Cell-to-cell transfer of Leishmania amazonensis amastigotes is mediated by immunomodulatory LAMP-rich parasitophorous extrusions.

The last step of Leishmania intracellular life cycle is the egress of amastigotes from the host cell and their uptake by adjacent cells. Using multidimensional live imaging of long-term-infected macrophage cultures we observed that Leishmania amazonensis amastigotes were transferred from cell to cell when the donor host macrophage delivers warning signs of imminent apoptosis. They were extruded from the macrophage within zeiotic structures (membrane blebs, an apoptotic feature) rich in phagolysosomal membrane components. The extrusions containing amastigotes were selectively internalized by vicinal macrophages and the rescued amastigotes remain viable in recipient macrophages. Host cell apoptosis induced by micro-irradiation of infected macrophage nuclei promoted amastigotes extrusion, which were rescued by non-irradiated vicinal macrophages. Using amastigotes isolated from LAMP1/LAMP2 knockout fibroblasts, we observed that the presence of these lysosomal components on amastigotes increases interleukin 10 production. Enclosed within host cell membranes, amastigotes can be transferred from cell to cell without full exposure to the extracellular milieu, what represents an important strategy developed by the parasite to evade host immune system.

Introduction

Leishmania infections, which affect around 2 million people globally each year (WHO, 2010), are transmitted to vertebrate hosts by infected insect vectors. In the infected mammalian host, Leishmania are predominantly sheltered within macrophage-like cells. Thus, the mechanism involved in their macrophage-to-macrophage transfer in the cutaneous or visceral lesions is an important area of study. However, the steps of the intracellular life cycle in mammalian hosts that involve the obligatory egress of Leishmania amastigote forms from host cells in order to the spread to new host cells and other tissues (tropism) and organisms are likely the least known aspect of the biology of this parasitic protozoan. A search of the early literature revealed that authors emphasized a ‘lytic’ cycle for this parasite, mainly based on histopathological observations fragmented in space and time (Theodorides, 1997; Dedet, 2007; Florentino et al., 2014). The preferential, almost exclusive, presence of oval-shaped parasites inside macrophages was the most puzzling aspect of leishmaniasis for the first histopathologists who studied the organism (Wright, 1903; Christophers, 1904), highlighting the uniqueness of Leishmania cell infection and supporting a concept of a specialized parasite, with a limited repertoire of cells able to host them. For decades, leishmaniasis was considered a disease almost exclusively of the host macrophage system (Meleney, 1925; Heyneman, 1971).

The first attempt to unveil Leishmania egress from infected host cells appears to be one study published in 1980, in which parasites were observed lying free on the edge of cellular infiltration as product of host cell lysis (Ridley, 1980). Macrophage lysis or the presence of extracellular amastigotes were not observed in infected tissues presenting decreased inflammatory response. These findings suggested that amastigote release is a consequence of the cytolytic environment modulated by host immune response and may be not actively promoted by parasites. The egress of amastigotes was revisited in the literature in the late 1990s (Rittig et al., 1998; Rittig and Bogdan, 2000). The long-term observation of cells hosting Leishmania major by live microscopy revealed that ‘after several uneventful days, small vacuoles suddenly accumulated asymmetrically at the periphery of the infected phagocytes’ where amastigotes were ‘constantly released over a period of several hours, leaving the somewhat shrivelled remnants of their host cells’. An alternative view of parasite egress was proposed, in which amastigotes would be released in a synchronized fashion, through an exocytosis-like process; it assumes that Leishmania egress does not necessarily require host cell lysis by an amastigote multiplication burst.

In this report, using live imaging microscopic evidence, we revisited and further investigated the previously described amastigote exit from host cells (Rittig et al., 1998; Rittig and Bogdan, 2000); we propose that host cell exit of Leishmania takes place from damaged host cells, in a process mediated by parasitophorous extrusions. These structures fully or partially surrounded amastigotes and were rich in host phagolysosomal components, especially lysosome-associated membrane proteins (LAMPs), which stimulated the production of anti-inflammatory cytokines.

Results

Amastigotes are transferred from cell to cell during host cell death

The continuous live cell recordings of bone marrow-derived macrophages (BMDMØ) infected with Leishmania amazonensis-DsRed2 amastigotes for the first 48 h of infection in vitro did not provide evidence of cell-to-cell transference of the intracellular form of the parasite (Real and Mortara, 2012). We decided to examine L. amazonensis-DsRed2 cell-to-cell transfer in long-term cultivated BMDMØ (> 4 days) considering that (i) these cells can be cultivated in vitro for several days, with minimum multiplication (Rabinovitch and De Stefano, 1973; Eischen et al., 1991), (ii) some host macrophages will display a controlled cell death programme in prolonged cultivation, and (iii) an exhausted host cell population, identified by controlled cell death markers, would drive amastigote exit and transference between cells.

First, the pro-apoptotic Bax gene expression assessed by qPCR as a marker of ongoing apoptotic process in macrophage cultures was studied to define the time window when amastigote cell-to-cell transfers would potentially occur. The expression of Bax is increased after 4 and 10 days of cultivation in infected and non-infected cultures (Fig. 1A). Second, we co-incubated BMDMØ that had been infected in vitro for 20 days with L. amazonensis-DsRed2 with fresh uninfected RAW 264.7 macrophage-like cells. RAW cells were observed to selectively take-up, or rescue, extruded amastigotes from damaged infected macrophages (Fig. 1B, Video S1A). Recipient RAW macrophages interact with donor BMDMØ in the imminence of apoptosis process; the long-term infected BMDMØ shrunk and became zeiotic, forming membrane blebs (Fig. 1B, arrowheads). Amastigotes rescued by adjacent BMDMØ or by fresh RAW cells survive in the new environment and develop spacious PVs in the recipient host cytoplasm. Recipient RAW cells or BMDMØ usually migrate towards donor host macrophages, extending their pseudopodia to and selectively reaching extruded amastigotes (Video 1A at time point 1d05:45, and Fig. 1C). Amastigote transfers between non-affected viable cells were not observed. Using BMDMØ-GFP infected with L. amazonensis-DsRed2, we observed the host cell and spacious PV shrinkage, with concomitant relocation of some amastigotes into the zeiotic structures (Fig. 1D arrowheads and Video S1B).

