Functionalizing DNA Nanostructures for Therapeutic Applications

Abstract

Recent advances in nanotechnology have enabled rapid progress in many areas of biomedical research, including drug delivery, targeted therapies, imaging, and sensing. The emerging field of DNA nanotechnology, in which oligonucleotides are designed to self-assemble into programmable 2D and 3D nanostructures, offers great promise for further advancements in biomedicine. DNA nanostructures present highly addressable and functionally diverse platforms for biological applications due to their ease of construction, controllable architecture and size/shape, and multiple avenues for chemical modification. Both supramolecular and covalent modification with small molecules and polymers have been shown to expand or enhance the functions of DNA nanostructures in biological contexts. These alterations include the addition of small molecule, protein, or nucleic acid moieties that enable structural stability under physiological conditions, more efficient cellular uptake and targeting, delivery of various molecular cargos, stimulus-responsive behaviors, or modulation of a host immune response. Herein, various types of DNA nanostructure modifications and their functional consequences are examined, followed by a brief discussion of the future opportunities for functionalized DNA nanostructures as well as the barriers that must be overcome before their translational use.

Graphical Abstract

1. INTRODUCTION

When Nadrian Seeman pioneered the field of DNA nanotechnology in the early 1980s (Kallenbach et al., 1983; Nadrian C. Seeman, 1982; N.C. Seeman & Kallenbach, 1983) his motivation was to use DNA to construct a highly ordered scaffold for the precise positioning of proteins that are otherwise difficult to crystallize on their own. Although multiple approaches to the design of DNA nanostructures have been developed, including Seeman’s tile-based approach (T. J. Fu & Seeman, 1993; Gu et al., 2010; LaBean et al., 2000), Rothemund’s DNA origami (Rothemund, 2006), and Yin’s single-stranded “brick” motifs (Ke et al., 2012; Wei et al., 2012), all share a conceptually similar design principle: Oligonucleotides programmably self-assemble into 2D and 3D nanostructures based on Watson-Crick base pairing, with the shape and size specified by Holliday junctions that bridge the DNA helices in a predetermined manner (Dietz et al., 2009; Shawn M. Douglas et al., 2009; D. Han et al., 2011; Ke et al., 2012; Rothemund, 2006; Wei et al., 2012). These advances in DNA structural design have caused the field to explode in the last 30 years and inspired applications far beyond Seeman’s original vision.

One such area where DNA nanotechnology offers tremendous potential is in biomedicine and targeted therapy. DNA nanostructures (DNs) are attractive platforms for drug delivery in particular due to their biocompatibility, biodegradability, nontoxicity (Jiang et al., 2012; Ko et al., 2008; Schüller et al., 2011; Zhang et al., 2014), and ability to form 2D and 3D structures with tunable size and shape. Additionally, DNA strands can be readily modified, both chemically and biologically, which in turn allows them to be used as drug delivery vehicles (S. M. Douglas et al., 2012; S. Li et al., 2018; Zhang et al., 2014), artificial lipid membrane channels (Burns et al., 2016; Langecker et al., 2012), and platforms for enzymatic and chemical reactions (J. Fu et al., 2012; Voigt et al., 2010). Due to their highly addressable nature, DNs have several advantages over traditional carriers: They are monodisperse and allow control over their size and shape to suit diverse cargoes (even within the same carrier); they enable the precise spatial arrangement of a variety of ligands with controlled valency; and they can be modified with functional moieties on their surface to influence their biological function.

Despite these advantages, the widespread use of DNs in vivo is limited by their stability under physiological conditions (Hahn et al., 2014) and their entrapment into endosomal compartments upon cellular uptake (P. Wang et al., 2018). Nevertheless, to date there have been a myriad of examples of DNs used in biological contexts. Some of these limitations have been mitigated by functionalization of the DNs themselves, including surface decorations that promote stability (Agarwal et al., 2017; Auvinen et al., 2017; Mikkila et al., 2014; Perrault & Shih, 2014; Ponnuswamy et al., 2017), targeting moieties to enhance delivery specificity (Q. Li et al., 2017; Xia et al., 2016), or promote subcellular trafficking (Chan & Lo, 2014; L. Liang et al., 2014), and stimulus-responsive triggers (Banerjee et al., 2013; Keum & Bermudez, 2012) to confer greater control over their behavior in vivo. In this review, we will describe the ways in which DNs have been functionalized to facilitate these properties. Although a large area of biological DN research is dedicated to sensing and diagnostics, this review will focus on their therapeutic applications. We refer the reader to other reviews that cover these other topics in greater detail (Bujold et al., 2018; J. B. Lee et al., 2010; Mathur & Medintz, 2019). Furthermore, we will focus primarily on examples where DNs were functionalized with non-oligonucleotide moieties (both covalently and non-covalently) to impart stability or bioactivity. Although we are aware that DNA nanostructures can be made by alternative methods, such as those made by rolling circle amplification (Jing Li et al., 2019; W. Zhao et al., 2008), this review is concerned only with monodisperse, self-assembled DNs designed with a discrete number of oligonucleotides. Our aim is to convey both the wide range of chemical strategies that can be used to modify DNA nanoscaffolds, as well as to survey the range of applications that these modifications enable.

2. FUNCTIONAL APPLICATIONS OF DNA NANOSTRUCTURES

2.1. In Vivo Stabilization

The self-assembly of DNA nanostructures often necessitates the close packing of DNA helices, resulting in electrostatic repulsion between their negatively charged phosphate backbones. To avoid DN denaturation, high concentrations (5–20 mM) of divalent cations, such as Mg²⁺, are required structural stability. These concentrations are approximately an order of magnitude higher than what is found in physiological fluids, such as human serum (which generally contains less than 1 mM of divalent cations) (Hahn et al., 2014). Additionally, DNs are particularly susceptible to the activity of nucleases and risk degradation in cell media and in vivo. Multiple studies have shown that nuclease activity results in rapid degradation upon DN incubation with 10% fetal bovine serum (FBS) (Gait et al., 2013; Hahn et al., 2014; Surana et al., 2013). In 2014, Hahn et al. determined that the addition of nuclease-inhibitory actin and up to 6 mM Mg²⁺ in cell culture media helped retain DN integrity without any observable effects on cell growth and viability (Hahn et al., 2014). Furthermore, the authors demonstrated that low-Mg²⁺ denaturation occurs in a design-dependent manner. Unfortunately, this approach is only applicable to in vitro cell culture conditions. More experimental studies are necessary to probe the in vivo behavior and integrity of DNs if they are to be used for biomedical purposes.

Some structures, such as the DNA tetrahedron (Goodman et al., 2005), were originally shown to survive under cell culture conditions without any stabilizers, making them a long-favored and often-implemented design for intracellular studies of DNs (Keum & Bermudez, 2009; Walsh et al., 2011). Their resilience was attributed to a less dense packing of DNA helices in concert with the relatively short length of the tetrahedral edges, which limit the available space for nuclease binding. In contradiction to this popular belief, recent work by the Sleiman group has demonstrated major stability issues with this structure that were originally overlooked due to flaws in the design of intracellular fluorescence experiments that evaluated tetrahedron uptake (see Challenges and Outlook section for a more thorough discussion of these findings and the importance of careful design of cellular uptake experiments) (Lacroix et al., 2019). The simple geometry and small size of DNA tetrahedrons, however, limit their capacity for multi-functionalization and internal cargo loading, necessitating methods for structural stabilization of alternative DN architectures. Recent reports by Benson et al. (Benson et al., 2015) or Veneziano et al. (Veneziano et al., 2016) further demonstrate that more complex wireframe structures can also be stable at physiological salt concentrations, although the enzymatic stability of these structures was not probed. Notably, it has been reported that an origami nanostructure made solely of a single strand of RNA is resistant to RNase I digestion and remains stable after overnight incubation in mouse serum. The authors attributed this remarkable stability to the compact nature of the structure as well as the lack of internal nick positions that are characteristic of traditional DNA origami fabrication methods (Qi et al., 2020).

