RNA nanostructures for targeted drug delivery and imaging

ABSTRACT

The RNA molecule plays a pivotal role in many biological processes by relaying genetic information, regulating gene expression, and serving as molecular machines and catalyzers. This inherent versatility of RNA has fueled significant advancements in the field of RNA nanotechnology, driving the engineering of complex nanoscale architectures toward biomedical applications, including targeted drug delivery and bioimaging. RNA polymers, serving as building blocks, offer programmability and predictability of Watson-Crick base pairing, as well as non-canonical base pairing, for the construction of nanostructures with high precision and stoichiometry. Leveraging the ease of chemical modifications to protect the RNA from degradation, researchers have developed highly functional and biocompatible RNA architectures and integrated them into preclinical studies for the delivery of payloads and imaging agents. This review offers an educational introduction to the use of RNA as a biopolymer in the design of multifunctional nanostructures applied to targeted delivery in vivo, summarizing physical and biological barriers along with strategies to overcome them. Furthermore, we highlight the most recent progress in the development of both small and larger RNA nanostructures, with a particular focus on imaging reagents and targeted cancer therapeutics in pre-clinical models and provide insights into the prospects of this rapidly evolving field.

KEYWORDS:

• RNA nanostructures
• RNA therapeutics
• targeted drug delivery
• bioimaging
• aptamers
• tumour targeting
• personalized nanomedicine
• bioconjugation
• cancer

Introduction

Cancer is one of the main causes of mortality worldwide, and conventional therapy has proven insufficient with regard to safety, targeting accuracy toward diseased tissues, and efficacy. To mitigate these obstacles, the idea of a drug designed to selectively eliminate diseased tissue or cells in the human body like a ‘magic bullet’ led to the foundation of targeted drug delivery strategies proposed by Paul Ehrlich more than a century ago [1]. Unlike conventional therapeutics, targeted therapies aim at improving the therapeutic index by increasing local drug concentration at the site of the targeted diseased tissues or organs and, concurrently, reducing toxicity in healthy tissues. By the same principle, non-invasive imaging techniques, such as those relying on fluorescence or radiolabeled tracers can significantly benefit from targeted approaches when used for diagnostic and prognostic purposes. As a result, high-contrast imaging and real-time detection can provide a more detailed molecular characterization of the diseased stage [2] and potentially improve the prediction of a patient’s outcome.

Targeted multifunctional nanoparticles (NP) can be built to have both diagnostic and therapeutic capabilities, also termed theranostics [3]. This requires platforms that can be functionalized with targeting ligands, payloads, and tracers to yield a single delivery system. Beyond combining therapeutic and diagnostic elements, these delivery platforms may also include measures to extend circulation time, reduce toxicity, and activate the drug at the diseased site [4]. One class of delivery platforms can be built by the use of non-covalent hydrophobic or ionic interactions to form lipid NPs and polyplexes, respectively. These delivery systems protect loaded drugs, such as nucleic acid, from enzymatic degradation and can be functionalized for targeted delivery and imaging (reviewed elsewhere [5,6]). Besides these widely applied systems, other types of platforms have emerged from the recent development of micro- and nanomaterials for the construction of scaffolded multifunctional nanostructures.

This review focuses on an emerging class of delivery platforms based on nucleic acid biopolymers, specifically DNA and RNA. Their unique programmability, arising from canonical Watson-Crick and non-canonical base pairing rules, facilitates the self-assembly of nanostructures with diverse sizes and shapes. Additionally, co-transcriptional folding into complex secondary and tertiary structural motifs like hairpins, stems, loops, bulges, and crossovers [7–9], gives rise to structures such as aptamers, ribo- and DNA-zymes, nanomachines, and more sophisticated ‘origami’ structures [10]. These biopolymers can readily be chemically modified with various compounds to protect against nucleases and decorated with functionalities through bioconjugation. In certain cases, precise control over spatial arrangement and valence of conjugated compounds at the nanoscale level can be achieved, resulting in structures with proven therapeutic effects [11,12]. In nature, RNA plays a vital role in numerous crucial cellular processes, including the storage of information as messenger RNA (mRNA) and regulation of gene expression through RNA interfering (RNAi) elements such as microRNA (miRNA) and small interfering RNA (siRNA). This offers intriguing functional advantages over DNA yielding a broader spectrum of possible functionalities. One of the most favorable benefits is RNA’s biocompatibility, as most RNA sequences are generally welltolerated in contrast to DNA, which usually triggers a strong immune response when present in the cytoplasm. This advantage confers potential to RNA-based nanostructures (RNs) to become biocompatible therapeutic and targeting agents to engage with biological processes. For instance, aptamers and RNAi elements could be designed and incorporated into an RNA-based scaffold, to specifically interact with a membrane receptor and hinder mRNA translation, respectively. The prominent versatility of RNA and its biocompatibility have paved the way for the application of RNA nanotechnology in targeted delivery in vivo through the construction of RNs for precision therapeutic, diagnostic, or theranostic medicine.

In this educational review, we discuss considerations to take into account in the design and delivery of structurally defined RNs by introducing the barriers encountered in vivo and solutions adopted to fully unlock the potential of RNs for delivery. Then, we focus on the advancements in nucleotide modifications and RNA functionalization, which have fuelled the success of targeted delivery of RNs in vivo, by conferring them more stability and additional functionalities. Lastly, we discuss recently developed RNs that have been applied to pre-clinical studies of cancer-specific targeted imaging and drug delivery and conclude with a brief outlook on the future clinical translation of RNs.

Considerations for the design of RNs applied in vivo

Prior to designing and assembling RNs for in vivo applications, it is crucial to understand the biological and physical barriers these RNs must overcome once administered (overview in Figure 1). To yield RNs with efficient targeting and delivery into tissues in vivo, the design of the RN should ensure structural stability under physiological conditions, minimize immunogenicity, optimize pharmacokinetics and biodistribution, and allow functionalizing the nanostructure with one or more targeting ligands, detection elements or drugs [13]. Additionally, the design of nanostructures intended for intracellular delivery should enable efficient cellular internalization and endosomal escape. Finally, in preparation for in vivo testing of RNs delivery, careful consideration should be given to selecting the optimal administration routes, which is crucial to maximize the RNs accumulation at the target site and minimize adverse side effects. This section focuses on how these prerequisites are challenged by biological barriers and can be addressed.

Figure 1. Overview of barriers to RNs delivery in vivo. A) The design of a targeted RN for the delivery of a drug (for instance RNAi elements), for molecular imaging, or theranostic purposes, should take into account the physical and biological barriers encountered in vivo. (1) Once in the bloodstream, the RNs are susceptible to degradation by endo- and exonucleases. (2) They can be intercepted by the immune system via macrophages or tagged by opsonins. (3) This process leads to the formation of a protein corona, which, together with other serum proteins adhering to the nanostructure’s surface (4) may affect the bioavailability and biodistribution of the RNs. B) To reach its target site, (5) the RNs must extravasate the endothelial cells lining the blood vessels . For drug delivery, cell internalization through receptor-mediated endocytosis is crucial. (6) In order to release the payload, the RNs must therefore escape the endolysosomal pathway to execute its mode of action.

Stability of the RNA biopolymer

RNA is unstable and prone to chemical and enzymatic degradation, warranting protective measures that ensure the stability of the RNA and allow evasion from nucleases [14,15] after administration in vivo (Figure 1 A-1). To address this, chemically modified nucleotides are introduced into the RNs’ sequence to extend their stability, thus prolonging their half-life in vivo. This has been demonstrated through the incorporation of modifications, either alone or in combination, for siRNAs [16,17], aptamers [18], or RNs [19]. Chemical modifications can be introduced either during chemical synthesis or enzymatically during in vitro transcription [20,21], using engineered RNA polymerase mutants [22–24]. While the extensive number of chemical modifications surpasses the scope of this review, we briefly highlight the most common modifications of the phosphate backbone, sugar moiety, or nucleobase (Figure 2).

