Abstract

Owing to their capacity for accurate structural control and complex programmability, DNA molecules have been extensively studied in relation to the construction of nanodevices. However, the existing logic gate sections based on DNA self-assembly were independent of each other, which hampered the development of large-scale integrated DNA circuits. Herein, we have explored a logic operation device with excellent scalability based on assembling and selectively releasing AuNPs on DNA origami, and have performed YES gate, OR gate, AND gate, and three-input composite gate. In the experiment, the logic operation result is detected by gel analysis and TEM image. The resolution of the output signals was greatly improved by determining the releasing of AuNPs from two-layer honeycomb origami. Our study provides a promising approach for building more complex large-scale DNA logic circuits.

1. Introduction

DNA self-assembly has provided a versatile tool to build bottom-up nanostructures with predesigned shapes, including lattices [1], filaments [2], spheres [3], barrels [4], and other complex three-dimensional structures [511]. Especially, Rothemund had firstly developed DNA origami, which was fabricated by approximately 200 short single-stranded DNAs (ssDNA) and one long single-stranded M13 scaffold to form complex nanostructures [12], providing a new effective way to design a wide range of artificial nanostructures. DNA origami not only offers an effective protocol to create various nanostructures with diverse shapes and sizes [1315] but can also serve as a template to precisely arrange nanoparticles [1619]. Based on the specific recognition of nucleotide sequence, extended DNA strands on origami can be regarded as DNA tags to detect and capture functional nanoparticles.

Consequently, various components such as proteins [2022], quantum dots [23, 24], carbon nanotubes [25, 26], and metal particles [2729] can be precisely arranged and manipulated at the specific positions on the DNA origami surfaces. Among various functional particles, owing to its structural and optical properties, gold nanoparticles (AuNPs) have attracted more attention from the fields of biosensing and nanodevices [3038]. For example, Sharma et al. used lipoic acid-modified ssDNA to construct AuNP-DNA conjugates with a monothiolate-Au linkage [39]. This strategy provides an effective method to fix the size of DNA nanostructures carrying a discrete number of AuNPs at designed positions. Moreover, Shen et al. developed DNA origami nanomachines by exploiting the optical property of AuNPs [40]. Although these works revealed that NPs can be well assembled with DNA origami statically, the dynamic operation of an NP/DNA origami device is still a concern for various researchers. Our previous study realized a NP/DNA origami system, in which AuNPs could be dynamically released from a DNA origami in response to specific DNA signals [41]. In this work, the yield of the output structure was not very high, because a monofunctionalized DNA/AuNP conjugate was hybridized with one extended strand from the origami template. Besides, the resolution of the output signals was not very obvious, resulting from the little molecular weight difference between the displaced product and the initial product.

To address these problems, we constructed a logic system based on the arrangement of AuNPs on two-layer honeycomb origami (TL), in which bifunctionalized AuNPs could be logically and dynamically released from DNA origami in response to input signals. Using this strategy, “YES,” “OR,” “AND,” and three-input composite logic gates were established by controlling the releasing of AuNPs from TL origami. In the experiment, the logic operation result is detected by gel analysis and TEM image. The resolution of the output signals was greatly improved by determining the releasing of AuNPs from TL origami. Our study provides a promising approach for building more complex large-scale DNA logic circuits.

2. Materials and Methods

2.1. Preparation of DNA Origami

As a computer-aided design tool for DNA origami, caDNAno was employed to design and produce staples. The design drawing of TL is shown in Figure S6. All the sequences of DNA strands (Tables S2 and S3) were exported from caDNAno and synthetized by Sangon Biotech Co. Ltd. To generate the TL structure, 2.5 nM of M13mp18 strands were mixed with ~200 staples at a ratio of 1: 10 in (Tris: 40 mM; acetic acid: 20 mM; EDTA: 2 mM; and magnesium acetate: 12.5 mM, pH 8.0). Composite samples were annealed from 85°C to room temperature within 16 h [42]. It is noteworthy that staples used to form binding sites should increase the concentration, so as to ensure the yield of binding sites. Purified through a 100 K ultracentrifugal filter (MWCO, Amicon, Millipore), which was able to eliminate the excess staple strands, TL was prepared to react with AuNP (Table S4). The concentration of TL was determined from the optical absorbance at 260 nm.

