Design and Assembly of a Cargo-Agnostic Hollow Two-Lidded DNA Origami Box

Abstract

DNA origami, a method of folding DNA into precise nanostructures, has emerged as a powerful tool for the design of complex nanoscale shapes. It has great potential as a technology to encapsulate and release cargos spanning small molecules through large proteins, while remaining stable in a variety of ex vivo processing conditions and in vivo environments. While DNA origami has been utilized for drug delivery applications, the vast majority of these structures have been flexible, flat 2D or solid 3D nanostructures. There is a crucial need for a hollow and completely enclosed design capable of holding and eventually releasing a variety of cargos. In this paper, we present the design and assembly of a hollow DNA origami box with two lids. We characterize the isothermal conditions for structural assembly within minutes. We demonstrate that passive loading of small molecules is charge dependent. We also outline an approach to design staple extensions pointing into the cavity or outside of the hollow DNA origami, allowing for the active loading of protein or the potential for decoration with passivating or targeting molecules.

1. Introduction

The need for nanoparticle cargo delivery systems is growing. Several key features are important to consider when designing these systems. They should be able to encapsulate and retain cargo like drugs for timed or triggered release. They should be able to target particular parts of the body and release drugs gradually over time. Ideally, this controlled release would be in response to environmental cues, potentially performing logic operations on multiple environmental signals. Other desirable features include ease of assembly, stability in biological environments and tunable immunogenicity. − Several types of nanoparticles have been developed from diverse materials, such as lipids, polymers, or inorganic molecules. While useful in numerous applications, these systems lack the ability to finely tune the nanoparticle structure. DNA origami presents a nanoparticle system with a tremendous ability to finely tune shape and generate mechanically sensitive or movable components. DNA nanostructures are being utilized across several fields from drug delivery to sensors to mechanical devices. −

DNA origami is a method of folding DNA into specific shapes. The single-stranded DNA (ssDNA) sequence called the “scaffold” provides a backbone for the designs, while shorter ssDNA strands called “staples” hold the designs in place. DNA origami has tremendous potential for drug delivery due to its ability to form a variety of mechanically movable shapes, its stability, nonimmunogenic properties and ease in chemically functionalizing the structure for various purposes. ,,− For DNA origami drug delivery constructs to be a viable option in medicine, they need to be designed to encapsulate and confine a variety of cargoes that can be released in response to environmental cues. Additionally, the assembly and purification methods need to be optimized for ease of production and separation from excess staple strands. This has proven challenging as most protocols for DNA origami assembly have used changes in temperature over long time periods during assembly, which limits production to small volumes assembled in a thermocycler. − There are some examples of isothermal assembly, but incubation times are long. , DNA origami have been shown to be reasonably stable, but this seems to depend highly on the design of the structure, , necessitating an examination of the stability for each design.

The vast majority of DNA origami structures are flexible, flat 2D structures of different geometries. These have been used to deliver or release drugs or model molecules including doxorubicin, platinum-based molecules and antibody fragments to name a few. ,,,− Molecules have been attached to the surface of the DNA origami structure through electrostatic interactions, as well as through covalent linkages. For instance, doxorubicin and similar molecules have been bound electrostatically to the surface of a 2D DNA surface. ,, Folic acid and other molecules have been covalently linked to similar 2D DNA origami shapes. 2D DNA origami structures are easy to design, assemble, and functionalize compared to 3D structures, but 2D structures have limitations. 2D structures do not offer an opportunity to enclose cargo. It also is challenging to encode sensitivity to environmental cues in 2D structures, whereas 3D DNA origami structures can be designed to specifically to respond to environmental signals, , potentially using that signal to release cargo.

3D DNA origami structures have potential to encapsulate cargo completely with release mechanisms for environmental cues. DNA origami box, cube and sphere designs with lids that can open in response to nucleic acid keys or light have been fabricated. − While these boxes were shown to open on demand, their loading with cargos was not demonstrated with the exception of the “enzyme vault” which was loaded with one protein. Other DNA origami designs show hollow tubes or wireframe structures that can gradually release drug when delivered, but the release is uncontrolled due to the open cage designs. ,,, Wireframes that can release enzyme in response to temperature have been designed and DNA origami tubes lacking lids that can be opened along its length based on logic nucleic acid keys (similar to what was mentioned above) have also been designed. Each of these tube designs still suffer from nonspecific delivery due to open-ended or wireframe structures. ,,, Enclosed 3D DNA origami structures have been designed with pH responsive lids. However, the loading potential of this design is minimal due to edges that are multiple helices thick, resulting in a small internal volume. There is a need for actuatable “box” designs that can be loaded with sufficient cargo and for which numerous alternative release mechanisms in response to specific environmental cues could control delivery.

Our long-term goal is to design 3D hollow DNA origami structures to entrap cargo that can later be released in response to a variety of specific environmental cues. To address this need, hollow DNA origami boxes were designed with two actuatable lids that allow loading and release of cargo ranging from small molecules to large proteins. We used Atomic Force Microscopy (AFM) and Dynamic Light Scattering (DLS) to characterize the newly designed DNA origami structures. We developed an isothermal assembly approach that dramatically diminishes the assembly time and optimizes purification methods for hollow DNA origami boxes. Stability was characterized under relevant processing and biological conditions, including various temperatures and pH values, as well as in the presence of solvent and serum, showing good stability for these structures under a variety of conditions. We demonstrated the loading of various cargos spanning small molecules to proteins. We used both passive and active loading strategies, leveraging a design approach for determining the optimal staple extension length and position that allows for protein entrapment within the DNA origami box or the decoration of protein on the outside of the DNA origami box. In summary, we outline the design, fabrication, and characterization of a hollow DNA origami box for drug delivery with two actuatable lids for potential release by two distinct environmental cues.

