Antisense inhibition of expression of the light subunit (35 kDa) of the Gal/GalNac lectin complex inhibits Entamoeba histolytica virulence

cDNA representational difference analysis

Using the technique described by Hubank and Schatz (1994), after three rounds of amplification–hybridization, transcripts more abundant in the ‘tester’ than in the ‘driver’, identified in the second and third round of amplification-hybridization, were gel purified and cloned in pGEM-5Zf(+) vector. According to the insert size, two independent inserts (clones 2.1 and 2.2) were isolated from the second round of amplification–hybridization, and two independent inserts (clones 3.1, 3.2) were isolated from the third round of amplification–hybridization. Southern blots of the starting representations (PCR product stage) were probed with the inserts from cloned subtracted products to confirm their uprepresentation in the HM-1:IMSS strain. As a control of basal expression, we used a probe for actin. Figure 1A shows that clone 2.1 is significantly more abundant (three times) in the HM-1:IMSS representation. To confirm that this uprepresentation reflects a higher expression of the corresponding mRNA in the HM-1:IMSS strain, dot-blot quantification and comparison of mRNA transcripts in HM-1:IMSS and Rahman strains was performed using the 2.1 insert as a probe. Figure 1B shows that mRNA of clone 2.1 is more abundant (two times) in the HM-1:IMSS strain compared with the Rahman strain. The insert of clone 2.1 (380 bp) was sequenced, and the resulting sequence was compared with the GenBank database using the blast program (Altschul et al., 1990) in which it was 100% homologous to the mRNA of the E. histolytica 35 kDa subunit of the Gal–lectin complex (accession no. L06065) (McCoy et al., 1993).

Analysis of the 35 kDa subunit expression in avirulent strain Rahman

In order to find out whether the lower level of transcription of the 35 kDa subunit found in the avirulent strain Rahman is correlated with a lower level of 35 kDa subunit expression, we determined and compared by Western blot the intensity of the band patterns in strains HM-1:IMSS and Rahman. After SDS–PAGE of total lysates from Rahman and HM-1:IMSS trophozoites performed under reducing conditions, Western immunoblotting was performed with a polyclonal monospecific antibody raised against a recombinant 35 kDa subunit. As shown previously by Western immunoblots of SDS–PAGE with a monoclonal antibody against the 35 kDa subunit (McCoy et al., 1994), three distinct bands can be seen in the 35 kDa region in lysates of HM-1:IMSS trophozoites (Fig. 1C). An additional band of around 31 kDa is also recognized by the polyclonal antibody used in this study. In strain Rahman, two of the three bands around 35 kDa have lower intensity, and all three bands have a slightly lower molecular mass. In addition, the fourth band is strongly shifted to a lower molecular weight (Fig. 1C). The differences between the band patterns of the 35 kDa isoforms found in Western blots of strains HM-1:IMSS and Rahman could be the result of some structural differences, such as post-translational modifications. The lower intensity of two out of three bands in Rahman trophozoites compared with HM-1:IMSS trophozoites indicates that strain Rahman also has a deficiency in the expression of the 35 kDa subunits. This deficiency is in agreement with the lower transcription level of 35 kDa subunits seen in strain Rahman compared with HM-1:IMSS (Fig. 1B). Western blots reacted with a polyclonal antibody against the Gal–lectin 170 kDa subunit revealed that the levels of Gal–lectin heavy subunit in trophozoites of strains Rahman and HM-1:IMSS were similar (Fig. 1C). Under non-reducing conditions, Western blots reacted with antibody against the Gal–lectin 170 kDa subunit or against the 35 kDa subunit revealed that the level of Gal–lectin 260 kDa heterodimers in strains Rahman and HM-1:IMSS trophozoites were similar (Fig. 4).

