Catalase impairs Leishmania mexicana development and virulence

ABSTRACT Catalase is one of the most abundant enzymes on Earth. It decomposes hydrogen peroxide, thus protecting cells from dangerous reactive oxygen species. The catalase-encoding gene is conspicuously absent from the genome of most representatives of the family Trypanosomatidae. Here, we expressed this protein from the Leishmania mexicana Β-TUBULIN locus using a novel bicistronic expression system, which relies on the 2A peptide of Teschovirus A. We demonstrated that catalase-expressing parasites are severely compromised in their ability to develop in insects, to be transmitted and to infect mice, and to cause clinical manifestation in their mammalian host. Taken together, our data support the hypothesis that the presence of catalase is not compatible with the dixenous life cycle of Leishmania, resulting in loss of this gene from the genome during the evolution of these parasites.


Introduction
Leishmania (Kinetoplastea: Trypanosomatidae [1,2]) is a genus of about two dozen species that are responsible for leishmaniasis, a vector-borne disease affecting humans and animals in tropical and subtropical regions [3]. The infection is predominantly transmitted by phlebotomine sandflies and manifests in three different forms -cutaneous, mucocutaneous, and visceral leishmaniasis -depending on the infecting species, geographical location, and the immune status of the host. Currently, the World Health Organization estimates the annual burden of leishmaniasis at about 700,000 to 1 million new cases and up to 65,000 deaths [4]. Although leishmaniasis is a poverty-related disease, climate change, migration caused by the military conflicts, and ever-growing international travel have resulted in the spread of the disease into new areas. The currently employed drugs have numerous side effects and no efficient vaccine is available to eradicate leishmaniasis [5,6]. Thus, functional studies of Leishmania are important for our understanding of disease etiology, identification of novel virulence factors, and the development of an efficient vaccination and/or treatment strategy.
Leishmania are dixenous (= two hosts) parasites with two main developmental stages: extracellular promastigotes in the insect gut and intracellular amastigotes within vertebrate macrophages. Their life cycle in the invertebrate host is confined to the digestive tract, where elongated nectomonad promastigotes escape from the endoperitrophic space and transform into short leptomonad promastigotes. In the midgut, leptomonads differentiate either into haptomonads attached to the stomodeal valve or into mammal-infective metacyclic promastigotes that are transmitted during blood meals of infected sandflies [7,8]. In the blood, metacyclic promastigotes are phagocytized by neutrophils and macrophages and transform into non-motile amastigotes inside phagolysosomes [9]. Upon ingestion by sandflies, Leishmania amastigotes differentiate into procyclic promastigotes inside the insect's midgut endoperitrophic space, and the whole cycle repeats [10][11][12].
It is generally accepted that dixenous parasites have evolved from their monoxenous (= one host) relatives [13], but how extensive genomic changes during the evolution of Trypanosomatidae species happened to allow the emergence of such a complex lifestyle is still a matter of debate. On the one hand, the infective potential of presumably monoxenous species was supported by their ability to adapt to the warm-blooded environment in coinfections [14][15][16][17] or single infections [18], and further corroborated by the thermoresistance of some trypanosomatid species [19][20][21]. On the other hand, comparative genomic analyses of monoxenous and dixenous species have revealed largescale gene gains and losses, highlighting crucial differences between these groups of parasites [22][23][24]. The gene for the antioxidant enzyme catalase is on this list. This protein is responsible for the decomposition of H 2 O 2 and, thus helps to protect from dangerous reactive oxygen species (ROS). We have recently performed a broad genomic screening for the presence of catalase in Euglenozoa [25,26]. The catalase-encoding gene was absent in the common ancestor of trypanosomatids and acquired independently, at least twice, by monoxenous Leishmaniinae and members of the Blastocrithidia/ "jaculum" clade [27]. Surprisingly, it was secondarily lost from all Leishmania spp. genomes sequenced so far. This agrees with studies pointing to the signaling role of H 2 O 2 in regulating Leishmania differentiation [28,29]. Taken together, these results prompted us to speculate that the loss of catalase could have been a cause or an effect of the dixenous life cycle evolution in Leishmania. To investigate this, we expressed catalase from the monoxenous Leishmaniinae Leptomonas pyrrhocoris in the dixenous Leishmania mexicana. Of note, expression of catalase in Trypanosoma brucei and T. cruzi is detrimental for the development of these parasites in vivo [30,31]. These cases are not directly relevant for Leishmania, because, in contrast to the latter, the common ancestor of trypanosomes did not possess a catalase-encoding gene [32].
Numerous molecular tools for regulated gene expression and/or protein (de)stabilization have been developed over the year, allowing researchers to tackle different aspects of Leishmania biology [33][34][35][36]. However, most of these systems have shown insufficient efficacy in developmental studies. The main reason for this is the unconventional mechanism of gene expression in trypanosomatids, which relies on the formation of poly-cistronic mRNAs processed further via trans-splicing [1,37]. Gene expression heavily depends on the regulatory elements in the untranslated regions (UTRs), controlling the transcript processing, stability, and subsequent translation in each of the main developmental stages of Leishmania (procyclic promastigotes, metacyclic promastigotes, and amastigotes) [38][39][40][41]. The UTRs in most commonly used Leishmania vectors drive expression in specific stages of the parasite's life cycle [34,42], making developmental studies not possible.
In this work, we applied a new strategy to express catalase throughout Leishmania development. Our approach relied on the 2A self-cleaving peptide derived from Teschovirus A (previously Porcine teschovirus-1) [43]. We demonstrated that catalase-expressing L. mexicana are defective both in vitro and in vivo (in sandflies and mice infection models), further supporting our premise that catalase presence is not compatible with the dixenous life cycle of Leishmania.

