Comparative Susceptibility of Aedes albopictus and Aedes aegypti to Dengue Virus Infection After Feeding on Blood of Viremic Humans: Implications for Public Health

Aedes albopictus is secondary to Aedes aegypti as a vector of dengue viruses (DENVs) in settings of endemicity, but it plays an important role in areas of dengue emergence. This study compared the susceptibility of these 2 species to DENV infection by performing 232 direct blood-feeding experiments on 118 viremic patients with dengue in Vietnam. Field-derived A. albopictus acquired DENV infections as readily as A. aegypti after blood feeding. Once infected, A. albopictus permitted higher concentrations of DENV RNA to accumulate in abdominal tissues, compared with A. aegypti. However, the odds of A. albopictus having infectious saliva were lower than the odds observed for A. aegypti (odds ratio, 0.70; 95% confidence interval, .52–.93). These results quantitate the susceptibility of A. albopictus to DENV infection and will assist parameterization of models for predicting disease risk in settings where A. albopictus is present.

Dengue is the most important arboviral infection of humans, with an estimated 00 million clinically apparent infections annually [1,2]. In dengue-endemic countries, Aedes aegypti (Linnaeus) is widely accepted to be the primary vectors of dengue viruses (DENVs), with Aedes albopictus (Skuse) regarded as a secondary vector. Yet A. albopictus can clearly transmit DENV at scales that are important to public health. For example, outbreaks of dengue  [3][4][5][6][7][8][9][10]. These outbreaks demonstrate that, although A. albopictus can be a competent vector of DENV, there must be biological features that render A. albopictus generally less well equipped to transmit DENV, compared with A. aegypti.
Aedes albopictus originated in Asia but has geographically spread through global trade, particularly via used car tires infested with A. albopictus eggs and larvae [11]. It is now distributed throughout the United States, Central America, and South America and in temperate African and European countries [12][13][14]. Numerous features have supported the expansion of A. albopictus, including the ability of the eggs to undergo diapause [15]. There is also speculation around their ability to outcompete established species of mosquitoes [16,17].
The expansion of A. albopictus has led to concern of an associated increase in the range of dengue transmission.
Lambrechts et al reviewed vector competence literature and reported that, although A. albopictus were more susceptible to DENV midgut infection, rates of virus dissemination to other tissues were significantly lower in A. albopictus than in A. aegypti [18]. Additionally, laboratory-reared A. albopictus became increasingly susceptibility to DENV, which may have been a confounding variable in the literature. Furthermore, the comparative vector competence literature has been derived entirely from laboratory experiments using artificial blood meals and laboratory-passaged viruses. How well these laboratory conditions replicate the pathogenesis of DENV transmission from viremic humans to mosquitoes in the field is uncertain.
The benefits of understanding why A. albopictus is largely a secondary vector of DENV in settings of endemicity are numerous. First, existing risk-analysis models of the likelihood of dengue outbreaks in southern Europe and the United States could be improved with quantitative estimates of A. albopictus vector competence [19,20]. Second, probability maps of dengue occurrence could be refined with better estimates of the relative vector competence of A. albopictus versus A. aegypti. Third, if the historical literature is correct and A. albopictus is more resistant than A. aegypti to disseminated DENV infection, then this provides an opportunity to identify speciesspecific antiviral defense mechanisms. Last, novel dengue control efforts using Wolbachia-infected A. aegypti will be better informed with an understanding of the vector competence of A. albopictus in candidate intervention settings [21]. To these ends, the current study compared the susceptibility of A. aegypti and A. albopictus to initial and disseminated DENV infection after direct blood-feeding episodes on viremic patients with dengue.

Ethics Statement
All participants provided written informed consent to participate in the study. The study protocols were reviewed and approved by the Scientific and Ethical Committee of the Hospital for Tropical Diseases (CS/ND/12/15) and the Oxford Tropical Research Ethical Committee (OxTREC 29-12). All investigations were conducted in accordance with the principles expressed in the Declaration of Helsinki.

Patient Cohorts
The study was performed at the Hospital for Tropical Diseases in Ho Chi Minh City, Vietnam, between September 2012 and November 2013. The inclusion criteria were age of ≥15 years, fever duration of ≤96 hours, and clinical suspicion of dengue; positive result of an NS1 rapid test; and provision of written informed consent. The exclusion criteria were current pregnancy, determined by clinical examination or a urine dipstick test for β-human chorionic gonadotropin; current intensive care unit stay; intellectual disability; a history of severe reactions to mosquito bites; and severe dermatological conditions. Demographic and clinical information were recorded prospectively in a standard case report form.

