Genomic instability and mono-parental expression mitigate genomic shock in a cross-subgenus Leishmania hybrid

Abstract

Hybridization, the merging of distinct genomes, is increasingly recognized as a major evolutionary force among eukaryotic pathogens, including facultatively sexual protist parasites like Leishmania and Trypanosoma. While it may contribute to pathogen virulence and drug resistance, hybridization remains poorly characterized, particularly how genetic distance between parental cells influences genomic compatibility and which compensatory mechanisms ensure hybrid viability. Here, we report the in vitro generation of an unusual sexual hybrid between Leishmania species infecting mammals (L. infantum) and reptiles (the Sauroleishmania L. tarentolae), and used this unique genetic model system to address these open questions. Our data provide evidence of genomic compatibility between even highly divergent Leishmania species, offering new insights into the evolutionary potential of Leishmania and related pathogens. We demonstrate that the genomic shock caused by the fusion of distinct genomes can be mitigated by two key mechanisms: (i) at the genomic level, chromosome loss allows the establishment of mosaic aneuploidy in the newly formed hybrid, and (ii) at the post-transcriptional level, preferential mono-parental allelic expression acts as a secondary compensatory mechanism. Our findings establish genome instability and post-transcriptional regulation as central processes in Leishmania hybridization, which may be of broad relevance to other biological systems undergoing genetic exchange.

Graphical Abstract

Graphical Abstract.

Introduction

Unicellular eukaryotes can reproduce either asexually through clonal reproduction, giving rise to a genetically identical progeny [1, 2], or sexually, with two parental genomes being combined by a cellular hybridization process to create a new one. These forms of reproduction are not always exclusive, as observed in protists and fungi [3–6]. This applies to parasites of the genus Leishmania, the causative agents of leishmaniases [7]. These vector-borne diseases are worldwide distributed in tropical and sub-tropical regions and present a range of clinical manifestations, from self-limiting cutaneous lesions to visceral complications leading to death if untreated [8]. At least 20 different Leishmania species have been identified, infecting a wide range of hosts and exhibiting different tissue tropisms and clinical outcomes [9].

Leishmania parasites have a diphasic life cycle, with extracellular promastigotes multiplying asexually within the digestive tract of their insect hosts, the phlebotomine sand flies, and intracellular amastigotes that multiply asexually within the macrophages of the vertebrate hosts [10]. Clonal reproduction was for long regarded as the only reproductive strategy of Leishmania. However, it is now clear that they also possess a cryptic, sexual reproductive cycle leading to the production of hybrids, as first hypothesized by the observation of field isolates [11–18] and later demonstrated through experimental hybridization studies [19]. This type of facultative sex creates new genotypes with unpredictable phenotypes, which may lead to changes in tissue tropism, virulence, or the emergence of drug resistance [20]. This is the case of L. infantum/L. donovani hybrids, recently identified as the cause of re-emerging human leishmaniasis in the North of Italy [21]. The possibility of selfing—a form of sexual reproduction where mating occurs between parasites within the same population—has recently been demonstrated in Leishmania spp., further challenging the notion of its clonal propagation [22].

Leishmania hybridization follows a meiosis-like process, as shown by whole-genome sequencing of experimental hybrids from successive generations [23] that involves some of the classical actors of meiosis, such as the fusogen Hap2 and Hop1, a protein participating in the formation of the synaptonemal complex [24–26]. This process takes place inside the sand fly gut [19]; however, the underlying mechanisms remain poorly understood. In particular, the specific cell types involved in mating remain elusive, and the evolutionary consequences of hybridization on parasite adaptation have not been elucidated. Both in experimental settings and in nature, hybridization events occur within and between species, suggesting the absence of a strict mating barrier among Leishmania species [20]. Nevertheless, the influence of genetic distance between parental strains on hybrid formation and fitness, as well as the compensatory molecular mechanisms allowing the hybrid survival had yet to be investigated and are the focus of our study.

Different strains and/or species of Leishmania frequently circulate in the same area, and co-infections have been recorded in different hosts, creating the possibility of hybrid generation [27]. This is the case for L. tarentolae and L. infantum, two genetically distant species circulating in the Mediterranean basin. The saurian-associated L. tarentolae, which infects reptiles and is transmitted by Sergentomyia minuta sand flies [28–30], is non-pathogenic to mammals, whereas L. infantum is transmitted by various Phlebotomus sub-species and causes visceral leishmaniasis in humans and dogs, its main reservoir [31, 32]. While the co-occurrence of both species inside the same sand fly, and thus the potential for hybrid formation, is possible, as supported by the detection of human and dog blood in engorged S. minuta females [33], no such extreme, inter-species—and even inter-subgenus—hybrid has been so far identified in nature. This raises the question if such hybrids just escaped detection due to low frequency and limited sampling, or whether the genetic distance between the possible parents precludes the formation of viable hybrids through a pronounced “genomic shock,” defined as the genome-wide disruption of genic and epigenetic regulations resulting from the sudden merging of divergent genomes [34].

In this study, we addressed this important open question and tested whether L. tarentolae and L. infantum were able to hybridize using an in vitro hybridization protocol [24, 35]. We provide phenotypic and genomic evidence for the successful creation of a viable hybrid strain, demonstrating genetic compatibility between these two highly divergent Leishmania species. We used this unique genetic model strain to investigate the mechanisms involved in the compensation of the genomic shock associated with Leishmania hybridization. The hybrid exhibits important genomic restructuring (in particular chromosome loss) and parent-specific allelic expression likely regulated at post-transcriptional level, revealing both changes in gene dosage and messenger RNA (mRNA) turnover as key adaptive mechanisms. These findings demonstrate that Leishmania employ both genomic plasticity and post-transcriptional regulation during and after hybridization as adaptive strategies to mitigate the genomic shock resulting from the fusion of two parental genomes; mechanisms that may likewise operate in other organisms undergoing facultative sexual cycles or alternative forms of genetic exchange.

