RNA interference (RNAi) is a widespread and widely exploited phenomenon. Here, we show that changing inositol 1,4,5‐trisphosphate (IP3) signalling alters RNAi sensitivity in Caenorhabditis elegans. Reducing IP3 signalling enhances sensitivity to RNAi in a broad range of genes and tissues. Conversely up‐regulating IP3 signalling decreases sensitivity. Tissue‐specific rescue experiments suggest IP3 functions in the intestine. We also exploit IP3 signalling mutants to further enhance the sensitivity of RNAi hypersensitive strains. These results demonstrate that conserved cell signalling pathways can modify RNAi responses, implying that RNAi responses may be influenced by an animal's physiology or environment.
Reducing intracellular IP3 signaling enhances RNAi sensitivity in C. elegans through a function of IP3 in the intestine. This suggests that an animal's physiology or environment may influence its RNAi responses.
Altering intracellular IP3 signalling modifies sensitivity to RNAi in different tissues.
Tissue‐specific rescue experiments suggest IP3 functions in the intestine to control RNAi sensitivity throughout the animal.
IP3 signalling mutations can further enhance the sensitivity of existing RNAi hypersensitive strains.
RNA interference (RNAi) is a widespread and widely exploited phenomenon which has potential as a strategy for both the treatment of disease and pest control. RNAi results in down‐regulation of a specific gene in response to the production of small interfering RNAs (siRNAs). RNAi is one of a family of processes mediated by small non‐coding RNAs , . In Caenorhabditis elegans, and in a number of other organisms, RNAi is systemic so that the introduction of dsRNA into one tissue triggers gene silencing in other tissues , , , , . Furthermore, systemic RNAi enables C. elegans and other organisms to exhibit environmental RNAi . For example, feeding C. elegans on bacteria expressing dsRNA initiates a widespread RNAi response , . Studies in C. elegans and other organisms have provided mechanistic insights into RNAi , , , , , although the role of exogenous RNAi in the normal life of C. elegans and other animals remains unclear .
Whilst C. elegans mounts a robust and widespread RNAi response for many genes, it is clear that its sensitivity to RNAi can be modified. Enhancers of RNAi such as rrf‐3  and eri‐1  are believed to act cell autonomously and are thought to enhance exogenous RNAi by releasing components that are normally shared between the exogenous and endogenous RNAi pathways . A group of retinoblastoma (Rb) pathway genes also give rise to enhanced sensitivity when mutated , . These genes and others  appear to influence RNAi sensitivity through core sncRNA pathways, gene regulatory mechanisms or RNA transport. Whether the response can also be altered by the broader physiological state of an animal remains unclear. Some hints that this may be the case come from observations that environmental or other factors may influence RNAi . For example, temperature can affect the RNAi response to particular genes in certain backgrounds , , and it is a common observation that temperature can affect the results of RNAi experiments.
Inositol 1,4,5‐trisphosphate (IP3) is an important second messenger in animals. IP3 is generated by the action of phospholipase C (PLC) in response to a diverse range of extracellular stimuli, including neurotransmitters and hormones acting on G‐protein or tyrosine kinase‐coupled receptors (Fig 1A). IP3 production leads to Ca2+ release from the endoplasmic reticulum (ER) through a ligand‐gated ion channel receptor, the IP3R, which regulates a wide range of processes in animals including C. elegans . Here, we show that reducing or increasing IP3 signalling enhances or suppresses the sensitivity of C. elegans to RNAi in a broad range of genes and tissues. Tissue‐specific rescue suggests that IP3 signalling acts non‐cell autonomously and that it acts in the intestine. Our results imply that an animal's exogenous RNAi response may be influenced by its physiology or environment.
