Real time PCR detection of common CYP2D6 genetic variants and its application in a Karen population study

Background Plasmodium vivax malaria is characterized by relapses arising from the hypnozoite stages in the liver. The only currently registered drug for radical treatment to prevent relapse is primaquine. Primaquine, a prodrug, requires metabolism through the liver cytochrome CYP2D6 isoenzyme to its active metabolite. Mutations in the CYP2D6 gene may thus affect primaquine efficacy. A SNPs genotyping technique was developed to characterize the CYP2D6 genetic variants and tested this in the patients with Plasmodium vivax infection collected in a Karen population on the Thailand–Myanmar border, where P. vivax malaria is endemic. Methods Direct sequencing of PCR-reamplified products (DSP) was used to uncover exonic CYP2D6 sequence variations. Subsequently, an allele-specific oligonucleotide probe real-time SNPs genotyping (ASO) assay was developed for rapid detection of the four clinically relevant CYP2D6 variants occurring in this population. These two in-house developed assays were used to genotype CYP2D6 mutations in blood samples obtained from 70 Karen adults. Results Results showed a high degree of concordance between the DSP and ASO methods. Six CYP2D6 point mutations were identified within the Karen population: C100T, C1039T, G1661C, G1846A, C2850T and G4180C, at frequencies of 0.43, 0.43, 0.76, 0.02, 0.32 and 0.76, respectively. The CYP2D6*2, *4, *5, *10 and *36 allelic frequencies were 0.33, 0.02, 0.03, 0.40 and 0.01, respectively. Alleles conferring an intermediate CYP2D6 metabolizer phenotype comprised 46% of the total number of alleles. Conclusion The newly developed ASO assay is a reliable and rapid tool for large-scale CYP2D6 genotyping. The high frequency of the CYP2D6*10 allele in the Karen population warrants further assessment of its association with the radical curative efficacy of primaquine. Electronic supplementary material The online version of this article (10.1186/s12936-018-2579-8) contains supplementary material, which is available to authorized users.

mutations in CYP2D6 can potentially affect primaquine efficacy [12,13]. Accurate genotyping of the CYP2D6 gene is difficult because of the presence of highly homologous pseudogenes and the highly polymorphic character of the gene. The CYP2D8 and CYP2D7 flanking pseudogenes display over 90% nucleotide sequence homology compared to the active CYP2D6 gene [14,15], potentially resulting in co-amplification during the gene amplification process [16,17]. Consequently, multiple homologous PCR templates and mis-primed sequences containing several inactive mutations can easily result in incorrect genotype assignments [14,18,19]. In addition, there is wide genetic variability in CYP2D6, including single nucleotide substitutions, insertion/deletion, partial gene conversions [20], CYP2D7/2D6 hybrid tandems [21], copy number variations (CNVs) [16,22,23] and complex structural rearrangements [17,24,25]. This has resulted in the characterization of over one-hundred CYP2D6 variant alleles, which further complicates genotyping. Furthermore, the prevalence of allelic variants associated with impaired CYP2D6 catalytic activity have been found to vary widely across ethnic populations [26][27][28]. In the present study, a CYP2D6 genotyping protocol was developed to screen for variants with known clinical significance in Southeast Asia and tested the method on P. vivax infection samples collected from a Karen population on the Thailand-Myanmar border.

Study population and DNA preparation
Seventy anonymized blood samples were collected in the Karen population living in Tak, one of the western provinces in Thailand. All patients were diagnosed by microscopy of thick and thin blood smears, examined by qualified laboratory technicians. Only patients with P. vivax mono-infections were included in the study. Among vivax patients, 39 (56%) were females, the median age was 16 years old (IQR 11-24 years old), the median weight was 40 kg (IQR 18-50 kg), and the median parasitaemia was 4024 parasites/µl (IQR 1592-10,793 parasites/µl). Genomic DNA extraction was by the QIAamp DNA Mini Kit (Qiagen, Germany), according to the manufacturer's guidelines. The isolated genomic DNA samples were stored at 4 °C until further processing. Allele designation followed the Human Cytochrome P450 (CYP) Allele Nomenclature Database (http://www. imm.ki.se/CYPal leles /) ( Table 1). The study was approved by the Ethics Committee of the Faculty of Tropical Medicine, Mahidol University (EC Submission No.: TMEC 15-095).

