In 2D culture, human neuroblastoma SH-SY5Y cells exhibit a neuroblast-like morphology with a few short neurite-like projections (arrowheads in the right panel of Fig. 1A). At 70–80% confluence, the trophozoites of B. mandrillaris were added to the culture medium of human neuroblastoma cells and plated on a monolayer of human neuroblastoma SH-SY5Y cells. Within 30 min, nearly all of the trophozoites became attached to the human cells. At 24 hours post coculture, a human cell-free area appeared and increased in size in a time-dependent manner (the dotted line in Fig. 1B). At higher magnification, there were round cells at the edge of the human cell-free zone (arrowheads in inset #1 in the right panel of Fig. 1B and the left panel of Fig. 1C), while the irregularly shaped, cytoplasm-protruding trophozoites were located at the center of the human cell-free area (arrow in inset #2 in the right panel of Fig. 1B and the right panel of Fig. 1C). Live imaging of the inverted phase contrast microscope shows a trophozoite attached to human cells and protruding pseudopods. The human cell-adhering trophozoite also rotated, while its pseudopods occasionally elongated and shortened (Additional file 1).
To examine the cytotoxicity of the B. mandrillaris trophozoites, live cell-impermeable trypan blue dye was added at 24 hours post coculture. The trypan blue-positive human neuroblastoma cells were proximal to the round trophozoites (inset #1–3 in Fig. 1D), indicating cell death due to loss of cell membrane integrity. In contrast, the trophozoite-free human cells were viable (inset #4 in the right panel of Fig. 1D). Downregulation of anti-apoptotic BCL2 transcripts were observed at one hour post coculture, while levels of apoptotic BAX transcripts tended to be upregulated but not significant difference (Fig. 1E). After 2 days, the monolayer of human neuroblastoma cells was entirely destroyed (upper panel of Fig. 1F). Cell debris, pseudopod-protruding trophozoites and floating round shaped cells were scattered throughout the culture well (insets#1–4, Fig. 1F). Together, the findings showed that the clinically isolated B. mandrillaris trophozoites were cytotoxic to human neuroblastoma SH-SY5Y cells.
To observe the mechanism by which the trophozoites ingest human cellular components in real time, a holotomographic microscope was used to capture 360-degree views of the 2D-cultured cells (Fig. 2A). The 2D view of cells in the X-Y plane showed a difference in intensity from high to low (white to black, Fig. S1). The dense white-colored circles observed in the 2D image of human cells were lipid droplet-like structures (arrowhead in Fig. S1). Based on morphology, the white oval shape was distinct from the elongated shape with protruding pseudopods. However, without the cell morphology, it was difficult to distinguish between the human cells and trophozoites in the 2D image (Fig. S1A).
Given that different types of cell components exhibit distinct RIs, the holotomographic microscope is capable of visualizing cell components in a 3D manner. Based on the RIs of the cell components, pseudocolors were assigned for each cell component. As shown in the X-Y plane (left panel in Fig. 2B), the cytoplasm of human neuronal cells exhibited a cyan color, while that of trophozoites displayed a blue gray color. The red dots that had an RI similar to that of the lipid droplets were observed along the X-Y, X-Z and Y-Z planes (arrowheads in all panels of Fig. 2B and Additional file 2). For the second representative cell image, the X-Y plane showed oval-shaped cells with a protruding pseudopod-like structure. When observing the lateral view along the Y-Z planes, there was a bridge between the human neuroblastoma cells and B. mandrillaris trophozoites (arrows in the middle and left panels of Fig. 2C and Additional file 3). Thus, the difference in the RI between human neuroblastoma cells and B. mandrillaris trophozoites allows the identification of the cell type and connecting point.
To observe the movement of the trophozoites, live imaging was performed using holotomography to observe cell-to-cell interactions. The 3D live imaging revealed that the B. mandrillaris trophozoite twisted its extending pseudopod (Additional file 4). When trophozoite rotation was imaged for 6 seconds, the snapshot image shows anchoring sites (arrows in the left, middle and right panels of Fig. 2D). Notably, the number of cyan-colored components increased in the pseudopod-protruding trophozoite when observed for 6 seconds (arrowhead in the middle and right panels of Fig. 2D). Taken together, the results showed that the B. mandrillaris trophozoite anchors to the human cell, followed by twisting its extending cytoplasm around the cytoplasmic bridge.
Given the similarity between the RIs of the human cells and the parasite, it was difficult to observe the invasion of the B. mandrillaris trophozoites. Thus, differential cell tracking was performed to observe a cell-to-cell interaction. Two fluorescence probes, CMFDA and DiD, were used to label amines and lipids, respectively (Fig. 3A). For human neuroblastoma SH-SY5Y cells, CMFDA was observed as a smear pattern dispersed throughout the cytoplasm, while DiD was observed as a granule-like pattern in the cytoplasm and on the cell membrane (Fig. S2A). In contrast, the cytoplasm of B. mandrillaris trophozoites exhibited a dispersed pattern of DiD-labeled lipids. There was no cellular component labeled with CMFDA (Fig. S2B). Hence, the CMFDA and DiD fluorescent probes were used for tracking human neuroblastoma cells and B. mandrillaris trophozoites, respectively (Fig. S2C).
