Recent Trends in the Delivery of RNA drugs: beyond the liver, more than vaccine

Abstract

RNAs are known for versatile functions and therapeutic utility. They have gained significant interest since the approval of several RNA drugs, including COVID-19 mRNA vaccines and therapeutic agents targeting liver diseases. There are increasing expectations for a new class of RNA drugs for broader applications. Successful development of RNA drugs for new applications hinges on understanding their diverse functions and structures. In this review, we explore the last five years of literature to understand current approaches to formulate a spectrum of RNA drugs, focusing on new efforts to expand their applications beyond vaccines and liver diseases.

Graphical Abstract

1. Introduction

Nucleic acids have been envisioned as a drug for decades. The development has faced early setbacks due to the risks associated with viral vectors [1, 2]. Meanwhile, the challenges prompted the development of non-viral carrier systems, and the focus has shifted from DNA to RNA, eliminating the risk of insertional mutagenesis. RNAs are now recognized for diverse functions relevant to treating infectious diseases, cancers, and genetic disorders. With the advent of several RNA drugs, including COVID-19 mRNA vaccines, RNA therapeutics are gaining explosive interest from the public, academia, and industry.

As RNA biology continues to advance, bringing new discoveries like CRISPR-Cas9 technology, expectations are soaring for a new class of RNA drugs applicable to a broader spectrum of diseases. The cornerstone of this development lies in effective delivery strategies tailored to the structures and functions of RNAs. The drug delivery field has come a long way to meet these requirements, bringing various approaches, from RNA modification and ligand conjugation to nanoparticle encapsulation. Nevertheless, the field of therapeutic RNAs awaits a breakthrough to enable the delivery to organs beyond the liver and address diverse diseases that respond suboptimally to traditional drugs.

To understand the current state of the art and identify new opportunities in this quest, we have reviewed the literature concerning RNA drug delivery published in the past five years. This article does not intend to be an exhaustive summary of each delivery technology, which has been thoroughly discussed elsewhere [3-11]. Instead, we overview different classes of RNA pursued as drugs (Fig. 1) and survey emerging delivery technologies as well as the current efforts to expand the applications of RNA drugs.

Fig. 1. Journey of RNA drugs to cell targets.

Therapeutic RNAs in varying sizes and conformations have diverse bioactivities. RNA drugs are chemically modified or encapsulated in natural or synthetic nanoparticles. Some of the delivery barriers have been overcome by these formulation strategies, but the control of biodistribution and delivery to intracellular targets remain challenging. Created with BioRender.com.

2. RNA as a drug

2.1. Antisense oligonucleotide (ASO)

ASOs are single-stranded DNA (ssDNA) or single-stranded RNA (ssRNA), 15-25 nucleotide long, which base-pair with complementary mRNAs to modify protein expression [12, 13]. An ASO binds to its target mRNA, forms an RNA-DNA heteroduplex, and prompts ribonuclease (RNase) H to cleave the RNA part of this duplex [14]. This process leads to the degradation of the target mRNA, thereby reducing the production of the associated protein [14]. Alternatively, ASOs may block ribosomal binding to mRNA, destabilize a precursor form of mRNA (pre-mRNA), or disrupt pre-mRNA splicing [15]. Since unmodified ASOs are prone to enzymatic degradation and undergo fast systemic clearance, they are modified on the phosphodiester backbone, 2’-OH of sugar moiety, or nucleobase to improve the stability and half-life [10]. Currently, ten ASO therapies were approved by the U.S. Food and Drug Administration (US FDA) for various indications, including Duchenne muscular dystrophy and spinal muscular atrophy [16]. ASO therapies are also actively investigated for anticancer applications in clinical trials and preclinical studies [16].

2.2. MicroRNA (miRNA)

miRNAs are small non-coding ssRNA molecules, typically 21-23 nucleotide long [17]. They regulate gene expression by controlling mRNA translation. miRNA in the cytoplasm is loaded onto the RNA-induced silencing complex (RISC) to guide its binding to the complementary mRNA, leading to mRNA inhibition or degradation [18]. The dysregulation of miRNA is implicated in various diseases, including cancer, hepatitis C infection, and cardiovascular diseases [19]. miRNAs promoting tumor proliferation and metastasis, such as miR-21 and miR-155 [19-21], called oncomiR, are targeted by an anti-miRNA antisense oligonucleotide or an artificial mRNA decoy [19, 22]. On the other hand, tumor-suppressive miRNAs such as miR-34, miR-200, and let-7 are downregulated in cancer cells, thus allowing cancer proliferation [23-27]. When the downregulation of miRNA contributes to the disease progression, synthetic miRNA is complemented to attenuate the process.

2.3. Small interfering RNA (siRNA)

siRNA is a short double-stranded (dsRNA), 20-25 base pairs in length, which suppresses specific gene expression by inducing mRNA degradation. siRNA is split from long dsRNA precursors and incorporated into the RISC, which binds and cuts the target mRNA, impeding its translation [28]. Due to its high efficiency and specificity, siRNA has been actively pursued as a drug to treat cancers and viral infections [29]. In 2018, the US FDA approved the first siRNA product, patisiran, indicated for adult individuals suffering from polyneuropathy caused by hereditary transthyretin amyloidosis (hATTR). Since then, four more siRNA agents, namely givosiran (2019), lumasiran (2020), inclisiran (2021), and vutrisiran (2022), have received the FDA approval. Givosiran is indicated for adult patients diagnosed with acute hepatic porphyria, lumasiran for pediatric and adult patients with primary hyperoxaluria type 1, inclisiran for adult patients with heterozygous familial hypercholesterolemia or clinical atherosclerotic cardiovascular disease [6, 30], and vutrisiran for hereditary transthyretin-mediated amyloidosis [31].

