RAG2 mutants alter DSB repair pathway choice in vivo and illuminate the nature of ‘alternative NHEJ’

DNA double-stranded breaks (DSBs) can be repaired by several mechanisms, including classical NHEJ (c-NHEJ) and a poorly defined, error-prone process termed alternative NHEJ (a-NHEJ). How cells choose between these alternatives to join physiologic DSBs remains unknown. Here, we show that deletion of RAG2's C-terminus allows a-NHEJ to repair RAG-mediated DSBs in developing lymphocytes from both c-NHEJ-proficient and c-NHEJ-deficient mice, demonstrating that the V(D)J recombinase influences repair pathway choice in vivo. Analysis of V(D)J junctions revealed that, contrary to expectation, junctional characteristics alone do not reliably distinguish between a-NHEJ and c-NHEJ. These data suggest that a-NHEJ is not necessarily mutagenic, and may be more prevalent than previously appreciated. Whole genome sequencing of a lymphoma arising in a p53−/− mouse bearing a C-terminal RAG2 truncation reveals evidence of a-NHEJ and also of aberrant recognition of DNA sequences resembling RAG recognition sites.

1 Supplementary Information Figure S1. Generation of RAG2 FS/FS knock-in mice.
B. Mice were generated in inGenious Targeting Laboratory Inc. A~11.5kb region used to generate the targeting vector was first sub cloned from a positively identified C57/Bl6 BAC clone. Two types of mutations were generated in exon 3 utilizing overlap extension PCR. The first mutation comprised deletion of base 1082 (T) to generate a frameshift. The second mutation is located 67bp 3 of the T deletion and comprised replacement of the last 435bp of coding sequence with the sequence-AAGCGGCCGCGACTCTAG followed by the 3 UTR sequence. First, primers located 5 and 3 to two unique Kpn1 (K) sites that flank the location of the mutations were used to amplify a 1.6kb fragment. The mutations were introduced into this fragment, which was then reintroduced back into the construct via ligation into the Kpn1 sites, thus replacing the wild type Kpn1 fragment with the mutated Kpn1 fragment. The Neo cassette is inserted 228bps 5 to the ATG in exon 3 using Red/ET recombineering technology with a short homology arm that extends showed an additional amino acid change at T296M that occurred during the targeting.
Nevertheless, this change did not affect the FS mutant functionality in our cell system assay allowing us to still use it as a mouse model to investigate our questions.    Sequence analysis of purified PCR products (n=4 RAG2 FS/FS mice). Capital letters at the middle of the junction represents N nt, Bold italic are microhomology, deletions are indicated in parentheses and small letters are Sanger sequence that did not align to mouse mm9 database.

Figure S6. Generation of Adjacent Direct Repeats -ADRs
A. Scheme showing how small ADRs (adjacent direct repeats) might be generated. The first step may or may not be a mechanistic constraint however it is necessary in the identification of ADRs.
We show homologies between ends having been generated by TdT or perhaps another polymerase (53). Terminal homologies can also be revealed between two ends by resection but were excluded in our analysis because prior existing sequences between ends that can generate ADRs are indistinguishable from simple direct joining products and difficult to score. The next step is stabilizing end-to-end interaction via complementary bases in the two single strand extensions compensating for the lack of Ku80. We suggest that annealing occurs near or at the termini. For the sake of parsimony, we have depicted Fen-1 as the flap removal factor prior to the gap-filling and strand displacement steps. ADRs are then generated when a gap-filling polymerase with strand displacement activity is present. This could be supplied by pol lamda (39), which has sufficient strand displacement activity in vitro with the cooperation of Fen-1, to invade three or four bp into a duplex, or by pol beta, which display stronger strand displacement on its own (38).
(Annotations are as given in Fig.2).
Endogenous antigen receptor rearrangements detected by whole genome sequencing. Seven endogenous genomic rearrangements detected in tumor 13422. All rearrangements were in the coding end configuration. End1 represents the coding end of a 23RSS; End 2 represents the coding end of 12RSS. Capital letters in the middle are N nt and Bold are P nt.

Supplementary Methods V(D)J recombination assay
The 293T cell line was grown in DMEM supplemented with fetal bovine serum (10%), nonessential amino acids and penicillin-streptomycin. Cells were grown at 37 C in the presence of 5% CO2. To assess V(D)J recombination, 0.5 g of the indicated murine Rag1, murine Rag2 and recombination substrate were transfected into cells using a FuGENE 6:DNA ratio of 3:1. Fortyeight hours after transfection cells were harvested and fluorescent intensity was measured using BD LSR II for FACS readout.

Spectratyping
Immunoglobulin heavy chain repertoire analysis was performed using CDR3 spectratyping, as described (54). Briefly, genomic DNA from splenocytes was purified using PureGene and amplified using the J606.1 and the J558.85 VH primers and a fluorescent labeled JH2 reverse primer. 2 L of PCR product per reaction were resolved by capillary electrophoresis on an ABI 3100 analyzer (Applied Biosciences Inc). Peak scanner v. 1.0 was used to generate and analyze the spectratypes (Applied Biosciences Inc) as described (54).