The impact of individual human immunodeficiency virus type 1 protease mutations on drug susceptibility is highly influenced by complex interactions with the background protease sequence

Machine learning-guided design of potent darunavir analogs targeting HIV-1 proteases: A computational approach for antiretroviral drug discovery

In the pursuit of novel antiretroviral therapies for human immunodeficiency virus type-1 (HIV-1) proteases (PRs), recent improvements in drug discovery have embraced machine learning (ML) techniques to guide the design process. This study employs ensemble learning models to identify crucial substructures as significant features for drug development. Using molecular docking techniques, a collection of 160 darunavir (DRV) analogs was designed based on these key substructures and subsequently screened using molecular docking techniques. Chemical structures with high fitness scores were selected, combined, and one-dimensional (1D) screening based on beyond Lipinski's rule of five (bRo5) and ADME (absorption, distribution, metabolism, and excretion) prediction implemented in the Combined Analog generator Tool (CAT) program. A total of 473 screened analogs were subjected to docking analysis through convolutional neural networks scoring function against both the wild-type (WT) and 12 major mutated PRs. DRV analogs with negative changes in binding free energy (ΔΔ G bind ) compared to DRV could be categorized into four attractive groups based on their interactions with the majority of vital PRs. The analysis of interaction profiles revealed that potent designed analogs, targeting both WT and mutant PRs, exhibited interactions with common key amino acid residues. This observation further confirms that the ML model-guided approach effectively identified the substructures that play a crucial role in potent analogs. It is expected to function as a powerful computational tool, offering valuable guidance in the identification of chemical substructures for synthesis and subsequent experimental testing.

Inherently interpretable position-aware convolutional motif kernel networks for biological sequencing data

Artificial neural networks show promising performance in detecting correlations within data that are associated with specific outcomes. However, the black-box nature of such models can hinder the knowledge advancement in research fields by obscuring the decision process and preventing scientist to fully conceptualize predicted outcomes. Furthermore, domain experts like healthcare providers need explainable predictions to assess whether a predicted outcome can be trusted in high stakes scenarios and to help them integrating a model into their own routine. Therefore, interpretable models play a crucial role for the incorporation of machine learning into high stakes scenarios like healthcare. In this paper we introduce Convolutional Motif Kernel Networks, a neural network architecture that involves learning a feature representation within a subspace of the reproducing kernel Hilbert space of the position-aware motif kernel function. The resulting model enables to directly interpret and evaluate prediction outcomes by providing a biologically and medically meaningful explanation without the need for additional post-hoc analysis. We show that our model is able to robustly learn on small datasets and reaches state-of-the-art performance on relevant healthcare prediction tasks. Our proposed method can be utilized on DNA and protein sequences. Furthermore, we show that the proposed method learns biologically meaningful concepts directly from data using an end-to-end learning scheme.

Temporal Trends in HIV-1 Mutations Used for the Surveillance of Transmitted Drug Resistance

In 2009, a list of nonpolymorphic HIV-1 drug resistance mutations (DRMs), called surveillance DRMs (SDRMs), was created to monitor transmitted drug resistance (TDR). Since 2009, TDR increased and antiretroviral therapy (ART) practices changed. We examined the changing prevalence of SDRMs and identified candidate SDRMs defined as nonpolymorphic DRMs present on ≥ 1 expert DRM list and in ≥0.1% of ART-experienced persons. Candidate DRMs were further characterized according to their association with antiretrovirals and changing prevalence. Among NRTI-SDRMs, tenofovir-associated mutations increased in prevalence while thymidine analog mutations decreased in prevalence. Among candidate NRTI-SDRMs, there were six tenofovir-associated mutations including three which increased in prevalence (K65N, T69deletion, K70G/N/Q/T). Among candidate NNRTI-SDRMs, six that increased in prevalence were associated with rilpivirine (E138K/Q, V179L, H221Y) or doravirine (F227C/L) resistance. With the notable exceptions of I47A and I50L, most PI-SDRMs decreased in prevalence. Three candidate PI-SDRMs were accessory darunavir-resistance mutations (L10F, T74P, L89V). Adding the candidate SDRMs listed above was estimated to increase NRTI, NNRTI, and PI TDR prevalence by 0.1%, 0.3%, and 0.3%, respectively. We describe trends in the prevalence of nonpolymorphic HIV-1 DRMs in ART-experienced persons. These data should be considered in decisions regarding SDRM list updates and TDR monitoring.