Materials and methods

Parasite strains and in vitro hybridization

The following parasite strains were used: L. infantum LLM320 (MHOM/ES/92/LLM-320; isoenzyme typed MON-1 [36], L. infantum LEM 2259 (MHOM/FR/91/LEM2259, isoenzyme typed MON-1, gift from the International Leishmania Cryobank and Identification Center in Montpellier, France), and L. tarentolae R011, isolated from a gecko (see [30]). The three strains were electroporated using an Amaxa 2D nucleofector (Lonza Bioscience) with integrative expression vectors carrying both a resistance cassette and a fluorescence marker: the pA2-GFP-Neo plasmid, carrying a Neomycin resistance cassette and a GFP fluorescence marker for the L. infantum LEM 2259 strain or the pA2-RFP-Hyg vector, carrying a Hygromycin cassette and an RFP reporter resistance for the L. infantum LLM320 and the L. tarentolae strains [37]. The resulting transgenic parasites, referred in the text as Linf-GFP, Linf-RFP, and Ltar-RFP, were cloned by limiting dilution and one single clone was used for subsequent use. The promastigotes were cultured at 26°C in Ld1S promastigote (pro) medium: M199 (Life Technologies), 10% fetal bovine serum (Biowest), 20 mM HEPES (Sigma–Aldrich), 100 µM adenine (Sigma–Aldrich), 2 mM L-glutamine (Merck), 10 µg/ml folic acid, 13.7 µM hemin (Merck), 4.2 mM NaHCO₃ (Sigma–Aldrich), 1xRPMI1640 vitamins (Sigma–Aldrich), 8 µM 6-biopterin (Sigma–Aldrich), 100 units penicillin, and 100 µg/ml streptomycin (Life Technologies), pH 7.4. For the in vitro generation of hybrids, we used our previously published protocol [24, 35]. Briefly, equal volumes of cultures of two parental lines at day 1 post-inoculation were mixed and distributed into 96-well plates at a total volume of 100 µl per well. After 24 h, each co-culture was transferred to a single well of a 24-well plate containing 900 µl of Ld1S pro medium supplemented with Hygromycin B (Invitrogen, 25 µg/ml) and Geneticin (Neomycin, Sigma–Aldrich, 50 µg/ml). Double-drug resistant parasites were passed in new selective medium and analyzed by Flow Cytometry to confirm their expression of GFP and RFP fluorescent markers. Exposure to X-rays (10 Gy) and to hydrogen peroxide (H₂O₂, 200 µM) was used to increase the frequency of hybrid production, as previously described [24]. Hybrid lines were cloned by limiting dilution, and a single clone from each hybrid line was used for genome and transcriptome sequencing. The daily concentration of promastigotes and the corresponding growth curves were determined by counting a 1/100 culture dilution from three independent experiments.

Flow cytometry

A CytoFlex Cytometer (Beckman Coulter) was used to observe the parasites fluorescence and to determine their ploidy. The data were analyzed with the CytExpert software v2.4.0.28. The ploidy of hybrid parasites was evaluated using propidium iodide staining, using the diploid parental lines as normalization controls, as previously described [24, 35, 36, 38].

Morphological analysis of the parasites

Bright-field images of Linf-GFP, Ltar-RFP, and Linf/Ltar hybrid parasites were captured following fixation with 4% paraformaldehyde (PFA) using an EVOS M5000 imaging system. Body length, body width, and flagellum length were measured for 100 parasites from each strain using Fiji software. The Mann–Whitney Student’s t-test was applied for statistical analysis.

Mouse BMDM preparation and infection

Murine bone marrow-derived macrophages (BMDMs) were prepared from the tibias and femurs of three mice as previously described [39]. Briefly, bone marrow cells were first plated at a density of 3 × 10⁷ cells/12 ml in Dulbecco's Modified Eagle Medium (DMEM) medium (Pan-Biotech) already containing 4.5 g/l glucose and 110 mg/l sodium pyruvate, plus 2 mM L-glutamine (Merck), 3.7 g/l NaHCO₃ (Pan Biotech), supplemented with 10% heat-inactivated FBS (Biowest), 50 µg/ml streptomycin and 50 IU/ml penicillin (Life technologies), 50 mM 2-β-mercaptoethanol (Sigma–Aldrich), 10 mM HEPES, and 50 ng/ml of recombinant mouse colony-stimulating factor 1 (rmCSF-1, ImmunoTools), and incubated at 37°C, 7.5% CO₂ in tissue culture treated dishes (Falcon 100 mm TC-treated 353003). After overnight incubation, the medium containing the non-adherent cells was collected, diluted in fresh medium containing 50 ng/ml of rmCSF-1 at a concentration of 1 × 10⁶ cells/ml, and transferred to hydrophobic Petri dishes (Greiner Bio-One 664161). After six days of culture, the medium was removed, and adherent cells were incubated with pre-warmed PBS (pH 7.4) containing 25 mM EDTA (Sigma–Aldrich) for 30 min at 37°C, 7.5% CO₂. BMDMs were detached by gentle flushing, collected, and resuspended in complete medium supplemented with 30 ng/ml of rmCSF-1 at a concentration of 2.5 × 10⁵ cells/ml. BMDMs were then plated at 2.5 × 10⁴ cells/well in 96-well plates (PhenoPlate^™-96, Revvity) and incubated overnight before infection. The BMDMs were then incubated with 20 parasites/BMDM for each of the different parasite strains or with three zymosan particles/BMDM (Invitrogen^™ Zymosan A S. cerevisiae BioParticles^™, Alexa Fluor^™ 488 conjugate) as a control for phagocytosis control. The parasites used for infections were prepared by differential centrifugation: promastigote cultures in the late stationary phase (day 5) were centrifuged at 1000 × g for 5 min to collect the supernatant containing metacyclic forms. The supernatant was further centrifuged at 3000 × g for 10 min, parasites were counted, and the culture volume was adjusted to reach a parasite-to-cell ratio of 20:1. Phagocytic and infection assays in murine BMDMs were performed in triplicates with the Linf-GFP, the Ltar-RFP, and the Linf/Ltar strains at 4, 24, 48, and 144 h post-infection. The proportion of infected cells and parasitic load (number of parasites per infected cell) at the different time points were determined using the OPERA Phenix Plus high content imaging system (Revvity) after labeling the cells and parasites. Images from nine fields/well were acquired per replicate and a mean of 2300 cells/well was counted. Image analysis was performed using a dedicated script on the Signals Image Artist (SImA) application. The percentage of infected cells, the number of parasites/infected cells, and the subsequent number of parasites per 100 cells were determined for each time point and each mouse. The results represent the mean values obtained per time point post-infection from the three independent mice.

