Genotypic Prediction of Co-receptor Tropism of HIV-1 Subtypes A and C

Consequences of HIV infection in the bone marrow niche

Dysregulation of the bone marrow niche resulting from the direct and indirect effects of HIV infection contributes to haematological abnormalities observed in HIV patients. The bone marrow niche is a complex, multicellular environment which functions primarily in the maintenance of haematopoietic stem/progenitor cells (HSPCs). These adult stem cells are responsible for replacing blood and immune cells over the course of a lifetime. Cells of the bone marrow niche support HSPCs and help to orchestrate the quiescence, self-renewal and differentiation of HSPCs through chemical and molecular signals and cell-cell interactions. This narrative review discusses the HIV-associated dysregulation of the bone marrow niche, as well as the susceptibility of HSPCs to infection by HIV.

AI-based multi-PRS models outperform classical single-PRS models

Polygenic risk scores (PRS) calculate the risk for a specific disease based on the weighted sum of associated alleles from different genetic loci in the germline estimated by regression models. Recent advances in genetics made it possible to create polygenic predictors of complex human traits, including risks for many important complex diseases, such as cancer, diabetes, or cardiovascular diseases, typically influenced by many genetic variants, each of which has a negligible effect on overall risk. In the current study, we analyzed whether adding additional PRS from other diseases to the prediction models and replacing the regressions with machine learning models can improve overall predictive performance. Results showed that multi-PRS models outperform single-PRS models significantly on different diseases. Moreover, replacing regression models with machine learning models, i.e., deep learning, can also improve overall accuracy.

Prediction of Coreceptor Tropism in HIV-1 Subtype C in Botswana

It remains unknown whether the C-C motif chemokine receptor type 5 (CCR5) coreceptor is still the predominant coreceptor used by Human Immunodeficiency Virus-1 (HIV-1) in Botswana, where the HIV-1 subtype C predominates. We sought to determine HIV-1C tropism in Botswana using genotypic tools, taking into account the effect of antiretroviral treatment (ART) and virologic suppression. HIV-1 gp120 V3 loop sequences from 5602 participants were analyzed for viral tropism using three coreceptor use predicting algorithms/tools: Geno2pheno, HIV-1C Web Position-Specific Score Matrices (WebPSSM) and the 11/25 charge rule. We then compared the demographic and clinical characteristics of people living with HIV (PLWH) harboring R5- versus X4-tropic viruses using χ2 and Wilcoxon rank sum tests for categorical and continuous data analysis, respectively. The three tools congruently predicted 64% of viruses as either R5-tropic or X4-tropic. Geno2pheno and the 11/25 charge rule had the highest concordance at 89%. We observed a significant difference in ART status between participants harboring X4- versus R5-tropic viruses. X4-tropic viruses were more frequent among PLWH receiving ART (χ2 test, p = 0.03). CCR5 is the predominant coreceptor used by HIV-1C strains circulating in Botswana, underlining the strong potential for CCR5 inhibitor use, even in PLWH with drug resistance. We suggest that the tools for coreceptor prediction should be used in combination.

Determination of HIV Tropism in Patients with Antiretroviral Therapy Failure in Arkhangelsk Region

The aim of the study was to determine the tropism of the human immunodeficiency virus in patients with virological failure of antiretroviral therapy (ART) from the Arkhangelsk Region based on the analysis of the env gene V3 loop nucleotide sequence.

