Fusion of histone variants to Cas9 enhances homology-directed repair

As a versatile genome editing tool, the CRISPR-Cas9 system induces DNA double-strand breaks at targeted sites to activate mainly two DNA repair pathways: HDR which allows precise editing via recombination with a homologous template DNA, and NHEJ which connects two ends of the broken DNA, which is often accompanied by random insertions and deletions. Therefore, how to enhance HDR while suppressing NHEJ is a key to successful applications that require precise genome editing. Histones are small proteins with a lot of basic amino acids that generate electrostatic affinity to DNA. Since H2A.X is involved in DNA repair processes, we fused H2A.X to Cas9 and found that this fusion protein could improve the HDR/NHEJ ratio. As various post-translational modifications of H2A.X play roles in the regulation of DNA repair, we also fused H2A.X mimicry variants to replicate these post-translational modifications including phosphorylation, methylation, and acetylation. However, none of them were effective to improve the HDR/NHEJ ratio. We further fused other histone variants to Cas9 and found that H2A.1 exhibited the improved HDR/NHEJ ratio better than H2A.X. Thus, the fusion of histone variants to Cas9 is a promising option to enhance precise genome editing.


Introduction
Histones are proteins that constitute eukaryotic chromosomes and have five (H2A.X phosphorylated at S139) in response to DNA double-strand breaks (DSBs) (5). 49 Then, mediator of DNA damage protein checkpoint protein 1 (MDC1) binds to γH2A.X 50 to initiate the DNA repair process by recruiting various DNA repair factors (6). K134 51 dimethylation by the histone methyltransferase SUV39H2 is also correlated with γH2A.X.

52
The K134A mutation that prevents this dimethylation reduces the expression of γH2A.X 53 (7). In addition, H2A.X acetylated at the lysine (K) 5 by TIP60 histone acetylase is 54 released from chromatin in DNA damage sites and binds to DNA damage response factors 55 to modulate DNA repair response (8-10). Thus, various post-translational modifications 56 play important roles in DNA repair.

57
DNA repair in response to DSBs mainly relies on two pathways: homology-58 directed repair (HDR) mediated by recombination with a homologous template that yields 59 precise repair products identical to the DNA sequence of the template, and non-60 homologous end joining (NHEJ) that brings the two broken DNA ends together often 61 with random insertions or deletions (11). However, mammalian cells preferentially adopt 62 NHEJ over HDR by the following mechanisms: NHEJ is active through the cell cycle, 63 whereas HDR is restricted to the S/G2 phases; NHEJ is faster than HDR (12). We have 64 observed the same trend in genome editing by CRISPR-Cas9 (13). Therefore, strategies 65 to enhance HDR over NHEJ are required. 66 Here, we fused H2A.X to Cas9 to see if the HDR activity could be enhanced. In further improve the HDR/NHEJ ratio. We also fused other H2A and H3 variants to Cas9 70 and found that some of them improved the HDR/NHEJ ratio. The transfection experiments were performed in triplicates (three biological replicates).

75
Statistical significance was assessed by a two-tailed Student's t-test to compare the 76 differences between two different conditions.   The digital PCR-based assay to detect the HDR and NHEJ activities was described 120 previously (13). In this study, we used RBM20-2, RBM20-g1, GRN-2, and GRN-g2 121 gRNAs described previously (16). Therefore, we used ddPCR primers and probe sets of 122 RBM20 Assay 3 (for RBM20-2), Assay 1 (for RBM20-g1), and GRN Assay 2 (for GRN-123 2 and GRN-g2) described previously (16). . By using N-GS and guide RNAs (gRNAs) previously designed (RBM20-2, RBM20-132 g1, GRN-2, and GRN-g2, (16)), we introduced two pathogenic point mutations: RBM20 133 R636S and GRN R493X in HEK293FT cells (Fig 1B). We found that N-GS induced less 134 NHEJ compared to the normal Cas9 while keeping the HDR level comparable with 135 RBM20-2 and GRN-g2 (Fig 1C, S4 Table). These results indicated that the fusion of 9 136 H2A.X to Cas9 could enhance the HDR/NHEJ ratio and prompted us to further optimize 137 the design of the fusion.  The arrows indicate the transcription start sites. (GGGGS) 5 linker to tether H2A.X to the N-or C-terminus of Cas9 (Fig 1A). We named 154 these fusion proteins N-GS, N-GS3, N-GS5, C-GS, C-GS3, and C-GS5 depending on the 155 length and position of the linkers (Fig 1A). We examined the HDR and NHEJ activities 156 of these fusion proteins. We found that most of the fusion proteins improved the 157 HDR/NHEJ ratio compared to using Cas9 alone, but N-GS3 was the best overall ( Fig 1C,   158 S4 Table). Therefore, we decided to further modify and improve N-GS3.

