A new high-resolution melting analysis for the detection and identification of Plasmodium in human and Anopheles vectors of malaria

Among vector-borne diseases malaria is the leading cause of morbidity in the world, with more than 200 million cases per year and a large number of deaths. The techniques traditionally used for the detection of Plasmodium in humans and Anopheles mosquitoes include microscopy, IRMA, ELISA, antibody or molecular assays, and anopheline dissection. However, these techniques are limited by their requirement of skilled personnel, low sensitivity or long processing times. A PCR-based high-resolution melting (PCR-HRM) analysis was developed for the detection and identification of P. falciparum, P. vivax and P. malariae that infect humans and Anopheles. In 41 human samples PCR-HRM detected 14 samples positive for P. vivax, 17 for P. falciparum, three for P. malariae, three mixed infections for P. vivax/P. malariae and four negative samples. Whereas benchmarking assays of microscopy and nested PCR had false positive detections. Additionally, PCR-HRM was able to detect natural infection with Plasmodium spp. in An. darlingi and An. mattogrossensis. The PCR-HRM presented is the first single assay developed for the detection and identification of P. vivax, P. falciparum and/or P. malariae in human and Anopheles. This method improves on currently available assays as it is easy-to-use, rapid, sensitive and specific with a low risk of contamination.

Among vector-borne diseases malaria is the main cause of morbidity in the world. It is estimated that in 2015 there were 212 million new cases and 429,000 mortalities due to malaria 1,2 . The main malaria species that infect humans are Plasmodium vivax, P. falciparum, P. malariae and P. ovale, of which, the first two, represent 95% of infections 1,3 . Plasmodium malariae has been reported in a limited number of cases in Colombia 4 , P. ovale predominates in Sub-Saharan Africa 5 and recent studies have shown that P. knowlesi, P. brasilianum and P. simiun cause malaria in some human populations [6][7][8] . In Colombia in 2016, 83,356 malaria cases were reported, of which 33,055 (39.7%) were from P. vivax, 47,497 (57.0%) from P. falciparum and 2,804 (3.3%) from mixed malaria infections 9 .
Mosquitoes belonging to the genus Anopheles are the vectors of malaria in humans. This genus comprises approximately 465 species, of which it is estimated that 41 have epidemiological importance in malaria transmission 10 . Traditionally one of the most widely used methods to detect Plasmodium infection in the Anopheles has been dissection of the midgut and salivary glands to detect sporozoites under a microscope 11 . Subsequently, new techniques for the detection of natural infections in anophelines began to emerge, including immunoradiometric assays (IRMA) 12,13 and the detection of the CS protein (of the circumsporozoite) by enzyme-linked immunosorbent assays (ELISA) 14 . However, these methods are time-consuming, require highly skilled personnel and have low sensitivity.
The diagnosis of malaria in humans is traditionally based on the microscopic detection of Plasmodium parasites in blood smears. This method is considered the gold standard diagnostic test for human malaria and is the most commonly used due to its low cost and simplicity 15 . However, disadvantages to this test include the

