Introduction

Japanese encephalitis virus (JEV), endemic to most South Asian countries and the Pacific Islands, has recently emerged as a significant public and animal health threat in Australia. According to estimates from the World Health Organization, approximately 68,000 cases of Japanese encephalitis and an estimated 15,000 deaths occur annually worldwide1. It is prevalent in the Western Pacific, including the Philippines, Indonesia, and the northern islands of Australia, such as the Tiwi Islands and Badu Island2. JEV is a single-stranded positive-sense RNA virus belonging to the genus Orthoflavivirus (family Flaviviridae). Five different genotypes of the virus have been recognised (GI-GV)3. Genotypes I and III are the dominant genotypes in Asia; however, the last decade has been associated with the emergence of previously rare genotypes IV and V and the expansion of JEV into previously naïve regions4.

While most human JEV infections are asymptomatic or present with mild flu-like symptoms, some cases are associated with encephalitis characterised by headaches, fever, seizures, and abnormal behaviours5. These can be followed by haemorrhagic lesions in the brain, paralysis, and death (can be as high as 30%). Additionally, those who recover may experience long-term neurological disorders such as mental impairment and physical illness5. Similar to several other flaviviruses, such as dengue virus (DENV) and Zika virus, JEV is transmitted through mosquito bites. The virus from an infectious blood meal enters a mosquito’s midgut cells, where it replicates and is then released into the mosquito’s haemocoel, infecting other mosquito tissues, most notably the salivary glands. The virus is then transmitted to other vertebrate hosts in saliva during subsequent blood-feeding events.

In the natural cycle of JEV, the virus is transmitted between waterbirds, which are the natural reservoirs, and Culex mosquitoes. Pigs serve as amplifying hosts. While infected sows show mild or no symptoms, if pregnant, they may abort or give birth to mummified, stillborn, or weak piglets, some of which may exhibit neurological symptoms5. Piglets infected after birth may develop encephalitis. Additionally, transmission between pigs can occur through virus excretion in saliva and nasal secretions6. Humans, cattle, and horses can become infected by JEV-infected mosquitoes and develop encephalitis, but these are regarded as dead-end hosts5. Vaccines against JEV exist, but they are expensive, their availability is limited, and concerns have been raised about their effectiveness in protecting against different genotypes of the virus and the longevity of their protection7,8. Currently, there are no approved anti-viral medications for JEV.

An unprecedented outbreak of JEV occurred across Australia in 2022, creating a public health emergency for the population that is largely naïve to the virus9. Previously, small, sporadic outbreaks of JEV genotypes I and II occurred that were geographically restricted to far northern Australia. In 2022, an outbreak of JEV genotype IV occurred across most mainland states, which was associated with 44 confirmed human cases and seven deaths. The rapid range expansion is thought to have occurred due to the migration of JEV-infected wading birds to temporary wetlands that emerged after La Niña rainfall and the subsequent infection of competent local mosquito species.

Fourteen mosquito species have been confirmed as vectors of JEV, while 11 species are considered potential vectors10. Among the confirmed vectors, JEV was isolated from four species in Australia: Aedes vigilax, Culex annulirostris, Cx. gelidus, and Cx. sitiens, with Cx. annulirostris being implicated as the main Australian vector11,12. A recent investigation has shown that an Australian strain of Culex annulirostris is a competent vector of the JEVNSW/22 strain, which emerged in Australia in 202213. This mosquito is widespread in mainland Australia, can fly around 5 km per day, and serves as a highly competent vector for several other viruses in the region, including Murray Valley encephalitis virus, Ross River virus, and Barmah Forest virus14. Although transcriptomic analyses have explored the effects of viral infections in various mosquito vectors, similar insights into the JEV-mosquito interaction remain unavailable. Notably, the genome and transcriptome sequences for Cx. annulirostris, a significant vector of JEV, are not presently available.

The initial step in viral infection involves virions attaching to the host’s cellular receptor(s), which facilitates viral endocytosis. The specificity of the virus for the receptor(s) contributes to the host range and tissue tropism. Identifying viral receptors is crucial for developing anti-viral treatments that target viral entry. However, the cellular receptor for JEV in mosquitoes remains unknown.

In this study, we determined the transcriptome of Cx. annulirostris through de novo assembly of RNA-Seq data and identified differentially expressed mosquito genes associated with JEV infection for the first time. Furthermore, we identified potential mosquito receptor/entry factors of JEV in Culex cells by overexpressing JEV’s envelope protein, affinity-purifying the potential receptors/entry factors, and sequencing interacting mosquito proteins. Overall, this study has highlighted the JEV-mosquito interactions at the molecular level and identified novel candidates for transmission-blocking interventions.

Results

De novo transcriptome assembly of Cx. annulirostris

In the absence of a genome sequence for Cx. annulirostris, we de novo assembled the mosquito’s transcriptome using the 18 RNA-Seq libraries generated in this study. These included JEV-infected and uninfected mosquitoes, which were used to determine differential gene expression (see below). We obtained between 40,433,248 and 51,447,298 raw reads per library (Supplementary Fig. 1A). After trimming, there were 36,339,961 to 47,063,713 clean reads per library. All clean reads from the 18 libraries were combined using a Trinity-based de novo assembler in OmicsBox to produce a single, combined reference transcriptome assembly for Cx. annulirostris. Out of about 716 million trimmed reads, a total of 137,427 contigs were assembled, of which approximately 27,317 contained potential coding regions (14,847 complete and 12,470 partial). The assembly had an N10 of 4433 bp, an N50 of 1593 bp, and a median sequence length of 936 bp, indicating the high quality and completeness of the transcriptome. The length of the assembled supertranscripts ranged from 500 to 26616, with 56774 transcripts >1000 bases. The completeness of the assembled transcriptome was assessed by BUSCO (benchmarked universal single-copy orthologs), with 2105 (64.08%) complete single-copy genes and 178 (5.42%) complete duplicated genes (Supplementary Fig. 1B). The remaining transcripts included 415 (12.63%) fragmented and 586 (17.87%) missing transcripts (returned as not classified).

