Advanced approaches to overcome biological barriers in respiratory and systemic routes of administration for enhanced nucleic acid delivery to the lung

Abstract

Introduction:

Numerous delivery strategies, primarily novel nucleic acid delivery carriers, have been developed and explored to enable therapeutically relevant lung gene therapy. However, its clinical translation is yet to be achieved despite over 30 years of efforts, which is attributed to the inability to overcome a series of biological barriers that hamper efficient nucleic acid transfer to target cells in the lung.

Areas covered:

This review is initiated with the fundamentals of nucleic acid therapy and a brief overview of previous and ongoing efforts on clinical translation of lung gene therapy. We then walk through the nature of biological barriers encountered by nucleic acid carriers administered via respiratory and/or systemic administration routes. Finally, we introduce advanced strategies developed to overcome those barriers to achieve therapeutically relevant nucleic acid delivery efficiency in the lung.

Expert opinion:

We are now stepping close to the clinical translation of lung gene therapy, thanks to the discovery of novel delivery strategies that overcome biological barriers via comprehensive preclinical studies. However, preclinical findings should be cautiously interpreted and validated to ultimately realize meaningful therapeutic outcomes with newly developed delivery strategies in humans. In particular, individual strategies should be selected, tailored, and implemented in a manner directly relevant to specific therapeutic applications and goals.

1. Introduction

1.1. Two pillars of gene therapy

Gene therapy is the process of introducing foreign nucleic acids into host cells to elicit therapeutic benefits. Two major components essential for therapeutically relevant gene therapy are nucleic acid payloads and delivery systems. The former provides therapeutic biological functions to address pathological disease developments but suffer from poor bioavailability with rapid clearance by host immune system and endogenous nuclease during their journey from the administration site to the target site [1,2]. Further, negatively charged nucleic acids are electrostatically repelled by the negatively charged cell membrane, which is of particular concern for large nucleic acids, such as DNA and messenger RNA (mRNA) [3,4]. Thus, the latter serves a critical role to enable efficient delivery of nucleic acid payloads to target tissues and cells while protecting them from premature clearance [5–9].

1.1.1. Types of therapeutic nucleic acids

Different types of nucleic acids have been investigated to execute three primary approaches of gene therapy, including gene addition, knockdown/silencing, or editing. Gene addition is induced with DNA (e.g., plasmid DNA or pDNA) or mRNA that generates functional proteins in host cells. Gene knockdown is generally achieved with RNA interference by antisense oligonucleotide (ASO), small interfering RNA (siRNA), microRNA (miRNA), or pDNA expressing the siRNA or miRNA [10]. Gene editing requires multiple components, including mRNA encoding gene editing enzymes, such as clustered regularly interspaced palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9), single guide RNA (sgRNA), and/or template DNA [11]. While pDNA, following the delivery into the cell nuclei, enables preferential and/or prolonged gene expression in target cells by various promoters [12], mRNA, ASO, and oligo RNAs act in the cytoplasm and thus are characterized by rapid and relatively short-lived therapeutic actions [13,14]. Nucleic acid payloads are selected and designed to mediate desired therapeutic outcomes depending on the etiological nature of a target disease.

1.1.2. Nucleic acid delivery carriers

Two major classes of nucleic acids delivery systems are viral and non-viral vectors. Viral vectors, derived from naturally occurring viruses, efficiently transfer genetic information into host cells and have long been at the forefront of nucleic acids delivery research, demonstrating remarkable nucleic acid delivery efficiency and promising clinical outcomes [5,15]. Three primary types of viral vectors widely utilized in pulmonary nucleic acid delivery applications are adenovirus (AdV), adeno-associated virus (AAV), and lentivirus (LV) [16,17]. AAV vectors have been approved by the US FDA for their uses in treating non-lung genetic disorders over the past few years [18–20]. To date, a majority of the inhaled gene therapy clinical trials for treating lung diseases have evaluated viral vectors. However, the intrinsic limitations, including high immunogenicity, inflammatory side effects, moderate clinical benefits, cargo capacity limitation, and extremely high manufacturing cost, have spurred interest in developing non-viral vector systems as an alternative [21–24]. Non-viral vectors, including those based on polymers, peptides, cationic/ionizable lipids, and inorganic materials, offer another avenue for encapsulating and delivering therapeutic nucleic acids [25]. In particular, positively charged carrier materials that facilitate the formation of nanoparticles (NPs) via electrostatic interactions with negatively charged nucleic acids have been widely studied in preclinical and clinical settings [7,26,27]. Recently, mRNA and lipid NPs (LNPs) have garnered unprecedented attention due to the groundbreaking success of the mRNA/LNP vaccines against COVID-19 developed by Moderna and BioNTech/Pfizer [28]. These various delivery vehicles play a vital role in improving the stability, protection, and targeted delivery of nucleic acid payloads, paving the way for highly versatile and cost-effective therapeutic interventions in gene therapy and nucleic acid-based vaccination.

1.2. Where do we stand on lung gene therapy?

1.2.1. Previous clinical studies

Owing to the discovery of monogenic nature of a few hereditary lung disorders, including cystic fibrosis (CF) and α-1 antitrypsin (AAT) deficiency [29,30], lung gene therapy has been extensively pursued. The anatomical features of the lung allow utilization of the inhaled administration route via nebulization or intratracheal/intranasal administration for direct therapeutic exposure to and deposition in lung compartments. In addition, lung is highly vascularized and thus is reachable via the systemic route [31]. Recombinant AAV serotype 1 and 2 (AAV1 and 2) were tested in patients with CF or AAT deficiency and found to be safe in multiple clinical trials, but no significant improvement in lung function was reported [32–36]. Likewise, a non-viral system based on poly-L-lysine (PLL) conjugated to 10 kDa polyethylene glycol (PEG), developed by Copernicus Therapeutics, was given to CF patients via nebulization and found well tolerated by those patients but the primary clinical endpoint was not reached [26]. More recently, Genzyme lipid 67-based liposome (GL67A), carrying CpG-free pDNA encoding human CF transmembrane conductance regulator (CFTR), was clinically tested via monthly nebulized application by the UK CF Gene Therapy Consortium (CFGTC) (NCT01621867) [27,37]. GL76A formulated with GL67, 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), and 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-methoxy(polyethylene glycol)-5000 (DMPE-PEG5000) was well tolerated but showed significant yet marginal improvement in the forced expiratory volume in one second (FEV₁) at 12 months follow-up at best, not reaching a meaningful clinical improvement [27]. The first mRNA/LNP for inhaled application, MRT5005, was recently investigated in a Phase I/II clinical study of CF gene therapy by Translate Bio (RESTORE-CF, NCT03375047) [38]. Thirty-one patients received MRT5005 composed of imidazole cholesterol ester, DOPE, 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG2000), and codon-optimized mRNA encoding CFTR via nebulization with various treatment regimens, including daily dosing for 5 days [38]. Although it was generally safe and well tolerated 28 days after the last dose, beneficial effect on percent predicted FEV₁ (ppFEV₁) was not achieved [38], similar to the observations in previous trials. On the other hand, oligo nucleic acids, such as siRNA and ASO, were developed and investigated as career-free inhaled formulations for treating asthma, respiratory syncytial virus injection, or CF but were discontinued in early clinical trials due to lack of significant therapeutic benefits [39–42].

1.2.2. Current effort to implement lung gene therapy in clinical trials

Early clinical studies revealed that inhaled nucleic acid carriers were unable to efficiently transduce/transfect fully differentiated lung airway epithelial cells from the apical surface due to the low density of cell surface receptors and a low rate of endocytosis compared to the basolateral membrane [43,44]. Thus, nucleic acid delivery strategies have been primarily devised to increase the efficiency of carrier delivery into airway epithelium. The 4D Molecular Therapeutics is leading a Phase I/II clinical trial for CF gene therapy (NCT05248230) with 4D-710 identified by directed evolution of AAV capsids to target and transduce airway epithelial cells. Three patients who were ineligible/intolerant to CF modulators were followed for 9 – 12 months after a single treatment with 4D-710 via nebulization. Analysis of bronchoscopy sample results demonstrated widespread and consistent expression of the CFTR transgene protein in 92% - 99% of lung airway cells at significantly higher levels than in normal control and CF lung-derived lung tissue samples. The ppFEV₁ was significantly improved in a participant with moderate ppFEV₁ impairment at baseline and ppFEV1 remained stable at baseline in participants with normal or mild impairment [45]. As a similar airway epithelial targeting strategy, the UK CFGTC is focusing on organizing a Phase I/IIa inhaled CF gene therapy trial investigating a psedotyped LV vector, Sendia virus protein-pseudotyped simian immunodeficiency virus (SIV) [46].

