Multisite phosphorylation dictates selective E2-E3 pairing as revealed by Ubc8/UBE2H-GID/CTLH assemblies

Abstract

Ubiquitylation is catalyzed by coordinated actions of E3 and E2 enzymes. Molecular principles governing many important E3-E2 partnerships remain unknown, including those for RING-family GID/CTLH E3 ubiquitin ligases and their dedicated E2, Ubc8/UBE2H (yeast/human nomenclature). GID/CTLH-Ubc8/UBE2H-mediated ubiquitylation regulates biological processes ranging from yeast metabolic signaling to human development. Here, cryoelectron microscopy (cryo-EM), biochemistry, and cell biology reveal this exquisitely specific E3-E2 pairing through an unconventional catalytic assembly and auxiliary interactions 70-100 Å away, mediated by E2 multisite phosphorylation. Rather than dynamic polyelectrostatic interactions reported for other ubiquitylation complexes, multiple Ubc8/UBE2H phosphorylation sites within acidic CK2-targeted sequences specifically anchor the E2 C termini to E3 basic patches. Positions of phospho-dependent interactions relative to the catalytic domains correlate across evolution. Overall, our data show that phosphorylation-dependent multivalency establishes a specific E3-E2 partnership, is antagonistic with dephosphorylation, rigidifies the catalytic centers within a flexing GID E3-substrate assembly, and facilitates substrate collision with ubiquitylation active sites.

Keywords: CK2; CTLH complex; E2 ubiquitin conjugating enzyme; E3 ubiquitin ligase; GID complex; UBE2H; Ubc8; cryo-EM; phosphorylation; ubiquitin.

Figure 1:. Dedicated GID-Ubc8 E3-E2 pair is configured for channeling a flexibly-tethered folded substrate

A. Overlay of representative cryo-EM maps (right, class I and II in Figure S2A) illustrating flexing of Chelator-GID^SR4 and repositioning of the centrally-captured Fbp1 substrate (PDB 7NS5). Close-up (left) depicts varying positions of Ubc8~ubiquitin active sites (stars) and preferential Fbp1 target lysines (shown as spheres). Models fit into class I and II are, respectively, colored grey or color-coded (Gid2, sky blue; Gid9, navy; Ubc8, orange; ubiquitin, yellow). B. Cartoon representing substrate channeling permitted by: (1) spring-like helical connections between substrate receptor-scaffolding (SRS), catalytic (Cat) and supramolecular assembly (SA) modules, (2) flexible tethering of folded Fbp1 domains displaying target lysines with two N-terminal degrons (black arrows) anchored to two opposing substrate receptor Gid4 molecules (red), (3) constrained configuration of the catalytic module. C. Bipartite interactions between Ubc8~ubiquitin and Gid2-Gid9 catalytic module revealed in a 5-Å-resolution focused-refined map (around regions indicated with a black dashed line in (A)). The map was docked with models of Cat and SRS modules (Gid1-Gid5-Gid8 in grey, Gid4 in red), and Ubc8~ubiquitin. A meandering electron density around Gid2-Gid9 coiled-coil was assigned as Ubc8 C-terminal extension (orange dashed line). See also Figures S1 and S2, Table S1, and Video S1.

Figure 2:. Molecular determinants of Ubc8~ubiquitin-Gid2-Gid9 catalytic assembly

A. Segmented 3.5-Å-resolution focused refined map of Ubc8~ubiquitin-bound Gid2-Gid9, sharpened with DeepEMhancer (left). The corresponding model (right) illustrates conserved and unique E3-E2~ubiquitin interactions mediating recruitment of ubiquitin-conjugated Ubc8’s UBC domain (top) and its allosteric activation (bottom). The bowl-shaped cavity cradling the UBC domain is marked with a black dashed line. B. Probing conserved Ubc8’s UBC-Gid2 RING interactions with in vitro ubiquitylation assay of a fluorescently labeled model peptide substrate (pep*, harboring N-terminal Mdh2-degron connected to a single target lysine placed at position 27 via a flexible linker). C. In vitro ubiquitylation assay as in (B) but with Gid2 and Gid9 mutants of residues recruiting and activating the Ubc8~ubiquitin conjugate (LP – ‘linchpin’ residue, E2B – canonical E2-binding interface, NRP – non-RING priming element). D. Focused refined map as in (A) overlayed with low-resolution electron density of Ubc8 C-terminal extension (transparent) extracted from the map shown in Figure 1C. The trajectory of Ubc8 C-terminus (indicated with an orange dashed line) complements the positively-charged patch within the GID E3 catalytic module (electrostatic potential surface, right). E. Close-up of high-resolution electron density corresponding to a distal portion of Ubc8 C-terminal extension (orange, transparent) that enabled building coordinates for its 11 residues, including sidechains of the C-terminal-most phosphoserines (pSer) 202 and 207 (orange sticks). See also Figures S1 and S2, Table S1.

