Individual-based dengue virus surveillance in Aedes aegypti mosquitoes collected concurrently with suspected patients in Tarlac City, Philippines

Background Dengue virus (DENV) infection continues to be a major public health concern throughout tropical and subtropical regions of the world where Aedes aegypti mosquitoes, its primary vector, dwell. In the context of DENV transmission, effective control is reliant not only on knowledge of mosquito abundance, but also on mosquito infection. In the 2015 dengue season, we conducted a one-month entomological surveillance of adult Aedes aegypti mosquitoes around households of suspected dengue patients in Tarlac City, Philippines to assess the DENV infection rate in the local mosquito population, and to identify the DENV genotypes and serotypes concurrently co-circulating in mosquitoes and patients. Methods We performed a one-step multiplex real-time RT-PCR assay for the simultaneous detection and serotyping of DENV in patients and in individual female Aedes aegypti mosquito. Consequently, we performed sequencing and phylogenetic analyses to further characterize the detected DENVs in mosquitoes and patients at the genotype level. Results We collected a total of 583 adult Aedes aegypti mosquitoes, of which we tested 359 female mosquitoes individually for the presence of the DENV. Ten mosquitoes (2.8%) from amongst 359 female mosquitoes were conrmed to be positive for the presence of the DENV. We detected DENV-1, DENV-2, and DENV-4 in the eld-collected mosquitoes, which were consistent with the serotypes concurrently infecting patients. Sequencing and phylogenetic analyses of the detected DENVs based on the partial envelope ( E ) gene revealed three genotypes concurrently present in the sampled mosquitoes and patients during the study period, namely: DENV-1 genotype IV, DENV-2 Cosmopolitan genotype and DENV-4 genotype II. Notably, we observed a random geographic distribution of DENVs in the study area suggesting the occurrence of active DENV transmission within and outside the vicinities of Tarlac City. Conclusions In this study, we demonstrate the utility of an individual-based DENV surveillance in eld-collected mosquitoes and the importance of incorporating mosquito virus data in phylogenetic studies. Analyzing virus sequences from vector and host could potentially improve our understanding of the dynamics of DENV transmission. was obtained, blood was collected and then serum was separated. The presence of DENV NS1 antigen was initially tested using PanBio® Dengue Early Rapid Kit (Alere Medical Co. Ltd., Massachusetts, USA) using serum. Laboratory diagnosis of dengue cases were conrmed based on virus isolation using Vero 9013 (African green monkey) cells. Ten microliters of serum were inoculated in Vero 9013 cells in minimum essential medium (MEM) supplemented with 10% fetal bovine serum (FBS) and 100 U/ml of penicillin. Plates were incubated in 34 o C, 5% CO 2 and infected culture uid (ICF) was harvested after day 7 and 14 of incubation period. Viral RNA was extracted from serum and ICF using QIAamp MinElute Virus Spin Kit (Qiagen, Hilden, Germany) based on the manufacturer’s protocol. DENV detection and serotyping were performed using a multiplex real-time RT-PCR method [30]. RT-PCR amplication of the DENV E gene followed by sequencing was also done as additional diagnostic evidence. eld-collected mosquitoes. This method is capable of detecting all 4 DENV serotypes in a single mosquito in one run. Our results from the RT-PCR amplication and sequencing of the DENV partial E gene revealed an infection rate of 27.9 DENV-infected females per 1000 female Aedes aegypti mosquitoes in Tarlac City during the one-month DENV surveillance. This result was comparable to the calculated (female) infection rates in previous studies where pooled mosquitoes were analyzed and mosquito collections were done in longer periods (Table S5). The relatively high infection rate we obtained may be attributed to the purposive surveillance of mosquitoes we implemented around dengue suspected cases, and also to the individual-based DENV detection approach we employed in this study. Individual-based virus detection in eld-collected mosquitoes is seldom performed in mosquito surveillance studies due primarily to nancial costs. However, in this study, we opted to employ an individual-based virus detection approach in order for us to more accurately estimate the infection rate in the study area, and to assess the circulating genotypes and serotypes among mosquitoes and patients at the individual level. Hospital, and to the staff of the Local Government Unit of Tarlac City for their help and support in the hospital-based patient surveillance and mosquito surveillance. Our deepest thanks also go to the Tohoku-RITM Collaborative Research Group for their assistance in the virus isolation, detection and sequencing of DENV in patient samples as well as their helpful comments in the detection and characterization of DENV in eld-collected mosquitoes. We are also grateful to Katherine Viacrusis for her technical assistance with the mosquito surveillance.