Although clinical trials of lung gene therapy to date rely predominantly on respiratory (i.e., local) administration, a number of recent preclinical studies have highlighted lung-targeted strategies with increased efficiency of pulmonary nucleic acid delivery via systemic administration. To this end, we here aim to delineate the biological obstacles associated with either or both delivery route(s). Subsequently, we introduce various delivery methodologies, specifically those tailored to overcome respiratory, systemic, and immune barriers. Further, our comprehensive review attempts to critically evaluate the distinct strategies employed in addressing the physiological and biological challenges, leveraging on our current understanding of those barriers.

2. Biological barriers to pulmonary nucleic acid delivery

2.1. Barriers in respiratory route

The fluid layer on the airway lumen, named air surface liquid, consists of two distinct components, including the luminal mucus layer and the periciliary layer (PCL) (Figure 1A). The mucus layer, which is the initially encountered biological barriers to inhaled foreign matters, is a dense viscoelastic gel containing high molecular weight (2 – 200 MDa) polymeric glycosylated mucin fibers [47]. The glycosylated regions of mucin fibers carry a net negative charge and are separated by periodic hydrophobic protein domains. Thus, the inhaled matters, including particles and pathogens, are captured in the mucus via adhesive interactions, such as electrostatic interactions, hydrophobic forces, and hydrogen bonding [48]. It has been demonstrated that polystyrene NPs coated with non-adhesive PEG as large as 200 – 300 nm, but not 500 nm, are capable of penetrating airway mucus collected from patients with or without lung diseases such as CF and chronic obstructive pulmonary disease (COPD) [49–51], underscoring the existence of a pore size cut-off that restricts particle mobility in mucus. The physicochemical nature of the mucus layer as a barrier may vary depending on the disease state [50,52]. Of note, most of the viral and non-viral vectors tested in clinical trials of inhaled gene therapy to date were found unable to efficiently penetrate human airway mucus [53–56]. Particulate matters trapped in the airway mucus via adhesive interactions and/or physical obstruction are rapidly cleared from the lung via mucociliary clearance driven by the PCL at a rate of 3.6 mm per minute (in humans) [57–59]. The PCL is located immediate underneath the mucus gel layer and contains a tethered mucin brush on the surface of the ciliated cells [60], thereby serving as a secondary protective barrier to inhaled foreign substances. Dextran molecules and non-PEGylated polystyrene NPs possessing hydrodynamic diameters of 40 nm were excluded from the PCL established in air-liquid interface (ALI) cultures of human bronchial epithelial (HBE) cells [61]. In another study, AAV sized 25 – 30 nm penetrated into the PCL of the ALI cultures, whereas AdV sized ~100 nm was completely excluded from the layer [60]. These reports suggest that PCL likely possesses tighter meshes compared to the luminal mucus layer. Of note, 50% - 80% of airway epithelial cells are ciliated [62], and thus the PCL may not constitute a critical delivery barrier throughout the non-ciliated epithelial surfaces.

Figure 1. Nucleic acid carriers administered via respiratory or systemic route encounter distinct and shared biological barriers.

(A) The carriers administered via the respiratory route must traverse the luminal mucus gel layer and the periciliary layer (PCL) and subsequently get into the target cells in tightly sealed airway epithelium. (B) The carriers administered via the systemic route must accumulate and/or penetrate the pulmonary endothelium while avoiding liver uptake and persistent circulation. (C) Regardless of the administration route, the carriers must evade the host immune system, including macrophages and neutralizing antibodies, prior to reaching target lung parenchymal cells. MCC: mucociliary clearance; PCL: periciliary layer; ATIl: alveolar epithelial type I; ATII: alveolar epithelial type II; RBC: red blood cell. This figure has been created with BioRender.com.

The airway epithelium beyond the air surface liquid layer acts as a delivery barrier [63] as well as a therapeutic target. Airway epithelium is composed of various cells, including ciliated cells, mucus-producing goblet cells, and progenitor basal cells (Figure 1A). These cells separate the external environment from the epithelial and subepithelial tissue through intercellular junctions that prevent the access of inhaled particulate matters to the basolateral side [64]. This aspect has been a critical challenge to clinically tested viral vectors, such as AAV2 and AdV, of which primary cell surface receptors are located on the basolateral surface of airway epithelium [43,65]. However, these intercellular junctions act as dynamic structures that can open or close in response to physiological and pathological stimuli, and thus the barrier property varies with disease states [66,67].

2.2. Barriers in systemic route

The lung is one of the most vascularized organs [31]. In addition, the first capillary bed encountered after intravenous injection into a peripheral vein is that of the lungs [68], which makes the lung accessible via the systemic route. However, intravenously injected particulate matters, such as viral vectors and NPs, readily accumulate in liver and other mononuclear phagocyte system (MPS) organs and are rapidly eliminated from the bloodstream [69,70]. In particular, liver tissue is extensively perfused by hepatic sinusoids, characterized by endothelium perforated with transcellular pores, named fenestrae (50 – 300 nm in diameters) [71], and the total hepatic blood flow accounts for approximately 25% of resting cardiac output while the liver mass constitutes only 2.5% of the total body weight [72]. These features collectively lead to prolonged liver resident time of circulating matters and their markedly greater interactions with hepatic cells, including but not limited to the liver-resident macrophages, Kupffer cells, compared to peripheral cells (Figure 1B) [73]. Consequently, only a limited fraction of the systemically administered nucleic acid carriers reaches other non-MPS organs, including the lung. Nucleic acid carriers that manage to evade MPS clearance must traverse the pulmonary endothelium to reach the target lung cells. Pulmonary microvasculature is composed of non-fenestrated blood capillaries sealed with intercellular junctions with an upper limit pore size of approximately 5 nm (Figure 1B) [74] and serves as a barrier that regulates penetration of circulating matters into the pulmonary interstitium [31]. Both viral and non-viral vectors tested in clinical and preclinical settings are markedly larger than this size cut-off and thus are unlikely to traverse the unperturbed pulmonary endothelium via the intercellular route. However, AAV vectors and positively charged liposomes have been shown to transduce/transfect lung epithelial cells following systemic administration [75–77], suggesting that other non-intercellular mechanisms that permit the penetration of circulating matters across the pulmonary endothelium exist. Of note, permeability of the pulmonary endothelium is altered by pro-inflammatory cytokines/chemokines and other inflammatory mediators upregulated in various inflammatory lung diseases, including but not limited to COPD and asthma [78–81].

2.3. Immune barriers

Regardless of the administration route, the host immune system poses a critical hurdle that hampers efficient delivery of nucleic acid carriers to target lung parenchymal cells. Lung-resident macrophages that serve to engulf harmful foreign particulate matters are prevalent throughout the lung compartments, including airways and alveoli, and are the most abundant cell type in the alveolar space (Figure 1C) [82]. Macrophages favor micro-sized particles [83], and conventional nucleic acid carriers are much smaller, rarely larger than 200 nm, but can be attracted to macrophages via chemical and/or molecular interactions depending on their surface properties [21,84,85]. Indeed, alveolar macrophages have been shown to preclude delivery of viral and non-viral vectors to target lung cells either by direct phagocytosis or by opsonin-dependent mechanisms [86,87]. Phagocytosis of nucleic acid carriers triggers pro-inflammatory responses [87,88], resulting in the recruitment of innate immune cells that further promote removal of nucleic acid carriers via phagocytosis [89]. In addition, this innate immune cascade can induce carrier-specific adaptive immune responses, including the production of therapeutic-inactivating neutralizing antibodies [90] (Figure 1C), which is of a particular concern for viral vectors [21]. It should be noted that significant fractions of healthy individuals and those with CF are readily seropositive to various AAV serotypes, including the clinically tested AAV1 and 2 [91,92]. These seropositive populations are unlikely to respond to gene therapy based on the respective AAV vectors in a similar manner observed with patients receiving repeated administration of an identical vector [21].

