Long-term hematopoietic stem cells trigger quiescence in Leishmania parasites

Addressing the challenges of quiescence and post-treatment relapse is of utmost importance in the microbiology field. This study shows that Leishmania infantum and L. donovani parasites rapidly enter into quiescence after an estimated 2–3 divisions in both human and mouse bone marrow stem cells. Interestingly, this behavior is not observed in macrophages, which are the primary host cells of the Leishmania parasite. Transcriptional comparison of the quiescent and non-quiescent metabolic states confirmed the overall decrease of gene expression as a hallmark of quiescence. Quiescent amastigotes display a reduced size and signs of a rapid evolutionary adaptation response with genetic alterations. Our study provides further evidence that this quiescent state significantly enhances resistance to treatment. Moreover, transitioning through quiescence is highly compatible with sand fly transmission and increases the potential of parasites to infect cells. Collectively, this work identified stem cells in the bone marrow as a niche where Leishmania quiescence occurs, with important implications for antiparasitic treatment and acquisition of virulence traits.


Introduction
Visceral leishmaniasis (VL) is a lethal neglected tropical disease caused by the obligate intracellular protozoan Leishmania [1,2] and transmitted through the bites of infected female phlebotomine sand flies [3,4].Leishmania parasites alternate between two main morphological forms during their life cycle: a long flagellated extracellular promastigote within the sand fly and a non-flagellated obligate intracellular amastigote in the vertebrate host that resides within the monocyte-derived cells of the liver, spleen and bone marrow (BM) and eventually causes lifethreatening complications [5][6][7][8].
The current antileishmanial drugs have many disadvantages and post-treatment relapse rates are increasing [9].In many instances, relapse does not relate to reinfection, drug quality, drug exposure or resistance [10], but is rather due to persistence for which mechanistic information is lacking.Persistent infections can occur in a variety of host sanctuary tissues or cellular niches, such as hepatocytes (Plasmodium vivax), skeletal muscle and neurons (Toxoplasma gondii), adipose tissue (Trypanosoma brucei and T. cruzi) and the BM (Mycobacterium tuberculosis) [11][12][13][14][15].Long-term hematopoietic stem cells (LT-HSC) in the BM were identified as a relapse niche for VL infection.LT-HSC become readily infected with extreme parasite burdens accompanied with low reactive oxygen species (ROS) and nitric oxide (NO) levels and a specific Stemleish transcriptional profile [16].A recent dual-scRNA-seq analysis in a chronic L. donovani infection model corroborated the proportional importance of HSC, identifying them as the main parasitized cell type in the bone marrow [17].
Besides persistence linked to cellular niches, treatment failure can also be associated with the adaptive behavior of parasites.In response to stress, certain microorganisms employ socalled quiescence to increase their chances of survival [18].The quiescent state is characterized by a lowered metabolic activity and renders a microorganism tolerant to antibiotics at the expense of becoming non-proliferative [19,20].Hence quiescent cells are phenotypic variants of the wildtype, and their dormancy can be reversed when stressors are alleviated.Given its discernible clinical implications, microbial quiescence has gained considerable interest and has been the subject of intense research for certain pathogens, especially bacteria [19][20][21][22][23].In contrast, quiescence in Leishmania has only recently been discovered and its role in drug tolerance, infection relapse and general parasite biology remains poorly understood.A recent study demonstrated that Leishmania quiescence can be induced by various triggers (e.g.antimonial drug pressure or stationary phase growth [24]).Furthermore, transcriptomic and metabolomic analyses of quiescent stages corroborated an overall downregulation of biosynthetic processes as a hallmark of quiescence [24].Despite these important insights, the molecular determinants orchestrating the phenotypic transition to the quiescent state in Leishmania remain thus far unknown.
The present study started with the observation that visceral Leishmania amastigotes inside LT-HSC rapidly enter a quiescent state.Although the induction is unrelated to drug pressure, we demonstrate that these quiescent parasites benefit from an enhanced survival of antileishmanial treatment.To better understand the molecular basis underlying amastigote quiescence, we performed an unbiased total RNAseq, revealing an overall decreased gene expression as a hallmark of parasitic quiescence in the LT-HSC niche.In addition, we show that transitioning through a quiescent state has a profound impact on parasite infectivity and transmissibility.

PLOS PATHOGENS
The results provide important information on the in situ acquisition of quiescence and its downstream effects on parasite biology (survival under drug pressure, infection, and transmissibility).

