Nonviral nucleic acid delivery systems: specific features of polymeric carriers

Recent years have witnessed rapid development of therapeutic methods based on the use of various types of nucleic acids (NAs) in diagnosis, therapy, and vaccination.[1] The real application of NAs in medicine began in 1998, when the drug fomivirsen (Vitravene®) was approved by the U.S. Food and Drug Administration (FDA) for the treatment of human cytomegalovirus retinitis.[2] This drug was a synthetic antisense oligonucleotide (ASO) and was administered without additional delivery agents. In the subsequent years, large-scale research was undertaken with the aim of expanding the range of therapeutic NAs.

A key problem of NA application in medicine is to provide their targeted and effective delivery to the cells of a living organism. The biological methods for NA delivery (Fig. 1) are based on the use of viral agents (transduction).[3] The nonviral approaches to the delivery of NAs,[4][5] which are called transfection, make use of physical and chemical methods that are also shown in Fig. 1. Most physical methods are aimed at increasing the permeability of the cell membrane and involve the direct introduction of NAs into the cell; chemical methods are based on the use of various chemical compounds that act as NA carriers.[6]

Fig. 1

Key approaches to delivery of nucleic acids into cells

The delivery of NAs into a cell or the cell nucleus may serve various purposes; the mechanisms of gene therapy are discussed in detail in a number of biomedical reviews.[7-9] Irrespective of the target effect, the delivery of NAs into a cell is a challenging interdisciplinary problem. The development of viral delivery systems is a laborious task,[10] which typically takes a lot of time. Furthermore, these agents raise concerns related to immuno­genicity (insertional mutagenesis) and limited therapeutic cargo.[11][12] An example of effective and fast solution of the pressing challenge of protecting the population during the SARS-CoV-2 coronavirus pandemic was the development of the Sputnik V vaccine based on the Ad26 and Ad5 adenoviruses[13][14] and its analogues.[15][16] The use of adeno-associated viruses and lentiviruses reduces the risk of mutagenesis (a vivid example is the Zolgensma® drug developed by Novartis[17]); however, the wide use of such drugs is considerably held up by their cost (~$3.2 million for a treatment course with Elevidys® manufactured by Sarepta Therapeutics and ~$2 million for Zolgensma®).

Obvious drawbacks of physical transfection methods (see Fig. 1) are the need for specialized equipment and low throughput; the practical application of such methods in vivo is limited.[18] At the current stage of research, chemical methods and approaches are usually utilized for effective NA delivery into a living organism. In scientific periodicals and extensive review literature dealing with the development of chemical NA delivery systems,[19-22] these vectors are traditionally subdivided into three main types: nanoparticles (NPs), liposomes (including lipid nanoparticles, LNPs), and polymers.

The role of a chemical carrier is to bind NAs and counterbalance their negative charge, protect NAs from degradation during the delivery, and ensure NA penetration into the cell. The first examples of compounds capable of forming cell-penetrating NA complexes are diethylaminoethyl dextran (DEAE-dextran, 1965)[23-25] and nano-sized hydroxyapatite (HAp, 1973) co-precipitated with NAs.[26] Note that the application of HAp is still relevant,[27-29] and other inorganic[30-33] and hybrid organic-inorganic[34] carriers based on NPs are also intensively studied.

DEAE-dextran (Fig. 2) is a cationic polymer that successfully binds and compacts NAs; however, the transfection efficiency of these complexes turned out to be low. Chitosan is a readily available natural analogue of DEAE-dextran, which has attracted attention of researchers since 1998[35] and up to now.[36] The synthesis of cationic lipids, which are still widely used today, was a milestone in the development of transfection agents. N-[1-(2,3-Dioleyloxy)propyl]-N,N,N-trimethylammonium (DOTMA) reported in 1987[37] was the first example of such a carrier. The further development of the concept of cationic lipids led to the design of ionizable lipids, that is, amphiphilic molecules that contain aliphatic amino groups that can be protonated.[38-42] The amphiphilic nature of cationic and ionizable lipids accounts for the liposomal features of morphology of their complexes with NAs in physiological media.

Fig. 2

Structures of DEAE-dextran, chitosan, DOTMA, PLL and PDMAEMA polymers, branched PAMAM and bPEI, dimeric ionizable lipid, and the Lipofectamine 3000 formulation

In the late 1980s and early 1990s, various amino-containing oligomers and polymers were investigated as carriers for NAs. Among these, mention should be made of poly(L-lysine) (PLL, 1988),[43] dendrimer polyamidoamine (PAMAM, 1993),[44] polyethylenimine (PEI, 1995),[45] and poly[2-(dimethylamino)­ethyl methacrylate] (PDMAEMA, 1996),[46] which have formed the toolbox of polymeric transfection agents for many years to come (see Fig. 2).

In the subsequent years, large-scale research was undertaken to expand the range of therapeutic NAs and develop methods for their delivery; during this time, effective two- and multi-component transfection agents were obtained.[19][39][47-53] An example of widely used two-component transfectant is Lipofectamine 3000, which is based on ionizable lipids, DOSPA {2,3-dioleyloxy-N-[2-(sperminecarboxamido)ethyl]-N,N-dimethyl-1-propylammonium hydrochloride} and DOPE (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine) taken in ~3 : 1 ratio (see Fig. 2).[54]

The advances in NA delivery techniques using LNPs led to the development of patisiran (Onpattro) drug based on small interfering RNA (siRNA), which was approved by FDA in 2018.[55] This led to the exceptionally rapid development of SARS-CoV-2 vaccines based on messenger RNA (mRNA) by Moderna and Pfizer/BioNTech.[56] According to a report by the American Society for Gene & Cell Therapy, approximately 1250 NA-based drugs were undergoing clinical trials in 2025.[57] The issues of practical use of NA therapy are addressed in a number of reviews (see, e.g., Refs [8], [58-60]). The total number of FDA-approved gene therapy protocols and drugs is currently approaching 40 (Table 1); however, the proportion of nonviral agents among them is low.

