Role of carbohydrate-mediated adherence in cytopathogenic mechanisms of Acanthamoeba.

Acanthamoeba keratitis is a vision-threatening corneal infection. The mannose-binding protein of Acanthamoeba is thought to mediate adhesion of parasites to host cells. We characterized the amoeba lectin with respect to its carbohydrate binding properties and the role in amoeba-induced cytopathic effect (CPE). Sugar inhibition assays revealed that the amoeba lectin has the highest affinity for alpha-Man and Man(alpha1-3)Man units. In vitro cytopathic assays indicated that mannose-based saccharides which inhibit amoeba adhesion to corneal epithelial cells were also potent inhibitors of amoeba-induced CPE. Another major finding was that N-acetyl-D-glucosamine (GlcNAc) which does not inhibit adhesion of amoeba to host cells is also an inhibitor of amoeba-induced CPE. The Acanthamoebae are thought to produce CPE by secreting cytotoxic proteinases. By zymography, one metalloproteinase and three serine proteinases were detected in the conditioned media obtained after incubating amoebae with the host cells. The addition of free alpha-Man and GlcNAc to the co-culture media inhibited the secretion of the metalloproteinase and serine proteinases, respectively. In summary, we have shown that the lectin-mediated adhesion of the Acanthamoeba to host cells is a prerequisite for the amoeba-induced cytolysis of target cells and have implicated a contact-dependent metalloproteinase in the cytopathogenic mechanisms of Acanthamoeba.

Acanthamoeba keratitis is a painful, vision-threatening corneal infection. It is caused by parasites of the genus Acanthamoeba (1,2). The mechanism by which Acanthamoebae produce keratitis has not been fully elucidated. Contact lens wear is thought to be the leading risk factor. However, the disease also occurs, if less frequently, in non-contact lens wearers. It is believed that minor corneal surface injury caused by contact lens wear or other noxious agents and exposure to contaminated solutions, including lens care products and tap water, are two major factors in the pathogenesis of the keratitis. The adhesion of the parasite to the host cells is thought to be a critical first step in the pathogenesis of infection (3)(4)(5). Studies aimed at the characterization of the molecular mechanisms by which amoebae adhere to corneal epithelium have shown that: (i) the adhesion of Acanthamoebae to corneal buttons in organ culture and to monolayer cultures of corneal epithelium can be inhibited by methyl-␣-D-mannopyranoside (␣-Man) but not by several other monosaccharides (6,7), and (ii) a mannose-binding protein is present on the surface membranes of Acanthamoeba (8). In vitro, Acanthamoeba parasites have been shown to produce a cytopathic effect (CPE) 1 on a variety of cell types (9 -11) including rabbit and human corneal stromal and epithelial cells (5,12,13). The parasite's ability to cause cytolysis and necrosis of host tissues is clearly a key component of Acanthamoeba infection but our understanding of how this is achieved is still rudimentary. In the present study, we show that the Acanthamoeba lectin binds ␣-Man and ␣1-3-D-mannobiose with highest affinity and that the lectinmediated adhesion of the amoeba to host cells is a prerequisite for the amoeba-induced cytolysis of target cells.

Characterization of the Carbohydrate-binding Specificity of Acanthamoeba Lectin
An Acanthamoeba strain derived from an infected human cornea (MEEI 0184; Acanthamoeba castellanii based on morphological characteristics) was used throughout this study. The parasites were axenically cultured in a proteose peptone/yeast extract/glucose medium (14). To characterize the sugar binding properties of the amoeba lectin a solidphase assay was used (8). Briefly, wells of microtiter plates were coated with bovine serum albumin-␣-D-mannopyranosylphenylisothiocyanate (Man-BSA (Sigma), 15-25 mol of ␣-D-mannopyranoside/mol of albumin, 0.03 g/ml, 50 l/well in 0.1 M sodium carbonate buffer, pH 9.6, 4°C, overnight); nonspecific binding was blocked with 1% BSA in phosphatebuffered saline (1 h, room temperature), a 50-l aliquot of 35 S-labeled Acanthamoebae (5 ϫ 10 6 cells/ml in phosphate-buffered saline; 1-2 cpm/parasite, Ͼ95% trophozoites) was added to each well, and the plates were incubated for 2 h at room temperature. After the cells were rinsed with phosphate-buffered saline to remove unbound Acanthamoebae, 0.1 ml of 2% SDS was added to each well, and the radioactivity in aliquots of solubilized material was determined in a scintillation counter. To characterize the specificity of the amoeba lectin, a range of saccharides were tested for their ability to inhibit the adhesion of amoebae to Man-BSA (see Table I). The 25 saccharides tested were purchased from Sigma, V-Labs Inc. (Covington, LA), and Pfanstiehl Laboratories Inc. (Waukegan, IL), and included ␣1-3-D-mannobiose, ␣1-6-D-mannobiose, ␣1-3,␣1-6-D-mannotriose, and ␣1-3,␣1-6,␣1-3,␣1-6-D-mannopentose (Table I). Serial dilutions of sugars were tested to calculate the concentration required for 50% inhibition of the binding.
