The authors have declared that no competing interests exist.
Conceived and designed the experiments: SJLvW ASJM SJdV HTMT AMJJB. Performed the experiments: SJLvW ASJM. Analyzed the data: SJLvW ASJM SJdV HTMT AMJJB. Contributed reagents/materials/analysis tools: HTMT AMJJB. Wrote the paper: SJLvW ASJM SJdV HTMT AMJJB.
Current address: Institute of Biochemistry II and Buchman Institute for Molecular Life Sciences (FMLS), Goethe University School of Medicine, University Hospital, Frankfurt am Main, Germany.
Current address: Biomolecular Dynamics, Department of Physics T38, Technical University München, Garching, Germany.
Ubiquitination relies on a subtle balance between selectivity and promiscuity achieved through specific interactions between ubiquitin-conjugating enzymes (E2s) and ubiquitin ligases (E3s). Here, we report how a single aspartic to glutamic acid substitution acts as a dynamic switch to tip the selectivity balance of human E2s for interaction toward E3 RING-finger domains. By combining molecular dynamic simulations, experimental yeast-two-hybrid screen of E2-E3 (RING) interactions and mutagenesis, we reveal how the dynamics of an internal salt-bridge network at the rim of the E2-E3 interaction surface controls the balance between an “open”, binding competent, and a “closed”, binding incompetent state. The molecular dynamic simulations shed light on the fine mechanism of this molecular switch and allowed us to identify its components, namely an aspartate/glutamate pair, a lysine acting as the central switch and a remote aspartate. Perturbations of single residues in this network, both inside and outside the interaction surface, are sufficient to switch the global E2 interaction selectivity as demonstrated experimentally. Taken together, our results indicate a new mechanism to control E2-E3 interaction selectivity at an atomic level, highlighting how minimal changes in amino acid side-chain affecting the dynamics of intramolecular salt-bridges can be crucial for protein-protein interactions. These findings indicate that the widely accepted sequence-structure-function paradigm should be extended to
During their life, proteins undergo various modifications ranging from structural marking or signaling to degradation. One major biochemical process involves ubiquitin, a small and evolutionary conserved protein. This regulatory protein serves as a tag that, when attached to a protein substrate, alters its function, cellular sub-location or commits the labeled protein to destruction in the proteasome. The high specificity of the ubiquitination pathway is achieved through interactions between two large protein families, E2 and E3, that ensure the efficient covalent conjugation of ubiquitin. By comparing two “almost identical” E2 enzymes, we identified a single minute substitution that, operated by a dynamic network of salt-bridges, functions as a subtle switch that controls interaction selectivity toward E3 proteins. Using a combination of bioinformatics and modeling techniques, complemented by mutagenesis and experimental screening of E2-E3 interactions, we unraveled an equilibrium between an “open”, binding-competent and a “closed”, binding-incompetent state. Subtle modifications in this network are sufficient to switch the selectivity profile. These findings should serves as a cautionary tale and raises new challenges for bioinformatics analysis, modeling and experimental engineering of protein-protein interactions. The dynamic nature of the identified regulatory switch suggests that the widely accepted sequence-structure-function paradigm should be extended to
Biological systems critically rely on selective and specific protein interactions, creating building blocks that cooperatively form the basis of functional complexity
The molecular basis for this selectivity is provided by surface residues of E2 enzymes and by “cross-braced”, zinc-binding domains of the Really Interesting New Gene (RING) sub-family of E3 ligases
Efficient ubiquitin-conjugation depends on specific interactions between E2 and E3 enzymes. The high degree of sequence and structure conservation in E2s and, to some extent, among the E3 RING-finger domains and their reported E2-E3 interfaces, remains compatible with a highly selective binding for their cognate E2 or E3 partners
Based on this, physical E2-E3 interactions can be regarded as primary determinants for efficient conjugation reactions. We demonstrate here that a subtle and minute change – an aspartic to glutamic acid substitution (one methylene group difference) – is sufficient to completely change the selectivity profile of an E2 enzyme toward E3 RING-finger domains. Using molecular modeling and molecular dynamics simulations, a network of intra-molecular salt-bridges was identified that controls the balance between a binding-competent and a binding-incompetent state. Perturbation of network components located both in and outside the classical interaction surface resulted in a switch in the E3 interaction binding profile. These results suggest a new and delicate mechanism of how protein-protein interaction selectivity is achieved within the promiscuous ubiquitination system.