Results
Design of primers for pCR-HRM. Using the sequences reported for the 18S rRNA gene of Plasmodium spp. two pairs of primers were designed, Endmal18sF-R and Amzmal18sF-R (Fig. 1). The Endmal18sF-R primers ( Fig. 1A) were designed to target single nucleotide polymorphisms (SNPs) unique to each Plasmodium spp., whereas Amzmal18sF-R targeted Plasmodium species-specific SNPs, as well as, size varying insertions/deletions (Fig. 1B). For primer Amzmal18sR three degenerate bases were incorporated in order to increase the amplification success of the three Plasmodium species.
The primers Endmal18sF-CCTGGTTGATCTTGCCAGTAG and Endmal18sR-GCTACAGCTTTTCCGTAG TTATCC amplified a 162 bp fragment, whereas the primer pair Amzmal18sF-GGATAACTACGGAAAAGCTG and Amzmal18sR TTAYGTGTYRCTTCTTTGTT amplicons varied in length from 132 to 137 bp, the length depending on the species of Plasmodium targeted (Fig. 1B). As both primer pairs amplified species-specific regions, the effect of these sequence variations were investigated to establish whether this allowed species discrimination through melting curve analyses. theoretical estimation of the melting temperatures of the amplicons obtained for P. vivax, P. falciparum and P. malariae. The  optimal pCR annealing temperatures. Using conventional PCR the primers Endmal18sF-R and Amzmal18sF-R demonstrated a good ability to amplify DNA isolated from the positive control of P. falciparum 3D7. The Endmal18sF-R amplicons were generated from 45.4 °C to 60 °C, whereas the Amzmal18sF-R amplicons were generated from 45.4 °C to 53 °C. We determined the optimal annealing temperatures for the PCR-HRM assay as 55 °C or 52 °C for the Endmal18sF-R or Amzmal18sF-R primers, respectively.
Assay specificity and sensitivity of pCR-HRM for Plasmodium spp. The standard curve in the PCR-HRM assay using the Endmal18sF-R and Amzmal18sF-R primers showed good correlation (0.99), with an efficiency of 93%. Our PCR-HRM assay achieved a good limit of detection to Plasmodium spp. parasitemia levels, where the amplification range was 10 1 to 10 6 copies ( Supplementary Fig. S1). Additionally, it was determined that the ratio of 1:100 P. falciparum:interfering DNA did not inhibit the PCR-HRM reaction. However, an effect on the amplification of the PCR does occur, which is observed by a change in the Ct value. For both primer pairs the Ct value is higher in the presence of the interfering DNA ( Supplementary Fig. S2). Our PCR-HRM assay for both primer pairs showed specificity for Plasmodium DNA as there was no amplification with human DNA or An. darlingi DNA. We also conducted an in silico specificity analysis demonstrating that the Endmal18sF-R and Amzmal18sF-R primers were non-specific to the genomes of human and Anopheles spp. Furthermore, no similarity was observed against the genomes of Leishmania spp., Leptospira interrogans, Rickettsia spp., Brucella spp., Mycobacterium tuberculosis, dengue serotype 1-4, Zika virus, Chikungunya virus and Yellow fever virus. Additionally, only partial similarity of the two primer pairs was observed against the DNA of Toxoplasma gondii and Trypanosoma cruzi.
The PCR-HRM assay using the primers Endmal18sF-R and Amzmal18sF-R succeeded in distinguishing the three species of Plasmodium studied. The melting curves were plotted as normalised melting curves and peaks using the LightCycler ® 96 software v1.1.0.1320. The melting curves and experimental T m showed a clear A.          separation between the evaluated species, allowing each species to be determined unambiguously using the Endmal18sF-R (p < 0.001) and Amzmal18sF-R (p < 0.0001) primers (Table 1 and Fig. 2). The experimental T m obtained from the PCR-HRM analyses performed on the amplicons obtained with the Endmal18sF-R primers differentiated P. vivax, P. falciparum and P. malariae with a minimum difference around 0.38 °C between the three species. However, although a similar overall result using the Amzmal18sF-R primers was obtained they showed a more marked difference in the T m between each of the three species. This allowed greater discrimination (greater difference in the T m ) between P. vivax and P. malariae than the Endmal18sF-R primers (Table 1).

