Computational design of non-porous pH-responsive antibody nanoparticles

Targeted delivery of therapeutic cargoes requires nanoparticles that encapsulate cargo, recognize target cells via specific receptors, and then disassemble to release cargo under defined intracellular conditions. Prior designed protein nanoparticles often use one or two static building blocks and are porous, limiting efficient packaging and retention of payloads. Antibodies are attractive targeting moieties, and previous platforms incorporated antibodies into nanoparticles but often left large pores. The research question here is whether modular, non-porous antibody-bearing nanoparticles can be computationally designed to assemble cooperatively from separate components, package protein or nucleic acid cargo, and disassemble in response to endosomal or tumor-associated acidic pH, with tunable transition points. The authors focus on transforming a previously porous octahedral Fc/tetramer assembly (042.1) by adding a symmetry-matched, pH-responsive trimeric plug at the three-fold axes to close pores and enable pH-triggered disassembly.

The study builds on extensive work in protein nanomaterial design and delivery. Prior designed cages and multi-component assemblies demonstrated accurate geometries and some cargo interactions but were typically porous and comprised of one or two components (e.g., King et al. 2014; Bale et al. 2016; Hsia et al. 2016, 2021). Antibody incorporation into delivery platforms has been explored in viral and non-viral systems, including vector retargeting and modular antibody-binding nanoplatforms (Rujas et al. 2021; Kim et al. 2016; Volpers et al. 2003). The authors’ previous 042.1 antibody nanocages assembled antibodies into polyhedral architectures but had large 13 nm pores, complicating cargo retention (Divine et al. 2021). pH-responsive protein design using histidine networks provides a route to tunable disassembly (Boyken et al. 2019). This work integrates symmetry-matched plug design with helical repeat fusions, advanced docking (RPXDock), and Rosetta sequence optimization to create a three-component, programmable, non-porous nanoparticle.

