Generation of Leishmania Hybrids by Whole Genomic DNA Transformation

Genetic exchange is a powerful tool to study gene function in microorganisms. Here, we tested the feasibility of generating Leishmania hybrids by electroporating genomic DNA of donor cells into recipient Leishmania parasites. The donor DNA was marked with a drug resistance marker facilitating the selection of DNA transfer into the recipient cells. The transferred DNA was integrated exclusively at homologous locus and was as large as 45 kb. The independent generation of L. infantum hybrids with L. major sequences was possible for several chromosomal regions. Interfering with the mismatch repair machinery by inactivating the MSH2 gene enabled an increased efficiency of recombination between divergent sequences, hence favouring the selection of hybrids between species. Hybrids were shown to acquire the phenotype derived from the donor cells, as demonstrated for the transfer of drug resistance genes from L. major into L. infantum. The described method is a first step allowing the generation of in vitro hybrids for testing gene functions in a natural genomic context in the parasite Leishmania.


Introduction
Leishmaniasis is a complex of diseases caused by parasites of the genus Leishmania for which there is an estimated 12 million people infected in tropical and subtropical areas of the world [1]. The clinical manifestations of the disease are often correlated with the infecting Leishmania species and range from self-healing cutaneous sores (cutaneous leishmaniasis) to deadly visceral pathologies (visceral leishmaniasis) [1]. The Leishmania genomes are characterized by a high degree of gene synteny and contain a surprisingly low number of species-specific genes relative to clinical diversity [2,3]. While Leishmania parasites are usually considered as diploid, recent evidence revealed a sizable variation in chromosome copy numbers between species [3,4]. Hence, the tropism of leishmaniasis is likely to come from a combination of species-specific genes identified to date [2,5] and from differences in gene copy number and expression between species [6,7].
The occurrence of a sexual cycle for Leishmania has long been debated but the abundance of hybrid parasites described in nature [8][9][10] suggested that the exchange of genetic material can occur in the field. These observations received experimental confirmations where genetic crosses were made between Leishmania strains in the sand fly vector [11][12][13]. While powerful, achieving the generation of hybrid parasites in vitro would bring a distinct advantage for studying the contribution of genetic loci to the expression of particular phenotypic traits. Since its inception in the early 90's, gene transformation by electroporation has changed the field of parasitology, enabling constant progress in reverse genetic tools. However, these tools allow functional studies mostly at the level of single genes, at least in the natural genomic context [14].
Also, highly homologous sequences are required between the donor and recipient DNAs for successful recombination in Leishmania [15], which prevent, in general, the use of single constructs for inactivating the same genes in different species. Cross-species gene replacement was nonetheless described once in L. donovani [16]. Given the frequency of hybrid parasites in the field and their importance in shaping Leishmania population's heterogeneity, we assessed the possibility of generating cross-species recombinants in vitro by heterologous genomic DNA (gDNA) transfection. We describe a knock-in protocol based on whole genome transformation (WGT) that will be useful for studying the role of species-specific genomic loci and of nucleotide polymorphisms pertaining to the biology of Leishmania. We show that genomic regions up to 45 kb can be selectively transferred between Leishmania species and that inactivating the mismatch repair gene MSH2 can further facilitate the recovery of cross-species hybrids.

Methods
Cell lines, culture conditions and transfections L. major Friedlin, L. infantum JPCM5 and L. infantum (MHOM/ MA/67/ITMAP-263) promastigotes were grown at 25uC in SDM-79 medium supplemented with 10% heat inactivated fetal bovine serum and 10 mg/ml hemin. Transfectants were selected with 300 mg/ml of hygromycin (HYG); 1,500 mg/ml of paromomycin (PM); 100 mg/ml of blasticidin (BLA); 120 mg/ml of puromycin (PUR); or 40 mg/ml of neomycin (NEO). Electroporation was done in 2-mm cuvette using 400 ml of cells (1610 8 of parasites) resuspended in HEPES-NaCl buffer [17] at 500 mF, 450 V (2.25 kV/cm) as previously described [18]. Transfection efficiency was evaluated by transfecting 1610 8 parasites with 20 mg of gDNA or 5 mg of linear digested DNA, taking into account the number of colonies obtained and the number of parasites transfected. Following electroporation, parasites were grown in drug-free media overnight and then plated on SDM-agar (1% Noble Agar, Nunc.) containing the appropriate drug using the same concentration as in liquid SDM-79 medium. Colonies were counted from SDM-agar plates after growing 10 to 14 days at 25uC and the transformation efficiencies expressed per 10 mg of DNA.
