How do proteins tie a knot?

Polypeptide chains, like yarn and headset wires, can get entangled and knotted, but in a defined way. While such an idea was inconceivable for structural biologists two decades ago, recent structural genomic initiatives and developments of protein knot detection algorithms have helped identify hundreds of knotted proteins with different knotted topologies, ranging from a simple trefoil knot to a very complex Stevedore’s knot (Figure 1). Using MJ0366 from Methanocaldococcus jannaschii, the smallest knotted protein known to date, as a model system, we systematically investigate its folding equilibrium, kinetics, and internal dynamics under native and chemically denatured states.

HFSP Career Development Award holder Danny Hsu and colleagues
authored on Mon, 18 May 2015

MJ0366 contains only 92 amino acids in length. Its folding pathway involves the formation of a native-like and monomeric intermediate before completing the rate-limiting step to form a unique ribbon-helix-helix MJ0366-like fold that is homodimeric (Figure 2) [1]. Compared to other proteins of similar sizes, however, MJ0366 exhibits exceedingly slow unfolding kinetics, which is on a timescale of hours. The very slow folding kinetics may be associated with the need to explore the vast conformational space to attain its complex knotted folding topology. Our finding also highlights the gap between experiments and theoretical calculations, which have managed to fold MJ0366 using a very realistic model within a microsecond [2]. While protein folding experiments and simulations are beginning to converge for small globular proteins, there is evidently a lot to be learnt for knotted proteins.

Figure 1: Protein knots are difficult to visualize directly from the ribbon representations of their three-dimensional structures. A protein knot detecting server, pKNOT (, can deduce knotted elements from the complex protein structures by backbone smoothing as illustrated for YibK, a bacterial RNA methyltransferase, and UCH-L1, a human deubiquitinase.

In fact, the slow folding kinetics are also found in other knotted proteins. For example, human ubiquitin C-terminal hydrolyase, UCH-L1, which is an abundant neuron-specific protein that contains a Gordian knot with five projected crossings in its backbone topology, also unfolds on a timescale of hours. We have shown that familial mutations in UCH-L1 that are associated with Parkinson’s disease significantly destabilize its folding stability, accelerate its unfolding rates and increase its aggregation propensity [3]. There is also good evidence to suggest that the loss of knotted elements due to N-terminal truncation greatly affects the folding and function of UCH-L1 [4].

Figure 2: Folding pathway of MJ0366 deduced from a multimetric biophysical approach.

Through systematic comparisons of the folding pathways of different knotted proteins using a multimetric biophysical approach, we aim to provide atomic insights into the mechanisms by which proteins thread themselves into defined knotted topologies. A more challenging question that remains to be answered would be: why Nature evolved such complex protein knots?


[1] Unraveling the folding mechanism of the smallest knotted protein, MJ0366. Wang, Chen and Hsu (2015) J. Phys. Chem. B. 119, 4359-70.

Link to article

[2] Knotting a protein in explicit solvent. Noel, Onuchic and Sulkowska (2013) J. Phys. Chem. Lett. 4, 3570-3.

Link to article

[3] The effect of Parkinson's disease associated mutations on the deubiquitinating enzyme UCH-L1. Andersson, Werrell, McMorran, Crone, Das, Hsu and Jackson (2011) J. Mol. Biol. 407, 261-72.

Pubmed link

[4] N-terminal truncated UCH-L1 prevents Parkinson’s disease associated damage. Kim et al., (2014) PLoS ONE 9, e99654.

Pubmed link