Immunofluorescence on infected BMDMs

Immunofluorescence was performed on infected murine BMDMs fixed in 4% PFA (Sigma–Aldrich) in a 96-well plate for at least 30 min, followed by a 15-min incubation with 50 mM NH₄Cl in PBS. A blocking step was performed with 5% goat serum in PBS-saponin (0.1 mg/ml) for 30 min. The cells were then incubated with a rat IgG anti-LAMP 1 (0.5 mg/ml from eBioscience) diluted at 1:100 in PBS–saponin–gelatin (PBS 1× + 0.1 mg/ml saponin + 0.25% gelatin) for 45 min, washed three times in PBS–saponin, and incubated with a goat anti-rat IgG Fab’2-Alexa 647 (Jackson ImmunoReseach) for 45 min. The cells were washed again three times with PBS–saponin, followed by staining with the nuclear dye Hoechst 33 342 (Sigma–Aldrich) at a final concentration of 5 µg/ml for 15 min. All incubation and washing steps were performed at room temperature. After two final PBS washes, PBS was added to the wells (200 µl/well), and the plates were stored at 4°C until image acquisition.

Dog PBMC preparation and infection

10 ml of blood/dog collected in heparin-lithium tubes from three healthy Beagles were purchased from Marshal BioResources, France. Peripheral blood mononuclear cells (PBMCs) were obtained following the procedure described by [40] with modifications. Blood samples were mixed with DMEM medium at a 1:2 ratio (9 ml blood + 18 ml DMEM = 27 ml). The mixture was then layered on an equal volume (27 ml) of PANCOLL (1.077 g/ml, Dutscher) and centrifuged at 700 × g for 30 min at room temperature to separate the mononuclear cells. The cells of interest, found at the interface of the two layers, were transferred in a new tube, washed twice with 15 ml of warm DMEM medium and centrifuged at 300 × g for 10 min at room temperature. The recovered PBMCs were suspended in warm DMEM medium supplemented with FBS, antibiotics (100× Penicillin–Streptomycin) and human recombinant CSF-1 (Proteintech). Cells were seeded at a density of 5 × 10⁵ cells/ml into 24-well plates for 5 days and infected on the fifth day with the three parasite lines at a ratio of ten parasites per cell, or incubated with zymosan, bioparticles derived from yeast surface components recognized by phagocytic cells such as macrophages and used here as a phagocytic control, three particles per cell. Based on zymosan phagocytic tests, about 7% of the PBMCs matured into macrophages after 5 days of culture. The percentage of infected cells and the number of parasites per infected cell were counted daily. All experiments were conducted in technical triplicates for each time point and each dog. At 4 and 24 h PI, microscopical images were taken after fixation with 4% PFA and nuclei staining with Hoechst 33342.

Viral presence assessment

We explicitly screened our datasets for Leishmania RNA viruses using a targeted approach. For that we retrieved a curated set of 90 LRV1 and LRV2 reference sequences, including all sequences reported in two previous studies [41, 42], and mapped both DNA-seq and RNA-seq reads against this reference set. No reads mapped to any LRV1 or LRV2 sequence under these conditions.

Reference genomes

The genomic and transcriptomic sequencing were mapped to the L. infantum and L. tarentolae genomes. For L. infantum, both the genome sequence and annotation were obtained from NCBI (accession GCA_900500625.2), consisting of 36 chromosomes and a total size of 32.8 Mb. For L. tarentolae, the genome sequence and annotation were retrieved from NCBI (accession GCA_009731335.1), consisting of 179 contigs with a total size of 35.4 Mb. To study the hybrid samples, we constructed a hybrid reference by concatenating the genome sequences and annotations of L. tarentolae and L. infantum (https://doi.org/10.5281/zenodo.15608973). While bioinformatics analyses including mapping were performed with the complete L. tarentolae genome, we also established a curated version containing only the 57 contigs assigned to known chromosomes (accounting for 32.2 Mb, 97.3% of the coding genes, and 8703 coding genes) to simplify downstream visualizations (https://doi.org/10.5281/zenodo.15608973).

Whole genome sequencing and analysis

DNA was extracted from parasite cultures in log-phase using the DNA blood and tissue kit from MACHEREY-NAGEL, according to the manufacturer’s instructions. Genomic sequencing data were generated using an Illumina NovaSeq 6000 platform, producing paired-end reads with a length of 150 base pairs (2 × 150 bp). The sequencing data exhibit high quality, with an average Phred quality score of 35. Paired-end reads were aligned to the L. infantum JPCM5 v63 reference genome available on TritrypDB (tritrypdb.org) [43] using the BWA-MEM aligner v0.7.17 with default parameters. Single-nucleotide polymorphisms (SNPs) were determined using the PAINT software suite designed for studying inheritance patterns in aneuploid genomes [44]. PAINT was also used to find and extract the homozygous SNP marker differences between the parental cell lines and to estimate the chromosome copy numbers (somy). Chromosome somies were determined by calculating the normalized median read depth multiplied by 2 (for a diploid genome) using the ConcatenatedPloidyMatrix utility with a 5-kb window size. Genomic regions of multiple sequence repeats and high copy number variation (CNV) were filtered out from the analysis by eliminating positions with coverage levels ≥2-fold and ≤0.5-fold the chromosome mean coverage. In the case of the polyploid hybrids (≥3n), the somy values were divided by 2 and multiplied by the ploidy estimated from the DNA content analysis (PI staining) and the parental contribution profile. Allele frequencies of <0.15, read depth <10 or represented by <25% of reads in either forward or reverse direction were filtered out of the analysis. The allelic inheritance of each homozygous parental SNP in the hybrid progenies was determined using the getParentAllelFrequencies PAINT utility. The parental allele frequencies were formatted to be compatible with Circos software v0.69 [45], and inheritance circos plots were generated with 1 257 650 homozygous marker differences between Linf-GFP and Ltar-RFP, and 109 307 homozygous marker differences between Linf-GFP and Linf-RFP cell lines labeled in blue and red, respectively. To estimate the chromosomal somies of the parental strains, sequencing data from each parent were mapped to their respective reference genomes using the Sequana Mapper pipeline (v1.3.1, https://github.com/sequana/mapper [46]) with default parameters. For the hybrid strain, to assess the contribution of each parent, its genomic data were mapped to the hybrid reference (see reference genome section). Parental contributions were then quantified using the somy-score tool from the Sequana Python library [46]. The parental contribution somies were then extracted for subsequent analysis. The file containing the raw counts was deposited on Zenodo ( https://doi.org/10.5281/zenodo.15608973).

