Gag P2/NC and pol genetic diversity, polymorphism, and drug resistance mutations in HIV-1 CRF02_AG- and non-CRF02_AG-infected patients in Yaoundé, Cameroon

In HIV-1 subtype-B, specific mutations in Gag cleavage sites (CS) are associated with treatment failure, with limited knowledge among non-B subtypes. We analyzed non-B HIV-1 gag and pol (protease/reverse-transcriptase) sequences from Cameroonians for drug resistance mutations (DRMs) in the gag P2/NC CS, and pol major DRMs. Phylogeny of the 141 sequences revealed a high genetic diversity (12 subtypes): 67.37% CRF02_AG versus 32.6% non-CRF02_AG. Overall, 7.3% transmitted and 34.3% acquired DRMs were found, including M184V, thymidine analogue mutations (T215F, D67N, K70R, K219Q), NNRTIs (L100I, Y181C, K103N, V108I, Y188L), and PIs (V82L). Twelve subjects [10 with HIV-1 CRF02_AG, 8 treatment-naïve and 4 on 3TC-AZT-NVP] showed 3 to 4 mutations in the Gag P2/NC CS: S373Q/T/A, A374T/S/G/N, T375S/A/N/G, I376V, G381S, and R380K. Subjects with or without Gag P2/NC CS mutations showed no significant difference in viral loads. Treatment-naïve subjects harboring NRTI-DRMs had significantly lower CD4 cells than those with NRTI-DRMs on ART (p = 0.042). Interestingly, two subjects had major DRMs to NRTIs, NNRTIs, and 4 mutations in the Gag P2/NC CS. In this prevailing CRF02_AG population with little exposure to PIs (~3%), mutations in the Gag P2/NC CS could increase the risk of treatment failure if there is increased use of PIs-based therapy.

surveillance component of HIV drug resistance (HIVDR), which includes in-country monitoring of early warning indicators of HIVDR 17 , assessing the threshold of transmitted or pretreatment DRMs and monitoring acquired HIVDR 16,18 .
As in other SSA countries, ART scale-up is effective in Cameroon, with an increasing national coverage (from 0% in 2003 to 22% in 2014) 19,20 . Therefore, it is critical to monitor HIV-infected Cameroonians for DRMs that could affect ART efficacy. Previous studies of HIV-infected subjects in Cameroon showed treatment failure among some patients on ART, with some of these patients having DRMs, while others did not show any major mutation known to be associated with treatment failure 21 . However, these previous studies of DRMs in Cameroon mainly focused on the viral reverse transcriptase (RT) and protease 21 . Of note, the protease cleaves the 55-kDa viral group specific antigen (Gag) precursor protein (p55) into six structural proteins: the matrix (p17), capsid (p24), spacer peptide-1 (p2), nucleocapsid (NC, p7), spacer peptide-2 (p1) and p6 22,23 . This enzyme also cleaves the 160-kDa GagPol polyprotein precursor into structural proteins and three enzymes: RT, protease, and integrase 22,23 . Protease cleavage occurs at specific cleavage sites on the Gag and GagPol polyproteins 24 , and it has been demonstrated that mutations in Gag cleavage sites can induce resistance to protease inhibitors (PIs) [25][26][27] and Nucleoside/Nucleotide Reverse Transcriptase Inhibitors (NRTIs) 28,29 independently of mutations in the protease, resulting in poor treatment outcomes 27,30 .
The recombinant HIV-1 CRF02_AG is the predominant viral strain circulating in West and Central Africa, including Cameroon (52-80%) [31][32][33][34] ; but there has been no study, to our knowledge, of Gag DRMs in settings with such HIV molecular epidemiology, and likewise, no study assessing the association between Gag mutations and DRMs in the polymerase, or viremia, and patients' immunological status in these settings. We therefore sought to ascertain the potential effects of Gag P2/NC cleavage site mutations and polymerase (protease and RT) major DRMs among HIV-infected Cameroonians on treatment outcomes, as well as the possible effects of such interactions on patients' viral loads and CD4 cell counts, including comparative analyses of CRF02_AG versus non-CRF02_AG.

