DNA-Pkcs Phosphorylation at the T2609 Cluster Alters the Repair Pathway Choice During

bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 DNA-PKcs phosphorylation at the T2609 cluster alters the repair pathway choice during

2 immunoglobulin class switch recombination

4 Jennifer L. Crowe*,† Xiaobin S. Wang*,†, Zhengping Shao*, Brian J. Lee*, Verna Estes* and Shan

5 Zha*,‡

7 * Institute for Cancer Genetics, Vagelos College of Physicians and Surgeons, Columbia

8 University, New York City, NY 10032

9 † Graduate program of Pathobiology and Molecular Medicine, Vagelos College of Physicians

10 and Surgeons, Columbia University, New York City, NY 10032

11 ‡ Division of Pediatric Oncology, Hematology and Stem Cell Transplantation, Department of

12 Pediatrics, Vagelos College of Physicians & Surgeons, Columbia University, New York City,

13 NY 10032

14 † equal contribution

15 Running title: Class switch recombination without DNA-PKcs T2609 phosphorylation

16 Address Correspondence to Shan Zha at [email protected]

17 Word counts (without space): 42,902

18 Abstract

19 The DNA-dependent protein kinase (DNA-PK), composed of the KU heterodimer and the

20 large catalytic subunit (DNA-PKcs), is a classical non-homologous end-joining (cNHEJ)

21 factor. Naïve B cells undergo class switch recombination (CSR) to generate antibodies with

22 different isotypes by joining two DNA double-strand breaks at different switching regions via

23 the cNHEJ pathway. DNA-PK and the cNHEJ pathway play important roles in the DNA repair

24 phase of CSR. To initiate cNHEJ, KU binds to DNA ends, and recruits and activates DNA-PK.

25 DNA-PKcs is the best-characterized substrate of DNA-PK, which phosphorylates DNA-PKcs at

26 both the S2056 and T2609 clusters. Loss of T2609 cluster phosphorylation increases radiation

27 sensitivity, suggesting a role of T2609 phosphorylation in DNA repair. Using the DNA-PKcs5A

28 mouse model carrying an alanine substitution at the T2609 cluster, here we show that loss of

29 T2609 phosphorylation of DNA-PKcs does not affect the CSR efficiency. Yet, the CSR

30 junctions recovered from DNA-PKcs5A/5A B cells reveal increased chromosomal translocation,

31 excess end-resection, and preferential usage of micro-homology – all signs of the alternative

32 end-joining pathway. Thus, these results uncover a role of DNA-PKcs T2609 phosphorylation

33 in promoting cNHEJ repair pathway choice during CSR.

35 Key points

36 Loss of T2069 cluster phosphorylation of DNA-PKcs promotes Alt-EJ-mediated CSR.

38 Introduction

39 DNA-dependent protein kinase (DNA-PK), composed of the KU70 and KU86 (Ku80 in

40 mouse) heterodimer (KU) and a large catalytic subunit (DNA-PKcs), is a member of the classical

41 non-homologous end-joining (cNHEJ) pathway. As a major DNA double-strand break (DSB)

42 repair pathway in mammalian cells, cNHEJ entails both end-processing and end-ligation. KU

43 initiates cNHEJ by binding to the double-stranded DNA (dsDNA) ends, which in turn recruits

44 and activates DNA-PKcs that, among other functions, activates Artemis endonuclease for end-

45 processing (1, 2). KU also interacts with and stabilizes the LIG4/XRCC4/XLF complex for end-

46 ligation. In animal models, loss of cNHEJ mediated end-ligation (e.g., Lig4-/-) leads to severe

47 neuronal apoptosis, which causes embryonic lethality in the case of Xrcc4-/- and Lig4-/- mice (3).

48 In contrast, loss of DNA-PKcs or Artemis increases radiation sensitivity but does not cause

49 measurable neuronal apoptosis, consistent with their relatively limited roles in direct end-

50 ligation, if any (4-6). Several new cNHEJ factors (e.g., PAXX, and CYREN/MRI) are

51 characterized based on their interaction with KU. Loss of PAXX or CYREN/MRI increases

52 radiation sensitivity, but only abrogates end-ligation if XLF, a non-essential cNHEJ factor, is

53 also lost simultaneously (7-11).

54 In addition to general DSB repair, developing lymphocytes undergo programmed DSB

55 break and repair events to assemble and modify the immunoglobulin heavy chain (IgH) genes.

56 Specifically, V(D)J recombination, which occurs in progenitor lymphocytes, assembles the

57 variable region exon from individual V, D, and J gene segments exclusively via cNHEJ. DNA-

58 PKcs and Artemis are required for opening the hairpin at the coding ends during V(D)J

59 recombination. Correspondingly, DNA-PKcs-/- (null) or Artemis-/- mice are born of normal size at

60 the expected ratio but develop severe combined immunodeficiency (SCID) (4-6). Patients with

61 hypomorphic mutations in DNA-PKcs or Artemis also develop SCID (12, 13). Mature B cells

62 undergo CSR, which replace the initially expressed IgM constant region with another

63 downstream constant region encoding a different isotype, to generate antibodies with different

64 effector functions. The cNHEJ pathway plays an important role in CSR. But in cells lacking a

65 core cNHEJ factor (e.g., Lig4 or Xrcc4), up to 25-50% of CSR can be achieved by the alternative

66 end-joining (Alt-EJ) pathway that preferentially uses microhomology (MH) at the junctions.

67 Consistent with DNA-PKcs and Artemis being dispensable for direct end-ligation, DNA-PKcs-/-

68 B cells only have moderate defects in CSR (14, 15). Nevertheless, in recent high-throughput

69 sequence analyses, we found that CSR junctions recovered from DNA-PKcs-/- B cells contained

70 increased chromosomal translocations and extensive end-resection, and preferentially used MHs

71 (16), suggesting that the seemingly robust CSR achieved in DNA-PKcs-/- cells is primarily

72 mediated by the Alt-EJ pathway like those in Lig4-/- or Xrcc4-/- B cells (17, 18). Accordingly,

73 human patients with spontaneous mutations in DNA-PKcs show severe defects in both V(D)J

74 recombination and CSR and increased MH in the residual CSR junctions(12, 19).

75 Active DNA-PK phosphorylates many substrates, among which, DNA-PKcs is the best-

76 characterized (20). In a previous study, we showed that mice carrying a knock-in kinase-dead

77 mutation of DNA-PKcs (DNA-PKcsKD) die in utero with severe neuronal apoptosis, like Lig4-/-

78 or Xrcc4-/- mice. KU deletion rescued the embryonic development of DNA-PKcsKD/KD mice,

79 suggesting that DNA-PKcs has an end-capping function at the DNA ends that is regulated by its

80 kinase activity. Correspondingly, DNA-PKcsKD/KD mature B cells display severe CSR defects

81 like Lig4-/- B cells (16) and the residual CSR junctions from DNA-PKcsKD/KD B cells are all

82 mediated by the Alt-EJ pathway

83 Three DNA damage-induced phosphorylation clusters on DNA-PKcs have been

84 characterized: the S2056 cluster, the T2609 cluster, and the S3590 cluster (Fig. 1A)(21-24).

85 Among them, the T2609 cluster can also be cross phosphorylated by ATM upon ionizing

86 radiation (25, 26) and by ATR upon UV radiation (27). Human DNA-PKcs with alanine

87 substitutions at the S2056 cluster and/or the T2609 cluster, cannot restore radiation resistance in

88 DNA-PKcs-deficient Chinese hamster ovary (CHO) cells (22, 28-31), suggesting a role of DNA-

89 PKcs phosphorylation in DSB repair. The mouse model carrying alanine substitutions in the

90 S2056 cluster (DNA-PKcsPQR/PQR) has no defects in chromosomal V(D)J Recombination or CSR,

91 despite moderate radiation sensitivity (32). Mouse models with alanine substitutions of three or

92 all five threonines at the T2609 cluster (DNA-PKcs3A/3A or DNA-PKcs5A/5A) developed neonatal

93 bone marrow failure that abolishes both myeloid and lymphoid progenitors (33-35). Tp53-

94 deficiency restores lymphocyte development and peripheral mature lymphocyte counts in DNA-

95 PKcs3A/3A and DNA-PKcs5A/5A mice and v-abl kinase transformed DNA-PKcs3A/3A and DNA-

96 PKcs5A/5A B cells undergo robust chromosomal V(D)J recombination in culture (35, 36),

97 suggesting that T2609 phosphorylation is not required for V(D)J recombination.

