Homology search within the E. histolytica genome revealed a total of 86 genes coding for putative peptidases. These comprise 50 cysteine peptidases of various families, all of which belonging to clan CA. In addition, 4 aspartic, 10 serine and 22 metallo peptidase genes were identified (Figure 1, Table 1). Structural details of the various E. histolytica cysteine peptidases have been described recently 12.

Primary structure prediction of the other 3 groups of proteolytic enzymes are as follows:

The 4 aspartic peptidases (EhAsP22-1 to EhAsP22-4) share 30 to 40% sequence identity and are homologous to intramembrane-cleaving peptidases (clan AD, family A22). All of them have the specific active site residues TyrAsp and GlyLeuGlyAsp and contain 7 or 8 transmembrane domains but only EhAsP22-1 and EhAsP22-2 have recognizable signal peptides, whereas EhAsP22-3 contains a predicted signal anchor motif. EhAsp22-1, -2 and -3 have significant homology to signal peptide peptidases of various organisms including Trypanosoma cruzi and Arabidopsis thaliana and in addition, EhAsP22-1 and EhAsP22-3 contain the signal peptide peptidase-specific motif GlnProAlaLeuLeuTyr 2425. The primary structure of EhAsp22-4 revealed highest identity (35–40%) to putative presenilins of various organisms including Dictyostelium discoideum, Arabidopsis thaliana and Homo sapiens, but a signal peptide or signal anchor is absent.

Of the 10 E. histolytica genes coding for putative serine peptidases, 5 are predicted to belong to clan SC, family S9 (Figure 1, Table 1), with the active site residues Ser, Asp, His. According to the amino acid residues adjacent to the active site Ser (GGSYGG), EhSP9-1, -2, and -3 can be grouped into subfamily C. The sequences of EhSP9-1 and EhSP9-2 are identical except for a 12 amino acid insertion present in EhSP9-2. In contrast, EhSP9-3, -4 or -5 share only 20% sequence identity with EhSP9-1 or EhSP9-2. The active site residues of EhSP9-4 are not conserved and for EhSP9-5 only a partial sequence of 102 amino acid residues is available. Thus, a reliable assignment of these two enzymes to a specific S9 subfamily is not possible. Signal peptides were identified only for EhSP9-1, EhSP9-3 and EhSP9-4, respectively.

Another 3 enzymes were classified into clan SC but represent most likely members of family S28 of serine peptidases (EhSP28-1, -2, -3), which are also known as lysosomal Pro-Xaa carboxypeptidases. All 3 molecules have a predicted signal peptide and are of similar size comprising betwen 457 and 480 amino acid residues. EhSP28-1 (EhSp1) and EhSP28-2 (EhSp2) have been previously characterized 23. Both are highly similar as they share 89% sequence identity, but only 35% to EhSP28-3.

Two other serine peptidases (EhSP26-1 and EhSP26-2) have homology to members of the signal peptidase family S26B (clan SF) containing the catalytic dyad Ser and His. EhSP26-1 has a calculated molecular mass of approximately 20 kDa and contains 2 hydrophobic regions located near the N- and C-terminus, respectively. Sequence similarity to other members of this family is approximately 45%. In contrast, EhSP26-2 shares only 20% sequence identity with members of the S26 family. Moreover, it does not contain predicted transmembrane regions and the active site is not conserved.

A considerable number of 22 E. histolytica genes are predicted to encode putative metallo peptidases. These are relatively diverse and can be attributed to 7 different clans and 11 different families (Figure 1, Table 1). Six of the enzymes group into clan MA, with the characteristic zinc binding-motif consisting of two histidine residues encompassing the sequence HEXXH. One member is assigned to family M1 (EhMP1-1) and two others to family M3 (EhMP3-1, EhMP3-2). The latter are known as Glu-zincins with the third Zn-binding site being a glutamate residue.