Table 1

\[ \]

List of gene therapy drugs approved by FDA and/or other organizations (according to Refs 8, 58 – 60)

(1)

Table 2

\[ \]

Table 1 (continued)

(2)

A large number of reviews published in the last decade address general issues related to transfection[61-63] and the prospects for its use in specific applications.[64][65] There are publications devoted to nonviral[66-69] delivery of NAs by means of NPs,[20] including cationic LNPs [19][47-51] and LNPs based on ionizable lipids.[39] Particular attention has been given to mRNA delivery systems.[41][70-74] Some reviews consider the general issues of NA delivery using polymeric carriers[75-83] and specific types of these carriers based on poly(amino esters),[84][85] including poly(β-amino esters) (PβAE),[86-90] PEI,[91][92] PAMAM,[93] and chitosan.[36][94-96] In the context of the present review, mention should be made of earlier publications in Russian Chemical Reviews,[97-100] which address issues directly related to the development of chemical delivery systems for nucleic acids.

This review consists of two logical parts. The first part is introductory (Section 2) and describes the key issues of transfection that are common to polymeric and low-molecular-weight carriers (nonviral vectors). The second part is devoted to high-molecular-weight carriers and NA delivery using them (so-called polyfection), including the design and synthesis of polymeric NA carriers (Section 3) and to the diversity of their characteristics, which make it possible to control the amphiphilicity, buffer capacity, polymer chain topology, and, ultimately, the efficiency of NA delivery (Section 4). The summarizing Section 5 compares lipofection and polyfection and highlights relevant trends in the development of polymeric NA carriers. A distinctive feature of the present review is the attempt to identify the key trends of polyfection, since most studies on transfection are devoted to lipid carriers, whereas important experimental studies on polymeric carriers are often overlooked in broader reviews. While preparing this review, we sought to pay particular attention to chemical aspects of polyfection at all stages of the research ranging from polymer design and synthesis to the identification of structure–property relationships as applied to specific NA delivery tasks.

Along the path to their destination, exogenous NAs encounter a few barriers, including

(1) the negatively charged phospholipid membrane of the cell, which prevents the penetration of negatively charged hydrophilic NA molecules;

(2) endosomal (or lysosomal) vesicles in which the captured NAs are cleaved;

(3) the nuclear envelope, which prevents DNA from entering the nucleus.

These barriers can be overcome only through fine-tuning of interactions between NA molecules, the carrier, and cell compartments.

The initial stage of cellular internalization of a nonviral vector is binding to the cell membrane. This binding can be achieved in two ways: through specific interaction between the receptor and the ligand and through nonspecific interaction between the negatively charged outer surface of the plasma membrane and the positively charged NA – carrier complexes. In the former case, the complexes should be additionally modified with specific ligands, which enables the targeted delivery of NAs to particular cells and tissues. Thus, a well-known strategy is the mannosylation of lipid, peptide, and other carriers to achieve targeting of the cells that express the mannose receptors: macrophages, immature dendritic cells, and endothelial cells.[101] In the latter case, the cationic complexes are attached to the cell membrane via Coulombic attraction forces.^a[102] As the charge of the complex increases, the efficiency of binding to the cell increases due to interactions with negatively charged phosphatidylserine, sulfated glycosaminoglycans, and sialic acids.[103][104] Weiss et al.[105] showed that for a series of cationic carbon nanoparticles with similar ζ-potentials, the cellular uptake and toxicity are better correlated with the surface charge density rather than with the absolute ζ-potential value.

After adhesion of the nanoparticle carrier on the cell membrane, the next stage is internalization of either the whole particle or the NA being delivered. The simplest form of internalization, that is, direct fusion of the carrier with the cell membrane, occurs in rare cases, e.g., under the action of cell-penetrating peptides (CPP).[106] They constitute a family of peptides not more than 30 amino acids long, enriched with positively charged groups; for example, the R9 peptide carries only guanidine groups as substituents.

The decoration of various polymers ranging from poly(trimethylene carbonate) to polynorbornene with guanidine-containing groups is a known strategy for the design of NA carriers possessing CPP-mimetic behaviour.[107-109] As an example, consider the class of NA carriers reported by Jeon et al.,[110][111] that is, poly(oxanorbornene)imides with guanidine groups in the side chain (PONI-Guan) (Fig. 3). For these systems, a model of direct internalization via fusion was proposed instead of the classical endocytic model of cellular uptake (see below). In 2023, it was shown[110] that ~170 nm ^b polyplexes (сomplexes of NA and polymers) produced a diffuse cytosolic signal from Cy3-labelled siRNA after incubation with RAW264.7 (macrophage-like cells). After 6 h, the complexes showed little colocalization with LysoTracker, while 92.4% of the cells were Cy3-positive, which was inconsistent with conventional endosomal accumulation. Functionally, the use of these polymers was accompanied by a more than 70% inhibition of eGFP expression (GFP is green fluorescent protein) in genetically modified RAW264.7:eGFP cells at a guanidinium : phosphate ratio of ~30. The developed system surpassed the Lipofectamine RNAiMAX commercial delivery system by ~25%. The key mechanistic argument of the authors[110] was based on the fact that depletion of membrane cholesterol under the action of nystatin and methyl-β-cyclodextrin ^c decreased the gene knockdown by more than 75%, whereas endocytosis inhibitors had only a minor effect (< 10%). This fact served as evidence supporting the hypothesis of direct internalization of polyplexes depending on membrane fluidity. For therapeutic siRNAs against the TNF-α gene, the carrier showed a >65% decrease in TNF-α secretion ^d and a ~ 80% decrease in the amount of mRNA for TNF-α in vitro, and also a >80% inhibition of TNF-α expression in vivo in lipopolysaccharide (LPS)-stimulated mouse cells at a low dose of siRNA (0.28 mg kg^–1).

