A Modified Digestion-Circularization PCR (DC-PCR) Approach to Detect Hypermutation- Associated DNA Double-Strand Breaks SARAH K. DICKERSON AND F. NINA PAPAVASILIOU Laboratory of Lymphocyte Biology, The Rockefeller University, New York, New York 10021, USA ABSTRACT: Hypermutation of immunoglobulin variable region genes occurs in B cells during an immune response, and is vital for the development of highaffinity antibodies. The molecular mechanism of the hypermutation reaction is unknown, but it seems to correlate with the generation of locus-specific, doublestrand breaks. These DNA breaks have been measured by ligation-mediated polymerase chain reaction (LM-PCR), a technique that relies upon the ligation of a small linker to DNA breaks followed by the specific amplification of such breaks using combinations of locus- and linker-specific primers. Here, we describe a modified version of the digestion-circularization PCR (DC-PCR) technique, which can be used to amplify and measure DNA breaks directly. KEYWORDS: B lymphocyte; immunoglobulin gene; somatic hypermutation; DNA double-strand breaks Support for the existence of DNA strand lesions in somatic hypermutation (SHM) comes from recent experiments describing the constitutively hypermutating Ramos cell line, and the ability of terminal deoxynucleotidyl transferase (TdT)-transfected Ramos cells to accumulate untemplated nucleotide additions near mutational hotspots within the Ig V region. 1 Since TdT can add untemplated nucleotides to either single- or double-strand breaks, 2 the nature of the breaks remained uncertain. To address this issue, we and others 3,4 have probed genomic DNA from Ramos cells 3 or from germinal center cells from immunized mice 4 for locus-specific, double-strand breaks (DSBs) by ligation-mediated polymerase chain reaction (LM- PCR) (FIG. 1A). 5 Blunt DSBs were readily detectable over the Ramos Ig heavy and light chain V regions (V H and V l ) (FIG. 1B), 3 but could not be amplified from the heavy chain constant region or from a panel of other genomic loci. 3 In addition, these same breaks could not be amplified with reverse orientation primers (which would detect the downstream end of the break, defined as the end lying on the Address for correspondence: F. Nina Papavasiliou, Laboratory of Lymphocyte Biology, The Rockefeller University, Box 39, 1230 York Avenue, New York, NY 10021.Voice: 212-327-7857; fax: 212-327-7319. papavasiliou@rockefeller.edu Ann. N.Y. Acad. Sci. 987: 135 139 (2003). 2003 New York Academy of Sciences. 135
136 ANNALS NEW YORK ACADEMY OF SCIENCES FIGURE 1. LM-PCR amplification of DSBs from the Ramos heavy-chain, variable-region gene. (A) Schematic of the principle behind LM-PCR: a double-stranded DNA linker is ligated to DNA ends, and the region proximal to the end is amplified through two rounds of PCR using locus-specific and linker-specific nested primers. (B) Dilution analysis for the quantitation of variable-region DSBs using the indicated cell equivalents of linker-ligated DNA (two rounds of PCR: 10 cycles in the first round, and 30 cycles in the second). 3 Amplification products representing DNA DSBs over the Ramos VH are run on a polyacrylamide gel that is directly stained with SybrGreen and visualized on a FluorImager. Generic PCR amplification of the entire VH (30 cycles) serves as a DNA loading control. piece of DNA separated from the promoter 3 ). Hence, DSB formation in somatic hypermutation results in asymmetric products: upstream ends that are usually blunt, and downstream ends that almost always contain a modification that prevents their detection. An alternative explanation for the asymmetry would be that the process actually creates single-strand gaps. 6 If such gaps were largely confined to one strand of the DNA, and if the linker could ligate to the gaps, we would expect an asymmetric distribution of LM-PCR products, such as we observe. To determine the ligation efficiency of a double-stranded linker to a gap, we generated two substrates (one with a 19-nt gap and the other with a DSB) that could be detected by the same specific PCR primers. In competitive LM-PCR experiments, gaps were detected 100- to 1000-fold less efficiently than DSBs (FIG. 2), indicating that our LM-PCR assay would detect at most 1% of gaps. Since the assay detects a product from about 4 5% of Ig variable region alleles in Ramos cells, gaps would have to exist at a rate of at least 400 per 100 alleles to account for our data. This is improbable, but not impossible. To deal with this issue, we devised a modified digestion-circularization PCR (DC-PCR) assay to look at the same DSBs. We reasoned that if the upstream ends of the hypermutation-associated DSBs were true blunt DSBs, one should be able to digest the DNA upstream of the V region with a restriction enzyme that produces blunt ends: subsequent ligation then would seal the two blunt ends and produce a circle (FIG. 3A). One can amplify the artificial joint by choosing appropriate oligonucleotide primers. We digested 100 µl of agarose-embedded Ramos DNA with EcoRV, which recognizes a site 3 kb upstream from the Ramos V region (FIG. 3B). Then, 24 hours lat-
