The complex life cycle of the human parasite Leishmania mexicana requires rapid translational adaptation for survival in two distinct environments: the insect vector and the mammalian host. These protists lack conventional transcriptional control due to their unusual genome organization. Consequently, tRNA modifications may represent an additional mechanism for post-transcriptional regulation of gene expression. One such modification is queuosine (Q), which is incorporated at the anticodon wobble position 34 of specific tRNAs. Here, we demonstrate that Q-tRNA levels increase substantially during Leishmania differentiation from the insect stage to the mammalian-infective stage, implying an important role for virulence. Hence, we generated mutant cells lacking the enzyme responsible for Q incorporation, tRNA-guanine transglycosylase (TGT), which exhibited substantial changes in the proteome during differentiation in vitro. Specifically, downregulated proteins were enriched in NAU codons, whereas upregulated proteins predominantly contained NAC codons. Although LmxTGT knockout parasites exhibited normal growth and differentiation in vitro, they demonstrated impaired survival within macrophages and reduced pathogenicity in mice, highlighting the role of the Q-tRNAs under stress conditions. To our knowledge, we present here the first direct evidence that queuosine tRNA modification controls the infectivity of Leishmania via codon-biased translation. To date, gene expression regulation in Leishmania and other trypanosomatids has been attributed mostly to RNA stability and processing; however, our findings demonstrate that tRNA modifications also play a key regulatory role. Specifically, the Q-tRNA modification provides a novel layer of gene expression regulation, maintaining translational balance and supporting the parasite’s ability to adapt to changing environments, and contributing to Leishmania virulence.

Fig 1. LmxTGT2 is necessary for Q-tRNA formation.

(A) Schematic representation of the formation of queuosine (Q) tRNA modification in eukaryotes by the heterodimeric tRNA-guanine transglycosylase (TGT) complex, which catalyzes the replacement of guanine (g) with queuine (q) in four specific tRNAs (Created in BioRender. Paris, Z. (2026) https://BioRender.com/fyga397). (B) Growth curves of wild-type (WT), LmxTGT2-KO (KO), and add-back (AB) promastigote cells exhibit no differences under standard exponential growth conditions with regular dilution (n = 3). (C) Northern hybridization showing the effect of LmxTGT2 gene knock-out on Q-modified tRNA levels. Total RNA from WT, LmxTGT2-KO, and AB cells was separated on an APB affinity gel and hybridized with ³²P-labeled tRNA specific probes. As a result, Q-tRNAs migrate slower on the gel, leading to separation of modified (Q) and unmodified (G) tRNAs. Periodate oxidation (OX), which eliminates the mobility shift by oxidizing cis-diols (queuosine as well as terminal 3′ ribose), serves as a negative control. The SL RNA is used as a loading control. The numbers below the blot indicate the percentage of Q-modified tRNA, calculated by dividing the intensity of the Q band by the sum of the Q and G bands and multiplying by 100. The results are representative of three independent experiments. The underlying data can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003887.g001

Cell culture

The three cell lines wild-type (WT), LmxTGT2 knock-out (KO), and add-back (AB) strains of L. mexicana (isolate MNZC/BZ/62/M379) were grown in M199 medium (Sigma-Aldrich, Cat. No. M0393-10X1L), which was supplemented with 2 μg/ml biopterin (Sigma-Aldrich, Cat No. B2517-25MG), 2 μg/ml hemin (Sigma-Aldrich, Cat No. 51280-5G), 25 mM HEPES (Applichem, Cat. No. A3268), 100 units/ml penicillin, 100 μg/ml streptomycin (Sigma-Aldrich, Cat. No. P4333-100ML), 10% fetal bovine serum (FBS) (Sigma-Aldrich, Cat. No. F7524-500ML), and 100 μg/ml hygromycin (Invitrogen, Cat. No. ant-hg-1) at 25 °C. The medium used for the cultivation of the LmxTGT2-KO cell line was also supplemented with 100 μg/ml nourseothricin (Jena Bioscience, Cat No. AB-102L) and 50 μg/ml puromycin (Invitrogen, Cat No. ant-pr-5), serving as selection markers. To determine the cell density, samples were fixed with a LeishFix solution (3.6% formaldehyde, 150 mM NaCl, 15 mM Na₃C₆H₅O₇) and cell count was determined using a Neubauer hemocytometer (Sigma-Aldrich, Cat No. BR717805-1EA). Bone marrow-derived macrophages were cultured in DMEM/F12 medium (Invitrogen, Cat. No. 10565), supplemented with 100 units/ml penicillin and 100 μg/ml streptomycin (Sigma-Aldrich, Cat. No. P4333-100ML) and 10% FBS at 37 °C (prior to infection) or 34 °C (post-infection) and 5% CO₂.

