Bioactive multi-protein adsorption enables targeted mast cell nanotherapy

Proteins readily and often irreversibly adsorb to nanomaterial surfaces, resulting in denaturation and loss of bioactivity1,2. Controlling this process to preserve protein structure and function has remained an elusive goal that would enhance the fabrication and biocompatibility of protein-based bioactive nanomaterials3-7. Here, we demonstrate that poly(propylene sulfone) (PPSU)8 nanoparticles support the controlled formation of multi-component enzyme and antibody coatings while maintaining their bioactivity. Simulations indicate that hydrophobic patches9 on protein surfaces induce site-specific dipole relaxation on PPSU surfaces to noncovalently anchor proteins without disrupting hydrogen bonding or protein structure. As proof-of-concept, a nanotherapy for enhanced antibody-based targeting of mast cells and inhibition of anaphylaxis3,4 is demonstrated in a humanized mouse model. The ratio of co-adsorbed anti-Siglec-610,11 and anti-FcεRIα antibodies is systematically optimized to effectively inhibit mast cell activation and degranulation. Protein immobilization on PPSU surfaces therefore provides a simple and rapid platform for the development of targeted nanomedicines.

Proteins readily and often irreversibly adsorb to nanomaterial surfaces, resulting in denaturation and loss of bioactivity 1,2 . Controlling this process to preserve protein structure and function has remained an elusive goal that would enhance the fabrication and biocompatibility of protein-based bioactive nanomaterials [3][4][5][6][7] . Here, we demonstrate that poly(propylene sulfone) (PPSU) 8 nanoparticles support the controlled formation of multi-component enzyme and antibody coatings while maintaining their bioactivity. Simulations indicate that hydrophobic patches 9 on protein surfaces induce site-specific dipole relaxation on PPSU surfaces to noncovalently anchor proteins without disrupting hydrogen bonding or protein structure. As proof-of-concept, a nanotherapy for enhanced antibody-based targeting of mast cells and inhibition of anaphylaxis 3,4 is demonstrated in a humanized mouse model. The ratio of co-adsorbed anti-Siglec-6 10,11 and anti-FcεRIα antibodies is systematically optimized to effectively inhibit mast cell activation and degranulation. Protein immobilization on PPSU surfaces therefore provides a simple and rapid platform for the development of targeted nanomedicines.
Nanomaterials that preserve the bioactivity of anchored proteins hold promise for a wide range of applications as nanozymes, diagnostic sensors, and targeted delivery systems [5][6][7] . Although various synthetic platforms have been developed, unmodified nanoparticles (NPs) remain challenging in interfacing with a wide range of proteins without affecting protein folding or function 7,12 . For example, NPs with hydrophobic surfaces tend to irreversibly adsorb protein adlayers, leading to denaturation and associated loss of protein bioactivity 1,2 . It is therefore mandatory to involve hydrophilic "anti-fouling" polymers, such as poly(ethylene glycol) (PEG) 13 and zwitterionic moieties 14,15 , within NP surfaces to prevent this non-specific adsorption.
Given their highly hydrophilic surfaces after modifications, most NPs require the tedious process of chemical coupling to reliably interface with proteins to harness their bioactivity [3][4][5][6][7] . This method has simultaneously enabled the surface display of targeting antibodies but also limited the control to typically a single antibody type with minimal control over the surface density. Protein adsorption thus presents a comparatively facile noncovalent approach to rapidly engineer bioactive NP surfaces with multiple antibodies and proteins. However, application of this approach is hampered by the continued desorption or loss of activity of the protein layer [16][17][18] . Of note, controlling the composition and stability of anchored proteins, particularly for multiple antibodies, has remained challenging even when employing chemical coupling strategies, which have conjugation yields dependent on the reactivity of the NP surface and accessibility of binding sites within proteins.
We report on PPSU NPs that irreversibly adsorb nonspecific proteins without compromising protein function. This ready-to-use protein immobilization platform is effective for both single and multi-component antibody and enzyme coatings with well-controlled composition. To further demonstrate this design freedom, we engineer PPSU NPs into the first antibody-based nanomedicine for desensitizing mast cells (MCs) to allergen. The nanotherapy requires the controlled co-adsorption of two separate antibodies at an optimized ratio to tune the simultaneous signaling of Siglec-6 and FcεRIα receptors in close proximity on MC surfaces. This therapeutic strategy administers allergen immunotherapy without triggering anaphylaxis in a humanized mouse model.