Primers and probes
Two genomic sequences retrieved from NCBI, CYP2D6 (GeneBank accession number M33388.1) and CYP2D7/8 (GeneBank accession number M33387.1), were used as genomic reference templates. Optimal primer sequences of PCR-reamplified products were selected by Primer3 software version 0.4.0 and were then synthesized by Macrogen Inc. (Korea). Variant-specific primer and probe sets of real-time PCR were designed and supplied directly by Applied Biosystems (Thermal Fisher Scientific, Inc.). Wild-type and mutant CYP2D6 probes were labeled at the 5′ end with VIC and FAM respectively, and both probes included a non-fluorescent quencher. The sequences of primers and probes of all PCRs performed in each step are listed in Table 2. Positive controls are well-characterized samples sequenced across the entire CYP2D6 gene and redetected by real-time PCR assay. A summary of the genotyping method is provided in Fig. 1 and can be summarized in the following four steps.

Determination of CYP2D6 gene duplications/ multiplications and gene deletions by extra-long range polymerase chain reaction (XL-PCR)
In a single tetra-primer XL-PCR mixture, two separate reactions including the CYP2D6 gene duplications/multiplications and deletions, both producing a 3.5 kb long PCR fragment, were determined simultaneously with P34S, S486T -the amplification of a 5.1 kb fragment encompassing the entire CYP2D6 gene. The generated CYP2D6 fragments were then used as templates for genotyping as described previously [29].

Discrimination of functional CYP2D6 and nonfunctional CYP2D8 and CYP2D7 genes by intron 2 sequencing (INT2)
In each XL-PCR run, the newly produced fragments were randomly sampled to be reamplified and sequenced in the

Detection of CYP2D6 coding region mutations by direct sequencing of PCR-reamplified products (DSP)
The isolated CYP2D6 gene, which consists of nine exons, was PCR-reamplified by nested DSP primers. The nested DSP reactions were conducted on a Mastercycler-pro (Eppendorf, Hamburg, Germany

CYP2D6 genotyping assay
Successfully produced entire CYP2D6 fragments (XL-PCR) served as templates in the assay (Fig. 2). Full-length intron 2 sequencing (INT2), which primers anneals internally to the XL-PCR products, distinguished between CYP2D7/8 pseudogenes and the functional CYP2D6 gene. All nucleotide sequences of the intron 2 region showed more than 95% sequence similarity to the CYP2D6 gene, indicating that the desired templates were generated correctly (see Additional file 1). Paired forward-reverse reads of 9 exons in each individual sample demonstrated that the sense CYP2D6-specific strands were perfectly matched baseby-base to their antisense strands, implying that the

CYP2D6 variability in the Karen population
CYP2D6-specific XL-PCR products were successfully amplified in all seventy patients, of which 2 (3%) carried multiple active functional CYP2D6 genes and 4 Fig. 2 Overview of CYP2D locus arrangements and typical recombinant events. a A single CYP2D6 gene is located downstream of the CYP2D locus, with two pseudogenes CYP2D8/2D7 flanking at the 5′ end of the CYP2D6 gene. The position of each forward and reverse pair for amplifying the target fragments are represented by different colored arrows. b In the CYP2D6*5 deletion allele, the entire CYP2D6 is deleted from the CYP2D locus producing the new hybrid (REP DEL) containing 5′ yellow box sequence and 3′ green box sequence. c The general duplication arrangement, the hybrid REP DUP containing 5′ green box sequence and 3′ yellow box sequence were formed at 5′ of the CYP2D6 gene in case of multiple CYP2D6 gene copies occur d the 5.1 kb XL-PCR products encompass the entire CYP2D6 gene consisting of 9 exons with the four known clinically relevant point mutations located at different polymorphic sites. The 3.5 kb long fragment represents either duplications/multiplication or deletion (6%) a whole CYP2D6 gene deletion (Fig. 2). Combining all direct sequencing results of full-length CYP2D6 coding regions, at least one mutation was detected in all exons except in exon 4, 5, 7 and 8. Three non-synonymous substitutions (C100T: Exon1, C2850T: Exon6 and G4180C: Exon9), two synonymous substitutions (C1039T: Exon2 and G1661C: Exon3) and a mutation at the splice junction (G1846A) were detected. Frequencies of each allele and genotype are summarized in Additional file 3. Of the six mutations identified, SNPs at loci G1661C and G4180C were most frequent (0. In contrast, genotype frequencies of the splice site defect 1846G/A 3 (0.04) and 1846A/A 0 (0.00) were very low. CYP2D7 exon 9 conversion, assessed through entire exon 9 sequencing and associated with decreased enzymatic activity, was observed in only one individual (see Additional file 4). Data generated from the DSP/ASO genotyping assays, the 6 distinct alleles and 11 genotype frequencies, are summarized in Table 3. Other alleles and genotypes described in the literature were not observed in this Karen study population.

