Highly E cient CRISPR/Cas9 System in Plasmodium Falciparum Using Cas9-expressing Parasites and a Linear Donor Template

The CRISPR/Cas9 system is a powerful genetic engineering technology for Plasmodium falciparum. We here report further improvement of the CRISPR/Cas9 system by combining the Cas9-expressing parasite with a liner donor template DNA. The Cas9-expressing parasite was generated by inserting the cas9 gene in the genome by double crossover recombination. The site-directed mutagenesis and the fusion of fluorescence protein was achieved within two weeks with high efficiency (> 85%), by transfecting the schizonts of the Cas9-expressing parasite with the liner donor template and the plasmid carrying the sgRNAs. Notably, there were neither off-target mutations in the resultant transgenic parasites nor unexpected recombination, that are the technical problems of the current CRISPR/Cas9 system. Furthermore, with our system, two genes on different chromosomes were successfully modified in single transfection. Because of its high efficiency and robustness, our improved CRISPR/Cas9 system will become a standard technique for genetic engineering of P. falciparum, which dramatically advances future studies of this parasite.


INTRODUCTION 1
Malaria is still a global public health threat, and its burden exceeds 200 million infections every 2 year, resulting in more than 400,000 deaths annually 1 . Plasmodium falciparum is the most lethal 3 human parasite among the five human malaria parasites and is responsible for most of those deaths. 4 There is not yet an effective vaccine against P. falciparum, and drug resistance against all antimalarials 5 has already emerged in endemic areas. Thus, countermeasure against infection are urgently required. 6 Genetic engineering of this parasite is an essential technology for investigating the function of genes, 7 allowing the exploration of drug targets and vaccine antigens. Recently, the CRISPR/Cas9 system was 8 developed for P. falciparum and employed for genetic engineering 2,3 . In this system, the gene is 9 modified through two steps as follows: the targeted genomic locus is cleaved by the Cas9-single guide 10 RNA (sgRNA) complex, and the induced double-strand break is then repaired by homology-directed 11 recombination (HDR) using donor template DNA. Three essential components, i.e., the Cas9 gene, 12 sgRNA, and donor template DNA, are currently cointroduced by electroporating the ring form of the 13 parasites 4 . Alternatively, these components are preloaded into red blood cells (RBCs), which are 14 infected with parasites, and then introduced into parasites by uptake 5 . Because of the large size of the 15 Cas9 gene, these components have to be introduced using two plasmids 2,3,6-9 . However, since the 16 efficiencies of both DNA transfer methods are low, it has been challenging to cointroduce two 17 plasmids into the parasites, which is a technical obstacle for the current CRISPR/Cas9 system in P. 18 falciparum. In addition, the use of two plasmids requires the use of two kinds of drugs for selecting 19 transgenic parasites, which decreases their growth rate and thus delays the establishment of transgenic 20 parasites. Moreover, the usage of a circular plasmid causes another serious technical problem; 21 following the HDR between the cleaved genomic locus and donor template, the entire circular plasmid 22 donor templates and a plasmid with two sgRNAs, showing the applicability of our system. This 1 improved system solved the current technical problems associated with the CRISPR/Cas9 system in 2 P. falciparum and will thus be the standard method for genetic engineering of this parasite. Fully mature schizonts are known to be the developmental stage of Plasmodium parasites suitable 7 for DNA transfer. High DNA transfer efficiency can be achieved by using this stage in rodent malaria 8 parasites, such as P. berghei 12 . However, because RBCs containing fully mature schizonts of P. 9 falciparum rupture spontaneously in vitro, there are only a few numbers of full mature schizonts in 10 culture: fully mature schizonts usually account for less than 0.05% of total parasites. Thus, we enriched 11 fully mature schizonts by tightly synchronizing the cell cycle of parasites. Briefly, mature and 12 immature schizonts were purified using a Percoll-sorbitol gradient and subsequently cultured for 4 13 hours with fresh RBCs, followed by treatment with 5% sorbitol. The cell cycle of the parasites was 14 synchronized to a window of 4 hours with these procedures. We further repeated those procedures 15 two times and eventually obtained tightly synchronized parasites. The final ratio of fully mature 16 schizonts to total parasites