TRIAD1 and HHARI bind to and are activated by distinct neddylated Cullin-RING ligase complexes

Abstract

RING (Really Interesting New Gene)-in-between-RING (RBR) enzymes are a distinct class of E3 ubiquitin ligases possessing a cluster of three zinc-binding domains that cooperate to catalyse ubiquitin transfer. The regulation and biological function for most members of the RBR ligases is not known, and all RBR E3s characterized to date are auto-inhibited for in vitro ubiquitylation. Here, we show that TRIAD1 and HHARI, two members of the Ariadne subfamily ligases, associate with distinct neddylated Cullin-RING ligase (CRL) complexes. In comparison to the modest E3 ligase activity displayed by isolated TRIAD1 or HHARI, binding of the cognate neddylated CRL to TRIAD1 or HHARI greatly stimulates RBR ligase activity in vitro, as determined by auto-ubiquitylation, their ability to stimulate dissociation of a thioester-linked UBCH7∼ubiquitin intermediate, and reactivity with ubiquitin-vinyl methyl ester. Moreover, genetic evidence shows that RBR ligase activity impacts both the levels and activities of neddylated CRLs in vivo. Cumulatively, our work proposes a conserved mechanism of CRL-induced Ariadne RBR ligase activation and further suggests a reciprocal role of this special class of RBRs as regulators of distinct CRLs.

Figure 1

Characterization of TRIAD1 ubiquitin ligase activity. (A) HEK293 lysates were subjected to immunoprecipitation with pre-immune IgG or anti-TRIAD1 antibody and immunoblotted with the indicated antibodies. (B) In vitro nickel-NTA precipitations were performed using recombinant His₆-TRIAD1 and untagged UBCH7. Binding was detected by immunoblot analysis of the assay supernatants (S/N) and pellets with the indicated antibodies. (C) GFP-tagged wild-type or mutant TRIAD1 containing the amino-acid substitutions indicated in the upper schematic (showing the domain structure of the protein, including the Ariadne domain) was stably expressed in HEK293 cells and immunoadsorbed using anti-GFP agarose. The inputs and immunopellets were immunoblotted as indicated. The anti-TRIAD1 antibody detects both the endogeneous and slower migrating exogeneous proteins. (D) UBCH7∼ubiquitin thioester was incubated with increasing concentrations of TRIAD1 (0.15–3.6 μM) at 37°C for 60 min. Reaction products were resolved on non-reducing SDS–PAGE gels and visualized by SimplyBlue staining (upper panel) or immunoblot analysis using anti-UBCH7 antibody. (E) UBCH7∼ubiquitin hydrolysis was assayed in the presence of 1.8 μM TRIAD1 at 37°C for indicated time points and visualized by SimplyBlue staining. (F) TRIAD1 auto-ubiquitylation assay (t=90 min) with UBCH7 and the indicated mutants in the TRIAD1 RING2 domain. Also included are control assays lacking ATP or TRIAD1. (G) TRIAD1 auto-ubiquitylation (t=90 min) using UBCH7 with WT and mutant forms of ubiquitin to establish ubiquitin linkage preference. A lysine-less ubiquitin (K all R) was used, as were single-lysine versions in which six of ubiquitin’s seven lysines were mutated to arginine (K all R except) with the number representing the remaining Lys residue. Upper panel represents an anti-ubiquitin immunoblot; lower panel is SimplyBlue stained to determine equal ubiquitin input.