Fig 1

Cell-to-cell transfer of Leishmania amazonensis amastigotes occurred after host cell death.A. Pro-apoptotic Bax gene mRNA expression measured by qPCR in infected or non-infected BMDMØ after 4 h, 4 and 10 days after L. amazonensis infection. The data are presented as the relative quantification 2−ΔΔCt against β-actin gene expression. There was an increase in Bax expression after 4 and 10 days, independent of infection. anova, P < 0.05.B. Live cell imaging of BMDMØ infected with L. amazonensis-DsRed2 (red) for 30 days in vitro and co-cultured with uninfected RAW 264.7 macrophage-like cells. RAW cell interacts with infected BMDMØ and rescue several amastigotes after macrophage collapse. Host cell death presented zeiosis (arrowheads), a typical feature of apoptosis. The time of image acquisition is represented by days:hours:minutes:seconds:milliseconds (d:hh:mm:ss:sss). Image acquisition started after 2 h of RAW cell addition. Bar = 10 μm.C. Field-emission scanning electron microscopy (FE-SEM) of a BMDMØ culture infected with L. amazonensis-DsRed2 for 20 days, showing a macrophage (coloured in blue) interacting with an oval-shaped structure (red) extruded from a vicinal macrophage. This structure presents dimensions compatible with an amastigote. Bar = 10 μm.D. Amastigotes (red) are extruded from dying BMDMØ-GFP (green) within zeiotic structures (blebs). Images are representative of two z-stacks of the same macrophage and the arrowheads indicate parasite-containing extrusions. Bar = 10 μm. Results are representative of two experiments.

Our observation of BMDMØ that had been infected in vitro for 15 days with L. amazonensis-DsRed2 followed by additional 3 days of live microscopy revealed several cases in which amastigotes were rescued from affected macrophage (donor cells) by vicinal, recipient cells of the same lineage (Fig. 2A, Video S2). Transfers occurred in the first hours of host cell alteration and involve the mobilization of recipient cells towards donor cells. These events were quantified in live cell recordings, in parallel with automatic cell quantification based on Hoechst stain of macrophage nuclei (Fig. 2B, Video S2). Although the numbers of infected and non-infected macrophages decreased at similar rates, the live cell recordings from 15 to 18 days post-infection started with a higher number of cells per field in the infected sample (Fig. 2B, graph). Quantitative PCR for the expression of anti-apoptotic gene mRNAs (IGF-I and Bcl-2) revealed that during long-term cultivation (10 days in vitro) infected samples express more anti-apoptotic gene mRNAs than non-infected cultures (Fig. 2C). During the additional 70 h of infection in which macrophage live imaging was recorded, a median of 12 amastigote transfers per field was counted, occurring throughout the entire acquisition period (Fig. 2D).

Fig 2

Quantification of amastigote transfer events and host cell numbers during multidimensional acquisition of BMDMØ infected with L. amazonensis-DsRed2.A. Example of L. amazonensis-DsRed2 cell-to-cell transfer recorded between BMDMØ infected for 15 days. The image shows the macrophages involved in transfer (arrowheads), classified as donor (a macrophage from which parasites are extruded and transferred after host cell death) and recipient (viable macrophages in the vicinity which rescue extruded parasites from donor). Upper panel: at time point 1d07:45 amastigotes are relocated to a macrophage extrusion; recipient macrophages take up these extruded parasites at time points 1d08:45 and 1d09:00 (asterisks indicate regions of parasite transfers). L. amazonensis-DsRed2 in red, Hoechst 33342 staining in cyan channel. Lower panel: amastigotes were tracked during transfer, in an interval of 2 h after amastigote extrusion; parasites displacement in x, y and z dimensions is depicted as tracing lines in each time point. Tracing lines display a colorimetric range, relative to the start (extrusion, t = 0 h, time point 1d07:45, black-blue tracing) and the end of tracking (complete transfer, time point 1d09:45, yellow-white tracing). Macrophage nuclei (Hoechst 33342 staining) in cyan, software-detected amastigotes (isospots) in red squares. Displayed time corresponds to image acquisition elapsed time, after 15 days of infection. Bars = 20 μm.B. Example of a microscopic field of infected culture showing in the first DIC image merged with red (L. amazonensis-DsRed2) and blue (Hoechst 33342) fluorescence channels and in the second image the automated recognition of macrophage nuclei by software analysis (yellow = counting hit), employed to quantify the number of host cells during multidimensional imaging. This automatic quantification was applied to 10 multidimensional images of infected and non-infected cultures, acquired after 15 of infection, for additional 70 h. Right panel (graph): quantified nuclei of infected and non-infected BMDMØ cultures were plotted in the graph, which shows the mean (with standard error) number of macrophage nuclei per microscopic field (40× objective, 228 × 228 μm).C. Anti-apoptotic genes IGF-I and Bcl-2 mRNA expression measured by qPCR in infected or non-infected BMDMØ after 4 h, 4 and 10 days after L. amazonensis infection. The data are presented as the relative quantification 2−ΔΔCt against β-actin gene expression. Infected cell cultures express higher levels of IGF-I and Bcl-2 after 10 days of infection. anova, P < 0.05.D. Quantification of amastigote transfer events (red dots) per microscopic field (40× objective, 228 × 228 μm, n = 9). A median of 12 transfers per field was observed in live cell recordings.

Laser-induced cell death of infected macrophages hasten the amastigote extrusion and transmission

Considering that amastigotes exit from apoptotic-like macrophages, we investigated whether induction of host cell apoptosis would trigger L. amazonensis extrusion and its rescue by adjacent macrophages. Our challenge was to selectively induce apoptosis in some cells, sparing vicinal macrophages from death in order to allow viable cells to rescue extruded amastigotes. To do so, the nuclei of some infected BMDMØ-GFP in the microscopic field were micro-irradiated by near UV laser (405 nm). A concentrated pulse of UV lasers (351 nm and 364 nm) on the HeLa cell nucleus is known to induce the destruction of these cells via apoptosis (Soustelle et al., 2008).

