|Home | About | Journals | Submit | Contact Us | Français|
Entamoeba histolytica, the protist that causes amebic dysentery and liver abscess, has a truncated Asn-linked glycan (N-glycan) precursor composed of seven sugars (Man5GlcNAc2). Here, we show that glycoproteins with unmodified N-glycans are aggregated and capped on the surface of E. histolytica trophozoites by the antiretroviral lectin cyanovirin-N and then replenished from large intracellular pools. Cyanovirin-N cocaps the Gal/GalNAc adherence lectin, as well as glycoproteins containing O-phosphodiester-linked glycans recognized by an anti-proteophosphoglycan monoclonal antibody. Cyanovirin-N inhibits phagocytosis by E. histolytica trophozoites of mucin-coated beads, a surrogate assay for amebic virulence. For technical reasons, we used the plant lectin concanavalin A rather than cyanovirin-N to enrich secreted and membrane proteins for mass spectrometric identification. E. histolytica glycoproteins with occupied N-glycan sites include Gal/GalNAc lectins, proteases, and 17 previously hypothetical proteins. The latter glycoproteins, as well as 50 previously hypothetical proteins enriched by concanavalin A, may be vaccine targets as they are abundant and unique. In summary, the antiretroviral lectin cyanovirin-N binds to well-known and novel targets on the surface of E. histolytica that are rapidly replenished from large intracellular pools.
Entamoeba histolytica causes amebic dysentery and liver abscess in the developing world (10, 20, 29). We are interested in E. histolytica glycoproteins containing Asn-linked glycans (N-glycans) for numerous reasons. E. histolytica makes an N-glycan precursor that contains 7 sugars (Man5GlcNAc2-PP-dolichol) rather than 14 sugars (Glc3Man9GlcNAc2-PP-dolichol) made by most animals, plants, and fungi (21, 31, 44). E. histolytica N-glycans are used for quality control of glycoprotein folding in the endoplasmic reticulum (ER) lumen, and there is positive selection for sites of N-linked glycosylation in secreted and membrane proteins of E. histolytica (5, 11, 53).
Unprocessed Man5GlcNAc2, by far the most abundant E. histolytica N-glycan, is present on the plasma membrane and vesicular membranes (31). The antiretroviral lectin cyanovirin-N, which is specific for α-1,2-linked mannose present on unprocessed N-glycans, binds E. histolytica N-glycans and forms aggregates or caps on the surface of E. histolytica trophozoites (1, 25, 31, 44, 45). E. histolytica glycoproteins are also capped by the plant lectin concanavalin A (ConA), which has a broader carbohydrate specificity (mannose and glucose) than cyanovirin-N (3, 16, 18, 19). Heavy subunits of the Gal/GalNAc lectin, the most important E. histolytica vaccine candidate, have 7 to 10 potential sites for N-linked glycosylation (32, 39, 43). Inhibition of N-glycan synthesis results in Gal/GalNAc lectins that are unable to bind to sugars on host epithelial cells.
Carbohydrates appear to be an important target on the surface of E. histolytica as anti-proteophosphoglycan (PPG) monoclonal antibodies bind to O-phosphodiester-linked glycans and protect animal models from amebic infection (6, 33, 35, 40, 48). Lectin affinity columns are a powerful method for enriching unique parasite glycoproteins that may be identified by mass spectrometry (MS) of tryptic fragments (17, 55). For example, we recently used the plant lectin wheat germ agglutinin to dramatically enrich glycoproteins with short N-glycans of Giardia (42).
The goal of the present studies was to explore further the interaction of the antiretroviral lectin cyanovirin-N with E. histolytica trophozoites in vitro. Questions asked included the following: Are E. histolytica glycoproteins with N-glycans replenished on the plasma membrane after capping with cyanovirin-N? What is the effect of cyanovirin-N capping on other amebic virulence factors and/or vaccine candidates (e.g., the Gal/GalNAc lectin and PPG)? Is capping by cyanovirin-N mediated by actin, as described for capping by the Gal/GalNAc lectin and ConA? What is the effect of the cyanovirin-N on amebic phagocytosis of mucin-coated beads, a surrogate assay for virulence? Which trophozoite glycoproteins are potential targets of cyanovirin-N (identified by mass spectrometry of lectin-enriched E. histolytica proteins)? Are any of them potential vaccine candidates?
