Dynamic carriers for therapeutic RNA delivery

Abstract

Carriers for RNA delivery must be dynamic, first stabilizing and protecting therapeutic RNA during delivery to the target tissue and across cellular membrane barriers and then releasing the cargo in bioactive form. The chemical space of carriers ranges from small cationic lipids applied in lipoplexes and lipid nanoparticles, over medium-sized sequence-defined xenopeptides, to macromolecular polycations applied in polyplexes and polymer micelles. This perspective highlights the discovery of distinct virus-inspired dynamic processes that capitalize on mutual nanoparticle–host interactions to achieve potent RNA delivery. From the host side, subtle alterations of pH, ion concentration, redox potential, presence of specific proteins, receptors, or enzymes are cues, which must be recognized by the RNA nanocarrier via dynamic chemical designs including cleavable bonds, alterable physicochemical properties, and supramolecular assembly–disassembly processes to respond to changing biological microenvironment during delivery.

At present, the future looks bright for RNA therapeutics and nanomedicine. But only 10 y ago, expectations were modest (1, 2); only extreme optimists would have predicted the current advances in realizing a global medical impact of RNA therapeutics as recently observed for messenger RNA (mRNA) lipid nanoparticles (LNPs). In hindsight, progress was not triggered by one spontaneous breakthrough. It developed steadily over 60 y, with first RNA transfer by synthetic means already achieved in the 1960s using infectious virus RNA and basic proteins or polycationic diethylaminoethyl-dextran (3, 4), demonstration of in vivo mRNA expression after direct administration into muscle in 1990 (5), and numerous further steps. Obviously, groundbreaking discoveries such as RNA interference and inventions such as improved nucleic acid chemistry or genome editing technologies boosted the translation of research into the clinic. This perspective reviews current RNA carriers based on cationic lipids, polymers, and sequence-defined xenopeptides. In particular, it highlights that a dynamic (supra)molecular design is a key property of nanocarriers to meet the challenges for efficient RNA delivery.

Lessons on RNA Delivery from Viruses

Viral entry into cells is a dynamic process, proceeding stepwise in a timely controlled fashion (6). RNA viruses protect RNA by compaction and packaging into protein capsids and optionally additional lipid envelopes. They contain surface domains directly or indirectly interacting with cell surface receptors. In most cases, this triggers active cellular uptake by endocytosis into endosomes, endoplasmic reticulum (ER), or related compartments. After escape into the cytosol, viruses or viral capsids travel to the site of replication in the cytosol (most RNA viruses) or the nucleus (most DNA viruses) and release the nucleic acid. What lessons can we learn from delivery by viruses?

Evolution yielded both “naked”, protein capsid RNA viruses (e.g., rhinovirus, poliovirus) and lipid-enveloped RNA viruses (e.g., influenza virus, coronavirus). Viruses are metastable nanoparticles; several sequential signals are needed for virus uncoating triggered by numerous virus–host cell interactions. This is well illustrated by the fascinating “assembly–disassembly paradox” as outlined by virologists Ari Helenius, Urs Greber, and Yohei Yamauchi (6–8): How can a virus first be assembled in an infected cell and then fall apart (disassemble) during entry into another cell? They provide three types of explanations: i) most commonly, viruses are structurally modified after leaving their host cells; ii) for alphaviruses, they keep unchanged, but virus-producing and uninfected cells are different; and iii) compartments of assembly and disassembly are different. In a nutshell, neither virus nor cell alone is sufficient for virus uncoating, but a dynamic interplay is essential. Yamauchi and Greber compare the viral uncoating process with the snooker game, where numerous timely shots (cues) to the ball (virus) are required for successful delivery. Cues are biological triggers of receptors, enzymes, and chemicals that act directly on the virus particle to alter its structure, trafficking, and infectivity.

