Infection rate of Aedes aegypti mosquitoes with dengue virus depends on the interaction between temperature and mosquito genotype

Abstract

Dengue fever is the most prevalent arthropod-transmitted viral disease worldwide, with endemic transmission restricted to tropical and subtropical regions of different temperature profiles. Temperature is epidemiologically relevant because it affects dengue infection rates in Aedes aegypti mosquitoes, the major vector of the dengue virus (DENV). Aedes aegypti populations are also known to vary in competence for different DENV genotypes. We assessed the effects of mosquito and virus genotype on DENV infection in the context of temperature by challenging Ae. aegypti from two locations in Vietnam, which differ in temperature regimes, with two isolates of DENV-2 collected from the same two localities, followed by incubation at 25, 27 or 32°C for 10 days. Genotyping of the mosquito populations and virus isolates confirmed that each group was genetically distinct. Extrinsic incubation temperature (EIT) and DENV-2 genotype had a direct effect on the infection rate, consistent with previous studies. However, our results show that the EIT impacts the infection rate differently in each mosquito population, indicating a genotype by environment interaction. These results suggest that the magnitude of DENV epidemics may not only depend on the virus and mosquito genotypes present, but also on how they interact with local temperature. This information should be considered when estimating vector competence of local and introduced mosquito populations during disease risk evaluation.

1. Introduction

Globalization fosters long-distance transport of commercial goods and widespread movement of humans and other organisms between geographical regions. This movement creates abundant opportunities for unintentional transportation of mosquitoes and other arthropods, as well as their associated pathogens, such as arthropod-borne viruses (arboviruses). As a consequence, vector-borne diseases that were previously geographically restricted can be readily introduced to new areas and to disease-naive human populations. This situation becomes exacerbated under certain climate change scenarios that predict regional temperature increases that would result in shifts or expansion of the geographical range of vectors of human pathogens, placing large numbers of humans at risk of infection.

Temperature restricts the range at which poikilothermic arthropod vectors can survive, breed, reproduce and blood-feed [1–7]. Temperature also affects vector susceptibility to arbovirus infection, the length of the extrinsic incubation period, arbovirus transmission rates and adult longevity [7–13]. Thus, temperature is a primary determinant of the distribution of arbovirus diseases and magnitude of outbreaks.

Dengue fever is the most widespread arbovirus disease in humans, with 390 million estimated infections per year and annual cases on the rise [14]. Every year, 3.6 billion people are at risk of infection and an effective approved vaccine is not yet available. Dengue virus (DENV) is a positive-sense single-stranded RNA virus in the family Flaviviridae. DENV is transmitted by female mosquitoes of the genus Aedes, mainly by the urban-dwelling mosquito Aedes aegypti. A complete life cycle of Ae. aegypti requires temperatures in the range of 15–37°C [15], with survival possible above 4°C [1]. Within these thermal limits, Ae. aegypti can be found in regions with widely different temperature regimes. Studies have shown that mosquitoes are predisposed to viral infections when reared at low temperatures or when exposed to low ambient temperatures after infection [8,16–18]. In insects and plants, the RNA silencing-mediated defence pathway (RNAi) against viral diseases [19,20] is compromised at low temperatures [21,22]. Temperature is also known to impact several aspects of the innate immune system in malaria-transmitting mosquitoes [23,24]. As temperature would compromise the ability of mosquitoes to survive, it is likely that local adaptation of Ae. aegypti takes place; thus mosquitoes from warmer regions may be more susceptible to viral infection at cold temperatures than mosquitoes from cold regions, i.e. cold-adapted mosquitoes should have less low-temperature suppression of the RNAi antiviral pathway. Alternatively, in the absence of local adaptation, temperature should impact viral infection rates equally in populations of Ae. aegypti, regardless of their local temperature. Additionally, because viral genotypes are known to differ in virulence [25,26], viral adaptation to the mosquito host or to the local temperature is also plausible and can potentially impact the mosquito infection rate.

