Improved CRISPR-Cas9 Gene Editing in Leishmania to Modify Multiple Genes, Create Chromosome Changes, and Study DNA Repair

We previously showed that the addition of an oligonucleotide donor improved CRISPR-Cas9 gene disruption efficiency in L. donovani (7). In some cases, however, a gene-editing task, such as the introduction of an epitope tag or a point mutation, may involve a site where only low-activity gRNAs can be designed, such as the gRNAc-targeting site in the L. donovani miltefosine transporter gene MT (7). We therefore wanted to determine whether sequential transfections of an oligonucleotide donor containing 25-nucleotide-long flanking homologous sequences to the Cas9 break site would increase the gene-editing efficiency and improve the isolation of the edited mutants (Fig. 1A; see also Data Set S1 in the supplemental material). The MT gene was selected for evaluating CRISPR-Cas9 gene-editing efficiency (frequency), since mutations (insertions, deletions, and selected point mutations) in the MT gene lead to survival (resistance) in the presence of miltefosine (7). Twenty-one days following transfection of vectors expressing gRNAc and Cas9 in L. donovani (21 days is the time required to select for cells with stable expression), sequential transfections with an oligonucleotide donor containing stop codons were carried out every 3 days for a total of four transfections (Fig. 1B). The miltefosine resistance rate was determined 3 days after each oligonucleotide donor transfection. As shown in Fig. 1C, multiple transfections increased the miltefosine resistance rate significantly, compared to gRNA expression alone. At the end of the fourth donor oligonucleotide transfection, the final miltefosine resistance rate in these gRNAc-expressing cells increased from 0.4% without a donor to nearly 10%, representing a 25-fold increase. This demonstrated that sequential transfections of an oligonucleotide donor can significantly improve CRISPR-Cas9 gene editing (oligonucleotide donor-directed repair) efficiency to make it easier to isolate the edited mutants. Since the growth rate can be altered in edited mutants, it is necessary to clone transfectants into 96-well plates within 2 days after the third or fourth donor transfection. We have successfully used this strategy to generate L. major and L. mexicana centrin gene null mutants (data not shown) and the single-amino-acid substitution mutant for L. donovani RAD51 (see below).

One of the key advantages of CRISPR-Cas9 technology is that the Cas9/gRNA complex will continually scan the genome and generate double-strand DNA breaks until all the targeting sites in the genome have been deleted or mutated. This has been used successfully to inactivate multicopy family genes and endogenous retroviruses in other cell types (9, 10). A2 is a multicopy gene family and an important virulence factor for visceral Leishmania infection (6, 8, 11–13). There are at least 11 copies of the A2 gene of different sizes in L. donovani 1SC12D (8, 12) (Fig. 2A). Due to these multiple copies that also alternate with another gene termed A2rel, it was not possible to delete this gene family from L. donovani using the conventional gene-targeting approach (8). We therefore attempted to delete all copies of this multigene family from chromosome 22 of L. donovani by using the CRISPR-Cas9 method. A2 gene-coding sequences are mainly composed of 30-nucleotide repeat sequences encoding 10 amino acid repeats (11, 14) (Fig. 2A). To avoid these repeated sequences from being used as an HDR template, two gRNAs targeting the unique sequences (one targeting near the 5′ end of the A2 coding sequence and the other targeting the 3′ untranslated region) (Fig. 2A) in each of the A2 genes were designed and cloned into the dual gRNA expression vector (Fig. 2B; Data Set S2). In addition to targeting A2 genes, we also investigated whether it was possible to increase the efficiency of A2 gene deletion by coselecting for cells with CRISPR-Cas9 activity. Coselection for CRISPR-Cas9 activity was performed by targeting the L. donovani miltefosine transporter gene (LdMT) and selecting for miltefosine resistance at the same time as targeting the A2 genes. The rationale was that if one gRNA were used to target the LdMT gene and the other gRNA(s) targeted a different gene of interest (in this case, A2), following selection for miltefosine resistance the A2 gene would be targeted with a higher frequency. Thus, the LdMT gRNAa coding sequence was added into the gRNA A2a+b construct to generate a triple gRNA expression vector to determine whether A2 genes could be more efficiently deleted in these miltefosine-resistant cells (Fig. 2B).

