Generation and functional characterization of Plasmodium yoelii csp deletion mutants using a microhomology-based CRISPR/Cas9 method

Abstract

CRISPR/Cas9 is a powerful genome editing method that has greatly facilitated functional studies in many eukaryotic organisms, including malaria parasites. Due to the lack of genes encoding enzymes necessary for non-homologous end joining (NHEJ) DNA repair pathway, genetic manipulation of malaria parasite genomes is generally accomplished through homologous recombination (HR) requiring the presence of DNA templates. Recently, an alternative double-strand break (DSB) repair pathway, microhomology-mediated end joining (MMEJ), was found in the Plasmodium falciparum parasite. Taking the advantage of the MMEJ pathway, we developed a MMEJ-based CRISPR/Cas9 (mCRISPR) strategy to efficiently generate multiple mutant parasites simultaneously in genes with repetitive sequences. As proof of principle, we successfully produced various size mutants in the central repeat region (CRR) of the Plasmodium yoelii circumsporozoite surface protein (CSP) without the use of template DNA. Monitoring mixed parasite populations and individual parasites with different sizes of CSP-CRR showed that the CSP-CRR play role in the development of mosquito stages, with severe developmental defects for parasites with large deletions in the repeat region. However, the majority of the csp mutant parasite clones grew similarly to the wild type P. yoelii 17XL parasite in mice. This study develops a useful technique to efficiently generate mutant parasites with deletion or insertion and shows that the CSP-CRR plays a role in parasite development in mosquito.

1. Introduction

Malaria parasites are organisms that can survive immune and drug killing by changing their genome sequences or by altering the expression of target molecules. The major mechanisms of genetic changes include nucleotide substitution, deletion/insertion, genetic recombination, or epigenetic modifications. Double-strand breaks (DSBs) of DNA can result in changes in the genome and lead to chromosome translocations, chromosome loss, cell cycle stalling, and cell death (Moynahan and Jasin, 2010). Repair of the DSBs plays important roles in parasite biology and survival. Whereas faithful repair of damaged essential genes is necessary for parasite survival; polymorphisms due to DNA repair error at some specific antigen genes may help the parasite overcome host immune and drug pressures. Three major DSB repair mechanisms have been described in eukaryotic organisms, including non-homologous end joining (NHEJ), homologous recombination (HR), and microhomology-mediated end joining (MMEJ) (Lieber, 2008; McVey and Lee, 2008; Moynahan and Jasin, 2010; Chang et al., 2017). Because no genes encoding enzymes for the NHEJ pathway can be found in the genomes of malaria parasites, the major DSB repair pathway in Plasmodium parasites is HR that has been experimentally verified by numerous HR based genetic modification (Lee et al., 2014). An alternative end joining pathway was also described in the human parasite Plasmodium falciparum (Kirkman et al., 2014), which is likely to be mediated through MMEJ (Singer et al., 2015). The MMEJ pathway may be responsible for the high rates of small indel mutation in AT-rich malaria genome sequences (Hamilton et al., 2017) and elevated recombination events associated with G-quadruplex motifs (Stanton et al., 2016).

CRISPR/Cas9 is a gene editing technique that has led to great advances in basic research, biotechnology, and therapy (Komor et al., 2017). The CRISPR/Cas9 system generally employs a prokaryotic RNA programmable nuclease that can introduce DSB at a target site identified by sequence-specific guide RNA (sgRNA) (Jinek et al., 2012). This powerful technique has been successfully adapted to edit malaria parasite genomes (Ghorbal et al., 2014; Wagner et al., 2014; Zhang et al., 2014; Lu et al., 2016; Mogollon et al., 2016; Crawford et al., 2017; Zhang et al., 2017). Unfortunately, because malaria parasites lack the critical components of NHEJ DNA repair pathway, such as Ku70/80 and DNA Ligase IV9 (Gardner et al., 2002), HR is considered to be the primary DSB repair pathway in this parasite (Straimer et al., 2012). Therefore, CRISPR/Cas9 methods for malaria parasites also require homologous DNA templates flanking the targeted DSB site that are typically provided in a plasmid. The requirement of template DNAs can greatly increase the workload in plasmid construction, particularly when editing a large number of genes or members of a gene family.

