Identification of field caught Anopheles gambiae s.s. and Anopheles arabiensis by TaqMan single nucleotide polymorphism genotyping

Anopheles gambiae sensu lato (s.l.) is a species complex composed of seven biological species, including two widely distributed and important vectors of human malaria in subSaharan Africa (An. gambiae sensu stricto and Anopheles arabiensis, whose geographical ranges broadly overlap [1‐5]. Two species (Anopheles melas and Anopheles merus) are restricted to brackish water environments of coastal Africa, and are of regional importance as vectors; the other 3 species (Anopheles bwambae, and Anopheles quadriannulatus species A and species B) are species of localized distribution [6]. For purposes of basic and operational research and as part of malaria control programmes, it is often necessary to identify field caught An. gambiae s.s. and An. arabiensis, but they are not distinguishable morphologically. Cytological methods based upon chromosome inversion patterns were the original method to distinguish the species [3]; other methods have included isoenzyme electrophoresis and cuticular hydrocarbon analysis [7]. Elucidation of ribosomal DNA (rDNA) sequences allowed development of probes and primers, permitting application of molecular based methods for species identification, including DNA-DNA hybridization and polymerase chain reaction [8‐10]. Scott et al. [8] aligned ca. 1,000 base pairs (bp) of the An. gambiae s.l. intergenic spacer (IGS; GenBank U10135), situated between the 3' 28S and 5'IGS regions of the genome, and from these alignments designed a universal forward primer (i.e., a primer conserved amongst five species in the complex) and reverse primers specific for five of the seven known species. Their use in a standard polymerase chain reaction yields amplicons of species-diagnostic sizes in gel electrophoresis. This method has become the standard tool for species identification in this complex, but it has the drawback when applied to large scale field studies of being laboriously slow. Nucleic acids from each specimen must be extracted, and then a standard PCR reaction carried out on each specimen, followed by gel electrophoresis. For large numbers of individuals, and given an unknown success rate of the reaction, the amount of time required can significantly slow the progress of a study. Often, a subsample of the entire sample is done, and then the proportion of either species is inferred to represent accurately the larger sample [11, 12]; in other cases, the entire sample is analysed [13‐15]. Rafferty et al. [10] suggested a remedy to speed up processing by standard PCR, based upon the use of a 96-pin bacterial replicator to transfer samples and reagents. Additionally, in practice, the reaction often leads to failures, repeated attempts at PCR with the same specimens, and uncertain interpretations for negative samples [16].

Genotyping single nucleotide polymorphisms (SNPs) as alleles using automated PCR in a quantitative format such as TaqMan™, has become a useful tool in detection of nucleotide sequences in which one or a few nucleotides vary in a genomic nucleotide sequence as in SNPs, and where these variations define alleles or genotypes whose presence, or frequency, in a population is of interest [17‐20]. The method requires quantitative PCR to a determined number of amplification cycles, and hybridization of fluorescently-labeled probes binding to the sequences of interest within the amplicon. The 5' nuclease activity of the Taq polymerase liberates the fluorescent label (or "fluorophore") conjugated to the hybridized probe, thereby unquenching the fluorophore, and causing it to fluoresce. Intensity of fluorescence is quantified spectrophotometrically. If two probes designed to hybridize to different nucleotide sequences of alleles operating at the same locus (or, in this application, to sequence variants occurring along the same nucleotide consensus sequence) are labeled with different fluorophores at the 5' end of the sequence, then hybridization will yield fluorescence of one fluorophore or the other. The amount of fluorescence of one fluorophore to the other becomes an X, Y bivariate whose values indicate the likelihood of one, the other, or both genotypes being present in the sample [19, 20]. Efficiency of probe hybridization is enhanced by conjugation of minor groove binding ligands to the 3' end of the probe and to the 3' direction of the quencher [18, 21]. Statistical algorithms based upon cluster analysis and maximum likelihood estimation provide the basis for classification of these fluorescence values to the correct allele or DNA sequence, referred to in technical jargon as "calling" the alleles [20].

