Improvement to Gene Editing in Neospora Caninum: Knockout ku80 to Increase the Eciency of Homologous Recombination

Background: The CRISPR/Cas9 technology based on homologous recombination has been widely used for gene editing in organism. Gene deletion of ku80 complex can increase the eciency of homologous recombination(HR), or gene editing by CRISPR/Cas9 in Arabidopsis thaliana and Toxoplasma gondii, which has remained elusive in Neospora caninum. Methods: Here, we knocked out the ku80 in Nc1 strain, named Δku80, using CRISPR/Cas9. In vitro, phenotype assays, including plaque formation, invasion, replication and egress were carried out to determination of the growth of Δku80. In vivo, we respectively injected BALB/c mice with different doses of NcKU80 and Nc1, and measured the survival curves and parasites burdens of mice infected. To test the eciency of HR, Δku80 and Nc1 strains were transfected with the same HA-tagged plasmids, and the percentage of parasites with HA-tagged was investigated. Results: It is showed that the growth and virulence of Δku80 kept unaffected, while the eciency of gene targeting via double-crossover at several genetic loci was increased 2 to 3 fold in Δku80. Conclusions: It is concluded that Δku80 can be used as an effective strain for rapid gene editing in N. caninum.


Background
The global prevalence of Neosporosis poses serious harm, causing abortion in cattle, neuromuscular disorder in calves, neuromuscular injury and hind limb paralysis in puppies. Its causative pathogen is Neospora caninum, an obligate intracellular apicomplexan parasite, which is capable of parasitizing nucleated cells of legion mammalian species. N. caninum is highly similar to Toxoplasma gondii in life cycle, morphological and biological characteristics.
To prevent and control Neosporosis, in-deep analysis N. caninum gene function study closely associated with gene editing technology is required. However, most gene editing methods have common problems in practice application, such as off-target binding and low e ciency of spontaneous homologous recombination. This is also an existing issue in N. caninum' study, which is the reason why numerous researches focus on technological optimization. Lei et al. constructed the overexpressing strain of NcROP18, but failed to knock out it [1]. Ma et al. completely knocked out ROP5 and ROP16 genes of N. caninum using homologous recombination, although the screening e ciency was only 1.4% [2,3]. In recent years, CRISPR/Cas9 technology has become the rst choice of genome editing as a transformative DNA genome editing system. Arranz-Solís and Yang successively applied CRISPR/Cas9 technology in N. caninum to improve the effectiveness of genome editing successfully, promoting the study of this parasite [4,5]. In present, several studies have applied CRISPR/Cas9 technology to disrupt gene in N. caninum.
The CRISPR/Cas9 nuclease complex consists of guide sgRNA and Cas9 endonuclease. Cas9 can generate DNA double-strand breaks (DSB) at a speci c target gene site under sgRNA guidance. Once a DSB is produced, it must be repaired because a single unrepaired DSB can result in cell death. Three pathways contribute to DSB repair in eukaryotes: homologous recombination(HDR or HR), classical nonhomologous end joining (c-NHEJ) and alternative end joining (alt-NHEJ). Among these,HR is the primary DSB repair pathway in yeasts and prokaryotes, while higher eukaryotes including N. caninum use c-NHEJ priorly. However, CRISPR/Cas9 technology is based on HR, which generates gene insertion and point mutations by rearranging DNA sequences in the presence of homologous DNA ank sequences.
These factors lead to the low e ciency of CRISPR/Cas9 gene editing in N. caninum [6,7]. To improve the HR e ciency of CRISPR/Cas9 technology in N. caninum, we committed to block c-NHEJ.
KU protein binds on DSB ends to initiate c-NHEJ. Studies have shown that knocking out ku80 signi cantly improves the e ciency of gene editing in Toxoplasma gondii without changing its biological peculiarities [8,9]. In this study, ku80 knockout strain of N. caninum was constructed to inhibit c-NHEJ using CRISPR/Cas9 technology, and the probability of HR during DNA repair was increased.