1) Leishmania infection of mouse and human stem cells triggers amastigote quiescence
Serendipitously we discovered by flow cytometry that already after 24 hours of Leishmania infection in LT-HSC, there is a presence of two distinct DsRed + amastigote populations (DsRed hi and DsRed lo ) suggesting different metabolic states (Fig 1A ) in the dsRed-transformed parasite line.Both populations remain at relatively constant proportions from 24 hours post infection (hpi) onwards.The decreased DsRed signal indicates reduced expression from the 18S rDNA locus, previously reported as an indicator of entry into a quiescent state [25].In situ amastigote quiescence was shown to occur independently of strain (L.infantum LEM3323 or clinical isolate LLM2346) or species (L.infantum and L. donovani) (Fig 1A

2) Amastigotes enter quiescence following in situ proliferation in mouse and human stem cells
To uncover why LT-HSC trigger the rapid development of quiescent amastigotes, a CFSE labelling of L. infantum LEM3323 promastigotes was performed to assess the number of parasite divisions before entering into quiescence (i.e.acquire a DsRed lo phenotype).The number of divisions was calculated based on curve-fitting of the cellular CFSE-intensity using the Proliferation Analysis tool of FlowLogic.The 0 hps (hours post staining) peak was considered as generation 0 and an unstained sample as baseline (Fig 2A

PLOS PATHOGENS
proliferation rate in LT-HSC is associated with quiescence.To confirm clinical relevance of these findings, amastigote quiescence was tested in human HSPC using a recent L. infantum clinical isolate (LLM2346 WT PpyRE9/DsRed ).In Fig 2C and 2D, CFSE labelling of promastigotes was again performed to assess the number of divisions.After 24 hours of co-incubation, the

PLOS PATHOGENS
DsRed hi fraction divided about 0-1 time compared to the control (0 hours), which was comparable to the promastigote culture after 24 hours.The amastigote fraction that ultimately becomes quiescent divided about 1-3 times (Fig 2D).The observed difference in division rate between LEM3323 and LLM2346 can be linked to intrinsic growth rate differences (S3A and S3B Fig) .Collectively these data show that Leishmania quiescence arises mostly after 2-3 divisions in mouse and human stem cells.From Fig 2E it is apparent that transitioning promastigotes (asterisk) can already be detected as early as 1 hpi.Moreover, 6 hpi, intracellularly dividing amastigotes can be already clearly observed (red asterisk).By 24 hpi, the cell contains large numbers of amastigotes.These data unambiguously show that promastigotes rapidly convert into proliferating amastigotes in the HSC niche.This further documents the peculiar interaction of Leishmania with this specific cell type.

3) In situ quiescent amastigotes undergo vast transcriptional changes
To unravel the molecular basis for quiescence in LT-HSC amastigotes, unbiased total RNA sequencing was performed on three independent samples of 10,000 DsRed lo (quiescent) and DsRed hi (non-quiescent) amastigotes that were isolated and flow sorted from mouse LT-HSC in three independent infection experiments.Principal component analyses (S4A and S4B Fig) revealed distant profiles of the quiescent and DsRed hi samples, supporting the observed difference between both parasite phenotypes.Consistent with a previous quiescence study (24), ribosomal genes were strongly downregulated.The ribosomal genes (194 genes) and transfer RNA (37 genes) were removed for the downstream differential expression analysis.The total number of reads and median mapped read length were much lower in quiescent samples as compared to DsRed hi samples (S1 Data).Strongly reduced transcript levels were confirmed by RT-dPCR on a set of 18 genes, indicating an average 9.3-fold decrease (S4C Fig) .The used default mapping conditions (MAPQ 10 and no read length limit) were discovered unsuitable due to a very high proportion of fragmented reads in quiescent samples resulting in stochastic RNA read counting.Kraken2 read classification did not indicate any significant contaminations from bacteria, human, or mouse.Based on these observations, very strict criteria (MAPQ 60 and mapped read length greater than 90 bases) were implemented and only 167 nuclear genes with high-confidence reads were identified in the three independent quiescent samples (S1 Data).These genes were enriched in GO biological processes related to microtubule-based movement and cholesterol/ergosterol biosynthetic processes (S4D Fig and S1 Data).While a generalized reduction of transcript levels was evident, no obvious cellular stress response based on heat shock protein expression (only LINF_360027200 -Hsp60 was detected in the three quiescent samples) and no potential positive markers or drivers of quiescence could be identified.

4) Transition through in situ quiescence enhances parasite survival, infectivity, and transmission potential
Next, we wondered whether adoption of a quiescent state would have an impact on various aspects of parasite biology.To investigate whether amastigote quiescence in LT-HSC is linked to survival of treatment and consequent occurrence of relapse, sorted/infected cells were treated with 250 μM paromomycin (PMM) or 7. this study demonstrates that quiescence occurs naturally in LT-HSC in the absence of drug pressure.
To assess whether going through a reversible quiescent state influences subsequent transmission, sand flies were infected with promastigotes derived from DsRed hi and quiescent amastigotes.In Fig 3C, parasite load in the gut was compared at different time points, with parasites having transitioned through a quiescent state showing a slightly enhanced sand fly infectivity (p <0.05 at day 12).A similar analysis was performed for parasites derived from human HSPC, where infectivity in the sand fly vector remained the same (Fig 3E).
As shown before sand fly passage, quiescent parasites post-sand fly passage showed a lower DsRed signal (Fig 3G), indicating that some quiescence-associated phenotypic changes are stable after transmission.Infectivity was evaluated by co-incubation with peritoneal macrophages for 96 hours.Interestingly, promastigotes derived from quiescent (DsRed lo ) amastigotes had a significantly higher infectivity compared to their DsRed hi counterparts (p <0.0001) or the original promastigote culture before LT-HSC passage (p <0.0001).This difference in infectivity was even more pronounced after sand fly passage (Fig 3D).The infectivity in macrophages was also observed to be significantly higher for promastigotes derived from the quiescent strain recovered from human HSPC (p <0.0001) than for DsRed hi promastigotes (Fig 3F).The in vitro growth curves of L. infantum LEM3323 and LLM2346 promastigotes comparing quiescent and DsRed hi phenotypes show no biologically significant differences, suggesting that growth rate is not a determining factor contributing to the observed higher infectivity (S3A and S3B Fig) .These results highlight that a transition through quiescence not only affects treatment but also significantly influences important life cycle features such as infectivity and transmissibility.

