Microhomology-mediated CRISPR/Cas9-based method for genome editing in fission yeast

The CRISPR/Cas9 system enables the editing of genomes of numerous organisms through the induction of the double-strand breaks (DSB) at specific chromosomal targets. We improved the CRISPR/Cas9 system to ease the direct introduction of a point mutation or a tagging sequence into the chromosome by combining it with the microhomology mediated end joining (MMEJ)-based genome editing in fission yeast. We constructed convenient cloning vectors, which possessed a guide RNA (gRNA) expression module, or the humanized Streptococcus pyogenes Cas9 gene that is expressed under the control of an inducible promoter to avoid the needless expression, or both a gRNA and Cas9 gene. Using this system, we attempted the MMEJ-mediated genome editing and found that the MMEJ-mediated method provides high-frequency genome editing at target loci without the need of a long donor DNA. Using short oligonucleotides, we successfully introduced point mutations into two target genes at high frequency. We also precisely integrated the sequences for epitope and GFP tagging using donor DNA possessing microhomology into the target loci, which enabled us to obtain cells expressing N-terminally tagged fusion proteins. This system could expedite genome editing in fission yeast, and could be applicable to other organisms.

The cells that did not grow on each EMM selective medium plates were selected and 205 stocked. For swi6-W104A mutagenesis, the colonies were streaked on YES plates 206 containing of 10 µg/mL of thiabendazol (TBZ) to evaluate the TBZ sensitivity. Using the 207 cells exhibiting TBZ sensitivity, colony PCR was performed to amplify the region of swi6 208 gene with primers KT1957-KT1958 and the PCR products were subjected to sequencing. 209 For mrc1-S604A mutagenesis and knock-in experiments, all procedures were performed 210 similar to in swi6 mutagenesis experiment, except for the confirmation of the mutant 211 phenotype and the insertion confirmation by microscopic observation and electrophoresis. 212 Since mrc1 mutants exhibit hydroxyurea (HU) sensitivity (Tanaka and Russell 2001), the 213 transformants were streaked on YES plates containing 5 mM HU to identify the 214 mrc1-deficient cells introduced by the MMEJ-mediated genome editing. HU-sensitive cells 215 were used as templates for colony PCR, and PCR products amplified with primers 216 KT27-KT90 were subjected to sequencing. For knock-in at the 5′ end of the stn1 + gene, 217 colony PCR products amplified from the region near the first ATG with primers 218 KT2049-KT2050 were separated by 2 % agarose gel electrophoresis to check the Flag 219 insertion. PCR products were sequenced to confirm the in-frame 2 × Flag insertion. For 220 GFP knock-in at the 5′ end of the reb1 + gene, transformants were observed by fluorescent 221 microscopy to see the GFP signals. Colony PCR products amplified from the region near 222 the first ATG with primers KT2125-2126 were separated by 0.8 % agarose gel 223 electrophoresis to check the insertion and subjected to sequencing. 224 For genome editing with a single vector expressing both a gRNA and the Cas9 225 protein, the cells cultured in EMM5S -thiamine medium were subjected to transformation. 226 Transformation was carried out as described above. To obtain a homogenous mutant 227 population, the transformants were streaked on EMM-selective medium plates (without 228 thiamine) and incubated at 32 ˚C for 3 days to select the cells carrying the vector. The cells 229 were streaked on YES plates at 32 ˚C for 2 days to get single colonies. To confirm the loss 230 of vector, single colonies were streaked on YES and EMM-selective medium plates, 231 followed by colony PCR and sequencing as described above. 232

Immunoblot analysis 233
The protein extracts were prepared using alkaline-TCA method as previously 234 described (Knop et al. 1999). 1 × 10 7 cells growing in early log phase were harvested and 235 used to prepare the cell extracts followed by Immunoblot analysis. The 3 × Flag tagged 236 Cas9 protein and the 2 × Flag tagged Stn1 were detected using monoclonal anti-Flag M2 237 antibody conjugated with HRP (Sigma, A8592). Anti-Cdc2 (Santa Cruz, SASC53) was 238 used as a loading control. 239

Microscopy 240
The preparation of living cells expressing GFP fused proteins was carried out as 241 described previously

RESULTS 254
