A synthetic protein-level neural network in mammalian cells

• DOI: 10.1126/science.add8468
• Publication Date: 2024-12-12T19:00:41Z
• Authors: Zibo Chen; James M. Linton; Shiyu Xia; Xinwen Fan; Dingchen Yu; Jinglin Wang; Ronghui Zhu; Michael B. Elowitz
• Keywords: 0303 health sciences; 03 medical and health sciences; Article

Artificial neural networks provide a powerful paradigm for nonbiological information processing. To understand whether similar principles could enable computation within living cells, we combined de novo-designed protein heterodimers and engineered viral proteases to implement synthetic circuit that performs winner-take-all network classification. This "perceptein" combines weighted input summation through reversible binding interactions with self-activation mutual inhibition irreversible proteolytic cleavage. These collectively generate large repertoire of distinct species stemming from up eight coexpressed starting species. The complete system achieves multi-output signal classification tunable decision boundaries in mammalian cells can be used conditionally control cell death. results demonstrate how protein-based programmable cells.