A synergy of activity, stability, and inhibitor-interaction of HIV-1 protease mutants evolved under drug-pressure

A clinically-relevant, drug-resistant mutant of HIV-1 protease (PR), termed Flap+_(I54V) and containing L10I, G48V, I54V and V82A mutations, is known to produce significant changes in the entropy and enthalpy balance of drug-PR interactions, compared to wild-type PR. A similar mutant, Flap+_(I54A) , which evolves from Flap+_(I54V) and contains the single change at residue 54 relative to Flap+_(I54V) , does not. Yet, how Flap+_(I54A) behaves in solution is not known. To understand the molecular basis of V54A evolution, we compared nuclear magnetic resonance (NMR) spectroscopy, fluorescence spectroscopy, isothermal titration calorimetry, and enzymatic assay data from four PR proteins: PR (pWT), Flap+_(I54V) , Flap+_(I54A) , and Flap+_(I54) , a control mutant that contains only L10I, G48V and V82A mutations. Our data consistently show that selection to the smaller side chain at residue 54, not only decreases inhibitor affinity, but also restores the catalytic activity.

NMR and MD studies combined to elucidate inhibitor and water interactions of HIV-1 protease and their modulations with resistance mutations

Over the last two decades, both the sensitivity of NMR and the time scale of molecular dynamics (MD) simulation have increased tremendously and have advanced the field of protein dynamics. HIV-1 protease has been extensively studied using these two methods, and has presented a framework for cross-evaluation of structural ensembles and internal dynamics by integrating the two methods. Here, we review studies from our laboratories over the last several years, to understand the mechanistic basis of protease drug-resistance mutations and inhibitor responses, using NMR and crystal structure-based parallel MD simulations. Our studies demonstrate that NMR relaxation experiments, together with crystal structures and MD simulations, significantly contributed to the current understanding of structural/dynamic changes due to HIV-1 protease drug resistance mutations.

Structural studies on molecular mechanisms of Nelfinavir resistance caused by non-active site mutation V77I in HIV-1 protease

Background

The human immunodeficiency virus (HIV-1) is a retrovirus causing acquired immunodeficiency syndrome (AIDS), which has become a serious problem across the world and has no cure reported to date. Human immunodeficiency virus (HIV-1) protease is an attractive target for antiviral treatment and a number of therapeutically useful inhibitors have been designed against it. The emergence of drug resistant mutants of HIV-1 poses a serious problem for conventional therapies that have been used so far. Until now, thirteen protease inhibitors (PIs), major mutation sites and many secondary mutations have been listed in the HIV Drug Resistance Database. In this study, we have studied the effect of the V77I mutation in HIV-PR along with the co-occurring mutations L33F and K20T through multi-nanosecond molecular dynamics simulations. V77I is known to cause Nelfinavir (NFV) resistance in the subtype B population of HIV-1 protease. We have for the first time reported the effect of this clinically relevant mutation on the binding of Nelfinavir and the conformational flexibility of the protease.

Results

Two HIV-PR mutants have been considered in this study - the Double Mutant Protease (DBM) V77I-L33F and Triple Mutant Protease (TPM) V77I-K20T-L33F. The molecular dynamics simulation studies were carried out and the RMSD trajectories of the unliganded wild type and mutated protease were found to be stable. The binding affinity of NFV with wild type HIV-PR was very high with a Glide XP docking score of -9.3 Kcal/mol. NFV showed decreased affinity towards DBM with a docking score of -8.0 Kcal/mol, whereas its affinity increased towards TPM (Glide XP score: -10.3). Prime/MM-GBSA binding free energy of the wild type, DBM and TPM HIV-PR docked structures were calculated as -38.9, -11.1 and -42.6 Kcal/mol respectively. The binding site cavity volumes of wild type, DBM and TPM protease were 1186.1, 1375.5 and 1042.5 ų respectively.