2.4. RNA aptamers

RNA aptamers are short ssRNA oligonucleotides with a length of 56 to 120 nucleotides that act similarly to monoclonal antibodies [32, 33]. RNA aptamers consist of two distinct regions: a variable region positioned in the center, with a length from 20 to 80 nucleotides, and constant regions located on both sides, approximately 18 to 20 nucleotides each [32, 33]. Once in the cells, RNA aptamers form a folded structure that binds to the disease-associated target proteins with high affinity and specificity [32-34]. RNA aptamers offer the advantages of both RNAs and antibodies, the flexible conformation and the functional binding characteristics, respectively. Furthermore, RNA aptamers may be connected to other aptamers or therapeutic RNAs to target disease-associated cells [32]. In 2004, the US FDA approved Pegaptanib (Macugen), a polyethylene glycol-modified (PEGylated) RNA aptamer targeting isoform 165 of vascular endothelial growth factor A (VEGF-A), for the treatment of age-related macular degeneration (AMD) by intravitreal injection, the first RNA biotherapeutic in clinic [32].

2.5. RNA-based immune adjuvants

dsRNA and ssRNA are pathogen-associated molecular patterns (PAMPs) recognized by the pattern recognition receptors (PRRs) of innate immune cells [35], such as endosomal toll-like receptors (TLRs), retinoic acid-inducible gene-I (RIG-I)-like receptors (RLRs), absent in melanoma 2 (AIM2)-like receptors (ALRs) [35, 36]. For example, synthetic analogs of dsRNA, polyinosinic:polycytidylic acid (polyI:C, 200-1500 base pairs [37]), activate TLR3 [38] as well as cytosolic receptors RIG-I and MDA-5. Guanosine- or uridine-rich ssRNA activates TLR7/8 [39]. In using these RNAs as immune adjuvants, a critical challenge is to target antigen-presenting cells (APCs) and prevent nonspecific interactions with non-hematopoietic cells.

2.6. Messenger RNA (mRNA)

Messenger RNA (mRNA) is a ssRNA that carries coding information for protein synthesis. mRNA is translated in the cytosol and, thus, avoids the need for nuclear transport and the risk of insertional mutagenesis linked to DNA [9]. In vitro-transcribed (IVT) mRNA is a synthetic version of mRNA developed for therapeutic applications. IVT mRNA was developed in 1984 [40] and applied in vivo in 1990 [41]; since then, several derivatives of IVT mRNA, such as self-amplifying mRNA, trans-amplifying RNA and circular mRNA, have been developed to reduce the immunogenicity and increase the protein expression efficiency [42]. Conventional IVT mRNA is derived from cyclic plasmid DNA (pDNA), linearized enzymatically and amplified by the polymerase chain reaction. It consists of a 5′ cap, 5′ and 3′ untranslated regions (UTRs), an open reading frame (ORF), and a poly(A) tail [43]. On the other hand, self-amplifying mRNA has a gene encoding replicase within the ORF, which allows for autonomous multiplication and extended protein expression [44-46]. Trans-amplifying RNA comprises two mRNA strands, one with a replicase-encoding gene and the other with a gene of interest [47, 48]. Circular mRNA is generated from a linear mRNA precursor through an intramolecular ring-forming reaction. With the closed-loop structure, circular mRNA is resistant to exonucleases and, hence, more stable than linear mRNA. mRNA has been used to encode viral antigens for vaccine applications, including SARS-CoV-2, and is actively investigated for tumor vaccines in clinical settings [49, 50]. Moreover, mRNA has been designed to encode cytokines and tumor suppressors to modulate the tumor microenvironment (TME) for cancer immunotherapy [42].

2.7. CRISPR-Cas9

The clustered regularly interspaced short palindromic repeats (CRISPR) - CRISPR-associated (Cas) proteins is a genome editing technology derived from a natural immune defense mechanism of bacteria [51, 52]. Bacteria combat viral infection by capturing fragments of viral DNA and incorporating them into their DNA in a genetic memory bank called CRISPR arrays. In the event of subsequent viral attacks, the bacteria generate RNA segments from the CRISPR arrays, which bind to specific regions within the viral DNA. Subsequently, the bacteria employ a Cas enzyme to cleave the viral DNA and inactivate the virus [51, 52]. Cas9 has been the most widely used enzyme in gene editing applications.

A classic CRISPR-Cas9 system involves a dual-guide RNA and Cas9 enzyme. The dual-guide RNA is composed of a CRISPR RNA (crRNA, 42 nucleotides), complementary to the target DNA, and a trans-activating crRNA (tracrRNA, 75 nucleotides) [53], which triggers Cas9 to cleave DNA. Together, the dual guide RNA directs Cas9 endonuclease to introduce double-strand breaks in the target DNA, like the bacterial RNA segments from the CRISPR arrays do to the viral DNA [53]. Following the DNA cleavage, the DNA repair machinery joins the DNA ends, resulting in gene disruption, or introduces a custom-designed DNA sequence as a replacement [54]. For the simpler application of CRISPR-Cas9, the crRNA and tracrRNA may be fused to make a single-guide RNA (sgRNA, ~130 nucleotides) [53]. Due to the versatility of designing the guide RNA and the precision of targeting a specific DNA sequence, CRISPR-Cas9 has quickly become the leading genome-editing tool. CRISPR-Cas9 may be designed in three formats [55, 56]: (i) a ribonucleoprotein complex of Cas9 protein (160 kDa) and sgRNA [57, 58]; (ii) a plasmid DNA encoding Cas9 enzyme and sgRNA [59]; and (iii) a combination of Cas9 mRNA (4300 nucleotides) and sgRNA [60].

