Optimization of HIV Sequencing Method Using Vela Sentosa Library on Miseq Ilumina Platform

1. Introduction

Human immunodeficiency virus (HIV) infection remains a major public health problem. Human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS) has caused the deaths of 40.4 million people since the start of the epidemic in the 1980s [1]. According to the latest available data, in 2022, 630,000 people died of HIV/AIDS-related illnesses, and 1.3 million people contracted HIV worldwide [1]. In France, 6.5 million HIV serology tests were performed by medical diagnostic laboratories in 2022 [2]. According to the estimates made by the French Ministry of Health, between 4200 and 5700 persons were tested positive for HIV in 2022. This number has remained stable for several years, regardless of the mode of contamination or place of birth of those who were diagnosed to be HIV-positive. In 2022, 28% of HIV infections were discovered at an advanced stage of infection in France, and this proportion has remained stable over the past few years.

The World Health Organization (WHO) recommends two nucleoside reverse transcriptase inhibitors (NRTIs) plus a non-nucleoside reverse transcriptase inhibitor or an integrase inhibitor for the standard first-line antiretroviral (ARV) therapy in adults and adolescents [3]. In adults, the second-line ARV therapy consists of two NRTIs combined with a ritonavir-boosted protease inhibitor. ARV treatment works best when taken as prescribed by medical practitioners. Poor compliance can lead to drug resistance, enabling HIV to multiply and the disease to progress.

In addition to poor compliance, another major obstacle to a successful ARV therapy is the development and acquisition of drug resistance due to mutations in HIV strains. Genotypic tests for ARV resistance are widely used to improve the therapeutic management of patients [4,5]. They are recommended at the time of diagnosis, i.e. before the initiation of treatment, in order to search for possible transmission of resistance, which is detected in around 10% of new infections for non-NRTIs and 3% for anti-proteases [4]. Further genotypic tests are strongly recommended in the event of ARV treatment failure. Genotypic tests analyse the sequences of genes that code for viral reverse transcriptase, protease, and integrase, in comparison with those of a wild-type HIV reference strain, to identify the mutations associated with resistance [6,7,8,9].

At the Versailles Hospital Centre in Versailles, France, personal genomic machine (PGM) Ion Torrent high-throughput sequencing technology has been used, together with a commercial kit from Vela Diagnostics, since 2021 to diagnose HIV. Compared with the older diagnostic techniques, such as the Sanger dideoxy method, Next-Generation Sequencing (NGS) enables simultaneous reads of millions of copies of HIV genomes per run, enabling the detection of low-frequency mutations. In addition, NGS approaches enable the detection of admixtures, which is particularly important for patient follow-up [10]. In 2019, Vela Diagnostics NGS obtained the United States Food and Drug Administration (FDA) approval for HIV genotyping and resistance testing for in vitro diagnostic purposes [11].

A recent Italian study that evaluated the Vela Diagnostics system showed better results with this NGS technology, compared to Sanger sequencing, in terms of success rate, accuracy, repeatability, and reproducibility of results [12]. In the present study, for the first time, we used a Vela Sentosa kit to sequence HIV genomes using the Illumina Miseq platform. The aim of our study was to show that it is possible to adapt the Vela kit to Miseq and obtain better results at a lower cost than with the PGM Ion Torrent sequencer.

2. Materials and Methods

2.1. Test validation samples

The reference HIV samples for quality control (QC) and test validation were obtained from the French National Reference Centre for HIV (Centre National de Référence HIV; CNR) in Bichat hospital, Paris, France. These samples are produced from the supernatants of the reference HIV strains under in vitro cultivation. Each year, QC samples are distributed and tested in several laboratories in France and at the Pasteur Institute (Paris, France) to assess the quality of HIV tests being performed. In addition to the QCs, we added a positive HIV control (AcroMetrixTM Genotyping HIV RT/PR mutant control) and a negative control (HIV-free PCR).

2.2. RNA extraction

All samples were centrifuged at 17000 g for 2 h at 4°C. After centrifugation, 400 µl of the supernatant were used to extract the viral RNA using the EZ1 virus mini kit (Qiagen S.A.S., Courtaboeuf, France) as recommended by the manufacturer.

2.3. PCR amplification

Reverse transcriptase-polymerase chain reaction (RT-PCR) was performed using the Sentosa^® SQ HIV drug resistance mutations (DRM) library kit (Vela Diagnostics, Hamburg, Germany) according to the manufacturer's recommendations. Four primer pools were used to amplify the protease (PR), reverse transcriptase (RT), and integrase (INT) genes. Each sample was tested in quadruplicate according to the primer pool used in a single PCR program.

