Gene discovery for the carcinogenic human liver fluke, Opisthorchis viverrini

Features of the dataset

Abundantly expressed transcripts

Gene ontology assignments of ESTs from O. viverrini and related flukes

Three hundred and eighty three (383) of the total 1,932 OvAEs (19.8%) could be assigned GO classifications (Figure 1). The most abundant groups represented under the molecular function category were linked to binding (34.8%), catalytic activity (27.9%) and structural molecule activity (13.9%). Other sequences of interest identified in this category were inferred to relate to caspase activity (0.2%) and transporter activity (5.3%). The most abundant groups represented under the biological process category corresponded to physiological processes (41.2%), cellular processes (39.7%) and unknown biological processes (15.6%).

We then undertook a comparative assessment of GO assignments of sequences from O. viverrini and two other trematode parasites of humans in Asia – the liver fluke, C. sinensis (2,679 contigs) and the blood fluke, S. japonicum (107,427 contigs). In general, the percentages of ESTs allocated to each GO category among these three flukes was similar (Figure 2). However, some categories were over- or under-represented in one species. For example, contigs encoding structural proteins were ~four times more abundantly represented in the two liver flukes than in S. japonicum, whereas contigs encoding motor proteins were ~three times more abundant in O. viverrini than they were in C. sinensis or S. japonicum. Sample sizes were too small to determine whether these differences were statistically significant. Both structural and motor proteins are important components of fluke teguments [15], playing roles in surface maintenance and turnover in schistosomes [16, 17] and liver flukes [18]. Therefore, these molecular differences might reflect the specialised niches and physiological requirements of each parasite. From just 1932 OvAEs, 15 different contigs had ORFs encoding components of the dynein complex of motor proteins, a category of motion- related, and surface and gut-localized EF-hand motif- containing antigens which, at least in schistosomes, represent potent allergens and targets of protective immunological responses [17, 19].

Evolutionary relationships between O. viverrini and other platyhelminths

To assess the evolutionary relationships between O. viverrini and other platyhelminths (both parasitic and free-living), we used SimiTri [20] to plot the relative similarities of predicted polypeptide sequences (Figure 3). OvAEs shared most sequence identity with sequences from C. sinensis (2,679 publicly available ESTs), and S. japonicum (107, 427 ESTs). OvAEs were less similar to sequences from the free-living turbellarian platyhelminth, Schmidtea mediterranea (171,472 ESTs) [21], which was not altogether surprising given the phylogenetic distance between parasitic and free-living members of the phylum Platyhelminthes [22, 23]. The bulk of this phylum (including those species analysed here) represents a monophyletic group based on 18S rDNA sequences [22] and morphological characters [24], and is often referred to as the Rhabditophora [22]. Therefore, the members of the Rhabditophora are considered to be more closely related to each other than to other turbellarian clades, such as the Polycladida [22]. A total of 105 OvAEs had homologues in the ESTs from the two parasitic flukes but not in the free-living Schmidtea ([see additional file 1]; selected examples are shown in the table in Figure 3), suggesting that at least some of these are parasitism-specific genes. Thirty-three (33) of the conserved parasitic fluke genes were novel and did not have homologues of known function. Predicted proteins of known function included homologues of legumain, fatty acid binding proteins, myoglobin and potential anti-inflammatory proteins such as Ly6/UPAR domain-containing proteins. Of the parasitic fluke-specific genes, 38 encoded ORFs with N-terminal signal sequences; 14 of these OvAEs had homologues in just S. japonicum and 24 had homologues in both S. japonicum and C. sinensis.

Secreted and membrane proteins

We conducted an analysis of ORFs containing an N-terminal signal peptide or signal anchor. A total of 164 OvAEs contained ORFs with signal sequences. The dataset was divided into three categories – sequences that were (i) novel; (ii) platyhelminth-specific; (iii) conserved across multiple phyla. Novel sequences constituted 55.4% of the total, but only 5.2% of them encoded proteins with a signal sequence. Conserved sequences constituted 36.3% of the total, and 10.7% of these encoded proteins with signal sequences. Finally, the sequences inferred to be platyhelminth-specific accounted for 8.3% of the total dataset, but 20.6% of these encoded proteins with a signal sequence (Figure 4). It should be noted, however, that not all of the OvAEs contained full-length nt sequences, and therefore the true percentage of sequences with signal peptides cannot be definitively inferred in the absence of full genome coverage.

Proteases

As with other parasitic helminth transcriptomes [11–13, 25, 26], proteins with catalytic activity were abundantly represented in the O. viverrini dataset (27.9% of contigs that were assigned GO molecular functions). Many of these enzymes encoded endo- and exo-proteases belonging to established families (MEROPS classification), but which have not yet been described from liver flukes (Table 3). Of particular interest were members of the S1A serine protease family with sequence similarity to kallikrein and chymotrypsin, and, therefore, potentially involved in feeding or tissue migration [27]. Other proteases included homologues of enzymes that digest hemoglobin in blood-feeding helminths, including cathepsin D-like aspartic and cathepsin B-like cysteine proteases [28–30] as well as an asparaginyl endopeptidase, which is known to activate the gastrodermal cathepsin B enzyme, and probably other gut proteases in S. mansoni [31]. We also identified O. viverrini homologues of the cell death enzyme, caspase-2, and the neutral cysteine protease from the tegument of schistosomes, calpain.