Differential gene expression of Cx. annulirostris mosquitoes following JEV infection

To evaluate the impact of JEV infection on the transcriptome of Cx. annulirostris mosquitoes, female mosquitoes were fed blood with or without JEV. They were then collected at 2, 7, and 12 days post-inoculation, followed by RNA extraction and RNA-Seq analysis (Fig. 2A). The selected timepoints span early, mid, and late stages of JEV infection in Cx. annulirostris, with 12 dpi chosen to capture the late dissemination phase. To confirm the successful infection of mosquitoes with JEV, an Indirect Fluorescent Assay (IFA) was conducted. The results showed progressively higher levels of viral dissemination in mosquito sections (Fig. 1). In addition, reads from the infected libraries were mapped to the JEV genome. Results demonstrated a gradual and significant increase in JEV reads over time (Fig. 2B), with 3.5% of reads mapped to the viral genome at day 12 (Supplementary Data 1). No reads mapped to the viral genome in mock-infected libraries (Supplementary Data 1). The top BLAST hit species in our analysis included Culex fatigans, Culex pipiens pallens, and Aedes albopictus. Their presence as top hits suggests a high level of similarity with closely related species in the analysed sequences, providing confidence in the relevance and accuracy of the assembled transcriptome.

Fig. 1: Detection of JEV infection in Cx. annulirostris mosquitoes using Indirect Fluorescent Assay.
figure 1

Mosquito sections collected at various times post-JEV infection were processed by IFA using an anti-flavivirus NS1 antibody 4G4. m midgut, th thorax, h head. Green fluorescence (indicated with arrows) represents JEV-related signals. DNA is stained using DAPI (blue).

Full size image
Fig. 2: Confirmation of JEV infection in Cx. annulirostris mosquitoes and the differential expression of genes following JEV infection.
figure 2

A A schematic diagram of the experimental workflow. B The Violin plot illustrates an increase in the number of reads mapped to the JEV genome at 2, 7, and 12 days post-JEV infection. One-way ANOVA with Tukey’s multiple comparisons test was used to compare the data. Asterisks denote significant differences between the compared samples. *p < 0.05; ***p < 0.001; ****p < 0.0001. C Volcano plot analysis. Red circles represent differentially expressed genes in response to JEV infection (Fold change >2 and FDR < 0.05). Green circles indicate the top ten DEGs, which are labelled. DPI, days post. D Venn diagram illustrating the number of differentially expressed genes at three distinct time points after JEV infection.

Full size image

Out of the 27,317 assembled transcripts, 1008 showed differential expression (FC > 2, p < 0.05) in at least one of the conditions, comprising 350, 322, and 451 genes at 2, 7, and 12 dpi (Fig. 2C and Supplementary Data 2). Principal component analysis (PCA) revealed the separation of infected and uninfected samples at each time point (Supplementary Fig. 2). Four genes were differentially expressed across the three time points, with 26 to 52 genes overlapping between different time points (Fig. 2D). The four overlapping genes are uricase (urate oxidase), glycine-rich cell wall structural protein 1.8-like, cytochrome P450, and an uncharacterised protein K02A2.6-like. The first two genes were upregulated, while the latter two were downregulated across the three time points post-infection.

Among the DEGs, several odorant-binding protein genes and odorant receptors were differentially expressed. In total, 14 genes were differentially expressed across the three time points following infection; eight were downregulated at 2 dpi, five were downregulated and one was upregulated at 7 dpi, and five were downregulated while two were upregulated at 12 dpi. Overall, across the three time points, the genes for odorant-binding proteins and odorant receptors were predominantly downregulated.

Regarding immune-related genes, lysozyme was upregulated at all time points post-infection, while spätzle, serpin1, defensin A, and neprilysin were upregulated at 12 dpi. Glutathione-S-transferase and galectin were upregulated at 7 dpi. A scavenger receptor and phenoloxidase activating factor 2 were upregulated and downregulated, respectively, at 2 and 7 dpi. A C-type lectin was upregulated at both 2 and 12 dpi, and melanization protease and dopamine D2-like receptor were both upregulated at 2 dpi. Eleven leucine-rich repeat-containing proteases were differentially expressed (1 up and 1 down at 2 dpi; 3 up and 3 down at 7 dpi; 4 up and 2 down at 12 dpi). Seven clip-domain containing serine proteases were differentially expressed (1 up at 2 dpi; 1 up and 1 down at 7 dpi; 3 down and 2 up at 12 dpi). A fibrinogen-related gene was also upregulated at 12 dpi.

Among the metabolic-related genes, uricase was found to be upregulated, while a zinc metalloprotease was downregulated across the three time points. GATA exhibited downregulation at 2 and 12 dpi, whereas vitellogenin showed upregulation at 2 and 7 dpi. Both trypsin1 and chymotrypsin1 were upregulated at 2 and 7 dpi, and a cytochrome P450 was upregulated at 12 dpi. An ATP synthase gene was also upregulated at 12 dpi in JEV-infected mosquitoes, alongside several mitochondrial-related genes that were differentially expressed.

Gene ontology analysis of differentially expressed genes

All the differentially expressed genes were submitted to OmicsBox for Gene Ontology (GO) analysis to functionally group the DEGs. The majority of sequences (55.36%, 558 sequences) are fully annotated with GO databases, including InterPro, EggNOG, and enzyme code (EC) mapping, providing detailed functional insights into biological processes, molecular functions, and cellular components. A smaller proportion (28.27%, 285 sequences) represents sequences with significant BLAST hits, indicating homology to known sequences in databases and serving as a basis for further functional predictions. The remaining 16.37% (165 sequences) have undergone GO mapping, establishing connections to functional categories but without complete annotation. Accordingly, 328, 292, and 114 GO terms were identified in biological process, molecular function, and cellular components, respectively. The most common categories among DEGs were associated with metabolic process, signalling, small molecule binding, hydrolase activity, intracellular anatomical structure, membrane, organelle, extracellular region, cytoplasm, and cell periphery, among others (Supplementary Fig. 3). Fisher’s exact test highlighted the most overexpressed GO terms for each time point. At 2 dpi, the most enriched genes were related to cellular functions, including multicellular organismal process, and membrane-bounded organelle, with a lesser extent to sensing, including odorant binding, sensory perception of chemical stimulus, sensory perception, and nervous system (Fig. 3). At 7 dpi, cellular functions, including multicellular organismal process, and membrane-bounded organelle, were also among the most enriched genes, along with genes related to RNA polymerase and transcription, such as cis-regulatory region sequence-specific DNA binding, RNA poly II transcription regulator, and DNA-binding transcription factor activity, RNA poly II-specific (Fig. 3). At 12 dpi, most of the enriched differentially expressed genes were related to metabolism, such as protein metabolic process, catalytic activity, proteolysis, peptidase activity, endopeptidase activity, extracellular region, serine-type endopeptidase activity, and monooxygenase activity (Fig. 3).