Results
Demographic and clinical characteristics of study subjects. We analyzed plasma samples obtained between 2008 and 2015 from 283 HIV-infected subjects in Yaoundé, Cameroon; 157 samples were from individuals with undetectable viremia, and 126 samples were from subjects with detectable viremia. We successfully amplified and sequenced 113 (89.68%) of the 126 samples from subjects with detectable viremia, and 28 (17.8%) of the 157 samples from subjects with undetectable viremia. Of these 28 samples, we successfully amplified both pol and gag in 8 samples, but could only amplify pol in 4 samples, and gag in 16 samples. Of the total 141 samples amplified and sequenced, 109 (77%) were from ART-naïve subjects. Subjects' demographics and clinical characteristics are summarized in Table 1.

Phylogenetic analysis by HIV-1 genomic regions
Genetic diversity of HIV-1 gag. We successfully amplified gag sequences from 137 (97.16%) of the 141 samples. Analysis showed that CRF02_AG was the predominant subtype, with 93 subjects (67.15%) harboring HIV-1 CRF02_AG and 44 subjects (32.85%) harboring non-CRF02_AG subtypes [9 (6.56%) CRF11_cpx, 6 (4.37%) CRF22_01A1, 6 (4.37%) subtype G, 4 (2.91%) CRF18_cpx, 4 (2.91%) subtype F2, 4 (2.91%) subtype D, 3 Drug resistance mutations in ART-naïve subjects. Our analysis of pol sequences from 109 treatment-naïve subjects showed 8 subjects (7.3%) (including 6 infected with HIV-1 CRF02_AG) with transmitted DRMs. These transmitted DRMs included major resistance mutations to NRTIs such as M184V, T69D, T215F, K65R, and Y115F; and major resistance mutations to non-Nucleoside/Nucleotide Reverse Transcriptase Inhibitors (NNRTIs) such as Y181C, K103N, P225H, V108I, K101E, Y188L, E138Q, and L100I (Table 2). Four of the 8 treatment-naïve subjects were infected with viruses (HIV-1 CRF02_AG) that had major resistance mutations to both NRTIs and NNRTIs (Table 2). M184V was the most frequent NRTIs resistance mutation and was detected in 3 of the 8 patients (37.5%) ( Table 3). Other major NRTIs resistance mutations detected included the isolates from Cameroon (NACMR or NA2CMR IDs) were aligned using ClustalW, and phylogenetic analysis performed using the neighbor-joining method of MEGA.v.5 software as described in the Methods section. The reference sequences were from the Los Alamos database and included HIV-1 isolates from twelve countries (country letters code precedes reference accession number). Some references have been omitted to enable better visualization of the new Cameroon HIV sequences: CRF02_AG (blue) and non-CRF02_AG (green) subtypes. The Bootstrap value of 1000 replicates of at least 70% was used to determine the HIV-1 subtype. The scale bar represents 2% genetic distance. isolates from Cameroon (NACMR or NA2CMR IDs) were aligned using ClustalW, and phylogenetic analysis performed using the neighbor-joining method of MEGA.v.5 software as described in the Methods section. The reference sequences were from the Los Alamos database and included HIV-1 isolates from twelve countries (country letters code precedes reference accession number). Some references have been omitted to enable better visualization of the new Cameroon HIV sequences: CRF02_AG (pink) and non-CRF02_AG (green) subtypes. The Bootstrap value of 1000 replicates of at least 70% was used to determine the HIV-1 subtype. The scale bar represents 2% genetic distance. multidrug resistance mutation T69D, the thymidine analogue mutation (TAM) T215F, and the non-TAMs K65R and Y115F (Table 3). K103N was the most frequent NNRTIs resistance mutation detected and was present in 4 of the 8 naïve patients (50%) (

Treatment status Subtypes
Total n (%) ART (n) Naïve (n) 02_AG (n) Non-02_AG (n) NRTI M184V 9 3 6 5 11(58)  (Tables 2 and 3). Six of the 11 subjects with major DRMs harbored viruses with the thymidine analogue mutations (TAMs) D67N, K70R, K219E/Q, and T215F (Tables 2 and 3). Nine of these 11 subjects harbored viruses with major resistance mutations to both NRTIs and NNRTIs (Table 2). One subject (NA2CMR157) harbored viruses that had major resistance mutations to NRTIs, NNRTIs, and PIs ( Table 2). None of the 4 subjects on second line ART harbored viruses with major resistance mutation to PIs, but one of these subjects (NACMR039) harbored viruses with major resistance mutations to both NRTIs and NNRTIs, and a minor/secondary resistance mutation to PIs (K20I) ( Table 2). Four subjects on ART also had viruses harboring the NNRTIs secondary mutations V179D, A98G, E138A, V179E, and F227L, including subject NA2CMR151 who had both A98G and the major NNRTI resistance mutation Y181C. Most of the subjects with major resistance mutation to NRTIs and/or NNRTIs harbored viruses that also had secondary resistance mutations to PIs: L10I, V11I, K20V, and K20I (Table 2). Overall, M184V and K103N were the most prevalent NRTIs and NNRTIs transmitted or acquired DRMs, both for subjects infected with HIV-1 CRF02_AG and those infected with non-CRF02_AG subtypes ( Table 3).

