An atlas of protein homo-oligomerization across domains of life.

Protein structures are essential to understanding cellular processes in molecular detail. While advances in artificial intelligence revealed the tertiary structure of proteins at scale, their quaternary structure remains mostly unknown. We devise a scalable strategy based on AlphaFold2 to predict homo-oligomeric assemblies across four proteomes spanning the tree of life. Our results suggest that approximately 45% of an archaeal proteome and a bacterial proteome and 20% of two eukaryotic proteomes form homomers. Our predictions accurately capture protein homo-oligomerization, recapitulate megadalton complexes, and unveil hundreds of homo-oligomer types, including three confirmed experimentally by structure determination. Integrating these datasets with omics information suggests that a majority of known protein complexes are symmetric. Finally, these datasets provide a structural context for interpreting disease mutations and reveal coiled-coil regions as major enablers of quaternary structure evolution in human. Our strategy is applicable to any organism and provides a comprehensive view of homo-oligomerization in proteomes.

All-RNA-mediated targeted gene integration in mammalian cells with rationally engineered R2 retrotransposons.

All-RNA-mediated targeted gene integration methods, rendering reduced immunogenicity, effective deliverability with non-viral vehicles, and a low risk of random mutagenesis, are urgently needed for next-generation gene addition technologies. Naturally occurring R2 retrotransposons hold promise in this context due to their site-specific integration profile. Here, we systematically analyzed the biodiversity of R2 elements and screened several R2 orthologs capable of full-length gene insertion in mammalian cells. Robust R2 system gene integration efficiency was attained using combined donor RNA and protein engineering. Importantly, the all-RNA-delivered engineered R2 system showed effective integration activity, with efficiency over 60% in mouse embryos. Unbiased high-throughput sequencing demonstrated that the engineered R2 system exhibited high on-target integration specificity (99%). In conclusion, our study provides engineered R2 tools for applications based on hit-and-run targeted DNA integration and insights for further optimization of retrotransposon systems.

De novo protein identification in mammalian sperm using in situ cryoelectron tomography and AlphaFold2 docking.

To understand the molecular mechanisms of cellular pathways, contemporary workflows typically require multiple techniques to identify proteins, track their localization, and determine their structures in vitro. Here, we combined cellular cryoelectron tomography (cryo-ET) and AlphaFold2 modeling to address these questions and understand how mammalian sperm are built in situ. Our cellular cryo-ET and subtomogram averaging provided 6.0-Å reconstructions of axonemal microtubule structures. The well-resolved tertiary structures allowed us to unbiasedly match sperm-specific densities with 21,615 AlphaFold2-predicted protein models of the mouse proteome. We identified Tektin 5, CCDC105, and SPACA9 as novel microtubule-associated proteins. These proteins form an extensive interaction network crosslinking the lumen of axonemal doublet microtubules, suggesting their roles in modulating the mechanical properties of the filaments. Indeed, Tekt5 -/- sperm possess more deformed flagella with 180° bends. Together, our studies presented a cellular visual proteomics workflow and shed light on the in vivo functions of Tektin 5.

Structural basis of the Integrator complex assembly and association with transcription factors.

Integrator is a multi-subunit protein complex responsible for premature transcription termination of coding and non-coding RNAs. This is achieved via two enzymatic activities, RNA endonuclease and protein phosphatase, acting on the promoter-proximally paused RNA polymerase Ⅱ (RNAPⅡ). Yet, it remains unclear how Integrator assembly and recruitment are regulated and what the functions of many of its core subunits are. Here, we report the structures of two human Integrator sub-complexes: INTS10/13/14/15 and INTS5/8/10/15, and an integrative model of the fully assembled Integrator bound to the RNAPⅡ paused elongating complex (PEC). An in silico protein-protein interaction screen of over 1,500 human transcription factors (TFs) identified ZNF655 as a direct interacting partner of INTS13 within the fully assembled Integrator. We propose a model wherein INTS13 acts as a platform for the recruitment of TFs that could modulate the stability of the Integrator's association at specific loci and regulate transcription attenuation of the target genes.

ERMA (TMEM94) is a P-type ATPase transporter for Mg2+ uptake in the endoplasmic reticulum.

