Our method is based on parallel PCR amplification of the target sequences, followed by primer-extension-mediated minisequencing using fluorochrome-labeled dideoxynucleoside triphosphates (ddNTPs). Subsequent base calling occurs on a microarray upon sequence-specific hybridization. The flow chart in Fig. depicts schematically the principle of the parallel SNP analysis system.
FIG. 1. Flow diagram of the analytical procedure, starting from blood samples collected in the field. DNA is prepared from blood samples and is amplified by nested PCR; subsequently all amplicons are combined, and nucleotides are eliminated by SAP. Primer extension (more ...) Analyzed material.
To establish and evaluate the technique, cultured material from strains 3D7 and K1 and samples collected during drug efficacy studies using molecular analyses of drug resistance markers in Papua New Guinea have been used. Ethical clearance has been granted for this study by the Papua New Guinea Medical Research Council.
Blood samples and DNA preparation.
Blood samples were collected either in EDTA Microtainer tubes (BD Biosciences, Allschwil, Switzerland) or on Isocode sticks (Schleicher and Schuell, Dassel, Germany). Plasma was separated from blood samples in EDTA by centrifugation, and red blood cell pellets were stored frozen until use. DNA from cultures and field samples was extracted from 50- to 100-μl blood pellets using QIAamp DNA blood kits (QIAGEN, Hombrechtikon, Switzerland) according to the manufacturer's instructions. DNA was eluted from Isocode sticks according to the manufacturer's recommendations.
We analyzed the following 36 polymorphisms in five genes at 32 SNP sites: pfmdr1 codons N86Y, Y184F, S1034C, N1042D, and D1246Y; pfcrt codons C72S, K76T, H97Q, T152A, S163R, A220S, Q271E, N326D/S, I356L/T, and R371I; pfdhfr codons A16V, N51I, C59R, S108N/T, and I164L; pfdhps codons S436A, A437G, K540E, A581G, A613T/S, I640F, and H645P; and pfATPase6 codons S538R, Q574P, A623E, N683K, and S769N. Oligonucleotides for amplification, extension, and arraying are shown in Table S1 in the supplemental material. To cover all SNP sites, we performed 10 PCRs with the amplification primers listed in Table S1 in the supplemental material. The amplification reaction mixture contained 1× PCR buffer with MgCl2 in a final concentration of 3 mM, 0.2 mM deoxynucleoside triphosphates, and 0.2 μM of each primer. Reactions were carried out in 50 μl containing 2.5 μl DNA and 2.5 U Taq polymerase (Firepol; Solis BioDyne, Tartu, Estonia). Cycling conditions were 96°C for 3 min followed by 20 cycles of 96°C for 30 s, 52°C for 90 s, and 72°C for 90 s.
As our aim also was to identify SNPs in asymptomatic samples from community-based surveys, we performed nested PCR for the highest sensitivity. Nested PCRs were carried out in 100 μl with 5 μl primary PCR product and 5 U Taq polymerase. The buffer and cycling conditions were identical to those described above, but nested PCR primers were used (see Table S1 in the supplemental material).
To eliminate nonincorporated nucleotides, all nested PCR products of one blood sample were pooled, and 5 μl of a 1:10 dilution of the pooled PCR product was digested with 2 U shrimp alkaline phosphatase (SAP) (Amersham Biosciences, Freiburg, Germany) in a reaction volume of 12 μl for 1 h at 37°C. SAP was inactivated by incubating samples for 15 min at 90°C.
Since most microarray scanners support only dual-fluorescence measures simultaneously, a strategy of two parallel reactions had to be applied. Two primer extension reactions were carried out per sample. The reaction mixes differed in their combinations of Cy3- and Cy5-labeled ddNTPs (Perkin Elmer, Schwerzenbach, Switzerland), and extension primers were added as shown in Table S2 in the supplemental material. Thus, it was possible to detect all possible SNP permutations by using only two fluorochromes. All primer extension reactions for one sample were carried out in two aliquots of 20 μl containing 1× Sequenase buffer, extension primer mix 1 or 2, ddNTP mix 1 or 2 (see Table S2 in the supplemental material), and 2 U Thermo Sequenase (Termipol; Solis). The concentration of ddNTPs in both mixes was 0.25 μM, and primers were diluted to a concentration of 6.25 nM each. The extension reaction was cycled 35 times at 94°C for 30 s and at 50°C for 10 s, with an initial cycle of 1 min at 94°C. After the extension reaction was performed, both mixtures were pooled and 6 μl denaturing solution (3% sodium dodecyl sulfate [SDS] in 40 mM EDTA, pH 8.0) was added. The sample was denatured at 95°C for 60 s and subsequently was kept on ice until hybridization onto the microarray.
