Genome streamlining and pathway refactoring are powerful strategies for constructing controllable microbial chassis for both fundamental studies and applications. While rational design benefits from reduced genetic complexity, adaptive laboratory evolution (ALE) thrives on metabolic redundancy, creating a mismatch between optimal hosts for design and evolution. Here, we introduce a dual chassis framework (DUET) in which rational pathway construction and adaptive evolution are first carried out in an evolution‑competent host, and the resulting optimized designs are subsequently transferred into a genetically stable chassis for deployment. Using the naturally evolvable bacterium Acinetobacter baylyi ADP1 and its genome-stabilized derivative (ISx), we applied this framework to the β-ketoadipate pathway, a central hub for aromatic compound catabolism. We first streamlined the native network by deleting individual pathway branches and then engineered a minimal synthetic route that merges protocatechuate and catechol metabolism. Subsequent ALE enabled efficient growth through this synthetic pathway, and reverse-engineering identified key adaptive mutations underlying functional recovery. Both the synthetic pathway and the mutations were transferred unchanged into ISx, where robust growth was maintained without further adaptation. These results demonstrate that DUET enables portable, host-independent deployment of rational metabolic streamlining combined with evolution, providing a generalizable strategy for building reduced yet robust microbial platforms.

Encapsulins are prokaryotic self–assembling protein nanocages with promise as nanovaccine scaffolds. Their utility as modular platforms require tolerance to surface engineering, high-yield soluble production, formulation stability, and controlled antigen (co–)display. Herein, a previously uncharacterized encapsulin from Alkaliphilus metalliredigens is engineered into a SpyCatcher–decorated nanoscaffold (Am–S) that enables controlled surface display of SpyTagged antigens. Cryo-EM confirms that the native encapsulin forms a T = 1 icosahedral nanocage, and that C-terminal SpyCatcher fusion yields Am–S without compromising nanocage assembly, symmetry, or structural integrity. Notably, Am–S exhibits high-yield soluble production in Escherichia coli, remains monodisperse after freeze–thaw and extended storage, and supports efficient SpyTagged peptide conjugation for single– and multi–antigen display. As a proof–of–concept, Am-S is functionalized with Alzheimers disease-associated beta–amyloid and/or hyperphosphorylated tau epitopes to generate single–target nanocages displaying either antigen and dual–target mosaic nanocages co–displaying both. In mice, Am–S antigen display enhances antigen–specific IgG responses relative to free antigens and induces predominantly IgG1–biased humoral immunity. Mosaic nanocages elicit antibodies against both targets, with immune sera selectively recognizing amyloid–beta and phosphorylated tau-associated pathology in ex vivo brain sections from Alzheimers disease mouse models. These findings position Am–S as a manufacturable scaffold for developing multi–targeting nanovaccines against complex diseases.

While bacterial sensor histidine kinases (SHKs) are widespread as natural molecular biosensors, tools for high-throughput characterization of SHK signaling phenotypes are limited, hindering wide scale implementation of bacterial-based sensing. Here, we developed a synthetic two-component signaling system that reports chimeric SHK signaling via a standardized fluorescence readout. With this synthetic system, we screened a library of chimeric DcuS/EnvZ SHKs to characterize sequence-function relationships within in the DcuS sensory and transmembrane domains. We quantified the effects of 1,173 mutations on signaling outputs in the presence of fumarate, a native DcuS ligand, as well as aspartate for which DcuS has minimal affinity for. We identified eleven positions across the DcuS domains which significantly alter aspartate responsiveness and selectivity and further observed a role for cytoplasmic N-terminal residues in determining signaling outputs. In future studies, this framework will expedite design of biosensors for novel ligands by enabling high-throughput screening of mutagenized libraries of natural SHKs.

Lignocellulosic biomass represents a promising renewable feedstock for sustainable biorefinery applications, yet efficient enzymatic saccharification remains challenging due to the recalcitrant structure of plant cell walls. This study presents a comprehensive comparative analysis of enzymatic activities, saccharification performance, and secretome composition of six fungal species cultivated under solid-state fermentation (SSF) and submerged fermentation (SmF) conditions using untreated flax shives as substrate. While SmF yielded approximately 4-fold higher total protein concentrations (0.38 ± 0.13 g.L-1 vs. 0.08 ± 0.02 g.L-1), SSF-derived enzymes demonstrated superior specific enzymatic activities, particularly for endo-xylanase and endo-cellulase, resulting in more efficient biomass saccharification. Proteomics analysis revealed distinct secretome profiles between fermentation modes, with SSF showing higher proportions of polysaccharide metabolism proteins (71.0%) compared to SmF (49.3%), while SmF exhibited greater enzyme diversity including more lytic polysaccharide monooxygenases (LPMOs) and auxiliary activity enzymes. Trichoderma species consistently demonstrated the highest saccharification efficiency, with glucose yields reaching 2.37 mM under SSF conditions. A Scheffe simplex-lattice mixture design comprising 65 enzyme cocktail combinations revealed significant synergistic interactions between several cocktails, with the binary mixture of Trichoderma 2SA21 and P. chrysogenum achieving 54% synergy - in terms of higher sugar release above expectations - and the highest total monosaccharide release (1.80 mM). These findings provide practical guidance for developing cost-effective enzyme cocktails for lignocellulosic biorefinery applications, emphasizing the importance of fermentation mode selection and strategic strain combination over enzyme supplementation complexity. The methodology established here, combining systematic screening, comparative proteomics, and statistical mixture design, offers a robust framework for optimizing fungal enzyme systems across diverse biomass substrates.