Materials and methods. We used blood plasma samples obtained from 76 HIV-infected persons from the Arkhangelsk Region with virological failure of antiretroviral therapy. The nucleotide sequences of the HIV env gene C2-V3-C3 region were studied by PCR followed by sequencing. The genotype of the studied strains was determined based on the analysis of their phylogenetic relations with reference sequences from the international GenBank database, as well as using specialized programs. To predict viral tropism, the Garrido rule and the online bioinformatic tool Geno2Pheno[coreceptor] were used. The Geno2Pheno[coreceptor] algorithm, determines the false positive rate (FPR) based on the analysis of the env gene V3 loop nucleotide sequence. Results and discussion. Significantly lower representation of R5X4/X4-tropic HIV variants in long-term infected persons with subsubtype A6 virus compared to subtype B virus has been shown. For all FPR cut-off algorithms, a significant correlation between subtype and HIV tropism was observed (p=0.0014 and p=0.013 for FPR 10 % and FPR 20 %, respectively). While among subtype B strains, at least 57 % were identified as R5X4/X4-tropic variants (for an FPR of 10 %), including two strains classified as X4-tropic; among HIV subsubtype A6 even at an FPR of 20 %, the frequency of R5X4/X4-tropic samples only slightly exceeded 22 %. It can be assumed that the dynamics of changes in HIV tropism depends on the virus subtype. Significant differences in the distribution of amino acid residues of the V3 region sequences in the examined group between R5-tropic and R5X4/X4-tropic strains of subsubtype A6 for positions 18 (χ2=7.616, p=0.0058), 21 (χ2=7.281, p=0.007), 24 (χ2=5.587, p=0.0181), and 34 (χ2=5.144, p=0.0233) have been demonstrated. Among the R5X4/X4-tropic strains of the A6 subsubtype, amino acid substitutions were registered at positions 6, 19, 21, 26, 29, 30, which were not found in the R5-tropic A6 strains. The high occurrence frequency of a number of mutations previously described as presumably associated with resistance to maraviroc and similar drugs may indicate a natural polymorphism characteristic of the A6 subsubtype, which does not correlate with resistance to CCR5 co-receptor antagonists.

HIV-1 Infection of Long-Lived Hematopoietic Precursors In Vitro and In Vivo

Latent reservoirs in human-immunodeficiency-virus-1 (HIV-1)-infected individuals represent a major obstacle in finding a cure for HIV-1. Hematopoietic stem and progenitor cells (HSPCs) have been described as potential HIV-1 targets, but their roles as HIV-1 reservoirs remain controversial. Here we provide additional evidence for the susceptibility of several distinct HSPC subpopulations to HIV-1 infection in vitro and in vivo. In vitro infection experiments of HSPCs were performed with different HIV-1 Env-pseudotyped lentiviral particles and with replication-competent HIV-1. Low-level infection/transduction of HSPCs, including hematopoietic stem cells (HSCs) and multipotent progenitors (MPP), was observed, preferentially via CXCR4, but also via CCR5-mediated entry. Multi-lineage colony formation in methylcellulose assays and repetitive replating of transduced cells provided functional proof of susceptibility of primitive HSPCs to HIV-1 infection. Further, the access to bone marrow samples from HIV-positive individuals facilitated the detection of HIV-1 gag cDNA copies in CD34+ cells from eight (out of eleven) individuals, with at least six of them infected with CCR5-tropic HIV-1 strains. In summary, our data confirm that primitive HSPC subpopulations are susceptible to CXCR4- and CCR5-mediated HIV-1 infection in vitro and in vivo, which qualifies these cells to contribute to the HIV-1 reservoir in patients.

Predicting HIV drug resistance using weighted machine learning method at target protein sequence-level

Acquired immune deficiency syndrome (AIDS) is a fatal disease caused by human immunodeficiency virus (HIV). Although 23 different drugs have been available, the treatment of AIDS remains challenging because the virus mutates very quickly which can lead to drug resistance. Therefore, predicting drug resistance before treatment is crucial for individual treatments. Here, based on HIV target protein sequence information, we analyzed 21-drug resistance caused by mutated residues using machine learning (ML) methods. To transform target sequences into numeric vectors, seven physicochemical properties were used, which can well represent the interacting characteristics of target proteins. Then, principal component analysis (PCA) method was adopted to reduce the feature dimensionality. Random forest (RF) and support vector machine (SVM) based on three different kernel functions, including linear, polynomial and radial basis function (RBF), were all employed. By comparisons, we found that RBF-based SVM method gives a comparative performance with RF model. Further, we added the weight information to RBF-based SVM method by four different weight evaluation methods of RF, eXtreme Gradient Boosting (XGB), CfsSubsetEval and ReliefFAttributeEval, respectively. Results show that the RF-weighted RBF-based SVM yield the superior performance and 13 out of 21 drug models provide the correlation coefficients (R²) over 0.8 and 3 of them are higher than 0.9. Finally, position-specific importance analysis indicates that most of the mutation residues with high RF weight scores are proved to be closely related with drug resistance, which has been revealed in previous reports. Overall, we can expect that this method can be a supplementary tool for predicting HIV drug resistance for newly discovered mutations. Here, based on HIV target protein sequence information, we analyzed 21-drug resistance caused by mutated residues using machine learning (ML) methods by fusing the weight information of different mutation positions.

