PAXX and XLF DNA repair factors are functionally redundant in joining DNA breaks in a G1-arrested progenitor B-cell line

Significance

Classical nonhomologous end joining (C-NHEJ) is a major mammalian DNA double–strand break (DSB) repair pathway. During V(D)J recombination in progenitor (pro)-B lymphocytes, C-NHEJ joins programmed DSBs in antibody gene loci to form complete antibody genes. C-NHEJ also protects mammalian cells from the harmful effects of exposure to ionizing radiation. We now find that the recently identified paralogue of XRCC4 and XLF (PAXX) DNA repair factor, like the related XLF repair factor, is dispensable for V(D)J recombination. However, combined loss of these two factors in pro–B-cell lines totally abrogates V(D)J recombination DSB joining and greatly sensitizes the cells to ionizing radiation. These findings show that PAXX can provide critical C-NHEJ functions that are normally masked by functional redundancy with XLF.

Abstract

Classical nonhomologous end joining (C-NHEJ) is a major mammalian DNA double-strand break (DSB) repair pathway. Core C-NHEJ factors, such as XRCC4, are required for joining DSB intermediates of the G1 phase-specific V(D)J recombination reaction in progenitor lymphocytes. Core factors also contribute to joining DSBs in cycling mature B-lineage cells, including DSBs generated during antibody class switch recombination (CSR) and DSBs generated by ionizing radiation. The XRCC4-like-factor (XLF) C-NHEJ protein is dispensable for V(D)J recombination in normal cells, but because of functional redundancy, it is absolutely required for this process in cells deficient for the ataxia telangiectasia-mutated (ATM) DSB response factor. The recently identified paralogue of XRCC4 and XLF (PAXX) factor has homology to these two proteins and variably contributes to ionizing radiation-induced DSB repair in human and chicken cells. We now report that PAXX is dispensable for joining V(D)J recombination DSBs in G1-arrested mouse pro-B–cell lines, dispensable for joining CSR-associated DSBs in a cycling mouse B-cell line, and dispensable for normal ionizing radiation resistance in both G1-arrested and cycling pro-B lines. However, we find that combined deficiency for PAXX and XLF in G1-arrested pro-B lines abrogates DSB joining during V(D)J recombination and sensitizes the cells to ionizing radiation exposure. Thus, PAXX provides core C-NHEJ factor-associated functions in the absence of XLF and vice versa in G1-arrested pro–B-cell lines. Finally, we also find that PAXX deficiency has no impact on V(D)J recombination DSB joining in ATM-deficient pro-B lines. We discuss implications of these findings with respect to potential PAXX and XLF functions in C-NHEJ.

To repair DNA double-strand breaks (DSBs), mammalian cells use two major DSB repair pathways: homologous recombination (HR) and classical nonhomologous end joining (C-NHEJ). HR requires a long homologous template for repair and is active in S/G2 cell cycle phase (1). In contrast, C-NHEJ, which directly ligates DSBs with short (1–4 bp) or no homologies, functions throughout interphase, including during the G1 cell cycle phase (2–4). C-NHEJ repairs diverse DSBs, including those arising ectopically by ionizing radiation as well as DSBs generated as intermediates during V(D)J recombination in developing progenitor (pro)-B and -T lymphocytes and Ig heavy chain (IgH) class switch recombination (CSR) in activated mature B lymphocytes (5). The C-NHEJ pathway has a number of characterized members, including “core” C-NHEJ factors that were so-named in part based on their requirement for joining known types of ends via C-NHEJ and their evolutionary conservation (6). Such core factors include the Ku70 and Ku80 proteins, which form the “Ku” end recognition complex, and the XRCC4 and DNA Ligase4 proteins, which form the C-NHEJ ligation complex (5, 6).

Exons that encode Ig and T-cell receptor (TCR) variable region exons are assembled from germline V, D, and J gene segments. This V(D)J recombination process, which takes place in pro-B and -T lymphocytes, occurs within the G1 cell cycle phase (7). Each V(D)J gene segment is flanked by a recombination signal sequence (RSS) that is a target of the RAG1/2 (RAG) endonuclease. RAG generates DSBs precisely between the RSS and flanking coding gene segment to create hair-pinned coding ends (CEs) and blunt, 5′-phosphorylated RSS ends (also termed signal ends) (8). Subsequently, the joining of both CEs and signal ends during V(D)J recombination is absolutely dependent on core C-NHEJ factors (5). Mice lacking any core C-NHEJ component cannot assemble antigen receptor variable region exons and correspondingly, have SCID that manifests as complete lack of mature B and T cells (6).

Activated cycling mature B cells use CSR to exchange their initially expressed IgM heavy-chain constant (Cμ) region exons for a set of downstream exons (e.g., Cγ, Cε, Cα, etc.) that encodes a different IgH Cμ region and antibody effector functions. CSR is a cut and paste process that is initiated specifically by activation-induced cytidine deaminase (AID) (9). In this regard, the various sets of CH exons are each flanked upstream by long repetitive switch (S) regions. During CSR, AID introduces deamination lesions into Sμ and a targeted downstream acceptor S region. Subsequently, these S region deamination lesions are converted into DSBs that are end joined to fuse Sμ and a downstream S region to complete CSR (5). Notably, whereas core C-NHEJ likely contributes substantially to end joining during CSR, in their absence, this reaction can be mediated at nearly 50% of WT levels by alternative end joining (A-EJ) pathways. A-EJ tends to more frequently use microhomologies (MHs) than C-NHEJ during CSR (10). A-EJ also substantially contributes to joining other types of DSBs in core C-NHEJ–deficient cycling cells (11, 12).

There are various C-NHEJ factors that are not required as broadly as core factors. In this regard, absence of either DNA-dependent protein kinase catalytic subunit (DNA-PKcs) or Artemis abrogates V(D)J CE joining, at least in part because of the role of these factors in hairpin opening and processing, but has much less impact on signal end joining (13). Functional redundancies with other factors can also impact on the requirement for certain C-NHEJ factors with respect to joining various classes of DSBs (6). For example, XLF deficiency has no measurable impact on chromosomal V(D)J recombination (14, 15) because of functional redundancy with the ataxia telangiectasia-mutated (ATM) DNA DSB response (DSBR) protein (6). Thus, although ATM deficiency only mildly impacts V(D)J recombination, this process is abrogated in developing pro-B cells dually deficient for XLF and ATM or downstream DSB response factors (16–18). XLF also is functionally redundant with DNA-PKcs in V(D)J recombination signal end joining (19). Potential processes in which XLF and DSBR factors may be functionally redundant are not well-characterized but may include tethering ends or facilitating their joining (6, 16). Notably, XLF also has functional redundancy with a truncation mutant of RAG2 for CE joining during V(D)J recombination, potentially implicating the RAG2 protein in some aspect of shepherding the V(D)J recombination joining reaction specifically to C-NHEJ (20, 21).

The paralogue of XRCC4 and XLF (PAXX; also known as c9ORF142 and XRCC4-like small protein) recently has been implicated as a C-NHEJ factor based on its structural similarity to XRCC4 and XLF (22–24). In this regard, PAXX deficiency conferred a range of ionizing radiation sensitivity in various human or chicken cell lines. In addition, although XLF deficiency modestly impacts V(D)J joining in extrachromosomal substrates in nonlymphoid cells (14), PAXX deficiency has been found to accentuate the requirement for XLF for this process (25). To further elucidate PAXX function in C-NHEJ, we now have assayed for potential unique roles of PAXX and potential functionally redundant roles of PAXX with XLF.