Logarithmic-phase trophozoites of the genome project HM1 strain of E. histolytica were chilled to release adherent organisms, concentrated by low-speed centrifugation, and washed in chilled phosphate-buffered saline (PBS) (29). For surface labeling, trophozoites were incubated for 30 min at 4°C in cyanovirin-N labeled with either Alexa Fluor 488 (green) or Alexa Fluor 585 (red) (1, 31). Cyanovirin-N-labeled trophozoites were washed three times in PBS and then fixed for 10 min at 4°C in 2% paraformaldehyde in 100 mM phosphate, pH 7.4. For capping experiments, trophozoites labeled with cyanovirin-N were warmed to 37°C for 15 min prior to fixation.
To determine whether N-glycans are replenished on the surface of trophozoites capped with cyanovirin-N, we treated capped and fixed organisms with PBS containing 2% bovine serum albumin (BSA) to quench free aldehydes and then labeled them with cyanovirin-N conjugated to a different Alexa Fluor dye. To demonstrate actin fibrils, we permeabilized capped and fixed organisms with 0.1% Triton X-100 and then stained them with 0.1 mg/ml phalloidin conjugated to Alexa Fluor 480 for 1 h at 4°C (18). To determine whether cyanovirin-N cocaps other E. histolytica antigens, we incubated capped and fixed E. histolytica with an Alexa Fluor-labeled mouse monoclonal antibody to the Gal/GalNAc lectin (a generous gift of William Petri) (32, 39). Alternatively, capped and fixed E. histolytica organisms were incubated with an Alexa Fluor-labeled mouse monoclonal antibody to the E. histolytica PPG (a generous gift of Michael Duchêne) (33).
For internal labeling with cyanovirin-N, we fixed E. histolytica trophozoites for 10 min at 4°C, and Triton X-100 was added to a final concentration of 0.1% for 1 min. Cells were gently pelleted by centrifugation, washed with PBS-2% BSA, and then incubated with cyanovirin-N, as described above. Similar methods were performed for labeling the surface and interior of E. histolytica with anti-Gal/GalNAc antibodies and for determining whether Gal/GalNAc lectins are replenished on the parasite surface after capping.
The nuclei of E. histolytica cells labeled with cyanovirin-N or the anti-Gal/GalNAc antibody were stained with 0.1 μg/ml 4′,6′-diamidino-2-phenylindole (DAPI), SlowFade antifade solution (Invitrogen) was added, and organisms were visualized with a DeltaVision deconvoluting microscope (Applied Precision, Issaquah, WA) with channels for each fluorochrome. Images were taken at a primary magnification of ×100 and deconvolved using Applied Precision's softWoRx software.
Assays for E. histolytica phagocytosis of mucin-coated spheres were preformed, as described previously (18). Briefly, E. histolytica trophozoites (105/ml) were incubated with microspheres (107/ml) in culture medium for 15 min at 37°C and then fixed in 2% paraformaldehyde. Phagocytosed beads within 100 cells in each group were counted with a fluorescence microscope. The results were plotted using a modified box and whiskers plot to illustrate the significant shift in phagocytosed beads between the cyanovirin-N-treated and untreated trophozoites. In addition, an analysis of variance (ANOVA) was used to evaluate the statistical significance of the differences in phagocytosis and calculate the P value.
Logarithmic-phase E. histolytica trophozoites were harvested on ice, washed in PBS, and sonicated in an ice-water slurry containing 0.1% Triton X-100 and EDTA-free Complete protease inhibitor cocktail (Roche). Insoluble material was removed by centrifugation (at >12,000 × g). Soluble proteins were applied to a ConA-Sepharose column (EY Laboratories, Inc.) (17, 55), and the column was subsequently rinsed with PBS. To avoid collecting proteins that were nonspecifically bound to the ConA resin, we selectively eluted E. histolytica glycoproteins with 50 mM α-methyl mannoside rather than with SDS. Proteins eluted from the ConA column were run on SDS-PAGE gels containing a 4 to 20% gradient of acrylamide (Bio-Rad). In a parallel lane were E. histolytica proteins that were treated twice with 1,000 units of peptide:N-glycanase F (PNGaseF; New England Biolabs) for 9 h at 37°C in NEB G7 phosphate buffer. E. histolytica proteins were transferred to nitrocellulose membranes by electroporation, incubated with horseradish peroxidase (HRP)-conjugated cyanovirin-N, and developed with ECL chemiluminescent substrate (Pierce).