98 Here we analyzed the impact of DNA-PKcs T2609 phosphorylation on CSR in mature B

99 cells isolated from DNA-PKcs5A/5A mice carrying Tp53-deficiency (heterozygous or

100 homozygous). While CSR efficiency is not affected by T2609A substitution, high throughput

101 sequence analyses reveal that the CSR junctions from DNA-PKcs5A/5A B cells contain increased

102 chromosomal translocations, evidence of excessive end-resection, and increased MHs – all

103 signatures of the Alt-EJ pathway. Therefore, our findings uncover a role of DNA-PKcs T2609

104 phosphorylation in repair pathway choice during CSR.

105 Materials and Methods

106 Generation and characterization of the mouse models

107 DNA-PKcs-/-, p53-/- and the pre-arranged IgH heavy and light chain knock-in alleles (HLk/k) that

108 bypass the need for DNA-PKcs in early B cell development, have been described previously (4,

109 37-39). The DNA-PKcs5A allele and the generation of the DNA-PKcs5A/5A mouse model were

110 recently published (35). The DNA-PKcs5A mutation converts all 5 threonines in the T2605

111 cluster (T2609 in human) encoded by exon 58 of murine DNA-PKcs to alanines (Fig.1A).

112 Briefly, DNA sequences (5’ and 3’ arms) from the genomic DNA-PKcs (Prkdc) locus were

113 isolated via PCR and cloned into the pGEMT vector. We synthesized a ~500bp fragment

114 containing all of the mutations (Genewiz), inserted it into an extended 3’arm (4.4kb) in

115 pGEMT, and confirmed by sequencing. The 5’ arm and the extended 3’arm with the

116 mutations were then inserted into the pEMCneo vector that carries a Neo-Resistance (NeoR)

117 gene flanked by a pair of FRT sites. The Sma1 linearized plasmids were electroporated into

118 murine ES cells (129Sv background) and NeoR+ clones were isolated and screened by

119 Southern blotting and Sanger sequencing (Fig.1A) (35). Upon injection and germline

120 transmission, the resultant DNA-PKcs+/5AN mice (N for NeoR containing) were crossed with

121 constitutively FLIpase expressing Rosa26aFLIP/FLIP mice (Jax stock number 003946, also in

122 129Sv background) to remove the NeoR cassette and allow the expression of the DNA-PKcs-

123 5A allele. All animal work was conducted in a specific pathogen-free facility and all the

124 procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at

125 Columbia University Medical Center.

126 Lymphocyte development analyses

127 Single-cell suspensions were prepared from the thymus, bone marrow, and spleen of

128 young adult (6-8 weeks) mice. Splenocytes were treated with red blood cell lysis buffer (Lonza

129 ACK Lysis Buffer) for 1-2 minutes at room temperature. Approximately 1 × 105 cells were

130 stained using fluorescence-conjugated antibodies and analyzed by flow cytometry (11, 40). The

131 following antibody cocktails were used for B cell (FITC anti-mouse CD43, Biolegend, 553270;

132 PE Goat anti-mouse IgM, Southern Biotech, 1020-09; PE-Cyanine5 anti-Hu/Mo CD45R (B220),

133 eBioScience, 15-0452-83; and APC anti-mouse TER119, Biolegend, 116212) and T cell (PE rat

134 anti-mouse CD4, Biolegend, 557308; FITC anti-mouse CD8a, Biolegend, 100706; PE/Cy5 anti-

135 mouse CD3e, eBioscience, 15-0031-83; and APC anti-mouse TCRβ, BD Pharmingen, 553174)

136 analyses. The dead cells and debris were excluded based on their high side scatter and low

137 forward scatter. For bone marrow and spleen with significant and sometimes variable

138 erythrocytes remaining after red blood cell lysis, we used the Ter119 (an erythrocyte marker)

139 antibody to remove the red blood cells before the analyses of B, T, and myeloid cell-specific

140 markers as in Figure 1.

141 Class switch recombination analyses

142 For the CSR assay, CD43- splenocytes were isolated with magnetic beads (MACS,

143 Miltenyi Biotec) and cultured (5-10 x 105 cells ml-1) in RPMI (GIBCO) supplemented with 15%

144 FBS (Hyclone) and 2 µg ml-1 anti-CD40 (BD Bioscience) plus 20 ng ml-1 of IL-4 (R&D). Cells

145 were maintained daily at 1 x 106 cells ml-1 and collected for flow cytometry after staining for

146 IgG1 and B220 (FITC anti-IgG1, BD Pharmingen, and PE Cy5 anti- B220, eBioscience). Flow

147 cytometry data were collected on a FACSCalibur flow cytometer (BD Bioscience) and processed

148 using the FlowJo V10 software package. For the CTV staining, purified B cells were incubated

149 at 6 × 106 cells ml−1 in 1 mL of PBS with 1 uL of 5 mM Cell Trace Violet (CTV) dye (Thermo

150 Scientific). The CTV-stained B cells were washed and activated as detailed above and subjected

151 to flow cytometry analyses 4 days after activation.

152 High throughput genomic translocation sequencing (HTGTS) of CSR junctions

153 HTGTS was performed as described (16, 41). Briefly, genomic DNA was collected from

154 activated B cells 4 days after stimulation, sonicated (Diagenode Bioruptor), and amplified with

155 an Sµ specific biotinylated primer (5’/5BiosG/CAGACCTGGGAATGTATGGT3’) and nested

156 primer (5’CACACAAAGACTCTGGACCTC3’). AflII was used to remove germline (non-

157 rearranged) sequences. Since all mice were of pure 129 background, the IgH switch region (from

158 JH4 to the last Cα exon, chr12: 114,494,415-114,666,816) of the C57/BL6 based mm9 genome

159 was replaced with the corresponding region of the AJ851868.3 129 IgH sequence (1415966-

160 1592715) to generate the mm9sr (switch region replacement) genome and the sequence analyses

161 were performed as detailed previously (41, 42). The best-path searching algorithm (related to

162 YAHA (43)) was used to identify optimal sequence alignments from Bowtie2-reported top

163 alignments (alignment score > 50). The reads were filtered to exclude mispriming, germline

164 (unmodified), sequential joints, and duplications. To plot all of the S-region junctions, including

165 those within the repeats and thus have low mappability but unequivocally mapped to an

166 individual switch region, we combined the ones filtered by a mappability filter but unequivocally

167 mapped to S regions with ‘good’ reads passing both the mappability filters (both de-duplicated)

168 (41). MHs are defined as regions of 100% homology between the bait and prey-break sites.

169 Insertions are defined as regions containing nucleotides that map to neither the bait nor prey

170 break site. Blunt junctions are considered to have no MHs or insertions. We calculated the

171 AGCT or RGYW motif frequency in each switch region using the IgH region sequence from the

172 129/Sv strain (accession number AJ851868.3). The mutation rate was calculated by a customized

173 Excel integrated VBA script. Specifically, we analyzed the mutations in the bait portion by

174 comparing the actual bait sequence of each IgH junction with the germline bait sequence from

175 the 129/Sv strain (accession number AJ851868.3). Only true mismatches (no insertion or

176 deletion) were counted as mutations. The HTGTS data reported in this paper have been

177 deposited in the Gene Expression Omnibus (GEO) database, https://www.ncbi.nlm.nih.gov/geo/

178 (accession no. GSE117628 and others).

179

180 Results

181 Loss of the Tp53 tumor suppressor gene restores peripheral B cells in DNA-PKcs5A/5A

182 mice

183 The human DNA-PKcs T2609 phosphorylation cluster contains five TQ sites, among

184 which, T2609, T2638 and T2647 (T2605, T2634, and T2643 in mouse, respectively) are

185 conserved (Fig. 1A) (20, 29, 30). To interrogate the impact of DNA-PKcs T2609 cluster

186 phosphorylation, we replaced all five threonine residues in this region with alanines and named

187 the allele -5A. Similar to the previously published 3A allele (which replaces the three conserved

188 threonines)(33), the DNA-PKcs5A/5A mice were born alive with hyper-pigmented toes, but

189 succumbed to bone marrow failure by 3 weeks with very few mature lymphocytes (35)(Fig.

190 1B, and Supplementary Fig. 1A and 1B). Tp53-deficiency, both heterozygous and

191 homozygous, rescues the growth retardation, corrects the hyperpigmentation, and most

192 importantly, restores peripheral splenic B cell frequency (Fig.1B, Supplementary Fig. 1A, and

193 1B), consistent with the dispensible role of T2609 phosphorylation in V(D)J recombination

194 (35, 36).