Figure 2. Examples of commonly used chemically modified nucleotides. RNA nucleotides can be chemically modified in the phosphate backbone (red), nucleobase (blue), or through substitution of the 2’-OH (green) to improve the thermal and serum stability of oligonucleotides. Common modifications: phosphorothioate RNA (PS-RNA); 2’-O-methyl RNA (2’-OMe); 2’-fluoro RNA (2’-F RNA); 2’-O-methoxyethyl RNA (2’-MOE); locked nucleic acid (LNA); phosphoramidite morpholino oligomer (PMO); N1-methylpseudouridine (m1Ψ). Chemical structures are created in the software ChemDraw professional 17.1.

Phosphorothioate (PS) modifications constitute a replacement of the non-bridging oxygen of the phosphate backbone with sulphur and were the first to be implemented in oligonucleotides. PS modifications confer nuclease resistance [25] and reduced immunogenicity [26]. PS modifications are widely used in oligonucleotide therapeutics, such as in the FDA-approved siRNA Givosiran, by Alnylam Pharmaceuticals. Interestingly, PS-modified oligonucleotides are known to associate with serum proteins [27,28], and with careful design, this can be exploited to extend their circulation half-life [29].

One of the main characteristics of RNA in comparison to DNA is the presence of the 2’-OH group in the sugar moiety. This group is involved in phosphodiester exchange and cleavage of the original phosphodiester bond and renders the RNA unstable at higher pH and in environments where RNases are present [30]. Substitution of the 2’-OH group with 2’-fluoro (2’-F), 2’-O-methyl (2’-OMe) or 2’-O-methoxyethyl (2’-MOE) can prevent RNA degradation, increase the melting temperature of double-stranded regions and reduce immunogenicity [15]. Another commonly used 2’ modification is the locked nucleic acid (LNA) where a methylene bridge connects the 2’ and 4’ carbons of the sugar moiety and locks the ribose in C3’-endo conformation in dsRNA. When hybridized with complementary RNA, this modification confers a greater thermostability and higher nuclease resistance [31].

An alternative type of modification that has been used in the development of several antisense oligonucleotides (ASOs), including eteplirsen (Exondys 51) and golodirsen (Vyondys 53) by Sarapta Tx for blocking RNA splicing and translation, is the phosphorodiamidate morpholino oligomer (PMO) [15]. Here, the ribose moiety is replaced with a morpholino subunit, which neutralizes the charge of the phosphate backbone, leading to poor association with serum proteins. Due to this and its incompatibility with RNase H activity, PMO has gained much interest in engineering ASOs.

Chemical modification of the nucleobase is a strategy adopted with the overall aim of improving binding interactions, stability of nucleic acid-based drugs, and reduce immunogenicity. In the early 2000’s Karikó and Weissman incorporated enzymatically naturally occurring nucleosides, such as the pseudouridine (Ψ) and its analogs, into mRNAs to evade toll-like receptor 7 (TLR7) and TLR8 mediated inflammatory response and improve translational efficiency in mammalian cell culture and mice [32,33].

Immunogenicity, pharmacokinetics, and biodistribution

The immune system protects from harmful foreign agents to maintain the safeguard of the body. RNs delivered in vivo may represent an alarming trigger for the innate immune response, recruiting cells of the immune system, such as macrophages (Figure 1 A-2), to detect and eliminate RNs via their so-called pattern recognition receptors (PRR), such as TLRs, RIG-I-like receptor (RLR) and protein kinase R (PKR) [34,35]. Although the activation mechanisms of immunoreceptors are not entirely elucidated, it is well-established that the immune response triggered by RNs is highly dependent on their size, shape, and composition [36–39]. In some cases, the triggered immune response can be exploited by tailoring the RN to confer immunomodulatory properties that redirect the immune response towards recognition of certain antigens found on cancer cells [40]. While a few interesting examples of this beneficial immune stimulation are showcased in the subsection ‘Molecular imaging and delivery of immunotherapies and chemotherapeutic drugs’ (within ‘RNA nano- and microstructures’), the general interest for RNs applied in vivo is to be well-tolerated or able to evade the immune system. An evasion approach generally applied to NPs, is the coating with ‘self’ biomolecules or peptides such as those computationally designed from the human CD47 by Rodriguez and colleagues [41], which reduced macrophage-mediated clearance of the NPs. As mentioned in the previous section, another strategy to reduce immunogenicity is the incorporation of modified nucleotides into the scaffold of RNs. This allows the RN to evade the immune system, extend its circulation time, and reach therapeutic concentrations at the diseased site [16,42,43].

The composition of an RN can largely affect its physicochemical properties, thereby dictating how the nanostructure will interact with the biological environment in circulation. In some cases, a protein corona can form around the nanostructure through non-covalent interactions with serum proteins (Figure 1A-3) [44,45]. Depending on its composition and distribution, the protein corona can have beneficial or adverse effects on RN functionality [46], which sparked significant efforts dedicated to understanding protein corona formation around NPs. These include studies focusing on the effects of the corona on the NP’s physicochemical properties and activity through incubation with blood plasma [47], but also proteomic profiling approaches focusing on the effects NP properties have on corona protein adsorption [48]. The protein corona mediates adverse effects including shielding of targeting moieties displayed by RNs [46], causing inefficient binding and potential off-target accumulation, and tagging the nanostructure with immunoglobulins and complement proteins (opsonins) [49] to target the mononuclear phagocytic system (MPS), which rapidly sequesters them from the blood, reducing their bioavailability and activity [50,51]. On the other side, the beneficial effects include adsorption to dysopsonins, such as serum albumin and clusterin, resulting in increased circulation time of smaller RNs and improved targeting ability, stealth, and safety profile for in vivo applications [50,52,53]. This adsorption can also be achieved through the introduction of a hydrophobic group like palmitoyl in the RNA scaffold which can stably bind albumin and lipoproteins [54,55]. An alternative strategy successfully applied to different types of NPs [45], is precoating or functionalizing the RNs with proteins such as albumin to potentially improve pharmacokinetics [56], as well as targeting, and internalization.

The increase in size due to the protein corona coating the surface of the RN could also influence its biodistribution, bioavailability, and clearance leading to high accumulation in organs such as the liver, spleen, and kidneys [13] (Figure 1A-4). For example, NPs with a diameter less than 5–6 nm are subject to rapid glomerulus filtration in the kidneys [57], while larger NPs (>100 nm) are retained or intercepted by cells of the MPS in the liver, spleen, and lymph nodes [57,58]. To address this, Jasinski et al. have investigated how the control of RNs’ size and shape could influence their biodistribution upon intravenous (i.v.) injection [59]. They found that squares measuring 5 nm were primarily secreted by the kidneys while 10 and 20 nm-large RNs appeared to have affinity for serum proteins and display a tendency to interact with hepatic macrophages, leading to a slower clearance. Along with size, the shape of RNs seemed to also influence clearance as triangle-, square- and pentagon-shaped RNs (sharing the same size along their edges) exhibited different biodistributions upon 12 hours of administration, highlighting the involvement of different secretion pathways. While the triangle was barely visible in clearing organs, the square primarily accumulated in the kidneys and the pentagon was substantially concentrated in the spleen.

Unfavorable pharmacokinetics and clearance could be corrected by accessorizing RNs with polyethylene glycol (PEG) molecules [60], or cholesterol [61] conferring stealth properties and a prolonged circulation time, overcoming renal filtration [49].

RN functionalization with targeting ligands and imaging tracers

RNs can be rationally designed for targeted drug delivery and imaging through functionalization with targeting ligands and detection elements (Figure 1B-5) to create a new and promising generation of multifunctional RNs. The following sections describe the most popular conjugation strategies adopted to functionalize RNs and the most relevant targeting ligands and imaging tracers that have been used for targeted drug delivery and imaging in pre-clinical settings.

Strategies for functionalization of RNA

The functionalization of RNs is crucial for targeting, delivering payloads, or visualizing specific tissues for biomedical applications [62]. Functionalization of RNA can be achieved non-covalently, for instance, through charged interactions or annealing with other oligonucleotides, or covalently, by conjugating functional groups either internally or at the 5’ and/or 3’ ends of the RNA sequence. Building on a versatile platform of commercially available modified oligonucleotides, RNA molecules can be functionalized at specific sites through the conjugation of payloads, tracers, or targeting ligands.

