Development of DNA nanostructures for high affinity binding to human

The development of nucleic acid therapeutics has been hampered by issues ... DNA structures with strong binding affinity to Human Serum Albumin (HSA)...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/JACS

Development of DNA Nanostructures for High-Affinity Binding to Human Serum Albumin Aurélie Lacroix,† Thomas G. W. Edwardson,† Mark A. Hancock,‡ Michael D. Dore,† and Hanadi F. Sleiman*,† †

Department of Chemistry and Centre for Self-Assembled Chemical Structures (CSACS), McGill University, 801 Sherbrooke Street West, Montreal, Quebec H3A 0B8, Canada ‡ SPR-MS Facility, McGill University, 3655 Promenade Sir William Osler, Montreal, Quebec H3G 1Y6, Canada S Supporting Information *

ABSTRACT: The development of nucleic acid therapeutics has been hampered by issues associated with their stability and in vivo delivery. To address these challenges, we describe a new strategy to engineer DNA structures with strong binding affinity to human serum albumin (HSA). HSA is the most abundant protein in the blood and has a long circulation halflife (19 days). It has been shown to hinder phagocytosis, is retained in tumors, and aids in cellular penetration. Indeed, HSA has already been successfully used for the delivery of small-molecule drugs and nanoparticles. We show that conjugating dendritic alkyl chains to DNA creates amphiphiles that exhibit high-affinity (Kd in low nanomolar range) binding to HSA. Notably, complexation with HSA did not hinder the activity of silencing oligonucleotides inside cells, and the degradation of DNA strands in serum was significantly slowed. We also show that, in a site-specific manner, altering the number and orientation of the amphiphilic ligand on a self-assembled DNA nanocube can modulate the affinity of the DNA cage to HSA. Moreover, the serum half-life of the amphiphile bound to the cage and the protein was shown to reach up to 22 hours, whereas unconjugated single-stranded DNA was degraded within minutes. Therefore, adding protein-specific binding domains to DNA nanostructures can be used to rationally control the interface between synthetic nanostructures and biological systems. A major challenge with nanoparticles delivery is the quick formation of a protein corona (i.e., protein adsorbed on the nanoparticle surface) upon injection to biological media. We foresee such DNA cage−protein complexes as new tools to study the role of this protein adsorption layer with important implications in the efficient delivery of RNAi therapeutics in vitro and in vivo.



labeled serum albumin is found at the tumor site.3,5 The encapsulation of drugs into albumin nanoparticles and coupling of small molecules to the protein have led to FDA-approved drugs for cancer treatment. Albumin-bound chemotherapeutics are currently the most common type of nanoparticles treatment tested in clinical trials.6 For example, Abraxane is an albumin paclitaxel nanoparticle currently available for the treatment of breast, lung, and pancreatic cancers.7 Currently in Phase III clinical trials, Aldoxorubicin (INNO-206) releases doxorubicin from albumin within the acidic environment of cancer cells.8 Moreover, Lippard and co-workers have designed a series of platinum prodrugs to interact non-covalently with HSA; their complexes protected the drugs from degradation, and significantly higher therapeutic potential was reported.9 HSA is also known to interact with GP60, an endothelial cell-surface receptor that mediates its uptake to the tumor interstitium via transcytosis.8,10,11 This mechanism allows even distribution of the drug inside tumor tissues. All the examples noted above

INTRODUCTION Human serum albumin (HSA) is the most abundant protein in blood, accounting for 50−60% of total serum proteins. This 66 kDa plasma protein transports different compounds throughout the body, especially long chain fatty acids that are otherwise insoluble in serum. HSA has an extraordinarily long circulatory half-life (19 days)1 and unique binding propertieseach molecule is capable of binding up to seven equivalents of fatty acids at multiple binding sites.2 HSA has also been shown to bind other important, biologically-relevant molecules such as bilirubin, metal ions, and a variety of therapeutics drugs.3 This non-covalent binding of such ligands to HSA increases their circulatory half-lives in the blood and improves biodistribution throughout the body. HSA has emerged as a promising tool for drug delivery due to its important binding capacity, low cost, and high bioavailability. It has been shown to accumulate and be retained into solid tumors tissues due to the enhanced permeability and retention (EPR) effect.4 This unique characteristic of solid tumor tissues causes preferential accumulation of macromolecules, as evidenced by studies where 3% to 25% of dye-labeled or radio© 2017 American Chemical Society

