Modular Platform for Therapeutic Drug Delivery Using Trifunctional Bio-Orthogonal Macromolecular Conjugates

Abstract

Targeted delivery of macromolecular therapeutics holds great promise for overcoming the limitations of conventional small molecules, enabling the modulation of protein–protein interactions and precise genome editing. However, efficient, safe, and cell type-specific delivery remains a major challenge. To address this, we developed a modular platform for synthesizing heterotrifunctional bio-orthogonal macromolecular conjugates (BMCs) by engineering diverse combinations of targeting ligands, cell-penetrating peptides (CPPs), and bioactive cargos. We optimized facile bioconjugation chemistries to generate BMCs with improved yields, structural integrity, and activity. Modular BMCs accommodate diverse components, including antibodies and receptor ligands for targeting, CPPs for intracellular trafficking, and optical probes, therapeutic peptidomimetics, and CRISPR-Cas9 nuclease as cargo to confer specific biological activities. We assayed their utility across multiple applications: BMCs with fluorescently labeled cargo revealed endosomal escape and intracellular accumulation; peptidomimetic MYB transcription factor inhibitor BMCs exhibited potent antileukemic activity against acute myeloid leukemia cells; and Cas9 BMCs achieved rapid delivery and cell type-specific gene editing in human cells. The BMC approach enables the customizable delivery of functional macromolecules, nominating BMCs as a broadly applicable platform for biomedical applications.

Graphical Abstract

INTRODUCTION

The therapeutic potential of proteins and polypeptides—ranging from enzymes and antibodies to growth factors and transcription modulators—has sparked widespread interest in their clinical applications.^1-4 Protein therapeutics exhibit distinctly high specificity and affinity to their biological targets, which are frequently “undruggable” by traditional synthetic small molecule drugs.^5-8 Furthermore, protein therapeutics are typically biodegradable and less likely to generate toxic metabolites, contributing to a more favorable safety profile. Their large surface areas and structural molecular diversity also allow for the design of tailored therapeutics that can be engineered with desired pharmacokinetic properties and conditional activation in tissues.^9,10

However, efficient intracellular delivery of these generally membrane-impermeable macromolecules remains a formidable challenge. A variety of protein delivery systems have been developed, including liposomes, polymeric nanoparticles, viral vectors, and protein-based carriers, to enhance cellular uptake or improve pharmacokinetic bioavailability.^11,12 Despite significant progress, current protein delivery platforms face several critical bottlenecks. In particular, poor selectivity and nonspecific uptake by off-target tissues can lead to systemic toxicity and reduced bioavailability at the intended therapeutic sites of action and thus often compromise the therapeutic efficacy of protein therapeutics.

To overcome these limitations, several bioengineering strategies have been explored to improve the selectivity of protein therapeutics. Incorporating targeting ligands into protein therapeutics offers a strategic advantage in enhancing tissue and cell type specificity. These ligands bind selectively to surface receptors expressed on diseased cells and facilitate receptor-mediated endocytosis, resulting in reduced off-target effects and enhanced cellular uptake.¹³ While targeting ligands have been conventionally assembled with cytotoxic drugs in the form of antibody-drug conjugates to convey targeted delivery of small molecule toxins,^14-16 their direct use for protein therapeutics has not been systematically explored. A key barrier in this approach is the absent or hindered endosomal escape of macromolecular drugs upon internalization through receptor-based endocytosis. As endosomal escape is required for the biological activity of therapeutics targeting nuclear or cytoplasmic compartments,¹⁷ approaches to facilitate endosomal escape with minimal cellular toxicity are required.

Cell-penetrating peptides (CPPs) are relatively short peptides with distinct physicochemical properties that have emerged as effective tools to enhance the intracellular delivery of protein therapeutics.^18-21 CPPs such as TAT, penetratin, and R9 have been widely studied for their ability to transport various macromolecules—including enzymes, antibodies, and transcription factors—into the cytosol or nucleus of target cells.^22-26 Recent advances have focused on designing novel CPPs with enhanced membrane penetrating capabilities and reduced cytotoxicity.^27,28 Although their mechanisms of cell entry are not fully understood, transient membrane disruption is thought to be required for penetration and subcellular translocation. Indeed, we and others have recently reported improved CPPs with enhanced nuclear and cytoplasmic delivery exceeding hundreds of millions of molecules per human cell while maintaining minimal membrane disruption and negligible toxicity in vitro.²⁷

Here, we report a modular synthetic platform for assembling macromolecular protein therapeutic cargos with both targeting ligands and CPPs in the same construct to achieve carrier-cargo-CPP trifunctional bio-orthogonal macromolecular conjugates (BMCs). Using bio-orthogonal synthetic chemistry with widely available reagents, we investigated three different molecular engineering strategies to assemble diverse BMCs with high yield, purity, and biological activity. We show that the modular BMCs platform can incorporate diverse targeting ligands, CPPs, and functional cargo—including an optical probe for subcellular imaging, a therapeutic peptidomimetic for modulating transcription factors, and CRISPR-Cas9 nuclease for therapeutic gene editing—to render specific biological activities. These results nominate BMCs as a broadly accessible and versatile approach for cell type-specific intracellular delivery of bioactive macromolecules for biomedical applications.

RESULTS

Modular Engineering of Bio-Orthogonal Macromolecular Conjugates (BMCs).

First, we sought to design a modular strategy for engineering and synthesizing macromolecular conjugates that would enable cell type-specific delivery, efficient intracellular internalization, and sufficient endosomal escape of bioactive macro-molecules. We envisioned the tripartite BMCs, containing (i) targeting ligand such as antibody or receptor ligand (e.g., plasma protein transferrin) to serve as a carrier for targeted delivery, (ii) CPP to confer membrane translocation and endosomal escape, and (iii) macromolecular cargo to execute the designed intracellular activity (Figure 1A). To enable specific and efficient synthesis, we used bio-orthogonal coupling chemistry, including amines reacting with N-hydroxysuccinimide (NHS) esters, thiols reacting with maleimides, asymmetric disulfide bond formation, and strain-promoted azide–alkyne cycloaddition (SPAAC).^29-32 These reactions were selected due to their high efficiency under mild native conditions, which enabled the preservation of the structure and function of bioactive macromolecules (Figure 1B).