increased to approximately 1-2%, demonstrating 20-to 40-fold enrichment 17 (Fig. 1A). Subsequently, the fully mature schizonts were purified along with immature schizonts and 18 then used for transfection. The purified parasites (1.0 × 10 8 cells) containing the fully mature schizonts 19 were electroporated with 5 µg of the centromere plasmid pFCENv1 13 ; transgenic parasites were 20 detected 10 days posttransfection, and the number of independently transfected parasites was 21 calculated to be 3.6 × 10 3 based on the multiplication rate (3.7/cell cycle) and percentage of 22 expression cassette, and its transcription was controlled by the promoter of pfhsp70 (PF3D7_0818900) 1 and the 3'-UTR of pbhsp70 (PBANKA_0711900). The nuclear localization signal and the FLAG tag 2 sequences were introduced at the N-terminus of Cas9. The Cas9 expression cassette was flanked with 3 two partial sequences of the kahrp gene (PF3D7_0202000), which was used as the target genomic 4 locus. The guide RNA (gRNA), specific for the kahrp gene, was cloned into the psgRNA1_cen 5 plasmid, which contains the centromere of P. falciparum, and was transcribed by the promoter of U6 6 spliceosomal RNA (PfU6: PF3D7_1341100) ( Fig. 2A). The resultant plasmid was named 7 psgRNA1_cen_kahrp. Twenty-five micrograms of each linearized Cas9 expression cassette and the 8 psgRNA1_cen_kahrp plasmid were cointroduced into fully mature pfcen_cas9 schizonts by 9 electroporation. Transfection experiments were carried out in duplicate to obtain biologically 10 independent transgenic parasites. Since the psgRNA1_cen_kahrp and pCen_cas9 plasmids had human 11 dihydrofolate reductase and blasticidin deaminase genes, respectively, as drug-selectable markers, 12 transgenic parasites that had two plasmids were screened by treatment with those two drugs. The 13 transgenic parasites became visible in the culture approximately 4 weeks after treatment and were then 14 harvested. To examine whether the Cas9 expression cassette was incorporated in the kahrp locus, we 15 analysed their genotypes by PCR using the primer set p1 and p2 (Supplemental data 1). The results 16 showed specific amplification of a 7.2-kb DNA fragment, indicating the incorporation of the Cas9 17 expression cassette (Fig. 2B). Subsequently, to remove the psgRNA1_cen_kahrp and pCen_cas9 18 plasmids from the obtained transgenic parasites, we cultured them in the absence of drug for 6 weeks. 19 Following long-term cultivation, we cloned plasmid-free transgenic parasites, i.e., drug-selectable 20 marker-free parasites, by limiting the dilution procedure. We eventually obtained 4 parasite clones that 21 lost both the pCen9_cas9 and psgRNA1_cen_kahrp plasmids. We selected one of these plasmid-free 22 clonal parasites and named it pfcas9. To confirm whether the Cas9 expression cassette was integrated 1 only at the kahrp locus in pfcas9, we performed Southern hybridization analysis using the Cas9 gene 2 as probe DNA. The signal was detected solely at 4.8 kb in pfcas9, indicating that the Cas9 expression 3 cassette was integrated as a single copy at the kahrp locus in the genome (Fig. S1). Western blot 4 analysis using a FLAG antibody confirmed that the Cas9 was expressed without any degradation (Fig.  5 2C). The pfcas9 parasites could multiply in erythrocytes at growth rates comparable to those of the 6 parental strain 3D7 (Fig. 2D): the multiplication rates of pfcas9 and strain 3D7 were estimated to be 7 5.2 and 5.3 per cell cycle, respectively. Female and male gametocytes of pfcas9 were detected 8 microscopically; in addition, exflagellation of the male gamete was induced by xanthic acid 9 (Supplemental Mov. 1). These results showed that there was no obvious defect in asexual or sexual 10 development in pfcas9 due to the constitutive expression of Cas9. 11 To examine the effect of the constitutive expression of Cas9 on genome integrity, we conducted 12 whole-genome sequencing analysis of the pfcas9 parasite and examined the accumulation of mutations 13 caused by Cas9 during maintenance. The genomic DNA used for analysis was purified from pfcas9 14 that had been maintained over one month in culture and then sequenced to a depth of approximately 15 64.7× coverage, followed by comparison to the reference genome sequence of P. falciparum strain 16 3D7 deposited in PlasmoDB (https://plasmodb.org/plasmo/). A total of 165 SNPs and indels were 17 called (Supplemental data 2), and 127 of them were found in intergenic regions, subtelomeric regions 18 (Supplemental data 3), and introns. The SNPs and indels called in those regions may have been false 19 positives because mapping errors frequently occur in these regions due to their low sequence 20 complexity. Although 38 clear SNPs and indels were called in pfcas9, they might not have been caused 21 by the constitutive expression of Cas9. The parental parasite used for the generation of pfcas9 in this 22 study had been cultured for a long time, e.g., several months, which allowed for the accumulation of 1 mutations that did not participate in multiplication in RBCs. As shown later, these mutations are 2 commonly found in the transgenic parasite, supporting our speculation (Supplemental data 4). 