AND logic-gated CRISPR/Cas9 and hybridization chain reaction system for precise ctDNA detection

Abstract

Circulating tumor DNA (ctDNA) is a critical biomarker for liquid biopsies, enabling the non-invasive acquisition of cancer-related information from blood samples. Precise detection of ctDNA, particularly the identification of single-nucleotide variations (SNVs), is crucial for early cancer diagnosis, therapeutic monitoring, and prognostic evaluation. However, current ctDNA detection methods often encounter challenges such as complex procedures, difficult data analysis, and false-positive signals during pre-amplification. In this study, we introduce a novel detection method based on AND logic-gated integration of interspaced short palindromic repeats and associated proteins (CRISPR/Cas9) system with hybridization chain reaction (HCR) isothermal amplification. This strategy enhances the specific and sensitive detection of ctDNA. The incorporation of the AND logic gate effectively minimizes the off-target effects of Cas9 and enables the differentiation of single-nucleotide mutations, such as KRAS G12D, even in complex serum environments. Our system exhibits high sensitivity and specificity, achieving a limit of detection as low as 1 fM and capable of identifying SNVs mutations with allele fractions as low as 0.1% among wild-type sequences. Furthermore, we validated the specificity of our approach by successfully detecting various mutations, including KRAS G12C, KRAS G12D, EGFR T790M and TP53 R273H, in simulated clinical samples. These findings highlight a reliable method for precise ctDNA detection, offering high specificity, selectivity, and accuracy, thus paving the way for potential cancer diagnostic application.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-025-03912-y.

Introduction

Early tumor screening is a critical strategy for cancer prevention and treatment [9, 33]. Liquid biopsy has emerged as a promising tool for cancer detection, offering a non-invasive method to monitor tumor dynamics and treatment responses in patients through analyzing circulating tumor DNA (ctDNA), circulating tumor cells (CTCs), and other biomarkers in body fluids. Compared to conventional tissue biopsies, liquid biopsy provides real-time monitoring and is simpler and less invasive [12, 28]. Among the various biomarkers, ctDNA has become one of the important biomarkers in liquid biopsy. ctDNA originates from circulating cell-free DNA (cfDNA) fragments released by tumor cells through necrosis, apoptosis, or active secretion [2, 7, 29]. These fragments circulate in the bloodstream and carry genetic information about the tumor, such as mutations and gene deletions [6]. The detection of ctDNA enables early cancer diagnosis, treatment monitoring, and prognosis assessment, offering a non-invasive approach to obtaining cancer-related information from blood samples.

In cancer diagnosis, specific ctDNA mutations, such as those found in KRAS (Kirsten rat sarcoma viral oncogene homolog), TP53 (tumor protein 53 gene), and EGFR (epidermal growth factor receptor gene) genes, are crucial for identifying tumor subtypes and guiding treatment decisions. For example, KRAS mutations (e.g. G12D and G12C) are prevalent in various cancers and contribute to tumor proliferation and metastasis, particularly in colorectal cancer [1, 15]. Similarly, the R273H mutation in TP53, a key tumor suppressor gene, leads to uncontrolled cell growth and cancer progression, making these mutations important targets for precision medicine and personalized treatment [1, 42]. For instance, in non-small cell lung cancer (NSCLC), patients who initially respond to tyrosine kinase inhibitors (TKIs) may later develop the EGFR T790M mutation, leading to drug resistance [26, 30]. Thus, precise detection of ctDNA mutations allows for timely treatment adjustments, improving patient outcomes.

Despite the promise of ctDNA as a biomarker, its low abundance, especially in early-stage cancer, poses a significant challenge [27]. Highly sensitive and reliable detection methods are needed to overcome this limitation. Although next-generation sequencing (NGS) and quantitative real-time PCR (qPCR) are the current gold standards for detecting DNA mutations, these methods are time-consuming, costly and complex, with limited multiplexing capabilities [11, 18, 23]. Moreover, the low concentration of ctDNA in non-invasive samples, such as blood or urine, often falls below the detection limits of these methods [31]. Therefore, developing a simple, cost-effective, and highly sensitive approach for precise detection of ctDNA remains a critical research focus.

Isothermal amplification methods, including strand displacement amplification (SDA), rolling circle amplification (RCA) and hybridization chain reaction (HCR), offer an alternative by rapidly amplifying nucleic acids at a constant temperature [14, 39]. HCR, in particular, is an enzyme-free amplification technique that can achieve high sensitivity and specificity by utilizing repeated hybridization reactions between DNA hairpins [8]. The ease of operation, combined with its capacity to amplify small amounts of target molecules, make HCR a powerful tool for early cancer detection and treatment monitoring [3, 25, 34, 38, 41]. However, isothermal amplification methods are prone to have non-specific amplification, limiting their practical application [39]. Recently, the integration of isothermal amplification with CRISPR/Cas systems (e.g. Cas9, Cas12, Cas13, and Cas14) has revolutionized nucleic acid detection in point-of-care testing (POCT) diagnostics [5, 16, 21, 22, 35–37, 40]. These systems use programmable guide RNA (gRNA) to direct Cas proteins to specific DNA or RNA sequences, enabling highly sensitive detection through signal amplification. However, CRISPR-based methods face challenges in detecting subtle genetic alterations, such as single-nucleotide variations (SNVs). Besides, off-target effects of Cas proteins and the need for pre-amplification processes can introduce polymerase-derived errors, reducing the fidelity of CRISPR-based detection. A strategy that allows direct detection of the original sequence, minimizes off-target effects, and maintains high sensitivity is essential for precise ctDNA detection.

