Determination of Permethrin Resistance Allele Frequency of Human Head Louse Populations by Quantitative Sequencing

Abstract

A quantitative sequencing (QS) protocol that detects the frequencies of sodium channel mutations (M815I, T917I, and L920F) responsible for knockdown resistance in permethrin-resistant head lice (Pediculus humanus capitis De Geer) was tested as a population genotyping method for use as a preliminary resistance monitoring tool. Genomic DNA fragments of the sodium channel α-subunit gene that encompass the three mutation sites were polymerase chain reaction (PCR)-1 amplified from individual head lice with either resistant or susceptible genotypes, and combined in various ratios to generate standard DNA template mixtures for QS. After sequencing, the signal ratios between resistant and susceptible nucleotides were calculated and plotted against the corresponding resistance allele frequencies. Quadratic regression coefficients of the plots were close to 1, demonstrating that the signal ratios are highly correlated with the resistance allele frequencies. Resistance allele frequencies predicted by QS, using either "pooled DNA" (DNA extracted from individual louse specimens and pooled) or "pooled specimen DNA" (DNA simultaneously extracted from multiple louse specimens), agreed well with those determined by individual sequencing, confirming the reliability and accuracy of QS as a population genotyping method and validating our approach of using the pooled specimen DNA as the DNA template for QS. Our protocol for QS was determined to be highly reliable for the prediction of resistance allele frequencies higher than ≈7.4% at the 95% confidence level. According to the resistance allele frequencies determined by QS, pyrethroid resistance varies substantially among different geographical regions, emphasizing the importance of early resistance detection and proper management strategies.

human head lice, sodium channel, kdr, pyrethroid resistance, quantitative sequencing

Pediculosis, caused by the human head louse, Pediculus humanus capitis De Geer, is the most prevalent ectoparasitic infestation of humans, particularly in children, worldwide (Gratz 1999). Although head lice do not transmit diseases like body lice, P. h. humanus, and the infestation symptoms are relatively mild, the social, mental, and economic impacts of pediculosis are substantial. Most people find lice intolerable and often repeatedly and prophylactically apply pediculicides without realizing the potential hazard and toxicity if misused or overused. Extensive use of pediculicides has resulted in rapid development of resistance. Louse resistance to most commercial pediculicides is widespread and increasing in frequency (Pollack et al. 1999), particularly to DDT, malathion, and permethrin (Yoon et al. 2004, Pittendrigh et al. 2006). Knockdown resistance (kdr) is a major factor in all permethrin-resistant lice worldwide and supports the claim that treatment failure is largely due to resistance (Lee et al. 2000).

Three point mutations (M815I, T917I, and L920F) in the voltage-sensitive sodium channel (VSSC) α-subunit gene have been identified in permethrin-resistant head lice (Lee et al. 2000, Lee et al. 2003). Allele frequencies of the T917I and L920 F mutations were well correlated with actual levels of permethrin resistance in head lice from California, Florida, and Texas (Gao et al. 2003). Recently, functional expression in conjunction with electrophysiological recording of sodium channel variants with or without these mutations has confirmed that sodium channel sensitivity is three- to four-fold reduced by both M815I and L920F mutations, whereas it is almost abolished by the T917I mutation (Yoon et al. 2008). Previous reports indicate that permethrin resistance is global, but it varies in intensity and is not yet uniform (Gao et al. 2003). Early resistance detection is critical for proper management decisions that can delay and reverse this trend. To this end, rapid detection of mutations responsible for kdr in permethrin-resistant head lice is essential in both monitoring and population genetic issues, such as the origin or evolution of resistance in geographically distinct head louse populations.

We have previously used the serial invasive signal amplification reaction (SISAR) as a high-throughput genotyping method to detect the T929I and L932F mutations responsible for permethrin resistance in individual lice (Kim et al. 2004). Most recently, fluorescence-based assays such as real-time TaqMan assays and high-resolution melt analysis have been developed to detect kdr alleles in Anopheles gambiae Giles (Bass et al. 2007). These individual genotyping techniques are very accurate and efficient in obtaining detailed information on resistance allele frequency and allelic zygosity. Such genotyping techniques, which are based on obtaining individual genomic DNA samples from ethanol-stored or frozen louse specimens, are particularly useful for resistance monitoring of head lice because obtaining a sufficient number of live lice for bioassay is difficult. They require, however, a large number of samples to ensure accurate estimation of resistance allele frequency, which in many instances reduces its practicality as a routine resistance monitoring tool. Moreover, if the size of specimen is very small, as for lice, individual genomic DNA extraction and subsequent analysis are laborious and costly. Population genotyping methods based on pooled DNA samples (Sham et al. 2002), such as real-time polymerase chain reaction (PCR) amplification of specific allele (rtPASA) (Germer et al. 2000), real-time PCR with allele-specific TaqMan probes (rtPASA-TaqMan) (Livak 1999), and quantitative sequencing (QS) (Amos et al. 2000), have been developed to reduce the cost and time associated with the determination of the frequencies of single nucleotide polymorphisms (SNPs). These methods have been used widely in the determination of the allele frequencies of SNPs associated with human genetic disease (Dasgupta et al. 2003, Rollinson et al. 2004) and insecticide resistance (Kwon et al. 2004). Among these methods, QS was evaluated as the simplest method to use and was found to be highly accurate in population genotyping (Wilkening et al. 2005).

