Development and Validation of Reverse Transcription Loop-Mediated Isothermal Amplification (RT-LAMP) for Rapid Detection of ZIKV in Mosquito Samples from Brazil

The rapid spread of Zika virus (ZIKV) represents a global public health problem, especially in areas that harbor several mosquito species responsible for virus transmission, such as Brazil. In these areas, improvement in mosquito control needs to be a top priority, but mosquito viral surveillance occurs inefficiently in ZIKV-endemic countries. Quantitative reverse transcription PCR (qRT-PCR) is the gold standard for molecular diagnostic of ZIKV in both human and mosquito samples. However, the technique presents high cost and limitations for Point-of-care (POC) diagnostics, which hampers its application for a large number of samples in entomological surveillance programs. Here, we developed and validated a one-step reverse transcription LAMP (RT-LAMP) platform for detection of ZIKV in mosquito samples. The RT-LAMP assay was highly specific for ZIKV and up to 10,000 times more sensitive than qRT-PCR. Assay validation was performed using 60 samples from Aedes aegypti and Culex quinquefasciatus mosquitoes collected in Pernambuco State, Brazil, which is at the epicenter of the Zika epidemic. The RT-LAMP had a sensitivity of 100%, specificity of 91.18%, and overall accuracy of 95.24%. Thus, our POC diagnostics is a powerful and inexpensive tool to monitor ZIKV in mosquito populations and will allow developing countries to establish better control strategies for this devastating pathogen.


Results
Detection of ZIKV in Aedes aegypti under controlled conditions. First, we determined the ability of RT-LAMP to detect ZIKV in A. aegypti under controlled conditions. To this end, crude lysate of uninfected mosquitoes were spiked to result in either a high (1 × 10 6 PFU/mL) or low viral load (1 × 10 3 PFU/mL) in order to mimic physiological concentrations of ZIKV in these vectors. Spiked samples were processed for RT-LAMP without RNA isolation. RT-LAMP assay for ZIKV were positive in both viral loads tested. As expected, non template control (NTC) samples (water) and negative control (crude lysate of uninfected A. aegypti) tested negative ( Fig. 1A-C). RNA extraction did not improve RT-LAMP detection (data not shown). RT-LAMP results were confirmed by qRT-PCR, through which the Ct value was 12.1 and 26.8, for high viral and low viral load, respectively (Figs 1 and S4). The same results were obtained with viral spike in C. quinquefasciatus homogenates (data not shown).
In order to mimic a real world scenario of ZIKV surveillance in mosquitoes, we determined the capacity of the RT-LAMP to detect ZIKV in A. aegypti mosquitoes experimentally infected by oral feeding on rabbit blood spiked with ZIKV. In this study, mosquitoes fed on unspiked rabbit blood were also included as controls. Crude mosquito lysates were used for RT-LAMP assay without RNA isolation. After incubation, the RT-LAMP was able to detect ZIKV only in infected mosquitoes, but not controls (Fig. 1D-F), suggesting the test may be useful for ZIKV detection in entomological samples. RNA extraction did not improve RT-LAMP detection (data not shown).