Figure 2

TRIAD1 associates with cullin-5. (A) Table showing proteins identified by mass spectrometry in GFP immunoprecipitates from a GFP-TRIAD1 expressing cell line. (B) Immunoadsorption of endogeneous CUL5 and TRIAD1 from HEK293 cell lysates, followed by immunoblotting with the indicated antibodies. The slower migrating of the two CUL5-reactive bands is the neddylated form of the protein (CUL5-N8). (C) Immunoblots of anti-HA immunopellets and cell lysates from HEK293 cells stably expressing the indicated HA-tagged cullins. (D) Anti-GFP immunopellets and cell lysates from HEK293 cells stably expressing the indicated GFP-tagged TRIAD1 truncation constructs. (E) Alignment of the N-terminus of human (Hs) TRIAD1 with its mouse (Mm), chicken (Gg), Xenopus (Xl) and Drosophila (Dm) homologues, as well as the closely related ligase HHARI from the same species. (F) Wild-type or mutant HA-CUL5 was immunoprecipitated from stably transfected HEK293 cells and immunoblotted with the indicated antibodies. HA-CUL5 mut1: R417E, K418E, K423E, K424E mutant; HA-CUL5 mut2: R417E, K423E, R683E mutant; HA-CUL5 mut3: K676E, K679E, K682E, R683E mutant.

Figure 3

The TRIAD1/CUL5 interaction is neddylation dependent. (A) HEK293 cells stably expressing GFP-TRIAD1 or GFP-only control were treated overnight with 1 μM MLN4924 before immunoprecipitation with anti-GFP antibodies and immunoblotting as indicated. (B) Immunoblots of cell lysates and anti-HA immunopellets from HEK293 cells stably expressing HA-tagged WT and mutant CUL5. (C) Recombinant MBP-TRIAD1 was used to capture non-neddylated and neddylated recombinant CUL5–RBX2 complex in an in vitro binding assay. (D) Gel filtration chromatography analysis of TRIAD1 mixed either with non-neddylated (upper graph) or with neddylated CUL5–RBX2 (lower graph). (E) GFP-tagged WT and mutant TRIAD1 were immunoprecipitated from HEK293 cells and immunoblotted to detect CUL5 binding. LVW/AAA refers to the triple mutant L84A/V86A/W100A. *Indicates a non-specific band. (F) NEDD8- or ubiquitin-agarose beads (or beads only control) were incubated with recombinant wild-type TRIAD1 (left panel) or FLAG-UBXN7 (right panel). Binding was detected by immunoblotting.

Figure 4

Neddylated CUL5–RBX2 complex stimulates TRIAD1 E3 ligase activity in vitro. (A) UBCH7∼Ub discharge assays with 0.25 μM TRIAD1 were performed in the absence or presence of 30 nM CUL5–RBX2 (lane 4), 30 nM neddylated CUL5–RBX2 (N8-CUL5–RBX2) (lane 5) or free NEDD8 (lane 7). Reaction products were resolved on SDS–PAGE and visualized by SimplyBlue. (B) SimplyBlue-stained gel showing time course of UBCH7∼Ub discharge by TRIAD1 in the absence or presence of either 30 nM CUL5–RBX2 or 30 nM N8-CUL5–RBX2. (C) UBCH7∼Ub discharge reactions involving 0.25 μM of the indicated TRIAD1 variants were performed in the presence of 30 nM N8-CUL5–RBX2 and reaction products were analysed as described in (A). (D) Ubiquitylation reactions with UBCH7 and 0.25 μM TRIAD1 were performed in the absence (mock, top panel) or presence of either 30 nM CUL5–RBX2 (middle panel) or 30 nM neddylated CUL5–RBX2 (bottom panel) and measured at indicated time points with anti-ubiquitin antibody. (E) Quantification of ubiquitin conjugation as shown in (D) was obtained from chemiluminescent immunoblots using ImageJ software analysis. Standard errors of the mean are given from two independent experiments. (F) Ubiquitylation reactions with UBCH7 and 0.3 μM TRIAD1 were performed in the absence (mock) or presence of 100 nM N8-CUL5–RBX2 for the indicated times. TRIAD1 auto-ubiquitylation and CUL5 ubiquitylation were analysed by immunoblot analyses. (G) Ubiquitylation reactions with UBCH7 and a fixed concentration of 100 nM N8-CUL5–RBX2 were performed in the absence (mock) or presence of 0.3 μM TRIAD1 for the indicated times. TRIAD1 auto-ubiquitylation and CUL5 ubiquitylation were analysed by immunoblot analysis. Ub_x-TRIAD1, ubiquitylated TRIAD1; Ub_n-CUL5, ubiquitylated CUL5.