In this work, we have evaluated the potential of QS as a high-throughput population genotyping method to detect the three mutations in the human head louse VSSC gene responsible for permethrin resistance. It is envisioned that such methods will be used as the initial resistance monitoring tool for the routine detection of permethrin resistance in field populations of lice.

Materials and Methods

Head Lice Rearing and Specimen Collection.

Permethrin-resistant (FL-HL), moderately susceptible (PA-HL), and susceptible (EC-HL) strains of head lice have been maintained using the in vitro membrane feeding system as described previously (Yoon et al. 2006). Head louse samples were collected from various geographic locations, immersed in 95% ethanol immediately, and shipped to our laboratory. Louse specimens from different patients were combined and stored at -20°C in 95% ethanol until use.

Development of QS Protocol for Detection of Resistance Mutations.

Genomic DNA from individual lice (FL-HL or EC-HL strains) was extracted by DNeasy tissue kit (QIAGEN, Valencia, CA) according to manufacturer’s instructions. In brief, an individual louse sample was homogenized with 100 μl of ATL buffer and treated by proteinase K at 55°C for 3 h. The homogenate was mixed with 100 μl of AL buffer and 100 μl of 96% ethanol. The mixture was centrifuged through the spin column and genomic DNA eluted by Tris-EDTA buffer.

A 908-bp genomic DNA fragment from the head louse VSSC α-subunit gene, which encompassed the three mutation sites (M815I, T917I, and L920 F), was PCR-amplified from individual genomic DNAs obtained from both FL-HL and EC-HL strains using sequence-specific primer sets (5′HL-QS: 5′-ATTTTGCGTTTGGGACTGCTGTT; 3′-HL-QS: 5′CCATCTGGGAAGTTCTTTATCCA) with 35 cycles of 95°C/30 s, 63°C/30 s, 68°C/1min 30 s. Ten individual PCR products (eight EC-HL and two FL-HL samples) were sequenced to confirm the genotypes at each mutation site and to identify any sequence polymorphisms in the intron regions. PCR products were purified using a PCR purification kit (QIAquik, QIAGEN) and quantified by using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE). The standard DNA templates were mixed in the following molar ratios: 0:10, 1:9, 3:7, 5:5, 7:3, 9:1, and 10:0 (resistant allele:susceptible allele at each mutation site). Standard DNA template mixtures (10 ng) were sequenced with an ABI Prism 3730 DNA sequence analyzer (Applied Biosystems, Foster City, CA) by using Big Dye Terminator cycle sequencing kit (Applied Biosystems) and two sets of sequencing primers (5′QSMI, 5′-TGTGGCCTTACTTGTATTCGA and 3′QSMI, 5′-CCCCCCGCATTAAAATTAAAT for the sense- and antisense-directional sequencing of M815I mutation, respectively; 5′QSTILF, 5′-AAATCGTGGCCAACGTTAAA and 3′QSTILF, 5′-TTACCCGTGTAATTTTTTCCA for the sense- and antisense-directional sequencing of T917I and L920F mutations, respectively) (NICEM Sequencing Facility, Seoul National University, Seoul, Korea). The nucleotide signal intensities of the resistant and susceptible alleles at each mutation site were measured from the sequence chromatogram by using Chromas version 2.31 software (Technelysium Pty Ltd., Tewantin, Australia) and the signal ratios [resistant nucleotide signal/(resistant nucleotide signal + susceptible nucleotide signal)] were calculated. Five individual DNA templates obtained from the lice that had been previously determined heterozygous were also sequenced as an internal reference to the 5:5 standard DNA template. The signal ratios of template DNA mixtures were normalized by multiplying them with the normalization factor (signal ratio of the heterozygous DNA template/signal ratio of the 5:5 standard DNA template). The series of normalized signal ratios were plotted against the corresponding resistance allele frequencies, and standard regression equations together with lower and upper prediction equations were generated using the SigmaPlot version 10.0 (Systat Software Inc., San Jose, CA) for the estimation of resistance allele frequencies of unknown samples and their prediction intervals at the 95% confidence level.