As shown by Western blotting analysis in Fig. 2C, with prolonged culture and without cloning, the A2 genes could be completely deleted (inactivated) with this CRISPR system (Fig. 2C, lane 5). However, coselection for CRISPR-Cas9 activity with miltefosine significantly reduced the time needed to delete all A2 genes, from 4 months with no selection (Fig. 2C, lane 5) to 6 weeks with miltefosine selection (Fig. 2C, lane 4). In contrast, in the absence of miltefosine there remained detectable but diminished A2 protein expression at 6 weeks (Fig. 2C, lanes 2 and 3). Six weeks is the minimum time to establish miltefosine-resistant cells in culture. The complete deletion of A2 genes in these L. donovani cells was further confirmed by PCR analysis with A2-specific primers (Fig. 2D). This demonstrated that this CRISPR system can be used to delete multicopy family genes and that coselection for CRISPR-Cas9 activity can significantly reduce the time needed to obtain these deletion mutants. This is valuable, since there are many tandem gene arrays and supernumerary chromosomes in various Leishmania species and isolates (3–6).

This coselection observation is similar to what was reported for Caenorhabditis elegans and in human cells, where the co-CRISPR strategy has greatly facilitated detection of genome-editing events; particularly, CRISPR cotargeting the hypoxanthine phosphoribosyltransferase (HPRT) gene in human cells followed by 6-thioguanine selection highly enriched the cotargeting gene edited mutants (15–17). This also agrees with the observation of a bimodal distribution of CRISPR inactivation of porcine endogenous retroviruses (PERVs), where only 10% of clones exhibited complete disruption of all 62 copies of PERV pol genes and the remaining clones exhibited no or a low level of editing (10).

Targeted chromosomal translocations have been successfully generated in various organisms by using CRISPR-Cas9 technology, including various cancer models (18–21). The ability to generate targeted chromosomal translocations in Leishmania will help to investigate gene expression and identify mechanisms for the initiation and termination of polycistronic transcription and chromosome stability (22–24). To determine whether specific chromosomal translocations could be generated in Leishmania, a dual gRNA expression plasmid was constructed where one gRNA was targeted to the nonessential multidrug resistance gene Ld241510↗ in chromosome 24 (25) (Data Set S3) and the other was targeted to the miltefosine transporter gene (LdMT) in chromosome 13 (7) (Fig. 3A). The hypothesis was that although most of the dsDNA breaks (DSBs) generated by Cas9 nuclease in chromosomes 13 and 24 would be repaired by intrachromosomal joining, some interchromosomal (translocation) joining could also occur. Following transfection and miltefosine selection, genomic DNA was extracted from the surviving cells, and various PCR primer pairs (one primer specific for chromosome 13 and the other for chromosome 24, each close to one of the two DSB sites) (see Data Set S3 for details), were used to investigate chromosomal translocation. As shown in Fig. 3B, it was possible to detect all four types of chromosomal translocation events resulting from the two DSBs generated simultaneously in chromosomes 13 and 24. This demonstrated that these translocations can occur regardless of the polycistronic transcription direction or the size of the new chromosome generated by translocation. In addition, following cotransfection of two oligonucleotide donors that promoted competitive translocation events (type II or type IV), both translocation PCR products were detected at similar frequencies (Fig. 3C). This further suggests that the frequency of chromosomal translocation events may not be affected by the direction of polycistronic transcription.