Microarrays carried short oligonucleotides (20 to 35 bp) corresponding to the antisense DNA of the extension primers (see Table S1 in the supplemental material). All oligonucleotides possessed a C7-aminolinker and were spotted onto aldehyde-activated glass slides (Genetix, Munich, Germany). Prior to the spotting of oligonucleotides, a mask with 12 circular wells (diameter, 8 mm) was applied to the surface of each slide (MaProline GmbH, Starrkirch-Wil, Switzerland). Oligonucleotides were spotted in triplicate, and anchor oligonucleotides prelabeled with Cy3 and Cy5 as well as four oligonucleotides with a random sequence were added as positive and negative controls, respectively.
Slides were spotted using a VersArray ChipWriterPro system (Bio-Rad Laboratories, Hercules, CA). Oligonucleotides were dissolved in 180 mM phosphate buffer, pH 8.0, and 0.5 nl of a 50 μM solution was spotted onto the slides. Slides were stored desiccated and in the dark until used for hybridization.
Twenty-three microliters of the pooled and denatured primer extension reaction mixture was transferred to a well of a microarray glass slide, and 6 μl 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) was added. Hybridization was carried out in a humid chamber at 50°C for 60 to 90 min. After hybridization, the slide was washed at room temperature in 2× SSC plus 2% SDS for 20 min, followed by another wash with 2× SSC for 20 min, and then a final wash with 2× SSC plus 2% ethanol for 2 min. The slides were dried with compressed air and stored in the dark until scanned.
Hybridized slides were scanned at 635 and 532 nm using an Axon 4100A fluorescence scanner (Bucher Biotec AG, Basel, Switzerland). Cy3 and Cy5 images were acquired and analyzed using the Axon GenePix Pro (version 6.0) software. This software generates data points using pixel intensity after background subtraction. We developed software for further analysis of raw data that produces an output determining whether an infection is wild type, mutant, or mixed, and it also determines the dominant genotype in the latter case. Each signal was classified either as wild type, mutant, or mixed based on the expression intensities of the scanned image. The grouping was done according to the following algorithm: fluorescence intensities below 9,000 (Cy3) or 10,000 (Cy5) U (mean intensities minus background) were regarded as negative. For measures above these cutoff values, we considered the ratio of Cy5 intensity to Cy3 intensity to discriminate between wild-type, mutant, or mixed infection.
To determine an optimal algorithm to translate the output of the GenePix Pro software into predictions about the genotypes present in analyzed samples, we used two singly infected blood samples that previously were sequenced at 29 SNP sites. Sequence data showed that the samples from 3 of 29 SNP sites differed (C59R, S108N, and A437G). Either single or mixed samples were analyzed with the chip in various proportions (1:2, 1:4, 1:8, and 1:16). With this approach, we could empirically determine the following threshold values: for Cy5/Cy3 ratios below 0.7, the sample was classified according to whether the wild type or mutant was labeled with Cy3. Ratios between 0.7 and 2.4 were assigned to mixed genotypes, and ratios above 2.4 were assigned to the Cy5-labeled genotype.
To estimate the above-mentioned threshold parameters and to determine the predictive accuracy of our method, we used three of four identical but independently processed microarrays to estimate the threshold value to distinguish positive from negative signals so that the results would match the sequence data as closely as possible. The fourth microarray then was used to apply this algorithm to determine the predictive accuracy of the method. This procedure was repeated four times in all possible combinations. Finally, we applied this algorithm to samples that were genotyped by sequencing and PCR-RFLP to determine the sensitivity and specificity of our method.
PCR products were purified by size-selective polyethylene glycol precipitation (12
) and directly sequenced using one of the respective nested PCR primers. Cycle sequencing (25 cycles of 96°C for 30 s, 50°C for 15 s, and 60°C for 4 min) was performed using the ABI PRISM Dye Terminator cycle sequencing ready reaction kit (Perkin Elmer), and sequences were analyzed using an ABI PRISM 310 genetic analyzer and the ABI PRISM software.