To determine whether any other lectin, besides the mannose-binding protein is present on the amoeba surface, the solid-phase assays were performed using microtiter wells coated with a number of neoglycoproteins including N-acetyl-D-glucosamine-BSA(GlcNAc-BSA), N-acetylgalactosamine-BSA, fucose-BSA, and galactose-BSA. These neoglycoproteins contained 15-30 mol of sugar/mol of albumin. * This work was supported by the National Institutes of Health Grants EY07088 and EY09349. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. *

Characterization of the Role of the Mannose-binding Protein in the Cytopathogenic Mechanisms of Acanthamoeba
To characterize the role of carbohydrate-mediated host-parasite interactions in the cytopathogenic mechanisms of Acanthamoeba, in vitro cytopathic assays were performed using cell cultures of corneal epithelium as target cells.
Immortalization of Rabbit Corneal Epithelial Cells-Since primary cultures of corneal epithelium have limited life span and can only be passaged 2 to 3 times without significant alteration in cell morphology, corneal epithelial cells with extended life span were produced. For this, the primary corneal epithelial cell cultures prepared as described earlier (15,16) were infected with SV40 large T antigen cDNA (17). Retroviruses were used to transduce both the 3Ј-phosphotransferase gene conferring G418 antibiotic resistance and the SV40 large T antigen cDNA into the primary cultures. For infection, irradiated producer cells (⌿CRIP-SV40, 1 ϫ 10 6 cells/100-mm dish) were added to confluent primary cultures of corneal epithelium. Two to three days after infection, the transformed cells were selected by adding G418 (500 g/ml) into the culture medium. To promote the growth of the selected cells, initially they were co-cultured with irradiated fibroblast cells. After 10 -12 days, when the cultures reached 30 -40% confluence, G418 as well as fibroblasts were withdrawn. As of this stage, cells were cultured in Dulbecco's modified Eagle's medium/Ham's F-12 (1:1) containing fetal bovine serum (15%), dimethyl sulfoxide (0.5%), gentamicin (40 g/ml), epidermal growth factor (10 ng/ml), insulin (5 g/ml), and cholera toxin (0.1 g/ml). When these cells reached over 50% confluence, they were passaged with a split ratio of 1:2. Upon phase-contrast microscopy, no fibroblastic cells could be detected subsequent to the first passage. The entire cell population was epithelioid. The cultures reacted positively with mAb AE5 which reacts with keratin K3, a marker for corneal epithelial cells. This further established the epithelial origin of the immortalized cells.
Cytopathic Assay-To evaluate the Acanthamoeba-induced CPE on corneal epithelial cells, the parasites (Ͼ95% trophozoites) were rinsed three times in a serum-free medium supplemented with 0.4% BSA (18) and aliquots of the parasite suspension (0.2 ϫ 10 5 to 1 ϫ 10 6 parasites/ ml; 300 l/well for 24-well plates; 1 ml/well for 6-well plates) were added to wells of confluent cultures of epithelium which had been rinsed and preincubated in the serum-free medium for 2 h. The plates were then incubated at 37°C in a CO 2 incubator and were periodically examined under a phase-contrast microscope for the presence of cellfree plaques in the monolayer for up to 28 h. Control wells contained the target cells in the medium without the parasites. To evaluate the effect of saccharides on Acanthamoeba-induced CPE, the assays were performed in the presence of sugars (0.03-100 mM). To estimate the relative inhibitory potency of each sugar, the CPE assays were terminated when, based on observations under a phase-contrast microscope, the culture wells incubated with the amoeba in the absence of saccharides exhibited approximately 50% destruction of the monolayer (10 -14 h). At the end of the incubation period, the cultures were stained with Giemsa (Diff-Quik, Dade Diagnostic Inc., Aguada, PR). Approximate cell density in each well was estimated by scanning the stained plates in a calibrated, computer-assisted Bio-Image Scanner (Mllipore, Ann Arbor, MI). CPE was expressed as percentage of the optical density by the following formula. To determine whether the mechanisms modulating the CPE produced by intact amoeba and the amoeba-conditioned medium (ACM) are distinct, the CPE assays were also performed using the ACM in the presence and absence of saccharides. To produce the ACM, exponentially growing parasites (Ͼ95% trophozoites) in the protease peptone/ yeast medium were rinsed and were then incubated in the serum-free medium supplemented with 0.4% BSA (2 ϫ 10 5 to 1 ϫ 10 6 cells/ml, 37°C, 24 h) in the presence and absence of saccharides (␣-Man and GlcNAc, 50 mM). At the end of the incubation period, the parasites were removed by centrifugation and the supernatants were analyzed for their ability to produce CPE. In some cases saccharides were added to the ACM after it was produced.

