H. sapiens (TGS101) and S. cerevisiae Vps23 proteins sequences (access numbers: Q99816 and P25604, respectively), retrieved from the Uniprot database (https://www.uniprot.org) were used as query to search an E. histolytica Vps23 protein (EhVps23) in the AmoebaDB database (https://amoebadb.org/amoeba/app). Structural domains of the candidates were identified using the SMART genomics server (http://smart.embl-heidelberg.de/) and the motif tool of KEGG (https://www.genome.jp/kegg/).The predicted amino acid sequences of the putative EhVps23 (C4M9S4) were aligned with orthologous sequences. Data were submitted to phylogenetic analysis by UPGMA using the MEGA 5.05 software (Tamura et al., 2011). Bootstrapping was performed for 1000 replicates.

To obtain the EhVps23 3D model, we used the crystal structure of the Vps23 protein from S. cerevisiae (3R3Q:A) as a template on the I-TASSER server (https://zhanglab.dcmb.med.umich.edu/I-TASSER/) (Roy et al., 2010). After selecting the most energetically stable model, we evaluated their quality by the RAMPAGE server (http://mordred.bioc.cam.ac.uk/~rapper/rampage.php). The amino acid sequences of HsTSG101 were analyzed using the RaptorX server (http://raptorx.uchicago.edu/) to obtain the HsTSG101 3D structures (Källberg et al., 2012). The structures were visualized using the UCSF Chimera software. Similitudes beteween the UEV and Vps23 core domains of both proteins was analyzed using the RaptorX Structure Alignment tool.

E. histolytica trophozoites, strain HM1:IMSS, were axenically grown at 37°C in TYI‐S‐33 medium (Diamond et al., 1978) and harvested at logarithmic growth phase. To perform the experiments, the culture flasks were chilled at 4°C and trophozoites were collected by centrifugation. All experiments were performed at least three times by duplicate.

As primary antibodies we used: mouse monoclonal α-histidine (α-His; Roche), mouse monoclonal α-BtUb (α-Ub; SantaCruz), rabbit α-EhADH (α-EhADH18) (Díaz-Hernández et al., 2021), mouse α-EhVps32 (α-EhVps32) (Avalos-Padilla et al., 2015), mouse α-LBPA (α-LBPA; Echelon Bioscience) and mouse monoclonal α-human actin (α-actin) (kindly given by Dr. Manuel Hernández, Cell Biology Department, CINVESTAV IPN). Secondary antibodies were: HRP-labelled α-rabbit, mouse and rat IgGs (Zymed) for western blot. FITC-labelled α-rabbit and α-mouse IgGs and TRITC-labelled α-rat IgGs (Life Technologies) for immunofluorescence. For immunoelectron microscopy experiments, we used α-rat IgG, conjugated with 10 nm gold particles and α-mouse IgG, conjugated with 20 nm gold particles (TED Pella Inc).

Total RNA was isolated from trophozoites using TRIzol reagent (Invitrogen), according to manufacturer’s recommendations. Complementary DNA (cDNA) was synthesized using oligo dT primers and the Superscript II reverse transcriptase (Invitrogen). PCR amplifications were carried out using Q5 ^® High-Fidelity DNA Polymerase (Biolabs) with 20 ng of cDNA as templates, and specific primers. Products were separated by electrophoresis in 1% agarose gels, and stained with ethidium bromide. As controls for PCR amplifications, instead of DNA, nuclease-free water was used.