Generation of LmxTGT2-KO and AB cell lines

The LmxTGT2-KO cell line was generated by CRISPR/Cas9 approach using the cloning strategy described by Ishemgulova and colleagues [35]. The LmxTGT2 gene specific single guide RNA (sgRNA) (sequence: GAAGATGGTGCAGGTAAGCG│CGG) was designed using the eukaryotic pathogen CRISPR guide RNA/DNA design tool (Peng and Tarleton, 2015; http://grna.ctegd.uga.edu), amplified by PCR using specific primers (S1 Table), and cloned into a modified pLEXSY-SAT2.1 vector (Jena Bioscience, Cat. No. EGE-274) containing a U6 promoter and a nourseothricin resistance marker. The vector was kindly provided by Prof. Vyacheslav Yurchenko (Life Science Research Centre, Faculty of Science, University of Ostrava). The repair template was PCR-amplified using corresponding oligonucleotides (S1 Table), generating a cassette containing the puromycin resistance gene flanked by 30 bp LmxTGT2 homology arms at the Cas9 cleavage sites. Ten micrograms of both the sgRNA construct and repair template were co-transfected into the L. mexicana Cas9-expressing cell line using the Amaxa Nucleofector program U-033. LmxTGT2 ablation was confirmed by PCR on genomic DNA using specific primers (S1 Table). The add-back cell line was generated using expressing a Cas9-resistant version of the LmxTGT2 gene cloned in a modified pLEXSY vector, which contains UTRs of the A600 gene, constitutively expressed throughout all parasite life stages.

Northern blot analysis

Total RNA was isolated by phenol–chloroform extraction as previously described [36]. For aminophenylboronic acid (APB) affinity electrophoresis, total RNA samples were deacylated by incubating them in 100 mM Tris (pH 9.0) for 30 min. To create an oxidation control, RNA was further treated with 2 mM NaIO₄ in 50 mM Tris (pH 5.0) in the dark for 2 hours at 37 °C, preventing APB binding by modifying the ribose hydroxyl groups. The oxidation reaction was subsequently quenched with 2.5 mM glucose before use. A 10 ml gel mixture (8 M urea, 8% polyacrylamide) was supplemented with 25 mg of APB (Sigma-Aldrich) before polymerization [37]. Electrophoresis of APB gels was conducted for ~6 hours at 75 V and 4 °C, followed by electroblotting onto Zeta-Probe membranes under standard transfer conditions. The membrane was then UV-cross-linked for 1 min. For hybridization, γ-32P ATP-labeled oligonucleotide probes were prepared using T4 Polynucleotide Kinase (NEB, Cat. No. M0201S) and purified using a Sephadex G-25 column. The probes were denatured at 98 °C for 5 min before being added to the hybridization buffer (5× SSC, 20 mM phosphate buffer, 7% SDS, 1× Denhardt’s solution, and 1 mg/ml salmon sperm DNA). The membrane was hybridized at 48 °C overnight, then washed twice at 48 °C (Wash I: 3× SSC, 5% SDS, 25 mM NaH₂PO₄, 10× Denhardt’s solution; Wash II: 1× SSC, 1% SDS). Hybridized membranes were exposed overnight to a phosphor-imager screen (GE Healthcare) and analyzed using a Typhoon FLA 9000 scanner with ImageQuant TL software (GE Healthcare). Probes utilized for Northern hybridization are listed in S1 Table.

In vitro differentiation

To induce differentiation of L. mexicana from the procyclic promastigote stage into infective metacyclic forms, cultures were initiated at a density of 2 × 10⁵ cells/ml and maintained in M199 medium (pH 7) supplemented with 10% FBS for 10 days without dilution. Axenic amastigote forms were obtained by transferring the metacyclic culture into SDM medium containing 20% FBS and incubating at 32 °C for an additional four days [38]

qPCR

To assess stage-specific gene expression by quantitative reverse transcription PCR (qPCR), RNA was harvested on days 0, 8, 10, 13, and 17 of differentiation. Reverse transcription was performed using the QuantiTect Reverse Transcription Kit (Qiagen, Cat. No. 205311). cDNA was then used for qPCR with 60S primers as a control, PfrD1 primers, Sherp primers, and Amastin primers (S1 Table). Reactions were carried out using Power SYBR Green PCR Master Mix (Thermo Fisher Scientific, Cat. No. 4367659) according to the manufacturer’s instructions and executed on the QuantStudio 6 PCR system (Thermo Fisher Scientific).