Measurement of Acanthamoeba-induced Target Cell Lysis
Cell lysis was quantified by measuring 51 Cr release from prelabeled cells. For radiolabeling, confluent monolayer cultures of corneal epithelium in 24-well plates were incubated in cell culture medium containing Na 51 CrO 4 (460 mCi/mg, 12 Ci/ml, NEN Life Science Products) for 18 -20 h. At the end of the labeling period, the cells were extensively rinsed and were then incubated with Acanthamoeba (1 ϫ 10 6 parasites/ ml) in the presence or absence of saccharides (50 mM ␣-Man or GlcNAc). The cultures were periodically examined under a phase-contrast microscope for CPE. When approximately 50 -70% cell loss occurred in the monolayers, aliquots of cell-free conditioned media from each well were analyzed for specific 51 Cr release. To determine specific 51 Cr release values, counts/min released in control wells incubated in media alone without the parasites were deducted from the counts/min released in the test wells. Percent specific release was calculated using a formula: E/T ϫ 100, where E is counts/min released in test wells minus the counts/min released in control wells, and T is the total counts/min (counts/min cells ϩ counts/min supernatant in control wells). Counts/ min of cells was determined by incubation of the control wells with 0.5% Triton X-100. Conditioned media from each well was also analyzed for lactate dehydrogenase release using an lactate dehydrogenase assay kit purchased from Sigma. In each case, there were four wells/group. The same procedure was used to determine whether ACM possesses cytolytic activity.

Analysis of Proteinases by Zymography
To determine whether the mannose-mediated host-parasite interactions influence the levels of secretory proteinases, conditioned media obtained by incubating the amoebae with primary cultures of corneal epithelium in the presence and absence of saccharides (␣-Man and GlcNAc, 50 mM, 6 -8 h) were analyzed by zymography. The zymography was performed using separating gels containing 10% acrylamide and 2 mg/ml gelatin, essentially as described by Kleiner and Stetler-Stevenson (19). Samples to be analyzed were diluted in the electrophoresis sample buffer (20) without mercaptoethanol and were then applied to the gels. After electrophoresis, the proteinases separated on gels were renatured by incubating the gels in 2.5% Triton X-100 in 50 mM Tris-HCl, pH 7.5, to remove SDS. Following this, the gels were incubated for 18 h in a developing buffer (50 mM Tris-HCl, pH 7.5, containing 10 mM CaCl 2 ) and were then stained with Coomassie Brilliant Blue. Areas of digestion were visualized as nonstaining regions of the gel. In some experiments, samples were pretreated with phenylmethylsulfonyl fluoride (PMSF, 1 mM), an inhibitor of serine proteinases (21), for 1 h prior to electrophoresis, or 1,10-phenanthroline (1 mM), an inhibitor of metalloproteinases (22), was added to the developing buffer.
To determine whether GlcNAc and/or ␣-Man influence the levels of secretory proteinases of Acanthamoeba by contact-independent mechanisms, ACM prepared in the presence and absence of the saccharides (50 mM) were also analyzed by zymography.