In this study, inspired by molecular logic gates [4, 10, 17, 20], we present a novel method for detecting single-nucleotide mutations by integrating the CRISPR/Cas9 system with HCR isothermal amplification through an AND logic gate. Without requiring pre-amplification, this approach enables the direct detection of ctDNA with high specificity and sensitivity. The CRISPR/Cas9 system is engineered to specifically bind target sequences, while hairpin probes recognize and bind to the single-stranded region generated by Cas9 nickase (nCas9) cleavage. This AND-gated strategy minimize the off-target effects of Cas9 and enhances detection specificity for single-nucleotide mutations. Once bound, the hairpin probe triggers HCR, amplifying the signal and enhancing detection sensitivity. This method achieves high sensitivity and single-nucleotide resolution for low-abundance ctDNA, with a limit of detection (LOD) as low as 1 fM, and demonstrates stability over 10 days with 97.5% accuracy in 40 samples. Furthermore, we successfully detected cancer-related mutations, including KRAS G12C, KRAS G12D, EGFR T790M and TP53 R273H, in physiological serum samples (50% serum). Our results demonstrate the promising clinical potential for precise ctDNA detection with high-specificity, high-sensitivity, and high-fidelity.

Experimental section

Preparation of DNA targets

The mutant sequences and corresponding wild-type sequences, along with their complementary sequences, were obtained from the NCBI database and commercially synthesized (Sangon, Shanghai). The mutant targets and wild-type targets were prepared by PCR amplification or by annealing two complementary oligonucleotides in 1× TE/Mg²⁺ buffer using a thermal cycler, with the temperature decreasing from 92 °C to 25 °C. Subsequently, the dsDNA targets were aliquoted and stored at −20 °C.

HCR experiment

DNA sequences were designed and analyzed using NUPACK (Nucleic Acids Package) software. DNA H1 monomer sequences and DNA H2 monomer sequences were designed as hairpin structures with a sticky end of 7 nt, a stem length of 15 nt, and a loop length of 7 nt. Trigger chain I was designed to open the hairpin structure to initiate the HCR. The lyophilized powder of hairpin DNA H1, hairpin DNA H2, and trigger chain I DNA sequences were dissolved in 1×PBS solution and diluted to prepare samples with a series of different gradients of hairpin DNA H1, hairpin DNA H2, and trigger chain I. The samples were then stored at −20 °C. Before conducting the HCR, the DNA H1 monomer and DNA H2 monomer were diluted to 1 µM and denatured at 95 °C for 5 min in a PCR machine, followed by gradually cooling down to room temperature at a rate of 1 °C per minute. This step was performed to ensure that DNA H1 and H2 monomers could form stable hairpin structures.

To verify whether the designed sequences could efficiently trigger HCR at low ratios of trigger chain I, series of concentration gradients of trigger chain I were prepared at 1 µM, 0.5 µM, 0.1 µM and 0.05 µM, respectively. The entire reaction system contained 10 µL of DNA H1 monomer (1 µM), 10 µL of DNA H2 monomer (1 µM), 10 µL of trigger chain I at different concentrations, and 20 µL of 1×PBS buffer, with a total volume of 50 µL. Five groups were set up based on the different ratios of ssDNA trigger chains, the ratios of trigger strands (I) to DNA monomers (H1, H2) were 1:1, 1:2, 1:10, 1:20, and a reference group without trigger strands, respectively.

Staining of HCR products with SYBR green I and characterization

After the HCR, 20 µL of HCR product was mixed evenly with 1 µL of SYBR Green I working solution and incubated in a constant temperature shaker at 37 °C for 30 min. Characterization was conducted using 1.5% agarose gel electrophoresis. The gel was then removed and placed in a gel imaging system, where bands were observed and photographed under both ultraviolet and blue light conditions.

The fluorescence emission spectra of the reaction products were measured using a F-7100 Fluorescence Spectrophotometer. 20 µL of each reaction product was mixed with 1 µL of SYBR Green I working solution and diluted to 1000 µL, followed by incubation at 37 °C for 30 min. Quartz cuvettes were used as measurement containers, with both excitation and emission slit widths set to 10 nm. The scan speed was set to 240 nm/min, PMT Voltage was set to 700 V, and the excitation wavelength was set to 493 nm. Fluorescence spectra were collected and analyzed in the wavelength range of 510 nm to 600 nm.

