Search

Chromosomal translocations are a hallmark of hematopoietic malignancies. CG motifs within translocation fragile zones (typically 20 to 600 bp in size) are prone to chromosomal translocation in lymphomas. Here we demonstrate that the CG motifs in human translocation fragile zones are hypomethylated relative to the adjacent DNA. Using a methyltransferase footprinting assay on isolated nuclei (in vitro), we find that the chromatin at these fragile zones is accessible. We also examined in vivo accessibility using cellular expression of a prokaryotic methylase. Based on this assay, which measures accessibility over a much longer time interval than is possible with in vitro methods, these fragile zones were found to be more accessible than the adjacent DNA. Because DNA within the fragile zones can be methylated by both cellular and exogenous methyltransferases, the fragile zones are predominantly in a duplex DNA conformation. These observations permit more-refined models for why these zones are 100- to 1,000-fold more prone to undergo chromosomal translocation than the adjacent regions., (Copyright © 2015, American Society for Microbiology. All Rights Reserved.)

The t(14;18) chromosomal translocation typically involves breakage at the bcl-2 major breakpoint region (MBR) to cause human follicular lymphoma. A theory to explain the striking propensity of the MBR breaks at three CpG clusters within the 175-bp MBR region invoked activation-induced deaminase (AID). In a test of that theory, we used here minichromosomal substrates in human pre-B cell lines. Consistent with the essential elements of the theory, we found that the MBR breakage process is indeed highly dependent on DNA methylation at the CpG sites and highly dependent on the AID enzyme to create lesions at peak locations within the MBR. Interestingly, breakage of the phosphodiester bonds at the AID-initiated MBR lesions is RAG dependent, but, unexpectedly, most are also dependent on Artemis. We found that Artemis is capable of nicking small heteroduplex structures and is even able to nick single-base mismatches. This raises the possibility that activated Artemis, derived from the unjoined D to J(H) DNA ends at the IgH locus on chromosome 14, nicks AID-generated TG mismatches at methyl CpG sites, and this would explain why the breaks at the chromosome 18 MBR occur within the same time window as those on chromosome 14.

During V(D)J recombination, RAG targeting to correct sites versus off-target sites relies on both DNA sequence features and on chromatin marks. Kinetic analysis using the first highly active full-length purified RAG1/RAG2 complexes has now allowed us to define the important catalytic features of this complex. We found that the overall rate of nicking, but not hairpinning, is critical for the discrimination between correct (optimal) versus off-target (suboptimal) sites used in human T-cell lymphomas, and we show that the C-terminal portion of RAG2 is required for this. This type of kinetic analysis permits us to analyze only the catalytically active RAG complex, in contrast to all other methods, which are unavoidably confounded by mixture with inactive RAG complexes. Moreover, we can distinguish the two major features of any enzymatic catalysis: the binding constant (K(D)) and the catalytic turnover rate, k(cat). Beyond a minimal essential threshold of heptamer quality, further suboptimal heptamer deviations primarily reduce the catalytic rate constant k(cat) for nicking. Suboptimal nonamers reduce not only the binding of the RAG complex to the recombination site (K(D)) but also the catalytic rate constant, consistent with a tight interaction between the RAG complex and substrate during catalysis. These features explain many aspects of RAG physiology and pathophysiology.

Upon transcription of some sequences by RNA polymerases in vitro or in vivo, the RNA transcript can thread back onto the template DNA strand, resulting in an R loop. Previously, we showed that initiation of R-loop formation at an R-loop initiation zone (RIZ) is favored by G clusters. Here, using a purified in vitro system with T7 RNA polymerase, we show that increased distance between the promoter and the R-loop-supporting G-rich region reduces R-loop formation. When the G-rich portion of the RNA transcript is downstream from the 5' end of the transcript, the ability of this portion of the transcript to anneal to the template DNA strand is reduced. When we nucleolytically resect the beginning of the transcript, R-loop formation increases because the G-rich portion of the RNA is now closer to the 5' end of the transcript. Short G-clustered regions can act as RIZs and reduce the distance-induced suppression of R-loop formation. Supercoiled DNA is known to favor transient separation of the two DNA strands, and we find that this favors R-loop formation even in non-G-rich regions. Most strikingly, a nick can serve as a strong RIZ, even in regions with no G richness. This has important implications for class switch recombination and somatic hypermutation and possibly for other biological processes in transcribed regions.

