Protein aggregation is a feature of numerous neurodegenerative diseases. However, regulated, often reversible, formation of protein aggregates, also known as condensates, helps control a wide range of cellular activities including stress response, gene expression, memory, cell development and differentiation. This review presents examples of aggregates found in biological systems, how they are used, and cellular strategies that control aggregation and disaggregation. We include features of the aggregating proteins themselves, environmental factors, co-aggregates, post-translational modifications and well-known aggregation-directed activities that influence their formation, material state, stability and dissolution. We highlight the emerging roles of biomolecular condensates in early animal development, and disaggregation processing proteins that have recently been shown to play key roles in gametogenesis and embryogenesis.

NR4A receptors, including NUR77 (NR4A1), NURR1 (NR4A2) and NOR-1 (NR4A3), form a family of nuclear receptors that act as transcription factors to regulate many physiological and pathological processes such as cell cycle and apoptosis, lipid metabolism, inflammation, carcinogenesis, vascular and neuronal functions. In the absence of known endogenous ligand modulating their physiological functions, the NR4A family remains a class of orphan receptors. However, several post-translational modifications (PTMs), including SUMOylation, have been shown to regulate the expression and/or activity of these receptors. Addition of Small Ubiquitin-like MOdifier (SUMO) proteins is a dynamic and reversible enzymatic process that regulates multiple essential functions of proteins, including nuclear receptors. This review aims at summarizing what is known about the impact of SUMOylation on NR4A family member transcriptional activities and physiological functions.

Although RNA-binding proteins (RBPs) are known to be enriched in intrinsic disorder, no previous analysis focused on RBPs interacting with specific RNA types. We fill this gap with a comprehensive analysis of the putative disorder in RBPs binding to six common RNA types: messenger RNA (mRNA), transfer RNA (tRNA), small nuclear RNA (snRNA), non-coding RNA (ncRNA), ribosomal RNA (rRNA), and internal ribosome RNA (irRNA). We also analyze the amount of putative intrinsic disorder in the RNA-binding domains (RBDs) and non-RNA-binding-domain regions (non-RBD regions). Consistent with previous studies, we show that in comparison with human proteome, RBPs are significantly enriched in disorder. However, closer examination finds significant enrichment in predicted disorder for the mRNA-, rRNA- and snRNA-binding proteins, while the proteins that interact with ncRNA and irRNA are not enriched in disorder, and the tRNA-binding proteins are significantly depleted in disorder. We show a consistent pattern of significant disorder enrichment in the non-RBD regions coupled with low levels of disorder in RBDs, which suggests that disorder is relatively rarely utilized in the RNA-binding regions. Our analysis of the non-RBD regions suggests that disorder harbors posttranslational modification sites and is involved in the putative interactions with DNA. Importantly, we utilize experimental data from DisProt and independent data from Pfam to validate the above observations that rely on the disorder predictions. This study provides new insights into the distribution of disorder across proteins that bind different RNA types and the functional role of disorder in the regions where it is enriched.

ABCG1 is an ATP binding cassette (ABC) transporter that removes excess cholesterol from peripheral tissues. Despite its role in preventing lipid accumulation and the development of cardiovascular and metabolic disease, the mechanism underpinning ABCG1-mediated cholesterol transport is unknown. Here we report a cryo-EM structure of human ABCG1 at 4 Å resolution in an inward-open state, featuring sterol-like density in the binding cavity. Structural comparison with the multidrug transporter ABCG2 and the sterol transporter ABCG5/G8 reveals the basis of mechanistic differences and distinct substrate specificity. Benzamil and taurocholate inhibited the ATPase activity of liposome-reconstituted ABCG1, whereas the ABCG2 inhibitor Ko143 did not. Based on the structural insights into ABCG1, we propose a mechanism for ABCG1-mediated cholesterol transport.