Hollow PPSU NPs (Fig. 1a) were fabricated at scale (Supplementary Fig. 1) according to our previous report 8 . As prepared, the NPs are negatively charged on their surfaces with a zeta potential of ~ -39 mV in water. We ascribed this phenomenon to spontaneous orientation of propylene sulfone (PS, Fig. 1b) at the interface between water and PPSU due to inherent amphiphilicity within the repeating unit. To explore this postulate, explicit solvent all-atom molecular dynamics (AAMD) simulations of PPSU in water were performed to construct a hollow aggregate that mimics the experimentally observed PPSU NPs. Hollow NPs formed in both the 125-chain and 600-chain simulation systems (Fig. 1c) without severe deformation or collapse of shell structures ( Supplementary Fig. 2). By calculation of the Coulombic (polar) and Lennard-Jones (nonpolar) interactions between PPSU and water ( Supplementary Fig. 3), we confirmed increased surface hydrophilicity for simulated NPs when compared to a single PPSU chain (Fig. 1c). Thus, we concluded spontaneous orientation of PSs at the water-PPSU interface.
The simulations showed that the PPSU NPs exhibit comparable or even higher surface hydrophilicity compared to that of PEG. However, we were surprised to observe the rapid adsorption of nonspecific proteins by PPSU NPs in phosphate buffered saline (PBS) ( Supplementary Fig. 4). Protein surfaces are chemically diverse with characteristic distributions of surface accessible hydrophobic patches (Fig. 1d) 9 . Because the hydrophilic state of PPSU surfaces is induced by the aqueous solvent molecules, we inferred that the dominant orientations of interfacial PSs could also be directed by the hydration of adjacent patches experienced at protein interfaces. That is, a hydrophobic protein patch is likely to flip the interfacial PSs while a hydrated hydrophilic patch would preserve the hydrophilic state of PPSU surfaces (Fig. 1e). Such local hydrophilic/hydrophobic switching events thermodynamically favor the screening of electrostatic repulsion among sulfone groups, leading to small site-to-site heterogeneous NP surfaces at PPSUprotein interfaces.
To investigate the mechanism of protein adsorption, the AAMD simulation of the 600-chain PPSU NP was proceeded by adding 6 molecules (the limit set by space constraints) of trypsin in the aqueous system. The last simulation snapshot is shown in Fig. 2a and Supplementary Movie 1. All six trypsin molecules were adsorbed by the PPSU NP. Calculations of trypsin-NP interactions explicitly supported that the hydrophobic interactions were a major determinant of the adsorption process (Fig. 2b). The 3D structure of adsorbed trypsin was preserved at the PPSU surface, supported by the small root-mean-square deviation of approximately 1.5 Å for each trypsin molecule during the simulation ( Supplementary Fig. 5) 19 . This is in contrast with traditional hydrophobic surfaces where proteins were often found to spread and unfold upon hydrophobic binding ( Fig. 1e) 2 . Furthermore, the 6 trypsin molecules were randomly oriented after anchoring (Fig. 2c, Supplementary Fig. 6). The non-specific orientation of trypsin allowed the exposure of their active sites for 5 out of the 6 molecules, demonstrating retention of enzymatic activity even when shielding by further corona formation was taken into account 16 . Dipole relaxation of the PPSU backbone at all 6 PPSU-trypsin interfaces was confirmed via calculating the surface percentage of sulfone groups, revealing significantly enhanced surface hydrophobicity triggered by trypsin adsorption (Fig. 2d, Supplementary Fig. 7). Simulations also showed that water lubricated the PPSU-trypsin contact regime via PPSU-water-trypsin H-bonds (Fig. 2a, inset). Furthermore, the hydration of trypsin, described by the amounts of trypsin-water H-bonds and water neighbors, remained unchanged despite hydrophobic adsorption (Fig. 2e). This retention of protein hydration is an appealing feature for hydrophilic surfaces and essential for preserving the bioactivity of proteins 19 . Taken together, the simulation results suggested that restructuring of the PPSU surface, rather than denaturing the adsorbed trypsin, led to controlled interfacial hydrophobic interactions for preserving protein structure. Of note, similar results were reported by simulating the PPSU NP in a bovine serum albumin (BSA) solution ( Supplementary Fig. 8).
