|Home | About | Journals | Submit | Contact Us | Français|
Heritable RNA interference (RNAi), triggered from stably expressed transgenes with an inverted repeat (IR) configuration, is an important tool for reverse genetic studies. Here we report on the development of stable RNAi in Anopheles stephensi mosquitoes, the major vector of human malaria in Asia. Trans genic mosquitoes stably expressing a RNAi transgene, designed to produce intron-spliced double-stranded RNA (dsRNA) targeting the green fluorescent protein EGFP gene, were crossed to an EGFP-expressing target line. EGFP expression was dramatically reduced at both the protein and RNA levels. The levels of gene silencing depended upon the RNAi gene copy number and its site of integration. These results demonstrate that specific RNAi-mediated knockdown of gene function can be achieved with high efficiency in Anopheles. This will be invaluable to systematically unravel the function of Anopheles genes determining the vectorial capacity of the malaria parasite.
Stable and heritable RNA interference (RNAi), a form of post-transcriptional gene silencing that results in the sequence-specific degradation of mRNA by homologous dsRNA molecules (1,2), has been utilized in numerous organisms including Caenorhabditis elegans (3), trypanosomes (4,5), Drosophila (6–8), plants (9) and, recently, by expressing short-interfering RNAs in mammalian cells (10) to investigate the function of target genes. Because of its flexibility, RNAi technology has become a powerful tool for reverse genetic studies in organisms where generating loss-of-function phenotypes by the direct manipulation of target genes has proven difficult or impossible. In Anopheles mosquitoes, the vectors of human malaria, genetic transformation is routinely achieved through a transposon-mediated integration process which occurs at random locations in the genome (11,12), thus preventing the targeted disruption or manipulation of any gene of choice. As an alternative to gene targeting by conventional methods, RNAi in Anopheles has been employed in the form of injection of in vitro synthesized dsRNA molecules into adult Anopheles gambiae mosquitoes (13). However, experimental evidence indicates that the silencing effects of injected dsRNA are generally limited in time and space (14,15). The development of a heritable and robust RNAi technology for anopheline mosquitoes would provide the opportunity to exploit the information generated by the recent completion of the A.gambiae genome sequencing project (16). The application of this RNAi technology for reverse genetic screens could be used to assess the role of putative genes involved in interactions with Plasmodium parasites, as well as expanding the knowledge of important biological processes, such as embryogenesis, sex determination and behavior.
To validate heritable RNAi technology in Anopheles, we targeted an EGFP gene integrated in the genome of transgenic Anopheles stephensi mosquitoes (17). We anticipated that this approach would facilitate the assessment of the efficiency of the technology without the bias introduced by the loss of function of endogenous genes. For this purpose A.stephensi lines were developed using a red fluorescent protein DsRed-marked transposon, which contained an RNAi gene (18) designed to encode an intron-spliced dsRNA targeting the green fluorescent protein EGFP gene. Transgenic lines were crossed to EGFP-expressing mosquitoes, and the progeny of the crosses analyzed for EGFP expression at both the protein and the RNA levels. Various degrees of inhibition were observed, ranging from marginal to high level silencing. The reduction in EGFP expression was dependent upon the site of integration of the inverted repeat (IR) transgene and its copy number.
Plasmid pIR-EGFP was developed as described previously (19). Briefly, EGFP IR were produced by duplicating, in the opposite direction, the EGFP gene (Clontech) by directionally inserting a 465 bp (from position 219 to 683) PCR product into the plasmid pCR2.1v (Invitrogen). To link the two arms of the inverted repeat, a 67 bp PCR product corresponding to intron 1 from the A.gambiae lysozyme gene (GenBank accession no. U28809) was amplified from A.gambiae genomic DNA and inserted between the EGFP-IR (Fig. (Fig.1a),1a), as previously described (19). This IR-EGFP cassette was then inserted between the Actin5C promoter and Hsp70 terminator elements from Drosophila melanogaster and cloned as a NotI cassette into a minos-based plasmid that contained the DsRed marker gene, also cloned under the control of the Actin5C promoter (Fig. (Fig.11a).
Wild-type A.stephensi embryos (strain sd 500) were injected, essentially as described previously (11), with a mixture of helper plasmid pHSS6hsILMi20 (100 µg/ml) (20) and pIR-EGFP (400 µg/ml) (19). Homozygous lines were developed as described previously (11).
