Structural and functional asymmetry of RING trimerization controls priming and extension events in TRIM5α autoubiquitylation

Abstract

TRIM5α is an E3 ubiquitin ligase of the TRIM family that binds to the capsids of primate immunodeficiency viruses and blocks viral replication after cell entry. Here we investigate how synthesis of K63-linked polyubiquitin is upregulated by transient proximity of three RING domains in honeycomb-like assemblies formed by TRIM5α on the surface of the retroviral capsid. Proximity of three RINGs creates an asymmetric arrangement, in which two RINGs form a catalytic dimer that activates E2-ubiquitin conjugates and the disordered N-terminus of the third RING acts as the substrate for N-terminal autoubiquitylation. RING dimerization is required for activation of the E2s that contribute to the antiviral function of TRIM5α, UBE2W and heterodimeric UBE2N/V2, whereas the proximity of the third RING enhances the rate of each of the two distinct steps in the autoubiquitylation process: the initial N-terminal monoubiquitylation (priming) of TRIM5α by UBE2W and the subsequent extension of the K63-linked polyubiquitin chain by UBE2N/V2. The mechanism we describe explains how recognition of infection-associated epitope patterns by TRIM proteins initiates polyubiquitin-mediated downstream events in innate immunity.

Fig. 1. Recombinant RING constructs mimic distinct relative RING arrangements in TRIM5α.

a Conserved tripartite motif (TRIM) domain arrangement in TRIM5α. b NMR structure of the human TRIM5α RING domain monomer (PDB: 2ECV). c Cartoon representation of the TRIM5α dimer, in which the two RING domains are presumed to exist as monomers. d The monomeric R1-WT RING construct used in this study. e Crystal structure of the rhesus TRIM5α RING (PDB: 4TKP) reveals the four-helix-bundle dimer interface. The isoleucine residue in the core of the four-helix bundle is highlighted in cyan here (I77) and in panel a (I76). f RING dimerization is thought to occur when multiple TRIM dimers come into proximity upon association with their binding partners. g RING dimerization is stabilized in the R2-WT tandem RING construct. h Crystal structure of the B-box trimer (PDB: 5IEA) reveals the structural arrangement of three B-box domains at the vertices of the honeycomb-like TRIM5α assemblies (i)^, . j Proximity of three RING domains at the vertices of the TRIM5α honeycomb was mimicked in the R3-WT construct by fusing RING domain to the T4 fibritin foldon, a compact trimerization domain.

Fig. 2. Conformational repertoire of distinct RING constructs as revealed by NMR spectroscopy and circular dichroism.

a ¹⁵N-TROSY HSQC spectra of different RING constructs reveal spectral signatures of RING dimerization. b Three representative spectral regions from (a) showing NMR signals that display distinctive chemical shift changes upon transition between the monomeric and the dimeric states of the RING domain. c Dependence of molar ellipticity on protein concentration reveals that almost three-quarters of total alpha-helical content is lost in the R1-WT construct as protein concentration is lowered from 100 μM to 6 μM (n = 2 independent experiments). d Cartoon representation of the distinct conformational and oligomeric states observed in different RING constructs.

Fig. 3. The contribution of RING dimerization to the catalytic enhancement of ubiquitin discharge from the UBE2N~Ub and UBE2W~Ub conjugates can be quantified using a FRET ubiquitin discharge assay.

a A cartoon schematic of the FRET ubiquitin discharge assay used in this study. b Kinetics of RING-catalyzed ubiquitin discharge from the heterodimeric UBE2N~Ub/V2 (0.1 µM) is well approximated by an exponential decay function in agreement with the Michaelis-Menten equation (Supplementary Methods). c In contrast, ubiquitin discharge from UBE2W~Ub (0.1 µM) is not well described by the Michaelis-Menten model. d UBE2N/V2 ubiquitin discharge rates are plotted against RING concentration and the initial slopes (k_cat/K_m) can be used for quantification of the catalytic activity (n = 2 independent experiments). e Quantification of UBE2W discharge (see “Methods” and Supplementary Methods for more details) establishes that UBE2W is also activated by RING dimerization, albeit less potently than UBE2N. [RING] concentration is shown here as total concentration of RING domains in each sample irrespective of the oligomerization state (e.g., 15 μM sample of the tandem RING construct R2-WT has [RING] = 30 μM) (n = 2 independent experiments). f Cartoon representation of the dimerization-dependent activation of both UBE2N~Ub and UBE2W~Ub by TRIM5α RING.