The Ehvps23_{1123-1485 bp} or Ehvps23_{1-341 pb} sequences were PCR-amplified using cDNA as template and specific primers: for Ehvps23_{1123-1485 bp} : sense 5′-GGGGTACCATGGAAGAGTCTGAAGAAATACTTCATG-3′, antisense 5′-CCGGATCCTTATTCAGTTATGCAATACTTTGCATGAA-3′; for Ehvps23_{1-341 bp} 5´-GGGAGCTCATGCAACCTATAAACAATGAAAAGAATATTAAC-3´ and the antisense 5′- CCGGTACCAACTTTCCTAATAACACCATTTTCATCTAC-3. Underlined sequences correspond to the enzyme restriction sites added to the primers. Fragments were cloned in pColdI for producing a recombinant protein tagged with 6X His, and in pL4440 for silencing experiments, generating the pColdIEhvps23_{1123-1485 pb} and pL4440Ehvps23_{1-341 pb} , plasmids, respectively. The quality of the constructs were verified by restriction enzyme analyses and automatic DNA sequencing.

E. coli BL21 (DE3) bacteria were transformed with the pColdIEhvps23_{1123-1485 bp} plasmid to produce the His-tagged EhVps23_{375-494 aa} recombinant protein (rEhVps23). Protein expression was induced by 0.4 mM isopropyl beta-D-thiogalacto pyranoside (IPTG) in the LB medium for 16 h at 16°C. After lysis, the rEhVps23 protein was recovered from the inclusion bodies, using the solubilization buffer containing 20 mM Tris-HCL pH 7.5; 8 M Urea, 0.5 M NaCl, 1 mM β-mercaptoehtanol and 5 mM imidazole. rEhVps23 protein was purified through chromatography in Ni-Sepharose 6 columns (GE Healthcare in the NGC Q Chromatographic System, Bio-Rad) (de la Cruz et al., 2019). Identity and integrity of the purified rEhVps32 protein was verified by 10% SDS-PAGE and western blot assays.

150 μg of purified rEhVps23 protein were emulsified in Titer-Max Gold adjuvant (1:1) (Sigma) and subcutaneously and intramuscularly inoculated in Wistar rats (Avalos-Padilla et al., 2015). One more dose without Titer-Max was injected after 4 weeks. Pre-immune sera were obtained before immunizations.

Total extracts from trophozoites (prepared in the presence of 100 mM PHMB, 2.7 mM E64 and inhibitors cocktail) and bacterial lysates (prepared in the presence of 100 mM PMSF), were separated by SDS-PAGE gels, transferred onto PVDF or nitrocellulose membranes, and probed with rat α-EhVps23 (1:500), mouse α-His (1:500), mouse α-Ub (1:100), rabbit α-EhADH18, (1:500), mouse α-EhVps32 (1:500) or mouse or α-human actin (1:3,500) antibodies. Membranes were incubated with the species-specific horseradish peroxidase (HRP)-labelled secondary antibodies (1:10,000), and developed with the ECL Prime detection reagent (GE-Healthcare). Pre-immune serum was used as a control and competition experiments were performed using the recombinant protein to assure the proper identification of EhVps32 in trophozoites lysates.

For phagocytosis assays the trophozoites were incubated for 0, 2, 5, 15, and 30 min with RBCs (1:25) at 37°C. At different times, trophozoites were prepared for immunofluorescence (García-Rivera et al., 1982) and observed through the laser confocal microscope. Other preparations were stained by Novikoff technique (Novikoff et al., 1972) and ingested erythrocytes were counted in 100 trophozoites through the light microscope (Axiolab, Zeiss). For pulse-chase experiments, incubation with RBCs was carried out for 2 min at 37°C. Then, preparations were incubated with TYI-water (2:1) for 5 min at 37°C to remove the adhered and non-ingested erythrocytes, cell mixtures were again incubated at 37°C for different times and samples were treated for immunofluorescence and observed through the Carl Zeiss LMS 700 laser confocal microscope.