2.1.1. Cationic Polymers

Therapeutically-oriented DN research has developed several approaches to address the challenge of DN stability in denaturing or enzymatically hydrolyzing environments (Bila et al., 2019). In 2016, cationic block copolymers were synthesized to electrostatically adhere to and encase DNA origami structures (Kiviaho et al., 2016). These polymer-coated DNs exhibited very low cell toxicity after nine hours of incubation with A549 human epithelial cells. Furthermore, luciferase proteins encapsulated by DNs demonstrated limited enzyme activity when coated compared to their bare counterparts due to the restriction of substrate accessibility. The extent of substrate inaccessibility was dependent upon polymer structure, suggesting that these coatings may serve future applications in tuning enzymatic reaction rates.

Shih and co-workers demonstrated that electrostatic coating of DNs with a PEGylated oligolysine decamer (K₁₀-PEG_5K) affords an approximately 400-fold increase in DN half-life (~36 h) in 10% FBS and protects DN structural integrity in as low as 0.6 mM Mg²⁺ (Ponnuswamy et al., 2017). The charged polymer effectively replaced magnesium to stabilize the structures, and the PEG polymers blocked nuclease accessibility to the surface, thereby reducing degradation rates. Additionally, the authors demonstrated that the K₁₀-PEG_5K coating did not hinder accessibility to single-stranded (ss)DNA handles on the DN, suggesting that these structures can still be decorated with surface features if needed. The coated structures showed a 5-fold greater half-life in vivo in murine models compared to the uncoated structures, possibly due to their increased resistance to nuclease degradation and denaturation. Even greater stability (>48 h of incubation with DNase I) was achieved in a follow-up investigation in which the amine side chains of the oligolysine coating were covalently cross-linked with glutaraldehyde, which was hypothesized to decrease mobility and/or dissociation of the coating from the DN (Anastassacos et al., 2020) (Figure 1A). It was also shown that DNs with the cross-linked coating were internalized into cells more readily than those coated with the non-cross-linked counterpart. A comparable cationic PEG-polylysine polymeric coating molecule designed by Agarwal et al. (Agarwal et al., 2017) displayed similar protective benefits when incubated with 10% FBS, DNase I, or in low salt concentrations (no Mg²⁺ and 20 mM NaCl). Importantly, this study illustrated that the coating can be removed from DNs via competitive complexation of polyanionic dextran sulfate (Figure 1B), which could be problematic in an in vivo environment that contains anionic polymers (e.g. heparin).

Figure 1. Electrostatic coating strategies for in vivo DNA nanostructure stabilization.

A) Schematic of PEGylated oligolysine coating and subsequent crosslinking by glutaradehyde. Figure adapted from (Ponnuswamy et al., 2017) and (Anastassacos et al., 2020). B) Block copolymer protection strategy. DNA origami is degraded in the absence of polymer (shown in gray/green) but protected from nucleases after polymer complexation. Decomplexation is made possible via competition with dextran sulfate (shown in red). Figure adapted from (Agarwal et al., 2017). C) Schematic view of the two types of peptoid surface coatings of octahedral DNA origami (brush-type and block-type) and their chemical structures. Figure adapted from (S.-T. Wang et al., 2020). D) Lipid micellization of an octahedral DN with lipid-conjugated DNA handles in a solution of surfactant and lipids (liposomes, DOPC, and PEG-PE). Figure adapted from (Perrault & Shih, 2014). E) Schematic of BSA conjugated to a synthetic binding domain (G2) and its dendritic chemical structure. Figure adapted from (Auvinen et al., 2017).

Molecular dynamics simulations of a non-PEGylated oligolysine decamer (K₁₀) interacting with a DNA origami rectangle showed that the peptide adheres nonspecifically and reversibly to the structure, stabilizing it by binding across adjacent helices with its flexible cationic side chains (Roodhuizen et al., 2019). One additional finding to consider in future electrostatic coating studies is that K₁₀ binding also causes a slight decrease in aspect ratio and change in DN conformation, albeit to a lesser extent than the common DNA-stabilizing cations Mg²⁺ and Na⁺. Although a highly promising method of DN stabilization against low salt concentrations and nuclease degradation, the nonspecific binding mechanism of oligolysine prevents the modulation of site-specificity or density of the coating. Additionally, this peptide-based coating could be susceptible to protease activity, a concern that was not addressed in these investigations. Overall, although electrostatic coatings are a highly promising—and potentially scalable—method for functionalizing and stabilizing DNs, they fundamentally alter the nanostructure charge and the chemical identity of the surface. As such, it will be critical to conduct in-depth studies on their biological properties prior to widespread adoption. These studies include: (1) probing their immunogenicity; (2) determining their stability to in vivo conditions, both in terms of protease degradation and potential “stripping” of the coating by other anionic molecules in the biological milieu; and (3) evaluating the nonspecific, charge-mediated uptake (or worse yet, toxicity) of these structures due to the reduction in anionic charge.

Employing polycationic peptoid molecules for DN protection may offer a solution to the issue of protease degradation for peptide-based coatings. Peptoids are peptidomimetic N-substituted glycine oligomers that offer many of the same benefits as peptides, including biocompatibility, low-cost synthesis, and high chemical addressability, but without susceptibility to proteases. Wang et al. designed a series of PEGylated peptoid molecules and tested their ability to protect octahedral DNs from biological adversaries (S.-T. Wang et al., 2020). In this work, peptoids with two different binding modes were designed: brush-type and block-type. Brush-type peptoids exhibited alternating amine side chains (Nae) and PEG side chains (Nte) and bound to DNs with the full backbone against the DN structure. The block-type peptoids were designed with Nae and Nte clustered into blocks of repeating units that resulted in binding of the Nae end of the chain to the DNs while the Nte end protruded out and away from the structure (Figure 1C). It was determined that brush-type peptoids exhibited the greatest DN protection against physiological conditions, including 1.25 mM MgCl²⁺ or the presence of DNase I or 10% FBS over 24 hours. Furthermore, peptoid-coated DNs conferred a 26% increase in protection of their encapsulated protein cargo when incubated with trypsin compared to uncoated DN encapsulation, demonstrating that these modifications can serve purposes beyond protection of the scaffold alone. The chemical versatility of the peptoid coating was also demonstrated by functionalizing them with various moieties, including peptides and antibodies, via click chemistry.

There is a concern, however, that in vivo application of peptoids as coating molecules could generate an unwanted immune response, as they have been indicated as effective haptens for vaccines (Desmond et al., 2013). Though peptoids themselves have not been shown to elicit any humoral immune response (Astle et al., 2008), their covalent linkage to immunogenic carrier proteins has been shown to evoke anti-peptoid, anti-linker, and anti-carrier antibodies in rabbits (Case & Desmond, 2016). The lack of hydrogen bonding in peptoid backbones confers a high degree of conformational flexibility, which may at least partially account for the lack of a specific antibody response against peptoids unless anchored to carriers. It has yet to be determined whether a noncovalent adherence to DNs, which would similarly limit this conformational flexibility, may also stimulate an undesirable humoral response against the peptoid-complexed structures. It is, however, unlikely that a cell-mediated response against peptoids could occur since T cell recognition is dependent upon their proteolytic degradation and presentation in antigen-presenting cells.