Multiple membrane-spanning proteins

Predicted proteins with multiple membrane spanning domains were identified. Tetraspanins, an abundantly represented family of four-transmembrane proteins in the tegument of schistosomes [32, 33], were identified from O. viverrini (Figure 5). These proteins are thought to stabilize the cell membrane by forming a network of interactions, called the tetraspanin web, with other membrane-bound and -associated proteins, particularly on the surface of cells of the immune system [34]. A homologue of the six-transmembrane domain family of water channel proteins, aquaporin [35], was identified. Seven transmembrane proteins are common drug targets [36], and at least three distinct members of this family were identified, including receptors for dystroglycan and lamin b, and a protein with homology to a the DC-STAMP receptor from the surface of dendritic cells (Table 3).

Host-parasite "cross talk"

Parasitic helminths receive host-derived signals for growth and reproduction via surface receptors for host ligands, [37–39]. Convergent evolution of extracellular parasite proteins to promote their interactions with host tissues is well documented [40, 41], and we identified O. viverrini ORFs encoding membrane and secreted proteins, some of which were clearly more similar to vertebrate than to invertebrate homologues (Table 3). Transforming growth factor-beta (TGF-β) regulates cell growth and differentiation and is acquired on the cell surface by specific TGF-β receptors [42]. An ORF encoding a member of the TGF-β receptor type Ib family was identified in O. viverrini. The ORF included a 28 amino acid insertion absent from other type I TGF-β receptors, except for TR1 from the hydatid tapeworms of the genus Echinococcus (also members of the phylum Platyhelminthes) [43]. However, these two insertions did not share sequence identity (Figure 6A). Unlike many of the ESTs identified for which the closest homologues were from parasitic trematodes, the O. viverrini TGF-β receptor type I was divergent from SmRK-I of S. mansoni [44] and instead grouped more closely with proteins from Echinococcus multilocularis and from parasitic and free-living nematodes (Figure 6B). In pairwise sequence comparisons, however, the O. viverrini partial ORF was more similar to pig and macaque sequences (44% identity over 181 amino acids) than it was to Echinococcus TR1 (40% over 180 residues) or SmRK1 (40% over 182 residues). SmRK-I is known to bind to human TGF-β [40], suggesting that the O. viverrini receptor might also bind host growth factors for maturation and reproduction. Another OvAE encoding a protein which is potentially involved in the acquisition of host signals (and subsequent signaling) for growth and development was a fibroblast growth factor (FGF) receptor substrate 2. Parasitic flatworms induce fibrosis (and FGF) [45], and the parasites might acquire and utilize the host FGF that they induce for development and reproduction. Indeed, schistosomes are dependent upon FGF and transferrin for growth and maturation in vitro [46]. Of the sequences presented in Table 3, another OvAE which shared greatest identity with vertebrate homologues, encoded for calumenin, an EF-hand calcium binding protein localized to the secretory pathway. Calumenin is an inhibitor of the gamma-carboxylation system [47] and is expressed in thrombin-activated thrombocytes. It has a modulating effect on the organization of the actin cytoskeleton and may be involved in the pathophysiology of thrombosis or in wound healing [48]. The predicted calumenin of O. viverrini was most similar to rat and frog orthologues/paralogues, suggesting that it might interact with actin on the surface of host cells which are damaged during parasite feeding and migration.

Molecules associated with cancer?

O. viverrini is the major cause of CCA in South-East Asia [1]. The molecular mechanisms underlying induction of O. viverrini-induced CCA are thought to be multi-factorial (reviewed in [49]), but recent evidence suggests that O. viverrini secretes mitogenic proteins into host tissues [1, 8]. OvAEs encoding secreted proteins with prospective mitogenic activity were identified in the EST dataset. Of note, first, progranulin (pgrn) is a pluripotent secreted growth factor that mediates cell cycle progression, cell motility [50] and wound repair [51]. We identified an OvAE (OvAE1732) that shared sequence identity with pgrn (data not shown). Of particular importance is that pgrn has been implicated in regulating the proliferation of tumour cells, and its expression is up-regulated in more aggressive cancers (reviewed in [50]). The kallikrein-like serine proteases are another family of enzymes whose over-expression has been linked to cancer. The expression of some kallikreins in prostate cells leads to changes indicative of an epithelial to mesenchymal transition, an important process in cancer progression [52]. An OvAE (OvAE1918) with sequence identity to kallikrein-like secreted proteases is present in the new O. viverrini gene catalogue. Phospholipase A2 (PLA-2) regulates the provision of arachidonic acid to both cyclooxygenase- and lipoxygenase-derived eicosanoids (reviewed by [53]), and the upregulation of cyclooxygenase-2 is thought to be an important feature of cholangiocarcinogenesis in both humans and experimental rodent models [49, 54, 55]. We identified an OvAE (OvAE1644) that encodes a secreted PLA-2 which shared greatest sequence identity with PLA-2 from venom of Heloderma (Gila monster) and an EST from C. sinensis (Table 3). Parasites utilize secreted serine proteases [56] and PLA-2s [57] to invade host tissues, and homologues of these proteins (and granulin) are potentially secreted by O. viverrini into host tissues where they might promote cell proliferation, mutagenesis and ultimately carcinogensis. Ongoing studies in our laboratories are now focused on the physiological roles of these putative carcinogens in the host-parasite relationship and in cholangiocarcinogenesis induced by O. viverrini infection.

