Functional diversity and structural disorder in the human ubiquitination pathway

Abstract

The ubiquitin-proteasome system plays a central role in cellular regulation and protein quality control (PQC). The system is built as a pyramid of increasing complexity, with two E1 (ubiquitin activating), few dozen E2 (ubiquitin conjugating) and several hundred E3 (ubiquitin ligase) enzymes. By collecting and analyzing E3 sequences from the KEGG BRITE database and literature, we assembled a coherent dataset of 563 human E3s and analyzed their various physical features. We found an increase in structural disorder of the system with multiple disorder predictors (IUPred - E1: 5.97%, E2: 17.74%, E3: 20.03%). E3s that can bind E2 and substrate simultaneously (single subunit E3, ssE3) have significantly higher disorder (22.98%) than E3s in which E2 binding (multi RING-finger, mRF, 0.62%), scaffolding (6.01%) and substrate binding (adaptor/substrate recognition subunits, 17.33%) functions are separated. In ssE3s, the disorder was localized in the substrate/adaptor binding domains, whereas the E2-binding RING/HECT-domains were structured. To demonstrate the involvement of disorder in E3 function, we applied normal modes and molecular dynamics analyses to show how a disordered and highly flexible linker in human CBL (an E3 that acts as a regulator of several tyrosine kinase-mediated signalling pathways) facilitates long-range conformational changes bringing substrate and E2-binding domains towards each other and thus assisting in ubiquitin transfer. E3s with multiple interaction partners (as evidenced by data in STRING) also possess elevated levels of disorder (hubs, 22.90% vs. non-hubs, 18.36%). Furthermore, a search in PDB uncovered 21 distinct human E3 interactions, in 7 of which the disordered region of E3s undergoes induced folding (or mutual induced folding) in the presence of the partner. In conclusion, our data highlights the primary role of structural disorder in the functions of E3 ligases that manifests itself in the substrate/adaptor binding functions as well as the mechanism of ubiquitin transfer by long-range conformational transitions.

Figure 1. Predicted disorder of the main classes of human E3 ubiquitin ligases.

We used the IUPred disorder prediction method for predicting structural disorder in 563 human E3 ligases, and calculated the average percent of disordered residues for proteins in the different sub-classes. The functional classification tree for the E3 family is shown above the bars. The specific functional characteristics for each main branch are indicated in boxes, such as interaction with E2 enzyme and/or with the substrate (‘S’), transient covalent binding to ubiquitin (‘covalent’), or functioning as a scaffold or adaptor/substrate recognition subunit in msE3s.

Figure 2. Distribution of the disorder content of E3s.

Residue-level structural disorder was predicted for all 563 human E3 ligases by IUPred, and the percent of disordered residues was calculated for each protein. A) The distribution of E3 proteins as a function of their disorder content. The superposed line shows the disorder tendency for the human proteome (30% sequence redundancy). B) The average percent of disordered residues as a function of the mean sequence length for different E3 families.

Figure 3. Structural disorder of E3 ligases as a function of their connectivity in the interactome.

Disorder content for the three connectivity groups of human E3s (hub: k≥25, ICP: 4≤k≤24, non-hub: k≤3). Green circles represent individual proteins. The bottom and top borders of the boxes represent the 25% and 75% of the data while the bottom and top whiskers indicate 10% and 90% of the data, respectively. The bold line indicates median value.

Figure 4. Induced folding of human E3 ligases in interactions with their partner molecules.

PDB structures are presented in which a disordered segment of a human E3 ubiquitin ligase binds to the folded domain of a human partner protein (neither an E1/E2/E3 enzyme nor a substrate for the given E3). A) Interaction between E3 ligase CBL-B (CBLB) and CD2-associated protein (CD2AP; PDB 2J6F). B) Interaction between E3 ligase CBL-B (CBLB) and SH3K1 (SH3 domain-containing kinase-binding protein 1; PDB 2BZ8). C) Interaction between E3 ligase MDM2 and UBP7 (Ubiquitin carboxyl-terminal hydrolase 7, also USP7; PDB 2FOP). D) Interaction between E3 ligase AMFR2 and TERA (Transitional endoplasmic reticulum ATPase, also VCP; PDB 3TIW). On all four panels the domain maps for the whole chain of both interaction partners are also shown, next to the PDB structure: the upper map is for the E3 ligase, the bottom one is for the partner. In the structures, the disordered E3 chains are represented as purple cartoon while the partner molecule is rendered in surface representation. The domain maps show the lengths and names of the proteins and their domains. The regions predicted to be disordered by IUPred are marked in purple, the ordered segments are in white; the regions present in the PDB structures are delimited by asterisks.