In previous studies, global E2-E3 (RING) interactions have been mapped by high-throughput yeast two-hybrid screening (Y2H), revealing that E2 enzymes that show high conservation within their E3 interface region interact with similar cohorts of E3 domains
To understand how these minimal variations in amino acids composition can cause divergent E3 interaction patterns, residues in UbcH8 were substituted with their counterparts from UbcH6 and experimentally tested by yeast two-hybrid screening for interaction against 250 human isolated E3-RING domains. In agreement with our previous study
To investigate the underlying molecular mechanisms at the origin of these differences, the structural environment of the differentiating residues in the unbound structures of both E2 enzymes was evaluated in more detail. Structural analysis of the respective PDB files (UbcH6: PDBid 3BZH; UbcH8; PDBid 1Y6L) revealed that the side-chains of H1 D58/E66 are remote from and not pointing toward the E3 interaction interface. Because of this, we decided to concentrate on the L1 D-to-E substitution. Inspection of UbcH6/UbcH8 E105 and D113 revealed that both amino acids are surface-exposed with their side-chains pointing toward the predicted E2-E3 interaction surface.
To dissect potential effects of the D-to-E substitution at the E2-E3 interface, E2 structures were studied in complex with the E3 RING domain. The RING-finger domain of the ubiquitin ligase TOPORS, previously reported to interact with UbcH6 but not with UbcH8, was used. Since at present, there are no experimental structures of UbcH6 or UbcH8 in complex with a RING domain, we used existing E2-E3 structures as templates, combining three-dimensional alignments, protein threading and refinement in explicit solvent using HADDOCK
To further study the impact of the D-to-E substitution, extended molecular dynamics (MD) simulations of the free forms UbcH6 and Ubch8 were performed. Cross-RMSD values indicated no significant conformational changes (
The dynamics of this salt-bridge network was further studied for both E2 enzymes by monitoring the distances between E105-K109 and K109-D137 in UbcH6 and D113-K117 and K117-D145 in UbcH8. We observed a slightly higher population of the binding-impaired conformation (D113-K117 salt-bridge formed) in UbcH8 WT than UbcH6 WT. Furthermore, the exchange toward the open form seems to be more frequent in UbcH6 WT (
The observed differences between the two E2 enzymes are clearly not sufficient to explain the experimental differences in selectivity profile. Still the MD simulations were key in identifying the network components. From these observations, we hypothesized that differences in side-chain length introduced by the D-to-E mutation are able to gently shift the balance of the labile conformational equilibrium between the open (binding-competent) and the bridged (binding-impaired) conformations. Accordingly, the key in controlling E3 interaction profiles between UbcH6 and UbcH8 should reside in the subtle intra-molecular dynamics of residues in L1, which, in turn, are influenced by residues remotely located from the interaction surface that are part of the identified salt-bridge network. If this hypothesis is true, then mutating residues in this network should also have impact on the selectivity profile of these E2s, something that can be experimentally tested.
The hypothesis of a side-chain equilibrium controlling UbcH8 and UbcH6 E2 interactions with E3 RING-finger domains was tested experimentally by perturbing key residues involved in the intra-molecular salt-bridge network.
