Structure of E6AP in complex with HPV16-E6 and p53 reveals a novel ordered domain important for E3 ligase activation

Summary

High-risk human papillomavirus E6 oncoprotein is a model system for the recognition and degradation of cellular p53 tumor suppressor protein. There remains a gap in the understanding of the ubiquitin transfer reaction, including placement of the E6AP catalytic HECT domain of the ligase concerning the p53 substrate and how E6 itself is protected from ubiquitination. We determined the cryo-EM structure of the E6AP/E6/p53 complex, related the structure to in vivo modeling of the tri-molecular complex, and identified structural interactions associated with activation of the ubiquitin ligase function. The structure reveals that the N-terminal ordered domain (NOD) in E6AP has a terminal alpha helix that mediates the interaction of the NOD with the HECT domain of E6AP and protects the HPV-E6 protein from ubiquitination. In addition, this NOD helix is required for E6AP ligase function by contributing to the affinity of the E6-E6AP association, modulating E6 substrate recognition, while displacing UbcH7.

eTOC blurb:

Kenny et al. characterized the complex structure of full-length E6AP with E6 and p53, a prerequisite for p53 ubiquitination by E6AP. This structure highlights an N-terminal ordered domain (NOD) of E6AP. Additional interactions between NOD and E6 mediate the E6AP activation mechanism by E6, explaining the effective degradation of p53.

Graphical Abstract

Introduction

The hierarchical enzymatic cascade comprising E1 (Ub-activating enzyme), E2 (Ub-conjugating enzyme), and E3 (Ub-ligating enzyme) results in the covalent tagging of host proteins with ubiquitin (Ub) with protean consequences, including modification of cell cycle, proteasomal degradation, DNA repair, mitophagy, and various signaling pathways^1–4. The human genome encodes 2 E1s, around 40 E2s, and more than 600 E3 ligases^5–7. Since E3 ligases mainly mediate the recognition of the substrates targeted for ubiquitination, they often play key regulatory roles^8,9. Concomitantly, deregulation of ubiquitination, either by targeting the substrates or the conjugation machinery itself, is often implicated in the development of many pathophysiological conditions such as cancer, neurodegeneration, and viral infections including those mediated by the human papillomaviruses (HPVs)^5–7. The high-risk HPVs are etiological agents of cervical cancer, other anogenital cancers, and a growing proportion of head and neck cancers^10–13. Oncogenic transformation of cells by this mucosal virus relies on two viral proteins, E6 and E7, which compromise different host tumor suppressor pathways¹⁴. Specifically, E6 oncoprotein hijacks a cellular ubiquitin ligase termed E6-associated protein (E6AP, product of the UBE3A gene) and promotes degradation of p53 and additional cellular proteins by the ubiquitin-proteasome system. It facilitates evasion of the host antiviral response by recruiting E6AP to polyubiquitinate p53 and abrogate the latter’s apoptotic activity^15,16.

E6AP, a multi-modular protein, is the prototypical HECT E3 ligase (Homologous to E6AP Carboxyl Terminus) that accepts Ub from its cognate E2-conjugating enzyme, UbcH7. An intrinsic activity relayed by the catalytic cysteine (Cys843^E6AP) facilitates the formation of Ub-conjugated thioester immediately before the transfer of the Ub cargo onto its substrates^17–23. The structure of the catalytic domain of E6AP (HECT domain) in complex with UbcH7²⁴, solved over two decades ago, showed that the HECT domain itself has an amino-terminal lobe separated by a short linker from a carboxy-terminal catalytic lobe. The E2 enzyme (UbcH7) that ubiquitinates the HECT domain of E6AP binds to the N-lobe of the HECT domain in an orientation that makes it possible for the flexible C-lobe of the HECT domain to participate in the Ub hand-off. A previous structural study characterized the interaction of p53 and E6 (from this point on we refer to HPV16-E6 as E6 unless otherwise noted) in the presence of an E6AP peptide containing the leucine-rich LxxLL motif to reveal fundamental insights into how the three proteins may associate²⁵. However, by reducing the nearly 900 amino acid E6AP ligase to a 15-residue peptide, the structure could not explain how or if different additional binding regions in E6AP, in addition to the LxxLL motif, might interact with E6 and p53 in the ternary complex. Evidence for additional interactions in addition to LxxLL-E6 was observed in yeast two-hybrid studies by Drews and Vande Pol^26–28 as well as studies based on cross-inked peptides between E6 and E6AP by mass spectrometry²⁹. These studies identified previously undescribed regions from both the N- and C-terminal regions of E6AP that contribute to its interaction with E6 and activation of its ligase activity. In addition, evidence exists for E6AP in association with the proteasome by a high-affinity interaction between the N-terminal AZUL domain of E6AP and hRpn10, a subunit of the 26S proteasome³⁰.

Despite the years of research studying this system, the molecular mechanisms underlying the conformational dynamics that lead to polyubiquitination and proteasomal degradation of p53 remain poorly understood. Another question that remains unanswered is how, if at all, E6 is protected from self-ubiquitination and subsequent degradation. To gain further structural insights into the E6-mediated polyubiquitination of p53 by E6AP, we sought to capture the ternary complex between a functional E6AP protein with E6 and p53 as analyzed by cryo-EM. Our work successfully models the multiple intermolecular interactions of the E6AP/E6/p53 complex, characterizing new E6AP-E6 interaction hotspots in addition to the previously described LxxLL motif of E6AP. We have also identified a novel N-terminal ordered domain (NOD) in E6AP spanning from residues 236 to 515, concluding with a long α-helix before transitioning to the HECT domain. This NOD terminal long helix contributes to the interaction with E6 by making contacts with an E6 helix that connects the zinc-structured domains of E6. The formation of the E6-E6AP complex displaces the ubiquitin-charging E2 ligase UbcH7 from the HECT domain. The 3D reconstruction of this ternary complex, combined with in vitro p53 ubiquitination turnover assays and yeast two-hybrid assays, provides new insights. At the same time, it is inspiring additional questions about the protein dynamics relevant to ubiquitination. During the course of this study, three other studies reporting the complex of E6AP/E6 in complex with and without p53 were published.^31–33 Our cryo-EM structure reiterates the observations made by these studies. The structure along with the results from our yeast hybrid studies provide fresh insights into the interplay between the different components of the ligation machinery and subsequent conformational dynamics that leads to the ubiquitination of p53.

Results

Expression and characterization of an active E6AP/E6/p53 complex.

In agreement with previous reports, we observed that full-length E6 protein precipitated when purified without E6AP³⁴. E6 (1–151, Uniprot ID: A0A384LI46) and E6AP (isoform 2, 1–875, Uniprot ID: Q05086) were co-expressed and co-purified to achieve expression of the soluble protein in bacteria. A p53 construct lacking the tetramerization domain (p53_1–312) for a higher yield in protein expression was used to produce the ternary complex. The ternary complex was formed by adding excess amounts of purified p53 to the E6AP/E6 complex, followed by size-exclusion chromatography (SEC) to remove free p53. This approach was based on a previous report³⁵ showing that the three proteins assemble in a 1:1:1 stoichiometric fashion to produce the heterotrimeric E6AP:E6:p53 complex, suggesting preferential assembly of monomeric p53 into the ternary complex instead of tetramerization. A peak eluting earlier than p53 alone (Fig. 1A) corresponded to the E6AP/E6/p53 complex. When subjected to negative-stain EM, this size-excluded complex showed an apparent uniform distribution of particles (Fig. 1B), encouraging cryo-EM analysis. The vitrified sample allowed us to capture several different projections (Fig. 1C) of the ternary complex, which could be used for 3D reconstruction. This sample resulted in a map of the ternary complex at 3.54 Å global resolution, measured by the gold standard Fourier Shell Correlation (FSC) of 0.143. The local resolution of the map showed that regions of lower resolution are observed near the solvent-exposed regions where the protein components exhibit structural flexibility (Fig. 1D). The C-lobe of the HECT domain where the catalytic Cys843 of E6AP is located also exhibited lower local resolution, likely due to conformational dynamics of that domain compared to the rest of protein (E6AP_236–760).

Figure 1. Reconstruction of the E6AP/E6/p53 ternary complex.

