Main

Tumour-targeting strategies are plagued by the lack of ideal tumour-specific membrane receptors1,2. The tumour necrosis factor (TNF) receptor superfamily (TNFRSF) plays essential roles in mammalian physiology3,4,5. However, the ubiquitous expression of TNFRSF receptors on human cells prevents the development of biomolecules exclusively targeting specific cell populations via them. Nevertheless, ligands or antibodies of specific TNFRSF members have been developed to activate the apoptosis machinery of cancer cells6,7. Recent studies, however, revealed that these ligands or antibodies lack efficacy due to their inability to precisely tune the oligomerization of the receptors on which the apoptosis cascades rely8,9,10,11.

By conjugating ligands to different scaffolds, an effective way to oligomerize certain members of TNFRSF can be achieved12,13,14,15. Furthermore, to tune the inter-ligand distance with molecular precision, we and others have recently designed flat sheet-like DNA origamis for patterning ligands of death receptor 5 (DR5) or Fas receptors, revealing that hexagonal ligand patterns with specific inter-ligand distance at the nanoscale substantially drive apoptosis16,17,18. However, these ligand patterns also interact with healthy cells. It is thus expected that, when applied in vivo, they risk promoting on-target, off-tumour toxicities, including neurotoxicity19, increased susceptibility to infections20 and immune disorders21.

Cells of solid tumours exorbitantly consume oxygen and other nutrients, causing low oxygen tension and anaerobic glycolysis that ultimately leads to increased acidity22. In this Article, to address the potential side effects of TNFRSF ligand patterns when used in vivo, we have developed a pH-sensitive three-dimensional DNA origami switch that can sense the high acidity of solid tumours. The device autonomously switches on surface display of six ligands as a sub-10 nm hexagonal pattern, while at higher pH this pattern remains hidden inside a cavity.

Design of the origami robotic switch

We designed the origami as an asymmetric double cylinder with a 24-nm-tall hollow head and a 15-nm-tall solid stem (Fig. 1a and Supplementary Fig. 1). The head has a 14-nm-deep cavity, where specific oligonucleotides (called ‘mini-scaffolds’ because they function as replacement for the scaffold in the origami design) are located to act as bridges between the typical origami staples and a binding site for ligand-decorated oligos. The distance between two adjacent mini-scaffolds is 6 nm. Each mini-scaffold is then used as a site to hybridize a peptide ligand-functionalized oligo, bearing one sequence that forms the double helix (the hybridization region) with the mini-scaffold as well as bearing an additional sequence of triplex-forming oligo (TFO).

Fig. 1: Design principles and simulations of the pH-responsive origami switch for ligand pattern hiding and display.
figure 1

a, A three-dimensional origami has a head part with a cavity that physically hides six peptide ligands conjugated to DNA strands. b, Illustration of the mechanism of how a peptide ligand hides inside the head cavity via the entropic spring imparted by the random coiling of the TFO. c, When the pH drops, the TFO of the peptide–DNA conjugate forms a tsDNA that forces the six peptides that were originally hidden in the cavity to get displayed as a hexagonal pattern on the top surface of the origami. d, Close-ups of the tsDNA. e, The mean structure of the head part and cavity (the whole mean structure is shown in Supplementary Fig. 2), calculated as an average from multiple oxDNA simulations. TFOs are coloured in cyan, and their 5′ termini are in orange. The top (left) and side (right) views are presented. f, Schematic of how the distance from the TFO terminus or the peptide to the origami surface plane is defined by two vectors. g, Real-time distances, from the TFO terminus (dots in orange) or from the peptide surface to the origami surface (dots in magenta, recalculated from the dataset of orange points via inclusion of the orientation information of the TFO terminus and the physical diameter of the peptide), along the simulation. h,j, Distance frequency histograms of the 5′ termini (h) or peptides (j) of the six TFOs to the origami surface. i,k, Probability that multiple 5′ termini (i) or peptides (k) of the six TFOs are simultaneously outside the origami. For g, h and j, positive values indicate that the tracked points are at the origami side of the defined plane (inside the origami cavity), while negative values mean that the tracked points are on the opposite side (outside the cavity).

Source data

Full size image

At lower pH, the TFO can wind along the duplex to form a triple-stranded DNA (tsDNA) that occurs when purines pair simultaneously from both their Watson–Crick and Hoogsteen interfaces (Fig. 1b). Via controlling the AT content, the triggering pH of tsDNA can be finely tuned23. Because of its high programmability, Ng et al. used this mechanism to spatially organize plasmids into raster-filled structures24. In addition, Sachenbacher et al. elucidated the structural basis of triplex domains and duplex–triplex crossovers in origami25. As we can customize the mini-scaffolds, we can subsequently control the responsive pH of tsDNA.

Our design intent was to create a structure that could display an apoptosis-driving ligand pattern in solid tumours (pH 6.5) while physically hiding ligands under normal physiological conditions (pH 7.4) (Fig. 1c). Previous studies have utilized tsDNA to switch the overall configuration of origami23,26,27,28. In these cases, the formation of tsDNA under low pH brings two domains of origami closer while the high pH releases them. For example, Ijäs et al. prepared the reconfigurable DNA origami nanocapsule26. However, the cargo got displayed under high pH while hidden under low pH, which is the opposite of what would be needed for drug exposure in acidic conditions. Our design reverses this.