Results and Discussion
IP3 receptor mutants have enhanced RNAi sensitivity
In C. elegans, IP3 receptors (IP3Rs) are encoded by a single gene itr‐1 (Fig 1A) , , . Whilst investigating itr‐1, we observed that an unexpectedly large number of genes showed apparent interactions with itr‐1 in RNAi experiments. This led us to hypothesise that itr‐1 reduction‐of‐function mutants have enhanced RNAi responses. To test this, we selected a group of RNAi targets, which had been reported to be refractory to RNAi in wild‐type (WT) worms but sensitive to RNAi in hypersensitive strains , . We selected genes that act in a range of processes and tissues, including genes that cause embryonic lethality (apr‐1, qua‐1 and hmr‐1), sterility (arf‐3 and mys‐1), defects in vulval development (lin‐1 and lin‐31), a dumpy phenotype (dpy‐13) or neuronal Unc phenotypes (Table 1). RNAi was performed by feeding in an itr‐1 temperature‐sensitive allele itr‐1(sa73) (Table 1). Assays were carried out at 20°C at which temperature itr‐1(sa73) exhibits a partial reduction‐of‐function phenotype but is reasonably healthy. Other widely used RNAi‐sensitive strains were also tested (Table 1). We found that itr‐1(sa73) worms generally showed stronger and more penetrant RNAi phenotypes than wild‐type animals (Table 1, Fig 1B). In comparison with the RNAi‐sensitive strain rrf‐3, itr‐1(sa73) often showed similar sensitivity (e.g. qua‐1 RNAi caused 23.3 ± 12% lethality in itr‐1 and 22.9 ± 2.5% in rrf‐3) but often showed less enhancement than eri‐1; lin‐15b worms (e.g. lin‐31 RNAi caused 22.8 ± 0.6 multivulval worms in itr‐1 and 56.1 ± 6.1% in eri‐1; lin‐15b). Thus, itr‐1(sa73) animals show a broad enhancement of the RNAi effect in a range of target tissues including the nervous system.
In our standard assay, adult worms are placed on RNAi plates and allowed to lay eggs, which subsequently develop on the same plate. We see the same effect within a single generation (Fig 1C), that is the effect is independent of any generational effects. itr‐1(sa73) animals are slow growing and constipated . These phenotypes might increase RNAi by increasing exposure to dsRNA. Analysis of RNAi sensitivity in animals carrying mutations which cause itr‐1‐like phenotypes in defecation (kqt‐3)  and growth (dbl‐1) (Supplementary Fig S1) showed that these do not result in increased sensitivity to lin‐1 RNAi (Fig 1B).
To test whether the RNAi sensitivity of itr‐1 mutants correlates with the degree of reduction in itr‐1 function, we tested other alleles of itr‐1. We used lin‐1 as a standard test RNAi target. lin‐1 is involved in vulval development, and depletion causes a multivulval (Muv) phenotype . itr‐1 is not known to be involved in vulval development so that any effect of IP3 signalling on the lin‐1 Muv phenotype should be independent of vulval development. itr‐1(sa73) worms show a significantly stronger RNAi response to lin‐1 RNAi than wild‐type animals (Fig 1B and D). Two putative null or near null alleles, itr‐1(tm902) and itr‐1(n2559), showed significantly stronger RNAi sensitivity than itr‐1(sa73). Expression of a genomic itr‐1(+) transgene in itr‐1(sa73) restores normal RNAi sensitivity (Fig 1D). Thus, itr‐1 mutants show increased sensitivity to RNAi, which is proportional to the degree of itr‐1 function.
Increases in RNAi phenotypes in itr‐1 mutants result from reduced target gene expression
To confirm that the increased sensitivity of itr‐1 mutants did indeed result from reduction in gene expression, we used two approaches. First, we used qRT–PCR to demonstrate that RNAi of dpy‐13 causes a further reduction in mRNA levels in itr‐1 and, as a positive control, eri‐1;lin‐15b mutants compared to that seen in wild‐type animals (Fig 1E). Secondly, we used a direct read‐out of gene expression by performing RNAi of GFP in animals carrying GFP markers. In one system an unc‐47p::GFP transgenic reporter expressed in GABAergic neurons  was used to test knock‐down in the nervous system. Wild‐type animals expressing this construct show very little reduction in GFP fluorescence after feeding of GFP dsRNA, whereas in eri‐1(mg366) animals, the number of fluorescent neurons is significantly reduced after GFP RNAi  (Fig 1F and J). itr‐1(sa73); unc‐47p::GFP animals showed a similar reduction in response to GFP RNAi (Fig 1F and J). In a second system (Fig 2A), we measured RNAi of GFP expressed in the body wall muscles using a myo‐3p::GFP construct , . Again itr‐1 mutants show an increased effect (Fig 1I).