Design: Starting from an octahedral antibody nanoparticle (042.1) formed by six C4-symmetric designed tetramers on four-fold axes and 12 IgG1 Fc dimers on two-fold axes, the authors designed a C3 trimeric plug to occupy the unfilled three-fold axes that presented 13 nm pores. They extended a previously designed pH-responsive trimer by fusing helical repeat proteins to generate over 80,000 distinct C3 trimer fusions using WORMS helical fusion. These were docked into the nanoparticle’s three-fold pore using RPXDock axle docking, aligning C3 axes and sampling translations/rotations and arm lengths. Designability was scored by RPX residue pair transform (rpx) and interface shape complementarity; 6,000 top docks underwent Rosetta sequence optimization at fusion junctions and interfaces. Forty-five trimer–tetramer pairs were selected. Expression and assembly screening: Trimer and tetramer were initially expressed bicistronically and co-purified by IMAC; 16/45 trimers co-eluted with tetramer by SDS-PAGE. Adding sfGFP-Fc to lysates and IMAC yielded three-component co-purification for 5/16. For in vitro control, trimer and tetramer genes were subcloned into separate vectors with His-tags and purified by SEC. Three-component assemblies were formed by mixing purified trimeric plug, tetramer, and Fc or full IgG at optimized stoichiometry (per protomer 1.1:1:1 trimer:tetramer:Fc), incubating overnight, and purifying by SEC. Biophysical characterization: SEC, DLS, and NS-EM confirmed assembly. For design 0432-17, DLS showed hydrodynamic diameter 34 nm with Fc and 40 nm with full IgG; NS-EM showed monodisperse particles and plug-like density in three-fold views. Assembly was cooperative, requiring all three components; mixtures of any two did not form full nanoparticles. Cryo-EM: Pre-SEC samples of 0432-17-Fc were vitrified and imaged on a Titan Krios (300 kV, K3, pixel size 0.42 Å). After preprocessing and ab initio 3D reconstruction without symmetry, heterogeneous refinement yielded four classes, all fully plugged. Non-uniform refinement with octahedral symmetry gave a 7.05 Å map (EMDB EMD-29602). The design model was rigid-body docked and backbone-refined in Rosetta into density, showing close agreement; pores reduced to ~3 nm from 13 nm. Cargo packaging assays: Electrostatic tuning of the plug interior generated positively (0432-17(+)) and negatively (0432-17(-)) charged variants by Rosetta-guided substitutions at interior surface residues not at oligomer or inter-component interfaces. Nucleic acid packaging: Assembled 0432-17(+) with tetramer and cetuximab IgG and a 154-nt pegRNA. Non-denaturing agarose electrophoresis with SYBR Gold and Coomassie detected co-migration; nuclease protection was assessed with Benzonase or RNase A. Protein cargo packaging: Assembled 0432-17(-) with Fc and pos36GFP. SEC with dual wavelength monitoring (A280, A488) quantified co-elution and estimated cargo copy number in 200 mM vs 1 M NaCl; NS-EM confirmed assemblies. pH-responsive disassembly and tuning: Nanoparticles were assembled with AF647-labeled trimer variants and sfGFP-Fc, immobilized on Myc-peptide-coated beads or yeast, and incubated in citrate-phosphate buffers (pH 4.2–7.5). Flow cytometry quantified loss of AF647 (plug) and sfGFP (Fc) signals upon disassembly. Histidine network count and hydrophobic core destabilizing mutations (e.g., 3HIS, I57V, L75A) tuned apparent pKa of dissociation. Cargo release was tested by immobilizing nanoparticles on Myc-displaying yeast, washing, incubating at pH 4.2, and measuring supernatant fluorescence (mRuby2-Fc as assembly marker; GFP as cargo). Cellular targeting and uptake: EGFR-targeted nanoparticles (0432-17-CTX; AF647-trimer, tetramer, mRuby2-Fc plus cetuximab) and non-targeted controls (0432-17-Fc) were incubated with A431 and HeLa cells. Epifluorescence/confocal imaging quantified percentage of labeled cells and integrated intensity per cell area; EGFR-knockout HeLa served as negative control. Data and code: Cryo-EM map EMD-29602; design code at https://github.com/erincyang/plug_design; docking and fusion tools at worms and rpxdock GitHub repositories.

- Successful design of a three-component octahedral nanoparticle (termed 0432) comprising 12 Fc dimers on two-fold axes, six designed tetramers on four-fold axes, and eight designed pH-responsive trimeric plugs on three-fold axes. The selected construct 0432-17 assembled cooperatively only when all three components were present. - Biophysical properties: DLS hydrodynamic diameter ~34 nm for Fc-containing assemblies (PDI 0.05) and ~40 nm for full IgG assemblies (PDI 0.07). NS-EM showed monodisperse particles with plug density along three-fold views. - Cryo-EM structure: 7.05 Å global resolution map with octahedral symmetry (EMD-29602). The density-refined model closely matches the design; pore diameter reduced to ~3 nm from 13 nm in the original 042.1. RMSD values: 1.6 Å between 0432-17 design and density-refined model; 1.9 Å between 0432-17 and 042.1 density-refined models; 4.2 Å between 042.1 design and density-refined model. - Cargo packaging and protection: Positively charged variant 0432-17(+) co-migrated with a 154-nt pegRNA on native gels and protected RNA from Benzonase digestion; RNA was not protected from RNase A, indicating pore size excludes ~60 kDa Benzonase but admits 14 kDa RNase A. Plugless 042.1 did not protect RNA. - Protein cargo encapsulation: Negatively charged 0432-17(-) packaged pos36GFP; SEC co-elution showed approximately 9–10 GFP molecules per nanoparticle in 200 mM NaCl (~40% interior volume occupancy). Packaging abolished in 1 M NaCl, indicating electrostatic driving forces. 042.1 did not co-migrate with GFP. - pH-responsive disassembly tuning: By increasing histidine hydrogen-bond networks and introducing core-destabilizing substitutions, apparent pKa for plug release was increased across variants. Apparent pKa (plug AF647 fluorescence, 50% signal): 0432-17(-)_3HIS_I57V at pH 6.1; 0432-17(-)_3HIS_L75A at pH 5.9; 0432-17(-)_3HIS_I57V_L75A at pH 6.7. Nanoparticle disassembly (sfGFP-Fc release) occurred at lower pH (e.g., pH 5.3, 4.7, ~5.0, respectively). Control variants with 0–2 histidine networks disassembled near pH ~5.0–5.1. - pH-triggered cargo release: Immobilized 0432-17(-) released encapsulated GFP upon incubation at pH 4.2, with significantly higher GFP release than 042.1, consistent with packaging enabled by the plug. - Receptor-mediated uptake: EGFR-targeted nanoparticles (0432-17-CTX) showed substantially higher cellular association than non-targeted controls. In A431 cells, 44% of cells were positive for 0432-17-CTX vs 5% for control. In HeLa cells, EGFR-dependent labeling observed: WT HeLa 45% (with serum) and 68% (without serum) vs EGFR knockout 10% and 7%, respectively.

