- Methodology article
- Open Access
A Bead-based Normalization for Uniform Sequencing depth (BeNUS) protocol for multi-samples sequencing exemplified by HLA-B
BMC Genomics volume 15, Article number: 645 (2014)
Human leukocyte antigen (HLA) is a group of genes that are extremely polymorphic among individuals and populations and have been associated with more than 100 different diseases and adverse drug effects. HLA typing is accordingly an important tool in clinical application, medical research, and population genetics. We have previously developed a phase-defined HLA gene sequencing method using MiSeq sequencing.
Here we report a simple, high-throughput, and cost-effective sequencing method that includes normalized library preparation and adjustment of DNA molar concentration. We applied long-range PCR to amplify HLA-B for 96 samples followed by transposase-based library construction and multiplex sequencing with the MiSeq sequencer. After sequencing, we observed low variation in read percentages (0.2% to 1.55%) among the 96 demultiplexed samples. On this basis, all the samples were amenable to haplotype phasing using our phase-defined sequencing method. In our study, a sequencing depth of 800x was necessary and sufficient to achieve full phasing of HLA-B alleles with reliable assignment of the allelic sequence to the 8 digit level.
Our HLA sequencing method optimized for 96 multiplexing samples is highly time effective and cost effective and is especially suitable for automated multi-sample library preparation and sequencing.
To date, several high-throughput HLA typing methods using next-generation sequencing (NGS) have been developed [1–8]. In our previous study, we completely sequenced long-range PCR amplicons encompassing entire regions of each of the HLA genes (HLA-A, -C, -B, -DRB1, -DQB1, and -DPB1). PCR amplicons were subjected to transposase-based library construction and multiplex sequencing with the MiSeq sequencer. Paired-end reads of 2 × 300 bp enabled us to demonstrate phase-defined allele determination (also defined as HLA gene haplotypes) in 33 HLA homozygous samples, 11 HLA heterozygous samples, and 3 parent–child families. Our sequencing protocol and pipeline provided essentially complete phase-defined HLA gene sequences; however, it required complicated and labor-intensive workflows especially in the library preparation step. Most importantly, the method is not well adapted for processing multiple samples. In the present study, we developed a new library preparation method for NGS and applied it to 96 samples. Long-range PCR products of HLA-B spanning from promoters to 3′-UTRs were prepared and sequenced with the MiSeq sequencer via transposase-based library preparation. In the previous protocol, although the DNA amount of each library was strictly measured and the library size was validated using a BioAnalyzer before the sequencing step, it was difficult to equalize the DNA amount and molecular size of the libraries, resulting in variable numbers of reads in each sample. We observed dropouts of samples owing to insufficient reads. Here, we developed a Bead-based Normalization for Uniform Sequencing (BeNUS) procedure using three steps of bead purification. BeNUS can easily and precisely normalize the molar concentrations of up to 96 samples, not only simplifying the library preparation step but also permitting automation of HLA typing using NGS.
Results and discussion
PCR amplification of HLA-Band library preparation
We applied long-range PCR to amplify HLA-B, which is known to be highly polymorphic. HLA-B-specific amplification products were obtained from 96 individuals. All 96 PCR amplicons were subjected to transposase-based library construction using the Nextera kit, which simultaneously fragments the DNA and adds adaptors needed for multiplex sequencing. We developed protocol steps using altered AMPure XP beads to normalize the molar concentrations of 96 samples. Each PCR amplicon was subjected separately to transposase-based library construction, whereby a sample-specific index was introduced. The Nextera kit can construct libraries of a broad size range. For phase-defined HLA sequencing, a library range of 500–1,000 bp is desirable. In the previous protocol , the library size selection was achieved by cutting of agarose gel and checking using a BioAnalyzer, which is a very laborious step, especially for preparing multiple samples.