Conclusion

In this study, we have studied the structural roles of the two HIV-PR mutations by conducting molecular dynamics simulation studies of the wild type and mutant HIV-1 PRs. The present study proposes that DBM protease showed greater flexibility and the flap separation was greater with respect to the wild type protease. The cavity size of the MD-stabilized DBM was also found to be increased, which may be responsible for the decreased interaction of Nelfinavir with the cavity residues, thus explaining the decreased binding affinity. On the other hand, the binding affinity of NFV for TPM was found to be enhanced, accounted for by the decrease in cavity size of the mutant which facilitated strong interactions with the flap residues. The flap separation of TPM was less than the wild type protease and the decreased cavity size may be responsible for its lower resistance, and hence, may be the reason for its lower clinical relevance.

C-5-Modified Tetrahydropyrano-Tetrahydofuran-Derived Protease Inhibitors (PIs) Exert Potent Inhibition of the Replication of HIV-1 Variants Highly Resistant to Various PIs, including Darunavir

We identified three nonpeptidic HIV-1 protease inhibitors (PIs), GRL-015, -085, and -097, containing tetrahydropyrano-tetrahydrofuran (Tp-THF) with a C-5 hydroxyl. The three compounds were potent against a wild-type laboratory HIV-1 strain (HIV-1(WT)), with 50% effective concentrations (EC50s) of 3.0 to 49 nM, and exhibited minimal cytotoxicity, with 50% cytotoxic concentrations (CC50) for GRL-015, -085, and -097 of 80, >100, and >100 μM, respectively. All the three compounds potently inhibited the replication of highly PI-resistant HIV-1 variants selected with each of the currently available PIs and recombinant clinical HIV-1 isolates obtained from patients harboring multidrug-resistant HIV-1 variants (HIVMDR). Importantly, darunavir (DRV) was >1,000 times less active against a highly DRV-resistant HIV-1 variant (HIV-1DRV(R) P51); the three compounds remained active against HIV-1DRV(R) P51 with only a 6.8- to 68-fold reduction. Moreover, the emergence of HIV-1 variants resistant to the three compounds was considerably delayed compared to the case of DRV. In particular, HIV-1 variants resistant to GRL-085 and -097 did not emerge even when two different highly DRV-resistant HIV-1 variants were used as a starting population. In the structural analyses, Tp-THF of GRL-015, -085, and -097 showed strong hydrogen bond interactions with the backbone atoms of active-site amino acid residues (Asp29 and Asp30) of HIV-1 protease. A strong hydrogen bonding formation between the hydroxyl moiety of Tp-THF and a carbonyl oxygen atom of Gly48 was newly identified. The present findings indicate that the three compounds warrant further study as possible therapeutic agents for treating individuals harboring wild-type HIV and/or HIVMDR.

IMPORTANCE

Darunavir (DRV) inhibits the replication of most existing multidrug-resistant HIV-1 strains and has a high genetic barrier. However, the emergence of highly DRV-resistant HIV-1 strains (HIVDRV(R) ) has recently been observed in vivo and in vitro. Here, we identified three novel HIV-1 protease inhibitors (PIs) containing a tetrahydropyrano-tetrahydrofuran (Tp-THF) moiety with a C-5 hydroxyl (GRL-015, -085, and -097) which potently suppress the replication of HIVDRV(R) . Moreover, the emergence of HIV-1 strains resistant to the three compounds was considerably delayed compared to the case of DRV. The C-5 hydroxyl formed a strong hydrogen bonding interaction with the carbonyl oxygen atom of Gly48 of protease as examined in the structural analyses. Interestingly, a compound with Tp-THF lacking the hydroxyl moiety substantially decreased activity against HIVDRV(R) . The three novel compounds should be further developed as potential drugs for treating individuals harboring wild-type and multi-PI-resistant HIV variants as well as HIVDRV(R) .

