Programming multicellular assembly with synthetic cell adhesion molecules

The study addresses whether cell-cell adhesion can be systematically programmed in metazoan cells to control multicellular organization, with potential applications in development, neurobiology, immunology, tissue repair, and therapeutic cell design. Native cell-cell interactions are mediated by complex cell adhesion molecules (CAMs) that combine extracellular binding with intracellular cytoskeletal coupling. It is unclear to what extent these functions can be uncoupled and recombined to create new cell-cell connectivities. Building on indications of modularity from previous work, the authors propose to fuse orthogonal extracellular binding domains (ECDs) with intracellular domains (ICDs) from endogenous CAMs to create synthetic CAMs (synCAMs), then test their ability to form interfaces and program multicellular assemblies.

Background highlights include: CAMs such as integrins and cadherins mediate adhesion and cytoskeletal remodeling; prior chimeric receptor studies suggest modularity between extracellular recognition and intracellular signaling; adhesion strength and cell sorting relate to cortical tension and cytoskeletal coupling; recent synthetic biology efforts in bacteria and mammalian systems used surface tethering and orthogonal nanobody-antigen pairs to control multicellular assembly. The authors aim to extend these by integrating orthogonal ECDs with native CAM ICDs to generate tunable, orthogonal, and modular adhesion in mammalian cells.

• Constructed heterophilic synCAMs by fusing orthogonal binding ECDs (e.g., GFP/anti-GFP nanobody pair) to transmembrane (TM) and ICDs from multiple CAMs: E-cadherin (ECAD), integrin β1 (ITGB1), integrin β2 (ITGB2), ICAM-1, DLL1, JAM-B, NCAM-1, and MUC-4. Expression matched to controls.
• Compared synCAM-mediated interfaces in L929 fibroblasts using confocal microscopy after mixing cells expressing cognate synCAM pairs in ultra-low-attachment plates; controls included native WT ECAD and a TM-tether lacking ICD.
• Quantified interface geometry via contact angle (apparent surface tension metric correlated with interface size) and measured receptor enrichment fraction (fraction of GFP-tagged synCAM localized to interface vs total membrane).
• Assessed contributions of ECD affinity vs ICD identity by creating an affinity series of GFP nanobodies (KDs spanning ~0.7 nM to micromolar) and by deleting ICDs; performed similar tests with ITGB1 ICD.
• Validated adhesion preferences using competition sorting assays where different anti-GFP synCAM variants (ICAM-1 ICD) compete to co-sort with GFP synCAM bait cells.
• Surface-spreading assay: plated anti-GFP synCAM-expressing L929 cells on immobile GFP-coated surfaces; after 75 min–2 h, fixed and stained with Phalloidin to visualize actin; characterized spreading morphology (expansive vs protrusive) and kinetics; quantified cell footprint and area.
• Perturbed cytoskeletal regulators with small molecules (latrunculin B, blebbistatin, CK666) to assess roles of actin polymerization, contractility, and Arp2/3-mediated lamellipodia.
• Performed mutational analysis of ICD interaction motifs to confirm roles of adaptor recruitment (e.g., β-catenin, talin, vinculin, ERM; PDZ scaffolds and lipid raft interactions).
• Built a panel of orthogonal heterotypic synCAMs with multiple antibody–antigen ECD pairs (HA/anti-HA scFv, MBP/anti-MBP nanobody, CD19/anti-CD19 scFv, MET/anti-MET nanobody, mCherry/anti-mCherry nanobody, EGFR/anti-EGFR nanobody); validated orthogonality via co-sorting assays; quantified exclusion of WT cells.
• Programmed defined multicellular assemblies (alternating A+B, bridging A–B–C, cyclic A→B→C→A) by expressing appropriate orthogonal synCAMs; analyzed nearest-neighbour distributions with Harmony software; time-lapse imaging captured assembly dynamics.
• Engineered homotypic synCAMs using antiparallel leucine zipper coiled-coils (Aph4, IF1) as ECDs with an intervening fibcon FN3 spacer to favor trans over cis binding; expressed with ICAM-1 TM/ICD; assessed differential sorting and emergent structures in mixed populations, including three-way mixes with WT ECAD.
• Intercalation with native adhesions: created anti-P-cadherin (PCAD) scFv synCAM fused to ICAM-1 TM/ICD and tested integration into PCAD-mediated spheroids.
• Tested synCAM functionality in primary human dermal fibroblasts, human mesenchymal stromal cells, and iPS-derived smooth muscle cells using GFP/anti-GFP-ICAM-1 synCAMs vs tether controls.
• Tissue remodelling assays: mixed L929 cells expressing WT ECAD vs WT PCAD, introduced heterophilic synCAMs of varying strength (tether, ECAD ICD, ICAM-1 ICD) to force integration; similarly bridged PCAD-L929 spheroids with MDCK epithelial monolayers using GFP/anti-GFP synCAMs (tether, ECAD ICD, ICAM-1 ICD) and imaged emergent lattice-like networks; analogous experiments in primary tissues.