Mosquitoes for Blood-Feeding Experiments
All of the mosquitoes (A. aegypti and A. albopictus) that fed on patients with dengue were F3 generation and derived from 2 independent pooled larval collections, each sampled from 3 locations within 5 km of each other in Ho Chi Minh City. The A. aegypti used were distinct from those in our previous study [22]. Briefly, field-caught larvae were pooled and fed commercial dry fish and dog food. Mosquitoes were housed as described previously [22]. Briefly, adults (F1 generation) were kept in cages containing males and females in an environmental chamber with 12-hour cycles of light and dark at 27°C and 70% relative humidity. F1 females were provided blood meals by direct feeding on afebrile healthy human volunteers for multiple gonotrophic cycles over 45 days, with 15% sucrose provided freely in addition. Eggs from F1 females were hatched and reared and the subsequent F2 females provided with human blood meals as described above. When female F2 mosquitoes were 12 days old, they were killed and pooled into groups of 10 mosquitoes.
Each pool was homogenized and tested, along with appropriate controls, by reverse transcription-polymerase chain reaction (RT-PCR) to confirm the absence of DENV, Japanese encephalitis virus, and chikungunya virus. Eggs collected and stored from F2 females were the source of F3 females that were used for direct feeding on patients with dengue.

Experimental Exposure of Patients to Mosquitoes
Each patient was assigned a schedule of 2 exposures to mosquitoes on 2 different study days during the first 4 days after enrollment. The patient's forearm was exposed to 30-40 A. aegypti and A. albopictus aged 3-7 days (using a 2:3 ratio of A. aegypti to A. albopictus because preliminary experiments indicated that a higher fraction of A. aegypti took a blood meal) contained in a mesh-covered plastic cup that was held against the patient's forearm for 5 minutes. After 5 minutes, mosquitoes were returned to the insectary and subjected to cold anesthesia at 4°C for 45 seconds. Engorged mosquitoes were transferred to 500-mL plastic cups, separated by species, and maintained in an environmental chamber with 12-hour cycles of light and dark at 27°C and 70% relative humidity for 14 days. The number of dead mosquitoes was recorded daily.

Clinical Adverse Events
Postexposure severe adverse events were defined as any event that was clinically significant (ie, requiring clinical intervention, prolonged hospitalization, or admission to an intensive care unit) and possibly, probably, or definitely related to experimental exposure to mosquitoes.

Detection of DENV in Mosquito Tissues and Saliva
Technicians blinded to clinical and virological details of the participants performed laboratory assays of the mosquitoes. Mosquitoes were killed by cold exposure. The abdomen was dissected from the rest of the mosquito body and suspended in 0.5 mL of mosquito diluent (2% [v/v] heat-inactivated fetal calf serum [FCS] in Roswell Park Memorial Institute 1640 medium, antibiotics, and antimycotics). Individual mosquito abdomens were homogenized with 1-mm zirconia/silica beads for 15 minutes at 30 Hz by using a TissueLyser II system (Qiagen), as described previously [22]. Mosquitoes were scored as being infected with DENV, using a previously described quantitative RT-PCR analysis of homogenized tissue, with the results expressed as copies per abdomen [23]. Infectious virus in the saliva of individual mosquitoes was detected as described previously [22]. Briefly, the proboscis of dewinged and delegged mosquitoes was inserted into the end of a micropipette tip containing 6 µL of filtered saliva medium (a 1:1 solution of 15% [v/v] sucrose in normal saline and inactivated FCS). After 30 minutes, the 6 µL of saliva medium was ejected and then drawn into a pointed glass capillary tube (tip diameter, <0.3 µm), after which the volume of saliva medium derived from 1 mosquito was injected into the thorax of 4-6 A. aegypti (age, 4-7 days; volume injected, approximately 1 µL per mosquito). These injected mosquitoes were maintained for 7 days in an environmental chamber as described above. After 7 days, the mosquitoes were killed and the bodies pooled, homogenized, and tested by quantitative RT-PCR for DENV infection, with saliva samples scored as positive or negative depending on this result.

