Codon-specific Ramachandran plots show amino acid backbone conformation depends on identity of the translated codon

A dataset of alternately located segments in protein crystal structures

Protein Data Bank (PDB) files list the relative spatial location of atoms in a protein structure as the final output of the process of fitting and refining to experimentally determined electron density measurements. Where experimental evidence exists for multiple conformations, atoms are modelled in alternate locations. Programs reading PDB files commonly ignore these alternate conformations by default leaving users oblivious to the presence of alternate conformations in the structures they analyze. This has led to underappreciation of their prevalence, under characterisation of their features and limited the accessibility to this high-resolution data representing structural ensembles. We have trawled PDB files to extract structural features of residues with alternately located atoms. The output includes the distance between alternate conformations and identifies the location of these segments within the protein chain and in proximity of all other atoms within a defined radius. This dataset should be of use in efforts to predict multiple structures from a single sequence and support studies investigating protein flexibility and the association with protein function.

Translation Rates and Protein Folding

The mRNA coding sequence defines not only the amino acid sequence of the protein, but also the speed at which the ribosomes move along the mRNA while making the protein. The non-uniform local kinetics - denoted as translational rhythm - is similar among mRNAs coding for related protein folds. Deviations from this conserved rhythm can result in protein misfolding. In this review we summarize the experimental evidence demonstrating how local translation rates affect cotranslational protein folding, with the focus on the synonymous codons and patches of charged residues in the nascent peptide as best-studied examples. Alterations in nascent protein conformations due to disturbed translational rhythm can persist off the ribosome, as demonstrated by the effects of synonymous codon variants of several disease-related proteins. Charged amino acid patches in nascent chains also modulate translation and cotranslational protein folding, and can abrogate translation when placed at the N-terminus of the nascent peptide. During cotranslational folding, incomplete nascent chains navigate through a unique conformational landscape in which earlier intermediate states become inaccessible as the nascent peptide grows. Precisely tuned local translation rates, as well as interactions with the ribosome, guide the folding pathway towards the native structure, whereas deviations from the natural translation rhythm may favor pathways leading to trapped misfolded states. Deciphering the 'folding code' of the mRNA will contribute to understanding the diseases caused by protein misfolding and to rational protein design.

Variability in SOD1-associated amyotrophic lateral sclerosis: geographic patterns, clinical heterogeneity, molecular alterations, and therapeutic implications

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by progressive loss of motor neurons, resulting in global health burden and limited post-diagnosis life expectancy. Although primarily sporadic, familial ALS (fALS) cases suggest a genetic basis. This review focuses on SOD1, the first gene found to be associated with fALS, which has been more recently confirmed by genome sequencing. While informative, databases such as ALSoD and STRENGTH exhibit regional biases. Through a systematic global examination of SOD1 mutations from 1993 to 2023, we found different geographic distributions and clinical presentations. Even though different SOD1 variants are expressed at different protein levels and have different half-lives and dismutase activities, these alterations lead to loss of function that is not consistently correlated with disease severity. Gain of function of toxic aggregates of SOD1 resulting from mutated SOD1 has emerged as one of the key contributors to ALS. Therapeutic interventions specifically targeting toxic gain of function of mutant SOD1, including RNA interference and antibodies, show promise, but a cure remains elusive. This review provides a comprehensive perspective on SOD1-associated ALS and describes molecular features and the complex genetic landscape of SOD1, highlighting its importance in determining diverse clinical manifestations observed in ALS patients and emphasizing the need for personalized therapeutic strategies.

Re-examining Correlations Between Synonymous Codon Usage and Protein Bond Angles in Escherichia coli

