INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Entamoeba histolytica is a protozoan parasite that causes dysentery and liver abscess in developing countries which cannot prevent its fecal-oral spread (56). Amebae have three virulence-associated properties, which are of interest to cell biologists (43). First, amebae survive anaerobic conditions in the lumen of the colon and tissue abscesses by means of fermentation enzymes that resemble those of anaerobic bacteria (31, 48, 57, 60). Indeed, amebae lack enzymes of oxidative phosphorylation and have an atrophic mitochondrion-derived organelle, which resembles the petite mitochondria of yeast cells grown under anaerobic conditions (11, 20, 41). Second, amebae have a vesicle/vacuole-filled cytosol like that of macrophages (63). Amebae phagocytose bacteria in the colonic lumen and epithelial cells and erythrocytes (RBC) when parasites invade tissues and cause dysentery (46, 56). Third, amebae form a chitin-walled cyst, which is the infectious form of the parasite, because cysts are resistant to stomach acids (5, 15).

Phagocytosis is the best studied of the amebic virulence mechanisms. Parasites attach to bacteria, epithelial cells, and RBC via amebic lectins, which recognize Gal or GalNAc sugars on the surface of target cells (42, 45). Amebae kill bacteria or lyse host cells within their phagolysosomes via oxygen-independent mechanisms including lysozyme, cysteine proteases, and amebapores (also known as pore-forming peptides) (8, 34, 64). Lysosomal proteins, which have been identified in supernatants of cultured trophozoites, include cysteine proteases, acid phosphatase, collagenases, glycosidases, and esterases but not pore-forming peptides (1, 35, 68). One cysteine proteinase (CP-5) is present on the surface E. histolytica trophozoites but is absent from the surface of avirulent E. dispar (26). Amebic phagocytosis is disrupted by wortmannin and by overexpression of hyperactive amebic p21^rac, a ras-family protein involved in the site selection of actin polymerization (18, 23, 36). In animal models, phagocytosis mutants of amebae, selected by consumption of bromodeoxyuridine-loaded bacteria, are less virulent than wild-type parasites (58).

Secretion by amebae is much less well understood than phagocytosis. Amebae have on their surface vaccine candidates, including the Ser-rich E. histolytica protein (SREHP) and Gal or GalNAc lectin, which presumably get there when secretory vesicles fuse with the plasma membrane (42, 45, 51, 52, 61, 67, 73). Small and large subunits of the Gal or GalNAc lectin have signal sequences that are cleaved at sites predicted by the "3,1" rule of von Heijne (Table 1) (50). Signal sequences as well as propeptide sequences are cleaved from pore-forming peptides and cysteine proteases (Table 1) (8, 34). An E. histolytica gene has been cloned that encodes a 54-kDa peptide of the signal recognition particle, a ribonuclear protein that binds N-terminal signal sequences on secreted proteins (55). An amebic gene has also been cloned that encodes an endoplasmic reticulum (ER) retention receptor (ERD2), a cis-Golgi-associated transmembrane protein (62). ERD2 binds C-terminal KDEL peptides on proteins such as the 70-kDa heat shock protein BiP, which is a chaperonin in the lumen of the ER (10, 22). We recently cloned the E. histolytica bip gene and confirmed that the predicted BiP has a C-terminal KDEL peptide (16a). Although amebae lack a Golgi with tight lamellae, a putative Golgi was identified by confocal microscopy with NBD-ceramide and by transmission electron microscopy with thiamine-pyrophosphatase (44).

We recently used molecular cloning methods to identify amebic chitinases, which are secretory proteins expressed by encysting parasites (14). Each amebic chitinase contains a series of acidic and hydrophilic repeats between an N-terminal signal sequence and a C-half catalytic domain (50). As E. histolytica trophozoites are difficult to encyst in axenic culture, cyst formation has for the most part been studied by using the reptilian pathogen E. invadens, which converts to chitin-walled cysts within 2 days when deprived of glucose (15). During E. invadens cyst formation, chitin synthase and chitinase are both expressed, and cyst formation is inhibited by the chitin synthase inhibitors polyoxin D and Nikkomycin and by the chitinase inhibitor allosamidin (6, 13, 72).