Virus attachment to cellular receptors often activates cellular signaling processes that trigger enhanced endocytosis. For example, influenza A virus, after binding to (noninternalizing) sialic acids, induces receptor tyrosine kinase signaling and endocytosis via the epidermal growth factor receptor. Binding of picornaviruses to surface receptors induces conformational changes in the capsid that forms a pore in the endosomal membrane through which the viral RNA is translocated into the cytosol. Adenoviruses, in contrast, cause the lysis of endosomes, allowing escape of the whole capsid (9). An estimated internal pressure of 30 atmospheres within the capsid of adenoviruses or herpes viruses built up by electrostatic repulsion within negatively charged DNA assists in virus disassembly. Polyomaviruses take advantage of ER-associated degradation pathways to enter the cytosol. In the case of influenza virus, release of viral capsids into the cytosol involves membrane fusion, mediated by the glycoprotein hemagglutinin HA2. Conformational change of HA2 at low pH exposes amphipathic peptides that initiate the endosomal membrane fusion, which inspired application of such peptides in synthetic delivery systems (10, 11). The LTR retrotransposon homolog, PEG10, can bind its own mRNA or recombinant mRNA via defined untranslated flanking sequences and facilitates packaging into virus-like particles (12). Exosomes present another natural but nonviral process for cellular exchange of RNA. In sum, numerous creative natural transfer mechanisms serve as instructive role models for designing "synthetic RNA viruses" (10, 13–16).

Delivery Barriers and How to Overcome Them in a Timely Fashion

The viral assembly–disassembly paradox can be easily bypassed in the design of synthetic delivery systems. In the assembly step (carrier synthesis and formulation), chemistries and solvents can be used that do not exist in living systems. This provides an enormous chemical space and versatility for supramolecular assembly of nanoparticles. Dynamics of molecules may include i) physical changes in charge, solubility, or conformation, triggering self-assembly or disassembly, ii) specific molecular interactions (e.g., ligand/receptor), and/or iii) chemical bond cleavage. The design needs to chemically program intended changes for overcoming biological delivery barriers in a bioresponsive fashion. Extracellular nanoparticle stability and nucleic acid protection but disassembly at the place of function are required. Steps that need to be activated in sequential mode are tissue and cellular targeting, facilitated uptake, escape from endosomes, desired intracellular cargo release kinetics and functions. Such alterations need to capitalize on specific triggers in the biological microenvironment, including chemical pH, ionic, or redox states, or contact with specific proteins, receptors, or enzymes (17–20).

Dynamic Lipoplexes and LNPs

The overwhelming impact of LNPs in mRNA vaccines (21) as well as small interfering RNA (siRNA) (22) and other RNA therapies (23) follows a continuous development of liposomal therapeutics over five decades. Initial lipoplexes (24, 25) were designed as complexes of cationic lipids and nucleic acid, stabilizing and compacting the nucleic acid cargo into nanoparticles by electrostatic and hydrophobic forces (26). In LNPs (27–30), combination of the cationic lipid with cholesterol, phospholipids, and a polyethylene glycol (PEG)-lipid merges the advantages of lipoplexes and liposomes: robust encapsulation of RNA into nanoparticles by artificial cationic lipids on the one hand, complemented by the properties of natural lipid membrane forming components and protective PEG-lipid on the other hand, and optionally additional modifications for chemical or ligand targeting. Conceptual benefits of LNPs include their controlled self-assembly in well-established production processes based on low-molecular weight natural or artificial lipid components. The combinatorial space of four lipid components and nucleic acid provides potential for fast optimization. In contrast to standard liposomes or polyplexes, the LNP architecture is highly adaptable for efficient packaging of different nucleic acid cargos, ranging from small siRNA (22) to large CRISPR/Cas9 mRNA (23).

The LNP assembly according to common practice is already a dynamic two-phase process. The lipid mixture dissolved in ethanol is turbulently mixed with an aqueous acidic (pH 4) nucleic acid solution. By this, the lipids convert into small cationic micelles and vesicles based on protonation of the ionizable lipid, complexing the nucleic acid. Specific buffer conditions can modulate the assembly process (31). In the second maturation step, removal of ethanol by dialysis against physiological buffer (pH 7.4) deprotonates the ionizable lipid and stabilizes LNPs. Inherently, LNPs contain at least five additional kinds of dynamic elements that determine their potency in a bioresponsive fashion; i) a PEG-lipid that, based on short (commonly dimyristyl) lipid anchors, is sheddable upon interaction with serum proteins in blood circulation; ii) a cholesterol/phosphatidylcholine envelope that triggers the formation of an apolipoprotein E-rich surface protein corona, facilitating receptor-mediated endocytosis into liver hepatocytes via the low-density lipoprotein receptor; iii) a cationizable tertiary amine-containing lipid (Fig. 1A), which after cellular uptake into acidifying endosomes becomes positively charged and changes physical properties; iv) interaction of cationized and anionic lipids trigger membrane destabilization promoting crossing of the endosomal barrier (Fig. 1B); and v) merging nanoparticle lipids with the cellular lipid membrane lead to release of the nucleic acid payload into the cytosol to a significant extent in carrier-free, bioavailable form (32–34). Consequently, the LNP composition strongly influences protein corona formation, endogenous receptor targeting, and functional activity (35).

