The threat of antimicrobial resistance (AMR) does not originate from a single pathogen or gene, but from a complex spectrum of resistant organisms and determinants: CRE (carbapenem-resistant Enterobacteriaceae), VRE (vancomycin-resistant Enterococcus), MRSA (methicillin-resistant Staphylococcus aureus), and key resistance genes such as bla, van, and mecA. Attempting to address all of them simultaneously would dilute focus, hinder closed-loop validation in modeling and experiments, and obscure innovation.

Therefore, the goal of DBTL1 was to establish a stepwise filtering logic: from epidemiological burden at the pathogen level → mechanistic centrality at the gene level → detectability and translational potential, ultimately converging on the most rational template for our project.

(1) Pathogen-level screening

We constructed a two-strain SIRD (Susceptible–Infectious–Recovered–Deceased) competition model to simulate the dynamics of “resistant strain vs. susceptible strain” within the same population. This allowed us to compare epidemiological outcomes across candidate resistant pathogens.

• Equations (two-strain SIRD system):

• S*: susceptible individuals
• I_m: infections caused by resistant strain
• I_s: infections caused by sensitive strain
• R*: recovered population
• D*: deceased population
• β*: transmission rate
• γ*: recovery rate
• μ*: mortality rate
• N: total population size

By substituting parameters from different candidate pathogens, we directly compared their epidemic peaks, duration, and cumulative mortality—thereby identifying which resistant strains impose the heaviest burden.

(2) Gene-level screening

At the genetic level, we evaluated candidate resistance determinants (bla*, van, mecA) using isolates from the NCBI Pathogen Detection database (2024–2025).

• Method: Logistic regression was applied with gene status as predictor and resistance phenotype (resistant vs. susceptible) as outcome. Performance was assessed by ROC-AUC, sensitivity/specificity, odds ratios (OR), and Fisher’s exact test. Subgroup analyses (OXA, FOX, combined indicators) and cross-validation ensured robustness.
• Objective: To quantify the discriminative power of each gene and identify the most decisive molecular target.

1. Pathogen-level results Simulations using parameter sets for resistant vs. susceptible strains demonstrated that resistant pathogens produce higher epidemic peaks, longer circulation, and greater mortality burdens. Specifically, in the MRSA vs. MSSA comparison, MRSA consistently exhibited the heaviest impact, reflecting the strong link between resistance and increased transmissibility and severity.

• Community MRSA: γ ≈ 0.0016/day; R₀ ≈ 1.3; 30-day CFR 20–24%.
• Community MSSA: γ ≈ 0.011/day; R₀ ≈ 1.1–1.2; 30-day CFR 10–12%.
• ICU MRSA: R₀ ≈ 0.28–0.34; 3-month CFR 27–30%.
• ICU MSSA: R₀ < 1; 3-month CFR 20–23%.

2. Gene-level results

• Methicillin-class composite (MET∪OXA∪FOX): AUC = 0.967, sensitivity ≈ 95.8%, specificity ≈ 97.6%.

• OXA subgroup: AUC = 0.944, sensitivity ≈ 96.8%, specificity ≈ 92.1%.

• Logistic regression OR ≈ 352 (p < 0.001).

These results show that mecA positivity is nearly equivalent to phenotypic resistance, while mecA-negative isolates rarely exhibit resistance. In other words, mecA is both the mechanistic core and the most reliable clinical marker.

3. Literature validation Epidemiological studies and surveillance reports consistently highlight that MRSA, compared with MSSA, causes higher mortality and greater healthcare burden in both community and hospital settings. Meanwhile, the mecA → PBP2a pathway has been repeatedly confirmed as the hallmark mechanism of methicillin resistance. These external findings align with and reinforce our modeling and data-driven results.

For further details, see the Responsiveness section of our iHP documentation.

• Our choice emerged from a stepwise narrowing process: broad-spectrum screening → epidemiological modeling → gene-level prediction → literature validation.
• Among many resistant pathogens and genes, MRSA × mecA uniquely satisfies all criteria: high transmissibility, heavy burden, mechanistic centrality, and clinical measurability..

In DBTL1, we have chosen the mecA gene as a representative antibiotic resistance gene, which combines a stable molecular mechanism of resistance with dynamic gene mobility (1,2).

To address this issue, we proposed a new strategy: eliminate the resistance gene itself, thereby removing the resistance and preventing its horizontal transfer to other bacteria. We selected CRISPR/Cas9 technology for the targeted cleavage of the mecA gene, given its high feasibility for directly disrupting the mecA gene (3). This approach provides a fundamental solution to combat the spread of antibiotic resistance at the genetic level.

To design the target sequence, the ChopChop tool (https://chopchop.cbu.uib.no) was used (Figure 1). After evaluating multiple potential sequences, we selected the rank 2 target sequence: TGGCAAAAAGATAAATCTTGGGG, with 0 mismatches and an efficiency of 67.83% (Figure 2).

Figure 1. Ranking and evaluation of candidate CRISPR target sequences for mecA generated by ChopChop. The ChopChop output displays the distribution of candidate target sequences across the input genomic region. Each candidate is ranked according to predicted efficiency, GC content, and self-complementarity. Among the top 10 ranked sequences, the rank 2 sequence (TGGCAAAAAGATAAATCTTGGGG) was selected for further use.

Figure 2. Off-target analysis of the selected CRISPR target sequence. Potential off-target sites were evaluated using the ChopChop tool. The analysis identified one exact match (0 mismatches) at genomic location MRSA_Scaffold1:649055.

Based on the mecA target sequence (TGGCAAAAAGATAAATCTTG), we identified the corresponding guide region (UGGCAAAAAGAUAAAUCUUG). We then ordered the sgRNA-ΔmecA from Genscript, which was adapted for the Cas9 system and contained this guide region.

After the sgRNA-ΔmecA/Cas9 complex was formed, it was incubated with the mecA gene fragment. The cleavage was assessed by performing DNA gel electrophoresis. In the experimental group where the mecA gene was incubated with the sgRNA-ΔmecA/Cas9 complex, the 2004bp mecA gene fragment was cleaved into two fragments of approximately 700bp and 1300bp (Figure 3). In all other control groups, only a single band of approximately 2000bp was observed, indicating no cleavage.

Figure 3. 1% agarose gel electrophoresis analysis of the mecA gene treated with different components. A. mecA gene treated with the Cas9 nuclease reaction buffer. B. mecA gene treated with only Cas9 protein. C. mecA gene treated with only sgRNA-ΔmecA. D. mecA gene treated with sgRNA-ΔEGFP/Cas9 complex. E. mecA gene treated with sgRNA-ΔmecA/Cas9 complex.

Since the target sequence begins approximately at the 1300th nucleotide of the 2004bp-long mecA gene, the gene was expected to be cleaved into two fragments of around 1300bp and 700bp, which aligns with our experimental results. This confirms that the designed guide region is effective and can successfully target the mecA gene for cleavage using the CRISPR/Cas9 system.

However, direct application of the sgRNA/Cas9 complex to target and cleave antibiotic resistance genes in resistant bacteria is not feasible. The complex must overcome cell membrane barriers and face challenges of poor stability (4). Therefore, an effective delivery system is crucial to ensure that the CRISPR/Cas9 complex can successfully reach the target and function in vivo.

References

1. Fergestad ME, Stamsås GA, Angeles DM, Salehian Z, Wasteson Y, Kjos M. Penicillin-binding protein PBP2a provides variable levels of protection toward different β-lactams in Staphylococcus aureus RN4220. MicrobiologyOpen. 2020 Aug 1;9(8):e1057.

2. Lowy FD. Antimicrobial resistance: the example of Staphylococcus aureus. J Clin Invest. 2003 May 1;111(9):1265–73.

3. Hsu PD, Lander ES, Zhang F. Development and Applications of CRISPR-Cas9 for Genome Engineering. Cell. 2014 June;157(6):1262–78.

4. Seijas A, Cora D, Novo M, Al-Soufi W, Sánchez L, Arana ÁJ. CRISPR/Cas9 Delivery Systems to Enhance Gene Editing Efficiency. International Journal of Molecular Sciences. 2025 Jan;26(9):4420.