Analysis of the kDNA inheritance

Using bwa mem with default parameters, DNA Illumina reads were mapped to the L. major Fn kinetoplast DNA (kDNA) reference sequence (available at http://leish-esp.cbm.uam.es/) [47]. SAM files were coordinated-sorted and converted into BAM files using samtools (v1.21). A SNP-heavy maxicircle sequence of 1.1 kb was chosen to illustrate parental allelic contribution in the hybrid. The L. major kDNA reference was used to facilitate the identification of a large number of SNPs. For the minicircle analysis, read counts per reference minicircle classes were obtained using samtools idxstats (v1.21). Minicircle presence in each sample was defined using a threshold of >10 mapped reads per class. Parent-specific minicircle classes were identified by comparative presence or absence analysis between the two parental species. Inheritance patterns in the hybrid were evaluated using the same read-count threshold. Set intersections among parents and hybrid were visualized in R using the UpSetR package (v1.4.0).

Transcriptomic analysis

Total RNA was extracted using Nucleospin RNA isolation kit (MACHEREY-NAGEL) from the Linf/Ltar hybrid and its parents from three different biological replicates on the second day of growth culture, according to the manufacturer’s instructions. The quality and concentration of RNA were measured by Bioanalyzer DNA1000 Chips (Agilent, #5067-1504). Transcriptomic sequencing data were generated using an Illumina NextSeq 2000 platform, producing single-end reads with a length of 65 bp. The sequencing run yielded 25–80 million reads per sample, except for one hybrid replicate that yielded 8 million reads. According to a PCA analysis, this replicate behaves like the two others and was kept in the analysis. The RNA-seq analysis was performed using Sequana v0.17.3 [46]. Specifically, we used the RNA-seq pipeline (v0.20.0, https://github.com/sequana/sequana_rnaseq) built on top of Snakemake v6.7.0 [48]. Reads were mapped to the aforementioned hybrid reference using Bowtie2 (v2.4.4) [49]. FeatureCounts v2.0.0 [50] was used to generate the count matrix by assigning reads to features based on the hybrid annotation. The effective number of reads used after read counts was 5–40 million reads per sample with an average of 20 million reads per sample. The file containing the raw and normalized counts was deposited on Zenodo (https://doi.org/10.5281/zenodo.15608973).

Orthogroup analysis

To study the relative contribution of each parental genome to the genome and transcriptome of the Linf/Ltar hybrid at the gene level, it was necessary to identify the gene orthologs between the two genomes. For that, we selected the genes from the annotation and genome reference using Sequana library [46] and converted the genomic sequences to amino acid sequences using Bioconvert v1.1.0 [51]. We then used OrthoFinder v2.5.5 [52] and could assign 16 413 genes (94.4% of the total) into 7 627 orthogroups. We excluded the orthogroups composed of species-specific genes only (e.g. gene duplication on one species, gene loss in the other). Indeed, 94 orthogroups (corresponding to 353 genes) were found to be specific to L. infantum and 76 (corresponding to 342 genes) to L. tarentolae. We also excluded orthogroups where genes are not located on the same chromosome (within a species or between species), removing 140 additional orthogroups. This left 7 317 orthogroups for the final analysis, encompassing 15 189 genes (87.4% of the total). The file containing orthologs was deposited on Zenodo (https://doi.org/10.5281/zenodo.15608973).

Z-score and enrichment calculation

FeatureCounts v2.0.0 [50] was used to count the number of reads mapped to each gene, applied to both genomic and transcriptomic data. Counts were summed across orthogroups. No normalization was performed since DNA and RNA contributions were calculated as the ratio of L. tarentolae reads to the total contribution (L. infantum + L. tarentolae). The distribution of orthogroups in a two-dimensional RNA–DNA contribution plot revealed a clear relationship centered around equal contributions from both parents. To identify outliers from this distribution, we calculated a z-score based on the Mahalanobis distance, which measures the distance of each point from the meanwhile accounting for the covariance structure of the data. This Mahalanobis-based z-score extends the concept of a standard z-score into two dimensions. Unlike a 1D z-score, which indicates the number of standard deviations a point lies from the mean along a single axis, the Mahalanobis distance considers correlations between dimensions and provides a single, non-negative value for the overall distance from the mean. Despite this difference, there is an approximate relationship between the two: for example, a 1D z-score of 1.5 corresponds roughly to a 2D z-score of 2. This relationship allows us to interpret the Mahalanobis-based z-scores in terms of familiar 1D Gaussian statistics. By using this metric, we identified outliers in the RNA–DNA contribution space for further enrichment analysis. Enrichment analysis of outlier orthogroups was performed using the TriTrypDB database [43] with GO enrichment features. Lists of enriched molecular functions, cellular components, and biological processes were retrieved and visualized using custom scripts and Sequana library (e.g. GFF annotation file manipulation). These scripts also included functionality from the Bioservices library v1.12.1 to programmatically access the QuickGO website and generate a graph of GO term relationship [53].