Fig. 3

Structures of guanidine-containing polymers

Endocytosis is a more frequent mechanism for NA uptake than direct translocation of NA into the cytoplasm. The main endocytosis types include clathrin-mediated endocytosis (CME), caveolae-mediated endocytosis (CvME), and macropinocytosis; implementation of a particular mechanism depends on the type of cytoplasmic proteins involved and/or composition of lipid rafts (Fig. 4).[49][112] The initiation of a particular mechanism is also influenced by the physical characteristics (size, shape, and surface charge) and chemical properties of the complexes (composition of the carrier and ligands on the nanoparticle surface) as well as by the type of cells.[113][114]

Fig. 4

Illustration of the main types of endocytosis

The clathrin-mediated endocytosis consists in the assembly of clathrin proteins at the ligand – receptor binding site and the formation of a plasma membrane invagination, followed by the formation of vesicles with a diameter of 70 – 150 nm. This mechanism is implemented in the uptake of nutrients (such as transferrin) and is often used to capture small nanoparticles. In addition, CME is preferred for cationic NPs due to strong electrostatic interactions with the negatively charged cell surface. The vesicles formed during CME are acidified and transformed into late endosomes and lysosomes.

The caveolae-mediated endocytosis is characterized by the formation of flask-shaped invaginations (50 – 100 nm in size) rich in caveolins, cavins, cholesterol and sphingolipids. Caveosomes can fuse with early endosomes or be directed toward the endoplasmic reticulum or the Golgi apparatus, in some cases avoiding lysosomal degradation. CvME can be initiated by bacterial toxins, albumin, folic acid, and other compounds. This mechanism is preferred for neutral or negatively charged nanoparticles.

Finally, macropinocytosis is a non-selective process that forms large macropinosomes (0.2 – 5 μm), which capture both liquids and particles without a strict receptor dependence. Subsequently, the contents of the macropinosome may recirculate to the plasma membrane and develop into lysosomes. Macropinocytosis is stimulated by growth factors, phosphoinositides, diacylglycerol, and other compounds. This route is characteristic of large nanoparticles and aggregates (>500 nm); it is more probable for charged NPs an for NPs carrying so-called protein coronas,[115] structures that are formed on the NP surface in biological media (for details, see Section 4.6.3).

Both the physical factors (particle shape, size, and surface topography) that influence the initiation of a specific endocytosis pathway[113][115] and the chemical factors related to the particle composition and surface charge are adequately covered in the scientific literature. The effect of the carrier charge on the mechanism of NA internalization by lipid and polymeric vectors deserves separate consideration. Zhao et al.[116] investigated a series of LNPs based on 1,2-dioleoyl-sn-glycero-3-phospho­choline (DOPC) and 3-[N-(N',N'-dimethylaminoethyl)­carbamoyl]cholesterol (DC-Chol). By varying the proportion of the cationic lipid, the researchers set different ζ-potentials for NPs (from +32 to +43 mV), while the NP size was invariable (~ 100 nm). It was shown that an increase in the positive charge is accompanied by increasing efficiency of LNP uptake by cells and decreasing nanoparticle stability and tendency to capture proteins from the medium to form their own corona. However, the mechanism of cellular uptake remains the same, even on switching to negatively charged LNPs based on DOPC, 1,2-dioleoyl-sn-glycero-3-phosphoglycerol (DOPG), and cholesterol. In both cases, CME virtually does not take place, and the uptake of NPs is largely determined by the interaction of the corona with heparan sulfate proteoglycans. Thus, the addition of dextran sulfate to the system caused an approximately twofold decrease in the cellular uptake of both positively and negatively charged LNPs.

The relationship between the endocytosis mechanism and the charge as applied to polyplexes the characteristics of which differ from those of LNPs was addressed by Mott et al.[117] The authors studied the efficiency of the delivery of plasmid DNA as a PGA – PEI–plasmid ternary polyplex [PGA is poly(α-glutamic acid)] to the HEK293, U-87MG, and HeLa cells,^e the effect of the component ratio in the mixture on the choice of the endocytosis pathway, and particle aggregation with bovine serum albumin (BSA). The best results for all three tested cell lines were obtained for a polyplex with a component ratio of 1.5 : 3 : 1 (by mass), a ζ-potential of +11 mV, a hydrodynamic diameter of approximately 200 nm, and moderate aggregation with BSA. Thus, the introduction of PGA partially suppressed the particle aggregation with BSA (isoelectric point of 4.9, negative charge at physiological pH), which simulated the anionic protein environment in the cellular medium. According to experiments, transfection proved to be less effective for negatively charged polyplexes, despite their internalization in amounts similar to those of positively charged polyplexes. Hence, the efficiency of transfection is limited not by the internalization, but rather by differences in the intracellular transport of the polyplexes. The positively charged polyplexes were mainly detected in endosomal compartments containing caveolin, whereas nearly half of the negatively charged polyplexes were transported to acidic late endosomes or lysosomes, which was a result of clathrin-dependent endocytosis.