2. Results

2.1. Design, Assembly, and Purification

2.1.1. Design and Temperature Gradient Assembly

A DNA origami design with multiple lids can allow for multiple responses to environmental cues and enhances molecular transport into or out of the hollow DNA origami structure. Cadnano (square) was used to design the DNA origami box with two actuatable lids (Figure S1). Staple sets are available in Supporting Information (Figure S2 and Table S1). The DNA origami box consists of an uninterrupted square-shaped conduit with two hinged lids that can attach to the respective opposite edges of the conduit to close the box. DNA helices are flexible with a persistence length around 50 nm or 15 base pairs, so as to mechanically strengthen the edges in the design, the DNA staples crossed helices at each edge. This ensures all the helices are connected at the edge. The design features a 40 nm × 40 nm × 40 nm box with a 3D diagonal across the box of approximately 69 nm. This design places bounds on measurements of dimensions between 40 and 69 nm. Cadnano designs with the closed DNA origami box, as well as versions with one lid or two lids open were rendered using CanDo and UCSF ChimeraX (Figure A and S3A). The size for each design was measured by using Dynamic Light Scattering (DLS) (Figure B). Closed boxes matched our expectations for size dimensions (40–69 nm) and were smaller than either the one-lid-open or two-lids-open versions, which were larger due to the flexible nature of the lids. The polydispersity index for particles is defined as the square of the coefficient of variation and was less than 0.05 for all designs and was the smallest for closed boxes, indicating highly monodisperse particle sizes. To visualize individual structures, we took Atomic Force Microscopy (AFM) images of each design (Figure C). In-solution AFM was used to visualize these fragile hollow DNA origami boxes that tend to collapse and aggregate when dried, unlike most 2D DNA origami structures (Figure C). The estimated sizes matched our expected size from Cadnano predictions and the DLS measurements. Quantitative analysis of the AFM images was completed, and the sizes of each DNA origami design were similar to what was seen with the DLS measurements (Figure S3B). The width for all boxes was ∼40 nm and the lengths were ∼50, 70, and 80 nm, slightly different than the predicted 40, 80, and 120 nm. However, boxes can adsorb to the surface in many different conformations. The height was somewhat less than expected, but these structures are quite flexible and deformations due to AFM tip interactions with the DNA origami are likely. The DNA origami boxes with closed lids, one-lid-open, or two-lids-open showed characteristically distinct shapes (Figure D). Closed boxes were symmetric in shape. One-lid-open boxes were frequently asymmetric with one side with a shorter height (the lid) than the other (the box). Two-lids-open boxes were symmetric or asymmetric but usually contained a tall structure in the center (the box) with two shorter structures (the lids) on the sides. Quantification of the shapes representing each structure in each assembly (closed box, one-lid-open and two-lids-open) showed that the largest fraction of structures for each assembly matched the predicted shape (Figure D). These results showed that we can assemble hollow DNA origami boxes that are closed or that have either one lid or two lids open by using traditional temperature gradient assembly.

1.

Characterization of hollow DNA origami boxes with two openable lids. (A) CanDo renderings of cadnano files based on the design of the DNA origami box with no lids open, one-lid-open, and two-lids-open. (B) DLS graphs of the size distribution after gradient-temperature assembly and filter purification. The average diameter distribution from five independent assemblies is noted as the black curve and the peak diameter is denoted on the graph (purified at 5000 g for 20 min, 3 μM DNA origami). (C) AFM images of each gradient-temperature assembly. Scale bar represents 200 nm. (D) Shape quantification of the AFM images with sample diagrams and images for each expected assembly shape. Counting was performed on four AFM images from each of two independent assembly experiments per design. The graph shows the fraction of the shapes within an assembly.

2.1.2. Isothermal Assembly

Since the successful assembly of DNA origami boxes was achieved, we were interested in whether we could use other assembly methods to fabricate hollow DNA origami boxes. As demonstrated, small-scale assembly of hollow box DNA origami boxes for structural studies is relatively straightforward, but larger quantities needed for in vivo experiments or even in vitro drug release experiments are difficult to make because of the complex assembly requirements. DNA origami is nearly always assembled with a thermal gradient of −1 °C/min from 95 °C to room temperature, which is accomplished in a thermocycler in 50 μL aliquots. This tedious process wastes time and material. Therefore, we explored whether isothermal assembly, which allows for larger batch production, could produce these hollow DNA structures. Hollow DNA origami structures were assembled for 2 h at a range of assembly temperatures from 4 to 90 °C after heating the assembly mixture to 95 °C for 10 min, the samples were resolved by electrophoresis through an agarose gel (Figure A). Normal gradient assembly results in a band with lower mobility than the scaffold (∼5000 bp). In addition, there is a weak secondary band lower mobility (10,000 bp to 20,000 bp). High temperatures (>80 °C) resulted in a smear in the lane instead of a clear band for the assembled DNA origami boxes, whereas lower temperatures (<80 °C) produced a band at the same size as that produced using the temperature gradient assembly method. Isothermal assembly resulted in DNA origami structures with sizes similar to those achieved with a thermal gradient as measured by DLS and AFM (Figure S4). These results suggested that hollow DNA origami structures are produced during the isothermal assembly (Figure A).

2.

Temperature-dependent assembly of hollow DNA origami boxes under isothermal conditions. (A) Agarose gel of different assemblies over a range of temperatures after heating the assembly mixture for 10 min at 95 °C. (B) Apparent size in equivalent dsDNA base pairs and band intensity of the DNA origami assembly across various assembly temperatures. Red is the closed box assembly. Light blue is one-lid-open assembly. Dark blue is two-lids-open box assembly. (C) Apparent size in equivalent dsDNA base pairs and band intensity of the secondary DNA band that runs more slowly than the DNA origami assembly. Red is the closed box assembly. Light blue is one-lid-open assembly. Dark blue is two-lids-open box assembly. The gray shaded regions are the apparent size/intensity ranges calculated using 95% confidence intervals when assembly is done using a thermal gradient of −1 °C/min starting from 95 °C. Solid lines are melting temperature isotherm of complementary DNA binding fit to the data (c = 5.00 nM, ΔH = −200 kJ/mol, ΔS = −605 J/mol, C _p = −214 J/mol K). The thin line shows the distribution of melting temperatures of the staple strands.

To better characterize the isothermal assembly, we quantified both the band intensity and the apparent DNA size across numerous isothermal assembly temperatures and plotted the distribution of melting temperatures of staple strands (T _m,i ) on the same graph (Figure B). Isothermal assembly temperatures below the majority of the individual staple strand melting temperatures result in DNA origami structures with the same intensity and size that we see with the gradient assembly (Figure B). No significant differences were observed in the isothermal assembly efficiency between closed box, one-lid-open boxes, or two-lids-open boxes. Others have shown isothermal assembly for simple 2D and nonhollow 3D DNA origami designs using a handful of annealing temperatures from 15 to 60 °C with annealing times ranging from 5 min at lower temperatures to 4 h at higher temperatures. ,, Modeling has shown a thermodynamic preference for DNA origami annealing at temperatures less than the staple strand melting temperatures. In addition to the primary band, we observed a secondary band. A similar temperature dependence indicative of isothermal assembly was observed with the secondary band normalized intensity and apparent size. Normalized intensity peaked at around 50 °C, while the apparent size was a minimum at a similar temperature (Figure C). The maximum intensity and the minimum apparent size were at slightly lower temperatures than the peak in staple strand melting temperatures. This secondary band could be dimerized boxes or structures with incomplete assembly (Figure S5). At high temperatures, the dimerization could be inhibited due to higher internal energy, whereas at lower temperatures, the number of misfolded structures due to poor base pairing could be limited. We assessed the kinetics of isothermal assembly with the intent of diminishing the incubation time. , Notably, hollow DNA origami structures assembled very quickly on the order of a minute or so between the temperatures of 4 and 90 °C (Figure S6). DLS confirmed that the DNA origami boxes are present in solution after only 10 min of assembly at 4 °C (Figure S4A and B). These results show that we are able to assemble hollow DNA origami structures isothermally on the order of minutes if the assembly temperature is below the majority of the staple strand melting temperatures.

