Introduction
The HIV/AIDS pandemic has been a global crisis for four decades [
1,
2]. At the end of 2020, there were 37.6 million people living with HIV globally, 1.5 million new HIV-1 infections, and 690,000 AIDS-related deaths [
3]. South Africa has 7.9 million people living with HIV-1 (PLWH), making it the country with the highest number of infections in the world [
4,
5]. This highlights the need to better understand HIV-1 natural immune responses in order to bolster the efforts of developing effective HIV-1 vaccines.
During early HIV-1 infection, effective cytotoxic T-lymphocyte (CTL) immune responses play an important role in the control of viraemia, by contributing to suppression of the viral load (VL) to a set-point [
6]. CTL responses target and kill virus-infected cells, via recognition of viral peptide epitopes and releasing cytokines and cytotoxic granules that facilitate cell killing [
7‐
9]. Viral peptide epitopes are presented on the surface of infected cells by the human leukocyte antigen type I (HLA I) trans-membrane proteins encoded by
HLA-A,
B and
C alleles [
10‐
12]. The targeted epitopes undergo immune selection pressure, which leads to the emergence of viral escape mutations [
13,
14]. This results in evasion of the immune system, loss of immune control and ultimately progression to AIDS [
7,
10]. The majority of epitopes that play a role in the control of viraemia have been extensively studied and reported for HIV-1 Gag [
15,
16]. However, some studies show that HIV-1 Pol also harbours important CTL epitopes that contribute to the control of HIV-1 viraemia [
17‐
19], hence
pol is included in some HIV-1 vaccine candidates [
20‐
25]. There are limited data on the characterisation of HIV-1 Pol CTL epitopes in early and chronic HIV-1 disease stages. The aim of this study was to characterise and assess the evolution of CTL epitopes encoded by HIV-1
pol during early and chronic stages of HIV-1 infection, using deep sequencing methods.
Materials and methods
Study population
This was a retrospective study that used stored plasma samples that were collected from individuals in the Tshwane district of South Africa. Participants enrolled in this study had a confirmed diagnosis of early or chronic HIV-1 infection and were antiretroviral therapy (ART) naïve. They were identified in a study that screened for early HIV-1 infection in individuals who had a negative rapid test at point-of-care facilities, and diagnosis was confirmed through HIV-1 VL, antibody (enzyme-linked immunoassay), p24 antigen, Western blot and avidity testing [
26]. Most of the participants were followed up, and thus samples at two time-points were available (at baseline and follow-up) for analysis.
Nucleic acid extraction and amplification of HIV-1 pol
Total nucleic acids were manually extracted from plasma samples using the QIAamp UltraSens Virus kit (Qiagen, Hilden, Germany). Samples with a VL > 1000 copies/ml were extracted from a plasma input volume of 500 µl, which was adjusted to 1 ml using phosphate-buffered saline (PBS). Samples with a VL ≤ 1000 copies/ml were extracted directly from 1 ml plasma input volume to increase nucleic acid yield. Extraction was performed according to the manufacturer’s instructions except for the first centrifugation step, which was optimized to 800 relative centrifugal force (RCF) instead of the recommended 1200 RCF. This modification facilitated a more efficient dissolution of the pellet. Nucleic acids were eluted in a volume of 60 µl and all the eluates were stored at − 70 °C immediately after extraction.
The complete HIV-1
pol gene was amplified in all samples using an in-house nested PCR method, employing the SuperScript™ III One-Step RT-PCR System with Platinum™
Taq High Fidelity DNA Polymerase (Invitrogen, Carlsbad, California, USA). Each PCR reaction was performed in a 50 µl reaction, which included 5 µl of extracted RNA / DNA template, 2X reaction mix containing 2.4 mM magnesium sulphate (MgSO
4), 0.4 mM of each deoxynucleotide triphosphate (dNTPs), 10 mM sense primer, 10 mM anti-sense primer, 5 units (U) of enzyme mix for (first round) or 1U of Platinum Taq polymerase (for second round), and nuclease-free water. The first-round amplicon was used as template for the second-round amplification (
www.lifetechnologies.com) (Additional file
1: Table S1). One or two extracted HIV-1-negative controls were included in each experiment to assess potential contamination.