DICKERSON & PAPAVASILIOU: DC-PCR TO DETECT DNA BREAKS 137 FIGURE 2. Comparison of the ligation efficiency of a double-stranded DNA linker to a gap vs. a DSB. We created a 550-nt DNA fragment with recognition sequences for N.BstNB1, a site-specific endonuclease that catalyzes a single-strand break 4 nts beyond the 3 side of the recognition sequence, GAGTCNNNN/. N.BstNB1 cleaves twice in the substrate, and as the sites of cleavage are on the same strand and 19 nts apart, the reaction results in the formation of a 19-nt gap, 290 nts from the one end of the 550-nt double-stranded fragment. Formation of the gap can be monitored, as it is accompanied by the loss of a BsmF1 restriction enzyme site that lies within the 19-nt gapped region. In competitive LM-PCR experiments, gaps were detected with 100- (left) to 1000-fold (right) lower efficiency than DSBs (left: compare GAP and DSB signal in lanes 7,8; right: compare DSB and GAP signal in lanes 4 and 5). in: input DNA. er, we diluted the digest to 600 µl and ligated the reaction overnight, followed by PCR amplification ( 30 cycles) of the generated joints. We could easily amplify bands only from agarose-embedded Ramos DNA when we would digest with EcoRV prior to ligation. The appearance of the bands was dependent on ligation (FIG. 3A, lanes 2 and 3 vs. lanes 5 and 6), and the bands disappear if, in addition to EcoRV, the genomic DNA was digested with SnaB1, an enzyme that is predicted to cleave the DNA between the EcoRV site and the hypermutation-associated DSBs. Sequencing of the PCR products revealed that the location of these DSBs is not random, but like the LM-PCR amplified breaks, these DSBs also tended to co-localize with mutational hotspots (Ref. 3 and unpublished results). Because of its dependence on a particular restriction site, this assay is less promiscuous in DSB detection when compared with LM-PCR. In addition, because it does not depend on linker ligation, it only detects true blunt DSBs (and not gaps). We could not amplify specific joints between the downstream end and the end generated by blunt-cutting enzymes located downstream of the V region DSB (FIG. 3B). These data strongly suggest that the SHM-associated lesions we detect are not nicks or gaps but true blunt DSBs, which are asymmetrically detectable. It is possible that a subset of DSBs (perhaps consisting of the few that are detected from both directions) is related to normal breakage and repair processes active in rapidly replicating cells and, as such, are unrelated to SHM. But given the frequency and the location
138 ANNALS NEW YORK ACADEMY OF SCIENCES FIGURE 3. DC-PCR amplification of DSBs from the Ramos heavy-chain, variable-region gene. (A) Schematic representation of the modified DC-PCR assay. The locus is cut with a locus-specific, blunt restriction enzyme (in this case, EcoRV). In addition, the locus carries a mutation-generated DSB, the upstream end of which is blunt. Self-ligation of the region will lead to the circularization of the EcoRV-generated ends and of the hypermutation-generated ends. Appropriately chosen primers can then amplify the artificial joint. (B) Amplification of variable-region DSBs from 50,000 cell equivalents (lane 5) and from 5,000 cell equivalents (lane 6) (single round of PCR, 30 cycles). Amplification was with appropriate primers, and products representing DNA DSBs over the Ramos VH are run on a polyacrylamide gel that is directly stained with SybrGreen and visualized on a FluorImager. of the DSBs over the mutating V regions, as well as the absence of such breaks over other genomic loci, we conclude that in Ramos cells, DSBs over the variable regions are tightly linked to somatic hypermutation. Thus, there is a body of data which suggests that SHM involves the generation and processing of DNA DSBs. But the identity of the putative nuclease(s), as well as the manner by which it is specifically targeted to immunoglobulin loci, are crucial issues yet to be resolved. Clearly, our understanding of the mechanism of SHM remains rudimentary, but significant advances may come quickly through a better definition of the processes that target and generate the relevant DNA lesions. REFERENCES 1. SALE, J.E. & M.S. NEUBERGER. 1998. TdT-accessible breaks are scattered over the immunoglobulin V domain in a constitutively hypermutating B cell line. Immunity 9: 859 869. 2. GRAWUNDER, U. & M. LIEBER. 1997. A complex of RAG-1 and RAG-2 proteins persists on DNA after single-strand cleavage at V(D)J recombination signal sequences. Nucleic Acids Res. 25: 1375 1382. 3. PAPAVASILIOU, F.N. & D.G. SCHATZ. 2000. Cell-cycle-regulated DNA double-stranded breaks in somatic hypermutation of immunoglobulin genes. Nature 408: 216 221.
DICKERSON & PAPAVASILIOU: DC-PCR TO DETECT DNA BREAKS 139 4. BROSS, L. et al. 2000. DNA double-strand breaks in immunoglobulin genes undergoing somatic hypermutation. Immunity 13: 589 597. 5. SCHLISSEL, M. & D. BALTIMORE. 1989. Activation of immunoglobulin kappa gene rearrangement correlates with induction of germline kappa gene transcription. Cell 58: 1001 1007. 6. KONG, Q. & N. MAIZELS. 2001. DNA breaks in hypermutating immunoglobulin genes: evidence for a break-and-repair pathway of somatic hypermutation. Genetics 158: 369 378.