• synCAMs recapitulate native-like adhesion: Several ICDs (ECAD, ITGB1, ITGB2, ICAM-1, MUC-4) produced extensive interfaces comparable to native cadherin; tethers lacking ICDs formed only minimal point contacts.
• Two phenotypic classes emerged: expansive (large interfaces, higher contact angles; ICDs ECAD, ITGB1, ITGB2, ICAM-1, MUC-4) and protrusive (small but highly receptor-enriched interfaces; ICDs NCAM-1, JAM-B, DLL1; MUC-4 and ICAM-1 showed hybrid).
• ICDs dominate adhesion mechanics: Decreasing ECD affinity across >10^3-fold (e.g., ICAM-1 synCAM KD from 0.7 nM to 3 µM) gradually reduced contact angle but still allowed expanded interfaces; deleting the ICD abolished interface formation even with high-affinity ECD. Similar modest dependence on ECD KD for ITGB1 ICD (0.7–110 nM). NCAM-1 synCAM maintained high interface enrichment across KD 0.7–600 nM.
• Competition sorting corroborated ICD dominance: Higher-affinity anti-GFP (ICAM-1 ICD) variants preferentially co-sorted with GFP bait; increasing synCAM expression (GFP-ICAM-1/anti-GFP-ICAM-1) increased contact angle, whereas higher expression of tethers did not.
• Surface spreading mirrored two classes: Expansive ICDs drove uniform circular spreading with cortical actin at periphery; protrusive ICDs produced ‘fried-egg’ morphology with lamellipodial/filopodial protrusions and actin localized away from periphery. Spreading required actin polymerization (blocked by latrunculin B); contractility modulated structure (blebbistatin uncoupled structured assembly); Arp2/3 inhibition (CK666) disrupted lamellipodia for protrusive ICDs.
• Mechanistic model: Expansive ICDs recruit adaptors (β-catenin, talin, vinculin, ERM) that engage cortical actin to expand interfaces; protrusive ICDs recruit PDZ scaffolds/lipid rafts leading to clustered assemblies that trigger N-WASP/Arp2/3-driven protrusions; mutational analyses of ICD motifs supported these roles.
• Asymmetric interfaces: Large interfaces require balanced expansive ICDs on both sides; pairing two protrusive ICDs maintained enrichment; unbalanced pairs (protrusive vs expansive) yielded asymmetric wrapping (protrusive cell encircles expansive cell). Tether on one side disrupted interface formation.
• Orthogonal ECD toolkit: Multiple distinct antibody–antigen ECDs produced functional, orthogonal synCAMs enabling programmed assemblies: alternating A+B pairs, A–B–C bridges, and cyclic A→B→C→A rings forming predicted minimal 3–4 cell motifs; synCAM-defined connectivity dominated nearest-neighbour distributions.
• Homophilic synCAMs: Coiled-coil ECDs (Aph4, IF1) with fibcon spacer yielded predictable sorting and multi-compartment assemblies; in pairwise mixes, IF1 cells localized to centers vs ECAD or Aph4; ECAD and Aph4 formed barbell structures; in three-way mixes, ECAD–Aph4 barbell with IF1 core emerged.
• Interfacing with native tissues: Anti-PCAD synCAM (ICAM-1 ICD) cells intercalated into PCAD-dependent spheroids, whereas non-synCAM cells were excluded.
• Function in primary/iPS-derived cells: ICAM-1-based synCAMs relocalized to heterotypic interfaces in primary human dermal fibroblasts, mesenchymal stromal cells, and iPS-derived smooth muscle cells; tethers did not.
• Tissue remodelling: In ECAD vs PCAD L929 mixtures, stronger synCAMs (ICAM-1 or ECAD ICD) converted segregated bilobed structures into integrated mixed compartments; weaker tethers produced core–shell layering with modest increase in heterotypic contacts. Bridging PCAD-L929 spheroids to MDCK monolayers with stronger synCAMs transitioned from segregated spheroids (no synCAM) to aster-like bumps (ECAD ICD) to cooperative lattice-like networks (ICAM-1 ICD) with reduced MDCK confluence, indicating emergent mechanical coupling.

Findings demonstrate that CAM function is modular: extracellular recognition specifies connectivity and affinity, while intracellular domains dictate cytoskeletal engagement, interface mechanics, and morphology. ICD identity plays a dominant role over ECD affinity in determining interface strength, size, and enrichment, aligning with models where cortical tension and actin coupling drive adhesion and sorting. The synCAM approach generalizes across diverse ICDs and ECDs, enabling orthogonal, tunable control of both bonding patterns and interface phenotypes. Combining synCAMs and native CAMs allows construction of complex, mechanically coupled tissues that exhibit emergent organization (e.g., lattice networks), offering a platform to probe and program multicellular self-organization. The work also provides evolutionary insight: simple extracellular recognition modules could have been co-opted by metazoans via recombination with cytoskeleton-coupling ICDs to produce diverse adhesion systems.

The study introduces a versatile, modular toolkit of synthetic cell adhesion molecules that decouple extracellular connectivity from intracellular mechanical outputs. synCAMs can: (1) create native-like, tunable cell-cell interfaces; (2) program specific multicellular architectures using orthogonal heterophilic and homophilic interactions; (3) intercalate into and remodel tissues formed by native CAMs; and (4) function across cell types, including primary and iPS-derived cells. Future work could expand the repertoire of orthogonal ECDs and ICDs, incorporate regulatory features of natural ECDs (cis-oligomerization, catch bonds, allostery), refine in vivo applicability, and explore therapeutic applications in tissue repair, immune cell trafficking, and neural circuit modulation.

• The engineered ECDs are simplified recognition modules and lack regulatory sophistication found in many natural CAM ECDs (e.g., cis-oligomerization, catch bonds, allosteric conformational changes), which may influence adhesion dynamics in vivo.
• Experiments were conducted primarily in vitro using model cell lines and controlled surfaces; generalizability to complex tissues and in vivo contexts remains to be established.
• Adhesion strength and outcomes depend on expression levels and cellular cytoskeletal state; potential variability across cell types and physiological conditions warrants further study.