Chemical Science

Ligand binding to intrinsically disordered proteins resists description in terms of conventional binding pockets, yet it can be analysed as a dynamic process in which ligands move across transient surface interaction sites. Here we characterise a pathway-based representation in which ligand binding is described as a sequence of transitions between residue-defined microstates, enabling ligand-specific effects to be distinguished from intrinsic properties of the peptide conformational ensemble. Using all-atom molecular dynamics simulations of A{beta}42 and the C-terminal region of -synuclein in complex with chemically diverse small molecules, we construct transition matrices that encode ligand movement across the peptide surface and use Markov state models to identify dominant binding pathways and relative binding propensities. Pairwise enrichment-factor and AUC analyses reveal strong conservation of the highest-ranked pathways across chemically diverse ligands, with enrichment factors of 15-45 for the top-ranked states and AUC values typically [≥]0.75, well above random expectation. These dominant pathways are also preserved across changes in pH and temperature, whereas a urea control, included as a non-specific binder, shows reduced enrichment, indicating that ligands primarily modulate pathway weights rather than define the underlying network topology. Ensemble docking across chemically diverse libraries further supports the presence of recurrent ligand-accessible hotspots within the peptide conformational ensemble. Building on this framework, we apply a prospective screening pipeline to A{beta}42, combining MSM-derived hotspots with sequence-based Ligand-Transformer scoring and Gnina docking across 1.66 million compounds, to nominate 19 candidates for prospective experimental evaluation. Together, these results indicate that disordered protein sequences give rise to conformational ensembles that exhibit characteristic binding pathways for small molecules.

Protein stabilization is a "Holy Grail" of biocatalysis, and stability design is an area of intense research interest. While it is increasingly feasible to effectively increase enzyme thermostability, optimization without compromising activity or selectivity remains a significant challenge. Here, we use full-atom protein sequence design with sidechain conditioning (FAMPNN) to engineer thermostable variants of the borneol dehydrogenase from Salvia rosmarinus (SrBDH1), an enzyme from a family where unselective enzymes dominate, and selectivity is determined by dynamical considerations. By combining FAMPNN design with residue conservation analysis and avoiding active site residues, we were able to computationally design SrBDH1 variants with up to 10 {degrees}C enhanced thermostability and strongly increased half-life time at elevated temperature, while retaining selectivity towards (+)-borneol. This design framework, integrating de novo and physics-based protein design tools, demonstrates that stability can be enhanced without disrupting functionally relevant dynamics, providing a route to engineer robust and selective biocatalysts. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=198 SRC="FIGDIR/small/719482v1_ufig1.gif" ALT="Figure 1"> View larger version (97K): org.highwire.dtl.DTLVardef@1a35073org.highwire.dtl.DTLVardef@f6c56dorg.highwire.dtl.DTLVardef@11b965forg.highwire.dtl.DTLVardef@2d6eef_HPS_FORMAT_FIGEXP M_FIG Graphical Abstract C_FIG

The LC3/GABARAP protein family is a promising target for selective inhibition of autophagy and for targeted protein degradation. LC3/GABARAP proteins are challenging targets for small-molecule drug development due to their long, shallow binding grooves. In this work, we evaluate multiple approaches to stabilizing the extended structure of the native binding motif, producing N-methylated peptides and stapled peptides with low nanomolar affinity. A crystal structure and molecular dynamics simulations support a model where the N-methylation pre-organizes the motif into an extended, strand-like structure. N-methylation allowed minimization of the binding motif to a tetrapeptide that retained sub-micromolar affinity while minimizing charge and overall molecular weight. The truncated, N-methylated tetrapeptide showed moderate passive permeability. These results highlight more drug-like space for the development of LC3/GABARAP ligands with high affinity and selectivity.

