Background

Parasitic flatworms (phylum Platyhelminthes; class Trematoda) are responsible for neglected tropical diseases (NTDs) that affect more than 750 million humans and other mammals throughout the world, particularly in Southeast Asia and the Western Pacific regions [1,2,3]. Important trematode species include Clonorchis sinensis and Opisthorchis spp. [3, 4]. The consumption of fish containing the infective developmental stage (metacercaria) of these liver flukes leads to an infection that, if chronic, can cause serious hepatobiliary diseases in humans, including clonorchiasis (C. sinensis) or opisthorchiasis (Opisthorchis spp.), particularly in Asia [3, 5,6,7]. For example, C. sinensis impacts more than 35 million people across China, Japan, Korea and Vietnam [8, 9], and has been classified as a Class I carcinogen by the International Agency for Research on Cancer (IARC) [10]. Despite the importance of this parasite, the molecular mechanisms that govern or modulate the interactions between C. sinensis and its host animals remain largely unknown.

When piscivorous mammals (e.g. humans, dogs and cats) consume fish infected by C. sinensis, metacercariae excyst in the duodenum, and juveniles migrate and develop to adult flukes in the biliary system [11, 12]. The adult flukes can alter biliary duct pathology via mechanical irritation and by releasing molecules to facilitate feeding [13]. Furthermore, as the adult flukes grow and migrate, they can obstruct the bile duct and elevate bile duct pressure [6]. Clonorchiasis often leads to chronic hepatobiliary illness and can induce cholangiocarcinoma (CCA), a malignant cancer of the biliary system [1, 5, 14]. To better understand the pathogenesis of clonorchiasis and CCA, and to assist in efforts to control the parasite causing these diseases, researchers have explored the function of C. sinensis proteins, including proteins likely to be important for establishment and survival in the human biliary system.

In bile, cholesterol and phospholipids are abundant, and lipid-binding proteins (LBPs) have been proposed to play an important role in maintaining the chemical homeostasis of liver flukes in the bile duct [15]. Interestingly, gene duplication events have led to more than 20 copies of genes encoding homologues of a lipid-binding protein, Niemann-Pick type C2 (NPC2), in the genomes of C. sinensis and O. viverrini [16,17,18]. In most eukaryotes, NPC2 is encoded by a single copy gene. In some mammals, this protein binds cholesterol and other lipids and transports them out of the lysosome to other parts of the cell [19]. In arthropods, duplication and genetic differentiation of NPC2 genes [20,21,22,23] have led to new molecular functions, including chemoreception via binding to semiochemical and other volatile compounds [23,24,25]. Related proteins that contain a conserved myeloid differentiation factor-2 (MD-2)-related lipid-binding (ML) domain are also reported to play diverse roles in lipid metabolism, innate immunity and/or chemoreception in arthropods [20, 26]. Despite the substantial expansion of the NPC2-like protein family in liver flukes, almost nothing is known, at the molecular level, about their biological functions in such parasites and/or the regulation of host-pathogen interactions. Here, we employed a bioinformatic workflow to predict, curate and annotate NPC2-like proteins encoded in the draft genomes of a Korean and a Chinese isolate of C. sinensis.

Methods

Inference of NPC2-like protein sequences

Two C. sinensis genomes, one assembled from a Korean isolate (gene accession numbers beginning with “Cs-k2”; BioProject ID: PRJNA386618) [17] and one from a Chinese isolate (gene accession numbers beginning with “csin”; BioProject ID: PRJNA72781) [18], and their gene annotations were downloaded from the WormBase ParaSite database (v.13; accessed 20 May 2019) [27]. NPC2-like homologues in C. sinensis were initially identified (Fig. 1, steps 1-3). First, homologues of nine curated NPC2 proteins available in the SWISS-PROT database (Table 1; accessed 20 May 2019) [28] were identified in the C. sinensis proteomes using BLASTp v.2.2.29 (E-value cut-off: 10) [29]. Second, a reciprocal BLASTp (E-value cut-off: 10) search of homologues of NPC2 proteins from SWISS-PROT against the NCBI non-redundant protein database (NCBI-nr; accessed 20 May 2019) [30] was performed. Clonorchis sinensis NPC2-like protein homologues matching proteins submitted to NCBI-nr and annotated as “Niemann-pick C2 protein”, “NPC intracellular cholesterol transport 2”, “Epididymal secretory protein E1” or “Epididymal secretory protein E1-like”, or that lacked a protein description (unnamed or hypothetical proteins), were retained. Third, we searched for conserved domains in the predicted C. sinensis proteome using the program InterProScan v.5.15.54 [31], utilising the Pfam database v.27.0 [32]. Proteins with a conserved ML domain (Pfam identifier PF02221) were retained. Numbers of C. sinensis proteins with NPC2 homologues and/or a conserved ML domain were displayed in a Venn diagram employing the Intervene tool [33]. Heatmaps of BLASTp and InterProScan bit scores were created using the R package ggtree v.1.16.6 [34].