Results

PAXX Is Dispensable for End Joining During V(D)J Recombination.

To elucidate PAXX functions in C-NHEJ during V(D)J recombination, we used CRISPR-Cas9 to delete the entire ORF of murine PAXX in a previously characterized WT bcl-2 transgenic v-Abl kinase-transformed pro–B-cell line (16) (hereafter referred to as v-Abl cells) (Fig. S1 A and B) and assayed two independent clones of PAXX^−/− v-Abl cells. Treatment of v-Abl lines with Abl kinase inhibitor STI-571 leads to G1 arrest, induction of RAG1/RAG2 protein expression, and V(D)J recombination at endogenous RAG target loci as well as chromosomally integrated reporter substrates. The bcl-2 transgene circumvents STI-571–induced apoptosis to allow analysis of induced V(D)J recombination (26). Using the same Southern blot probe, coding joins (CJs) and unrepaired CEs can be measured in cells containing either the pMX-DEL-CJ or pMX-INV substrates, whereas signal joins (SJs) and unrepaired signal ends can be measured in cells containing pMX-DEL-SJ substrates (26) (Fig. 1A and Fig. S2 A and C). Based on assays with these three chromosomally integrated substrates, as expected, STI-571–treated WT v-Abl cells accumulated a substantial fraction of CJs and SJs with no detectable CEs and signal ends, whereas STI-571–treated XLF^−/−ATM^−/− and Ligase4^−/− v-Abl cells accumulated unrepaired CEs and signal ends in the absence of readily detectable CJs or SJs. In both PAXX^−/− clones, assays of both types of integrated CJ/CE substrates and the SJ/signal end substrate revealed WT levels of CJ and SJ formation with no detectable evidence of CEs or signal ends (Fig. 1B and Fig. S2 B, D, and E). We conclude that PAXX is dispensable for repair of CEs and signal ends during V(D)J recombination within chromosomal V(D)J recombination substrates.

Fig. S1.

Targeted deletions and complementation for v-Abl cell lines. Two independent PAXX^−/− clones were generated from WT, XLF^−/−, and ATM^−/− parental v-Abl lines. (A) Southern blot analysis of NcoI-digested genomic DNA for WT, PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, ATM^−/−, and ATM^−/−PAXX^−/− v-Abl cells. The 5′PAXX probe detects germline (2.9 kb) and the targeted deletion (1.5 kb), which deletes a portion of the 5′ UTR and the entire coding region of PAXX. (B) Western analysis of the same v-Abl lines described in A probing for total PAXX, ATM, XLF, and Ku80 (loading control) proteins. Oligos to generate Cas9/gRNAs, probe, and PCR for targeted deletion are listed in Table S2. (C and D) Western analysis showing complementation of (C) 3xFLAG-XLF and (D) 3xFLAG-PAXX in XLF^−/−PAXX^−/− v-Abl cells probed with XLF or PAXX antibodies, respectively, and Ligase1 antibody as a loading control. Long Exp., long exposure; Short Exp., short exposure.

Fig. 1.

PAXX deficiency does not affect coding joining during V(D)J recombination. (A) Overview of the pMX-DEL-CJ retroviral construct. The 12RSS and 23RSS are convergently oriented to monitor deletional CJs via Southern blotting assays. The CD4 probe (red dashed lines) combined with EcoRV-digested genomic DNA reveal germline unrearranged fragments (URs), CJs, and free CEs. LTR, viral long-terminal repeat; RV, EcoRV; SE, signal end. (B) Southern blot analysis of CJs and CEs for WT, PAXX^−/−, and Ligase4^−/− v-Abl cells. Numbers 1 and 2 indicate independent clones assayed for a given genotype.

Fig. S2.

PAXX is dispensable for generation of SJs formed by deletion or CJs formed by inversion. (A) Schematic of the retroviral pMX-DEL-SJ V(D)J recombination construct. The RAG 12RSS and 23RSS are divergently oriented to monitor chromosomal signal joining by Southern blot. (B) Southern blot analysis of SJs and signal ends from WT, PAXX^−/−, and Ligase4^−/− v-Abl cells. (C) Schematic of the pMX-INV V(D)J recombination construct. The 12RSS and 23RSS are separated by a spacer and present in the same orientation, which enforces recombination by inversion. In this context, the CD4 probe can detect CEs and deletional hybrid joins (HJs; aberrant joins between signal ends and CEs) with EcoRV-digested DNA and inversional CJs as well as deletional HJs with an EcoRV/NcoI digest. (D) Southern blot analysis of EcoRV-digested DNA to detect the presence of free CEs or HJs from WT, PAXX^−/−, and Ligase4^−/− v-Abl cells. (E) Southern blot analysis of EcoRV/NcoI-digested DNA to reveal inversional CJs and deletional HJs. Numbers 1 and 2 indicate independent clones assayed for a given genotype. LTR, viral long-terminal repeat; N, NcoI; RV, EcoRV; SE, signal end; UR, unrearranged fragment.

PAXX Contributes to Core C-NHEJ During V(D)J Recombination in the Absence of XLF but Not in the Absence of ATM.

The lack of detectable impact of PAXX deficiency on V(D)J recombination indicates either that PAXX is dispensable for this process or that PAXX functions can be compensated for by other factors. Given the structural similarity shared between PAXX, XRCC4, and XLF (22, 23), we hypothesized that potential functions of PAXX in V(D)J recombination might be masked by functional redundancies with XLF or the ATM-dependent DSBR. To test this hypothesis, we generated two separate clones of XLF^−/−PAXX^−/− and two clones of ATM^−/−PAXX^−/− v-Abl cells from XLF^−/− and ATM^−/− parental v-Abl lines (Fig. S1 A and B) and performed V(D)J recombination substrate analyses as described above for PAXX^−/− v-Abl lines. Consistent with prior observations (26), ATM^−/− v-Abl cells accumulated CJs and a fraction of unrepaired CEs with both tested substrates and generated SJs without detectable unrepaired signal ends (Fig. 2 A and B and Fig. S3). Likewise, both clones of ATM^−/−PAXX^−/− v-Abl cells accumulated CJs, CEs, and SJs similarly to ATM^−/− cells (Fig. 2A and Fig. S3). Thus, there seems to be little or no functional overlap between ATM and PAXX with respect to V(D)J recombination.

Fig. 2.

XLF^−/−PAXX^−/− but not ATM^−/−PAXX^−/− v-Abl cells are abrogated for coding and signal end joining block during V(D)J recombination. Southern analysis of CJs and SJs from (A) pMX-DEL-CJ or (B) pMX-DEL-SJ containing ATM^−/−, ATM^−/−PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, and Ligase4^−/− v-Abl cells. Southern analysis of the (C) pMX-DEL-CJ or (D) pMX-DEL-SJ recombination substrates in XLF^−/−PAXX^−/− v-Abl clone 1 complemented with empty vector, XLF, or PAXX compared with XLF^−/− and PAXX^−/− 1 v-Abl lines transduced with empty vector. Complementation results for XLF^−/−PAXX^−/− 2 are shown in Fig. S4. SE, signal end; UR, unrearranged fragment.