Ethics statement
Animals were maintained and handled in the animal facility of Charles University in Prague in accordance with institutional guidelines and Czech legislation (Act No. 246/1992 and 359/2012 coll. on protection of animals against cruelty in present statutes at large), which complies with all relevant EU guidelines. All the experiments were approved by the Committee on the Ethics of Laboratory Experiments of the Charles University and were performed under permission No. MSMT30397/ 2019-3 of the Czech Ministry of the Environment. All efforts were made to minimize the number and suffering of experimental animals during the study.

Sandflies and mice
The colony of Lutzomyia longipalpis (originated from Brazil) was maintained in the insectary of the Department of Parasitology, Charles University in Prague, under standard conditions as described previously [45]. BALB/c mice (AnLab, Prague, Czech Republic) were maintained in the animal facility of the Department of Parasitology with a 12/12 light/ dark photoperiod, at 22-25°C, and 40-60% humidity.

Expression vectors and transfection
The genetic nomenclature used in this work followed the guidelines of Clayton et al., 1998 [50]. All constructs described below were sequenced before transfection.
To analyze the expression of the B-TUBULIN locus, a construct to replace LmxM. 32.0792 with the hygromycin B (HYG) resistance gene ORF was generated by fusion PCR [51,52]. 5´and 3´Β-TUBULIN UTRs, and HYG ORF (from pLEXSY-hyg2, Jena Bioscience GmbH) were amplified using primer pairs A1/B1, C1/D1, and E1/F1 (hereafter all primer sequences are listed in S1 Table). Nested primers G1/H1 were used for the last PCR amplification step. The PCR product was gel purified and transfected into the log phasegrown procyclic promastigotes of L. mexicana as described before [53]. The transfectants were selected in cultivation medium supplemented with 100 μg/ml of hygromycin B (Carl Roth GmbH, Karlsruhe, Germany) for 10 days.
The Teschovirus A P2A DNA fragment was generated by conventional PCR with long overlapping oligonucleotides (A2/B2). It was fused with LmxM.32.0792 5 ´and 3´ UTRs (amplified with C2/D2 and E2/F2, respectively) using the nested primers G2/H2. The final product was cloned into the SwaI-digested pLEXSY-sat2.1 (pLEXSY_LmxM.32.0792_UTR). Tagged nourseothricin resistance gene (V5::SAT), mCHERRY (mCHERRY::HA), and CATALASE (CAT::HA) ORFs were PCR amplified from the pLEXSY-based constructs using primers A3/B3, C3/D3, and E3/F3, respectively. They were cloned into the pLEXSY_LmxM.32.0792_UTR using restriction enzymes EcoRI and HindIII for V5::SAT, and BglII and KpnI for CAT::HA and mCHERRY::HA, thus producing cassettes referred to as SAT::2A::CAT and SAT::2A:: mCHERRY. Restriction sites for cloning were introduced into the primer sequences. The resultant plasmids were digested with SwaI, DNA products were purified and used for transfection as described above. The transfectants were selected in cultivation medium supplemented with 100 μg/ml of nourseothricin (Jena Bioscience GmbH) for 10 days.  [56]. In total, 13 reads were picked and separately reassembled with Canu. This procedure yielded a 6,602 bp-long contig, which represented the allele variant of the Β-TUBULIN locus with one Β-TUBULIN gene replaced with SAT::2A::mCHERRY sequence. As overall coverage of this contig was low, the assembler was not able to correct all sequencing errors in a consensus sequence. Alignment of the 6,602 bp-long contig with Β-TUBULIN locus was done manually.