Fig. 1
figure 1

Summary of workflow and results for the prediction, curation, and annotation of NPC2-like proteins in the genomes of a Korean and Chinese isolate of Clonorchis sinensis. Initially, putative C. sinensis NPC2-like proteins were identified using BLASTp sequence homology searches against curated NPC2 proteins submitted to the SWISS-PROT database (step 1). Selected proteins with a reciprocal best match to NPC2-like proteins submitted to NCBI-nr database were then identified (step 2). Next, C. sinensis predicted proteins with a conserved myeloid differentiation factor-2 (MD-2)-related lipid-binding (ML) domain (PF02221) were identified using InterProScan (step 3). The intersect between NPC2 protein sequence homology and retention of a conserved ML domain was then used to select putative NPC2-like proteins. Reciprocal BLASTn sequence homology across gene regions was used to infer orthology and incomplete homologous sequences were complemented in both isolates of C. sinensis using BLAT and Exonerate (step 5). Phylogenetic relationships among identified C. sinensis NPC2-like protein-encoding genes was then determined by employing MACSE2, TrimAl and MrBayes (step 6). Last, all NPC2-like proteins identified were functionally annotated using SignalP-5.0, SWISS-PROT and InterProScan databases, as well as I-TASSER, to predict tertiary structures. Transcription in different development stages and adult tissues was inferred using HISAT2 and FeatureCounts (step 7). The number of Korean (blue) and Chinese (red) C. sinensis sequences retained in each step are indicated

Full size image
Table 1 Nine curated Niemann-Pick C2 (NPC2) proteins in the SWISS-PROT database used for identification of C. sinensis NPC2-like protein homologues. Protein length, signal peptide position and the most similar solved crystal structure are given for each sequence
Full size table

Curation of sequences

All inferred C. sinensis NPC2-like proteins were individually curated in three steps (Fig. 1). First, gene regions (including exonic and intronic nucleotide sequences) encoding C. sinensis NPC2-like proteins were extracted from each of the two draft genomes (representing Chinese and Korean isolates) via the corresponding “general feature format” (GFF) file employing the gffread tool v.0.11.4 [35]. A reciprocal BLASTn v.2.2.29 search against the C. sinensis NPC2-like gene sets (for both isolates) was then conducted to infer gene orthology. Second, C. sinensis NPC2-like proteins from each isolate (Chinese or Korean) were reciprocally mapped to the genome of the alternative isolate using BLAT v.34x12 [36], and a new gene model was inferred based on this mapping employing the program Exonerate v.2.2.0 [37].

Analysis of transcription

Available RNA-Seq data (testis, sucker, muscle, ovary, adult, 8-week adult, metacercaria, 2-week juvenile; ENA/SRA accession numbers: ERR604978–ERR604981, SRR189060, SRR6188894–SRR6188896) for C. sinensis were mapped to each reference genome using HISAT2 [38]. From these mapped data (stored in the BAM format), read counts were inferred using the corresponding GFF files and employing FeatureCounts v.1.6.4 [39]. Read counts from each library were then normalised to counts per million (CPM) using the edgeR package v.3.26.8 [40]. A heatmap matrix of CPM per gene per library was created using the ggtree R package v.1.16.6 [34].

