Monoallelic knockout of r2d2 affects the antiviral RNAi response to Mayaro virus and the reproductive potential in Aedes aegypti
Zachary J. Speth, Vivek Pokhrel, Kyah M. Featherston, David G. Rehard, William R. Reid, Alexander W. E. Franz
TL;DR
This study shows that knocking out the r2d2 gene in Aedes aegypti mosquitoes affects their antiviral response and reproduction, with monoallelic disruption increasing Mayaro virus replication.
Contribution
The study reveals the essential role of r2d2 in RNAi and reproduction, and how its partial loss affects viral infection and compensatory piRNA activity.
Findings
Complete loss of r2d2 is lethal in Aedes aegypti, preventing homozygous knockout.
Monoallelic r2d2 disruption increases Mayaro virus replication and vpiRNA abundance.
R2D2 is essential for RISC loading and siRNA pathway function, but dicer-2 remains unaffected.
Abstract
Aedes aegypti is an important vector for several human-pathogenic arboviruses. RNAi is the principal antiviral immune pathway in mosquitoes. Key steps of antiviral RNAi are processing of long dsRNAs into siRNA duplexes by dicer-2; loading of the siRNA duplexes onto Argonaute-2 with the help of R2D2; RISC formation via incorporation of Argonaute-2, which contains an siRNA; RISC-mediated targeting and degradation of homologous viral RNAs. Here, we generated an r2d2 knockout mosquito line to reveal how RNAi impairment during RISC loading complex (RLC) formation would affect arbovirus infection of Ae. aegypti. CRISPR/Cas9 gene editing has been used to knock out r2d2 in Ae. aegypti. Crossing experiments were conducted to reveal the effects of loss of r2d2 function on fecundity and fertility. Mayaro virus (Togaviridae: MAYV) infection and RNAi pathway gene expression levels were monitored…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9- —National Institutes of Health - National Institute for Allergy and Infectious Diseases (NIH-NIAID)
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsMosquito-borne diseases and control · RNA regulation and disease · RNA Interference and Gene Delivery
Background
The yellow fever mosquito, Aedes aegypti, is the principal vector for important human-pathogenic arboviruses such as dengue virus (DENV), Zika virus (ZIKV), and yellow fever virus (YFV), belonging to the Flaviviridae, and chikungunya virus (CHIKV), belonging to the Togaviridae [1, 2]. These viruses can cause explosive outbreaks in tropical regions worldwide, especially when circulating in urban disease cycles between humans and Ae. aegypti mosquitoes [3]. Mayaro virus (Togaviridae; MAYV) is a New World alphavirus, which is closely related to the Old World Semliki Forest virus (SFV) [4]. Disease symptoms caused by MAYV in humans resemble those caused by SFV or CHIKV. Sylvan vectors for MAYV are mosquitoes of the genera Mansonia and Haemagogus [5, 6], while Ae. aegypti has been implicated as an urban vector for MAYV [7–9]. A mosquito vector acquires an arbovirus during feeding on a viremic vertebrate (human) host (reviewed in [10]. During infection of the mosquito, arboviruses are confronted with several tissue barriers in the insect as well as antiviral immune responses regulated by Toll, Imd, JAK-STAT, and RNAi pathways, the latter of which plays the principal role in restricting arbovirus infection [11, 12]. The RNAi pathway in mosquitoes consists of three branches: the siRNA, miRNA, and piRNA pathways [13]. The exogenous siRNA (exo-siRNA) pathway is the major branch controlling RNA viruses such as arboviruses, whereas the miRNA pathway regulates endogenous gene expression, and the germline-associated piRNA branch provides epigenetic stability by suppressing transposable element (TE) mobilization in the germline. When somatically active in mosquitoes, the piRNA pathway targets (arbo)viral RNA genomes and processes them into small RNAs designated v(iral)piR-Nas [14–16].
The exo-siRNA pathway is triggered by the presence of long dsRNA molecules such as the replicative intermediate form of an RNA virus. Dicer-2 is the siRNA pathway’s molecular sensor and processor, whose RNA III domain cleaves long dsRNA into characteristic 21-bp duplexes with 2 nt overhangs at the 3′OH ends [13, 17, 18]. The complex consisting of dicer-2 and siRNA duplex then interacts with the small dsRNA binding protein, R2D2, to form a complex (RISC loading complex, RCL), which helps to load the siRNA duplex into the RNAse H-type endonuclease Argonaute-2 (AGO2) [19–24]. Within RCL, R2D2 binds to the thermodynamically more stable 5′ end of the 21-bp RNA and in this way determines which strand will become the passenger strand. Accordingly, AGO2 unwinds the 21-bp RNA into the guide and passenger strands, the latter of which gets discarded, while the former guides AGO2 to a long single-stranded viral RNA (or mRNA) molecule with perfect sequence homology to the guide strand. Upon base-pairing between the AGO2-associated guide RNA and the target RNA, RISC is formed to carry out the degradation of the targeted viral RNA [24].
In mosquitoes, the PIWI-interacting RNA (piRNA) pathway silences TE activity in the germline and responds to acute virus infections in the soma by targeting and degrading viral RNA genomes, including those of arboviruses [13–15]. piRNA biogenesis appears to be independent of siRNA production. The hallmark of this pathway is ~ 24–30-nt piRNAs, which are generated from ssRNA substrates in a dicer-independent manner and associate with PIWI proteins to form piRISCs [25]. Key players involved in the piRNA pathway are endonucleases of the PIWI subclass of Argonaute proteins, such as AGO3 and PIWI5 [15, 26]. A central component of the piRNA pathway is the so-called ping-pong amplification loop, which is initiated by PIWI5 associated with primary antisense piRNA precursors showing a 5′ U1 bias [13–15, 27, 28]. The resulting piRISC then binds to complementary sense-strand secondary piRNA precursors, which, following their cleavage, exhibit a 5′ A10 bias [29, 30]. These secondary piRNAs associate with AGO3, which then cleaves complementary antisense 5′ U1 piRNA precursors to maintain the ping-pong amplification loop. The precise origin of the viral genome-derived antisense piRNA precursors that initiate the ping-pong amplification loop is still not clear. Endogenous viral elements (EVEs), consisting of snippets of retrotranscribed viral RNA sequences that have been integrated into the mosquito genome in so-called piRNA clusters, have been proposed as sources of piRNA biogenesis [31–33].
Several studies have been conducted to analyze how manipulating RNAi pathway genes in Ae. aegypti affect the mosquito’s vector competence for arboviruses. For example, the siRNA pathway genes dcr2, r2d2, and ago2 were transiently silenced in mosquitoes via intrathoracic injections or transgene-mediated expression of long dsR-Nas targeting these genes [34–37]. These early efforts showed that silencing of RNAi pathway genes in Ae. aegypti increased the replication of blood meal-acquired DENV2, O’nyong’nyong, and Sindbis viruses (SINV) and shortened their extrinsic incubation periods in the vector. These studies relied on a functional RNAi pathway that temporarily impairs itself, thus turning RNAi against itself. Recently, dcr2 and ago2 have been stably disrupted in Ae. aegypti by using TALEN or CRISPR/Cas9 [38–41]. Mosquitoes in which both alleles were defective for dcr2 were significantly more susceptible to infections with a range of alphaviruses and orthoflaviviruses than WT controls [39, 40]. Importantly, depending on the infecting virus, survival rates (longevity) of the females were significantly diminished, and highly increased piRNA production as a response to compensate for the loss of virus-derived siRNAs (vsiRNAs) did not prevent a drop in survival. Similarly, stable knockout of ago2 in Ae. aegypti significantly increased titers of orally acquired DENV2, ZIKV, and MAYV at 14 days post-infection (dpi) compared with the WT control [41]. Although vsiRNA production was not affected in these ago2 knockout mosquitoes, their survival upon infection with any of the three viruses was significantly reduced. Increased levels of viral hyperinfection in those mutant females led to increased cell lysis and apoptosis. Here, we used CRISPR/Cas9-mediated gene editing to disrupt the RNAi pathway gene r2d2 in Ae. aegypti at a monoallelic level and show how its impairment affected arbovirus replication, RNAi pathway function, and mosquito fecundity and fertility.