Cell and Parasites Culture
Vero cells and HFF cell lines were transduced and cultured with 6% and 10% fetal bovine serum (FBS, Gibco, USA) DMEM (Dulbecco's Modi ed Eagle's Medium). N.caninum was cultured in the above two cells with 1% fetal bovine serum DMEM. The culture conditions are 37℃ and 5% CO 2 .

Plasmid and Transgenic Parasite Construction
Primers for PCR ampli cation are shown in Table 1. The gRNA targeting site for NcKU80 (5'-GTTCAGCCTTCAGCTCTCCG-3') was designed in the EuPaGDT Library of ToxoDB, and pCRISPR-Cas9-NcKU80 was constructed based on pSAG1-Cas9-NLS-GFP-SAG1 plasmid preserved in the Key Laboratory of Animal Parasitology (Beijing, China). The method is as described by Yang [5]. The gRNA was intergrated into the plasmid downstream of the 3 'Flank sequence of the U6 promoter by ligation transformation (Vazyme, Nanjing, China).
The knockout plasmid was constructed on the basis of plasmid pTCR-CD retained in the laboratory. 5' and 3' homologous anks of NcKU80 were ampli ed using Nc1 genome as a template. To facilitate screening, the chloramphenicol resistance gene and red uorescent protein gene (RFP) were inserted between 5' and 3' homologous anks, and ligated into plasmid backbone pTCR-CD with anks. The plasmid CRISPR and the plasmid pTCR-NcKU80 were co-transfected into Nc1. And used the limited dilution method to cloned NcΔku80 after chloramphenicol treatment.
The lab-built plasmid CRISPR and pTCR of NcGrxC5-oe, NcGrxC5-HA, NcGrxS14-oe, NcGrxS14-HA was constructed by the method as above. The difference was that these pTCR plasmids were integrated the DHFR gene inserted and the strains were treated with pyrimethamine after co-transfect.
Plaque Assay 200 tachyzoites were added into 12-well plate HFF cells and xed with 4% paraformaldehyde after 8 days, and then stained with 0.2% crystal violet dye. The plaque area was calculated using Photoshop CC 2018 software (Adobe, United States). The calculation was repeated 3 times to take the average. Three independent trials were carried out.

Invasion assay
A total of 1×10 5 tachyzoites were added to the HFF cell that cultured on 12-well plates. After 1 hour, the parasites that didn't attach or invade to cells were washed with PBS. Immuno uorescence assay was performed after cultured for 24 hours. The 12-well plates were observed using a uorescence microscope, and the number of HFF cells and PV were calculated by randomly selecting 5 visual elds from the top, bottom, left, right and middle positions respectively. The invasion capacity of the parasite is the number of PV in the eld divided by the number of HFF cells. The calculation was repeated 3 times to take the average. Three independent trials were carried out.

Replication assay
Cultivate parasites and IFA in the same way as invasion assay. 100 PVs were randomly calculated using a uorescence microscope, and then the number of PVs containing different amounts of tachyzoites was calculated. The calculation was repeated 3 times to take the average. Three independent trials were carried out.

Egress assay
Cultivate parasites in the same way as invasion assay and treated with calcium ionophore A23187 (Sigma-Aldrich, USA) for 3 minutes after 24 hours. IFA was then utilized. The escaping ability of tachyzoites is expressed as the proportion of escaping PVs to the total PVs. Calculate 100 PVs and the calculation was repeated 3 times to take the average. Three independent trials were carried out.

Virulence tests in mice
The 6-week-old female BALB/c mice were purchased from Jinmuyang Laboratory Animal Breeding Co., Ltd. (Beijing, China) and were kept in a sterile cage for a week, ad libitum access to clean food and water.
Five per group mice were intraperitoneally injected with different doses of tachyzoites and were observed for 30 days.
Quantitative Real-Time PCR Parasite quantitative detection in mice by qPCR was carried out following the method Collantes-Fernández reported [10]. The mice were humanely sacri ced at 30 days, then collected brains and extracted DNA using the DNA Extraction Kit (Beijing Aide Lai Biotechnology Co. Ltd., Beijing, China). Primers were designed to quantitatively detect 28S RNA of mice and Nc5 gene of N.caninum by qPCR. Established the standard curve of 28S RNA and Nc5. The parasite load was expressed as the content of N.caninum Nc5 per nanogram of mouse 28S rRNA.