Figure 5. Induced folding in the interaction of E3 ligases and their substrates.

Three PDB structures are presented in which induced folding or mutual induced folding (cofolding, synergistic folding) occurs upon interaction of a human E3 ligase with its substrate. A) Interaction between E3 ligase SMURF1 and its substrate SMAD1 (SMA and mothers against decapentaplegic homolog 1; PDB 2LAZ) is a case of co-folding of two disordered regions. B) Interaction between E3 ligase RING2 and RYBP (RING1 and YY1-binding protein; PDB 3IXS) is also an example of co-folding. C) Interaction between E3 ligase MDM2 and P53 (P53 tumor suppressor protein, also TP53; PDB 1YCR), here the substrate undergoes induced folding upon binding to the folded SWIB domain of MDM2. On all three panels PDB structures and domain maps of the two proteins (E3 on top) are shown. On the domain maps, the names of domains, their positions and total length of the protein are indicated. The regions are color coded according to their IUPred disorder status: regions predicted to be disordered are in purple, ordered segments are in light grey. The regions present in the PDB structures are delimited by asterisks. In the PDB structures the disordered segments of partners are shown as purple cartoon whereas the E3 ligase is rendered in surface representation; disordered regions (mapped from disorder predictions on the unbound form) being light grey, and ordered regions white.

Figure 6. Structural organization and molecular dynamics analysis of an E2-E3-substrate complex.

Structural and molecular dynamics analysis of the complex (PDB code: 4A4C) between human CBL, ubiquitin-conjugating enzyme E2, and a peptide derived from the CBL substrate ZAP-70. (A) Structural organization of CBL, as seen in the crystal structure. The E3 molecule is in blue, E2 in dark grey and the ZAP-70 substrate peptide is in red. The predicted disorder profiles of the CBL sequence present in the crystal structure using (B) IUPred, and (C) FoldIndex, respectively. Vertical lines represent the linker helix region (CYS353-CYS381). In the IUPred plot, peaks represent the predicted disordered region(s), whereas in FoldIndex the negative values correspond to unfolded/disordered regions. The disorder calculations were run for the entire CBL sequence (UniProtKB: P22681), but the figure only shows the peptide segment (PRO48 – ASP435) present in the crystal structure. (D) Sequence of CBL with blue color indicating regions with high crystal B-factors (>100Ų). (E) RMSF plot from the 50ns MD simulation. (F) Distance between the center-of-masses of the substrate-binding TKB domain of CBL and the E2 as a time-series plot from the MD simulation. (G) Distance between the E2 catalytic CYS and the N-terminal SER of the ZAP-70 peptide. (H) Two orientations (“open” and “closed” forms) of the E2-E3-substrate peptide complex obtained from the NM simulation. They correspond to two extreme configurations (along the lowest frequency normal mode), showing the bending around the linker helix region that acts as a hinge/lever. The “open” configuration is colored dark grey, and the “closed” configuration is colored blue (E3), and orange (E2). The catalytic CYS85 and the substrate peptide are shown in spacefill representation. CYS85 are shown for both the open and closed forms of the structure, whereas the substrate peptide is shown only for the closed form (for clarity). The TKB domains of the two different configurations are structurally superposed using the C-alpha atoms. The TKBD is aligned with very low RMSD, whereas the RING-domain and the E2 have moved significantly in the two conformations (in the direction pointed by the curved arrow).

Figure 7. Structural disorder enables intramolecular diffusion in E3 action.

A simplified scheme of the linker (entropic chain) function of disordered regions in E3 ligases (for molecular recognition function, see text and Figures 4 and 5). Several ligases of the ssE3 family have a substrate-binding domain (SBD, can also be a disordered motif) and an E2-binding domain (shown as RING here, can be also a U-box or HECT domain) separated by a disordered linker region (dashed line). Due to the conformational freedom of the disordered linker, the bound substrate (S) and ubiquitin-charged E2 (E2∼Ub, ubiquitin shown in red) can diffuse toward and away from each other, without dissociating from the E3. This “intramolecular diffusion” mechanism enables proximity of substrate and E2∼Ub for ubiquitin transfer and also subsequent replacement of E2 with E2∼Ub in a more open conformation, i.e. (re)charging of the ligase. In principle, the flexibility of the linker enables the polyubiquitiniation or multiple monoubiquitination of the substrate, which may explain processivity of the ligation reaction.