The genomes of malaria parasites contain large numbers of simple sequence repeats such as microsatellites in non-coding and minisatellites in coding regions (Su et al., 1999; Gardner et al., 2002). One example is the gene encoding sporozoite surface protein (CSP). The CSP protein contains two small conserved regions (Regions I and II) and a central repeat region (CRR) containing immunodominant epitopes (Lal et al., 1987). Antibodies specific for the CSP-CRR are important for the protective immunity (Sinnis and Nardin, 2002). Indeed, a subunit malaria vaccine candidate composed of the CSP repeats (RTS, S) has been shown to be partially protective, validating the CSP repeats as a vaccine target (Rts et al., 2012; Triller et al., 2017). A total deletion of the CRR of Plasmodium berghei CSP leads to defective sporozoites, but appeared to have normal oocysts (Menard et al., 1997; Ferguson et al., 2014). Replacement of P. berghei CSP gene with that of the avian Plasmodium gallinaceum parasite also results in non-infective sporozoites (Aldrich et al., 2012). Additionally, CSP is reported to interfere with the import of host NFκB into the nucleus and to influence the expression of over 1,000 host genes (Singh et al., 2007). The highly conserved region I is cleaved by a papain family cysteine protease, which is required for sporozoite invasion into host cells (Coppi et al., 2005). Although CSP is primarily expressed in sporozoites and is essential for sporozoite development (Menard et al., 1997; Ferguson et al., 2014), the effects of CRR size variation on sporozoite development are not clear. This study has two major goals: One is to test the principle of Plasmodium MMEJ pathway in repairing CRISPR/Cas9 mediated DSBs and to develop a simple MMEJ mediated CRISPR/Cas9 method (mCRISPR) for parasite genome editing. Another goal is to use the method to generate CSP-CRR deletion mutants for studying CSP-CRR functions in parasite development and potentially for vaccine development. We successfully developed and verified the functionality of a mCRISPR method that does not require homologous DNA templates. We then used the mCRISPR to generate P. yoelii CSP-CRR mutants, cloned ten parasites with different CRRs, and evaluated their development in mice and mosquitoes. The majority of the PyCSP-CRR mutant clones grew similarly to the wild type (WT) 17XL parasite in mice. Parasites with large deletion (172 AA) in the CRR or deletion with reading frame shift cannot develop into infective sporozoites. The results show that a minimum length of CRR is required for sporozoite development, providing new insights for the role of CSP-CRR in the development of mosquito stages.

2. Materials and Methods

2.1. Malaria parasite, mice and mosquitoes

P. yoelii lethal line 17XL derived from the nonlethal isolate 17X was used in this study (Pattaradilokrat et al., 2008). Female outbred ICR mice and inbred BALB/c mice, 6–8 weeks old, used to maintain and evaluate parasite growth, respectively, were purchased from Xiamen University Laboratory Animal Center or Shanghai Laboratory Animal Center, CAS (SLACCAS). A colony of Anopheles stephensi mosquitoes (Hor strain) was raised at 23°C and 75% humidity under a 12:12 light-dark illumination cycle and fed with 5% sucrose solution. All animal experiments were performed in accordance with an approved protocol (#XMULAC20150080) by the Laboratory Animal Management and Ethics Committee of Xiamen University.

2.2. Plasmid vector constructs and parasite transfection

Four sgRNAs were designed to target multiple sites in the CRR of the Pycsp gene (accession no. PY17X_0405400, 1284 bp) (Supplementary Fig. S1 and Table S1). Cas9-sgRNA plasmids without homologous donor templates were generated based on the pYC vector that contains the gene encoding Cas9 enzyme from Streptococcus pyogenes (Zhang et al., 2017). Plasmid without sgRNA served as a control was generated as well. Transfection and selection of transformed parasites with pyrimethamine were performed as described previously (Nair et al., 2017; Zhang et al., 2017). Briefly, 0.5–1 × 10⁷ purified schizonts were mixed with 100 μl cytomix (120 mM KCl, 0.15 mM CaCl₂, 5 mM MgCl₂, 25 mM Hepes, 2 mM EGTA, 10 mM KH₂PO₄, pH 7.6) containing 5–10 μg plasmid DNA in an electroporation cuvette and transfected using program T-016 in the Nucleofector device (Amaxa). Transfected parasites were immediately injected into a tail vein of a naïve mouse and subsequent selection with pyrimethamine provided in drinking water at a concentration of 7 mg/L from day 2 post-injection. Drug resistant and transformed parasites usually appear 5 to 7 days post-injection.