It is interesting that the two gene clusters from chromosomes 13 and 24 were joined together back to back in the type I chromosomal translocation, despite transcription going in opposite directions. In contrast, the other two gene clusters were joined together head to head in the type II chromosomal translocation. Since these two gene clusters in type I chromosomal translocation lose their corresponding transcription initiation sequences, it will be interesting to see how the transcription levels of these gene clusters from the parent chromosomes are affected by these chromosomal translocations. To our knowledge, this could be the first example to show that all four types of chromosomal translocations can be generated (18–21). Interestingly, we also detected a chromosomal translocation event (Fig. 3D) in Leishmania cells transfected with the triple gRNA expression vector described above in Fig. 2B, in which two gRNAs were targeted to the A2 loci in chromosome 22 and one gRNA was targeted to the MT locus in chromosome 13.

As expected, in the absence of oligonucleotide donors, all the chromosomal translocations were joined by MMEJ. It is interesting that two MMEJs were observed in both the detected type III and IV chromosomal translocations. In the type III chromosomal translocation detected, one MMEJ (TCCAC) joined sequences from each side of the break in chromosome 13 before the second MMEJ made the chromosomal translocation joint, which used only a 3-bp microhomology sequence (CCA). In the detected type IV chromosomal translocation, a section of chromosome 24 sequence (more than 200 bp) was reversed, which was likely caused by a flip of the single-strand DNA created by an end resection after the double-strand break and the intramolecular MMEJ (ACGACACCAT). Though the chromosomal translocation events were relatively rare and were enriched by cotargeting the LdMT gene in the current study, it should be feasible to isolate chromosomal translocation mutants in other specific targeting sites by using donors containing drug selection markers.

Due to its precise transcription initiation and termination, the RNA polymerase III U6 promoter has been widely used to drive gRNA expression in higher level eukaryotic cells and other protozoan parasites, including L. major (26–30). In contrast, the RNA polymerase I rRNA promoter used for gRNA expression has only been reported in L. donovani (7) and Trypanosome cruzi (31). We therefore compared genome-editing efficiency when gRNA expression was under control of these two different RNA polymerase promoters. An LdMT (L. donovani miltefosine transporter gene) gRNAa expression vector using the L. donovani U6 promoter was constructed and compared to the rRNA promoter in the pSPneogRNAaH vector, which expresses the same LdMT gRNAa previously described (7). In addition, we made an expression vector in which the human U6 promoter was used to direct LdMT gRNAa expression (Fig. 4A; Data Set S4). These LdMT gRNAa expression vectors were transfected into Cas9-expressing L. donovani promastigotes, and resistance to miltefosine was determined. As shown in Fig. 4B, the miltefosine resistance rate was much lower in the LdU6 promoter vector-transfected cells than in the rRNA promoter vector-transfected cells at 32 days posttransfection, indicating that the rRNA promoter is more efficient at driving gRNA expression in L. donovani. Interestingly, while the miltefosine resistance rate for cells using the rRNA promoter appeared to be stabilized at 12 to 25% in prolonged culture, the miltefosine resistance rate in cells using the LdU6 promoter was able to reach a similar level after a much longer time period of 73 days posttransfection (Fig. 4B). It is also important to note that because the rRNA promoter is a stronger promoter than the LdU6 promoter (32, 33), it was much easier to obtain G418-resistant transfectants from the rRNA promoter vector than from the LdU6 promoter vector-transfected Leishmania cells (data not shown). Surprisingly, the human U6 promoter was also functional in Leishmania and could mediate gRNA expression, though with a much lower targeting efficiency. Taken together, this comparison demonstrated that the rRNA promoter is much better than the U6 promoter in driving gRNA expression in Leishmania. Thus, it would be interest to explore the RNA polymerase I promoter for improving gRNA expression in other organisms, including human cells.