Intracellular Mg2+ (iMg2+) is bound with phosphometabolites, nucleic acids, and proteins in eukaryotes. Little is known about the intracellular compartmentalization and molecular details of Mg2+ transport into/from cellular organelles such as the endoplasmic reticulum (ER). We found that the ER is a major iMg2+ compartment refilled by a largely uncharacterized ER-localized protein, TMEM94. Conventional and AlphaFold2 predictions suggest that ERMA (TMEM94) is a multi-pass transmembrane protein with large cytosolic headpiece actuator, nucleotide, and phosphorylation domains, analogous to P-type ATPases. However, ERMA uniquely combines a P-type ATPase domain and a GMN motif for ERMg2+ uptake. Experiments reveal that a tyrosine residue is crucial for Mg2+ binding and activity in a mechanism conserved in both prokaryotic (mgtB and mgtA) and eukaryotic Mg2+ ATPases. Cardiac dysfunction by haploinsufficiency, abnormal Ca2+ cycling in mouse Erma+/- cardiomyocytes, and ERMA mRNA silencing in human iPSC-cardiomyocytes collectively define ERMA as an essential component of ERMg2+ uptake in eukaryotes.

Implications of a multiscale structure of the yeast nuclear pore complex.

Nuclear pore complexes (NPCs) direct the nucleocytoplasmic transport of macromolecules. Here, we provide a composite multiscale structure of the yeast NPC, based on improved 3D density maps from cryogenic electron microscopy and AlphaFold2 models. Key features of the inner and outer rings were integrated into a comprehensive model. We resolved flexible connectors that tie together the core scaffold, along with equatorial transmembrane complexes and a lumenal ring that anchor this channel within the pore membrane. The organization of the nuclear double outer ring reveals an architecture that may be shared with ancestral NPCs. Additional connections between the core scaffold and the central transporter suggest that under certain conditions, a degree of local organization is present at the periphery of the transport machinery. These connectors may couple conformational changes in the scaffold to the central transporter to modulate transport. Collectively, this analysis provides insights into assembly, transport, and NPC evolution.

Elucidation of E3 ubiquitin ligase specificity through proteome-wide internal degron mapping.

The ubiquitin-proteasome system plays a critical role in biology by regulating protein degradation. Despite their importance, precise recognition specificity is known for a few of the 600 E3s. Here, we establish a two-pronged strategy for identifying and mapping critical residues of internal degrons on a proteome-scale in HEK-293T cells. We employ global protein stability profiling combined with machine learning to identify 15,800 peptides likely to contain sequence-dependent degrons. We combine this with scanning mutagenesis to define critical residues for over 5,000 predicted degrons. Focusing on Cullin-RING ligase degrons, we generated mutational fingerprints for 219 degrons and developed DegronID, a computational algorithm enabling the clustering of degron peptides with similar motifs. CRISPR analysis enabled the discovery of E3-degron pairs, of which we uncovered 16 pairs that revealed extensive degron variability and structural determinants. We provide the visualization of these data on the public DegronID data browser as a resource for future exploration.

Identification and functional characterization of transcriptional activators in human cells.

Transcription is orchestrated by thousands of transcription factors (TFs) and chromatin-associated proteins, but how these are causally connected to transcriptional activation is poorly understood. Here, we conduct an unbiased proteome-scale screen to systematically uncover human proteins that activate transcription in a natural chromatin context. By combining interaction proteomics and chemical inhibitors, we delineate the preference of these transcriptional activators for specific co-activators, highlighting how even closely related TFs can function via distinct cofactors. We also identify potent transactivation domains among the hits and use AlphaFold2 to predict and experimentally validate interaction interfaces of two activation domains with BRD4. Finally, we show that many novel activators are partners in fusion events in tumors and functionally characterize a myofibroma-associated fusion between SRF and C3orf62, a potent p300-dependent activator. Our work provides a functional catalog of potent transactivators in the human proteome and a platform for discovering transcriptional regulators at genome scale.

Novel machine learning approaches revolutionize protein knowledge.