Received: March 23, 2017 Published: May 5, 2017 7355

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362

Article

Journal of the American Chemical Society

Figure 1. Binding of D-DNA to human serum albumin. (A) Structure of synthetic D1 molecule. (B) D-DNA molecules binding to HSA in 1X PBS at room temperature. (C) Electrophoretic mobility shift assay reveals that free D1 (lower bands) binds HSA (upper bands) in a dose-dependent manner. D1 (100 nM) was incubated with increasing equivalents of HSA (0, 0.10, 0.23, 0.43, 0.67, 1.0, 1.5, 1.85, 2.3, 3.0 equiv). The lane 11 is HSA loaded with no DNA, showing that HSA does not stain the gel. (D) Surface plasmon resonance demonstrates saturable, dose-dependent binding of D1 (0−500 nM; 2-fold dilution series) to amine-coupled HSA in HBS-EP running buffer.

illustrate how optimizing the interaction between small molecules and albumin can improve their therapeutic properties in vivo: protection from degradation, enhanced accumulation within cancer tissues, and improved cellular uptake. In contrast, significant challenges have impeded the development of methods for the in vivo delivery of nucleic acid-based therapeutics such as siRNAs, antisense oligonucleotides, and nucleic acid-based delivery systems such as DNA nanostructures.12,13 When injected into the human body, oligonucleotides are typically degraded within minutes by serum nucleases. Clearance by the kidney, difficulty crossing the epithelial barrier, and/or poor cellular uptake also considerably reduce the amount of nucleic acid which reaches its target.14 To date, very few reports in the literature have examined the potential of HSA to circumvent such issues for oligonucleotide delivery. For example, Manoharan and co-workers conjugated ibuprofen to an antisense oligonucleotide, which in turn bound to HSA with micromolar affinity.15 White and co-workers also conjugated siRNAs to a free cysteine residue on HSA, allowing for extravasation of the siRNA and silencing effect in cardiomyocytes.16 To improve upon the half-life of DNA-based therapeutics and DNA nanostructures, their biodistribution, and cellular uptake, we wanted to develop a new strategy to complex DNA with HSA. To allow possible release of the oligonucleotide from the protein, we have designed a new amphiphilic ligand that is able to bind HSA non-covalently. Binding to this highly abundant serum

protein can also provide key insights about the role of protein coating and protein corona on nucleic acid delivery. The rapid formation of a protein corona affects the pharmacokinetics and physiological effects of nanoparticles.17,18 For example, opsonization is a protein labeling process that allows fast uptake into the mononuclear phagocyte system and clearing of the pathogens or, here, the nanostructure.19 On the other hand, HSA can act as a dysopsonizing protein by creating a protective shield and thus preventing phagocytosis of nanocarriers.20 Because of its complexity, protein coating constitutes a major hurdle that slows down the approval and commercialization of nanoparticles as therapeutics.21 Harnessing their interaction with HSA provides a novel tool to control the pharmacokinetic properties of drug delivery systems. In this contribution, the discovery of a supramolecular interaction between amphiphile-decorated DNA structures and HSA is presented. DNA nanotechnology has emerged as promising method to produce targeted drug-delivery vehicles22 thanks to the ease of manipulating their structural parameters such as size, shape, rigidity, and presentation of functional groups. Capitalizing upon this flexibility, we used a minimal DNA scaffold, a DNA nanocube, to present different numbers and geometries of protein binding ligands to tailor the interaction between a synthetic particle and native proteins. By optimizing the structural properties of the ligand, the multivalent presentation of the amphiphiles led to high-affinity binding 7356

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362

Article

Journal of the American Chemical Society

Figure 2. Cube assembly, site-specific decoration with D1 molecules, and binding to HSA. A cube is assembled from four DNA clips in one-pot. The eight single-stranded regions of the cube allow site-specific positioning of the D-DNA molecules. Dashed-line box is representative of cubes readily binding to HSA at room temperature in 1X PBS. This scheme is not drawn to scale.

