Comparative analysis of NGS and Sanger sequencing methods for HLA typing at a Russian university clinic

Introduction

The Major Histocompatibility Complex (MHC) is among the most polymorphic genetic systems in humans. Over last decade, extensive research in HLA (Human Leukocyte Antigens) has revealed hundreds of new HLA alleles through intensive application of immunogenetic sequencing methods, including monoallelic Sanger-sequencing method, or, more recently, next-generation sequencing. In September 2018, the database of the World Health Organization (WHO) Nomenclature Committee for Factors of the HLA System (IPD-IMGT/HLA Database) contained information on the nucleotide sequences of 20272 diff erent HLA alleles, of which 14800 were HLA class I and 5288 founded for the HLA class II alleles [1-3].
During last 20 years, the automated Sanger technique has become a prevalent approach to genome sequencing in humans, animals, bacteria, and viruses. However, a need for more rapid routine genome screening required some novel technologies of multiplex DNA sequencing. It depicts these modern methods as the second-generation approaches (Next-Generation Sequencing, NGS). Th ese technological platforms based on diff erent strategies, regarding unique preparations of DNA templates, their sequencing, registration, retrieval and evaluation of the nucleotide sequences with novel bioinformatics approaches [4]. A principal benefit of the new-generation sequencing is an opportunity get large databases of multiple defi ned oligonucleotide sequences within a short time period with low costs.
Out of all known HLA loci, the relatively important and most commonly used for transplantation of hematopoietic cells are HLA – A, B, C, DRB1 and DQB1 (Fig. 1). The American Society for Histocompatibility and Immunogenetics (ASHI) established a catalogue of common and well‐documented (CWD) HLA. It is very commonly used now around the world as a great tool for resolving typing ambiguities in tissue transplantation or for checking the universality of any HLA allele in the world [5]. There established catalogues (database). Th e total number of CWD alleles is similar in the EFI (N = 1048) and ASHI (N = 1031) catalogues [6] (http://igdawg.org/cwd.html).
The importance of only Exons 2 and 3 for the Class I and Exon 2 for Class II is very well-known and designated as coding proteins involved in antigen presentation in the major histocompatibility complex (MHC) receptor grove in-between the two helices accommodates peptides and interaction between an alloantibody IgG complex.
HLA alleles having nucleotide sequences that encode the same protein sequence for the peptide binding domains (exon 2 and 3 for HLA class I and exon 2 only for HLA class II alleles) designated by an upper case ‘P’ which follows the allele designation of the lowest numbered allele in the group. HLA alleles that have identical nucleotide sequences for the exons encoding the peptide binding domains (exon 2 and 3 for HLA class I and exon 2 only for HLA class II alleles) designated by an upper case ‘G’ which follows the allele designation of the lowest numbered allele in the group.
The first two digits describe the allele family, which oft en corresponds to the serological antigen carried by the allotype. The third and fourth digits assigned in the order in which the sequences have been determined. Alleles whose numbers diff er in the fi rst four digits must diff er by one or more nucleotide substitutions that change the amino-acid sequence of the encoded protein. Alleles that differ only by synonymous nucleotide substitutions within the coding sequence distinguished by the use of the fifth and sixth digits. Alleles that only differ by sequence polymorphisms in introns or in the 5’ and 3’ untranslated regions that fl ank the exons and introns distinguished by the use of the seventh and eight digits [7].

Figure 1. Current mapping of HLA loci on the chromosome 6 [Robinson J et al.] http://www.hla.alleles.org/alleles/index.html

No wonder that the general NGS approach adapted for HLA typing proved to be a breakthrough in molecular biology applications being quite promising to the transplantation clinics and bone marrow donor registries. However, to promote the NGS implementation, we need specialized typing strategies and digital program algorithms. The sequencing costs per single run sharply decreased with NGS approach, which may be accessible to the small size tissue typing laboratories in a sooner time [8].
However, despite higher resolution of NGS [9], it was necessary to conduct a comparative analysis of control samples with "rare" genotypes. It is also important to understand the cost-eff ectiveness of diff erent methods. Hence, the aim of this pilot study was to evaluate the comparative advantages of using NGS and the Sanger sequencing approach, to identify rare HLA alleles, and to estimate the costs for the both different methods.

