Homo sapiens and Saccharomyces cerevisiae Vps35 protein sequences (access Q96QK1 and P34110, respectively), retrieved from the UniProt database (https://www.uniprot.org), were used as query to search the E. histolytica Vps35 proteins (EhVps35) in the AmoebaDB database (https://amoebadb.org/amoeba/app). The structural domains of the candidates were identified using the SMART genomics server (http://smart.embl-heidelberg.de/) and the KEGG server (https://www.genome.jp/kegg/). The predicted amino acid sequences of the five E. histolytica putative EhVps35 proteins (access C4M8Z4, C4M4G0, A0A8U0WP17, C4LWG4, and C4M3E8) were aligned with the orthologous sequences of H. sapiens (access Q96QK1), S. cerevisiae (access P34110), Arabidopsis thaliana (access Q7X659, F4I0P8, and A8R7K9), Plasmodium falciparum (access Q8IIQ6), T. gondii (access S7USD8), Trypanosoma brucei (access Q38C17), and Giardia lamblia (access Q6Y0Y2).

The complementary DNA (cDNA) was obtained as described (Galindo et al., 2021). The Ehvps35_{1–1,428 bp} and Ehvps35_{1–543 bp} DNA fragments were PCR-amplified using cDNA as template and specific primers.

Underlined sequences correspond to the enzyme restriction sites added to the primers. Fragments were cloned in pColdI to produce recombinant proteins tagged with 6× His and in pL4440 for the silencing experiments. Thus, we generated the pColdI/Ehvps35_{1–1,225 bp} and pL4440/Ehvps35 plasmids, respectively. The quality of the constructs was verified by restriction enzyme analysis and automatic DNA sequencing.

Escherichia coli BL21 (DE3) bacteria were transformed with the pColdI/Ehvps35_{1–1,225 bp} plasmid to produce the His-tagged EhVps35_{1–476 aa} recombinant protein (rEhVps35). Protein expression was induced by 0.4 mM isopropyl β-d-thiogalactopyranoside (IPTG) in Luria–Bertani (LB) medium for 16 h at 16°C. The identity and integrity of the purified rEhVps35 protein were verified by 10% SDS-PAGE and Western blot assays.

To generate mouse polyclonal antibodies against a specific EhVps35 peptide (N₃₀-ESEIMNAALNNKDLSK-C₄₅), male Balb/c mice (from an already existing collection in UPEAL-CINVESTAV) were immunized with 80 µg of the peptide resuspended in TiterMax Gold adjuvant (1:1; Sigma-Aldrich, St. Louis, MO, USA). Thereafter, three more immunizations were performed at 15-day intervals, followed by bleeding to generate the α-EhVps35 antibody. Pre-immune serum was obtained before immunizations.

The other primary antibodies used were as follows: mouse monoclonal α-biotin (Santa Cruz, Dallas, TX, USA), hamster α-EhADH (Galindo et al., 2022), mouse α-heavy subunit Hgl of Gal/GalNac lectin (kindly given by Dr. William A. Petri Jr., University of Virginia, Charlottesville, VA, USA), mouse α-Hist (Santa Cruz), mouse monoclonal α-human actin (kindly given by Dr. Manuel Hernandez, CINVESTAV IPN, Mexico), and rat α-EhVps23 (Galindo et al., 2021). As secondary antibodies, the following were used: horseradish peroxidase (HRP)-labeled α-hamster, α-rat, and α-mouse immunoglobulin G (IgG) (Zymed, San Francisco, CA, USA) for the Western blot assays and fluorescein isothiocyanate (FITC), Cy5, or the Pacific Blue α-mouse and Alexa 647 α-hamster IgG (Life Technologies, Carlsbad, CA, USA) for the immunofluorescence assays.

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

The trophozoite lysates (35 µg) were electrophoresed in 10% SDS-PAGE, transferred to nitrocellulose membranes, and probed with either α-EhVps35 (1:500), α-Hist (1:500), α-actin (1:3,000), α-Hgl Gal/GalNac lectin (1:100), α-EhADH (1:500), or α-EhVps23 (1:500) antibodies. The membranes were washed, incubated with the respective HRP-labeled secondary antibodies (1:1,000; Sigma), and revealed using the ECL Prime detection reagent (GE Healthcare, Chicago, IL, USA) according to the manufacturer’s instructions.