The Amoeba Lectin Has the Highest Affinity for ␣-Man and Man(␣1-3)Man Units
In solid-phase assays, binding of amoebae to Man-BSA was dose-dependent ( Fig. 1). Amoebae did not bind to a number of other neoglycoproteins including galactose-BSA, fucose-BSA, N-acetyl-D-galactosamine-BSA, and GlcNAc-BSA (Fig. 1). To characterize the carbohydrate binding properties of the amoeba lectin, sugar inhibition studies were performed. Concentrations of sugars required for 50% inhibition were calculated from inhibition curves and are listed in Table I. Among various monosaccharides tested, ␣-Man was the most potent inhibitor of amoeba binding to Man-BSA. Compared with ␣-Man the inhibitory potency of methyl-␤-mannopyranoside and D-mannose was 8-and 10-fold less, respectively. Epimers of D-mannose, i.e. altrose (C-3) and talose (C-4), were not inhibitory but glucose (C-2) was a weak inhibitor. Compared with D-mannose, the inhibitory potency of D-glucose was 3.5-fold lower. D-Lyxose, which is equivalent to D-mannose with the C-5 substituent missing, was not inhibitory. The reduction of D-mannose to D-mannitol by sodium borohydride completely destroyed its ability to inhibit the amoeba binding to Man-BSA. Neither D-mannosamine nor N-acetyl-D-mannosamine were inhibitory. D-Glucosamine was also not inhibitory. However, GlcNAc and ␣-L-fucose were weak inhibitors. Their inhibitory potencies were 26-and 14-fold lower, respectively, compared with that of ␣-Man. Both mannobioses, the mannotriose, and the mannopentose were also potent inhibitors; the inhibitory potency of ␣1-3-D-mannobiose was approximately three times greater than that of ␣1-6-D-mannobiose and ␣-Man alone. ␣1-3,␣1-6-D-mannotriose and ␣1-3,␣1-6,␣1-3,␣1-6-D-mannopentose were slightly better inhibitors than ␣1-3-D-mannobiose (Table I).

Mannose-mediated Host-Parasite Interactions Play a Role in Acanthamoeba-induced Cytopathogenic Mechanisms
Acanthamoeba trophozoites have been shown to produce a CPE on a variety of cultured mammalian cells (5, 9 -13). To determine whether the carbohydrate-mediated adhesion of Acanthamoeba to corneal epithelium is a prerequisite for the parasite-induced target cell damage, CPE assays were performed in the presence and absence of various saccharides. The pathogenic ocular isolate of Acanthamoeba used in the present study produced extensive CPE on both immortalized (Fig. 2) as well as primary (Fig. 3) rabbit corneal epithelial cells in culture. Within 8 -10 h incubation with the parasites, small cellfree plaques were seen in the monolayers (Fig. 2, 8 h). With the continued incubation with the amoebae, the size of the cell-free areas increased (Fig. 2, 16 h), and eventually the monolayer surrounding the large plaques lifted entirely from the culture dish resulting in almost complete loss of the cell layer (Fig. 2, 24 h). The extent of the monolayer destruction depended on the concentration of amoebae, the nature of cell cultures (primary versus immortalized), the length of the incubation period, and the composition of the media used for the assay. When immortalized cultures were used, an amoeba concentration of 2 ϫ 10 5 cells/ml was required to completely destroy the monolayer within 20 h. All CPE assays were performed in a serum-free medium supplemented with 0.4% BSA. If BSA was omitted from the media, the destruction of the monolayer occurred approximately twice as fast (data not shown).
Upon phase-contrast microscopy, it was evident that when amoebae were incubated with monolayers, they adhered to the cells within minutes. All saccharides which inhibited amoeba binding to Man-BSA also inhibited the amoeba binding to corneal epithelial cells (not shown). In vitro CPE assays, for the most part, revealed a direct correlation between the ability of the sugar to inhibit the Acanthamoeba binding to Man-BSA and to inhibit the amoeba-induced CPE (Table I). ␣-Man, both mannobioses, the mannotriose, and the mannopentose which were potent inhibitors of amoeba binding to Man-BSA were also potent inhibitors of amoeba-induced CPE; saccharides such as mannitol, mannosamine, N-acetyl-D-mannosamine, methyl-␣-and methyl-␤-galactopyranoside which did not inhibit amoeba binding to Man-BSA also did not inhibit amoebainduced CPE. One striking exception was GlcNAc which was a weak inhibitor of the amoeba adhesion to Man-BSA but a potent inhibitor of CPE (see below).