Rosenberg AA, Marx A, Bronstein AM (Codon-specific Ramachandran plots show amino acid backbone conformation depends on identity of the translated codon. Nat Commun. 2022:13:2815) recently found a surprising correlation between synonymous codon usage and the dihedral bond angles of the resulting amino acid. However, their analysis did not account for the strongest known correlate of codon usage: gene expression. We re-examined the relationship between bond angles and codon usage by applying the approach of Rosenberg et al. to simulated protein-coding sequences that (i) have random codon usage, (ii) codon usage determined by mutation biases, and (iii) maintain the general relationship between codon usage and gene expression via the assumption of selection-mutation-drift equilibrium. We observed correlations between dihedral bond angle and codon usage when codon usage is entirely random, indicating possible conflation of noise with differences in bond angle distributions between synonymous codons. More relevant to the general analysis of codon usage patterns, we found surprisingly good agreement between the analysis of the real sequences and the analysis of sequences simulated assuming selection-mutation-drift equilibrium, with 91% of significant synonymous codon pairs detected in the former were also detected in the latter. We believe the correlation between codon usage and dihedral bond angles resulted from the variation in codon usage across genes due to the interplay between mutation bias, natural selection for translation efficiency, and gene expression, further underscoring these factors must be controlled for when looking for novel patterns related to codon usage.

Quantifying negative selection on synonymous variants

Widespread adoption of DNA sequencing has resulted in large numbers of genetic variants, whose contribution to disease is not easily determined. Although many types of variation are known to disrupt cellular processes in predictable ways, for some categories of variants, the effects may not be directly detectable. A particular example is synonymous variants, that is, those single-nucleotide variants that create a codon substitution, such that the produced amino acid sequence is unaffected. Contrary to the original theory suggesting that synonymous variants are benign, there is a growing volume of research showing that, despite their "silent" mechanism of action, some synonymous variation may be deleterious. Here, we studied the extent of the negative selective pressure acting on different classes of synonymous variants by analyzing the relative enrichment of synonymous singleton variants in the human exomes provided by gnomAD. Using a modification of the mutability-adjusted proportion of singletons (MAPS) metric as a measure of purifying selection, we found that some classes of synonymous variants are subject to stronger negative selection than others. For instance, variants that reduce codon optimality undergo stronger selection than optimality-increasing variants. Besides, selection affects synonymous variants implicated in splice-site-loss or splice-site-gain events. To understand what drives this negative selection, we tested a number of predictors in the aim to explain the variability in the selection scores. Our findings provide insights into the effects of synonymous variants at the population level, highlighting the specifics of the role that these variants play in health and disease.

Augmenting tomato functional genomics with a genome-wide induced genetic variation resource

Induced mutations accelerate crop improvement by providing novel disease resistance and yield alleles. However, the alleles with no perceptible phenotype but have an altered function remain hidden in mutagenized plants. The whole-genome sequencing (WGS) of mutagenized individuals uncovers the complete spectrum of mutations in the genome. Genome-wide induced mutation resources can improve the targeted breeding of tomatoes and facilitate functional genomics. In this study, we sequenced 132 doubly ethyl methanesulfonate (EMS)-mutagenized lines of tomato and detected approximately 41 million novel mutations and 5.5 million short InDels not present in the parental cultivar. Approximately 97% of the genome had mutations, including the genes, promoters, UTRs, and introns. More than one-third of genes in the mutagenized population had one or more deleterious mutations predicted by Sorting Intolerant From Tolerant (SIFT). Nearly one-fourth of deleterious genes mapped on tomato metabolic pathways modulate multiple pathway steps. In addition to the reported GC>AT transition bias for EMS, our population also had a substantial number of AT>GC transitions. Comparing mutation frequency among synonymous codons revealed that the most preferred codon is the least mutagenic toward EMS. The validation of a potato leaf-like mutation, reduction in carotenoids in ζ-carotene isomerase mutant fruits, and chloroplast relocation loss in phototropin1 mutant validated the mutation discovery pipeline. Our database makes a large repertoire of mutations accessible to functional genomics studies and breeding of tomatoes.

Mechanisms and pathology of protein misfolding and aggregation

Despite advances in machine learning-based protein structure prediction, we are still far from fully understanding how proteins fold into their native conformation. The conventional notion that polypeptides fold spontaneously to their biologically active states has gradually been replaced by our understanding that cellular protein folding often requires context-dependent guidance from molecular chaperones in order to avoid misfolding. Misfolded proteins can aggregate into larger structures, such as amyloid fibrils, which perpetuate the misfolding process, creating a self-reinforcing cascade. A surge in amyloid fibril structures has deepened our comprehension of how a single polypeptide sequence can exhibit multiple amyloid conformations, known as polymorphism. The assembly of these polymorphs is not a random process but is influenced by the specific conditions and tissues in which they originate. This observation suggests that, similar to the folding of native proteins, the kinetics of pathological amyloid assembly are modulated by interactions specific to cells and tissues. Here, we review the current understanding of how intrinsic protein conformational propensities are modulated by physiological and pathological interactions in the cell to shape protein misfolding and aggregation pathology.