2.4. Library preparation for Ion Torrent PGM platform

The library was prepared according to the Sentosa^® SQ HIV DRM (Experienced User Card V1.0) manual protocol (Vela Diagnostics, Hamburg, Germany). After amplification, the PCR products were normalized and purified using the Sentosa^® SQ virus solution prep kit. For each sample, 13 µl of the purified PCR product were used for enzymatic fragmentation using the Sentosa^® SQ HIV DRM library kit. The ligation step was performed using 1 µl adapter and 1 µl barcode in the Ion Xpress Barcode Adapters kit (Thermo Fisher Scientific, Paris, France). The ligation program was 20°C for 15 min, 65°C for 5 min, and 85°C for 2 min. After ligation, each ligated PCR product was purified with magnetic beads and eluted with 25 µl elution buffer. For each sample, 10 µl were taken to constitute a pool of all the libraries. The pool was then diluted by 1/4 with the elution buffer, and PCR and emulsion enrichment were performed on the Ion Chef instrument prior to sequencing. The three viral genes, PR, RT, and INT, were sequenced on a Sentosa^® SQ 316 chip (Ion Torrent PGM) using the dedicated Sentosa^® reagent kits.

2.5. Library preparation for Illumina Miseq platform

After PCR, 20 µl of the purified product were used for enzymatic fragmentation using the same Sentosa^® SQ HIV DRM manual protocol. Ligation was performed using 10 µl of TruSeq DNA UD Indexes v2 from the Illumina platform (Illumina Inc., San Diego, CA, USA) under the following conditions: 20°C for 20 min, 65°C for 5 min, and 85°C for 2 min. After ligation, each ligated PCR product was eluted with 18 µl elution buffer. For each sample, 5 µl were taken to form a pool of all the libraries. The concentration of the pooled library was set at 20 nM, and no dilution was performed. The pool was denatured with 0.05 N NaOH and diluted with HT1 to 30 pM. This diluted pool was incubated at 96°C for 4 min using a heat block, and 15% PhiX control was added. The library products were analysed by sequencing using the MiSeq v2 reagent kit (cycle 300) (Illumina Inc.) on the Illumina MiSeq platform in 2 × 151 cycles and a micro flow cell.

2.6. Sequence data analysis

The bioinformatic analyses were performed using the Advanced Sequencing Platform (ASP) v3.13.0 available from SmartGene (SmartGene, Lausanne, Switzerland). The SmartGene HIV-1 application is based on the proprietary IDNS (Integrated Database Network System) technology of SmartGene and is accessible via a secure web interface, which can handle base called sequencing files generated through different sequencing technologies. Related SmartGene patents are EP05700367 and EP07816282. *.bam or *.fastq files obtained through the Sentosa SQ platform (Vela) or through the MiSeq platform (Illumina) were uploaded to ASP. The application first performs the following steps in an automated manner:

(1) paired-end detection of read files, if uploaded in the same batch;

(2) technology-specific quality filtering to trim or remove low-quality reads; and

(3) establishment of a work list of the batch.

The user then selects the analysis pipeline to be used, in this case the HIV-1 PR+RT+INT targeted workflow, and selects the appropriate cut-off for ambiguous bases (background noise filtering, here 0.5%). The automated analysis pipeline performs the following actions:

(1)

quality filtering: R1 and R2 files are merged for Illumina data, then a sliding window (size of 25 nt) is applied to reads, enhancing the trimming of poor quality sections having a low Phred score (< 23 for the present study) and filtering short reads as well (< 20 nt in this study);

(2)

read mapping: reads passing the quality filters are mapped against HIV-1 profiles for PR, RT, and INT, generating a frame-aware nucleic acid alignment, using a proprietary alignment method of combined global and local alignments developed by SmartGene, followed by translation of the corrected alignments into amino acid sequences;

(3)

mapped reads are used for the creation of a frequency matrix;

(4)

mutations are determined and validated with respect to the reference (here HxB2, accession no. K03455) and their frequencies are calculated.

To generate the final report, the user selects the mutation cut-off (0.5–20%) above which mutations are assessed for resistance using one of the embedded HIV resistance algorithms, as well as a minimal read depth. In the present study, the algorithm developed by the French National Agency for AIDS Research (ANRS) was used [13], with a cut-off value of 20% and a minimum depth of 200. The drug resistance profile was then characterised, and the mutation frequencies were expressed as a percentage of the total number of HIV strains analysed. The final results were summarized in a list of all mutations per sample with their respective frequencies.

2.7. Statistical analysis

All statistical tests were carried out using Microsoft Excel version 16 (Microsoft Corp., Redmond, WA, USA). The cost per sample per run (eight samples) was calculated for the sequencing preparation workflow of Illumina Miseq and Ion Torrent PGM platforms. The prices of all kits used from PCR to sequencing were included, taking into account the total number of samples that can be processed per kit.