The L. major Friedlin MF80.3 mutant was selected from a cloned parental population using a stepwise selection until they were resistant to 80 mM of miltefosine (MF) [19]. Miltefosine (MF) was purchased from Cayman Chemical (Ann Harbor, USA) and trivalent antimony (SbIII) was purchased from Sigma (Saint Louis, USA). N-methyl-N9-nitro-N-nitrosoniguanidine (MNNG) was purchased from Sigma (Saint Louis, USA) and was dissolved in 100% DMSO. Growth curves were obtained by measuring absorbance at 600 nm as previously described [20].

DNA manipulations
Total DNA was isolated using DNAzol reagent (Invitrogen) as recommended by the manufacturer. For quantitative Southern blots, the genomic DNA was digested with appropriate restriction enzymes and migrated in 0.8% agarose gels. Southern blots, hybridizations and washes were performed following standard protocols [21].
For every gene inactivation cassette, two pairs of primers were used to amplify regions of 500-600 bp upstream and downstream of the target gene. These DNAs were then fused to the NEO, HYG or BLA genes by overlap extension PCR [22,23]. After electroporation, the integration of the inactivation cassette was confirmed by PCR and Southern blots. The inactivation of the gene pyridoxal kinase (LmjF30.1250) in L. major was previously described [19]. The pSP72-aBLAa-LRP plasmid was generated by cloning the Leucine Rich Repeat Protein (LRP) gene amplified from L. major with primer F-LRP (59 GCG GATATCGCTGTTGGTGTTCGTGTCGTC) and Primer R-LRP (59 GCG ATCGATCAGAGGCG-GAGTGGGCTGTCC) into EcoRV and ClaI restriction sites of the pSP72-aBLAa-vector. The pSP-aPURa-MSH2 plasmid was generated by cloning the MSH2 gene amplified from L. infantum with primer F-MSH2 (59 CGCTCTAGACGCACATGCACC-TACGCACG) and Primer R-MSH2 (59 CGCTCTAGACAAA-CAAGGATAGCGAGAAG) into the single XbaI restriction site of the psp72-aPURa vector. All the primers used for these constructs are listed in Table S1.
The pulse field gel electrophoresis was performed as previously described [24]. Briefly, we prepared low-melting point agarose blocks containing Leishmania parasites resuspended in HEPES buffer at a cell density of 1610 8 parasites/ml. Parasites were lysed by incubating the blocks in lysis buffer (0.5 M EDTA [pH 9.5]; 1% SLS; 50 mg/ml of proteinase K). Chromosomes were resolved by a BioRad (Hercules, California, USA) contour-clamped homogeneous electric field (CHEF) mapper at a constant temperature of 14uC. Saccharomyces cerevisiae chromosomes were used as molecular weight markers. Chromosomes were revealed by ethidium bromide staining.

DNA amplification and sequencing
Multilocus sequencing analyses were performed with Leishmania hybrids by amplifying the genes located in the vicinity of the integrated selection marker using primers specific for genomic regions conserved between L. major and L. infantum. Primers were designed with Primer3 [25] and are listed in Table S2. PCR products were amplified using Phusion High Fidelity DNA polymerase (New England Biolabs, Inc.) and sequenced by conventional Sanger sequencing. Sequences were analyzed with the Lasergene software (DNASTAR, Inc.) and compared to the L. major and L. infantum sequences available at GeneDB (www. genedb.org).