Primary cultures were relatively less susceptible to amoebainduced CPE in that 1 ϫ 10 6 parasites/ml were required to completely destroy the monolayer within 24 h. This parasite concentration is five times higher than that required to produce the same degree of CPE when immortalized cultures were used as target cells. Only a limited number of saccharides were tested for their ability to inhibit the amoeba-induced CPE on primary cultures. At a sugar concentration of 5 mM or less, ␣-Man, both mannobioses, the mannotriose, the mannopentose, and GlcNAc inhibited CPE. Methyl-␣-D-galactopyranoside, methyl-␤-D-galactopyranoside, and ␣-L-fucose were not inhibitory up to 100 mM concentration (Fig. 3).

␣-Man and GlcNAc Modulate Acanthamoeba-induced Cytopathic Effect by Distinct Mechanisms
As described above, GlcNAc, a weak inhibitor of amoebae adhesion to Man-BSA, was a potent inhibitor of amoeba-induced CPE. The inhibitory potency of GlcNAc in the amoeba adhesion assay was 26 times less than that of ␣-Man (Fig. 4, top panel; Table I). In contrast, in the CPE assays, the inhibitory potency of both ␣-Man and GlcNAc was nearly identical (Fig. 4, bottom panel; Table I). Amoebae did not adhere to the cell a Immortalized rabbit corneal epithelial cell cultures were used as target cells. In a limited number of CPE assays with primary cultures, methyl-␣-mannopyranoside, both mannobioses, the mannotriose, the mannopentose, and N-acetyl-D-glucosamine were inhibitory at a sugar concentration of 5 mM or less; ␣-galactose, ␤-galactose, and ␣-L-fucose were not inhibitory up to 100 mM concentration. b ND, not determined. c Note that GlcNAc is a weak inhibitor of amoeba binding to Man-BSA but a potent inhibitor of CPE. monolayer when ␣-Man was present. In addition, ␣-Man almost completely inhibited amoeba-induced CPE. Cell-free plaques were not detected in the presence of ␣-Man (Figs. 3 and Fig. 5, top panel, group M). ␣-Man, however, did not inhibit the lifting of the intact monolayer from the cell substratum upon "prolonged" (24 -28 h) incubation of the cultures with amoebae ( Fig. 5, middle and bottom panels, groups M). In the absence of ␣-Man, nearly complete loss of cell layer occurred around 20 h; in the presence of ␣-Man (50 mM), after an approximate 24-h incubation with the amoebae, the cell layer began to lift from the periphery (Fig. 5, middle panel, group M) and continued to roll inwards until the entire layer was lifted from the dish (Fig.  5, bottom panel, group M). This occurred more than 4 h after the complete loss of the cell layer had occurred in the cultures incubated without the sugar. When GlcNAc (50 mM) was present in the media, amoebae adhered tightly to the monolayer. Despite this, the cell layer remained attached to the culture dish during the entire assay period of 28 h (Fig. 5, middle and bottom panels, groups G). GlcNAc markedly delayed, but did not completely inhibit plaque formation. Upon careful examination of the cultures incubated with amoebae in the presence of GlcNAc for 28 h, small plaques could be seen in the monolayer (Fig. 5, bottom panel, group G). Since CPE assays were performed in the serum-free media, incubations were not continued beyond 28 h. 51 Cr release revealed that Acanthamoeba possesses cytolytic activity (Fig. 6,  EϩA). The extent of cytolysis, however, did not correlate with the extent of CPE; when approximately 50% cell loss had occurred in the CPE assays, percent specific 51 Cr release was about 12.5%. Greater than 70% of amoeba-induced cytolysis was inhibited by ␣-Man (50 mM) (Fig. 6, EϩAϩM). In contrast, GlcNAc was only a weak inhibitor of amoeba-induced cytolysis with inhibition values ranging between 10 and 20% (Fig. 6,  EϩAϩG). Similar observations were made using LDH release assays (data not shown).

