Protein Science

A challenge in synthetic biology is to transfer the knowledge acquired in vitro to in vivo applications. For instance, binding affinity between peptide binding domains (PBDs) and their binding peptides is usually measured in simple buffers and in the absence of the cell machinery responsible for synthesis and proteostasis. Consequently, there is often a discrepancy between interaction quantified in vitro and in vivo. To identify strongly binding PBD-peptide pairs for in vivo applications, we aim to better understand the determinants of synthetic interaction in living yeast cells. We designed libraries of peptides expected to bind to PBDs based on data collected by phage-display. Using an in vivo protein interaction assay, we assessed the interaction strength of >6,500 PBD-peptide variants in yeast. We found that most substitutions disrupt binding, but that 3.4% increased binding strength to levels higher than expected from the phage-display data. We asked if, by using our mutational data, it was possible to optimize peptide sequences to gain even more binding strength and show additivity effect up to four substitutions. We found that PBD specificity is consistent between in vitro and in vivo assays. We hypothesized that in vivo peptide abundance could contribute to the discrepancy between the PCA and the phage-display results. We performed assays to detect in vivo peptide availability, reporting on peptide abundance, and we found that peptides with a stronger binding also show a higher availability in cells. We show that peptide availability, which seems mainly defined by peptide hydrophobicity, is key for the design of strong synthetic interactions.

Structure-aware proteolysis prediction often ranks candidate cleavage sites using solvent accessibility from static protein structures, yet this assumption is rarely tested against native-state cleavage data. We examined this problem using AprX, a serralysin-family protease that readily hydrolyses caseins but shows lower activity toward native whey proteins. Across six bovine milk proteins, sequence compatibility and local scissile-bond accessibility did not separate caseins from whey proteins. In contrast, accessibility fell progressively in whey proteins as the required contiguous substrate window increased, reaching 0% across the working P4-P4' footprint, while all evaluable casein windows remained accessible in isolated-chain ensembles. Local remodelling showed that whey segments could still adopt AprX-compatible backbone geometry. An external thermolysin benchmark rejected a rigid whole-protein clash metric, but did not alter the contiguous-accessibility result. These findings identify footprint-scale accessibility as a distinct structural variable and show why single-residue solvent exposure can overestimate protease accessibility in folded proteins.

Human Ca2+/calmodulin (CaM)-dependent protein kinase II alpha (CaMKII) is a highly abundant and conserved protein in the brain. It is a multifunctional serine/threonine kinase that phosphorylates cellular substrates across subcellular compartments, activating signaling pathways essential for long-term memory storage. However, a detailed understanding of protein structure and enzyme-substrate interaction mechanisms require a method to produce active, stable, and pure proteins at scale. Since phosphorylation can alter catalytic activity and substrate affinity, examining the non-phosphorylated state enables an unbiased characterization of basal enzymatic function. Therefore, producing the non-phosphorylated form of the kinase is important for biochemical and structural studies aimed at understanding kinase activation mechanisms. We describe a detailed protocol for the expression and high-yield purification of full-length, non-phosphorylated human CaMKII in bacterial cells. CaMKII was co-expressed with {lambda}-phosphatase to maintain it in a non-phosphorylated state. Protein expression was induced by autoinduction at low temperature to enhance the stability and solubility of CaMKII. Recombinant CaMKII was purified by Ni-NTA affinity chromatography, followed by gel filtration chromatography. Although CaMKII has been considered difficult to purify in its non-phosphorylated form, we obtained it in high yield, and the protein was predominantly non-phosphorylated. Our results show that combining co-expression with autoinduction improves both the solubility and yield of the homogeneous non-phosphorylated form of CaMKII, which could be suitable for subsequent biochemical and structural studies.

De novo design of protein binders requires a target structure. However, for flexible targets, such as intrinsically disordered proteins, this structure does not exist until the binder has stabilized the interaction. Such targets are therefore difficult for methods that separate structure generation from sequence design. We introduce GuideFlip, which co-designs structure and sequence through guided discrete flow matching: binder residues are assigned progressively while the complex is re-predicted at each step, allowing the evolving interface to affect the design process. GuideFlip reduces the hydrophobic bias of direct AlphaFold optimization and improves in silico success rates over existing approaches. We release a database of binder candidates for 177 human disordered proteins. Experimentally, we obtain de novo binders to the C-terminus of -synuclein and the disordered amino terminus of RBX1 with hit rates of 13.5% and 41.7%, respectively, and we confirm the epitopes of selected binders by NMR and mutagenesis. Applying GuideFlip to flexibility on the binder side, we design a nanobody that binds the agonist-bound {beta}1-adrenergic receptor in the active state, but not the receptor in its inactive state, with a 75% hit rate and cryo-EM structure confirming the design. GuideFlip enables protein design where bound structures emerge only upon binding.