Whole genome sequencing and analysis
Genomic DNAs were prepared from mid-log phase cultures of L. infantum 263 transfected with gDNA derived from L. major MF80.3DLmjF13.1540::NEO/LmjF13.1540 and from the clone 1 of L. infantum JPCM5 transfected with gDNA derived from L. major LmDLRP::NEO/LRP as previously described [26]. Their sequences were determined by Illumina HiSeq1000 101-nucleotides paired-end reads which assembled into 4849 and 2774 contigs of at least 500 nucleotides for the L. infantum 263 and L. infantum JPCM5 hybrids, respectively. Sequence reads from each hybrid were aligned to the reference genome L. infantum JPCM5 [2] available at TriTrypDB (version 4.0) [27] using the software bwa (bwa aln, version 0.5.9) with default parameters [28]. The maximum number of mismatches was 4, the seed length was 32 and 2 mismatches were allowed within the seed. The detection of single nucleotide polymorphisms (SNPs) was performed using samtools (version 0.1.18), bcftools (distributed with samtools) and vcfutils.pl (distributed with samtools) [29], with a minimum of three reads to call a potential variation prior to further analysis. The sequence data are available at the EMBL European Nucleotide Archive (http://www.ebi.ac.uk/ena), accession number ERP001431 (samples ERS138995 and ERS138996 corresponding to the L. infantum 263 hybrid transfected with gDNA derived from LmMF80.3DLmjF13.1540::NEO/LmjF13.1540 and the L. infantum JPCM5 hybrid transfected with LmDLRP::NEO/ LRP gDNA, respectively). Several python (version 2.4.3) scripts and bash (version 3.2) scripts were created to further analyze the data. The quality assessment software samstat (v1.08) was used to generate quality reports [30].

Donor line generations
Our work was carried out with L. major and L. infantum, two species respectively responsible for cutaneous and visceral leishmaniasis that have an estimated 20-100 million years of

Author Summary
Leishmania spp. are pathogenic protozoa characterized by a substantial diversity in pathogenesis and virulence despite their considerable synteny at the genome level. The existence of genetic exchange was recently proven experimentally in the sand fly vector where hybrid parasites were isolated and generated. Here, we show the feasibility of generating Leishmania hybrids by electroporating genomic DNA of donor cells into recipient Leishmania parasites. This methodology was made possible by introducing a drug resistance marker in the donor Leishmania cells that could be used for selecting recombinant recipient parasites. Integrations of exogenous DNA fragments as large as 45 kb were possible for several chromosomal regions and took place at homologous loci in recipient Leishmania strains. Our observations are the first step for the generation of in vitro hybrids for assessing gene function under natural genomic contexts and this technology may be applicable to other pathogens. divergence [31,32]. Our strategy for selecting hybrid parasites was to introduce selectable markers (usually the NEO gene) into specific L. major genes, to electroporate total gDNA extracted from these recombinant parasites into recipient cells and to recover recombinant hybrids thriving under selection pressure. The first locus that was studied encodes for a leucine rich repeat protein (LRP) on chromosome 34 (LmjF34.0550) that we showed to be involved in resistance to antimonials [33], the chemotherapeutic mainstay against Leishmania. We generated mutants haploid for LRP by inserting NEO cassettes made to target either the LRP gene of L. major Friedlin (LmjF34.0550) or L. infantum strain 263 (LinJ34_V3.0570) (Figure 1), giving rise to the LmDLRP::NEO/ LRP and Li263DLRP::NEO/LRP haploid lines, respectively ( Figure 1B, 1D, lanes 2). We also successfully obtained LRP null mutants for both species (LmDLRP::NEO/NEO and Li263DLRP::-NEO/NEO) by loss of heterozygosity [34] after one round of allelic inactivation and higher drug selection ( Figure 1B, 1D, lanes 3). Consistent with the role of LRP in antimonial resistance, the LmDLRP::NEO/NEO and Li263DLRP::NEO/NEO mutants were more sensitive to antimonials (Figure 1 and data not shown). We also generated haploid mutants for genes present on chromosome 1 (LmjF01.0315), 5 (LmjF05.0610), 13 (LmjF13. 1540) and 30 (LmjF30.1250) of L. major Friedlin. With the exception of the L. major LmDPK::NEO/PK mutant haploid for LmjF30.1250 (pyridoxal kinase gene) that has already been published [19], all other single NEO disruptions were generated for this study and molecular evidence for the haploid mutants can be found in Figure S1.