with the carboxyl group.28 While these native ligands for HSA possess methyl terminated aliphatic chains, our D1 modification presents a hydroxyl group at the termini. To assess the importance of the alcohol group on the dodecanol chain for HSA binding, we synthesized D2, a dodecane version of D1 (more hydrophobic) (Supplementary Figure 4). By using mobility shift assays, we did not observe significant changes on the binding affinity (Supplementary Figure 6). Therefore, the alcohol group does not impede binding to the protein. We also synthesized shorter, less flexible, and more hydrophobic D-DNA amphiphiles removing the C6 (hexane) chains (D3 and D4) (Supplementary Figure 4).23 We hypothesized that less flexible molecules may not fit as well in the different binding pockets of the protein, and that this could lower the affinity. EMSA indicated that D3 and D4 had a slightly lower binding affinity than D1 and D2 (Supplementary Figure 6). Sequences and structures are detailed in Supplementary Table I. Overall, these modifications failed to significantly alter the EMSA results observed with D1, and thus the dendritic C12 (dodecane) chains appear to be the most important structural element for binding to HSA. D1 remained our ligand of choice for the construction of all subsequent D-DNA and cube structures utilized in this study. EMSA assays are typically considered qualitative because the samples are not at equilibrium and various parameters can influence and disrupt the binding.26 For example, the difference in charge between the ligand and HSA and their respective molecular weights, the conformation of the binding, and/or the temperature are all important parameters that can disrupt binding during the electrophoretic process. Therefore, we used surface plasmon resonance (SPR) to quantitatively examine binding between D1 and HSA. SPR has emerged as a powerful technique to examine protein interactions in real-time without the use of fluorescent or radioactive labels.29 Dose-dependent binding of D1 to amine-coupled HSA surfaces was observed (Figure 1D), whereas equimolar concentrations of unmodified DNA strand (negative control) failed to interact with HSA (Supplementary Figure 9). Tested over multiple surface densities, the interaction between D1 and HSA exhibits fast association and dissociation kinetics with strong overall affinity in the low nanomolar range (41 ± 11 nM) (Supplementary Figure 10). When all available binding sites on HSA are occupied by D1,

(i.e., low nanomolar range) of the cube to HSA. We showed that the amphiphilic oligonucleotide exhibited significantly higher stability in serum being protected by both the nanocage and the binding to HSA. Cellular assays confirmed that the amphiphilic ligand fully retained its biological activity inside mammalian cells.



RESULTS AND DISCUSSION Binding of D-DNA to Human Serum Albumin. We have developed dendritic DNA branching C-12 chains conjugated to DNA (called D-DNA)23 that show promise for binding noncovalently to HSA (Figure 1A). Such amphiphiles present hydrophobic alkyl chains yet do not cause aggregation of the oligonucleotide-lipid structure. They are very easily synthesized using phosphoramidite chemistry and standard solid-phase synthesis on a DNA synthesizer.24,25 Binding of D-DNA to HSA (molecule D1, Figure 1B) was first examined qualitatively by electrophoretic mobility shift assays (EMSA).26 This is a technique commonly used to study DNA−protein interactions, where the bound complexes migrate with reduced mobility compared to free DNA in an electrophoretic field. Fixed amounts of D1 were mixed with increasing amounts of HSA protein, separated by polyacrylamide gel electrophoresis (PAGE), and stained with GelRed (DNA intercalator). A discrete band with clearly reduced mobility relative to the DNA band was observed as we increased the concentration of the protein, suggesting the formation of a well-defined D1−HSA complex (Figure 1C). A single-stranded unmodified DNA sequence of the same length did not bind HSA (Supplementary Figure 5). Additionally, we incubated D1 with human serum, a complex mixture of all serum proteins at physiologically relevant concentrations. The electrophoretic shift observed was the same as that with pure HSA (Supplementary Figure 11). Binding to HSA was specific since other common serum components (i.e., IgG, the most abundant antibody in human body fluids and albumin-depleted serum samples) failed to generate similar band shifts in additional EMSA assays (Supplementary Figure 11). HSA was reported to have seven binding sites for fatty acids (FAs).27 These sites, displaying different affinities and positioned asymmetrically, still have features in common such as a deep hydrophobic cavity accommodating the methylene part of the FA and polar residues located on the protein surface interacting 7357

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362

Article

Journal of the American Chemical Society

Figure 3. Binding of D1-decorated DNA nanocubes to HSA. (A) EMSA showed little or no detectable complexes between undecorated cube, cube +1, or cube + 2 with HSA, but significant complex formation was detected with cube + 4top and cube + 8. (B) Surface plasmon resonance screen revelated fast association and very slow dissociation of the cage, and binding behaviors similar to those of EMSA. The pink line corresponds to the signal of undecorated cube (not binding).

