Introduction

Respiratory viral infections pose significant multisystem threats by employing diverse pathogenic mechanisms, including direct cytopathic effects, immune dysregulation, and metabolic disturbances. These infections induce acute respiratory damage while potentially causing systemic complications and long-term sequelae via immune-metabolic dysfunction and multiorgan dissemination. Notably, influenza virus (IFV)1, human parainfluenza virus (HPIV)2,3, and SARS-CoV-24 have particularly high infection rates and clinical risk profiles5, warranting special consideration in respiratory disease management.

Notably, children under 5 years of age, as well as those aged 6–10, are particularly susceptible not only to single-pathogen infections but also to co-infections. This reflects a concerning trend of cross-transmission and a high prevalence of mixed infections within these age groups. For example, a Guangzhou-based study conducted between 2017 and 2022 investigated the burden of pediatric respiratory infections and the epidemiological characteristics of common respiratory pathogens identifying 4,973 cases of mixed infections. Among these, 1,033 cases involved co-infections of IAV, IBV, and Mycoplasma pneumoniae (MP), whereas 1,022 cases featured concurrent IAV and IBV infections6,7. The persistently high infection rates of these viruses continue to drive significant demand for in-hospital diagnostic testing.

Although real-time quantitative polymerase chain reaction (qPCR) is widely regarded as the gold standard for viral detection due to its exceptional sensitivity, specificity, and standardized protocols8, it presents several limitations that hinder its broad implementation in primary healthcare settings. Specifically, qPCR is highly dependent on specialized instrumentation, requires considerable processing time (typically 2–4 h), and involves complex operational procedures9. The rapid diagnosis of pediatric respiratory infections demands portable, user-friendly, and highly efficient detection technologies. While recombinase polymerase amplification (RPA) eliminates the need for thermal cycling and enables rapid amplification (< 30 min), it still relies on complex fluorescence detection systems10. Additionally, multiplex RPA amplification alone results in significant non-specific background signals11.

Currently, the integration of CRISPR technology with various detection platforms—such as surface plasmon resonance (SPR), fiber-optic sensing, electrochemical sensing, and graphene field-effect transistors (GFET)—has significantly enhanced detection performance in terms of sensitivity, specificity, speed, and portability. These advancements have introduced groundbreaking solutions for the diagnosis of genetic diseases, infectious disease surveillance, and mutation analysis12,13,14,15. Among these, CRISPR-Cas12a technology has emerged as a pivotal strategy in the development of innovative biomolecular detection methods, drawing significant attention due to its high efficiency and specificity. This system provides essential technical support for the advancement of next-generation diagnostic tools16,17,18. Among its variants, CRISPR/Cas12a—a type V-A RNA-guided DNA endonuclease—exhibits unique molecular recognition capabilities. Guided by a crRNA, Cas12a specifically binds to complementary double-stranded DNA (dsDNA) targets, subsequently initiating its trans-cleavage activity, which results in the non-specific degradation of single-stranded DNA (ssDNA) reporter molecules19. Currently, researchers have successfully harnessed the trans-cleavage activity of CRISPR-Cas12a in nucleic acid detection systems20,21,22. This integration facilitates visual signal output by enabling the specific cleavage of fluorophore-labeled reporter molecules. However, despite these advancements, the sensitivity of CRISPR-based detection remains substantially inferior to that of both PCR and isothermal amplification techniques21,23, presenting a significant challenge for its clinical application.

In recent years, considerable research has focused on the development of integrated biosensor platforms that couple the signal amplification capabilities of CRISPR/Cas12a with lateral flow assay (LFA) technology24,25,26,27,28,29,30. This innovative integration leverages the molecular recognition precision of CRISPR/Cas12a while harnessing the visual readout advantages of LFA, thereby eliminating the reliance on specialized instrumentation typically by conventional fluorescence-based CRISPR detection systems. The key advantages of this approach include(1) preservation of the high target specificity of CRISPR, (2) rapid visual interpretation without the need for optical devices, and (3) enhanced suitability for point-of-care applications. This technological convergence represents a significant advancement in molecular diagnostics, particularly for use in resource-limited settings.