Autonomous heavy-chain variable (VH) domains are the smallest functional antibody fragments, and they possess unique features, including small size and convex paratopes, which provide enhanced targeting of concave epitopes that are difficult to access with larger conventional antibodies. However, human VH domains have evolved to fold and function with a light chain partner, and alone, they typically suffer from low stability and high aggregation propensity. Development of autonomous human VH domains, in which aggregation propensity is reduced without compromising antigen recognition, has proven challenging. Here, we used an autonomous human VH domain as a scaffold to construct phage-displayed synthetic libraries in which aspartate was systematically incorporated at different paratope positions. In selections, the library yielded many anti-EphA1 receptor VH domains, which were characterized in detail. Structural analyses of a parental anti-EphA1 VH domain and an improved variant provided insights into the effects of aspartate and other substitutions on preventing aggregation while retaining function. Our naïve libraries and in vitro selection procedures offer a systematic approach to generating highly functional autonomous human VH domains that resist aggregation and could be used for basic research and biomedical applications.

Many of the world's most important food crops such as rice, barley and maize accumulate silicon (Si) to high levels, resulting in better plant growth and crop yields. The first step in Si accumulation is the uptake of silicic acid by the roots, a process mediated by the structurally uncharacterised NIP subfamily of aquaporins, also named metalloid porins. Here, we present the X-ray crystal structure of the archetypal NIP family member from Oryza sativa (OsNIP2;1). The OsNIP2;1 channel is closed in the crystal structure by the cytoplasmic loop D, which is known to regulate channel opening in classical plant aquaporins. The structure further reveals a novel, five-residue extracellular selectivity filter with a large diameter. Unbiased molecular dynamics simulations show a rapid opening of the channel and visualise how silicic acid interacts with the selectivity filter prior to transmembrane diffusion. Our results will enable detailed structure-function studies of metalloid porins, including the basis of their substrate selectivity.

Vertebrate organisms express a diversity of protein receptors that recognize and respond to the presence of pathogenic molecules, functioning as an early warning system for infection. As a result of mutation or dysregulated metabolism, these same innate immune receptors can be inappropriately activated, leading to inflammation and disease. One of the most important receptors for detection and response to RNA viruses is called RIG-I, and dysregulation of this protein is linked with a variety of disease states. Despite its central role in inflammatory responses, antagonists for RIG-I are underdeveloped. In this study, we use invitro selection from a pool of modified DNA aptamers to create a high affinity RIG-I antagonist. A high resolution crystal structure of the complex reveals molecular mimicry between the aptamer and the 5'-triphosphate terminus of viral ligands, which bind to the same amino acids within the CTD recognition platform of the RIG-I receptor. Our study suggests a powerful, generalizable strategy for generating immunomodulatory drugs and mechanistic tool compounds.

Type III protein secretion is widespread in Gram-negative pathogens. It comprises the injectisome with a surface-exposed needle and an inner membrane translocase. The translocase contains the SctRSTU export channel enveloped by the export gate subunit SctV that binds chaperone/exported clients and forms a putative ante-chamber. We probed the assembly, function, structure and dynamics of SctV from enteropathogenic E. coli (EPEC). In both EPEC and E. coli lab strains, SctV forms peripheral oligomeric clusters that are detergent-extracted as homo-nonamers. Membrane-embedded SctV9 is necessary and sufficient to act as a receptor for different chaperone/exported protein pairs with distinct C-domain binding sites that are essential for secretion. Negative staining electron microscopy revealed that peptidisc-reconstituted His-SctV9 forms a tripartite particle of ∼22 nm with a N-terminal domain connected by a short linker to a C-domain ring structure with a ∼5 nm-wide inner opening. The isolated C-domain ring was resolved with cryo-EM at 3.1 Å and structurally compared to other SctV homologues. Its four sub-domains undergo a three-stage "pinching" motion. Hydrogen-deuterium exchange mass spectrometry revealed this to involve dynamic and rigid hinges and a hyper-flexible sub-domain that flips out of the ring periphery and binds chaperones on and between adjacent protomers. These motions are coincident with local conformational changes at the pore surface and ring entry mouth that may also be modulated by the ATPase inner stalk. We propose that the intrinsic dynamics of the SctV protomer are modulated by chaperones and the ATPase and could affect allosterically the other subunits of the nonameric ring during secretion.

