An atlas of amyloid aggregation: the impact of substitutions, insertions, deletions and truncations on amyloid beta fibril nucleation

Distinct Aggregation Behavior of N-Terminally Truncated Aβ4-42 Over Aβ1-42 in the Presence of Zn(II)

The deposition of amyloid-β (Aβ) aggregates and metal ions within senile plaques is a hallmark of Alzheimer's disease (AD). Among the modifications observed in Aβ peptides, N-terminal truncation at Phe4, yielding Aβ4-x, is highly prevalent in AD-affected brains and significantly alters Aβ's metal-binding and aggregation profiles. Despite the abundance of Zn(II) in senile plaques, its impact on the aggregation and toxicity of Aβ4-x remains unexplored. Here, we report the distinct aggregation behavior of N-terminally truncated Aβ, specifically Aβ4-42, in the absence and presence of either Zn(II), Aβ seeds, or both, and compare it to that of full-length Aβ1-42. Our findings reveal notable differences in the aggregation profiles of Aβ4-42 and Aβ1-42, largely influenced by their different Zn(II)-binding properties. These results provide insights into the mechanisms underlying the distinct aggregation behavior of truncated and full-length Aβ in the presence of Zn(II), contributing to a deeper understanding of AD pathology.

Systematic characterization of indel variants using a yeast-based protein folding sensor

Gene variants resulting in insertions or deletions of amino acid residues (indels) have important consequences for evolution and are often linked to disease, yet, compared to missense variants, the effects of indels are poorly understood and predicted. We developed a sensitive protein folding sensor based on the complementation of uracil auxotrophy in yeast by circular permutated orotate phosphoribosyltransferase (CPOP). The sensor reports on the folding of disease-linked missense variants and de-novo-designed proteins. Applying the folding sensor to a saturated library of single-residue indels in human dihydrofolate reductase (DHFR) revealed that most regions that tolerate indels are confined to internal loops, the termini, and a central α helix. Several indels are temperature sensitive, and folding is rescued upon binding to methotrexate. Rosetta and AlphaFold2 predictions correlate with the observed effects, suggesting that most indels destabilize the native fold and that these computational tools are useful for the classification of indels observed in population sequencing.

Oxidative stress via UVC irradiation on the structural rearrangement of hen egg white lysozyme

Oxidative stress is a physiological condition where oxygen radicals are responsible for the conformational restructuring and loss of functionality of important biomacromolecules. Among the various external agents, UV irradiation is one of the sources that can induce oxidative stress. Here, we report an in vitro study to gauge the effect of ROS on the structural rearrangement of hen egg white lysozyme, a hydrolytic enzyme, via UVC exposure studied via various biophysical techniques. The investigations revealed a rise in the β-sheet content of the protein at the expense of a decrease in α-helix within ten minutes of exposure, thereby showing rapid changes in the secondary structure. While the unexposed sample showed partial reversibility after being subjected to a heating and cooling cycle, the newly formed structures via irradiation, on the other hand, were found to be more thermally stable. The aging of the samples via UVC exposure was reflected in both the UV-vis and PL spectra of the samples, as well as the loss of spectral features in the aliphatic and aromatic regions in the magnetic resonance spectrum. Finally, the increase in the hydrodynamic diameter of the samples shows cross-linking taking place due to the generated oxygen radicals.

Site-saturation mutagenesis of 500 human protein domains

Missense variants that change the amino acid sequences of proteins cause one-third of human genetic diseases1. Tens of millions of missense variants exist in the current human population, and the vast majority of these have unknown functional consequences. Here we present a large-scale experimental analysis of human missense variants across many different proteins. Using DNA synthesis and cellular selection experiments we quantify the effect of more than 500,000 variants on the abundance of more than 500 human protein domains. This dataset reveals that 60% of pathogenic missense variants reduce protein stability. The contribution of stability to protein fitness varies across proteins and diseases and is particularly important in recessive disorders. We combine stability measurements with protein language models to annotate functional sites across proteins. Mutational effects on stability are largely conserved in homologous domains, enabling accurate stability prediction across entire protein families using energy models. Our data demonstrate the feasibility of assaying human protein variants at scale and provides a large consistent reference dataset for clinical variant interpretation and training and benchmarking of computational methods.