The following sets of primers were used for amplification of the complete pol gene: SM-F1 (outer forward) 5′-GCG GCT ACA TTA GAA GAA ATG ATG-3′ (HXB2 1807–1830) and SM-R1 (outer reverse) 5′-GCC AAG TAT TGT AGA GAT CCT ACC T-3′ (HXB2 5462–5488), SM-F3 (inner forward) 5′-AGA TTG TTA AAT GTT TCA ACT GTG G-3′ (HXB2 1952–1976) and SM-R3 (inner reverse) 5′-CTC CTG TAT GCA AAC CCC AAT A-3′ (HXB2 5245–5266). The nested primers amplified a fragment size of 3268 bp.
Sanger sequencing and sequence analysis
Sanger sequencing was performed on PCR amplicons, but only for baseline samples. Sequencing was performed commercially in six overlapping regions covering the entire
pol ORF (Inqaba Biotechnical Industries, Tshwane, South Africa) (Additional file
1: Table S1). Sequences were edited in CLC Main Workbench 21 software (Qiagen, Hilden, Germany) and consensus sequences were generated, viewed in BioEdit 7.2.5 (
https://bioedit.software.informer.com/download/) and aligned using the online version of the MAFFT program (
https://mafft.cbrc.jp/alignment/server/). The HIV-1 HXB2 reference sequence (K03455.1) was used for nucleotide numbering. Phylogenetic analysis was performed on the multiple alignment and the ratio of nonsynonymous to synonymous (dN-dS) mutations within the HIV-1 Pol region was computed using MEGA software (
https://www.megasoftware.net/). A dN-dS ratio
\(\ge\) 5 was used to identify codons under high selection pressure and these sites were subsequently mapped to CTL epitopes documented in the Los Alamos HIV database [
27].
Illumina sequencing and analysis of consensus sequences
Sequencing of PCR amplicons was performed at Inqaba Biotechnical Industries (Pretoria, South Africa), a commercial next generation sequencing (NGS) service provider. Briefly paired-end libraries (2 × 300 bp) were prepared using the NEBNext® Ultra™ II DNA library prep kit for Illumina® (New England Biolabs, USA) and sequencing was performed on an Illumina MiSeq instrument (Illumina, USA).
Some samples (n = 12) that were viewed to have poor quality scores were sent for repeat sequencing to the National Institute for Communicable Diseases (NICD), Sequencing Core Facility, Johannesburg, South Africa. Multiplexed paired-end libraries (2 × 300 bp) were prepared using the Nextera™ XT DNA Library Sample Preparation kit (Illumina Inc., California, United States) according to the manufacturer’s instructions and sequencing was carried out on an Illumina MiSeq instrument (Illumina Inc., California, United States).
Illumina sequencing data was analysed in the PASeq program (
https://paseq.org/) using the feature that excludes APOBEC-induced hypermutations, and consensus sequences were generated. The quality of the Illumina reads was assessed in PASeq and CLC Genomics Workbench 21 (Qiagen, Hilden, Germany). PASeq-generated Illumina consensus sequences were viewed in BioEdit 7.2.5 software and aligned to HIV-1 reference sequences obtained from the GenBank database, in the MAFFT program available online. Sanger and Illumina consensus sequences obtained from baseline samples were compared through phylogenetic analysis in MEGA software and were observed to correctly cluster together (Additional file
4: Figure S1).
Analysis of Illumina sequencing reads for minority variants
Deep sequencing reads were mapped to an HIV-1 subtype C reference (AY162225.1) in CLC Genomics. Minority variants were analysed using a low-frequency variant detection tool in CLC Genomics, setting the cut-off for significance at 1% and the minimum frequency for variants at 5%. Minority variants were assessed in all participants and only nonsynonymous variants were mapped to HIV-1 CTL epitopes in the Los Alamos HIV database, to identify the epitope they fall within. Thus, in this study, minority variant epitopes were defined as epitopes existing at proportions above 5% but less than 20%. The Los Alamos HIV database was also used to predict human leukocyte antigen (HLA) allotypes that may be involved in presenting the identified epitopes to CTLs. Evolution within Pol CTL epitopes was assessed by comparing the proportion of minority variants between baseline and follow-up samples, and in early and chronic HIV-1 disease stages.