Introduction
The increasing incidence and expanding geographical range of dengue virus (DENV) infections are causes for international concern. DENVs are transmitted through a human-to-mosquito-to-human cycle throughout tropical and subtropical regions of the world, with the Aedes aegypti mosquito as the primary vector [1]. While DENVs isolated from patients is vital in dengue disease surveillance, the complementary data from mosquitoes, including viral sequences, mosquito infection rate and serotype/genotype prevalence, has the potential to provide additional information in understanding the transmission dynamics of the DENV. For this reason, virus surveillance in eld-collected mosquitoes is useful in tracking virus activity and in implementing control measures [2,3,4,5].
Detection of DENV in adult female Aedes aegypti mosquitoes remains a challenge due to the low infection rate (typically about 0.1%) observed in adult female mosquitoes [6]. However, recent advancements in molecular virus detection techniques, particularly nucleic acid ampli cation tests such as RT-PCR and real-time RT-PCR assays, have enabled researchers to directly detect DENV RNA in eld-collected mosquitoes [3,4,5,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21]. The sensitivity, speci city and speed of nucleic acid ampli cation tests provide an alternative to the traditional method (i.e. virus isolation by cell culture) of detecting DENV. Current testing of mosquito populations for DENVs has been limited to RT-PCR of mosquito pools. Pool screening of mosquitoes has been widely utilized in mosquito-based virus surveillance programs due to its cost-effectiveness, and also, in part, due to the small amount of viral RNA recovered from a single mosquito. However, one important consideration of pool screening is the choice of pool size as the inappropriate choice of pool size may lead to inaccurate estimation of infection rates. Given that there is no generalized procedure for size pooling of mosquitoes, indicators used for estimating levels of mosquito infection, minimum infection rate (MIR) and maximal likelihood estimation (MLE), cannot warrant the accurate estimation of proportions of infected mosquitoes [22,23].
In order to monitor the infection rate in eld-collected mosquitoes with higher precision, an individual-based approach would be useful. Individual-based DENV detection using RT-PCR has been reported to be technically possible using laboratory-infected mosquitoes [7,10]. Thus far, only one eld study employed an individual-based RT-PCR approach in detecting DENV in local mosquitoes [8]. Utilizing an individualbased approach not only allows a more accurate estimation of infection rate, but also allows the direct sequencing of DENV RNA from a single mosquito that could potentially improve the robustness of constructing DENV phylogenetic trees.
By utilizing DENV sequences from mosquitoes and patients, taxa-inclusive phylogenetic analysis comprising both vector and host [24] could be obtained potentially improving our understanding of how DENV moves in a human-to-mosquito-to-human cycle. Knowledge of spatial patterns of DENV infection is generally limited because most studies only emphasize symptomatic infections and do not account for asymptomatic infections [25]. These asymptomatic infections are possibly an important component of the overall burden of dengue because these infections may likely serve as hidden reservoir hosts for mosquito infection [26] as asymptomatic people can be infectious to mosquitoes despite their lower level of viremia [27]. Viral data from eld-collected mosquitoes has the potential to detect these asymptomatic infections [15] providing additional information in understanding DENV dynamics. In this context, incorporating data on infected female Aedes aegypti mosquitoes into the current patient-based dengue surveillance system might aid in increasing its sensitivity by enhancing its ability to predict and prevent an outbreak as well as detect silent virus [7,15,19]. Moreover, analyzing DENV sequences isolated from mosquitoes and patients has the potential to reveal whether DENV is geographically constrained in the household-or community-level.
To this end, we conducted a one-month DENV surveillance in mosquitoes collected around households of suspected dengue patients during the 2015 dengue peak season in Tarlac City, Philippines in order to assess the distribution of DENVs present in the local mosquito population. We utilized mosquito virus data to serve as supporting evidence to the DENVs detected in humans during the same period. Our objectives were two-fold: (1) provide a more accurate estimate of DENV infection rate in mosquitoes by employing an individual-based one-step multiplex real-time RT-PCR assay, and (2) assess the DENV serotype and genotype distribution circulating in mosquitoes and patients during the same period. In this study, we highlight the importance of detecting and characterizing the DENV in a naturally infected mosquito and the potential of mosquito virus data alongside patient virus data in inferring DENV transmission dynamics.