(A) Size-exclusion chromatography purification of the E6AP/E6/p53 complex. The figure also shows an SDS-PAGE image of the fractions indicated in the chromatogram, demonstrating the purity and separation of excess p53 from the ternary complex. (B) Negative stain imaging of the purified ternary complex indicating apparent homogeneity of the sample. (C) Representative 2D projections of the ternary complex calculated from extracted particles. (D) Local resolution map of the ternary complex following refinement of the 3D reconstruction. Regions with the highest resolution are colored blue, while those with the lowest resolutions are in red. (E) The drop-off assay shows E6AP-mediated ubiquitination of p53 in the presence of E6, which is indicated by higher molecular weight species. (F) Ubiquitination of p53 by E6AP were compared without (left panel) and with E6 (right panel). (G) Ubiquitinated lysine residues on p53 were detected using mass spectrometry.

To ensure that the ternary complex was active, the purified complex was tested for its ability to ubiquitinate p53 in combination with the E1 Ub-activating enzyme Uba1, the E2 enzyme UbcH7, and Ub, as shown by a drop-off assay (Fig. 1E). While some level of p53 ubiquitination could be observed in the absence of E6 under the assay conditions, robust ubiquitination of p53 was detected as a high molecular-weight species in the presence of E6, confirming E6’s role in facilitating p53 recognition by the E3 ligase. P53 ubiquitination reactions with and without E6 (Fig 1F) were performed to confirm the role of E6 as an E6AP activator. The p53 ubiquitination reaction is significantly accelerated in the presence of E6, and the catalytically inactive E6AP_C843A fails to ubiquitinate p53 even when added at 10-fold excess compared to the WT protein. E6 levels also remain constant in the reaction with E6 added, indicating that E6 is not ubiquitinated and remains protected in this complex. While autoubiquitination of E6AP was also detectable, the E6 protein band intensity was not altered after one hour, indicating protection from ubiquitin modification.³² To confirm the polyubiquitination pattern on p53, the reaction samples were digested by trypsin and subjected to LC-MS/MS. Several ubiquitinated peptides of p53 were identified (Fig. S1), most of which were consistent with a previous report³⁵; the peptides containing Lys101, Lys120, Lys164, and Lys305 of p53 were revealed as the dominant sites of ubiquitination (Fig. 1G).

E6AP has been reported to associate with the proteasomal subunit hRpn10.³⁰ We introduced hRpn10 into our sample to increase the size of the complex for cryo-EM studies. To ensure that hRpn10 binding did not occlude E6 binding, we performed a pulldown experiment, wherein HA-tagged hRpn10 was used to pull down the ternary complex (Fig. S2). We observed that hRpn10 could bind the E6AP/E6/p53 ternary complex, as judged by the co-elution of the three components of the complex from the immunoprecipitate of HA-hRpn10. This observation suggests that the viral factor can hijack E6AP on the proteasome. Despite this observation, the E6AP-hRpn10 association interface (E6AP_AZUL-hRpn10_RAZUL) is in an unresolved region of E6AP in our map and thus is unresolved in our cryo-EM map.

Cryo-EM model of the E6AP/E6/p53 complex.

The structure of the ternary complex was determined using the pipeline described in Figure S3A. AlphaFold 2^36,37 was used to aid in modeling the structures of E6AP and p53 since the full-length structures of these two proteins are yet to be determined experimentally. The predicted models of these two proteins were used to fit into the volume generated from single particle reconstruction. Residues 1–118 and 171–235 in E6AP and residues 1–93 and 293–312 in p53 could not be modeled in the cryo-EM map. These segments are also modeled as loops in their respective AlphaFold models. The fitting of E6AP and p53 onto the cryo-EM map revealed an unoccupied density region between the two proteins. Using the p53 structure already placed in the cryo-EM map as a reference unit, we aligned it with a previously solved crystal structure of p53 DNA binding domain bound to E6 (PDB ID: 4XR8²⁵), which resulted in E6 fitting in the unoccupied volume of the map. This ternary complex was modeled further by feeding the aligned coordinates of E6AP-E6 from the previous step as a template into ColabFold³⁸. This directed prediction led to an ordered LxxLL helix of E6AP sitting between the two lobes of E6, as shown in previous structures²⁵. Overall, the cryo-EM model of E6AP/E6/p53 complex shows the positioning of p53 with respect to the catalytic Cys843 of E6AP and the placement of E6, where the viral protein is hidden behind the ubiquitination face of the complex. Most of E6AP, except some regions towards the N-terminal end of the ligase, were modeled with high confidence (pLDDT > 90, Fig. S4).

The structure of E6AP reveals an ordered domain N-terminus of the HECT domain.

The cryo-EM structure of the ternary complex reveals a novel structured domain of E6AP. We define this region, which stretches from residues 116–518, as the N-terminal Ordered Domain of E6AP (E6AP_NOD). This NOD domain is organized into four distinct regions (Fig. 2A). Region 1 is a stack of 9 α–helices of various lengths spanning from residues 119–170 and 236–386 (Fig. 2A, green). Immediately following this region is a 41-residue-long largely disordered loop (residues 386–426, region 2). A short structured helical segment is located within this disordered region, presenting the LxxLL motif previously recognized as sufficient to mediate E6 association with E6AP (Fig. 2A, blue). Region 3 is a smaller α-helical stack spanning from residues 427–485 (Fig. 2A, lime) that connects the LxxLL bearing loop (region 2) to the fourth region of E6AP_NOD (Fig. 2A, red), a 30-residue long α-helix (E6AP residues 486–515), hereafter referred to as the “NOD long helix”, which bridges the N-terminal domain of E6AP to the beginning of the catalytic HECT domain of E6AP (E6AP_HECT: residues 517–875). The protein backbone of the E6AP_NOD region, excluding the flexible region encompassing the LxxLL helix, fits into the cryo-EM density (Fig. 2B).

Figure 2. N-terminus ordered domain folds independently of the HECT domain.

(A). A novel N-terminal ordered domain is observed in our E6AP structure. The domain can be divided into four regions: helix stack 1 (region 1, green), LxxLL helix loop (region 2, cyan), helix stack 2 (region 3, lime), and NOD long helix (region 4, red). (B) Density fit of regions 1, 3, and 4. Region 2 is not shown due to its flexible nature and weak density. (C) E6AP residues 742–755 bridge the E6AP_HECT and E6AP_NOD domains via electrostatic interactions. The residues participating in electrostatic interactions are presented in a table. (D) A yeast two-hybrid experimental designed to probe for inter-domain interactions. Individual haploid yeast strains expressing the lexA fusion proteins were mated to strains expressing the indicated B42-transactivator fusion proteins. The formation of a complex between the lexA fusion protein and a B42 transactivator fusion protein creates a hybrid transcription factor that activates a lexA-responsive lacZ reporter, resulting in a blue colony on XGAL-containing plates. (E) Yeast two-hybrid study showing association of E6AP_NOD with E6AP_HECT (more on Fig. S5). Individual haploid yeast strains expressing the indicated lexA fusion proteins were dotted horizontally and mated to strains expressing the indicated B42-tranactivator fusion proteins dotted vertically to create diploid strains expressing both lexA and B42 fusion proteins. Protein-protein association between the lexA and B42 fusion proteins is indicated by patches of blue yeast on XGAL plates.

E6AP_NOD sits on top of the E6AP_HECT, interacting with residues 742–755 within the HECT domain (Fig. 2C). This short span of residues connects the regulatory N-lobe and catalytic C-lobe of E6AP_HECT²⁴, while mediating several electrostatic interactions with residues from region 3 of E6AP_NOD. We predict that this N-terminal ordered domain of E6AP is potentially a substrate-binding domain of E6AP. Previous studies have shown that known substrates of E6AP bind outside E6AP_HECT^39,40, requiring amino acid residues upstream of the HECT domain. Furthermore, a previous yeast two-hybrid assay study shows that deletions longer than the first 310 amino acid residues of E6AP compromise the general catalytic activity of E6AP in both yeast and mammalian cells, possibly due to folding defects. These observations suggest that residues from E6AP_NOD could be critical for substrate recruitment or proper folding of the HECT domain.

