2.1. The Fundamental Principle of the Toeless Triplex-Mediated Strand Displacement Reactions
Firstly, we validated the basic principle of the reaction. We labeled the quencher BHQ1 at the end of Output 1 (O1) (denoted as O1-BHQ1), and the fluorophore FAM at the end of O1*. We hybridized O1 and O1* at a concentration of 1 µM as the reaction substrate. We added 1 µM unlabeled Invading strand 1 to experimental group 1, and added simulated leaky strand 2 without the triplex binding region to experimental group 2. Both experimental groups received 1 µM of reaction substrate at the same time. In addition, we set 1 µM reaction substrate as the negative control group and 1 µM O1* strand as the positive control group. Subsequently, we conducted the reactions for 100 minutes at pH 7 and 37°C for each group, while detecting the fluorescence intensity of each group at 1-minute intervals (with the positive control group set as the reference value of 1 or 100% in all experiments). Figure 1b indicates that, the reaction rate of experimental group 1 is significantly faster, compared to that of experimental group 2, suggesting that Domain A can mediate the toeless DNA displacement reaction. Next, to confirm that the reaction mechanism of toeless DNA displacement reaction is indeed as shown in Fig. 1a, we labeled fluorophore FAM at the boundary of Domain A and Domain B of Invading strand 1(denoted as Invading strand I-FAM). In the experiment, we added 1 µM Invading strand 1-FAM to the prehybridized substrate of 1 µM O1-BHQ1 and O* as the experimental group, and 1 µM Invading strand 1-FAM was set as the control group. Subsequently, we conducted the reactions for 120 minutes at pH 7 and 37°C for each group, while detecting the fluorescence intensity of each group at 1-minute intervals. As shown in Fig. 1c, compared with the control group, the fluorescence intensity of the experimental group showed a significant decrease followed by an increase, indicating that the rapid binding of Domain A and polypurine sequence mediated the subsequent strand displacement reaction. We believe that the length of Domain A is a key factor affecting the kinetics and thermodynamics of toeless triplex-mediated DNA displacement reaction. Increasing the length of Domain A not only enhances the binding strength between Domain A and Domain T, but also increases the difficulty of dissociation of Output 1 strand. Therefore, theoretically, there is an equilibrium point for the dissociation and binding tendency, which corresponds to a moderate value of the length of Domain A. Deviation from the optimal length in Domain A, either longer or shorter, leads to a decrease in reaction efficiency. Therefore, we further vary the length of Domain A and Domain T to explore the effect of triplex domain length on the degree of reaction. According to the results in Figure S1, this equilibrium is reached when the length of triplex domain is 9 nucleotides. In conclusion, we have successfully constructed a toeless triplex-mediated DNA strand displacement reaction with aPHB.