Antisense inhibition of the 35 kDa subunit of the E. histolytica Gal/GalNac lectin complex

In order to find out whether the deficiency in the synthesis of the 35 kDa subunit is an important factor responsible for amoebic avirulence, we used the antisense technology, developed previously to inhibit cysteine proteinase expression in E. histolytica (Ankri et al., 1998), to inhibit the expression of the 35 kDa subunit in virulent strain HM1:IMSS. A plasmid construct pSA20 (Fig. 2A) containing a segment of the coding region (842 bp) of the 35 kDa subunit gene (obtained from plasmid gEh-35/1 kindly provided by Dr E. Tannich, Hamburg, Germany) was inserted in the opposite orientation between the 5′ and 3′ untranslated regions of the E. histolytica gene coding for ribosomal protein RP-L21 (ehg34 gene copy) (Petter et al., 1994). pSA20 was transfected into E. histolytica strain HM-1:IMSS and, after a gradual increase in the resistance to G418, the transfected trophozoites were subcultured continuously in the presence of 6 μg ml⁻¹ or up to 150 μg ml⁻¹ G418. Plasmid DNA was rescued from pSA20 trophozoites grown in the presence of 6 μg ml⁻¹ or 60 μg ml⁻¹ G418 and introduced into Escherichia coli cells to yield ampicillin-resistant transfectants. Restriction patterns of the rescued plasmid and the pSA20 plasmid were similar, showing that the plasmid remained intact in E. histolytica trophozoites (data not shown). The transformation efficiency between E. coli transformed with plasmid DNA rescued from pSA20 trophozoites grown in the presence of 6 μg ml⁻¹ or 60 μg ml⁻¹ G418 was ≈10 times, and this is in agreement with the finding of Haman et al. (1995) that the plasmid copy number is proportional to the level of resistance of the cultures to G418. In order to determine whether the antisense mRNA of the 35 kDa subunit is transcribed, the blot was probed with an enriched fraction of the sense single strand or double strand of the 35 kDa subunit gene (Fig. 2B).The mRNA bands in the antisense-transfected trophozoites were more intense for both probes, indicating that they transcribe an antisense mRNA. As a control, the blot was also probed with a double-stranded neo fragment and probed with a double-stranded actin fragment. As shown in 2Fig. 2B, the pSA20-transfected trophozoites grown with 60 μg ml⁻¹ G418 specifically transcribe Neo mRNA. The basal level of transcription of the actin genes is similar in pSA20-transfected trophozoites and in the untransfected strain HM-1:IMSS.

The level of 35 kDa subunit gene expression in the pSA20-transfected trophozoites was determined by Western immunoblot under reducing conditions by enhanced chemiluminescence (Fig. 3), as well as by radioactivity determination after the interaction of blots with ¹²⁵I-labelled protein A (data not shown). In both determinations, trophozoites transfected with pSA20 and grown in the presence of 60 μg ml⁻¹ G418 had a significant decrease in the intensity of the three 35 kDa bands (≈60% less) compared with the intensity of the bands found in the wild-type strain HM-1:IMSS. This result indicates that the expression of the 35 kDa subunit is significantly inhibited in the pSA20 transfectant. On the other hand, the levels of expression of the Gal–lectin 170 kDa in pSA20 transfectant grown in the presence of 60 μg ml⁻¹ G418 were similar to that present in HM-1:IMSS trophozoites. This result is in agreement with the above-mentioned observation made on the levels of light and heavy Gal–lectin subunits in strain Rahman (see above) and again confirm that a deficiency in 35 kDa subunit expression does not affect the level of expression of the heavy 170 kDa Gal–lectin subunit. Western blots reacted under non-reducing conditions with antibody against the Gal–lectin 170 kDa subunit (Fig. 4C) or against the 35 kDa subunit (Fig. 4D) revealed in both cases that the amounts of the 260 kDa Gal–lectin heterodimers present in the pSA20 transfectant were lower than those present in strain HM-1:IMSS. Monomeric forms of the 170 kDa Gal–lectin subunit were not detected under these conditions in the pSA20 transfectant (Fig. 4C).

As an additional control, the levels of enzymatic activity of alcohol dehydrogenase were determined in lysates of pSA20-transfected trophozoites (60 μg ml⁻¹ G418) and untransfected trophozoites and found to be very similar (49.3 and 49.6 U), indicating that general protein synthesis was not affected in the transfected cells. Furthermore, the growth rate in Diamond's TYI-S-33 medium of pSA20 transformants growing in the presence of 60 μg ml⁻¹ G418 was very similar (generation time of 8 h) to various pEhAct-Neo or other transfected trophozoites grown at the same concentration of G418 (Ankri et al., 1998), suggesting that the 35 kDa subunit is not a critical component for growth in culture.

