Three laboratory reference strains of An. gambiae s.s. were used to verify the specificity and sensitivity of the assay, and these were used as controls in subsequent experiments. R70 from Tanzania possesses the wildtype susceptible (104L) genotype; VKPR, homozygous for the L104F West African kdr allele; and RSP, homozygous for the L104S East African kdr allele. In addition, artificial heterozygotes were created by combining DNA extracted from individuals of susceptible and resistant strains, L(R70)/F(VKPR) for the West African kdr and L(R70)/S(RSP) for the East African kdr.

Field studies were conducted in villages in the lower Moshi area on the slopes beneath Mt Kilimanjaro in northeast Tanzania.

Individual mosquitoes, dried on silica gel or processed for CSP ELISA, were homogenized in 50 μl STE buffer (1 mM EDTA, 10 mM Tris-HCl, 50 mM NaCl), incubated at 95°C for 12 minutes then centrifuged at 13,000 rpm for four minutes at room temperature. Aliquots of the supernatant containing suspended DNA were transferred into fresh tubes and stored at -20°C until use. DNA extraction was carried out in individual tubes or in 96-well PCR plates using a clean pipette tip for homogenization of each sample. DNA extracted from individual, dried mosquitoes using the method described by Collins et al. 20 was used for comparison. Ten microlitres of the DNA extract from two individuals were pooled in each well, and this pooled DNA was used in the polymerase chain reaction.

Forward and reverse primers developed by Kolaczinski et al. 21 were used to amplify a 216 bp fragment of the voltage-gated sodium channel gene, with a biotin modification of the reverse primer at the 5' end (MWG Biotech, Riskov, Denmark). Each 20 μL PCR reaction consisted of 0.25 mM each dNTP, 0.1 μM each primer, one unit HotStarTaq polymerase (Qiagen, Albertslund, Denmark) in buffer containing 1.5 mM MgCl₂ (Qiagen) and 2 μL extracted DNA. Reaction conditions were 94°C for 15 minutes followed by 35 cycles of 94°C for one minute, 55°C for one minute and 72°C for one minute, with a final extension at 72°C for four minutes; the PCR product was kept at 4°C until use. Amplifications were performed in 96-well PCR plates and the reaction mixture was overlaid with one drop of mineral oil. PCR products from controls and several samples from each plate were confirmed by electrophoresis on a 1.5% agarose gel. Species identifiation of members of the An. gambiae complex was performed on individual mosquitoes according to Scott et al. 22.

The ELISA plates (Maxisorp; Nunc, Roskilde, Denmark) were coated with streptavidin in phosphate buffered saline (PBS) (1 g/mL), covered, and left overnight at 4°C. Prior to use, the plates were washed three times in washing buffer (1 × PBS containing 0.05% Tween 20). The PCR products were diluted 1:1 in water in a 96-well PCR plate, denatured at 95°C for five minutes, and immediately thereafter cooled to 4°C until use. The 3'-end digoxigenin-conjugated SSOPs (MWG Biotech, Riskov, Denmark) (shown in Table 1) were diluted to a 4 nM concentration in tetramethyl ammonium chloride (TMAC; Sigma Aldrich, Dorset, UK) solution (3 M TMAC, 50 mM Tris, pH 8.0, 0.1% sodium dodecyl sulfate, 2 mM EDTA, pH 8.0), heated to 53°C, and 100 μL was then added to each well of the ELISA plate. Two microlitres of the diluted PCR products was subsequently added. Replicate ELISA plates were made to enable simultaneous probing with SSOPs targeting all three kdr genotypes. The plates were incubated in a hybridization oven (AH Diagnostics, Aarhus, Denmark) at 53°C on a shaking device for one hour and washed three times in washing buffer. This was followed by two rounds of washing and incubation (14 minutes per round for 104L and 104S probes, 15 minutes per round for the 104F probe) in TMAC solution at 68°C. To remove TMAC, the plates were then washed three times in washing buffer, and peroxidase-conjugated anti-digoxigenin antibody in dilution buffer (1:1,000) (Roche Diagnostics, Mannheim, Germany) was added to each well. After incubation for one hour at room temperature, the plates were washed three times in washing buffer and 100 μL of room temperature TMB substrate (Sigma Aldrich, Dorset, UK) was added to the plates. The reaction was stopped after five minutes by adding 0.5 M H₂SO₄ and the optical density (OD) at 450 nm was measured in an ELISA reader. A flow diagram of the methodology is presented in Figure 1.