␣-Man but Not GlcNAc Is a Potent Inhibitor of Amoebainduced Cytolysis-Cell lysis as measured by
GlcNAc but Not ␣-Man Inhibits ACM-induced CPE-Early studies have shown that cell-free ACM also has the ability to produce CPE. The CPE-inducing ability of ACM is likely to be independent of host-parasite interactions, and related to the cytotoxic secretory factors of Acanthamoeba. To determine the role of secretory factors of the parasites in the amoeba cytopathogenic mechanisms, 51 Cr release and cytopathic assays were also performed using cell-free ACM. 51 Cr release assays revealed that ACM lacks cytolytic activity (Fig. 6, EϩACM). Also, ACM failed to produce cell-free plaques in the monolayer ( Fig. 7, top panel). However, approximately 6 -8 h following the incubation of the monolayer with the ACM, the cell sheet at the edges of the monolayer began to lift and roll inwards (Fig. 7, top  panel, 8 h) and eventually the entire monolayer lifted en bloc from the dish (Fig. 7, top panel, 16 h). These observations were similar to those seen when the monolayers were incubated with trophozoites in the presence of ␣-Man for prolonged periods (Fig. 5). When saccharides (50 mM) were added to ACM after it was produced, neither ␣-Man nor GlcNAc had an inhibitory effect on the ACM-induced CPE (Fig. 7, middle panel, groups ACMϩM and ACMϩG). Also, ACM produced in the presence of ␣-Man exhibited potent CPE (Fig. 7, middle panel, group ACMϩMЈ). In contrast, the ACM produced in the media containing GlcNAc did not produce CPE (Fig. 7, middle panel, group ACMϩGЈ). To further determine whether GlcNAc inhibits the ACM-induced CPE directly by itself or indirectly by influencing the expression of CPE producing factors of amoeba, filtrate and retentate obtained after filtering the ACM in Centricon 3 microconcentrators were tested for their ability to produce CPE. From ACM produced in the absence of sugar, almost all of the CPE inducing ability was recovered in the retentate obtained after Centricon filtration (Fig. 7, bottom  panel). The ACM produced in the presence of GlcNAc could not be caused to acquire the CPE inducing ability after removing GlcNAc by Centricon filtration (Fig. 7, bottom panel).
Mannose-mediated Adhesion of Acanthamoebae to Host Cells Induces Expression and/or Secretion of a Metalloproteinase-To characterize the mechanism by which the mannosemediated host-parasite interactions influence the amoeba-induced CPE, conditioned media obtained after incubating amoebae with the host cells for 6 -8 h in the presence and absence of ␣-Man were analyzed for proteinases by zymography on gelatin gels. Four components (P1, 230-kDa; P2, 97-kDa; P3, 80-kDa; and P4, 55-kDa; average values from four gels) were detected in the samples obtained after incubating amoebae with primary cultures of corneal epithelium in media alone for 6 h (Fig. 8, panel A, lane EϩA). Of these four components, the expression and/or secretion of the component P3 was dependent on the mannose-mediated adhesion of amoeba to host cells; if the adhesion of amoebae to the host cells was prevented by adding ␣-Man in the media, component P3 was not detected (Fig. 8, panel A, lane EϩAϩM). Nor was proteinase P3 detected in media conditioned by Acanthamoeba or epithelial cells alone (Fig. 8, panel A, lanes ACM and EM). Presence of GlcNAc in the co-culture either did not alter or reduced only slightly the levels of P3 (not shown), but had a dramatic effect on the levels of P1, P2, and P4. In the ACM prepared in media alone, mainly components P1 and P4 with trace amounts of component P2 were seen (Fig. 8, panels A and  B, lanes ACM). The presence of ␣-Man in the media used to prepare ACM did not influence the levels of P1, P2, or P4 (not shown, pattern indistinguishable from ACM lanes shown in panels A and B). In contrast, in the ACM prepared in the presence of GlcNAc, P2 was not detected and P1 and P4 were seen in markedly reduced amounts (Fig. 8, panel B, lane ACMϩG). Components P1, P2, and P4 were susceptible to PMSF (Fig. 8, panel C, lanes ACM and EϩA), but were resistant to 1,10-phenanthroline (Fig. 8, panel D, lanes ACM and EϩA). In contrast, component P3 was susceptible to 1,10-phenanthroline and resistant to PMSF (Fig. 8, panels C and D, lanes EϩA). In the media conditioned by epithelial cells alone, two components were seen (E1, 93-kDa; E2, 63-kDa; average values of two gels) (Fig. 8, panel A, lane EM), both of which were resistant to PMSF (Fig. 8, panel C, lane EM) and susceptible to 1,10-phenanthroline (Fig. 8, panel D, lane EM). Proteinases P1, P2, and P4 were susceptible whereas P3, E1, and E2 were resistant to aprotinin, another serine proteinase inhibitor (not shown, data identical to those shown for PMSF).