Machine Learning-Guided Protein Engineering

Recent progress in engineering highly promising biocatalysts has increasingly involved machine learning methods. These methods leverage existing experimental and simulation data to aid in the discovery and annotation of promising enzymes, as well as in suggesting beneficial mutations for improving known targets. The field of machine learning for protein engineering is gathering steam, driven by recent success stories and notable progress in other areas. It already encompasses ambitious tasks such as understanding and predicting protein structure and function, catalytic efficiency, enantioselectivity, protein dynamics, stability, solubility, aggregation, and more. Nonetheless, the field is still evolving, with many challenges to overcome and questions to address. In this Perspective, we provide an overview of ongoing trends in this domain, highlight recent case studies, and examine the current limitations of machine learning-based methods. We emphasize the crucial importance of thorough experimental validation of emerging models before their use for rational protein design. We present our opinions on the fundamental problems and outline the potential directions for future research.

An amino-domino model described by a cross-peptide-bond Ramachandran plot defines amino acid pairs as local structural units

Protein structure, both at the global and local level, dictates function. Proteins fold from chains of amino acids, forming secondary structures, α-helices and β-strands, that, at least for globular proteins, subsequently fold into a three-dimensional structure. Here, we show that a Ramachandran-type plot focusing on the two dihedral angles separated by the peptide bond, and entirely contained within an amino acid pair, defines a local structural unit. We further demonstrate the usefulness of this cross-peptide-bond Ramachandran plot by showing that it captures β-turn conformations in coil regions, that traditional Ramachandran plot outliers fall into occupied regions of our plot, and that thermophilic proteins prefer specific amino acid pair conformations. Further, we demonstrate experimentally that the effect of a point mutation on backbone conformation and protein stability depends on the amino acid pair context, i.e., the identity of the adjacent amino acid, in a manner predictable by our method.

cdsBERT - Extending Protein Language Models with Codon Awareness

Recent advancements in Protein Language Models (pLMs) have enabled high-throughput analysis of proteins through primary sequence alone. At the same time, newfound evidence illustrates that codon usage bias is remarkably predictive and can even change the final structure of a protein. Here, we explore these findings by extending the traditional vocabulary of pLMs from amino acids to codons to encapsulate more information inside CoDing Sequences (CDS). We build upon traditional transfer learning techniques with a novel pipeline of token embedding matrix seeding, masked language modeling, and student-teacher knowledge distillation, called MELD. This transformed the pretrained ProtBERT into cdsBERT; a pLM with a codon vocabulary trained on a massive corpus of CDS. Interestingly, cdsBERT variants produced a highly biochemically relevant latent space, outperforming their amino acid-based counterparts on enzyme commission number prediction. Further analysis revealed that synonymous codon token embeddings moved distinctly in the embedding space, showcasing unique additions of information across broad phylogeny inside these traditionally "silent" mutations. This embedding movement correlated significantly with average usage bias across phylogeny. Future fine-tuned organism-specific codon pLMs may potentially have a more significant increase in codon usage fidelity. This work enables an exciting potential in using the codon vocabulary to improve current state-of-the-art structure and function prediction that necessitates the creation of a codon pLM foundation model alongside the addition of high-quality CDS to large-scale protein databases.