For each probe, the non-complementary control strains served as negative controls (e.g. for the 104L probe, wells that contained R70 served as positive controls while wells on the same plate that contained VKPR or RSP served as negative controls). Some between-experiment variation in the OD values of positive and negative controls was apparent, possibly due to marginal differences in the probe binding strength and the washing force during high stringency washes. While this rarely compromised specificity, no fixed threshold could be specified for SNP tests and for each experiment a threshold of positivity was set for each SSOP corresponding to twice the maximum negative control OD value. Samples with OD values exceeding this threshold for each SSOP were considered positive for the corresponding SNP, or combination of SNPs in the case of heterozygotes.

A set of 12 samples, consisting of six individual controls, five mixtures of controls and one field-collected sample were used in a double blind trial to test the consistency of the SSOP-ELISA method in relation to the multiplex PCR methods described by Martinez-Torres et al. 4 and Ranson et al. 6. The susceptible Dondotha strain of An. arabiensis was used in addition to R70 as a wildtype control.

Kdr primers with additional 5' EcoR1 and 3' Not1 site were used to generate PCR products, which were cloned into pAcGP67 (BD biosciences). Plasmids were prepared using MiniPrep spin columns (Omega Biotech). Sequencing was done on an ABI Prism 377 (Perkin-Elmer) using the Big Dye terminator reaction mix (Perkin-Elmer) and ABI Prism proof-reading and translation software.

To evaluate the specificity and sensitivity of the ELISA, the three reference strains of An. gambiae s.s., R70, VKPR and RSP known to carry the different kdr genotypes, as well as the artificial heterozygote mixtures of these reference strains, were tested. DNA extracted from samples by STE or the Collins method worked equally well, and further evaluation was carried out using DNA extracted by the STE method. Figure 2 shows that the assay correctly identifies the SNPs of the different strains using SSOP for 104L, 104F and 104S for single individuals. The difference in OD value between the threshold of positivity and the positive reactions, ΔOD, was always greater than 1.3. For mixtures of two individuals representing artificial heterozygotes (Figure 3), the OD obtained with kdr-specific probes (104F or 104S) was slightly lower than the OD obtained with corresponding probes for single individuals; however, the ΔODs for the different SNPs were at least 0.9. The OD using the wildtype probe (104L) was lower in both mixtures compared to single individuals, with a ΔOD of 0.9 for the L(R70)/F(VKPR) mixture and 0.8 for the L(R70)/S(RSP) mixture; however, the signal strength was sufficient for visual detection. Figure 4 shows the visual results of the SSOP-ELISA test for control strains, individually, in combination as artificial heterozygotes, and with an additional dilution, in three columns of an ELISA plate representing the different SSOP.

In two dilution series with pools containing four or five mosquitoes possessing the resistant and susceptible genotypes, the SSOP-ELISA could reliably detect one resistant individual in a pool containing three other susceptible individuals (1F:3L or 1S:3L), with the ΔOD in relation to the negative control of 0.9 for 104F and 0.7 for 104S. However, when the pool size was increased to five individuals, sensitivity to detect one resistant mosquito in a pool containing four other susceptible individuals (1F:4L or 1S:4L) was reduced, and the ΔOD for the minority genotype was <0.5.

Results of the double blind trial are presented in Table 2. All six individual controls were successfully genotyped by multiplex PCR and correctly identified by SSOP-ELISA. For five artificial heterozygotes, multiplex PCR failed to amplify the 137 bp susceptible band in both L/S mixtures, and failed to amplify the 195 bp resistant band in one of two L/F mixtures, while the F/S mixture was successfully genotyped. This was in contrast to the SSOP-ELISA which detected both SNPs in all mixtures with success. The field sample of unknown genotype was identified as 104L by multiplex PCR and SSOP-ELISA.

To ensure detection of heterozygous individuals in field-collected specimens, pools of two individuals only were used representing a total of four alleles. The individuals from positive pools were tested separately for confirmation. All 978 individuals from susceptibility tests and experimental huts in Mabogini tested homozygous for the wildtype susceptible (104L) genotype. However, the SSOP-ELISA detected two out of 642 individuals from Msitu wa Tembo village that were heterozygous for the L104F kdr genotype (allele frequency = 0.16%). These L104F kdr genotypes were confirmed by DNA sequencing. It was not possible to correlate the kdr genotype with insecticide susceptibility status since these mosquitoes had been collected by light trap catch for the purpose of sporozoite rate measurement.