Trophozoites (grown on coverslips) were fixed with 4% paraformaldehyde at 37°C for 1 h, permeabilized with 0.2% Triton X‐100 and blocked with 10% fetal bovine serum in PBS. Preparations were incubated at 4°C overnight (ON) with α-EhADH18 (1:50), or α-EhVps32 (1:50), α-LBPA (1:30), or α-EhVps23 (1:50) or α-Ub (1:15) antibodies, followed by incubation for 30 min at 37°C with TRITC-labelled α-rat IgG for α-EhVps23, FITC-labelled α-rabbit IgG for α-EhADH18 and α-EhVps32, or FITC-labeled α-mouse IgG for α-LBPA and α-Ub (1:100). Preparations were preserved using Vectashield antifade reagent (Vector) and then, 0.5 µm laser sections were examined through the confocal microscope and processed with ZEN 2009 Light Edition Software (Zeiss). To evaluate the co-localization between molecules, Pearson coefficients were obtained from at least 25 confocal images using the ImageJ 1.45v software and the JACoP plugin.

Samples were prepared for TEM as described (Bolaños et al., 2016). Briefly, trophozoites were fixed with 4% paraformaldehyde and 0.5% glutaraldehyde in PBS for 1 h at room temperature. Samples were embedded in LR White resin (London Resin Co) and polymerized under UV at 4°C for 48 h to obtain thin sections (60 nm) that were mounted on Formvar‐covered nickel grids followed by ON incubation with α‐EhVps23 (1:30) and α-LBPA (1:10). The thin sections were incubated ON with secondary antibodies (1:50) conjugated to 10‐nm (for α‐EhVps23) or 20-nm (for α-LBPA) gold particles, contrasted with uranyl acetate and lead citrate and observed through a Joel JEM‐1011 transmission electron microscope.

The 3D predicted and refined structures of EhADH and EhVps32 proteins were used for docking experiments (Montaño et al., 2017) whereas the 3D structure of EhVps23 protein was generated in this work. The snapshots were obtained using the clustering analysis in the last 50 ns of the MDS with the Carma software (Koukos and Glykos, 2013). The protein-protein docking analysis were done employing different conformers with the Cluspro server (Comeau et al., 2004a; Comeau et al., 2004b; Kozakov et al., 2013). The conformers with the highest cluster members and the lowest energy, calculated in FireDock (Mashiach et al., 2008), were taken for analysis on the PDBSum server (Laskowski et al., 1997). Visualization of 3D structures was performed by VMD (Humphrey et al., 1996). To obtain the binding site between EhVps23 and EhUbiquitin, the crystal structure of EhUbiquitin (PDB:4GSW: B) and Vps23 3D structure from yeast were structurally aligned to get the coordinates of the binding site between the EhVps23 with the EhUbiquitin.

The LBPA structure was obtained and optimized as reported (Castellanos-Castro et al., 2016b). Molecular docking was performed using Autodock4 (Morris et al., 2009), Autodock Vina (Trott and Olson, 2009) and AutoDock Tools 1.5.7, and performed using a grid box of 80 Å and a grid space of 0.375 Å. Then, polar hydrogen atoms and Kollman charges (Singh and Kollman, 1984) were localized in EhVps23. For scoring sampling, the Lamarckian Genetic Algorithm using an initial randomized population of 100 individuals and a maximum number of energy evaluations of 1 x 10⁷ runs were performed. The predicted lowest energy binding position was considered for the analysis.

Trophozoites were lysed in the presence of 10 mM Tris‐HCl, 50 mM NaCl, and proteases inhibitors, by cycles of freeze‐thawing in liquid nitrogen and vortexing. Immunoprecipitation assays were performed using 200 μl of protein G‐agarose (Invitrogen) and α-EhVps23 antibody. Immunoprecipitated proteins were analyzed by western blot assays using α-EhVps23, α-EhADH18 α-EhVps32, and α-Ub, as described above.