DOCK 6: Impact of new features and current docking performance

This manuscript presents the latest algorithmic and methodological developments to the structure-based design program DOCK 6.7 focused on an updated internal energy function, new anchor selection control, enhanced minimization options, a footprint similarity scoring function, a symmetry-corrected root-mean-square deviation algorithm, a database filter, and docking forensic tools. An important strategy during development involved use of three orthogonal metrics for assessment and validation: pose reproduction over a large database of 1043 protein-ligand complexes (SB2012 test set), cross-docking to 24 drug-target protein families, and database enrichment using large active and decoy datasets (Directory of Useful Decoys [DUD]-E test set) for five important proteins including HIV protease and IGF-1R. Relative to earlier versions, a key outcome of the work is a significant increase in pose reproduction success in going from DOCK 4.0.2 (51.4%) → 5.4 (65.2%) → 6.7 (73.3%) as a result of significant decreases in failure arising from both sampling 24.1% → 13.6% → 9.1% and scoring 24.4% → 21.1% → 17.5%. Companion cross-docking and enrichment studies with the new version highlight other strengths and remaining areas for improvement, especially for systems containing metal ions. The source code for DOCK 6.7 is available for download and free for academic users at http://dock.compbio.ucsf.edu/.

Reverse transcriptase backbone can alter the polymerization and RNase activities of non-nucleoside reverse transcriptase mutants K101E+G190S

Previous work by our group showed that human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) containing non-nucleoside RT inhibitor (NNRTI) drug resistance mutations has defects in RNase H activity as well as reduced amounts of RT protein in virions. These deficits correlate with replication fitness in the absence of NNRTIs. Viruses with the mutant combination K101E+G190S replicated better in the presence of NNRTIs than in the absence of drug. Stimulation of virus growth by NNRTIs occurred during the early steps of the virus life cycle and was modulated by the RT backbone sequence in which the resistance mutations arose. We wanted to determine what effects RT backbone sequence would have on RT content and polymerization and RNase H activities in the absence of NNRTIs. We compared a NL4-3 RT with K101E+G190S to a patient-isolate RT sequence D10 with K101E+G190S. We show here that, unlike the NL4-3 backbone, the D10 backbone sequence decreased the RNA-dependent DNA polymerization activity of purified recombinant RT compared to WT. In contrast, RTs with the D10 backbone had increased RNase H activity compared to WT and K101E+G190S in the NL4-3 backbone. D10 virions also had increased amounts of RT compared to K101E+G190S in the NL4-3 backbone. We conclude that the backbone sequence of RT can alter the activities of the NNRTI drug-resistant mutant K101E+G190S, and that identification of the amino acids responsible will aid in understanding the mechanism by which NNRTI drug-resistant mutants alter fitness and NNRTIs stimulate HIV-1 virus replication.

Darunavir is a good third-line antiretroviral agent for HIV type 1-infected patients failing second-line protease inhibitor-based regimens in South India

Eleven protease mutations have been associated with reduced susceptibility to darunavir. In this study of 87 HIV-1-infected patients experiencing virological failure to second-line regimens containing protease inhibitors boosted with ritonavir (viral load >1,000 HIV RNA copies/ml), we observed a low prevalence (3%) of ≥3 darunavir resistance-associated mutations, indicating that this drug may be a good option for third-line antiretroviral therapy in southern India.

Loss of the protease dimerization inhibition activity of tipranavir (TPV) and its association with the acquisition of resistance to TPV by HIV-1