Sequence Amplification and Sequencing of the Gene Encoding DENV Envelope Protein
Viral RNA was extracted as previously described from plasma and mosquito abdomen tissues [22]. Complementary DNA (cDNA) was synthesized from 8 µL of viral RNA by using the Superscript III First-Strand Synthesis System for RT-PCR according to manufacturers' instructions. The genomic region spanning the genes encoding premembrane and envelope proteins was amplified from the cDNA in 16 amplimers. Subsequent steps were performed according to the manufacturers instructions for 454 GS-Junior Next Generation Sequencing method (Roche). Briefly, the primers in this first PCR round contained universal tails at the 5′ end to allow the addition of 454 sequencing-specific nucleotides and isolate-specific multiplex identifiers (also known as "barcodes") in a second PCR round. The first-round and second-round PCR analyses used FastStart High Fidelity polymerase (Roche). The long-read sequencing performance of the 454 GS-Junior (between 400 and 500 bases), in combination with a sample pooling strategy that uses barcoded amplicons, was used for parallel analysis of pooled samples. GS Mapping software was used for primer trimming and alignment of reads against a reference sequence DENV-1/VN/BID-V2732/2007 (GenBank accession number GQ199773.1). Sequence quality was measured using Phred (Q) scores with a minimum acceptable threshold of 95% of sequencing reads having Q scores of >20 (1/100 errors per base).

Dengue Diagnostic Tests
Serological responses were detected using immunoglobulin M (IgM) and immunoglobulin G (IgG) antibody-capture enzyme-linked immunosorbent assays in accordance with the manufacturer's instructions (Panbio, Australia). In accordance with the manufacturer's instructions, we classified serological profiles as "probable secondary" when >22 U of IgG were detected in either acute or early convalescent samples. If acute or early convalescent samples were IgM positive but IgG negative, serological profiles were classified as "probable primary." When IgM or IgG tests result were equivocal, we classified the serological profile as "indeterminate." DENV plasma viremia levels were measured by a validated quantitative RT-PCR assay that has been described previously [23].

Statistical Analysis
The probability of successful human-to-mosquito transmission was compared between A. aegypti and A. albopictus, using marginal logistic regression models. Models for assessment of abdomen samples were adjusted for the patient's log 10 -transformed viremia level, and models for assessment of saliva samples were further adjusted for the abdominal tissue viremia level. Analyses were performed for all patients and stratified by serotype. To account for potential within-patient correlation, model-robust sandwich standard error estimates were used throughout to construct confidence intervals (CIs) and P values. We derived 50% mosquito infectious doses (MID 50 values; defined as plasma viremia levels corresponding to a 50% probability of infection) for abdomen infection on the basis of marginal logistic regression coefficients for each mosquito type, and corresponding 95% CIs were calculated using the delta rule. Since the proportion of mosquitoes with infectious saliva was <50%, MID 50 values were not estimated. In the initial analysis of the transmission of DENV to saliva, mosquitoes with uninfected abdomens were assigned a log value of 0. This analysis was also run with exclusion of mosquitoes with uninfected abdomen's. The relationship between various covariates and the probability of successful human-to-mosquito transmission was assessed for each mosquito type, using a similar multivariable marginal regression model. The covariates assessed were day of illness, plasma viremia level, serotype, serological result, and abdominal viral burden. The number of mutations in the DENV-1 consensus sequence was compared between mosquito types, using a stratified version of the Wilcoxon test with stratification by patient [24]. All the analyses were performed using the R statistical software package, version 2.13.2 (R Foundation for Statistical Computing; Vienna, Austria).

Study Population Characteristics
Between September 2012 and November 2013, 120 patients with dengue were enrolled and experimentally exposed to field-derived A. aegypti and A. albopictus on 2 randomly allocated days within the first 4 study days. The patient enrollment flowchart is shown in Supplementary Figure 1. The final cohort for analysis comprised 118 DENV viremic patients with 232 independent mosquito exposure events. The baseline characteristics of the patients enrolled are shown in Table 1. Experimental exposure to mosquitoes was well tolerated, and no patient experienced a severe adverse event or required withdrawal from the study. DENV serotype 1 (DENV-1) and DENV-4 were responsible for 26% and 44% of cases, respectively, with DENV-2 (13%) and DENV-3 (16%) also represented.