www.nature.com/scientificreports www.nature.com/scientificreports/ Analytical sensitivity of Rt-LAMp for detection of ZIKV. First, we sought to optimize the RT-LAMP assay conditions, reactions were performed at temperatures ranging from 59 °C to 75 °C following an incubation that ranged from 10 min to 60 min. The best amplification results were obtained at 72 °C for 40 min, but incubation time as short as 20 minutes was sufficient for detecting positive samples. Therefore, all assays were carried out using 40-min incubation time. The analytical sensitivity (limit of detection) of RT-LAMP was determined in crude lysate of A. aegypti spiked with a 10-fold serial dilution of ZIKV ranging from 10 5 PFU to 10 −7 PFU without RNA isolation. RT-LAMP was able to detect a broad range of virus concentration (from 10 5 to 10 −5 PFU), including viral loads found in naturally infected mosquitoes 39 . Considering 10 independent replicates per protocol developed, the probit regression analysis revealed that the limit of detection at 95% probability for each RT-LAMP was −2,98 log 10 PFU of ZIKV (~1/1000 PFU) with confidence interval from −3,62 to −1,64 (Table 2 and Fig. S6). Additionally, viral RNA extracted from the same dilutions tested by RT-LAMP was assayed by the widely used ZIKV qRT-PCR method developed by Lanciotti 40 . For qRT-PCR assay, the lower detection limit was   42 . Of these, 31 samples were ZIKV negative as determined by qRT-PCR and 29 were positive, including naturally and experimentally infected mosquitoes ( Table 3). The Ct value in these samples ranged from 27.0 to >40.0. From the total of 60 samples, the RT-LAMP assay was able to detect ZIKV in 32 samples,  www.nature.com/scientificreports www.nature.com/scientificreports/ including the 29 samples already determined to be positive by qRT-PCR (Fig. 4). Moreover, samples that were at the detection threshold by qRT-PCR (Ct values ranging from 37.5 to 40.3) were tested positive by the RT-LAMP assay result (Fig. 5), highlighting the sensitivity of the test in mosquito samples.
The diagnostic performance of ZIKV RT-LAMP for mosquito samples was determined by statistical analysis using qRT-PCR as the gold standard technique. The overall ZIKV prevalence in the samples was 46.03% (95% CI 33.39% to 59.06%). The RT-LAMP assay had a diagnostic sensitivity of 100% (95% CI 88.06% to 100.00%) and diagnostic specificity of 91.18% (95% CI 76.32% to 98.14%). The positive predictive value, which is probability that the virus is present when the test is positive, was 90.62% (95% CI 76.64% to 96.61%), whereas the negative predictive value, which indicates the probability that the virus is absent when the test is negative, was 100%. The overall accuracy of the RT-LAMP test was determined to 95.24% (95% CI 86.71% to 99.01%) ( Table 4), highlighting the practical value of RT-LAMP for ZIKV detection in entomological samples.
To confirm the identity of ZIKV RT-LAMP positive samples, we sequenced positive samples from field-caught Aedes spp. and Culex spp. mosquitoes by the Sanger method. Sequencing results and BLAST analysis demonstrated that ZIKV RT-LAMP amplicons match 100% with virus circulating in Brazil (Fig. 6), confirming the specificity of the RT-LAMP for ZIKV.
Together, these results indicated that our ZIKV RT-LAMP assay represents a robust and affordable diagnostic platform that can be used as a surveillance tool for mosquitoes infected with ZIKV.