3 Therefore, we concluded that the constitutive expression of Cas9 did not cause unexpected mutations 4 in the parasite genome. 5 6 Genetic modification using pfcas9 and a linear donor template. 7 Next, we attempted to engineer a gene by cointroducing the linear donor template and the plasmid 8 carrying the sgRNA into the fully mature schizonts of pfcas9 (Fig. 3A). As an initial attempt, we 9 introduced a single nucleotide insertion in the coding sequence of the transcription factor PfAP2-G, 10 which is involved in gametocytogenesis (Fig. 3B). The sgRNA designed in the middle of its AP2 11 domain was cloned in the psgRNA1_cen plasmid, and the resultant plasmid was named 12 psgRNA1_cen_ap2g. The linear donor template with single nucleotide insertion was generated by 13 PCR. In addition, to prevent re-cleavage by the Cas9-sgRNA complex after homologous 14 recombination, a shield mutation was introduced into the PAM sequence in the donor template DNA. 15 The fully mature schizonts of pfcas9 were purified using a Percoll-sorbitol gradient and then 16 cotransfected with 25 µg each of psgRNA1_cen_ap2g and the linear donor template DNA. The 17 transfected parasites were maintained after electroporation in the absence of drug for 3 days, followed 18 by pyrimethamine treatment for 10 days. The transgenic parasites were visible in the culture 2 days 19 after withdrawal of drug and then harvested. The target region was amplified from genomic DNA 20 purified from harvested parasites using primers p3 and p4 (Supplemental data 1) and sequenced. This 21 analysis confirmed that shield mutations were introduced with almost 100% efficiency: the wild-type 22 PAM sequence was not detected in this analysis (Fig. S2). However, some of the harvested parasites 1 did not have an additional A residue between nucleotide positions 6563-6564: we detected minor 2 chromatograms of the wild-type sequence downstream of nucleotide position 6563 (Fig. S2). These 3 results suggested that the majority of the transgenic parasites had both shield mutations and inserted 4 A residues, but there was a minor population that possessed only shield mutations. HDR with a linear 5 donor template occurred fully in the obtained transgenic parasites after cleavage of the target site by 6 the Cas9-sgRNA complex, but it might accidentally terminate in the minor parasite population before 7 reaching the site where a single nucleotide was inserted. Seven clonal parasites were established by a 8 limiting dilution procedure, and their mutations were then examined by sequencing analysis (Fig. 3C). 9 This analysis showed that all of them possessed the shield mutation, but one clonal parasite did not 10 have an A nucleotide residue, supporting the possibility described above. We estimated the efficiency 11 of this genetic manipulation to be 85% based on this result. The clonal parasites with disruption of 12 pfap2-g, named pfap2-g-ko, completely lacked gametocyte production capability (Fig. 3D). 13 Subsequently, we examined by whole-genome sequencing whether any off-target sites were 14 mutated in pfap2-g-ko. A total of 170 SNPs and indels were called except for the single nucleotide 15 insertion and the shield mutation in the pfap2-g gene by comparison to the genomic sequence of the 16 parental pfcas9 parasite (Supplemental Data 4). In total, 165 SNPs and indels were shared between 17 pfap2-g-ko and pfcas9, indicating that they were inherited from the parental pfcas9. This analysis 18 further called two indels unique in the exons of PF3D7_0505000 and PF3D7_0818700. Both indels 19 were found in repetitive sequences; in addition, no sequences around the indels were similar to the 20 sgRNA, which suggested that they were false positives due to low sequence complexity. Therefore, 21 we concluded that no off-target mutations were caused by genetic engineering using our CRISPR/Cas9 22

system. 1