Methods
CRISPR/Cas9-mediated editing of Ae. aegypti r2d2
The Higgs’ white eye (HWE) strain of Ae. aegypti [42] was used as the recipient for CRISPR/Cas9-mediated disruption of r2d2 (AAEL011753). The sgRNA (“#6”) target sequence (5′ GAGCACGAAGACTCCAATAA 3′) corresponded to the first exon of the r2d2 gene at nucleotide position 21, downstream of the ATG start codon (Fig. 1A). Disruption of r2d2 was achieved by site-specifically inserting the marker/reporter encoding donor plasmid shown in Fig. 1B. The donor plasmid also encoded a cassette in which the sgRNA was placed under control of the Ae. aegypti U6:3 snRNA promoter (AAEL017774). DNA plasmids were constructed using standard DNA cloning techniques and purified using the Zymo Plasmid Midiprep kit (Zymo Research, Irvine, CA, USA). The injection mix for germline transformation consisted of 2.7 µg “ΔR2D2-U6” donor plasmid (420 ng/ml), 70 ng/µl dsRNA_Ku70 (to temporally suppress non-homologous end joining of genomic DNA), and 300 ng/µl Cas9-NLS protein (PNA Bio, Thousand Oaks, CA, USA). Aligned preblastoderm Ae. aegypti HWE embryos were micro-injected with the injection mix as described [43] using a Leica micro-manipulator and a fabricated micro-capillary, which was connected to a Femtojet (Eppendorf, Hamburg, Germany) air compressor. Injected eggs were maintained for 5 days at 28^o^ C before hatching. Every hatched G0 individual developing into a male was singly outcrossed to 10 HWE females. Following a 3-day mating period, crossed individuals were combined in large (64 oz.) cartons, each containing up to 200 females. Surviving G0 females were pooled in numbers up to 30 and outcrossed to 4–5 HWE males. Mosquitoes were given artificial blood meals consisting of defibrinated sheep blood (Colorado Serum Co., Denver, CO, USA), and cartons were provided with paper-fitted egg cups for oviposition. After those eggs hatched, all resulting G1 individuals were screened for ECFP eye marker expression. These were selected, reared to adulthood, and further outcrossed to HWE (now G2) to establish a transgenic line. For the “copycat” procedure, the injection mix consisted of 100 ng/µl sgRNA #6, 70 ng/µl dsRNA_Ku70, and 300 ng/µl Cas9-NLS protein, which was micro-injected into [male ΔR2D2-U6 G1 x HWE female] preblastoderm embryos. Transgenic lines were established as described above, with the exception that only male G0s were outcrossed (since the transgenic marker was only present on male chromosome 1), and only female G1s that were 3xP3-eCFP positive were selected. Intercrossed r2d2 knockout mosquitoes of generations G6–G13 (designated ΔR2D2^(+/-)^) were used for the following experiments to describe the phenotype associated with r2d2 impairment.Fig. 1Aedes aegypti r2d2 as a target for CRISPR/Cas9-mediated gene knockout. A Genomic structure of r2d2 in Ae. aegypti. The gene is located on chromosome 1 at nucleotide position 147,407,747–147,421,473. Two small exons (each < 550 kb in size) are separated by a ~ 12,500 nt intron. The red star indicates the insertion site of the transgene shown in (B), resulting in the disruption of r2d2. B Transgene depicted as a donor plasmid for HDR-mediated insertion at genome position nt 147,407,898 to disrupt r2d2 function. Shown are the left and right homology arms partly consisting of r2d2 exon 1 sequence (depicted in pink); the open reading frame of mCherry as a reporter for r2d2 expression; transcription termination signals form Simian virus 40; the ECFP eye marker under control of the 3xP3 promoter; the U6-sgRNA expression cassette downstream of the right homology arm. The backbone of the transgene consists of the pSLfa1180fa plasmid vector. C Linkage of the r2d2 allele to the M locus (male sex determination locus) on chromosome 1; the diagram has been modified from Matthews et al. (2018) [67]
Fertility and fecundity assays
The following crosses were established between 5–6-day-old males and females: (1) ΔR2D2^(+/-)^ males (G13) x HWE females; (2) ΔR2D2^(+/-)^ females (G13) x HWE males; (3) ΔR2D2^(+/-)^ males x ΔR2D2^(+/-)^ females; (4) HWE males x HWE females. Each mosquito group was placed in a large (64 oz.) carton and, following a 3-day mating period, provided with a non-infectious artificial blood meal consisting of defibrinated sheep blood (Colorado Serum Co.). Fully engorged females were then placed singly into small (16 oz.) cartons (n = 19–20/group), each of which contained an egg cup lined with a strip of paper towel for oviposition. Following a 5-day oviposition period, the egg liners were collected and maintained under slightly moist conditions. Meanwhile, females received another blood meal to complete a second gonotrophic cycle. Using a stereo light microscope, egg counts for each female were recorded to estimate fecundity. Egg liners from each female were then separately hatched in large plastic OP cups. Larvae were fed with flakes of tropical fish food (Tetramin, Melle, Germany). When reaching the pupa stage, individuals were counted to assess male/female fertility for each cross and screened for ECFP eye marker expression using a Leica MZ10 fluorescent stereo microscope (Leica, Wetzlar, Germany).
Confocal microscopy on ovarian tissues
Mosquito ovaries were dissected from fully engorged female Ae. aegypti at 24 h and 48 h post-blood-feeding using defibrinated sheep blood. Pairs of ovaries from individual females were dissected and placed into 200 µl fixative buffer (4% paraformaldehyde diluted in 1 × Dulbecco’s phosphate buffered saline [PBS]) and stored for up to 2 weeks at 4 °C. Following fixation, ovaries were washed 3 × in PBS prior to permeabilization for 1 h at room temperature in PBST (1 × PBS, 0.2% Triton X-100; Sigma-Aldrich, St. Louis, MO, USA). Samples were washed 3 × in PBS prior to incubation with Alexa Fluor 488 Phalloidin (Thermo-Fisher Scientific, Waltham, MA, USA) diluted 1:200 in staining buffer (100 µl PBS, 1% porcine serum albumin [Sigma-Aldrich], 0.1% Triton X-100) for 1–2 h in the dark at room temperature. Nucleic DNA staining was performed by adding 0.5 µl of 0.5 µg/µl DAPI and incubating tissues for 30 min at room temperature in the dark. Prior to mounting, the tissues were washed again 3 × in PBS. Pairs of ovaries from individual females were then placed in 20 µl of DAPI-Fluoromount-G (Electron Microscopy Sciences, Hatfield, PA, USA) on PTFE Printed Microscope Slides (6-well, 8 mm well diameter, Electron Microscopy Sciences). After placement of cover slips, the slides were sealed with clear nail polish and stored at 4 °C in the dark prior to imaging. The imaging was performed by the University of Missouri Advanced Light Microscopy Core on a Leica TCS SP8 STED confocal microscope.
