Engineering a scalable and orthogonal platform for synthetic communication in mammalian cells

Engineering synthetic communication between mammalian cells enables distributed information processing, reduced resource burden in single cells, and improved specificity in diagnostics and therapeutics. Existing approaches using cell-to-cell contact, small molecules, or evolved natural proteins face limitations in scalability, orthogonality, and receptor-level logic. This study asks whether a fully orthogonal, scalable platform using diffusible ligands can activate synthetic receptors and compute Boolean logic directly at the receptor level. The authors propose functionalizing GEMS synthetic receptors with orthogonal coiled-coil (CC) pairs to achieve programmable, selective receptor dimerization, enabling intercellular communication and logic while maintaining modularity across intracellular signaling pathways.

Prior mammalian synthetic communication has used: (1) juxtracrine systems with synthetic receptors (e.g., SynNotch) performing logic via cell contact; (2) small-molecule-based circuits enabling complex biocomputation but limited in scalability and lacking receptor-level logic; (3) directed evolution of natural proteins (e.g., orthogonal IL-2/IL-2Rβ) providing selective signaling but tied to specific cytokine pathways and labor-intensive to generate orthogonal pairs. There is no current platform for diffusible-ligand-based communication that is inherently scalable, orthogonal to native signaling, and capable of Boolean logic at the receptor level. Coiled-coils offer a designed, orthogonal, and tunable binding toolbox previously used intracellularly for logic and assembly, suggesting potential for receptor-level control and intercellular signaling without re-engineering for each ligand-receptor pair.

- Receptor engineering: GEMS receptors based on an erythropoietin receptor (EpoR) transmembrane scaffold inert to erythropoietin were functionalized extracellularly with designed CC peptides. Intracellular domains from IL-6RB (JAK/STAT pathway) or VEGFR2 (PLCG pathway) were used, enabling pathway rerouting to SEAP reporters via minimal promoters. - CC pairs and linkers: Orthogonal CC heterodimers (A:A′, B:B′, Γ:Γ′/I:I′ nomenclature in text) were fused to EpoR via glycine-serine linkers of varying lengths (0 aa, (GS)2, (GS)4, (GSS)9). Orthogonality and linker effects were tested by co-expressing cognate and non-cognate receptor pairs. - Ligand design and production: Two classes of soluble ditopic ligands were created: (i) synthetic A′-A′ dipeptides via N-termini linkage using a homo-bifunctional bismaleimide crosslinker on cysteine-functionalized CC monomers (SUMO-fusion expression in E. coli, SUMO cleavage, anion-exchange purification); (ii) genetically encoded ditopic ligands linking CCs by protein linkers (combinations of rigid (EAAAK)n and flexible (SGSSGS)n modules; l1–l3) expressed in E. coli with/without SUMO tags and purified. Variants with helix-interrupting PG linkers were also tested. - Cell systems: HEK293T used for transient expression of CC-GEMS receptors, STAT3 (when JAK/STAT used), and SEAP reporters. For PLCG signaling, VEGFR2 intracellular domain and an NFAT-responsive SEAP reporter were used. - Assays: SEAP activity quantified colorimetrically (pNPP to PNP) and expressed in U/L; dose-response titrations of ligands; confocal microscopy with Cy3/Cy5/AF488-labeled ligands to confirm receptor-ligand colocalization and receptor membrane expression; SDS-PAGE and Native-PAGE to confirm ligand dimerization and complex assembly; SEC for AND-complex verification. - Intercellular communication: Sender HEK293S GnTI TetR cells engineered via lentiviral transduction to secrete SUMO-tagged ditopic ligands (A′-A′ for two-cell system; A′-Γ and Γ′-A′ for three-cell AND system) under doxycycline-inducible CMV-Tet operator control; secretion quantified via western blot calibration against bacterial standards. Receiver HEK293T cells transiently expressed cognate receptors and SEAP reporters; co-cultures tested with and without doxycycline. - Therapeutic output: A-typeJAK/STAT receptor cells expressing STAT3 and IL-10 under a STAT3-responsive promoter; IL-10 quantified by ELISA after treatment with cognate vs non-cognate ligands. - Statistics: n=3 independent samples; unpaired two-sided t-tests or one-way ANOVA with Tukey/Šidák/Dunnett as appropriate; P<0.05 considered significant.