Data Analysis
All data were analyzed using GraphPad Prism 8.0.2 (San Diego, CA, USA), and the results were displayed as mean ± standard deviation and tested by two-way ANOVA and the Student's t test. P-values are indicated as asterisks: * p < 0.05, * * p < 0.01, * * * p < 0.001 and * * * * p < 0.0001. We consider all p < 0.05 to be signi cant.

Deletion of ku80 in N. caninum
The KU domain-containing protein of the N. caninum was searched in the ToxoDB genomic resource database, and two genes were found (NCLIV_056045 and NCLIV_064150). BLASTp analysis of these two genes showed that NCLIV_056045 had a higher percent identity with KU80 protein of Arabidopsis and Toxoplasma gondii, respectively 24.06% and 64.81% (E-value of 3e-10 and 0.0), named Ncku80.
The CRISPR/Cas9 technology platform constructed by Yang et al. [5] was used to delete ku80 in the N. caninum wild-type (WT) strain (Nc1) parasites (Fig. 1A). Identi cation of ku80 genewas carried out using PCR. Two primer pairs on the upstream and downstream of ku80 gene were designed to identify the absence of the target gene. Further, two primer pairs were designed to verify the integration in 5' ank and 3' ank (Fig. 1B). Two gene-deletion strains, named Δku80, were obtained by screening. A red uorescent protein (RFP) coding sequence was added to the 5' end of the chloramphenicol acetyltransferase (CAT) gene, uorescent knockout strains were captured with an inverted uorescence microscope (Fig. 1C).
Knocking out ku80 did not in uence the phenotype of N. caninum in vitro The ideal result of this study is to construct a parent strain that can be used in practical study. Therefore, in addition to increase the e ciency of HR, the morphological, growth rate and virulence of the constructed strain cannot be changed. In terms of morphological characteristics, Δku80 and Nc1 were observed under optical microscope and uorescence microscope after IFA. Both Δku80 and Nc1 strains can form elliptic parasitophorous vacuole (PV), and the tachyzoites have a classical crescent-shaped the same as Nc1 (Fig. 2A).
In terms of growth, N. caninum has a complete lytic cycle in which the tachyzoite rst invades the host cell, and the tachyzoites maintained proliferating in the host cell until su ciently large PV is produced [11]. Subsequently, the PV bursts, and the released tachyzoites invaded other cells. In order to analyze whether knocking out KU80 affects N. caninum' lytic cycles and virulence, phenotype assays, including plaque formation, invasion, replication and egress should be carried out. Plaque formation assays can comprehensively re ect the parasites develop ability in vitro. Both Nc1 and Δku80 led to uniform, clear morphological rules and plaques (Fig. 2B). We randomly selected 50 plaques and calculated their area. T-test analysis showed that Δku80 kept the parallel capability of plaque formation from its parental strain. (P = 0.4029, Fig. 2C).
For invasion assay, IFA showed Δku80 invade to HFF cells normally, which suggests there is no barrier or delay to invasion in the knockout strain (P = 0.3478, Fig. 2D). For replication assay, 100 vacuoles were randomly selected to calculate the number of tachyzoites inside each vacuole. There were no difference between Nc1and Δku80 on the numbers of vacuoles carrying 2, 4, 8, 16 or more tachyzoites (P = 0.9814, Fig. 2D). For the egress assay, Nc1 and Δku80 were stimulated with the calcium ionophore A23187 at the same time. It was observed that Δku80 had similar speed of escaping from HFF cells as Nc1. (P = 0.2748, Fig. 2D). Overall, the above experiments proved that compared with Nc1, knocking out NcKU80 had no effect on the N. caninum phenotype in vitro.
Knocking out ku80 did not in uence the virulence in mice of N. caninum To estimate the effect of knocking out ku80 on the virulence of N. caninum in mice, we respectively injected BALB/c mice with different doses of Δku80 and Nc1. With the injection of 1.5x 10 7 tachyzoites, it was found that two of the ve mice died on 7dpi and one died on 10dpi in Nc1 group. Similarly, ve mice have two deaths on 7dpi and one death on 18dpi in Δku80 group. Statistical analysis using Graph Pad 8.0 found no signi cant difference between the two groups (P = 0.9119, Fig. 3A). With the injection of 2 x 10 7 tachyzoites, two of ve mice respectively died on 7dpi and 13dpi in Δku80 group. While three of the ve mice respectively died on days 7, 9, and 18 in Nc1 group. Statistical analysis demonstrated that the two groups also share the same survival status after injection with same doses (P = 0.6576, Fig. 3B). The mice injected with 1×10 7 tachyzoites in the two groups survived 30 days of continuous observation. On the 31st day, brain tissues of the surviving mice were collected for qPCR to detect the parasite burden in the brain, and the result showed there was no difference in the parasite burden between the two groups (P = 0.1363, Fig. 3C).
Results 2 and 3 illustrated that the growth and the virulence of Δku80 kept unaffected.

