Comparative analyses reveal discrepancies among results of commonly used methods for Anopheles gambiae molecular form identification

Table 1 lists the indoor-resting female samples processed in this study, the dates of collections and the numbers of specimens genotyped. Figure 2 shows the location of the sampling sites. The specimens from Guinea Bissau and The Gambia [21] were selected based on inconsistent results from IGS [16] and SINE200 insertion [18] genotyping.

DNA was extracted from either legs or other parts of the carcasses not including the abdomen, to avoid possible biases associated to the risk of contamination with DNA from sperm harboured in spermathecae.

Anopheles gambiae samples were first identified based on results from PCR-RFLP approach (from now on PCR-RFLP⁵⁸¹) recognizing a T/C SNP (T = M-form; C = S-form) at position 581 of IGS rDNA region (from now on IGS⁵⁸¹[16]). Subsequently, the following genotyping approaches were applied (Figure 1 shows the position of primers and restriction sites on the IGS amplicon sequence): i) the PCR-RFLP approach (from now on PCR-RFLP⁶⁹⁰) recognizing a A/T SNP (A = M-form; T = S-form) at position 690 of IGS rDNA region (from now on IGS⁶⁹⁰[17]); ii) the PCR approach using allele-specific primers designed to detect the IGS⁵⁸¹ SNP (from now on AS-PCR [13]); iii) the PCR approach based on the specific and irreversible single-locus insertion of a SINE200 transposable element in the X-chromosome centromeric region (from now on SINE-PCR [18]), about 1 Mb from the IGS rDNA region including the IGS form-specific SNPs. A sub-sample of specimens from Angola, Burkina Faso, The Gambia and Guinea Bissau were also identified by a PCR approach utilizing Intentional Mismatch Primers containing single base mismatches at the third nucleotide from their 3' end (from now on IMP-PCR[14]).

An IGS fragment of 367 bp (from now on "IGS-amplicon") was amplified using UN and GA primers by Fanello et al[16] (Figure 1) from selected specimens and sequenced using ABI Big Dye Terminator v.2 chemistry and an ABI Prism 3700 DNA Analyser. Chromatograms were inspected for double peaks by eye. PCR and sequence analyses were carried out in Rome and/or Lisbon. Selected samples were analysed in both laboratories for results validation.

QSVanalyzer software - which allows the extraction of quantitative sequence variant (QSV) information from sequence electropherograms - was applied to estimate the relative proportions of the double peaks (i.e., c opy n umber p roportions: CNP)[22]) observed in electropherograms of IGS amplicon at positions 581 [16] (hereafter CNP⁵⁸¹) and 690 [17] (hereafter CNP⁶⁹⁰) in sequences of the IGS locus from single A. gambiae specimens. The programme analyses each trace and adjusts it in relation to the peak heights of upstream/downstream nucleotides, allowing rapid batchwise analysis of DNA sequence traces for estimation of the relative proportions of two QSVs at a given site. Kruskal-Wallis and multiple comparison tests were carried out by STATISTICA 6.1 (StatSoft, Inc. 2003).

A higher number of MS heterozygous patterns resulted from AS-PCR than from PCR-RFLP genotyping, mostly in specimens from Burkina Faso, Guinea Bissau and The Gambia. This is due to a low specificity of the AS-PCR approach (using form-specific primers differing only for the SNP variant at 3' end) which is affected by the inability of this single 3' mismatch to prevent extension of the non-specific primer by the polymerase [23]. In fact, this low specificity has been recently circumvented using primers containing an additional intentional mismatch at the third nucleotide from the 3' end which increases the power of Taq polymerase to extend from the 'right' primer and to partly optimize the reaction thermodynamics when both primers anneal on the template, thus providing more power to identify MS hybrids (IMP-PCR [14]). The comparison between the results of the AS-PCR and of IMP-PCR highlighted a higher specificity of the latter. In fact, IMP-PCR produced patterns consistent with those obtained either by sequencing or by the two PCR-RFLP approaches in all specimens tested (N = 146), with the exception of two S-form specimens from Gambia genotyped as MS by IMP-PCR (Table 2, line 18).

An A/C heterozygous pattern was observed in the recognition sequence of the enzyme utilised in IGS⁶⁹⁰ PCR-RFLP (i.e. MseI) in five out of 32 M-form specimens from Angola. This polymorphism did not allow the complete cleavage of the M-specific PCR-amplified band, thus producing a false heterozygous MS⁶⁹⁰ pattern.

A few specimens from Burkina Faso, Cameroon, Guinea Bissau and The Gambia were incorrectly genotyped as MS by PCR-RFLP (IGS⁵⁸¹: N = 8; IGS⁶⁹⁰: N = 5), due to incomplete digestion of the PCR-amplified fragment during restriction. A second round of PCR-RFLP reactions did not change the observed PCR-pattern and the specimens were confirmed to be homozygous at each site by sequencing.