Next, we attempted to construct reaction mode 2 using PHB. And based on reaction mode 2, we further constructed a pH-responsive and reversible reaction. As shown in Fig. 1d,e, at pH around 5, Invading strand 2 forms PHB with O2* and displaces Output 2. We designed a dissociation domain Toehold* at the end of Output 2, and the complementary domain is denoted as Toehold. When the pH increases to around 9, the PHB in the triplex domain dissociates. At this point, the Toehold can mediate the re-hybridizing of Output 2 with O2* and displace Invading strand 2, thus completing the reverse of the reaction. Before conducting the reversible reaction study, we aimed to investigate the influence of the length of Domain C, which forms the triplex structure, on the reaction kinetics and thermodynamics, as shown in Fig. 1e. 1 µM Invading strand 2 of different Domain C lengths is used to react with 1 µM of Output 2-O2* duplex complex as experimental groups. We also set a negative control group of 1 µM Output 2-O2* duplex complex and a positive control group of 1 µM Output 2 only. Subsequently, we conducted the reactions for 360 minutes at pH 7 and 37°C for each group, while detecting the fluorescence intensity of each group at 1-minute intervals. We discovered that the reaction could only occur when the length of Domain C was less than 6 nt (Fig. 1f). Moreover, the reaction was most efficient when the length of Domain C was reduced to 0 nt. We further studied the reactions when the length of Domain C was 6 nt, 5 nt, and 4 nt (Figure S2), which confirmed that the length of Domain C required for the reaction to occur was indeed at least 6 nt. Similar to reaction mode 1, we also explored the effect of reducing the length of Domain A on the efficiency of the reaction. When the length of Domain C was fixed at 0 nt and the dissociation region length was constant, we found that the efficiency of reaction mode 2 rose as the length of Domain A decreased (Figure S3). Through these two experiments, we demonstrated that the difficulty of output strand dissociation from the original triplex domain significantly impacted on the efficiency of the reaction mode 2, which provided feasibility for accurate manipulation of this reaction. Based on the above experiments, we explored the reversibility of this reaction mode by adjusting the pH. As shown in Fig. 1g, we adjusted the pH of the system from 5 to 9 and continued to detect fluorescence at 1-minute intervals, 37°C. We observed a rapid decrease in fluorescence signal, indicating that the reverse reaction occurred and the reaction system was reverse. This result validated the feasibility of the reaction strategy shown in Fig. 1d. In summary, we successfully constructed a reversible toeless strand displacement reaction using PHB.
2.2. Enhancement of Kinetic Efficiency in Toeless Triplex-Mediated Strand Displacement Reactions
The above two reaction modes have achieved toeless blunt-end DNA strand displacement reaction. However, their reaction kinetics are not ideal, and the reaction degree is generally less than 30%, which limits their further application and expansion. Hoogsteen-bond must rely on the Waston-Crick bond to exist stably. Thus, both of the aforementioned toeless triplex-mediated strand displacement reactions involve a repeated process of Hoogsteen-bond dissociation and formation, which greatly increases the energy barrier of each strand displacement step and leads to a decrease in the reaction rate (Fig. 2a). Therefore, through further optimization, we designed the following two reaction modes to solve the above problems. First, the principle of reaction mode 3 is shown in Fig. 2b. After the 1# domain of Invading strand 3 forms a PHB with the 1* strand, the 2# domain will replace the Output 3 − 1 (2*) strand and bind to the 3 strand. At the same time, the 2* strand will react with the reporter system labeled by FAM fluorophore and BHQ1 quencher, generating a fluorescence signal. This reaction does not involve the repeated process of Hoogsteen-bond formation and dissociation. The energy barriers that need to be overcome for each base substitution step in this strand displacement reaction are lower. Thus, theoretically, the efficiency of reaction mode 3 is higher than mode 1 and mode 2, and the reaction rate is also faster. At the same time, the strength of PHB is related to the degree of protonation of the pyrimidine in strand 1*. Hence, by gradient adjusting the length of the triplex domain in the strand displacement reaction and the dissociation region length in the strand migration reaction, we can achieve precise control over the kinetics and thermodynamics of the reaction. We added 1 µM of Invading strand 3, which have different lengths of 10–20 nt in Domain 1, to the pre-formed 1*-2*-3* duplex complex with a concentration of 1 µM, and added 1 µM of reporter system to each group as the experimental group. The negative control group consisted of 1 µM 1*-2*-3* complex without Invading strand 3 and 1 µM reporter system, and the positive control group consisted of only 1 µM Output 3 − 1 and 1 µM reporter system. Subsequently, we allowed each group to react for 180 minutes at pH 5 and 37°C, and detected the FAM fluorescence intensity of each group every 1 minute. The experimental results in Fig. 2c show that the reaction efficiency of reaction mode 3 is generally between 70% − 90%, and can be completely reacted within 2 hours. Compared with reaction modes 1 and 2, reaction mode 3 does have a higher reaction efficiency and a faster reaction rate. At the same time, we observed that as the length of Domain 1 in strand 3 increased, the efficiency of the triplex strand displacement reaction decreased. We speculate that this is due to a significant increase in the invading reaction energy barrier with an increase in the length of the triplex domain. This result reveals that an increase in the length of the triplex domain significantly increases the binding energy barrier between the invading strand and the reaction substrate. In subsequent similar strand displacement experiments, the length of the triplex region to 10 nt will be set by default.