Another two clan MA members (EhMP8-1, EhMP8-2) are homologous to metzincins, which are characterized by a C-terminal His residue being a third zinc-binding site. The two enzymes share 34% sequence identity and both contain a predicted C-terminal transmembrane region but only EhMP8-1 has a signal sequence.

A further clan MA member belongs to family M48 (EhMP48-1) and contains a predicted signal anchor and 6 additional transmembrane domains. The structure of EhMP48-1 is homologous to ste24, an endopeptidase from yeast. Like the yeast enzyme, the amoeba molecule contains the conserved HEXXH zinc-binding motif, located between the fourth and the fifth transmembrane domain.

Another putative metallo peptidase was assigned to clan ME, family M16C containing the characteristic zinc-binding motif HXXEH. Members of this family are falcilysin from Plasmodium falciparum, eupitrilysin from Homo sapiens and CYM1 peptidase from Saccharomyces cerevisiae.

A group of 6 enzymes (EhMP24-1 to EhMP24-6) was predicted to constitute metallo peptidases of clan MG, family 24, which usually represent cytosolic exopeptidases that require co-catalytic ions such as cobalt or manganese. Another 6 peptidases were identified, with homology to metallo peptidases of clan MH. Of these, 2 constitute most likely aspartyl aminopeptidases belonging to family M18 (EhMP18-1, EhMP18-2). They share 40% identity and approximately 35% with members of this family from other organisms. The other 4 amoeba enzymes of clan MH were attributed to family 20 (EhMP20-1 to EhMP20-4). In general, enzymes of this family hydrolyse the late products of protein degradation to complete the conversion of proteins into free amino acids.

The deduced amino acid sequence of a further amoeba gene revealed homology to clan MK, family 22 of metallo peptidases. The only enzyme belonging to this family known so far is the O-sialoglycoprotein endopeptidase from Pasteurella haemolytica. At present, the nature of the active site residues is unknown 26. Like the amoeba homologue, the bacterial peptidase does not possess a signal peptide.

In addition, one E. histolytica enzyme was identified belonging to family M49, clan M. The mammalian homologues are cytosolic dipeptidyl peptidases, which sequentially release N-terminal dipeptides 27. Moreover, an enzyme designated EhU48-1 were annotated, which is similar to EhMP48-1. Like EhMP48-1, it contains a signal anchor sequence and six transmembrane domains. Nevertheless, homology search grouped this peptidase into the U48 family. However, the specificities of the two families are overlapping but not identical 2829.

To allow detailed expression analyses of the various E. histolytica peptidase genes, a small microarray was designed. This array contains 86 specific oligonucleotides representing 4 different E. histolytica houskeeping genes, 3 peptidase-inhibitor genes as well as 79 of the 86 identified peptidase genes. Genes coding for the serine peptidase EhSP9-4 or for the cysteine peptidases EhCP-A7, EhUBP, EhUCH, EhUlp-1, EhUlp-2 and EhUlp-3, respectively, were not included because it was either not possible to design a specific oligonucleotide or they were identified after the array was already spotted. In a first attempt, labelled cDNA from the widely used laboratory strain HM-1:IMSS was hybridized to the array (Figure 2). The results from multiple experiments using RNA preparations from cells grown under standard axenic culture conditions were highly reproducible and indicated that only 3 peptidase genes were expressed at high levels (mean spot intensity >8000), all of them encoding cysteine proteinases (EhCP-A1, EhCP-A2, EhCP-A5). A set of 17 peptidase genes revealed intermediate expression levels (mean spot intensity 800 to 3000). This group comprised the genes for the cysteine peptidases EhCP-A6, -A10, -A11, -B2, -C4 and EhCALP1, the aspartic peptidase EhAsP22-1, the serine peptidase EhSP9-2 and the metallo peptidases EhMP1-1, 16-1, 18-1, 20-3, 20-4, 24-1, 24-2, 24-6 and 48-1, respectively. All other peptidase genes were expressed at levels below the detection limit of Northern blots (mean spot intensity <700). The reliability of the results obtained by array hybridization was confirmed by qRT-PCR using a set of 22 pairs of primers amplyfing cDNAs of the 3 highly expressed genes as well as a representative number of the intermediate or low expressed peptidase genes (Table 2).