Fig. 1.

LNPs—(A) Examples for cationizable lipids, lipidoids, and lipopeptides, containing cationizable tertiary amines (highlighted in blue). (B) Interaction of cationized lipids with endosomal membrane lipids, leading to endosomal escape. Image credit: Created with BioRender.com.

As multicomponent assemblies, LNPs can be improved by combinatorial variation of phospholipids, PEG-lipid, cholesterol analogs, and/or the ionizable lipid. Optimizing the cationizable lipid structure (Fig. 1A) by modifying hydrophilic head group and hydrophobic tails, optionally including degradable bonds, was key for the LNP efficacy (27, 36–38). First breakthrough was the replacement of permanent cationic lipids such as DOTMA or DOTAP by ionizable tertiary amine-containing lipids. The dynamic reversible protonation both enhances efficacy and eliminates potential toxicity problems of permanently cationic materials. The evolution of siRNA LNPs proceeding from the ionizable lipid 1,2-dioleoyl-3-dimethylammonium propane (DODAP) to D-Lin-MC3-DMA (applied in the siRNA product Onpattro) was associated with a more than 3 log units improved potency, reducing the effective dose (ED50) for gene silencing to 0.005 mg/kg (39). As alternative to cationizable lipids, other tertiary amine-containing lipid-like structures were applied in LNPs (40–43). Such compounds termed "lipidoids" were identified by screening of large combinatorial libraries generated by Michael addition or epoxide coupling of lipidic agents to oligoamines or small peptides, yielding potent molecules such as L98-N12-5 (40), C12-200 (41), or lipopeptide cKK-E12 (42). Based on structure-activity analysis of around 1400 degradable lipidoids, Whitehead identified four criteria that predict high in vivo gene silencing potency of corresponding siRNA LNPs in hepatocytes; i) surface pKa values of LNPs >5.4; ii) at least one tertiary amine; iii) at least three alkyl-ester tails; and iv) upon screening alkyl-esters (between C10 and C14), C13 as optimum alkyl chain length (43).

Gaurav and colleagues substituted cholesterol with β-sitosterol, which led to LNPs with polyhedral shape and enhanced endosomal escape (44). Siegwart and coworkers introduced cationizable phospholipids (45). For lipidic carriers, targeting a specific cell type outside the liver from a systemic administration route is a big challenge. Physicochemical engineering of LNPs, for example, with other helper lipids and PEG-lipid ratios, can alter the physical and functional in vivo distribution (“chemical targeting”) with preferential expression in lung, spleen, or tumor (35, 46–54). Alternatively, “active receptor targeting” has been achieved, for example, using peptide-modified LNPs for the retina (55) or antibody-modified LNPs against CD4 (56), CD117 (57), or CD38 (58) for in vivo delivery into hematopoietic stem cells and immune cells. Chemically retargeted mRNA LNPs to nonhepatocyte cell types including endothelial cells, T cells, or tumor cells were identified by barcoded functional in vivo screening (52–54).

In sum, LNPs and related lipoplexes present a highly potent carrier platform based on dynamic supramolecular assembly and disassembly. Current efforts underline their applicability beyond use for vaccination and transfer to hepatocytes. This raises the question: Would we need additional technologies for RNA delivery at all? A simple answer would be yes; natural viruses—with or without lipid envelope—contain their nucleic acid compacted with proteins. Historically, combinations of lipidic and polycationic carriers in nanoparticles termed lipopolyplexes, such as lipid-protamine-mRNA (59) have been explored with encouraging efficacy in tumor models. And ionizable amino-polyesters have already been applied within LNPs for tissue-selective mRNA delivery (60). Such lipid-polymer combinations appear as interesting solution for targeting nonhepatic tissues.

Dynamic Polyplexes

Natural evolution has also generated lipid-free viruses based on naked protein capsids, convincingly demonstrating that lipid-free delivery is a viable option for nucleic acid transfer. A strong rationale for use of cationic polymer-based systems named "polyplexes" and "polymer micelleplexes" (15, 25, 61–64) is a highly effective spontaneous packaging of polyanionic nucleic acids into compact, protected form. The design of medium to high molecular weight polycations provides a direct chemical control on the nucleic acid spatial compaction. These advantages, however, are associated with more demanding syntheses and analytics of macromolecular carriers.