Target site selection and sequence design of SgRNA

In the CRISPR/Cas system, designing a sgRNA enables the Cas9 protein to bind to the target dsDNA under the guidance of sgRNA, thereby achieving specific cleavage of the target DNA. The sgRNA sequence is composed of two parts: the direct repeat sequence and the spacer sequence. When designing the sgRNA sequence, the first step is to identify the PAM sequence on the target sequence. The PAM sequence recognized by the Cas9 protein is NGG. Then, the target sequence was designed based on the downstream 20-nt sequence of the PAM sequence. The spacer sequence of the sgRNA needs to be complementary to this sequence to determine the spacer sequence of the sgRNA. The direct repeat sequence can form a stable binary complex with the Cas9 protein. nCas9 was used as the functional protein of Cas9. Primers containing the T7 promoter, target recognition site, and primer complementary region were designed. The template DNA was formed by the complementary interaction of the upper and lower primers, and the template DNA is transcribed into the desired sgRNA sequence.

Synthesis of template DNA

The obtained nucleic acid sequence powder was dissolved in nuclease-free water, diluted to a concentration of 10 µM as measured by the Nanodrop, then stored in a −20 °C refrigerator for backup. After the designed primers hybridize and form complexes, Taq DNA polymerase was used to extend the primers to form template DNA, and PCR amplification was then performed. The upstream primer (10 µM) and downstream primer (10 µM) were mixed in reaction buffer (10 mM Tris-HCl, 50 mM NaCl, 1.5 mM MgCl₂, and 0.4 mM dNTP, pH 7.9), and Taq DNA polymerase (0.2 U/µL) was added, with a total reaction volume of 50 µL. The PCR reaction was conducted on a PCR machine with the following program: 95 °C for 5 min, followed by 95 °C for 30 s, 56 °C for 30 s, and 72 °C for 30 s, repeated for 40 cycles, then 72 °C for 10 min, and finally stored at 4 °C. The synthesized template DNA (116 bp) was purified using a purification kit for centrifugation purification.

In vitro transcription of SgRNA

The synthesized template DNA from the previous step contains a T7 promoter, allowing for the synthesis of the desired sgRNA through transcription by T7 RNA polymerase. 500 ng of template DNA was mixed in reaction buffer (10 mM Tris-HCl, 50 mM NaCl, 1.5 mM MgCl₂, and 2 mM dNTP, pH 7.9), and T7 RNA polymerase (2.5 U/µL) was added, with a total reaction volume of 20 µL. The reaction mixture was then incubated at 37 °C for 2 h. Here, the reaction product still contained template DNA, which required digestion with RNase-free DNaseI. 1 µL of DNaseI and 3 µL of 10× reaction buffer was added, and nuclease-free water is added to bring the reaction volume to 30 µL. The mixture is then incubated at 37 °C in a PCR machine for 1 h. Incubate the mixture in the PCR machine at 37 °C for 1 h.

Assembling the CRISPR/Cas9 system

sgRNA and target DNA were synthesized through in vitro experiments. nCas9 (600 µM) and sgRNA were reacted in reaction buffer (10 mM Tris-HCl, 50 mM NaCl, 5 mM MgCl₂, pH 7.9) at 15 °C for 10 min to generate CRISPR RNP complexes with a final concentration of 100 nM. Without purification of RNPs, 50 nM of target DNA was added, and the mixture was incubated at 37 °C for 30 min, with a total volume of 20 µL, to assemble the CRISPR/Cas9 system.

AFM characterization of CRISPR/Cas9 complexes

Fix the brown mica sheet (20 × 50 mm) to a flat circular metal disk (15 mm) with double-sided tape and flatten the mica surface by peeling off the double-sided tape. Next, pipette 3 µL of diluted reaction product (1 nM) onto the center of the smooth mica sheet and let it stand for 5 min. Subsequently, rinse the mica surface with nuclease-free water and quickly dry it. The prepared samples were placed on the AFM sample stage, tapping mode was selected for imaging, and the images were processed and analyzed using NanoScope Analysis software.

Design and screening of molecular beacon probes

The hairpin recognition probe sequences were designed and analyzed by NUPACK software. The hairpin recognition probe with a stem length of 13-nt and a ring length of 17-nt was named “HP-s13 test”, where the 17-nt region of the ring matches the sgRNA recognition region. Test the designed probes using NUPACK software with the following parameters: nucleic acid type as DNA; temperature at 25 °C; nucleic acid concentration at 1 µM; Na⁺ concentration at 50 mM; Mg²⁺ concentration at 10 mM. Ten different probes were designed and named HP-s6, HP-s7, HP-s8, HP-s9, HP-s10, HP-s11, HP-s12, HP-s13, HP-s14, and HP-s15, as well as a linear probe Probeline as the reference. NUPACK 4.0 (https://nupack.org/) was used to calculate whether a spontaneous HCR occurs between each probe and H1 monomer and H2 monomer and comparing to reference. Delete the hairpin recognition probes that spontaneously trigger reactions.