Genetic source completeness of HIV-1 circulating recombinant forms (CRFs) predicted by multi-label learning

MOTIVATION

Infection with strains of different subtypes and the subsequent crossover reading between the two strands of genomic RNAs by host cells' reverse transcriptase are the main causes of the vast HIV-1 sequence diversity. Such inter-subtype genomic recombinants can become circulating recombinant forms (CRFs) after widespread transmissions in a population. Complete prediction of all the subtype sources of a CRF strain is a complicated machine learning problem. It is also difficult to understand whether a strain is an emerging new subtype and if so, how to accurately identify the new components of the genetic source.

RESULTS

We introduce a multi-label learning algorithm for the complete prediction of multiple sources of a CRF sequence as well as the prediction of its chronological number. The prediction is strengthened by a voting of various multi-label learning methods to avoid biased decisions. In our steps, frequency and position features of the sequences are both extracted to capture signature patterns of pure subtypes and CRFs. The method was applied to 7185 HIV-1 sequences, comprising 5530 pure subtype sequences and 1655 CRF sequences. Results have demonstrated that the method can achieve very high accuracy (reaching 99%) in the prediction of the complete set of labels of HIV-1 recombinant forms. A few wrong predictions are actually incomplete predictions, very close to the complete set of genuine labels.

AVAILABILITY AND IMPLEMENTATION

https://github.com/Runbin-tang/The-source-of-HIV-CRFs-prediction.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

THETA: a new genotypic approach for predicting HIV-1 CRF02-AG coreceptor usage

MOTIVATION

The circulating recombinant form of HIV-1 CRF02-AG is the most frequent non-B subtype in Europe. Anti-HIV therapy and pathophysiological studies on the impact of HIV-1 tropism require genotypic determination of HIV-1 tropism for non-B subtypes. But genotypic approaches based on analysis of the V3 envelope region perform poorly when used to determine the tropism of CRF02-AG. We, therefore, designed an algorithm based on information from the gp120 and gp41 ectodomain that better predicts the tropism of HIV-1 subtype CRF02-AG.

RESULTS

We used a bio-statistical method to identify the genotypic determinants of CRF02-AG coreceptor use. Toulouse HIV Extended Tropism Algorithm (THETA), based on a Least Absolute Shrinkage and Selection Operator method, uses HIV envelope sequence from phenotypically characterized clones. Prediction of R5X4/X4 viruses was 86% sensitive and that of R5 viruses was 89% specific with our model. The overall accuracy of THETA was 88%, making it sufficiently reliable for predicting the tropism of subtype CRF02-AG sequences.

AVAILABILITY AND IMPLEMENTATION

Binaries are freely available for download at https://github.com/viro-tls/THETA. It was implemented in Matlab and supported on MS Windows platform. The sequence data used in this work are available from GenBank under the accession numbers MK618182-MK618417.