To validate the potential interaction between the E6AP_NOD and the E6AP_HECT, we performed yeast two-hybrid assays to test the association between these two ordered domains. Truncation constructs of the C-terminus of E6AP were fused with a LexA DNA binding domain (bait constructs), while truncations of the N-terminus of E6AP were fused with a B42 transactivation domain (prey constructs) (Fig. 2D). The results show that the E6AP_NOD and the E6AP HECT domain (residues 518–875) are indeed associated as two independent domains (Fig. 2E, red box). Removal of the interacting regions, specifically regions 3 or 4 (Fig. 2A), shows a loss of association between the E6AP_HECT and E6AP_NOD, as shown by white colonies. When the same experiment was performed using the HECT domain of E6AP now fused to the DNA binding protein LexA, and the E6AP_NOD is fused to the B42 transactivation domain, full-length E6AP is again reconstituted in trans. We also probed to see what would happen if the N-lobe and the C-lobe of E6AP_HECT were truncated, forming a construct with an exposed region 3 of E6AP_NOD while still retaining some residues from the HECT domain (forming E6AP_47–654). The HECT domain loop (742–755), absent in the E6AP_47–654 construct, provides electrostatic and hydrogen bonding interactions (Fig. 2C) and thereby bridges the N- and the C-lobes of the E6AP HECT domain. Testing the interaction between E6AP_47–654 and the HECT domain (Fig. 2E, blue box) unexpectedly revealed two different interaction behaviors when comparing a catalytically active HECT domain with an inactive HECT domain. The active HECT domain, with the catalytic cysteine (Cys843) intact, interacted with E6AP_47–654. In contrast, the inactive HECT domain (HECT_C843A) did not (Fig. 2E). Since ubiquitination is an active post-translational modification in yeasts, the difference in colony phenotype when the active HECT domain and inactive HECT domain are tested for interaction with E6AP_NOD suggests a second mode of interaction between E6AP_NOD and E6AP_HECT dependent on the ubiquitination activity of E6AP, which needs to be investigated further in future studies. These observations show an N-terminal ordered domain of E6AP that associates with the HECT domain independently to constitute full-length E6AP.

E6 interacts with E6AP through additional sites beyond the LxxLL motif.

Previous findings reported that residues starting at position 291 of E6AP are required for E6-dependent ubiquitination of p53. ⁴¹ The study also identified an 18 amino acid residue segment containing the LxxLL motif sufficient for E6 binding to E6AP, which was structurally characterized years later by co-crystallizing the peptide with E6 and p53.⁴² Using the full-length E6AP in preparing the ternary complex revealed additional interfaces on E6AP that mediate E6 recognition (Fig. 3A). This structural state is in agreement with the previously solved crystal structure of E6 (PDB ID: 4GIZ⁴²) and E6 in complex with LXXLL peptide and p53 (PDB ID: 4XR8²⁵), where two zinc-structured domains (E6N and E6C, respectively) are connected by the E6 connecting alpha helix (Fig. 3B). The apposition of Leu50_E6 to the LQELL motif of E6AP is consistent with the interactions identified by the yeast two-hybrid assay.²⁸ The cryo-EM structure reveals two additional binding interfaces between E6 and E6AP. At Interface I, Region 3 of E6AP_NOD (Fig. 3C) interacts with the E6 region C-terminal to the connecting helix (Fig. 3E helix 3, residues 64 to 78). Thr87_E6 and Tyr81_E6 approach within interacting distance, potentially through hydrogen bonding, to Glu472_E6AP and Glu470_E6AP (Fig. 3C). Another observation is the interactions mediated by Arg77_E6 (Fig. 3D), located at Interface II. This residue is within hydrogen-bonding distance from Glu406_E6AP (located in the disordered loop that houses the LQELL motif of E6AP) and Glu743_E6AP (located in the flexible segment between the N- and the C-lobes of the HECT domain). By mediating these interactions with two different regions of E6AP_NOD on either side of the LQELL motif, the viral protein strengthens its association with the ligase (Fig. 3D). Also at Interface II, The C-terminal end of the E6AP_NOD long helix (Region 4) participates in several hydrophobic interactions, specifically between Tyr76_E6 and Ile73_E6, with Ile507_E6AP and Phe69_E6 with Leu510_E6AP and Val514_E6AP (Fig. 3E). This helix-helix interaction between the connecting helix of E6 and the long helix of E6AP_NOD is stabilized by interactions between Lys65_E6 and Gln515_E6AP on one end, Lys72_E6 and Tyr511_E6AP at the center of the helix, and Tyr76_E6 and Arg506_E6AP on the other end. Interestingly, these key interacting residues are conserved between the high-risk and low-risk strains of HPV²⁸, which likely explains the ability of E6AP to recruit both high-risk and low-risk HPV E6 (Fig. S6). The interface between E6AP and E6 is also defined by several hydrophobic interactions between α3_E6 and the C-terminal end of the E6AP_NOD long helix. The E6AP LxxLL motif which is located in the middle of a long and unstructured loop (Fig. 2A) is structured when bound to E6. This long disordered loop may provide a higher degree of freedom for the E6AP LxxLL helix to engage the E6 binding pocket, in turn allowing LxxLL-E6 to reposition itself with respect to the E6AP_NOD. This model shows that E6 binds to the back side of E6AP (HECT N-lobe is designated as the front side of E6AP). With regard to the whole complex, this model shows the ligase engaging the p53 diametrically opposite to the C-lobe of E6AP_HECT, the catalytic center of the ligase where Ub is accepted before being transferred onto p53.

Figure 3. E6AP_NOD provides additional interfaces for E6-E6AP interactions.

(A) Left panel: E6/p53 is modeled within the cryo-EM map, revealing two additional interaction interfaces beyond the LQELL peptide. Right panel: E6AP is depicted with the same color scheme as that used in Fig 2 (region 1: green, region 2: blue, region 3: lime green, region 4: red). (B) The interaction between the LQELL peptide in our model (left) is very similar to the E6AP_LQELL/E6/p53 contact reported in the 4XR8 crystal structure²⁵. In addition, E6 also interacts with region 3 of E6AP_NOD (C), the HECT domain and disordered region 2 of E6AP_NOD (D), and region 4 of E6AP_NOD. (E) Interactions between the NOD long helix and E6 connecting helix show both hydrophobic and polar interaction modes. (F) Yeast two-hybrid assay validating E6 residues essential for interaction with E6AP_NOD. LexA_E6 and the indicated mutant E6 fusions were spotted horizontally and mated to the indicated B42 fusions proteins, and diploids were selected and assayed for protein association on XGAL plates. PTPN3 connotes a B42 fusion to the PTPN3 phosphatase that associates with E6 by a PDZ domain interaction; it serves as a positive control interaction for the association with E6 or mutant E6 proteins⁵⁰. LQELL or LQELS refers to a B42 fusion to the 12 amino acid LXXLL peptide of the wild type E6AP (LQELL) that binds E6 or mutant LQELS peptide that fails to bind E6. As indicated, the WT LQELL or mutant LQELS peptide was also expressed in the context of the full-length B42-E6AP_C843A protein. E6 cannot interact with LQELS but can when E6 is in the context of full E6AP, indicating additional associations between E6 and E6AP; the boxed mutations indicate where mutations in the E6 connecting helix disrupt that additional association²⁸. (G) Yeast three-hybrid assay suggests p53 ubiquitination rescue when E6 residues in the NOD long helix-interacting domain are mutated. LexA baits were co-transfected with B42-fusions as indicated in the horizontal rows, then mated to strains expressing wild-type unfused E6 or the indicated E6 mutants in vertical columns. Diploid yeast mating products were plated onto XGAL indicator plates. Mutation of the E6 connecting helix alters both ubiquitin ligase and substrate recognition of NHERF1.