Fig. 3: Gene Ontology (GO) enrichment analysis of differentially expressed genes in Cx. annulirostris mosquitoes upon JEV infection using Fisher’s exact test highlights the most overexpressed GO terms for each time point.
figure 3

Point size represents the number of enriched genes in each term, and the red-coloured scale indicates the significance by p value.

Full size image

Identification of the Culex cellular entry factor for JEV

To identify potential mosquito cell entry factors of JEV, we first expressed His-tag JEV E protein in E. coli. The JEV E protein was detected by Western blot with antibodies to flavivirus E protein at 4 h post-IPTG induction and not at 0 h (Fig. 4A). The molecular weight was close to the predicted size, with a His-tagged of about 55 kDa recognised by a monoclonal antibody to the E protein. We tested two available monoclonal antibodies, M2-8E7 and 4G2, with M2-8E7 displaying stronger reactivity. The JEV E protein was then purified with Ni-NTA resin (Fig. 4B), and 1μg of purified protein was mixed with 1 mg of Hsu cell lysate to allow interactions of JEV protein with cell proteins. M2-E87 monoclonal antibody was then added to co-immunoprecipitate cell proteins bound to JEV E protein. The co-immunoprecipitated proteins were resolved on an SDS-PAGE, with bands specific to the anti-E protein at 22, 36, and 60 kDa being cut and processed for mass spectrometry. Several proteins were sequenced and identified by MS (Supplementary Data 3). Among these, three proteins (annexin, 14-3-3ε, and prohibitin), which have the highest hits in the 36 kDa band, have previously been shown to be involved in host-virus interactions, either as receptors or in facilitating the entry of viruses15,16,17. Therefore, we focused on these three proteins to determine their potential role in JEV entry into Culex cells.

Fig. 4: Overexpression of JEV E protein.
figure 4

A Western blot analysis of E. coli overexpressing JEV E protein at 0 h and 4 h after induction with IPTG. Bacterial lysates were run in duplicate, and the blot was then cut in half, where the molecular weight (MW) marker was included. Subsequently, the blots were probed using the two available monoclonal antibodies against flavivirus E proteins (4G2 and M2-8E7). B SDS-PAGE analysis of the overexpressed, purified E protein using the Ni-NTA resin beads.

Full size image

RT-qPCR analysis of RNA from Hsu cells infected with JEV demonstrated significant induction of annexin (Kruskal-Wallis, p = 0.0088), 14-3-3ε (One-way ANOVA, p < 0.0001), and prohibitin (One-way ANOVA, p < 0.0001) at 1 and 2 dpi (Fig. 5A–C). An increase in virus replication in the samples was also confirmed by titrating the virus (p < 0.0001) (Fig. 5D). The results also showed a statistically significant linear relationship between the expression levels of the genes and JEV levels (Fig. 5E–G). These findings suggest a potential role of the three proteins in JEV infection of Hsu cells. Among the three genes, only annexin showed significant increases in read numbers between mock and JEV-infected mosquitoes at 7 and 12 dpi (1.5 and 2.25 folds; Supplementary Data 2). However, the cell line results are not directly comparable with whole mosquitoes, as a cell line comprises more homogeneous cells than mosquitoes, which consist of various tissues that become infected at different stages of infection.

Fig. 5: Transcriptional changes of annexin, 14-3-3ε, and prohibitin following JEV infection.
figure 5

A–C RT-qPCR analysis of RNA extracted from Hsu cells after infection (1 MOI) with JEV at 1 and 2 days post-infection (dpi) using primers for the three genes. The graphs illustrate gene fold changes after infection compared to the mock. D Focus forming assay of media collected from cells used in (A–C), quantifying JEV virions. Kruskal-Wallis with Dunn’s multiple comparisons test was used for A, and One-way ANOVA with Tukey’s multiple comparisons test for (B–D). The error bars indicate the standard error of the mean (SEM) of biological replicates. Asterisks denote significant differences between compared samples. ns, not significant; *p < 0.05; ***p < 0.001; ****p < 0.0001. E–G General linear regression analysis was performed for the three genes and JEV replication on all data points (n = 12). Error bars indicate 95% confidence interval of the linear regression. E–G graphs correspond to (A–C) graphs. The raw data for the qPCR reactions are provided in Supplementary Data 4.

Full size image

Silencing the genes for JEV E-interacting proteins reduced JEV replication

To assess the impact of knocking down the putative entry factor genes of JEV on its replication, dsRNAs were synthesised against prohibitin, annexin, and 14-3-3ε. Following two subsequent transfections of dsRNA into Hsu cells (2 days each), the cells were infected with 1 MOI JEV and harvested at 2 dpi. RT-qPCR analysis of RNA extracted from the cells confirmed knockdown of the genes (Fig. 6A–C; annexin 97%, 14-3-3ε 92%, prohibitin 79% reductions). RT-qPCR analysis also revealed significant reductions (t-test; p < 0.0001) in genomic RNA replication of JEV compared to dsGFP for all three genes (Fig. 6D). These findings suggest that these proteins act as entry factors or are involved in the virus replication process.

Fig. 6: Effect of silencing the potential entry factors of JEV on its replication and antibody neutralisation assay.
figure 6

dsRNAs were designed against annexin, 14-3-3ε, and prohibitin and double transfected into Hsu cells. A–C Confirmation of silencing of the three genes using RT-qPCR. D Two days after the second transfection, cells were infected with 1 MOI of JEV and analysed 2 dpi using RT-qPCR. The data presented are combined from two independent experiments. E Antibody neutralisation assay using anti-prohibitin and anti-Flag (control) alongside JEV infection. In Mock+JEV, no antibody was used. RT-qPCR was used to determine the fold changes in the gRNA levels of JEV. Unpaired t-test was used for data analysis in (A–D), and One-way ANOVA with Tukey’s multiple comparisons test was used for (E). The error bars indicate the standard error of the mean (SEM) of biological replicates. Asterisks denote significant differences between compared samples. *p < 0.05; ***p < 0.001; ****p < 0.0001. The raw data for the qPCR reactions are provided in Supplementary Data 4.