Tipranavir (TPV), a protease inhibitor (PI) inhibiting the enzymatic activity and dimerization of HIV-1 protease, exerts potent activity against multi-PI-resistant HIV-1 isolates. When a mixture of 11 multi-PI-resistant (but TPV-sensitive) clinical isolates (HIV(11MIX)), which included HIV(B) and HIV(C), was selected against TPV, HIV(11MIX) rapidly (by 10 passages [HIV(11MIX)(P10)]) acquired high-level TPV resistance and replicated at high concentrations of TPV. HIV(11MIX)(P10) contained various amino acid substitutions, including I54V and V82T. The intermolecular FRET-based HIV-1 expression assay revealed that TPV's dimerization inhibition activity against cloned HIV(B) (cHIV(B)) was substantially compromised. The introduction of I54V/V82T into wild-type cHIV(NL4-3) (cHIV(NL4-3(I54V/V82T))) did not block TPV's dimerization inhibition or confer TPV resistance. However, the introduction of I54V/V82T into cHIV(B) (cHIV(B)(I54V/V82T)) compromised TPV's dimerization inhibition and cHIV(B)(I54V/V82T) proved to be significantly TPV resistant. L24M was responsible for TPV resistance with the cHIV(C) genetic background. The introduction of L24M into cHIV(NL4-3) (cHIV(NL4-3(L24M))) interfered with TPV's dimerization inhibition, while L24M increased HIV-1's susceptibility to TPV with the HIV(NL4-3) genetic background. When selected with TPV, cHIV(NL4-3(I54V/V82T)) most readily developed TPV resistance and acquired E34D, which compromised TPV's dimerization inhibition with the HIV(NL4-3) genetic background. The present data demonstrate that certain amino acid substitutions compromise TPV's dimerization inhibition and confer TPV resistance, although the loss of TPV's dimerization inhibition is not always associated with significantly increased TPV resistance. The findings that TPV's dimerization inhibition is compromised with one or two amino acid substitutions may explain at least in part why the genetic barrier of TPV against HIV-1's development of TPV resistance is relatively low compared to that of darunavir.

Short communication: Phenotypic protease inhibitor resistance and cross-resistance in the clinic from 2006 to 2008 and mutational prevalences in HIV from patients with discordant tipranavir and darunavir susceptibility phenotypes

To test tipranavir (TPV) or darunavir (DRV) as treatment options for patients with phenotypic resistance to protease inhibitors (PIs), including lopinavir, saquinavir, atazanavir, and fosamprenavir, the PhenoSense GT database was analyzed for susceptibility to DRV or TPV among PI-resistant isolates. The Monogram Biosciences HIV database (South San Francisco, CA) containing 7775 clinical isolates (2006-2008) not susceptible to at least one first-generation PI was analyzed. Phenotypic responses [resistant (R), partially susceptible (PS), or susceptible (S)] were defined by upper and lower clinical cut-offs to each PI. Genotypes were screened for amino acid substitutions associated with TPV-R/DRV-S and TPV-S/DRV-R phenotypes. In all, 4.9% (378) of isolates were resistant to all six PIs and 31.0% (2407) were resistant to none. Among isolates resistant to all four first-generation PIs, DRV resistance increased from 21.2% to 41.9% from 2006 to 2008, respectively, and resistance to TPV remained steady (53.9 to 57.3%, respectively). Higher prevalence substitutions in DRV-S/TPV-R isolates versus DRV-R/TPV-S isolates, respectively, were 82L/T (44.4% vs. 0%) and 83D (5.8% vs. 0%). Higher prevalence substitutions in DRV-R/TPV-S virus were 50V (0.0% vs. 28.9%), 54L (1.0% vs. 36.1%), and 76V (0.4% vs. 15.5%). Mutations to help predict discordant susceptibility to DRV and TPV in isolates with reduced susceptibility to other PIs were identified. DRV resistance mutations associated with improved virologic response to TPV were more prevalent in DRV-R/TPV-S isolates. TPV resistance mutations were more prevalent in TPV-R and DRV-S isolates. These results confirm the impact of genotype on phenotype, illustrating how HIV genotype and phenotype data assist regimen optimization.

A Perturbation Method for Inference on Regularized Regression Estimates

Analysis of high dimensional data often seeks to identify a subset of important features and assess their effects on the outcome. Traditional statistical inference procedures based on standard regression methods often fail in the presence of high-dimensional features. In recent years, regularization methods have emerged as promising tools for analyzing high dimensional data. These methods simultaneously select important features and provide stable estimation of their effects. Adaptive LASSO and SCAD for instance, give consistent and asymptotically normal estimates with oracle properties. However, in finite samples, it remains difficult to obtain interval estimators for the regression parameters. In this paper, we propose perturbation resampling based procedures to approximate the distribution of a general class of penalized parameter estimates. Our proposal, justified by asymptotic theory, provides a simple way to estimate the covariance matrix and confidence regions. Through finite sample simulations, we verify the ability of this method to give accurate inference and compare it to other widely used standard deviation and confidence interval estimates. We also illustrate our proposals with a data set used to study the association of HIV drug resistance and a large number of genetic mutations.

