bs_bs_banner
Insect Molecular Biology (2014) 23(2), 175–184
Insect Molecular Biology doi: 10.1111/imb.12074
dsRNA uptake and persistence account for tissue-dependent susceptibility to RNA interference in the migratory locust, Locusta migratoria
D. Ren*, Z. Cai*, J. Song*, Z. Wu† and S. Zhou* *State Key Laboratory of Integrated Management of Pest Insects and Rodents, Institute of Zoology, Chinese Academy of Sciences, Beijing, China; †School of Life Sciences, University of Science and Technology of China, Hefei, China Abstract RNA interference (RNAi) by introducing doublestranded RNA (dsRNA) is a powerful approach to the analysis of gene function in insects; however, RNAi responses vary dramatically in different insect species and tissues, and the underlying mechanisms remain poorly understood. The migratory locust, a destructive insect pest and a hemimetabolic insect with panoistic ovaries, is considered to be a highly susceptible species to RNAi via dsRNA injection, but its ovary appears to be completely insensitive. In the present study, we showed that dsRNA persisted only briefly in locust haemolymph. The ovariole sheath was permeable to dsRNA, but injected dsRNA was not present in the follicle cells and oocytes. The lack of dsRNA uptake into the follicle cells and oocytes is likely to be the primary factor that contributes to the ineffective RNAi response in locust ovaries. These observations provide insights into tissue-dependent variability of RNAi and help in achieving successful gene silencing in insensitive tissues. Keywords: RNAi, dsRNA, siRNA, ovary, migratory locust. Introduction RNA interference (RNAi) triggered by target specific double-stranded RNA (dsRNA) has been employed as a First published online 6 December 2013. Correspondence: Shutang Zhou, Institute of Zoology, Chinese Academy of Sciences, 1 Beichen West Road, Beijing 100101, China. Tel.: 86-1064806293; fax: 86 10 64807099; e-mail:
[email protected]
© 2013 The Royal Entomological Society
powerful reverse-genetic tool in the study of gene function in insects, particularly the insects in which the mutation is unavailable (Bellés, 2010; Terenius et al., 2011; Scott et al., 2013). RNAi has also been proposed as a potential approach to insect pest control (Baum et al., 2007; Mao et al., 2007; Price & Gatehouse, 2008); however, although the response to RNAi has been reported in the orders of Hymenoptera, Diptera, Coleoptera, Lepidoptera, Hemiptera, Dictyoptera, Isoptera and Orthoptera, the sensitivity to RNAi varies dramatically among insect species and tissues (Bellés, 2010; Terenius et al., 2011). The fruit fly Drosophila melanogaster has been found to be relatively less sensitive to RNAi as gene silencing is solely observed in its haemocytes after injection of dsRNA into the larvae (Miller et al., 2008; Bellés, 2010). Low susceptibility to RNAi has also been demonstrated in many lepidopteran insects, including the silkworm Bombyx mori and tobacco hornworm Manduca sexta (Terenius et al., 2011). Limited success of RNAi by injection or feeding of dsRNA or small interfering RNA (siRNA) has been reported in the honey bee Apis mellifera (Jarosch & Moritz, 2011; Jarosch et al., 2011; El Hassani et al., 2012). In contrast, injection of dsRNA into the larval body cavity leads to gene knockdown in all tissues tested in the beetle Tribolium castaneum (Tomoyasu & Denell, 2004; Miller et al., 2008, 2012; Bai et al., 2011; Rulifson et al., 2013). In the German cockroach Blattella germanica, a robust response to RNAi has been observed in all developmental stages including moulting, metamorphosis and oogenesis (Bellés, 2010; Lozano & Bellés, 2011; Huang et al., 2013; Irles et al., 2013). Tissue-dependent sensitivity to RNAi has been observed in a number of insect species. In the mosquito Anopheles gambiae, salivary glands show a relatively lower response to RNAi than other tissues (Boisson et al., 2006), while in the mosquito Aedes aegypti, the head and ovary appear to be less responsive to RNAi compared with the fat body, midgut and abdomen (Telang et al., 2013). In lepidopteran species, more success of RNAi and higher knockdown efficiency have been 175