R-loops form cotranscriptionally in vitro and in vivo at transcribed duplex DNA regions when the nascent RNA is G-rich, particularly with G clusters. This is the case for phage polymerases, as used here (T7 RNA polymerase), as well as RNA polymerases in bacteria, Saccharomyces cerevisiae, avians, mice, and humans. The nontemplate strand is left in a single-stranded configuration within the R-loop region. These structures are known to form at mammalian immunoglobulin class switch regions, thus exposing regions of single-stranded DNA for the action of AID, a single-strand-specific cytidine deaminase. R-loops form by thread-back of the RNA onto the template DNA strand, and here we report that G clusters are extremely important for the initiation phase of R-loop formation. Even very short regions with one GGGG sequence can initiate R-loops much more efficiently than random sequences. The high efficiencies observed with G clusters cannot be achieved by having a very high G density alone. Annealing of the transcript, which is otherwise disadvantaged relative to the nontemplate DNA strand because of unfavorable proximity while exiting the RNA polymerase, can offer greater stability if it occurs at the G clusters, thereby initiating an R-loop. R-loop elongation beyond the initiation zone occurs in a manner that is not as reliant on G clusters as it is on a high G density. These results lead to a model in which G clusters are important to nucleate the thread-back of RNA for R-loop initiation and, once initiated, the elongation of R-loops is primarily determined by the density of G on the nontemplate DNA strand. Without both a favorable R-loop initiation zone and elongation zone, R-loop formation is inefficient.

R-loops have been described in vivo at the immunoglobulin class switch sequences and at prokaryotic and mitochondrial origins of replication. However, the biochemical mechanism and determinants of R-loop formation are unclear. We find that R-loop formation is nearly eliminated when RNase T(1) is added during transcription but not when it is added afterward. Hence, rather than forming simply as an extension of the RNA-DNA hybrid of normal transcription, the RNA must exit the RNA polymerase and compete with the nontemplate DNA strand for an R-loop to form. R-loops persist even when transcription is done in Li(+) or Cs(+), which do not support G-quartet formation. Hence, R-loop formation does not rely on G-quartet formation. R-loop formation efficiency decreases as the number of switch repeats is decreased, although a very low level of R-loop formation occurs at even one 49-bp switch repeat. R-loop formation decreases sharply as G clustering is reduced, even when G density is kept constant. The critical level for R-loop formation is approximately the same point to which evolution drove the G clustering and G density on the nontemplate strand of mammalian switch regions. This provides an independent basis for concluding that the primary function of G clustering, in the context of high G density, is R-loop formation.

The mechanism by which the cytidine deaminase activation-induced deaminase (AID) acts at immunoglobulin heavy-chain class switch regions during mammalian class switch recombination (CSR) remains unclear. R-loops have been proposed as a basis for this targeting. Here, we show that the difference between various forms of the Smu locus that can or cannot undergo CSR correlates well with the locations and detectability of R-loops. The Smu R-loops can initiate hundreds of base pairs upstream of the core repeat switch regions, and the area where the R-loops initiate corresponds to the zone where the AID mutation frequency begins to rise, despite a constant density of WRC sites in this region. The frequency of R-loops is 1 in 25 alleles, regardless of the presence of the core Smu repeats, again consistent with the initiation of most R-loops upstream of the core repeats. These findings explain the surprisingly high levels of residual CSR in B cells from mice lacking the core Smu repeats but the marked reduction in CSR in mice with deletions of the region upstream of the core Smu repeats. These studies also provide the first analysis of how R-loop formation in the eukaryotic chromosome depends on the DNA sequence.