DNA origami was selected as the delivery method for the CRISPR/Cas9 system, given its low immunogenicity, programmability, and stimuli-responsive release (1). We referenced the literature and decided to use rectangular-shaped DNA origami structures because the yield of well-formed rectangles was high, with minimal breakage or folding (Figure 1) (2).

Figure 1. Design of the basic rectangular DNA origami structure (DO). DO is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands (S, black short single strands). Each numbered position in the figure represents a specific location on the scaffold strand, essential for the accurate design of the rectangular structure.

The next step was to modify both the sgRNA and the DNA origami to enable the interaction between the DNA origami and the sgRNA/Cas9 complex. Considering that it is unsafe to conduct experiments with MRSA and mecA models, we chose Escherichia coli (E. coli) MG1655 as the model organism and lacZ gene as the target gene to ensure feasibility and safety. We created a modified version of sgRNA targeting lacZ, termed sgRNA_L, by adding an extra linker sequence (GGGUGGUGCCCAUCCUGGUC) at its 3' end. This linker was designed to base-pair with complementary sequences on the DNA origami, ensuring stable binding. Additionally, we incorporated complementary sequences matching the linker into selected staple strands of the origami. Preliminary calculations indicated that the diameter of the Cas9 protein is approximately one-fourth of the width of the DNA origami rectangle. To provide sufficient space for Cas9 protection within the complex environment of the wound, we deliberately excluded staples near the edges. After balancing the available space and the need for efficient loading, we selected six staples in the middle. These complementary sequences, along with a PAM-capturing sequence, were incorporated into the same staple strands (S-PAM-cap) of the origami to facilitate binding with the modified sgRNA_L and to enhance Cas9 recruitment by increasing the local concentration of potential Cas9 binding sites (Figure 2).

Figure 2. Design of the basic rectangular DNA origami structure with S-PAM-cap and PAM-rich (DO_PAMGH). DO_PAMGH is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. The modified staple strands, shown in red, are labeled as S-PAM-cap, which captures PAM-rich. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

To enable visual tracking of the assembly and loading process, the DNA origami was labeled with Cy5 (red) by introducing F-cap and Cy5-labeled F-H (Figure 3), while the sgRNA_L/Cas9 complex was labeled with FITC (green).

Figure 3. Design of DO_PAMGH with F-cap, and F-H (Cy5-labeled DO_PAMGH). Cy5-labeled DO_PAMGH is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. The modified staple strands, shown in red, are labeled as S-PAM-cap, which captures PAM-rich. Yellow dots indicate the F-cap strands that capture the F-H strand. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

We obtained the transcription template for sgRNA_L from GenScript and amplified it using PCR. The sgRNA_L was then transcribed in vitro, and we purified it to obtain high-quality RNA and labeled it with FITC. The DNA origami was assembled by slowly annealing the M13mp18 scaffold, S, S-PAM-cap, PAM-rich, F-cap, and F-H. Once assembled, the FITC-sgRNA_L/Cas9 complex was loaded onto the DNA origami.

To confirm the stability of DO, we built the DNA origami and conducted molecular dynamics simulations within the dry lab. Using oxDNA, a platform specific for molecular dynamics simulation of DNA origami, we acquired figures of "Trajectory Inspection", "Energy Monitoring", "RMSD and Alignment", "RMSF and Mean Structures", "Base-Pairing Statistics", and "Global Metrics". The results illustrated high stability of the origami, though set in high temperature and with more than ten thousand steps of simulation.

Atomic force microscope (AFM) imaging was used to verify the successful assembly of DO_PAMGH. In AFM imaging, the DO_PAMGH maintained the rectangular shape (Figure 4A). Height analysis of a selected single rectangular DO_PAMGH shows that the length and the width of DO_PAMGH are also in line with our expectations (Figure 4B). These structural values indicate the successful assembly of DO_PAMGH.

Figure 4. AFM image and Height analysis of a single rectangular DO_PAMGH. A. A schematic diagram of the designed DO_PAMGH (100nm × 80nm × 2nm) and an AFM image showing a selected assembled rectangular nanostructure. B. Height analysis for the length (upper) and width (below) of DO_PAMGH by AFM. The left picture shows the drawn lines, and the right picture shows the height variation curve along the line, which changes with distance. Scale bar: 100nm.

To validate the successful loading of the sgRNA_L/Cas9 complex onto the DNA origami, we used UV-Vis spectroscopy to measure the absorbance of the system. The DNA origami carrying the sgRNA_L/Cas9 complex exhibited absorption peaks at both 495 nm (FITC) and 650 nm (Cy5), confirming the presence of both components (Figure 5). This result validated that the sgRNA_L/Cas9 complex had been successfully incorporated into the DNA origami, and the system was ready for further application in gene editing.

Figure 5. The UV absorption spectra of Cy5-labeled DNA origami loaded with FITC-sgRNA_L/Cas9 (Cy5-labeled DO_PAMR^FITCC), Cy5-labeled DNA origami (Cy5-labeled DO_PAM), and FITC-sgRNA_L/Cas9 (R^FITCC). The absorption peaks at both 495 nm and 650 nm of Cy5-labeled DO_PAMR^FITCC confirm the presence of both DNA origami and sgRNA_L/Cas9, indicating successful loading.

The successful design and assembly of the DNA origami structure, along with the functional sgRNA_L/Cas9 complex, demonstrated the feasibility of using DNA origami as a delivery platform for CRISPR/Cas9 systems. The incorporation of fluorescence labeling allowed for the precise tracking of the complex, confirming that the loading and assembly processes were successful. The AFM imaging and UV-Vis spectrometry results further validated the structural integrity of the system. Moving forward, the next step will be to overcome bacterial cell membrane barriers for in vivo delivery. However, the current results lay a solid foundation for optimizing the delivery mechanism and further exploring the potential of DNA origami as a gene-editing tool.

• DBTL3.1: Modify sgRNA
• DBTL3.2: Modify DNA origami
• DBTL3.3: Label sgRNA_L/Cas9 complex with FITC
  • DBTL3.3.1: FITC-Cas9
  • DBTL3.3.2: FITC-sgRNA_L
• DBTL3.4: Loading Verification
  • DBTL3.4.1: Fluorescence Agarose Gel
  • DBTL3.4.2: UV-Vis Absorbance Verification

References

Tang W, Tong T, Wang H, Lu X, Yang C, Wu Y, et al. A DNA Origami-Based Gene Editing System for Efficient Gene Therapy in Vivo. Angewandte Chemie International Edition. 2023;62(51):e202315093.

Rothemund PWK. Folding DNA to create nanoscale shapes and patterns. Nature. 2006 Mar;440(7082):297–302.

We designed a modified version, termed sgRNA_L, by adding an extra linker (sequence: GGGUGGUGCCCAUCCUGGUC) at its 3′ end. This linker is intended to form complementary base pairing with specific sequences on the DNA origami, thereby enabling stable binding. The reason behind this design was to utilize an intrinsic mechanism within bacterial cells to release the sgRNA/Cas9 complex. RNase H is a key enzyme that is widely present in bacteria and essential for maintaining genomic stability, as it cleaves the DNA-RNA hybrid region (1). This enzyme naturally recognizes and cleaves DNA-RNA hybrids, which is exactly what would occur between the sgRNA_L and the DNA origami. Although RNase H does not require specific sequences, it prefers certain structural features of DNA. Our design, based on prior studies, aligns with these features, allowing RNase H to efficiently recognize and release the sgRNA/Cas9 complex within the bacterial system.

To eventually test whether DNA origami carrying the sgRNA_L/Cas9 complex can be delivered into bacteria for targeted gene editing, we initially considered MRSA and its resistance gene mecA as a potential model. However, carrying out experiments with MRSA presents significant biosafety risks. After an extensive literature review, we chose Escherichia coli (E. coli) MG1655 as the model organism instead to ensure feasibility and safety. We targeted the lacZ gene because its phenotypic changes are easy to monitor: on X-gal indicator plates, colonies with intact lacZ appear blue, whereas those with disrupted lacZ appear white (2). This provides a straightforward and reliable readout for evaluating gene-editing outcomes.