Transition-metal ion Forster resonance energy transfer (tmFRET) measures short-range distances and conformational changes in proteins, but its distance sensitivity depends on the donor-acceptor pair. Here, we establish the metal-chelating noncanonical amino acid bipyridylalanine (bpyA) as a flexible, genetically encoded acceptor platform whose Forster distance (R0) can be tuned without changing protein labeling. Cu2+, Cu2+ complexed with nitrilotriacetic acid (NTA), and Fe2+ form distinct acceptors at the same bpyA site, allowing distance sensitivity to be adjusted by changing the solution conditions. We incorporated bpyA at two sites in maltose binding protein (MBP) and labeled nearby engineered cysteines with the donor fluorophore BODIPY. The resulting [Cu(bpyA)]2+, [CuNTA(bpyA)]2+, and [Fe(bpyA)]2+ acceptors provided R0 values ranging from 13.9 to 21.2 [A]. Analysis of time-resolved tmFRET data recovered donor-acceptor distance distributions and maltose-dependent changes consistent with structural predictions. At the 237/295 site pair, both copper and iron acceptors resolved the maltose-dependent decrease in distance. At the 309/322 site pair, [Cu(bpyA)]2+ did not appreciably resolve the distance increase, whereas adding NTA shifted R0 toward shorter distances and enabled detection of an increase from 12.1 to 14.4 [A]. These results demonstrate that a single bpyA-labeled protein preparation supports multiple tmFRET acceptors, providing a cysteine-independent acceptor-labeling strategy with adjustable distance sensitivity.

Protein language models (PLMs) have transformed our ability to learn from evolutionary sequence space, but protein engineering ultimately asks a different question: not what evolution selected, but what we should build next. Zero-shot likelihoods therefore provide useful, but not universal, measures of fitness and can misalign with engineering objectives. Experimental supervision redirects these priors towards properties of interest, enabling target-specific prediction and closed-loop optimization. PLMs thereby complement structure-based design: structural methods provide geometric control, while PLMs integrate experimental feedback to optimize functional and developability properties. Realizing this potential requires evaluation beyond retrospective predictive accuracy, towards extrapolation, multi-objective optimization, and prospective experimental success. We argue for recurring competitions combining scalable predictive benchmarks with prospective experimental challenges. Ultimately, progress should be measured not by predicting existing experiments, but by enabling successful new ones.

Thermophilic bacterial lipases must combine structural stability with the conformational flexibility required for interfacial activation, yet the local organization of their metal-binding sites remains incompletely understood. Here, we determined two X-ray crystal structures of the family I.5 lipase from Geobacillus kaustophilus (GkLip). The 2.47-[A] orthorhombic model contains a Ca2+ ion at the conserved metal-binding site, whereas the 2.30-[A] anisotropic triclinic model contains four non-equivalent protein chains in which no Ca2+ ion was modeled. GkLip retains the canonical /{beta}-hydrolase fold, the Ser114-Asp318-His359 catalytic triad, the Phe17/Gln115 oxyanion-hole region, an extended lid, and a conserved Zn2+-binding site. In the orthorhombic structure, Gly287, Glu361, Asp366, and Pro367 contribute to the Ca2+ coordination environment. Pro367 adopts a C{gamma}-endo pucker in the Ca2+-containing structure; three Ca2+-free chains also adopt the endo pucker, whereas one adopts a C{gamma}-exo pucker. The presence of both Pro367 pucker states in the Ca2+-free structure is consistent with calcium preferentially stabilizing a pre-existing, coordination-compatible conformation. This local organization is further supported by an Arg35-centered network of direct and water-mediated contacts. Comparison with the 94%-identical open G. thermocatenulatus lipase reveals substantial lid rearrangement despite strong conservation of the protein core. Together, these structures highlight localized conformational variability at the calcium-binding region within an otherwise highly conserved thermophilic lipase scaffold.