Figure 3A and Supplementary Video S3 demonstrate that, in the same microscopic field, non-irradiated macrophages were able to rescue amastigotes from apoptotic-like irradiated cells, suggesting that amastigotes benefit from host cell death to spread among other cells. Amastigotes were rescued by adjacent macrophages in 21 out of 31 apoptotic, micro-irradiated BMDMØ (total of 47 micro-irradiated host cells, n = 3 experiments). Amastigote transfer started after 6 h post-micro-irradiation. The morphology of macrophage nuclei during L. amazonensis extrusion was followed in the multidimensional images of Hoechst staining; nuclei shrinkage and condensation of chromatin in the periphery of the nuclei were observed. Additionally, cell shrinkage and zeiosis were also observed, suggesting an apoptosis-like host cell death in the process of amastigote transfer. In experiments employing 48 h infected BMDMØ micro-irradiated in the presence of the fluorescent probes YO-PRO-1 and Annexin V-CF633, we confirmed that micro-irradiation of BMDMØ nuclei induced apoptosis (Annexin-positive/YO-PRO-1-negative staining) then late necrosis (double positive). Figure 3B and Supplementary Video S4 show one case in which an infected BMDMØ, micro-irradiated for 7 min enters apoptosis 12 h post-irradiation and late necrosis 1 h thereafter (Fig. 3B). An extruded, annexin-positive L. amazonensis-DsRed2-containing PV is observed during macrophage apoptosis (arrowhead).

Fig 3

Micro-irradiation of host cell nuclei induced cell death and amastigote transfer.A. Time-lapse frames of a micro-irradiated BMDMØ-GFP transferring amastigotes (arrowheads) to non-irradiated macrophages. Upper panels: Leishmania amazonensis-DsRed2 in red and Hoechst 33342 in cyan, merged with DIC image. Lower panels: green (GFP signal from macrophages) and red channels of the same region, showing GFP leakage of the micro-irradiated cells (contour) during amastigote transfers (arrowhead). The micro-irradiation spot is indicated by a white circle. Image acquisition started after 120 h of infection plus 5 min of micro-irradiation.B. Multidimensional live imaging of micro-irradiated, 48 h infected BMDMØ in the presence of the phosphatidylserine (PS) probe Annexin V (conjugated to CF633 flurophore, cyan) and the cell membrane-impermeant nucleic probe YO-PRO-1 Iodide (yellow). At time point 0d12:53, during macrophage apoptosis (Annexin V-positive, YO-PRO-1-negative staining), L. amazonensis–DsRed2 amastigotes (red) are relocated to zeiotic structures which expose PS (arrowhead); the host cell starts necrosis around 0d14:00 (YO-PRO-1 staining is indicative of loss of membrane integrity). Time of image acquisition is represented by days:hours:minutes (d:hh:mm). Bar = 10 μm.

Our observation of micro-irradiation-induced amastigote transfers suggests that these events occurred during macrophage GFP leakage (Fig. 3A, white cell contours), evidencing the lack of host cell membrane integrity. To examine if damage to infected macrophage plasma membrane could trigger amastigote transfer, long-term infected BMDMØ-GFP cultures were treated with the bacterial toxin SLO, which is known to form pores on cholesterol-containing membranes (Tweten, 2005). Since the membrane of amastigotes contains ergosterol instead of cholesterol, SLO should not be active on the membrane of the parasites (Fernandes et al., 2011). The quantification of amastigote transfer events by the observation of multidimensional images of infected BMDMØ-GFP treated with SLO revealed an increase in amastigote transfer events when compared with non-treated controls (Supplementary Fig. S1A and B, Video S5A). We also treated infected BMDMØ-GFP with nocodazole, which interferes in microtubule polymerization and has been shown to induce the extrusion of the intracellular bacteria Chlamydia (Hybiske and Stephens, 2007). In the majority of macrophages, nocodazole induced the formation of unstable blebs containing amastigotes but neither host cell lysis, as attested by retention of GFP fluorescence in the macrophages, nor cell-to-cell transfer of amastigotes were increased in comparison to untreated control (Supplementary Fig. S1B and C, Video S5B). Classical activation of the macrophages by treatment with IFN-γ and LPS did not trigger amastigote extrusions or transfers.

Amastigotes are transferred in association with phagolysosomal components

Although host cell pores might be involved in amastigote exit, the parasite remains associated with the host membrane during extrusion and transfer. The spacious PVs developed by L. amazonensis are rich in phagolysosomal membrane markers such as LAMP and Rab7 proteins, and there is an intimate interaction between amastigotes of this species and the PV membrane (Courret et al., 2002; Real and Mortara, 2012). Thus, we investigated if the PV membrane markers LAMP1 and Rab7 would remain associated with the amastigote surface during amastigote extrusion and rescue by adjacent macrophages. In these experiments, we infected LAMP1-GFP- or Rab7-GFP-transfected RAW macrophages with L. amazonensis-DsRed2 amastigotes and recorded events of parasite extrusion by multidimensional imaging (Fig. 4A and B, Video S6A–D). We observed the extrusion of amastigotes fully covered by LAMP1 or Rab7 proteins and the transfer of amastigotes that remained associated with (fully or partially covered by) these phagolysosome/PV markers (Fig. 4A, arrowheads). LAMP1 or Rab7 proteins surround extruded amastigotes for minutes, or even hours, after host macrophage zeiosis; then these phagolysosomal markers concentrate on amastigote in a polarized fashion (Fig. 4B, arrowhead). When rescued by adjacent macrophages, amastigotes maintained these phagolysosome markers but not F-actin when actin-RFP-transfected RAW macrophages were investigated under the same experimental conditions (Video S6E).

