DNA Nanostructures for Targeted Antimicrobial Delivery

Abstract We report the use of DNA origami nanostructures, functionalized with aptamers, as a vehicle for delivering the antibacterial enzyme lysozyme in a specific and efficient manner. We test the system against Gram‐positive (Bacillus subtilis) and Gram‐negative (Escherichia coli) targets. We use direct stochastic optical reconstruction microscopy (dSTORM) and atomic force microscopy (AFM) to characterize the DNA origami nanostructures and structured illumination microscopy (SIM) to assess the binding of the origami to the bacteria. We show that treatment with lysozyme‐functionalized origami slows bacterial growth more effectively than treatment with free lysozyme. Our study introduces DNA origami as a tool in the fight against antibiotic resistance, and our results demonstrate the specificity and efficiency of the nanostructure as a drug delivery vehicle.


Modifications to the 5-well frames
The following oligonucleotides were modified to carry aptamers that can bind E.coli and B.subtilis [2] . The aptamers sequence is CAT ATC CGC GTC GCT GCG CTC AGA CCC ACC ACC ACG CAC C (in red in the table below).

Sequence Name
Sequence Bases 5wf-009  1NO-apt   AAA ATT TTA GCC TTT ATT TCA ACG TTT TTC TAC TAA TAG TAG TAA AAA  GGT GGC ATC AAT TTT TT C ATA TCC GCG TCG CTG CGC TCA GAC CCA  CCA CCA CGC ACC  5wf-015  2KL-apt   GAG AGA TCT GGA GCA AAC AAG AGA CAA TCA TAT GTA CCC CAG   The following oligonucleotides were functionalised with Alexa 647 molecules.

Atomic Force Microscopy (AFM)
Origami tiles were diluted ten times in origami buffer and 25 µl of the sample were deposited on freshly cleaved mica and incubated at room temperature for 10 minutes. The samples were then To load the streptavidin functionalised origami tiles with the antimicrobial enzyme, the tiles were incubated with 1mg/ml biotinylated lysozyme (Chicken Lysozyme protein, Egg whites, GeneTex) for 10 minutes at room temperature after which they were filtered and imaged as above.

AFM data analysis
Images of all datasets were plane-fitted using the speed-optimised plane correction function of the SPIP software (Image Metrology A/S, Hørsholm, Denmark), which fits each line in the horizontal axis to a polynomial equation. SPIP was also used for calculation of the volumes of proteins attached to the DNA origami tiles. The "inspection window" feature of SPIP was used to zoom into individual tiles and then the "circular area of interest" tool was used to allow the software to calculate only the volume of the protein rather than that of the whole tile, according to the following equation: where Zmaterial volume is the volume of all pixels inside the shape's contour with a Z value greater than zero: where dx and dy are the point spacings in the X and Y directions of the image, respectively. Zvoid volume is the volume of all pixels inside the shape's contour with a Z value lower than or equal to zero: where dx and dy are the point spacing in the X and Y directions of the image, respectively.
For cross-sections of sample features, tile dimensions measurements, as well as for the 3D rendering of the images, Nanoscope 1.9 software (Bruker) was used. Volume histograms were drawn with bin widths chosen according to Scott's equation [3] , using GraphPad Prism.
"Theoretical" molecular volumes of proteins based on molecular mass were calculated using the equation by Schneider et al [4] : where M0 is the molecular mass, N0 is Avogadro's number, V1 and V2 are the partial specific volumes of protein and water, respectively, and d is the extent of protein hydration. The partial specific volume of a typical protein (V1) is considered to be 0.74 cm 3 g -1 , and the extent of protein hydration (d) has been estimated to be 0.4 g of water/g of protein.
The partial specific volume of water (V2) is 1 cm 3 g -1

Fluorescence microscopy experiments
The fluorescence microscopy experiments performed on the origami structures were conducted on a custom-built microscope based on an Olympus (Center Valley, PA) IX-73 frame with a 100x 1.49 NA oil objective lens (Olympus UAPON100XOTIRF) and a 647-nm laser (MPB Communications Inc. VFL-P-300-647-OEM1-B1). The samples were imaged in total internal reflection fluorescence (TIRF) mode and dSTORM images collected on an Andor iXon Ultra 897 camera as described previously [5] .

dSTORM on origami structures
16000 frames were acquired for the dSTORM reconstructions, each recorded at 20 ms exposure time using an EM gain of 200 over a 256x256 pixel region. The pixel size in the image plane was measured to be 118 nm. The raw single molecule data sets were reconstructed using ThunderSTORM [6] , and visualized as averaged shifted histograms with a magnification factor of 10.
The peak-to-peak distance between the fluorophores tethered to the origami structures was measured by taking a cross-sectional profile in Fiji/ImageJ [7] between two bright spots in different regions of interest in the reconstructed image, and using a custom MATLAB (Natick, MA) script to measure the average distance between two peaks. Representative large fields of view can be seen in Figure