Long-range protein electron transfer (ET) often depends on tryptophan and tyrosine residues acting as radical relay sites. For example, cytochrome c peroxidase (CcP) generates a W191*+ radical to increase ET from cytochrome c (Cc) to the active center. W191 substitution to Tyr reduces ET rates, but introduction of an adjacent general base at position 232 (as Glu or His) recovers activity. E232 fluorination shifts the ET pH dependence to lower values, verifying that a hydrogen bond elevates the Y191* formal potential for effective ET. Photoinitiated ET between Zn-porphyrin (ZnP) CcP (ZnCcP) and Cc also depends on activating Y191 with a basic residue, but through a different mechanism than for the peroxide-driven system. In ZnCcP, pH dependencies and solvent isotope effects indicate that proton-coupled electron transfer to the basic residue and ZnP*+, respectively, facilitate Y191* formation. Replacing Cc with the irreversible oxidant [Co(NH3)5Cl]2+ isolates distinct protein radicals for characterization by Electron Paramagnetic Resonance (EPR) spectroscopy. Radical distributions reveal that W191*+ lies [~]15 mV in potential below ZnP*+ and that the two radicals exchange on a slow time scale despite their close separation. Remarkably, ZnCcP Y,G191:E,H232 variants propagate radicals differently to peripheral sites depending on the nature of the 232 residue. QM/MM calculations support radical exchange between ZnP*+/Trp*+ and the importance of a hydrogen bond to Y191* for maintaining a high potential to oxidize peripheral donors. These resolved reactivity patterns of CcP/ZnCcP have general relevance for engineering proton management to separate and migrate charge in proteins and potentially other molecular systems.

Pseudomonas aeruginosa is an adaptable organism, frequently found in chronic infections, and for which antimicrobial resistance is a growing concern. Therefore, there is an urgent need for alternative therapeutic strategies. Cationic antimicrobial peptides (AMPs) offer potent bactericidal activity but suffer from limited selectivity and potential host toxicity. To enhance species-specific targeting, we designed two prodrug variants of the AMP D-Bac8CLeu2,5 - EEEE-D-Bac8CLeu2,5 and ELEG-D-Bac8CLeu2,5 -- engineered for activation by the P. aeruginosa extracellular aminopeptidase PaAP. While both prodrug motifs effectively neutralized the positive charge of D-Bac8CLeu2,5 and prevented DNA-peptide complex formation, EEEE-D-Bac8CLeu2,5 showed negligible antimicrobial activity due to slow and incomplete activation. In contrast, ELEG-D-Bac8CLeu2,5 underwent rapid PaAP-mediated activation, restoring bactericidal activity in planktonic cultures and biofilms. PaAP contributed significantly to complete prodrug activation, particularly within biofilms, where the accumulation of partially activated intermediates correlated with biphasic killing kinetics. The prodrug showed reduced activity against other ESKAPEE pathogens, demonstrating selective activation by P. aeruginosa. Experiments selecting resistant bacteria revealed distinct mutations in lipopolysaccharide biosynthesis pathways for D-Bac8CLeu2,5 and the prodrug, with limited cross-resistance. These findings establish aminopeptidase-activated AMP prodrugs as a promising approach for species-selective antimicrobial therapy and highlight the feasibility of exploiting bacterial enzymes for controlled antimicrobial peptide activation. Table of contents graphic O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=99 SRC="FIGDIR/small/715093v1_ufig1.gif" ALT="Figure 1"> View larger version (35K): org.highwire.dtl.DTLVardef@4a5505org.highwire.dtl.DTLVardef@13e578org.highwire.dtl.DTLVardef@3e3080org.highwire.dtl.DTLVardef@e24266_HPS_FORMAT_FIGEXP M_FIG C_FIG

DNA-encoded libraries (DELs) enable ultra-large screening of billions of molecules simultaneously. However, various limitations of DELs have prompted interest in training machine learning (ML) models on these large datasets to extrapolate predictions to non-DEL compounds. A recent NeurIPS competition revealed that even top performing ML models trained on DEL data failed at generalizing to out-of-distribution (OOD) chemical space. We investigated whether integrating structural modeling could bridge this generalization gap. We systematically assessed state-of-the-art ML, docking, and co-folding methods with three biologically diverse protein targets screened against libraries containing multiple DEL synthesis formats, and show that while ML excels in-distribution, the optimal approach for OOD hit discrimination performance is both target and ligand dependent. We conclude that, regardless of performance reported in aggregated benchmarks, rigorous, system-dependent pilot testing is critical for reliable virtual screening predictions. We provide these workflows and analysis tools in an open-source package: DEL-iver. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=118 SRC="FIGDIR/small/719394v1_ufig1.gif" ALT="Figure 1"> View larger version (25K): org.highwire.dtl.DTLVardef@d9d299org.highwire.dtl.DTLVardef@913f59org.highwire.dtl.DTLVardef@1d5f69borg.highwire.dtl.DTLVardef@316102_HPS_FORMAT_FIGEXP M_FIG C_FIG