In addition to single nucleotide insertion, we performed another type of genetic engineering: 2 fluorescent protein tagging ( Fig. 3E and S3A). We fused GFP with the transcription factor PfAP2-I, 3 which is essential for asexual multiplication. The sgRNA was designed at the region proximal to its 4 terminal codon and cloned in the psgRNA1_cen plasmid, resulting in the psgRNA1_cen_ap2-i 5 plasmid. The linear donor template encoding the gfp gene and the psgRNA1_cen_ap2-i plasmid were 6 cointroduced into pfcas9. The transgenic parasites emerged 2 days after drug treatment for 10 days. 7 To examine the fusion of pfap2-i with gfp, PCR analysis of the harvested parasites was performed 8 using the primer sets P5 and P6. The results showed the amplification of an approximately 2.0-kbp 9 fragment derived from the modified genomic locus, which confirmed GFP fusion (Fig. 3F). In contrast, 10 the estimated 1.0 kbp fragment from wild-type parasites was not amplified in the pooled parasite 11 population, suggesting that GFP was fused to PfAP2-I with almost 100% efficiency. We subsequently 12 cloned parasites by a limiting dilution procedure and then named them pfap2-i::gfp. Sequencing which confirmed its proper localization and expression profile (Fig. 3G). Collectively, these genes 20 could be modified by cotransfecting pfcas9 with linear donor template DNA and plasmids containing 21 sgRNA without unexpected recombination, showing that the technical problems of the current 22 CRISPR/Cas9 system in P. falciparum could be solved. 1 2 Double genetic engineering using the improved CRISPR/Cas9 system. 3 Our sequence analysis showed that the wild-type parasites were not present in the parasite 4 population emerging in culture after cotransfection with the linear donor template and plasmid DNA 5 containing the sgRNA. P. falciparum does not have the canonical nonhomologous end joining 6 (cNHEJ) pathway; if a double-strand break is not repaired by HDR using a donor template, the parasite 7 will die, probably due to instability of the cleaved chromosome, resulting in the observed elimination 8 of wild-type parasites. Hence, if multiple genomic sites are cleaved by Cas9 with sgRNA 9 corresponding to each target site, only transgenic parasites in which all sites are repaired by HDR may 10 survive, resulting in multiple genetic modifications. To validate this concept, we modified two genes 11 simultaneously by transfecting pfcas9 with two sgRNAs and two linear donor templates (Fig. 4A). To 12 this end, we generated the centromere plasmid psgRNA2_cen, which expressed two sgRNAs. Each 13 sgRNA including tracrRNA was transcribed by the promoters of U6 spliceosomal RNA of P. 14 falciparum and P. berghei (PbU6: PBANKA_1354380). In this attempt, we introduced expression 15 cassettes for two fluorescent proteins, GFP and mCherry, into two genomic loci on different 16 chromosomes; the GFP and mCherry expression cassettes were integrated into the pfcsp gene on 17 chromosome 3 and the pfpalm gene on chromosome 6, respectively. Moreover, the gfp and mcherry 18 genes were transcribed sex-specifically under the control of the promoters of the dynein heavy chain 19 (Male: PF3D7_1023100) and CCP2 (Female: PF3D7_1455800), respectively. The expression 20 cassettes of GFP and mCherry were flanked with sequences used for HDR by PCR, resulting in each 21 donor template DNA. The gRNAs specific for the pfcsp and pfpalm genes were designed and cloned 22 into the psgRNA2_cen plasmid, resulting in psgRNA2_cen_csp:palm. The psgRNA2_cen_csp:palm 1 plasmid and the two donor templates containing GFP and mCherry expression cassettes were 2 cointroduced into the pfcas9 parasites. Transgenic parasites were harvested after becoming visible in 3 the culture 14 days after transfection. PCR-based genotype analysis indicated that the GFP cassette 4 was integrated into the genomic locus of the pfcsp gene with almost 100% efficiency but the mCherry 5 cassette with lower efficiency; the fragments were amplified from not only the modified pfpalm locus 6 (2.8 kbp) but also the wild-type pfpalm locus (1.1 kbp) (Fig. 4B). We considered that this less efficient 7 mCherry fusion was probably due to less efficient cleavage of the Cas9 complex and the sgRNA for 8 the pfpalm locus. Subsequently, we obtained the transgenic parasite Pfg_red/green, in which both GFP 9 and mCherry protein expression cassettes were integrated into the corresponding locus. The 10 integration of both cassettes was confirmed in the Pfg_red/green parasites by PCR and sequence 11 analyses. In addition, fluorescence microscopic analysis showed that male and female gametocytes of 12 Pfg_red/green expressed GFP and mCherry proteins, respectively (Fig. 4C); in contrast, there was no 13 fluorescence in parasites at asexual stages, such as the ring form, trophozoite, and schizont stages. 14 These results demonstrated that multiple genetic modifications could be carried out simultaneously by 15 utilizing the CRISPR/Cas9 system developed in this study. 16

DISCUSSION 18