Full size image

Incubation of Hsu cells with anti-prohibitin antibodies reduced JEV infection

Given the availability of an anti-prohibitin antibody, we conducted an antibody neutralisation assay by incubating Hsu cells with anti-prohibitin antibody and an anti-FLAG antibody as a control. After two hours, the medium was removed and the cells were infected with JEV, then collected 24 h post-infection. RT-qPCR analysis of the RNA extracted from the cells showed a significant reduction (One-way ANOVA, p = 0.0003) in JEV gRNA levels in the cells incubated with anti-prohibitin antibodies compared to the controls (Fig. 6E). These results suggest that prohibitin may act as an entry factor for JEV in Culex cells.

Discussion

While several studies have investigated the effect of JEV on the transcriptional response of mammalian hosts to infection (reviewed in ref. 18), to our knowledge, none have focused on mosquito infection. Furthermore, there are limited reports on JEV’s molecular interactions with its mosquito vectors, including receptors or entry factors for JEV. Currently, 17 species are confirmed as vectors for JEV, with another 10 species identified as potential vectors. In endemic areas, Cx. tritaeniorhynchus and Cx. annulirostris are recognised as the primary vectors for JEV19,20. Given the significance of Cx. annulirostris in JEV transmission and the lack of molecular information concerning JEV and the mosquito, we aimed to shed light on this knowledge gap.

We generated the first de novo assembled transcriptome of Cx. annulirostris, comprising 137,427 assembled sequences, with 27,317 containing potential complete or partial coding regions. Among these 27,317 genes, 1008 displayed differential expression at least at one of the post-infection time points. Similar to previous studies investigating the effects of viral infection on mosquito transcriptions, including DENV and ZIKV infections in Aedes aegypti mosquitoes21,22, DEGs in response to JEV infection in Cx. annulirostris comprised immune-related genes, metabolism, cellular processes, and odorant-binding proteins. Based on GO analysis, the enriched genes showed a trend over time. At 2 dpi, the enriched DEGs are related to cellular functions, such as multicellular organismal process, and membrane-bounded organelle, and to a lesser extent, to sensing. At 7 dpi, the enriched genes are more related to cellular functions, including multicellular organismal process, and membrane-bounded organelle, as well as genes associated with RNA polymerase and transcription. At 12 dpi, most of the enriched DEGs are related to metabolism and enzyme activities.

An ATP synthase gene was upregulated at 12 dpi in JEV-infected mosquitoes, and several genes related to mitochondria were differentially expressed. Viral infections may affect the morphology and function of mitochondria, causing an imbalance between fusion and fission (reviewed in ref. 23). These events occur due to changes in the expression of genes involved in the fission or fusion of mitochondria or the relocation of proteins from the cytoplasm into mitochondria. Elongated mitochondria were observed in Ae. aegypti Aag2 cells infected with chikungunya virus (CHIKV), an alphavirus24. The authors found that proteins related to mitochondrial function were differentially present in cells infected with CHIKV, including those associated with ATP synthesis. In Ae. albopictus C6/36 cells infected with DENV, fusion of mitochondria is also supported25. The effects of JEV infection on the morphology and function of mitochondria in mosquitoes remain to be determined.

Cathepsins are lysosomal proteases that participate in various cellular functions, including protein degradation, energy metabolism, apoptosis, immune responses, and virus-host interactions26. The cathepsin L-like peptidase in Ae. aegypti interacts with the antiviral protein, Aecystatin, which is induced in the mosquito’s salivary glands following DENV infection, and has been shown to correlate negatively with DENV titre27. Cathepsin-B was also upregulated at 8 and 24 hpi in Ae. aegypti mosquitoes infected with DENV-2. Silencing Cathepsin-B in mosquitoes led to reduced DENV-2 replication. In another study, the silencing of Cathepsin-B also led to lower DENV titers in the salivary glands of Ae. aegypti mosquitoes28. This suggests that the protein may facilitate DENV-2 replication in the mosquito. In this study, the cathepsin-B homologue was also induced in JEV-infected mosquitoes at 2 and 7 dpi.

Based on several studies investigating the effects of flaviviral infections on mosquito transcriptomes, specifically DENV and ZIKV in Ae. aegypti, researchers commonly find differential expression of immune-related genes21,22. These genes include those associated with major immune pathways, such as Toll and IMD, as well as the melanisation pathway. Likewise, several immune-related genes, including spätzle, serpin1, defensin A, phenoloxidase activating factor 2, and clip-domain containing serine proteases, were identified as differentially expressed in Cx. annulirostris mosquitoes infected with JEV. Similar to ZIKV infection in Ae. aegypti22, the RNAi and JAK/STAT pathway genes did not change in JEV-infected Cx. annulirostris mosquitoes.

In ZIKV-infected Ae. aegypti mosquitoes, the expression of several genes coding for odorant-binding proteins was suppressed at 2 dpi22. Similarly, we found that eight and five odorant-binding protein genes were downregulated at 2 and 7 dpi, respectively, while two were upregulated at 12 dpi. It is speculated that virus infection may reduce host-seeking behaviour during the early stages of infection and promote host-seeking when the mosquito becomes infectious (around 12–14 dpi)28. In Plasmodium-infected Anopheles mosquitoes, host-seeking behaviour is also suppressed early in the infection and increases at later stages when the mosquitoes can transmit the parasite29,30. Therefore, manipulating host behaviour during mosquito infection may be a common phenomenon to maximise the spread of pathogens transmitted by vectors.