CRISPR combined with HCR

Optimization of experimental conditions, including HCR monomer concentration, reaction temperature and time. The concentration of the HCR monomers (H1 and H2) affects the efficiency of amplification, and unsaturated amplification reactions can affect the accuracy of detection. Therefore, we set up 6 experimental groups with trigger chain concentrations set at 0.1 M, while the concentrations of hairpin probes ranged from low to high (1 nM, 5 nM, 10 nM, 50 nM, 100 nM, 500 nM). Fluorescence intensity increased with the concentration of monomers, but the signal level stabilized after the concentration exceeded 100 nm. Therefore, 100 nM was considered as the optimal concentration.

The synthesized 150 bp KRAS G12D sequence was used as the target DNA when exploring optimal temperature. Recognition experiments were set up at 25 ℃ and 37 ℃. When the KRAS G12D sequence was detected, a strong fluorescence signal was produced. Compared with 25 °C, higher fluorescence intensity was shown at a reaction temperature of 37 °C, proving that 37 °C was the optimal reaction temperature.

nCas9 (600 µM) and sgRNA were incubated in reaction buffer (10 mM Tris-HCl, 50 mM NaCl, 5 mM MgCl₂, pH 7.9) at 15 °C for 10 min to generate a final concentration of 100 nM of CRISPR RNP complex. Different concentrations of target DNA were then added and incubated at 37 °C for 30 min, with a total volume of 20 µL, to assemble the CRISPR/Cas9 system. Subsequently, hairpin recognition probes (100 nM) were added to hybridize with the exposed non-complementary strand and incubated at 37 °C for 60 min. HCR monomers H1 and H2 (100 nM) were then added and incubated at 37 °C with shaking for 120 min. The binding products were analyzed using 8% PAGE gel electrophoresis.

By adding the target DNA into 25% or 50% FBS, the obtained simulated samples were applied in the experimental detection procedure. It was the first to complete the recovery experiment of ctDNA in diluted serum. The detection procedure was conducted as described above, with the addition of 1 U/µl of RNase inhibitor.

Results

Principle of AND logic-gated CRISPR/Cas9 and HCR system for precise ctdna detection

As illustrated in Fig. 1, the detection system begins with the formation of a ribonucleoprotein (RNP) complex, comprising nCas9 protein with a single guide RNA (sgRNA). nCas9 is a variant of Cas9 nuclease, designed to create single-strand breaks (nicks) in DNA rather than double-strand breaks. The sgRNA contains a 20-nucleotide (nt) sequence that specifically target a region of the ctDNA. The programmability of the CRISPR/Cas9 system allows customization of the recognition sequence by modifying the sgRNA, making it highly versatile. When the RNP complex is introduced to a sample containing ctDNA, the sgRNA precisely recognizes and binds to the target sequence (protospacer) on the ctDNA. The successful RNP binding could trigger nCas9 to cleave one strand of the DNA, exposing a single-stranded DNA (ssDNA) region. A hairpin probe (HP), designed to complement this exposed ssDNA, is then introduced, forming the basis of the AND logic gate. Only when the ctDNA perfectly matches the designed sgRNA sequence does the complementary region of the probe HP able to bind to the exposed ssDNA, opening the hairpin structure and revealing the probe’s sticky end (red colored). This opening initiates the HCR by enabling the sticky end to interact with two other hairpin monomers (H1 and H2), leading to the formation of a long double-stranded DNA (dsDNA) through a cascade of hybridization. The amplified dsDNA product is then detected using SYBR Green I, a fluorescent dye that selectively binds to dsDNA, allowing for both qualitative and quantitative detection of ctDNA. However, if CRISPR/nCas9 RNPs bind to a sequence with a mismatch, or bind and cut untargeted sequence due to its off-target effect, the exposed ssDNA can not be recognized by probe HP, thus preventing further HCR for signal output. This ensures high specificity in the AND logic gate operation, effectively discriminating against single-nucleotide mismatched sequences.

Fig. 1.

Schematic illustration of the AND logic-gated CRISPR/nCas9 and HCR system for precise detecting of ctDNA with SNVs

Construction and optimization of AND logic-gated CRISPR/Cas9 and HCR system for DNA detection