Verification of genome integration by
Similarly, we performed PacBio sequencing (but with lower sequencing depth, 13,068 polymerase reads) for L. mexicana-CAT. We mapped these reads on the L. mexicana genome with BWA tool v. 0.7.12 [57], using "bwa mem" alignment algorithm with "-x" option set to "pacbio". Average mapped read length was 11 kbp. Only 5% of the genome was not covered and 72% of the genome was covered with less than seven reads. From 16 reads that overlapped with the Β-TUBULIN tandem array on chromosome 32 (coordinates 281,233-282,564), only three contained SAT::2A::CAT fusion surrounded by proper sequences. These reads were picked and further examined with BLASTN against L. mexicana chromosome 32, NCBI non-redundant nucleotide collection, and SAT::2A::CAT fusion sequence. BLASTN high scoring pairs were used to annotate PacBio sequences to ensure that they contain a SAT::2A::CAT fusion, surrounded by proper Β-TUBULIN sequences from the chromosome 32. To ensure that we did not miss any other unmapped PacBio reads bearing SAT::2A::CAT fusion sequence (either due to the presence of non-genomic sequence introducing long deletion in alignment or read mapping in other genomic location), fusion sequence was BLASTed against all PacBio reads, but only three reads that had been already annotated, were returned.

RNA isolation and sequencing
Total RNA was isolated as described previously [23]. Six transcriptomic libraries were prepared using TruSeq stranded mRNA kit (Illumina, San Diego, USA) and were sequenced on Illumina platform in paired-end mode with read length of 150 bp at Macrogen Europe (Amsterdam, Netherland). An average sequencing depth was 31 × 10 6 reads per sample. RNA sequencing reads were deposed in NCBI SRA under BioProject PRJNA660365.

Differential expression analysis
RNA sequencing reads for each sample were trimmed using Trimmomatic v. 0.39 [58]. Trimmed reads were mapped on the TriTrypDB L. mexicana genome assembly with Bowtie2 v. 2.3.4.1 [59] using default options (end-to-end mapping). Read alignments were processed with SAMTools and BEDTools [60] to extract read counts per genes. Read counts data were analyzed with DESeq2 package v. 1.18 [61]. Read counts were log-normalized by DESeq2 "rlog" function.

Catalase activity assays
Catalase activity was determined using a catalase calorimetric activity kit (Life Technologies/Thermo Fisher Scientific) according to the manufacturer's instructions. The absorbance was read at 560 nm on Epoch Microplate Spectrophotometer (BioTek/Agilent, Winooski, USA). All experiments were performed in triplicate.
For fast evaluation of enzymatic activity, L. mexicana cells (5 × 10 6 ) were spun down at 3,000 rpm for 5 min, washed with PBS, re-suspended in 10 µl of PBS, and placed on a microscopic slide. A drop of 3% H 2 O 2 was added to the cells, and the formation of oxygen bubbles indicated catalase activity.

Cell viability assay
The cytotoxic effect of H 2 O 2 was determined using the fluorimetric method based on the conversion of resazurin (Alamar Blue) by live cells [64,65]. Briefly, serial 2× dilutions of H 2 O 2 (4 to 0.03125 mM) were prepared in 96-well plates. The log phase-grown procyclic promastigotes were added to the wells to the final concentration of 5 × 10 5 cells/ml and the plates were incubated for 1 h. The 0.5 mM stock of resazurin was added to the final concentration of 50 µM. Fluorescence was read after 24 h of incubation at 23⁰C on Infinite M200 (Tecan, Männedorf, Switzerland) using excitation and emission wavelengths of 540 and 590 nm, respectively. The data were analyzed using Prism (GraphPad Software, San Diego, USA) with nonlinear regression and sigmoidal dose-response analysis with variable slope to obtain EC 50 values. All experiments were performed in triplicate.