Receptor tyrosine kinases (RTK) bind growth factors and are critical for cell proliferation and differentiation. Their dysregulation leads to a loss of growth control, often resulting in cancer. Epidermal growth factor receptor (EGFR) is the prototypic RTK and can bind several ligands exhibiting distinct mitogenic potentials. Whereas the phosphorylation on individual EGFR sites and their roles for downstream signaling have been extensively studied, less is known about ligand-specific ubiquitination events on EGFR, which are crucial for signal attenuation and termination. We used a proteomics-based workflow for absolute quantitation combined with mathematical modeling to unveil potentially decisive ubiquitination events on EGFR from the first 30 seconds to 15 minutes of stimulation. Four ligands were used for stimulation: epidermal growth factor (EGF), heparin-binding-EGF like growth factor, transforming growth factor-α and epiregulin. Whereas only little differences in the order of individual ubiquitination sites were observed, the overall amount of modified receptor differed depending on the used ligand, indicating that absolute magnitude of EGFR ubiquitination, and not distinctly regulated ubiquitination sites, is a major determinant for signal attenuation and the subsequent cellular outcomes.

Protein kinase G (PKG) is a major receptor of cGMP, and controls signaling pathways distinct from those regulated by cAMP. However, the contributions of the two substituents that differentiate cGMP from cAMP (i.e. 6-oxo and 2-NH2) to the cGMP-versus-cAMP selectivity of PKG remain unclear. Here, using NMR to map how binding affinity and dynamics of the protein and ligand vary along a ligand double-substitution cycle, we show that the contributions of the two substituents to binding affinity are surprisingly non-additive. Such non-additivity stems primarily from mutual protein-ligand conformational selection, whereby not only does the ligand select for a preferred protein conformation upon binding, but also, the protein selects for a preferred ligand conformation. The 6-oxo substituent mainly controls the conformational equilibrium of the bound protein, while the 2-NH2 substituent primarily controls the conformational equilibrium of the unbound ligand (i.e. syn versus anti). Therefore, understanding the conformational dynamics of both the protein and ligand is essential to explain the cGMP-versus-cAMP selectivity of PKG.

Our poor understanding of the mechanism by which the peptide-hormone H2 relaxin activates its G protein coupled receptor, RXFP1 and the related receptor RXFP2, has hindered progress in its therapeutic development. Both receptors possess large ectodomains, which bind H2 relaxin, and contain an N-terminal LDLa module that is essential for receptor signaling and postulated to be a tethered agonist. Here, we show that a conserved motif (GDxxGWxxxF), C-terminal to the LDLa module, is critical for receptor activity. Importantly, this motif adopts different structures in RXFP1 and RXFP2, suggesting distinct activation mechanisms. For RXFP1, the motif is flexible, weakly associates with the LDLa module, and requires H2 relaxin binding to stabilize an active conformation. Conversely, the GDxxGWxxxF motif in RXFP2 is more closely associated with the LDLa module, forming an essential binding interface for H2 relaxin. These differences in the activation mechanism will aid drug development targeting these receptors.

Most eukaryotic transmembrane and secreted proteins contain N-terminal signal peptides that mediate insertion of the nascent translation products into the membrane of the endoplasmic reticulum. After membrane insertion, signal peptides typically are cleaved from the mature protein and degraded. Here, we tested whether a small hydrophobic protein selected for growth promoting activity in mammalian cells retained transforming activity while also acting as a signal peptide. We replaced the signal peptide of the PDGF β receptor (PDGFβR) with a previously described 29-residue artificial transmembrane protein named 9C3 that can activate the PDGFβR in trans. We showed that a modified version of 9C3 at the N-terminus of the PDGFβR can function as a signal peptide, as assessed by its ability to support high level expression, glycosylation, and cell surface localization of the PDGFβR. The 9C3 signal peptide retains its ability to interact with the transmembrane domain of the PDGFβR and cause receptor activation and cell proliferation. Cleavage of the 9C3 signal peptide from the mature receptor is not required for these activities. However, signal peptide cleavage does occur in some molecules, and the cleaved signal peptide can persist in cells and activate a co-expressed PDGFβR in trans. Our finding that a hydrophobic sequence can display signal peptide and transforming activity suggest that some naturally occurring signal peptides may also display additional biological activities by interacting with the transmembrane domains of target proteins.