Breakthrough methods in machine learning (ML), protein structure prediction, and novel ultrafast structural aligners are revolutionizing structural biology. Obtaining accurate models of proteins and annotating their functions on a large scale is no longer limited by time and resources. The most recent method to be top ranked by the Critical Assessment of Structure Prediction (CASP) assessment, AlphaFold 2 (AF2), is capable of building structural models with an accuracy comparable to that of experimental structures. Annotations of 3D models are keeping pace with the deposition of the structures due to advancements in protein language models (pLMs) and structural aligners that help validate these transferred annotations. In this review we describe how recent developments in ML for protein science are making large-scale structural bioinformatics available to the general scientific community.

Alternating access of a bacterial homolog of neurotransmitter: sodium symporters determined from AlphaFold2 ensembles and DEER spectroscopy.

Neurotransmitter:sodium symporters (NSSs) play critical roles in neural signaling by regulating neurotransmitter uptake into cells powered by sodium electrochemical gradients. Bacterial NSSs orthologs, including MhsT from Bacillus halodurans, have emerged as model systems to understand the structural motifs of alternating access in NSSs and the extent of conservation of these motifs across the family. Here, we apply a computational/experimental methodology to illuminate the conformational landscape of MhsT alternating access. Capitalizing on our recently developed method, Sampling Protein Ensembles and Conformational Heterogeneity with AlphaFold2 (SPEACH_AF), we derived clusters of MhsT models spanning the transition from inward-facing to outward-facing conformations. Systematic application of double electron-electron resonance (DEER) spectroscopy revealed ligand-dependent movements of multiple structural motifs that underpin MhsT's conformational cycle. Remarkably, comparative DEER analysis in detergent micelles and lipid nanodiscs highlights the profound effect of the environment on the energetics of conformational changes. Through experimentally derived selection of collective variables, we present a model of ion and substrate-powered transport by MhsT consistent with the conformational cycle derived from DEER. Our findings not only advance the understanding of MhsT's function but also uncover motifs of conformational dynamics conserved within the broader context of the NSS family and within the LeuT-fold class of transporters. Importantly, our methodological blueprint introduces an approach that can be applied across a diverse spectrum of transporters to describe their conformational landscapes.

Modeling Flexible Protein Structure With AlphaFold2 and Crosslinking Mass Spectrometry.

We propose a pipeline that combines AlphaFold2 (AF2) and crosslinking mass spectrometry (XL-MS) to model the structure of proteins with multiple conformations. The pipeline consists of two main steps: ensemble generation using AF2 and conformer selection using XL-MS data. For conformer selection, we developed two scores-the monolink probability score (MP) and the crosslink probability score (XLP)-both of which are based on residue depth from the protein surface. We benchmarked MP and XLP on a large dataset of decoy protein structures and showed that our scores outperform previously developed scores. We then tested our methodology on three proteins having an open and closed conformation in the Protein Data Bank: Complement component 3 (C3), luciferase, and glutamine-binding periplasmic protein, first generating ensembles using AF2, which were then screened for the open and closed conformations using experimental XL-MS data. In five out of six cases, the most accurate model within the AF2 ensembles-or a conformation within 1 Å of this model-was identified using crosslinks, as assessed through the XLP score. In the remaining case, only the monolinks (assessed through the MP score) successfully identified the open conformation of glutamine-binding periplasmic protein, and these results were further improved by including the "occupancy" of the monolinks. This serves as a compelling proof-of-concept for the effectiveness of monolinks. In contrast, the AF2 assessment score was only able to identify the most accurate conformation in two out of six cases. Our results highlight the complementarity of AF2 with experimental methods like XL-MS, with the MP and XLP scores providing reliable metrics to assess the quality of the predicted models. The MP and XLP scoring functions mentioned above are available at https://gitlab.com/topf-lab/xlms-tools.

More than just pattern recognition: Prediction of uncommon protein structure features by AI methods.