Mass spectrometry of E. histolytica proteins was performed using two different methods, as two different mass spectrometers were used. For the linear trap quadrupole (LTQ) ProteomeX ion trap mass spectrometer (Thermo Finnigan) present at the Boston University Proteomics Core Facility, E. histolytica peptides were prepared and analyzed using methods that were essentially the same as those used to identify peptides from the E. histolytica cyst wall (54) or from lectin affinity preparations of Giardia glycoproteins (42, 56). In addition, some samples were run on a similar Thermo Finnigan mass spectrometer at the Cancer Center at the Massachusetts Institute of Technology (MIT). Mass spectra were compared to tryptic digests of E. histolytica proteins predicted from whole-genome sequencing using SEQUEST, GPM (The Global Proteome Machine Organization [www.thegpm.org]) open source software, or Mascot software (13, 22, 23).
Two-dimensional protein gels were simulated from mass spectrometry data using GPM, where the position of each protein was determined by its predicted pI and mass, not including posttranslational modifications, and the size of the spot was proportional to the number of observed ions corresponding to that protein. These two-dimensional gels highlighted relative abundances of secreted and plasma membrane proteins (defined by either an N-terminal ER-targeting sequence or a transmembrane helix [TMH]) (24, 36) versus nucleocytoplasmic proteins (defined by the absence of these features). The Excel files in the supplemental material each show the merged results of four mass spectrometric experiments using Mascot software. Proteins previously identified as hypothetical because they showed no homology to other eukaryotic proteins were assigned simple names based upon their topology (e.g., unique nucleocytosolic protein, unique secreted protein, unique type 1 membrane protein, unique glycosylphosphatidylinositol [GPI]-anchored protein, etc.). GPI anchors were predicted using the algorithms of Eisenhaber et al. (15). Where there seemed a good match in the nonredundant (NR) database as demonstrated by a high score with BLASTP (2), we renamed the E. histolytica protein (e.g., “cysteine proteinase” or “disulfide isomerase” rather than “conserved hypothetical protein”).
To identify occupied N-glycan sites, we used a two-dimensional chromatography approach. A peptide mixture from the tryptic digestion of ConA-enriched E. histolytica glycoproteins was treated with PNGaseF to remove N-glycans and to convert Asn to Asp. PNGaseF-treated peptides and an untreated control were separated using strong cationic exchange (SCX) chromatography prior to Nanoflow reversed-phase high-performance liquid chromatography (HPLC)-coupled tandem mass spectrometry (MS/MS). SCX chromatography was performed on a Beckman Coulter ProteomeLab PF2D using a PolySulfoethyl A column. The buffers used were the following; buffer A, 7 mM KH2PO4, pH 2.65, 30% acetonitrile (ACN; vol/vol); buffer B, 7 mM KH2PO4, 350 mM KCl, pH 2.65, 30% ACN (vol/vol); buffer C, 50 mM K2HPO4, 500 mM NaCl, pH 7.5. Peptides were separated using a linear gradient from 0% to 70% of buffer B in 30 min, from 70% to 100% of buffer B in 10 min, and then 100% of buffer B for 6 min. The flow rate used was 0.5 ml/min. Thirteen 2-min fractions were collected. Each fraction was dried to eliminate acetonitrile before LC-MS/MS.
LC-MS/MS was performed using a nanoAcquity ultra-performance liquid chromatography (UPLC) capillary system (Waters Corp., Milford, MA), coupled to an LTQ-Orbitrap hybrid mass spectrometer (ThermoFisher Scientific, San Jose, CA) equipped with a TriVersa NanoMate ion source (Advion, Ithaca, NY). Sample concentration and desalting were performed online using a nanoAcquity UPLC trapping column (180 μm by 20 mm; packed with 5-μm, 100-Å-pore-size Symmetry C18 material; Waters Corp.) at a flow rate of 15 μl/min for 1 min. Separation was accomplished on a nanoAcquity UPLC capillary column (100 μm by 100 mm; packed with 1.7-μm,130-Å-pore-size bridged ethyl hybrid [BEH] C18 material; Waters Corp.). A linear gradient of A and B buffers (buffer A, 3% ACN–0.1% formic acid [FA]; buffer B, 97% ACN–0.1% FA) from 7% to 45% buffer B over 124 min was used at a flow rate of 0.5 μl/min to elute peptides into the mass spectrometer. Columns were washed and reequilibrated between LC-MS/MS experiments. Electrospray ionization was carried out at 1.7 kV using the NanoMate, with the LTQ heated capillary set to 150°C.