P. vivax C A A A G A A G C G A C A C G T A A
We also demonstrated that our PCR-HRM assay could experimentally distinguish a mixed infection (P. vivax/P. falciparum, P. malariae/P. vivax and P. malariae/P. falciparum) with the primers Amzmal18sF-R (Fig. 3). However, this distinction was not observed with the Endmal18sF-R primers as the melting curves and the T m obtained were ambiguous showing similar values to a monoinfection. preliminary validation of the pCR-HRM in human clinical and mosquito samples. In the 41 clinical samples of DNA extracted from human blood, PCR-HRM detected 14 samples positive for P. vivax, 17 for P. falciparum, three for P. malariae, three mixed infections for P. vivax/P. malariae and four negative samples. These results did not concur with two benchmarking assays. The first by counting Plasmodium parasites in blood smears under a microscope, which only observed 26 positive samples: ten P. vivax, 15 P. falciparum and one mixed infection of P. vivax/P. malariae. The second, nested PCR 16 had 31 positive samples: 13 P. vivax, 15 P. falciparum, two P. malariae and one mixed infection of P. vivax/P. malariae (Table 2). Additionally, the Kappa index between blood smears and our PCR-HRM with the primers Endmal18sF-R was k = 0.373 IC 95% and Amzmal18sF-R was k = 0.549 IC 95%. Between our PCR-HRM and nested PCR we observed with the primers Amzmal18sF-R a Kappa index of k = 0.739 IC 95%, and with Endmal18sF-R was k = 0.396 IC 95%.
Similarly, 342 anophelines captured from an endemic area with high malaria transmission, thus having a high suspicion of being infected naturally with Plasmodium spp., were evaluated. Our PCR-HRM assay with primers Endmal18sF-R and Amzmal18sF-R detected two An. darlingi positives with P. vivax and P. falciparum, and eight An. mattogrossensis positives with P. vivax. The remaining samples were negative ( Table 2, Supplementary Figs S3, S4  and Supplementary Table S2). The positive PCR amplicons were confirmed by sequencing.

Discussion
Three of the four Plasmodium spp. -P. falciparum, P. vivax and P. malariae -traditionally known to infect humans and causing malaria are present in Colombia. Commonly, the techniques used to detect and identify these species include microscopy or by PCR-based assays using genus -and/or species-specific primers 16,25,[32][33][34] . Microscopy, although an economical and rapid technique that can be performed in rural areas with simply equipped hospitals or health centres, still requires an experienced microscopist and a high-level of infection in the sample; low or mixed parasitemia are not easy to detect leading to false negative tests. Typically, the detection  and identification of Plasmodium spp. that infect humans by PCR requires the running of two independent reactions, which increases costs and extends test turnaround times. We present here the results of the implementation of a PCR-HRM assay that allows the rapid and accurate detection and identification of three Plasmodium spp., in either clinical samples or Anopheles vector samples. Our assay differs from those PCR-HRM already published 35,36,42 that have only been used to detect Plasmodium spp. in clinical samples. The PCR-HRM assay developed was successful in detecting and identifying three of the main species of Plasmodium, with two sets of independent primer pairs Endmal18sF-R and Amzmal18sF-R. The Amzmal18sF-R primer pair were more accurate than the Endmal18sF-R primers. This a consequence of the Amzmal18sF-R design incorporating both SNPs and insertions/deletions leading to amplicon length differences that gave more effective discrimination between each of the Plasmodium species.
The technique presented improved on established PCR assays that test for Plasmodium infection 15,16,25,26,28,34 as it offers the ability to detect and discriminate between the three Plasmodium spp. using a single set of primers in a single round of PCR. An additional benefit of our assay using the Amzmal18sF-R primers was demonstrated by its sensitivity to detect and differentiate mixed infections of P. falciparum/P. vivax, P.falciparum/P. malariae and P. vivax/P. malariae (Fig. 3). This would be highly relevant in diagnostic healthcare provision as mixed infections of P. vivax/P. falciparum is the most predominant 1 . Further assay development is required to support the in silico predications that our PCR-HRM assay also has the potential to detect and discriminate between infections for P. simium and P. brasilianum, species genetically related with P. vivax and P. malariae (Supplementary  Table S3). If experiments support this in silico prediction our PCR-HRM will improve Plasmodium testing as current PCR-HRM cannot discriminate between P. simium and P. brasilianum 25,26 .
We demonstrate that estimating the theoretical T m is a good gauge to predict whether a PCR-HRM primer pair will be species-specific. Although the absolute values of the theoretical T m differed from those obtained experimentally with both our primer pairs, the difference between the values was consistent in magnitude and allowed species discrimination.
It is important in diagnostics that any new test has equivalent sensitivity to established assays. We demonstrated that our PCR-HRM assay developed with both primer pairs had excellent sensitivity, being able to detect one parasite. This is an improvement on the sensitivity of microscopy that at best has a minimum sensitivity of 10-30 parasites/µl in blood 15,16 . Our PCR-HRM is also comparable to the nested PCR method of Snounou et al. 16 -a benchmarking test for mosquito vector incrimination -and other established PCR-based assays 29,33,43 , having a limit of detection of at least one parasite (equivalent to 6 copies/μl). Eight samples of An. mattogrossensis and two of An. darlingi were negative by Snounou's PCR 16 . However, with the PCR-HRM developed in this research these samples were positive for Plasmodium (Table 2), confirming the high sensitivity of our assay.  Our PCR-HRM was found to be within the expected range of efficiency of between 90-110% 44 . This allowed us to infer that amplicon generation with both primer pairs occurred without contamination. Although our assay is not as cost efficient as those established tests such as microscopy, ELISA and other PCR-based methods, it does present a number of advantages. Our assay is able to detect and identify Plasmodium spp. in a single round of PCR, which is not possible with the method of Snounou et al. 16,25 , and it minimises the presence of false positives as commonly encountered when testing by ELISA.
To the best of our knowledge, we report the first natural infection of P. vivax in An. mattogrossensis in Colombia. This anopheline has been found to be positive for P. vivax in Brazil and Peru 45,46 . We collected An. mattogrossensis in San Pedro de Tipisca and Doce de Octubre, Puerto Nariño, Amazonas. These localities have a high transmission of malaria throughout the year. We therefore purport that this species together with An. darlingi play a role in the transmission of malaria in these communities, which current mosquito programs do not recognise.
In conclusion, the PCR-HRM developed in this research is the first test developed in Colombia that can be used for the detection and identification of three species of Plasmodium that infect both human and anopheline vectors. We suggest the preferential use of Amzmal18sF-R for the PCR-HRM, as this primer pair can be used to detect and identify P. vivax, P. falciparum, P. malariae and mixed infections. Additionally, this methodology has multiple advantages with respect to established methods such as nested PCR 16 . Our PCR-HRM is easy-to-use, rapid, sensitive and specific, uses a single PCR reaction and has a low risk of contamination. Finally, this research is currently aimed at epidemiology-based research centres, but has the potential for diagnostic development and approval.