Results

Generation of inter-subgenus L. infantum/L. tarentolae hybrids in vitro

In order to test whether L. infantum and L. tarentolae—two Leishmania species circulating in the same geographic area but belonging to different phylogenetic subgenera—are able to hybridize, we set up in vitro crosses between a GFP+, neomycin-resistant L. infantum strain (Linf-GFP) and either an RFP+, hygromycin-resistant L. tarentolae strain (Ltar-RFP) or an RFP+, hygromycin-resistant L. infantum strain (Linf-RFP), used as a control for in vitro hybridization (Fig. 1A). We conducted six independent crosses for both Linf/Ltar and Linf/Linf, totaling 504 culture wells per cross type. All Linf/Linf crosses produced hybrids (9 total), while only one Linf/Ltar cross yielded a single hybrid, reflecting a minimum frequency of mating-competent cells in the parental cultures ranging from 3.17E-9 to 1.32E-9 for the Linf/Linf and 2.16E-9 to 1.49E-9 for the Linf/Ltar crosses (Table 1). Flow cytometry confirmed the hybrid nature of the double-drug resistant parasites, which simultaneously express the fluorescent markers of both parental strains (Fig. 1B). The three parasite strains exhibit significantly different morphologies in culture: Ltar-RFP promastigotes are smaller in body size compared to Linf-GFP, while the hybrid displays an intermediate morphology with greater variability (Supplementary Fig. S1A and B). Both parental strains present a similar growth profile in culture, attaining stationary growth phase four days after inoculation and reaching a maximum density at 4.7E7 cells/ml. In contrast, the Linf/Ltar hybrid shows a slight growth defect, reaching stationary phase only three days later at a density of only 6.7E6 cells/ml (Supplementary Fig. S1C). We then analyzed the ploidy of the different hybrids by flow cytometry. Both the Linf/Ltar hybrid and Linf/Linf hybrids present a typical DNA content pattern: a large peak of cells in the G₀/G₁ phase and a second, smaller peak of cells in G₂/M phase with a doubled quantity of DNA. Hybrids present a ploidy close to 4n in the G₀/G₁ phase, suggesting a contribution of each parental genome close to 2n (two complete sets of chromosomes) (Fig. 1C and D). Together, these data provide first evidence of experimental F1 hybrid production between L. infantum and L. tarentolae, which seems a very rare event even in vitro. This low frequency could be due to the genetic difference of such distant species causing a strong “genomic shock” [34].

Figure 1.

L. infantum and L. tarentolae can produce a viable hybrid progeny in vitro. (A) Schematic of the experimental strategy used for the generation of hybrids. (B) Validation of the hybrids by flow cytometry on fluorescent markers from the parental lines. (C and D) Observed ploidy of the Linf/Ltar hybrid (C) and one representative Linf/Linf hybrid (D) relative to their parental strains.

Table 1.

Summary of the in vitro cross series performed.

		Linf-GFP × Ltar-RFP					Linf-GFP × Linf-RFP
		Initial # of parasites/well		Proportion of positive wells*	Min. frequency of mating-competent cells		Initial # of parasites/well		Proportion of positive wells*	Min. frequency of mating-competent cells
Cross	Condition	Linf-GFP	Ltar-RFP		Linf-GFP	Ltar-RFP	Linf-GFP	Linf-RFP		Linf-GFP	Linf-RFP
1	Untreated	8.0E + 06	5.5E + 06	1/84	1.49E-09	2.16E-09	8.0E + 06	7.5E + 06	2/84	2.98E-09	3.17E-09
2	Untreated	8.0E + 06	5.5E + 06	0/84	–	–	8.0E + 06	7.5E + 06	2/84	2.98E-09	3.17E-09
3	H2O2 (200 µM)	9.0E + 06	6.5E + 06	0/84	–	–	9.0E + 06	9.0E + 06	1/84	1.32E-09	1.32E-09
4	H2O2 (200 µM)	9.0E + 06	6.5E + 06	0/84	–	–	9.0E + 06	9.0E + 06	1/84	1.32E-09	1.32E-09
5	X-rays (10 Gy)	8.0E + 06	5.0E + 06	0/84	–	–	8.0E + 06	8.5E + 06	1/84	1.49E-09	1.40E-09
6	X-rays (10 Gy)	8.0E + 06	5.0E + 06	0/84	–	–	8.0E + 06	8.5E + 06	2/84	2.98E-09	2.80E-09
			Total0.19% (1/504)					Total1.78% (9/504)

*wells where the growth of GFP + RFP + hybrid parasites was observed

Phenotypic characterization of the Linf/Ltar hybrid and its parental strains

To evaluate the infectivity of the Linf/Ltar hybrid and both its parents in mammalian macrophages, we next compared the phagocytic uptake of the three parasite strains by murine bone marrow derived macrophages (BMDMs) and dog PBMCs after differentiation into macrophages (Fig. 2). In the murine BMDMs, we observed a similar dynamic of infection for the three strains, with a decrease over time observed for the proportion of infected cells and the number of parasites/100 cells, with clearance observed for the Linf-GFP parent at 144 h post-infection (PI), and for the Ltar-RFP parent and the Linf/Ltar hybrid at 24 h PI (Fig. 2A and B). Of note, a higher parasitic load was observed for the Linf-GFP strain at 4 and 24 h PI. For the dog cells, as for the murine cells, a higher proportion of infected cells was observed with the Linf-GFP strain (from 68.4% at 4 h PI to 62% at 24 h PI) compared to the Ltar-RFP and the hybrid strains (from 62.5% and 57.8% at 4 h PI to 38.4% and 21.9% at 24 h PI, respectively), and intracellular parasites were detected at both time points for all strains (Fig. 2C). However, none of the strains produce sustained infection in mouse BMDMs, though they persist in dog PBMCs at 24 h PI. Of note, both parental strains and the hybrid were found negative for currently described LRV1 and LRV2 viruses.

Figure 2.