Using common characteristics of DENV and West Nile virus (WNV) receptors on mosquito cells, Ren et al. speculated that a 74 kDa heat shock cognate protein 70 (HSC70) could function as a receptor for JEV, although no experimental evidence was provided to support this claim31. However, another study showed that HSC70B was significantly downregulated by JEV infection, suggesting it may have an anti-viral role. The antimicrobial peptide defensin was found to enhance JEV infection by facilitating its adsorption and entry through the downregulation of HSC70B in C6/36 cells32. Therefore, HSC70 does not seem to be the receptor for JEV in mosquito cells. In mammalian cells (mouse neuronal Neuro2a cells and human epithelial Huh-7 cells), glucose regulated protein 78 (GRP78), a member of the HSP70 family, facilitates JEV entry into the cells33. Boonsanay and Smith performed an overlay assay of membrane proteins using Ae. albopictus C6/36 cells and found several protein bands that bound to JEV. However, their mass spectrometry did not identify any proteins34. To date, no reports exist regarding JEV receptors in mosquito cells. Using co-immunoprecipitation combined with mass spectrometry, we identified three Culex proteins interacting with the recombinant JEV E protein. These were prohibitin, annexin, and 14-3-3ε. All three of these genes were upregulated in JEV-infected Culex Hsu cells and their silencing, using RNAi, reduced JEV replication in the cells. The induction of genes that may be proviral, along with the corresponding relationship between the abundance of these genes and viral gRNA, provide additional evidence that they are potential entry factors. Furthermore, antibody neutralisation with anti-prohibitin antibodies resulted in reduced JEV replication.

14-3-3 proteins are conserved scaffold proteins that interact with many other proteins involved in signalling pathways related to several biological processes such as the cell cycle and proliferation, apoptosis, and immunity (reviewed in ref. 35). In insects, particularly the Ae. aegypti mosquito, two isoforms of the protein have been identified, Ɛ and ζ36. The role of 14-3-3 proteins as host factors in the replication of several RNA and DNA viruses in mammalian systems has been documented. Examples include Hepatitis C virus, Herpes Simplex virus Type 1, Parainfluenza 5 virus, Human Immunodeficiency virus, Epstein-Barr virus, JEV, ZIKV and coronavirus (reviewed in ref. 17). However, these roles are mainly intracellular, involving participation in cellular pathways or binding to viral proteins to suppress anti-viral immunity rather than binding and internalising viruses. Whether 14-3-3ε plays a role in the replication of JEV requires further investigation.

Annexins, particularly Annexin II, are implicated in infections caused by various viruses. Human papillomavirus (HPV), enterovirus 71, respiratory syncytial virus, and cytomegalovirus utilise Annexin II during cell attachment and penetration (reviewed in ref. 37). Annexin II has been identified as a cellular protein that interacts with DENV-2 in Vero cells, facilitating binding to these cells. Antibodies against this protein, when used for antibody-mediated inhibition or its depletion, reduced DENV-2 infection and production levels15. This protein also facilitates the entry and intracellular trafficking of HPV type 16 in human keratinocytes38. Annexins are multifunctional proteins that contribute to endocytosis, trafficking, and the organization of vesicles during endocytosis and exocytosis, as well as in ion channel formation, immunity, and apoptosis39,40.

Prohibitins are conserved proteins involved in various cellular functions, including mitochondrial function and maintenance, senescence, apoptosis, cell proliferation, and immunity (reviewed in ref. 41). In addition to being present in the inner mitochondrial membrane, the nucleus, and the cytoplasm42,43, prohibitin is also found in the cell membrane44,45 of mammalian cells. In insects, prohibitin has been found on the cell membrane of Spodoptera frugiperda Sf9 cells46, Helicoverpa armigera47, and Ae. aegypti midgut cells48.

In addition to the functions mentioned, prohibitin has been shown to act as a receptor for certain viruses, including DENV16,49, and chikungunya virus (CHIKV)50. Paingankar et al. (2010) identified six proteins that bound to DENV-2 in mosquito samples, one of which was prohibitin51. In their study, the expression of the prohibitin gene was reduced in C6/36 cells following DENV-2 infection. This contrasts with the observations in Cx. quinquefasciatus Hsu cells infected with JEV, which may be attributed to variations in both the host and virus used in the respective studies. The authors hypothesised that prohibitin may serve as a modulator of innate immunity in mosquitoes rather than merely as a receptor. In another study, Kuadkitkan et al. demonstrated that prohibitin acts as a receptor protein mediating DENV-2 entry into mosquito cells16. The protein bound to DENV-2 in Ae. albopictus mosquito C6/36 cells and CCL-125 cells as well as adult Ae. aegypti mosquitoes. The protein was shown to co-localise and co-immunoprecipitate with DENV-2 E protein in C6/36 cells. Similar observations have been made in mammalian cells. In a proteomic study involving DENV-2 and DENV-4 with HEK293T/17 (human embryonic kidney) cells, six proteins were found to be coordinately regulated in both serotypes49. Among these, prohibitin 1 was downregulated by infection with both serotypes of DENV. This is similar to what was observed in mosquito cells with DENV-251. When prohibitin was overexpressed, replication of both DENV strains increased. Further, the study showed the co-localisation of DENV E protein with prohibitin and a direct interaction of the protein with E protein of DENV-2, but not DENV-4. Similarly, prohibitin was identified as a receptor for CHIKV, interacting with its E protein in the human microglial CHME-5 cell line50.

In summary, we identified differentially expressed genes in Cx. annulirostris mosquitoes in response to JEV infection based on the first de novo assembled transcriptome of this significant vector of JEV. The majority of the DEGs common to the three time points following infection are involved in membrane-bounded organelle, protein metabolic process, catalytic activity, proteolysis, and peptidase activity. Furthermore, immunoprecipitation and mass spectrometry identified prohibitin, annexin, and 14-3-3ε as mosquito-interacting proteins with JEV’s E protein, which could facilitate its entry or replication in mosquito cells. An antibody neutralisation assay using an anti-prohibitin antibody demonstrated decreased JEV replication. As our findings were determined from bulk RNA sequencing of whole mosquitoes, a limitation of the study, further insights into the mosquito immune response to JEV infection may be gained using high-resolution RNA sequencing methods, including single-cell RNA sequencing (scRNA-Seq)52. Overall, the findings from this study are the first to be reported in the context of JEV-mosquito molecular interactions and provide valuable insights for future investigations in this field of research.

