Combinational antiretroviral therapy (cART) has significantly slowed the AIDS pandemic and reduced the incidence of HIV infections (UNAIDS, 2020). However, treatment intensification exerts selective pressure and drives viral adaptation resulting in the emergence of HIV-1 drug resistance mutations (DRMs). The selection of drug-resistant viruses is based on HIV’s high mutation rate, followed by the generation of a genetically diverse pool of HIV quasispecies in each patient, and the selection and outgrowth of the fittest variants, adapted to the given condition (Feder et al., 2021). Under antiretroviral treatment, there is a strong purifying selection for drug-resistant viruses, which can be traced by viral genomic analyses of the drug target’s respective genomic regions, primarily found in the pol region (Feder et al., 2021). Drug resistance mutations represent a major barrier to effective therapy (Hamers et al., 2018). People infected with HIV-1 who have acquired DRMs are less likely to achieve viral suppression, are more likely to discontinue treatment, acquire new DRMs, and experience virological failure or death (WHO, 2019). Also, naturally occurring polymorphisms (NOPs) can modulate the magnitude of drug resistance, paving the way to developing DRMs, or have intrinsic resistance potential themselves (Wainberg and Brenner, 2012). NOPs are most abundant but least studied in regions like West and Central Africa where highly diverse non-B clade viruses circulate (Wainberg and Brenner, 2012; Hemelaar et al., 2020). Non-clade B infections make up most HIV-1 infections worldwide, with genetic variation among clades up to 35% (Hemelaar et al., 2020). Naturally occurring polymorphisms, their diversity and plasticity, cART-associated viral adaptations, and the growing proportion of DRMs are areas of intense clinical interest and still unfolding. Comprehensive, population-wide studies are required to leverage a deeper understanding of these mutations and NOPs, including effective prevention and treatment strategies. We provided such an analysis for Cameroon, a West-Central African country with a population of ∼27 million and ∼3.7% HIV prevalence in 2021, and we considered all functional, non-clonal reverse transcriptase (RT), protease (PR), and integrase (IN) sequences that have been deposited to the LANL database (8,130 out of ∼10,600 submitted sequences in total) (Figure 1).

cART programs in sub-Saharan Africa are based on WHO guidelines that have traditionally recommended reverse transcriptase and protease inhibitors (RTIs, PIs) for first- and second-line therapy, and only recently have begun transitioning to integrase strand transfer inhibitor (InSTI) usage in some countries (UNAIDS, 2017b). Effective access to cART in Cameroon started in 2004 following the WHO/UNAIDS “3 by 5” initiative to provide three million people living with HIV (PLHIV) in low and middle-income countries with life-prolonging cART by the end of 2005 (WHO, 2003). Before 2004, only a few generic antiretroviral drugs [lamivudine (3TC), zidovudine (ZDV), stavudine (d4T), and nevirapine (NVP)] were available at a very low scale in main cities such as the capital Yaoundé (Bourgeois et al., 2005; Aghokeng et al., 2011). Until May 2007, treated patients had to pay for cART (US $23–$100 monthly), laboratory tests (US $58–$85 per viral load assay and $19–$27 per CD4 cell count), and physician’s consultation ($1.5–$15), thus limiting the number of people accessing cART (Laurent et al., 2006; Boyer et al., 2009). cART usage in Cameroon expanded in June 2007 with free access to cART for eligible patients dependent on CD4 count-based guidelines, and since 2016 to all HIV-infected individuals following the implementation of the “test and treat” UNAIDS’s initiative (WHO, 2005; National AIDS Control Committee, 2015). The scale-up of cART considerably improved the lives of PLHIV globally, including sub-Saharan Africa, but West-Central Africa remains far behind in all three categories of the UNAIDS 90-90-90 targets, i.e., diagnosis, treatment, and viral suppression (UNAIDS, 2017a). Furthermore, the implementation