It is then crucial to understand how ambient temperature impacts DENV infection of Ae. aegypti and whether these responses are population (genotype) specific. Genotype by genotype (G×G) interactions have been studied in the context of virus genotypes adapting to the local mosquito population, but evidence of local adaptation is not consistent [27,28]. Chikungunya virus transmission is affected by a three-way interaction between viral strain, Ae. albopictus mosquito genotype and environmental temperature [29]. Whether these same interactions occur for DENV and Ae. aegypti is unknown.

Here, we evaluate the effect of Ae. aegypti genotype (G_M), DENV serotype 2 (DENV-2) strain (G_V) and temperature (E) on the virus infection rate, as well as the presence of possible interactions among these effects using a common-garden experiment (figure 1). Our goal is to determine whether G_M × G_V × E interactions occur in order to determine the parameters that should be considered to accurately estimate the vector competence of local and introduced mosquito populations for disease risk evaluation.

Figure 1.

Experimental design of Ae. aegypti mosquito challenges with DENV-2. DENV-Hanoi (6H) and DENV-HCM (434S).

2. Material and methods

(a). DENV-2 strains and culture conditions

Two DENV-2 strains were provided by Dr Robert B. Tesh (University of Texas Medical Branch, Galveston, USA), from the World Reference Center for Emerging Viruses and Arboviruses collection hosted by the Center for Biodefense and Emerging Infectious Diseases. Both strains were isolated from human sera and passaged once on C6/36 cells (American Type Culture Collection (ATCC) #CRL-1660) derived from Ae. albopictus (table 1). Strain 6H was isolated in Hanoi, Vietnam; strain 434S was isolated from Long An Province, Vietnam (less than 50 km from Ho Chi Minh City). We refer to these as DENV-Hanoi and DENV-HCM, respectively, consistent with the corresponding mosquito populations (see below). Upon arrival in our laboratory, lyophilized aliquots were resuspended in 1 ml serum-free Dulbecco's modified Eagle's minimum essential medium (DMEM), and 400 µl of the suspension was used for in vitro infections to generate virus stocks on C6/36 cells (see the electronic supplementary material, methods). Viral stock titres (plaque-forming units (pfu) per millilitre) were estimated via plaque assay (see the electronic supplementary material, methods).

Table 1.

Description of dengue virus serotype 2 (DENV-2) strains from Vietnam used in this study.

strain	geographical origin	collection date	PFU/mla	infectivityb
DENV-Hanoi (6H)	Hanoi	5 Nov 1986	2.08 × 107	0.0159
DENV-HCM (434S)	Long An Province	28 Apr 1988	3.77 × 107	0.0184

^aPFU/ml = plaque-forming units per millilitre.

^bInfectivity = number of infectious particles divided by the total number of viral particles detected by real-time PCR.

(b). In vitro DENV-2 infections and temperature treatments

In vitro experiments were used as a simplified system to address the effect of DENV-2 genotype (G_V) and temperature (E) on viral growth. In this system, cellular complexity is reduced to a relatively homogeneous cell monolayer, compared with the richer mosquito midgut environment in which a variety of cell types and morphologies are present [30] and where the microbiota and secreted housekeeping enzymes may influence pathogen virulence [31–33]. These experiments aim to provide a basal understanding of viral growth curves and uncover any differences among the DENV-2 strains that may arise independently of the mosquito genotype. In vitro infections were performed on confluent cell monolayers (approx. 8.6 × 10⁵ cells well⁻¹ of a six-well plate) of Aag-2 cells, derived from Ae. aegypti. Experiments were carried out in triplicate from a common DENV-2 stock using 200 µl DENV-2 aliquots prepared from a frozen stock (table 1) to achieve initial infection conditions at a multiplicity of infection (MOI) equal to 0.5 virus particles per cell (see the electronic supplementary material, methods). Aliquots (200 µl) were taken on days 1, 3, 5 and 7 post-infection, replacing them with fresh growth media, and frozen at −80°C for subsequent titration by RT-PCR.

