Lung-specific supramolecular nanoparticles for efficient delivery of therapeutic proteins and genome editing nucleases

Significance

Protein therapeutics hold immense promise for treating infectious diseases and genetic disorders. However, their efficacy is substantially limited by the challenge of delivering proteins to specific tissues, as most nanoparticles designed for this purpose tend to accumulate in the liver. Herein, we report the design of lung-specific supramolecular nanoparticles (LSNPs) that enable the delivery of a diverse range of proteins and CRISPR-based genome-editing nucleases directly to the lungs after systematic administration. The lung-targeting efficacy of LSNPs was rigorously validated in different animal species, demonstrating great potential for tissue-specific genome editing and acute bacterial pneumonia treatment. LSNPs represent a significant advancement in developing tissue-specific protein delivery approaches for advancing protein therapies.

Abstract

Protein therapeutics play a critical role in treating a large variety of diseases, ranging from infections to genetic disorders. However, their delivery to target tissues beyond the liver, such as the lungs, remains a great challenge. Here, we report a universally applicable strategy for lung-targeted protein delivery by engineering Lung-Specific Supramolecular Nanoparticles (LSNPs). These nanoparticles are designed through the hierarchical self-assembly of metal-organic polyhedra (MOP), featuring a customized surface chemistry that enables protein encapsulation and specific lung affinity after intravenous administration. Our design of LSNPs not only addresses the hurdles of cell membrane impermeability of protein and nonspecific tissue distribution of protein delivery, but also shows exceptional versatility in delivering various proteins, including those vital for anti-inflammatory and CRISPR-based genome editing to the lung, and across multiple animal species, including mice, rabbits, and dogs. Notably, the delivery of antimicrobial proteins using LSNPs effectively alleviates acute bacterial pneumonia, demonstrating a significant therapeutic potential. Our strategy not only surmounts the obstacles of tissue-specific protein delivery but also paves the way for targeted treatments in genetic disorders and combating antibiotic resistance, offering a versatile solution for precision protein therapy.

Proteins play pivotal roles in a wide range of cellular functions, from catalyzing biochemical reactions to facilitating communications between cells (1, 2). Their dysregulation is implicated in a variety of pathologies, including malignancies and genetic disorders (3, 4), underscoring the therapeutic potential of directly modulating cell signaling and communications using proteins (5, 6). Despite this promise, the therapeutic application of proteins is often hindered by their delicate structure, sensitive activity, and cell membrane impermeability, presenting significant delivery challenges (7–9). Traditional methods such as genetic introduction of cell-penetrating domains have been used to deliver protein into cells (10), yet their in vivo effectiveness is restricted by low stability and poor tissue specificity. Recent advancements in nanoparticle-based drug delivery, such as lipid nanoparticles (LNPs) and polymers, have shown promise for protein delivery, while they face challenges in tissue-specific delivery, particularly for extrahepatic tissues like the lung (11, 12). Effective pulmonary protein delivery is crucial for treating respiratory diseases and cancers, yet current methods frequently suffer from low delivery efficiency and lung specificity (13–16). Recently, nebulized LNPs and the systemic administration of selective organ-targeting (SORT) LNPs have marked significant success in lung-targeted delivery, particularly for delivering mRNA-based therapeutics (17, 18). However, the translation of these successes from tissue-specific delivery of mRNA to protein faces unique challenges due to the intrinsic differences between protein and nucleic acid. For instance, proteins are more susceptible and instable toward the stresses encountered during nebulization than mRNA (19). While SORT technology offers a promising approach for delivering mRNA specifically to the lungs, its effectiveness in delivering proteins remains largely unexplored and presents considerable challenges. Therefore, novel strategies for lung-specific protein delivery are crucial not only for deepening our understanding of protein function in complex biological settings but also for improving therapeutic interventions.

In this study, we report a universally applicable strategy for lung-specific protein delivery through the design of Lung-Specific Supramolecular Nanoparticles (LSNPs), derived from the hierarchical self-assembly of metal-organic polyhedra (MOP). By utilizing positively charged and adamantane (Ada)-functionalized MOP (20, 21), and controlling their self-assembly with β-cyclodextrin (β-CD)-conjugated polyethyleneimine (PEI) through the host–guest interactions between Ada and β-CD, we have designed LSNPs capable of encapsulating and delivering proteins with varied physiochemical properties specifically to the lungs (Fig. 1). We elucidate how the surface chemistry of LSNPs and the formation of a protein corona abundant in vitronectin (Vtn) determines the lung-specificity of protein delivery. Remarkably, our LSNPs have successfully facilitated the delivery of genome-editing nucleases, such as CRISPR/Cas13d ribonucleoproteins (RNPs) for precise gene regulation within lungs. Additionally, we demonstrate the efficacy of LSNPs in delivering catalase to mitigate lung inflammation and antimicrobial peptides, like the crucian carp c-type lysozyme peptide (CCL), to alleviate acute bacterial infections caused by drug-resistant Pseudomonas aeruginosa strains across various animal species. Our results indicate that lung-specific protein delivery via LSNPs significantly decreases bacterial load and enhances animal survival rates, offering an excellent alternative and improvement to traditional antimicrobial therapy. Notably, our approach for lung-specific protein delivery shows effectiveness across various species, including mice, rabbits, and dogs. The implications of our strategy extend beyond merely addressing the challenges of tissue-specific protein delivery, it further allows for targeted interventions across a spectrum of diseases while minimizing systemic side effects.

Fig. 1.

Self-assembly of LSNPs for lung-specific protein delivery. (A) Schematic illustration of the assembly of metal-organic polyhedra (MOP) into lung-specific supramolecular nanoparticles (LSNPs) for protein delivery. (B) Lung-specific protein delivery for acute bacterial pneumonia treatment and genome editing.