Structural and energetic analysis of stabilizing indel mutations

Amino acid insertions and deletions (indels) are among the most common protein mutations and necessitate changes to a protein's backbone geometry. Examining how indels affect protein folding stability (and especially how indels can increase stability) can help reveal the role of backbone energetics on stability and introduce new protein engineering strategies. Tsuboyama et al. measured folding stability for 57,698 single amino acid insertion or deletion mutants in 405 small domains, and this analysis identified 103 stabilizing mutants (ΔΔGunfolding > 1 kcal/mol). Here, we use computational modeling to analyze structural and energetic changes for these stabilizing indel mutants. We find that stabilizing indel mutations tend to have local structural effects and that stabilizing deletions (but less so insertions) are often found in regions of high backbone strain. We also find that stabilizing indels are typically correctly classified as stabilizing by the Rosetta energy function (which explicitly models backbone energetics), but not by an inverse folding (ESM-IF)-based analysis (Cagiada et al. 2024) which predicts absolute stability (ΔGunfolding).

Comprehensive deletion scan of anti-CRISPR AcrIIA4 reveals essential and dispensable domains for Cas9 inhibition

Delineating a protein's essential and dispensable domains provides critical insight into how it carries out its function. Here, we developed a high-throughput method to synthesize and test the functionality of all possible in-frame and continuous deletions in a gene of interest, enabling rapid and unbiased determination of protein domain importance. Our approach generates precise deletions using a CRISPR library framework that is free from constraints of gRNA target site availability and efficacy. We applied our method to AcrIIA4, a phage-encoded anti-CRISPR protein that robustly inhibits SpCas9. Extensive structural characterization has shown that AcrIIA4 physically occupies the DNA-binding interfaces of several SpCas9 domains; nonetheless, the importance of each AcrIIA4 interaction for SpCas9 inhibition is unknown. We used our approach to determine the essential and dispensable regions of AcrIIA4. Surprisingly, not all contacts with SpCas9 were required, and in particular, we found that the AcrIIA4 loop that inserts into SpCas9's RuvC catalytic domain can be deleted. Our results show that AcrIIA4 inhibits SpCas9 primarily by blocking PAM binding and that its interaction with the SpCas9 catalytic domain is inessential.

Massive experimental quantification of amyloid nucleation allows interpretable deep learning of protein aggregation

Protein aggregation is a pathological hallmark of more than fifty human diseases and a major problem for biotechnology. Methods have been proposed to predict aggregation from sequence, but these have been trained and evaluated on small and biased experimental datasets. Here we directly address this data shortage by experimentally quantifying the amyloid nucleation of >100,000 protein sequences. This unprecedented dataset reveals the limited performance of existing computational methods and allows us to train CANYA, a convolution-attention hybrid neural network that accurately predicts amyloid nucleation from sequence. We adapt genomic neural network interpretability analyses to reveal CANYA's decision-making process and learned grammar. Our results illustrate the power of massive experimental analysis of random sequence-spaces and provide an interpretable and robust neural network model to predict amyloid nucleation.

Production of Distinct Fibrillar, Oligomeric, and Other Aggregation States from Network Models of Multibody Interaction

Protein aggregation can produce a wide range of states, ranging from fibrillar structures and oligomers to unstructured and semistructured gel phases. Recent work has shown that many of these states can be recapitulated by relatively simple, topological models specified in terms of multibody interaction energies, providing a direct connection between aggregate intermolecular forces and aggregation products. Here, we examine a low-dimensional network Hamiltonian model (NHM) based on four basic multibody interactions found in any aggregate system. We characterize the phase behavior of this NHM family, showing that fibrils arise from a balance between elongation-inducing and contact-inhibiting forces. Complex oligomers (including annular oligomers resembling those thought to be toxic species in Alzheimer's disease) also form distinct phases in this regime, controlled in part by closure-inducing forces. We show that phase structure is largely independent of system size, and provide evidence of a rich structure of minor oligomeric phases that can arise from appropriate conditions. We characterize the phase behavior of this NHM family, demonstrating the range of ordered and disordered aggregation states possible with this set of interactions. As we show, fibrils arise from a balance between elongation-inducing and contact-inhibiting forces, existing in a regime bounded by gel-like and disaggregated phases; complex oligomers (including annular oligomers resembling those thought to be toxic species in Alzheimer's disease) also form distinct phases in this regime, controlled in part by closure-inducing forces. We show that phase structure is largely independent of system size, allowing generalization to macroscopic systems, and provide evidence of a rich structure of minor oligomeric phases that can arise from appropriate conditions.

