Oligolysine-based coating protects DNA nanostructures from low-salt denaturation and nuclease degradation

DNA nanostructures have evoked great interest as potential therapeutics and diagnostics due to ease and robustness of programming their shapes, site-specific functionalizations and responsive behaviours. However, their utility in biological fluids can be compromised through denaturation induced by physiological salt concentrations and degradation mediated by nucleases. Here we demonstrate that DNA nanostructures coated by oligolysines to 0.5:1 N:P (ratio of nitrogen in lysine to phosphorus in DNA), are stable in low salt and up to tenfold more resistant to DNase I digestion than when uncoated. Higher N:P ratios can lead to aggregation, but this can be circumvented by coating instead with an oligolysine-PEG copolymer, enabling up to a 1,000-fold protection against digestion by serum nucleases. Oligolysine-PEG-stabilized DNA nanostructures survive uptake into endosomal compartments and, in a mouse model, exhibit a modest increase in pharmacokinetic bioavailability. Thus, oligolysine-PEG is a one-step, structure-independent approach that provides low-cost and effective protection of DNA nanostructures for in vivo applications.


Supplementary
Effect of different lengths of oligolysine on the stability of DN1 in low-salt buffer at 37 °C. Cy5 labelled DN1 (coloured red) was coated with different lengths of oligolysine at various N:P ratios, depicted in the figure and then diluted with RPMI-1640 such that the overall Mg 2+ concentration is 0.6 mM. The samples were incubated at 37 °C for 1 hour, then loaded onto a 2% agarose gel and ran in 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 2 hours. Longer lengths require lower N:P to stabilize the DNs and cause aggregation of DNs at higher N:P. Figure 6| Negative stain TEM images of DN1 coated with different lengths of oligolysine in low-salt buffer at 37 °C. DN1 was coated with different polyamines at various N:P ratios, depicted in the figure and then diluted with RPMI-1640 such that the overall Mg 2+ concentration is 0.6 mM. The samples were incubated at 37 °C for 1 hour. a) DN1 + K 4 (100:1); b) DN1 + K 6 (N:P = 10:1); c) DN1 + K 10 (N:P = 0.5:1); d) DN1 + K 12 (N:P = 0.4:1); e) DN1 + K 16 (N:P = 0.2:1); f) DN1 + K 20 (N:P = 0.05:1). Longer lengths require lower N:P to stabilize the DNs and cause aggregation of DNs at higher N:P. DNs were stained using 2% uranyl formate for 30 seconds. Scale bar is 100 nm in all images except 6e, where it is 200 nm. Figure 7| AGE of different DNs. DN1-DN8 were annealed as described in the methods and purified using the glycerol gradient ultracentrifugation method. 10µL of 10 nM purified samples were loaded onto a 2% agarose gel and ran in a 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 2 hours.

Supplementary
Supplementary Figure 8| Negative stain TEM images of purified DN1-DN8. DNs were stained using 2% uranyl formate. Scale bars are 500 nm for DN7 and 100 nm for all other panels. Figure 9| Chondroitin sulfate removes oligolysine coating from DNs without damaging the structures. This is a useful operation, as it allows gel analysis of oligolysine-coated DN structures that have been subjected to some environmental condition while N:P = 1; without stripping the oligolysine, such structures have neutral charge and won't enter the gel. Bare, K 10 coated and K 10 -PEG 5K coated DN1 were incubated with different excess of chondroitin sulfate relative to the amount of amines in oligolysine for 2 hours at 37 °C. The Mg 2+ concentration was adjusted to 10 mM. The samples were loaded on a 2% agarose gel and ran in 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 1 hour. Intact structures were observed under negative stain TEM for DN1 when incubated with 100x chondroitin sulfate. Note that DN1 has been purified of excess staples using glycerol gradient and that chondroitin sulfate is stained by ethidium bromide. Scale bar is 100 nm. Figure 10| Dialysis of DN1-DN8 into zero-Mg 2+ buffer (5mM Tris, 1mM EDTA, pH 7). 50µL of bare, K 10 coated and K 10 -PEG 5k coated DN1-DN8 were dialyzed into 1L of zero-Mg 2+ buffer overnight at room temperature. 