Susceptibility of A. aegypti and A. albopictus to Acquisition of DENV Infection After Direct Feeding
There was a dose-response relationship between plasma viremia levels at the time of mosquito feeding and the proportion of mosquitoes of either species with DENV-infected abdomens 14 days later (Figure 1). The overall DENV MID 50 for A. aegypti was lower than that observed for A. albopictus, but this difference was not statistically significant (7.0 log 10 copies/mL [95% CI, 6.77-7.23] vs 7.1 log 10 copies/mL [95% CI, 6.9-7.3]; Table 2). Viral RNA concentrations of all 4 DENV serotypes were significantly higher in abdomen tissues from infected A. albopictus than in those from A. aegypti (Supplementary Table 1). Given that A. albopictus accommodated higher DENV RNA concentrations once infected, we examined whether this could result in greater virus sequence diversity. As a test case, the consensus nucleotide sequence of the gene encoding DENV-1 envelope protein was determined directly by analysis of plasma specimens from 10 patients and abdominal tissues from 20 A. aegypti and 20 A. albopictus that had taken blood meals from these cases (ie, 2 A. aegypti and 2 A. albopictus per patient). The median number of nucleotide differences between the consensus sequences of the gene recovered from plasma specimens and the gene recovered from  Viremia level, log 10 copies/mL 8.1 (6.9-8.7) 7.6 (7.1-8.9) 6.6 (5. Severe dengue 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Others 0 (0) 0 (0) 1 (5. Abbreviation: ICU, intensive care unit. a The viremia level for 1 patient was below the limit of detection.
b Data for 1 patient were missing.
c Data for 1 patient with DENV-4 infection were missing. d One patient was categorized as having a "viral infection" but was actually infected with DENV-3.
e One patient was categorized as having a "viral infection" but was actually infected with DENV-4.

Acquisition of Infectiousness Among A. aegypti and A. albopictus
As was observed in abdomen tissues, there was a dose-response relationship between the plasma viremia level and the proportion of mosquitoes with infectious saliva 14 days after blood feeding ( Figure 2). However, the likelihood of detecting infectious saliva differed by mosquito species and DENV serotype. Table 3 summarizes the odds of abdomen and saliva infection and demonstrates that the detection of infectious saliva was less likely in blood-fed A. albopictus, compared with blood-fed A. aegypti, in both unadjusted analysis and analysis that adjusted for plasma viremia level (adjusted odds ratio [OR], 0.70; 95% CI, .52-.93). By serotype, the odds of A. albopictus having infectious saliva were significantly lower for blood meals involving uptake of DENV-2 and DENV-4, compared with those involving uptake of DENV-1 or DENV-3 (Table 3). These data identified the odds of A. albopictus becoming infectious as lower than the odds of A. aegypti becoming infectious after feeding on the blood of viremic patients. This analysis used all engorged mosquitoes as the denominator. Supplementary Table 2 shows the odds of having infectious saliva, using mosquitoes with DENV-positive abdomens as the denominator. This analysis confirms that the odds of A. albopictus becoming infectious were lower than those for A. aegypti after viremic blood feeding (OR, 0.69; 95% CI, .49-.96).

Covariates and Their Association With Successful Human-to-Mosquito Transmission
In multivariable regression analysis, a higher plasma viremia level at the time of exposure was independently associated with a greater likelihood of DENV transmission to abdominal tissue for both mosquito types. Each 1-log increase in the   Table 4). These data highlight plasma viremia level as a risk factor for infectiousness among A. aegypti and A. albopictus. Supplementary Table 3 shows the association between these covariates and successful transmission, with mosquitoes with DENV-negative abdomens excluded from the model.