Proteinase Inhibitors Inhibit both Amoebaand ACM-induced CPE
To further determine the role of proteinases in the cytopathogenic mechanisms of Acanthamoeba, the effect of proteinase inhibitors on amoeba-and ACM-induced CPE on corneal epithelial cells was analyzed. Serine proteinase inhibitors, PMSF (0.5 mM) as well as aprotinin, (5 units/ml), almost completely inhibited both amoeba-and ACM-induced CPE (not shown, results identical to those of CPE assays performed in the presence of GlcNAc, Fig. 5). The metalloproteinase inhibitor 1,10phenanthroline was toxic to corneal epithelial cells, and therefore it was not possible to evaluate its CPE inhibitory potential. DISCUSSION The goals of the present study were to define more precisely the carbohydrate binding properties of the Acanthamoebae mannose-specific lectin and to determine whether carbohydrate-mediated adhesion of amoebae to host cells is a necessary prerequisite for the parasite-induced CPE. The results of the  Table I). Bottom, both ␣-Man and GlcNAc are equally potent inhibitors of Acanthamoeba-induced CPE. Confluent cultures of corneal epithelium in 24-well plates were incubated with Acanthamoebae in the presence and absence of varying concentrations of ␣-Man and GlcNAc for 16 h. At the end of the incubation period, the plates were stained with Giemsa and scanned to estimate approximate cell density. A value of 1.0 was assigned to the cell density of the plates incubated in media alone (Cont.). The values for cultures incubated with amoebae are expressed as change in the density with respect to control plates. Note that both ␣-Man and GlcNAc inhibited the amoeba-induced CPE in a dose-dependent manner; the inhibition was: 60% at 0.31 mM concentration for both sugars, 99% at 1.2 mM ␣-Man, and 90% at 1.2 mM GlcNAc. Mean values are reported. n ϭ 2 or more in each case. saccharide inhibition assays revealed that among monosaccharides, the amoebic lectin has the highest affinity for ␣-Man. The inhibitory potency of D-mannose was 10-fold less than that of ␣-Man. Our results, that the epimers of D-mannose, altrose (C-3), and talose (C-4) were not inhibitory and glucose (C-2) was only a weak inhibitor, suggest that the configuration of free hydroxyl groups at C-2, C-3, and C-4 positions of D-mannose is necessary for optimal binding interactions. The fact that D-lyxose was not inhibitory supports the notion that the CH 2 OH substituent is also essential for binding. The results of the inhibitory potencies of both mannobioses, the mannotriose, and the mannopentose suggest that Man(␣1-3)Man disaccharide is the most complimentary to the carbohydrate-binding site of the amebic lectin being three times more potent than the corresponding ␣1-6-disaccharide. Elongation of the Man(␣1-3) chains by additional mannose residues increased the affinity only slightly.
In vivo, pathology of Acanthamoeba infection is characterized by the infiltration and necrosis of the infected tissue (1, 2) and, as described earlier, the parasite has also been shown to cause CPE on a variety of cell types in vitro (5, 9 -13). Although it has been presumed that the adhesion of amoeba to host cells is a critical first step in the pathogenesis of infection, several studies have reported that cell-free ACM contain destructive proteinases (23)(24)(25) and phospholipases (26). Thus, one could argue that if the amoebae elaborate digestive enzymes, target cell loss may well be independent of host-cell contact. The present study revealed that the trophozoite-induced target cell loss resulted from at least four sequential steps: (i) adhesion of amoeba to target cells; (ii) cytolysis of target cells; (iii) development of cell-free plaques in the monolayer which increase in size with time; and (iv) detachment of the monolayer surrounding the plaques from the culture dish.
What is the likely mechanism by which amoeba produce cell-free plaques in the monolayer? It appears that the first step in the plaque formation is carbohydrate-mediated contactdependent cytolysis. The cytolysis, in turn, makes the cell substratum in the localized areas accessible to one or more proteinases responsible for the detachment of the cells from the culture dish. This mechanism of plaque formation is supported by our findings that: (i) inhibition of the parasite-induced cytolysis by ␣-Man also inhibited the plaque formation and (ii) ACM which lacks cytolytic activity did not produce cell-free plaques. The fact that despite the lack of the ability to induce plaque formation, the ACM results in the detachment of the monolayer suggests that the factors responsible for the detachment of the monolayer are secreted by amoebae and are independent of the mannose-mediated adhesion of amoeba to the host cells. This would explain why prolonged incubation of target cultures with amoebae resulted in the detachment of the monolayer from the substratum despite the presence of ␣-Man in culture media which inhibited: (i) adhesion of amoebae to host cells; (ii) cytolysis of target cells; and (iii) plaque formation.