Gene deletion and/or insertion events were first detected by PCR amplification using Taq DNA polymerase (#AP111, TransGen Biotech) with primer pairs F1/R1 or F1/R2 flanking the Pycsp repeat region (Supplementary Table S1). PCR cycles and set up were: initial denaturation at 94°C for 3 min; cycling 94°C for 20s, 45°C for 20s, 60°C for 70s for 30–35 cycles; and a final extension at 60°C for 3 min. PCR products were separated on 1.5% agarose gels, visualized and photographed on a UV box (Tanon-2500). PCR products, amplified from parasite mixtures (without cloning) or cloned parasites, were sequenced directly by a commercial company (Xiamen Borui Biological Technology Co., Ltd).

2.3. Cloning PCR products in TA vector and DNA sequencing

To detect deletions or insertions in the CRR among parasite populations after CRISPR/Cas9 editing, DNA from blood stage parasite mixtures transfected with mCRISPR plasmids were amplified, and the PCR products were cloned into TA vector. The TA plasmids were used to transform Escherichia coli (E. coli). Individual bacterial colonies were picked and grown in LB-broth overnight. Plasmid DNAs were extracted from the bacterial cultures for DNA sequencing.

2.4. Parasite cloning and identification

Limiting-dilution cloning of parasite was followed if mutations in Pycsp gene were detected in a transfected parasite population after PCR amplification and/or DNA sequencing. Infected red blood cells (iRBCs) from donor mouse blood were diluted to 1 parasite per 200 μl PBS (inoculum size) and were i.v. injected into 20–30 recipient mice. Mice were monitored for the presence of parasites on day 5–9 after the injection. Positive blood samples with a parasitemia of 1–5% were collected for DNA extraction using phenol/chloroform method and for parasite preservation in −80°C refrigerator or liquid nitrogen. Individual clones with CRISPR/Cas9 mediated mutations were identified by PCR assay and confirmed by DNA sequencing.

2.5. Characterization of asexual blood and mosquito stages

Daily parasitemia of asexual blood stages of WT 17XL and PyCSP-CRR mutant parasites, and host mortality rates were evaluated in female BALB/c mice (4–5 mice per single parasite clone). Each mouse was injected i.v. with an inoculum containing approximately 1 × 10⁶ iRBCs. Giemsa-stained thin tail blood films were made daily from day 2 post-infection (pi). Parasitemias were monitored and measured by microscopic examination and recorded in Excel.

For blood feeding of mosquitoes, 200 μl PBS containing 1 × 10⁶ iRBCs from a donor mouse was injected i.v. into a recipient ICR mouse. Parasitemia and gametocytemia were monitored by thin blood smears and Giemsa stain. The infected mouse was fed to 150–200 female mosquitoes for 30 min day 3–4 pi. Mosquito midguts were dissected on day 8 after blood meal and examined for oocysts under a light microscope. Salivary glands were dissected on day 17–19 and harvested for homogenization; released sporozoites were counted using a hemocytometer. To measure the infectivity of salivary gland sporozoites, naïve mice were infected through mosquito bites.

2.6. Statistic tests

Pair comparisons were tested using two tailed t-test in Excel or Prism. P-value ≤0.05 is considered significant.