Elucidating the structure and dynamics of insulin and its analogs has been of broad interest, while presenting challenges due to the unique structural dynamics of insulin (composed itself of two multiply cross-linked peptides A and B) and its ability to assemble in a variety of oligomeric structures under physiological conditions. Here, we present two distinct X-ray crystallographic structures of the long-acting human insulin analog detemir (INSD) resolved in hexameric and dodecameric (or dihexameric) states at 2.85 [A] and 2.70 [A] resolution, respectively, using diffraction data collected under ambient temperature conditions. Characterization of the collective dynamics of these oligomers using the Gaussian Network Model (GNM) reveals several key features: (i) Oligomerization imparts high cooperativity in structural dynamics evidenced by dissection of the cross-correlations at various hierarchical levels; (ii) detemir monomers conformational flexibility is highly suppressed within oligomeric constructs, the effect being particularly strong in the dihexamer due to the asymmetric packing of the hexamers and the presence of myristoyl groups at B peptides termini whose interactions imparts further heterogeneities; and (iii) a number of key residues retain, however, their intrinsic dynamics, to be deployed upon release from the oligomers. We distinguish in particular residues serving as hinge sites that mediates the conformational dynamics of the asymmetric units (dimers) and monomers (I2A-V3A and Y19 A -C20A, and L11B-L15B and Y26B of the respective peptides A and B), or as anchors supporting structural stability (disulfide-bridge forming cysteines, plus selected residues such as L16A, G8B and R22B-F24B. Overall, this study provides a structural-dynamic framework for gaining new insights into the dynamics of long-acting analog INSD and helps identify actionable sites for modulating insulin (analogs) dynamics toward designing more effective therapeutics.

OXA-48 is a globally disseminated class D serine {beta}-lactamase that efficiently confers resistance to a range of {beta}-lactam antibiotics, including carbapenems, the most potent such agents versus Enterobacterales (Escherichia coli and relatives). Here we characterise the interactions of OXA-48 with the acyl-enzyme complex intermediates formed on its reaction with the carbapenems meropenem and ertapenem using X-ray crystallography and molecular dynamics (MD) simulations. X-ray crystal structures identify acyl-enzymes in both the {Delta}1-imine and {Delta}2-enamine pyrroline tautomeric forms. MD simulations show the epimeric {Delta}2 tautomers of meropenem and ertapenem to more frequently adopt binding poses competent for hydrolysis, i.e. with an appropriate orientation of the carbapenem 6-hydroxyethyl group and positioning of the water molecule required for deacylation; the results indicate that the {Delta}2 tautomers are preferred for deacylation over the {Delta}1-tautomer. MD simulations based on the crystal structures show that, compared to OXA-48, acyl-enzyme complexes of OXA-519 (a natural OXA-48 variant with a single Val120Leu substitution adjacent to the catalytic general base) more frequently sampled conformations favouring hydrolysis, or formation of the alternative {beta}-lactone deacylation product. MD simulations of complexes derived from quantum mechanics/molecular mechanics (QM/MM) simulations show the meropenem-derived {beta}-lactone product is better retained in the OXA-48 active site than hydrolysed meropenem, consistent with reversible {beta}-lactone formation. Overall, our results demonstrate how acyl-enzyme tautomerisation, dynamics and hydration collectively modulate degradation of 1{beta}-methyl carbapenems by class D {beta}-lactamases of the OXA-48 group, and how subtle changes in active site structure potentiate such effects in the OXA-519 variant.