3. Advanced strategies to overcome the biological barriers to pulmonary nucleic acid delivery

3.1. Strategies to tackle respiratory delivery barriers

Early clinical studies revealed that clinically tested viral and non-viral vectors were unable to efficiently transduce/transfect airway epithelial cells from the apical surface after administration via the respiratory route [43,44,93]. The lesson ignited efforts to develop nucleic acid carriers capable of engaging with, and internalizing into, these target cells from apical surface to intracellularly introduce therapeutic nucleic acid payloads. In the meantime, mucus gel layer covering the lung airways has been identified and appreciated as one of the greatest challenges in achieving therapeutically relevant inhaled nucleic acid transfer to the lung airways via the respiratory route [94–96] To this end, we here discuss diverse strategies developed and evaluated to overcome these barriers for efficient nucleic acid delivery to the target cells in the lung (Table 1). We note that this task can be realized by modulating either nucleic acid carriers or delivery barriers such as airway mucus. We here focus on the former and encourage readers to refer to other comprehensive reviews for the latter [96,97].

Table 1.

Strategies to tackle respiratory delivery barriers.

Strategy		Carrier	Cargo	Target disease	Model	Administration	Reference
Viral surface modification for airway transduction	Selection and rational design of viral capsids	AAV2 mutant (THALWHT)	ssDNA (lucifera se)	-	ALI cultures of HAE (Calu-3)	Apical	[108]
				CF	ALI cultures of CF HBE (CFBE41o-)	Apical
		AAV2 mutant (KKFNKPFVFLI)	ssDNA (ATT)	-	BALB/c mice	Intratracheal	[110]
		AAV8 mutant (Y773F)	ssDNA (PEDF)	-	C57BL/6 mice	Intratracheal	[113]
		AAV6	ssDNA (EGFP or luciferase-YFP)	CF	ALI cultures of CF HBE (CFBE41o-)	Apical	[114]
				CF	ALI cultures of primary CF HBE	Apical
				CF /COPD	Scnn1b-Tg mice	Intratracheal
	Directed evolution	AAV2.5T	ssDNA (truncated CFTR)	CF	ALI cultures of primary CF HBE	Apical	[120]
				CF	Neonate or juvenile ferrets	Intratracheal	[121]
		AAV2H22	ssDNA (truncated CFTR)	-	ALI cultures of primary PAE	Apical	[122]
				CF	CF pigs	Aerosol inhaled
	Capsid/envelope pseudotyping	AAV2/HBoV1	ssDNA (luciferase)	-	ALI cultures of primary HAE	Apical	[126]
			ssDNA (CFTR)	CF	ALI cultures of primary CF HAE	Apical
			ssDNA (luciferase)	-	ALI cultures of primary HAE	Apical	[127]
					Neonate or juvenile ferrets	Intratracheal
		GP64-FIV	ssDNA (luciferase or β-galactosidase)	-	ALI cultures of primary HAE	Apical	[130]
				-	BALB/c mice	Intranasal
			ssDNA (luciferase or mCherry)	-	ALI cultures of primary HAE	Apical	[133]
				-	ALI cultures of primary PAE	Apical
				-	Pigs	Intranasal
			ssDNA (erythropoietin, β-galactosidase, or luciferase)	-	BALB/c mice	Intranasal	[135]
		GP64-FIV mutant (E45K/T259A)	ssDNA (luciferase)	-	ALI cultures of primary HAE	Apical	[136]
		F/HN-SIV	ssDNA (GFP or luciferase)	-	C57BL/6N mice	Intranasal	[137]
			ssDNA (luciferase or GFP-CFTR)	-	ALI cultures of primary HAE	Apical
			ssDNA (GFP or luciferase)	-	C57BL/6N mice	Intranasal	[132]
				-	ALI cultures of primary nasal HAE	Apical
				-	Human nasal brushings and human/sheep lung slices	-
			ssDNA (EGFP or luciferase)		ALI cultures of primary HBE	Apical	[138]
Chemical Surface modification for enhanced mucus penetration	PEGylation	PEGylated PEI NP	pDNA (luciferase or CFTR)	-	BALB/c mice	Intranasal	[140]
				CF	ALI cultures of primary CF HAE	Apical
		PEGylated CPP NP	pDNA (luciferase or GFP)	-	BALB/c mice	Intratracheal	[142]
		PEGylated PBAE NP	pDNA (EGFP)	-	BALB/c mice	Intratracheal	[141]
			pDNA (thymulin)	Asthma	BALB/c mice challenged with intratracheal OVA	Intratracheal	[143]
			pDNA (luciferase, GFP, or shENaC)	CF/COPD	Scnn1b-Tg mice	Intratracheal	[144]
				CF	ALI cultures of primary CF HBE	Apical
			pDNA and oligo DNA (OVA and CpG)	Lung cancer	BALB/c mice implanted with OVA-expressing LLC cells	Intratracheal	[145]
		PEGylated PLGA /cationic lipid-like molecule NP	siRNA (IL-11)	Idiopathic pulmonary fibrosis	C57BL/6 mice challenged with intratracheal bleomycin sulfate	Inhalation	[146]
		PEGylated TPFE NP	pDNA (ZsGreen1)	-	C57BL/6 mice	Intratracheal	[147]
				CF/COPD	Scnn1b-Tg mice	Intratracheal
		PEGylated LNP	mRNA (luciferase or Cre recombinase)	-	BALB/c mice and Ai9 mice	Aerosol inhaled	[148]
			mRNA (CFTR or luciferase)	CF	CFTR KO-Tg mice	Aerosol inhaled
		PEGylated AdV	ssDNA (β-galactosidase)	-	C57BL/6 mice	Intratracheal	[149]
		PEGylated AAV2	ssDNA (β-galactosidase)	-	C57BL/6 mice	Intratracheal	[150]
	Other surface modifications	HA-coated PBAE NP	siRNA (TNF-α)	Acute lung injury	C57BL/6 mice challenged with intratracheal LPS	Intranasal	[151]
		Fluorinated cationic polypeptide NP	siRNA (TNF-α)	Acute lung injury	Balb/c mice challenged with intratracheal LPS	Intratracheal	[154]
Material screening		NLD1 LNP	mRNA (membrane-anchored nanoluciferase)	-	BALB/c mice	Aerosol inhaled	[155]
			mRNA (membrane-anchored FI6)	Influenza infection	BALB/c mice infected with influenza A virus (H1N1/PR8)	Aerosol inhaled
		Ionizable lipid LNP	mRNA (luciferase)	-	C57BL/6J mice	Intratracheal	[158]
			mRNA and sgRNA (SpCas9 and sgAi9L/sgAi9R)	-	Ai9 mice	Intratracheal
		PBAE NP	mRNA (luciferase)	-	BALB/c and DBA/2	Aerosol inhaled	[159]
				-	LVG golden Syrian hamsters	Aerosol inhaled
				-	Neutered and de-scented Fitch ferrets	Aerosol inhaled
				-	Holstein calf	Aerosol inhaled
				-	Rhesus macaques	Aerosol inhaled
			mRNA and crRNA (Cas13a and N3.2)	SARS-CoV-2 infection	LVG golden Syrian hamsters Infected with SARS-CoV-2	Aerosol inhaled
Piggybacking on EVs		EVAAV6	ssDNA (luciferase-YFP)	-	C57BL/6 mice	Intratracheal	[171]
				-	ALI cultures of primary HBE	Apical
				CF	ALI cultures of primary CF HBE	Apical

ssDNA: single-strand DNA; PEDF: pigment-epithelial-derived factor; shENaC: short hairpin RNA against epithelial sodium channel; Scnn1b-Tg: transgenic mice overexpressing epithelial sodium channel β-subunit; PAE: pig airway epithelial; CPP: cell-penetrating peptide; OVA: ovalbumin; LLC: Lewis lung carcinoma; PLGA: poly(lactic-co-glycolic acid); TPFE: tetra(piperazino)fullerene epoxide; LPS: lipopolysaccharide

