During the maturation of the HIV virus, the HIV-1 capsid protein (CA) assembles with thousands of copies to forms the capsid core [ref 1] with a characteristic conical shape (see Figure 1 and Figure 2C). This core encloses the RNA viral genome. Upon the entry of the HIV in host cells, the capsid core is released into the cytoplasm, and it dissociates in connection with the reverse transcription in a not completely understood process. This leads to the importation of DNA viral genome in the host cell’s nucleus, where it is integrated in the host DNA to finalize the infection.
The critical role of CA protein, in early and late stages of the viral replication life cycle, has led to recent efforts on drug development, targeting the mature form of the protein. Currently, none of these molecules are used in clinic, and some face natural polymorphism and resistant mutations [ref 2]. Therefore, continued development of drugs targeting the CA protein is still needed.
CA protein consist of a sequence of 231 amino acids which folds into 3 different domains (Figure 2A): The N-terminal domain (N-ter), the linker, and the C-terminal domain (C-ter). This protein chain complexes with other chains to form hexamers (Figure 2B) or pentamers; which assemble together to form the fullerene-cone shape of the capsid core (Figure 2C). There are several models of the core assembly, but all are composed of ~200 hexamers, and exactly 12 pentamers.
The FightAIDS@Home team is working with World Community Grid to find active compounds which could attach to the CA proteins and mediate the assembly of the capsid core. This computational experiment will be performed using the docking software AutoDock VINA [ref 3].
Thanks to the volunteers, around 2 million molecules will be screened across ~50 conformations of the capsid protein, and hopefully lead to a reduced selection of molecules. This will be the starting point of a drug discovery process targeting the HIV-1 capsid protein.
This computational experiment will be performed using the docking software AutoDock VINA [ref 3].
With the support of our collaborators from the HIV Interaction and Viral Evolution (HIVE), experimental biding assays and infectivity assays will be conducted to determine if the selected compounds could be optimized as a promising drug candidate.
Based on X-ray structures of CA protein, models of the core, and computational analysis of their flexibility, four pockets of interest have been selected on the surface of the hexamer assembly (see Figure 3).
These pockets involve either one monomer (as pocket 2 along the linker domain), at the interface of two monomers (pocket 1 & 4), or at the six-fold interface (pocket 3).
Mutagenesis experiments revealed that core stability is fine-tuned to allow ordered disassembly during early stage of virus replication cycle [ref 4]. This is why selection of compounds will be done either for molecules which could stabilize or destabilize the hexamer; assuming that both actions could have impacts on the equilibrium of the core.
Before being included in FA@H, this capsid project began as a smaller virtual screening focused on pocket 1; the one which overlaps the binding site of the known inhibitor PF74.
This previous virtual screening of the HIV-1 capsid consisted of docking experiments (DE) using 60,393 commercially available compounds across 6 conformations of the mature capsid. DE were performed using the docking software AutoDock VINA. Each DE was replicated 5 times in order to analyse the consistency of the results. This led to 1.8 million DE in total.
This work was presented as a poster at the 2016 annual NIGMS AIDS-Related Structural Biology meeting which took place at the NIH (Bethesda, Maryland, USA).
Out of this screening, 27 compounds were selected for experimental assays. With the support of our collaborators from the HIVE (Sarafianos' Lab), 5 compounds were identified in thermal shift binding assays, and 1 of these 5 showed an effect in infectivity assays (unplubished results). Further investigations are currently being performed to study the potential optimisation of this compound.
A total of 36 different structures of the HIV-1 capsid protein are used for the computations.
They were selected after analysis of the flexibility of the backbone and the side chains in the vicinity of each of the 4 pockets. Contexts of both pentamer and hexamer complexes were investigated.
See the different steps of this target structure selection.
A total of 1,677,767 commercially available compounds are being virtually screened against the target structures.
This ligand library consists of subsets from the ZINC database including : maybridge, chembridge, fda approved drugs, and human metabolite database.
See some statistics on the ligand library.
After the computation, the results of each docking experiment are processed. The obtained ligand's pose is analysed in terms of ligand/receptor interactions, predicted affinity, consistency across replicates/targets, etc., in order to be compared to other ligand results, and finally to be ranked.
Based on these analyses, only a small set of compounds are selected for experimental assays.
See flowchart and details on the post-processing analysis of the results.
|BATCH NUMBER||COMPUTATION COMPLETED||POST-PROCESING COMPLETED|
1000012 -> 1001269
1001270 -> 1002527
1002528 -> 1003785
1003786 -> 1005043
1005044 -> 1006301
Each batch represents a set of 100,000 docking experiments (1 docking experiment = 1 ligand vs 1 pocket in 1 target).