Fig. S3.

PAXX deficiency abrogates CJ formation during inversional recombination in XLF^−/− but not in ATM^−/− v-Abl cells. Southern blot analysis of the pMX-INV V(D)J construct as described in Fig. S2C for ATM^−/−, ATM^−/−PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, XLF^−/−ATM^−/−, and Ligase4^−/− v-Abl cells using (A) EcoRV-digested DNA or (B) EcoRV/NcoI-digested DNA. Hybrid joining does not appear to be dramatically altered in ATM^−/−PAXX^−/− v-Abl cells compared with in the parental ATM^−/− v-Abl line. Numbers 1 and 2 indicate independent clones assayed for a given genotype. Fig. 1 and Fig. S2 have abbreviations and schematics for hybrid joins (HJs), CJs, and free CEs. UR, unrearranged fragment.

As expected (14, 20), XLF^−/− v-Abl cells assayed with both CJ substrates showed a largely normal accumulation of CJs, with very minor levels of unrepaired CEs (Fig. 2 A and B and Fig. S3), and when assayed for SJs, they also showed profiles that appear largely normal (Fig. 2B). Strikingly, however, both clones of XLF^−/−PAXX^−/− v-Abl cells as assayed with both CJ constructs and the SJ construct showed a complete absence of CJs and SJs and a corresponding accumulation of CEs and signal ends (Fig. 2 A and B and Fig. S3). Indeed, the defect in V(D)J recombination CE and signal end joining appeared as severe as that of similarly treated XLF^−/−ATM^−/− or Ligase4^−/− v-Abl lines (Fig. 2 A and B and Fig. S3). To confirm that the inactivation of PAXX in XLF^−/− v-Abl cells was, indeed, the cause of this dramatic end joining defect, we retrovirally complemented both of the XLF^−/−PAXX^−/− clones with either PAXX or XLF and found that ectopic expression of either of these factors was sufficient to restore both CE and signal end joining to XLF^−/− or PAXX^−/− levels, respectively (Fig. 2 C and D and Fig. S4). These findings indicate that PAXX and XLF serve functionally redundant roles in core C-NHEJ during V(D)J recombination.

Fig. S4.

Ectopic expression of either XLF or PAXX restores coding and signal joining in a second XLF^−/−PAXX^−/− clone. In parallel experiments with XLF^−/−PAXX^−/− clone 1, XLF^−/−PAXX^−/− v-Abl clone 2 was complemented with empty vector, XLF, or PAXX and assayed for joining with the pMX-DEL-CJ or pMX-DEL-SJ cassettes as shown in Fig. 2 C and D. The samples displayed here were run and analyzed from the same gel as the samples shown in Fig. 2 C and D. SE, signal end; UR, unrearranged fragment.

PAXX Is Dispensable for CSR C-NHEJ in the Presence or Absence of XLF.

To determine whether PAXX deficiency, either alone or in the absence of XLF, affects end joining of DNA DSBs beyond those associated with V(D)J recombination, we tested the impact of PAXX deficiency on IgH CSR. The CH12F3 (CH12) B-cell lymphoma cell line can be induced in culture to undergo robust CSR to IgA by treatment with a combination of anti-CD40, IL-4, and TGF-β (27). For these analyses, we used a Cas9/guide-RNA (gRNA) approach to generate independent lines of PAXX^−/−, XLF^−/−, and XLF^−/−PAXX^−/− CH12 cells (Fig. S5 A and B). PAXX^−/− CH12 cells underwent CSR to IgA at levels comparable with those of the parental WT CH12 line (average: 55 and 65% for PAXX^−/− vs. 50 and 62% for WT cells on days 2 and 3 of stimulation, respectively) (Fig. 3A and Fig. S6). Consistent with studies of CSR to IgG1 in XLF^−/− primary B cells (14), XLF^−/− CH12 cells underwent CSR to IgA at about 32 and 39% of WT levels on days 2 and 3 of stimulation, respectively. This level of IgH class switching was very similar to that of CH12 cells deficient for the core C-NHEJ factor Ligase4 (Fig. 3A and Figs. S5 C and D and S6). In the latter context, XLF^−/−PAXX^−/− CH12 cells, perhaps not surprisingly, showed no obvious impairment in CSR to IgA beyond that observed for XLF^−/− CH12 cells (Fig. 3A and Fig. S6).

Fig. S5.

Targeted deletions for CH12 cell lines. (A and B) Two independent PAXX^−/− and XLF^−/− clones were generated from WT CH12 cells; two independent XLF^−/−PAXX^−/− CH12 clones were generated from PAXX^−/− clone 2. (A) Southern blot analysis of NcoI-digested genomic DNA for WT, PAXX^−/−, XLF^−/−, and XLF^−/−PAXX^−/− CH12 cells using (Left) the XLF probe or (Right) the PAXX probe. Fig. S1 shows PAXX genotyping details. The XLF probe detects germline (5.0 kb) and the targeted deletion [3.3 kb; exons 4 and 5 of XLF (14)]. (B) Western blot analysis to confirm loss of XLF and PAXX proteins from the cell lines described in A. Antibody against Ku80 was used as a loading control. Western blot analysis was performed with protein lysates from steady-state cultures. (C) Southern blot analysis revealing deletion of the Ligase4 coding region from a WT CH12 clone. Strategy and predicted sizes are indicated. (D) Western analysis to verify loss of Ligase4 protein. Antibody against actin is used as a loading control. Oligos to generate Cas9/gRNAs, probe, and PCR for targeted deletion are listed in Table S2.

Fig. 3.

PAXX deficiency does not affect end joining during CSR. (A) IgA staining quantified by flow cytometry for WT, PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, and Ligase4^−/− CH12 B cells stimulated for CSR on day 2. Fig. S6A shows CSR analysis on day 3. (B) Metaphase FISH analysis for IgH locus breaks of stimulated WT, PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, and Ligase4^−/− CH12 cells from day 2 stimulated CH12 cells; 50–60 metaphases were analyzed for each replicate across all clones. (C) Junction structure distributions between Sµ and Sα for WT (red square), PAXX^−/− (blue triangle), XLF^−/− (orange diamond), XLF^−/−PAXX^−/− (light blue downward triangle), and Ligase4^−/− (black circle) CH12 cells as determined by linear amplification-mediated high-throughput genome-wide translocation sequencing (LAM-HTGTS) assay (30) from day 3 stimulated CH12 cells (n = 3 for each clone). Table S1 shows individual clone distributions. (D) Mean MH lengths are reported for WT, PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, and Ligase4^−/− CH12 cells. (A, B, and D) All graphs are displayed as means ± SD. Clones 1 and 2 are indicated for each genotype by white and purple circles, respectively; n ≥ 3 independent stimulations for each clone.

Fig. S6.

Day 3 CSR levels and representative CSR FACS plots of CH12 cell lines used in this study. (A) IgA staining quantified by flow cytometry for WT, PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, and Ligase4^−/− CH12 B cells stimulated for CSR on day 3. Clones 1 and 2 are indicated for each genotype by white and purple circles, respectively; n ≥ 3 independent stimulations for each clone. (B) Representative FACS plots of CH12 cells stimulated for CSR and assayed for IgA on days 2 and 3 of stimulation. y axis, forward scatter height (FSC-H); x axis, IgA.

