Multi-ethnic cytochrome-P450 copy number profiling: novel pharmacogenetic alleles and mechanism of copy number variation formation.

To determine the role of CYP450 copy number variation (CNV) beyond CYP2D6, 11 CYP450 genes were interrogated by multiplex ligation-dependent probe amplification and quantitative PCR in 542 African-American, Asian, Caucasian, Hispanic and Ashkenazi Jewish individuals. The CYP2A6, CYP2B6 and CYP2E1 combined deletion/duplication allele frequencies ranged from 2 to 10% in these populations. High-resolution microarray-based comparative genomic hybridization (aCGH) localized CYP2A6, CYP2B6 and CYP2E1 breakpoints to directly oriented low-copy repeats. Sequencing localized the CYP2B6 breakpoint to a 529-bp intron 4 region with high homology to CYP2B7P1, resulting in the CYP2B6*29 partial deletion allele and the reciprocal, and novel, CYP2B6/2B7P1 duplicated fusion allele (CYP2B6*30). Together, these data identified novel CYP450 CNV alleles (CYP2B6*30 and CYP2E1*1Cx2) and indicate that common CYP450 CNV formation is likely mediated by non-allelic homologous recombination resulting in both full gene and gene-fusion copy number imbalances. Detection of these CNVs should be considered when interrogating these genes for pharmacogenetic drug selection and dosing.

INTRODUCTION

The significance of structural variation in human disease and phenotypic diversity has increasingly become recognized[1-3] and several studies have generated catalogs of copy number variants (CNVs) in the human genome to facilitate a better understanding of their functional relevance.[4-9] It is estimated that ~12% of the human genome contains CNVs, which range in size from less than 1 kb to several megabases.[10] Several mechanisms of CNV formation have been proposed including non-allelic homologous recombination (NAHR), non-homologous end joining, fork stalling and template switching, and microhomology-mediated break-induced replication.[11-12] Like single nucleotide polymorphisms (SNPs), CNVs not only influence disease susceptibility but also interindividual differences in drug response, as several important pharmacogenetic genes are known to have variable copy number (e.g., CYP2D6, GSTT1, GSTM1, SULT1A1 and UGT2B17).[13-15]

The polymorphic cytochrome-P450 (CYP450) system is a superfamily of 56 functional genes (and over 50 highly homologous pseudogenes) that are principally involved in drug metabolism and bioactivation.[16] Inherited genetic variation among the CYP450 genes contributes to disease susceptibility and interindividual differences in drug response. While SNPs represent an important form of CYP450 variation that can influence enzyme activity, CNVs also can affect CYP450 activity by altering gene expression.[17-18] However, most studies on CYP450 CNVs have focused on CYP2D6[18-21] and to a lesser extent on CYP2A6,[22-23] with limited CNV population data on other CYP450 family members. This is primarily because CNV detection is challenging for some CYP450 genes due to the presence of homologous gene family members, pseudogenes, and complex rearrangements including chimeras and gene conversions. For example, CYP2D6 CNV characterization and metabolizer phenotype assignment is convoluted by the presence of CYP2D6/2D7P1 hybrid alleles[21] and the fact that not all duplication alleles encode functional enzymes (e.g., increased function CYP2D6*2xN; reduced function CYP2D6*10xN; and nonfunctional CYP2D6*4xN).[18, 24–26]

Thus, this study aimed to determine the prevalence of CNVs in CYP450 genes other than CYP2D6 in a large multi-ethnic cohort of healthy, unrelated African-American, Asian, Caucasian, Hispanic, and Ashkenazi Jewish (AJ) individuals. Copy number was determined for 11 CYP450 genes by multiplex ligation-dependent probe amplification (MLPA) and validated by quantitative PCR (qPCR) copy number assays. Additionally, a subset of samples were subjected to high-resolution microarray-based comparative genomic hybridization (aCGH) to accurately map the size of identified CNVs and localize breakpoint intervals. In addition to revealing the diverse frequency profiles of CYP450 CNVs across multi-ethnic populations, these studies also identified two novel CYP450 alleles, CYP2B6*30 and CYP2E1*1Cx2. Moreover, the frequency data offer insight into the adaptive evolution of CYP450 CNVs and the breakpoint analyses provide insight into their mechanism of formation.

MATERIALS AND METHODS

Study population

Peripheral blood samples from healthy donors who indicated their racial/ethnic background and gave informed consent for the use of their DNA for research were obtained from the New York Blood Center with Institutional Review Board (IRB) approval.[27-29] In addition, blood samples were obtained with informed consent from unrelated, healthy 100% AJ individuals from the greater New York metropolitan area as previously described.[28, 30–32] All personal identifiers were removed, and isolated DNA samples were tested anonymously. Genomic DNA was isolated using the Puregene® DNA Purification kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Individuals from five different populations were subjected to CYP450 CNV testing: African-American (n=105), Asian (n=102), Caucasian (n=103), Hispanic (n=109), and AJ (n=123).