3. RNA formulations

RNA drug development faces several unique challenges, such as enzymatic instability, proinflammatory activity (in non-vaccine applications), and unfavorable biopharmaceutical properties due to the size and charge. RNA is vulnerable to RNases and activates the innate immune cells via PRRs, which are naturally evolved defense mechanisms against foreign RNAs [11]. Moreover, the highly negative charge subjects RNAs to rapid clearance from the circulation; coupled with the large size (ranging from 14 kDa of siRNA to ~1.5 MDa of CRISPR-Cas9 RNA combinations), the negative charge further limits RNA transport into the cells (Fig. 1).

The enzymatic instability and proinflammatory activity have been tackled by chemical modification of RNA. Strategies to improve RNA stability and overcome innate immunogenicity, a topic for a separate review on its own [6, 8, 10, 61], include the modification of the phosphodiester backbone (phosphorohioate, methylphosphonate, or phosphotriester), nucleobase (pseudouridine, 1-methylpseudouridine, or 5-methylcytidine), or 2’-OH of the ribose moiety (2’-O-methyl, 2’-methoxyethyl, or 2’-fluoro). The significance of these efforts has been recognized by the 2023 Nobel Prize in Physiology or Medicine, awarded to Kariko and Weissman for their discovery of nucleoside base modifications [62, 63].

For relatively small RNAs, the chemical modification can address some of the biopharmaceutical challenges, enabling parenteral delivery without additional carriers. For example, ~20 nucleotide ASOs with a phosphorothioate backbone become hydrophobic enough to bind serum albumin to cross the lipid bilayer and escape the endosome [11]. However, large or double-stranded RNAs rely on various delivery approaches, ranging from ligand conjugates to nanoparticles (Fig. 1). Representative nanoparticle carriers include lipid-based formulations, such as liposomes or lipid nanoparticles (LNPs). Liposomes are vesicles of lipid bilayers made of cholesterol and one or more phospholipid, which encapsulate aqueous cores [64, 65]. LNPs consist of cholesterol, a helper phospholipid, a PEGylated lipid, and an ionizable amine-containing lipid [66]. Unlike liposomes, LNPs have solid cores in which the lipids and RNA form a worm-like or a disordered inverse hexagonal structure [67, 68]. These lipid-based nanoparticles as well as other commonly used delivery approaches, including the approved products and those in the pipeline, are discussed in several review articles [3-9, 69]. Not intending to replicate the existing efforts, we introduce the approaches published in the last five years in the following.

3.1. miRNA or anti-miRNA

An early effort to deliver miRNA (or anti-miRNA) employs liposomes with optional modification with a ligand that can improve target cell interaction [70]. An amphoteric liposome – cationic at acidic pH for complexation with miRNA and anionic in neutral pH to enable systemic circulation [71] – was developed for systemic delivery of a tumor suppressor miR-34a to the liver (and liver tumors). The liposomal miR-34a was found to be effective in suppressing tumor growth in two mouse models of orthotopic liver [72]; nevertheless, the first-in-human phase I study was terminated early due to fatal immune-related adverse events [73].

Derivatives of cationic polymers have been used for miRNA delivery to tumors, showing favorable preclinical outcomes [74-76]. Monomethoxy(polyethylene glycol)–poly(d,l-lactide-co-glycolide)–poly(l-lysine) triblock copolymer (mPEG–PLGA–PLL) was used to co-deliver paclitaxel (PTX) and miR-7 for systemic tumor therapy. Here, miR-7 was included to downregulate the epidermal growth factor receptor (EGFR) and extracellular signal-regulated kinase (ERK) pathway due to PTX [75]. In another study, polyamidoamine G5 dendrimer was complexed with a self-assembled RNA triple helix, comprising a miRNA mimic duplex and an anti-miRNA, which restore tumor suppressor function and inhibit oncogenic miRNA, respectively [76]. The 50 nm electrostatic complex of polyamidoamine dendrimer and RNA triple helix mediated the RNA uptake via macropinocytosis. Upon local delivery to solid tumors via a dextran hydrogel, the RNA-dendrimer complex suppressed tumor growth to a greater extent than the hydrogels with conventional chemotherapeutic agents, showing a gene expression profile consistent with each miRNA activity in the tumor [76].

miRNAs are also delivered by non-cationic carriers with modifications to attract the RNA. Hyaluronan sulfate-calcium complex was used to deliver a miR-21 mimic to macrophages for an anti-inflammatory effect [77]. Systemically administered, the NPs accumulated in the infarcted heart via interactions with macrophages, promoted the secretion of anti-inflammatory cytokines, enhanced angiogenesis, and suppressed adverse ventricular remodeling compared to the blank NPs [77]. Another example is mesoporous silica nanoparticles (MSNs) modified with amines [78, 79]. Additional carriers, such as cyclodextrin/PEG hydrogel, were used for local delivery of the miR to the heart [78] (Fig. 2a, 2b), or iRGD-conjugated lipid coating for systemic delivery to integrin-expressing tumors (Fig. 2c, 2d) [79]. Other inorganic NPs comprising SiO₂, CaO, and P₂O₅ have also been considered for miRNA delivery, where an osteoblast-promoting miRNA (miRNA-5106) was loaded via the interaction with Ca²⁺ and achieved greater transfection efficiency than lipofectamine or polyethylenimine [80]. A derivative of these inorganic NPs, clustered via a polyethylenimine template, afforded 19 times higher loading of miRNA-5106 than the NPs due to the large pore size and was shown to be effective when surgically implanted in the rat cranial defect model [81].