Coupling to biomolecules, such as antibodies, proteins, or peptides, is frequently done by reacting to primary amines (-NH₂) and thiols (−SH) on the side chain of lysine residues or N-terminus and cysteine residues, respectively. Primary amines can react with many reactive groups via acylation or alkylation. An example is the rapid reaction between -NH₂ and an activated N-hydroxysuccinimide (NHS) ester that forms a stable amide bond with high yields [63] (Figure 3B). Thiols can rapidly react with maleimide or haloacetyl (iodoacetyl, bromoacetyl) groups to form stable thioether bonds via a Michael-type addition reaction at pH 6.5–7.5 (Figure 3A). Prior to conjugation to thiol-containing compounds, a reduction step is required to prevent the formation of disulfide bonds [63,64]. Despite the popularity of maleimide-thiol reactions, the instability of the thioether bond in the presence of excess thiols, as in biological environments, confers some limitations to this method and makes it less suitable for bioconjugate production for in vivo purposes [64].

Figure 3. Bioconjugation reaction schemes. A) Amine-NHS reaction. B) Thiol-maleimide reaction. C) Copper(I)-catalyzed azide-alkyne addition (CuAAC). D) Strain-promoted azide-alkyne cycloaddition (SPAAC). E) Inverse-electron demand diels-alder reaction (iEDDA). R1 and R2 indicate the position of RNA oligonucleotide and any molecule, respectively, with which RNA can be chemically functionalized. Chemical structures were created in ChemDraw professional 17.1.

Bioorthogonal reactions prevent undesired cross-reactivity with other compounds present in a biological system. The technique ‘click’ chemistry [65,66] not only offers orthogonality but also fast reaction kinetics under physiological conditions and high yields of one major product [67]. In copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC), the triazole ring formation is catalyzed by Cu(I) (Figure 3C) [63,67]. However, the toxicity of copper makes this type of click chemistry less compatible for use in biological systems [68]. As an alternative solution, Cu-free, strain-promoted azide-alkyne cycloaddition (SPAAC) [69] (Figure 3D) has gained popularity due to its biocompatibility and simple reaction conditions [67]. Here, the reaction kinetics are driven by ring strain in alkyne-containing strained cyclic rings, such as dibenzocyclooctyne (DBCO) or bicyclo[6.1.0]nonyne (BCN), that destabilizes the alkyne to push the equilibrium [70].

Inverse electron demand Diels – Alder (iEDDA) biorthogonal reaction exceeds both CuAAC and SPAAC with regard to kinetics, orthogonality, and biocompatibility [70,71]. iEDDA reactions occur between tetrazine- and trans-cyclooctene (TCO)-containing compounds to form a stable dihydropyridazine bond (Figure 3E). One major disadvantage of iEDDA is the large size of the reactive groups (tetrazine and TCO) compared to those used in CuAAC or SPAAC, which may affect the functionality of the conjugated biomolecules [67].

Targeting ligands

Specific targeting can be achieved by identifying and exploiting biomarkers such as surface proteins with upregulated or specific expression in only diseased cells or tissue. To confer RN targeting functions, natural or artificial ligands, such as small molecules, carbohydrates, peptides, proteins, monoclonal antibodies (mAbs), and aptamers binding with high affinity to specific biomarkers, can be chemically conjugated to RNA, and enable selective delivery of payloads or tracers to target tissues [72,73].

The choice of the best conjugation strategies is often dictated by the type of functional molecules selected to confer the RN targeting, detection, or therapeutic activity. For this reason, the selection of appropriate and specific biomarkers is paramount in the field of precision medicine, particularly in oncology. Biomarkers may be physiologically expressed in several tissues but are mutated or highly overexpressed on the cell surface of well-characterized tumor cells [74–76]. However, the intratumoral heterogeneity often constitutes a large variability in the extent and homogeneity of the expression. This might affect the efficiency of the targeting action of the developed RN and, particularly for therapy, introduce off-target effects and adverse reactions [77]. Therefore, ligands binding their cognate receptors with high affinity and specificity are essential to drive the targeting action of an RN primarily to diseased tissues. Such high-performance binders can be identified by applying stringent conditions to the selection of mAbs, antibody fragments, peptides [78], and aptamers [79]. Alternatively, the affinity and specificity of weaker ligands can be enhanced by multivalent and multispecific displays on the surface of the RN. For instance, when the targeting action is directed to cellular receptors, a multivalent system could improve the overall binding strength, also referred to as avidity, to the target cell and additionally favor cellular internalization of drug payload via receptor-mediated endocytosis (RME) [80].

Among the class of protein ligands, mAbs serve as excellent targeting agents due to their high specificity, in vivo stability, and affinity towards a multitude of targets. Owing to the hybridoma technology [81] and phage display, numerous mAbs have been selected against biologically relevant biomarkers [82], and many function as therapeutics in cancer therapy [83–85]. However, functionalization of RNs with mAbs is not a commonly employed strategy due to their large size (~150 kDa). Moreover, their random conjugation through primary -NH₂ onto RNs impairs control of the conjugation site, potentially affecting their targeting activity, and/or exposing their Fc regions to macrophages belonging to the MPS [86,87]. The smaller size of antibody fragments such as the variable domain of heavy-chain-only antibodies, also known as nanobodies (12–15 kDa), favors ligand multimerization on an RN. Nanobodies are less immunogenic and exhibit structural and binding characteristics similar to those of conventional antibodies [88]. Their facile production in bacteria and yeast facilitates recombinant engineering to introduce chemical handles for site-directed bioconjugation [89,90], allowing better control of their orientation. Despite the limited literature documenting the multimerization of nanobodies on RNA scaffolds for in vivo applications, successful work in vivo of genetically fused or scaffolded nanobodies for multivalent and multiparatopic displays [91,92], encourages optimism about its potential to create powerful RNs for targeted delivery.

Another class of small protein ligands suitable for multimerization on RNA scaffolds are peptides, consisting of short chains of linear or cyclic amino acids (<100 amino acids) with low immunogenicity and high stability. Peptides with strong affinity and specificity can derive from natural ligands or be selected from artificial libraries against various molecular targets, often by phage display technology [93].

As previously mentioned, RNA can not only act as a scaffold or as a drug but it can also fold into small structures and be able to function as a ligand, binding molecular targets with great affinity and specificity in the low nanomolar range. This class of binders is known as aptamers, which are composed of 30–100 base pairs (bp) single-stranded DNA (ssDNA) or RNA folding into unique tertiary shapes that are critical for target recognition and interactions [94]. Aptamers can be selected against virtually any type of biological molecular target by the Systematic Evolution of Ligands by Exponential Enrichment (SELEX) technology [95]. Numerous therapeutic aptamers with inhibitory functions have been selected yielding two FDA-approved aptamers [96,97] and many others in pre-clinical testing for various diseases [98,99]. Compared to other targeting ligands, aptamers production is benefited by the low batch-to-batch variability inherent to nucleic acid synthesis. Further advantages are their unmatched structural and chemical flexibility for site-specific conjugation and controlled functionalization through the introduction of chemically modified nucleotides. In addition, RNA aptamers can be readily incorporated into the RNs through co-transcription of the targeting aptamer with the scaffold and other RNA-based elements. However, the binding affinities of aptamers under physiological conditions are not always as robust and consistent as those of mAbs, nanobodies, and peptides, in part due to the in vitro selection conditions, variations in the methods used for validating binding specificity, and determination of K_D values. It is therefore recommended to rigorously validate the binding ability of the selected aptamer under physiological conditions

Another class of ligands is represented by small molecules naturally capable of targeting cancer cells, such as the vitamin folic acid, that binds to the folate receptor α (FR-α) overexpressed particularly in breast, ovary, and lung cancer [100]. Interaction of folate with the FR-α facilitates RME, which can be exploited for targeted intracellular delivery of payload in cancer therapy and bioimaging [101]. One well-characterized small ligand is the N-acetyl galactosamine (GalNAc) targeting the asialoglycoprotein receptor (ASGPR) overexpressed on hepatocytes (liver cells) [102]. An example of successful trivalent GalNAc-mediated (triGalNAc) targeting and internalization is the drug Givosiran (Alnylam Pharmaceuticals), which efficiently delivers conjugated siRNA in the liver for the treatment of acute hepatic porphyria [103].

Despite the advancements in transcriptomics and proteomics, which have significantly propelled biomarker discoveries in oncology, many cancer types and diseases still lack well-characterized molecular targets ideal for targeted delivery and have to rely on delivery systems exploiting other characteristics of tumors, such as low intratumoral pH [104] or hypoxia [105].