Our new method, BeNUS, which is described in detail in the Methods section, circumvents the gel cutting method and employs bead-based steps for library size selection as well as equalization of DNA molar concentrations in up to 96 samples. More specifically, three bead steps were performed: first, 20 μl of altered beads suspended in 20% PEG and 2.5 M NaCl solution was added to 50 μl of diluted PCR product, and the supernatant fraction containing the desired fragments (<1,000 bp) was collected. Second, 5 μl of beads was added to the collected supernatant. Desired fragments of larger than 500 bp were bound to the beads, while smaller fragments remaining in the supernatant were discarded (Figure 1). After these two steps, the desired DNA fragments (500–1,000 bp) were selected (actual size: 492 to 1,625 bp, average size: 924 bp) (Additional file 1: Figure S1). Finally, 20-fold diluted beads were added to the size-fractionated library. Small numbers of beads can bind saturated amounts of DNA because bound DNA is in proportion to the number of beads (Additional file 2: Figure S2). Eventually, the final DNA size, concentration, and thereby molar concentration were equalized (Additional file 3: Figure S3).
Complete sequencing of HLA-Bfor 96 samples
The Nextera-treated libraries from 96 HLA-B PCR amplicons were subjected to NGS sequencing. The distribution of reads among the 96 multiplexed indexes ranged from 0.2% to 1.55% as percentages of 24.6 million reads and 1.04 Â± 0.32% on average. Technical improvements in libraries preparation were clarified by comparison between our previous method, cutting of agarose gel followed by the BioAnalyzer, and BeNUS method. The distribution of respective read numbers by the previous method was 0.06% to 2.74% and the value by BeNUS method was 0.2 to 1.55% (Figure 2). Sequence reads of HLA-B, were aligned to the reference sequence at an average mapped rate of 99.63%, ranging from 99.11 to 99.87%. The average depth of the alignment before phasing was 6495.3× and the phased haplotype depth was 2097.8×. Of the 96 samples, heterozygous SNVs and indels were not observed in 8 samples (0605, 0607, 0645, 1057, 1169, 1184, 1199, and 1229) indicating completely homozygous haplotypes. Generally, MiSeq yields base errors at a rate of 0.8% with PCR free library preparation . In the result of these homozygous samples, the total error rate of sequencing reaction and PCR amplification was estimated to be as less than 1.2%. In our simulated alignment result, the maximum error rates in 300× and 200× average depth were 4.8% and 6%, respectively, and the minimum depth for complete phasing was approximately 800× average depth. We recognize phasing result is not only dependent on average depth but also on distance between two heterozygous SNVs and long insert library covering the two heterozygous SNVs. As a general method, we propose that sequence reads of at least 800× depth is valid for providing the phase-defined HLA sequences with for high accuracy (less than 5% error rate). All samples were completely phased by the phase-defined sequencing pipeline , although six samples showed only 1 to 7 SNVs in one exon: samples 0785, 1018, 1117, and 1175 had 4, 7, 3, and 3 heterozygous SNVs in exon 2, respectively. Sample 1224 had 3 heterozygous SNVs in exon 3, whereas sample 0979 had one heterozygous SNV in exon 5 (Figure 3). All the above samples were also phased using partial phasing (Additional file 4: Figure S4). Allelic imbalance as a result of PCR is manifested by skewed allelic calls after HLA sequencing. Allelic imbalance of the PCR amplification of HLA-B was negligible, as judged by the ratio of sequencing depth between the two phased alignments; 1:1.68 at the maximum and 1:1.19 on average in heterozygous samples (Table 1). Consequently, 192 haplotype sequences of 96 individuals, which include 28 different haplotype sequences, were constructed as phased allelic HLA-B sequences. In general, allelic imbalance after PCR amplification has been occasionally observed. Evidently, this problem is not specific to NGS analyses. If capture, NGS, or the analytical step are causes of allelic imbalance, we would observe the discrepancy between our NGS typing and SSO typing, which was not the case in the current study. The current protocol aimed at minimizing the disparity of each sample; this was achieved although not perfectly. The minimum depth could be important to obtain phase-defined sequences because lower depth could leave unphased region. However, complete phase-defined sequencing is dependent on the allelic type of the sample, thereby it is not easy to give an exact depth number to accomplish the sequencing in general. In the current study, the most important point of the current study was to obtain complete phase-defined sequences of 96 samples without any dropout, which was achieved using the current protocol.