Fig. 2. miRNA delivery with amine-modified MSNs.

(a) Schematic illustration of miRNA loading into amino (−NH₂) and trimethylamine (−N(CH₃)₃, TMA)-co-modified MSNs, and the preparation of the dual crosslinked hydrogel through the hydrophobic interactions (step 1) and Schiff base formation (step 2), which could selectively release the MSN/miR-21-5p complex at pH 6.8 (step 3). (b) The efficacy of miR-21-5p loaded in MSN and hydrogel evaluated in a porcine model of myocardial infarction. Representative images of hearts with infarct areas and tissue necrosis, histological analysis, and quantitative analysis of infarct area, fibrotic area, and miRNA transfection efficiency analyzed 28 days after the treatment. Reprinted with permission from [78]. Copyright (2021) AAAS. (c) Schematic illustration of iRGD-modified MSN for combination delivery of siPlk1 and miR-200c. The cytosolic release of the cargo is promoted by the reactive oxygen species generated by the photothermal therapy in the endosome. (d) Treatment schedule, tumor growth curve, and bioluminescence images of mice with metastatic tumors. Antitumor activity of nanoparticles in the orthotopic MDA-MB-231 breast cancer model in NCG mice. Mice were treated IV twice a week for 3 weeks (si-Plk1: 1 mg/kg; miR-200c: 1 mg/kg ; ICG: 720 μg/kg). Mice were treated with ICG co-loaded with si-Plk1 and/or miR-200c and/or negative RNA (NC) loaded in iRGD-modified MSN (iMSN) or regular MSN. Reprinted with permission from [79]. Copyright (2020) American Chemical Society.

Another popular carrier is exosomes, a natural carrier of miRNA and other small RNAs [82]. Exosomes are a subset of extracellular vesicles with a diameter of 40 to 160 nm, originated from endosomes and secreted into the extracellular space, and ferry various intracellular components (including miRNA) to other cells [83]. Given the origin and function, exosomes make a carrier of choice for therapeutic miRNA delivery. In contrast to synthetic carriers, exosomes face minimal immune clearance and take advantage of natural ligands for receptor-mediated tissue targeting [83]. Cells are coaxed by soluble factors or genetic modification to enrich specific miRNA of interest in exosomes [84] (a topic outside the scope of this review). Nevertheless, the reproducibility, scale-up, and inclusion of unknown bioactive molecules remain a challenge [85, 86]. Therefore, exosomes are obtained from relatively well characterized cells, or derivatives of exosomes (e.g., cytoplasm-free ghost vesicles) are derived from cell membrane [87]. These vesicles are manipulated post hoc to encapsulate the miRNA or other small RNA of interest. Representative methods include sonication, transfection, or electroporation [88, 89].

For example, miR-124 with pro-neurogenic and angiogenic activity has been delivered to the ischemic cortex in a mouse model using MSC-derived exosomes [90]. Here, the MSCs were engineered to express rabies virus glycoprotein in the exosomes to target neurons, and the miRNA was loaded by electroporation [90]. Electroporation has also been used to co-load 5-fluorouracil and anti-miR-21, which reverses drug resistance, in Her2-targeting exosomes for tumor specific delivery [91]. Alternatively, miRNA has been modified with cholesterol and co-loaded with doxorubicin in disintegrin-metalloproteinase 15-expressing exosomes by hydrophobic interaction with the lipid bilayer [92]. Similarly, anti-miRNA has been integrated into exosomes as a cholesterol conjugate by hydrophobic interaction [93]. Additional components, such as anticancer drug (doxorubicin), superparamagnetic iron oxide, and endosomolytic peptide, have been co-loaded on exosomes via hydrophobic interaction, affinity interaction, and electrostatic interaction, respectively, for systemic application in a mouse tumor model [93].

3.2. siRNA

4 out of 5 approved siRNA products (e.g., givosiran, lumasiran, inclisiran, and vutrisiran) are siRNA conjugates with multivalent N-acetyl-D-galactosamine (GalNac) molecules. GalNac interacts with the asialoglycoprotein receptor, taking siRNA to hepatocytes where the receptor is overexpressed [6]. GalNac-conjugated siRNA has shown a durable gene silencing effect that lasts for months [94], due to the stability in acidic intracellular compartments [95]. LNPs have also been used for siRNA delivery, making the first FDA-approved siRNA product (patisiran, Onpattro^®). Patisiran LNP consists of cholesterol, phospholipid, PEG-lipid as well as an ionizable amino lipid DLin-MC3-DMA (MC3, dilinoleylmethyl-4-dimethylaminobutyrate) with a pKa of 6.4, which allows for systemic administration and endosomal escape [96]. The MC3 LNP in circulation binds to apolipoprotein E (ApoE), which interacts with low-density lipoprotein receptors of hepatocytes [97]. Due to the dominant interactions with hepatocytes, both GalNac-conjugated siRNA and LNP-formulated siRNA products are indicated for liver diseases [6].