Study area
This study was conducted in Tarlac City, the provincial capital of Tarlac province located in Central Luzon, Philippines. The city is situated at the center of Tarlac province and is a densely populated peri-urban area that encompasses 274.66 km² with a total population of 342, 493 inhabitants in 2015 [28]. The population density is 1,247 habitants per square kilometer. The city is composed of 76 barangays (i.e. village equivalent); of these, 19 barangays comprise the urban area as de ned by the 2000 Census of Population and Housing [29]. Maps were created using the QGIS 3.6 software and data were acquired from the Philippine GIS Data website (www.philgis.org).

Patients' recruitment and laboratory diagnosis
In 2015, high prevalence of dengue occurred in Tarlac City with a total number of 1577 dengue cases (no reported deaths). For this study, febrile inpatients within 5 days from onset of symptoms and suspected of having dengue infection (Dengue fever onset: August 1 to October 31, 2015) in the Tarlac Provincial Hospital were recruited. After the informed consent was obtained, blood was collected and then serum was separated. The presence of DENV NS1 antigen was initially tested using PanBio® Dengue Early Rapid Kit (Alere Medical Co. Ltd., Massachusetts, USA) using serum. Laboratory diagnosis of dengue cases were con rmed based on virus isolation using Vero 9013 (African green monkey) cells. Ten microliters of serum were inoculated in Vero 9013 cells in minimum essential medium (MEM) supplemented with 10% fetal bovine serum (FBS) and 100 U/ml of penicillin. Plates were incubated in 34 o C, 5% CO 2 and infected culture uid (ICF) was harvested after day 7 and 14 of incubation period. Viral RNA was extracted from serum and ICF using QIAamp MinElute Virus Spin Kit (Qiagen, Hilden, Germany) based on the manufacturer's protocol. DENV detection and serotyping were performed using a multiplex real-time RT-PCR method [30]. RT-PCR ampli cation of the DENV E gene followed by sequencing was also done as additional diagnostic evidence.

Mosquito collection
Surveillance of Aedes aegypti mosquitoes was purposely carried out in households of dengue suspected cases from August 26 to September 30, 2015. Households were categorized using the following conditions: (category 1) households of patients who tested positive for DENV NS1 antigen using the PanBio® Dengue Early Rapid kit (Alere Medical Co. Ltd., Massachusetts, USA) at Tarlac Provincial Hospital at the time of the mosquito collection period, (category 2) households proximal (< 150 m) to households of patients from category 1, and (category 3) households of suspected dengue patients reported by barangay health workers 15 days prior to the commencement of the mosquito collection period. For category 3, selection of barangays was based on the previous dengue epidemiology record provided by the city's health o ce. Identi ed barangays were San Isidro, San Miguel, San Sebastian, Maliwalo, Dalayap, San Rafael, San Nicolas, Ligtasan, San Vicente, Binauganan and Matatalaib. Based on previous years, these barangays had high number of reported cases in the city. All households were provided with informed consent for their voluntary participation in the mosquito surveillance. In the case of category 1, once participants agreed, mosquitoes were immediately collected within 24-48 hours after positive DENV NS1 antigen detection. Direct contact with the head of the household for house visitation and mosquito collection was done.
Commercially available mosquito UV-light traps (Mosquito Trap®, Jocanima Corporation, Metro Manila, Philippines) were used to collect mosquitoes daily from early afternoon to early morning (14:00-07:00), and the traps were installed either inside or outside the premises of the surveyed households. Inspection of installed mosquito traps and gathering of trapped mosquitoes were done daily during the morning (07:00-11:00). Sampled mosquitoes were sorted, labeled, identi ed and separated as male and female based on pictorial keys [30]. Identi ed Aedes aegypti mosquitoes were individually kept in a 1.5 ml tube containing 1.0 mL of RNAlater® (Ambion®, Invitrogen, California, USA), and stored at -20 °C until processed.