Minor effects of C-NHEJ deficiencies on IgH CSR cannot, in some cases, be readily apparent to assays that measure IgH class switching by surface staining. For example, Artemis deficiency has little or no impact on IgM to IgG1 switching relative to WT as measured by surface staining but does have an impact on CSR that can be observed via an FISH assay to quantify levels of unrepaired IgH CH locus breaks. In addition, this assay directly indicates end joining defects as opposed to the surface staining assay that measures impacts on CSR that could occur at various levels (28, 29). Thus, we used IgH FISH to look for IgH chromosomal breaks (and translocations) in CSR-activated WT and mutant CH12 cells after 2 d of activation for IgA class switching. In these studies, the levels of chromosomal IgH breaks were not distinguishable between activated WT and PAXX^−/− CH12 cells (∼2% in both) (Fig. 3B). In contrast, CSR-activated XLF^−/− CH12 cells showed a marked increase in chromosomal breaks that was not significantly increased in XLF^−/−PAXX^−/− CH12 cells (about 10% in both genotypes) (Fig. 3B). Notably, the level of chromosomal breaks in Ligase4^−/− CH12 cells (25%) (Fig. 3B) was much greater than that of XLF^−/− CH12 cells (Fig. 3B). Additional work will be needed to understand why relative levels of IgH breaks are greater in Ligase4^−/− CH12 cells relative to XLF^−/− CH12 cells given their very similarly reduced levels of IgA class switching. We note, however, that mechanisms for this type of divergence that do not necessarily relate to the CSR mechanism per se have been implicated in different CSR stimulation conditions and other factors (29, 30).

Prior studies of limited numbers of CSR junctions indicate that more than 90% of CSR joins in core C-NHEJ factor-deficient activated B cells are not direct and use MHs as opposed to those from WT B cells that generally have a much larger fraction of direct junctions (31–33). To assess the structure of a large number of Sμ to Sα CSR junctions from the various WT and mutant CH12 cells after 3 d of activation, we used our recently developed high-throughput CSR junction assay (30). Based on analysis of thousands of CSR junctions from WT and PAXX^−/− CH12 cells (Table S1), we observed no significant difference in use of junctional MHs, with nearly 25% of junctions being direct and an overall mean MH length of ∼1.8 bp in both cell types (Fig. 3 C and D and Table S1). Consistent with prior observations of MH bias in Sμ-Sγ1 junctions in XLF^−/− primary B cells (14), analysis of thousands of junctions from XLF^−/− CH12 cells revealed a profound MH bias, with only ∼7% of junctions using direct joins and an overall mean MH length of 2.8 bp (Fig. 3 C and D and Table S1). Indeed, the shift from direct to MH-mediated junctions was indistinguishable from that observed for Ligase4-deficient CH12 cells (Fig. 3A), consistent with the similar impact of XLF and this core C-NHEJ factor on IgH class switching as measured by the FACS surface staining assay. In this context, loss of PAXX did not impact the MH bias in either direction beyond the severe effect found for XLF deficiency, because analysis of thousands of CSR junctions from XLF^−/−PAXX^−/− CH12 cells showed ∼7% direct join use and a mean MH length of 2.8 bp (Fig. 3 C and D and Table S1).

Table S1.

MH distribution of Sα prey junctions cloned from Sμ bait DSBs in stimulated CH12 cells

Genotype and replicate	MH length											Junction total	Mean MH length
	Direct	1	2	3	4	5	6	7	8	9	10
WT
Exp. 1	195	183	149	100	66	32	4	0	3	0	2	734	1.73
Exp. 2	171	173	137	102	87	33	4	3	3	1	0	714	1.88
Exp. 3	182	167	139	113	70	41	5	1	4	2	0	724	1.87
PAXX−/− 1
Exp. 1	224	257	187	149	96	47	5	2	4	1	5	977	1.88
Exp. 2	334	322	276	225	120	58	9	3	4	1	3	1,355	1.82
Exp. 3	339	358	261	220	132	56	9	3	4	1	2	1,385	1.79
PAXX−/− 2
Exp. 1	143	148	126	99	60	17	3	3	0	1	1	601	1.80
Exp. 2	200	182	161	140	71	25	7	3	2	0	1	792	1.80
Exp. 3	208	185	148	153	84	27	6	3	3	1	0	818	1.83
XLF−/− 1
Exp. 1	68	122	223	209	163	79	12	8	3	2	2	891	2.73
Exp. 2	39	70	159	138	116	46	5	4	1	0	2	580	2.73
Exp. 3	45	78	161	136	114	62	9	3	4	0	4	616	2.79
XLF−/− 2
Exp. 1	20	44	100	97	62	37	7	2	1	1	0	371	2.81
Exp. 2	30	44	102	88	57	24	7	4	2	0	0	358	2.64
Exp. 3	24	38	83	76	65	37	6	2	0	0	3	334	2.87
XLF−/−PAXX−/− 1
Exp. 1	42	70	193	171	142	71	10	6	1	1	0	707	2.84
Exp. 2	54	79	145	108	100	49	10	7	2	1	2	557	2.68
Exp. 3	63	99	201	167	144	87	19	10	4	0	1	795	2.82
XLF−/−PAXX−/− 2
Exp. 1	26	60	124	100	81	45	8	3	2	0	1	450	2.78
Exp. 2	37	85	161	174	120	75	13	6	5	0	1	677	2.89
Exp. 3	49	88	190	174	147	74	13	7	3	1	1	747	2.84
Ligase4−/− 1
Exp. 1	61	100	198	202	161	85	15	8	5	0	0	835	2.82
Exp. 2	97	204	403	331	247	112	15	7	9	2	4	1,431	2.67
Exp. 3	107	221	408	371	273	139	16	9	8	2	6	1,560	2.72
Ligase4−/− 2
Exp. 1	25	50	78	79	71	36	6	3	0	0	2	350	2.81
Exp. 2	87	91	111	115	96	42	14	1	0	1	1	559	2.44
Exp. 3	31	41	89	112	90	49	11	5	3	0	2	433	3.01

LAM-HTGTS–generated junctions from Sμ bait DSBs are enumerated for each replicate experiment in a given genotype CH12 clone. Sα prey junction totals encompass direct joins and 1- to 10-bp MHs and were used to calculate mean MH length.

PAXX and XLF Are Functionally Redundant for Repair of Ionizing Radiation-Induced DSBs in G1 Phase-Enriched v-Abl Lines.

To test for potential roles of PAXX, in either the presence or the absence of XLF, in the repair of more general DSBs, we assayed cycling WT and the various mutant CH12 cells for ionizing radiation sensitivity. These assays revealed that PAXX^−/− CH12 cells did not detectably exhibit ionizing radiation sensitivity beyond that of WT CH12 cells (Fig. 4A), indicating that PAXX is dispensable for the repair of this general class of DSBs in CH12 cells. However, XLF^−/− cycling CH12 cells revealed ionizing radiation sensitivity that was clearly greater than that of WT but in contrast to their CSR defect, not as severe as that of Ligase4^−/− CH12 cells (Fig. 4A). Moreover, with respect to ionizing radiation sensitivity, XLF^−/−PAXX^−/− CH12 cells showed no greater sensitivity to ionizing radiation than that of XLF^−/− cells alone (Fig. 4A). We conclude that PAXX has no major functional redundancy with XLF with respect to sensitivity to ionizing radiation-induced DNA damage in cycling CH12 cells.