(c). Virus quantification via real-time PCR

Real-time PCR is a more efficient method of titrating viral samples than plaque assays, but they should be calibrated owing to differences in the number of infectious virus particles per total number of viral genomes among different virus genotypes. To calibrate the RT-PCR assay used in this study, we artificially generated a 541 bp fragment of DENV-2 RNA by transcribing a PCR product amplified from an infected Ae. aegypti mosquito with the primers listed in electronic supplementary material, table S1, using the MEGAscript kit (Life Technologies, Carlsbad, CA, USA). This fragment was used to generate the standard curve for reference. The concentration of viral particles was assayed via real-time PCR with the SuperScript III Platinum One-Step Quantitative RT-PCR System (Life Technologies) using primers and probes described in [34] for DENV-2 (electronic supplementary material, table S1). DENV-2 growth in cell culture and the presence or absence of DENV-2 infection in Ae. aegypti mosquitoes were subsequently surveyed using this method. Briefly, viral RNA was extracted from frozen samples (50 µl) using the KingFisher™ Flex Purification System (ThermoFisher Scientific, Waltham, MA, USA) with the Mag-Bind^® Viral DNA/RNA Kit (Omega Bio-tek, Norcross, GA, USA) and eluted in a final volume of 50 µl of 1/10X TE buffer (Tris–EDTA). Two microlitres of this elution were then used for the RT-PCR assay described above. Samples were only thawed once. Infectivity of DENV-2 was defined as the percentage of infectious particles per millilitre (as determined from the plaque assay; see the electronic supplementary material, methods) relative to the total number of viral copies in the sample calculated from the real-time PCR data (table 1).

(d). DENV-2 strain sequencing

Sequencing of the two DENV-2 strains was performed to confirm that they were genetically different. Viral RNA was extracted from the lysate of infected cell cultures using the QIAamp Viral RNA mini Kit (Qiagen, Hilden, Germany), followed by cDNA synthesis with SuperScript II Reverse Transcriptase (Life Technologies) and Random Hexamer primers, according to the manufacturer's instructions. Three segments of the DENV-2 genome were then amplified using the primers described in electronic supplementary material, table S1. Fragments were Sanger sequenced with an Applied Biosystems 3730xl DNA Genetic Analyser at the DNA Analysis Facility at Science Hill at Yale University. As only partial sequences were obtained, segments were concatenated for analysis. Consensus sequences were then aligned using MUSCLE [35], as implemented by Geneious v. 5.3 [36].

(e). Aedes aegypti populations

Aedes aegypti mosquitoes (provided by Dr Duane J. Gubler, Duke-National University of Singapore, Singapore) arrived at Yale University as F1 eggs from females collected in Hanoi and Ho Chi Minh City (HCM) in Vietnam (figure 2), between August and September of 2013. To collect eggs, they were blood-fed at the National Institute of Hygiene Epidemiology insectary located in Hanoi, Vietnam. Colonies were established with more than 500 individuals from these F1 eggs at the Yale School of Public Health insectary (see the electronic supplementary material, methods). Adults from laboratory generations six to nine from both populations were used in this study.

Figure 2.

Map of Vietnam showing sample locations and their corresponding temperatures. Ave, average; min–max, minimal and maximal temperature as reported by WeatherBase (http://www.weatherbase.com).

(f). Aedes aegypti genetic differentiation

The two Ae. aegypti populations from Vietnam were genotyped at 12 microsatellite loci to confirm that they were distinct genetic units, as previously described [37–39] (electronic supplementary material, methods). Population differentiation was evaluated in Vietnam and Asia using the Bayesian clustering method implemented by the software STRUCTURE v. 2.3 [40] using K = 2 and K = 4 clusters, respectively, and microsatellite data from Gloria-Soria et al. [37]. STRUCTURE assigns individuals to K clusters with no a priori information of sample location. Each run assumed an admixture model and correlated allele frequencies using a burn-in value of 100 000 iterations followed by 500 000 repetitions. Discriminant analysis of principal components (DAPC) on allele frequencies was used as an alternative to STRUCTURE to infer genetic clusters in the same groups. This multivariate statistical approach partitions the sample variance into between-group and within-group components in order to maximize discrimination between groups [41]. Linear combinations of alleles that optimize the discrimination between groups are called discriminant functions (DA) [41]. DAPC was performed with the ADEGENET package [42] in R v. 3.2.0. [43]. Pairwise genetic distances (F_ST) were calculated in Genodive 2.0b.27 [44].

