Clinical isolates were collected in Peru at the Institute of Tropical Medicine Alexander von Humboldt between 2001 and 2004 within the framework of the multiregional LeishNatDrug project (Yardley et al., 2006). Nine of them were selected for the present study, according to the following inclusion criteria: i) first episode of leishmaniasis, ii) cutaneous form of the disease, iii) no previous treatment before attending the clinic, iv) complete therapy with antimonial post-diagnosis, v) no concomitant disease, and vi) infecting species typed as Leishmania braziliensis. One strain of Leishmania lainsoni was included in the study as control: this is a line highly susceptible to antimonials, for which the existence of quiescence after PAT exposure was demonstrated (Jara et al., 2022). For each isolate, enhanced green fluorescent protein (EGFP) was integrated within the 18S ribosomal DNA locus, further called rEGFP with the use of the pLEXSY-hyg system (Jena Bioscience, Jena, Germany) as previously reported elsewhere (Bolhassani et al., 2011; Jara et al., 2022). After the generation of the transgenic rEGFP parasites, clones were obtained with the “micro-drop” method, as described elsewhere (Van Meirvenne et al., 1975; Jara et al., 2022). All the work presented here was made with these clones, further called rEGFP strains.

Promastigotes were maintained in complete M199 (M199 medium at pH 7.2 supplemented with 20% fetal bovine serum, hemin 5 mg/L, 50 μg/mL of hygromycin Gold, 100 units/mL of penicillin, and 100 μg/mL of streptomycin) at 26°C with passages done twice per week. To obtain axenic amastigotes, 1 mL of stationary promastigotes was centrifuged (1, 500×g, 5 min), the pellet was resuspended in 5 mL of complete MAA (M199 at pH 5.5, supplemented with 20% fetal bovine serum, glucose 2.5 g/L, 5 g of tryptic soy broth, hemin 5 mg/L, 25 μg/mL of hygromycin Gold, 100 units/mL of penicillin, and 100 μg/mL of streptomycin), and parasites were incubated at 34°C. After observing morphological evidence of amastigogenesis on day 3, the lines were subcultured every 4 days, and growth was monitored microscopically over 3 weeks. If axenic amastigotes were not able to proliferate, three subsequent attempts for axenization were performed: the generation of axenic amastigotes was successful for three rEGFP strains (PER094, 122, and 362). Samples for further experimental procedures were prepared in complete M199 or MAA in the absence of hygromycin. The growth curves were monitored by daily counting of parasites, and the generation time was calculated with the following formula: G = t/(logb − logB)/log2), where t = time interval in hours, B = number of parasites at the beginning of a time interval, and b = number of parasites at the end of the time interval.

Exponentially growing promastigotes were exposed to 9 µg/mL of potassium antimonyl tartrate (PAT) for 48 h after which an aliquot of 50 µL was sub cultured in a medium without drug pressure (post-PAT). For each strain, a control without exposure to drug pressure and instead treated with PBS was maintained in parallel (post-PBS). The cell cultures were monitored microscopically, and if positive cellular growth was observed, additional subcultures for the evaluation of drug susceptibility and pellets for sequencing were prepared. Throughout the paper, lineages that were exposed to PAT were referred to as “post-PAT,” while those that were not previously exposed to PAT were referred to as “post-PBS.”

The drug susceptibility was measured with the resazurin test. Briefly, exponentially growing parasites were plated into 96-well plates containing a serial dilution of PAT to reach final concentrations ranging from 83 µg/mL to 0.1 µg/mL. Each plate included controls without PAT and controls for monitoring proliferation and autofluorescence of the medium. Resazurin sodium salt (200 µg/mL, Sigma, Darmstadt, Germany) was added at two different time points: at the time of plating for the proliferation controls (T1) and 20 h post- drug pressure for the other wells (T2). After 4 h of incubation with resazurin, the fluorescence of its reduced form resorufin was recorded using the Victor X3 Multilabel Reader (Perkin Elmer, Waltham, USA) exciting at 560 nm and measuring emission at 590 nm. The test was considered valid for further analysis only if the fluorescence for the controls without drug pressure at T2 was at least 2- fold the fluorescence for the proliferation controls at T1. All experiments with the different lines were repeated three times with three technical replicates each. The blank-subtracted data expressed in relative fluorescence units (RFU) were exported to GraphPad Prism 8 to calculate the 50% inhibitory concentration (IC₅₀), using a sigmoidal dose–response model with variable slope. Statistical analysis and data visualization were performed in R Studio with built-in functions and ggplot2, respectively.

