Introduction

Accurate identification of biomolecular markers is fundamental to advancing biomedical research, facilitating early disease diagnosis, enabling precise monitoring of disease progression, and informing therapeutic responses1,2,3. However, the detection of single biomarkers often provides limited diagnostic value, as most pathological conditions are governed by intricate networks of molecular interactions. In this context, multiplexed analytical platforms that allow simultaneous detection of multiple biomolecules—including nucleic acids, proteins, metabolites, and cellular signals—have garnered significant attention in both research and clinical settings4,5,6. Complex diseases such as cancer and infectious disorders are characterized by heterogeneous biomarker landscapes that span both endogenous molecular signatures and microenvironmental factors7,8. Multiplex detection thus offers improved diagnostic resolution, enhances sensitivity and specificity, minimizes false positives/negatives, and reduces turnaround time for clinical decisions.

Contemporary multiplexing strategies—such as multiplex PCR9, microarrays10, and next-generation sequencing (NGS)11,12—have enabled high-throughput molecular profiling for applications ranging from genomic analysis and drug discovery to clinical diagnostics. Despite their widespread adoption, these approaches are often constrained by intrinsic limitations, including suboptimal sensitivity, operational complexity, and labor-intensive protocols. In recent years, barcode-based molecular technologies have emerged as transformative tools in multiplexed bioanalysis. These platforms leverage unique identifiers—such as nucleic acid barcodes13,14,15,16, protein barcodes17,18, and mass spectrometry barcodes19 to enable precise, facile, and high-throughput discrimination of multiple targets within a single assay. Their versatility and adaptability hold great promise for expanding the frontiers of multiplex diagnostics and personalized medicine.

Fluorescent barcoding has encoded targets by assigning distinct spectral addresses and decoding multicolor intensity patterns. Recent implementations span nanoparticle- or bead-based barcodes, intensity-ratio and lifetime schemes, and enzyme-assisted digital readouts, yet spectral approaches remain constrained by dye photophysics, inter-channel crosstalk, and demanding calibration20,21. DNA nanotechnologies mitigate some of these limits: DNA origami places fluorophores with nanometer precision to add geometric addressability and internal standards; hybridization chain reaction22 boosts signal and quantitative consistency in high-background samples; DNA points accumulation for imaging in nanoscale topography23,24 recasts encoding from colors to sequences for high-plex/single-color readout; and dynamic designs such as color-changing fluorescent barcodes (CCFB)25, quaternary four-bit encoding (FLUCO)26, and thermal-plex27 expand coding space without adding fluorescence channels. Still, practical bottlenecks persist. Complex assemblies and multi-component DNA probe sets increase the risk of dye loss, misplacement, and structural defects of DNA nanostructures during preparation and imaging. Multidimensional outputs that mix spectra, intensities, and lifetimes complicate decoding and invite misclassification, eroding specificity. Most panels are also single-use, consuming dozens of strands per run, which raises cost and limits scalability.

Here, we present a spatial fluorescence barcode (SFB) strategy based on DNA beads for multiplexed nucleic-acid detection. Each target is encoded by the bead’s physical position and read out using a single fluorescence channel, eliminating the need for multiple dye calibration and spectral unmixing, thereby simplifying the decoding process. The simplified probe architecture reduces both preparation complexity and cost. Enzyme-mediated bead resetting streamlines the encoding process and enhances practicality by allowing repeated assays without the need to re-prepare costly probe combinations. The primary sensing components are transiently luminescent DNA beads (TLDBs), which use strand displacement to trigger fluorophore activation upon target binding. Built-in nucleases degrade the target, resetting the probe for reuse. Sequential imaging then generates a spatial codebook, enabling accurate target identification through bead position. Using a single fluorophore, SFB distinguished at least seven targets and was validated for pathogen detection in blood and miRNA profiling in cancer tissue, offering a compact, resettable, and scalable solution for multiplex molecular diagnostics and spatial genomics.

Results

Validation of DNA target-responsive Y-shaped probes

The core sensing units—TLDBs—comprise streptavidin-coated polystyrene beads bearing Y-shaped DNA probes. Upon target binding, toehold-mediated strand displacement (TMSD) activates localized fluorophores via releasing the quencher-labeled strand; built-in, activity-tunable nucleases then degrade the bound target to reset the probe for subsequent detection (Fig. 1a). A spatial codebook is generated by sequential evanescent-field imaging, assigning distinct pseudo-color to each type of TLDB (Fig. 1b). Immobilized bead patterns minimize crosstalk from thermodynamic instability or nonspecific interactions, and unknowns are identified by colocalization against the reference codebook (Fig. 1c).