195 DNA-PKcs5A/5A B cells undergo slightly delayed, but efficient Class Switch Recombination

196 To understand the role of DNA-PKcs T2609 phosphorylation in CSR, we purified splenic

197 B cells from DNA-PKcs5A/5ATp53+/-(or -/-) and control DNA-PKcs+/+ or Tp53+/-(or -/-) mice and

198 cultured them with anti-CD40 and interleukin 4 (IL-4) that induce robust CSR to IgG1 and IgE.

199 Tp53-deficiency does not affect CSR efficiency or CSR junctions (16, 17, 44, 45). About 20%

200 DNA-PKcs+/+ B cells were IgG1+ after 3 days of stimulation and ~37% IgG1+ by day 4 (Fig.1C

201 and 1D). In comparison, DNA-PKcs5A/5ATp53+/-(or -/-) B cells achieved ~15% IgG1+ B cells by day

202 3 and eventually 30-40% IgG1+ by day 4 of activation (Fig.1C and 1D). Successful CSR

203 requires cell proliferation. Cell division analyses via the Cell Trace Violet (CTV) dye did not

204 find proliferation defects in DNA-PKcs5A/5ATp53+/-(or -/-) B cells (Supplementary Fig. 1C),

205 consistent with the largely normal weight and peripheral lymphocyte numbers of DNA-

206 PKcs5A/5ATp53+/-(or -/-) mice (Supplementary Fig. 1A) (35). These data suggest that

207 phosphorylation of DNA-PKcs at the T2609 cluster is not required for efficient CSR, despite a

208 moderate delay in kinetics.

209 T2609 cluster phosphorylation alters the distribution of the IgH junctions

210 Given DNA-PKcs-/- B cells undergo efficient CSR to IgG1 via the Alt-EJ mediated repair

211 pathway, as detected by the HTGTS of Sμ- Sμ internal deletions and Sμ-Sγ1 CSR junctions (16),

212 we decided to analyze the CSR junctions in DNA-PKcs5A/5Ap53+/-(or -/-) B cells using the sensitive

213 HTGTS. Briefly “translocation junctions” from a single bait site at the 5’S to various targets

214 were isolated after linear amplification and sequenced (Fig. 2A) (16, 41). Altogether, we

215 analyzed 4545 junctions from B cells derived from 5 different DNA-PKcs5A/5Ap53+/-(or -/-) mice

216 and over 10,000 junctions total from control DNA-PKcs+/+, Tp53+/- and DNA-PKcs-/- mice. As

217 shown in Figure 2B, nearly 90% of all junctions isolated from DNA-PKcs+/+, Tp53-/- or Tp53+/-,

218 DNA-PKcs5A/5Ap53+/-(or -/-) and DNA-PKcs-/- mature B cells are joined with a target (referred as

219 “Prey”) within the IgH locus (defined as a 400kb region including all D, J elements and Constant

220 regions). We further defined the orientation of Prey junctions as (+) if it reads from centromere

221 to telomere after the junction, or as (–) if it reads from telomere to centromere. Since the IgH

222 locus resides on the (-) strand of murine chromosome 12 and the bait at the 5’ S reads from

223 telomere to centromere, the normal internal deletion (Sμ-Sμ) and class switch recombination

224 (Sμ-Sγ1, or Sμ-Sε) junctions are in the (-) orientation (Fig. 2A). Indeed, > 80% of all preys from

225 DNA-PKcs+/+ (WT) and Tp53-deficient B cells fall on the (-) strand of IgH (83% for WT and

226 82% for Tp53-deficient). Meanwhile, only 10% and 11% of all preys fall on the (+) strands of

227 IgH in the DNA-PKcs+/+ (WT) and Tp53-deficient cells (Fig. 2A). In contrast, only 75% of all

228 preys from DNA-PKcs5A/5A B cells are on the IgH (-) strand and an increased 17% fall on the IgH

229 (+) strand, similar to those in control DNA-PKcs-/- B cells (68% IgH+ and 22% IgH-) (Fig. 2B

230 and 2C) (16). To understand the significance of the relative increases of the IgH+ preys among

231 all IgH preys, we examined the orientation of the inter-chromosomal preys outside of the IgH

232 locus (almost all to another chromosome beyond chromosome 12). Non-IgH (nIgH) preys fall

233 equally on (+) and (-) strands in all genotypes tested (Fig. 2B), suggesting that the negative

234 strand bias among the IgH preys reflects a strong preference for intra-chromosomal deletion

235 joining. Thus, we hypothesized that the relative increase of IgH preys falling on the (+) strand

236 in DNA-PKcs5A/5A and DNA-PKcs-/- B cells reflect either increased inversional events (Fig 2C)

237 (as illustrated with blue arrows in Fig. 2A) or increased inter-sister chromatid or inter-

238 chromosomal translocations (between IgH of two homologous chromosomes 12). A similar

239 phenotype has been noted for B cells lacking cNHEJ factors (e.g., Xrcc4-/-) or DNA damage

240 response factors (e.g., Atm-/- and Tpr53BP1-/-) (16, 46, 47).

241 The cytokine combination anti-CD40 and IL-4 induce CSR to IgG1 and IgE. Accordingly,

242 ~42% of all junctions from DNA-PKcs+/+, and Tp53-/- or Tp53+/- mice were located in Sε (-)

243 (potentially IgE switching), 15~20% in Sγ1(-) (potential IgG1 switching) and 18-23% in Sμ (-)

244 (internal deletion) (Fig. 2D). IgH junctions from both DNA-PKcs-/- and DNA-PKcs5A/5A B cells

245 have a greatly increased Sμ (-) portion (28%), and a corresponding reduction in Sε (-) preys (9%

246 and 27%, respectively) (Fig.2D). The overall proportion changes are more dramatic in those from

247 DNA-PKcs-/- B cells than those from DNA-PKcs5A/5A B cells (16). Together these data suggest that

248 despite normal CSR efficiency, the junctions from DNA-PKcs5A/5A B cells contain increased

249 translocations (to the (+) strand) and preferential loss of Sμ-Sε junctions – both features of cNHEJ

250 deficiency, similar to those from DNA-PKcs-/- cells. Correspondingly, Florenscence In situ

251 Hybridization with a telomere probe to increase the sensitivity to IgH locus breaks near the

252 telomere described before (48) also revealed increased genomic instability, including both

253 chromosome breaks at both sister chromatids, and chromatid breaks in activated DNA-PKcs5A/5A

254 B cells (Fig.S2A, S2B, and S2C)

255 Loss of T2609 cluster phosphorylation increase MH usage in CSR junctions

256 In the absence of cNHEJ, including the loss of DNA-PKcs, significant CSR is achieved via

257 the Alt-EJ pathway that preferentially uses MHs at the junction (16-18). In a recent analysis, we

258 identified two features associated with Alt-EJ dependent CSR using HTGTS (16) - increased

259 usage of MH at the junctions and evidence of extensive resection in switch regions. Indeed,

260 junctions from DNA-PKcs-/- B cells with only moderate CSR defects contain both features (14, 15).

261 Among CSR junctions (including all IgH preys) recovered from DNA-PKcs+/+ or Tp53-deficient

262 (alone) cells, ~25% are blunt, 25% have 1nt MH and another 25% have 2-3nt MH. In contrast,

263 less than 15% of junctions from DNA-PKcs-/- and 20% from DNA-PKcs5A/5A cells were blunt and

264 only another 15% have 1nt MH (Fig. 2E). A prominent 40% of junctions from DNA-PKcs-/- B cells

265 and 32% of the IgH junctions from DNA-PKcs5A/5A B cells have 2-3nt MH. Indeed, cumulatively,

266 nearly 20% of IgH junctions from DNA-PKcs-/- and DNA-PKcs5A/5A B cells have >4 nt MH (in

267 contrast to ~10% in DNA-PKcs+/+ or Tp53-deficient alone) (Fig. 2E). The overall frequency of

268 insertions within IgH junctions did not differ between genotypes (Fig.2E). Together these findings

269 indicate that loss of the T2609 phosphorylation of DNA-PKcs unleashed robust MH-mediated

270 CSR that compensates for the moderate loss of cNHEJ.

271 Loss of T2609 cluster phosphorylation increases end-resection

272 Next, we mapped the junctions within and around each specific switch region. The (+) and

273 (-) oriented junctions were plotted above (blue) and below (red) the centerline, respectively (Fig,

274 2A, 3, 4, and 5). Since the location of the bait is fixed within the 5’ Sμ (telomeric), the majority of

275 the Sμ joins from DNA-PKcs+/+ and Tp53-deficient B cells are located near the bait site only ~10%

276 beyond the core Sμ region (Fig.3). In contrast, over 15% of Sμ junctions from DNA-PKcs5A/5A B

