MICB Allele Genotyping on Microarrays by Improving the Specificity of Extension Primers

Major histocompatibility complex (MHC) class I chain-related gene B (MICB) encodes a ligand for activating NKG2D that expressed in natural killer cells, γδ T cells, and αβ CD8+ T cells, which is associated with autoimmune diseases, cancer, and infectious diseases. Here, we have established a system for genotyping MICB alleles using allele-specific primer extension (ASPE) on microarrays. Thirty-six high quality, allele-specific extension primers were evaluated using strict and reliable cut-off values using mean fluorescence intensity (MFI), whereby an MFI >30,000 represented a positive signal and an MFI <10,000 represented a negative signal. Eight allele-specific extension primers were found to be false positives, five of which were improved by adjusting their length, and three of which were optimized by refractory modification. The MICB alleles (*002:01, *003, *005:02/*010, *005:03, *008, *009N, *018, and *024) present in the quality control panel could be exactly defined by 22 allele-specific extension primers. MICB genotypes that were identified by ASPE on microarrays were in full concordance with those identified by PCR-sequence-based typing. In conclusion, we have developed a method for genotyping MICB alleles using ASPE on microarrays; which can be applicable for large-scale single nucleotide polymorphism typing studies of population and disease associations.


Introduction
The product of major histocompatibility complex (MHC) class I chain-related gene B (Gene ID: 4277; MICB) is expressed as three extracellular domains, a transmembrane segment, and a carboxy-terminal cytoplasmic tail [1]. The flexible extracellular domains-α1, α2, and α3together serve as a ligand for activating NKG2D that is expressed on natural killer (NK) cells, γδ T cells, and αβ CD8 + T cells [2]. As shown in the IMGT/HLA database (http://www.ebi.ac. uk/imgt/hla), over 40 polymorphic alleles have been found in exon 2 to 6 of MICB. Specific polymorphisms are associated with autoimmune diseases, cancer, and infectious diseases, as well as with the success of hematopoietic stem cell transplantation. For example, MICB Ã 004 allele is significantly associated with susceptibility to rheumatoid arthritis in Spanish Caucasians [3], MICB Ã 005:02 allele is negatively associated with cervical cancer in the Thai population [4], and MICB Ã 008 allele (designated as MICB Ã 0106 allele in 2004) is positively associated with ulcerative colitis in the Han Chinese population of central China [5]. MICB Ã 008 allele is also associated with a reduction in severity of dengue fever [6]. In addition, MICB Ã 008 and Ã 002 alleles are associated with susceptibility to celiac disease in Europeans [7]. Moreover, recipients in need of a transplant had significantly improved survival probabilities when MICA and MICB alleles matched between the donor and recipient [8].
In previous studies, genotyping MICB alleles has been carried out using polymerase chain reaction with sequence-specific primers (PCR-SSP) and PCR-sequence-based typing (PCR-SBT) [9,10]. Methods for detecting single nucleotide polymorphisms (SNPs) on a microarray platform include single base extension (SBE), allele-specific primer extension (ASPE), oligonucleotide ligation, allele-specific oligonucleotide direct hybridization, allele-specific cleavage of a flap endonuclease, and 5 0 nuclease assay [11][12][13]. In this study, we developed a high-throughput genotyping system for MICB alleles by optimizing allele-specific extension primers for ASPE on glass slides.

Genomic DNA extraction
Genomic DNA was extracted from peripheral blood samples taken from 200 healthy Korean individuals (staff and students of the College of Medicine, The Catholic University of Korea) using an AccuPrep Genomic DNA Extraction Kit (Bioneer Corporation; Daejeon, Korea), following the manufacturer's instructions. Briefly, the cells were lysed in 100 μL K buffer (50 mmol/L KCl, 10 mmol/L Tris-HCl (pH 8.3), 2.5 mmol/L MgCl 2 , 0.5% Tween 20, and 100 μg/mL Proteinase K) for 60 min at 56°C and then inactivated for 10 min at 95°C. After extraction, the DNA concentration was adjusted to 100 ng/μL and the resulting DNA samples were used as the PCR template in the genotyping assays.