3. Results

3.1. Generation of PyCSP-CRR mutants using mCRISPR method

To study the effects of PyCSP-CRR size polymorphism on parasite development, we developed a simple and efficient CRISPR/Cas9 method taking the advantage of the MMEJ DNA repair mechanism observed in malaria parasites recently (Kirkman et al., 2014) (Singer et al., 2015). The CSP protein of 17XL parasite contains two types of tandem repeats [(QGPGAP)₂₅ and (PPQQ)₇] in the CRR. We designed four guide RNAs (sgRNA, Supplementary Table S1 and Fig. S1) from the CRR region and constructed four plasmids that transcribes each of the sgRNA and the Cas9 enzyme without homologous DNA templates to generate targeted DSBs (Fig. 1A and 1B). The sgRNA-2 is expected to cleave three sites at the CRR because it has three corresponding target sequences, whereas the other three sgRNAs have one cleavage site (Supplementary Fig. S1A). Without the presence of homologous DNA templates, overlapping repetitive sequences may anneal randomly, leading to cleavages of overhanging sequences and sealing of the gaps (Supplementary Fig. S1B) (McVey and Lee, 2008). Target DNA samples were extracted from the 17XL parasite in mouse bloods when the parasitemia reached 1–5% after transfection and pyrimethamine selection and were amplified using primer pairs F1/R1 (or R2) flanking the PyCSP-CRR (Fig. 1B). Multiple bands could be detected in parasites transfected with sgRNA-2 and sgRNA-3, whereas a major band similar to that of 17XL WT was observed in the parasites transfected with sgRNA-1 and sgRNA-4 (Fig. 1C). We cloned the amplified DNAs from transfectants with four individual sgRNAs into TA-vector, transformed bacteria, randomly selected 11 colonies for DNA sequencing, and obtained seven unique sequences with deletion of 234 to 881 bp and two 17XL WT sequences (Table 1). We also amplified the PyCSP-CRR from WT 17XL parasite, cloned the PCR product into TA vector, and sequenced DNA samples from six randomly picked bacterial colonies. No change in the CRR sequence was observed (data not shown), suggesting that PyCSP-CRR is relatively stable in E. coli. Interestingly, three of the sequences had deletions leading to reading frame shift, suggesting that a complete functional CSP protein is not required for blood stage survival (also see cloned mutant parasites below). These results demonstrate that the P. yoelii parasite can repair DSBs generated by CRISPR/Cas9 through the MMEJ pathway without the need of homologous DNA templates, although we cannot rule out that some of the cloned mutant sequences were due to deletion during cloning in E. coli.

Figure 1.

Demonstration of a microhomology-based CRISPR/Cas9 strategy to generate size mutants. (A) A simplified diagram of transfection plasmid for generation of double strand break. (B) Csp gene structure showing the central repeat region, the conserved region I and II, and the location of primers used in PCR. (C) An agarose gel showing bands of PCR products using primer F1/R2 from day 3 parasite mixtures after mCRISPR modification and drug selection. The sgRNAs used and no sgRNA control are as indicated.

Table 1.

CSP deletion clones obtained after mCRISPR editing and TA-cloning.

Clone	sgRNA	# of bp	# bp del.	# of AA	# AA del.
17XL		1284	/	427	/
C1–4	sg1	403	881	33*	394
C1–21	sg1	1284	0	427	0
C2–7	sg2	888	396	295	132
C2–8	sg2	1284	0	427	0
C2–9	sg2	1050	234	349	78
C3–31	sg3	771	513	256	171
C4–36	sg4	1032	252	343	84
C4–37	sg4	404	880	74*	353
C4–39	sg4	530	754	98*	329

Note: Clone, bacterial colony after TA-cloning;

sgRNA, single guide RNA used in CRISPR editing;

# of bp, numbers of nucleotide in the amplified segments;

# bp del., numbers of base pairs deleted or inserted, compared with that of wild type;

# of AA, number of amino acids in the amplified segments;

# AA del., numbers of amino acids deleted or inserted.

^*,mutation in open reading frame causing premature termination.

3.2. Fitness of parasites with different CSP-CRR in mice and mosquitoes

To evaluate effects of CSP-CRR size polymorphism on parasite fitness, we transfected the 17XL parasite with plasmids containing genes encoding the Cas9 enzyme and sgRNA-2 or sgRNA-3, respectively. After injection of the transfected parasites into three mice followed with drug selection (7 mg/L pyrimethamine treatment starting from day 2 pi), we extracted DNA from parasite mixtures day-1 to day-7 post injection. We then PCR-amplified the repetitive region using primers F1/R2 and detected DNA bands on agarose gels. At least six major bands could be detected in the day-1 DNA sample from parasite mixture transfected with sgRNA-2, with the majority of new major bands being smaller than that of WT band (Fig. 2A). Interestingly, most of the bands became weaker from day-2; and by day-4, a band of ~900 bp (smaller than the WT band) became the dominant band. The WT band disappeared likely due to continuing cleavage of the WT DNA by Cas9 because it contained the sgRNA sequence, whereas the 900 bp band may not have the sgRNA sequence thereby avoiding further cleavages. Direct sequencing of the 900 bp band showed a mixture of multiple sequences. Cloning and sequencing of individual DNA from the 900 bp product confirmed the absence of the sgRNA sequences in the sequences (data not shown), which partially explains the persistence of parasites with the 900 bp DNA. We also performed experiments using sgRNA-2 or sgRNA-3, treated the transfected parasites with pyrimethamine for 7 days starting from day 2 pi, passed the parasites to naïve mice without further drug treatment, and obtained similar results (Fig. 2B and 2C). Without drug pressure, traces of WT band could be detected after day 2, although the ~900 bp products were still the dominant bands. In the sgRNA-3 group, more than six bands were detected day 1 post injection, however, three major bands were retained from day-2 to day-5 (Fig. 2C; mice died earlier without pyrimethamine treatment). These results suggest that parasites with large deletions in the CSP-CRR may not grow as well as the parasite with ~900 bp size or WT CSP in mouse blood, although we cannot rule out that negative ‘off-target’ effects may also contribute to the disappearance those smaller bands. The results suggest that it is necessary to clone parasites as early as possible in order to obtain mutants with different deletion sizes.