Drug Resistance Prediction Using Deep Learning Techniques on HIV-1 Sequence Data

The fast replication rate and lack of repair mechanisms of human immunodeficiency virus (HIV) contribute to its high mutation frequency, with some mutations resulting in the evolution of resistance to antiretroviral therapies (ART). As such, studying HIV drug resistance allows for real-time evaluation of evolutionary mechanisms. Characterizing the biological process of drug resistance is also critically important for sustained effectiveness of ART. Investigating the link between "black box" deep learning methods applied to this problem and evolutionary principles governing drug resistance has been overlooked to date. Here, we utilized publicly available HIV-1 sequence data and drug resistance assay results for 18 ART drugs to evaluate the performance of three architectures (multilayer perceptron, bidirectional recurrent neural network, and convolutional neural network) for drug resistance prediction, jointly with biological analysis. We identified convolutional neural networks as the best performing architecture and displayed a correspondence between the importance of biologically relevant features in the classifier and overall performance. Our results suggest that the high classification performance of deep learning models is indeed dependent on drug resistance mutations (DRMs). These models heavily weighted several features that are not known DRM locations, indicating the utility of model interpretability to address causal relationships in viral genotype-phenotype data.

HIV-1 subtype C predicted co-receptor tropism in Africa: an individual sequence level meta-analysis

Background

Entry inhibitors, such as Maraviroc, hold promise as components of HIV treatment and/or pre-exposure prophylaxis in Africa. Maraviroc inhibits the interaction between HIV Envelope gp120 V3-loop and CCR5 coreceptor. HIV-1 subtype C (HIV-1-C) is predominant in Southern Africa and preferably uses CCR5 co-receptor. Therefore, a significant proportion of HIV-1-C CXCR4 utilizing viruses (X4) may compromise the effectiveness of Maraviroc. This analysis examined coreceptor preferences in early and chronic HIV-1-C infections across Africa.

Methods

African HIV-1-C Envelope gp120 V3-loop sequences sampled from 1988 to 2014 were retrieved from Los Alamos HIV Sequence Database. Sequences from early infections (< 186 days post infection) and chronic infections (> 186 days post infection) were analysed for predicted co-receptor preferences using Geno2Pheno [Coreceptor] 10% FPR, Phenoseq-C, and PSSMsinsi web tools. V3-loop diversity was determined, and viral subtype was confirmed by phylogenetic analysis. National treatment guidelines across Africa were reviewed for Maraviroc recommendation.

Results

Sequences from early (n = 6316) and chronic (n = 7338) HIV-1-C infected individuals from 10 and 15 African countries respectively were available for analyses. Overall, 518/6316 (8.2%; 95% CI 0.7–9.3) of early sequences were X4, with Ethiopia and Malawi having more than 10% each. For chronic infections, 8.3% (95% CI 2.4–16.2) sequences were X4 viruses, with Ethiopia, Tanzania, and Zimbabwe having more than 10% each. For sequences from early chronic infections (< 1 year post infection), the prevalence of X4 viruses was 8.5% (95% CI 2.6–11.2). In late chronic infections (≥ 5 years post infection), X4 viruses were observed in 36% (95% CI − 16.3 to 49.9), with two countries having relatively high X4 viruses: South Africa (43%) and Malawi (24%). The V3-loop amino acid sequence were more variable in X4 viruses in chronic infections compared to acute infections, with South Africa, Ethiopia and Zimbabwe showing the highest levels of V3-loop diversity. All sequences were phylogenetically confirmed as HIV-1-C and clustered according to their co-receptor tropism. In Africa, Maraviroc is registered only in South Africa and Uganda.

Conclusions

Our analyses illustrate that X4 viruses are present in significantly similar proportions in early and early chronic HIV-1 subtype C infected individuals across Africa. In contrast, in late chronic infections, X4 viruses increase 3–5 folds. We can draw two inferences from our observations: (1) to enhance the utility of Maraviroc in chronic HIV subtype C infections in Africa, prior virus co-receptor determination is needed; (2) on the flip side, research on the efficacy of CXCR4 antagonists for HIV-1-C infections is encouraged. Currently, the use of Maraviroc is very limited in Africa.