Recombinase polymerase amplification (RPA), the CRISPR/Cas system, and lateral flow assay (LFA) each present distinct advantages and limitations in nucleic acid detection, with their strengths being highly complementary. The integration of RT-RPA with CRISPR/Cas and LFA technologies enables rapid and visual detection of individual viral targets. However, during seasonal outbreaks involving pathogens such as influenza virus (FLV), SARS-CoV-2, or human parainfluenza virus (HPIV), the need for multiple reagent additions and sequential testing complicates the diagnostic workflow. This highlights the urgent need for a multiplexed, rapid detection system capable of simultaneously differentiating multiple pathogens within a single assay.

To address this challenge, this study introduces a CRISPR-based multiplex nucleic acid detection system integrated with real-time visual biochip technology. By combining CRISPR-Cas12a, isothermal amplification, and a microfluidic chip platform, we developed a passively driven chip capable of highly sensitive, rapid, and visually interpretable detection of multiple nucleic acid targets. Experimental validation demonstrated that this system can detect multiple pathogen nucleic acids within 35 min(excluding sample processing time.), with a detection limit as low as 10 copies, and enables visual readout via a colloidal gold-based LFA (Fig. 1).

Our RT-RPA/CRISPR-Cas12a combination offers distinct practical advantages over the prevalent LAMP-based platforms because it operates at a lower temperature (37–42 °C), enables faster results, and requires simpler primer design. This technology offers several key advantages, including operational simplicity, low cost, and high portability, making it well-suited for point-of-care testing, home-based self-diagnosis, and large-scale field screening. This passively driven chip presents a novel and practical solution for multiplexed, rapid molecular diagnostics, with substantial potential for clinical and public health applications.

Fig. 1
figure 1

Schematic diagram of the passively driven chip: (A, B, C) depict the fluid flow pathways and associated reactions at each site.

Full size image

Materials and methods

Protein expression and purification

Recombinant plasmids encoding target genes for proteases, single-stranded DNA-binding proteins, promoter factor A, promoter factor B, DNA polymerase, endonuclease IV, and exonuclease III were transformed into the E. coli expression strain BL21(DE3). Monoclonal colonies displaying uniform growth were selected and cultured in liquid medium under shaking conditions (37 °C, 220 rpm) until the optical density at 600 nm (OD600) reached 1.0. Protein expression was induced by the addition of isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM, followed by incubation at 20 °C and 220 rpm for 16 h. After induction, bacterial cells were harvested via centrifugation and lysed using ultrasonication in lysis buffer (50 mM NaH2PO4, 150 mM NaCl, 10 mg/mL lysozyme, and 10 mg/mL protease inhibitor, pH 7.2). The lysate was centrifuged at 12,000×g for 20 min to separate soluble proteins from insoluble inclusion bodies. Both the supernatant and pellet fractions were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to identify the strain exhibiting the highest target protein expression. The selected high-expression strain was subsequently subjected to a multistep purification process, including affinity chromatography, ion-exchange chromatography, and size-exclusion chromatography, to obtain the purified target protein. The Cas12a protein was subsequently purified using an analogous protocol (for further details, refer to the Supplementary Information and Figure S1).