De novo-designed enzymes have recently achieved high catalytic activity and stereoselectivity while demonstrating exceptional thermostability in entirely novel protein scaffolds. Among these, -helical barrel protein scaffolds are attractive structures for biocatalysis due to their structural simplicity, high thermostability, and rationalizable sequence patterning. However, enabling major structural reengineering of these scaffolds while maintaining the structure, stability and catalytic activity while also improving soluble protein production remain major challenges and pose the fundamental question how engineerable a de novo backbone-sequence pair is. Here, we combine deep learning based and classic computational protein design to modify and optimize de novo -helical barrel biocatalysts. Using the previously reported six-helical barrel 6H5L as a model scaffold, AlphaFold2-guided RosettaRemodel enabled the design of a truncated variant, whose crystal structure closely matches the design model. Additional sequence-redesign using ProteinMPNN generated a variant with a tenfold increase of soluble protein yield in Escherichia coli. Biochemical, biophysical, and structural analyses showed that both variants retained the overall barrel architecture, high thermal stability, and catalytic activity for both purified protein and whole-cell systems. Detailed kinetic analysis on the variants showed both variation in kcat and Km, reflecting changes in catalytic turnover and substrate binding. Together, these approaches provide new insights and possibilities for the further engineering of functional de novo -helical barrels, their ability to withstand dramatically large sequence changes and their broader application in biocatalysis and biotechnology.

AlphaFold 3 (AF3) has broadened the computational structural biology landscape by predicting all-atom complexes across proteins, nucleic acids, small-molecule ligands, and ions, including chemical modifications. For non-experts, Google provides a fully web-based service to utilize AlphaFold 3. Unfortunately, despite a flexible and rich interface for inputs, the server's outputs are incomplete and complicated for many users, as witnessed hands-on. Not all confidence metrics are displayed, and those that are shown do not present much interactivity; besides, only model number 1 is shown among the 5 produced by the software, and upon download the user finds the models are provided only in CIF format, which is not yet well-known especially among highly practical users. Here, we present the AlphaFold 3 Multi-Model Inspector (AF3 Inspector, https://pdbms.altervista.org/af3viewer/afviewer8.html), an open, client-side, zero-install web application designed to dissect, align, and interactively explore the complete ensemble of structural models produced by AlphaFold 3. Part of the PDB Manipulation Suite (https://pdbms.altervista.org/), the AF3 Inspector operates entirely within web browsers, meaning it is available out of the box in all devices and operating systems. AF3 Inspector delivers 3D visualization in various styles and colors for all models, overlay with backbone alignment if requested, interactive plots for pLDDT profiles, pAE matrices and contact maps, chain-pair interaction heatmaps, automated detection of ligands, ions and post-translational modifications, and rapid mmCIF-to-PDB conversion allowing users to download the more familiar files. The platform addresses critical practical considerations highlighted in recent community benchmarks, such as CASP16, and extends the lightweight, client-side biophysical toolkit paradigm established by the PDB Manipulation Suite (PDBMS).

Small multifunctional viral membrane proteins called viroporins are attractive drug targets but are difficult to structurally characterize in a lipid-bilayer environment. We have successfully reconstituted the full-length Envelope (E) protein from SARS-CoV-2 in a lipid-bilayer, co-assembled with a native host factor that binds to E, the PDZ domain 2 of ZO1. We have leveraged the enhancement afforded by dynamic nuclear polarization NMR to determine that the ensemble-averaged oligomer state of E is a pentamer. Complexing the protein with a host PDZ domain did not impact the oligomer state, but resulted in structuring of the last four residues of E, and immobilization of the PDZ domain near the lipid surface. Combined with secondary structure information derived from the chemical shifts, these data were used to model an E protein pentamer together with the PDZ protein dimer. Inclusion of a relevant host factor, as well as a complex lipid bilayer that mimics conditions in the ER-Golgi intermediate compartment, approaches native conditions, and addresses the ongoing controversy regarding the functional oligomer state in an infected cell.