Trophozoites were grown on coverslips, fixed with 4% paraformaldehyde at 37°C for 1 h, permeabilized (or not) with 0.2% Triton X-100, and blocked with 10% fetal bovine serum in phosphate-buffered saline (PBS). The preparations were incubated at 4°C overnight (ON) with the primary antibodies (1:50). For some experiments, α-EhVps35 labeled with a FITC fluorochrome kit (Molecular Probes/Thermo Fisher, Waltham, MA, USA) was also used. The preparations were incubated with the corresponding secondary antibodies for 30 min at 37°C: FITC-labeled α-mouse IgG for α-EhVps35, Cy5-labeled α-mouse IgG for α-biotin, Pacific Blue-labeled α-mouse IgG for α-Hgl Gal/GalNac, and Alexa Fluor 647-labeled α-hamster IgG for α-EhADH. The nuclei were labeled with 4′,6-diamidino-2-phenylindole (DAPI) for 5 min at room temperature (RT), and F-actin was detected by incubation with rhodamine phallodin (Invitrogen, Carlsbad, CA, USA) for 20 min at 37°C. All preparations were preserved using a VECTASHIELD antifade reagent (Vector, Newark, CA, USA). Subsequently, 0.5-µm laser sections were obtained and examined through a Carl Zeiss LMS 700 confocal microscope and then processed with ZEN 2009 Light Edition software (Zeiss, Jena, Germany). To evaluate the co-localization between molecules, Pearson’s coefficients were obtained from at least 30 confocal images using ImageJ 1.45v software and the JACoP plug-in.

For the phagocytosis assays, the trophozoites were incubated at 37°C for 2, 5, and 30 min with red blood cells (RBCs; 1:25) from an already existing collection. At different times, the trophozoites were prepared for immunofluorescence (García-Rivera et al., 1982) and observed through a laser confocal microscope. Other preparations were stained using the Novikoff technique (Novikoff et al., 1972), and adhered or ingested RBCs were counted through a light microscope (Zeiss Axiolab) in 100 randomly chosen trophozoites. In other experiments (i.e., pulse-chase), the trophozoites were incubated with RBCs for 2 min at 37°C. Afterward, the cell mixtures were washed with TYI/water (2:1) for 5 min at 37°C to remove the adhered and non-ingested RBCs. Subsequently, the cells were incubated at 37°C for different times (3, 13, 28, 58, or 88 min) and processed for Western blot or immunofluorescence assays.

To visualize the acidic structures, the trophozoites were incubated with LysoTracker™ Blue DND-22 (2 μg/ml; Invitrogen) at 37°C for 2 h and were then prepared for the immunofluorescence assays as described above. In other cases, to inhibit de novo protein synthesis, the trophozoites were incubated for 30 min at 37°C with cycloheximide (CHX; 100 µg/ml), washed with PBS, and then used for the experiments after verifying that the viability was greater than 98%, which was evaluated using trypan blue exclusion (0.2%) and examined under a light microscope (Bañuelos et al., 2012).

The trophozoites in basal condition and those incubated for 5 min with RBCs (1:25) at 37°C were prepared for TEM as described (Talamás-Lara et al., 2021). Briefly, the trophozoites were fixed with 4% paraformaldehyde and 0.5% glutaraldehyde in PBS for 1 h at RT, washed with PBS, and dehydrated with increasing concentrations of ethanol. After infiltration, the samples were embedded in LR White resin (London Resin Co., London, UK) and polymerized under UV at 4°C for 48 h to obtain thin sections (60 nm), which were mounted on Formvar-coated nickel grids followed by ON incubation at 4°C with the α-EhVps35 antibody (1:50). Thereafter, the samples were incubated ON with the corresponding secondary antibodies (1:50) conjugated to 20-nm gold particles (TED Pella Inc., Redding, CA, USA), contrasted with uranyl acetate and lead citrate, and observed through a Joel JEM-1011 transmission electron microscope.

The knockdown (KD) of the Ehvps35 gene was performed using the bacterial expressed double-stranded RNA (dsRNA) that was soaked with the trophozoites as described (Solis et al., 2009). Briefly, HT115 bacteria were transformed with pL4440/Ehvps35 and grown at 37°C in LB broth in the presence of ampicillin (100 mg/ml) and tetracycline (10 mg/ml) (Takiff et al., 1989). The expression of Ehvps35 dsRNA was induced by 1 mM IPTG overnight at 37°C. The dsRNA was then isolated from bacteria using a TRIzol reagent (Invitrogen) according to the manufacturer’s recommendations. DNase I (Invitrogen) and RNase A (Ambion, Austin, TX, USA) were added to remove the single-stranded RNA (ssRNA) and double-stranded DNA (dsDNA) molecules. The Ehvps35 dsRNA was washed with isopropanol and ethanol, analyzed using agarose gel electrophoresis, and its concentration determined by spectrophotometry. Finally, the trophozoites (3 × 10⁴) in the TYI-S-33 medium were incubated with purified Ehvps35 dsRNA molecules to a final concentration of 5 μg/ml and the cultures left for 48 h at 37°C, the time at which we previously determined was the silencing effect. Cells growing under standard conditions (without dsRNA) were used as the control.