Fig. 4.

PAXX provides G1 phase-specific compensation for XLF in cellular resistance to ionizing radiation. Asynchronous CH12 and v-Abl lines and G1-arrested v-Abl cell lines were subjected to different doses of ionizing radiation and serially diluted to assess clonogenic survival (Materials and Methods). In some cases, clonal growth was below the detection of our assay and thus, is not plotted. (A) Asynchronous WT (n = 5; red), PAXX^−/− (n = 10; two clones; blue), XLF^−/− (n = 8; two clones; orange), XLF^−/−PAXX^−/− (n = 8; two clones; light blue), and Ligase4^−/− (n = 5; purple) CH12 cells. (B) Asynchronous v-Abl cell genotypes are indicated: WT (n = 7; red), PAXX^−/− (n = 8; two clones; blue), XLF^−/− (n = 4; orange), XLF^−/−PAXX^−/− (n = 8; two clones; light blue), and Ku70^−/− (n = 7; purple) v-Abl cells. (C) STI-571 G1-arrested genotypes are indicated: WT (n = 7), PAXX^−/− (n = 9; two clones), XLF^−/− (n = 4), XLF^−/−PAXX^−/− (n = 8; two clones), and Ku70^−/− (n = 8). Retrovirally complemented 3xFLAG-XLF (n = 5; blue dashed line) or 3xFLAG-PAXX (n = 5; orange dashed line) in XLF^−/−PAXX^−/− and STI-571 G1-arrested v-Abl cells is indicated. (D) PD0332991 G1-arrested pro–B-cell genotypes are indicated: WT (n = 3), PAXX^−/− (n = 3), XLF^−/− (n = 3), XLF^−/−PAXX^−/− (n = 3), and Ku70^−/− (n = 3). Mean ± SEM plotted for each ionizing radiation dose. IR, ionizing radiation.

Our finding that the functional redundancy of PAXX and XLF with respect to V(D)J recombination in G1-phase v-Abl lines does not extend to ionizing radiation sensitivity in cycling CH12 cells might reflect cell type and stage differences. To test this possibility, we used clonogenic survival assays to test ionizing radiation sensitivity of the various WT and mutant v-Abl transformed lines that were ionizing radiation treated while cycling. As expected for core C-NHEJ deficiency (14), the survival of cycling Ku70^−/− v-Abl cells was profoundly inhibited by ionizing radiation treatment compared with that of ionizing radiation-treated WT cells (Fig. 4B). In contrast, cycling PAXX^−/− v-Abl cells exhibited no detectable ionizing radiation sensitivity compared with WT, a result consistent with the dispensability of PAXX for V(D)J recombination, CSR, and ionizing radiation resistance in cycling CH12 cells. As also expected (14), cycling XLF^−/− v-Abl cells exhibited intermediate ionizing radiation sensitivity relative to WT and Ku70^−/− v-Abl cells (Fig. 4B). However, cycling XLF^−/−PAXX^−/− cells did not exhibit ionizing radiation sensitivity beyond that of XLF^−/− cells (Fig. 4B). We conclude that neither cycling CH12 cells nor cycling v-Abl cells show functional redundancy of PAXX and XLF for repair of ionizing radiation-induced DSBs.

The are several possible explanations for the apparent discrepancy between the functional redundancy observed for PAXX and XLF in V(D)J recombination in G1-arrested v-Abl cells and the absence of functional redundancy in mediating ionizing radiation resistance in cycling v-Abl and CH12 cells. One possibility would be that this functional redundancy only occurs in the context of DSB repair during V(D)J recombination. Alternatively, the functional redundancy of these two factors might apply more broadly to DSB repair but would only be manifested for DSBs that are generated in the G1 cell cycle phase. To distinguish between these two possibilities, we treated v-Abl lines with ionizing radiation, when they were G1 arrested via STI-571 treatment (Fig. S7), and then, we subsequently released them into cycle to perform clonogenic survival assays (Materials and Methods). Based on this assay, G1-arrested Ku70^−/− cells were extremely sensitive to ionizing radiation treatment compared with WT v-Abl cells, whereas G1-arrested PAXX^−/− v-Abl cells had ionizing radiation sensitivity indistinguishable from that of WT (Fig. 4C). G1-arrested XLF^−/− v-Abl cells had an intermediate level of ionizing radiation sensitivity compared with WT and Ku70^−/− counterparts, but strikingly, G1-arrested XLF^−/−PAXX^−/− v-Abl cells also were extremely ionizing radiation sensitive relative to XLF^−/− or PAXX^−/− G1-arrested v-Abl cells (Fig. 4C). However, we could not directly compare the ionizing radiation sensitivity between Ku70^−/− and XLF^−/−PAXX^−/− v-Abl cells, because in both cases, clonogenic survival under these conditions was beyond the sensitivity of our current assays (Fig. 4C). Finally, we note that the extreme ionizing radiation hypersensitivity of XLF^−/−PAXX^−/− v-Abl cells revealed by these experiments was rescued to levels observed in PAXX^−/− or XLF^−/− cells by complementation via ectopic expression of either XLF or PAXX, respectively (Fig. 4C and Fig. S1 C and D). These findings indicate that PAXX and XLF have functional redundancy for repair of ionizing radiation-induced DSBs generated in G1-phase v-Abl cells.

Fig. S7.

G1 enrichment profiles in the presence of STI-571 or PD0332991. WT, PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, and Ku70^−/− v-Abl cells were treated with STI-571 for 60 h or PD0332991 for 24 h to induced G1 arrest and compared against their asynchronous counterparts. Nuclear content (using 7AAD) and nascent nucleotide incorporation were measured by flow cytometry (Materials and Methods). G1-phase populations were identified by nuclear content (2N) and no BrdU incorporation, whereas S-phase populations were identified by BrdU incorporation.

We considered the possibility that the functional redundancy that we observed in STI-571–treated v-Abl cells might be an effect related to inhibition of v-Abl as opposed to G1 arrest per se. To test this notion, we used the cyclin-dependent kinase 4,6 inhibitor PD0332991 as an independent means of G1 arrest (34) (Fig. S7). We then assessed clonogenic survival of irradiated WT, PAXX^−/−, XLF^−/−, XLF^−/−PAXX^−/−, and Ku70^−/− v-Abl cells that were G1 synchronized with PD0332991 and subsequently released from G1 arrest after irradiation. Similar to observations in STI-571–treated Ku70^−/− v-Abl cells, PD0332991 treatment of Ku70^−/− v-Abl cells also induced a profound hypersensitivity under these conditions (Fig. 4D). Moreover, similar to observations with STI-571 treatment, PD0332991-treated XLF^−/−PAXX^−/− cells exhibited a marked hypersensitivity relative to similarly treated XLF^−/− cells (Fig. 4D). Thus, using two independent methods of G1 arrest, our results indicate that the observed functional redundancy for PAXX and XLF is, indeed, because of a G1-phase enrichment. We conclude that the functional redundancy between XLF and PAXX in v-Abl lines extends to the joining of more general DNA DSBs but is only manifested in G1 phase-enriched cells.