Nevertheless, the growing success of targeted delivery undoubtedly amplifies the endeavors toward the discovery of novel biomarkers, thereby emphasizing the need for targeting ligands that meet the demands for the design of more advanced and sophisticated targeting systems.

Fluorophores and radionuclides for molecular imaging

Molecular imaging of tumors employs targeting ligands to visualize diseased tissues and quantify the extent of the prognostic molecular target expression to evaluate the stage of development. Targeted RNs can be designed to function as tracers used in molecular imaging by functionalizing them with tracers such as fluorophores and radionuclides.

Optical systems such as fluorescence-based detection are the most popular choices to image molecular targets in preclinical models thanks to their large accessibility and straightforward readout. A large variety of fluorophores are commercially available bearing different handles for chemical conjugation, making the functionalization of RNs simple and adaptable for in vivo applications. Fluorophores in the visible spectrum could be particularly useful in applications such as image-guided surgery for tumor resection [106] but molecular imaging of deeper tissues is strongly affected by the interactions between tissue and photons hampering light propagation. These interactions mostly occur in the form of photon scattering and absorption, which decrease the imaging depth, resolution, and signal detection [107]. Illumination in the visible range is also less desirable due to significant autofluorescence generated within tissues [108]. Yet, the near-infrared (NIR) range of the spectrum has favorable propagation properties that enable imaging up to several centimeters deep into the tissue which is sufficient in many murine models [109,110]. This improved transparency due to reduced absorbance, scattering, and autofluorescence results in enhanced spatial resolution of the imaging [106].

There are two classes of NIR dyes; traditional NIR-I dyes with excitation (Ex) between 700–900 nm; and newly introduced NIR-II organic dyes with Ex between 1000–1700 nm (see overview in Table 1). The physicochemical characteristics of fluorophores, such as type of conjugation chemistry, numbers of fluorophores and their distribution, charges, and hydrophilicity may alter the RN’s biodistribution [116–118], confer toxicity [119], affect ligand binding, internalization, and the RN’s dynamic response in vivo [120]. For more information on the physicochemical properties of NIR dyes and their applications, we refer the reader to the following reviews [121,122].

Table 1. Fluorescence-based bioimaging.

NIR-I Dyes
Name	Ex/Em (nm)	Properties	Conjugation chemistry
Indocyanine green (ICG) [111,112]	807/822	Negatively charged; Amphiphilic to lipophilic; low toxicity in vivo	NHS/DBCO
Non-sulphonated Cyanines [113]	747/774 (Cy7) 788/808 (Cy7.5)	Positively charged; hydrophobic, poor water solubility	NHS/DBCO/Maleimide/Azide/Alkyne/Amine/Carboxylic acid
IRDye800-CW [114]	774/789	Negatively charged, highly water soluble	NHS/DBCO/Maleimide/Azide/Alkyne/Carboxylic acid
NIR-II Dyes
CH1055-PEG [115]	750/1055	Soluble in DMSO	− COOH functional groups for conjugation to amine

To date, fluorescence imaging is not suitable for whole-body scans of larger animals and humans due to its poor tissue penetration [106]. Alternatively, radioactive tracers such as radionuclide-labelled molecules can be used to achieve improved sensitivity with very high penetration depth. The signal can be visualized through detection systems like single-photon emission computed tomography (SPECT) and positron emission tomography (PET) [123]. Both systems lack anatomical information and rely on the integration of computed tomography (CT) or magnetic resonance imaging (MRI) for reconstruction of the spatial distribution of the radioactive tracer inside the body [123,124].

Radiolabeling of RNs destined for imaging purposes should generally involve the use of radionuclides characterized by a relatively short half-life (t_½). SPECT relies on the use of radionuclides emitting gamma (γ) rays, such as technetium-99 m (^99mTc, t_½ = 6.02 h), while PET detects beta plus (β⁺) emitting radionuclides, such as fluorine-18 (¹⁸F, t_½ = 109.8 min) and gallium-68 (⁶⁸Ga, t_½ = 67.6 min) [125]. Radiolabeling of targeted RNs with alpha (α), beta minus (β⁻) or Auger e⁻ emitters could also find theranostic application in radiotherapy against cancer, concurrently allowing the visualization of the tumor size reduction and therapeutics accumulation at the target site [126,127]. Table 2 reports the radionuclides commonly used in pre-clinical and clinical fields for imaging and radiotherapy with the relevant examples of radiolabeling on RNA-based material.

Table 2. Radionuclides for bioimaging.

Radionuclide	Half-life (t½)	Decay	Imaging technique	Oligonucleotide radiolabeling
^99mTc [128,129]	6.02 h	γ	SPECT	Direct Chelator
^68Ga [130]	67.6 min	β^+	PET	Chelator
^18F [131]	109.8 min	β^+	PET	Chemical conjugation

There are different techniques for RNA radiolabeling. Incorporation of ¹⁸F into the 5’ position of oligonucleotides was initially demonstrated by the use of (¹⁸F]fluoromethyl)phenyl isothiocyanate by Hedberg and Längström in 1997 [132], or more recently by alkylating agents screened by de Vries and colleagues to radiolabel antisense oligonucleotides [133]. Direct radiolabeling of RNA aptamers was also demonstrated by attaching ^99mTc without the need for any chelating agent [125]. However, the covalent tethering of bifunctional chelators to the RN would offer a more efficient radiolabeling of nucleic acids with radionuclides commonly used in clinics [125,134]. Depending on the technique applied for the loading of the radionuclide on the chelator, harsh conditions such as high temperatures, and extremely acidic or alkaline pH can cause changes to the stability or the physicochemical properties of the nanostructure, and therefore, affect its biodistribution and biological activity. Hence, radiolabeling is best applied to chemically stabilized RNs, such as those incorporating 2’-F or 2’-OMe-modified nucleotides. Additionally, particular care must be dedicated to the choice of radionuclide, optimizing the radiolabeling conditions, and validating that the nanostructure’s properties and functions are preserved [123,135].

Internalization and endosomal escape

In the case of delivery of intracellular payloads, RNs must extravasate the endothelial cells lining the blood vessels and be internalized by cells. The large size of RNs and the negative charge of the backbone prevent the unassisted crossing of the cellular membrane. Most targeted drug delivery approaches take advantage of RME for cell entry (Figure 1B-6) [136]. However, one major challenge related to this strategy is the entrapment of the payload within intracellular compartments, such as endosomes or lysosomes, where the RNs with payload are either recycled back to the surface or destined for degradation in the endolysosomal pathway [136].

Decades of research studying endosomal entrapment and strategies to overcome it and achieve successful nucleic acid delivery resulted in two main strategies [137,138]. The first one is inspired by the delivery of RNA therapeutics co-administered with small molecule endolytic agents like chloroquine or guanabenz, which are known to destabilize the endosomal membrane and ultimately lead to endosomal rupture, to enable cytosolic release [137]. A recent example is the advancement of an endolytic small molecule, UNC7938, which showed efficient delivery of PMO-modified oligonucleotides to the lungs in mouse models [139]. However, endolytic agents are associated with high cytotoxicity due to lysis of many endosomes and lysosomes. The second strategy involves the use of larger molecules such as cationic polymers, which have been investigated for their ability to facilitate endosomal escape through the ‘proton-sponge’ effect, or peptides such as the cell-penetrating peptides (CPP) HIV-1-TAT, able to interact with the membrane inducing their transient destabilization [140,141]. An alternative strategy aims to delay the maturation of early endosomes into late endosomes and lysosomes to extend the window for escape. One strategy aims at inhibiting proteins crucial for endosome maturation and trafficking of endosomal cargo, such as the endosomal sorting complex required for transport-I (ESCRT-1) or Niemann-Pick type C1 (NPC1) [138]. Another approach requires the functionalization with cell-internalizing aptamers that when incorporated in a larger RN, can persistently retain, as an intact structure, in early endosomes for more than two hours [142].

In spite of many years of research, cellular internalization and endosomal escape remain two of the most critical barriers to be overcome, and newer approaches involving the encapsulation of drugs (including RNs) within lipid-based or polymer-based complexes are under investigation [62,143].