of cART has accelerated the development of DRMs. The overall prevalence of drug-resistant strains in cART-naïve and treated individuals has increased dramatically, from 2.2 and 40.7% in 2010, reaching > 10% and 60% in recent years, respectively (Nanfack et al., 2015, 2017; WHO, 2019). Cameroon and other West-Central African countries have mainly used a triple cocktail of two nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs) and one non-NRTIs (NNRTIs) as first-line treatment supported by PIs in the second line (WHO, 2015). Although targeting the same protein (RT), NRTIs and NNRTIs exert differential drug pressure, which results in complementary DRM profiles (Supplementary Table 1). For example, RT mutations M184V/I and T215F/Y are linked with NRTIs (e.g., lamivudine, emtricitabine, zidovudine, or stavudine) and K103N/S with NNRTIs (e.g., nevirapine or efavirenz). The intensive use of InSTIs (Boyer et al., 2009; Landman et al., 2014), particularly dolutegravir (DTG), effectively started in 2020, whereas in preceding years, it was distributed in 10 treatment centers only out of more than 160 functional treatment units. InSTIs potently suppress viral load in diverse HIV-1 clade infections, yet subtype-specific differences in efficacy and acquisition of DRMs exist (Wainberg and Brenner, 2012), as has become known for other antiretroviral drug classes (Theys et al., 2019; Stanford University HIV Drug Resistance Database, 2020).

While computational methods have been developed for RTIs, PIs, and more recently, for InSTIs, to quantify the genetic barrier to the acquisition of DRMs (Gotte, 2012; Theys et al., 2019), population-based time-series analyses are needed to show the actual adaptive changes to the viral genomic landscape in countries under treatment pressure. Here, we present such a study for Cameroon, a country undergoing progressive cART scale-up since 2004 and known as the epicenter of the HIV disease where an immense diversity of HIV strains exists (Hahn et al., 2000; Hemelaar et al., 2020). Among a high number of subtypes, circulating, and unique recombinant forms (CRFs, URFs), the globally most abundant recombinant form, CRF02_AG, is predominant (Nanfack et al., 2017; Hemelaar et al., 2020). To understand the emergence and impact of NOPs and DRMs, we use computational and structural methods to compare HIV-1 pol genomic plasticity in RT, PR, and IN regions and dissect the cross-sectional changes in HIV-1 strains in Cameroon over time.

This study provides a thorough overview of HIV pol mutations and genotypic drug resistance in an entire population. In an ecological analysis of pooled viral genomic data from Cameroon where HIV prevalence stands at ∼3.7% in 2021, we used computational and structural methods to assess the genomic plasticity of HIV-1 pol over time and its implication on treatment. Treatment exerts selective pressure on the swarm of viruses present in every patient and the entire population and drives viral adaptation resulting in the emergence of HIV-1 DRMs. The cross-sectional nature of our study on a representative set of 8,130 sequences from Cameroon (all LANL-deposited RT, PR, and IN sequences that are functional and non-clonal) enabled us to compare sufficient numbers of mutations and NOPs statistically over time. Fluctuating sample numbers and the inability to follow and assess individuals longitudinally were limitations. Our study revealed high plasticity in HIV-1 pol on the population level, which appears to be profoundly shaped by regionally applied cART protocols. Effective cART in Cameroon started in 2004. Hereafter, substantial scale-ups occurred in 2007 when cART became free of charge for eligible patients (CD4 count < 200 cells/mm³) and with the “test and treat” initiative implemented in 2016. Cameroon’s HIV treatment guidelines have primarily relied on regimens with two NRTIs and one NNRTI, which resulted in an increase in pre-treatment NRTI/NNRTI DRM rates up to >10%, and in patients failing first or second-line cART up to >60% (Nanfack et al., 2015, 2017; WHO, 2019), jeopardizing the success of national HIV