For the recycling assays, we followed the protocol described by Ivaska et al. (2002) and Roberts et al. (2001), with some modifications. The trophozoites were incubated for 1 h at 37°C in serum-free TYI medium and then washed once with cold PBS. The cell surface proteins were labeled with 0.2 mg/ml of EZ-Link™ sulfo-NHS-SS-cleavable biotin (Thermo Scientific, Waltham, MA, USA) at 4°C for 30 min (Biller et al., 2014). After this time, the trophozoites were washed with cold PBS and incubated in TYI complete medium at 37°C for 15 min to allow internalization of the labeled proteins. To remove labeling from the cell surface proteins, the trophozoites were incubated in 100 mM dithiothreitol (DTT) for 30 min at 4°C, washed once with cold PBS, incubated again with TYI complete medium for 20 or 40 min at 37°C, and then observed through a laser confocal microscope.

For analysis of EhADH protein and Gal/GalNac lectin recycling, the protocol described above was followed, omitting the treatment with biotin and DTT and using an incubation step of 40 min at 37°C to follow the recycling of the proteins. Thereafter, the samples were processed to confocal microscopy using the antibodies and labeled secondary antibodies in order to follow the recycling of EhADH and Gal/GalNac. Moreover, for some experiments, the trophozoites were incubated with CHX (100 µg/ml) at 37°C for 30 min. Lastly, the samples were prepared for laser confocal microscopy and Western blot assays.

The three-dimensional (3D) model of EhVps35 was obtained on the I-TASSER server using the Mus musculus (PDBID:6VAB: B) and H. sapiens (PDB:7BLN:C) Vps35 as templates. The 3D model was refined through 200 ns of molecular dynamics stimulation (MDS) by NAMD2.8 (Phillips et al., 2020) with the force field CHARMM36 to create the topologies of the protein (Huang and Mackerell, 2013). The TIP3 model was applied for the water molecules. The system was solvated using the psfgen plug-in in the VMD program (Humphrey et al., 1996). Thereafter, 64,480 water molecules and 12 sodium ions were added to neutralize the system, which was minimized for 10,000 steps, followed by equilibration under constant temperature and pressure (NPT) conditions for 1 ns with the protein and lipid atoms restrained. Afterward, MDS was run for 200 ns, considering EhVps35 as a soluble protein, without position restraints under periodic boundary conditions (PBCs) and using an NPT ensemble at 310 K and 200 ns of MD simulation. The structures were visualized using the UCSF Chimera software. Once the best EhVps35 model was chosen, 3D structural alignment with the Vps35 from H. sapiens (PDB:7BLN:C) was developed to compare both proteins.

The published 3D predicted and refined structures (Montaño et al., 2017) of EhADH were used for docking experiments, whereas the 3D model of the EhVps35 protein was generated in this work. Snapshots were obtained using clustering analysis of the 200-ns MDS with the Carma software (Koukos and Glykos, 2013). Protein–protein docking was performed employing different conformers with the ClusPro server (Comeau et al., 2004; Kozakov et al., 2013). The conformers with the highest cluster members and the lowest energy were analyzed on the PDBSum server (Laskowski et al., 1997), while 3D model visualization was performed by the VMD program (Humphrey et al., 1996).

Immunoprecipitation was carried out using trophozoite lysates prepared in the presence of 10 mM Tris-HCl, 50 mM NaCl, and 100 mM protease inhibitors (i.e., PHMB, IA, NEM, and TLCK), followed by freeze–thawing cycles in liquid nitrogen and vortexing (Avalos-Padilla et al., 2015). In parallel, 200 μl of the recombinant protein G-agarose (rProtein-G; Invitrogen) was incubated with 100 μg of the mouse αEhVp35 antibodies or the pre-immune serum for 2 h at 4°C, with gentle stirring. Thereafter, the rProtein-G beads were washed with 0.5% bovine serum albumin (BSA) in PBS, followed by additional washes with PBS for 5 min, under gentle stirring, and centrifuged at 11,600 × g for 2 min. The trophozoite lysates (1 mg) were pre-cleared with 200 μl of rProtein-G (previously blocked with 2% BSA) and incubated for 2 h at 4°C under gentle stirring. The samples were then centrifuged at 11,600 × g to obtain the supernatant that was added to the rProtein-G previously incubated with antibodies. The preparations were incubated ON at 4°C and the beads recovered by centrifugation. After washes with PBS, 60 μl of 4× sample buffer (40% glycerol, 240 mM Tris-HCl, pH 6.8, 8% SDS, 0.04% bromophenol blue, and 5% β-mercaptoethanol) was added. The samples were boiled for 3 min and centrifuged again at 11,600 × g for 2 min at 4°C. The supernatant (30 μl) was loaded into 12% SDS-PAGE, subjected to Western blot assays, and revealed using the α-EhVps35, α-EhADH, α-Hgl Gal/GalNac, and α-actin antibodies.