Moreover, to reduce potential steric hindrance from the origami and improve the accessibility of the exposed hexagonal ligand pattern to corresponding receptors on cells, we introduced eight extra bases between the ligand and the TFO (Fig. 1d).

Ligand hiding estimation using molecular dynamics simulation

We performed molecular dynamics simulations using oxDNA29,30, to estimate if peptides can be hidden inside the origami cavity. Because of entropic spring effects, single-stranded DNA (ssDNA) tends to coil up rather than stay linearly stretched31. It indeed reveals that TFOs typically stay in coiled configurations, hidden in the origami (Fig. 1e and Supplementary Fig. 2).

To calculate the distances from the 5′ terminus of a TFO or from the conjugated peptide to the origami surface, during simulations, a plane on the origami surface is defined using two vectors, and the normal vector of the plane points towards the origami (Fig. 1f). Orientation of the TFO terminus and the van der Waal’s diameter (2.8 nm) of the peptide were introduced to calculate the relative distance between the peptide and the defined plane (Supplementary Fig. 3). The real-time distances indicated that, in most cases, both the 5′ termini of TFOs and the peptides remained inside the origami (Fig. 1g and Supplementary Fig. 4).

Each 5′ terminus can, with a low probability, transiently protrude outside the origami cavity (Fig. 1h), but the probability that more than three termini simultaneously end up at or above the surface is close to zero (Fig. 1i). The same holds true when also accounting for the added estimated position of the peptide (Fig. 1j,k). Our previous work with flat sheet-like origami indicated that, to be cytotoxic, more than three of the six hexagonally patterned peptides needed to be presented18. The very low probability for more than three peptides staying outside the origami thus predicts that, when the tsDNA does not form, our design can be assumed to not be cytotoxic, supporting our aim to use pH as a regulator for cytotoxicity.

Origami characterization

Origami was stabilized using the ultraviolet (UV) cross-linking method32,33. We placed extra thymidines (Ts) at the termini and crossovers of staples, creating sites for cyclobutane pyrimidine dimer bonds. The ssDNA regions of the mini-scaffolds were designed to not contain any Ts and Cs. Samples, before and after cross-linking, under transmission electron microscopy (TEM) similarly showed the expected homogeneous particles (Supplementary Figs. 5–7). In 1× phosphate-buffered saline (PBS), the original origami showed some tendency to disassemble (Supplementary Fig. 8), while the cross-linked one kept its shape (Supplementary Figs. 9 and 10).

The cryogenic electron microscopy (cryo-EM) data of the cross-linked origami yielded an intermediate-resolution reconstruction of 7 Å, sufficient to confirm the structural consistency between the experimentally observed and designed origami (Fig. 2a–c and Supplementary Fig. 11). The terminal regions of the individual helices could not be seen due to their extensive flexibility. However, the flexibility of the core part could be quantified and visualized using the 3DFlex tools in CryoSPARC34,35. This analysis showed the continuous flow of helices spanning across ~20 Å distance transversally (Fig. 2d), while no considerable flexing was observed longitudinally.

Fig. 2: Structure characterization of the UV cross-linked origami and its stability.
figure 2

a, Example of cryo-EM micrograph of the UV cross-linked origami structure. Scale bar, 100 nm. b, Representative 2D class averages of the UV cross-linked origami structure. c, Cryo-EM density map. d, Frames picked from the 3DFlex analysis. e, A 2% agarose gel electrophoresis of origami incubated in freshly prepared DMEM with 10% FBS at 37 °C. L, 1 kb DNA ladder; M, DMEM with 10% FBS; C, origami before adding to the medium. f, Normalized agarose gel intensity versus time for origamis. Data are presented as mean ± s.d., n = 2 independent measurements. g, Protrusion sites (coloured in blue) surrounding the origami for site-specific PEGylation. h, The PEGylation mechanism, via hybridization of ssDNA–PEG5K to 20 single-stranded protrusions of the origami. i, Electrophoresis of origamis in a 2% agarose gel. L, ladder; S, p8064 scaffold; UV, UV cross-linked origami, six-peptides-loaded device (pH_O6p) and PEGylated version (PEG_pH_O6p). j, Negative-stain TEM image of the PEG_pH_O6p. Scale bar, 100 nm. k, Two-dimensional class averages of the PEG_pH_O6p. Scale bar, 100 nm.

Source data

Full size image

Stability tests in cell culture medium showed that around 30% of the cross-linked origami survived up to 48 h (Fig. 2e,f), indicating that the origami would be reliable for hours of cellular experiments.

