In addition to acting as a key tool for molecular studies, RNAi plays an important role in regulating normal gene expression. Endogenous short non-coding RNA substrates such as microRNAs (miRNAs) engage the RNAi pathway leading to gene silencing either through mRNA destruction or the suppression of protein synthesis (Chua et al. 2009; Liu et al. 2008). Five miRNAs have been identified in S. japonicum (Xue et al. 2008).

A simple model for RNAi is based on two steps, each centering on a ribonuclease enzyme. First, the trigger RNA (either dsRNA or miRNA primary transcript) is processed into a short inhibitory RNA (siRNA) by RNaseIII enzymes called Dicer and Drosha, with dsRNA binding domain (dsRBD) proteins acting as cofactors. In the second step, siRNAs are loaded into the effector protein complex called the RNA-induced silencing complex (RISC) (Pratt and Macrae, 2009). The siRNA is unwound in a strand-specific manner during RISC assembly. This single-stranded siRNA locates its cognate mRNA target by Watson–Crick base pairing. Gene silencing results from the nucleolytic degradation of the targeted mRNA by the RNaseH enzyme Argonaute (also called Slicer) (Pratt and Macrae, 2009). If the siRNA/mRNA duplex contains mismatches at the scissile site, as is often the case for miRNAs, the mRNA is not cleaved and gene silencing results from translational inhibition (Siomi and Siomi, 2009). In addition, small RNAs targeted to gene promoters can control gene expression not only through transcriptional suppression but also, in some cases, by activation (Janowski et al. 2007; Morris et al. 2008).

From gene suppression work, it is clear that schistosomes possess the molecular machinery that mediates RNAi. Furthermore, a number of potential protein participants in the RNAi pathway in schistosomes have been identified from analysis of transcriptome and genomic sequences. For instance, a single Dicer homologue (SmDicer) has been characterized from S. mansoni (Krautz-Peterson and Skelly, 2008a). At an estimated 2641 amino acids in size, it is the largest Dicer protein yet described. SmDicer contains all domains that are characteristic of metazoan dicers including an amino terminal helicase domain, DUF283, a PAZ domain, two RNAse III domains and an RNA binding domain (Krautz-Peterson and Skelly, 2008a). A Drosha gene has been identified in the S. mansoni genome which, through alternative splicing, is predicted to generate 2 protein homologues (Gomes et al. 2009). These proteins, SmDrosha1 & 2, comprise 1531 and 1577 amino acids, respectively, and possess conserved endonuclease domains and double-stranded RNA binding motifs (Gomes et al. 2009). Four Argonaute homologues have been identified (SmAgo1–4) with SmAgo 3 and 4 being splice variants (Gomes et al. 2009). Each SmAgo is ~900 amino acids in size and all possess conserved PAZ and Piwi domains (Gomes et al. 2009). Finally, analysis of the schistosome transcriptome has revealed several expressed sequence tags with homology to components of the RNAi pathway of other organisms (Verjovski-Almeida et al. 2003). These analyses permit us to derive here a draft of potential RNAi pathways in schistosomes.