Design of primers and CrRNA

In accordance with the RPA primer design guidelines of TwistAmp, the conserved regions of the influenza virus A, influenza virus B, novel coronavirus, and human parainfluenza virus genes were downloaded from the NCBI database.Following the design principles of TwistDx Ltd. (which specify, for example, a primer length of 30–35 nt and avoidance of long homopolymers and complex secondary structures), the corresponding RPA primers were designed manually via Primer Premier 5 software. BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) and IDT Oligo Analyzer (https://sg.idtdna.com/pages) were used to verify whether the primers complied with the various design principles, and highly efficient and specific primer pairs were selected. The designed primer pairs need to be able to amplify the target sequence. The target amplicon lengths were designed to be within the range of 100–200 bp. To enable virus detection using CRISPR-Cas12a, we designed specific crRNAs targeting conserved genomic regions of each virus. The crRNA directs the Cas12a protein to cleave a specific target DNA sequence, a process that requires the presence of a protospacer adjacent motif (PAM) with the sequence 5´-TTTN-3´31. However, for the purpose of this study, all target sites were designed with a stringent 5’-TTTT-3’ PAM to simplify the experimental design and analysis. In this design, the core target sequence within the amplicon recognized by the crRNA is typically 20–25 bp in length. However, to optimally target highly conserved regions, one crRNA (for influenza A virus) was designed with an 18 bp spacer, which demonstrated excellent performance in our validation assays. The specific target sequences used in this study are as follows: the matrix (M) gene (5’-GGA AAT GTT TGC AGC AGT–3’, 18 bp) for influenza A virus, the nucleoprotein (NP) gene (5’-GGA GGA GGA TTA GAT GAA ACG A-3’, 22 bp) for influenza B virus, the nucleocapsid (N) gene (5’-TTA CAA GGA CCA GGA GGA TGT-3’, 21 bp) for SARS-CoV-2, and the hemagglutinin (H) gene (5’-GTG ATG AAG TCA CTT CTT TGA-3’, 21 bp) for human parainfluenza virus (HPIV). All crRNAs were designed based on a 5’-TTTT-3’ PAM located upstream of their respective target sequences. The sequences of the primers used for these four respiratory viruses along with the corresponding crRNAs are shown in the supporting information (Table S1). The designed RPA primers and crRNA were synthesized by BiOligo Biotechnology Co., Ltd. (Shanghai) and diluted with enzyme-free water. They were then stored at − 20 °C until use.

RT-RPA-CRISPR/Cas12 reaction system and test strips

In the conventional two-step method, the probe typically consists of a 12-nt random sequence, which is prone to degradation by DNA polymerase within the RPA system. In this study, a shortened and optimized 5-nt probe (5’-TTATT-3’) was employed, offering enhanced stability and increased susceptibility to Cas12a-mediated cleavage. Additionally, the system configuration that exhibited minimal interference with Cas12a activity was selected. The AMP Future buffer was identified as optimal for fluorescence detection, as it effectively reduces the initial activity of Cas12a, thereby ensuring the efficient progression of both amplification and detection reactions within a single-tube format. The total volume of this reaction system was 30 µL. The specific reaction system is illustrated in the Supporting Information (Table S2). The reaction mixture was added to the passively driven chip reaction chamber for freeze-drying. When used, 60 µL of the mixture containing the test sample was added for reconstitution.

To evaluate the trans-cleavage activity of Cas12a on the ssDNA reporter, the ssDNA was dual-labeled at the 5’ and 3’ ends with FAM and biotin, respectively (5’-/6-FAM/TTATT/biotin/-3’). For visual detection via a lateral flow assay, colloidal gold conjugated with anti-FAM antibody was applied to the binding pad, while streptavidin was immobilized on the T line to capture intact probes bearing the biotin label. The C line was coated with an anti-species secondary antibody. In the absence of the target (negative sample), Cas12a remains inactive, and the ssDNA reporter is not cleaved, allowing the intact probe to bind at the T line, which results in a visible signal. Conversely, in the presence of the target (positive sample), the RNA sensor activates Cas12a, resulting in the cleavage of the ssDNA reporter. As a result, the fragmented probe cannot be captured at the T line, and no color development is observed at that position.