Fig 4

Amastigotes were transferred from cell to cell associated with phagolysosomal components.A. Multidimensional imaging of RAW 264.7 cells expressing GFP-tagged LAMP1 (in green) and infected with Leishmania amazonensis-DsRed2 (red) for 24 h. Arrowheads indicate an extruded amastigote rescued by vicinal macrophage. Time of image acquisition is represented by days:hours:minutes (d:hh:mm). Image acquisition started after 24 h of infection plus 2 min of micro-irradiation. Bars = 10 μm.B. The temporal sequence of the extrusion event presented in A shows that amastigote is surrounded by LAMP1 (hue-saturation-value filter, colorimetric scale relative to mean fluorescence intensity) during the first hours of extrusion. Later, the extrusion concentrates LAMP1 on an amastigote pole (arrowhead). Relative time of extrusion is represented by hours:minutes (h:mm). Extruded amastigote is indicated by an isosurface (grey) constructed by imaging software, based on DsRed2 signal. Bar = 5 μm.C. TEM of two amastigotes (arrowheads) extruded from an apoptotic macrophage. The host cell presents apoptotic features such as shrinkage and peripheral chromatin condensation in a deformed nucleus. Amastigotes are fully (left arrowhead) or partially (right arrowhead) exposed to extracellular milieu. Bar = 2 μm.D. Correlative imaging of infected macrophages extruding amastigotes. The image presents two examples (upper and lower panels) of macrophages observed under confocal and then FE-SEM. For the observation by confocal microscopy, cells were immunostained with combined LAMP1/LAMP2 antibodies (red channels), L. amazonensis-specific antibody (2A3-26, green channels) and stained with DAPI (blue channel). Extruded amastigotes are indicated by letters a, b and c. In the lower panel, the confocal images are presented in different focal planes to show the indicated amastigote extrusions. The membrane covering the extruding amastigotes (A) is observed at higher magnification and resolution in these two examples and displays a smooth aspect with no evident pores. The same extrusion observed by confocal microscopy revealed that extruded amastigotes are fully covered with LAMPs. Bars = 2 μm.Results are representative of four (A–B) and two (C–D) experiments.

The ultrastructure of amastigote extrusions confirmed an association between L. amazonensis extrusion and host cell membrane. The parasite posterior pole faced the macrophage and was generally associated with macrophage components; the anterior pole faced extracellular milieu when not fully covered by macrophage membrane (Fig. 4C). We observed amastigote extrusion in correlative images of confocal and scanning electron microscopy (Fig. 4D). We found that amastigotes extruded from affected macrophages were fully covered by LAMP1/LAMP2 phagolysosomal proteins, and the amastigote-containing extrusions displayed a smooth surface in contrast to the usually ruffled macrophage surface and with no apparent host membrane pores (Fig. 4D, insets). Amastigotes were wrapped so tightly by host membrane that the extrusions took on the morphology of the parasite (Fig. 4D, extruded parasites are indicated by letters a, b and c). Host macrophages in late necrosis were usually associated with fully exposed amastigotes but some extruded amastigotes still associated with LAMP1 (Supplementary Fig. S2A and B, arrowheads, b). Isotype controls for LAMP1/LAMP2 immunostaining confirmed antibody specificity (Supplementary Fig. S2C and D).

Fibroblast-derived amastigotes displaying LAMP on their surfaces stimulate the production of IL-10 by macrophages

In addition, we examined the importance of PV components on L. amazonensis amastigote surface for parasite establishment within recipient macrophages. These host cell components associated with amastigotes could modulate anti-inflammatory immune responses such as IL-10 and TGF-β production, in a fashion similar to the apoptotic mimicry described in L. amazonensis infections (Wanderley et al., 2006).

To obtain isolated amastigotes associated with PV membranes, parasites were collected after mechanical rupture of host cell cultures or footpad lesion tissues. Host cell remnants, including lysosomal membrane proteins, were previously described to be associated with amastigotes after tissue rupture (Saraiva et al., 1983; Winter et al., 1994). When isolated from footpad lesions (2 months after inoculum), around 20% of amastigotes remained associated with LAMPs. Incubation of amastigotes under agitation for 2 h decreases this percentage to approximately 4% (Supplementary Fig. S3A).

Amastigotes were also covered by host cell membranes rich in LAMP when isolated from mouse embryonic fibroblasts (MEF) after 48 h of transient infection (Supplementary Fig. S3B and C, Fig. 5A). LAMP immunostaining was associated with amastigote posterior poles as observed by confocal microscopy; scanning electron microscopy (SEM) confirmed the presence of remnants associated with the parasite poles, similar to caps. Amastigotes isolated from LAMP1/LAMP2 double-knockout MEF did not display these lysosomal components on their surfaces but conserved host cell debris on posterior pole (Fig. 5B). The host debris or caps on the amastigote surface were of lysosomal origin, as revealed by the immunostaining of another lysosomal component, lysosomal integral membrane protein-1 (LIMP1).

Fig 5

Fibroblast-derived amastigotes associated with LAMP stimulate interleukin-10 (IL-10) production by macrophages.A. L. amazonensis-DsRed2 amastigotes (red) isolated from wild-type mouse embryonic fibroblasts (MEF-WT) preserved LAMP1 and LAMP2 proteins (immunostained with LAMP1/LAMP2 antibodies, green) on their surfaces. First image: DIC merged with green, red and blue (DAPI) channels, bar = 5 μm. Second and third images: three-dimensional reconstructions of an amastigote showing LAMP cap at the parasite posterior pole (arrow). The position of parasite nucleus (N) and kinetoplast (k) indicate amastigote polarity. Bar = 1 μm. Inset: field-emission scanning electron microscopy (FE-SEM) of an amastigote and associated membrane debris (arrow) at the posterior pole, bar = 2 μm.B. L. amazonensis-DsRed2 amastigotes (red) isolated from LAMP1/LAMP2 knockout MEF. First image show DIC merged green, red and blue (DAPI) channels of amastigotes incubated with LAMP1/LAMP2 antibodies (green): amastigotes are devoid of a LAMP cap on their surfaces, bar = 5 μm; second image acquired by FE-SEM reveals membrane debris at their posterior poles, bar = 2 μm. Third and fourth images: immunostaining of LIMP1 (green) revealed that the membrane cap on the amastigote surface conserves lysosomal components; three-dimensional reconstruction of amastigotes (fourth image) shows LIMP cap on parasite poles (arrows), bar = 5 μm.C. Flow cytometry quantification of the percentage of amastigotes positive for LIMP-1. Amastigotes were isolated from MEF-WT or MEF-LAMP1/2KO, after 48 h of infection. LIMP-1 is presented on approximately 20% of amastigotes (P > 0.05, anova).D. IL-10 production by BMDMØ cultures infected with amastigotes isolated from MEF-WT (LAMP-associated) or from MEF-LAMP1/LAMP2 KO (LAMP-devoid amastigotes); medium alone, non-infected macrophage cultures and amastigotes cultivated under agitation for 2 h (stirred amastigotes) were employed as experimental controls. LAMP-associated amastigotes stimulate a higher production of IL-10 when compared with LAMP-devoid parasites, which, in turn, induce a higher production of the cytokine when compared with control groups. anova, P < 0.05. Data representative of three independent experiments.