'v' refers to an undialyzed positive control and 'C' refers to the dialyzed bare DNs. a) An aliquot of the undialyzed positive control and the dialyzed samples were run on a 2% agarose gel in 0.5x TBE containing 11 mM Mg 2+ for 3 h at 65 V. The release of the constituent staple strands can be observed upon dialysis for bare DNs. b) An aliquot of the undialyzed positive control and the dialyzed samples were first incubated with 50 mM chondroitin sulfate (at 100x amine concentration) at 37 °C for 3 hours to remove the oligolysie shell ( Supplementary Fig. 9) and then ran on a 2% agarose gel in 0.5x TBE containing 11 mM Mg 2+ for 3 h at 65 V. Structural integrity of DNs was preserved when they were coated with K 10 and K 10 -PEG 5K . Figure 11| TEM of K 10 coated DNs dialyzed in zero-Mg 2+ buffer (5mM Tris, 1mM EDTA, pH 7). Negative stain TEM images of K 10 coated DN1-DN8 after being dialyzed into zero-Mg 2+ buffer overnight at room temperature. Scale bars are 500 nm for DN7 and 100 nm for all other panels. Figure 12| TEM of K 10 -PEG 5K coated DNs dialyzed in zero-Mg 2+ buffer (5mM Tris, 1mM EDTA, pH 7). Negative stain TEM images of K 10 -PEG 5K coated DN1-DN8 after being dialyzed into zero-Mg 2+ buffer overnight at room temperature. Scale bars are 500 nm for DN7 and 100 nm for all other panels. Figure 13| TEM of K 10 -PEG 5K coated DNs dialyzed in distilled water. Negative stain TEM images of K 10 -PEG 5K coated DN1-DN8 after being dialyzed into distilled water overnight at room temperature. Scale bars are 500 nm for DN2 and DN4 and 100 nm for all other panels. Figure 14| Nuclease degradation of DN1 by DNase I in RPMI-1640 at 37 °C is dependent on the overall Mg 2+ concentration. DN1 (final concentration 1 nM) was incubated with 1 U/mL of DNase I in RPMI-1640 buffer at 37 °C for different time points. RPMI-1640 buffer was either used as it is (contains 0.4 mM Mg) or was supplemented with additional Mg 2+ such that the final concentration is 10 mM. Incubated samples were then loaded on a 2% agarose gel and ran in 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 2 hours. Gel intensity can be used as a measure for structural degradation. We observe that nuclease degradation of DN1 by DNase 1 is Mg 2+ dependent. We hypothesize this is primarily due to the preservation of overall structure at higher Mg 2+ concentrations, and that low-salt denatured structures are much better substrates for serum nucleases, perhaps due to their more open and accessible configurations. Figure 15| Nuclease degradation of plasmid ds DNA (p5542) by DNase 1 in RPMI-1640 at 37 °C is independent of the overall Mg 2+ concentration. Plasmid p5542 (final concentration 1 nM) was incubated with 1 U /mL of DNase I in RPMI-1640 buffer at 37 °C for different time points. RPMI-1640 buffer was either used as it is (contains 0.4 mM Mg) or was supplemented with additional Mg 2+ such that the final concentration is 10 mM. Incubated samples were then loaded on a 2% agarose gel and ran in 0.5x TBE buffer containing 11 mM Mg at 65 V for 2 hours. Gel intensity can be used as a measure for structural degradation. In contrast to DN1, no effect of magnesium on the rate of degradation of p5542 was observed. Figure 16| Effect of length of oligolysine on the DNase I degradation of DN1. DN1 (final concentration 1 nM) coated with different oligolysines at N:P ratios depicted in the figure was incubated with 1 U/mL of DNase I in RPMI-1640 buffer at 37 °C for different time points. Incubated samples were then loaded on a 2% agarose gel and ran in 0.5x TBE buffer containing 11 mM Mg at 65 V for 2 hours. Gel intensity can be used as a measure for structural degradation. Higher N:P of oligolysine seems to offer better protection against nuclease degradation. With longer oligolysines, higher N:P can not be afforded as they lead to aggregation of the nanostructures.  Figure 18| Effect of size of conjugated PEG in K 10 -PEG nK on aggregation of structures. DN1 was coated with 1:1 N:P ratio of K 10 -PEG 1K / 5K or 20K and imaged on TEM using negative stain to check for aggregation. Scale bars are 500 nm.

Supplementary
Supplementary Figure 19| Deactivation of DNase I by EGTA and βmercaptoethanol in RPMI-1640 at 37 °C. p7308 (final conc 2 nM) was incubated with 50U/mL DNase I in RPMI-1640 at 37 °C for 1 hour. Different amounts of EGTA (ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-tetraacetic acid) and β-mercaptoethanol were added to the above samples to deactivate DNase I and prevent degradation of p7308 scaffold. The best deactivation of DNase I was observed with 5 mM EGTA and 10% βmercaptoethanol.