Figure 3:. Multisite phosphorylated Ubc8/UBE2H C-terminal extension mediates functional partnership with GID/CTLH E3

A. C-terminal extensions of yeast Ubc8 and human UBE2H possess numerous acidic residues (bold black) and putative phosphorylation sites (bold red), flanked by irregularly distributed hydrophobics (bold brown). B. Intact mass spectrometry analyses revealing multisite phosphorylation of insect cell-expressed (IC) C-terminally Twin-Strep-tagged yeast Ubc8 (left) and human UBE2H (right) as compared to their unphosphorylated lambda phosphatase-treated (λPP) and bacterially expressed (BC) versions. 40-Da-difference between the BC and λPP-treated IC E2s stems from N-terminal acetylation of the latter. C. Coomassie-stained Phos-tag SDS-PAGE gels examining phosphorylation status of point mutants and truncations (indicated in (A)) of IC Ubc8 (top) and UBE2H (bottom) C-terminal extensions (CTE). D. Assessing phosphorylation of ectopically-expressed WT and mutant 3xFLAG-UBE2H in HEK293 cells. UBE2H in Phos-tag and corresponding SDS-PAGE gels was detected by anti-FLAG immunoblotting. E. Qualitative binding test examining capacity of different versions of C-terminally Twin-Strep-tagged Ubc8 (Ubc8–2xS) or UBE2H (UBE2H-2xS) to co-purify their recombinant untagged cognate E3s from insect cell lysates. Strep-Tactin pull-down fractions were examined with Coomassie-stained SDS-PAGE. F. Determining binding affinity (equilibrium dissociation constant, K_D) of UBE2H for the CTLH E3 with Octet BioLayer Interferometry (BLI). The steady-state responses upon E3 binding to GST-UBE2H (right) were determined based on reference-subtracted BLI sensorgrams shown left and in Figure S4A. The unphosphorylated BC and phosphomimetic S>D UBE2H variants bound too weakly to accurately estimate K_Ds. SD (n≥3). G. Quantitative mass-spectrometry analysis of proteins immunoprecipitated with ectopically-expressed WT and CTE S>A mutant 3xFLAG-UBE2H from HEK293 cells. Volcano plot visualizes the -log10 p-value and log2 fold change between UBE2H variants. Dotted line represents 5% q-value significance cutoff. Q-value was calculated using the Benjamini-Hochberg method for multiple testing correction. CTLH subunits are outlined with a grey oval and color-coded as in (E). H. In vitro assays probing the roles of phosphorylated Ubc8/UBE2H C-terminal extension for ubiquitylation of the fluorescent model peptide substrates (pep*) designed as described in Figure 2B and harboring either Mdh2 degron or human GID4 recognition sequence PGLWRS at their N-termini for reactions mediated by GID^SR4 (left) and CTLH^SR4 (right), respectively. I. Kinetic parameters of substrate ubiquitylation mediated by GID^SR4 and phosphorylated (IC), unphosphorylated (BC) or phosphomimetic (BC, S,T>D) Ubc8, estimated based on plots shown in Figure S4. The corresponding fold changes between catalytic efficiencies (k_obs/K_m) are calculated relative to BC Ubc8. SD, n=3. J. In vivo glucose-induced degradation of exogenously expressed Fbp1–3xFLAG (quantified as a fraction of substrate remaining at different time points after switch from glucose-deplete to glucose-rich medium, normalized to the level of DHFR) in WT and Ubc8 mutant yeast strains. Error bars represent SD (n=3), points indicate the mean. See also Figures S3 and S4.

Figure 4:. Multiple phosphorylation sites potentiate Ubc8 and UBE2H activity

A. Sequence of Ubc8 C-terminal extension indicating three subsets of phosphosites (colored brown, green and magenta) and positions of truncations used for mutational analysis. The structurally-resolved portion of the C-terminal extension is marked with an orange box. B. Qualitative ubiquitylation assays of fluorescent model peptide (pep*) (as in Figure 2B) examining importance of individual phosphosite subsets within IC Ubc8 C-terminal extension (color-coded as in (A)) and effects of its truncations. C. Quantitative kinetic analysis of IC Ubc8 phosphosite clusters mutants. Fitting to the plots showing the fraction of ubiquitylated Mdh2 as a function of Ubc8 concentration (left) or the time-course of Mdh2 ubiquitylation (right) represented as the fraction of remaining unmodified substrate (S₀) yielded values of K_m and k_obs, respectively (bottom table, SD, n=3). The fold change of catalytic efficiencies (k_obs/K_m) was calculated relative to WT IC Ubc8. Full kinetic plots and representative scans of SDS-PAGE gels are shown in Figure S4. D. Effects of mutating subsets of Ubc8 phosphosites on in vivo degradation of yeast Fbp1–3xFLAG after 120 min of glucose recovery (relative to timepoint 0), quantified using the promoter reference technique. Error bars represent SD (n≥3). E. Close-up of Ubc8~ubiquitin-Gid2-Gid9 model highlighting the constellation of Gid2 basic patch residues (shown as sticks) interacting with the structurally visualized portion of Ubc8 C-terminal extension (left). Mutating these Gid2 residues impeded ubiquitylation of fluorescent model peptide substrate (pep*) in reactions with WT but not C-terminally truncated (Δ197–218) IC Ubc8 (right). F. Impact of mutating Gid2 basic patch (shown in (E)) on in vivo degradation of Fbp13xFLAG after 120 minutes of glucose recovery, assayed with the promoter reference technique. Error bars represent SD, n=3. G. In vitro assays (bottom) testing effects of mutating individual UBE2H phosphoserine subsets and progressive truncations of its C-terminal extension (indicated in the amino acid sequence, top) on CTLH-mediated ubiquitylation of fluorescently labeled model peptide (pep*, as in Figure 3G). H. Assessing binding of WT ectopically expressed 3xFLAG-UBE2H and its C-terminal extension mutants to endogenous CTLH. Immunoblots detect the core (RANBP9) and catalytic (MAEA) CTLH subunits as well as various 3xFLAG-UBE2H versions in the input HEK293 lysate and samples after FLAG IP. See also Figure S4.