2.1.3. Purification

When DNA origami is assembled, excess staples are used to ensure complete assembly. After assembly, purification of DNA origami by removal of excess staples eliminates the chance that staples interfere with characterization assays. In these studies, purification was done using concentrating filters with a MWCO of 100,000 Da. Based on the filter’s recommendations, the hollow DNA origami was filter purified from staples using 20,000 g centrifugal force for 5 min. However, this did not remove the excess staples from preparations of closed boxes, one-lid-open boxes, or two-lid-open boxes (Figure A). Concerned that we were crushing the structures, we decreased the centrifugal force to 5000 g for 20 min to minimize its effects on these fragile hollow DNA origami boxes. At this lower speed and longer run time, the hollow DNA origami structures were retained, and excess staples were successfully eliminated (Figure B and C). To determine whether there was damage to the boxes, we imaged them using AFM (Figure D). We observed similar numbers of DNA origami boxes by AFM but more staples (small particles <30 nm) after separation by 20,000 g compared with 5000 g (Figure D and E). Failure to remove excess staples seems to be specific to these hollow DNA origami boxes, as 2D DNA origami structures can be purified at high centrifugal force. We propose that the higher spin speeds crush a small fraction of hollow DNA origami on the surface of the filter. This clogs the filter and eliminates the ability to filter the excess staples at high spin speeds. Accordingly, lower spin speeds would not crush the boxes and, thereby, allow for the removal of more excess staples. Consequently, we have identified proper purification conditions that allow for the removal of excess staples from these hollow DNA origami boxes.

3.

Hollow DNA origami box separation is sensitive to centrifugal force. (A) Agarose gel showing the lack of staples when the assemblies are purified at 5000 g for 20 min compared to 20,000 g for 5 min. (B) DNA origami and (C) normalized staple band fluorescence intensities for closed box assemblies (red), one-lid-open box assemblies (light blue), and two-lids-open box assemblies (dark blue) shown in (A). DNA origami and staple intensities were normalized by dividing the intensity of a particular band by the sum of all band intensities for a particular sample in one lane. (D) In solution AFM images of the origami structures obtained after purification at each of the two centrifugal forces (3 μM DNA origami). Solid white boxes show DNA origami structures within the expected size. Dashed white boxes show particles too small to be DNA origami structures. The scale bar represents 200 nm. (E) Quantification of number of DNA origami boxes vs small box fragments. Counting was performed on one AFM image from a single gradient-temperature assembly experiment per condition.

2.2. Stability

Once we were able to assemble and purify the hollow DNA origami structures, we were interested in examining their stability. Stability is an important aspect of drug delivery systems. For DNA origami to be used in drug delivery, it should be stable in different environments. Temperature stability will determine whether refrigeration is needed for storage and if the drug delivery construct will disassemble at body temperature. Stability at different pHs will determine its potential for application in different tissues or in different compartments within the cell like lysosomes, where acidic conditions are present. Finally, DNA origami drug carriers need to be compatible with other components of the drug formulation, such as solvents such as dimethyl sulfoxide (DMSO). Our base case was DNA origami assembled in the refrigerator (4 °C) at neutral pH (7.5) with 12.5 mM Mg­(OAc)₂. We probed the stability of the DNA origami boxes at different temperatures (4, 18, and 37 °C) and pH values (pH 5 and pH 7.5) for up to 14 days and in solvents (10% DMSO) for up to 3 days by changing only one experimental parameter from the base case at a time. After each incubation time, samples were collected and assessed using electrophoretic resolution through agarose. There was no change in either apparent size or band intensity for DNA origami structures under base conditions (Figure A, B, E, and F), whether unpurified (Figure A and E) or purified (Figure B and F). Altering the temperature to 18 or 37 °C or changing the pH to 5 showed neither a change in apparent size or intensity over the entire incubation time (Figure E and F and Figure S7A and B). To test the stability of the DNA origami structures in the presence of a common drug carrier, we assessed stability in DMSO. Others have shown a gradual decline in stability up to DMSO concentrations of 50%. Adding 10% DMSO did not change the apparent size or band intensity over 3 days (Figure C and G). This demonstrates that these boxes are stable to temperatures as high as body temperature and as low as pH 5 for at least 2 weeks and in the presence of DMSO for at least 3 days.

4.

Hollow DNA origami boxes are temperature, pH and solvent stable. Electrophoretic resolution through agarose gels of hollow closed box DNA origami isothermal assemblies held for various times (A) at 4 °C, pH 7.5 and 12.5 mM Mg­(OAc)₂, (B) at 4 °C, pH 7.5 and 12.5 mM Mg­(OAc)₂ (purified), (C) in 4 °C, pH 7.5 and 12.5 mM Mg­(OAc)₂ 10% DMSO, and (D) assembled in 4 °C, pH 7.5, 12.5 mM Mg­(OAc)₂ and incubated in 10% FBS. Quantification of the DNA origami band intensity for DNA origami assemblies kept at different (E) temperatures, (F) pH (purified), and (G) solvent or serum conditions. Each dot represents the average of at least three independent assembly experiments. (H) Quantification of the apparent DNA origami band size in equivalent dsDNA base pairs normalized to the no serum exposure DNA origami band size as measured in equivalent dsDNA base pairs. DNA origami assemblies kept in 10% DMSO, assembled in 12.5 mM Mg­(OAc)₂ and incubated in 10% FBS, and assembled in 100 mM NaCl and incubated in 10% FBS. Each bar represents the average of three independent assembly experiments. All error bars represent 95% confidence intervals.