Multiplex Ligation-Dependent Probe Amplification (MLPA)

MLPA was performed using the recently developed Cytochrome P-450 MLPA kit (P128-B1; MRC-Holland, Amsterdam, The Netherlands) as per the manufacturer’s instructions. This probe mix included 2 to 5 MLPA probes each for CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4 and CYP3A5. In brief, all DNA samples were diluted to 200 ng with tris-EDTA buffer and denatured in a thermalcycler for 5 min at 98 °C. After cooling to 25°C, probemix and MLPA buffer were added to each sample, mixed and incubated for 1 min at 95°C followed by 16 hour hybridization at 60°C. The ligation reaction was performed at 54°C by adding 32 µl of Ligase-65 mix to each tube followed by heating for 5 min at 98°C. PCR buffer, water, and MLPA ligation reactions were mixed in new tubes and maintained in a thermalcycler at 60°C while polymerase mix was added to each tube. Exon-specific probes with universal tagged primers underwent PCR which consisted of 35 amplification cycles (95°C for 30 sec, 60°C for 30 sec, and 72°C for 60 sec), followed by a 20 min incubation at 72°C. Amplified products were separated by capillary gel electrophoresis and analyzed using GeneMarker v1.90 software (SoftGenetics, State College, PA). After quality control and data normalization to 12 autosome reference probes (Supplemental Table S1), copy number was determined according to the following peak ratio ranges: one copy >0.25 and <0.75; two copies >0.75 and <1.25; three copies >1.25 and <1.7; four copies >1.7 and <2.0. The copy numbers of CYP2A6, CYP2B6, and CYP2E1 alleles were determined using the average values of interrogated intragenic probes (four, three and three probes, respectively).

Copy Number Analyses using Quantitative PCR (qPCR)

The copy number of selected genes was interrogated using commercially available TaqMan® real-time qPCR Copy Number Assays (Applied Biosystems, Carlsbad, CA) as per the manufacturer’s instructions. In brief, FAM™-labeled CYP2A6, CYP2B6, and CYP2E1 TaqMan® minor groove binding probes and unlabeled PCR primers (Supplemental Table S2) were individually run in a duplex qPCR with a VIC™-labeled RNase P TaqMan® Copy Number Reference Assay (catalogue number: 4403326; Applied Biosystems). Quadruplicate experiments were each performed in 10 µl reactions containing ~10 ng of DNA, 1X TaqMan® Genotyping Master mix, 0.5 µl each of TaqMan® Copy Number and Reference Assays in 384 well plates. Covered plates were run in a 7900HT Fast Real-Time PCR System (Applied Biosystems) and the amplification consisted of a denaturation step at 95°C for 10 min followed by 40 amplification cycles (95°C for 15 sec and 60°C for 60 sec). Data was captured using absolute quantitation with a manual CT threshold and autobaseline, followed by analysis using CopyCaller™ v1.0 Software (Applied Biosystems) where the number of copies of target sequence was determined by relative quantitation with the comparative CT (ΔΔCT) method. This method measures the CT difference (ΔCT) between target and reference sequences, and then compares the ΔCT values of test samples to a calibrator sample known to have two copies of the target sequence. The copy number of the target was calculated to be two times the relative quantity.

Array-based Comparative Genomic Hybridization (aCGH)

To map the breakpoint regions of identified CNVs, a custom oligonucleotide microarray was designed using eArray (https://earray.chem.agilent.com/earray/; Agilent Technologies, Santa Clara, CA) that included unique high-density catalog probes corresponding to the February 2009 human genome assembly, NCBI Build 37 (hg19). Specifically, CYP2A6, CYP2B6, and CYP2E1 were assessed by interrogating the following chromosome nucleotide intervals: chr19:41,255,363 – 41,551,068 (1,224 probes; ~242 bp resolution) and chr10:134,507,520 – 135,534,747 (7,995 probes; ~128 bp resolution). In addition, a whole-genome backbone of 3,000 interspersed probes was included on the array with 300 whole-genome probes replicated five times to provide a feature extraction quality control measure.

All aCGH experiments were performed according to the manufacturer's instructions and as previously described.[33-34] In brief, ~1 µg of experimental and reference DNAs were double digested with AluI and RsaI restriction endonucleases (Promega) and fluorescently labeled with Cyanine 5-dCTP (Cy-5; experimental) and Cyanine 3-dCTP (Cy-3; reference) using the Genomic DNA Labeling kit (Agilent Technologies). Labeled experimental and reference DNAs were purified, combined, denatured, pre-annealed with Cot-1 DNA (Invitrogen, Carlsbad, CA) and blocking reagent (Agilent Technologies), and hybridized to arrays in a rotating oven (20 rpm) at 65°C for 24 hr. After hybridization and recommended washes, the arrays were scanned at 5 µm resolution with a G2505C Agilent Microarray Scanner. Images were processed and the data analyzed with CytoGenomics 1.5 software (both from Agilent Technologies). All array data passed the recommended quality control metrics. Breakpoints were defined as inter-probe regions with surrounding average log2 ratios that indicated a change in copy number from two to one or three copies.