160
Mimicry variants of H2A.X S139 phosphorylation or K134 methylation did not 161 improve the HDR/NHEJ ratio.

162
The post-translational modifications of H2A.X have been reported to be involved in DNA 163 damage repair (5,17). In particular, γH2A.X (H2A.X phosphorylated at S139) is the most 164 well-known marker of DNA damage and functions as a platform for the recruitment of 165 DNA damage response (DDR) signaling factors, but its specific involvement in the HDR 166 and/or NHEJ pathways has not yet been reported. Therefore, we generated an S139D 167 phosphorylation mimic mutant of H2A.X fused to Cas9 (SD-Cas9, Fig 2A). However, 168 the HDR and NHEJ activities of SD-Cas9 were comparable to those of N-GS3 with 169 RBM20-2, RBM20-g1, and GRN-2. With GRN-g2, both HDR and NHEJ activities of SD-Cas9 were increased compared to N-GS3, but the HDR/NHEJ ratio was still 171 comparable ( Fig 2B, S5 Table). We also fused Cas9 and H2A.X with the S139A non-172 phosphorylatable mutation (SA-Cas9), but the HDR/NHEJ ratio was not significantly 173 altered by SA-Cas9 either (Fig 2B, S5 Table). It has been reported that dimethylated K134 174 is critical for H2A.X S139 phosphorylation (7), although it is still debatable (18). Cas9, and KA_SD-Cas9, respectively (Fig 2A). We measured the HDR and NHEJ 182 activities of these fusion proteins with the four gRNAs. However, the HDR/NHEJ ratios 183 of all the fusion Cas9s with H2A.X with the post-translational modification mimic 184 mutations were comparable to that of N-GS3 (Fig 2B, S5 Table). prevents the accumulation of the DNA repair factors (9) (Fig 2A). To investigate whether 201 H2A.X mimicries of acetylation of K5 can alter the balance of HDR and NHEJ, we 202 mutated K5 to glutamine (K5Q, KQ) as an acetylation mimicry variant, and K5 to 203 arginine (K5R, KR) as a non-acetylatable H2A.X variant, respectively (Fig 3A). We 13 204 found that the NHEJ activity in KQ-Cas9 was slightly increased compared to N-GS3 with 205 GRN-2 and GRN-g2 gRNAs, but no such trend was observed with RBM20-2 and 206 RBM20-g1 gRNAs. The NHEJ activity in KR-Cas9 was slightly decreased compared to 207 N-GS3 with RBM20-g1, but this trend was not observed with the other gRNAs (Fig 3B,   208 S6 Table). These results overall indicate that the fusion of Cas9 with a mimicry of K5 209 acetylation or a non-acetylatable variant of H2A.X did not result in an improvement of 210 the HDR activity. the H2A variants and H2B, H2A.1-Cas9 showed decreased NHEJ with RBM20-2, 232 RBM20-g1, and GRN-2 gRNAs, but comparable HDR with RBM20-2 and GRN-2 233 gRNAs compared to N-GS3 (Fig 4B, S7 Table).  We also tested Cas9 fusion proteins with the H3 variants. As for the H3 variants, H3.1, 253 H3.2, and H3.3 differ from H3 by only a few amino acid residues (Fig 4C). Compared to 254 N-GS3, H3.3 showed decreased NHEJ with all gRNAs and also decreased HDR with 16 255 RBM20-g1 and GRN-2 gRNAs (Fig 4D, S7 Table). These suggested that H2A.1 is the 256 best option to achieve the highest HDR/NHEJ ratio by fusion of Cas9 and histone variants 257 tested in this study. In this study, we initially found that H2A.X tethered to Cas9 with GGGGS 261 linkers improved the HDR/NHEJ ratio compared to Cas9 alone. Therefore, we 262 investigated whether mimicries of post-translational modifications of H2A.X could 263 further improve the HDR/NHEJ ratio, but none of them were effective. However, we 264 found that the H2A.1 variant improved the HDR/NHEJ ratio better than H2A.X when 265 fused to Cas9. There have been several reports of the fusion of HDR factors with Cas9 to 266 increase the HDR activity (15,20-23), but this is the first report that the fusion of histones 267 to Cas9 can improve the HDR/NHEJ ratio. 268 S139 phosphorylated H2A.X (γH2A.X) rapidly accumulates at the sites of DNA 269 damage and plays a role in DNA repair (5,17). Therefore, in this study, we generated a 270 mimicry variant of γH2A.X by substituting S139 with an aspartic acid and fusing it to 271 Cas9 (SD-Cas9) (Fig 2A). However, unfortunately, we found that SD-Cas9 did not 17 272 improve the HDR/NHEJ ratio (Fig 2B, S5 Table). We also examined whether the fusion 273 of Cas9 and H2A.X methylation mimicry improves the HDR/NHEJ ratio since it is known 274 that K134 of H2A.X is dimethylated by SUV39H2, resulting in the γH2A.X production 275 (7). Therefore, we generated fusion proteins of Cas9 with H2A.X variants of K134 276 methylation mimicry and non-methylatable mutant of K134 (Fig 2A). However, these 277 fusion proteins were not effective compared to N-GS3 either (Fig 2B, S5 Table).

278
In addition, since acetylation at K5 of H2A.X is important for assembling DNA 279 repair proteins to damaged sites (9), we tested whether the fusion of Cas9 with mimicry 280 of K5 acetylation of H2A.X, playing this role, could enhance the HDR activity. Contrary 281 to our expectations, however, the fusion of Cas9 with mimicry of K5 acetylation did not 282 directly improve the HDR/NHEJ ratio. Acetylation at K5 and phosphorylation at S139 of 283 H2A.X are important components of the cellular response to DNA damage. Further 284 studies are necessary to understand how these post-translational modifications of H2A.X 285 are involved in DNA repair and apply this knowledge to improving precise genome 286 editing.

287
As mentioned above, H2A.X was known to accumulate at damaged DNA sites 288 after phosphorylation and be responsible for DNA repair, but little is known about other 18 289 histone variants. In this study, we discovered that H2A.1 improved the HDR/NHEJ ratio 290 compared to other histone variants (Fig 4, S7 Table). The component ratios of H2A.1 and 291 H2A.2 are known to change with aging and differentiation in rat liver tissue and human 292 fibroblasts (24,25). However, the improvement of the HDR activity with H2A.1 was 293 found for the first time in this study, suggesting a previously unknown role in DNA repair 294 for this histone variant.

295
In conclusion, we found that the fusion of histone variants H2A.1 improves the 296 HDR/NHEJ ratio induced by CRISPR-Cas9. These findings will lead to the development 297 of more precise genome editing platforms.