PLC‐β/EGL‐8 mutants show increased RNAi responses
IP3 is produced from phosphatidylinositol 4,5‐bisphosphate (PIP2) by a family of phospholipase C enzymes which are activated by cell surface receptors (Fig 1A). To investigate whether ITR‐1 modifies RNAi through a canonical IP3‐mediated pathway, we tested whether phospholipase C (PLC) also interacts with the RNAi pathway. We tested mutants of each of the five C. elegans PLC genes  for increased sensitivity to lin‐1 dsRNA (Fig 1G). Only loss of PLC‐β, egl‐8 resulted in a significant increase in multivulval animals. Three different alleles of egl‐8 showed increased RNAi sensitivity (Fig 1G) with the strongest phenotype in the putative null allele egl‐8(e2917), and lesser effects in two partial loss‐of‐function mutants . egl‐8(e2917) animals show increased sensitivity to a wide range of genes (Table 1). egl‐8(e2917) animals also showed significant increases in sensitivity in both the unc‐47p::GFP reporter (Fig 1F) and body wall muscle GFP systems (Fig 1I). Thus, egl‐8 also modulates RNAi sensitivity. The level of reduction in the null allele egl‐8(e2917) is not as severe as that in itr‐1 null alleles (compare Fig 1G and D); thus, other PLCs may be compensating for the loss of EGL‐8. RNAi tests of the remaining PLCs using lin‐1 in an egl‐8 background revealed that only knock‐down of PLC‐γ, plc‐3 enhanced the RNAi response further. We therefore tested plc‐3(tm753); egl‐8(n488) worms and found that they also have increased RNAi sensitivity over either single mutant (Fig 1G). Thus, plc‐3 is able to compensate for the loss of egl‐8 in RNAi sensitivity. PLC‐β is usually activated by heterotrimeric G‐proteins acting downstream of GPCRs . Thus, signalling through a GPCR may be important to the alterations in RNAi sensitivity.
Increased IP3 signalling causes RNAi resistance
To ascertain whether IP3 signalling is capable of modulating the RNAi response in both directions, we tested whether increasing IP3 signalling could reduce RNAi sensitivity. Initially, we increased expression of the IP3R by introducing transgenes carrying the whole itr‐1 gene into wild‐type worms. Such transgenes tend to be toxic when introduced at high level, and thus, overexpression is likely to be modest. To test for reduced RNAi, we used RNAi of unc‐15 by feeding, which produces an intermediate phenotype in wild‐type worms (Fig 1H). Wild‐type worms carrying extra itr‐1 genes show reduced expression of the unc‐15 phenotype. To confirm this result, we used ipp‐5 mutants. IP3 is metabolised to IP2 and IP4 by the enzymes inositol polyphosphate 5‐phosphatase encoded by ipp‐5 and IP3 3‐kinase encoded by lfe‐2, respectively (Fig 1A). ipp‐5 loss‐of‐function mutant animals are therefore assumed to have increased IP3 levels (see  for discussion). ipp‐5 mutants show a substantially reduced response to unc‐15 RNAi (Fig 1H). lfe‐2 mutants which may also have increased IP3 levels in some tissues did not change sensitivity to unc‐15 RNAi. To test whether reduced sensitivity in ipp‐5 mutants was due to changes in gene expression, we used the body wall muscle GFP reporter and showed reduced knock‐down in ipp‐5(sy605) animals (Fig 1I). Thus, increased IP3 signalling leads to decreased RNAi sensitivity, and decreased IP3 signalling causes increased sensitivity demonstrating that IP3 signalling is able to modulate RNAi sensitivity in worms. We note that we and others have used RNAi in IP3 signalling mutants to dissect IP3‐mediated signalling pathways . The results of such studies should now be reviewed in the light of these results.