The technical limitations and problems of the current CRISPR/Cas9 system in P. falciparum include 19 the difficulty of introducing two plasmids containing Cas9, the sgRNA, and the donor template into 20 parasites, the requirement for two kinds of drugs for the selection of transgenic parasites and the 21 unexpected recombination of the plasmid DNA used to deliver the donor template into the parasite 22 genome. In the present study, we solved all of these issues by developing an efficient DNA transfer 1 technique using fully mature schizonts and by transfecting Cas9-expressing parasites with a linear 2 donor template. The desired transgenic parasites were generated approximately 2 weeks after 3 electroporation, and no unexpected recombination was found in the resultant parasites. 4 The linear form of DNA has to be used to avoid unexpected recombination; however, it is readily lost 5 from the parasites during nuclear division due to its low segregation, probably disappearing from the 6 parasite during the first cell cycle after electroporation. Thus, for genetic engineering using a linear 7 donor template, the HDR between the cleaved genome and the linear donor template must be 8 completed as quickly as possible after transfection. To this end, a linear donor template has to be 9 transferred with high efficiency, but current DNA transfer techniques are not sufficiently efficient. 10 Thus, we consider that a DNA transfer technique using fully mature schizonts is essential for the 11 CRISPR/Cas9 system using a linear donor template at present. In addition, the pre-expression of Cas9 12 allows efficient recombination, which prompts the integration of the linear donor template into the 13 genome, allowing the completion of HDR before the parasite loses the template DNA. The pre-14 expressed Cas9 can form a complex with sgRNA immediately after the introduction of the plasmid 15 carrying sgRNA, and this immediate cleavage prompts the subsequent HDR with the linear donor 16 template by efficiently recruiting the molecule responsible for recombination. Collectively, our 17 CRISPR/Cas9 system is based on three technical elements: the usage of a linear donor template, a 18 direct transfection technique using fully mature schizonts, and Cas9-expressing parasites. If any of 19 these elements are missing, high accuracy and efficiency cannot be achieved. 20 High efficiency of DNA transfer into the parasite can be achieved by using fully mature schizonts. 21 In contrast to fully mature schizonts, immature schizonts are sensitive to electroporation and thus 22 readily die from electric pulses, resulting in the failure of DNA introduction. The fully mature 1 schizonts contain invasive merozoites, which are released as a result of the disruption of two 2 membranes, belonging to parasitophorous vacuoles (PVMs) and RBCs (RBCMs). The merozoites are 3 wrapped with either PVM or RBCM, suggesting that both become fragile during schizont maturation, 4 including proteolytic digestion of membrane proteins 15 , and that one membrane is retained by chance. 5 This remaining membrane might be readily disrupted by electroporation, and transfected merozoites 6 would invade new RBCs immediately, resulting in high transfection efficiency. 7 The Cas9 nuclease-sgRNA complex binds to double-stranded DNA if there are three to five base 8 pair mismatches in the PAM-distal region of the sgRNA sequence. Thus, it can cleave other genomic 9 sequences, i.e., off-target sites, other than the desired target site. This cleavage at off-target sites is 10 repaired in eukaryotic cells by the cNHEJ pathway, causing a small deletion or insertion. On the other 11 hand, the Plasmodium genus, including P. falciparum, lacks the cNHEJ pathway. Thus, if off-target 12 sites are cleaved in Plasmodium parasites, they will not be repaired by the cNHEJ pathway. These off-13 target cleavages make the genome unstable, resulting in the death of parasites. As a result, parasites 14 with off-target cleavages may be eliminated from the transgenic parasite population. The whole-15 genome sequencing in this study suggested that there were no off-target mutations in the resultant 16 transgenic parasite clone. Furthermore, similar results were obtained in our previous study in the 17 rodent malaria parasite P. berghei. Therefore, we consider that genetic engineering can be performed 18 by the CRISPR/Cas9 system without off-target mutations in the Plasmodium genus. 