Given the results from V(D)J recombination and ionizing radiation sensitivity assays, we tested whether our findings might result from differential expression levels of PAXX in cycling vs. G1-arrested v-Abl cells. For this purpose, we used Western blotting to assay relative PAXX protein levels in cycling WT v-Abl cells vs. WT v-Abl cells arrested in G1 via treatment with PD0332991 (Fig. S8). These analyses revealed no marked differences in relative PAXX protein levels between the two populations. In this regard, although differential PAXX proteins levels do not seem to explain the increased ionizing radiation sensitivity of G1-arrested vs. cycling cells, PAXX posttranslational modifications and subcellular localization are among many other conceivable mechanisms that could contribute to explaining this phenomenon.

Fig. S8.

PAXX protein levels in G1-phase v-Abl cells are similar to their cycling counterparts. Western analysis of WT cycling, PD0332991 [cyclin-dependent kinase 4,6 inhibitor (CDKi)]-treated, and PAXX^−/− cycling v-Abl cell lysates using antibodies recognizing (Top) PAXX and (Middle) cyclin A2. Cyclin A2 is expressed in cycling but not in G1-arrested cells and serves as a control for these populations. For these experiments, 10E6 cells were lysed in a total volume of 200 μL 1× lysis buffer (Materials and Methods) and serially diluted 1:1 in 1× lysis buffer before loading. Percentage input is indicated above each lane. (Bottom) Ponceau S staining of the entire membrane is shown as a loading control for total protein content in each sample.

Discussion

We find that PAXX is dispensable for repair of V(D)J recombination-associated DSBs in G1-enriched v-Abl cells, repair of IgH CSR DSBs in cycling CH12 cells, and mediating normal resistance to ionizing radiation in cycling cells. However, we further find that PAXX is strictly required in the absence of XLF to join V(D)J recombination-associated DSBs in G1-enriched v-Abl cells. Likewise, PAXX deficiency greatly increases the ionizing radiation sensitivity of XLF-deficient G1-enriched v-Abl cells. These findings provide clear evidence to strongly support the prior conclusion that PAXX functions as a C-NHEJ factor (22, 23) and further show that PAXX is capable of mediating core C-NHEJ functions. Notably, the synthetic defects associated with combined PAXX and XLF deficiencies are manifested only in the context of DSBs generated in XLF^−/−PAXX^−/−, G1 phase-enriched v-Abl pro-B cells. Although more work will be required to understand why the synthetic core C-NHEJ function conferred by XLF and PAXX has, thus far, manifested only in G1-arrested v-Abl pro-B cells, the differential expression of compensatory activities in cycling vs. G1 cells is one possibility. We do note that the ionizing radiation sensitivity phenotype of the XLF^−/−PAXX^−/−, G1 phase-enriched v-Abl pro-B cells was observed after these cells were released from their G1 block. In this regard, we speculate that, in the absence of PAXX and XLF, ionizing radiation treatment leads to high levels of unrepaired genomic DSBs in the G1-arrested cells that would normally be repaired by core C-NHEJ and that such DSBs may overwhelm the G1 DSB checkpoint that would likely be engaged on release (35), resulting in widespread apoptotic death of released cells.

No IgH chromosome breaks accumulated in activated PAXX^−/− CSR-stimulated CH12 cells, showing that PAXX is not required for end joining during CSR. In contrast, substantially increased levels of such breaks accumulated in XLF^−/− CH12 cells, confirming the requirement for XLF in CSR end joining (14). In this regard, we find that XLF deficiency in CH12 cells has a similar impact to Ligase4 deficiency or other core C-NHEJ factor deficiencies (31, 32) on CSR with respect to both the overall decrease in class switching levels and the greatly increased MH use in CSR junctions between different S regions. In the latter context, it has been noted that S regions may be better substrates for such highly MH-biased joining because of their highly repetitive structure and being resection-prone (30). Overall, based on the two types of assays outlined above, XLF seems to provide functions associated with core C-NHEJ factors, at least in some aspects of CSR. However, we note the potential caveat in this interpretation, discussed above, regarding the divergence in relative levels of IgH chromosome breaks that accumulate in CSR-activated XLF^−/− vs. Ligase4^−/− CH12 cells. Finally, it is notable that XLF deficiency alone does not have as severe an impact on other processes that use DSB repair via C-NHEJ [e.g., V(D)J recombination and ionizing radiation resistance] as it does on IgH class switching. In this regard, compensation for XLF DSB repair functions by PAXX, ATM, and even factors, such as RAG2, that are not considered DSB repair factors clearly obviates a strict requirement for XLF for DSB joining in some of these contexts (16, 20) Whatever the case, it is notable that PAXX is not functionally redundant with ATM for V(D)J recombination. In this regard, PAXX and ATM may have distinct overlapping activities with respect to XLF functions, or alternatively, PAXX may operate downstream of ATM with respect to V(D)J recombination.

Prior studies have shown that PAXX is required for repair of ionizing radiation-induced DSBs in human and chicken cells (22, 23). In this context, our findings that the PAXX-deficient lymphoid cell lines that we studied are not detectably impaired for V(D)J recombination and IgH CSR and also, when cycling, do not show increased ionizing radiation sensitivity were unexpected. However, such apparently divergent findings may reflect differences in the types of assays used and/or differences in the contribution of PAXX to particular joining pathways in the cell types analyzed. As one example of the latter possibility, XLF-deficient mouse embryonic fibroblasts (MEFs) and ES cells have a markedly greater defect in end joining of V(D)J recombination substrates compared with WT MEFs or ES cells when assayed with ectopically introduced episomal V(D)J recombination substrates (14). Likewise, in XLF-deficient v-Abl pro-B cells, there is a greater defect in V(D)J joining within chromosomally integrated as opposed to episomal V(D)J recombination substrates (14). In any case, based on our current findings, we would predict that PAXX^−/− mice will be competent for V(D)J recombination and CSR and that XLF^−/−PAXX^−/− mice will likely have an SCID phenotype because of inability to join V(D)J recombination-associated DSBs. Finally, it is also conceivable that variations in PAXX activity might contribute to the variable immunodeficiency phenotype observed in XLF-deficient humans (6, 14, 16).

Materials and Methods

Chromosomal V(D)J Recombination Assays.

v-Abl lines (details on gRNAs used for line generation are in Table S2) containing the pMX-DEL-CJ, pMX-DEL-SJ, or pMX-INV cassettes (26) underwent V(D)J recombination and are described in SI Materials and Methods.

Table S2.