Inhibition of the Early-Stage Cross-Amyloid Aggregation of Amyloid-β and IAPP via EGCG: Insights from Molecular Dynamics Simulations

Amyloid-β (Aβ) and islet amyloid polypeptide (IAPP) are small peptides that have the potential to not only self-assemble but also cross-assemble and form cytotoxic amyloid aggregates. Recently, we experimentally investigated the nature of Aβ-IAPP coaggregation and its inhibition by small polyphenolic molecules. Notably, we found that epigallocatechin gallate (EGCG) had the ability to reduce heteroaggregate formation. However, the precise molecular mechanism behind the reduction of heteroaggregates remains unclear. In this study, the dimerization processes of Aβ40 and IAPP peptides with and without EGCG were characterized by the enhanced sampling technique. Our results showed that these amyloid peptides exhibited a tendency to form a stable heterodimer, which represented the first step toward coaggregation. Furthermore, we also found that the EGCG regulated the dimerization process. In the presence of EGCG, well-tempered metadynamics simulation indicated a notable shift in the bound state toward a greater center of mass (COM) distance. Additionally, the presence of EGCG led to a significant increase in the free energy barrier height (∼15k B T) along the COM distance, and we observed a transition state between the bound and unbound states. Our findings also unveiled that the EGCG formed a greater number of hydrogen bonds with Aβ40, effectively obstructing the dimer formation. In addition, we carried out microseconds of all-atom conventional molecular dynamics (cMD) simulations to investigate the formation of both hetero- and homo-oligomer states by these peptides. MD simulations illustrated that EGCG played a significant role in preventing oligomer formation by reducing the content of β-sheets in the peptide. Collectively, our results offered valuable insight into the mechanism of cross-amyloid aggregation between Aβ40 and IAPP and the inhibition effect of EGCG on the heteroaggregation process.

Comprehensive deletion scan of anti-CRISPR AcrIIA4 reveals essential and dispensable domains for Cas9 inhibition

Delineating a protein's essential and dispensable domains provides critical insight into how it carries out its function. Here, we developed a high-throughput method to synthesize and test the functionality of all possible in-frame and continuous deletions in a gene of interest, enabling rapid and unbiased determination of protein domain importance. Our approach generates precise deletions using a CRISPR library framework that is free from constraints of gRNA target site availability and efficacy. We applied our method to AcrIIA4, a phage-encoded anti-CRISPR protein that robustly inhibits SpCas9. Extensive structural characterization has shown that AcrIIA4 physically occupies the DNA-binding interfaces of several SpCas9 domains; nonetheless, the importance of each AcrIIA4 interaction for SpCas9 inhibition is unknown. We used our approach to determine the essential and dispensable regions of AcrIIA4. Surprisingly, not all contacts with SpCas9 were required, and in particular, we found that the AcrIIA4 loop that inserts into SpCas9's RuvC catalytic domain can be deleted. Our results show that AcrIIA4 inhibits SpCas9 primarily by blocking PAM binding, and that its interaction with the SpCas9 catalytic domain is inessential.

Effects of anthocyanidins on the conformational transition of Aβ(1-42) peptide: Insights from molecular docking and molecular dynamics simulations

Recent evidence from in vitro and in vivo studies has shown that anthocyanins and anthocyanidins can reduce and inhibit the amyloid beta (Aβ) species, one of the hallmarks of Alzheimer's disease (AD). However, their inhibition mechanisms on Aβ species at molecular details remain elusive. Therefore, in the present study, molecular modelling methods were employed to investigate their inhibitory mechanisms on Aβ(1-42) peptide. The results highlighted that anthocyanidins effectively inhibited the conformational transitions of helices into beta-sheet (β-sheet) conformation within Aβ(1-42) peptide by two different mechanisms: 1) the obstruction of two terminals from coming into contact due to the binding of anthocyanidins with residues of N- and second hydrophobic core (SHC)-C-terminals, and 2) the prevention of the folding process due to the binding of anthocyanidin with the central polar (Asp23 and Lys28) and native helix (Asp23, Lys28, and Leu34) residues. These new findings on the inhibition of β-sheet formation by targeting both N- and SHC-C-terminals, and the long-established target, D23-K28 salt bridge residues, not with the conventional central hydrophobic core (CHC) as reported in the literature, might aid in designing more potent inhibitors for AD treatment.