In the absence of siRNA products for other target organs, extrahepatic delivery of siRNA has been pursued in preclinical studies. One approach is to replace GalNac with lipid compounds [98-100]. A recent study reports an siRNA conjugate with DLin-MC3-DMA, added to enhance the endosomal escape [101]. Although this conjugate did not avoid liver accumulation, it showed a tendency to accumulate in the vascular compartment of tissues, suggesting an effect of conjugate structure on its intratissue distribution [101]. Alternatively, the surface of LNP is modified with a ligand interacting with specific organs (e.g., a tumor-homing peptide (tLyp-1) for tumor delivery [102]). Derivatives of LNPs have also been produced for extrahepatic delivery of siRNA. NPs made of an ionizable low–molecular weight polymer (7C1) and PEG-lipid delivered siRNA to the endothelium of multiple organs (lung, heart, retina, kidneys) after intravenous infusion in a primate model [103]. Hybrids of polymer and lipid NP were proposed to extend the circulation half-life of siRNA and improve tumor accumulation [104]. A similar polymer–lipid hybrid NPs with a macrophage-targeting peptide (S2P) have been developed to deliver siRNA to the macrophages in atherosclerotic lesions [105].

Our lab has reported another approach to enhance the extrahepatic delivery of siRNA. To avoid lipid or ligand-mediated liver tropism, we developed a nanocapsule based on polydopamine, called “Nanosac” [106]. Nanosac is produced by sequential coating of mesoporous silica nanoparticles (MSN) with siRNA and polydopamine, followed by the removal of the sacrificial MSN core (Fig. 3a). The product is a hollow polydopamine capsule encapsulating siRNA inside (Fig. 3b), which offers two unique features: (i) polyphenol surface, which co-opts serum albumin in circulation to exploit albumin transport mechanism for biodistribution (Fig. 3c), and (ii) softness, which helps navigate through the organs of the reticuloendothelial system and infiltrate into tumors (Fig. 3d, 3e). Nanosac delivering siRNA targeting PD-L1 outperformed a hard counterpart with the MSN core (Fig. 3f) and anti-PD-L1 antibody (Fig. 3g) in a mouse model of CT26 tumor [106].

Fig. 3. Polyphenol nanocapsules (Nanosac) for siRNA delivery.

(a) Schematic illustration of Nanosac preparation. siRNA was loaded on cationized MSN, which was coated with polydopamine (pD), followed by removal of the MSN core. (b) Transmission electron microscope image of Nanosac. (c) The most abundant proteins in the protein corona bound on the MSN^a , MSN^a/pD, or Nanosac, after incubating in 50% fetal bovine serum, analyzed by LC-MS/MS. (d) Young’s moduli of MSN^a/pD and Nanosac measured by atomic force microscopy. (e) Time-lapse intravital microscopic images of Cy5-labeled MSN^a-cy5/pD and Nanosac circulating in CT26 tumor-bearing BALB/c mice. Green: Dextran-FITC (locating blood vessel); and red: cy5-labeled NPs. (f) Antitumor activity of 5% dextrose (D5W), MSN^a/siPD-L1/pD, and Nanosac in CT26 tumor-bearing mice. Mice were treated every two days for ten times (siPD-L1: 0.75 mg/kg/time, IV injection). (g) Antitumor activity of D5W, anti-PD-L1 antibody, and Nanosac in Balb/c mice bearing CT26 tumors. Mice were treated every two days for five times (anti-PD-L1 antibody: 200 μg/mouse/time, intraperitoneal injection; siPD-L1:1.5 mg/kg/time, IV injection). Reprinted with permission from [106]. Copyright (2021) American Chemical Society.

3.3. Immune adjuvant (polyI:C)

PolyI:C has been exploited in several advanced phase clinical trials, such as Hiltonol^®, polyI:C stabilized with poly(L-lysine) and carboxymethylcellulose, for local administration in cancer patients [107], and Ampligen^® (rintatolimod), mismatched polyI:C with additional uracil (U) [39], for systemic application [108, 109]. To improve the access to its intracellular receptors (TLR3, RIG-I, or MDA-5), polyI:C has been delivered via lipid, polymer, or inorganic nanoparticles. As an immune adjuvant, polyI:C is often co-delivered with antigens or drugs inducing antigen production in situ. The diversity and complexity of carrier systems result from the combinations and their respective functions.

For example, polyI:C was co-loaded with tumor antigens in DOTAP (N-[1-(2,3-Dioleoyloxy)propyl]-N,N,N-trimethylammonium chloride)-based cationic liposomes to activate innate immune cells and promote antigen presentation, thereby eliciting antitumor immunity [110, 111]. The liposome surface was further modified with glycocholic acid (bile acid) and mannose to promote intestinal absorption of the liposomes and interaction with dendritic cells, respectively, enabling oral delivery of tumor vaccines [111]. Cationic polymers, synthetic or natural origin, are used as a standalone carrier or a component for co-delivery of poly I:C and antigens [112-115]. Chitosan [114, 115] appear frequently in the literature as a carrier of polyI:C, enhancing antigen-presenting cell uptake of antigens and the activation of antigen-presenting cells.

Non-cationic carriers, such as poly(lactic-co-glycolic acid) (PLGA) NPs, are also used for the delivery of nucleic acid adjuvants [116, 117]. PolyI:C has been co-encapsulated with antigens in PLGA NPs, coated with a cationic derivative of cholesterol and hyaluronic acid, to control the release of the nucleic acid adjuvant and enhance the uptake by antigen-presenting cells, respectively, proving effective as a preventative and therapeutic tumor vaccine [116]. A recent study describes poly I:C encapsulated PLGA NPs conjugated on the surface of bone marrow-derived macrophages (BMDM) to polarize the macrophages and endogenous antigen-presenting cells toward tumoricidal phenotype for antitumor application [117].

