Protein–Antibody Conjugates (PACs) – A Plug-and-Play Strategy for Covalent Conjugation and Targeted Intracellular Delivery of Pristine Proteins

Abstract

We report here on protein-antibody conjugates (PACs) that are used for antibody-directed delivery of protein therapeutics to specific cells. PACs have the potential to judiciously combine the merits of two prolific therapeutic approaches, viz. biologics and antibody-drug conjugates. We utilize spherical polymer brushes to construct PACs using the combination of two simple and efficient functionally orthogonal click chemistries. In addition to the synthesis and characterization of these nanoparticles, we demonstrate that PACs are indeed capable of specifically targeting cells based on the presence of target antigen on the cell surface to deliver proteins. The potentially broad adaptability of PACs opens up new opportunities for targeted biologics in therapeutics and sensing.

Graphical Abstract

Protein-antibody conjugates (PACs) have the potential to judiciously combine the merits of two prolific therapeutic approaches, viz. biologics and antibody-drug conjugates for antibody-directed delivery of protein therapeutics to specific cells.

This manuscript describes the design and construction of protein-antibody conjugates (PACs) with the objective of enabling antibody-directed cell-specific delivery of therapeutic proteins. We were inspired by the success of two emergent fields of therapeutics, antibody-drug conjugates (ADCs) and biologics. ADCs have greatly impacted cancer therapy, as they selectively ferry cytotoxic small molecule drugs that are covalently attached to targeting antibodies to cancer cells.^[1–9] With the identification of receptors that are specifically overexpressed on pathogenic cell surfaces, ADCs have garnered considerable attention as a versatile form of targeted therapy.^[10,11] Similarly, the advent of biologics offered the promise of directly addressing cellular functional deficiencies through the direct use of biologics as therapeutics. We define biologics as protein cargos that upon delivery to cells result in a therapeutic response. Biologics have been mainly targeted for extracellular deficiencies or functions, such as with insulin or antibody-based therapeutics. Use of biologics for intracellular function greatly expands the repertoire of biologics, including in cancer therapy, regenerative medicine, gene editing and subunit vaccine.^[12–16] In this paper, we describe the concept of PACs, where the antibody-directed cell-specific intracellular delivery of proteins strategy promises to harness key advantages of both ADCs and biologics. We outline our preliminary findings in this largely unexplored area.

In addition to capturing the advantages of ADCs and biologics, the design of PACs also addresses some of their inherent limitations. For example, utilizing biologics for intracellular therapeutics require the proteins to be efficiently delivered into the cytosol or subcellular organelles with retention of activity.^[17–22] The bottleneck here is that the large and hydrophilic nature of globular proteins render them impermeable to the cell membrane. Furthermore, actively targeted delivery of the protein therapeutics into specific cells is also highly desired to improve the bioavailability at the lesion. Learning from ADCs, proteins can be simply conjugated to an antibody as a fusion protein.^[23–31] Note however that one of the limitations of ADCs is their low drug-to-antibody ratio of ~4 in most formulations. With the antibody-protein fusion analog, this is even more amplified, as a reasonable protein-to-antibody ratio (PAR) would be just 1. The PACs promise to address these limitations, thus offering a promising strategy for antibody-directed cell-specific delivery of protein cargo (Figure 1).

Figure 1.

Protein-Antibody Conjugates - combine the advantages of protein therapeutics and antibody-drug conjugates while addressing their limitations

To ensure that the PAC platform is versatile while addressing these limitations, we stipulate that our system satisfy the following criteria: (i) the platform is amenable for encapsulating a broad range of proteins; (ii) the PAR value is high; (iii) the encapsulated proteins are protected from proteolysis, as this impacts the stability of the cargo during circulation; and (iv) the methodology is synthetically efficient and carried out under mild conditions.

The molecular design that satisfies all these requirements is shown in Scheme 1. Note that the targeting antibody and the protein cargo are both large hydrophilic proteins. But, the spatial placement of these two macromolecules in our nanoscale system must be precise. The antibodies must decorate the surface of the nanoparticle, while the cargo protein must be present inside such that it is not accessible to the degradative proteases. We envisaged the possibility of utilizing spherical polymer brushes^[32,33] for this purpose. To specifically place the antibody and the protein on these brushes, we use two efficient orthogonal click reactions, viz. the ultrafast boronic acid-salicylhydroxamate click^[21,32–36] for covalent loading of the protein cargo and antibody conjugation via a strain-promoted [3+2] cycloaddition.^[37–40]

Scheme 1.