The goal of the present studies was to visualize structures involved in secretion (secretory vesicles, ER, and Golgi) in amebic trophozoites and encysting organisms. These studies were patterned after similar studies of Giardia lamblia, a second intestinal protozoan parasite, which forms cysts with an acid-resistant wall (19, 39). Secretory vesicles and Golgi apparatus of G. lamblia trophozoites (motile forms) are very difficult to visualize. In contrast, secretory vesicles, which contain cyst wall proteins, are so prominent in encysting giardia that they were given a special name, encystation-specific vesicles or ESV (47). Similarly, encysting G. lamblia show increased expression of Golgi proteins and the ER-associated protein BiP (22, 37, 38). Here amebic secretory vesicles, putative ER, and putative Golgi, which were visualized by the various probes presented in Table 2, were prominent in trophozoites, as well as in encysting parasites (secretory vesicles and Golgi). Further, targeting of chitinase in transfected parasites was modified by removal of an N-terminal signal sequence or addition of a C-terminal, ER retention peptide (KDEL).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Parasites used and conditions for pinocytosis, phagocytosis, and encystation. The HM-1 strain of E. histolytica, from which chitinase, ariel, arf, and bip genes were identified, was used to study secretion in amebic trophozoites (14, 40). The IP-1 strain of E. invadens, a generous gift of Dan Eichinger of New York University, was used to study secretion during encystation (15). E. histolytica and E. invadens were grown at 37 and 25°C, respectively, in the same axenic TYI-S33 medium.

Cloning of E. histolytica and E. invadens arf genes. Segments of the E. invadens and E. histolytica arf genes were isolated from DNA of E. invadens IP-1 strain and E. histolytica HM-1:IMSS strain genomic DNAs by using the PCR and degenerate primers to conserved peptides in ADP-ribosylating factor (ARF) of other eukaryotes (3, 21, 28, 32). A degenerate sense primer [AGAAT(CT)(CT)T(AT) ATGGT(AT)GG] was to RILMVG, while an antisense primer [GG(AT)A(AG) (AG)TCTTGTTT(AG)TT(AT)(AG)CG] was to F(AV)NKQDL. PCR products were cloned in TA-vector and sequenced by dideoxy methods. As expected, the E. histolytica arf PCR product was more AT-rich in the third position (89%) than the E. invadens arf PCR product (60%) (8, 14, 31, 36, 40, 53, 55, 60, 62). The predicted E. histolytica ARF was compared with proteins in the GenBank database by using BLAST2 (2). An alignment of ARF and ARF-like sequences was made by using PIMA-2, and a neighbor-joining (NJ) tree was constructed by using Treecon (66, 71). The reproducibility of the resulting topology was determined by replicating the analysis on 100 bootstrap resamplings of each alignment.

RT-PCR of arf, erd2, and chitinase mRNAs of E. histolytica and E. invadens parasites. Total RNA was extracted from E. histolytica trophozoites and from encysting E. invadens by lysing organisms in saturated guanidinium and centrifuging the lysate through a cesium chloride cushion. First-strand cDNA synthesis was performed with E. histolytica RNA by using reverse transcriptase (RT) and antisense primers specific for arf (ATTTCTCATCTCATCTTC), erd2 (TCAATATGGCAAAACGAATTT), or chitinase (TTAACATTTCTCAATTAG) genes (14, 62). PCRs were then performed with these antisense primers and sense primers specific for E. histolytica arf (GGACTTGATGCTGCCGG), erd2 (ATGGTGTTTAATCTTTTTAGA), and chitinase (ATGTCACTACTGGCATTAAT) genes, respectively. Controls included PCR without prior RT. RT-PCR was performed in a similar manner with RNAs from encysting E. invadens by using sense (GTGACATCGTCCCAAGAA) and antisense (GGGCACAGTACAGACAAA) primers to E. invadens chitinase genes, and the same sense and antisense primers were used to amplify E. histolytica arf cDNAs (14).