Thus, at the level of interaction between cells and nanocarriers, similar selection mechanisms operate for LNPs and polyplexes, ensuring the NA transport via a particular endocytosis mechanism. This mechanism, in turn, determines the rate and degree of acidification of the arising vesicle and the probability of successful NA delivery.

^a Vasir and Labhasetwar[102] directly measured the adhesion force between cells and PLL-modified nanoparticles by atomic force microscopy.

^b Here and below, the presented particle diameter was determined using dynamic light scattering (DLS) unless otherwise specified.

^c Nystatin is a polyene macrolide that binds to membrane sterols; methyl-β-cyclodextrin is a cyclic oligosaccharide with a hydrophobic cavity capable of extracting cholesterol from the plasma membrane.

^d TNF-α is the tumour necrosis factor alpha, a key pro-inflammatory cytokine produced by macrophages.

^e HEK293 are human embryonic kidney cells, U-87MG are glioblastoma (brain tumour) cells.

Despite the differences in the internalization mechanisms discussed above, it is necessary to bear in mind that both macropinosomes and caveosomes can fuse with early endosomes, which predetermines their evolution into lysosomes. Therefore, endosomal escape is a common bottleneck for virtually any transfection system. This challenge is addressed in a number of recent reviews.[118-122]

During maturation, the endosome pH gradually decreases from a physiological value of 7.4 down to ~6.5 in the early endosome, 6.0 – 4.8 in the late endosome, and 4.5 in the lysosome.[123] The carrier inability to bring NAs out of the endo(lyso)somal compartments results in the elimination of loaded NAs via exocytosis or in their enzymatic degradation. Thus, to achieve a productive release from these compartments is a key condition for effective transfection. According to some studies, the fraction of nanoparticles that escape from endosomes is exceptionally low (1 – 2% for siRNA[124] and < 10% for mRNA in experiment using easy-to-transfect HEK293 cells)[125] even in systems such as lipid nanoparticles that are already used in clinical practice.

Different nanocarriers tend to use different mechanisms of endosomal escape. In lipoplexes, the major component, ionizable lipid, is responsible for this process. At physiological pH values, the ionizable lipid is neutral and insoluble, but it becomes charged and water-soluble as pH decreases. Electrostatic interactions arising between cationic lipids and anionic lipids of the endosome give rise to cone-shaped ion pairs and trigger a phase transition of the endosomal membrane lipids from a bilayer structure (lamellar phase in Fig. 5) to an inverted micelle structure (hexagonal phase).

Fig. 5

Mechanism of endosomal escape of lipids

The descriptions of the endosomal escape of NAs transported by polymeric carriers take into account specific effects of the carrier macromolecules: polymer swelling, polycation intercalation into the endosome membrane, and the proton sponge effect (Fig. 6), which is widely discussed in scientific literature.[126]

Fig. 6

Schematic picture of the proton sponge effect.¹²⁶ Published under the CC-BY 4.0 license.

The proton sponge concept is based on the buffering action of polymers the pK_a values of which fall within a physiologically relevant range. This model is often mentioned in studies dealing with PEI, PLL, PAMAM, and other polyamines.[127] This effect is due to the formation of a buffer system as the polymer amino groups are being protonated during endosomal acidification: the ongoing activity of the vacuolar H⁺-ATPase results in the accumulation of protons, the entry of counter-ions (primarily Cl^–), and the osmotic influx of water. The arising swelling of the endosomal compartment and increasing tension of its membrane, together with the destabilizing effect of the protonated polymer on the membrane lead to the loss of integrity of the endosomal membrane and escape of NAs into the cytosol.[128] The buffering function of polyamines (particularly PEI) in the endosome was confirmed by in situ experiment.[129] The transfection efficiency was shown to markedly decrease on going from PEI to its quaternized derivatives.[130] However, the proton sponge concept remains controversial. For example in the above-mentioned study by Mott et al.,[117] transfection was also accomplished by a PEI-based carrier, but the addition of bafilomycin A1, an endosome acidification inhibitor, had no effect on this process. Quite a few arguments pro et contra were summarized by Vermeulen et al.[131] under the title The Proton Sponge Hypothesis: Fable or Fact? Currently, the proton sponge concept has not been abandoned, but has appreciably changed. It is believed that the buffer capacity of a polymer is indeed important and can serve as a criterion for selecting an effective polymeric carrier,[132] but it does not by itself guarantee effective transfection:[133] Nucleic acids are released not as a result of complete lysosome breakdown, but due to local leaks of its contents.[134][135] The polymer itself may possess endosomolytic properties and directly participate in membrane destabilization.[136]

Another difficulty in the targeted delivery of NAs is overcoming the nuclear membrane. The passive transport of NAs into the nucleus is possible during mitosis due to the temporary disassembly of the nuclear envelope; this mechanism makes a higher contribution to the transfection of proliferating cells and a lower contribution for slowly dividing cells such as neurons.[137] The active transport is driven by importin proteins, which recognize specific peptide sequences structurally similar to CPPs, known as nuclear localization signals (NLS). Modification of the carrier with NLS allows nuclear delivery of molecules with a molecular weight (M) of up to 40 kDa; meanwhile, usually, particles larger than 9 nm, which corresponds to M ≈ 1 kDa, cannot pass through nuclear pore complexes.[138]

