Abstract
Background
The mosquito Aedes aegypti transmits two of the most serious mosquito-borne viruses, dengue virus (DENV) and Zika virus (ZIKV), which results in significant human morbidity and mortality worldwide. The quickly shifting landscapes of DENV and ZIKV endemicity worldwide raise concerns that their co-circulation through the Ae. aegypti mosquito vector could greatly exacerbate the disease burden in humans. Recent reports have indicated an increase in the number of co-infection cases in expanding co-endemic regions; however, the impact of co-infection on viral infection and the detailed molecular mechanisms remain to be defined.
Methods
C6/36 (Aedes albopictus) cells were cultured in Dulbecco's modified Eagle medium/Mitsuhashi and Maramorosch Insect Medium (DMEM/MM) (1:1) containing 2% heat-inactivated fetal bovine serum and 1× penicillin/streptomycin solution. For virus propagation, the cells were infected with either DENV serotype 2 (DENV2) strain 16681 or ZIKV isolate Thailand/1610acTw (MF692778.1). Mosquitoes (Ae. aegypti UGAL [University of Georgia Laboratory]/Rockefeller strain) were orally infected with DENV2 and ZIKV through infectious blood-feeding.
Results
We first examined viral replication activity in cells infected simultaneously, or sequentially, with DENV and ZIKV, and found interspecies binding of viral genomic transcripts to the non-structural protein 5 (NS5). When we challenged Ae. aegypti mosquitos with both DENV2 and ZIKV sequentially to probe similar interactions, virus production and vector susceptibility to infection were significantly enhanced.
Conclusions
Our results suggest that DENV2 and ZIKV simultaneously establishing infection in the Ae. aegypti mosquito vector may augment one another during replication. The data also implicate the homologous NS5 protein as a key intersection between the flaviviruses in co-infection, highlighting it as a potential target for vector control.
Similar content being viewed by others
Background
Mosquito-borne diseases are one of the most significant public health burdens [1,2,3,4,5]. Human activities, urbanization, and climate change are increasingly bringing more human hosts in contact with disease vectors [2, 6,7,8,9]. Currently, half of the global population is at risk for dengue virus (DENV) infection [2, 8, 10]. In the aftermath of the 2015–2016 Zika virus (ZIKV) outbreak, which exposed more than 130 million people to infection, the virus remains endemic to tropical and subtropical regions [11].
Sharing a common vector in Aedes aegypti, DENV and ZIKV endemicity may expand in concert, resulting in widespread co-circulation [12, 13]. Thus, DENV-ZIKV synergy presents a bleak outlook for the near future with a growing proportion of the global population living under threat of simultaneous infection by both flaviviruses. As DENV-ZIKV co-circulation expands worldwide to affect currently low-risk or virus-free regions, mosquito vectors will have increased opportunities to receive and transmit both viruses [5]. Recent reports of DENV-ZIKV co-infection corroborate this prediction and reflect a proliferating synergy that has been overlooked. This may be the result of systemic underreporting engendered by difficulties in the differential diagnosis and detection of asymptomatic infections [14,15,16].
One cross-sectional study of the ZIKV epidemic in Colombia detected 8.8% of the DENV-ZIKV serotype among 34 co-infection cases [15]. Another study in southern Mexico randomly sampled a cohort of pregnant women during a non-epidemic period and found a relatively high proportion (2%) with DENV-ZIKV co-infection [17]. A previous report revealed that DENV infections occurred during the same period, highlighting the concerning extent of silent transmission of ZIKV with DENV [17, 18]. Besides underreporting, underlying this phenomenon may be antibody-dependent enhancement (ADE) between DENV and ZIKV in co-endemic areas. Studies support ADE of ZIKV infection by anti-DENV antibodies [4, 19,20,21], which not only may increase disease severity, but also may drive ZIKV transmission into primarily DENV-endemic areas. Similar results have been suggested for DENV, in which ZIKV-mediated ADE increases the propensity for severe disease and enables DENV to persist and proliferate in primarily ZIKV-endemic regions [ The replicating DENV2 genome interacts with the ZIKV-NS5 protein in co-infected mosquito cells. ZIKV-NS5 interacting with DENV2 and ZIKV genomic RNA was precipitated from cell lysates at 4 days using CLIP-PCR, followed by RT-PCR with primers specific for DENV and ZIKV genomes. Ribosomal protein S7 was used as a loading control
Co-infection of Ae. aegypti with DENV2 and ZIKV results in differential viral genome expression
Finding significant enhancement of viral replication by DENV2-ZIKV co-infection in vitro possibly arising from cross-species interactions, we challenged adult female Ae. aegypti sequentially with both viruses to quantify the effects of DENV2-ZIKV co-infection on vector competence in vivo. So far, existing studies into DENV-ZIKV co-infection have only challenged Ae. aegypti simultaneously, presenting both viruses in the same blood meal (BM). Contrarily, we elected to challenge mosquitos with DENV2 and ZIKV sequentially to maximize ecological validity. It is much more likely that in a DENV-ZIKV co-endemic area, a female mosquito would acquire co-infection in sequential feeding episodes: by feeding first on a DENV-infected host, then a ZIKV-infected host, or vice versa. The possibility of a mosquito obtaining both viruses from a DENV-ZIKV co-viremic human host through a single BM is much lower. Critically, we also consider the findings of studies reporting that non-infectious BM prior to subsequent viral BM can promote viral replication, as the initial non-infectious BM induces physiological changes in the midgut epithelium, rendering it more permissible to dissemination through the basal lamina [28, 29, 33, 34].