Oligos used in this study

Name	Sequence 5′→3′	Purpose
gRNAs
5′PAXX_F	CAC CGC AGG AGC CAA GTT AGA TGT	BbsI digest from pX330; PAM: GGG; pair with 3’PAXX gRNA
5′PAXX_R	AAA CAC ATC TAA CTT GGC TCC TGC
3′PAXX_F	CAC CGC AAA GCC TGT TAC TTG GCC	BbsI digest from pX330; PAM: AGG; pair with 5’PAXX gRNA
3′PAXX_R	AAA CGG CCA AGT AAC AGG CTT TGC
5′XLF_F	CAC CGC CTT AGC AGA GAT GTT AAG T	BbsI digest from pX330; PAM: AGG; pair with 3’XLF gRNA
5′XLF_R	AAA CAC TTA ACA TCT CTG CTA AGG C
3′XLF_F	CAC CGT TGT GGT CCA GAT ATG GAA T	BbsI digest from pX330; PAM: GGG; pair with 5’XLF gRNA
3′XLF_R	AAA CAT TCC ATA TCT GGA CCA CAA C
5′LIG4_F	CAC CGC TTT ATC AGT TCA AAC CGG	BbsI digest from pX330; PAM: AGG; Pair with 3’LIG4 gRNA
5′LIG4_R	AAA CCC GGT TTG AAC TGA TAA AGC
3′LIG4_F	CAC CGT ATT TGC TTT AGA GCT TGC	BbsI digest from pX330; PAM: TGG; Pair with 5’LIG4 gRNA
3′LIG4_R	AAA CGC AAG CTC TAA AGC AAA TAC
Genotype
PAXX_F	CAA CAG TTT GGA GGC CCT GC	PCR screen for PAXX; germline: 2.2 kb, deletion: 850 bp
PAXX_R	TGC TAC CTG GGT GGG GCT TG
PAXX_Fs	CAG CCT TGG GAA GAC TTC TG	513-bp PAXX probe for Southern; NcoI digest compatible
PAXX_Rs	CTT GTG CTC TTC CGC TTT TC
PAXX_GL_F	GAG CTC TGG AGC ACC TGC TT	PCR screen to detect PAXX germline sequence, 532 bp
PAXX_GL_R	TGC CAG CCG CCT CTC TAG GTT
XLF_F	GAG TCT GGA TAT GAG CGC TCA G	PCR screen for XLF; germline: 2.6 kb, deletion: 820 bp
XLF_R	GGT TGC AGC CTT AGA AAA GTG G
XLF_Fs	GAG TCT GGA TAT GAG CGC TCA G	687-bp XLF probe for Southern; NcoI digest compatible
XLF_Rs	AGC CTT GGG AAG GTC AGT TG
XLF_GL_F	CCT GCC TCT ACT TGA GTG CC	PCR screen to detect XLF germline sequence, 694-bp fragment
XLF_GL_R	GGT TGC AGC CTT AGA AAA GTG G
Ligase4_F	GTA TTC GGT GCT AAA TAG TC	PCR screen for Ligase4; germline: 3.3 kb, deletion: 536 bp
Ligase4_R	GCA CTT CTT GAA GCC ATA G
Ligase4_Fs	CTA TGG CTT CAA GAA GTG C	544-bp Ligase4 probe for Southern; NcoI digest compatible
Ligase4_Rs	AGC CAG AGC ACC CAG TCT CT
Ligase4_GL_F	CCC TCT GTA GGG CTT AGT GAC	PCR screen to detect Ligase4 germline sequence, 511-bp fragment
Ligase4_GL_R	CTA TGC GGC CTT GTA TGG GTG
Molecular cloning
PAXX_cDNA_F	ATG GCT CCT CCG TTG TCG	PAXX cloning from murine G1-arrested v-Abl cDNA pool
PAXX_cDNA_R	TCA GGT CTC ATC AAA GTC TAC ACC
XLF_cDNA_F	GCC CGG GCG GGA TCC ATG GAA GAG CTA GAG CAA GAC C	XLF cloning from murine cDNA pool
XLF_cDNA_R	TAA CCT CGA GAT GCG GAT CCT CAC ACG CGT TTA ACT GAA GAG TCC CCT GG
MSCV_MCS_F	CAT GGT CAG GCG GCC GCC ACT ACG CGT GTG AGG ATC CGC ATC	XhoI, BamHI, MluI, EagI, NcoI multiple cloning site
MSCV_MCS_R	TCG AGA TGC GGA TCC TCA CAC GCG TAG TGG CGC CCG CCT GAC
Gibson assembly
G_MSCV_3xFLAG_F	ATC CAT GGT CAG GCG GCC GCC ACT ACG CGT GCC ACC ATG GAT TAC AAG GAT G	30-bp Overlap with vector end MluI; kosak plus 3xFLAG; ∼100-bp fragment
G_3xFLAG_PAXX_R	CAA CGG AGG AGC CAT GGA TCC CGC CCG GGC TTT ATC	15 bp Each shared overlap; ∼100-bp fragment; BamHI
G_3xFLAG_PAXX_F	GCC CGG GCG GGA TCC ATG GCT CCT CCG TTG TTG TC	15 bp Each shared overlap; ∼650-bp fragment; BamHI
G_PAXX_MSCV_R	TAA CCT CGA GAT GCG GAT CCT CAC ACG CGT TCA GGT CTC ATC AAA GTC TAC ACC	30-bp Overlap with vector end MluI; ∼650-bp fragment
G_3xFLAG_XLF_F	GCC CGG GCG GGA TCC ATG GAA GAG CTA GAG CAA GAC C	15-bp Each shared overlap; ∼900-bp fragment; BamHI
G_XLF_MSCV_R	TAA CCT CGA GAT GCG GAT CCT CAC ACG CGT TTA ACT GAA GAG TCC CCT GG	30-bp Overlap with vector end MluI; ∼900-bp fragment

CH12 CSR, IgH Metaphase FISH, and LAM-HTGTS.

CH12 cells underwent CSR as previously described (36); IgH metaphase FISH has been previously described (28). Junction structure analysis of IgH S regions using LAM-HTGTS has been described previously (30). Additional details are provided in SI Materials and Methods.

Clonogenic Ionizing Radiation Sensitivity Assays.

Clonogenic survival of v-Abl and CH12 cells after ionizing radiation treatment is described in SI Materials and Methods.

SI Materials and Methods

Cell Lines and Culture Conditions.

v-Abl pro-B cells were cultured at 37 °C and 5% CO₂ in RPMI supplemented with 15% (vol/vol) FCS, 50 U/mL penicillin/streptomycin (ThermoFisher), 2 mM l-glutamine (ThermoFisher), 1× MEM-NEAA (ThermoFisher), 1 mM sodium pyruvate (ThermoFisher), 50 μM 2-mercaptoethanol (Sigma), and 20 mM Hepes (pH 7.4). The following cell lines were used to target PAXX (mRNA accession no. NM_153557; protein accession no. NP_705785.1) deletion: WT (clone 8,411) (16), ATM^−/− (clone 8,653) (26), and XLF^−/− ATM^flox/flox (clone 8,269) (16). Ku70^−/−LIG4^flox/flox mice were obtained through breeding, and v-Abl pro–B-cell lines were obtained as previously described (26). All of the aforementioned cell lines were derived from mice also containing the Eμ-Bcl2 transgene. The entire predicted ORF corresponding to the human homolog of PAXX was deleted using 5′PAXX and 3′PAXX gRNA pairs (Table S2). Deletions were PCR screened (Table S2) (60 °C annealing temperature; germline: 2.2 kb; deletion: 850 bp). Locus alteration and clonality were verified by Southern blot (Table S2). Cas9/gRNAs were nucleofected (2.5 μg each/3 × 10⁶ cells) using 4-D nucleofector, SF solution, and program DN-100 and cultured clonally 24 h later. In some instances, 0.5 µg pMAX-GFP (Lonza) was included in the nucleofection, and highly expressing GFP-positive cells were sorted before clonal plating. On day 8 of clonal culture, colonies were picked and transferred to new 96-well dishes containing fresh media; genomic DNA for each clone was isolated by resuspending each cell pellet (with ∼10 µL residual volume) in a 10 µL 2× DNA lysis solution containing 2% Tween-20, 0.2 mM EDTA (pH 8.0), 2× Taq PCR Buffer (Qiagen), and 1 mg/mL proteinase K followed by incubation at 56 °C for 2 h and 95 °C for 10 min. Genotyping was performed using 2 µL genomic DNA in a standard 10 μL PCR using Taq Polymerase and 1× Coralload PCR Buffer (Qiagen). Clones positive for deletion by PCR were expanded to further confirm deletion by Southern analysis (Table S2) and Western blot.

