
Even within a given locus, there can be a wide range of DSB mechanisms. The SCL and LMO2 loci predominantly both sustain V(D)J-type DSBs, but one-third or more of the DSBs are incompatible with the sequence requirements for V(D)J-type DSBs, and these may be due to free radical damage, ionizing radiation, or topoisomerase failures. Different loci within a single cell are therefore prone to different types of DSB mechanisms.
During DNA replication, deletions can arise due to slippage of the synthesizing strand on the template strand. Chromosomal rearrangements that occur at specific hotspots, whether in cancer in somatic cells or during gametogenesis/initial developmental divisions as constitutional translocations, are called recurrent translocations that can be seen across many patients. Nonrecurrent translocations are those that occur at different locations from one patient to another but alter or inactivate a gene that causes a disease. Unlike the recurrent translocations that we have discussed in cancer above, the mechanisms that cause the strand exchange in nonrecurrent translocations appear to involve template switching during replicative DNA synthesis. These template switches can occur at small regions of DNA sequence homology, such as 5 bp. This template switching has been called microhomology-mediated breakage-induced replication (MMBIR) or Fork Stalling and Template Switching (FoSTeS). For nonrecurrent translocation junctions that involve several long stretches of sequence from regions of the genome that are normally separated from one another, multiple template switching events has been proposed as a mechanism [38, 39].