Effects of antisense inhibition of the 35 kDa subunit on cytopathic activity, cytotoxic activity and adherence of the trophozoites

Intact trophozoites of strain HM-1:IMSS are known to destroy baby hamster kidney (BHK) monolayer cells (Bracha and Mirelman, 1984). As shown previously (Burchard and Mirelman, 1988), no significant cytopathic activity was measured for avirulent strain Rahman (Fig. 5A). A 65% decrease in the cytopathic activity was observed for the pSA20-transfected trophozoites of strain HM-1:IMSS compared with untransfected trophozoites or with pSA8-transfected trophozoites that are deficient in cysteine proteinase (Ankri et al., 1998) (Fig. 5A). Interestingly, pSA20 transfectants that were grown for 3 months in the absence of G418 recovered 90% of their cytopathic activity (Fig. 5A). Examination of the amounts of pSA20 plasmid present inside the trophozoite population that was grown in the absence of G418 for 3 months revealed that it was significantly lower (90%), and these trophozoites had higher expression levels of the 35 kDa subunit (data not shown). This indicates that the observed phenotype in the pSA20 transfectants correlates with the levels of antisense transcript. The fact that the pSA8 transfectants, as shown previously (Ankri et al., 1998), are able to destroy BHK monolayers but not the pSA20 transfectant suggests that inhibition of expression of the 35 kDa subunit in the pSA20 transfectant is responsible for the loss of cytopathic activity. This also proves that the inhibition of cytopathic activity is not caused by a squelching effect of the plasmid 5′ and 3′ Ehrpg34 flanking sequences that are present at high copy number.

Incubation of E. histolytica trophozoites in suspension with mammalian cells resulted in killing of the mammalian cells as analysed by trypan blue uptake (Ravdin et al., 1980). A 55% decrease in the cytotoxic activity to BHK cells was observed for the pSA20-transfected trophozoites of strain HM-1:IMSS compared with untransfected trophozoites or with pSA8-transfected trophozoites that are deficient in cysteine proteinase (Ankri et al., 1998) (Fig. 5B). Interestingly, incubations performed in the presence of the polyclonal monospecific antibody (40 μg ml⁻¹) against the recombinant 35 kDa subunit did not inhibit the cytotoxic activity of trophozoites of strain HM-1:IMSS to BHK cells (Fig. 5B).

The Gal-inhibitable adhesion of the trophozoites to the target cells is an event preceding the cytopathic and cytotoxic effect (Petri, 1996). In order to test whether the low cytopathic and cytotoxic activity in the pSA20-transfected trophozoites results from a defect in adherence, we measured the adhesion of pSA20-transfected trophozoites to BHK cells (Fig. 6A) as well as to ¹⁴C-labelled Escherichia coli serotype 055 (Fig. 6B). No significant difference was found between the levels of adhesion of strain Rahman, pSA20-transfectant, pSA8-transfectant and strain HM-1:IMSS to BHK cells (Fig. 6A). The adhesion was inhibitable for all the strains at 200 mM galactose (≈50–70% of inhibition) (Fig. 6A), which indicates that most of the adhesion in all the tested trophozoites continues to be Gal–lectin mediated (Petri, 1996). Furthermore, no significant difference was found in the adherence of E. coli serotype 055, a bacteria known to attach to the amoebic Gal-surface lectin (Bracha et al., 1982), to the strain Rahman, pSA20 transfectant, pSA8 transfectant or HM-1:IMSS untransfected trophozoites (Fig. 6B).

Effect of antisense inhibition of the 35 kDa subunit on cysteine protease activity and amoebapore

Cysteine proteinase and amoebapores have been described as amoebic molecules that cause damage to the host cells (Leippe, 1997; Que and Reed, 1997). In order to see whether the inhibition of expression of the 35 kDa subunit has an effect on these amoebic molecules, we compared the cysteine proteinase activity as well as the amount of amoebapore A and B between the pSA20-transfected trophozoites and the HM-1:IMSS strain. No difference was found between these types of trophozoites in the level of cysteine proteinase (± 45 U) or in the amounts of amoebapore A and B detected on Western blots by specific antibodies obtained from Dr M. Leippe (Hamburg, Germany) (data not shown).