Production of in vitro transcribed long dsRNA and transient silencing of r2d2
For transient silencing of r2d2, we generated a ~ 500-bp dsRNA derived from the coding sequence of the gene, as described before [35]. Total RNA was extracted from HWE mosquitoes using TRI-reagent (Sigma-Aldrich). Using gene-specific FWD and REV primers containing additional T7 promoter sequences at their 5′ ends (Supplemental Table S1) and the OneTaq One-Step RT-PCR kit (New England Biolabs, Ipswich, MA, USA), a PCR amplicon was generated as transcription template for dsRNA synthesis. Following purification of the PCR amplicon using the NucleoSpin Gel and PCR Clean-up Kit (Takara Bio USA Inc., Mountain View, CA, USA), the PCR amplicon was in vitro-transcribed into dsRNA using the MEGAscript TM7 Transcription Kit (Thermo-Fisher Scientific) while increasing the reaction volume (including input DNA, 3 µg) by three-fold. The produced r2d2 dsRNA was purified using a standard 1:1 (vol/vol) phenol/chloroform extraction method. The purified r2d2 dsRNA was intrathoracically injected into 5-day-old HWE or ΔR2D2^(+/-)^ females at a concentration of 700 ng/µl using the Nanoject II injection system (Drummond Inc, Broomall, PA, USA). Two days post-injection of ~ 100 ng dsRNA/individual (in a 140-nl volume), females were challenged with a MAYV containing a blood meal and further analyzed as described below.
Virus infection of mosquitoes
One-week-old HWE and ΔR2D2^(+/-)^ females (G6, G7) were orally challenged with MAYV strain IQT 4235 Peru (NCBI GenBank no. MK070491). MAYV was propagated in Vero cells at an M.O.I. of 0.01 for 32–36 h before culture supernatants from the infected cells were mixed with defibrinated sheep blood (Colorado Serum Co.) at a 1:1 ratio. The artificial blood meals containing MAYV were administered to female mosquitoes as described [8]. Non-infected control mosquitoes received artificial blood meals in which defibrinated sheep blood was mixed with non-infected cell culture supernatant at a 1:1 ratio. Following feeding, engorged females were maintained in large cartons and analyzed as whole bodies for virus infection and endogenous gene expression levels at defined timepoints.
RT-qPCR assays for the detection of viral RNA and endogenous/transgenic transcripts.
Quantitative reverse transcription PCR (RT-qPCR) assays were conducted to assess expression patterns of the endogenous RNAi genes r2d2 and dcr2 in both ΔR2D2^(+/-)^ and HWE females. Additionally, expression of the transgene-associated mCherry reporter was assessed in ΔR2D2^(+/-)^ females following MAYV infection. The oligonucleotide primers used for the RT-qPCR assays are listed in Supplemental Table S1. The RT-qPCR reaction was conducted using the SYBR green-based Luna Universal One-Step RT-qPCR kit (New England Biolabs) following the manufacturer’s protocol. Per reaction, 150 ng total RNA from individual female whole bodies was used as template. The thermal cycling condition comprised a reverse transcription (RT) step at 55 ℃ for 10 min and RT inactivation at 95 ℃ for 1 min followed by 40 cycles of PCR amplification at 95 ℃ for 10 s and 60 ℃ for 1 min. RT-qPCR reactions were performed in triplicate using the Step One Plus Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). Endogenous gene and transgene expression levels were analyzed using the ΔΔCT method [44] with normalization to the steady-state expression of Ae. aegypti ribosomal protein S7 (RPS7; AAEL009496). Viral RNA transcript equivalents were measured against a standard curve produced from tenfold serial dilutions of a plasmid in which a segment of the MAYV nsp1 gene had been inserted (for primers, see: Supplemental Table S1) and regression of the standard curve data to a four-parameter logistic (4PL) curve. The Ct values of the samples were then converted into the number of moles of RNA; these were then further converted into RNA copy number equivalents by accounting for the initially loaded RNA amount per sample.
Small RNA profiling
Whole-body tissues were collected per replicate from three ΔR2D2^(+/-)^ (G10) and three HWE females (control) at 5 days post-blood-feeding with a MAYV-containing blood meal (blood meal titer: ~ 1 × 10^7^ PFU/ml). Total RNA from homogenized tissue samples was extracted using Tri-Reagent (Sigma-Aldrich) following the manufacturer’s protocol. Total RNA yield per sample was at least 500 ng. Small RNA library preparations and sequencing were performed at the University of Missouri Genomics Technology Core. The small RNA libraries were prepared using the NEBNext Multiplex Small RNA Library Prep Set for Illumina (New England Biolabs) followed by size selection for products corresponding to < 100 nt RNAs. In total, three biological replicates were sequenced for the virus-infected ΔR2D2^(+/-)^ and HWE samples, respectively.
Sequencing was performed on an Illumina Novaseq 6000 (S4- PE100) platform. An average of 12.36 million reads were sequenced per sample (minimum: 11.06 million; maximum: 13.6 million). Sequences were trimmed using cutadapt (v4.8) [45] with specification of the NEBnext adaptor sequences AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC for the Read 1 sequence data and GATCGTCGGACTGTAGAAC for the Read 2 data. The following cutoffs were also applied during adaptor trimming (of paired-end reads): -m 10 -M 64 –max-n 0.05 -q 15. Of the processed reads, three of the HWE replicates and two of the ΔR2D2^(+/-)^ replicates exhibited read length distributions characteristic of dipteran small RNAs and were further analyzed. Indexing and alignment files were prepared using bowtie v1.3.1 [46]. Alignments were performed against the genome sequence of the MAYV strain IQT 4235 used for the mosquito infections. Datasets were then sorted, indexed, and filtered with SAMtools v1.22 [47] for the small RNA partitions of interest (i.e. 18–34 bp, 21 bp, and 24–34 bp RNA fractions). Graphical analysis of the small RNA sequences within the 18–34-bp range, including generation of histograms and alignments to the MAYV reference genome, was performed in R studio using the following programs: Rsamtools, rtracklayer, GenomicRanges, and bamsignals [48–51]. Read count normalization per sample was calculated from the RPM values of small RNA reads mapping to the viral reference genome. The RPM scaling factor of total read counts per sample was applied following adaptor trimming and Q score filtering of the paired-end reads.
Statistical analyses
Statistical analyses were performed using the GraphPad 10.2.0 software suite. Fecundity and fertility were analyzed using the non-parametric Kruskal-Wallis test followed by the Mann-Whitney U-test. Viral RNA copy numbers and endogenous gene/transgene expression levels were analyzed in pairwise comparisons using the non-parametric Mann-Whitney U-test or the Kruskal-Wallis test when more than two groups were compared.
