Phenotypic co-receptor tropism and Maraviroc sensitivity in HIV-1 subtype C from East Africa

Genotypic tropism testing (GTT) for co-receptor usage is a recommended tool for clinical practice before administration of the CCR5-antagonist maraviroc. For some isolates, phenotypic tropism testing (PTT) revealed discordant results with GTT. In this study, we performed a comparative study between GTT and PTT in HIV-1C from East Africa (HIV-1CEA) and compared the data with HIV-1B and 01_AE and described the maraviroc susceptibility in the CCR5-tropic strains. Patient-derived HIV-1 envgp120 region was cloned into a modified pNL4-3 plasmid expressing the luciferase gene. rPhenotyping dissected single clones from 31 HIV-1CEA infected patients and four strains with known phenotype. Additionally, 68 clones from 18 patients (HIV-1B: 5, 01_AE: 7, HIV-1CEA: 6) were used to determine the PTT in GHOST cell line. The respective V3-sequences were used for GTT. R5-tropic strains from HIV-1CEA (n = 20) and non-C (n = 12) were tested for maraviroc sensitivity in TZMbl cell line. The GTT falsely called a higher proportion of X4-tropic strains in HIV-1CET compared to PTT by both rPhenotyping and the GHOST-cell assay. When multiple clones were tested in a subset of patients’ samples, both dual-tropic and R5-tropic strains were identified for HIV-1C. Relatively higher EC50 values were observed in HIV-1C strains than the non-C strains (p = 0.002).

treatment modality in non-B settings, mainly low-and middle-income countries (LMICs). Therefore, optimal use and a high predictability of GTT on non-B subtypes is required to treat the patient with MVC.
Several recent studies from countries, where HIV-1 subtype HIV-1C is dominating, indicated an increase in predicted X4-tropic strains over time 8,9 . Most of these genotypic and phenotypic tropism correlation studies were performed on HIV-1C sequences from Southern Africa or India. Therefore, training sets and genotypic prediction for HIV-1C were solely based on sequences from those two regions. In contrast, a recent study showed a significant disagreement when these genotypic tools were applied to HIV-1C sequences from Ethiopia 9 , revealing that data on GTT and PTT for East African HIV-1C (HIV-1C EA ) strains are largely lacking. Given the higher heterogeneity among the East African strains 10 , which can significantly affect the sequence-based tropism prediction 1,4,11 , we hypothesized in this study that the current GTT tools for subtype C overestimate the X4-tropism in HIV-1C EA .
Therefore, the present study aims to phenotypically verify the co-receptor usage in HIV-1C EA and compare it using the current versions of several genotypic tools. Further, we sought to study Maraviroc susceptibility among the phenotypically determined R5-tropic stains.

Results
Clonal PTT (cPTT) using the virus produced using the individual clones in GHOST cell lines were performed with 68 individual clones which were infectious from 180 clones tested, obtained from 18 patients samples infected with HIV-1B (n = 5), HIV-1C (n = 6) and 01_AE (n = 7) subtypes. The clonal GTT by sequencing individual clones (cGTT) falsely identified a higher proportion of X4-tropism in HIV-1C compared to phenotypic tropism testing by cPTT (Fig. 1A). Presence of both dual tropic and R5-tropic strains were observed in HIV-1C when multiple clones from the same subset of patients' samples were tested (Fig. 1). In some patients' samples (Pt#2, Pt#7) there are no changes in the V3 sequences, but one clone showed R5 tropic and other as dual-tropic. Virus generated from the plasmids pMJ4 (R5-tropic) and pNL43 (X4-tropic) showed R5-tropic and X4-tropic respectively.
The replicative phenotyping (rPhenotyping, population PTT, cPTT) by pooling several clones was performed with 31 patients' samples infected with HIV-1C and four HIV-1C QC-samples using a co-culture of transfected 293 T cells with the SXR5 reporter cell. cPTT was also performed with six QC viruses with known tropism (97ZA009, 97ZA003, 93IN101, 94KE105, DEMC08NG001, and DEMC09ZA001.S1). All QC viruses identified showed 100% concordance with cGTT and population GTT (pGTT) performed by bulk population sequncing by Sanger Sequencing method. In contrast, discordances were observed between different tools for cGTT/pGTT and cPTT or rPhenotyping. In concordance with cGTT, pGTT also identified more X4-tropic viruses than rPhenotyping (Fig. 2).
Next, a maraviroc susceptibility assay was performed on phenotypically determined R5-tropic viruses that identified a wide range of susceptibility in HIV-1 C viruses. When comparing with HIV-1B or 01_AE, the HIV-1C EA strains showed reduced susceptibility (p = 0.002) for maraviroc (Fig. 3).