19 Genetic modification at two different genomic loci was performed by our CRISPR/Cas9 system. In 20 the present study, we used this method to integrate two fluorescence protein expression cassettes into 21 different chromosomes. In addition, it can be used for various genetic modifications, such as double 22 gene targeting and tagging and gene targeting of two different genes. Furthermore, this method will 1 be useful for deleting or replacing kbp-scale genomic regions, which has been difficult to accomplish 2 by using one sgRNA. In general, after the target sites are cleaved by Cas9, the DNA sequence around 3 the 5' end on either strand is trimmed, generating 3' overhangs. These overhangs invade the 4 complementary donor template, initiating HDR. When one sgRNA is used for kbp-scale genetic 5 modification, there is a distance between the cleaved genomic locus and the regions used for HDR. 6 Due to this distance, it is difficult to generate overhangs possessing complementary sequences to the 7 regions used for HDR, resulting in failure. However, when two sgRNAs are used for similar genetic 8 modification, each cleaved genomic locus will be proximal to the regions used for HDR. The 3' 9 overhang sequences that are complementary to the region used for HDR will be readily generated in 10 this case, resulting in successful modification. Kilobase-scale genetic modifications can be utilized for 11 a wide range of experiments, such as generating complete null mutants by deletion of entire gene 12 regions, including the coding region, 5'-, and 3'-UTR; replacing promoter regions with a synthetic 13 DNA fragments; and deleting specific genomic loci with unique epigenetic marks. Thus, we anticipate 14 that genetic modification using two sgRNAs will be useful for generating transgenic parasites with 15 complex genetic modifications. 16 When GFP and mCherry expression cassettes were integrated into the pfcsp and pfpalm loci, 17 respectively, by our CRISPR/Cas9 system using the psgRNA2_cen plasmid, we found that some 18 transgenic parasites maintained the wild-type pfpalm sequence, including the site targeted by the 19 sgRNA. This suggested that cleavage by the Cas9-sgRNA complex was less efficient at the pfpalm 20 gene than at the pfcsp gene. Because the sgRNA for pfpalm was controlled under the PbU6 promoter, 21 its transcriptional activity in P. falciparum might be weaker than that stimulated by the PfU6 promoter. 22 This weaker transcriptional activity of the PbU6 promoter might cause less efficient cleavage of the 1 targeted sequence of the pfpalm gene, resulting in failure of integration of mCherry cassette. We 2 consider that the transcriptional activity of the promoter used for the sgRNA may be a determinant of 3 the efficiency of genetic engineering by the CRISPR/Cas9 system. Thus, an appropriate promoter 4 derived from P. falciparum should be used for the transcription of sgRNA. 5 In conclusion, our new CRISPR/Cas9 system overcame all technical problems in the current 6 system for P. falciparum. Furthermore, since our system dramatically elevates the efficiency with 7 which transgenic parasites were generated, it can not only accelerate studies in P. falciparum but also 8 enable us to perform complicated gene editing, such as editing two loci at once and achieving large-9 scale editing, which has never been accomplished with previous systems. If the same CRISPR/Cas9 10 system could be developed in strain NF54, which is widely used for parasite transmission experiments 11 in mosquito vectors, the functional analysis of genes would be expanded throughout the life cycle. 12 Therefore, our CRISPR/Cas9 system will open new avenues in molecular genetics and postgenomics 13 in P. falciparum and become the standard method for genetic modification of P. falciparum. 14 15

Parasites and culture 17
The pfcen_cas9 parasite, which contains the cas9 expression centromere plasmid pfCas9_cen, 18 was generated from P. falciparum strain 3D7 in our previous study 14 and used for the generation of 19 the pfcas9 parasite in the present study. The pfcas9 parasite will be deposited at the Malaria Research 20 and Reference Reagent Resource Center, MR4 (https://www.beiresources.org/About/MR4.aspx). All 21 parasites were cultured in vitro under low oxygen concentrations as described previously. 22 1 Transfection of fully mature schizonts 2 Parasites were roughly synchronized by treatment with 5% sorbitol prior to tight synchronization. 3 When most of the parasites had developed into schizonts, they were purified using a 40%-70% 4 discontinuous Percoll gradient solution (GE Healthcare Life Sciences) with 6% sorbitol. Purified 5 schizonts were cultured with fresh RBCs for four hours and then treated with 5% sorbitol. The 6 resulting parasites were synchronized within a window of approximately four hours. These Percoll 7 and sorbitol synchronizations were repeated three times, resulting in tightly synchronized parasites. 8 The emergence of fully mature schizonts was monitored via microscope for 88 hours after the final 9 synchronization, and the ratios of mature schizonts to total schizonts were determined every two hours. 10 When the ratio of fully mature schizonts to total schizonts reached a maximum number, the parasites 11 were purified again using a discontinuous Percoll gradient. Purified schizonts consisted of both where T is the total number of RBCs in culture (5 ml medium with Ht 2%); D is the number of days 9 after transfection; P is the percentage of parasitemia at day D; and I is the number of independently 10 transfected parasites. 11 12