Approximately 30 μg of the immunoprecipitated proteins was enzymatically digested according to the protocol reported by Ramírez-Flores et al. (2019). Afterward, the peptides were loaded into the Symmetry C18 Trap V/M precolumn (Waters, Milford, MA, USA), 180 μm × 20 mm, 100-Å pore size, and 5-μm particle size, and desalted using mobile phase A (0.1% formic acid in H₂O) and mobile phase B (0.1% formic acid in acetonitrile) under the following isocratic gradient: 99.9% mobile phase A and 0.1% mobile phase B at a flow of 5 μl min⁻¹ for 3 min. Subsequently, the peptides were loaded and separated on an HSS T3 C18 column (Waters), 75 μm × 150 mm, 100-Å pore size, and 1.8-μm particle size, using a UPLC ACQUITY M-Class (Waters) with the same mobile phases under the following gradient: 0 min with 7% mobile phase B, 121.49 min with 40% mobile phase B, 123–126.46 min with 85% mobile phase B, and 129–130 min of 7% mobile phase B, at a flow of 400 nl min⁻¹ at 45°C. The spectra were acquired with a mass spectrometer using electrospray ionization and ion mobility separation with Synapt G2-Si (Waters) using a data-independent acquisition approach in HDMSE mode (Waters). The generated raw files containing the MS and MS/MS spectra were deconvoluted and compared using the ProteinLynx Global SERVER (PLGS) v3.0.3 software (Li et al., 2009) against a reversed E. histolytica database (downloaded from UniProt). All identifications had a reliability ≥95%.

The values for all experiments were expressed as the mean and standard error of at least three independent assays carried out in duplicate. Statistical analyses were performed with the GraphPad Prism v5.01 software using 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 Federal Register no. 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 is under 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 protocol approval involving the use of laboratory animals, reviewed and approved all animal experiments (protocol no. 0505-12, CICUAL 001).

The canonical retromer in yeast and mammals is formed by Vps26, Vps29, Vps35, and SNX dimer (Vps5 and Vps17 in yeast and SNX1, SNX2, SNX5, and SNX6 in mammals). In contrast, only the EhVps26, EhVps29, EhVps35, and EhSNX1 proteins have been identified in E. histolytica (Nakada-Tsukui et al., 2005; Watanabe et al., 2020). In addition, unlike other SNX orthologs, EhSNX1 presents only the PX and not the BAR domain ( Figure 1A ). From what is known so far, EhVps26 is the only retromeric protein that binds to EhRab7A ( Figure 1A ) (Nakada-Tsukui et al., 2005), in contrast to the retromer of other species where Vps35 also interacts with Rab7 on the endosomal membranes ( Figure 1A ) (Seaman et al., 2009; Liu et al., 2012; Purushothaman et al., 2017).

In the E. histolytica genome, five EhVps35 isoforms (A–E) were found, but only two of them (EhVps35A and EhVps35C) were conserved in the complete PRLYL motif ( Figure 1B ), which allows the binding of Vps26 to Vps35 in yeast and mammalian cells. For this study, EhVps35C was selected as, so far, it is the only E. histolytica protein with experimental evidence of direct interaction with EhVps26 and EhVps29 (the other two proteins of the CSC complex) and it is translationally active (Marion et al., 2005; Nakada-Tsukui et al., 2005; Biller et al., 2014; Perdomo et al., 2015). However, the functionality of the other four EhVps35, found in AmoebaDB, cannot be discarded. In fact, there are reports on a functional Vps35 isoform without the complete PRLYL motif in A. thaliana (Jaillais et al., 2007; Hashiguchi et al., 2012; Zelazny et al., 2013).

