A LD block is found in the HapMap Caucasians (CEU) spanning a 14-kb region from the proximal promoter (rs3824260) to the 3'-downstream (rs10504255) of the CYP7A1 gene (Figure 1. CEU-B1). A similar LD block from rs3824260 to the 3'-downstream was also reported in a Swedish population 18. About 4.4 kb upstream from rs3824260, there is another LD block (CEU-B2) crossing a 3-kb region at the distal promoter region. Recombination between the two blocks is 0.84. Only five haplotypes with a frequency > 2% exist in CEU-B1 (Figure 2. CEU-B1H1 to CEU-B1H5). CEU-B1H1 and CEU-B1H2 are two common haplotypes, together representing a total of 68% of the haplotype frequency in CEU-B1. CEU-B1H1 carries common alleles at all markers except rs8192879 in 3'-UTR, whereas CEU-B1H2 is composed of less common alleles at 5 out of 8 loci. In CEU-B2, there are only two types of haplotypes (CEU-B2H1 and CEU-B2H2). CEU-B2H1 carries common alleles at all SNP loci, whereas CEU-B2H2 has less common alleles. A similar LD pattern is found in the HapMap African YRI (Figure 1), but the larger LD block (YRI-B1) is slight shorter (9 kb from rs8192879 to rs3824260) than CEU-B1. The haplotype structure is also similar between YRI and CEU, however, the frequency of each haplotype is different. The most common haplotype (55%) in YRI-B1, YRI-B1H1, has identical haplotype structure with CEU-B1H2, whereas the second common haplotype in YRI-B1, YRI-B1H2 (22.5%), is the same as CEU-B1H1. YRI-B1H1 and YRIB1H2 together add up to 77.5% of the total haplotypes in YRI-B1. In YRI-B2, the dominant haplotype YRI-B2H1 has the same haplotype structure as CEU-B2H2, whereas less common haplotype YRI-B2H2 is identical to the common haplotype CEU-B2H1. A similar recombination (0.81) is also found between YRI-B1 and YRI-B2. In CHB and JPT, only one LD block is found from the distal promoter to a part of the CYP7A1 gene. Although the JPT-B1 (16 kb, from rs8192879 to rs1023649) is larger than CHB-B1 (10 kb, from rs1457043 to rs1023650), LD is weak between rs8192879 in intron 4 and rs1457043 in intron 2 in JPT. CHB and JPT share almost the same haplotype structure within the block. JPT-B1H1, JPT-B1H2 and JPT-B1H3 are the same as CHB-B1H1, CHB-B1H2 and CHB-B1H3, respectively. Only CHB-B1H4 (6%) is unique in CHB.

In comparison of LD and haplotype structure among the HapMap populations, strong LD is found from the distal promoter region to intron 2 of the CYP7A1 gene across the HapMap populations. Two common haplotypes with complete opposite alleles at all loci (common-versus-less common alleles) within this region count for more than 85% of total haplotype frequencies in all four HapMap populations. A diverted LD degree exists between intron 2 and the 3'-downstream region from high to low across CEU, YRI, JPT, and CHB.

Because of the strong LD in the CYP7A1 genes, some markers correlate 100% with each other in a population. Only a subset of representative SNPs is necessary for defining a haplotype. These SNPs can tag either neighboring markers or a set of common haplotypes within an LD block. The htSNPs in CYP7A1 were selected using Tagger, implemented in the HaploView 3.12, which combines the simplicity of pairwise methods with the potential efficiency of multimarker approaches 25. The CYP7A1 htSNPs are different in the various populations (see upside-down red triangles in Figure 2 for each HapMap population). Some markers tag on all three populations, but others for only one or two. It has been suggested that the populations genotyped in the HapMap project may serve as reference populations for the selection of htSNP markers in association studies 26.

Nine SNP markers and one short deletion marker were selected (see detail in Table 3), in which, eight are htSNP markers defined by the HapMap populations, including rs3808607, a functional polymorphism at A-278C in the promoter region. Two functional mutations were also included. One is a two-base deletion in exon 6 (1302 delTT) causing a frame shift and CYP7A1 enzyme deficiency 20. The other one is a C/T SNP in exon 3, causing an amino acid change at Asn233Ser. This is the only non-synonymous SNP reported in NCBI SNP database in the CYP7A1 gene.