SCOTCH: subtype A coreceptor tropism classification in HIV-1

Motivation

The V3 loop of the gp120 glycoprotein of the Human Immunodeficiency Virus 1 (HIV-1) is considered to be responsible for viral coreceptor tropism. gp120 interacts with the CD4 receptor of the host cell and subsequently V3 binds either CCR5 or CXCR4. Due to the fact that the CCR5 coreceptor is targeted by entry inhibitors, a reliable prediction of the coreceptor usage of HIV-1 is of great interest for antiretroviral therapy. Although several methods for the prediction of coreceptor tropism are available, almost all of them have been developed based on only subtype B sequences, and it has been shown in several studies that the prediction of non-B sequences, in particular subtype A sequences, are less reliable. Thus, the aim of the current study was to develop a reliable prediction model for subtype A viruses.

Results

Our new model SCOTCH is based on a stacking approach of classifier ensembles and shows a significantly better performance for subtype A sequences compared to other available models. In particular for low false positive rates (between 0.05 and 0.2, i.e. recommendation in the German and European Guidelines for tropism prediction), SCOTCH shows significantly better prediction performances in terms of partial area under the curves and diagnostic odds ratios compared to existing tools, and thus can be used to reliably predict coreceptor tropism for subtype A sequences.

Availability and implementation

SCOTCH can be downloaded/accessed at http://www.heiderlab.de.

HIV-1 tropism prediction by the XGboost and HMM methods

Human Immunodeficiency Virus 1 (HIV-1) co-receptor usage, called tropism, is associated with disease progression towards AIDS. Furthermore, the recently developed and developing drugs against co-receptors CCR5 or CXCR4 open a new thought for HIV-1 therapy. Thus, knowledge about tropism is critical for illness diagnosis and regimen prescription. To improve tropism prediction accuracy, we developed two novel methods, the extreme gradient boosting based XGBpred and the hidden Markov model based HMMpred. Both XGBpred and HMMpred achieved higher specificities (72.56% and 72.09%) than the state-of-the-art methods Geno2pheno (61.6%) and G2p_str (68.60%) in a 10-fold cross validation test at the same sensitivity of 93.73%. Moreover, XGBpred had more outstanding performances (with AUCs 0.9483, 0.9464) than HMMpred (0.8829, 0.8774) on the Hivcopred and Newdb (created in this work) datasets containing larger proportions of hard-to-predict dual tropic samples in the X4-using tropic samples. Therefore, we recommend the use of our novel method XGBpred to predict tropism. The two methods and datasets are available via http://spg.med.tsinghua.edu.cn:23334/XGBpred/. In addition, our models identified that positions 5, 11, 13, 18, 22, 24, and 25 were correlated with HIV-1 tropism.

Performance of Geno2Pheno[coreceptor] to infer coreceptor use in human immunodeficiency virus type 1 (HIV-1) subtype A

BACKGROUND

Assessment of human immunodeficiency virus type 1 (HIV-1) coreceptor usage is required prior to treatment with the CCR5 antagonist maraviroc to exclude the presence of CXCR4-using (X4) strains. Genotype-based interpretation systems are mostly designed on subtype B and have been reported to be less accurate for subtype A/CRF02_AG.

OBJECTIVES

To evaluate the performance of the widely used Geno2Pheno[coreceptor] (G2P[c]) algorithm for prediction of coreceptor usage with subtype A/CRF02_AG vs. subtype B.

STUDY DESIGN

Co-receptor tropism of 24 subtype A/CRF02_AG and 24 subtype B viruses was measured phenotypically by a homebrew single-cycle assay and genotypically by using G2P[c]. Samples with discrepant genotype-phenotype results were analyzed by next generation sequencing (NGS) and interpreted by the NGS Geno2Pheno algorithm (G2P[454]).

RESULTS

At 10% false positive rate (FPR), the G2P[c]/phenotype discordance rate was 12.5% (n = 3) for subtype A/CRF02_AG and 8.3% (n = 2) for subtype B. Minority X4 species escaping detection by bulk sequencing but documented by NGS explained the two subtype B and possibly one subtype A/CRF02_AG discordant case. The other two subtype A/CRF02_AG miscalled by G2P[c] could be explained by X4 overcalling at borderline FPR and/or by algorithm failure.