A timely nanoparticle disassembly as part of the intracellular delivery process has to be considered to a higher extent in polymer design than for natural lipid-based systems. Of note, viruses release their fragile nucleic acid only after reaching the intracellular location of replication (6). The stability of purely electrostatic polyelectrolyte complexes is highly dependent on the size and type of nucleic acid as complex formation is driven by entropy increase due to released counterions. Hereby, nucleic acid properties (e.g., number of nucleotides, single versus double-stranded, RNA versus DNA) play an important role. Especially for small cargos such as siRNA, an increase of polyplex stability by incorporation of hydrophobic domains has been found beneficial. For larger nucleic acid materials such as mRNA, compaction into smaller nanosized structures is important at least for initial delivery steps. Furthermore, the different intracellular target sites and functions of cargos have a critical impact on the choice of polymeric carriers (65, 66). For these reasons, polymeric carriers that were initially developed for DNA delivery often cannot be directly applied for mRNA or siRNA delivery. Not surprising, screening of libraries of polymers, such as poly(amine-co-esters) or poly(2-ethyl-2-oxazoline)/polyethylenimine (PEI) copolymers, identified differing compositional optima for different nucleic acid cargos (67, 68).

Potency of polymeric carriers has been strongly linked with bioresponsive, dynamic actions. Two critical steps, i) overcoming the endosomal lipid membrane barrier and ii) intracellular disassembly of the compacted nucleic acid complex at the site of action, have to be handled in a different dynamic mode than lipidic systems that apply lipid mixing with cellular membranes both for cytosolic transition and nucleic acid release (32, 33). The first cationizable polymers such as PEI or polyamidoamine dendrimers demonstrated far superior transfection activity over permanently cationic polymers, largely due to improved endosomal escape (69–74). Due to their high density of amines, these polymers are only partially protonated at physiological pH. Endosomal acidification facilitates further carrier cationization, which supports cation-triggered membrane disruption. A less regarded but important physical characteristic of protonation is the enhanced aqueous solubility of polymers. For example, for hydrophobic stabilization of siRNA polyplexes, Zuber and colleagues (74) applied tyrosine-modified PEI (PEIY). The endosomal pH-triggered dissolution of PEIY self-aggregates was key to the effectiveness as it allows both endosomal rupture and siRNA release.

A series of other dilemmas in polyplex delivery (75) requests dynamic solutions. Surface-shielding of PEI polyplexes with PEG molecules blocks transfection activity ("PEG-dilemma"). Remedies are cellular receptor targeting strategies as reviewed in (76, 77), and triggered PEG removal upon exposure to endogenous stimuli (78, 79). The "polyplex dilemma" refers to the critical balance between sufficiently high polyplex stability in the extracellular environment, but efficient intracellular nucleic acid release. Itaka et al. demonstrated that both linear (LPEI) and branched PEI (BPEI) polyplexes mediate endosomal escape, but subsequent disassembly of LPEI/plasmid DNA (pDNA) polyplexes and gene expression was superior compared with BPEI polyplexes (80). In contrast to BPEI, LPEI polyplexes were also able to transfect nondividing cells (81). Moreover, nucleic acid-specific effects became obvious. LPEI presents an excellent carrier for pDNA but not siRNA or mRNA; succinylated branched PEI, for example, constitutes a far better carrier for the latter cargos (82, 83). Dohmen et al. noted that a small change of PEI-like oligoamines by introducing a methylene group into every second aminoethylene unit strongly enhanced mRNA transfer due to improved balance of cationic properties (84).