277 cells fall beyond the core Sμ region, a similar trend to what has previously been described in DNA-

278 PKcs-/- cells (Fig. 3). DNA-PKcs5A/5A junctions show a similarly wider distribution within Sγ1, with

279 1.66% falling downstream of Sγ1 (versus 0.14-0.54% in the controls) (Fig. 4), and within Sε (9.13%

280 in DNA-PKcs5A/5A junctions versus ~6% in the controls) (Fig. 5). Altogether, the extensive use of

281 distal switch regions suggests end-resection, another feature of Alt-EJ, and cNHEJ deficiency in

282 DNA-PKcs5A/5A B cells.

283 Loss of T2609 phosphorylation reduced G-mutations in the 5’Sµ bait region of IgH

284 Finally, we analyzed mutations in the 5’Sµ region among the junctions (Fig. 6). To

285 compare with prior studies using Sanger sequencing, we present the data based on the top strand

286 relative to the S germline transcription (Fig. 6A). The estimated mutation rate for the MiSeq

287 platform is 0.1% (10 mut/10,000nt). The overall mutation rate (mut/10,000nt) and frequency of

288 reads with mutations were both higher in IgH junctions derived from DNA-PKcs+/+ and Tp53-/-

289 cells (~27 mut/10,000nt) than those from DNA-PKcs5A/5A cells (~16 mut/10,000nt) that were also

290 Tp53 deficient (~16 mut/10,000nt) (Fig. 6D). Junctions from DNA-PKcs-/- cells also had a

291 slightly lower mutation rate than DNA-PKcs+/+ and Tp53-/- cells (Fig. 6D). Despite a similar

292 distribution of nucleotide types (Fig. 6B), G was most frequently mutated in the 5’Sµ region of

293 DNA-PKcs+/+ and Tp53-/- B cells (47 mut/10,000 G). Lower G mutation rates in DNA-PKcs-/-

294 cells (20 mut/10,0000 G) and DNA-PKcs5A/5A cells (25mut/10,000 G) account for a lower overall

295 mutation rate in those cells (Fig. 6F). The pattern of G mutations (relative frequency to A, T, C)

296 did not change, suggesting the reduction is likely due to AID targeting, rather than differences in

297 secondary processing by uracil-DNA glycosylase (UNG), Mismatch repair (MMR) or other

298 pathways (Fig. 6C). Moreover, there is also an overall reduction in the frequency of reads with

299 mutations in DNA-PKcs5A/5A cells (Fig. 6E). Altogether, we note a reduction of G nucleotide

300 mutations in the 5’Sμ region in DNA-PKcs5A/5A and DNA-PKcs-/- cells. Similar changes have

301 been noted in cNHEJ deficient, Xrcc4-/- and DNA-PKcsKD/KD B cells.

302 Discussion

303 DNA-PKcs deficiency has been linked to reduced CSR efficiency and a preference for

304 MH in patients and mouse models(16, 19). The T2609 cluster phosphorylation is one of the most

305 prominent post-translational modification of DNA-PKcs and has been associated with increased

306 radiation sensitivity (22, 28-31), which its role in CSR remains elusive. In this context, the

307 T2609 phosphorylation has been implicated in telomere biology (34) and ribosomal biogenesis

308 (35) which both could indirectly contribute to radiation sensitivity. To understand the role of

309 T2609 phosphorylation in CSR and DSB repair specifically, we evaluated the role of T2609

310 phosphorylation in physiological DSB repair during CSR using germline mouse models. As in

311 the case of V(D)J recombination (33, 35, 36), T2609A mutation also does not affect CSR

312 efficiency. But CSR junctions from DNA-PKcs5A/5A B cells revealed a preferential loss of Sμ-Sε

313 junctions, increased MH usage, and evidence of extensive end resection and increased inter-

314 chromosomal translocations ((+) strand preys)– all consistent with cNHEJ deficiency and

315 compensatory Alt-EJ-mediated CSR. Together these findings support the role of DNA-PKcs

316 T2609 phosphorylation in promoting cNHEJ-mediated CSR. In the absence of T2609

317 phosphorylation, the increased dependence on the Alt-EJ pathway, together with the extensive

318 length of the S1 region (12.5kb in 129SV strain), the dense RGYW/AGCT sequence motifs

319 (Fig. 4), and the sequence-independent 3D interactions (synapses) between Sµ-S1 (49-52), lead

320 to robust Alt-EJ-mediated CSR to IgG1 in DNA-PKcs5A/5A B cells. Thus, our analyses identify a

321 novel role of DNA-PKcs T2609 phosphorylation in repair pathway choice during CSR.

322 The estimated error rate of the Illumina Miseq platform is 0.1%, which is lower than the

323 estimated and reported somatic hypermutation (SHM) rate (53) and therefore allows for analyses

324 of SHM. As previously reported (54, 55), G mutations are two times more frequent than C

325 mutations on the top strand of the 5’Sµ region (the non-template strand relative to Sµ germline

326 transcription). Interestingly, the T2609A mutation attenuated this preference of G mutations,

327 while the pattern of G mutations (relative to A, C, T) is unaffected, suggesting that the T2609A

328 mutation and its associated repair defects might compromise AID targeting to 5’Sµ without

329 affecting U:G mismatch processing. Consistent with this model, the frequency of reads with

330 mutations also decreased in DNA-PKcs5A/5A B cells. The bias of APOBEC family enzymes to

331 deaminate the lagging strand during DNA replication (53, 56) and the relative abundance of

332 sense vs anti-sense transcripts (57) could all contribute to this strand bias.

333 The CSR junctions from DNA-PKcs5A/5A and DNA-PKcs-/- cells share many features,

334 including the end-resection and increased MH. T2609 phosphorylation is also dispensible for

335 hairpin opening and V(D)J recombination(33, 35, 36, 40), suggesting a model in which the

336 T2609 cluster phosphorylation promotes end-ligation, but is dispensable for end-processing and

337 Artemis recruitment. The increased use of MHs and end-resection in DNA-PKcs5A/5A cells are

338 also much less prominent than those from DNA-PKcsKD/KD cells (16, 40). Together with the

339 normal CSR in DNA-PKcsPQR/PQR B cells with alanine substitutions at the S2056 cluster(32),

340 these results suggest that lack of DNA-PKcs phosphorylation alone is not sufficient to explain

341 the strong cNHEJ-defects in DNA-PKcsKD/KD cells. The possibilities include other

342 phosphorylation sites (23) or phosphorylation-independent effects of catalysis itself on the

343 structure or organization of DNA-PKcs protein. Recent structural analyses of DNA-PK

344 holoenzymes have provided evidence for allosteric changes upon DNA-PK activation (58-60),

345 which might link the catalysis directly with DNA-PKcs turnover and its end-protection function.

346 In summary, our findings unequivocally demonstrate an important distinction between

347 auto-phosphorylation and catalysis on DNA-PKcs, indicating that auto-phosphorylation at

348 T2609, is largely dispensable for lymphocyte development. Despite the robust CSR

349 efficiency, the CSR junctions recovered from DNA-PKcs5A/5A cells reveal a strong bias toward

350 Alt-EJ, suggesting a role of T2609 in promoting cNHEJ at the price of Alt-EJ during CSR.

351

352 Acknowledgment

353 We thank Dr. Wenxia Jiang for her input on the DNA-PKcs KD mouse model and Ms.

354 Verna Estes for assistance with the colony management. We thank Dr. Chyuan-Sheng (Victor)

355 Lin for technical assistance and advice on the generation of DNA-PKcs5A/5A mouse models. We

356 thank other members of the Zha lab for helpful discussions and technical advice. We apologize

357 to colleagues whose work could not be cited due to space limitations and was covered by reviews

358 instead.