Figure 1.

Scheme for the design and synthesis of bio-orthogonal macromolecular conjugates (BMCs). (A) Graphic illustration of the BMC platform for therapeutic macromolecule delivery. (B) Key reagents used in assembling tripartite BMCs. All essential functional groups for bio-orthogonal conjugation are labeled with their use in the corresponding reactions. (C–E) Synthetic strategies 1, 2, and 3 to construct tripartite BMCs. Arrows with dashed lines represent reactions occurring between the pointed functional groups.

To assemble diverse BMC architectures with different structural features, we investigated three synthetic strategies to assemble tripartite BMCs (Figure 1C-E). These three strategies differed by module assembly sequence and were distinguished by the last module assembled into the BMCs, enabling the efficient construction and screening of BMCs optimized for specific biological activities. Namely, Strategies 1, 2, and 3 can be used to define the functional contribution of specific carriers, cargos, and CPPs, respectively. Importantly, we designed these strategies in a convergent manner and established a concise three-step modular conjugation scheme to produce the final BMC products (Figure 1C-E). To enable this, we designed CPPs with three different functional groups: (i) maleimide to react with the cysteine side-chain embedded in the cargo protein, (ii) azide to enable the bio-orthogonal SPAAC click chemistry, and (iii) desthiobiotin (DTB) functional group to enable quantitative validation of conjugation efficiency and/or affinity-based purification, based on desthiobiotin’s specific binding to streptavidin and elution with biotin,³³ respectively (Figure 1B).

For Strategies 1 and 2 (Figure 1C,D), BMC synthesis proceeds with thiolation of the carrier (1) using sulfosuccinimidyl 6-(3′-(2-pyridyldithio)propionamido)hexanoate (sulfo-LC-SPDP) linker to form the carrier-SPDP intermediate (3). In parallel, specific cargo (2) is reacted with the trifunctionalized CPP via thiol-maleimide reaction to produce the cargo-CPP intermediate (4). Strategies 1 and 2 differ in the specific orientation of the dibenzocyclooctyne-PEG-thiol (DBCO-PEG-SH) linker, which can be conjugated to the cargo versus the carrier, respectively (Figure 1C,D). For Strategy 1, the DBCO-PEG-SH linker is reacted with the cargo-CPP intermediate (4) using SPAAC click chemistry to generate the cargo-CPP-SH (5), which is then conjugated with the carrier-SPDP intermediate (3) via asymmetric disulfide bond formation to generate the final tripartite BMC (6). For Strategy 2, the DBCO-PEG-SH linker is attached to the carrier-SPDP intermediate (3) via asymmetric disulfide bond formation, and the resultant carrier-DBCO (7) is then assembled with cargo-CPP (4) via SPAAC click chemistry to form the final tripartite BMC (6). Both Strategies 1 and 2 enable site-specific installation of the carrier-CPP-cargo tripartite BMC, with explicit control of the cargo valency for functional and therapeutic applications.

In contrast, Strategy 3 leverages the initial modification of the cargo (2) with an NHS-PEG24-DBCO linker to install a DBCO functional group as a handle for bio-orthogonal CPP conjugation at the final step (Figure 1E). The resultant cargo-DBCO intermediate (9) is directly conjugated to a thiolated carrier (8) through the formation of an asymmetric disulfide bond. Lastly, the obtained carrier-cargo-DBCO intermediate (10) is then assembled with an azide-functionalized CPP to construct the final tripartite BMC (11). Importantly, Strategy 3 provides enhanced versatility for screening different CPPs and enables variable CPP valency by controlling the stoichiometry of the NHS-DBCO handle installation.

To investigate the potential versatility and generalizability of the designed BMC synthesis strategies, we chose to test several different cargo, carrier, and CPP components. For cargo, we tested two candidate macromolecules with varying sizes and applications. As the CRISPR-Cas9 system is revolutionizing the field of gene editing, its cell type-specific delivery is highly desirable for biomedical research and therapeutic applications.^34-36 In addition, as a 160 kDa protein, Cas9 represents a large macromolecular cargo with stringent requirement for accurate folding and structural integrity to confer specific activities.^37,38 As a challenging cargo, delivery of Cas9 has been achieved mostly as a nanoparticle (nanocluster, nanocomplex, etc.) format, while most of them lacked specificity to defined cell populations and often accumulated in the liver when administered.^34-36,39,40 Similarly, intracellular protein–protein interactions are a central mechanism of biological signaling, while many of these interactions involve targets frequently considered “undruggable” with small molecule therapeutics.^6,41-43 Here, we selected the 5 kD selective peptidomimetic inhibitor of the MYB:CBP/P300 transcription coactivation complex, termed CRYBMIM, which specifically blocks oncogenic gene expression in leukemia cells while relatively sparing normal hematopoietic progenitor cells.^44,45 Importantly, both Cas9 and CRYBMIM have limited cell-penetrating activities on their own, necessitating improved strategies for efficient intracellular delivery.

Furthermore, to pursue cell type-specific or broad tissue delivery, we evaluated various targeting ligands as carriers, including immunoglobulin antibody and blood plasma carrier proteins transferrin and albumin, given their established use in current medical practice. For the antibody, we specifically selected one targeting the cell surface receptor KIT, given its specific expression by blood stem and progenitor cells, which are important targets for therapeutic gene editing approaches currently being explored to correct monogenic blood disorders, such as sickle cell disease.^46-48

Lastly, we utilized structurally diverse CPPs, including the prototypical cationic CPP TAT, as well as recently developed improved CPPs, TAT-Ebola, TAT-EBV, and A5K (Table S1).^27,49 These CPPs represent different physicochemical classes, including cationic, amphipathic, and chimeric penetrator molecules. In all, this battery of cargos, carriers, and CPPs constitutes a varied set of components with distinct chemical and structural properties intended to elucidate the structure—activity principles of the BMC approach.

BMCs Enable Enhanced Biological Activity of Therapeutic Macromolecules.