Effects of viral subtypes and drug resistance mutations on patients' viral load and CD4+ cell count.
Analysis of viral and immunological parameters showed no differences in CD4+ cell counts of treatment-naïve and subjects on ART, either for subjects infected with CRF02_AG, or non-CRF02_AG viruses (Table 4). There was no difference in CD4+ cell counts of treatment-naïve subjects with NNRTIs resistance mutations, compared to subjects on ART that had NNRTIs resistance mutations (Table 4). However, CD4+ cell counts were significantly lower in treatment-naïve subjects with NRTIs resistance mutations (148.3 ± 79.5 cells/µl), compared to subjects on ART that had NRTIs resistance mutations (291.4 ± 108 cells/µl) ( Table 4, P = 0.042). Treatment-naïve subjects infected with HIV-1 CRF02_AG had higher viral loads (4.48 ± 1.59 log copies/ ml) than subjects on ART infected with HIV-1 CRF02_AG (3.31 ± 1.34 log copies/ml) ( Table 4, P = 0.0049). Treatment-naïve subjects infected with non-CRF02_AG viruses had higher viral loads (4.6 ± 1.2 log copies/ml) than non-CRF02_AG infected subjects on ART (2.8 ± 1.56 log copies/ml) ( Table 4, P = 0.0002). There was no significant difference in viral loads of treatment-naïve subjects with NRTIs DRMs, compared to infected subjects on ART that had NRTIs DRMs (Table 4, P = 0.256). However, treatment-naïve subjects with NNRTIs DRMs tended to have higher viral loads (4.57 ± 1.64 log copies/ml) than infected subjects on ART with NNRTIs DRMs (3.13 ± 1.5 log copies/ml) ( Table 4, (Fig. 4a). No mutation was found in the Q379 residue. Of the subjects analyzed, 16 (11.67%), 8 (5.83%), and 1 (0.72%) harbored viruses with T375A, S373Q, and S373P mutations respectively in the P2/NC cleavage site (Fig. 4a). Three subjects had mutations at both HIV-1 Pol and Gag P2/NC cleavage sites. Subject NA2CMR151 harbored viruses with both the Gag cleavage site mutations S373Q, T375G, I376V, G381S; the major NRTIs resistance mutations T69N, K70R, M184V, K219Q; the major NNRTIs resistance mutation Y181C, and the secondary NNRTIs resistance mutations A98G. Subject NA2CMR176 harbored viruses with both Gag cleavage site mutations A374T, T375G, I376V, G381S, the major NRTIs resistance mutations M184V, and the major NNRTIs resistance mutation Y188L. Subject NACMR050 harbored viruses with both Gag cleavage site mutation T375A and the secondary NNRTI resistance mutation V179E.  both treatment-naïve and subjects on ART, although both groups often showed differences in the percentage of subjects with each specific mutation (Fig. 4b). Naïve subjects showed some mutations in the P2/NC cleavage site that were absent in ART-experienced patients, including S373P, A374G/V/H/Q, T375P/H/I, I376A, M377L, and M378L/V. All treated subjects with Gag P2/NC cleavage site mutations were on first line ART regimen. The R380G mutation was present only in subjects on ART. Further analysis to identify subjects with more than two mutations in the Gag P2/NC cleavage site showed that Gag sequences from 12 of the 137 (8.75%) patients harbored viruses with 3 to 4 mutations in the P2/NC cleavage site (Table 5). Eight were treatment-naïve and 4 were on 3TC-AZT-NVP (Table 5). Eleven of these 12 subjects with ≥3 mutations in the Gag P2/NC cleavage site also had secondary PIs resistance mutations, including 10 subjects with K20I (Table 5).
