The ability to precisely identify and manipulate specific cells is a cornerstone of modern medicine and biology. From cancer immunotherapy to regenerative medicine, targeting unique molecular signatures on a cell's surface holds immense therapeutic promise. However, a persistent challenge has been the lack of true specificity. Most strategies rely on targeting a single surface antigen, a method that often fails in complex biological systems where healthy and diseased cells share markers, leading to dangerous off-target effects. The field has long sought a "smart" molecular system capable of integrating multiple inputs to make a logical decision before acting—a tool that can distinguish not just a single feature, but a unique cellular identity defined by a combination of markers.

The Path to Precision: A Brief History of Protein Engineering

The journey toward such a system is built on decades of advances in protein engineering. A pivotal discovery was protein splicing, a natural process where an intervening protein sequence, or "intein," excises itself and ligates the flanking sequences, known as exteins [2]. Scientists quickly realized the potential of this molecular cut-and-paste tool. By splitting an intein into two inactive halves, they created "split inteins," which could reassemble and function only when brought into close proximity. This laid the groundwork for conditional protein ligation, or protein trans-splicing (PTS), offering a way to make protein interactions dependent on a specific event. However, early applications were often limited by slow kinetics, a lack of robustness in extracellular environments, and the absence of a generalized framework for performing complex logical operations on the surface of living cells. The challenge remained: how to build a modular, reliable, and programmable decision-making device from these components.

The SMART Breakthrough: On-Demand Protein Synthesis on Cell Surfaces

A landmark paper from Tom W. Muir's lab at Princeton University, published in Nature, introduces a powerful solution: the SMART (Splicing-Modulated Actuation upon Recognition of Targets) platform [1]. This work elegantly overcomes the limitations of previous technologies by creating an autonomous system that performs programmable protein synthesis directly on the surface of target cells.

The Core Innovation: Proximity-Gated Logic

The SMART system re-imagines protein engineering as a form of cellular computation. The core of the system consists of two separate, inactive protein fragments. Each fragment is fused to a distinct targeting moiety—such as an antibody fragment or a nanobody—that recognizes a specific cell-surface antigen. These two components are inert on their own and circulate harmlessly.

The magic happens only when a cell expresses both target antigens simultaneously. The two SMART components bind to their respective targets on the same cell surface, bringing them into close proximity. This proximity triggers the engineered split-intein domains to recognize each other, refold, and execute protein trans-splicing. In this process, the intein fragments excise themselves and covalently ligate the two previously separate protein fragments, creating a single, fully active protein in situ. This functions as a highly specific "AND" logic gate: the functional output is generated only if Input A AND Input B are present.

Demonstrating Unprecedented Specificity and Modularity

The researchers rigorously validated the SMART platform's capabilities. In one key experiment, they designed a system to assemble the protein "SpyCatcher" on cells expressing both HER2 and EGFR, two common cancer markers. The results were striking: functional SpyCatcher was generated with high efficiency exclusively on the dual-positive (HER2+/EGFR+) cells, with virtually no activity detected on single-positive or negative cells. This demonstrated the system's exquisite selectivity, a critical feature for minimizing off-target effects in therapy.

Crucially, the SMART platform is highly modular. The authors showed its compatibility with various targeting domains (antibody fragments, DARPins) and its ability to generate different functional outputs. Beyond simply assembling a protein, they engineered a SMART system to produce and release the cytokine IL-1β. This "release-type" configuration allows the system to locally activate immune signaling pathways only in the microenvironment of the target cells, opening new avenues for precision immunotherapy.

The Broader Impact: From Targeting to Programming Biology

The SMART system represents a paradigm shift from simple molecular targeting to programmable biological actuation. Instead of just delivering a pre-made drug to a cell, this technology enables the on-site manufacturing of a functional protein, contingent on a logical assessment of the cell's identity. This has profound implications across multiple fields.

1. Safer and More Effective Therapeutics: By requiring multiple antigens for activation, SMART-based therapies could dramatically improve the safety of potent treatments like cytokine therapies or antibody-drug conjugates, directing their effects only to diseased cells and sparing healthy tissue.
2. Advanced Synthetic Biology: The platform provides a robust tool for building complex synthetic circuits that mediate cell-cell communication. One can imagine programming cells to sense their neighbors and respond by producing specific signaling molecules, enabling the engineering of synthetic tissues and complex biological systems.
3. Next-Generation Research Tools: SMART can be adapted to deliver proximity-labeling enzymes or fluorescent reporters, allowing researchers to map the proteomes or visualize specific cell subpopulations that are defined by complex surface codes, rather than a single marker.

Realizing the full potential of this modular platform will involve exploring a vast design space of targeting domains, payloads, and linker configurations to optimize for different applications. This necessitates a move away from traditional, low-throughput cloning. Exploring this vast design space will require high-throughput methods to build and test countless construct variants. Platforms that streamline this design-build-test-learn cycle, such as AI-native DNA Coding and self-selecting vector libraries, will be instrumental in accelerating the development of next-generation SMART-based therapeutics.

While challenges such as in vivo delivery and potential immunogenicity of the protein components must be addressed, the SMART platform provides a powerful and versatile framework for engineering cellular interactions with unprecedented precision. It moves the field one step closer to the ultimate goal of programming biology as reliably as we program computers.

References

1. Kofoed, C., Erkalo, G., Tay, N. E. S., Ye, X., Lin, Y., & Muir, T. W. (2025). Programmable protein splicing on the surface of living cells. Nature. https://www.nature.com/articles/s41586-025-09287-2
2. Topilina, N. I., & Mills, K. V. (2023). An overview of the intein landscape. Mobile DNA, 14(1), 22. https://doi.org/10.1186/s13100-023-00306-0