Rapid and high throughput molecular identification of diverse mosquito species by high resolution melting analysis

Mosquitoes are a diverse group of invertebrates, with members that are among the most important vectors of diseases. The correct identification of mosquitoes is paramount to the control of the diseases that they transmit. However, morphological techniques depend on the quality of the specimen and often unavailable taxonomic expertise, which may still not be able to distinguish mosquitoes among species complexes (sibling and cryptic species). High resolution melting (HRM) analyses, a closed-tube, post-polymerase chain reaction (PCR) method used to identify variations in nucleic acid sequences, has been used to differentiate species within the Anopheles gambiae and Culex pipiens complexes. We validated the use of PCR-HRM analyses to differentiate species within Anopheles and within each of six genera of culicine mosquitoes, comparing primers targeting cytochrome b ( cyt b), NADH dehydrogenase subunit 1 (ND1), intergenic spacer region (IGS) and cytochrome c oxidase subunit 1 ( COI) gene regions. HRM analyses of amplicons from all the six primer pairs successfully differentiated two or more mosquito species within one or more genera ( Aedes ( Ae. vittatus from Ae. metallicus), Culex ( Cx. tenagius from Cx. antennatus, Cx. neavei from Cx. duttoni, cryptic Cx. pipiens species), Anopheles ( An. gambiae s.s. from An. arabiensis) and Mansonia ( Ma. africana from Ma. uniformis)) based on their HRM profiles. However, PCR-HRM could not distinguish between species within Aedeomyia ( Ad. africana and Ad. furfurea), Mimomyia ( Mi. hispida and Mi. splendens) and Coquillettidia ( Cq. aurites, Cq. chrysosoma, Cq. fuscopennata, Cq. metallica, Cq. microannulatus, Cq. pseudoconopas and Cq. versicolor) genera using any of the primers. The IGS and COI barcode region primers gave the best and most definitive separation of mosquito species among anopheline and culicine mosquito genera, respectively, while the other markers may serve to confirm identifications of closely related sub-species. This approach can be employed for rapid identification of mosquitoes.

Mosquitoes are among the most important disease vectors, known to transmit and maintain the circulation of pathogens that cause both global and neglected tropical diseases in humans and animals 1 . The correct identification of different field-collected mosquito species, endemic to distinct ecologies, with high parasite and arthropodborne virus (arbovirus) diversities is crucial to the planning of targeted vector control strategies to mitigate disease transmission 2 . The last and most comprehensive Afrotropical mosquito identification keys were published in 1941 for culicines 3 and in 1987 for anophelines 4 . Molecular approaches that efficiently differentiate conspecific mosquitoes such as the barcode region 5 improve identification accuracy considerably 6 , but are time consuming, expensive in terms of post-polymerase chain reaction (post-PCR) processing and depend heavily on DNA sequencing.
Recent approaches have taken advantage of the unique melting profiles generated by homologous PCR products with small sequence differences during high resolution melting (HRM) analysis 7,8 . Indeed, PCR-HRM has been used to differentiate mosquito transmitted arboviruses [9][10][11] and malaria Plasmodium 12, 13 , vertebrate blood meals of mosquitoes 10 , between two members of the Anopheles gambiae complex 14 and amongst three members of the Culex pipiens complex 15 . HRM analysis has proven to offer higher resolution of PCR product based species identification on sequence variants than electrophoretic methods by revealing even single nucleotide polymorphisms (SNPs) in the simple sequence repeats (SSRs) among products of similar sizes 16,17 . Conventional detection of specific PCR products sequence relies on costly molecular probes and/or product sequencing 18 . For species identification 16 , only representative samples with distinct HRM profiles need to be sequenced, thereby reducing reagent and sample consumption costs [10][11] . Combining HRM analysis of barcode region sequences (Bar-HRM) has been successfully used to rapidly and accurately distinguish between closely related antelope species 19 and medicinal plants 20,21 and to authenticate the source of vegetable oils 22 .
Although HRM has been successfully used to differentiate between specific Anopheles and Culex mosquitoes, the approach's broader applicability and most suitable markers have not been evaluated. Previously, only the ribosomal DNA was targeted for An. gambiae sensu lato (s.l.) 14 and only the acetylcholinesterase gene was used in distinguishing the Cx. pipiens complex 15 . This study aimed at validating the use of HRM analysis for high throughput molecular culicine and anopheline mosquito identification and differentiation, comparing the utility of one ribosomal IGS (previously used to differentiate An. gambiae s.l.) 14 and three mitochondrial (COI, ND1, cyt b) gene markers.

