Tudor-SN Promotes Early Replication of Dengue Virus in the Aedes aegypti Midgut

Summary

Diseases caused by mosquito-borne viruses have been on the rise for the last decades, and novel methods aiming to use laboratory-engineered mosquitoes that are incapable of carrying viruses have been developed to reduce pathogen transmission. This has stimulated efforts to identify optimal target genes that are naturally involved in mosquito antiviral defenses or required for viral replication. Here, we investigated the role of a member of the Tudor protein family, Tudor-SN, upon dengue virus infection in the mosquito Aedes aegypti. Tudor-SN knockdown reduced dengue virus replication in the midgut of Ae. aegypti females. In immunofluorescence assays, Tudor-SN localized to the nucleolus in both Ae. aegypti and Aedes albopictus cells. A reporter assay and small RNA profiling demonstrated that Tudor-SN was not required for RNA interference function in vivo. Collectively, these results defined a novel proviral role for Tudor-SN upon early dengue virus infection of the Ae. aegypti midgut.

Graphical Abstract

Highlights

• •Tudor-SN is upregulated in the Ae. aegypti midgut early upon dengue virus infection
• •Tudor-SN promotes viral replication in vitro and in vivo
• •Tudor-SN localizes to the nucleolus in mosquito cells
• •Tudor-SN is not required for RNAi function in vivo

Genetics; Virology

Introduction

The mosquito Aedes aegypti transmits a wide range of pathogens to humans, many with severe consequences on public health, including dengue, Zika, and chikungunya viruses (Gould et al., 2017). For instance, dengue virus (DENV) infects 390 million people annually (Bhatt et al., 2013) and 50% of the world's population is at risk for infection (Brady et al., 2012). DENV belongs to the Flaviviridae family and has a positive-sense, single-stranded RNA genome. DENV exists as four genetic types (DENV-1, -2, -3 and 4) that are phylogenetically related and loosely antigenically distinct (Katzelnick et al., 2015). In the wild, mosquitoes acquire DENV by feeding on a viremic host. After the infectious blood meal, DENV infection is first established in the mosquito midgut before spreading systematically and reaching the salivary glands, where the virus engages in further replication (Raquin and Lambrechts, 2017) before being transmitted to the next host via the saliva released during the bite (Salazar et al., 2007, Black et al., 2002).

The primary prevention strategy against arboviral diseases relies on the control of vector populations. Current vector control methods are mainly based on insecticides. Despite having been applied for decades, the burden of arboviral diseases keeps increasing (Messina et al., 2019). Human travel, urbanization, climate change, and geographic expansion of mosquito vectors increase pathogen transmission and spread (Weaver, 2013). Over the last two decades, research efforts have led to the production of laboratory-engineered mosquitoes that either suppress wild vector populations or render them incapable of transmitting pathogens (Champer et al., 2016, Yakob et al., 2017). As the methods for genetic modification of mosquitoes develop, the need to identify optimal target genes that are naturally involved in mosquito antiviral defenses or required for viral replication also increases. Preferably, such pro- or antiviral target genes would act early during the course of an infection, and, when engineered, would permit early blocking of virus replication, at the level of the midgut cells. This would hinder viral dissemination and make further transmission of the virus impossible.

The majority of our knowledge about insect antiviral immunity originates from investigations in the model organism Drosophila melanogaster (Merkling and Van Rij, 2013, Mongelli and Saleh, 2016), whereas studies in mosquito vectors remain more limited (Bartholomay and Michel, 2018, Simoes et al., 2018, Lee et al., 2019). The Toll, IMD, and Jak-Stat pathways have been implicated in insect innate immune responses to bacteria, fungi, viruses, and parasites. Their activation triggers translocations of NF-κB-like or Stat transcription factors to the nucleus, inducing the expression of an array of immune genes encoding antimicrobial peptides and virus restriction factors, among others (Bartholomay and Michel, 2018, Simoes et al., 2018, Lee et al., 2019, Merkling and Van Rij, 2013, Mongelli and Saleh, 2016). Another major branch of insect innate immunity is RNA interference (RNAi), which encompasses several pathways leading to the production of small RNA molecules of different characteristics, such as small interfering RNAs (siRNAs), microRNAs (miRNAs), and P element-induced wimpy testis (PIWI)-interacting RNAs (piRNAs) (Miesen et al., 2016). The siRNA pathway is hitherto considered as the cornerstone of antiviral immunity in insects. It is initiated with the sensing and cleavage of viral double-stranded RNA (dsRNA) into 21-nucleotide-long siRNAs by the endonuclease Dicer-2. These siRNAs are loaded in the RNA-induced silencing complex (RISC) that guides Ago2-mediated cleavage of viral target sequences (Miesen et al., 2016). Numerous studies reported that depletion of siRNA pathway components in mosquitoes resulted in increased arbovirus replication (Campbell et al., 2008, Keene et al., 2004, Myles et al., 2008, Sanchez-Vargas et al., 2009, Franz et al., 2006).