IP3 signalling modifies RNAi induced by internally produced dsRNA
Widespread RNAi induced, as above, by the feeding of bacteria carrying dsRNA consists of a number of steps in which itr‐1 might function. First dsRNA is absorbed from the environment through the intestine. Next, the RNAi signal is transported between cells through a process requiring the production, export and import of the RNAi signal. Finally, cell autonomous processes leading to mRNA destruction are required.
To narrow down the step at which itr‐1 functions, we asked whether sensitivity to RNAi induced by dsRNA introduced by other methods was also altered. We used a system designed to assay RNAi spreading (Fig 2A)  in which GFP is expressed in the pharynx (myo‐2p::GFP) and body wall muscle (bwm) (myo‐3p::GFP). RNAi is then induced by expressing sense and antisense RNA for GFP in the pharynx (myo‐2p::dsRNAGFP) . Thus, knock‐down of GFP in the pharynx is, presumably, primarily cell autonomous whilst knock‐down in the body wall muscle requires RNAi spreading. We observed increased knock‐down in itr‐1 mutants in both the pharynx and bwm (Fig 2B). Similarly, ipp‐5 mutants show decreased sensitivity in both the pharynx and bwm (Fig 2C). Thus, itr‐1 mutants show increased sensitivity to internally induced RNAi.
itr‐1 acts in the intestine to modify RNAi responses
We sought to further clarify the mechanism of IP3 action by investigating the site of action at a tissue level. We used tissue‐specific promoters to express an itr‐1 cDNA that had previously been shown to rescue other phenotypes in both neurons and the intestine (Ford, Peterkin and Baylis, unpublished data). Using the unc‐47p::GFP system and RNAi by feeding, we tested for the ability of itr‐1 to restore normal sensitivity in the target cells. Expression of itr‐1 in unc‐47‐expressing neurons (unc‐47p::itr‐1) or in the nervous system in general (unc‐119p::itr‐1) failed to restore normal RNAi sensitivity (Fig 3A). Whilst this suggests that itr‐1 does not act in the target cells, we cannot exclude other possibilities such as insufficient expression, although the ability of the unc‐119p::itr‐1 construct to rescue other neuronal phenotypes makes this less unlikely. In contrast, expression from a well‐characterised intestine‐specific promoter, vha‐6p , was able to restore normal sensitivity (Fig 3A). Furthermore, vha‐6p::itr‐1 partially rescued sensitivity to lin‐1 RNAi by feeding (Fig 3B). In both of these experiments, we induced RNAi by feeding, and thus, itr‐1 might function in the gut to improve dsRNA uptake. We therefore tested whether expression in the intestine could rescue sensitivity induced in response to an internal dsRNA trigger. We used the GFP pharynx and bwm system in an itr‐1 background. As shown above, itr‐1 mutants show increased sensitivity in both the pharynx (producing tissue) and bwm (non‐producing tissue) (Fig 2B and C). Expression of itr‐1 in the intestine restores normal sensitivity in both tissues (Fig 3C).
Thus, IP3 signalling in the intestine appears to play a role in the modulation of RNAi sensitivity. Since intestinal itr‐1 function influences RNAi induced by either external or internal dsRNA, it seems unlikely that this is exclusively due to modified uptake from the gut although we cannot exclude such a role. For example, even in the internal dsRNA system, it is possible that dsRNA is released into the environment and subsequently taken up by other worms. In the internal dsRNA system, RNAi responses in both producing and target cells are increased in itr‐1 mutants. This suggests that IP3 signalling is not altering spreading, although again we cannot exclude an indirect route to RNAi in the producing cells. Overall, therefore we consider the most likely explanation for our data is that IP3 signalling in the intestine is involved in the production or release of some humoral signal, which then modifies RNAi responses in cells elsewhere in the body.