DENV detection in mosquitoes
Individual female mosquitoes were manually homogenized with a sterile plastic pestle in 200 µl of 1X phosphate-buffered saline (Takara Bio Inc., Shiga, Japan) in a 1.5 mL microcentrifuge tube. Total RNA was subsequently extracted from the homogenate using ISOGEN (Nippon Gene Co., Ltd., Toyama, Japan), following the manufacturer's protocol. The crude RNA was then treated with DNase using the TURBO DNAfree Kit (Ambion®, Thermo-Scienti c Massachusetts, USA). DNAse-treated RNA was eluted in 30 µl nuclease-free molecular biology reagent water (Sigma-Aldrich Co., Missouri, USA) and stored at -80 °C pending analysis. Total RNA quantity and quality were veri ed for each sample with NanoDrop measurement (Thermo-Scienti c, Massachusetts, USA).
A one-step multiplex real-time RT-PCR method [31] was adapted as the method for detection of DENV in individual Aedes aegypti mosquitoes. The assay was performed using the Bio-Rad CFX96 Touch™ Deep Well Real-Time PCR Detection System (Bio-Rad, California, USA). Primer and probe sequences for DENV-2 were modi ed, and the DENV-3 probe was labeled with Cy5.5 and BHQ2, instead of Texas Red and BHQ2 (Table S1). All assays were performed using the iTaq Universal Probes One-step Kit (Bio-Rad, California, USA) and carried out in 25 µl-reaction mixtures containing 5 µl total RNA, 1X reaction mix, 200 nM each of DEN-1, DEN-2, DEN-3 and DEN-4 primers, and 180 nM of each probe.
Cycling conditions for all primer sets were 50 °C for 30 min, and 95 °C for 2 min, followed by 45 cycles of 95 °C for 15 sec and 60 °C for 1 min.
Negative template controls consisted of water as template. A sample was de ned as positive if the average threshold cycle (Ct) value of the sample replicates was above cycle 15 and below cycle 37.

DENV nucleotide sequencing
The DENV E gene of both mosquito (partial sequence) and patient (full-length) samples were ampli ed using the primers described in Table  S2. Brie y, reverse transcription of the total RNA using random primers was carried out using Superscript® III First-Strand Synthesis Supermix (Invitrogen, California, USA), and subsequent PCR ampli cation of the DENV E gene using the resulting cDNA as template was carried out using Phusion® High-Fidelity DNA Polymerase (New England Biolabs, Massachusetts, USA). RT-PCR and gene-speci c PCR were performed using the Bio-Rad T100 Thermal Cycler (Bio-Rad, California, USA).
Amplicons were puri ed using the QIAquick PCR Puri cation Kit (Qiagen, Hilden, Germany), according to instructions from the manufacturer.
Puri ed PCR products of mosquito samples were sent to Euro ns Genomics, Tokyo, Japan for Sanger sequencing. For the patient samples, cycle sequencing was performed using a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied BioSystems, Foster City, CA, USA) in the TaKaRa PCR Thermal Cycler Dice. Sequencing reactions were puri ed using a BigDye XTerminator Puri cation Kit (Applied BioSystems) followed by loading into Genetic DNA Analyzers 310, 3130, or 3730xl (Applied BioSystems). Bidirectional sequencing was performed using primers listed in Table S3 to resolve the full-length DENV E gene.