Freeze-drying of the reagent reaction system

Owing to the limited stability of liquid reagents, they typically require refrigerated or frozen storage and transportation. Moreover, the manual addition of these reagents not only complicates the operational workflow but also imposes stringent requirements on storage and handling conditions, thereby limiting their applicability for rapid on-site deployment. To simplify the procedure and enable storage and transport at ambient temperature, both isothermal amplification reagents and CRISPR/Cas components were preloaded into the reaction chambers of the passively driven chip and lyophilized under optimized conditions. The test results for the protein in its freeze-dried state and its stability are presented in the supporting information (Figure S2). Prefreezing Stage: After the reagent systems were supplemented with 5% trehalose or 4% mannitol, 30 µL of the solution was dispensed into each of the four reaction chambers of the passively driven chip. The chip was then transferred to an–80 °C ultralow temperature freezer and maintained for 30 min to ensure complete solidification of the reaction mixture and the formation of a stable ice crystal structure. Concurrently, the freeze-dryer was pre-cooled to − 25 °C and equilibrated for at least 1 h. Freeze-drying protocol: Following cryofixation, the chip was immediately transferred to a pre-cooled freeze-dryer under vacuum. Primary Drying: the initial drying phase involved sublimation under controlled conditions. The system temperature was gradually increased to − 10 °C and maintained for 18 h, allowing complete sublimation of the ice crystals under vacuum. Precise temperature and pressure control were maintained throughout the mixture to preserve protein integrity. For secondary drying, the desorption phase was conducted at 5 °C for 5 h to remove residual bound water molecules, thereby increasing product stability. Post-Lyophilization Processing: The freeze-dried chip was promptly transferred to a nitrogen-purged glove box for encapsulation to prevent moisture absorption and maintain detection sensitivity.

Preliminary assessment of analytical sensitivity and specificity tests

To perform a preliminary qualitative assessment of the assay’s analytical sensitivity, serial dilutions of quantified viral nucleic acids (influenza A, influenza B, HPIV, and SARS-CoV-2) were prepared. The dynamic range of detection was probed using 10-fold dilution steps, which ranged from spanning from 100 to 103 copies/µL. These samples were tested in parallel using both the fluorescent probe-based method (reference) and the gold nanoparticle-based probe method (test). The results were compared to determine the relative performance of the two detection modalities at each concentration.

In this detection system, viral typing is accomplished through a Cas12a-mediated cleavage reaction initiated by the recognition of specific target gene sequences by the CRISPR/Cas12a system. To assess the specificity of each reaction system for its corresponding target, Cas12a-based assays were conducted on the passively driven chip. The crRNAs and associated reagents specific to each virus were lyophilized within the respective reaction chambers. Additionally, an off-chip simulated detection was performed using a fluorescent probe in place of the colloidal gold probe. To evaluate the interaction between each RNA template and its corresponding crRNA on the chip, the research team introduced positive control samples for SARS-CoV-2, human parainfluenza virus, influenza A virus, and influenza B virus. Off-chip fluorescence detection was subsequently carried out. Furthermore, equimolar mixtures of two or four viral positive viral controls were tested to examine potential cross-reactivity. The analytical specificity of the detection system was evaluated based on results obtained from both lateral flow test strips and fluorescence-based detection.

Clinical sample detection

To evaluate the clinical performance of the passively driven chip, we conducted a validation study using 80 nasopharyngeal swab samples collected from pediatric patients at Shenzhen Children’s Hospital between March and April 2025. As shown in Fig. 1A, patient samples were collected by swabbing the bilateral palatine tonsils and posterior pharyngeal wall using sterile synthetic fiber swabs to obtain epithelial cells. The swab was then immediately immersed in a centrifuge tube containing 3 mL of nucleic acid lysis buffer (SPL4801, EZassay Ltd.), which was subsequently vortexed for 30 s and incubated at room temperature for 10 min. This lysis and release process enables the direct use of the crude lysate as a template for amplification without any purification. For our detection system, a measured volume of this prepared sample is directly loaded into the sample port of the microfluidic chip for subsequent on-chip nucleic acid amplification. The sample cohort comprised 20 influenza A virus-positive, 15 influenza B virus-positive, 18 human parainfluenza virus (HPIV)-positive, 11 SARS-CoV-2-positive, and 16 negative control samples, as previously confirmed by Sanger sequencing (gold standard method). All experiments involving human subjects and/or human tissue samples were performed in accordance with the Declaration of Helsinki and relevant national and institutional guidelines and regulations. The experimental protocols were approved by the Medical Ethics Committee of Shenzhen Children’s Hospital. Informed consent was obtained from all individual participants included in the study. For participants under the age of 18 or those with impaired decision-making capacity, written informed consent was obtained from their legal guardians.All methods were carried out in accordance with the relevant guidelines and regulations.