To validate the detection system, a 230 bp segment of Lambda DNA was synthesized and used as the target sequence. The sgRNA, designed to recognize this specific DNA sequence (Table.S1), was transcribed in vitro (Fig.S1). nCas9 was assembled into the CRISPR/Cas9 complex, which facilitated specific cleavage of the target dsDNA, guided by the sgRNA (Fig. 2a). Electrophoresis confirmed the successful formation of the RNP complex. To further visualize RNP binding, atomic force microscopy (AFM) was employed. Two sgRNAs targeting different regions of the Lambda DNA sequence were tested, and AFM images (Fig. 2b) demonstrated the precise binding of the CRISPR/Cas9 complex to the target DNA, indicating the potential for multiplex detection of different mutations. Next, a 17-nt hairpin probe (HP), with sequences complementary to the sgRNA recognition region was designed (Fig.S2). The structures of hairpin probes with stem lengths ranging from 6 to 15 nucleotides (Table.S2) were predicted using NUPACK software (named HP-s*) (Fig.S2). The analysis indicated that probes with stems shorter than 13-nt were prone to spontaneous opening, leading to non-specific HCR. Therefore, HP-s13 was selected as the candidate for the subsequent validation experiments. The HP-s13 probe initiates the HCR by interacting with hairpin monomers H1 and H2. Agarose gel electrophoresis (Fig.S3) showed successful cascading chain reactions under various conditions, with optimal efficiency observed when the monomer concentration was ten times higher than the initial probe concentration. SYBR Green I was also successfully used to label the final HCR product, producing a fluorescence signal proportional to the target DNA concentration (Fig.S3).

Fig. 2.

Construction and characterization of the detection system. AFM images are performed by the labeled CRISPR/Cas9 system of single-site (a) and dual-site (b). (c) Methods for optimizing hair recognition probes to reduce background signal. (d) 1.5% agarose gel electrophoresis characterization of CRISPR/Cas9-triggered HCR process. Lane M: marker; Lane A: 230 bp target dsDNA; Lane B: Lane A + nCas9 RNPs; Lane C: Lane B + HP-s13; Lane D: Lane C + H1 and H2 monomers (500 nM); Lane E: nCas9 RNPs + HP-s13 + H1 and H2 monomers (500 nM). (e) 1.5% agarose gel electrophoresis to verify the effect of optimized probes

The AND logic-gated CRISPR/Cas9 and HCR system was successfully constructed to demonstrate its feasibility for ctDNA detection. Initially, we characterized the system using 1.5% agarose gel electrophoresis. As shown in Fig. 2d, the sgRNA guides the nCas9 RNP complex to the target site, where it cleaves the dsDNA to expose ssDNA. This process produced a gradually upward-migrating band (Lanes A-C). Upon introducing the H1 and H2 monomers, a band accumulation at the top of the gel (Lane D) indicated successful HCR amplification. However, a smear band was also observed in Lane E, where no target DNA was present, indicating some degree of spontaneous HCR initiation by the hairpin probe, which could lead to background signal interference and affect detection accuracy. Thus, further optimization of the hairpin probe design is essential to enhance specificity and ensure high-fidelity ctDNA detection.

The background signal was likely caused by unintended binding between the sgRNA and the hairpin probe HP, triggering non-specific HCR. To address this issue, we optimized the HP sequence to improve binding specificity. The loop length was maintained at 17-nt to match the length of the ssDNA produced by nCas9 cleavage. Compared to previous probes, the optimized HP (Table.S3) had an extra segment of stems that could bind to sgRNA (Fig. 2c). Optimized probes (optimized HP-1) and (optimized HP-2) were selected for further experiments. These probes were mixed with H1 and H2 monomers at a ratio of 1:10:10 to assess their specificity in triggering HCR. Results from the gel electrophoresis (Fig. 2e and Fig.S4) demonstrated that HCR occurred only in the presence of the target DNA, while the hairpin probes alone did not initiate HCR. Compared to the original probe, the signal of optimized HP-1 and HP-2 was amplified 1.3-fold and 1.4-fold, respectively, while the original group produced only a minimal S/N ratio of 1.05 (Fig.S5). This confirmed the high specificity and minimized background interference of the optimized system, making it suitable for precise ctDNA detection. Therefore, optimized HP-2 was selected as the HP for next experiments.

Precise detection of single-nucleotide variations in cancer-related ctDNA

After optimizing the AND logic-gated CRISPR/Cas9 and HCR system, we selected the wild-type KRAS gene fragment as a model target for SNVs detection. KRAS is a key oncogene frequently mutated in human cancers, particularly colorectal cancer (CRC), and detecting SNVs such as KRAS G12D and KRAS G12C is critical for diagnosis and personalized medicine. As shown in Fig. 3a, a 60 bp dsDNA fragment of KRAS was synthesized, and specific mutations were inserted at positions 19, 15, 7, and 1 within the target region. These modified DNAs were designated as Mis-19-DNA, Mis-15-DNA, Mis-7-DNA, and Mis-1-DNA, respectively. The relevant sequences have been provided in Table.S4. Notably, the mutation at position 7 corresponds to the KRAS G12D variant. Since nCas9 cleaves the target DNA and exposes a 17-nt ssDNA segment, the variation site of Mis-1-DNA is located outside the exposed ssDNA region, allowing us to evaluate whether our system can detect mutations at any location within the 20-nt sgRNA recognition sequence (protospacer). Successful synthesis of the 60 bp dsDNA fragments, both wild-type and those with mutations, was confirmed by 8% native PAGE gel electrophoresis (Fig.S6). Optimal conditions for HCR, including a probe HP concentration of 10 nM, monomer concentrations of 100 nM, and a reaction temperature of 37 °C, were established (Fig.S7). Besides, in a series of experiments, the optimal subsequent incubation time was determined to be 2 h (Fig.S8).