Given that these DNA origami structures were stable under different temperatures and pHs and with the addition of DMSO, we decided to test the stability in the presence of 10% FBS over 3 days using gel electrophoresis. In the presence of 10% FBS at 4 °C, the DNA origami band showed no change in intensity or band smearing (Figure D and G). Interestingly, early during exposure to 10% FBS, the DNA origami band shifts to higher apparent sizes, only to return to a normal size after 3 days (Figure D and H). This behavior was observed whether DNA origami was purified from the staples or not (Figure S8). This may suggest that the hollow DNA origami boxes are either transiently coated by serum components or transiently aggregate. ,− Stability in 10% FBS at 37 °C was different. Both purified and unpurified DNA origami structures were disassembled or degraded after 1 day (Figure S8), likely due to serum containing nucleases. , Similarly, other groups have shown that simpler DNA origami structures are completely dissembled or degraded after 1 day in the presence of 10–20% FBS. ,, Some recent evidence suggests that changing the counterion type from magnesium to another ion can increase the stability. However, there appear to be contradictory effects of counterion type and concentration on DNA origami stability. ,, Some studies showed similar folding yields, whereas others showed lower folding yields compared to commonly used magnesium buffers. We probed whether these hollow DNA boxes can be assembled in sodium chloride in the absence of magnesium salts and whether this affected stability in 10% FBS. No dramatic changes in either band intensity or apparent size were observed when the DNA origami boxes were assembled in NaCl (see no exposure condition in Figure G and H and Figures S7C and S8). We examined whether assembly in NaCl affects the stability of DNA origami boxes exposed to 10% FBS at 4 °C (Figure G and H, Figure S7C and Figure S8). The DNA origami boxes were stable under these conditions over the time scale of 3 days. As was seen with Mg­(OAc)₂, early during exposure of DNA origami assembled in NaCl to 10% FBS results in shifts to higher apparent sizes for both purified and unpurified DNA origami early after exposure (Figure H, Figure S7C and Figure S8), but this returns to normal values at 3 days. Interestingly, assembly of DNA origami structures in NaCl seems to inhibit the serum-mediated disassembly or degradation seen in Mg­(OAc)₂ at 37 °C (Figure S8). Overall, the DNA origami boxes are outstandingly stable in serum at 4 °C over 3 days and assembly in NaCl can dramatically increase the stability in serum at 37 °C.

2.3. Small Molecule Loading

Once we determined the stability of the hollow DNA origami boxes, we were interested in whether we could load them with a variety of cargos, including small molecules and large proteins. Around 90% of drugs sold globally are small molecules. For DNA origami to be a viable vehicle for drug delivery, it must be able to be loaded with small molecules. The easiest way to load molecules is to passively load them during assembly. To test whether our hollow DNA origami boxes could be loaded with small molecules, we selected several different fluorescent small molecules as model cargos: fluorescein (negatively charged), calcein (neutral), and malachite green (positively charged). These were chosen to span the spectrum of charges of potential drug cargos. The fluorescent small molecules were added to the DNA origami assembly mixture at a concentration of 10 mg/mL (∼15–30 mM). Negatively charged fluorescein and neutral calcein did not show evidence of loading into the boxes when evaluated by agarose gel electrophoresis (Figure A and B). No band was detected in the dashed white box. Free dye is seen at the bottom of the gel in both the presence and absence of a DNA loading stain. In contrast, malachite green consistently showed evidence of loading as demonstrated by a band in the absence of DNA loading stain (Figure C). Note that free malachite green, which is positively charged, is not found at the bottom of the gel, unlike fluorescein and calcein. Furthermore, the DNA origami band is shifted to higher apparent sizes and appears in the dashed green box rather than the dashed white box. This is likely due to the change in charge imparted on the hollow DNA origami by the positively charged malachite green, decreasing the overall negative charge of the hollow DNA origami. The intensities of the bands of loaded hollow DNA origami boxes differed depending on whether the fluorescent molecule was added during or after box assembly (Figure C and D). The band intensity within the dashed green box was analyzed (Figure C). The condition where scaffold (M13mp18), fluorophore (malachite green), and staples was designated as “entrapped and bound to the outside of the box” (Figure D). The condition where fluorophore (malachite green) was added after DNA origami box assembly was designated as “bound to outside of box”. Indeed, the fluorescence intensity of the band observed when the fluorescent molecule was added after box assembly was lower than the fluorescence intensity of the band observed when fluorescent dye is added during assembly, suggesting that there is entrapped dye within the DNA origami box. The fluorescence intensity of the DNA origami boxes changed in proportion to the concentration of malachite green present during box assembly, whereas the fluorescence intensity of DNA origami boxes incubated with malachite green postassembly saturated at or before 5 mg/mL (Figure D and Figure S9). The difference between these intensities and the matched boxes assembled in the presence of malachite green increased monotonically with malachite green concentration. These results suggest saturation of binding sites on the outside of the box while loading inside the box continued to increase with increasing malachite green concentration. The shift in box position on the gel is correlated with the amount of externally adsorbed malachite green and not with the amount of entrapped malachite green or with the sum of the two (Figure E). While others have passively loaded DNA origami structures with fluorophores, these have tended to be labeled dextrans and we know of no other exploration of passive loading of DNA origami as a function of charge. These results demonstrate that positively charged molecules can be passively loaded into hollow DNA origami boxes in a concentration dependent manner.

5.

Hollow DNA origami loading is dependent on the charge of the cargo. Fluorescent molecules with (A) negative, (B) no, and (C) positive charge were loaded in hollow DNA origami boxes, assembled isothermally, and then separated use PEG purification. Chemical structure and charge in parentheses (left) and agarose gels of 10 mg/mL fluorescent molecule loaded in 3 μM hollow DNA origami boxes (right) are shown. The dotted green box indicates the location of loaded DNA origami boxes. The dotted white box indicates the expected location of the unloaded DNA origami boxes. (D) Quantification of the band intensity referring to the DNA origami box loaded with fluorophore. Dark green refers to the signal when fluorophore was added to preassembled boxes and light green refers to the signal when fluorophore was added during DNA origami assembly. Each bar represents the average of three independent assembly experiments. (E) Quantification of the shift of the band induced by the DNA origami box loaded with fluorophore compared to DNA origami box with no fluorophore. Each square represents the average of three independent assembly experiments. All error bars represent 95% confidence intervals.