Sources of antibodies against chitinases, Ariel antigens, ADH1, ARF, and -COP. E. histolytica and E. invadens chitinases each have a series of degenerative repeats, which are hydrophilic and acidic, between signal peptides and catalytic domains (14). Because the chitinase repeats differ from each other, two sets of rabbit polyclonal, monospecific anti-chitinase antibodies were prepared (Table 2) (24). The first antibodies were against a multiantigenic peptide (MAP) containing E. histolytica chitinase repeats (HESSEIKPDSSESKHESSEK). The second antibodies were against a MAP containing E. invadens chitinase repeats (PKPEESSEQPKPEESSEEKK). Rabbit antibodies were purified on affinity columns containing each chitinase MAP and checked against Western blots of amebic proteins.

Expression of chitinase genes under an actin promoter in transfected E. histolytica trophozoites. E. histolytica trophozoites, which do not encyst in axenic culture, do not express their chitinase gene (14). To identify secretory vesicles in E. histolytica trophozoites, we expressed the E. histolytica chitinase gene in transfected amebae under an E. histolytica actin 1 gene promoter (Table 2) (17, 18). Briefly, the coding region of the E. histolytica chitinase gene was isolated from genomic DNA by PCR. The sense primer (S1 = GCGGTACCATGTCACTACTGGCATTAAT) contained a KpnI site (italics) and encoded the first six amino acids at the amino terminus of the parasite's chitinase (MSLLAL [underlined]). The antisense primer (A1 = GCGGATCCTTAACATTTCTCAATTAG) contained a BamHI site (italics) and encoded five amino acids at the C terminus of the organism's chitinase (LIEKC [underlined]). The chitinase gene was cloned into the pJST4 amebic transformation vector and electroporated into E. histolytica HM-1 strain amebae (17). Parasites were step selected in 10 to 100 µg of G418 per ml (4 to 6 weeks), when the expression of chitinase mRNA was checked by RT-PCR. Overexpression of intact chitinase or modified chitinases (described below) had no effect on amebic viability, growth rate, pinocytosis, or phagocytosis.

Confocal microscopy. To immunolocalize secretory proteins on the surface of E. histolytica and E. invadens, we fixed parasites with 2% paraformaldehyde for 10 min at 4°C. To visualize secretory vesicles or Golgi-associated proteins, we permeabilized amebae by incubation with 0.1% Triton X-100 for 5 min at room temperature. Similar results were obtained when parasites were permeabilized with 0.1% saponin. Amebae were immunostained for 60 min with polyclonal rabbit antibodies to chitinases, Ariel, or -COP, diluted 1:100 or 1:200 in PBS with 1 mg of bovine serum albumin per ml (24). Organisms were washed four times and immunodecorated for 60 min with a Texas red-conjugated goat anti-rabbit antiserum. Controls included parasites stained with preimmune rabbit serum. Similar methods were performed with mouse monoclonal anti-ARF and anti-myc antibodies, each diluted 1:200. Lysosomes were visualized by incubating living parasites in Arg-Arg-4-methoxy-2-naphthylamide, a substrate of cysteine proteinases that fluoresces when it is cleaved (64). Fluorescently labeled parasites were observed with a Leica NT-TCS confocal microscope fitted with argon and krypton lasers. Images of amebae were recorded in 512 image size format with a ×40 or ×100 Planapo objective. The number of vesicles per ameba labeled with anti-ARF, anti--COP, or anti-chitinase antibodies in parasites transfected with the chitinase-KDEL construct was determined by making serial optical sections through the parasites with the confocal microscope. To illustrate some of these structures, composite figures were made that combined multiple optical sections. The number of secretory vesicles, which were identified with antibodies to intact chitinase or Ariel, was too many to count accurately.