Mass spectra were acquired in the Orbitrap in the positive-ion mode over the range of m/z 300 to 2,000 at a resolution of 60,000. Mass accuracy after internal calibration was within 4 ppm. Simultaneously, tandem MS spectra were acquired using the LTQ for the five most abundant, multiply charged species in the mass spectrum with signal intensities of >8,000 noise levels. MS/MS collision energies were set at 35%, using helium as the collision gas, and MS/MS spectra were acquired over a range of m/z values dependent on the precursor ion. Dynamic exclusion was set such that MS/MS for each species was acquired a maximum of twice. All spectra were recorded in profile mode for further processing and analysis.
Xcalibur software was used for MS and MS/MS data analysis, while peptide and protein assignments were conducted using Mascot to search against the E. histolytica database employing an error window of 6 ppm on the precursor ions and 0.6 Da on the fragment ions. Table 1 shows occupied N-glycan sites where the predicted Asn was converted to Asp by PNGaseF treatment, resulting in a shift in mass of +1 Da.
Mass spectrometric data have been deposited in AmoebaDB (4).
Cyanovirin-N, which labels α-1,2-linked mannose residues in unprocessed N-glycans, evenly stains the surface of E. histolytica trophozoites either kept at 4°C to prevent capping or fixed prior to labeling (see Fig. S1 in the supplemental material). Glycoproteins containing N-glycans are capped on the surface of E. histolytica trophozoites when cyanovirin-N-labeled trophozoites are warmed to 37°C (Fig. 1A and C, large arrows). Subsequent labeling of fixed parasites with cyanovirin-N conjugated to a different Alexa Fluor dye shows that many glycoproteins containing N-glycans are replenished on the surface of E. histolytica trophozoites away from the cap (Fig. 1B and C, small arrows). Similarly, Gal/GalNAc lectins that are capped by a monoclonal antibody are replenished on the surface of E. histolytica away from the cap (Fig. 1D to F; see also Fig. S1 in the supplemental material).
The source of the new glycoproteins with N-glycans on the E. histolytica surface after capping is likely the large intracellular pool of glycoproteins containing N-glycans. These glycoproteins are clearly visible on cyanovirin-N labeling of fixed and permeabilized E. histolytica trophozoites (Fig. 1G). Cyanovirin-N binds in a reticular or membrane pattern to glycoproteins of permeabilized E. histolytica. In contrast, the anti-Gal/GalNAc lectin monoclonal antibody labels the membranes and the contents of numerous secretory vesicles throughout the E. histolytica trophozoite (Fig. 1H). A model for the replenishment of capped E. histolytica surface glycoproteins from large intracellular pools is shown in Fig. 1I and discussed below.
Consistent with the presence of 7 to 10 N-glycan sites on each heavy subunit of the Gal/GalNAc lectin, cyanovirin-N cocaps the Gal/GalNAc lectin (Fig. 2A to C). The presence of anti-Gal/GalNAc lectin antibody labeling in areas away from the cap (Fig. 2B and C) is consistent with spontaneous replenishment of the Gal/GalNAc lectin from the large intracellular pools concurrent with the capping event. Cyanovirin-N also cocaps glycoproteins recognized by the anti-PPG antibodies (Fig. 2D to F). There is binding of the anti-PPG in areas away from the cap (Fig. 2E), consistent with replenishment of the PPG from large intracellular pools (data not shown).
Filamentous actin, which is labeled by the fungal toxin phalloidin, is important for amebic motility, capping, and phagocytosis. Actin filaments accumulate in the region of the cyanovirin-N induced cap (Fig. 2G to I), as has been shown for caps by the plant lectin ConA and by the monoclonal antibody to the Gal/GalNAc lectin.