We designed 20 ssDNA protrusions surrounding the origami for the docking of ssDNA–polyethylene glycol (PEG)5K polymers (Fig. 2g,h). PEGylation is expected to improve the stability and circulation half-time of DNA origami when applied in vivo36,37. Our site-specific PEGylation here renders each origami a much lower PEG5K coating density than the previous oligolysine-mediated PEGylation37. We hypothesized, though, that a site-specific PEGylation could help us avoid interference in TFO. The PEGylated UV-cross-linked origami migrated slower than the bare origami (Fig. 2i) on gel, as expected following successful PEGylation. TEM imaging (Fig. 2j) and averaged 2D classes (Fig. 2k) agreed with its morphology.

pH-dependent tsDNA origami switch

We detected the ssDNA regions of mini-scaffolds using Cy5-tagged complementary oligo. It gave a stepwise pattern corresponding to the abundance of mini-scaffold per origami (Supplementary Fig. 12), indicating that mini-scaffolds were integrated.

We then introduced TFO to the mini-scaffold. Cy5 and Cy3 were tagged to the 5′ end and 3′ end, respectively, of the longer oligo that was hybridized to the mini-scaffold, allowing us to follow the switch-on and switch-off of the tsDNA via measuring Förster resonance energy transfer (FRET) (Fig. 3a). Its switch-on takes the 5′ end of the oligo to the surface of the origami, bringing the donor chromophore and acceptor chromophore in proximity. It showed that, when the AT content was 50%, the tsDNAs were mostly in the switch-on state at pH below 6.8 while they were in the switch-off state at pH 7.4 (Fig. 3b,c and Supplementary Fig. 13), demonstrating their responses to the extracellular acidity (pH 6.5–6.8) of solid tumours. When we changed the pH to 7.4 and 6.5, it showed a reversible change between the switch-off and switch-on states over several consecutive rounds (Fig. 3d).

Fig. 3: pH-sensitive FRET and peptide quantification.
figure 3

a, The principle of using FRET to verify the pH-dependent formation of the tsDNA inside the DNA origami. When a tsDNA forms, the donor chromophore (Cy3) and acceptor chromophore (Cy5) get in proximity, showing FRET. The sequence of the tsDNA is indicated to the right. b, Changes of normalized (to FRET efficiency under pH 7.4) FRET efficiency ratio in buffers with different pH (n = 3 independent measurements). c, With an excitation of 520 nm, the emission spectrum of the origami samples under pH 7.4 or 6.5. d, Normalized (to FRET efficiency under pH 7.4) FRET efficiency ratio when we consecutively exchanged the origami samples between buffer with pH 7.4 and 6.5 (n = 2 independent measurements). e, Illustration of how we quantify peptides in origami. After loading the ssDNA–peptide conjugate into the origami, the peptide was detected using Cy5-labelled oligo that targets the TFO of the conjugate. f, Peptide quantification of origami structures with different numbers of mini-scaffold with the principle indicated in e (n = 3 independent measurements). RFU, relative fluorescence units. Data are presented as mean ± s.d.

Source data

Full size image

Quantification of ssDNA–ligand conjugates in origami

We conjugated the 5′ end of TFO with the TNF-related apoptosis-inducing ligand (TRAIL)-mimicking peptide ligand of DR5 (Supplementary Fig. 14). The affinity of the conjugate to DR5 stayed similar under pH 7.4 (KD = 37.8 nM) or 6.5 (KD = 52.8 nM), supported by surface plasmon resonance (SPR) measurements (Supplementary Fig. 15). Via hybridization, we assembled the conjugates into the cross-linked origami.

To verify the stoichiometry, we prepared origamis with different numbers of mini-scaffold and incubated them with the conjugate. We used a Cy5-oligo to target the TFO (Fig. 3e), showing that the fluorescence increased stepwise with the corresponding increase in mini-scaffolds (Supplementary Fig. 16). The analysis showed a clear correlation between the designed number of mini-scaffolds per origami and the fold change of the Cy5 signal (Fig. 3f), supporting that the peptide abundance per origami followed our design. Origami switch, containing six peptides, is referred to as pH_O6p. Its PEGylated version is referred to as PEG_pH_O6p.

pH-driven ligand pattern induces DR5 clustering

The affinity (to DR5) of the origami always displaying the peptide pattern on surface was slightly higher than the case of non-structured peptides (Supplementary Fig. 17), which we hypothesize is positively, and negatively, contributed to by: (1) the patterning of peptides and (2) the large molecular mass of origami, respectively. We performed a pH-controlled cell experiment to verify PEG_pH_O6p (Fig. 4a). SK-BR-3 cells were treated with Cy5-labelled PEG_pH_O6p structures for 4 h and then analysed using flow cytometry. Cells at pH 6.5 showed a notably higher Cy5 signal than did cells at pH 7.4 (Fig. 4b and Supplementary Fig. 18). This confirmed that the peptides were presented outside the origami when the pH was low enough to form tsDNA, mediating origami structure to cells. Meanwhile, when it was under pH 7.4, cells incubated with empty origami had similarly low signals as those treated with PEG_pH_O6p, indicating the absence of DR5-mediated interactions and that peptides can be reliably hidden under pH 7.4.