Gag P2/NC cleavage site mutations in HIV-1 CRF02_AG and non-CRF02_AG subtypes. Ten of the 12 subjects with ≥3 mutations in the Gag P2/NC cleavage site were infected with HIV-1 CRF02_AG (Table 5). Subtype analysis showed higher percentage of mutations in the P2/NC cleavage site amino acid residues S373, A374, T375, I376, M378 and G381 in Gag sequences from subjects harboring HIV-1 CRF02_AG compared to subjects infected with Non-CRF02_AG viruses (Fig. 4c). HIV-1 CRF02_AG isolates showed lower percentage of mutations in the P2/NC cleavage site residues M378 and 380 and no mutations in the gag residues M377 and N382 (Fig. 4c). Mutations at residues 377 and 382 were observed only in non-CRF02_AG isolates and these non-CRF02_AG subtypes also showed a higher percentage of mutations at residue R380 (Fig. 4c). Both HIV-1 CRF02_AG and Non-CRF02_AG isolates showed no mutation at residue Q379.
Gag P2/NC cleavage site mutations, viral loads and CD4 cell counts. We performed additional analyses to determine whether mutations in the Gag P2/NC cleavage site had any effects on viral loads and CD4 cell counts. Of the 137 subjects with Gag sequences, 11 had 3 or more mutations in the Gag P2/NC cleavage site while 126 had less than 3 mutations in the P2/NC cleavage site. There was no significant difference in the viral loads of subjects with ≥3 mutations and subjects with less than 3 mutations in the P2/NC cleavage site ( Table 6). Analysis of all subjects with a mutation in the Gag P2/NC cleavage site showed higher viral loads in treatment-naïve subjects compared to subjects on ART (Table 6, P = 0.0005). There was no difference in CD4 cell counts of subjects with ≥3 mutations and those with less than 3 mutations in the P2/NC cleavage site; and no difference in CD4 cells counts of subjects with Gag P2/NC mutations that were treatment-naïve or were on ART ( Table 6).
to antiretroviral drugs and resistance to PIs 25,36 . Other studies of subjects infected with HIV-1 subtypes B, A, C, G, and D 30,37-39 , showed that mutations in Gag associated with resistance to PIs (LPV, and SQV/r) and treatment failure included mutations at amino acid residues 373 (including S373Q/P), 374, 375, and 378 at the P2/NC cleavage site. These mutations were observed before and after treatment failure, and resulted in the emergence of genetically distinct viruses at the time of treatment failure 38 . Viral genotype could also influence mutations and treatment efficacy. A study of patients infected with subtype-B and non-B HIV-1 on LPV/r monotherapy showed that at baseline, non-B subtypes were significantly more likely to harbor mutations, and the presence of more than 2 mutations in the P2/NC cleavage site at baseline predicted virologic failure 40 . This suggests that the 12 subjects identified in our current study with 3 to 4 mutations in the Gag P2/NC cleavage site could be more susceptible to treatment failure if given PIs. These subjects also had secondary mutations in the protease, including K20I that was previously shown to be associated with failure of PIs-based ART when simultaneously present with mutations in the Gag amino acids residues 373, 374, and/or 375 30,38 .
In our current study, mutations in the P2/NC cleavage site did not significantly affect viral loads or CD4 cell counts. This could result from mutations in Gag being more likely to affect PIs efficacy, whereas only 4 of the 141 subjects were on PIs, and none of the subjects with ≥3 mutations in the Gag P2/NC cleavage site had been on PIs. In fact, it has been shown that mutations in the Gag cleavage sites, as well as mutations outside the cleavage sites, change the structure of the Gag substrate, and this can reduce the PIs' affinity to the Gag-binding cleft and render PIs inefficient, resulting in restoration of the viral fitness and treatment failure 41,42 . In addition to affecting PIs' efficacy, mutations in the Gag cleavage sites can increase viral infectivity and resistance to NRTIs 29 .