Rare variants in long non-coding RNAs are associated with blood lipid levels in the TOPMed Whole Genome Sequencing Study

Long non-coding RNAs (lncRNAs) are known to perform important regulatory functions. Large-scale whole genome sequencing (WGS) studies and new statistical methods for variant set tests now provide an opportunity to assess the associations between rare variants in lncRNA genes and complex traits across the genome. In this study, we used high-coverage WGS from 66,329 participants of diverse ancestries with blood lipid levels (LDL-C, HDL-C, TC, and TG) in the National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) program to investigate the role of lncRNAs in lipid variability. We aggregated rare variants for 165,375 lncRNA genes based on their genomic locations and conducted rare variant aggregate association tests using the STAAR (variant-Set Test for Association using Annotation infoRmation) framework. We performed STAAR conditional analysis adjusting for common variants in known lipid GWAS loci and rare coding variants in nearby protein coding genes. Our analyses revealed 83 rare lncRNA variant sets significantly associated with blood lipid levels, all of which were located in known lipid GWAS loci (in a ±500 kb window of a Global Lipids Genetics Consortium index variant). Notably, 61 out of 83 signals (73%) were conditionally independent of common regulatory variations and rare protein coding variations at the same loci. We replicated 34 out of 61 (56%) conditionally independent associations using the independent UK Biobank WGS data. Our results expand the genetic architecture of blood lipids to rare variants in lncRNA, implicating new therapeutic opportunities.

The structural plasticity of polyglutamine repeats

From yeast to humans, polyglutamine (polyQ) repeat tracts are found frequently in the proteome and are particularly prominent in the activation domains of transcription factors. PolyQ is a polymorphic motif that modulates functional protein-protein interactions and aberrant self-assembly. Expansion of the polyQ repeated sequences beyond critical physiological repeat length thresholds triggers self-assembly and is linked to severe pathological implications. This review provides an overview of the current knowledge on the structures of polyQ tracts in the soluble and aggregated states and discusses the influence of neighboring regions on polyQ secondary structure, aggregation, and fibril morphologies. The influence of the genetic context of the polyQ-encoding trinucleotides is briefly discussed as a challenge for future endeavors in this field.

Computational and artificial intelligence-based methods for antibody development

Due to their high target specificity and binding affinity, therapeutic antibodies are currently the largest class of biotherapeutics. The traditional largely empirical antibody development process is, while mature and robust, cumbersome and has significant limitations. Substantial recent advances in computational and artificial intelligence (AI) technologies are now starting to overcome many of these limitations and are increasingly integrated into development pipelines. Here, we provide an overview of AI methods relevant for antibody development, including databases, computational predictors of antibody properties and structure, and computational antibody design methods with an emphasis on machine learning (ML) models, and the design of complementarity-determining region (CDR) loops, antibody structural components critical for binding.

Machine learning classifiers predict key genomic and evolutionary traits across the kingdoms of life

In this study, we investigate how an organism's codon usage bias can serve as a predictor and classifier of various genomic and evolutionary traits across the domains of life. We perform secondary analysis of existing genetic datasets to build several AI/machine learning models. When trained on codon usage patterns of nearly 13,000 organisms, our models accurately predict the organelle of origin and taxonomic identity of nucleotide samples. We extend our analysis to identify the most influential codons for phylogenetic prediction with a custom feature ranking ensemble. Our results suggest that the genetic code can be utilized to train accurate classifiers of taxonomic and phylogenetic features. We then apply this classification framework to open reading frame (ORF) detection. Our statistical model assesses all possible ORFs in a nucleotide sample and rejects or deems them plausible based on the codon usage distribution. Our dataset and analyses are made publicly available on GitHub and the UCI ML Repository to facilitate open-source reproducibility and community engagement.

Implementing computational methods in tandem with synonymous gene recoding for therapeutic development

Synonymous gene recoding, the substitution of synonymous variants into the genetic sequence, has been used to overcome many production limitations in therapeutic development. However, the safety and efficacy of recoded therapeutics can be difficult to evaluate because synonymous codon substitutions can result in subtle, yet impactful changes in protein features and require sensitive methods for detection. Given that computational approaches have made significant leaps in recent years, we propose that machine-learning (ML) tools may be leveraged to assess gene-recoded therapeutics and foresee an opportunity to adapt codon contexts to enhance some powerful existing tools. Here, we examine how synonymous gene recoding has been used to address challenges in therapeutic development, explain the biological mechanisms underlying its effects, and explore the application of computational platforms to improve the surveillance of functional variants in therapeutic design.