Effect of antisense inhibition of the 35 kDa subunit on erythrophagocytosis activity and resistance of trophozoites to lysis by serum

The Gal/GalNac-inhibitable amoebic lectin activity is involved in the attachment and agglutination of human erythrocytes (Ravdin et al., 1985). In order to see whether the inhibition of expression of the 35 kDa subunit in the transfectant has an effect on the rates of erythrophagocytosis of human red blood cells, we compared the erythrophagocytosis activity between the pSA20-transfected trophozoites and the HM-1:IMSS strain. No difference between the level of erythrophagocytosis was found between each type of trophozoites (± 12 erythrocytes per amoeba after 10 min of incubation at 37°C).

Virulent amoebae activate the complement system but are able to evade this host defence mechanism (Hamelmann et al., 1992). This evasion has been shown to be mediated by the ability of the Gal–lectin to bind complement components C8 and C9, thereby abrogating assembly of the membrane attack complex (Braga et al., 1992). No significant difference was found in the susceptibility of pSA20 transfectants to human serum in comparison with the control of pEhAct-Neo or untransfected strain HM-1:IMSS. In all cases, lysis was approximately ± 70%.

Effect of antisense inhibition of the 35 kDa subunit on liver abscess formation in the hamster

In order to test whether the decrease in cytopathic activity of pSA20 is correlated with a general decrease in virulence that can also be observed in liver abscess formation, hamsters were inoculated intrahepatically with 5 × 10⁵ trophozoites/animal of non-transfected HM-1:IMSS strain, pTS-4 transfectants or pSA20 transfectants grown in the presence of 150 μg ml⁻¹ G418. All the hamsters inoculated with HM-1:IMSS strain or with pTS-4 transfected amoebae presented extensive necrotic lesions (> 2 cm). Out of the six hamsters injected with pSA20 transfectants, only one showed a very minor lesion (< 0.2 cm). As a control, we rescued and determined in transformed E. coli cells the amounts of pSA20 plasmid that were present in transfected trophozoites grown in TYI-33 medium in the presence of 150 μg ml⁻¹ G418 and then cultivated for 1 week in TYI-S-33 medium in the absence of G418. The numbers of E. coli colonies obtained before or after the removal of G418 was approximately the same (data not shown).

E. histolytica strains

Trophozoites of E. histolytica strains HM-1:IMSS and Rahman were cultured under axenic conditions in Diamond's TYI-S-33 medium (Diamond et al., 1978). Trophozoites in log phase of growth were used in all experiments. Avirulent strain Rahman is a well-characterized E. histolytica strain that has been known for quite a number of years (Sargeaunt et al., 1980; Burchard and Mirelman, 1988). It was originally isolated in England in 1964 by Dr Robinson from the stool of a sailor arriving from Calcutta and suffering from dyspepsia. It was subsequently cultured by the late Dr Ralph Neal in London and axenized in the laboratory of Dr L. S. Diamond (personal communication). In the beginning, the strain had virulent properties but, after axenization, an avirulent culture was obtained, and the original virulent parental strain is no longer available.

E. histolytica strains HM-1:IMSS and Rahman were tested before use for their ability to produce liver abscesses in hamsters (Burchard et al., 1988; Navarro-Garcia et al., 1995) and for their cytopathic activity on baby hamster kidney (BHK) cell monolayers (Bracha and Mirelman, 1984). E. histolytica strain HM-1:IMSS was fully virulent, whereas strain Rahman had no cytopathic activity (Fig. 5A) and did not induce the formation of liver abscesses in hamsters (data not shown). E. histolytica pSA8 transfectant (ehcp5 antisense RNA; Ankri et al., 1998) was used in this study as a positive control for the cytopathic activity assay. E. histolytica pTS4 transfectant (CAT reporter gene flanked by the upstream and downstream regions of Ehg34; Moshitch-Moshkovitch et al., 1998) was used in this study as a positive control for the induction of amoebic liver abscesses in hamsters.

cDNA representational difference analysis (cDNA RDA)