To knock-down the Ehvps23 gene we used bacterial double-stranded RNA (dsRNA) and parasite soaking experiments (Solis et al., 2009). Briefly, the competent RNase III-deficient E. coli strain HT115 (rnc14:DTn10) was transformed with the pL4440Ehvps23_{1-341 pb} . Bacteria were grown at 37°C in LB or 2YT broth for dsRNA expression, in the presence of ampicillin (100 mg/ml) and tetracycline (10 mg/ml), using 2 mM (IPTG) for induction, ON at 37°C. Then, the bacterial pellet was mixed with 1 M ammonium acetate and 10 mM EDTA, incubated with phenol:chloroform:isoamyl alcohol (25: 24:1) and centrifuged. The supernatant was mixed with isopropanol, centrifuged, and the nucleic acid pellet was washed with 70% ethanol. DNase I (Invitrogen) and RNase A (Ambion) were added to eliminate ssRNA and dsDNA molecules; Ehvps23-dsRNA was washed again with isopropanol and ethanol, analyzed by agarose gel electrophoresis and concentration was determined by spectrophotometry. Purified Ehvps23-dsRNA (5 μg/ml) were added to the trophozoites (3.0 × 10⁴) growing in TYI-S-33 complete medium tand incubated at 37°C for 72 hr. Cells growing in standard conditions (without dsRNA, or with an unrelated dsRNA) were used as controls.

Values for all experiments were expressed as the mean and standard error of at least three independent assays, carried out by duplicate. Statistical analyses were performed using the GraphPad Prism v5.01 software by a paired Student’s t test. *p< 0.05; **p< 0.01, and ***p< 0.001.

CINVESTAV fulfills the standard of the Mexican Official Norm (NOM-062-ZOO-1999) “Technical Specifications for the Care and Use of Laboratory Animals”, based on the Guide for the Care and Use of Laboratory Animals (“The Guide,” 2011, NRC, USA with the Federal Register Number BOO.02.03.02.01.908), awarded by the National Service for Agrifood Health, Safety and Quality (SENASICA). This organization verifies the state of compliance of such NOM in Mexico and belongs to the Ministry of Agriculture and Rural Development. The Institutional Committee for Animal Care and Use (IACUC/Ethics committee) from CINVESTAV, the regulatory office for research protocols approval involving the use of laboratory animals, reviewed, and approved all animal experiments (Protocol Number 0505-12, CICUAL 001).

In humans and yeast, ESCRT-I is an elongated heterotetrameric complex (1:1:1:1) that interacts by one side with the HRS/STAM (ESCRT-0), and by the other with the Vps36 protein (ESCRT-II). It is formed by TSG101 (named Vps23 in yeast), Vps28, Vps37, and UBAP1 (MVB12 in yeast) proteins ( Figure 1A ). Vps23 from Saccharomyces cerevisiae and Homo sapiens orthologues have been already crystallized and their role in vital cellular functions has been studied (Kostelansky et al., 2007; Im et al., 2010; Ren and Hurley, 2011; Flower et al., 2020). In E. histolytica, the ESCRT-III proteins have been studied, but little is known on the other complexes of the ESCRT machinery. Using as templates the Vps23 core and the UEV (referred to the ubiquitin E2 variant domain found in TSG101 protein and in many other ubiquitin-binding proteins) domains of the S. cerevisiae Vps23 protein, we uncovered two contigs in the E. histolytica genome with intronless open reading frames, predicting 494 and 283 amino acid sequences ( Figure 1B ), but the smaller one (Leung et al., 2008), lacks the Vps23 core domain, that is the Vps23 proteins signature, thus, it was not considered for this work. Additionally, we did not detect consensus sequences with domains described for Vps28, Vps37, and MVB12 (Kostelansky et al., 2006; Kostelansky et al., 2007). Lopez-Reyes (Lopez-Reyes et al., 2010) found a Vps37D protein, but its tertiary structure lacks the alpha helixes necessary for the EhVps23 functions (Kostelansky et al., 2007). However, given the variation of the ESCRT proteins of E. histolytica, compared with their orthologues (Lopez-Reyes et al., 2010; Avalos-Padilla et al., 2018) ( Figures S1, S2 ), the existence of other ESCRT-I members cannot be discarded.