Liquid–liquid phase separation (LLPS) plays a central role in cellular regulation, with its dysregulation linked to numerous diseases. LLPS is also increasingly implicated in various biological contexts, such as virus replication. These findings have driven the development of numerous computational predictors to screen and identify phase-separating proteins from sequence and/or structural models. Despite the need for these tools, their performance across diverse biological contexts remains incompletely understood, complicating tool selection and result interpretation. We performed a systematic comparative analysis of 9 LLPS prediction algorithms using multiple curated datasets comprising both LLPS-positive (LLPS+) and LLPS-negative (LLPS−) proteins. The datasets span multiple biologically relevant scenarios, including intrinsically disordered proteins, folded proteins, proteins with LLPS-abolishing variants, benchmark datasets, and viral proteins. We observed substantial variability in predictive performance across algorithms when assessing proteins of different structural classes such as LLPS+ and LLPS− folded proteins, LLPS-abolishing mutations, and viral proteins.  These results demonstrate that LLPS predictor performance is strongly context dependent, leading to different predictors being optimal for different biological questions. For overall protein assessment, DeePhase and MolPhase provided the most consistently accurate predictions, being the least impacted by structural bias. For assessing the impact of small mutations on LLPS propensity, PSPHunter, a built-for-purpose algorithm, reliably predicts mutation impacts, with structure-informed algorithms PSPire and PICNIC also providing strong insight. Across all evaluated datasets, the findings highlight the need for well-benchmarked training and testing data that encompass a broad and diverse range of protein classes.

Bacterial-mediated cancer therapy shows therapeutic potential yet remains constrained by bioavailability and biosafety challenges. We introduce a scalable, modular biosurface-engineering platform for functionalizing wild-type magnetotactic bacteria (Magnetospirillum magneticum, AMB-1), enabling the integration of diverse materials, including small molecules, polymers, nanoparticles, and metal–organic frameworks. As a proof-of-concept, AMB-1@Fe3+-PDA (poly dopamine) exemplifies a bacteria-mediated therapeutic strategy that merges tumor-targeted delivery, immunosuppressive tumor microenvironment (TME) reprogramming, and innate immune activation with photothermal-chemodynamic therapy. The Fe3+-PDA coating counteracts deep-tissue immunosuppression and primes adaptive immunity, while AMB-1 enhances penetration into resistant TME niches. These mechanisms synergistically suppress aggressive 4T1 breast tumor progression, metastasis, and recurrence while establishing immunological memory. In murine models, two treatment cycles prolonged median survival from 45 to 67 days and elicited systemic antitumor immunity, including abscopal effects targeting untreated distant tumors. This platform establishes a bridge between materials science and synthetic biology, providing a generalizable blueprint for advancing the clinical translation of engineered bacterial therapies.

Protein-based biosensors offer unique advantages over conventional analytical methods by enabling real-time detection of target analytes with minimal sample preparation. However, efficiently coupling molecular recognition to a reliable output signal remains a key challenge in biosensor design. Here, we present a plug-and-play strategy using the de novo switch platform, LOCKR, which enables direct transduction of a binding event into a defined signal output. The LOCKR architecture supports modular reconfiguration: the recognition domain can be swapped to detect a desired analyte, while the reporter module can be interchanged to tune the output format. We expand the range of LOCKR-compatible readouts beyond split luciferase to include ratiometric Förster-type resonance energy transfer and β-lactamase-based colorimetry. By integrating computationally designed high-affinity binders as interchangeable recognition elements, we demonstrate sensitive detection of glucagon, neuropeptide Y, and peptide YY with limits of detection in the picomolar range. Together with the expanding landscape of de novo designed binders and reporters, the LOCKR platform bridges the gap between molecular recognition and signal generation, enabling versatile biosensor development tailored for user-defined applications.

Genetic constructs meant for metabolic engineering in nonmodel microbes often use similar genetic parts to those familiar to E. coli work. The typical workflow is to clone these parts into plasmids in E. coli before they are transferred to the nonmodel host or its genome. In many cases, the metabolic burden of these constructs is stronger in the E. coli cloning phase of the workflow than in the eventual host, possibly resulting in mutation or other failure during cloning. Here, we apply generic knockdown of a range of popular expression systems, using CRISPR interference, by targeting guide RNAs to either promoters or RBSs that are commonly used in metabolic engineering. Generic targeting of a constitutive promoter series, combined with genome integration of CRISPR components, allows the use of only one or a few specific cloning strains to achieve strong knockdown of a wide range of constructs. Further, we present a recombinase-based workflow for easily adding guide RNAs with custom targets, so that users can knock down any desired promoter or ORF. Together, this group of strains comprises easy-to-use cloning strains meant for increasing success rates of difficult or burdensome cloning reactions, ultimately allowing more ambitious genetic constructs to reach their intended context.

Understanding how sequence context influences the likelihood of mutation at a given genomic locus is critical for deciphering evolutionary processes. It is well established that the immediately adjacent bases influence mutation rates, but the role of more distal bases remains poorly understood. Here, we analyze over 100,000 mutations from 32 E. coli mutation accumulation experiments, encompassing strains with varying DNA proofreading and mismatch repair capabilities. By quantifying the frequency of each nucleotide up to 6 bp around mutation sites, we reveal complex mutational biases that extend beyond the immediately adjacent bases and are unique to each type of base pair substitution. Furthermore, which sequence contexts contribute most to mutational bias depends on what repair mechanisms are active and which strand serves as the leading versus lagging template during replication. Mononucleotide runs are prominent mutational hotspots–we systematically characterize which types of runs are the most mutagenic, including a previously undescribed hotspot for G:C→C:G transversions that can increase their frequency by up to four orders of magnitude. Remarkably, extending our analysis to 1,000 bp from mutation sites reveals that sequence context can influence mutational bias over unexpectedly long distances. These findings expose the intricate interactions between extended sequence context and DNA repair systems that shape spontaneous mutagenesis, and shed light on the mechanistic origins and evolutionary consequences of mutational signatures.

Understanding the relationships between genotype and phenotype is key to many areas of biological research and to the development of synthetic cells. We describe an image-based screening and sorting workflow that explores the phenotypes of gene-expressing vesicles within nonclonal populations and selects the desired variants. Using automated confocal microscopy and real-time, neural network–assisted image analysis, we demonstrate that liposomes can be selected for fluorescence intensity, protein localization, membrane morphology, and dynamic behaviors, and their phenotype can be linked to genetic content. This approach could substantially accelerate the evolution of cellular functions in a minimal synthetic context.

For decades, natural product biosynthetic enzymes have challenged conventional logic and expanded the repertoire of biological catalysis. Understanding these enzymes that catalyze such diverse and powerful reactions is critical for understanding biological pathways, leveraging biomimetic or biocatalytic chemistry, and discovering novel natural products. Uncovering these enzymes is a challenging and historied pursuit, where the landscape has shifted greatly with advances in bioinformatics and unprecedented access to large datasets. These new omics technologies have unlocked new strategies, complementing older strategies, for the prioritization of candidate enzymes for characterization. This Highlight examines several established and emerging strategies for uncovering novel enzymatic chemistry in the context of natural product biosynthesis, focusing on representative examples that define current capabilities and outline key challenges.

Accurate prediction of enzymes' optimal catalytic temperature (Topt) is crucial in biotechnology, as enzymes with extreme Topt values are highly desirable for reactions at extreme temperatures and for their general stability. However, experimental determination of Topt is costly, labor-intensive, and time-consuming. Meanwhile, existing computational methods suffer from small and imbalanced datasets, suboptimal predictions at extreme temperatures, and insufficient validation. In this study, we address these challenges by expanding the Topt dataset and validating on an independent test set based on sequence similarity. We further tackle these limitations by comparing multiple resampling techniques to improve predictions at extremes and by considering diverse protein representations and multiple machine learning architectures. Overall, the best performing models reached R2 approximately 0.64 with MAE approximately 7-8 degrees C, while extreme resampling improved tail performance, reducing tail MAE by up to approximately 1.8 degrees C. Notably, our models show improved performance over state-of-the-art prediction models. We also demonstrate that accurate prediction of Topt is achievable even in the absence of organism growth temperature (OGT). Our Topt prediction models are made freely available as AdventML on GitHub.

Artificial intelligence (AI) has transformed protein engineering by leveraging deep learning, protein language models, and knowledge graphs to decode relationships between sequence, structure, and function. Models like AlphaFold2 achieve near-experimental accuracy in structure prediction, while transformer-based language models facilitate de novo sequence design under functional constraints. AI enhances therapeutic protein engineering, enzyme catalysis, and synthetic biology, accelerating the transition from in silico design to experimental validation. These advances accelerate experimental validation across healthcare and industrial biotechnology. Despite algorithmic successes, challenges remain in model interpretability, training data biases, and experimental validation rates. This review examines the computational methodologies shaping protein design, benchmarking metrics, and the integration of machine learning with experimental pipelines.