Results and discussion

Analysis of protein expression levels

The verified plasmids encoding proteinase, single-stranded DNA-binding protein, promoter factors A and B, DNA polymerase, endonuclease IV, and exonuclease III were transformed into the E. coli BL21(DE3) strain. The expression of all target proteins was subsequently induced.

Cellular lysates were subjected to ultrasonication and subsequently separated into soluble and insoluble fractions, followed by SDS-PAGE analysis. As shown in Fig. 2A(The gel image in Fig. 2A has been cropped to optimize layout and focus, and the corresponding full, uncropped image can be found in Figure S3), strains with high expression levels of the target proteins were selected. Electrophoretic analysis confirmed the successful expression of all the recombinant proteins, including proteinase, single-stranded DNA-binding protein, promoter factors A and B, the large fragment of DNA polymerase, DNA endonuclease IV, and exonuclease III. Notably, a portion of each recombinant protein was detected in the soluble fraction. To increase functional efficiency and stability, the protein sequence was optimized, as illustrated in Fig. 2B. Both the original-sequence recombinase (a) and the sequence-optimized recombinase (b) were stored at 37 °C for 0, 7, 14, and 21 days, followed by storage at 4 °C. The optimized recombinase retained 80% of its enzymatic activity even after 21 days at 37 °C.

Fig. 2
figure 2

SDS-PAGE of core proteins and functional comparison of homologous recombinase before and after sequence optimization: (A) M-marker, L1-recombinase protein, L2-single-strand binding protein, L3-promoting factor A, L4-promoting factor B, L5-DNA polymerase, L6-endonuclease lV, and L7-exonuclease lll. (B) Group a: Recombinant protein with the original homologous recombinase sequence stored at 37 °C for 0, 7, 14, and 21 days (labeled a-0, a-7, a-14, a-21). Group b: Recombinant protein with an optimized homologous recombinase sequence stored at 37 °C for 0, 7, 14, or 21 days (labeled b-0, b-7, b-14, or b-21). All samples were subsequently stored at 4 °C and subjected to comparative recombinase isothermal amplification assays.

Full size image

Results of recombinase purification

Following purification via affinity chromatography, ion exchange chromatography, and size-exclusion chromatography, the purity of the target proteins exceeded 95%, indicating a highly successful purification process capable of supplying high-quality recombinant proteins for subsequent RPA reactions. The production yields of all recombinant proteins are summarized in Table 1. The reaction times were all greater than 10,000 times, meeting the experimental requirements. Among them, recombinant proteinase can support more than 5.8 million reactions; single-linked binding protein can support more than 64 million reactions; promoting factor B can support more than 108 million reactions; DNA polymerase can support more than 3.66 million reactions; endonuclease IV can support more than 49 million reactions; and exonuclease III can support more than 34 million reactions. The production yields of these recombinant proteinases substantially exceed the threshold required for 10,000 reactions, fully meeting experimental demands. Notably, the yield from a single fermentation batch is sufficient to support many RPA reactions, thereby satisfying both research-scale and industrial-scale application requirements.

Table 1 Purification results and consumption levels.
Full size table

Optimization of the reagent reaction system

In this study, the detection principle of the test strips is the signal depletion format. A signal-depletion format was employed, wherein the T-line disappears when the target analyte concentration exceeds a defined threshold. The quantity of ssDNA probe used in the system is a critical parameter. Excessive probe concentration may hinder proper visualization of negative results, whereas insufficient probe amounts may lead to incomplete cleavage, resulting in unintended signal retention on the T-line. To balance these factors and ensure optimal detection sensitivity, the minimum ssDNA dosage visible to the naked eye was determined. Starting from 800 nm/test, a series of twofold serial dilutions was performed. The dilution factors tested were 4, 8, 16, 32, and 64. At a 64-fold dilution, the T-line appeared faint and indistinct. To maintain experimental reproducibility while preserving visibility and detection accuracy, a 32-fold dilution of the ssDNA probe was selected for use in this study.