Protein annotation

Signal peptides and their cleavage sites were inferred for C. sinensis NPC2-like proteins using the SignalP-5.0 [41]. The structures of the mature C. sinensis NPC2-like proteins (i.e. without their predicted signal peptide domain) were modelled and annotated using the program I-TASSER v.4.4 [42] and compared with the crystal structures of NPC2s from Bos taurus (cow; Protein Data Bank (PDB) accession 2HKA chains A and C) [19] and Camponotus japonicus (Japanese carpenter ant; PDB accession 3WEA chain A) [43] employing the program UCSF Chimera v.1.9 [44].

Phylogenetic analysis

The mRNA sequences encoding C. sinensis NPC2-like proteins were extracted from each assembled genome sequence via their corresponding GFF files employing gffread. Codons of C. sinensis NPC2-like sequences were aligned using a translated protein sequence alignment employing MACSE v.2.03 [45]. Gaps were removed from the nucleotide alignment using trimAl v.1.4.rev15 [46] using the -gappyout option. The Akaike Information Criteria (AIC) test in ModelFinder [47] selected the general time reversible model of evolution for subsequent phylogenetic analyses. Bayesian phylogenetic inference (BI) was determined using Markov chain Monte Carlo (MCMC) analysis in MrBayes [48]. Two million generations of MCMC analysis were performed, and trees were recorded every 200th generation. At this point, the standard deviation of split frequencies was < 0.01, and the potential scale reduction factor (PSRF) approached 1. Consensus trees (50% majority rule) were generated using the final 75% of trees. Trees were annotated and enhanced using the ggtree R package v.1.16.6 [34], and nodal support values on trees were indicated as posterior probabilities (pp).

Results

NPC2-like proteins of Clonorchis sinensis

Based on predicted protein sequence homology, 72 of 14,538 Korean C. sinensis proteins and 71 of 13,634 Chinese C. sinensis proteins were homologous to one or more NPC2 proteins in SWISS-PROT (Fig. 1, Additional file 1: Table S1). Sequence homology between NPC2 proteins and their best matched C. sinensis proteins was mostly low (20.0–47.6 % amino acid identity), and sequence alignment coverage was highly variable (14–100%). In total, 55 Korean and 56 Chinese NPC2 homologues were reciprocal BLASTp matches (E-value cut-off: 10) to NCBI-nr proteins annotated as NPC2-like or those without a protein description and were thus retained (Fig. 1, Additional file 1: Table S1). Of those, 35 Korean and 27 Chinese C. sinensis proteins contained at least one ML domain, with an average conserved domain length of 107 amino acids (Fig. 1, Additional file 1: Table S1). Two conserved ML domains were predicted in three Korean C. sinensis protein sequences (Cs-k2.gene14549, Cs-k2.gene14290 and Cs-k2.gene14112). Based on protein sequence homology to NPC2 and the presence of a conserved ML domain (Fig. 1), 35 and 27 NPC2-like proteins were predicted to be encoded in the Korean and Chinese C. sinensis gene sets, respectively. One Chinese protein sequence (csin112467) that shared no significant sequence homology to NPC2 proteins, but contained a conserved ML domain, was also retained for subsequent curation (Fig. 1).

Gene models

Reciprocal nucleotide alignments of Korean and Chinese C. sinensis NPC2-like gene regions (including introns) and mapping to the alternative genome assemblies identified paired orthologues between the two isolates. Nine pairs of NPC2-like gene models were consistent in gene model structure and overall sequence length. Based on reciprocal nucleotide matches across exonic and intronic regions, we identified 11 additional orthologous pairs with unresolved variation in the first exon positions and lengths. Three of the Korean C. sinensis genes identified by reciprocal nucleotide alignments (i.e. Cs-k2.gene992, Cs-k2.gene8673 and Cs-k2.gene14547) were removed as they did not encode a conserved ML domain or share amino acid sequence homology with NPC2 proteins from SWISS-PROT. In contrast, four Chinese C. sinensis genes (csin101111, csin103126, csin111538 and csin111895) were added based on mapping of Korean NPC2-like gene sequences to the genomic region encoding these genes. Six Korean gene models, although aligning to the Chinese genome, only aligned partially to the corresponding Chinese gene model in this genomic region, and available sequence data were insufficient to resolve a single gene model for both isolates. Eight Korean C. sinensis genes were not identified in the Chinese genome, and four Chinese C. sinensis genes were not identified in the Korean genome. Taken together, 35 Korean and 32 Chinese NPC2-like protein genes were retained.