Exponentially proliferating cells (promastigotes and amastigotes) were exposed to PAT at concentrations ranging from 1 µg/mL to 83 µg/mL, and both the cell viability and the rEGFP expression were measured by flow cytometry at three time points: before drug exposure and after drug pressure at 24 h and 48 h. The rEGFP expression is a negative marker of quiescence: highly expressed in proliferative cells and downregulated during quiescence (Jara et al., 2019; Jara et al., 2022). The cell viability was evaluated by preincubating cells with the NucRed Dead 647 (Thermo Fisher Scientific, Waltham, USA) for the staining of dead cells and Vybrant Dye Cycle violet (Thermo Fisher Scientific) for the staining of DNA in all cells. The Vybrant Dye Cycle violet was used to select the subpopulation of cells having a healthy pattern for their DNA. A wild-type (non-rEGFP) strain and a non-stained sample were included as negative controls, together with a sample exposed to thermal shock as a positive control for cellular death. The samples were analyzed with a calibrated flow cytometer BD FACS Verse ™ in the medium flow rate mode. In order to compare the rEGFP relative fluorescence units (RFU) among the samples, the acquisitions were made with the same settings during all the experiments. The FCS files were analyzed with the FCS 5 Express Plus Research edition. The subpopulation of cells with good cell viability was selected by sequential gating. Briefly, single cells were selected by pulse geometry gate; a first gate was selected by plotting the SSC-W vs. SSC-H, and a second gate was created by plotting the FCS-H and FCS-A. Among the single cells, a third gate was created by plotting the rEGFP vs. the NucRed and selecting the subpopulation NucRed-negative. Finally, a fourth gate was created by plotting the rEGFP vs. the Vybrant Dye Cycle violet and selecting the subpopulation having a healthy DNA pattern. The cell viability for each sample was estimated by multiplying the percentage of cells NucRed-negative and Vybrant DyeCycle-positive. For each developmental stage of Leishmania and for each time point post-exposure to PAT, the effect of the drug pressure over the rEGFP expression in the subpopulation of viable cells was evaluated with a two-way ANOVA considering the PAT drug pressure and strains as explanatory variables. Subsequently, a Tukey’s HSD test was used to evaluate the statistical significance of differences between pairwise comparisons.

The measures of cell viability obtained after the flow cytometry assay over increasing concentrations of PAT were used to estimate the lethal doses of PAT that kill 10%, 25%, and 50% of the population (LD₁₀–LD₅₀). The LDs were estimated using the drc R package (Ritz et al., 2015). Briefly, the data were fitted using a log-logistic model with a lower limit at 0 and having as a formula the % of cell viability as explained by the concentration of PAT and the line of the parasite as the grouping factor. Finally, the desired LDs were estimated having as a reference the upper limit and setting the confidence interval to 0.95.

For each lineage (post-PBS and post-PAT), DNA was isolated from a sample maintained without drug pressure, and a sample recovered after 48 h of PAT drug pressure, using the QIAamp DNA Micro Kit (Qiagen, Hilden, Germany). The DNA concentration was assessed with the Qubit DNA broad-range DNA quantification kit (Thermo Fisher). Library preparation and sequencing were performed at BGI with a DNBSEQ-WGS-PCR free library index and the DNBSEQ PE150 platform, respectively. The FASTQ files containing paired reads of 150 bp were subsequently analyzed for the assessment of sequence variants and ploidy. Briefly, the quality of the reads was evaluated with SAMtools and the multiqc command. Samples were aligned against the reference genome of L. braziliensis MHOM/BR/75/M2904 (Van den Broeck et al., 2020). The per position mean depth and the breadth of coverage were estimated with SAMtools. Variants (SNPs and INDELs) were called jointly for all 18 samples with BCFtools and the mpileup command (Danecek et al., 2021). Subsequently, the multisample vcf file was filtered according to the following criteria: overall variant quality (QUAL) higher than 200, the individual read depth per locus (FMT/DP) higher than 5, and the average mapping quality (MQ) and genotype quality (GQ) higher than 40. The annotation of the genomic location for the variants was done with SnpEff (Cingolani et al., 2012). Somy values were estimated by using the median read depth across each chromosome as reported elsewhere (Dumetz et al., 2017). A maximum likelihood phylogenetic tree was reconstructed with IQ-TREE v1.6.12 (Nguyen et al., 2015) reusing the HKY+G substitution model and 1,000 ultrafast bootstrap approximation (Hoang et al., 2018).