CYP2B6 and CYP2E1 Sequence Analyses

The breakpoint regions (~3 kb) of the CYP2B6/2B7P1 fusion alleles (CYP2B6*29 and *30) were assessed by breakpoint-spanning PCR using overlapping primers (Supplemental Table S3) and the SequalPrep™ Long PCR Kit (Invitrogen). In addition, long-range PCR was performed to generate CYP2B6 and CYP2B7P1 gene fusion-specific amplicons of 18 kb and 11 kb spanning exons 1–5 and exons 4–9, respectively, which were used as templates for exon and exon-intron junction sequencing using primers listed in Supplemental Table S3. DNA samples from four controls with two copies of CYP2B6 by MLPA and qPCR were included as negative controls. Long-range PCR reactions were performed in 20 µl containing ~100 ng of DNA, SequalPrep™ 1X PCR buffer, 0.4 µl DMSO, SequalPrep™ 1X Enhancer A, 0.5 µM of forward and reverse primers (Supplemental Table S3), and 1.8 units of SequalPrep™ Long Polymerase. Amplification consisted of an initial denaturation step at 94°C for 2 min followed by 10 amplification cycles (94°C for 10 sec, 58°C for 30 sec, and 68°C for 1 min/kb), 30 amplification cycles (94°C for 10 sec, 58°C for 30 sec, and 68°C for 1 min/kb (+20 sec/cycle) and a final extension at 72°C for 5 min. Additionally, for selected subjects, all CYP2B6 exons and exon/intron boundaries were individually amplified and sequenced using exon primers listed in Supplemental Table S3.

For subjects harboring a CYP2E1 duplication, long-range PCR was performed to generate a CYP2E1 specific amplicon of ~3 kb spanning the promoter region and exon 1 using conditions detailed above and primers listed in Supplemental Table S4. Exons 2 to 9 of CYP2E1 were also amplified and directly sequenced using primers listed in Supplemental Table S4.

All amplicons were digested with 3.0 units of shrimp alkaline phosphatase (SAP) and 2.0 units of Exonuclease I (both from USB Corporation, Cleveland, OH) and bidirectionally sequenced using amplification and/or sequencing primers. Sequencing was performed on an ABI 3700 Sequencer (Applied Biosystems) and sequence chromatograms were analyzed using Mutation Surveyor software v3.30 (SoftGenetics).

RESULTS

CYP450 Copy Number Profiling

CNVs were interrogated with an MLPA panel of 11 CYP450 genes (CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4 and CYP3A5) among 542 healthy, unrelated African-American, Asian, Caucasian, Hispanic, and AJ individuals. In addition to CYP2D6 (data not shown), CNVs were detected in CYP2A6, CYP2B6, and CYP2E1. Representative MLPA data from samples with variable CYP450 copy number are shown in Table 1 and Supplemental Table S1, and the identified CYP2A6, CYP2B6, and CYP2E1 allele frequencies are summarized in Table 2. Wide variability in CYP2A6, CYP2B6, and CYP2E1 deletion/duplication allele frequencies was observed across the tested populations ranging from 0 to 8.3% for deletions and 0 to 4.1% for duplications. The total CNV allele frequencies were 9.0% for African-Americans, 10.3% for Asians, 4.4% for Caucasians, 5.5% for Hispanics, and 2.4% for the AJ (Table 2). All detected CYP2A6, CYP2B6, and CYP2E1 CNVs were validated by TaqMan® qPCR copy number assays, indicating that the multiplexed MLPA assay was robust at detecting CYP450 CNVs. The genomic locations of MLPA and TaqMan® probes are illustrated in Supplemental Figure S1.

TABLE 1

Representative MLPA data from selected samples with variable CYP450 copy number

		Samples (Peak Ratiosa)
Gene	MLPAProbe	AA404	AA410	HISP287	C048	C106
CYP2A6	exon 1	0.983	1.540	0.924	0.521	0.974
	exon 2a	1.029	1.487	0.953	0.575	1.038
	exon 2b	1.036	1.430	0.934	0.580	0.966
	exon 5	1.001	1.386	1.057	0.549	1.033
CYP2B6	exon 2	0.576	1.035	0.963	1.060	1.008
	exon 3	0.563	1.111	0.928	1.022	0.963
	exon 4	0.535	1.030	0.950	1.022	1.003
CYP2E1	exon 5	0.943	0.916	1.521	0.899	1.412
	exon 6	0.999	1.016	1.519	0.954	1.406
	exon 8	1.024	0.926	1.514	0.998	1.466

See ‘Materials and Methods’ for peak ratio calculation.

Shading indicates MLPA probes with variable copy number (one copy > 0.25 and < 0.75; three copies >1.25 and < 1.7; four copies > 1.7 and <2.0).

AA: African-American; C: Caucasian; HISP: Hispanic.