The t(14;18) chromosomal translocation is the most common translocation in human cancer, and it occurs in all follicular lymphomas. The 150-bp bcl-2 major breakpoint region (Mbr) on chromosome 18 is a fragile site, because it adopts a non-B DNA conformation that can be cleaved by the RAG complex. The non-B DNA structure and the chromosomal translocation can be recapitulated on intracellular human minichromosomes where immunoglobulin 12- and 23-signals are positioned downstream of the bcl-2 Mbr. Here we show that either of the two coding ends in these V(D)J recombination reactions can recombine with either of the two broken ends of the bcl-2 Mbr but that neither signal end can recombine with the Mbr. Moreover, we show that the rejoining is fully dependent on DNA ligase IV, indicating that the rejoining phase relies on the nonhomologous DNA end-joining pathway. These results permit us to formulate a complete model for the order and types of cleavage and rejoining events in the t(14;18) translocation.

The most common chromosomal translocation in cancer, t(14;18) at the 150-bp bcl-2 major breakpoint region (Mbr), occurs in follicular lymphomas. The bcl-2 Mbr assumes a non-B DNA conformation, thus explaining its distinctive fragility. This non-B DNA structure is a target of the RAG complex in vivo, but not because of its primary sequence. Here we report that the RAG complex generates at least two independent nicks that lead to double-strand breaks in vitro, and this requires the non-B DNA structure at the bcl-2 Mbr. A 3-bp mutation is capable of abolishing the non-B structure formation and the double-strand breaks. The observations on the bcl-2 Mbr reflect more general properties of the RAG complex, which can bind and nick at duplex-single-strand transitions of other non-B DNA structures, resulting in double-strand breaks in vitro. Hence, the present study reveals novel insight into a third mechanism of action of RAGs on DNA, besides the standard heptamer/nonamer-mediated cleavage in V(D)J recombination and the in vitro transposase activity.

Activation-induced deaminase (AID) is essential for class switch recombination and somatic hypermutation, and it has the ability to deaminate single-stranded DNA at cytidines. Mammalian class switch regions form R-loops upon transcription in the physiological orientation. The displaced DNA strand of an R-loop is forced to wrap around the RNA-DNA hybrid; hence, it may not have complete exposure to proteins. A fundamental question concerns the extent to which AID is accessible to the displaced strand of a transcription-generated R-loop. We used a minimal R-loop to carry out high-resolution analysis of the precise locations of AID action. We found that AID deaminates on the displaced DNA strand across the entire length of the R-loop. Displaced strand locations with a WRC (where W is A or T and R is A or G) sequence are preferred targets, but there are clear exceptions. These WRC deviations may be due to steric constraints on the accessibility of AID to these sites as the displaced strand twists around the RNA-DNA duplex. This phenomenon may explain the lack of WRC site preference at the mutations surrounding class switch recombination junctions.

Endonuclease G (endo G) is one of the most abundant nucleases in eukaryotic cells. It is encoded in the nucleus and imported to the mitochondrial intermembrane space. This nuclease is active on single- and double-stranded DNA. We genetically disrupted the endo G gene in mice without disturbing a conserved, overlapping gene of unknown function that is oriented tail to tail with the endo G gene. In these mice, the production of endo G protein is not detected, and the disruption abolishes the nuclease activity of endo G. The absence of endo G has no effect on mitochondrial DNA copy number, structure, or mutation rate over the first five generations. There is also no obvious effect on nuclear DNA degradation in standard apoptosis assays. The endo G null mice are viable and show no age-related or generational abnormalities anatomically or histologically. We infer that this highly conserved protein has no mitochondrial or apoptosis function that can discerned by the assays described here and that it may have a function yet to be determined. The early embryonic lethality of endo G null mice recently reported by others may be due to the disruption of the gene that overlaps the endo G gene.

In all of the transposition reactions that have been characterized thus far, synapsis of two transposon ends is required before any catalytic steps (strand nicking or strand transfer) occur. In V(D)J recombination, there have been inconclusive data concerning the role of synapsis in nicking. Synapsis between two 12-substrates or between two 23-substrates has not been ruled out in any studies thus far. Here we provide the first direct tests of this issue. We find that immobilization of signals does not affect their nicking, even though hairpinning is affected in a manner reflecting its known synaptic requirement. We also find that nicking is kinetically a unireactant enzyme-catalyzed reaction. Time courses are no different between nicking seen for a 12-substrate alone and a reaction involving both a 12- and a 23-substrate. Hence, synapsis is neither a requirement nor an effector of the rate of nicking. These results establish V(D)J recombination as the first example of a DNA transposition-type reaction in which catalytic steps begin prior to synapsis, and the results have direct implications for the order of the steps in V(D)J recombination, for the contribution of V(D)J recombination nicks to genomic instability, and for the diversification of the immune repertoire.