We quantified the percentage of MEF-derived amastigotes associated with PV components by flow cytometry, using LIMP1 as PV marker. Around 25% of amastigotes were LIMP1-positive, indicating association with host cell PV membranes; this percentage was found in amastigotes isolated from MEF-WT and MEF-LAMP1/2 KO (Fig. 5C). Flow cytometry quantification of LIMP1-positive amastigotes employed antibody isotype and secondary immunostaining controls for LIMP1 (Supplementary Figs S2E and F and S3D). Thus, we obtained preparations of amastigotes associated with host phagolysosomal components (LIMP1 positive), mimicking parasite-containing host cell extrusions, which display or are devoid of LAMP1/LAMP2 proteins (after amastigote isolation from MEF-WT and MEF-LAMP1/2KO respectively).

These preparations of LAMP-positive or LAMP-negative amastigotes were added to fresh BMDMØ cultures and IL-10 production was assessed (Fig. 5D and E). When interacting with amastigotes displaying LAMP1/LAMP2-rich host membrane caps, macrophages produce higher amounts of IL-10 when compared with the interaction with amastigotes devoid of LAMP (Fig. 5D). Stirred amastigotes (cultivated under mild agitation), devoid of host membrane caps, induced a lower IL-10 production (comparable to non-infected macrophages). In some experiments LAMP-devoid amastigotes induced an intermediate IL-10 production by macrophages, not so pronounced as that induced by LAMP-positive amastigotes but higher than baseline production of non-infected macrophages. This suggests that additional host cell membrane components attached to amastigotes could stimulate IL-10 production by macrophages.

Discussion

The spreading of amastigotes to vicinal cells and tissues represent a much-unexplored topic in Leishmania biology. Here we demonstrate that L. amazonensis amastigotes are extruded from affected macrophages presenting apoptotic-like features and are often rescued by vicinal macrophages in vitro. We propose that this intracellular parasite carries host cell components during its extrusion, an event that has important implications for the pathogenesis of Leishmania and possibly other intracellular parasites adapted to and interactive with parasitophorous vacuoles. By conserving host cell components on their surfaces, parasites could evade host immune systems, modulating cytokine production.

Cell-to-cell transfer of Leishmania could be the preferential way of parasite spreading in vivo. The exchange of amastigotes between host cells would occur in a cascade of events, involving diversified cell lineages and culminating in the intracellular growth within macrophages. Peters and colleagues studied the context of transfer of L. major promastigote forms from the sand fly Phlebotomus duboscqi to mice (Peters et al., 2008). They observed by multiphoton intravital microscopy (MP-IVM) the migratory behaviour of host neutrophils at the site of insect bite and concluded that these cells are essential for the establishment of the parasite after host-to-host transfer. Additionally, they present flow cytometry data showing that parasites once hosted by neutrophils after 18 h of post-intradermal inoculation of promastigotes were found in macrophages after 6 days. Ng and colleagues using the same live imaging techniques have shown that dendritic cells responded to L. major infection in a similar manner (Ng et al., 2008). Thus, it is possible to conceive neutrophils, dendritic cells and other host cells as amastigote deliverers to macrophages, sheltering the parasite in customized PVs until the transmission. For instance, Leishmania donovani promastigotes are maintained in non-lytic compartments of neutrophils and are transferred to macrophages when the latter internalize infected and apoptotic neutrophils, a process regarded as a Trojan horse tactic (Gueirard et al., 2008). These data suggest that cell-to-cell parasite transfer without exposure to extracellular milieu could represent an important requirement, at least in the first hours of infection, for parasite establishment.

We present evidence that, although unable to halt the expression of the pro-apoptotic Bax gene in long-term cultivated macrophages, L. amazonensis amastigotes increased the expression of anti-apoptotic genes such as Bcl-2 and IGF-I. Our results suggest an elegant strategy of Leishmania intracellular parasitism: delay or inhibit host cell apoptosis to endure, but ultimately exploit and benefit from host cell apoptosis-like damage to spread. The association between host cell apoptosis and Leishmania spreading for other parasite species and cell lineages requires further investigation (Getti et al., 2008). The importance of host cell apoptosis for amastigote cell-to- cell spreading was strongly suggested by micro-irradiation of infected macrophages, which selectively induce apoptosis-like cell death of predefined macrophages, allowing and triggering non-irradiated vicinal macrophages to rescue parasites from dying, irradiated cells.

Our results discard a parasite multiplication burst of host cells. We hypothesize that amastigote populations hosted by macrophages for long periods of time would stabilize due to nutrient restriction (after parasite population increase) or by expression of a number of host cell factors. In this line, in experimental visceral leishmaniasis in hamsters, Leishmania chagasi amastigote apoptosis was observed in spleen and liver in the late period of infection, suggesting a parasite population control (Lindoso et al., 2004). Further, parasites may present different multiplication rates and spread more or less efficiently among cells in vitro depending on the host cell lineage (Hsiao et al., 2011).

Pores on macrophage membrane could be formed during this stabilization of amastigote population, without necessarily trigger host cell death. Although Leishmania parasites are able to form pores on biological membranes (Noronha et al., 2000), and pore formation of macrophages takes place in amastigote transfers (inferred by GFP leakage during transfers and increase in transfer occurrence after SLO treatment of host cells) we did not find evidence of host membrane pores localized on parasite-containing extrusions. Macrophage pores on infected cells possibly participate in parasite cell-to-cell transmission by releasing chemotactic signals to other viable macrophages, which rescue extruded amastigotes. Considering that (i) long-term infected macrophages express increased amounts of IGF-I mRNA, and (ii) IGF-I is implicated in Leishmania infections (Reis et al., 2013) and in the increased mobilization of inflammatory mononuclear cells to infected sites in footpad lesions (Gomes et al., 2000), it is reasonable to assume a participation of IGF-I not only as an important anti-apoptotic factor (Kooijman et al., 2002), but also as a chemotactic factor promoting the migration of viable cells towards infected, damaged macrophages. Further studies are needed to explore this specific field that may involve a number of host cell factors influencing migration of cells and phagocytic process.