Results

The Design and Self-Assembly of LSNPs For Protein Delivery.

Our synthesis of LSNPs begins with the coordination-driven self-assembly of adamantane-functionalized M₂L₄-type MOP (Ada-MOP), utilizing platinum nitrate and a bidentate dipyridine ligand (Ada-Dpy) derived from tetraphenylethylene (TPE) (Fig. 1 and SI Appendix, Fig. S1). The self-assembly of positively charged Ada-MOP into LSNPs not only encapsulates protein through electrostatic interactions, but also enhances lung specificity for in vivo delivery by tuning the surface chemistry of LSNPs. Meanwhile, the inclusion of Ada-conjugated polyethylene glycol (Ada-PEG) greatly enhances the stability of LSNPs in biological settings. In addition, the introduction of β-cyclodextrin-modified polyethyleneimine (β-CD-PEI) not only facilitates the cross-linking of Ada-MOP and the formation of LSNPs by making use the host–guest interactions between β-CD and Ada, but also promotes endosomal escape of LSNPs through the proton-sponge effect of ionizable PEI moiety (22), ensuring effective cytosolic protein release. To self-assemble LSNPs, we introduced a modular strategy that combines Ada-MOP, Ada-PEG, and β-CD-PEI in a weight ratio of 1:10:10 to form well-dispersed nanoparticles. The structural integrity and the essential role of β-CD and Ada interactions for LSNPs formation were first verified through ¹H NMR spectroscopy. The downfield shifts in the adamantyl proton signals from Ada-MOP and Ada-PEG in the ¹H NMR spectrum of LSNPs suggest their incorporation within the β-CD cavities (Fig. 2A). Moreover, the presence of the [M-3NO₃]³⁺ peak in the ESI-MS spectrum of LSNPs that ascribed to intact Ada-MOP confirms the structural integrity of Ada-MOP during the self-assembly of LSNPs (Fig. 2B). In addition, fluorescence analysis of Ada-MOP remained in the supernatant of the self-assembly mixture of LSNPs indicated that more than 90% of Ada-MOP was assembled into LSNPs (SI Appendix, Fig. S2). To study the potency of LSNPs for protein encapsulation, bovine serum albumin (BSA), was utilized as a model protein. It was found that BSA was encapsulated within LSNPs in an efficiency greater than 90% when it was coassembled with the LSNPs components (SI Appendix, Fig. S3). The size of protein-encapsulated LSNPs, characterized using dynamic light scattering (DLS) and transmission electron microscopy (TEM), exhibited an average diameter of approximately 100 nm (Fig. 2C and SI Appendix, Fig. S4). Additionally, we found that incorporating Ada-PEG into LSNPs significantly enhances the stability of BSA-loaded LSNPs in biological settings, a crucial factor for in vivo protein delivery. LSNPs/BSA nanoparticles exhibited remarkable stability over 24 h, even in the presence of serum proteins, while those without Ada-PEG incorporation showed greatly decreased stability (SI Appendix, Fig. S4).

Fig. 2.

Characterization of LSNPs and their intracellular protein delivery study. (A) ¹H NMR spectrum (300 MHz, 298 K) indicates the host–guest interaction of Ada-PEG and Ada-MOP with β-CD-PEI in LSNPs. (B) Experimental (red) and calculated (blue) ESI-TOF-MS spectra of the [M-3NO₃]³⁺ peak corresponding to Ada-MOP in LSNPs. (C) Transmission electron microscopy (TEM) image of LSNPs/BSA nanoparticles. (D) CLSM images of MDA-MB-231 cells treated with LSNPs/BSA-FITC nanoparticles containing 50 nM BSA-FITC for 6 h (Scale bar: 20 μm.) (E) Comparison of cellular uptake efficiency between free BSA-FITC and LSNPs/BSA-FITC by MDA-MB-231 cells. Data are presented as mean ± SD (n = 3). (F) Intracellular β-Gal activity assay in MDA-MB-231 cells treated with free β-Gal and LSNPs/β-Gal nanoparticles containing 50 nM β-Gal (Scale bar: 50 μm.)

The biocompatibility of LSNPs was first evaluated by treating MDA-MB-231 human breast cancer cells with BSA-loaded LSNPs, where high cell viability was observed across a broad range of nanoparticle concentrations (SI Appendix, Fig. S5). Moreover, the hemolytic activity of LSNPs was studied by exposing red blood cells to different concentrations of LSNPs at 37 °C for 1 h. The observation of less than 1% hemolysis after LSNP treatments indicates their biocompatibility for protein delivery both in cellulo and in vivo (SI Appendix, Fig. S6). The protein delivery capability of LSNPs was assessed in cultured cells by delivering fluorescein-labeled BSA (FITC-BSA) into MDA-MB-231 cells. Confocal laser scanning microscopy (CLSM) imaging revealed significant intracellular fluorescence (Fig. 2D), indicating a successful delivery at a protein concentration of 50 nM, starkly contrasting with the minimal fluorescence observed in cells treated with free FITC-BSA (SI Appendix, Fig. S7). Flow cytometry further corroborated these results, showing a substantial proportion of fluorescence-positive cells (over 90%) at even the lowest protein concentration of 30 nM (Fig. 2E). Moreover, the potential of LSNPs to deliver functional enzymes was studied in MDA-MB-231 and HeLa cells by treating the cells with LSNPs encapsulating β-galactosidase (β-Gal). We found a significant intracellular enzymatic activity upon the delivery of β-Gal (50 nM) using LSNPs, as evidenced by the hydrolysis of the chromogenic substrate X-Gal that generates distinct blue punctuates within the cells (Fig. 2F and SI Appendix, Fig. S8).

