Cells and Materials Are Learning to Compute: The Rise of Engineered Biocomputability

Living systems are, at their core, computational machines. From single molecules inside a cell to entire microbial populations, biology continuously senses a shifting barrage of chemical and physical signals, interprets them, and produces functional outputs with remarkable fidelity. Scientists have long admired this capability; now they are learning to reprogram it. A new review published in Nature Chemical Engineering by Ryan Gharios, Murial L. Ross, Annabella Li, Shivani P. Kottantharayil and Cole A. DeForest of the University of Washington maps the rapidly expanding field of biocomputation, in which cells, proteins, nucleic acids and small molecules are hijacked, evolved and repurposed to perform programmable logic. The authors argue that two research communities, synthetic biology and stimuli-responsive biomaterials, have been converging on the same underlying goal from opposite directions, and that their union is now producing systems with unprecedented complexity and multiplexability.

The conceptual framework at the heart of the review is Boolean logic, the same binary mathematics that underpins silicon computing. Inputs such as a drug, a light pulse, a metabolite or a disease-associated enzyme are treated as binary signals, and engineered circuitry combines them through operators such as AND, OR and NOT to generate a defined output, whether that is gene expression, protein release, material degradation or a therapeutic response. The review opens by cataloguing logical operators and their associated truth tables, providing a common design language for circuits built from living components. This abstraction is powerful because it allows biologists, chemists and engineers to describe, compare and compose biological information processing in the same way electrical engineers describe integrated circuits, even when the underlying hardware is a strand of RNA, a protease or a hydrogel crosslink.

Within living cells, the most mature layer of the biocomputing stack is transcriptional regulation. The field’s foundations were laid more than two decades ago with landmark constructions in Escherichia coli, including a synthetic genetic toggle switch and a repressilator, an oscillatory network of transcriptional regulators, both reported in Nature in 2000. Since then, combinatorial synthesis of genetic networks, modular AND gates that integrate environmental signals, and chimeric transcription factors built from orthogonal LacI/GalR family repressors have expanded the designer’s toolbox. In mammalian cells, large-scale design efforts have produced robust circuits with multiple inputs and outputs, and synthetic multistability has been demonstrated, allowing cells to occupy multiple stable gene-expression states. More recently, recombinase-based systems have pushed cells toward true state-machine behavior, recording event histories in DNA, and preprint work has even simulated for-loop-like counting programs in the human genome, counting to three through cascaded recombination events.

Nucleic acids themselves have emerged as an extraordinarily versatile computational medium. Engineered riboregulators enabled post-transcriptional control of gene expression as early as 2004, and higher-order RNA devices soon followed, with synthetic riboswitches processing multiple inputs inside cells. Tandem riboswitch architectures found in nature exhibit complex gene-control functions that synthetic designs have begun to emulate. The ribocomputing era arrived when complex cellular logic was demonstrated with RNA devices that integrate several sensors in a single transcript, and CRISPR-based tools such as Cas13 have added programmable RNA sensing and editing to the repertoire. Machine learning is now accelerating this design cycle dramatically: deep learning has been used to generate novel promoters in E. coli, cell-type-targeting cis-regulatory elements, ribosome binding sites, optimized untranslated regions for mRNA-delivered gene editing, and even de novo RNA aptamers predicted from structure. A 2026 study introduced Toehold-VISTA, a machine learning approach for deciphering programmable RNA sensor-target interactions, while automated platforms such as Cello 2.0 bring genuine design automation to genetic circuit engineering.

Proteins offer a faster, more compact computational layer than gene expression, and the review devotes detailed attention to protein-based biocomputability within mammalian cells. Programmable protein circuits built from modular interaction domains can process signals in minutes rather than hours, and designed protease-based signaling networks use enzymes such as tobacco etch virus protease as reversible logic elements that cleave specific peptide substrates. De novo designed protein logic gates, reported in Science in 2020, showed that computational behavior can be engineered directly into protein structures, and recent work from the DeForest group demonstrated Boolean logic-gated protein presentation through autonomously compiled molecular topology. Directed evolution platforms such as YESS and its successors have improved protease catalytic efficiency and specificity, while optogenetic tools including photocleavable proteins like PhoCl, light-inducible dimerizers, and single-component light-assisted uncaging switches give researchers exquisite spatial and temporal control. Opto-PROTACs extend this logic to targeted protein destruction, and engineered condensates provide programmable compartments where these reactions can be organized.