Having understood the distinct protein affinity of PPSU surfaces, we established a facile and versatile process for coating PPSU NPs with protein adlayers (Fig. 3a). In brief, PPSU NPs were incubated with excess proteins in PBS for 5 min at room temperature, followed by thorough washing to remove dynamically adsorbed proteins as well as residual proteins in solution. Saturation of adsorbed proteins on the NPs avoided aggregation in PBS ( Supplementary Fig. 9). The obtained protein-coated NPs showed improved colloidal stability, with their zeta potential dependent on the adsorbed proteins (Fig. 3b). Using BSA as a model protein, small-angle X-ray scattering (SAXS) measurements revealed that the shell thickness of NPs increased from 5.3 nm to 7.1 nm after protein coating (Fig. 3c). The BSA-coated NPs were imaged by transmission electron microscopy (TEM) and cryogenic scanning TEM (cryo-STEM), showing consistent sizes and morphologies with that of the pristine PPSU NPs (Supplementary Fig. 10). To probe the displacement of adsorbed proteins by competitive serum proteins, we coated PPSU NPs with fluorescein isothiocyanate-tagged BSA (FITC-BSA) and incubated the complex of FITC-BSA@NP in pooled human plasma for 48 h at 37 °C. FITC-BSA fluorescence was retained on pelleted NPs but was undetectable in the supernatant, confirming the irreversibility of the protein adsorption process (Supplementary Fig. 11).
We explored the effect of irreversible adsorption on protein bioactivity using various adsorbed proteins as models. Trypsin and green fluorescent protein (GFP) were coated onto PPSU NPs to form complexes of trypsin@NP and GFP@NP, respectively. The stable surface presentation of adsorbed trypsin was demonstrated by matrix assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry (Fig. 3d). To assess the enzyme activity of trypsin@NP, we quantified the significant increase in fluorescence resulting from the proteolytic cleavage of quenched FITC-BSA solutions into FITC-labeled peptides ( Supplementary Fig. 12). This assay confirmed the ability of trypsin@NP to cleave BSA (Fig. 3e). We previously reported that GFP had no detectable fluorescence after encapsulation within PPSU NPs, likely due to exposure to DMSO during NP formation 8 . In contrast, the structure-dependent fluorescence of GFP was readily detectable for GFP@NP (Fig. 3f), verifying surface adsorption to be a reliable and facile process for loading proteins without exposure to denaturing organic solvents. We next assessed the retention of antibody binding affinity, using low mass ratios of antibodies to achieve near 100% adsorption efficiency. Given the incomplete surface coverage of the antibody-adsorbed NPs, BSA as a blocking agent, was subsequently coated to avoid non-specific interactions with unoccupied regions on the NP surface. This two-step pre-adsorption protocol allows us to incorporate various and even multiple antibodies into protein adlayers with wellcontrolled composition (Supplementary Fig. 13). Staining anti-CD4/BSA@NP with secondary gold-coupled antibodies verified the surface presence as well as immunological recognition by the pre-adsorbed primary antibodies (Fig. 3g, Supplementary Fig. 14). Using CD3 as a cellular target, anti-CD3/BSA@NP significantly increased uptake into T cells compared to controls (Fig. 3h).
MCs are tissue-based granulocytes involved in allergic and other responses and are historically difficult to selectively target with therapeutics [20][21][22] . Perhaps the most significant barrier to developing selective therapies for MCs is the difficulty identifying specific surface targets to mediate inhibitory signaling upon FcεRIα activation following allergen binding. Siglec-6 has been identified as such a target 10,11 , but optimizing inhibitory Siglec-6 signaling during FcεRIα engagement has not been achieved. We hypothesized that co-engagement of Siglec-6 with preadsorbed antibodies on PPSU NPs could be optimized to inhibit MC secretion for the prevention of anaphylaxis. To test this hypothesis, PPSU-based nanomedicines that consist of co-adsorbed anti-Siglec-6 and anti-FcεRIα antibodies were prepared with a controlled surface density of anti-Siglec-6 antibodies (Fig. 4a). We confirmed the bioactivity of pre-adsorbed anti-FcεRIα antibodies to bind and activate primary human skin MCs in vitro (Supplementary Fig. 15). Importantly, inhibited secretion was observed when the nanomedicines were administered (Fig. 4b-e,  Supplementary Fig. 16). This result suggested the ability of co-localized engagement to inhibit FcεRI-mediated activation of primary human mast skin cells (Fig. 4c).