Comparison of homologous recombination e ciency between Δku80 and Nc1
Since KU80 plays an important role in NHEJ, it is hypothesized that knocking out Δku80 can block NHEJ, thus increasing the probability of HR. In the studies of N. caninum, overexpression and mutation are the main editing technology. The e ciency of the gene replacement events in the ku80 knockout background can be evaluated by introducing HA-tag into plasmids permits. The HA-tagged mutation (HA) and overexpression (oe) plasmids of glutaredoxins C5(GrxC5) and S14(GrxS14) were constructed previously.
Previous experiments had shown apicoplast localization of GrxC5 and GrxS14, and neither of their overexpression nor gene deletion affected the growth of the N. caninum (unpublished data). Therefore, to determine if the gene replacement e ciency is increased in the ku80 knockout background, we chose the lab-built HA-tagged overexpression to targeted into UPRT locus, and HA-tagged mutation plasmids to GrxC5 and GrxS14 locus. 1×10 7 Δku80 and Nc1 tachyzoites were transfected with an equal volume of HA-tagged plasmid and CRISPR/Cas9 plasmid at the same time. All strains (NcGrxC5-oe, NcGrxC5-HA, NcGrxS14-oe, NcGrxS14-HA) were veri ed by PCR for the insertion of endogenous HA-tags correctly. IFA was used to measure the percentage of HA-tagged in 300-tachyzoites after 10 days of cultivation (Fig. 4A). Each plasmid was transfected 5 times.
The statistical results showed that the gene replacement events of Δku80 was remarkably increased more than twice or even three times that of Nc1, showing a signi cant difference (Fig. 4B-4E). It is worth noting that using Δku80 as parent strains success 90.75% gene replace, while that is 63.25% in Nc1, although no signi cant difference was found in the two groups ( Fig. 4F, P = 0.0678). Above all, our results proved that the knockout of ku80 could signi cantly increase the e ciency of HR in N.caninum.