Comparative analysis of CYP2D6 variants in different populations
The genotyping results was compared to the list of global genetic mutations in CYP2D6 from the 1000 Genome Project (http://www.ensem bl.org) (Fig. 3).
Although this study was conducted in P. vivax infected patients, the genotypic pattern observed in the Karen study population was similar to other patterns reported from East Asia, including China, Japan and Vietnam. The exception was C2850T, which showed a notably higher frequency in this study population ( Table 4). As the geographical distance increased from the study area, including Bangladesh, India, Pakistan and Sri Lanka, the similarity between genotypes decreased. Similarity in the distributions of genotype frequencies was further reduced when comparing populations in Africa, America and Europe ( Fig. 3 and Additional file 5).

Discussion
Primaquine is required for preventing P. vivax malaria relapses [1]. The radical curative efficacy of primaquine is thought to be mainly dependent on CYP2D6-mediated metabolism [12]. In addition to primaquine, its long half-life analogue tafenoquine has recently been developed for radical cure of P. vivax malaria. However, there are somewhat conflicting results whether differences in CYP2D6 metabolism confers differences in therapeutic efficacy. A clinical study showed that tafenoquine efficacy in P. vivax-infected patients was not affected by the changes in CYP2D6 activity [30]. In contrast, tafenoquine pharmacokinetic profiles in CYP2D knockout mice were differed significantly from those of wild-type mice, suggesting that tafenoquine could be possibly affected by the CYP2D metabolism [31]. Further studies will be needed to assess the importance of CYP2D6 mutations in tafenoquine biological activity. Since the CYP2D6 gene is highly polymorphic with hundreds of variant alleles described, it potentially affects a large range of clinically used drugs metabolized by the encoded enzyme [32]. Related to this, there is a growing interest for the development of user-friendly CYP2D6 genotyping platforms with sufficiently high throughput to characterize clinically relevant genetic variations in the CYP2D6 gene. Techniques previously described include restriction fragment length polymorphism (PCR-RFLP) [33,34], single strand conformation polymorphism (SSCP) [35,36], multiplex allele-specific PCR (Multiplex PCR) [37,38], and allelespecific oligonucleotide hybridization (PCR-ASO) [39]. However, these techniques are mostly laborious to execute, time-consuming, error-prone and characterize only a limited number of alleles. Real-time PCR-based strategies enable detection of a larger number of mutations and are more rapid, but often deploy in-house developed primers with limited specificity, resulting in unwanted co-amplification of pseudogenes [40][41][42]. More recent techniques including pyrosequencing [43], denaturing high-performance liquid chromatography (DHPLC) [44] and Luminex-xTag [45] perform better with shorter run-times, but all require highly advanced equipment often not available in malaria endemic countries. Highthroughput microarray technology, such as GeneChip CYP450, Amplichip CYP450 and the DMET microarray [46][47][48], have excellent performance and allele coverage, but is also technically difficult and costly.
In the current study, several steps were taken to increase performance of the assay: Firstly, the problem of co-amplification of CYP2D with high sequence similarity was overcome by using CYP2D6-specific amplification primers (XL-PCR) and a nested PCR approach (INT2). Secondly, introduction of direct sequencing of PCR-reamplified products (DSP) lowered the chance of missing non-targeted variations within the exonic CYP2D6 sequences. Thirdly, addition of the ASO assay enabled rapid identification of four allelic sites observed in the Karen study population, requiring approximately 90 min for parallel identification. Overall, our customizable ASO assay showed high accuracy with very high SNP call rates for each genotype and absence of contamination errors. Moreover, it yields high-intensity fluorescent signals and clearly separating allelic clusters. Reproducibility of the assay was not formally assessed, since this would involve in the results generated by interlaboratory tests. However, the DSP and ASO assays were repeatedly genotyped, at least 2 times on different days, to assure the repeatability of the assays, and were analysed simultaneously with known CYP2D6 genotypes. The replications of CYP2D6 genotyping results were in 100% concordance. In order to rule out contamination, negative samples, consisting of nuclease free water, were evaluated in parallel within a single analytical run together with the patients' samples. To reduce the chance of