To further simplify this genome-editing system in Leishmania, single coexpression vectors (pLdCN and pLdCH) were constructed where the transcription of gRNA, Cas9, and a drug selection maker (neomycin or hygromycin resistance gene) were placed under control of the same rRNA promoter (Fig. 5A; Data Set S5). The LdMT gRNAa sequence was used to test whether these vectors could function as predicted. As shown in Fig. 5B, although there were some variations in the targeting efficiency, miltefosine-resistant cells were obtained with both pLdCNgRNAa and pLdCHgRNAa, demonstrating that the gRNA and Cas9 nuclease were properly expressed in both of these coexpression vectors. It is important to note that the L. donovani ribosomal promoter also functions well in other Leishmania species (32, 33). So far, we have successfully used the pLdCN vector to delete genes in L. major, L. mexicana, and L. donovani (data not shown). Therefore, a single appropriately designed gRNA construct could be used to target a conserved site (a 20-bp conserved sequence plus NGG, known as the protospacer-adjacent motif) in all three and perhaps more Leishmania species.

We previously observed that DSBs created by CRISPR-Cas9 in Leishmania were repaired by interallelic HDR, which results in error-free repair, or by MMEJ, which results in deletion mutations (7) (see Fig. 7B, below). RAD51 is a DNA recombinase required for HDR during DSB repair and is not involved in MMEJ (34–37). To determine whether inhibition of the HDR pathway would increase the frequency of MMEJ-mediated DSB repair to improve CRISPR-Cas9-directed gene inactivation efficiency, we attempted to use this CRISPR system to disrupt the L. donovani RAD51 gene. As shown in Fig. 6A (see also Data Set S6), a plasmid carrying an L. donovani RAD51-specific gRNA was transfected into Cas9-expressing L. donovani cells, and subsequently a single-strand oligonucleotide donor with stop codons or a bleomycin resistance marker donor PCR product was cotransfected into these cells as described for the experiments shown in Fig. 1.

Although it was possible to disrupt one of the L. donovani RAD51 alleles by using the donors with or without phleomycin selection, as verified by PCR analysis after cloning, one wild-type RAD51 allele remained in the surviving cells (Fig. 6A and B). Even after two RAD51 alleles were disrupted when we used a combination of the stop codon oligonucleotide donor and the bleomycin marker gene donor, a third wild-type RAD51 allele still persisted (Fig. 6A and B). The RAD51 gRNA-expressing cells did however proliferate slower than control gRNA-expressing cells, likely because the RAD51 gRNA and Cas9 complex were continually targeting the remaining RAD51 allele (Fig. 6C). Indeed, after cloning the RAD51 gRNA-expressing cells (Ld RAD51⁺/⁻) in 96-well plates, at least 10 of these single-cell clones died out after continuous culture for 2 weeks. Interestingly, after cloning, many of these putative RAD51 null mutants were able to continue multiplying slowly, as a clump to as many as 100 parasites before crashing, indicating it would take some time to dilute and degrade the remaining wild-type RAD51 mRNA and proteins in these null mutant cells (Fig. 6D). Taken together, this demonstrated that like in mammalian cells, the RAD51 gene is essential for L. donovani.

In contrast to L. donovani, a RAD51 gene null mutant has been generated in L. infantum (37). Since a low frequency of homologous recombination events could still be detected in L. infantum RAD51 null mutants, L. infantum may have developed ways (other recombinases) to compensate for RAD51 deficiency (37). Interestingly, while it was possible to generate RAD51 null mutants in Trypanosoma brucei and L. infantum, RAD51 could also be essential for L. major and T. cruzi, as no RAD51 null mutants have been reported in these latter parasites despite attempts (38-41).

Although it was not possible to disrupt or introduce stop codons into both wild-type RAD51 alleles in surviving L. donovani cells, we were able to edit both RAD51 alleles with an oligonucleotide donor designed to generate a single conserved amino acid substitution (A189G) (Fig. 6A, E, and F). This may be the first example of engineering a chromosomal single amino acid change in Leishmania, further revealing the importance and versatility of CRISPR-Cas9 gene editing for Leishmania.

