Analysis of ubiquitin recognition by the HECT ligase E6AP provides insight into its linkage specificity

Abstract

Deregulation of the HECT-type ubiquitin ligase E6AP (UBE3A) is implicated in human papilloma virus-induced cervical tumorigenesis and several neurodevelopmental disorders. Yet the structural underpinnings of activity and specificity in this crucial ligase are incompletely understood. Here, we unravel the determinants of ubiquitin recognition by the catalytic domain of E6AP and assign them to particular steps in the catalytic cycle. We identify a functionally critical interface that is specifically required during the initial formation of a thioester-linked intermediate between the C terminus of ubiquitin and the ligase-active site. This interface resembles the one utilized by NEDD4-type enzymes, indicating that it is widely conserved across HECT ligases, independent of their linkage specificities. Moreover, we uncover surface regions in ubiquitin and E6AP, both in the N- and C-terminal portions of the catalytic domain, that are important for the subsequent reaction step of isopeptide bond formation between two ubiquitin molecules. We decipher key elements of linkage specificity, including the C-terminal tail of E6AP and a hydrophilic surface region of ubiquitin in proximity to the acceptor site Lys-48. Intriguingly, mutation of Glu-51, a single residue within this region, permits formation of alternative chain types, thus pointing to a key role of ubiquitin in conferring linkage specificity to E6AP. We speculate that substrate-assisted catalysis, as described previously for certain RING-associated ubiquitin-conjugating enzymes, constitutes a common principle during linkage-specific ubiquitin chain assembly by diverse classes of ubiquitination enzymes, including HECT ligases.

Keywords: enzyme mechanism; post-translational modification; ubiquitin; ubiquitin ligase; ubiquitylation (ubiquitination).

Figure 1.

C-lobe of E6AP interacts with ubiquitin in trans. A, weighted and combined chemical shift perturbations, Δδ(¹H¹⁵N), of E6AP C-lobe resonances induced by a 12.5-fold molar excess of ubiquitin, plotted over the E6AP residue number. Resonances that undergo line broadening (Lys-801 and Thr-819) are marked by an asterisk. B, weighted, combined chemical shift perturbations of ubiquitin resonances induced by a 12.5-fold molar excess of the E6AP C-lobe plotted over the ubiquitin residue number. C, structures of the E6AP C-lobe (extracted from PDB code 1C4Z (30)) and ubiquitin (PDB code 1UBQ (94)) are shown in ribbon representation. The nitrogen atoms of backbone amide groups whose resonances display binding-induced shift perturbations, Δδ(¹H¹⁵N) > 0.04, or undergo line broadening (Lys-801 and Thr-819 of E6AP) are highlighted as balls (magenta). The side chain of the catalytic cysteine, Cys-820, is also displayed. D, determination of an apparent dissociation constant, K_D, for the C-lobe–ubiquitin interaction based on FP measurements. The mean FP signal and standard deviations from three independent experiments using a fluorophore-labeled ubiquitin variant were plotted as a function of the C-lobe concentration and fitted to a single-site binding model (line).

Figure 2.

Most of the residues identified by NMR are important for thioester transfer of ubiquitin from UBE2L3 to E6AP. A, thioester transfer of ubiquitin from the E2 (UBE2L3) to the E6AP HECT domain, followed in single-turnover, pulse-chase assays at three time points, as indicated, and monitored by nonreducing SDS-PAGE and anti-ubiquitin Western blotting. The thioester-linked HECT domain–ubiquitin conjugate (E6AP∼Ub) and, in some cases, the thioester-linked E2-ubiquitin precursor (UBE2L3∼Ub) are visible. The input amount of HECT domain (E6AP) is monitored by reducing SDS-PAGE and Coomassie staining. Note that no auto-ubiquitination of the HECT domain occurs within the tested time range. B, analogous assays as in A, monitoring the effect of mutations in ubiquitin on thioester transfer to the E6AP HECT domain. For the I36A and L71A variants, silver staining (right) was used in lieu of anti-ubiquitin Western blotting (left), because these variants are not detected well by the antibody (P4D1) used here.

Figure 3.

Thioester transfer of ubiquitin requires similar residues in E6AP and NEDD4-type ligases; alternative residues are involved in isopeptide bond formation. A, thioester transfer of ubiquitin from the E2 (UBE2L3) to the E6AP HECT domain interrogating additional variants of ubiquitin (Q40A) and E6AP (L814A and A842I). The mutation sites are homologous to critical residues in the donor interface of NEDD4-type ligases (18, 31, 40, 41). B, isopeptide bond formation assays comparing the activities of the E6AP HECT domain WT and those variants tested in Fig. 2A that do not display defects in thioester formation. Activities are monitored at three time points, as indicated, by reducing SDS-PAGE and Western blotting against E6AP (HECT domain auto-ubiquitination marked as E6AP_Ub) and ubiquitin (di-ubiquitin reaction product marked as Ub₂), respectively. Time point 0 denotes samples before ATP addition. The amounts of Ub₂ and E6AP_Ub were quantified and normalized to the input amount of E6AP. Quantifications are based on three independent experiments; the means and standard deviations were plotted for the 60-min time point.