Thus, we included cohorts of mosquitoes presented with an initial naïve BM before challenge with DENV2 or ZIKV to account for this possible confounder in evaluating co-infection effects on viral replication (Fig. 4A). Overall, two co-infection schemes were used in vivo, in which Ae. aegypti were either challenged first with DENV2 and then ZIKV on a second BM (DENV2 → ZIKV), or vice versa (ZIKV → DENV2). These cohorts were compared against mosquitos single-infected with DENV and ZIKV, through either one or two BMs, as well as a mock cohort (BM-BM).
Co-infection of Ae. aegypti mosquitos with DENV and ZIKV results in differential viral genome expression. A Time course of experimental oral challenge with DENV2 and ZIKV. Mosquitos were captured 12 h ahead of day 0 and presented with blood meal (BM), DENV2, ZIKV, or maintained on sugar feeding (Sugar). On day 5, one group was given a second BM (BM-BM), and others were challenged with DENV2 or ZIKV. At 7 dpi (day 12), whole mosquito bodies were collected and homogenized. Relative viral genome expression of B DENV2 and C ZIKV was determined by qRT-PCR analysis, with normalization to the endosomal S7 protein. Each of six oral challenge schemes consisted of at least five biologically independent cohorts, and post hoc comparisons between groups were performed using Tukey’s multiple comparisons test; **P < 0.01; ***P < 0.001; ****P < 0.0001
Collecting mosquitos at 7 dpi, which we previously determined to be an optimal time point for viral genomic analysis, we quantified via qRT-PCR the relative expression of DENV2 and ZIKV genomes, respectively (Fig. 4A). We found that, contrary to our observations in vitro, DENV2 expression was significantly downregulated in both co-infection scenarios, particularly when compared with those challenged twice (DENV2 → DENV2) with DENV2 (Fig. 4B). Also, a significant difference in expression between DENV2 single-infected mosquitos presented with (BM → DENV2), and without (sugar-fed → DENV2), an initial naïve BM suggests that BM-induced modifications do indeed promote viral replication by encouraging dissemination from the midgut. Genomic expression of ZIKV was significantly elevated in both co-infection scenarios, even in comparison to mosquitos infected twice (ZIKV → ZIKV) with ZIKV (Fig. 4C). Between the two co-infection groups, ZIKV → DENV2 mosquitos expressed genomic ZIKV at significantly higher levels than did their DENV2 → ZIKV counterparts. There was also a significant difference in expression between sugar-fed → ZIKV and BM → ZIKV individuals, as with DENV2, again highlighting the initial BM’s ability to promote viral replication in and past the midgut. Moving forward, we evaluated the implications of co-infection on vector competence, quantifying the infectivity of virus particles produced in vivo.