Based on light microscopic observation, the whole trophozoites reportedly entered African green monkey kidney cells . To confirm the cell-in-cell formation, CMFDA and DiD were used to label the respective human cells and the B. mandrillaris trophozoites to detect human cell-invading B. mandrillaris trophozoites (Fig. 3A). After 40 min of coculture on a glass slide, the X-Y plane of the 2D-captured image showed DiD-labeled B. mandrillaris trophozoites (magenta, Fig. 3B) located proximal to the CMFDA-labeled human neuronal cells (green, in Fig. 3B). The trophozoites were positioned at the top of human neuroblastoma cells. Some trophozoites were partially adhered to the human cells (insets 1 and 3 of Fig. 3B), while some spread their cytoplasm across a few human cells (inlet 2 of Fig. 3B). There were no yellow-colored merged areas, suggesting that none of the trophozoites entered the human cells.
To observe the invasion by B. mandrillaris trophozoites into human neuroblastoma cells, differential cell tracking was performed as mentioned above (Fig. 3A). At 40 min post coculture, magenta-colored holes were observed in the cytoplasm of human cells (arrows in all panels of Fig. 3C), indicating the presence of DiD-labeled lipids in B. mandrillaris trophozoites. By using the Z-stack, the X-Z and Y-Z planes allow a lateral view of the imaged cells (left panel of Fig. 1C). Thus, both the X-Z and Y-Z planes revealed that the DiD-labeled compartment was surrounded by the human cytoplasm (green) and was proximal to the human nucleus (left panel of Fig. 3C). Around the cytoplasmic hole, the intensity of CMFDA fluorescence was higher than that in other parts of the human cytoplasm (arrowheads in the middle panel of Fig. 3D). The lipid-containing cytoplasm of the B. mandrillaris trophozoite protruded into human cells, a structure mimicking invadopodia (yellow arrows in the right panel of Fig. 3C). A 3-D view of the X-Y-Z plane shows the invadopodia-like structure of B. mandrillaris trophozoites protruding into human neuroblastoma cells (yellow arrow in Fig. 3D and Additional file 5).
Following invasion, the cellular uptake of the B. mandrillaris trophozoites was examined. To observe components of human cells inside the trophozoites, both CMFDA and DiD fluorescence probes were preincubated with human neuroblastoma cells, while the B. mandrillaris trophozoites remained unstained (Fig. 3E). In the cytoplasm of the B. mandrillaris trophozoites (dotted line in Fig. 3F), the amine-containing cell components appeared as well-defined granules with irregular sizes. In contrast, the lipid components were nonuniformly dispersed throughout the cytoplasm of B. mandrillaris trophozoites (Fig. 3F). The pattern of human-derived lipids in the cytoplasm of the trophozoites was similar to that of B. mandrillaris-derived lipids (Fig. S2B). The majority of amines and lipids showed no overlap, indicating the existence of a distinct cellular compartment (merge panel of Fig. 3F). Moreover, the trophozoite contained sparse granular nuclei with smaller sizes than human nuclei (arrowheads in Fig. 3F). When the cellular uptake was examined, the percentages of CMFDA- and DiD-positive trophozoites adhering to human cells increased to 96% at 40 min post coculture (Fig. 3G). However, there were no changes in the proportion of human cell-ingesting floating trophozoites from 10–40 min. However, after 6 hours of coculture, the B. mandrillaris trophozoites were mostly detached and floated in the culture medium. The floating trophozoites had ingested the human cell components, which appeared as amine-containing, irregularly shaped granules and lipid-diffused areas in the trophozoite cytoplasm (Fig. S3). Altogether, these results suggest that the B. mandrillaris trophozoite protrudes its cytoplasm into human cells, forming a structure mimicking invadopodia, prior to ingesting human cell components.
To visualize ingestion of the human cell cytoplasm in real time, the human cells were cultured as a 2D monolayer and incubated with the amine-binding fluorescent dye CMFDA (Fig. 4A). The nonlabeled trophozoites were cocultured with CMFDA-labeled human cells. At 6 hours post coculture, sparse green circular shapes were observed in the cytoplasm of the trophozoites. Live imaging showed that green vesicles appeared in the area proximal to the human cells (Additional file 6). Within 13 min, the colored area in the human cytoplasm increased, suggesting loss of the cytoplasmic compartment (inlets in Fig. 4B). Subsequently, the human cells lost cell membrane integrity, likely due to apoptosis (Fig. 4C). Thus, the B. mandrillaris trophozoites ingested the human protein constituent in the form of spherical granules, a finding similar to the 2D snapshot confocal images in Fig. 3F.
The phosphatidylinositol-3 kinase (PI3K) pathway plays an important role in trogocytosis, a process in which parasitic amoebae ingest portions of human cells . To investigate whether human cell ingestion by B. mandrillaris relies on the PI3K signal, human trophozoites were preincubated with wortmannin, a PI3K inhibitor, prior to 40-min coculture with CMFDA-labeled human cells (Fig. 4C). Number of the CMFDA- and DiD-positive trophozoites indicated that the B. mandrillaris trophozoites incubated with wortmannin retained the capability of ingesting human amine-containing components, indicating that ingestion by B. mandrillaris trophozoites was trogocytosis independent (Fig. 4D).