to tune the interaction of DNA cages with HSA and create novel protein−DNA nanostructures for drug delivery. We hypothesized that the high affinity of D1 to HSA would help overcome the electrostatic repulsion of the DNA cage to the negatively charged protein. We decided to use the cubic scaffold previously designed in our laboratory.23 The cubic scaffold is composed of four 96-mer clips, self-assembling quantitatively in one-pot. It has the unique property that it displays eight single-stranded regions, four on the top, four on the bottom, that can be further decorated with complementary DNA sequences, as shown in Figure 2. One can therefore design different sequences and modulate the number and orientation of the D-DNA chains on the cube, positioning up to eight D1 molecules on the different faces of the cube. Structures displaying different geometries were prepared: cubes with none (cube), one (cube + 1), two (cube + 2), and four D1s on the same side (cube + 4top), a total of four D1s on opposite sides (cube + 4opp), and finally four D1s on each side, making a total of eight D1s per cube (cube + 8). (Figure 2) Once again, EMSA assays were first used to test for binding of the DNA cubes to HSA. The undecorated cube and the cube decorated with one or two D1s do not bind to the protein (Figure 3 and Supplementary Figure 12). However, when four D1s are positioned on one cube face, we observe a clear shift and the formation of a well-defined DNA cube−HSA complex (Figure 3). The formation of a clear band suggested the formation a 1:1 complex. The smeared band below is possibly due to the dissociation of the complex and intrinsically due to the nature of

the observed binding responses (i.e., signal shift scaled in resonance units, RU) correlated well with the theoretical Rmax values predicted (see Supporting Information section IV-d) assuming a 1:1 binding stoichiometry. Unlike other methods to modify albumin covalently, our specific and non-covalent binding between D-DNA and HSA is complete within minutes at room temperature. Moreover, previous hydrophobic modifications (e.g., ibuprofen,15 cholesterol, tocopherol,30 or phosphorothioate backbone31 introduced in DNA) have yielded low affinity (micromolar) binding to HSA, whereas our current SPR data are consistent with previous literature describing high affinity (nanomolar) binding of fatty acids to HSA (107−108 M−1 association constants).32 The dendritic alkyl structure may help to overcome the electrostatic repulsion between the DNA and the protein by providing a multivalent ligand to interact with the multiple sites on HSA. Binding of D-DNA-Decorated DNA Cages to Human Serum Albumin. Using DNA as a building block to construct highly programmable nanostructures offers opportunities to develop novel drug delivery systems.33 Our group and others have successfully built DNA cages34−39 and shown that these biocompatible constructs are more stable in serum and resistant to degradation than their single-stranded components.40 These cages can be designed to open in response to stimuli,41 for example releasing therapeutics such as small molecules23 or siRNAs42 in response to a DNA sequence. However, limitations concerning cellular uptake and biodistribution still need to be addressed.43 We thus investigated if our D1 ligand could be used 7358