- CC-driven receptor activation and orthogonality: Co-expression of cognate CC-GEMS receptor pairs (e.g., A with A′) robustly activated SEAP via JAK/STAT; non-cognate pairs showed negligible activation, confirming orthogonality. Linker lengths from 0 to 27 aa did not critically affect activation for cognate pairs. - Soluble ligand activation and linker dependence: Synthetic A′-A′ dipeptides activated A-typeJAK/STAT receptors; monomeric A′ did not. Ligand-induced activation vs monomer control showed fold increases by receptor linker: α0: 31.8×, α1: 103.6×, α2: 47.0×, α3: 15.7×. Dose-response exhibited a bell-shaped curve with a broad plateau between ~0.03–0.5 µM ligand. - Genetically encoded ditopic ligands: Recombinant A′-A′ ligands with protein linkers activated receptors; relative to monomeric A′: l1 (no SUMO): 24.7×, l1 (SUMO): 20.7×; l2 (SUMO): 16.6× (similar to synthetic A′-A′ reference: 15.5×); l3 (SUMO): 8.4×, l3 (no SUMO): 7.5×. SUMO tags did not prevent activation. Confocal imaging confirmed ligand-receptor colocalization for cognate pairs. - Pathway versatility: A-typePLCG receptors (VEGFR2 intracellular domain) showed 7.6× SEAP increase upon 0.12 µM A′-A′ ligand, demonstrating rerouting beyond JAK/STAT. - Scalability and orthogonality across CC sets: Only cognate ligand-receptor pairs activated; relative activation magnitudes correlated with known CC affinities: A′-A′→A-type ~30×, I′-I′→I-type ~16×, B′-B′→B-type ~5×. Adding mixtures of non-cognate ligands did not inhibit cognate activation. - Boolean logic at receptor level: - OR gate: Cells co-expressing A-, B-, and Γ-type receptors activated with B′-A′ or A′-Γ′ ligands individually; co-application (0.5 µM total) yielded ~10.8× increase vs baseline, while either ligand alone at 0.5 µM yielded ~3×. Heterogeneous receptor co-expression reduced individual receptor availability and activation. - AND gate: Inter-ligand dimerization using A′-Γ and Γ′-A′ ligands formed A′-Γ:Γ′-A′ complexes (Native-PAGE, SEC). Only the presence of both ligands produced receptor activation: ~30–45× vs either alone or no ligand. - Intercellular communication: - Two-cell system: Co-culture of sender cells secreting A′-A′ (doxycycline-inducible) with A-typeJAK/STAT receivers yielded a 6.6× SEAP increase (dox vs no dox) and 9.7× vs control senders. Secreted ligand measured ~0.14 µM at 48 h. Activation was lower than exogenous bacterial ligand addition, attributed to gradual secretion kinetics. - Three-cell AND: Receivers plus two senders secreting A′-Γ and Γ′-A′ achieved 7–9.6× increases only when both senders were present. Estimated complex concentration ~0.06 µM (two SUMO tags per complex and 0.12 µM SUMO signal in medium). - Therapeutic output control: IL-10 secretion under STAT3 control increased 15.5× with cognate A′-A′ ligand vs non-cognate and 17× vs no ligand, demonstrating therapeutic actuation via CC-GEMS.

The study addresses the need for an orthogonal, scalable, diffusible-ligand-based communication platform in mammalian cells by leveraging designed coiled-coil heterodimers to drive GEMS receptor dimerization and signaling. Orthogonality and modularity allow multiple independent channels of communication and combinatorial logic directly at the receptor, avoiding added transcriptional layers that increase burden and latency. The platform functions across distinct intracellular pathways (JAK/STAT, PLCG), and synthetic ligands can be produced in bacteria or secreted by mammalian senders to support intercellular communication. Boolean OR and AND operations achieved at the receptor level validate the capacity for distributed computing across cell populations. The ability to trigger therapeutic protein secretion (IL-10) underscores the translational potential for designing responsive therapeutic consortia. Observed differences in activation due to linker lengths, receptor density, and ligand presentation highlight design parameters for optimization. Overall, CC-GEMS provides a blueprint for constructing complex, multi-channel cell networks with precise, orthogonal control over communication and actuation.

This work establishes CC-GEMS, a modular mammalian synthetic communication platform in which orthogonal coiled-coil ligands dimerize engineered receptors to activate programmable outputs. The system: (i) supports multiple, orthogonal CC pairs with predictable selectivity; (ii) operates through distinct intracellular pathways; (iii) performs receptor-level Boolean OR and AND logic; (iv) enables sender-receiver intercellular signaling with inducible ligand secretion; and (v) controls therapeutic protein release. Future directions include expanding the CC toolbox with varied affinities and multi-input designs (e.g., 4-helix bundles), optimizing linker architectures and receptor surface density, extending to alternative receptor chassis (MESA, DocTAR) to minimize native pathway cross-talk, and deploying in primary immune cells and in vivo models to realize programmable therapeutic consortia and quorum-sensing-like behaviors.

- Linker effects: Although not critical within tested lengths for cognate co-expression, ligand-induced activation varied with linker length; broader exploration may refine performance and geometry. - Activation magnitude differences: Ligand-induced activation was lower than co-expressed cognate receptor pairing, likely due to receptor availability on the membrane and inability to dimerize pre-trafficking; also bell-shaped dose-responses constrain dynamic range. - Secretion kinetics and concentration: Sender-derived ligand led to lower activation than bolus addition despite similar nominal concentrations, indicating that gradual secretion and medium dynamics impact effective receptor engagement. - Receptor expression heterogeneity and density: Co-expression of multiple receptors reduces availability of each, limiting logic gate activation; optimizing expression balance and membrane density is needed. - Temperature considerations: Some experiments at 30 °C (due to B:B′ and Γ:Γ′ melting temperatures around 40 °C) may not fully reflect physiological conditions. - Cell model scope: Demonstrations were in HEK293 derivatives; generalization to primary cells and in vivo settings remains to be validated.