Sample collection and identification
We used 109 mosquitoes (Table 1 and Table 2) that were collected in 2012 during the rainy seasons near Lake Baringo from March 2-4, July 16-24 and October 12-21 and Lake Victoria from April 2-15, May 18-31 and November 12-29 during a mosquito diversity study around the islands and mainland shores of Lake Baringo in GenBank accessions are provided only for samples with confirmed identity and from which the COI DNA sequences were obtained during a previously published mosquito diversity study 6 .  GenBank accessions are provided only for samples with confirmed identity and from which the COI DNA sequences were obtained during a previously published mosquito diversity study 6 .
Baringo County (Table 1) and Lake Victoria in Homa Bay County ( These mosquitoes were morphologically identified during a previous study 6 . Baringo County is a known hotspot for arbovirus outbreaks 23 , while Homa Bay County is endemic to malaria and is located in a region with a history of arbovirus activity 10 . One sample each of Anopheles gambiae sensu stricto (s.s.) and An. arabiensis, Aedes aegypti and Culex pipiens from laboratory colonies maintained in the Insectary of the International Centre of Insect Physiology and Ecology (icipe), Nairobi, Kenya, were used as controls. Also, specimens with confirmed identity that have been previously sequenced and submitted to GenBank (Table 1 and Table 2) were used as both controls and samples.

DNA extraction
From each mosquito, we extracted DNA according to the hot sodium hydroxide and Tris (HotSHOT) DNA extraction protocol 24 from a single mosquito leg that was detached from the rest of the body using a pair of forceps and dissecting pin. Without crushing, the mosquito leg was put in a 0.2 ml microcentrifuge tube containing 30 µl of Alkaline Lysis buffer (25 mM NaOH (Thermo Fisher Scientific, Pittsburgh, USA), 0.2 mM disodium EDTA (Thermo Fisher Scientific), pH 8.0) and incubated in a thermocycler at 95°C for 30 minutes and cooled at 4°C for 5 minutes. Then, 30 µl neutralising solution (40 mM Tris-HCl (Thermo Fisher Scientific)) was added. The resulting DNA was stored at -20°C until required as templates for PCR assays.  Table 3). The COI AnophF primer was initially designed specifically for Anopheles mosquitoes to be used with the HCO2108R primer 26 , but tested on other species as well. Using samples of morphologically and molecularly identified Culex, Aedeomyia, Mimomyia, Coquillettidia, Mansonia, Aedes, and Anopheles mosquito species (Table 1 and Table 2), we amplified different gene regions of their genomes using six pairs of primers (

Results
We differentiated 12 mosquito species in the Aedes (two), Anopheles (two), Culex (six), and Mansonia (two) genera by HRM analyses ( Table 4). The COI sequences of some of the mosquito samples analyzed and differentiated were obtained during a previously published mosquito diversity study 6 and their respective GenBank Accession numbers are listed in Table 1 and Table 2.
Despite the fact that the COI-AnophF/HCO2198R primers were originally designed based on Anopheles mitochondria genome alignments, they were most efficient in differentiating among Mansonia (Ma. africana and Ma. uniformis ( Figure 1A)), Culex (Cx. neavei and Cx. duttoni, Cx. tenagius and Cx. antennatus, and two genetic variants of Cx. pipiens (Figure 2A)), and Aedes (Ae. vittatus and Ae. metallicus (Figure 3)) mosquitoes (Table 4). Indeed, the DNA sequences flanked by the COI-AnophF/HCO2198R primers included multiple polymorphic sites in species within these genera ( Figure 4). Although there are SNPs within species DNA that resulted to the slight changes observed in their HRM profiles, the SNPs across species were enough to distinguish between them.
Mansonia africana and Ma. uniformis could also be differentiated by Mos-COI-JV ( Figure 1B Table 4). However, unlike the COI HRM profiles (Figure 2A, B), the ND1 HRM profiles ( Figure 2D) of Cx. pipiens amplicons showed a melting temperature shift of to the right (higher temperature) compared to the Culex sp. GPA amplicons, possibly due to greater GC  richness of Cx. pipiens at this locus 28 . Similarly, the IGS primers (AgamUni) differentiated Anopheles gambiae s.s. from An. arabiensis ( Figure 5). In addition, the COI-AnophF/HCO2198R primers were also used to separate Cx. neavei from Cx. duttoni (Figure 2A)