In order to determine the extend of inter-strain variation in the expression of peptidase genes, HM-1:IMSS was compared with 6 different E. histolytica isolates all of them cultivated under axenic conditions. These isolates originated from different parts of the world and were obtained from patients with different forms of amoebic disease or in at least one case from an asymptomatic E. histolytica carrier. Pairwise comparison of the various isolates with HM-1:IMSS revealed only minor differences in the expression of the various peptidase genes (Table 3). Three isolates including the one from an asymptomatic carrier showed no differences and two isolates differed only in the expression of one gene. In isolate HK-9, expression of the of the gene for cysteine peptidase EhCP-A5 was decreased by 2.3 fold and in isolate DRP expression of the the gene for the metallo peptidase EhM48-1 was increased by 5.2 fold. The only exception was isolate EGG, which revealed differences in expression for 4 peptidase genes. This isolate was obtained from a patient who simultaneously developed amoebic colitis and liver abscess. Compared to HM-1:IMSS isolate EEG showed decreased expression of the genes for the cysteine peptidase EhCP-A1 as well as for the metallo peptidase EhMP20-3 by about 2 fold, and an increase in the expression of the genes for serine peptidase EhSP9-2 and for the metallo peptidase EhMP20-1 by about 2.8 and 8.6 fold, respectively.

Previous studies have suggested that the level of expression of a number of cysteine peptidase genes is sensitive to heat shock 3031. To further characterize the influence of heat stress on the expression pattern of the various E. histolytica peptidase genes, amoeba were cultured at 42°C for 4 hours and compared with amoebae cultivated under standard culture conditions at 36°C. The results indicated that only 5 of the 79 peptidase genes investigated were differentially expressed upon heat shock. The amount of RNA for the highly expressed genes ehcp-a1 and ehcp-a2 was found to be decreased by about 6 and 4 fold, respectively, whereas the expression of ehcp-a5, ehcp-a6 or ehmp8-2 was increased by approximately 2 fold (Table 4). Similar results were obtained by qRT-PCR. However, there were no significant differences in expression for the remaining 74 peptidase genes.

Seven E. histolytica isolates were used in this study. Strain HM-1:IMSS was isolated in 1967 from a patient with amoebic dysentery, strain NIH:200 was isolated in 1949 from a patient with colitis, strain HK-9 was isolated from a patient with amoebic dysentery (year unknown), strain DRP was isolated in 1985 from a patient with an amoeboma, strain EGG was isolated in 1988 from a patient with colitis and amoebic liver abscess, strain 452 was isolated in 1983 from an asymptomatic carrier. Origin of strain 32 is unknown. HM-1:IMSS, NIH:200 and HK-9 are standard laboratory strains obtained from the American Type Culture Collection. They were original isolated in Mexico, India and Korea respectively. All other strains were isolated in Brasil and kindly provided by Prof. E. F. Silva, University of Minas Gerais, Belo Horizonte, Brasil. Different genotypes of the E. histolytica strains were confirmed by PCR-based genotyping based on variation in the numbers of short tandem repeats that are linked to E. histolytica tRNAs 47. Trophozoites of the various isolates were cultured axenically in TYI-S-33 medium supplemented with 10% adult bovine serum 48. Cells were harvested by chilling on ice and subsequent centrifugation at 430 × g at 4°C for 5 min. The resulting pellet was washed twice with phosphate-buffered saline (6.7 mM NaHPO₄, 3.3 mM NaH₂PO₄, 140 mM NaCl, pH7.2). For heat shock experiments incubation temperature of cultures was shifted from 36°C to 42°C for 4 hours.