First, we used a fluorescently labeled peptidic molecule as cargo and transferrin as a carrier to establish the initial macromolecular conjugation methods. To monitor the intracellular internalization of macromolecular conjugates, we developed a dual fluorophore labeling system to specifically label cargo and carrier molecules, thereby enabling their analysis in live human acute myeloid leukemia (AML) cells using confocal fluorescence emission microscopy (Figures 2A and S1). We first labeled transferrin with a near-infrared fluorophore DyLight800 using NHS-amine chemistry, followed by thiolation using sulfo-LC-SPDP. The resultant Tf-Dy-Light800-SPDP was then reduced using tris(2-carboxyethyl)-phosphine (TCEP) to reveal the thiol, which was then reacted with TAT cargo peptide containing fluorescein amidite (FAM) and a cysteine residue protected with 3-nitro-2-pyridinesulfenyl (Npys), Cys(Npys)-TAT-FAM (Table S1), via asymmetric disulfide bond formation to assemble the final transferrin-DyLight800-TAT-FAM BMC.

Figure 2.

Efficient intracellular delivery of MYB mimetic inhibitor cargos with transferrin BMCs to target AML. (A) Synthetic strategy for constructing Tf-DyLight800-TAT-FAM BMC. Red and green spheres denote DyLight800 and FAM fluorophore, respectively. (B) Time-dependent intracellular internalization of Tf-DyLight800-TAT-FAM BMC and subsequent cargo release in MV-411 AML cells. Representative live cell confocal fluorescence emission microscopy photographs were shown at 30 min, 3 h, and 20 h after 40 nM Tf-DyLight800-TAT-FAM BMC treatment. Mitotracker (magenta) was used to define live cells, and Hoechst dye (blue) was used to define nuclei, with TAT-FAM shown in green and DyLight800 shown in red. Bottom right insets show the magnification of representative individual cells. Scale bar = 20 μm. (C) Synthetic strategy for constructing Tf-CRYBMIM BMC. (D) Antileukemic potency of Tf-CRYBMIM BMC (red squares), compared with CRYBMIM (blue circles) and Tf alone (gray triangles) using MV-411 AML cells treated for 5 days, with cell viability monitored using CellTiter-Glo luminescence, normalized to cells treated with PBS control. Symbols and whiskers show the mean and standard deviation for three biological replicates. Tf stands for transferrin.

Multicolor fluorescence emission live cell confocal microscopy revealed time-dependent intracellular internalization of transferrin-DyLight800-TAT-FAM BMC in cultured MV-411 human AML cells, followed by subsequent cargo release and nuclear accumulation (Figure 2B). Intracellular internalization of transferrin BMCs demonstrated that the modified transferrin carrier remained competent for binding and endocytosis via cell surface transferrin receptors. We also confirmed that the observed intracellular accumulation was not due to cellular toxicity or nonspecific membrane rupture, as cells exhibiting BMC internalization also had normal mitochondrial activity, as measured using the MitoTracker Red FM dye (magenta), a specific fluorescent probe of cellular mitochondrial activity.⁵⁰ Similarly, we confirmed nuclear accumulation of BMC-delivered Cys-TAT-FAM cargo (Figure 2B), as identified by colocalization with nuclear chromatin, as measured with the membrane-permeable fluorescent DNA-intercalating Hoechst 33342 dye (blue).

Likewise, we also validated that the disulfide-conjugated BMC cargo was competent for endosomal reduction and release, likely conferred via the membrane destabilizing activity of TAT CPP.^23,24,26 As indicated in Figure 2B, puncta could be observed within 30 min of BMC treatment with overlapping fluorescence emission (yellow) from transferrin DyLight800 (red) and cargo CPP FAM (green), demonstrating cellular internalization of intact BMCs. Notably, after 3 h of BMC exposure, Cys-TAT-FAM cargo began to appear separately from the transferrin carrier signals and diffusely throughout the cytosol and nucleus. Ultimately, after 20 h of BMC exposure, no cargo and carrier colocalized puncta were observed, consistent with complete release and nuclear delivery. Thus, transferrin BMCs enabled endosomal escape and intracellular cargo delivery to live human cells.

Having confirmed the intracellular delivery of transferrin BMCs, we synthesized transferrin BMCs containing the peptidic MYB inhibitor CRYBMIM (Figures 2C and S2). We have validated the on-target activity of CRYBMIM to suppress MYB-dependent oncogenic gene expression in AML cells using biochemical disassembly of the MYB:CBP chromatin complex, suppression of MYB-dependent gene expression using transcriptomics, and on-target antileukemia activity using functional genetic resistance studies.^44,45 We synthesized CRYBMIM containing 3-nitro-2-pyridinesulfenyl (Npys)-protected terminal cysteine residues (Table S1) and revealed the thiol immediately before conjugation using reduction with TCEP. Then, we thiolated transferrin using the aforementioned SPDP linkage, and reacted CRYBMIM-Cys under oxidizing conditions via asymmetric disulfide bond formation to form the transferrin-CRYBMIM BMC. Treatment of MV-411 leukemia cells with transferrin-CRYBMIM BMC demonstrated more than 4-fold improvement in antileukemia potency as compared to free CRYBMIM and measured using CellTiter-Glo luminescent cell viability assay (IC₅₀ = 1.6 ± 0.17 versus 7.1 ± 0.50 μM, respectively; t-test p = 4.4 × 10⁻³; Figure 2D). Thus, the BMCs' strategy is able to enhance the biological activity of therapeutic macromolecules, due to improvements in their intracellular delivery.

Convergent Construction of Tripartite BMCs with Large Macromolecular Cargos.

To investigate the utility of BMC engineering for cell type-specific delivery of large macromolecular cargo, we synthesized Cas9 tripartite BMCs using all three designed strategies (Figure 1C-E). We used Cas9 proteins engineered with three nuclear localization signals (termed 3xNLS-SpCas9) as the initial template, based on prior work that optimized this construct for intracellular gene editing activity by electroporation (Figure 3A).⁵¹ We reasoned that the activity of Cas9 BMCs may depend on the position of tethering of the carrier to the Cas9 nuclease, given its requirement for complex formation with a guide RNA (gRNA) to identify its target site and the necessity of conformational reorganization of its nuclease domains upon DNA target site licensing to achieve DNA cleavage.^37,38,52-54 Thus, based on the atomic-resolution structure of the gRNA and DNA-bound Cas9,^38,53,55 we engineered three different 3xNLS-SpCas9 variants (V1, V2, and V3) containing single cysteine residue at different positions within Cas9, likely to be amenable to macromolecular tethering based on prior fluorescent probe coupling studies or avoid interference in binding with the gRNA and DNA substrate (Figure 3A,B, Tables S2, and S3).^52,53

Figure 3.