TABLE 2

Variable CYP450 copy number allele frequencies

		African-American(n = 210)		Asian(n = 204)		Caucasian(n = 206)		Hispanic(n = 218)		Ashkenazi Jewish(n = 246)
Gene	Copies/Allele	Freq.	95% CI	Freq.	95% CI	Freq.	95% CI	Freq.	95% CI	Freq.	95% CI
CYP2A6	2	0.010	0.000–0.023	0.015	0.000–0.032	-	-	-	-	0.020	0.000–0.038
	1	0.952	0.923–0.981	0.902	0.861–0.943	0.985	0.969–1.000	0.986	0.971–1.000	0.980	0.962–1.000
	0	0.038	0.012–0.064	0.083	0.045–0.121	0.015	0.000–0.031	0.014	0.000–0.029	-	-
CYP2B6	2	-	-	0.005	0.000–0.015	-	-	-	-	-	-
	1	0.995	0.986–1.000	0.995	0.985–1.000	1.000	1.000-1.000	1.000	1.000-1.000	1.000	1.000-1.000
	0	0.005	0.000–0.014	-	-	-	-	-	-	-	-
CYP2E1	2	0.038	0.012–0.064	-	-	0.029	0.006–0.052	0.041	0.015–0.067	0.004	0.000–0.012
	1	0.962	0.936–0.988	1.000	1.000-1.000	0.971	0.948–0.994	0.959	0.933–0.985	0.996	0.988–1.000
	0	-	-	-	-	-	-	-	-	-	-

n: number of alleles; CI: confidence interval.

Breakpoint Identification by aCGH

The CYP2A6, CYP2B6 and CYP2E1 breakpoint intervals were further refined by aCGH using a custom high-density oligonucleotide microarray on a subset of 46 samples with CYP450 CNVs as determined by MLPA and qPCR (Table 3). The University of California Santa Cruz (UCSC) Genome Browser (http://genome.ucsc.edu/; human genome build GRCh37/hg19) was used to visualize the location of known human genes from the NCBI reference sequence collection (RefSeq), structural variants from the Database of Genomic Variants (DGV; http://projects.tcag.ca/variation/), segmental duplications/low copy repeats (LCRs), and low-complexity repetitive elements (e.g., SINEs, LINEs).

TABLE 3

Summary of CYP450 aCGH analyses

Gene	Chr	CNV Interval(Mb)	CNV size(kb)	Proximal LCR	Distal LCR	Repetitive elements
CYP2A6	19	41.33 – 41.36	~32	41,336,557 – 41,361,522	41,367,421 – 41,393,307	SINEs and LINEs in directly oriented LCRs; CYP2A7 homologous gene
CYP2B6	19	41.45 – 41.51	~70	41,426,446 – 41,460,038	41,494,916 – 41,528,965	SINEs in directly oriented LCRs; CYP2B7P1 pseudogene
CYP2E1	10	135.250 – 135.381	~130	135,236,151 – 135,249,933	135,380,989 – 135,394,059	SINEs in directly oriented LCRs; SPRN and SPRNP1 pseudogene

Based on human genome build GRCh37/hg19. CNV: Copy number variant; kb: kilobases; LCR: low copy repeat (segmental duplication); LINEs: long interspersed elements; Mb: megabases; SINEs: short interspersed elements (e.g., Alu).

CYP2A6

Both CYP2A6 deletion and duplication alleles were detected by MLPA with copy number probes in exons 1, 2 and 5. The deletion allele was detected in all populations except the AJ and was most common among Asians (8.3%). In contrast, the duplication allele frequency was highest among the AJ (2.0%) but not detected in the Caucasian or Hispanic populations. Two TaqMan® qPCR assays (exons 1 and 7; Supplemental Figure S1) confirmed the MLPA results and indicated that the identified deletions and duplications likely encompassed the entire CYP2A6 gene (Figure 1). However, any CYP2A6/2A7 gene conversions in the 3’ flanking region[35] would not be detected by these MLPA or qPCR assays.

FIGURE 1

CYP2A6 copy number confirmation and breakpoint region identification

(A) TaqMan® qPCR copy number assays located in CYP2A6 exons 1 and 7 were used to confirm CNVs detected by MLPA. Note the homozygous duplications and deletions for AJ 080 and Asian 035, respectively. (B) Representative CYP2A6 aCGH data, indicating that the CNV breakpoints localized to two segmental duplications surrounding the CYP2A6 gene, resulting in full gene deletion and duplication. Red, black and blue dots represent oligonucleotide aCGH probes with log2 ratios less than −0.20, in between −0.20 and 0.20, and >0.20, respectively. (C) Illustration of the local genomic architecture surrounding CYP2A6 with enlarged views of the interval illustrated in panel B (~41.32 to 41.40 Mb). The location of known human genes from NCBI RefSeq, aCGH probes, identified CNV interval, segmental duplications, repetitive elements (e.g., SINEs, LINEs, LTRs), and structural variants from the Database of Genomic Variants (DGV; blue=copy number gain; red=copy number loss; brown=copy number gain and loss) are represented. The CNV breakpoints map to a cluster of neighboring repetitive Alu and LINE elements present in two directly oriented LCRs (>90% similarity). (D and E) Dot Matrix plots showing regions of similarity based upon alignment using BLAST (http://blast.ncbi.nlm.nih.gov). Both the x- and y-axes represent nucleotide sequence (in kilobases) corresponding to (D) the genomic interval displayed in panels B and C, and (E) the proximal and distal segmental duplications highlighted by red boxes in panel C. Alignments are shown in the plots as lines, whereby directly oriented homologies are slanted from the bottom left to the upper right corner. Approximate location of CYP2A6 CNV region is highlighted in panel D with a red box.