Indels allow antiviral proteins to evolve functional novelty inaccessible by missense mutations

Antiviral proteins often evolve rapidly at virus-binding interfaces to defend against new viruses. We investigated whether antiviral adaptation via missense mutations might face limits, which insertion or deletion mutations (indels) could overcome. We report one such case of a nearly insurmountable evolutionary challenge: the human anti-retroviral protein TRIM5α requires more than five missense mutations in its specificity-determining v1 loop to restrict a divergent simian immunodeficiency virus (SIV). However, duplicating just one amino acid in v1 enables human TRIM5α to potently restrict SIV in a single evolutionary step. Moreover, natural primate TRIM5α v1 loops have evolved indels that confer novel antiviral specificities. Thus, indels enable antiviral proteins to overcome viral challenges inaccessible by missense mutations, revealing the potential of these often-overlooked mutations in driving protein innovation.

A turn for the worse: Aβ β-hairpins in Alzheimer's disease

Amyloid-β (Aβ) oligomers are a cause of neurodegeneration in Alzheimer's disease (AD). These soluble aggregates of the Aβ peptide have proven difficult to study due to their inherent metastability and heterogeneity. Strategies to isolate and stabilize homogenous Aβ oligomer populations have emerged such as mutations, covalent cross-linking, and protein fusions. These strategies along with molecular dynamics simulations have provided a variety of proposed structures of Aβ oligomers, many of which consist of molecules of Aβ in β-hairpin conformations. β-Hairpins are intramolecular antiparallel β-sheets composed of two β-strands connected by a loop or turn. Three decades of research suggests that Aβ peptides form several different β-hairpin conformations, some of which are building blocks of toxic Aβ oligomers. The insights from these studies are currently being used to design anti-Aβ antibodies and vaccines to treat AD. Research suggests that antibody therapies designed to target oligomeric Aβ may be more successful at treating AD than antibodies designed to target linear epitopes of Aβ or fibrillar Aβ. Aβ β-hairpins are good epitopes to use in antibody development to selectively target oligomeric Aβ. This review summarizes the research on β-hairpins in Aβ peptides and discusses the relevance of this conformation in AD pathogenesis and drug development.

Minimum information and guidelines for reporting a multiplexed assay of variant effect

Multiplexed assays of variant effect (MAVEs) have emerged as a powerful approach for interrogating thousands of genetic variants in a single experiment. The flexibility and widespread adoption of these techniques across diverse disciplines have led to a heterogeneous mix of data formats and descriptions, which complicates the downstream use of the resulting datasets. To address these issues and promote reproducibility and reuse of MAVE data, we define a set of minimum information standards for MAVE data and metadata and outline a controlled vocabulary aligned with established biomedical ontologies for describing these experimental designs.

Fibrinaloid Microclots and Atrial Fibrillation

Atrial fibrillation (AF) is a comorbidity of a variety of other chronic, inflammatory diseases for which fibrinaloid microclots are a known accompaniment (and in some cases, a cause, with a mechanistic basis). Clots are, of course, a well-known consequence of atrial fibrillation. We here ask the question whether the fibrinaloid microclots seen in plasma or serum may in fact also be a cause of (or contributor to) the development of AF. We consider known 'risk factors' for AF, and in particular, exogenous stimuli such as infection and air pollution by particulates, both of which are known to cause AF. The external accompaniments of both bacterial (lipopolysaccharide and lipoteichoic acids) and viral (SARS-CoV-2 spike protein) infections are known to stimulate fibrinaloid microclots when added in vitro, and fibrinaloid microclots, as with other amyloid proteins, can be cytotoxic, both by inducing hypoxia/reperfusion and by other means. Strokes and thromboembolisms are also common consequences of AF. Consequently, taking a systems approach, we review the considerable evidence in detail, which leads us to suggest that it is likely that microclots may well have an aetiological role in the development of AF. This has significant mechanistic and therapeutic implications.