Conserved domains of cysteine-, serine-, metallo-, or aspartic-peptidases were used for homology search 49 against the E. histolytica genome as provided by The Sanger Centre and The Institute of Genomic Research 505152. With the help of MEROPS 53 the identified enzymes were grouped into the corresponding peptidase clans and families.

For microarray experiments a 60-base oligonucleotide array was designed containing probes for 79 of the 86 identified putative peptidase genes. The various oligonucleotides contain similar GC-contents of 35.5% and an average T_m of 71.6°C, with a standard deviation of 1.17 (range 66–74°C). The oligonucleotides were designed and synthesized by Eurogentec. Each oligonucleotide was printed in quadruplicate on glass slides (Advalytix Epoxy AD100) in a concentration of 50 μM. The spotting procedure was done in cooperation with the University of Marburg (Genomic Solutions OmniGrid). The oligonucleotide sequences are listed [see Additional file 1].

Total amoeba RNA was isolated using TRIZOL reagent (InVitrogen) according to standard protocols. For microarray analysis 5 μg of total RNA was used. Two biological replicates including dye swap experiments were performed. The reverse transcription of RNA into cDNA was performed according to the Atlas Superscript Fluorescent Labeling Kit (TaKaRa) followed by indirect labelling.

The cDNA labelling was performed with the Cy3- and Cy5-monoreactive dyes (Amersham). In a typical oligoarray the control cDNA was labelled with Cy3, while the experimental second cDNA was labelled with Cy5 (and vice versa for dye swap experiment). Prehybridization and hybridization was performed using standard protocols.

Each array was scanned at 550 nm (Cy3) and at 650 nm (Cy5) at a resolution of 5 μm. Calculation and output of the data was done using ScanArray software, version 3.0 (PerkinElmer). For calculation, the mean signal intensity (pixel) minus local background (pixel) of each spot was used. Flagged spots were eliminated. Two methods for normalizing the data were applied. i) normalization among the totality of genes and ii) normalization among housekeeping genes. For the first normalization method, the signal intensities of each spot of the experiment as well as of each spot of the control were totalised. The sum of the experiment over the sum of the control gives the normalization factor. In order to normalise the control, the signal intensity of each spot of the control was divided by this normalization factor. For housekeeping normalization, the calculation was performed in the same manner as described above but on the basis of the housekeeping gene spot intensities of the control. Spots with a signal intensity (pixel) = 300 were excluded from data analysing. Genes with a ratio of more than 2 and less than 0.5 were considered as differentially expressed.

In order to validate the results obtained by microarray analyses, quantitative RT-PCR (qRT-PCR) was performed by random sampling. Sense and antisense primers were designed to amplify approximately 100 base pairs [see Additional file 2]. These primers were designed independently from the oligonucleotides used on the microarray. Thus, they represent different regions of the same gene. cDNA synthesis was carried out with SuperScriptIII Reverse Transcriptase (InVitrogen). In a final volume of 20 μl, 1 μg of RNase-free DNase-treated total RNA was mixed with 5 × First-Strand buffer, 500 μM dNTPs, 500 nM OdT-T71 (5'-GAG AGA GGA TCC AAG TAC TAA TAC GAC TCA CTA TAG GGA GAT₂₄), 2 mM DTT, 40 U RnaseOut (Invitrogen) and SuperScriptIII (200 U/μl). Incubation was performed for 1 h at 42°C. Quantitative amplification was performed in a Rotor-Gene (Corbett) using RealMasterMix (Eppendorf) SYBR Green kit. 1 μl of the synthesized cDNA was mixed with 2,5× RealMasterMix/20 × SYBR, 5 pmol/μl of the respective sense-primer and antisense-primer to a final volume of 20 μl. Amplification conditions were as follows: 35 cycles at 95°C for 15 s, 58°C for 20 s and 68°C for 20 s and an adjacent melting step (42°C–95°C). Two biological replicates were analysed in triplicate. Relative quantification was carried out with the use of the delta delta ct method provided by the Rotor-Gene software 54 and E. histolytica actin gene RNA as normalizer.