2.1.2. Lipid and Protein-Based Coatings

Although cationic polymers present a promising solution to the issue of DN stability, other methods have shown similar potential. Perreault and Shih, for example, developed a method of lipid micellization of DNA nanostructures by utilizing lipid-DNA conjugates that anneal to external DNA handles on the structure in a surfactant solution (Perrault & Shih, 2014) (Figure 1D). The virus-inspired strategy resulted in up to 85% protection of the encapsulated DNs from DNase I digestion over 24 hours. Furthermore, immune activation against the micellized DNs was decreased by two orders of magnitude compared with controls, as measured by inflammatory cytokine secretion and splenocyte uptake. Lipid micellization was also shown to significantly increase the biodistribution and pharmacokinetic half-life of DNs in murine models. Though this encapsulation strategy was not explicitly tested for efficacy of DN protection in low-salt conditions, the nanostructure behavior in murine models suggests that they were not denatured under physiological salt concentrations.

Another concern regarding the use of DNA nanostructures as intracellular delivery vehicles is their transfection efficiency. A protein-based coating was designed to enhance DN transfection and biological stability by chemically modifying bovine serum albumin (BSA) with a cationic dendrimer to enable nonspecific DNA binding (Auvinen et al., 2017) (Figure 1E). This coating not only enhanced transfection into human embryonic kidney (HEK293) cells by a factor of 2.5, but notably reduced the levels of inflammatory cytokine secretion by primary mouse splenocytes. Another interesting observation is that the BSA-coated structures seemed to exhibit enhanced endosome escape capabilities compared to the uncoated structures, a useful attribute if cytosolic delivery of DN therapeutic cargos is required.

2.2. Cellular Uptake & Targeting

Successful utilization of DNA nanostructures as therapeutic delivery vehicles necessitates adequate transfection and targeting capabilities. Investigations on cellular uptake efficiency have shown that DNs can be internalized by various mechanisms of endocytosis depending on their mass, shape, and target cell line. Analysis of the endocytic pathways for tetrahedral DN internalization demonstrated a caveolin-dependent mechanism (L. Liang et al., 2014). Caveolin-dependent endocytosis is receptor-mediated and engrosses material up to about 60 nm in diameter (Rennick et al., 2021). However, receptor-mediated clathrin-dependent endocytosis is the major pathway for the internalization of roughly spherical extracellular materials 100–200 nm in diameter. Receptor-independent clathrin-mediated endocytosis enables internalization of particles at a similar size distribution, albeit at a slower rate than its receptor-dependent counterpart. Although one study has shown that tetrahedral DNs demonstrate caveolin-mediated endocytosis (L. Liang et al., 2014), it has been suggested that multiple endocytic mechanisms may contribute to DN internalization (Hu et al., 2019), particularly for particles greater than 60 nm in diameter. In 2018, Wang et al. demonstrated that the uptake efficiency of four distinct DNs varied between three human cancer cell lines (P. Wang et al., 2018). The authors further determined that larger structures were more readily endocytosed, and they hypothesized that this was due to a greater DN surface area interacting with the cell membrane. However, they were unable to determine specific endocytic pathways for these structures using chemical endocytosis inhibition assays (see Section 3, Challenges and Outlook). Later that year, Bastings et al. provided further evidence of the effect of mass and shape on DN cellular uptake, finding that larger, more compact particles were preferentially internalized compared to more elongated structures (Bastings et al., 2018). This study also showed that DNs were internalized with various efficiencies depending on cell type, with bone marrow dendritic cells (BMDCs) exhibiting superior uptake compared to HEK293 and human umbilical vein endothelial cells (HUVECs). The authors hypothesized that these results are due to the fact that BMDCs are particularly adept at taking up materials from their extracellular environment.

2.2.1. Increasing Uptake Efficiency

Although some structures, such as tetrahedral DNs, are often efficiently endocytosed without the aid of transfection agents, the anionic nature of DNA discourages association with negatively charged cell membranes. Thus, a number of groups have investigated ways to enhance intracellular delivery of DNs as a way to enhance delivery of therapeutic cargos. By using the cationic capsid proteins from the cowpea chlorotic mottle virus (CCMV) to coat the surface of a DN, a 13-fold increase in delivery was observed compared to bare DNs, a number that far surpassed transfection with commercially available agents like Lipofectamine 2000 (Mikkila et al., 2014) (Figure 2A). Little is known about the endocytic mechanism of CCMV host infection, but the authors proposed that the capsid’s positively charged N-terminus mitigates charge repulsion between the virus and host cell membrane, enhancing uptake (Roenhorst, 1989).

Figure 2. Methods for cell-specific targeting and uptake of DNA nanostructures.

A) Viral capsid proteins for electrostatic coating and enhancement of cellular uptake of DNA origami structures. Figure adapted from (Mikkila et al., 2014). B) DNA nano-bundles modified with a varying number of cholesterol moieties (depicted in orange) to enhance DN interaction with the cell membrane and increase internalization. Figure adapted from (Whitehouse et al., 2019). C) Tumor penetrating peptide conjugation with a Dox-intercalated DNA tetrahedron that allows for cancer-specific DN uptake. Figure adapted from (Xia et al., 2016). D) Folate-modified DNA nanotube labelled with a fluorescent dye (Cy3) capable of targeting folate receptor-overexpressing cancer cells. Image adapted from (Ko et al., 2008). E) Nucleolin-specific aptamer (AS1411)-modified tetrahedral DNs (TDN) elicit increased internalization into cancer cell lines compared to non-aptamer modified TDNs. Adapted from (Q. Li et al., 2017).

Aside from charge neutralization, DN uptake has been increased in other ways. Whitehouse et al. engineered a 6-helix bundle nanostructure outfitted with hydrophobic cholesterol groups to increase DN association with the cell membrane (Whitehouse et al., 2019) (Figure 2B). This method dramatically enhanced internalization of the bundles (up to 10-fold) compared to unmodified DNs, and increasing number of cholesterol moieties (from one to two, or three) further facilitated the kinetics of DN uptake into HeLa cells. A subsequent investigation using this cholesterol-mediated approach with a tetrahedral DN yielded similar results (Jorge et al., 2018). Interestingly, the authors observed that adherence to the cell membrane was significantly enhanced when DNs were delivered in cell culture media containing fetal bovine serum (FBS) versus protein-depleted media (Whitehouse et al., 2019). The authors suggested that this may be caused by the electrostatic interaction of serum proteins with the DNs, thereby mediating DN interaction with the cell. It was further observed that the cholesterol moieties may aid in this putative protein-DN association. This increased interaction with the cell membrane, however, did not directly correspond to enhanced internalization, although it was noted that the larger size of the complexes as well as charge neutralization may influence their trafficking into the cell.

2.2.2. Enhancing Target Specificity

A common concern in the design of any biological delivery vehicle is its capacity to elicit undesirable off-target effects. Thus, there has been a considerable research effort to enhance therapeutic specificity by the addition of specific cell-targeting moieties to delivery vehicles. In addition to directing delivery vehicles to specific cell types, a complementary approach involves ensuring that the therapeutic payload is only released in the correct cellular environment or context. For example, drug delivery vehicles for cancer can be designed to only release their cargo in the context of the tumor microenvironment to enhance tumor-specific delivery. Enhancing specificity also often results in an increase in cellular internalization due to a higher local concentration around the target cells or tissues. To date, a variety of methods have been reported to accomplish targeted delivery with DNA nanostructures, including the use of aptamers, cell surface receptor ligands, subcellular trafficking peptides, and intracellular small molecule recognition.