Although several pathways involved in antiviral immunity have been characterized in mosquitoes, several aspects of anti-DENV defense remain elusive. For example, the siRNA pathway was shown to inefficiently restrict DENV replication in the Ae. aegypti midgut (Olmo et al., 2018). Besides, most of previous studies have focused on mosquito antiviral or restriction factors that antagonize DENV, but little is known about mosquito host factors with a proviral function, that is, factors enhancing DENV propagation. Several human factors required for DENV infectivity were recently discovered through genome-wide CRISPR screens (Savidis et al., 2016, Zhang et al., 2016, Marceau et al., 2016), whereas only a handful of DENV host factors have been identified in mosquitoes to date (Londono-Renteria et al., 2015, Jupatanakul et al., 2014, Sessions et al., 2009, Raquin et al., 2017). Although CRISPR screens cannot be readily carried out in live mosquitoes, transcriptome analysis by high-throughput RNA sequencing is a powerful method to identify DENV host and restriction factors in vivo (Sigle and Mcgraw, 2019). For example, novel DENV restrictions factors (DVRF-1 and -2) that depend on the Jak-Stat pathway activation have been uncovered by overlapping transcriptional profiles of mosquitoes infected with DENV and mosquitoes with a hyperactive Jak-Stat pathway (Souza-Neto et al., 2009).

In this study, we exploited a unique transcriptomic dataset that we previously generated by performing RNA sequencing on individual midguts in a field-derived Ae. aegypti population during early DENV-1 infection (Raquin et al., 2017). In addition to a conventional pairwise comparison of gene expression between DENV-infected and uninfected controls, we also used an approach to detect correlations between viral RNA load and gene expression. Of 269 candidate genes identified by either method, only four were differentially expressed upon DENV-1 infection and had expression levels that correlated with viral RNA load in infected mosquitoes (Raquin et al., 2017). Among the four candidate genes identified by both methods was a gene encoding a member of the Tudor protein family, Tudor Staphylococcal Nuclease (abbreviated Tudor-SN or TSN), which we selected for further investigation in the present study. Using RNAi-mediated gene knockdown in vivo, we found that reduced TSN expression resulted in lower viral loads in vitro and in vivo. Immunofluorescence assays revealed that TSN localized to the nucleolus and did not colocalize to DENV replication sites in DENV-infected cells. Finally, we used a reporter assay and small RNA profiling to show that TSN was not involved in RNAi function in the midgut of adult mosquitoes. Altogether, our results demonstrate that TSN has an early proviral effect on DENV replication in the midgut and could be considered as a target to develop genetically modified mosquitoes that are refractory to DENV infection.

Results

TSN Expression Is Upregulated upon DENV-1 Infection and Positively Correlates with Viral Loads

Our previous transcriptomic analysis revealed that TSN (AAEL000293) expression was significantly upregulated upon DENV-1 infection relative to mock controls 1 day after exposure to the infectious blood meal (Figure 1A) but not 4 days post blood meal (Figure 1B). Inversely, we found that TSN expression was not significantly correlated with DENV-1 viral loads 1 day post blood meal (Figure 1C) but was positively correlated with DENV-1 RNA loads 4 days post blood meal (Figure 1D). Thus, we concluded that TSN expression was induced by DENV-1 infection within 24 h after the infectious blood meal and that subsequently, its expression was positively correlated with DENV-1 replication. The positive correlation was suggestive of a proviral role for TSN upon DENV-1 infection.

Figure 1.

TSN Is Upregulated upon DENV-1 Infection and Correlates Positively with Midgut Viral Loads

(A and B) TSN midgut expression levels on day 1 and day 4 post DENV-1 exposure. Log₂-transformed TSN normalized RNA-seq counts are shown in mock-infected (n = 6) and DENV-1-infected (n = 16) midguts. p values of the pairwise t tests are indicated.

(C and D) Correlation of TSN expression level and viral load in DENV-1-infected midguts on day 1 and day 4 post virus exposure. Log₂-transformed TSN normalized RNA-seq counts are shown as a function of the log₁₀-transformed midgut viral load. Black lines represent the linear regression and light purple shaded areas represent the 95% confidence intervals of the regression. Pearson's coefficients of determination (r) and p values of the linear regression coefficient are indicated.