(g). Mosquito oral infections and temperature treatments

Batches of approximately 100 Ae. aegypti females were simultaneously challenged with either DENV-Hanoi or DENV-HCM virus and incubated for 10 days at three different temperatures (25, 27 and 32°C), as indicated in figure 1. Each experimental treatment was repeated three times. The replicates were performed on different days, with new batches of mosquitoes, and new virus aliquots. The presence of virus in bodies was assayed by RT-PCR after RNA extraction, as described above. Mosquitoes with more than 100 virus copies per microlitre were considered positive for infection, as measured by the RT-PCR standard curve. Reported infection rates are the percentage of mosquitoes tested that were positive for virus in the body. Details of the in vivo infections can be found in the electronic supplementary material, methods.

(h). Statistical analysis

Differences in DENV-2 replication (concentration) among treatments in cell culture were analysed using a two-way mixed factorial design analysis of variance (ANOVA), considering replicate measures as error terms. All paired comparisons were performed as Tukey contrasts with the multcomp package [45] available for R [43]. Results from the in vivo treatments were analysed by considering infection as a binary response (positive/negative) in a full-factorial generalized linear model that included mosquito population, virus strain and temperature as factors, as well as their interactions, incorporating replicates (of the experimental treatment) as a random effect. The model was fitted using the logit function and statistical significance was assessed via ANOVA with a χ²-test at a threshold of p < 0.05. Comparisons within specific temperatures to test for equality of proportions were performed using a non-parametric method (χ²-test) with continuity correction. All analyses were implemented using the R software [43].

3. Results

(a). Different DENV-2 genotypes have similar infectivity rates in vitro

The two DENV-2 strains were characterized genetically and phenotypically. A total of 3075 bp of good-quality sequence was obtained from the three fragments amplified from each DENV-2 strain. Alignment of the concatenated sequences uncovered 17 fixed polymorphic sites (SNPs) between the DENV-Hanoi and DENV-HCM strains, with six being non-synonymous substitutions, confirming that the two DENV-2 isolates are distinct genotypes (electronic supplementary material, file S1). Comparison between the plaque assays and the real-time PCR analysis found that the two strains produced similar numbers of infectious particles (table 1). This information allowed us to perform direct comparisons of real-time PCR data on both strains in subsequent analyses.

(b). Warmer temperatures promote DENV-2 growth in cell culture

The growth dynamics of DENV-Hanoi and DENV-HCM in Ae. aegypti-derived Aag-2 cells were compared following incubation at 25, 28 and 32°C for 7 days, with sampling on days 1, 3, 5 and 7 (electronic supplementary material, figure S1). Analysis of DENV-2 titres showed an effect of incubation temperature (F_{d.f. 1,69} = 103.511; p < 0.001) and day post-infection (F_{d.f. 1,69} = 7.033; p = 0.01) on the number of virus particles present in the culture supernatant, but no effect of DENV-2 strain (F_{d.f. 1,69} = 3.135; p < 0.0813); see electronic supplementary material, figure S1. Virus concentration was affected by the interaction between temperature and day post-infection (F_{d.f. 1,69}= 15.673; p = 0.0002), but not by the interaction between temperature and viral strain (F_{d.f. 1,69} = 0.089; p = 0.76583).

At all temperatures, DENV-2 concentration peaked after 5 days post-infection (electronic supplementary material, figure S1), although no significant difference in titre was detected after day 3 (z_D3−D5 = 2.553, p = 0.0521; z_D3−D7 = 1.071, p = 0.7071 and z_D5−D7 = −1.482, p = 0.4485). No difference in the number of virus particles between the DENV-Hanoi and DENV-HCM strains was observed at this time point ( p = 0.4457). Virus concentration was the lowest at 25°C and increased with temperature (z_25–28 = 3.69, p < 0.001 and z_28–32 = 7.787, p < 0.001; electronic supplementary material, figure S2).