Ethics statement
The study was approved by the Catholic University Institutional Review Board (IRB), with written informed consent from all subjects (Institutional Review Board number: MC13SISI0126). polymorphisms with an additional 18 bp oligo(dT) spacer at the 5 0 end, was dissolved in a buffer containing 350 mmol/L sodium bicarbonate (pH 9.0) ( Tables 3 and 4). The primers were then spotted onto standard microscope glass slides coated with aldehyde (SCHOTT; Jena, Germany) using a MicroGridII (BioRobotics; Cambridge, MA, USA) controlled with an MCM-310 operating system. The primers were immobilized as NH 2 -modified oligonucleotides. Distilled water was used to hydrate the primers on the slides at 25°C for 1 h in a humid chamber, and then the slides were baked at 120°C for 1 h.
Next, to deoxidize the aldehyde residue, the slides were washed for 30 min, first with 0.2% sodium dodecyl sulfate (Sigma; Dorset, UK), followed by distilled water, and then with an NaBH 4 solution containing 1 g NaBH 4 , 300 mL PBS, and 100 mL EtOH. Finally, the slides were rinsed with 0.2% sodium dodecyl sulfate followed by distilled water and then dried. The processed slides were stored at −70°C in a silica-gel box.

Primer extension on glass slides
The PCR products were treated with 10× λ exonuclease buffer containing 3.5U λ exonuclease (Fermentas; Burlington, Canada) and 1.75U shrimp alkaline phosphate (Qiagen; Hamburg, Germany). After the PCR products had been cleaved into approximately 50-bp fragments, they were hybridized with a mixture that contained 25× Themo Sequenase buffer (Enzymonics; Daejeon, Korea), 3.5U Thermo Sequenase (diluted 1:8 in Sequenase dilution buffer; Enzynomics; Daejeon, Korea), and 0.1 mmol/L each of both dATP/dTTP/dGTP and Cy5-dCTP at 60°C for 1 h. To remove non-specific primers, the slides were washed for five minutes with a solution containing 0.1% Alconox and then rinsed twice with distilled water, for 5 min each time. References from IMGT/HLA databases [14]. § +, successful primer -, failed primer; ND, not determined in our control panel (Table 1).
Underlines, MICB polymorphic sites.  The data were obtained using the EasyScan software after scanning with NanoDscan (Nanostorage; Seoul, Korea) and then analyzed for genotyping using Microsoft Excel. The background mean fluorescence intensity (MFI) was subtracted from the MFI of each primer; cut-off values were determined using quality control samples (Fig 1).
The MICB exon 6 primers were designed to differentiate MICB Ã 005:02 allele from MICB Ã 010 allele in exon 6, with the forward primer as 5 0 -GGGCAACTGAAGAGAGAAAAG-3 0 and the reverse primer as 5 0 -CAGGAGCAGTCGTGAGTTTG-3 0 . The PCR conditions were as follows: an initial denaturation of 95°C for 5 min; 30 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s, and extension at 72°C for 30 s; and a final extension at 72°C for 10 min. Then the reactions were held at 20°C.
Purified PCR products were sequenced using an ABI 377 DNA sequencer; the MICB primers for PCR amplification were also used as the sequencing primers. The sequencing data were analyzed using the ABI Factura software and the ABI Sequence Navigator program (PE Biosystems; Mississauga, Canada).

Calculation of melting temperature and statistical analysis
The melting temperature (Tm) for sequences longer than 13 bases was calculated as follows: Tm ¼ 64:9 þ 41 Â ðyG þ zC À 16:4Þ wA þ xT þ yG þ zC W, x, y, and z indicate the number of adenine, thymine, guanine and cytosine (A, T, G, and C nucleotides), respectively, in the sequence.
The Tm and GC content in primers from the successful group of allele-specific extension primers were compared to those from the failed group using the unpaired Student's t-test. We corrected for statistical analyses calculated from five or less observed scores using Fisher's exact test.

Control extension primers for quality control
For quality control, 20 extension primers were designed against sequences in MICB exon 2, 3, 4, and 5. The primers were 19 bp long, with a Tm of 53.0 ± 3.0°C, and a GC content of 57.4 ± 7.4% ( Table 2). The MFI value of the N0 primers was higher than the maximum value (65,535 MFI) of N0 primers measured by the EasyScan-1 scanning machine, and the MFI value for the N4 primers was >30,000. By contrast, the N1 primer group produced a very weak signal, whereas N2 and N3 displayed refractory extension (Fig 1). As the N0 and N4 primers showed strong positive signals, they were used as positive controls, and the primers in the other groups were used as negative controls. Based on these results, we set very strict and reliable cut-off values for MFI: an MFI >30,000 was considered as a positive signal, and an MFI <10,000 was considered a negative signal. Using these cut-offs, it was possible to select high quality allele-specific extension primers.