In the current study, analysis of RT and protease sequences from 109 HIV-infected and treatment-naïve Cameroonians showed 8 (7.3%) with transmitted DRMs, including major resistance mutations to NRTIs (M184V and T215F) and NNRTIs (L100I, Y181C, K103N, V108I, and Y188L). Previous analysis of pol sequences from 116 43 and 216 44 treatment-naïve subjects in Cameroon showed a 13.9% 43 and 8.2% 44 rates of transmitted DRMs. Our data confirm these previous findings and show ongoing transmission of viral populations with DRMs in Cameroon.
Analysis of Pol sequences from the 32 patients on ART also showed 11 (34.3%) with major DRMs. In addition to the resistance mutations to NRTIs (M184V and the TAM T215F) and NNRTIs (L100I, Y181C, K103N, V108I, and Y188L) observed in naïve subjects, these subjects on ART also had the TAMs D67N, K70R, and K219Q; and 9 of those 11 subjects had major resistance mutations to both NRTIs and NNRTIs. These higher proportions of DRMs in subjects on ART confirm increased prevalence of acquired resistance mutations to NRTIs and NNRTIs in Cameroon. Increased prevalence of acquired DRMs and transmission of drug resistant mutants to other Cameroonians could result in increased risk of treatment failure and pose a major challenge to the local and global efforts against HIV/AIDS. M184V was the most common NRTIs resistant mutation in both treatment-naïve and subjects on ART. M184V mutation reduces the incorporation of nucleotide analogs into DNA, resulting in increased resistance to 3TC and emtricitabine [45][46][47] . The TAMs T215F, D67N, K70R, and K219Q induce resistance to the thymidine analogues AZT, stavudine, and other NRTIs by enhancing ATP-mediated excision and hydrolytic removal of the drug incorporated into viral DNA, thereby unblocking the viral DNA chain and enabling continuation of viral replication 48,49 .
The major NNRTIs resistance mutations identified in both treatment-naïve and subjects on ART included L100I, Y181C, K103N, V108I, and Y188L. These mutations are known to decrease the binding affinity of EFV and NVP to the viral target, resulting in resistance to these antiretroviral drugs and increased risk of virologic failure 47,50,51 . All subjects on ART in our study whose viral sequences showed major resistance mutations to NRTIs, NNRTIs, or more than 2 mutations in the Gag P2/NC cleavage site were on regimens including 3TC; and most of these subjects were on regimens including AZT, NVP, or EFV, suggesting a potential risk of future virologic failure in those patients. This risk would further increase in subjects harboring viruses with multiple DRMs. In fact, 5 of the 8 treatment-naïve subjects with transmitted DRMs and 9 of the 11 subjects with acquired DRMs had major resistance mutations to both NRTIs and NNRTIs. One subject had major resistance mutations to NRTIs, NNRTIs, and PIs, and two subjects had major resistance mutations to NRTIs, NNRTIs, and 4 mutations in the Gag P2/NC cleavage site. These multiple DRMs could increase the risk of treatment failure. In fact, analysis of pol sequences from 216 HIV-infected subjects starting ART showed that 80% of patients failing first line ART harbored viruses with at least 1 resistance mutation to two antiretroviral drug classes, and 36% of those failing second line ART harbored viruses with at least 1 resistance mutation to three antiretroviral drug classes 44 .
Despite this risk of drug resistance, ART is necessary and is beneficial for all infected subjects. Our data showed overall lower viral loads in subjects on ART, better immune recovery with significantly higher CD4 counts in subjects with major resistance mutations to NRTIs who were on ART, compared to treatment-naïve subjects with major resistance mutations to NRTIs. Our data also showed better viral control and borderline significant lower viral load in subjects with major resistance mutations to NNRTIs who were on ART, compared to treatment-naïve subjects with major resistance mutations to NNRTIs. One subject had major resistance mutations to NRTIs, NNRTIs and PIs; 9 subjects had major resistance mutations to both NRTIs and NNRTIs; two subjects had major resistance mutations to NRTIs, NNRTIs, and 4 mutations in the Gag P2/NC cleavage site. It is possible that such multiple DRMs could negatively impact treatment efficacy, viremia, and immune recovery, but studies with a larger sample size would be required to assess these effects. It is also likely that multiple mutations in the Gag cleavage sites, as shown in 12 subjects with 3 to 4 mutations in the Gag P2/NC cleavage sites, could affect the efficacy of PIs-based ART.

Conclusions and recommendations.