Data analysis
Descriptive statistics was used to present median values and interquartile range (IQR) for age, HIV-1 VL and sequencing depth. Fisher’s exact test was used to assess if there was an association between the distribution of mutations and HIV-1 disease stage for epitopes identified through Sanger sequencing. Two sample t-test was used for comparing the proportions of minority variant epitopes among the different regions of the Pol protein. A p-value of ≤ 0.05 was considered statistically significant. All statistical tests were performed on the STATA 16.0 software package (StataCorp LP, College Station, TX, USA). The statistical analyses of dN–dS ratios were computed on the MEGA programme using the HyPhy test for selection. Due to the small sample size, codons with dN–dS ratios of ≥ 5 were considered for assessment of epitopes under immune selection pressure. GraphPad Prism v.9.1.2 (GraphPad Software, San Diego, California, USA) was used to prepare graphs for the frequently recognised minority variant epitopes among the different regions of the Pol protein, and also for showing the dynamics of these epitopes in sample pairs obtained during early and chronic HIV-1 infection stages.
Discussion
To our knowledge, this is the first study to assess the evolution of HIV-1 Pol CTL epitopes in samples obtained during early and chronic disease stages in a setting where HIV-1 subtype C is predominant. We identified epitopes within all three regions of Pol, unlike previous studies that have only characterised epitopes within RT and PR [
16,
29,
33]. This highlights the advantage of using a PCR method that amplifies the complete
pol gene. Baseline and follow-up sequences for each participant correctly clustered together, indicating accurate amplification and sequencing of viral genes belonging to the same participant. It was expected that the majority of sequences would cluster with HIV-1 subtype C as this is the most common subtype in the southern African region [
34,
35]. A small number of non-subtype C strains was detected, which could have been introduced through migration, and this has also been reported in previous South African studies [
36,
37].
Many epitopes were identified through the analysis of minority variants obtained by deep sequencing compared to majority variants obtained by Sanger sequencing. This shows the advantage of using deep sequencing for characterisation of Pol CTL epitopes as this method can detect variants that exist below 20% of the total virus population, which are not readily detected by Sanger sequencing [
38,
39]. Some identified minority variant epitopes have never been reported for HIV-1 subtype C and were possibly missed by previous studies that employed Sanger sequencing. The use of Sanger sequencing in most studies has most likely underestimated the presence of CTL epitopes in Pol. Other studies have also shown increased detection of CTL epitopes when using NGS methods [
40,
41]. This shows an advantage in the use of NGS for characterising HIV-1 CTL epitopes in future research studies.
In this study, CTL epitopes were identified in all regions of Pol, but mostly in RT and IN. Some of these epitopes such as GKKAIGTVL (PR 68–76) and IAMESIVIW (RT 375–383) (Table
3) have been previously reported to induce CTL responses [
42,
43]. These data indicate that all three regions of Pol are immunogenic and may play a role in the control of viraemia and the establishment of a viral set-point during early HIV-1 infection, in agreement with previous studies that assessed CTL responses against the HIV-1 proteome [
42‐
44]. In a study that assessed CTL responses during acute HIV-1 infection, Kim et al.showed that the larger proportion of targeted epitopes were within Pol as compared to other regions of the HIV-1 proteome, and some of these epitopes stimulated strong CTL responses [
42]. Ojwach et al.reported that CTL responses towards epitopes in RT and IN during acute and chronic HIV-1 infection induced mutations that decreased viral replicative fitness, indicating that RT and IN harbour crucial CTL epitopes [
45]. Fewer epitopes were identified in PR especially during early HIV-1 infection and this could indicate that this protein harbours more subdominant CTL epitopes that are mostly recognised later during the course of infection [
46]. Thus, PR might have a limited role in control of viraemia during early HIV-1 infection. CTL epitopes encoded by
pol have also been reported to play a role in the control of viraemia in long-term non-progressors (LTNPs) and elite controllers (ECs) [
47,
48].