As shown in Fig. 2d, based on the aPHB not relying on protonated cytosine, we designed reaction mode 4 that is not influenced by pH changes. The A# domain of Invading strand 4 formed aPHB with A* and C strands, while its B# domain displaced the Output 3 − 2(B*) strand. The resulting B* strand reacted with the reporter system containing FAM-BHQ1 pair to generate fluorescence (related reporter system sequences shown in Supplementary Information 109 to 110). In addition to Invading strand 4 that can undergo complete strand displacement reaction with B*, we also designed Invading strands 4 with decreasing length at the 3' end. These strands can create a dissociation region at the end of the A*-B*-C* complex, for observing the length effect of the dissociation region on the reaction degree. We shortened the B# 3' ends by 1 nt for each chain, denoted as Invading strand 4-0nt to Invading strand 4-5nt (related sequences shown in Supplementary Information 10 − 3 to 10 − 8). Each of the Invading strand 4-n (n = 0–5 nt) strands was added to the 1 µM A*-B*-C duplex complex at a concentration of 1 µM as the experimental group. The negative control group was the 1 µM A*-B*-C* complex without Invading strand 4-n and the reporter system at a concentration of 1 µM, while the positive control group was only the 1 µM reporter system with 1 µM Output 3 − 2. Each group was allowed to react for 360 minutes at pH 7 and 37°C, and the FAM fluorescence intensity was detected at 1-minute intervals. As shown in Fig. 2e, the reaction degree of reaction mode 4 gradually decreased as the length of the dissociation region reserved at the B# 3' end increased. The highest reaction degree, which reached about 65%, was achieved when the length of the dissociation region was 0 nt, suggesting a complete chain displacement reaction in the system. Under reaction mode 4, the triplex structure has two binding patterns: T-A:T and A-A:T. We found that the rate of strand displacement reaction with T-A:T binding was slightly faster than that with A-A:T binding, as the result shown in Figure S4. Yet, when the length of the triplex domain and the dissociation region were kept constant, the final reaction degree was close to 65% regardless of whether we mixed adenine(A) or thymine(T) bases into the invading region of Invading strand 4. Therefore, we conclude that the principle of reaction mode 4 has been validated, indicating that our toeless triplex-mediated strand displacement reaction can occur independently of pH changes.
After verifying the reaction mechanisms of the aforementioned two reaction modes, we explored the reversibility of reaction mode 3. As seen in Fig. 3a, we added a 6-nt dissociation region to the 3 strand of reaction mode 3 (denoted the new strand as 3**) which can be used as a toehold-mediated strand displacement reaction in the reverse process. In addition, we also modified the reporter pair into a Cy5-BHQ2 combination to limit the effect of pH changes on the fluorescence of the reaction. We conducted a reaction at pH 5 and 37°C using 1 µM of Invading strand I and 1 µM of the 1*-2**-3 duplex complex, which eventually showed a similar forward reaction degree as reaction mode 3. The negative control group consisted of a 1 µM 1–2*-3** complex without the addition of Invading strand I and a 1 µM reporter system, whereas the positive control group consisted of only 1 µM Output 3–3 and 1 µM reporter system. After adjusting the pH of the system from 5 to 9, we immediately recorded the fluorescence intensity of each group at 1-minute intervals at 37°C. As shown in Fig. 3b, there is a rapid decrease in the fluorescence signal in the system from the platform value, indicating the reversible nature of reaction mode 3. Furthermore, we evaluated the correlation between reaction mode 3's reaction thermodynamics and pH. As illustrated in Fig. 3c and 3d, we observed that the degree of the above forward reaction decreased with increasing pH in the range of pH 5 to pH 6. This demonstrates that we can precisely regulate the reaction degree of reaction mode 3 by adjusting pH. Additionally, based on this principle, we demonstrated that the output strand of toeless triplex-mediated strand displacement reaction can serve as input for the same reaction, i.e., this type of strand displacement has the potential for modular self-compatibility, enabling the construction of cascade reactions. The sequence design and reaction results are shown in Figure S5a, 5b, 5c.