Retinoblastoma-binding protein 1 (RBBP1) is involved in gene regulation, epigenetic regulation, and disease processes. RBBP1 contains five domains with DNA-binding or histone-binding activities, but how RBBP1 specifically recognizes chromatin is still unknown. An AT-rich interaction domain (ARID) in RBBP1 was proposed to be the key region for DNA-binding and gene suppression. Here, we first determined the solution structure of a tandem PWWP-ARID domain mutant of RBBP1 after deletion of a long flexible acidic loop L12 in the ARID domain. NMR titration results indicated that the ARID domain interacts with DNA with no GC- or AT-rich preference. Surprisingly, we found that the loop L12 binds to the DNA-binding region of the ARID domain as a DNA mimic and inhibits DNA binding. The loop L12 can also bind weakly to the Tudor and chromobarrel domains of RBBP1, but binds more strongly to the DNA-binding region of the histone H2A-H2B heterodimer. Furthermore, both the loop L12 and DNA can enhance the binding of the chromobarrel domain to H3K4me3 and H4K20me3. Based on these results, we propose a model of chromatin recognition by RBBP1, which highlights the unexpected multiple key roles of the disordered acidic loop L12 in the specific binding of RBBP1 to chromatin.

Protein modification by ubiquitin or SUMO can alter the function, stability or activity of target proteins. Previous studies have identified thousands of substrates that were modified by ubiquitin or SUMO on the same lysine residue. However, it remains unclear whether such overlap could result from a mere higher solvent accessibility, whether proteins containing those sites are associated with specific functional traits, and whether selectively perturbing their modification by ubiquitin or SUMO could result in different phenotypic outcomes. Here, we mapped reported lysine modification sites across the human proteome and found an enrichment of sites reported to be modified by both ubiquitin and SUMO. Our analysis uncovered thousands of proteins containing such sites, which we term Sites of Alternative Modification (SAMs). Among more than 36,000 sites reported to be modified by SUMO, 51.8% have also been reported to be modified by ubiquitin. SAM-containing proteins are associated with diverse biological functions including cell cycle, DNA damage, and transcriptional regulation. As such, our analysis highlights numerous proteins and pathways as putative targets for further elucidating the crosstalk between ubiquitin and SUMO. Comparing the biological and biochemical properties of SAMs versus other non-overlapping modification sites revealed that these sites were associated with altered cellular localization or abundance of their host proteins. Lastly, using S. cerevisiae as model, we show that mutating the SAM motif in a protein can influence its ubiquitination as well as its localization and abundance.

Converging evidence points to the N-terminal domain comprising the first 17 amino acids of the Huntingtin protein (Nt17) as a key regulator of its aggregation, cellular properties and toxicity. In this study, we further investigated the interplay between Nt17 and the polyQ domain repeat length in regulating the aggregation and inclusion formation of exon 1 of the Huntingtin protein (Httex1). In addition, we investigated the effect of removing Nt17 or modulating its local structure on the membrane interactions, neuronal uptake, and toxicity of monomeric or fibrillar Httex1. Our results show that the polyQ and Nt17 domains synergistically modulate the aggregation propensity of Httex1 and that the Nt17 domain plays important roles in shaping the surface properties of mutant Httex1 fibrils and regulating their poly-Q-dependent growth, lateral association and neuronal uptake. Removal of Nt17 or disruption of its transient helical conformations slowed the aggregation of monomeric Httex1 in vitro, reduced inclusion formation in cells, enhanced the neuronal uptake and nuclear accumulation of monomeric Httex1 proteins, and was sufficient to prevent cell death induced by Httex1 72Q overexpression. Finally, we demonstrate that the uptake of Httex1 fibrils into primary neurons and the resulting toxicity are strongly influenced by mutations and phosphorylation events that influence the local helical propensity of Nt17. Altogether, our results demonstrate that the Nt17 domain serves as one of the key master regulators of Htt aggregation, internalization, and toxicity and represents an attractive target for inhibiting Htt aggregate formation, inclusion formation, and neuronal toxicity.