Methods
The clinical samples (N = 41) and Anopheles mosquitoes (N = 342) were collected in the municipality of Puerto Nariño, communities of San Juan de Tipisca and Doce de October, during 2015 to 2016 as part of the project "Implementation of an Early Warning System for the Prevention and Control of the main ETV in the department of Amazonas, Colombia". During this two year period the communities were visited in total eight times, once every three months. The clinical samples were obtained by active search of patients, or concentration of persons with malaria symptomatology or persons that belonged to malaria endemic areas. In parallel to human sample collection, Anopheles spp. were collected with CDC traps located overnight inside and outside of houses, for a minimum of two nights per location. samples for pCR-HRM. Genomic parasite DNA of P. vivax, P. falciparum 3D7 and human P. malariae were supplied by the Malaria-SIU research laboratory of the University of Antioquia. For the clinical samples we used DNA extracted from blood using Qiagen ® DNA extraction kit. For each individual we took two blood samples. Following the official malaria diagnostic protocol in Colombia, the first sample was used for blood smear 47 . The additional second sample was stored on FTA cards (Whatman ® ) for molecular techniques. For the mosquito malaria vector DNA, 342 samples of Anopheles spp. were captured from indigenous settlements in Puerto Nariño, Amazonas, Colombia, an area endemic with high malaria transmission. Subsequent to the morphological identification of the specimens, the head and thorax of each mosquito was separated from the abdomen. The Plasmodium DNA extraction was made by macerating the specimens (head-thorax) for 30-40 seconds in 20 µl of Milli-Q water. Then 20 µl of 10% Chelex ® was added and heated at 56 °C for 10 minutes followed by 99 °C for 10 minutes. Finally, the solution was centrifuged at 12,000 rpm for 5 minutes and the supernatant (DNA) was separated and stored at −20 °C. The established methods of Plasmodium counts from blood smear under a microcope 47 and nested PCR described by Snounou et al. 16 were used as benchmarking assays to establish the detection of Plasmodium spp.
Genomic target selection and primer design for pCR-HRM. PCR-HRM primers were designed by targeting both a highly conserved and a variable region of 18S rRNA of Plasmodium spp. We used GenBank (NCBI, Bethesda MD, USA) sequence accession numbers M19172 for P. falciparum, M54897 for P. malariae and X13926 for P. vivax. Primer3Plus (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) was used to design two pairs of primers, Endmal18sF-R and Amzmal18sF-R, to specifically target the three Plasmodium spp. The design quality of the oligonucleotides was evaluated by OligoAnalyzer v3.1 (https://www.idtDNA. com/calc/analyzer) to avoid homodimers and heterodimers. Primer specificity was analysed by Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/).

