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 . 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 defined by the 2000 Census of Population and Housing . Maps were created using the QGIS 3.6 software and edited in Inkscape (http://www.inkscape.org) , with some figures created with BioRender (http://biorender.com). Data for creating the map 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 confirmed 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 34oC, 5% CO2 and infected culture fluid (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 . RT-PCR amplification of the DENV E gene followed by sequencing was also done as additional diagnostic evidence.
Surveillance of Ae. 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 office. Identified 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 as previously described [30,31]. The trap emits UV light, and generates heat and CO2 gas via a photocatalytic reaction on the TiO2 coated funnel. Decoyed mosquitoes enter the trap through the capture windows and then strongly drawn into the capture net by a strong current produced by the ventilator. The mosquito UV-light traps collected mosquitoes daily from early afternoon to early morning (14:00 – 07:00), and were installed either inside or outside the premises of the surveyed households. For each household, one mosquito trap was installed. Inspection of installed mosquito traps and gathering of trapped mosquitoes were performed daily each morning (07:00-11:00). Sampled mosquitoes were sorted, labeled, identified and separated as male and female based on pictorial keys . Identified Ae. 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 DNA-free Kit (Ambion®, Thermo-Scientific 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 verified for each sample with NanoDrop measurement (Thermo-Scientific, Massachusetts, USA).
A one-step multiplex real-time RT-PCR method  was adapted as the method for detection of DENV in individual Ae. 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 modified (Table S1) in this protocol from the original method , with a few nucleotide bases either revised or deleted based on the consensus sequence of currently major circulating DENV-2 strains. 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. The one-step multiplex real-time RT-PCR assay was performed once in duplicates. 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 defined 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 amplified using the primers described in Table S2. Briefly, 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 amplification 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-specific PCR were performed using the Bio-Rad T100 Thermal Cycler (Bio-Rad, California, USA).
Amplicons were purified using the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany), according to instructions from the manufacturer. Purified PCR products of mosquito samples were sent to Eurofins 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 purified using a BigDye XTerminator Purification 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.
Mosquito-derived and patient-derived partial E gene sequences together with DENV reference sequences (Table S4) were aligned using ClustalW 2.1  and were manually edited using Mesquite 3.3 . 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-fit substitution model was determined using the jModeltest  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  and the reliability of the analyses was calculated using 1000 bootstrap replications. No outgroups were used, and DENV isolates grouped accordingly to genotypes as previously described . Trees were visualized and edited in FigTree 1.4.4  and Inkscape (http://www.inkscape.org).