The macrophage microtubule network acts as a key cellular controller of the intracellular fate of Leishmania infantum

The parasitophorous vacuoles (PVs) that insulate Leishmania spp. in host macrophages are vacuolar compartments wherein promastigote forms differentiate into amastigote that are the replicative form of the parasite and are also more resistant to host responses. We revisited the biogenesis of tight-fitting PVs that insulate L. infantum in promastigote-infected macrophage-like RAW 264.7 cells by time-dependent confocal laser multidimensional imaging analysis. Pharmacological disassembly of the cellular microtubule network and silencing of the dynein gene led to an impaired interaction of L. infantum-containing phagosomes with late endosomes and lysosomes, resulting in the tight-fitting parasite-containing phagosomes never transforming into mature PVs. Analysis of the shape of the L. infantum parasite within PVs, showed that factors that impair promastigote-amastigote differentiation can also result in PVs whose maturation is arrested. These findings highlight the importance of the MT-dependent interaction of L. infantum-containing phagosomes with the host macrophage endolysosomal pathway to secure the intracellular fate of the parasite.

Introduction promastigotes from 1 h to 4 h PI, accompanied by the rapid appearance of the intermediate parasite forms. Subsequently, the number of intermediate forms steadily decreased from 4 to 18 h PI, at the same time as the number of amastigotes increased. In addition, CLSM showed the replication of L. infantum amastigotes, in which the PV membrane elongated preceding its fission, generating two new tight-fitting PVs, each containing one progeny amastigote ( Fig  1D).

Biogenesis of tight-fitting L. infantum LEM 5700-containing PVs hosted in macrophage-like RAW 264.7 cells
Wilson and co-workers [15][16][17][18] showed the decoration of L. infantum chagasi-containing vacuoles by several bona fide markers of sub-compartments of the host-cell endolysosomal pathway at various times post-infection (PI) in promastigote-infected human monocyte-derived macrophages, human U937 monocytic cells, and mouse bone-marrow-derived macrophages. This included transient decoration by early endosome antigen-1 and Rab5 GTPase and stable decoration by membrane-associated protein-1, belonging to the Lamp1/2 family of proteins distributed among LEs and lysosomes [19]. We revisited and extended these observations by conducting a time-course experiment to assess and quantify the association of a large set of endolysosomal markers with the parasites insulated in mouse macrophage-like RAW 264.7 cells infected with L. infantum promastigotes. We examined the classical markers, membraneassociated Rab7 GTPase, a marker of the transition between early endosomes (EEs) and LEs [20], membrane-associated protein-2 (Lamp-2) [19], and cathepsin D (CatD), an aspartyl protease that represents a major category of luminal lysosomal hydrolases [21]. In addition, we followed the Qb-SNAREs (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) Vti1a and Vti1b (vesicle transport through t-SNARE Interaction) [22]. Knowledge of the membrane composition of tight-fitting PVs that insulate species of the L. donovani complex and L. major in terms of functional proteins is limited [9][10][11][12]. We also examined, for the first time, the lysosomal membrane-associated class III glucose/fructose transporter, GLUT8 (Slc2A8) [23,24] and polypeptide transporter associated with antigen processing-like (TAPL; ABCB9) [24,25].
Multi-channel confocal laser scanning microscopy (CLSM) acquisition and 3D-reconstruction surface rendering allowed the simultaneous observation of immunolabeled endolysosomal markers and parasites in infected cells. Quantification of decorated intramacrophage parasites during a time-course PI (Fig 2A) showed a rapid increase in the percentage of Rab7-positive parasites, reaching a maximum at 4 h PI, which than tended to decrease. The percentage of parasites decorated with Lamp-2 increased rapidly, reaching a maximum at 4 h PI, and remained stable up to 24 h. The percentage of parasites positive for Vti1a and Vti1b increased rapidly, reaching a maximum at 4 h PI, and then declined significantly. The percentage of CatD-positive parasites increased more slowly, reaching a maximum at 8 h PI, and remained stable at the subsequent time-points. We observed that the percentage of GLUT8-and TAPLpositive L. infantum parasites increased similarly, reaching a maximum at 8 h PI, and then decreased significantly.
3D-reconstruction surface rendering confocal micrographs and videos showed various patterns of decoration of insulated L. infantum parasites by endolysosomal markers, including randomly distributed small patches of various sizes and large patches wrapping densely or almost entirely around the parasite body. Fusiform-like parasites were decorated by large Rab7-positive patches, whereas amastigotes were only decorated by small patches (Fig 2B). Lamp-2-positive tube-like structures and patches densely decorated both fusiform-like parasites and amastigotes (Fig 2C and S1 Video). Large Vti1a-positive patches decorated the fusiform-like parasites and amastigotes ( Fig 2D). Both fusiform-like parasites and amastigotes were sparsely decorated by small randomly distributed Vti1b-positive patches (Fig 2E). Large GLUT8-and TAPL-positive patches decorated the fusiform-like parasites, whereas amastigotes were sparsely decorated by small randomly-distributed patches (Fig 2F and 2G, respectively). In contrast, both fusiform-like forms and amastigotes were almost entirely enwrapped by continuous patches positive for the lysosomal marker CatD (Fig 2H and S2 Video). The observed decoration of L. infantum parasites by Rab7 and Lamp-2 is reminiscent of their previously observed presence in the membrane of tight-fitting L. donovani-containing PVs [8,26]. The observed decoration of L. infantum parasites with Rab7, Lamp-2, and CatD is consistent with the previously observed presence of Rab7p, Lamp-1, and cathepsin B in tight-fitting L. major-containing PVs [27,28]. Loading of infected cells with LysoTracker Red, a small membrane-permeable lysosomotropic molecule that labels acidic vacuolar compartments, showed the intracellular ovoid and round parasite forms of L. infantum to be strongly enwrapped by LysoTracker Red fluorescence (Fig 2I).
We next measured the size of the luminal space of phagosomes/PVs that insulated the L. infantum parasites. We took advantage of the CatD immunofluorescence, which almost completely enwrapped the parasites, and which is known to localize to the luminal space of PVs [28]. Measurement of the volume of the parasite body and that of CatD immunolabeling by isosurface rendering analysis showed there to be a small vacuolar space between the parasite body and phagosome/PV membrane (S1 Fig), in accordance with the known tight-fitting nature of the phagosome/PV insulating parasites of the L. donovani complex and L. major [3,4].