The CASP14 experiment demonstrated the extraordinary structure modeling capabilities of artificial intelligence (AI) methods. That result has ignited a fierce debate about what these methods are actually doing. One of the criticisms has been that the AI does not have any sense of the underlying physics but is merely performing pattern recognition. Here, we address that issue by analyzing the extent to which the methods identify rare structural motifs. The rationale underlying the approach is that a pattern recognition machine tends to choose the more frequently occurring motifs, whereas some sense of subtle energetic factors is required to choose infrequently occurring ones. To reduce the possibility of bias from related experimental structures and to minimize the effect of experimental errors, we examined only CASP14 target protein crystal structures determined to a resolution limit better than 2 Å, which lacked significant amino acid sequence homology to proteins of known structure. In those experimental structures and in the corresponding models, we track cis peptides, π-helices, 310-helices, and other small 3D motifs that occur in the PDB database at a frequency of lower than 1% of total amino acid residues. The best-performing AI method, AlphaFold2, captured these uncommon structural elements exquisitely well. All discrepancies appeared to be a consequence of crystal environment effects. We propose that the neural network learned a protein structure potential of mean force, enabling it to correctly identify situations where unusual structural features represent the lowest local free energy because of subtle influences from the atomic environment.

Analysis of the complex between MBD2 and the histone deacetylase core of NuRD reveals key interactions critical for gene silencing.

The nucleosome remodeling and deacetylase (NuRD) complex modifies nucleosome positioning and chromatin compaction to regulate gene expression. The methyl-CpG-binding domain proteins 2 and 3 (MBD2 and MBD3) play a critical role in complex formation; however, the molecular details of how they interact with other NuRD components have yet to be fully elucidated. We previously showed that an intrinsically disordered region (IDR) of MBD2 is necessary and sufficient to bind to the histone deacetylase core of NuRD. Building on that work, we have measured the inherent structural propensity of the MBD2-IDR using solvent and site-specific paramagnetic relaxation enhancement measurements. We then used the AlphaFold2 machine learning software to generate a model of the complex between MBD2 and the histone deacetylase core of NuRD. This model is remarkably consistent with our previous studies, including the current paramagnetic relaxation enhancement data. The latter suggests that the free MBD2-IDR samples conformations similar to the bound structure. We tested this model of the complex extensively by mutating key contact residues and measuring binding using an intracellular bioluminescent resonance energy transfer assay. Furthermore, we identified protein contacts that, when mutated, disrupted gene silencing by NuRD in a cell model of fetal hemoglobin regulation. Hence, this work provides insights into the formation of NuRD and highlights critical binding pockets that may be targeted to block gene silencing for therapy. Importantly, we show that AlphaFold2 can generate a credible model of a large complex that involves an IDR that folds upon binding.

Systematic identification of conditionally folded intrinsically disordered regions by AlphaFold2.

The AlphaFold Protein Structure Database contains predicted structures for millions of proteins. For the majority of human proteins that contain intrinsically disordered regions (IDRs), which do not adopt a stable structure, it is generally assumed that these regions have low AlphaFold2 confidence scores that reflect low-confidence structural predictions. Here, we show that AlphaFold2 assigns confident structures to nearly 15% of human IDRs. By comparison to experimental NMR data for a subset of IDRs that are known to conditionally fold (i.e., upon binding or under other specific conditions), we find that AlphaFold2 often predicts the structure of the conditionally folded state. Based on databases of IDRs that are known to conditionally fold, we estimate that AlphaFold2 can identify conditionally folding IDRs at a precision as high as 88% at a 10% false positive rate, which is remarkable considering that conditionally folded IDR structures were minimally represented in its training data. We find that human disease mutations are nearly fivefold enriched in conditionally folded IDRs over IDRs in general and that up to 80% of IDRs in prokaryotes are predicted to conditionally fold, compared to less than 20% of eukaryotic IDRs. These results indicate that a large majority of IDRs in the proteomes of human and other eukaryotes function in the absence of conditional folding, but the regions that do acquire folds are more sensitive to mutations. We emphasize that the AlphaFold2 predictions do not reveal functionally relevant structural plasticity within IDRs and cannot offer realistic ensemble representations of conditionally folded IDRs.

Cross-linking mass spectrometry discovers, evaluates, and corroborates structures and protein-protein interactions in the human cell.