To design suitable sgRNA target sites for lacZ, we used the ChopChop tool (https://chopchop.cbu.uib.no). After evaluating multiple potential sequences, we selected the target sequence CTGGGGAATGAATCAGGCCACGG. The decision was supported by the repair outcome predictions (Figure 1). Specifically, this site exhibited a high frameshift frequency (70.08%), indicating that cleavage at this locus is highly likely to disrupt the reading frame of lacZ. Moreover, the high microhomology deletion frequency (68.79%) suggests that error-prone repair is effectively activated at this site, helping E. coli survive CRISPR/Cas9-induced double-strand breaks that might otherwise be lethal (3). Taken together, these statistical features strongly support the use of this sequence as a robust target for functional knockout of lacZ.

Figure 1. Prediction of repair outcomes for the selected lacZ target sequence. The statistical predictions of repair outcomes for the lacZ target sequence (CTGGGGAATGAATCAGGCCACGG) were obtained using the ChopChop tool. The analysis reveals a high frameshift frequency (70.08%), indicating a strong likelihood of disrupting the reading frame of lacZ upon cleavage. Additionally, the microhomology deletion frequency is relatively high (68.79%), suggesting that error-prone repair pathways are activated efficiently, minimizing the risk of lethal double-strand break repair outcomes in E. coli. These predictions support the selection of this sequence for functional knockout of lacZ.

Based on the lacZ target sequence (CTGGGGAATGAATCAGGCCA), we identified the corresponding guide region of sgRNA (CUGGGGAAUGAAUCAGGCCA). Considering both synthesis time and cost, we decided to produce sgRNA_L through in vitro transcription. We first designed a transcription template containing a T7 promoter, the guide region of the sgRNA, the gRNA scaffold, and the linker sequence (Figure 2). An additional G was placed immediately downstream of the T7 promoter to enhance transcription efficiency. Between the sgRNA sequence and the linker, we inserted four thymidines (TTTT), which function as a spacer region to minimize potential structural interference and ensure that the linker and sgRNA retain their respective functions.

Figure 2. Design of the transcription template for sgRNA_L production. The sgRNA_L was designed for in vitro transcription using a template containing a T7 promoter, the sgRNA guide region, the gRNA scaffold, and the linker sequence. To enhance transcription efficiency, an additional G was added immediately downstream of the T7 promoter. A spacer region consisting of four thymidines (TTTT) was inserted between the sgRNA sequence and the linker to prevent structural interference, ensuring the proper function of both the linker and the sgRNA.

We obtained the transcription template from GenScript and amplified it by PCR. Subsequently, we used in vitro transcription and purification kits from ToloBio to transcribe the template into large quantities of sgRNA_L. The transcription products were then purified, yielding high-quality sgRNA_L suitable for downstream applications (Figure 3).

Figure 3. Overview of the in vitro transcription process for sgRNA_L production. The process begins with the design of the transcription template, including a T7 promoter, guide region, and linker sequence (left). The template is amplified by PCR (middle left), followed by in vitro transcription (IVT) using T7 polymerase, which synthesizes large quantities of sgRNA_L from the template in the presence of NTPs (middle right). The final step involves product purification, yielding high-quality sgRNA_L suitable for downstream applications (right).

We assessed the concentration and purity of the purified sgRNA_L using a nanodrop spectrophotometer. The measured concentration was 2813.8 ng/μL, with an A260/A280 ratio of 2.05 and an A260/A230 ratio of 2.55, indicating high purity and suitability for downstream experiments.

To further investigate whether the addition of a linker at the 3′ end affects the function of the sgRNA_L, we first PCR-amplified a 2001 bp DNA fragment from the lacZ gene region of E. coli MG1655, which contains the sgRNA target site. Subsequently, the sgRNA_L was incubated with Cas9 protein to assemble a ribonucleoprotein (RNP) complex. This RNP complex was then incubated with the lacZ PCR fragment under appropriate reaction conditions. To evaluate whether the lacZ fragment was successfully cleaved, we performed agarose gel electrophoresis on the reaction products (Figure 4). Given that the target sequence is located at positions 657-677 within the 2001 bp fragment, successful cleavage is expected to yield two DNA fragments of approximately 670 bp and 1330 bp. The observed band pattern on the gel matched these theoretical sizes, confirming that the addition of a 3′ linker does not impair the function of the sgRNA.

Figure 4. 1% agarose gel electrophoresis analysis of lacZ gene treated with different components. A. lacZ gene treated with sgRNA_L/Cas9 complex. B. lacZ gene treated with buffer.

The cleavage results confirmed that the sgRNA_L retained full function even after the addition of a 3′ linker, as evidenced by the expected ~670 bp and ~1330 bp bands observed in agarose gel electrophoresis. This indicates that the 3′ extension does not interfere with the formation or functionality of the sgRNA/Cas9 complex.

However, modifying the sgRNA alone is not sufficient to enable hybridization with the DNA origami structure. To achieve stable base-pairing between the sgRNA and DNA origami, complementary sequences must also be incorporated into selected staple strands of the origami that match the linker region of the sgRNA.

References

Hang T, Zhang X, Wu M, Wang C, Ling S, Xu L, et al. Structural insights into a novel functional dimer of Staphylococcus aureus RNase HII. Biochemical and Biophysical Research Communications. 2018 Sept 10;503(3):1207–13.

Zhao D, Yuan S, Xiong B, Sun H, Ye L, Li J, et al. Development of a fast and easy method for Escherichia coli genome editing with CRISPR/Cas9. Microb Cell Fact. 2016 Dec;15(1):205.

Huang C, Ding T, Wang J, Wang X, Guo L, Wang J, et al. CRISPR-Cas9-assisted native end-joining editing offers a simple strategy for efficient genetic engineering in Escherichia coli. Appl Microbiol Biotechnol. 2019 Oct 1;103(20):8497–509.

To enable base-pairing between the modified sgRNA_L and the DNA origami structure, we designed complementary sequences that match the 3′ linker region of the sgRNA and incorporated them into selected staple strands of the DNA origami. These complementary sequences were added to the 5′ end of the chosen staples, facilitating the hybridization of the sgRNA with the origami through Watson-Crick base pairing. Preliminary calculations indicated that the diameter of the Cas9 protein is approximately one-fourth of the width of the DNA origami rectangle. To provide sufficient space for Cas9 protection within the complex wound environment, we deliberately excluded staples near the edges. After balancing the available space and the need for efficient loading, we chose six staple strands in the middle (102, 103, 104, 113, 114, and 115) as the insertion sites.

In addition, to further enhance the recruitment and stable binding of the Cas9 protein, we inserted a PAM-capturing sequence between the complementary region and the original staple sequence (Figure 1). This intermediate segment is designed to hybridize with PAM-rich DNA strands, thereby increasing the local concentration of potential Cas9 binding sites and promoting more stable association of the sgRNA/Cas9 complex with the origami scaffold. These six modified staple strands are collectively referred to as S-PAM-cap.

Figure 1. Design of the basic rectangular DNA origami structure with S-PAM-cap and PAM-rich (DO_PAM). DO_PAM is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. The modified staple strands, shown in red, are labeled as S-PAM-cap, which captures PAM-rich. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

We mixed M13mp18, S, and S-PAM-cap together and assembled the basic scaffold structure by slow annealing. Afterward, we added the PAM-rich strands and slowly annealed them to integrate into the structure.

AFM imaging was used to verify the successful assembly of DO_PAM. In AFM imaging, the DO_PAM maintained the rectangular shape (Figure 2A). Height analysis of a selected single rectangular DO_PAM shows that the length and the width of DO_PAM are also in line with our expectations (Figure 2B). These structural values indicate the successful assembly of DO_PAM.

Figure 2. AFM image and Height analysis of a single rectangular DO_PAM. A. A schematic diagram of the designed DO_PAM (100nm × 80nm × 2nm) and an AFM image showing a selected assembled rectangular nanostructure. B. Height analysis for the length (upper) and width (below) of DO_PAM by AFM. The left pictures show the drawn lines, and the right pictures show the height variation curve along the line, which changes with the distance. Scale bar: 100nm.