Generation of parasite hybrids
Whole genome transformation (WGT) has proven useful when working with naturally transformable microorganisms [35,36] and electroporation was tested for rendering Leishmania cells competent to receive total gDNAs from other Leishmania strains. We first electroporated total gDNA derived from L. infantum Li263DLRP::-NEO/LRP into wild-type (WT) L. infantum 263 recipients cells and recovered recombinant clones resistant to paromomycin, a drug for which NEO bestows resistance. The NEO marker integrated at the appropriate locus in most clones (Table 1), as determined by Southern blot analysis of digested DNA (data not shown) but also of chromosome sized blots since a NEO probe hybridized to chromosome 34 in recombinant clones but not in the initial L. infantum 263 WT cells ( Figure 2A, lanes 1 and 2). Similar results were obtained for L. major Friedlin, where electroporation of gDNA derived from L. major LmDLRP::NEO/LRP single disruptant into L. major Friedlin WT parasites led to transfectants growing in the presence of paromomycin (data not shown). Again, the NEO gene integrated properly at chromosome 34 in the recovered recombinants, as determined by Southern blot analysis of chromosome sized blots (data not shown). We also succeeded in electroporating total gDNA derived from L. infantum Li263DLRP::-NEO/LRP into L. infantum JPCM5 WT recipients and isolated recombinants which have integrated the NEO marker at the level of chromosome 34 ( Figure 2A, compare lanes 5 and 6). The latter results indicated that gDNAs electroporated in Leishmania can integrate at homologous loci, at least in the same strains or species.
We then tested whether we could cross the species barrier and electroporated total gDNA derived from L. major LmDLRP::NEO/ LRP into either L. infantum 263 or JPCM5 WT cells. This indeed seemed to have occurred since a NEO marker was found to have integrated at the proper chromosome in both L. infantum strains (Figure 2A, lanes 3 and 7). This is significant as the construct used for inactivating the targeted allele is species-specific but the use of gDNAs allowed crossing the species barrier. The integration of the NEO marker into the homologous locus did not occur in all transfectant clones however, as the smear obtained in PFGE blots for one L. infantum 263 transfectant suggested a circularization of a NEO containing DNA fragments ( Figure 2A, lane 4). This was indeed confirmed by the recovery of circular episomes from this transfectant by alkaline lysis (data not shown). In most transformation experiments the majority of clones had donor DNA integrated but episomal DNA was also observed in other clones (Table 1). While L. major and L. infantum are highly syntenic [2], there are nonetheless differences between homologous genes at the nucleotide level (mean 90% genome identity) and we capitalized on these natural single nucleotide polymorphisms (SNPs) to measure the extent of exchanged DNA between both species. By targeted sequencing of genomic regions located upstream and downstream of the NEO gene in one hybrid clone of L. infantum 263 and two hybrid clones of L. infantum JPCM5, we could infer that DNA fragments ranging from 20 kb to 40 kb were exchanged at the LRP locus between L. major and L. infantum ( Figure 2B). For example, when we sequenced the LmjF34.0540 gene (LinJ34_V3.0560) in the 263 hybrid clone 1 we found an hybrid sequence with one allele with L. major and one allele with L. infantum sequences (Table S3). Sequencing of PCR fragments was done similarly with genes upstream and downstream of LinJ34_V3.0560 to find the first genes (LinJ34_V3.0530 and LinJ34_V3.0630) upstream and downstream NEO with exclusively L. infantum sequences. The genes in between were hybrids between L. infantum and L. major (Table S3). In this specific hybrid transfectant, the contiguous DNA transferred was estimated to be at least 40 kb. We used a similar strategy of multilocus sequencing to