management. Nevirapine was included in most first-line regimens in Cameroon between 2004 and 2016 and was replaced in 2016 by efavirenz in the preferential first-line regimen in resource-constrained settings (tenofovir disoproxil fumarate/lamivudine/efavirenz) (National AIDS Control Committee, 2015; WHO, 2015). The long-term use of NRTIs and NNRTIs explains the high proportion of RTI DRMs observed in our study. Starting from low levels of RT/PR-associated DRMs in the cART-naïve population (≤2004), a significant increase occurred at multiple protein residues post-implementation of cART. RT mutations M184X, K103X, and T215X rose most significantly, both in CRF02_AG (Figure 4) and non-CRF02_AG (Figure 7 and Supplementary Table 1), which is in line with recent DRM monitoring studies in Cameroon and worldwide (Nanfack et al., 2015, 2017; WHO, 2019). These mutations strongly affect the efficacy of NRTIs lamivudine/emtricitabine (M184V/I) and zidovudine/stavudine (T215F/Y), and NNRTIs nevirapine/efavirenz (K103N/S). Notably, these mutations were significantly linked among the study sequences (Figure 6), implying that combinational NRTI/NNRTI treatment exerts simultaneous selection pressure and induces mutations in different RT regions at comparable rates depending on the applied drugs. The functional consequence of the co-existence of drug resistance mutations remains to be studied in further detail. Based on the recent introduction of InSTIs in Cameroon, the absence of DRMs to InSTIs was expected.

Antiretroviral drugs, including PIs, NRTIs/NNRTIs, and InSTIs, have mainly been tested for efficacy against subtype B viruses. Although most of these drugs are expected to act on diverse subtypes, genetic differences can impact drug resistance pathways or kinetics of DRM development (Theys et al., 2019). Although only marginally assessed, reduced susceptibility to antiretroviral drugs has been described in CRF02_AG infections (Wainberg and Brenner, 2012). Reduced PI susceptibility at baseline has been associated with subsequent virological failure on lopinavir/ritonavir monotherapy in antiretroviral-naïve patients harboring CRF02_AG viruses (Sutherland et al., 2015). In the DAYANA trial, a simplified regimen of tenofovir plus lopinavir/ritonavir used for early treatment attained poor viral suppression (defined as HIV-1 RNA viral load ≥ 100 copies/ml between weeks 24 and 96) in a CRF02_AG study population (Landman et al., 2014). In IN, the G118R mutation was associated with NOPs at codons 74 and 118 in clades C and CRF02_AG and conferred resistance against raltegravir and DTG (Malet et al., 2011; Brenner et al., 2016). In addition to canonical DRMs, we found emergent CRF02_AG polymorphisms accumulating during cART scale-up. NOPs have been reported to alter or impair susceptibility to RTIs, PIs, and even InSTIs (Wainberg and Brenner, 2012). Comparable to the observed DRM imprints upon cART in Cameroon, the NOP landscape changed most strikingly in RT, followed by PR and IN. It implies that resistance issues might be involved besides compensatory mutations and evolutionary/epidemiological trends. Protein destabilization is a potential drug resistance mechanism (Barouch-Bentov and Sauer, 2011; Chang and Torbett, 2011; Sheik Amamuddy et al., 2018); however, as our DRM data show, there is no strict association with resistance, and only a subset of DRMs exert destabilizing effects. Consequently, emerging treatment-associated mutations showed a differential pattern. Notably, a set of >10 emerging RT treatment-associated mutations had strong destabilizing potential. For example, RT T362X mutations significantly increased post-cART, and both T362N/S were shown to be strongly destabilizing (Figures 5, 6). T362 is located at the DNA-interacting RT connection domain, and mutations have been associated with NNRTI treatment, possibly affecting RNAse H activity (Julias et al., 2003). In the IN region, we observed a high baseline occurrence of low-resistance mutation L64M, known as a specific feature of CRF02_AG (Rhee et al., 2003). L64M further increased post-cART, though not significantly.