Construction of PACs with polymer brush as the vehicle for interior protein encapsulation and surface antibody conjugation.

The PAC nanoparticles were synthesized through a ‘graft from’ approach. Silica nanoparticles (~180 nm) were used as the template on which a layer of crosslinked poly(hydroxylethyl methacrylate) (PHEMA) was coated to obtain SiO₂@PHEMA particles. The surface abundant hydroxyl functional groups were transformed to ATRP initiator moieties through an esterification reaction that installed alkyl bromide groups (Figure 2a & 2b; Scheme S1, Figure S1–S3 and Discussion – D1 in the SI). The SiO₂@PHEMA-Br particles were then used to initiate polymer brush formation using two methacrylate monomers (Scheme 1 and S1), one with hydrophilic PEG units and the second bearing trityl-salicylhydroxamate functionalities. The resultant SiO₂@PHEMA-g-PEG particles revealed a well-defined core-shell structure (Figure 2c) with a size of ~320 nm (Figure 2a). Next, the alkyl bromide after polymerization was converted to a clickable azide functional group via a simple nucleophilic substitution reaction. Finally, the salicylhydroxamate group was subsequently liberated with deprotection of trityl groups in the presence of TFA (Scheme S1). Composition of these particles and the corresponding functional group transformations were characterized by a series of techniques, viz. TEM, DLS, IR, and TGA (Figure 2 and S1–S4).

Figure 2.

Characterization of the particles. a) DLS measurement for different particles. b) FT-IR for functional group transformation. c) TEM image of SiO₂@PHEMA-g-PEG. d) TGA curve of different particles.

We then conjugated the antibody onto the surface of nanoparticles for targeted delivery. We chose anti-CD4 as the model antibody to target CD4+ mT-ALL cells, because CD4+ T-cells play a crucial role in the immune system and are involved in many malignancies.^[41–44] First, the antibody was modified with a DBCO-bearing linker (Scheme S1) and the modification was quantified to be ~ 0–2 DBCO group/antibody (Figure S5). This functionalized antibody was clicked onto the polymer brush with a conjugation efficiency of 68%, as assessed by fluorescence quantification (Figure S6). This translated to the average antibodies on each particle to be ~1120 (Figure S6).

Prior to protein conjugation, we tested the selectivity of these antibody-conjugated nanoparticles towards CD4+ mT-ALL cells. To image these particles, before conjugation with the antibody, a small percentage of the terminal azide moieties on the polymer was labelled with a sulfonated Cy5-DBCO dye. We used flow cytometry to evaluate the cell binding ability of this Cy5-labelled anti-CD4-decorated polymer brush toward CD4+ mT-ALL cells at 4 °C. The Cy5 labelled polymer brush without the antibody conjugation was used as a control. The antibody-decorated nanoparticles increased the surface staining of CD4+ mT-ALL cells by more than 12-fold, compared to the control CD4- mT-ALL cells (Figure 3a, –3b) and DO11.10. clone 2 cells (Figure S7, S8). Furthermore, the staining process also showed a dose-dependent relationship. Additionally, we also tested the selectivity of antibody conjugated nanoparticles towards human epidermal growth factor receptor 2 (HER2) in SKBR3 cells using MCF10 A cells as negative control. As observed with anti-CD4 conjugated nanoparticles, trastuzumab-conjugated nanoparticles stained only SKBR3 cells at 4 °C and resulted in higher cellular uptake at 37 °C (Figure S9, S10). These results were promising as the selectivity after antibody conjugation can significantly enhance the interaction between the particles and the cells, potentially leading to better cell uptake.

Figure 3.

Antibody conjugation and protein encapsulation. a) flow cytometry of the polymer brushes surface staining the targeted CD4+ mT-ALL cells at 4 °C, b) quantified cell surface staining from flow cytometry (PB indicates polymer brush. PB^Ab indicates antibody conjugated polymer brush), c) SDS-PAGE gel for GFP-BA encapsulation with varying polymer brush to GFP-BA ratios, d) Protease stability of GFP under trypsin digestion (PACs are Protein-antibody conjugates, cargo protein: GFP)

Encouraged by these results, we then evaluated the protein encapsulation efficiency of polymer brushes to generate the targeted PACs. We used the salicylhydroxamate-boronic acid reversible click reaction,^[21,45] for covalent complexation between the protein cargo and polymer chains. To facilitate encapsulation, the protein surface was modified with an arylboronic acid terminated linker and then simply incubated with SiO₂@PHEMA-g-PEG particles. We tested the protein encapsulation efficiency using SDS PAGE gel with GFP as the model protein. The dose dependent reaction was studied via variation of the ratio between the nanoparticles and the boronic acid modified GFP (GFP-BA). The free GFP-BA protein band gradually faded with an increase in the amount of the PB particles, suggesting protein encapsulation (Figure 3c). When the feed ratio was increased to ~15:1 (meaning ~7% loading capacity), the protein was completely encapsulated. Based on this encapsulation, the amount of GFP in each particle was calculated to be ~1.72*10⁴ (Figure S6).