MC degranulation cell surface markers (CD107a and CD63) were quantified via flow cytometry to assess the extent of mast cell activation. The experiments showed that lower densities of anti-Siglec-6 on NP surfaces resulted in greater receptor engagement and inhibition (Fig. 4d). The treatment was further evaluated and compared to a variety of controls, including a mixture of free form anti-FcεRIα with anti-Siglec-6/BSA@NP to assess the ability of a nanotherapy to inhibit preor co-activated MCs. The data indicated that localized co-engagement of both Siglec-6 with FcεRIα is necessary to achieve a significant reduction in degranulation (>60%; p < 0.001), which was compared to control formulations where anti-FcεRIα was delivered without being co-adsorbed or without anti-Siglec-6 ( Fig. 4e). Notably, these in vitro data demonstrated the need for both antibodies to be present on the NPs simultaneously.
For in vivo validation, we tested the optimized nanomedicine formulation in a humanized MC mouse model by intravenous injection (Supplementary Fig. 17). Anaphylactic reactions were measured by monitoring changes in body temperature using a digital rectal thermometer and a clinical scoring system (Fig. 4f) 23 . Mice that experienced anaphylaxis had significantly greater drops in body temperature and increased clinical scores (ΔT > 5°C, clinical scores > 2, p < 0.001)). Successful inhibition of anaphylaxis was manifested by responses that were statistically indistinguishable to the PBS negative control (ΔT ≤1°C, clinical score <0.5). Mice treated with the nanomedicine displayed near-complete inhibition of anaphylaxis, whereas other formulations which did not have antibodies co-adsorbed or were freely solubilized did not display any significant inhibition of anaphylaxis.
Using a combination of in silico, in vitro and in vivo methods and a simple mixing protocol, we demonstrate controlled, irreversible adsorption of multiple proteins simultaneously to PPSU NP surfaces while preserving protein function. Highly dynamic PPSU surfaces were capable of siteto-site hydrophilic/hydrophobic switching in the presence of adsorbing protein, which allowed facile binding of bioactive enzyme and antibody combinations. This design freedom provided the flexibility to optimize a novel nanotherapy for anaphylaxis, achieving the first targeted inhibition of FcεRI receptor activation on human mast cells via Siglec-6 signaling. Our work validates PPSU as a robust platform for the facile engineering of bioactive NPs and allows customization of the complex and combinatorial capabilities of biologics in the treatment of disease.

Fig. 1 Protein surfaces induce site-to-site dipole rotation of interfacial propylene sulfone. (a)
Simultaneously recorded cryo-STEM and cryo-SEM images showing a typical PPSU hollow nanoparticle and its surface, respectively. (b) Schematic illustration of the dipole moment of amphiphilic propylene sulfone, the repeating unit of PPSU homopolymer (blue indicates hydrophilic and grey indicates hydrophobic for all the schemes in the work, unless stated otherwise). (c) Atomistic simulations demonstrating enhanced surface hydrophilicity as the aggregation of PPSU chains become disordered. PEG is included for comparison. (d) As seen in trypsin using glycine as the reference, the surfaces of water-soluble proteins are heterogeneous, with characteristic patch size distributions based on hydrophobicity. (e) Protein inducing the formation of locally heterogeneous PPSU surfaces that mimic protein surfaces. Hydrophobic interactions at the interface between this PPSU surface and the adsorbed protein can be weak enough not to outcompete the forces governing protein folding despite irreversible adsorption. Fast spreading of protein on a hydrophobic surface is included for comparison 2 .   (d) Optimizing nanomedicine formulation via adjusting the surface density of anti-Siglec-6. In vitro results showing that lower anti-Siglec-6 density is more effective in suppressing CD107a and CD63 expression. (e) In vitro results demonstrating the importance of binding Siglec-6 in close proximity and time with engagement of FcεRI in suppressing CD107a and CD63 expression by mast cells. The optimized formulation (0.01 wt% of anti-Siglec-6) is used for the nanomedicine; ctr-a: Anti-Siglec-6/BSA@NP + Anti-FcεRIα; ctr-b: Anti-Siglec-6 + Anti-FcεRIα + BSA; ctr-c: BSA@NP + Anti-FcεRIα. (d-e) Inhibition is normalized from mast cells receiving only anti-FcεRIα and expressing a positive population mean of 52.7 ± 5% for CD63 and 71.2 ± 8% for CD107a (n = 3). (f) The optimized nanomedicine (0.01 wt% of anti-Siglec-6) succeeds in administering allergen immunotherapy without triggering anaphylaxis in a humanized mouse model. Results were from 2 combined datasets (total n = 6). Statistical significance is determined by Tukey-post hoc test: * p < 0.05, ** p < 0.01, *** p < 0.005, **** p < 0.001.