3.1.1. Viral surface modification for airway transduction

3.1.1.1. Selection and rational design of viral capsids

Numerous fundamental studies have deepened the understanding of the AAV vector capsid property that plays critical roles on the vector interactions with biological entities and thus the transduction efficiency [19]. AAV2 binds to heparin sulfate proteoglycan (HSPG) [98] present on the basolateral surface of airway epithelial cells [43,99], demonstrating limited transduction when administered via the respiratory route to target the apical surface [100,101]. The reality led to multiple preclinical studies that investigated other AAV serotypes potentially for enhanced inhaled gene therapy applications. In particular, AAV1, 5, and 6 have been extensively shown to provide greater in vivo airway transduction compared to AAV2 [100,102], likely attributed to their ability to interact with sialic acids [103–105] found on the apical surface of airway epithelium [104,106]. To improve the apical airway transduction efficiency of AAV2, a targeting moiety has been incorporated on the capsid at specific regions which do not interfere with capsid assembly or genome packaging. A 7-mer peptide, THALWHT, which was discovered by phage display on human bronchial epithelium-derived cell line (i.e., 16HBE14o−) [107], was inserted on the AAV2 capsid by genetic engineering [108]. This AAV2 mutant showed significantly greater transduction compared to wild-type AAV2 in ALI cultures of human airway epithelial (HAE) and CF HBE cells [108]. Likewise, AAV2 mutants engineered to display a long serpin ligand, KKFNKPFVFLI [109], on the capsid exhibited significantly greater human ATT transgene expression compared to wild-type AAV2 in mouse lungs following intratracheal or intranasal administration [110]. However, the levels of transgene expression mediated by these mutants were markedly lower than the level observed with AAV5 possessing an inherent apical airway binding affinity [110].

AAV2 exhibits poor mobility in pathological human mucus (i.e., CF sputum) [54], presumably due to the abundance of HSPG in CF sputum [111]. In comparison, an AAV2 mutant harboring two mutations that reduce heparin binding, R585A and R588A [112], has demonstrated significantly enhanced diffusion in CF sputum compared to wild-type AAV2 [54]. More recently, an AAV8 harboring a single mutation, Y773F, was found to more rapidly diffuse in CF sputum compared AAV2, which was correlated with its enhanced transgene expression in healthy mouse lungs compared to AAV2 following intratracheal administration [113]. However, it is yet to be determined whether the observation reflects the inherent ability of wild-type AAV8 to provide enhanced mucus penetration compared to AAV2 or the single mutant implemented on the AAV8 capsid has contributed to it. Duncan et al. also showed that AAV6 efficiently penetrated CF sputum with significantly greater diffusion rates compared to AAV1 [114] which exhibited poor mobility in CF sputum similarly as AAV2 [54]. Interestingly, one point mutation implemented on the AAV6 capsid that confers AAV6 with AAV1-like binding affinity to glycans, K531E [115], resulted in the loss of its diffusion benefit in human airway mucus [114]. Thanks to the mucus-penetrating ability, AAV6 provided markedly greater transduction efficiency in ALI cultures of primary CF HBE cells and in the lungs of a transgenic mouse model characterized by mucus hypersecretion and stasis [114], a pathological hallmark of CF and COPD [116,117].

3.1.1.2. Directed evolution

Directed evolution of AAV involves screening of a library of AAV mutants with a relevant in vitro or in vivo model to identify variant(s) providing superior transduction efficiency in a target cell population [118–120]. Excoffon et al. established a library of AAV mutants possessing highly diverse chimeric capsids based on AAV2 and 5 using DNA shuffling and error-prone PCR and screened on ALI cultures of primary HAE cells [120]. Five rounds of selection revealed a single AAV variant, AAV2.5T, possessing a chimeric structure of AAV2 (aa1-128) and AAV5 (aa129–725) with a point mutation (A581T), that provided over 100-fold greater transduction in the ALI cultures compared to wild-type AAV2 and 5 [120]. In a subsequent study, AAV2.5T was shown to efficiently transduce ferret airway epithelial cultures and to produce human CFTR mRNA at 2 – 3-fold greater levels compared to endogenous ferret CFTR mRNA in ferret lungs [121]. Another library of chimeric AAV mutants was prepared by shuffling of AAV1, 2, 4, 5, 6, 8, and 9 capsid genes and error-prone PCR and subjected to sequential screening on ALI cultures of primary pig airway epithelial cells (PAE) and in non-CF 6-week-old pigs [122]. A single lead AAV variant identical to AAV2 except for five mutations, including E67A, S207G, Q598L, I648V, and V708I, named AAV2H22, was identified to provide a 240-fold greater transduction efficiency compared to AAV2 in vitro and partially restored CFTR function in CF pig airways [122]. As introduced earlier, a clinical trial of AAV mutant identified by the direct evolution technique has been recently initiated to potentially treat CF patients who cannot use CFTR modulators (4D-710; NCT05248230).

3.1.1.3. Capsid/envelope pseudotyping

Approximately two decades ago, human bocavirus 1 (HBoV1) was discovered to cause acute respiratory tract infection [123] and was shown to productively infect ALI cultures of HAE cells from the apical surface even at extremely low multiplicity of infection [124,125]. Leveraging on this high airway tropism of HBoV1, Yan et al. went on to pseudotype AAV2 with HBoV1 capsid to engineer a novel chimeric vector, AAV2/HBoV1 [126]. AAV2/HBoV1 apically transduced the ALI cultures of HAE cells at 5.6- and 70-fold greater efficiency than AAV1 or AAV2, respectively, and apical treatment of ALI cultures of CF HAE cells with AAV2/HBoV1 carrying CFTR gene resulted in efficient correction of CFTR-dependent chloride transport [126]. Further, AAV2/HBoV1 was capable of efficiently transducing the lungs of both neonate and juvenile ferrets following intratracheal administration, and prior vector exposure appeared not to compromise the transduction efficiency of AAV2/HBoV1 [127]. LV vector envelops were also extensively pseudotyped with various glycoproteins from diverse viral families with known lung epithelium tropism, including filovirus [128], coronavirus [129], baculovirus [130], and Sendai virus [131,132]. In a glycoprotein screening study, feline immunodeficiency virus (FIV) vector pseudotyped with glycoprotein GP64 of baculovirus (GP64-FIV) demonstrated most efficient apical transduction of ALI cultures of HAE and PAE compared to other glycoprotein candidates [133]. In addition, GP64-FIV was able to transduce pig airways in vivo following intratracheal administration [133]. Encouragingly, due to the relatively lower immunogenicity of LV vectors compared to other viral vectors [134], consecutive daily intratracheal dosing of GP64-FIV resulted in a linear increase in reporter transgene expression in mouse airway epithelial cells without the generation of local or systemic neutralizing antibodies [135]. As a combinatorial strategy, directed evolution was implemented on GP64-FIV to identify a variant with two point mutations, E45K and T259A, which showed 8-fold greater transduction compared to the parent GP64-FIV in ALI cultures of HAE cells [136]. The UK CF CFGTC is leading effort on clinical development of SIV pseudotyped with Sendai virus fusion protein/hemagglutinin and neuraminidase protein (F/NH-SIV) [46]. Griesenbach et al. demonstrated that a single intranasal administration of F/HN-SIV resulted in persistent reporter transgene expression in mouse lungs for the animal lifetime of about 2 years without incurring chronic toxicity [132,137]. F/HN-SIV also transduced ALI cultures of primary HAE cells to a greater extent compared to the clinically tested non-viral system, GL67A [132,137]. More recently, F/HN-SIV was found to efficiently transduce both ciliary and non-ciliary airway cells using sialylated glycans abundant on the apical surface of human lung airway cells, which underscores the suitability of the vector for human lung gene therapy applications [138].