Universal Lung-Specific Protein Delivery Using LSNPs.

We next evaluated the in vivo protein delivery efficacy of LSNPs by intravenously injecting Cy5-labeled BSA in the format of LSNPs to BALB/c mice at a protein dosage of 1 mg kg⁻¹. The LSNPs were assembled at various β-CD-PEI to Ada-MOP ratios along with Cy5-BSA, allowing us to study the influence of assembly ratios of LSNPs on the efficiency and tissue specificity of protein delivery. Six-hour postinjection, the major organs of mice were harvested for ex vivo fluorescence imaging. As shown in Fig. 3A, the β-CD-PEI to Ada-MOP ratio indeed greatly affected both the efficiency and tissue specificity of protein delivery using LSNPs. A noteworthy observation was the redirection of the fluorescent signal from the liver to the lungs as the ratio of β-CD-PEI to Ada-MOP increased, peaking at 10:1, indicating the tissue-selective delivery potential of LSNPs. In contrast, the complexation of Cy5-labeled BSA with either β-CD-PEI or β-CD-PEI/Ada-PEG under the same condition did not show significant lung-targeting effect after intravenous injections (SI Appendix, Fig. S9). Furthermore, we used inductively coupled plasma mass spectrometry (ICP-MS) to quantify the distribution of platinum in various organs, and its clearance in the lung after systematic injections. Our results confirmed the lung-specificity of LSNPs by showing a pronounced enrichment of platinum in the lung, compared to other organs (SI Appendix, Fig. S10A). In addition, the highest platinum level in the lungs was observed 24 h after the administration, it was decreased down to a significantly low-level 144 h postinjections (SI Appendix, Fig. S10B). To further study the in vivo circulation of LSNPs, we evaluated the pharmacokinetics following intravenous injection of LSNP and analyzed the change of platinum level in plasma using ICP-MS at different time points after systematic administration. The circulation half-life of LSNPs was measured to be approximately 6.6 h (SI Appendix, Fig. S10C), suggesting its clearance and therefore biocompatibility for in vivo protein delivery. To assess the versatility of LSNPs in delivering proteins with different physicochemical properties to the lungs, four proteins with different molecular weight, including alcohol dehydrogenase (ADH, 150 kDa), DUF5 (55 kDa), catalase (CAT, 250 kDa), and β-Galactosidase (β-Gal, 430 kDa) were fluorescently labeled with Cy5 and encapsulated into LSNPs for in vivo studies. The intravenous administration of LSNPs/Cy5-labeled protein indeed resulted in significant lung-targeting effect, which is not observed when free proteins were administrated under the same condition (Fig. 3B and SI Appendix, Fig. S11).

Fig. 3.

Lung-specific protein delivery using LSNPs. (A) Lung-specific protein delivery using LSNPs assembled at different β-CD-PEI to Ada-MOP ratios nanoparticles. The mice were intravenously injected with Cy5-labeled BSA at a dosage of 1.0 mg kg⁻¹ encapsulated in LSNPs. The mice tissues were harvested for ex vivo fluorescence imaging 6 h after injections. (B) Lung-specific delivery of different proteins, including ADH (alcohol dehydrogenase), DUF5, CAT (catalase), and β-Gal (β-galactosidase). (C) Schematic illustration of the analysis of β-Gal activity in mouse tissues. After the administration of LSNPs/β-Gal nanoparticles (3 mg kg⁻¹ β-Gal-Cy5), the major mouse organs, including the heart, liver, spleen, lung, and kidney were collected, dissected, lysed, and subsequently analyzed for β-Gal activity. (D) Quantitative assessment of β-Gal activity in different tissue lysates, confirming lung-specific β-Gal delivery facilitated by LSNPs. (n = 3 biologically independent mice). (E) β-Gal activity assay of the lysates of different organs of dogs injected with LSNPs/β-Gal (3 mg kg⁻¹ β-Gal) nanoparticles. (n = 2). Unless otherwise stated, data are presented as mean ± SD.

The functionality of lung-specific protein delivery using LSNPs was evaluated using β-Gal as a model protein. After the administration of LSNPs/β-Gal nanoparticles, we observed significant β-Gal activity within the mouse lungs, as indicated by the hydrolysis of X-Gal in the presence of the lysates of lung tissues (Fig. 3 C and D) (23). The measured β-Gal activity in the lung exhibited about 87% of the total enzymatic activity measured in all tissues, nearly eightfold higher than that in the liver, highlighting the capability and high specificity of LSNPs for lung-targeted delivery of protein in its active form. Notably, the lung-targeting potency of LSNPs was general and highly effective across different animal species. Following the intravenous administration of LSNP/β-Gal-Cy5 to beagle dogs, we observed a significant accumulation of the protein in the lung of dogs, while minimal β-Gal-Cy5 fluorescence was observed in other organs, such as the liver (SI Appendix, Fig. S12). Additionally, an assessment of β-Gal activity across different tissues of the dog indicated significant capability of the lung lysates to hydrolyze X-Gal compared to other tissues, further validating the lung-specific delivery of functional protein using LSNP in larger animal models (Fig. 3E). Together, these findings underscore the versatility and effectiveness of LSNPs in delivering proteins specifically to the lungs across different animal species, and remaining the biological activity of the delivered proteins within specific tissues.

Mechanistic Study of Lung-Specific Protein Delivery Using LSNPs.

