To verify the activity of DSN in our system, we tested the cleavage activity of DSN enzyme on the linear probe-miRNA complex at 37℃. The results showed that (Fig1) the experimental group produced clear sgRNA bands (indicated by the arrow), which indicates that DSN can function normally and efficiently in the first-generation system we designed. Fig1. Denaturing PAGE. The arrow denotes sgRNA.

        In order to verify the feasibility of CONAN system for detecting trace nucleic acids, we simply repeated the experiment of CONAN system (Fig2). We put 1uM Cas12a, 50nM sgRNA, 500nM assistant DNA and 1uM scgRNA-F in the experimental groups, but not sgRNA in the control group. The data was obtained by enzyme labeling instrument. Through four groups of parallel controls and one group of blank control, we found that the CONAN system has certain signal amplification ability.         Fig2. The CONAN system is used for exponential magnification. 20uL reaction system: 1uM Cas12a, 50nM sgRNA, 500nM assistant DNA, 1uM scgRNA-F. The rapid and obvious increase of fluorescence signal could be seen in the experimental group.

        Considering the cost and environmental protection of the detection system, our team decided to improve the CONAN system to find a better reaction concentration of Cas12a. We preset the sgRNA of 200nM in the system, add 5nM, 10nM, 20nM, 40nM, 100nM, 200nM (sufficient) Cas12a respectively, and fully combine Cas12a with sgRNA through incubation. Combined with assistant DNA, trans-cleavage activity is activated to cleave scgRNA-F, and the exponential amplification reaction of CONAN system occurs (Fig3).
        From the results, it can be seen that when the concentration of Cas12a in the system is 100nM, it is not very different from 200nM (saturation). Considering the convenience and cost of adding liquid, the 100n MCas12a reaction system was often used in the following experiments.         Fig3. The use of Cas12a trans-cleaving activity of ssDNA-F to explore the optimal amount of Cas12a. 50uL reaction system: 200nM sgRNA, 500nM assistant DNA, 1uM ssDNA-F. Input different concentrations of Cas12a as shown in the illustration, and comprehensively determine that the best working concentration of Cas12a is 100nM.

        After determining the dosage of Cas12a, in order to predict the detection threshold of CONAN system under our laboratory conditions, we set a series of sgRNA concentration gradients in each experimental group, which were 10E-4nM, 10E-2nM, 10E-1nM, 0nM and 1nM, and then added 100nM Cas12a and 100nM scgRNA-F for reaction. The results show that 10E-4nM, 10E-2nM and 0nM almost overlap (Fig4). So at least it can be presumed that the threshold of CONAN is less than or equal to 10E-1nM under our laboratory conditions.         Fig4. Explore the threshold of CONAN system in detection. 50uL reaction system: 100nM Cas12a, 200nM assistant DNA, 1uM scgRNA-F. Input different concentrations of sgRNA as shown in the illustration, and the threshold in this reaction environment is about 10E-1nM.

        After verifying the feasibility of the system and obtaining the prediction of Cas12a dosage and threshold, we tested different concentrations of miRNA. 10E-4nM, 10E-2nM, 10E-1nM and 1nM miRNA were added into the reaction system and fully combined with probe- α _ LP, and then sgRNA was released by DSN enzyme cleavage. Cas12a captured the sgRNA signal and conducted the CONAN system reaction to release the fluorescence signal. (Fig5)
        Unfortunately, the results showed that there is no significant difference in the output signal when the system detects different concentrations of miRNA. The rule is also irregular, and even the curve of low concentration group is above the curve of high concentration group, which means we cannot correctly distinguish different concentrations of input microRNA.         Fig5. The concentration of target miRNA was measured by using the combined system of CONAN system, DSN and circular probe. 50uL reaction system: 100nM Cas12a, 50nM probe-α_LP, 200nM assistant DNA, 1uM scgRNA-F. Add the target miRNA as shown in the legend. The differentiation between groups is not good.
        We analyzed the reason for the poor effect of the first time, and speculated that it’s the functional redundancy of the components of the CONAN system that led to the poor detection effect between groups. In order to change the detection object from DNA to RNA and further optimize the detection threshold, we introduced the DSN enzyme and probe- α. We found that both scgRNA-F and the introduced probe can undertake the task of exponential amplification, and probe- α can also realize the linear amplification, so the existence of scgRNA-F is superfluous. Therefore, we removed the scgRNA-F from the CONAN system and launched our DRJ system.