TSN Is a DENV Proviral Factor In Vitro

First, we sought to test the proviral role of TSN in vitro by using Ae. aegypti Aag2 cells in culture. We transfected Aag2 cells with dsRNA to trigger RNAi-mediated knockdown of TSN or an exogenous green fluorescent protein (GFP) sequence (Table 1) and subsequently inoculated them with DENV-1 at a multiplicity of infection of 1. We measured TSN expression levels by reverse transcription quantitative PCR (RT-qPCR) at 0, 12, 24, 36, 48, 72 and 96 h post infection and found that TSN knockdown efficiency ranged from ~50% to 80% and was statistically significant at most of the time points (Figure 2A). We visualized TSN protein levels by western blotting using an antibody directed against the human ortholog of TSN named SND1, which also reacted against the Ae. aegypti TSN. We confirmed that TSN knockdown reduced TSN protein levels by 70%–80% in Aag2 cells at 24 and 48 h post DENV-1 infection, compared with the GFP control (Figure 2B). To determine whether TSN also augmented viral infection in vitro, we measured both DENV-1 RNA levels (Figure 2C) and DENV-1 infectious titers by focus-forming assay (Figure 2D) over the course of infection. We found that DENV-1 RNA levels were significantly reduced upon TSN knockdown relative to control levels at 24 h post infection (Figure 2C, p < 0.01). Moreover, DENV-1 RNA levels were consistently lower in TSN-depleted cells from 24 to 96 h post infection. DENV-1 infectious titers were also significantly reduced upon TSN knockdown at 24 and 48 h post infection (Figure 2D, p < 0.05 and p < 0.001, respectively). Overall, these data demonstrated a proviral role of TSN in vitro.

Table 1.

List of Oligonucleotide Primers and Molecular Probes Used in This Study

Organism	Primer/Probea	Sequence (5′-3′)	Product Size (bp)	Reference
Ae. aegypti	rp49-F	ACAAGCTTGCCCCCAACT	97	(Gentile et al., 2005)
	rp49-R	CCGTAACCGATGTTTGGC
	TSN-F	CTGCAGATGAACGTCGAGTA	100	This study
	TSN-R	CATCGCTGACCAGTTCCTT
	dsTSN-Fb	taatacgactcactatagggAAAGGCAAATGGAGCGACT	312	This study
	dsTSN-Rb	taatacgactcactatagggGACGTCACGTTGCAGCAG
	dsGFP-Fb	taatacgactcactatagggATGGTGAGCAAGGGCGAG	501	This study
	dsGFP-Rb	taatacgactcactatagggTTACTTGTACAGCTCGTC
	dsLuc-Fb	taatacgactcactatagggCGCCCTGGTTCCTGGAAC	556	This study
	dsLuc-Rb	taatacgactcactatagggAGAATCTCACGCAGGCAGTTC
DENV-1	NS5-F	GGAAGGAGAAGGACTCCACA	105	(Fontaine et al., 2016)
	NS5-R	ATCCTTGTATCCCATCCGGCT
	NS5-Probe	CTCAGAGACATATCAAAGATTCCAGGG
DENV-3	NS5-F	AGAAGGAGAAGGACTGCACA	105	This study
	NS5-R	ATTCTTGTGTCCCAACCGGCT
CHIKV	CHIK_10366_F	AAGCTCCGCGTYCTTTACCAAG	208	(Modified from Pastorino et al., 2005)
	CHIK_10574_R	CCAAATTGTCCYGGTCTTCCT

^aF stands for forward and R stands for reverse.

^bT7 sequences are written in bold.

Figure 2.

TSN Promotes DENV-1 Infection in Ae. aegypti Aag2 Cells

(A) TSN expression levels in Aag2 cells transfected with dsRNA targeting TSN (dsTSN) or GFP (dsGFP), over the course of DENV-1 infection at a multiplicity of infection of 0.1. Aag2 cells were transfected 3 and 1 day prior to virus infection. TSN expression was measured by RT-qPCR and normalized to rp49. Data represent mean and standard deviation of three biological replicates. *p < 0.05; **p < 0.01; ***p < 0.001 (Student's t test). Data are from the same experiment shown in (C) and (D).

(B) Western blot analysis of TSN expression in Aag2 cells 24 and 48 h after viral infection. An anti-β-actin monoclonal antibody was used for loading control. Molecular mass in expressed in kilodaltons (kDa).