To study the cellular location and follow the movement of the EhVps35C (hereinafter EhVps35) protein in the trophozoites, we generated specific antibodies against an EhVps35 peptide (N₃₀-ESEIMNAALNN KDLSK-C₄₅) located at the amino terminal region of the protein and absent in other isoforms ( Figures 2A, B ). This polypeptide was commercially synthetized and injected into mice to generate antibodies (α-EhVps35). An EhVps35 fragment (1–476 amino acids) was also used to obtain the recombinant EhVps35 polypeptide (rEhVps35), labeled with a histidine tag (His-tag). The rEhVps35 was used to verify the specific recognition of the α-EhVps35 antibody in the Western blot assays. The α-Hist and α-EhVps35 antibodies recognized the same 48-kDa band in the lysates obtained from transformed bacteria ( Figure 2C ). Similarly, the α-EhVps35 antibody recognized a 71-kDa band in the trophozoite lysates ( Figure 2C ). The expected molecular weight of the EhVps35 protein, calculated from its amino acid sequence, was 87.2 kDa. However, the aberrant migration of EhVps35 was not surprising as this has been widely reported for the Vps35 protein in humans (Haft et al., 2000; Damen et al., 2006; Zhao et al., 2007; Fjorback et al., 2012) and for other EhVps in E. histolytica (Avalos-Padilla et al., 2015). The aberrant migration behavior could be explained by the protein structure and charge.

With a confocal microscope, in permeabilized trophozoites, the antibodies were shown to detect the EhVps35 protein in clusters, some of them located in vesicles scattered in the cytoplasm and near the plasma membrane ( Figure 2D ). Intriguingly, in non-permeabilized trophozoites, the EhVps35 protein appeared in the plasma membrane ( Figure 2D ), and TEM assays confirmed that it was located on the outer face of the plasma membrane and in vesicle membranes ( Figure 2E , e1). Interestingly, EhVps35 was observed in elongated structures, suggesting that membrane remodeling possibly occurred at these points, an event necessary for protein recycling ( Figure 2F , f1). The scheme in Figure 2G shows a diagram of the vesicle projections that could be forming the tubular retromer structures described in other systems (Bonifacino and Hurley, 2008; van Weering et al., 2010), and Figure 2H corresponds to the control treated with secondary antibody.

Phagocytosis is one of the landmarks of trophozoite virulence. This function requires high activity in membrane remodeling to form the phagocytic cups and channels, the endosomes, the multivesicular bodies (MVBs), and other vesicles. At the same time, protein recycling is needed in the cell to reuse certain molecules and save energy. Thus, being a complex involved in the selection and recycling of proteins, the retromer may also be involved in phagocytosis.

We explored the fate of the EhVps35 protein under the RBC stimulus, tracking the protein with the α-EhVps35 antibodies and the FITC-labeled α-mouse secondary antibodies. Confocal images revealed that after 2 min contact of the trophozoites with RBCs, the EhVps35 protein was translocated at the contact site of both cells and was visualized decorating the adhered RBCs, as has been previously reported for the EhSNX1 and EhVps26 retromeric proteins (Nakada-Tsukui et al., 2005; Srivastava et al., 2017; Watanabe et al., 2020). EhVps35 was also found in the ingested RBCs, which confirms the high mobility of this protein from the vesicle membranes to the plasma membrane and vice versa, suggesting that the retromer acts in concert with other molecules during phagocytosis.

In some experiments, after 2 min of phagocytosis, the adhered and non-ingested RBCs were removed by mild osmotic shock and then incubated again in TYI medium. These pulse-chase experiments were carried out to follow the fate of the already ingested cells, without the noise of other RBCs recently adhered and ingested. In these experiments, at 5 (2 + 3) and 15 (2 + 13) min, the antibodies again detected the EhVps35 protein in cytoplasmic vesicles, as well as on and in ingested erythrocytes. At 30 (2 + 28) min, the protein was located on and around the ingested erythrocytes and in phagosomes ( Figure 3A ). Our experiments showed that EhVps35, as are other retromeric proteins ( Figure 3B ) (Nakada-Tsukui et al., 2005, 2009; Watanabe et al., 2020), is located in the vesicle membranes that surround the ingested erythrocytes. Our results also evidenced that EhVps35, as are other proteins involved in phagocytosis (Avalos-Padilla et al., 2015; Galindo et al., 2021), follows the path of the ingested target cell.

Intriguingly, the Western blot assays of the lysates obtained from the trophozoite–RBC mixture revealed that the amount of EhVps35 protein apparently decreased during phagocytosis ( Figure 3C ). Densitometry analysis confirmed that, after 10 min of phagocytosis, the EhVps35 protein significantly decreased between 20% and 40% with respect to the 0 time ( Figure 3D ). This condition could be due to the partial degradation of the cellular amount of this protein during phagocytosis. Actin was used as a control to normalize the data for quantification ( Figures 3C, D ).