        In order to control variables and observe the effect of the new system, we added different concentrations of sgRNA to simulate the cleaving amplification effect of DSN enzyme. Controlling the concentration of Cas12a to 100nM, we put 10nM, 1nM, 10E-1nM and 0nM sgRNA into DRJ system respectively, and observed the results by fluorescence (Fig6).
        The results showed that compared with the first iteration, the system has a correct rule for different concentrations of sgRNA, and there is a more obvious degree of differentiation when the input concentration of sgRNA is greater than or equal to 10E-1nM.         Fig6. Input sgRNA to simulate the cleaving effect of DSN, and use DRJ system to amplify exponentially. 50uL reaction system: 100nM Cas12a, 50nM probe-α_LP, 200nM assistant DNA, 400nM scgRNA-F, and add sgRNA as shown in the illustration. The Cas-only group contained only Cas12a, assistant DNA and ssDNA-F, but no sgRNA or probe, as is the Case in the later experiment.

        After the simulation, we used this system to detect different concentrations of miRNA. MiRNA with the concentration of 10nM, 1nM and 10E-1nM was added into the system respectively and fully combined with probe-α_LP, and then sgRNA was released by DSN enzyme cleavage. Cas12a captured the sgRNA signal and conducted the DRJ system reaction to release the fluorescence signal (Fig7).
        Unfortunately, there was no significant difference between different gradients, and the system still didn’t not correctly reflect the amount of miRNA input. We analyzed the results and doubted the effect of sgRNA modified by DNA, so we decided to use pure RNA sgRNA in the new round of project.         Fig7. Use the DRJ system for exponential amplification. 50uL reaction system: 100nM Cas12a, 50nM probe-α_ LP, 200nM assistant DNA, 400nM scgRNA-F. There was no significant difference between different gradients when miRNA was added as shown in the illustration.

        In order to control variables and observe the effect of the new system, In order to control variables and observe the effect of the new system, we added different concentrations of sgRNA to simulate the cleaving amplification effect of DSN enzyme. Controlling the concentration of Cas12a to 100nM, we put 10nM, 1nM, 10E-1nM and 0nM sgRNA into DRJ system respectively, and observed the results by fluorescence.
        Similar to the second iteration, the system has a correct rule for different concentrations of sgRNA. When the input sgRNA concentration is greater than or equal to 10E-1nM, there is a more obvious degree of differentiation. However, when the concentration is low, the differentiation is still not obvious. Fig8. Add sgRNA to simulate the cleaving effect of DSN, and use DRJ system to amplify exponentially. 50uL reaction system: 100nM Cas12a, 50nM probe-β_ LP, 200nM assistant DNA, 400nM scgRNA-F, and add sgRNA as shown in the illustration.