These results indicate that the two DENV-2 strains respond equally to temperature under the simplified conditions of cell culture.

(c). Aedes aegypti populations from Vietnam are genetically distinct

Genotyping the two Ae. aegypti populations from Vietnam with 12 microsatellite loci confirmed that they were genetically distinct. Bayesian clustering analysis showed the two populations as genetically closer to each other than to other Asian populations (figure 3a). However, clustering analysis within Vietnam distinguished Hanoi and HCM as individual Ae. aegypti populations (figure 3b), consistent with the DAPC shown in figure 3c,d. Differentiation between the two Vietnam populations was further confirmed by significant F_st values (F_st = 0.1076, p < 0.05), while genetic differentiation across generations of each mosquito population used for this study (Gen 6–9) was minimal (F_ST < 0.017 and 0.020 for Hanoi and HCM populations, respectively).

Figure 3.

Genetic structure of Ae. aegypti populations within Asia and in Vietnam. (a,b) STRUCTURE bar plots indicating relatedness among geographical locations based on 12 microsatellite loci. Each vertical bar represents an individual. The height of each bar represents the probability of assignment to each of K = 4 and K = 2 clusters, respectively, as determined using the Delta K method. Each cluster is depicted with a different colour. (c) Plot of the discriminant analysis of principal components (DAPC) on microsatellite allele frequencies for Asia. Discriminant functions 1 and 2 are shown. (d) DAPC of Hanoi and Ho Chi Minh populations. Plot shows the densities of individuals from each population given discriminant function 1. Hanoi: red, HCM: black.

(d). Two-way interaction between Aedes aegypti genotype and temperature determines infection rate by DENV-2

The in vivo responses of DENV-2 strains from Vietnam to changes in temperature were different from the in vitro responses. We evaluated the effect of Ae. aegypti populations, DENV-2 strains and extrinsic incubation temperature (EIT), and their interactions, by performing a common-garden experiment that included all 12 possible combinations of mosquito × virus genotypes (figure 1). Viral infection was surveyed in 821 Ae. aegypti females from Hanoi and Ho Chi Minh City challenged with two strains of DENV-2 virus (DENV-Hanoi and DENV-HCM) and incubated at three different temperatures (25, 27 and 32°C) for 10 days. All four possible mosquito and virus combinations were present in each temperature block (table 2), and each temperature block was repeated on 3 different days. None of the control mosquitoes run simultaneously was positive for DENV-2. The percentage of mosquitoes infected was directly affected by the EIT and DENV-2 strain, but not by the Ae. aegypti population (tables 2 and 3). However, the proportion of infected mosquitoes was influenced by the interaction between the mosquito population and the EIT (table 3 and figure 4; electronic supplementary material, figure S3).

Table 2.

Results of Ae. aegypti oral infections with DENV-2 for each experimental treatment.

Aedes aegypti population	temperature (oC)	Na		infectedb
		DENV-Hanoi	DENV-HCM	DENV-Hanoi	DENV-HCM
Hanoi	25	94	98	4 (4.2)	8 (8.1)
	27	55	100	5 (9.1)	13 (13)
	32	80	47	3 (3.7)	2 (4.2)
total		229	245	12 (5.2)	23 (9.4)
Ho Chi Minh	25	74	65	8 (10.8)	16 (24.6)
	27	70	51	2 (2.8)	5 (9.8)
	32	22	65	0 (0)	5 (7.7)
total		166	181	10 (6.0)	26 (14.4)

^aNumber of mosquitoes challenged.

^bNumber of mosquitoes infected (percentage of total challenged).

Table 3.

ANOVA table for Ae. aegypti infections with DENV-2 strains from Vietnam. Significant values are depicted in italics. d.f. = degrees of freedom; p-values as calculated with the χ²-test.

parameter	d.f.	deviance	p-value
temperature	2	7.0619	0.0292
Aedes aegypti	1	2.1591	0.1417
DENV-2 strain	1	10.2603	0.0013
temperature: Ae. aegypti	2	9.9156	0.0070

Figure 4.