The parasitophorous extrusions conserve PV components such as LAMPs and Rab7, which are internalized by recipient macrophages together with the rescued amastigotes. The presence of LAMP and possibly other PV components on amastigotes have increased the secretion of IL-10 by recipient macrophages. IL-10 cytokine is known to be secreted by monocytes in response to apoptotic cells (Voll et al., 1997), and by neutrophils in response to L. amazonensis amastigotes (Carlsen et al., 2013). The abnormal presence and/or topology of LAMP in the content of phagolysosomes (as observed during L. amazonensis transfer events) could act as a danger associated molecular pattern (DAMPs), triggering IL-10 production; it could also activate signalling pathways related to recognition of apoptotic/necrotic bodies, leading to anti-inflammatory immune responses which involves IL-10 and TGF-β.

IL-10 is implicated in susceptibility and chronicity of Leishmania mexicana in vivo (Buxbaum and Scott, 2005). Wanderley et al. (2006) demonstrated that the induction of IL-10 mRNA expression and increased TGF-β secretion due to L. amazonensis infection was correlated with the increase of phosphatidylserine (PS) exposure on amastigote surface, thus establishing that amastigotes mimic an apoptotic cell to subvert host immune functions and prevail. Mass spectrometry evidenced that Leishmania promastigotes do not constitutively produce PS but its presence is conceded on other growth phases (Weingärtner et al., 2012). However the observation made here that amastigotes remain associated with immunomodulatory host membrane remnants suggests that part of PS displayed by parasites comes from an exogenous origin.

Although we present a solid evidence of an anti-inflammatory stimulus mediated by exposition of LAMP on amastigotes, distinct immunomodulatory features of amastigotes extracted from different host cells must not be discarded.

The parasitophorous extrusions displayed by L. amazonensis were similar to other pathogen host cell egress mechanisms. These mechanisms, involving or not involving host cell death, include pathogen exocytosis, extrusion or expulsion from the host. The bacteria Chlamydia exit host cells after host cell lysis mediated by proteases or after a non-lytic, actin-dependent extrusion from the host cell. In this process, the bacteria are extruded in a double-membrane vesicle generated from the bacterial intracellular vacuole and cytoplasmic membranes (Hybiske and Stephens, 2007). Legionella pneumophila express two proteins (LepA and LepB) homologous to host SNARE proteins. This homology allows the subversion of native host cell intracellular trafficking and vesicle fusion by these bacterial products, favouring the non-lytic exocytosis of L. pneumophila from host cells (Chen et al., 2004; 2007,). During in vitro cell-to-cell transmission, Mycobacterium marinum and M. tuberculosis are ejected from host cells through an F-actin-rich structure denominated ejectosome (Hagedorn et al., 2009); the bacteria also remain associated with membrane remnants in the process. Viruses such as human immunodeficiency virus (HIV) have also developed non-lytic strategies to egress from host cells. In macrophages, HIV egress can be mediated by Trojan exosomes, originating in multivesicular bodies (MVB) that retain common endosomal and MVB proteins (Gould et al., 2003; Chertova et al., 2006). The fungus Cryptococcus neoformans escape from host cells by an active, pathogen-driven extrusion/expulsion process, apparently without damaging the host (Alvarez and Casadevall, 2006). The fungus can be transferred from cell to cell without exposure to the extracellular milieu; however, the process is still poorly understood and the underlying mechanisms unknown (Alvarez and Casadevall, 2007).

This experimental work proposes a novel mechanism of host cell exit for Leishmania amastigotes, an issue extensively neglected in the study of these parasites. It is now possible to investigate this dynamic phenomenon due to live imaging advances, some of them employed very advantageously in this study, during in vitro experiments. To extend these observations in ex vivo live imaging of infected tissue slices and intra-vital imaging are future challenges and will certainly reveal additional aspects of amastigote egress and reinfection.

Experimental procedures

Parasites and host cells

Leishmania amastigotes were isolated from BALB/c mice foot-pad lesions after 2 months of amastigote subcutaneous inoculation (Rabinovitch et al., 1986). The parasites employed in this study were wild type (WT) or Discosoma sp. red fluorescent protein-2 (DsRed2)-transfected L. (L.) amazonensis LV79 strain (MPRO/BR/72/M1841) and green fluorescent protein (GFP)-transfected L. (L.) amazonensis M2269 strain (MHOM/BR/1973/M2269).

BALB/c, C57BL/6 and C57BL/6-GFP mice (8 weeks of age) were used as the source of bone marrow macrophage precursors, which differentiate into macrophages on plastic Petri dishes in the presence of RPMI 1640 medium supplemented with 20% fetal bovine serum, 100 U m−1 penicillin, 100 μg ml−1 streptomycin and 30% of L929 cell conditioned medium for 6 days at 37°C, 5% CO2 (De Souza Leao et al., 1995). Differentiated bone marrow-derived macrophages (BMDMØ) were detached from Petri dishes by incubation with 0.01 M phosphate-buffered saline (PBS) pH 7.2 supplemented with 20 mM Hepes and 1% EDTA for 20 min. BMDMØ were collected by flushing dishes with warmed Hanks' Buffered Salt Solution (HBSS, Sigma-Aldrich). Cells were counted and distributed in 24-well plates, Mattek (Mattek Corporation) or ibidi Hi-Q4 (ibidi GmbH) round dishes, both pre-treated for cell culture, where they were cultivated in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 U m−1 penicillin, 100 μg ml−1 streptomycin (complete medium) and 5% of L929 cell conditioned medium.

RAW 264.7 macrophage-like lineage cells (ATCC) were cultivated in complete medium; the liposomal transfection of RAW 264.7 macrophage-like cells was performed using FuGene HD transfection reagent (Roche Applied Science) and the following plasmids: LAMP1-GFP (kindly donated by Dr Norma Andrews, Maryland University, USA), Rab7-GFP and LifeAct-RFP F-actin marker (ibidi GmbH). Macrophages were cultivated on Mattek or ibidi Hi-Q4 round dishes suitable for live cell imaging and placed in incubators fitted on microscope stages.

Immortalized C57BL/6 mouse embryonic fibroblasts (MEF) were kindly donated by Dr Paul Saftig (Biochemisches Institute, Christian-Albrechts-Universität zu Kiel, Germany). In this study we employed the PS1 wild-type fibroblasts (MEF-WT) and the #79 LAMP1/LAMP2 double knockout fibroblasts (MEF-LAMP1/LAMP2 KO), which were immortalized by the same protocol (Eskelinen et al., 2004). These cells were maintained in 75 cm2 culture bottles, pre-treated for cell culture.