As the induction of immunogenic cell death has gained interest in cancer immunotherapy, polyI:C has been co-delivered with other agents inducing cell death. Carriers are designed to accommodate multiple components with distinct properties. For example, poly I:C and photosensitizer chlorin e6 were co-delivered to treat triple-negative breast cancer, where tumors are illuminated to induce photodynamic effects on tumor [112]. Here, chlorin e6 was conjugated to cationic polymer polyethylenimine, which attracted polyI:C, and combined with a phenylboronic acid derivative and polyvinyl alcohol, making reactive oxygen species (ROS)-sensitive ester links in NP structure. The irradiated photosensitizer generated ROS, which cleaved the ester bond to expose negative charges, promoting immunogenic cell death and polyI:C release, respectively. Intravenously administrated, NPs delivered more polyI:C to the tumor than free polyI:C, activating antitumor immune activity, which manifested as prolonged survival time and the abscopal effect on tumors inoculated after the treatment [112]. Another example is the combination of polyI:C and PTX nanocrystals. The nanocrystals were stabilized by cationic β-lactoglobulin, on which poly I:C and mannuronic acid (MA) were loaded via electrostatic interaction. The components play distinct roles: PTX inducing cell death, polyI:C activating cytosolic RIG-I, and MA promoting tumor cell uptake, which led to superior antitumor effect compared to the combination of Taxol and free poly I:C, with the evidence of immunostimulatory effect [118].

3.4. mRNA

Therapeutic delivery of mRNA has been pursued for several decades. Following the success of COVID-19 mRNA vaccines, mRNA therapeutics and its best-known carrier LNPs have received a burgeoning public interest. Of the components of LNP mentioned earlier (ionizable amino lipid, helper phospholipid, cholesterol, and PEG-conjugated lipid), the pH-sensitive ionizable lipids facilitate mRNA encapsulation and endosomal escape, and the other lipids contribute to the LNP stability and size control. The mRNA delivery efficiency is determined by the composition, in particular, the ionizable lipid [119], as well as other formulation factors such as buffer choice [120] and mixing process [121]. Recent review articles provide comprehensive summaries of the evolution, capabilities, and limitations of existing LNPs [61, 122, 123]. Ongoing research in mRNA delivery focuses on extending its utility beyond vaccine applications. For this purpose, efforts are made to control the biodistribution of mRNA formulations by systemic administration [8].

It is reported that the addition of supplemental lipids changed the biodistribution profile of intravenously administered LNPs, allowing them to deliver mRNA to organs beyond the liver, such as spleen and lungs [119, 124]. The additional components, termed SORT (selective organ targeting) helper lipids, include cationic lipids, anionic lipids, and ionizable lipids (Fig. 4a). The organ tropism is explained by serum proteins binding to the SORT molecules, which are exposed after the desorption of PEG lipids in circulation (Fig. 4b) [125, 126]. For example, spleen-tropic LNPs and lung-tropic LNPs in mouse serum preferentially captured β2-glycoprotein I and vitronectin, respectively, whereas conventional LNPs predominantly distributed to the liver showed ApoE binding (Fig. 4c) [125].

Fig. 4.

(a) SORT nanoparticles for tissue-specific mRNA delivery have unique biodistribution and ionization behavior. A supplemental SORT molecule (an ionizable cationic lipid (DODAP), an anionic lipid (18PA), or a cationic quaternary ammonium lipid (DOTAP)) is added to a conventional, four-component LNP (mDLNP) for tissue-specific mRNA delivery based on the chemical structure of the included SORT molecule. Ex vivo fluorescence of Cy5-labeled mRNA in major organs extracted from C57BL/6 mice IV injected with SORT LNPs that incorporate increasing percentages of different SORT molecules (0.5 mg/kg mRNA/body weight, 6 h). (b) A proposed three step endogenous targeting mechanism for tissue-specific mRNA delivery by SORT LNPs: 1) PEG lipid desorption 2) enables distinct plasma proteins to bind SORT LNPs, 3) resulting in cellular internalization in the target tissues by receptor-mediated uptake; SDS–PAGE of the plasma proteins adsorbed to the surface of mDLNP, liver SORT, spleen SORT, and lung SORT LNPs, showing distinct protein binding patterns of LNPs with different organ-targeting properties. (c) The average abundance of proteins with distinct biological functions in the protein coronas of mDLNP and liver, spleen, and lung SORT LNPs. The choice of SORT molecule leads to large-scale differences in the functional ensemble of plasma proteins which bind the LNP; Isoelectric point distribution for the most enriched proteins constituting 80% of the protein corona of the LNPs, influenced by the headgroup structure of SORT molecules; The top five most abundant plasma proteins that bind different SORT LNPs. The chemical structure of SORT molecule affects the number one plasma protein that is most highly enriched on the surface of SORT LNPs. Reprinted with permission from [125]. Copyright (2021) National Academy of Sciences of the United States of America.