Structural characterization of amyloid aggregates with spatially resolved infrared spectroscopy

The self-assembly of proteins and peptides into ordered structures called amyloid fibrils is a hallmark of numerous diseases, impacting the brain, heart, and other organs. The structure of amyloid aggregates is central to their function and thus has been extensively studied. However, the structural heterogeneities between aggregates as they evolve throughout the aggregation pathway are still not well understood. Conventional biophysical spectroscopic methods are bulk techniques and only report on the average structural parameters. Understanding the structure of individual aggregate species in a heterogeneous ensemble necessitates spatial resolution on the length scale of the aggregates. Recent technological advances have led to augmentation of infrared (IR) spectroscopy with imaging modalities, wherein the photothermal response of the sample upon vibrational excitation is leveraged to provide spatial resolution beyond the diffraction limit. These combined approaches are ideally suited to map out the structural heterogeneity of amyloid ensembles. AFM-IR, which integrates IR spectroscopy with atomic force microscopy enables identification of the structural facets the oligomers and fibrils at individual aggregate level with nanoscale resolution. These capabilities can be extended to chemical mapping in diseased tissue specimens with submicron resolution using optical photothermal microscopy, which combines IR spectroscopy with optical imaging. This book chapter provides the basic premise of these novel techniques and provides the typical methodology for using these approaches for amyloid structure determination. Detailed procedures pertaining to sample preparation and data acquisition and analysis are discussed and the aggregation of the amyloid β peptide is provided as a case study to provide the reader the experimental parameters necessary to use these techniques to complement their research efforts.

Investigating Cu(I) binding to model peptides of N-terminal Aβ isoforms

Amyloid beta (Aβ) peptides and copper (Cu) ions are each involved in critical biological processes including antimicrobial activity, regulation of synaptic function, angiogenesis, and others. Aβ binds to Cu and may play a role in Cu trafficking. Aβ peptides exist in isoforms that vary at their C-and N-termini; variation at the N-terminal sequence affects Cu binding affinity, structure, and redox activity by providing different sets of coordinating groups to the metal ion. Several N-terminal isoforms have been detected in human brain tissues including Aβ1-40/42, Aβ3-42, pEAβ3-42, Aβ4-42, Aβ11-40 and pEAβ11-40 (where pE denotes an N-terminal pyroglutamic acid). Several previous works have individually investigated the affinity and structure of Cu(I) bound to some of these isoforms' metal binding domains. However, the disparately reported values are apparent constants collected under different sets of conditions, and thus an integrated comparison cannot be made. The work presented here provides the Cu(I) coordination structure and binding affinities of these six biologically relevant Aβ isoforms determined in parallel using model peptides of the Aβ metal binding domains (Aβ1-16, Aβ3-16, pEAβ3-16, Aβ4-16, Aβ11-16 and pEAβ11-16). The binding affinities of Cu(I)-Aβ complexes were measured using solution competition with ferrozine (Fz) and bicinchoninic acid (BCA), two colorimetric Cu(I) indicators in common use. The Cu(I) coordination structures were characterized by X-ray absorption spectroscopy. The data presented here facilitate comparison of the isoforms' Cu-binding interactions and contribute to our understanding of the role of Aβ peptides as copper chelators in healthy and diseased brains.

Oxidative stress induced conformational changes of human serum albumin

Oxidative stress, generated by reactive oxygen species (ROS), is responsible for the loss of structure and functionality of proteins and is associated with several aging-related diseases. Here, we report an in vitro study to gauge the effect of ROS on the structural rearrangement of human serum albumin (HSA), a plasma protein, through metal-catalyzed oxidation (MCO) at physiological temperature through various biophysical techniques like UV-vis absorption, circular dichroism (CD), differential scanning calorimetry (DSC), MALDI-TOF, FTIR, and Raman spectroscopy. The UV-vis spectra of oxidized HSA show an early blueshift, signifying the unfolding of the protein because of ROS followed by the broadening of the absorption peak at a longer time. The DSC data corroborate the observation, revealing an exothermic transition for the oxidized sample at a longer time, suggesting in situ aggregation. The CD and FTIR spectra indicate the associated secondary structural changes occurring with time, depicting the variation of the helical content of HSA. The amide-III analysis of Raman data also complements the structural changes, and MALDI-TOF data show the mass distribution with time. Overall, this work might help determine the effect of oxidation on the biological activity of serum albumin as it can impact the physiological properties of HSA.