        After the simulation, we used this system to detect different concentrations of miRNA. MiRNA with the concentration of 10nM, 1nM and 10E-1nM was added into the system respectively and fully combined with probe- β _ LP, and then sgRNA was released by DSN enzyme cleavage. Cas12a captured the sgRNA signal and conducted the DRJ system reaction to release the fluorescence signal.
        The results showed that the difference between groups was more obvious than before, and there was no overlap between groups, indicating that the sgRNA provided by probe- β was better than that of probe- α. However, the rules between groups are irregular, which provoked our reflective thoughts. It occurred to us that the sequence of sgRNA will self-fold. It means that the linear probe- β with complete sgRNA sequence cannot maintain the linear state in space, but will perform base pairing, self-folding to form a more mature sgRNA, and then bind and activate Cas12a, which leads to the generation of this inexplicable curve. In order to solve the problem of probe- β _ LP self-folding, we decided to cyclize it in the next round of engineering iteration.         Fig9. Use the DRJ system for exponential magnification. 50uL reaction system: 100nM Cas12a, 50nM probe-β_LP, 200nM assistant DNA, 400nM scgRNA-F. Add miRNA as shown in the legend, and the rule between the groups was irregular.

        Through literature review, we found two methods to cyclize probes: (1) Use T4 DNA Ligase to cyclize with the assistance of a short primer; (2) Use Cric Ligase to cyclize. (Fig10) Fig10. Two cyclization methods for DNA-RNA chimeric probe.
        In order to save cost, we first replaced the ribonucleotide in the probe with deoxyribonucleotide, synthesized a phosphorylated linear probe sequence, and tried to cyclize it with Cric Ligase, which is most commonly used for single-strand nucleic acid cyclization. However, as is shown in the urea denaturation PAGE results (Fig 11), the cyclization efficiency of Cric Ligase is not high enough and the cost of this cyclization method is high. Fig11. Denaturing PAGE. The bands shift upward after probe circularization.
        To reduce the cost and further improve the cyclization efficiency, T4 DNA ligase was used and a sequence specific primer was introduced to cyclize the probe. PAGE results (Fig12 a) and HPLC results (Fig12 b-c) showed that the cyclization method using T4 DNA ligase had very high efficiency, reaching nearly 100% cyclization efficiency. Next, we purified the circular product using Exonuclease III and Exonuclease I. Because we also wanted to use Exonuclease I for further processing of sgRNA in the DSN system (see design), we retained the activity of Exonuclease I and did not inactivate it after purification of cyclization products.         Fig12. a. Denaturing PAGE. The bands shift upward after probe circularization. b - d. Results of HPLC. The cyclization products were cross-validated by HPLC. b. Experimental group: Composition of the product which is cyclized by T4 DNA ligase for two hours. c. Composition of the product purified by exonuclease. The substance with a peak time of about 30min is circular DNA-RNA chimeric probe. d. The control group. The substance with a peak time of about 26 minutes is the primer, and the substance with the peak time of about 29.5 minutes is the linear DNA-RNA chimeric probe to be circularized.
        Next, we cyclized the circular DNA-RNA chimeric probe with T4 DNA ligase and obtained the purified circular product (Fig13).          Fig13. Denaturing PAGE. The bands shift upward after probe circularization. Arrow denotes circular DNA-RNA chimeric probe.

        In order to verify the activity of DSN in the modified system, we tested the cleavage activity of DSN enzyme on the circular probe-miRNA complex at 37℃ (Fig14).The results showed that the experimental group produced a distinct sgRNA bands(the arrow indicated band), indicating that DSN can function normally and efficiently in our modified system. We also compared the linear probe with the circular probe. Under the same reaction conditions, after the miRNA-linear probe complex was cut by DSN, the amount of remaining linear probes was reduced, but no obvious sgRNA band was observed. After the miRNA-circular probe complex cleaved by DSN, an obvious sgRNA band was generated. This shows that the circular probe is more stable and more efficient. Fig14. Denaturing PAGE. Arrow denotes sgRNA

        In order to control variables and observe the effect of the new system, we added different concentrations of sgRNA to simulate the cleaving amplification effect of DSN enzyme. Controlling the concentration of Cas12a to 100nM, we put 10nM, 1nM, 10E-1nM and 0nM sgRNA into DRJ system respectively, and observed the results by fluorescence (Fig15).
        To our excitement, the new system not only has fast response speed, observing a significant and reasonable difference between groups in around 2000 s, but also made the negative control group at a significantly lower level than that of the experimental group. This gave us great confidence and data support for our final experiment. We finally decided to challenge a lower threshold.         Fig15. Input sgRNA to simulate the cutting effect of DSN, and use DRJ system to amplify exponentially. 50uL reaction system: 100nM Cas12a, 50nM probe-β_CP, 200nM assistant DNA, 400nM scgRNA-F, add sgRNA as shown in the illustration.