For the cDNA RDA, we used the same protocol as that described by Hubank and Schatz (1994). Briefly, total RNA was prepared using an RNA isolation kit RNAZOL B (Cinna/BioTech Laboratories). mRNA from strains HM-1:IMSS and Rahman was isolated twice by purification over an oligo(dT) cellulose column (Ausubel et al., 1987). Double-stranded cDNA was prepared by reverse transcription of the poly(A) RNA using an oligo(dT) primer (Ausubel et al., 1987). To reduce its complexity, we restricted each cDNA with the 4 bp cutting enzyme DpnII. After ligation of the synthetic oligonucleotides (referred to in the report by Hubank and Schatz, 1994) to restricted cDNAs and filling in of the ends, cDNAs were amplified by PCR (Taq DNA polymerase; Boehringer Mannheim) to obtain the HM-1:IMSS and Rahman representations. Assuming that E. histolytica strain HM-1:IMSS specifically expressed some additional genes that are involved in virulence compared with the avirulent strain Rahman, strain HM-1:IMSS was chosen as the ‘tester’ and strain Rahman as the ‘driver’. Three rounds of amplification–hybridization were performed. The overabundant selected products were cloned in the pGEM-5Zf(+) vector using the pGEM-T vector systems (Promega).

Construction of hybrid plasmids for transfection

All common techniques and routine DNA manipulations, including transformation, plasmid preparation and gel electrophoresis, were carried out according to standard procedures (Sambrook et al., 1989). Restriction and modifying enzymes (Boehringer Mannheim and New England Biolabs) were used in accordance with the manufacturer's recommendations.

Plasmid pSA20 (Fig. 3A) was constructed as follows: the 35-kDa subunit of the E. histolytica Gal-inhibitable lectin was amplified by PCR from the plasmid gEh-35/1 (a kind gift from Dr E. Tannich, Hamburg, Germany) using the sense primer catgccatggttatattagtcttattgata and the antisense primer gcgcgtcgacttatgcaaacacaggaataa. The 867 bp PCR product obtained was sequenced and was fully homologous to the E. histolytica Gal–lectin 35 kDa subunit. This fragment was digested with Bgl II and ligated to our previously reported plasmid pSA8 (Ankri et al., 1998) after removal of its EhCP5 insert by Bgl II and NotI digestion. After blunt ending with T4 DNA polymerase, the plasmid was closed by ligation.

Transfection of E. histolytica cells

Exponentially growing trophozoites of E. histolytica strain HM-1:IMSS were transfected as described previously (Vines et al., 1995). The cells were allowed to recover for 2–3 days, and the G418 concentration was gradually increased until all the trophozoites in the control tubes, which underwent electroporation in the absence of plasmid, died. The media was removed carefully every 24–48 h to reduce the amount of debris in the culture until the transfected trophozoites recovered. The G418 concentration was then raised gradually to 150 μg ml⁻¹ within a period of 2–3 weeks.

Southern and Northern blot hybridization

For Northern blot hybridization, total RNA was prepared using an RNA isolation kit RNAZOL B (Cinna/BioTech Laboratories), and 10 μg of total RNA was size fractionated under denaturing conditions on agarose gels containing formaldehyde. The DNA or RNA was transferred to a nylon membrane and hybridized under stringent conditions as described previously (Bracha et al., 1995). Probes were prepared from the appropriate PCR fragments. For preparation of the single-strand probe, used in the Northern blot hybridization, asymmetric PCR was performed using the E. histolytica galactose-inhibitable lectin (g-35/1) gene (accession number M96024) as template. A labelled double strand probe was prepared with the Ready Prime II kit (Amersham) using the same template.