A key advantage proposed for the PACs involves the possibility of attaining high PAR in this format. Based on the estimated number of surface antibodies and the number of GFP proteins encapsulated per nanoparticle, the PAR value is ~15 for these PACs. (Figure S6) This PAR is already quite high compared to one that is typical for PACs based on fusion proteins. Furthermore, PAR can be simply controlled the number of surface antibody per nanoparticle and the theoretical maximum of the PAR value can reach to 1.72*104 for one antibody per particle (Figure S11). PAR can also be further increased through optimization of protein encapsulation inside the particles.

The encapsulation process is complete in less than 12 h (Figure S12). This reaction kinetics is understandably slower than the corresponding linear polymer, where the conjugation was complete in ~10 min.^[21] The reaction is expected to be slower here because the salicylhydroxamate groups are sterically more encumbered in these spherical polymer brush nanoparticles. This densely packed environment, on the other hand, potentially offers proteolytic stability for the encapsulated proteins. We used the intrinsic fluorescence of GFP to monitor its integrity in the presence of a protease – trypsin (Figure 3d). Loss of the fluorescence was attributed to the proteolysis of GFP by trypsin.^[46] Both GFP and GFP-BA showed fast degradation in the presence of trypsin. However, the encapsulated GFP-BA inside the PB showed very little degradation, indicating that our PAC architecture indeed protects the protein from proteases.

With these promising encapsulation results, we tested the PACs for intracellular protein delivery. SKBR3 (HER2+) cells and MCF10A (HER2-) cells were incubated with GFP-BA loaded polymer brushes with and without antibody functionalization. Negligible green fluorescence was observed in MCF10A cells which was comparable to unfunctionalized GFP-BA loaded polymer brushes in SKBR3 cells. On the contrary, GFP delivery in SKBR3 cells was improved upon conjugation of trastuzumab to polymer brushes i.e., in the PACs format (Figure S13). These results suggest that the protein delivery efficiency of the bare polymer brush nanoparticle alone is very low. The lack of cellular uptake can be attributed to the high density of exposed PEG chains on the surface of the particles that are quite non-fouling (Figure S14, S15).^[47–51] Like most nanoparticle systems, PACs bear the possibility of uptake and clearance by macrophages in the body. However, we expect that the PEG functionalities and the surface antibodies on the nanoparticles will improve their pharmacokinetics and enable cell-receptor specific uptake while minimizing off-target toxicity and clearance by macrophages.

We then demonstrated the effects of antibody conjugation in the targeted intracellular delivery of proteins in the PAC format using yet another antibody i.e., anti-CD4. The GFP-loaded PACs were incubated with CD4+ mT-ALL cells at 37 °C for 4 h. The cell uptake was then quantified by flow cytometry (Figure 4a,4b). A significantly high GFP fluorescence was seen in the CD4+ mT-ALL cells with the anti-CD4 PACs, compared to both GFP and GFP-BA versions of the protein by themselves. The polymer nanoparticle without the antibody showed slightly higher GFP fluorescence than the bare proteins, suggesting that the mT-ALL cells have slightly better propensity to take up the bare nanoparticles. Note however that this fluorescence is still negligibly small, compared to that observed from the anti-CD4 PACs. As a further support for the antibody-directed cellular uptake, the extent of GFP trafficking was also found to be dose dependent. To additionally test whether the covalently conjugated encapsulation is necessary for the observed antibody-directed uptake, we also tested a physical mixture of GFP with the nanoparticle with and without the antibody. Neither of these combinations yielded any cellular uptake (Figure 4b). We also confirmed the cell uptake through imaging flow cytometry (Figure 4c). The PACs indeed exhibited the highest cell uptake as seen by the bright green color dispersed throughout the cells, while the bare nanoparticles did not induce significant cell uptake. We used the imaging flow cytometry data to obtain statistical curves (Figure S16) and these results correlated well with the flow cytometry data in Figures 4a and 4b. Further, the targeting specificity was also tested in CD4-negative mT-ALL cells, where the PACs and all the control formulations showed negligible cell uptake (Figure S17).