Accurate prediction of mutation-induced changes in protein-protein affinity remains a central challenge in computational biophysics and protein engineering. Here we present a physics-based scoring method for predicting the effects of single amino acid substitutions on protein-protein and protein-peptide binding affinity. The method is based on MMPBSA energy calculations and implemented in a user-friendly, publicly available workflow (https://github.com/Hoecker-Lab/mmpbsa_scoring). To derive the scoring function, MMPBSA energy terms were fitted using a large and diverse dataset of experimentally measured binding affinity changes upon point mutation. The terms were then evaluated on independent systems. Specifically, we assessed the ability to predict mutational effects at the interface of SARS-CoV-2 spike protein with its cellular receptor, as well as computationally predicted Armadillo-repeat protein-peptide complexes. The approach performs well across diverse targets and shows promising predictive performance on both experimentally determined and computationally predicted structures. It matches or exceeds the accuracy of established physics-based protein design software and state-of-the-art deep learning-based scoring functions when applied to data outside their training domain. The method provides a computationally efficient and interpretable framework for prioritizing interface mutations in protein engineering and binding affinity optimization.

In many neurodegenerative and systemic disorders, proteins can form insoluble protein aggregates called amyloids. Identification of amyloid forming regions in a protein remains central to understanding of its aggregation behaviour. Many predictors have been formed in the past and until recently which to estimate the aggregation-prone regions, which may not correspond to total residues found in actual disease associated fibril structures. To address this, we formed a sequence-based based predictor to ascertain structural amyloid-core propensity. The predictor was developed using experimentally determined fibril structures obtained from Amyloid Atlas. Ordered residues in the experimental structures were designated as core, and unresolved sampled from the same proteins as matched controls as non-core. Core and non-core regions were encoded using various physicochemical descriptors and protein language model (PLM) embeddings (ESM-2, ANKH, ProtT5) and evaluated using protein-grouped cross-validation. PLMs consistently outperformed physicochemical descriptors, with the best models reaching AUROC values of ~0.88 and AUPRC values of ~0.85. Locked full-length protein scans further localized experimental cores, with ESM-2/ExtraTrees W21 achieving AUROC 0.833, AUPRC 0.751, and a mean peak distance of 12.4 residues. In comparative benchmarking, our predictor showed better performance metrics than CrossBeta and AggrescanAI. The framework therefore provides residue-resolved prediction of structurally incorporated amyloid-core regions directly from sequence. AmyloCore-ML is easily accessible through an interactive Google Colab notebook, enabling sequence-based amyloid fibril-core prediction without local installation or dedicated computing infrastructure.

Cryptic binding pockets; sites absent or occluded in a proteins resting-state structure that become druggable only in specific, transiently populated conformations; represent one of the largest untapped opportunities in structure-based drug discovery. Large-scale structural surveys estimate that cryptic pockets occur in on the order of one in six protein families genome-wide, and that accounting for them could expand the druggable fraction of the disease-associated human proteome from roughly 40% to nearly 80%. Detecting cryptic pockets computationally has historically forced a choice btw physics-based conformational sampling, which is reliable but too computationally expensive to deploy across more than a handful of targets, and fast machine-learning pocket predictors trained on static structures, which frequently over-predict and lose the precision needed for practical triage. We report Kryptix-1, an in-house pipeline developed at Covenant Biosciences that combines conformational ensemble generation with a consensus pocket-scoring algorithm, benchmarked here against a matched static-structure baseline and against the published literature on a 10-protein sample from CryptoBench, an independently curated cryptic-pocket reference dataset. On the fields own standard residue-overlap metric (Jaccard index 0.5), Kryptix-1 succeeds on 80% of benchmark proteins, compared to 20% for the static baseline and approximately 40-45% for the best-performing methods reported in an independent comparative evaluation using the same metric. We report these results alongside an explicit accounting of where Kryptix-1s advantage is smaller or reverses, and we are explicit throughout that this is a small pilot evaluation, not a fully powered validation study.

