DNA nanotechnology excels at the bottom-up fabrication of engineered nanostructures. DNA duplexes can be manipulated into user-defined shapes by exploiting the base-pairing rules for duplex formation [1
]. Discrete nanostructures can be assembled in two and three dimensions with sub-nanometer control using dedicated design software [4
]. Chemical diversity and functionality can be incorporated into structures site-specifically using, for example, solid phase DNA synthesis [6
], or non-specifically via intercalation [8
], or electrostatic interactions [10
]. This rapidly evolving field has transformed materials science with wide-ranging applications, including the generation of DNA origami devices for optical sensing [12
], controlled single molecule synthesis using a lab-on-a-chip DNA board [13
], computation devices [14
], and finite sub-nm movement of DNA-based robots using DNA ligands [16
DNA nanotechnology applied to the life sciences is gaining traction. DNA nanostructures can help control processes within cells, or at the membrane interface to advance biological understanding [17
]. This progress includes the generation of novel diagnostic tools [21
], the enhancement of existing drugs [24
], and devices with novel therapeutic action [25
]. Recently, intracellular DNA-based delivery vehicles have been used to transport biomolecules. Engineered DNA cages that encapsulate small molecule drugs [25
], mRNA [26
], peptides, and proteins [27
] have been developed to deliver biomolecular cargo. DNA nanostructures can be internalized in specific mammalian cells, although the nature of the design appears to play an important role [29
]. Coating the nanostructures in certain chemical groups can improve cellular uptake [10
To fulfill desirable biomolecular functions, the DNA nanostructures have to be stable. DNA origami-based nanostructures have been studied previously in vitro and in vivo [33
]. Generally, the origami constructs withstand diverse biology conditions under short time durations. Yan and colleagues have recently shown intact and functional DNA origami in the renal system of a mouse model [34
]. However, other reports have identified significant degradation and unfolding of DNA origami structures in biological media [35
]. This instability has been attributed to the low level of Mg2+
ions—essential to stabilize DNA origami nanostructures—and digestion from enzymes including DNAses. The susceptibility to degradation appears to be design-specific, with tubular designs proving more resilient [37
]. Other strategies can be employed to help stabilize the nanostructures, including chemical ligation of DNA nicks [38
], the introduction of non-native base pairs, such as LNA, PNA, and XNA [39
]. Alternatively, cationic peptides [10
], polymers [32
], or intercalators [9
] can be used to improve structural stability.
DNA-based nanopores are the most recent class of membrane channels which can potentially offer a unique degree of control at the membrane interface [40
]. Naturally occurring nanopores are usually composed of proteins or peptides to help regulate ion transport across cell membranes [45
]. However, it can be challenging to de novo design amino acid-based nanopores due to unexpected protein misfolding [46
]. In contrast, utilizing DNA as a construction material can help overcome this issue. To date, DNA nanotechnology has produced nanopores with highly customizable properties, including channel diameter, length, functionalized groups within the lumen, and ligand-controlled pore opening [40
]. For future biological applications, including pore-mediated drug delivery, nanopore stability and solubility within biological media must be maintained.
To investigate DNA nanopore stability, this study employs the DNA nanopore NP-3C (Figure 1
]. The pore is assembled from six single strands (Tables S1 and S2, Figures S1 and S2
), which form six interwoven DNA duplexes to generate a six helical barrel. Three cholesterol lipid anchors are site-specifically incorporated to the exterior of the bundle to facilitate membrane binding and nanopore behavior. The assembled pore punctures the membrane to generate a toroidal pore to enable ion transport across the lipid bilayer (Figure 1
]. However, the hydrophobic lipid anchors can also mediate other undesired behavior, including intermolecular oligomerization (Figure 1
]. To help distinguish the lipid anchor effect, a cholesterol-free version, NP-0C, was assembled to serve as a negative control (Tables S1 and S2, Figure S1
Cell media is composed of complex ions and nutrients which help to maintain cell homeostasis and phenotype. For biological applications, the amphiphilic DNA nanostructures must remain structurally stable within the used medium. Therefore, this report assayed biologically compatible media to identify the pore’s structural stability and membrane activity, including phosphate-buffered saline (PBS), bacterial growth medium lysogeny broth (LB), mammalian cell media Dulbecco’s modified Eagle medium (DMEM), and DMEM supplemented with 10% v
fetal bovine serum (FBS) (Table 1
]. Serum is required for specific cell types to maintain cell function, and is composed of a wide array of entities including proteins, hormones, and electrolytes. The total protein concentration in FBS is ~0.3–0.5 mg/mL [53
]. Albumin, globulin, and fibrinogen make up the majority of proteins found in serum, at approximately 55, 38, and 7%, respectively. In addition, over a thousand other regulatory proteins exist at much smaller levels. Metal ions must also be considered. Positively charged metal ions coordinate with DNA ionically to stabilize duplexes. Therefore, a range of metal cations was assayed to identify the counterion stabilization on the nanostructures [33
]. We tested monovalent sodium and potassium ions typically used for single channel current recordings used to study nanopores [54
], and divalent magnesium ions, conventionally used for the stabilization of DNA origami constructs. The nanopore formation was determined using gel electrophoresis. To identify the thermal stability at physiological temperatures in biological media, the melting temperatures of the constructs were identified using fluorescence spectroscopy [55
]. Further, our study identified the aggregation extent of the nanostructures using fluorescence spectroscopy and confocal microscopy over time. Finally, to confirm membrane activity of the nanopore, binding to model membranes was determined using fluorescence microscopy. With the knowledge gained using our approach, new pore formulations and folding protocols can be established which should help provide insights for future applications across the life sciences.