Results
Generation of ΔR2D2 mosquitoes and the observed linkage of the r2d2 allele to the sex-determination locus
Initially, we micro-injected 1146 preblastoderm HWE embryos with the ΔR2D2-U6 donor plasmid shown in Fig. 1B, dsRNA targeting Ku70, and Cas9 protein. We obtained 34 female and 47 male G0 survivors, resulting in a 7.1% survival rate. The female survivors were mated in bulk to 34 HWE males (and blood fed), while the 47 male survivors were individually outcrossed to HWE females. Following a 3-day mating period, 23 and 24 of the 47 males were combined into two separate large (64 oz.) cartons for blood-feeding. Following blood-feeding, the female pool (F1) did not produce any transgenic offspring among 1575 pupae that had been screened for ECFP eye marker expression (Table 1). Similarly, no transgenic offspring among 6120 screened pupae were observed from the first pool (M1) consisting of 23 G0 males outcrossed to HWE females. However, 162/4243 screened larvae and pupae from the second male G0 pool (M2) showed ECFP eye marker expression (Table 1; Supplemental Fig. S1). All ECFP-positive pupae developed into males (G1). Crossing out 20 transgenic G1 males to an equal number of HWE females resulted in 53% transgenic offspring, all of which developed into males. In a repeated experiment, 22 ECFP-positive males were outcrossed to an equal number of HWE females. In this experiment, 58% of the offspring showed eye marker expression; again, all of those developed exclusively into males.Table 1G1 hatch data following microinjection of the ΔR2D2-U6 donor and backcrossing of 34 female and 47 male G0 survivors to HWE mosquitoesPool, blood feed #ECFP male progeny ECFP female progeny No eye marker (m/f progeny)F1, blood feed #100455F1, blood feed #200470F1, blood feed #300650M1, blood feed #1002120M1, blood feed #2003500M1, blood feed #300500M2, blood feed #112002160M2, blood feed #22001280M2, blood feed #3220803M2, BF #2: The 20 eye marker-positive G1 males were crossed with 20 HWE females. G2: 30/57 (53%) individuals were eye marker positive (all males). M2, BF #3: The 22 eye marker-positive males were crossed with 25 HWE females. G2: 144/250 (58%) individuals were eye marker positive—all males
These observations indicated that in Ae. aegypti, the r2d2 allele is in a strong linkage disequilibrium with the mosquito’s chromosomal sex determination locus. As shown in Fig. 1C, r2d2 is located on chromosome 1q, ~ 5 Mb downstream of the dominant male sex determination (M) locus Nix [52–57]. We then reasoned that “copy-pasting” the transgene to the other allele, which is linked to the female (m) locus on chromosome 1, would result in the development of transgene-bearing females. Therefore, we micro-injected 870 preblastoderm [male ΔR2D2^(+/-)^ (G1) x female HWE] embryos with a mixture consisting of Cas9 protein, Ku70 dsRNA, and sgRNA #6. According to Gantz and Bier (2016) [58], this approach is termed “copycat.” We obtained 35 G0 survivors, which expressed ECFP in their eyes, and all developed into males (~ 4% survival). Each male was then mated with 10 HWE females. Following a 3-day mating period, the 350 females were then consolidated into two large (64 oz.) cartons (~ 175 females each). Females received five consecutive blood meals. Between 35 and 48% of all G1 pupae originating from the females that had received the five consecutive blood meals showed ECFP eye marker expression. Furthermore, most of these pupae were male, with the exception of 25 eye marker-expressing pupae (< 3%) that were female (Table 2). Although only a few transgenic females were obtained, our results show that the “copycat” approach to insert the gene-knockout transgene into the r2d2 gene sequence located on the m locus-bearing allele was successful. When 16 G1 ΔR2D2^(+/-)^ females (blood feed #4) were crossed with four G1 ΔR2D2^(+/-)^ males, 226 female (36%) and 403 male (64%) eye marker-positive G2 progeny were obtained. This allowed us to intercross ΔR2D2^(+/-)^ males and females from G2 onward.Table 2G1 hatch data following “copycat” Cas9 and sgRNA microinjection into [male ΔR2D2^(+/-)^ x female HWE] embryos and backcrossing of resulting G0 male survivors to HWE femalesPool 1 + 2, G1, blood feed #ECFP male progenyECFP female progenyNo eye marker (m/f progeny)G1; blood feed #1107901593G1; blood feed #2137021701G1; blood feed #3145241568G1; blood feed #439916736G1; blood feed #54073545G2; 1. blood feeding403226n/a^^The 16 eye marker-positive G1 females from blood feed #4 were crossed with four eye marker-positive G1 males
Monoallelic r2d2 impairment compromises female fecundity and fertility
Fecundity and fertility data were collected from two consecutive gonotrophic cycles based on the crosses described in the Methods section. Looking at both gonotrophic cycles combined, we observed that [male ΔR2D2^(+/-)^ x female ΔR2D2^(+/-)^] and [male HWE x female ΔR2D2^(+/-)^] mosquitoes produced significantly fewer eggs (63 eggs on average; p = 0.0101 and p = 0.0207) than [male HWE x female HWE] individuals producing 77 eggs on average (Fig. 2A). However, each gonotrophic cycle, when looked at separately, did not affect mosquito fecundity (Supplemental Figure S2A). Fertility was reduced by 27% (average of both gonotrophic cycles; p < 0.0001) in [male ΔR2D2^(+/-)^ x female ΔR2D2^(+/-)^] individuals compared to [male HWE x female HWE] mosquitoes (Fig. 2B). In the first gonotrophic cycle, the cross [male HWE x female ΔR2D2^(+/-)^] also had significantly (p = 0.0204) reduced fertility (by 15% on average) compared with the [male HWE x female HWE] cross, while there was no significant difference in fertility between the crosses [male ΔR2D2^(+/-)^ x female HWE] and [male HWE x female HWE] (Supplemental Figure S2B). While average pupa development rates among the outcrosses and HWE intercrosses ranged from 57 to 75% in both gonotrophic cycles, the average pupa development rates of ΔR2D2^(+/-)^ intercrosses did not reach 50%. Our data suggest that ΔR2D2^(+/-)^ females, but not so ΔR2D2^(+/-)^ males, contributed to the significant reductions in fecundity and fertility and that r2d2 behaves as a haplo-insufficient gene.Fig. 2. Effects of monoallelic r2d2 KO on mosquito fecundity and fertility. A Fecundity over two gonotrophic cycles (combined) of individual ∆R2D2^(+/-)^ and HWE females (n = 20) when outcrossed or intercrossed. Each data point represents the number of eggs produced by a single female during her first and second gonotrophic cycles. B Fertility over two gonotrophic cycles (combined) of individual ∆R2D2^(+/-)^ and HWE females (n = 20) when outcrossed or intercrossed. Each data point represents the relative quantity of pupae developed from all eggs of a single female produced during her first and second gonotrophic cycles. Bars represent mean values. *p < 0.05; **p < 0.01; ****p < 0.0001; Mann-Whitney U-test