Sandfly infection
Promastigotes from log-phase cultures (day 3-4) were washed twice in saline and re-suspended in heatinactivated rabbit blood at a concentration of 1 × 10 6 promastigotes/ml. Sandfly females (5-9 days old) were infected by feeding through a chick-skin membrane (BIOPHARM, Jílové u Prahy, Czech Republic) on the promastigote-containing suspension. Engorged sandflies were separated and maintained under the same conditions as the colony. Females were dissected on day 2, 5, and 9 post-blood meal (PBM) and the abundance and location of L. mexicana parasites in the sandfly digestive tract were examined under the light microscope. Leishmania mexicana parasites were found in five locations in the sandfly digestive tract: endoperitrophic space (E.SP.), abdominal midgut (AMG), thoracic midgut (TMG), cardia (CA) and the stomodeal valve (SV). Parasite loads were estimated by two methods [66]. Gut infections in situ were graded as light (<100 parasites/gut), moderate (100-1,000 parasites/gut) and heavy (>1,000 parasites/gut). For exact counting of parasites, 10 guts from each group were transferred into 100 µl of 1% formaldehyde solution, re-suspended and counted using a hemocytometer. Experimental infections were repeated independently four times.

Transmission by bite
Anesthetized BALB/c mice were placed individually into small 20 × 20 cm cages and 10-15 infected L. longipalpis females (days 8-13 PBM) were allowed to feed on the whole body of mice for one hour. Two researchers followed each experiment; one recorded biting sites and feeding time, while the second collected engorged sandflies by an aspirator immediately after terminating their blood meal. After exposure, mice were euthanized; the biting sites excised and skin biopsies were stored at −20°C until DNA extraction. Engorged sandfly females were dissected immediately and their guts checked for the presence of Leishmania.

Macrophage culture and infection
Bone marrow-derived macrophages were isolated and infected as described previously [67]. Briefly, cells were released with trypsin-EDTA solution (Sigma-Aldrich), washed in 0.9% saline solution and seeded into Greiner Cellstar 24-well plates (Sigma-Aldrich) at a density of 4 × 10 5 cells/ml. Macrophages were exposed to the stationary-phase grown parasites at a promastigote to macrophage ratio of 6 to 1. Two hours later, complete uptake of Leishmania was verified under a light microscope and cells were incubated for 3 days in complete RPMI 1640 (Life Technologies/Thermo Fisher Scientific) supplemented as above, with a combination of IFN-γ (50 U/ml, AbD Serotec/Bio-Rad, Kidlington, UK) and LPS (0.5 µg/ml, Sigma-Aldrich) for classically stimulated macrophages, or IL-4 (25 ng/ml, eBioscience/Thermo Fisher Scientific, San Diego, USA) for alternatively stimulated macrophages. Infected macrophages were then lysed in 0.016% SDS (Sigma-Aldrich) for 5-10 min at room temperature. The suspension was passed through the 1 ml insulin syringe, washed with 0.9% saline solution, and spun down at 3,000 × g at room temperature. The pellet was re-suspended in 100 µl of 0.9% saline solution and amastigotes were counted with a hemocytometer.
All experiments were performed independently twice and samples were analyzed in triplicate.

Nitrite analysis to measure NO production
The accumulation of nitric oxide produced by cultured macrophages over a 72 h period was determined in a microplate assay using Griess reagent as described previously [68].

Mice infection
Leishmania mexicana cultures (passage 1 or 2 after recovering from mice lesion) were enriched for metacyclic forms by inoculation into cultivation medium with the pH adjusted to 5.5 [44] and harvested 10 days later. Eight weeks old female BALB/c mice (10 per WT; 10 per L. mexicana-mCHERRY; 20 per L. mexicana-CATALASE lines) were anesthetized with the mixture of ketamine and xylazine (62 and 25 mg/ kg, respectively) and infected by intradermal inoculation of 1 × 10 5 parasites in 5 µl of sterile saline into ear pinnae. Control mice were inoculated with the same volume of saline without parasites. Mice were checked weekly for external signs of the disease until week 12 post-infection (p.i.) when they were subjected to xenodiagnostic experiment and sacrificed.