characterize the extent of DNA exchange in the hybrids JPCM5 clone 1 (Table S4) and clone 2 (Table S5). Interestingly, both L. infantum alleles were replaced by the L. major sequences in the JPCM5 clone 1 hybrid parasite, most probably by loss of heterozigocity ( Figure 2B and Table S4). We also performed paired-ends next generation sequencing of the whole genome of the L. infantum JPCM5 hybrid clone 1. Sequence analysis revealed a single stretch of 1055 SNPs (381 within coding sequences) derived from L. major that spanned positions 212,950 to 234,918 (48 SNPs/kb). In contrast, only 131 SNPs were detected for the rest of the 1.8 Mb sequence of chromosome 34 upstream and downstream of the integrated L. major DNA fragment (0.07 SNP/ kb). Most importantly, analysis of the sequenced genome revealed that no other genomic fragments derived from L. major Friedlin were co-transferred elsewhere in the genome ( Figure 2B, hybrid JPCM5 (1)). Thus, targeted sequencing confirmed that we could generate L. infantum hybrid parasites containing up to 40 kb of contiguous L. major Friedlin sequence and whole genome sequencing confirmed this and also informed that no other DNAs inserted elsewhere in the genome. The integration was stable and maintained in the absence of selective pressure (not shown). Importantly, hybrids acquired the phenotype of the donor cells since every L. infantum 263 or L. infantum JPCM5 hybrid clones with a LRP/NEO integrated DNA had an increased sensitivity to SbIII ( Figure 1F and data not shown).
The ability to select for the integration of NEO-marked gDNA from L. major into L. infantum recipients was not unique to the LRP locus. We recently inactivated one allele of the pyridoxal kinase gene in L. major (LmjF30.1250) [19] and the electroporation of gDNA derived from this haploid mutant (LmDPK::NEO/PK) into L. infantum 263 and JPCM5 WT cells allowed recovering hybrid transfectants thriving under paromomycin pressure. Hybridization of chromosome sized blots confirmed the proper integration at the level of chromosome 30 ( Figure 3A) and targeted sequencing of individual genes upstream and downstream of the NEO marker in both L. infantum hybrid strains (Tables S6 and S7) estimated that DNA fragments of 12 and 18 kb derived from L. major Friedlin had been exchanged at the PK locus of both L. infantum strains ( Figure 3D). In the L. infantum 263 recipient, we observed both integration at the level of the chromosomal locus and a circular amplicon ( Figure 3A, lane 2).
To test the generality of hybrid formation, we generated haploid L. major Friedlin mutants for the LmjF01.0315 (LmDLmjF01.0315:: NEO/LmjF01.0315) or LmjF05.0610 (LmDLmjF05.0610::NEO/ LmjF05.0610) genes ( Figure S1), both coding for proteins of unknown function. Electroporating L. infantum 263 WT cells with gDNA derived from LmD01.0315::NEO/LmjF01.0315 yielded paromomycin-resistant hybrids that had properly integrated the NEO gene at the level of chromosome 1 ( Figure 3B, lane 4). Using our targeted DNA sequencing approach of individual genes around the NEO marker (Table S8), we could estimate that a 40 kb contiguous L. major DNA fragment replaced the L. infantum sequence in one allele ( Figure 3E). Two bands were observed when hybridizing with the probe specific to chromosome 1 ( Figure 3B), but this is probably related to the size polymorphism already reported for this chromosome [37]. It is hence probable that the NEO marker integrated in the chromosome with higher molecular weight ( Figure 3B, lane 4). The integration of the NEO marker did not occur in the chromosome 1 of L. infantum JPCM5 however, as the smear obtained in PFGE blots for this transfectant indicated a circularization of the NEO cassette ( Figure 3B, lane 2). However, the same L. infantum cells transfected with gDNA from LmDLmjF 05.0610::NEO/LmjF05.0610 failed to lead to any transfectants ( Figure 4F). This suggested that hybrid formation differed between genomic loci targeted by WGT.