SDS–PAGE and Western immunoblotting

E. histolytica trophozoites were solubilized with 1% NP-40 in PBS in the presence of 50 μM protease inhibitor E-64. Proteins from whole E. histolytica were resolved for 35 min at 200 V on 10% polyacrylamide gels (25 μg lane⁻¹) under reducing conditions (Laemmli, 1970) or for 60 min at 200 V on 8% polyacrylamide gels (25 μg lane⁻¹) under non-reducing conditions (without β-mercaptoethanol) and transferred electrophoretically to nitrocellulose membranes. The blots were reacted with a suspension of polyclonal anti-Gal–lectin 170 kDa (1:5000) (a gift from Dr S. L. Stanley, St Louis, MO, USA) or with a suspension of polyclonal monospecific anti-35 kDa subunit (1:250) (see below) and then subjected to interaction with an HRP-conjugated goat anti-rabbit antibody (1:5000) and developed by enhanced chemiluminescence. Detection was done by autoradiography. The amount of bound polyclonal monospecific anti-35 kDa subunit antibody was determined by incubating the blots with ¹²⁵I-labelled protein A (10⁻⁶ Ci ml⁻¹; Amersham) for 1 h. Detection and quantification was carried out using a phosphorimager.

For Western immunoblotting of the amoebapore, trophozoites were resuspended in saline buffer containing protease inhibitors (10 μM E-64, 6.4 mM benzamidine, 0.5 mM PMSF) at a concentration of 10⁷ trophozoites ml⁻¹ and disrupted by freeze–thaw (three times). Homogenates were sedimented at 100 000 × g at 4°C for 15 min. Proteins from the soluble fraction (5 μg lane⁻¹) were resolved on 20% polyacrylamide gels under non-reducing conditions and transferred electrophoretically to nitrocellulose membranes. The blot was reacted with a suspension of polyclonal anti-amoebapore A or B (1:1000) (a gift from Dr M. Leippe, Hamburg, Germany) and then subjected to interaction with an HRP-conjugated goat anti-rabbit antibody (1:5000).

Preparation of a polyclonal monospecific anti-35 kDa subunit antibody

A recombinant 35 kDa subunit was prepared from the prokaryotic expression vector system pGEX-2T (Pharmacia). The 35 kDa subunit of the E. histolytica Gal-inhibitable lectin was amplified by PCR from the plasmid gEh-35/1 (a kind gift from Dr E. Tannich, Hamburg, Germany) using the sense primer cgcggatccaagactcaagacggaaaagat and the antisense primer gcgcgtcgacttatgcaaacacaggaataa. The PCR product obtained was digested with Bgl II and BamHI and cloned into the pGEX-2T plasmid at the BamHI site of the polylinker. This allowed for the inducible expression of a glutathione-S-transferase–35 kDa subunit. Intracellular fusion protein was recovered from E. coli lysates and purified by affinity chromatography using a glutathione Sepharose gel matrix by direct exchange with reduced glutathione (1 mM). The glutathione-S-transferase (GST)–35 kDa subunit fusion protein was used to inoculate rabbits for the generation of polyclonal anti-35-kDa subunit antibodies. Monospecific anti-35 kDa subunit antibodies were obtained by affinity chromatography of the rabbit antiserum over a Sepharose-4B column (Pharmacia) conjugated to the GST–35 kDa subunit fusion protein. The specific recognition of the polyclonal monospecific anti-35 kDa subunit antibodies to the GST–35 kDa subunit fusion protein (data not shown) or to the E. histolytica 35 kDa Gal–lectin subunits (Fig. 3) was verified by Western blot analysis and compared with the specific recognition of a monoclonal antibody for the 35 kDa subunit (a kind gift from Dr B. Mann, VA, USA) as a positive control. Preimmune serum was not able to detect either the GST–35 kDa fusion protein or the E. histolytica 35 kDa subunit (data not shown).

Cytopathic activity

The destruction rate of tissue-cultured monolayers of baby hamster kidney (BHK) cells by transfected and untransfected viable trophozoites (2 × 10⁵) was performed essentially as described previously (Bracha et al., 1984). After 60 min of interaction between the trophozoites and BHK cells, the experiment was stopped by placing the plates on ice and washing off the trophozoites with cold saline. The percentage of living BHK cells remaining in the wells was determined using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide]-based assay (Mosmann, 1983).