Fig. 1: Workflow and principle of the monochromatic Spatial Fluorescence Barcode (SFB) system.
Fig. 1: Workflow and principle of the monochromatic Spatial Fluorescence Barcode (SFB) system.
Full size image

a Fluorescence activation and autonomous resetting of DNA beads. Y-shaped DNA probes labeled with Cy5 and dual BHQ3 quenchers are immobilized on streptavidin-coated polystyrene beads. Upon target recognition, toehold-mediated strand displacement (TMSD) removes the quencher-labeled strand, triggering Cy5 fluorescence which is captured using highly inclined and laminated optical sheet (HILO) microscopy. Built-in nucleases then degrade the hybridized target strand, autonomously resetting the probe for subsequent detection cycles. The figure was created with BioRender. b Generation of the monochromatic spatial fluorescence barcode codebook (mSFBC) reference. A mixture of DNA beads with different probes is sequentially incubated with the corresponding targets. Fluorescence patterns are recorded by microscopy, pseudo-colored, and merged to generate a reference image, where each bead’s spatial position uniquely encodes its corresponding target. c Target identification in unknown samples. Fluorescence images of unknown sample SFB are compared to the mSFBC reference using spatial colocalization analysis. The probe reset mechanism enables repeated and multiplexed detection.

Y-shaped probes are initially characterized for their responsiveness to DNA targets. As illustrated in Fig. 2a and Supplementary Fig. 1, the Y-shaped probe consists of four strands: F1, F2, F3, and F4. Two BHQ3 quenchers on F2 and F3 flank the Cy5 fluorophore on F4, enabling efficient dual-sided quenching and minimizing background fluorescence. The modular design uses F3 and F4 as shared components across all codes, while F1 and F2 are target-specific, allowing flexible sequence substitution for cost-effective multiplexing. This structure offers high stability, reusability, and compatibility with enzyme-mediated resetting, making it well-suited for multiplex detection under sensitive imaging conditions. All DNA strands in the Y-shaped probes have 3’ phosphonothioate modifications, preventing nuclease degradation. The DNA targets trigger TMSD, releasing Structure 2 with dual-BHQ3 quenchers, while activating the fluorescence of Structure 1 labeled with Cy5. Preloaded ExoIII simultaneously degrades the DNA targets, facilitating Structure 2 rehybridization and the autonomous reset of the Y-shaped probes. Furthermore, Y-shaped probe assembly and reaction performance were analyzed by 12% polyacrylamide gel electrophoresis (PAGE). Supplementary Fig. 2a shows a clear band at the expected position in lane 6, indicating successful Y-shaped probe assembly. The reaction mechanism validation is presented in Supplementary Fig. 2b: lane 1 shows the Y-shaped probe-target TMSD reaction; lane 2, TMSD products after ExoIII treatment; lane 3, the DNA target alone; lane 4, the assembled Y-shaped probe; lane 5, Structure 1 alone; and lane 6, Structure 2 alone. The results confirm effective TMSD-triggered probe activation and ExoIII-mediated target cleavage and probe reset, validating the probe design and reaction pathway.

Fig. 2: Design and characterization of Y-shaped probes.
Fig. 2: Design and characterization of Y-shaped probes.
Full size image

a Four-strand Y-shaped probe labeled with a Cy5 fluorophore and dual-BHQ3 quenchers, modified with a 3’ phosphonothioate group. Target recognition triggers TMSD, releasing the BHQ3 quencher-labeled Structure 2 and activating the Cy5 fluorescence of Structure 1. ExoIII digests the DNA target, allowing Structure 2 to rehybridize and fluorescence to reset autonomously. b Activation-reset cycling stability of single Y-shaped probe IV over six repeated activation-quenching cycles. Data are mean ± S.D. (n = 3 independent experiments), and the shaded areas represent error bars. The figure contains elements created by BioRender. c Orthogonality analysis among seven Y-shaped probes I-VII. The heatmap shows normalized AUC. d Sequential activation-reset cycling of seven Y-shaped probes I-VII. The probe mixture was tested across 14 cycles, with each probe triggered twice sequentially. The concentration of all DNA in the reaction system was 50 nM, the ExoIII concentration was 20 U mL⁻¹, the reaction temperature was 37 °C, and the buffer composition consisted of 10 mM Bis-Tris-Propane-HCl, 10 mM MgCl2, and 1 mM DTT (pH 7). Data are mean ± S.D. (n = 3 independent experiments), and the shaded areas represent error bars. The figure contains elements created by BioRender. Source data can be found in the Source Data file.

