Intrinsically disordered proteins (IDP) are important in a broad range of biological functions and are involved in many diseases. An understanding of intrinsic disorder is key to develop drugs against IDPs. Experimental characterization of IDPs are expensive and less efficient and demand the development of computational tools. Here, we present ADOPT, a new predictor of protein disorder. ADOPT is a deep bidirectional transformer, which extracts dense residue level representations from Facebook’s Evolutionary Scale Modeling (ESM) library. Using the experimentally designed CheZod database as a training and test dataset for protein disorder, it predicts Z scores and protein disorder with new state-of-the-art performance in a few seconds. We show that ADOPT offers substantial improvement in comparison to previous predictors with a Spearman correlation coefficient between experimental and computational Z scores of 0.69. We identify the coordinates which are relevant for the prediction performance and show that good performance can already gained with less than 100 features. We believe that ADOPT will be a useful tool for all experimental scientists working with intrinsically disordered proteins. It is available as a standalone package at https://github.com/PeptoneInc/ADOPT.git.

Bacterial flagella are reversible rotary motors that rotate external filaments for bacterial propulsion. Some flagellar motors have diversified by recruiting additional components that influence torque and rotation, but little is known about the possible diversification and evolution of core motor components. The mechanistic core of flagella is the cytoplasmic C ring, which functions as a rotor, directional switch, and assembly platform for the flagellar type III secretion system (fT3SS) ATPase. The C ring is composed of a ring of FliG proteins and a helical ring of surface presentation of antigen (SPOA) domains from the switch proteins FliM and one of two usually mutually exclusive paralogs, FliN or FliY. We investigated the composition, architecture, and function of the C ring of Campylobacter jejuni, which encodes FliG, FliM, and both FliY and FliN by a variety of interrogative approaches. We discovered a diversified C. jejuni C ring containing FliG, FliM, and both FliY, which functions as a classical FliN-like protein for flagellar assembly, and FliN, which has neofunctionalized into a structural role. Specific protein interactions drive the formation of a more complex heterooligomeric C. jejuni C-ring structure. We discovered that this complex C ring has additional cellular functions in polarly localizing FlhG for numerical regulation of flagellar biogenesis and spatial regulation of division. Furthermore, mutation of the C. jejuni C ring revealed a T3SS that was less dependent on its ATPase complex for assembly than were other systems. Our results highlight considerable evolved flagellar diversity that impacts motor output, biogenesis, and cellular processes in different species. IMPORTANCE The conserved core of bacterial flagellar motors reflects a shared evolutionary history that preserves the mechanisms essential for flagellar assembly, rotation, and directional switching. In this work, we describe an expanded and diversified set of core components in the Campylobacter jejuni flagellar C ring, the mechanistic core of the motor. Our work provides insight into how usually conserved core components may have diversified by gene duplication, enabling a division of labor of the ancestral protein between the two new proteins, acquisition of new roles in flagellar assembly and motility, and expansion of the function of the flagellum beyond motility, including spatial regulation of cell division and numerical control of flagellar biogenesis in C. jejuni. Our results highlight that relatively small changes, such as gene duplications, can have substantial ramifications on the cellular roles of a molecular machine.

The ability to design and construct structures with atomic level precision is one of the key goals of nanotechnology. Proteins offer an attractive target for atomic design because they can be synthesized chemically or biologically and can self-assemble. However, the generalized protein folding and design problem is unsolved. One approach to simplifying the problem is to use a repetitive protein as a scaffold. Repeat proteins are intrinsically modular, and their folding and structures are better understood than large globular domains. Here, we have developed a class of synthetic repeat proteins based on the pentapeptide repeat family of beta-solenoid proteins. We have constructed length variants of the basic scaffold and computationally designed de novo loops projecting from the scaffold core. The experimentally solved 3.56-Å resolution crystal structure of one designed loop matches closely the designed hairpin structure, showing the computational design of a backbone extension onto a synthetic protein core without the use of backbone fragments from known structures. Two other loop designs were not clearly resolved in the crystal structures, and one loop appeared to be in an incorrect conformation. We have also shown that the repeat unit can accommodate whole-domain insertions by inserting a domain into one of the designed loops.computational protein design | synthetic repeat proteins | de novo backbone design | coarse-grained model D uring the course of evolution, natural proteins may be recruited to new unrelated functions conferring a selective advantage to the organism (1, 2). This accretion of new features and functions is likely to have left behind complex interlocking amino acid dependencies that can make reengineering natural proteins difficult and unpredictable (3). For this reason, we and others hypothesize that it is more desirable to design de novo proteins because these proteins provide a biologically neutral platform onto which functional elements can be grafted (4). Artificial proteins have been designed by decoding simple residue patterning rules that govern the packing of secondary structural elements, and this technique has been particularly successful for α-helical bundle proteins (5-7). An alternative approach is to assemble de novo folds from backbone fragments of known structures or idealized secondary structural elements and use computational protein design methods to design the sequence (4,(8)(9)(10). Both the computational and simpler rules-based design approaches have concentrated on designing proteins consisting of canonical secondary structure linked with loops of minimal length.A class of proteins that has attracted considerable interest is artificial proteins based on repeating structural motifs due to their intrinsic modularity and designability (11). Repeat proteins have applications that include their use as novel nanomaterials (12-14) and as scaffolds for molecular recognition (15, 16). These proteins may be designed using sequence consensus-based rules (17) or computational prot...

Intrinsically disordered proteins (IDPs) are important for a broad range of biological functions and are involved in many diseases. An understanding of intrinsic disorder is key to develop compounds that target IDPs. Experimental characterization of IDPs is hindered by the very fact that they are highly dynamic. Computational methods that predict disorder from the amino acid sequence have been proposed. Here, we present ADOPT (Attention DisOrder PredicTor), a new predictor of protein disorder. ADOPT is composed of a self-supervised encoder and a supervised disorder predictor. The former is based on a deep bidirectional transformer, which extracts dense residue-level representations from Facebook’s Evolutionary Scale Modeling library. The latter uses a database of nuclear magnetic resonance chemical shifts, constructed to ensure balanced amounts of disordered and ordered residues, as a training and a test dataset for protein disorder. ADOPT predicts whether a protein or a specific region is disordered with better performance than the best existing predictors and faster than most other proposed methods (a few seconds per sequence). We identify the features that are relevant for the prediction performance and show that good performance can already be gained with <100 features. ADOPT is available as a stand-alone package at https://github.com/PeptoneLtd/ADOPT and as a web server at https://adopt.peptone.io/.

Electron cryo-tomography and subtomogram averaging enable visualization of protein complexes in situ, in three dimensions, in a near-native frozen-hydrated state to nanometer resolutions. To achieve this, intact cells are vitrified and imaged over a range of tilts within an electron microscope. These images can subsequently be reconstructed into a three-dimensional volume representation of the sample cell. Because complexes are visualized in situ, crucial insights into their mechanism, assembly process, and dynamic interactions with other proteins become possible. To illustrate the electron cryo-tomography workflow for visualizing protein complexes in situ, we describe our workflow of preparing samples, imaging, and image processing using Leginon for data collection, IMOD for image reconstruction, and PEET for subtomogram averaging.