Synthetic biology has long promised programmable cells, living therapeutic devices that sense disease biomarkers, compute a response, and deliver a treatment. The vision is compelling, but it has run into a fundamental engineering barrier: complex logic requires many parts, each additional part consumes cellular resources, and the total genetic payload that can be delivered to a human cell is sharply limited by viral vector capacity.
A team at The Hebrew University of Jerusalem, led by corresponding author Lior Nissim, has now demonstrated a new approach that changes the scaling curve. In a paper published June 16 in Nature Communications, the researchers show that by moving logic computation from the transcriptional level to the RNA trans-splicing level, they can implement a full binary adder, a circuit that adds three input bits and produces a sum and carry output, using just three modules in a single computational layer.
The Parts Problem
Current synthetic circuits in mammalian cells work mostly at the transcriptional level: transcription factors activate or repress promoters in cascaded layers. A simple AND gate requires two input signals to converge on a single output promoter. A full adder, three inputs in, two outputs out, requires multiple layers of cascaded regulation, each with its own set of orthogonal transcription factors, promoters, and binding sites. As complexity grows, the number of parts grows faster than the number of functions, and the cell’s transcription and translation machinery becomes increasingly burdened.
The Nissim lab’s solution repurposes RNA trans-splicing, a natural process in which two separate pre-mRNA transcripts are spliced together to form a single functional mRNA. In the team’s design, a gene is split into two exons, each placed under a different promoter on a separate plasmid. The exons carry complementary binding domains that cause them to associate and splice together, but only when both promoters are active. The result is a molecular AND gate at the RNA level, requiring no additional transcription factors.
From AND Gates to Full Adders
The team built four increasingly complex circuits to demonstrate the framework:
A three-input combinatorial gate implementing the Boolean function (I1 AND I2) AND (NOT I3) achieved a 14- to 144-fold difference between the expected ON state and all seven other input combinations.
A half-adder, two inputs computing SUM (XOR) and CARRY (AND), correctly processed all four binary combinations (00, 01, 10, 11) using a pair of mutually repressing synthetic microRNAs to implement the XOR function.
A full adder, three inputs computing SUM and CARRY across all eight binary states, was implemented in a single computational layer using just 10 plasmids across three modules. The circuit correctly computed every state: single input activates SUM only; two inputs activate CARRY only; three inputs activates both.
A dynamic 3-to-1 multiplexer with a Selector Overload Status (SOS) safety output, claimed as the first implementation of its kind in mammalian cells, routes one of three input signals to a single output based on two selector bits, while the SOS output fires only when both selectors are simultaneously active, functioning as an overload detection safety switch.
The Core Innovation
The enabling technology is a set of orthogonal binding domains, short RNA sequences (30,50 base pairs) that bind specifically to their complementary partner but not to any other sequence in the cell. The team validated six such pairs, achieving off-target chimera rates below 0.013% by RNA sequencing. Where previous designs required increasingly complex transcription factor cascades for each new input, the RNA trans-splicing approach adds complexity at the RNA level, where binding domains can be combined combinatorially without requiring new protein parts.
The output of each logic module is a synthetic transcription factor (GAD, a GAL4-VP16 fusion), which then drives any desired downstream output through GAL4-responsive promoters. The team demonstrated this flexibility by placing the immunotherapeutic cytokine IL-15 under circuit control, ensuring it is secreted only in the precise input state that triggers the circuit.
What This Means for Cell Therapy
The implications go beyond elegant circuit design. Therapeutic cells, CAR-T cells, engineered stem cells, microbial delivery vehicles, have limited genetic cargo capacity. Adeno-associated virus vectors, the most common delivery system for gene therapy, carry roughly 4.7 to 8 kilobases. Each additional transcriptional layer consumes precious space.
An RNA-level circuit that can perform full binary addition in a single layer using a fraction of the parts represents a real engineering advantage. The team’s circuits are designed as modular, swappable components: input sensors, RNA binding domains, splice signals, and output effectors are all independently interchangeable. The design rules for creating new orthogonal binding domains are explicitly provided.
The work also opens a path toward therapeutic cells that can integrate multiple disease-relevant biomarkers, detecting, for example, the simultaneous presence of a tumour antigen, a hypoxia signal, and an immune checkpoint molecule, before triggering a therapeutic response. The SOS multiplexer adds a safety layer: the cell can detect when too many inputs are active and trigger a kill switch.
The authors have filed patent applications (US and EU) for the technology, and Nissim is a co-founder of MeatoLogic, which has licensed the patents. Thirty-eight plasmids have been deposited with Addgene for distribution to the research community.
Sources:
1. Roas, K., Kovalski, I., Mouhadeb, O. et al. “Modular Scalable Synthetic Gene Circuits for Complex Functions Within Minimal Computational Layers in Human Cells.” Nature Communications, June 16, 2026. DOI: 10.1038/s41467-026-74408-y
2. Addgene plasmid repository. 38 plasmids from the Nissim lab deposited for distribution. https://www.addgene.org/