2.2. Modification of AuNP

Fifteen-nanometer AuNP colloids were mixed with bis(p-sulfonatophenyl)phenylphosphine dihydrate dipotassium salt (BSPP) and incubated in a shaking bed overnight. The treated AuNP solution reacted with thiolated ssDNA, A-SH, or B-SH to form AuNP/DNA conjugates Au-A or Au-B, respectively. Additionally, oligoethylene glycols (OEG) were utilized to make surface-unsaturated AuNPs stable in high concentration magnesium ion. Subsequently, excess ssDNA strands were removed the same way as that employed by the TL DNA origami. Finally, freshly prepared gold conjugates and TL were mixed at a ratio of 30:  1, which formed three kinds of AuNP/origami composite structures: TL-A, TL-B, and TL-AB.

2.3. Loading AuNPs on TL

The mixed solution was incubated for two hours at room temperature. After adding trigger ssDNA strands, samples were characterized by 1% agarose gel (stained with ethidium bromide) electrophoresis in (45 mM Tris, 45 mM boric acid, and 1 mM EDTA; pH 8.0) containing 6 mM MgCl2 for 1.5 h in an ice-water bath. The target-assembly bands were sliced out and ground so that a little liquid could be absorbed for TEM imaging.

2.4. TEM Analysis

Two microlitres of target samples were deposited on a carbon-coated TEM grid (100 mesh, Ted Pella, Inc.) and dyed by uranyl acetate (Electron Microscopy, Beijing, China). Whereafter, the excess solution was removed by using a piece of filter paper. The grid was stored at room temperature to allow drying. TEM images were obtained by a Hitachi H-7650 transmission electron microscope. TEM statistics were completed manually (Table S5). MATLAB was employed to generate bar graphs and fitting curves.

3. Results and Discussion

The scheme of the logic gate system is depicted in Figure 1(a). The TL origami with dimensions of provided a template for the nanodevice, in which DNA-coated AuNPs could bind on by hybridizing with extended DNA staples. Six AuNP-binding sites, composed of two extended ssDNA per site, were evenly distributed on the TL’s surface with a distance of ~21 nm between every two sites. Extended sequences LogA and LogB were arranged alternately on these sites. In other words, every TL-A or TL-B was able to bind three gold particles, respectively, and TL-AB was able to bind six AuNPs. The TL structure is described in Figure S3, including an unfolding map, a cross-section profile, all the sequences of the staples, and the specific locations of the binding sites. Typical TEM images of TL are shown in Figure S7. Besides, two types of 15 nm AuNPs were modified with thiolated ssDNA, A-SH, or B-SH, which were marked as Au-A or Au-B, respectively. Figure 1(b) illustrates components of the logic gate device by gel images. Lane 1 indicates bare AuNPs without modified DNA strands, whose band has more mobility than that of Au-A or Au-B (lane 2 and lane 3, respectively). As a reference for TL running in lane 5, M13mp18 ran in lane 4. Both of them could be observed under UV light after being stained by ethidium bromide (EB).

In the logic gate system, logic operations were completed by inputting ssDNA which could displace AuNPs arranged on the TL’s surface selectively and programmably. The addition of a trigger strand was equivalent to a logical input, with possible states of adding or not adding (0 or 1, respectively). Then, we defined AuNPs-bound-on TL structure as the original state (0), while the AuNP-free TL structure was defined as the triggered state (1).

In YES gate A, Au-A was set as the potential released particle. When input strand A was presented, extended staples LogA preferentially hybridized to A owing to the 5 nt long toehold (Figure S1). As a result, Au-A was released. Comparing lane 6 and lane 7, we could easily observe the disappearance of TL-A in the output area of lane 7, on account of the perfect displacement of Au-A. Interestingly, in the UV image, we found that a triggered structure exhibited a faster movement by getting rid of Au-A. Furthermore, without the influence of gold nanoparticles, DNA origami appeared much more brightly. To further study the effect of strand A, we performed a concentration-gradient experiment. As shown in Figure S2, with the increase of the concentration of A, the replacement effect became better. Logic gate B was a little different from gate A in hybridization and displacement details. As shown in Figure S1, there were two toeholds in the hybrid structure of B-SH and LogB. Both 3-end and 5-end toeholds could be recognized by B, leading to the strand displacement. The reactions similar to Au-A displacement occurred in lane 8 and lane 9. Profiting from the two-time toehold, the displacement efficiency of B was higher than that of A, which was especially evident in the 4.5 μM band (Figure S3). Furthermore, Figures 1(c) and 1(d) demonstrated several times the TEM images of Au-B-bound TL (TL-B) and Au-B-released TL.

