Yellow Fever Virus (YFV) Detection in Different Species of Culicids Collected During an Outbreak in Southeastern Brazil, 2016-2019

1. Introduction

Yellow Fever (YF) is a severe disease caused by the arbovirus Orthoflavivirus flavi (former Yellow Fever Virus (YFV), a member of the Flaviviridae family, and remains a significant public health concern in parts of Africa and the Americas [1]. YF may cause significant morbidity and mortality rates in the human populations, as well as impact the neotropical primates population [2]. Despite the availability of the live attenuated 17-DD vaccine, a high case fatality rate (CFR) of 40% to 60% persists, particularly in South America [1,3]. In Brazil, after the eradication of the urban YFV cycle in 1942 transmitted by Aedes aegypti mosquitoes, YFV is maintained by a sylvatic transmission cycle involving several species of neotropical primates (NTPs) and forest canopy-dwelling mosquitoes, mainly Haemagogus-spp and Sabethes-spp, and human cases are caused by a spillover process in green areas [4]. YF surveillance is based on confirmation of epizootic events through virus detection by RT-qPCR and or immunohistochemistry in accordance with the Ministry of Health Guidelines [5].

Seasonal climatic variations significantly influence YFV transmission by affecting mosquito population dynamics and viral amplification. During the rainy season, abundant precipitation creates numerous larval habitats while elevated temperatures and high humidity accelerate mosquito development and viral replication, leading to surges in sylvatic vectors such as Haemagogus and Sabethes spp. These conditions enhance virus amplification among non-human primates and elevate the risk of spillover to humans [6]. Conversely, in the dry season, reduced rainfall limits breeding sites and diminishes vector densities, though desiccation-resistant eggs permit a low-level virus circulation that can rapidly rebound once rains return [7].

From mid-2016 until late 2018, Brazil faced one of the largest YF outbreaks in recent decades, mainly in the southeastern region [8,9,10,11]. São Paulo state, located in southeast Brazil, is the most densely populated state of the country, containing one of the world’s largest urban conurbations [12]. A total of 875 cases of YFV in NTPs between July 2016 and November 2019 and 624 cases of YFV in humans between January 2017 and 18th November 2019 were reported. This outbreak was caused by the 1E lineage belonging to South American I (SA-I) genotype that originated in the Amazon basin which has later disseminated from northern São Paulo into geographically neighboring areas of western MG and into the south of the state [14]. Some epizootic events in Callithrix monkeys were confirmed in large urbanized cities in proximity of urban green areas where Haemagogus and Sabethes mosquitoes were not found [15], indicating that synanthropic mosquitoes were likely involved in viral transmission in these areas. Considering that entomological investigation is a complementary tool to better understand eco-epidemiological aspects of YF after notification of suspected epizootic events, here we describe different Aedini and Sabethini mosquitoes found positive with YFV by RT-qPCR and its ecological factors, showing the continuous threat of reurbanization of YFV in Brazil.

2. Materials and Methods

2.1. Study Area

The study was conducted in the state of São Paulo, Brazil, which comprises 645 municipalities organized into 15 administrative regions. The state spans approximately 248,196.960 square kilometers and has a population of 44,749,699 inhabitants, primarily concentrated in the coastal region. São Paulo encompasses two distinct biomes: the Cerrado and the Atlantic Forest, both of which have suffered significant deforestation in recent years.