Xenodiagnosis
Xenodiagnostic experiments were performed on week 12 p.i. Five to seven-day-old L. longipalpis females were allowed to feed on the inoculated ear (the site of inoculation of parasites) of anesthetized mice. Mice were covered with the cotton bag, so that only the ear pinnae were accessible to sandflies and placed into a small cage with 30-50 sandfly females. After one hour, engorged females were separated and maintained at 26°C on 50% aqueous sucrose solution. Females were dissected and their guts examined under the light microscope for presence of L. mexicana 8 days PBM.

Morphometry of parasites
Midgut smears of sandflies infected with L. mexicana dissected by day 9 PBM were fixed with methanol, stained with Giemsa (Sigma-Aldrich), examined by light microscopy (Olympus DP70, Olympus, Shinjuku, Tokyo, Japan) with an oil immersion objective and photographed. Body length, flagellar length and body width of 500 randomly selected promastigotes from 5 to 6 samples were measured for each group using Image J software [69]. Three morphological forms were distinguished, based on the criteria of [70,71]: (a) nectomonad promastigotes, body length ⩾ 14 µm; (b) metacyclic promastigotes, flagellar length >2 times body length and body length <14 µm, and (c) leptomonad promastigotes, body length <14 µm and flagellar length ⩽ 2 times body length. Haptomonads were not distinguished as they may remain attached to the gut and can be underrepresented on gut smears.

Statistical analysis
Differences in intensities of infections and percentage of morphological forms found in infected flies were analyzed by the χ2 test using SPSS Statistics 23 package. Differences in parasite loads in sandflies were tested by the non-parametric Mann-Whitney U test. The correlation between lesion size and Leishmania strain (categorical variable) over time (continuous independent variable) were tested using R software by fitting multilevel linear regression model (package "nlme").

Novel 2A self-cleaving peptide-based expression system in L. mexicana: proof of principle and revised structure of the Β-TUBULIN locus
Our attempts to express a catalase from L. pyrrhocoris H10 in three developmental stages of L. mexicana using a T7 polymerase-driven, tetracycline-inducible gene expression system failed [42,53]. The main reason for this is that the inducible gene expression system depends on exogenous UTRs flanking both the gene of interest and the T7 polymerase integrated into the 18S rRNA locus, thereby influencing the mRNA and protein levels. To overcome this limitation, we searched for genes, which are transcribed and translated in all three developmental stages of L. mexicana [22,72,73]. The Β-TUBULIN is such a gene. Indeed, it encodes a major structural element of microtubules, which play an important role in numerous biological processes. It is well documented that TUBULIN is expressed in all developmental stages of L. mexicana [74]. While the transcription is fairly stable, the protein is more abundant in the procyclic promastigotes than in amastigotes [75,76].
Our model organism is L. mexicana MNYC/BZ/ 1962/M379, an isolate, which is different from the TriTrypDB reference MHOM/GT/2001/U1103. We previously reported genomic sequencing of this isolate using Illumina-based methodology [79], but the repeated nature of the TUBULIN-encoding locus did not allow us to assemble its sequence properly. To overcome this hurdle, we used a long-read generating PacBio sequencing platform, which allowed us to assemble a 990 kb-long contig, containing over 70% of the L. mexicana chromosome 32 (the final assembly was 33.9 Mb, with N50 of 640 kb). We found that this contig contains seven almost identical tandemly arranged copies of Β-TUBULIN (GenBank Acc. No. We decided to take advantage of this genomic organization and replaced an ORF for the first gene in the tandem (LmxM.32.0792) by the HA-tagged hygromycin B resistance gene. Western blotting analysis revealed that this gene is stably expressed throughout L. mexicana development (S2 Fig). To avoid introduction of exogenous UTRs into the expression cassette (so the expression would be regulated only by 5ʹ and 3ʹ UTRs of the stable expressed LmxM. 32.0792), we used the 2A self-cleaving peptide. This is a short, 19-22 aa long sequence, which is recognized by the eukaryotic ribosome and skipped during translation. As a result, two separate proteins, located upstream and downstream of the 2A linker sequence, are produced simultaneously [80,81]. From the most widely used 2As, we chose that one derived from Teschovirus A because of its high cleavage efficiency [82]. Importantly, such bicistronic expression is regulated by endogenous UTRs and co-expression level of both proteins is equal. Of note, another system, which relies on a 2A peptide of Thosea asigna virus, has been previously reported in L. mexicana and used to tag endogenous proteins [83,84].
To prove the concept, we first generated the expression cassette SAT:: Due to the high rate of sequencing errors of PacBio technology, high sequence identity of the Β-TUBULIN genes within a cluster on chromosome 32, and average read length being shorter than that of the Β-TUBULIN cluster size, it was not possible to determine with confidence, which of the 7 copies of the gene was replaced.
The L. mexicana-mCHERRY cells were differentiated in vitro [85] and mCHERRY expression was assayed by RT-qPCR (S3B Fig) and Western blotting (S3C Fig). The differentiation was controlled by RT-qPCR analysis of the selected marker genes PFR1D, SHERP, and AMASTIN for promastigotes (both pro-and metacyclic), metacyclic promastigotes, and amastigotes, respectively [86] and compared to the wild type L. mexicana cells (S5 Fig, panels A and B), confirming no major defect in differentiation in vitro. While the mCHERRY RNA level was stable, we noticed a drop in mCHERRY protein level in amastigotes. Nevertheless (and as a significant improvement over previously published system [42]), this approach allowed us to express the protein of interest throughout L. mexicana developmental cycle. This was further confirmed by immunofluorescent microscopy, detecting fluorescent mCHERRY in axenically differentiated procyclic/metacyclic promastigotes and amastigotes (S3D Fig). Importantly, the 2A peptide-mediated cleavage appeared to be very efficient, as no higher molecular weight un-cleaved protein was detected by anti-HA Western blotting (S3C Fig). In conclusion, the novel bicistronic expression system from the Β-TUBULIN endogenous locus reported here allowed us to express and follow a protein of interest during promastigote-to-amastigote development.