We analyse paths through the regulatory networks that control gene-expression patterns in Yeast, in five different physiological states: cell cycle, DNA damage, stress response, diauxic shift, and sporulation. The network in each state is specified as a directed graph, containing different sets of edges connecting pairs selected from a combined set of 1475 nodes. Each network contains some nodes that have no parents, and others that have no children. We call these, respectively, 'source' and 'sink' nodes. For each network we enumerate paths between source and sink nodes. In a previous paper (Lesk and Konagurthu, 2020), we defined, extracted and compared the neighbourhoods of each transcription factor in different physiological states, and how the system reconfigures itself. Here we compare the usage of nodes and edges by different networks, and how they are assembled into paths. The picture that emerges is that the networks are not disjoint but show substantial sharing of nodes and edges; however, they assemble these materials into different sets of paths. Four of the networks, other than the cell-cycle network, contain paths between only a small fraction (<13%) of possible source-sink pairs. Although the cell-cycle network is not an outlier in terms of total number of nodes and edges, and number of sink nodes, it is very much an outlier in having a greater proportion of source-to-sink paths than the other networks.

Accurate predictions of the three-dimensional structures of proteins from their amino acid sequences have come of age. AlphaFold, a deep learning-based approach to protein structure prediction, shows remarkable success in independent assessments of prediction accuracy. A significant epoch in structural bioinformatics was the structural annotation of over 98% of protein sequences in the human proteome. Interestingly, many predictions feature regions of very low confidence, and these regions largely overlap with intrinsically disordered regions (IDRs). That over 30% of regions within the proteome are disordered is congruent with estimates that have been made over the past two decades, as intense efforts have been undertaken to generalize the structure-function paradigm to include the importance of conformational heterogeneity and dynamics. With structural annotations from AlphaFold in hand, there is the temptation to draw inferences regarding the "structures" of IDRs and their interactomes. Here, we offer a cautionary note regarding the misinterpretations that might ensue and highlight efforts that provide concrete understanding of sequence-ensemble-function relationships of IDRs. This perspective is intended to emphasize the importance of IDRs in sequence-function relationships (SERs) and to highlight how one might go about extracting quantitative SERs to make sense of how IDRs function.

Stunning advances have been achieved in addressing the protein folding problem, providing deeper understanding of the mechanisms by which proteins navigate energy landscapes to reach their native states and enabling powerful algorithms to connect sequence to structure. However, the realities of the in vivo protein folding problem remain a challenge to reckon with. Here, we discuss the concept of the "proteome folding problem"-the problem of how organisms build and maintain a functional proteome-by admitting that folding energy landscapes are characterized by many misfolded states and that cells must deploy a network of chaperones and degradation enzymes to minimize deleterious impacts of these off-pathway species. The resulting proteostasis network is an inextricable part of in vivo protein folding and must be understood in detail if we are to solve the proteome folding problem. We discuss how the development of computational models for the proteostasis network's actions and the relationship to the biophysical properties of the proteome has begun to offer new insights and capabilities.

Traditional sequence analysis algorithms fail to identify distant homologies when they lie beyond a detection horizon. In this review, we discuss how co-evolution-based contact and distance prediction methods are pushing back this homology detection horizon, thereby yielding new functional insights and experimentally testable hypotheses. Based on correlated substitutions, these methods divine three-dimensional constraints among amino acids in protein sequences that were previously devoid of all annotated domains and repeats. The new algorithms discern hidden structure in an otherwise featureless sequence landscape. Their revelatory impact promises to be as profound as the use, by archaeologists, of ground-penetrating radar to discern long-hidden, subterranean structures. As examples of this, we describe how triplicated structures reflecting longin domains in MON1A-like proteins, or UVR-like repeats in DISC1, emerge from their predicted contact and distance maps. These methods also help to resolve structures that do not conform to a "beads-on-a-string" model of protein domains. In one such example, we describe CFAP298 whose ubiquitin-like domain was previously challenging to perceive owing to a large sequence insertion within it. More generally, the new algorithms permit an easier appreciation of domain families and folds whose evolution involved structural insertion or rearrangement. As we exemplify with α1-antitrypsin, coevolution-based predicted contacts may also yield insights into protein dynamics and conformational change. This new combination of structure prediction (using innovative co-evolution based methods) and homology inference (using more traditional sequence analysis approaches) shows great promise for bringing into view a sea of evolutionary relationships that had hitherto lain far beyond the horizon of homology detection.