Figure 2.

Selection of Pycsp central repeat mutant alleles during development in the blood and in mosquito. (A) Various Pycsp central repeat alleles amplified from day 1–7 parasite mixtures after transfection and drug selection. (B-C) Various Pycsp central repeat alleles amplified from blood stages without drug selection after passing the parasites to naïve mice. B, transfected with sgRNA-2; C, transfected with sgRNA-3. (D) Pycsp central repeat alleles amplified from blood stages (BS), day 8 oocysts (D8 ooc), day 10 oocysts (D10 ooc), day 12 oocysts (D12 ooc), day 17 sporozoites (D17 spz) or blood stages after infection with sporozoites (BS-2). (E) A similar experiment as in (D) with sporozoites that had a Pycsp central repeat of ~900 bp and failed to infect mice. Amplification products in (A-E) were obtained using primer pair F1/R2.

We next fed parasite mixtures to mosquitoes and monitored parasite population dynamics while developing in mosquitoes. We injected a CRISPR/Cas9 (CSP-sg2) transfected parasite mixture (1×10⁶ iRBCs) from frozen stock into mice. On day 4 pi when gametocytes could be detected in blood smear, we fed the parasite mixture to mosquitoes and collected DNA samples from day 8, 10, 12 oocysts, day 17 sporozoites, and blood stages after infecting mice with the sporozoites through mosquito bites. Interestingly, the WT band and the original dominant blood stage bands of ~900 bp became the major bands in oocysts (Fig. 2D). After infecting mice, the WT band was the only major band, suggesting that parasite with the WT CSP-CRR can develop better in mosquito and/or in the liver than the mutant parasites. In a separate experiment, the WT band did not become the dominant band in the mosquito stages likely due to lack of WT parasite; instead, the ~900 bp bands remained to be the dominant bands (Fig. 2E). The results suggest that some parasites with mutant CSP-CRR may not develop well, or at all, in mosquito and/or in the liver.

3.3. Cloning parasites with different csp repeat lengths

To further characterize the functional roles of PyCSP-CRR in parasite development, we attempted to clone parasites with different lengths of CRR. We performed limiting dilution cloning by injecting ~1 (in 100 μl PBS) parasite transfected with the sgRNA-2 or sgRNA-3 CRISPR/Cas9 plasmid into 120 mice in five independent experiments (Supplementary Table S2). Amplification of DNA samples from these parasites day 7 pi showed that the majority of the parasite clones carried the WT CSP-CRR, except one clone with smaller repeat region (clone sg2–4-1, Fig. 3A). There were also several clones (sg2–2-1, sg2–2-2, sg2–3-5, sg2–4-3, and sg2–7-3) that might still have mixed parasites with a dominant WT parasite population and a minor population with smaller CRR (Fig. 3A). Because these parasites were cloned from a frozen stock with a parasitemia of >1%, the parasite with WT repeat might have outgrown the other parasites with mutant CRRs during rounds of growth in mice. To improve the efficiency in obtaining CSP-CRR mutant parasites, we transfected mice with the mCRISPR plasmids again and cloned parasites directly as soon as parasites could be detected in blood smear (<0.1%). We obtained more parasite clones with different CSP repeat lengths on agarose gel (Fig. 3B). After DNA sequencing of PCR products from 25 cloned parasites of different transfections, we obtained nine independent parasite clones with different sizes of the PyCSP-CRRs, carrying deletions from 46 amino acids (AAs) to 172 AAs (Fig. 3C, Supplementary Fig. S2 and Fig. S3). One of the clones (sg3–4-4) had changes leading to disruption of open reading frame (ORF) and a truncated protein. These cloning results confirm that parasites with different CSP-CRRs can still grow in mouse blood. In the experiments without cloning, the parasite with WT CSP could be affected by the presence of CRISPR/Cas9 activity, and therefore, may not become the dominant population.