The review then pivots from living cells to engineered materials, a domain where the same logical principles are being embedded into synthetic and biological matter. Synthetic hydrogels have been programmed as molecular logic gates: in a 2018 Nature Chemistry study, modular biomaterial logic gates responded to combinations of environmental triggers to release therapeutics only under the correct conditions, and subsequent work extended logical stimuli-triggered delivery to small molecules and site-specifically modified proteins. Programmable polymer libraries have enabled stimuli-responsive nanocarriers containing logic gates, and AND-gated prodrug micelles have been shown to locally stimulate antitumor immunity, while Boolean logic prodrug nanoparticles support combination cancer immunotherapy. Piezoelectric enzyme systems such as MnTiO3 have been used for programmed catalysis of tumor cell death, and pH-responsive doped carbon nanodots have demonstrated multilevel, dual-readout logic operations, showing that biocomputation extends well beyond organic polymers.

DNA nanotechnology represents perhaps the most elegant materials-based computing platform. Since Nadrian Seeman’s foundational vision of nucleic acid junctions and lattices in 1982, and the synthesis of a DNA cube in 1991, researchers have folded DNA into arbitrary nanoscale shapes through DNA origami and engineered logic directly into DNA crystals. Logic-gated DNA nanorobots have been shown to transport molecular payloads to target cells, aptamer-encoded extracellular vesicles perform multi-parameter logic for colorectal cancer diagnosis, and wavelength-dependent orthogonal photoregulation enables complex DNA liberation operations. DNA circuits loaded into biomimetic ZIF-8 nanoparticles, equipped with biological amplifiers, have accurately identified specific cancers in vivo, and site-occupying interactions between DNA and metal-organic frameworks have yielded molecular sensing and logic systems. These systems translate molecular recognition directly into mechanical or chemical action, effectively making the material itself the computer.

Protein-based materials complete the materials triad. Inteins, the self-splicing protein elements, have been used to trigger artificial protein hydrogels and to assemble covalently stabilized single-chain protein nanostructures with ultra-high stability. Photocleavable proteins such as PhoCl underpin dynamic hydrogels and injectable recombinant materials like PhoCoil, which degrade on demand for minimally invasive delivery. Light-responsive ligand tethers regulate stem cell mechanics, cyclic stiffness modulation responds to user-specified stimuli, and LOVTRAP-based systems create light-responsive dynamic protein networks. Sortase-mediated functionalization, SpyTag/SpyCatcher chemistry and the broader catcher/tag toolbox allow modular, covalent assembly of protein materials, and recent work has produced biohybrid material circuits with binary decoder functionality as well as Boolean logic-based controlled release of bioactive proteins with diversified inputs. Supramolecular hydrogel-enzyme hybrids have even been taught to respond to a variety of biological substances through installed logic-gate behavior.

The most forward-looking section of the review concerns engineered living materials, where computation and matter genuinely merge. Living materials grown from engineered microbial co-cultures exhibit programmable functionalities, pattern-forming bacteria have been used to assemble pressure sensors, and de novo bacterial lifestyle programs encode entire behavioral states. Bacteria are increasingly deployed as living biosensors, including for DNA detection, and shelf-stable living materials permit continuous biosensing of oligosaccharides. Bacteriogenic protocells assembled from living material, bio-hybrid robots, and frameworks for building synthetic cells together point toward a future in which the boundary between computing substrate and living tissue dissolves. The authors also invoke the concept of collective intelligence as a unifying idea spanning scales and substrates, suggesting that the computational principles being engineered into cells and gels may illuminate how biological intelligence itself operates.

What emerges from this synthesis is a discipline in adolescence: the individual components, genetic circuits, RNA devices, protease logic, DNA nanostructures, responsive hydrogels, are mature, but their integration into robust, multiplexed, deployable systems remains the central challenge. The convergence of synthetic biology and biomaterials engineering promises applications across advanced chemical processing, additive manufacturing, tissue engineering, organ-on-chip disease modeling, and living therapeutics that sense disease markers and respond only where and when appropriate. As machine learning continues to compress design cycles and protein design tools such as ProteinMPNN improve expression, stability and function of engineered components, the vision articulated by Gharios, Ross, Li, Kottantharayil and DeForest comes into focus: a future in which computation is not confined to silicon but is woven into the living and material fabric of the world around us, executing logic wherever biology already thrives.

Subject of Research: Engineering Boolean logic-based biocomputation in living cells and stimuli-responsive biomaterials

Article Title: Engineering biocomputability in living and material systems

Article References: Engineering biocomputability in living and material systems. (n.d.). https://doi.org/10.1038/s44286-026-00433-9

Image Credits: AI Generated

DOI: 10.1038/s44286-026-00433-9

Keywords: biocomputation, synthetic biology, biomaterials, Boolean logic gates, hydrogels, DNA nanotechnology, protein circuits, genetic circuits, optogenetics, stimuli-responsive materials, engineered living materials, riboswitches