To perform genotyping of the 10 markers in the Mexican-American and African-American populations, a high-throughput and inexpensive SNP genotyping platform was developed using thin-film biosensor chips. We have reported a microarray platform for genotyping both SNPs and microsattelite repeat on thin-film biosensor chips 2728. The thin-film biosensor chip has excellent sensitivity of detection and extremely low non-specific binding, making it an excellent platform for discrimination of polymorphisms 29. A positive reaction (blue color signal) can be visualized over the unreacted background (gold color) by an unaided human eye, without any instrumentation. Once the chips are printed, they are robust. Several thousands of genotypes can be performed in a 96-well plate in a laboratory with a standard molecular genetics setting within a few hours. Cost for reagents and materials to genotype 10 CYP7A1 htSNPs, including genomic DNA isolation, PCR reaction, and SNP genotyping on the thin-film biosensor chips, is ~US$0.20 per SNP per sample. It is relative less expensive than other high-throughput genotype platforms, such as TaqMan or Real-time PCR.

To verify genotyping specificity on the thin-film biosensor chips, a pool of the synthetic targets for allele 1 or allele 2 was applied to a chip for hybridization and ligation. After signals were developed, the result images were captured by a black-white camera on a Nucleosite™ Image Analyzer (Biostar, Inc., Louisville, Colorado). High specificity was achieved on these synthetic targets with unambiguous genotypes (see images in Figure 3B and 3C). A negative control showed the signals are target dependent (Figure 3D). As a positive control for genotyping, 12 HapMap DNA samples were purchased from Coriell Cell Repositories (Camden, NJ), which are one family trio from YRI (NA18500, NA18501 and NA18502); one family trio from CEU (NA06985, NA06991, and NA06993); three independent individuals from CHB (NA18524, NA18526, and NA18529); and three independent individuals from JPT (NA18940, NA18942, and NA18943). Genotypes of the 8 HapMap htSNPs in the 12 HapMap samples were determined on thin-film biosensor chips. A 100% concordance was obtained between the 96 genotypes generated by thin-film biosensor chips and the 96 genotypes downloaded from the HapMap database which are generated by Illumina Bead Assay.

To define the LD pattern and haplotype structures of CYP7A1 gene in Mexican-American and African-American populations, DNA samples from 90 healthy individuals for each population were randomly selected from our DNA bank. These DNA samples were collected by other research projects on alcoholism in the Mexican-American population 30 and pharmacogenomics of CYP enzymes in both Mexican-Americans and African-Americans 3132. Genotypes of the 10 selected markers on the 90 Mexican-American and 90 African-American subjects were determined by using the thin-film biosensor chip platform. Representative images of the different genotypes from different individuals are shown in Figure 3E, 3F, and 3G. Genotypes of each individual on the 10 markers were saved in linkage format and uploaded to HaploView. Observed genotype frequencies, allele frequencies, expected heterozygosity, and Hardy-Weinberg p-value of the 10 markers is summarized in Table 5. No significant HW p-values (<0.0010) were found. No TT deletion mutation at 1302 and C mutation at rs8192874 were detected in these two population samples. This indicates that these mutations have very low frequencies in the general populations. The genotyping data of the 8 htSNPs were uploaded to HaploView 3.12 to define LD patterns and haplotypes structures of CYP7A1. In Mexican-Americans, three LD blocks were identified. In comparison to the CEU LD blocks, MA-B3 in the distal promoter region has the same pattern as CEU-B2, but haplotype frequencies are different. MA-B3H1 has a frequency (78%) higher than CEU-B2H1 (55%). Unlike one big block in CEU, the CYP7A1 gene is divided by two LD blocks in the Mexican-American population. MA-B2 covers from proximal promoter to intron 2, whereas MA-B1 extents from 3'-UTR to 3'-downstream. The recombination frequencies between the blocks are 80–90%. In African-Americans, two LD blocks were recognized. AA-B2 has the identical structure as YRI-B2 and frequencies of the two haplotypes (AA-B2H1 and AA-B2H1) in the African-American population are almost the same as YRI-B2H1 and YRI-B2H2. AA-B1 is shorter than YRI-B1. The HapMap htSNPs are necessary SNP markers to capture all haplotypes in the MA and AA populations.