Figure 5

HHARI binds neddylated cullin-1 complexes. (A) HEK293 cell lines stably expressing GFP or GFP-HHARI were either mock (−) treated or treated overnight with 1 μM MLN4924 (+) before immunoprecipitation with anti-GFP antibodies and subsequent immunoblotting with the indicated antibodies. (B, C) Stably expressed GFP-tagged WT, N-terminally truncated (aa 101–557) and mutated HHARI (as indicated) were immunoprecipitated from HEK293 cells with anti-GFP antibodies and immunoblotted to detect CUL1 and UBCH7 binding. VIW/AAA refers to a V123A/I124A/W140A mutant. (D) Gel filtration chromatography analysis of recombinant HHARI mixed either with non-neddylated (left graph) or with neddylated CUL1–RBX1 (right graph).

Figure 6

Neddylated CUL1–RBX1 activates auto-inhibited HHARI. (A, B) Ubiquitylation reactions were carried out ±ATP with UBCH7 and non-conjugatable lysine-less ubiquitin in the absence or presence of HHARI, N8-CUL1–RBX1, N8-CUL1^CTD–RBX1, and NEDD8 as indicated. Reaction products were visualized by SYPRO Ruby staining. (C) Ubiquitylation reactions were carried out as in (A) and (B) and reaction products were subjected to immunoblot analysis with HHARI antibody. (D) SYPRO Ruby-stained gel showing time-dependent covalent modification of HHARI with Ub-vinyl methyl ester (Ub-VME) in the absence or presence of CUL1–RBX1, N8-CUL1–RBX1, and NEDD8. (E) Covalent modification of HHARI (ΔAri) with Ub-VME in the absence (−) or presence (+) of either isolated wild-type or mutated (F430A, E431A, E503A) Ariadne domain over indicated time. Reaction products were visualized by SYPRO Ruby staining. (F) SYPRO Ruby-stained gel showing time-dependent formation of TRIAD1∼Ub in the absence or presence of CUL5–RBX2, N8-CUL5–RBX2, and NEDD8. *Unspecific bands.

Figure 7

Dominant-negative mutants of TRIAD1 and HHARI impact on cullin neddylation in vivo. (A, B) GFP, or GFP-tagged WT or mutant TRIAD1 containing the indicated amino-acid substitutions was stably expressed in HEK293 cells and immunoprecipitated using anti-GFP agarose. (C) HEK293 cell lines expressing GFP, GFP-TRIAD1, or RING2-mutated GFP-TRIAD1(C310S) were treated with 3 μM MLN4924 inhibitor over the indicated time periods and subjected to immunoblot analyses. (D) GFP-immunoprecipitations were performed from HEK293 cell lines stably expressing GFP, GFP-HHARI, or RING2-mutated GFP-HHARI(C357S). Samples were analysed for cullin neddylation by immunoblotting. The ratios of neddylated to unmodified cullins were obtained from these autoradiograms using ImageJ analysis software. (E, F) HEK293 cell lines expressing GFP, GFP-HHARI, or GFP-HHARI(C357S) were treated with 0.3 μM MLN4924 for the indicated time periods and subjected to immunoblot analyses. (G) HEK293 cells expressing GFP-HHARI (wild-type or C357S mutant) were treated with thymidine or nocodazole, or left untreated (asynchronous) and subjected to immunoblot analyses. (H) The indicated GFP-HHARI constructs were expressed in HEK293 cells and subjected to immunoprecipitation with anti-GFP antibodies before immunoblot analysis with the indicated antibodies.

Figure 8

Model for CRL-induced activation of Ariadne RBR ligases. (A) Auto-inhibited Ariadne RBR ligase based on the HHARI structure (Duda et al, 2013). (B) Illustration of the possible activation of the Ariadne RBR ligase by neddylated cullin as described in Discussion. Orange ball, catalytic Cys; AC, acidic domain; UBA-like, ubiquitin-association (UBA)-like domain.