Another common hurdle that must be overcome to enable DNA nanostructure-mediated delivery applications is that once endocytosed, DNs generally become trapped in endosomal compartments (Kocabey et al., 2014; D. S. Lee et al., 2016; Whitehouse et al., 2019). However, several reports have shown that DNs can be trafficked to other subcellular locations. Liang et al. covalently attached a nuclear localization signal to a tetrahedral DNA nanostructure, which successfully shuttled it to the nucleus (L. Liang et al., 2014). The authors also demonstrated that tetrahedral DNs were trafficked in a microtubule-dependent manner. In 2013, transport to both mitochondria and nuclei was demonstrated when DNA nanocages were site-selectively outfitted with subcellular localizing peptides for each respective organelle (Chan & Lo, 2014). In this investigation, vertical silicon nanoneedle technology was used to directly deliver DNs to the cytosol by physically puncturing the cells and injecting the structures. In this way, the researchers were able to circumvent the issue of endosomal entrapment by avoiding the need for endocytosis altogether, while still preserving the structural integrity of the DNs. This study provides a new platform to study the intracellular behavior of DNs, but is unlikely to be feasible for DN-based therapeutic administration because of this method’s low throughput, and the fact that many tissues are difficult to physically reach with this technology.

Peptides can also be used to target specific cell surface markers to direct DNs on a cellular, rather than subcellular, level. For example, click chemistry was used to modify a DNA tetrahedron with tumor-penetrating peptides (TPPs) serving as ligands for neuropilin-1, a transmembrane co-receptor for VEGF (Xia et al., 2016) (Figure 2C). Since VEGF is strongly expressed on the surface of certain cancer cells (in this case, glioblastoma cells) and the angiogenic blood vessels associated with tumor growth, TPPs have been proven to specifically bind and penetrate cancer cells (Xia et al., 2016, p.). It was demonstrated that TPP functionalization actually increased DN uptake in both healthy and cancer cells, but with a pronounced enhancement in the latter. In this study, antitumor drug-loaded TPP-DNs were successfully delivered to glioblastoma cells and exhibited greater cytotoxicity than delivery with the drug-loaded DN without TPPs. Similarly, Wang et al. utilized a cell-penetrating peptide conjugated to the external surface of a cargo-loaded tubular nanostructure to enhance its cellular uptake and tumor retention in vivo (Z. Wang et al., 2020).

Small molecule moieties have been demonstrated to help DNs target specific cell types as well. For example, folate, the ligand for a receptor that is overexpressed in cancer cells, can be easily conjugated to oligonucleotides through NHS chemistry and thereby incorporated into DNs. In an early study, DNA nanotubes were functionalized with folate and delivered to cancer cells, exhibiting superior adherence to cell membranes and overall internalization (Ko et al., 2008) (Figure 2D). Increasing the number of folate moieties displayed on the nanostructure resulted in an increase in internalization of structures with up to 10% folate composition, after which internalization efficiency plateaued. This folate-mediated DN targeting method has been implemented in subsequent studies, with varying levels of success (Kocabey et al., 2014; Raniolo et al., 2018).

Antibodies are commonly used as targeting elements to impart specificity, with antibodies against a myriad of therapeutically relevant targets identified, characterized, and made available commercially. In theory, it is possible to discover or raise an antibody against nearly any target, making them particularly attractive as targeting agents. A multitude of conjugation techniques have been designed to covalently attach antibodies (or antibody single-chain variable fragments) to oligonucleotides (Maerle et al., 2019; Rosier et al., 2017; J. Wang et al., 2017), many of which are described extensively in a review by Zhao et al. (D. Zhao et al., 2020). In the context of DNA nanotechnology, however, most examples of antibody-DN conjugates serve diagnostic and sensing purposes rather than for targeting and delivery.

Nucleic acid aptamers are highly compatible moieties for functionalization of DNs because they only require the addition of a sequence that is complementary to a ssDNA handle on the DN. Like antibodies, aptamers are highly diverse and capable of binding a variety of targets. Although many therapeutically relevant aptamers have already been identified, discovering new specific or higher affinity aptamer sequences is no trivial task, often involving screening massive libraries of oligonucleotides against a specific target of interest. Nevertheless, there exist many examples in the literature of aptamer-modified DNs. Li et al. designed a tetrahedral DN partially hybridized to a single aptamer specific for nucleolin, which is overexpressed on the surface of tumor cells, and showed substantial accumulation of the tetrahedra in the nucleus of MCF-7 human cancer cells (as compared with non-modified DNs) under hypoxic conditions that mimic the tumor microenvironment (Q. Li et al., 2017) (Figure 2E). Under the same conditions in noncancerous murine L929 cells, comparatively fewer DNs were internalized overall, with no difference observed in the accumulation of aptamer-modified versus unmodified DNs. In another study, DNA tetrahedra were outfitted with anti-HER2 aptamers (Ma et al., 2019). HER2 is an overexpressed cell surface marker in breast cancers, which stimulates proliferation and differentiation in breast cells and thus augmenting breast cancer malignancy. In this study, anti-HER2 aptamer-DNs were demonstrated to bind to the receptor, resulting in HER2-mediated endocytosis digestion of both the receptor and the DNs in the lysosome, effectively reducing the amount of HER2 on the surface of the cell (Ma et al., 2019, p. 2). This reduction induced apoptosis and arrested cell growth of HER2+ breast cancer cells, with no inhibitory effect on HER2- breast cancer cells. The authors also demonstrated that association with the DN increased circulation time compared to bare aptamer injection in murine models.

Aptamers have also been implemented in DNA nanostructures to bind non-cellular targets. Kwon et al. developed a DNA star-shaped design that displayed ten dengue envelope protein targeting aptamers spatially arranged to match the geometry of these proteins on the dengue virus (DENV) surface (Kwon et al., 2020). This design increased avidity of DENV-aptamer binding and served as a potent inhibitor of DENV by electrostatically trapping virions from the host cell membrane via the negative charges of the DN. This method of viral inhibition can be easily tailored to bind geometries of other viral particles, so long as aptamers specific for the virus of interest exists.

2.3. Cargo Delivery

The aforementioned strategies to enhance DN physiological stability, uptake, and specificity all help improve the functionality of these carriers in biological environments. The ultimate goal for many of these constructs is delivery of therapeutic cargo to the desired cell types. Although the research on DN behaviour in vivo is limited, many experiments have been conducted to evaluate the feasibility of DN-enabled delivery of a variety of cargos, including genetic material, proteins, and small molecule therapeutics. Below, we cover key examples of each of these payloads and how to best incorporate them into a DN carrier.

