AlphaFold predicts novel human proteins with knots

Abstract

The fact that proteins can have their chain formed in a knot is known for almost 30 years. However, as they are not common, only a fraction of such proteins is available in the Protein Data Bank. It was not possible to assess their importance and versatility up until now because we did not have access to the whole proteome of an organism, let alone a human one. The arrival of efficient machine learning methods for protein structure prediction, such as AlphaFold and RoseTTaFold, changed that. We analyzed all proteins from the human proteome (over 20,000) determined with AlphaFold in search for knots and found them in less than 2% of the structures. Using a variety of methods, including homolog search, clustering, quality assessment, and visual inspection, we determined the nature of each of the knotted structures and classified it as either knotted, potentially knotted, or an artifact, and deposited all of them in a database available at: https://knotprot.cent.uw.edu.pl/alphafold. Overall, we found 51 credible knotted proteins (0.2% of human proteome). The set of potentially knotted structures includes a new complex type of a knot not reported in proteins yet. That knot type, denoted 63 in mathematical notation, would necessitate a more complex folding path than any knotted protein characterized to date.

Knots and θ-Curves Identification in Polymeric Chains and Native Proteins Using Neural Networks

“…7,8 The properties of Fold, 21 and Evolutionary Scale Modeling (ESMFold) 22 to collectively release millions of 3D protein structure predictions covering most of the proteomic databases. A survey of AlphaFold database featuring a majority of UniProtKB entries revealed around 700,000 potentially knotted proteins 23 including knot types not seen before in proteins, such as 5 1 knot, 19,24 complex knots such as 6 3 , 25 7 1 , 19,24 8 3 , 24 and even a composite knot 3 1 #3 1 19 (this one is already confirmed experimentally by Bruno da Silva et al 18 ). To our knowledge, no one has checked the AlphaFold Database from θ-curves point of view, but since structures deposited in the PDB represent less than 1% of all known proteins, it is to be expected that more complex θ-curves may exist.…”

AlphaFold Blindness to Topological Barriers Affects Its Ability to Correctly Predict Proteins’ Topology

“…The 5 1 and 7 1 knots identified by Brems et al [26] are single-domain structures with no visible dense residue packing. The 6 3 knot found by Perlinska et al [27] is a double-domain knot; however, the domains are swapped. A similar effect was already observed in the analysis of a possible evolutionary pathway of the deep 4 1 knot, another topologically symmetric structure [28], which opens the pathway for the existence of the 6 3 -knotted protein.…”

“…In this spirit, among the structures predicted by AlphaFold, Brems et al identified two new knot topologies, with five and seven crossings in minimal crossing projection (5 1 and 7 1 knots); the latter, if experimentally verified, would be the most complex protein knot to date [26]. Yet another knot type, a symmetric knot with six crossings (6 3 knot), was found in AlphaFold predictions by Perlinska et al [27]. It is worth mentioning that all of those knots, if also identified in an experiment, would disprove the long-standing hypothesis that all protein knots are formed by single threading through a twist loop (so-called "twist knots") [28].…”

Mathematical topology and geometry-based classification of tauopathies

“…Methods from mathematical topology can provide more accurate measurable characterization of protein structure complexity both locally and globally 19 – 23 . In particular, methods from knot theory enable the characterization of complexity of both unknotted and knotted proteins 21 , 23 – 28 . Recent work has shown a connection between novel topological metrics and protein kinetics, namely, that experimental protein folding rates correlate with the topological structural complexity of the native state of simple, 2-state proteins without knots or slipknots 23 .…”