High yield synthesis of Cas9 tripartite BMCs. (A) Design schematic of the 3xNLS-SpCas9 nuclease, and its variants (V1, V2, and V3), showing three nuclear localization signals (NLS), 6xHis purification tag, and native C107 residue. (B) Structural model of Cas9 nuclease (gray), with its gRNA (purple) and DNA (green) substrates, based on the PDB 4oo8 structure, highlighting V1 (red), V2 (yellow), and V3 (blue) amino acid substitutions. (C) Synthesis strategy 3 used for engineering Cas9 tripartite BMCs. (D and E) Analysis of the reaction intermediates (Cas9 and Ab) and final products (BMC) using SDS-PAGE with silver staining (D) and streptavidin staining (E). TAT peptide functionalized with desthiobiotin is labeled in red. (F) Quantitation of desthiobiotin moieties in Cas9 BMCs to estimate CPP valency, based on the molar ratio of CPP and BMC for BMCs with V1, V2, or V3 Cas9 as cargo, Tf or Ab as carrier, and TAT as CPP. Serial dilution of stock solution of desthiobiotin TAT peptides was used as the reference standard to establish calibration curves, with the transferrin reaction intermediate (Tf-SPDP) used as the negative control. Symbols and whiskers represent means and standard deviations of 3 replicates. (G) Mass photometry histograms of the reaction intermediates and final BMC products measured at 250 nM concentrations, showing relative abundance as a function of measured molecular mass, for the expected tripartite (mean mass = 305 kDa), Cas9-Ab-Cas9 oligomer (mean mass = 465 kDa), and tripartite BMC dimer (mean mass = 610 kDa). (H) Quantitative analysis of the mass photometry data to estimate relative BMC yields, containing V1 Cas9 as cargo, Tf or Ab as carrier, and with or without TAT as CPP. Molecular binding events of tripartite, oligomer, and tripartite dimer were combined in calculating the product yield based on the consumption of Cas9 instead of the antibody/transferrin, which were used in slight excess to Cas9 in the synthesis.

Accordingly, we used Strategies 1 and 2 to construct Cas9 tripartite BMCs. First, trifunctionalized TAT CPP containing the N-terminal desthiobiotin and azide plus C-terminal maleimide functional groups, as well as the dual-functionalized DBCO-PEG-SH linker, were incorporated to tether the cargo Cas9 with the carrier KIT antibody to form the tripartite BMC (Figure 1C,D). We confirmed the site-specific synthesis of the tripartite BMC using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) followed by silver staining for protein analysis under denaturing conditions (Figure S3), as well as quantitative mass photometry of BMC complexes in aqueous solution under native conditions (Figure S4).^56,57 We observed the reaction yields of tripartite BMC were modest with Strategies 1 and 2 (mean 7 and 8%, respectively), which were not improved substantially by varying reaction stoichiometries, temperature, duration, buffer conditions, or linker rigidities and lengths (Figures S3-S5).

In contrast, with Strategy 3, we achieved over 90% reaction yield in assembling Cas9 tripartite BMC, as evident from almost complete consumption of the starting Cas9 material and specific generation of high-molecular mass conjugates (Figures 3D and S6). We confirmed the correct assembly of the Cas9 tripartite BMCs using SDS-PAGE (Figure 3D) and mass photometry (Figure 3G,H), which demonstrated the generation of monovalent carrier:cargo BMC complexes with a mean molecular mass of 305 kDa, in comparison to the mean 160 and 145 kDa mass of the reactants, corresponding to Cas9 cargo and KIT antibody carriers, respectively. We also observed higher-order assemblies that likely represent the coupling of more than one Cas9 protein to the antibody.

To orthogonally validate the correct BMC assembly and the intended CPP installation, we leveraged the desthiobiotin-containing CPP design (Table S1), as it is particularly challenging to measure CPP conjugation solely by changes in electrophoretic mobility or mass photometric light scattering, given TAT’s 2 kDa molecular mass, corresponding to just 7% of 305 kDa total mass of the conjugate. Thus, we confirmed BMC incorporation of desthiobiotin-containing CPP, consistent with the intended BMC design, as assessed by Western blotting using fluorescently conjugated streptavidin (Figure 3E).

With this established method, we synthesized V1, V2, and V3 Cas9-Ab-TAT tripartite BMCs, achieving reaction yields of 98, 97, and 95%, respectively, as estimated by SDS-PAGE silver stain (Figure S7). Thus, BMC assembly occurs efficiently for cysteines located at all three different surface positions on Cas9, suggesting similar steric accessibility and chemical reactivity for tripartite BMC formation. Leveraging the desthiobiotin-containing CPP design (Table S1), we estimated the BMC valency of installed CPPs using a colorimetric desthiobiotin quantitation assay with free TAT-DTB peptides as calibration reference standards and identified mean CPP valencies of 11, 10, and 9 for the V1, V2, and V3 variants of Cas9-Ab-TAT BMCs, respectively (Figure 3F). Multivalent CPP conjugation may enable enhanced BMC membrane permeabilization and endosomal escape of macromolecule cargos.

To test the robustness and potential generalizability of the BMC strategy, we used two additional carriers, transferrin and albumin, to generate the respective Cas9 tripartite BMCs, and all constructs were achieved with high reaction yields (Figures 3H, S8, and S9). This produced Cas9-Tf-TAT BMCs with TAT valencies of 5, 6, and 5, for the V1, V2, and V3 variants, respectively, consistent with the differences in the accessibility of the DBCO handle (Figure 3F). Likewise, we constructed BMCs containing different CPPs (TAT-Ebola, TAT-EBV, and A5K) spanning different physicochemical classes (Figures 3D, S7, and Table S1). This included relatively hydrophobic CPPs with limited aqueous solubility such as TAT-Ebola and A5K, for which we implemented a buffer system containing Pluronic F-127 poloxamer based on our prior studies for native protein extraction^37,38,58 (Figure S10). Overall, this effort enabled the successful synthesis of diverse BMCs, which incorporated different cargos, carriers, and CPPs, demonstrating the versatility and generalizability of this modular platform technology.