Virus production and vector susceptibility are enhanced in Ae. aegypti challenged with both DENV2 and ZIKV
Having characterized the genomic expression profiles of co-infected mosquitos, we next assessed the quantity and infectivity of viruses produced therein. Performing plaque assays on isolated virus from individuals collected at 7 dpi from each treatment group (Fig. 4A), we found that viral titers were significantly higher in co-infected mosquitos than in the DENV single-infected cohort (Fig. 5A). Notably, these increases were observed for both DENV2 → ZIKV and ZIKV → DENV2 co-infection groups against all three DENV single-infection cohorts: those challenged either once (sugar/BM → DENV2) or twice (DENV2 → DENV2) with DENV2. Accordingly, the corresponding infection rates were markedly higher in co-infected mosquitos, increasing by as much as 36.7% (Fig. 5B). Compared with ZIKV single-infected cohorts, viral titers were also higher in co-infected individuals. Infection rates also saw increases by about 10% except when compared with the sugar → ZIKV cohort, a result likely attributable to individual variability. Despite the unexpectedly high infection rate, the median viral titer for this cohort was at least twofold lower than those of co-infected cohorts. Overall, mosquitos challenged sequentially with DENV and ZIKV produced more infectious virus, which induced a greater extent of infection ex vivo.
Virus production and vector susceptibility are enhanced in DENV2/ZIKV co-infected mosquitos. A Mosquitos were challenged with virus as described in Fig. 4, and then collected at 7 dpi (day 12) for plaque assay. Geometric means (PFU/ml) are plotted, and each of eight viral challenge schemes comprised at least three biological cohorts. B Sample size n, infection rate, and median PFU/ml corresponding to the experimental groups. Infected samples had positive (> 0) PFU/ml values; uninfected, negative samples are represented on the log scale as positive (PFU/ml = 1) only for visual interpretation. Comparison between groups post hoc was performed using Dunn’s multiple comparisons test; *P < 0.05; **P < 0.01; ***P < 0.001
Mosquitos twice-challenged with ZIKV produced higher viral titers than those co-infected, an unsurprising result following consecutive ZIKV infection. Secondary infection is greatly assisted by the existing RC infrastructure and an infected, compromised midgut epithelium and basal lamina (ZIKV → ZIKV). Indeed, viral titers were higher in co-infected mosquitos compared with those challenged once (sugar/BM → ZIKV). Infection rates of co-infected mosquitos also increased, suggesting greater susceptibility to infection and virus propagation, i.e., enhanced vector competence. As for mosquitos twice-challenged with DENV2, the clear enhancement of viral replication by co-infection is much more readily apparent, as these cohorts were entirely refractory to infection (0% infected, DENV2 → DENV2). Across DENV2 single-infected cohorts, infection rates peaked at 11.1%. The fact that vector infection rates were elevated to 44.1% and 46.7%, respectively, in co-infected cohorts indicates that DENV2-ZIKV co-infection positively, mutually modulates viral replication.
Of note, mosquitos given an initial naïve BM neither produced significantly higher viral titers nor were infected at higher rates than those directly challenged with DENV2/ZIKV. This suggests that while an initial non-infectious BM may correlate with higher viral genomic expression by assisting virus dissemination from the midgut, it ultimately neither enhances the production of viable, infectious particles nor promotes co-infected Ae. aegypti susceptibility to infection.
Taken together, our results show that DENV2-ZIKV co-infection significantly enhances virus production and vector susceptibility to infection. At the cellular level, viral replication is mutually enhanced owing to the overlap of highly homologous flaviviral replication machinery. We observed therein that DENV2 transcripts engage ZIKV-NS5. Overall, our findings raise grave concerns about DENV2 and ZIKV co-circulation, which threatens to strain healthcare resources and exacerbate transmission and disease as mosquitos vector these flaviviruses with greater efficiency.
Discussion
DENV and ZIKV are flaviviruses transmitted by mosquitos of the Aedes genus, primarily Ae. aegypti [35]. They co-circulate in overlap** endemic areas. Consequently, as the spread of DENV and ZIKV expands rapidly, mosquitos will have increased opportunities to acquire simultaneous and/or mixed infections with different types of flaviviruses [35]. This may occur following an infectious BM from a single human co-viremic for DENV and ZIKV, or when mosquitos acquire sequential BMs from two individuals, each infected with a different virus. DENV and ZIKV share a highly conserved non-structural protein repository consisting of five enzymes/subunits (NS1–5), which associate closely to form a tightly-regulated RC. We hypothesized that DENV and ZIKV interact through their homologous RC components during replication, which has significant implications for viral replication and vector competence.