In summary, the toeless triplex-mediated strand displacement reaction modes that we have invented completely circumvents the leakage problem of the traditional single-strand base-exposed sequence in W-C sticky end toehold mediated strand displacement reaction, and incorporates both reversible and regulatory capabilities. Next, we shall illustrate its strong compatibility in DNA logic circuits.
2.3. Toeless Triplex-Mediated Strand Displacement Reactions for the Development of Reversible DNA Logic Gates
Based on reaction mode 3, we aimed to explore the possibility of constructing simple DNA circuit components, And gate and Or gate. As shown in Fig. 4a, the AND gate consists of Substrate A, Medium-A, Output-A, and two types of Input 1 strands. When Domain A on the Input 1A strand binds with the established substrate complex, Domain B in the migration region serves as a mediator, facilitating Input 1B strand to form Hoogsteen-bonds with Domain B. Therefore, only when both attached Input 1A and Input 1B are present in the system, can Domain A mediate the invading of the Input 1B strand and replace the Output-A strand. Subsequently, the Output-A strand and Cy5-BHQ2 reporter system react to produce a Cy5 fluorescence signal. We reserved a dissociation region at the end of the Substrate A strand. When the system pH was adjusted to 9, all triplex region will be dissociated. At this point, Input A falls off naturally, and Output-A can replace Input 1B strand for the reaction to be reversed. We employed the Substrate A/Medium-A/Output-A duplex complex as the reaction substrate. In the first experimental group, 1 µM Input 1A and 1 µM Input 1B were introduced to the 1 µM substrate as signal (1 1). In experimental group 2 (1 0) and experimental group 3 (0 1), we introduced either 1 µM Input 1A or 1 µM Input 1B respectively to the same 1 µM substrate. The reporter system was added at a concentration of 1 µM in each experimental group. We used the 1 µM substrate complex without any input and 1 µM reporter as negative controls (0 0), and the system only with 1 µM Output-A and 1 µM reporter as positive controls. We adjusted the pH to 5 in each group, and recorded the fluorescence intensity every 1 minute at 37°C (recorded every 5 seconds within 6 minutes after each addition). As shown in Fig. 4b, we observed a rapid and significant fluorescence signal in the (1 1) reaction group, while each individual Input 1A or 1B alone did not generate a significant signal. In addition, there was a rapid reverse of the reaction system after pH adjustment to 9. These experimental results demonstrate the successful construction of our reversible toeless triplex-mediated AND gate.
The reaction mechanism of the OR gate is illustrated in Fig. 4c. We used two partially identical sequences, Input 1C and Input 1D, to represent two input signals 1. Both Input signals can trigger the same strand displacement reaction and lead to the same Output-O output. We defined Output-O as responding to the Cy5-BHQ2 reporter system and generating Cy5 fluorescence as output signal 1. This logic gate can also be reversed by the toehold domain reserved on the substrate O strand when the reaction system is raised to pH 9. We combined Substrate O/Medium-O/Output-O duplex complex as reaction substrate. In experimental group 1 (1 1), 1 µM substrate, 1 µM Input 1C, and 1 µM Input 1D were added. In experimental group 2 (1 0) and group 3 (0 1), we added 1 µM substrate and 1 µM of either Input 1C or 1 µM Input 1D, respectively. Moreover, we added 1 µM reaction substrate and 1 µM reporter system to the negative control group (0 0). At the same time, we added 1 µM Output-O and 1 µM reporter system to the positive control group. Finally, we measured fluorescence intensity every minute at pH 5 and 37°C for each group (recorded every 5 seconds within 6 minutes after each addition). The results shown in Fig. 4d demonstrated that all experimental groups exhibited significant fluorescence signals, while the negative control group showed almost no signal. Once the pH was adjusted to 9, all reaction systems were reversed. This experimental result proved that we have successfully constructed a reversible toeless triplex-mediated OR gate.