Construction of sgRNA-expressing plasmid 13
The gRNA was designed as described previously. Briefly, a 19-bp sequence was designed upstream 14 of the protospacer-adjacent motif (PAM), and a pair of complementary oligonucleotides was 15 synthesized for each target site. Since the U6 promoter requires a guanosine nucleotide to initiate 16 transcription, a guanosine was added at the 5' end of the designed oligonucleotide that encoded the 17 sense sequence. In addition, the oligonucleotides were designed to generate overhangs used for 18 cloning into BsmBI-or BsaI-digested plasmids, as described below. Two synthesized complementary 19 oligonucleotides were annealed and cloned into plasmids. 20 A centromere plasmid for expressing sgRNA was generated from the pf-gRNA plasmid 11 . The pf-21 gRNA contains a sgRNA expression cassette in which transcription of sgRNA is controlled by the 22 PfU6 (U6 spliceosomal RNA, PF3D7_1341100) promoter. Two recognition sites of BsmBI are 1 introduced between the PfU6 promoter and tracrRNA and used for cloning the gRNA. This plasmid 2 also contains hdhfr, a drug-selectable marker gene, which is driven by the P. berghei elongation factor 3 1α (PBANKA_1133300, PBANKA_1133400) promoter. The centromere of P. falciparum 4 chromosome 5 was excised from the pfCas9_cen plasmid 14 by BamHI and NotI digestion and then 5 cloned into pf-gRNA, resulting in the psgRNA1_cen plasmid. The annealed gRNA oligonucleotides 6 were cloned into the BsmBI-digested psgRNA1_cen plasmid. 7 To generate a centromere plasmid expressing two sgRNAs targeting different genomic loci, 8 another sgRNA expression cassette was incorporated into the psgRNA1_cen plasmid. The sgRNA 9 expression cassette was amplified from the psgRNA2 plasmid previously reported by Shinzawa et al. 10 The cassette is composed of the PbU6 (U6 spliceosomal RNA, PBANKA_1354380) promoter and 11 tracrRNA scaffold, and two BsaI recognition sites are included between them to clone the gRNA. The 12 β-lactamase gene, which is a well-known selectable marker in E. coli, contains the BsaI site, which 13 was eliminated by introducing a synonymous mutation before cloning the sgRNA expression cassette. 14 The PbU6-driven sgRNA cassette was then integrated into the mutated psgRNA1-cen at the BamHI 15 site by In-Fusion cloning, and the resulting plasmid was named psgRNA2_cen. Two gRNAs were 16 cloned into BsmBI and BsaI, and the resultant plasmid expressing two sgRNAs was used for the 17 multiple genetic modification experiments. 18 19

Preparation of donor template DNA 20
The pfhsp70 (PF3D7_0818900) promoter and the cas9 gene with the 3'-UTR of pbhsp70 21 (PBANKA_0711900) were amplified from the genomic DNA of strain 3D7 and the pfCas9_cen 22 plasmid, respectively. These two DNA fragments were then fused by overlap PCR, digested with 1 BamHI and SalI, and cloned in tandem into BamHI and SalI-digested pBluescript SK(+) using a DNA 2 Ligation Kit (Takara). Two partial sequences were amplified from the kahrp locus and cloned into the 3 plasmid containing the cas9 expression cassette. These sequences flanked the cas9 expression cassette 4 on both sides. The resultant plasmid with the Cas9 expression cassette and two partial sequences of 5 kahrp was linearized by digestion with KpnI and NotI restriction enzymes and used as donor template 6 DNA to generate the pfcas9 parasite. 7 A donor DNA template for pfap2-g (PF3D7_1222600) gene knockout was produced by overlap PCR. 8 The donor template DNA contained the following two mutations: an adenosine insertion at position 9 6563 of pfap2-g and a single nucleotide substitution at the PAM sequence. For fusion of gfp to pfap2-10 i (PF3D7_1007700), donor template DNA containing the gfp gene flanking two homologous regions 11 of pfap2-i was produced by overlap PCR. Six nucleotides encoding Ala and Ser residues were 12 introduced between pfap2-i and gfp as a linker sequence. The male-and female-specific reporter 13 cassettes were generated using GFP and mCherry, respectively. The pfdynein (dynein heavy chain, 14 PF3D7_1023100) and pfccp2 (LCCL domain-containing protein, PF3D7_1455800) promoters were 15 used as male-and female-specific promoters, respectively. The pfcsp (circumsporozoite protein, 16 PF3D7_0304600) locus was used as the target site for integration of the male-specific reporter cassette 17 with GFP. The pfpalm (liver merozoite formation protein, PF3D7_0602300) locus was used as the 18 target site for the female-specific reporter cassette with mCherry. The transcription of gfp and mCherry 19 was terminated by the 3'-UTRs of pfhsp90 (PF3D7_0708400) and Pfhsp70, respectively. DNA 20 fragments of the male-and female-specific reporter cassettes were generated by overlap PCR. The 21 male-specific reporter cassette was then cloned into the EcoRV recognition sites in pBluescript SK(+). 