Macrophage isolation and in vitro infection

Bone marrow-derived macrophages were isolated from BALB/c mice as described previously [39]. Murine macrophages were infected with L. mexicana parasites following a previously described protocol, with minor modifications [39]. Prior to infection, differentiated metacyclic L. mexicana were labeled with Cell Tracker^TM Orange CMRA dye (Invitrogen, Cat No. C34551) as described by Michael and colleagues [40]. Macrophages were primed with IFN-γ (5 ng/ml; Invitrogen, Cat. No. PMC4031) 24 hours prior to infection and subsequently activated with a combination of IFN-γ (5 ng/ml) and LPS (10 ng/ml; Sigma-Aldrich, Cat. No. L2630-10MG). Subsequently, macrophage cultures were exposed to parasites at a 1:7 ratio in a final volume of 2 ml for 4 hours at 34 °C, 5% CO2. The noninternalized parasites were removed with four successive washes with 1x PBS and subsequently replaced with fresh medium. This point was considered the initial time of infection (day 0). Cultures of infected macrophages were incubated for up to 5 days under the same conditions. To assess infection rates, cells were seeded onto coverslips placed in 6-well plates. At designated time points (days: 0, 1, 2, 3, and 5), coverslips were removed, and cells were fixed and permeabilized with ice-cold methanol for 10 min. Following fixation, cells were washed three times with 1x PBS to remove residual methanol. Subsequently, cells were stained with DAPI (Thermo Fisher Scientific, Cat No. P36941). Infection rates were determined by counting 100 cells per group under a fluorescence microscope, identifying nuclei stained with DAPI and parasites labeled with CMRA dye (Thermo Fisher Scientific, Cat No. C34551).

Dual luciferase reporter constructs design and assay

Renilla and Firefly luciferases were encoded within a single open reading frame and expressed as an in-frame fusion protein from one mRNA. While the Renilla luciferase (Rluc) sequence remained unaltered as an internal control, the cognate codons of Q-tRNAs in Firefly luciferase (Fluc) were either left unchanged (50% NAC, 50% NAU) or modified to exclusively C-ending (100% NAC) or U-ending (100% NAU) codons. These two recoded constructs were synthesized by Eurofins Genomics. All three reporter constructs were cloned into the pLEXSY vector under the regulation of the A600 UTR and introduced into L. mexicana WT and LmxTGT2-KO cell lines via electroporation. Positive clones were selected using puromycin. The reporter assay was performed using Promega Dual Luciferase kit, as per manufacturer’s protocol. Briefly, 2 × 10⁴ cells were washed with PBS and resuspended in Passive Lysis Buffer (PLB) in a 96-well transparent flat-bottom plate. The plate was incubated at room temperature for 15 min with continuous shaking. Firefly luciferase activity (FLuc) was measured following the addition of the LARII substrate using a Tecan Spark luminometer. Subsequently, the Stop and Glo reagent, which quenches Firefly luminescence while activating Renilla luminescence, was added, and RLuc activity was recorded. The FLuc/RLuc ratio was calculated, and values were normalized to the control construct (50% NAC + 50% NAU).

Proteomics

WT, LmxTGT2-KO, and AB cells were collected at distinct differentiation time points (days: 0, 8, 10, 13, and 17) and analyzed at the CEITEC Proteomics Facility, Brno, Czech Republic. Briefly, proteins extracted in SDT buffer were processed using filter-aided sample preparation as described elsewhere [41]. The resulting peptides were analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) using the nanoElute system (Bruker) coupled to the timsTOF Pro 2 spectrometer (Bruker). Prior to LC separation, tryptic digests were online concentrated and desalted using a trapping column (Acclaim PepMap 100 C18, Thermo Fisher Scientific) and eluted onto an analytical column (Aurora C18, Ion Opticks) via a 90-min linear gradient. MS/MS data were acquired in data-independent acquisition (DIA) mode with a precursor isolation range of m/z 400–1,012. DIA data were processed using DIA-NN (version 1.8.1) [42] in library-free mode against the UniProtKB L. mexicana proteome database (UP000007259). Protein MaxLFQ intensities were further processed in a containerized computational environment (OmicsWorkflows, version 4.7.7a), following standard preprocessing, log2 transformation, and differential expression analysis using LIMMA. Mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium via PRIDE [43] partner repository with the dataset identifier PXD061534.

Proteomic data analysis and GO enrichment methodology

Proteomic data analysis was performed using R for visualization, including UpSet plots, principal component analysis (PCA), and Gene Ontology (GO) term plots. Differentially expressed proteins were identified by comparing LmxTGT2-KO with WT and AB across multiple time points, focusing on those consistently up- or downregulated during differentiation. GO enrichment analysis, conducted using TriTrypDB (https://tritrypdb.org), assessed the functional impact of these proteins by mapping GO terms to relevant biological processes.