Genomic DNA from transformed adult mosquitoes (G3 and G9 generations) was digested with either EcoRI and NotI or HincII restriction endonucleases. Digested genomic DNA (~4 µg/lane) was separated on a 0.8% (w/v) agarose gel and transferred onto a nylon membrane. Membranes were hybridized overnight at 65°C with the following 32P-labeled probes. Probe M: a PCR product encompassing the terminal inverted repeats of the left and right minos arms, as described previously (11). Probe EGFP-S: a 0.65 kb PCR product complementary to the sense arm of IR-EGFP, amplified using the primer combination pEGFP-S465for (5′-CCCGGATCCCCGCTACCCCGACCA-3′) and pEGFP-S465com (5′-CCCGATATCGCGGCGGTCACGAAC-3′). Probe HST: a 0.9 kb PCR product corresponding to the D.melanogaster Hsp70T gene, amplified using the primer combination pHSTfor (5′-CCCCTCTAGATGTACGAACACGCATTTATC-3′) and pHSTcom (5′-CCCCGCGGCCGCCACACGGGCGAGCGACCA-3′). Between different hybridizations, the filter was washed with a 1.5 M NaCl, 0.5 N NaOH solution for 10 min to remove the hybridized probe.
Digital images of larvae were captured on a Nikon inverted microscope using an attached Nikon DXM1200 digital camera. Fluorescent gene expression was quantified using the Lucia G image processing and analysis software (version 4.61; Nikon UK) as previously described (21). EGFP expression was calculated for 20 individual F1 and F1 Backcross larvae per experiment, 5 days post-hatching. For statistical analyses, unpaired t-tests were performed; the null hypothesis was rejected at P ≥ 0.05. χ2 tests were performed to demonstrate that the observed level of gene expression in the F1 Backcross crosses did not statistically deviate from the expected (50%) level of EGFP expression in F1 progeny when a second copy of the RNAi gene was supplied.
Total RNA was extracted from 10 fourth instar larvae from the F1 Backcross crosses. Briefly, larvae were homogenized in RNA extraction buffer (1% SDS, 1 mg/ml proteinase K, 100 mM EDTA, 125 mM NaCl, 50 mM Tris–HCl, pH 7.6) and incubated at 50°C for 1 h. Homogenates were phenol–chloroform extracted and RNA selectively precipitated from solution by three successive 8 M lithium chloride precipitations. Samples were treated with DNase RQ1 (0.2 U/µg total RNA) (Promega) and used for cDNA synthesis. An aliquot of 1 µg of DNase-treated total RNA was reverse transcribed using oligo(dT)12–18 and SuperScript™ II (Invitrogen) according to the manufacturer’s instructions.
EGFP mRNA quantification was performed on an ABI 7700 PRISM™ Sequence Detector (Applied Biosystems). EGFP MGB-TaqMan™ primers and probes (ABI) were designed using Primer Express software (ABI) in the 5′ coding sequence upstream of the IR-EGFP, to amplify a 69 bp PCR product (nucleotides 66–133). The sequences of the primers were as follows: EGFP-F, 5′-CGTAAACGGCCACAAGTTCAG-3′; EGFP-R, 5′-GGGTCAGCTTGCCGTAGGT-3′. The probe (EGFP-P, 5′-CCCTCGCCCTCGC-3′, nucleotides 95–108) was labeled with the reporter dye 6-carboxyfluorescein (6′-FAM) at the 5′ end and with the quencher dye 6-carboxytetramethylrodamine (TAMRA) at the 3′ end. EGFP expression values were normalized to expression levels of the A.stephensi ribosomal S7 gene. Primers and a FAM/TAMRA probe were designed from the orthologous A.gambiae S7 gene (22) to amplify an 84 bp product: S7-F, 5′-GGTGCACCTGGATAAGAACCA-3′; S7-R, 5′-CGGCCAGTCAGCTTCTTGTAC-3′; S7-P, 5′-AGACCACCATCGAACACAAAGT TGACACGT-3′. cDNA prepared from F1 progeny of the control cross between line VB and wild-type mosquitoes was used to develop standard curves for each gene and to determine the relative EGFP expression in the experimental crosses. Measurements of gene expression were taken in quadruplicate from two separate experiments and their mean used for further analyses. In each run a negative control (wild-type mosquito cDNA) and an RNA sample without a reverse transcriptase step (to determine genomic DNA contamination) was included.