Antimicrobial peptides (AMPs) that also form functional amyloids exhibit remarkable environmental sensitivity, yet the physicochemical rules governing their structural switching remain unresolved. Here, we investigate how surfactant charge and assembly dynamics regulate the antimicrobial-amyloidogenic transition of Uperin 3.5, a 17-residue amphibian AMP with pronounced conformational plasticity. Using an integrated approach combining all-atom molecular dynamics simulations with circular dichroism and thioflavin T fluorescence assays, we systematically probe the effects of surfactant identity, concentration relative to the critical micelle concentration (CMC), peptide stoichiometry and ionic strength. We show that -helical stabilisation and antimicrobial-like behaviour scale directly with surfactant charge: anionic Sodium dodecyl sulphate (SDS) induces the highest helicity in monomeric Uperin 3.5 ({approx}80-90%), followed by zwitterionic dodecyl-phosphocholine (DPC) ({approx}35-45%), while cationic Cetyltrimethylammonium bromide (CTAB) fails to stabilise secondary structure. This charge-ordered trend is mirrored in oligomer remodelling, with SDS driving the most rapid dissociation of {beta}-sheet tetramers, DPC inducing slower partial disassembly and CTAB exhibiting minimal effect. Above the CMC, micellar environments stabilise amphipathic -helical states and efficiently dissolve amyloid assemblies. In striking contrast, under below-CMC conditions, limited SDS availability combined with peptide crowding promotes cooperative aggregation, where surfactant monomers act as dynamic scaffolds that nucleate N-terminal {beta}-sheet interactions--an effect strongly accelerated by physiological salt. Large-scale simulations reveal mixed /{beta} aggregates whose formation is governed by electrostatic screening and surfactant-mediated co-assembly. Together, these findings establish surfactant charge and assembly state as quantitative, environment-dependent regulators of functional amyloidogenesis in antimicrobial peptides. More broadly, they suggest that controlled modulation of membrane-mimetic environments can be exploited to bias peptides toward antimicrobial or amyloidogenic states, offering conceptual avenues for therapeutic strategies targeting peptide misfolding and neurodegenerative disorders.

Cyclic peptides have emerged as a pivotal modality for next-generation therapeutics, due to their superior biocompatibility, high selectivity, and structural stability. While AI-driven peptide design has advanced rapidly, conventional optimization algorithms are often constrained by initialization biases, which impede the efficient exploration of the vast chemical space. Here, we propose a novel methodology that integrates the protein language model ESM-2 with cyclic permutation averaging of embeddings to resolve this bottleneck. This approach establishes a comprehensive "peptide space", a high-dimensional vector representation that encapsulates the physicochemical and structural attributes of cyclic peptides. Our analysis reveals that random sequence selection results in a heterogeneous distribution within this space, potentially underrepresenting specific functional regions. Conversely, navigating this defined peptide space enables the selection of libraries that uniformly span diverse molecular properties. In a proof-of-concept study designing binders for {beta}2-microglobulin ({beta}2m), we demonstrate that initial sequences uniformly sampled from our peptide space yield superior candidates more efficiently than those derived from random selection. Furthermore, this framework facilitates the quantitative assessment of mutational perturbations on global peptide properties, supporting rational decision-making for both broad exploration and local optimization. This "peptide space" concept provides a foundational framework for defining appropriate search boundaries and enhancing computational efficiency in AI-mediated drug discovery. Graphic Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=172 SRC="FIGDIR/small/710724v1_ufig1.gif" ALT="Figure 1"> View larger version (48K): org.highwire.dtl.DTLVardef@1dd903eorg.highwire.dtl.DTLVardef@128f941org.highwire.dtl.DTLVardef@1041e13org.highwire.dtl.DTLVardef@1527b25_HPS_FORMAT_FIGEXP M_FIG C_FIG

Mycobacterium tuberculosis fatty acid synthase I (Mtb FAS-I) is a multifunctional hexameric complex essential for fatty acid (FA) synthesis. The need of a hexameric structure for activity of the complex in Mtb remains elusive. Here, we model a conformation of the functionally active complex with acyl carrier protein (ACP) at ketoacyl synthase (KS). Our model reveals a crucial cross-dome dependence in the mechanism of FA synthesis at the condensation step. Using molecular dynamics simulation, we identify key ACP and KS residues that mantain persistent interactions. ACPs phosphopantetheine (PPT) arm adopts several conformations while accessing KSs catalytic pocket, including two distinct conformations that correlate with volumes of ACP and KS pockets. A PHE residue, reported as a gatekeeper of the KS pocket in other species, also shows open and closed orientations in our simulation. Our results provide crucial insights that are essential for a mechanistic undersanding of the Mtb FAS-I complex.