The improved CRISPR/Cas9 system efficiently introduces mutations in fission yeast 255 To simplify genome editing using the CRISPR/Cas9 system in fission yeast, we 256 modified the gRNA expression vector described in Jacob et al. NLSs (Ran et al. 2013) was cloned into pSLF273 (nmt41 promoter, ura4 + marker) vector 274 to be expressed under the control of the nmt41 promoter ( Figure 1B, pAH235). We further 275 generated a vector expressing both of a gRNA and the Cas9 protein to modify the genome 276 through a single transformation using one marker gene, either LEU2 or ura4 + (pAH237: 277 LEU2 marker, pAH243: ura4 + marker, Figure 1C and 1D). To ensure Cas9 expression, we 278 generated the vectors that expressed the Cas9 using the nmt41 promoter on each marker 279 gene's vector. Immunoblot analysis was performed using anti-Flag antibodies to validate 280 the expression of the Cas9 protein in the cells carrying pAH233 in the presence or absence 281 of thiamine ( Figure 2A). The result showed that the expression of the Cas9 protein by the 282 nmt41 promoter was induced in the absence of thiamine, whereas its expression was 283 repressed in the presence of thiamine ( Figure 2A). We also analyzed the expression of the 284 Cas9 protein from pAH237 and pAH243 in the absence of thiamine and observe that the 285 Cas9 protein was stably expressed by the nmt41 promoter ( Figure 2B). 286 To examine the efficiency of the improved CRISPR/Cas9 system, we edit the ade6 + 287 gene. The ade6 gRNA target sequence was designed near ade6-M210 mutation site to 288 evaluate the mutation efficiency based on the frequency of the appearance of red color 289 colonies (Jacobs et al. 2014). At the beginning, we tested a combination of ade6-gRNA 290 vector (ade6-gRNA-LEU2, pAH244) and inducible Cas9 vector (nmt41p-Cas9-ura4 + , 291 pAH235). We sequentially transformed the wild type of ade6 + strain with the inducible 292  (Table 1 and Figure S1A). We sequenced the ade6 gene in eight 300 red-colored transformants and found 1-3 bp indels in the ade6 gene (Table S4), which 301 indicates that the CRISPR/Cas9 system efficiently induced an ade6 mutation comparable 302 with that of a previously reported system (Jacobs et al. 2014). We also transformed 303 wild-type cells expressing the Cas9 protein with the rrk1 promoter-CspCI-ade6 gRNA 304 cassette that has previously been used (pMZ284, derived from pMZ283, Addgene ID 305 52225) and cloned them into a LEU2 marker vector (pAH234; pRE-pMZ284) to validate 306 our modified system. The transformants carrying pRE-pMZ284 exhibited slow growth 307 23 ( Figure S2). It suggests that the cells had growth defects compared with the cells carrying 308 pAH244, while the efficiency of ade6 mutagenesis is 97 % (Table 1). 309 Subsequently, we transformed the cells with the single vector expressing both the 310 ade6-gRNA and the Cas9 protein. The wild-type cells were cultured in EMM5S without 311 thiamine before the transformation. The cells were transformed with either the LEU2-or 312 ura4-marked single vector and plated on EMM selective medium plates for 6-7 days. 313 Transformants were streaked on YE plates to observe the colors of their colonies. The 314 results revealed that the transformation with the LEU2-marked vector introduced the 315 mutation at a high frequency (83 %, Table 2), whereas the transformation with the 316 ura4-marked vector introduced the mutation at low frequency (13 %, Table 2 (1/10, Figure 3D and Table S4). However, most of the mutations introduced by the 15 bp 368 homologous oligonucleotides were indel mutations, and some of mutants had 369 rearrangements near the cleavage site (Table S4). Notably, 20 bp and 25 bp homologous 370 oligonucleotides introduced swi6-W104A mutations at 70 % (7/10) and 100 % (8/8) of the 371 frequency, respectively ( Figure 3D). As expected, PCR fragments containing swi6-W104A 372 mutation could introduce an appropriate mutation into the chromosome at 100 % (16/16) 373 frequency. Therefore, even using very short oligonucleotides, the MMEJ pathway can 374 introduce mutations at high frequency, which is comparable with the HR pathway. We also 375 designed another swi6 gRNA vector (+51 bp far from swi6-W104 position, #5 in Figure  376 3D) and tested the frequency of introduction of swi6-W104A mutation by the HR pathway 377 using same PCR products ( Figure 3D). However, the mutation frequency was not high  Figure 3F and Table S4). The ssOligoFw introduced the mutation at higher frequency than 391 the ssOligoRv, implying a difference in repair pathways occurred within the sense and 392 antisense strands. Remarkably, the mixed oligos and dsOligo could highly introduce the 393 swi6-W104A mutation at 90 % and 100 % frequency ( Figure 3F and Table S4). 394

29