BMCs Maintain Cas9 Nuclease Activity for Sequence-Specific Gene Editing.

To evaluate whether Cas9 retains native folding and function after BMC synthesis, we assessed the nuclease activity of its tripartite BMCs using synthetic DNA substrates in vitro, and human genome editing in cells (Figure 4A). First, we used a linearized plasmid DNA containing a single Cas9 target site as a substrate for an in vitro DNA cleavage assay to determine the effects of BMC synthesis on Cas9 ribonucleoprotein (RNP) formation with its complementary single guide RNA (sgRNA) and the overall nuclease activity. Next, we electroporated Cas9 BMC RNPs into human Kasumi-1 traffic light reporter (TLR) cells^59,60 —where sequence-specific genome editing induces expression of mCherry fluorescent marker—to quantify intracellular editing efficiency by fluorescence-activated cell scanning (FACS) and confocal fluorescence emission microscopy.

Figure 4.

BMC bioengineering strategy preserves the bioactivity of Cas9 BMC cargos. (A) Schematic of the plasmid cleavage assay of BMC nuclease activity in vitro and upon electroporation in cells. (B and C) In vitro cleavage efficiency of Cas9 BMCs containing V1, V2, or V3 Cas9 variants as cargo, Tf or Ab as carrier, and TAT as CPP. The cleavage efficiency was assessed using gel electrophoresis analysis of linearized DNA cleavage as quantified by image densitometry, normalized to a DNA loading control. Unconjugated native Cas9 variants and 3xNLS-SpCas9 parent molecule were used as baseline controls for comparison. (D-I) Cellular BMC activity assessed upon electroporation in Kasumi-1 TLR cells and measured by FACS of mCherry-expressing cells. Cas9 BMCs containing V1 (D and E), V2 (F and G), or V3 (H and I) Cas9 variants as cargo, Ab as carrier, and TAT, TAT-EBV, TAT-Ebola, and A5K as CPP were tested. Unconjugated native Cas9 variants were used as baseline control for comparison. 3xNLS: 3xNLS-SpCas9; V1/V2/V3_Unconj: unconjugated native Cas9 variants. Data show 3 replicates.

We first validated that the editing efficiency of all three engineered V1, V2, and V3 Cas9 variants matched the parent 3xNLS-SpCas9 in apparent nuclease activity in vitro (Figure S11A-E). Additionally, we also demonstrated that the engineered Cas9 variant possessed adequate stability in serum-containing medium, cell lysates (Figure S11F,G), and storage at 4 °C (Figure S13), as needed for BMC production for biomedical applications. We then tested V1, V2, and V3 Cas9-tripartite BMCs containing four different CPPs, TAT, TAT-EBV, TAT-Ebola, and A5K, and two carriers, KIT antibody and transferrin. All BMC constructs retained enzymatic nuclease activity in vitro (Figures 4B,C, S12, Tables S4, and S5) and robust human gene-editing efficiency upon electroporation in cells (Figures 4D-I and Table S6). Interestingly, we observed differences between nuclease activities in vitro and in cells for distinct BMCs, suggesting that specific BMC configurations may have compensatory effects on RNP assembly in vitro with chromatin accessibility and cofactor activity in cells (Figures 4H,I, S12, Tables S5, and S6). Similarly, conjugating a single highly cationic CPP TAT directly to Cas9-Cys via maleimide—thiol chemistry modestly reduced gene editing activity in vitro and upon electroporation in cells (Figure S14). In contrast, their corresponding tripartite BMCs mostly maintained full biological activity upon electroporation in cells (Figure 4D-G and Table S6). These observations suggest that CPP-induced perturbations of the Cas9 RNP stability can be apparently compensated by carrier conjugation within tripartite BMCs. Collectively, these findings demonstrate that the BMC synthetic approach preserves the Cas9 conformational integrity and enzymatic nuclease activity in diverse tripartite BMCs.

BMCs Enable Rapid and Cell Type-Specific Intracellular Delivery of Functional Macromolecular Cargos.

To assess the real-time intracellular delivery of Cas9 tripartite BMCs, we utilized DyLight488-labeled Cas9 constructs to treat Kasumi-1 TLR cells (Figure 5A). Because Cas9 single-guide RNAs (sgRNAs) engineered for sequence-specific gene editing are 100 nucleotides in length, with only a small portion bound by Cas9 and the remainder solvent exposed, thus potentially interfering with the membrane penetration of Cas9 BMCs, we preloaded cells with sgRNA by electroporation, followed by washing to remove excess sgRNA, and treated sgRNA-loaded cells with Cas9 BMCs (Figure 5A). We used gRNAs labeled with ATTO647 fluorophore to track their subcellular localization using time-lapse live cell confocal fluorescence emission microscopy. After just 30 min of exposure to Cas9 BMCs, where the excess BMCs were removed by cell washing, Cas9 BMC-treated cells displayed significantly more intracellular DyLight488 Cas9 puncta as compared to cells treated with Cas9 alone (Figure 5B,C, mean 10 and 0.80 puncta/nucleus, respectively; one-way ANOVA with Tukey’s test p = 7.6 × 10⁻⁴). In particular, BMC-treated cells exhibited specific colocalization between DyLight488 Cas9 puncta with both Hoechst-stained nuclear chromatin and ATTO647-labeled gRNA, confirming intended RNP assembly and nuclear import of BMC-delivered Cas9 (Figure 5B). Using time-lapse imaging, we observed a time-dependent increase in nuclear accumulation of Cas9, culminating in successful gene editing to induce fluorescent mCherry expression upon 72 h of culture, which is exclusively in BMC-treated cells (Figure 5D,E and Movie S1). In contrast, unmodified Cas9 failed to generate any gene-editing reporter signal despite the observation of occasional intracellular puncta (Figure 5D and Movie S1).

Figure 5.

