Dynamic conformational changes in the rhesus TRIM5α dimer dictate the potency of HIV-1 restriction

Abstract

The TRIM5α protein from rhesus macaques (rhTRIM5α) mediates a potent inhibition of HIV-1 infection via a mechanism that involves the abortive disassembly of the viral core. We have demonstrated that alpha-helical elements within the Linker 2 (L2) region, which lies between the SPRY domain and the Coiled-Coil domain, influence the potency of restriction. Here, we utilize single-molecule FRET analysis to reveal that the L2 region of the TRIM5α dimer undergoes dynamic conformational changes, which results in the displacement of L2 regions by 25 angstroms relative to each other. Analysis of restriction enhancing or abrogating mutations in the L2 region reveal that restriction defective mutants are unable to undergo dynamic conformational changes and do not assume compact, alpha-helical conformations in the L2 region. These data suggest a model in which conformational changes in the L2 region mediate displacement of CA bound SPRY domains to induce the destabilization of assembled capsid during restriction.

Keywords: Coiled coil; Dimer; HIV-1; Molecular dynamics simulation; Restriction factor; Single molecule FRET; TRIM5alpha; Tripartite Motif; smFRET.

Figure 1. Structural Organization of TRIM5α Assemblies

A. The domain structure of TRIM5α, with color coded domain structure used throughout the manuscript. B. Putative structure of TRIM5α dimer, assembled from a homology model of the CC-L2 dimer of rhTRIM5α and the structures of individual domains (Abe et al., 2007; Biris et al., 2012; Goldstone et al., 2014). C. Putative domain structures within hexagonal TRIM5α assembly (Ganser-Pornillos et al., 2011).

Figure 2. Determinants of HIV-1 restriction within the L2 region of rhTRIM5α

A. Hela cells stably expressing rhTRIM5α were infected with serial dilution of HIV-1 GFP. GFP expression was measured 48 hours after infection. B. Summary of the rhTRIM5α characteristics for WT and the indicated mutants, described in Sastri et al (Sastri et al., 2014). Dimerization was assessed by glutaraldehyde crosslinking. % α-helicity was assessed using circular dichroism. Restriction was assessed as in A. CA binding was assessed by measuring co-precipitation with in vitro assembled CA. C. homology model of the rhTRIM5α CC-L2 dimer, similarly described in (Sastri et al., 2014). D. Location of fluorophore conjugation in dually labelled (A488, A594) rhTRIM5α dimers, as described in the text. E. Dually labelled rhTRIM5α dimers were crosslinked with 0, 1, or 2 mM glutaraldehyde for 5 minutes, subjected to electrophoresis and assessed via fluorescent imaging, as indicated. Results are representative of three or more individual experiments.

Figure 3. Three conformational states are observed in the rhTRIM5α dimer

A. Representative fluorescence intensity (donor green, acceptor red) and FRET efficiency (blue) traces obtained from individual CC-L2 dimers via TIRF microscopy. B. A composite histogram of FRET efficiencies compiled from individual traces and fit to three Gaussian distributions centered around 0.2, 0.6 and 0.8 (dashed lines). The overall fit is shown as a red line. C. Idealized FRET traces were obtained by fitting each trace to a three state Markov Model. D. Transition density plot generated from the modeled FRET trajectories.

Figure 4. Restricting enhancing or disrupting mutations in the CC-L2 region alter the stability of individual conformations of the rhTRIM5α dimer

A. Representative fluorescence intensity (donor green, acceptor red) and FRET efficiency (blue) traces obtained from individual HKN271-273AAA CC-L2 dimers via TIRF microscopy. A composite FRET histogram (right panel) compiled from individual traces was fit to three Gaussian distributions centered around 0.2, 0.6 and 0.8 (dashed lines). The overall fit is shown by the red line. B. Corresponding data for RRV275-277AAA CC-L2 dimers.

Figure 5. Molecular dynamics simulations of FRET states assumed by the rhTRIM5α dimer

A. Molecular dynamics simulations were used to generate models of each FRET state observed experimentally. Individual cysteines were introduced after rhTRIM5α residue 296 in our previously described homology model of the rhTRIM5α dimer (Sastri et al., 2014). Individual models were generated by separating the C-terminal cysteines by the indicated distance along the long axis of the dimer and artificially melting H3 and H3′ and allowing the model to relax in a 100 ns simulation. The coiled coil domains are colored orange and light orange, while residues in the L2 region are labelled blue and light blue. B. Steered molecular dynamics simulations were used to generate force profiles resisting the pulling of the C-terminal cysteines along the axis of the dimer at constant velocity. Force profiles, plotted against the distance between C-terminal cysteines (top panel) or simulation time (bottom panel), using both the 43 Å model shown in A (residues 132–296) or a previously published homology model of the rhTRIM5α dimer incorporating residues 297–302, are shown. Force measured on each arm of the dimer are shown as black and red traces, respectively. C. Structural intermediates observed during the simulation shown in B, at the indicated time and separation distance between C-termini. The coiled coil domains are colored orange and light orange, while residues in the L2 region are labelled blue and light blue.