Impairment of the maturation of tight-fitting L. infantum LEM 5700-containing phagosomes/PVs
We developed a pharmacological approach to block the maturation of L. infantum-containing phagosomes/PVs that occurs through heterotypic fusion with LEs and lysosomes. We used the small trafficking inhibitor Retro-2, known to block ricin and Shiga-toxin trafficking at the EEs-trans-Golgi network (TGN) interface [29,30]. For our time-dependent analysis, we took into account the previously [15][16][17][18] and above reported rapid insulation of L. infantum parasites within early tight-fitting phagosomes and subsequent mature tight-fitting PVs, also previously observed with tight-fitting PV-insulated L. major [27,31] and L. donovani [7,32]. Moreover, we also accounted for the fact that Retro-2 exerts direct parasiticide activity against axenic L. amazonensis at concentrations of 10 to 100 μM [33]. We thus infected the cells with L. infantum LEM 5700 in the continuous presence of the small compound at sub-parasiticide concentrations of 1 to 5 μM. We therefore ensured that both the number of L. infantum promastigote-infected cells and the number of promastigotes internalized within each cell were not modified relative to untreated infected cells (S2 Fig). In addition, we controlled that the continuous presence of Retro-2 at 1 μM promoted the re-localization of the Qa-SNARE syntaxin 5 (Stx5) from its normal perinuclear localization toward the cytoplasm (S3 Fig) previously observed when the small inhibitor blocks the retrograde transport of the toxins [29].
3D-reconstruction surface rendering confocal microscopy showed the absence of decoration by the LE-associated Rab7 of the fusiform-like and ellipsoidal forms of the parasites present in 18 h-infected Retro-2-treated cells (Fig 3A and 3D). For Lamp-2, 3D-reconstruction surface rendering confocal microscopy and video showed the absence of decoration of fusiform-like and ellipsoidal parasite forms by Lamp-2-positive tube-like and small patches in 8 h-, 18 h-and 24 h-treated infected cells (Fig 3B and 3D, and S3 Video). Similarly, parasite forms present in 8 h-, 18 h-and 24 h-treated infected cells showed the absence of decoration by Vti1a and Vti1b (Fig 3C and 3D). Quantification of Lamp-2 showed the decreased association of LE-associated markers with the parasites to develop as a function of the concentration of Retro-2 ( Fig 3E).
3D-reconstruction surface rendering confocal micrographs also showed there to be a marked decrease or the absence of decoration of fusiform-like and ellipsoidal parasite forms by the lysosomal membrane-associated GLUT8 (Fig 4A and 4D) and TAPL (Fig 4B and 4D) transporters in 8 h-, 18 h-and 24 h-infected Retro-2-treated cells relative to untreated infected cells. Consistent with these results, 3D-reconstruction surface rendering confocal micrographs and quantification showed the almost total absence of decoration by the lysosomal luminalassociated CatD hydrolase of fusiform-like and ellipsoidal parasite forms hosted in 8 h-, 18 hand 24 h-infected Retro-treated cells relative to untreated infected cells (Fig 4C and 4D, and S4 Video). Quantification of the number of intramacrophage parasites decorated by CatD showed that the decrease of parasite decoration developed as a function of the concentration of Retro-2 ( Fig 4E). For the CatD-decorated fusiform-like parasites hosted in Retro-2-treated infected cells (Fig 4C), scanning of the CatD RFI in fusiform-like parasites showed the level of decoration per parasite to be dramatically lower than that of fusiform-like parasites hosted in untreated infected cells (Fig 4F to 4H). In addition, LysoTracker Red fluorescence did not enwrap the parasites in cells infected in the presence of Retro-2 ( Fig 5A and 5B), contrasting with the extensive enwrapping of parasites hosted in infected untreated cells (Fig 2H). The observed decrease of association with LE-and lysosome-associated molecules was not the consequence of Retro-2-triggered alteration of the cellular presence or distribution of LES and lysosomes, as CLSM images and quantification showed that the cell expression and distribution of Lamp-2-and CadD-positive vesicles were not modified in Retro-2-treated cells relative to that of untreated cells (S4 Fig).
Overall, these results show that Retro-2 impaired the acquisition of host cell LE-and lysosome-associated markers by L. infantum-containing phagosomes hosted in macrophage-like cells, suggesting that the small trafficking blocker arrested the heterotypic interactions between these vacuolar compartments. To confirm that Retro-2 triggers failure of the heterotypic  interaction of donor vacuoles with lysosomes, we conducted a control experiment in a noninfectious cellular autophagy model in which the fusion of donor autophagosomes with recipient lysosomes leads to the formation of the final degradative vacuolar compartment, autolysosomes [34,35]. We used HeLa cells stably expressing the autophagy marker microtubuleassociated protein 1 light chain 3 (LC3) coupled to green fluorescent protein (GFP-LC3). As expected [36], nutrient starvation induced the strong presence of autolysosomes in the cell cytoplasm ( https://doi.org/10.1371/journal.pntd.0008396.g005 [36]. Western blotting showed the abundance of LC3-II protein to be no higher in nutrientstarved Retro-2-treated cells in the presence of CQ than nutrient-starved Retro-2-treated cells without CQ (S5C Fig), showing that Retro-2 blocks autophagic flux.