Beyond the basic pathway leading to mRNA cleavage or translational inhibition, additional molecular components bear upon how efficient the process is in different organisms or tissues. Foremost among these is the ability of the RNAi effect to spread from the point of first engagement with dsRNA. In the nematode Caenorhabditis elegans RNAi is systemic (Whangbo and Hunter, 2008). The effect can spread from a site of RNA introduction to suppress a targeted gene at a different site or throughout the animal and, in some cases, in the animal's progeny (Whangbo and Hunter, 2008). Furthermore, RNAi can take effect when worms are simply soaked in solutions of dsRNA or if they are fed bacteria that express dsRNA (Conte and Mello, 2003). It is clear therefore that dsRNA can enter C. elegans from the environment and spread internally. In C. elegans, mutants that were defective for spreading of the RNAi effect were identified (and designated systemic RNAi defective (Sid)) (Winston et al. 2002). This led to the identification of the sid-1 gene that encodes a multiple membrane-spanning protein (Winston et al. 2002). This protein has been shown to be required for uptake of dsRNA in C. elegans and is expressed in all cells sensitive to systemic RNAi (Winston et al. 2002; Feinberg and Hunter, 2003). Furthermore, expression of SID-1 in Drosophila cells results in marked potentiation of dsRNA soaking-induced RNAi by facilitating dsRNA uptake from the growth medium (Shih et al. 2009). Current data suggest that SID-1 functions as an energy-independent channel capable of importing dsRNA into cells (Shih et al. 2009). RNAi triggered by environmental exposure to dsRNA has also been documented in several other invertebrates including schistosomes (Skelly et al. 2003; Skelly, 2006; Ndegwa et al. 2007). This suggests that these parasites also possess mechanisms to transport dsRNA into and between cells. An examination of the S. mansoni sequence database revealed a clear SID-1 homologue which we describe in this work.

Fig. 2 examines the variation in the level of suppression achieved in individual adult female parasites following RNAi treatment using SmAP siRNA #1 by electroporation. The level of expression of the SmAP gene in 8 individual adult female worms treated with control siRNA (Fig. 2, grey bars) is compared with SmAP gene expression in 8 parasites treated with SmAP siRNA (Fig. 2, white bars). It is clear that there is a large variation in relative SmAP gene expression (as much as two fold) between different individual control worms. While there are likewise differences in the levels of gene suppression achieved in individual worms treated with SmAP siRNA, it is clear that all SmAP siRNA treated worms exhibit robust (>80%) gene suppression.

Fig. 3 shows the level of suppression of the SmAP gene that is achieved after adult female parasites are soaked in siRNA. Greater suppression is achieved following exposure to higher levels of siRNA and a clear dose response is recorded. However the level of suppression achieved is lower than that seen when siRNAs are delivered by electroporation; soaking parasites in 2·5 μg siRNA leads to ~60% SmAP gene suppression (Fig. 3), whereas electroporating the parasites with this amount routinely yields >80% suppression (Fig. 2).

Male/female parasite pairs were electroporated with increasing amounts of siRNA while the parasites were in copula. Fig. 4 shows the level of SmAP gene suppression achieved in separated male and female parasites 8 days later. It is clear that robust suppression (>80%) is seen in all cases. However, at lower doses of siRNA (e.g. 1·25 and 0·6 μg) the level of gene suppression, while still substantial, declines in females.

Fig. 5 compares the level of SmAQP gene suppression achieved following schistosomula electroporation with siRNA either in electroporation buffer (Fig. 5, left) or in RPMI medium (Fig. 5, right). Comparable levels of suppression (both ~95%) are seen using both approaches.

An examination of the predicted S. mansoni protein database at S. mansoni genedb (http://www.genedb.org/genedb/smansoni/) for homologues of the C. elegans RNA channel SID-1 revealed a match coded on sequence Smp_152020. This sequence potentially encodes a 1018 amino acid protein, here designated SmSID-1, with a predicted size of 115 556 and pI of 8·3. The SmSID-1 sequence is shown in Fig. 6A along with an alignment of this protein with diverse members of this protein family generated using ClustalW. Conserved motifs are shaded. SmSID-1 exhibits high sequence similarity (81%) with its counterpart from S. japonicum (Genbank accession number BAH22347), moderate similarity (40%) with a human homologue (NP_060169) and lower similarity (19%) with C. elegans SID-1 (AF478687) (Fig. 6A). Amino acids 1–21 are predicted to constitute a signal peptide (¹MIRLLFAIAFAVSCVIFPTYS²¹) at http://www.cbs.dtu.dk/services/SignalP-3.0/ and this is boxed in Fig. 6A. In common with the other members of this protein family, SmSID-1 is predicted to contain 11 transmembrane domains (TM1-11, Fig. 6A) and to possess a large (here >600 amino acid) extracellular N-terminal domain.