A yeast two-hybrid assay ²⁸ was used to probe the role for NOD long helix in a construct of E6AP LQELL motif mutated to LQELS. In this mutant, the association between E6AP/E6 would rely on interactions involving residues outside region 2 of E6AP_NOD (E6AP residues 387–426). Residue Ile73_E6 was indeed crucial for the binding of E6 with E6AP, as mutating this residue to an alanine resulted in a loss of binding between E6 with both the LQELL and LQELS variant of E6AP (Fig. 3F). Residues Phe69_E6 and Arg77_E6 were also shown to be crucial for the interaction with region 4 of E6AP_NOD, despite the residues being away from the LQELL helix binding site (Fig. 3F). Arg77_E6 coordinates two hydrogen bonds between different regions of E6AP_NOD (Fig. 3D) and showed reduced interactions to region 4 only when neighboring residues were also mutated (Fig. 3G, column 12). This observation identifies Arg77_E6 as a key residue for activation of E6AP. This residue makes hydrogen-bonding interactions with two glutamates (Glu406 and Glu743) from E6AP_NOD. We predicted that mutating either one of these glutamates to alanine in E6AP would result in impaired interaction of the mutant ligase with E6. We used Biolayer Interferometry (BLI) to quantify the binding affinity of E6AP (WT and mutants) with E6. All three mutants (E6AP_E406A, E6AP_E743A, and the double mutant E6AP_E406A/E743A) showed impaired binding to E6 when compared to the wild-type E6AP (Fig. S7A). The reduced binding is most likely due to the loss of hydrogen-bonding interactions with Arg77_E6. Interestingly, the affinity of E6AP_E743A to E6 and the double mutant E6AP_E406A/E743A to E6 show a similar degree of binding affinity impairment, which likely hints at the prevailing importance of the Arg77_E6 with Glu743_E6AP interaction compared to Arg77_E6 with Glu406_E6AP.

We further evaluated the effect of these mutations in their ability to polyubiquitinate p53 via the in vitro p53 ubiquitination assay. The assay this time round was conducted differently with the concentrations of the reactants reduced considerably compared to that used in Fig. 1E. For this assay, we incubated Uba1 (250 nM), UbcH7 (1 mM), E6AP (200 nM), MBP-16E6 (200 nM), p53_1–312 (1 mM) with Ub (15 mM) and initiated the reaction by adding ATP to a final concentration of 1 mM. A control sample where E6AP (WT or mutants) was excluded from the reaction was also included. Samples at different time points (as indicated) were quenched with 5X SDS-PAGE dye. 4mL of each reaction time point was resolved by SDS-PAGE and probed for p53 by immunoblotting against α-p53 (DO-1) antibody (cat. no. MAB1355, R&D Systems). Results show that within the first 10 minutes of the reaction wild-type E6AP is able to polyubiquitinate most of the p53 (Lane 2; Fig S7B - observed by the disappearance of p53 signal relative to the no-E6AP control where p53 stays unmodified during the course of the reaction) The three mutants, however, are not as efficient as the wild-type protein although by the 30 minute time-point even the mutants are able to modify most of the p53 in the reaction. This indicates that the weakened binding exhibited by the mutants impairs their ability to ligate ubiquitin onto p53. Collectively, these findings validate our results from the yeast-three hybrid assay (Fig. 3).

To probe concordance between the ternary cryoEM structure and the traits of E6 mutants in vivo, we used yeast cells expressing all three components, leveraging p53’s activity as a transcription factor in a yeast three-hybrid system. p53 contains a transcriptional activation domain, which, if brought close to a LexA-lacZ reporter gene, results in transcription of the lacZ gene, forming blue colonies on the X-gal plate. As we have shown previously²⁸, when p53 + E6 + LexA-E6AP_C843A (catalytic cysteine on E6AP mutated to produce catalytically inactive mutant) are expressed in yeast containing a lexA-responsive lacZ reporter gene, blue colonies arise because the p53 transactivation domain is recruited to the reporter gene by LexA-E6AP_C843A + E6 (Fig. 3G). However, with a catalytically active E6AP in the LexA-E6AP construct, white colonies arise due to the E3 ligase activity of WT E6AP in ubiquitinating p53, leading to its degradation and loss of the transactivation of the reporter gene. (Fig. 3G). B42-activation domain fused NHERF1 was used as a control in place of p53 to describe other substrates recruited to E6AP by HPV E6 proteins^28,43. When testing the same mutants of E6 from the E6AP-E6 association assay, mutation of some residues exhibited substrate-specific and unexpected behavior. The mutation of Arg77_E6 to a glutamate residue resulted in a blue colony in the presence of WT E6AP, indicating that although the E6 mutant could recruit p53, the ubiquitination and subsequent degradation of p53 by E6AP was impeded. Interestingly, mutating residues in the vicinity of Arg77_E6, such as Lys72_E6 and Glu75_E6, did not change the ability of E6 to activate E6AP for ubiquitination of p53 but ablated the ability of E6 to recruit the substrate NHERF1 to LexA-E6AP_C843A without altering ubiquitin ligase activation as seen with R77E_E6 mutation.

Association of the E2 UbcH7 with the ligase complex

Ubiquitination of p53 by E6AP requires the recruitment of a Ub-charged E2 (Ub~E2) onto the N-lobe of the E6AP HECT domain. We attempted incubations of UbcH7~Ub with the ternary E6AP/E6/p53 complex, but cryo-EM studies of this sample led to the reconstitution of the E6AP/E6/p53 ternary complex without any density for UbcH7~Ub. All attempts to co-immunoprecipitate UbcH7 and E6 together with E6AP were not successful. These results are consistent with a prior TAP-tagged E6 affinity purification report showing E6AP association but not the UbcH7⁴⁴. However, the structure of the E6AP HECT domain in complex with UbcH7 is known (PDB ID: 1C4Z²⁴) and enabled modeling of the UbcH7-E6AP interactions in the E6AP/E6/p53 ternary complex. Also, an alternative approach using protein crosslinking mass spectrometry studies with disuccinimidyl sulfoxide (DSSO) was employed to obtain evidence for a transient five-protein complex in solution. DSSO crosslinks lysine residues within 25 Å of each other⁴⁵. Hence, any distance beyond 25 Å could indicate transient longer-range interactions in solution. The experimental conditions compared the DSSO crosslinking of E6AP/E6/p53 alone versus the E6AP/E6/p53 complex incubated with UbcH7~Ub. UbcH7~Ub is produced by 5,5-dithio-bis-(2-nitrobenzoic acid) (DTNB) crosslinking Ub_G76C to UbcH7_C17A/C137A. Several additional crosslinks were formed in the presence of UbcH7~Ub. The C-lobe of the HECT domain crosslinks extensively with UbcH7 as expected (Fig. 4A, 4B) to position the essential catalytic cysteine (Cys843_E6AP) to transfer Ub from UbcH7, as reported previously by Kamadurai et. al. for HECT-type E3 ligases.²⁰ Crosslinks were also observed between the C-lobe and the N-lobe of the HECT domain, suggesting the possible movement of the C-lobe, and by extension the Ub-charged C-lobe, to the N-lobe. Of note is an intermolecular crosslink observed between region 1 of the E6AP_NOD with p53 (Fig. 4A, 4B) in the presence of UbcH7. This crosslink occurs on Lys164_p53, the dominant site of ubiquitination on p53 (Figs. 1G and S1). The solution dynamics suggesting the movement of p53 elucidated by crosslinking is an interesting subject to be investigated further and will be part of subsequent studies of this system. The UbcH7 association with the E6AP-E6 complex was investigated further in the yeast system. When the transactivation domain of Gal4 was fused to UbcH7, E6AP-UbcH7 interaction was observed, but not in the presence of E6 protein (Fig. 4C). These data reveal a mechanism previously not described, wherein the association of E6 with E6AP helps recruit substrates such as p53, NHERF1, and SCRIB, but also represses UbcH7 binding to E6AP. To test this E6 role in modulating binding partners to E6AP, E6 mutants with impaired E6AP binding were used to rescue the association of UbcH7 with E6AP in the yeast three-hybrid system. Indeed, the E6 mutants that impaired interaction with region 4 of E6AP_NOD rescued UbcH7 interaction with E6AP. Of interest was the Arg77_E6 mutation, which showed impairment of p53 ubiquitination, which also reversed the ability of UbcH7 to bind E6AP (Fig. 4D, red box E14).

Figure 4. Dynamics of E2 association with E6AP.

(A) Residues crosslinked by DSSO that are over 10 Å apart in the ternary complex + UbcH7~Ub model compiled into a table, and visually represented in panel (B). (C) Yeast three-hybrid assay showing UbcH7 binding repression by E6. (D) A yeast three-hybrid assay between mutants of E6 was used to observe the effects of the mutation on UbcH7 binding to E6AP. (E) Comparison of UbcH7 binding repression by E6 proteins from different genera of papillomaviruses.