The actual mechanisms of Acanthamoeba-induced cytolysis and CPE are not known. It has been reported that pathogenic strains of Acanthamoebae produce significantly higher quantities of phospholipases (26), fibrinolytic enzymes (23), and cysteine proteinases (25) compared with nonpathogenic strains. Moreover, a correlation between pathogenicity and proteinase activity has been reported for a number of protozoans (27) including Entamoeba histolytica (28,29), Giardia lamblia (30), Leishmania amazomen (31), and Trypanosoma cruzi (32). In the present study, we found, not only that Acanthamoebae secrete proteinases but also that carbohydrate-mediated adhesion of amoebae to host cells has a profound effect on the nature of the proteinases secreted by the parasite and/or host cells. We found that a specific metalloproteinase (P3) was secreted into the culture media only upon mannose-mediated direct contact of the parasite to target cells. Although it remains to be determined whether P3 is produced by the host cells or the parasite and whether it is the expression, the activity, or merely the secretion of P3 which is elevated upon the adhesion of amoeba to the host cells, it is clear that contact-mediated cross-talk between the parasite and the host cells is a key component of Acanthamoeba infection.
Studies by Alizadeh and co-workers (33) have shown that the Acanthamoeba-induced cell death is due, at least in part, to apoptosis. A recent study has shown that target cells killed by E. histolytica also undergo DNA fragmentation characteristic of apoptotic death (34). Killing of the host cells by E. histolytica FIG. 7. Top, intact amoebae but not ACM induce plaque formation. Confluent cultures of corneal epithelium were incubated with: media alone (Cont), Acanthamoeba (2 ϫ 10 5 parasites/ml), and ACM prepared by incubating amoebae alone in serum-free media for 24 h (2 ϫ 10 5 parasites/ml, 37°C). Note that intact amoebae (A) but not ACM induce plaque formation. However, approximately 6 h following the incubation with the ACM, the cell sheet at the edges of the monolayer began to lift and roll inwards and eventually the entire monolayer lifted en bloc from the dish. Middle, GlcNAc but not ␣-Man inhibits ACM-induced CPE. Confluent cultures of corneal epithelium were incubated for 16 h with ACM ϩ ␣-Man (M) and ACM ϩ GlcNAc (G). Groups M and G, sugars added after ACM was prepared in the media alone. Groups MЈ and GЈ: ACM was prepared in the media containing the sugar. Note that ␣-Man did not inhibit the ACM-induced CPE and GlcNAc inhibited the CPE only if it was present in the media used for the preparation of ACM. GlcNAc did not inhibit the CPE if it was added to the ACM prepared in the media alone. Bottom, GlcNAc inhibits the ACM-induced CPE indirectly. Confluent cultures of corneal epithelium were incubated with filtrate and retentate obtained by filtration of ACM and ACM ϩ GЈ in Centricon tubes. Note that: (i) almost all of the CPE-inducing ability of ACM was retained in the retentate and (ii) removal of GlcNAc from the ACM produced in the presence of the sugar by Centricon filtration did not result in the loss of CPE inhibitory activity of the ACM .   FIG. 8. A, mannose-mediated adhesion of Acanthamoeba to host cells induces expression and/or secretion of a metalloproteinase. Primary cultures of rabbit corneal epithelium were incubated with amoebae (1 ϫ 10 6 parasites/ml) in the serum-free media alone (EϩA) or in the media containing 50 mM ␣-Man (EϩAϩM) for 6 h. At the end of the incubation period, the conditioned media were collected and analyzed for proteinases by zymography on gelatin gels. Note that: (i) four major components (P1, 230-kDa; P2, 97-kDa; P3, 80-kDa; and P4, 55-kDa) were seen in conditioned media prepared by incubating Acanthamoebae with corneal epithelial cells; (ii) component P3 was not detected when ␣-Man was added to the co-culture; nor was it seen in ACM or in media conditioned by epithelial cells alone (EM). B, GlcNAc markedly diminishes the ability of parasites to produce and/or secrete proteinases. Note that P1, P2, and P4 were present in markedly diminished levels in the ACM prepared in the serum-free media containing 50 mM GlcNAc (lane ACMϩG) compared with that prepared in the media alone (lane ACM). In contrast, the zymography pattern of the ACM prepared in the presence of ␣-Man was indistinguishable from the ACM prepared in media alone (not shown). C, components P1, P2, and P4 were susceptible to PMSF, an inhibitor of serine proteinases. D, component P3, E1, and E2 were susceptible to 1,10-phenanthroline, an inhibitor of metalloproteinases.