The polyplex dilemma has been addressed by numerous investigators, generating biodegradable polymeric carriers, for example, based on hydrolyzable ester bonds, acidic, or redox cues (85–87). A dynamic polymer degradation also reduces toxicity problems of polycationic materials. For mRNA delivery, Nuhn and colleagues designed block copolymers with disulfide-linked amines. The cationized amines allowed effective polyplex formation with mRNA and subsequent release under the reductive conditions of the cytosol (88). Kaczmarek et al. synthesized poly(β-amino ester) (PBAE) terpolymers by Michael addition of hydrophobic diacrylates with various amines, alkylamines, and end caps. Optimized coformulation of mRNA with these PBAEs (low molecular weight of ~3400 Da) and PEG-lipid resulted in strongly (~100-fold) increased mRNA transfer to the lung endothelium and pulmonary immune cells after intravenous administration in mice (89). Yan et al. generated a combinatorial biodegradable polyester library incorporating amino- and alkyl-thiol groups. Intravenous application of mRNA polyplexes stabilized with 5% pluronic F127 enabled marker gene expression in mouse lungs (90). Kataoka and coworkers optimized systemic delivery of mRNA complexed by polyaspartamides (pAsp) with PEI-like oligoamines in the side chain, providing “proton sponge” character. They reported that the number of aminoethylene repeats critically impacts mRNA binding and transfection efficiency (91, 92). pAsp(TEP) presents a slowly biodegradable cationic block with tetraethylene pentamine side chains. A cholesterol moiety was attached to the ω-terminus of the copolymer for enhanced blood circulation upon systemic application. Intravenous administration of therapeutic mRNA polyplexes resulted in growth inhibition of pancreatic cancer in mice (93). Alternatively, the same lab complexed mRNA with poly(L-ornithine), observing an enhanced stability compared to mRNA complexed with poly(L-lysine) which differs in molecular structure by one methylene group only. The mRNA polyplexes were coated with a charge-conversion polymer, cis-aconityl pAsp(DET), negatively charged at extracellular pH but turning positive at endosomal acidic pH to disrupt the endosomal membrane (94). The following subsections report on further examples of polymeric transfection carriers that form dynamic nucleic acid complexes.

Virus-Inspired Polymer for Endosomal Release (VIPER).

Pun and colleagues designed a bioresponsive polymer called VIPER (95, 96). This methacrylate block copolymer (Fig. 2) contained i) a cationic poly(2-(dimethylamino)ethyl methacrylate) domain for binding and compaction of nucleic acid, ii) a pOEGMA [poly(oligo(ethylene glycol) monomethyl ether methacrylate)] block for nanoparticle hydration and surface shielding, and iii) a reversibly hydrophobic domain, containing pDIPAMA [poly(2-diisopropylaminoethyl methacrylate)] and pDSEMA [poly(pyridyl disulfide ethyl methacrylate)]. The activated disulfide groups were used for coupling the lytic peptide melittin. pDIPAMA and related polymer blocks with tertiary amino groups have been used to modulate polymer micelle stability in a highly pH-specific manner (97). Upon VIPER polyplex uptake into acidifying endosomes, pDIPAMA provides a sharp phase transition from hydrophobic to hydrophilic state. Following exposure of pDSEMA-linked melittin peptides leads to endosomal lipid membrane disruption and polyplex delivery into the cytoplasm. VIPER formulations enabled gene transfer into KB tumors after intratumoral application of pDNA (95) and gene silencing in lungs after intratracheal administration of siRNA (96). Bee venom–derived melittin is a very potent pore-forming peptide, raising concerns about cytotoxicity (98). Studies of VIPER analogs with pH-specific fusion peptides instead of melittin demonstrated that the pH-specific exposure of melittin compensated for the lack of inherent pH-specificity of the peptide (99). The findings remind of the pH-triggered “spring-loaded” conformational change in the influenza hemagglutinin, exposing the terminal HA2 fusion peptide. Also, other investigators applied the endosomal responsiveness of pDIPAMA derivatives for nucleic acid delivery, for example, in the design of PD-L1 siRNA micelleplexes for cancer immunotherapy (100).

Fig. 2.

VIPER. (A) Endosomal pH-responsive methacrylate copolymer VIPER. (B) Upon endosomal protonation, the hydrophobic diisopropylamino domain is solubilized, exposing the membrane bilayer disruptive peptide melittin, which promotes release into the cytosol. Reproduced from Cheng et al. (95) with permissions of John Wiley and Sons. 2016 Wiley-VCH Verlag GmbH & CO. KGaA, Weinheim.

mRNA Transfer by Charge-Altering Releasable Transporters (CARTs).

Waymouth, Wender, and colleagues developed amphiphilic diblock copolymers for mRNA transfer, named CARTs (101–103). The carriers provide a polycarbonate block containing esters of lipophilic alcohols and cationic hydroxyethyl glycine repeat units (Fig. 3). Optimized carriers form mRNA complexes by electrostatic and hydrophobic interactions. Intracellular polyplex disassembly and mRNA release into the cytoplasm is triggered by charge-neutralizing intramolecular rearrangement, degrading the cationic block to neutral small molecules by self-immolation. Libraries of CARTs with different incorporated lipidic side groups yielded mRNA complexes that successfully transfected lymphocytes in vivo (102) and upon intratumoral injection triggered antitumor immune responses (103). Recently, the researchers replaced the hydroxyethyl glycine (G) motif by a hydroxyethyl lysine (K) motif (104). Upon systemic application of mRNA polyplexes in mice, the distinct CART carriers showed different organ specificity (K-CART >90% lung selectivity; G-CART >99% spleen selectivity). Importantly, the dynamic characteristics of polycation degradation reduce potential toxicity problems with persistent polycationic materials.