The scarcity of available HIV-1 IN sequences from Cameroon after the roll-out of DTG-based regimens in Africa (2017) limits our sequence-based conclusions on InSTI treatment in Cameroon. Between 2017 and November 2020, only three complete and two fragmented IN sequences have been deposited to LANL. The latter two are from the same HIV-1 CRF18_cpx-infected individual who developed multi-drug InSTI resistance in 2017 based on G140A, Q148R, and E157Q DRMs, which further aggravated in 2019 additionally involving E138K/Q and S147G (Fokam et al., 2020). Our structural simulations suggest that CRF02_AG IN has a comparable quaternary structure with clade B, including surface and drug-binding site charge distribution (Supplementary Figures 13, 14). Due to DTG’s high genetic barrier to resistance, the emergence of DRMs in treatment-naïve patients is extremely rare, which renders DTG highly effective across clades (Rhee et al., 2019). Increased usage of InSTIs including the long-acting cabotegravir for maintenance therapy and pre-exposure prophylaxis is anticipated (Swindells et al., 2020). Emerging IN drug resistance is an imminent threat requiring broad and widespread monitoring efforts across Cameroon and beyond (Inzaule et al., 2018; Lubke et al., 2019). Particular attention should be paid to patients that previously used Raltegravir-containing regimens as in the case of DTG and Darunavir/r multi-drug resistance in HIV-1 CRF18_cpx infection described recently in Cameroon (Fokam et al., 2020). Furthermore, recent data suggest a reduced InSTI efficacy in patients with DRMs in RT (Siedner et al., 2020), stressing the importance of studying full pol. Even beyond pol, there have been reports of InSTI resistance caused by five mutations in the nef region (Malet et al., 2017). In addition, mutations in gag have been shown to contribute to PI resistance, and findings suggest a tight interdependency between Gag structural proteins and the protease during the development of PI resistance (Clavel and Mammano, 2010; Codoñer et al., 2017; Datir et al., 2020). Most recently, mutations in the envelope glycoprotein (Env) have been associated with resistance to antiretrovirals, including the InSTI DTG (Van Duyne et al., 2019; Hikichi et al., 2021). Consequently, future studies should aim at analyzing DRMs and NOPs over the full genome and in the context of clinical drug resistance data to detect drug resistance across drug classes more comprehensively and inform clinical management more accurately. The low number of full-genome CRF02_AG sequences from Cameroon (61 deposited in the LANL database) and the lack of associated clinical data impeded such analyses for Cameroon at this time. In summary, the better characterization of DRMs and NOPs including changes over time in the context of diverse HIV-1 clades and applied cART regimens is paramount to understanding the underlying mechanisms of emerging and evolving antiretroviral drug resistance. The association between genomic imprints and usage of antiretroviral drugs targeting the respective pol regions underline the ability of ecological analyses to reveal viral adaptation processes and underlying purifying selection pressures, to eventually guide clinical management and targeted therapies. Our results support the applicability of InSTIs for cART in Cameroon but stress the necessity of tight surveillance of DRMs when increasing drug pressure is exerted. The current study focused on CRF02_AG and the comparison of pre- and post-cART periods in Cameroon. Future studies will need to intensify longitudinal characterizations involving different clades and regions of the world under consideration of the applied treatment protocols to dissect HIV-1 pol and full-genome adaptations and their clinical implications on a global scale. Phenotypic testing of emerging treatment-associated mutations in different clades with or without co-occurring DRMs will reveal their direct and indirect impact on drug resistance.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

JNT and RD conceived research goals and analyses. JSB, PY, SD, X-PK, and RD performed formal analyses. RD acquired funding for the project. PY, SD, X-PK, and RD developed methodologies, designed and/or implemented computer codes. AJN, JNT, and RD supervised the research activity and validated the research output. JSB and RD verified the underlying data and prepared figures and tables. JSB, AJN, and RD wrote the manuscript. All authors reviewed and edited the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The funders had no role in the design or conduct of the study.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.