22 Two partial sequences of pfcsp were amplified from genomic DNA of strain 3D7 and cloned on each 1 side of the male-specific reporter cassette in the plasmid using In-Fusion cloning kit. The female-2 specific reporter cassette was cloned into pBluescript SK(+) digested with XhoI and HindIII, and two 3 partial sequences of pfpalm were also amplified and cloned at each side of the female-specific reporter 4 cassette in a similar manner as the male-specific cassette. The male-and female-specific reporter 5 cassettes flanking those sequences used for HDR were amplified from the resultant plasmids by PCR, 6 and the resultant linear DNA fragments were used for the transfection experiment. 7 8

Southern blot analysis 9
Genomic DNA used was purified from pfcas9 parasites by the standard phenol/chloroform method 10 (Iwanaga et al., 2012). Briefly, parasite pellets were dissolved in HMW buffer, which was 10 mM 11 Tris-HCl, pH 8.0, 150 mM NaCl, 10 mM EDTA, and 0.1% SDS, and then treated with 40 µg/ml RNase 12 (Takara) for 30 min, followed by treatment with 200 µg/ml Proteinase K (Wako) for 90 min. Genomic 13 DNA was extracted once with phenol, followed by extraction with phenol-chloroform-isoamyl alcohol. 14 After precipitation with ethanol, the DNA was dissolved in TE buffer. Genomic DNA was digested 15 with EcoRI and EcoRV for 8 hours. The digested DNA was separated on 1% agarose gels and blotted 16 onto nitrocellulose membranes (Amersham Hybond-N+, Merck). Probe DNA labelling and detection 17 were carried out using DIG High Prime DNA Labeling and Detection Starter Kit II according to the 18 manufacturer's instructions. The hybridized signals were detected using ChemiDoc MP (Bio-Rad). 19 All other Southern hybridization analyses were performed in a similar manner as described above. 20 21

Western blotting analysis 22
Infected red blood cells were lysed with red blood cell lysis buffer (150 mM NH4Cl, 10 mM 1 NaHCO3, and 1 mM EDTA). After red blood cell lysis, the parasites were recovered by centrifugation 2 and dissolved in 1x SDS-loading buffer containing 5% 2-mercaptoethanol, followed by boiling for 5 3 min. Western blotting was performed as described previously 31  The parasitemia of parasites was adjusted to 0.1% and cultured in complete medium as described 16 previously. The progression of arasitemia was examined every 48 hours using Giemsa-stained thin 17 smears. The average parasitemia between pfcas9 and strain 3D7 was evaluated using a t-test. The 18 growth rate was calculated based on the approximate growth curve. The curve was represented by the 19 following equation; 20

P = Ae xD 21
Where P is the parasitemia; A is the constant value; D is the day of the postinfection; and e x is the 22 For parasite growth during asexual development, the values are presented as the mean ± SEM 1 from at least three biological replicates and were statistically compared using unpaired Student's t-test. Whole-genome sequencing data were deposited in the DDBJ database under accession numbers 7 DRA011698. All relevant data are available from the authors upon request. was comparable to that of wild-type parasites (black, box). Positive and negative errors were 24 calculated from the standard error of the mean from biological triplicates. Distributions for each day 1 were compared using the unpaired t-test (not significant). Two linear donor templates, which contained male-and female-specific reporter cassettes, and the 16 psgRNA2_cen plasmid containing two sgRNAs were cointroduced into pfcas9 parasites. A male-17 specific reporter cassette with the gfp gene was integrated at the pfcsp locus. A female-specific cassette 18 with the mCherry gene was performed at the pfpalm locus. If the cleaved genomic loci at pfcsp and 19 pfpalm were repaired with donor templates by HDR, the parasite would survive. However, if one of 20 them was not repaired, parasites would die due to instability of the cleaved chromosome. (B) 21 Genotyping PCR was performed using genomic DNA purified from Pfg_red/green before and after 22