To investigate the efficiency of the MMEJ-mediated mutagenesis at other loci of the 395 chromosomes, mrc1 + gene was mutagenized by oligonucleotides using a similar procedure. 396 The mrc1 + gene encodes mediator of replication checkpoint protein 1, which is required for 397 Rad3-dependent activation of checkpoint kinase Cds1 in response to replication fork arrest 398 (Tanaka and Russell 2001) ( Figure 4A). Mrc1-S604 is located in SQ repeats, which are 399 potential substrates of Rad3/Tel1 kinase, and the mrc1-S604A mutant exhibited HU 400 sensitivity due to reduced Cds1-T11 phosphorylation, which is required for the interaction 401  Figure 4D). The frequency of the transformants exhibiting HU sensitivity was very high 407 (96 % of transformants (23/24) exhibited HU sensitivity). Sequence analysis was 408 performed using colony PCR products to determine the mutation sites of the HU-sensitive 409 30 transformants. The results showed that the frequency was as high as that for the swi6 410 mutagenesis: 60 % for ssOligoFw, 20 % for ssOligoRv, 90 % for mixed ssOligos, and 411 100 % for dsOligo ( Figure 4C and Table S4 (88 % and 96 % frequency, respectively, Figure 5C). In contrast, ssOligoFw and ssOligoRv 434 introduced the insertion but at low frequency (21 % and 4 %, respectively, Figure 5C). 435 Sequence analysis showed that the frequency of in-frame knock-in by transformation with 436 mixed ssOligos and annealed dsOligo were high (82 % and 89 %, respectively, Figure 5C). 437 Some of the knock-in strains contained the in-frame 3 × Flag insertion when the cells were 438 transformed with the mixed ssOligos and dsOligo. Although the insertion frequency was 439 32 low, ssOligoFw also introduced in-frame knock-in at high fidelity (80 % frequency, Figure  440 5C). We did not obtain the knock-in strain by the transformation with ssOligoRv ( Figure  441 5C). We also found that the frequency of in-frame Flag tag insertion without indel was high 442 (89 %) in the transformants obtained by transformation with dsOligo ( Figure 5C and Table  443 S4). The results suggest that the transformation with dsOligo would lead to precise 444 knock-in without indel when using the MMEJ-mediated CRISPR/Cas9 system. We 445 performed immunoblot analysis to confirm the expression of 2 × Flag-Stn1 protein. Flag 446 epitope tag fused Stn1 protein was detected with anti-Flag antibody ( Figure 5D), indicating 447 that Flag epitope sequence was efficiently and precisely introduced at the 5′ end of the 448 stn1 + gene. 449 We further attempted to perform GFP gene knock-in at the 5′ end of the reb1 + gene 450 using MMEJ-mediated knock-in. Reb1 is an rDNA-binding protein that is required for the 451 termination of the transcription by RNA polymerase I (Zhao et al. 1997) ( Figure 5E). Reb1 452 fused to GFP at the C terminus is located in the nucleus and the nucleolus (Hayashi et al. 453 2009). To generate a knock-in strain expressing GFP-fused Reb1 protein, we prepared a 454 33 PCR fragment with GFP sandwiched between 24 bp upstream of the first ATG of the reb1 + 455 gene and 24 bp downstream of the first ATG of the reb1 + gene ( Figure 5E). Endonuclease 456 activity of the reb1 gRNA/Cas9 complex was evaluated on the basis of the frequency of the 457 mutations introduced in the reb1 + gene. We performed a sequence analysis of colony PCR 458 products amplified near the first ATG of the reb1 + gene and found that transformants had 459 mutations in the reb1 gene at a 75 % frequency (6/8, Table S4). The cells carrying the Cas9 460 vector were co-transformed with both of the reb1 gRNA vector and PCR donor fragment, 461 and transformants were analyzed as described above. The result showed that 7 out of 48 462 transformants had GFP gene insertion at a frequency of 15 % ( Figure 5F and 5G). The 463 heated denatured PCR fragment (single-stranded PCR product) could induce GFP insertion, 464 but the frequency was lower than that of the double-stranded PCR fragment (2/24, 8 %, 465 regions of homology requires serial PCR steps and, therefore, is thus time-consuming. 511 Using long (~100 bp) oligonucleotides for PCR can save the time required for preparation 512 of a donor DNA. However, the cost of synthesizing long oligonucleotides is high, and the 513 shorter homologous sequence lengths result in lower targeting efficiency. The 514 MMEJ-mediated genome editing method reported here could save time and costs by using 515 short synthesized oligonucleotides as a donor DNA. We demonstrated that the MMEJ 516 38 pathway is highly effective for the introduction of point mutations into two genes using the 517 CRISPR/Cas9 system with ~50 bp of short oligonucleotides. 