Following exposure of a schistosomula population to dsRNA, the outcome for each parasite is not uniform and this is likely due to differences in dsRNA uptake and/or RNAi pathway activation and/or target gene expression in different individuals. For instance, following cathepsin B gene suppression in a population of schistosomula, the level of cathepsin B protein present in individuals within the population was assessed after 6 days by immunostaining using anti-cathepsin B antibodies. While most parasites (63%) lacked detectable protein, a sizable percentage (37%) stained clearly and some of these (5%) very brightly (Skelly et al. 2003). To monitor individual variability directly in gene suppression achieved following dsRNA exposure, the level of SmAP gene expression was assessed in each of eight adult worms that had been subjected, as a group, to RNAi. This work reveals some variability (up to 2 fold) in normal SmAP gene expression in control parasites as well as a remarkably robust suppression of SmAP gene expression (by 80–95%) in each worm tested. This high level of gene suppression in each individual is encouraging and may lead to a more uniform and informative phenotypic manifestation following gene suppression in adult parasites.

In an effort to simplify the RNAi protocol and minimize excessive handling of the parasites, which can damage them, parasite pairs were not separated before treatment. Instead, worms in copula were subjected to RNAi by electroporation. Eight days later the level of gene suppression achieved was assessed in now separated males and females. It is clear that both genders are highly susceptible to RNAi following this protocol and that separation of females from their male partners is not essential for efficient gene suppression. However, at the lower doses of siRNA used, females do display some diminution in the effect suggesting that at these levels their residence within the gynaecophoric canal of their male partners lessens their exposure to sufficient siRNA for maximal gene suppression.

It has been reported that dsRNA can be delivered to schistosomes by electroporation in RPMI culture medium (Morales et al. 2008). In work described here, the efficiency of gene suppression was compared when siRNA was delivered by electroporation to parasites in RPMI medium versus in commercial electroporation buffer (from Ambion Inc). Very high and comparable levels of suppression of the SmAQP target gene were observed in both cases. The use of medium makes each experiment more economical given its considerably lower cost versus electroporation buffer. It remains possible that differences in efficiency, following the use of medium versus electroporation buffer, may be revealed at lower concentrations of dsRNA or for different gene targets.

The preferred protocol for RNAi in schistosomes that is followed in this laboratory involves treating parasites with relatively low doses (2·5 μg) of gene-specific siRNA by electroporation. This routinely leads to robust target gene suppression. In work described here, the level of suppression achieved by soaking parasites in siRNA was assessed. In a Drosophila cell line cultured in medium containing dsRNAs, it was reported that longer dsRNAs are preferentially taken up versus siRNAs (Saleh et al. 2006). Here we show that schistosomes can take up siRNAs in a dose-dependent manner. However, the level of suppression achieved is substantially lower than that seen using electroporation. Approximately 55% suppression of SmAP gene expression is achieved after soaking parasites in 2·5 μg siRNA versus >80% following electroporation. Electroporation likely delivers dsRNA more efficiently into the body of the parasite. Since schistosomes can take up siRNA by soaking this suggests that the parasites in the blood stream will be amenable to RNAi if they are exposed to a sufficiently high dsRNA dose. It has previously been shown that gene expression in schistosomes can be efficiently suppressed by soaking the parasites in long dsRNA (Ndegwa et al. 2007; Skelly et al. 2003). Long dsRNA cannot be used in vivo to suppress gene expression in intravascular parasites since nucleases in the blood are known to rapidly degrade any introduced long dsRNAs (Aagaard and Rossi, 2007). Additionally, long dsRNAs are rapid inducers of a systemic/nonspecific interferon response when delivered in vivo, resulting in cell apoptosis (Aigner, 2006). This implies that siRNAs will need to be used for use with blood-dwelling schistosomes and it was recently reported that injecting siRNAs into schistosome-infected mice could suppress gene expression in parasites in the blood (Pereira et al. 2008). Three siRNAs targeting the parasite hypoxanthine–guanine phosphoribosyltransferase (HGPRTase) gene were injected in phosphate buffered saline intravenously into schistosome-infected mice and 6 days later, parasites were recovered. The total number of parasites was reduced by ~27% after treatment and the level of HGPRTase gene expression was significantly reduced in the HGPRTase siRNA-treated group versus controls (Pereira et al. 2008).