DENV infection rate in mosquitoes
The number of DENV positive mosquitoes per 1000 mosquitoes was determined from the DENV partial E gene PCR and sequencing results. The infection rate was calculated as the number of DENV-positive female mosquitoes divided by the total number of female mosquitoes analyzed in the study area multiplied by 1000.

Phylogenetic analyses
Mosquito-derived and patient-derived partial E gene sequences together with DENV reference sequences (Table S4) were aligned using ClustalW 2.1 [32] and were manually edited using Mesquite 3.3 [33]. Nucleotide sequences of the DENV isolates were submitted to the GenBank database under accession numbers MK268743-MK268752 (mosquito-derived sequences) and LC553202-LC553256 (patient-derived sequences). Phylogenetic analyses of DENV-1, DENV-2 and DENV-4 isolates were conducted using the Maximum Likelihood (ML) method.
The best-t substitution model was determined using the jModeltest [34] by Bayesian Information Criterion. ML trees were inferred using the TN93 + G parameters for DENV-1 (300 bp) and DENV-2 (258 bp), and GTR + I parameter for DENV-4 (486 bp). ML trees were constructed using PhyML 3.1 [35] and the reliability of the analyses was calculated using 1000 bootstrap replications. No outgroups were used, and DENV genotypes were grouped accordingly [36]. Trees were visualized and edited in FigTree 1.4.4 [37] and Inkscape (http://www.inkscape.org).

Mosquito collection and DENV detection
In this study, 421 patients were screened for DENV NS1 antigen at the Tarlac Provincial Hospital from August to October 2015. Of the 421 patients, 187 patients tested positive for the presence of the DENV NS1 antigen. Amongst the 187 patients, 32 patients, who were residents of Tarlac City, were detected in September. During the same period, mosquito surveillance was conducted around dengue suspected cases. In brief, Aedes aegypti mosquitoes were collected at 48 households where dengue-infected mosquitoes were suspected to be present (Fig. 1). Of the 48 households, 12 households were grouped under category 1, 15 households were grouped under category 2, and 21 households were grouped under category 3.
A total of 583 adult Aedes aegypti mosquitoes were collected; of which, 383 (65.7%) were females whereas 200 (34.3%) were males. Out of the 383 female mosquitoes, 359 were processed for DENV detection due to low RNA quality and quantity of some extracted mosquito samples. Of the 359 female mosquitoes tested, 14 mosquito samples (3.9%) showed positive multiplex real-time RT-PCR results. When RT-PCR ampli cation and sequencing of the DENV partial E gene were performed, 10 mosquito samples (2.8%) were validated to be DENVpositive (Table 1). There was no clear difference in the Ct values or total RNA concentrations between the 10 validated mosquitoes and the other 4 mosquitoes that were not validated by RT-PCR ampli cation and sequencing.
The DENV infection rate during the one-month mosquito surveillance was calculated to be 27.9 DENV-infected mosquitoes per 1000 female Aedes aegypti. Six out of the 10 DENV-positive mosquitoes harbored DENV-2, 3 mosquitoes harbored DENV-4 and 1 mosquito harbored DENV-1 ( Table 2). No DENV-3 was detected in the analyzed mosquito samples as there were only two DENV-3 cases detected in patients during the study period (data not shown). Notably, 4 DENV-infected mosquitoes were collected from 3 households of patients tested positive for DENV NS1 antigen, and the DENV serotypes detected from all 4 mosquitoes did not coincide with the serotypes of the patients residing in the same household where the mosquitoes were collected (Table 1). Table 1 Mosquito samples validated to be DENV-infected using RT-PCR ampli cation and sequencing of the DENV partial E gene.  Phylogenetic relationships among DENVs isolated from mosquitoes and patients Phylogenetic analysis revealed 3 serotypes and genotypes co-circulating in the local mosquito population during the study period, namely: DENV-1 genotype IV, DENV-2 Cosmopolitan genotype and DENV-4 genotype II (Fig. 2). Notably, the same serotypes and genotypes were present in the analyzed patient serum samples.
All DENV-1 patient samples (n = 35) and mosquito sample (n = 1) belong to Genotype IV ( Fig. 2A). Phylogenetic analysis revealed no geographical segregation of DENV sequences within the survey sites in Tarlac City, with some sequences from different barangays or municipalities outside Tarlac City clustering together in the tree (Fig. 2) and have especially high nucleotide identities (up to 100%). Furthermore, reference strains from East Asia (China, Taiwan and Japan) and Southeast Asia (Indonesia and Singapore) were closely related to the DENV detected in mosquitoes and patients from Tarlac City.