2. Materials and Methods
All reagents were purchased from Sigma Aldrich (UK) unless stated otherwise. The DNA nanopore was published previously (information on the sequences, including 2D maps and dimensions is provided in the Supporting Information
]. The DNA nanopores were assembled by mixing an equimolar mixture of the component DNA strands (0.5 µM, unless stated otherwise) (Integrated DNA Technologies, Coralville, IA, USA) containing the stated buffer or media. The nanopores were folded by heating the solution from 95 °C for 2 min, and cooling to 20 °C at a rate of 5 °C per min. The folded DNA nanopore constructs were stored at room temperature, and vortexed for 2 s before use. Where stated, n
-octyl-oligo-oxyethylene (OPOE) (Enzo Life Sciences, Exeter, UK) was added to the folding mixture prior to nanopore assembly (1.5% v
Buffer and reagents. Na: NaCl 300 mmol/L, tris 15 mmol/L, pH 8.0. K: KCl 300 mmol/L, tris 15 mmol/L, pH 8.0. Mg: MgCl2
14 mmol/L, tris 40 mmol/L, acetic acid 20 mmol/L, ethylenediaminetetraacetic 1 mmol/L, pH 8.3. PBS: NaCl 137 mmol/L, KCl 2.7 mmol/L, Na2
8 mmol/L, KH2
2 mmol/L, pH 7.4. LB: tryptone 10 mg/mL; yeast extract 5 mg/mL; NaCl 10 mg/mL values taken from [56
]. D components include CaCl2
2.4 mmol/L, MgSO4
0.8 mmol/L, KCl 5.4 mmol/L, NaHCO3
44.0 mmol/L, NaCl 109.5 mmol/L, NaH2
0.9 mmol/L. Neat fetal bovine serum components include bilirubin 2.4 mg/L; Cholesterol 340 mg/L; creatinine 27.3 mg/L; urea 260 mg/L; Na+
142 mmol/L; Cl−
155.5 mmol/L; K+
8 mmol/L; Ca2+
3 mmol/L; Mg2+
1.1 mmol/L; PO43−
2.3 mmol/L; Fe 1.6 mg/L; glucose 550 mg/L; protein 36 g/L; albumin 17 g/L; α-globulin 17 g/L; β-globulin 2 g/L; γ-globulin 1 g/L, values taken from [57
DNA nanopore folding was characterized using 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Bio-Rad, Watford, UK) with standardized buffers typically applied to proteins. The gel was thermally equilibrated at 8 °C prior to loading. The gel was run at 140 V for 70 min. The bands were visualized by first removing SDS with deionized water, then stained using ethidium bromide solution. A 100 base-pair DNA marker (New England Biolabs, Hitchin, UK) was used as a reference.
The Förster resonance energy transfer (FRET) characteristics of the fluorescein (FAM) and Cy3 labeled nanopore constructs were identified using a Varian Eclipse fluorescence spectrophotometer (Agilent, Stockport, UK). 20 µL of the various DNA nanostructures (folded in PBS at 1 µM) (see Tables S1 and S2
for strand information) was added to PBS (180 µL) in a quartz cuvette with a path length of 1 cm. The samples were analyzed by excitation at 495 nm, and the emission monitored between 505–700 nm. A 5 nm slit width and 600 PMT voltage was applied, along with a scanning rate of 600 nm per min; the scan was performed 3 times and averaged.
The melting transitions of the DNA nanostructures were identified using a MyIQ real-time PCR (Bio-Rad, Watford, UK). The nanostructures were assembled containing FAM and Cy3 FRET pairs (folded at 1 µM in PBS). The DNA constructs were diluted into the stated buffer systems to give a final DNA concentration of 0.1 µM (total volume of 25 µL) in a 96-well thin wall fluorescence plate (Bio-Rad, Watford, UK). Optical quality sealing tape (Bio-Rad, Watford, UK) was placed on top to prevent evaporation. The sample was heated from 30–85 °C at a rate of 0.5 °C per min. The melting temperature was determined from taking the derivative of the donor fluorescence profile. Errors were identified from 3 independent experiments.