To decipher the underlying mechanisms contributing to the lung-specificity of LSNPs-mediated protein delivery, we performed a comprehensive analysis of various factors influencing the tissue tropism of nanoparticles. These included an examination of nanoparticles surface charge, size, and protein corona formed in vivo. We first compared the physicochemical properties of nanoparticles showing liver (assembled at β-CD-PEI to Ada-MOP of 6:1) and lung (assembled at β-CD-PEI to Ada-MOP of 10:1)-targeting effect. As shown in SI Appendix, Fig. S13A (SI Appendix), both types of nanoparticles exhibited an average diameter of around 110 nm. However, there was a significant difference exists between the nanoparticle surface charge: lung-targeted LSNPs exhibited a higher positive zeta potential (19.8 mV) compared to liver-targeted LSNPs (7.81 mV). This disparity in charge led us to hypothesize that it significantly influences plasma protein adsorption on nanoparticle surface in circulation, thereby dictating tissue specificity. To verify this hypothesis, we performed proteomic profiling of the surface-absorbed protein of LSNPs incubated with murine serum. After isolating surface-adsorbed proteins, mass spectrometry analysis revealed distinct protein corona composition for these nanoparticles. Lung-targeted nanoparticles harbored 375 proteins, while liver-targeted ones contained 433 proteins, with 277 proteins common to both (SI Appendix, Fig. S13B). A Venn diagram illustrated the protein distribution (SI Appendix, Fig. S13B), indicating 98 unique proteins to lung-targeted nanoparticles. Among the top twenty most abundant proteins, only nine were shared between the two types of nanoparticles. Notably, liver-targeted nanoparticles predominantly exhibited apolipoprotein A (ApoA), aligning with prior reports of the role of ApoA in hepatic-targeting delivery (SI Appendix, Fig. S13C) (24, 25). In contrast, lung-targeted LSNPs were enriched with vitronectin (Vtn), constituting about 13% of their corona, a finding of particular significance given the known interaction of Vtn with α_vβ₃ integrin on lung endothelial cells (Fig. 4A) (26). Indeed, flow cytometry analysis on the cells isolated from the lung of mice received LSNPs/Cy5-labeled BSA delivery indicated that approximately 80% of lung endothelial cells are fluorescence positive, suggesting the determinative role of Vtn and α_vβ₃–integrin interaction on enabling the lung-targeting of LSNPs (Fig. 4B).

Fig. 4.

Mechanistic study of lung-specific protein delivery of LSNPs. (A) Protein corona analysis of LSNPs/BSA-Cy5 NPs. Quantification of percentage of total proteins of the top five protein components in the protein corona of the lung-targeted nanoparticles is shown. (B) Flow cytometry quantification of Cy5-positive cells in lung tissues. Mice were injected with LSNPs/BSA-Cy5 nanoparticles at a single dose of 1 mg kg⁻¹ BSA-Cy5. Endothelial cells were stained by fluorescein isothiocyanate (FITC) labeled-CD31 antibody, immune cells were stained by FITC-CD45 antibody, and epithelial cells were stained by FITC-CD326 antibody. (C) Schematic illustration of vitronectin (Vtn)-mediated intracellular uptake of LSNPs. (D) Representative CLSM images of HepG2, U87MG, and HFL-1 cells treated with LSNPs/BSA-FITC nanoparticles containing 2.5 nM BSA-FITC for 6 h. To verify the essential role of Vtn in receptor-medicated uptake of nanoparticles, the nanoparticles were preincubated with 500 ng mL⁻¹ Vtn before adding to cells (Scale bar: 20 µm.) (E) Comparison of the cellular uptake efficiency of LSNPs/BSA-FITC nanoparticles preincubated with different concentrations of Vtn by HepG2, U87MG, and HFL-1 cells. Unless otherwise stated, data are presented as mean ± SD. (n = 3 biologically independent experiments).

Following the identification of Vtn at notably high levels in the protein corona of lung-targeting LSNPs, we conducted further experiments to corroborate its pivotal role in enhancing lung specificity through α_vβ₃ integrin-mediated cellular uptake of LSNPs (Fig. 4C). To this end, LSNPs were preincubated with varying concentrations of Vtn prior to their exposure to cells exhibiting different levels of α_vβ₃ integrin expression. We found that the cellular uptake efficiency of LSNPs/BSA-FITC nanoparticles (containing 2.5 nM protein) by α_vβ₃ integrin-positive U87MG cells and lung-derived HFL-1 cells was significantly higher than that by α_vβ₃ integrin-negative HepG2 cells (Fig. 4 D and E). In addition, the cellular uptake efficiency of LSNPs by α_vβ₃ integrin-positive cells increased proportionally with the concentration of Vtn precoated with nanoparticles, confirming the crucial role of Vtn and α_vβ₃ integrin interaction in determining the lung-specificity of LSNPs.

Lung-Specific Genome Editing Via LSNPs-Mediated Protein Delivery.

A critical challenge in advancing therapeutic genome editing is developing delivery strategies that enable precise targeting of specific tissues, such as the lung. Having demonstrated the high efficacy and tissue specificity of LSNPs to deliver protein to the lungs, we expanded our investigation to targeted genome editing by selectively delivering programmable nucleases to specific tissues. For this purpose, we first evaluated its potency for genome editing in cultured cells. HEK293 cells stably expressing green fluorescent protein (GFP) were treated with LSNPs encapsulating recombinant Cas9 protein and single-guide RNA (sgRNA) targeting GFP. Our results indicated that the delivery of Cas9/sgGFP ribonucleoproteins (RNPs) (40 μg mL⁻¹ Cas9 and 2.5 μg mL⁻¹ sgGFP) using LSNPs effectively knocked down the expression of GFP, increasing the percentage of GFP-negative cells up to approximately 70% compared to untreated cells (Fig. 5A). In addition, T7 endonuclease I (T7EI) assay indicated that the INDEL efficiency in HEK293-GFP cells following LSNPs/Cas9/sgGFP treatment was increased up to 38.5%, which was significantly higher than negative controls (Fig. 5B).