The phylogenetic analysis showed two genes in E. dispar, one in E. invadens and two in E. moshkovskii grouped in a clade close to Trypanosoma cruzi and T. brucei, in which E. histolytica is between E. dispar and E. moshkovskii ( Figure 1C ).

The 3D structure of EhVps23 was obtained from the I-TASSER server. It was selected according to its C-score and the best Ramachandran plot values after MDS during 70 ns in a soluble environment (Figures 2A–D). The Ramachandran plot (https://zlab.umassmed.edu/bu/rama/) (Hooft et al., 1997) showed 76.2% amino acids in the favored regions, 97.3% in the allowed regions, and 2.6% in the outlier ones (see Figure S-3 ), suggesting that the refinement of torsion angles of certain amino acids through MDS, occurred. RMSD calculations indicated that EhVps23 reached the equilibrium after 50 ns ( Figure 2A ), and Rg data revealed that EhVps23 is an extended protein that compacts through the dynamics ( Figure 2B ). The RMSF values showed a peak of fluctuation in a disordered region from Q200 to P334 amino acids, formed by random coils. Another fluctuation peak was located at the carboxy-terminus from N358 to E494 amino acids, where an α-helix is formed ( Figure 2C ). Thus, these analysis showed that EhVps23 has the 3D structure described for its orthologues (Kostelansky et al., 2007; Im et al., 2010; Ren and Hurley, 2011; Flower et al., 2020) with fourteen alpha helices and six beta-sheets ( Figure 2D ).

The UEV ( Figures 2F, G ) and Vps23 core ( Figures 2I, J ) domains of EhVps23 ( Figure 2D ) and human TSG101 ( Figure 2E ) exhibited 82% (RMSD: 2.44) and 90% (RMSD: 0.84) structural identity, respectively ( Figures 2H–K ). Besides, the carboxy terminus (Vps23 core domain) of EhVps23 presents the alpha helixes described as responsible for the ESCRT-I heterotetramer formation, suggesting that these structures could contact amoebic proteins that we did not identify here. The analysis of the 3D structures strongly suggests that EhVps23 is indeed an orthologue of yeast Vps23 and human TSG101 proteins.

To study the location and follow the movement of EhVps23 in the trophozoites, we produced specific antibodies against the carboxy terminus of the protein. A fragment containing 1123 to 1485 bp of the gene, absent in the truncated contig ( Figure 3A ), was used to obtain the recombinant EhVps23_{375-494 aa} polypeptide (rEhVps23) in Escherichia coli, labelled with a histidine tag (His-tag). The polypeptide was injected into rats to generate antibodies. In western blot assays, antibodies recognized a 15 kDa band ( Figure 3B ). The rat α-EhVps23_{375-494 aa} antibodies (α-EhVps23) ( Figure 3B ) reacted with the rEhVps23 polypeptide in bacteria and revealed a 54 kDa band in trophozoites lysates ( Figure 3B ), the molecular weight deduced from the amino acids sequence. Through the confocal microscope, these antibodies detected EhVps23 clustered, probably in vesicles ( Figure 3C ).

To explore the fate of EhVps23 under the stimulus of phagocytosis, the protein was tracked using the α-EhVps23 and TRITC-labelled anti-rat secondary antibodies ( Figure 3D ). Confocal images of trophozoites in contact with red blood cells (RBCs) revealed the displacement of EhVps23 at the site of contact with the target cell. At 2 and 5 min, the label was localized in the adhered and ingested RBCs ( Figure 3D , 2 and 5 min). In other experiments, the cell mixture was incubated at 37°C for 2 min, and adhered and non-ingested RBCs were removed by mild osmotic shock to avoid the noise of new ingested cells, and trophozoites were again incubated at 37°C. In these experiments, at 5 min (2 + 5), the antibodies detected larger vesicles in the cytoplasm and around the ingested erythrocytes. At 2 + 30 and 2 + 60 min, some of the protein appeared in cytoplasm vesicles not in contact with the ingested RBCs, whereas another part of the label remained surrounding the RBCs ( Figure 3D ). Elongated and round vesicles were observed close to or inside of large phagosomes and MVBs ( Figures 3D, E ). Magnification of the images revealed chains of relatively ordered vesicles of different sizes, as if they were forming specific cellular structures. Many of them, seem to be emerging from or fusing to other vesicles, however other type of experiments are needed to prove this ( Figure 3E ). In synthesis, the analysis of many confocal images revealed that, after the RBCs stimulus, EhVps23 moves from the cytoplasm to the adherence site at the plasma membrane, then, after ingestion, it borders the erythrocytes-containing phagosomes and MVBs; later, it appears close to or inside of vesicles, giving a panoramic of the movement of EhVps23 during the active vesicular trafficking and phagocytosis of trophozoites.