Supplementary Figure 20| Effect of size of conjugated PEG in K 10 -PEG nK on the nuclease degradation of DN1 in fresh culture medium at 37 °C. The stability of DN1 coated with K 10 -PEG 1K / 5K or 20K was tested. Samples were incubated with freshly prepared BMDC culture medium (RPMI-1640 supplemented with 10% FBS, 2mercaptoethanol and GM-CSF) for time intervals between 0h -48h and ran on a 2% agarose gel in 0.5x TBE buffer containing 11 mM Mg 2+ . Note that the nucleases were first deactivated using 5 mM EGTA and 10% β-mercaptoethanol and then the samples were incubated with 100x chondroitin sulfate prior to agarose gel to remove the PEG shell. All the PEG shells offer comparable stability towards nuclease degradation.
Supplementary Figure 21| Effect of K 10 -PEG 5K coating on the DNase I degradation of DN1 in RPMI-1640 at 37 °C. Bare DN1 (final concentration 1 nM) and DN1 coated with K 10 -PEG 5K at N:P of 1:1, 2:1 and 4:1 was incubated with different U/mL of DNase I in RPMI-1640 buffer at 37 °C. Incubated samples were then loaded on a 2% agarose gel and ran in 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 2 hours. Note that the PEG shells have not been removed in this experiment, as the DNAse I above 50U/mL could not be fully deactivated. Gel intensity can be used as a measure for structural degradation. All the PEG N:P ratios tested offer comparable stability towards DNase I degradation.
Supplementary Figure 22| Comparison of nuclease degradation of bare and K 10 -PEG 5K coated DN1 in mouse serum at 37 °C. DN1 (final concentration 1 nM) coated with K 10 -PEG 5K at N:P of 1:1 was incubated with mouse serum at 37 °C Time points are shown in minutes. Note that the PEG shells have not been removed in this experiment, as the mouse serum could not be fully deactivated. Gel intensity can be used as a measure for structural degradation. While the bare DN1 degraded very rapidly, K 10 -PEG 5K coated DN1 maintains structural integrity until the time tested, i.e. 3 hours. Scale bar is 100 nm. Figure 23| AGE and TEM characterization of DN9. DN9 was annealed using the protocol described in the methods. Three Mg 2+ concentrations (6mM, 8 mM, 10 mM) were tested during annealing, all of which gave comparable yields. 10µL of 10 nM crude samples were loaded onto a 2% agarose gel and ran in a 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 2 hours. All the Mg concentrations tested show comparable folding yield. Negative stain TEM of DN9 showed expected helical structures. The structures were stained with 2% uranyl formate. Scale bar is 100 nm.

Supplementary
Supplementary Figure 24| Stability of K 10 -PEG 5K coated DN9 against DNase1 degradation in RPMI-1640 at 37 °C. Bare DN9 (final concentration 1 nM) and DN9 coated with K 10 -PEG 5K at N:P of 1:1 was incubated with different U/mL of DNase I in RPMI-1640 buffer at 37 °C. Incubated samples were then loaded on a 2% agarose gel and ran in 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 2 hours. Note that the DNase I was first deactivated using 5 mM EGTA and 10% β-mercaptoethanol and then the samples were incubated with 100x chondroitin sulfate prior to agarose gel to remove the PEG shell. DN9 exhibited ~five fold greater nuclease resistance when coated with K 10 -PEG 5K . This result suggests that the nuclease resistance offered by K 10 -PEG 5K is in part due to steric shielding provided by the PEG layer that hinders binding of nucleases to the DNA. Figure 25| Stability of handles on DN1 towards nuclease degradation. The stability of handles DN1 was tested by placing the handles either on the outside or on the inside or at the core of the DN1 nanostructure. For the outside and inside handles, we incorporated either twelve 21mer ssDNA handles on the outside of the DN1 or twelve 21mer ssDNA handles on the inside of the DN1, to be bound by Cy5 labelled DNA anti-handles recruited from bulk solution. For the core handles, the 5´ of the inner staple oligos of DN1 were directly functionalized with Cy5 fluorophore. All the above DN1 samples were then coated with K 10 -PEG 5K at N:P of 1:1. Samples were incubated with freshly prepared BMDC culture medium for time intervals between 0h -24h and ran on a 2% agarose gel in 0.5x TBE buffer containing 11 mM Mg 2+ . Measurable differences in degradation are only