Methods

Mosquitoes, mosquito cell line, and virus infection

Culex annulirostris mosquitoes were maintained at 28 °C, with a 12 h light/12 h dark cycle that included 30 min crepuscular periods, and at 75% relative humidity in the QIMR Berghofer insectary. They were provided with a 10% sugar solution on cotton wool pledgets. The mosquitoes were starved for 24 h prior to the virus challenge. Female mosquitoes (2 to 5 d old) were aspirated from the cage, and groups of approximately 60 mosquitoes were placed into 750 ml plastic containers with gauze lids. Three containers of mosquitoes were fed per treatment (JEV or mock treatment). Within the QIMR Berghofer biosecurity level 3 insectary, each container of mosquitoes was provided with 2 ml of JEVNSW/22 culture supernatant or cell culture medium only (mock treatment) in washed defibrinated sheep blood, at a ratio of 1:4 virus (or media) to blood, using glass feeders53. JEVNSW/22 (GenBank: OP904182) was obtained from an infected porcine neonate by Dr. Peter Kirkland (Elizabeth Macarthur Agriculture Institute, New South Wales, Australia). Freshly dissected mouse (C57BL/6) skins were applied as the membrane, which were harvested from cadavers sourced under the QIMR Berghofer Animal Facility tissue sharing provisions. The blood:virus mixtures were maintained at 37 °C using a recirculating water bath, and the mosquitoes were allowed to feed for one hour. A sample of the blood/virus mixture was collected before and after feeding to determine the mean titre of JEV fed to the mosquitoes. The mosquitoes were anaesthetised with CO2 and placed on a petri dish on ice, with engorged mosquitoes separated from non-engorged ones. The feeding rate per cup ranged from 61.7 to 75.8%. Engorged mosquitoes were transferred to gauze-covered containers, provided with a 10% sucrose solution on cotton balls, and maintained in an environmental chamber (Fitoclima 600, Aralab, Portugal) under the conditions described above. JEV-infected and uninfected mosquitoes were collected at 2, 7, and 12 days post-infection for RNA extraction and RNA-Seq.

The JEV titre in the pre- and post-blood meal samples, and for MOI calculations in Hsu infection studies, was determined using cell culture ELISA on C6/36 cells and an immunoplaque assay in Vero76 cells, respectively, using 4G4 anti-flavivirus NS1 monoclonal hybridoma supernatant for detection54. The mean titre of JEV fed to the mosquitoes was determined to be 7.6 log CCID50 in C6/36 cells.

Culex quinquefasciatus Hsu cells55 were maintained as cell monolayers in a flask using Leibovitz-15 (L-15) medium supplemented with 10% fetal bovine serum (FBS) at 27 °C. For infection, cells were infected with JEV (NSW-22 genotype 4) in 12-well plates at a multiplicity of infection (MOI) of 1. Cells and media were collected for RNA extraction or titration at specified times indicated in the relevant experiments.

Overexpression of JEV E protein and affinity purification of Culex entry factors

The full-length ORF for the JEV E protein (NCBI accession number OP904182) was amplified and cloned into the pET vector using BamHI and XhoI restriction sites fused with a His-tag, and expressed in E. coli BL21/DE3 competent cells. The protein, determined as soluble, was purified using nickel beads and the protocol provided by the manufacturer (Qiagen). Briefly, 1 L of bacterial culture was induced with IPTG (Isopropyl ß-D-1-thiogalactopyranosid) at a final concentration of 2 mM at 37 °C for 4 h and harvested by centrifugation at 4000 × g for 20 min at 4 °C. The pellet was resuspended in 10 ml sonication buffer (50 mM Na-phosphate pH 8.0, 300 mM NaCl), sonicated on ice (1 min bursts/1 min cooling/250 Watt), centrifuged at 10,000 × g for 20 min, and the collected supernatant was mixed with 300 μl Ni-NTA resin by stirring on ice for 1 h. The resin mix was loaded onto a column and washed with the sonication buffer as well as the wash buffer (300 mM NaCl, 10% glycerol, pH 6.0). Protein was eluted in 200–300 mM imidazole in the wash buffer.

Western blot analysis

Protein samples were separated on a 12% SDS-PAGE and transferred onto a nitrocellulose membrane (1 h at 100 V) using a BioRad Mini Protean gel system. Following blocking in 20 ml 1×TBST (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Tween20) containing 5% non-fat dry milk for 1 h, the membrane was washed three times (5 min each) in 20 ml 1×TBST at room temperature. Then, the membrane was incubated with anti-E protein (1:2000) monoclonal antibodies in 1% non-fat dry milk in 20 ml 1×TBST for 2 h at room temperature. The monoclonal anti-E protein antibodies (M2-8E7 and 4G2) were kindly provided by Dr Jody Peters from the School of Chemistry and Molecular Biology. Following three washes (5 min each) in 20 ml 1×TBST at room temperature, the primary antibodies were detected using anti-mouse antibodies conjugated with alkaline phosphatase (1:10,000; Sigma) (1 h at room temperature), and the blots were washed in 10 ml 1 × TBST (5 min each) and developed with 2 ml of NBT/BCIP (ThermoFisher).

Immunoprecipitation (IP)