Although the nuclear membrane is a double bilayer, the mechanism of crossing the membrane is similar to that involved in crossing the cytoplasmic membrane. The positive charge of the carrier ensures adhesion of the complex on the membrane, which may be followed by penetration of either free DNA molecule[139] or the whole NA – carrier complex[140] into the cell nucleus. This enables transfection of post-mitotic, non-dividing cells.[141][142]

Synthetic carriers optimized for one type of nucleic acids are often ineffective for transporting NAs of other types. This may be due to differences in physical properties, such as the size of the biomolecule, or to the need to deliver particular NAs to different intracellular compartments. Kauffman et al.[143] showed that a lipid system optimized toward siRNA is not optimal for the therapy using mRNA. A similar fact was established by Kulkarni et al.[144] for RNA and plasmid DNA (pDNA). Thus, each major type of NA requires a distinct delivery approach. Specific features of the delivery of mRNA, pDNA, short two-stranded RNA (siRNA, microRNA), and ASO are considered below.

Plasmids are circular DNA molecules ranging in length from 2 – 3 to 20 – 100 thousand base pairs, carrying one or more genes. The delivery of DNA vectors opens up the possibilities for three different therapeutic applications:

(1) inhibition of protein expression via RNA interference,

(2) temporary protein production by means of expression vectors,

(3) stable expression or knockout of proteins through gene editing.

For functioning of plasmid DNA, it is necessary that it should enter the nucleus where the transcription apparatus is located. However, the first significant problem along this line is cytoplasm. First, the diffusion of DNA molecules, possessing relatively large size and charge, in viscous cytoplasm is severely limited. Second, DNA is subject to the action of nucleases: it was shown that out of the 2 – 100 thousand copies of the plasmid delivered to the cytoplasm by polyplexes and lipoplexes, only 1 – 10% enter the nucleus.[145]

The efficiency of nonviral DNA delivery largely depends on characteristics of the DNA vector. As a rule, small size of the vector leads to higher transfection: indeed, the efficiency of lipofection of AoSMC cells proved to be 77 times lower for a plasmid consisting of 52 500 base pairs (bp) than for minicircle DNA (2900 bp).[146] Thus, the absence of a bacterial scaffold (such as a bacterial origin of replication and antibiotic resistance genes) makes the use of minicircle DNA a promising strategy for improving the therapeutic efficacy of DNA vectors.[147] The possible causes for this effect include the increased diffusion coefficient and improved nuclear internalization, as well as more effective transcription of gene expression cassettes with minicircle DNA.[47]

Recent publications report successful in vivo delivery of large plasmids (10 – 15 kbp) using polyplexes (e.g., those based on PAMAM-modified polymethacrylate),[148] lipoplexes (based on DOPE and polycationic OH4 lipid),[149] and hybrid systems.[150] All of the above approaches are combined in a common strategy: increasing the positive charge density of the carrier for more efficient compaction of large DNA molecules, which includes the use of dendrimer cationic moieties within a polymer, ethyleneimine trimer as a part of OH4 lipid, and polyethylenimine as the condensing core of the hybrid carrier nanoparticle. This assumption is consistent with the published data[151][152] on the importance of DNA compaction for penetration through the nuclear pore complex of postmitotic cells. A relatively less frequent strategy for improving the efficiency of large-size plasmid delivery is the destabilization of the nuclear membrane, achieved either by enlarging the nuclear pores under the action of dexamethasone or by transfecting cells synchronized in the S-phase.[153]

The delivery object studied most intensively in recent years[41][70-74] is mRNA, a single-stranded molecule with a length of 1000 – 5000 nucleotides. The mRNA configuration includes a cap at the 5'-end, 5'-untranslated region, which is followed by open reading frame, 3’-untranslated region, and 3'-poly(А)-tail (Fig. 7a). Untranslated regions play a key role in regulating mRNA degradation and translation efficiency, with the 3'- and 5'-untranslated regions influencing the mRNA half-life and efficiency of translation initiation, respectively. The issue of mRNA stability is highly acute even in comparison with other typical therapeutic NAs: due to the action of cellular and serum nucleases, the half-life of RNA administered in vivo is approximately 70 s;[154] for pDNA and siRNA, this value is in the range of 5 – 10 min,[155][156] while that for ASO is >1 h.[157] The efforts to enhance mRNA stability include nucleoside modifications [replacement of uridine by pseudouridine (Ψ), especially by N¹-methylpseudouridine (m¹Ψ) (see Fig. 7c)[158]],^f codon optimization, and the use of methyl-guanosine cap analogues for protection against enzymatic decapping.[159] The clinical applications of mRNA include, first of all, mRNA vaccines and mRNA replacement therapy. Meanwhile, mRNA does not need to enter the nucleus; hence, it can be expressed in non-dividing cells, such as liver cells, muscles, and immune cells. This fact accounts for the high value of mRNA for therapy and vaccine development. Irrespective of the particular application, the agent that performs the medicinal function is the encoded protein.

Fig. 7

Structure of therapeutic mRNA: cap modification and replacement of mRNA nucleotides by modified nucleotides important for evading the innate immunity and reducing the rate of hydrolysis (a), chemical formula of 5'-cap: eukaryotic methylated caps^g are usually modified at the first base (designated by А), triphosphate (В), or the second base (C) (b), and pseudouridine structures (с).