Materials and methods

The potential donor's test samples were obtained from the Bone Marrow Donor Registry at the First I. P. Pavlov State Medical University of St. Petersburg, Russia, Raisa Gorbacheva Memorial Institute of Children's Hematology, Oncology, and Transplantation, and various hematology patients undergoing HLA SBT testing for planned allogeneic hematopoietic stem cell transplantation.
Genomic DNA was isolated from peripheral blood leukocytes using MagNA Pure System (Roche Life Science). The target DNA concentration was from 10 to 140 ng/μL. Quantity and quality estimation of the isolated DNA was performed with Quantus Fluorometer TM (Promega Corp., USA). The main steps of the NGS as performed with Illumina platform (MiSeq, USA) using NGSgo protocol were as follows:
1. HLA locus-specifi c amplifi cation: the complete sequences of HLA genes are amplifi ed with allele-specifi c primers in a single reaction for each locus using Long-Range DNA polymerase;
2. DNA quantifi cation by Quantus Fluorometer and pooling of amplicons according to the volumes calculated by the NGSgo Pooling Calculation Sheet (provided by GenDx, Netherlands);
3. Double-stranded DNA fragmentation by means of specific fragmentase optimized by its size for the specifi c HLA locus, end repair, 5’ phosphorylation of poly-A and poly-T ends and adapter ligation (Fig. 2);
4. DNA cleanup and size selection with 0.45x SPRI beads using 80% ethanol (Beckman Coulter, AMPure XP);
5. Indexing PCR products using a unique combination of i5 and i7 primers for each sample;
6. Plate-based DNA cleanup and size selection with 0.6x SPRI beads using 80% ethanol (Beckman Coulter, AMPure XP);
7. Plate-based library pooling, library quantifi cation performed using Qubit Fluorometer and loading to the NGS sequencer (MiSeq, Illumina, USA);
8. Next-generation sequencing by MiSeq and data analysis.
The libraries are sequenced on an Illumina NGS platform. The FastQ data can be analyzed with an HLA typing soft ware package to determine the HLA typing (for example, NGS engine). To assign the HLA alleles, the soft ware allows communicating with updated IMGT database (Fig. 3).
The NGS method allows performing sequencing of all exons in the A, B, and C HLA loci and three exons (from second to fourth) of DQB1 and DRB1 (Fig. 4) which, however, has its limitations. Th e allele imbalances can be observed in some
rare cases:
• NGSgo HLA-DRB1: allele imbalances for DRB1*01, DRB*04, and DRB1*14 alleles can occur in case of imbalanced amplifi cation.
• NGSgo HLA-DRB4: allele imbalances for DRB2 exon 2 and exon 3 can occur in case of imbalanced DRB4 amplifi cation. In the case of an HLA-DRB4 exon 3 amplicon dropout, limit the analysis to exon 2 only.
• NGSgo HLA-DRB3/4/5: allele imbalances for heterozygous DRB3/4/5 samples can occur in case of imbalanced amplicon pooling. Analysis of DRB3/4/5 has been optimized in NGSengine v2.1 (and higher), which applies a split-analysis of the individual DRB3/4/5 loci to improve HLA typing.
Th e main steps of Sanger sequencing when performed with Applied Biosystems Genetic Analyzer (USA) 3500xl genetic analyzer using PROTRANS HLA SBT Class I and Class II S4 (Hockenheim, Germany, http://www.protrans.info/nano.cms/en/products/MainCatID/9/). Single Allele, Allele- Group and Locus Specific Sequencing. Fourteen specific primer mixes pre-pipetted in 8 and 16 well strip, in order of sequencing the Exons 1, 2, 3 and 4 for Class I, Exon 2 for DRB1 and Exon 2, 3 for DQB1, according to the manufacturers’ recommendations.