Protein function is fundamentally reliant on inter-molecular interactions that underpin the ability of proteins to form complexes driving biological processes in living cells. Increasingly, such interactions are recognised as being formed between proteins that exist on a broad spectrum of dynamic conformational states and levels of intrinsic disorder. Additionally, the sizes of the structures formed can range from simple binary complexes to large dynamic biomolecular condensates measuring 100 nm or more. Understanding the parameters that govern such interactions, how they form, how they lead to function and what happens when they take place in unintended manners and lead to disease, represent some of the core questions for molecular biosciences. In light of recent advances made in solving the protein folding problem by machine learning methods, we discuss here the challenges and opportunities brought by these new data-driven approaches for the next frontiers of biomolecular science.

After decades of progress in computational protein design, the design of proteins folding and functioning in lipid membranes appears today as the next frontier. Some notable successes in the de novo design of simplified model membrane protein systems have helped articulate fundamental principles of protein folding, architecture and interaction in the hydrophobic lipid environment. These principles are reviewed here, together with the computational methods and approaches that were used to identify them. We provide an overview of the methodological innovations in the generation of new protein structures and functions and in the development of membrane-specific energy functions. We highlight the opportunities offered by new machine learning approaches applied to protein design, and by new experimental characterization techniques applied to membrane proteins. Although membrane protein design is in its infancy, it appears more reachable than previously thought.

The use of force probes to induce unfolding and refolding of single molecules through the application of mechanical tension, known as single-molecule force spectroscopy (SMFS), has proven to be a powerful tool for studying the dynamics of protein folding. Here we provide an overview of what has been learned about protein folding using SMFS, from small, single-domain proteins to large, multi-domain proteins. We highlight the ability of SMFS to measure the energy landscapes underlying folding, to map complex pathways for native and non-native folding, to probe the mechanisms of chaperones that assist with native folding, to elucidate the effects of the ribosome on co-translational folding, and to monitor the folding of membrane proteins.

Traditionally, structural biologists approach the complexity of cellular proteomes in a reductionist manner. Proteomes are fractionated, their molecular components purified and studied one-by-one using the experimental methods for structure determination at their disposal. Visual proteomics aims at obtaining a holistic picture of cellular proteomes by studying them in situ, ideally in unperturbed cellular environments. The method that enables doing this at highest resolution is cryo-electron tomography. It allows to visualize cellular landscapes with molecular resolution generating maps or atlases revealing the interaction networks which underlie cellular functions in health and in disease states. Current implementations of cryo ET do not yet realize the full potential of the method in terms of resolution and interpretability. To this end, further improvements in technology and methodology are needed. This review describes the state of the art as well as measures which we expect will help overcoming current limitations.

The need to make sense of the thousands of genetic variants uncovered every day in terms of pathology or biological mechanism is acute. Many insights into how genetic changes impact protein function can be gleaned if three-dimensional structures of the associated proteins are available. The availability of a highly accurate method of predicting structures from amino acid sequences (e.g. Alphafold2) is thus potentially a great boost to those wanting to understand genetic changes. In this paper we discuss the current state of protein structures known for the human and other proteomes and how Alphafold2 might impact on variant interpretation efforts. For the human proteome in particular, the state of the available structural data suggests that the impact on variant interpretation might be less than anticipated. We also discuss additional efforts in structure prediction that could further aid the understanding of genetic variants.

The prediction of highly ordered three-dimensional structures of amyloid protein fibrils from the amino acid sequences of their monomeric self-assembly precursors constitutes a challenging and unresolved aspect of the classical protein folding problem. Because of the polymorphic nature of amyloid assembly whereby polypeptide chains of identical amino acid sequences under identical conditions are capable of self-assembly into a spectrum of different fibril structures, the prediction of amyloid structures from an amino acid sequence requires a detailed and holistic understanding of its assembly free energy landscape. The full extent of the structure space accessible to the cross-β molecular architecture of amyloid must also be resolved. Here, we review the current understanding of the diversity and the individuality of amyloid structures, and how the polymorphic landscape of amyloid links to biology and disease phenotypes. We present a comprehensive review of structural models of amyloid fibrils derived by cryo-EM, ssNMR and AFM to date, and discuss the challenges ahead for resolving the structural basis and the biological consequences of polymorphic amyloid assemblies.