DISCUSSION
There is speculation on the arboviral disease risks posed by the invasion of A. albopictus into southern Europe and the United States [25]. Central to calibrating these risk assessments is an understanding of A. albopictus vector competence. There is conflicting literature on whether A. albopictus is less susceptible than A. aegypti to disseminated DENV infection [26][27][28][29][30][31][32]. Furthermore, all of the previous work in this area has used laboratory methods that do not mimic natural infection. Here, in experiments involving direct feeding on viremic patients with dengue, we provide evidence that A. albopictus had a susceptibility to acquiring DENV infection that was similar to that of A. aegypti. However, blood-fed A. albopictus was significantly less likely to become infectious with DENV-2 and DENV-4, suggesting an interspecies difference of epidemiological importance. A first step in the establishment of DENV infection in mosquitoes is attachment of virus particles to luminal receptors on the midgut epithelium, followed by virus entry via endocytic pathways. There are few insights into the nature of these receptors for DENV, although it has been proposed that the R67 and R80 proteins may act as midgut receptors leading to subsequent systemic infection [33]. Our finding that the plasma viremia MID 50 values for A. aegypti and A. albopictus were broadly similar suggests that DENV uses common mechanisms to attach and initially infect the midgut epithelium in these specimens. The MID 50 values for DENV-1 and DENV-4 measured in this study were concordant with those observed previously, whereas the MID 50 values differed for DENV-2 and DENV-3 [22]. We speculate that this reflects the smaller numbers of patients with DENV-2 and DENV-3 in this current study and hence decreased accuracy.
Our results identified that A. albopictus and A. aegypti were equally likely to develop infectious saliva containing DENV-1 and DENV-3. However, DENV-2 and DENV-4 were detected in saliva from A. albopictus less frequently than in saliva from A. aegypti. This might simply reflect intrinsic differences between the virus types in their virulence for A. albopictus, with DENV-1 and DENV-3 being fitter in this instance. However, it is interesting to note that, while DENV-1 was the dominant serotype in the Key West outbreak of dengue with A. aegypti as the vector, DENV-1 has not become further established in the southern United States, where A. albopictus is prevalent [34].
We explored the relationship between various covariates and the likelihood of successful human-to-mosquito transmission. We confirmed the importance of the plasma viremia level in successful human-mosquito transmission: each 1-log increase in plasma viremia level was associated with an approximately 5-fold increase in the odds of successful human-to-mosquito transmission for both A. aegypti and A. albopictus [22]. This finding again suggests that interventions that reduce DENV viremia during natural infection could have both an individual benefit and a public health benefit through reducing human infectiousness and thus reducing the risk of further transmission [19].
DENV RNA concentrations were significantly higher in abdominal tissues of A. albopictus than A. aegypti for all serotypes. When examined for DENV-1, this was also associated with higher levels of sequence drift away from the consensus sequence of the gene encoding envelope protein in the patient's plasma sample. Plausibly, by being more permissive to DENV replication in its abdominal tissues, A. albopictus might contribute to the genetic diversity that exists within DENV populations. Despite A. albopictus harboring higher viral burdens in abdominal tissues, significantly fewer A. albopictus had infectious saliva resulting from DENV-2 or DENV-4 infections. These data are broadly consistent with previous observations demonstrating that A. albopictus was more susceptible to fulminant DENV infection of the mosquito body but that viral dissemination to saliva was less than in A. aegypti [18]. That such nuanced interspecies differences can exist is perhaps not surprising, given the literature reporting variation in the susceptibility of A. aegypti populations to DENV infection, albeit under laboratory conditions [35][36][37][38][39]. Future studies to understand the basis for these phenotypes, particularly when using field-derived mosquitoes and direct feeding on human viremic hosts, will require large sample sizes to overcome the intrinsic variance in this system. While our findings have identified interspecies differences in susceptibility to DENV infection, the absolute differences between the mosquito types are nonetheless small. This points to factors other than outright susceptibility to DENV infection as important reasons why A. albopictus is a marginal contributor to dengue transmission in urban and peri-urban settings in dengue-endemic countries. Behavioral attributes are also important for vector competence, and thus we speculate that the tendency of A. aegypti to live proximate to humans is the major reason why it is the primary DENV vector.
Our study has limitations [18]. First, for practical reasons we measured the phenotype of blood-fed, field-derived mosquitoes at 14 days after feeding; data acquisition at other time points might have led to additional insights. Second, the results were limited in scope to the virus serotypes and genotypes in circulation during the study period, and consequently we acquired sparse data on DENV-2 and DENV-3. Additionally, there may be specific interactions between DENV types and mosquito genotypes that influence infection outcome [40]. Studies in other settings should be encouraged to understand the generalizability of this work.
In summary, we are the first to demonstrate that A. albopictus are less likely than A. aegypti to develop an infectious phenotype 14 days after direct blood feeding on viremic patients with dengue due to DENV-2 or DENV-4 infection. These results will enable more-accurate parameterization of DENV transmission models in regions where dengue is endemic and those that are at risk for endemicity. In addition, we have confirmed the central importance of plasma viremia in determining the likelihood of mosquito tissue infection, suggesting that interventions that attenuate viremia will have both individual and community benefits.

Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.