2.2. Epizootic Events and Mosquito Collection

Between November 2016 and June 2019, a total of 3,731 mosquito pools from the Aedini Tribe and Sabethes genus were collected in 192 municipalities with ongoing epizootic events and adjacent cities. Briefly, frozen carcasses of NTPs were sent to Adolfo Lutz Institute for YFV detection, according to the Brazilian Ministry of Health Guidelines as previously described [8]. Mosquitoes were captured at ground level between 9 a.m and 3 p.m using entomologic nets and bottle-type manual vacuums in forested and green areas, and Nasci Aspirator in urban dwellings. After sampling, mosquitoes were frozen, transferred to cryogenic tubes, and stored in liquid nitrogen containers for transport. Identification was performed based on morphological characteristics by the Pasteur Institute (formerly the Superintendence for Control of Endemic Diseases - SUCEN). The mosquitoes were subsequently sorted into pools containing 1 to 50 individuals per pool, according to species, collection date, and location. Molecular detection for YFV was carried out in non-engorged mosquitoes (n=3,376) at the local reference laboratory for arthropod-borne viruses at Instituto Adolfo Lutz (IAL) in São Paulo. Pools were triturated in FastPrep-24 5G Instrument (MP Biomedicals, Ohio, USA) and in Magna Lyser (Roche) in 1 mL of phosphate-buffered saline solution with 0.75% bovine albumin, penicillin (100 units/mL) and streptomycin (100 µg/mL). The resultant suspension was centrifuged at 1800×g for 15 min, and the supernatant was withdrawn and frozen at -70ºC until further use.

2.3. YFV RNA Detection and Statistical Analysis

Viral RNA was extracted using QIAamp Viral RNA Mini Kit following the manufacturer’s instructions (QIAGEN, Hilden, Germany). Detection of YFV RNA was performed using an RT-qPCR protocol [11]. Results with Cycle Threshold (CT) values ≥ 35 were retested. If the new result had a CT value ≤ 38, the pool was considered positive for YFV. The Kruskal-Wallis test was conducted exclusively among YFV-positive mosquito pools to evaluate differences in viral load by mosquito species, as indicated by their CT values. To assess differences in Yellow Fever Virus (YFV) viral loads among different mosquito species, a Generalized Linear Model (GLM) was performed with Ct value as the dependent variable. Ct values were used as a proxy for viral load, where lower Ct values indicate higher viral loads.

The primary YVF vector Hg. leucocelaenus was set as the reference category to compare viral loads across species. The model was specified as Ct ~ Species, where Ct value was assumed to follow a Gaussian (normal) distribution with an identity link function. The analysis reports estimated mean differences in Ct values (β coefficients) for each species compared to Hg. leucocelaenus. To evaluate whether seasonal variation (rainy vs. dry) influenced Ct values, season was included as an additional predictor in the GLM. An interaction term (Species × Season) was also tested to assess potential species-specific seasonal effects. All p-values <0.05 were considered significant. All analysis were performed using Rstudio v.2023.12.1, ggplot2 package [16].

3. Results

A total of 3,731 mosquito pools were collected during the outbreak (Table 1), of which 46 pools (1.4%) from 9 mosquitos species tested positive for yellow fever virus (YFV), representing 22 municipalities (8.7%) (Table 2). Additionally, epizootic events were confirmed by RT-qPCR in 82 cities (Supplementary Material 1). The Ct values of YFV-positive pools ranged from 16 to 38, with a median of 32 (Figure 1).

Among the species collected, Aedes scapularis accounted for 26.46% of all mosquitoes, with 0.67% of pools testing positive, followed by Aedes albopictus (21.66%, 0.41% positive) and Psorophora ferox (11.20%, 1.32% positive). Haemagogus leucocelaenus represented 8.09% of the total, with 5.83% of its pools testing positive, while Haemagogus janthinomys/capricornii comprised 3.4%, with 5.51% positive. Other species testing positive for YFV included Aedes serratus (5.72%, 2.07% positive), Sabethes albiprivus (2.67%, 15.78% positive), Sabethes purpureus (0.80%, 2.08% positive), and Sabethes identicus (0.74%, 1.75% positive).