Validation of QS as a Predictor of Resistance Allele Frequency.

Three louse populations (Pop A, B, and C), with different resistance allele frequencies, were artificially prepared by mixing the ethanol-stored FL-HL and PA-HL lice randomly in different ratios (Pop A, 19 PA-HL females + one FL-HL female; Pop B, 17 PA-HL females + three FL-HL females; and Pop C, seven PA-HL females + three FL-HL females) and used in the following validation experiments. Ethanol-stored female louse specimens were cut into two parts (head-thorax and abdomen). The head-thorax pieces were combined and processed together by extracting genomic DNA with 400 μl of the ATL buffer and designated as "pooled specimen DNA." The abdomens were individually processed by extracting genomic DNA with 100 μl of the ATL buffer. Equal amounts (22 ng) of the individual genomic DNA samples from each of the three populations were combined to prepare "pooled DNA." The 908-bp genomic DNA fragments were PCR amplified from the pooled specimen DNA, "pooled DNA." and each of the individual DNA samples as described above. The PCR products from all 50 individual lice DNA samples (20 Pop A, 20 Pop B, and 10 Pop C DNA samples originated from 43 PA-HL and seven FL-HL lice) were directly sequenced to determine actual genotypes at the three mutation sites. Both of the pooled specimen DNA and pooled DNA were processed for QS as described above, and signal ratios at each mutation site were obtained from sequence chromatogram. The resistance allele frequencies were predicted by substituting signal ratios into the standard regression equations generated from the sense-directional sequencing (Table 1).

Table 1

Quadratic regression and prediction equations of the resistant nucleotide signal ratio versus corresponding resistant allele frequency at the M815I, T917I, and L920F mutation sites

To investigate whether DNA preparation from different developmental stages can affect the performance of QS, three different stages (second and third instars, male and female adults) of ethanol-stored FL-HL lice were sorted and used for the preparation of template DNAs for QS.

Evaluation of Resistance Allele Frequencies of Field-Collected Louse Specimens.

Louse specimens (≈14-52 individuals per population), at the same developmental stage (first instar, third instar, male or female adults), were sorted from ethanol-stored louse stocks and combined. Genomic DNA was extracted from the combined specimens with DNeasy tissue kit as described previously. The 908-bp genomic DNA fragments were amplified, quantified, and analyzed by sense-directional QS as described previously. Resistance allele frequencies were predicted by using the standard regression equations (Table 1).

Results and Discussion

Exon–Intron Structure of the 908-bp Genomic DNA Fragment.

The exon–intron structure of the 908-bp genomic DNA fragment from the VSSC α-subunit gene that contains the three kdr mutations is illustrated in Fig. 1. In total, three exons and four introns were found. The M815I mutation was located in the 143-bp exon 1, whereas both the T917I and L920 F mutations were located in the 164-bp exon 3. Sequence comparison of 10 individual DNA fragments revealed that there were extensive insertion/deletion sequence polymorphisms in the introns, which resulted in size variations of ≈77-85, ≈84-86, ≈86-88, and ≈83-85 bp in introns 1, 2, 3, and 4, respectively (sequence data not shown). Such extensive intron variation seems unique to the head louse because only low levels of intron variations were observed at the same VSSC region from other insects, including the diamondback moth, Plutella xylostella (L.), and the malaria mosquito, Anopheles sinensis Wiedemann (D.H.K. and S.H.L., unpublished data). In contrast, no sequence polymorphisms, except for those at the three mutation sites, were found in the three exon regions. Sequence analysis of 10 individual DNA clones revealed that all three of the resistance mutations occurred together in FL-HL, whereas no mutations were found in EC-HL. The alternative exon c/d that is widely conserved in this fragment in most insects (Davies et al. 2007) was not found in the head louse. Because of the extensive size and sequence variations in the introns, sequence primers for QS were designed within the same exons that contain the mutation sites of interest or from a conserved intron region (Fig. 1). From this finding, it is imperative that exon-intron polymorphisms in the target genomic DNA region be inspected before designing the sequencing primers for QS.