First, the significance of K117 was assessed in UbcH8, since K117 was predicted to play a central role in alternatively contacting D113 and D145, thereby, preventing D113 of interacting with the RING domain. By substituting UbcH8 K117 with a histidine, the E3 interaction pattern of UbcH8 was reverted to that of UbcH6 (
We next investigated the role of D145, the most remote (as seen from the E3 interaction site) component of the network. Interaction screening of an UbcH8 D145K mutant demonstrates that D145K interacts in a similar manner as UbcH8 WT (
In contrast, the introduction of D145E in UbcH8 alters the interaction profile of UbcH8 WT to that of UbcH6 WT. Here, the longer glutamine side-chain can more effectively contact K117, forming a stable salt-bridge, and consequently freeing D113 for forming contacts with RING. To ascertain that this indeed involves D113, the double mutant UbcH8 D113A/D145E was also characterized, showing the loss of the previously gained E3 interactions (
Taken together, these results point toward a delicate network of intra-molecular salt-bridges that dynamically control the positioning of a crucial E3-interacting residue (schematically represented in
Although the catalytic UBC-fold of E2 enzymes is characterized by high levels of sequence and structural similarities, it is adapted to selectively recognize RING-finger domains to allow transfer of Ub to substrates
In the majority of E2 enzymes, residues involved in establishing E3 interaction specificity are not concentrated in a single hotspot, but dispersed over the N-terminal helix one and two relatively flexible and divergent loop regions (L1 and L2)
Finally, the N-terminal extension of the tested E2 enzymes, not present in the crystal structures, might also affect their E3 binding preference. This could explain why reverse mutations in UbcH6 that mimic UbcH8 did not restore the E3 interaction profile of UbcH8. The presence of another intermolecular salt-bridge in the UbcH6-TOPORS complex generated by HADDOCK, involving the residues LYS43 (UbcH6) and GLU28 (TOPORS) (data not shown), could explain why UbcH6 E105D did not affect the selectivity profile of UbcH6. This salt-bridge cannot be formed with UbcH8, where LYS43 is replaced by ALA51 in the sequence. Comparing the sequences of UbcH6 and UbcH8 indeed reveals amino acid differences at several key positions in this N-terminal extension known to be primarily involved in the interaction with the E1 ubiquitin-activating enzyme, but also to play a role in E3 interaction selectivity.
We should note that Y2H screening of physical protein-protein interactions between E2 enzymes and E3 RING-finger domains does not directly address biological functionality, e.g. the ability to transfer the activated Ub. However, we previously demonstrated that E2-RING E3 interactions found by LexA-B42 yeast two-hybrid assays are good predictors for enzymatic functionality
Despite their high percentage of identity, it has been reported that UbcH6 plays a major role in ubiquitin-conjugation while UbcH8 has only a minor role in ubiquitination but rather is the key conjugating enzyme of the ISGylation pathway
In conclusion, a dynamic equilibrium of conserved residues in two highly homologous E2 enzymes was identified, that mediate RING interactions. Amino acids located both within the classical interaction surface as well as residues that are remote from this surface are actively involved in modulating side-chain conformations and thus availability for binding of crucial residues. The subtle and dynamic nature of the identified regulatory switch suggests new ways how protein interactions can be controlled. Furthermore, the observation that minimal sequence differences between two highly similar proteins can control protein interaction networks serves as a cautionary tale and raises new challenges for bioinformatics analysis and modeling of protein interactions. Finally, these findings indicate that the widely accepted sequence-structure-function paradigm should be extended to
High-copy yeast-two hybrid (Y2H) shuttle plasmids expressing human UbcH6 (UBE2E1) and UbcH8(UBE2E2) and RING-finger domains of 250 human RING-type E3 ubiquitin protein ligases as fusions with the
Both wild-type and mutant UbcH6 and UbcH8 were used as starting structures. Wild-type UbcH6 and UbcH8 structures were taken from the Protein Data Bank (PDB) (PDB-ID
Structural models of UbcH6 and UbcH8 in complex with the RING-finger domain of TOPORS were generated by structural alignment of single components onto the bound structures of existing E2-E3 complexes (PDB:
Electrostatic potential surfaces were determined using the Adaptive Poisson-Boltzmann Solver package (APBS)
Manipulation of yeast cells and two-hybrid techniques are described in
(TIF)
(TIF)
The authors thank Prof. Rolf Boelens (University Utrecht) for helpful discussions.