Fig. 5.

Lung-specific genome editing using LSNPs. (A) Delivery of LSNPs/Cas9/sgGFP nanoparticles into HEK-GFP cells for genome editing. HEK293-GFP cells were treated with Cas9/sgGFP alone, LSNPs alone, and LSNPs/Cas9/sgGFP nanoparticles (40 μg mL⁻¹ Cas9 and 2.5 μg mL⁻¹ sgGFP). The GFP expression profiles of cells which received different treatment was quantified 48 h post protein delivery. Data are presented as mean ± SD (n = 3). (B) T7EI indel analysis of DNA isolated from HEK293-GFP cells with different treatment as indicated. The black arrows indicate cleavage bands. N.D., nondetectable. (C) Schematic illustration of the delivery of ProTα-Cre to activate tdTomato expression in Ai9 mice by genetically deleting the stop cassette. (D) Lung-specific delivery of LSNPs/ProTα-Cre selectively induces tdTomato expression in the mouse lung. Ex vivo fluorescent images of the major organs of Ai9 mice intravenously injected with ProTα-Cre alone or LSNPs/ProTα-Cre nanoparticles at a protein dosage of 1 mg kg⁻¹ (n = 3 biologically independent mice). (E) Schematic depiction of selective knockdown of Ctsl in mouse lungs through LSNPs-mediated Cas13d RNPs delivery. (F) Ctsl levels in the major organs of mice treated with LSNP/Cas13d/gCtsl. Mice were injected intravenously with PBS, free Cas13d/gCtsl, or LSNPs/Cas13d/gCtsl nanoparticles at a protein dosage of 1 mg kg⁻¹. After 72 h, the major organs were collected and analyzed for Ctsl expression by western blotting. The levels of Ctsl were normalized to GAPDH. Data are presented as mean ± SD (n = 3 biologically independent mice), statistical significance was determined by unpaired two-tailed Student t-tests. **P < 0.01.

We next studied whether LSNPs-enabled protein delivery was effective for lung-specific genome editing. To this end, we first studied the delivery of Cre recombinase to lungs for gene regulation using LSNP. We engineered a protein construct, ProTα-Cre, by fusing Cre recombinase with a negatively charged human prothymosin alpha (ProTα), to enhance the encapsulation of Cre within LSNPs (27). The effectiveness of LSNPs/ProTα-Cre for tissue-specific gene recombination was evaluated using a Cre/LoxP reporter featuring Ai9 mice, where a LoxP-flanked stop cassette blocks the expression of the red fluorescent protein tdTomato (Fig. 5C) (28). Our studies revealed that LSNPs-mediated delivery of Cre recombinase successfully excised the stop cassette, activating tdTomato expression exclusively in the lung, as revealed by the ex vivo fluorescence imaging of mouse tissues following LSNPs/ProTα-Cre delivery (Fig. 5D). This lung-targeted gene recombination and tdTomato expression were also confirmed through the fluorescence imaging of lung sections, which shows significantly high percentage of tdTomato-positive cells (SI Appendix, Fig. S14). However, an effective gene recombination in the lung was not observed for mice received the injections of free ProTα-Cre, underscoring the necessity and high efficacy of LSNP for lung targeting.

To further explore the potential of LSNPs for lung-specific genome editing, we studied the delivery of CRISPR/Cas13d RNPs to knock down specific genes in the lungs using LSNPs. We selected protease cathepsin L (Ctsl, Fig. 5E), a potential target for treating respiratory diseases and preventing SARS-CoV-2 infections, as our model gene (29). We first studied whether LSNPs could effectively deliver Cas13d RNPs to the lung. To this end, recombinant Cas13d protein was encapsulated into LSNPs along with Cy5-labeled RNA, and administered it to mice intravenously. Indeed, we observed significant lung-specific accumulation of Cas13d RNPs following intravenous injections, a stark contrast to the ineffective lung-targeting of free Cas13d RNPs administration (SI Appendix, Fig. S15). The lung-specific delivery of Cas13d RNPs targeting Ctsl resulted in significant RNA knockdown in the lung of mice compared to the nontreated controls (SI Appendix, Fig. S16A). Detailed analysis of Ctsl expression in different tissues further indicated an efficient and selective knockdown of Ctsl in the lungs (Fig. 5F), while no efficient Ctsl editing was observed in other tissues (SI Appendix, Fig. S16B), demonstrating the great potency of LSNPs for tissue-specific genome editing and targeted gene therapy.

Alleviating Lung Inflammation Via Targeted Catalase Delivery.