To assess the feasibility of targeted gene silencing through the in vivo expression of dsRNA from stably integrated RNAi genes in Anopheles mosquitoes, we employed EGFP as the target gene (17). We reasoned this fluorescent marker gene would provide a simple and impartial model for assessing the efficacy of gene silencing by allowing analysis at the protein level both visually and quantitatively using image processing software. Transgenic mosquito lines were engineered to stably express an EGFP-targeting RNAi gene. A total of 610 A.stephensi embryos were injected with transformation construct pIR-EGFP (19), which contained the EGFP RNAi gene under the transcriptional control of the Actin5C promoter from D.melanogaster and a DsRed selectable marker expression unit (Fig. (Fig.1a).1a). The progeny of the 60 surviving adults showed the presence of 17 fluorescent phenotypes derived from a minimum of four G0 individuals. Southern blot analyses on HincII-digested genomic DNA (G3 generation) were performed and hybridized with probe M, spanning the left and right inverted repeats of minos (Fig. (Fig.1b).1b). A total of 13 individual transgenic lines were detected (Fig. (Fig.1b),1b), representing a minimum integration frequency of 6.7%. In these lines, two bands of variable size hybridized with probe M in each case (Fig. (Fig.1b),1b), suggesting the occurrence of a single minos-mediated transformation rather than random integration of the whole plasmid. Six independent lines (lines eIR-A1, eIR-A2, eIR-A3, eIR-B1, eIR-C2 and eIR-191) were selected for further studies. To determine the structural stability of the IR-EGFP integrated transgene, EcoRI/NotI-digested genomic DNA was also hybridized with probe EGFP-S, corresponding to the sense arm of IR-EGFP. On the basis of the specific restriction pattern of pIR-EGFP (Fig. (Fig.1a),1a), a single internal band of 4.5 kb was expected. Surprisingly, no band was detected in two (lines eIR-A1 and eIR-C2) of the six transgenic lines (Fig. (Fig.1c,1c, upper panel), indicating a deletion within the IR-EGFP cassette. Hybridization with probe HST, spanning the terminator sequence present in both the DsRed and IR-EGFP cassettes, gave two bands of the expected sizes in four out of the six lines. However, only one band, corresponding to the Hsp70T element of the DsRed expression unit, was obtained in lines eIR-A1 and eIR-C2 (Fig. (Fig.1c,1c, lower panel), demonstrating that the deletion involved the entire RNAi gene cassette. Southern blot analysis was also performed on the same six transgenic lines at generation G9 using the same three probes. Hybridizations identical to those obtained in generation G3 were observed (data not shown).
To estimate the silencing efficiency of the integrated RNAi gene, homozygous male mosquitoes from the six selected transgenic lines (eIR-A1, eIR-A2, eIR-A3, eIR-B1, eIR-C2 and eIR-191) were crossed to virgin females of transgenic line VB, homozygous for EGFP (11). These six RNAi lines were selected as they exhibited various phenotypes in terms of the spatial distribution of the marker and intensity of its fluorescence and carried a single copy insertion of the transgene. Genetic crosses had proven that these IR-EGFP lines contained the RNAi gene integrated into a chromosome different from that carrying the EGFP gene of target line VB (data not shown), to ensure two copies of the RNAi transgene and one copy of the EGFP target gene could be obtained in backcross experiments. Quantitative analysis of EGFP protein levels in the F1 progeny of the crosses, heterozygous for both the RNAi gene and EGFP target gene, was performed using image processing and analysis software (21). Among the four lines that Southern blot analysis had shown to contain the RNAi gene, eIR-B1 and eIR-191 mediated a significant inhibition of gene expression, reducing EGFP levels by 43 and 37%, respectively, as compared to heterozygous line VB larvae (Fig. (Fig.2a).2a). As expected, no inhibition was observed in lines eIR-A1 and eIR-C2, where the RNAi gene had been deleted. Heterozygous adult females from each cross were then backcrossed to males homozygous for the RNAi gene, and individuals possessing two copies of the RNAi gene and one copy of the EGFP target gene were analyzed for EGFP levels (F1 Backcross). Lines eIR-B1 and eIR-191 consistently mediated a highly significant reduction in EGFP expression, averaging 73 and 65%, respectively (Fig. (Fig.2a).2a). Remarkably, EGFP levels in these crosses were approximately halved with respect to those obtained in the F1 cross (χ2 = 0.7). EGFP expression in lines eIR-A1 and eIR-C2, which did not contain the RNAi unit, was again unaffected. Finally, lines eIR-A2 and eIR-A3 did not show a significant reduction in EGFP levels in either the F1 or F1 Backcross progeny (Fig. (Fig.2a).2a). Interestingly, in these lines the spatial pattern of DsRed expression did not co-localize perfectly with EGFP expression from line VB (Fig. (Fig.3),3), thus suggesting that the RNAi gene and its target were not transcribed in the same cells. As previously demonstrated (19), DsRed expression did not affect EGFP levels, as determined by EGFP intensity in the crosses between line VB and lines MinRED1, eIR-A1 and eIR-C2, which contained the DsRed marker cassette but not the RNAi unit.