V(D)J recombination is directed by recombination signal sequences. However, the flanking coding end sequence can markedly affect the frequency of the initiation of V(D)J recombination in vivo. Here we demonstrate that the coding end sequence effect can be qualitatively and quantitatively recapitulated in vitro with purified RAG proteins. We find that coding end sequence specifically affects the nicking step, which is the first biochemical step in RAG-mediated cleavage. The subsequent hairpin formation step is not affected by the coding end sequence. Furthermore, the coding end sequence effect can be ablated by prenicking the substrate, indicating that the coding end effect is specific to the nicking step. In reactions in which both 12- and 23-substrates are present, a suboptimal coding end sequence on one signal can slow down hairpin formation at the partner signal, a result consistent with models in which coordination between the signals occurs at the hairpin formation step. The coding end sequence effect on nicking and the coupling of the 12- and 23-substrates explains how hairpin formation can be rate limiting for some 12/23 pairs, whereas nicking can be rate limiting when low-efficiency coding end sequences are involved.

A central unanswered question concerning the initial phases of V(D)J recombination has been at which step the 12/23 rule applies. This rule, which governs which variable (V), diversity (D), and joining (J) segments are able to pair during recombination, could operate at the level of signal sequence synapsis after RAG-HMG1 complex binding, signal nicking, or signal hairpin formation. It has also been unclear whether additional proteins are required to achieve adherence to the 12/23 rule. We developed a novel system for the detailed biochemical analysis of the 12/23 rule by using an oligonucleotide-based substrate that can include two signals. Under physiologic conditions, we found that the complex of RAG1, RAG2, and HMG1 can successfully recapitulate the 12/23 rule with the same specificity as that seen intracellularly and in crude extracts. The cleavage complex can bind and nick 12x12 and 23x23 substrates as well as 12x23 substrates. However, hairpin formation occurs at both of the signals only on 12x23 substrates. Moreover, under physiologic conditions, the presence of a partner 23-bp spacer suppresses single-site hairpin formation at a 12-bp spacer and vice versa. Hence, this study illustrates that synapsis suppresses single-site reactions, thereby explaining the high physiologic ratio of paired versus unpaired V(D)J recombination events in lymphoid cells.

DNA-dependent protein kinase (DNA-PK) is the only eukaryotic protein kinase known to be specifically activated by double-stranded DNA (dsDNA) termini, accounting for its importance in repair of dsDNA breaks and its role in physiologic processes involving dsDNA breaks, such as V(D)J recombination. In this study we conducted kinase and binding analyses using DNA-PK on DNA termini of various lengths in the presence and absence of Ku. We confirmed our previous observations that DNA-PK can bind DNA termini in the absence of Ku, and we determined rate constants for binding. However, in the presence of Ku, DNA-PK can assume either a productive or a nonproductive configuration, depending on the length of the DNA terminus. For dsDNA greater than 26 bp, the productive mode is achieved and Ku increases the affinity of the DNA-PK for the Ku:DNA complex. The change in affinity is achieved by increases in both the kinetic association rate and reduction in the kinetic dissociation rate. For dsDNA smaller than 26 bp, the nonproductive mode, in which DNA-PK is bound to Ku:DNA but is inactive as a kinase, is assumed. Both the productive and nonproductive configurations are likely to be of physiologic importance, depending on the distance of the dsDNA break site to other protein complexes, such as nucleosomes.