observed at the 24h time point. The handles at the core of DN1 are the most stable followed by handles placed on the inside and finally handles placed on the outside. Error bars represent standard deviation (n=2). Figure 26| Oligolysine coating of DNs does not prevent accessibility of ss DNA handles. We assayed the ability of ssDNA handles protruding of the surface of DNs to interact with ssDNA anti-handles in solution. For this purpose, we incorporated 12 inward-facing 21mer ssDNA handles on the inside of DN1, to be bound by Alexa-750 labelled DNA anti-handles recruited from bulk solution. We incubated DN1 containing the ssDNA handles, bare versus K 10 coated versus K 10 -PEG 5K coated, with Alexa-750 labelled DNA anti-handle for 2 hours at 37ºC. The oligolysine coating was removed using 100x chondroitin sulfate to recover the original origami mobility. The samples were then loaded on a 2% agarose gel ran in 0.5x TBE buffer containing 11 mM Mg 2+ at 65 V for 2 hours. Intensity of the Alexa channel (red) was then compared for the 3 variants of DN1 and incorporation efficiency was calculated relative to the bare DN1. As a control to non-specific adsorption of the Alexa-750 on the DNA surface, DN1 with mismatched handle was also included as a negative control in the experiment. No Alexa-750 was incorporated when mismatched handles were used. Error bars represent standard deviation (n=2).

Supplementary Figure 27 | Uptake of DN6 into HUVE cells, monitored by FACS, as a function of cRGD ligands.
Each helix in DN6 that is coloured in yellow is labelled with 6 cRGD ligands using handle-antihandle binding. Left) Normalized fluorescence intensities of HUVE cells after 12 hour incubation with 1 nM Cy5-labeled DN6 with and without DNase I treatment. Blue bars represent cells untreated by DNase I (i.e. internally + externally associated particles) and red bars represent cells treated by 40 U of DNase I (i.e. internally associated particles only). Error bars represent standard deviation (n=3). Centre) Fluorescence distribution for HUVE cells incubated with 1 nM Cy5-labeled DN6 untreated by DNase I as measured by FACS. Right) Fluorescence distribution for HUVE cells incubation with 1 nM Cy5 labelled DN6 treated by 40 U of DNase I as measured by FACS. Figure 28| K 10 -PEG 5K coated structures are taken up by BMDC and the coating is non-toxic. Left: Bare and Right: K 10 -PEG 5K coated Cy5 labelled DN1 (5 nM, pink) was incubated with Bone-marrow derived dendritic cells (BMDC) for 12 hours (final concentration 5 nM DN1, therefore ~90 nM fluorophores). The Cy5 fluorescence intensity within the cells was measured using confocal microscopy at 12 hours postwashing. Figure 29| K 10 -PEG 5K coated structures are inside the cells and not stuck on the surface. K 10 -PEG 5K coated Cy5 labelled DN1 (5 nM, pink) was incubated with Bone-marrow dendritic cells (BMDC) for 12 hours, after which the medium was changed to remove DNs that were not taken up. The Cy5 fluorescence intensity within the cells was measured using confocal microscopy at 12 hours post-washing. Images (af) where a is at the top of the stack were collected at 0.2 µm intervals. Figure 30| K 10 -PEG 5K coated Cy5 labelled DN1 are mostly localized in vesicular compartments such as endosomes and endo-lysosomes. Left: DN1 (5 nM) and fluorescein-transferrin (green) was incubated with BMDC for 12 hours, after which nucleus-stain Hoechst (blue) was added and the sample was incubated for 1 hour prior to confocal imaging. Right: DN1 (5 nM) was incubated with BMDC for 12 hours, after which fluorescein-transferrin (green) and nucleus-stain Hoechst (blue) was added and the sample was incubated for 1 hour prior to confocal imaging. We observe that K 10 -PEG 5K coated Cy5 labelled DN1 are mostly localized in vesicular compartments similar to fluorescein-transferrin. Figure 31| Retention of Cy3 and Cy5 labelled DN1 within BMDC at 37 °C. DN1 (5 nM) was incubated with BMDC for 12 hours, after which the medium was changed to remove DNs that were not taken up. The Cy3 and Cy5 fluorescence intensity within the cells was monitored via confocal at 0h, 6h, 12h and 24h. Top panel : Cy5 fluorophore (pink), bottom panel : Cy3 fluorophore (green). Half-life of DN1 within BMDC is about 8 hours post-washing. Fluorescence intensity was