RADAR Platform

Our project builds on RADAR (RNA sensing via Adenosine Deaminases Acting on RNA), a technology developed in the Xiaojing Gao Lab at Stanford. RADAR translates the presence of a target RNA transcript into a protein output by exploiting the RNA editing activity of ADAR ("adenosine deaminase acting on RNA") enzymes. When a "sensor" RNA is designed to pair with a specific "target" RNA, ADAR modifies a strategically placed adenosine in the sensor strand. This edit converts a premature stop codon into a start codon, leading to translation of a downstream protein. In this way, the detection of RNA inside a cell is directly coupled to the production of a measurable or therapeutic or diagnostic output.

Programmable Logic System

We extended this platform into a programmable logic system. Instead of relying on a single input, we designed two independent RADAR sensors, each tuned to a distinct RNA transcript. One sensor detects a conserved HBV integration-derived transcript, while the other detects mRNA elevated in hepatocellular carcinoma: GPC3, AKR1B10, or RP11.

Safety & AND-Gate Design

To ensure safety, neither input alone is sufficient to activate therapy. Each sensor was engineered to control the expression of only half of a split protein. Only when both RNAs are present in the same cell are both protein halves expressed. These halves then self-assemble into a functional protein, enforcing an intracellular "AND-gate" that ensures output is triggered only when two independent disease signals coincide.

Proof of Concept

As a proof of concept, we validated this mechanism using a split GFP reporter. Each RADAR sensor was linked to one of two fragments of GFP. When only one input RNA was present, cells expressed an incomplete protein. When both inputs are present, GFP assembles, producing fluorescence and providing a quantifiable readout of gate activation. This system not only demonstrates the feasibility of our RNA-based logic gating in living cells, but also provides a platform to optimize design parameters such as sensor efficiency, and background editing.

Modular Design

Beyond GFP, our design is modular. Any protein that can be split computationally can serve as an output. Potential outputs for targeting cancer cells include caspases for controlled apoptosis, therapeutic antibodies, or immune-modulating proteins for eliciting antitumor immunity.

Computational Pipeline

To identify the most reliable post-integration HBV markers, we developed a computational pipeline that systematically screened RNA-seq datasets from HBV-induced HCC cell lines. By comparing transcripts produced only after viral integration in patient samples, we identified conserved and highly expressed regions suitable for sensor design. This ensured that the HBV-derived signal was both specific to diseased hepatocytes and generalizable across patients.

Taken together, our implementation transforms RADAR from a single-signal detection system into a logic-gated therapeutic platform. By requiring simultaneous detection of two independent inputs, we minimize off-target activation and achieve a level of precision unattainable with conventional single-biomarker approaches.