Optimization of allele-specific extension primers
The ability to discriminate MICA alleles from MICB alleles was verified using primer location and a cross over test. Primers for amplification of exons 2-5 of MICB alleles did not exist in the MICA gene (Gene ID: 100507436). The PCR products of MICB exons 2-5 did not showed any positive signals on the MICA allele-specific extension primers of microarray for MICA allele genotyping [17].
Eighteen samples including 13 MICB alleles previously identified in Koreans and other populations were selected as the quality control panel for ASPE on microarrays. We defined MICB alleles of 163 samples in the UCLA International KIR and MICA Exchange Report program. Among them, six samples including three MICB alleles of MICB Ã 005:01, Ã 005:06 and Ã 024 alleles which did not find in Korean population were added in Table 1. MICB alleles of the other 157 samples were also found in Korean population. A validation of the amplification reaction has been performed for the alleles (MICB Ã 002:01, Ã 003, Ã 004:01, Ã 005:01, Ã 005:02/ Ã 010, Ã 005:03, Ã 005:06, Ã 008, Ã 009N, Ã 014, Ã 018, Ã 019, and Ã 024 alleles) included in the quality control panel. To identify the 41 known MICB alleles, 36 allele-specific extension primers were designed with a length of 18 to 21 bases. Among these primers, the specificity and sensitivity of only 22 primers could be determined by the quality control panel used in this study (Table 3). Of these primers, 14 showed excellent sensitivity and specificity with a Tm of 52.9 ± 2.1°C and a GC content of 57.2 ± 5.1%. The eight failed primers had a Tm of 50.6 ± 1.5°C and a GC content of 50.0 ± 2.8%. The Tm and GC content were significantly different between the successful extension primer group and the failed primer group (P < 0.02 for each category).

Genotyping of MICB alleles
The ASPE on microarrays were evaluated using the arrays that genotyped the 18 SNPs in the MICB alleles. The samples of known MICB alleles from about 200 Koreans and six individuals of other populations were analyzed for genotyping. Scatter plots were resulted in numerical presentation about one clustering of three genotypes. The similar values between the signals from the normal allele-specific reactions were divided into three districts depending on the genotype; 20 SNPs were over 30,000 MFI (Fig 3). Scatter plots showed that the allele-specific  (Table 5), the other 10 SNPs of MICB did not detect and not discriminate all alleles, existed many ambiguous allele combinations. These SNPs are present only in very rare alleles which have been not reported in published papers and are not found in 163 samples of the UCLA International Exchange Report program. Therefore, it may have some limitations for application compared with direct Genotyping by ASPE on microarrays was fully concordant with MICB genotyping using PCR-SBT in the sample-to-sample comparison ( Table 6). MICB Ã 002:01 and Ã 004:01 alleles, which were determined using ASPE on microarrays, were identified as MICB Ã 002:01:01 and MICB Ã 004:01:01 alleles by PCR-SBT. Our SBT method could discriminate MICB Ã 010 allele  Development of Microarrays for MICB Genotyping Using ASPE from MICB Ã 005:02 allele because these two alleles are different in exon 6. We could identify only MICB Ã 005:02 allele in Korean population using PCR-SBT (n = 99). However, we showed MICB Ã 005:02/ Ã 010 allele because the primer sets used to amplify templates for ASPE on microarrays were not designed to differentiate MICB Ã 005:02 allele from MICB Ã 010 allele in exon 6. MICB Ã 018 and Ã 019 alleles, which have not previously been reported in Koreans, were found in this study.