Fig. 4. Proximity of three RING domains promotes each of the two distinct steps in the autoubiquitylation of TRIM5α: the N-terminal monoubiquitylation (priming) and the extension of the TRIM5α-anchored K63-linked polyubiquitin chain.

a, b RING autoubiquitylation products visualized by WB of FLAG-tagged RING constructs. [RING] = 5 µM, [Ub] = 10 µM, [E2] = 0.1; 0.2; 0.5; 1 µM. 30 min incubation. a Monoubiquitylation of RING in the presence of the increasing amounts of UBE2W. b RING polyubiquitylation in the presence of both UBE2W and the heterodimeric UBE2N/V2. c Ubiquitylation products of different E2/E3 combinations analyzed side-by-side on the same SDS-PAGE gel. [RING] = 5 µM, [Ub] = 10 µM, [E2] = 1 µM. 30 min incubation. WB was performed with either anti-FLAG antibodies (top panel) or anti-K63-linked-Ub antibodies (bottom panel). d N-terminal anchoring of ubiquitin in the R3 WT construct is confirmed by mass spectrometry. e N-terminal monoubiquitylation is preferred over other modification sites by more than two orders of magnitude. f Kinetics and products of ubiquitin discharge from the AF594-UBE2W~AF488-Ub conjugate (0.1 μM) catalyzed by R2-WT ([RING] = 1 μM; [dimer] = 0.5 μM) or R3-WT constructs ([RING] = 0.5 μM; [trimer] = 0.17 μM) in the presence of [N5-Ub] = 5 μM. g Cartoon representation of the UBE2W reactions and the corresponding products. h Kinetics and products of ubiquitin discharge from the AF594-UBE2N~AF488-Ub/UBE2V2 conjugate (0.1 μM) catalyzed by R3-WT ([RING] = 0.5 μM; [trimer] = 0.17 μM) or Ub-R3-WT constructs ([RING] = 0.5 μM; [trimer] = 0.17 μM) in the presence of [N5-Ub] = μM. i Cartoon representation of the UBE2N/V2 reactions and the corresponding products.

Fig. 5. Transient proximity of three RING domains promoted by the association of TRIM5α with the retroviral capsid explains how TRIM5α autoubiquitylation is activated by capsid binding.

The honeycomb-like TRIM5α assembly on the surface of the retroviral capsid is required for the antiviral activity of the protein (see a recent comprehensive review for detailed bibliography). The architecture of the TRIM5α assembly suggests that the cooperativity between SPRY:capsid interactions and B-box:B-box interactions forms the basis of the pattern recognition functionality and promotes capsid binding. B-box trimerization at the vertices of the TRIM5α honeycomb brings three N-terminal RING domains into proximity. Here we show that proximity of three RINGs creates an asymmetric structural arrangement, which explains how RING can act as both the catalytic activator and the substrate in the ubiquitin transfer reactions. Fluorescent ubiquitin discharge assays reveal that proximity of three RINGs strongly enhances the rates of the two distinct ubiquitin transfer processes involved in the autoubiquitylation of TRIM5α: the N-terminal priming of TRIM5α with a single ubiquitin by UBE2W and the subsequent extension of the TRIM5α-attached K63-linked polyubiquitin chain by UBE2N/V2. The findings imply that both the priming and the extension are enhanced by the association with the capsid. Relative arrangement of protein subunits in the complexes mediating these events can be modeled based on extensive existing structural data (see “Methods”)^{, , –, –,} . The models (insets in the upper right and lower right corners of the figure) illustrate how proximity of the third RING and the flexibility of its backbone facilitate ubiquitin transfer in the two complexes. Collectively, the findings elucidate how the symmetry mismatch between B-box trimerization and RING dimerization enables the dual activator and substrate functionality of the RING and activates TRIM5α autoubiquitylation upon capsid binding.