Permanent residence of L. infantum LEM 5700 parasites in immature phagosomes negatively affects parasite differentiation and survival
We next examined whether impairment of the heterotypic interaction between the L. infantum-containing phagosomes and the macrophage LEs and lysosomes has an impact on parasite differentiation. Quantitative imaging showed Retro-2-treated infected cells to contain mostly intermediate forms that showed a marked delay in transformation into amastigotes ( Fig 6A). This observed delay of differentiation of intramacrophage parasites was not simply a direct effect of the compound on the parasites, as the differentiation of Retro-2-treated axenic parasites was not altered (S6 Fig). CLSM analysis showed a high number of L. infantum intermediate and amastigote forms residing in Retro-2-treated infected cells with a highly-altered cell surface (Fig 6B and 6C), which strongly differed from that of the regular body surface of parasites hosted in untreated infected cells (Fig 1). RT-qPCR quantification showed marked time-dependent lowering of the parasite and amastigote burden in Retro-2-treated infected cells ( Fig 6D). Overall, these results provide evidence that the maturation of L. infantum-containing phagosomes is necessary for L. infantum parasites to successfully complete promastigote-to-amastigote differentiation and survive.

Routing of L. infantum LEM 5700-containing phagosomes toward LEs and lysosomes is microtubule (MT) dependent
The functional maturation process of phagosomes implies dynein-dependent movements of early phagosomes along MT tracts followed by their fusion with lysosomes controlled by SNAREs-containing membrane fusion platforms [6]. The above observed lack of decoration of poorly differentiated L. infantum parasites by LE and lysosomal markers in Retro-2-treated infected macrophage-like cells may result from the absence of L. infantum-containing phagosome movement along MT tracks or downstream impairment of SNAREs-dependent membrane fusion with LEs and lysosomes. Thus, we first examined the organization of the MT network in Retro-2-treated infected macrophage-like cells. Unexpectedly, 3D-reconstruction surface rendering confocal micrographs revealed the dense tubulin network in L. infantum LEM 5700-infected macrophage-like cells (Fig 7A and 7C) to be highly disorganized in cells infected for 24 h in the continuous presence of Retro-2 ( Fig 7B and 7C). In these cells, large spaces devoid of tubulin and dispersed fragmented tubulin-positive bar-like structures were evident. We validated this unexpected observation by conducting a control experiment with epithelial HeLa cells in which the MT network consisted of straight MT filaments well-oriented from the cell nucleus toward the membrane (Fig 7D). CLMS analysis showed that, as in macrophage-like cells, 24 h of treatment with Retro-2 triggered dramatic concentrationdependent disorganization of the HeLa MT network, characterized by the concentrationdependent appearance of dispersed tubulin-positive bar-like and vesicle/aggregate structures (Fig 7E and 7F). We finally confirmed the ability of Retro-2 to induce concentration-dependent MT disassembly using an in vitro tubulin polymerization assay (Fig 7G).

Nocodazole and siRNA dynein silencing mimic the effects of Retro-2
We confirmed the pivotal role of MTs in the maturation process of L. infantum LEM 5700-containing phagosomes/PVs by infecting macrophage-like cells in the continuous presence of the reference MT-disassembling agent (MDA), nocodazole [37]. 3D-reconstruction surface rendering analysis showed the elevated presence of hosted fusiform-like and ellipsoidal parasites in nocodazole-treated 24 h-infected cells (Fig 8A). The maturation of L. infantum-containing phagosomes/PVs hosted in macrophage-like cells infected in the continuous presence of nocodazole was impaired, shown by the scarcity of CatD decorating the hosted parasites (Fig 8A and 8B). The continuous treatment of cells with nocodazole impaired parasite differentiation, shown by the prolonged elevated presence of the intermediate forms of the parasite, with the concomitant delayed appearance of amastigotes (Fig 8C), as in Retro-2-treated infected cells (Fig 6A). Moreover, the parasites hosted in nocodazole-treated infected cells had an altered cell surface (Fig 8A and 8D), such as those in infected cells treated with Retro-2 (Fig 6B and 6C).
We completed our analysis by silencing the cellular gene dynein. This gene codes for the MT-associated motor minus-end-directed dynein [38], which generates rapid unidirectional movement of phagosomes towards lysosomes along MTs [39]. RT-qPCR showed siRNAmediated silencing of the dynein gene to be effective in macrophage-like RAW 264.7 cells ( Fig  9A). The siRNA dynein-transfected infected cells showed a dramatic reduction in the maturation of the L. infantum LEM 5700-containing phagosomes/PVs and a delay in the differentiation of the insulated parasites (Fig 9B and 9C), as observed in Retro-2-treated infected cells (Figs 2 and 3, and Fig 6A) and nocodazole-treated infected cells (Fig 8A to 8C). Moreover, the fusiform-like parasites hosted in siRNA dynein-transfected infected cells expressed a strong altered surface (Fig 9B and 9C), similar to that of the parasites hosted in Retro-2-treated ( Fig  6B and 6C) and nocodazole-treated (Fig 8D and 8E) infected cells. In addition, RT-qPCR quantification showed that the parasites in siRNA dynein-transfected infected cells died, shown by marked lowering of the amastigote burden (Fig 9D), as found in infected cells treated with Retro-2 ( Fig 6D).

Discussion
Our work provides evidence that the maturation of L. infantum-containing phagosomes in infected macrophages is dependent on microtubule dynamics and highlights the pivotal role of the fusion of parasite-containing phagosomes to the host cell endolysosomal vacuolar

PLOS NEGLECTED TROPICAL DISEASES
compartments to allow the full promastigote-to-amastigote differentiation crucial for parasite survival and development. Our results add to existing knowledge on the maturation process of L. infantum-containing phagosomes/PVs by newly examining the time-dependent appearance/disappearance of two membrane-associated SNAREs, Vti1a and Vti1b, and two membrane-associated lysosomal transporters, GLUT8 and TAPL (Fig 10) in comparison with the appearance/disappearance of the classical bona fide endolysosomal markers Rab7, Lamp-2, and CatD, necessary for the morphological differentiation of L. infantum insulated in tight-