Significant recent advances in structural biology, particularly in the field of cryoelectron microscopy, have dramatically expanded our ability to create structural models of proteins and protein complexes. However, many proteins remain refractory to these approaches because of their low abundance, low stability, or-in the case of complexes-simply not having yet been analyzed. Here, we demonstrate the power of using cross-linking mass spectrometry (XL-MS) for the high-throughput experimental assessment of the structures of proteins and protein complexes. This included those produced by high-resolution but in vitro experimental data, as well as in silico predictions based on amino acid sequence alone. We present the largest XL-MS dataset to date, describing 28,910 unique residue pairs captured across 4,084 unique human proteins and 2,110 unique protein-protein interactions. We show that models of proteins and their complexes predicted by AlphaFold2, and inspired and corroborated by the XL-MS data, offer opportunities to deeply mine the structural proteome and interactome and reveal mechanisms underlying protein structure and function.

The sensor of the bacterial histidine kinase CpxA is a novel dimer of extracytoplasmic Per-ARNT-Sim domains.

Histidine kinases are key bacterial sensors that recognize diverse environmental stimuli. While mechanisms of phosphorylation and phosphotransfer by cytoplasmic kinase domains are relatively well-characterized, the ways in which extracytoplasmic sensor domains regulate activation remain mysterious. The Cpx envelope stress response is a conserved Gram-negative two-component system which is controlled by the sensor kinase CpxA. We report the structure of the Escherichia coli CpxA sensor domain (CpxA-SD) as a globular Per-ARNT-Sim (PAS)-like fold highly similar to that of Vibrio parahaemolyticus CpxA as determined by X-ray crystallography. Because sensor kinase dimerization is important for signaling, we used AlphaFold2 to model CpxA-SD in the context of its connected transmembrane domains, which yielded a novel dimer of PAS domains possessing a distinct dimer organization compared to previously characterized sensor domains. Gain of function cpxA∗ alleles map to the dimer interface, and mutation of other residues in this region also leads to constitutive activation. CpxA activation can be suppressed by mutations that restore inter-monomer interactions, suggesting that inhibitory interactions between CpxA-SD monomers are the major point of control for CpxA activation and signaling. Searching through hundreds of structural homologs revealed the sensor domain of Pseudomonas aeruginosa sensor kinase PfeS as the only PAS structure in the same novel dimer orientation as CpxA, suggesting that our dimer orientation may be utilized by other extracytoplasmic PAS domains. Overall, our findings provide insight into the diversity of the organization of PAS sensory domains and how they regulate sensor kinase activation.

The Origin and Evolution of Sex Peptide and Sex Peptide Receptor Interactions.

Post-mating responses play a vital role in successful reproduction across diverse species. In fruit flies, sex peptide binds to the sex peptide receptor, triggering a series of post-mating responses. However, the origin of sex peptide receptor predates the emergence of sex peptide. The evolutionary origins of the interactions between sex peptide and sex peptide receptor and the mechanisms by which they interact remain enigmatic. In this study, we used ancestral sequence reconstruction, AlphaFold2 predictions, and molecular dynamics simulations to study sex peptide-sex peptide receptor interactions and their origination. Using AlphaFold2 and long-time molecular dynamics simulations, we predicted the structure and dynamics of sex peptide-sex peptide receptor interactions. We show that sex peptide potentially binds to the ancestral states of Diptera sex peptide receptor. Notably, we found that only a few amino acid changes in sex peptide receptor are sufficient for the formation of sex peptide-sex peptide receptor interactions. Ancestral sequence reconstruction and molecular dynamics simulations further reveal that sex peptide receptor interacts with sex peptide through residues that are mostly involved in the interaction interface of an ancestral ligand, myoinhibitory peptides. We propose a potential mechanism whereby sex peptide-sex peptide receptor interactions arise from the preexisting myoinhibitory peptides-sex peptide receptor interface as well as early chance events both inside and outside the preexisting interface that created novel sex peptide-specific sex peptide-sex peptide receptor interactions. Our findings provide new insights into the origin and evolution of sex peptide-sex peptide receptor interactions and their relationship with myoinhibitory peptides-sex peptide receptor interactions.

Arsenate reductase of Rufibacter tibetensis is a metallophosphoesterase evolved to catalyze redox reactions.