2.4. Protein Loading

2.4.1. Predicting Staple Extension Orientation

Once we were able to load the DNA origami boxes with small molecules, we were interested in whether we could load proteins into the boxes. Passive loading, as demonstrated with small molecules, requires high concentrations of the small molecule in the assembly mix. This is likely not feasible for proteins due to solubility and expense constraints. Therefore, an active loading process was used, which involved cutting specific staples at a particular position and extending them with biotinylated 3′ or 5′ ends that could capture biotin-binding cargos. With the knowledge that the average helical twist has approximately 10.4 nt, not considering relaxed or supercoiled DNA, we designed four staples, each with starting positions that were shifted three nucleotides along the helix within an 11 nucleotide (nt) repeating unit and a biotinylated 5 nt extensions of the 3′ or 5′ ends (Figure A). The extensions were exposed to streptavidin labeled beads, and the beads were washed to remove unbound material (Figure B). Scaffold from DNA origami that attached to the beads was extracted by exposure to NaOH (Figure B). The NaOH was neutralized with HCl and scaffold in the two fractions, extract and beads, were resolved by electrophoresis through an agarose gel (Figure C) and quantified (Figure D). DNA origami structures with outside extending staples should bind to beads, and the scaffold should be shown in the bead binding gels. If binding between the DNA origami and beads is complete, then there should also be depletion of scaffold in the extract; however, if binding is not complete, there could be scaffold in the extract as well. Boxes with staple extensions on positions 1 and 4 bound to the beads, whereas boxes with staple extensions on positions 2 and 3 largely remained in the buffer, indicating that staple extensions on positions 1 and 4 were on the outside of the boxes and staple extensions at positions 2 and 3 were on the inside of the box. It is known that staple extensions as short as 3 nt may penetrate through the lattice. , To ensure that longer staple extensions do not penetrate the lattice, we replicated the experiment with a 20 nt staple extension for positions 1 and 3 (Figure S10). The long, 20 nt extension showed bead binding behavior similar to that of the short, 5 nt extension. Using this information, we correctly predicted two more inside staple positions (Figure S11). Staple extensions can now be added anywhere on the DNA origami box with an accurate prediction of whether they extend into or out of the hollow DNA origami. This capability introduces the possibilities of actively loading protein or of tethering molecules to the outside of DNA origami.

6.

Staple extensions from the hollow DNA origami box can be designed for capture of cargo inside or for decoration with targeting molecules outside, using 5 nt extensions. (A) Schematic of 5nt staples extending inside and outside of the hollow DNA origami box (left). Zoomed view of a single DNA helical twist, showing the direction of extension of different staples. (B) Schematic of the experimental process of attaching hollow DNA origami boxes with biotinylated extended staples to streptavidin coated beads to determine if the extension was directed toward the inside or outside. (C) Agarose gel showing the binding capabilities of hollow DNA origami boxes with staple extensions at different positions. Light blue refers to staples oriented outside and purple refers to staples oriented inside. The no bead binding wells are samples taken from the supernatant after beads were spun down. The bead binding wells are samples taken from the solution after beads were exposed to NaOH, removing the biotinylated hollow DNA origami boxes from the streptavidin coated beads and subsequently neutralized with HCl. (D) Quantification of the scaffold band intensity of the hollow DNA origami box. Each bar represents the average of three independent isothermal assembly experiments. All error bars represent 95% confidence intervals.

2.4.2. Streptavidin Loading

Once we determined the orientation of the extended staple tethers, we loaded the hollow DNA origami boxes with biologics. Biologics (proteins, antibodies, etc.) are used for a number of disease treatments, so a viable way of loading proteins in delivery systems is needed. As mentioned above, passive loading of large molecules is prohibitive due to the challenges of solubility limits and cost. Active loading, using staple extensions that can attach to proteins through an affinity tag, provides an effective way to load hollow DNA origami boxes. As a model system, we used biotinylated staple extensions and streptavidin. A variety of other nucleotide-protein or functionalized nucleotide tethers including thiol-modified staples, complementary ssDNA strands, zinc “fingers”, antibody fragments and more could be used to actively load protein into DNA origami boxes. ,,,, Staple extension lengths likely determine whether a protein can bind to the staple extension. Indeed, previous studies have shown that the length of the tether of surface-attached ligands influences the kinetics of protein attachment, with an optimal tether-length providing for the highest affinities. , We used the two staple extensions of 5 nt (short) and 20 nt (long) that were used above to determine the staple extension orientation (Figure A). To visualize streptavidin, we used Atto 488-biotin to occupy on average two of the four sites on the protein, which was optimized by titration (Figure S12A). The remaining sites would be available for binding to the biotinylated staple extensions.

7.

Hollow DNA origami loading requires inside staple extensions of a minimum length. (A) Schematic showing that short biotinylated staples on DNA origami boxes, assembled isothermally, do not attach to streptavidin, while long biotinylated staples on DNA origami boxes can attach to streptavidin. Components not shown to scale. (B) Agarose gels of DNA origami boxes with short, outside (left) and long, inside (right) extensions. Green dashed box refers to the hollow DNA origami box with attached streptavidin bound to fluorescent biotin. Schematics of other species of bound streptavidin are shown to the left.

We loaded the hollow DNA origami boxes with the streptavidin conjugate by adding Atto 488-biotin–streptavidin to the assembly mixture. The assembly mixture is initially heated to 95 °C. Therefore, we considered the temperature sensitivity of the streptavidin–biotin interaction. While streptavidin is only thermostable to 75 °C, its thermal stability increases to 112 °C upon biotin binding. In addition, we showed that streptavidin heated to 95 °C does not bind to Atto 488-biotin, but Atto 488-biotin–streptavidin heated to 95 °C is indistinguishable from nonheated Atto 488-biotin–streptavidin when run on an agarose gel (Figure S12B). DNA origami boxes with inside staple extensions of 5 and 20 nt assembled in the presence of Atto 488-biotin–streptavidin were resolved through an agarose gel (Figure B) in the presence or absence of a DNA loading stain. A band at the DNA origami box size (green dashed box) was observed in the absence of DNA loading dye for only the 20 nt staple extension (Figure ). Similarly, DNA origami boxes with outside staple extensions of 5 and 20 nt assembled in the presence of Atto 488-biotin–streptavidin were resolved through an agarose gel (Figure S13) in the presence or absence of a DNA loading stain. A band at the DNA origami box size (green dashed box) was observed in the absence of DNA loading dye for only the 20 nt staple extension, even after 6 h of incubation (Figure S13). Consequently, we concluded that the 5 nt length extension does not allow for binding of Atto 488-biotin–streptavidin to either inside or outside staple extensions. There is likely steric hindrance inhibiting the binding of the Atto 488-biotin–streptavidin to the DNA origami boxes as the 5 nt extension is approximately 1.7 nm in length compared to the 5 nm hydrodynamic radius of the streptavidin. Biotin binding with streptavidin results in almost an entire engulfment of the biotin as opposed to an adsorption, explaining biotin’s extremely high affinity binding with streptavidin. When the staple is extended to 20 nt (6.6 nm), streptavidin can more easily attach to the biotinylated staple both inside and outside of the DNA origami boxes. These binding experiments indicate that staple extensions of longer than 5 nt are needed to either bind targeting proteins on the outside of the hollow DNA origami box or actively load protein cargo into the cavity of the hollow DNA origami box. Taken together, these experiments outline the design principles associated with tethering targeting molecules to the outside of a hollow DNA origami box and actively loading it with protein cargo.