Fig. 3.

CARTs. Mechanism of polyplex disassembly and release of mRNA by carrier degradation. Reproduced from Blake et al. (104) with permission of the American Chemical Society. 2023 ACS Publications, Washington D.C.

Evolution of Sequence- and Topology-Defined Lipo-Xenopeptides

The complexity and precise functionality of natural proteins is based on their evolutionary optimization as sequences, with storage of information in the form of nucleic acids. Analogous chemical evolution strategies have been adopted for optimizing sequence-defined nucleic acid carriers (105). The ease and precision of solid-phase assisted peptide synthesis (SPPS) enabled the design of peptide (106–110) or xenopeptide (111–115) libraries with clear sequence-activity relationships, providing valuable information for nucleic acid complexation and transfection. For example, Mixson and colleagues systematically evaluated linear and branched peptide libraries containing lysines (for nucleic acid binding) and histidines (for endosomal buffering) in various ratios and architectures (e.g., linear, four or eight branches) as carriers for siRNA and mRNA transfer (107, 108). Lu and collaborators used SPPS with artificial oligoamine building blocks for the assembly of cationic lipopeptides as carriers for antitumoral siRNA (111).

Inspired by the known “proton sponge” transfection properties of PEI and shorter oligoethylenimines, Schaffert et al. designed artificial oligo(aminoethylene) amino acids such as succinoyl tetraethylene pentamine (Stp), with terminal Fmoc and internal Boc amine protecting groups (115), for SPPS assembly into sequence-defined oligoaminoamides (OAA). Libraries of OAA sequences were synthesized in different architectures, including linear (115, 116), two-arm (115), three-arm (115), or four-arm (115, 117) topologies. Branching points were provided by the two (α, ε) amines of lysine and incorporation of histidines tuned efficacy by increasing the capacity for endosomal protonation. This strongly enhanced gene transfer both in vitro and in vivo in tumors (117).

T-Shaped Lipo-OAA Carriers for siRNA, CRISPR-Cas9/single guide RNA (sgRNA) Ribonucleoprotein (RNP), and mRNA Delivery.

Especially for siRNA as small nucleic acid cargo, stabilization of polyplexes by incorporation of hydrophobic domains has been found to be beneficial. For this purpose, lipo-OAAs were designed with i-shape, T-shape, or U-shape topology (115) by introducing two or four fatty acids or other lipidic residues. In several different topologies, terminal tyrosine tripeptides plus cysteines served as additional stabilizing motifs (118). T-shaped lipo-OAA carriers (Fig. 4) were highly suitable for siRNA delivery. In combination with ligand-PEG surface modification, tumor-targeted siRNA polyplexes demonstrated tumoral EG5 gene silencing in mouse lymphoma or hepatoma tumor models in vivo (119, 120).

Fig. 4.

T-shaped lipo-OAAs for siRNA, sgRNA, and mRNA delivery. Optimization by structural variations both in backbone and side chains (dashed lines indicate optional incorporations/variations of distinct motifs). C, cysteine; FA, fatty acid; H, histidine; K(N₃), azido-lysine; LinA, linoleic acid; NonOcA, 8-nonanamido octanoic acid; OHSteA, hydroxystearic acid (hydroxyl group at position C9 or C10); OleA, oleic acid; Y₃, tyrosine-tripeptide; ssbb, cystamine disulfide building block; Stp, succinoyl tetraethylene pentamine. Image credit: Created with BioRender.com.