Passively driven chip packaging and flow of fluid process detection

This study presents a four-channel passively driven chip fabricated from polymethyl methacrylate (PMMA) using computer numerical control (CNC) machining, with the potential for cost-effective mass production via injection molding. As illustrated in Fig. 3, the device architecture comprises: (1) a sample inlet port for test sample introduction; (2) reaction chambers preloaded with lyophilized reagents, including isothermal amplification components and CRISPR/Cas12a detection systems; and (3) a detection zone integrated with lateral flow assay (LFA) strips. The system is sealed with a hydrophobic membrane. This configuration maintains a controlled internal environment—airtight to liquids and contaminants—for reliable fluid manipulation, while permitting passive pressure equilibration. This integrated approach reduces evaporation and prevents contamination. Furthermore, the microchannel connecting the reaction chamber to the vent is designed with a depth of approximately 100 μm to further minimize vapor diffusion. The operational workflow is as follows: the chip is positioned vertically, and 280–320 µL of sample is introduced through the inlet port. Driven by gravitational and capillary forces, the sample is metered into 60 ± 2 µL aliquots within each quantitative structure, while excess fluid is directed to an overflow chamber. The dispensed volume was confirmed gravimetrically by transferring the liquid from each chamber to a pre-weighed container and measuring its mass with a microbalance (ten replicates per chamber); the complete dataset is provided in Supplementary Table 3. The S-shaped channel design and sealed chamber bottoms prevent premature fluid transfer. The controlled opening of air vents then enables sequential fluid flow—first into the reaction chambers, where lyophilized reagents are rehydrated, followed by a 30-min incubation at 39 °C for isothermal amplification and CRISPR-mediated cleavage. The reaction products are subsequently directed into the detection chambers, which contain containing the LFA strips. A visual readout of pathogen-specific results is obtained within 5 min of the fluid reaching the detection zone.

Fig. 3
figure 3

Structural diagram and operating flow of the four-channel passively driven chip: (A) Front view of the four-channel PMMA microfluidic chip featuring the sample inlet and LFA detection zones, hermetically sealed with PCR-grade films. (B) Back view of the chip, featuring S-shaped flow channels, control air holes (for triggering liquid transfer), overflow chambers, and four reaction chambers preloaded with lyophilized reagents. The sealed bottom design prevents premature liquid flow into downstream compartments. (C) Side view of the four-channel PMMA microfluidic chip. (D) The sample flows through the channel to fill the sample quantitative structure, and the excess liquid flows into the overflow cavity. (E) The air holes of the reaction chamber were opened, and the quantified liquid in the sample cup structure entered the reaction chamber through the first connection channel. (F) The air holes in the detection chamber are opened, and the reaction products enter the detection chambers through the second connection channel. (H) Test completed and interpretation of the results.

Full size image

Preliminary assessment of the analytical sensitivity of the passively driven chip

Preliminary assessment of analytical sensitivity revealed that in dilution series experiments, the real-time fluorescent CRISPR/Cas12a assay generated detectable signals at the lowest concentration tested, In contrast, the lateral flow assay (LFA) results were below the visual detection threshold at this level (Fig. 4 A, C, E, G). At a relatively high concentration, both methods yielded positive results for all four target viruses (Fig. 4 B, D, F, H). These findings indicate that, under these experimental conditions, the operational sensitivity of the LFA system is comparable to that of the real-time fluorescent CRISPR/Cas12a assay at the higher concentrations, but is less sensitive at the lowest tested level.

The observed performance difference can be explained by the signal generation mechanism. The real-time fluorescent CRISPR/Cas12a assay curves at the lowest concentration did not reach the plateau phase, indicating that the cleavage of the reporter molecules was incomplete. Consequently, the quantity of activated reporter generated was insufficient to produce a visible test line in the LFA format.