Tertiary structures and functional annotation

The tertiary structures of 35 Korean and 32 Chinese C. sinensis NPC2-like proteins were modelled using I-TASSER (Additional file 1: Table S2). The I-TASSER model confidence (C-) scores ranged from -5 (lowest confidence) to 1.29 (highest confidence). For comparison, I-TASSER models were also inferred for nine curated NPC2 proteins from SWISS-PROT (Table 1). For these proteins, the model C-scores ranged from 1.30 to 1.46. Eleven Korean and nine Chinese C. sinensis NPC2-like proteins had predicted structures with C-scores of ≥ 1. These high-confidence models were retained for further analysis. Two Korean and three Chinese NPC2-like proteins had C-scores of < -4 and were thus low-confidence predictions. Proteins Cs-k2.gene6404 (Korean isolate) and csin102672 (Chinese isolate) had the highest paired ortholog C-scores (1.26). Chinese NPC2-like proteins csin107773 and csin111438 had the highest (1.29) and lowest C-score (-5), respectively. Of note, six Korean and Chinese NCP2-like proteins with a C-score of < -4 were not predicted to retain a signal peptide region (Additional file 1: Table S2). Based on protein structurally close to the target in PDB, 2HKA chain A (NPC2 from cow) and 3WEA chain A (NPC2 from the Japanese carpenter ant) had the highest and second-highest structural similarity to C. sinensis NPC2-like models (representing 44 and 5 of all structures predicted, respectively; Additional file 1: Table S2). The most commonly predicted ligand (73%) was cholesterol (C3S).

Phylogenetic relationships

The phylogenetic relationship among C. sinensis NPC2-like proteins was determined using aligned coding domains; the resulting tree (Fig. 2a) was annotated with experimental data (Fig. 2b-h). NPC2-like proteins of C. sinensis clustered in four well-supported groups (pp = 0.81–1.0) that contained 27 paired orthologues, with eight and four proteins being unique to the Korean and Chinese isolates, respectively (Fig. 2b). The Korean and Chinese C. sinensis proteins most similar to NPC2 proteins from SWISS-PROT clustered within group 2 (E-value 1.23E−20–2.91E−24), whereas C. sinensis sequences that were least similar to NPC2 proteins from SWISS-PROT were within group 1 (E-value 2.64E−10–4.65) or group 4 (E-value 6.12E−23–5.28). Interestingly, NPC2-like proteins were frequently encoded on the same genome scaffold (Fig. 2c). For example, four groups of two to six Korean NPC2-like proteins within group 4 were encoded on the same scaffold, whereas one and two groups of Korean proteins (n = 2–5) in group 3 and group 1, respectively, were encoded on the same scaffold.

Fig. 2
figure 2

Phylogenetic relationships among curated NPC2 proteins identified in two genomes of Clonorchis sinensis (a). Curated SWISS-PROT NPC2 proteins and known PDB structures for NPC2-like proteins were used as an outgroup. For each C. sinensis gene, their predicted Chinese (red) and Korean (blue) orthologues are presented as pairs in the phylogenetic tree. For each gene, the phylogenetic grouping (b) and the encoding scaffold (c) are listed. For all encoded protein sequences, the presence of a signal peptide (d), homology to canonical SWISS-PROT NPC2 proteins (e) and the presence of a conserved Pfam ML domain (PF02221) (f) are indicated. For representative pairs of orthologous proteins, the predicted quality of the I-TASSER model (C-score; “confidence score”) and the best matched PDB structure is shown (g). Transcription levels for each NPC2-like protein-encoding gene in available developmental and tissue-specific RNA-Seq libraries are included as a heatmap (h)

Full size image

Based on available C. sinensis transcriptomic sequence data, NPC2-like protein-encoding genes within group 2 showed evidence of transcription in all developmental stages (metacercaria, juvenile and adult) and adult tissues (testis, ovary, muscle and sucker) (Fig. 2h). Most NPC2-like protein-encoding genes in C. sinensis showed moderate to high transcription in the metacercarial stage. Genes in group 1 showed the lowest (overall) transcription levels across all stages and tissues. The gene Cs-k2.gene7762 (Korean isolate) showed the highest transcription overall, and was highly transcribed in all stages and tissues studied.