Cyanovirin-N inhibits phagocytosis of mucin-coated beads by E. histolytica trophozoites, a surrogate assay for amebic virulence (Fig. 3A to C). While untreated E. histolytica trophozoites phagocytose 43 ± 22 (mean ± standard deviation [SD]) mucin-coated beads, cyanovirin-N-treated trophozoites phagocytose 9 ± 11 (mean ± SD) beads (P 0.005). The inhibition of phagocytosis by cyanovirin-N is comparable to that caused by overexpression of a dominant negative p21rac mutant that interferes with localization of actin filaments during phagocytosis (18).
Lectin affinity chromatography was performed with ConA-Sepharose because glycoproteins can be eluted with excess α-methyl mannoside. In contrast, glycoproteins bound to cyanovirin-N-–epharose may only be eluted with SDS that introduces nonspecifically bound contaminants. Western blotting showed that cyanovirin-N conjugated to horseradish peroxidase binds to E. histolytica glycoproteins that were enriched by ConA affinity chromatography (Fig. 4A). In contrast, cyanovirin-N no longer binds to Entamoeba glycoproteins treated with PNGaseF to remove N-glycans (Fig. 4A). These results confirm that cyanovirin-N is binding only to E. histolytica N-glycans.
In the absence of ConA affinity chromatography, the vast majority (87%) of 302 E. histolytica proteins identified by mass spectrometry of tryptic peptides are nucleocytosolic (Fig. 4B and C, labeled blue). For example, when proteins are listed by their Mascot score, there are 41 nucleocytosolic proteins before the first secreted protein, a cysteine proteinase (see Excel file S1 in the supplemental material). While they are not the focus of the present study, nucleocytosolic proteins (many of which have greater than 50% peptide coverage) include enzymes involved in fermentation, glycolysis, and protein synthesis as well as chaperones and cytoskeletal proteins.
Following ConA affinity enrichment, the majority of E. histolytica proteins identified (52%) were membrane or secreted, as shown by the presence of N-terminal signals and/or transmembrane helices (Fig. 4B and C, labeled red). For example, 25 of the 30 proteins with the highest Mascot scores are secreted or membrane proteins rather than nucleocytosolic proteins (see Excel file S2 in the supplemental material). These glycoproteins (many of which have greater than 50% peptide coverage) include well-characterized virulence factors such as all three subunits of the Gal/GalNAc adherence lectin, as well as lysosomal proteases and phosphatases (see Excel file S2) (7, 10, 29, 32, 39). ER chaperones, protein disulfide isomerases, peptidyl-prolyl cis-trans isomerases, and calreticulin are all abundant. Of particular interest for discovery of potential vaccine candidates are 27 unique type 1 membrane proteins and six unique GPI-anchored proteins (see the FASTA file in the supplemental material). In the absence of information with regard to the location of any of these unique proteins, in the Excel files in the supplemental material the proteins with TMHs were arbitrarily assigned to the plasma membrane while proteins with an N-terminal signal peptide and no TMHs were assigned to the lysosome.
ConA-enriched glycoproteins were treated with PNGaseF, and peptides in which the predicted Asn was converted to Asp were identified by a shift in mass of +0.984 Da (Table 1). These modified peptides (32 total), which represent occupied N-glycan sites, are absent from E. histolytica proteins that have not been treated with PNGaseF. Glycoproteins (26 total) with occupied N-glycan sites include numerous well-characterized virulence factors and/or vaccine candidates (heavy and intermediate subunits of the Gal/GalNAc lectin, serine and cysteine peptidases, and a receptor kinase) (Table 1) (8, 10, 32). Other glycoproteins with occupied N-glycan sites include 17 unique proteins that are secreted, membrane associated, or GPI anchored. Because some of these unique E. histolytica proteins with occupied N-glycan sites are both short and abundant (e.g., EHI_077530 is 206 amino acids long with 56% peptide coverage and EHI_161040 is 180 amino acids long with 42% peptide coverage), it is likely that they would make good vaccine candidates. A list of unique E. histolytica glycoproteins is shown in the FASTA file in the supplemental material.