Rapid intracellular delivery of Cas9 tripartite BMCs. (A) Schematic workflow of experimental measurements of subcellular localization of BMC components using live cell confocal fluorescence emission microscopy. (B) Representative live cell confocal fluorescence emission microscopy photographs of the Kasumi-1 TLR cells treated with ATTO647-labeled gRNA (purple) only versus Cas9-DyLight488 (green) versus Cas9-Ab-TAT-DyLight488 BMC revealed the internalization of the fluorescently labeled Cas9 tripartite BMCs. Hoechst (blue) was used to define the nuclei. z = 0.85 μm. Scale bar = 10 μm. (C) Quantitative image analysis of Cas9-DyLight488 puncta per Hoechst-stained nucleus of cells treated with PBS, gRNA only, Cas9-DyLight488, or Cas9-Ab-TAT-DyLight488 BMC. An average of 56, 46, and 25 nuclei were identified from the gRNA-only, Cas9-DyLight488, and Cas9-Ab-TAT-DyLight488 treated samples, respectively, in triplicate measurements. Symbols and whiskers represent standard deviations of three technical replicates. (D and E) Representative live cell confocal microscopy photographs of cells 72 h post Cas9 tripartite BMC treatment (E) in comparison to cells only (D). mCherry expression (red) was observed in Cas9 tripartite BMC treated cells, indicating successful nuclear delivery of functional Cas9 cargo, conferring sequence-specific gene editing to induce mCherry expression. Scale bar = 10 μm. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001 for ANOVA Tukey’s test.

We next quantified sequence-specific BMC-induced gene editing in cells by measuring the ratio of mCherry-fluorescent cells via flow cytometry, which served as the ultimate readout of the successful BMC nuclear delivery of functional Cas9 (Figure 6A). We compared Cas9-Ab-TAT BMCs comprised of all three Cas9 variants and concentrations using 30 min of BMC treatment followed by cell washing to remove excess BMC. At 4 μM concentration, Cas9 V3 BMCs bearing TAT CPP achieved over 60% gene-editing efficiency as compared to 3.3% with unmodified Cas9 (two-way ANOVA with Tukey’s test, p = 3.0 × 10⁻⁷; Figure 6B,C). Cas9 V3 variant consistently outperformed V1 and V2, suggesting that the respective cystine positions and conjugation sites can influence BMC internalization efficiency (Figures S15 and S16).

Figure 6.

Cell type-specific human gene editing using Cas9 tripartite BMCs. (A) Schematic workflow of Kasumi-1 TLR or HEK393 TLR cells loaded with specific sgRNA by electroporation, followed by treatment with Cas9 tripartite BMCs and analysis of gene editing by flow cytometry and DNA sequencing. (B) Gene editing efficiency of unconjugated Cas9 (gray) versus BMCs (blue) in Kasumi-1 TLR cells as a function of increasing concentration for tested constructs. The Cas9 tripartite BMCs contained V1, V2, and V3 Cas9 as cargo, KIT antibody (Ab) as carrier, and TAT or TAT-Ebola (Eb) as CPP. Bars and whiskers represent means and standard deviations of three biological replicates. (C) Editing efficiencies of 4 μM Cas9 tripartite BMCs in Kasumi-1 TLR cells. Cas9 tripartite BMCs contained V1, V2, or V3 Cas9 variants as cargo, KIT antibody (Ab) as carrier, and TAT (yellow) or TAT-Ebola (Eb, blue) as CPP. Unconjugated Cas9 variants (gray) were included as baseline control for comparison. Bars and whiskers represent means and standard deviations of three biological replicates. (D) Editing efficiencies assessed in Kasumi-1 TLR versus HEK293T TLR cells, treated with 2 μM Cas9 tripartite BMCs containing V3 Cas9 as cargo, transferrin (Tf) or KIT antibody (Ab) as carrier, or TAT or TAT-Ebola (Eb) as CPPs. Bars and whiskers represent means and standard deviations of three biological replicates. (E and F) TIDE analysis of Sanger sequencing of the TLR locus in BMC-treated mCherry-positive Kasumi-1 TLR cells, showing sequence-specific DNA editing. * p ≤ 0.05,**p ≤ 0.01,*** p ≤ 0.001 for ANOVA ’s test.

Comparative analysis of TAT-Ebola, TAT-EBV, and A5K CPPs revealed that TAT-Ebola conferred improved human gene-editing activities across all Cas9 BMC variants (mean editing efficiency of V3 BMC at 4 μM of 58, 1.5, and 2.1%, respectively; one-way ANOVA with Tukey’s test p = 6.6 × 10⁻⁶ and 6.7 × 10⁻⁶ for TAT-Ebola versus TAT-EBV and A5K for V3 BMC, respectively; Figures 6B,C, S15, and S16).

To verify the accurate BMC gene-editing activity in cells, we purified mCherry-positive cells using fluorescence-activated cell sorting (FACS) and analyzed their genomic DNA (gDNA) using Tracking of Indels by Decomposition (TIDE),⁶¹ which confirmed intended gene-editing at the targeted DNA position within the TLR locus with a total efficiency of 35.2% (R² = 0.84; Figure 6E,F). Thus, both the macromolecular cargo conjugation geometry and CPP identity represent important parameters for effective BMC-mediated intracellular delivery and Cas9 nuclease genome-editing activity in cells.

Finally, we investigated cell type-specific BMC delivery and carrier-dependent targeting effect by assessing BMCs with the specific targeting ligand KIT antibody versus the broad targeting ligand transferrin. We also utilized two different lineage human cell lines to test cell type specificity: Kasumi-1 cells with high KIT and transferrin receptor expression, whereas HEK293T cells had low KIT and moderate transferrin receptor expression. We analyzed four different Cas9 tripartite BMCs, with KIT antibody or transferrin serving as carriers, and TAT or TAT-Ebola serving as CPPs (Figures 6D and S17). We found that the KIT antibody conferred strong specificity and enabled targeted delivery of the corresponding Cas9 tripartite BMCs, with V3-Ab-TAT and V3-Ab-TAT-Ebola tripartite BMCs specifically internalized in KIT-positive Kasumi-1 TLR cells as compared to KIT-negative HEK293T TLR cells (mean gene-editing efficiency of 38 and 33 versus 1.5 and 3.0%, respectively; one-way ANOVA with Tukey’s test p = 1.1 × 10⁻⁷ and 1.1 × 10⁻⁶ for V3-Ab-TAT and V3-Ab-TAT-Ebola, respectively; Figures 6D and S17). In contrast, given the broader expression of its receptor, transferrin carrier BMCs were able to convey delivery into both cell types (21 and 28% versus 5 and 20% for V3-Tf-TAT and V3-Tf-TAT-Ebola BMCs in Kasumi-1 TLR and HEK293T TLR cells, respectively; Figures 6D and S17). Thus, the carrier-dependent cell specificity, combined with modular flexibility in selecting CPP and cargos, demonstrates the versatility of the BMC platform and its potential for cell type-specific intracellular delivery of macromolecular therapeutics.