5) In vivo relapse parasites share characteristics with parasites that transitioned through quiescence
Using a previously described reproducible post-treatment relapse model [16], mice were sacrificed at 6 weeks post infection (4 weeks post PMM treatment) and BM was collected for promastigote back-transformation (Fig 4A).Infection of mouse peritoneal macrophages revealed an enhanced infectivity of relapse parasites (Fig 4B ) as was found for quiescent parasites.Infection phenotypes can often be attributed to different levels of metacyclic parasites (FSC lo , [26]).To exclude such bias, flow cytometry analysis and visual inspection of the cultures were performed, revealing similar FSC/SSC profiles and visually confirming > 90% metacyclics (S6

Discussion
Current resources for new drug development for VL are scarce.Despite the imminent need to combat the increasing relapse rates, little is known on the underlying cause and suitable methods to study this phenomenon.Notably, post-treatment relapse is mostly not due to reinfection, drug quality, drug exposure or drug-resistant parasites [10,27], but rather due to persistence of the pathogen.This persistence causing subclinical infections and subsequent relapse has been widely described across the microbiological spectrum [11][12][13][14][15][28][29][30].

PLOS PATHOGENS
Generally, two aspects can be the cause of this: pathogens residing in sanctuary sites or niches, or pathogens switching to a quiescent phenotype.The first allows pathogens to survive and escape treatment or immunity without genetic or phenotypic changes.The latter relies on phenotypic diversity, e.g.quiescent or dormant forms [31].Based on our observations, treatment failure in VL most likely results from an intrinsic ability of the Leishmania parasite to benefit from both quiescence and the occupation of sanctuary sites.We have previously identified a relapse-prone cellular niche in the BM in which treatment is less effective [16].Here, we formally link quiescent parasites residing in this particular sanctuary site to drug tolerance.
In literature there is no clear consensus on the definition of persistence or quiescence, and additional synonyms such as dormancy and latency are used interchangeably as conceptually related terms.More commonly, quiescence refers to genetically drug susceptible, dormant (non-growing or slow growing) organisms that survive exposure to a given cidal drug and have the capacity to regrow (or resuscitate and grow) under highly specific conditions [32].Several external triggers may induce quiescence, such as host immunity, drug pressure and/or nutritional and energetic stress [21,33].It is possible that the lowered production of ROS and NO in infected LT-HSC as documented in our previous work [16] creates a more hospitable environment for Leishmania survival and multiplication [34], and underly the specific host cell characteristics that support the emergence of parasitic quiescence.Investigation of the physiology of this phenotype in vivo is very challenging due to their scarcity and difficulty of detection.A review by Barrett et al. envisages several methods to study quiescence such as fluorescent probes, DNA replication probes, sorting of quiescent vs replicating cells, and several omics approaches [31].Studies in various model organisms have been proposed to gain insights in the enigmatic basis of quiescence, such as transgenic fluorescent hypnozoites for P. vivax [35], multiple-stress model to study M. tuberculosis [36], and GFP reporter gene expression from an 18S rDNA locus in cutaneous leishmaniasis [25].
Quiescence in (cutaneous) leishmaniasis has been observed in several experimental systems since the first observation in 2015 by Kloehn et al. [37].A form of quiescence has been described in axenic cultures of cutaneous Leishmania strains treated with antimonials or grown under stress conditions [24,25).A recent study on L. mexicana dermal granulomas describes a mosaic of metabolically active and semi-quiescent parasites during acute phases of infection, linking the phenotype to treatment failure [38].Another study uses in vivo labeling with bromodeoxyuridine (BrdU) to visualize persistent slow-growing L. major parasites in macrophages in the skin [39].To date, a knowledge gap exists for naturally elicited quiescent parasites recovered from a host cell niche, especially for VL.The present study provides unprecedented insights into quiescence of VL parasites in LT-HSC.The Leishmania amastigote stage had already been described as a less active state that may represent an adaptive response to a growth-restrictive intracellular microenvironment in granulomas [37,40].Here, we compared amastigotes within the common macrophage host cell to those within the LT-HSC niche and found distinct differences.Within LT-HSC, quiescence occurred within 6 hours following an estimated 2-3 divisions that affect amastigote size and are linked to rapid genetic alterations as observed by the erosion of the DsRed expression cassette.This stably reduced expression of DsRed that is also present in the promastigote form is possibly linked to aneuploidy, with chromosomal copy number reduction corresponding to reduced transcript and protein levels [41].This indicates that the high intracellular parasite proliferation rate in the stem cell is associated with the early emergence of quiescence.