176
D. Ren et al.
seen in pheromone glands and silk glands, followed by the fat body, haemocytes, midgut and brain, whereas wing discs and larval epidermis seem to be insensitive (Terenius et al., 2011). In the honey bee A. mellifera, injection of dsRNA in adult workers has a significant effect on target gene expression in fat bodies but not in ovaries (Jarosch & Moritz, 2011); however, in another study, feeding with siRNA specific to an isoform of gene gemini transcripts is able to disrupt its expression and leads to activation of worker bee ovaries (Jarosch et al., 2011). In the desert locust Schistocerca gregaria, significant knockdown efficiency has been observed in the fat body, brain, optic lobes, suboesophageal ganglion, prothoracic glands and malpighian tubules by injection of dsRNA into the hemocoel (Badisco et al., 2011; Marchal et al., 2011, 2012; Tobback et al., 2012; Van Wielendaele et al., 2013). The female and male reproduction systems, particularly the ovaries of desert locusts are less susceptible to dsRNA injection (Wynant et al., 2012). Despite these observations of variability of RNAi sensitivity among insect tissues, the underlying molecular and cellular mechanisms have not been clearly revealed. The migratory locust Locusta migratoria, is one of the most destructive agricultural insect pests (Lomer et al., 2001). It is also a representative of evolutionarily primitive insects with incomplete metamorphosis and with panoistic ovaries (Wyatt & Davey, 1996). Previous studies have shown that injection of dsRNA in the hemocoel leads to efficient gene knockdown in locust fat body, head, brain and legs (Guo et al., 2011; Ma et al., 2011; Luo et al., 2012; Wu et al., 2012), but not in the ovary (Song et al., 2013). In the present study, we performed multiple experiments to determine the factors that contribute to the ineffective RNAi response in locust ovary. We found that dsRNA persists only briefly in locust haemolymph. Confocal microscopy showed that the ovariole sheath is permeable to dsRNA and siRNA, but injected dsRNA and siRNA are not present in the follicle cells and oocytes. Furthermore, we investigated the possible involvement of SID-1 and Clathrin heavy chain in RNAi. We propose that the ineffective RNAi response in locust ovary is attributable to inefficient dsRNA uptake by follicle cells and oocytes. Results Locust ovary is completely insensitive to injected dsRNA Previously, we have reported that Dicer1 and Argonaute1 could not be disrupted in locust ovaries via dsRNA injection (Song et al., 2013). We speculated that the insensitivity of locust ovaries to RNAi could be gene-dependent. We performed further studies using multiple genes with different functions and expression patterns. Two of those
genes were Methoprene-tolerant (LmMet; GenBank: KF377825), which is ubiquitously expressed, and vitellogenin receptor (LmVgR; GenBank: KF377827), which is specifically expressed in the ovary. The fat body was used as the control tissue as it plays a crucial role in female locust reproduction (Wyatt & Davey, 1996) and is highly susceptible to RNAi by dsRNA injection (Song et al., 2013). Quantitative PCR (qPCR) demonstrated that LmMet was continuously expressed in the fat body and ovary of female adults after eclosion, and LmMet transcripts were more abundant in the fat body at 0, 2, 8 days after eclosion (Fig. 1A). Injection of LmMet dsRNA (dsMet) resulted in an average of 72% knockdown of LmMet expression in the fat body (Fig. 1B), but LmMet mRNA levels had no apparent change in the ovary (Fig. 1C). LmVgR was continuously expressed in ovaries during oocyte maturation, though its mRNA levels fluctuated at different stages (Fig. 1D). Injection of LmVgR dsRNA (dsVgR) did not alter LmVgR expression in the ovary (Fig. 1E).