We conducted confocal microscopy on ovarian tissue preparations obtained from intercrossed ΔR2D2^(+/-)^ females to reveal any abnormalities in follicle morphology/development that might be associated with the mutant genotype. Indeed, at 24 h post-blood meal, primary follicles were considerably less evenly developed within the ovaries of ΔR2D2^(+/-)^ individuals than in the HWE control (Fig. 3). Up to around one-third of the visible follicles per tissue sample obtained from ΔR2D2^(+/-)^ females appeared to be very small/underdeveloped and distorted. It can be speculated that many of these distorted follicles would not develop into viable oocytes, which would align with the profound decreases in fertility observed for the [male ΔR2D2^(+/-)^ x female ΔR2D2^(+/-)^] mosquitoes.Fig. 3. Confocal microscopy images showing primary follicles of HWE and ∆R2D2^(+/-)^ females at 24 h post-blood meal. A Primary follicles associated with different ovarian samples collected from six HWE females. B Primary follicles associated with different ovarian samples collected from six ∆R2D2^(+/-)^ females. Blue: DAPI staining of nuclear DNA in nurse cells and follicular epithelia; green: phalloidin staining specific for F-actin subunits. Images were taken at 10 × magnification using a Leica TCP SP8 MP confocal microscope
Biallelic loss of r2d2 function results in a recessive lethal phenotype
Based on individual sibling crosses (12–15 individual crosses per crossing group) following the crossing scheme shown in Fig. 4 and described above, we compared the number of eye marker-expressing individuals among the hatched progeny. On average, 41.4–50.1 progeny pupae per individual sibling cross were obtained from each of the four crossing groups. It became apparent that ΔR2D2^(+/-)^ males (almost) exclusively generated male eye marker-positive progenies because of the linkage of the eye marker (transgene)-bearing allele to the M locus, while, as expected, outcrossed ΔR2D2^(+/-)^ females generated ~ 50% of female eye marker positive progeny (Fig. 4). Considering that in the [male ΔR2D2^(+/-)^ x female ΔR2D2^(+/-)^] intercross both ΔR2D2 parents were hemizygous, a non-sex linked, non-essential gene on an autosome would be expected to be inherited by 75% of the male and female progeny. Instead, 95% of the male progeny and only 50% of the female progeny showed eye marker expression in repeated intercrossing experiments. These results strongly suggest that although r2d2 is sex-linked, a biallelic loss of function of the gene results in a recessive lethal phenotype. Indeed, results of repeated experiments showed that it was not possible to generate viable (homozygous) ΔR2D2^(^^−/−)^ individuals via intercrossing.Fig. 4. Sex ratios and ECFP eye marker expression of F1 progeny resulting from sibling crosses (intercrosses/reciprocal outcrosses) between HWE and ∆R2D2^(+/-)^ individuals. Data were collected at pupa stage. Between 12 and 25 individual sibling crosses were analyzed for each of the four parental crosses. The number of progeny ranged from 538 [∆R2D2^(+/-)^ males x ∆R2D2^(+/-)^ females] to 764 [∆R2D2^(+/-)^ males x HWE females]. Bars in blue indicate the number of progeny (separated by sex) showing ECFP eye marker expression; bars in orange indicate the number of progeny (separated by sex) showing white eyes (no transgenic eye marker) expression
Monoallelic knockout of r2d2 increases MAYV replication and r2d2 expression in infected individuals
We conducted a 10-day time course study to observe MAYV replication and r2d2 expression in virus-infected ΔR2D2^(+/-)^ females. In addition, we intrathoracically injected dsRNA targeting r2d2 into ΔR2D2^(+/-)^ and HWE females at 48 h prior to MAYV infection via artificial blood meals to assess whether r2d2/RNAi function in the ΔR2D2^(+/-)^ mosquitoes could be further impaired, resulting in an altered infection phenotype. At 4 h post-infection (hpi), we observed that in PBS-injected ΔR2D2^(+/-)^ females (n = 9), median MAYV RNA copy number equivalents were significantly increased (p = 0.0025) compared with the PBS injected HWE control (n = 8) (Fig. 5A). Notably, at this early timepoint, a fraction of the overall detected viral RNA could have still originated from the input (blood meal acquired) virus and therefore may not represent de novo synthesized virus. At 4 dpi, no statistically significant difference was observed between ΔR2D2^(+/-)^ females (n = 9–10) and the HWE control (n = 9–12). However, in the r2d2_dsRNA-injected samples, the median viral RNA copy numbers were ~ 5-fold more than in the HWE control. Median MAYV RNA copy numbers were significantly increased (p = 0.0012) at 10 dpi in those ΔR2D2^(+/-)^ females (n = 9) that had been injected with r2d2_dsRNA. During the 10-day time course, median MAYV RNA copy number equivalents significantly increased from ~ 10^7^ to ~ 10^9^ in both HWE (p < 0.0001) and ΔR2D2^(+/-)^ mosquitoes (p = 0.0008).Fig. 5. Effects of monoallelic r2d2 KO on MAYV replication and relative r2d2 expression levels in virus-infected ∆R2D2^(+/-)^ and HWE mosquitoes at 4 hpi, 4 dpi, and 10 dpi. A MAYV RNA genome copy number equivalents (per µg total RNA) in HWE and ∆R2D2^(+/-)^ females at 4 hpi, 4 dpi, and 10 dpi. B Relative r2d2 expression levels in HWE and ∆R2D2^(+/-)^ females at 4 hpi, 4 dpi, and 10 dpi, which had been intrathoracically injected with long dsRNA targeting r2d2 or with PBS (control) prior to receiving a MAYV containing blood meal (titer in the blood meal: 10^7^ PFU/ml). Expression levels of r2d2 were normalized to the average expression of rsp7. A, B Each data point represents a whole-body female. Bars indicate median values. *p < 0.05; **p < 0.005; ****p < 0.0001; Mann-Whitney U-test
We then observed r2d2 expression during the 10-day time course in MAYV-infected ΔR2D2^(+/-)^ females. In r2d2_dsRNA-injected HWE females at 4 hpi, r2d2 expression was reduced by ~ 50% (p = 0.0101), matching similar levels as observed in ΔR2D2^(+/-)^ females (Fig. 5B). At this timepoint, r2d2_dsRNA injection did not reduce r2d2 expression levels in ΔR2D2^(+/-)^ females any further, while at 4 dpi, a significant reduction in r2d2 expression was observed between ΔR2D2^(+/-)^ females and the HWE control, irrespectively of whether or not r2d2_dsRNA had been injected (range: p < 0.0001 to p = 0.0031). At 10 dpi, an additional reduction of r2d2 expression due to r2d2_dsRNA injection was no longer apparent among HWE or ΔR2D2^(+/-)^ mosquitoes. Nonetheless, expression levels of r2d2 were significantly reduced in the ΔR2D2^(+/-)^ females (p < 0.0001). Overall, median expression levels of r2d2 (and almost linearly) increased significantly (p < 0.0001) in the virus-infected and r2d2_dsRNA or PBS-injected HWE females during the 10-day observation period but remained at similar levels in the r2d2_dsRNA-injected ΔR2D2^(+/-)^ females. Thus, in the non-compromised HWE mosquitoes, increasing r2d2 expression followed a pattern over time that corresponded to that of increasing MAYV replication. In the HWE control, r2d2_dsRNA injection led to silencing of r2d2 at 4 hpi, but no longer at 4 dpi and 10 dpi, whereas monoallelic knockout of the gene resulted in significantly reduced r2d2 expression levels compared with the HWE control at all three timepoints (range: p < 0.0001 to p = 0.0031).
