Engineering polymer-binding bispecific antibodies for enhanced pretargeted delivery of nanoparticles to mucus-covered epithelium

Abstract

Mucus represents a major barrier precluding sustained and targeted drug delivery to mucosal epithelium. Ideal drug carriers should not only rapidly diffuse across mucus, but also bind the epithelium. Unfortunately, ligand-conjugated particles often exhibit poor penetration across mucus. Here, we explored a two-step “pretargeting” approach through engineering a bispecific antibody that binds both cell-surface ICAM-1 and polyethylene glycol (PEG) on the surface of nanoparticles, effectively decoupling cell targeting from particle design and formulation. When tested in a mucus-coated Caco-2 culture model that mimics the physiological process of mucus clearance, pretargeting increased the amount of PEG-particles binding to cells by ~2-fold or more compared to either non-targeted or actively targeted PEG-particles. Pretargeting also markedly enhanced particle retention in mouse intestinal tissues. Our work underscores pretargeting as a promising strategy to improve the delivery of therapeutics to mucosal surfaces.

Table of Contents Text:

Enhancing transmucosal drug delivery: Densely PEGylated particles can rapidly diffuse across mucus but fail to specifically bind the underlying cells (see picture; top). Here, through engineering a novel PEG-binding bispecific antibody, we introduce a two-step, “pretargeted” concept to increase delivery of PEG-coated particles to mucus-covered epithelium (bottom).

Sustained and targeted delivery of therapeutics to the underlying epithelium of mucus membranes remains a long-standing challenge in the field of drug delivery. The challenge is particularly exacerbated by the continuously secreted layer of mucus, a highly viscoelastic and adhesive gel that traps drug carriers and quickly eliminates them via natural clearance mechanisms.^[1] A breakthrough for increasing particle delivery through mucus came over a decade ago with the discovery that a dense coating of polyethylene glycol (PEG) on particles can effectively minimize particle adhesion to mucins.^[2] This led to the development of “mucus-penetrating” particles (MPP) that could quickly traverse mucus, which in turn achieved more uniform distribution at mucosal surfaces and prolonged retention compared to conventional mucoadhesive particles, and consequently superior drug delivery performance.^[3]

For targeted delivery to the underlying epithelium, ideal drug carriers should not only quickly traverse the viscoelastic gel layer, but also bind to cells in the underlying epithelium. Unfortunately, the PEG coating on MPP reduces their interactions with target cells,^[4] leading to limited binding and uptake by the underlying epithelium. Although conjugating ligands, such as lectins or cell-penetrating peptides, can facilitate particle association to cells, their incorporation on the surface of particles can result in non-specific adhesive interactions between the particles and mucins, leading to reduced permeation through mucus.^[5]

To enable both rapid mucus penetration and efficient association to the epithelium simultaneously, we decided to explore a two-step, “pretargeting” approach that decouples the cell targeting from particle design and formulation. This approach is enabled by “bispecific” pretargeting molecules that can bind both specific epitopes on target cells and the drug-carrier particles. When delivered first, the pretargeting molecule will rapidly diffuse through mucus and accumulate on the cell surface, and capture subsequently dosed particles that permeate through the mucus layer. Conceptually, the pretargeting approach satisfies both requirements of rapid mucus penetration and cell-binding. In practice, pretargeting involves careful engineering of the pretargeting molecule.

Here, as a proof-of-concept that we can facilitate targeted delivery of MPP to epithelial cells, we generated a bispecific antibody against two epitopes, ICAM-1 that is constitutively expressed on the luminal surface of gastrointestinal (GI) epithelium,^[6] and PEG on the surface of MPP. By binding directly to PEG, the design and formulation of our nanoparticles is substantially simplified, and we avoid potentially compromising the ability of densely PEGylated nanoparticles to rapidly penetrate mucus. To overcome the traditional shortcomings of bispecific antibody production, including low yields and purity due to misassembly of heavy and light chains during co-expression,^[7] we engineered a “Fab-IgG” bispecific antibody comprising distinct Fab domains separated by a flexible peptide linker with each Fab possessing site-specific amino acid mutations in the heavy–light chain interface to impart orthogonal heavy and light chain pairings (Figure 1 a & b).^[8] This approach ensures proper assembly of heavy and light chain pairs, and consequently high yields and purity (Figure 1 c) from simple co-expression of a single heavy and two light chain plasmids, while largely retaining native antigen binding affinities. Indeed, our engineered α-ICAM1 x α-PEG Fab-IgG binds to ICAM-1 and PEG with comparable affinities as native α-ICAM1 and α-PEG IgG, respectively (Figure 2 a & b). We further verified the ability of Fab-IgG to bind PEG on the surface of 100 nm fluorescent MPP (>2 PEG/nm²; Table S1) via an immunoblot assay (Figure 2 b inset).