Fig. 1

Exon–intron structure of the 908-bp head louse VSSC genomic region that contains the M815I, T917I, and L920 F mutations. Shaded boxes and solid lines indicate exons and introns, respectively. Locations of three resistance mutations are marked with black circles. Vertical arrows indicate the approximate locations of intron polymorphisms. Horizontal arrows indicate the locations of the QS primers.

Relationship between Nucleotide Signal Intensity and Allele Frequency for the Prediction of Resistance.

When a set of standard DNA mixtures with different ratios of resistant and susceptible alleles were sequenced, the signal intensity of each resistant allele increased as the resistance allele frequency increased (Fig. 2). At the 5:5 standard DNA template mixture, the signal intensities of the resistant alleles were not equal to those of susceptible alleles, but they were lower at all three mutation sites, resulting in signal ratios that ranged from 0.279 to 0.469, depending on the nature of nucleotide substitution at the mutation site and the sequencing direction. When the DNA templates from individual heterozygous lice (internal references) were sequenced, signal ratios (0.239–0.405) were similar to those obtained from the 5:5 standard DNA template mixture (data not shown), confirming that the unequal signal intensities at the mutation sites are in fact due to the different efficiencies in the dideoxy nucleotide terminator incorporation between the resistant and susceptible nucleotides in the sequencing reaction (Korch and Drabkin 1999). Nevertheless, the slight difference in the signal ratio between the 5:5 standard DNA mixture and heterozygous DNA templates substantiates the necessity for normalization of signal ratios to guard against any biases in the sequence signal intensity.

Fig. 2

Sequencing chromatograms of the standard template DNA mixtures with different ratios of resistant and susceptible alleles at the M815I, T917I, and L920F mutation sites. The numbers on the top of each column indicate the resistance allele ratios at each mutation site. The relative intensities of the resistance allele signals are indicated with arrows.

Nucelotide signal ratios were plotted against the corresponding resistance allele frequencies at each mutation site and fitted to quadratic equations (Fig. 3). All the resulting regression curves showed high correlation coefficients (r² = 0.9928≈0.9997), demonstrating that the nucleotide signal ratio is highly proportional to the resistance allele frequencies (Table 1). The correlation coefficients varied slightly, however, depending on the sequencing direction, and resulted in different prediction equations at each mutation site (Fig. 3; Table 1). Sequencing of the M815I and T917I mutations in the sense direction generated prediction equations that had considerably narrower intervals (Fig. 3A and B), whereas the prediction equations generated from the L920F sequencing were almost identical regardless of sequencing direction (Fig. 3C and F). Considering both higher regression coefficients and narrower prediction intervals, regression equations generated from the sense-directional sequencing were chosen for the prediction of all three mutation frequencies, with the lower and upper prediction equations used for the determination of prediction intervals at the 95% confidence level (Table 1).

Fig. 3

Resistance sequence signal ratios obtained from the sense (A, B, and C)- and antisense (D, E, and F)-directional sequencing were plotted with corresponding resistance allele frequencies at the M815I, T917F, and L920F mutation sites. Quadratic regression lines are indicated by solid black lines with the upper and lower 95% prediction lines indicated by dotted red lines. Nucleotide signal ratio (x-axis) was calculated as [resistant nucleotide signal/(resistant nucleotide signal + susceptible nucleotide signal)].

Using the lower and upper 95% prediction equations (Table 1), the M815I mutation allele frequencies could be accurately predicted within the range of 9.7≈90.9% at the 95% confidence level. Likewise, the detection limit for the T917I mutation allele frequency was determined as low as 8.0% and as high as 91.9 at the 95% confidence level. The detection range for the L920F mutation allele frequency was determined as 4.4≈94.4% at the 95% confidence level. In summary, the lower detection limits for the three resistance mutations were 4.4≈9.7% (7.4 ± 2.7%), suggesting that QS can be used as a preliminary resistance monitoring tool for the detection of resistance allele frequencies higher than ≈7.4% at the 95% confidence level.

The plot properties as determined in Fig. 3 remained statistically unchanged when the DNA template quantity for QS was doubled (20 ng) or halved (5 ng) (data not shown), indicating that template amount is not a critical factor in QS performance as long as nucleotide signal ratio rather than the absolute signal intensity is used as the predicting parameter. The low impact of template DNA quantity on the performance of QS makes it a more robust technique compared with the rtPASA protocol, where accurate template quantification is a prerequisite for the successful prediction of allele frequency (Kwon et al. 2004). Additionally, virtually the same regression equations were repetitively generated from the same set of standard DNA templates as long as an identical sequencing instrument is used for QS (data not shown). Once established, therefore, a set of standard regression and prediction equations can be repetitively used for the prediction of allele frequencies of unknown samples without sequencing the standard DNA templates each time.