Loss of protease dimerization inhibition activity of darunavir is associated with the acquisition of resistance to darunavir by HIV-1

Dimerization of HIV protease is essential for the acquisition of protease's proteolytic activity. We previously identified a group of HIV protease dimerization inhibitors, including darunavir (DRV). In the present work, we examine whether loss of DRV's protease dimerization inhibition activity is associated with HIV development of DRV resistance. Single amino acid substitutions, including I3A, L5A, R8A/Q, L24A, T26A, D29N, R87K, T96A, L97A, and F99A, disrupted protease dimerization, as examined using an intermolecular fluorescence resonance energy transfer (FRET)-based HIV expression assay. All recombinant HIV(NL4-3)-based clones with such a protease dimerization-disrupting substitution failed to replicate. A highly DRV-resistant in vitro-selected HIV variant and clinical HIV strains isolated from AIDS patients failing to respond to DRV-containing antiviral regimens typically had the V32I, L33F, I54M, and I84V substitutions in common in protease. None of up to 3 of the 4 substitutions affected DRV's protease dimerization inhibition, which was significantly compromised by the four combined substitutions. Recombinant infectious clones containing up to 3 of the 4 substitutions remained sensitive to DRV, while a clonal HIV variant with all 4 substitutions proved highly resistant to DRV with a 205-fold 50% effective concentration (EC(50)) difference compared to HIV(NL4-3). The present data suggest that the loss of DRV activity to inhibit protease dimerization represents a novel mechanism contributing to HIV resistance to DRV. The finding that 4 substitutions in PR are required for significant loss of DRV's protease dimerization inhibition should at least partially explain the reason DRV has a high genetic barrier against HIV's acquisition of DRV resistance.

HIV-1 protease mutations and protease inhibitor cross-resistance

The effects of many protease inhibitor (PI)-selected mutations on the susceptibility to individual PIs are unknown. We analyzed in vitro susceptibility test results on 2,725 HIV-1 protease isolates. More than 2,400 isolates had been tested for susceptibility to fosamprenavir, indinavir, nelfinavir, and saquinavir; 2,130 isolates had been tested for susceptibility to lopinavir; 1,644 isolates had been tested for susceptibility to atazanavir; 1,265 isolates had been tested for susceptibility to tipranavir; and 642 isolates had been tested for susceptibility to darunavir. We applied least-angle regression (LARS) to the 200 most common mutations in the data set and identified a set of 46 mutations associated with decreased PI susceptibility of which 40 were not polymorphic in the eight most common HIV-1 group M subtypes. We then used least-squares regression to ascertain the relative contribution of each of these 46 mutations. The median number of mutations associated with decreased susceptibility to each PI was 28 (range, 19 to 32), and the median number of mutations associated with increased susceptibility to each PI was 2.5 (range, 1 to 8). Of the mutations with the greatest effect on PI susceptibility, I84AV was associated with decreased susceptibility to eight PIs; V32I, G48V, I54ALMSTV, V82F, and L90M were associated with decreased susceptibility to six to seven PIs; I47A, G48M, I50V, L76V, V82ST, and N88S were associated with decreased susceptibility to four to five PIs; and D30N, I50L, and V82AL were associated with decreased susceptibility to fewer than four PIs. This study underscores the greater impact of nonpolymorphic mutations compared with polymorphic mutations on decreased PI susceptibility and provides a comprehensive quantitative assessment of the effects of individual mutations on susceptibility to the eight clinically available PIs.

The evolution of protease mutation 76V is associated with protease mutation 46I and gag mutation 431V

Recently, first-line lopinavir failure was observed due to protease mutation 76V. In the present study, we found 76V associated with protease mutation 46I and gag cleavage-site mutation 431V. Longitudinal analysis of patients failing protease inhibitor therapies demonstrated that 76V strictly occurs either together with 46I and/or 431V or in HIV isolates already harbouring one of both mutations. Therefore, all three mutations seem to cooperate in terms of protease inhibitor resistance.