The TEM images, using gold-labeled secondary antibodies, confirmed that, after 5 min of phagocytosis, the EhVps35 protein was located at the contact sites of the trophozoites with RBCs, as observed in confocal images ( Figure 4A , a1). Subsequently, EhVps35 was observed in the RBC-containing phagosomes. The gold label appeared mainly in the endosome and phagosome membranes, where it could have undergone an eventual membrane remodeling, which is necessary for protein recycling ( Figures 4B–D , b1 - d1, respectively). In TEM assays, the RBC-containing phagosomes presented white intricate arrangements (WIAs) that have been reported in other studies on proteins involved in phagocytosis (Avalos-Padilla et al., 2015; Galindo et al., 2021). These WIAs could partially correspond to the digested hemoglobin that is inside these vesicular structures, which have not been thoroughly described. However, the location of the EhVps35 protein in the WIAs could also suggest that the retromer carried out its function of selecting proteins to recycle them to different cellular compartments ( Figure 4E , e1). More experimental evidence is necessary to confirm this hypothesis.

To experimentally demonstrate that recycling occurred in the trophozoites in this study and that the EhVps35 protein is involved in this function, we knocked down the Ehvps35 gene using dsRNA (Solis et al., 2009). Western blot assays using silenced trophozoites (Ehvps35-KD) evidenced that they expressed approximately 60% less EhVps35 than the control ( Figures 5A, B ). The results of the immunofluorescence assays corroborated these findings ( Figures 5C, D ). Subsequently, we used biotin (S-NHS-SS-Biotin) to label the plasma membrane proteins (Biller et al., 2014). Under a confocal microscope, the biotinylated trophozoites, maintained at 4°C, displayed the plasma membrane proteins uniformly labeled in both the control and Ehvps35-KD trophozoites ( Figure 5E , 1). Thereafter, the trophozoites were incubated for 15 min at 37°C and the label internalized in both cell types ( Figure 5E , 2). Interestingly, after 20 min incubation at 37°C, the labeled proteins returned to the plasma membrane in the control trophozoites, but not in the Ehvps35-KD trophozoites ( Figure 5E , 4). The incubation time at 37°C was augmented to 40 min; however, recycling in the Ehvps35-KD trophozoites remained poor ( Figure 5E , 4). After quantification of the fluorescence intensity in the plasma membrane, these assays confirmed that significant recycling reductions of 50% and 60% occurred in the Ehvps35-KD trophozoites compared to the control ( Figure 5F ). These results strongly suggest that the EhVps35 protein, even when located in the plasma membrane, does not participate in the introduction of the target cell, but has a vital role in protein recycling.

To analyze the interaction of EhVps35 with the other membrane proteins involved in phagocytosis, the complete EhVps35 3D model was used. A previous report showed the 3D design of the EhVps35 carboxy terminus (Srivastava et al., 2017), but not the complete protein. Here, we obtained the full 3D protein structure from the I-TASSER server ( Figure 6D ). The structural alignment of this model exhibited a root mean square deviation (RMSD) of 1.07 Å with the HsVps35 (PDB:7BLN:C) crystal ortholog ( Figure 6F ). Analysis of the MDS trajectory measured using RMSD indicated that EhVps35 has an unstable trajectory oscillating ±3 Å, and it is independent of the simulation time ( Figure 6A ). The radius of gyration (R _g) data confirmed that EhVps35 is an extended protein that is compacted during the MDS in the first 25 ns, after which is compacted again at 100 ns and expanded after 100–150 ns, remaining unchanged until the end of the simulation ( Figure 6B ). However, the root mean square fluctuation (RMSF) values showed a high movement of the protein in the N-terminus region from 1 to 199 amino acids and at the C-terminus from 501 to 757 amino acids ( Figure 6C ). This analysis showed that EhVps35 has the 3D model described for its orthologs as an α-solenoid protein ( Figure 6E ) (Hierro et al., 2007; Lucas et al., 2016) containing 33 α-helices ( Figure 6D ), confirming that EhVps35 is indeed an ortholog of the HsVps35 protein in humans.

Once the EhVps35 model was obtained, docking analysis was carried out between EhVps35 and EhADH using the 3D EhADH model already reported (Montaño et al., 2017). The docking results predicted that EhVps35 interacts with EhADH with a global free energy of ΔG = −920.6, given by two salt bridges and 19 hydrogen bonds ( Figure 7A ). The EhVps35 residues that interacted with EhADH were H271, E312, L315, D316, Q320, R321, N323, S329, and Y233. On the other hand, in EhADH, these were K98, K100, Y445, Y457, R663, Q673, Q674, P675, Y676, G679, T680, and N681. The binding of EhADH to EhVps35 did not interfere with the Bro1 or adhesion domains in the EhADH protein as, except for the K98, K100, and Y457 amino acids that appeared located in these domains, most of the EhADH residues appeared situated at the C-terminus after the adhesion domain.