3. Discussion

In this paper, we showed the design, fabrication, and characterization of a two-lidded hollow DNA origami box. Measurements using AFM and DLS resulted in DNA origami box sizes that matched the predicted size. Isothermal DNA origami assembly was observed at ∼60 °C, which is the average melting temperature of the staple set. Temperatures between 40 and 60 °C, resulted in misfolded or partially assembled structures that produced high apparent size. Stability of the DNA origami structures was characterized under a variety of conditions, showing a relatively stable structure across several temperatures at pH of 7 and across several pH values at a temperature of 4 °C. The addition of FBS resulted in some degradation at 37 °C, but this was dramatically tempered when the DNA origami was assembled in the presence of sodium rather than magnesium. Finally, we demonstrated active and passive loading strategies. DNA origami could be loaded with positively charged small molecule cargo. Biotinylated tethers could be designed to be pointing inward or outward locating streptavidin on the inside or outside, and tether lengths of greater than 5 nt are needed for attaching streptavidin.

Most protocols for DNA origami assembly have used changes in temperature over long time periods during assembly, but this limits production to small volumes assembled in a thermocycler. − Staple sequences have slightly different melting temperatures; therefore, ramps allow for higher melting temperature sequences to assemble first. Slow ramps then give time for staples to find their complementary sequences within the scaffold. However, staple sets are rarely designed with the order of assembly in mind, so using the long-time temperature ramp seems to be a decision borne from convention rather than necessity. There are some examples of isothermal assembly, but incubation times are long , or use relatively simple DNA origami systems. We show that isothermal assembly can be used for this DNA origami structure and present data to indicate that assembly might be quite fast, thereby challenging the convention that slow ramps are needed.

DNA origami constructs have been shown to be reasonably stable, but this seems to depend highly on the design of the structure, , necessitating an examination of stability for each design. DNA origami has been probed for stability in fetal bovine serum. Some wireframe designs seem to show significant degradation in 10% FBS at 37 °C after 6 h whereas others are stable up to 24 h. Hollow 3D DNA origami boxes have been shown to be instantaneously degraded in 1% FBS at 37 °C. We showed that our hollow 3D DNA origami box design was also degraded by 1 day after incubation with 10% FBS at 37 °C, but it was stable in 10% FBS at 4 °C for 3 days. DNA origami structures have shown to be more stable when assembled with various ions, even when exposed to some physiological conditions. , Consequently, we assessed stability in sodium ions rather than magnesium ions. We found good stability in 10% FBS over 3 days, even at 37 °C and in the absence of a protecting agent. ,

While DNA origami has been loaded with cargo, several issues remain. Wireframes that can release enzyme in response to temperature have been designed. DNA origami tubes that can be opened along their length using logic nucleic acid keys have also been designed. However, wireframe and open designs still suffer from nonspecific delivery due to their porous structures. ,,, Hollow and enclosed containers can overcome this problem. DNA origami box, cube and sphere designs with lids that can open in response to nucleic acid keys or light have been fabricated. − While these boxes were shown to open on demand, their loading with cargos was not demonstrated with the exception of the “enzyme vault” which was loaded with one protein via click chemistry. In this paper, we show the design and assembly of a two-lidded DNA origami box and load it with both small molecules as well as protein cargo. Our ability to both passively load small molecules or actively load protein through tethers positions these structures to be used in future work controlling release. The two-lid design can be coupled with the design of different environmentally sensitive locks to release cargo in response to environmental cues for a variety of applications.

4. Materials and Methods

4.1. DNA Origami Design

The design of staple strands to assemble the hollow DNA origami box was done using a cadnano square. The predicted 3D structure was created using CanDo and UCSF Chimera imaging. − M13mp18 circular bacteriophage DNA was used as the scaffold for the staple design. The full staple set is shown in Table S1.

4.2. DNA Origami Assembly

Hollow DNA origami boxes were assembled in a TAE (40 mM Tris base, 20 mM glacial Acetic acid and 0.8 mM EDTA) buffer with Mg­(OAc)₂ (12.5 mM) or NaCl (100 mM). M13mp18 circular DNA (P-107, Bayou Biolabs) was used as the scaffold for the staples (IDT). Scaffold (1 to 3 μM) was used with staples at a concentration between a 1:2 to 1:5 scaffold:staple molar ratio. For gradient assembly, the assembly solution was split into 50 μL aliquots and placed into a thermocycler to be heated at 95 °C for 5 min and cooled at −1 °C/min until 25 °C. The aliquots were combined after cooling. For isothermal assembly, the assembly solution was heated in a dry bath at 95 °C for 10 min and then cooled to 4–90 °C for 30–120 min. For quick isothermal assemblies (approximately 1 min or less), the assembly solution was heated as normal, placed in ice for 1 min, then immediately placed into an agarose gel.

4.3. DNA Agarose Gel Electrophoresis

Agarose gels (1%) were run at 5.3 V/cm for 60 min. Prestain or loading stain was used to visualize the DNA bands. If prestain was used, 6 μL of prestain (SmartGlow Prestain, E4500-PS, Accuris Instruments) was added to 60 mL of agarose gel. If loading stain was used, an appropriate volume of 6x loading stain (SmartGlow Loading stain, E4500-LD, Southern Labware) was added to bring the sample to 1X loading stain before loading into the gel. For fluorescent cargo experiments DNA stains were not used, and in place of a 6x loading stain, a 50% (v/v) glycerol-in-water solution was added to the sample at the same volume as the loading stain. Gels were analyzed using the Gel Analyzer tool in ImageJ. The analysis area for each gel extended from immediately after the well to immediately after the staple bands. The intensities of the bands were determined by finding the area under each peak of the averaged linescan, which accounts for background fluorescence calculated locally.