Figure 2. Gene library preparation with NGSgo – LibX and NGSgo – IndX referred from https://www.gendx.com [10]

Figure 3. Data analysis software (NGSengine, https://www.gendx.com)

Figure 4. Target generation with NGSgo – AmpX [10]

Direct automated fl uorescent DNA sequencing was performed by a 24-channel automated capillary electrophoresis system, and fl uorescent detection of DNA fragments using an Applied Biosystems GA3500xl Genetic Analyzer. Capillary electrophoresis proceeded in the POP-7 polymer under denaturation conditions. Th e data on nucleotide sequences were retrieved at a stationary computer in the Data Collection program, then having been analyzed by Protrans SEQUENCE PILOT soft ware (Hockenheim, Germany).
DNA amplification kits for Sanger sequencing are designed to provide high-resolution identifi cation of alleles of the human HLA-A, -B, -C, -DRB1, -DQB1 genes.

Results

The aim of our pilot study was a comparison of two methods and an evaluation of their eff ectiveness. To achieve our purpose, we selected a group of 35 persons (see Materials and methods), and conducted analysis by Sanger and NGS method in parallel. Th e NGS method allowed detecting rare variants of alleles when performing data analysis with NGSengine soft ware (Fig. 5).
We have conducted two sets of experiments. Mean coverage in the fi rst experiment was 881x – (1010x, 897x, 768x, 807x, 923x, respectively for A-, B-, C-, DRB-, DQB- loci), and 992x in a second experiment (1194x, 856x, 698x, 1001x, 1346x, respectively for A-, B-, C-, DRB-, DQB- loci).
Mean percentage of aligned reads to the total read number was 96.5% in the fi rst set (DRB locus, 92.6%, other loci, >97.2%). In the second set, an appropriate percentage of aligned reads to the total read number was 95.0% (DRB locus, 91.6%, other loci, >95.5%) (Fig. 6). This metrics shows a high quality of the sequencing that was performed according to the manufacturer’s instructions.
To perform a more detailed analysis of each sample, the NGSengine soft ware contains the sections of «typing results» and «visualization», where the coverage for diff erent regions may be registered in more details, or a nucleotide position of interest should be found (Fig. 7).

Figure 5. A typical data evaluation table presented by NGSengine software (Genome Diagnostics, Netherlands)

It presents information for each locus (HLA-A, -B, -C, -DRB1, -DQB1) for single samples. Data on total read number and percentage of aligned reads for the given locus, mean read length, mean coverage, alleles identifi ed and presence of synonymous substitutions in coded [Ex] and it also displays non-coding [In] regions.

Figure 6. Statistics for samples (percentage aligned reads from total number of reads and number of reads mapped to the reference per strand) in NGSengine software

Figure 7. The results of the NGS sequencing all alleles for sample “3” and visualization of HLA-A locus for sample “3” in NGSengine software

It displays typing results for all HLA loci assayed. Th e cases of ambiguous results shown as Allele Ambiguities. In our series, no ambiguities were detectable for any locus. The figure visualization allows us to look at the visual segment (it shows exons in yellow). It indicates the sequencing coverage of the given locus below (marked gray). The vertical ticks seen at appropriates points of HLA loci in cases of synonymous nucleotide substitutions.

Table 1. Comparison of allele sequenced by NGS and Sanger’s method – 100% homology results (there are no differences in 2nd and 3rd exons sequences)

