Soluble and multivalent Jag1 DNA origami nanopatterns activate Notch without pulling force

Notch signaling is an evolutionarily conserved cell-to-cell communication pathway essential for cell fate decisions, tissue patterning, angiogenesis, and neurogenesis. Mammalian Notch receptors (NOTCH1–4) have extracellular domains (NECD) comprising 29–36 EGF repeats followed by the negative regulatory region (NRR) near the membrane. Canonically, activation involves ligand binding followed by sequential proteolytic cleavages at S2 (by ADAM proteases) and S3 (by γ-secretase), releasing NICD to drive transcriptional programs. A prevalent model posits that ligand endocytosis in the ligand-presenting cell exerts a pulling force on Notch, exposing the S2 site within the NRR. Experimental evidence has shown that forces of approximately 4–12 pN can activate Notch, with observations of catch-bond behavior for JAG1 and DLL4. Excessive pulling or disruption of the heterodimerization domain can shed the ECD and expose the receptor for activation. However, several observations complicate a strictly force-dependent model: reports of long-range activation via secreted ligands, activation using bead- or Fc-clustered ligands in vitro, and results from synthetic Notch constructs. While microscale ligand patterning on surfaces has been studied, precise solution-phase patterns had not been tested. The present study re-examines activation mechanisms by presenting Jag1 ligand nanopatterns from solution using DNA origami to stimulate endogenous Notch in human neuroepithelial-like stem cells, probing whether multivalency and binding time can activate Notch without a pulling force.

Prior work has supported a force-dependent activation mechanism for Notch: tension gauge tethers, magnetic tweezers, and force clamp spectroscopy demonstrated that piconewton forces can activate Notch and suggested catch-bond interactions with Jagged and Delta ligands. Structural studies have elucidated the ligand-binding domain and the NRR, with models often depicting a stretched, rod-like EGF repeat array implying that force transmission is required to affect the membrane-proximal NRR. Yet, evidence for long-range or secreted ligand-mediated activation, activation by ligand-coated beads or Fc-clustered ligands in solution, and data from synthetic Notch systems (SynNotch and SNIPR) pose challenges to an exclusively force-based model. Moreover, EM and high-resolution structural studies suggest flexible, non-linear EGF regions, potentially favoring compact tertiary arrangements rather than a rigid rod, which could allow alternative activation mechanisms such as slow conformational unraveling or kinetic segregation effects.