Assessment of the phagocytic uptake of the Linf/Ltar hybrid by macrophages. (A) Microscope images of infected mouse BMDMs. The nuclei were labeled with Hoechst 33342 (white) and the MHC class II staining LAMP-1 was used to visualize the parasitophourous vacuoles (red). (B) Percentage of infected cells and number of parasites per 100 infected mouse BMDMs at 4, 24, 48, and 144 h PI. (C) Percentage of infected cells and number of parasites per infected dog PBMCs at 4 and 24 h PI.

The Linf/Ltar hybrid is a full-genome heterozygous, but the parental contribution to its karyotype varies depending on the chromosome

Whole genome sequencing of the single Linf/Ltar hybrid and of seven of the nine Linf/Linf hybrids revealed in each case a biparental inheritance of all the homozygous SNPs that are different between the two parents, resulting in a full-genome heterozygosity (Fig. 3A and B), as previously reported for experimental hybrids produced in sand flies and in vitro [15, 19, 20, 30, 44]. In both parents most chromosomes were disomic (i.e. with a somy score close to 2), though aneuploidies were observed for chromosomes 1, 10, 25, 29, 31, and 33 for Ltar-RFP (Fig. 3C), and chromosomes 1, 2, 3, 5, 11, 13, 20, 23, 26, 29, 31, and 33 for Linf-GFP (Fig. 3D). For most chromosomes, the tetraploid Linf/Ltar hybrid had inherited all the chromosome copies from each of the parents, often resulting in a 2:2 ratio (e.g. chromosomes 6, 8, 9, 13, 14, and 16) (Fig. 3E). However, for several chromosomes, the somy score in the hybrid indicates the loss of one parental copy and thus unequal chromosome contribution. This was observed for the copies of chromosomes 2, 5, 7, 21, and 26 from the Ltar-RFP parent and for the copies of chromosomes 1, 11, 15, 17, 18, 19, 22, 25, 28, 32, 24, and 35 from the Linf-GFP parent. Chromosomes 3 and 10 both are disomic in the hybrid, with a single copy of each chromosome coming from each of the parental strains. In Leishmania hybrids, mitochondrial DNA inheritance appears uniparental for maxicircle kDNA but biparental for minicircle kDNA [54]. Consistently, minicircles from both parental strains were detected in the Linf/Ltar hybrid (Supplementary Fig. S2), whereas the maxicircle sequence matches only the Ltar parent (Fig. 3F).

Figure 3.

The Linf/Ltar hybrid is a full-genome heterozygous, but the parental contribution to its karyotype varies depending on the chromosome (A and B) Circos plot representation of the inheritance pattern of the homozygous SNP differences existing between their parental strains in the Linf/Ltar hybrid (A) and 7 selected independent Linf/Linf hybrids (B). (C–E) Individual somies of the Ltar-RFP (C), the Linf-GFP (D) strains, and the Linf/Ltar hybrid (E). (F) Representative section of the maxicircle sequence of the Linf/Ltar hybrid, illustrating the parental allelic contribution to the hybrid kDNA.

These results show that the genome of the Linf/Ltar hybrid is not a simple addition of the genetic content of its parents, suggesting that genomic rearrangements during the formation and early growth of hybrids could be important adaptative mechanisms to overcome the genomic shock associated with hybridization.

The Linf/Ltar hybrid transcriptome is expressed from both parental genomes

We questioned whether the Linf/Ltar hybrid transcriptome fully reflects its genomic composition or if certain RNAs are preferentially expressed from one parent. To test this, we isolated RNA from axenic promastigotes of Linf-GFP, Ltar-RFP, and the Linf/Ltar hybrid, performed Illumina sequencing and mapped the RNA reads on the L. infantum reference genome, the L. tarentolae genome or the hybrid reference (both genomes combined). As expected, the RNA reads from Linf-GFP align perfectly to the L. infantum reference (98.9% of alignment) but quite poorly to the L. tarentolae reference (18.2%) whereas the Ltar-RFP strain shows an inverse trend (18.2% of alignment to the L. infantum reference and 98.7% to the L. tarentolae reference). In the hybrid, 50.3% of the reads align to the L. infantum reference genome and 66.8% to the L. tarentolae genome. Using an in silico-generated hybrid reference (both genomes combined), the alignment increases to 99% reads, indicating that both parental genomes are transcribed and contribute to a mixed hybrid transcriptome (Fig. 4A). The PCA plot established from the RNA analysis (mapped on the hybrid reference) also shows a consistent clustering of the experimental replicates from each genotype, with the hybrid clustering in between the Linf-GFP and the Ltar-RFP samples (Fig. 4B).

Figure 4.

The Linf/Ltar hybrid transcriptome is expressed from both parental genomes. (A) Alignments rate of the RNA reads from the Linf/Ltar hybrid and its parents to the L. infantum reference genome (left), the L. tarentolae reference genome (middle), or both genomes combined (right). (B) PCA plots of the three replicates of Linf-GFP, Ltar-RFP, and Linf/Ltar hybrid transcriptomes, mapped on both genomes combined. (C) Proportion of the mRNA reads originating from the L. tarentolae genome in the Ltar/Linf hybrid at the chromosome level. The measured somy ratio for each chromosome, calculated with the DNA sequencing data, is indicated by the black diamonds. (D) Regression plot between the proportion of reads originating from the L. tarentolae genome in the hybrid at the DNA (x-axis) and the RNA (y-axis) levels for each individual chromosome.

We then investigated the transcriptomic contribution of each parental genome to the overall expression of the hybrid by comparing the proportion of reads mapping on the L. tarentolae reference [Ltar/(Ltar + Linf)] at both DNA and RNA levels. At the level of individual chromosomes, we observed a very high correlation between the RNA and the DNA ratios (Fig. 4C and D), indicating no preferential expression of one parental genome compared to the other. These data indicate that the global transcriptome of the Linf/Ltar largely results from the expression of both its parental genomes.