We also monitored dcr2 expression levels over the 10-day time course to determine whether there was an interdependence (feedback loop) in the RNAi response between dcr2 and r2d2. We did not observe that dcr2 significantly changed its expression pattern during the time course in MAYV-infected HWE or ΔR2D2^(+/-)^ females (Supplemental Figure S3). This suggests that dicer-2 activity does not rely on proper Ago2-RISC function.
Use of the ΔR2D2(+/-) mosquitoes to monitor RNAi pathway activation
As shown in Fig. 1, our r2d2 KO transgene contained the CDS of mCherry followed by a transcription terminator derived from SV40. Thus, insertion of the transgene near the 5′ end of exon 1 was expected to place mCherry expression under control of the endogenous r2d2 promoter of Ae. aegypti. Indeed, due to antiviral RNAi pathway activation, an increased r2d2 expression in MAYV-infected ΔR2D2^(+/-)^ mosquitoes correlated with a significantly (p = 0.0010) increased mCherry reporter expression at 4 dpi compared with its expression levels in non-infected ΔR2D2^(+/-)^ mosquitoes (Supplemental Figure 4A). However, there were no differences in mCherry expression levels at various timepoints post MAYV infection, including 4 hpi, 4 dpi, and 10 dpi (Supplemental Figure 4B). The samples tested were from the same groups as used to detect r2d2 expression levels. It was not possible to visualize any specific mCherry expression in whole bodies, dissected midguts, or ovaries that looked clearly distinct from unspecific background illumination under a fluorescent microscope. The likely reason for this was the generally low mCherry expression levels, which were quantifiable by RT-qPCR but could not be visually confirmed in whole-body mosquitoes or dissected tissues using fluorescent microscopy.
Monoallelic r2d2 knockout alters the RNAi response to MAYV
We obtained small RNA profiles from ΔR2D2^(+/-)^ (two replicates) and HWE whole-body females (three replicates) at 5 dpi with 10^7^ PFU/ml MAYV via artificial blood meals. Analyzed small RNA sizes ranged from 18 to 34 nt (Figs. 6A, D). A large peak was apparent for both ΔR2D2^(+/-)^ and HWE at 21 nt RNA size, representing vsiRNAs that were processed by dicer-2 as a hallmark for an active antiviral RNAi response [13, 59]. Average normalized counts of 21 nt vsiRNAs were not significantly different between MAYV-infected HWE (830 RPM mean ± 438) and ΔR2D2^(+/-)^ (~ 576 RPM mean ± 226) mosquitoes (Fig. 7A). Both sense- and antisense oriented vsiRNAs aligned to the entire length of the MAYV genome, and there was no clear distinction in the alignment profiles between the ΔR2D2^(+/-)^ and the HWE control samples (Fig. 6B, E). In the two ΔR2D2^(+/-)^ samples, however, there was a slight tendency of a larger read count (up to two-fold) of positive sense vsiRNAs aligning to the subgenomic viral RNA compared with the HWE samples. Overall, read counts of unique vsiRNAs rarely exceeded 20 counts in any of the samples. These data indicate that production of siRNAs was not significantly affected in R2D2-deficient mosquitoes.Fig. 6. Small RNA profiling. A Small RNA reads mapping to the MAYV genome in three infected HWE samples and D two infected ∆R2D2^(+/-)^ samples at 5 dpi. B Read coverage of 21-nt vsiRNA across the genome of MAYV in three infected HWE samples and E two infected ∆R2D2^(+/-)^ samples. C Read coverage of 24–34-nt RNA across the genome of MAYV in three infected HWE samples and F two infected ∆R2D2^(±)^ samples. Each sample consisted of total RNA extracted from three whole-body females that had acquired a MAYV-containing blood mealFig. 7Analysis of MAYV-derived vpiRNA profiles in ∆R2D2^(+/-)^ and HWE mosquitoes. A Average normalized small RNA read counts (RPM) distinguishing between 21-nt and 24–34-nt RNA reads from two infected ∆R2D2^(+/-)^ and three infected HWE samples. Error bars show the standard deviation. B Peaks of 24–34-nt RNA reads aligning to the MAYV genome representing “hotspots,” which were further characterized according to their nucleotide sequences. Upward arrows indicate peaks containing putative secondary piRNAs showing A10 bias and matching the virus sense RNA genome (see Table 3 for further details). Secondary piRNA reads, which were identified in each of three samples (∆R2D2^(+/-)^ #1, #2, and HWE #3), are marked by asterisks. The position of each “hotspot” relative to the MAYV genome is indicated below the graph
In Ae. aegypti, most piRNAs are distributed within a range of 25–30 nt in length [60]. For this study, small RNA profiles with a size range between 24 and 34 nt were analyzed. In all five samples, virus-derived 24–34-nt RNAs had a strong bias for the sense viral RNA strand and were distributed across the entire viral genome (Fig. 6C, F**, Fig. 7B). In addition, 24–34-nt RNA reads corresponding to the subgenomic RNA encoding part of the viral RNA genome were > 2-fold enriched compared with the average number of reads matching the remainder of the viral genome in those five biological samples. However, there was a clear distinction in the 24–34-nt RNA profiles between the two ΔR2D2^(+/-)^ samples and 2/3 HWE samples. Small (24–34 nt) RNA reads were enriched in the ΔR2D2^(+/-)^ samples, showing on average 2.3-fold more normalized 24–34-nt RNA reads (1453 RPM mean ± 12.59) than in the HWE control samples (622 RPM mean ± 423) (Fig. 6C, F, Fig. 7). In contrast to the observed increase in MAYV-derived 24–24-nt RNA read counts among the pooled ΔR2D2^(+/-)^ samples, read counts of RNAs with the same size range when produced from the Ae. aegypti flamenco locus were remarkably similar among all sample groups (Fig. 8). The flamenco locus on chromosome 2 contains clusters of endogenous viral elements (EVEs) and is a primary source of piRNA production from the Ae. aegypti genome [33]. In those five small RNA libraries, 1.9–4.2% of the measured total 24–34-nt RNAs originated from the flamenco locus.Fig. 8. Small RNAs in MAYV-infected HWE and ΔR2D2^(+/-)^ females aligning to the flamenco locus (chromosome 2), a major source for endogenous viral elements (EVEs). A Positions and B** read length distributions of small RNAs (18–34 bp) aligning to the Aedes aegypti flamenco locus in HWE and ΔR2D2^(+/-)^ females at 5 dpi with MAYV. The Ae. aegypti AaegL5.0 assembly was used as the reference genome
We then analyzed the MAYV-associated small 24–34-nt RNA sequences from those peaks showing ≥ 25 read counts in at least one sample (Figs. 6F, 7B) for the presence of the “A10” bias, indicative of ping-pong loop-generated secondary piRNAs derived from the viral RNA genome (vpiRNAs). This way, 13 putative secondary vpiRNAs, ranging in size from 26–32 nt, were repeatedly identified across separate samples (Table 3). These data show the level of sequence conservation and a strong bias toward the production of vpiRNAs specific to viral sequence “hotspots.” Six (asterisks, Fig. 7B) of the 10 read count peaks (arrows, Fig. 7B) in ΔR2D2^(+/-)^ sample #1 contained eight vpiRNA sequences, which were also found among ΔR2D2^(+/-)^ sample #2 and the HWE #3 control sample (Table 3). In pair-wise comparisons, there were positive correlations between unique piRNA counts greater than zero and the three sample groups (Pearson correlation coefficients: 0.67, 0.08, and 0.70), meaning that the level of abundance of individual vpiRNAs followed a similar trend among the three sample groups. The identified vpiRNA sequences matched to MAYV genome regions encoding nsp1 (2 x), nsp2 (3 x), capsid (1 x), E3 (1 x), E2 (4 x), and E1 (2 x).Table 3. Small RNAs with vpiRNA signatures aligning to the MAYV genomevpiRNA (putative)Matching strand (±)Length (bp)Aligns to (position)Viral geneCounts in sample #R2D2 KO (#1)R2D2 KO (#2)HWE (#1)gaaugucgaaagucuuuguagauaucg + 2775–101nsp121239aaugucgaaagucuuuguagauaucg + 2676–101nsp1141517acgacuuggauuuaggacuaccgccu + 263678–3703nsp2203020acgacuuggauuuaggacuaccgccuaaugcu + 323678–3709nsp271623aacgguagaagaacaguaacccugcauccu + 303992–4021nsp2868aaagaagcaaccacguagaaagaaacc + 277599–7625capsid114–gugacagcuaugugccuucuggcgaau + 278197–8223E31533acugugcagacuguggcaugggccau + 268435–8460E2756gacaagacuaucaauagcugcaccguugaca + 319022–9052E2193–aagacuaucaauagcugcaccguugaca + 289025–9052E2132–aguacuacuacgggcugcauccuacgacgacc + 329449–9480E2162818aaacauggcaaaaagacagggacucacc + 2810550–10577E196–uccggucagagcgaugaauugcgcugu + 2710620–10646E1129–Counts are shown for each of the two small RNA read replicates (#1, #2) from the ∆R2D2 mosquitoes and HWE sample #1A piRNA signature was considered to be an “A” at the 10th nucleotide position for reads matching the viral RNA sense strand
Discussion
When attempting to generate a line of ΔR2D2 KO mosquitoes, we obtained only males that showed gene disruption and eye marker expression. Outcrossing of these males resulted in progeny where, again, only males carried the donor, and this pattern of heredity was repeated over successive generations. This is a clear indication that in Ae. aegypti, the r2d2 allele is in a strong linkage disequilibrium with the sex-determining locus on the q-arm of chromosome 1, a region that likely does not recombine [52–55, 57]. There is also evidence that recombination suppression is extensive on chromosome 1 between base pairs 61–211 Mbp [56]. Beyond this region, chromosome 1 has the features of a regular autosome, allowing recombination to occur with recombination rate estimates between 0.42 and 0.90 cM/Mbp. At the lower recombination rate estimate and in the absence of any recombination suppression, we would have expected to observe the r2d2 KO transgene in ~ 2% of the female progeny from outcrossed transgenic males. However, r2d2 is separated by ~ 5 Mbp chromosomal DNA from the essential component of the M locus, Nix, which inhibits the (m/m) female-specific splicing pattern of the sex determination gene dsx in the (M/m) males [52]. In our CRISPR/Cas9-mediated gene editing experiment, the donor was initially inserted by chance into the M locus-bearing allele. With a similar probability, the donor could have been inserted into the m locus instead of the M locus-bearing allele, resulting in female-only (m/m) carriers of the r2d2 disrupting transgene. Insertion of our donor into the M locus-bearing allele prompted us to “push” the transgene to the other m locus-bearing allele (where Nix is absent), allowing the generation of female transgene-carrying progeny. Applying this “copycat” procedure [58], we co-injected hemizygous embryos with Cas9 and sgRNAs and outcrossed the survivors to the HWE recipient strain. We had to add sgRNA to the injection mix because the donor plasmid linked U6-sgRNA cassette originally used for the generation of the transgenics was located outside the homology arms and therefore not inserted into the M locus bearing allele (Fig. 1).
Outcrossed eye marker-positive “copycat” survivors produced progeny, of which males and females were eye marker positive, although not at a 1:1 ratio, as substantially fewer eye marker-carrying females than males were generated. Furthermore, generational intercrosses between individual eye marker-positive male and female siblings did not result in the generation of homozygous progeny, suggesting that a biallelic loss of r2d2 function results in recessive lethal alleles for the embryo. This was further substantiated by the observation that the fertility of ΔR2D2^(+/-)^ females when mated to ΔR2D2^(+/-)^ males was reduced by > 40%, indicating an additional semi-dominant effect on fertility. In Drosophila, it was found that r2d2 is not just an essential component of the RNAi machinery but also involved in follicular development [61]. Loss of r2d2 function in Drosophila resulted in dramatically reduced (> 90%) female fertility. Specifically, during follicular development at embryogenesis, the stalk cells between follicles were absent, causing follicles to appear fused to one another. Furthermore, the follicular epithelium was abnormally developed, showing the presence of large gaps. Both defects impaired proper oocyte formation and/or caused follicle degeneration. Additional monoallelic knockout of dcr1 caused female flies to be completely sterile. We observed an accumulation of distorted primary follicles among regular-looking follicles in our intercrossed ΔR2D2^(+/-)^ females (Fig. 3), although we could not confirm the absence of stalk cells between follicles and the presence of large gaps in the follicular epithelia, as observed in Drosophila [61]. Nonetheless, our observations allow us to conclude that in Ae. aegypti, r2d2 has a similar function during embryogenesis/oogenesis as observed in Drosophila.
Monoallelic disruption of r2d2 in Ae. aegypti impaired its RNAi response to the alphavirus MAYV and resulted in a significant, although relatively minor, increase in viral RNA copy number equivalents at 4 hpi and 10 dpi, while at 4 dpi, there was a tendency toward increased viral replication in the ΔR2D2^(+/-)^ females. Overall, the effect of r2d2 impairment on MAYV replication was relatively modest during the time course, likely because (i) RNAi function was impaired by only 50%; (ii) in the mosquito, impairment of the siRNA pathway may have been compensated by an increased piRNA pathway-mediated response to the virus, which then prevented a stark increase in viral RNA levels. In an earlier work, the transient silencing of r2d2 in Ae. aegypti via intrathoracic injection of long dsRNAs led to a significant (~ 3-fold) increase in DENV2 titers at 7 dpi and a shortened extrinsic incubation period for the virus in the treated mosquitoes [35]. The study demonstrated that partial (transient) r2d2 impairment via gene silencing in Ae. aegypti can be sufficient to alter the mosquito’s infection phenotype.
The additional injection of r2d2 targeting dsRNA to further impair r2d2 function did not consistently enhance the increased MAYV infection phenotype in the HWE control or in ΔR2D2^(+/-)^ females. It can be assumed that, to be fully effective, long dsRNA-mediated gene silencing would require a completely functional siRNA machinery. However, in our ΔR2D2^(+/-)^ mosquitoes, the siRNA machinery is compromised at the stage of dicer-2/R2D2-mediated RISC loading complex (RLC) formation [19, 22], with the potential to reduce overall gene silencing efficiency in these mosquitoes. This was apparent at 4 hpi when long dsRNA targeting r2d2 had no effect on the expression level of the endogenous gene. However, at 4 dpi, a silencing effect of the supplied r2d2_dsRNA was observed in ΔR2D2^(+/-)^ females. Expression levels of r2d2 in the MAYV-infected HWE control, while generally low, increased almost linearly during the 10-day observation period and followed a similar trajectory of viral RNA genome replication at 4 hpi and 4 dpi, while at 10 dpi, viral replication seemed to reach a plateau. Thus, the mosquito’s natural RNAi pathway does not prevent viral RNA replication per se, but it might keep viral replication below a threshold.