Cytotoxic activity

The activity was determined by the vital dye exclusion assay (Ravdin et al., 1980). Freshly harvested BHK cells (9 × 10⁵) with and without amoebae (1.5 × 10⁵) were placed in 450 μl of TYI-S-33 media without serum and incubated with rotation at 37°C for 1 h. Samples (30 μl) were taken at different time points, treated with trypan blue (0.1% final concentration) and examined microscopically in a haemocytometer chamber. Viability was calculated by counting the BHK cells that included or excluded trypan blue. Three separate counts were made for each sample, and the experiment was repeated twice. Standard deviation of the mean was calculated and found to be in the range of ± 5%. The cytotoxic activity of strain HM-1:MMS was also determined in the presence of added polyclonal antibodies, monospecific against the 35 kDa subunit (40 μg ml⁻¹).

Adhesion assay to BHK cells

For adhesion assays, the BHK monolayers were fixed with 5% formaldehyde, washed twice with PBS, incubated with 250 mM glycine for 30 min and washed with PBS. Trophozoites (2 × 10⁵) were added to wells containing fixed monolayers in 1 ml of Dulbecco's modified Eagle medium (DMEM) without serum and incubated at 37°C for 30 min. The number of trophozoites adherent to BHK cells was determined by counting under the microscope the trophozoites that remained adhered to cell monolayers after gentle decanting (twice) of the non-adhered trophozoites with 37°C warmed DMEM. Inhibition of adhesion was determined in the presence of 200 mM galactose.

Adhesion assay of E. coli serotype O55

The assay was performed according to Bracha et al. (1982). [¹⁴C]-E. coli (10⁹) (300 c.p.m. 10⁻⁶ bacteria) were incubated with trophozoites (10⁶) in 1 ml of saline solution for 30 min at 4°C. Separation between bacteria that attached to the trophozoites and those that did not was performed by discontinuous density gradient centrifugation with Percoll (Pharmacia). Unattached bacteria layered between 100% and 60% Percoll, whereas amoebae layered between 60% and 20%. The bacteria attached to trophozoites were counted in a scintillation fluid using a liquid scintillation counter.

Erythrophagocytosis assay

Erythrophagocytosis assay was measured as described previously (Mora-Galindo et al., 1997). HRBC (10⁸) and trophozoites (10⁶) were mixed and incubated for 10 min at 37°C. Phagocytosis was stopped by the addition of distilled water. The average number of erythrocytes inside the trophozoites was quantified using a calibration curve and by reading the absorbance at 397 nm after resuspending the pellet of parasites in 90% formic acid.

Alternative pathway killing of E. histolytica by non-immune human sera

Determination of alternative pathway killing of E. histolytica was as described previously (Hamelmann et al., 1992). Human sera (0.1 ml) was mixed with 5 × 10⁵ trophozoites in 0.15 ml of 20 mM phosphate buffer, pH 7.4, containing 135 mM NaCl, 10 mM EGTA and 2.5 mM MgCl₂. After incubation for 15 min at 37°C, the number of dead amoebae was determined (sum of those lysed and stained with 0.1% eosin). The control group contained heat-inactivated sera (56°C for 30 min).

Cysteine proteinase activity

CP activity was measured in total lysates of trophozoites (10⁶) lysed in 1 ml of Nonidet P-40 (1% in PBS). CP activity was determined by following the digestion of the chromophoric substrate benzyloxycarbonyl-l-arginyl-l-arginine-p-nitroanilide (Z-Arg-Arg-pNA; Bachem) (Leippe, 1995). One unit (U) of activity is defined as the number of μmol of substrate digested min⁻¹ mg⁻¹ protein.

Alcohol dehydrogenase (ADH) activity

ADH activities were determined as described previously (Leippe, 1995). One unit (U) is defined as the number of μmol of substrate digested min⁻¹ mg⁻¹ protein.

Induction of amoebic liver abscesses

Syrian golden hamsters (6 weeks old) were inoculated intrahepatically with 5 × 10⁵ (i) E. histolytica HM1:IMSS trophozoites; (ii) E. histolytica HM1:IMSS trophozoites transfected with pSA20 and grown in the presence of 150 μg ml⁻¹ G418; and (iii) E. histolytica HM1:IMSS trophozoites transfected with pTS-4 (Moshitch-Moshkovitch et al., 1998) and grown in the presence of 150 μg ml⁻¹ G418. Hamsters (at least four in each group) were sacrificed 1 week after intrahepatic inoculation, and the formation of lesions was evaluated.