3.1.2. Chemical surface modification for enhanced mucus penetration

3.1.2.1. PEGylation

Hydrophilic and neutrally charged PEG has been most widely utilized to endow nucleic acid carriers with non-mucoadhesive surfaces, thereby enhancing their penetration through airway mucus [96,139]. In particular, PEGylation has been employed to various non-viral vectors based on cationic carrier materials, including polymers, such as polyethyleneimine (PEI) [140] and poly(β-amino esters) (PBAE) [141], and peptides, such as PLL [140] and cell-penetrating peptide [142], and have demonstrated markedly enhanced mobility in CF sputum. These DNA-loaded mucus-penetrating particles (DNA-MPPs) were shown to provide widespread airway distribution, prolonged lung retention, and/or enhanced pulmonary reporter transgene expression in mouse lungs compared to otherwise identical positively charged, and thus mucus-impermeable, counterparts following intranasal or intratracheal administration [140–142]. Subsequently, the therapeutic relevance of PBAE-based DNA-MPP was established in multiple preclinical studies [143–145]. Intratracheal administration of DNA-MPPs carrying pDNA encoding thymulin [143] or short hairpin RNA against epithelial sodium channel [144] resulted in near-complete normalization or marked reduction of relevant disease phenotypes in a mouse model of asthma or CF/COPD, respectively. Further, DNA-MPP was engineered to carry pDNA encoding a model antigen and a nucleic acid-based adjuvant and shown to induce markedly enhanced antigen-specific local and systemic immune responses following intratracheal administration, compared to standard vaccination via intradermal injection or intramuscular electroporation [145]. Similar to DNA-MPPs, PEGylated siRNA delivery NPs composed of biodegradable polymer and cationic lipid-like molecule demonstrated enhanced mucus penetration and therapeutic relevance; siRNA against interleukin-11 (IL-11) packaged and intratracheally administered in these NPs mediated efficient IL-11 silencing and therapeutic efficacy in a mouse model of pulmonary fibrosis [146]. PEGylation also rendered the surface of inorganic nucleic acid delivery NPs, based on tetra(piperazino)fullerene epoxide, non-mucoadhesive to markedly improve the distribution and overall level of reporter transgene expression in mouse lungs [147]. More recently, LNPs with a greater PEGylated lipid (i.e., DMG-PEG2000) content exhibited enhanced diffusivity in a mucin suspension and provided widespread mRNA expression in mouse lungs after nebulization [148]. PEGylation of viral vectors, including AdV [149] and AAV2 [150], was shown to enhance or prolong transgene expression in mouse lungs following intratracheal administration. Although these early findings were primarily attributed to the ability of PEG to reduce immunogenicity, improved mucus penetration by the PEGylation might have contributed to the outcomes, given the limited ability of these viral vectors to penetrate the airway mucus [53,54].

3.1.2.2. Other surface modifications

As an alternative to PEG, hyaluronic acid (HA) has been used to shield positively charged siRNA delivery NPs to promote airway mucus penetration. Specifically, an optimized surface coating with HA was shown to near-neutralize positively charged siRNA-loaded PBAE NPs and enhance their penetration through purified CF sputum [151]. Accordingly, mice with acute lung injury treated intratracheally with HA-coated NPs carrying siRNA against tumor necrosis factor-α (TNF-α) exhibited significantly greater TNF-α knockdown and therapeutic effects compared to those treated with uncoated NPs [151].

Fluorinated dendrimers were previously shown to mediate efficient nucleic acid delivery in serum-containing media [152] and this unique serum resistance was attributed to the lipophobic as well as hydrophobic nature of fluorocarbon compounds [153]. Inspired by this finding, Ge et al. validated their hypothesis that fluorocarbon modification would minimize mucoadhesion to enhance mucus penetration of nucleic acid carriers [154]. Specifically, carriers based on fluorinated cationic polypeptides, named P3F16 and P7F7, exhibited more than two orders of magnitude greater diffusion rates in diluted (5%) CF mucus compared to the otherwise identical unfluorinated carriers [154]. Likewise, P3F16 and P7F7 carrying siRNA against TNF-α provided comprehensive and significantly greater anti-inflammatory effects in the lungs of mouse model of acute lung injury following intratracheal administration compared to the unfluorinated counterparts [154].

3.1.3. Material screening

It is quite challenging to rationally design nucleic acids carriers capable of simultaneously tackling multiple independent biological barriers encountered in the respiratory route. Thus, in vivo screening of carrier material variants and relative compositional ratios has been employed to determine lead formulations that provide efficient lung nucleic acids delivery following administration via the respiratory route. Lokugamage et al. conducted in vivo cluster-based iterative LNP screening study with four variables that they hypothesized to affect nebulized LNP delivery; those included the amount and structure of PEG-lipid, the charge of helper lipid, and the presence or absence of cholesterol, while holding an ionizable lipid, 7C1, as a constant [155]. Based on successive rounds of screening with three selection criteria, including small size (< 200 nm in hydrodynamic diameter), stability (low dispersity), and nebulized mRNA delivery efficiency, NLD1, composed of 7C1 (35%), 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (DMPE-PEG2000, PEG-lipid; 55%), cholesterol (5%), and 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP; positively charged helper lipid; 5%), was identified as a lead formulation [155]. NLD1 exhibited significantly greater reporter mRNA expression in the lung following nebulized administration compared to previously developed LNPs, including the DLin-MC3-DMA (MC3)-based LNP (analogous to ONPATTRO^®; [156]) and protection against Influenza virus when antibody-encoding mRNA was loaded and delivered [155]. Notably, the amount of PEG-lipids in NLD1 is far greater than that of clinically used formulations (e.g., COVID-19 vaccines, etc.; 1.5% - 1.6% PEG-lipid) [157], which may have enhanced the ability of the formulation to penetrate airway mucus. Recently, 720 LNPs formulated with distinct and newly synthesized ionizable lipids were sequentially screened for reporter mRNA expression after intramuscular injection and then intratracheal administration [158]. A LNP formulation based on a lead ionizable lipid (RCB-4–8; 30%), DMPE-PEG2000 (1%), cholesterol (30%), and DOTAP (39%) showed approximately 100-fold greater reporter mRNA expression in the lung following intratracheal administration compared to the MC3-based LNP, presumably due to the biodegradable nature of RCB-4–8 [158]. It is conceivable that a LNP formulation based on RCB-4–8 with a greater PEG-lipid content may provide enhanced mRNA delivery and expression in the lung.

The in vivo screening approach was also applied to a polymer-based library for nebulized mRNA delivery to the lung [159]. Specifically, out of 166 distinct PBAE polymer variants, a thiolated polymer, P76, demonstrated greater reporter mRNA expression in healthy mouse lungs compared to non-thiolated polymers and robust mRNA expression in the lungs of other species, including hamster, ferret, cow, and macaque [159]. Further, P76 NPs carrying Cas13a-encoding mRNA and anti-SARS-CoV-2 CRISPR RNA (crRNA) enhanced the treatment of a SARS-CoV-2-challenged hamster model compared to a non-thiolated PBAE NPs previously shown to mediate efficient mRNA expression in mouse and hamster lungs [160,161]. Overall, in vivo screening studies revealed novel formulations that provided improved nucleic acid delivery to the lung via the respiratory route, but the mechanisms by which the enhancements were achieved warrant further investigation.

3.1.4. Piggybacking on extracellular vesicles

Over the past decade, extracellular vesicles (EVs) have been widely explored as a therapeutic delivery platform due to their inherent ability to shuttle various biological cargoes between cells [162,163]. On the other hand, EVs are found in human airway mucus to mediate intercellular crosstalk [164–166], and mucosal EVs play key pathological roles in various chronic respiratory diseases, such as CF, COPD, and asthma [167–170]. The findings allude that EVs likely possess inherent ability to penetrate human airway mucus. To this end, Kwak et al. recently investigated AAV6 stably associated with EVs (EVAAV6) released from AAV-producing cells as a hybrid nucleic acid delivery system for inhaled gene therapy applications (Figure 2) [171]. EVAAV6 exhibited enhanced ability to penetrate CF sputum and transduce mucus-free HBE cell line [171]. Accordingly, EVAAV6 provided significantly greater transduction of mucus-covered ALI cultures of primary non-CF and CF HBE cells and healthy mouse lungs following intratracheal administration compared to AAV6 previously shown to outperform AAV1 [114,171]. Of note, a physical mixture of individually prepared EVs and AAV6 was unable to provide enhanced gene transfer efficacy, underscoring that the stable EV-AAV6 association was required for the desired outcome [171].

Figure 2. AAV6 stably associated with EVs efficiently penetrates human airway mucus and provides marked enhanced transduction of mucus-covered airway epithelium in vitro and in vivo following apical and intratracheal administration, respectively, compared to AAV6.