PLOS NEGLECTED TROPICAL DISEASES
fitting PVs hosted in macrophage-like cells. Vacuolar fusion with lysosomes occurs by two processes, the first consisting of a continuous cycle of transient membrane contacts that mainly allow the transfer of the lysosomal luminal content and the second, by full membrane fusion, allowing both the transfer of lysosomal membrane-associated proteins and lipids and luminal content [40]. Desjardins [41] proposed that phagolysosome biogenesis involves heterotypic fusion with host-cell lysosomes by a kiss-and-run mechanism of fusion. The observed acquisition of lysosomal membrane-associated GLUT8 and TAPL by L. infantum-containing PVs, together with the previously observed presence of the lysosomal membrane-associated vATPase in L. donovani-containing PVs [9,10], suggests that a certain amount of lysosomal membrane is probably acquired during membrane fusion. Early acquisition of the lysosomal membrane GLUT8 and TAPL transporters by the L. infantum-containing PVs, followed by their disappearance, is surprising but may be due to the fact that they may be deleterious for the intravacuolar lifestyle of the parasite. Indeed, these transporters function in lysosomes as exporters for the recycling of degradation products [42]. Thus, they may deplete the mature PVs of nutrients useful for parasite growth. Whether Leishmania excludes these lysosomal exporters by the same mechanism that leads to the exclusion of the vATPase from the PV membrane [10] is yet to be demonstrated.
We show that the maturation of L. infantum-containing phagosomes in tight-fitting PVs is MT-and dynein-dependent (Fig 10), but the precise nature of the interaction of the L. infantum-containing phagosomes with the MT-associated dynein remains unclear. It is possible that it interacts with dynein similarly to phagosomes, for which fusion with lysosomes involves the membrane-bound Rab7 GTPase. In this case, following clustering within phagosome membrane microdomains, Rab7 GTPase recruits the MT-associated dynein-dynactin complex that powers minus-end transport along MTs [39,[43][44][45]. As we found here for L. infantum, Rab7 is present at the membrane of tight-fitting PVs that insulate L. major and L. donovani [9,27,32,46]. This GTPase appears to play an important role in the maturation of L. donovanicontaining phagosomes, as they become low-fusogenic when it is missing [32]. It has been elegantly demonstrated that the lipophosphoglycan (LPG) of L. donovani inhibits the motion of parasite-containing phagosomes by altering the membrane-associated microdomains [7,8,10,11,32,47]. Our observation of a MT-and dynein-dependent mechanism controlling the maturation process of L. infantum-containing phagosomes could explain how LPG delays the fusion of L. donovani-containing phagosomes with LEs and lysosomes [7,47]. It is possible that LPG arrests the movement of L. donovani-containing early phagosomes by affecting the known positioning of the dynein motor within the cell membrane microdomains [39], resulting in its disconnection from with the PV membrane-associated Rab7.
Insulated within infected host mammalian cells, Leishmania promastigotes encounter various stressful conditions and environmental changes [48,49] and undergo differentiation, characterized by large morphological and functional changes. The final transformation into amastigotes is crucial, since this form of the parasite is particularly adapted to live within the acidic, hydrolytic, and nutrient-poor environment and resist attacks by the defensive mechanisms of the macrophage [50]. Moreover, metabolic adaptation of the parasite is particularly important, as it includes the downregulation of many surface nutrient transporters and changes in central carbon metabolism [51]. We show that permanent inhibition of the interaction of the L. infantum-containing phagosomes/PVs with host cell LEs and lysosomes delays their maturation, resulting in immature phagosomes/PVs (Fig 10). This situation is extremely deleterious for the insulated parasites, as they do not complete promastigote-to-amastigote differentiation and die. Microbial pathogens and parasites insulated in intracellular vacuoles have developed specific sophisticated strategies to import host-cell cytosol nutrients into the lumen of the pathogen-containing phagosomes/PVs to support their intracellular growth [52].