Importantly, this was a preliminary qualitative assessment. A formal LOD study with sufficient replicates and statistical analysis (e.g., using probit analysis to determine the 95% detection limit) is needed to establish a statistically robust sensitivity for the assay. While the sensitivity of the LFA could potentially be enhanced by optimizing parameters such as the reporter concentration and reaction time, such adjustments would need to be balanced against the risks of increased non-specific background and prolonged assay duration.

Fig. 4
figure 4

Preliminary analysis of detection sensitivity. Representative results from serial dilution experiments for influenza A, B, human parainfluenza virus, and SARS-CoV-2 are shown. (A, C, E, G) Fluorescence detection signals. (B, D, F, H) Corresponding results from the passively driven microfluidic chip. The viral template inputs for each row, from top to bottom, were 1 × 103, 1 × 102, 1 × 101, and 1 × 100 copies/µL. NCT: negative control.

Full size image

Specificity of the passively driven chip

This study evaluated the specificity of four respiratory nucleic acid detection systems by examining their performance against positive controls for four distinct respiratory viruses: SARS-CoV-2, human parainfluenza virus, influenza A virus, and influenza B virus. As shown in Fig. 5, each detection system demonstrated exclusive reactivity to its corresponding viral target, with no detectable cross-reactivity to the remaining three viral positive viral controls (Fig. 5A-H), indicating high assay specificity. Moreover, when tested with equimolar mixtures containing two or four viral targets, each detection system accurately produced its respective positive signal without mutual interference (Fig. 5I–L). These results collectively confirm the high specificity of the developed multiplex detection platform.

Fig. 5
figure 5

Specificity of the passively driven chip detection system: (A-H) Representative fluorescence detection signals and passively driven chip results for individual viruses: (A, B) SARS-CoV-2 (COVID-19), (C, D) human parainfluenza virus (PIV), (E, F) influenza A virus (IAV), and (G, H) influenza B virus (IBV). (I) Fluorescence detection signal and (J) passively driven chip results for the multiplex detection of IAV and IBV viruses in a mixed sample. (K) Fluorescence detection signal and (L) passively driven chip results for the multiplex detection of all four viruses (SARS-CoV-2, HPIV, IAV, and IBV) in a mixed sample.

Full size image

Clinical sample validation

To evaluate the diagnostic performance of the passively driven chip detection system for respiratory virus identification, a total of 80 clinical samples were analyzed. The signal intensity of the test line (T-line) was quantitatively assessed using a numerical scoring system (0 = negative; 0.5–1 = positive), as illustrated in Fig. 6. Sanger sequencing, revealed that sample 23 tested positive; however, it yielded a negative result with the proposed detection system. For all remaining samples, the results were fully consistent with those obtained by Sanger sequencing. The virus-specific detection outcomes are summarized as follows: influenza A virus (IAV) was detected in 19 samples (samples 1, 3, 6, 11, 16, 29, 30, 31, 36, 38, 47, 53, 54, 56, 60, 73, 75, 77, 80), whereas influenza B virus (IBV) was detected in 15 samples (samples 4, 8, 14, 15, 20, 26, 33, 42, 43, 44, 62, 63, 64, 65, 66). Human parainfluenza virus (HPIV) was detected in 18 positive samples (samples 2, 5, 7, 9, 10, 12, 13, 39, 40, 41, 57, 58, 59, 68, 70, 71, 76, 79), and SARS-CoV-2 was detected in 11 positive samples (samples 17, 18, 21, 22, 24, 32, 46, 49, 50, 72, 74). These results were in complete concordance with the Sanger sequencing data, which are considered the gold standard. Notably, all 16 negative control samples were accurately identified, underscoring the system’s high specificity. The false-negative result (sample 23) is likely attributable to viral loads falling below the detection threshold. Using the Clopper-Pearson exact method, we calculated 95% confidence intervals. The assay demonstrated a sensitivity of 98.44% (95% CI: 91.6%–99.96%) and a specificity of 100% (95% CI: 79.4%–100%). These results validate the system’s high specificity and highlight its potential for reliable clinical application in respiratory virus detection (Fig. 6).

Fig. 6
figure 6

Validation of the passively driven chip in clinical samples.