Administration routes

The route of administration of RN should be carefully considered to be compatible with the stability of the nanostructure and its function, and allow it to efficiently reach the target tissue. The way the RNs enter the body can strongly influence their bioavailability, activity, and clearance. For these reasons, preliminary in vivo experiments should verify that the chosen route of administration for the designed RN ensures favorable pharmacokinetics, biodistribution, and clearance.

Delivery by i.v. injection ensures the highest bioavailability, rapid response, and enables the agent to reach multiple organs before being cleared by renal excretion [144,145], which is particularly favorable for non-invasive imaging. A slower release of the injected agent can be accomplished by intradermal or subcutaneous administration, or by intramuscular injection [146]. Compared to i.v. delivery, intratumoral injection offers a compelling alternative in cancer therapy as it circumvents systemic-specific barriers and ensures faster, more effective drug release, low off-target distribution, and increased tumor-specific accumulation of the therapeutic agent [147,148]. However, intratumoral administration is limited by the accessibility of the target tissue and the invasiveness of the procedure, especially if the effective concentration of therapeutics requires multiple injections. This could cause the drug to leak to nearby tissues and the systemic circulation, if not uniformly distributed and absorbed by cancer cells [149].

Applications of RNs for tissue targeting

Over the past two decades, there has been significant growth in the development of tailored RNs coupled with targeting ligands, drugs, and imaging tracers for application in nanomedicine as therapeutics, imaging tracers, or theranostic agents. The following section showcases the established and most recent advancements in the development and pre-clinical application of RNs for tumor treatment and diagnostics.

Aptamer chimeras

The advancement of cell-type specific aptamers has led to their use as both targeting moieties and efficient delivery vehicles of siRNAs, short hairpin RNA (shRNA), miRNA, chemotherapeutic and imaging agents for targeted therapy and diagnostic purposes. Such systems are referred to as aptamer chimeras and can find application in targeted therapy and imaging.

In 2006, McNamara and colleagues introduced one of the first RNA-based aptamer-siRNA chimeras (AsiC), A10-Plk1 (Figure 4A) [150]. The A10 aptamer targeted and mediated internalization of the construct into prostate cancer cells binding to the prostate-specific membrane antigen (PSMA). The siRNA targeting the survival genes polo-like kinase 1 (PLK1) was responsible for gene silencing through RNAi pathway. Intratumoral administration of A10-Plk1 in mice bearing PSMA-positive tumor xenografts resulted in a successful reduction of tumor volume by 2.21-fold compared to control chimeras or saline group. A few years later, Dassie and co-workers reported on an improved design of the A10-Plk1 chimera [151], obtained by truncating the aptamer (A10–3.2) and incorporating 2’-F pyrimidine modifications, extending them to part of the siRNA sequence for improved serum and thermal stability. From five different chimera designs, the ‘blunt’ and ‘swap’ chimeras showed improved knockdown efficiency in vivo than the first aptamer-siRNA generation (Figure 4B). In particular, the ‘swap’ AsiC, designed with a co-transcribed anti-sense (guide strand) strand extending from the A10–3.2 aptamer and ending with a 3’ UU overhangs, showed the best tumor reduction with an overall 2.3-fold slower growth rate and up to 8-fold decrease in volume of mouse xenografts after 10 days of treatment.

Figure 4. Aptamer-chimeras for drug delivery and bioimaging. A) Secondary structure of PSMA-specific aptamer-siRNA chimera (AsiC) designed by McNamara and co-workers consisting of the A10 aptamer portion and siRNA against either PLK1 [150]. B) Optimized designs of A10-Plk1 reported by Dassie and colleagues from (A) resulted in two best-performing designs: blunt A10–3.2-Plk1 without overhangs on the 3’ end of the siRNA; swap A10–3.2-Plk1 contains UU overhang on the 3’ end of the siRNA and the passenger and guide strands are swapped [151]. C) The structure of a larger AsiC designed by Liu et al. featured with a bivalent PSMA aptamer with dual siRNA targeting survivin and EGFR (PSEP). Both siRNAs contain a 2-nt overhang at the 3’ end. The PSEP structure is composed of three strands: one containing the A10–3.2 aptamer with anti-survivin antisense siRNA strand; one with A10–3.2 aptamer and two sense siRNA strands connected by a UUUU-linker; and one anti-sense EGFR siRNA strand [152]. D) Schematic illustration of the anti-Axl aptamer-let-7 g miRNA chimera reported by Esposito et al. [153]. E) The structure of prostate-specific E3 aptamer conjugated to the cytotoxic drug MMAF designed by Gray et al [154]. F) The construct reported by Cheng and colleagues including an ME07 RNA aptamer targeting the extracellular domain of EGFR, chemically coupled to ¹⁸F-fluorobenzyl for PET imaging of mice bearing tumor xenografts of varying EGFR expression [155]. Figures are adapted from the cited articles.

These studies soon inspired the construction of larger and more complex AsiC (~59 kDa), such as the one reported by Liu et al., consisting of bivalent anti-PSMA aptamers A10–3.2 and two siRNAs – one against endothelial growth factor receptor (EGFR) and one against survivin, an inhibitor of apoptosis – also referred to as PSMA aptamer-survivin siRNA-EGFR siRNA PSMA aptamer (PSEP) (Figure 4C) [152]. The entire chimera was in vitro transcribed with 2’-F pyrimidines for improved serum stability. Here, the authors exploited the effect of multivalency by using bivalent aptamers for enhanced specificity, cell association, and internalization. The complex PSEP chimera was able to effectively reduce PSMA-positive tumour xenografts up to 4-fold compared to the saline group or chimera with scrambled siRNAs following intraperitoneal treatment over 21 days. Moreover, tumors treated with PSEP appeared less vascularized and smaller in size compared to those in the control groups. This observation was further confirmed by histological and immunohistochemistry analyses of the tumour xenografts. Tumor xenografts treated with PSEP chimera showed a significant downregulation in both EGFR and survivin expression confirming the functionality of the dual siRNA design. As intended, this led to a reduction of signalling cascades involved in cell survival and promoted induction of apoptosis in PSMA-positive cells.

Along with the fundamental development of AsiCs to combat prostate cancer, several research groups have expanded their focus on advancing AsiCs also for the treatment of breast cancer [156], cancer stem cells [157], and for its application in cancer immunotherapy [158]. In all cases, the engineered AsiCs presented excellent stability in vivo and the ability to target genes of interest for RNAi effect to ultimately hamper tumor growth. Owing to the attributes of cell-specific aptamers, targeted delivery of AsiCs resulted in efficacious and cell-specific therapeutic effects. This was accommodated with less toxicity in healthy cells or tissues and a lowered dose required for the administration of the AsiCs.

Cell-specific aptamers can also be used for delivery of shRNAs that for example inhibit DNA repair pathways in prostate cancer cells to make them more responsive to radiotherapy [159] or impede the function of nonsense-mediated mRNA decay (NMD) to improve anti-tumor immunity in B cell lymphoma [160].

Another generation of aptamer chimeras was engineered for the first time by Esposito and colleagues for the delivery of therapeutic miRNAs for cancer therapy. Here, the authors designed a serum stable chimera (GL21.T-let) (Figure 4D) composed of a 34-nt RNA aptamer, GL21.T, specific for the cancer biomarker Axl tyrosine kinase receptor, which is linked with poor prognosis and tumor metastases, and a miRNA, let-7 g, that acts as a tumor suppressor by targeting genes involved in cell cycle, proliferation, and apoptosis [153]. Mice treated with the GL21.T-let chimera or GL21.T alone displayed a 1.4-fold reduction in tumor growth in Axl-positive tumors and an increased expression level of let-7 g miR. Immunohistochemistry staining against the proliferation marker Ki-67 of tumor sections further confirmed the anti-tumor effects of the chimera. Despite conflicting results regarding the specificity of the aptamer towards Axl [161,162], in this work, the aptamer itself demonstrated superior antagonistic properties as it proved to be extremely efficient in inhibiting tumor growth on its own. Such dual action mechanism is attractive and highly beneficial in developing novel, multifunctional aptamer chimeras, as it results in the combination of two potent therapeutic agents: the aptamer, which binds to the oncogenic receptor, inhibits cell growth by interfering with the downstream signalling pathway, the RNAi element amplifies the tumour suppressor effect by silencing mRNAs encoding for oncogenic proteins.