We previously described an evolutionary split in papillomaviruses between a group whose E6 proteins cluster together in evolution and preferentially interact with MAML1 to repress Notch signaling and a distinct evolutionary cluster (termed the super-alpha group that includes both low and high-risk HPV types) that preferentially associate with E6AP and stimulates its ubiquitin ligase function⁴⁶. E6 proteins from these two groups differed in the ability to displace UbcH7 from E6AP, as demonstrated in Fig. 4E. All the E6 proteins from the super-alpha group displaced UbcH7 from E6AP yeast. None of the MAML1-associated E6 proteins did so (Fig. 4E, columns 11–12 vs. 13–16), making the correlation between those E6 proteins that activate E6AP degradation and those that displace UbcH7 from E6AP complete, including both high-risk and low-risk HPV E6 proteins (Fig. 4E, columns 3–4 vs. 5–7). Remarkably, the repression of UbcH7 association with E6AP by E6 proteins in the super-alpha group is observed in papillomaviruses affecting distant species (Fig. 4E, columns 20–24). It is possible that the ectopically expressed E2, primarily in uncharged form, has a lower affinity for E6AP when it is bound to E6, which warrants future investigations.

Discussion

HPV-E6-mediated degradation of p53, a common denominator among several human cancers, has been the focus of research for many years now. Despite some concurrent structural studies,^31,32 our mechanistic understanding of how the viral factor manipulates E6AP to tag p53 with Ub has remained fragmented. The studies here bridge this gap in knowledge by determining the structure of the E6AP ligation machinery through cryo-EM studies. A key feature of in our report reveals an N-terminal ordered domain (which we named E6AP_NOD) that can be further subdivided into 4 regions. There are two helical bundles (regions 1, residues 119–170 and 236–386, and region 3 (residues 427–485) that are separated by a long disordered loop containing the LQELL sequence (region 2). Region 4 (which we also termed the NOD long helix) is a 30-residue helix at the end of the NOD that connects the E6AP_NOD to the E6AP_HECT domain (Fig. 2A). The model predicts that region 3 is necessary for the association of E6AP_NOD to E6AP_HECT through several electrostatic and hydrogen bonding interactions. AlphaFold3⁴⁷ was used to model the E6AP_NOD assocation with E6AP_HECT when the two sequences are inputted independently as two separate interacting proteins (Fig. S8). The solvent-exposed faces of regions 3 and 4 exhibit additional interactions with E6. NOD long helix interaction is driven by hydrophobic packing and hydrogen bonding interactions along the entire length of the helix. In our structure, we note how residue Arg77 on E6 facilitates electrostatic interactions with two different regions of E6AP – one located within region 2 (disordered loop) and another within the HECT domain. The interacting residue on E6AP_HECT, Glu743, is noteworthy because it is a part of the loop that connects the N-lobe and C-lobe of the HECT domain and has a role to play in the association of E6AP _NOD to the HECT domain of the ligase (Fig. 3D).

The results from the yeast three-hybrid assays reveal that mutating Arg77_E6 to any other residue extricates p53 from ubiquitination despite retaining the ability to recruit p53 onto E6AP. Given that Arg77_E6 bridges the interaction between the LxxLL helix loop and the HECT domain of E6AP, this interaction is expected to be key in the activation of the ligase for p53 ubiquitination. Interestingly, Arg77 and His78 are highly conserved among E6 proteins from the alpha supergroup (including low-risk HPV (Fig. S9) but not among E6 proteins that preferentially associate with MAML1 and do not stimulate E6AP degradation⁴⁶. The repression of UbcH7 binding to E6AP by E6 explains the ligase’s activation mechanism. Rapid turnover of UbcH7 after Ub is transferred to E6AP prevents inhibition of the E6AP HECT C-lobe catalytic cysteine from its required motion for subsequent Ub transfer to p53. This feature might distinguish high-risk and low-risk HPV strain E6 protein’s ability to catalyze p53 ubiquitination. However, HPV11 E6 and HPV7 E6 (low-risk strains) also suppress UbcH7 binding to E6AP.

The crosslinking mass spectrometry data suggests a conformational change that brings p53 closer to E6AP_NOD and the N-lobe of E6AP_HECT. Based upon this observation, it is reasonable to suggest that Ub transfer onto p53 occurs around the N-lobe of the E6AP_HECT domain. Possibly, the motion that brings p53 close to the HECT C-lobe might be a processive process caused by the binding of UbcH7 to E6AP, which competes with E6 binding to E6AP. Future work will include further detailed trials to collect more cross-linked data to deduce complex dynamics. Since the N-lobe of E6AP_HECT, where UbcH7 binds, is connected to E6 through the Glu743_E6AP-Arg77_E6 interaction (Fig. 3D), it is possible that binding of UbcH7 to the ligase destabilizes this interaction to enforce the intermolecular interaction of Arg77_E6 and Glu403_E6AP. Glu403_E6AP is located in the long flexible loop of E6AP_NOD (region 2)–the interaction with the flexible loop could lead to its motion in bringing p53 closer to the site of Ub transfer by the E6AP_HECT C-lobe. Further studies that probe the dynamics of complex assembly and disassembly more sensitively are required to confirm this activation mechanism hypothesis.

Recent work from the Scheffner lab showed that some of the substrate proteins, like NHERF1, recruited to the E6AP during their in vitro binding assays are common to both high and low-risk alpha genera E6 proteins⁴⁸. The observations here implicate the interaction of the E6 connecting helix with the NOD long helix in the recruitment of NHERF1 by E6 to E6AP. NHERF1 interacts with E6AP²⁸ via E6 and in turn displaces UbcH7 (Fig. 4C). A screen for HPV16 E6 mutants that failed to degrade NHERF1 identified E6 mutations at position Lys72_E6 that ablated degradation of NHERF1 (but not p53) by E6²⁸ and interaction of NHERF1 with HPV16 E6 and E6AP in yeast (Fig. 3G). The results suggest that the interaction between the E6 connecting helix and the NOD long helix is important for substrate recruitment and ubiquitin ligase activation. It remains unclear if NHERF1 interaction directly involves contacts with Lys72_E6 or if the E6 connecting helix with the E6AP NOD long helix induces an altered E6AP confirmation to enhance the recruitment of substrates for E6AP.

Dysregulation of E6AP is often related to Angelman syndrome, a rare neurogenetic disorder resulting in developmental problems at around 6–12 months of age. While most familial Angelman mutations are nonsense or splice variants, three missense mutations in the region 4 NOD long helix have been described (at positions 488, 491, and 497). Interestingly, two single-point mutants indeed affect E6AP function (Fig. 5); the N497V mutant resulted in an inactivated E6AP, as indicated by blue colonies shown when the N497V mutant is introduced to E6AP_WT, while the Y494V mutant resulted in an increased affinity between UbcH7 to E6AP, possibly impairing the activation of E6AP. The T488A mutant showed no change in activity. These structure-function correlates using the yeast hybrid study implicate the potential reason why E6AP mutants in the long helix result in Angelman syndrome. Even so, the long helix interaction doesn’t seem to be the only factor in causing Angelman’s disease, as indicated by our observation of the T488A mutant.

Figure 5. Angelman Disease residue mutation tested on yeast hybrid studies.

Whilst this manuscript was in preparation, independent studies from three different labs, describing the complexes of E6AP/E6 with and without p53 were published. The first paper published by JCK Wang et al ³¹ revealed the structure of the E6AP-E6-p53 complex, highlighting the additional surfaces formed between full length E6AP and E6. This additional interaction increased the previously characterized 939 Ų interacting surface between the LxxLL helix with E6, to now be 2361 Ų. This interaction is similarly observed in our structure, where region 4 (site 2 in their manuscript) and region 3 (site 3 in their manuscript) of the E6AP_NOD contributes to the E6AP-E6 interaction. JCK Wang and co-authors also reported that the affinity increase as measured by SPR increased from a K_D of 2–4 μM between E6 to LxxLL helix only, to 144 pM in the case of E6 to full length E6AP. Further SPR analysis on E6 mutants also highlighted the importance of several residues, namely Arg77 (referred as R84 due to different isoform numbering), which was shown to significantly decrease p53 ubiquitination and E6AP activity upon mutation.³¹ Their study provided the first structural and biophysical corroboration of the yeast two-hybrid studies. The second paper by Z Wang et al characterized the structure of E6AP in complex with E6 alone and found that the the E6AP-E6 complex is found as a dimer ³². The paper characterized the terminal residues of the region 4 long helix acts as a dimerization interface, in addition to the interaction between the HECT C-lobe to the inter-subunit E6 N-lobe. Z Wang and co-authors suggested that this dimer state of E6AP is the active form, where E6, a natural substrate of E6AP, can be ubiquitinated. Inferring from the different states of the structure, they also suggested that when E6AP_HECT C-lobe is detached from the E6 N-lobe, p53 can be docked into the structure without clashes. This suggests that p53 is ubiquitinated in this form of the E3 ligase complex. The authors generated a dimerization inhibiting mutant of E6, E6_{D25A/Y43A/R141A}, and showed that significant decrease of activity was observed. Although not observed in this paper, the 2:2:2 E6AP:E6:p53 oligomerization state was observed in a pre-print published in 2023. Sandate et al observed the formation of higher-order complexes when the three constituents of the ternary complex were present in equimolar amounts. Interestingly, when the tetramerization domain of p53 is present, only the dimeric state of the ternary complex was observed. Furthermore, the dimerization state captured by Sandate et al is different from Z Wang et al’s structure. The alternate dimerization state observed here is built on interactions between the E6AP_NOD of two E6AP subunits. Whether one dimerization state is more predominant than another is still a matter of study for the future.