In summary, the human CYP7A1 gene and its surrounding sequences have constant genetic structures across all populations. This genetic structure can be divided into three components: (1) the distal promoter region, about 7-kb upstream from the transcriptional start code, there is a 3-kb LD block highly conserved across all populations. Only two haplotypes exist in this region in the most populations, except YRI. The most common haplotype in CEU and Mexican-American becomes the second most common haplotypes in YRI, African-American, CHB, and JPT populations. (2) A relative conserved LD block is present in the proximal part of CYP7A1 gene from the proximal promoter region (about 500 bp from the transcriptional start code) to intron 2 of CYP7A1. The two most conserved haplotypes count for up to 80 to 90% of the haplotype frequencies in all populations in this region. (3) A much diverted LD pattern is observed in the lower part of the CYP7A1 gene (from intron 4 to 3'-downstream). In CEU and YRI, a complete or partial LD block is merged to the block in the proximal part of the CYP7A1 gene. In Mexican-Americans, a LD block in this region is separated from the block in the proximal part of the CYP7A1 gene. In JPT, a weak linkage makes the proximal part block extended into the 3'-UTR. In CHB and African-Americans, there is no LD existing in this area.

The DNA samples for the HapMap come from a total of 270 people: 90 individuals from the Yoruba of Ibadan, Nigeria (YRI), (30 sets of trios, each trio with samples from two parents and an adult child); 90 individuals (30 sets of trios) from U.S. residents with northern and western European ancestry collected by the Centre d'Etude du Polymorphisme Humain (CEU); 45 unrelated individuals from the Tokyo area in Japan (JPT); and 45 unrelated individuals from Beijing, China (CHB). Two American population samples of Mexican-Americans and African-Americans were also used in this study. These DNA samples were collected by other research projects on alcoholism in the Mexican-American population 30 and pharmacogenomics of CYP enzymes in African-Americans 32. Studies on these human subjects were approved by the Human Subjects Committee of the Kansas University Medical Center. Ninety DNA samples of healthy individuals were randomly selected from each population.

LD patterns and haplotype structures of CYP7A1 gene in the HapMap populations were analyzed by using genotype data from the HapMap database. In the HapMap Phase II database, a total of about 5.9 million SNPs (about 1 SNP every 500 bp across the genome) are typed in the four HapMap populations 21. Genotypes of each selected SNP in the 270 HapMap population samples can be downloaded from its database 34. Fourteen SNPs are genotyped by the HapMap project in a total of 25 kb region with 10 kb of 5'-upstream flanking, 5 kb of 3'-downstream flanking, and the CYP7A1 gene sequences. The SNP density is about 1.8 kb per SNP. The CYP7A1 A-278C (or A-204C) promoter polymorphism is included with an ID number of rs3808607. A close promoter polymorphism C-554T, which was identified together with A-278C by Wang et al. 7, is also included as rs3824260. Chromosomal positions and locations in the CYP7A1 gene regions of the 14 SNPs are listed in Table 1 with polymorphic allele 1 for common allele and allele 2 for less common allele in CEU. Genotypes of each individual samples for the 14 SNP markers were dumped from the HapMap database and saved as a HapMap formatted file that can be opened directly by HaploView 3.12 for defining LD patterns and haplotype structure 25. Four HapMap files were separately saved for CEU, CHB, JPT, and YRI, respectively. By uploading the files into HaploView, frequencies of allele 2, frequencies of observed heterozygous genotypes, and Hardy-Weinberg p-value for each marker were summarized for each population (see Table 2). No marker has a HW p-value smaller than the cutoff value of 0.0010 in the four populations. The LD between any two markers was defined by HaploView 3.12. A standard color scheme is used to display LD in Figure 1. A LD block was created by confidence intervals 25 if 95% of the informative comparisons are in strong LD using default algorithms of 95% confidence bounds on D prime. Haplotypes structure was defined by using an accelerated EM algorithm, similar to the partition/ligation method 33. This creates highly accurate population frequency estimates of the phased haplotypes, based on the maximum likelihood as determined from the unphased input. Haplotypes with frequency > 2% in a block in CEU, YRI, JPT, and CHB are displayed in Figure 2. Alleles with blue boxes and red boxes represent common alleles and less common alleles in CEU, defined as Allele 1 and Allele 2, respectively. In the crossing areas, a value of multiallelic D' is shown to represent the level of recombination between the two blocks.