2.3.1. Nucleic Acid Therapeutics

DN-mediated gene modulation can be achieved through RNA interference (RNAi) because these genes are comprised of short RNA sequences that can be easily designed to hybridize to a ssDNA handle on the nanostructure. Alternatively, they may comprise the structural components of the DN while simultaneously serving as the genetic cargo. In 2012, a multifunctional DNA tetrahedron was developed for targeted in vivo delivery of small interfering (si)RNA (H. Lee et al., 2012). These DNs were outfitted with multiple folate moieties of defined geometrical orientation for cancer cell specificity and a site-specifically hybridized to a single siRNA molecule to elicit an over 60% knockdown of GFP expression (Figure 3A). Interestingly, the authors found that orientation of the ligands on the DN dramatically affected gene silencing. Maximizing the local density of folate ligands on the DN exhibited the greatest degree of GFP silencing even though orientation was shown not to affect intracellular uptake. The authors hypothesized that the high local folate density may influence the mechanism of DN endocytosis. Furthermore, as corroborated in other studies, DN-mediated delivery increased the circulation time of the siRNA when injected via the tail vein in a murine model. In another study, Wang et al. successfully co-delivered— both in vitro and in vivo— two distinct siRNAs along with chemotherapeutic doxorubicin by loading them into a tubular DN (Z. Wang et al., 2020). Co-delivery of the two siRNAs substantially reduced protein levels of their target genes and elicited potent antitumor activity in murine models. DNs have also been used to deliver antisense oligonucleotide (ASO) therapeutics to cells. Sleiman et al. designed 3D DNA prisms that integrate phosphorothioated ASOs into their structure (Fakhoury et al., 2014) (Figure 3B). The group found that ASO incorporation into the DN increased their nuclease resistance, allowing for enhanced gene silencing capacity compared to phosphorothioated single- and double-stranded controls. These ASO-DNs not only significantly maintained gene knockdown in HeLa cells over 72 hours but slowed knockdown recovery after 48 hours.

Figure 3. DNA nanostructure-mediated cargo delivery methods.

A) siRNA hybridized to a tetrahedral DN during assembly to enable cellular delivery of the RNAi therapeutic. Adapted from (H. Lee et al., 2012). B) Antisense oligonucleotides (ASOs) with phosphorothioate (PS)-modified backbones are hybridized to DNA cage in varying amounts. Adapted from (Fakhoury et al., 2014). C) A kite-shaped, Dox-intercalated triangular DN outfitted with two disulfide-linked tumor suppressor genes (p53) for the delivery of a combination gene and chemotherapy. Figure adapted from (J. Liu et al., 2018). D) Cross-sectional and perspective view of an aptamer-gated DNA nanorobot loaded with a protein payload. Image adapted from (S. M. Douglas et al., 2012). E) “Mirror-image” DNA tetrahedron assembled from L-DNA outfitted with streptavidin that enables modular loading of biotinylated species onto the DN. Figure adapted from (K.-R. Kim et al., 2018). F) Luciferase-loaded DNs are taken up into cells and their uptake efficiency is measured by a luminescence assay performed on cell lysate. Adapted from (Ora et al., 2016). G) Triangular DN intercalated with Dox and delivered via tail vein injection into murine models. Figure adapted from (Zhang et al., 2014). H) DNA tetrahedron self-assembled from cholesterol- and chemotherapeutic 5-fluoro-2’-deoxyuridine (FdU)-modified DNA strands. Adapted from (Jorge et al., 2018).

RNAi-based gene therapy aims to knock down and silence target genes. Intracellular delivery of whole genes, by contrast, can be used to rescue or enhance target expression. The DN-mediated delivery of a linear p53 tumor suppressor gene was reported in 2018 (J. Liu et al., 2018). The system was multi-functional, featuring incorporation of doxorubicin (a DNA-intercalating chemotherapeutic drug), MUC1 aptamers for targeted tumor cell delivery, and a disulfide linker for the attachment and controlled release of the capped p53 gene (Figure 3C). This delivery system combined gene and chemotherapeutic technologies to effectively inhibit the growth of multidrug resistant tumors both in vitro and in vivo with no evidence of systemic toxicity. This study elegantly demonstrated the potential and versatility of DNA nanotechnology when coupled with diverse functional moieties in biological systems.

2.3.2. Proteins

Peptides and proteins are becoming an increasingly large portion of the therapeutic market. However, their use is often limited by delivery due to barriers like stability and bioavailability. These molecules are susceptible to protease degradation, agglutination, or limited absorption into target tissues (Bruno et al., 2013). Another issue for these molecules is the inadvertent recognition by, and activation of, the immune system. As a result, a considerable amount of research has been conducted to develop drug delivery vehicles with the capacity to mitigate these problems.

In 2012, Church et al. described a DN “nanorobot”, i.e. a “smart” nanostructure capable of transporting molecular payloads in a targeted, conditional manner (S. M. Douglas et al., 2012). This logic-gated nanorobot utilized aptamer “locks” that, upon binding to their targets, triggered the opening of a DNA clamshell-like structure (Figure 3D). Once open, the nanorobot exposed fluorescently labelled antibody fragment payloads specific for various cell surface markers that could bind to and label cells. The two aptamer locks were designed in such a way that each had a distinct target, and the structure only opened if both aptamers were engaged with their ligand, imparting logic-gated AND functionality to the DN. This study also demonstrated this system’s ability to alter cell signaling in vitro. Not only did this work result in the development of an aptamer-encoded logic gate targeting system for DNs, but it is one of the earliest examples of DNs used to deliver protein cargos.

In a separate study, this nanorobot platform was extended to deliver chemotherapeutic molecules to cancer cells in vivo (S. Li et al., 2018). In this work, aptamers against the cancer cell marker nucleolin were used as the locking mechanism to protect the thrombin payload from being exposed to the physiological environment in absence of the nucleolin-expressing tumor cells. Exposing the thrombin to the tumor microenvironment by the activated nanorobot promoted vascular thrombosis and concomitant tumor necrosis and inhibition of growth in murine models. Furthermore, this chemotherapeutic platform was proven to be immunoquiescent and nontoxic in mice and Bama miniature pigs, showing great potential as a clinical cancer treatment.

The above examples of protein delivery utilized bifunctional linkers to conjugate payloads to DNA handles that hybridize to the DN. An alternative approach for protein loading into DNs is to implement biotin-modified strands in the DN design and biotin-modified protein cargo units. Addition of tetravalent streptavidin “glue” then links the DN with its cargo. Indeed, this approach was accomplished by Kim et al. using a L-DNA tetrahedron (K.-R. Kim et al., 2018). L-DNA tetrahedra were selected because they have been shown to exhibit increased intracellular stability (since L-DNA is not recognized by nucleases or other proteins) and uptake (K.-R. Kim et al., 2014). Using this DN-streptavidin system, three distinct biotin-modified proteins were successfully delivered to cells and executed various functions (e.g. triggering apoptosis, gene modification, and carbohydrate hydrolysis) (K.-R. Kim et al., 2018) (Figure 3E). This platform was also shown to localize to tumor sites in murine models. A similar approach was implemented in which biotinylated hexagonal tube DNs were loaded with streptavidin-conjugated Lucia luciferase enzyme and successfully delivered to HEK293 cells without altering luciferase activity during transfection (Ora et al., 2016) (Figure 3F).

2.3.3. Small Molecules

Small molecules comprise the vast majority of the pharmaceutical market. Unfortunately, their use is limited in drug-resistant cells, necessitating alternative delivery approaches to circumvent this issue. DNs are particularly ideal carriers for DNA-intercalating small molecule drugs, such as doxorubicin. Doxorubicin (Dox) is a well-known chemotherapeutic, and as such has been implemented in multiple studies regarding DN-mediated small molecule delivery. For example, Dox-loaded DNA triangles and tubes were delivered to a drug-resistant cancer cell line and exhibited enhanced cytotoxicity to these cells compared to Dox-loaded duplex DNA or free Dox (Jiang et al., 2012). The DNs circumvented Dox resistance by enhancing cell uptake of the drug while simultaneously inhibiting lysosomal acidification, an approach known to reverse drug resistance. In a follow-up study, Dox-loaded DNA triangles were delivered via tail vein injection to murine models and demonstrated passive tumor-targeting capability with long-lasting accumulation in tumor regions (Zhang et al., 2014) (Figure 3G). These structures successfully exhibit antitumor activity without any noticeable immune recognition or systemic toxicity, as measured by cytokine secretion, whole blood analysis, and murine body weight measurements.