Determination of codon bias in proteomics data

To evaluate the impact of codon bias on protein expression, proteins significantly up- or downregulated in the proteomics data were analyzed for codon usage of tyrosine (Tyr), asparagine (Asn), aspartate (Asp), and histidine (His). The corresponding mRNA sequences were retrieved from the NCBI database using UniProt IDs. For each coding sequence (CDS), the number of Q-containing codons for each amino acid was calculated as the “observed codon frequency”. The Z-score for codon usage was determined using the formula: z-score of codon = (Observed frequency of codon − Expected frequency of codons)/ Standard deviation of observed frequency of codon, where the expected frequency represents the genome-wide average frequency of the respective codon. Genome codon frequencies were obtained from DNA Hive [44]. Analysis was performed in RStudio (version 4.3.3), with ggplot2 used for data visualization.

Ethical statement

Mice were housed and handled in the animal facility of the Biology Centre in accordance with institutional guidelines and Czech legislation (Act No. 246/1992 and 359/2012). All experiments were approved by animal ethics committee and conducted under permit No. PP 56_2022P from the Czech Ministry of Environment.

Mice infection

Six-week-old BALB/c female mice were subcutaneously inoculated in the right ear with 1 × 10⁶ of WT, LmxTGT2-KO, and AB strains of metacyclic L. mexicana. M199 medium, serving as a blank control, was injected into the ears of the control group mice. Six mice per group were used for the infection study. Ear thickness was measured using a Teclock standard-type thickness gauge, with a range of 0–20 mm and a scale interval of 0.01 mm. Lesion size was monitored weekly. To determine the lesion thickness, the thickness of the uninfected ear was subtracted from the thickness of the infected ear for each experimental animal.

Antibody ELISA

To prepare the antigen, L. mexicana parasites were washed three times with 1× PBS and subjected to three freeze-thaw cycles, alternating between liquid nitrogen and a 37 °C water bath. The protein concentration was determined using a BCA assay (Pierce BCA Protein Assay Kit, Thermo Fisher Scientific). Medium-binding 96-well plates were coated with 5 μg/ml of freeze-thawed L. mexicana antigen and incubated overnight at 4 °C. The following day, plates were washed with PBS-Tween (0.05% Tween-20 in 1× PBS, pH 7.4) and blocked with 5% milk in PBS-Tween for 1 hour at 37 °C. Blood samples were collected from infected BALB/c mice via submandibular vein bleeding every two weeks. Serum was separated by centrifugation and used for the antibody ELISA assay. After washing with PBS-Tween, serum samples were added in duplicate to the first well, followed by serial dilution across the plate. The plates were incubated for 2 hours at 37 °C, followed by washing. The plates were then incubated with HRP-conjugated antibodies IgG1 (BD Biosciences, Cat#559626, clone X56) or IgG2a (BD Biosciences, Cat#553391, clone R19-15) at a final concentration of 0.2 μl in 1 ml of PBS-FBS (3:1) for 1 hour at 37 °C, washed again, and incubated with TMB solution (Fisher Scientific) until color change was observed. The reaction was stopped with 5% phosphoric acid. Absorbance was measured at 450 nm using a Molecular Devices SpectraMax M3 microplate reader, and SoftMax Pro software was used to determine antibody titers.

Parasitic burden

For in vivo studies, ears and draining lymph nodes were collected at the time of harvest (12 weeks post-infection for BALB/c mice). The tissues were homogenized using a cell strainer in 3 ml of Schneider’s Insect medium (Gibco, USA) supplemented with 20% heat-inactivated FBS and 1% penicillin/streptomycin. Each ear was separated into two sheets of dermis, washed in PBS containing 2% penicillin/streptomycin, and processed into small pieces. The tissue was then incubated in HBSS (Sigma-Aldrich, Cat: H9394) supplemented with 2 mM EDTA, 2% FBS, and 1% penicillin/streptomycin at 37 °C for 30 min, shaking at 250 rpm. After centrifugation, the cell pellet was incubated in DMEM + 2 mg/ml Collagenase A + 5% FBS + 1% penicillin/streptomycin at 37 °C for 1 hour, shaking at 250 rpm. The enzymatic reaction was stopped by adding FBS, and the tissue was mashed through a 70 μm strainer, then washed with DMEM + 20% FBS + 1% penicillin/streptomycin. The resulting lymph node or ear cell suspensions were serially diluted (1:20) in duplicates in two 96-well plates, ensuring that each sample was diluted across 24 wells. After 7 days of incubation at 26 °C, the plates were examined using an inverted microscope at 40x magnification. Parasite burden was determined from all six mice in the WT group, whereas four mice were available for the LmxTGT2-KO and AB groups due to loss of samples during processing of the mouse tissue to obtain axenic L. mexicana cells. The values reported in the graphs represent the highest logarithmic dilution showing viable parasites.