4.4. DNA Origami Purification

Purification was done using Amicon 100,000 and 50,000 Da MWCO filters (UFC5100, Sigma). The manufacturer recommends that the filters be spun in the centrifuge at approximately a centrifugal force of 20,000 g for 5 min, which was later lowered to 5000 g for 20 min. AFM was used to observe differences between the purification speeds. Size analysis was done on particles, where any particle with a length of less than 30 nm was considered small. Alternatively, the following procedure for PEG purification was used for loading experiments to prevent clogging of the filters and to limit expense. Large amounts of dye used in loading experiments required several filter passes. This resulted in either clogged filters if the same filter was used or the necessity to use multiple filters. PEG (50 μL, 17 wt % 8000 Da with TAE and Mg­(OAc)₂ (12.5 mM)) was added to 400 μL of assembled DNA origami. The solution was gently mixed and placed at 4 °C for 10 min before being spun at 12,600 g for 30 min. The top 400 μL was removed, and 350 μL of TAE with Mg­(OAc)₂ (12.5 mM)­was added along with another 50 μL of PEG solution. This was repeated until the visible dye was dramatically diluted (1–4 repetitions).

4.5. Dynamic Light Scattering (DLS)

DLS on DNA origami was performed using the Zetasizer Nano-ZS (Malvern). With ZEN0040 equiv cuvettes, 50 μL of either filter purified from staples or unpurified DNA origami boxes was used to determine the size. Water was selected as the dispersant, and a refractive index of 1.53 was used for the DNA origami size determination. The system temperature was set to 25 °C with an equilibration time of 120 s for data in Figure . For the isothermal data in Figure S4, the system temperature was set to 4 °C.

4.6. Atomic Force Microscopy

4.6.1. Sample Preparation

A muscovite mica (grade V1) substrate was subjected to a fresh cleavage process to achieve optimal surface conditions for Atomic Force Microscopy (AFM) imaging. Subsequently, a 200 μL volume of a purified DNA origami sample was combined with 100 μL of a TAE buffer containing Mg­(OAc)₂ (12.5 mM). The resultant mixture was gently deposited on the surface of the mica substrate. Following deposition, adsorption was allowed to proceed for 35 min at room temperature before the AFM experiment.

4.6.2. Instrumentation

AFM imaging was conducted by utilizing a BioScope Resolve system (Bruker), equipped with a SCANASYST-FLUID probe featuring a spring constant of 0.4 N/m and tip radius 20 nm. Imaging was executed in the ScanAsyst mode. The BioScope Resolve AFM sitting inside a vibration and noise isolation chamber reduced the noise level in nanoscale measurements. It was also sitting on top of a vibration isolation table, which needed to be switched on prior to each experiment. There was a magnetic contact on the base plate/sample stage used to lock the stainless-steel discs that were attached to the mica used for sample imaging. All imaging procedures occurred in solution within a controlled environmental chamber, which was isolated to minimize potential sources of drift and temperature fluctuations.

4.6.3. Imaging Parameters

AFM images were acquired at a variable scan rate spanning between 0.3 and 0.4 Hz to prevent damage to the DNA origami sample and to yield topographical and structural details. The scan size was modulated within the range of 3–7 μm to capture diverse regions of interest in the muscovite mica substrate, and the AFM height images were zoomed in on locations with significant features. A consistent peak force set point of 200 pN was maintained to preserve the structural integrity of DNA origami. However, some compression of the structure was detected in the z-direction. Imaging utilized a resolution of 256 pixels per line. The peak force amplitude set point and peak force frequency were kept constant at 250 nm and at 1 kHz respectively during imaging.

4.6.4. Data Analysis

Nanoscope Analysis and WSxM 5.0 software were employed to process and analyze the AFM images acquired during the experiment. Structural attributes, including dimensions and topographical features of the DNA origami structures, were extracted from these images. Shape characteristic data were found by manually counting the number of distinct shapes across four images for two independent assemblies per DNA origami box design. Purification data was quantified by manually counting the number of small molecules (size less than ∼25 nm) and DNA origami boxes (size approximately 30–45 nm) in one image per condition. For data shown in Figure S3, ImageJ was used to measure the particles’ lengths and widths using the “line” and “plot profile” tools. The length was the longest side of a rectangle fit to the object. The width was the length scale perpendicular to the length. Both length and width were measured at half-height of the object. Five random particles were selected for each measured image (two-three images per DNA origami design).

4.7. Stability

Stability at different temperatures was tested as follows with a base case of 4 °C, pH 7.5, with Mg­(OAc)₂ (12.5 mM) and no DMSO or FBS. For example, when the temperature was changed, pH and ion remained the same as the base case. Unpurified DNA origami closed boxes were kept at 4 °C, 18 and 37 °C for up to 14 days. Equal volumes of each sample from each time point for each temperature were loaded into the gel (approximately 10 μL/sample). Stability as a function of pH was tested as follows. Filter-purified DNA origami closed boxes were kept at pH 7.5 and pH 5 for up to 14 days. Both buffers were TAE with Mg­(OAc)₂ (12.5 mM). The pH 5 buffer was adjusted using acetic acid, and the pH 7.5 buffer was not adjusted. DNA concentrations of filter-purified DNA origami boxes were measured using a spectrophotometer to check for stability based on mass. The pH 5 samples were at a 4:1 volumetric ratio of pH 5 TAE buffer to filter-purified DNA origami boxes in pH 7.5 TAE. Equal mass of DNA origami was loaded into the gels for each time point for each condition. Stability in DMSO and FBS was tested as follows. Unpurified DNA origami closed boxes were kept in 10% DMSO or 10% FBS (nonheat-inactivated, A5209401, Gibco) for up to 3 days. In experiments using DMSO, the DNA origami closed boxes were assembled with Mg­(OAc)₂ (12.5 mM). In experiments using FBS, the DNA origami closed boxes were assembled with Mg­(OAc)₂ (12.5 mM) or with NaCl (100 mM). Equal volumes from each stability sample from each time point for each condition were loaded into the gel (approximately 10 μL/sample).