Discussion
We have presented in this paper the utility of an individual-based DENV surveillance scheme in inferring serotype and genotype distributions in local mosquito populations and the signi cance of utilizing virus sequences from both vector and host in generating DENV phylogenetic trees.
In this study, we have shown that an individual-based virus detection approach using a one-step multiplex real-time RT-PCR assay can directly detect and serotype DENV in the RNA extracts of eld-collected mosquitoes. This method is capable of detecting all 4 DENV serotypes in a single mosquito in one run. Our results from the RT-PCR ampli cation and sequencing of the DENV partial E gene revealed an infection rate of 27.9 DENV-infected females per 1000 female Aedes aegypti mosquitoes in Tarlac City during the one-month DENV surveillance. This result was comparable to the calculated (female) infection rates in previous studies where pooled mosquitoes were analyzed and mosquito collections were done in longer periods (Table S5). The relatively high infection rate we obtained may be attributed to the purposive surveillance of mosquitoes we implemented around dengue suspected cases, and also to the individual-based DENV detection approach we employed in this study. Individual-based virus detection in eld-collected mosquitoes is seldom performed in mosquito surveillance studies due primarily to nancial costs. However, in this study, we opted to employ an individual-based virus detection approach in order for us to more accurately estimate the infection rate in the study area, and to assess the circulating genotypes and serotypes among mosquitoes and patients at the individual level.
Our ndings showed concurrent co-circulation of similar serotypes and genotypes in mosquitoes and patients, which is similar to the results of a previous study that detected the DENV in both Aedes albopictus mosquitoes and viraemic patient in Catalonia, Spain [38]. Sequencing and phylogenetic analyses showed that the detected DENVs grouped primarily into three genotypes, namely: (1) DENV-1 genotype IV, (2) DENV-2 Cosmopolitan genotype, and (3) DENV-4 genotype II, suggesting the hyperendemicity of dengue in Tarlac City, Philippines. These results are also consistent with the reported multiple genotypes currently co-circulating in the Philippines [39]. No DENV-3 was detected in mosquitoes at the time of the study period and this may be attributed to the low number of DENV-3 infected individuals in the study area in 2015 (data not shown). Our results suggest that there is continuous circulation of the same DENV genotypes in the Philippines over the last years since their rst reported isolation implying that DENV genotype distributions remained unchanged. Since 1956, dengue has been considered a noti able disease [40] in the Philippines, and a national program directed towards community-based prevention and control has been implemented nationwide in 1998 to combat dengue [41]. While there has been a notable increase in the reported incidence of dengue through the years, the amount of published dengue research in the country is still limited [42]. Moreover, to date, no report has been published about the circulating DENV serotypes and genotypes in local mosquito populations in the Philippines. To our knowledge, our study is the rst report of a mosquito-based virus surveillance around dengue suspected cases in the Philippines. Our results underline the need for enhanced DENV surveillance to monitor the DENV transmission dynamics in the Philippines.
The major hallmark of the phylogenetic analyses we conducted in this study was the overall clustering of DENVs from various geographical locations (i.e. barangay, municipality) (Fig. 2). Notably, genetically related DENVs detected from mosquitoes and patients were not geographically segregated. Rather, these DENVs were randomly distributed within and outside the vicinities of Tarlac City during the same period suggesting that there is active movement of DENV in the area that may involve the transfer of both infected humans and mosquitoes. This hypothesis was further supported by the mismatch in the detected DENV serotypes between mosquitoes and patients residing in the same household (Table 1). A study conducted in Brazil harbored the same result [12] as our study indicating that most infections are obtained at other houses or public spaces such as schools or workplaces [43,44,45]. However, we also consider that asymptomatic infections may have signi cant implications to our results. Individuals residing in the same household with dengue patients may be asymptomatic and may have harbored the same serotype to that of the mosquito collected in the same household. In order to detect asymptomatic dengue infections, people in the same household of dengue cases must also be tested for DENV [46], highlighting the need for a more detailed surveillance and contact tracing of dengue index cases [47].
Tarlac City is the primary urban center of business and trade within the Tarlac province, with high concentration of commercial activities, the city likely attracts more people on a regular basis, creating hubs for disease transmission and spatial spread [48]. However, since only partial DENV E gene sequences were analyzed and only a subset of mosquitoes and patients in Tarlac City were used in this study, further investigation is necessary to con rm our hypothesis. A wider geographic analysis of the phylogeny of DENV in both vector and host is needed as this has considerable epidemiological signi cance.
We were able to show how mosquito-based virus surveillance around dengue suspected cases can be useful in the evaluation of local virus activity patterns in a de ned time period and area. However, our study's small sample size is not enough to permit de nitive conclusions regarding our hypothesized active transmission of DENV in Tarlac City. Also, the mosquito surveillance we conducted proved to be challenging due to limited access to patients' residences. For these reasons, our current ndings suggest areas for further study and future application of mosquito surveillance around dengue suspected cases conducted with larger sample size and longer time scales to fully establish the impact of mosquito-based virus surveillance data in the prevention of human dengue cases. We were unable to obtain complete DENV E gene sequences from all mosquito samples due to variations of their ampli cation e ciency. It should also be noted that we analyzed both blood-fed and unfed mosquitoes. Detection of virus from a blood-fed female mosquito may not represent an actual occurrence of infection, but may only indicate that the mosquito had ingested viremic blood [49].