CH12F3 (CH12) cell lines were cultured as similarly described for v-Abl pro–B-cell lines but using only 10% (vol/vol) FCS. PAXX, XLF, and Ligase4 deletions were generated using the Cas9/gRNAs and screened (Table S2). Cas9/gRNAs were nucleofected as similarly described for v-Abl cells, except that the program CA-137 was used. Genomic DNA was isolated from individual clones as described above for v-Abl cells.

Vectors and cDNA Cloning.

RNA was isolated from G1-arrested Abl pro–B-cell RNA (Tripure Reagent; Roche). PAXX cDNA was amplified (Table S2) from the resulting cDNA pool (Invitrogen Superscript Plus II) and ligated into EcoRV-digested and CIP-treated pBluescript II (KS+). The pMSCV-puro plasmid containing the Ligase4 cDNA, which was a gift from Kefei Yu, Michigan State University, East Lansing, MI, was modified by releasing the insert with XhoI and NcoI and ligating a custom multiple cloning site (MSCV_MCS) (Table S2) to generate pMSCV-puroVK. The N-terminal 3XFLAG and PAXX cDNA were cloned into the MluI site of pMSCV-puroVK using Gibson assembly (37) (Table S2) to generate pMSCV-3xFLAG-PAXX-puroVK. The murine XLF cDNA was similarly cloned from a pool of cDNA (Table S2); the resulting cDNA was cut with BamHI and cloned into BamHI cut pMSCV-3xFLAG-PAXX-puroVK.

Retroviral Transduction and Western Analysis.

Retroviral supernatants for PAXX and XLF murine cDNAs were generated from calcium phosphate transfection of pMSCV-3xFLAG-PAXX-puroVK and pMSCV-3xFLAG-XLF-puroVK into Phoenix-Eco cells. v-Abl cells were infected at 2 × 10⁶ cells per 1 mL viral supernatant in the presence of 8 μg/mL polybrene and centrifuged at 1,250 × g for 1.5 h at room temperature before further culturing at 37 °C and 5% CO₂. After 48 h, cells were selected in 4 μg/mL puromycin. For Western analysis, cells were lysed by boiling/vortexing in 1:1 volumes of PBS and 2× sample buffer containing 0.125 M Tris⋅Cl (pH 6.8), 4% SDS, 20% glycerol, bromophenol blue, and 2-mercaptoethanol (freshly added to 10%) and loaded directly for SDS/PAGE analysis. Primary antibodies used were β-Actin (A2066l; Sigma), ATM (1:2,000; 2C1 1A1; Ab78; Abcam), Ku80 (1:500; 2753; Cell Signaling Technology), Ligase 1 (1:2,000; N-13; Santa Cruz), PAXX (1:500; ab126353; Abcam), anti-CyclinA2 (1:500; sc-596; Santa Cruz), and XLF (1:1,000; A300-730A; Bethyl). Ligase4 antibody (rabbit; 1:2,000) was a gift from David Schatz, Yale University, New Haven, CT. In some cases, Ponceau S staining solution was used to assess total protein content on the transferred membrane.

Chromosomal V(D)J Recombination Assays.

v-Abl lines were infected with the pMX-DEL-CJ, pMX-DEL-SJ, or pMX-INV cassettes and enriched using human CD4 MACS (Miltenyi). G1 arrest, DNA isolation, and Southern blot were performed as previously described (16). Briefly, v-Abl lines were treated with 3 μM STI-571 to initiate G1 arrest and V(D)J recombination; untreated samples and samples treated with STI-571 for 48 and 96 h were harvested for genomic DNA isolation. Southern blot was performed using a 900-bp HindIII/NheI fragment of the CD4 sequence as probe (isolated from pMX-DEL-SJ).

CH12 CSR, IgH Metaphase FISH, and LAM-HTGTS.

CH12 cells underwent CSR as previously described (36). Briefly, cells were stimulated at 5 × 10⁴ cells per 1 mL in the presence of 20 ng/mL IL-4, 1 μg/mL αCD40, and 1 ng/mL TGF-β, and cells were harvested after 48 or 72 h of stimulation for downstream assays. Metaphases for FISH analysis were obtained by adding colcemid (Karyomax) at 50 μg/mL overnight to CH12 cells stimulated for 48 h, with metaphase preparations and FISH studies performed as previously described (28). For HTGTS-based IgM to IgA end joining studies of stimulated CH12 lines, genomic DNA was isolated after 72 h. LAM-HTGTS was performed as previously described (30) and mapped against the mm9 genome build; Sα region junctions mapping only within positions 114,500,088–114,502,837 on chromosome 12 were used for MH analysis to obviate any contributions of potential local joining events from the noncoding allele, which contains a hybrid Sμ-Sα (27). MH analysis was performed as previously described (12).

Cell Cycle Analysis and Clonogenic Ionizing Radiation Sensitivity Assays.

Cell cycle analysis for v-Abl lines was measured by assaying transient BrdU incorporation as a function of nuclear content. Briefly, cells were pulsed for 20–60 min with a cell proliferation reagent (2 μL/mL; GE Healthcare) before fixation in 2% paraformaldehyde (diluted in PBS; Electron Microscopy Sciences) at 4 °C overnight. Cells were permeabilized and BrdU epitope exposed using freshly made 3 N HCl/0.5% Tween-20 for 20 min at room temperature and neutralized in 0.1 M sodium tetraborate. FITC–anti-BrdU (1:100; BD Pharmingen) and 7AAD (1:250; BD Pharmingen) were incubated for 15 min each at room temperature in the dark. Labeled cells were assayed for cell cycle analysis using flow cytometry (BD FACSCalibur). Optimal culture durations necessary for uniform G1 arrest of v-Abl lines in the presence of 3 μM STI-571 or 10 μM PD0332991 (Palbociclib) were 60 and 24 h, respectively.

Clonogenic ionizing radiation sensitivity assays were performed for v-Abl lines that were either asynchronous or G1 arrested with either STI-571 or PD0332991 before treatment. Cells were treated with 1, 2.5, or 5 Gy ionizing radiation and compared with an untreated control for each cell line assayed. Briefly, 5 × 10⁶ v-Abl cells that were either G1 arrested or asynchronous before treatment were centrifuged at 300 × g for 5 min, washed once with fresh medium, resuspended in 8 mL culture medium without G1 arrest agents, and aliquoted equally into separate tubes for the four conditions assayed. Cells were then either left untreated or treated with ionizing radiation, and then, for each treatment, they were transferred to the first 8-well column of a 96-well dish prefilled with culture media (100 μL per well—excluding the first column—using the Liquidator 96; Mettler Toledo) at ∼6 × 10⁴ cells per well. Cells from each of these 8 wells were 1:1 serially diluted across an additional 23 wells (2 × 96-well dishes) and cultured at 37 °C and 5% CO₂. Clonogenic survival after ionizing radiation treatment was measured relative to untreated cells after 10–11 d of culture.