Figure 4.

Targeted delivery of GFP to CD4+ mT-ALL cells. a) Flow cytometry for GFP delivery. b) Quantified cytometry data. Concentration of GFP is 4 μg/mL, unless stated otherwise c) Imaging flow cytometry images (GFP-BA concentrations: 2 μg/mL). (PB indicates polymer brush. PB^Ab indicates antibody conjugated polymer brush.)

To demonstrate the potential utility of this system for targeted delivery of biologics, RNase A was used as a therapeutic protein. Our strategy requires surface modification of RNase A with boronic acid functionality for encapsulation inside the polymer brush. However, as RNase A is a small protein, the modification of surface lysines could decrease or even turn off its activity. ^[21,22] In order to ensure that its activity inside the cells after delivery is retained, we utilized a self-immolative linker between the lysine and the boronic acid moiety (Figure 5a).^[21] The linker is based on a strategically placed disulfide moiety, which undergoes self-immolation for traceless release of the pristine protein (Figure 5a).^[21] We then utilized such boronic acid linker functionalized RNase A (RNaseA-BA) for encapsulation into the polymer brush matrix through the reversible click reaction similar to GFP-BA encapsulation. Dose-dependent study showed ~10% encapsulation capacity of RNaseA-BA inside the particles (Figure 5b) and time dependent study revealed a completed encapsulation in 8 h (Figure S18). With these results, we also calculated the number of RNase A in each of the particles to be 4.62*10⁴ with a PAR of ~41 (Figure S6). With an efficient encapsulation of the protein, we next evaluated the release of RNase A under reducing conditions that mimic the intracellular environment. The protein was gradually released from particles as the appearance of the RNase A band in SDS-PAGE gel (Figure 5c) and the release kinetics suggested the reaction was complete in ~6 h. Further, we tested the activity of the released RNase A (Figure 5d). Compared to the original RNase A, RNase A-BA with surface modification of boronic acid group revealed a significant decrease in activity and subsequent encapsulation inside PACs totally turned off its activity. When the PACs were subjected to the reducing conditions, the released RNase A showed fully restored activity comparable to native RNase A (Figure 5d). These results suggested that the cleavage of the disulfide bond triggers a self-immolation process for pristine protein release with restored activity. We then evaluated the intracellular performance of the PACs using RNase A as a cargo. The PACs indeed induced a concentration dependent toxicity, whereas the RNaseA-BA loaded nanoparticles without antibody conjugation did not induce any toxicity (Figure 5e). For control, the bare nanoparticles with or without antibody conjugation did not exhibit any toxicity, which also reveals the biocompatibility of the carrier itself and that the cytotoxicity is indeed due to the RNase A cargo (Figure S19). We also performed the delivery of the RNaseA-BA loaded particles toward a CD4-negative mT-ALL cell line. Both of the nanoparticles with or without antibody conjugation did not induce any of the cell toxicity (Figure S20). These results highlight the antibody-directed specificity towards the target cells.

Figure 5.

Targeted delivery of RNase A. a) Traceless release of RNase A inside cells through a self-immolation process under a reduction condition. b) SDS-PAGE gel for RNase A-BA encapsulation inside polymer brushes with different protein to polymer brush ratios after 24 h reaction. c) RNase A release kinetics under reduction condition. d) activity assay of RNase A. e) cytotoxicity study toward CD4+ mT-ALL cells.

In summary, we demonstrate here the concept of protein-antibody conjugates (PACs) using a spherical polymer brush-based nanoparticle scaffold. This unique architecture allows the simultaneous loading of the protein therapeutics inside their interior and conjugating antibodies to its surface using two highly efficient and functionally orthogonal click reactions. Significant amount of protein payloads can be encapsulated to achieve high PAR values. The high surface density of PEG functionalities offered protection to the encapsulated proteins, as discerned by the enhanced resistance proteolysis. That same feature also likely contributed to minimizing non-specific cellular uptake, which in turn enhanced the cellular specificity of intracellular delivery of proteins using PACs. Overall, our strategy offers a versatile pathway for antibody-directed protein delivery to specific cells. The strategy is conveniently extendable to most globular proteins, as demonstrated by its use with three different proteins of different size and pI values. These features of antibody-directed delivery of proteins to specific cells open up new opportunities for the emergent therapeutic area of biologics for intracellular targets.