Improving recipient cells
The efficiency of generation of hybrids containing up to 40 kb of sequences from another species was found to be low. Indeed, while the efficiency of transfection for NEO disruption PCR constructs targeting single genes was calculated to be 1610 27 , this was decreased about 10 fold when transfecting whole gDNA ( Table 1). The lack of integration observed for some gDNAs, for example the gDNA derived from L. major LmDLmjF05.0610::NEO/LmjF05.0610 into L. infantum ( Figure 4F), could be due to several reasons but it could possibly occur if recombination between non identical DNAs was made more permissive. We first assessed whether overexpressing the RAD-51 recombinase gene (LinJ28_V3.0580) in L. infantum parasites [38] could increase recombination efficiency and lead more easily to hybrids but this has not been the case (results not shown). We then assessed whether the generation of hybrids with heterologous DNA would be more efficient in cells impaired for the mismatch repair (MMR) machinery since it was showed to prevent recombination between divergent sequences [39,40]. In the related trypanosomatid parasite Trypanosoma brucei, the inactivation of the MMR gene MSH2 increased the efficiency of recombination between mismatched DNAs [41]. L. infantum has a single MSH2 gene (LinJ33_V3.0420) and we generated a L. infantum 263 MSH2 null-mutant (Li263DMSH2::HYG/BLA) by two successive rounds of allelic inactivation using HYG and BLA inactivation cassettes ( Figure 4A) conferring resistance to hygromycin and blasticidin, respectively. The inactivation of MSH2 in Li263DMSH2::HYG/BLA was confirmed by Southern blot analysis of restricted DNA and by PCR using specific MSH2 primers ( Figure 4B). A MSH2 chromosomal null mutant complemented in trans for MSH2 was also generated by transfecting Li263DMSH2::HYG/BLA with the rescue plasmid pSP72a-PURa-MSH2 (data not shown).
Impairing with the MMR machinery is usually associated with an increased tolerance to N-methyl-N9-nitro-N-nitrosoniguanidine (MNNG), an alkylating agent that interferes with the proper  replication of DNA by methylating the O 6 position of guanine. While the L. infantum MSH2-deficient line had no significant growth difference in comparison to wild-type cells, we observed a small but significant increase in resistance to MNNG ( Figure S2). Interestingly, the MSH2 null mutant was more proficient in recombination for small heterologous linear DNA fragments ( Figure 4D) but not for those using homologous fragments ( Figure 4C). Most importantly, the inactivation of MSH2 in L. infantum also allowed recovering a higher number of recombinants following electroporation with cross-species gDNA derived from L. major LmDLRP::NEO/LRP ( Figure 4E). By sequencing PCR fragments for genes upstream and downstream of NEO (Tables S9  and S10), we could determine that fragments of 35-45 kb were transferred in two independent clones of the L. infantum MSH2 null mutant hybrids ( Figure 3F). Transfection in the MSH2 null mutant also allowed recovering recombinants when transfecting with gDNAs that otherwise did not led to hybrids with WT recipients. Indeed, while we could never recover paromomycin-resistant    (Figure S1), the transformation of the L. infantum MSH2 null mutants consistently yielded recombinant parasites ( Figure 4F). The hybridization of chromosome sized blots confirmed that the NEO gene integrated in the proper chromosome for every aforementioned WGTs ( Figure 3C).
Application of whole genome transformation to study the role of a point mutation in drug resistance The described procedure could prove very useful for applications regarding the role of genomic loci or SNPs in conferring a particular phenotypic trait like virulence or drug resistance. In Leishmania, the miltefosine transporter (MT) is a phospholipid flippase located at the plasma membrane of the parasite [42]. Point mutations in MT are correlated to resistance to MF [19,42], a drug used for the treatment of antimonial-resistant infections in endemic regions [1]. We previously showed that the L. major  (Table S11) estimated the size of the exchanged DNA to approximately 25 kb ( Figure 5B). Furthermore, degenerate primers allowing the amplification of MT from both L. major (LmjF13.1530) and L. infantum (LinJ13_V3.1590) were used to amplify the MT locus in one representative L. infantum hybrid and the cloning of these amplified MT fragments into the pGEM-T-easy plasmid revealed a allele frequency of 60/40% for L. infantum WT/LmjF-MF80.3 among E. coli clones ( Figure 5C). The L. infantum 263 hybrid thus integrated the M547del mutation from LmjF-MF80.3 on one of its MT allele while maintaining the other allele unaltered. This was confirmed by paired-ends next generation sequencing of the whole genome of this hybrid parasite, which further revealed that no other genomic fragment from LmMF80.3DLmjF13.1540::NEO/LmjF13.1540 integrated elsewhere in the genome. Whole genome sequencing revealed a single stretch of 29,2 kb with 541 SNPs (301 within coding sequences) derived from L. major that spanned positions 596,277 to 625,504 (19 SNPs/kb) on one allele of chromosome 13 that were transferred to L. infantum 263. In contrast only 35 SNPs were detected for the rest of the 645 kb sequence of chromosome 13 upstream and downstream the integrated L. major DNA fragment (0.05 SNP/kb).