Factors contributing to the costs arising for the in-depth sequencing

The reagents for the entire HLA-sequencing process include those used for routine pre-analytic steps (e.g., DNA extraction, quality assessment, and initial low-resolution typing step). Additional expenditures are subject to some ambiguities, due to diff erent prices for reagents and equipment off ered by distinct manufacturers. Moreover, it should be addressed that all the commercial NGS platforms off er their closed-type systems, thus causing broad variations in prices for the entire NGS procedure per single DNA sample, strongly depending on the annual capacity of the given HLA typing laboratory.
However, even considering maintenance costs (about 10% equipment cost), enrolling third-party core facilities or shared equipment, the Sanger sequencing (220 K) proves to be twice more expensive than NGS (variable, but still less than Sanger technique), as shown elsewhere [11]. Hence, the sample preparation costs remain the same whereas the sequencing tends to decrease as discussed in [9]. Calculations of economic efficiency for HLA typing in Russia by Sanger technique versus NGS were among our major tasks. Therefore, we have performed a pricing for the sequencing kits at a company providing reagents to this purpose. Th e request was made twice (February 2017 and August 2018). The reagent price in Euros did not change suffi ciently. Of note, the sequencing kits by Sanger are produced for 25 or 100 tests, whereas NGS kits are off ered for 24 and 96 tests.
Clear benefit of Sanger approach is that a single locus may be sequenced in the sample, being, however, economically ineff ective when using NGS technology.
Hence, we have compared the panels for 100 tests covering fi ve main HLA loci, i.e., AlleleSEQR HLA-A PCR/Sequencing Mix, AlleleSEQR HLA-B PCR/Sequencing Mix, AlleleSEQR HLA-C Plus PCR/Sequencing Mix, AlleleSEQR HLA-DRB1 PCR/Sequencing Mix, AlleleSEQR HLA-DQB1 PCR/Sequencing Mix). Each set was purchased for 6250 Euros. Hence, the total cost of locus-specifi c reagents for 5 loci was 31250 Euros, thus providing 312.50 Euros per 1 human DNA sample (ca. 24,000 roubles as per October 2018). The prime costs should also include disposables for the core sequencing procedure. E.g., if performing 100 tests for 5 HLA loci, we run 2500 reactions with a 24-channel Applied Biosystems GA3500xl Genetic Analyzer ABI 3500. To start the process, the following general items are needed: 26 plates for the gene analyzer (MicroAmp 96 Well Reaction Plate); three universal polymers. For capillary electrophoresis, POP-7 for 960 samples, fi ve Formamide packs (25 mL each), containers with anode and cathode buff ers etc., at a total price of 4500 Euros. Hence, the prime costs of Sanger reagents, when sequencing 5 main loci at the full-load regimen and usage of a 100-test kit, makes about 35750 Euros (ca. 2681250 roubles), excluding costs for pipette tips, microtubes, gloves and other inexpensive disposables. Th at means ca. 27000 roubles per one human DNA sample.
When applying NGS approach, the number of samples taken into analysis is quite suffi cient, since even a high-throughput MiSeq machine may perform sequencing of up to 269 samples in parallel, using a standard 4.5-Gb cartridge.
To calculate costs of comparable NGS analysis, we have chosen a reagent set for sequencing of 96 samples which incuded the following items: NGSgo®-AmpX HLA-A, B, C, DRB1, DQB1, NGSgo®-LibrX Library Preparation (2 kits), NGSgo ®-IndX Adapter & Indices (4x24) RUO Illumina, Agencourt AMPure XP 5 mL Kit, GenDx LongRange polymerase (3 kits), MiSeq Reagent Kit v2. A total sum for typing 5 loci made ca. 14 800 Euros, thus comprising 155 Euros (>10000 Roubles) per sample.
However, in case of spared use of the reagents for sample preparation by two-fold decrease in reaction volume (as proven by our experience), the prime costs per a single test dropped to 90 Euros. Th e self-cost is here provided without accessory disposables.
Hence, the costs of sequencing reagents, even without their sparing, is suffi ciently cheaper when using NGS technologies. Moreover, the procedure takes 2-fold less time for its performance than the Sanger technique.

Conclusion

Comparison of allele sequenced by NGS and Sanger’s method yielded 100% homology results. Hence, our work is in accordance with previously published data [9], which demonstrate the advantage and effi ciency of NGS, as compared to Sanger sequencing.
NGS-based HLA analysis is performed with a 100% reliability, and well fi ts the tasks of HLA typing in unrelated donors, in concordance with EFI and ASHI policies. Th is work process well corresponds to the working schedules for medium- and high-capacity laboratories, thus being potentially attractive to the donor registries. Recently introduced next-generation sequencing techniques have a facilitating potential for the high-resolution genotyping via a decrease of general ambiguity of end results, like as due to more extended sequencing regions. In near future, the NGS approaches will be an eff ective and cost-eff ective technology when evaluating histocompatibility parameters and immunogenetic interactions.

Conflict of interest

The authors have declared no conflicting interests.

Funding

This research was funded by Russian Science Foundation grant №14-50-00069.