359 Reference 360 1. Ma, Y., U. Pannicke, K. Schwarz, and M. R. Lieber. 2002. Hairpin opening and overhang 361 processing by an Artemis/DNA-dependent protein kinase complex in nonhomologous 362 end joining and V(D)J recombination. Cell 108: 781-794. 363 2. Falck, J., J. Coates, and S. P. Jackson. 2005. Conserved modes of recruitment of ATM, 364 ATR and DNA-PKcs to sites of DNA damage. Nature 434: 605-611. 365 3. Frank, K. M., J. M. Sekiguchi, K. J. Seidl, W. Swat, G. A. Rathbun, H. L. Cheng, L. 366 Davidson, L. Kangaloo, and F. W. Alt. 1998. Late embryonic lethality and impaired 367 V(D)J recombination in mice lacking DNA ligase IV. Nature 396: 173-177. 368 4. Gao, Y., J. Chaudhuri, C. Zhu, L. Davidson, D. T. Weaver, and F. W. Alt. 1998. A 369 targeted DNA-PKcs-null mutation reveals DNA-PK-independent functions for KU in 370 V(D)J recombination. Immunity. 9: 367-376. 371 5. Taccioli, G. E., A. G. Amatucci, H. J. Beamish, D. Gell, X. H. Xiang, M. I. Torres 372 Arzayus, A. Priestley, S. P. Jackson, R. A. Marshak, P. A. Jeggo, and V. L. Herrera. 373 1998. Targeted disruption of the catalytic subunit of the DNA-PK gene in mice confers 374 severe combined immunodeficiency and radiosensitivity. Immunity. 9: 355-366. 375 6. Rooney, S., J. Sekiguchi, C. Zhu, H. L. Cheng, J. Manis, S. Whitlow, J. DeVido, D. Foy, 376 J. Chaudhuri, D. Lombard, and F. W. Alt. 2002. Leaky Scid phenotype associated with 377 defective V(D)J coding end processing in Artemis-deficient mice. Mol.Cell 10: 1379- 378 1390. 379 7. Hung, P. J., B. Johnson, B. R. Chen, A. K. Byrum, A. L. Bredemeyer, W. T. Yewdell, T. 380 E. Johnson, B. J. Lee, S. Deivasigamani, I. Hindi, P. Amatya, M. L. Gross, T. T. Paull, D. 381 J. Pisapia, J. Chaudhuri, J. J. H. Petrini, N. Mosammaparast, G. K. Amarasinghe, S. Zha, 382 J. K. Tyler, and B. P. Sleckman. 2018. MRI Is a DNA Damage Response Adaptor during 383 Classical Non-homologous End Joining. Mol Cell 71: 332-342 e338. 384 8. Ochi, T., A. N. Blackford, J. Coates, S. Jhujh, S. Mehmood, N. Tamura, J. Travers, Q. 385 Wu, V. M. Draviam, C. V. Robinson, T. L. Blundell, and S. P. Jackson. 2015. DNA 386 repair. PAXX, a paralog of XRCC4 and XLF, interacts with Ku to promote DNA double- 387 strand break repair. Science 347: 185-188. 388 9. Xing, M., M. Yang, W. Huo, F. Feng, L. Wei, W. Jiang, S. Ning, Z. Yan, W. Li, Q. 389 Wang, M. Hou, C. Dong, R. Guo, G. Gao, J. Ji, S. Zha, L. Lan, H. Liang, and D. Xu. 390 2015. Interactome analysis identifies a new paralogue of XRCC4 in non-homologous end 391 joining DNA repair pathway. Nat Commun 6: 6233. 392 10. Kumar, V., F. W. Alt, and R. L. Frock. 2016. PAXX and XLF DNA repair factors are 393 functionally redundant in joining DNA breaks in a G1-arrested progenitor B-cell line. 394 Proc Natl Acad Sci U S A 113: 10619-10624. 395 11. Liu, X., Z. Shao, W. Jiang, B. J. Lee, and S. Zha. 2017. PAXX promotes KU 396 accumulation at DNA breaks and is essential for end-joining in XLF-deficient mice. Nat 397 Commun 8: 13816. 398 12. van der Burg, M., H. Ijspeert, N. S. Verkaik, T. Turul, W. W. Wiegant, K. Morotomi- 399 Yano, P. O. Mari, I. Tezcan, D. J. Chen, M. Z. Zdzienicka, J. J. van Dongen, and D. C. 400 van Gent. 2009. A DNA-PKcs mutation in a radiosensitive T-B- SCID patient inhibits 401 Artemis activation and nonhomologous end-joining. J Clin Invest 119: 91-98. 402 13. Woodbine, L., A. R. Gennery, and P. A. Jeggo. 2014. The clinical impact of deficiency in 403 DNA non-homologous end-joining. DNA Repair (Amst) 16: 84-96.

404 14. Bosma, G. C., J. Kim, T. Urich, D. M. Fath, M. G. Cotticelli, N. R. Ruetsch, M. Z. Radic, 405 and M. J. Bosma. 2002. DNA-dependent protein kinase activity is not required for 406 immunoglobulin class switching. J.Exp.Med. 196: 1483-1495. 407 15. Manis, J. P., D. Dudley, L. Kaylor, and F. W. Alt. 2002. IgH class switch recombination 408 to IgG1 in DNA-PKcs-deficient B cells. Immunity. 16: 607-617. 409 16. Crowe, J. L., Z. Shao, X. S. Wang, P. C. Wei, W. Jiang, B. J. Lee, V. M. Estes, F. W. Alt, 410 and S. Zha. 2018. Kinase-dependent structural role of DNA-PKcs during 411 immunoglobulin class switch recombination. Proc Natl Acad Sci U S A. 412 17. Yan, C. T., C. Boboila, E. K. Souza, S. Franco, T. R. Hickernell, M. Murphy, S. 413 Gumaste, M. Geyer, A. A. Zarrin, J. P. Manis, K. Rajewsky, and F. W. Alt. 2007. IgH 414 class switching and translocations use a robust non-classical end-joining pathway. 415 Nature. 449: 478-482. 416 18. Boboila, C., C. Yan, D. R. Wesemann, M. Jankovic, J. H. Wang, J. Manis, A. 417 Nussenzweig, M. Nussenzweig, and F. W. Alt. 2010. Alternative end-joining catalyzes 418 class switch recombination in the absence of both Ku70 and DNA ligase 4. J Exp Med 419 207: 417-427. 420 19. Bjorkman, A., L. Du, K. Felgentreff, C. Rosner, R. Pankaj Kamdar, G. Kokaraki, Y. 421 Matsumoto, E. G. Davies, M. van der Burg, L. D. Notarangelo, L. Hammarstrom, and Q. 422 Pan-Hammarstrom. 2015. DNA-PKcs Is Involved in Ig Class Switch Recombination in 423 Human B Cells. J Immunol 195: 5608-5615. 424 20. Block, W. D., Y. Yu, D. Merkle, J. L. Gifford, Q. Ding, K. Meek, and S. P. Lees-Miller. 425 2004. Autophosphorylation-dependent remodeling of the DNA-dependent protein kinase 426 catalytic subunit regulates ligation of DNA ends. Nucleic Acids Res 32: 4351-4357. 427 21. Chan, D. W., B. P. Chen, S. Prithivirajsingh, A. Kurimasa, M. D. Story, J. Qin, and D. J. 428 Chen. 2002. Autophosphorylation of the DNA-dependent protein kinase catalytic subunit 429 is required for rejoining of DNA double-strand breaks. Genes Dev 16: 2333-2338. 430 22. Chen, B. P., D. W. Chan, J. Kobayashi, S. Burma, A. Asaithamby, K. Morotomi-Yano, E. 431 Botvinick, J. Qin, and D. J. Chen. 2005. Cell cycle dependence of DNA-dependent 432 protein kinase phosphorylation in response to DNA double strand breaks. J Biol Chem 433 280: 14709-14715. 434 23. Douglas, P., X. Cui, W. D. Block, Y. Yu, S. Gupta, Q. Ding, R. Ye, N. Morrice, S. P. 435 Lees-Miller, and K. Meek. 2007. The DNA-dependent protein kinase catalytic subunit is 436 phosphorylated in vivo on threonine 3950, a highly conserved amino acid in the protein 437 kinase domain. Mol Cell Biol 27: 1581-1591. 438 24. Merkle, D., P. Douglas, G. B. Moorhead, Z. Leonenko, Y. Yu, D. Cramb, D. P. Bazett- 439 Jones, and S. P. Lees-Miller. 2002. The DNA-dependent protein kinase interacts with 440 DNA to form a protein-DNA complex that is disrupted by phosphorylation. Biochemistry 441 41: 12706-12714. 442 25. Chen, B. P., N. Uematsu, J. Kobayashi, Y. Lerenthal, A. Krempler, H. Yajima, M. 443 Lobrich, Y. Shiloh, and D. J. Chen. 2007. Ataxia telangiectasia mutated (ATM) is 444 essential for DNA-PKcs phosphorylations at the Thr-2609 cluster upon DNA double 445 strand break. J Biol Chem 282: 6582-6587. 446 26. Douglas, P., G. P. Sapkota, N. Morrice, Y. Yu, A. A. Goodarzi, D. Merkle, K. Meek, D. 447 R. Alessi, and S. P. Lees-Miller. 2002. Identification of in vitro and in vivo 448 phosphorylation sites in the catalytic subunit of the DNA-dependent protein kinase. 449 Biochem J 368: 243-251.