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362

Article

Journal of the American Chemical Society

with the HSA binding pocket as compared to when all four D1 ligands are oriented in the same direction. Also of special note, repeating our SPR experiments using low, medium, and high HSA surface densities (i.e., 900−4500 RU) and multiple buffer systems (i.e., 1X TAMg pH 8.0 to match cage assembly and EMSA data, as well as physiologically relevant HEPES-buffered saline at pH 7.4) yielded similar outcomes (Supplementary Tables III and IV). Lastly, we performed competitive SPR assays where sensorbound cubes (i.e., to immobilized HSA during the association phase) were competed with solution-phase HSA during the dissociation phase (Supplementary Figure 18). We found that the addition of free HSA during the dissociation phase was most effective at displacing cube + 2 from immobilized HSA surfaces faster compared to buffer-only dissociation. As the number of D1 ligands increased, however, the solution-phase HSA was less effective at displacing cube + 4 and was unable to displace cube + 8. These competitions are consistent with our earlier SPR affinity assays where increases in decoration correlated with decreases in the dissociation rate constants (i.e., more stable complex formation, higher affinity overall). Even in the presence of excess competitor, the very slow dissociation kinetics for cube + 8 may indicate that most or all of its alkyl chains are pointing toward the protein (i.e., stronger binding affinity compared to the cube + 4 with four D1s on one side). We have previously shown that the chains on cube + 8 were flexible enough to meet inside and form a hydrophobic core inside the DNA cage.23 We suspect that, in the presence of albumin, these chains will orient outside the cube and strongly bind to the protein in a multivalent fashion. Increased Serum Stability upon Binding to HSA. One of the major challenges currently faced by oligonucleotides therapeutics is their high susceptibility to degradation by nucleases in vivo. To overcome this problem, nucleosides or backbone modifications have traditionally been introduced.44 Since these solutions remain synthetically complex, we decided to examine the resistance of our DNA-HSA constructs toward nucleases. Based upon the method previously developed by Conway et al.40 we incubated our complexes with Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) at 37 °C. Aliquots were collected at different time points and separated by PAGE, and the resultant band intensities were quantified to predict the half-life of our constructs. Overall, the D1−HSA complex was 3 times more stable compared to unmodified DNA (Supplementary Figure 20), suggesting that the protein protects the strand from enzymatic degradation. In a similar manner, the half-life of D1 bound to the cage was then investigated. After incubation, we denatured the DNA cube structures in the collected aliquots by adding formamide.45 However, the D1−HSA interaction remains unchanged under denaturing conditions, thus permitting quantification of both band intensities from the DNA cube clips and from the bound amphiphile D1 to HSA (Figure 4). The stability measured corresponds to the remaining full-length cube strands or to D1 attached to HSA after degradation at a chosen time point. D1 ligands participating in the binding are now protected by both the DNA cage and the binding to HSA, providing a double steric protection. Our results show that under these conditions, D1 serum stability increases from minutes to hours. Indeed, when four D1s are positioned on the top of the cage, the half-life of HSA-bound D1 was 10 h. When eight D1s were positioned (both faces of the cube decorated), the half-life of HSA-bound D1 halflife reached 22 h. D-DNA half-life thus potentially allows the

the electrophoretic assay. The upper band appearing at high HSA concentration is most likely caused by the formation of HSA aggregates (Supplementary Figure 3). We used dynamic light scattering (DLS) experiments to further confirm the 1:1 binding to HSA. The D1-decorated cube gave a peak of low polydispersity and an increase in size when bound to protein. The measured radius is consistent with the expected radius of a 1:1 complex (Supplementary Figure 19). Interestingly, the cube with two D1s on the top face and two on the bottom displays affinities between the two D1s and the four D1s on the top. This is shown by the appearance of a smeared band that reflects dissociation and hence lower binding affinity (Supplementary Figure 14). We also performed a supershift assay26 with an anti-HSA antibody to confirm the identity of the complex. The detection of a lower mobility band indicates the binding of the antibody to the DNA cube−protein complex. (Supplementary Figure 15). To cross-validate our EMSA and supershift data, complementary SPR assays were also performed. Rapid, fixedconcentration screening showed that the undecorated cube did not interact with HSA, cube + 1 and cube + 2 generated specific signal shifts, and the largest shifts were observed when four or eight D1s were positioned (Figure 3). Subsequent dosedependent titrations (see Supplementary Figure 16 for singlecycle kinetics, Supplementary Figure 17 for multi-cycle kinetics) revealed a distinct trend in the cube geometries: in general, the dissociation rate decreased and the overall affinity became stronger as the degree of decoration increased (Table 1). Notably, the cube with four D1s on one face (cube + 4top) or the cube with eight D1s exhibited the strongest equilibrium dissociation constants (Kd values of 8 ± 2 and 5 ± 2 nM). Consistent with our EMSA data, however, cube + 4opp exhibited weaker overall affinity. We suspect that when D1 ligands are oriented in opposite directions, this weakens the cage’s interface Table 1. Surface Plasmon Resonance Results for Different Cube Structuresa

a

Surface plasmon resonance revealed fast association and very slow dissociation of the cage. Dose-dependent titrations in HBS buffer allowed us to calculate the binding affinities of the different structures. The affinity can be tuned by changing number and position of the ligand. Apparent equilibrium dissociation constants (Kd) were determined by SPR as described in the Supporting Information. The values shown (± standard deviation) represent the average of duplicate injections acquired from the three surface densities. 7359

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362

Article

Journal of the American Chemical Society

Figure 4. Serum stability experiments. HSA-bound D1 (alone or decorated cube) was incubated with FBS and collected at different time points. (A) Denaturing PAGE allows quantification of half-life of D1 bound to HSA. We can distinguish both bands of the cube clips and the HSA-bound D1. (B) Normalized band intensity was plotted to estimate half-life of the constructs.