The standard set of 20 genetically-encoded amino acids (C20) exhibits a statistically non-random distribution in primarily two structurally-relevant physicochemical properties: hydrophobicity and molecular volume, and to a lesser extent charge. It remains an open question, however, whether evolutionary pressures similarly optimized the same alphabet for the distribution of functionally-relevant properties, such as reactivity. In this study, we used semi-empirical quantum chemistry simulations to calculate the highest occupied molecular orbital and lowest unoccupied molecular orbital (HOMO-LUMO) gaps for 84 xeno amino acids and constructed 10 million random 20-mer amino acid alphabets to determine where C20 fit amongst this background. The HOMO-LUMO gap measurements demonstrated that C20, similar to hydrophobicity and volume, also exhibits a non-random distribution. However, unlike hydrophobicity and volume, this distribution is non-random across an unevenly broad range. The results expand upon previous theory and suggest HOMO-LUMO gap energies as one synthetic biologists may consider when developing novel protein design tools or designing functional xeno amino acid alphabets. HighlightsO_LILifes amino acid alphabet is non-randomly distributed within an expanded computationally-generated chemistry space generated from large-scale quantum chemistry simulations. C_LIO_LIAmino acid alphabet coverage theory applies beyond structurally-relevant physicochemical descriptors to include functionally-relevant properties like reactivity as measured by frontier molecular orbitals C_LIO_LIFindings here provide a theoretical framework to guide the design of novel proteins and development of synthetic amino acid alphabets. C_LI

Cyclic peptides are recognized as versatile scaffolds for therapeutic and functional applications due to their structural stability and resistance to degradation. Despite this promise, systematic analysis and prediction of their thermal stability remain limited by fragmented data resources, inadequate sequence comparison methods, and the lack of cyclicity-aware computational models. We provide a comprehensive, multi-scale computational framework to characterize cyclic peptides. First, we unified four fragmented public repositories of cyclic peptides into a single largest curated resource of 930 cyclic peptides, Cyclome930. This integrates cyclic topology, sequence, experimental structural coordinates, and source organism annotations into a consistently featurized dataset. Cyclome930 thus expands the dataset of annotated cyclic peptides by [~]3.4 fold (from 276 to 930). Second, we developed a novel cyclic sequence alignment algorithm that explicitly accounts for rotational symmetry and knot topology, enabling more accurate scoring of sequence similarity than conventional linear alignments. Third, we investigate the thermal stability of cyclic peptides using extensive all-atom replica-exchange molecular dynamics (100ns; REMD) simulations, allowing conformational sampling across 298 K - 400 K and track its stress tensors with increasing temperature. Finally, these simulation-derived thermo-stability metrics were used to train a machine learning model to predict cyclic peptide melting points from sequence and topology (STop2Melt). Crucially, the model introduces cyclicity-aware embeddings derived from ESMc representations coupled with cyclic offset vector, capturing the peptides knot topology. STop2Melt achieved strong predictive performance on held-out peptides and outperforms baseline methods that neglect cyclic structure. Finally, we scored Cyclome930 (cyclic ligands) for critical mineral metal binding using a multi-classifier model (CritiCL). To our knowledge, Cyclome930 represents the first effort in peptide literature to integrate physics-based temperature ramped simulations, cyclic sequence similarity scoring, machine learning for thermal stability prediction and scoring them for critical metal binding. Cyclicity-aware computational toolchains (cyclome930.studio/) provide a foundational resource for computational design of stable cyclic peptide prototype libraries thereby annotating and expanding genomic islands linked to critical mineral recovery.