Infection of host cells

Leishmania amazonensis amastigotes were added to macrophage cultures in multiplicity of infection (moi) of 5:1 and incubated at 34°C, 5% CO2 in complete medium for 2 h for host cell-parasite interaction. Amastigotes at moi 20:1 were added to RAW 264.7 cultures after 24 h of transfection and incubated at 34°C, 5% CO2 in complete medium for 6 h for interaction. Amastigotes at moi 20:1 were added to MEF cultures in culture flasks and incubated at 34°C, 5% CO2 in complete medium for 24 h for interaction.

After this interaction period, free parasites were removed from cell cultures by washing with HBSS; then cultures were incubated in complete medium at 34°C, 5% CO2 atmosphere for additional 24 h (MEF and RAW 264.7 cultures) or for up to 30 days (BMDMØ cultures).

For long-term experiments (> 4 days), BMDMØ cultures were maintained at 34°C, 5% CO2 and medium was replaced every 4 days and also before each microscopic recording. Viability of L. amazonensis amastigotes hosted by BMDMØ during long-term cultivation was inferred by the maintenance of enlarged parasite-containing PVs and amastigote fluorescence (in the case of parasites expressing fluorescent proteins).

Infection of BMDMØ with amastigotes associated with or devoid of host cell LAMP

MEF cells were used as transient hosts for Leishmania amastigotes for 48 h, cultivated in DMEM medium supplemented with 10% fetal bovine serum, 100 U m−1 penicillin, 100 μg ml−1 streptomycin at 34°C, 5% CO2. After this incubation period, infected MEF cells were scrapped from culture bottles and mechanically disrupted through a 15 ml potter homogenizer. Then the suspension containing disrupted cells was passed 10 times through 30G needles. Amastigotes were collected from suspensions of disrupted MEF-WT or MEF-LAMP1/LAMP2 KO cells by employing the standard amastigote isolation protocol (Rabinovitch et al., 1986), without overnight or stirred incubation to preserve host cell membranes associated with amastigotes (Saraiva et al., 1983). The amastigotes isolated from MEF cells were added to C57BL/6 BMDMØ cultures in 24-well plates at 20:1 multiplicity, which were then incubated for 72 h at 34°C, 5% CO2 in complete medium containing 5% L929 conditioned medium and 100 ng ml−1 of lipopolysaccharide (LPS, Sigma-Aldrich). After this period, BMDMØ supernatants were collected and stored at −80°C until cytokine production evaluation.

Multidimensional live cell imaging

Multidimensional imaging of infected macrophage cultures was performed using a Leica SP5 II Tandem Scanner system (Leica Microsystems). Images were acquired from cells cultivated and infected in Mattek or ibidi dishes, placed in the incubator coupled to the microscope. The incubator maintains temperature and CO2 condition throughout the image acquisition.

Fluorescence and differential interference contrast (DIC) images were acquired at resonant scanning mode, 512 × 512 or 1024 × 1024 resolution, using 63× (HCX PL APO 63×/1.40–0.60 CS λ-BL) or 100× (HCX PL APO 100×/1.44 CORR CS) oil immersion objectives and z-stacks ranging from 0.3 to 0.5 μm. To avoid phototoxicity, lasers were set to no more than 5% of potency; additionally, the duration of z-stack acquisition, for each position and in each time point, was 30 s at maximum (to avoid prolonged exposure).

After acquisition, images were processed by Imaris v.7.4.2 software (Bitplane AG, Andor Technology), which allowed for the construction of multidimensional images integrating spatial (tridimensional) and temporal data (acquisition of the same microscopic field in defined time intervals). Images were visualized applying Blend, maximum intensity projection (MIP) or hue-saturation-value filters.

In some experiments, macrophages were stained with Hoechst 3 3342 (Invitrogen, Life Technologies) in order to identify host cell nuclei. Macrophage nuclei fluorescence signals allowed for automatic quantification of host cells in multidimensional images by the Imaris program. Based on Hoechst staining, the software creates isosurfaces representative of macrophage nuclei (Real and Mortara, 2012) and automatically counts these isosurfaces per time point in different microscopic fields. Software attributes a yellow square to each counted nuclei. Parameters for Hoechst-based isosurfaces construction include background subtraction, track surfaces and no split objects. Transfer of amastigotes between macrophages was counted by direct observation of multidimensional images. A transfer event is counted when one macrophage transfers amastigotes to another one, independent of how many amastigotes were transferred. Tracking of transferred amastigotes were performed in some cases, to improve the visualization of the events: in this approach, the DsRed signal of L. amazonensis amastigotes hosted by donor macrophages was tracked using Imaris isosurface tracking tools. The software attribute isospots for each amastigotes and a line tracing of isospots displacement in multidimensional images.

Multichamber dishes (Hi-Q4 dishes, ibidi GmbH, pre-treated for cell culture) were used for experiments in which more than one experimental condition was observed in the same microscope acquisition session. Multichamber dishes were used in experiments with infected and non-infected macrophages and in experiments with infected macrophage cultures (20 days of infection) treated or not treated with 50 ng ml−1 Streptolysin O (SLO), which was washed from cultures after a pulse of 20 min in calcium-free PBS. Production and purification of SLO were performed as established, using plasmids kindly donated by Dr Rodney Tweten (University of Oklahoma, USA) and kindly made available to us by Dr Norma W. Andrews (University of Maryland, USA) (Idone et al., 2008).

Immunostaining and ELISA assays

Previous to cell cultivation, glass coverslips were washed with neutral detergent, water and ethanol, and autoclave sterilized. Coverslips containing infected macrophages or amastigote suspensions in 1.5 ml microcentrifuge tubes (derived from MEF host cells or mouse footpad lesions) were washed and fixed for 1 h with 3.5% paraformaldehyde in PBS at room temperature. LAMP1 and LAMP2 proteins were immunostained using monoclonal antibodies obtained from DSHB (1D4B and ABL-93 hybridomas, Iowa University, USA). LIMP1 immunostaining of these structures was performed using anti-CD63 (LIMP1) monoclonal antibodies (US Biological, Massachusetts, USA). LAMP and LIMP antibody specificities were confirmed by employing isotype control antibodies (rat IgG1 and IgG2A primary antibodies) or only secondary antibodies. Amastigotes were immunostained with 2A3-26 antibody conjugated to FITC (kindly donated by Dr Eric Prina, Institut Pasteur, France). Primary and secondary antibodies (Alexa Fluor™) were incubated for 1 h before an additional stain with 10 μM 4′,6-diamidino-2-phenylindole (DAPI, Invitrogen, Life Technologies) for 15 min. Confocal images of the samples were acquired with a 63× or 100× oil immersion objective in the Leica SP5 II TS system and processed by the Imaris program using Blend filters. Flow cytometry analysis of immunostained amastigotes were performed on BD FACSCanto or BD Accuri C6 instruments (BD Biosciences).

