Reconstituting Cellular Intelligence in Nonliving Biomolecular Matter

Living cells convert environmental and internal information into physical and biochemical action. This dissertation focuses on two layers of cell-like control that can be rebuilt in nonliving biomolecular systems: active physical machinery, where cytoskeletal architecture turns molecular force generation into mechanics and transport, and molecular sense-and-response, where compartment architecture turns environmental cues into biochemical output. The common problem is control. Active cytoskeletal materials remodel while generating force, so force transmission must remain predictable in a scaffold that also acts as a machine. Synthetic compartments face a related boundary problem: a compartment must preserve internal state while admitting inputs, sustaining production, and releasing functional output. Study 1 asks how nanoscale motor forces become network-scale mechanical work. We use PRC1-mediated microtubule crosslinking to tune architecture in kinesin-driven active materials and show that connectivity alone does not guarantee mechanical output. Low-connectivity networks dissipate motor activity locally. Connected but mechanically floppy networks reorganize collectively but fail to sustain long-range force propagation or perform substantial work. Mechanically persistent networks support coherent contraction and extractable work. Velocity correlations reveal a transition from local motion to collective organization, while work measurements and a semiflexible-network model separate connectivity from rigidity. The result is a force-to-work rule for active matter: motors generate force, but crosslinkers determine whether force dissipates locally or becomes coherent, extractable mechanical work. Study 2 asks how spatial patterning can make active matter predictable enough for task execution despite nonlinear dynamics. Motor--microtubule active fluids generate autonomous motion, but useful engineering requires flow fields specifying where material moves, which objects separate, and how fluids mix. We use light to pattern contracting motor--filament networks and use predictive modeling to compose hydrodynamic outputs into micrometre-scale flow fields. Programmed flows move and separate cell clusters, impose extensional deformation on polymers and giant vesicles, and mix fluids at low Reynolds number. The result is a force-to-flow rule: optical geometry converts internally generated active forces into predictable transport functions. Study 3 builds an acellular sense-and-response system. A giant unilamellar vesicle, or GUV, supplies a cell-sized lipid boundary for preserving internal biochemical state. Cell-free transcription--translation, or TX-TL, supplies a nonliving biochemical engine for reading DNA and synthesizing proteins outside a cell. We engineer input-output separation across the compartment boundary: small-molecule access admits lactate and nutrients, lactate sensing switches internal productive state, TX-TL synthesizes a bispecific T-cell engager and a genetically encoded membrane gate, and controlled release exports macromolecular output. Lactate-conditioned GUVs recruit T cells to kill CD19-positive target cells. Study 3 brings selective input, productive internal state, controlled output, and target-cell killing into one nonliving compartment. Across the three studies, we rebuild cellular functions outside living cells. Molecular architecture controls force-to-work conversion, optical patterning controls force-to-flow conversion, and compartment architecture controls cue-to-output conversion. In each case, nonliving biomolecular matter uses molecular, spatial, or compartmental structure to link input with output.