DISCUSSION

Our study did not demonstrate relevantly higher G2P[c] inaccuracy with subtype A/CRF02_AG with respect to subtype B. Genotype/phenotype discordances can be due to different reasons, including but not limited to, algorithm inaccuracy. Very large genotype/phenotype correlation panels are required to detect and explain the reason for any consistent difference in genotypic tropism prediction for subtype A/CRF02_AG vs. subtype B.

Molecular Epidemiology of HIV-1 Virus in Puerto Rico: Novel Cases of HIV-1 Subtype C, D, and CRF-24BG

HIV-1 subtype B virus is the most prevalent subtype in Puerto Rico (PR), accounting for about 90% of infection in the island. Recently, other subtypes and circulating recombinant forms (CRFs), including F(12_BF), A (01_BF), and CRF-39 BF-like, have been identified. The purpose of this study is to assess the distribution of drug resistance mutations and subtypes in PR. A total of 846 nucleotide sequences from the period comprising 2013 through 2017 were obtained from our "HIV Genotyping" test file. Phylogenetic and molecular epidemiology analyses were performed to evaluate the evolutionary dynamics and prevalence of drug resistance mutations. According to our results, we detected a decrease in the prevalence of protease inhibitor, nucleoside reverse transcriptase inhibitor (NRTI), and non-NRTI (NNRTI) resistance mutations over time. In addition, we also detected recombinant forms and, for the first time, identified subtypes C, D, and CRF-24BG in PR. Recent studies suggest that non-subtypes B are associated with a high risk of treatment failure and disease progression. The constant monitoring of viral evolution and drug resistance mutation dynamics is important to establish appropriate efforts for controlling viral expansion.

Next generation sequencing reveals a high frequency of CXCR4 utilizing viruses in HIV-1 chronically infected drug experienced individuals in South Africa

BACKGROUND

Entry inhibitors, such as Maraviroc, bind to CCR5 inhibiting entry of CCR5 utilizing viruses (R5 viruses). In the course of HIV infection, CXCR4 utilizing viruses (X4 viruses) may emerge and outgrow R5 viruses, and potentially limit the effectiveness of Maraviroc. The use of Maraviroc is reserved for salvage therapy in South Africa.

OBJECTIVE

In this study, we examined the frequency of R5 and X4 viruses, using next generation sequencing, in patients under treatment to draw inferences on the utility of Maraviroc in a South African population.

STUDY DESIGN

Proviral DNA was isolated from peripheral blood mononuclear cells (PBMC) of 72 chronically HIV infected patients on antiretroviral treatment. HIV V3 loop gene was amplified and sequenced on an Illumina MiniSeq platform. Viral subtypes were determined by the jumping profile Hidden Markov Model (jpHMM) and REGA genotyping tools. De Novo consensus sequences were derived for the majority and minority populations for each patient using Geneious^® software version 8.1.5. HIV-1 tropism was inferred using PSSMsinsi, Geno2pheno and Phenoseq-C web-based tools.

RESULTS

Quality V3 loop sequences were obtained from 72 patients, with 5 years (range: 0-16) median duration on treatment. Subtypes A1, B and C viruses were identified at frequencies of 4% (3/72), 4% (3/72) and 92% (66/72) respectively. Fifty four percent (39/72) of patients exclusively harboured R5 viral quasispecies; and 21% (15/72) exclusively harbored X4 viral quasispecies. Twenty five percent of patients (18/72) harbored dual/mixture of R5X4 quasispecies. Of these 18 patients, about 28% (5/18) harbored the R5+X4, a mixture with a majority R5 and minority X4 viruses, while about 72% (13/18) harbored the R5X4+ mixture with a majority X4 and minority R5 viruses. The proportion of all patients who harbored X4 viruses either exclusively or dual/mixture was 46% (33/72). Thirty-five percent (23/66) of the patients who were of HIV-1 subtype C harboured X4 viruses (χ² = 3.58; p = .058), and 57% of these (13/23) harbored X4 viruses exclusively. CD4⁺ cell count less than 350 cell/μl was associated with the presence of X4 viruses (χ² = 4.99; p = .008).