High-resolution aCGH mapped the mean size of the CYP2A6 deletion and duplication to ~32 kb at 19q13.2 (~41.33 – 41.36 Mb) and localized the CNV breakpoints to a cluster of Alu and LINE elements within the 3’ flanking regions of CYP2A6 and the more distal homologous CYP2A7 gene (Figure 1). The identified breakpoints mapped to two directly oriented LCRs: chr19:41,336,557 – 41,361,522 (proximal) and chr19:41,367,421 – 41,393,307 (distal) (Table 3). Sequence alignment indicated that the two LCRs share >90% sequence homology and encompass both CYP2A6 and CYP2A7 (Figure 1).

CYP2B6

Identification of the CYP2B6/2B7P1 Duplicated Fusion Allele (CYP2B6*30)

Both CYP2B6 deletion and duplication alleles were detected in the African-American and Asian populations, respectively, by MLPA with copy number probes in exons 2, 3 and 4. Two TaqMan® qPCR assays were used to confirm these results, one in exon 4 and a second located 12.6 kb downstream of the CYP2B6 gene (Supplemental Figure S1). No commercial TaqMan® probes were available that specifically interrogated exons 5 to 9 of CYP2B6. Interestingly, whereas the exon 4 TaqMan® probe confirmed the MLPA results, the downstream TaqMan® probe indicated a normal balanced copy number, suggesting the possibility of partial CYP2B6 deletion and duplication (Figure 2).

FIGURE 2

CYP2B6 copy number confirmation and breakpoint region identification

(A) TaqMan® qPCR copy number assays located in CYP2B6 exon 4 and 3’ downstream of the gene were used to confirm CNVs detected by MLPA. (B) Representative CYP2B6 aCGH data, indicating that the CNV breakpoints localized to two segmental duplications surrounding CYP2B6 and the CYP2B7P1 pseudogene, resulting in both partial gene deletion and duplication. Red, black and blue dots represent oligonucleotide aCGH probes with log2 ratios less than −0.20, in between −0.20 and 0.20, and >0.20, respectively. (C) Illustration of the local genomic architecture surrounding CYP2B6 with enlarged views of the interval illustrated in panel B (~41.41 to 41.54 Mb). Genome browser tracks are as detailed in Figure 1. The CNV breakpoints map to a homologous region of introns 4 of both CYP2B6 and the CYP2B7P1 present in two directly oriented LCRs (>90% similarity). (D and E) Dot Matrix plots showing regions of similarity based upon alignment using BLAST (http://blast.ncbi.nlm.nih.gov). Both the x- and y-axes represent nucleotide sequence (in kilobases) corresponding to (D) the genomic interval displayed in panels B and C, and (E) the proximal and distal segmental duplications highlighted by red boxes in panel C. Alignments are shown in the plots as lines, whereby directly oriented homologies are slanted from the bottom left to the upper right corner. Approximate location of CYP2B6 CNV region is highlighted in panel D with a red box.

High-resolution aCGH mapped the size of the CYP2B6 deletion and duplication to ~70 kb at 19q13.2 (~41.45 – 41.51 Mb) and localized the CNV breakpoints to a homologous region within introns 4 of both CYP2B6 and the proximal CYP2B7P1 pseudogene (Figure 2). The identified breakpoints mapped to two directly oriented LCRs: chr19:41,426,446 – 41,460,038 (proximal) and chr19:41,494,916 – 41,528,965 (distal) (Table 3). Sequence alignment indicated that the two LCRs share >90% sequence homology (Figure 2). These results suggested that the partial deletion allele represented the previously reported CYP2B6*29 allele[36] and the partial duplication represented a reciprocal, and novel, CYP2B6/2B7P1 duplicated fusion allele (Figures 2 and 3). The data for this allele, including the sequencing described below, were reviewed by the Cytochrome P450 Allele Nomenclature Committee[37-38] and it subsequently has been designated CYP2B6*30 (http://www.cypalleles.ki.se/cyp2b6.htm).

FIGURE 3

Schematic representation of the CYP2B6*29 and *30 fusion allele formation and location of PCR amplicons

Exons for CYP2B7P1 and CYP2B6 are shown in white and black boxes, respectively. The high degree of sequence homology between CYP2B6 and CYP2B7P1 facilitates unequal crossover at the intron 4/exon 5 region (noted by asterisks) leading to the formation of the CYP2B6*29 partial deletion allele and the reciprocal CYP2B6*30 duplicated fusion allele. Long-range PCR fragments used to sequence the recombination site and exons are illustrated.