Potential vaccines

Parasite material

Adult O. viverrini were collected from experimentally infected hamsters (Mesocricetus auratas) maintained at the animal facility of the Khon Kaen University Faculty of Medicine. Protocols approved by the Khon Kaen University Animal Ethics Committee were used for all animal research conducted in this study. Briefly, metacercariae of O. viverrini were collected from naturally infected cyprinoid fish by pepsin digestion. Metacercariae (100 per hamster) were administered intragastrically to hamsters. Hamsters were euthanazed 6 weeks after inoculation, and adult worms were flushed with saline from the bile ducts [69]. Worms were washed extensively with sterile phosphate-buffered saline (pH 7.2), after which they were snap frozen and stored in liquid nitrogen or employed immediately as the source of fluke RNA.

Construction and mass excision of cDNA library

Total RNA from adult O. viverrini was extracted using Trizol (Invitrogen), following the manufacturer's instructions. Ten μg of O. viverrini total RNA was used as a template for the synthesis of double-stranded cDNA using the SMART cDNA kit (BD Bioscience), after which the cDNA modified with adapters was cloned into the Sfi I site of the pTriplEx2 plasmid (BD Bioscience) and packaged into λ arms. The titer and percentage of recombinant phages in the library were determined using the protocols recommended by the manufacturer. Escherichia coli strain BM25.8 cells were transduced with recombinant phage, from which the excision of the pTriplEx phagemid library was accomplished.

Clone selection, sequencing and annotation

Five thousand clones were randomly selected from the phagemid library and grown overnight in Luria Bertani (LB) broth supplemented with ampicillin to a final concentration of 25 μg/ml. Overnight cultures were shipped at 4°C in LB broth/ampicillin to the University of Melbourne (Department of Veterinary Science). The sequencing was performed by AgGenomics Inc., Australia, using a 3730xl DNA analyzer (Applied Biosystems). The TempliPhi™ DNA Sequencing Template Amplification system (GE Healthcare) was used to sequence each clone using the 5'λ TriplEx2 sequencing primer.

Bioinformatic analyses

Edited sequences were condensed into contigs or singletons using TGICL [70] with the default parameters of 40 bp overlap, a minimum of 95% identity and a 30 bp maximum mismatched overhang. Sequences of less than 100 nt were discarded. Sequences were named using the same convention as that used for the human blood fluke, Schistosoma mansoni [13]; OvAE for O. v iverrini A ssembled E ST. Sequences were compared with those available in the NCBI non-redundant protein and nucleotide databases using BLASTx and BLASTn. searches, respectively in October 2006. The dbEST database was queried using BLASTn and tBLASTx searches. BLAST alignments with an E-value of ≤ 1.0 × 10^-5 were reported. OvAEs were functionally categorized by querying a local copy of the Gene Ontology (GO) database [71] (downloaded November, 2006) with an E-value cutoff of 1.0 × 10^-5. All ESTs from C. sinensis [9] and Schistosoma japonicum [11, 12] were downloaded from NCBI [72], and the same methodology was used to derive ontology classifications for the C. sinensis ESTs. ORF predictions were performed using GENSCAN [73] using the HumanIso parameter set. Signal sequence prediction was accomplished using SignalP 3.0 [74], incorporating both hidden Markov models and neural networks. Positive signal sequence predictions from either method and positive signal anchor predictions using Markov models were reported. Predictions of transmembrane domains were conducted using TMPred [75]. All multiple sequence alignments were carried out using ClustalW. Clan and family assignments of proteolytic enzymes were analyzed via the MEROPS protease database [76]. Putative phosphorylation sites were predicted using the NetPhos 2.0 server [77].

Phylogenetic trees

Multiple sequence alignments were assembled using ClustalW. Only regions which completely overlapped with partial ORFs of O. viverrini ESTs were used for tree construction. Alignments were imported into PAUP version 4.0 beta [78] to construct trees using the neighbour joining and maximum parsimony methods. Robustness was assessed by bootstrap analysis using 100 replicates. Clades with more than 50% support were denoted with bootstrap values on the branches.

Cross-taxon similarity analysis

OvAEs were compared with all entries for other organisms in the NCBI dbEST database using tBLASTx. The highest BLAST scores (above a cut-off value of 50) were used to generate SimiTri plots [20] using software developed in-house (J. Mulvenna, unpublished).