(A) Representative transmission electron micrographs of AAV6, EVs, EVAAV6 and EV+AAV6. (B) Median MSD values of EVs and EVAAV6 in sputum samples spontaneously expectorated by CF patients. MSD is a square of distance traveled by an individual particulate matter within a predetermined time interval (i.e., time scale; τ = 1 s) and thus is directly proportional to the particle diffusion rate. The red dashed line indicates the MSD value of AAV6 previously measured in CF sputum [114]. n.s.: no significance (two-tailed Student’s t-test). (C) Luciferase activity measured in lysates of mucus-free HBE cells (16HBE14o-) treated with EVs, EVAAV6 or EV+AAV6. n.s.: no significance, ****p < 0.0001 (one-way ANOVA). (D) Representative confocal images demonstrating reporter transgene expression in ALI cultures of primary wild-type (WT) or CF HBE cells treated with normal saline, AAV6, EVAAV6, or EV+AAV6. Green: YFP, Blue: nuclei. Image-based quantification of (E) coverage and (F) intensity of YFP transgene expression in the ALI cultures of primary WT (black) and CF (red) HBE cells treated with normal saline, AAV6, EVAAV6, or EV+AAV6. n.s.: no significance, ***p < 0.001, ****p < 0.0001 (two-way ANOVA). (G) Representative confocal images demonstrating reporter transgene expression throughout the whole left lung lobes of mice intratracheally treated with normal saline, AAV6, EVAAV6 or EV+AAV6. Green: YFP, Blue: nuclei. (H) Luciferase activity measured in lysates of the whole mouse lungs treated with normal saline, AAV6, EVAAV6, or EV+AAV6. n.s.: no significance, **p < 0.01, ***p < 0.001, ****p < 0.0001 (one-way ANOVA). This figure has been adopted and modified from [171] which has been distributed under a CC BY-NC 4.0 License.

3.2. Strategies to tackle systemic delivery barriers

As introduced earlier, despite the highly vascularized nature of lung tissue [123], rapid clearance of circulating matters by liver and other MPS organs [69,70] makes systemic nucleic acid delivery to the lung challenging. Thus, the efforts to improve pulmonary nucleic acid delivery via the systemic route have centered around the strategies to enhance lung endothelial targeting or accumulation of nucleic acid carriers while minimizing their interactions with MPS organs (Table 2). Various relevant strategies are discussed here.

Table 2.

Strategies to tackle systemic delivery barriers.

Strategy	Carrier	Cargo	Target disease	Model	Administration	Reference
Molecular lung targeting	AAV9 mutant (AAV9.452sub.LUNG1)	ssDNA (EGFP)	-	C57Bl/6J mice	Intravenous (retro-orbital)	[77]
	GALA-decorated liposome	siRNA (podoplanin)	-	C57BL6/J mice	Intravenous	[181]
	PECAM-1 antibody-conjugated LNP	mRNA (luciferase or EGFP)	-	C57BL/6 mice	Intravenous (retro-orbital)	[183]
Electrostatic lung accumulation	AAV5-PK2	ssDNA (AAT)	-	C57BL6/J mice	Intravenous	[184]
	DOTMA lipoplexes	siRNA (luciferase)	Metastatic lung cancer	BALB/c and C57BL/6 mice	Intravenous	[185]
	DOTAP lipoplexes		Lung metastasis of breast cancer	BALB/c mice	Intravenous	[187]
	DOTAP LNP	mRNA (luciferase)	-	C57BL/6 mice	Intravenous	[182]
		mRNA (Nls-Cre, EGFP, or tdTomato)	-	Ai14 and C57BL/6 mice	Intravenous	[186]
		mRNA and/or sgRNA (mCherry, luciferase, or Cas9/sgRNA)	-	C57BL/6 mice	Intravenous	[189]
Material screening	PBAE/PEG-lipid NP	mRNA (luciferase)	-	C57BL/6 mice	Intravenous	[191]
	PBAE/lipid NP	mRNA and DNA (luciferase)	-	C57BL/6 mice	Intravenous	[192]
Red blood cell hitchhiking	Red blood cell-anchored AAV9	ssDNA (luciferase or EGFP)	-	BALB/c and C57BL/6 mice	Intravenous	[198]

3.2.1. Molecular lung targeting

To engineer a novel AAV vector for lung targeting via the systemic route, Cre recombination-based AAV targeted evolution (CREATE) selection process [172] was implemented on a library of AAV9 variants [77]. Specifically, AAV9 variants were generated by substituting the surface-exposed capsid residues aa452–458 with a 7-mer randomized amino acids and administered via retro-orbital injection to be screened via CREATE in Tek-Cre transgenic mice which preferentially express Cre recombinase in the lung [77]. After two rounds of selection, 426 lung tropic variants were discovered and one variant with an amino acid motif KDNTPGR, named AAV9.452sub.LUNG1, was enriched nearly an order of magnitude more than any other variants [77]. Subsequently, AAV9.452sub.LUNG1 was shown to mediate 18- and 60-fold greater reporter transgene expression compared to AAV9 and 5, respectively, in the lungs of wild-type mice following systemic administration [77]. Interestingly, transgene expression mediated by AAV9.452sub.LUNG1 was more prominent in alveolar epithelial type II (ATII) cells compared to type I (ATI) cells, but the relative fraction of ATII cells expressing transgene was comparable to that of AAV5 and 9 [77]. To enhance the transduction of lung parenchymal cells, such as ATI and ATII cells, via the systemic route, viral vectors must traverse pulmonary endothelium. Collectively, enhanced transduction of ATII cells is likely attributed to the ability of AAV9.452sub.LUNG1 to interact preferentially with and penetrate pulmonary endothelium rather than specific ATII cell tropism. Indeed, Cre recombinase expression in Tek-Cre mice is driven by endothelial-specific promoter/enhancer [173], underscoring that AAV9.452sub.LUNG1 may have been enriched in pulmonary endothelial cells. AAV2 variants harboring random 7-mer peptide library were also screened over 4 – 5 rounds for lung endothelial targeting following intravenous injection, which revealed ESGHGYF as a lead lung targeting peptide displayed on the AAV capsid [174]. This AAV2 variant with ESGHGYF mediated strong and lung-specific transgene expression, primarily in endothelial cells throughout the entire pulmonary microvasculature, whereas wild-type AAV2 mediated expression mainly in the liver [174]. The study did not determine whether the variant mediated transgene expression in other cell types in the lung. In an early study, pulmonary endothelial re-targeting strategy was implemented also on AdV which naturally accumulate in liver. Specifically, AdV surface was decorated with antibody against angiotensin converting enzyme (ACE), a membrane bound ectoenzyme highly expressed on pulmonary endothelium [175]. This ACE-targeted AdV exhibited 20-fold increase in both DNA localization and transgene expression in rat lungs compared to untargeted AdV following systemic administration with 80% reduction in liver accumulation [175]. More recently, ACE-targeted AdV expressing small hairpin RNA against tryptophan hydrocylase-1 (Tph1) was shown to knockdown Thp1 in pulmonary arterial endothelial cells and attenuate hypoxia-induced pulmonary hypertension in rats following systemic administration [176].

A 30-mer synthetic peptide based on a glutamic acid-alanine-leucine-alanine peptide repeat, GALA, was originally designed to mimic the function of the influenza viral protein, hemagglutinin, which promotes the acidic pH-dependent destabilization of endosomal membrane [177,178]. However, the peptide was also expected to function as a targeting ligand against sialic acid-terminated sugar chain [177] present on the lumen of pulmonary endothelium [179]. Thus, liposomes with surface GALA decoration were prepared for anti-CD31 siRNA delivery and shown to accumulate and mediate gene knockdown in pulmonary endothelium following systemic administration, thereby eradicating lung metastasis in a tumor model [180]. More recently, this formulation, unlike unmodified liposomes, demonstrated the ability to traverse the endothelial cells via transcytosis and reach ATI, ATII, and alveolar macrophages following systemic administration; flow cytometric analysis revealed that GALA-decorated liposomes were accumulated in 70% and 30% of pulmonary endothelial and ATI cells, respectively [181]. As a result, systemic administration of the formulation carrying anti-podoplanin siRNA reduced the respective mRNA level approximately by 30% in ATI cells [181]. Systemic administration of mRNA-loaded LNPs predominantly accumulate in liver [182]. To alter this fate, Parhiz et al. conjugated monoclonal antibody against platelet-endothelial cell adhesion molecule-1 (PECAM-1 or CD31) to LNPs and investigate the effect of the modification on the particle biodistribution [183]. Systemic administration of the PECAM-1-targeted LNPs resulted in profound inhibition of hepatic uptake while enhancing the mRNA delivery to and expression in the lung by ~200- and 25-fold, respectively, compared to non-target counterparts [183]. Overall, various molecular targeting strategies have validated the concept of pulmonary retargeting and hepatic detargeting of viral and non-viral gene vectors.