Discussion
The ASPE method depends for identifying alleles on the sequence-specific extension of immobilized allele-specific extension primers that differ at the 3 0 ends of their SNPs [11]. However, certain mismatches are not refractory to extension. The systemic characteristics and optimization of the ASPE procedure have been shown to be influenced by annealing temperature, template concentration, and Mg 2+ concentration [18]. Of the extension primers examined in this study, which were designed to have a length of 18 to 21 bases, 14 showed excellent sensitivity and specificity in the control panel (Table 3). Allele-specific PCR with Taq polymerase offers the greatest template discrimination (40-to 100-fold) against mismatches to thymine, guanine, or cytosine (T, G, or C) at the 3 0 end of a primer, but not against mismatch to adenine (A) [19]. To improve the accuracy, we used the thermostable Thermo Sequenase which was engineered to catalyze the incorporation of dNTPs with higher efficiency than other DNA polymerases on SBE-based microarray [20]. For primers with cytosine (C) as the 3 0 nucleotide, Thermo Sequenase was highly specific for template complementarity to this base (Table 4). By contrast, primers with thymine (T) at the 3 0 end were less efficiently amplified regardless of the corresponding nucleotide on the template strand. These results are directly applicable to the design of primers for SNP detection and have made it possible to develop a general procedure to increase the specificity of extension primers by specifying an additional variable nucleotide at the 3 0 end of the extension primer.
Different approaches have been proposed to improve the specificity of ASPE. As the specificity of primer-directed extension is not sufficient for quantitative SNP analysis, artificial mismatched bases have been introduced into the 3 0 end regions of the specific primers as a way of improving the switching characteristics of the primer extension reactions. The best position in the primer for such artificial mismatched bases is the third position from the 3 0 end of the primer [21]. In our control extension primers, mismatched control extension primers containing errors in the 2nd and 3rd bases from the 3 0 end gave a completely negative signal, whereas primers containing an error in the 1st base from the 3 0 end gave a weak signal (Fig 1). The specificity of extension primers targeting SNP region was also increased by the modification for refractory extension (Fig 2; Table 4). For further improvement of primer quality, apyrase, a nucleotide-degrading enzyme to the extension reaction, should be introduced to DNA microarrays [22,23]. We did not use apyrase in our assay and the quality of primers was improved by optimal modifications instead. We retested the reactions if the MFI value was between 10,000 and 30,000. When the same values appeared over three trials, the results were confirmed using other methods such as PCR-SBT (data not shown).
Although we designed 36 primers to detect almost all of the known 41 MICB alleles, only 22 primers could be defined or improved using the control panel in this study. Therefore, we could not define the other rare alleles and it is necessary to confirm the quality of the 14 remaining primers using samples from other populations. In Korean population, the MICB alleles detected by the genotyping microarray were fully in concordance with those detected by PCR-SBT and we compared the MICB population allele frequency (Korean population) with the test sample frequency in Table 6 (2n = 400). Two of the alleles found in this study, MICB Ã 018 and Ã 019 alleles, had not been previously found in Koreans [16]. Therefore, it was possible to identify 13 MICB alleles in this study using verified allele-specific extension primers. Our microarray did not include SNP in exon 6 because these SNPs are present only in very rare alleles which have been not reported in population data of published papers and are not present even in 163 samples of the UCLA International Exchange Report program. Although we designed primers to detect SNPs in exon 2 to 5 of MICB, new alleles have recently been defined by SNPs in exon 6 [24]. MICB Ã 010 and Ã 005:02 alleles could not be discriminated by ASPE on microarrays because the primers had not been designed to differentiate MICB Ã 005:02 allele from MICB Ã 010 allele in exon 6.
If the frequency of the minor allele is greater than 1%, such variants are called polymorphisms. These 13 MICB alleles were found more than 1% in reported ethnic populations including Welsh, Spanish, and Chinese and the other alleles are present less than 1% only in particular populations [5,10,16,[24][25][26]. Therefore, this method may be useful for disease association studies in various populations, even though it can not detect rare alleles.
We had previously developed a typing method of the SNPs of cytokine genes using ASPE on a fluorescence bead array [27]. According to this study and the bead array study, we showed that allele typing could not only be possible on liquid phase but also on solid phase. Although Sanger and next generation sequencing is readily available in labs and new alleles can be detected, this DNA chip-based assay allows for a multiplex assay to detect several mutations in addition to a being relatively rapid and cost-effective means of detecting defined mutations for genetic diagnosis.

Conclusions
We have established a system for genotyping MICB alleles using ASPE on microarrays. Available operation of control and allele-specific extension primers was representatively confirmed by schematic patterns of successful allele-specific extension primers that could discriminate 13 MICB alleles (S1 and S2 Figs). In conclusion, our method for genotyping MICB alleles using ASPE on microarrays could be applicable for large-scale SNP typing studies of population and disease associations.