4.8. Loading Hollow DNA Origami Boxes with Fluorescent Small Molecules

DNA origami boxes were loaded with molecules of different charge: fluorescein (−2) (863200, Carolina Biological Supply), calcein (0) (C0875, Sigma), or malachite green (+2) (32745, Sigma). Closed DNA origami (3 μM, with 12.5 mM Mg­(OAc)₂) boxes were assembled either with fluorescent molecules to measure how much was entrapped and bound to the DNA structure or after assembly to measure how much was bound to the DNA structure. Final concentrations of fluorescent molecules present during assembly ranged from 1 to 10 mg/mL. Loaded DNA origami boxes were purified using up to four rounds of PEG purification to reduce background fluorescence. Some loaded DNA origami samples showed a mobility shift due to a difference in charge between the loaded DNA origami structure and DNA origami structure with no dye. Under some conditions, the DNA origami structure moved more slowly than the largest ladder band (approximately 20 kb). These structures were quantified in pixels rather than by estimates of apparent size. Gel pictures were taken at a consistent height using the SmartDoc Gel Imaging System (Accuris Instruments) with Blue Light Illumination.

4.9. Extending DNA Staples

Extended DNA staples were designed by cutting the original staple at a designed position and extending it with a 3′ or 5′ poly-A or poly-T tail of lengths 5 or 20 nt long. The remaining small fragment was not used in the assembly. The cut was made to minimize the length of the original sequence that was not used in the assembly. The unused sequence was no more than 10 nt long. The polyA or polyT segments were not complementary to the scaffold. Once a position was identified as either inward facing or outward facing, one could determine other inward or outward facing positions along a single helical twist by counting 11 nt in the cadnano design, based on the assumption of normal tension and pitch. The extended staples were biotinylated on the 3′ or 5′ end of the extensions that would be inside or outside the DNA origami boxes. All staples extended for Figure were biotinylated at the 3′ end. All staples extended for Figure S11 were biotinylated at the 5′ end. The 5 nt sequences are as follows: Figure position 1 staple extension 1:5′-TGC CGT CGA TTT CGG AAC CTA TTA AAA A-3′. Figure position 2 staple extension 1:5′-TGC CGT CGA TTT CGG AAC CTA TTA TTA AAA A-3′. Figure position 3 staple extension 1:5;- TGC CGT CGA TTT CGG AAC CTA TTA TTC TAA AAA-3′. Figure position 4 staple extension 1:5′- TGC CGT CGA TTT CGG AAC CTA TTA TTC TGA AAA AA-3′. Figure S11B staple extension: 5′-TTT TTA GAG CTC CAA CGT CAA CCT CAG AAC CGC CAC CC-3′. Figure S11C staple extension: 5′- TTT TTA CAT GGC TTT TGA TGA TAC AGG ATT GGC CTT-3′. For experiments identifying inward or outward facing extensions, the DNA origami boxes were exposed to streptavidin-coated beads (MyOne Dynabeads, 65001, Thermofisher) and gently mixed for 10 min, and the supernatant (no bead binding supernatant) was collected after pulling down the beads with a magnet. Then the beads were exposed to NaOH (10 μL, 100 mM), gently mixed for 10 min and the supernatant (bead binding supernatant) was collected and added to HCl (10 μL, 100 mM) to neutralize the solution. No bead binding and bead binding supernatants were loaded into agarose gels and analyzed as described above.

4.10. Loading Hollow DNA Origami Boxes with Fluorescently Labeled Protein

There were short poly-A extensions of 5 nt and long poly-A extensions of 20 nt from the original staple extensions that were directed inward or outward from the DNA origami boxes (listed above) (3 μM, with 12.5 mM Mg­(OAc)₂). The extended staples were biotinylated on the 3′ end of the extensions that would be inside or outside of the DNA origami boxes. This would allow them to bind to streptavidin (21122, Pierce). Streptavidin was bound with Atto 488-biotin (30574–1 mg-f, Atto) (1:2 molar ratio of streptavidin:Atto 488 biotin). Streptavidin-Atto 488-biotin conjugate was added at a 1:1 molar ratio to the biotinylated extended staples. When using the short, outside extended staples, DNA origami boxes with the staples were exposed to the Atto 488-biotin–streptavidin conjugate for approximately 6 h. When using the long, outside extended staples, DNA origami boxes with the staples were exposed to the Atto 488-biotin–streptavidin conjugate for approximately 30 min. When using the short or long inside extended staples, isothermal assembly of the DNA origami boxes at 4 °C for 30 min was done with the Atto 488-biotin–streptavidin conjugate.

5. Conclusions

In this article, we reported the design and assembly of a unique hollow DNA origami structure with multiple lids. We have characterized the sizes and shapes of these structures using a variety of techniques including in-solution AFM. We have developed an approach to isothermally assemble hollow DNA origami structures across a variety of temperatures over the time scale of minutes, providing a means of rapidly assembling many boxes. While these hollow DNA origami structures appear to be sensitive to centrifugal force, they are exceedingly stable at different temperatures and pH and in different salts as well as in the presence of both solvents and biologically relevant fluids, particularly when assembled in sodium rather than magnesium counterions. We demonstrated that these hollow DNA origami boxes are agnostic to cargo size, loading both small molecules and proteins. However, the charge of the small molecule governs the ability to passively load hollow DNA origami boxes. Active loading of large proteins can be facilitated by extending staples into the hollow DNA origami box cavity; however, staple extension length must be optimized to enhance affinity and to overcome steric hindrance. The internal tethers could be temperature labile, binding cargo at cold temperatures during assembly with release under higher temperatures after the DNA origami box is closed, priming the cargo for release upon lid opening. These DNA origami boxes with multiple lids have the potential to be used with locking mechanisms that can be opened via different inputs to release a variety of cargo in drug delivery applications.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsabm.5c00907.

• Details of staple sequences and DNA origami design and figures showing additional characterization of DNA origami structure, stability, and loading (PDF)

A.K.: Conceived and conducted the experiments, analyzed the data and interpreted the data across all figures, wrote the article, prepared the figures, and edited the article. N.M.: Conducted the experiments and interpreted the data in Figures 1, 3, S3, S4, and S5 and edited the article. J.V.H.: Conducted the experiments, analyzed the data, and interpreted the data in Figures 4, S6, and S7. C.S.: Conducted the experiments in Figures 1, 3, and S4. M.N.-H: Participated in discussions of the data, their interpretations, made suggestions for experimental design, and edited the article. A.S.: Conceived and conducted the experiments and interpreted the data in Figures 1, 3, S3, S4, and S5, and edited the article. I.C.S.: Conceived the experiments, analyzed the data, and interpreted the data across all figures, wrote the article, prepared the figures, and edited the article.

This work was supported by the College of Engineering and the Department of Chemical and Biological Engineering at Iowa State University. A.K. was supported with a graduate research fellowship from the National Science Foundation (2022329640). Work was supported by NSF award 2333556.

The authors declare no competing financial interest.