Fig. 4: Interaction of the origami with breast cancer cells and induction of DR5 clustering.
figure 4

a, Experimental setup of cancer cell treatment with Cy5-labelled origami at pH 7.4 or 6.5. b, The Cy5 signal of SK-BR-3 cells treated with 5 nM empty origami, PEG_pH_O6p at pH 7.4 or 6.5, measured by flow cytometry. c, Localization of Cy5-labelled origamis (magenta) relative to DR5 (cyan; detected using Alexa488-labelled anti-DR5 monoclonal antibody) on SK-BR-3 cells. The co-localization of the magenta and cyan is shown in grey. Maximum intensity projections from z-stacks. Scale bars, 20 μm.

Full size image

We previously found that with the treatment of a sub-10 nm hexagonal pattern of peptide ligands, DR5 clustered into structurally diverse assemblies in the membranes of apoptotic cells18. We thus examined if the same effect was achieved with the origami robotic switch. This revealed that, with SK-BR-3 cells, treatment with PEG_pH_O6p under pH 6.5 caused notably more clustering of DR5 than the treatment with empty origami or PEG_pH_O6p under pH 7.4 (Fig. 4c). Under pH 6.5, the co-localization of signals from origami and DR5 clusters on cells supported that the receptor clustering was induced by the origami treatment. Furthermore, the morphology of cells treated with PEG_pH_O6p under pH 6.5 showed shrinkage, which is one of the typical characteristics of apoptosis.

Based on a previous study38, we developed a repair qPCR protocol (Supplementary Fig. 19) to quantify origami associated with cells. It showed that, as expected, the total amount of PEG_pH_O6p per cell was 16-fold higher under pH 6.5 versus both pH 7.4 and the case of empty origami. We also analysed intracellular origami with the same protocol following enzymatic DNA degradation of the extracellular material. This similarly showed a notable increase in internalization at low pH. Together, these assays demonstrated that the tumour microenvironment (TME)-corresponding acidity promoted the presentation of the ligand pattern from the origami, which in turn resulted in the binding to and clustering of DR5.

Cytotoxicity of the origami switch is pH dependent

We first assessed if the pH itself could impact cell growth. There were no viability differences between the cells maintained under the original cell culture medium and medium at pH 7.4 or 6.5. SK-BR-3 cancer cells treated with PEG_pH_O6p at pH 7.4 did not show an impact on cell viability. In contrast, cell viability at pH 6.5 continuously decreased after a 4 h incubation. After 24 and 48 h, less than 20% and 10% of cells survived, respectively (Fig. 5a). The further analysis on apoptosis demonstrated that substantially more cells incubated with PEG_pH_O6p at pH 6.5 were in an early (46.8% versus 5.8%) or late (32.2% versus 4.5%) apoptotic state compared with cells incubated with PEG_pH_O6p under pH 7.4, empty origami or control (Fig. 5b,c). We thus verified that the PEG_pH_O6p switched on its apoptosis-inducing and cytotoxic bioactivity in a pH-dependent manner.

Fig. 5: Cell cytotoxicity is controlled by acidic conditions.
figure 5

a, Cell viability assay on SK-BR-3 cancer cells after treatment with PEG_pH_O6p origami at pH 7.4 or 6.5. b,c, Cancer cell apoptosis assay of DNA origami treatment at pH 7.4 or 6.5, measured by flow cytometry (Q1, necrotic cells; Q2, late apoptotic cells; Q3, early apoptotic cells; Q4, live cells). d–g, Cell viability assay at pH 7.4 of breast cancer cells (d) and human non-cancer cells, including T lymphocytes (e), kidney cells (f) and endothelial cells (g). n = 3 independent experiments. Data are presented as mean ± s.d.

Source data

Full size image

When considering in vivo treatment, the origami would interact with circulating immune cells or cells in healthy organs and tissues before accumulating in TME, and that DR5 is expressed in almost all human cells. If PEG_pH_O6p cannot keep the cytotoxic ligand pattern hidden before reaching TME, it might cause unwanted side effects. At pH 7.4, apart from cancer cells (Fig. 5d), we tested the device for safety with human T lymphocytes (Fig. 5e), embryonic kidney cells (Fig. 5f) and cerebral microvascular endothelial cells (Fig. 5g). We also fabricated origami samples in which peptides were always hidden (negative control) or always displayed (positive control) as controls. Viability profiles of all these cells showed similar trends. Treatments using the negative control and treatment with PEG_pH_O6p did not impact on cell viability, while cells in the positive control were decreasingly viable. This supports the notion that the designed peptide pattern can generally induce apoptosis-based cytotoxicity of human cells rather than being specific for cancer cells. Importantly, PEG_pH_O6p showed similar effects as the negative control, demonstrating its biosafety before exposure to TME.

Solid tumours get suppressed by PEG_pH_O6p

We tested the origami switch on nude mice implanted with SK-BR-3 xenografts. We administered the origami samples via intravenous (i.v.) or intratumoural (i.t.) injection.

Biodistributions were assessed via tracking Alexa750-modified origami. Following the i.v. injection, the PEG_pH_O6p spread throughout the whole body and displayed a rapid renal clearance (Fig. 6a). Its dominant signal accumulated in the bladder, where it reached a maximum after 1 h, and was excreted thereafter. Some liver accumulation was also noticeable. These results are similar to origami biodistribution experiments in a previous study using mice without tumours37. We observed no signs of selective accumulation in the solid tumour, which could be explained by the lack of tumour-targeting motifs in the origami. For the i.t. injection, origami signals mainly stayed in the tumour for more than 8 h, although they slowly decreased over time (Fig. 6b). Renal clearance was also evident, with the maximal signal in bladder after 2 h, but there was no strong accumulation in liver.