Analysis of Ct values among YFV-positive mosquito pools revealed significant differences in viral loads (Figure 1). Haemagogus species consistently exhibited the lowest Ct values, indicating higher viral loads, while Sa. albiprivus, Ae. albopictus, Ae. serratus, and Ps. ferox had higher Ct values, suggesting lower viral loads. The distribution of Ct values varied across species, with some species displaying a wider range, indicating heterogeneity in infection levels within the same species. While Ae. scapularis pools generally showed high Ct values, two pools recorded Ct values of 25 and 28, suggesting moderate viral loads. The Kruskal-Wallis test confirmed a statistically significant difference in Ct values among species (p = 0.0002). These differences are visually represented in Figure 1.

The Generalized Linear Model (GLM) analysis identified significant differences in Ct values among mosquito species. Ha. leucocelaenus exhibited the lowest Ct values and was used as the reference species. Compared to Ha. leucocelaenus, Ha. janthinomys-capricornii showed a moderate increase in Ct values (β = 4.71, p = 0.039).

Mosquito species from the Aedes, Psorophora, and Sabethes genera exhibited significantly higher Ct values, indicating lower viral loads. Ae. scapularis had a β coefficient of 11.83 (p < 0.001), while Ps. ferox and Ae. albopictus showed β values of 13.80 and 15.67, respectively (p < 0.001). Among Sabethes species, Sa. purpureus, Sa. albiprivus, and Sa. identicus exhibited the highest Ct values (β = 14.00 to 16.00, p < 0.01).

These results indicate species-specific differences in YFV viral loads, with Haemagogus species displaying lower Ct values compared to other genera. A full summary of the GLM estimates is presented in Table 3. The effect of season (rainy vs. dry) on Ct values was not statistically significant (p = 0.173). The interaction between mosquito species and season also did not significantly influence Ct values (p > 0.3 for all species).

Out of the 46 positive mosquito pools, 24 (52.2%) were collected during the rainy season (18th October – 4th April), and 22 (47.8%) during the dry season (Supplementary Material 2). Aedes species were predominantly collected during the rainy season, whereas Haemagogus spp. and Psorophora ferox were mostly collected during the dry season. Notably, Sa. albiprivus and Sa. identicus tested positive exclusively in the dry season. All YFV-positive mosquito pools were collected within the Atlantic Forest biome (Supplementary Material 3).

4. Discussion

Brazil is an endemic country for YFV, with the Amazon region acting as a source for viral diversity and dipersal across the country. Although YFV circulation has been documented in southeastern Brazil since the early 21st century, the 2016-2018 outbreak caused by the SA-I genotype, particularly in São Paulo state, was unexpected due to the high number of positive cases reported in both humans and animals [14,17]. Notably, during this outbreak, nine different species of Culicidae, including mosquitoes from the Aedes, Psorophora and Sabethes genera, tested positive for YFV. All these mosquitos were collected in the Atlantic Forest biome, where Haemagogus leucocelaenus act as the primary vector [18,19,20]. While Sabethes spp. are traditionally considered secondary vectors limited information is available regarding their role in YFV transmission in this region.

Our findings confirm that YFV viral loads varied accross Culicidae species, with Hg.janthinomys/capricornii. and Hg. leucocelaenus having the highest viral loads. Notably, two pools of Ae. scapularis also had viral loads comparable to those of Hg.janthinomys/capricornii., indicating that this species may play a more relevant role in YFV transmission than previously thought. These pools were collected in Urupês on February 15, 2017, and in Araçatuba on November 25, 2016, at the municipal Zoo, yet neither location reported epizootic events at the time. The presence of YFV in these areas could be attributed by different susceptibility of NTP, as some Callithryx sp. may be less susceptible to the disease [2]. Ae. scapularis, which was the most abundant specie collected in this study, is considered a generalist in its use of habitats, occurring in both sylvatic and human-dominated areas. Adult females are opportunistic in their behaviour, feeding especially on mammals [21,22]. Considering the wide host breadth and feeding habitats, coupled with synanthropic adaptions, it is possible that Ae. scapularis may be an important bridge vector for human and animal viruses. Thus, our data suggest that this species may have played a secondary role in the YF outbreak.