Specific mRNA species are preferentially expressed from one of the parental genomes

Finally, we investigated the transcriptomic contribution of each parental genome at the level of individual genes. Due to differences in annotation between L. infantum and L. tarentolae genomes that precludes a one-to-one mapping, we first derived orthogroups that allowed us to select 7317 shared genes for our analysis (see Materials and methods for the selection criteria). This analysis showed that most orthologous parental genes are expressed at similar levels (Fig. 5A) but also that outliers deviate from the central distribution, indicating that some specific mRNA species could be preferentially enriched from one or the other parental genome.

Figure 5.

Specific mRNA species are preferentially expressed from one of the parental genomes. (A) Proportion of the mRNA reads originating from the L. tarentolae genome for each orthogroup, when orthogroups are grouped by chromosome. (B) Comparison of the proportion of reads originating from the L. tarentolae genome for each orthogroup, at the DNA level (x-axis) and the RNA level (y-axis). The z-score for each orthogroup has been color-coded, with z-scores > 2 indicating significance (outliers). (C) Genomic distribution of the outlier orthogroups on chromosomes 14 and 33. The position of the polycistronic transcriptional units (PTUs) found on these chromosomes is also indicated (bottom band). (D) Gene Ontology analysis of all orthogroups presenting a z-score > 2.

To better characterize this phenomenon, we plotted the RNA versus DNA ratios of Ltar/total reads for each orthogroup with a normalization step accounting for the somy score. This adjustment provided a representation of the entire orthogroup dataset, where outliers were more distinctly visible (Fig. 5B). We calculated z-scores, color-coded in Fig. 5B, with outliers—corresponding to z-scores superior to 2—represented in red. We investigated the genomic distribution of the outliers along the genome and observed no spatial clustering, as exemplified with chromosomes 14 and 33 (Fig. 5C). The genomic distribution of the outliers for each chromosome is given in Supplementary Fig. S3; with a more detailed version accessible here https://doi.org/10.5281/zenodo.15608973. Different types of outliers can be observed: outliers along the diagonal likely represent gene copy number variations where orthogroups are differently represented between the two genomes, with mRNA abundance reflecting the genomic composition and asymmetric parental gene CNVs. The middle-top quadrant reflects equal genomic content but higher L. tarentolae expression, and the middle-bottom shows lower L. tarentolae expression despite balanced genomic content. We performed a GO term enrichment analysis on outliers that revealed several enriched GO terms spanning molecular function (MF), cellular component (CC), and biological process (BP) categories. Notably, terms such as “transporter activity,” “structural constituent of cytoskeleton,” and “microtubule-related functions” were identified. A complete listing of these terms is provided in Fig. 5D. We performed additional GO term enrichment analyses on various outlier categories (Supplementary Fig. S4), and on individual chromosomes (https://doi.org/10.5281/zenodo.15608973). The GO term “hydrolase” was notably found enriched in the outliers from the middle-top quadrant (i.e. with a balanced genomic content but an higher expression of the L. tarentolae transcripts), and the terms “phosphorylation/kinase” in the middle-bottom category (i.e in orthogroups with a balanced quantity of DNA reads from both parents but a higher proportion of RNA reads expressed from the L. infantum genome). These results suggest potential differences in the activation and regulation of signaling cascades between the hybrid and its parents (Supplementary Fig. S4).

Given the well-established and diverse roles of GP63 proteins as virulence factors [55], we next focused our analysis specifically on this gene family. In the reference annotations used, Leishmania tarentolae has 59 GP63 genes, compared to 13 in L. infantum. Due to annotation inconsistencies and the complex nature of the GP63 gene cluster organization, GP63 orthogroups were excluded from our initial analysis. Specifically, while all L. infantum GP63 genes are located on chromosome 10, many L. tarentolae GP63 genes are found on smaller, unplaced contigs not assigned to chromosome 10. We generated a curated list of GP63 genes located on chromosome 10 in both species (excluding smaller contigs), resulting in 42 GP63 genes in L. tarentolae and 13 in L. infantum. Based on this subset, the expression ratio in the hybrid (calculated as the L. tarentolae contribution divided by the total expression) is approximately 70% and aligns closely with the expected DNA copy ratio of ∼76% [42/(42 + 13) × 100]. Finally, since our analysis effectively excluded “orphan” genes (those present in one parent but lacking an ortholog in the other), we also examined the RNA and DNA read counts for all genes in the hybrid. This revealed no substantial differences between genes within orthogroups and the orphans (Supplementary Fig. S5).

Overall, our results and analyses indicate that while most transcripts are expressed in a balanced fashion from the parental genomes in the Ltar/Linf hybrid, certain mRNA species are preferentially expressed from one or the other parental genomes. Thus, beyond genomic restructuring, post-transcriptional regulation may serve as an additional mechanism for the establishment of viable hybrid phenotypes that may promote parasite evolvability and fitness gain.

Discussion

In this study, we demonstrated that L. infantum and L. tarentolae, two highly divergent Leishmania species belonging to distant clades, can produce sexual hybrid F1 progeny, at least in vitro. As far as we are aware, this is the most distant hybrid ever reported, either from field isolates or experimental in vivo or in vitro production. The occurrence of this hybrid not only emphasizes that there seems to be no species barrier in Leishmania hybridization, even if the frequency of successful hybridization events may depend on the combination of parasite strains and their genetic distance but also provided us the unique opportunity to investigate the molecular mechanisms involved in the obtention and survival of Leishmania hybrids.

Despite being close to tetraploidy the Linf/Ltar hybrid does not simply cumulate the genomic content of its diploid parents, as one or several chromosome copies from one or the other parental genomes are lost, as previously reported for other experimental hybrids [23, 24, 35, 56]. This observation, together with the very low frequency of hybrid formation, clearly emphasizes that hybridization in Leishmania is much more than a simple fusion of two cells and addition of their genomes. A previous study based on the analysis of the genomic content of experimental hybrids from the first and second generations demonstrated that hybrid production results from a meiotic-like process [19, 23]. However, the identity and more specifically the ploidy of the parental cells fusing is still not established. The hybrids produced in this study were all close to tetraploid, and an important proportion of hybrids previously generated in vitro are polyploids (3n or 4n) [24, 35, 56]. These “young” hybrid genomes contrast those of field-isolated hybrids that are mainly diploids, which could indicate that fusion of diploid parental cells is the initial step of hybridization, generating a tetraploid hybrid offspring that can later undergo a meiotic reduction of its chromosome content. Another hypothesis would be that diploid hybrids result from a classical meiotic cycle with fusion of haploid gametes (whose existence is so far not demonstrated in Leishmania), whereas polyploid hybrids result from the abnormal fusion of a diploid cell with a haploid or a diploid cell.