Like the use of DNA, the use of mRNA is faced with the problem of inverse relationship between transfection efficiency and the length of the genetic construct. Teko-Agbo et al.[160] demonstrated that upon switching from mRNA with a length of approximately 1 kb to mRNA ~4.8 kb in length, expression of the target protein decreases by two orders of magnitude for the same CPP carrier. The issue of delivery of large RNA molecules is especially relevant in view of the growing interest in self-amplifying RNA (saRNA), in which the target gene is encoded together with alphavirus components. This modification results in higher protein expression per delivered RNA molecule compared to conventional mRNA, which provides a higher efficacy of saRNA vaccines. However, since the size of saRNA is as long as 10 kb, their delivery requires special transfection agents: macromolecular linear PEI[161] or its copolymers with poly(propyleneimine) (PPI),[162] poly(β-amino ester) with increased molecular weight,[163-165] or LNPs with optimized composition.[166-168]

In 2025, branched PEI with M = 22.5 kDa was used to develop saRNA vaccines.[169] According to this study, the key to effective for PEI formulations for saRNA is to switch to a specific format of compact monomolecular polyplexes. In the presence of excess polyethylenimine, some saRNA chains (~3 MDa) rearrange from an extended conformation to a dense globule (Fig. 8), thus forming ultrasmall particles (~30 nm) with a high packing density and relatively low proportion of the polymer in the complex. An increase in the ratio of nitrogen-containing (N) groups of the reagent to the phosphate groups (P) of NA (especially if N : P ≥100) leads to increasing fraction of ‘dissolved’ (not precipitating) saRNA molecules in polyplexes and monomolecular species, as determined by AF4-MALS and AUC analytical methods ^h and a combination of small-angle neutron scattering (SANS) and small-angle X-ray scattering (SAXS); this finding is correlated with the biological activity. Functionally, this is manifested as a gradual increase in luciferase expression in in vitro and in vivo experiments with increasing N:P ratio; in vaccination models, this means that comparable humoral and T-cell responses can be elicited at markedly lower doses of saRNA (in terms of the administered RNA, the activity is enhanced as this ratio increases).[169] Under optimized conditions, an approximately tenfold increase in the activity compared to that of conventional saRNA – PEI polyplexes was demonstrated, along with the absence of noticeable toxicity.

Fig. 8

Schematic picture of mRNA and PEI chains and polyplexes they form.¹⁶⁹ Copyright Nature 2025.

Another important aspect of the cited study is detailed physicochemical verification of the compactness and composition of the complexes: the gyration radius (R_g) of the particles was ~12 – 13 nm and the particles had signs of core – shell structure with PEI-enriched shell, and changes in the RNA secondary structure were observed upon complex formation. In addition, the authors demonstrated the practical ‘pharmaceutical’ applicability of these polyplexes, particularly their high colloidal stability, retention of integrity (or activity) upon dilution and after freezing or drying, protection of RNA from degradation, and the absence of release into serum. In combination, these factors define the PEI-induced saRNA compaction as a versatile tool for the preparation of ultrasmall, stable, and highly active polymeric RNA vaccines or carriers.

Among numerous studies focused on optimization of the composition of the carrier, few studies that propose new strategies can be distinguished. One such strategy is the design of macromolecules directed toward decreasing immune or allergic reactions, e.g., those containing polysarcosine,[170] instead of widely used hydrophilic poly(ethylene glycol) (PEG), or carriers with variable positive charge[171][172] (these issues are described in more detail in Section 4.1).

^f The use of m¹Ψ in mRNA-1273 and BNT162b2 is considered to be a key factor for the high expression and low innate immunogenicity of vaccines.

^g mRNA cap is a modified 5'-end of the molecule usually represented by 7-methylguanosine connected to the first mRNA nucleotide by an unusual 5' – 5' triphosphate bond. The cap protects mRNA from exonucleases, participates in the ribosomal recognition, and enhances translation efficiency.

^h AF4-MALS is asymmetrical flow field-flow fractionation coupled with multi-angle light scattering, AUC is analytical ultracentrifugation.

Small interfering RNA (siRNA, microRNA) are short (~21 – 23 bases) double-stranded RNA able to induce silencing of particular genes. To achieve the desired effect, siRNA must penetrate into the cytoplasm of the target cells; this is followed by cleavage of the duplex with Dicer endonuclease to form the nucleoprotein RNA-induced silencing complex (RISC) suppressing mRNA targets, which subsequently cuts the complementary mRNA target. Small interfering RNA and related microRNA and short CRISPR sequences (crRNA) are specific delivery objects. On the one hand, their small size facilitates penetration into tissues (some cells can take up such oligonucleotides by themselves), and, on the other hand, small-sized NAs are extremely vulnerable to nucleases and are faster eliminated from the body. Effective intracellular delivery of siRNA requires either chemical modification of the molecules or the use of nanoparticles as carriers. Currently, six siRNA drugs have been approved, and only one of them, patisiran (siRNA targeting the transthyretin gene), is delivered using LNPs. Other drugs are amphiphilic conjugates carrying a trivalent N-acetylgalactosamine ligand (Fig. 9). This modification of the siRNA molecule made it possible to develop a series of carrier-free drugs designed for regulating gene expression in the liver (for example, the cholesterol-lowering drug inclisiran).[173] Other conjugates comprise lipids, peptides, sugars, and various functional groups.[174][175] This modification makes it possible to tailor the tissue specificity of the siRNA agent and facilitate cellular uptake by using amphiphilic conjugates. For example, siRNA decorated with a lipid based on distearoylglycerol is not only far superior to nonmodified siRNA in the circulation time and bioavailability, but also surpasses LNPs with the commercial jetPEI carrier regarding the siRNA accumulation rate in a tumour.[176]