Since we were not able to generate a RAD51 null mutant, as it is essential for L. donovani, we attempted to inhibit RAD51 activity by using the RAD51 inhibitors B02 and RI-1 (42–44). We reasoned that impairing HDR with these inhibitors could induce higher levels of MMEJ and increase CRISPR-Cas9 gene inactivation efficiency. However, there was no increase in LdMT gene inactivation through MMEJ in LdMT gRNAc-expressing cells after we used various concentrations of these RAD51 inhibitors (Fig. 7A), suggesting that HDR may play a less dominant role than previously anticipated for DSB repair in Leishmania (7). The B02 and RI-1 inhibitors did, however, confirm that RAD51 is essential, since L. donovani cells could not survive in medium containing more than 15 µM B02 or 70 µM RI-1, though the general toxicity of these RAD51 inhibitors might also have contributed in part to the death of these Leishmania cells.

The role that HDR and MMEJ play in DSB repair was further investigated by comparing the repair rates of MT genes in wild-type L. donovani (LdMT^+/+) cells and L. donovani cells with one MT gene replaced with a bleomycin gene (LdMT^+/−) by using traditional gene replacement. LdMT^+/+ and LdMT^+/− mutants were then transfected with vectors expressing Cas9 and three different LdMT gene-targeting gRNAs designated a, b, and c (7). If HDR indeed played a major role in DSB repair, the miltefosine resistance rate (i.e., mutation rate) would be expected to be much higher in LdMT^+/− cells than in wild-type LdMT^+/+ cells. This is because there is no wild-type MT allele available as an HDR template, and cleavage of the single wild-type MT allele would result in miltefosine resistance either through HDR or through MMEJ (Fig. 7C). In the wild-type LdMT^+/+ cells, the majority of the DSBs generated in one of the MT alleles would be repaired by HDR (error free) with the remaining intact wild-type allele as the template, and it would therefore take longer for resistance to develop (Fig. 7B). Interestingly, however, there was no increase in miltefosine resistance rates in LdMT^+/− cells compared to the wild-type LdMT^+/+ cells (Fig. 7D). This strongly suggests that MMEJ is more efficient than HDR in DSB repair in Leishmania, which agrees with the above RAD51 inhibition data.

Given that significant increases in gene-editing efficiency have been observed when using oligonucleotide and double-stranded DNA donors with only 25 nt (bp) microhomology flanking sequences in this and our previous studies (7), these donors could actually be the ideal templates for MMEJ rather than the HDR templates (45, 46). This interesting alternative is important, as we and authors of study reports on other organisms have previously attributed HDR as the sole mechanism of oligonucleotide donor-directed DSB repair (7, 47, 48). This would not, however, explain why the double-stranded donors with long homology arms (100 to 1,000 bp) were often less efficient in DSB repair than the short single-strand oligonucleotide donors (47, 48). These results also argue that Cas9 cleavage is quite efficient; it can quickly cleave the second allele following the first allele cleavage or it can generate DSBs on both alleles simultaneously.

In summary, we have described the optimization and simplification of CRISPR-Cas9 genome editing in Leishmania. Sequential transfections of oligonucleotide donors and cotargeting MT significantly increase gene-editing efficiency. The single gRNA and Cas9 coexpression vectors that were successfully used in all three tested species further simplify this CRISPR-Cas9 system for Leishmania. With this optimization, it was possible to delete the multicopy A2 gene family and to generate targeted chromosomal translocations, which will further advance studies on pathogenesis and polycistronic transcription control in Leishmania. By directly observing the death of CRISPR-targeted RAD51 null mutant clones, we demonstrated that the RAD51 DNA recombinase is essential for L. donovani, confirming that this CRISPR system can be effectively used to determine the essentiality of a Leishmania gene. We have presented evidence to argue that MMEJ plays a more important role than HDR in DSB repair in L. donovani. The CRISPR-Cas9 technology described within has greatly improved the ability to manipulate the Leishmania genome and provides new knowledge about DNA repair in L. donovani. We foresee that CRISPR-Cas9 will soon be widely used in Leishmania research and will eventually help control and eliminate leishmaniasis.