3.2.2. Electrostatic lung accumulation

The lung endothelium glycocalyx is enriched in negatively charged sialic acids [179] which can promote its electrostatic trapping of circulating nucleic acid carriers managed to reach the pulmonary endothelium. Stiles et al. engineered an AAV5 mutant by inserting two lysine residues into the variable loop VIII of the capsid to test their hypothesis that the addition of positively charged amino acid residues would enhance AAV accumulation in the lung following systemic administration [184]. This AAV5 mutant, named AAV5-PK2, was then shown to provide significantly greater lung accumulation compared to AAV2 and 5 while exhibiting two orders of magnitude decreased liver accumulation compared to AAV5 [184].

The concept of electrostatic lung accumulation was also employed to lipid-based nucleic acid carriers, including lipoplexes and LNPs [185–187]. Positively charged lipoplexes composed of 1,2-di-O-octadecenyl-3-trimethylammonium propane (DOTMA) and DOPE exhibited reporter mRNA expression predominantly in the lungs of wild-type mice with minimal expression in the liver and spleen [185]. Interestingly, gradual decrease of the cationic lipid (i.e., DOTMA) content shifted the expression from the lung towards the spleen [185]. Likewise, positive charged LNPs comprising a cationic lipid, DOTAP, was shown to mediate sequential mRNA-mediated Cre recombinase production and reporter TdTomato expression in CD31-positive pulmonary endothelial cells in Ai14 mice following systemic administration [186]. Systemic administration of positively charged siRNA-carrying lipoplexes based on DOTAP or another cationic lipid, dimethyldioctadecylammonium bromide (DDAB), were also shown to accumulate and mediate gene knockdown in lung-metastasized tumors [187]. However, the enhanced permeability and retention effect may have contributed to the observation given the leaky nature of tumor vasculature [188]. More recently, Chen et al. introduced a strategy termed selective organ targeting (SORT) where increasing the molar composition of permanently cationic lipids with high pK_a, such as DOTAP and DDAB, shifted accumulation of systemically administered LNPs from liver to spleen and eventually to lung [189]. However, SORT LNPs comprising an ionizable cationic lipid with a relatively low pK_a, 1,2-dioleoyl-3-dimethylammonium-propane (DODAP), demonstrated enhanced liver delivery of mRNA payloads [189], similar to conventional LNPs comprising other ionizable cationic lipids [182]. Adsorption of apolipoprotein E (ApoE) on these conventional LNPs has been shown to drive their predominant liver accumulation via interactions with low-density lipoprotein receptors abundant on the surface of hepatocytes [190]. However, the surface of SORT LNPs with high cationic lipid content was found to be enriched with vitronectin rather ApoE after incubation in mouse plasma [182], suggesting potential role of distinct protein corona on lung targeting of these LNPs beyond the electrostatic effect.

3.2.3. Material screening

PEGylated PBAE NPs formulated with PBAE polymers and 7% PEG-lipid, unlike non-PEGylated counterparts, were shown to mediate robust reporter mRNA expression primarily in the lung to a similar extent as jetPEI following systemic administration [191]. Built upon this study, Kaczmarek et al. synthesized multiple PBAE polymer variants, formulated mRNA- or pDNA-carrying NPs at a fixed 7% PEG-lipid, and conducted a series of in vitro and in vivo screening [192]. In this study, two distinct formulations based on different PBAE polymer variants with structural resemblance, including DD-90-C12–103 and DC-90-C12–103, were identified to exhibit the most efficient reporter mRNA and pDNA expression, respectively, in the lung following systemic administration [192]. A library of poly(trimethylolpropane allyl ether-co-suberoyl chloride), in conjunction with a varying degree of surface coating with Pluronic^® F127, was also sequentially screened in vitro and in vivo to determine a lead formulation that provides efficient systemic delivery of mRNA payloads to the lung [193]. Of note, Pluronic^® F127 is a triblock copolymer composed a hydrophobic segment flanked by two PEG arms that endows NPs with surface PEG coatings via physical adsorption of the central hydrophobic segment on the particle surface [194]. One formulation composed of a polymer variant, PE4k-A17–0.33C12, and 5% F127 was revealed to mediate most efficient and predominant reporter mRNA expression in the lung [193]. In these studies, increasing the surface PEG coatings compromised that ability of the formulations to provide efficient reporter mRNA expression in the lung [191,193]. These findings suggests that excessive masking of positively charged surface of cationic polymer-based nucleic acid carriers with PEG may not be desired for their systemic delivery to the lung. Of note, the mechanisms by which the lead polymer variants exhibited superior performances over others warrant further investigation.

3.2.4. Red blood cell hitchhiking

Inspirated by the findings that naturally occurring biological entities are shuttled by circulating red blood cells (RBCs), RBC hitchhiking strategy was applied to drug delivery NPs [195]. Anselmo et al. mixed NPs with RBCs ex vivo, which caused NP adsorption onto the RBCs, and serendipitously found that the RBC-bound NPs accumulated in the lung following systemic administration (Figure 3) [196]. The observation was attributed to the dislodgement of NPs from RBCs due to the high shear stress experienced by the circulating RBCs in narrow pulmonary capillaries [197]. The strategy was recently employed to viral vector where AAV9 anchored onto RBCs was predominantly delivered to and mediate efficient transduction in the lung [198]. Quantitatively, RBC-anchored AAV9 (RBC-AAV9) exhibited 4 – 5-fold enhancement of transgene expression in the lung compared to free AAV9 following systemic administration [198]. Given the abundance of negatively charged sialylated glycoproteins of the RBC membrane [199], RBC hitchhiking strategy is likely well suited for systemic delivery of positively charged non-viral nucleic acid carriers, which can readily adsorbed onto RBC surface via electrostatic interactions, to the lung.

Figure 3. RBC hitchhiking enables preferential delivery of AAV to and reporter transgene expression in mouse lungs following systemic administration.

(A) Biodistribution of AAV at different time points after intravenous administration. (B-D) Luciferase transgene expression in mouse lungs following a single-dose intravenous administration. (B) Schematic showing the administration and assessment schedule of this single-dose study. (C) Quantification of the luciferase transgene expression by bioluminescence in and (D) In Vivo Imaging System (IVIS) images of mouse lungs, 40 days after the administration. (E-G) Luciferase transgene expression in mouse lungs following a two-dose intravenous administration. (E) Schematic showing the administration and assessment schedule of this two-dose study. (F) Quantification of luciferase transgene expression by bioluminescence in and (G) IVIS images of mouse lungs, 59 days after the first administration (i.e., 31 days after the second administration). *p < 0.05, **p < 0.01, and ****p < 0.0001 (Student’s t-test). This figure has been adopted and modified from [198] which has been distributed under a CC BY 4.0 License.

3.3. Strategies to tackle immune barriers

As introduced earlier, nucleic acid carriers administered via respiratory or systemic route encounter immune barriers that compromise their performances en route to or in the lung. We here introduce strategies implemented on nucleic acid carriers to overcome the primary immune barriers, including lung-resident macrophages (if not desired target cells) and therapy-inactivating antibodies.

3.3.1. Resistance to phagocytic uptake by lung-resident macrophages

Positive surface charges promote recognition and phagocytosis of nucleic acid carriers by lung-resident macrophages [87], which is a particular concern for carriers based on cationic materials. PEGylation is arguably the most extensively explored strategy to minimize phagocytic uptake of drug delivery NPs by macrophages of various origins and types, particularly those encountered by systemically administered NPs [139]. However, reports that have directly investigated the effect of PEGylation on the ability of inhaled nucleic acid carriers to resist phagocytosis in the lung are rare. One study showed that PEGylation markedly reduced the uptake of inorganic nucleic acid delivery NPs by mouse alveolar macrophage cells in vitro, which likely contributed to the enhanced transgene expression compared to non-PEGylated counterparts following intratracheal administration [147]. Alternatively, surface decoration with the “marker of self” membrane protein CD47 or its synthetic peptide analogue has been shown to reduce macrophage-mediated clearance and thus to prolong circulation of systemically administered NPs [200]. However, this phagocytosis-evading approach is yet to be validated for inhaled therapeutic delivery applications, including nucleic acid delivery. Of note, both approaches should be fine-tuned or combined with a lung-targeting strategy for systemic nucleic acid delivery to the lung to circumvent persistent circulation that lacks lung accumulation.