Insulated Leishmania spp. require a diverse range of host energy sources, including sugars, amino acids, and lipids to survive intracellularly [53]. The acquisition of host-cell nutrients likely occurs through the constant fusion of Leishmania spp.-containing PVs with the host cell lysosomes containing degraded cargo [27,28,31,46], thus providing the carbon sources necessary for growth of the parasite [54,55]. We hypothesize that the observed death of the L. infantum parasite occurs by nutrient starvation. This is based on our observation that pharmacologically induced failure of the MT-dependent fusion between L. infantum-containing phagosomes and lysosomes results in their accumulation in a permanently immature state. This leads to an impoverished nutritional status because of the lack of acquisition of degraded lysosomal products. Further experiments are needed to demonstrate the scarcity of nutrients in the permanently immature phagosomes/PVs, although it is recognized that the direct measurement of nutrient levels in this vacuolar compartment is difficult [54].
The cellular vacuolar trafficking of several toxins, bacterial pathogens, and viruses is negatively affected by the small inhibitor Retro-2 [56]. Knowledge of the regulatory mechanisms that control such trafficking suggests that the inhibitory effects of Retro-2 are the result of its actions on various targets. Stx5-dependent effects [29,57,58] relate to a Retro-2-induced default of Stx5 localization to the TGN because it impairs Sec16A-dependent cycling of this SNAREs between the Golgi and the ER [30] and inhibits the ASNA1-mediated ER targeting and insertion of tail-anchored proteins [59]. Moreover, it promoted a default of the docking of EEs carrying Shiga toxin onto the TGN because the EE-associated Shiga toxin-trafficking chaperone GPP130 can no longer bind to its TGN-associated Stx5 receptor [30]. However, effects of Retro-2 at sites other than the TGN-ER have been reported, such as the blocking of cell entry of Herpes Simplex virus [60] and filovuruses [61], and the decrease of the size of the very large, communal PV insulating L. amazonensis which is formed by fusion with LEs and lysosomes [33]. The observation that Retro-2 disassembles the cellular MT network raises certain questions. Indeed, this network is required for endogenous vacuolar trafficking essential for cellular homeostasis [62,63]. As a consequence, Retro-2 would be expected to have negative pleiotropic effects and yet, this is not the case [29]. Identical questions have been raised [64] concerning the blocking effect of Retro-2 at the ER [30,59]. For the effect on MT network it is possible that Retro-2 selectively affects one or several existing stable and dynamic sub-populations of αβ-tubulin dimers, or stable MTs with numerous post-translational modifications of tubulin, or various cellular pools of MTs [65,66]. Future studies are required to resolve this issue.
Our study and others [33,67,68] demonstrate that Leishmania spp.-containing PVs constitute a druggable target for the treatment of leishmaniasis [69,70]. This is especially important in the context of the current therapeutic arsenal that is rapidly becoming ineffective due to the accelerated propagation of drug resistance in Leishmania spp. [71].