The AFM images indicate that the DNA origami was correctly folded according to the design, confirming the structural accuracy of our design. However, the functionality of the structure remains to be further verified, particularly whether the origami can effectively load sgRNA_L/Cas9 for its intended function.

To investigate whether our designed DNA origami can correctly load the sgRNA_L/Cas9 complex, we fluorescently modified the sgRNA_L/Cas9 complex and DNA origami with different fluorophores. If we successfully detect the two distinct fluorescence signals, it will indicate that the loading process was successful. Firstly, we considered labeling the sgRNA_L/Cas9 complex.

To characterize the sgRNA_L/Cas9 complex, different colored fluorescent groups were intended to be attached to it. FITC (green, excitation wavelength = 495nm, maximum emission wavelength = 525nm) is a common fluorescent-modified molecule (Comparatively, the reason why we don't directly use EGFP is that it is too large and does not meet the size requirements of our DNA origami structure). The isothiocyanate moiety (-N=C=S) on FITC reacts with amino-terminal and primary amine groups on the target biomolecule to form a covalent thiourea bond, and the resulting fluorescein conjugates can be used as specific probes in several applications (1).

For the sgRNA_L/Cas9 complex, we initially planned to label Cas9 via FITC. The reason is that Cas9 has many lysines distributed across its surface, providing abundant, accessible sites for labeling with NHS-ester dyes like FITC. This makes the reaction efficient and reliable. Meanwhile, the FITC-labeled protein technology has become very mature, and we have found that in the previous studies for CRISPR systems, the Cas9 protein was usually selected as the target for labeling, also highlighting that it achieved a high "degree of labeling".

We first perform desalting on the purchased Cas9 to replace the original solvent provided by the supplier with PBS. Subsequently, Cas9 was diluted to the desired reaction concentration, and the corresponding concentrations of FITC-labeled reagent and labeling buffer were added according to the pre-designed reaction system. After incubation at 4°C for 1 hour in the dark, the termination buffer was added, and binding continued for an additional 15 minutes. The Cas9 (160 kDa) samples were then purified using a G-25 Mini Deionization Column to remove unbound FITC, thereby minimizing potential fluorescence interference in subsequent experiments. The eluate was collected, and the absorbance at 280 nm and 495 nm was measured using a Nanodrop and a UV spectrophotometer. Finally, the recovery rate, product concentration, and labeling efficiency were calculated.

To prevent the accidental loss of valuable samples, small-scale desalting was carried out. However, even after multiple optimizations, the recovery rate is still not ideal, with a significant sample loss rate (Table 1).

Table 1. Profiles of one example set of desalting pretreatments of Cas9 protein. The table presents the comparison of the spCas9 protein and the desalted sample, showing the measurements of A₂₈₀ (absorbance at 280 nm), volume, mass, and the recovery rate after desalting. The recovery rate of the desalted sample was calculated as the ratio of the recovered mass (26.9 µg) to the initial mass (50 µg), resulting in a recovery rate of 53.8%.

By recovering and measuring the filtration from the purified column, a simple recovery graph was obtained, in which the first peak is the FITC-Cas9 we needed (Figure 1). The desired product was recovered at the bottom of the first peak, but the loss rate was extremely high, and the yield was not ideal (Table 2). The SDS-PAGE electrophoresis of the products did not reveal any bands. One possible reason could be the insufficiently loaded protein.

Figure 1. Elution profile of FITC-Cas9 in Sephadex G-25 column. 20 tubes were used to collect the eluted solution, with each tube containing 1 ml, for a total of 20 ml. Absorbance was quantified for each 1 ml fraction, and the data were plotted. The A₂₈₀ curve (black) represents protein content, while the A₄₉₅ curve (red) indicates the presence of FITC. Two distinct peaks were observed. The separation of these peaks reflects the size-based separation of proteins, confirming the presence of both labeled and unlabeled Cas9 proteins, as anticipated. The larger FITC-labeled Cas9 protein eluted first. Consequently, the six tubes corresponding to the first peak (3-8 ml) were collected, as indicated by the dashed lines, yielding the desired FITC-labeled Cas9 protein.

Table 2. Profiles of labeling and purification of FITC-Cas9 protein. (C_Cas9: molar concentration of Cas9, C_FITC: molar concentration of FITC, V: volume, E: labeling efficiency; R: recovery rate.)

After measuring the absorbance at 280 nm and 495 nm, a significant loss of the Cas9 protein was observed during the pre-desalting and purification steps, resulting in a yield that did not meet the expected value. Meanwhile, the flow-through efficiency of the G-25 column was relatively low, likely due to the small sample volume. Given the experimental conditions, we did not use molecular sieves; instead, we manually collected each milliliter of the eluate until no protein remained. As a result, the purification process was both complex and time-consuming. Furthermore, the volume of the eluate was substantial, and the product concentration was relatively low, requiring further concentration before it could be used in subsequent experiments. Due to the high price of the Cas9 protein, the amount of protein available to us is very limited. This poses a challenge for us to further optimize the experimental conditions and expand the reaction system. To label the sgRNA_L/Cas9 complex, an alternative choice was to label the sgRNA_L.

Due to the high price of Cas9 and the large amount needed to achieve a high concentration of FITC-Cas9, we decided to give up using FITC-modified Cas9. Instead, we reconsidered the strategy of using FITC modification on the sgRNA_L. But the target sites for modification are far more limited to nucleic acids, compared to proteins. One approach is to modify the amino groups attached to the tail end of the nucleic acid during the synthesis process. The other method is to directly modify the amino groups on the A and C bases (2). The latter method has lower labeling efficiency but more potential labeling sites, which was selected for use. Meanwhile, considering that the efficiency of modifying nucleic acids is generally lower than that of modifying Cas9, we replaced FITC with FITC-NHS ester to enhance the modification efficiency.

The pre-synthesized sgRNA_L and other pre-configured FITC-NHS labeling reagents and labeling buffer were mixed and incubated at 37°C for an hour. The FITC-sgRNA_L was purified by the ethanol precipitation method. During the first purification step, the unreacted fluorescein will remain in the supernatant, while the fluorescein-labeled precipitate will appear dark brown. The presence of a white precipitate indicates that the reaction was unsuccessful. This can act as secondary evidence to examine if the labeling is successful. Similarly, the absorbance values at 260nm and 495nm were measured using a nanodrop after purification, and the recovery rate, product concentration, and labeling efficiency were calculated.

After the purification and separation process, a clear separation of solid and liquid phases, as well as a dark brown precipitate, was observed (Figure 1). The purified product was also found to have two distinct absorption peaks at 260nm and 495nm wavelength, respectively corresponding to sgRNA_L and FITC (Figure 2). The molarity and labeling efficiency reached ideal values that satisfy the requirements of DNA origami assembly (15 nM) and fluorescence imaging (Table 1).

Figure 1. Fluorescence during the ethanol purification process. The three pictures from left to right correspond respectively to situations after the first centrifugation, after the supernatant has been removed, and after the second washing solution has been added. The presence of dark brown sediment can be clearly seen in all pictures.

Figure 2. UV absorption spectrum of one example FITC-sgRNA_L product. There were two distinct absorption peaks at 260nm and 495nm, respectively, corresponding to sgRNA_L and FITC.

Table 1. Profiles of one example set of FITC-labeled sgRNA_L product. The molar concentration (absorbance/ extinction coefficient) was measured to be 11.43 μM, with a purification yield of 5 μL volume. The labeling efficiency of FITC (molarity of FITC/ molarity of sgRNA_L) was calculated to be about 6.70. The recovery rate of sgRNA (mass after purification/ mass before purification) was about 66.70%. Details of the calculation can be seen in the protocol.

In this FITC labeling experiment, we first attempted to label Cas9 with FITC, only to find significant sample loss and low recovery rates during the desalting and purification processes. Worse still, the high cost and limited availability of the Cas9 protein posed a considerable challenge. After switching to labeling sgRNA_L with FITC-NHS ester, we achieved ideal molarity and labeling efficiency, which met the requirements for DNA origami assembly and fluorescence imaging. This shift in strategy underscored the necessity of considering alternative approaches when faced with technical and budgetary limitations.