Nucleotide sequence accession numbers. Nucleotide and derived amino acid sequences of E. histolytica and E. invadens arf gene segments have been submitted to GenBank under accession numbers AF082517 and AF082518.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Chitinase is an appropriate reporter protein for studying secretion in transfected E. histolytica because it is not expressed by trophozoites. GFP, which is a popular epitope tag used to localize proteins in living eukaryotic cells, does not work in transfected amebae (unpublished observations) (54). This is because GFP fluorescence is dependent upon the presence of free oxygen, which is not present in amebic cultures. Amebic chitinase was chosen here to study secretion in transfected trophozoites because chitinase contains a series of antigenic repeats and is normally secreted by encysting parasites (14). Chitinase mRNAs, detected by RT-PCR, were present in extracts of encysting E. invadens parasites but were absent in extracts of either E. histolytica or E. invadens trophozoites (Fig. 1). In contrast, mRNAs of the Golgi-associated protein ARF were present in extracts of trophozoites of E. histolytica and E. invadens and in extracts of encysting E. invadens (Fig. 1 and see further discussion below) (12). ERD2 mRNAs were also present in extracts of E. histolytica trophozoites (Fig. 1) (62).

View larger version (74K): [in this window] [in a new window]   FIG. 1.   RT-PCR of arf, erd2, and chitinase (chit) mRNAs in nontransfected E. histolytica trophozoites and E. invadens encysting for 0 to 48 h. Ethidium staining of PCR products, separated on agarose gels, was electronically reversed for ease of reproduction.

Chitinase-associated secretory vesicles in encysting E. invadens parasites are numerous and small. Antibodies to antigenic repeats of the E. invadens chitinase were used to demonstrate hundreds of mostly small secretory vesicles in encysting E. invadens parasites (Fig. 2 and Table 2) (14, 61). In most cells, these chitinase-associated secretory vesicles were so abundant that their exact number could not be determined. Chitinase was also present in patches on the surface of encysting parasites. Chitinase was absent from trophozoites, which lack detectable messages for this gene (as above). The chitinase-associated vesicles of encysting amebae are similar in their appearance to ESV previously described in encysting G. lamblia (19, 39, 47).

View larger version (57K): [in this window] [in a new window]   FIG. 2.   Confocal micrographs of chitinase-associated secretory vesicles of encysting E. invadens identified with anti-chitinase antibodies. (A) Chitinase-associated secretory vesicles were absent from trophozoites (data not shown) but were widely distributed in amebae encysting for 24 h. (B) Anti-chitinase antibodies in nonpermeabilized amebae mark secreted products on the parasite surface. Bars, 5 µm.

Secretory vesicles of E. histolytica trophozoites, marked by anti-chitinase antibodies in transfected parasites, are numerous and small like vesicles containing Ariel antigens or cysteine proteinases. Axenic amebae, which were transfected with an unmodified chitinase gene under an actin gene promoter, had numerous small chitinase-associated vesicles (Fig. 3 and Table 2). Again, these secretory vesicles in most cells were too abundant to make an accurate count. Double-labeling studies, which were performed with Texas red-labeled chitinase and FITC-dextran or GFP-labeled E. coli, showed that chitinase-associated vesicles did not fuse with pinocytotic or phagocytotic vacuoles. Chitinase-associated secretory vesicles were about the same size as lysosomes (identified with Arg-Arg-4-methoxy-2-naphthylamide, a fluorescent substrate of the cysteine proteases) (64). Because the signal from the cysteine proteinase was present in both FITC and Texas red channels, it was not possible to colocalize the lysosomes and the chitinase-containing vesicles. Chitinase was absent from the surface of transfected parasites (data not shown), suggesting that the surface of trophozoites lacked chitin-binding activity present on the surface of encysting parasites.

View larger version (23K): [in this window] [in a new window]   FIG. 3.   Confocal micrographs of E. histolytica trophozoites transfected with their own chitinase gene under an actin promoter. Anti-chitinase antibodies (red) demonstrated secretory vesicles in transfected, fixed, and permeabilized E. histolytica trophozoites. These vesicles were independent of pinocytotic vacuoles marked with FITC-dextran (green in panel A) or phagocytotic vacuoles marked with GFP-labeled bacteria (green in panel B). Anti-chitinase antibodies failed to stain the surface of nonpermeabilized, transfected parasites, thus demonstrating that the enzyme is secreted and shed (data not shown). Chitinase-associated vesicles were about the same size and abundance as lysosomes, as marked by the fluorescent substrate (Arg-Arg-4-methoxy-2-naphthylamide) of cysteine proteases (yellow in panel C). Bars, 5 µm.