DNA ends are generated during double-strand-break repair and recombination. A p70-p86 heterodimer, Ku, accounts for the DNA end binding activity in eukaryotic cell extracts. When one or both subunits of Ku are missing, mammalian cells are deficient in double-strand-break repair and in specialized recombination, such as V(D)J recombination. Little is known of which regions of Ku70 and Ku86 bind to each other to form the heterodimeric complex or of which regions are important for DNA end binding. We have done genetic and biochemical studies to examine the domains within the two subunits important for protein assembly and for DNA end binding. We found that the C-terminal 20-kDa region of Ku70 and the C-terminal 32-kDa region of Ku86 are important for subunit-subunit interaction. For DNA binding, full-length individual subunits are inactive, indicating that heterodimer assembly precedes DNA binding. DNA end binding activity by the heterodimer requires the C-terminal 40-kDa region of Ku70 and the C-terminal 45-kDa region of Ku86. Leucine zipper-like motifs in both subunits that have been suggested as the Ku70-Ku86 interaction domains do not appear to be the sites of such interaction because these are dispensable for both assembly and DNA end binding. On the basis of these studies, we have organized Ku70 into nine sequence regions conserved between Saccharomyces cerevisiae, Drosophila melanogaster, mice, and humans; only the C-terminal three regions are essential for assembly (amino acids [aa] 439 to 609), and the C-terminal four regions appear to be essential for DNA end binding (aa 254 to 609). Within the minimal active fragment of Ku86 necessary for subunit interaction (aa 449 to 732) and DNA binding (aa 334 to 732), a proline-rich region is the only defined motif.

We have analyzed a large collection of coding junctions generated in human cells. From this analysis, we infer the following about nucleotide processing at coding joints in human cells. First, the pattern of nucleotide loss from coding ends is influenced by the base composition of the coding end sequences. AT-rich sequences suffer greater loss than do GC-rich sequences. Second, inverted repeats can occur at ends that have undergone nucleolytic processing. Previously, inverted repeats (P nucleotides) have been noted only at coding ends that have not undergone nucleolytic processing, this observation being the basis for a model in which a hairpin intermediate is formed at the coding ends early in the reaction. Here, inverted repeats at processed coding ends were present at approximately twice the number of junctions as P nucleotide additions. Terminal deoxynucleotidyl transferase (TdT) is required for the appearance of the inverted repeats at processed ends (but not full-length coding ends), yet statistical analysis shows that it is virtually impossible for the inverted repeats to be polymerized by TdT. Third, TdT additions are not random. It has long been noted that TdT has a G utilization preference. In addition to the G preference, we find that TdT adds strings of purines or strings of pyrimidines at a highly significant frequency. This tendency suggests that nucleotide-stacking interactions affect TdT polymerization. All three of these features place constraints on the extent of junctional diversity in human V(D)J recombination.

Substrates for studying V(D)J recombination in human cells and two human pre-B-cell lines that have active V(D)J recombination activity are described. Using these substrates, we have been able to analyze the relative efficiency of signal joint and coding joint formation. Coding joint formation was five- to sixfold less efficient than signal joint formation in both cell lines. This imbalance between the two halves of the reaction was demonstrated on deletional substrates, where each joint is assayed individually. In both cell lines, the inversional reaction (which requires formation of both a signal and a coding joint) was more than 20-fold less efficient than signal joint formation alone. The signal and coding sequences are identical in all of these substrates. Hence, the basis for these differential reaction ratios appears to be that coding joint and signal joint formation are both inefficient and their combined effects are such that inversions (two-joint reactions) reflect the product of these inefficiencies. Physiologically, these results have two implications. First, they show how signal and coding joint formation efficiencies can affect the ratio of deletional to inversional products at endogenous loci. Second, the fact that not all signal and coding joints go to completion implies that the recombinase is generating numerous broken ends. Such unresolved ends may participate in pathologic chromosomal rearrangements even when the other half of the same reaction may have proceeded to resolution.

V(D)J recombination has been examined in several X-ray-sensitive and double-strand break repair-deficient Chinese hamster cell mutants. Signal joint formation was affected in four mutants (xrs 5, XR-1, V-3, and XR-V9B cells, representing complementation groups 1 through 4, respectively) defective in DNA double-strand break rejoining. Among these four, V-3 and XR-V9B were the most severely affected. Only in V-3 was coding joint formation also affected. Ataxia telangiectasia-like hamster cell mutants (V-E5 and V-G8), which are normal for double-strand break repair but are X ray sensitive, were normal for all aspects of the V(D)J recombination reaction, indicating that X-ray sensitivity is not the common denominator but that the deficiency in double-strand break repair appears to be. The abnormality at the signal joints consisted of an elevated incidence of nucleotide loss from each of the two signal ends. Interestingly, in complementation groups 1 (xrs 5) and 2 (XR-1), signal joint formation was within the normal range under some transfection conditions. This suggests that the affected gene products in these two complementation groups are not catalytic components. Instead, they may be either secondary or stochiometric components involved in the later stages of both the V(D)J recombination reaction and double-strand break repair. The fact that such factors can affect the precision of the signal joint has mechanistic implications for V(D)J recombination.