The LmxTGT2 subunit is essential for the formation of Q-tRNAs in L. mexicana

We identified the LmxTGT1 (LmxM.29.1800) and LmxTGT2 (LmxM.27.0030) genes in the genome of L. mexicana, which show 67% and 42% identities to their orthologs from T. brucei, respectively. To study Q-tRNA modifications in L. mexicana, we focused on the LmxTGT2 subunit, as our previous studies in T. brucei demonstrated that both TbTGT1 and TbTGT2 subunits are essential for Q-tRNA formation [24]. We generated a gene knockout of the LmxTGT2 subunit in the promastigotes of L. mexicana using CRISPR/Cas9 (S1A Fig). The absence of the LmxTGT2 gene was confirmed by PCR analysis of genomic DNA (S1B Fig). Subsequently, an AB cell line was constructed using the pLEXSY vector containing a recoded Cas9-resistant copy of the LmxTGT2 gene flanked by the 5′ and 3′ untranslated regions (UTRs) of the A600 gene (LmxM.33.3640), which was previously reported to maintain stable expression throughout Leishmania differentiation [45].

To assess the effect of LmxTGT2 depletion, we monitored growth of the cultured procyclic promastigote stage, which showed no significant differences in growth rates among the WT, LmxTGT2 KO, and LmxTGT2 AB strains (Fig 1B). Next, we investigated the involvement of LmxTGT2 in the formation of the Q modification of tRNAs. Total RNA from WT, KO, and AB cell lines was analyzed using acrylamidophenyl boronic acid (APB)-affinity electrophoresis, followed by Northern blotting [37], which allows separation of the Q-modified tRNAs from their unmodified counterparts. The steady-state levels of Q-tRNA modifications in WT procyclic promastigotes varied among different tRNAs: Q-tRNA^Asp, Q-tRNA^Asn, and Q-tRNA^Tyr showed modification levels ranging from 42% to 66%, whereas Q-tRNA^His was close to the detection limit of the assay.

The complete absence of the bands corresponding to Q-modified tRNAs in the LmxTGT2 KO strain, along with a recovery of 44%–83% of WT levels (depending on the tRNA species) observed in the AB strain (Fig 1C), illustrates the direct role of LmxTGT2 in Q-tRNA formation.

Q-tRNA levels increase during the differentiation from the promastigote to the amastigote stage

To determine whether the levels of Q-tRNA modification change during the life cycle and whether its depletion affects the development of the parasite, we induced differentiation in the axenic culture. To simulate differentiation in vitro, promastigote cultures are left proliferating until they reach stationary phase and transform into metacyclic promastigotes. Subsequently, cells are transferred into Schneider’s insect medium (pH 5.5) and incubated at 32 °C, which induces further differentiation into the axenic amastigote stage [38] (Fig 2A). However, the absence of Q-tRNA did not result in any noticeable changes in the growth or morphology of the LmxTGT2-KO parasites when growth was assessed daily up to day 13, within the same flask, and compared to the WT and AB strains (Fig 2B).

Although no obvious defects were observed during the differentiation in LmxTGT2-KO or AB, we validated the developmental progress by detecting stage-specific markers. As expected, expression of the amastigote-specific marker amastin increased during the differentiation into amastigotes. No significant changes in amastin expression were observed across the different cell lines (Fig 2C). Similarly, no cell line-specific differences were observed in the expression of the other stage-specific markers (S2 and S3 Figs). Collectively, these findings imply that although Q-tRNA modification increases during the life cycle, the absence of LmxTGT2 and subsequent lack of this modification does not substantially impact the differentiation process in vitro.

During the differentiation, we examined variations in the Q-tRNA modification steady-state levels across different life stages of the WT. Using APB electrophoresis of the total RNA, we assessed the Q content of tRNA^Tyr, tRNA^Asn, tRNA^Asp, and tRNA^His at various differentiation time points: day 0, day 10 (metacyclic promastigotes), day 13 (early amastigotes), and day 17 (amastigotes). The results revealed a global increase in the Q-tRNA modification levels throughout the life cycle, peaking at the amastigote stage. The increase was the most prominent in the tRNA^Tyr, and less striking but still prominent in tRNA^Asp and tRNA^Asn (Fig 2D and 2E). Similar to the previous result (Fig 1), Q modification in tRNA^His remained almost undetectable.