Fig. 6: In vivo biodistribution and anti-tumour effects.
figure 6

a, Biodistribution imaging after i.v. injection with Alexa750-labelled PEG_pH_O6p origami in nude mouse bearing a SK-BR-3 tumour at different timepoints. The tumour is indicated using a dashed circle. b, Imaging after i.t. injection with Alexa750-labelled PEG_pH_O6p origami in nude mouse bearing a SK-BR-3 tumour at different timepoints. The tumour is indicated using a dashed circle. c, Treatment schedule and SK-BR-3 tumour growth curves of mice receiving i.v. injections of PBS (n = 2), PEG_mutated_O6p (non-switchable control; n = 4) or PEG_pH_O6p (n = 4). d, Treatment schedule and SK-BR-3 tumour growth curves of mice receiving i.t. injections of PBS (n = 2), PEG_mutated_O6p (n = 4) or PEG_pH_O6p (n = 4). Statistical analysis was performed with one-way ANOVA followed by Tukey post-tests. Data are presented as mean ± s.d.

Source data

Full size image

Finally, we treated mice with PEG_pH_O6p to suppress the implanted tumours. Apart from the PBS group, PEG_mutated_O6p was included as another control group. To get the PEG_mutated_O6p control, the TFO sequence of PEG_pH_O6p was mutated, making PEG_mutated_O6p almost chemically identical to the switchable PEG_pH_O6p, but just the minute sequence mutations in the TFO region renders the control structure not sensitive to pH and, thus, always hiding peptides inside the origami.

Using i.v. administrations, the PEG_pH_O6p started showing its effects after three injections, and it gradually slowed down the growth of tumour volumes by around 30% by the endpoint (Fig. 6c and Supplementary Fig. 20). For i.t. injections, the PEG_pH_O6p suppressed solid tumours. Volumes of tumours receiving PEG_pH_O6p substantially decreased; compared with the control groups, its tumour suppression efficacy was around 70% (Fig. 6d and Supplementary Fig. 21). With both routes of administration, tumours receiving PEG_mutated_O6p treatments grew similarly to those in mice receiving PBS treatment, supporting that the anti-tumour mechanism of PEG_pH_O6p relied on its pH sensitivity. Biomarkers of apoptosis, cleaved caspase-3 and cleaved caspase-8, of the tumours treated with PEG_pH_O6p treatments were more abundant (Supplementary Fig. 22). Altogether, these treatment results demonstrated the apoptosis induction-based therapeutic effects of the DNA origami nanorobotic switch in solid tumours.

Conclusions

We report a DNA origami switch that senses TME and then displays a cytotoxic ligand pattern. This method of autonomous actuation of cytotoxic ligand patterns could improve cancer treatment efficacy and reduce side effects, raising the possibility of developing new drugs. The addressability of DNA origami enables the precise patterning of biomolecules39,40,41. It has gained increasing interest for use in deciphering mechanisms of biology. However, limiting these patterns to specific conditions is difficult. Reconfigurable DNA nanostructures, which expose ligands upon opening, can help, but they face challenges in achieving high yields in both preparation and activation. Our pH-sensitive tsDNA-based origami switch overcomes these issues, showing substantial effectiveness in inducing apoptosis under acidic environments.

Our strategy aims to reduce on-target/off-tumour side effects, confirmed in vitro using non-cancer cells such as human T lymphocytes and others at pH 7.4. However, in vivo verification in mouse models is challenging due to species-specific peptide ligand binding. In our study, the peptide ligand did not affect non-tumour cells in BALB/c nude mice xenografts, but future studies should investigate this in humanized mouse models for a more accurate assessment.

DNA origami production costs are now reduced due to enzymatic and bacteriophage-based staple production, bringing the price to about $200 g−1 (refs. 42,43). Although our study used more expensive solid-phase synthesized staples for prototyping, using these cost-effective techniques can lower the cost for scale-up. Additionally, the recent availability of affordable peptide synthesis will further aid in scaling up DNA nanorobotics production for therapeutic tests in the future.

Methods

DNA origami preparation

DNA origami folding and purification

We used the phage-amplification-in-bacteria protocol to produce p8064 scaffold DNA18. Staple oligonucleotides were ordered from Integrated DNA Technologies (see ‘DNA oligo sequences of the origami.xlsx’ in Supplementary Data). A standard 100 μl folding mixture contained 20 nM p8064 scaffold DNA, 100 nM of each staple oligonucleotide, 22 mM MgCl2, 5 mM Tris and 1 mM EDTA. DNA origami formed during an annealing programme from 60 °C to 24 °C in 14 h. We used the PEG precipitation method described by previous papers44 to purify DNA origami. We resuspended the DNA origami pellet with the buffer containing 5 mM MgCl2, 5 mM NaCl, 5 mM Tris and 1 mM EDTA at 37 °C on a shaker, at 300 rpm, for 4 h.