Figure 3.

Cloning and verification of parasites with different sizes of Pycsp central repeat. (A) PCR products of parasites cloned from a mouse blood injected with mCRISPR modified parasite mixtures. Blood from the mouse receiving the transfected parasites was first frozen, and a stock of 100 μl was thawed and injected into a second mouse. Day-3 blood was collected for limiting-dilution cloning. On top of the gel are parasite names. For example, sg2–1-1 indicates clone 1–1 from the mixture transfected with sgRNA-2. (B) Similar parasite cloning and PCR amplification as in (A), except that the parasites were cloned the at low parasitemia (<1%) directly from the mouse receiving electroporated parasites. (C) Summary of parasite clones with different numbers of amino acid deleted in the Pycsp central repeat region (CRR). Note: primer pair F1/R1 was used to amplify the products in (A) and (B).

3.4. Effects of different CSP-CRRs on blood stage growth

We next infected mice with ten parasite clones (clone sg2–2-12, sg2–2-21, sg2–2-22, sg2–2-24, sg2–4-1, sg2–4-2, sg3–2-63, sg3–4-4, sg3–5-3, sg3–7-1, and WT 17XL) carrying nine different PyCSP-CRRs (sg2–2-24 and sg2–4-1both have 126 AA deletion; Fig. 3) to evaluate effects of individual CRR mutant on parasite development. Eight of the ten mutant parasites grew similarly to the WT parasite, except clone sg2–2-22 (114 AA deletion) and sg3–7-1 (46 AA deletion) that had significantly lower parasitemia than those of the WT parasite on day 3–5 pi (Fig. 4A and 4B). Mice infected with these two parasites also survived longer than those infected with WT and other mutant parasites (Fig. 4C and 4D; euthanized on day 20). To evaluate the target site specificity, we searched the 17X genome for sequences homologous to the Pycsp sgRNA sequences we used. We found one and three potential off-target sites for sgRNA-2 and sgRNA-3, respectively (Supplementary Table S4). All of the four off-target sequences had ≥ 3 nucleotide mismatches. DNA sequencing of PCR amplified products of the potential off-target sites detected no mutations at these sites (Supplementary Table S4). However, we cannot rule out that the changes in growth for these two parasites were caused by other ‘off-target’ effects or unknown mutations in the genome because the parasitemia of the parasites with larger deletions or even disruption of open reading frame (sg3–4-4) were not affected by the deletions. These results appear to contradict the observations from those of un-cloned parasite mixtures, which showed that a parasite with WT or a PyCSP-CRR slightly smaller than that of WT parasite grew better than parasites with other mutant CRRs. One possible explanation is that the growth advantage of the parasites can only be observed in a mixture directly competing for survival in the same mouse. However, the mechanism for the disappearance of some small PyCSP-CRR bands in transfection mixture over time requires further investigations.

Figure 4.

Parasitemia and host survival rates of wild type (WT) 17XL and Pycsp repeat mutants. Individual parasite clones (1×10⁶) were injected i.v., and blood smears were prepared and counted as described in Methods. (A-B) Parasitemia of BALB/c mice infected with WT 17XL and six sgRNA-2 Pycsp central repeat mutant parasites (A) or four sgRNA-3 Pycsp central repeat mutant parasites (B). (C-D) Survival curves of 17XL and the Pycsp central repeat mutant parasites. Means and standard errors were calculated from four or five mice for each parasite clone; unpaired two-tailed t-test, **P < 0.01; ***P < 0.001.