LmxTGT2-KO compromises the ability of L. mexicana to infect macrophages in vitro

Although the lack of Q-tRNAs did not affect cell proliferation in the axenic cell culture, the levels of Q-tRNAs changed during differentiation, suggesting a possible importance in the infectious stages of the parasite. Thus, we hypothesized that the absence of Q-tRNAs could be more relevant if the parasites were subjected to a challenging environment within their mammalian host cells. To test the potential impact of LmxTGT2-KO on the infectivity of the parasites, we conducted an in vitro infection of bone marrow-derived murine macrophages. The differentiated parasites were fluorescently labeled, and infection was monitored over five days following the initial infection. At the start of the experiment, 60%–70% of macrophages were infected (day 0) across all groups. However, the difference between the three tested groups became significant from day 2 to day 5 of infection (Fig 3A and 3B). Over time, the number of macrophages infected with LmxTGT2-KO parasites decreased substantially, resulting in only 11% infected macrophages on day 5, compared to 41% in the WT group (P = 0.0157) (Fig 3B). Correspondingly, the number of parasites detected inside a macrophage was comparable at the beginning of the infection but progressively decreased for LmxTGT2-KO parasites. WT parasites maintained higher intracellular parasite numbers, whereas the AB line displayed an intermediate phenotype and was not significantly different from either WT or KO at later time points (Fig 3C). Overall, these data indicate that LmxTGT2-KO parasites can initially infect macrophages at the same level as WT, but they are unable to survive inside the host cells for several days.

Q-tRNA is important for cytosolic codon-biased translation

To investigate the role of Q in cytosolic codon-biased translation, we employed the dual luciferase reporter system, where the Renilla luciferase (RLuc) is fused to the Firefly luciferase (FLuc) with a linker (Fig 4A) [46]. Since these enzymes use different substrates, it is possible to quantify differential expression of these proteins simultaneously, in a luminescence assay. We generated codon re-engineered constructs, where the RLuc sequence was kept unaltered, but in the FLuc sequence, the cognate codons for all four Q-containing tRNAs (tRNA^His_GUG, tRNA^Asp_GUC, tRNA^Asn_GUU, tRNA^Tyr_GUA) were changed either to all NAC (100% NAC) or all NAU (100% NAU) codons, as opposed to the control 50% NAC + 50% NAU FLuc, which contained approximately equal number of NAC and NAU codons. The total numbers of NAC and NAU codons are enumerated in Fig 4B. We hypothesized, based on available literature and our previous results in T. brucei, that Q might be more efficient than G, in decoding of U-ending codons [24]. Thus, in the absence of Q-tRNAs we might observe a decreased translation of NAU luciferase, and subsequently, a decrease in the luminescence. We transfected these three constructs into WT and LmxTGT2-KO cells and measured Fluc luminescence, normalizing it to Rluc. The absence of LmxTGT2 resulted in an overall defect in translation, since luminescence dropped to 75% (P = 0.01) in KO cells with the control construct (50% NAC + 50% NAU). While the NAC construct did not show significant differences in signal between WT and LmxTGT2-KO cells, the NAU construct exhibited a significant decrease in Fluc luminescence in the absence of LmxTGT2 (Fig 4C). This suggests that NAU codons may be translated less efficiently without Q-tRNAs, as hypothesized above.

The loss of Q‐tRNA modification alters proteome of L. mexicana during differentiation

Since the luciferase assay indicated an effect of Q-tRNAs on codon-biased translation, we performed a proteomic analysis to investigate the consequences of Q-tRNA depletion on the proteome during the L. mexicana life cycle. This analysis had two main objectives: first, to identify differentially expressed proteins in LmxTGT2-KO relative to WT and AB at different time points; and second, to examine proteins that showed consistent changes throughout differentiation in LmxTGT2-KO (relative to day 0). Our results revealed significant differences in proteome composition, particularly on days 8, 13, and 17 of the differentiation process (Fig 5).

We performed PCA on the proteomics data collected across all time points (days 0, 8, 10, 13, and 17) to assess global proteomic changes during differentiation and evaluate the relationship between the WT, LmxTGT2-KO, and AB cell lines (Fig 5A). The first principal component (PC1), which accounted for 37% of the total variance, primarily reflected progression through in vitro differentiation. Samples from all cell lines shifted along PC1 over time. All three cell lines clustered tightly on day 0, indicating highly similar baseline proteomic profiles prior to differentiation. As differentiation progressed, WT and AB samples continued to cluster closely together, whereas KO samples diverged from this trajectory. Notably, the AB line consistently aligned with the WT line at later time points, demonstrating the restoration of a WT-like proteomic profile with the reintroduction of LmxTGT2. These data suggest that the observed proteomic differences in KO are differentiation-dependent and specifically attributable to the loss of LmxTGT2.