Figure 5. Silencing experiments. (A) Cells were transfected with 100 nM of DNA structures and incubated 24 h before Luciferase assays. The antisense sequences with and without modifications have a strong silencing effect. However, the controls showed no silencing. (B) Cells were transfected with 100 nM of DNA structures and incubated 24 h before Luciferase assays. Control in the q-RT-PCR experiment is cells treated with Lipofectamine 3000 without DNA. The modified sequences keep a strong silencing activity.

molecule sufficient time for circulation and distribution in vivo. Once again, the increases in half-life appear to correlate with the increase in decoration (i.e., stronger binding affinities). Silencing Potency of the Complexes. Next, we wanted to examine if the desirable protective effect of HSA (i.e., increased serum stability) would inadvertently interfere with the biological activity of our amphiphilic ligand. We synthesized a new conjugate, D-Luciferase, with an antisense oligonucleotide targeting Luciferase Firefly mRNA. The antisense sequence contains a phosphorothioate backbone, and the sequence is given in Supplementary Table V. Using Lipofectamine 3000, our DNA−HSA complex was transfected in HeLa cells expressing Luciferase to assay its biological activity. We observed similar silencing activity of the antisense and the DDNA-antisense, confirming that binding to HSA does not inhibit the silencing effect of the bound DNA (Figure 5). The negative controls are the same sequences but without a phosphorothioate backbone (Supplementary Table V). We also designed a D-siRNA ligand for silencing of apolipoprotein B. We incubated the transfected strand for 24 h with HepG2 cells. Total RNA was collected and analyzed by quantitative PCR. We observed 70% greater silencing with the D-siRNA in comparison to the untreated

control (Figure 5). The silencing effect was somewhat reduced compared to siRNA with no modification (Figure 5). These experiments confirmed that conjugated nanostructures do not prevent the DNA from being recognized by the silencing machinery inside the cell.



CONCLUSION We report the synthesis of a new dendritic, amphiphilic ligand that can be used to complex DNA to HSA with high affinity (low nanomolar range). The non-covalent association happens within minutes under physiological conditions. HSA binding improves the stability of the DNA in serum without affecting its silencing activity). We have also shown that the affinity between DNA cubes and HSA can be tuned by modulating the number and positions of ligands on the cage. The D-DNA structure is hydrophobic enough to show strong binding to albumin, while having very low propensity for aggregation. Non-covalent binding of a DNA nanostructure to HSA is attractive, because it can protect the nanostructure in circulation, but the protein can be shed as the nanostructure penetrates constrained environments, such as many tumors. DNA cages offer the opportunity to position the D1 ligands into discontinuous, spatially well-defined 7360

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362

Article

Journal of the American Chemical Society

scholarship. The McGill SPR-MS Facility thanks the Canada Foundation for Innovation (CFI) for infrastructure support.

patterns to systematically examine and optimize multivalent binding. The ability to use ligand patterning and multivalency is an important step toward the use of DNA nanostructures as drug delivery systems. For example, HSA could serve as the carrier to help DNA cages accumulate within tumors in cancer patients, and the cage itself would deliver therapeutics. We envision that the binding affinity could be tuned to control release rates of oligonucleotides in serum. In addition, while four ligands on one cube face were enough to bind HSA with high affinity, the four single-stranded binding sites remaining on the other face could be decorated with targeting ligands such as folic acid, aptamers, or antibodies. The cage can also encapsulate drugs by intercalation (for example, the well-characterized doxorubicin46) incorporate sequences that possess inherent therapeutic effect (antisense or siRNAs) or contain a hydrophobic core for the encapsulation of small molecules.23 It is well established that nanoparticles, whether inorganic, liposomal, polymer, or nucleic acid based, can interact with serum proteins.20 The rapid formation of such a protein corona can be detrimental to the pharmacokinetics and physiological effects of nanoparticles by altering their size, surface charge, or functionality of the targeting ligands attached and by tagging them for uptake by the mononuclear phagocyte system.17 Since our current findings demonstrate that we can tune and control the interaction of a DNA nanostructure with highly abundant HSA, and since HSA can reduce phagocytosis,20 this may present a strategy to control the protein corona as well as to understand the role of protein coating on biodistribution and cellular uptake. One could envision decorating the cube with other ligands to target other serum proteins and better understand the role of each. Further in vitro and in vivo investigations are being performed to better understand the impact of serum protein coating on the biodistribution and cellular uptake of DNA nanostructures into cancer tissues.