Sequence Analyses of CYP2B6*29 and *30

A long-range PCR-based strategy was employed to interrogate the breakpoint regions of the identified CYP2B6*29 deletion and CYP2B6*30 duplication fusion alleles (Figure 3). All breakpoint spanning amplicons were sequenced and the putative breakpoint regions were identified by multiple sequence alignment using CLUSTAL (http://www.ebi.ac.uk/clustalw). Based on the alignment data, the breakpoint region was localized to a 529 bp interval within introns 4 and exons 5 of both CYP2B6 and CYP2B7P1 where the sequences of the two genes are nearly identical. Breakpoint junctions for CYP2B6*29 and *30 were defined at the locations where the nucleotide sequence changed from CYP2B7P1 to CYP2B6 and from CYP2B6 to CYP2B7P1, respectively (Figure 3; Supplemental Figures S2 and S3). The homologous introns 4 of CYP2B6 and CYP2B7P1 also contain four directly-oriented Alu elements (AluSc, AluSx, AluSx, and AluSg/x).

In addition, all CYP2B6*29 and *30 exons and exon-intron junctions were sequenced. The two alleles were translated using ExPasy (http://web.expasy.org) and aligned with CYP2B6 to identify polypeptide sequence alterations. Despite the unique breakpoint region identified in our study, the coding sequence of our CYP2B6*29 allele was consistent with that previously reported[36] (http://www.cypalleles.ki.se/cyp2b6.htm), sharing complete sequence identity with exons 1 to 4 of CYP2B7P1 and exons 5 to 9 of CYP2B6. Conversely, CYP2B6*30 shared complete sequence identity with exons 1 to 4 of CYP2B6 and exons 6 to 9 of CYP2B7P1. Given the high sequence homology between exons 5 of CYP2B6 and CYP2B7P1, it is unclear which gene this exon is derived from for CYP2B6*30. Fifteen amino acid alterations were detected in CYP2B6*30 (p.A279P, p.N289I, p.L290I, p.R326K, p.I332V, p.E339A, p.H341D, p.Y354R, p.S360A, p.R378X, p.I381T, p.H397R, p.T423N, p.M459V and p.R487C), indicating that the duplicated CYP2B6*30 allele is most likely nonfunctional. Sequencing the CYP2B6 exons not included in *30 identified the p.Q172H and p.K262R variants; however, the phase of these alleles could not be determined. As such, the downstream CYP2B6 copy in tandem with *30 could either be a *1 or a *4 (p.K262R), *6 (p.[Q172H+K262R]), or *9 (p.Q172H) variant allele.

CYP2E1

Identification of the Duplicated CYP2E1 Allele

Duplicated CYP2E1 alleles were detected by MLPA with probes in exons 5, 6 and 8 in all tested populations except Asians, and were most frequent among Hispanics (4.1%). No CYP2E1 deletion carriers were identified. Six TaqMan® qPCR assays targeting exons 1 and 9, and two each upstream (45.0 kb and 16.8 kb) and downstream (1.8 kb and 15.2 kb) of CYP2E1 were used to confirm the MLPA data (Supplemental Figure S1). All qPCR copy number data were consistent with those obtained by MLPA and indicated full gene duplication (Figure 5). High-resolution aCGH mapped the mean size of the CYP2E1 duplication to ~130 kb at 10q26.3 (~135.250 – 135.381 Mb) and localized the CNV breakpoints to two directly oriented LCRs: chr10:135,236,151 – 135,249,933 (proximal) and chr10:135,380,989 – 135,394,059 (distal) (Table 3). Sequence alignment indicated that the two LCRs share >98% sequence homology and encompass the majority of the SPRN gene and its distal SPRNP1 pseudogene (Figure 4).

FIGURE 4

CYP2E1 copy number confirmation and breakpoint region identification

(A) TaqMan® qPCR copy number assays located in exons 1 and 9 and surrounding CYP2E1 were used to confirm CNVs detected by MLPA. (B) Representative CYP2E1 aCGH data, indicating that the CNV breakpoints localized to two segmental duplications surrounding CYP2E1, resulting in full gene duplication. Red, black and blue dots represent oligonucleotide aCGH probes with log2 ratios less than − 0.20, in between −0.20 and 0.20, and >0.20, respectively. (C) Illustration of the local genomic architecture surrounding CYP2E1 with enlarged views of the interval illustrated in panel B (~135.21 to 135.41 Mb). Genome browser tracks are as detailed in Figure 1. The CNV breakpoints map to highly homologous directly oriented LCRs (>98% similarity). (D and E) Dot Matrix plots showing regions of similarity based upon alignment using BLAST (http://blast.ncbi.nlm.nih.gov). Both the x- and y-axes represent nucleotide sequence (in kilobases) corresponding to (D) the genomic interval displayed in panels B and C, and (E) the proximal and distal segmental duplications highlighted by red boxes in panel C. Alignments are shown in the plots as lines, whereby directly oriented homologies are slanted from the bottom left to the upper right corner. Approximate location of CYP2E1 CNV region is highlighted in panel D with a red box.