The variation of ionizable lipids can also enhance preferential delivery of mRNA-loaded LNPs to specific cells or organs. Combinatorial libraries of ionizable lipids are generated by varying the tails and the head groups. Screening of 720 ionizable lipids has identified a lead lipid that increased LNP transport to the lung epithelial cells, facilitating mRNA delivery potentially for the treatment of airway diseases [127]. Another recent study reported the delivery of VEGF-encoding mRNA to the cells in the placenta for the treatment of defective placental vasodilation during pregnancy via LNPs containing an ionizable lipid selected from a library [128]. Additionally, surface modification of LNPs has shown to be effective in mRNA delivery to the bone marrow [129]. LNPs were modified with an antibody against CD117, a marker expressed on both human and mouse hematopoietic stem cell (HSPCs) surfaces. With the optimal antibody clone and relatively long alkyl chain PEG-lipid, the antibody-conjugated LNPs delivered Cre recombinase mRNA to almost all HSPCs within the bone marrow [129].

Biodegradable polycations have been explored for systemic delivery of mRNA, varying the composition and surface properties of the formulation. mRNAs encoding interferon regulatory factor 5 (IRF5) and its activating kinase IKKβ, complexed with a poly(β-amino esters) (PBAE) and modified with mannose, were delivered to tumor-associated macrophages (TAMs) for their reprogramming to an M1-like phenotype and the activation of antitumor immune responses [130]. The NPs were delivered to tumors by local (intraperitoneal for ovarian cancer) or systemic routes, showing evidence of TAM phenotype change and promoting antitumor immune responses [130]. The same system was used for in situ delivery of chimeric antigen receptor (CAR) or T cell receptor (TCR) mRNA to circulating T cells for their transient reprogramming [131]. Systemically infused, the NPs induced T cells in blood to express tumor-specific CARs or virus-specific TCRs, resulting in disease regression comparable to ex vivo engineered lymphocytes in mouse models [131]. To expand the utility of PBAEs in systemic mRNA delivery to broader cell populations and organs, the polymer was modified with alkyl amine and other hydrophobic components, and the mRNA/polymer complex was coated with PEG-lipid [132]. Upon intravenous administration, the PEGylated NPs showed predominant mRNA expression in the lungs, despite broad mRNA distribution throughout major organs including the liver [132].

Other polycation-based mRNA delivery employs a block-copolymer (PEG-poly(glycidyl butylamine, PEG-PGBA), forming polyion complex micelles with mRNA [133]. The polycation segment of PEG-PGBA, with a polyether backbone, is more flexible than poly(l-lysine) with a backbone based on peptide bonds, achieving superior mRNA complexation to PEG-PLL and increasing protection from enzymatic degradation and polyanions. Preliminary in vivo studies suggest that the advantages of PEG-PGBA relative to PEG-PLL translated to the improvement of mRNA bioavailability and protein translation efficiency [133]. A recent study reports a new polymer for mRNA delivery, which relies on π–π stacking in addition to electrostatic interaction in the complexation with mRNA. The polymer was produced by coupling tyrosine to PEGylated polyglycerol (PEG-PG), where the tyrosine pendants interacted with mRNA via NH₃⁺ and hydroxyphenyl groups to form <100 nm micelles. The mRNA/PEG-PG-Tyr-based micelles showed a longer circulation half-life than mRNA/PEG-poly(L-lysine) micelle and the evidence of tumor accumulation following intravenous injection [134].

3.5. CRISPR-CAS9

Due to the challenges associated with large protein delivery [135, 136], Cas9 has been delivered as mRNA along with sgRNA, taking advantage of common chemical features of RNA. Therapeutic delivery of the CRISPR-Cas9 RNA combination mirrors the mRNA delivery in both approaches and limitations. The first in vivo gene editing applied in human patients targets the TTR gene of hepatocytes, which produces misfolded transthyretin (TTR) protein causing its abnormal buildup in various tissues, particularly the nerves and heart [137, 138]. The treatment (NTLA-2001) consists of an sgRNA targeting human TTR and a human-codon-optimized mRNA encoding bacterial Cas9, encapsulated in LNPs. NTLA-2001 took advantage of the liver tropism of LNPs due to ApoE corona to deliver the RNA combination to the target hepatocytes, reducing the mean serum TTR by 52% at a dose of 0.1 mg/kg and by 87% at 0.3 mg/kg at 28 days from a single intravenous administration [137].

Preclinical studies further support the feasibility of delivering Cas9 mRNA and sgRNA targeting abnormal genes by LNPs [139, 140]. For example, sgRNA targeting Angiopoietin-like 3 (Angptl3) gene and Cas9 mRNA were co-delivered by LNPs containing tail-branched bioreducible lipidoid to the liver hepatocytes [140]. The optimized LNP formulation lowered blood lipid levels better than those delivered by the commercial benchmark MC3 LNP, lasting > 100 days following a single dose intravenous injection without any indications of toxicity [140]. Another case reports systemic delivery of mRNA encoding ABE8.8 adenine editor and sgRNA targeting the PCSK9 gene in hepatocytes via LNPs [141]. The PCSK9 loss-of-function and the long-lasting (8 months) reduction of PCSK9 and LDL cholesterol levels following a single dose treatment were demonstrated in cynomolgus monkeys [141].

CRISPR-Cas RNA delivery to the organs beyond the liver has been tried, but cases are limited. Cas9-encoding mRNA and sgRNA targeting PLK1, a kinase essential to mitosis, were delivered by LNPs to orthotopic brain tumor and ovarian tumors disseminated in the peritoneal cavity and extended the survival of tumor-bearing mice [60]. The CRISPR-Cas RNA LNPs were administered locally: by intracerebral injection to orthotopic glioblastoma multiforme, and intraperitoneal injection to the ovarian tumors. In the latter, the LNPs were decorated with antibodies for EGFR to enhance selective delivery to EGFR-overexpressing ovarian tumors [60]. As discussed earlier, the variation of ionizable lipids [127, 142] or the inclusion of SORT helper lipids [119, 124] can modify the biodistribution profiles of LNPs delivering CRISPR-Cas RNA combinations. Still, the alternative organs are currently limited to the lungs and spleen.