To determine the levels of EGFP transcripts, a random sample of 10 F1 Backcross larvae from each cross was used for RNA extraction and cDNA synthesis. The abundance of EGFP mRNA in each cross was determined by real-time PCR, using a probe and primers to amplify a 69 bp region of the EGFP gene not common to the IR-EGFP sequence. Results were normalized to the A.stephensi ribosomal S7 gene (22) and relative EGFP mRNA levels calculated.
In accordance with the analysis of EGFP protein expression, F1 Backcross larvae derived from crosses involving lines eIR-B1 and eIR-191 exhibited a significant reduction in EGFP transcripts, averaging 58 and 78%, respectively, as compared to heterozygous line VB control larvae (Fig. (Fig.2b).2b). Line eIR-A2 also mediated a significant reduction in EGFP mRNA levels (34%). In agreement with the protein data, no significant reduction in EGFP transcripts was observed in crosses with lines eIR-A3, eIR-A1 and eIR-C2.
In order to investigate the feasibility of establishing a stable and heritable RNAi technology in mosquitoes, EGFP from a transgenic A.stephensi line was selected as the target gene. It was reasoned that EGFP could provide an impartial model to determine the feasibility of stable RNAi in Anopheles. The rationale for this was based on the consideration that silencing of the EGFP transgene would produce a neutral phenotype for the mosquito, avoiding the risks of selection bias associated with high silencing activity against important or vital endogenous genes.
The RNAi gene was designed to produce intron-spliced dsRNA, which our previous studies had demonstrated could mediate transient RNAi with high efficiency against EGFP (19). In that study, EGFP silencing achieved in transient transfection experiments of Anopheles cells and larvae averaged 93 and 69%, respectively. When stably expressed, this RNAi gene suppressed EGFP expression by up to 73% at the protein level and 78% at the RNA level. The intron-containing RNAi gene was utilized here on the assumption that it would be less susceptible to the genomic rearrangements reported to occur in constructs containing uninterrupted repeats (23). Our analysis, however, revealed the complete absence of the RNAi unit in two of the transgenic lines studied. This finding could be explained by recombination events that took place either within the plasmid itself or genomically. Recombination may have been triggered by the presence of the symmetrically parallel RNAi and DsRed transcription units (Fig. (Fig.1a),1a), sharing identical promoter and terminator sequences, rather than as a direct consequence of the presence of the IR in the RNAi gene itself. Importantly, the integrated constructs showed genomic stability beyond nine generations, which argues for construct instability at the time of injection.
Our data clearly indicate that silencing effects were improved by doubling the RNAi gene copy number, in accordance with similar studies using IR transgenes in Drosophila (7,8). Levels of EGFP protein expression were approximately halved when a second copy of the IR transgene was provided in two of the crosses. Furthermore, silencing efficiency appeared to be dependent upon the site of integration of the RNAi gene. Silencing was only observed in those lines where the expression pattern of the EGFP target gene and the RNAi gene, as determined by DsRed expression, overlapped. This is in agreement with studies of transgene-mediated RNAi in Drosophila (23,24), indicating that RNAi is not systemic and does not spread across mosquito cells. Whether this suggests a lack of dsRNA amplification in Diptera, common to Neurospora (25), plants (26) and C.elegans (27), is yet to be established, but it may simply reflect the postulated inefficiency of dsRNA formation from transgenes, possibly because nuclear RNA-binding proteins prevent dsRNA annealing (28).
While it is impossible to extrapolate the results of this study to endogenous mosquito genes, the level of gene silencing achieved here is encouraging despite the fact that we were unable to completely abolish the EGFP phenotype. Since RNAi only knocks down target gene expression, it could offer an advantage in performing functional genomic studies on vital genes and produce a range of useful phenotypes that are missed in classical knockout studies. Because minos reliably creates multiple genetic lines, an array of useful hypomorph phenotypes is likely to be generated for most studies.
The results reported here represent the first example of stable RNAi in Anopheles. This approach can now be applied to study mosquito molecules that have been postulated to function as receptors for malaria parasite development. Furthermore, stable RNAi could greatly contribute to the understanding of important mosquito physiological functions, including chemoreception, locomotion, osmoregulation and homeostasis, taking advantage of the recently published A.gambiae genome sequence (16).
A.E.B. was supported by the BBSRC, and L.B. and F.C. were supported by the Wellcome Trust.