In the present study, we observed significant enhancements in virus production and vector susceptibility following DENV2-ZIKV co-infection in Ae. aegypti, and we demonstrated the cellular and molecular bases of these effects. In co-infected mosquito cells, we found that DENV2 expression was significantly enhanced, whereas ZIKV was coincidentally markedly suppressed and virus production was significantly increased for both. The finding that flaviviral RCs co-localize extensively in the cytoplasm, we determined whether DENV and ZIKV participate in cross-species interactions during replication. We found that the replicating DENV2 genome engages ZIKV-NS5. This surprising interaction readily explains the drastic enhancement of DENV2 expression in vitro, which suggests that DENV2 engages the ZIKV RC competitively. DENV-ZIKV interactions during replication provide a basis for the mutual increase in secretory virus particles observed. Next, we challenged Ae. aegypti adult females with both viruses to determine whether similar DENV2-ZIKV interactions also modulate viral replication in vivo. We found that while genomic expression at 7 dpi of DENV2 was markedly downregulated, with ZIKV upregulated, there was again a mutual enhancement of viral replication as indicated by significantly elevated levels of infectious virions produced in co-infected mosquitos, which were also more susceptible to infection.
Our findings describe for the first time the potent mutualistic outcomes of flaviviral co-infection for viral replication, and suggest that vector competence may be enhanced as a result. Vector competence in the arboviral sylvatic cycle is defined by the extent to which the mosquito permits a virus to utilize its circulatory system. The virus must first establish infection in the midgut and produce virions that disseminate to secondary tissues, which must then sustain replication throughout the organ system until freshly propagated virions breach the salivary glands. We found that viruses produced in DENV-ZIKV co-challenged mosquitos were significantly more abundant compared with those infected with only one virus, particularly DENV2. Moreover, these mosquitos were infected at a higher rate, suggesting that co-infected mosquitos are more susceptible to infection, perhaps resulting from the convergence of immune suppression pathways of both replicating viruses, particularly in the midgut. These results strongly suggest that co-infection enhances vector competence because, at a collection date of 7 dpi, either virus will have already disseminated from the midgut and commenced replication throughout the entire body to produce viable, infectious particles. Although employing more collection points to visualize replication kinetics may allow for a clearer spatio-temporal resolution, for the purpose of understanding co-infection in terms of viral replication and vector competence, it was sufficient to isolate virus from mosquitos at 7 dpi for the plaque assay. The implications of our results for vector competence are somewhat limited, however, absent organ-specific analysis. Specifically, quantifying virus present in the salivary glands may permit more direct observation of outcomes for vector competence, as infectious viral particles must be secreted into the saliva during a BM for transmission [56]. However, the variability of flaviviral infection in vivo among individual mosquitos largely precluded such an investigation, as salivary glands dissected and analyzed individually would likely vary greatly in viral titer. Collecting whole bodies enabled an individual analysis without sacrificing statistical integrity. Additionally, we studied replication dynamics in vivo using a blood meal challenge, opting not to infect mosquitos by intrathoracic injection to preserve the critical barrier to vector competence manifest in the midgut’s physical and immunological fortifications. Quantifying virus propagated through sequential co-infection by oral challenge allowed us to observe the consequences of DENV-ZIKV interaction directly through the entire course of infection within the vector, beginning with ingestion in the midgut. Furthermore, our findings are highly relevant to the evolving landscape of DENV-ZIKV endemicity, as our sequential oral infection model is closely aligned with actual vector activity. Mosquitos are much more likely to acquire flaviviral co-infection sequentially from individual hosts than from a single host viremic for both DENV and ZIKV. Individual variability in feeding opportunity, i.e., extent of engorgement, however, likely negates many real differences in virus intake between laboratory and field.
To date, exploratory studies into arboviral co-infection have only established that Ae. aegypti are susceptible to infection by more than one type of virus and they may simultaneously transmit multiple viruses. One study reported that Ae. aegypti simultaneously challenged with a combination of two or three arboviruses including DENV2, ZIKV, and chikungunya (CHIKV), were frequently double- or triple-infected, which indicated that the mosquitos are susceptible to co-infection. Inoculation of saliva in vitro confirmed the potential for co-transmission of all three viruses [12]. Another study found that mosquitos simultaneously co-infected with DENV and ZIKV preferably transmit the latter [24]. The co-infection and co-transmission potential of ZIKV-CHIKV [36, 37] and DENV-CHIKV [38] have also been supported. With respect to arboviral co-infection, our study contributes novel insight into its implications for viral propagation in Ae. Aegypti and demonstrates that DENV-ZIKV co-infection mutually enhances viral replication. Our results also suggest that vector competence may be enhanced following DENV-ZIKV co-infection, as indicated by increased infection rates. We also provide the first account of molecular mechanisms underlying co-infection effects by reporting that DENV engages ZIKV-NS5 during replication.