In silico simulation of melting curves of P. vivax, P. falciparum and P. malariae for pCR-HRM.
In silico simulation of the melting curves was based on the sequence region amplified by Endmal18sF-R and Amzmal18sF-R primers. The theoretical melting temperatures (T m ) were calculated using OligoCalc (http://biotools.nubic.northwestern.edu/OligoCalc.html) and uMelt v2.0.2 (http://www.dna.utah.edu/umelt/um.php). evaluation of the optimal primer annealing temperature. A PCR-based temperature gradient was created to ensure that the two pairs of primers developed were able to amplify the target region in the three Plasmodium species, avoiding non-specific amplification or primer dimers that interfere with the interpretation of the subsequent results in the PCR-HRM analysis.
Target sequences were amplified for each primer pair independently. Reaction conditions included: 12.5 μl PCR Master Mix 2X(Thermo Fisher Scientific), 1.25 μl of each primer (10 μM), 9 μl of DNase-free water and 1 μl of DNA (10-20 ng/μl) for a final volume of 25 μl per sample. Thermocycling conditions were one cycle of 95 °C × 5 minutes, followed by 35  PCR-HRM assay sensitivity and specificity. To estimate the limit of detection of the PCR-HRM the amplicons derived from P. falciparum 3D7 using the primers Endmal18sF-R and Amzmal18sF-R were purified and cloned into the vector pGEM ® -T Easy (Promega, Madison WI) according to the manufacturer's specification. Serial ten-fold dilutions of a solution of the purified clone were made. The range of concentrations used was between 10 1 to 10 6 copies of the plasmid containing the amplified P. falciparum/μl. Each DNA concentration was assayed in duplicate by PCR-HRM.
To evaluate the specificity of the PCR-HRM, DNA from An. darlingi and from uninfected human DNA were used. The assay was performed using 50 ng of these DNAs per PCR-HRM reaction. We evaluated whether the presence of interfering DNA from An. darlingi and from humans could inhibit the PCR-HRM reaction. 10 ng of P. falciparum 3D7 was used in a 1:1 and 1:100 ratio with the interfering DNA.
In silico analysis was performed to evaluate the specificity of the assay against DNA of different microorganisms with febrile syndrome: Toxoplasma gondii, Leishmania spp., Trypanosoma cruzi, Leptospira interrogans, Rickettsia spp., Brucella spp., Mycobacterium tuberculosis, dengue serotype 1-4, Zika virus, Chikungunya virus and Yellow fever virus. statistical analysis. We tested the presence of atypical value using the Tukey test. We analysed the data using GraphPad Prism v7.04. The diagnostic of the concordance of the clinical samples was calculated using the Kappa index between blood smear, nested PCR and PCR-HRM. We also calculated summary measures using ANOVA with a statistical significance of p < 0.05.

ethics approval and consent to participate. This protocol was approved by the Human Experimentation
Ethics Committee of the University of Antioquia. For all cases we obtained informed consent of the patients included in the study. The study protocol was carried out in accordance with the ethical standards of the Declaration of Helsinki 1975 and Resolution 008430 of 1993 of the Ministry of Health of Colombia.

Data Availability
Patient information are not public due to the confidential nature of the collected personal data, but are available from the corresponding author upon reasonable request.