To study the potential mechanisms underlying Leishmania quiescence, transcriptomic analyses were conducted on sorted DsRed hi and DsRed lo parasites by RT-dPCR and an unbiased total RNAseq.RT-dPCR on a set of 18 tested genes, revealed an order of magnitude lower expression (average 9.3-fold) in quiescent parasites as compared to DsRed hi samples.RNAseq corroborated a much lower number of reads and median mapped read length in quiescent samples.We observed a high proportion of fragmented short reads of lower mapping quality in DsRed lo samples and therefore implemented more strict mapping criteria to exclude the stochastic low-quality reads.As a result, reads for only 167 nuclear genes were detected in the three independent quiescent samples, which were enriched in GO biological processes related to microtubule-based movement and cholesterol/ergosterol biosynthetic processes.No obvious signs of a stress response with the expression of stress related HSPs were noted [42][43][44].The overall downregulation of RNA, including ribosomal protein genes, is a signature of quiescence well described in literature [24,37,45,46], as ribosome biosynthesis is one of the most energy intensive processes in the cell and thus a measure of the metabolic state.Studies in Plasmodium spp.and Toxoplasma gondii persisters have shown that DNA replication, general transcription and protein synthesis are decreased [31].Axenic amastigote forms of L. mexicana and L. braziliensis showed downregulated synthesis of ATP, ribosomal components, proteins and alterations in membrane lipids [25,40].Mitochondrial gene expression has also been described to be reduced in an artificial model of quiescence [24].In contrast to many other species, Leishmania does not have transcriptional regulation and as such is mainly compensated by post-transcriptional mechanisms [47].Indeed, Leishmania is known for its extreme genomic and phenotypic variability, whereby a rapid shift in the gene repertoire is one of the key mechanisms for swift adaptation to changing environments [48], some of which have been associated with increased fitness in stress conditions or drug resistance [49].Characterizing quiescent VL parasites recovered from LT-HSC showed an increased cellular infectivity and a high capacity to colonize the sand fly gut.Consistent with the quiescent phenotype, parasites recovered from the BM of relapsed mice, of which LT-HSC were previously shown to be the main parasitized cells [16], show an increased fitness with an elevated macrophage infectivity combined with a high sand fly infectivity.This suggests that the selected phenotype, typically associated with transition through a quiescent state, may pose an additional threat to leishmaniasis control programs.An increased infectivity associated with relapse of L. donovani infection has already been shown for miltefosine and antimonial treatment in the Indian subcontinent [10,50,51].To prevent treatment failure, we advocate that the quiescent state of Leishmania should be considered in the early stages of the drug discovery process.Given its broad relevance across the microbial spectrum, further exploration of parasitic quiescence in the LT-HSC niche is warranted to identify potential liabilities for therapeutic intervention.

Ethical statement
The use of laboratory rodents was carried out in strict accordance with all mandatory guidelines (EU directives, including the Revised Directive 2010/63/EU on the Protection of Animals used for Scientific Purposes that came into force on 01/01/2013, and the declaration of Helsinki in its latest version) and was approved by the Ethical Committee of the University of Antwerp, Belgium (UA-ECD 2019-04).Human bone marrow aspirate rest samples, obtained as a diagnostic sample without a written informed consent, were available for in vitro infection experiments following approval by the Committee of Medical Ethics of the Antwerp University Hospital (B3002021000027).In accordance with Article 20, §1 of the Belgian decree of 19 December 2008, body material that remains after a diagnostic examination or an intervention (residual material or residual tissue) can be used for scientific research.Patients or legal

PLOS PATHOGENS
representatives have the right to refuse this at any time and can communicate this refusal to the responsible doctor or the medical director.

Laboratory animals and sand fly colony
Female BALB/c mice (6-8 weeks old) were purchased from Janvier (Genest-Saint-Isle, France) and accommodated in individually ventilated cages in SPF conditions.They were provided with food for laboratory rodents (Carfil, Arendonk, Belgium) and water ad libitum.Animals were subdivided in experimental groups based on simple randomization.Mice were kept in quarantine for at least 5 days before starting the experiment.Euthanasia was performed in CO 2 chambers followed by cervical dislocation, and tissues were collected under aseptic conditions.
A Lutzomyia longipalpis sand fly colony was initiated with the kind help of NIH-NIAID (Prof.Shaden Kamhawi and Prof. Jesus Valenzuela) and maintained at the University of Antwerp under standard conditions (26˚C, > 75% humidity, in the dark) with provision of a 30% glucose solution ad libitum [54].For infection experiments, 3-to 5-day old females from generations 31 to 44 were used.