Fig. 3.

Validation of the system for detecting SNVs. (a) Schematic representation of synthesizing DNA with single nucleotide mutations at different positions (Mis-19-DNA, Mis-15-DNA, Mis-7-DNA, and Mis-1-DNA). (b) 8% PAGE gel electrophoresis results for detecting single nucleotide differences using only the CRISPR/Cas9 system. Lane M: marker; Lane A: fully matched DNA (Full-DNA); Lane B: Mis-19-DNA; Lane C: Mis-15-DNA; Lane D: Mis-7-DNA; Lane E: Mis-1-DNA. (c) Detection of single nucleotide differences using the AND logic-gated CRISPR/Cas9 and HCR system and characterization of mutations with 1.5% agarose gel electrophoresis. Lane M: marker; Lane A: blank control; Lane B and C: Full-DNA; Lane D: Mis-19-DNA; Lane E: Mis-15-DNA; Lane F: Mis-7-DNA; Lane G: Mis-1-DNA. (d) Fluorescence intensity detection for the different single-nucleotide variant sequences, with stronger fluorescence observed for fully matched DNA compared to the mutation-containing samples. (e) Statistical analysis of fluorescence values measured at the maximum emission wavelength of 525 nm. The data represent the mean ± standard deviation from three independent replicates, with statistical significance determined using a two-tailed t-test: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

To assess the system’s ability to precisely differentiate mutations, we compared it to a control system that employed CRISPR/Cas9 without HCR initiation by the probe HP (Table.S5). As shown in Fig. 3b, electrophoresis results for fully matched DNA (Lane A) were indistinguishable from those for DNAs with mutations (Lanes D and E), demonstrating that CRISPR/Cas9 alone cannot differentiate between matched DNA and mutation samples. In contrast, when the HCR was integrated into the CRISPR/Cas9 system, forming an AND logic-gated operation (Fig. 3c), clear differentiation was observed between the target DNA (Lanes B and C) and DNAs with mutations (Lanes D-G). For the target DNA, a distinct band stacking at the well of the gel indicated successful HCR amplification. However, for Mis-19-DNA, Mis-15-DNA, and Mis-7-DNA (Lanes D, E, and F, respectively), only faint bands were detected, showing significant differentiation from the target DNA. The Mis-1-DNA (Lane G) produced a band similar to the target DNA, which we attribute to the location of the mutation. Since the variation is located outside the exposed single-stranded recognition region, the chain reaction proceeds as it would with fully matched DNA. This suggests that our system can precisely differentiate single-nucleotide mutations within the exposed region recognized by the hairpin probes.

To further confirm the differentiation ability, we stained the HCR product with SYBR Green I and collected fluorescence spectra within the range of 510 to 600 nm. As shown in Fig. 3d, the target DNA produced strong fluorescence, while mutation-containing DNAs generated significantly lower signals (Mis-19-DNA, Mis-15-DNA, and Mis-7-DNA). Statistical analysis (Fig. 3e) revealed a clear distinction in fluorescence intensity between the target DNA and the mutation samples, confirming that the system has a high level of discrimination for mutation the exposed region recognized by the hairpin probe. However, the system’s ability to detect mutations located outside this region is reduced, as demonstrated by the Mis-1-DNA result, which is consistent with the gel electrophoresis results. These findings indicate that the AND logic-gated integration of CRISPR/Cas9 and HCR, guided by hairpin probes, provides high accuracy and specificity in distinguishing single-nucleotide mutations, making it a promising approach for precise cancer-related ctDNA detection.

Detection performance evaluation of the AND logic-gated CRISPR/Cas9 and HCR system

To evaluate the detection performance of the AND logic-gated CRISPR/Cas9 and HCR system, 150 bp fragments of wild-type KRAS and KRAS G12D, matching the typical length of ctDNA found in clinical samples, were synthesized (Table.S6 and S7). A specific sgRNA targeting a 20-nt sequence of the KRAS G12D variant was designed and transcribed in vitro, along with a corresponding hairpin probe that could be selectively opened by CRISPR/Cas9-mediated cleavage. The sensitivity of the system was assessed by diluting KRAS G12D to concentrations ranging from 10 nM to 50 fM. As shown in Fig. 4a, the fluorescence intensity at 525 nm increased with rising KRAS G12D concentrations, demonstrating a strong linear relationship (Fig. 4b). The limit of detection (LOD) for KRAS G12D was approximately 1 fM (calculated as three times the standard deviation of blank control signals), indicating the system’s high sensitivity for target DNA detection.

Fig. 4.