The nucleotide sequences of interest within the IGS region aligned by Scott et al. [8] show sufficient variation for application of the TaqMan genotyping technology for identification of the two major malaria vector species in the An. gambiae s.l. species complex, namely An. arabiensis and An. gambiae s.s. There is insufficient variation for application of the method to the remaining species in the complex, although further research may reveal regions of the rDNA that would be useful in this regard. The purpose was to develop a method of higher throughput, with sensitivity and specificity equal to or greater than that of standard PCR.

The results of the three experiments reported here demonstrate that the TaqMan genotyping system provided a high reaction success rate and acceptable sensitivity in identification of An. gambiae s.s. and An. arabiensis from field caught An. gambiae s.l. adults in western Kenya. Because the rDNA sequence information used to develop the primers and probes was based upon that used for standard PCR for An. gambiae s.l. throughout Africa [8], it is very likely that the method would apply to other specimens from other geographic areas. The fact that no reactions occurred with a sample of culicine mosquitoes suggests that the method is specific as well. The reactions here all involved a single leg of each specimen, from which DNA was extracted with use brief use of a common sonicator bath, leaving the remainder of the specimen for analysis of other attributes, such as parasite infection, parity determination, body size measurement, etc.

The results from experiment 1 and 2 indicated a relatively higher success rate than in experiment 3; an explanation for this difference may be related to the handling of a much larger number of specimens in the latter study. Occasionally, it was observed that legs of individual mosquitoes would adhere to the side of the wells of the PCR tray during sonication, just above the 40 ul volume of extraction buffer; and thus DNA would not have been extracted in those few cases. The most parsimonious explanation for the slightly decreased reaction success rate in experiment 3 is this form of systematic operator error, and unlikely to be any aspect of the quantitative PCR or genotyping procedures thereafter. Indeed, in experiments 1 and 2, where fewer specimens were tested and they were handled more carefully, success rate was higher, ranging from 90 to 97%. Others have considered the issue of poor DNA quality of field-collected and archived samples, and have suggested a restriction fragment length polymorphism approach to species identification [26, 27], or have not observed it to be a problem [10]. The TaqMan method here obviates the need for several steps that normally are required in standard PCR, including multiple sample transfers; use of a standard thermal cycler; and preparing, visualizing, interpreting agarose gels in gel electrophoresis, and disposal of ethidium bromide-contaminated waste.

The proportions of the two species in the field populations agree with other studies in western Kenya, where these two species commonly overlap in geographic distribution and larvae occur in the same habitats [11, 15, 16, 28]. Results of experiment 2, in which An. arabiensis was about 10% of the population of An. gambiae, were in complete agreement with those of Huang et al. [22], from which study those specimens came. Thus, the TaqMan genotyping method appears to repeat well the standard PCR method when archived specimens are retrieved, although standard PCR was repeated here on those samples for purposes of comparison, in a blinded fashion. Of interest was that a small number of specimens were determined to be both species in this study. For example, in experiment 3, twenty-nine of 3,041 or 0.9% of individuals were called hybrids. It cannot safely be concluded that these were true hybrids, vs. an artifact of laboratory contamination leading to DNA from the two species ending up in a single extraction; these were field collections that were handled and sorted in multiple ways until they were ultimately processed as described here. Nonetheless, others have suggested that genes introgress from one species to the other [29], and indeed An. arabiensis/An gambiae s.s. hybrids have been observed in nature but at a very low rate (ca. 0.1%) [30]. When 11 of those 29 classified as hybrids by TaqMan genotyping from experiment 3 were then processed by standard PCR, only one showed both amplicons in the gel phenotype, suggesting a true hybrid; or possibly, it represented a sample with contaminating DNA. It was a female, which (in contrast with males) could have both alleles such that hybrids could be detected with this method [30]. The remainder of the females was An. arabiensis in standard PCR, which result suggests that the TaqMan genotyping method may err slightly in a false positive direction for both alleles when the true species is An. arabiensis. The 18 other specimens were males, and thus a hybrid result from them must have been from contaminating DNA. Nonetheless, the error rate was less than 1% and quite acceptable.