Emerging fungal pathogens represent a concerning threat to both global health and food security. In this study, we aimed to address our rising vulnerability to fungal pathogens through the development of the Fung-AI pipeline: an AI/ML-driven approach for antifungal discovery. A generative adversarial network (GAN) was trained to generate novel candidate antifungal peptide sequences. Next, in silico antifungal and hemolytic classifiers were built to further prioritize AI-generated peptides for experimental validation. From a pool of [~]10,000 candidates, thirteen peptides were selected for testing over two-stages of experimentation. Five peptides were found to display mild antifungal activity against the wheat pathogen, Fusarium graminearum, with minimal inhibitory concentrations (MICs) ranging from 250 {micro}g/mL to 500 {micro}g/mL. Four of the five peptides also showed activity against the human pathogen, Candida albicans (MIC: 500 {micro}g/mL). Two of our AI-generated antifungal peptides additionally demonstrated low cytotoxicity in HepG2 human liver carcinoma cells (LC50 > 704.2 {micro}g/mL) indicating that they may be useful as scaffolds for future optimization for therapeutic applications. None of our peptides were found to considerably inhibit the emerging pathogen C. auris, suggesting the need for pathogen-specific down-selection of candidate peptides. Overall, we present a proof-of-principle, generative-AI-based approach for the rapid design of de novo antifungal peptides.

Cytochrome P450s catalyze a diverse array of reactions including crosslinking of aromatic side chains in the biosynthesis of ribosomally synthesized and post-translationally modified peptides (RiPPs). ApyO is a cytochrome P450 enzyme that forms a C-C bond between two tyrosines in a YLY motif in the substrate ApyA, the precursor peptide of the RiPP aminopyruvatide. We utilized cell-free translation to generate ApyA variants and probe the substrate tolerance of ApyO. Through Alphafold-based modelling and in vitro assays, we show that ApyO accepts the 10 C-terminal residues of ApyA and requires a conserved Arg/Lys in the substrate peptide. Inspired by substrate sequences found in orthologous biosynthetic gene clusters, we substituted one of the tyrosine residues with a tryptophan and observed that ApyO catalyzed the formation of an N-C bond between the indole of Trp and the C{varepsilon}2 of Tyr. ApyO unexpectedly catalyzed formation of a C-O bond between the two tyrosine residues when we substituted the leucine residue in the YLY motif with tyrosine and tryptophan. We also show that a peptide containing a biaryl linkage and the C-terminal aminopyruvate displayed sub-nanomolar inhibitory activity against selected proteases. Overall, this study demonstrates plasticity in the manner of macrocyclization catalyzed by the P450 ApyO and provides a starting point for chemoenzymatic approaches towards producing diverse macrocyclic scaffolds.

We introduce the Serna Bio GenAI platform, a generative chemistry and multiparametric optimization platform for the design of RNA-targeting small molecules. Targeting RNA with small molecules has proven historically challenging but offers notable potential upsides, including access to unique mechanisms of action and the ability to target otherwise untargetable genes. We consider a major challenge here to be designing chemistry specific to RNA-targeting. Molecular design is a valuable application of AI in drug discovery, but many publicly available models use training data focused on protein-targeting - the modality best historically explored in drug discovery. We showcase the difference and value in building a specifically RNA-targeting platform, comparing its performance to state-of-the-art public chemical generators and experimentally validating its chemical designs in comparison to chemistry designed by a human expert.

Therapeutic peptides combine high target specificity with potent biological activity.1 However, treatment success is often limited by rapid clearance and the need for frequent injections.2, 3 This challenge is particularly acute for therapeutic peptides used in obesity, where clinical benefit must be balanced against dose-dependent adverse effects. In nature, these constraints are overcome by storing hormones as reversible fibrils,4 but pharmacokinetic control is essential for widespread adoption of bio-inspired self-assembled depots for therapeutic peptides. Here, we show that tuneable pharmacokinetics can be achieved and modelled by mapping the fundamental chemical parameters of reversibly self-assembly in vitro. We demonstrate this approach for the amylin analogue pramlintide. Amylin analogues are under development for the next generation of diabetes and obesity treatments, with improved mechanism of action e.g. preserving lean body mass.5-8 Pramlintide is an approved drug with a well-established safety profile, however, it has a comparable half-life to native amylin.8-12 In a pilot study, we achieve in vitro-in vivo correlation, increasing the half-life of pramlintide 20-82-fold in rats, while controlling burst release. These findings demonstrate that the optimisation of pharmacokinetics can be decoupled from peptide engineering, establishing a generalisable framework for generating long-acting peptide formulations by emulating native storage mechanisms.