DISCUSSION

We have established a convergent, modular platform for synthesizing tripartite BMCs that integrates diverse macromolecular cargos—including small molecule optical probes, peptidomimetics, and macromolecular nucleases —with targeting ligands, including antibodies and blood carrier proteins, and cell penetrating peptides conferring membrane permeabilization via a three-step bio-orthogonal synthetic strategy. In particular, the described Strategy 3 consistently achieved greater than 90% synthetic yields across diverse cargo types, from short peptides to 160 kDa macromolecular enzymes. Biochemical and cellular assays confirmed that both inhibitory and catalytic macromolecular cargos retained biological activity upon BMC synthesis, while live-cell confocal microscopy imaging demonstrated rapid cytosolic and nuclear BMC macromolecular delivery on the time scale of minutes, yielding therapeutic functional effects, including anticancer effects of peptidomimetic transcription factor inhibitor and sequence-specific genome editing by engineered Cas9 nuclease in human cells. Lastly, the carrier choice of cell surface receptor-specific KIT antibody and blood plasma transferrin conferred cell type-selective macromolecular delivery to receptor-expressing but not nonexpressing cells.

While BMCs demonstrate robust performance in vitro and in cultured cells, their pharmacokinetics, biodistribution, and immunogenicity in vivo remain to be explored. Likewise, the current study relies on chemical linkers that may exhibit variable stability under physiological redox and proteolytic conditions.Although we observed efficient endosomal escape of BMC cargos, the exact mechanisms and potential side effects of membrane disruption by BMC-installed CPPs will require further study. While we derived the initial structure–activity relationships of tripartite BMCs with respect to the effects of linker length, CPP valency, and carrier density in vitro, their effects on tissue penetration and potential toxicities in vivo need to be defined. Lastly, though the reported BMC synthetic strategy achieved over 90% yield, future studies will need to optimize this approach for large-scale manufacturing and scalability of multicomponent BMCs for therapeutic and clinical-grade applications.

Nevertheless, the developed bio-orthogonal macromolecular conjugate (BMC) platform offers distinct molecular engineering features with important advances for cell type-specific delivery of membrane-impermeable macromolecules, including those targeting protein–protein interactions or conferring enzymatic activities in various subcellular compartments in cells. In addition, adapting cleavable or stimuli-responsive linkers may enhance and control macromolecular cargo release while minimizing systemic exposures. Expanding the BMC cargo repertoire to include enzymes that reprogram metabolic pathways, peptidomimetics, and proteolysis and proximity targeting chimeric molecules (PROTACs) that block and modulate intracellular signaling, synthetic scaffold proteins, and transcriptional effectors, among others, should broaden research applications and translational potential of BMCs. Likewise, coupling BMCs with imaging agents could enable theranostic applications, while combining multiple cargos may enable dynamic and synergistic therapies. Ultimately, the modular BMC platform offers a generalizable strategy for targeted intracellular delivery of diverse macromolecular drugs with promising applications in gene therapy, oncology, metabolic disorders, and many other biomedical applications dependent on macromolecular modulators and therapeutics.

EXPERIMENTAL SECTION

Synthesis of Transferrin-CRYBMIM BMC.

All reactions were carried out in reaction buffer containing 20 mM phosphate, 100 mM NaCl, and 1 mM EDTA, pH 7.5, with 10% glycerol. The buffer pH was adjusted by using NaOH. All reactions were set up in 1.5 or 5.0 mL capped tubes and rotated end over end at the indicated temperature for incubation. The concentrations of the starting materials, intermediates, and final products were quantified using a NanoDrop 2000 spectrophotometer. Extinction coefficients are listed in Tables S1 and S2.

The cargo CRYBMIM-Cys was freshly prepared by reducing CRYBMIM-Cys(Npys) at 10 mg/mL with 20 mM TCEP, followed by solid phase extraction (SPE) purification to remove excess reducing agent. For SPE purification: column, C18 SPE cartridge (Thermo Fisher Scientific); mobile phase A, H₂O with 0.1% trifluoroacetic acid (TFA); mobile phase B, acetonitrile (ACN) with 0.1% TFA; method: 2 CV 100% B for activation, 3 CV 5% B for equilibration, 3 CV5% B after loading for removing TCEP, 3 CV 80% B for elution. The collected elution fractions containing CRYBMIM-Cys were dried in a vacuum evaporator (Genevac EZ-2 Elite) and used immediately for conjugation.

The carrier transferrin was first thiolated with sulfo-LC-SPDP at 250 μM in a molar ratio of 1:10 for 3 h at room temperature. Ultrafiltration was performed using an Amicon 30k filter unit to remove excess unreacted sulfo-LC-SPDP. The achieved Tf-SPDP was then reacted with 4 mol equiv of CRYBMIM-Cys at 45 μM overnight at 4 °C. The achieved Tf-CRYBMIM BMC was purified with IEX to get rid of all unreacted Tf-SPDP. For IEX: column, strong cation exchange spin columns (Thermo Fisher Scientific); buffer A, 50 mM MOPS, 50 mM NaCl, pH 6.5, 10% glycerol; buffer B, 50 mM MOPS, 1 M NaCl, pH 6.5, 10% glycerol; method, 10 mL buffer A for equilibration, 3 × 10 mL buffer A after loading for removing unreacted Tf-SPDP, 3 × 10 mL buffer B for elution. The collected eluents were further purified by ultrafiltration with 3 × 15 mL reaction buffer using Amicon 30 kDa filters to remove unreacted CRYBMIM-Cys, and the buffer was replaced with storage buffer 20 mM phosphate, 100 mM NaCl, 1 mM EDTA, pH 7.5, with 10% glycerol.