calculated using ImageJ independently for both the channels and normalized to 0h time point. The error bars represent standard deviation. Figure 32| Schematic representation of the cross section of DN1 functionalized with 18 Cy3 (green stars) and 18 Cy5 (red stars) fluorophores and 12 biotins (yellow spheres). Biotin functionalization was used for the deposition of the DNs on a streptavidin coated glass surface. The samples were coated with K 10 -PEG 5K at an N:P of 1:1. In order to simulate a cellular environment, the samples were incubated in BMDC culture medium. After ~99% photobleaching of Cy5 fluorophore, a 102.5% ± 2.4 enhancement in Cy3 fluorescence was observed. Error bars represent standard deviation. Figure 33| Schematic representation of the cross section of DN1 functionalized with 6 Cy3 (green stars) and 6 Cy5 (red stars) fluorophores and 12 biotins (yellow spheres). Biotin functionalization was used for the deposition of the DNs on a streptavidin coated glass surface. a) Positive FRET control: The Cy3 and Cy5 fluorophores are within Förster distance and can exhibit FRET. b) Negative FRET control: The Cy3 and Cy5 fluorophores are beyond Förster distance and cannot exhibit FRET. The samples were coated with K 10 -PEG 5K at an N:P of 1:1. In order to simulate a cellular environment, the samples were incubated in BMDC culture medium. Enhancement of Cy3 signal after ~99% photobleaching of Cy5 fluorophore is only observed for the positive FRET system. Error bars represent standard deviation. Figure 34| Cy3 enhancement on Cy5 photobleaching within BMDC at 0h post-washing of cells. DN1 (5 nM) was incubated with BMDC for 12 hours, after which the medium was changed to remove DNs that were not taken up. The Cy3 and Cy5 fluorescence intensity within the cells post-washing before and after photobleaching of the Cy5 fluorophore at 0h. Figure 35| Cy3 enhancement on Cy5 photobleaching within BMDC at 6h post-washing of cells. DN1 (5 nM) was incubated with BMDC for 12 hours, after which the medium was changed to remove DNs that were not taken up. The Cy3 and Cy5 fluorescence intensity within the cells post-washing before and after photobleaching of the Cy5 fluorophore at 6h. Figure 36| Cy3 enhancement on Cy5 photobleaching within BMDC at 12h post-washing of cells. DN1 (5 nM) was incubated with BMDC for 12 hours, after which the medium was changed to remove DNs that were not taken up. The Cy3 and Cy5 fluorescence intensity within the cells post-washing before and after photobleaching of the Cy5 fluorophore at 12h. Figure 37| Cy3 enhancement on Cy5 photobleaching within BMDC at 24h post-washing of cells. DN1 (5 nM) was incubated with BMDC for 12 hours, after which the medium was changed to remove DNs that were not taken up. The Cy3 and Cy5 fluorescence intensity within the cells post-washing before and after photobleaching of the Cy5 fluorophore at 24h. Figure 38| Biodistribution profile of Alexa 750 labelled DN1 injected retro-orbitally in mice. Fluorescence images of mice injected with either (1) Alexa750 DNA oligo or (2) DN1 bare or (3) DN1 coated functionalized with 36 Alexa750 fluorophores at 2 minutes post injection. Alexa 750 fluorescence was monitored every 2 minutes for 1 hour using Perkin Elmer In Vivo Imaging System (IVIS). Figure 39| Distribution of fluorescence in the mice organs. The mice were injected with either (1) Alexa750 DNA oligo or (2) DN1 bare or (3) DN1 coated functionalized with 36 Alexa750 fluorophores. The mice organs were harvested and analyzed 1h post injection. K 10 -PEG 5K coated DN1 showed the highest accumulation in the liver. Error bars represent standard deviation (n=3). Figure 40| Fitting of the biodistribution data. The fluorescence intensity of the dye oligo, DN1 bare and DN1 coated was normalized to the to the maximum fluorescence intensity observed in the bladder for the dye oligo. The data of dye oligo and DN1 bare was fitted to an exponential using MATLAB. For DN1 coated , the data fit is not very reliable since it did not reach 100% saturation in the bladder at 60 minutes; we simply take the 50% normalized intensity as the provisional renal clearance half-life of ~45 min in this case. Figure 48|  In our cell experiments (HeLa, HUVE and BMDC) DNs were not found to be toxic at a maximum tested concentration of 10 nM.