Discussion
The rapid detection of ZIKV in mosquito samples can help to understand the dynamics of the disease in areas that have favorable conditions for virus transmission 20 . In this context, we developed a rapid molecular test for the detection of ZIKV in mosquito samples that may be a valuable tool for vector surveillance. The RT-LAMP assay described here is straightforward, inexpensive, and enables ZIKV detection even in the absence of RNA extraction. To our knowledge, this is the first validation of a ZIKV RT-LAMP assay using experimentally and naturally infected A. aegypti and C. quinquefasciatus mosquitoes collected at the epicenter of the Zika epidemic in Brazil.
Currently, the gold standard technique for detection of ZIKV in mosquito samples is qRT-PCR. This assay is specific for detecting the virus in both human and mosquito samples 21,40 . However, its prohibitive cost makes qRT-PCR unfit for testing a large number of mosquitoes collected in entomological surveillance programs 41 . Another potential limitation of qRT-PCR is the inability to detect low viral titers, which may occur especially during interepidemic periods. The limit of detection for the assay described by Faye was 0.05 plaque forming unit (PFU) or 32 genome-equivalents and the one developed by Lanciotti was 25 RNA copies 21,40 . Recently, other research groups have developed methodologies using the LAMP approach for the detection of ZIKV using mosquito samples [34][35][36] . However, these studies used only a handful of mosquito samples and the lowest virus concentration detected was 10 3 PFU. Our RT-LAMP was evaluated using 60 and revealed to be about 10,000 fold more sensitive than the qRT-PCR, detecting virus concentrations as low as 10 −5 PFU. The large amount of infectious and non-infectious ZIKV RNA released into the culture supernatant explains the ability of RT-LAMP to detect     www.nature.com/scientificreports www.nature.com/scientificreports/ less than 1 PFU even without RNA extraction 43 . The analytical sensitivity of both our qRT-PCR and RT-LAMP differed from previously published studies which developed the primers 30,40 . There are a number of reasons that might have accounted for this variation, including differences in kits and research suppliers, viral strains, type of biological samples, and detection systems.
Several mosquito-borne arboviruses, including ZIKV, DENV and CHIKV, are endemic and co-circulate throughout the Northeast Brazil 44,45 . One possible limitation of diagnostic tests for ZIKV is the possibility of cross-reactivity with other flaviviruses, particularly DENV 40,46,47 . Here, we showed no cross-reactions with other arboviruses including four serotypes of DENV, YFV or CHIKV and sequencing of RT-LAMP amplicons from naturally infected A. aegypti and C. quinquefasciatus confirmed ZIKV identity.
We validated the RT-LAMP assay using samples obtained from experimentally and naturally ZIKV-infected A. aegypti and C. quinquefasciatus. The RT-LAMP had a sensitivity of 100%, specificity of 91.18%, and overall accuracy of 95.24% as compared to qRT-PCR. Importantly, the ZIKV RT-LAMP could undoubtedly detect ZIKV RNA in mosquito samples that had been previously tested as negative by qRT-PCR. These samples were at the detection threshold by the qRT-PCR with Ct value ranging from 38.6 to 40.3. In contrast with our findings, some studies have reported that the analytical sensitivity of the RT-LAMP assay is lower when compared to the gold standard diagnostic test (qRT-PCR) 32,38 . However, recently published studies have corroborated our findings that the analytical sensitivity of the RT-LAMP assay is superior than qRT-PCR 36,48 .
The RT-LAMP assay can be performed through either a two-step assay or one-step protocol. Two-step RT-LAMP requires the addition of the reverse transcriptase (RT) enzyme together with the DNA polymerase enzyme, which may be wild-type Bst DNA polymerase or Bst 2.0 polymerase 2.0 WarmStart. Several studies report the need for RNA extraction before performing the RT-LAMP assay and the use of the two-step RT-LAMP [49][50][51] . However, the two step protocol is longer, more expensive, and requires additional sample  Table 4. Diagnostic performance of ZIKV RT-LAMP for mosquito samples. www.nature.com/scientificreports www.nature.com/scientificreports/ handling, which increases the chances of pipetting errors and contamination. The use of Bst 3.0 Polymerase 3.0 WarmStart overcomes these concerns. This enzyme possesses high activity of reverse transcriptase and polymerase in a single-temperature incubation which allows the assay to be performed in a one-step. Additionally, the Bst 3.0 DNA polymerase is a robust enzyme capable of maintaining its activities even in the presence of inhibitors 33 . This is especially relevant for viral survey in entomological samples which are notorious to harbor amplification inhibitors 52 .
Recently, Yaren et al. reported a diagnostic test based on RT-LAMP for detection of ZIKV in mosquito samples 35 . Nonetheless, the need for RNA extraction limits its applications for POC diagnostics. In another study, Lamb et al. reported a low-cost molecular diagnostic test method based on RT-LAMP for detection of ZIKV in mosquito samples without RNA isolation 36 . However, the authors tested only five experimentally infected A. aegypti and did not validate the technique using naturally infected mosquitoes.
Other groups have also developed several technologies for molecular detection of ZIKV [30][31][32][33]35,50,51,[53][54][55][56][57][58] . However, many of these technologies still have limitations for POC diagnostic applications, including the need for RNA isolation or the use of sophisticated and proprietary hardware and software, which limits its applicability in the developing world.
The main advantages of the RT-LAMP assay described here is the ability to detect ZIKV without the need for pretreatment or RNA extraction from the mosquito samples. Importantly, positive samples can be diagnosed in just 20 minutes and the result can be easily interpreted visual examination. Given its simplicity, the assay can be run by individuals without specialty training. The cost per sample was inferior to $1, which is considerably lower than qRT-PCR. These advantages suggest that our diagnostic assay to detect ZIKV is suitable for use in viral surveillance in mosquitoes in remote areas or low resource countries affected by the ZIKV epidemics or at risk of viral introduction.