Synthesis of Cas9 Tripartite BMC.

The Cas9 tripartite BMCs were synthesized using bioorthogonal strategies, including the amine-NHS ester reaction, thiol-maleimide reaction, unsymmetrical disulfide bond formation, and DBCO-azide SPAAC click reaction. All reaction steps were carried out in the reaction buffer containing 50 mM phosphate (sodium phosphate monobasic), 300 mM NaCl, and 50 mM L-arginine hydrochloride. The buffer pH was adjusted to 7.0 using NaOH. All reactions were set up in PCR tubes or 1.5 mL Eppendorf tubes and rotated at 600 rpm on an Eppendorf thermomixer at the indicated temperature for incubation. The concentrations of the starting materials, intermediates, and final products were quantified using a NanoDrop 2000 Spectrophotometer. Extinction coefficients are listed in Tables S1 and S2.

For strategies 1 and 2, a 50 μM recombinant Cas9 variant was first reacted with trifunctionalized TAT via thiol-maleimide reaction at a molar ratio of 1:12 to prepare the Cas9-TAT conjugates. After 2.5 h of reaction at room temperature, ultrafiltration was performed using an Amicon ultra centrifugal filter unit with a molecular weight cutoff of 30k (in short as Amicon 30k) to remove excess unreacted TAT. Thirty μM KIT antibody (Ab) and 100 μM apo-transferrin (Tf), respectively, were reacted with Sulfo-LC-SPDP at a molar ratio of 1:100 to prepare the Ab-/Tf-SPDP. After 1 h of reaction at 37 °C, ultrafiltration was performed using an Amicon 30k filter unit to remove excess unreacted Sulfo-LC-SPDP.

For strategy 1, the 50 equiv of linker DBCO-PEG-SH was first ligated with 50 μM Cas9-TAT conjugates via SPAAC click reaction. After 3 h of reaction at 37 °C, ultrafiltration was performed using an Amicon 30k filter unit to remove excess unreacted DBCO-PEG-SH linker. The achieved Cas9-TAT-SH was further reacted at 30 μM with 1.4 equiv of Ab-/Tf-SPDP to form the final Cas9-TAT-Ab or Cas9-TAT-Tf tripartite BMC for 2 h at room temperature.

For strategy 2, 50 equiv of linker DBCO-PEG-SH was first ligated with 100 μM Ab-/Tf-SPDP via asymmetrical disulfide bond formation. After 1 h of reaction at 37 °C, ultrafiltration was performed using an Amicon 30k filter unit to remove excess unreacted DBCO-PEG-SH linker. Finally, 100 μM Ab-/Tf-DBCO was further reacted with 50 μM Cas9-TAT at a molar ratio of 1.4:1 via SPAAC click reaction to form the final Cas9-TAT-Ab or Cas9-TAT-Tf tripartite BMC for 3 h of reaction at 37 °C.

For strategy 3, 50 μM recombinant Cas9 variant was first reacted with DBCO-PEG24-NHS ester at a molar ratio of 1:20. Thirty μM Ab and 100 μM Tf, respectively, were reacted with SPDP-PEG36-NHS ester at a molar ratio of 1:100. Both reactions were incubated for 1 h at 37 °C, and ultrafiltration was performed using an Amicon 30k filter unit to remove excess unreacted reagents. 50 μM of the Cas9-PEG-DBCO was then conjugated with the thiolated KIT antibody or transferrin at a ratio of 1:2 to form the Cas9-Ab/Tf dipartite constructs after 1.5 h incubation at 37 °C. Finally, 30 μM of the Cas9-Ab/Tf dipartite was reacted with 12 equiv of azide-functionalized CPP peptides at 4 °C overnight. For CPPs with limited aqueous solubility, such as TAT-Ebola and A5K, peptides were first dissolved in buffer with 5% (w/v) Pluronic F-127, and then mixed with Cas9-Ab/Tf dipartite to confer SPAAC reaction similarly as described above. The Cas9 tripartite BMC was purified with ultrafiltration using an Amicon 30k filter unit to remove excess peptides.

Cell Viability Analysis.

MV-411 cells were resuspended in fresh media and plated at a density of 2 × 10⁵ cells per well with 95 μL media in a 96-well plate. Five μL of the stock solution for the tested samples was added to reach the indicated final concentrations. After 5 days of treatment, cell viability was assessed using the CellTiter-Glo Luminescent Viability assay, according to the manufacturer’s instructions (Promega). Luminescence was recorded using an Infinite M1000Pro plate reader with an integration time of 250 ms (Tecan).

Intracellular Delivery of Cas9 Tripartite BMC.

The sgRNA stock was mixed with electroporation enhancer, 250,000 Kasumi-1 TLR cells, and resuspension buffer R to a final volume of 14.5 μL. The amount of sgRNA corresponds to 2.0 equiv of the Cas9 tripartite that would be added later. The final concentration of electroporation enhancer is 1.8 μM. The Neon tube was filled with 3 mL of electrolytic buffer E. Ten μL of the sgRNA/cell mixture was taken into the Neon pipet tip and inserted into the Neon tube. The electroporation parameters were set as 1600 V, 10 ms, and 2 pulses. The electroporated cells were resuspended in antibiotic-free RPMI media and incubated at 37 °C for 3 h. The cells were then resuspended in fresh complete RPMI media containing 1% penicillin-streptomycin and spiked with a corresponding amount of Cas9 tripartite, to a final volume of 20 μL. The tripartite/cell mixture was incubated at 37 °C for 30 min, and the cells were spun down and resuspended in fresh complete RPMI media. The cells were further incubated at 37 °C for 3 days before flow cytometry.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.bioconjchem.5c00440.

Materials and methods and supporting Figures S1–S17 and Tables S1–S6 (PDF)

Observation of a time-dependent increase in nuclear accumulation of Cas9 using time-lapse imaging, culminating in successful gene editing to induce fluorescent mCherry expression (AVI)