Genotype by genotype by environment (temperature) interactions between Ae. aegypti and DENV-2 from Vietnam. Filled lines in blue: Hanoi mosquitoes; dashed lines in red: Ho Chi Minh mosquitoes. Shaded areas are standard errors of the estimated sample proportion (s.e._p = sqrt[p(1 − p)/n]). *Infections with DENV-2 from Hanoi; †infections with DENV-2 from HCM. Points are connected to facilitate visualization. See Figure S3 for an alternative representation of the data.

Overall, these results indicate that the mosquito population from warmer climate (Ho Chi Minh) was more susceptible to infection at a lower EIT than the mosquito population from a cooler climate (Hanoi), when infected with DENV-HCM (χ² = 7.16, p < 0.05). No statistically significant differences in the infection rate between the Hanoi and HCM mosquitoes were found at other EIT, regardless of the virus strain.

4. Discussion

We analysed the effects of the Ae. aegypti genotype, DENV-2 strain and temperature on the infection rate of mosquitoes derived from geographical regions that experience different temperature regimes (figure 2). Results of in vitro experiments indicate that temperature influences viral growth rates. A direct effect of temperature on the DENV-2 infection rate has been previously reported in studies with Ae. aegypti [13] and is known to affect infection rates of Culex tarsalis with western equine encephalomyelitis virus [11]. Our experiments also revealed an effect of viral strain on the infection rate in vivo but not in vitro, indicating that midgut infection is a more complex process than infecting a cell monolayer and that DENV strains vary in their ability to establish the initial infection in the mosquito. Thus, tissue culture systems do not necessarily predict dengue infection outcomes in intact mosquitoes. The effect of the DENV-2 strain on the infection rate could explain in part the pattern of lineage replacement observed typically in DENV strains around the world, as some genotypes would be more successful at initiating the infection than others in a novel environment [46,47]. Our data also show that mosquito infection rate was influenced by an interaction between temperature and mosquito genotype, a genotype by environment effect (G_M × E). Failloux et al. [48] reported variation in DENV-2 dissemination among Ae. aegypti populations, but the incubation temperature was not reported. In our experiments, the effect of mosquito genotype was detected only in the context of temperature. Lambrechts et al. [28] and Fansiri et al. [27] have reported a G_M × G_V interaction between DENV-1 virus and Ae. aegypti populations from Thailand. However, we did not observe this interaction in our comparison involving DENV-2 strains and mosquitoes from Vietnam. Rapid DENV turnover [49,50] may explain why we did not detect a G_M × G_V interaction. Despite their common geographical origin, the virus isolates and mosquito samples in this study were collected in different years. The DENV-2 strains used circulated in the mosquito population approximately 20 years ago and have been replaced by new DENV genotypes since then. This makes it unlikely that these particular viruses have adapted to the mosquito populations currently inhabiting the same geographical area. Nevertheless, the daily temperature range in Vietnam has not changed much during the intervening 20 years and the overall temperature profile (figure 2) of both Hanoi and HCM has been steady [51]. Under these circumstances, temperature adaptation of the virus to the local environment could have occurred. However, unlike for chikungunya virus [29], we did not see a three-way interaction between Ae. aegypti, DENV-2 genotype and temperature. The low DENV infection rates from our study may have prevented us from detecting minor sources of variation due to a relatively small sample size of infected mosquitoes.

It is possible that the different pattern of infection we observed among the Ae. aegypti populations at the low temperature reflects local temperature adaptation. Low larval rearing temperatures are known to increase chikungunya competence [8], and fluctuations around low average temperatures, but not at high temperatures, increase vector transmission [9,12]. Cold temperatures are known to destabilize the RNAi pathway and make mosquitoes more susceptible to chikungunya and yellow fever virus infection [52]. In C. tarsalis, this is also the case for infections with western equine encephalomyelitis virus [11]. Annual temperatures in Hanoi are 5°C lower than in Ho Chi Minh City and the difference can reach up to 7°C during the winter months (figure 2). It is conceivable that adaptation of the Hanoi population of Ae. aegypti to colder temperatures leads to a more robust immune system that is not compromised under low-temperature conditions, resulting in lower DENV-2 infection success when compared with the Ae. aegypti from Ho Chi Minh City, which may spend less time in such temperature extremes. Alternatively, these mosquito populations may differ in their feeding strategies. Population differences in the total volume of blood ingested may impact the infection rate in a dose-dependent manner [53]. Whether the G_M × E effect in our data reflects adaptation of the mosquito population to the local temperature regime can only be hypothesized, because temperature adaptation of the Ae. aegypti populations was not measured in this study.