Primary mouse cells
Mouse BM was collected from BALB/c mice using two distinct techniques, based on pilot studies comparing alternative methods in terms of yield and quality.For both techniques, mice were sacrificed, and hind legs aseptically removed.Isolated femurs and tibias were cleaned by removing soft tissue from the bone using 70% ethanol-soaked cloth and tweezers.
For the crushing technique, the protocol was adapted from Lo Celso and Scadden [55].Briefly, bones were crushed with mortar and pestle in ammonium-chloride-potassium (ACK) buffer (0.15 M NH 4 Cl, 1.0 mM KHCO 3 , 0.1 mM Na 2 EDTA) for erythrocyte lysis.Single cell suspensions were obtained by filtering through MACS SmartStrainers (100 μm, Miltenyi Biotec), centrifuged at 500×g for 10 min (4˚C) and resuspended in phosphate-buffered saline (PBS) + 0.2% bovine serum albumin (BSA).For efficient depletion of mature lineage-positive hematopoietic cells and to specifically isolate the preferred lineage-negative cells (i.e.undifferentiated progenitor cells), the Direct Lineage Cell Depletion Kit (Miltenyi Biotec) was employed according to manufacturer's instructions.Following lineage depletion, cells were counted in PBS using a KOVA counting chamber and resuspended in PBS + 0.2% BSA buffer to 2×10 7 cells/mL.Cells were kept on ice during all procedures.
The centrifugation method, adjusted from the protocol described by Amend et al. [56] and Dobson et al. [57], was used for subsequent macrophage and dendritic cell differentiation.Briefly, a 0.5 mL microcentrifuge tube was perforated at the bottom with a 21G needle and PLOS PATHOGENS nested inside a 1.5 mL tube (both from Eppendorf).After collection of femurs and tibias, one proximal end (knee epiphysis) was cut-off and placed in the 0.5 mL tube.Nested tubes were centrifuged in a microcentrifuge at 10,000×g for 15 sec, resulting in a visible pellet in the 1.5 mL tube.This pellet was then resuspended in ACK buffer for erythrocyte lysis.
Primary peritoneal macrophages were obtained from Swiss mice after inoculation of 1 mL 2% starch solution in PBS.Macrophages were seeded in a 96-well plate (6×10 4 cells/well) and kept at 37˚C and 5% CO 2 to allow adhesion.After 48 hours, macrophages were infected as described below.

Primary human BM cells
Human BM aspirate was obtained from the iliac crest using BD Vacutainer Plastic K3EDTA Tubes, initially collected for diagnostics, and delivered as residual sample.The BM was subjected to erythrocyte lysis twice using ACK buffer.Single cell suspensions were obtained by filtering through MACS SmartStrainers (100 μm, Miltenyi Biotec), centrifuged at 300×g for 10 min (4˚C) and resuspended in PBS + 0.2% BSA.Cells were counted in PBS and diluted to 2×10 7 cells/mL for flow cytometric analysis.Cells were kept on ice during these procedures.

In vitro and in vivo Leishmania infections
Parasite density was assessed using a KOVA counting chamber.For in vitro infections, macrophages, LT-HSC and human hematopoietic stem and progenitor cells (HSPC) were co-cultured with stationary-phase promastigotes at a multiplicity of infection (MOI) of 5 for a minimum of 24h at 37˚C with 5% CO 2 .Parasite cultures were inspected visually and measured by flow cytometry to confirm the presence of > 90% metacyclics, to exclude bias as a result of differences in metacyclogenesis in the cultures (S6 Fig) .For post-passage infections (both after sand fly and in vitro HSC infections, vide infra), parasites were recovered in HOMEM medium at 25˚C and checked daily for growth.Parasites were then transferred to a T25 flask in parallel and passaged once before determining the percentage of metacyclics using flow cytometry.For in vivo infection, stationary-phase parasites were centrifuged for 10 min at 4,000×g (25˚C) and resuspended to 1×10 9 parasites/mL in sterile RPMI medium.Mice were infected intravenously (i.v.) in the lateral tail vein with 1×10 8 parasites in 100 μL of RPMI medium.Animals were monitored using in vivo bioluminescence imaging (BLI) at selected time points.Imaging was performed 3 min after intraperitoneal (i.p.) injection of 150 mg/kg D-Luciferin (Beetle Luciferin Potassium Salt, Promega) in the IVIS Spectrum In Vivo Imaging System under 2.5% isoflurane inhalation anesthesia using 15 min exposure.Images were analysed using LivingImage v4.3.1 software by drawing regions of interests (ROIs) around specific organs to quantify the luminescent signal as relative luminescence units (RLU).