In our exploration of the therapeutic potential of lung-specific protein delivery using LSNPs, we next used a murine model with lung inflammation and studied whether the delivery of protein therapeutics could be effective for alleviating inflammation. The disease model was established via intraperitoneal injection of lipopolysaccharide (LPS), which induces a significant increase in reactive oxygen species (ROS) level in the mouse lung, thereby mimicking inflammatory lung conditions (30). Given the critical role of ROS in mediating lung inflammation, we hypothesized that the lung-specific delivery of catalase (CAT), an enzyme known for ROS-scavenging properties using LSNPs may be anti-inflammatory effective. To verify this potential, the LSNPs/CAT nanoparticles were injected intravenously to the mice with LPS-induced lung inflammation. Indeed, LSNPs/CAT administration led to a notable decrease in ROS levels within the lung, as evidenced by fluorescence imaging of the ROS level using a previously reported method (Fig. 6A) (31). LPS-treated mice exhibited an approximate 1.8-fold increase in ROS, as indicated by enhanced fluorescence intensity in the lung. In contrast, this enhancement was effectively alleviated in mice administered with LSNPs/CAT nanoparticles, showing ROS levels comparable to that of negative control mice without LPS pretreatment. This result was not observed in LPS-challenged mice administrated with free CAT, likely due to its inefficient targeting and accumulation within the lung. Furthermore, LSNPs/CAT treatment significantly reduced the levels of proinflammatory cytokines, including interleukin-6 (IL-6) and interleukin-1β (IL-1β), as shown in Fig. 6 B and C. The reduction in cytokine release corroborates the anti-inflammatory capabilities of LSNPs/CAT nanoparticles. Histological examination indicated that lung sections from LPS-challenged mice showed extensive leukocyte infiltration, indicating of inflammation, while those from LSNPs/CAT-treated mice exhibited minimal infiltration, highlighting their therapeutic promise as anti-inflammatory agents (Fig. 6D).

Fig. 6.

Effective alleviation of LPS-induced lung inflammation by LSNPs/CAT delivery. (A) Comparative fluorescence intensity in the lung of mice received different treatments, as indicated by the change of ROS levels post LPS-challenge. The LPS-challenged mice were injected with different protein formulations as indicated before ROS level imaging using a ROS probe. (B and C) Quantification of IL-6 and IL-1β levels in the lung after various treatments, highlighting the anti-inflammatory effect of LSNPs/CAT delivery. (D) Hematoxylin and eosin-stained lung sections of mice treated with PBS, free CAT, and LSNPs/CAT nanoparticles at a protein dosage of 1 mg kg⁻¹ before LPS challenging. Data are presented as mean ± SD (n = 3 biologically independent mice per group), statistical significance was determined by unpaired two-tailed Student t-tests (A–C). *P < 0.05, **P < 0.01, ***P < 0.001.

Therapeutic Application of LSNPs for Treating Acute Bacterial Pneumonia.

Acute bacterial pneumonia is a severe respiratory infection predominantly caused by bacterial pathogens and characterized by severe lung inflammation. It poses a major clinical challenge, particularly due to the infection caused by multidrug-resistant bacterial strains. Having demonstrated the high efficacy of LSNPs in lung-specific delivery of proteins and genome-editing nucleases, we extended our study to explore their therapeutic application in the treatment of acute bacterial pneumonia. To this end, we designed a protein construct by fusing crucian carp c-type lysozyme peptide (CCL), known for its potent antimicrobial properties with ProTα to enhance its encapsulation and lung-specific delivery using LSNPs (32). The antimicrobial efficacy of ProTα-CCL was first validated through its capability to prohibit the growth of multidrug-resistant P. aeruginosa (PA) (Fig. 7A) (33, 34). Subsequent intravenous administration of LSNPs/ProTα-CCL-Cy5 nanoparticles in a murine model showed significant accumulation of fluorescently labeled protein in the lungs (Fig. 7B). In vivo therapeutic efficacy was further evaluated in a murine model of acute bacterial pneumonia induced by PA infection. PA-infected mice were treated with phosphate-buffered saline (PBS), free ProTα-CCL, LSNPs, or LSNPs loaded with ProTα-CCL intravenously at a protein dosage of 4 mg kg⁻¹ (Fig. 7C). Notably, the LSNPs/ProTα-CCL treatment resulted in an impressive 85% reduction in lung bacterial load, quantified at 1.4 × 10⁷ CFU per gram tissue, significantly surpassing the efficacy of other treatments, including negative controls (Fig. 7 D and E). Remarkably, mice treated with LSNPs/ProTα-CCL exhibited a 90% survival rate over seven days, significantly outperforming negative controls and other treatments, including ciprofloxacin, a commonly used small molecule-based antibiotic for treating infections (Fig. 7F). Histological analysis further demonstrated that the LSNPs/ProTα-CCL treatment exhibited markedly reduced lung damage compared to control groups, therefore reinforcing its efficacy in addressing antibiotic resistance (SI Appendix, Fig. S17).

Fig. 7.

LSNPs/ProTα-CCL delivery for the treatment of acute bacterial pneumonia. (A) Antimicrobial activity assay of ProTα-CCL (0.5 mg mL⁻¹) against P. aeruginosa. (B) Ex vivo fluorescence imaging of the major organs of mice intravenously injected with PBS, free ProTα-CCL, and LSNPs/ProTα-CCL nanoparticles at a protein dosage of 4 mg kg⁻¹ (n = 3 biologically independent mice per group). (C) Schematic illustration of the antibacterial study of LSNPs/ProTα-CCL nanoparticles. Mice were infected intratracheally with P. aeruginosa and then injected intravenously with LSNPs/ ProTα-CCL nanoparticles. (D) Comparison of bacterial loads in the lungs of mice received different treatments as indicated. (E) Colony-forming units for P. aeruginosa found in the lungs of mice after different treatments at 24 h postinfection (n = 4 biologically independent mice per group). (F) Survival rate of mice infected with P. aeruginosa and treated with free ProTα-CCL, ciprofloxacin, and LSNPs/ProTα-CCL at a protein dosage of 4 mg kg⁻¹ (n = 10 biologically independent mice per group). (G) Schematic illustration of the study of lung-specific protein delivery in a rabbit model using LSNPs. Rabbits were administered intravenously with LSNPs/ProTα-CCL-Cy5 nanoparticles. 10 h after the injections, the rabbits were killed, and the major organs were excised for fluorescent imaging. (H) Fluorescence images of major organs of rabbits intravenously injected with PBS, free ProTα-CCL, and LSNPs/ProTαCCL nanoparticles at a protein dose of 6 mg kg⁻¹ (n = 3 biologically independent rabbits per group). (I) CFU for P. aeruginosa found in the lungs of rabbits after different treatments at 24 h postinfection. Data are presented as mean ± SD (n = 5 biologically independent rabbits per group). Statistical significance was determined by unpaired two-tailed Student t-tests. ****P < 0.0001.