An arsenate reductase (Car1) from the Bacteroidetes species Rufibacter tibetensis 1351T was isolated from the Tibetan Plateau. The strain exhibits resistance to arsenite [As(III)] and arsenate [As(V)] and reduces As(V) to As(III). Here we shed light on the mechanism of enzymatic reduction by Car1. AlphaFold2 structure prediction, active site energy minimization, and steady-state kinetics of wild-type and mutant enzymes give insight into the catalytic mechanism. Car1 is structurally related to calcineurin-like metallophosphoesterases (MPPs). It functions as a binuclear metal hydrolase with limited phosphatase activity, particularly relying on the divalent metal Ni2+. As an As(V) reductase, it displays metal promiscuity and is coupled to the thioredoxin redox cycle, requiring the participation of two cysteine residues, Cys74 and Cys76. These findings suggest that Car1 evolved from a common ancestor of extant phosphatases by incorporating a redox function into an existing MPP catalytic site. Its proposed mechanism of arsenate reduction involves Cys74 initiating a nucleophilic attack on arsenate, leading to the formation of a covalent intermediate. Next, a nucleophilic attack of Cys76 leads to the release of As(III) and the formation of a surface-exposed Cys74-Cys76 disulfide, ready for reduction by thioredoxin.

A model for polyomavirus helicase activity derived in part from the AlphaFold2 structure of SV40 T-antigen.

The mechanism used by polyomavirus and other viral SF3 helicases to unwind DNA at replication forks remains unknown. Using AlphaFold2, we have determined the structure of a representative SF3 helicase, the SV40 T-antigen (T-ag). This model has been analyzed in terms of the features of T-ag required for helicase activity, particularly the proximity of the T-ag origin binding domain (OBD) to the replication fork and the distribution of basic residues on the surface of the OBD that are known to play roles in DNA unwinding. These and related studies provide additional evidence that the T-ag OBDs have a role in the unwinding of DNA at the replication fork. Nuclear magnetic resonance and modeling experiments also indicate that protonated histidines on the surface of the T-ag OBD play an important role in the unwinding process, and additional modeling studies indicate that protonated histidines are essential in other SF3 and SF6 helicases. Finally, a model for T-ag's helicase activity is presented, which is a variant of the "rope climber." According to this model, the hands are the N-terminal OBD domains that interact with the replication fork, while the C-terminal helicase domains contain the feet that bind to single-stranded DNA. IMPORTANCE: Enzymes termed helicases are essential for the replication of DNA tumor viruses. Unfortunately, much remains to be determined about this class of enzymes, including their structures and the mechanism(s) they employ to unwind DNA. Herein, we present the full-length structure of a model helicase encoded by a DNA tumor virus. Moreover, this AI-based structure has been analyzed in terms of its basic functional properties, such as the orientation of the helicase at replication forks and the relative locations of the amino acid residues that are critical for helicase activity. Obtaining this information is important because it permits proposals regarding how DNA is routed through these model helicases. Also presented is structural evidence that the conclusions drawn from our detailed analyses of one model helicase, encoded by one class of tumor viruses, are likely to apply to other viral and eukaryotic helicases.

Predicting ion mobility collision cross sections using projection approximation with ROSIE-PARCS webserver.

Ion mobility coupled to mass spectrometry informs on the shape and size of protein structures in the form of a collision cross section (CCSIM). Although there are several computational methods for predicting CCSIM based on protein structures, including our previously developed projection approximation using rough circular shapes (PARCS), the process usually requires prior experience with the command-line interface. To overcome this challenge, here we present a web application on the Rosetta Online Server that Includes Everyone (ROSIE) webserver to predict CCSIM from protein structure using projection approximation with PARCS. In this web interface, the user is only required to provide one or more PDB files as input. Results from our case studies suggest that CCSIM predictions (with ROSIE-PARCS) are highly accurate with an average error of 6.12%. Furthermore, the absolute difference between CCSIM and CCSPARCS can help in distinguishing accurate from inaccurate AlphaFold2 protein structure predictions. ROSIE-PARCS is designed with a user-friendly interface, is available publicly and is free to use. The ROSIE-PARCS web interface is supported by all major web browsers and can be accessed via this link (https://rosie.graylab.jhu.edu).