Detection performance of the AND logic-gated CRISPR/Cas9 and HCR system in KRAS samples. (a) Fluorescence response plots of different target DNA concentrations (KRAS G12D). (b) Logarithmic dependence of fluorescence intensity (at 525 nm) for different target DNA concentration. Error bars indicate the standard deviation from three measurements. (c) Fluorescence intensity at 525 nm for the KRAS G12D mutant sequence at different ratios mixed with wild-type sequences (n = 3 replicates). Statistical significance was determined using a two-tailed t-test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars represent the standard deviation of three measurements. (d) Stability assessment of the detection system for discriminating single-nucleotide mutations. Error bars indicate the standard deviation from three measurements. (e) Accuracy evaluation of the detection system

Given the extremely low concentrations of ctDNA typically found in biological fluids, it is crucial for the detection system to accurately identify target ctDNA even in the presence of excess wild-type sequences. To evaluate the system’s sensitivity for detecting SNVs, KRAS G12D was mixed with wild-type KRAS DNA (100 nM) at various ratios (0%, 0.001%, 0.01%, 0.05%, 0.1%, 0.5%, 1%, and 10%). The results (Fig. 4c) demonstrated that the system was capable of detecting KRAS G12D at an allele frequency as low as 0.1%, based on the fluorescence intensity at 525 nm. This confirms the system’s ability to sensitively detect low-abundance ctDNA and differentiate between mutations.

To assess stability, KRAS G12D was mixed with wild-type KRAS (100 nM) at allele fractions of 10%, 1%, and 0.1%, and the samples were tested over 10 days at two-day intervals, with three replicates per test. The coefficient of variation (CV) for the 10%, 1%, and 0.1% allele fractions were 3.6%, 4.8%, and 2.6%, respectively, indicating high stability in ctDNA detection (Fig. 4d). The accuracy of the system was further evaluated through 40 blinded experimental tests of KRAS G12D DNA concentrations ranging from 0.1 nM to 10 nM. The detection protocol achieved an accuracy of 97.5% based on positive test results (Fig. 4e).

In addition to KRAS, the system’s capability was validated for other clinically relevant mutations. The EGFR T790M mutation, commonly associated with acquired resistance in lung cancer, and the TP53 R273H mutation, which promotes cancer cell survival, were selected for testing. The relevant sequences have been provided in Table.S6 and S7. The LOD for the EGFR T790M variant was approximately 1.5 fM, while the LOD for TP53 R273H was 1.3 fM (Fig. 5a and d). Accuracy evaluations, performed through 20 blinded experimental tests for each mutation, confirmed the system’s high repeatability and specificity (Fig. 5e). Overall, these results demonstrate that the AND logic-gated CRISPR/Cas9 and HCR system is highly sensitive, stable, and accurate for detecting ctDNA at low concentrations and can reliably differentiate single-nucleotide mutations at low allelic fractions, making it suitable for precise ctDNA detection in clinical applications.

Fig. 5.

Detection performance of the AND logic-gated CRISPR/Cas9 and HCR system in EGFR and TP53 samples. Fluorescence response plots and logarithmic dependence of fluorescence intensity (at 525 nm) for different target DNA concentrations in EGFR T790M (a and b) and TP53 R273H (c and d) samples. Error bars represent the standard deviation of three measurements. (e) Accuracy test of the detection system

Detection performance in serum samples

To evaluate the potential clinical application of our AND logic-gated CRISPR/Cas9 and HCR system for detecting ctDNA, we conducted simulated clinical assays using serum samples. Initial experiments demonstrated effective ctDNA detection in diluted serum samples (Table.S8), revealing minimal interference from serum concentrations of 0%, 25%, and 50% (Fig.S9). This suggests that our approach can detect target ctDNA without the need for prior DNA isolation. Next, we aimed to simulate a clinical environment by detecting KRAS G12D mutant sequences at various allele fractions in 50% serum, which closely resembles physiological conditions. Results indicated a gradual increase in fluorescence intensity with higher allelic fractions of the KRAS G12D mutant fragments. A significant difference was observed between the 0.1% and 0% groups, confirmed by a two-tailed t-test. This finding demonstrates that our strategy can successfully identify target ctDNA sequences at low concentration ratios in a serum environment, highlighting its potential for clinical application.

To further assess specificity, we tested various single-nucleotide mutation samples in 50% serum, including the KRAS wild-type sequence, KRAS G12D sequence, KRAS G12C sequence, and a blank control. Notably, there is only a single-nucleotide difference between the KRAS wild-type and KRAS G12D sequences, while the KRAS G12C sequence differs by two bases. Using the corresponding sgRNA and hairpin probe for KRAS G12D detection, we observed significantly lower fluorescence signals for the wild-type and G12C sequences compared to the target KRAS G12D (Fig. 6b). The KRAS G12C exhibited very weak fluorescence, similar to the blank control, likely due to strict base pairing requirements. Conversely, when using sgRNA and hairpin probe specific to KRAS G12C, the target G12C showed strong fluorescence, while KRAS G12D exhibited a weak signal (Fig. 6c). These results align closely with sequencing outcomes (Fig. 6a), confirming the assay’s high accuracy. Overall, our AND logic-gated CRISPR/Cas9 and HCR system allows for precise single-nucleotide genetic analysis, effectively distinguishing target ctDNA from other DNA sequences with high reproducibility and stability.