Both of the paper published this year performed MD simulations to validate the interactions observed and concluded independently that the E6 interaction to E6AP stabilizes the LxxLL helix significantly, forming a more stable complex which can be primed for increased E3 ligase activity. In this stable complex, Z Wang et al further elaborates that the dimerization is favorable, given the stabilization of not only the LxxLL helix, but also the region 4 long helix (referred to as α1-helix) when it transforms from a short helical structure into a long extended helix.

The comparison of the published structures with our structure is presented in Figure S10. All in all, the E6AP/E6 placement in the structures are similar between the three structures, with slight differences (RMSD below 3 Å with both structures) that can be attributed to different modeling approaches (AlphaFold2 vs. ab initio atomic modeling vs. AlphaFold2 followed by ISOLDE). We observe four key differences captured between the structure reported in this paper to the published structures. Firstly, density was present for the E6AP HECT C-lobe, thus allowing its modeling in our structure (Fig. S10C). Secondly, the conformation of the HECT C-lobe captured in our structure differs from those in the dimeric E6AP-E6 complex (Fig. S10D, top circle), which highlights the dynamics of the C-lobe – an essential feature of this catalytic domain. Thirdly, overlaying our structure with the dimeric structure, we could observe how p53 would be placed concerning the HECT C-lobe. This observation suggests how E6AP ubiquitinates more efficiently as a dimer, although dimerization-deficient E6 mutants are still able to ubiquitinate p53 albeit with lower activity.

The cryo-EM structure reported here recapitulates and builds on prior observations with novel yeast hybrid studies to provide new insights into the ternary complex’s conformational dynamics, especially the interplay with UbcH7. The results inspire a proposal for a mechanism of p53 ubiquitination (Fig. 6) that does not require dimerization. Our observations suggest that the process for E6AP activation starts with the recruitment of p53 by E6AP/E6, adopting the structure of the ternary complex. Following this step, Ub~UbcH7 engagement occurs, with rapid dissociation of UbcH7 from the N-lobe of E6AP_HECT. Subsequent E2 association to E6AP/E6/p53 could destabilize HECT-E6 interaction, which we hypothesize leads to E6 dissociation from the HECT domain and moving with the long-disordered loop of E6AP_NOD region 2. This motion would place p53 in proximity to the catalytic cysteine, allowing Ub transfer to occur. These steps would be repeated when Ub~UbcH7 is recruited to the complex repeatedly, leading to Ub chain elongation on p53, a requirement for proteasomal degradation. Z Wang and co-authors proposed an active form of the E6AP-E6 complex which requires the dimerization of an E6AP/E6 complex. The model proposed in this paper gives insight into the hypothesis of activity mechanism by a monomeric E6AP-E6 complex, which can explain how dimerization-deficient mutants of E6 could still catalyze the ubiquitination of p53. When E6AP is not in complex with E6, it is plausible to reason that monomeric E6AP should be able to catalyze ubiquitin transfer. The rapid dissociation of UbcH7 observed plays a key role in any oligomerization state of E6AP.

Figure 6. Proposed Mechanism of p53 ubiquitination through E6 activation of E6AP.

In conclusion, our study captured the ternary complex and validated the complex through yeast hybrid experiments. The introduction of E6 activates the ligase, rapidly changing the rate of reaction and association kinetics. This property of E6 makes this viral factor a molecular glue⁴⁹ since interactions between E6AP and the neosubtrate p53 exist but alone are not meaningful enough to induce ubiquitination. The work here uniquely identifies Arg77 in E6 as a key residue playing a vital role in this activation of E6AP and show that E6 association competes with UbcH7 binding. This model leads to a mechanistic proposal for how E6 promotes E6AP ubiquitination of p53, setting the stage for future studies to capture the dynamics of this system.

Resource Availability

Lead Contact:

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Chittaranjan Das (cdas@purdue.edu).

Materials Availability:

This study did not generate new unique reagents.

Data and Code Availability:

• The cryo-EM structure has been deposited in the Protein Data Bank and Electron Microscopy Data Bank (PDB: 9CHT, EMDB: EMD-45601). The mass spectrometry data of the P53 Ubiquitination Assays have been deposited (Massive: MSV000096177). These data are publicly available as of the date of publication. Accession codes are also stated in the key resources table.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
α-p53 (DO-1) antibody	R&D Systems	MAB1355
Bacterial and virus strains
E. coli strain BL21 (DE3)	Novagen	69450
Chemicals, peptides, and recombinant proteins
Ampicillin sodium salt	Fisher Bioreagents	BP1760–25
Isopropyl b-D-1-thiogalactopyranoside (IPTG)	VWR	97063–282
cOmplete, EDTA-free protease inhibitor cocktail tablet	Roche	11836170001
Lysozyme	Sigma Aldrich	L6876
KCl	VWR	470302-088
SP-Sepharose resin	Cytiva	17072910
Tris-HCl	Roche	10812846001
NaCl	VWR	SIAL71376-5KG
Sodium Dodecyl Sulfate (SDS)	Sigma Aldrich	L3771
Glycerol	Thermo Scientific	A16205.0F
Dithiothreitol (DTT)	GoldBio	DTT10
Bromophenol Blue	Thermo Scientific	A18469.09
Ni-NTA resin	Cytiva	17526801
NP40	Thermo Scientific	85124
Tween 20	Thermo Scientific	85113
Bovine Serum Albumin (BSA)	Sigma Aldrich	A2153
PierceTM HA-Tag IP/Co-IP Kit	ThermoFisher	26180
DSSO	Thermo Scientific	A33545
DMSO	Thermo Scientific	D12345
TrypZean®	MilliporeSigma	T3449
Pierce C18 spin tips	Thermo Scientific	PI84850
Deposited data
E6AP/E6/p53	This Study	PDB: 9CHT, EMDB: EMD-45601
P53 Ubiquitination Assay Spectra	This Study	Massive: MSV000096177
Experimental models: Organisms/strains
All lexA based yeast hybrid systems	Gyuris et al.51	N/A
lexA-E6AP and lexA-E6 variants	Drews et al.28	N/A
Oligonucleotides
Primers for mutagenesis, see Table S2	This Study	N/A
Recombinant DNA
pET28a-E6AP	Masuda et al.35	N/A
pACYCDuet-E6	Masuda et al.35	N/A
pET21a-p531-312	Peter Wright Lab	N/A
pET28a-HA-hRpn10	Dan Finley Lab	N/A
pET-SUMO	Invitrogen	K300-01
pRSETA	Invitrogen	V35120
Software and algorithms
MotionCor2	Zheng et al.52	https://emcore.ucsf.edu/ucsf-software
CryoSPARC	Punjani et al.53	https://cryosparc.com/
Phenix	Punjani et al.54	https://phenix-online.org/
Coot	Emsley et al.55,56	https://www2.mrclmb.cam.ac.uk/personal/pemsley/coot/
PyMOL	Schrödinger	https://pymol.org/2/
UCSF ChimeraX	Pettersen et al57,58	https://www.cgLucsf.edu/chimerax/
PyXlinkViewer	Schiffrin et al.59	https://github.com/BobSchiffrin/PyXlinkViewer
Other
Quantifoil R1.2/1.3, 300 mesh, holey carbon grids covered with thin amorphous carbon film	Quantifoil	Q350AR1.3
BLI Ni-NTA biosensor	Sartorius	18-5101

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Bacterial culture

Bacterial strains used in this study are listed in key resources table. Bacterial growth conditions can be found in method details.