CONCLUSION

The effectiveness of Maraviroc as a component in salvage therapy may be compromised for a significant number of chronically infected patients harboring CXCR4 utilizing viruses.

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-1C_EA) 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-1C_EA infected patients and four strains with known phenotype. Additionally, 68 clones from 18 patients (HIV-1B: 5, 01_AE: 7, HIV-1C_EA: 6) were used to determine the PTT in GHOST cell line. The respective V3-sequences were used for GTT. R5-tropic strains from HIV-1C_EA (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-1C_ET 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).

An Integrative Approach to Virus-Host Protein-Protein Interactions

Since cell regulation and protein expression can be dramatically altered upon infection by viruses, studying the mechanisms by which viruses infect cells and the regulatory networks they disrupt is essential to understanding viral pathogenicity. This line of study can also lead to discoveries about the workings of host cells themselves. Computational methods are rapidly being developed to investigate viral-host interactions, and here we highlight recent methods and the insights that they have revealed so far, with a particular focus on methods that integrate different types of data. We also review the challenges of working with viruses compared with traditional cellular biology, and the limitations of current experimental and informatics methods.

Prediction of coreceptor usage by five bioinformatics tools in a large Ethiopian HIV-1 subtype C cohort

Background

Genotypic tropism testing (GTT) has been developed largely on HIV-1 subtype B. Although a few reports have analysed the utility of GTT in other subtypes, more studies using HIV-1 subtype C (HIV-1C) are needed, considering the huge contribution of HIV-1C to the global epidemic.

Methods

Plasma was obtained from 420 treatment-naïve HIV-1C infected Ethiopians recruited 2009–2011. The V3 region was sequenced and the coreceptor usage was predicted by five tools: Geno2Pheno clinical–and clonal–models, PhenoSeq-C, C-PSSM and Raymond’s algorithm. The impact of baseline tropism on antiretroviral treatment (ART) outcome was evaluated.

Results

Of 352 patients with successful baseline V3 sequences, the proportion of predicted R5 virus varied between the methods by 12.5% (78.1%-90.6%). However, only 58.2% of the predictions were concordant and only 1.7% were predicted to be X4-tropic across the five methods. Compared pairwise, the highest concordance was between C-PSSM and Geno2Pheno clonal (86.4%). In bivariate intention to treat (ITT) analysis, R5 infected patients achieved treatment success more frequently than X4 infected at month six as predicted by Geno2Pheno clinical (77.8% vs 58.7%, P = 0.004) and at month 12 by C-PSSM (61.9% vs 46.6%, P = 0.038). However, in the multivariable analysis adjusted for age, gender, baseline CD4 and viral load, only tropism as predicted by C-PSSM showed an impact on month 12 (P = 0.04, OR 2.47, 95% CI 1.06–5.79).

Conclusion

Each of the bioinformatics models predicted R5 tropism with comparable frequency but there was a large discordance between the methods. Baseline tropism had an impact on outcome of first line ART at month 12 in multivariable ITT analysis but only based on prediction by C-PSSM which thus possibly could be used for predicting outcome of ART in HIV-1C infected Ethiopians.

Improving fold resistance prediction of HIV-1 against protease and reverse transcriptase inhibitors using artificial neural networks

Background

Drug resistance in HIV treatment is still a worldwide problem. Predicting resistance to antiretrovirals (ARVs) before starting any treatment is important. Prediction accuracy is essential, as low-accuracy predictions increase the risk of prescribing sub-optimal drug regimens leading to patients developing resistance sooner. Artificial Neural Networks (ANNs) are a powerful tool that would be able to assist in drug resistance prediction. In this study, we constrained the dataset to subtype B, sacrificing generalizability for a higher predictive performance, and demonstrated that the predictive quality of the ANN regression models have definite improvement for most ARVs.