The average values of depths and the numbers of reads were compared using the two-tailed Student t-test. The significance level was set at P < 0.05.

Cohen's kappa coefficient was calculated to estimate the degree of agreement between the results of Ion torrent PGM platform and Miseq Illumina platform [14].

3. Results

A total of six QCs and the single positive control were correctly sequenced, and all mutations were found with both platforms. For the Ion torrent PGM platform, the total number of reads ranged from 185,069 to 752,343 (Figure 1). The average depth was 21,955, with an average mutation frequency of 99.3% (Figure 2). All HIV strains showed mutations that are associated with drug resistance. CQ022022, CQ032022, CQ012023, and CQ042023 strains had subtype B, while CQ022023 strain had subtype A, and CQ012022 strain had subtype F2 (Additional File S1).

For the Miseq Illumina platform, the total number of reads ranged from 642,162 to 2,074,028. The average depth was 46,856, with an average frequency of 99.5%. CQ022022, CQ032022, CQ012023, and CQ042023 strains were subtype B, while CQ022023 strain was subtype A1, and CQ012022 strain was subtype F2 (Additional File S2).

Concerning the number of mutations, those found in the INT gene were 100% concordant (Cohen’s kappa coefficient, 1; perfect agreement) for all samples. However, a few small, non-significant discrepancies (P > 0.05) were observed in the PR and RT genes (Table 1). These discordant nucleotides have not been reported to be associated with drug resistance.

The comparison of mean reads showed a statistically significant difference between Miseq (mean total 577,718) and PGM (mean total 445,178) (P = 0.0017). Although the average depth was higher with the Illumina Miseq platform than with the Ion torrent PGM platform, the difference did not reach statistical significance (P = 0.10).

Furthermore, the total duration of the preparation run was 26 h 43 min with the PGM and 26 h 28 min with Miseq. In all cases, the handling of the Illumina protocol was simpler and shorter (Figure 3). The price of a sequencing run on the Illumina Miseq platform was 1640 euros or about 1790 US dollars (exchange rate on 5 January 2024), and 1685 euros or about 1833 US dollars on the Ion Torrent PGM platform (Table 2).

4. Discussion

The NGS technology has become a powerful tool in the public health context of virus surveillance. The study of HIV drug resistance by sequencing is a key step in HIV/AIDS therapeutic management and strategy. Sequencing performed in the search for mutations in RT, PR, and INT genes is indicated in several clinical situations to optimise patient management [6,7,8,9]. Since the introduction of PGM Ion torrent sequencing technology at the Hospital Centre in Versailles in 2021, this technique has been used to study HIV resistance in over 677 patients. Ion Torrent sequencing technology is known for its simplicity, speed, and lower cost, due to its semiconductor chips [15].

The switch to a new sequencer, i.e. the Illumina Miseq platform, at our laboratory prompted us to make adjustments and adapt the Sentosa^® kit supplied by Vela to the Miseq sequencer on the Illumina platform. Vela's Sentosa kit is usually used for the PGM; this is the first time that it has been adapted for use with the Miseq Illumina. The advantage of using the Vela kit is that a single PCR can amplify all three HIV genes of interest, namely RT, PR, and INT [16], compared with other techniques which require PCR amplifications of individual genes [17].

Overall, the results showed a perfect concordance (Cohen's kappa coefficient ranged from 0.97 to 1: perfect agreement) of mutations between the two techniques, validating the Miseq sequencing technique on the Illumina platform. Indeed, the latter assay performed with minor modifications met our predefined validation criteria based on recommended methods for validating an in-house genotyping assay for HIV drug resistance monitoring (Figure 3). Consequently, the test was deemed suitable for our needs and capable of replacing the Ion Torrent PGM test in our laboratory.

Interestingly, the total number of reads for each sample and the average depth were higher with the Miseq Illumina platform than with the Ion Torrent platform. This difference can be explained by the fact that the chip in Ion Torrent platform used for sequencing generates around 5 million reads, whereas Illumina platform's micro flow cell generates around 8 million reads. In addition, the preparation of the library was simpler and faster with Miseq Illumina platform, allowing time flexibility for laboratory technicians and rapid delivery of results. Moreover, the price of reagents was 1790 USD with the Miseq Illumina platform and 1833 USD with the Ion Torrent PGM platform. This small difference in price is due to the necessity of using several Thermo Fisher Scientific kits for library enrichment on the Ion Chef automated system prior to PGM sequencing. This enrichment step is specific to and necessary for the Ion Torrent platform, but not for the Miseq Illumina platform.

5. Conclusions

For the first time, we have shown that it is possible to use the Vela library to sequence HIV genome on the Miseq sequencer with Illumina platform. This platform is simpler and faster and provides high quality results similar to those of the Ion Torrent PGM platform.