Most importantly, and as a proof-of-principle of our facile strategy to introduce knock-ins in Leishmania, an increased resistance to MF was specifically observed for the L. infantum hybrid that acquired the M547del from LmjF-MF80.3 at their MT locus ( Figure 5D).

Discussion
Homologous gene targeting is a powerful reverse genetic approach allowing to test the functions of specific gene products in trypanosomatid parasites like Leishmania [14]. However, available tools only allow performing studies at the level of genes or small DNA fragments and do not allow investigating the role of large genomic loci or of easy assessment of specific point mutations in a natural genomic context. The most critical parameters for successful homologous recombination in Leishmania are the degree of homology and the length of homologous sequences between the donor and recipient DNAs [15]. While gene content and synteny is highly conserved between species of Leishmania, the high degree of variations observed at the level of nucleotide sequences [2,3] usually preclude the use of unique DNA constructs for targeting homologous loci between species (Figures 4D, 4F). Genetic exchange among natural populations of Leishmania has long been suspected [10,43] and recently received experimental confirmation [11], with some hybrids even reported to acquire increased fitness or transmission potential [44]. In rare cases are these hybrids crossing the species barrier but natural hybrids between L. major and L. infantum have been described [10].
In this study, we describe a protocol based on WGT that enables the transfer of large DNA fragments between strains and species of Leishmania. Integrations occurred at different genomic loci on distinct chromosomes in recipient cells and up to 45 kb of heterologous gDNA was exchanged between species. Such size limitation could be due to several factors like breakage of the high molecular weight DNA during transformation or structural variations in the genome of recipient cells that would disrupt the progress of recombination. This contrasts with the genome-wide heterozygosity that seems to be happening for hybrids generated in the sand fly vector either in a natural context [10] or in experimental settings [11][12][13]. Notwithstanding, genetic crosses performed in the sand fly vector normally only delimits phenotypic traits to loci covering tens of genes and our approach should thus reveal a valuable complement for narrowing down the list of candidates. Targeted sequencing analyses indicated that recombination events took place between and within orthologous genes in hybrid recombinants, which is consistent with the extensive synteny of Leishmania genomes [2]. Interestingly, we noticed that in the two events where the whole genomes of hybrids were sequenced those recombination events occurred in regions of highest homology (recombination sites were in regions of 95% identities while the average region was 90% identical).
The presence of a selection marker on either one or two alleles in the donor gDNAs did not affect the efficiency of transformation (results not shown) but the rates of targeting were different depending on the donor and recipient strains of Leishmania. Indeed, our results are consistent with the degree of divergence at the nucleotide level between Leishmania species [2,45] since we were unable to obtain L. major or L. infantum hybrids when using donor gDNAs derived from two independent L. (V.) braziliensis lines having a NEO marker integrated at distinct genomic locations,(result not shown). This correlates with the recombination events described in natural populations of Leishmania, which mainly implicates closely related species like L. major/L. infantum [10], L. panamensis/L. braziliensis or L. panamensis/L. guyanensis [46][47][48]. We focused our transformation experiments with L. infantum as the recipient cell of heterologous DNA. We succeeded once in creating a L. major hybrid with L. infantum DNA (results not shown) but this was more difficult and transformation of other L. infantum gDNA donors in L. infantum never led to L. major transformants being either integrated or episomal (data not shown). This may relate to the ten-fold lower efficiency of transformation of L. major [49,50] and one might thus be successful at obtaining L. major recombinant hybrids by improving transfection efficiencies.