(C and D) Analysis of DENV-1 RNA levels (C) or DENV-1 infectious titers (D) over the course of DENV-1 infection following TSN or GFP knockdown. RNA levels were measured by RT-qPCR on RNA from cell extracts, and infectious titers were determined by focus-forming assay on cell culture supernatants. Data represent mean and standard deviation of three biological replicates. *p < 0.05; **p < 0.01; ***p < 0.001 (Student's t test).

TSN Is a DENV Proviral Factor In Vivo

To confirm the proviral role of TSN in vivo, we experimentally reduced TSN expression in adult female mosquitoes by intrathoracic injection of dsRNA and subsequently exposed them to an infectious blood meal containing 10⁷ focus-forming units (FFU)/mL of DENV-1 (Figure 3A). First, we monitored TSN expression levels in individual mosquitoes by RT-qPCR on days 0, 1, and 4 after the infectious blood meal. On day 0, which corresponds to 3 days after injection of dsTSN, TSN expression was significantly knocked down relative to mosquitoes injected with a control dsRNA targeting GFP (Figure 3B, p < 0.001). Reduced TSN expression persisted over time through day 1 (Figure 3C, p < 0.0001) and day 4 (Figure 3D, p = 0.01) after exposure to the infectious blood meal. Importantly, reduced TSN expression did not significantly impact the survival of mosquitoes during the seven days following injection compared with the dsGFP control (Figure 3E, p = 0.54). We also measured TSN expression in head, thorax, abdomen, ovary, and midgut tissues in sugar-fed or blood-fed mosquitoes and found that TSN expression was significantly upregulated in midguts 1 day after a blood meal, suggesting a tissue-specific role within the first day after a blood meal (Figures S1A and S1B, related to Figure 3). Next, we measured DENV-1 RNA loads by RT-qPCR and found an ~50% reduction of viral loads in mosquito midguts depleted for TSN, compared with the GFP control, 4 days after the infectious blood meal (Figure 3F, p < 0.0001). We also measured DENV-1 RNA loads one day after the infectious blood meal and did not observe a decrease of viral loads. However, this is most likely due to the presence of viral RNA in the undigested blood still present in the midgut at this time point, as shown in a previous report (Raquin et al., 2017). These results confirmed the proviral role of TSN during DENV-1 infection in the mosquito midgut 4 days after the infectious blood meal. However, we found no evidence that TSN knockdown had an impact on infection prevalence after DENV-1 exposure. Among the mosquito midguts analyzed by RT-PCR on day 4, we found that 88% and 90% were positive for DENV-1 RNA in the TSN knockdown and the dsGFP control groups, respectively (Figure 3G, p = 0.59). In addition to DENV-1, we assessed the proviral role of TSN upon infection by another DENV serotype, DENV-3, and the alphavirus chikungunya virus (CHIKV). We injected adult mosquitoes with dsRNA against TSN or luciferase as a negative control and offered them an infectious blood meal 3 days later. We confirmed TSN knockdown (Figures S2A, related to Figure 3) and observed a significant reduction in DENV-3 RNA levels in individual mosquito midguts (Figures S2B, related to Figure 3) 4 days post DENV-3 exposure. We confirmed TSN knockdown 2 and 4 days post CHIKV exposure (Figures S2C and S2E, related to Figure 3), but despite a slight reduction of CHIKV RNA levels at both time points the difference with controls was not statistically significant (Figures S2D and S2F, related to Figure 3). Therefore, we found that TSN acted as a proviral factor for two DENV types, DENV-1 and DENV-3, but not for the alphavirus CHIKV. Overall, our data demonstrated that, although TSN does not influence the probability of DENV infection, it promotes early DENV replication in the mosquito midgut.

Figure 3.

TSN Promotes DENV-1 Infection in the Mosquito Midgut

(A) Experimental scheme of the gene-silencing assays in vivo.

(B–D) TSN expression levels following gene knockdown on day 0 (B), day 1 (C), and day 4 (D) after exposure to DENV-1 infectious blood meal. Mean percentage of gene expression knockdown on day 0 (B), day 1 (C), and day 4 (D) after DENV-1 exposure are indicated. Boxplots show TSN expression normalized by rp49 and expressed as 2^−dCt values in n = 12–24 individual mosquito midguts per group. Individuals with less than 50% gene expression knockdown are shown as empty dots. Data are representative of three separate experiments. p values above the graph indicate statistical significance assessed with a Wilcoxon test.

(E) Percentage of survival following dsRNA injection and/or DENV-1 exposure. Mosquitoes were injected with dsRNA targeting TSN (n = 127), targeting GFP (dsGFP, n = 110) 3 days prior to DENV-1 exposure. Non-injected mosquitoes that fed on an infectious (n = 71) or a non-infectious blood meal (n = 62) were used as controls. No significant difference in mortality was detected between dsGFP and dsTSN mosquitoes according to a Cox model (p = 0.54).