DNA origami irradiation with UV

We adjusted concentrations of DNA origami to be at or below 100 nM. On an ice bath, we irradiated samples with UV light at 310 nm wavelength for 2 h. After the UV irradiation, we washed the DNA origami with Amicon 100K filter tubes, using the buffer containing 5 mM MgCl2, 5 mM NaCl, 5 mM Tris and 1 mM EDTA, four rounds (7,000g for 1.5 min per round).

Peptide–ssDNA conjugate

We purchased ssDNA with a dibenzocyclooctyne modification at its 5′ end from Biomers (Supplementary Data). The conjugation of the peptide to the DNA was performed via copper-free click chemistry in 1× PBS. In the reaction, the molar ratio of peptide to ssDNA was 50 to 1. We purified the conjugate by using a proFIRE system (Dynamic Biosensors).

The affinity of the peptide–ssDNA conjugate to DR5

The binding kinetics of the peptide–ssDNA conjugates to DR5 were studied by SPR using a Biacore T200 instrument (GE Healthcare). First, streptavidin (Sigma-Aldrich, S4762) was dissolved in 10 mM sodium acetate pH 4.5 and immobilized on a CM3 chip (GE Healthcare, BR100536) via amine coupling following the manufacturer’s instructions. The biotinylated anchor oligonucleotides, complementary to the peptide–ssDNA conjugates, were captured at 150 RU by injecting them over the streptavidin surface for 660 s at a flow rate of 10 µl min−1, followed by a 60 s wash with running buffer 1× HBS-EP+ (GE Healthcare, BR100669). The peptide–ssDNA conjugates were immobilized by DNA complementarity at 300 RU by injecting them at a flow rate of 5 µl min−1 and washing the excess of conjugates for 60 s at 30 µl min−1 with the same running buffer. As a negative control, a parallel flow cell was modified with biotinylated anchors but without peptide–ssDNA. The DR5 analyte (R&D Systems, 10140-T2) was diluted from 516 nM to 32.25 nM and injected as a single-cycle kinetics experiment, with a flow rate of 30 µl min−1, and 500 s contact time for each concentration. After the last concentration, the dissociation phase lasted 1,000 s at the same flow rate, and the surface was regenerated with two 30 s injections of 50 mM NaOH. The single-cycle kinetics runs were carried out in HBS-EP+ buffer at pH 7.4 or 6.5, for both the empty negative flow cell and the peptide–ssDNA modified flow cell. The sensogram data were analysed with the BIA-evaluation 3.0 software assuming a 1:1 Langmuir binding model.

The affinity of the peptide origami to DR5

This interaction was also analysed by SPR. As described for the peptide–ssDNA conjugates, streptavidin and biotinylated oligonucleotides, complementary to the anchoring oligonucleotides in the origami, were immobilized on a CM3 chip. By DNA complementarity, the surface was saturated with the peptide origami at around 600 RU by injecting it at flow rate of 10 µl min−1, and washing for 60 s at 30 µl min−1 with the same running buffer. As a reference sample, a parallel reference flow cell was saturated with empty origami. The DR5 analyte (R&D Systems, 10140-T2) was diluted from 1,032 nM to 64.5 nM and injected as a single-cycle kinetics experiment, with a flow rate of 30 µl min−1, and 500 s contact time for each concentration. After the last concentration, the dissociation phase lasted 1,000 s at the same flow rate, and the surface was regenerated with two 30 s injections of 50 mM NaOH. Three replicates of the single-cycle kinetics were run with HBS-EP+ buffer supplemented with 5 mM MgCl2 at pH 7.4, for both the peptide origami and the empty origami flow cells. The sensogram data were analysed with the BIA-evaluation 3.0 software assuming a 1:1 Langmuir binding model.

Peptide–ssDNA conjugate and PEG5K–ssDNA hybridization to DNA origami

A 20-fold excess of conjugate to each mini-scaffold on the DNA origami was mixed with UV-irradiated DNA origami. We kept the mixture under 37 °C for 1 h followed by room temperature overnight. To wash away the free peptide–ssDNA conjugate from DNA origami, we used Amicon 100K filter tubes, with six rounds of dilution and concentration using 1× PBS (7,000 rcf for 1.5 min per round). To modify the surface of the origami with a certain number of PEG5K, we added an excess of PEG5K–ssDNA polymers to the origami with ssDNA protrusions on its surface (the molar ratio of ssDNA–PEG5K to each protrusion on the origami was 5 to 1). The structures were finally purified with 1× PBS using Amicon 100K filter tubes.

DNA origami characterization

Gel electrophoresis

We prepared 2% agarose gels in 0.5× Tris-borate-EDTA (TBE) buffer supplemented with 10 mM MgCl2 and 0.5 mg ml−1 ethidium bromide. On an ice bath, we ran electrophoresis in 0.5× TBE buffer supplemented with 10 mM MgCl2 at 90 V for 3 h. Gels were imaged under a GE LAS 4000 imager.