HLA class I allotypes play an important role in the control of HIV-1 viraemia [
29]. Several HLA class I alleles that encode allotypes predicted to present epitopes identified in this study are known to be present in the South African population, including HLA-B*58:01 and HLA-A*02 that present SAAVKAACW (IN 123–131) and MASDFNLPPIV (IN 22–31), respectively [
28,
30,
49]. Past gene association studies showed that alleles within the
HLA-B group correlate with greater viral control than alleles in the
HLA-A and
HLA-C groups [
28,
29,
50]. Payne et al.found that
HLA-B*58:01 was associated with slower HIV-1 disease progression and was protective against HIV-1 infection [
51], which could explain why some participants such as 8047 and 6743 recognised more epitopes, as they also recognised an epitope reported to be presented by HLA-B*58:01. A hierarchy in epitope presentation may be an explanation for recognition of more epitopes during chronic HIV-1 infection. Immunodominant epitopes could be preferentially presented earlier during the course of infection, followed by presentation of subdominant epitopes later [
52]. Immunodominant epitopes are often located in variable domains whereas subdominant epitopes usually map to more conserved regions of the HIV-1 proteome [
18,
53]. The HIV-1
pol gene is more conserved than
gag and
env [
54] and therefore harbours more subdominant epitopes [
18]. Examples of subdominant responses being more effective in controlling viraemia have been shown in past studies [
55,
56].
The evolution of Pol CTL epitopes was monitored by assessing minority variants in sample pairs. Minority variant epitopes that increased in proportion from baseline to follow-up may have replaced the wild-type epitopes over time and facilitated immune escape [
57]. Fitness cost is the likely explanation for minority variant epitopes that decreased in proportion over time, and these may have been out-competed by new variants [
57]. The proportion of some minority variant epitopes remained constant between the two time-points, and these could also represent variants that do not replicate efficiently and are hence maintained at lower levels within the HIV-1 quasispecies [
40,
57]. Some variant sites were located outside of reported epitopes (Additional file
3: Table S3), and these may represent new unreported epitopes or could highlight important adjacent sites that may affect epitope processing and presentation [
13,
58]. Viral evolution was also assessed by comparing sequences from early and chronic HIV-1 stages. Data from this study showed that evolution within CTL epitopes began quite early following infection, as epitopes with escape mutations were detected in participants who had early HIV-1 infection [
59]. Many epitopes with escape mutations were detected in participants in the chronic HIV-1 stage, indicating that viral evolution of HIV-1 results from immune selection pressure that occurs throughout the course of infection and diverse viral populations evolve that constantly evade immune responses [
57,
60].
All three expressed proteins (PR, RT and IN) of Pol harbour CTL epitopes and this highlights the importance of including
pol in the design of a vaccine candidate [
2,
20]. Some researchers have included RT and IN in the design of HIV-1 vaccines [
22,
23,
53]; this is supported by the findings of our study as the majority of epitopes mapped to these two proteins. Conserved epitopes have been shown to provide cross-clade protection against HIV-1 infection [
2,
18]. Some of the conserved Pol epitopes such as IETVEPVKL (RT 5–12) and SVPLDEGFRK (RT 117–126) (Table
3), that were previously observed in a study that looked at using a vaccine candidate with conserved immunogens were also identified in our study [
18]. The immune responses directed towards epitopes in conserved regions are usually subdominant but may provide protection against multiple HIV-1 subtypes [
17,
53]. Many studies that included Pol epitopes in vaccine candidates showed stimulation of effective CTL responses [
18,
56,
61,
62]. Ahmed et al., showed control of HIV-1 replication in vitro through stimulation of Pol CTL responses by a vaccine that contained subdominant epitopes [
56].
Limitations of this study include the small sample size and fewer participants with early HIV-1 infection. The duration of infection in participants with chronic HIV-1 infection was not known. HIV-1 VL was not measured in the participants at follow-up, therefore no correlation between VL and the number of minority variant epitopes detected could be made. Functional CTL responses against identified epitopes were not determined and genomic DNA samples were unavailable to conduct HLA class I genotyping. Longitudinal samples were only obtained at two time-points for most of the participants at a median interval of 4 weeks and this might not be sufficient for assessing viral evolution.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.