References

1. Chaganti LK, Venkatakrishnan N, Bose K. An efficient method for FITC labelling of proteins using tandem affinity purification. Biosci Rep. 2018 Dec 21;38(6):BSR20181764.

2. Nasri M, Mir P, Dannenmann B, Amend D, Skroblyn T, Xu Y, et al. Fluorescent labeling of CRISPR/Cas9 RNP for gene knockout in HSPCs and iPSCs reveals an essential role for GADD45b in stress response. Blood Adv. 2019 Jan 8;3(1):63–71.

To label the DNA origami with fluorescence, we selected 36 staple strands and added 15 thymine (T) bases at their 5′ ends (denoted as F-cap), which are designed to capture DNA strands with 15 adenine (A) bases labeled with Cy5 fluorescence at 3' ends (denoted as F-H) (Figure 1).

Figure 1. Design of DO_PAM with F-cap, and F-H (Cy5-labeled DO_PAM). Cy5-labeled DO_PAM is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. The modified staple strands, shown in red, are labeled as S-PAM-cap, which captures PAM-rich. Yellow dots indicate the F-cap strands that capture the F-H strand. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

We mixed M13mp18, S, S-PAM-cap, and F-cap together and assembled the scaffold structure by slow annealing. Afterward, we added the PAM-rich and F-H strands and slowly annealed them to integrate these two strands into the structure. Finally, we introduced the FITC-sgRNA_L/Cas9 complex and annealed it onto the origami scaffold; the final product was denoted as Cy5-labeled DO_PAMR^FITCC.

We ran the assembled origami structure through a DNA gel electrophoresis and then imaged the gel using a laser imaging system. If the origami appeared orange (a combination of green from FITC and red from Cy5 fluorescence), it would indicate successful loading. However, we encountered an issue: the DNA marker itself also exhibited green fluorescence (Figure 1). As a result, it was unclear whether the green fluorescence observed in the origami came from the DNA itself or from the FITC-sgRNA_L/Cas9 complex.

Figure 1. Agarose gel electrophoresis analysis of the DO_PAMRC based on Cy5-labeled DO_PAM and FITC-sgRNA_L. The individual strip on the left is the marker. The three bands on the right side are Cy5-labeled DO_PAMRC, DO_PAMR^FITCC, and Cy5-labeled DO_PAMR^FITCC.

This method proved to be unfeasible due to the interference caused by the green fluorescence of the DNA marker, which made it difficult to distinguish whether the observed green fluorescence in the origami structure came from the DNA itself or from the FITC-sgRNA_L/Cas9 complex. We suspected that the gel stain or the loading buffer could be affecting the fluorescence, so we removed these factors and repeated the experiment, but the issue remained unresolved. We speculated that the issue was caused by mismatched instrument channels for FITC and Cy5, but replacing the instrument was impractical, so we abandoned this method. Considering both FITC and Cy5 fluorescence exhibit absorption peaks at specific wavelengths, we decided to explore utilizing these absorption peaks to investigate whether the sgRNA_L/Cas9 complex can be loaded onto our designed DNA origami.

After reviewing the literature, we found that the maximum absorbance of FITC is at 495nm, while Cy5 has a maximum absorbance at 650nm (1,2). Based on this, we employed a UV-Vis assay to verify the successful loading of the sgRNA_L/Cas9 complex onto the DNA origami. If absorbance peaks were observed at both 495 nm and 650 nm, it would confirm that the loading was successful.

Same as 3.4.1.

We used a Nanodrop spectrophotometer to measure absorbance across the range of 190 nm to 850 nm (Figure 1). The absorbance peaks at 495nm and 650nm indicate correct loading.

Figure 1. The UV absorption spectra of Cy5-labeled DNA origami loaded with FITC-sgRNA_L/Cas9 (Cy5-labeled DO_PAMR^FITCC), Cy5-labeled DNA origami (Cy5-labeled DO_PAM), and FITC-sgRNA_L/Cas9 (R^FITCC). The absorption peaks at both 495 nm and 650 nm of Cy5-labeled DO_PAMR^FITCC confirm the presence of both DNA origami and sgRNA_L/Cas9, indicating successful loading.

The DNA origami carrying sgRNA_L/Cas9 exhibited absorption peaks at both 495 nm and 650 nm, indicating the presence of FITC and Cy5, respectively. This confirms the successful incorporation of sgRNA_L/Cas9 onto the origami structure, thereby validating the effectiveness of our design and methodology. However, this delivery system still faces challenges in overcoming the barriers posed by bacterial cell membranes and cell walls. Therefore, we need to incorporate elements that can disrupt these barriers to facilitate the delivery to the bacteria.

References

1. Schaub JM, Best QA, Zhao C, Haack RA, Ruan Q. Three sample-sparing techniques to estimate the molar absorption coefficient of luminescent dyes. Biochemistry and Biophysics Reports. 2025 June 1;42:101971.

2. Eggleton P, Nissim A, Ryan BJ, Whiteman M, Winyard PG. Detection and isolation of human serum autoantibodies that recognize oxidatively modified autoantigens. Free Radical Biology and Medicine. 2013 Apr 1;57:79–91.

To overcome the barriers posed by bacterial cell membranes and cell walls, we chose to use G4/hemin complexes since DNA origami provides a programmable scaffold for spatially organizing multiple G4/hemin motifs. The G4 (G-quadruplex) structure, when combined with hemin, forms a DNAzyme for catalytic oxidation of H₂O₂ to generate ROS, which aids in disrupting the bacterial wall and membrane (1). To integrate G4 into the DNA origami, we selected specific staple strands and appended the G4 array sequence GGGTAGGGCGGGTTGGG to the 3' end of these staples, and these modified staples are called S-G4 (Figure 1). The specific site choices, density, quantity, and order of assembly for G4/hemin lack normalized standards corresponding to different forms of DNA-origami. General principles guiding site choice for functional staple strands (2) advise against locations that are near the edge of the origami structure to avoid structural errors. To affirm whether connecting to the 3' end or the 5' end will result in the staple protruding towards the desired side, we referred to statistics provided in DNA-nanotube models (3). Crowding of G4/hemin leads to a lowering of its catalytic efficiency (4), so we must compromise between the density and quantity of G4/hemin. Finally, to refine the synthesis protocol, we decided to first allow the assembly of G4 and the DNA-origami and then add hemin, instead of assembling G4/hemin beforehand, given the instability of the G4/hemin complex (5). After analyzing the characteristics and interaction between G4/hemin and DNA-origami, we comprehensively designed a new DNA-origami with 136 copies of G4. Considering that fluorescence tracking is not required for practical applications of the DNA origami, the staples corresponding to the fluorescent sites from DBTL3 were also selected and modified by adding the G4 array.

Figure 1. Design of DO_PAM with S-G4 (DO_PAMG). DO_PAMG is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. S-PAM-cap strands are shown in red, which capture PAM-rich. Blue dots represent the sites of S-G4. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

We mixed M13mp18, S, S-G4, and S-PAM-cap and assembled the scaffold structure by slow annealing. Afterward, we added the PAM-rich strands and slowly annealed them to integrate into the structure. Considering that the incorporation of sgRNA_L/Cas9 did not contribute to the evaluation of the membrane-disrupting effect of the G4/hemin complex, the DNA origami used in the following experiments did not carry sgRNA_L/Cas9. Finally, we added hemin to the assembled DNA origami to form the G4/hemin complex.

G4s are introduced to disrupt the bacterial membrane. However, they are known to exhibit a strong tendency to dimerize through stacking interactions, which can influence the structure and functions of DNA origami (6). To examine whether the new DNA origami structure exhibited dimerization, we subjected DO_PAMG to agarose gel electrophoresis (Figure 2). The results revealed a similar single band of DO_PAMG to that of DO_PAM, with an approximate size double that of M13, indicating correct folding and the absence of dimerization.

Figure 2. 1% agarose gel electrophoresis analysis of M13, DO_PAM, and DO_PAMG. A. M13 scaffold solution. B. DO_PAM solution. C. DO_PAMG solution.