Fig. 9

Structures of TriGalNAc (above) and siRNA lipoconjugate (below)

Despite the great clinical success of conjugates, the delivery of siRNA using synthetic carriers is also being actively studied.[110][132][177-181] The double-stranded nature is a key distinction of siRNA from mRNA, which must be taken into account in the development of transfection agents. Hayashi et al.[182] showed that the rigidity of nucleic acid alone can hinder the secondary self-assembly of polyplexes even for the same nature and charge stoichiometry. A comparison of single-stranded RNA (ssRNA, flexible chain) and siRNA (rigid short duplex) with the same sequence of 21 bases showed that at low concentrations, both systems form 1 : 1 adducts with the PEG-PLL block copolymer (i.e., one polymer molecule per NA molecule) and with single polyion complexes (uPIC). However, above the critical association concentration, the behaviour of uPIC is appreciably different: ssRNA – uPIC are aggregated to micelles of ~30 – 70 nm size, while siRNA – uPIC retain stability as small complexes (~10 nm). This difference is crucially important for systemic administration:ⁱ small and well-determined size of uPIC should ensure more predictable pharmacokinetics, decrease the tendency toward concentration-dependent aggregation, and reduce the probability of non-specific capture of particles by the mononuclear phagocytic system compared to that for larger and more polydisperse associates.

Apart from conjugation, an integral feature of siRNA is the possibility of chemical modification of the proper oligonucleotide chain, with the goal to enhance binding to target sequences, increase the stability against nucleases, optimize the pharmacokinetic properties, and minimize side effects. Modifications may affect the phosphate backbone, sugar residues, and nitrogenous bases (Fig. 10). The following changes are most popular in the siRNA chemistry: replacement of the ribose 2'-OH group by 2'-OMe and 2'-F groups; replacement of phosphate groups by phosphorothioate groups, typically two at the 5'-end of the sense strand and at both ends of the antisense strand (Fig. 11).[183]

Fig. 10

Structures of nucleotides and their modified analogues

Fig. 11

Structures of siRNA-based drugs

In addition, various oligomerization strategies based on covalent or ionic binding are used to increase the stability of therapeutic siRNA in the body. At the laboratory level, good results were achieved for disulfide cross-linked linear poly-siRNA, thiol–maleimide cross-linked branched poly-siRNA, and siRNA oligomerized owing to short complementary A(5 – 8)/T(5 – 8) 3'-overhanging ends.[184]Oligomeric RNAs are the subject of a review by Kim et al.[185] published in 2022. Oligomerization can be achieved by introducing a thiol group at the 5'-end of the sense and antisense sequences (Fig. 12a),[186] which gives rise to siRNA as a mixture of oligomers sensitive to redox processes. ‘Polymerized’ siRNA forms smaller and more stable NPs with PEI and ensures more effective decrease in the target gene expression than ‘monomeric’ siRNA. Polyacrylamide gel electrophoresis reveals a broad molecular weight distribution of ‘polymerized’ siRNA and the regeneration of the monomeric form on treatment with dithiothreitol (see Fig. 12b).

Fig. 12

siRNA oligomerization strategies: introduction of a thiol group at the 5'-end of the sense and antisense sequences (a) and results of polyacrylamide gel electrophoresis of ‘polymerized’ siRNA (b).¹⁸⁶ Copyright Elsevier 2010.

Hong et al.[187] demonstrated the effect of branching reagents (Fig. 13). As the number of branches increased, the size of the polyplexes of microhydrogels with linear PEI decreased: ~100 nm for YY-siRNA, ~215 nm for M-siRNA, and micrometre-scale aggregates for ‘monomeric’ siRNA. Simultaneously, the efficiency of knockdown increased. For example, upon transfection of the MDA-MB-435 cells with the YY-siRNA – PEI system, the expression of the GFP gene decreased to 27.9% compared to the control, whereas the use of M-siRNA and ‘monomeric’ siRNA resulted in expression values of 70.5 and 87.7%, respectively.[187]

Fig. 13

Schematic diagram of the preparation of microhydrogels based on 3’-mercapto siRNA derivative involving dimeric and trimeric maleimide linkers.¹⁸⁷

ⁱ Administration methods of therapeutic agents fall into two main categories: systemic (affecting the whole body) and topical (targeting a particular site).

Antisense oligonucleotides (ASO) are short (usually, 15 – 25 nucleotides long) single-stranded molecules complementary to definite mRNA targets (Fig. 14). These compounds do not trigger the RNA interference mechanism; the mechanism of their action involves blocking of translation or change in splicing.[188] Currently, they are considered to be the most numerous class among approved NA-based pharmaceuticals; and many of them are gapmers, that is, compounds in which two series of modified ribonucleotides are separated by a sequence of deoxyribonucleotides (Fig. 15). The chemical structure of ASOs generally undergoes more extensive chemical modification than siRNAs. As a rule, phosphorothioate linkers completely replace the phosphate groups, and if the nucleoside is not a deoxyribose derivative, the 2'-hydroxyl group is always protected (most often, by the methoxyethyl protection). In addition, the 5-position of the pyrimidine base is often modified in ASO.