In summary, our data confirmed previous findings of HIV-1 CRF02_AG predominance in Cameroon (52-80%) [32][33][34] and showed the presence of transmitted (7.3%) and acquired (34.3%) resistance mutations to both NRTIs and NNRTIs. Our data also showed mutations in Gag P2/NC cleavages sites, with 12 subjects [8 treatment-naïve and 4 on 3TC-AZT-NVP] showing 3 to 4 mutations in the Gag P2/NC cleavage site. These results have clinical implications because the Gag P2/NC cleavage site mutations identified have been associated with resistance to PIs in subjects infected with HIV-1 subtype-B [25][26][27]35 . The presence of these Gag mutations in treatment-naïve and subjects not previously exposed to PIs could result in increased risk of virologic failure with use of PIs-based ART. Only 4 subjects in this study were on PIs-based ART, but with the potential increased use of second line regimen in Cameroon, subjects should be monitored for mutations in Gag cleavage sites that could affect the efficacy of PIs-based ART.

Study limitations.
We do not know whether the small numbers of non-CRF02_AG subtypes played a role in the lower frequency of mutations observed in non-CRF02_AG, compared to HIV-1 CRF02_AG sequences. Like in other countries in West and Central Africa, the molecular epidemiology of HIV in Cameroon is characterized by the predominance of HIV-1 CRF02_AG (over 67% of our samples), and despite the high viral genetic polymorphism, the frequency of the other 10 HIV-1 subtypes and URFs identified was low (1 to 7%), with some subtypes identified in only 1 or 2 subjects. Additional non-parametric regression analyses could have identified potential correlation between DRMs and individual subtypes, gender, age, and ART regimens. But with small sample sizes for non-CRF02_AG subtypes, we could not obtain sufficient statistical power for such analyses. For similar reasons, we were not able to analyze the correlation between Gag P2/NC cleavage site mutations and Pol DRMs. In fact, only 3 subjects with Gag P2/NC cleavage site mutations also had mutations in the Pol region. Future studies with larger sample size for subtypes that characterize the HIV molecular epidemiology in Cameroon would enable such correlation analyses.
The present study focused on Gag P2/NC, a primary cleavage site that play a critical role in virions production and maturation; there are four other cleavage sites on Gag and six cleavage sites on the GagPol polyproteins that are also involved in virions production, maturation, and fitness 26,[52][53][54] . Mutations in those sites could also result in drug resistance and virologic failure for subjects on PIs-based therapy 26,30,53,55,56 . Our subsequent studies will analyze these Gag and GagPol cleavage sites for the presence of DRMs and their impact on treatment outcomes. The presence of a major PI resistance mutation in only one subject in our study is likely due non-exposure to this drug class (only about 3% had been on PIs-based therapy). With the increasing use of second line regimens and PIs-based ART in SSA, studies of DRMs on the Gag and GagPol cleavage sites are important; such mutations would be of clinical and therapeutic consequences.

Materials and Methods
Study design, population, and ethical considerations. A cross-sectional analysis was conducted on plasma samples from 283 HIV-1 infected individuals in Yaoundé, Cameroon, between 2008 and 2015. These samples were collected as part of an ongoing project aimed at analyzing the influence of HIV genetic diversity on viral neuropathogenesis. This study was performed in accordance with guidelines of the Helsinki Declaration and was approved by the Cameroon National Ethics Committee, as well as the Institutional Review Board of the University of Nebraska Medical Center. Written informed consent was obtained from all participants and data were processed using unique identifiers to ensure confidentiality.
HIV serology, CD4 cell counts, and viral load. Sample collection and analyses were performed in the Hematology laboratory of the Yaoundé University Teaching Hospital, Cameroon. Venous blood samples were collected and stored at room temperature in the Hematology laboratory, and analyses performed within 6 hours of blood collection. The HIV status of each participant was determined using the Alere Determine HIV-1/2 antigen/antibodies Combo (Jouy-En-Josas, France), and the Murex HIV antigen/antibody Combination ELISA (Abbott Diagnostics, Chicago, IL, USA), according to the manufacturer's instructions. Each batch of reagents was quality controlled with known samples before used. A participant was considered HIV-positive if he/she tested positive for the two tests, HIV-negative if non-reactive for both tests and discordant if positive for only one test. No discordant result was observed in this study.
