Molecular computing is an emerging paradigm with important roles in data storage, biocomputing, and clinical diagnostics, showing promising development prospects. Meanwhile, synthetic DNA, as a programmable polymer, is capable of storing and processing information, from logic circuits to neural networks, to realize multifunctional computing devices.
Therefore, in order to realize accurate nucleic acid detection and automated intelligent analysis, we constructed a classifier based on DNA integrated circuits that performs computations based on artificial biochemical amplification circuits at the molecular level for medical diagnosis.
In terms of design, we constructed a universal detection platform for nucleic acid markers with polymerase self-enhancement and CRISPR two-step reaction, using polymerase-mediated strand displacement reaction, toe-mediated strand displacement reaction and CRISPR-Cas system as the core principles. The platform has the advantages of no PCR, high accuracy and adjustable high sensitivity.
The platform enables the identification, amplification, fluorescence reporting and single-molecule level decision-making of nucleic acid targets, and ultimately accurately transforms nucleic acid signals into intuitive two-dimensional results, thus realizing accurate diagnosis of diseases.
In the field of early disease detection, it is crucial to select appropriate biomarkers. microRNAs are a class of endogenous, non-coding small RNA molecules with a length of about 20-25 nucleotides. They are widely found in living organisms and play important roles in a variety of physiological and pathological processes. Compared with biomarkers such as mRNAs or proteins, microRNAs are more stable in the blood and are less likely to be degraded, making them more suitable as markers for in vitro detection. At the same time, the relatively small molecular structure of microRNAs makes them relatively simple and convenient for sample processing, extraction and detection. Therefore, we chose microRNAs as biomarkers for accurate disease detection.
We obtained microRNA expression profile data of the corresponding diseases from NCBI, calculated the correlation coefficient between the expression of each each microRNA and the patient's disease status, and screened out the targets with Top5 correlation coefficients. Then, machine learning algorithms such as Linear Regression, Logistic Regression, etc. were used to evaluate the detection effect of these candidate targets. The ROC curve, AUC value and confusion matrix are comprehensively utilized to select the target with the best effect.
In the signal recognition stage, we designed Rep strand as the recognition element.The 3' end sequence of Rep strand is complementary to the microRNA target and can specifically bind to the target microRNA. This binding not only improves the accuracy of recognition, but also ensures the smoothness of the subsequent signal amplification and detection process.
In the primary amplification system, we first utilize the polymerase-mediated strand displacement reaction for initial signal amplification. Specifically, when the miRNA target binds specifically to the Rep chain, the Rep chain undergoes a polymerase-mediated strand displacement reaction under the action of Bst polymerase, thereby displacing the H chain. Subsequently, the free H strand binds complementarily to the 3' end sequence of the DDSD long probe strand, and under the action of Bst polymerase, a polymerase-mediated strand displacement reaction occurs again, displacing 1 I strand and a specific number of wt strands. This process achieves an initial amplification of the signal and improves the sensitivity of the assay.
In order to convert the amplified signals into detectable fluorescent signals, we designed two kinds of reporter probes modified with fluorescent groups and quenching groups, which were used in the secondary signal amplification system based on the strand substitution reaction and the CRISPR system, respectively. When the interaction between the fluorescent group and the quenching group of the reporter probes is broken, the generated fluorescence signal is gradually enhanced. This change in fluorescence signal can be monitored and recorded in real time by single-molecule fluorescence microscopy.
- Polymerase self-enhanced binding chain substitution system
In this system, we introduced the quenching probe RP chain and the fluorescence probe F chain. In the initial state, the 3' end of the RP chain binds complementarily to the F chain, and the fluorescent group on the F chain is quenched due to its proximity to the quenching group on the RP chain. When the wt chain generated by the primary amplification system binds to the 3' end of the RP chain, due to the length of the wt chain is longer than the F chain, it will displace the F chain through the toe-mediated chain displacement reaction, which separates the fluorescent moiety from the quenching moiety, and thus restores fluorescent signal. This process realizes the secondary amplification of the signal and enhances the accuracy of detection. - Polymerase self-enhanced binding CRISPR system
In this system, the wt strand binds to cas12A enzyme together with crRNA to activate its non-specific DNA endonuclease paracrine activity. Subsequently, the cas12A enzyme cleaves the reporter probe modified with a fluorophore and a quenching moiety, cutting the reporter probe into two parts, resulting in the separation of the fluorophore from the quenching moiety and its dispersion in solution, which restores and significantly enhances the fluorescent signal. This mechanism not only realizes the secondary amplification of the signal, but also ensures the accuracy of the detection results through the high specificity of CRISPR technology.
In order to optimize the efficiency of fluorescent signal generation, we conducted several experiments and adjusted the concentration, structure, and reaction conditions of the fluorescent probe. The experimental results show that under the optimal conditions, there is a good linear relationship between the fluorescence signal and the concentration of microRNA target, and the background noise is low, which meets the needs of high sensitivity detection.
In order to more accurately quantify the biomarker content in the samples, we designed a single-molecule fluorescence resonance energy transfer (smFRET)-based reporter system and used total internal reflection fluorescence microscopy (TIRFM) for the determination. The system consists of the following key steps:
Sample preparation and fixation: In order to solve the problem of free and immobile fluorescent molecules in liquid environment, we used physicochemical means to fabricate slides modified with APTE negatively charged groups. Utilizing the negative electronegativity of the APTE groups, we immobilized the fluorescent molecules on the slide surface. At the same time, in order to control the volume of the test sample and to minimize errors, we created “chambers” immobilized on the surface of the slide to hold a fixed volume of the solution to be tested.
Continuous dynamic photography: During the detection process, we utilize a single-molecule fluorescence super-resolution microscope to photograph the samples in the “chamber”. By randomly selecting multiple fields of view and taking 100 consecutive pictures in each field of view in a dynamic manner, we ensure that the fluorescent spots are captured comprehensively and the data are reliable. This method of capture not only reduces the errors associated with field of view selection and single shots, but also improves the accuracy and repeatability of the test results.
Image processing and analysis: After shooting, we use the single-molecule fluorescence image processing program developed by our team to automatically process and analyze the images. The program can enhance and extract features from the captured images to identify effective fluorescent spots in the single-molecule fluorescence images. By calculating the number of effective fluorescent dots per unit field of view area, we obtained the exact amount of biomarkers in the sample.