After obtaining phase-defined HLA gene sequences for 192 haplotype sequences, we attempted to assign HLA allele numbers to these sequences by searching for known allele sequences in the IMGT/HLA database. We used phased HLA-B haplotype sequences that spanned all of the intronic and exonic regions as queries against genomic and CDS sequences in the database. The determined HLA-B allele calls in all samples were consistent with PCR-SSO and Omixon Target, although the PCR-SSO determined HLA allele numbers were limited to a 4-digit resolution.
We established a simple, high-throughput, high-resolution, and high-fidelity HLA sequencing and genotyping method, as a combination of the Nextera kit and our new BeNUS method. We successfully applied our method for HLA-B sequencing in 96 samples, without dropouts. By developing BeNUS, it becomes feasible to construct multi-libraries without agarose gel size selection and DNA density control of each library. This method would be greatly advantageous for clinical applications that require a user-friendly and cost-effective protocol, with high throughput and accuracy. Our protocols open a way to prepare NGS libraries for large-scale HLA gene sequencing and typing using an automated system.
A total of 96 unrelated healthy Japanese control subjects were recruited at the Health Evaluation and Promotion Center of Tokai University Hospital. All subjects gave written informed consent for the study. Ethical approvals for this study protocol were obtained from the IRBs of National Institute of Genetics and Tokai University School of Medicine.
DNA samples were extracted from peripheral blood using a DNA extraction kit Genomix (Biologica, Nagoya, Japan) using the manufacturer’s instructions.
HLA genotyping with PCR-SSO method
We genotyped HLA-B using the Luminex assay system and HLA typing kits (WAKFlow HLA Typing kits, Wakunaga, Osaka, Japan or LABType SSO, One Lambda, Canoga Park, CA, USA).
HLA-B was amplified using locus-specific primers by long-range PCR . Each amplification reaction contained 20 ng of genomic DNA, 0.25 unit of PrimeSTAR® GXL DNA polymerase (TAKARA BIO Inc., Shiga, Japan), 1× PrimeSTAR® GXL buffer (Mg2+ concentration 1 mM), 0.2 mM of each dNTP, and 0.2 μM of each primer in a 10 μl reaction volume. Cycling parameters were as follows: initial denaturation of 94°C for 2 min followed by 30 cycles of 98°C for 10 s, 60°C for 15 s, and 68°C for 5 min. Each PCR product concentration was measured with a Qubit dsDNA BR Assay Kit (Life Technologies, Carlsbad, CA, USA). PCR products were subjected to library preparation with a Nextera DNA Sample Preparation Kit (Illumina, San Diego, CA, USA) and a KAPA Library Amplification kit (Kapa Biosystems, Inc., Wilmington, MA, USA). The KAPA kit was used for library amplification because of its advantage of coverage depth in high-GC-content regions during library amplification (Additional file 5: Figure S5). Each sample was dual indexed and normalized with modified AMPure XP beads (Beckman Coulter Inc., Brea, CA, USA) method, which include optimal size selection and normalization of DNA concentration (Figure 1).