Clinical and Therapeutic Evaluation of the Ten Most Prevalent CRB1 Mutations

Mutations in the Crumbs homolog 1 (CRB1) gene lead to severe inherited retinal dystrophies (IRDs), accounting for nearly 80,000 cases worldwide. To date, there is no therapeutic option for patients suffering from CRB1-IRDs. Therefore, it is of great interest to evaluate gene editing strategies capable of correcting CRB1 mutations. A retrospective chart review was conducted on ten patients demonstrating one or two of the top ten most prevalent CRB1 mutations and receiving care at Columbia University Irving Medical Center, New York, NY, USA. Patient phenotypes were consistent with previously published data for individual CRB1 mutations. To identify the optimal gene editing strategy for these ten mutations, base and prime editing designs were evaluated. For base editing, we adopted the use of a near-PAMless Cas9 (SpRY Cas9), whereas for prime editing, we evaluated the canonical NGG and NGA prime editors. We demonstrate that for the correction of c.2843G>A, p.(Cys948Tyr), the most prevalent CRB1 mutation, base editing has the potential to generate harmful bystanders. Prime editing, however, avoids these bystanders, highlighting its future potential to halt CRB1-mediated disease progression. Additional studies investigating prime editing for CRB1-IRDs are needed, as well as a thorough analysis of prime editing's application, efficiency, and safety in the retina.

Machine learning approaches demonstrate that protein structures carry information about their genetic coding

Synonymous codons translate into the same amino acid. Although the identity of synonymous codons is often considered inconsequential to the final protein structure, there is mounting evidence for an association between the two. Our study examined this association using regression and classification models, finding that codon sequences predict protein backbone dihedral angles with a lower error than amino acid sequences, and that models trained with true dihedral angles have better classification of synonymous codons given structural information than models trained with random dihedral angles. Using this classification approach, we investigated local codon-codon dependencies and tested whether synonymous codon identity can be predicted more accurately from codon context than amino acid context alone, and most specifically which codon context position carries the most predictive power.

Local Backbone Geometry Plays a Critical Role in Determining Conformational Preferences of Amino Acid Residues in Proteins

The definition of the structural basis of the conformational preferences of the genetically encoded amino acid residues is an important yet unresolved issue of structural biology. In order to gain insights into this intricate topic, we here determined and compared the amino acid propensity scales for different (φ, ψ) regions of the Ramachandran plot and for different secondary structure elements. These propensities were calculated using the Chou–Fasman approach on a database of non-redundant protein chains retrieved from the Protein Data Bank. Similarities between propensity scales were evaluated by linear regression analyses. One of the most striking and unexpected findings is that distant regions of the Ramachandran plot may exhibit significantly similar propensity scales. On the other hand, contiguous regions of the Ramachandran plot may present anticorrelated propensities. In order to provide an interpretative background to these results, we evaluated the role that the local variability of protein backbone geometry plays in this context. Our analysis indicates that (dis)similarities of propensity scales between different regions of the Ramachandran plot are coupled with (dis)similarities in the local geometry. The concept that similarities of the propensity scales are dictated by the similarity of the NCαC angle and not necessarily by the similarity of the (φ, ψ) conformation may have far-reaching implications in the field.

Non-Equilibrium Protein Folding and Activation by ATP-Driven Chaperones

Recent experimental studies suggest that ATP-driven molecular chaperones can stabilize protein substrates in their native structures out of thermal equilibrium. The mechanism of such non-equilibrium protein folding is an open question. Based on available structural and biochemical evidence, I propose here a unifying principle that underlies the conversion of chemical energy from ATP hydrolysis to the conformational free energy associated with protein folding and activation. I demonstrate that non-equilibrium folding requires the chaperones to break at least one of four symmetry conditions. The Hsp70 and Hsp90 chaperones each break a different subset of these symmetries and thus they use different mechanisms for non-equilibrium protein folding. I derive an upper bound on the non-equilibrium elevation of the native concentration, which implies that non-equilibrium folding only occurs in slow-folding proteins that adopt an unstable intermediate conformation in binding to ATP-driven chaperones. Contrary to the long-held view of Anfinsen's hypothesis that proteins fold to their conformational free energy minima, my results predict that some proteins may fold into thermodynamically unstable native structures with the assistance of ATP-driven chaperones, and that the native structures of some chaperone-dependent proteins may be shaped by their chaperone-mediated folding pathways.