Cell staining, flow cytometry and fluorescence-activated cell sorting (FACS)
Parasite cultures were analysed on a MACSQuant Analyzer 10 (Miltenyi Biotec) after a 10 min centrifugation at 4,105×g and resuspension in PBS + 0.2% BSA buffer.Analyses were performed using FlowLogic Software (Miltenyi Biotec) using a specific gating for singlet parasites expressing dsRed, for which the non-transfected parental parasite line served as a control.In some experiments, parasites were stained with 5-(and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE; Cell Division Tracker Kit, BioLegend) according to manufacturer's instructions.Briefly, lyophilized CFSE was reconstituted with DMSO to a stock concentration of 5 mM.This stock solution was diluted in PBS to a 5 μM working solution.Promastigotes at a concentration of 10 8 cells/mL were centrifuged at 4,000×g for 10 min and resuspended in CFSE working solution for 20 min at 25˚C.The staining was quenched by adding 5 times the original staining volume of cell culture medium containing 10% FBS.Parasites were centrifuged again and resuspended in pre-warmed HOMEM medium for 10 min.After incubation, CFSE labeled parasites were used for infection and determining in situ proliferation in macrophages, LT-HSC and human HSPC.
Sorting of mouse LT-HSC and human HSPC was performed, and quality confirmed as described previously [16].Briefly, BM cell suspensions (2×10 7 /mL concentration) were treated with FcɣR-blocking agent (anti-CD16/32, clone 2.4G2, BD Biosciences) for 15 min, followed by a washing step using 400×g centrifugation and resuspension in PBS + 0.2% BSA buffer.Next, cells were incubated for 20 min at 4˚C with a mix of fluorescent conjugated anti-mouse antibodies at optimized concentrations.DAPI Staining Solution (Miltenyi Biotec) was used to assess viability.A 96-well plate (Greiner Bio-One) was prepared for sorting by adding RPMI 1640 medium supplemented with 1% NEAA, 100 U/mL penicillin, 100 μL streptomycin, 500 μg/mL gentamycin, 2 mM L-glutamine, 1 mM sodium pyruvate and 10% iFCS to the wells in which 10,000 LT-HSC/well were sorted using FACSMelody (BD Bioscience) following specific gating strategies, confirmed with fluorescence minus one (FMO) controls and compensated using single stains, as described and shown in S7 Fig and Tables 1-3 of our previous report [16].For visualizing infection, LT-HSC were collected on slides by Cytospin, fixed using methanol and stained for 15 min with Giemsa (Sigma Aldrich).Microscopic images were acquired using the UltraVIEW VoX dual spinning disk confocal system (PerkinElmer).For image analysis, z-stack imaging was included and a z-projection using the max intensity method was created for each image.The region-of-interest (ROI) was then manually drawn around the DsRed + signals of the amastigotes, allowing quantification of the area and fluorescence intensity using the FIJI software.For analysis of dsRed and/or CFSE levels on amastigotes at designated timepoints, infected macrophages and LT-HSC were recovered from the cultures.Cells were centrifuged at 400×g for 10 min and in PBS + 0.2% BSA.Host cell membranes were disrupted by 3 passages through a 25G needle.Amastigotes were collected in the supernatant after centrifugation at 250×g for 10 min and subsequently pelleted at 3,000×g for 10 min and resuspended in 500 μL PBS + 0.2% BSA for analysis by flow cytometry.

RNA isolation
Total RNA was extracted from three independent samples of 10,000 FACSMelody-sorted DsRed hi amastigotes or quiescent amastigotes (L.infantum LEM3323 PpyRE9/DsRed ) obtained from 5,000 sorted and infected LT-HSC.Extraction was performed with the QIAamp RNA Blood Mini kit (Qiagen), according to the manufacturer's instructions.To exclude gDNA, an additional step using gDNA elimination columns (Monarch) was performed.RNA samples were stored in aliquots at -80˚C.

RT-Digital PCR (RT-dPCR)
RT-dPCR was conducted using a QIAcuity dPCR system (QIAGEN), employing nanoplate PCR plates (8.5 k partitions, 96 well) and the QIAcuity OneStep Advanced EG Kit (cycling conditions: reverse transcription at 50˚C for 40 minutes, initial denaturation at 95.0˚C for 2 minutes, followed by 40 cycles of denaturation at 95.0˚C for 10 seconds, annealing/extension at 60.0˚C for 30 seconds, and a final extension at 40.0˚C for 5 minutes.Imaging was with an exposure duration of 500 ms and gain set to 6. Duplicate assays were performed for 18 different genes (S2 Table ) on RNA extracts from sorted DsRed hi and DsRed lo amastigotes.From the obtained data, extracted RNA copy numbers per 10 4 parasites were calculated.

Sand fly infections and evaluation of parasite load
Sand fly females (L.longipalpis) were fed with heat-inactivated heparinized mouse blood containing 5×10 6 /mL promastigotes from log-phase cultures through a chicken skin membrane.Groups were randomized by an independent researcher until data analysis to avoid bias.Blood-fed females were separated 24 h after feeding, kept in the same conditions as the colony and dissected on 5, 7, 9, and 12 days post blood meal to microscopically check the presence of parasites.Following disruption of the total gut in 50 μL PBS, the parasite load was quantified microscopically using a KOVA counting chamber [64,65].Parasites isolated from sand flies on day 12 post blood meal were cultured at 25˚C in HOMEM promastigote medium supplemented with 5% penicillin-streptomycin and checked daily for growth, to obtain post-sand fly PLOS PATHOGENS cultures.Parasites were then transferred to a T25 flask in parallel and passaged once before measuring DsRed expression or before infection studies (vide supra).The latter was carried out for all conditions synchronously, using stationary phase promastigote cultures, to normalize infection experiments.