Conclusions
In conclusion, we identi ed the DENV genotypes and serotypes concurrently co-circulating in mosquitoes and patients and revealed that there was high infection rate of DENV in the local Aedes aegypti population during the 2015 dengue peak season in Tarlac City, Philippines. Taken together, our results highlight the importance of DENV surveillance in eld-collected mosquitoes, especially in evaluating local virus activity patterns in a de ned time period and area. Integrating mosquito and patient viral data for phylogenetic analysis could potentially improve our understanding of how DENV moves in a human-to-mosquito-to-human cycle in a speci c locality. Further studies utilizing data from naturally infected mosquitoes may provide additional information in understanding DENV transmission dynamics. In particular, DENV transmission and the evolution of DENV RNA populations may be better described by considering not only viruses from severe cases (hospitalized), but also from mild cases (outpatients), asymptomatic infections, as well as viruses from mosquitoes.

Declarations
Ethics approval and consent to participate All participants were informed of the aims of the study and procedures involved in study participation at enrolment, and written informed consent was received before sample collection. This study was approved by the ethics committee of Tohoku University Graduate School of Medicine (2020-1-098) and the institutional review board of the Research Institute for Tropical Medicine of the Philippines (2013-017).

Consent for publication
Not applicable Availability of data and material All data generated or analyzed during this study are included in this published article and its supplementary les. All generated sequences are available in GenBank with accession numbers: MK268743-MK268752 (mosquito-derived sequences) and LC553202-LC553256 (patientderived sequences).

Figure 2
Phylogenetic trees of the partial E gene of DENV-1 (A), DENV-2 (B) and DENV-4 (C). The trees were inferred with the maximum likelihood criterion. Node support was evaluated with 1000 bootstrap replicates. Bootstrap values more than 70% were shown on the branches. DENV mosquito (black star) and patient sequences from Tarlac City (black circle) and other neighboring municipalities (grey circle) collected from August 1 to October 31, 2015 were included in the tree. Reference sequences are labeled by their NCBI accession numbers, two-letter ISO country code, and corresponding year of isolation. Scale bar indicates nucleotide substitutions per site.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.