This has been already hypothesized by Caputo et al[21] based on the inconsistent results from PCR-RFLP⁵⁸¹ and SINE-PCR on samples from Guinea Bissau and The Gambia, where a secondary contact zone between the two molecular forms has been hypothesized based on the high frequencies of MS putative hybrids reported [19, 20]. The results obtained confirm this hypothesis and highlight the technical bias which emerged when the same samples were identified by PCR-RFLP⁶⁹⁰. In fact, the restriction enzyme used for the PCR-RFLP⁵⁸¹ (i.e. HhaI) recognizes a S-specific restriction site, while the enzyme used for the PCR-RFLP⁶⁹⁰ (i.e MseI) recognizes a M-specific restriction site. It is possible to hypothesize that the PCR-amplification of individuals characterized by a number of copies of the M-IGS type higher than of S-IGS type exponentially increases this difference, producing a strong M⁵⁸¹ band and a weak S⁵⁸¹ one. The latter may not be visible on the agarose gel after the restriction step resulting in a MM⁵⁸¹/MS⁶⁹⁰ RFLP pattern. Conversely, individuals characterized by a number of copies of the S-IGS type higher than of M-IGS type are likely to produce a MS⁵⁸¹/SS⁶⁹⁰ RFLP pattern. This hypothesis is further supported by the relative high frequency of MM⁵⁸¹/MS⁶⁹⁰ (9%) and MS⁵⁸¹/SS⁶⁹⁰ (12%) specimens in the sample analysed, and by the absence of SS⁵⁸¹/MS⁶⁹⁰ and MS⁵⁸¹/MM⁶⁹⁰ genotypes. The QSV analysis of IGS sequences confirms that MM⁵⁸¹/MS⁶⁹⁰ or MS⁵⁸¹/SS⁶⁹⁰ individuals have proportions of array copy number intermediate between those of either MM⁵⁸¹/MM⁶⁹⁰ and MS⁵⁸¹/MS⁶⁹⁰ or SS⁵⁸¹/SS⁶⁹⁰ and MS⁵⁸¹/MS⁶⁹⁰ individuals, respectively (Table 2).

The comparison between the results of the IGS-genotyping (including direct sequencing, in case of inconsistencies among the approaches utilized) and of SINE-PCR showed consistent identifications in all samples, with the exception of those from Guinea Bissau and The Gambia. In these populations mismatches were observed, mostly due to SS and MM SINE-homozygotes with a heterozygous MS^IGS genotype or, less frequently, to MS SINE-heterozygotes with MM^IGS and SS^IGS genotypes, while no opposite MM-SINE/SS^IGS or SS-SINE/MM^IGS were found. As discussed in Caputo et al[21], the former individuals are likely to represent Fn progenies of inter-form crosses occurring in this "secondary contact zone", where the reproductive isolation mechanisms between M- and S-forms appear to be less effective than in the rest of the molecular forms sympatric distribution range. In fact, discrepancies between results from PCR-RFLP⁵⁸¹ and SINE-PCR led to hypothesize that the high frequencies of MS⁵⁸¹ patterns found in Guinea Bissau and in The Gambia were due to the presence of both M- and S-arrays in the multi-copy IGS rDNA region of single individuals, suggesting inter-locus recombination [21]. In this scenario, the SINE-PCR genotyping allows to discriminate putative MS hybrids from progenies of Fn-backcrosses (i.e. MM or SS SINE-homozygotes showing both M- and S-specific IGS arrays). In fact, the SINE-PCR genotyping of four MS⁵⁸¹ specimens reported in della Torre et al[7] (from Benin, Mali, Guinea and The Gambia) confirmed their putative hybrid origins.

On the other hand, the finding of high frequencies of consistent MS IGS/SINE patterns in larval samples from Burkina Faso, led Riehle et al.[24] to carry out a deeper genetic characterization of these individuals and to hypothesize that they may represent a new A. gambiae "sub-form" highly differentiated from M and S. This "sub-form" seems to be also characterized by a MS SINE-polymorphism in Hardy-Weinberg equilibrium consistent with IGS-patterns, a very unexpected scenario which needs to be taken into consideration when speculating on the origin of this putative "sub-form". In fact, based on their evolutionary dynamics, both IGS and SINE markers are expected to undergo rapid fixation in a randomly mated diverging taxa rather than being found at equilibrium in a taxon separated from M- and S-form.

Overall, the results here presented, as well as those by Riehle et al[24], do not only highlight limits in the approaches currently applied to discriminate M- and S-forms, but also on the actual definition of the two molecular forms, which might not fully correspond to the two A. gambiae incipient species in their entire geographical range. The M and S molecular forms are, in fact, defined specifically based on SNPs in the IGS region, which were initially used to discriminate between Mopti and Savanna/Bamako chromosomal forms in Mali and Burkina Faso [15] and, later, to identify two incipient species in other geographical regions, where the correlation with specific karyotypes was more complex [1, 5, 7, 25]. Since their initial description, all data on the genetic, ecological and behavioural divergence of M and S forms were obtained based on the IGS diagnostics, leading to a general acceptance of the IGS-SNPs as form-specific characters possibly linked to genes or genomic regions instrumental to the speciation process. This view was reinforced by the fact that the IGS lies within X-chromosome centromeric region, where most genetic divergence between M- and S-forms is observed [11, 12, 25, 26] and by the consistent almost complete absence of MS^IGS genotypes in nature. The finding of different number of copies of M- and S- IGS-arrays in single individuals from the western extreme of the molecular form range [21] highlighted how the genetic definition of the two A. gambiae incipient species is not fully tenable along their entire range. The recent sequencing of the genome of M- and S-colonies from Mali [11] and the availability of affordable SNP micro-array platforms [12], will probably allow in the near future a relatively easy processing of A. gambiae populations from the entire range. Moreover, the likely detection of multiple markers along their M-and S-form genome and their association will possibly allow a more precise definition of the two incipient species, as in the case of the allelic variant of TEP1 immune gene found to be fixed in M samples from Mali and Burkina Faso but absent in sympatric S populations [27].