A structurally related chemotherapeutic, daunorubicin, has also been delivered via intercalation into DNs (Halley et al., 2016). Rod-like DNs loaded with daunorubicin were shown to circumvent efflux-pump-mediated drug resistance in leukemia cells. Similar to the Dox studies, intercalation into DNs enhanced cellular uptake of the drug, resulting in increased drug efficacy. Notably, it was determined that overloading DNs had a detrimental effect on drug efficacy, likely caused by DN aggregation. Small molecule delivery can also be achieved through means other than drug intercalation. For example, 5-fluoro-2’-deoxyuridine (FdU) oligomers were incorporated into a DNA tetrahedral scaffold featuring cholesterol anchors that improved cellular internalization (Jorge et al., 2018) (Figure 3H). FdU is a more effectively cytotoxic derivative of the extensively used chemotherapeutic 5-fluorouracil. FdU-scaffolded DNs were shown to bypass the low sensitivity of colorectal cancer cells to 5-FU in part by increasing the intracellular concentration of the drug. Taken together, the results of these studies illustrate the immense potential and versatility of DN therapeutic delivery systems.

2.4. Environmental Stimulus Response

The inclusion of stimulus-responsive elements into DN design would enable greater control in biological contexts, especially for spatiotemporally precise cargo release. Physiologically relevant stimuli include pH, redox, light, and small molecule triggers. Although many stimulus-responsive moieties have been designed for DNA-based systems, very few have been demonstrated to work inside a cell. One such trigger shown to operate in a biological context is aptamer-triggered dissociation of DN subunits (Banerjee et al., 2013) or conformational changes (S. M. Douglas et al., 2012; S. Li et al., 2018). Banerjee et al. demonstrated such a system in which the biologically relevant molecule, cyclic-di-GMP (cdGMP), was used as a chemical trigger for controlled release of fluorescent cargo encapsulated by an icosahedral DNA nanostructure (Banerjee et al., 2013). In this design, two halves of the icosahedron were held together by strands that were also aptamers for cdGMP; binding to the small molecule then pried apart the two halves and facilitated cargo release. Although not demonstrated in a biological context, this promising strategy can be applied to other caged DNs encapsulating various cargos.

2.4.1. pH

Modulation of DNs by pH is a physiologically relevant approach, owing to the fact that the endosomal compartments that DNs initially enter display an acidic pH around 5.5. pH responsivity is also therapeutically relevant, as the tumor microenvironment also has an acidic pH (Reshkin et al., n.d.). Some researchers have implemented well-characterized acid-dependent nucleic acid secondary structures, such as the i-motif, (Day et al., 2014) to enable DN reconfiguration in acidic environments. Keum and Bermudez used the i-motif to demonstrate pH-controlled assembly and disassembly of a tetrahedral DN (Keum & Bermudez, 2012) (Figure 4A). This platform also enabled the controlled release of EGFP under acidic conditions. In a comparable study, the i-motif was introduced into a tetrahedral DN encapsulating RNase A via direct conjugation to the nanostructure (S. H. Kim et al., 2017). The attachment enabled reversible pH-dependent accessibility to the protein cargo over a pH range of 6.0–8.3. Hoogsteen triplex interactions (Figure 4B) have also been implemented to modulate the assembly of a DNA tetrahedron in a pH range of 5.0–8.0, where DN self-assembly is triggered under acidic conditions, in contrast to the behavior of the i-motif designs (Z. Liu et al., 2013). Although these methods demonstrate great promise for controlled drug delivery, to date, there are few examples of pH-triggered cargo release in biological systems (S. Liu et al., 2020).

Figure 4. Stimulus-responsive moieties in DNA nanostructures.

A) Chemical structure of DNA i-motif cytosine-cytosine⁺ base pairing (left) and a schematic of an i-motif-incorporated, DNA tetrahedron that dissociates in an acidic environment (right). Adapted from (Keum & Bermudez, 2012). B) Chemical structures of pH-dependent Hoogsteen triplex thymine-adenine-thymine (TAT, top left) and cytosine⁺-guanosine-cytosine (C⁺GC, bottom left) base pairing and a schematic of a DNA tetrahedron designed to utilize these interactions to assemble and disassemble according to environmental pH (right). Figure adapted from (Z. Liu et al., 2013). C) DNA tetrahedron assembly via oxidation of cysteamine-modified DNA. Subsequent disassembly can be achieved via glutathione (GSH)-mediated reduction of the resulting disulfide bond. Figure adapted from (B. Wang et al., 2019). D) Azobenzene UV-triggered trans to trans cis isomerization (left) and a schematic of the light-droven opening and closing of azobenzene-incorporated DNA nano-tweezers (right). Adapted from (X. Liang et al., 2008).

2.4.2. Oxidation state

Considering the ubiquitous existence of reductive or oxidative environments in physiological contexts, the development of redox-sensitive DNs would provide many therapeutically relevant applications. A recent, generalizable example of such a system was demonstrated by Wang et al. (B. Wang et al., 2019). DNA tetrahedra were electrostatically assembled with the aid of positively charged cystamine, which features a redox-active disulfide bond in the center of its chemical structure. In a reducing environment, this disulfide bond breaks to restore the monomeric cysteamine, which bears only a single positive charge, thereby reducing its capacity to hold together the DN assembly. This platform was shown to be reversibly regulated in the presence of either reducing agent glutathione (GSH) or oxidizing agent hydrogen peroxide (Figure 4C). A subsequent study showed the reversible modular assembly and disassembly of tubular DNs using disulfide bond redox chemistry as well (Del Grosso et al., 2020), although neither of the aforementioned nanostructures were tested in a cellular context. However, Wang et al. recently demonstrated a nanostructure that exhibited GSH-dependent opening and cargo release in vivo by implementing disulfide-bonded “locking” strands that keep the tubular nanostructure closed until it is exposed to the reducing environment of the cytosol (Z. Wang et al., 2020). Given the stark difference in reduction potential between the cytosol and the extracellular environment, it is likely that redox-switchable DNs will find increasing application as other issues (e.g. endosomal escape) are addressed.

2.4.3. Light

Photoresponsivity is a unique functional trait that would allow remote spatiotemporal control over DN behavior by simple irradiation with light. Photons are also efficient inputs for nanostructure modulation due to their high spatial and temporal precision, and the ease of accessibility compared with diffusion of a molecular trigger. One example of photon-powered DN conversion is the azobenzene-intercalated DNA “tweezer” structures reported by Fan and coworkers (X. Liang et al., 2008) (Figure 4D). The UV-inducible trans to cis isomerization of intercalated azobenzene results in a destabilization of the tweezer’s DNA duplexes, resulting in “opening” of the tweezer upon exposure to light. The closed conformation is reversibly achieved by visible light irradiation that allows rehybridization of the azobenzene DNA duplex. Photoresponsivity has also been utilized with azobenzene moieties to induce conformational changes in DNA tetrahedra via a similar mechanism of alternating wavelengths of light (Da Han et al., 2011). For implementation of these systems into biological applications, future studies must be conducted to evaluate the biosafety of azobenzene-incorporated DNA. Other barriers to clinical use of an azobenzene photo-switchable delivery vehicle include poor tissue permeability and automatic cis-trans relaxation (Mulatihan et al., 2020).