References

1. Robinson J, Halliwell JA, Hayhurst JH, Flicek P, Parham P, Marsh SGE. Th e IPD and IMGT/HLA database: allele variant databases. Nucleic Acids Research. 2015; 43:D423-431.
2. Marsh SGE, Albert ED, Bodmer W, Bontrop RE, Dupont B, Erlich HA, Fernández-Viña M, Geraghty DE, Holdsworth R, Hurley CK, Lau M, Lee KW , Mach B, Maiers M, Mayr WR, Müller CR, Parham P, Petersdorf EW, Sasazuki T, Strominger JL, Svejgaard A, Terasaki PI, Tiercy JM, Trowsdale J. Nomenclature for factors of the HLA system, 2010. Tissue Antigens. 2010; 75(4):291-455.
3. Marsh SGE. Nomenclature for factors of the HLA system, update July 2017. Human Immunol. 2017; 78(11-12):758- 761.
4. Holcomb CL, Höglund B, Anderson MW, Blake LA, Böhme I, Egholm M, Ferriola D, Gabriel C, Gelber SE, Goodridge D, Hawbecker S, Klein R, Ladner M, Lind C, Monos D, Pando MJ, Pröll J, Sayer DC, Schmitz-Agheguian G, Simen BB, Th iele B, Trachtenberg EA, Tyan DB, Wassmuth R, White S, Erlich HA.A multi-site study using high-resolution HLA genotyping by next generation sequencing. // Tissue Antigens. 2011;77(3):206-217. doi: 10.1111/j.1399-0039.2010.01606.x.
5. Mack SJ1, Cano P, Hollenbach JA, He J, Hurley CK, Middleton D, Moraes ME, Pereira SE, Kempenich JH, Reed EF, Setterholm M, Smith AG, Tilanus MG, Torres M, Varney MD, Voorter CE, Fischer GF, Fleischhauer K, Goodridge D, Klitz W, Little AM, Maiers M, Marsh SG, Müller CR, Noreen H, Rozemuller EH, Sanchez-Mazas A, Senitzer D, Trachtenberg E, Fernandez-Vina M. Common and well-documented HLA alleles: 2012 update to the CWD catalogue. Tissue Antigens. 2013 Apr;81(4):194-203. doi: 10.1111/tan.12093.
6. A. Sanchez‐Mazas, J. M. Nunes, D. Middleton, J. Sauter, S. Buhler, A. McCabe, J. Hofmann, D. M. Baier, A. H. Schmidt, G. Nicoloso, M. Andreani, Z. Grubic, J.‐M. Tiercy, K. Fleischhaue.r Common and well‐documented HLA alleles over all of Europe and within European sub‐regions: A catalogue from the European Federation for Immunogenetics. HLA Volume 89, Issue2.February 2017 Pages 104-113.
7. S. G. E. Marsh, E. D. Albert, W. F. Bodmer, R. E. Bontrop, B. Dupont, H. A. Erlich, M. Ferna´ndez-Vin˜ a, D. E. Geraghty, R. Holdsworth, C. K. Hurley, M. Lau, K. W. Lee, B. Mach, M. Maiers, W. R. Mayr, C. R. Mu¨ ller, P. Parham, E. W. Petersdorf, T. Sasazuki, J. L. Strominger, A. Svejgaard, P. I. Terasaki, J. M. Tiercy & J. Trowsdale. Nomenclature for factors of the HLA system, 2010 Tissue Antigens 75, 291-455.
8. Kuzmich EV, Alyanskiy AL, Tyapushkina SS, Nasredinova AA, Ivanova NE, Zubarovskaya LS, Afanasyev BV. Identifi cation of the new HLA-B*44:02:45, DQB1*02:85, DQB1*06:210, DRB1*01:01:30 alleles by monoallelic Sanger sequencing. Cell Th er Transplant. 2018. 7(1):62-66. 9. Serov YA, Barkhatov IM, Klimov AS, Berkos AS. Current methods and opportunities of next-generation sequencing (NGS) for HLA typing // Cell Th er Transplant. 2016; 5(4): 63-70. doi: 10.18620/ctt-1866-8836-2016-5-4-63-70.
10. https://www.gendx.com
11. Baxter-Lowe LA. Tailoring NGS for smaller volume labs. Proc. 42nd ASHI Annual Meeting. Abstract: Sept 28, 2016.