The impact of Gag non-cleavage site mutations on HIV-1 viral fitness from integrative modelling and simulations

Abstract

The high mutation rate in retroviruses is one of the leading causes of drug resistance. In human immunodeficiency virus type-1 (HIV-1), synergistic mutations in its protease and the protease substrate - the Group-specific antigen (Gag) polyprotein - work together to confer drug resistance against protease inhibitors and compensate the mutations affecting viral fitness. Some Gag mutations can restore Gag-protease binding, yet most Gag-protease correlated mutations occur outside of the Gag cleavage site. To investigate the molecular basis for this, we now report multiscale modelling approaches to investigate various sequentially cleaved Gag products in the context of clinically relevant mutations that occur outside of the cleavage sites, including simulations of the largest Gag proteolytic product in its viral membrane-bound state. We found that some mutations, such as G123E and H219Q, involve direct interaction with cleavage site residues to influence their local environment, while certain mutations in the matrix domain lead to the enrichment of lipids important for Gag targeting and assembly. Collectively, our results reveal why non-cleavage site mutations have far-reaching implications outside of Gag proteolysis, with important consequences for drugging Gag maturation intermediates and tackling protease inhibitor resistance.

Keywords: Group-specific antigen (Gag); HIV-1; Integrative modelling; Multiscale simulation; Protease inhibitor drug resistance.

Graphical abstract

Fig. 1

Modelling HIV-1 Gag MA-CA-SP1. (A) The crystal structure of the MA trimer (PDB: 1HIW). (B) The crystal structure of the immature CA hexamer (PDB: 5L93). (C) The hexamer of trimers model for the MA domains as supported by EM. (D) A model of the MA-CA-SP1 hexamer. The central subunits of the MA trimers are connected to the CA domains and coloured as in (B), whereas the peripheral MA subunits are coloured grey.

Fig. 2

HIV-1 Gag MA-CA-SP1 model. (A) An atomic model of MA-CA-SP1 hexamer built using the crystal structures of MA trimer and immature CA hexamer shown in cartoon representation. The MA domain (cyan), CA domain (green), and SP1 domain (dark grey) are shown in different colours. The positions of seven unique non-cleavage site mutations are shown in red. Residue numbering for each domain and their approximate positions are shown on the right. (B) The final snapshot from one of the 15 µs CG simulations of a membrane-bound WT MA-CA-SP1 hexamer showing contraction of the linker connecting MA and CA domains. Protein backbone and lipids are shown in licorice representation. N-terminal myristate (red), PC (white), PE (light grey), PS (pink), PIP2 (orange), sphingomyelin (purple) and cholesterol (dark green) are shown in different colours. (C) Solvent accessible surface area (SASA) of the MA-CA cleavage site region (residue 128–137) over the course of 15 µs CG simulations. Thin lines represent the six individual subunits and thick lines show the running average. Data were averaged over four independent simulations. (D) Average percentage of contact made by the cleavage site residues with the rest of the protein, mapped onto the backbone of a single subunit of MA-CA-SP1 from the WT simulations. Data were averaged over six subunits and four independent simulations. The cut-off distance used for contact analysis was 0.6 nm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Interactions between non-cleavage site mutations and cleavage site residues. (A) Snapshot at the beginning and end of the CG simulations of WT MA-CA-SP1 showing the interaction of residue G123 (sphere representation) and cleavage site residues (thick licorice representation). Two subunits are shown in red and green. (B) Percentage of contacts across the entire CG simulation sampling made between G123/E123 with each of the residues of the MA-CA cleavage site. This is averaged over the six subunits of MA-CA-SP1 and the four independent trajectories. The cut-off distance used for contact analysis was 0.6 nm. (C) Atomistic model, derived from the final snapshot of one of the CG simulations, highlighting E123 in mutant Gag (cyan) and nearby cleavage site residues (pink). (D) Snapshot at the beginning and end of the CG simulations showing the interaction between H219 from one subunit to cleavage site residues on an adjacent subunit. (E) Contact analysis similar to (B) for H219/Q219. (F) Atomistic model derived from the CG simulations of WT Gag highlighting H219 in WT Gag (cyan) and surrounding cleavage site residues (pink). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