The protein folding problem was first articulated as question of how order arose from disorder in proteins: How did the various native structures of proteins arise from interatomic driving forces encoded within their amino acid sequences, and how did they fold so fast? These matters have now been largely resolved by theory and statistical mechanics combined with experiments. There are general principles. Chain randomness is overcome by solvation-based codes. And in the needle-in-a-haystack metaphor, native states are found efficiently because protein haystacks (conformational ensembles) are funnel-shaped. Order-disorder theory has now grown to encompass a large swath of protein physical science across biology.

Characterizing the three-dimensional structure of macromolecules is central to understanding their function. Traditionally, structures of proteins and their complexes have been determined using experimental techniques such as X-ray crystallography, NMR, or cryo-electron microscopy-applied individually or in an integrative manner. Meanwhile, however, computational methods for protein structure prediction have been improving their accuracy, gradually, then suddenly, with the breakthrough advance by AlphaFold2, whose models of monomeric proteins are often as accurate as experimental structures. This breakthrough foreshadows a new era of computational methods that can build accurate models for most monomeric proteins. Here, we envision how such accurate modeling methods can combine with experimental structural biology techniques, enhancing integrative structural biology. We highlight the challenges that arise when considering multiple structural conformations, protein complexes, and polymorphic assemblies. These challenges will motivate further developments, both in modeling programs and in methods to solve experimental structures, towards better and quicker investigation of structure-function relationships.

The announcement of the outstanding performance of AlphaFold 2 in the CASP 14 protein structure prediction competition came at the end of a long year defined by the COVID-19 pandemic. With an infectious organism dominating the world stage, the developers of Alphafold 2 were keen to play their part, accurately predicting novel structures of two proteins from SARS-CoV-2. In their blog post of December 2020, they highlighted this contribution, writing "we've also seen signs that protein structure prediction could be useful in future pandemic response efforts". So, what role does structural biology play in guiding vaccine immunogen design and what might be the contribution of AlphaFold 2?

I outline how over my career as a protein scientist Machine Learning has impacted my area of science and one of my pastimes, chess, where there are some interesting parallels. In 1968, modelling of three-dimensional structures was initiated based on a known structure as a template, the problem of the pathway of protein folding was posed and bets were taken in the emerging field of Machine Learning on whether computers could outplay humans at chess. Half a century later, Machine Learning has progressed from using computational power combined with human knowledge in solving problems to playing chess without human knowledge being used, where it has produced novel strategies. Protein structures are being solved by Machine Learning based on human-derived knowledge but without templates. There is much promise that programs like AlphaFold based on Machine Learning will be powerful tools for designing entirely novel protein folds and new activities. But, will they produce novel ideas on protein folding pathways and provide new insights into the principles that govern folds?

Recent progress in structure-prediction methods that rely on deep learning suggests that the atomic structure of almost any protein may soon be predictable directly from its amino acid sequence. This much-awaited revolution was driven by substantial improvements in the reliability of methods for inferring the spatial distances between amino acid pairs from an analysis of homologous sequences. Improved reliability has been accompanied, however, by a reduced ability to detect amino acid relationships that are not due to direct spatial contacts, such as those that arise from protein dynamics or allostery. Given the central importance of dynamics and allostery to protein activity, we argue that an important future advance would extend modeling beyond predicting a single static structure. Here, we briefly review some of the developments that have led to the remarkable recent achievement in structure prediction and speculate what methods and sources of information may be leveraged in the future to develop a modeling framework that addresses protein dynamics and allostery.

Protein aggregation is a widespread phenomenon with important implications in many scientific areas. Although amyloid formation is typically considered as detrimental, functional amyloids that perform physiological roles have been identified in all kingdoms of life. Despite their functional and pathological relevance, the structural details of the majority of molecular species involved in the amyloidogenic process remains elusive. Here, we explore the application of AlphaFold, a highly accurate protein structure predictor, in the field of protein aggregation. While we envision a straightforward application of AlphaFold in assisting the design of globular proteins with improved solubility for biomedical and industrial purposes, the use of this algorithm for predicting the structure of aggregated species seems far from trivial. First, in amyloid diseases, the presence of multiple amyloid polymorphs and the heterogeneity of aggregation intermediates challenges the "one sequence, one structure" paradigm, inherent to sequence-based predictions. Second, aberrant aggregation is not the subject of positive selective pressure, precluding the use of evolutionary-based approaches, which are the core of the AlphaFold pipeline. Instead, amyloid polymorphism seems to be constrained by the need for a defined structure-activity relationship in functional amyloids. They may thus provide a starting point for the application of AlphaFold in the amyloid landscape.