Promastigote growth
Promastigote growth curves were made as described before [66] to compare the in vitro growth of quiescent vs non-quiescent strains.After passage through fine needles (21G and 25G) to break clustering, the promastigotes were diluted in PBS and counted by KOVA counting chamber.Exactly 5×10 5 log-phase promastigotes/mL were seeded in 5 mL HOMEM and their number was determined by microscopic counting every 24 h for a total of 10 days.Three independent repeats of each strain were run in parallel.

Statistics and reproducibility
Statistical analyses were performed using GraphPad Prism version 9.0.1.Tests were considered statistically significant if p <0.05.Growth curves were statistically compared using Wilcoxon matched-pairs signed rank test.Parasite load in sand fly infections were tested using Unpaired t test.Infectivity in macrophages was tested using Ordinary one-way ANOVA.MFI of DsRed expression was compared using Mann-Whitney test.
) and was recorded in both mouse LT-HSC and human hematopoietic stem and progenitor cells (HSPC) (Fig 1B).In contrast, amastigotes purified from infected macrophages cluster in one homogenous DsRed hi population (Fig 1C).Promastigote back-transformation was used to confirm viability of sorted DsRed lo parasites, the capacity to regain proliferative capacity and stability of the DsRed lo phenotype.DsRed expression in the quiescent state remained lowered after transformation into the promastigote form (Fig 1D), which was also confirmed for derived monoclonal lines.Quiescent parasite cultures lost DsRed-expression after promastigote back-transformation with a frequency between 1.96% (1/51 clones) and 4.76% (2/42 clones) (S1 Fig), suggesting that parasites undergo a rapid evolutionary adaptation response and genetic rearrangements upon entry and exit from quiescence.Loss of the dsRed gene was demonstrated at RNA and DNA level by qPCR (S1B Fig).In contrast, no clones derived from DsRed hi parasites lost the dsRed construct.Interestingly, quiescent amastigotes exhibit a significantly reduced size compared to DsRed hi amastigotes (Fig 1E).Confocal fluorescence microscopy with z-stacking confirmed heterogenous DsRed signals and variable amastigote size in the LT-HSC (Fig 1F and 1G).This was corroborated by flow cytometry, as the mean fluorescence intensity of the DsRed signal increased with the apparent amastigote size (forward scatter-FSC, Figs 1E, S2A and S2B), illustrating that acquisition of quiescence is associated with several cellular changes.
, left panel).The CFSE profile at time 0 has negligible differences in initial parasite size, and most are metacyclic (FSC lo ) with evenly distributed CFSE signal (S2C Fig).CFSE versus FSC profile does not show a correlation between CFSE intensity and event size (S2D Fig).After 6 hours of co-incubation, the DsRed hi amastigote fraction divided about 1 time compared to the control (0 hours), which was comparable to promastigote proliferation in 6-hour in vitro cultures.In contrast, the amastigote PLOS PATHOGENS fraction that would eventually acquire a DsRed lo phenotype displayed a more diverse pattern, ranging between 1, 2 and 3 in situ divisions (Fig 2A).The highest proportion of amastigotes underwent 3 divisions to enter into quiescence (Fig 2B).These data indicate that the high

Fig 2 .
Fig 2. Number of divisions associated with quiescence in amastigotes from mouse LT-HSC and human HSPC.(a) Number of divisions as calculated by CFSE staining and defined by curve-fitting of the cellular CFSE-intensity using the Proliferation Analysis tool of FlowLogic (left panel).Controls are unstained and CFSE L. infantum (LEM3323 WT PpyRE9/DsRed ) promastigote cultures.Sorted mouse LT-HSC were infected with CFSE labelled L. infantum promastigotes and amastigotes were recovered after 6 hours of co-incubation (right panel).(b) Percentage of cells in each division range from (a).Results are based on three independent repeats.(c) Number of divisions as calculated by CFSE staining using the Proliferation Analysis tool (left panel).Controls include unstained and CFSE-stained L. infantum (LLM2346 WT PpyRE9/DsRed ) promastigote cultures.Sorted human HSPC were infected with CFSE labelled L. infantum promastigotes and amastigotes were recovered after 24 hours of co-incubation (right panel).(d) Percentage of cells in each division range from (c).Results are based on three independent repeats.(e) Early infection of L. infantum (LEM3323 WT PpyRE9/DsRed ) in mouse LT-HSC at 1, 3, 6, and 24 hours post infection visualised by Giemsa staining.Amastigotes (arrow), promastigotes (dotted arrow), promastigotes transitioning to amastigotes, already without flagellum (asterisk), dividing amastigotes (red asterisk).https://doi.org/10.1371/journal.ppat.1012181.g002