Figure 4.

Model of a NEDD4-type complex between the C-lobe of E6AP and ubiquitin, as required for thioester formation. Ubiquitin was modeled by structural superposition of the C-lobe of the ubiquitin-bound HECT domain of NEDD4 (PDB code 4BBN, chain A (18)) with the C-lobe of the HECT domain of E6AP (chain A, extracted from PDB code 1C4Z (30)), using the PyMOL Molecular Graphics System, Version 2.0, Schrödinger, LLC. Ubiquitin and the E6AP C-lobe are displayed in ribbon representation; the side chains of residues relevant for thioester formation are displayed as balls and sticks. The side chain of the catalytic cysteine, Cys-820, is also displayed. The backbone nitrogen atoms of additional residues that experience perturbations upon ubiquitin addition are shown as spheres.

Figure 5.

C-terminal tail of E6AP interacts with the donor ubiquitin upon thioester formation. A, schematic of the protein samples used to monitor the interactions of the ¹⁵N-enriched C-lobe of E6AP and ubiquitin in trans or in the context of a disulfide-linked complex containing ubiquitin G76C (left). Weighted and combined chemical shift perturbations, Δδ(¹H¹⁵N) of C-lobe resonances, were induced by ubiquitin in trans (12.5-fold molar excess) or cis plotted over the E6AP residue number (right). The functionally important C-tail region is marked. B, sections of the ¹H-¹⁵N BEST-TROSY spectra (as analyzed in A) featuring resonances that originate from the C-tail. Each section shows a superposition of spectra in the presence (blue and rose, respectively) and absence (black) of ubiquitin.

Figure 6.

C-terminal tail of E6AP is essential for ubiquitin chain elongation. Isopeptide bond formation assays compare the activities of the E6AP HECT domain WT and C-terminal tail variants. Activities are monitored at three time points, as indicated, by reducing SDS-PAGE and Western blotting against E6AP (HECT domain auto-ubiquitination marked as E6AP_Ub) and ubiquitin (di-ubiquitin reaction product marked as Ub₂), respectively. Time point 0 denotes samples before ATP addition. The amounts of Ub₂ and unmodified E6AP were quantified and normalized to the input amount of E6AP. In addition, the contributions of mono-ubiquitination (1 ubiquitin moiety) and chain formation/multimono-ubiquitination (≥2 ubiquitin moieties) were quantified for each variant individually (with the total amount set to 1). Quantifications are based on three independent experiments; the means and standard deviations were plotted for the 60-min time point.

Figure 7.

C-terminal tail of E6AP directs Lys-48 specificity of ubiquitin linkage formation. A–C, assays monitoring the activities of the E6AP HECT domain WT and C-terminal tail variants toward ubiquitin WT, single Lys-to-Arg variants, or a lysine-free (K0) variant, respectively, by reducing SDS-PAGE and Western blotting against ubiquitin (di-ubiquitin reaction product marked as Ub₂). Assays were conducted for 15 min; the input amount of HECT domain (E6AP) in the absence of ATP is monitored by reducing SDS-PAGE and Coomassie staining (left). The amount of Ub₂ was quantified, and the ratio of the amounts of mutated Ub₂ to WT Ub₂ was plotted. Quantifications are based on three independent experiments; the means and standard deviations were plotted for the 60-min time point (right).

Figure 8.

Quantification of E6AP-catalyzed ubiquitin linkage types upon mutation of the C-terminal tail of the ligase. AQUA mass spectrometric analysis of Ub₂ species formed by the E6AP HECT domain WT (A), F849A (B), and Δ4 (C) variants. Results were normalized to the total amount of ubiquitin for each linkage type detected; the means and standard deviations from three replicates are shown (in %). The corresponding data are provided in Fig. S4.

Figure 9.

Distinct surface regions are required by the acceptor and donor ubiquitin during E6AP-catalyzed ubiquitin linkage formation. A, structure of ubiquitin (PDB code 1UBQ (94)) is shown in ribbon and surface representation. The area of the donor ubiquitin, including the C-terminal tail, that contacts the HECT C-lobe in the canonical mode is colored rose; hydrophobic residues important for E6AP activity are in green (49); and residues in proximity of Lys-48, as studied in B–D, are in purple. B, quantification of the isopeptide bond formation activity of the E6AP HECT domain toward ubiquitin variants, based on three independent experiments (for experimental data, see Fig. S5). For free chain formation (left) and auto-ubiquitination (right), the amount of reaction product (Ub₂ and E6AP_Ub, respectively) at 60 min was normalized to that of input E6AP. C, schematic of HECT ligase-mediated linkage formation between an enzyme-linked donor and an acceptor ubiquitin (left), and the two ubiquitin substrates employed in the ΔGG assay: WT ubiquitin and His₆-tagged, truncated ubiquitin (residues 1–74; Ub^ΔGG) (middle); validation of the assay using the E6AP HECT domain and equimolar mixtures of the indicated ubiquitin substrates. The Ub₂ reaction products were analyzed by reducing SDS-PAGE and Coomassie staining. The input amount of E6AP in the absence of ATP serves as a control (right). D and E, ΔGG assay with ubiquitin variants, performed and analyzed as in C. The amount of both Ub₂ products combined were quantified and normalized to the amount of input E6AP, and the means and standard deviations from three independent experiments were plotted (right).