Figure 1.

Bispecific “Fab-IgG” antibody used for pre-targeting MPP to mucus-covered epithelial cells. a) Schematic of the structure of “Fab-IgG” that binds ICAM-1 and PEG. Orthogonal amino acid mutations, denoted using the Kabat numbering scheme, were introduced to ensure proper heavy and light chain pairings. b) Corresponding sequence map of the one heavy chain and two light chains used to produce “Fab-IgG” via simple co-expression. c) SDS-PAGE (non-reduced and reduced) stained with Coomassie shows molecular weights and purity of bispecific (bsAb) “Fab-IgG” compared to native IgG controls (α-ICAM1 and α-PEG IgG). The additional Fabs in bispecific Fab-IgG increase its overall molecular weight to ~ 250 kDa. Intravenous immunoglobulin (IVIG) was used as IgG standards.

Figure 2.

Binding specificities for bispecific Fab-IgG. ELISA assays show Fab-IgG (α-ICAM1 x α-PEG) specifically binds a) ICAM-1 and b) PEG with affinities that are similar to their respective IgG controls. Anti-PEG IgG and anti-ICAM1 IgG serve as negative controls for ICAM1 and PEG binding, respectively. Inset: Immunoblot assay confirming Fab-IgG specifically bind PEG on the surface of MPP. Epithelial cell uptake of nanoparticles in the absence of mucus measured by flow cytometry, reported as c) median fluorescence intensity (MFI) and as d) percentage of epithelial cells binding and taking up nanoparticles out of entire gated population. Differences in pretargeted and active targeted particles were not significant (n.s.) as determined by Tukey’s HSD test for multiple comparisons following one-way ANOVA. **** indicates p < 0.0001 of indicated samples vs. both pretargeted and active targeted particles.

We next verified the ability for our Fab-IgG pretargeting molecule to facilitate cellular uptake of MPP using monolayers of non-polarized Caco-2 cells (Figure 2 c & d). Pretargeting with Fab-IgG increased the amount of particles associated with Caco-2 cells by more than 15-fold (p < 0.0001) compared to non-targeted PS-PEG (Figure 2 c), with nearly 100% of the cell population internalizing particles (Figure 2 d). The extent of pretargeted particle uptake is comparable to “active-targeted” particles prepared by pre-mixing the Fab-IgG with PEGylated MPP prior to addition to cells.

We next sought to validate our assumption that the presence of targeting ligands on actively targeted particles would impede particle diffusion through mucus, whereas pretargeted MPP, due to absence of cell-binding ligands on the particle surface, would quickly permeate mucus. We collected fresh, undiluted GI mucus secretions from mice, and performed high resolution multiple particle tracking microscopy to quantify the mobility of different particle formulations in mucus gel.^{[2a, 9]} Unmodified PS-PEG readily penetrate mouse GI mucus, as evident by their diffusive, Brownian trajectories that spanned many microns over the course of 20 s movies (Figure 3 a). In contrast, actively targeted particles exhibited highly constrained, non-Brownian time-lapse traces in the same mouse GI mucus secretions. The time-scale-dependent ensemble mean-squared displacement (<MSD>) of actively targeted PS-PEG was ~65-fold lower than that for unmodified PS-PEG at a time scale of 1 s (Figure 3 b), with most of the actively targeted PS-PEG (~ 85%) effectively trapped in mucus (exhibiting traces less than their own diameters; Figure 3 c & d).

Figure 3.

Transport of PEGylated nanoparticles (Non-Targeted PS-PEG) versus the same particles modified with surface ligands (Active Targeted PS-PEG) in fresh, undiluted mouse GI mucus. a) Representative trajectories for particles exhibiting effective diffusivities within one SEM of the ensemble average at a time scale of 0.2667 s. b) Ensemble-averaged geometric mean-square displacements (<MSD>) as a function of time scale. c) Distributions of the logarithms of individual particle effective diffusivities (D_eff) at a time scale of 0.2667 s. The criterion used to classify particles as effectively trapped (left of the dotted line) was a displacement of less than ~ 100 nm (i.e., less than the particle diameter) within a time scale of 0.2667 s. d) Ensemble-averaged geometric effective diffusion coefficients (<D_eff>) at a time scale of 0.2667 s from different GI mucus samples (indicated by different shaped and colored markers). * indicates statistically significant difference (p < 0.05) using paired, two-tailed Student’s t-test. Data represents five independent experiments with greater than 100 particle traces per experiment.