To verify whether hemin was successfully incorporated into the DNA origami to form the G4/hemin complex, we performed a UV–Vis assay on the DNA origami (Figure 3). Since hemin exhibits an absorbance peak around 400nm in the presence of DO_PAMG (7), the presence of a distinct peak around this wavelength in our samples confirmed the successful loading of hemin and the formation of the G4/hemin complex.

Figure 3. The UV absorption spectra of DO_PAMG and DO_PAMGH. DO_PAMG (blue) exhibited an absorption peak at 260nm (DNA), while DO_PAMGH (orange) exhibited two absorption peaks at 260nm and 400nm, corresponding to DNA and hemin, respectively.

To further evaluate the peroxidase-mimicking activity of the G4/hemin complex, we performed an ABTS assay. The DNA origami was incubated with ABTS, and then the absorbance at 415 nm was measured (8) (Figure 4). This absorbance reflects the concentration of oxidized ABTS, which correlates with the peroxidase-mimicking activity of the G4/hemin complex. Our results demonstrated that the G4/hemin loaded on DNA origami exhibited higher peroxidase-mimicking activity compared to an equivalent amount of free G4/hemin complex.

Figure 4. The absorbance of each group at 415nm. Data represent the mean ± s.d. from three independent replicates.

Though the peroxidase was generated efficiently, we simulated in silico that it won't harm the DNA origami itself, sgRNA, and Cas9 protein loaded on it. We performed a cross-section where they are most likely to be harmed (Figure 5). Then, with simulation for more than 10 thousand steps, there was no or limited ROS entering the origami, and even less ROS that reached the protein. The final percentage of the harm of protein and sgRNA is lower than 3 percent, respectively.

Figure 5. ROS Diffusion and Damage Probability Analysis of the sgRNA/Cas9-DNA Origami System. The result showed that sgRNA damage probability is 0.70%, Cas9 damage probability is 0.40%, and DNA damage probability is 11.15%, and the average ROS lifetime is 7.22 × 10⁻⁹ s.

To directly assess the effect of DNA origami on bacterial membrane disruption, we employed the NPN assay. In this assay, increased membrane permeability allows greater uptake of NPN, leading to enhanced fluorescence intensity. After incubating the bacteria with the DNA origami structures, we added NPN and measured the fluorescence as an indicator of membrane permeability (Figure 6). The results showed that the DNA origami induced a significant increase in membrane permeability, with an effect even stronger than that of EDTA, the positive control.

Figure 6. The background-subtracted fluorescence values of four groups. The background-subtracted fluorescence values are expressed as relative fluorescent units. Data represent the mean ± s.d. from three independent replicates.

The results above demonstrate that the DNA origami structure we designed is correctly assembled and does not undergo dimerization, indicating that both our design and assembly methods are effective. Additionally, the results show that the DNA origami, when modified with G4 and hemin, exhibits strong peroxidase-like activity and can disrupt bacterial cell membranes to increase membrane permeability. This property enhances the likelihood of successfully delivering sgRNA_L/Cas9 into the cells.

Although the DNA origami structures have demonstrated promising membrane-disrupting abilities, the incorporation of aptamers for targeting can further enhance membrane permeability. Aptamers, by enabling selective targeting of specific cells, could improve the efficiency of membrane disruption while potentially minimizing damage to human cells. This approach could be particularly useful in improving the precision and safety of DNA delivery systems.

References

1. Mergny JL, Sen D. DNA Quadruple Helices in Nanotechnology. Chem Rev. 2019 May 22;119(10):6290–325.
2. Zhan P, Peil A, Jiang Q, Wang D, Mousavi S, Xiong Q, et al. Recent Advances in DNA Origami-Engineered Nanomaterials and Applications. Chem Rev. 2023 Apr 12;123(7):3976–4050.
3. Berengut JF, Berengut JC, Doye JPK, Prešern D, Kawamoto A, Ruan J, et al. Design and synthesis of pleated DNA origami nanotubes with adjustable diameters. Nucleic Acids Res [Internet]. 2019 Dec 16 [cited 2025 Sept 10];47(22):11963–75. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7145641/
4. Yang B, Wang R, Li W, Wang J, Liu H. On-Origami Molecular Crowding Control of G-Quadruplex DNAzymes. Small Methods [Internet]. 2025 [cited 2025 Sept 10];9(6):2401401. Available from: https://onlinelibrary.wiley.com/doi/abs/10.1002/smtd.202401401
5. Monte Carlo AR, Fu J. Inactivation Kinetics of G‐Quadruplex/Hemin Complex and Optimization for More Reliable Catalysis. ChemPlusChem. 2022 July;87(7):e202200090.
6. Rosi BP, Libera V, Bertini L, Orecchini A, Corezzi S, Schirò G, et al. Stacking Interactions and Flexibility of Human Telomeric Multimers. J Am Chem Soc. 2023 July 26;145(29):16166–75.
7. Tokura Y, Harvey S, Chen C, Wu Y, Ng DYW, Weil T. Fabrication of Defined Polydopamine Nanostructures by DNA Origami‐Templated Polymerization. Angew Chem Int Ed. 2018 Feb 5;57(6):1587–91.
8. Re R, Pellegrini N, Proteggente A, Pannala A, Yang M, Rice-Evans C. Antioxidant activity applying an improved ABTS radical cation decolorization assay. Free Radical Biology and Medicine. 1999 May 1;26(9):1231–7.

To measure the membrane permeability of bacterial cells incubated with DO_PAMGH, we performed an NPN uptake assay. The hydrophobic fluorescent probe N-Phenyl-1-naphthylamine (NPN), which fluoresces weakly in aqueous environments but exhibits strong fluorescence in phospholipid conditions, has been widely used to evaluate the membrane integrity of bacteria (1). Due to the presence of lipopolysaccharide in the outer leaflet of the outer membrane, the lipophilic fluorescent probe cannot thoroughly permeate the intact bacteria. When the outer membrane is disrupted by permeabilizers, NPN can access the phospholipids and exhibits surges in fluorescent intensity (1).

We conducted a preliminary NPN assay to explore the optimal experimental conditions. Given that constructing DNA origami is time-consuming, the pre-experiment only included the negative control (buffer; buffer + cell) and a positive control (EDTA).

The fluorescence intensity of NPN in each well was measured (Figure 1; Figure 2).

Figure 1. The Background-Subtracted Fluorescence Values of Each Group. The background-subtracted fluorescence values are expressed as relative fluorescent units. Data represent the mean ± s.d. from three independent replicates.

Figure 2. The NPN uptake factors of each group. The NPN uptake factor of the buffer group is set to 1. The uptake factors for the other groups are calculated as the ratio of the background-subtracted fluorescence of each group to the background-subtracted fluorescence of the buffer group. This method allows for the comparison of NPN uptake between different experimental conditions.

The NPN uptake factor for the EDTA group was significantly lower than the values reported in the literature (1). After ruling out other possible causes, we suspect that the high fluorescence of the buffer group or the low fluorescence of the EDTA group may be the reason for this discrepancy. Upon reviewing a large amount of literature, we found that the high fluorescence of the buffer could be due to the use of transparent 96-well plates (1,2), and the low fluorescence in the EDTA group might be due to the use of PBS as the buffer, as the effect of EDTA in PBS is weak and may not effectively enhance bacterial cell membrane permeability (3). To avoid these problems, the black plate and Tris buffer should be used.

After identifying the potential causes, we switched from transparent 96-well plates to black plates to reduce background fluorescence. Additionally, we replaced the EDTA group's buffer with Tris-HCl to enhance the effectiveness of EDTA in increasing bacterial cell membrane permeability. We then performed the NPN uptake assay again using these adjusted conditions.

We conducted the NPN uptake assay. This time, we included DO_PAMGH.

The fluorescence intensity of NPN in each well was measured (Figure 1; Figure 2).

Figure 1. The Background-Subtracted Fluorescence Values of Each Group. The background-subtracted fluorescence values are expressed as relative fluorescent units. Data represent the mean ± s.d. from three independent replicates.

Figure 2. The NPN Uptake Factors of the Four Groups. The NPN uptake factor of the buffer group is set to 1. The uptake factors for the other groups are calculated as the ratio of the background-subtracted fluorescence of each group to the background-subtracted fluorescence of the buffer group. This method allows for the comparison of NPN uptake between different experimental conditions.