Design and assembly of Jag1 DNA origami nanopatterns: An 18-helix bundle (18HB) DNA origami nanorod was designed in caDNAno (hexagonal lattice), following prior designs. The scaffold was M13 p7560 ssDNA, folded with ~200 staples (20 nM scaffold; 100 nM staples; 13 mM MgCl2; Tris pH 8.5). Folding protocol: 65°C 4 min; 65→50°C at 0.7°C/min; 50→35°C at 1°C/h; hold at 20°C. Excess staples removed by buffer exchange (Amicon 100 kDa, PBS + 10 mM MgCl2). Protein production and conjugation: Jag1Fc (Jagged1 ectodomain fused to Fc; His-tagged) expressed in HEK293T by transient transfection, purified on HisTrap FF. Site-specific DNA conjugation used Bis-sulfone-DBCO reacting with 6xHis at C-terminus (4 h, RT), followed by click ligation to azide-DNA oligos. Conjugation oligo sequences: Jag1, CTCTCCTTCTTCCCTTTCTTT; BAI1, TTCGACAGCATGAACATCAGC. Nanopattern assembly: DNA origami nanorods carried 0, 1, 2, 3, 4, or 8 Jag1Fc attachment sites (28 nm spacing for 1–4x; 14 nm for 8x). Jag1Fc-DNA conjugates added at 20× excess per site and annealed (1 h at 37°C, then 14 h at 22°C). Excess ligands removed by FPLC size exclusion (Superose 6 Increase 10/300GL). Concentrations estimated by agarose gel band intensity versus standards. Characterization: Gel shift assays (2% agarose, 0.5× TBE + 10–11 mM MgCl2, EtBr) verified increasing mobility shift with higher loading. Negative-stain TEM (uranyl formate) imaged empty and 4x JNPs; additional electron tomography used for internal architecture comparison when extra helices were added. DNA-PAINT validated functionalization states: JNPs carried biotin handles for immobilization and docking-site extensions on Jag1 for imager binding. Imaging with TIRF microscopy; Picasso used for localization, RCC drift correction, and custom Python scripts for origami detection, ROI definition, occupancy quantification, position-to-position distance and linearity metrics. Occupancy distributions matched design; majority monomers (72.6%); intersite spacing ~28 nm. Binding kinetics by SPR: Biotinylated Notch1 EGF8–12 was immobilized on streptavidin CM3 chip (~200 RU). JNP analytes (2.5–10 nM in PBS + 10 mM MgCl2) injected at 5 µL/min, 35 µL. Multi-cycle kinetic analysis showed increased association and reduced dissociation with higher Jag1 valency, yielding lower apparent KD for higher-valency JNPs. Cell culture and stimulation: iPSC-derived neuroepithelial-like stem cells (It-NES) cultured on polyornithine/laminin in defined medium with bFGF/FGF. For stimulation, cells were seeded at ~18,750–19,000 cells/cm², allowed to attach 6 h, then treated for 3 h with JNPs normalized by total Jag1Fc protein concentration (accounting for partial labeling). Time-course with 3x JNPs (2–7 h) identified a peak HES1 response at 3 h. Readouts: qPCR quantified HES1 normalized to GAPDH; dose-response across 0.05–6.4 nM total Jag1Fc. Proximity ligation assay (PLA) detected cleaved Notch1 (Val1744) NICD epitope, with Duolink protocol, imaging by TIRF/epifluorescence, and automated analysis in CellProfiler. Immunostaining confirmed expression of Notch1–3. RNA-seq: TruSeq Stranded mRNA libraries, NextSeq550; quantification with Salmon; gene-level DE with tximport/DESeq2; GO enrichment with PANTHER. Controls for force-independent activation: Oligolysine K10 coating (0.5:1 N:P) to neutralize origami charge; zeta potential and gels confirmed charge shift; compared HES1 activation versus uncoated. Surface adsorption assay: qPCR targeting M13 scaffold after harsh NaOH wash quantified trace nonspecific surface-bound JNPs. Clathrin-mediated endocytosis inhibition: Pitstop2 (25 µM) added 2 h before stimulation; transferrin uptake validated inhibition; qPCR measured HES1 with/without inhibitor; RNA-seq examined endocytosis-related gene sets. Protease dependency: ADAM10-selective inhibitor GI 254023X applied at increasing concentrations 2 h prior to stimulation with 4x JNPs; relative inhibition used to estimate IC50 (~1.7 nM). Chimeric constructs to test avidity vs clustering: 1x Jag1Fc positioned centrally with cholesterol-DNA anchors (~70 nm on both sides) to enhance membrane binding; control lacked cholesterol. TEM confirmed monodispersity; confocal imaging of Cy5-labeled JNPs assessed apparent binding. BAI1/Jag1 chimeras: positions on the nanorod occupied by BAI1 (integrin/CD36-interacting) in place of Jag1 to increase avidity without engaging Notch directly; BAIl alone, and mixed compositions with 1x Jag1 were tested. Added DNA helices to increase mass/charge without membrane engagement as another control. Imaging and qPCR readouts assessed activation across these conditions. Statistics: One-way ANOVA with Dunnett’s test for multivalency comparisons; ANOVA with Tukey for PLA; normality assessed by Shapiro–Wilk. Imaging analysis was blinded. Data and code availability provided (nanobase, ArrayExpress E-MTAB-12439, Zenodo).