Cell culture and infection and treatment protocols
The mouse monocyte/macrophage-like cell line RAW 264.7 (Cell collection CNRS UMR 8076 BioCis, University Paris-Saclay) [75] was cultured in Dulbecco's modified Eagle's minimal essential medium (DMEM) (Invitrogen Life Technologies) supplemented with 10% heat-inactivated fetal bovine serum, at 37˚C in a humidified atmosphere containing 5% CO 2 . For infection, RAW 264.7 cells were placed in 24-well TPP tissue culture plates (ATGC, Marne la Vallée, France), with or without sterile coverslips, and infected with unopsonized, L. infantum LEM 5700 promastigotes (MOI of 10 parasites per cell) at 37˚C in an atmosphere containing 5% CO 2 . For treatments, cells were infected for the indicated time-periods in the continuous presence, or not, of compounds. At the indicated time-periods post-infection, samples were washed three times with cold EBSS to remove free parasites and treated for immunofluorescence labeling. Observation and image acquisition of fixed macrophage-like cultures by confocal microscopy were performed at various times after infection, ranging from 4 to 24 h, depending on the experiment.
HeLa cells and HeLa cells stably transfected with rat GFP-LC3 (GFP-LC3-expressing HeLa cells) were kindly provided by A.M. Tolkovsky (Department of Biochemistry, University of Cambridge, UK) [76]. Cells were seeded and grown in culture plates (TPP, ATGC Biotechnologie, Noisy Le Grand, France) containing coverslips and cultured in RPMI-1640 with L-glutamine (Life Technologies, Cergy, France) at 37˚C in an atmosphere containing 5% CO 2 . For autophagy induction by amino-acid deprivation, the cells were incubated for the indicated times in Earle's balanced salt solution (EBSS) (Sigma) at 37˚C in an atmosphere containing 5% CO 2 [36].