2.5. Immune Modulation

Modulating an immune response is a therapeutically relevant endeavor. Any foreign entity entering the body risks immune recognition and clearance by the host’s immune system, thus greatly decreasing its potential efficiency. Conversely, immunotherapeutics are designed to purposely stimulate an immune response against a specific disease-associated antigen in the interest of triggering immune clearance of the target pathogen, or to kill diseased cells that present a particular epitope. For an excellent discussion of the current research involving nucleic acids as adjuvants and nanocarriers for immune modulation, the reader is referred to a recent review by Comberlato et al. (Comberlato et al., 2019). Interestingly, in vivo studies in murine models have shown that DNs with architectures comprised of solely of DNA with no RNA counterparts have a very low immunogenic profile, with no detectable DN-triggered immune response as measured by cytokine and immunoglobulin presence (Hong et al., 2018; H. Lee et al., 2012; S. Li et al., 2018; Zhang et al., 2014). However, some exceptions to this trend exist (Perrault & Shih, 2014; Schüller et al., 2011), suggesting that architectural variations between DNs may play a significant role in determining their propensity to elicit an immune response. In a comprehensive study on the effects of nucleic acid nanostructure shape, sequence, connectivity, and composition on immune recognition potential, Hong et al. demonstrated that all tested structures were immunoquiescent (stimulating no interferon response) in the absence of transfection agents (Hong et al., 2018). Interestingly, after complexation with Lipofectamine, only structures comprised completely of DNA evaded the immune response, while RNA-containing structures exhibited varied immune stimulation. Delivery of Lipofectamine alone did not trigger an inflammatory immune response. In sum, the authors determined that Lipofectamine-mediated RNA nanostructure immunostimulation is dependent primarily on shape and connectivity rather than nucleic acid sequence. However, it has since been demonstrated by Qi et al. that a nanostructure comprised solely of RNA in the absence of any transfection agents has successfully activated an immune response (Qi et al., 2020). The variability in these experimental results for DNA-comprised nanostructures may be caused by differences between in vitro and in vivo models, illustrating the importance of robust testing in animal models for the assessment of DN physiological behavior and biosafety before any clinical applications can be realized. In addition, the great diversity and complexity of various DNA or RNA nanostructures could play a significant role in the immune response, so detailed immunologic studies will be necessary for each specific platform.

2.5.1. CpG Adjuvants

The innate immune system has developed many avenues for the clearance of foreign DNA from the cell in both prokaryotic and mammalian systems. We refer the reader to the DN-oriented review on immune compatibility by Surana et al. for an overview of these pathways (Surana et al., 2015). Of these pathways, the mammalian innate immune system’s Toll-like receptor (TLR)-mediated inflammatory response is perhaps the most therapeutically relevant, especially in the case of DNA-mediated immunostimulation. TLRs reside in endolysosomal compartments and serve to detect exogenous material taken up into the cell. Different classes of TLRs have unique specificities, including dsRNA, dsDNA, ssDNA, polysaccharides, and CpG DNA. Unmethylated CpG (cytosine-phosphate-guanosine) DNA is characteristic of bacterial genetic material and is known to elicit strong immune responses through recognition by TLR9. For this reason, CpG DNA is a commonly used adjuvant in vaccine design.

In vitro immune stimulation by CpG oligonucleotide structures has been evaluated for a few different structures. For example, polypod-like structured DNA comprised completely of a variable number of CpG motifs induced cytokine production in macrophage-like RAW264.7 cells (Mohri et al., 2012) (Figure 5A). Introducing greater amounts of CpG into the structure resulted in increased immunostimulatory activity, although the addition of DNA in this system exhibited an upper limit of efficacy due to a decline in serum stability. In a separate study, an origami tube with 62 CpG-decorated staple strands (a much higher concentration of CpG than the aforementioned study, which featured only up to 8 CpG motifs) was incubated with freshly isolated spleen cells and shown to induce inflammatory cytokine production as well as antigen-presenting immune cell activation (Schüller et al., 2011) (Figure 5B). This immunostimulation was greatly enhanced compared to incubation and Lipofectamine-aided transfection of equal amounts of CpG oligonucleotides. CpG-incorporated DNA tetrahedra have also been studied for their ability to stimulate the immune system (Jiang Li et al., 2011) (Figure 5C). The use of tetrahedral DNs in this system mitigates any need for transfection agents and avoids issues of cytotoxicity, and the structures were efficacious in triggering a robust CpG concentration-dependent inflammatory cytokine response as well.

Figure 5. Immune modulation methods using DN functionalization.

A) TLR9-induced immunostimulation by DNA nanotubes modified with three kinds of orthogonal cytosine-phosphate-guanine (CpG) oligonucleotides and handle sequences (CpG-H′s). CpG-H′ PTO-labelled strands feature stabilizing phosphorthioate (PTO)-modified backbones, and CpG-H′ chimera strands feature a PTO-modified CpG segment and a handle sequence whose backbone is unmodified. Figure adapted from (Schüller et al., 2011). B) CpG oligonucleotides arranged in polypod-like structures comprised of three (top left), four (top right), six (bottom left), or eight (bottom right) oligonucleotides. Adapted from (Mohri et al., 2012). C) A DNA tetrahedron bearing CpG motifs that induces TLR9-dependent immune activation. Adapted from (Jiang Li et al., 2011). D) Schematic of a synthetic DNA vaccine co-delivering CpG adjuvants and model antigens and the subsequent DN-mediated antigen presentation and activation of B and T cells. Figure adapted from (X. Liu et al., 2012). E) Fabrication of a dual adjuvant (dsRNa and CpG loop motif) and antigen-carrying DNA delivery vehicle (top) and atomic force microscopy images (bottom) of its unloaded open (left), cargo-loaded open (middle), and cargo-loaded closed state (right). Figure adapted from (S. Liu et al., 2020).

2.5.2. Vaccines & Immunotherapeutics

The World Health Organization estimates that vaccines prevent 2–3 million deaths per year (Immunization, 2020), and the recent Covid-19 pandemic has starkly highlighted the need for platforms that can rapidly trigger immunity against novel pathogens. However, despite the use of vaccines for well over 200 years, their design is still challenged with the trade-off between safety and efficacy. Subunit vaccines, composed of pathogenic peptide and/or polysaccharide antigens, are arguably the safest. However, they tend to generate weaker immune responses and provide no guarantee that the delivered antigen will undergo the correct type of antigen processing, which affects the type of T cell response (cytotoxic or effector) a given vaccine will elicit (Baxter, 2007; Kaufmann et al., 2014; Plotkin, 2003). In recent years, the unique properties of nanomaterials have been harnessed to develop nano-vaccines capable of targeting specific antigen processing pathways (C. G. Kim et al., 2019, p. 8). Nanotechnology has also presented many promising vaccine delivery strategies (Silva et al., 2013), owing in part to the fact that sub-micrometer particles are more easily phagocytosed by antigen presenting cells (APCs) (Peek et al., 2008). DNA nanotechnology offers an attractive and potentially safer platform for vaccine delivery due to the biocompatibility, biodegradability, and nontoxicity of the structures formed (Jiang et al., 2012; Ko et al., 2008; X. Liu et al., 2012; Zhang et al., 2014).