As for the apparent conflict between our in vitro and in vivo results in which DENV2 replication appeared be competitively promoted in co-infected cells (Figs. 1B, 2C), but drastically suppressed in the mosquito (Fig. 4), it is important to note that markedly reduced DENV2 expression in co-infected mosquitos reflected only the viral genome content at 7 dpi, not the amount of infectious DENV virions produced by the vector. Indeed, virus production was significantly higher in co-infected mosquitos than in those single-infected with DENV. In addition, co-infected cohorts were much more susceptible to infection. Low DENV2 genomic expression suggests that there may be competitive engagement between ZIKV and DENV2 due to interaction between ZIKV genomic transcripts with DENV2 RC proteins. We showed that DENV2 utilizes ZIKV-NS5 for transcription; thus, it is likely that ZIKV may reciprocally utilize DENV2-NS5 to its advantage. Another intriguing possibility is that flaviviral NS5 may have a dual purpose in the convergence of DENV-ZIKV co-infection by cross-species cap** of the respective RNA genomes by N-terminus methyltransferases to assist in evasion of the host immune response. This is an alternative (and not mutually exclusive) molecular premise for the observed increases in virus production and vector susceptibility in co-infected mosquitos. Further biochemical studies regarding the DENV-ZIKV interaction via NS5, a possible target for vector control and vaccine development, is warranted as the co-circulation of DENV and ZIKV broadens globally.
The clinical and epidemiological implications of expanding flaviviral co-circulation remain largely unexplored. Though it is not clear whether co-infected patients develop more severe disease [39,40,41,42], increased co-circulation and transmission of multiple flaviviruses will surely pose significant problems for diagnosis and surveillance because of the common clinical presentation, asymptomatic response, and cross-reactivity. Although progress is being made on the diagnostics front [43,44,45], effective vector control remains the most effective approach to managing and eliminating mosquito-borne diseases [3]. Central to vector control is a clear understanding of the pathogen-vector relationship as it evolves in real-time, with expanding arboviral co-circulation being a worrying trend that we have addressed. As the periphery of DENV-ZIKV co-circulation expands, it also encompasses other arboviruses, such as CHIKV [46]. As we demonstrated, co-infection with DENV and ZIKV mutually enhances viral replication within the vector. Further interaction with other arbovirus, such as yellow fever virus or CHIKV, may result in unknown synergistic effects. This may similarly threaten to facilitate widespread circulation and transmission of multiple deadly viruses by modulating the vector response. The interplay among these arboviruses in Ae. aegypti and its effects on viral replication and vector competence require further study. Future work in this area could incorporate the study of viral interactions with the mosquito microbiota [47, 48] and diverse RNA responses [49,50,51,52] of the mosquito. Whether differential interactions arise between the DENV serotypes in co-infection scenarios is also worth pursuing, as all four DENV serotypes (DENV1–4) are spreading throughout Asia, Africa, and the Americas [53]. It also remains to be determined whether arbovirus co-infection influences virus selection pressure and recombination events [54, 55]. In conclusion, this study presents the novel finding that DENV-ZIKV co-infection mutually enhances viral replication within the mosquito. This threatens to increase the disease burden in co-endemic areas, drive the “silent” transmission of strains not predominantly circulating, and introduce flaviviruses into communities not yet seen. With arboviral co-endemicity on the rise globally, the rapidly shifting vector–pathogen relationship must be further investigated, in which the pathogen itself bears many faces.
Conclusions
In this study, we determined the effects of DENV and ZIKV co-infection on viral replication in Ae. aegypti and to identify the specific molecular interactions involved. We first examined viral replication dynamics in cells infected simultaneously or sequentially with DENV and ZIKV. We report the interspecies binding of viral genomic transcripts to NS5. We then challenged Ae. aegypti mosquitos with both DENV2 and ZIKV sequentially to identify similar interactions, and found that virus production and vector susceptibility to infection were significantly enhanced. Our results suggest that DENV2 and ZIKV simultaneously establish infection in the Ae. aegypti vector, which may mutually augment one another during replication. The data also implicate the homologous NS5 protein as a key intersection between the flaviviruses in co-infection, highlighting it as a potential target for vector control.