Figure 5:. CK2 phosphorylates C-terminal extensions of Ubc8 and UBE2H

A. Acidophilic kinase (CK1, CK2 and GSK3) recognition motifs (red) and phosphorylation sites (black). B. Phos-tag SDS-PAGE gels assessing phosphorylation status of fluorescently labeled (*) BC Ubc8 (top) and UBE2H (bottom) after their incubation with ATP/MgCl₂-supplemented yeast and HEK293 lysates, respectively, with or without inhibitors of CK1, CK2 and GSK3. C-terminally truncated versions of BC Ubc8 (Δ157–218, ΔCTE) and UBE2H (Δ162–182, ΔCTE) lacking phosphorylation sites were included as negative controls. C. A Phos-tag mobility shift assay as in (B) but examining capacity of recombinant CK2 to phosphorylate the C-terminal extensions of BC Ubc8* (top) and UBE2H* (bottom) in vitro. D. Intact mass spectrometry examining phosphorylation status of CK2-treated BC Ubc86xHis and UBE2H. Numbers of conjugated phosphate groups are shown in red. E. In vivo assay testing the effect of individual deletions of two CK2α isoforms (encoded by CKA1 and CKA2 genes) on glucose-induced degradation of exogenous Fbp1–3xFLAG in yeast, assayed with the promoter-reference technique. Error bars represent SD (n=3), points indicate the mean. See also Figures S4 and S5.

Figure 6:. Multisite phosphorylated C-terminal extension of UBE2H stably engages the CTLH E3 basic patch

A. Segmented 3.4-Å-resolution focused-refined map of the UBE2H~ubiquitin-bound CTLH catalytic subunits (RMND5A, sky blue; MAEA, navy) sharpened with DeepEMhancer (left). UBE2H C-terminal extension docks into a unique RMND5A-MAEA composite basic patch (black circle) visible in the electrostatic potential surface of the catalytic module (right). B. Close-up of electron density corresponding to the CTLH-bound C-terminal extension of UBE2H (orange, transparent), highlighting coordinates for its all phosphoserines punctuated with acidic and hydrophobic residues (orange sticks). C. Details of interactions between multiphosphorylated UBE2H C-terminal extension and CTLH basic patch (left). RMND5A and MAEA basic residues mutated in (D), (E) and (F) are shown as sticks (right). D. In vitro ubiquitylation assays probing impact of mutating RMND5A-MAEA basic patch residues on ubiquitylation of a fluorescent model peptide substrate (pep*, described in Figure 3G) in reactions with WT IC or BC, and C-terminally deleted (ΔCTE, Δ162–183) IC UBE2H. E. Fluorescent scans of native PAGE gels examining binding of fluorescently labeled (*) UBE2H* and UBE2H*~ubiquitin to CTLH in their fully phosphorylated (phosphor-UBE2H*) and dephosphorylated (pre-treated with λPP) states to WT and basic patch mutant CTLH. F. Assessing susceptibility of phospho-UBE2H* and phospho-UBE2H*~ubiquitin to λPP-mediated dephosphorylation in the presence and absence of CTLH E3. Dephosphorylation reactions were quenched at indicated time points and resolved by Phos-tag SDS-PAGE gels. See also Figures S6 and S7, Table S1.

Figure 7:. Evolutionary conservation of phosphorylation-mediated GID/CTLH-Ubc8/UBE2H E3-E2 pairing

A. Catalytic modules of GID/CTLH E3 orthologs represented as their electrostatic potential surfaces and displayed as a phylogenetic tree. The models were generated with AlphaFold2 (for K. marxianus, D. melanogaster, D. rerio and X. laevis) or experimentally determined (for S. cerevisiae and H. sapiens, respectively). Black ovals indicate positions of basic patches. B. Alignment of Ubc8/UBE2H C-terminal extension sequences. Putative phosphorylation sites (red), acidic (black) and hydrophobic (brown) residues are bold and colored. Patterns of residues flanking the distal phosphosites are indicated above sequences (ϕ denotes hydrophobic residue).