Sequence Analyses of the CYP2E1*1Cx2 Duplication

Although the size of the CYP2E1 duplication (~130 kb) prevented the design of a duplication allele-specific PCR, common CYP2E1 variants were interrogated by direct sequencing of all samples harboring a CYP2E1 duplication. No CYP2E1*2, *3, *4, *5, *6, or *7 variants were identified among the duplication carriers; however, two 5’ promoter region repeat alleles were identified, *1C (6 repeats) and *1D (8 repeats).[39] For heterozygous *1C/*1D carriers (n=7; 30% of duplication carriers), the phase could not be determined so the duplicated allele for these cases would either be *1Cx2 or *1Dx2. However, for homozygous *1C/*1C subjects (n=16; 70% of duplication carriers), the duplicated allele must also be *1C. For these subjects, all nine CYP2E1 exons were sequenced and no additional coding region variants were identified. These data also have been reviewed by the Cytochrome P450 Allele Nomenclature Committee and the duplication allele has been designated CYP2E1*1Cx2 (http://www.cypalleles.ki.se/cyp2e1.htm). Notably, no CYP2E1*1D homozygotes were detected among the_CYP2E1 duplication subjects, suggesting that the CYP2E1*1DxN allele is either uncommon or non-existent in the tested populations.

DISCUSSION

The paucity of CNV data among the CYP450 genes, with the notable exception of CYP2D6, prompted our copy number survey of CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, and CYP3A5 in the African-American, Asian, Caucasian, Hispanic, and AJ populations. Of the 11 genes analyzed, CNVs were detected and confirmed in CYP2A6, CYP2B6 and CYP2E1, with multi-ethnic frequencies ranging from 0 to 8.3% for deletions and 0 to 4.1% for duplications. Interestingly, the discordant CYP450 deletion and duplication allele frequencies [e.g., CYP2A6 deletion (8.3%) and duplication (1.5%) frequencies among Asians] suggest that selective pressures likely have acted on certain alleles. CYP2D6 CNVs were also detected in all studied populations; however, these data are being integrated with CYP2D6 genotyping data for an independent manuscript in preparation. In addition to identifying the CNV frequencies for this panel of CYP450 genes, high-resolution aCGH and sequencing analyses facilitated the identification of the novel CYP2B6*30 duplicated fusion allele and CYP2E1*1Cx2. Moreover, these data indicate that common CYP450 CNV formation is likely mediated by NAHR resulting in both full gene and gene-fusion copy number imbalances.

CYP2A6 is a hepatic CYP450 enzyme primarily involved in the metabolism of nicotine, cotinine and nitrosamine precarcinogens.[40] The CYP2A6 gene is highly polymorphic and variant alleles have been associated with interindividual variability in nicotine metabolism, smoking behavior and tobacco-related cancer risk.[41] We detected wide variability in the distribution of CYP2A6 deletion alleles across the five tested populations with the highest frequency observed among Asians (8.3%). The duplication allele was detected at low frequencies (<2%) among the African-Americans, Asians and AJ, and was not detected in either the Caucasian or Hispanic cohorts. Although deletions are often under stronger purifying selection pressure than duplications to remove deleterious variants from the population,[42] the high CYP2A6 deletion frequency observed in our Asian population could be due to its previous association with lower cancer risks among Asians[43] and, therefore, may actually confer a protective effect. Moreover, it has been suggested that the CYP2A6 deletion is a relatively recent genetic event that likely occurred after the Caucasoid/Asian split ~35,000 years ago.[44]

Using high-resolution aCGH, we mapped the size of the CYP2A6 deletion and duplication to ~32 kb, with breakpoints localizing to the 3’ regions of CYP2A6 and the distal homologous CYP2A7 gene. Both CYP2A6 and CYP2A7 are encompassed by two directly oriented LCRs with >90% sequence homology and the identified breakpoints clustered to repetitive Alu and LINE elements. NAHR during meiosis can generate rearrangements as a consequence of recombination between regions of high sequence homology like LCRs and dispersed Alu repetitive elements.[45] For CYP2A6, the presence of two direct LCRs likely renders the region susceptible to rearrangements, leading to NAHR-mediated reciprocal full gene deletion and duplication.[22] As such, the different subtypes of the CYP2A6 deletion allele (CYP2A6*4A-H) previously reported in African populations[46] are likely the result of unique breakpoints mediated by different homologous Alu and/or other neighboring low complexity repeats. Moreover, gene conversions of the CYP2A6 3’ flanking region by CYP2A7,[35] which would not be detected by the methods employed in the current study, are also likely mediated by these low complexity repeats.

CYP2B6 is involved in the metabolism of anticancer agents (e.g., cyclophosphamide and ifosfamide), anesthetics (e.g., ketamine and propofol), and antiretrovirals (e.g., efavirenz and nevirapine). Commonly tested variant CYP2B6 alleles include p.K262R (*4), p.R487C (*5), and p.[Q172H+K262R] (*6).[47] Notably, the CYP2B6*6, *18, and *26 decreased activity alleles have been shown to influence the pharmacokinetics of the non-nucleoside reverse transcriptase inhibitor efavirenz,[48] which may have utility for CYP2B6 genotype-guided dose adjustment.[49] However, little is known about CNVs at this locus other than the CYP2B6*29 partial deletion allele.[36]