(F) DENV-1 RNA levels in mosquito midguts dissected from mosquitoes previously injected with dsGFP (n = 53) or dsTSN (n = 55). Boxplots represent the viral load measured by RT-qPCR on day 4 post exposure. Data represent three separate experiments combined. The negative effect of TSN knockdown on viral load was statistically significant in each of the three experiments. Viral loads are adjusted for differences between experiments and expressed in mean-centered DENV-1 RNA loads. The p value above the graph indicates statistical significance of the treatment effect assessed with an analysis of variance accounting for the experiment effect.

(G) DENV-1 infection prevalence in mosquito midguts measured by RT-qPCR on day 4 post virus exposure following injection with dsGFP (n = 53) and dsTSN (n = 43). Data from three separate experiments were combined after verifying the lack of a detectable experiment effect. Error bars represent 95% confidence intervals of the percentages. The p value above the graph indicates statistical significance of the treatment effect assessed with a logistic regression.

See also Figures S1 and S2.

TSN Localizes to the Nucleolus in Mosquito Cells

It was previously shown that TSN could interact with DENV RNA in mammalian cells (Lei et al., 2011), which led us to ask whether TSN co-localized with DENV-derived RNA in mosquito cells. Double-stranded RNA is produced during the replication of single-stranded RNA viruses like DENV and is a hallmark of RNA virus infection (Weber et al., 2006). Previous reports demonstrated that antibodies directed against dsRNA did not cross-react with cellular rRNA or tRNA and could be used to identify flavivirus replication complexes in infected cells (Emara and Brinton, 2007). To determine the subcellular localization of TSN, we performed immunofluorescence assays in mosquito cells derived from Ae. albopictus (C6/36, Figure 4A) or Ae. aegypti (Aag2, Figure 4B) using the anti-SDN1 antibody previously validated by western blotting (Figure 3B) and a monoclonal antibody targeting dsRNA (called αK1). In both cell types, we found that TSN was expressed and localized to the nucleolus. Indeed, it localized to the nucleus region but did not overlap with DAPI staining, which is reported to exclude the nucleolus (Sirri et al., 2008). Moreover, the staining was more intense at the nucleus-nucleolus interface where it formed a “ring.” TSN localization did not change upon DENV-1 infection, nor did its expression level. Six days after DENV-1 infection of C6/36 and Aag2 cells, dsRNA staining was readily detectable and mainly localized to cytoplasmic regions of infected cells, likely corresponding to viral replication sites. Since TSN localized to the nucleolus, and the dsRNA to the cytoplasm, we did not observe overlapping signals between both stainings. Thus, we conclude that TSN does not interact with DENV-1 RNA at its replication site. However, it remains possible that interactions occur with other forms of DENV-1 RNA (positive or negative single-stranded RNA) or viral proteins.

Figure 4.

TSN Localizes to the Nucleolus in Ae. aegypti and Ae. albopictus Cells

(A and B) Immunofluorescence assays in Ae. albopictus C6/36 (A) or Ae. aegypti Aag2 (B) cells 6 days after infection with DENV-1. Cells were stained with the nuclear stain DAPI, an antibody against TSN coupled to an Alexa 488 secondary fluorescent antibody, and an antibody against dsRNA (αK1) coupled to an Alexa 689 secondary fluorescent antibody. Cells were imaged on a confocal microscope at 63× magnification.