The study addresses the need for modular, targetable protein nanoparticles that both encapsulate cargo and disassemble under biologically relevant conditions. By designing a symmetry-matched, pH-responsive trimeric plug that interfaces with the existing tetramer-Fc architecture, the authors converted a porous two-component nanocage into a non-porous three-component system with tunable disassembly. Structural analysis by cryo-EM confirmed the intended architecture and reduced porosity. Engineering the plug’s interior electrostatics enabled selective packaging and protection of nucleic acids and positively charged proteins, while histidine network and core mutations tuned disassembly pH across the endosomal and tumor-microenvironment ranges (pH ~5–6.7). The modularity allows straightforward swapping of antibodies for targeting diverse receptors, and experimental data demonstrate receptor-dependent endocytosis into EGFR-expressing cells. Collectively, these results validate a generalizable strategy for programming environmental responsiveness, cargo handling, and targeting into designed protein nanomaterials. The system’s virus-like size and functions, yet modular multi-component architecture, make it a promising platform for precision biologics delivery, particularly for pH-triggered release in acidic microenvironments.

This work introduces a three-component, non-porous antibody nanoparticle platform (0432) designed to assemble cooperatively from independently purified parts, encapsulate protein and nucleic acid cargo electrostatically, and disassemble with tunable pH thresholds. Structural validation at ~7 Å resolution substantiates the designed architecture and reduced pore size. Cargo packaging, protection from nucleases, pH-triggered release, and receptor-targeted cellular uptake were all demonstrated. The platform’s modularity enables exchangeable antibodies for targeting and plug variants for tuning pH sensitivity and cargo specificity. Future research should integrate endosomal escape mechanisms to enhance cytosolic delivery, refine cargo loading efficiency and selectivity, increase structural resolution to guide further interface optimization, evaluate in vivo pharmacokinetics and biodistribution, and extend the approach to additional therapeutic payloads and disease-relevant targets.

- Endosomal escape is not engineered; efficient cytosolic delivery remains to be established. - Packaging efficiency and protection depend on electrostatics and pore size; small nucleases (e.g., RNase A) can access the interior, implying limited exclusion threshold despite plug insertion. - Some histidine-containing plug variants showed altered cargo interactions, reducing packaging of pos36GFP, which may complicate simultaneous optimization of pH tuning and cargo loading. - Stability in the presence of serum decreased labeling in some conditions, indicating potential sensitivity to complex biological fluids. - Cryo-EM resolution (~7 Å) limits atomic-level interpretation of interfaces; higher-resolution structures would further validate design nuances. - The study focuses on in vitro assembly and cell uptake; in vivo efficacy, immunogenicity, and safety remain to be demonstrated.