New Subtype B Containing HIV-1 Circulating Recombinant of sub-Saharan Africa Origin in Nigerian Men who have Sex with Men

Abstract

Background

HIV-1 circulating recombinant forms (CRF) containing subtype B are uncommon in sub-Saharan Africa. Prevalent infections observed during enrollment of a prospective study of men who have sex with men (MSM) from Lagos, Nigeria revealed the presence of a family of subtype B and CRF02_AG recombinants. This report describes the HIV-1 genetic diversity within a high-risk, high-prevalence, and previously undersampled cohort of Nigerian MSM.

Methods

Between 2013 and 2016, 672 MSM were enrolled at the Lagos site of the TRUST/RV368 study. Prevalent HIV-1 infections were initially characterized via pol sequencing and phylogenetic subtyping analysis. Samples demonstrating the presence of subtype B were further characterized by near full-length sequencing, phylogenetic, and Bayesian analyses.

Results

Within this cohort, HIV-1 prevalence was 59%. The major subtype was CRF02_AG (57%), followed by CRF02/B recombinants (15%), subtype G (13%), and smaller amounts of A1, B, and other recombinants. Nine clusters of closely related pol sequences indicate ongoing transmission events within this cohort. Among the CRF02_AG/B, a new CRF was identified and termed CRF95_02/B. Shared risk factors and Bayesian phylogenetic inference of the new CRF95_02B and the similarly structured CRF56_cpx, indicate a Nigerian or West African origin of CRF56_cpx prior to its observation in France.

Conclusion

With high HIV-1 prevalence, new strains, and multiple transmission networks, this cohort of Nigerian MSM represents a previously hidden reservoir of HIV-1 strains, including the newly-identified CRF95_02B and closely-related CRF56_cpx. These strains will need to be considered during vaccine selection and development in order to optimize the design of a globally effective HIV-1 vaccine.

INTRODUCTION

As of 2015, Nigeria has the largest population in Sub-Saharan Africa and second highest HIV-1 burden in the world.^1,2 National surveys have observed substantial differences in HIV-1 prevalence among all reproductive aged adults (3.4%)³ and key affected populations (KAPs), including female sex workers (8.6–19.4% depending on brothel-based or mobile business type), their clients, men who have sex with men (MSM; 22.9%), and people who inject drugs (3.4%).⁴ As a risk group, MSM in Nigeria face general social stigmatization and potential criminal prosecution.⁵ These negative motivators have led MSM to be less willing to seek HIV testing and engage in care, which has resulted in general undersampling^6,7 of the MSM population and a lack of MSM-specific molecular epidemiology data. Over the past decade, some pioneering studies^8–10 have begun to quantify the extent of HIV-1 prevalence among Nigerian MSM. However, studies intended to molecularly characterize HIV-1 strains among MSM cohorts have only recently begun. As part of efforts to address the monitoring and treatment needs of MSM in Nigeria, the TRUST/RV368 study was initiated to monitor, characterize, and provide HIV-1 treatment for cohorts of MSM and transgender women (TGW) in the cities of Abuja and Lagos.¹¹

Molecular characterization of an epidemic establishes the genetic background of HIV-1 subtypes or circulating recombinant forms (CRFs) that are typically found within that region. The genotypes of circulating HIV-1 strains guide vaccine development, testing kit reagent selection, and phylogenetic tracking of the epidemic’s progress. Most HIV-1 infections belong to group M, which contains 9 subtypes (A, B, C, D, F, G, H, J, and K) and greater than 96 CRFs. Of those, subtypes A, C, D, G and CRF02_AG represent the majority (89%) of the publicly available sequences from sub-Saharan Africa (hiv.lanl.gov), with the remaining fraction composed of other recombinants and smaller amounts of subtypes B, F, H, J, and K. Subtype C dominates in South Africa, the surrounding regions, and Ethiopia; subtypes A, C, and D compete for dominance in Kenya, Tanzania, and Uganda of east Africa; and CRF02_AG and subtype G comprise the major strains of Nigeria and its west African neighbors. In regions where multiple subtypes are found circulating in significant proportions, superinfections can lead to the development of unique recombinant forms (URFs) that can be transmitted and result in the establishment and growth of new CRFs. The HIV-1 epidemics of east and west Africa are particularly diverse and complex due to the presence of multiple co-circulating subtypes and a resulting plethora of URFs and CRFs.^12,13