Subsequently, we experimentally examined the EhVps35–EhADH binding with immunoprecipitation assays using the α-EhVps35 antibodies for precipitation and the total trophozoite extracts. The Western blot assays revealed that the α-EhADH antibodies recognized a 75-kDa protein in the immunoprecipitates, corresponding to the EhADH protein migration ( Figure 7B ). These results strongly suggest the association of EhVps35 with EhADH, a molecule that is actively mobilized during phagocytosis (Arroyo and Orozco, 1987; García-Rivera et al., 1999). Considering the docking analysis, the binding of EhVps35 and EhADH could occur by direct interaction; however, we cannot rule out that an indirect link can also happen.

In these experiments, we searched for the Gal/GalNac lectin, another membrane protein mobilized during virulence events (Petri et al., 2002; Frederick and Petri, 2005), and for actin as, during the retromer function, it binds to EhSNX-1 and EhSNX-1 interacts with the EhVps35 protein (Watanabe et al., 2020). In addition, Gal/GalNac lectin binds to actin and to other cytoskeletal proteins (McCoy and Mann, 2005). The EhVps35, EhADH, Gal/GalNac, and actin proteins appeared in the immunoprecipitates obtained with the α-EhVps35 antibody, strongly suggesting that they bind to each other. The pre-immune serum did not recognize any band ( Figure 7B ).

Once the association of EhVps35 with EhADH and Gal/GalNac was confirmed, we proceeded to investigate whether the retromer recycles these proteins using the Ehvps35-KD trophozoites and the respective antibodies against the two proteins. Firstly, under a confocal microscope, the α-EhVps35, α-EhADH, and α-Hgl Gal/GalNac antibodies in the control trophozoites were detected to co-localize in the plasma membrane and in some cytoplasmic regions ( Figure 8A ), suggesting the formation of EhADH and Gal/GalNac protein complexes associated with EhVps35. In contrast, in the Ehvps35-KD trophozoites, the fluorescent signals of the three antibodies were poor ( Figure 8A ). Weak labeling was also observed in dots scattered in the cytoplasm, without visible labels in the plasma membrane. The three antibodies also exhibited very poor co-localization in Ehvps35-KD trophozoites ( Figure 8 ). These experiments demonstrated that, when the EhVps35 protein is poorly expressed, the recycling of EhADH and Gal/GalNac lectin is affected, strongly supporting the participation of EhVps35 in this event.

Subsequently, to avoid interference of the proteins that were de novo synthesized and then transported to the plasma membrane, we used CHX to stop the protein synthesis and followed only those proteins present in the cell at the beginning of the recycling experiments. In the confocal images of control cells, the α-EhVps35, α-Hgl Gal/GalNac, and α-EhADH antibodies were again detected in the plasma membrane ( Figure 8B , arrow) and in some dots in the cytoplasm ( Figure 8B , arrowhead). The three proteins appeared to mainly co-localize in clusters in the plasma membrane, suggesting that CHX did not affect their interaction, nor the morphology of the trophozoites, as can be observed in the phase-contrast images ( Figures 8A, B ). Cell viability was neither affected, which was measured using trypan blue exclusion. In contrast, the Ehvps35-KD cells showed poor fluorescence labels, and their cellular location was also disturbed due to the proteins remaining scattered in the cytoplasm, but not in the plasma membrane ( Figure 8B ). Several Ehvps35-KD trophozoites presented a large cytoplasmic vesicle labeled by antibodies against the remnant EhVps35 and EhADH, and protein co-localization appeared poor and in small cytoplasmic vesicles ( Figure 8B ). These results confirmed that the recycling of EhADH and Gal/GalNac proteins was affected by the knockdown of the Ehvps35 gene.

To further explore the decrease of the EhADH and Gal/GalNac proteins observed in Ehvps35 knockdown, Western blot assays were performed using the lysates from untreated and CHX-treated trophozoites. In untreated Ehvps35-KD trophozoites, the Gal/GalNac protein significantly decreased ( Figures 8C, D, F ), but the EhADH and actin proteins did not show apparent changes ( Figures 8C, E, G ). In contrast, in CHX-treated Ehvps35-KD trophozoites, the EhADH, Gal/GalNac, and actin proteins significantly decreased between 20% and 60% with respect to the control trophozoites ( Figures 8C–G ). The α-EhVps23 antibody was used to normalize the data. These results suggest a possible degradation of EhADH, Gal/GalNac, and actin in the absence of the EhVps35 protein and the treatment with CHX.