Discussion
In this study, Ncku80 knockout strain was developed using CRISPR/Cas9 technology, and the phenotype and virulence experiment indicated that the deletion of Ncku80 had no signi cant impact on the growth or the virulence of N.caninum. Since KU80 is an important factor of c-NHEJ DNA repair pathway, the frequency of the gene replacement events of Δku80 was explored by transfecting HA-tagged plasmids. It is proved that the e ciency of HR in Δku80 has raise considerably. Therefore, this strain can be applied to the study of N.caninum to advance the e ciency of gene edition.
CRISPR/Cas9 technology has been introduced into the apicomplexan parasite, such as Toxoplasma, Plasmodium, Cryptosporidium [12] as well as N. caninum, which has considerably promoted the study on these parasites. However, due to the priority of c-NHEJ, the e ciency of HR is relatively low when CRISPR/Cas9 is used for gene insertion or deletion in many organisms. That is the reason why researchers attempted to construct ku gene deletion strains, and some results have been achieved. ku gene homologous sequence deletion strains have been constructed in various fungi [13] by blocking c-NHEJ, the e ciency of HR was increased to nearly 100%. In Arabidopsis, mutations in ku70 can result high-frequency gene targeting [14]. In Toxoplasma, ku80 gene deletion strains have also been constructed, markedly increased gene replacement e ciency to 300 to 400 fold compared to wild-type strains [9].While little efforts were made on ku80 knockout strain in N. caninum. We successfully lled in this gap, and proved the e ciency of HR was signi cantly increased in Δku80.
The e ciency of CRISPR/Cas9 technology has always been an obstacle [15]. Several parameters affect the e ciency of gene targeting in CRISPR/Cas9 technology, including the tness score of the target gene, drug selectable markers, plasmid, the targeting DNA ank lengths, the life force of parasite and so on.
Some methods have been reported to recover the defect, including substituting the N. caninum speci c promoter and terminator [5,15]. In this study, deletion of ku80 was proved to work well.
Although KU70/ KU80-mediated c-NHEJ plays a role in DSB repair in many protozoa, the c-NHEJ factor is often lost in parasites [16]. For example, in Plasmodium, Trypanosoma brucei, and African trypanosomes, the repair of DSB is mediated by a Ku-independent microhomology-mediated end Joining (MMEJ) [17][18][19]. While in N. caninum, three conserved factors of c-HNFJ were retained: KU70, KU80 and Lig4[16]. Our study con rmed the existence of KU80 protein in N. caninum, and it was revealed that the deletion of Δku80 decreases the rate of c-NHEJ, thus increasing HR's. An important issue is the virulence of Δku80, we determined that Δku80 can invade into the host cell, multiply in and release from it, pass across the blood-brain barrier and cause a same injury in the mice as Nc1. It means ku80 knockout background did not lead to develop defect in N. caninum, which are similar to fungi and Toxoplasma [9,13].
Constructing Δku80 without any drug tagging has been tried originally but ended in failure. In order to broaden the application range of this strain, the knockout of the chloramphenicol acetyltransferase gene is underway. It may be easier to successfully generate in the ku80 knockout background. When using this strain for gene editing, selectable markers other than chloramphenicol could be used for positive or negative selection. Positive selection including DHFR (resistance to pyrimethamine), HXGPRT (resistance to Mycophenoloic acid/xanthine), while negative selection including UPRT (resistance to 5uorodeoxyuracil), SNFR1 (resistance to Sinefungin) and HXGPRT (resistance to 6-thioxanthine) [20].
When four plasmids were used for transfecting in the ku80 knockout background, the success rate of gene replacement was unchanged, additionally the e ciency was more than doubled. It suggested that HR might play a less important role in the success rate of transfection than other factors, such as the target gene locus, the mass of the plasmid and drug screening marker. The dramatic improvement of HR e ciency in Δku80 makes it easier to screen the monoclonal strains. Moreover, the success rate of overexpression was higher than that of mutation, which may be due to different drug markers used ( oxuridine, FUDR in oe and DHFR in HA). It demonstrated the value of pharmacological markers in gene editing.

Conclusions
A strain was constructed with ku80 deletion, which improved the e ciency of homologous recombination and could improve the e ciency of gene editing in the study of N.caninum. Animal employee ID: 18049). The surviving mice infected with 1×10 7 doses were injected with atropine (0.02mg/kg) at 30dpi, and were humanely sacri ced by cervical dislocation.

Consent for publication
Not applicable Availability of data and materials Data supporting the conclusions of this article are included within the article.