Function-guided protein design by deep manifold sampling

Protein design is challenging because it requires searching through a vast combinatorial space that is only sparsely functional. Self-supervised learning approaches offer the potential to navigate through this space more effectively and thereby accelerate protein engineering. We introduce a sequence denoising autoencoder (DAE) that learns the manifold of protein sequences from a large amount of potentially unlabelled proteins. This DAE is combined with a function predictor that guides sampling towards sequences with higher levels of desired functions. We train the sequence DAE on more than 20M unlabeled protein sequences spanning many evolutionarily diverse protein families and train the function predictor on approximately 0.5M sequences with known function labels. At test time, we sample from the model by iteratively denoising a sequence while exploiting the gradients from the function predictor. We present a few preliminary case studies of protein design that demonstrate the effectiveness of this proposed approach, which we refer to as "deep manifold sampling", including metal binding site addition, function-preserving diversification, and global fold change.

Topological constraints of RNA pseudoknotted and loop-kissing motifs: applications to three-dimensional structure prediction

An RNA global fold can be described at the level of helix orientations and relatively flexible loop conformations that connect the helices. The linkage between the helices plays an essential role in determining the structural topology, which restricts RNA local and global folds, especially for RNA tertiary structures involving cross-linked base pairs. We quantitatively analyze the topological constraints on RNA 3D conformational space, in particular, on the distribution of helix orientations, for pseudoknots and loop-loop kissing structures. The result shows that a viable conformational space is predominantly determined by the motif type, helix size, and loop size, indicating a strong topological coupling between helices and loops in RNA tertiary motifs. Moreover, the analysis indicates that (cross-linked) tertiary contacts can cause much stronger topological constraints on RNA global fold than non-cross-linked base pairs. Furthermore, based on the topological constraints encoded in the 2D structure and the 3D templates, we develop a 3D structure prediction approach. This approach can be further combined with structure probing methods to expand the capability of computational prediction for large RNA folds.

In vitro 0N4R tau fibrils contain a monomorphic β-sheet core enclosed by dynamically heterogeneous fuzzy coat segments

Misfolding of the microtubule-binding protein tau into filamentous aggregates is characteristic of many neurodegenerative diseases such as Alzheimer’s disease and progressive supranuclear palsy. Determining the structures and dynamics of these tau fibrils is important for designing inhibitors against tau aggregation. Tau fibrils obtained from patient brains have been found by cryo-electron microscopy to adopt disease-specific molecular conformations. However, in vitro heparin-fibrillized 2N4R tau, which contains all four microtubule-binding repeats (4R), was recently found to adopt polymorphic structures. Here we use solid-state NMR spectroscopy to investigate the global fold and dynamics of heparin-fibrillized 0N4R tau. A single set of 13C and 15N chemical shifts was observed for residues in the four repeats, indicating a single β-sheet conformation for the fibril core. This rigid core spans the R2 and R3 repeats and adopts a hairpin-like fold that has similarities to but also clear differences from any of the polymorphic 2N4R folds. Obtaining a homogeneous fibril sample required careful purification of the protein and removal of any proteolytic fragments. A variety of experiments and polarization transfer from water and mobile side chains indicate that 0N4R tau fibrils exhibit heterogeneous dynamics: Outside the rigid R2–R3 core, the R1 and R4 repeats are semirigid even though they exhibit β-strand character and the proline-rich domains undergo large-amplitude anisotropic motions, whereas the two termini are nearly isotropically flexible. These results have significant implications for the structure and dynamics of 4R tau fibrils in vivo.

Solution structure of TbTFIIS2-2 PWWP domain from Trypanosoma brucei and its binding to H4K17me3 and H3K32me3

Posttranslational modifications (PTMs) of core histones, such as histone methylation, play critical roles in a variety of biological processes including transcription regulation, chromatin condensation and DNA repair. In T. brucei, no domain recognizing methylated histone has been identified so far. TbTFIIS2-2, as a potential transcription elongation factors in T. brucei, contains a PWWP domain in the N-terminus which shares low sequence similarity compared with other PWWP domains and is absent from other TFIIS factors. In the present study, the solution structure of TbTFIIS2-2 PWWP domain was determined by NMR spectroscopy. TbTFIIS2-2 PWWP domain adopts a global fold containing a five-strand β-barrel and two C-terminal α-helices similar to other PWWP domains. Moreover, through systematic screening, we revealed that TbTFIIS2-2 PWWP domain is able to bind H4K17me3 and H3K32me3. Meanwhile, we identified the critical residues responsible for the binding ability of TbTFIIS2-2 PWWP domain. The conserved cage formed by the aromatic amino acids in TbTFIIS2-2 PWWP domain is essential for its binding to methylated histones.

Rational design of proteins that exchange on functional timescales

Proteins are intrinsically dynamic molecules that can exchange between multiple conformational states, enabling them to carry out complex molecular processes with extreme precision and efficiency. Attempts to design novel proteins with tailored functions have mostly failed to yield efficiencies matching those found in nature because standard methods do not allow for the design of exchange between necessary conformational states on a functionally-relevant timescale. Here, we develop a broadly-applicable computational method to engineer protein dynamics that we term meta-multistate design. We used this methodology to design spontaneous exchange between two novel conformations introduced into the global fold of Streptococcal protein G domain β1. The designed proteins, named DANCERs, for Dynamic And Native Conformational ExchangeRs, are stably folded and exchange between predicted conformational states on the millisecond timescale. The successful introduction of defined dynamics on functional timescales opens the door to new applications requiring a protein to spontaneously access multiple conformational states.