In our MAYV-infected ΔR2D2^(+/-)^ females, monoallelic knockout of r2d2 had no effect on dcr2 expression levels, and, generally, dcr2 expression levels did not correspond to r2d2 expression levels in HWE and ΔR2D2^(+/-)^ females. This confirms that, despite their close molecular interactions during RLC formation, there is no direct feedback loop between r2d2 and dcr2 that would cause the latter RNAi component's expression levels to adjust to those of the former. Furthermore, the monoallelic impairment of r2d2 did not affect the production of MAYV-derived 21nt vsiRNA. This agrees with the findings gained from VSV or SINV-infected Δr2d2-KO^(−/−)^ Drosophila, showing that R2D2 seemed to be dispensable for the Dicer-2-mediated biogenesis of vsiRNAs [62]. Similarly, as shown by others, ago2 knockout did not negatively affect vsiRNA production in cells or mosquitoes [41, 63–65]. Nonetheless, the partial loss of r2d2 function in ΔR2D2^(+/-)^ females was reflected by an altered MAYV-derived small RNA profile showing a substantial (~ 2.7-fold) relative increase in the abundance of 24–34-nt small RNAs derived from the viral genome. This suggests, as mentioned above, that siRNA pathway impairment due to monoallelic r2d2 KO was compensated for by an increased piRNA pathway activity. Increases in abundance of virus-derived small RNAs (between 18 and 39 nt) have been previously observed in studies in which dcr2 or ago2 were knocked out in mosquitoes [39, 41].
To confirm that the observed increase in piRNA reads was not a sampling artifact or a result of increased global piRNA production in our ΔR2D2^(+/-)^ mosquitoes, the MAYV-derived vpiRNA read counts were compared with the piRNA read counts originating from the Ae. aegypti flamenco locus. This locus encodes clusters of EVEs, which are a major source of endogenously produced piRNAs [33]. Small RNA read alignments with the Ae. aegypti flamenco locus showed that up to ~ 4% on average of the 24–34 nt RNAs originated from this locus and that the small RNA profiles matching this locus were very similar among the HWE and ΔR2D2^(+/-)^ sample libraries. This indicates that monoallelic r2d2 KO did not affect the production of EVE-derived piRNAs in the germline but profoundly increased the generation of MAYV-derived vpiRNAs in the soma. Furthermore, as earlier reported for other alphaviruses such as CHIKV, SINV, and SFV [14, 25, 27, 28], we found that small RNAs in the piRNA size range matched almost exclusively to the MAYV sense RNA genome, and most putative vpiRNA peaks corresponded to the genomic region encoding the viral subgenomic RNA. Numerous MAYV-derived vpiRNAs had an “A10” bias indicative of secondary ping-pong loop-generated vpiRNAs, although we did not find them to be enriched in the vpiRNA fraction. The higher abundance of vpiRNAs matching the subgenomic region of the MAYV genome may be due to a higher expression level of subgenomic relative to genomic viral RNA, as demonstrated for SINV [14].
Taken together, our results suggest that the monoallelic KO of r2d2 compromises the mosquito’s siRNA machinery, which then leads to a ramp-up of vpiRNA production to compensate for the defective siRNA pathway. In support of this scenario, we did not observe a strong increase of MAYV replication in orally challenged ΔR2D2^(+/-)^ females, while an over-proportional increase in MAYV-derived vpiRNA reads was apparent. Thus, our observations differ from those made by Dong and Dimopoulos (2023) [41] regarding the biallelic knockout of ago2, showing a strong increase in both MAYV IQT 4235 titers and 21 nt vsiRNA reads along with vpiRNA reads at 3 dpi. A loss of ago2 function (ago2^(^^−/−)^) in Ae. aegypti prevents AGO2-RISC loading and assembly downstream of RLC formation [19, 22]. Biallelic knockout of dcr2, on the other hand, leads to an increased viral dsRNA substrate availability, providing a greater pool of viral RNAs to be processed by the piRNA pathway [40]. This then results in increased viral titers and strongly increased vpiRNA production, as earlier observed in mosquitoes [39] and in dcr2 defective Ae. albopictus C6/36 cells [66].
Conclusions
We aimed at knocking out r2d2, a key component of the siRNA pathway, to study the effects of RNAi machinery impairment at the point of RLC formation. Based on our results, there are several interesting takeaways from our study: (i) in Ae. aegypti, the r2d2 allele is linked to the sex determination locus on chromosome 1; (ii) a homozygous disruption of r2d2 has lethal consequences for the mosquito and thus is not possible; (iii) r2d2 function is not limited to RNAi but also affects mosquito fecundity and fertility; (iv) monoallelic knockout of r2d2 does not affect dicer-2 activity but impairs the RNAi pathway at the point of RLC formation; (v) increased piRNA activity along with increased vpiRNA production in MAYV-infected monoallelic r2d2 KO mosquitoes compensates for the defective siRNA pathway, leading to only moderate increases in viral replication.
Supplementary Information
Additional file 1: Figure S1: ECFP eye marker expression in ∆R2D2^(+/-)^ L2/L3 (G1) larvae of Aedes aegypti (M2, blood feed #1; see Table 1) resulting from male G0 survivors (harboring the “ΔR2D2-U6” donor plasmid), which had been backcrossed to the HWE recipient strain of Ae. aegypti. Figure S2: Effects of monoallelic r2d2 KO on mosquito fecundity and fertility. A Fecundity over two gonotrophic cycles of individual ∆R2D2^(+/-)^ and HWE females (n = 20) when outcrossed or intercrossed. Each data point represents the number of eggs per gonotrophic cycle produced by a single female. B Fertility over two gonotrophic cycles of individual ∆R2D2^(+/-)^ and HWE females (n = 20) when outcrossed or intercrossed. Each data point represents the relative quantity of pupae developed from all eggs a single female produced during the first or second gonotrophic cycle. Bars represent mean values. *p < 0.05; **p < 0.01; ***p < 0.0001; Mann-Whitney U-test. Figure S3: Relative dcr2 expression levels in MAYV-infected ∆R2D2^(+/-)^ and HWE females at 4 hpi, 4 dpi, and 10 dpi (virus titer in the blood meal: ~ 10^7^ PFU/ml). Each data point represents a whole-body female. Expression levels of dcr2 were normalized to the average expression of rsp7. Bars indicate median values. Statistical test: Mann-Whitney U-test. Figure S4: Relative mCherry expression levels in MAYV-infected ∆R2D2^(+/-)^ females as analyzed by RT-qPCR A at 4 dpi with MAYV; B at 4 hpi, 4 dpi, and 10 dpi with MAYV. Each data point represents a whole-body female that had acquired a MAYV-containing blood meal (titer in the blood meal: ~ 10^7^ PFU/ml) or a similar blood meal containing no virus. mCherry expression levels were normalized to the average expression of rsp7. Bars indicate median values. ***p = 0.0010; Mann-Whitney U-test. Supplemental Table S1: List of oligonucleotide primers used for the study.