RNAi Is Functional in TSN-Depleted Mosquitoes

Proteins containing Tudor motifs have been implicated in multiple aspects of RNA metabolism such as RNA splicing or small RNA pathways (Lasko, 2010, Siomi et al., 2010). TSN was shown to be a component of the RISC in Caenorhabditis elegans, Drosophila, and mammals (Caudy et al., 2003) and was suggested to participate in RNAi function in the tick Ixodes scapularis (Ayllon et al., 2015). Therefore, we asked whether TSN was involved in RNAi function in Ae. aegypti. We adapted a luciferase-based RNAi sensor assay developed in Drosophila to mosquitoes (Merkling et al., 2015a, Merkling et al., 2015b, Van Cleef et al., 2011). Adult females were intrathoracically injected with a mix of lipofectant along with Firefly luciferase reporter plasmid with Firefly luciferase-specific dsRNA and dsRNA targeting GFP (as a negative control), Ago2 (as a positive control), or TSN (Figure 5A). A reporter plasmid encoding a Renilla luciferase was used as an in vivo transfection control. Three days after injection, the efficiency of Firefly luciferase silencing was measured in whole-mosquito homogenates (Figure 5B). When reporter plasmids were injected together with control dsGFP, we observed a wide range of luminescence counts (likely due to variable in vivo transfection efficiency), but the average luciferase activity was about 100-fold higher than when dsRNA targeting Firefly luciferase (dsLuc) was co-transfected with the reporter plasmids and control dsGFP. The silencing of Firefly luciferase was partially restored upon knockdown of Ago2, a key gene of the RNAi pathway, demonstrating the validity of the reporter assay. Finally, we observed that the silencing of Firefly luciferase was maintained upon co-transfection with the dsRNA targeting TSN, suggesting that TSN does not enhance RNAi function in Ae. aegypti (Figure 5B). We measured expression levels of TSN and Ago2 upon co-transfection with reporter plasmids and dsRNA and verified that TSN and Ago2 expression levels were significantly reduced upon knockdown with their specific dsRNA (Figures 5C and 5D). Although we cannot exclude that residual TSN expression could suffice to maintain its activity, the knockdown efficiency was similar to that of Ago2. Overall, these results supported the conclusion that TSN is not a positive regulator of RNAi in Ae. aegypti. One caveat of this RNAi reporter assay is that we could only reliably assess a positive effect of TSN on RNAi activity (i.e., measure higher luminescence counts). Indeed, the efficiency of luciferase silencing was very high in the presence of dsRNA, which may have prevented our ability to detect a negative effect of TSN knockdown on RNAi activity (i.e., lower luminescence counts than the dsLuc + dsGFP control).

Figure 5.

TSN Does Not Enhance RNAi Function

(A) Experimental scheme. Adult mosquitoes were co-injected with Firefly luciferase (Fluc) and Renilla luciferase (Rluc) reporter plasmids, and dsRNA targeting Fluc in combination with dsRNA against GFP, TSN or Ago2. Luminescence was measured 3 days post injection.

(B) In vivo RNAi reporter assay. Reporter gene activity was measured in individual mosquitoes. Luminescence counts of Firefly luciferase were normalized to the Renilla luciferase counts. Mean and standard deviation are shown. *p < 0.05; **p < 0.01; ***p < 0.001 (Mann-Whitney t test).

(C and D) Expression levels of TSN (C) and Ago2 (D) measured by RT-qPCR on mosquitoes harvested 3 days after injection with plasmids and dsRNA. Gene expression was normalized by rp49 and expressed as 2^−dCt values.

*p < 0.05; **p < 0.01; ***p < 0.001 (Student's t test). Data are from the same experiment shown in panel (B).

Small RNA Profiling in TSN-Depleted Mosquitoes

To overcome the limitations inherent to the RNAi reporter assay, and further assess the impact of TSN depletion on RNAi activity, we deep sequenced small RNA populations in TSN-depleted mosquitoes infected with DENV-1. We first injected adult mosquitoes with dsRNA targeting TSN or luciferase as a control. Two days later, we exposed mosquitoes to DENV-1 via an infectious blood meal. Four days after exposure to the virus, we performed a second injection of dsRNA against TSN and luciferase to prolong gene silencing (Figure 6A). We verified TSN knockdown by measuring TSN expression levels on days 4 and 10 post infection (Figures S3A, related to Figure 6). We then selected mosquitoes that were infected with DENV-1 and displayed low TSN expression for sequencing, as well as control mosquitoes injected with dsRNA targeting luciferase that had equivalent DENV-1 RNA levels but normal TSN expression (Figure S3B, related to Figure 6). These mosquitoes were used to prepare and sequence small RNA libraries and compare the three canonical small RNA populations: miRNAs, siRNAs (21 nucleotides in length), and piRNAs (26–30 nucleotides in length). We first examined small RNAs mapping on the DENV-1 genome. As expected, 21-nucleotide (nt)-long siRNAs where highly abundant in both the TSN-depleted mosquitoes (Figure 6B) and control mosquitoes (Figure 6C), and they were distributed across the viral genome, on both strands, in both conditions (Figures S3C and S3D, related to Figure 6). Next, we examined small RNAs mapping on the mosquito genome. We found that TSN depletion did not affect the miRNA machinery, as miRNA abundance was very similar between both conditions tested (Figure 6D). Likewise, we found that TSN knockdown did not influence siRNA and piRNA biogenesis. The abundance of both histone-derived siRNAs (Figure 6E) and histone-derived piRNAs (Figure 6F) was similar between the TSN-depleted mosquitoes and controls.

Figure 6.