450 27. Yajima, H., K. J. Lee, and B. P. Chen. 2006. ATR-dependent phosphorylation of DNA- 451 dependent protein kinase catalytic subunit in response to UV-induced replication stress. 452 Mol Cell Biol 26: 7520-7528. 453 28. Ding, Q., Y. V. Reddy, W. Wang, T. Woods, P. Douglas, D. A. Ramsden, S. P. Lees- 454 Miller, and K. Meek. 2003. Autophosphorylation of the catalytic subunit of the DNA- 455 dependent protein kinase is required for efficient end processing during DNA double- 456 strand break repair. Mol Cell Biol 23: 5836-5848. 457 29. Cui, X., Y. Yu, S. Gupta, Y. M. Cho, S. P. Lees-Miller, and K. Meek. 2005. 458 Autophosphorylation of DNA-dependent protein kinase regulates DNA end processing 459 and may also alter double-strand break repair pathway choice. Mol Cell Biol 25: 10842- 460 10852. 461 30. Meek, K., P. Douglas, X. Cui, Q. Ding, and S. P. Lees-Miller. 2007. trans 462 Autophosphorylation at DNA-dependent protein kinase's two major autophosphorylation 463 site clusters facilitates end processing but not end joining. Mol Cell Biol 27: 3881-3890. 464 31. Uematsu, N., E. Weterings, K. Yano, K. Morotomi-Yano, B. Jakob, G. Taucher-Scholz, 465 P. O. Mari, D. C. van Gent, B. P. Chen, and D. J. Chen. 2007. Autophosphorylation of 466 DNA-PKCS regulates its dynamics at DNA double-strand breaks. J Cell Biol 177: 219- 467 229. 468 32. Jiang, W., V. M. Estes, X. S. Wang, Z. Shao, B. J. Lee, X. Lin, J. L. Crowe, and S. Zha. 469 2019. Phosphorylation at S2053 in Murine (S2056 in Human) DNA-PKcs Is Dispensable 470 for Lymphocyte Development and Class Switch Recombination. J Immunol 203: 178- 471 187. 472 33. Zhang, S., H. Yajima, H. Huynh, J. Zheng, E. Callen, H. T. Chen, N. Wong, S. Bunting, 473 Y. F. Lin, M. Li, K. J. Lee, M. Story, E. Gapud, B. P. Sleckman, A. Nussenzweig, C. C. 474 Zhang, D. J. Chen, and B. P. Chen. 2011. Congenital bone marrow failure in DNA-PKcs 475 mutant mice associated with deficiencies in DNA repair. J Cell Biol 193: 295-305. 476 34. Zhang, S., S. Matsunaga, Y. F. Lin, B. Sishc, Z. Shang, J. Sui, H. Y. Shih, Y. Zhao, O. 477 Foreman, M. D. Story, D. J. Chen, and B. P. Chen. 2016. Spontaneous tumor 478 development in bone marrow-rescued DNA-PKcs(3A/3A) mice due to dysfunction of 479 telomere leading strand deprotection. Oncogene 35: 3909-3918. 480 35. Shao, Z., R. A. Flynn, J. L. Crowe, Y. Zhu, J. Liang, W. Jiang, F. Aryan, P. Aoude, C. R. 481 Bertozzi, V. M. Estes, B. J. Lee, G. Bhagat, S. Zha, and E. Calo. 2020. DNA-PKcs has 482 KU-dependent function in rRNA processing and haematopoiesis. Nature 579: 291-296. 483 36. Lee, B. S., E. J. Gapud, S. Zhang, Y. Dorsett, A. Bredemeyer, R. George, E. Callen, J. A. 484 Daniel, O. Osipovich, E. M. Oltz, C. H. Bassing, A. Nussenzweig, S. Lees-Miller, M. 485 Hammel, B. P. Chen, and B. P. Sleckman. 2013. Functional Intersection of ATM and 486 DNA-PKcs in Coding End Joining During V(D)J Recombination. Mol Cell Biol. 487 37. Symonds, H., L. Krall, L. Remington, M. Saenz-Robles, S. Lowe, T. Jacks, and T. Van 488 Dyke. 1994. p53-dependent apoptosis suppresses tumor growth and progression in vivo. 489 Cell 78: 703-711. 490 38. Sonoda, E., Y. Pewzner-Jung, S. Schwers, S. Taki, S. Jung, D. Eilat, and K. Rajewsky. 491 1997. B cell development under the condition of allelic inclusion. Immunity 6: 225-233. 492 39. Pelanda, R., S. Schaal, R. M. Torres, and K. Rajewsky. 1996. A prematurely expressed 493 Ig(kappa) transgene, but not V(kappa)J(kappa) gene segment targeted into the Ig(kappa) 494 locus, can rescue B cell development in lambda5-deficient mice. Immunity 5: 229-239.

495 40. Jiang, W., J. L. Crowe, X. Liu, S. Nakajima, Y. Wang, C. Li, B. J. Lee, R. L. Dubois, C. 496 Liu, X. Yu, L. Lan, and S. Zha. 2015. Differential phosphorylation of DNA-PKcs 497 regulates the interplay between end-processing and end-ligation during nonhomologous 498 end-joining. Mol Cell 58: 172-185. 499 41. Dong, J., R. A. Panchakshari, T. Zhang, Y. Zhang, J. Hu, S. A. Volpi, R. M. Meyers, Y. 500 J. Ho, Z. Du, D. F. Robbiani, F. Meng, M. Gostissa, M. C. Nussenzweig, J. P. Manis, and 501 F. W. Alt. 2015. Orientation-specific joining of AID-initiated DNA breaks promotes 502 antibody class switching. Nature 525: 134-139. 503 42. Hu, J., R. M. Meyers, J. Dong, R. A. Panchakshari, F. W. Alt, and R. L. Frock. 2016. 504 Detecting DNA double-stranded breaks in mammalian genomes by linear amplification- 505 mediated high-throughput genome-wide translocation sequencing. Nat Protoc 11: 853- 506 871. 507 43. Faust, G. G., and I. M. Hall. 2012. YAHA: fast and flexible long-read alignment with 508 optimal breakpoint detection. Bioinformatics 28: 2417-2424. 509 44. Franco, S., F. W. Alt, and J. P. Manis. 2006. Pathways that suppress programmed DNA 510 breaks from progressing to chromosomal breaks and translocations. DNA Repair (Amst) 511 5: 1030-1041. 512 45. Boboila, C., V. Oksenych, M. Gostissa, J. H. Wang, S. Zha, Y. Zhang, H. Chai, C. S. 513 Lee, M. Jankovic, L. M. Saez, M. C. Nussenzweig, P. J. McKinnon, F. W. Alt, and B. 514 Schwer. 2012. Robust chromosomal DNA repair via alternative end-joining in the 515 absence of X-ray repair cross-complementing protein 1 (XRCC1). Proc Natl Acad Sci U 516 S A 109: 2473-2478. 517 46. Zhang, Y., M. Gostissa, D. G. Hildebrand, M. S. Becker, C. Boboila, R. Chiarle, S. 518 Lewis, and F. W. Alt. 2010. The role of mechanistic factors in promoting chromosomal 519 translocations found in lymphoid and other cancers. Adv Immunol 106: 93-133. 520 47. Panchakshari, R. A., X. Zhang, V. Kumar, Z. Du, P. C. Wei, J. Kao, J. Dong, and F. W. 521 Alt. 2018. DNA double-strand break response factors influence end-joining features of 522 IgH class switch and general translocation junctions. Proc Natl Acad Sci U S A 115: 762- 523 767. 524 48. Franco, S., M. Gostissa, S. Zha, D. B. Lombard, M. M. Murphy, A. A. Zarrin, C. Yan, S. 525 Tepsuporn, J. C. Morales, M. M. Adams, Z. Lou, C. H. Bassing, J. P. Manis, J. Chen, P. 526 B. Carpenter, and F. W. Alt. 2006. H2AX prevents DNA breaks from progressing to 527 chromosome breaks and translocations. Mol.Cell 21: 201-214. 528 49. Wuerffel, R., L. Wang, F. Grigera, J. Manis, E. Selsing, T. Perlot, F. W. Alt, M. Cogne, 529 E. Pinaud, and A. L. Kenter. 2007. S-S synapsis during class switch recombination is 530 promoted by distantly located transcriptional elements and activation-induced deaminase. 531 Immunity 27: 711-722. 532 50. Jhunjhunwala, S., M. C. van Zelm, M. M. Peak, and C. Murre. 2009. Chromatin 533 architecture and the generation of antigen receptor diversity. Cell 138: 435-448. 534 51. Zarrin, A. A., M. Tian, J. Wang, T. Borjeson, and F. W. Alt. 2005. Influence of switch 535 region length on immunoglobulin class switch recombination. Proc Natl Acad Sci U S A 536 102: 2466-2470. 537 52. Zarrin, A. A., F. W. Alt, J. Chaudhuri, N. Stokes, D. Kaushal, L. Du Pasquier, and M. 538 Tian. 2004. An evolutionarily conserved target motif for immunoglobulin class-switch 539 recombination. Nat Immunol 5: 1275-1281.