Polypeptoids (poly-N-substituted glycines) are synthetic peptidomimetic polymers with their sidechains attached to the backbone amide nitrogen rather than the -carbon in natural peptides. Peptoids display pronounced sequence-dependent conformational flexibility arising from the absence of backbone hydrogen bonding and slow cis/trans {omega}-dihedral isomerization. Despite growing interest in peptoid-based biomaterials, a coarse-grained (CG) model compatible with the modern MARTINI 3 framework is not yet available, limiting mesoscale simulation of peptoid structure and self-assembly. In this work, we develop the first MARTINI 3 compatible peptoid CG forcefield, covering 19 commonly used residue types. Extensive all-atom reference simulations employing parallel bias metadynamics (PBMetaD) were performed to ensure converged sampling of {omega}-dihedral transitions. Bonded parameters were derived from atomistic distribution functions via direct Boltzmann inversion (DBI), while nonbonded interactions were primarily adopted from the standard MARTINI 3 parameter library. The resulting CG model reproduces structural and thermodynamic properties in close agreement with all-atom simulations, while providing up to 57-fold enhanced computational efficiency. To facilitate its adoption by the research community, we have integrated all parameters and workflows to the MARTINI-based martinize2 tool, enabling automated generation of MARTINI 3 peptoid structures and topologies. This work establishes a transferable and computational efficient framework for simulating large-scale peptoid confirmations, assemblies, membrane interactions, and nanostructure formation, and supports the rational design of next-generation sequence-specific functional peptoid-based materials.

Antimicrobial peptides (AMPs) are emerging as promising alternatives to conventional antibiotics, and growing evidence indicates a fundamental link between antimicrobial activity and amyloid-like self-assembly. Many AMPs are known to form amyloid-like fibrils, while several amyloidogenic peptides exhibit intrinsic antimicrobial properties, suggesting shared underlying physicochemical determinants such as amphipathicity, {beta}-sheet propensity, and charge distribution. However, the rational design of peptides that simultaneously encode these dual functionalities remains a significant challenge. Here, we present amyAMP, a generative deep-learning framework based on a Wasserstein generative adversarial network with gradient penalty (WGAN-GP), designed to learn and generate peptides with integrated antimicrobial and amyloidogenic properties. Trained on curated datasets of antimicrobial and amyloid-forming peptides, amyAMP captures the latent sequence-property relationships governing dual functionality. Statistical and latent-space analyses demonstrate that the generated peptides closely overlap with biologically relevant peptide space while remaining distinct from random sequences, indicating successful learning of key biochemical features. To validate functional behavior, we performed extensive coarse-grained molecular dynamics simulations to probe membrane interaction, peptide self-assembly, and membrane disruption. The simulations reveal rapid membrane adsorption, stable amphipathic insertion, and strong peptide-peptide aggregation. Notably, cooperative clustering of peptides on membrane surfaces induces membrane thinning and curvature perturbations, highlighting a mechanistic coupling between aggregation and antimicrobial activity. Collectively, these results establish that amyAMP effectively captures the shared physicochemical principles underlying antimicrobial action and amyloid-like self-assembly. This work provides a generalizable framework for the AI-guided design of multifunctional peptides to advance the development of next-generation therapeutics targeting antimicrobial resistance.

Bacteriophytochromes (BphPs) find increasing interest as near-infrared (NIR) labels for imaging. Applications range from whole animal imaging to cell-to super-resolution microscopy. Here we present a comprehensive study of a BphP from Rhizobium etli (ReBphP) allowing mutant-based insights into BphP photophysics. This is complemented by QM/MM-optimized deep-learning structure predictions of ReBphP and variants in their photoswitched states, rationalizing the effects of variants. Pertaining to imaging, based on our study we identify a bright and far red-shifted BphP for imaging in mammalian cells as well as a fluorescent photoswitching BphP. Utilizing the latter, we introduce photoswitching fluorescence background suppression for in vivo whole animal fluorescence imaging achieving higher contrast over background than possible utilizing non-switching labels.