Injected dsRNA and siRNA are not taken up by follicle cells and oocytes To trace the fate of injected dsRNA in the fat body and ovary, 421-bp dsMet was 5’-labelled with Cy3 dye (Cy3dsMet) and intra-abdominally injected into female adults at 0 or 4 days after eclosion. Meanwhile, a 21-bp siRNA within the region of dsMet was 5′-labelled with TAMRA dye (TAMRA-siMet) and injected separately for parallel studies. The locust ovary is panostic, i.e. has no nurse cells. The follicle cells and oocytes were therefore selected for laser confocal microscopy in addition to the fat body. To facilitate the visualization of fluorescent signals of Cy3-dsMet and TAMRA-siMet in individual cells, nuclei and F-actin were stained simultaneously. Injection of nuclease-free H2O, which was used as the control, did not yield a positive signal except the background emitted from the cell itself. The fluorescent signals of Cy3-dsMet and TAMRA-siRNA were clearly seen in the fat body cells 24 h after injection (Fig. 2). In contrast, neither follicle cells nor oocytes had a detectable fluorescent signal of Cy3dsMet and TAMRA-siMet (Fig. 2). During the first gonadotrophic cycle of female locusts, vitellogenesis starts from ∼5 days after eclosion. In the vitellogenesis phase, a large amount of Vg is synthesized in the fat body and entered into maturing oocytes through the intercellular spaces termed ‘patency’ in the follicular epithelium and by receptor-mediated endocytosis at oocyte membranes (Raikhel & Dhadiallal, 1992; Wyatt & Davey, 1996). To assess whether the patency promotes the uptake of dsRNA or siRNA into oocytes and follicle cells, Cy3-dsMet and TAMRA-siMet were injected into 6-day-old female adults for further analyses. Fluorescent © 2013 The Royal Entomological Society, 23, 175–184
Tissue-dependent RNAi susceptibility
177
Figure 1. Expression and knockdown of LmMet and LmVgR. (A) LmMet expression profiles in the fat body (Fb) and ovary (Ov). **, P < 0.01 for fat body vs ovary. (B) LmMet knockdown efficiency in the fat body after injection of LmMet double-stranded (ds)RNA (dsMet). **, P < 0.01 compared with GFP dsRNA (dsGFP) control. (C) LmMet mRNA levels in the ovary after dsMet injection. (D) LmVgR expression profiles in the ovary. (E) LmVgR mRNA levels in the ovary after injection of LmVgR dsRNA (dsVgR). Values are mean ± SE; n = 8–12.
signals of Cy3-dsMet and TAMRA-siMet were again observed in the fat body cells (Fig. 3), but not in the follicle cells and oocytes (Fig. 4). Taken together, these observations suggest that dsRNA and siRNA are unable to enter
into the follicle cells and oocytes, irrespective of the developmental stages.
The ovariole sheath is permeable to dsRNA To determine whether the ovariole sheath functions as the barrier to prevent dsRNA entry, we established the explant culture of ovarioles which were incubated with Cy3-dsMet. The ovariole sheaths were either maintained or removed. A layer of fluorescent signals of Cy3-dsMet was clearly seen on the outside of follicular epithelium in both sheath-covered ovarioles and sheath-removed ovarioles (Fig. 5). The result suggests that the ovariole sheath is permeable to dsRNA. As expected, the fluorescent signals of Cy3-dsMet were observed in the fat body cells when the fat body was incubated with Cy3-dsMet (Fig. 5).
Figure 2. Uptake of double-stranded (ds)RNA and small interfering (si)RNA in the fat body vs ovary at the previtellogenic stage. Cy3-labelled LmMet dsRNA (Cy3-dsMet), TAMRA-labelled LmMet siRNA (TAMRA-siMet) or H2O (control) were separately injected into females adults at 0 or 4 days after eclosion. Fat bodies and ovarioles were collected and further stained with Hoechst 33342 for nuclei (blue) and with Phalloidin-Alexa Fluor 488 for F-actin (green). Arrow heads indicate Cy3-dsMet or TAMRA-siMet (red). Scale bars: 10 μm for fat body cells, 5 μm for follicle cells, and 20 μm for primary ooctyes.
© 2013 The Royal Entomological Society, 23, 175–184
Figure 3. Uptake of dsRNA and siRNA in the fat body at the vitellogenic stage. Cy3-labelled LmMet dsRNA (Cy3-dsMet), TAMRA-labelled LmMet siRNA (TAMRA-siMet) or H2O (control) were separately injected into 6-day-old adult females. The fat bodies were collected and further stained with Hoechst 33342 for nuclei (blue) and with Phalloidin-Alexa Fluor 488 for F-actin (green). Arrow heads indicate Cy3-dsMet or TAMRA-siMet (red). Scale bars = 20 μm.