For each selected SNP, target DNA molecules from each sample were amplified by PCR. PCR primers were designed based on the following criteria to make the PCR reaction uniform: (1) product size should be 120–200 bp with about 50–100 base flanking sequences around the SNP site in both directions, and (2) annealing temperature should be about 60°C for a standard PCR reaction condition. The best primer sets were selected by DS Gene Software version 1.5 (accelrys). The primer sequences for each SNP site are listed in Table 4. The selected primer sequences were synthesized by Invitrogen (Carlsbad, California). Multiple sets of the PCR products were amplified in a single PCR reaction.

For each SNP, three oligonucleotide probes were synthesized. A pair of allele specific P-1 oligos, differing only in their 3'-terminal nucleotide sequence, generally has 40 nucleotides complementary to the corresponding target sequences, and an additional 10-dA residue at their 5'-ends that constitutes a "spacer". Their 5'-terminal nucleotide is modified with an aldehyde group, allowing covalent attachment to the chip surface 27. A second oligonucleotide probe (biotin-P2) with 20 nucleotides immediately adjacent to the SNP nucleotide carries a biotin at the 3' end for detection, and a phosphate at its 5' end for ligation. To test genotyping specificity, a pair of oligonucleotide targets was also synthesized. The P1, P2 and target sequences for each SNP are listed in Table 4. The synthesized P1 oligos were dissolved to 100 μM in 0.1 M phosphate buffer, pH 7.8. A P1 working solution of 1 μM in 0.1 M phosphate buffer, pH 7.8, and 10% glycerol was prepared for each P-1 probe before spotting. Twenty nano liter of the P1 working solution was spotted on a 7 × 7 mm² chip in an 8 per row × 6 per column format, by a BioDot PC controlled dispense arrayer AD3200. A duplicate set of P1 probes were spotted on a chip with a spotting pattern shown in Figure 3A. After the spotted chips were incubated in a humidity-controlled chamber for at least 2 hrs, the chips were washed with 0.1% SDS, water, and air dried. A standard operating procedure for genotyping SNPs on the printed biosensor chips was described previously 27. An arrayed chip was assembled into a square well of a 96-well microtiter plate for hybridization. A ligation reaction was carried out in a microtiter plate well containing an arrayed chip. A reaction solution (100 μl) contained 100 femtomoles of each relevant PCR amplicon of the 10 CYP7A1 SNPs, 10 nM P-2 probe (one for each SNP) and 5 units of mutant Ampligase in a buffer of 20 mM Tris-HCl, pH 8.3, 25 mM KCl, 10 mM MgCl₂, 0.5 mM NAD, 0.01% Triton X-100, and 5 mg/ml alkaline treated casein. The ligation reaction was incubated for 20 min at 60°C. 96 chips in a 96-well plate were processed simultaneously. After a stringent wash (3 times in 0.01 M NaOH at room temperature and 3 times in 0.1 × SSC), the chips were incubated with an antibiotin-horse radish peroxidase (HRP) conjugate (1 μg/ml in hybridization buffer) for 10 min, and the chips were rinsed with 0.1 × SSC. 100 μl of a precipitate-generating HRP substrate TMB (BioFx) was added to each chip and incubated for 5 min, rinsed in ddH₂0, and air-dried.