Although G × E and G × G studies have been done before in vector–pathogen systems, this is the first study to investigate G × G × E effects on Ae. aegypti infection with DENV. Our study is also unique in that we went beyond inferring genetic distinctiveness of mosquito populations from their geographical location alone, or from the sequences of a couple of genes. Instead, we conducted population genetic analyses on 12 highly polymorphic loci distributed across the genome. Nevertheless, further experiments are needed to demonstrate that genetic divergence is related to temperature adaptation. Likewise, this study was restricted to the study of two DENV-2 isolates and two mosquito populations; additional work should focus on expanding this research to include additional geographical locations, dengue serotypes and genotypes, and Ae. aegypti populations, to determine the universality of these conclusions and subsequently elucidate the mechanisms behind the observed differences.

We emphasize that we only evaluated Ae. aegypti infection rates and did not measure viral dissemination and transmission. Infection of mosquito vectors is the first step required for endemic transmission of DENV [54] and thus a relevant parameter to consider when studying vector competence. Nevertheless, the transmission potential of a vector is determined by a complex network of intrinsic and mechanical factors, including immune pathways and tissue barriers that regulate arbovirus escape from the midgut, infection of salivary glands and ultimately the virus release into the saliva [54,55]. Subsequent experiments should go beyond measuring midgut infection rates, measuring dissemination to other tissues and the presence of infective virus in the saliva, to ultimately encompass G × G × E effects on dengue transmission potential.

Dengue incidence is expected to increase over the next decades as exotic Ae. aegypti populations and DENV genotypes are introduced to new areas through human migration and trade, and as Ae. aegypti expands its geographical distribution in response to global temperature increases that facilitate year-round survival at northern (colder) latitudes [56]. Studies of vector competence for arboviruses are usually performed at a standard 28°C temperature. Our results suggest that accurate vector competence assessments of these populations should incorporate the local temperature. Our data suggest that introduction of vectors from warmer to cooler areas can result in higher infection rates than in their native range depending on the virus genotype. Changes in local temperature patterns resulting from climate change will thus have a positive or negative impact on the epidemiology of arbovirus transmission based on the introduced vector and pathogen.

Ethics

This research did not involve humans or animals (vertebrates) that require animal care protocols. Aedes aegypti mosquitoes are not endangered or protected species. All infections were carried out within the Connecticut Agricultural Experiment Station BSL-3 laboratory under Arthropod Containment Level-3.

Data accessibility

The datasets supporting this article have been uploaded as part of the electronic supplementary material. DENV-2 DNA sequences and alignment: electronic supplementary material, File S1. Microsatellite genotypes: electronic supplementary material, File S2.

Authors' contributions

A.G.-S. participated in the design of the study, carried out the infections and the molecular work, conducted data analysis and drafted the manuscript. P.M.A. participated in the design of the study, and helped with mosquito infections and in drafting the manuscript. J.R.P. conceived and designed the study, and helped draft the manuscript. P.E.T. helped design the study, contributed to data analysis and helped draft the manuscript. All the authors gave their final approval for publication.

Competing interests

The authors declare no competing interests.

Funding

Financial support was provided by the US National Institutes of Health NIAID: RO1 AI101112 awarded to J.R.P. and R01 AI091646-01 awarded to P.E.T.; by the US National Science Foundation DEB-1021243 awarded to P.E.T.; and by the Centers for Disease Control and Prevention U50/CCU116806-01-1, the US Department of Agriculture Hatch Funds (CONH00773) and the Multistate Research Project (NE1443) to P.M.A. A.G.-S. was supported by NIAID 3R01AI091646-04S1.