The success of antibody-drug conjugates (ADCs) has inspired researchers to utilize aptamers in developing aptamer-drug conjugates (ApDC) for targeted cancer therapy [163]. This strategy can be exploited for in vitro delivery of the broadly used anthracycline, doxorubicin, loaded on a DNA aptamer-nanotrain system [164] or through chemical conjugation of antimetabolite, methotrexate, to anti-CD117 aptamer [165] for treatment of different types of leukaemia. In 2017, Gray et al. selected a 35-nt 2’-F pyrimidine RNA aptamer (E3 aptamer) to which the highly toxic tubulin inhibitors, monomethyl auristatin E (MMAE) and monomethyl auristatin F (MMAF), were chemically coupled for targeted therapy of prostate cancer with low androgen sensitivity (Figure 4E) [154]. While both ApDCs were able to be internalized and showed a high degree of cytotoxicity in different prostate cancer cell lines, MMAE-E3 uptake lacked specificity as demonstrated by the observed effects upon substitution of the E3 aptamer with a non-targeting control aptamer. However, the MMAF-E3 ApDC demonstrated a remarkable tumor co-localization in mice with 22Rv1 prostate tumour xenografts and was able to inhibit tumor growth by up to 2.5-fold compared to the control group following 35 days of treatment. Interestingly, the authors developed an antidote oligonucleotide capable of blocking the E3 aptamer to reduce cytotoxicity in non-target cells by using it as a safety switch, yet the use of this was not shown in vivo. A few years later, Kim and colleagues synthesized an ApDC constituting a 34-nt 2’-F pyrimidine RNA aptamer targeting the human epidermal growth factor receptor 2 (HER2) conjugated to DM1, another tubulin inhibitor used in chemotherapy, and 20K PEG for improved pharmacokinetics [166]. Breast cancer tumor xenografts in mice were strongly inhibited by both free DM1 (3.5-fold) and the HER2-aptamer-DM1 ApDC (5.4-fold) compared to PBS, with the latter being the most efficacious.

Aptamers not only can function as excellent delivery systems for drugs, but they also hold great potential as molecular imaging tracers for non-invasive diagnostic and theranostic purposes. In search of better binders for this purpose, several cancer-specific DNA and RNA aptamers have been under investigation for their ability to detect various types of tumors in mouse models with high detection sensitivity and specificity [167]. One such example was reported by Dassie et al. where the anti-PSMA RNA aptamer, A9g, was labeled with the NIR dye 800CW and able to specifically localize and be retained in PSMA-positive tumors in mice even 72 h after administration [168].

In the field of targeted nuclear imaging, Schmidt and colleagues were among the first to explore and show using scintigraphy that aptamers can deliver radioisotopes, such as ^99mTc, in mouse tumor xenografts [169]. To achieve enhanced visualization and a more precise evaluation of the stage of the tumor, Chen and colleagues developed an aptamer-based PET tracer for the detection of tumors with variable EGFR expression levels [155]. Using click chemistry, the 2’-F pyrimidine RNA aptamer, ME07, capable of recognizing the extracellular domain of human EGFR, was coupled to¹⁸F-fluorobenzoyl (ME07- ¹⁸F) (Figure 4F). PET imaging of mice bearing tumors with low or high EGFR expression revealed a significant accumulation in EGFR-positive tumors only 30 min following injection and with the highest tumor-to-muscle ratio observed in EGFR-positive A431 tumor. Biodistribution studies further confirmed the high tumor uptake and retention of ME07- ¹⁸F chimera in mouse xenografts with variable EGFR expression. Recently, Jiao et al. constructed a chimera composed of the previously mentioned PSMA-targeting A10–3.2 aptamer, a siRNA specific for mouse double minute 2 homolog (MDM2), and a bifunctional chelator for loading of the radioisotope ^99mTc for prostate cancer [170]. SPECT imaging of mice bearing PSMA-positive xenografts clearly showed tumor localization by the chimeras 30 min post-injection compared to PSMA-negative xenografts. Moreover, the chimeras were able to reduce tumor growth following 14 days of continuous treatment. This study is a great example of how aptamer chimeras with dual functions can be utilized to both diagnose and treat tumors. The advancement of radionuclides with variable half-lives, such as 68Ga, lutetium 177(¹⁷⁷Lu, t_½ = 6.6 days), and yttrium-90(⁹⁰Y, t_½ = 2.67 days) are being used in the clinics for diagnostics and therapy, but a rational combination of these in a single aptamer-chimera design or with any other RNs bearing a targeting ligand, alone or coupled to a therapeutic agent, can be used for the development of theranostic agents that allows simultaneous diagnosis and treatment of patients.

RNA nano- and microstructures

Delivery and visualization of RNAi drugs

The incorporation of multiple targeting and therapeutic elements within the RNA scaffold enables the assembly of larger and more complex multifunctional nanostructures. Similar to aptamer chimeras, these RNs can function both as effectors and molecular scaffolds to facilitate in vivo targeted imaging and drug delivery to specific cells and/or tissues. The synergistic effect of multivalent display using aptamers, combined with gene knockdown provided by multiple siRNAs and miRNAs, significantly enhances the efficacy of these RNs. As a result, they hold great potential for future applications in therapy and diagnostics offering improved targeting and specificity. The discovery of packaging RNA (pRNA) by Peixuan Guo in 1998 [171] paved the way for the implementation of this motif for the formation of 2’-F-modified pRNA-based RNs fused to targeting elements, RNAi drugs, and tracers to yield a theranostic nanostructure for the imaging and delivery of cancer-reducing drugs in vitro and in vivo. For example, the pRNA-based three-way junction (3WJ) scaffold was successfully functionalized with several aptamers targeting biomarkers such as PSMA, EGFR, CD133, or HER2 to deliver LNA-modified anti-miRNA 21 (miR-21, a cancer-related oncomiR) and inhibit the growth of prostate, triple-negative breast cancer [172–174]. In another work reported by Zhang and co-workers, the 3WJ NP was utilized to target tamoxifen-resistant human breast cancer [175]. The RN was decorated with an anti-HER2 aptamer and two siRNAs targeting the mediator subunit 1 gene (MED1), involved in tamoxifen resistance through crosstalk with the HER2 receptor (Figure 5A, (a)). The resulting nanostructure with a hydrodynamic size of 8.68 ± 1.87 nm exhibited high thermo- and biostability owing to the 2’-F modifications, and it was able to specifically accumulate in the tumor via HER2 targeting. When injected alone, the construct inhibited tumor growth resulting in a 2-fold reduction in volume compared to the negative control nanostructure, and prevented the insurgence of lung metastasis. This effect was further improved by an additional 2-fold volume reduction when the nanostructure was administered in combination with tamoxifen. A more recent application of the 3WJ involves functionalizing the nanostructure with an anti-EGFR aptamer, a fluorophore, and an siRNA targeting the kirsten rat sarcoma viral oncogene homolog (KRAS) in small cell lung cancer [176].

Figure 5. RNA nanostructures for visualization and delivery of RNAi drugs in tumours. A) the pRNA-based 3WJ nanoscaffold designed by Zhang and co-workers was functionalized with an anti-HER2 aptamer and two siMED1 sequences to target tamoxifen-resistant breast cancer [175]. B) The RNA nanoring reported by Afonin et al. embedded six siRNAs to achieve improved gene knockdown [177]. C) The miRNA-based stable triplex reported by Conde et al. could multimerize onto PAMAM molecules causing aggregation and formation of µm-sized NPs, which were subsequently embedded into dextran to form a hydrogel designed for the slow release of RNAi drugs [180]. Figures are adapted from the cited articles and partially created with BioRender.com.

The precise and stoichiometric spatial organization of small ligands, fluorophores, and RNAi elements, combined with the self-assembling property of RNA, allow it to form secondary structures, useful in the production of higher-order RNs or complex nanomaterials aimed at achieving the delivery of larger drug payloads. The RNA nanorings, reported by Afonin et al., represent an interesting example where siRNAs were successfully multimerized on the large ring-shaped nanoscaffold to obtain improved drug delivery and gene knockdown (Figure 5B) [177]. Six siRNAs targeting GFP were incorporated in the RNA construct, which displayed five copies of the J18 anti-EGFR aptamer and a biotinylated oligonucleotide for the following in vivo detection via streptavidin-phycoerythrin (PE). Five days after the intratumoral injection of the RNA nanoring into GFP- and EGFR-expressing tumor xenografts, ex vivo fluorescence quantification revealed up to 80% of GFP silencing in tumor tissue.