View larger version (32K): [in this window] [in a new window]   FIG. 4.   Confocal micrographs of nontransfected E. histolytica trophozoites which were fixed, permeabilized, and stained with antibodies to Ariel, a homologue of the vaccine candidate SREHP. While most Ariel-associated secretory vesicles (red) were independent of pinocytosed FITC-dextran (green in panel A), some localized with phagocytotic vacuoles marked with GFP-labeled bacteria (green in panel B). Anti-Ariel antibodies outlined the surface of nonpermeabilized trophozoites (red in panel C). Bars, 5 µm.

Targeting of chitinase in transfected E. histolytica parasites is changed when an N-terminal signal sequence is removed or a C-terminal ER retention signal is added. Amebic chitinase and other secretory or plasma membrane proteins have at their N-termini signal sequences, which fit the 3,1 rule of von Heijne (Table 1) (50). To test the necessity of the 12-amino-acid signal sequence of the E. histolytica chitinase, we transfected parasites with chitinase genes that encoded an intact chitinase and a truncated chitinase lacking the signal sequence (Fig. 5 and Table 2). Anti-chitinase antibodies demonstrated a vesicle-bright, cytosol-dark distribution of the intact chitinase that was similar to that of Ariel. In contrast, anti-chitinase antibodies demonstrated a reverse image of the truncated chitinase, in which vesicles were dark and the cytosol was bright. This cytosolic distribution, in which vesicles appear dark, like the holes in Swiss cheese, was also demonstrated with antibodies to ADH1, an abundant amebic fermentation enzyme (31, 41).

View larger version (103K): [in this window] [in a new window]   FIG. 5.   Confocal micrographs of amebic secretory vesicles, cytosol, and putative ER. (A) Intact chitinase was targeted in transfected E. histolytica trophozoites to secretory vesicles, which were small and numerous. (B) Similar vesicles were visualized with anti-Ariel antibodies in nontransfected parasites. (C) A truncated chitinase lacking an N-terminal signal sequence went to the cytosol, which has dark vesicles against a bright background. (D) This is the same appearance as that of nontransfected amebae stained with antibodies to ADH1. (E) A modified chitinase with a C-terminal KDEL peptide was retained in a putative ER, which was composed of many fewer vesicles than the secretory vesicles marked by unmodified chitinase or Ariel. (F) A similar pattern of staining was observed with myc-labeled BiP, which is retained by its native C-terminal KDEL in the putative ER. Bars, 5 µm.

E. histolytica and E. invadens arf genes encode a conserved Golgi-associated coatomer protein (COP) called ARF. To better understand the role of Golgi apparatus in secretion by amebic trophozoites and encysting parasites, we used the PCR and degenerate primers to conserve peptides in other ARF to clone segments of E. invadens and E. histolytica arf genes (Fig. 6) (3, 21, 28, 32). The predicted 113-amino-acid open reading frame (ORF) of the amebic arf genes, which encode about two-thirds of the expected 20-kDa ARF proteins, showed 96% positional identity with each other and 76 to 88% positional identities with other eukaryotic ARFs. Perfectly conserved in the amebic ARFs were guanine nucleotide-binding domains, including the pyrophosphate-binding loop (GLDAAGKT), switch region (DVGG), and guanine recognition motif (NKQD) (3, 21). In NJ trees, amebic ARF was easily distinguished from ARF-like proteins of parasites or humans (Fig. 7) (71).