L. donovani 1 S/Cl2D, L. major Friedlin v9, and L. mexicana (MNYC/BZ/62/M379) used in this study were routinely cultured at 27°C in M199 medium (pH 7.4) supplemented with 10% heat-inactivated fetal bovine serum, 40 mM HEPES (pH 7.4), 0.1 mM adenine, 5 mg liter⁻¹ hemin, 1 mg liter⁻¹ biotin, 1 mg liter⁻¹ biopterine, 50 U ml⁻¹ penicillin, and 50 µg ml⁻¹ streptomycin. Cultures were passaged to fresh medium at a 20-fold dilution once a week. Leishmania cells after transfection with the CRISPR vector and donor DNA were sometimes cultured at 33°C or 37°C for 2 to 3 days to improve gene-editing efficiency (7).

All primer sequences used in this study are listed in the supplemental material (). Data Sets S1 to S6

The LdU6gRNAaH plasmid was generated as follows. (i) A 184-bp PCR fragment containing the 98-bp L. donovani RNA polymerase III U6 promoter was amplified from L. donovani genomic DNA with primers LdU6pF and LdU6pR, which also contains the LdMT gRNAa guide coding sequence, and digested with HindIII and BbsI. (ii) The fragment from step 1 was cloned into the HindIII and BbsI sites of the pSPneogRNAH plasmid (7) after removing the 180-bp L. donovani rRNA promoter to create the LdU6gRNAaH plasmid.

The humanU6gRNAa plasmid was generated as follows. (i) A 740-bp PCR fragment containing the human U6 promoter and gRNA coding sequence was amplified from plasmid pX330 (26) with primers pX330gRNAF1 and pX330gRNAR. (ii) The fragment from step 1 was digested and cloned into the HindIII and BamHI sites of pSPneo vector to create HumanU6gRNA plasmid. (iii) The LdMT gRNAa guide coding sequence was then inserted into HumanU6gRNA plasmid after BbsI digestion.

The pLdCH plasmid was generated as follows. (i) A 495-bp PCR fragment containing the 180-bp L. donovani rRNA promoter, the gRNA and HDV ribozyme coding sequences, and the 92-bp pyrimidine track was obtained from the pSPneogRNAH plasmid (7) with primers LdrRNApNde1 and pSPneoRHind3. (ii) The PCR fragment from step 1 was cloned into NdeI and HindIII sites of the pSP72 vector. (iii) The 6,595-bp HindIII and BglII fragment containing the Cas9 and hygromycin resistance genes from pLPhygCas9 (7) was subsequently cloned into the corresponding sites of the pSP72 vector in step ii to generate the pLdCH plasmid.

The pLdCN plasmid was generated as follows. (i) A 495-bp PCR fragment containing the 180-bp L. donovani rRNA promoter, the gRNA and HDV ribozyme coding sequences, and the 92-bp pyrimidine track was obtained from the pSPneogRNAH plasmid with primers LdrRNApXho1 and pSPneoRHind3. (ii) The PCR fragment from step i was cloned into the XhoI and HindIII sites of pSP72 vector. (iii) The 4.4-kb CAS9 fragment from the pLPhygCas9 plasmid was cloned into the HindIII and BamHI sites of the pLPneo vector (49) to generate pLPneoCas9. (iv) The 6.4-kb HindIII and BglII fragment containing the Cas9 and neomycin resistance genes from the pLPneoCas9 plasmid was subsequently cloned into the corresponding sites of the pSP72 vector in step ii to generate the pLdCN plasmid.

The gRNA 241510+MT(gRNAa) coexpression vector was generated by inserting the 360-bp HindIII and BamHI fragment, which contains the LdrRNAP and gRNA 241510 coding sequence from the pSPneogRNA241510H plasmid (7), into the HindIII and BglII sites of the pSPneoHHgRNAaH plasmid (7) after removing the 180-bp LdrRNAP sequence.