To investigate the interaction of the JEV E protein with cellular proteins, IP was performed in vitro. The purified JEV E protein, expressed in E. coli, was incubated with Hsu total protein lysate from Hsu cells lysed in 1 × RIPA Cell Lysis buffer (ThermoFisher Scientific) and immunoprecipitated using an antiserum against E (M2-8E7)56. First, Hsu cells were pelleted at 840 × g for 10 min and washed three times with 1 × PBS. The cell pellet was then resuspended in 500 μl of RSB-NP40 (1.5 mM MgCl2, 10 mM Tris–HCl, 10 mM NaCl and 1% Nonidet P-40) in the presence of a protease inhibitor cocktail. Nuclei and debris were removed by centrifugation at 12,000 × g for 15 min at 4 °C. Purified JEV E protein (1 μg) and Hsu cell lysate protein (1 mg) were mixed in 500 μl of the binding buffer (20 mM Tris-Cl (pH 7.4), 0.1 M NaCl, 5 mM MgCl2, 0.1% NP-40, 200 mM phenyl methyl sulphonyl fluoride (PMSF), 20% glycerol (pH 7.4) for 4 h at 4 °C with gentle shaking. Protein concentrations were measured with the Bradford assay. The E-Hsu lysate mixture was incubated with either M2-8E7 or GFP (negative control) antibody overnight at 4 °C with gentle shaking. Protein A beads (100 μL) were added to the antibody-antigen lysate and incubated on ice (with shaking) for 1 h. Beads were collected by centrifugation at 10,000 × g for 1 min at 4 °C. The beads were washed with 500 µl of binding buffer, followed by four washes with IP wash buffer (NaCl 150 mM, EDTA 1 mM, Triton X-100 1% and Tris (pH 7.4) 10 mM). The captured antigens were eluted with 100 μl of 4 × LDS sample buffer (Novex, cat# B0007) and run on 8% SDS-PAGE protein gel. Distinct protein bands appeared after Coomassie staining and destaining, which were cut and purified for mass spectrometry using a previously described protocol57. Peptides were digested with trypsin and desalted using 0.6 μl C18 ZipTip (Merck) according to the manufacturer’s protocol. Eluted peptides were analysed on a TripleTOf 5600 instrument (SCIEX) using a Nanospray III interface. Peptides and proteins were identified by ProteinPilot v5.0.1 (SCIEX) using the Culex quinquefasciatus proteins.

Mosquito immunofluorescence analysis

JEV was detected in mid-sagittal sections of mosquitoes by immunofluorescence analysis (IFA) using established protocols58. Briefly, legs and wings were removed and mosquito bodies were fixed in 4% paraformaldehyde/0.5% Triton X overnight before mosquitoes were transferred to 70% ethanol. The mosquitoes were dehydrated through ascending graded alcohol and embedded in paraffin using standard procedures. Mid-sagittal paraffin sections (4 µM) were cut and fixed to positively charged slides, dried overnight at 37 °C, and dewaxed through descending alcohol grades to water. Antigen retrieval was performed by incubating sections in Dako Target Retrieval Solution (Dako, Denmark) pH 9.0, at 100 °C for 20 min using a decloaking chamber. Once cooled, the sections were washed in three changes of Tris-buffered saline/0.025% Tween 20 (TBSTW). Nonspecific antibody binding was inhibited by incubating the sections in Biocare Medical Background Sniper + 1% BSA for 15 min. Excess sniper was decanted, and sections were incubated with the primary antibody, 4G4 anti-flavivirus NS1 hybridoma supernatant (a kind gift from Dr Jody Peters at UQ), undiluted, at 27 °C overnight using a water bath. The sections were washed in three changes of TBSTW and incubated in Alexa Fluor 488-conjugated donkey anti-mouse secondary antibody, 1:300 dilution, for 60 min. Sections were washed three times in TBSTW, counterstained with DAPI, washed again, and mounted. Section staining was visualised using an Aperio Scanscope FL (Leica Biosystems, Illinois, USA) fluorescence slide scanner.

RNA extraction and RT-qPCR

To extract RNA from mosquitoes, they were lysed in pools of five as whole mosquitoes using TRIzol (ThermoFisher Scientific) with a TissueLyser (Qiagen) pulsed for 30 s at a frequency of 29.8 Hz per second. RNA was extracted from the lysates following the manufacturer’s instructions for TRIzol. Following quality control using an Epoch spectrophotometer (BioTek) and running aliquots of the samples on agarose gel, RNA samples were submitted to Genewiz for Illumina RNA sequencing. There were 18 samples in total, with three biological replicates for each time point, including the controls.

For cell lines, total RNA was extracted from cell pellets using QIAzol (QIAGEN) following the manufacturer’s instructions. DNase treatment was conducted using Turbo DNase (Ambion), and RNA was quantified using an Epoch spectrophotometer (BioTek). cDNA synthesis was done using M-MuLV reverse transcriptase (New England Biolabs) according to the manufacturer’s instructions. cDNAs were diluted 10 times and subjected to qPCR using QuantiFast SYBR Green PCR Kit (QIAGEN). The reactions were run on Rotor-Gene Q machine (QIAGEN), and PCR amplification and melt curve analysis for all products were performed according to the instructions. Primers used in this study are listed in Supplementary Table 1. All qPCR experiments were run with at least three biological and two technical replicates. Cx. quinquefasciatus 18S (NCBI accession number U22126.1) was used as a reference gene.

RNA-Seq and bioinformatics

Total RNA (300 ng, RIN > 6) was used to prepare poly(A)-enriched mRNA libraries using the VAHTS Universal V8 RNA-Seq Library Prep Kit for Illumina (Vazyme), following the manufacturer’s instructions. Adaptors were then ligated to both ends of the cDNA fragments. Adaptor-ligated DNA was size-selected using VAHTS DNA Clean Beads (Vazyme), and fragments of approximately 260–360 bp (insert size of ~200–300 bp) were recovered. The libraries were amplified by PCR for 12–15 cycles using P5 and indexed P7 primers. PCR products were purified using VAHTS DNA Clean Beads (Vazyme), validated using an Agilent 4200 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA), and quantified using a Qubit 3.0 Fluorometer (Invitrogen, Carlsbad, CA, USA). Indexed libraries were multiplexed and loaded onto an Illumina NovaSeq 6000 instrument, according to the manufacturer’s instructions (Illumina, San Diego, CA, USA). Sequencing was performed using a 2 × 150 bp paired-end configuration. Image analysis and base calling were conducted using NovaSeq Control Software (NVCS) and RTA 3.

The bioinformatics analyses were conducted using the CLC Genomics Workbench (version 21.0.3) and OmicsBox 3.2.2. Raw sequencing reads were processed using Trimmomatic v0.3959 to remove adapter sequences and trim low-quality bases. Quality filtering was based on a minimum average Phred score of Q30. Read quality was assessed before and after trimming using FastQC version 0.11.8 tool (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) accessed through OmicsBox. The trimmed reads were initially mapped to the JEV genome (OP904182.1), while unmapped reads were retained for de novo assembly.