        After the simulation, we used this system to detect different concentrations of miRNA. MiRNA with the concentration of 100nM, 1nM and 10E-2nM was added into the system respectively and fully combined with probe- β _ LP, and then sgRNA was released by DSN enzyme cleavage. Cas12a captured the sgRNA signal and conducted the DRJ system reaction to release the fluorescence signal (Fig16).
        It can be seen from the diagram that the system can correctly reflect the correct relationship of different input concentrations of miRNA, and has a better improvement than the previous three iterations, which indirectly showed that circular probe improves the work efficiency of DSN. At the same time, we found that there is still a large gap in the curves between 10E-2nM miRNA and 0Nm miRNA, indicating that under our laboratory conditions, the theoretical threshold of this system is lower than that of 10E-2nM, which is an exciting progress compared with the initial threshold of 10E-1nM.         Fig16. Use DRJ system for exponential amplification. 50uL reaction system: 100nM Cas12a, 50nM probe-β_CP, 200nM assistant DNA, 400nM scgRNA-F. Add miRNA as shown in the illustration and we can better distinguish the target miRNA of different concentrations.

        By modifying -SH at both ends of 30nt single-stranded DNA, we successfully achieved the cross-linking of free colloidal gold particles in 20nm diameter through the formation of Au-S bonds, and the good cross-linking effect can be seen under electron microscopy (Fig17 a、b).         Fig17. a: (TEM photo) The colloidal gold particles in the solution before crosslinking are relatively dispersed; b: (TEM photo) The colloidal gold particles are aggregated together after crosslinking.
The color of colloidal gold solution also changed significantly before and after cross-linking (Fig18a, b). We mixed the colloidal gold solutions before and after crosslinking in different proportions to simulate the possible solution color after cleaving of different amounts of cas12a output in the previous reaction. From our experimental results (Fig18c), it can be seen that the solution color did change. By controlling the concentration ratio of colloidal gold before and after cross-linking in the solution, we made a standard colorimetric card, and through machine learning, our APP obtained the ability to recognize different colors, that is, to recognize different concentrations of colloidal gold solutions.         Fig 18c. From left to right, the proportion of colloidal gold after crosslinking is 0% to 100%
Fig18. a: Before crosslinking, the solution color appears wine-red;
b: After crosslinking, the solution color appears purple;
c (from left to right): As is shown in the figure, the solution was added according to the ratio in the table above to simulate the possible solution color after different amounts of cas12a entered the solution to cleave ssDNA. The figure shows that the solution color did change.

        We used activated Cas12a to cleave the colloidal gold particles after crosslinking, hoping that the color change will return to the pre-crosslinking state. Unfortunately, it didn't turn out the way we had hoped (Fig 19). At the same time, we found that the enzyme buffer, colloidal gold precipitation, wall hanging and other factors would also cause unpredictable effects on the results.         Fig19. a:From left to right are 80uL colloidal gold before crosslinking, 80uL colloidal gold after crosslinking, 80uL colloidal gold after crosslinking，80uL colloidal gold after crosslinking，respectively.
b：Add cas12a, buffer, sgRNA, assistant DNA in the 2nd and 3rd tubes but 1st and 4th tubes without cas12a.And make up the difference volume with DEPC H2O.The picture shows the addition of instantaneous color changes.
c：The picture shows the color changes for the reaction lasting 30min.
d：The picture shows the color changes when the reaction lasting in the PCR instrument at 37 ℃ for 1 h.