By MLPA screening and qPCR assays, we identified both CYP2B6 deletion and duplication alleles that were limited to exons 1 to 4 of the CYP2B6 gene. High resolution aCGH mapped the size of the CYP2B6 deletion/duplication to ~70 kb and localized the CNV breakpoints to introns 4/exons 5 of CYP2B6 and the proximal CYP2B7P1 pseudogene, both within LCRs that share >90% sequence homology. Similar to CYP2A6, these LCRs likely mediate NAHR between CYP2B6 and CYP2B7P1, resulting in the previously reported CYP2B6*29 partial deletion allele and the reciprocal, and novel, CYP2B6/2B7P1 duplicated fusion allele (designated CYP2B6*30 by the Cytochrome P450 Nomenclature Committee). The CYP2B6*29 and *30 breakpoints were further narrowed by breakpoint-spanning PCR and sequencing, which indicated that unequal crossing over most likely occurred at a 529 bp interval within introns 4 and exons 5 of CYP2B6 and CYP2B7P1 where the two genes are nearly identical. Additionally, both introns 4 contains four directly oriented homologous Alu elements (AluSc, AluSx, AluSx, and AluSg/x) that may also facilitate NAHR at this locus.[50]

CYP2B6*30 is predicted to have 15 amino acid changes, including the p.R487C alteration consistent with the *5 allele.[51-52] More importantly, the exon 7 p.R378X nonsense alteration characteristic of the CYP2B6*28 allele is also encoded by *30, indicating that the translated *30 polypeptide would be truncated and lacking a heme binding site.[53] Taken together, these amino acid alterations indicate that the duplicated CYP2B6*30 allele is most likely nonfunctional. Detecting CNV alleles like CYP2B6*29 and *30 has significant pharmacogenetic implications when interpreting the metabolizer phenotype. For example, if just assessing copy number of exons 1 through 4, *30 could be misclassified as CYP2B6*1x2, which predicts an ultra-rapid metabolizer phenotype. Although the CYP2B6*29 (deletion) and *30 (duplication) alleles were rare in the tested populations (~1% in African Americans and Asians, respectively), these variant alleles have been identified in other population studies as evidenced by entries in the Database of Genomic Variants (DGV) for this region.

CYP2E1 metabolizes clinically important drugs such as acetaminophen, isoniazid and tamoxifen, and is involved in the activation of several procarcinogens. Additionally, CYP2E1 is the major component of the microsomal ethanol oxidizing system and accounts for ~10% of ethanol oxidation in the liver, indicating that this locus likely plays an important role in alcoholism and related disorders. For example, some studies previously have associated variant CYP2E1 alleles (e.g., *1D, *5B and *6) with alcoholism and cancer development in different ethno-racial groups.[54-58] Notably, although the Database of Genomic Variants (DGV) indicates that the 10q26.3 region is copy number variable, no CYP2E1 CNV alleles previously have been characterized by the Cytochrome P450 Nomenclature Committee (http://www.cypalleles.ki.se/cyp2e1.htm).

Interestingly, although we detected the CYP2E1 duplication (designated CYP2E1*1Cx2 by the Cytochrome P450 Nomenclature Committee) in all populations except Asians, no deletion alleles were identified in any of the studied racial or ethnic groups. Validation with qPCR assays confirmed full gene CYP2E1 duplications in all cases. High-resolution aCGH mapped the size of the CYP2E1 duplication to ~130 kb and localized the CNV breakpoints to two highly homologous directly-oriented LCRs that share >98% sequence homology and include SPRN and its distal SPRNP1 pseudogene. Of note, SPRN is a member of the prion protein family and variant SPRN alleles have recently been associated with Creutzfeldt-Jakob disease susceptibility.[59] Additionally, our aCGH data indicate that the CYP2E1 duplication region also includes the SYCE1 gene. SYCE1 is recruited to the synaptonemal complex during meiosis and, interestingly, microdeletions that encompass SYCE1 recently have been associated with premature ovarian failure.[60] Together, these data suggest that 10q26.3 CNVs may have additional clinical relevance beyond CYP2E1 pharmacogenetics and that adaptive selective pressures may partly explain why deletions in this region were not detected in any of our tested populations.

In conclusion, MLPA and TaqMan® qPCR assays were used to determine the frequencies of deletion and duplication alleles for 11 CYP450 genes in a multi-ethnic healthy cohort. In addition to CYP2D6, CNVs were detected in CYP2A6, CYP2B6 and CYP2E1, and their frequencies varied among the studied populations. Of note, the discordant deletion and duplication allele frequencies for certain CYP450 CNV alleles (e.g., CYP2A6 duplication, CYP2E1 deletion) suggest that selective pressures have likely acted on these alleles. The identified CYP2A6, CYP2B6 and CYP2E1 CNV alleles and their frequencies are strongly supported by the entries in the Database of Genomic Variants (DGV; http://projects.tcag.ca/variation/) from other population studies. Importantly, our study also identified the novel CYP2B6/2B7P1 duplicated fusion allele (designated CYP2B6*30) and the CYP2E1 duplication (designated CYP2E1*1Cx2). Moreover, high-resolution aCGH and sequencing analyses localized breakpoints to directly oriented LCRs with >90–98% homology, which indicate that common CYP450 CNV formation is likely mediated by NAHR resulting in both full gene and gene-fusion copy number imbalances. Detection of these CNVs should be considered when interrogating these genes for pharmacogenetic drug selection and dosing.