Two years later, a trivalent aptamer display was proposed by Li et al. through four self-assembled 2’-F-modified pRNA 3WJs, combined into an RNA tetrahedron NP with a size of ≈17 nm [178]. In vitro, the incorporation of siRNA against firefly luciferase into the sequence of the tetrahedron showed ≈ 90% silencing of EGFR-expressing cells MDA-MB-231. In vivo, the tetrahedron nanostructure was fused to three anti-EGFR aptamers, and subsequently labeled with an Alexa Fluor 647 fluorophore for in vivo fluorescence imaging of EGFR-positive tumors. The functionalized tetrahedron was injected systemically and after 8 hours, the organ scans revealed a high accumulation in the EGFR-positive tumor xenograft, with little to no background uptake in non-target organs.

The incorporation of a larger number of RNAi elements in RNA-based nanomaterials was reported by Kim and co-workers through the introduction of bubbled RNA-based cargo (BRC) RNs. The BRCs were obtained from the T7-mediated rolling-circle transcription of two circular DNAs, each encoding the sense or antisense sequences of an siRNA [179]. The nascent sense and antisense RNA are allowed to hybridize, alternated by contiguous unpaired sequences forming bubbles. The continuous polymer entangled into the higher-order BRC NP, reaching a size of 350 nm. In vitro treatment of BRCs with the Dicer enzyme showed up to ≈ 95% of the total siRNA release capacity after 48 hours, while the incorporation of Cy5-UTPs and the GFP-targeting siRNA sequence allowed for monitoring the BRCs uptake and effect on GFP silencing in cells, constitutively expressing the fluorescent protein. In vivo, the intratumoral administration of BRCs into GFP-positive tumor xenograft caused a reduction of the fluorescent signal intensity to 10% after 4 days, subsequently confirmed as the effect of BCRs-mediated gene silencing by RT-PCR quantification of mRNA encoding for GFP. Recently, BRC NPs have been successfully applied to target UBA6-specific E2 conjugating enzyme 1 (USE1) and reduce treatment of human lung cancer xenograft in mice upon multiple i.v. injections [181].

A larger and more complex RNA-based nanomaterial was reported by Conde et al. The self-assembly through Watson and Crick and Hoogsteen hydrogen bonds of two miRNAs, sense and antisense miR-205, and one anti-miRNA, antagomiR-221, resulted in a stable triple RNA helix, constituting the active component of engineered material (Figure 5C) [180]. The assembled helices were non-covalently complexed to the cationic surface of the polyamidoamine (PAMAM) dendrimer to form 3D RNs (size ≈50–60 nm), which are allowed to aggregate into 3–4 µm-sized particles. These were subsequently mixed with dextran aldehyde to form a millimeter-sized adhesive hydrogel intended for local and controlled in vivo miR-205 delivery. The correct triple-helix assembly of the sense and antisense miR-205 and the antagomiR-221, directed against triple-negative breast cancer, was monitored through a Förster resonance energy transfer (FRET) pair, conjugated at the 3’ end of antisense sequences. The 5’ end of the sense strand was conjugated to a quencher while its 3’ end was coupled to cholesterol to confer cell permeability and biostability to the RNA complex. Functionalization and assembly into the triple helix did not affect the cleavage activity of the Dicer and allowed the investigation of the pathway enabling the cellular internalization of the PAMAM-RNA complex. The RNA-based hydrogel was injected subcutaneously on the side of tumor xenografts implanted in mice and the release of miR and antagomiR was visually monitored for 14 days. After a single administration of the RNA-based hydrogel, a tumor size reduction of up to ≈ 90% was observed compared to tumors treated with the single miRs or chemotherapeutic drug-based hydrogels. The RNA release was localized in the tumor and no leakage was visible in non-target organs. Moreover, the single administration of the RNA-based hydrogel extended the survival rate up to 2-fold compared to other treated mice, along with a decreased expression of the targeted oncogenes.

Molecular imaging and delivery of immunotherapies and chemotherapeutic drugs

A flexible multifunctional RN for application in molecular imaging and drug delivery was reported by Andersen et al. creating a self-assembled modular four-way junction (4WJ), resembling a Holliday Junction (HJ) structure, which consisted of four, single-stranded oligonucleotides, fully modified with 2’-OMe and LNA nucleotides, that conferred higher thermostability, resistance in biological fluids, and low immunogenicity (Figure 6A) [182]. Each oligonucleotide has two six-bp sequences complementary to the two adjacent strands to form a ≈4 nm-sized nanoscaffold. The 5’ ends of each oligonucleotide module are modified with an amine group enabling the chemical functionalization of up to four different elements. These elements encompass drugs, small molecules, such as antigens for immune system activation [185], peptides, as well as fluorophores and chelators. This nanoscaffold shows strong potential for theranostic applications with dual functionality as both an imaging tracer and drug delivery vehicle. In this work, the HJ platform was utilized to enable liver tissue targeting, showing improved delivery of up to 1.4-fold by the multimerization of three triGalNac molecules compared to a single copy. In the following year, the HJ scaffold was implemented for the in vivo targeting and imaging of PSMA-positive prostate tumor xenografts by Omer et al. [162]. Up to three 2’-F modified anti-PSMA A9g or anti-Axl receptor aptamers were attached to three arms of the HJ via SPAAC conjugation (Figure 6B). The multivalent display of two and three aptamers demonstrated an increased accumulation of the nanostructure in PSMA-positive tumors in response to the number of targeting ligands. Compared to the monovalent construct, tumor accumulation increased up to 1.5-fold for the divalent and up to 2-fold for the trivalent construct.

Figure 6. RNA NPs for molecular imaging of tumours and delivery of chemotherapeutic drugs. A) Andersen et al.’s design of a self-assembled, modular, fully modified HJ-like structure for the conjugation of up to four molecules, among ligands, drugs, and tracers [182]. B) The following year, Omer et al. functionalized the HJ with up to three A9g anti-PSMA aptamers for improved multivalent targeted imaging of PSMA-positive tumor xenografts [162]. C) In combination with anti-CD3 antibody co-stimulation, the self-assembled X-polymer, displaying a CD28Apt7 dimer and a Del60 tetramer, recognizing CD28 and CTLA-4 respectively, was designed by Bai et al. to activate T-cells and direct their action towards cancer cells elimination [183]. D) Guo et al. implemented the RNA 4WJ to increase the solubility of the chemotherapeutic PTX by chemically conjugating a total of 24 PTX molecules to the four arms of the RN [184]. Figures are adapted from the cited articles.

The improved targeting activity of multimerized ligands is not limited to the delivery drugs or tracers, but it can also effectively influence extracellular biological pathways, such as those involved in immune system activation [186]. In 2020, Qi and co-workers characterized a transcribed, highly stable, and unmodified RNA origami rectangle (RNA-OG), as a good anticancer immunotherapeutic and adjuvant for vaccine development, thanks to its strong TLR3 agonist activity [187]. Recently, the immunostimulatory activity of the RNA-OG has been coupled to the scaffolding property of its structure to display multiple tumour neopeptides and generate an RNA-OG-peptide nanostructure to engage dendritic cells (DCs), induce peptide cross-presentation to CD8⁺ T cells, and elicit CD8-mediated antitumor immune activation [188]. To demonstrate the activity of the RNs, up to 13 azide-modified ovalbumin _257–264 peptides (pOVAs) were conjugated to the RNA-OG, resulting in the RNA-OG-pOVA nanostructure. In vitro studies showed that the stability of the NP was unchanged despite the peptide conjugation. MHC-I-mediated pOVA peptide presentation on primary bone marrow DCs (BMDCs) was verified by flow cytometry upon incubation with the RNA-OG-pOVA construct. Subsequently, coculturing RNA-OG-pOVA-treated BMDCs with pOVA-specific CD8⁺ T cells (isolated from transgenic OT-I mice) highlighted the high level of CD8⁺ T cell proliferation and activation resulting from the RN-induced DCs maturation and peptide-cross presentation. The antitumor activity of the RNA-OG-pOVA nanostructure was assessed in C57B1/g mice bearing B16-OVA tumor engraftment upon the subcutaneous injection of 3 doses of the construct every 5 days. The RN-treated mice showed the best survival rate with up to 60% of the tested mice living for up to 60 days. This effect was correlated to the tumor growth delay caused by the RN, however, prophylactic vaccination with the RNA-OG-pOVA nanostructure was demonstrated to inhibit tumor formation and progression, even upon a second tumor engraftment. However, treatment with the single RN was shown not to arrest completely, therefore the RNA-OG-pOVA nanostructures were combined with α-PD-1 immune checkpoint blockade (ICB) treatment, which improved long-term survival (>100 days) in 40% of tested mice. The RNA-OG-pOVA, alone or in combination with α-PD-1 ICB treatment, successfully converted the tumor microenvironment from immunosuppressive to immunogenic by significantly increasing CD8⁺ T cell proliferation and reducing immunosuppressive regulatory T cells. Finally, it also promoted the expansion of pOVA-specific central memory T cells in tumour-draining lymph nodes which can contribute to the long-term antitumor response.