Results

Trained regression ANNs were optimized for eight protease inhibitors, six nucleoside reverse transcriptase (RT) inhibitors and four non-nucleoside RT inhibitors by experimenting combinations of rare variant filtering (none versus 1 residue occurrence) and ANN topologies (1–3 hidden layers with 2, 4, 6, 8 and 10 nodes per layer). Single hidden layers (5–20 nodes) were used for training where overfitting was detected. 5-fold cross-validation produced mean R² values over 0.95 and standard deviations lower than 0.04 for all but two antiretrovirals.

Conclusions

Overall, higher accuracies and lower variances (compared to results published in 2016) were obtained by experimenting with various preprocessing methods, while focusing on the most prevalent subtype in the raw dataset (subtype B).We thus highlight the need to develop and make available subtype-specific datasets for developing higher accuracy in drug-resistance prediction methods.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-017-1782-x) contains supplementary material, which is available to authorized users.

Partial HIV C2V3 envelope sequence analysis reveals association of coreceptor tropism, envelope glycosylation and viral genotypic variability among Kenyan patients on HAART

Background

HIV-1 is highly variable genetically and at protein level, a property it uses to subvert antiviral immunity and treatment. The aim of this study was to assess if HIV subtype differences were associated with variations in glycosylation patterns and co-receptor tropism among HAART patients experiencing different virologic treatment outcomes.

Methods

A total of 118 HIV env C2V3 sequence isolates generated previously from 59 Kenyan patients receiving highly active antiretroviral therapy (HAART) were examined for tropism and glycosylation patterns. For analysis of Potential N-linked glycosylation sites (PNGs), amino acid sequences generated by the NCBI’s Translate tool were applied to the HIVAlign and the N-glycosite tool within the Los Alamos Database. Viral tropism was assessed using Geno2Pheno (G2P), WebPSSM and Phenoseq platforms as well as using Raymond’s and Esbjörnsson’s rules. Chi square test was used to determine independent variables association and ANOVA applied on scale variables.

Results

At respective False Positive Rate (FPR) cut-offs of 5% (p = 0.045), 10% (p = 0.016) and 20% (p = 0.005) for CXCR4 usage within the Geno2Pheno platform, HIV-1 subtype and viral tropism were significantly associated in a chi square test. Raymond’s rule (p = 0.024) and WebPSSM (p = 0.05), but not Phenoseq or Esbjörnsson showed significant associations between subtype and tropism. Relative to other platforms used, Raymond’s and Esbjörnsson’s rules showed higher proportions of X4 variants, while WebPSSM resulted in lower proportions of X4 variants across subtypes. The mean glycosylation density differed significantly between subtypes at positions, N277 (p = 0.034), N296 (p = 0.036), N302 (p = 0.034) and N366 (p = 0.004), with HIV-1D most heavily glycosylated of the subtypes. R5 isolates had fewer PNGs than X4 isolates, but these differences were not significant except at position N262 (p = 0.040). Cell-associated isolates from virologic treatment success subjects were more glycosylated than cell-free isolates from virologic treatment failures both for the NXT (p = 0.016), and for all the patterns (p = 0.011).

Conclusion

These data reveal significant associations of HIV-1 subtype diversity, viral co-receptor tropism, viral suppression and envelope glycosylation. These associations have important implications for designing therapy and vaccines against HIV. Heavy glycosylation and preference for CXCR4 usage of HIV-1D may explain rapid disease progression in patients infected with these strains.

Role of phenotypic investigation in the era of routine genotypic HIV-1 drug resistance testing

The emergence of drug resistance can seriously compromise HIV type-1 therapy and decrease therapeutic options. Resistance testing is highly recommended to guide treatment decisions and drug activity can be accurately predicted in the clinical setting through genotypic assays. While phenotypic systems are not suitable for monitoring drug resistance in routine laboratory practice, genotyping can misclassify unusual or complex mutational patterns, particularly with recently approved antivirals. In addition, phenotypic assays remain fundamental for characterizing candidate antiretroviral compounds. This review aims to discuss how phenotypic assays contributed to and still play a role in understanding the mechanisms of resistance of both licensed and investigational HIV type-1 inhibitors.