518 As a donor DNA, double-stranded DNA seems improve the efficiency of the 519 MMEJ-mediated genome editing. The annealed swi6-W104A oligonucleotides possessing 520 25 bp homologous sequences demonstrated higher frequencies during the introduction of 521 the swi6-W104A mutation (100 %, Figure 3D) than the ssOligos possessing 25 bp 522 homologous sequences (ssOligoFw: 71 %, ssOligoRv: 50 %, Figure 3F). The experiments 523 that introduced the mrc1-S604A mutation showed that the dsOligo possessing 25 bp of 524 homologous sequences edited the genome at a high frequency (100 %), in contrast, 525 ssOligos introduced the mrc1-S604A mutation at a 60 % (ssOligoFw) and 20 % 526 (ssOligoRv) frequency ( Figure 4C). For 2× Flag knock-in at the 5′ end of the stn1 + gene, 527 the dsOligo possessing 25 bp of homologous sequences at both ends was precisely inserted 528 at a high frequency (89 %, Figure 5C), however, both ssOligos introduced an in-frame 529 insertion at low frequency (ssOligoFw: 21 %, ssOligoRv: 0 %, Figure 5C). These results 530 39 suggest that the MMEJ-mediated genome editing in fission yeast is related to the HR 531 pathway, which takes place at G2 phase. 532 The frequency of GFP tagging, however, is lower (15 %, Figure 5G) than the 533 introducing frequency of the 2 × Flag knock-in at the 5′ end of the stn1 tagging. The length 534 of GFP gene is about 700 bp, however, which is much longer than the homologous 535 sequences required for annealing to the target genomic loci (24 bp each). Compared with 536 GFP fragment, the 2 × Flag the length of the 2× Flag repeats is 54 bp, and the homologous 537 sequences used for annealing were bidirectional 25 bp each. It highlights the significance of 538 the annealing efficiency. Although the molecular mechanism of the MMEJ pathway 539 requires further elucidation, the combination of the MMEJ and the CRISPR/Cas9 system in 540 fission yeast could be carried out using both of the HR and the SSA pathways (Decottignies 541 2007). 542 In addition to the annealing efficiency, the amount of a donor DNA transformed into 543 a cell may influence the efficiency of knock-in at the target locus of the chromosome. 544 GFP-reb1 PCR product is 763 bp, and we did not obtain any insertion when we used 545 100−200 µg of PCR product for transformation. As for the HR introduced swi6-W104A 546 mutation, 65-130 µg of PCR product (575 bp) was adequate for obtaining the mutant (swi6 547 gRNA #2 vector, Figure 3D). The observation could also be explained by the high 548 annealing efficiency of donor DNA since swi6-W104A PCR product has over 250 bp 549 homologous sequence at both ends. 550 We also observed that the ssOligos could induce the genome editing, and the forward 551 oligonucleotides exhibited higher frequency of precise genome editing than the reverse 552 oligonucleotides (Figures 3F and 4C). Notably, the mixed ssOligos could induce the 553 genome editing at high frequencies comparable with dsOligo, although the efficiency of 554 indel production was slightly higher (Figures 3F and 4C and Table S4). The differences in 555 the insertion and indel frequencies occasioned by the opposite directions of ssOligos could 556 be caused by the functions of different repair/replication proteins at the microhomologous 557 region due to the different orientations of Cas9 and gRNA binding (Engstrom et al. 2009;558 Lemos et al. 2018). We cannot exclude the possibility that the orientation of the transcripts 559 could be affected to the difference of the editing frequency. It remains unclear how variably 560 the repair/replication proteins act at microhomologous sequences near cleavage sites. As for 561 the repetitive sequence insertion, we obtained 3 × Flag tagged stn1 strain by transformation 562 with both of mixed ssOligo and dsOligo ( Figure 5B and Table S4). The DNA replication 563 machinery might generate an extra copy of Flag repeat sequence after the knock-in occurs 564 in the chromosome. 565 In this paper, we demonstrate that the MMEJ-mediated genome editing combined 566 with the CRISPR/Cas9 system is a powerful tool for generating point mutations and 567 knock-in strains without a selective marker gene. It may also be possible to improve the 568 current system using the Cas9 nickase to induce single-strand breaks, which may reduce       swi6-W104. The number of the cells described in total was analyzed to determine the mutation site by sequencing. As a control donor for the HR pathway, 575 bp length PCR product containing swi6-W104A mutation was co-transformed with the swi6 gRNA #2 vector and also the swi6 gRNA #5 vector that was designed to digest at +51 bp position from swi6-W104.
E. The swi6-W104A mutant exhibits TBZ sensitivity. Five-fold-diluted cultures of the indicated strains were plated onto YES non-selective medium (N/S) and YES containing 15 μg/mL of TBZ and incubated for 2-3 days at 32 ˚C.
F. The swi6-W104A mutagenesis by ssOligoFw, ssOligoRv, and mixed ssOligos. The number indicates the ratio of the number of the swi6-W104A mutants/total analyzed cells.