V(D)J recombination in lymphoid cells is a site-specific process in which the activity of the recombinase enzyme is targeted to signal sequences flanking the coding elements of antigen receptor genes. The order of the steps in this reaction and their mechanistic interdependence are important to the understanding of how the reaction fails and thereby contributes to genomic instability in lymphoid cells. The products of the normal reaction are recombinant joints linking the coding sequences of the receptor genes and, reciprocally, the signal ends. Extrachromosomal substrate molecules were modified to inhibit the physical synapsis of the recombination signals. In this way, it has been possible to assess how inhibiting the formation of one joint affects the resolution efficiency of the other. Our results indicate that signal joint and coding joint formation are resolved independently in that they can be uncoupled from each other. We also find that signal synapsis is critical for the generation of recombinant products, which greatly restricts the degree of potential single-site cutting that might otherwise occur in the genome. Finally, inversion substrates manifest synaptic inhibition at much longer distances than do deletion substrates, suggesting that a parallel rather than an antiparallel alignment of the signals is required during synapsis. These observations are important for understanding the interaction of V(D)J signals with the recombinase. Moreover, the role of signal synapsis in regulating recombinase activity has significant implications for genome stability regarding the frequency of recombinase-mediated chromosomal translocations.

Murine severe combined immune deficiency (scid) is marked by a 5,000-fold reduction in coding joint formation in V(D)J recombination of antigen receptors. Others have demonstrated a sensitivity to double-strand breaks generated by ionizing radiation and bleomycin. We were interested in establishing the extent of the defect in intramolecular and intermolecular DNA end joining in lymphoid and nonlymphoid cells from scid mice. We conducted a series of studies probing the ability of these cells to resolve free ends of linear DNA molecules having various biochemical end configurations. We find that the stable integration of linear DNA into scid fibroblasts is reduced 11- to 75-fold compared with that in normal fibroblasts. In contrast, intramolecular and intermolecular end joining occur at normal frequencies in scid lymphocytes and fibroblasts. This normal level of end joining is observed regardless of the type of overhang and regardless of the requirement for nucleolytic activities prior to ligation. The fact that free ends having a wide variety of end configurations are recircularized normally in scid cells rules out certain models for the defect in scid. We discuss the types of DNA end joining reactions that are and are not affected in this double-strand break repair defect in the context of a hairpin model for V(D)J recombination.

We examined a series of extrachromosomal DNA substrates for V(D)J recombination under replicating and nonreplicating conditions. Complete and partial replications were examined by monitoring the loss of prokaryote-specific adenine methylation at 14 to 22 MboI-DpnI restriction sites (GATC) on the substrates. Some of these sites are within 2 bases of the signal sequence ends. We found that neither coding joint nor signal joint formation requires substrate replication. After ruling out replication as a substrate requirement, we determined whether replication had any effect on the efficiency of V(D)J recombination. Quantitation of V(D)J recombination efficiency on nonreplicating substrates requires some method of monitoring the entry of substrate molecules into the cells. We devised such a method by monitoring DNA repair of substrates into which we had substituted deoxyuridine for 10 to 20% of the thymidine nucleotides in the DNA. The substrates which enter the lymphoid cells were repaired efficiently in vivo by the eukaryotic uracil DNA repair system. Upon plasmid harvest, we distinguished repaired (entered) from unrepaired (not entered) plasmids by cleaving unrepaired molecules with uracil DNA glycoylase and Escherichia coli endonuclease IV in vitro. This method of monitoring DNA entry does not appear to underestimate or overestimate the amount of DNA entry. By using this method, we found no significant quantitative effect of DNA replication on V(D)J recombination efficiency.