Validation of QS.

Sequencing 50 individual DNA samples (43 PA-HL and 7 FL-HL), which were used for preparing three populations (Pop A, B, and C), revealed that the allele zygosities at each of the three mutation sites were the same in a single individual (Table 2). For example, if an individual was homozygous susceptible at one mutation site, it was also homozygous susceptible at the other two mutation sites, etc. These results confirmed our previous findings that all three mutations exist en bloc, giving rise to only two haplotypes, one resistant possessing all three mutations and the other susceptible lacking the three mutations (Lee et al. 2003). Individual genotyping of 43 PA-HL revealed that a large number of heterozygotes (21), followed by homozygous susceptible (15) and resistant (7) individuals, are present in the population, resulting in a relatively high resistance allele frequency (40.7%) (Table 2). In contrast, the majority of FL-HL were homozygous resistant individuals (six of seven lice) followed by a small number of heterozygous (one of seven lice) but no susceptible homozygous individuals (Table 2). Because heterozygous individuals are phenotypically susceptible due to the recessive trait of the kdr factor (Yoon et al. 2003), the presence of a large number of heterozygous individuals, as in the case of PA-HL louse population, would not be detected by a traditional bioassay, highlighting the importance of DNA-based detection of resistance allele frequency and zygosity.

Table 2

Allele zygosity and frequency at the M815I, T917I and L920F mutation sites in three lice populations (Pop A, B and C) determined by sequencing

When the Pop A, B and C populations were prepared by randomly mixing lice from the PA-HL and FL-HL strains in different ratios and genotyped by sequencing, their actual resistance allele frequencies were calculated as 37.5, 60.0, and 45.0%, respectively (Table 2). When the pooled DNA samples from these three populations were analyzed by QS, the mean resistance allele frequencies of Pop A, B, and C were predicted as 39.8, 59.1, and 44.1%, respectively (Table 3). The actual resistance allele frequency of each population was included within the 95% prediction interval generated by QS, demonstrating the accuracy of QS to predict allele frequency and resistance (Table 3). When the pooled specimen DNA was analyzed by QS, the mean predicted allele frequencies of Pop A, B, and C were 39.6, 62.6, and 44.6%, respectively, with the 95% prediction interval of each estimate overlapping with the actual allele frequency. There were no significant differences in predicted allele frequencies between the pooled DNA and pooled specimen DNA samples as judged by the overlapping 95% prediction intervals (Table 3). This finding demonstrates that QS, using either pooled DNA or pooled specimen DNA, is as accurate and reliable as individual genotyping in the prediction of resistance allele frequency.

Table 3

Comparison of resistance allele frequencies at the M815I, T917I, and L920F mutation sites predicted by sense-directional QS of the pooled DNA and pooled specimen DNA from three head louse populations

The actual definition of DNA pooling is to combine equimolar amounts of individual DNAs from multiple specimens. If it is assumed, however, that the DNA content in a single specimen (size of louse specimens is similar) and DNA quality (louse specimens collected and stored in the same conditions) are identical, DNA samples extracted from combined multiple specimens as in the case of pooled specimen DNA should represent the same profile as pooled DNA. In our study, there was no significant difference in the resistance allele frequencies determined from the pooled DNA and pooled specimen DNA as described above. This finding validates our approach of using the pooled specimen DNA (DNA samples extracted from multiple louse specimens) as the DNA template for population genotyping by QS. The use of DNA extracted from combined multiple specimens for QS greatly reduces the overall cost and effort as repetitive DNA extraction from individual specimens, particularly from small specimens such as lice, is laborious and expensive.

To investigate whether different developmental stages (size) of lice can affect QS results, resistance allele frequencies were estimated using genomic DNA samples extracted from second and third instars and adult (female and male) stages. There were no significant differences in the predicted resistance allele frequencies at any of the three mutation sites among different developmental stages as each prediction was within the 95% prediction interval (Table 4). This result suggests that any developmental stage of lice can be used for QS as long as louse specimens in the same developmental stage (i.e., same size) are used. Pooled specimen DNA prepared from the louse specimens in mixed stage (i.e., different size) is not adequate for QS because the quantity of DNA from each individual would vary, resulting in misrepresentation of resistance allele frequency in the combined individuals.