Next, fluorescence kinetics were used to evaluate the target recognition and autonomous reset capability of Y-shaped probes. Supplementary Fig. 3 shows a rapid fluorescence increase upon the addition of target, followed by re-quenching within 10 min. The results indicate efficient activation and reset of Y-shaped probes. The activation-reset stability of each type of Y-shaped probe was analyzed over six repeated cycles (Fig. 2b and Supplementary Fig. 4). Quantitative analysis of fluorescence signals was performed using normalized area under the curve (AUC) (Supplementary Fig. 5a). The results indicate stable and consistent cycling performance across all Y-shaped probes. Meanwhile, orthogonality among the seven Y-shaped probes (I–VII) was validated using fluorescence kinetics (Fig. 2c and Supplementary Fig. 6) and native-PAGE (Supplementary Fig. 7). The fluorescence data show significant signals only for the matched probe-target pairs, with minimal interference from non-specific interactions. Native-PAGE analysis further confirmed orthogonality, as mobility-shifted bands were observed exclusively for the specific probe-target pairs, while non-specific interactions showed no band shift, indicating a lack of binding. Finally, the multiplex cycling stability of mixed Y-shaped probes I-VII was tested across 14 sequential activation cycles, with each type of Y-shaped probe activated twice. Normalized AUC confirms excellent cycling stability and independent fluorescence response of Y-shaped probes in the multiplex system (Fig. 2d and Supplementary Fig. 5b). Overall, the results validate the specificity, cycling efficiency, and multiplex compatibility of Y-shaped probes, supporting their application in constructing SFB systems. In addition, the fluorescence differences observed across different targets are due to sequence-dependent kinetics of the TMSD reactions, influenced by sequence context, GC content, and secondary structures, rather than variations in probe content or concentration28,29. It is noteworthy that these differences do not affect target identification, as recognition relies on spatial colocalization.

Characterization of transiently luminescent DNA beads

DNA beads are constructed by conjugating Y-shaped probes onto polystyrene beads (3–4 μm). TEM imaging and DLS analysis confirmed that beads exhibit uniform morphology and a narrow size distribution (average diameter 5.086 µm), ensuring good monodispersity and stable performance, as shown in Supplementary Fig. 8. As illustrated in Fig. 3a, streptavidin-coated polystyrene beads immobilize biotin-labeled Y-shaped probes. 12% native PAGE analysis confirmed successful Y-shaped probe conjugation onto bead surfaces, showing a significant decrease in free Y-shaped probe intensity after incubation with streptavidin-coated beads (Supplementary Fig. 9a). Zeta potential measurements further supported this, with a clear negative shift in surface charge from −26.9 mV (bare beads) to −40.2 mV (DNA beads) after probe attachment (Supplementary Fig. 9b). The Quartz Slides treated with 3-aminopropyltriethoxysilane (APTES) electrostatically immobilize these negatively charged beads for fluorescence microscopy (Fig. 3b). As illustrated in Fig. 3c, HILO microscopy records transient luminescence behavior of DNA beads. Upon adding DNA target to DNA beads with built-in ExoIII, the beads are initially illuminated, then return to dark within 25 min. In the absence of ExoIII, fluorescence continuously increases without re-quenching. Normalized fluorescence intensities matched these imaging observations (Fig. 3d). Flow cytometry was used to evaluate the fluorescence activation-reset performance of DNA beads at different ExoIII concentrations (Fig. 3e). Without ExoIII, fluorescence was activated without reset. Increasing ExoIII concentration (50, 200, and 500 U mL⁻¹) progressively enhances fluorescence reset. 500 U mL⁻¹ ExoIII was chosen for subsequent experiments. Furthermore, fluorescence imaging confirms orthogonality among the seven types of DNA beads (Fig. 3f and Supplementary Fig. 10). Strong fluorescence signals appear exclusively in matched probe-target pairs, with negligible cross-reactivity observed, confirming orthogonality among probes. The immobilization approach of multiple DNA beads effectively reduces the potential crosstalk signal caused by thermodynamic instability or nonspecific interactions.