Incorporation of hydrophobic fatty acid residues was also required for genome editing using Cas9 protein/sgRNA RNP polyplexes. Lächelt and colleagues (121) screened different fatty acids within T-shape structures and observed that hydroxystearic acid (OHSteA) was far superior to stearic acid regarding genome editing by Cas9/sgRNA RNPs. Lin et al. (122) further optimized the OHSteA Cas9 RNP carriers by incorporation of folic acid (FolA)-PEG for receptor-mediated uptake and improved gene editing. Targeting two immune checkpoint genes, PD-L1 and PVR, the FolA-modified RNP formulations achieved substantially higher gene editing rates after injection into CT26 colon cancer in vivo as compared with nontargeted nanoparticles. The dual disruption of PD-L1 and PVR induced CD8⁺ T cell recruitment and distinct CT26 tumor growth inhibition. Further tuning of Cas9/sgRNA RNP polyplex activity was achieved by systematic variation of the amphiphilic xenopeptide sequences (123), particularly the number and types of artificial oligoamino acids and applied fatty acids. The evaluation of 78 xenopeptides revealed a relationship between the logD_7.4 and genome editing potency. A highly potent amphiphilic carrier TFE-IDAtp1-LinA, containing a trifluoroethyl-iminodiacetic acid analog of Stp, achieved enhanced green fluorescent protein knockouts with an ED50 of 0.38 nM RNP. Mechanistic studies demonstrated that hydrophobic xenopeptides were more resistant to ionic stress and concentration-dependent dissociation, but a sweet spot of logD_7.4 existed for each xenopeptide architecture, since the activity dramatically dropped beyond the optimal range (too hydrophilic or too hydrophobic).

Similarly, for successful delivery of larger nucleic acid molecules such as pDNA and mRNA, too high lipopolyplex stability was found to negatively affect transfection efficiency. A careful balance between polyplex stabilization by lipidic residues and sufficient cargo release is required (83, 124). Screening of a library of T-shaped lipo-OAAs containing fatty acids of different chain lengths (C2 to C18) revealed that carriers with shorter fatty acids (C6 to C10) mediated lower polyplex stability, but higher endosomolytic potential and nucleic acid transfer activity than analogs with longer fatty acids (124). For mRNA lipopolyplexes, the most important measure was the incorporation of a bioreducible disulfide bond (i.e., ssbb) between the backbone and the lipidic side chain, leading to more effective mRNA release in the reductive intracellular environment (83). In carriers without this dynamic tuning element, less stabilizing shorter fatty acids were advantageous. Best performer of the library was a T-shape lipo-OAA containing histidines, ssbb and oleic acid. It effectively delivered mRNA to the lungs of mice upon intratracheal application.

Double pH-Responsive Lipo-Xenopeptides with Molecular Chameleon Characteristics.

Based on continuous chemical evolution, a class of novel, highly potent, nonviral nanocarriers for dynamic delivery of various nucleic acids (mRNA, pDNA, and siRNA) was recently generated via SPPS (Fig. 5). These carriers that combine the benefits of both polymer- and lipid-based delivery modules were found as up to several hundred-fold more potent than previous carrier generations (125). They were designed to contain at least two novel lipo amino fatty acids (LAFs) as cationizable motifs with tunable polarity in combination with polar cationizable Stp aminoethylene units. The building blocks were connected via lysines into different topologies (i.e., combs, T-shapes, bundles, and U-shapes) at different ratios of Stp to LAF. The switchable polarity of the LAF was implemented by a central tertiary amine, which disrupts the hydrophobic character once protonated, resulting in pH-dependent structural and physical changes, as evidenced by drastic changes in the logD from around +1 (hydrophobic) at pH 7.4 to −1 (hydrophilic) at pH 5.5. This "molecular chameleon character" turned out to be beneficial for dynamic and fast cargo delivery via lipopolyplexes utilizing lipid-mediated and cationic mechanisms. Screening of different topologies, Stp/LAF ratios, and LAFs identified tailor-made carriers for the various nucleic acid cargos. In the case of pDNA and mRNA, bundles with short LAF hydrocarbon chains as well as U-shapes, especially with LAFs of medium length, were favorable. For siRNA, the U-shape topology was found as most potent, even at very low siRNA doses (~170-fold lower dose compared to former positive controls). mRNA lipopolyplexes were highly potent in tumor cells even in the presence of 90% serum and at ultralow doses of 3 pg mRNA (~2 nanoparticles/cell), by this being comparable to viral vectors in terms of particle efficiency. Moreover, the carriers displayed high activity in vivo upon systemic application of 1 to 3 µg mRNA in mice, especially in spleen, tumor, lungs, and liver.

Fig. 5.

Double pH-responsive lipo-xenopeptides as molecular chameleons. (A) LAF-Stp carriers with both apolar (LAF) and polar (Stp) cationizable units. (B) pH-tunable polarity mediated by tertiary amines results in structural and physicochemical changes, evidenced by a drastic change in the logD value. (C) Effective U-shape and B2-bundle topologies. (D) Hypothetical mechanisms of endosomal escape of lipopolyplexes formed with LAF-Stp carriers. LAF, lipo amino fatty acid; Stp, succinoyl tetraethylene pentamine. Image credit: Created with BioRender.com.