Fig. 14

Structures of modified nucleotides

Fig. 15

Structures of pharmaceutical agents based on antisense oligonucleotides

Nucleoside analogues other than carbohydrates have also been used in the ASO design, e.g., morpholine derivatives (already FDA-approved), propylene glycol derivatives (glycol-nucleic acid, GNA), N-(2-aminoethyl)glycine derivatives (peptide-nucleic acid, PNA), etc., and the resulting compounds are being evaluated in clinical trials.[174][189][190] Chemical modi­fications endow ASO molecules with the ability to bind to blood plasma proteins and penetrate into cells without the need for special carriers (for example, nusinersen is administered into the cerebrospinal fluid as an aqueous solution). However, some approved ASO-based drugs are GalNAc conjugates, which gives them considerable advantages. For example, eplontersen modified at the 5'-end is administered to patients in a five times lower dose and four times less frequently than the analogous inotersen drug without a ligand.[188] Synthetic carriers, both lipid- and polymer-based ones, are still rarely used for ASO. They serve for addressing special issues such as

— targeted delivery of ASO to tumour tissues (an example is four-component LNP with PEG-lipid modified with iRGD peptide, which contains the arginine – glycine – aspartic acid, the integrin receptor recognition site of cell membranes),[191]

— delivery of ASO to sentinel lymph nodes to suppress metastasis (e.g., glycine – lysine oligopeptide with a PEG block optimized in length),[192] or

— crossing the blood – brain barrier in the central nervous system by ASO [e.g., copolymer of thiol- and nicotinamide-modified poly(L-lysine) with PEG].[193]

Macromolecules with positively charged groups are traditionally considered as polymeric carriers for NAs. The vast majority of these polymers contain amino groups or ammonium cations (R₃N⁺), although there are also examples of macromolecules with phosphonium (R₃P⁺) or sulfonium (R₂S⁺) cations. The main structural types of polymers that are of interest for NA delivery can be divided into two fundamentally different groups. The first group comprises polycations, which may optionally include a reactive group intended for subsequent functionalization, or a structural moiety that serves as a receptor for targeted delivery. The second group consists of amphiphilic macromolecules designed according to the all-in-one principle. These polymeric carriers for NAs combine all (or almost all) required functions in one macromolecule. A classic example of these systems are charge-altering releasable transporters (CART), polymeric delivery systems developed by Waymouth’s research team (see Section 4.1). The structural types of polymers of the former group can obviously be used for the design of more complex systems.

Both natural and synthetic polymers are used for NA delivery. Among natural polymers, a lot of attention has been devoted to proteins (in particular, histones enriched with ionizable L-lysine and L-arginine moieties, Fig. 16).[76] In a study published in 2020, Kuzmich et al.,[194] demonstrated high efficiency of the histone H2A adduct with the platelet-derived growth factor receptor (PDGFR) for the delivery of DNA to the PDGFR-positive gastric cancer cells. Histones are nontoxic; however, their clinical use raises reasonable concerns due to the risks of interfering with the natural processes of chromatin formation.

Fig. 16

Main structural types of polymeric NA carriers

Gelatin is much safer and more readily available than histones and is attractive because of the presence of RGD sequence, which promotes cell adhesion.[76] However, there are no recent publications on the NA delivery using gelatin, presumably due to the greater efficiency of synthetic polymers and chitosan (see below). The application of L-arginine-enriched protamine (see Fig. 16) for the NA delivery is the subject of a review by Pascolo and co-workers.[195] Furthermore, promising results in the cancer therapy were demonstrated by protamine-containing agents based on mRNA[196][197] and siRNA.[198][199] Protamine forms stable complexes with NAs and, therefore, it can be used as a part of transfecting agents that are effective even upon intranasal administration.[200]

DEAE-dextran (the structure is presented above), one of the first polymeric NA delivery agents, can also be classified as a polymer of natural origin.[23-25] The relatively low efficiency, inability to form stable polyplexes, and proven immunogenicity and cytotoxicity[201] resulted in the loss of interest in this carrier. However, in 2019, Siewert et al.[202] reported the use of DEAE-dextran as a model compound for the preparation of mRNA polyplexes. Cyclodextrin is not a polycation; however, when combined with the block copolymer composed of methyl­polyethylene glycol (mPEG), ε-caprolactone (εCL), and PEI, it forms a hydrogel for the delivery and controlled release of pDNA, which lasts for up to 7 days.[203]

Chitosan remains an intensively studied polycationic polymer of natural origin; it is the subject of a number of recent reviews.[36][94-96] The safety of chitosan-containing formulations is still an open issue, with potential complications associated with the use of chitosan being related to the nature of the source and the isolation and purification methods. This is a common problem in the biomedical application of naturally occurring polymers (a classic example is the contamination of silk fibroin with sericin),[204] which is an argument in favour of developing synthetic polymers with strictly defined composition, microstructure, and topology.

Structurally related PAMAM, PEI, PPI[91-93] should rather be classified as oligomers with a dendrimer topology; they represent relatively simple organic molecules containing a large number of ionizable moieties of primary, secondary, and tertiary aliphatic amines. These polycations have been widely studied and continue to be studied as NA delivery agents, as they have high complex-forming capacity. However, use of these compounds into clinical practice is reasonably complicated by the high toxicity characteristic of polyamines. Meanwhile, the application of these polymers in genetic techniques ex vivo is fairly efficient. In 2025, a manufacturer reported the registration of a linear PEI-based reagent for use in the production of adeno-associated viral vectors.[205] The benefits of PEI are synthetic availability and high buffer capacity. Numerous studies addressing the decrease in the toxicity of this polymer by introducing biodegradable moieties into the molecule are covered in a recent review.[206]