View larger version (32K): [in this window] [in a new window]   FIG. 6.   ARF alignment. Predicted ORF of a segment of the E. histolytica (Eh) arf gene aligned with ARF segments of E. invadens (Ei), G. lamblia (Gl; GenBank accession number S29008), Plasmodium falciparum (Pf, U57370), Dictyostelium discoideum (Dd; AJ000063), Saccharomyces cerevisiae (Sc; M35158), and Bos taurus (Bt; A45422). Dashes indicate identities with the E. histolytica ARF, while periods indicate gaps. Vertical boxes indicate locations of conserved pyrophosphate-binding loop (GLDAAGKT), switch region (DVGG), and guanine-recognition motif (NKQD) (3, 34).

View larger version (9K): [in this window] [in a new window]   FIG. 7.   NJ tree of ARF shown in Fig. 6 as well as ARF of Xenopus laevis (Xl; GenBank accession number U31350), Drosophila melanogaster [Dm(A); S62079], Cryptococcus neoformans (Cn; L25115), and Schizosaccharomyces pombe (Sp; L09551), and ARF-like proteins of D. melanogaster [Dm(AL); A40438], Homo sapiens (Hs; L28997), P. falciparum [Pf(AL); U57370], and Leishmania tarentolae (Lt; X97072). Branch lengths are proportional to differences between adjacent sequences, while numbers at each node indicate the extent of bootstrap support (of 100 replicates). Unmarked nodes occurred in fewer than 50 replicates.

The amebic Golgi and encystation are disrupted by Brefeldin A and okadoic acid. As discussed above, RT-PCR showed that mRNAs that encode the Golgi-associated protein ARF and the ER-retention receptor (ERD2) are expressed by E. histolytica trophozoites (Fig. 1) (12, 62). ARF was also expressed by E. invadens trophozoites and encysting E. invadens (Fig. 1). Anti-ARF antibodies identified a stage-independent, putative Golgi apparatus, which was composed of a few large perinuclear vesicles (5 to 20 per cell) (Fig. 8). Similar Golgi-associated vesicles were identified with antibodies to -COP (Fig. 9). This putative Golgi apparatus was disrupted into tiny vesicles Brefeldin A, which targets the GTPase of ARF (29). Brefeldin A also reduced the number of E. invadens parasites that encyst by 60%. Okadoic acid, which targets protein phosphatases and disrupts trans-Golgi of eukaryotic cells, disrupted the amebic Golgi and eliminated encystation completely (25). Wortmannin, which targets phosphoinositide 3-kinases and eliminates amebic pinocytosis and phagocytosis, also eliminated encystation (18). Previously, an amebic Golgi was seen with NBD-ceramide (as seen by fluorescence microscopy) and thiamine-pyrophosphatase (as seen by electron microscopy) (44).

View larger version (64K): [in this window] [in a new window]   FIG. 8.   Confocal micrographs of E. invadens parasites stained with heterologous anti-ARF antibodies. (A) ARF was present in large vesicles (presumed Golgi apparatus) surrounding the nucleus of trophozoites. (B) The ARF-stained vesicles of trophozoites were disrupted by Brefeldin A. (C) The Golgi apparatus of encysting parasites was similar when stained with anti-ARF. Micrographs shown in panels A and B represent a single section each, while the micrograph in panel C is a composite of multiple sections. Bars, 5 µm.

View larger version (34K): [in this window] [in a new window]   FIG. 9.   Confocal micrographs of E. histolytica and E. invadens parasites stained with antibodies to Golgi-associated coatomer protein -COP. Vesicles that contain -COP were relatively few and large in trophozoites of E. histolytica (A), trophozoites of E. invadens (B), or encysting E. invadens (C). The micrograph shown in panel A represents a single section, while the micrographs in panels B and C are each composed of multiple sections. Bars, 5 µm.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

These are the first arf genes identified in amebae, and this is the first use of transfection methods to localize amebic proteins and the first experimental demonstration of N-terminal signal and C-terminal ER retention sequences. Roughly drawn, the amebic cytosol is composed of small vesicles, which include those of the putative ER, secretory vesicles, lysosomes, pinocytotic vacuoles, and large vesicles, which include those of the putative Golgi and phagocytotic vacuoles.