METHOD DETAILS

Cloning, Expression, and Purification of Recombinant Proteins

The vectors for E6AP, E6, and p53 are kind gifts from the Masuda Lab (E6AP and E6) and the Schultz Lab (p53). pET28a (Novagen) carrying E6AP^FL isoform II and pACYC-Duet (Novagen) carrying E6 were co-transformed into BL21(DE3) Escherichia coli cells (Novagen). 4 L of LB media were inoculated with 1% of E6AP/E6 inoculum and allowed to grow to an OD₆₀₀ of ~0.6. Similarly, pET21a (Novagen) carrying p53^1–312 was transformed into BL21 (DE3) Escherichia coli cells (Novagen). 2 L of LB media was inoculated with 1% of the p53 inoculum and allowed to grow to an OD₆₀₀ of ~0.6. Cultures were induced with 350 μM Isopropyl ß-D-1-thiogalactopyranoside (IPTG), followed by incubation at 18 °C for 16–18 hours to allow protein expression. The cells were then harvested and resuspended in lysis buffer (1X PBS buffer supplemented with 0.4 M KCl + cOmplete, EDTA-free protease inhibitor cocktail tablet (Roche)). The cell resuspension, incubated with lysozyme for 30 minutes on ice, was lysed and the cellular debris pelleted by centrifugation (@ 75,000 × g) for one hour at 4°C. The clarified protein was purified by Ni²⁺-affinity chromatography using Ni-NTA resin (Cytiva), washed with wash buffer (1X PBS + 30 mM imidazole) and eluted with elution buffer (1X PBS + 300 mM imidazole) which was followed by size-exclusion chromatography (Superdex 75) with SEC buffer of 50 mM Tris pH 7.4 + 50 mM NaCl. The hexa-histidine tags at the N-termini of E6AP and p53 were left uncleaved.

UbcH7 was cloned into pET-SUMO plasmid (Invitrogen) and transformed into BL21 (DE3) E. Coli strain. Expression and purification of UbcH7 was the same as that followed for the ternary complex. Ubiquitin, cloned in pRSETA vector, a kind gift from Genentech Corporation (USA), was transformed into the BL21(DE3) strain of E. coli. Protein expression was carried out following the same protocol as described above. Harvested cells were resuspended in lysis buffer (50 mM sodium acetate pH = 4.5), followed by heat lysis at 70–80°C for 30 minutes before ultracentrifugation. Ub was purified by cation-exchange chromatography (SP-Sepharose; Cytiva) on a gradient of NaCl, running from 0 to 1M. Fractions containing pure Ub, confirmed by SDS-PAGE analysis, were pooled, concentrated, and further purified on Superdex 75 (SEC) with SEC buffer of 50 mM Tris pH 7.4 + 50 mM NaCl.

In vitro assays and mass-spectrometric analysis of p53 ubiquitination

p53 ubiquitination reactions containing 0.1 μM Uba1, 5 μM UbcH7, 1 μM E6AP or E6AP/E6 complex, 20 μM p53, 100 μM Ub, and 1 mM ATP were setup at 37 °C in a total reaction volume of 15 μL (reaction buffer = 25 mM Tris-HCl (pH 8), 50 mM NaCl, 1 mM DTT, 5 mM MgCl₂). The reactions, quenched with 5X SDS loading buffer (250 mM Tris·HCl, pH 6.8, 10% SDS, 30% (v/v) Glycerol, 10 mM DTT, 0.05% (w/v) Bromophenol Blue) at the indicated timepoints, were analyzed by 15% SDS-PAGE and Coomassie stained to detect protein bands. Similar reactions (with or without E6) were setup in a reaction volume of 60 μL for proteomics analysis of p53 ubiquitination sites. These reactions were allowed to incubate for 90 minutes at 37 °C and subjected to in-solution tryptic digest using the protocol described in⁶⁰, except for using chloroacetamide instead of iodoacetamide for alkylation of the cysteines.

Cryo-Electron Microscopy Data Collection

Freshly purified E6AP/E6/p53 sample were first visualized by negative-staining EM to check the sample quality. For cryo-EM grid preparation, aliquots of 3.5 μl samples at 2.5 mg/mL with 0.0002% NP40 (or glutaraldehyde crosslinked) were applied onto glow-discharged Quantifoil R1.2/1.3 holey carbon grids covered with thin amorphous carbon film. The grids were incubated with the sample for 10 s at 4°C and 100% humidity, blotted for 3.5 s, and plunged into liquid ethane using an FEI Vitrobot.

The grids were imaged by a Titan Krios microscope (FEI) running at 300 kV equipped with a Gatan K3 direct detection camera, a BioQuantum energy filter operated with a slit width of 20 eV. In total, 5,550 micrographs were collected using EPU in super-resolution mode at a nominal magnification of 105,000 (with a calibrated physical pixel size of 0.86 Å/pixel) with a fixed defocus value of −600 nm. Each micrograph was exposed for 1.92 s at a dose rate of 6 electron/pixel/sec and saved as 40 movie frames. Calculated defocus values are in a range of −0.6 to −1.5 μm.

Image Processing

Motion correction was done with MotionCor2⁵² within Appion⁶¹. All micrographs were imported to cryoSPARC⁵³ for further processing. CTF estimation was done with Patch CTF estimation. The CTF estimated micrographs were manually curated to remove micrographs that are outliers in various parameters. particle-picking was first performed with blob picker. Particles were extracted with a box size of 256 pixels that just covers the whole complex to avoid including neighboring particles. Multiple rounds of 2D classification were performed in cryoSPARC⁵³. After several rounds of cleaning, Ab initio reconstruction⁵³ and heterogeneous refinement⁵⁴ with multiple classes in cryoSPARC was used to analyze the heterogeneity. E6AP/E6/p53 shows conformation heterogeneity, which hinder high resolution determination. All resolution estimations were based on the gold-standard Fourier Shell Correlation (GSFSC) calculations using the FSC = 0.143 criterion in CryoSPARC.

Structure prediction, Model Building & Visualization of the ternary complex

AlphaFold predictions of E6AP and p53 were downloaded from the AlphaFold Protein Structure Database. Structure of E6AP/E6 complex was predicted using the ColabFold Notebook. Structure predictions were then used for modeling in COOT^55,62. The ternary complex was fit within the cryo-EM map using rigid body fitting in Chimera^57,58. Overlapping structures and loops were deleted and the missing fragments were manually built in COOT⁵⁶. The resulting model was subjected to real space refinement in Phenix⁶³. Figures were generated using PyMOL, Chimera⁵⁷, and ChimeraX⁵⁸.

Yeast hybrid assays

The lexA based yeast hybrid system is based upon strains and plasmids originally developed in the Brent lab⁵¹ with modifications to the strains and plasmids as described²⁸. The full length E6AP in yeast experiments is derived from isoform 3 mRNA (872 amino acids), but is labeled in accordance with isoform 2 mRNA (875 amino acids) so as to keep the active ubiquitination site labeled as C843, and the ubiquitin defective mutant labeled as C843A. Mutants labeled as E6AP NOD express aa 47–518 and mutants labeled as HECT express aa 515–875.

Biolayer Interferometry (BLI) Kinetics Study

Each of the His-tagged E6AP proteins (WT and the three mutants: E406A, E743A and E406A/E743A) were diluted in BLI buffer (1x PBS containing 0.05% v/v tween 20 and 0.1% w/v BSA) to a concentration of 25 μg/mL. The analyte, HPV16-E6 was also diluted in BLI buffer to 1mM. E6 was then diluted down to eight different concentrations with the same BLI buffer binding analysis. 40 μL of each analyte solution and buffer was added to a 384 tilted well plate. One Ni-NTA biosensor was used for each K_D measurement, dipping the E6AP protein loaded tip into wells that contained the lowest concentration of analyte first. The association and dissociation steps were carried out for 120 seconds and 100 seconds, respectively. BAL Octet Acquisition Software (version 12.2) was used to collect raw data for the association and dissociation curves. Association responses was plotted in BAL Octet Data Analysis Software. The data were fit to a non-linear regression one site – specific binding model to determine the K_D.