Small RNA Populations Are Unchanged Following TSN Knockdown

(A) Experimental scheme. Adult female mosquitoes were injected with dsRNA targeting luciferase or TSN 2 days before and 4 days after a DENV-1 infectious blood meal. Whole bodies were harvested on day 10 post infection for RNA extraction and deep sequencing of small RNAs.

(B and C) Size distribution of the total number of DENV-1-specific small RNA reads normalized to the total number of reads upon (B) TSN knockdown or (C) luciferase control.

(D–F) Linear relationship between the amount of viral and cellular small RNAs in TSN knockdown versus luciferase control mosquitoes 10 days post DENV-1 infection. Data points represent the normalized number of reads (coverage per 1M reads) corresponding to (D) miRNAs, (E) endo-siRNAs mapping on histone coding sequences, and (F) endo-piRNAs mapping on histone coding sequences. For miRNAs, each dot represents one miRNA. For siRNAs and piRNAs mapping on histone genes, each dot represents a histone gene. For siRNAs and piRNAs mapping on the DENV-1 sequence, each dot represents a 500-bp region of the viral genome. Lines represent the linear regression of each set of values. The equation and R² value of each regression are shown next to the line.

See also Figure S3.

Discussion

Antiviral immunity in Ae. aegypti mosquitoes remains poorly understood. Using a novel approach of transcriptomic analysis, we previously uncovered four genes that not only responded to DENV infection in the mosquito midgut but also had expression levels that correlated with viral loads in infected mosquitoes (Raquin et al., 2017). Here, we focused on one of these four genes, Tudor-SN, encoding a member of the Tudor protein family. TSN was induced upon DENV-1 infection, and its expression correlated positively with viral RNA load (Raquin et al., 2017). Using RNAi-mediated knockdown assays in vivo and in vitro, we demonstrated that TSN promotes both DENV-1 and DENV-3 replication. TSN knockdown also resulted in a slight decrease of CHIKV replication, but it was not statistically significant. We performed localization studies and discovered that TSN localizes to the nucleolus of Ae. aegypti and Ae. albopictus cells and does not colocalize with DENV-1 replication sites. Moreover, we found that, despite belonging to the Tudor family, TSN was not essential for RNAi function in adult mosquitoes.

TSN is a known component of the RISC, the RNAi protein complex that carries siRNAs and directs cleavage of complementary viral sequences in Caenorhabditis elegans, Drosophila, and mammals (Caudy et al., 2003). Additionally, previous work in Drosophila and other model organisms found essential functions for Tudor domain-containing proteins in piRNA biogenesis. A recent study describing a functional knockdown screen of all predicted Ae. aegypti Tudor proteins did not reveal a role for Tudor-SN in piRNA biogenesis (Joosten et al., 2019). This finding is consistent with our observations that Tudor-SN is not necessary for the piRNA pathway function in vivo in Ae. aegypti. Finally, Tudor-SN was also shown to be a conserved component of the basic RNAi machinery in Ixodes ticks (Ayllon et al., 2015). This study reported an effect of Tudor-SN on dsRNA-mediated gene silencing, which possibly involves the siRNA pathway. However, no evidence was obtained for a role of Tudor-SN in the response to microbial infection (Ayllon et al., 2015). Taken together, the data described in this study and discussed above did not find strong links between RNAi function and Tudor-SN in arthropods. Further studies using knockout mutants might be necessary to confirm these findings.

The mammalian ortholog of Tudor-SN is generally referred to as p100 and was identified as a host factor interacting with the 3′ untranslated region of the DENV genome (Lei et al., 2011). Moreover, p100 knockdown led to reduced levels of viral RNA and protein in mammalian cells, providing evidence that p100 was required for efficient DENV replication. Although these results are consistent with those reported here, in mammalian cells p100 was shown to interact with DENV genomic RNA and dsRNA replication intermediates, which we did not observe. Importantly, the subcellular localization of p100 in mammalian cells was perinuclear, whereas it was nucleolar in mosquito cells. This discrepancy in localization hints to a divergence in function between mammals, in which p100 interacts directly with viral RNA, and insects, for which evidence is lacking. Microscopy-based localization studies being limited in sensitivity and resolution, elucidating Tudor-SN function in mosquitoes will require further experiments to more definitely exclude interactions between viral RNA and Tudor-SN, such as protein immunoprecipitation and sequencing of associated RNA (RIP-seq).