(1) Sleep, D.; Cameron, J.; Evans, L. R. Biochim. Biophys. Acta, Gen. Subj. 2013, 1830 (12), 5526−5534. (2) Fanali, G.; di Masi, A.; Trezza, V.; Marino, M.; Fasano, M.; Ascenzi, P. Mol. Aspects Med. 2012, 33, 209−90. (3) Kratz, F. J. Controlled Release 2008, 132, 171−83. (4) Iyer, A. K.; Khaled, G.; Fang, J.; Maeda, H. Drug Discovery Today 2006, 11, 812−8. (5) Kratz, F.; Beyer, U. Drug Delivery 1998, 5, 281−99. (6) Mirkin, C. A.; Meade, T. J.; Petrosko, S. H.; Stegh, A. H. Nanotechnology-Based Precision Tools for the Detection and Treatment of Cancer; Springer: Berlin, 2015; Vol. 166. (7) Gradishar, W. J.; Tjulandin, S.; Davidson, N.; Shaw, H.; Desai, N.; Bhar, P.; Hawkins, M.; O’Shaughnessy, J. J. Clin. Oncol. 2005, 23, 7794− 803. (8) Elsadek, B.; Kratz, F. J. Controlled Release 2012, 157, 4−28. (9) Zheng, Y. R.; Suntharalingam, K.; Johnstone, T. C.; Yoo, H.; Lin, W.; Brooks, J. G.; Lippard, S. J. J. Am. Chem. Soc. 2014, 136, 8790−8. (10) Malik, C. T.; Wei, S.; Magnus, B.; Phillip, S.; Asrar, B. J. Biol. Chem. 1997, 272, 25968−75. (11) Merlot, A. M.; Kalinowski, D. S.; Richardson, D. R. Front. Physiol. 2014, 5, 299. (12) Gavrilov, K.; Saltzman, W. Yale J. Biol. Med. 2012, 85, 187−200. (13) Surana, S.; Shenoy, A. R.; Krishnan, Y. Nat. Nanotechnol. 2015, 10, 741−7. (14) Kanasty, R.; Dorkin, J. R.; Vegas, A.; Anderson, D. Nat. Mater. 2013, 12, 967−977. (15) Manoharan, M.; Inamati, G. B.; Lesnik, E. A.; Sioufi, N. B.; Freier, S. M. ChemBioChem 2002, 3, 1257−60. (16) Lau, S.; Graham, B.; Cao, N.; Boyd, B. J.; Pouton, C. W.; White, P. J. Mol. Pharmaceutics 2012, 9, 71−80. (17) Walkey, C.; Chan, W. C. Chem. Soc. Rev. 2012, 41, 2780−2799. (18) Monopoli, M. P.; Åberg, C.; Salvati, A.; Dawson, K. A. Nat. Nanotechnol. 2012, 7, 779−786. (19) Kumar, C. S. S. R. Nanomaterials for Medical Diagnosis and Therapy; Wiley-VCH: Weinheim, 2007. (20) Schöttler, S.; Landfester, K.; Mailänder, V. Angew. Chem., Int. Ed. 2016, 55, 8806−8815. (21) Zanganeh, S.; Spitler, R.; Erfanzadeh, M.; Alkilany, A. M.; Mahmoudi, M. Int. J. Biochem. Cell Biol. 2016, 75, 143−7. (22) Chen, Y.-J.; Groves, B.; Muscat, R. A.; Seelig, G. Nat. Nanotechnol. 2015, 10, 748−760. (23) Edwardson, T. G.; Carneiro, K. M.; McLaughlin, C. K.; Serpell, C. J.; Sleiman, H. F. Nat. Chem. 2013, 5, 868−75. (24) Edwardson, T. G. W.; Carneiro, K. M. M.; Serpell, C. J.; Sleiman, H. F. Angew. Chem. 2014, 126, 4655−4659. (25) de Rochambeau, D.; Barłóg, M.; Edwardson, T. G. W.; Fakhoury, J.; Stein, R.; Bazzi, H. S.; Sleiman, H. F. Polym. Chem. 2016, 7, 4998− 5003. (26) Hellman, L. M.; Fried, M. G. Nat. Protoc. 2007, 2, 1849−61. (27) Fasano, M.; Curry, S.; Terreno, E.; Fanali, G.; Narciso, P.; Notari, S.; Ascenzi, P. IUBMB Life 2005, 57, 787−796. (28) Curry, S.; Mandelkow, H.; Brick, P.; Franks, N. Nat. Struct. Biol. 1998, 5, 827−835. (29) Patching, S. Biochim. Biophys. Acta, Biomembr. 2014, 1838, 43−55. (30) Onishi, R.; Watanabe, A.; Nakajima, M.; Sekiguchi, M.; Kugimiya, A.; Kinouchi, H.; Nihashi, Y.; Kamimori, H. Anal. Sci. 2015, 31, 1255− 60. (31) Moreno, P. M.; Pêgo, A. P. Front. Chem. 2014, 2 (87), 1−7. (32) van der Vusse, G. J. Drug Metab. Pharmacokinet. 2009, 24, 300−7. (33) Seeman, N. C. Annu. Rev. Biochem. 2010, 79, 65−87. (34) McLaughlin, C. K.; Hamblin, G. D.; Aldaye, F. A.; Yang, H.; Sleiman, H. F. Chem. Commun. 2011, 47, 8925−7. (35) Fakhoury, J. J.; McLaughlin, C. K.; Edwardson, T. W.; Conway, J. W.; Sleiman, H. F. Biomacromolecules 2014, 15, 276−82.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b02917. DNA and D-DNA structures and synthesis, evaluation of binding to HSA by EMSA and SPR, competition experiments with serum components, DNA cube sequences, synthesis and preparation, evaluation of binding of DNA cubes to HSA, DLS experiments, serum stability, and silencing studies, including Supplementary Figures 1−25 and Tables I−V (PDF)