3.3.2. Avoidance of antibody neutralization

Therapy-inactivating neutralization by pre-existing or induced vector-specific antibodies is one of the greatest challenges to inhaled gene therapy with viral vectors regardless of administration route. Two primary types of strategies that address this issue include depletion of and evasion from antibodies. The former strategy involves plasmapheresis [201,202], enzymatic antibody degradation [203,204], or gene editing-based inhibition of antibody production [205]. Here, we introduce a couple of examples for the latter strategy, which include capsid modification and biomimetic camouflage.

3.3.2.1. Capsid modification

To minimize antibody neutralization, 127 different mutations were introduced at 64 positions on the external surface of AAV2 within the regions expected to bind antibodies [206]. The study revealed mutations at six positions, including Q263, S264, S384, Q385, E548, and V708, reduced binding or neutralization by a moue anti-AAV monoclonal antibody, A20, with two particular mutations, S264A and V708K, exhibiting the greatest effect against both binding and neutralization [206]. In parallel, two mutations, R471A and N587A, were found more resistant to neutralization by all three different AAV2-neutralizing antisera tested in this study compared to wild-type AAV2 [206]. Of note, two mutations resistant to A20, E548A and V708K, were also resistant to one or two of the sera [206]. Likewise, AAV2 capsid was modified by inserting a 14-mer peptide, QAGTFALRGDNPQG, at either of two different positions in the surface exposed loop regions, aa534 and aa573, which reduced the viral affinity to AAV antibodies up to 70% compared to wild-type AAV2 [207]. As an additional means to discover antibody-evading AAV mutants, directed evolution strategy was employed against neutralizing serum on an error-prone PCR-based AAV2 library [208]. In this study, four distinct mutants comprising T716A mutation were shown to provided 3-fold greater transduction compared to wild-type AAV2 in presence of serum [208]. More recently, a structure-guided evolution approach was applied to the mouse antigenic epitopes on AAV1 capsid to yield one lead variant, named CAM130, that effectively evaded polyclonal anti-AAV1 neutralizing sera from immunized mice [209]. The variant also displayed robust immune evasion in non-human primate and human serum samples at dilution factors as high as 1:5, meeting the criterion mandated in several clinical trials [209].

3.3.2.2. Biomimetic camouflage

In addition to the capsid modification strategy, AAV vectors shuttled by non-immunogenic biological entities have been shown to resist antibody neutralization. For example, EV-associated AAV (EVAAV) based on various serotypes, including AAV1 [210], 6 [211], and 8 [212], demonstrated greater resistance against neutralizing antibodies compared to respective EV-free AAV vectors. The RBC hitchhiking strategy introduced above was also shown to endow AAV vectors with the ability to evade antibody neutralization (Figure 3) [198]. Specifically, a prior treatment with AAV9 or RBC-AAV9 did not mitigate the ability of subsequently administered RBC-AAV9 to mediate reporter transgene expression in the lung following systemic administration [198]. A subsequent mechanistic study showed that anchoring AAV to RBCs did not significantly impact the AAV-specific humoral immune response [198], suggesting that the immune evasion may be attributed to delayed or reduced recognition of AAV vectors anchored on RBCs by neutralizing antibodies.

4. Conclusion

In this review, we sought to provide a concise yet comprehensive overview of lung gene therapy, primarily focusing on key biological barriers and strategies to overcome them to achieve therapeutically relevant gene transfer efficacy. We introduced the nature of biological barriers found in respiratory and/or systemic administration routes and discussed the innovation in delivery strategies spawned from the accumulated knowledge and historical scientific data. Since the first clinical trial in 1993 [213], we have endured a long tunnel of clinical disappointments but have made significant progress over the past three decades. In the meantime, gene therapy has become a viable therapeutic option for a few non-lung genetic disorders over the past few years, and clinical lung gene therapy investigation has been rekindled with an advent of advanced delivery strategies. We now embrace unprecedented hope for clinical translation of lung gene therapy, and if realized, the impact would be enormous given the unmet medical needs for numerous incurable chronic lung diseases.

5. Expert Opinion

A majority of the strategies explored to enhance lung cell transduction/transfection to date have targeted differentiated airway epithelial cells. However, these terminally differentiated airway cells have defined lifespan [214,215], thus the lifetime of gene therapy, necessitating repeated treatments for lifelong gene therapy of chronic lung diseases. While a newly engineered viral vector has shown promise in transducing airway cells in a clinical study [45], an initial treatment inevitably results in the induction of vector-specific immune responses, which renders subsequent treatments inefficacious. Non-viral vectors lacking immunogenicity [216], coupled with pDNA payload driven by a long-acting promoter, may serve as an alternative. Another means to address this limitation would be targeting multipotent progenitor basal cells that possess capacity to differentiate into airway parenchymal cells [217]. Specifically, permanent correction of basal cells with integrating LVs or gene editing techniques likely leads to continuous production of functional airway epithelium and effort to realize this goal has been commenced [218]. Of note, these basal cells are located underneath the airway basal membrane [219] and thus a means to breach the airway epithelial barrier from the apical surface would be needed or an access via the systemic route could be considered.

Above-mentioned high-throughput in vivo screening techniques, such as directed evolution of AAV [120] and cluster-based iterative screening of LNPs [155], constitute a powerful tool to identify nucleic acid carriers capable of overcoming a serious of biological barriers regardless of the administration route. However, these studies are essentially pursued with animal models, and thus their translatability in humans have been questioned [220,221]. Indeed, various AAV serotypes were previously shown to mediate distinct relative transduction efficiencies in airway epithelium from different species (e.g., rodent vs. human) [100]. More recently, an AAV variant identified to efficiently transduce pig airway epithelium failed to do so in human airway epithelium [122]. To this end, cross-validation of nucleic acid carrier candidates discovered via an in vivo screening with in vitro models, such as ALI culture [61], lung organoids [222], and lung-on-a-chip [223], based on primary human cells would enhance the relevance of those carriers to clinical translation.

As introduced earlier, PEGylation is a versatile strategy that has been widely employed to enhance the delivery performance of nucleic acid carriers for lung gene therapy. We note that the extent to which the approach should be implemented varies with the administration route. Specifically, it is now well appreciated that highly dense surface PEGylation is desired to avoid adhesive interactions with airway mucus and PCL after respiratory administration by rendering the carrier surface hydrophilic and near-neutrally charged [139]. On the other hand, less PEGylated, and thus positively charged, nucleic acid carriers were found accumulated in the lung via the systemic route in multiple independent studies [185–187,189,191,193]. The findings underscore that optimal surface PEGylation minimizing liver accumulation while promoting lung accumulation, rather than persistent circulation, after systemic administration may exist. Of note, excessive PEGylation may facilitate premature removal of systemically administered nucleic acid carriers by the accelerated blood clearance phenomenon [224,225].

Article highlights.

• Clinical trials of lung gene therapy to date have relied predominantly on administration via the respiratory route, but recent preclinical studies have highlighted lung-targeted strategies following systemic administration.

• Nucleic acid delivery carriers administrated via respiratory or systemic route encounter distinct or shared biological barriers.

• One of the most widely investigated strategies to overcome biological barriers is surface modification of delivery carriers to alter their interactions with biological entities and/or media.

• Lifelong gene therapy can be achieved by repeated treatments of target cells with carriers with negligible immunogenicity or targeted gene editing in multipotent progenitor basal cells.

• High-throughput screening techniques warrant their implementation on and/or validation in clinically relevant settings for clinical translation and development of newly identified nucleic acid delivery carriers.

• The surface modification of nucleic acid delivery carriers with polyethylene glycol is a versatile means to enhance the delivery efficiency but the approach should be employed in a case-sensitive manner depending on the route of the administration.