In a companion study, we introduced ProDive and used it to identify 318,289 high-confidence cross-family segment correspondences across all 25,545 Pfam families. Here we characterize the pervasiveness and biophysical context of the resulting mapped fragment resource. Graph-based community analysis shows that segment reuse is broadly distributed across the protein family universe, with thousands of mostly small modules rather than concentration in a small number of family pairs. Comparison between Pfam families and de novo designed proteins reveals an approximately 2.1-fold enrichment of cross-family segment correspondences relative to the Pfam-Pfam background. Mapped fragments are strongly enriched in helix-dominant secondary structures with intermediate solvent exposure, occupy a narrowed sequence-context entropy regime, and show a significant shift toward lower fragment-associated contact order relative to random control windows, directly supporting a folding-compatible local structural regime. Matched annotation analyses show depletion from known domain-based interaction interfaces and mild depletion from strict DisProt disorder. Direct Start2Fold/HDX and{varphi} -value comparisons are limited in scale but connect the same fragment class to folding-related protection: stability annotations support local structural stabilization that may facilitate folding, whereas folding annotations and transition-state enrichments provide more specific evidence for folding initiation or pathway involvement. Together, these results support a physical-constraint interpretation of local structural conservation, in which reusable compact fragments contribute prominently to local stability and folding-relevant local structure, with folding initiation supported in specific subsets rather than as an exclusive mechanism.

Antifreeze proteins and peptides adhere to ice crystals and inhibit their growth, resulting in freezing point depression. AFP6 is an abundant antifreeze peptide in the blood serum of winter flounder (Pseudopleuronectes americanus). This 37-residue peptide forms an amphiphilic -helix with a hydrophobic ice-binding surface. In this study, prediction of the peptide structure suggested the formation of an antiparallel dimer burying the hydrophobic surfaces of both subunits. The oligomerisation of AFP6 was therefore examined in aqueous solutions using size-exclusion chromatography and chemical crosslinking along with ice-binding evaluation. Crosslinking of non-amidated AFP6, followed by denaturing electrophoresis, revealed bands consistent with monomers and dimers, whereas results for the amidated form and a recombinant mutant suggested additional larger forms. Bands consistent with the presence of monomers and dimers were also predominant following non-amidated AFP6 crosslinking under a variety of solution conditions, with smaller proportions of peptide assembled up to tetrameric sizes; however, the larger oligomers appeared to be absent from AFP6 treated with sodium bicarbonate. Isolated oligomers of non-amidated AFP6 produced ice crystal morphologies and thermal hysteresis comparable to those for the unfractionated peptide suggesting similar interaction with ice. Oligomerisation of AFP6 may allow greater solubility of this amphiphilic -helical peptide in aqueous solutions.

Human {beta}-defensin 1 (hBD-1) is a small, disulfide-constrained antimicrobial peptide produced predominantly in the kidney and urogenital tract, where it contributes to epithelial innate defense. Genetic variation in the gene encoding hBD-1 has been associated with altered expression levels and susceptibility to infections, including infection following solid organ transplantation, yet the molecular consequences of coding substitutions in the active, mature form remain largely undefined. Here, we integrate conventional genomics, protein bioinformatics, structural modeling, and molecular dynamics to investigate the molecular consequences of 93 human-observed hBD-1 missense variants. Evaluation of hBD-1 missense variants across sequence-, structure-, and dynamics-based dimensions reveals variant-specific molecular perturbation signatures that complement conventional and static biophysical predictions. By treating naturally occurring substitutions as a perturbation series across a compact defensin scaffold, this study identifies multiple molecular routes through which sequence variation can remodel an endogenous antimicrobial peptide and prioritizes variants for direct structure-function testing. These findings provide a mechanistic framework for investigating how naturally occurring hBD-1 variation may influence peptide behavior and, ultimately, antimicrobial function.

Designed binders are routinely filtered on predicted interface confidence, which reports how self-consistently a model places an interface, not where it lies. We designed 200 nanobody loops against the survivin BIR-domain groove, co-folded 196, and checked epitope contact in every predicted complex. Confidence showed no detectable association with epitope engagement ({rho} = +0.100, 95% CI -0.04 to +0.24, n = 196); its highest and lowest quartiles failed alike (odds ratio 0.97, 95% CI 0.41 to 2.26). It tracked instead the run's other confidence estimates ({rho} = +0.61). Twelve-seed replication separated two candidates the criteria had not: one reached the complete epitope in 10 of 12 runs, the other never. Across 155 predictions the top-ranked model was no better than one drawn at random from its five (P = 0.610), though only 3 of 31 failed on every model. Nothing was assayed; nb_cand_057 is an in silico candidate only.