540 53. Haradhvala, N. J., P. Polak, P. Stojanov, K. R. Covington, E. Shinbrot, J. M. Hess, E. 541 Rheinbay, J. Kim, Y. E. Maruvka, L. Z. Braunstein, A. Kamburov, P. C. Hanawalt, D. A. 542 Wheeler, A. Koren, M. S. Lawrence, and G. Getz. 2016. Mutational Strand Asymmetries 543 in Cancer Genomes Reveal Mechanisms of DNA Damage and Repair. Cell 164: 538-549. 544 54. Neuberger, M. S., and C. Rada. 2006. Somatic hypermutation: activation-induced 545 deaminase for C/G followed by polymerase {eta} for A/T. J.Exp.Med. . 546 55. Xue, K., C. Rada, and M. S. Neuberger. 2006. The in vivo pattern of AID targeting to 547 immunoglobulin switch regions deduced from mutation spectra in msh2-/- ung-/- mice. 548 J.Exp.Med. 203: 2085-2094. 549 56. Hoopes, J. I., L. M. Cortez, T. M. Mertz, E. P. Malc, P. A. Mieczkowski, and S. A. 550 Roberts. 2016. APOBEC3A and APOBEC3B Preferentially Deaminate the Lagging 551 Strand Template during DNA Replication. Cell Rep 14: 1273-1282. 552 57. Lim, J., P. K. Giri, D. Kazadi, B. Laffleur, W. Zhang, V. Grinstein, E. Pefanis, L. M. 553 Brown, E. Ladewig, O. Martin, Y. Chen, R. Rabadan, F. Boyer, G. Rothschild, M. 554 Cogne, E. Pinaud, H. Deng, and U. Basu. 2017. Nuclear Proximity of Mtr4 to RNA 555 Exosome Restricts DNA Mutational Asymmetry. Cell 169: 523-537 e515. 556 58. Yin, X., M. Liu, Y. Tian, J. Wang, and Y. Xu. 2017. Cryo-EM structure of human DNA- 557 PK holoenzyme. Cell Res 27: 1341-1350. 558 59. Sibanda, B. L., D. Y. Chirgadze, D. B. Ascher, and T. L. Blundell. 2017. DNA-PKcs 559 structure suggests an allosteric mechanism modulating DNA double-strand break repair. 560 Science 355: 520-524. 561 60. Sharif, H., Y. Li, Y. Dong, L. Dong, W. L. Wang, Y. Mao, and H. Wu. 2017. Cryo-EM 562 structure of the DNA-PK holoenzyme. Proc Natl Acad Sci U S A 114: 7367-7372. 563

564 FIGURE LEGENDS

565 Figure 1: Class switch recombination is transiently delayed in DNA-PKcs5A/5A cells .

566 (A) Diagram of the DNA-PKcs protein and the organization of the T2609 cluster on human

567 DNA-PKcs and the corresponding T2605 cluster on mouse DNA-PKcs. The conserved T2609,

568 T2636 and T2647 are marked with asterisks. The DNA-PKcs5A allele contains alanine

569 substitutions of all 5 threonines, while the previously published DNA-PKcs 3A allele contains

570 alanine substitions at three conserved threonines. (B) Representative flow cytometry analyses of

571 B and T cell development in DNA-PKcs5A/5A and DNA-PKcs5A/5ATp53-/- mice, indicating Tp53-

572 deficiency restored peripheral mature B cells (IgM+B220+) in DNA-PKcs5A/5A mice. (C)

573 Representative flow cytometry analyses of class switch recombination to IgG1 at day 3 and day

574 4 after cytokine activation. (D) Quantification of IgG1 switching. The bars mark the standard

575 erros. There is no significant difference between DNA-PKcs5A/5A and DNA-PKcs+/+ B cells.

576 Figure 2: Class switch recombination in DNA-PKcs5A/5A cells have increased MH.

577 (A) Diagram of the murine IgH locus with the location of Sμ (beige), Sγ1 (blue) and Sε (green).

578 The bait site is marked on the lower right of the diagram. The red (deletional, (-) strand) and blue

579 (inversional, (+) strand) arrows indicate the orientation of potential junctions. (B) Relative

580 frequency of HTGTS junctions in the (-) and (+) strand orientation, and in the IgH and nonIgH

581 regions. We defined the junction as (+) orientation if it reads from the centromere to the

582 telomere, and as (-) if it reads from the telomere to the centromere. The IgH region is defined as

583 400kb including part of V, all D, all J segments and the entire switch region. The error bars

584 represent standard deviations. (C) Ratio between total number of junctions mapped to (-) strand

585 of IgH and (+) strand of IgH, The p <0.005 **, p<0.05 *, n.s. (Not significant) for unpaired

586 student’s t-test between the IgH(-) prey%. (D) Frequency of HTGTS junctions in the (-) and (+)

587 strand orientation, and in the IgH regions, subdivided into Sμ, Sγ1, and Sε. The data represent

588 the sum of data from 5 DNA-PKcs5A/5A mice, 3 DNA-PKcs+/+ mice (WT), 4 DNA-PKcs-/- and a

589 total of 4 (2 each) Tp53-/- or Tp53+/- mice. The p <0.001 ***, n.s. (Not significant) for unpaired

590 student’s t-test. (E) Frequency of junctions with microhomologies (MH), Blunt (BLU), and

591 insertions (INS). The p <0.0001 ****, n.s. (Not significant) for Kolmogorov-Smirnov test.

592

593 Figure 3: Distribution of CSR junctions within 10kb of the Sμ locus

594 The schematic of the region of interest is drawn at the top. For each genotype, the number of

595 junctions is indicated for the (+) strand (blue) and (-) strand (red). The (-) orientation junctions

596 are shown as negative percentages. Percent of junctional usage from the end of Sμ to the end of

597 the 10kb window (between chr12:114661500 – 114665500) is indicated with red arrows.

598 Junctions in this range are considered resections. Bin=50bp. Only the percentage of negative

599 strand junctions was used to calculate the resection %. The data represent the pool of 5 DNA-

600 PKcs5A/5A mice, 3 DNA-PKcs+/+ mice (WT), 4 DNA-PKcs-/- mice, and a total of 4 (2 each) Tp53-/-

601 or Tp53+/- mice.

602 Figure 4: Distribution of CSR junctions within 30kb of the Sγ1 locus

603 The schematic of the region of interest is drawn at the top. For each genotype, the number of

604 junctions is indicated for the (+) orientation (blue) and (-) orientation (red). The (-) orientation

605 junctions are shown as negative percentages. Percent of junctional usage from the end of Sγ1 to

606 the end of the 30kb window (between chr12: 114558000 – 11456600) is indicated with red

607 arrows (bin=100bp). Junctions in this range are considered resections. Only the percentage of

608 negative strand junctions was used to calculate the resection %. The data represent the pool of 5

609 DNA-PKcs5A/5A mice, 3 DNA-PKcs+/+ mice (WT), 4 DNA-PKcs-/- mice, and a total of 4 (2 each)

610 Tp53-/- or Tp53+/- mice.

611 Figure 5: Distribution of CSR junctions within 10kb of the Sε locus

612 The schematic of the region of interest is drawn at the top. For each genotype, the number of

613 junctions is indicated for the (+) strand (blue) and (-) strand (red). The (-) orientation junctions

614 are shown as negative percentages. Percent of junctional usage from the end of Sε to the end of

615 the 10kb window (between chr12:11450700 – 114513000 ) is indicated with red arrows (50bp

616 bin). Junctions in this range are considered resections. Only the percentage of negative strand

617 junctions was used to calculate the resection %. The data represent the pool of 5 DNA-PKcs5A/5A

618 mice, 3 DNA-PKcs+/+ mice (WT), 4 DNA-PKcs-/- mice, and a total of 4 (2 each) Tp53-/- or

619 Tp53+/- mice.

620 Figure 6: DNA-PKcs5A/5A HTGTS CSR junctions do not have a strong mutational pattern in

621 the IgH region.