The experimental results of the above AND and OR gates demonstrate that modular and reversible DNA circuit logic gate components can be constructed based on the toeless triplex-mediated DNA displacement mechanisms utilized in this study. This reverse does not generate any type of waste, and thus does not adversely affect subsequent computations.
2.4. Development of a DNA Walking Machine using Two-Orientation Toeless Triplex-Mediated Strand Displacement Reactions
The validation of the aforementioned reaction concept leads us to believe that our toeless triplex-mediated strand displacement reaction can undertake more complex tasks. Consequently, based on the toeless triplex-mediated strand displacement mechanism, we designed a DNA nano walking machine, namely DNA Walker. The operating concept of this Walker integrates reaction modes 3 and 4 (In PHB and aPHB mechanism, the orientation of invading domain is opposite). We accomplished the alternating movement of this walking structure by adjusting the design of input and the sticky end sequences of track strand. Figure 5a demonstrates that the walking strand is composed of Leg A, Leg B, and the connecting domain. We have labeled the quencher BHQ1 at both ends of Leg A and Leg B. The substrate Track complex has four free sticky single-strand ends, Track1 to Track4, with FAM fluorophore labeling on the free sticky single-strand ends of Track 2 to Track 4. Prior to the start of the reaction, we added the walking strand into the previously assembled Track complex as the initializing state. In State 1, Input 1 was introduced into Track complex to induce detachment of Leg A. In State 2, Leg A was hybridized with Track 3 resulting in quenching of the fluorophore at the end of Track 3. In State 3, Input 3 was added to unquench the fluorophore on Track 2. Subsequently, in State 4, Leg B was hybridized with Track 4 through the addition of Input 4 followed by quenching of the fluorophore on the Track 4. These four states constitute a complete walking cycle: Leg A and Leg B initially bind to Track 1 and Track 2, respectively, and ultimately move to Track 3 and Track 4, respectively. Given that the conformational changes of the Track complex may cause interference between FAM fluorescence signals (Figure S6), we set up three independent tubes. The Track complex structures in each tube are the same, but the positions of the FAM fluorophore located correspond to three different endpoints of the steps. We marked the observation of fluorescence at a certain position as "on" and the disappearance of fluorescence as "off". In the above reactions, the concentrations of each reactants were all 1 µM, and the reaction conditions were pH 5 and 37°C. The results in Fig. 5b showed that after adding of inputs at each step, the fluorescence signals of all three tubes exhibited the expected changes, which shows that each reaction proceeded as expected, and the walking cycle was finally completed. Additionally, the fast reactions between State 1/State 2 and State 3/State 4, as well as the relatively slow increase in fluorescence between State 2/State 3. This further confirmed that duplex hybridization, toeless triplex-mediated strand displacement, and another duplex hybridization occurred in the three State sequentially. The results suggest that an active DNA nanomachine can be constructed based on the reaction modes of our experiment. Theoretically, various cargoes can be connected to the connecting domain to form a cargo complex through covalent or Watson-Crick bonding modification. By designing the sequences of the sticky regions of Input No.n and Track No.n in each step, specific hybridization can be achieved, thereby controlling the stable transportation of the cargo complex on a Track complex of arbitrary length. The construction of the aforementioned logic gates and walker demonstrates that our reaction strategy can be flexibly applied to various DNA nanomachine, with high modularity and highly adjustable reaction processes, and has a great application potential.