Employing curated data sets (Fig. 2 and Additional file 1: Table S2), the paired orthologues Cs-k2.gene6404 and csin102672 were inferred to encode canonical NPC2 proteins. Transcriptomic evidence supported their constitutive transcription in all stages and in adult tissues. In addition, the proteins encoded by these genes were the most similar to curated NPC2 proteins from SWISS-PROT, and their predicted tertiary structures had the highest C-scores.

Evidence for structural conservation

Predicted high-confidence (C-score: ≥ 1) structures of 21 C. sinensis NPC2 and NPC2-like proteins (11 and 10 for Korean and Chinese isolates, respectively) were aligned with the two most similar PDB reference structures (2HKA and 3WEA) to assess conservation (Fig. 3). Most proteins were predicted to retain a conserved Ig-like β-sandwich fold conformation with seven-stranded β-sandwich folds fixed by three disulfide bonds (Cys-8-Cys-121, Cys-23-Cys-28, and Cys-74-Cys-80) and a large cavity in the interior of a protein barrel (Fig. 3a–c). In mammals, NPC2 binds cholesterol in the deep hydrophobic tunnel created by the βa and βb-βc loops (Fig. 3c) [19]. Predicted C. sinensis NPC2 and NPC2-like models were more similar to the resolved structure in the absence of bound cholesterol sulphate (2HKA chain A; Fig. 3b, c) than the resolved structure with an open pocket in the presence of bound cholesterol sulphate (2HKA chain C; Fig. 3d). Importantly, only Cs-k2.gene6404 and csin102672 proteins retained the three amino acid residues (Val-105, Tyr-109 and Phe-73) that are required for cholesterol binding [19] (Fig. 3a, c), further supporting their annotation as canonical NPC2 proteins.

Fig. 3
figure 3

Assessment of structural conservation in Clonorchis sinensis NPC2 and NPC2-like proteins. Predicted structures of 21 C. sinensis NPC2 and NPC2-like proteins were aligned to assess conservation relative to two reference NPC2 protein structures (2HKA and 3WEA). a Alignment of NPC2 and NPC2-like sequences and conserved barrel with seven-stranded β-sandwich folds (shown in green) fixed by three disulfide bonds (Cys-8-Cys-121, Cys-23-Cys-28, and Cys-74-Cys-80; indicated by black lines below the alignment). b Positioning of loops in C. sinensis NPC2 models was modelled on a closed sterol binding pocket. Conserved residues are shown in red, variable residues are shown in blue and fewer variable residues are shown in white. c Hydrophobic tunnel from an opening created by three β-sandwich loops and highlighting three amino acid residues important for binding cholesterol (Val-105, i; Tyr-109, ii; and Phe-73, iii). d A fully open pocket in the presence of bound cholesterol sulphate (orange)

Full size image

Discussion

This study conducted comparative genomic analyses to explore the substantial expansion of a NPC2-like protein family in C. sinensis. Using a bioinformatic workflow, we curated 35 and 32 C. sinensis NPC2-like proteins representing the Korean and Chinese isolates, respectively, clustered them into four distinct phylogenetic groups, predicted their tertiary structures and recorded transcriptional levels for the genes encoding these proteins in distinct developmental stages and tissues. These data provide evidence for the presence of a structurally conserved canonical form of NPC2 in C. sinensis and were used to explore the functional implications of genetic variation among members of the C. sinensis NPC2-like protein family.

Earlier characterisations of human NPC2 homologues [20] in liver flukes revealed an expansion of this protein family [16,17,18]. The present study predicted a higher number of NPC2-like proteins in C. sinensis compared with earlier studies and provides evidence of relatively high levels of genetic conservation between the Korean and Chinese isolates. To date, investigations of other platyhelminths or most other eukaryotes have reported only one or two copies of NPC2-like proteins [20], which likely retain a conserved role in sterol transport from the late endosome and/or lysosomes, together with conserved NPC1 proteins (C. sinensis gene identifiers Cs-k2.gene5262 and csin107525) [49]. The genetic mechanisms in C. sinensis that have led to a relatively rapid expansion and genetic diversification beyond a single, canonical NPC2-like gene remain to be investigated in detail. Clustering of C. sinensis NCP2-like proteins on the same draft genomic scaffolds (for each isolate) suggests gene duplication mechanisms are likely to be involved, including unequal crossing-over [50], retrotransposition [51], duplicated DNA transposition [52] and/or polyploidisation [53]. In arthropods, gene duplication events appear to have led to a similar expansion of ML domain-containing proteins (which includes NPC2-like proteins). For example, MD-2 or NPC2-like protein family expansions in insects have been recorded in Anopheles gambiae (13 copies), Aedes aegypti (15 copies), Tribolium castaneum (8 copies) and Drosophila melanogaster (8 copies) [20, 21]. The extent of NPC2 gene duplication events in other flatworms remains to be determined. Further characterisation of NPC2-like proteins encoded in all available genomes of all members of the phylum Platyhelminthes should provide useful insights into the evolution of this family of proteins as well as their functional roles in free-living and parasitic taxa.