Subtypes A, B, C, D, G and recombinants CRF01_AE and CRF02_AG represent over 90% of prevalent global HIV-1 infections.¹⁴ As a region, sub-Saharan Africa is home to approximately 70% of people living with HIV-1¹⁵ and although it contains a genetically diverse collection of HIV-1 strains, it does not have significant amounts of subtype B or CRF01_AE (0.29% and 0.24%, respectively, of sub-Saharan sequences as of April, 2018: hiv.lanl.gov). The global MSM networks are known to be a risk factor for HIV-1 transmission. CRF56_cpx, was discovered in year 2010 among French MSM with close links to West Africa indicated the existing network of MSM between the continents. We hypothesized that a molecular epidemiology study of HIV-1 strains in MSM enrolled in a study conducted in Lagos, Nigeria would provide new insight information on circulating and recently introduced strains to the area. Unexpectedly, this study provides a population level example of a hidden reservoir of infectious CRF02_AG/B recombinant strains within a well-known, yet understudied group of high-risk individuals. Among these recombinant strains, 4 of them shared similar recombinant structure and have been designated as the newly discovered CRF95_02B. Additionally, phylodynamic analysis is presented that these previously unknown subtype B and CRF02_AG recombinants share a common origin with CRF56_cpx.

METHODS

Study participants

Between 2013 and 2016, the Lagos site of the TRUST/RV368 prospective observational cohort study enrolled 672 people assigned the male sex at birth and followed them at 3 month intervals for 18 months. Males aged 18 years and older who reported receptive or insertive anal intercourse with a male partner within the twelve months prior to enrollment were recruited using respondent driven sampling as previously described.¹¹ Upon enrollment, 614 out of 672 participants were tested for HIV and were administered a questionnaire to assess risk behaviors. HIV testing was performed on finger-stick blood samples using the Government of Nigeria HIV diagnostic algorithm, which employs parallel sero-reactive rapid device tests: Determine® (Alere, Waltham, MA, USA) and Uni-gold® (Trinity Biotech, Co-Wicklow, Ireland) with Stat-Pak HIV (Chembio, Medford, NY, USA) serving as a tie-breaker.

Sequencing

Plasma samples from 150 out of 362 HIV-1 positive participants with a viral load ≥ 1000 copies/mL underwent viral RNA extraction, cDNA generation, and near-endpoint dilution prior to PCR amplification, purification, and sequencing with an ABI 3730XL capillary sequencer as previously described.¹⁷ Pol sequences were obtained using nested primers selected such that resulting PCR amplification products covered HXB2 referenced positions 2273–3869. Samples containing subtype B in pol were subjected to near-full genome sequencing via PCR amplification of complete (HXB2: 789–8713) or half-length (HXB2: 789–5852 and 4560–8713) genome products.

Phylogenetic analysis

Viral sequence subtypes were determined using a combination of the HIV-1 Genotyping Tool at NCBI,¹⁸ a modified local installation of the jumping profile Hidden Markov Model (jpHMM) tool¹⁹ which incorporated additional reference sequences from West Africa (see Reference List 1, Supplemental Digital Content), BLAST analysis (HIV BLAST, hiv.lanl.gov), and phylogenetic tree analysis. Recombinant genome structures were determined by first approximating the genome breakpoints using bootscanning with Simplot²⁰ v3.5.1 and the previously described jpHMM tool, followed by refinement of the breakpoints by visual inspection (SeqPublish, hiv.lanl.gov) using six reference sequences each for subtype B and CRF02_AG (see Reference List 2, Supplemental Digital Content). Confirmation of the parent subtype identity was provided via BLAST and bootstrap analysis of maximum likelihood trees constructed from the resulting sub-genome fragments. Recombinant genome structures were visualized using RecDraw.²¹ Times to most recent common ancestor (tMRCA) were calculated using the Bayesian Markov chain Monte Carlo algorithm within BEAST²² v1.8.4 with the GTR substitution model, strict clock, gamma site (5) heterogeneity model, exponential growth tree prior, and year of collection (study samples) or year of publication (reference sequences) as listed by the Los Alamos National Laboratory HIV sequence database (hiv.lanl.gov). Model parameters for tMRCA calculations were selected following comparisons of results which probed substitution models (GTR and HKY), site heterogeneity models (gamma and gamma+invariant), codon partitions (none and three), clock types (strict and relaxed with lognormal and gamma distributions), and tree priors (constant size and exponential growth). Chain lengths were selected such that the effective sample size of the posterior was greater than 200. The mean tMRCA and 95% highest posterior density values for each clade were reported by Tracer v1.6.0.²³ Multiple sequence alignments were generated with HIVAlign²⁴ using the Markov model²⁵ option. Maximum likelihood phylogenetic trees were calculated with DIVEIN²⁶ using the GTR model (with estimated invariant sites and gamma distribution), 100 replicates, and masked hypervariable regions in the variable loops of env. Transmission network calculations were performed with the TN93²⁷ algorithm (github.com/veg/tn93) using the average ambiguous base option, a genetic distance cut-off of 1.5%, and the pol sequences described above after masking of Drug-Resistance Surveillance sites.²⁸ The HIV-1 sequences described here are available under GenBank accession numbers: MH654824 - MH654973 for pol, and MH666153 - MH666162 for full-length genomes.