To obtain further data on the moving of EhVps23, we performed transmission electron microscopy (TEM) analysis using the α-EhVps23 and gold-labeled secondary antibodies. TEM images revealed the protein on the plasma membrane and in the extracellular space, close to several vesicles of about 0.6-0.9 µm, suggesting that EhVps23 could be secreted ( Figures 4A, B ) and inside MVBs and ILVs ( Figures 4C–E ), like the ones discovered in Dictyostelium discoideum treated with the drug U1866A (cationic sterol) that impairs fusion of endosomes with lysosomes, which causes the accumulation of membranous structures (Marchetti et al., 2004), and in E. histolytica overexpressing EhADH (Bañuelos et al., 2012). These vesicles appeared surrounded by other labelled vesicles ( Figures 4D, E ), like those observed through the confocal microscope. The different numbers and sizes of the vesicles could suggest a distinct maturation state of MVBs. Around some MVBs and vesicles, there were labelled networks ( Figure 4B, E ). Scarce label was detected free in the cytoplasm ( Figures 4A–F ), and it did not appear in controls treated with secondary antibodies ( Figure 4G ).

After 15 min contact with RBCs, TEM images showed convoluted membranous structures penetrating the phagosomes and ingested RBCs, and the protein was detected there ( Figures 4H, J ). Tubular structures appeared very close each other ( Figures 4I, J ). This and the white convoluted arrangements inside the RBCs suggest that some vesicles could be transporting different molecules including lytic enzymes, (necessary for digestion of the cargo) from one vesicle to another and that possibly their content are shed on the RBCs-containing phagosomes. The digested hemoglobin could account for the form and size of the arrangements observed in MVBs or phagosomes with ingested RBCs. Like in basal conditions, the labelled phagosomes were surrounded by other labelled vesicles ( Figure 4J ). By their appearance and their expansion inside ingested RBCs, lytic enzymes and lipids could be transported there.

According to Dupré et al., a yeast protein complex, containing the Vps23 protein, or its mammalian counterpart, TSG101, contains ubiquitin receptors (Dupré et al., 2001). E. histolytica possesses the ubiquitination machinery (Bosch and Siderovski, 2013), this, and the predicted ubiquitin domain in EhVps23 and its movement during phagocytosis ( Figure 5A ), suggested that EhVps23 could be also ubiquitinated or be in contact with ubiquitinated proteins. Thus, we used a commercial antibody against ubiquitin to explore this. In basal conditions, the α-ubiquitin antibodies detected ubiquitin in the cytoplasm of trophozoites, but reduced co-localization with EhVps23 was observed ( Figure 5A ). However, after 2 + 5 min of phagocytosis, an extensive co-localization of both proteins emerged, and after 2 + 60 min, when EhVps23 was returned to its basal state, both proteins were observed again separated ( Figures 5A, B ), suggesting that ubiquitination could activate EhVps23 to carry out its tasks during phagocytosis, which is supported by the fact that K50, K67, K100, K370, K403, K435, K436 and K445 amino acids are susceptible of ubiquitination. However, the presence of the UEV domain also makes probable that EhVps23 is interacting in a non-covalent way with ubiquitinated proteins, and that this interaction allows EhVps23 to perform its functions during erythrophagocytosis. Both events could be happening. After a certain time, ubiquitin and EhVps23 appeared again separated in the cell. In western blot assays, the α-ubiquitin commercial antibody recognized several bands ( Figure 5C ) corresponding to EhUbiquitin-conjugated proteins. The 10 and 15 kDa bands detected by the antibody exhibited the molecular weight deduced by the amino acids forming the E. histolytica ubiquitin (Wostmann et al., 1992; Arya et al., 2012).