Supplementary
We measured cell viability using the Thermofisher live/dead cell imaging kit. Supplementary Figure 49| Live/dead imaging of HeLa cells. Cells were incubated with 0.5 nM -10 nM DN in cell medium for 12 hours after which the live/dead dual probe was added. Cells were imaged using confocal microscopy at 488nm and 570nm.
Supplementary Figure 50| Live/dead imaging of HUVE cells. Cells were incubated with 0.5 nM -10 nM DN in cell medium for 12 hours after which the live/dead dual probe was added. Cells were imaged using confocal microscopy at 488nm and 570nm.

Supplementary Note 2
To validate that our FRET system could be used to monitor structural integrity, we measured the magnitude of Cy3 fluorescence enhancement induced by disruption of the arrangement between Cy3 and Cy5 dyes. For this initial experiment, the disruption was induced via treatment with 100 mM EDTA (Ethylenediaminetetraacetic acid) for 30 min at 37ºC, and fluorescence was monitored for particles in bulk solution. This treatment resulted in Cy3 fluorescence enhancement of ~700%. Conversely, for a negative control labelled with only Cy3, only 40% fluorescence enhancement was observed ( Supplementary Fig. 33). This result established reference behaviour for intact structures due to EDTA treatment of 700% Cy3 fluorescence. For example, a structure in which half the Cy3 dyes have lost their Cy5 partner, e.g. due to denaturation or degradation, would be expected to exhibit only 350% Cy3 fluorescence enhancement following EDTA treatment.
Next we examined a time course of DNAse I digestion, monitored via agarose-gel analysis, TEM analysis, and Cy3 fluorescence analysis, to observe how degradation of structure accumulated up to that time point correlated with Cy3 fluorescence enhancement induced by EDTA added after that time point (i.e. any loss of FRET induced by prior DNAse I degradation reduces the amount Cy3 fluorescence enhancement due to subsequent EDTA denaturation). We incubated DN1 coated with RPMI-1640 containing 100U/mL DNase I for different time points at 37ºC. In order to monitor the recovery of Cy3 fluorescence upon DNase I degradation, we first deactivated the DNase I using EGTA and -mercaptoethanol, stripped the K 10 -PEG 5K shell using chondroitin sulfate and then treated our samples with 100 mM EDTA for 30 min at 37ºC to induce denaturation. We note that at the t 1/2 for loss of fluorescence enhancement (6 hours in this system), structures were still intact as observed by AGE and TEM (Supplementary Fig 37). Unravelling of particles only was observed after ~70% of the fluorescence enhancement had been lost. This is consistent with a model in which DNA nanostructures can sustain many local lesions before global collapse occurs.
Supplementary Figure 51| DNs are intact at 50% fluorescence enhancement. a) DN1 labelled with 18 Cy3 fluorophores. b) DN1 labelled with 18 Cy3 and 18 Cy5 fluorophores. c) Cy3 fluorescence intensity before and after addition of EDTA (error bars represent standard deviation, n=2) d) DN1 coated with K 10 -PEG 5K was incubated with RPMI-1640 containing 100U/mL DNase I for different time points at 37ºC and analysed via AGE. In order to monitor the recovery of Cy3 fluorescence upon DNase I degradation, DNase I was first deactivated using 5 mM EGTA and 10%mercaptoethanol, the K 10 -PEG 5K shell was stripped using 100x chondroitin sulfate and then samples were treated with 100 mM EDTA for 30 min at 37ºC to induce denaturation. e) Percentage Cy3 enhancement upon addition of EDTA to the above samples. f) Negative stain TEM images of DN1 at 6h timepoint. At t 1/2 for loss of fluorescence enhancement (6 hours), structures were still intact as observed by TEM (crude samples were used for TEM imaging). This shows that DNA nanostructures can sustain many local cuts before global collapse occurs. Scale bars are 100 nm.

Supplementary Figure 52| Enhancement in Cy3 intensity upon EDTA denaturation
is only observed when the DNA nanostructure is intact. We did this experiment to confirm that enhancement in Cy3 signal within cells is only observed when the structures are intact and not when they are compacted in the endosomal compartments. Cy3 intensity before and after addition of EDTA is plotted for four samples. Only structurally intact nanostructures show enhancement in Cy3 signal upon denaturation whereas nanostructures compacted with polyethyleneimine do not. Error bars represent standard deviation (n=2). TEM images shows that PEI causes aggregation and irreversibly condenses the DNA nanostructure. The aggregation caused by PEI was not reversed even on addition of chondroitin sulfate.