REFERENCES

AUTHOR INFORMATION

Corresponding Author

*[email protected] ORCID

Hanadi F. Sleiman: 0000-0002-5100-0532 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to thank NSERC, CIHR, CFI, the Canada Research Chairs Program, FQRNT, and CSACS for funding. A.L. thanks the Canadian Institutes of Health Research (CIHR) for a Drug Development Training Program (DDTP) 7361

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362

Article

Journal of the American Chemical Society (36) Walsh, A. S.; Yin, H.; Erben, C. M.; Wood, M. J. A.; Turberfield, A. J. ACS Nano 2011, 5, 5427−5432. (37) Goodman, R. P.; Berry, R. M.; Turberfield, A. Chem. Commun. 2004, 1372−1373. (38) Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Subramani, R.; Mamdouh, W.; Golas, M. M.; Sander, B.; Stark, H.; Oliveira, C. L.; Pedersen, J. S.; Birkedal, V.; Besenbacher, F.; Gothelf, K. V.; Kjems, J. Nature 2009, 459, 73−6. (39) Chen, J. H.; Seeman, N. C. Nature 1991, 350, 631−3. (40) Conway, J. W.; McLaughlin, C. K.; Castor, K. J.; Sleiman, H. Chem. Commun. 2013, 49, 1172−4. (41) Bujold, K. E.; Fakhoury, J.; Edwardson, T. G. W.; Carneiro, K. M. M.; Briard, J. N.; Godin, A. G.; Amrein, L.; Hamblin, G. D.; Panasci, L. C.; Wiseman, P. W.; Sleiman, H. F. Chem. Sci. 2014, 5, 2449−2455. (42) Bujold, K. E.; Hsu, J. C. C.; Sleiman, H. F. J. Am. Chem. Soc. 2016, 138, 14030−14038. (43) Pinheiro, A. V.; Han, D.; Shih, W. M.; Yan, H. Nat. Nanotechnol. 2011, 6, 763−772. (44) Deleavey, G. F.; Damha, M. J. Chem. Biol. 2012, 19, 937−954. (45) Marmur, J.; Ts’o, P. O. P. Biochim. Biophys. Acta 1961, 51, 32−36. (46) Kim, K. R.; Kim, D. R.; Lee, T.; Yhee, J. Y.; Kim, B. S.; Kwon, I. C.; Ahn, D. R. Chem. Commun. 2013, 49, 2010−2.

7362

DOI: 10.1021/jacs.7b02917 J. Am. Chem. Soc. 2017, 139, 7355−7362