Using IP3 signalling mutants to produce further increases in RNAi sensitivity
RNAi‐sensitive strains have been important tools in the analysis of gene function. In some cases, sensitivity can be increased in an additive fashion by combining mutations, notably in the example of eri‐1 and lin‐15B and similar double mutant strains , , . We therefore tested whether itr‐1 is able to further enhance hypersensitivity in eri‐1 mutants. itr‐1(sa73); eri‐1(mg366) double mutants show an increased sensitivity to both lin‐1 and GFP dsRNA (Fig 4A and B). We therefore attempted to make a strain with further increased sensitivity over currently available sensitised strains. eri‐1(mg366); lin‐15b(n744) strains are commonly used in RNAi screens. Since itr‐1 mutants have pleiotropic phenotypes, we used egl‐8(e2917) as animals carrying this allele are relatively healthy. We produced two individual isolates of a strain with the genotype eri‐1(mg366); lin‐15b(n744); egl‐8(e2917), HB946 and HB947. Tests of these strains using lin‐1 and lin‐31 show that they have higher sensitivity than the eri‐1(mg366); lin‐15b(n744) strain (Fig 4C). Use of this strain may be advantageous in RNAi screening experiments, although it carries mutations in three pathways and would need to be used with care. These results also suggest that itr‐1 functions through a different mechanism than either eri‐1 or lin‐15B.
Our results provide the first clear example of a signal transduction pathway acting in the regulation of RNAi. This discovery raises the possibility that an animal's response to exogenous dsRNA may be modified by changes in the animal's environment or internal physiology as reflected in intercellular signals which require intracellular IP3 signalling. The discoveries that itr‐1 mutants strongly enhance the RNAi response and that this sensitivity can be further improved by combining itr‐1 with other RNAi‐sensitive mutations may prove useful in the application of RNAi to the treatment of disease and the control of pests, including other nematodes. For example, RNAi‐based strategies to treat human disease are currently being explored but at therapeutic doses tissue accessibility varies. For example, RNAi can efficiently reduce the function of a liver‐derived enzyme involved in cholesterol synthesis in hepatocytes , . In contrast, neurons are relatively inaccessible to nucleic acids hampering attempts to modify neurodegenerative diseases. The experiments presented in this work show that the IP3/calcium signalling pathway enables RNAi in refractory tissues. This pathway is conserved from invertebrates to humans, and therefore, our results may have relevance to developing methods of RNAi intervention in difficult‐to‐reach tissues in humans and other animals.
Materials and Methods
Detailed methods are given in the Supplementary Methods.
C. elegans culture and strains
RNAi induced by exogenous dsRNA
RNAi by feeding  was carried out using bacterial RNAi feeding strains from the Ahringer library . Adult animals were placed on plates seeded with bacteria expressing dsRNA and allowed to lay eggs for between 2 and 6 h before removal. The resulting progeny were scored for the relevant phenotypes.
RNAi induced by endogenous dsRNA, fluorescent microscopy and image analysis
We used a system similar to that developed by Hunter and colleagues in which GFP reporters are present in the pharynx and body wall muscle of the animals . L4 animals were imaged. Image collection was optimised independently for the pharynx and body wall muscle. In the case of the ipp‐5 experiments, worms were synchronised and starved to increase the RNAi effect  and thus the range of detection for resistance.
AIN, RPV‐M, RG and HAB designed experiments. AEH, AIN, CPC, HAB, MB, MDS, MH, ML, RG and RPV‐M performed experiments. AIN, HAB and RPV‐M wrote the manuscript. HAB instigated and oversaw project.
Conflict of interest
The authors declare that they have no conflict of interest.
We thank A. Fire, K. Ford, S. Mitani and H. Peterkin for the provision of plasmids and strains. Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). Other strains were provided by the Mitani Lab through the National Bio‐Resource Project of the MEXT, Japan. We are grateful to J. Ahringer, B. Olofsson and members of the Baylis group for helpful discussions. AIN was funded by Trinity Hall College, Cambridge and the Cambridge European Trust. The work of MDS and RPV‐M was partially funded by a Miguel Servet Grant (CP11/00090) from the Health Research Institute Carlos III, which is partially supported by the European Regional Development Fund. RPV‐M is a Marie Curie fellow (CIG322034). RG was funded by the MRC (G0601106).
FundingNIH Office of Research Infrastructure Programs P40 OD010440
- © 2015 The Authors