Generation of a L. mexicana line expressing CATALASE
We next generated a L. mexicana-CAT (CATALASE) clonal line using the same approach as outlined above (Figure 1a) and verified it by PacBio sequencing (S6 Fig). The CATALASE used in this study originated from Leptomonas pyrrhocoris. Sequence analysis of this enzyme revealed conservation of the amino acids involved in heme and NADPH binding, as well as those involved in active site charge relay (S7 Fig). In contrast to their human or yeast counterpart [87,88], trypanosomatid catalases lack a peroxisomal targeting signal and localized in the cytoplasm [30,53,89].
After differentiation of the transgenic parasites in vitro, no major differences with respect to the differentiation marker genes PFR1D, SHERP, and AMASTIN were detected between L. mexicana-CAT and L. mexicana-mCHERRY (S5 Fig), implying that expression of catalase does not influence transcription of these genes. The mRNA level of CATALASE was similar in procyclic and metacyclic promastigotes, and amastigotes (Figure 1b), while its protein level was lower in amastigotes (Figure 1c). This pattern of expression was similar to that of mCHERRY and may reflect intrinsic properties of the Β-TUBULIN locus [74,76] Expression of catalase has little effect on global transcription, or expression of genes involved in oxidative stress response or heme metabolism Differential expression analysis was carried out for all genes, but only 30 (including tRNAs and ncRNAs) had a fold change value of >2 (BioProject PRJNA660365). Catalase is a heme-containing enzyme involved in ROS protection, thus we specifically analyzed genes involved in heme metabolism and oxidative stress response (S2 Table). Only 3 and 11 genes involved in heme metabolism and oxidative stress response pathway, respectively, were differentially expressed (highlighted in S2 Table) with adjusted p-values below 0.05. However, the fold change for all these transcripts was below 2.