Finally, we sought to confirm whether pretargeting enhances nanoparticle targeting to cells with an overlaying mucus layer. Unfortunately, there are currently no in vitro cell culture models that can mimic continuous mucus clearance – a hallmark of physiological mucosal epithelium.^{[1a, 1b]} To partially recapitulate mucus clearance, we developed an in vitro culture model, whereby a uniform mucus layer is applied on transwell inserts, and placed on top of a monolayer of non-polarized Caco-2 cells (Figure 4 a). In this setup, mucus drainage is readily achieved by simply interchanging the transwell inserts. Using this model, we found the uptake for non-targeted and actively targeted nanoparticles were significantly reduced by nearly 2-fold or more (p < 0.0001) compared to pretargeted nanoparticles (Figure 4 b), a reflection of the mutually exclusive tradeoffs typically encountered in nanoparticle design and formulation when trying to incorporate the features of either rapid mucus penetration or efficient association to the underlying epithelium. We further compared the distribution and retention of non-targeted versus bispecific pretargeted nanoparticles in an ex vivo mouse intestinal perfusion model and found our pretargeted strategy to retain significantly more MPP while still preserving broad distribution of particles throughout mucosal tissues (Figure S1). These results underscore the potential promise of pretargeting to enhance particle delivery to mucosal surfaces.

Figure 4.

Uptake of nanoparticles by Caco-2 intestinal epithelial cells with an overlaying mucus layer. a) Schematic depicting the experimental design and samples. The mucus layer (green) is applied on a transwell insert (gray) and replaced after 2 hours (M1 to M2) to represent mucus clearance at the mucosal surface. Uptake measured by flow cytometry and reported as b) median fluorescence intensity (MFI). **** indicates p < 0.0001 of indicated samples vs. pretarget using bispecific Fab-IgG, as determined by Tukey’s HSD test for multiple comparisons following one-way ANOVA.

The concept of pretargeting was first introduced in the field of radioimmunotherapy, where pretargeting with bispecific fusion proteins has shown considerable promise in the preclinical and early phase clinical studies for treatment of blood cancers.^[10] Pretargeting has since been used for targeting a variety of solid tumors.^[11] Nevertheless, pretargeting has remained restricted to systemic application to date, and has never been explored for mucosal delivery. Here, by pretargeting with a carefully engineered, polymer-binding bispecific antibody, we demonstrate that MPP can bind specifically to the epithelium while retaining their ability to rapidly penetrate mucus. This dual functionality of our pretargeted approach holds promise over both muco-inert, non-targeted formulations as well as actively targeted formulations for prolonging retention and sustaining drug delivery from nanocarriers at mucosal tissues. A particular novelty of our system is the incorporation of antigen-binding fragment (Fab) that specifically recognizes and binds PEG, which allows us to utilize otherwise unmodified PEGylated MPP and ensure maximum flux of particles penetrating across the mucus layer. Our Fab-IgG bispecific pretargeting molecule design is compatible with various Fabs while maintaining individual Fab affinity, allowing us to swap in any cell- and polymer-specific Fabs of interest. This facilitates the development and screening of diverse libraries of pretargeting molecules spanning a range of antigens and bispecific formats to further optimize pretargeting of different MPP to specific mucosal surfaces. We believe pretargeting will likely further enhance the efficacy of a variety of therapeutics delivered to treat localized mucosal diseases, including inflammation, infections, and cancer.

Experimental Section

The general experimental methods were as follows; details are available in the Supporting Information. Bispecific and native wildtype monoclonal antibodies were cloned and expressed transiently in Expi293F mammalian cells followed by affinity chromatography purification, and characterized for molecular size, purity, and relative binding affinities. PEG-coated MPP were prepared by covalent conjugation of 3 kDa MW methoxy-PEG-amine to 100 nm fluorescent carboxylated PS particles,^[12] and characterized for size, surface charge, concentration, and PEG surface coverage. The transport kinetics of particles were assessed in fresh, undiluted mouse GI mucus using multiple-particle tracking.^{[2a, 9]} Cellular uptake of nanoparticles, in the presence and absence of a reconstituted mucin layer, was quantified by flow cytometry using confluent monolayers of non-polarized Caco-2 cells as model epithelium. Pretargeting antibody and fluorescent nanoparticles were dosed directly to the lumen of freshly excised healthy mouse small intestinal segments for ex vivo particle distribution and retention studies. These intestinal segments were then frozen, sectioned transversely onto glass slides, counterstained for cell nuclei with DAPI, and fluorescently imaged.