Furthermore, we assessed the biocompatibility of LSNPs/ProTα-CCL for treating acute bacterial pneumonia. Systematic blood chemistry analysis revealed no significant alterations in key liver function markers, including aspartate transaminase (AST), alanine transaminase (ALT), and bilirubin levels (SI Appendix, Fig. S18). Histological evaluations of lung tissues posttreatment showed preservation of normal tissue morphology, highlighting the biocompatibility and the suitability of LSNPs/ProTα-CCL for acute bacterial pneumonia treatment without adverse systemic effects (SI Appendix, Fig. S19). Additionally, to evaluate the potential inflammatory responses induced by LSNPs administration, we analyzed the change of various cytokines, including tumor necrosis factor alpha (TNF-α) and interleukin (IL)-1β in the serum of mice postinjection of LSNPs. There were no significant differences in cytokine levels between the LSNP-treated group and the control groups, indicating again the biocompatibility of LSNPs (SI Appendix, Fig. S20).

In light of the promising antimicrobial activity and therapeutic efficacy of LSNPs/ProTα-CCL in mouse models, we extended our investigation to a rabbit model to assess the generality and therapeutic potential of LSNPs across different animal species. To this end, rabbits were first intravenously injected with PBS, fluorescently labeled ProTα-CCL-Cy5 with or without LSNPs encapsulation (Fig. 7G). Ex vivo fluorescence imaging of the major organs, performed ten hours postadministration, showed significant accumulation of the protein in the lungs of rabbits (Fig. 7H), confirming the lung-specific delivery of antimicrobial protein using LSNPs. Furthermore, in a rabbit model of PA infection, the intravenous administration of LSNPs/ProTα-CCL at a dosage of 6 mg kg⁻¹ resulted in a dramatic reduction in lung bacterial load, decreasing it by three orders of magnitude to 3.7 × 10⁵ CFU per gram of lung tissue (Fig. 7I), showing significantly higher in vivo antibacterial efficacy compared to other treatments.

Discussion

The necessity for precision in delivering therapeutic proteins to the lung is paramount for treating a broad spectrum of diseases, including chronic obstructive pulmonary disease (COPD), pulmonary infections, and cancers (35–37). Despite the potential benefits, current methods for lung-specific protein delivery faces substantial challenges. Typically, the systemic administration of proteins in their native form leads to rapid systematic clearance and enzymatic degradation in the bloodstream. In addition, the inherent difficulty of proteins to cross cell membranes affects their ability to target intracellular mechanisms, such as those used for genome editing (38). Recent advancements have shown that surface-engineered nanoparticles can facilitate protein delivery, yet these often suffer from rapid clearance and nonspecific tissue accumulation, especially in the liver, due to the recognition and clearance by the mononuclear phagocyte system (MPS) following systemic administration (39). More recently, selective organ targeting (SORT) strategy using LNPs has shown promise in delivering mRNA and genome-editing ribonucleoproteins to the lungs through systemic administration (18). However, its effectiveness for lung-specific protein delivery and its applicability across large animal species remain underexplored. Moreover, while nebulization and inhalation of LNPs have been effective for lung-targeted mRNA delivery, these methods can negatively impact the stability and functionality of proteins, due to the unfolding and aggregation at the air–liquid interface during nebulization (19). The inhalation of peptide-encapsulated nanoparticles has also been observed to accumulate in the heart (40). Given these complexities, there is a great need for innovative strategies that are specially designed for lung-specific protein delivery. Such strategies must ensure not only precise organ specificity but also the preservation of the integrity and activity of proteins, which is essential for fully understanding protein functionality and advancing protein-based therapies.

Proteins usually show varied physicochemical properties, such as molecular weight, surface charge, and hydrophobicity, which collectively challenge an effective and universal tissue-specific protein delivery. In this context, our study introduces an innovative and versatile strategy for lung-specific delivery of therapeutic proteins through the design of LSNPs. The unique surface properties of LSNPs, characterized by their positively charged and dynamic protein corona, are crucial for their lung-targeting ability. Proteomic analysis of the protein corona of LSNPs has revealed a significant enrichment of vitronectin, a key factor for achieving lung specificity. This finding not only underscores the importance of protein corona composition in nanoparticle distribution in vivo but also paves the way for the design of advanced targeted protein delivery strategies. We demonstrated the universal applicability of LSNPs to deliver a broad range of proteins to the lung, including catalase to reduce lung inflammation in mice. Additionally, the successful delivery of Cre recombinase and CRISPR genome-editing nucleases enabled lung-targeted gene regulation. Particularly, our strategy of lung-specific delivery of therapeutic ProTα-CCL was notably more effective than small molecular antibiotics in reducing bacterial loads in acute bacterial pneumonia models. It has recently been reported that the efficiency of nanoparticles for drug delivery may vary across different animal models (41). The successful application of LSNPs for lung-targeted protein delivery in large animal models, including rabbits and beagle dogs, further emphasizes its scalability. Our study represents a comprehensive examination of lung-specific protein delivery in large animal models, filling a notable gap in current research and future clinical translation. The strategy and findings offer valuable insights into tissue-specific protein delivery, representing a substantial stride toward the development of effective therapeutic strategies for treating lung diseases, vaccination, and immunotherapy.