Conservation of secretory apparatus and protein-targeting sequences in amebae. Perhaps because amebae have such prominent pinosomes, phagosomes, and lysosomes, secretory structures of the parasite were not obvious in the absence of the specific probes used here (Table 2) (18, 43, 63). Chitinase expression in transfected parasites was used to show that amebic N-terminal signal sequences and C-terminal ER retention signals appear to follow the same rules as those established for higher eukaryotes (50, 61, 62). Indeed the 3,1 rule appears to hold for all signal sequences of amebic secretory and plasma membrane proteins which have been identified to date (Table 1). The amebic Golgi and encystation were disrupted by Brefeldin A and okadoic acid, which disrupt the Golgi of higher eukaryotes (25). Disruption of the G. lamblia Golgi, which is encystation specific, also inhibits cyst formation (37).

Unique properties of the amebic secretory pathway. Amebic secretion is different from that of higher eukaryotes in at least four ways. First, the putative amebic Golgi apparatus does not form tightly packed lamellae but instead forms a few large vesicles, which are adjacent to the nucleus (12, 44, 79). This is similar to the Golgi of G. lamblia, whereas Tritrichomonas foetus, another lumenal parasite, has a Golgi with tight lamellae (19, 37). Second, an amebic protein disulfide isomerase (PDI) lacks a C-terminal KDEL peptide, which usually retains other PDI in the ER (27). Third, a sharp distinction between amebic lysosomes and secretory vesicles has not yet been made (63). Fourth, signals that direct amebic proteins to the lysosomes cannot be the same as those of higher eukaryotes. Amebic lysosomal proteins lack signals for Asn-linked glycosylation, which is necessary for the addition of mannose-6-phosphates that might bind to lysosomal receptors (30, 61, 69). We are presently performing experiments to distinguish secretory vesicles from lysosomes and to identify signals that target amebic proteins to lysosomes. We plan to use immunoelectron microscopy to visualize amebic secretory vesicles, putative ER, and putative Golgi.

Implications of these findings for our understanding of amebic virulence. Four implications of these data may be important in understanding amebic virulence. First, the secretory apparatus, which targets proteins to the plasma membrane of amebic trophozoites, likely releases numerous proteins into the medium, even though to date only lysosomal enzymes have been identified in amebic supernatants (1, 35, 68). Second, the 3,1 algorithm might be used to identify secretory proteins in EST databases of amebic trophozoites or encysting parasites (50). Third, ARF in the cytosol of amebae may activate cholera toxin in individuals who are coinfected with Vibrio cholerae and amebae (49). Other targets for bacterial toxins previously identified in amebae include elongation factor 2 (diphtheria and pseudomonas toxins) and Ras-superfamily proteins (clostridial toxins) (36, 53). Fourth, stage-independent Golgi and ER in amebae suggest that these parasites may be susceptible to bacterial toxins, which enter the cytosol through the Golgi or ER (49).

Changing image of amebae and other lumenal parasites: from primitive to divergently evolved. Lumenal parasites (E. histolytica, G. lamblia, and Trichomonas vaginalis) branched early from the main eukaryotic tree, lack mitochondria and enzymes of oxidative phosphorylation, and have fermentation enzymes like those of bacteria (31, 33, 48, 57). These observations led to the idea that these microaerophilic parasites are "primitive" or living fossils from a time before eukaryotic cells were exposed to oxygen and captured the mitochondrial endosymbiont (7, 20, 57). The experiments described here show the similarity of the amebic secretory pathway to those of other eukaryotic cells, even if its appearance is somewhat different. Further, amebae, giardias, and trichomonads each have a gene encoding a homologue of the mitochondrial 60-kDa heat shock protein (Hsp60) (9, 11, 59). The presence of Hsp60 genes in "amitochondriate" parasites strongly suggests that all extant eukaryotes share a common ancestor that had a mitochondrion (20). In trichomonads, the mitochondrion-derived organelle was converted over time into a fermentation factory called the hydrogenosome (9, 48). The function of an atrophic mitochondrion-derived organelle of amebae is not known, while a mitochondrion-derived organelle has not yet been identified in giardias (41, 59).