Regardless of the ploidy of the fusing cells, the hybrids resulting from sexual reproduction need to survive, first in the environment where they originated (the sand fly gut) and later in any environment they will encounter. Hybridization, by sexual reproduction or through other types of genetic exchange such as transposable elements or plasmids, leads to a “genomic shock,” a concept first theorized in 1984 by Barbara McClintock [34]. This concept originating from plant studies posits that the stress and regulatory interference caused by combining two different genomes can destabilize the genome, potentially leading to significant genetic and epigenetic alterations. This raises the question of the compatibility of parental strains, as the hybridization of genetically more distant organisms could induce a stronger genomic shock, more difficult to overcome. The mechanisms allowing the homeostatic reprogramming of hybrids to establish a viable phenotype are not well understood and probably very different depending on the organisms and the type of genetic exchange. In the case of Leishmania products of sexual reproduction, our study indicates that several levels of gene expression control could be involved. At the genomic level, the adjustment of the number of chromosome copies appears as an important mechanism of fitness adaptation, whereas at the transcriptomic level differential mRNA abundance between parental alleles can further buffer possible genetic incompatibilities. Leishmania genes are organized in polycistronic units whose transcription initiation is not differentially regulated. The control of gene expression largely occurs at post-transcriptional levels via selective stabilization or degradation of mRNAs [57]. Thus, the mono-parental allelic expression we observed in the hybrid should be controlled by differential mRNA turnover. The molecular mechanisms that can distinguish between different parental mRNA alleles and the underlying structural requirements for differential mRNA turnover eludes us but may involve expression of non-coding RNAs such as snoRNAs, or epitranscriptomic modification of mRNA [58–60]. It remains to be determined whether the homeostatic adaptation of hybrids also involves additional regulatory mechanisms, such as translational control, protein turnover, or post-translational modifications.

A key question in organisms with a facultative sexual reproductive cycle is the role of sex in their evolution. Unlike clonal division, sexual reproduction is a costly and potentially risky process, as it generates entirely new genomes that may be non-viable or less adapted to the host environment than those of the parental organisms. However, this risk can also provide an evolutionary advantage, particularly for organisms facing harsh or fluctuating environments, as the new genome may quickly evolve towards expressing new and more robust adaptive phenotypes, notably in concert with the compensatory mechanisms discussed above. In the case of Leishmania, the frequent reports of natural hybrids among field isolates as well as the extraordinary genome plasticity of these parasites indicate that sexual reproduction is more frequent than initially anticipated and may play a central role in both the long-term evolution of Leishmania species and their short-term adaptative phenotypes. Further investigation—facilitated by experimental hybrid production—is needed to expand our currently limited understanding of the mechanisms and evolutionary consequences of the Leishmania hybridization process.

Finally, even though the model hybrid presented in this work has been generated in vitro, there are several lines of evidence that L. tarentolae and L. infantum, both present in Southern Italy, could engage in hybridization in nature. First, the recent molecular detection of L. tarentolae in human blood in central Italy [61] and by PCR and immunofluorescence antibody test (IFAT) in sheltered dogs in Italy [62] suggests that this parasite species could also circulate in mammals. Of note, no live L. tarentolae parasites were recovered from those dogs [62, 63]. Second, human and dog blood was detected in engorged S. minuta females, showing that these flies do not feed exclusively on reptiles [33]. Third, L. infantum amastigote-like forms and its DNA were detected in lizards and geckoes residing in or close to dog shelters in southern Italy [30]. Finally, experimental infections demonstrated that L. tarentolae is also able to complete its life cycle in P. perniciosus, a proven vector of L. infantum circulating in the Mediterranean basin [64]. Together with our study demonstrating that L. tarentolae and L. infantum are genetically compatible for hybridization in vitro, these arguments stress the importance of enhanced surveillance to help prevent future outbreaks of hybrid-derived strains.

Supplementary data

Supplementary data is available at NAR Molecular Medicine online.

Conflict of interest

The authors declare no competing interests.

Funding

This work was supported in part by an EMBO Scientific Exchange Grant to V.N.L.F., by the Agence Nationale pour la Recherche Labex “Integrative Biology of Emerging Infectious Diseases” [contract ANR-10-LABX-62-IBEID,S2I program to I.L.], the ANR JCJC program [grant SFLeisHyb to I.L.], the Roux-Cantarini and the HORIZON-TMA-MSCA-PF-EF post-doctoral programs [grant SF-Leishyb to I.L.], and the ERC SYNERGY project DecoLeishRN [grant agreement ID: 101071613 to T.C., P.P. and G.F.S.]. This research was supported in part by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH author(s) were made as part of their official duties as NIH federal employees, are in compliance with agency policy requirements, and are considered Works of the United States Government. However, the findings and conclusions presented in this paper are those of the author(s) and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services. The authors thank all the contributing programs for their financial support.

Data availability

The DNA and the RNA data generated in this study have been deposited on the European Nucleotid Archive under the reference E-MTAB-14984 and E-MTAB-14957, respectively. Intermediate data such as raw counts of the DNA and RNA reads as well as additional analyses performed in this study are accessible here: https://doi.org/10.5281/zenodo.15608973. Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Isabelle Louradour (isabelle.louradour@pasteur.fr).

Declaration of generative AI and AI-assisted technologies

During the preparation of this work, the authors used ChatGPT and Perplexity to improve the readability and ensure correct syntax in the manuscript.