• Precise DNA origami nanopatterns (JNPs) carrying 1–8 Jag1Fc ligands were assembled and validated by gel shift, DNA-PAINT (designed occupancy achieved; ~28 nm spacing), and TEM.
• SPR showed multivalency enhanced binding: higher Jag1 copy number led to faster association and slower dissociation, decreasing apparent KD (stronger avidity).
• In It-NES cells, Notch activation (HES1 by qPCR) increased with Jag1 multivalency: minimal effect for 1x, significant upregulation when >1 Jag1 per nanopattern, with activation increasing from 2x to 4x and strongest for 8x. ANOVA/Dunnett: P = 0.0026 (some groups) to P < 0.0001.
• Dose-response plateaued at ~1.7 nM total Jag1Fc for all JNPs, indicating multivalency, not bulk ligand concentration, is the dominant factor.
• PLA for cleaved NICD increased significantly with 8x JNPs, confirming receptor activation at the proteolytic level (ANOVA/Tukey P = 0.0011).
• RNA-seq: broad upregulation of Notch target genes after JNP stimulation. Dimerization-dependent targets (e.g., HES1, HES5) increased with higher Jag1 clustering, while HEY1 (likely monomeric NICD-dependent) reached maximal induction even with 1x JNP.
• Force-independent controls: • Neutralizing origami charge with K10 did not diminish HES1 induction, arguing against electrostatic repulsion-driven pulling. • qPCR of washed surfaces detected only trace, sub-linear-range amounts of nonspecifically adsorbed origami; K10 further reduced adsorption, arguing against traction from surface-bound JNPs. • Clathrin-mediated endocytosis inhibition (Pitstop2) did not alter multivalency-dependent activation; RNA-seq did not show marked changes in endocytosis gene expression.
• ADAM10 dependency: selective inhibitor GI 254023X had an IC50 ~1.7 nM (too low for ADAM17 inhibition), indicating ligand-dependent activation proceeds via ADAM10.
• Chimeric constructs support avidity/time-of-binding mechanism: • 1x Jag1Fc with two cholesterol anchors (membrane-binding) increased Notch activation and apparent binding, similar to 4x JNPs. • BAI1-only JNPs did not activate Notch, but combining 1x Jag1Fc with BAI1 on the same rod achieved activation comparable to 4x JNPs. • Increasing mass/negative charge by adding extra DNA helices did not increase activation, ruling out simple size/charge effects.
• Proposed models: prolonged binding allows time-dependent exposure of NRR (e.g., via EGF repeat “unraveling”) without external pulling; alternatively, kinetic segregation under the bulky origami might locally favor ADAM10 activity.

The study addresses whether Notch activation can be achieved from solution without mechanical pulling by the ligand. By presenting Jag1 as nanoscale multivalent patterns on DNA origami, the authors demonstrate robust activation of endogenous Notch signaling in neuroepithelial-like stem cells. The strength of activation scaled with ligand multivalency and with strategies that extended ligand residence time at the membrane, such as cholesterol anchoring or co-display of a membrane-binding protein (BAI1). Controls ruled out multiple potential sources of force: electrostatic repulsion, nonspecific surface adsorption and cell motility, and clathrin-mediated endocytosis. Moreover, inhibition profiles indicated ADAM10 dependence, consistent with ligand-driven activation. These findings suggest that increased avidity and prolonged binding time can allow the Notch ECD to undergo a slow conformational process that exposes the NRR/S2 site for ADAM10 cleavage, even in the absence of externally applied pulling forces. The results complement, rather than contradict, the force model, as pulling would be expected to accelerate such a process. Alternatively, a kinetic segregation mechanism—where the bulky origami and bound complex affect local composition to favor proteolysis—could contribute. The RNA-seq patterns, with differential dependence of targets on NICD dimerization, further support that multivalency modulates the qualitative nature of downstream transcriptional responses. Overall, the work reframes the necessity of force in Notch activation by revealing a binding-time–dominated pathway to activation, with potential implications for designing soluble Notch agonists and dissecting the structural dynamics of the Notch ECD.

Using molecularly precise DNA origami to present Jag1 in solution, the study shows that Notch can be activated without detectable pulling forces. Activation increases with ligand multivalency and with constructs that prolong residence time at the membrane, and is dependent on ADAM10, indicating bona fide ligand-driven activation. Chimeric constructs (cholesterol anchors; BAI1 co-display) replicate multivalency effects without additional Jag1 copies, arguing against a strict clustering requirement and supporting an avidity/time-of-binding mechanism. Two compatible mechanistic models are proposed: (i) slow EGF repeat “unraveling” that exposes the NRR, and (ii) kinetic segregation that favors ADAM10 activity under the nanostructure. Future directions include: developing monomeric Jag1-DNA conjugates to test the necessity of receptor dimerization; high-resolution structural/biophysical studies of full-length NECD dynamics during prolonged binding; testing across Notch receptor/ligand subtypes; and translating these principles to engineer potent, soluble Notch agonists for therapeutic applications.

• Ligand format: Jag1Fc is dimeric, so the minimal requirement for Notch dimerization could not be excluded; monomeric Jag1 constructs were not tested.
• Endocytosis blockade: Pitstop2 does not completely inhibit endocytosis; residual internalization-related forces cannot be entirely ruled out, though data argue against a major role.
• Surface adsorption assay sensitivity: Trace adsorption was below the linear range of detection; while unlikely, undetected contributions cannot be absolutely excluded.
• Structural mechanism remains indirect: The EGF “unraveling” and kinetic segregation models are inferred from functional data; direct structural visualization of NECD conformational changes during activation was not performed.
• System specificity: Experiments were performed in It-NES cells; responses in other cell types or with other Notch receptor/ligand pairs may vary.