Immunomodulation is also attempted by displaying multispecific targeting ligands on smaller nanostructures, such as the 3WJ construct (Figure 6C) designed by Bai and colleagues, which was coupling the CD28Apt7 aptamer dimer, directed to CD28, along with a Del60 aptamer tetramer, able to target the cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) [183]. The multivalent display of the six 2’-F-modified aptamers, combined with an anti-CD3 antibody co-stimulation, could induce a chimeric antigen receptor (CAR)-like effect that activates T-cells to proliferate and secrete cytotoxic factors. The modular RN, denoted as X-polymer, was successfully self-assembled embedding the six aptamers and a ssDNA module (TRS) labelled with a detection tracer (biotin or fluorophore) and folic acid, to achieve targeting in tumor cells expressing high levels of folate receptors (FRs). In vitro, the X-polymer was demonstrated to efficiently target FR-overexpressing melanoma and colon cancer cells. In combination with anti-CD3 stimulatory activity, the X-polymer promoted T-cell activation and CD4+ T-cell proliferation, and its strong stimulation proved to be highly effective in reversing the effect of the B7.1 inhibitor of T-cell activation, compared to the anti-CD3-CD28-CTLA-4 antibody-based treatment. The increased levels of interferon-γ (IFN- γ) and granzyme B in anti-CD3-X-polymer-activated T-cell supernatant suggested a strong killing activity, which was further confirmed in vitro by an evident increase in apoptotic melanoma cells. In vivo, a 2.6-fold reduction in volume of melanoma tumour xenografts was observed after four subcutaneous administrations of the X-polymer over 13 days, compared to non-targeting aptamers.

RNs have also been applied as an efficient strategy to solubilize and deliver hydrophobic drugs. To maximize the load and the solubility of the chemotherapeutic drug paclitaxel (PTX), Guo and co-workers presented a larger 2’-F-modified 4WJ loaded with 24 molecules of PTX, reaching a hydrodynamic size of 9.1 ± 0.1 nm [184]. The four oligonucleotides constituting the 4WJ were each functionalized with six alkyne groups for click reaction with azide-PTX molecules (Figure 6D). The drug cargo did not interfere with the assembly and stability of the 4WJ while the solubilization of PTX was highly improved. The incorporation of an anti-EGFR aptamer enabled the targeting of EGFR-overexpressing tumor xenografts and allowed the drug payload to localize primarily in the targeted tissues, resulting in low toxicity and up to 1.4-fold reduction of tumor xenografts masses compared to PTX treatment alone.

Recently, this strategy was coupled with the delivery of miRNA, as introduced by Wang et al. [189]. A 2’-F-modified six-way junction (6WJ), incorporating up to 24 molecules of PTX, was assembled to one copy of the miR-122 (a miRNA involved in hepatocarcinogenesis) and three GalNAc molecules, reaching a hydrodynamic size of 17.3 ± 5.6 nm. In vivo, the fully equipped 6WJ was injected six times at the clinically administered dosage of PTX formulated in Cremophor EL/ethanol over 22 days. The rubber-like stretchability of the RNs ensured further penetrability into liver tissue, which strengthened tumor homing but caused a low non-specific hepatic uptake of the naked 6WJ. The stronger efficacy of the fully equipped 6WJ was achieved through the synergistic effect of the miR-122 and PTX, exhibiting superior action by a 3-fold higher tumor growth inhibition compared to the modest tumor volume reduction exhibited by the same dosage of the chemotherapeutic agent alone. Moreover, the injected RN demonstrated enhanced safety thanks to the combination of two key benefits. Firstly, the low immunogenicity of the nanostructure was evidenced by the absence of detectable pro-inflammatory cytokine production in vivo with respect to PTX treatment alone. Secondly, the tumor-targeting specificity conferred by GalNAc, allowed the RN to primarily localize to the tumor, thus minimizing harm to non-target tissues.

Conclusion and outlook

The ever-growing field of RNA nanotechnology is widely explored to build nanostructures for targeted therapy, diagnostics, and theranostics, particularly applied in the treatment of cancer. Their success in pre-clinical applications has been driven by the progress in interdisciplinary fields converging efforts into increasing RNA stability in vivo thanks to chemically modified nucleotides, and the versatile yet controlled bioconjugation with drugs, targeting ligands, and imaging tracers. These advancements prospect the use of RNs in future cancer therapy and diagnostic imaging, however, several challenges remain to be addressed.

While we are gaining important insights into the biocompatibility of the delivered RNs and the impact of biological barriers on their efficacy, RN safety remains a limitation and demands an in-depth characterization of the physicochemical properties of the functionalized RNs, including pharmacokinetics, biodistribution, toxicity, off-target effects, clearance, side effects, and administration strategy [37,190–193]. Furthermore, RN production poses limitations to their clinical application due to the high cost associated with the synthesis and purification of chemically modified and bioconjugated RNA and the lack of standardized procedures to produce pyrogen-free RNs with minimal batch-to-batch variability and establish optimal storage conditions to preserve long-term stability and activity [194].

Another challenge lies in the selection of more specific targeting ligands that function specifically in vivo. Here, the selection of chemically stabilized aptamers against specific mouse or human biomarkers and samples proves advantageous in comparison to the selection of protein-based ligands as it offers a more accessible and cost-effective approach for researchers, not only to achieve high specificity for any relevant biomarker but also to generate cross-reactive ligands, to facilitate the clinical translation in the transition from mouse to human testing. Furthermore, aptamers would also facilitate multivalent and multiparatopic displays and can be co-transcribed along with RNA drugs within the scaffold’s sequence of the nanostructure, an advantage that renders RNA one of the most versatile and unique biopolymers for the construction of future therapeutic agents. However, one concern is the quality of the published landscape of aptamers [161]. Their claimed specificities are often questionable and careful testing should be performed before entering preclinical studies.

The size and complexity of RNs enable them to be harnessed with multifunctional capabilities but also call for careful validation of intended action, a prerequisite that unfortunately often is lacking in published studies. Carefully designed controls that confirm stability, targeting capacity, and gene regulatory effect must be included to ensure the future sustainability of the therapeutic RN field.

The success of mRNA vaccines against SARS-CoV-2 has inspired a stronger interest in the application of RNA nanotechnology in clinical settings. Especially smaller siRNA- and ASO-based constructs benefit from more extensive scientific investigation over the last decades and compatibility with existing production processes. Therefore, these structures are more likely to exhibit faster progress and approval for clinical use compared to more structurally complex RNs. However, it is increasingly acknowledged that the inherent design flexibility of the RNA biopolymer unleashes the capacity for structurally complex RNs to tailor the targeting and therapeutic activity of future RNA therapeutics.

Acknowledgments

We thank Maria Gockert for reviewing and editing this manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

L. T. was supported by the Novo Nordisk Foundation [NNF23OC0081177, RNA-META] and the Danish National Research Foundation (Centre for Cellular Signal Patterns, DNRF135). M.O. was funded, in part, by grant DNRF135 and by the Novo Nordisk Foundation [NNF23OC0082848].