BeNUS for 96-well plate-based library
We prepared altered AMPure XP beads by resuspending the beads in half of the original volume of 20% polyethylene glycol 8000 (PEG) and 2.5 M NaCl solution. The resuspended beads were twice as concentrated as the standard AMPure XP beads (Additional file 6: Figure S6). The optimal volume ratio of altered beads to DNA solution was determined by relation between the PEG–NaCl concentration and the selected DNA fragment size (Additional file 6: Figure S6). For library size selection, 20 μl of resuspended altered beads was added to 50 μl of diluted library (5 μl of PCR product and 45 μl of water), mixed well by pipetting at least 10 times, and incubated for 5 min at room temperature. The tube was placed on the NGS MagnaStand (NIPPON Genetics, Tokyo, Japan) to separate the beads from the supernatant. After separation, the supernatant was carefully transferred to a new tube. Five μl of the altered beads were added to the supernatant, mixed, and incubated, and the beads were then separated from the supernatant on the same conditions. The supernatant containing unwanted DNA was carefully removed. One hundred μl of 80% ethanol was added to the tube for washing and the supernatant was carefully discarded after incubation and separation. The beads were air-dried for 10 min while the tube was on the magnetic stand with the lid open. The target library was eluted from the beads into 22 μl of water (Figure 1). Next, to normalize DNA concentration, altered beads, which were diluted 20 fold with 20% PEG and 2.5 M NaCl solution, were used to capture a certain amount of DNA (Additional file 2: Figure S2). Twenty μl of 20-fold diluted altered beads and 20 μl of isopropanol were added to the 20 μl of size-selected library, mixed well by pipetting at least 10 times, and incubated for 5 min at room temperature. The tube was placed on the magnetic stand to separate the beads from the supernatant. After the supernatant was discarded, 100 μl of 80% ethanol was added to the tube kept on the magnetic stand and incubated at room temperature for 30 s, and then the supernatant was carefully discarded. The beads were air-dried, and then 20 μl of water was added to elute the normalized libraries (Figure 1). The combination of fragment size selection and normalization of DNA amount results in an equalized DNA molar concentration among 96 libraries.
Equal volumes of libraries were pooled and subjected to multiplex sequencing on the MiSeq sequencer (Illumina). The MiSeq flow cell of 2 × 300 bp paired-end reads resulted in 11.6 million read pairs corresponding to 6 Gbp of valid sequence data without adapter sequence.
Determination of the HLA-Bsequence
Sequence reads were distributed according to index information to assign samples. We used the phase-defined sequencing pipeline (http://p-galaxy.ddbj.nig.ac.jp) [1, 10], which include trimming low quality bases (Phred quality score < Q20), selection of only long (>200 bp) paired-end reads, alignment to reference sequence, SNVs and indel identification, and haplotype phasing. HLA-B sequence (UCSC hg19, chr6: 31,317,316 – 31,331,864, complement) was used as reference sequence. After phasing, two BAM files were created as phased HLA-B alignments. The IGV genome viewer  was used to visualize the alignment results. The consensus sequences in FASTQ or FASTA format were constructed for searching the IMGT/HLA (http://www.ebi.ac.uk/imgt/hla/) database to identify the HLA alleles. We used them as query for BLAT  search to known HLA allele sequences in the database as complete matches of genomic sequence or CDS sequence. For validation, HLA-B genotype calls were compared with the result of PCR-SSO and Omixon Target HLA Typing (Omixon Inc., Budapest, Hungary) [13, 14].
Availability of supporting data
The data set associated with this project has been submitted to DDBJ Sequence Reads Archive (DRA accession number: DRA001289).
Hosomichi K, Jinam TA, Mitsunaga S, Nakaoka H, Inoue I: Phase-defined complete sequencing of the HLA genes by next-generation sequencing. BMC Genomics. 2013, 14: 355-
Bentley G, Higuchi R, Hoglund B, Goodridge D, Sayer D, Trachtenberg EA, Erlich HA: High-resolution, high-throughput HLA genotyping by next-generation sequencing. Tissue Antigens. 2009, 74: 393-404.
Lind C, Ferriola D, Mackiewicz K, Heron S, Rogers M, Slavich L, Walker R, Hsiao T, McLaughlin L, D'Arcy M, Gai X, Goodridge D, Sayer D, Monos D: Next-generation sequencing: the solution for high-resolution, unambiguous human leukocyte antigen typing. Hum Immunol. 2010, 71: 1033-1042.