Fig. 6.

Detection of KRAS, EGFR and TP53 mutation genes in simulated clinical samples. Sequencing results for KRAS G12D/G12C (a), EGFR T790M (d), and TP53 (f). Fluorescence results in a physiological serum environment for single-nucleotide mutation samples: KRAS G12D (b), KRAS G12C (c), EGFR T790M (e), and TP53 (g)

Additionally, we investigated other hotspot mutations, including EGFR T790M and TP53 R273H, in 50% serum using specifically designed hairpin probes. The mutant sequences generated strong fluorescent signals compared to both the blank control and wild-type samples, demonstrating the capacity of our strategy to accurately detect EGFR T790M and TP53 R273H mutations in simulated clinical samples. 150 bp of TP53 R273H mutant DNA was mixed with the corresponding TP53 wild-type sequence (100 nM) at a 10% allelic fraction. The resulting DNA samples were diluted to create concentrations of 0.1 nM, 0.5 nM, 1 nM, 10 nM, 20 nM, 50 nM, 70 nM, and 100 nM, alongside two blank controls (one with 0 nM DNA and the other with 100 nM EGFR T790M mutant DNA). All samples were assigned random numbers, sequenced, and compared with our experiments. The same experimental conditions were used for EGFR T790M detection. Results depicted in Fig. 6d-g show varying fluorescence intensities across different groups, with two groups (the blank controls) exhibiting weak signals. Sequencing confirmed the identification of two negative and eight positive samples, indicating a favorable correlation between our experiment and sequencing results. This suggests that our method is highly accurate and possesses practical utility for ctDNA detection in clinical samples.

Conclusions

In this study, we developed a detection method based on the AND logic-gated integration of CRISPR/Cas9 system and HCR isothermal amplification for the highly specific and sensitive detection of ctDNA. Unlike conventional CRISPR-based sensors, which often rely on single-recognition events that can potentially lead to off-target effects, our system employs a dual-recognition mechanism. Signal amplification is triggered only when both specific Cas9 RNP binding and hairpin probe hybridization occur, thereby significantly minimizing false-positive signals. Our findings demonstrate that this system effectively distinguishes target mutations with high accuracy, even in complex serum environments, showcasing its potential for clinical applications in oncology.

Compared to existing methods, our detection system offers several advantages, including specificity, sensitivity, and reproducibility (Table S9). In terms of specificity, the AND logic gate enables superior discrimination of single-nucleotide variants (SNVs) compared to standard HCR-coupled assays [24] or non-gated CRISPR systems [13], which have false-positive amplifications due to off-target effects and unstable hairpin structures. The capability to discern single-nucleotide mutations with precision makes it a valuable tool for personalized medicine, enabling tailored treatment strategies based on individual genetic profiles. Regarding sensitivity, the synergistic combination of CRISPR recognition and HCR amplification achieves a lower limit of detection (LOD) of ~ 1 fM, enabling the identification of trace ctDNA even amid a vast excess of wild-type sequences. Furthermore, the workflow benefits from the non-enzymatic, isothermal nature of HCR, avoiding the complex thermal cycling required by PCR-based methods and the synthetic errors introduced by polymerase. Compared to traditional NGS and qPCR techniques, our process is more cost-effective. Notably, the system can detect low-abundance ctDNA amidst wild-type sequences, addressing a significant challenge in cancer diagnostics.

Despite these strengths, certain limitations persist, such as the reliance of the Cas9 protein on protospacer adjacent motif (PAM) sequences, which may restrict its application to specific genomic targets. However, recent advancements in CRISPR research, including the development of alternative Cas variants with broader PAM requirements [19, 32], may enhance the versatility of CRISPR/Cas9-based detection approaches. In conclusion, our AND logic-gated CRISPR/Cas9 and HCR system represents a significant step forward in ctDNA detection, combining high specificity and sensitivity with potential applications in clinical diagnostics. Continued research and development in this field will be essential to overcoming existing limitations and realizing the full potential of molecular diagnostics in oncology.

Author contributions

Tianliang Ji: Methodology, Resources, Validation, Visualization, Writing-original draft. Yujia Zhang, Yixiu Wang and Kaiyu Yuan: Methodology, Validation, Investigation, Data curation, Writing-original draft. Mingxiang Wang and Jingyi Ye: Resources, Validation, Data curation. Huan Zhang and Honglu Zhang: Conceptualization, Supervision, Funding acquisition, Writing-review & editing. Ning Zhang: Funding acquisition, Writing-review & editing.

Funding

This work was supported by the National Key R&D Program of China (2022YFB3808200), the National Natural Science Foundation of China (32571596, 82503795 and 22377076), the Agricultural Science and Technology Innovation Program (CAAS-CSCB-202402).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare no competing interests.