In order to quantitatively analyze the diversity of logic gate structures, we statistically counted all the TEM images, plotted bar graphs, and fitted curves utilizing MATLAB. Representational TEM images are displayed in Figure S8 (YES gate A) and Figure S9 (YES gate B). The numerical information of the statistical results is listed in Table S1. By statistical counting, the distribution histograms of various products are shown in Figure 1(e) (YES gate A) and Figure 1(f) (YES gate B). The narrower Gaussian distribution observed by TEM indicated the higher homogeneity of the sample. In Figure 1(e), blue bars represent the original state, whose peak corresponded to “3” on the abscissa, and the theoretical value of the number of AuNPs bound on per TL. Gray bars indicated the triggered state. It could be observed that 92.8% of TLs were not connected with AuNPs. Characteristics similar to Figure 1(e) were obtained in Figure 1(f). Particularly, 3-AuNPs-bound-on TL structures accounted for 79.4% in the original state sample and the AuNP-free TL occupied 92.6% of the triggered state.

It was worth mentioning that different products from the original or triggered state samples had conformed to different distribution laws. The histograms from the original state samples were fit with Gaussian distribution, while the triggered state samples were fit with exponential distribution. These results demonstrated well the selective release of AuNPs from the DNA origami. (The truth tables of YES gate A and YES gate B are listed in Table 1.)

Owing to the superior scalability of our computing system, the AND logic gate was also established via combining YES gate A and YES gate B. In the construction of the AND gate, TL origami with six alternately-arranged binding sites was obtained as described above. Au-A and Au-B were simultaneously assembled on TL (TL-AB).

As illustrated in Figure 2(a), the AuNPs were released when A and B were added, and the true output was defined by AuNP-free TL. Evidently, by displacing Au-A with A, TL-B was gained. Similarly, TL-A could be obtained separately by adding B which was capable of untying Au-B. In the output area of the gel result (Figure 2(b)), because a nanoparticle has little effect on band migration, the bands of TL-AB, TL-B, and TL-A were almost displayed at the same horizontal position. Only when Au-A and Au-B were simultaneously disassembled would the band disappear. Histogram statistics of TEM images described a matching result in Figure 2(c). The peak moved to “0” when both A and B were added to the system. Moreover, the distribution of purple bars representing an input-free state positioned a peak at ~“5,” rather than at the theoretical value of “6.” We guessed that intermolecular repulsion and steric hindrance affected the perfect assembly of TL-AB. Besides, we also guessed that, when Au-A or Au-B was separated, the particle spacing increased, which facilitated the rehybridization of excess AuNPs to the corresponding binding sites. The statistical result of the histogram supported these guesses: when either A or B was added into the system separately, which should have led to three AuNPs unbound, peaks identically shifted from “5” to “3,” instead of “2” or less. Representative TEM images of various output products are displayed in Figures 2(d) and S10. The truth table of AND gate is listed in Table 2.

To fabricate the OR gate, we altered the trigger ssDNA for TL-B. As illustrated in Figure 3(a), TL-B could be disassembled by both bst and bAu. Concretely, bst recognized the toehold region of the extended staple (light green) and could form a 12-pair-long complementary double-strand structure, which left a 3-pair-long unstable double-strand connection between the staple strand and B-SH. Due to the instability of the 3-pair complementary structure, Au-B was detached from TL, thereby reporting the output as “1.” The trigger strand bAu worked similarly: it detected the toehold of B-SH (orange) and preferentially displaced B-SH from LogB, so as to rid Au-B of the TL template. The detailed sequences are portrayed in Figure S1. Then, the logic OR gate system was characterized by gel electrophoresis (Figure 3(b)). When bst or bAu, respectively, reacted with the original conjugate, or both of them simultaneously participated in the reaction, the band indicating TL-B disappeared in the output area. Furthermore, concentration gradient experiments for the trigger strand were performed and the results are exhibited in Figure S4 and Figure S5. It was reasonable that bAu consumed more than bst in general, as a result of excessive Au-B in the solution. Statistical distribution of the TEM images agreed with the rules of OR logic. Peak migration occurred when either bst or bAu was added into the system as well as when both of these trigger strands were added. Typical TEM images of the outputs caused by diverse inputs are shown in Figures 3(d) and S11. The truth table of OR gate is listed in Table 3.