cross-reactivity in the experiments, 3 approaches were applied (a) all generated CYP2D6 templates were assayed by INT2 to confirm absence of cross-reactivity with the pseudogenes, (b) to minimize unintended binding to the pseudogenes, allele-specific primer and probe sets were designed using the public databases (NCBI and dbSNP), and (c) thermal cycling was optimized to reduce  A limitation of the assay is the interpretation of results in heterozygous individuals with gene multiplications; the current method does not assess specific gene allele duplication or quantify the copy number for each allele. However, whereas gene amplification might result in increased production of the active metabolite of primaquine potentially increasing its efficacy, reduced activity of the drug is mainly associated with gene mutations resulting in reduced CYP2D6 activity. The latter was the scope of the study. Initial use of the assay showed that the CYP2D6 allelic frequencies in this vivax-infected population corresponded well with the observed frequencies in nearby Asian populations, but were different from frequencies in Caucasian and African populations. This similarity suggests absence of selective pressure on the CYP2D6 genotype in this P. vivax infected population. The most abundant allele in the Karen population was the CYP2D6*10 allele, occurred at a frequency of 0.40, suggested that there is a high prevalence of individuals with reduced metabolic capacity for CYP2D6 dependent substrates. Whereas the CYP2D6*4 and *5 defective alleles, occurred at a frequency of 0.02 and 0.03 respectively, represents rare causes of reduced enzyme activity in this population. The CYP2D6*2 allele with a frequency of 0.33 was the most common functional allele in the Karen population, with high frequency of the 2850C/T and 2850T/T alleles. The number of CYP2D7 exon 9 conversion carriers (CYP2D6*36) was very small in the study, and a larger study is warranted to assess the frequency of this clinically important genotype in more detail. Since CYP2D6 allele frequencies vary markedly across ethnic populations, the ASO assay would have to be re-evaluated for other geographical areas. Additional file 1 shows the differences in CYP2D6 allele frequencies present in other areas including Africa, America, Europe and South Asian. The DSP approach providing the sequence the CYP2D6 region of the population of interest could inform which adaptations in a customized ASO assay would be necessary. Further studies are planned on CYP2D6 mutations in the Karen population using the ASO genotyping platform. Results will be compared to the data on the efficacy of primaquine in the study cohort to inform if efficacy proves to be compromised by CYP2D6 mutations associated with decreased CYP2D6 enzyme activity. Indeed, this is not a point of care test, but it is a simple method, which is easy to set up in molecular laboratories in tropical countries. Defining the impact of CYP2D6 mutations on primaquine dosing and efficacy will require a clinical trial, which can make use of the platform described here. The prevalence of mutation associated with defective CYP2D6 phenotypes in the Karen population has prompted our group to initiate such a study. Elimination of P. vivax will require wider deployment of radical cure with primaquine in an effective dose. This effective dose might differ according to the prevalent CYP2D6 mutations in the population. This relationship should be studied more extensively in different populations to ensure proper dosing. The described technique could facilitate this.

Conclusion
The ASO assay is a new CYP2D6 genotyping assay with high-accuracy and high-reproducibility for the detection of common CYP2D6 variant alleles, and is suitable for large-scale surveys. The high prevalence in the P. vivax infected patients in Karen population of the CYP2D6*10 allelic variant associated with reduced CYP2D6 enzyme activity could potentially affect the radical curative efficacy of primaquine and warrants more extensive evaluation.

Additional files
Additional file 1: Figure S1. Discrimination of functional and nonfunctional genes using intron 2 sequencing. The multiple sequence alignment of Intron 2 region (1.1 kb) of each CYP2D gene compared among individual's intron 2 sequence.
Additional file 2: Figure S2. Comparison of the two genotyping platforms. DSP (electropherograms) and ASO (amplification plot) detected the presence of four common genetic variations in the CYP2D6 gene including C100T, G1846A, C2850T and G4180C. The X axis of the amplification plot shows the relative fluorescence for wild-type alleles (green curve) and variant alleles (blue curve), respectively. Clusters of the homozygous wildtype, heterozygous and homozygous mutant are also shown in the allelic