Previous studies have shown that EhADH can be located in acidic vesicles during phagocytosis (Castellanos-Castro et al., 2016), suggesting that a portion of this protein is conducted to degradation and another portion is recycled, as shown here ( Figures 8A, B ). Thus, LysoTracker was used to investigate whether EhADH appeared in acidic vesicles in the control and the CHX-treated Ehvps35-KD trophozoites. In basal conditions, the control trophozoites showed that the EhVps35 and EhADH proteins co-localized, but their co-localization with LysoTracker was low. Interestingly, the CHX-treated Ehvps35-KD trophozoites showed higher co-localization of both proteins with LysoTracker ( Figures 9A–F ). These results suggest that a greater degradation of both proteins could have occurred in Ehvps35-KD trophozoites compared to the control. After phagocytosis (2 min), in the control trophozoites, the EhVps35 and EhADH proteins moved to the contact site with RBCs. In contrast, the Ehvps35-KD trophozoites appeared poorly labeled and the protein co-localization with LysoTracker was also poor ( Figures 9A–F ). In the control cells, after 30 min of phagocytosis, the EhVps35 and EhADH proteins re-localized to the plasma membrane and co-localized with LysoTracker in some RBC-containing phagosomes, suggesting that, during phagocytosis, a portion of them were conducted to degradation. In contrast, the CHX-treated Ehvps35-KD trophozoites presented fewer acidic vesicles, the EhADH protein did not return to the plasma membrane, and EhVps35 and EhADH remained in small points in the cytoplasm, with some of them co-localizing with LysoTracker ( Figures 9A, F ). Quantitative data of the fluorescence intensity of the EhVps35 and EhADH proteins and the acidic vesicle marker showed a significant reduction (60%–80%) in Ehvps35-KD trophozoites compared to the control cells ( Figures 9B–D ). These results strengthened the assumption that EhADH is degraded in the absence of the EhVps35 protein, and these remark the importance of the retromer in phagocytosis, recycling and preserving the proteins participating in these events.

The Western blot assays revealed a decrease of actin in Ehvps35-KD trophozoites ( Figures 8C, D ). To explore whether the actin cytoskeleton was affected by the Ehvps35 gene silencing, confocal microscopy assays were performed using CHX-treated trophozoites. In basal conditions, in the control trophozoites, EhVps35 and F-actin again co-localized mainly near the plasma membrane and in cytoplasmic structures, as described (Manich et al., 2018a). In contrast, Ehvps35-KD trophozoites presented the diminished fluorescence intensity of F-actin and disorganization of the actin cytoskeleton ( Figures 10A–C ). After 2 min of phagocytosis, the control trophozoites displayed the EhVps35 and F-actin proteins relocated to the contact site with the RBCs, decorating the phagocytic cups and channels ( Figure 10A ). However, these structures were not observed in the Ehvps35-KD trophozoites, suggesting that the silencing of the Ehvps35 gene affected the structuring of the actin cytoskeleton, which is necessary in the formation of phagocytic cups and channels ( Figures 10A, B ). Quantitative data of the fluorescence intensity of the EhVps35 and F-actin proteins showed a reduction of 40%–70% in Ehvps35-KD trophozoites compared to the control cells ( Figures 10B, C ). Co-localization analysis between EhVps35 and F-actin showed an increase in the control trophozoites after 2 min of phagocytosis, while the Ehvps35-KD trophozoites remained constant. Furthermore, in EhVps35-KD trophozoites, the co-localization between EhVps35 and F-actin was significantly lower compared to that in the control trophozoites ( Figure 10D ). These results suggest that CHX treatment did not affect the structure of the actin cytoskeleton. To obtain more insights on the relationship between the EhVps35 protein and actin cytoskeleton structuring, the immunoprecipitates with α-EhVps35 antibodies were analyzed using mass spectrometry. EhVps35-bound proteins were selected after comparison to previous proteomic analyses of the actin cytoskeleton in E. histolytica (Manich et al., 2018; Rath and Gourinath, 2020). The results revealed 33 cytoskeleton-associated proteins, including actin and cytoskeleton modulator proteins ( Table 1 ).

Finally, to examine the importance of the EhVps35 in phagocytosis, for its role in surface protein recycling and the actin cytoskeleton structure required for this event, the adhesion efficiency and the rate of phagocytosis were evaluated using the silenced trophozoites. The results showed a decrease of 50%–63% in the adhesion efficiency and 53%–61% in the rate of phagocytosis in Ehvps35-KD trophozoites at 5 and 30 min, respectively, with respect to the results of the control trophozoites, which were taken as 100% ( Figures 11A, B ). These results strengthened the assumption that the EhVps35 protein is involved in the phagocytosis of trophozoites, probably due to its active participation in protein recycling and in the structuring of the actin cytoskeleton.