Intrinsically disordered proteins (IDPs) constitute a broad set of proteins with few uniting and many diverging properties. IDPs-and intrinsically disordered regions (IDRs) interspersed between folded domains-are generally characterized as having no persistent tertiary structure; instead they interconvert between a large number of different and often expanded structures. IDPs and IDRs are involved in an enormously wide range of biological functions and reveal novel mechanisms of interactions, and while they defy the common structure-function paradigm of folded proteins, their structural preferences and dynamics are important for their function. We here discuss open questions in the field of IDPs and IDRs, focusing on areas where machine learning and other computational methods play a role. We discuss computational methods aimed to predict transiently formed local and long-range structure, including methods for integrative structural biology. We discuss the many different ways in which IDPs and IDRs can bind to other molecules, both via short linear motifs, as well as in the formation of larger dynamic complexes such as biomolecular condensates. We discuss how experiments are providing insight into such complexes and may enable more accurate predictions. Finally, we discuss the role of IDPs in disease and how new methods are needed to interpret the mechanistic effects of genomic variants in IDPs.

The protein folding problem was apparently solved recently by the advent of a deep learning method for protein structure prediction called AlphaFold. However, this program is not able to make predictions about the protein folding pathways. Moreover, it only treats about half of the human proteome, as the remaining proteins are intrinsically disordered or contain disordered regions. By definition these proteins differ from natively folded proteins and do not adopt a properly folded structure in solution. However these intrinsically disordered proteins (IDPs) also systematically differ in amino acid composition and uniquely often become folded upon binding to an interaction partner. These factors preclude solving IDP structures by current machine-learning methods like AlphaFold, which also cannot solve the protein aggregation problem, since this meta-folding process can give rise to different aggregate sizes and structures. An alternative computational method is provided by molecular dynamics simulations that already successfully explored the energy landscapes of IDP conformational switching and protein aggregation in multiple cases. These energy landscapes are very different from those of 'simple' protein folding, where one energy funnel leads to a unique protein structure. Instead, the energy landscapes of IDP conformational switching and protein aggregation feature a number of minima for different competing low-energy structures. In this review, I discuss the characteristics of these multifunneled energy landscapes in detail, illustrated by molecular dynamics simulations that elucidated the underlying conformational transitions and aggregation processes.

The axon initial segment (AIS) is a distinct neuronal domain, which is responsible for initiating action potentials, and therefore of key importance to neuronal signaling. To determine how it functions, it is necessary to establish which proteins reside there, how they are organized, and what the dynamic features are. Great strides have been made in recent years, and it is now clear that several AIS cytoskeletal and membrane proteins interact to form a higher-order periodic structure. Here we briefly describe AIS function, protein composition and molecular architecture, and discuss perspectives for future structural characterization, and if structure predictions will be able to model complex higher-order assemblies.

Protein design has come of age, but how will it mature? In the 1980s and the 1990s, the primary motivation for de novo protein design was to test our understanding of the informational aspect of the protein-folding problem; i.e., how does protein sequence determine protein structure and function? This necessitated minimal and rational design approaches whereby the placement of each residue in a design was reasoned using chemical principles and/or biochemical knowledge. At that time, though with some notable exceptions, the use of computers to aid design was not widespread. Over the past two decades, the tables have turned and computational protein design is firmly established. Here, I illustrate this progress through a timeline of de novo protein structures that have been solved to atomic resolution and deposited in the Protein Data Bank. From this, it is clear that the impact of rational and computational design has been considerable: More-complex and more-sophisticated designs are being targeted with many being resolved to atomic resolution. Furthermore, our ability to generate and manipulate synthetic proteins has advanced to a point where they are providing realistic alternatives to natural protein functions for applications both in vitro and in cells. Also, and increasingly, computational protein design is becoming accessible to non-specialists. This all begs the questions: Is there still a place for minimal and rational design approaches? And, what challenges lie ahead for the burgeoning field of de novo protein design as a whole?