While actin-mediated capping of amebic proteins has been described (3, 16, 18, 19, 50), this is the first demonstration, to our knowledge, of replenishment of surface antigens from large intracellular pools. This process is shown in the model in Fig. 1I, where the precap surface antigens are shown in green, and the precap internal pool of antigens is shown in red. During capping of the green antigens by cyanovirin-N or antibodies to the Gal/GalNAc lectin, the red antigens move from internal pools to cover the parasite surface.
Because replacement occurs so quickly, the effects of cyanovirin-N on amebic phagocytosis in vitro (shown here) and of antibodies to the Gal/GalNAc lectin (32, 39, 43) and to PPG (6, 33, 35, 40, 48) on amebic virulence in vivo are likely not simply based upon clearing the relevant proteins from the parasite surface. Instead, the effects of cyanovirin-N and of antibodies to the Gal/GalNAc lectin or to PPG are likely also mediated by perturbation of the cytoskeleton during capping (3, 16, 18, 19, 50) and/or by signals transduced by various receptors (Gal/GalNAc lectin and/or receptor kinases) (8). Conversely, it does not appear that E. histolytica trophozoites escape anti-Gal/GalNAc lectin or anti-PPG antibodies by capping and removing antigens from their surfaces as both the Gal/GalNAc and PPG are rapidly replenished from large intracellular pools.
While there was no surprise that the E. histolytica Gal/GalNAc lectin has occupied N-glycan sites (32), it was not possible in advance to predict that glycoproteins recognized by anti-PPG antibodies are also capped by cyanovirin-N (6, 33, 35, 40, 48). The latter result suggests that some E. histolytica glycoproteins contain both N-glycans and O-phosphodiester-linked glycans.
ConA affinity chromatography enabled the identification of >100 E. histolytica secreted and membrane proteins by mass spectrometry. The gel-free mass spectrometric methods used here are easier than cutting proteins from two-dimensional protein gels and result in relatively fewer cytosolic proteins identified than methods in which membranes or lysosomes are isolated (12, 26, 37, 51, 52, 55, 56). However, these other mass spectrometric studies reveal differences between virulent and avirulent strains of Entamoeba and demonstrate accessory proteins (e.g., Rabs) involved in vesicle sorting, endocytosis, and protein secretion.
Gal/GalNAc lectins are among the most abundant E. histolytica glycoproteins identified here, consistent with their prior identification by monoclonal antibodies and their importance in amebic pathogenesis (32, 39, 43). Dozens of unique and abundant E. histolytica glycoproteins identified here by mass spectrometry include new vaccine candidates (type 1 membrane proteins and GPI-anchored proteins) and/or new proteins involved in pathogenesis (secreted proteins). Of course, vaccine candidates and proteins involved in pathogenesis may be overlapping (e.g., the Gal/GalNAc lectin) (39). Recombinant versions of these unique E. histolytica glycoproteins, many of which are relatively small and not too Cys rich, might be used to vaccinate animal models and so add to the relatively short list of amebic vaccine candidates (Gal/GalNAc lectins, serine-rich E. histolytica protein [SREHP], and the 29-kDa protein) (9, 32, 39, 46, 47). Knockdown or knockout methods might be used to test the roles of these proteins in amebic virulence (27, 34).
Unprocessed Man5GlcNAc2, by far the most abundant E. histolytica N-glycan, is recognized by the antiretroviral lectin cyanovirin-N that has been overexpressed in Lactobacillus (1, 25, 28, 31, 41, 46). Cyanovirin-N-expressing lactobacilli (“yogurt-plus”) might be introduced into the gastrointestinal tract, where the bacteria may have an antiamebic effect. Other bacterial lectins that target high-mannose N-glycans of HIV (e.g., griffithsin and banana lectin [BanLec]) may have even greater efficacy than cyanovirin-N versus E. histolytica (38, 49). Conversely, it may be possible to vaccinate against amebic infection with high-mannose N-glycans present on Saccharomyces mutants (14, 30).
This work was supported by NIH grants AI44070 (to J.S.), GM31318 (to P.W.R.), and RR10888 (to C.E.C.). Support for D.M.R. was provided by the Training Program in Host Pathogen Interactions (T32 AI052070).
We thank Richard Cook of MIT for some mass spectrometry data.
†Supplemental material for this article may be found at http://ec.asm.org/.
Published ahead of print on 17 September 2010.