Future Steps and Challenges

Advances in nanocarrier engineering enable several practical options for protecting rather fragile therapeutic RNA, delivering it into target tissues, across cellular membrane barriers, and releasing it in intact bioactive form at its site of action. In vivo targeting of certain tissues such as the central nervous system or tumors is limited for nanoparticles and will require active processes to overcome pharmacological barriers. Carriers range from lipids over xenopeptides to polymers. For all classes, accumulated knowledge of structure-activity relationships is already available for further optimization. Nevertheless, a series of critical challenges has to be overcome.

Robust Chemistry and Stable Storage Despite Subsequent Dynamic Function.

We face a chemical assembly–disassembly paradox: The initial RNA nanoassembly, based on ideally simple components, must be designed in stable form that enables long shelf-life and storage without premature release or degradation. For example, recently developed COVID-19 mRNA vaccines require ultracold storage (126, 127). Current efforts aim at 4 °C storage, with the Moderna’s lyophilized CMV vaccine evaluated in a phase 3 trial. In contrast, subsequent RNA delivery should occur in a dynamic fashion; the cargo must be delivered in intact form, but nanoparticles and carrier components should dissociate into excretable, nontoxic fragments. Thus, we face the challenge of synthesizing dynamic carriers with high precision, at convenient large scale and with high yields. Suitable storage formulations including protective excipients, lyophilizates, or other innovative forms such as spray-dried nanoparticles (128) need to be developed.

In Vitro–In Vivo Discrepancy.

As demonstrated repeatedly, the predictive value of in vitro and cell culture research for in vivo efficacy is very limited (129, 130). Good in vitro activity may or may not correlate with intended in vivo performance. Even worse, a selection process of nanocarriers based on in vitro assays only might eliminate very potent in vivo candidates. Consequently, in vitro assays are needed that mimic the in vivo situation more realistically. In this context, it is recommended to combine several analytical and biological characterization methods to get a better understanding of the interactions at the nano-bio interface and in vivo nanoparticle characteristics (129). The choice of biofluid as well as standardized protocols are of great importance for more consistent, robust, and comprehensive preclinical studies in order to derive relevant structure-activity relationships and in vitro–in vivo correlations. Some information like in vivo biodistribution and off-target effects, however, are not provided by in vitro experiments. In this respect, new high-throughput in vivo screening methods such as the barcoding system of Dahlman et al. represent a more effective, economical, and ethical way compared to conventional in vivo studies (52–54, 131). Translatability from small to large animals and humans is another so far unmet challenge (132). Better in silico predictions due to advances in bioinformatics, as well as new technologies such as microfluidic “human-organ-on-a-chip”, may substitute animal studies in the future (132).

Bioinspired Chemical Evolution Including Machine Learning.

Viruses optimized by biological evolution for efficient nucleic acid transfer into host cells can serve as role models for virus-like synthetic delivery systems (16, 19). Bioinspired chemical and molecular evolution (16, 61, 105) is done by i) creating libraries of combinatorial/sequence-defined nanocarriers with different motifs and topologies by means of precise synthesis routes (e.g., SPPS) and modular design, followed by ii) cycles of high-throughput screening, iii) rational selections, and iv) systematic structural mutations. Combination with in silico simulations, computational/statistical predictions (e.g., design-of-experiments), and machine learning approaches (113, 133) may rationalize and accelerate the optimization process.

Conclusion

So far, LNPs represent the most advanced nonviral formulations for successful application of therapeutic RNA in vivo. At current stage, delivering RNA therapeutics to the liver can already be achieved. Also, vaccination with mRNA LNPs via local intramuscular injection has proven to be effective and safe in humans. However, addressing nonhepatic organs or tumors is still an unresolved challenge. Further optimization of the nonviral, synthetic delivery systems is required. In this regard, natural viruses can serve as models. Optimized by evolution, they feature numerous dynamic elements for successful cellular nucleic acid transduction. Learning from this, “synthetic viruses” can be developed as a delivery platform for RNA therapeutics and optimized in a bioinspired molecular and chemical evolution approach. Natural viruses consist of nucleic acid encapsulated in a peptide capsid that is in many cases additionally surrounded by a lipid envelope. Therefore, lipid–polymer combinations represent a sweet spot between LNPs/lipoplexes and polyplexes and may have a huge potential for targeting tissues beyond the liver.

Data, Materials, and Software Availability

There are no data underlying this work.