PIP2 is enriched around MA domain. (A) The percentage of PIP2 found within 0.6 nm of the MA domain during CG simulations of WT and mutant MA-CA-SP1 models. Error bars indicate standard deviation from four independent simulations. (B) Two-dimensional number density map of PIP2 for CG simulations of mutant 3 (Simulation 2 in Table S2) averaged over the last 5 µs. (C) Snapshots from one of the simulations of mutant 3 showing PIP2 binding at the interface of the MA trimer. The position of mutant residue, L75R, is shown in pink stick representation, whereas PIP2 is shown in orange surface representation. The rest of the MA domain is shown in grey. (D) Snapshot from simulations of mutant 7 (Simulation 4 in Table S2) showing PIP2 binding facilitated by E40K mutation, displayed in similar colours and representation as in (C). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5

The dynamics of CypA-binding loop. (A) Per-residue root mean square fluctuations (RMSF) of the CA from two independent 500 ns apo simulations. Thin lines show the RMSF of each individual CA subunit, whilst the thick lines show the average from each simulation. Simulation with neutral H219 is shown in black, positively charged H219 in blue, and mutant H219Q in red. Dotted box indicates the loop region where CypA binds. (B) Enlarged RMSF plot for CypA-binding loop. Average from the two simulations is shown and error bars indicate standard deviations between repeat simulations. (C) Atomic contact analysis performed between residues at position 219 and the rest of the CypA-binding loop, including E230 (indicated by red arrowhead). Average values from two simulations are shown and error bars indicate standard deviations between repeat simulations. Cut-off distance for contact analysis is 0.4 nm. (D) Representative structures of the CypA-binding loop from the simulations with positively charged H219 (left) and mutant H219Q (right) calculated using cluster analysis. The central structures (the structure with the lowest RMSD from all other structures) of the top clusters are shown (more details in Figure S8). In the former, H219 can form a salt bridge interaction with residue E230 found on the opposite side of the loop, represented as dotted line. (E) The distribution of minimum distance between the hydrogen atoms bonded to the two nitrogen atoms on the side chain of the protonated H219 and the two oxygen atoms on the side chain of E230. Data taken from all six CA subunits and both repeat simulations. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6

Atomistic simulations of mature CA with CypA. (A) Side and top views of mature CA hexamer (PDB: 6BHT) with each subunit bound to CypA (aligned using PDB: 1AK4). Residue H219 is shown in van der Waals representation, while IP6 bound in the center of the hexamer is shown in stick representation. (B) Buried surface area between CA and CypA throughout 500 ns simulations. Data taken from two independent simulations of CA with neutral H219 (black), protonated H219 (blue) and H219Q mutant (red). Thin lines indicate values from each of the six subunits, whilst the thick lines show the running averages. Red arrows show transient dissociation of CypA from CA in H219Q mutant simulations. (C) Residues on CypA that made contacts specifically with the side chains of neutral and protonated H219 (left and centre, respectively) from the WT CA simulations and Q219 (right) from the mutant CA simulations. CypA is shown in ribbon representation, the colour and thickness of which represent the percentage of contact made with CA during the simulations. A corresponding plot of per-residue contact percentages for each system is shown underneath, with key residues labelled. The cut-off distance used for contact analysis was 0.4 nm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7

The role of the Q199H mutation in CA oligomerisation. (A) (Left) A snapshot of two subunits of the mature WT CA hexamer in cartoon representation. The Q199 residue bridging the N-terminal domain (pink) of one subunit and the C-terminal domain (cyan) of an adjacent subunit is shown in van der Waals representation. (Right) Enlarged image of the hexameric interface showing residues found within 0.4 nm of the Q199 residue. (B) Percentage contact made by residue Q199 (WT), neutral or positively charged H199 (mutants) with residues from adjacent CA subunits. Contacts are averaged over all six subunits and two independent simulations. Error bars indicate standard deviations between repeat simulations. The cut-off distance used for contact analysis was 0.4 nm. (C) Short-range Coulombic pairwise interaction energies between residue at position 199 on one subunit and surrounding residues on the adjacent subunit. Thin lines indicate values from each of the six subunits and two independent simulations, while thick lines show the running averages. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)