In a pilot DN vaccine delivery experiment, a DNA tetrahedron was outfitted with CpG oligonucleotide adjuvants and electrostatically associated streptavidin model antigens (X. Liu et al., 2012) (Figure 5D). Cellular internalization studies showed enhanced codelivery of both antigen and adjuvant using the tetrahedral nanostructure. In a following experiment, mice were administered primary and secondary immunizations via tail vein injection of the DN vaccine and then challenged with streptavidin alone and tested for anti-streptavidin antibody and memory B cell responses. The DN vaccine exhibited approximately 2-fold higher IgG levels than free streptavidin + CpG, even after 70 days. Elevated levels of memory B cells in these mice demonstrate that the stronger, longer-lasting antibody response is due in part to the generation of streptavidin-specific memory B cells. Furthermore, this system reportedly required lower amounts of antigen and adjuvant to induce a specific immune response than reported elsewhere (Klinman et al., 1999), a strategy which may reduce the chance of unwanted nonspecific immune activation.

Very recently, a more complex and multifunctional DN-based vaccine delivery vehicle was reported (S. Liu et al., 2020). A modified version of the previously reported DNA nanorobot (S. M. Douglas et al., 2012; S. Li et al., 2018) (Figure 5E) was designed for intracellularly delivery of ovalbumin, a model antigen, with CpG and dsRNA motifs that stimulate distinct TLRs (TLR9 and TLR3, respectively) (S. Liu et al., 2020). The “locking” strands for this redesigned platform are comprised of pH-responsive i-motifs that allow nanostructure opening in the acidic environment of the endosome upon uptake. The nanodevice was successfully endocytosed into bone marrow dendritic cells and triggered cytotoxic T cell activation in vitro. Delivery of cancer antigens with this platform into murine melanoma models resulted in potent antigen-specific T cell responses and tumor regression. Furthermore, these T cell immune responses demonstrated long-term protection against tumor rechallenge in these models. The success of this highly functionalized, addressable DN vaccine platform may be realized as a useful tool in the development of personalized vaccines.

3. CHALLENGES & OUTLOOK

DNA nanostructures show great promise in addressing current challenges in biomedicine. The commercial feasibility and large-scale production of DNA origami structures of DNs as therapeutics has already been demonstrated to be achievable and cost-effective (Praetorius et al., 2017), as well as economically competitive with more traditional technologies (Weiden & Bastings, 2021). However, a number of key areas remain to be explored before these materials can translated to the clinic. For example, more systematic studies of nanostructure metabolism, bioavailability, pharmacokinetics, renal clearance, and safety are necessary, especially after significant modification of the original structures with polymeric or polypeptide coatings. Furthermore, to date there are no specific guidance documents from regulatory agencies for the clinical translation of therapeutic nucleic acids, the existence of which would encourage more standardized protocols for their synthesis and characterization, enabling a clearer path towards clinical implementation (Afonin et al., 2020). It is also important that the translational benefit of DNs relative to pre-existing therapies be clearly demonstrated via future side-by-side comparative experimental analyses.

It is clear from the present body of work that more comprehensive studies are required to improve the mechanistic understanding of DNs in vivo and to clarify contradictory results in several recent reports. For example, it has been noted in this review and elsewhere (Weiden & Bastings, 2021) that immune responses against naked DN administration in mouse models has been somewhat nebulous, being detected in some studies, but not others. Additionally, the Sleiman group has recently delineated many examples of conflicting data on the cellular internalization efficiency of DNA tetrahedron structures (Lacroix et al., 2019). The authors attribute this inconsistency to a misinterpretation of cellular fluorescence data as an indicator of cellular uptake, showing that cyanine-dye labelled DNs are likely degraded extracellularly, with intracellular fluorescence resulting from cleaved, non-attached dyes entering the cell due to their positive charge and accumulating in the mitochondria as opposed to intact structures. They conclude that intracellular fluorescence in previous studies (Charoenphol & Bermudez, 2014; Keum et al., 2011; D. S. Lee et al., 2016; Walsh et al., 2011) cannot be correlated with uptake of the intact structure. Furthermore they demonstrated that FRET-based experiments are similarly confounded due to a high level of “random FRET” that must be accounted for, caused in part by the high local concentration of cyanine dyes accumulated in the mitochondria (Lacroix et al., 2019). Although this report has shown that DNA tetrahedra do not enter cells as readily as previously thought, the authors note that this could serve as an advantage for drug delivery applications, for which nonspecific uptake would be undesirable. This investigation serves as an important warning, highlighting the necessity for careful design of fluorescence-based assays, and control experiments that directly probe the kind of degradation that could give rise to artifacts and inaccurate results.

Another under-investigated area of therapeutic DN research is the mechanisms by which they are internalized by cells. Over the last five years, our understanding of the processes and pathways of endocytosis, particularly for nanoparticle uptake, has made tremendous progress (Rennick et al., 2021). Advances in endocytosis research has shown that chemical endocytic pathway inhibitors traditionally used to study uptake pathways are nonspecific, convoluting previous data obtained using this technique (L. Liang et al., 2014; P. Wang et al., 2018). To further complicate the issue, it has been shown that pathway crosstalk can occur, such that some components traditionally associated with one pathway may be involved multiple pathways, and inhibition of one pathway may cause another to be upregulated in compensation. It has been suggested that future experiments probing nanoparticle endocytic mechanisms should implement genetic knockout or knockdown of distinct endocytic pathway components to avoid off-target effects of chemical inhibitors or should involve markers known to be specifically internalized by a particular pathway (Rennick et al., 2021). It is also important to note that particles internalized by the same mechanism are not necessarily trafficked to the same compartments and conversely different uptake mechanisms can still lead to particles trafficked to the same location. Furthermore, it has recently been discovered that some endocytic pathways are not exhibited in all cell lines, and the same particles may be internalized by different mechanisms in different cell lines. Modern investigations of the pathways involved in DN internalization that bear in mind the recent advances in our understanding of endocytic mechanisms can inform better cell and tissue targeting of DN-based therapeutics and avoid unexpected results when studying DNs in different cell lines.

Conclusion

In this review, we have demonstrated the diverse functionalities that can be realized by the modification of DNA nanostructures with both covalent and noncovalent strategies. DNA nanotechnology has a bright future in biomedical applications. The structural programmability and functional versatility of DNs provide almost limitless opportunities for new and improved biological functions. Future studies can explore additional combinations of functional moieties for highly specific applications, such as delivery vehicles that enable both cellular targeting and subcellular trafficking, or that enhance codelivery of distinct cargos. Photo- and redox-responsive DNs have yet to be implemented in biological systems, but once realized may allow sequential responsivity to multiple external stimuli. For example, a Russian doll-like mechanism of delivery can be designed, which upon internalization and input of a particular stimulus triggers release of a secondary nanostructure payload featuring its own cargo, trafficking, and responsivity profile. Another potential avenue of research related to immunomodulation is to utilize DNs to direct a specific type of T cell-mediated immune response by virtue of antigen delivery to either the exogenous or endogenous antigen processing and presentation pathways.

We have outlined key modifications to enhance the stability and functionality of DNA nanostructures in a biological context and discussed key barriers to their clinical translation. The past four decades of DNA nanotechnology research has demonstrated huge leaps in the field’s understanding of DN design elements and how these designs can interface with biology. Continued rapid progression and interdisciplinary collaboration in this field will promote exciting new directions for DNA nanotechnology for many years to come.