Negative-stain TEM

We mixed 8 µl of 1 M NaOH with 400 µl of 2% (w/v) uranyl formate. The mixture was then immediately centrifuged at 16,500g for 5 min. We added 5 µl of 5 nM DNA origami to a glow-discharged, carbon-coated, formvar resin grid for 1 min. We then blotted the sample drop away with filter paper before we applied 5 µl of uranyl formate solution to the grid for 40 s. The sample was air-dried and imaged in a 120 kV LaB6 microscope (Talos 120C G2 with Ceta-D detector) at different magnifications. The particle class-averaging processing was carried out in Scipion.

Cryo-EM

Vitrobot Mk4 (FEI) was used to prepare cryo specimens. We used the Quantifoil R 1.2/1.3 (LF, 200 Mesh, Gold) grid. The grid was glow-discharged under 25 mA for 30 s. We applied 3 µl of the highly concentrated (>500 nM) DNA origami solution to the grid. The grids were incubated at a relative humidity of 100% for 1–5 min and frozen in liquid ethane after blotting (3 s, drain time 1 s). The grids were kept in liquid nitrogen until imaging. Cryo-EM data were collected with EPU software using a Krios G3i TEM operated at 300 kV. Images were acquired in 81 kx nanoprobe EFTEM SA mode with a slit width of 10 eV using a K3 Bioquantum. Exposure time was 4.4 s, during which 45 movie frames were collected with a fluency of 0.90 e Å−2 per frame. Image pre-processing until particle picking was carried out in Warp45, with subsequent processing using CryoSPARC34. An outline of the data processing pipeline can be found in Supplementary Fig. 17. For density visualization only, the reconstructed map was post-processed using DeepEMhancer46 using tight mask target settings and dust hiding was performed using UCSF ChimeraX47. A comparison between the original and post-processed maps can be found in Supplementary Fig. 11.

Stability of the origami in cell culture medium

Five microlitres of 100 nM origami was mixed with 95 μl of Dulbecco’s modified Eagle medium (DMEM) containing 10% foetal bovine serum (FBS). The samples were kept in a humidified incubator containing 5% CO2 at 37 °C. At different timepoints, the samples were checked via 2% agarose gel electrophoresis in 0.5× TBE buffer supplemented with 10 mM MgCl2 at 90 V for 3 h. Gels were imaged under a GE LAS 4000 imager.

Quantification of mini-scaffold and peptide on origami

Origami structures containing different numbers of peptides were incubated with the Cy5-labelled oligonucleotides complementary to the mini-scaffold. The molar ratio of the Cy5-labelled oligo to each mini-scaffold of origami was 20 to 1. After purification, concentrations of origami samples were set to 10 nM. Each sample’s Cy5 signal of 100 µl was measured via a multimode microplate reader (Varioskan LUX).

Coarse-grained molecular dynamics simulation of the origami

The origami design file from Cadnano was converted to oxDNA format using tacoxDNA48. The oxDNA simulation started with a relaxation step for removing possible overlapping nucleotides, followed by a molecular dynamics simulation run to reduce overstretched bonds. Afterward, the structure was simulated for 1 × 108 steps, with a time step of 0.005 oxDNA time units. The simulations were run with the oxDNA2 model at 30 °C with a salt concentration of 0.15 M and an Anderson-like thermostat. Every 2 × 104 steps were saved as one simulation state. The mean structure was calculated using the oxDNA analysis tool29,30.

Cell experiments

Cell culture

Immortalized human cell lines, including SK-BR-3 (ATCC, HTB-30), HBEC-5i (ATCC, CRL-3245) and HEK293 (ATCC, CRL-1573), were purchased. SK-BR-3 cells were cultured in DMEM high glucose (Sigma-Aldrich, D5796) containing 10% FBS (Sigma-Aldrich, F2442) and 100 units ml−1 penicillin–streptomycin (Sigma-Aldrich, P4333) in a humidified incubator containing 5% CO2 at 37 °C. For the culture of HEK293, Eagle’s Minimum Essential Medium (ATCC, 30-2003) supplemented with 10% FBS and 100 units ml−1 penicillin–streptomycin was used; for the culture of HEEC-5i, the flasks or plates were pre-coated with 0.1% gelatin (ATCC, PCS-999-027) for 1 h. Cells were then seeded on the flasks and cultured in DMEM: F12 (ATCC, 30-2006) supplemented with 40 µg ml−1 endothelial growth supplement (Sigma, E2759) and 10% FBS.

Confocal assay

A total of 2.5 × 104 cells per well were seeded in an eight-well Millicell EZ SLIDE for 24 h. Before adding origami, the medium was replaced by fresh medium with a pH of 7.4 or 6.5. The final concentration of Cy5-labelled origami in 0.5 ml of medium was 5 nM. After a 4 h incubation in a humidified incubator containing 5% CO2 at 37 °C, cells were fixed with 4% formaldehyde. To detect the DR5, the cells were incubated with 66 nM human TRAIL R2/TNFRSF10B antibody (R&D Systems, MBA6311) overnight at 4 °C. The next day, goat anti-mouse IgG H&L (Alexa Fluor 488; Abcam, ab150113), at a concentration of 2 µg ml−1, was used as the secondary antibody to visualize the bound primary antibody against DR5. After staining with Fluoroshield Mounting Medium with 4′,6-diamidino-2-phenylindole (DAPI; Abcam, ab104139), we imaged the cells under LSM700 (Zeiss).