Table 4

Predicted resistance allele frequencies of lice from different global locations at the M815I, T917I, and L920F mutation sites using the QS population genotyping method

The virtually identical allele frequencies at the three mutation sites again substantiates that the three mutations exist together in a resistant haplotype in the head louse. This finding suggests that prediction of resistance allele frequency can be conducted with any of the three mutations. Considering that the T917I mutation alone renders the sodium channel practically insensitive to the action of permethrin, thereby playing a critical role in permethrin resistance (Yoon et al. 2008), the determination of the T917I mutation allele frequency alone seems appropriate as a representative indicator for permethrin resistance in head louse populations.

Population Genotyping by QS of Several Louse Populations from Various Global Locations.

Resistance allele frequencies varied substantially in lice from different global regions (Table 4). Resistance allele frequencies are completely saturated or near saturation in lice from Onkaparinga (Australia), Bristol (United Kingdom), Bobigny (France), and South Florida (United States) as judged by signal ratios of one and resulting predicted frequencies near 100%. Lice from California and Texas seem to be moderately resistant to pyrethroid as judged by their mean resistance allele frequencies (86.2 and 59.2%, respectively). The predicted resistance allele frequency was 3.1% in lice from Ecuador and 1.0% in lice from Guatemala and Thailand. These findings support our original contention that permethrin resistance in head lice is currently widespread but not uniform (Gao et al. 2003). Resistance alleles seem more prevalent in developed countries, which may be due to the more extensive use of permethrin-containing pediculicides in these countries.

In regions where resistance allele frequency is saturated or near saturation, pyrethroids use should be curtailed and alternative pediculicides with different mode of actions used instead. In regions where the resistance allele frequencies are low or near zero, pyrethroids should be used cautiously and in conjunction with a resistance monitoring program. This approach will extend the effective life span for this valuable group of pediculicides.

Detection of the early phase of resistance is crucial to long-term, efficient resistance management but is very difficult using conventional bioassay-based monitoring methods, particularly when resistance is recessive. More specifically, collecting large numbers of lice is often impossible and always difficult. Because the reliable lower detection limit for resistance allele frequency prediction by QS is ≈7.4% (roughly equivalent to the population with one homozygous resistant individual out of 14 louse specimens or one heterozygous resistant individual out of seven louse specimens), QS with small- to medium-sized populations (for example, 14–28 louse specimens per population) seems adequate and a practical approach for routine population genotyping. To this end, population genotyping by QS will greatly facilitate resistance-monitoring efforts in field populations of lice. If more precise determination of resistance allele frequency below the QS detection limit is required on a population basis, rtPASA can be used as a supporting monitoring step (Kwon et al. 2004). This method enabled the detection of the kdr allele frequency in the diamondback moth at levels as low as 0.02%. If the information on resistance allele zygosity as well as allele frequency in a population is required, more accurate individual genotyping methods such as SISAR (Kim et al. 2004) can be conducted as a secondary resistance monitoring tool on a much reduced number of populations. Although the proposed tire of molecular resistance monitoring tools including QS only detect the permethrin resistance mediated by kdr, considering that kdr is a major factor in all permethrin-resistant lice worldwide (Lee et al. 2000), they should be useful for screening a large number of wild louse populations as alternatives to conventional bioassay.

In conclusion, QS-based population genotyping can process a large number of louse populations simultaneously for the evaluation of resistance allele frequencies. It is expected that QS for 96 different population samples can be completed within 2 d in moderately equipped laboratories (1 d for genomic DNA extraction, 3 h for PCR to generate DNA templates for QS, and 5 h for sequencing). The technique dependency of QS is also relatively low compared with other population genotyping techniques such as rtPASA-TaqMan (Livak 1999) and rtPASA (Germer et al. 2000). Thus, the speed, simplicity, and accuracy of QS make it an ideal candidate for a routine primary resistance monitoring tool to screen a large number of wild louse populations as an alternative to conventional bioassay. These benefits are particularly true for human lice because obtaining a sufficient number of live individuals and maintaining them alive for bioassay is usually limited and impractical. The QS-based population genotyping protocol should be readily applicable for other insects, including other human louse species, as a routine resistance monitoring tool as long as the information on resistance-associated mutations in any target genes is available.

Acknowledgements

We thank to Drs. Arezki Izri and Rémy Durand for providing the French lice samples. We also thank to Jee Sun Min for preparing individual lice genomic DNA samples. This study was supported by National Institutes of health grant 5 R01 AI045082-05. D.H.K. was supported in part by Brain Korea 21 program.