622 (A) Diagram of the bait site and nucleotide counting scheme for mutational analysis (left). For

623 mutation analyses, we counted the mutations after the nested PCR primer until the end of the bait

624 sequence. There is a clear difference between mutation frequency in IgH junctions vs non-IgH

625 junctions even if only the bait sequence is compared, so only IgH junctions (both + and –

626 orientation) were counted. The number of independent libraries, reads, and nucleotides analyzed

627 for mutational distribution is shown on the right. (B) Distribution of nucleotide (A, T, C, or, G)

628 frequency in the bait portion used to calculate mutation frequency. (C) Frequency of mutations

629 of each nucleotide (A, T, C, or G) in all mutations obtained. (D) The overall frequency of

630 mutations per 10,000 nucleotides (nt). (E) Frequency of all IgH junctions (reads) containing >0

631 mutations. (F) Mutations in a particular nucleotide per 10,000 of that nucleotide type. The error

632 bars represent standard errors. The data represent the collection from 4 DNA-PKcs5A/5A mice, 4

633 DNA-PKcs+/+ mice (WT), 4 DNA-PKcs-/- and 2 Tp53+/- mice.

634

635

636 Footnotes

637 1) Grant Support

638 This work is in part supported by NIH/NCI 5R01CA158073, 5R01CA184187, and

639 R01CA226852. SZ is the recipient of the Leukemia Lymphoma Society Scholar Award. JC was

640 supported by NIH/NCI F31CA183504-01A1 and CTSA/NIH TL1 TR000082. This research was

641 funded in part through the NIH/NCI Cancer Center Support Grant P30CA013696 to the Herbert

642 Irving Comprehensive Cancer Center (HICCC) of Columbia University.

643 2) Author Contributions

644 JC, XW, SZ designed experiments. SZ wrote the paper with help from XW and JC. JC and XW

645 analyzed the DNA-PKcs 5A mice and cells for lymphocyte development and function. JC

646 generated the DNA-PKcs 5A mice. XW, JC, and BL established the HTGTS assay and

647 performed the HTGTS analyses of the CSR junctions for this study. VE helped with the breeding

648 and animal care.

649 3) Abbreviations

650 Alt-EJ: alternative end-joining

651 cNHEJ: classical non-homologous end-joining

652 CSR: class switch recombination

653 DNA-PK: DNA-dependent protein kinase

654 DSB: double-strand break

655 HTGTS: High throughput genomic translocation sequencing

656 MH: micro-homology

657 SHM: somatic hypermutation

1 PPP PP P PP P 4128 DNA-PKcs PI3K

T2609* S2612 S2620 T2638* T2647* Human veTQaSQgtlqtrTQegslsarwpvagqiraTQqqhdftlTQtadgrss Mouse ieTQaspsilhTQTQegplsdqrqkpgqvraTQqqydftpTQasverss T2605 S2614/16 T2634 T2643

Bone marrow Spleen B +/+ PKcs - DNA +/+ 5A/5A Tp53 PKcs - - / - DNA Tp53 Gr1 Gr1 B220 B220 CD4 IgM CD19 IgM CD19 CD8 C Day3 Day4 D 45 40 35 30 CTRL 25 20 15 % B220+IgG1+ 5A/5A 5A/5A -

/ 10 - 5 PKcs -

Tp53 0 Ave D3 Ave D4

DNA Ctrl (n=3) DNA-PK 5A/5A p53+/- (-/-) (n=4) B220 IgG1 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 2 A (+) (+) JH4 Cen C C1 C3 Cµ (-) (-) S S1 30k Sµ 10k 10k N=3 N=3 N=3 Bait site N=2 N=2 N=2 N=3 B C D 100% 100% ✱✱ n.s. *** 4% 3% 4% 6% n.s. 3% 3% 4% 5% ✱ 10% 11% 10 17% 18% 22% 23% 75% nIgH+ 8 75% 28%

) ns 28% Smu- + nIgH- (

H 6 Smu+ g

I 20%

/ 15% Sg1- IgH+ ) 50% - 50% ( 4 Sg1+

83% 82% H

IgH- g 18% Se-

75% I 68% 2 Se+ 25% 25% 26% 43% 42%

Freqyency among all Prey sites all Prey among Freqyency 0

- - sites Prey all IgH among Freqyency / A -/ 27% L + /5 A H 3 5 K n 5 P T p K 9% 0% W P 0% WT p53+/- PK-5A PK WT nHLp53+/-PK-5APK Null nHL Null

E 50% WT nHL **** ） ****

% DNA-PKcs -/- IgH 40% **** （ PK 5A/5A 30% p53+/-

20%

10% Frequency IgH Prey Frequency in 0% >15 10-14 8-9 6-7 4-5 2-3 1 BLU 1 2 3-4 5-6 7-8 9-10 >11

microhomology (bp) insertion (bp) bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 3 dsDNA AGCT AGCT Motif/50bp

Frequency of 114661500 114663500 114665500 114667500 114669500 114671500 2% 197 0% +/+ 905 -2% Bait site PKcs

- -4% 10.7% -6% DNA -8% 2% 506 0%

- 1501

+/ -2%

P53 -4% 9.86%

-6% junctions (50bp bin) (50bp junctions

μ -8% 2% - / - 1154 0% 2035 PKcs

- -2%

-4% DNA 29.4% -6%

-8%

Frequency S within prey of Frequency 2% 717

5A/5A 0% 1704 -2% PKcs - -4% 15.3% DNA -6%

-8% + strand - strand bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 4

114558000 114,565,700.00 ~12kb 114588000

10 RGYW 5 motif/100bp

0 motif/100bp dsDNA motif/100bp

Frequency of RGWY RGWY Frequency of 114558000 114563000 114568000 114573000 114578000 114583000 114588000 1% 192

+/+ 0% 2173 PKcs

- -1% 0.14%

DNA -2%

-3% 1% 547 0% -

+/ 6098 -1% P53 0.54% -2% junctions (100bp bin) (100bp junctions γ1 -3% 1% - / - 539 0%

PKcs 6194 - -1% DNA -2% 2.21%

Frequency S within prey of Frequency -3% 1% 285 0%

5A/5A 4348 -1% PKcs - -2% 1.66% DNA -3% + strand - strand bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 10kb Figure 5

114507000 ~.4.2 kb114517000

Sε

114507000 114509000 114511000 114513000 114515000 114517000 1% 53 0% +/+ 1238 -1% PKcs - -2%

DNA 6.14% -3%

-4% 1% 145 0% - +/ -1% 2655

P53 -2%

-3% junctions (50bp bin) 5.24%

Sε -4% 1%

- 35 / - 0% 498 -1% PKcs -

-2% DNA -3% 13.3% -4%

Frequency of Frequency withinprey 1% 34 0%

5A/5A 1260 -1%

PKcs -2% -

-3% 9.13% DNA -4% + strand - strand bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Figure 6 A. count mut in the Sµ bait - strand CEN

Prey Barcode Translocation Targets Sµ Bait Nested primers B. C. 100% 100%

90% 90%

80% 80% A% A-> 70% 70%

counted T-> T % 60% 60% seq C-> C% 50% 50% G-> G% 40% 40% Nucleotide/Mut 30% 30% 20%

Composition for GL Composition 20% 10% 10% 0% 0% WT P53 def PK KO 5A/5A D. WT P53 def PK KO PK 5A E. 35 12% 30 10% 25 8% 20 6% 15 4% 10 Mut/10000 nt Mut/10000 5 2% % of reads with mut reads% of 0 0% WT P53 def PK KO PK 5A WT P53 def PK KO PK 5A

60 F. WT 50 P53 def PK 5A 40 PK KO 30

NT Mut/10000 dNTP NT Mut/10000 10

0 A-> C-> G-> T-> bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.SupFigure 1

A B *

DNA-PKcs: +/+ 5A/5A 5A/5A 5A/5A DNA-PK+/+ DNA-PK5A/5A DNA-PK5A/5A p53: +/+ +/+ +/- -/- p53+/-

C CTRL 5A/5A 5A/5A Tp53-/- B220 B220 DNA-PKcs IgG1 CTV bioRxiv preprint doi: https://doi.org/10.1101/2020.04.23.057877; this version posted April 24, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.SupFigure 2

A B

45% 60% DNA-PKcs+/+ 40% ChrTB/MP 50% 35% DNA-PKcs5A/5A (1) ChrSb/MP 30% 40% DNA-PKcs5A/5A (2) 25% 30% 20%

15% 20% Abnormality/MP (%) Abnormality/MP

Abnormal Metaphase (%) Abnormal Metaphase 10% 10% 5%

0% 0%

C Normal Chromatid Chromosomal Metaphase Break Break