De novo assembly was performed using Trinity (version 2.15.1), a De Bruijn graph-based assembler, with a minimum contig length of 500. This method enabled the construction of super transcripts, as previously described60,61. The quality of the assembly was evaluated using BUSCO 5.6.0 (Benchmarking Universal Single-Copy Orthologs) assessments. This tool provides quantitative measures for transcriptome assembly completeness based on evolutionary expectations of gene content from near-universal single-copy orthologs. For this assessment, the Insecta lineage from OrthoDBv10 was used62,63.

TransDecoder (version 5.7.1) was employed to identify candidate coding regions within the transcript sequences, with a minimum protein length threshold of 100 amino acids and the universal genetic code. Homology searches were performed using the Pfam database to enhance the sensitivity of the predicted open reading frames64. The annotated super transcripts served as the reference for further transcriptome analyses. To align reads to the assembled reference, STAR (version 2.7.8a) was used with default parameters. STAR is a fast RNA-Seq mapper specifically designed for the accurate alignment of RNA-Seq reads to a reference genome, supporting splice junction and fusion read detection65.

Expression profiles of infected and non-infected samples at different time points (2-, 7-, and 12-days post-infection) were analysed using the CLC Genomics Workbench. This tool employs a statistical model similar to the EdgeR and DESeq packages for normalisation and differential expression analysis, based on a negative binomial Generalised Linear Model (GLM).

PCA graphs were generated for each time point after JEV infection, comparing control and infected samples. The expression levels used as input were normalised log CPM (Count Per Million) values.

Expression levels for all identified transcripts were calculated using the total mapped read counts and normalised to Transcripts Per Million (TPM) to enable accurate comparison across samples. Gene expression profiles for each sample were generated based on the calculated TPM values. For differential expression analysis, raw counts were analysed using a GLM based on a negative binomial distribution, as implemented in CLC Genomics Workbench. The Wald test in CLC Genomic Workbench was used to compare each sample against its control group (infected vs non-infected) at each time point. We considered genes with more than 2-fold changes and a p-value of ≤0.05 as significantly modulated genes. Venn diagrams were generated to visualise the overlap of differentially expressed transcripts across the three comparisons.

All differentially expressed genes were uploaded to the OmicsBox server for functional annotation and GO analysis. Functional annotation was conducted using BLAST (nt and nr databases), enzyme classification codes (EC), and InterProScan algorithms66 to identify the GO terms associated with differentially expressed sequences. EggNOG version 4.5 was employed for orthology prediction, focusing on Clusters of Orthologous Groups67. The most abundant GO terms were identified for the categories of molecular function, biological process, and cellular component. A statistical assessment of enriched GO terms was performed using the FatiGO package integrated into OmicsBox, which applies Fisher’s Exact Test for significance testing.

In vitro dsRNA synthesis and cell transfection

For the in vitro synthesis of dsRNA targeting specific genes and control, the MEGAscript T7 transcription kit (Invitrogen) was used. Gene-specific forward and reverse primers both containing T7 promoter sequences (Supplementary Table 1) were used for PCR amplification of products (500 nt), which were confirmed after sequencing. dsGFP was used as control. In the transcription reaction, 1 μg of purified PCR product was used along with all the required ingredients and incubated at 37 °C overnight. After DNase treatment, dsRNA was precipitated using Lithium chloride, washed with 70% ethanol, and finally dissolved in nuclease-free water. The quantity and quality of dsRNA were measured using an Epoch spectrophotometer (BioTek) and running a 1% agarose gel.

For transfection, Hsu cells were transfected with 1 μg of dsRNA using the Cellfectin transfection reagent according to the manufacturer’s instructions (Invitrogen). Two days after the initial transfection, the cells underwent a second transfection for two days and were subsequently infected with 1 MOI JEV. The cells were then collected 48 h post-infection (hpi) for RNA extraction and assessment of JEV genomic RNA levels using RT-qPCR.

Antibody neutralisation

Hsu cells were seeded into a 12-well plate at a density of 400,000 cells per well. Rabbit polyclonal anti-prohibitin 2 (Santa Cruz Biotech, sc-67045) or anti-FLAG-tag (Sigma-Aldrich) antibodies were mixed separately in L-15 medium and added to the wells (200 µl) at a final concentration of 1 µg antibody per well. After 2 h, the medium-antibody mixture was removed, and cells were infected with JEV at 1 MOI. Cells were collected 24 hpi, from which RNA was extracted to assess viral gRNA levels.

Focus forming assay

For JEV immunoplaque assay, Vero76 cells (2 × 10⁴/well) were seeded in 96-well plates and infected the following day with serial 10-fold dilutions of cell culture supernatant (undiluted, 10−1, 10−2, 10−3). For each condition, a single technical replicate per dilution was performed across four biological replicates. At 30 h post-infection, cells were fixed with 100 μL/well 80% acetone in PBS (1 h, −20 °C). After PBS washing and drying, plates were blocked with 150 μL/well Pierce™ Clear Milk solution (30 min). Plates were incubated with primary mouse anti-NS1 flavivirus antibody 4G4 (1:100, 50 μL/well, 1 h) followed by goat anti-mouse IRDye 800CW secondary antibody (1:2000, 50 μL/well, 1 h; LI-COR). All antibodies were diluted in Clear Milk/PBS-T (0.05% Tween 20). Five times 5 min PBS-T washes followed each incubation. Plates were scanned (Odyssey CLx, 42 μm; medium; 3.0 mm), and foci-forming units counted (Image Studio Lite v5.2.5) to calculate FFU/mL. The limit of detection was 40 FFU/mL.

Quantification and statistical analysis

GraphPad Prism version 10 was used for all statistical analyses and graph production. For data that met the normality test (Shapiro-Wilk test), a t-test or One-way ANOVA with Tukey’s post-hoc comparisons test was used to determine significance levels between two and three or more treatments, respectively. Non-parametric tests were used if the data did not pass the normality test. More details are provided in the relevant figure legends or the text. qPCR data were analysed using the relative expression ratio method (Ratio = (Etarget)ΔCPtarget(control – sample)/(Eref)ΔCPref(control – sample)) as described previously68. Gene expression and JEV gRNA levels in controls were set to 1, with transcript levels in treatments expressed as fold changes relative to the controls.