Flow cytometry

For the quantitative assay of origami binding to cells, 1 × 107 SK-BR-3 cells per well were cultured in six-well plates overnight. The medium of each well was replaced by 2 ml of fresh medium with pH 7.4 or 6.5 containing 5 nM of Cy5-labelled origami structures for 4 h treatments in a humidified incubator containing 5% CO2 at 37 °C. Cells were collected and washed twice with cold 1× PBS, followed by a flow cytometry measurement. For the cell apoptosis assay, with six-well plates, 1 × 107 SK-BR-3 cells were seeded in each well for culture overnight. The medium was removed, and 2 ml of fresh medium with pH 7.4 or 6.5 containing 5 nM origami was added to the cells for 24 h treatments. All cells (including detached cells in the medium) were collected and resuspended in cold 1× PBS. Cells were then stained with annexin V-fluorescein isothiocyanate (FITC) and propidium iodide (PI), sequentially, according to the protocol of a commercial apoptosis assay kit (Thermo Fisher Scientific, V13242). Cells were analysed using the Cytek Aurora flow cytometer.

Origami quantification with repair qPCR

A total of 1 × 107 SK-BR-3 cells per well were cultured in four-well plates overnight. The medium was replaced by 2 ml of fresh medium with pH 7.4 or 6.5 containing 5 nM origami structures for 4 h treatments. The cell layer in each well was gently washed with cold 1× PBS twice, which was followed with or without 5 U μl−1 Benzonase (Invitrogen) treatment. After cell washing and lysing using 100 μl DNAzol Reagent (Life Technologies), DNA pellets were collected according to a previously established protocol38. In 1× PBS, the DNA samples were denatured under 95 °C for 30 min. Then, the PreCR Repair Mix (New England Biolabs) was used according to the kit protocol to repair the DNA samples. The rationale for this step is that the structures were both UV-cross-linked and subjected to very high temperatures. The Luna Universal qPCR Master Mix (New England Biolabs) and the primer pair (forward: CTGGCTCGAAAATGCCTCT, reverse: ACCAGTATAAAGCCAACGCT) were used together to quantify the scaffold of origami in the samples.

Cell viability assay

A total of 5 × 104 cells per well were seeded on 96-well opaque white polystyrene microplates (Fisher Scientific, 10583534) and cultured for 24 h. The cell culture medium was replaced with 95 μl of fresh medium with pH 7.4 or 6.5, followed by the addition of 5 μl of 100 nM DNA origami. At different timepoints, the plate was transferred out of the culture incubator and incubated at room temperature for 20 min. To measure the cell viability, the CellTiter-Glo Luminescent Cell Viability Assays (Promega, G7571) were used. The luminescence was recorded on a multimode microplate reader (Varioskan LUX). The luminescence of the wells containing medium without cells was used as the background control. The following equation was used to calculate the relative cell viability: % viable cells = (luminescencesample − luminescencebackground)/(luminescencePBS − luminescencebackground) × 100.

Animal experiments

Ethics permit

All animal handling and experimental procedures were carried out according to local ethics guidelines and approved by Stockholm’s Animal Experimentation Ethics Committee (Stockholms djurförsöksetiska nämnd, Dnr 16041-2019).

Mouse model

Mice were ordered from Charles River and bred at the animal facility Comparative Medicine Biomedicum (KMB), Solna Campus, Karolinska Institutet. Female BALB/c Nude CByJ.Cg-Foxn1nu/J mice at the age of 29 days were used to establish the SK-BR-3 xenograft model. A total of 5 × 106 SK-BR-3 cells were subcutaneously inoculated into the upper left flank of mice in 0.1 ml of Matrigel (Corning, 354234) diluted in cold PBS (Matrigel:PBS, 1:1). After 1 week, mice were randomized into treatment groups for experiments.

Biodistribution

The Alexa750-labelled origami was administered to the mice intravenously into the tail (100 μl, 100 nM) or intratumourally (20 μl, 200 nM). To track the biodistribution of the origami, the live mice were imaged at different timepoints using a PerkinElmer in vivo imaging system.

Anti-tumour efficacy assay

PBS or origami (100 μl of 100 nM origami for i.v. injection; 20 μl of 200 nM origami for i.t. injection) was administrated according to the therapy regimes. We measured the sizes of tumours after each injection with a caliper and calculated the tumour volume using the following equation: tumour volume (mm3) = (length × width2) × ½.

Statistics and reproducibility

For quantitative analysis, a minimum of three biological replicates were analysed excluding the in vivo data of PBS (n = 2). Data were analysed by ordinary one-way analysis of variance (ANOVA) with Tukey post-tests and Student’s t-test. A P value <0.05 was considered statistically significant. No statistical method was used to pre-determine sample size. No data were excluded from the analyses. The investigators were not blinded to allocation during experiments and outcome assessment.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.