EhADH participates along the process of phagocytosis (García-Rivera et al., 1999). Associated with the EhCP112 cysteine protease, it forms the EhCPADH complex involved in the adherence to and destruction of target cells (García-Rivera et al., 1999). After the EhCPADH complex contacts the prey, it surrounds the ingested RBCs and penetrates the phagosomes, interacting with ESCRT-III proteins during ILVs formation (Avalos-Padilla et al., 2018). EhADH also binds to cholesterol and to the EhNCP1 and EhNCP2 proteins, responsible for cholesterol transport (Bolaños et al., 2016), and to the LBPA phospholipid (Castellanos-Castro et al., 2016a), acting as a scaffold molecule, performing distinct cellular functions. We explore here the relationship of EhVps23 with EhADH, EhVps32 and LBPA, three molecules involved in phagocytosis.

In basal conditions, EhADH and EhVps23 co-localized in regions close to the inner plasma membrane, as well as in small vesicles ( Figures 6A, B ). After the trophozoites sense the RBCs, both proteins move to the phagocytic cups ( Figure 6A ). At longer times, interaction persisted in the plasma membrane. Interestingly, we distinguished EhVps23 forming bunches of vesicles close to the membrane and in some labelled spots outside the cell with EhADH ( Figure 6A , 30 min). A recent proteomic study describes the presence of EhVps23 protein in secretion products (Sharma et al., 2020). The proximity and the cellular movement of these two proteins during phagocytosis support their involvement in vesicular traffick and phagocytosis. In contrast, we detected a minor interaction between EhVps23 and EhVps32 ( Figures 6C, D ). Interestingly, both proteins were distinguished inside MVBs, indicating a specific function of EhVps23 that could act as a carrier of molecules that are directed to these structures.

LBPA associates to EhADH and it is a marker of late endosomes (Castellanos-Castro et al., 2016a). We investigated whether this phospholipid interacts with EhVps23 in the convoluted membranous structures observed through TEM (see Figure 4 ). In basal conditions, LBPA was found in vesicles in the cytoplasm and as discrete spots in the plasma membrane. Interestingly, 5 min after the RBCs stimulus was given; both molecules were visualized around the ingested RBCs and MVBs, although we also observed many clumps of LBPA without any interaction with EhVps23 ( Figures 7A, B ). Inside large phagosomes, networks of strands stained by the α-LBPA antibody appeared close to EhVps23 covering big areas of the phagosomes, and its ramifications reached the plasma membrane and the extracellular space ( Figure 7A , 30 min). By TEM experiments, we also investigated whether LBPA was present in the convoluted membranous structures, co-localizing with EhVps23, using gold labeled antibodies of different sizes. TEM images showed both molecules together in the membranous structures ( Figure 7C a-d), in vacuoles, and around phagosomes ( Figure 7C ). These results together give a panoramic of the complex process of phagocytosis, from the prey’s capture up to its digestion, and evidence distinct molecules and structures participating in these functions, as other authors have also shown (García-Rivera et al., 1999; Petri et al., 2002; Perdomo et al., 2016; Javier-Reyna et al., 2019). They also evidenced the role of EhADH, ESCRT proteins and LBPA in phagocytosis.