Diffusion-based models have substantially advanced biomolecular structure prediction. Within this framework, restraint-guided inference (RGI) corrects the denoiser output at each reverse-diffusion step using a differentiable restraint loss, allowing available experimental information, stereochemical requirements, or a specified target conformational state to be incorporated into predictions without retraining the model. Here, we present RGI-Toolkit, a general-purpose Python library that makes RGI available for a range of structure predictors. The toolkit consolidates restraint definition, atom selection, loss evaluation, and coordinate optimization into a shared, backend-agnostic engine, limiting predictor-specific code to a thin preprocessing adapter and a single hook in the denoising loop. Restraints on distances, angles, and dihedrals, on deviations from reference structures, and on user-defined reaction coordinates, together with ligand conformer restraints, are all specified in a common configuration format. The toolkit currently supports six predictors, including AlphaFold3. Across all six predictors, conformer restraints eliminated chirality and cis/trans errors in the predicted ligands and improved their local geometry, while geometric and custom restraints systematically shifted the sampled protein conformations toward the specified target states. RGI-Toolkit thus provides a unified framework for integrating experimental information and prior geometric knowledge into structure prediction and for generating biomolecular structures with specified conformational and stereochemical properties.

Although protein toxins represent valuable pharmacological templates, predicting toxicity directly from primary sequences is inherently challenging because of their evolutionary dynamics. Active toxins and their benign homologues frequently share identical structural scaffolds and differ by only a few key residue substitutions, making global sequence similarity an unreliable indicator of function. Conventional predictors compound this challenge through flawed negative-set designs that sample non-toxic controls from arbitrary background proteins and reduce redundancy through identity-based clustering under random partitioning. This practice introduces global confounding shortcuts that inflate benchmark performance while failing to capture true functional boundaries. To address these limitations, we developed NeoToxPred, a sequence-only binary classifier that fine-tunes the pretrained ESM-C 600M protein language model end-to-end with a six-layer feed-forward classification head. Although the neural architecture is intentionally standard to ensure computational scalability, the core novelty of the framework lies in its rigorous training-data curation: (i) an orthologous negative set drawn from the same InterPro families under controlled taxonomic proximity to eliminate phylogenetic shortcuts, (ii) a length-stratified negative control set to eliminate sequence length as a predictive cue, and (iii) a strict family-disjoint splitting strategy to prevent performance inflation caused by pattern memorization. On held-out internal benchmarks, NeoToxPred achieved a Matthews correlation coefficient (MCC) of 0.897 and an F1-score of 0.949 on long sequences and an MCC of 0.869 on short peptides, decisively outperforming five recent predictors while maintaining stable performance across varying sequence lengths and evolutionary distances. In a genome-scale external validation using the previously unseen proteome of the redfin waspfish (Paracentropogon rubripinnis), NeoToxPred successfully recovered 11 of 16 empirically validated toxins without collapsing into a single-class prediction, whereas structure-dependent baselines identified substantially fewer positives. Computational alanine scanning further confirmed that the framework contextually localizes its predictions to functionally active residues. Taken together, these findings suggest that systematic negative-set design should be prioritized alongside algorithmic complexity in biological sequence classification.

High-throughput enzymology enables quantitative characterization of enzyme function across hundreds to thousands of sequence variants and experimental conditions. Nevertheless, the scale and complexity of these datasets create substantial challenges for analysis and quality control. High-Throughput Microfluidic Enzyme Kinetics (HT-MEK), for example, generates large microscopy datasets that must pass through multiple analytical stages, including image processing, initial-rate fitting, and kinetic modeling. Choices or errors made at any stage can propagate into the final kinetic parameters without being evident from fit statistics alone. Here, we provide a broadly applicable guide for analyzing high-throughput enzymology data from the HT-MEK platform, using Michaelis-Menten kinetics as a representative example. We first outline the conceptual workflow from raw fluorescence measurements to estimates of kcat, KM, and kcat/KM. We then discuss common experimental and analytical failure modes and present a framework for deciding when data should be refitted, filtered, qualified, or repeated. Finally, we provide a step-by-step workflow using Mercury, an open-source Python framework that integrates scalable HT-MEK data processing with traceable quality control and diagnostic visualization. This workflow preserves the connection between reported parameters and their underlying measurements and can be adapted to other kinetic models and biochemical assays.