Sabethes mosquitoes were observed in low abundance, distribution, and infection rates, suggesting a local or secondary role during the 2016–2018 outbreak in the Brazilian Southeastern region [23]. In our study, this genus accounted for 11.5% (n = 430 pools) of the collected mosquitoes, with 10% (n = 5) of positive pools, all exhibiting high Ct values, indicative of low viral loads. Similarly, during the 2009 YF outbreak in São Paulo, YFV was only isolated from a single pool of Hg. leucocelaenus in Buri, despite the collection of Sa. chloropterus, Sa. purpureus, and Sa. undosus in the same area [19]. However, the absence of RT-qPCR analysis in that study may explain the lack of positive detections among Sabethini mosquitoes. Conversely, during a YF epidemic and epizootic in Misiones, a northeastern province of Argentina, YFV was successfully isolated in cell culture from pools of Sabethes albiprivus [24]. This viral isolation indicates high viral loads, contrasting with the low viral loads observed in Sabethes specimens from the Atlantic Forest.

Sa. chloropterus has been identified as the primary YF vector during the dry season in the Cerrado biome of Minas Gerais [25]. In Espírito Santo, where the sylvatic YF cycle was first described in Brazil, Sa. chloropterus, Sa. soperi, Sa. identicus, Aedes aureolineatus, and Shannoniana fluviatilis were noted for their secondary roles in YFV transmission [26]. Additionally, Sa. albiprivus from Rio de Janeiro demonstrated high vector competence when inoculated with Brazilian YFV strains [27]. To better elucidate the role of Sabethes mosquitoes in the YF transmission cycle within São Paulo state, where the virus has now been established [28], additional studies are required.

Considering the Aedes genus, earlier studies suggested that Brazilian Ae. aegypti mosquitoes might not favor the establishment of an urban cycle of YF [29]. However, a more recent study demonstrated that both anthropophilic mosquitoes, Ae. aegypti and Ae. albopictus, are highly susceptible to American and African YFV strains [27]. In 2018, in Minas Gerais state, a single Ae. albopictus mosquito pool tested positive for YFV [30]. In our surveillance study, Ae. albopictus was the second most frequent species collected, accounting for 21.66% of the total, with three pools testing positive for YFV, all of which exhibited low viral loads. No Ae. aegypti mosquitoes were found positive. Despite the high number of human infections during the outbreak, no urban YF cases were reported. Given that YFV has demonstrated potential for adaptation to Ae. albopictus and can be transmitted between NTP [31,32] our findings underscore a potential threat to endemic areas in South America where these mosquitoes are present. With their widespread distribution and ecological plasticity, Ae. albopictus could serve as a bridge vector, facilitating virus transmission between urban environments and rural areas.

One objective of this study was to assess whether seasonal variation (rainy vs. dry) influenced YFV viral loads in mosquitoes. Despite previous reports showing seasonal peaks in mosquito abundance and transmission during rainy periods our results indicate that season was not a significant predictor of Ct values, suggesting that once a mosquito is infected, viral replication remains stable. Sacchetto and collaborators reported viral persistence during the non-epidemic dry season in NTP collected in Belo Horizonte, Minas Gerais state [34]. These results show the importance of continuous surveillance, regardless of seasonal variations.

Our study has some limitations. Specifically, our study involved triturating whole mosquitoes instead of processing solely the salivary gland. Additionally, the contents of the mosquitoes' digestive systems—whether engorged or not—were assessed solely through visual examination, and some of the positive results could came from a residual blood feeding. Nevertheless, the data obtained in the present study is relevant, as monitoring of virus circulation and characterizing vectors are fundamental elements for understanding the dynamics of vector-borne viruses, providing new insights for the establishement of control strategies and to prevent the risk of re-urbanization of YFV. More, new studies of vectorial competence, mainly in Ae. scapularis, are needed.