Figure 10.

Glu-51 of ubiquitin is a critical determinant of the Lys-48 linkage specificity of E6AP. A, comparison of the activity of the E6AP HECT domain toward ubiquitin WT and the E51A variant, respectively, monitored by reducing SDS-PAGE and Western blotting. Two different anti-ubiquitin antibodies were used: P4D1 to monitor all ubiquitin modifications, and D9D5 for Lys-48 linkages. The input amount of E6AP is monitored by reducing SDS-PAGE and Western blotting against E6AP (left). The amounts of Ub₂ assembled from the WT and E51A ubiquitin variants were quantified after 30 min, and the mean ratios and standard deviations from three independent experiments were plotted (right). B, amounts of Ub₂ species linked through individual lysine residues from reactions supplied with ubiquitin E51A, measured by AQUA MS. The values reflect the means and standard deviations from three biological replicates. Note that Lys-48 linkages could not be quantified reliably in this setup, due to the E51A mutation impacting peptide ionization. The corresponding data are provided in Fig. S7. For the corresponding data on WT ubiquitin, see Fig. 8A and Fig. S4.

Figure 11.

The N-lobe of E6AP interacts with ubiquitin, and residues in the exosite of this ligase are important for isopeptide bond formation. A, FP-based determination of an apparent dissociation constant, K_D, for the interaction of ubiquitin with the E6AP N-lobe and the HECT domain, respectively. The mean FP signal and standard deviations from three independent experiments were fitted to a single-site binding model (line). B, structural superposition of the E6AP N-lobe (extracted from PDB code 1C4Z (30)) with the ubiquitin-bound N-lobe of RSP5 (extracted from PDB code 3OLM (36)). Protein backbones are shown as ribbons and mutated residues as balls and sticks. The C-lobes are not displayed for clarity. C, K_D determination analogous to A, using exosite variants of the E6AP HECT domain. D, isopeptide bond formation assays comparing the activities of E6AP HECT domain variants. Activities are monitored at three time points, as indicated, by SDS-PAGE and Western blotting against E6AP (HECT domain auto-ubiquitination marked as E6AP_Ub) and ubiquitin (di-ubiquitin reaction product marked as Ub₂), respectively. Time point 0 denotes samples before ATP addition. The amounts of Ub₂ and E6AP_Ub at 60 min were quantified and normalized to the E6AP input, and the means and standard deviations from three independent experiments were plotted. E, thioester transfer of ubiquitin from the E2 (UBE2L3) to the E6AP HECT domain, followed in single-turnover, pulse-chase assays at three time points, as indicated, and monitored by reducing SDS-PAGE and anti-ubiquitin Western blotting. The input amount of E6AP is monitored by reducing SDS-PAGE and Coomassie staining. Note that no auto-ubiquitination of the HECT domain occurs within the tested time range.

Figure 12.

Schematic view summarizing the identified interactions and surface patches critical for Lys-48–linked ubiquitin chain formation by the HECT domain of E6AP. Our studies suggest that the C-lobe of E6AP utilizes canonical (NEDD4-type) contacts with the donor ubiquitin (rose) during thioester transfer of ubiquitin from the E2 to the E3. The C-terminal tail (C-tail) of E6AP is not required for this step (A). E6AP-mediated isopeptide bond formation between the thioester-linked donor–E6AP complex and an acceptor ubiquitin relies on surface regions that are distinct from those required during thioester transfer (B). The hydrophobic patch (green) is used by the donor ubiquitin for yet uncharacterized interactions with the N-lobe, the C-lobe, or the acceptor ubiquitin. The acceptor ubiquitin is critically dependent on a hydrophilic area (purple) around the acceptor residue (Lys-48), including Glu-51. The C-tail contacts the thioester-linked donor and confers linkage specificity in isopeptide bond formation, possibly by additional contacts with the N-lobe or the acceptor ubiquitin. Furthermore, we demonstrate that E6AP interacts with ubiquitin through the N-lobe. Mutations of residues in the exosite region (red) weaken this interaction and reduce isopeptide bond formation activity; however, they do not affect thioester formation. Whether the N-lobe–ubiquitin interaction resembles the exosite-mediated binding mode seen in NEDD4-type ligases awaits structural elucidation. It also remains unclear which functional ubiquitin moiety this interaction involves (e.g. a regulatory moiety or, possibly, the acceptor).