Multi-layered computational gene networks by engineered tristate logics.

So far, biocomputation strictly follows traditional design principles of digital electronics, which could reach their limits when assembling gene circuits of higher complexity. Here, by creating genetic variants of tristate buffers instead of using conventional logic gates as basic signal processing units, we introduce a tristate-based logic synthesis (TriLoS) framework for resource-efficient design of multi-layered gene networks capable of performing complex Boolean calculus within single-cell populations. This sets the stage for simple, modular, and low-interference mapping of various arithmetic logics of interest and an effectively enlarged engineering space within single cells. We not only construct computational gene networks running full adder and full subtractor operations at a cellular level but also describe a treatment paradigm building on programmable cell-based therapeutics, allowing for adjustable and disease-specific drug secretion logics in vivo. This work could foster the evolution of modern biocomputers to progress toward unexplored applications in precision medicine.

From waste to strength: Tailor-made enzyme activation design transformation of denatured soy meal into high-performance all-biomass adhesive.

Given the severe protein denaturation and self-aggregation during the high-temperature desolubilization, denatured soy meal (DSM) is limited by its low reactivity, high viscosity, and poor water solubility. Preparing low-cost and high-performance adhesives with DSM as the key feedstock is still challenging. Herein, this study reveals a double-enzyme co-activation method targeting DSM with the glycosidic bonds in protein-carbohydrate complexes and partial amide bonds in protein, increasing the protein dispersion index from 10.2 % to 75.1 % improves the reactivity of DSM. The green crosslinker transglutaminase (TGase) constructs a robust adhesive isopeptide bond network with high water-resistant bonding strength comparable to chemical crosslinkers. The adhesive has demonstrated high dry/wet shear strength (2.56 and 0.93 MPa) for plywood. After molecular recombination by enzyme strategy, the adhesive had the proper viscosity, high reactivity, and strong water resistance. This research showcases a novel perspective on developing a DSM-based adhesive and blazes new avenues for changes in protein structural function and adhesive performance.

Exosomal long non-coding RNA AU020206 alleviates macrophage pyroptosis in atherosclerosis by suppressing CEBPB-mediated NLRP3 transcription.

Recent studies have suggested exosomes (EXO) as potential therapeutic tools for cardiovascular diseases, including atherosclerosis (AS). This study investigates the function of bone marrow stem cell (BMSC)-derived exosomes (EXO) on macrophage pyroptosis in AS and explores the associated mechanism. BMSC-EXO were isolated from healthy mice and identified. RAW264.7 cells (mouse macrophages) were exposed to oxLDL to simulate an AS condition. BMSC-EXO treatment enhanced viability and reduced lactate dehydrogenase release of macrophages. An animal model of AS was established using ApoE-/- mice. BMSC-EXO treatment suppressed plaque formation as well as macrophage and lipid infiltration in mouse aortic tissues. Moreover, BMSC-EXO decreased concentrations of pyroptosis-related markers interleukin (IL)-1ß, IL-18, cleaved-caspase-1 and gasdermin D in vitro and in vivo. Long non-coding RNA AU020206 was carried by the BMSC-EXO, and it bound to CCAAT enhancer binding protein beta (CEBPB) to block CEBPB-mediated transcriptional activation of NLR family pyrin domain containing 3 (NLRP3). Functional assays revealed that silencing of AU020206 aggravated macrophage pyroptosis and exacerbated AS symptoms in mice. These exacerbations were blocked upon CEBPB silencing but then restored after NLRP3 overexpression. In conclusion, this study demonstrates that AU020206 delivered by BMSC-EXO alleviates macrophage pyroptosis in AS by blocking CEBPB-mediated transcriptional activation of NLRP3.

A strong and tough supramolecular assembled ß-cyclodextrin and chitin nanocrystals protein adhesive: Synthesis, characterization, bonding performance on three-layer plywood.

The development of a biomass adhesive as a substitute for petroleum-derived adhesives has been considered a viable option. However, achieving both superior bonding strength and toughness in biomass adhesives remains a significant challenge. Inspired by the human skeletal muscles structure, this study reveals a promising supramolecular structure using tannin acid (TA) functionalized poly-ß-cyclodextrin (PCD) (TA@PCD) as elastic tissues and chitin nanocrystals (ChNCs) as green reinforcements to strengthen the soybean meal (SM) adhesive crosslinking network. TA@PCD acts as a dynamic crosslinker that facilitates reversible host-guest interactions, hydrogen bonds, and electrostatic interactions between adjacent stiff ChNCs and SM matrix, resulting in satisfactory strength and toughness. The resulting SM/TA@PCD/ChNCs-2 adhesive has demonstrated satisfactory wet and dry shear strength (1.25 MPa and 2.57 MPa, respectively), toughness (0.69 J), and long-term solvents resistance (80 d). Furthermore, the adhesive can exhibit desirable antimildew characteristics owing to the phenol hydroxyl groups of TA and amino groups of ChNCs. This work showcases an effective supramolecular chemistry strategy for fabricating high-performance biomass adhesives with great potential for practical applications.

Engineered poly(A)-surrogates for translational regulation and therapeutic biocomputation in mammalian cells.

Here, we present a gene regulation strategy enabling programmable control over eukaryotic translational initiation. By excising the natural poly-adenylation (poly-A) signal of target genes and replacing it with a synthetic control region harboring RNA-binding protein (RBP)-specific aptamers, cap-dependent translation is rendered exclusively dependent on synthetic translation initiation factors (STIFs) containing different RBPs engineered to conditionally associate with different eIF4F-binding proteins (eIFBPs). This modular design framework facilitates the engineering of various gene switches and intracellular sensors responding to many user-defined trigger signals of interest, demonstrating tightly controlled, rapid and reversible regulation of transgene expression in mammalian cells as well as compatibility with various clinically applicable delivery routes of in vivo gene therapy. Therapeutic efficacy was demonstrated in two animal models. To exemplify disease treatments that require on-demand drug secretion, we show that a custom-designed gene switch triggered by the FDA-approved drug grazoprevir can effectively control insulin expression and restore glucose homeostasis in diabetic mice. For diseases that require instantaneous sense-and-response treatment programs, we create highly specific sensors for various subcellularly (mis)localized protein markers (such as cancer-related fusion proteins) and show that translation-based protein sensors can be used either alone or in combination with other cell-state classification strategies to create therapeutic biocomputers driving self-sufficient elimination of tumor cells in mice. This design strategy demonstrates unprecedented flexibility for translational regulation and could form the basis for a novel class of programmable gene therapies in vivo.

Selective knowledge sharing for privacy-preserving federated distillation without a good teacher.

While federated learning (FL) is promising for efficient collaborative learning without revealing local data, it remains vulnerable to white-box privacy attacks, suffers from high communication overhead, and struggles to adapt to heterogeneous models. Federated distillation (FD) emerges as an alternative paradigm to tackle these challenges, which transfers knowledge among clients instead of model parameters. Nevertheless, challenges arise due to variations in local data distributions and the absence of a well-trained teacher model, which leads to misleading and ambiguous knowledge sharing that significantly degrades model performance. To address these issues, this paper proposes a selective knowledge sharing mechanism for FD, termed Selective-FD, to identify accurate and precise knowledge from local and ensemble predictions, respectively. Empirical studies, backed by theoretical insights, demonstrate that our approach enhances the generalization capabilities of the FD framework and consistently outperforms baseline methods. We anticipate our study to enable a privacy-preserving, communication-efficient, and heterogeneity-adaptive federated training framework.