Ubiquitylation – Which came first, the chicken, or the egg?

Earlier this week we explored protein ubiquitylation (AKA ubiquitination). This is a “pre-Genome” field, ancient in terms of modern molecular biology, yet recently rejuvenated by the finding that, as small as it is, the 76-amino acid ubiquitin molecule is nonetheless capable of a remarkably large number of self-interactions. 8 at the last count. These involve the conjugation of the C-terminal glycine from one ubiquitin molecule to any one of 7 acceptor lysine side chains or to the amino terminus of a second ubiquitin. This is then repeated in a poorly understood mechanism to lead poly-ubiquitin chains that may be linear, or branched, or both. All of this is interesting because we now believe that each of these interactions probably leads to a distinct biological outcome.

But as old or new as the field may be, a paper recently published in PLOS Computational Biology (Wang et al, v 10 e1003691 (link is external)) raises an age-old question: which came first, the chicken, or the egg*? More specifically, does a non-covalent association between two ubiquitin monomers allow their conjugation by a ligase that recognises the dimeric complex, or is conjugation driven by a ligase binding to the monomeric ubiquitin molecule at the end of the growing poly-ubiquitin chain which then recruits a new monomer? Using molecular dynamics simulations and a model that has a remarkable ability to predict structures for which there is experimental evidence, Wang et al show that ubiquitin monomers are able to self-associate using a range of hydrophobic and electrostatic interactions to form non-linked dimers that nonetheless closely resemble each of the covalently linked di-ubiquitin molecules for which structural evidence exists. This raises the possibility that two ubiquitin monomers, one fixed and one free, self-associate to form a template surface that attracts a cognate ligase whose sole role is then to formalise their union with a covalent bond. This is attractive. It is possible to imagine how local changes in pH for example would drive two ubiquitin molecules to associate in slightly different ways by protonating/deprotonating particular interface side chains, changing the electrostatic interactions that are available. However this model is at odds with the prevailing model where an E3 ubiquitin ligase and its E2 conjugation partner would bind to a protein target and build a repeating chain of di-ubiquitin linkages, and is also based on a very artificial model (are ubiquitin molecules ever free?) that is furthermore, then allowed to run through a series of computer algorithms.
But maybe both models do co-exist, and a chain of randomly associated ubiquitins is actually what drives protein degradation, while orderly chains formed in a controlled way by a specific ligase are the ones that drive distinct biological outcomes? This idea might suit the more organised amongst you, but not me. I am trying to keep my absent-minded scientist image intact, despite our clean desk policy here at Avacta. As a firm believer in stochasticity, I think the answer will be more…messy, and complicated. The idea that there might be complexity beyond the linear chain is frankly much more appealing. Fortunately structural information (e.g. Kommander et al 2009, EMBO Reports 10: 466–473) and new tools are beginning to emerge that will allow us to address these questions, including our own Affimer technology that specifically distinguish between di-ubiquitin linkages, such as:
  • K6-linked di-ubiquitin
  • K33-linked di-ubiquitin
  • K48-linked di-ubiquitin

View these products, and more, in our catalogue

*It was the egg, because chickens evolved from egg-laying dinosaurs!