In terms of the overall proteome, we identified 7236 proteins. In KO relative to WT and AB, we observed 168 proteins downregulated on day 8, 72 proteins on day 13, and 101 proteins on day 17. In contrast, fewer proteins were upregulated, with a peak of 36 upregulated proteins on day 8, followed by 9 proteins on day 10, 4 proteins on day 13, and just 2 proteins on day 17 (2-fold change, P = 0.05; S4, S5B Figs). Notably, there were not many common proteins among different time points. Interestingly, only three proteins LmxTGT1, LmxTGT2, and an uncharacterized protein, were consistently downregulated across all time points, forming the only overlap between all groups (Fig 5B).

To assess proteomic alterations within LmxTGT2-KO during differentiation, we analyzed proteins that were consistently downregulated or upregulated relative to the LmxTGT2-KO starting point (day 0), while remaining unchanged in WT and AB. Between day 8 and day 17 of differentiation, a total of 344 proteins were significantly downregulated, while 131 proteins were significantly upregulated (2-fold change, P = 0.05).

To investigate the potential function of these downregulated proteins in LmxTGT2-KO cells, we performed the GO enrichment analysis. Overall, 313 proteins were mapped to GO terms, while 22 GO terms were excluded due to having only one associated protein. The most significantly enriched GO terms were related to RNA metabolism and processing, including RNA metabolic processes, RNA modification, ribosome biogenesis, and noncoding RNA processing, suggesting a disruption in RNA processing. Additionally, processes related to gene expression regulation, post-transcriptional control, and methylation were notably affected (Fig 5C).

To further investigate the impact of Q-tRNA on translation, we examined the correlation between changes in protein expression and the usage of Q-specific codons, focusing on NAC and NAU codons. We observed a slight positive correlation between NAC codon frequency and protein levels on both day 8 (r = 0.166, P = 2 x 10⁻¹⁶) and day 13 (r = 0.149, P = 2 x 10⁻¹⁶). In contrast, NAU codons showed a trend towards negative correlation on day 8 (r = −0.005, P = 0.653) and day 13 (r = −0.043, P = 0.000228) (S5A, S6A Figs).

Since the effect of Q-modification may be specific to individual tRNAs and could be obscured in the broader analysis of NAC and NAU codons, we next focused on the specific codons involved. This analysis concentrated on amino acids decoded by Q-tRNAs, specifically tyrosine (Tyr), asparagine (Asn), aspartate (Asp), and histidine (His). Codon usage deviations were evaluated using the Z-score [47], which compares the observed frequency of specific codons with their expected frequency based on genome-wide codon usage. The analysis revealed significant differences in the usage of Asp codons in the proteome of day 8, since GAU codon is enriched in downregulated proteins (Fig 6B). Furthermore, Tyr codons showed pronounced enrichment of UAU in downregulated proteins, and enrichment of UAC in up-regulated proteins on day 13 (Fig 6C). Differences in codon usage at other time points were less pronounced. In general, Q-tRNA-dependent U-ending codons were enriched in downregulated proteins, whereas C-ending codons were more prevalent in up-regulated proteins, in line with the original hypothesis (S5 and S6 Figs).

The ablation of LmxTGT2 impacts parasite infectivity and virulence in the mammalian host

To determine whether the decreased infection observed in macrophages could be replicated in vivo and prompted by the profound effects on the LmxTGT2-KO proteome, an infection study was performed in Balb/c mice to investigate a role of LmxTGT2 in L. mexicana virulence. Parasites were administered intradermally into an ear of each mouse. Lesions became apparent after 8 weeks in all groups and continued to increase in size. From the ninth week onward, notable variations in lesion size were observed, with mice infected with LmxTGT2-KO parasites developing lesions approximately half the size of those caused by WT and AB strains. This difference became statistically significant from week 10 onward (Fig 7A).

The assessment of parasite load through a serial dilution assay did not reveal significant changes in the number of parasites in the ear, despite the larger lesions observed in the WT and AB groups compared to LmxTGT2-KO (Fig 7B). Interestingly, a lower number of parasites was detected in the lymph nodes in LmxTGT2-KO compared to WT and AB (Fig 7C). To evaluate the inflammatory response in infected mice, we performed an enzyme-linked immunosorbent assay (ELISA) for IgG1 and IgG2a in the serum. Throughout the infection period, IgG1 levels steadily increased in all tested groups (Fig 7D). In contrast, IgG2a titers were significantly lower in LmxTGT2-KO animals compared to WT and AB mice (Fig 7E). The ratio of IgG2a to IgG1 was calculated to be significantly reduced in mice infected with the LmxTGT2-KO strain. This ratio is an important indicator of the immune response profile, where a higher IgG2a/IgG1 ratio typically reflects a Th1-biased (protective) immune response, whereas a lower ratio indicates a Th2 (anti-inflammatory) response [48,49] (Fig 7F).