To develop the capacity of the logic device for large-scale operations and to verify the extensibility of this capacity, a three-input composite logic gate was constructed through uniting YES gate A and OR gate. Figure 4(a) is the scheme of the composite gate. Au-A and Au-B were simultaneously assembled on TL (TL-AB), which was identical in the preparation of the AND gate. Subsequently, three input strands (A, bst and bAu) were utilized to unbind the AuNPs on TL. In the same way, we defined a Au-free TL as a true output and marked it as “1” in the truth table. Interestingly, the composite gate circuit could be the equivalent of an OR gate with a switch control, where A was regarded as the switching variable, while bst and bAu constituted a logical variable. First, when A was absent from the system, no matter how bst and bAu were used as input, there was no true output. As illustrated in Figure 4(b), if the values of A were 0, bands representing AuNPs-binging-on-TL products always appeared in the output area (lanes 1-4, Figure 4(b)). Then, A was added into the solution, leading to a logic rule which was concordant with the OR gate. Bands vanished in the output area of the gel image as long as bst, bAu, or both of them were added into the system (lanes 6-8, Figure 4(b)). The truth table of the composite gate is listed in Table 4.

4. Conclusions

To sum up, we have successfully explored a logic operation device with excellent scalability based on assembling and selectively releasing AuNPs on DNA origami. In the computing process, the arrangement of AuNPs on DNA origami was controlled by specific DNA signals, which realized the logic operations. Using this strategy, we performed YES gate, OR gate, AND gate, and three-input composite gate through DNA strand displacement. The models were confirmed by gel electrophoresis and TEM images including statistical distribution of specific structures. It is worth mentioning that we established a progressive relationship between different logic gates. For example, we obtained AND gate by combining two YES gates and three-input composite gate by combining YES gate and OR gate. These processes have proved that this strategy has outstanding scalability and splicing.

Moreover, compared with previous work, the true output was distinguished by determining that there were AuNPs on DNA origami or not, which allowed us to read the output signal by observing whether the band disappeared in a relatively large-scale output area instead of comparing subtle dissimilarities in migration. That is to say, the resolution of the logic nanodevice is greatly improved. In addition, as a result of abandoning the 15 nm particle, which was used as a marker in the previous computing system, we ameliorated the utilization rate of AuNP as an effective information. Finally, our prototype provides a promising approach for building a multibit arithmetic unit taking biomolecules as the material and could inspire advanced designs for large-scale DNA logic circuits.

Data Availability

All data included in this study are available upon request by contact with the corresponding author.

Conflicts of Interest

The authors declare that there is no conflict of interest regarding the publication of this paper.

Acknowledgments

TEM analysis was supported by the Beijing Key Laboratory of New Technology in Agricultural Application. This work was supported by the Joint Fund of the Equipment Preresearch Ministry of Education (6141A02033607, 6141A02033608), the National Key R&D Program of China (2017YFGH001465), the National Natural Science Foundation of China (61872002), Beijing Natural Science Foundation (4182027), and Fundamental Research Funds for the Central University (2019JG001).

Supplementary Materials

Materials, experimental methods, and additional experimental data are provided in the supplementary materials. Figure S1: the detailed sequences of connections between DNA origami and AnNPs. Figures S2 to S5: the gel images of the concentration gradients of the trigger strand. Figure S6: scaffold/staple layout of the two-layer rod origami (TL). Figure S7: TEM image of TL. Figures S8 to S11: representational TEM images of various output structures. Table S1: the statistical results of TEM images. Table S2: the sequences of TL’s staples. Table S3: other DNA sequences used in this work. Table S4: thermal annealing ramp for TL. Table S5: fitting parameters for the statistical histograms of various output structures. (Supplementary Materials)