Materials and Methods

Self-Assembly of LSNPs for Intracellular Protein Delivery.

LSNPs were self-assembled by stirring an aqueous solution containing β-CD-PEI (0.3 mg), Ada-PEG (0.305 mg), and Ada-MOP (0.03 mg) with or without different amounts of proteins (as shown in SI Appendix, Table S1) for one hour at room temperature. The resulted LSNPs/protein nanoparticles were separated through centrifugation (13,000×g, 15 min), washed using water, and redispersed in PBS at a concentration of 0.1 mg mL⁻¹ (calculated by Ada-MOP concentration) for further studies.

To self-assemble and optimize LSNPs for intracellular protein delivery, β-CD-PEI (0.3 mg) and BSA-Cy5 (0.375 mg) in 0.3 mL of aqueous solution was mixed with different concentration of Ada-MOP and Ada-PEG as show in SI Appendix, Table S1. The mixture was stirred for one hour at room temperature. The resulted LSNPs/BSA-Cy5 nanoparticles were purified through centrifugation. The resulted LSNPs were washed using water and redispersed in PBS at a concentration of 0.1 mg mL⁻¹ Ada-MOP for further studies. To determine the protein encapsulation efficiency of LSNPs, the supernatant of the self-assembly mixture was collected for measuring fluorescence intensity, which was normalized to that of free BSA-Cy5 to calculate protein encapsulation efficiency. Meanwhile, to determine the encapsulation efficiency of Ada-MOP in LSNPs, the fluorescence intensity of free Ada-MOP in the supernatant was measured and quantified.

Lung-Specific Delivery of LSNPs/BSA-Cy5 Nanoparticles.

To study the in vivo distribution of LSNPs/BSA-Cy5 nanoparticles, BALB/c mice were intravenously injected with 200 μL LSNPs/BSA-Cy5 nanoparticles at a dosage of 1 mg kg⁻¹ BSA-Cy5. Meanwhile, β-CD-PEI/BSA-Cy5 and β-CD-PEI/Ada-PEG/BSA-Cy5 (self-assembled from 0.3 mg β-CD-PEI, 0.305 mg Ada-PEG, 0.375 mg BSA-Cy5) were used as controls for studying the essential role of assembling LSNPs for lung-targeted delivery of BSA-Cy5. 6 h after the injections, the mice were killed, the major organs were harvested for fluorescence imaging under an excitation of 640 nm and emission filter of 680 nm. To quantify the platinum level in different tissues, a small portion of mouse organ was digested in 5 mL of nitric acid for analyzing the content of platinum in tissues at different time points after administration using ICP-MS. Meanwhile, mouse serum was collected for measuring cytokine level using enzyme-linked immunosorbent assay kit.

To further identify the specific cell populations in the mouse lung that were transfected with proteins, BALB/c mice were intravenously administrated with LSNPs/BSA-Cy5 nanoparticles at a dosage of 2 mg kg⁻¹ BSA-Cy5. Subsequently, lung tissues were harvested, minced, and treated with a mouse Lung Dissociation Kit (Miltenyi Biotec, USA). The resulting cell suspensions were filtrated through a 70-μm strainer, after centrifugation, the cells were treated with a red blood cell lysis buffer for 5 min, followed by an additional centrifugation step to collect the cell pellet. The isolated cells were resuspended in a flow cytometry staining buffer, and specific antibodies were introduced to the suspension. The mixture was incubated for 30 min on ice in the absence of light. After the incubation, the stained cells were washed twice using cold PBS, followed by resuspension in PBS. Subsequently, the fluorescence-positive cells were quantified on a Beckman Coulter CytoFLEX (Beckman Coulter, USA) by staining the different types of cells using the following antibodies: CD326-FITC, CD31-FITC, and CD45-FITC.

Treatment of PA-Induced Acute Pneumonia in Mouse Model.

The acute pneumonia mice model was performed following these steps: BALB/c mice were anesthetized by isoflurane inhalation, and the trachea was carefully exposed. Intratracheal instillation of PA was conducted by administering 4 × 10⁸ CFU mL⁻¹ in a 50 μL PBS solution, utilizing a BD insulin syringe (0.33 mm × 12.7 mm). Subsequent to a 1-hour interval, mice were subjected to intravenous injection of LSNPs/ProTα-CCL in PBS (at a ProTα-CCL dose of 4 mg kg⁻¹). In parallel, the control mice were intravenously injected with PBS, ProTα-CCL, or LSNPs at the same concentration as employed in LSNPs/ProTα-CCL nanoparticles. After 24 h, the mice were killed, and their lung tissues were collected for bacterial load quantification. For bacterial load analysis, the entire lung tissue was homogenized in PBS as described previously. The resulting homogenates underwent serial tenfold dilution in sterile PBS, and were then plated on agarose plates. These plates were incubated at 37 °C for 24 h, the bacterial colony-forming units (CFU) were measured and calculated. To assess the survival rate of PA-infected mice following different treatments, a lethal dose of PA culture containing 2 × 10⁹ CFU mL⁻¹ in 50 μL of PBS was intratracheally introduced per mouse. Meanwhile, ciprofloxacin (4 mg kg⁻¹) was intravenously injected into mice one-hour postbacterial inoculation. On the seventh day postinjections, the mice were killed, and their lung tissues were harvested for hematoxylin and eosin (H&E) staining. To evaluate the safety of LSNPs/ProTα-CCL nanoparticles for treating acute pneumonia, BALB/c mice were killed at 24 h after the intravenous administration of PBS and LSNPs/ProTα-CCL (4 mg kg⁻¹ of ProTα-CCL) for collecting mouse serum for liver function study and hematoxylin and eosin (H&E) staining.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.