After modifying the experimental conditions, the NPN uptake factor of the EDTA positive control increased, indicating that our improvements were effective. Additionally, we found that the NPN uptake factor for the DO_PAMGH group was even higher than that of EDTA, suggesting that DO_PAMGH effectively enhances bacterial cell membrane permeability.

References

1. Helander IM, Mattila‐Sandholm T. Fluorometric assessment of Gram‐negative bacterial permeabilization. J Appl Microbiol. 2000 Feb 1;88(2):213–9.

2. Fluorescence Intensity Measurements | BMG LABTECH [Internet]. [cited 2025 Sept 29]. Available from: https://www.bmglabtech.com/en/fluorescence-intensity/

3. Vaara M. Agents that increase the permeability of the outer membrane. Microbiol Rev. 1992 Sept;56(3):395–411.

To enhance the membrane disruption efficiency of DNA origami while minimizing damage to human cells, we plan to incorporate a targeted aptamer into the DNA origami structure. When screening for feasible aptamers, we pursued two approaches to identify suitable aptamer sequences. The first approach involved searching aptamer databases, with the hope of finding sufficient data to enable the application of machine learning techniques for building a more comprehensive mechanism. The second approach involved reviewing relevant literature to find aptamer data, which would allow us to leverage previous research ideas. Both methods yielded progress: we identified pertinent papers that provided the aptamer sequences we needed, and we also located a global nucleic acid aptamer database where we found a suitable aptamer (sequence: CCATCCACACTCCGCAAGGGTGCCCCGGGGGGCTGTTCAGCGTGGTGGTGGGATGCCGTGTTGGTCCTTAGTCTCCGTCGTCGGCTGCCTCTACAT). Unfortunately, the database contained too few aptamers for methicillin-resistant Staphylococcus aureus (MRSA), limiting our ability to proceed with machine learning-based work.

Given that the model used for experimental validation is Escherichia coli MG1655, we employed an aptamer sequence that has been validated in the literature to specifically target Escherichia coli (sequence: CATATCCGCGTCGCTGCGCTCAGACCCACCACCACGCACC) (1). To enable the aptamer and the DNA origami to be connected through base pairing, we added a linker to the 3' end of the aptamer (C-APT), designed to be complementary to the origami structure (linker sequence: TTTTTCGCTTATTATTATTATTATTA). Subsequently, we incorporated 12 staple strands on the DNA origami that contain complementary sequences to the aptamer linker sequence at its 5' end (Apt-cap), allowing it to pair with the aptamer (sequence: TAATAATAATAATAAGCGTTTTT) (Figure 1). This ensures that the aptamer can bind to the DNA origami through base pairing facilitated by the complementary sequences. Considering that aptamers have a more complex structure than linear DNAs, they are more likely to interfere with each other if placed in the middle of the DNA origami rectangle. Additionally, aptamers would bind to the target more effectively if they were not obstructed by other functional staples. Therefore, we positioned the aptamers along the short sides of the DNA origami. For the placement pattern, we referenced the work of Li et al. and added two additional adjacent aptamers in the middle of the short side to ensure efficient targeting (2).

Figure 1. Design of DO_PAMG with Apt-cap and C-APT (DO^A_PAMG). DO^A_PAMG is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. S-PAM-cap strands are shown in red, which capture PAM-rich. Blue dots represent the sites of S-G4. Apt-cap strands are shown in purple, which capture C-APT. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

To verify the targeting effect of the aptamer, we decided to assess it using an NPN assay indirectly. If the aptamer has a targeting effect, the origami with the aptamer should exhibit a better membrane disruption effect, as the aptamer can help the origami accumulate around the bacteria. Fluorescence was not required for this experiment, so the constructed origami did not include F-cap and F-H, which is consistent with Figure 1.

For the construction of the DNA origami for the NPN assay, we mixed M13mp18, S, S-G4, S-PAM-cap, and Apt-cap and assembled the scaffold structure through slow annealing. Afterward, we added the PAM-rich strands and aptamer strands, and slowly annealed them to integrate into the structure. Since the incorporation of sgRNA/Cas9 did not contribute to the evaluation of the membrane-disrupting effect of the G4/hemin complex, the DNA origami used in the subsequent experiments did not carry sgRNA/Cas9. Finally, we added hemin to the assembled DNA origami to form the G4/hemin complex.

To visually assess the targeting effect of the aptamer and whether it facilitates the internalization of the DNA origami, we decided to incubate the origami with or without the aptamer with bacteria and observe them using a laser confocal microscope (LSM). To determine whether the origami accumulates around the bacteria or enters the bacteria, both types of origami in the experiments needed to be fluorescently labeled. The assembly process of the origami was nearly identical to that used in the NPN assay, except that F-cap and F-H were included.

It is important to note that in the targeting validation experiment by LSM, we were primarily interested in observing whether the origami accumulates around the bacteria. If the origami enters the bacteria, it could interfere with the results. Therefore, in the origami constructed for the targeting validation experiment by LSM, we did not include S-G4 (Figure 2).

Figure 2. The DNA origami used to validate the targeting effect (Cy5-labeled DO^A_PAM). Cy5-labeled DO^A_PAM is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. S-PAM-cap strands are shown in red, which capture PAM-rich. Yellow dots represent the sites of F-cap, with capture Cy5-labeled F-H. Apt-cap strands are shown in purple, which capture C-APT. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

For the experiment investigating whether the aptamer can facilitate the internalization of the origami, both F-cap and S-G4 were required. However, all 36 binding sites of F-cap share the same site as S-G4. To ensure appropriate fluorescence intensity, we kept all 36 F-cap sites and added 100 S-G4 (Figure 3).

Figure 3. The DNA origami used to validate the internalization (Cy5-labeled DO^A_PAMG). Cy5-labeled DO^A_PAMG is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. S-PAM-cap strands are shown in red, which capture PAM-rich. Yellow dots represent the sites of F-cap, with capture Cy5-labeled F-H. Blue dots represent the sites of S-G4. Apt-cap strands are shown in purple, which capture C-APT. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

To verify whether the incorporation of the aptamer enhances the membrane-disrupting effect of DNA origami against E. coli, we incubated bacteria separately with DNA origami containing the aptamer and with DNA origami lacking the aptamer. Subsequently, we performed an NPN assay. The results demonstrated that the NPN fluorescence of bacteria incubated with aptamer-functionalized DNA origami was significantly higher than that of bacteria incubated with DNA origami without the aptamer (Figure 4). This indicates that the presence of the aptamer targeting E. coli indeed improves the membrane disruption efficiency of the DNA origami.

Figure 4. The background-subtracted fluorescence values of two groups. The background-subtracted fluorescence values are expressed as relative fluorescent units. Data represent the mean ± s.d. from three independent replicates.

To visually compare the targeting and internalization effects of DNA origami with and without the aptamer, we incubated E. coli with Cy5-labeled DNA origami containing or lacking the aptamer. Afterward, we washed off any non-targeted or non-internalized origami using PBS. The samples were then analyzed using laser confocal microscopy.

In the targeting validation, DNA origami with the aptamer was observed surrounding the bacteria, whereas DNA origami without the aptamer was randomly scattered around, confirming that the aptamer indeed facilitates targeted binding to E. coli (Figure 5). The quantitative analysis of the mean fluorescence intensity (MFI) co-localized with E. coli also indicated that the MFI of E. coli treated with Cy5-labeled DO^A_PAM is higher than that treated with Cy5-labeled DO_PAM, further confirming the targeted effect of the aptamer (Figure 6).

Figure 5. Representative confocal microscopy images of E. coli cells after the incubation of Cy5-labeled DO_PAM with or without the modification of the aptamer. Scale bar: 20µm × 20µm.

Figure 6. Quantification of the mean fluorescence intensity (MFI) of E. coli cells after the incubation of Cy5-labeled DO_PAM with or without the modification of the aptamer. - Aptamer: Cy5-labeled DO_PAM. + Aptamer: Cy5-labeled DO^A_PAM. Data represent the mean ± s.d.. Statistical significance was calculated by the unpaired two-tailed Student's t-test. (***P < 0.001).