Fig 8. Model illustrating the role of queuosine (Q) tRNA modification in the virulence of L. mexicana.

Q-tRNA levels increase significantly during the differentiation from the procyclic promastigotes (insect stage) to the axenic amastigotes (mammalian stage), underscoring the importance of Q-tRNA modification for parasite adaptation and infectivity. Loss of Q modification via LmxTGT2 knockout disrupts codon-biased translational control, leading to substantial changes in the proteome. Specifically, downregulated proteins were enriched in NAU codons decoded by Q-tRNA, while upregulated proteins predominantly contained NAC codons. The resulting codon-specific imbalance impairs the parasite’s ability to establish infection and maintain virulence. These findings demonstrate that Q modification functions as a novel post-transcriptional regulatory mechanism that fine-tunes translation during Leishmania life cycle, compensating for the lack of classical transcriptional control and promoting adaptation to the host environment.

https://doi.org/10.1371/journal.pbio.3003887.g008

S1 Fig. Design and confirmation of LmxTGT2 knockout in L. mexicana.

(A) Schematic illustration of the strategy used to generate and validate LmxTGT2 knockout (KO) cells. Positions of PCR primers used for genotyping are indicated (Created in BioRender. Paris, Z. (2026) https://BioRender.com/ceb39l3). (B) PCR-based amplification followed by DNA electrophoresis reveals the presence or absence of specific genomic sequences in wild-type (WT) and knockout (KO) cells. Product sizes (in base pairs) are indicated to confirm the specificity of the knockout strategy.

https://doi.org/10.1371/journal.pbio.3003887.s001

(TIFF)

S4 Fig. Overlap of up-regulated proteins during in vitro differentiation in LmxTGT2-KO parasites.

UpSet plot showing the number of up-regulated proteins at different time points of differentiation in LmxTGT2-KO relative to WT and AB. Bars indicate the size of each intersection, and connected filled circles denote the time points that share common proteins. When no bar is present for a given combination, no common proteins were identified among those groups. Differential expression was defined as a fold change of ≥2 with statistical significance, as determined by Student t test (P < 0.05, n = 3). The underlying data can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003887.s004

(TIFF)

S5 Fig. Regulation of translation via codon bias in the absence of LmxTGT2 (day 8).

(A) Pearson correlation between NAC and NAU codon frequency and protein abundance in LmxTGT2-KO cells on day 8, relative to WT and AB. The Pearson correlation coefficient (r) with P value and linear regression line (black) are indicated. (B–D) Codon frequency of Tyr, His, and Asnshown as Z-scores, in upregulated and downregulated proteins in LmxTGT2-KO cells on day 8, relative to wild-type (WT) and add-back (AB). (E) Codon frequency of all NAC and NAU codons, shown as Z-scores, in upregulated and downregulated proteins on day 8. The center line indicates the median, the box shows the interquartile range (upper and lower quartiles), and individual dots represent outliers (minimum and maximum values). The data are based on three independent biological replicates used for proteomic analysis and were analyzed using Student t test. Significant differences are indicated as follows: (**P < 0.01, * P < 0.05, ns = not significant). The underlying data can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003887.s005

(TIFF)

S6 Fig. Regulation of translation via codon bias in the absence of LmxTGT2 (day 13).

(A–C) Codon frequency of Asp, His, and Asn codons, shown as Z-scores, in upregulated and downregulated proteins in LmxTGT2-KO cells on day 13, relative to wild-type (WT) and add-back (AB). (D) Codon frequency of all NAC and NAU codons, shown as Z-scores, in upregulated and downregulated proteins in LmxTGT2-KO cells on day 13, relative to wild-type (WT) and add-back (AB). The center line indicates the median, the box shows the interquartile range (upper and lower quartiles), and individual dots represent outliers (minimum and maximum values). The data are based on three independent biological replicates used for proteomic analysis and were analyzed using Student t test. Significant differences are indicated as follows: (**P < 0.01, * P < 0.05, ns = not significant). The underlying data can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003887.s006

(TIFF)

S1 Data. Raw measurements underlying the quantitative analyses presented in Figs 1B, 1C, 2B–2E, 3B, 3C, 4C, 5A–5C, S4, 6A–6C, S5A, S5B–S5E, S6A–S6D, 7A–7F, S2A–S2C, and S3A.

https://doi.org/10.1371/journal.pbio.3003887.s008

(XLSX)