Proteasome subunit pulldown of E6AP/E6/p53

The plasmid encoding human 26S proteasome non-ATPase regulatory subunit 4 (also called human Rpn10, hRpn10) in the pET28a (Novagen) vector was a kind gift from Dr. Suzanne Elsasser. The proteasome subunit was purified using His-tag affinity chromatography similar to the expression and purification of the recombinant p53. This hRpn10 construct contains an N-terminus HA-tag, allowing it to be used for the pulldown experiment. HA-hRpn10 was used as the probe protein to pulldown the E6AP/E6/p53 complex in this experiment. 20 μL of a 50% slurry of anti-HA beads provided in the PierceTM HA-Tag IP/Co-IP Kit (ThermoFisher) were equilibrated with 10× bed volume of 1×PBS buffer; pH 7.4. The beads were then centrifuged for 5 min at 5000 rpm and the supernatant discarded. This wash step was repeated twice. 100 μL of 100 μM HA-hRpn10 was then used to charge the anti-HA beads. These were then incubated at 4 °C with end-over-end mixing for 4 h, to ensure that the probe protein bound to the beads. The loaded beads were centrifuged for 5 min at 5000 rpm and the flow through collected. The beads were washed as before, and the wash collected after each step. Following the washes, the charged beads were incubated with 100 μL of a 100 μM stock of E6AP/E6/p53 sample. Binding was allowed to proceed overnight at 4 °C with end-over-end mixing. After incubation, the beads were centrifuged and washed as above, again collecting the flow through and wash at every step. Proteins were eluted by boiling the beads in 20 μL of elution buffer (provided in the kit), followed by centrifugation. The elution step was repeated once more. The collected samples were analyzed by SDS- PAGE.

Crosslinking-Mass Spectrometry Sample Preparation and Analysis

Purified E6AP/E6/p53 complex samples were pooled and subjected to cross-linking using DSSO at different concentrations in molar excess to the complex. Reactions of DSSO with E6AP/E6/p53 at 100:1, 250:1, and 500:1 molar ratios were incubated in room temperature for 1 hour before quenching with a final concentration of 20 mM Tris (pH = 8.0). A negative control sample of E6AP/E6/p53 lacking DSSO (but containing an equal volume of solvent, DMSO) was subjected to the same experimental conditions. The sample was split with some sample being kept in solution while the rest was run on an SDS-PAGE gel. For the E6AP/E6/p53/UbcH7~Ub sample, a 1:2 molar ratio of E6AP/E6/p53 to UbcH7~Ub was incubated for 2 hours before introduction of DSSO. Reactions of DSSO with E6AP/E6/p53/UbcH7~Ub at 100:1, 250:1, and 500:1 molar ratios were incubated in room temperature for 1 hour before quenching with a final concentration of 20 mM Tris (pH = 8.0). A negative control sample of E6AP/E6/p53 lacking DSSO (but containing an equal volume of solvent, DMSO) was subjected to the same experimental conditions.

In-gel digestion was performed essentially as described⁶⁴ by first running ~2.5 μg crosslinked protein on SDS-PAGE and staining with Coomassie Blue. High molecular weight protein bands were excised, and destained with 50% acetonitrile in 25 mM ammonium bicarbonate. Proteins were reduced with 10 mM DTT for 30 min at 25°C followed by alkylation with 55 mM chloroacetamide for 1 h in the dark at 25°C. Samples were dehydrated with 100% acetonitrile, rehydrated in 20 μg/mL TrypZean^® (MilliporeSigma) in 50 mM ammonium bicarbonate, and incubated for ~16 h at 37°C. Peptides were extracted twice by adding acetonitrile to 60% and incubating 10 min, then desalted on Pierce C18 spin tips (Thermo Scientific, PI84850) and dried by vacuum centrifugation prior to MS analysis.

Gel-free trypsin digestion was performed as described⁶⁵. Cross-linked and control samples were supplemented with 6 M urea (from 8 M stock), 5mM tris(2-carboxyethyl) phosphine, and 30 mM chloroacetamide to a final volume of 200 μL and incubated 1 h at 37°C. Samples were diluted with 3 volumes (600 μL) of fresh 50 mM ammonium bicarbonate to reduce urea concentration (<2 M), supplemented with 0.5 μg TrypZean^® and incubated at 37°C. After 12 h, a second aliquot of 0.5 μg TrypZean^® was added and incubation continued for an additional 12 h. Digestion was quenched with 0.1% trifluoroacetic acid to a final volume of 850 μL. Peptides were desalted using Pierce C18 spin columns (Thermo Scientific, PI89870) and dried under vacuum.

Tryptic peptides were solubilized in 3% acetonitrile/0.1% formic acid and analyzed by reverse phase LC-ESI-MS/MS using a Dionex UltiMate 3000 RSLCnano system coupled to an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific, Waltham, MA). Peptides were initially loaded on a trap column (300 μm ID × 5 mm) packed with 5 μm particle/100 Å pore PepMap C18 resin at 5 μL/min. Peptides were eluted from the trap column into an Aurora UHPLC emitter column (25 cm × 75 μm ID) packed with 1.6 μm/120 Å C18 resin (Ionopticks, Victoria, Australia) at constant flow rate of 200nl/min and temperature of 40 °C. LC solvent A was 0.1% formic acid and solvent B was 0.1% formic acid/80% acetonitrile. Peptide elution was achieved with a 3-stage gradient of increasing solvent B: 80 min from 8–27% solvent B, followed by 20 min from 27–45%, and finally 5 min from 45–100%. Data were collected on the Lumos in positive ion modes, using a data-dependent acquisition method with the advanced peak determination function. MS1 and MS2 scan ranges were 375–1,500 and 300–1250 m/z, respectively. Precursor ions were fragmented by higher energy collision dissociation at a normalized collision energy setting of 30%. Orbitrap resolution was 120,000 and 7,500 for MS1 and MS2, respectively, with maximum injection time of 50 ms for MS1 and 20 ms for MS2. Dynamic exclusion was set at 60 sec with 10 ppm tolerance.

Identification of DSSO-crosslinked peptides was performed with MetaMorpheus V0.0.320⁶⁶. Raw data from a DMSO-treated control sample and a DSSO crosslinked sample were searched together against a custom database containing the sequences of the five subunits of the recombinant E6AP/E6/p53/UbcH7~Ub complex using the three-stage MetaMorpheus workflow: calibration, general peptide search, and crosslink search. Key search parameters included 15 ppm precursor mass tolerance, 40 ppm product mass tolerance, crosslinker type = DSSO, potential crosslinking amino acids = KSTY, fixed modification = carbamidomethyl (Cys), variable modification = oxidation (Met), maximum 4 missed cleavages, Dissociation type = HCD, quench methods = Tris, H₂O, minimum score allowed = 2 (full search parameters listed in Supplemental spreadsheet). Total protein sequence coverage was high (Supplemental spreadsheet). Cross-linked peptide pairs were identified at an overall FDR of 1%. For simplicity, tables of all unique interlinked and intralinked peptide pairs identified at least once are reported in the Supplemental Dataset (using the highest total XL score for each). Crosslinked amino acids were visualized using the xVis software⁶⁷ and PyXlinkViewer⁵⁹.

QUANTIFICATION AND STATISTICAL ANALYSIS

The cryo-EM datasets were processed using CryoSPARC v3.2.0 then analyzed using the software listed in the key resources table. Cryo-EM data collection and refinement statistics are summarized in Table S1.

Supplementary Material

1

Table S1. Cryo-EM data collection and refinement statistics, related to STAR methods

Table S2. List of oligonucleotides used for mutagenesis, related to STAR methods

2

Data S1. Excel file containing additional data too large to fit in a PDF, related to Figure 4

Highlights:

1. Cryo-EM structure of E6AP/E6/p53 reveals an ordered domain on E6AP termed the NOD.

2. E6 Arg77, important in E6AP activation, interacts with E6AP NOD and HECT domain.

3. The binding of E6 to E6AP likely promotes the dissociation of UbcH7 from E6AP.

4. Angelman’s disease mutants are often located in the E6AP NOD long helix.