4. Future perspectives

The year 2023 marks the 70th year since the discovery of DNA structure, which revolutionized our understanding of diseases and therapies. DNA rightfully received immediate attention, but RNA took longer to get the due recognition for its versatile functions and therapeutic utility. Once considered a dull intermediary between DNA and proteins, RNA is now known for its ability to control gene expression and edit the genome, offering significant opportunities for the therapy of undruggable diseases.

The initial challenges in RNA drug development involved enzymatic instability and inflammatory activities, which have been successfully controlled by chemical modification of RNA molecules. The drug delivery technologies initially developed for DNA, such as LNPs, were adopted for RNA delivery [143] to improve cell and organ-level delivery. Most delivery efforts focus on the shared features of RNA drugs coming from their building block, ribonucleotides, such as polarity and negative charge, which account for the widespread use of cationic or ionizable lipids and polymers as the main component. However, RNA drugs vary in size, conformation, and function; thus, it would be reasonable to tailor the delivery approach to keep up with the diversity of RNAs. As such, the following may be worth considering for the future design of RNA drug carriers.

First, the design criteria for a class of RNA drugs may not be generalized to all RNA drugs. For example, chemically stabilized siRNAs remain stable in acidic intracellular compartments, such as lysosomes [95], and use them as a depot storage, enabling months-long activity by a single dose [94]. In this case, fast endosomal escape may not weigh as much as in mRNA drugs. It is also important to understand that the carriers working for one class of RNA do not necessarily translate to others. An early report of mRNA encapsulation in LNPs states that the LNP formulation optimized for mRNA delivery did not improve siRNA delivery, indicating differences in formulation design spaces between the two types of RNAs [144]. Consistently, cationized MSNs serving for siRNA and DNA delivery were reported to be ineffective for mRNA delivery [145]. Although the mechanism is not fully understood, RNAs of different classes do not always share the same challenge and need distinct optimization strategies according to their structures and applications.

Second, immunostimulatory activities of the formulations remain to be controlled. LNPs are by far the most explored RNA carriers and considered safe, given the approval of patisiran and COVID-19 mRNA vaccines. However, LNPs have been reported to serve as an immune adjuvant [146] and exacerbate pre-existing inflammation [147]. The proinflammatory activities are linked to ionizable lipids [148-150]. They may not be a concern in healthy subjects and may be exploited for vaccines or immunostimulatory therapies [151]. However, for future RNA drugs with extended applications, it is worthwhile to explore alternative materials considering the adverse effects.

Third, the liver tropism of LNPs is attributed to their affinity for ApoE and its interaction with low-density lipoprotein receptors [97]. New LNPs targeting extrahepatic organs such as the spleen and lung do so by binding to alternative serum proteins [125]. These observations underscore the significance of protein corona as one of the main drivers of nanoparticle biodistribution [152]. The nanomedicine field is shifting the focus from avoiding the corona formation to co-opting specific proteins in situ by engineering the surface chemistry. RNA carriers would be no exception in this effort, and the carrier design should reflect the dynamic changes occurring in circulation.

Finally, in evaluating new carriers, it is worth noting the difference between in vitro and in vivo activities. We have reported that a polymeric gene carrier effective in cell models does not perform well in tumors due to the interference of tumor-associated macrophages [153]. Likewise, mRNA encapsulated in LNPs have shown inconsistent trends between in vitro and in vivo, with a poor performer in a placental cell model showing significant mRNA delivery to the liver and protein production [128]. While our understanding of such discrepancy is limited, it needs to be considered so that one does not miss out potentially promising candidates.

Fig. 5.

(a) Schematic illustration of CRISPR-LNP (cLNP) preparation. Microfluidic-based mixing of lipids forms cLNPs encapsulating Cas9 mRNA and sgRNA. (b) Therapeutic genome editing in 005 murine glioblastoma multiforme (GBM)-bearing mice. b-1. Schematic illustration of intracerebral injection to mouse brain; Confocal micrographs of brain sections analzyed six hours after intracerebral injection of Cy5.5-cLNPs into the tumor bed. Blue, DAPI; green, 005 GFP cells; yellow, Cy5.5 cLNPs. Scale bars, 50 μm; Percentage of gene editing events in the PLK1 locus as determined by NGS analysis, 48 hours after injection of PBS or 0.05 mg/kg of sgGFP-cLNPs or sgPLK1-cLNPs. b-2. Representative bioluminescence imaging of 005 GBM–bearing mice showing tumor growth inhibition by single-dose treatment with cLNPs; 005 tumor growth curve quantification; and survival curves of 005 GBM–bearing mice. (c) Therapeutic genome editing in OV8-bearing mice. c-1. Schematic illustration of targeted cLNP production by Anchored Secondary scFv Enabling Targeting. c-2. Representative fluorescence imaging of tumors extracted from mCherry-OV8–bearing mice. Top, mCherry OV8 tumors; bottom, Cy5.5-cLNP signal accumulation. c-3. Representative fluorescence imaging of OV8-bearing mice showing tumor growth inhibition by dual-dose treatment with cLNPs; OV8 tumor growth curve quantification; and survival curves of OV8-bearing mice. Reprinted from [60] under a CC BY-NC license.