Fluorescence spectroscopic analysis was performed on Cy3-modified DNA nanostructures using a Varian Eclipse fluorescence spectrophotometer (Agilent, Stockport, UK) with a fluorescence cuvette. The samples were analyzed by excitation at 540 nm, and by monitoring the emission from 550–600 nm, using a 10 nm slit width, 800 PMT voltage, scanning at 600 nm per min and taking the average of 3 repeat scans. The DNA nanostructures (2 µL, folded at 0.5 µM) in the stated buffers were scanned once the folding temperature reached 40 °C by diluting in the buffer systems (200 µL final volume). At the designated time points, the samples were centrifuged for 10 min at 16k revolutions per min at room temperature (Eppendorf, Stevenage, UK), and the supernatant was carefully extracted and the fluorescence monitored using the same dilution and settings as described.
Confocal laser scanning microscopy (CLSM) images were collected using a 60× oil objective FV-1000 Olympus microscope. Images were analyzed using ImageJ software. To image the DNA nanopore constructs, the folded pore containing a Cy3 dye (10 µL, 0.5 µM in PBS) was deposited on a fluorodish (World Precision Instruments, Sarasota, FL, USA), and left to settle for 20 min prior to imaging. For the vesicle-binding assays, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) giant unilamellar vesicles (GUVs) were prepared by modifying a published protocol [48
]. POPC (150 µL, 10 mM) in chloroform was added to a 1 mL glass vial, the solvent was removed under vacuum, and underwent rotation using a rotary evaporator. The thin film generated was resuspended in mineral oil (150 µL) by vortexing and sonicating for 10 min. Green fluorescent protein (5 µL,10 µM in PBS) was mixed with sucrose solution (20 µL, 400 mM), followed by addition of mineral oil (150 µL). The suspension was vortexed and sonicated for 10 min at room temperature, then carefully added to the top of a glucose solution (1 mL, 400 mM) in a plastic vial (1 mL). The vesicles were generated by centrifuging at 16K RPM at 4 °C for 10 min. The mineral oil top layer and the majority of the sucrose layer (~800 µL) were carefully removed. The remaining solution containing the pelleted vesicles was gently mixed with a pipettor, then transferred to a clean plastic vial. The POPC GUVs (5 µL) were added to a KCl solution (5 µL, 0.5 M), and then Cy3-labeled NP-3C (2 µL folded at 0.25 µM) in the stated buffers was mixed, and the solution deposited on the confocal slide and used within 24 h. For the serum time series assay, NP-3C (10 µL, 0.25 µM folded in PBS) was added to FBS (10 µL) for the stated time durations. The NP-3C-FBS solution (4 µL) was added to the GUV solution as described above. All images were collected after 20 min using identical settings.
DNA-based nanopores are a recent and exciting class of synthetic membrane channel. This construction approach provides a unique level of biophysical control across lipid bilayers. For DNA nanopores to provide functionality in the life sciences, for example, as biosensors, drug delivery vehicles, or cytotoxic-inducing agents, biocompatibility needs to be addressed. Essential parameters before in vitro and in vivo testing include confirming structural stability and solution-phase solubility in biologically compatible media. We have employed an archetypical DNA nanopore and tested its folding capabilities in diverse biological media routinely used to culture mammalian or bacterial cells. The conditions were deliberately chosen to be stringent as folding was performed in the cell media. It would also have been possible to fold pores in protein-free buffers and then add them to the protein-containing media. This differential treatment would have, however, made a fair comparison across all buffers and media more difficult. Our results suggest that pores folded efficiently and remained structurally stable in all media assayed. Importantly, the pore constructs displayed melting temperatures above physiological temperatures, even in the absence of divalent magnesium ions. With the exception of FBS-containing media, the DNA nanopores remained predominately water soluble in all conditions tested over 48 h. These results reinforce the suitability of DNA nanotechnology as a good building material for use in biological studies.
Serum proteins cause aggregation most likely due to complexation towards the DNA’s negatively charged phosphate backbone. Albumin serum proteins have been shown to complex to antisense oligonucleotides, which in some instances enhances the half-life of intravenously injected DNA [61
]. However, hydrophobic groups can further increase the protein complexation extent [63
]. In the case of amphiphilic DNA nanopores used in this study, the resultant complexes caused significant aggregation, drastically reducing the amount of membrane tethering action. This result is an important finding in the field of lipidated DNA nanostructures, and if the use of FBS is unavoidable, the DNA nanopores should be applied for short time periods of less than 1 h, or the cells transferred temporarily into other media, such as phosphate buffered saline. Alternatively, vesicle delivery agents may be employed to shield the nanostructures from aggregation. Other strategies include using DNA masking groups, such as coating the structures in polyethylene glycol, or carboxylic acids, both are known to improve the circulation time of biomolecules in the bloodstream. Cationic groups such as lysine and arginine-rich peptides [32
], polyamines [64
], and metal cations can be used to block the serum proteins from binding. In addition, charge-neutral peptide nucleic acids nanopore equivalents [39
] can be developed to prevent serum complexation. We expect lipidated DNA nanodevices and DNA nanopores to provide a useful foothold cell biology, and should find use as drug delivery gate-keepers, to function as drug molecules, such as immunosuppressants or immunoactivators, or act as novel tools in diagnostics and sensing.