CD4 T-lymphocyte count was quantified by flow cytometry, using a Fluorescence Activated Cell Sorting (FACS) Count Instrumentation System, BD FACSCount, according to the manufacturer's instructions (BD Biosciences, San Jose, CA, USA). The FACS instrument was calibrated and quality control tested before each experiment. Plasma samples were stored at −70 °C or lower. For viral load quantification, HIV RNA copies number in each plasma sample was quantified by reverse transcription polymerase chain reaction (RT-PCR), using Amplicor HIV-1 Monitor Test (Roche Diagnostic Systems, Pleasanton, CA), according to the manufacturer's protocol. The Amplicor HIV-1 Monitor Test detection limit was 50 viral copies/ml. Five microliters of each RT-PCR reaction product was used in a second/nested PCR, in a 50 μL reaction volume containing 7.5 pmoles each of the following forward (5′-GACAGGCTAATTTTT TAGGG-3′; 2078-2094 bp  of HXB2) and reverse (5′-GGC TCT TGA TAA ATT TGA TAT GT-3′; 3580-3561 bp of HXB2) primers for the  pol gene, under the following conditions: 93 °C, 12 min; 40 cycles of 94 °C, 30 s; 53 °C, 30 s; 72 °C, 2 min; and  a final extension step at 72 °C, 10  2599-2582 bp of HXB2). Each of these primers was used at 12.8 pmoles in a 20 μL reaction mix including 130 ng of the purified pol PCR product (protease: amino acid 1-99; and reverse transcriptase: amino acid 1-280). Pol sequencing was performed using the Big-Dye chemistry method (Perkin-Elmer).
Capillary electrophoresis was performed using an Applied Biosystems 3730 DNA sequencer (Applied Biosystems, Tokyo, Japan) and sequences were loaded and assembled into Pregap4 v.1.5 software to generate contigs 57 . Nucleotide sequences were aligned with subtype/CRFs reference sequences from the Los Alamos National Laboratory (LANL) database using the CLUSTAL.W integrated into Bioedit.7.2.5 software 58,59 . Following comparison of each sequence to the subtypes and CRFs reference sequences 60 (database accessed on 8/17/2016), gaps were removed from the final alignments. The phylogenetic tree was constructed by the neighbor-joining and Kimura's two-parameter methods 61 using the MEGA.v.5 software 62 . The reliability of the branching orders was determined using 70% bootstrap robustness for subtype assignation 63,64 . Recombination among HIV-1 subtypes was assessed by SCUEAL 65 , COMET 66 , SimPlot 67 , Splitstree 68 , and Rega subtyping tool v.3 69 . The results obtained using each of the five genotyping tools were similar and concordant.
Determination of drug resistance mutations (DRMs). DRMs were analyzed in the protease and RT regions using the Stanford algorithm V.8.3 47 and the International AIDS Society 2015 mutation list was used to confirm each mutation 70 . The prevalence of transmitted DRMs was assessed using the Surveillance Drug resistance Mutations worksheet developed from the 2009 WHO list of mutations for surveillance of transmitted DRMs to NNRTI, NRTI, and PIs 71 . In accordance to these WHO guidelines 71 , the presence of one or more major resistance mutations to any drug class in treatment-naïve patients was considered as transmitted DRM.
Analysis of Gag P2/NC cleavage site mutations. Mutations were identified in the P2/NC Gag cleavage site of each sequence using HXB2 as reference sequence. Differences in frequency of amino acid sequences and the percentage of patients having each mutation, as compared to the wild type HXB2 virus, were determined following the alignment of all sequences using the CLUSTAL.W integrated into Bioedit.7.2.5 software 58 . Each Gag sequence was analyzed for the presence of specific P2/NC cleavage site mutations known to be associated with resistance to PIs and treatment failure. Samples containing a mixture of wild type and mutant sequences were scored as mutants.
Statistical analysis. Statistical analysis was performed using GraphPad Prism 5.0b. (GraphPad Software, La Jolla, CA, USA). Data were analyzed by two-tailed unpaired t-test for two groups comparison and P-value ≤ 0.05 was considered statistically significant. Ethical approval and informed consent. This study was conducted in accordance with the Declaration of Helsinki. The study protocol was approved by the Cameroon National Ethics Committee, as well as the Institutional Review Board of the University of Nebraska Medical Center. All subjects gave written informed consent for inclusion before participating in the study.