Impaired development of L. mexicana-CAT in Lutzomyia longipalpis
In sandflies, the development of L. mexicana-mCHERRY and L. mexicana-CAT was followed on days 2, 5, and 9 post-blood meal (PBM). On day 2 PBM, heavy infections were observed in most females and parasites were localized in the endoperitrophic space, i.e. inside the bloodmeal surrounded by the peritrophic matrix. High infection rates were also documented post defecation of the blood remnants (days 3-4 PBM) and most females remained infected by day 9 PBM. At this time point, the intensity of infection was heavy and parasites were localized in the thoracic part of the midgut colonizing the stomodeal valve (Figure 2a). Females infected with L. mexicana-CAT, showed only slightly lower parasite burden during the early stages of infections (χ 2 = 1.193, df = 1, p = 0.379 and χ 2 = 4.559, df = 1, p = 0.066 on day 2 and 5 PBM, respectively). However, on day 9 PBM the percentage of infected females was about 75% for the L. mexicana-CAT strain, compared to 95% for the L. mexicana-mCHERRY (Figure 2a). This difference was statistically significant (χ 2 = 9.895, df = 1, p = 0.002).
Morphological analysis of the infections 9 days PBM (500 promastigotes per strain) revealed that development of L. mexicana-CAT was considerably impaired, compared to that of L. mexicana-mCHERRY (Figure 2b). This defect manifested in the significantly lower representation of macrophage-infective metacyclic promastigotes (and concomitant higher representation of elongated nectomonads) in sandflies infected with L. mexicana-CAT (2.6% versus to 20% in control group, χ 2 = 137.256 df = 2, p < 0.0001). Of note, a lesser impact of catalase expression on metacyclogenesis was observed in vitro, where percentage of metacyclic promastigotes at day 9 of differentiation were 35%, 36%, and 21% for WT, L. mexicana-mCHERRY, and L. mexicana-CAT, respectively (S9 Fig,  data in brackets). No differences were documented when comparing infectivity of WT and L. mexicana-mCHERRY lines (χ 2 = 0.459, df = 1, p = 1 for day 2; χ 2 = 0.564, df = 1, p = 1 for day 5; and χ 2 = 0.339, df = 1, p = 1 for day 9) (S10A Fig) and representation of morphological forms in guts of L. longipalpis on day 9 PBM (χ 2 = 0.228, df = 2, p = 0.893) (S10B Fig), indicating that neutral (mCHERRY) genetic manipulation of the locus itself has no effect on L. mexicana virulence in insects. We concluded that parasites, expressing catalase, retained the ability to produce heavy infection in the natural vector L. longipalpis with colonization of the stomodeal valve and production of metacyclic parasites. However, the infection rate and proportion of metacyclic promastigotes in these flagellates have been significantly decreased compared to the control group L. mexicana-mCHERRY by day 9 PBM.
Transmission experiments demonstrated that L. longipalpis infected with WT and L. mexicana-CAT did not differ in feeding time on BALB/c mice (Table 1), while significant differences were observed in transmission efficiencies (Figure 2c). Sandfly females infected with WT and L. mexicana-CAT transmitted parasites to mice ears in 15/17 (88%) and 13/21 (62%) cases, respectively. Moreover, the Leishmania load (the number of parasites) also differed significantly. It was over 1,000 parasites (maximum 74,300) for the WT-infected females, and mostly below 100 (maximum 211) parasites in females infected with L. mexicana-CAT (Table 2, Figure 2c).

Catalase compromises ability of L. mexicana to infect macrophages in vitro
Next, we tested the ability of L. mexicana-CAT to establish infection in mouse bone marrow-derived macrophages (BMMɸ). After 72 hours p.i. of macrophages with metacyclic promastigotes, the number of L. mexicana-CAT amastigotes was significantly lower as compared with Leishmania-mCHERRY parasites in unstimulated (p < 0.001), as well as classically-(p < 0.001) and alternatively-(p < 0.001) stimulated macrophages (Figure 3a). The observed differences in the number of amastigotes in unstimulated and classically stimulated macrophages were expected, as stimulation with IFN-γ + LPS generally leads to a considerable reduction in parasite load. The numbers of Leishmania-mCHERRY and L. mexicana-CAT amastigotes were reduced in macrophages, stimulated with IFN-γ + LPS (p < 0.001 for both Leishmania lines). Of note, no statistically significant difference in infection of macrophages was noted between WT and L. mexicana-mCHERRY lines (p = 0.535; 0.0859; and 0.303 for unstimulated, classically, and alternatively stimulated macrophages) (S11 Fig). No differences in NO production were observed for unstimulated and alternatively stimulated macrophages in both WT (p = 0.103) and CAT (p = 0.91) groups (S12 Fig), ruling out the potential role of NO in the documented phenotypic manifestation.
No extracellular parasites were detected 2 hours after infection that makes it reasonable to assume that defect in the ability of the L. mexicana-CAT cells to infect macrophages in vitro was not associated with their initial uptake.