178
D. Ren et al.
Figure 4. Double-stranded (ds)RNA and small interfering (si)RNA uptake in the ovary at the vitellogenic stage. Nuclei were stained with Hoechst 33342 (blue) and F-actin stained with Phalloidin-Alexa Fluor 488 (green). Note: Cy3-labelled LmMet dsRNA (Cy3-dsMet) or TAMRA-labelled LmMet siRNA (TAMRA-siMet) was not detected in follicle cells and oocytes. Control, H2O. Arrows indicate patency. Black and white images showing oocyte integrity and yolk granules. Scale bars: 10 μm for follicle cells and 50 μm for primary ooctyes.
SID-1 is not required for RNAi in female adult locusts
heavy chain (LmChc) (Gene ID: LM10043824) as Clathrin heavy chain is the major component of the polyhedral cage of coated vesicles. LmChc was continually expressed in the fat body and ovary after adult eclosion (Fig. 7A). Injection of LmChc dsRNA (dsChc) resulted in LmChc knockdown in the fat body but not in the ovary (Fig. 7B). Interestingly, depletion of LmChc led to an 88% decrease of LmMet mRNA levels in the fat body (Fig. 7C), indicating the pleiotropic effects of Clarthrin heavy chain. In addition to the role in endocytosis, Clarthrin heavy chain plays an important role in mitosis, protein secretion from the trans-Golgi network and many other cellular processes (McMahon & Boucrot, 2011). Injection of dsChc followed by dsMet resulted in a 78% decrease of LmMet expression in the fat body, which is not significantly different from that caused by injection of dsChc alone (Fig. 7C). LmMet mRNA levels had no significant change in the ovary because of the lack of RNAi response here (Fig. 7C).
Persistence of dsRNA in locust haemolymph We conducted an ex vivo assay by incubating 527-bp dsRNA of green fluorescent protein (dsRNA) with the haemolymph, haemolymph plasma, fat body homogenate and ovary homogenate for 4 h at room temperature,
Since the orthologue of nematode Caenorhabditis elegans SID-1 (systemic RNA interference deficient-1) has been reported to mediate dsRNA uptake in insects such as the honey bee and silkworm (Aronstein et al., 2006; Huvenne & Smagghe, 2010; Kobayashi et al., 2012), we examined the role of locust Sid-1(LmSid-1) (GenBank: JN676095) in RNAi in female adult locusts. LmSid-1 was expressed in both fat body and ovary after female adult eclosion, with relatively higher levels in the ovary (Fig. 6A). Injection of LmSid-1 dsRNA (dsSid-1) led to the disruption of LmSid-1 in the fat body but not in the ovary (Fig. 6B); however, LmSid-1 depletion did not impair the knockdown efficacy of LmMet in the fat body (Fig. 6C), while the transcript levels of LmMet and LmVgR had no significant change in the ovary (Fig. 6C, D). A previous study has shown that expression of LmSID-1 protein in Drosophila S2 cells could not enhance dsRNA uptake (Luo et al., 2012). These data indicate that LmSid-1 is unlikely to be involved in dsRNA uptake in female adult locusts.
Involvement of Clathrin heavy chain in RNAi All the orthologues of candidate genes related to endocytosis-mediated dsRNA uptake in Drosophila have been identified in locusts (Luo et al., 2013). We initially investigated the possible involvement of locust Clathrin
Figure 5. Permeability of the ovariole sheath to dsRNA. Fat bodies and ovarioles with sheaths (+sheath) or without sheaths (-sheath) were incubated with Cy3-labelled LmMet dsRNA (Cy3-dsMet) for 12 h. Nuclei were stained with Hoechst 33342 (blue) and F-actin stained with Phalloidin-Alexa Fluor 488 (green). Arrow heads indicate fluorescent signals of Cy3-dsMet. Upper panel is partially enlarged images of middle panel. Control, unlabelled dsMet. FC, follicle cell. PO, primary oocyte.