For each of the nine rEGFP strains of L. braziliensis used, we measured the drug susceptibility of promastigotes of the corresponding post-PBS lineages with a resazurin test. Values ranged from 0.2 µg/mL to 7.48 µg/mL, with a median of 1.98 µg/mL ( Figure 1A and Supplementary Figure 1 ). We then exposed all lines to the same concentration of PAT (9 µg/mL, representing ~ 4-fold the median IC₅₀ of all nine strains) in order to evaluate the potential differences in their adaptation to survive. i) The potential cytostatic effect of PAT on Leishmania cells was monitored with the resazurin test by measuring the signal of its reduced form resorufin before the drug treatment and after 24 h of exposure to PAT or PBS. In the absence of the drug, the resorufin log2 FC signal ranged from 0.8 (PER206 EGFP Cl2) to ~ 3.2 relative fluorescence units (RFU, equivalent to 9- fold the initial signal, in PER122 EGFP Cl1), indicating that all strains were proliferating albeit at different rates. Under drug pressure, the resorufin log2 FC signal was significantly decreased compared with the one measured in the corresponding control without the drug in all strains suggesting that under our conditions of drug pressure, parasites have very diminished growth ( Figure 1B ). ii) After 48 h of exposure to 9 µg/mL of PAT, parasite death was observed in each lineage, but parasites were all able to resume their proliferation after subculturing in a fresh medium without a drug. This indicated that at least a proportion of the cells survived and maintained their capability to switch to a proliferative state once they were in optimal conditions. iii) We found that the susceptibility to the drug did not change in the post-PAT lineages (median IC_50 = 2.35 µg/mL; interquartile range, IQR = 2.54) when compared with the corresponding post-PBS lineages (median IC_50 = 2.00 µg/mL, IQR = 2.70, Mann–Whitney U test, P = 0.9, Figure 1A ). Noteworthy, the capacity of Leishmania to survive concentrations of the drug beyond the IC₅₀ of 9 µg/mL represents ~ 45 times the IC₅₀ for the most susceptible line PER348 EGFP Cl1 (IC_50 = 0.2 µg/mL).

The expression of rEGFP is a negative biomarker of quiescence: it is high in proliferative cells, while it decreases or disappears in quiescent cells (Jara et al., 2022). i) In promastigotes, the exposure of proliferating cells to PAT (9.0 µg/mL) led to a significant reduction in the levels of rEGFP expression (RFU) among viable cells of each line when compared with the control treated with PBS at both 24 h and 48 h ( Figure 2A ). Furthermore, rEGFP expression was also different among the evaluated strains (two-way ANOVA: P < 0.001 for PAT treatment and P < 0.001 for strain). The same significant effect of the drug was observed in amastigotes. ii) We then explored if the reduction in rEGFP expression changes in response to increasing concentrations of PAT. The results showed that higher doses of PAT induced a more profound reduction of the rEGFP expression compared with the control without PAT treatment at both 24 h and 48 h post-PAT treatment in both promastigotes and amastigotes ( Figure 2B ). The reduction in rEGFP expression follows a dose–response model where there is a linear response that reaches a plateau approximately 9.0 µg/mL of PAT with an average relative rEGFP reduction of 44% and 40% after 48 h of PAT in promastigotes and amastigotes, respectively. iii) The analysis of relative rEGFP reduction also shows differences among L. braziliensis strains (ranging between 23.7% and 56.2% in promastigotes and between 33.7% and 49.9% in amastigotes).