3.5. Development of PyCSP-CRR mutants in mosquitoes

We also infected mosquitoes with the 10 individual CSP CRR mutants, counted day-8 oocysts and day-17 salivary gland sporozoites, and infected mice again through mosquito bites. Large variations in oocyst counts were observed between individual feedings, depending on each batch of mosquitoes. No difference in day 8 oocyst counts was observed when data from different feedings were combined (Fig. 5A). Significant reduction in sporozoite infectivity was found for sg2–2-12 (120 AA deletion) and sg2–4-2 (138 AA deletion) parasites after infection of mice through mosquito bites (Fig. 5B). Additionally, the clone with 172 AA deletion (sg3–2-63) and the one with reading frame shift (sg3–4-4) were not able to produce mature sporozoites to infect mice (Fig. 5B and Supplementary Table S3). These results show that a minimum length of CRR-CRR is required for the development of infective sporozoites.

Figure 5.

Oocyst and sporozoite counts from mosquitoes infected with 10 Pycsp central repeat mutants. (A) Day 8 oocyst counts for wild type 17XL and the Pycsp central repeat mutants. The middle level bars are mean oocyst counts, and the lower and upper bars are standard errors, respectively. (B) Day 17 salivary gland sporozoite counts for wild type 17XL and the Pycsp central repeat mutants. The numbers above each bar are the numbers of infected mice/numbers of challenged mice in each group. Means and standard errors were calculated from three to eight mice for each parasite clone; unpaired two-tailed t-test comparing sporozoite infectivity (percentage of mice infected) from each individual clone and 17XL, **P < 0.01; ***P < 0.001.

We next fed A. stephensi mosquitoes with the parasite mixtures having different csp repeat lengths to evaluate the impact of csp repeat variations on parasite growth in mosquitoes (Fig. 6). For the first mixture, all the three parasites could develop to mature sporozoites, but sg3–5-3 band became dominant in two of the three mice after mosquito bites (Fig. 6A). Similarly, sg3–5-3 became dominant and sg2–2-21 could not be detected in all three mice for the second mixture (Fig. 6B), suggesting that a 132 AA deletion may affect parasite migration to or development in the liver.

Figure 6.

Growth dynamics of Pycsp central repeat mutant clones in mosquito over time. 17XL, sg3–5-3 and sg2–4-1 (or sg2–2-21) parasites were counted, mixed at approximately at 1:1:1 ratio, injected into a mouse before feeding to mosquitoes. DNA from parasites before injection (D0), day 8 oocyst (D8), day 19 salivary gland sporozoites (D19) and blood of infected mice after mosquito bites (B2) were extracted and amplified for 27 cycles (for B2, 29 cycles to increase signals). A, Amplified PCR products from 17XL/sg3–5-3/sg2–4-1 parasite mixture. B, Similar experiments as in (A) from 17XL/sg3–5-3/sg2–2-21 parasite mixture. M1-M3 indicates mosquitoes fed on three individual mice. Each DNA sample was extracted from 20–30 dissected salivary glands.

4. Discussion

This study develops a MMEJ based CRISPR/Cas9 gene editing strategy to generate size polymorphisms in PyCSP-CRR. Various CRISPR/Cas9 methods have been developed for editing malaria parasite genomes; however, they all require the presence of homologous DNA templates (Ghorbal et al., 2014; Wagner et al., 2014; Zhang et al., 2014; Lu et al., 2016; Mogollon et al., 2016; Crawford et al., 2017; Zhang et al., 2017) because the absence of the NHEJ machinery in the parasite genomes (Gardner et al., 2002). The discovery of the alternative MMEJ mechanism in P. falciparum provides an opportunity for developing an efficient method to generate a large number of size mutants in genes with repetitive sequences (Kirkman et al., 2014; Singer et al., 2015). MMEJ based CRISPR methods have been developed and tested in various organisms such as pathogenic fungus Aspergillus fumigatus, Trypanosoma cruzi, and HEK293T cells (Nakade et al., 2014; Peng et al., 2014; Zhang et al., 2016). In T. cruzi, it was shown that DSB repair by HR was much more efficient than MMEJ if a template for DNA repair was provided (Peng et al., 2014). Additionally, the MMEJ-based CRISPR technique was also successfully employed to knock down the expression of members of a gene family T. cruzi, suggesting an important tool for studying gene families. Many malaria parasites have large number of repetitive sequences and gene families (Cunningham et al., 2010); genetic manipulation of gene families and repetitive sequences in the parasites have been a difficult task. The mCRISPR strategy we demonstrated may be applied to manipulate and study malaria gene families in a way similar to techniques used for random mutagenesis, although more work needs to be done to show that this technique is a practical method for modifying gene families in malaria parasites.