Evaluating generalizability of artificial intelligence models for molecular datasets

Deep learning has made rapid advances in modeling molecular sequencing data. Despite achieving high performance on benchmarks, it remains unclear to what extent deep learning models learn general principles and generalize to previously unseen sequences. Benchmarks traditionally interrogate model generalizability by generating metadata based (MB) or sequence-similarity based (SB) train and test splits of input data before assessing model performance. Here, we show that this approach mischaracterizes model generalizability by failing to consider the full spectrum of cross-split overlap, i.e., similarity between train and test splits. We introduce Spectra, a spectral framework for comprehensive model evaluation. For a given model and input data, Spectra plots model performance as a function of decreasing cross-split overlap and reports the area under this curve as a measure of generalizability. We apply Spectra to 18 sequencing datasets with associated phenotypes ranging from antibiotic resistance in tuberculosis to protein-ligand binding to evaluate the generalizability of 19 state-of-the-art deep learning models, including large language models, graph neural networks, diffusion models, and convolutional neural networks. We show that SB and MB splits provide an incomplete assessment of model generalizability. With Spectra, we find as cross-split overlap decreases, deep learning models consistently exhibit a reduction in performance in a task- and model-dependent manner. Although no model consistently achieved the highest performance across all tasks, we show that deep learning models can generalize to previously unseen sequences on specific tasks. Spectra paves the way toward a better understanding of how foundation models generalize in biology.

ProteinGym: Large-Scale Benchmarks for Protein Design and Fitness Prediction

Predicting the effects of mutations in proteins is critical to many applications, from understanding genetic disease to designing novel proteins that can address our most pressing challenges in climate, agriculture and healthcare. Despite a surge in machine learning-based protein models to tackle these questions, an assessment of their respective benefits is challenging due to the use of distinct, often contrived, experimental datasets, and the variable performance of models across different protein families. Addressing these challenges requires scale. To that end we introduce ProteinGym, a large-scale and holistic set of benchmarks specifically designed for protein fitness prediction and design. It encompasses both a broad collection of over 250 standardized deep mutational scanning assays, spanning millions of mutated sequences, as well as curated clinical datasets providing high-quality expert annotations about mutation effects. We devise a robust evaluation framework that combines metrics for both fitness prediction and design, factors in known limitations of the underlying experimental methods, and covers both zero-shot and supervised settings. We report the performance of a diverse set of over 70 high-performing models from various subfields (eg., alignment-based, inverse folding) into a unified benchmark suite. We open source the corresponding codebase, datasets, MSAs, structures, model predictions and develop a user-friendly website that facilitates data access and analysis.

Towards sequence-based principles for protein phase separation predictions

The phenomenon of protein phase separation, which underlies the formation of biomolecular condensates, has been associated with numerous cellular functions. Recent studies indicate that the amino acid sequences of most proteins may harbour not only the code for folding into the native state but also for condensing into the liquid-like droplet state and the solid-like amyloid state. Here we review the current understanding of the principles for sequence-based methods for predicting the propensity of proteins for phase separation. A guiding concept is that entropic contributions are generally more important to stabilise the droplet state than they are for the native and amyloid states. Although estimating these entropic contributions has proven difficult, we describe some progress that has been recently made in this direction. To conclude, we discuss the challenges ahead to extend sequence-based prediction methods of protein phase separation to include quantitative in vivo characterisations of this process.

DIMPLE: deep insertion, deletion, and missense mutation libraries for exploring protein variation in evolution, disease, and biology

Insertions and deletions (indels) enable evolution and cause disease. Due to technical challenges, indels are left out of most mutational scans, limiting our understanding of them in disease, biology, and evolution. We develop a low cost and bias method, DIMPLE, for systematically generating deletions, insertions, and missense mutations in genes, which we test on a range of targets, including Kir2.1. We use DIMPLE to study how indels impact potassium channel structure, disease, and evolution. We find deletions are most disruptive overall, beta sheets are most sensitive to indels, and flexible loops are sensitive to deletions yet tolerate insertions.