We delve into the study of EhVps23 putative interactions with EhUbiquitin, EhADH, EhVps32 and LBPA by docking analysis and immunoprecipitation assays. First, we aligned the homologous crystal structure of the complex Vps23-ubiquitn (from S. cerevisiae) (1UZX), with the EhUbiquitin crystal (4GSW) (Teo et al., 2004; Bosch and Siderovski, 2013). Then, we used the coordinates from this analysis to carry out a second alignment with the EhVps23 3D model obtained here. The structural alignment presented an RMSD of 1.577 ( Figure S4 ), whereas the interaction site between EhVps23 and EhUbiquitin matched with the binding site reported by the crystal structure of S. cerevisiae Vps23 interacting with Bos taurus ubiquitin (Teo et al., 2004). In docking analysis, the binding site between EhUbiquitin and the EhVps23 UEV domain appeared conserved ( Figure 8A ). The residues from EhVps23 binding to EhUbiquitin were: T44, S45, R46, I47, I91, H101, E106, N107, G108 and V109, while the residues from EhUbiquitin were: L11, I47, F48, A49, G50, K51, Q65, K66, E67, S68, T69, H71, V73 and R75 ( Figure 8A ).

Docking analysis also predicted that EhVps23 interacts with EhADH, EhVps32 and LBPA. The global free energy of EhVps23-EhADH interaction was calculated in -186.14 kcal/mol, given by two salt bridges and thirteen hydrogen bonds. The EhVps23 residues that interacted with EhADH were N64, R66, E96, E106, Q144, T149 and N339 located mainly in the UEV domain; meanwhile, in EhADH were R172, D219, K222, I227, T228, F231, Q278, K270, N309 and N310, located in the Bro1 domain ( Figure 8A ). The EhVps23-EhVps32 interaction showed a global free energy of -155.13 kcal/mol. The binding site was formed by four salt bridges and fifteen hydrogen bonds. The residues from EhVps23 interacting with EhVps32 were L62, K99, N132, R136, Y137, P138, R141, Q142, N145, H147, T149 and S175 located mainly in the UEV domain. The residues from EhVps32 were T105, K106, D112, D121, N127, C131, L137, G138, E139, D140, L141, Q142, I144, D145, E148, and E150, located in the Snf7 domain ( Figure 8A ).

EhVps23 bound to LBPA with an energy of -9.3 kcal/mol. The residues from EhVps23 interacting with LBPA were M1, Q2, I4, E7, K8, N11, E25, P323, G324, D326, L329, A330, N331 and E386, located in amino terminus and between UEV and Vps23 core domains ( Figure 8A ).

These interactions were experimentally tested by immunoprecipitation assays using α-EhVps23 antibodies and total extracts of trophozoites. EhUbiquitin, EhADH and EhVps32 appeared in the immunoprecipitates. As a negative control we used a preinmune serum ( Figure 8B ). These results evidenced the direct or indirect association of EhVps23 with these molecules involved in the phagocytosis process.

To obtain additional data on the function of EhVps23, we knocked-down the gene using dsRNA (Solis et al., 2009). Western blot assays using silenced trophozoites (Ehvps23-KD), evidenced that EhVps23 is expressed between 35 to 40% less in the knocked down trophozoites than in the control ( Figures 9A, B ). The immunofluorescence assays corroborated these results ( Figures 9C, D ). The rate of growth of the Ehvps23-KD trophozoites was retarded, but after five days of incubation at 37°C, cultures reached the same number of cells that the control did at four days ( Figure 9E ). These results are inconcordance with results obtained in other systems (Leung et al., 2008).

Next we evaluated the role of EhVps23 in the rate of phagocytosis using the silenced trophozites. Interestingly, at 15 min, the rate of phagocytosis of the Ehvps23-KD trophozoites was like the controls, but at 30 min, the Ehvps23-KD cells phagocytosed half of the RBCs than the control cells ( Figures 9F, G ). These results strengthen the assumption that the EhVps23 protein is involved in the phagocytosis of trophozoites, probably due to its active participation in vesicular trafficking.