Discussion
In this study using several bioinformatics tools for GTT and two PTT methods, we were able to demonstrate a higher false assignment to X4-tropism by GTT for HIV-1C viruses from East Africa. This indicates the need to train the machine-learning GTT tools with these strains. The maraviroc susceptibility assay further identified a wide range of drug susceptibilities in HIV-1C East African strains ex vivo and a statistically significantly reduced susceptibility compared to non-C viruses.
In our recent study, we showed that, despite strong monophyletic clustering, HIV-1C strains from East Africa possess a significantly higher diversity compared to HIV-1C strains from other geographical locations such as India or South Africa 10 . Our study also indicated an increase in predicted X4-tropic strains over the decade 10 . A more recent genotypic study further confirmed the findings in a large number of samples 9 . A study from India also indicated a marginal temporal increase in X4-tropism prevalence 8 . However, there is a high chance for false calls of X4-tropism in HIV-1C depending upon the machine-learning GTT tools use in the case of marginal increase 12 . A recent study comparing GTT and PTT, demonstrated the over-estimation of the X4-tropism in 01_AE and 02_AG isolates 11 . Also a study by Kalu et al. from Ethiopia showed the disagreements between the GTTs 9 . Earlier phenotypic studies had reported hardly any (6%) 13 or no 14 X4-tropic strains for Ethiopia. Our study now confirmed the over-estimation of the predicted X4-tropism by GTT in HIV-1C from East Africa which is in line with our original hypothesis. Our study also indicated that, despite the absence of any changes in the V3-loops, some strains are dual tropic. This is further supported by a recent study, in which the inclusion of the complete gp120 sequence improved the genotypic prediction for HIV-1A and C viruses 15 . Overall, these studies emphasize the need for improving the present algorithms for GTT, e.g. by including larger segments of gp120 in the tropism prediction.
In our study, a number of the HIV-1C strains from East Africa possessed a reduced drug sensitivity to the CCR5 antagonist maraviroc compared to isolates of the non-C HIV-1 subtypes B and 01_AE. Other studies have indicated that mutations in the V3 loop of gp120 could mainly be responsible for this variability in drug sensitivity to maraviroc [16][17][18] . However, this difference appears to be strain specific 16,19,20 . Of note, the various sensitivity to maraviroc in this study could not in all cases assigned to V3 changes, hinting the involvement of mutations in gp120 and gp41 regions other than the V3 loop sequence itself 21,22 .
In conclusion, our study demonstrates that there is a high need to re-train the machine-learning GTT tropism tools for HIV-1 non-B isolates, necessitating the inclusion of a larger number of relevant non-B strains for the accurate tropism prediction. As current GTT highly overestimates the X4-tropism e.g. in HIV-1C strains from East Africa, where a higher heterogeneity in the V3-loop is common, it may falsely limit the use of drugs such as maraviroc that are becoming available in these resource-limited areas of Africa and other places. Moreover, as the introduction of CCR5 antagonists such as maraviroc is becoming a therapeutic option there, additional studies are needed that virologically monitor the clinical response to maraviroc, particularly of HIV-C viruses.

Materials and Methods
Cell lines, viruses, and plasmids. TZM  Clinical specimens, Ethical Considerations and data availability. Stored patients' plasma samples included in this study were from two different cohorts: (1) Swedish InfCare Cohort (n = 35) 23 and (2) Ethiopian cohort (n = 25) 10 . All samples had been subjected to near full-length sequencing as described by us recently 24 . Only pure subtypes were selected for the study. Ethical permissions were obtained from the respective sites. Swedish samples:   the manuscript. All methods were performed in accordance with approved institutional guidelines. The patient identity was anonymised and delinked prior to analysis.
Phenotypic tropism testing. Phenotypic tropism testing was performed using two methods: (1) Tropism testing in GHOST cell-lines 27 of the viruses generated from individual clones (cPTT) and (2) replicative phenotypic tropism test rPhenotyping (pPTT) whereas described previously 25 . In cPTT co-receptor tropism was determined by measuring Renilla luciferase activity (relative light units [RLU]) using Bright-Glo ™ Luciferase Assay System (Promega, US). We consider a 10-fold shift in mean RLU of infected cells over non-infected. The pNL4-3 and pMJ4 were used as positive control for X4-and R5-tropic strains respectively. In addition to luciferase expression, in a subset of clones' green fluorescent protein (GFP) expression was also captured using confocal microscopy (Olympus Fluoview v2.0b). The rPhenotyping was performed with 31 patients' samples infected with HIV-1C and four QC-samples. The ligated mixture was transformed into TOP10 bacteria (Life Technologies) and inoculated directly in the LB broth supplemented with ampicillin to retain the viral diversity.