RESULTS

Study participant demographics, social environment, and prevalence of HIV-1, chlamydia, and gonorrhea for the combined locations of Abuja and Lagos have been previously published.^5,11,29,30 During enrollment, 614 participants were screened for HIV-1 and 362 prevalent infections were observed within the Lagos cohort yielding an HIV-1 prevalence rate of 59%. Of these, 205 had HIV-1 RNA ≥1000 copies/mL within 120 days of diagnosis and 150 underwent analysis of pol sequences. The participants who underwent viral sequencing had median age 23 (interquartile range [IQR] 20–27) years, median CD4 437.5 (IQR 344–547) cells/mm³, and median HIV-1 RNA 4.51 (IQR 4.15–4.92) log₁₀copies/ml. Among them, self-reported gender identity included 76.7% men, 14.7% women, and 8.7% other/unknown. Fifty-seven percent reported their sexual orientation as gay/homosexual while the others were bisexual. Most of the participants (70%) had senior secondary level and 26% had higher education. Their occupations ranged from not working (23%), student (25%), professional/self-employed (15%), entertainment (9%), labor (5%), and other/unknown (23%). The phylogenetic analysis results revealed an HIV-1 subtype distribution of 57% CRF02_AG, 15% CRF02_AG/B recombinants, 13% subtype G, 3% subtype A1, 2% subtype B, and smaller amounts of recombinants containing CRF02_AG, A1, or G.

Application of a genetic distance cut-off of 1.5% in pol identified nine transmission network clusters containing two to five participants each and composed of pure CRF02_AG (five clusters) or subtype G (four clusters) strains. A maximum likelihood tree showing the branching patterns of subtype and transmission network clusters within the pol sequences is shown in Figure 1.

Figure 1.

Phylogenetic relationships and transmission network clusters among participant pol sequences. Maximum likelihood (ML) tree constructed using subtype reference strains and pol sequences (HXB2: 2273–3869) from prevalent infections observed at study enrollment. Colored sample numbers denote CRF02_AG (pink) or subtype G (green) sequences belonging to proposed transmission networks (defined as having a pairwise genetic distance ≤1.5%). The cluster circled in blue represents the CRF02_AG/B recombinants in which the CRF95_02B strains were observed. ML bootstrap values at relevant nodes are shown. The scale bar indicates a genetic distance of 10%.

A total of 23 pol sequences were identified as recombinants between CRF02_AG and subtype B. The participants who harbored these recombinants did not have different demographic characteristics and geographic location compared to others infected with non-CRF02_AG/B recombinants (see Table 1, Supplemental Digital Content). From the cluster of 15 CRF02_AG/B recombinants circled in Figure 1, eight samples forming mini-clusters with similar recombination patterns in pol were selected for near full-length sequencing. Recombinant breakpoint analysis of the full-length genomes revealed parental subtype B contributions to the gag and pol genes of most of the selected samples amidst a background of CRF02_AG and a small amount of subtype G (in one sample). Five of the full-length genomes were identified as representatives of a new CRF, termed CRF95_02B, as shown in Figure 2a. This new CRF has a background of CRF02_AG with contributions from subtype B to the C-terminus of gag p17, thumb of RT,³¹ αA-αD motifs of RNase,³² and the catalytic core of integrase.³³ The remaining strains share various genomic breakpoints in common with the CRF95_02B strains, but have diverged through additional recombination or evolutionary events, Figure 2b.

Figure 2.

Recombinant genome structures. Genome structures of the eight strains (seven full-length and one half-length) observed while characterizing CRF95_02B. (a) The HXB2 referenced positions of the CRF95 subtype B fragments are shown relative to genomic markers in gag and pol (dash-dot lines). (b) The similarly structured genomes of the related non-CRF95 strains. The colors of the fragments correspond to their predicted parent subtypes: CRF02_AG (pink), subtype B (blue), subtype G (green).

Recombinant breakpoint analysis identified eight breakpoints within the HIV-1 genomes representing the new CRF95_02B. The breakpoints define nine fragments, which were aligned with subtype reference strains to build maximum-likelihood trees for subtype confirmation through bootstrap assessment (see Supplemental Figures 1–3, Supplemental Digital Content). Bootstrap values ≥ 85% support the assigned parent subtype for seven of nine fragments. Supportive bootstrap values for the two remaining fragments could not be obtained owing to the length of Fragment II (104 nucleotides) and the divergence of Fragment V from CRF02_AG and subtype A, consequently the parental subtypes for those two fragments were confirmed as subtype B and CRF02_AG, respectively, using BLAST similarity searching.