While we were able to recover recombinants for most of the genomic loci tested, some chromosomal locations were nonetheless targeted less efficiently. The MMR machinery plays a critical role in maintaining genetic stability by correcting for base mismatches that can arise through replication errors or chemical damage [51] and also influences the frequency of homologous recombination between divergent sequences [41,52]. Interestingly, interfering with the MMR machinery in Leishmania increased the number of hybrid clones for these loci less amenable to hybrid formation, probably by dimming the barriers for recombination between mismatched DNA (Figure 4). We also found a background of transfectants maintaining the selection marker as part of extrachromosomal amplicons for most loci (Figure 2A, lane 4; Figure 3B, lane 2, Table 1). The hybridization of chromosome-sized blots from several independent hybrid clones indicated that the relative abundance of chromosomal targeting and episomal maintenance of the selection marker varied depending on the donor gDNAs but in the majority of clones, the DNA was integrated (Table 1). Extrachromosomal circles can be generated by homologous recombination between repeated genomic sequences in Leishmania [53,54]. It is thus likely that these circles were generated by recombination between repeated sequences present on large DNA fragments including the NEO marker. Extrachromosomal circles were not stable and were lost in the absence of selective pressure (not shown).
Whole genome transformation in naturally competent bacteria was shown to lead to the acquisition of several distinct donor DNA segments that optimally replace up to 3% of the genome of recipient cells [35]. This is in contrast to the unique integration events observed in the genome of Leishmania hybrids ( Figure 2B, Figure 5B), for which homologous recombination were restricted to genomic loci surrounding the selection marker as shown by whole genome sequencing. It may be possible to further increase the efficiency of recombination by manipulating the expression of recombination enzymes and more loci could be targeted by the use of additional selection markers in the same cell. On the other hand, this controlled recombination prevents the likelihood of phenotypic artefacts due to surrogate DNA exchange events. The method presented here is now allowing the in vitro generation of hybrid parasites allowing for testing for gene functions in a natural genomic context. This technique of hybrid formation has also the potential to be useful for other microbial pathogens.     (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid 263 (1) of Figure 2 are listed by their respective position in the gene. (DOC) Table S4 Loci analyzed by multilocus sequencing typing genes. Primers forward and reverse were used for both DNA amplification and sequencing (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid JPCM5 (1) of Figure 2 are listed by their respective position in the gene. (DOC) Table S5 Loci analyzed by multilocus sequencing typing genes. Primers forward and reverse were used for both DNA amplification and sequencing (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid JPCM5 (2) of   (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid L. infantum 263 of Figure 3D are listed by their respective position in the gene. (DOC) Table S7 Loci analyzed by multilocus sequencing typing genes. Primers forward and reverse were used for both DNA amplification and sequencing (Table S2). The natural polymor-phisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid L. infantum JPCM5 of Figure 3D are listed by their respective position in the gene. (DOC )   Table S8 Loci analyzed by multilocus sequencing typing genes. Primers forward and reverse were used for both DNA amplification and sequencing (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid L. infantum 263 of Figure 3E are listed by their respective position in the gene. (DOC) Table S9 Loci analyzed by multilocus sequencing typing genes. Primers forward and reverse were used for both DNA amplification and sequencing (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid 263 (2) of Figure 3F are listed by their respective position in the gene.

(DOC)
Table S10 Loci analyzed by multilocus sequencing typing genes. Primers forward and reverse were used for both DNA amplification and sequencing (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid 263 (3) of Figure 3F are listed by their respective position in the gene. (DOC) Table S11 Loci analyzed by multilocus sequencing typing genes. Primers forward and reverse were used for both DNA amplification and sequencing (Table S2). The natural polymorphisms between L. major and L. infantum enabled mapping the size of exchanged DNA by sequencing. The SNPs found in the hybrid L. infantum 263 of Figure 5 are listed by their respective position in the gene. (DOC)