The predicted structure of Ae. aegypti Tudor-SN includes four Staphylococcal Nuclease (SN)-like domains and a Tudor domain embedded in a fifth SN domain. Tudor-SN homologs are found in diverse eukaryotic species such as plants, humans (Staphylococcal Nuclease and Tudor domain containing 1), and insects (Drosophila Tudor-SN). The very similar structure of eukaryotic Tudor-SN homologs is consistent with potentially conserved functions (Figure 7). However, Tudor-SN subcellular localization is variable in other species, which also hints toward species-specific function(s) of Tudor-SN proteins.

Figure 7.

Conserved Structure of TSN Homologs Among Diverse Eukaryotic Species

Structural domains provided in the UniProt database are shown for homologs of Ae. aegypti TSN (Q17PM3) in Drosophila melanogaster (Q9W0S7), Ixodes scapularis (B7QIP4), Arabidopsis thaliana (F4K6N0), and Homo sapiens (Q7KZF4). Diagrams on the right side indicate TSN subcellular localization (in green) in the cytoplasm, nucleus, and/or nucleolus of each species based on the literature (Fashe et al., 2013, Ku et al., 2016, Gutierrez-Beltran et al., 2015, Frei Dit Frey et al., 2010, Caudy et al., 2003).

Our observation that Ae. aegypti Tudor-SN localizes primarily to the nucleolus of mosquito cells makes it unlikely that its proviral effect on DENV relies on a direct action on viral genome stability or replication. The nucleolus is a multifunctional nuclear domain involved in ribosome biogenesis and several other cellular functions, such as cell cycle regulation, telomere metabolism, or DNA damage sensing and repair (Lam and Trinkle-Mulcahy, 2015). Various nucleolar alterations during viral infection have been documented (Salvetti and Greco, 2014). Interestingly, DENV non-structural protein 5 (NS5), which encodes the virus RNA-dependent RNA polymerase, was recently shown to localize to the nucleolus of infected mammalian cells (Fraser et al., 2016), where it interferes with precursor messenger RNAs (pre-mRNA) splicing to limit host antiviral response (De Maio et al., 2016). Human Tudor-SN was implicated in spliceosome assembly and, therefore, may influence splicing of pre-mRNAs and/or interact with DENV NS5 to facilitate viral RNA accumulation (Gao et al., 2012). More generally, Tudor-SN could promote viral replication through regulation of gene expression. For example, the Jak-Stat pathway protects Ae. aegypti against DENV infection (Souza-Neto et al., 2009) and Tudor-SN was shown to bind Stat proteins to modulate host gene transcription (Paukku and Silvennoinen, 2004). Also, Tudor-SN is a component of stress granules (Gao et al., 2014) and could interfere with their formation to facilitate DENV infection (Miller, 2011). Particularly interesting is the early proviral effect of TSN in the mosquito midgut, which might suggest a role for TSN in viral sensing, or early antiviral responses. For instance, TSN might sense the infection and, in the nucleus, alter the spliceosome to increase the availability of cellular resources that the virus requires to replicate. Its presence in the nucleolus and at the nucleus-nucleolus interface might enhance ribosome biogenesis and subsequently increase production of viral proteins.

Although several pathways involved in antiviral immunity have been characterized in mosquitoes, several aspects of anti-DENV defense remain to be elucidated, particularly during the early phase of infection. For example, the siRNA pathway was recently shown to inefficiently restrict DENV replication in the Ae. aegypti midgut (Olmo et al., 2018). The present work adds to the small number of studies that identified DENV proviral factors in mosquitoes (Londono-Renteria et al., 2015, Jupatanakul et al., 2014, Sessions et al., 2009, Raquin et al., 2017). Such host factors have been proposed as new targets for antiviral therapy in humans (Savidis et al., 2016, Zhang et al., 2016, Marceau et al., 2016). Although the development of novel vector control methods has focused on viral restriction factors so far (Flores and O'neill, 2018), targeting essential host factors could complement antiviral strategies in mosquitoes. We showed that Tudor-SN is such a host factor for DENV in the mosquito Ae. aegypti. Further studies will be necessary to elucidate the specific mechanisms underlying the role of Tudor-SN in DENV replication.

Limitations of the Study

We demonstrated a proviral role of TSN upon DENV infection in the mosquito midgut. Our study primarily relied on gene knockdown to diminish TSN expression in vitro and in vivo, and it will be useful in the future to confirm the results using knockout mosquitoes. Second, this study used a strain of Ae. aegypti from Thailand, and it remains to be determined whether the proviral role of TSN extends to other mosquito strains. Finally, additional in-depth functional studies are required to elucidate the exact role of TSN.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: February 21, 2020

Data and Code Availability

The accession number for the RNA-seq data set reported in this paper is SRA: PRJNA386455.