MMEJ repairs of DSBs requires 5–25 base pair (bp) microhomologous sequences that align at DSB ends and generally results in deletions (McVey and Lee, 2008). Our data from repair of DSBs in the PyCSP-CRR also showed that the majority of the DNA bands amplified from parasite mixtures and from cloned parasite mutants had deletions. We obtained cloned parasites with deletion length ranging from 138 to 516 bp. However, we also observed DNA bands larger than that WT from PCR amplification of mCRISPR modified parasite mixtures and obtained cloned sequences that were larger than the WT sequence, showing that both deletion and insertion mutants can be generated using mCRISPR. These results suggest the mCRISPR could be an important tool for generating genetic polymorphisms in genes with repetitive sequences or gene families of malaria parasites. For the MMEJ based repair, the sgRNAs are designed to target repetitive sequences, and one sgRNA can potentially target more than one site in the repeat region. Different sgRNAs may have different efficiency in cutting the DNA. sgRNA-2 has three targeting sites in the CSP-CRR region and could be more efficient than other sgRNAs in mediating DNA cleavage. The location of a sgRNA may affect cleavage efficiency too. sgRNA-2 and sgRNA-3 are located more centrally than sgRNA-1 and sgRNA-4 and appear to be more efficient in cutting the DNA. Differences in cleavage efficiency may influence the frequency of parasites with WT CSP-CRR region after transfection. Additionally, the lack of drug pressure during development in mosquito may increase the frequency of WT parasite if the WT CSP-CRR is not completely cleaved during blood stage development. There will be variations in deletion/insertion size and location between individual parasites, depending on which and how many sequences are cleaved.

The repetitive region of the CSP protein has been shown to encode protective epitopes, to contribute to modulation of host immune responses, and to be critical for sporozoite development and parasite transmission (Ferguson et al., 2014; Triller et al., 2017). Although isogenic parasite clones without the csp repeat or replacement of csp gene from a different malaria species have been generated using the traditional homologous replacement (Persson et al., 2002; Aldrich et al., 2012; Espinosa et al., 2013; Ferguson et al., 2014), the effect of size polymorphism of CSP-CRR on parasite development has not been carefully investigated. Different parasite strains often have size polymorphism in the CSP-CRR; however, field isolates or strains often have genetic differences in other parts of genomes, making it difficult to evaluate the roles of different lengths of CSP-CRR in parasite development and in modulation of host immune. The mCRISPR can be a good strategy for generating parasite clones that have different length of CSP-CRR with the same genome background for testing the effects of CSP-CRR length variation on parasite development in mosquitoes. We obtained two PyCSP-CRR mutant parasite clones that could not develop to infective sporozoites in mosquito, with one clone having interruption of coding region and the other having a large deletion (172 AA). It is understandable that parasites with a large deletion or disruption of open reading frame of CSP would have more severe consequences. The second largest deletion clone sg2–4-2 has 138 AA deletion. This parasite can still develop into salivary gland sporozoites in two of four batches, but not able to infect mice after mosquito bites. These results suggest that a minimum CSP-CRR length is required for the development in the mosquitoes. It would be interesting to obtain additional clones with deletion of which may still produce infective sporozoites, but may not be able to develop in the liver.

Additionally, the observations of selection for WT and/or the ~900 bp band during blood stage development suggest that either the PyCSP-CRR plays a role in blood stage development or that differential immune responses are triggered by PyCSP with different repeat lengths. However, since the genomes of the mutant parasites were not sequenced to detect potential unknown off-targets or other unknown mutations, it is unclear whether loci other than CSP were affected. Further investigations are necessary to dissect these possibilities.

Supplementary Material

Supplementary Fig S2

Amino acid sequences of mutant PyCSP central repeat region (CRR) obtained from individual parasite clones. The repeats are highlighted with different colors.

Supplementary Fig S3

DNA sequence alignments of the Pycsp obtained from individual parasite clones. sgRNA-2 and sgRNA-3 sequences are indicated by yellow and green background, respectively, based on the wild-type 17XL sequence.

Supplementary Fig S1

Locations of sgRNAs and PCR primers in the Plasmodium yoelii CSP gene sequence (A) and diagram showing MMEJ mediated CRISPR/Cas9 repair mechanism (B).