Mouse interleukin 10 Enzyme-Linked Immunosorbent Assay kit (ELISA Ready-SET-Go!, eBioscience) was used for the measurement of cytokine production of BMDMØ culture supernatants, according to manufacturer's instructions.

Micro-irradiation of infected cells

Macrophages obtained from C57BL/6-GFP mice or transfected RAW 264.7 macrophages were infected with L. amazonensis-DsRed2 amastigotes for 24 to 96 h and cultivated in round dishes, suitable for the incubator coupled to the Leica SP5 II TS system. Using the Leica FRAP Wizard tool provided by the acquisition software, we delimited an area or a region of interest (ROI), in which laser pulses were applied. These ROI corresponded to Hoechst-stained macrophage nuclear regions, adjusting the focal plane (z axis) to the centre of the nuclei; a concentrated near-ultraviolet (UV) laser [405 nm, at 100% power (50 mW) and 400 Hz frequency] was applied to the ROI for 300–500 s before multidimensional live recordings (Soustelle et al., 2008). The technique was also employed to hasten the extrusion of amastigotes from RAW transfected cells, which are not suitable for long-term in vitro infection (i.e. for more than 5 days).

To check whether micro-irradiation is inducing apoptosis, micro-irradiation of 48 h infected BMDMØ cultures was performed in complete medium supplemented with 5% of L929 cell conditioned medium, 2.5 mM CaCl2, 500 nM of nuclear probe YO-PRO-1 Iodide (Molecular Probes) and 2.5 μg ml−1 of Annexin V-CF633 (Biotium).

Electron microscopy and correlative imaging

Suspensions of macrophages infected for 20 days with L. amazonensis were fixed in a solution of 2.5% glutaraldehyde and 2% formaldehyde in 0.1 M sodium cacodylate buffer at room temperature for 1 h, then stored at 4°C before conventional transmission electron microscopy (TEM) procedures (Arruda et al., 2012). Samples were observed in a Jeol 100 CX II electron microscope. For SEM, infected macrophages or isolated amastigotes were incubated in Mattek dishes and fixed with 3.5% paraformaldehyde according to established preparation protocols. Briefly, fixed samples were washed in 0.1 M cacodylate solution, post-fixed with osmium tetroxide, treated with tannic acid, dehydrated with ethanol and then dried in a CPD 030 critical point dryer. The dishes were coated with a gold layer using a sputtering method. Samples were observed in a Field Emission FEI Quanta FEG 250 scanning electron microscope.

In order to observe and acquire images from the same infected macrophage in confocal and SEM instruments, cells were cultivated on gridded dishes (ibidi u-Dish 35 mm Grid-500 ibiTreat; ibidi GmbH) suitable for observation in the Leica SP5 II TS confocal as well as the Field Emission SEM. The grid presents alpha-numerical co-ordinates and gridded markings spaced by 500 μm. Samples were immunostained and observed by confocal microscopy prior to preparation for observation by electron microscopy.

Quantitative PCR for the expression of pro- and anti-apoptotic macrophage genes

To evaluate the expressions of pro- and anti-apoptotic macrophage genes, total RNA was extracted from 2 × 105 cells ml−1 using TRIzol (Invitrogen, USA), following the manufacturer's protocol (RNA integrity was determined as an OD260/280 absorption ratio between 1.8 and 2.1). Then 1 μg of purified RNA was mixed with 10 μl of a solution consisting of a basic buffer (100 mM Tris-HCl, pH 8.3, containing 500 mM KCl and 15 mM MgCl2 Invitrogen, USA), dNTP (10mM; Fermentas, USA), random primers (Invitrogen, USA), OligoDT primers (Invitrogen, USA), RNaseOUT recombinant ribonuclease inhibitor (40 U μl−1; Invitrogen, USA), M-MLV reverse transcriptase (100 U μl−1). The reactions were incubated at 37°C for 50 min and were denatured at 70°C for 15 min.

For real-time quantitative RT-PCR, the following primers set were designed: murine Bax gene (GenBank Accession No. NM007527), forward: 5′-GGC CTT TTT GCT ACA GGG TTT CAT-3′ and reward: 5′-TGC TGT CCA GTT CAT CTC CAA TTC-3′; for murine Bcl-2 gene, forward: 5′-GAC TGA GTA CCT GAA CCG GCA TCT-3′ and reward: 3′-AAG CCC AGA CTC ATT CAA CCA GAC-3′ (GenBank Accession No. NM009741), and for murine IGF-I, forward, 5′-TAC TTC AAC AAG CCC ACA GG-3′ and reward, 5′-AGT CTT GGG CAT GTC AGT GT-3′ (GenBank Accession No. NM010512). β-Actin (GenBank Accession No. NM00739) was used as a constitutively expressed control gene for normalization (primers: forward, 5′-GCC TTC CTT CTT GGG TAT GGA ATC-3′ and reward, 5′-ACG GAT GTC AAC GTC ACA CTT CAT-3′). The reactions included master mix Syber Green (2×) (Applied Biosystems, USA) and 1 μl cDNA (1 μg) template and were run in triplicate on a real-time PCR system (StepOne; Applied Biosystems, USA). The PCR conditions were the same for all primer combinations: 95°C for 10 min, 40 cycles of 92°C for 2 min, 57.5°C for 30 s and 70°C for 30 s. After PCR amplification, a melting curve was generated to confirm the specificity of the products. The data were presented as a relative quantification and were calculated using 2−ΔΔCt (Pfaffl, 2001).

Statistics

All statistical tests employed in this study were performed by SPSS software (SPSS). These tests included the Student's t-test and analysis of variance (anova) with Bonferroni post-hoc tests. Results represent data obtained in at least two experiments and at least in duplicate.