Structured Illumination Microscopy (SIM)
Images of the sample were collected using 3-color SIM for optical sectioning [8] . Cobolt). Images were acquired using custom SIM software described previously [9] .
An automated analysis routine for processing SIM images was written in MatLab. In order to

Section 4. Bacterial growth curves
Bacterial cell culture studies were conducted using E. coli BL21(DE3), expressing GFP and B. subtilis (BS168). All experiments were conducted in LB medium, supplemented with carbenicillin (100 µg/ml) for E. coli and chloramphenicol (25 μg/ml) for B. subtilis. Bacterial starter cultures were grown overnight, and the bacteria were then diluted 1:100 into 150µl LB, and grown over 16 hours in a shaking plate reader, at 37°C, with measurements taken every 5 minutes, in the following conditions: The OD values at 600nm were collected and used for the creation of growth curves. For each condition, 9 individual growth curves were analysed and averaged. Individual growth curves were fitted in MATLAB using the curve fitting toolbox, to a re-parameterised Gompertz growth model [10] , to extract growth rates.
DNA origami carrying lysozyme were prepared as described in Section 2.1 and added to the samples where appropriate. The growth rates for E. coli and B. subtilis grown in the presence of DNA origami without aptamers are presented in Figure

Section 5. Binding affinity of DNA nanostructures
We estimated the apparent dissociation constant Kd of the aptamer-functionalised nanostructures, to better understand their affinity for the bacterial targets and the impact of having many of them locally concentrated into a multivalent complex.
The aptamers used have a dissociation constant Kd of 27.2 nM for E.coli and 9.97 nM for B.subtilis according to Song et al [2] . Recently, Csizmar et al. [11] have used multivalent scaffolds to target tumour cells and have proposed the following equation to quantify the effect of the multiple valency, N, in a molecular scaffold, on its apparent affinity: where Kd,1 is the affinity of a single-target ligand, and Kd,N is the apparent affinity of the multipletarget ligand.
Our DNA nanostructure carries 14 aptamers; we thus obtain an apparent Kd,N for E. coli and B.
subtilis of 141 pM and 50 pM respectively.

Section 6. Cell viability assay
In order to explore the future potential of the DNA origami nanostructures to be used in vivo for selective bacterial targeting, we performed a mammalian cell viability assay. We used the CellTiter 96® AQueous One Solution Cell Proliferation Assay (Promega), to assess the effects of the DNA origami on mammalian cells. COS-7 cells were plated in a 96-well plate at concentration of 10,000 cells/well in 100µl of media (DMEM+10%FBS). 20µl of CellTiter 96® AQueous One Solution Reagent were added per well and the cells were incubated at 37°C for 2 hours in a humidified, 5% CO2 atmosphere. After 2 h, the absorbance at 490nm was measured, using a 96-well plate reader.
The measurements were performed in triplicates. In the case of lysozyme, it is possible that the loading of multiple enzymes on the same delivery platform leads to synergistic action that enhances its antimicrobial activity. Previous work has indeed indicated that the targeted delivery of several lysozyme molecules does increase antimicrobial activity locally. For example, it has been reported that all of Dextran-conjugated lysozyme [12] and chitosan-lysozyme [13] and selenium-lysozyme [14] nanoparticles loaded with increasing amounts of lysozyme increase activity of the enzyme. The activity of lysozyme has also been shown to increase through delivery via "Engineered Water Nanostructures" [15] . So overall there is plausible evidence that multiple loading sites provide cumulative benefits for antimicrobial applications, which will be explored in future work.
It is also possible that charge effects mediated by the DNA origami platform affect enzyme function. Although there is no available literature on lysozyme / DNA origami effects of this nature, an increased enzymatic activity was observed when other enzymes (i.e. not lysozyme) were coupled to DNA origami. For example, T. Morii's group used DNA origami to assemble ribulose biphosphate carboxylase/oxygenase (RuBisCO). They show that the enzymatic activity is retained upon binding and possibly enhanced [16] . Similarly, Zhao et al. showed that GOx/HRP enzyme pairs exhibit enhanced catalytic activity when bound to DNA nanocages [17] , while Ora et al. report intact activity of enzymes bound on DNA origami for delivery to mammalian cells [18] . Potentially these effects are indeed mediated by the charge of the DNA scaffold.