5 Fig 3 .
Fig 3. Phenotypic characteristics of quiescent amastigotes from LT-HSC.(a-b) Sorted LT-HSC were infected with L. infantum (LEM3323 WT PpyRE9/DsRed ) for 24 hours followed by treatment with 250 μM PMM or 7.5 μM MIL for 72 hours.To compare pre-and post-treatment distribution of quiescent parasites, amastigotes were isolated and remeasured on the FACSMelody.(c) Sand flies were infected by L. infantum LEM3323 promastigotes recovered from DsRed hi and quiescent amastigotes in mouse LT-HSC.The parasite load in the gut was assessed at days 5, 7, 9 and 12 after infection (blood meal).Sand fly infections were repeated three independent times.Unpaired t test, 10 < n < 30, *p <0.05.(d) L. infantum LEM3323 promastigotes of a control culture and promastigotes recovered from DsRed hi and quiescent parasites in mouse LT-HSC, pre-and post-sand fly passage (pre-and post-SF) were co-incubated with peritoneal macrophages for 96 hours and infectivity was assessed with Giemsa staining.Cultures were visually confirmed to contain > 90% metacyclics and normalized by counting to expose macrophages at a MOI of 5 (S6 Fig).The original promastigote culture was used as a control and infectivity was found to be significantly lower than those of pre-SF DsRed hi (p <0.01), post-SF DsRed hi (p <0.05), and pre-and post-SF quiescent parasites (p <0.0001).Ordinary one-way ANOVA, 30 < n < 130, ****p <0.0001.% infected cells and number of parasites per 100 Fig).Moreover, growth curves shown in S3C Fig demonstrate no discernable differences in promastigote growth over 10 consecutive days of culture.The percentage of reduction after in vitro PMM treatment of infected macrophages remained stable for relapse versus the parental parasites, demonstrating that these parasites did not acquire a drug resistant phenotype (Fig 4C).To check whether the parasites were still infective for sand flies, the gut parasite load was compared at different time points.No significant differences in sand fly parasite loads macrophages are included in S1 Table.(e) Sand flies were infected with L. infantum LLM2346 promastigotes, recovered from DsRed hi and quiescent amastigotes in human HSPC.The parasite load in the gut was assessed at days 5, 7, 9 and 12 after the infectious blood meal.Sand fly infections were repeated three independent times.10 < n < 34.(f) L. infantum LLM2346 promastigotes of DsRed hi and quiescent recovered promastigotes from human HSPC were co-incubated with peritoneal macrophages for 96 hours and infectivity was assessed with Giemsa.Mann-Whitney test, n = 100, ****p <0.0001.(g) DsRed expression of L. infantum LEM3323 DsRed hi and quiescent promastigotes recovered from mouse LT-HSC before or after passage through the sand fly, and L. infantum LLM2346 DsRed hi and quiescent promastigotes recovered from human HSPC.Mann-Whitney test, **p <0.01.All experiments are expressed as mean ± SEM. https://doi.org/10.1371/journal.ppat.1012181.g003PLOS PATHOGENS (Fig 4D) and in vitro promastigote growth (S3C Fig) were detected.The recorded high fitness of relapse parasites overlaps with the parasite phenotype after transition through a quiescent stage in LT-HSC.

Fig 4 .
Fig 4. Phenotypic characteristics of promastigotes from relapsed BM.(a) BALB/c mice were infected with 10 8 stationary phase promastigotes of L. infantum LEM3323 WT PpyRE9/DsRed .One group was treated with PMM 350 mg/kg per day (IP) for 5 consecutive days.Infection was followed up by BLI, BM was collected at 6 weeks post infection (wpi) from untreated and relapsed mice.(b) Stationary phase promastigotes recovered from relapsed BM (relapse) and untreated BM (infection) were co-incubated at a MOI of 5 with peritoneal macrophages for 96 hours and infectivity was assessed with Giemsa.Parasite cultures were counted and visually confirmed to contain > 90% metacyclics to exclude an infectivity bias based on the parasite culture (S6 Fig).Mann-Whitney test, n = 100, ***p <0.001.% infected cells and number of parasites per 100 macrophages are included in S1 Table.(c) L. infantum LEM3323 promastigotes recovered from relapsed BM (relapse) and untreated BM (infection) were co-incubated with peritoneal macrophages for 24 hours and treated with 120 μM, 250 μM or 350 μM PMM for 72 hours, infectivity was assessed with Giemsa.(d) Sand flies were infected by L. infantum LEM3323 promastigotes recovered from relapsed BM (relapse) and untreated BM (infection) and the parasite load in the gut of infected flies was assessed at days 5, 7, 9 and 12 after infection (blood meal).Sand fly infections were repeated three independent times.10 < n < 32.https://doi.org/10.1371/journal.ppat.1012181.g004