Conclusion
We have developed a low cost, point-of-care diagnostic platform based on the RT-LAMP assay to detect ZIKV in mosquito samples collected at the epicenter of the Zika epidemics in Brazil. The test is a robust, fast and inexpensive tool for surveillance of ZIKV in mosquito populations and will enable developing countries to establish better viral surveillance in vectors and improve the efficacy of control programs. Our results provide a potential new molecular diagnostic test for ZIKV in mosquito samples as a novel straightforward and inexpensive method for detection of ZIKV in arthropod vectors.  30 . In order to visualize positive reactions and prevent contamination, 1 μL of SYBR Green I (ThermoFisher Scientific) diluted 1:10 dilution in RNase-free water (Promega) was added to the center of the tube caps before the reaction and mixing afterwards. Reactions were incubated at 72 °C for 40 min in a heat block, and then inactivated at 80 °C for 5 minutes. To evaluated the robustness of the assay for POC applications, all set-up and execution of RT-LAMP reactions were done in a conventional lab bench using designated pipettes and filter tips. Imaging analysis took place in separate rooms. All experiments were independently replicated at least six times.

Cells and viruses. Vero cells were grown in
After the incubation, the RT-LAMP products reactions were detected using three different methods. In the first, the products were observed by naked eye under natural light and photographed using a conventional smartphone camera. A color change from orange to greenish yellow was used to identify positive sample, while a negative sample remained orange. The second method was visual analysis of reaction tubes under UV light irradiation (UV wavelength of 302-312 nm) using a transilluminator (model UVB LTB 20 × 20 STV, Loccus Biotecnologia, São Paulo, Brazil) coupled with a camera and connected to a computer. In this method, negative samples were dark blue and positive reactions were light fluorescent. In the third method, the RT-LAMP amplicons were analyzed by agarose gel electrophoresis (2.0%) in 1x TAE buffer, followed by ethidium bromide staining and gel visualization using transilluminator. For electrophoresis analysis, 1 kb Plus DNA Ladder (ThermoFisher Scientific) was used as a DNA size marker.

Detection of ZIKV in Mosquito samples Under Controlled Conditions.
To evaluate the ability of RT-LAMP to detect ZIKV in mosquitoes, pools of A. aegypti or C. quinquefasciatus mosquitoes (n = 10) were homogenized in 300 μL of RNA-free water. Crude lysates were then spiked with 100 µL of ZIKV so the final viral concentration in the lysates was either 10 6 or 10 3 PFU/mL, thus simulating a situation of high and low viral load, respectively. After incubation at 37 °C for 1 hour, samples were directly assay by RT-LAMP without RNA extraction.
In order to assess ZIKV detection by RT-LAMP in infected mosquitoes, we used samples from experimentally infected female A. aegypti mosquitoes. In brief, the Rec-Lab colony was maintained under standard conditions (temperature, 26 °C ± 1 °C, relative humidity of 60 to 80% and photoperiod 12:12 h C/E) at the Entomology Laboratory of the Institute Aggeu Magalhães (IAM). For artificial feeding, cell supernatant containing 10 6 PFU of ZIKV were mixed in 1:1 defibrinated rabbit blood and provided to starving mosquitoes for for 90 minutes as previously described 39 . Whole female mosquitoes were collected at 18 days post-infection, homogenized in 300 μL of RNA-free and processed for RT-LAMP. Mosquitoes independently fed on non-infected culture cells mixed to the defibrinated rabbit blood was used as controls. To evaluate the analytical sensitivity (limit of detection) of the RT-LAMP assay, ZIKV strain PE243 was 10-fold serially diluted in crude lysates of uninfected A. aegypti mosquito. Virus concentration in spiked mosquito samples ranged from 10 5 PFU to 10 −7 PFU. After dilution, samples were directly assayed by RT-LAMP without RNA isolation. To compare the results of RT-LAMP with a gold standard technique, viral RNA was extracted from the same dilutions using Trizol reagent (Invitrogen Carlsbad, USA) according the manufacturer's instructions and then assayed by the widely used ZIKV qRT-PCR method 40 .

Validation of Rt-LAMp for ZIKV detection in Mosquito samples.
To validate the performance of the RT-LAMP for the diagnosis of ZIKV relative to qRT-PCR, 60 samples from A. aegypti (n = 32) and C. quinquefasciatus (n = 28) previously assayed by qRT-PCR 11,41 were obtained from the Entomology Department and tested by RT-LAMP. The intrinsic diagnostic utility of the test was determined using several statistical parameters described below.
sequencing of LAMp fragments. The genetic characterization of the LAMP fragments from two field positives samples from A. aegypti and C. quinquefasciatus was performed by the Sanger sequencing method. Amplicons from RT-LAMP reaction were directly purified using illustra GFX PCR DNA and Gel Band Purification Kit (GE) according to the manufacturer´s instructions and eluted in 30 μL of water. Purified amplicons were directly sequenced using the primer FIP and the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems,USA) as established by the manufacturer and run on an ABI Prism 3100 Capillary Automatic DNA Analyzer. Sequences of fragments were analyzed using the Bioedit software, v7.0.5 and submitted to NCBI BLAST database (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi) to identify the most closely ZIKV strain. statistical analysis. Graphs were generated using the GraphPad Prism Software version 5.01 for Windows (GraphPad Software, La Jolla, California, USA). A probit regression was performed to calculate the limit of detection of the RT-LAMP for detection of ZIKV using MedCalc software (version 18.11, MedCalc Software, Ostend, Belgium). The estimation of the several diagnostic parameters (sensitivity, specificity, ZIKV prevalence, positive predictive value, negative predictive value and overall accuracy) of the RT-LAMP for detection of ZIKV was calculated using the web-based software MedCalc's Diagnostic Test Evaluation Calculator (https://www.medcalc. org/calc/diagnostic_test.php). This analysis was based on the results from 60 mosquito samples previously diagnosed by qRT-PCR.