Erlich RL, Jia X, Anderson S, Banks E, Gao X, Carrington M, Gupta N, DePristo MA, Henn MR, Lennon NJ, de Bakker PI: Next-generation sequencing for HLA typing of class I loci. BMC Genomics. 2011, 12: 42-
Wang C, Krishnakumar S, Wilhelmy J, Babrzadeh F, Stepanyan L, Su LF, Levinson D, Fernandez-ViÃ±a MA, Davis RW, Davis MM, Mindrinos M: High-throughput, high-fidelity HLA genotyping with deep sequencing. Proc Natl Acad Sci U S A. 2012, 109: 8676-8681.
Lank SM, Wiseman RW, Dudley DM, O’Connor DH: A novel single cDNA amplicon pyrosequencing method for high-throughput, cost-effective sequence-based HLA class I genotyping. Hum Immunol. 2010, 71: 1011-1017.
Lank SM, Golbach BA, Creager HM, Wiseman RW, Keskin DB, Reinherz EL, Brusic V, O'Connor DH: Ultra-high resolution HLA genotyping and allele discovery by highly multiplexed cDNA amplicon pyrosequencing. BMC Genomics. 2012, 13: 378-
Shiina T, Suzuki S, Ozaki Y, Taira H, Kikkawa E, Shigenari A, Oka A, Umemura T, Joshita S, Takahashi O, Hayashi Y, Paumen M, Katsuyama Y, Mitsunaga S, Ota M, Kulski JK, Inoko H: Super high resolution for single molecule-sequence-based typing of classical HLA loci at the 8-digit level using next generation sequencers. Tissue Antigens. 2012, 80: 305-316.
Quail MA, Smith M, Coupland P, Otto TD, Harris SR, Connor TR, Bertoni A, Swerdlow HP, Gu Y: A tale of three next generation sequencing platforms: comparison of Ion Torrent, Pacific Biosciences and Illumina MiSeq sequencers. BMC Genomics. 2012, 13: 341-
Nagasaki H, Mochizuki T, Kaminuma E, Watanabe S, Morizaki S, Kodama Y, Saruhashi S, Takagi T, Okubo K, Nakamura Y: DDBJ read annotation pipeline: a cloud computing based pipeline for high-throughput analysis of next generation sequencing data. DNA Res. 2013, 20: 383-390.
Thorvaldsdóttir H, Robinson JT, Mesirov JP: Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration. Brief Bioinform. 2013, 14: 178-192.
Kent WJ: BLAT–the BLAST-like alignment tool. Genome Res. 2002, 12: 656-664.
Major E, Rigó K, Hague T, Bérces A, Juhos S: HLA typing from 1000 genomes whole genome and whole exome illumina data. PLoS One. 2013, 8: e78410-
Kulski JK, Suzuki S, Ozaki Y, Mitsunaga S, Inoko H, Shiina T: In phase HLA genotyping by next generation sequencing - a comparison between two massively parallel sequencing bench-top systems, the Roche GS Junior and ion torrent PGM. HLA and Associated Important Diseases. Edited by: Xi Y. 2014, 143-181. InTech
This work was supported by MEXT KAKENHI Grant Numbers 22133002 and 221S0002 (K.H, I.I). We are grateful to Mr. Masayuki Tanaka, the Education and Research Support Center, Research and Development Division, Isehara, Tokai University, and Dr. Yasukazu Nakamura and Dr. Eli Kaminuma, Center for Information Biology and DNA Data Bank of Japan, National Institute of Genetics for their in silico work, and Dr. Timothy A. Jinam, Division of Human Genetics, National Institute of Genetics, for comments on our manuscript. We express our deepest gratitude to Mr. Ryota Sugimoto, Ms. Junko Kajiwara, and Ms. Yuko Okudaira for their technical assistance.
The authors declare that they have no competing interests.
KH and II designed the research and prepared the manuscript. KH and HN developed the bioinformatics method. SM performed HLA typing using the SSO method. All authors read and approved the final manuscript.