The gRNA A2a+b coexpression vector was generated as follows. (i) A 276-bp PCR fragment containing gRNAA2a, HDV, and hammerhead ribozymes and the gRNAA2b guide coding sequences was amplified with primers Ld220670a and Ld220670b from the gRNA 241510+MT coexpression vector. (ii) The PCR product from step i was digested with BbsI and inserted into the BbsI-digested pSPneogRNAH vector.

The gRNA A2a+b+MT(gRNAa) triple expression vector was generated by inserting the 579-bp HindIII and BamHI fragment which contained LdrRNAP, gRNA A2a+b, and ribozyme coding sequences from the gRNA A2a+b coexpression vector into the HindIII and BglII sites of pSPneoHHgRNAaH after removing the 180-bp LdrRNAP sequence.

Since current gRNA design tools developed from data for higher-order eukaryotic cells are not necessarily suitable for Leishmania (7), we selected gRNA guide sequences based on the relatively high activity scores in all of the following three design programs and no off-target site. The Eukaryotic Pathogen CRISPR guide RNA design tool (EuPaGDT) (http://grna.ctegd.uga.edu/↗) provides useful information on off-target sites and microhomology sequences flanking the DSB (9). The microhomology sequences are required for MMEJ but should be avoided if a donor will be used. Sequence scan for CRISPR (SSC) (based on human and mouse data; http://crispr.dfci.harvard.edu/SSC/↗) and CRISPRscan (based on zebrafish data; http://www.crisprscan.org/?page=sequence↗) are helpful for predicting gRNA activity based on the guide sequence alone (50, 51).

The primers for PCR were selected manually or via Primer3 (http://bioinfo.ut.ee/primer3/↗). All primers and oligonucleotides used in this study were ordered from Alpha DNA (Montreal, Canada).

The following experimental procedures were performed as previously described: single gRNA guide sequence cloning into various gRNA expression vectors (7); Leishmania transfection and a limiting dilution assay to determine miltefosine resistance rates (7); Leishmania genomic DNA extraction, PCR, and sequencing analysis (7); A2 Western blot analysis (14).

The stock solutions (10 mM) of Rad51 inhibitors B02 and RI-1(catalog numbers SML0364 and 1274; Sigma) were prepared in dimethyl sulfoxide and stored at 4°C (42–44). Immediately before use, the stock solutions were diluted with Leishmania culture medium to a 1 mM working concentration for RI-1 and 100 µM for B02. These inhibitors were then directly added into Cas9 and LdMT gRNAc-expressing Leishmania culture (1 × 10⁶ promastigotes per ml) to a final concentration 0 to 20 µM for B02 and 0 to 100 µM for RI-1. The proper concentrations of these inhibitors in culture were maintained by adding fresh inhibitors once every 3 days for a total 4 times in a 2-week period before measuring the miltefosine resistance rate.

The specific LdMT bleomycin-targeting fragment was generated by overlapping PCR. (i) The 666-bp LdMT 5′ flanking sequence with primers Ld1315905′F and Ld1315905′R, the 534-bp bleomycin expression cassette with primers 131590BleF and 131590 BleR, and the 656-bp LdMT 3′ flanking sequence with primers Ld1315903′F1 and Ld1315903′R were PCR amplified separately. (ii) The three PCR fragments from step i were mixed and used as PCR template for primers Ld1315905′F and Ld1315903′R to generate the specific 1,857-bp LdMT bleomycin-targeting fragment. L. donovani promastigotes transfected with this LdMT bleomycin-targeting fragment were selected with 100 µg/ml phleomycin. The LdMT^+/− single-knockout clones were verified by PCR with primer pair Ld1315905′F1 and 131590BleR.

The pLdCN, pLdCH, and pSPneogRNA241510+MT plasmids have been deposited in Addgene under numbers 84290, 84291, and 84292, respectively. The partial sequences for these plasmids are provided in the supplemental material (and). Data Sets S3 S5