Molecular characterisations of several arthropod ML-proteins support neofunctionalisation arising from gene duplication, with diversified ML proteins playing crucial roles in steroid biosynthesis [54], immunity [55] and chemoreception [23,24,25,26]. For example, arthropod ML proteins can act as receptors (e.g. “pattern recognition receptors”) or co-receptors for various ligands to modulate innate immune signalling pathways [21]. In addition, several NPC2-like proteins are highly expressed in the chemosensory organs of ants [43], ticks [24] and spiders [56], where they are reported to play a key role in chemoreception, by acting as carriers of semiochemicals [23,24,25]. As C. sinensis is taxonomically and evolutionarily distinct from arthropods, the gene expansion events in arthropods and liver flukes appear to be independent as they do not share common ancestry. Therefore, it is unlikely that the function of C. sinensis NPC2-like proteins can be inferred from amino acid sequence homology. However, the predominant transcription of most NPC2-like C. sinensis proteins in the metacercarial stage does provide support for a role in chemoreception; the infective metacercarial stage is exposed to a hostile environment which requires molecular mechanisms facilitating survival, adaptation, migration and development [57], and there is increasing evidence that bile stimulates the expression of particular genes involved in these processes [58]. Importantly, chemoreception has been shown to play a criticial role in the ability of newly excysted juveniles to locate the ampulla of Vater and migrate into the biliary system [15, 59]. The abundance of transcripts of most NPC2-like protein genes in the metacercarial stage might be indicative of an important role for these proteins during excystation, initial growth and development and/or migration. Whether these NPC2-like proteins are expressed and/or linked to the chemotactic behaviour of C. sinensis warrants detailed investigation. In addition to roles in chemotaxis, a high level of transcription of some of the NPC2-encoding genes in adult tissues (including reproductive tissues and sucker) might suggest broader roles in lipid metabolism, feeding and/or reproduction [16, 20]. In other parasites, including intracellular protists [60, 61] and schistosomes [62], which cannot synthesise cholesterol, these molecules may have functions in nutrient uptake, immune evasion and/or energy storage. If C. sinensis is unable to synthesise cholesterol, as is the case for the related liver fluke O. viverrini [16], canonical NPC2 in C. sinensis might assume similar functions.

In this study, our established bioinformatics workflow [63] assisted in the prediction of a conserved set (family) of C. sinensis NPC2-like proteins. Evidence that two pairs of NPC2-like protein orthologues (Cs-k2.gene11598/csin11538 and Cs-k2.gene9726/csin101111) undergo positive selection [17] lends support for a recent expansion of this protein family. Interestingly, a small number of gene encoding these proteins (n = 4–8) were present exclusively in the Korean or Chinese isolate; whether these genes/proteins have evolved recently and are indeed isolate-specific remains to be established. Resolving the final copy numbers of NPC2-like protein genes in C. sinensis will necessitate the sequencing and assembly of complete, chromosome-contiguous genomes using ‘third-generation’ technologies [64]. This would also pave the way to detailed comparative genome analyses.

Conclusions

In the present study, we used comparative genomics and functional annotation tools to explore a novel family of NPC2-like proteins in C. sinensis. Based on our findings, we propose that these proteins are involved in binding and transportation of sterols and/or other lipids throughout the life-cycle of this parasite. The curated set of these proteins presented herein provides a solid foundation for future investigations of the molecular functions of NPC2-like proteins in C. sinensis and other liver flukes.