Decreased virulence of L. mexicana-CAT in mice
We also tested L. mexicana-CAT infectivity in a murine model and documented a clear difference in lesions development between the mCHERRY and CATALASEexpressing L. mexicana. Lesions appeared in all animals infected with L. mexicana-mCHERRY, while only in 10 out of 20 in the CAT group. In addition, lesions produced by L. mexicana-CAT appeared 2 weeks later and developed slower than in the control group (p < 0.001) (Figure 3b and c). Of note, no statistically significant difference was observed in lesions development between the WT and mCHERRY-expressing L. mexicana (p = 0.547) (S11 Fig).
To assess the transmission-to-sandfly (xenodiagnosis) potential of Leishmania-CAT parasites we allowed  Table 3. All 10 mice infected with WT parasites were infectious to sandflies, and (as an average of two experiments) 28.5% of sandflies have acquired parasites. In sharp contrast, only 3 out of 10 animals were infectious to sandflies in the CAT group with only 4% of sandflies acquiring Leishmania.
In addition, we also confirmed that CATALASE was retained by L. mexicana and expressed after passaging through sandflies and mice, ruling out the possibility that parasites that no longer express catalase had selection advantage in vivo (S13 Fig). Interestingly, its expression level was lowered compared to that of in vitro cultivated promastigotes, again supporting our argument that presence of catalase is detrimental for Leishmania in vivo.

Discussion
One of the core questions of evolutionary biology is the origin of parasitism. The transition from a free living to a parasitic life style was accompanied by the gain and loss of multiple genes (often whole protein gene families) in order to adapt to the new environment within the host [90]. Understanding this process requires to compare the free-living and parasitic species [91]. In this respect, kinetoplastids are an ideal model group because they unite obligatory parasitic trypanosomatids and their free-living relatives, bodonids [32,92]. A large number of kinetoplastid genomes have been sequenced and analyzed [2], facilitating our understanding of the basic differences between these groups of organisms. The comparative analyses of the dixenous and monoxenous parasites illuminate the key features, which have allowed the former to adapt to the environment of vertebrate hosts [22,93,94].
In terms of the complex parasitic lifestyle, a recently established subfamily Leishmaniinae is particularly interesting because it encompasses phylogenetically related monoxenous and dixenous trypanosomatids [95,96]. Interestingly, the catalase gene was secondarily lost from all dixenous members of the Leishmaniinae [22,26,32]. The fact that hydrogen peroxide has been proposed as a signaling molecule in Leishmania amastigote differentiation [28,29] prompted us to hypothesize that catalase may be not compatible with the dixenous life cycle of Leishmania spp. To investigate this, we expressed the enzyme from the monoxenous L. pyrrhocoris in dixenous L. mexicana. In order to do that we had to circumvent the methodological difficulty of expressing a protein throughout the developmental cycle of L. mexicana. Numerous previous attempts failed, mainly because all routinely used genetic manipulation techniques in Leishmania rely on the incorporation of exogenous UTRs. These genetic elements greatly impact on the expression of transgenic DNA sequences [42,97,98]. In this work, we have solved this issue by using the stably expressed LmxM.32.0792 Β-TUBULIN locus and a modified genetic system, based on the selfcleaving 2A peptide. This approach has rendered unnecessary the requirement to use exogenous UTRs. The choice of the locus was primarily determined by its stable transcription and translation, yet, in retrospect, the multiple copies of Β-TUBULIN gene made it more complicated than expected [77]. The 2A self-cleaving peptide was discovered 30 years ago [99] and since then it has been used in many species, including zebrafish, mice, silkworm, and algae [82,100]. In this work, the use of the P2A self-cleaving peptide in L. mexicana resulted in efficient expression of both the antibiotic resistance and catalase genes, integrated into the Β-TUBULIN locus.
We hypothesized that the loss of the catalase gene was a prerequisite for the dixeny in Leishmania [26]. It has been indirectly supported by studies in other dixenous trypanosomatids, T. brucei and T. cruzi, where catalase was found to have a disadvantageous effect on their vitality in vivo [30,31]. To investigate this in detail, we established a L. mexicana line that expressed

Data Availability Statement
The data that support the findings of this study are openly available in NCBI BioProject at https://www.ncbi.nlm.nih. gov/bioproject/, reference number BioProject PRJNA660365.