When measuring the drug susceptibility with standard fluorometric procedures such as the resazurin test at a single point in time, it is not possible to address with certainty if the drug has cytostatic or cytotoxic effects. The decreased signal of resorufin in samples with the drug pressure compared with the control could result from cellular death, arrested growth, or likely, a mixture of both events. Therefore, we evaluated the cell viability at the single-cell level by flow cytometry and calculated the lethal doses 10, 25, and 50 (LD₁₀, LD₂₅, and LD₅₀), which are the concentrations of the drug that kill 10%, 25%, and 50% of the population, respectively. i) In promastigotes, the kinetics of the cell viability over different concentrations of PAT pressure suggests variable cytotoxic effects among the evaluated lines. The LD₅₀ among all strains ranged from 5.98 µg/mL in PER206 EGFP Cl2 to 83.3 µg/mL in PER006 EGFP Cl3 ( Figures 3A, B ). The median LD₅₀ among the nine strains was 28.2 µg/mL, which is approximately 10- fold the IC₅₀ that was previously measured with the resazurin test. Noteworthy, four out of the nine lines had LD₅₀ higher than 30 µg/mL of PAT, a non-physiological concentration. These strains may be considered highly tolerant to PAT compared with the other strains at the same parasite stage. ii) As promastigotes showed high variability in their rates of survival to increasing concentrations of PAT, we further evaluated the potential relationship between the capability to adopt a quiescent state as measured by the reduction in rEGFP expression and the success to overcome the drug pressure as measured by their cell viability. After 24 h of PAT treatment, there was a weak positive relationship at low concentrations of PAT (1 µg/mL to 3 µg/mL of PAT) that became strong at 9 µg/mL and 28 µg/mL of PAT ( Figure 3C ). iii) Amastigotes showed remarkably higher survival rates to PAT compared with the promastigotes of the respective strains. After twice the time of exposure to PAT (48 h), the highest concentration of the drug could not kill 50% of the population, and the estimation of the median LD₂₅ for the three strains was 46.7 µg/mL ( Figures 3D, E ). Because these results with amastigotes of L. braziliensis were surprisingly high, we evaluated the LDs from a highly susceptible line of L. lainsoni. The same experimental setup estimated that in amastigotes, the LD₂₅ and LD₅₀ were 0.49 µg/mL and 0.97 µg/mL, respectively, ruling out that the L. braziliensis results would be an experimental artifact ( Supplementary Figure 2 ). Accordingly, amastigotes of L. braziliensis are intrinsically more tolerant to PAT pressure compared with promastigotes. Further work is required to know if this could be due to a deeper quiescence state in amastigotes.

We evaluated the genome stability between pre- and post-PAT lines by WGS, at the levels of somy, SNPs, and INDELs. i) At the somy level, we observed variability of karyotypes in a strain-specific manner. The number of aneuploid chromosomes/strain ranged from 1 to 10. Trisomy of chromosomes 5, 11, 25, and 29 and tetrasomy of chromosome 31 were among the most frequent occurrences being present in at least three out of the nine strains. For each strain, the original overall ploidy of the post-PAT lineage was identical to the corresponding post-PBS lineage ( Figure 4A ). ii) Compared with the L. braziliensis reference genome (MHOM/BR/75/M2904), a total of 163,823 high-quality SNPs and 17,055 high-quality INDELs were called across our panel of 18 lineages. A maximum likelihood phylogenetic tree based on SNPs shows that each post-PAT lineage clusters with its corresponding control, post-PBS lineage ( Figure 4B ), reflecting the high genomic similarity between post-PAT and post- PBS lineages. Indeed, each post-PAT lineage showed differences from its corresponding control lineage at only 11 to 208 SNP loci and 76 to 119 INDEL loci, the majority of which (73.4% SNPs and 93.1% INDELs) occurred in the non-coding region of the genome ( Supplementary Data ). None of these variants occurred in a homozygous state; in other words, all variants were either absent in the post-PBS lineage and heterozygous in the post-PAT lineage, or vice versa. For each strain, the genomic analysis indicates that PAT-exposed lines are genomically very similar to their parental line and that, although a limited number of mutations occur, they are not in genes known to be related to drug resistance in Leishmania.

Persisters represent another manifestation of quiescence and DT; in bacteriology, these correspond to a subpopulation of DT cells that can survive antibiotic treatment much better than the majority of the population (Barrett et al., 2019). We explored the occurrence of persister-like cells in promastigotes of L. braziliensis PER191 EGFP Cl5 under two conditions of prolonged stress (see Figure 5A for a description of the experimental design): i) by long-term stationary culture, where cells were maintained without subculturing and without PAT over 14 days (Sta D14), and ii) by exposing exponentially growing cells to PAT and maintaining the drug pressure (exposure 1) over 14 days (Log PAT D14). We reasoned that in these cultures majority of the cells would be dead, and only the ones with persister-like properties would remain. In both stress conditions, we tested the resilience of the survivors on day 14, by treating them (exposure 2) with PAT or PBS as untreated control, over 24 h; we measured viability, rEGFP expression, and reversibility to proliferative state.