© 2013 The Royal Entomological Society, 23, 175–184
Tissue-dependent RNAi susceptibility
179
Figure 6. Effect of LmSid-1 knockdown on LmMet or LmVgR RNAi. (A) LmSid-1 expression profiles in the fat body (Fb) and ovary (Ov). *, P < 0.05 and **, P < 0.01 for fat body vs ovary. (B) LmSid-1 knockdown efficiency in the fat body and ovary after injection of LmSid-1 dsRNA (dsSid-1). **, P < 0.01 compared with GFP dsRNA (dsGFP) control. (C) LmMet mRNA levels in the fat body and ovary after injection of dsSid-1 followed by dsMet. **, P < 0.01 compared to dsGFP control. (D) LmVgR mRNA levels in the ovary after injection of dsSid-1 followed by dsVgR. Values are mean ± SE, n = 8–12.
followed by dsRNA recovery and gel electrophoresis. When incubated with the haemolymph or haemolymph plasma, the RNA band was not seen in the gel (Fig. 8A). In contrast, dsGFP retained its integrity when incubated with the fat body or ovary homogenate (Fig. 8A). To evaluate the duration of dsGFP persistence in the haemolymph, a time course study was carried out by incubating dsGFP with the haemolymph for 0.5 to 4 h. In the gel, the specific RNA band became faint after 1 h, and disappeared at 4 h (Fig. 8B). For in vivo assays, dsGFP was injected into 4-day-old female adults, extracted from the haemolymph and quantified by qPCR. To obtain an absolute quantitation, qPCR was calibrated by spiking haemolymph with serially diluted dsGFP (40 ng to 4 pg), followed by linear regression analysis of Ct values and Log10-transformed dsGFP quantity in the haemolymph (Garbutt & Reynolds, 2012). As shown in Fig. 8C and D, dsGFP residues in the haemolymph decreased dramatically and dropped to the detection limit (4 pg/μl haemolymph) 2 h after injection. © 2013 The Royal Entomological Society, 23, 175–184
Discussion The fate of injected dsRNA in the ovary and haemolymph The present study further demonstrated that injection of dsRNA into the hemocoel is unable to cause RNAi response in the locust ovary. The genes used in RNAi experiments have different functions and are either ubiquitously expressed or specifically expressed in the ovary, suggesting that the insensitivity of locust ovaries to RNAi is unlikely to be gene-dependent. Less susceptibility but clear responses to RNAi in ovaries have been seen in the desert locust S. gregaria (Wynant et al., 2012) and mosquito Ae. aegypti (Telang et al., 2013). In the honey bee A. mellifera, the susceptibility of its ovary to RNAi relies on the delivery methods (injection or ingestion) of dsRNA (or siRNA) and the genes targeted (Jarosch & Moritz, 2011; Jarosch et al., 2011). We studied the fate of dsRNA by injecting fluorescently labelled dsRNA into the hemocoel of female adults at both previtellogenic and vitellogenic
180
D. Ren et al.
Figure 7. Effect of LmChc knockdown on LmMet RNAi. (A) LmChc expression profiles in the fat body (Fb) and ovary (Ov). **, P < 0.01 for fat body vs. ovary. (B) LmChc knockdown efficiency in the fat body and ovary after injection of LmChc double-stranded (ds)RNA (dsChc). **, P < 0.01 compared to GFP dsRNA (dsGFP) control. (C) LmMet mRNA levels in the fat body and ovary after injection of dsChc followed by dsMet. **, P < 0.01 compared with dsGFP control. Values are mean ± SE, n = 8–12.
stages. Fluorescent signals of labelled dsRNA were clearly seen in the fat body cells, which is consistent with the observation of efficient gene knockdown in the fat body. In contrast, no fluorescently labelled dsRNA was detected in the follicle cells and oocytes, supporting the phenomenon of ineffective RNAi response in the ovary. We showed that the ovariole sheath is permeable to dsRNA. The ovariole sheath is the sieve-like structure freely permeable to molecules with the molecular weight ranging from 12 000 to 500 000 g/mol and with the dimension