Full size image

Discussion

Viral respiratory tract infections are among the most common pediatric illnesses, presenting with clinical manifestations that range from mild upper respiratory symptoms to severe, life-threatening conditions32. Human parainfluenza virus (HPIV) represents the predominant etiology of acute lower respiratory infections in children globally and is a major contributor to pediatric hospitalizations and mortality. Nearly all the children experienced HPIV infection before two years of age. During influenza seasons, pediatric populations demonstrate heightened susceptibility and typically experience more severe symptoms than adults do. Notably, although influenza—characterized by high fever and myalgia—and HPIV—commonly presenting with cough and dyspnea—share overlapping early clinical manifestations, their therapeutic strategies differ markedly33. For the detection of influenza A, influenza B, and HPIV, a range of well-established diagnostic methods targeting nucleic acids, antibodies, and proteins is available. Among these methods, reverse transcription polymerase chain reaction (RT-PCR) remains the gold standard for detecting viral nucleic acid34,35.

However, high-throughput nucleic acid testing poses several challenges, including prolonged turnaround times and heavy reliance on specialized equipment and trained personnel, which collectively impede timely outbreak containment. Therefore, there is an urgent need for rapid, sensitive, and cost-effective diagnostic tools that are suitable for point-of-care testing (POCT) and home-based self-testing. In addition to CRISPR-Cas-based fluorescent biosensors, electrochemical sensors have emerged as promising platforms for pathogen detection, enabling simultaneous analysis of nucleic acids, antigens, and antibodies36,37. Recent advancements have incorporated nanomaterials as signal reporters38, portable fluorimeters coupled with lateral flow immunoassays39, and microfluidic multiplex systems40, all aimed at enhancing signal amplification, reducing detection time, and improving sensitivity. Furthermore, the integration of chemiluminescence immunoassays with CRISPR technology could facilitate high-throughput, automated detection.

Our study employs RT-RPA coupled with CRISPR/Cas12a, building on the well-established foundation of isothermal amplification-based CRISPR diagnostics. Notably, the majority of existing platforms, particularly those designed for field deployment, have utilized loop-mediated isothermal amplification (LAMP) because of its high sensitivity and robustness41,42. This LAMP-CRISPR synergy has been successfully demonstrated in various formats, from simple lateral flow strips to integrated paper-based microfluidic devices for pathogen detection43,44. While LAMP is highly effective, our choice of RT-RPA offers distinct advantages for certain application scenarios. Specifically, RT-RPA operates optimally at a lower temperature (37–42 °C) than RT-LAMP (typically 60–65 °C), which significantly reduces the energy requirement and simplifies the heating instrumentation, making it more amenable to resource-limited or point-of-care settings. Furthermore, the reaction kinetics of RT-RPA are often faster, potentially leading to a shorter time-to-result. From an assay design perspective, RT-RPA requires only two primers, simplifying the design process and reducing development complexity compared with the multiple primers required for LAMP. Therefore, our RT-RPA-based CRISPR-Cas12a system presents a complementary and practical alternative to LAMP-based platforms, offering a favorable balance of speed, operational simplicity, and instrumental requirements for rapid, on-site molecular diagnostics.

This study developed a miniaturized, passively driven chip platform for high-throughput multiplex detection of respiratory viruses, integrating isothermal amplification and CRISPR-mediated target recognition within a unified reaction system. This method leverages the high amplification efficiency and trans-cleavage activity of RT-RPA, enabling sensitive, visual, and multiplex detection via microfluidic platforms. Key advantages include a simplified workflow (eliminating preamplification steps), high specificity through sequence-specific targeting by CRISPR, and potential for point-of-care applications. Clinical validation demonstrated the superior performance of the platform compared with conventional methods, although the limited sample size restricted the findings. Future research should consider integrating multiple detection modalities, employing more optimized probes, or combining diverse physical and chemical signal outputs to increase sensitivity and reduce costs. The platform also offers adaptability for scalable multiplex detection, enabling customization to detect varying numbers of pathogens based on specific diagnostic requirements.