Immunofluorescence labelling
L. infantum LEM 5700 and components of L. infantum LEM 5700-containing PVs were identified by indirect immunofluorescence labelling. Specimens were fixed with 3% paraformaldehyde in phosphate-buffered saline (PBS) for 5 min at room temperature, washed three times with PBS, treated with PBS containing 50 mM NH 4 Cl for 10 min to neutralize the aldehyde, and blocked by adding PBS containing 0.2% gelatin. Cells were permeabilized by incubation in 0.2% Triton X-100 in PBS for 4 min at room temperature and then washed three times with PBS. Intramacrophage L. infantum promastigotes and amastigotes were immunolabeled with mAb 2A3-26 (1/500). Immunolabeling of endolysosomal pathway markers was performed Microtubules and the actin cytoskeleton were immunolabeled with anti-α-tubulin (1/500) and anti-actin (1/100) antibodies, respectively. After immunolabeling with primary antibodies, the specimens were developed using appropriate conjugated secondary antibodies (Jackson Immunoresearch). Cells were also stained for 15 min with DAPI (100 μg/ml) (Invitrogen Life Technologies) to visualize the nucleus. Coverslips were mounted using Dako fluorescent mounting medium (Invitrogen Life Technologies) for CLSM examination.

LysoTracker Red labeling
Cells were labeled with LysoTracker Red (50 nM) by incubating them with the probe for 60 min. The labeled cells were washed three times with sterile PBS, and then fixed with 3% paraformaldehyde in PBS for 5 min at room temperature. Coverslips were mounted using Dako fluorescent mounting medium.

Confocal laser scanning microscopy (CLSM)
The samples were imaged with an inverted STED-gated Leica TCS SP8 microscope (Leica, Germany) using a HC PL APO CS2 63x/1.40 oil immersion objective lens. The instrument was equipped with a 405-nm diode for DAPI excitation and a WLL Laser. Blue, green, and red fluorescence emission was collected using 410-460, 505-550, and 560-760 nm wide emission slits in sequential mode, respectively. The pinhole was set to 1.0 Airy unit, giving an optical slice thickness of 0.89 μm. Twelve-bit numerical confocal micrographs were processed using Leica SP8 LAS X software (Version 2.0.1; Leica, Germany). Confocal optical sectioning was performed each 0.3 μm along the z axis.

Multidimensional imaging analysis and quantification
Three-dimensional volume renders (3D-reconstruction surface rendering) from confocal microscopy optical sections (z-stack) were obtained using Imaris Measurement Pro software (Bitplane AG, Zurich, Switzerland). Isosurface rendering was used to measure the volume of identified structures. To measure the volume of a defined object, confocal z-stack images (row data following the Nyquist sampling theorem) were used to perform 3D reconstruction views. The resulting volume rendering data sets were then processed to obtain a surface rendering calculation based on voxel intensities. A smoothing Gaussian filter of 0.09 was applied before thresholding. Imaris provides volumes and surfaces of the structure. Each channel was analyzed separately. Relative fluorescence intensities (RFIs) were determined on maximum-intensity projections using z-stacks. The level of a given host cell marker associated with an intramacrophage L. infantum parasite was quantified using the "segmented line" tool was to cleanly delineate the area relative to the immunolabeled parasite in the z-stack micrographs. Once the area of interest was determined, the "Plot profile" tool of ImageJ was used to obtain the RFIs (in arbitrary units) of the parasite and associated marker, given by Channel #1 or #2, respectively, of the confocal acquisition system. This procedure can reliably estimate the level of an endolysosomal marker associated with intramacrophage L. infantum-containing PVs. Cropping of movies was performed in ImageJ (ImageJ, NIH). The percentage of intramacrophage parasites positively associated with a host cell endolysosomal marker was determined manually using Imaris software. Quantitative values obtained for host cell and parasite numbers were exported to Excel for further analysis and graphical representations. All imaging analyses and quantifications were performed blindly to eliminate any possible bias.

Western-blot analysis
Cells are washed once with cold PBS and then treated for 15 min at 4˚C with extraction buffer (25 mM HEPES, 0.5% Triton, 150 mM NaCl, 2 mM EDTA) containing proteases and phosphatase inhibitors. Protein fractions were dissolved in the appropriate volume of Laemlli buffer and incubated at 100˚C for 5 min. The proteins were immediately separated on 12% SDS-polyacrylamide gels. For western-blot analysis, gels were transferred to polyvinylidene difluoride membranes (Perkin Elmer, Les Ullis, France). A primary rabbit anti-LC3 antibody was used to reveal LC3-I and LC3-II proteins. A primary mouse antibody specific for β-actin was used to verify the equal loading of lanes. Primary antibodies were revealed with anti-rabbit-or antimouse-peroxidase secondary antibodies. The abundance of LC3-II and β-actin proteins in the western blots was quantified by densitometry using ImageJ software.

