Moyamoya disease factor RNF213 is a giant E3 ligase with a dynein-like core and a distinct ubiquitin-transfer mechanism

[…]

1) The authors used auto-ubiquitination as a readout for E3 activity. It would be interesting to know if the ubiquitination of one of its substrates involved in MMD and focus on any potential differences between wild type and mutant RNF213. Have the authors done that? It might be possible that auto-ubiquitination is unaffected, but the ubiquitination of its substrate(s) changed.

The reviewers raise an important point. It would be very informative to test whether the ability of RNF213 to ubiquitinate its putative substrates is affected by MMD mutations, which do not seem to effect auto-ubiquitination activity. Unfortunately, to date no direct substrates have been identified. From the implicated targets, we have tested NFATC2, but could not detect substantial ubiquitination activity of RNF213. Most likely, the published association between NFATC2 and RNF213 is indirect, relying on other E3 enzymes operating in between.

In terms of RNF213 mechanism, our data demonstrate that the basal E3 activity, i.e. the ability to catalyze the transfer of ubiquitin from an E2 to a protein substrate, is preserved in the ∆RING deletion and the R4753K MMD mutants. We therefore consider it likely that either the RING domain that harbors many of the most severe MMD variants, or other pathologic mutations in the E3 module, affect ubiquitination by disrupting the substrate binding site of RNF213.

Moreover, we noted that the rate of self-ubiquitination is much higher than the rates of ubiquitination of other reaction components (E1, E2, BSA, free ubiquitin), with the majority of ubiquitin being attached to RNF213 when the reaction is running to completion (Author response image1). This apparent specificity for auto-ubiquitination may hint to a physiologically relevant activity, however, needs to be further explored.

Author response image 1

Download asset Open asset

2) The SEC elution profile shown in Figure 1—figure supplement 1B provides evidence that the sample is a single species but in the absence of a MW calibration curve or SEC-MALLS data does not provide any information about the monomeric/oligomeric state of the protein. Please either show such data if they are available or otherwise rephrase the sentence.

We have rephrased the text to clarify how the conclusion was derived: EM studies show homogenous monomeric particles, DLS measurement reveal the size and monodispersity of the analyzed monomer, as further confirmed by a SEC analysis. To clarify this point, we have now added the DLS data and have rephrased the corresponding paragraph in the main text:

“The initial cryo-EM class averages revealed a compact macromolecule with overall dimensions of 90 × 130 × 220ų, depicting the RNF213 ubiquitin ligase in its monomeric state. Size exclusion chromatography (SEC) and dynamic light scattering (DLS) analyses pointed to a monodisperse protein population (Figure 1—figure supplement 1BC), suggesting that the monomer is the dominant form of RNF213 in solution.”

3) The statement in subsection “RNF213 is a distinct E3 ubiquitin ligase” that UbcH7 is among the less reactive E2s is incorrect. UbcH7 is a cysteine-reactive E2. This section needs to be rephrased to make it explicitly clear what the difference between UbcH7 and other, lysine-reactive E2s is. And why the observation that RNF213 works with UbcH7 is so remarkable. Please make sure to include all relevant references to E3s that function via thioester intermediates.

Please also comment in the Results section if there is any structural or sequence homology between the E3 module of RNF213 and those E3s that contain a catalytic cysteine.

Based on this structure, could the authors speculate how RNF213 might work?

We thank the reviewers for clarifying the point of UbcH7 reactivity. The text has been rephrased accordingly, noting the inability of UbcH7 to discharge onto lysine residues.

“This finding is surprising, because UbcH7 is not known to collaborate with canonical RING-type E3s as it lacks intrinsic reactivity with lysine. Instead, UbcH7 is reactive only with cysteine residues and works together with transthiolation E3 enzymes such as HECT (Schwarz, Rosa and Scheffner, 1998), RBR (Wenzel et al., 2011), and RCR (Pao et al., 2018) ubiquitin ligases.”

Of note, when looking for structurally related domains of the E3 module in the entire PDB database, we did not find a single sub-domain matching more than 3 helices in any target, highlighting the unique nature of the RNF213 ubiquitination scaffold. Likewise, we could not detect sequence similarity to any other E3 enzyme, beyond the RING domain itself and short helical repeats. To account for these comparative analyses, we added a short statement in the Discussion part:

“Strikingly, the RING-less RNF213, which exhibits strong auto-ubiquitination activity, does not display structural similarity to any of the known ubiquitination enzyme in eukaryotes nor to the novel E3-like ligases (NEL, Maculins et al., 2016) that are found in bacterial pathogens and also operate in a RING-independent manner. […] Though the E3 mechanism, the identity of the active site cysteine and the role of the RING domain remain to be resolved, …”

Given the unknown identity of the E3 active site as well as of the nucleophile mediating the ubiquitin transfer from E2 to substrate, we cannot speculate on the molecular mechanism. Collaboration with UbcH7 suggests that the transthiolation reaction proceeds by a Cys-relay ping-pong mechanism. However, this is already stated in the manuscript.