By scanning the available schistosome DNA sequence databases, a number of groups have identified homologues of proteins in schistosomes that are known to be involved in the RNAi pathway in other organisms (Gomes et al. 2009; Lendner et al. 2008). This, coupled with our own work, has permitted the generation of a first draft of RNAi pathways in schistosomes that are represented in Fig. 7. The pathway engaged following the uptake of exogenous dsRNA is shown in Fig. 7A while that engaged by endogenous dsRNA is shown in Fig. 7B.

Exogenous dsRNA that is taken up (Fig. 7A (i)) engages SmDicer and an RNA-binding protein (Partner Dicer) (Fig. 7A (ii)). This interaction leads to dsRNA cleavage and siRNA generation. The siRNA duplex is unwound and one strand is preferentially loaded into the short interfering RNA induced silencing complex (siRISC) (Fig. 7A (iii)). The minimal RISC, sufficient for target RNA recognition and cleavage is simply an Argonaute protein bound to a small RNA (Rivas et al. 2005). However Argonaute proteins can have several associated binding partners (Pratt and Macrae, 2009). Here we postulate that in a schistosome RISC, Argonaute (SmAgo) associates with homologues of the RNA binding protein Fmr1 and the nuclease Tudor-SN (Fig. 7A (iii)); the S. mansoni genome is reported to contain homologues of these proteins and they exist in other invertebrate RISCs (Gomes et al. 2009; Lendner et al. 2008). The RISC, now loaded with a single stranded RNA (called the guide RNA), searches the transcriptome to find potential target mRNAs. Targets are engaged by Watson-Crick base pairing which directs the Argonaute endonuclease (SmAgo) to cleave them (Fig. 7A (iv)) (Siomi and Siomi, 2009). Many rounds of mRNA cleavage result in substantial suppression of target gene expression and effectively results in gene silencing.

In the case of the microRNA pathway (Fig. 7B), first, primary precursor miRNA transcripts (pri-miRNAs), generated in the nucleus, are cleaved by the RNaseIII SmDrosha (in association with the RNA binding protein Partner Drosha) (Fig. 7B (i)). These pri-miRNAs, containing stem-loop structures depicted in Fig. 7B, harbour the mature miRNAs in the 5′ or 3′ half of the stem. The Drosha-cleaved product is generally 60–70 nucleotides and is called a precursor miRNA (pre-miRNA). This pre-miRNA is then exported, via the carrier protein Exportin 5, to the cytoplasm. Here the SmDicer/Partner Dicer complex may engage it and cleave it to generate a mature miRNA (Fig. 7B (ii)). Next, one strand of the miRNA is loaded into the microRNA induced silencing complex (miRISC) whose composition may be similar, as depicted in Fig. 7B (iii), or different from the postulated siRISC. In other systems different proteins, including distinct Argonautes, comprise different RISCs (Okamura et al. 2004) and this may be the case for schistosomes too since miRNA engagement often results in gene suppression via translational repression (Fig. 7B (iv)) as opposed to target mRNA cleavage. In the S. mansoni genome 4 Argonaute homologues have been identified (Gomes et al. 2009) and it is possible that different forms occupy distinct RISCs, or perform other functions within the cell. It is noteworthy that of the 4 Argonaute proteins in humans only Ago2 is a catalytically active slicer (Liu et al. 2004).