Discussion
We compared six pairs of primers for their potential to differentiate at least two morphologically similar mosquito species within each of seven mosquito genera by PCR-HRM analysis and identified suitable markers for differentiating species within Anopheles, Aedes, Culex and Mansonia mosquitoes. However, none of the markers were suitable for HRM analysis to distinguish among species of Aedeomyia, Mimomyia or Coquillettidia genera mosquitoes. Also, Cx. watti, which can be misidentified morphologically as Cx. duttoni or Cx. pipiens, could not be differentiated by PCR-HRM analyses. Nonetheless, we were able to distinguish Ma. africana from Ma. uniformis, An. gambiae s.s. from An. arabiensis (sibling species of An. gambiae s.l.), Ae. vittatus from Ae. metallicus,   as well as Cx. neavei from Cx. duttoni, Cx. tenagius from Cx. antennatus and two cryptic sympatric species of morphologically identical Cx. pipiens. Most notably, the two Cx. pipiens species with distinct COI barcode sequences 6 were indeed first identified by HRM analysis of numerous samples 6 . Thus, the relative economy of HRM analysis compared to sequencing facilitates the rapid identification of cryptic species.
Surprisingly, HRM analysis of PCR products from the COI-AnophF/HCO2198R primers, which were designed for Anopheles, could not distinguish between these sibling species, yet were most effective in discriminating species within the Mansonia, Aedes and Culex genera, including between the cryptic Culex pipiens species. Anopheles gambiae and An. arabiensis were only distinguished using the IGS gene, which was also designed for An. gambiae 2 and is routinely used for distinguishing these sibling species by conventional PCR 29 and HRM analysis 14 . In contrast, species complexes of An. coustani and An. funestus were not   COI-AnophF/HCO2198R primers were most sensitive in discriminating morphologically indistinct species. This highlights the power of the COI barcode region for identifying diverse cryptic species 32 . The SNPs present in the COI genes of the ten mosquito species confirms that the COI gene is conserved and polymorphic enough to differentiate these species even in cases of morphological misidentification. The polymorphisms between species were enough to robustly separate them based on their HRM profiles, while sequence polymorphisms within species were too few to significantly alter their HRM profiles.
We, therefore, recommend the initial use of the COI-AnophF/ HCO2198R primers Bar-HRM to differentiate Mansonia, Culex and Aedes mosquito species and the IGS primers for anopheline mosquito identification 2,14,33 by HRM. The inability of all the six primer pairs to differentiate many mosquito species among all seven genera tested is an indication that the genetic diversity of many mosquito species is complicated and still poorly understood. Also, the number (sample size) of many of the analyzed mosquito species was small (<3) because these species were scarcely present in the study areas. More samples (≥3) should be used and more study areas should be sampled in subsequent studies to test genetic differentiation of mosquito species 34 . Additional polymorphic DNA loci should also be identified, tested and used in combination with existing ones for the identification of mosquito species, especially among species complexes and across genera.

Conclusions
This study shows that specific PCR markers can be used to distinguish closely related species of mosquitoes using HRM analysis. We distinguished two sibling species of An. gambiae s.l., two species each of Mansonia and Aedes, and six species, including cryptic species, of Culex using six pairs of primers targeting the mitochondrial and ribosomal genes. HRM is a low cost (<$1 per reaction), effective tool that enhances culicine and anopheline mosquito identification and may also reveal population differences in conserved mitochondrial sequences. This approach can improve vector surveillance associated with Plasmodium (malaria) or arbovirus transmission and inform targeted vector control strategies.

Data availability
All sequence data associated with this manuscript are freely available in GenBank. All relevant accession numbers are listed in Table 1 and Table 2. Author contributions YUA, DM, JV, and AM conceived of, designed and coordinated the study. YUA, DO and TOO did sample collection and identification. YUA and EM carried out the molecular genetic studies. YUA and JV carried out the sequence analyses and drafted the manuscript. DM, JV and YUA contributed materials used for the study. All authors were involved in the revision of the draft manuscript and have agreed to the final content.

Competing interests
No competing interests were disclosed.

Grant information
We gratefully acknowledge the financial support for this research by the following organizations and agencies: Swedish International Development Cooperation Agency (SIDA), grant number 75000529 to YUA as an African Regional Postgraduate Programme in Insect Science (ARPPIS) student; Funds from Training Health Researchers into Vocational Excellence (THRiVE) in East Africa (grant number 087540) funded by Wellcome Trust to JV and DM supported part of the field sampling. We also acknowledge funding from UK's Department for International Development (DFID); the Swiss Agency for Development and Cooperation (SDC); and the Kenyan Government.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Ajamma have an article on the use of the technique high resolution melt analysis towards the identification of morphologically indistinct species of mosquito. The specific goal is to expand the current set of primers in the research literature in order to identify more species from multiple genera of mosquitoes.
The methodology is clear with sufficient details for it to be reproduced by listing all appropriate reagents, DNA primer sequences and real-time PCR instrumentation. I do recommend adding the criteria by which the melt curves were deemed to be sufficiently different to allow identification of the species as compared to the "Did Not Separate" state as reported in Table 4.  table 4. I appreciate the authors efforts to repeat and report data from the previously published primers ("AgamUni") as a point of comparison. Appropriate controls were used with (1) water as a negative control for amplification and (2) samples from defined colonies and samples previously sequenced as positive controls. The most significant limitation is the number of replicates, and the diversity of sample collection points for each species. The authors clearly acknowledge these limitations in the conclusion and clearly state the need for additional samples to asses the intra-specific variation which is critically important information to make this method highly useful.
In summary, the paper is clearly and concisely written with 1 minor recommendations for additional information on the method. The goals of the research are clearly stated, and the results as well as the conclusions support the goals. The researchers have achieved the goals by identifying and confirming at least one primer pair for each of 4 genera that identify various species that are difficult to identify by morphology alone.

I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. Competing Interests: