Protein language models (pLMs) capture evolutionary sequence constraints but are limited in modeling underrepresented functional classes due to training data imbalance. Metalloproteins constitute a fundamental but sparsely represented class in sequence databases. We therefore assess whether structure-conditioned synthetic sequences can be used to specialize pLMs toward metal-binding functionality. We fine-tuned the generalist model ProtGPT2 on synthetic sequences generated by the inverse-folding model ProteinMPNN, constructing training sets with controlled variation in size and diversity. Fine-tuning increased recovery of canonical metal-binding motifs from 43% in the baseline model to 91% in the fine-tuned models. Generated sequences retained high predicted structural confidence and structural similarity to known folds, despite low sequence identity. Analysis of latent representations from ProtGPT2 indicated that fine-tuned models occupy distinct regions of embedding space relative to both the baseline model and structure-conditioned sequences, consistent with partial incorporation of structural constraints while preserving sequence diversity. A multi-step filtering pipeline applied to sequences lacking canonical motifs identified candidate metal-binding sites in four-helical bundle topologies not detected in a non-redundant subset of Protein Data Bank structures or in AlphaFold-predicted proteomes.  https://doi.org/10.5281/zenodo.18672158  https://huggingface.co/gsgueglia 

Bacillus subtilis is an important chassis for biotechnology, but its use in multiplex genome engineering is limited by low natural transformation efficiency. Here, we compared inducible promoter systems for synthetic activation of the competence regulator ComK and evaluated their effects on the comG operon competence reporter and transformation efficiency. Xylose- and mannitol-inducible systems outperformed IPTG-based constructs and shifted 96-99% of cells into a reporter-positive competent state. However, reporter activation alone did not predict transformation potential. Optimization of culture density and induction timing increased transformant yield 45-fold relative to the initial protocol and 2800-fold relative to the conventional Spizizen method. Disruption of native competence regulatory genes did not improve performance and often reduced transformation output, highlighting the importance of endogenous regulatory circuitry. Using the optimized strain and protocol, we achieved co-transformation frequencies of 11-18% and constructed multiplex spore-display libraries containing fluorescent protein fusions integrated at multiple loci. Screening identified strong dual-display combinations and showed that cargo loading depends on anchor protein, integration locus, and genetic background. SscA fusions supported the highest display capacity and promoted synergistic co-display. Together, these results show improvements in natural transformation-based genome engineering in B. subtilis and provide insight into the construction of multifunctional engineered spores.

Synthetic biology has driven growing interest in engineered bacteria for cancer therapy. Salmonella Typhimurium stands out as one of the most promising live biotherapeutic products (LBPs) because of its innate tumor-targeting and colonizing abilities. However, the pathogenicity and inconsistent performance of wild-type S. Typhimurium have hindered further clinical translation. Recent advances in programmable and modular genetic redesign now enable precise reprogramming of bacterial functions, facilitating the creation of customized LBPs with enhanced safety and efficacy. In this review, we propose a systematic engineering framework with adaptive optimization and therapeutic enhancement for transforming S. Typhimurium into an effective anticancer LBP. This integrated strategy encompasses multiple facets, including strain attenuation, targeted enhancement, colonization optimization, therapeutic production, lysis-controlled release, and auxiliary modules. This review summarizes the key methodologies for engineering S. Typhimurium, highlighting its evolution from a pathogen to an antitumor vehicle. Furthermore, we summarize the multimodule collaborative engineering design and applications and call for more combinatorial strategies to enhance therapeutic efficacy. We also discuss the challenges and bottlenecks of engineered S. Typhimurium from the perspectives of genetic stability and clinical translation. Finally, we highlight how these synthetic biology advances are refining the mechanistic understanding of bacteria-mediated tumor therapy and paving the way toward safer, more effective, and clinically controllable anticancer strategies.

The universe of possible protein sequences is astronomically large, yet our understanding of the sequence-structure relationship is confined to the infinitesimal fraction used currently by life. Determining whether "foldable" architectures are rare singularities or accessible solutions is critical for understanding protein evolution and designing novel proteins. Here, we map the structural landscape of random sequence space by screening one million synthetic proteins using a high-throughput in vivo FRET biosensor. We reveal that this space is structurally heterogeneous, populated not only by disordered chains and stress-inducing aggregates but also by "benign" compact structures that resemble globular proteins and evade cellular chaperone responses. By training machine learning models on these phenotypes, we show that structural potential is learnable and generalizes to natural proteomes. These findings demonstrate that biology-like folds are accessible from random sequences with surprising frequency, providing data required to expand generative protein design beyond evolutionary priors.

High-quality datasets that span broad sequence diversity are essential for understanding protein sequence-function relationships beyond local mutational landscapes. Here, we applied Growth-based Quantitative Sequencing (GROQ-seq) to measure function across an 11,722 member protease library, comprised of natural homologs and AI-shrunken variants. This library spans vast sequence diversity, with Levenshtein distances of up to 245 and a mean pairwise sequence identity of 41% to TEV protease S219V. We identified sequence-divergent TEV protease homologs that preserve function against the native TEV protease substrate. These findings reveal the robustness of protease activity across highly diverse sequences. Here, we demonstrate the aptitude of the GROQ-seq assay for screening large, diverse protein libraries for function, enabling efficient data generation at scale for training machine learning models across broad sequence landscapes.

Scalable genetic circuits are essential for implementing complex functions in living cells. Toward this goal, RNA regulators can provide a much-needed parts library with added benefits of low metabolic load, design flexibility, and logic capacity. However, despite the great potential of synthetic RNA circuits, constructing such circuits with wide dynamic ranges and multiplexed regulatory cascades remains a challenge. To address this, we introduce RATEX (Ribosome-Assisted Transcriptional EXpression controller) by integrating a translation-to-transcription converter with synthetic RNA regulators, enabling a compact and scalable RNA-programmed circuit architecture. The RATEX platform repurposes a large library of well-characterized translation regulators with up to 1,492-fold gene regulation, while leveraging natural ribosome-mediated sensing of diverse environmental inputs, such as metabolites. We demonstrated multi-input logic processing with up to a 6-input OR logic gate for RNA inputs and hybrid 3-input logic gates to sense diverse metabolite and small-molecule inputs alongside RNA signals. Signal amplification with multiplexed combinatorial control of RNA outputs was achieved through multiplexed signaling cascades. Finally, the RNA- and metabolite-sensing 3-input AND gates were used to control cellular morphology and intracellular spatial organization. Together, the RATEX platform, with its scalable and modular architecture, offers a broad potential design space for synthetic biology and biotechnology.