BLAST similarity searching of full-length and individual fragments of the CRF95_02B strains (as described above) indicated the presence of a potentially related recombinant: CRF56_cpx³⁴. Multiple sequence alignments from the gag-pol, int, and gp120-gp41 regions containing only a single subtype in common to both CRF95_02B and CRF56_cpx strains were used to construct maximum likelihood trees, which supported the genetic similarity implied by the BLAST results and genomic structures (see Supplemental Figure 4, Supplemental Digital Content). The results of tMRCA calculations using the same genomic regions described above in an alignment containing 108 CRF02_AG and 58 subtype B reference sequences (see Reference List 3, Supplemental Digital Content) displayed large differences in the predicted tMRCA for the CRF95_02B and CRF56_cpx clades. The mean tMRCA for the gag-pol, int, and gp120-gp41 regions ranged from 26.5 to 31.9 years for CRF95_02B and 7.3 to 8.8 years for CRF56_cpx. As a single clade, the mean tMRCA for those same regions ranged from 27.0 to 34.9 years, which largely overlaps the predicted tMRCA for the CRF95_02B clade. The mean and 95% highest posterior density interval for each genomic region per clade is shown in Supplemental Figure 5b, Supplemental Digital Content.

DISCUSSION

The TRUST/RV368 study was implemented to address the HIV-1 treatment needs of MSM in Nigeria and obtain the molecular information necessary to guide HIV-1 vaccine development efforts. The results presented in this report describe the molecular genetics of HIV-1 within a cohort of Nigerian MSM at the time of study enrollment. The large proportions of CRF02_AG (57%) and subtype G (13%) observed within this cohort are consistent with the predominant strains observed among all Nigerian people living with HIV¹² as well as other specific populations in Nigeria.³⁵ However, the presence of a significant proportion of CRF02_AG/B recombinants (15%) is a new finding within a West African cohort. Detailed examination of these strains revealed the presence of a new CRF termed CRF95_02B. Analysis of the CRF95_02B strains suggested a relationship with the CRF56_cpx¹⁶ strains observed in France in 2010. Calculations of tMRCA for shared regions across the genome show large differences between the mean tMRCA of CRF95_02B (26.5–31.9 years) and CRF56_cpx (7.3–8.8 years). The tMRCA of the CRF56_cpx strains likely represents sampling of closely related transmissions of CRF56_cpx (as evidenced by the similarity between the CRF56_cpx strains shown in Supplemental Figure 4, Supplemental Digital Content). However, the tMRCA of a combined clade of CFR95_02B and CRF56_cpx strains is nearly identical to the tMRCA for the CRF95_02B strains alone. When viewed in the context of shared risk factors,¹⁶ similar recombinant breakpoints, and similar phylogenetic clustering, the combined tMRCA calculations are supportive of a common origin for both CRF95_02B and CRF56_cpx. If correct, then the current structure of the CRF56_cpx genome is the result of further recombination events since that time. In the past, there has been strong correlation of MSM interaction between HIV-1 infected individuals of West Africa origin and the French MSM populations as supported by the existence of Unique Recombinant Forms (URFs) B/CRF02_AG, URF5_B/02/G and CRF56_cpx identified among MSM in France. The CRF02_AG/B recombinants identified in this cohort might be recombinant strains generated previously among MSM who traveled between West Africa and France.

The newly described CRF95_02B and its cluster of related strains was readily detected in this cohort. The diversity evident in the phylogenetic cluster patterns and tMRCA results shown here, indicate it has been in circulation within this cohort for decades. CRF95_02B is only newly identified due to under-surveillance of this group. Although only prevalent strains are presented here, the presence of transmission networks in this cohort indicates a possible exposure to this CRF in the community (manuscript in preparation). With high HIV-1 prevalence, new strains, and multiple transmission networks, this cohort of Nigerian MSM represents a previously hidden reservoir of HIV-1 strains that will need to be considered during vaccine immunogen selection and development. This study demonstrates the need to continue epidemiological monitoring of all groups engaged in high-risk activities, identifying their risks, and finding solutions for prevention. The cross-reactivity of both subtype C and mosaic HIV-1 vaccines currently in efficacy studies in southern Africa^36,37 to this new CRF, and related variants, will be critical to optimize the design of globally effective HIV-1 vaccines in key affected populations.

Supplementary Material

Supplemental Digital Content

Supplemental Figure 1

Supplemental Figure 2

Supplemental Figure 3

Supplemental Figure 4

Supplemental Figure 5

Reference List 1

Reference List 2

Reference List 3

Table 1.docx