Quantitative real-time PCR (RT-qPCR)
Total RNA was extracted from cells using TRIzol reagent (ThermoFisher Scientific) following the manufacturer's protocol. The level of L. infantum parasites in macrophage-like RAW 264. Amastin RT-qPCR was performed in a 20-μl volume with 2 μl template DNA under the same conditions as those already described using specific primers (forward [5'-GCCGTTCTTGA GGTTGGTT-3]' and reverse [5'-CGCTCGACGTGTTGATCT-T-3']). A clinical isolate typed L. infantum MON-1, kindly provided by S. Houze (AP-HP, Bichat-Claude Bernard hospital, Paris, France), was used as an amastigote positive control and cultured promastigotes as amastigote negative controls. The ddCt method was used to calculate the level of total parasites and amastigote load in macrophages. Genomic DNA isolated from L. infantum strain MON-1 served as the quantification standard for the Licytb qPCR assay. We considered 150 ng of leishmanial DNA to be equivalent to 1.5 x 10 6 parasites based on the conversion between the quantification of leishmanial DNA and parasites. This equivalence was used to prepare the standard curves based on serial dilutions of the Leishmania DNA standard in nuclease-free water, corresponding to a range of 5 × 10 7 to 5 × 10 −3 parasites, for which the equation was y = -1.431n(x) + 22.889 and R 2 = 0.9958.
The determination of dynein RNA levels in non-transfected and siRNA-transfected macrophage-like RAW 264.7 cells was performed by RT-qPCR using specific dynein primers (forward [5'-TACGCTGGCTACTTTGACCA -3'] and reverse [5'-CGTTCGTCAGCATTGGA GAG-3']). Data were normalized against that of the housekeeping gene GAPDH (forward [5'-CAAGAAGGT-GGTGAAGCAGG-3'] and reverse [5'-GCATCGAAGGTGGAAGAGTG-3']). One microgram of extracted RNA was treated with DNase I (Amplification Grade, Life Technologies) to eliminate the risk of contamination by mouse DNA before the RT reaction was carried out using the Superscript VILO commercial kit (Life Technologies). The RT-qPCR reaction consisted of a mixture of 20 μl, including 2 μl cDNA, 10 μl SensiFAST SYBR No Rox Mix (Bioline, Paris, France) and 400 nM primers. The PCR amplification conditions consisted of an incubation cycle of 2 min at 95˚C, followed by 40 cycles of amplification of 10 s at 95˚C, 10 s at 60˚C, and 20 s at 72˚C. A melting curve cycle was performed at the end of the cycle to verify the specificity of the amplicon.
RT-qPCR determinations were monitored using the Via 7 Real-Time PCR System (ViA 7 system, ThermoFisher Scientific). The efficiency of PCR amplification was evaluated for each pair of primer and was > 1.9.

Analysis of microtubule polymerization in vitro
Compounds were tested using the Tubulin Polymerization BK004P kit as specified by the manufacturer (Cytoskeleton, Inc., Denver, CO). Briefly, tubulin protein (> 99% purity) was suspended (300 μg/sample) in 100 μl G-PEM buffer (80 mM PIPES, 2 mM MgCl 2 , 0.5 mM EGTA, 1.0 mM GTP, pH 6.9) plus 5% glycerol, with or without the test compound, at 4˚C. Then, the sample mixture was transferred to a pre-warmed 96-well plate and polymerization measured by the change in absorbance at 340 nm every 1 min for 60 min (SpectraMAX Plus; Molecular Devices, Inc., Sunnyvale, CA) at 37˚C.

Presentation of the data and statistics
Each experiment was performed at least two times in duplicate or triplicate. Representative confocal micrographs or movies of typical cells were used for the Fig. For the final Fig, the presented confocal micrographs were resized using Adobe Photoshop CS6 software (San Jose, CA). Quantitative data are presented as the means ± standard error of the mean (SEM). Graphs were produced using Microsoft Excel software. Comparisons between the experimental groups were performed using the unpaired Student t-test. Significance was established when P < 0.01.