In the internalization validation, strong red fluorescence was observed inside the bacteria in the aptamer group, while the fluorescence was much weaker in the group without the aptamer (Figure 7). The MFI analysis co-localized with E. coli further confirmed this (Figure 8). This indicates that DNA origami with the aptamer enhances the internalization efficiency of the origami into E. coli.

Figure 7. Internalization of Cy5-labeled DO_PAMGH and DO^A_PAMGH on E. coli. With the addition of the aptamer, the amount of fluorescence entering the cells increased, and the area with fluorescence also expanded. Scale bar: 20µm × 20µm.

Figure 8. Quantification of the mean fluorescence intensity (MFI) of E. coli cells after the incubation of Cy5-labeled DO_PAMGH with or without the modification of the aptamer. - Aptamer: Cy5-labeled DO_PAMGH. + Aptamer: Cy5-labeled DO^A_PAMGH. Data represent the mean ± s.d.. Statistical significance was calculated by the unpaired two-tailed Student's t-test. (*P < 0.05).

Our experimental results demonstrate that the aptamer we used for targeting E. coli is effective and can increase the probability of DNA origami internalizing into the cells. However, the issue of poor stability of Cas9 remains unresolved. After reviewing the literature, we discovered that rolling rectangular DNA origami into tubular structures to encapsulate sgRNA/Cas9 can improve its stability (3). This approach has become our next step in the project.

References

1. Mela I, Vallejo‐Ramirez PP, Makarchuk S, Christie G, Bailey D, Henderson RM, et al. DNA Nanostructures for Targeted Antimicrobial Delivery. Angew Chem Int Ed. 2020 July 27;59(31):12698–702.
2. Li S, Jiang Q, Liu S, Zhang Y, Tian Y, Song C, et al. A DNA nanorobot functions as a cancer therapeutic in response to a molecular trigger in vivo. Nat Biotechnol. 2018 Mar;36(3):258–64.
3. Tang W, Tong T, Wang H, Lu X, Yang C, Wu Y, et al. A DNA Origami-Based Gene Editing System for Efficient Gene Therapy in Vivo. Angewandte Chemie International Edition. 2023;62(51):e202315093.

To enhance the stability of the sgRNA/Cas9 complex during application, we decided to introduce a new staple to roll the rectangular DNA origami structure into a tubular shape, referred to as S-Lock. We incorporated a DNA strand that is complementary to the frame at both ends of the origami, aiming to fold the rectangular structure into a tube (Figure 1). Additionally, we introduced thiol groups to each strand so that adjacent S-Lock structures could form disulfide bonds, further stabilizing the overall structure. To optimize the unutilized staples on the long sides, we placed disulfide bonds in the middle of 8 staples on each side. This arrangement strengthens the bond between the two sides, facilitating the transformation of the origami rectangle into a tube. Yin et al. used a similar number of staples for side-to-side binding, which demonstrated an appropriate density of functional staples (1). Furthermore, the length of their staples that do not pair with the scaffold is similar to ours. Based on this, we assume that the adjacent staples will not interfere with each other, and the likelihood of incorrect pairing is minimal.

Figure 1. Design of DO^A_PAMG with S-Lock (L-DO^A_PAMG). L-DO^A_PAMG is constructed using the M13mp18 phage genome as the scaffold strand (black long single strand), which is folded into a rectangular shape by staple strands. S-PAM-cap strands are shown in red, which capture PAM-rich. Blue dots represent the sites of S-G4. Apt-cap strands are shown in purple, which capture C-APT. S-Lock strands are shown in green. Each numbered position in the figure corresponds to a unique site on the scaffold strand, indicating precise staple placement.

The process of folding from a rectangle to a tube was explored through previous studies. Approaches such as thrombin-loaded tubular origami (2) and tissue plasminogen activator-loaded tubular origami (1) have been successfully achieved. Our cargo, the Cas9 protein, shares similarities with thrombin and tissue plasminogen activator, as all three proteins feature a high-density positively charged region. Additionally, it was highlighted that DNA backbones carry a negative charge, which repels negatively charged molecules (1). Based on this, we hypothesized that the DNA origami could fold into a complete tube, driven by the interaction between the positively charged cargo and the negatively charged DNA backbone.

We mixed M13mp18, S, S-G4, S-PAM-cap, and Apt-cap, and assembled the scaffold structure by slow annealing. Afterward, we added the PAM-rich strands and aptamer strands, slowly annealing them to integrate into the structure. Then, we incorporated the sgRNA/Cas9 complex and slowly annealed the mixture. Subsequently, we added the lock strands and performed slow annealing again. Finally, we added hemin to the assembled DNA origami to form the G4/hemin complex.

To verify the protective effect of the cylindrical structure on the sgRNA/Cas9 complex, we planned to incubate DNA origami with or without Lock, containing sgRNA/Cas9, with Proteinase K, followed by SDS-PAGE to assess the amount of Cas9 remaining. If the origami with the Lock strand showed more remaining Cas9, while the group without the Lock strand showed less, it would indicate that the Lock strand indeed provides a protective effect. However, when preparing for the experiment, we realized that the concentration of our DNA origami was too low (5 nM) and the amount of Cas9 protein was insufficient. Even without Proteinase K digestion, the Cas9 was not detectable by SDS-PAGE. Given the experimental conditions, increasing the origami concentration was not feasible. Therefore, we abandoned the wet-lab experiment to demonstrate the protective effect of the tubular structure on sgRNA/Cas9 and considered using a dry-lab approach instead.

In the dry-lab, we quantified the degradation process of the tubular DNA origami by multiple enzymes through an enzyme diffusion kinetics model combined with Monte Carlo simulation methods (3). The simulation process integrates parameter initialization, time-step iteration, result statistics, and visualization modules, with statistically significant results obtained through multiple independent runs. Tests reveal an overall enzyme entry probability of only 4.0%, with average final stability maintained at 1.000, indicating limited degradation effects. Neutrophil elastase exhibited the highest entry frequency due to its high concentration (1000 ng/mL) and small size (Rh = 1.96 nm), whereas MMP-2 (matrix metalloproteinase-2) failed to enter owing to low concentration and larger size.

The simulation process includes parameter initialization, time-step iteration, statistical analysis, and visualization. Multiple independent runs ensure statistical significance. After the simulation, the overall probability of enzyme entry is 4.0%. The average final stability of the tubes is 1.000, indicating limited degradation. Among the enzymes tested, neutrophil elastase displayed the highest entry frequency, which could be attributed to its high concentration (1000 ng/mL) and small size (Rh = 1.96 nm). In contrast, MMP-2 could not enter the nanotubes due to its low concentration and larger size. In summary, the result showed that little or no enzyme could enter the tubular DNA origami, which confirmed the protection of the origami and the application of our system.

Given the experimental conditions, we abandoned the wet-lab experiment to demonstrate the protective effect of the tubular structure on sgRNA/Cas9 and considered using a dry-lab approach instead. On the programmatic aspect, several optimizations and corrections were made during development: a parameter error in the Brownian motion visualization function (c → color) was fixed, type hints were enhanced to improve code reliability, and a minimum value protection was added to the diffusion coefficient calculation to prevent division-by-zero errors. Furthermore, the synergy matrix was expanded (e.g., the synergy factor between MMP-9 and elastase was set to 1.2), and the Brownian motion trajectory simulation function was enhanced, deepening the model's capability to describe complex biophysical processes.

From a biological perspective, though we already simulated typical and abundant enzymes that may harm the Cas9 protein, more complex situations still exist. In the future, we may conduct deeper research on different attributes of enzymes to better simulate their unique features. Also, we will increase the variety of enzymes to fit more practical applications.

References

1. Yin J, Wang S, Wang J, Zhang Y, Fan C, Chao J, et al. An intelligent DNA nanodevice for precision thrombolysis. Nat Mater. 2024 June;23(6):854–62.
2. Li S, Jiang Q, Liu S, Zhang Y, Tian Y, Song C, et al. A DNA nanorobot functions as a cancer therapeutic in response to a molecular trigger in vivo. Nat Biotechnol. 2018 Mar;36(3):258–64.
3. Harrison RL. Introduction to monte carlo simulation. In 2010. p. 17.