Fig. 3: Construction and characterization of TLDBs.
Fig. 3: Construction and characterization of TLDBs.
Full size image

a Schematic of DNA bead construction. Streptavidin-coated polystyrene beads (3–4 μm) immobilize biotin-modified Y-shaped probes. The figure contains elements created by BioRender. b Immobilization of DNA beads of microscopy slide. Positively charged APTES-treated quartz slides electrostatically adsorb negatively charged DNA beads for fluorescence imaging. The figure contains elements created by BioRender. c Fluorescence microscopy imaging of DNA beads captured every 5 min using HILO microscopy. Scale bar: 5 μm. Data are mean ± S.D. (n = 3 independent experiments). d Normalized average fluorescence intensity extracted from images in 3c. e Flow cytometry analysis of DNA beads at varying ExoIII concentrations (50, 200, and 500 U mL⁻¹). f Fluorescence imaging-based orthogonality analysis of seven DNA beads. Source data can be found in the Source Data file.

Additionally, Supplementary Fig. 11 demonstrates the effective autonomous resetting of DNA beads over five detection cycles. The PCC values for cycles 2–5 compared to cycle 1 are 0.94, 0.91, 0.87, and 0.85, respectively, indicating reliable target recognition and spatial consistency. The results confirmed the stable spatial addresses of DNA beads and their excellent reusability. The fluorescence signal for each target is driven by sequence-specific hybridization, with all beads functionalized with identical Y-shaped probes at uniform loading densities. The platform achieves a detection limit of approximately 1 pM, with fluorescence activation observed and complete signal quenching through autonomous probe resetting (Supplementary Fig. 12a). Microbeads are considered positive only if their fluorescence intensity exceeds the no-target control baseline by at least three standard deviations. Moreover, per-bead normalized fluorescence at 100 pM across independent FOVs showed comparable distributions, with no significant differences observed (F(2, 64) = 2.241, p = 0.1147, one-way ANOVA), supporting spatially uniform signal detection across regions (Supplementary Fig. 12b).

Multiplexed recognition based on SFB

The SFB system uses DNA beads with distinct sequences for multiplexed nucleic acid detection. As shown in Fig. 4a, a mixture of DNA beads was immobilized on an APTES-treated quartz slide and imaged before and after the addition of targets. Fluorescence imaging captures the spatial positions of the DNA beads, with synthetic ssDNA targets used to identify the nucleic acid information. Sequential introduction of Targets I-VII activates fluorescence on the corresponding beads, followed by probe resetting, enabling cyclic imaging. Each target is assigned a unique pseudo-color. Merged fluorescence images of all SFB construct monochromatic spatial fluorescence barcode codebook (mSFBC), encoding the spatial distribution of seven distinct targets, which serves as a reference for analyzing targets in unknown samples. Supplementary Fig. 13 demonstrates fluorescence imaging of mixed DNA beads in both the quenched and activated states by each target. The fluorescence of all targets is significantly higher than the quenched state, implying that there is no cross-reactivity between beads. After resetting the Y-shaped probes, multiplexed detection is performed on unknown samples (Fig. 4b), confirming the presence of Targets I, II, III, and VI in the sample. Once reset, the SFB system remains reusable for further detection. Figure 4c provides a bead-by-bead comparison of the TLDB fluorescence images between the unknown sample and the corresponding mSFBC. Figure 4d shows a colocalization analysis of TLDB fluorescence images across three independent fields of view (Fig. 4a and Supplementary Fig. 14), confirming the accuracy of target identification. The mean Pearson correlation coefficient (mPCC) values for Targets I-VII are 0.82, 0.79, 0.75, 0.87, 0.86, 0.66, and 0.67, respectively, all exceeding the 0.5 threshold, ensuring reliable target recognition30. These results confirm that the SFB system accurately decodes spatial fluorescence signals from targets, enabling reliable and efficient multiplexed detection. Notably, constructing a codebook is a standard step in multiplexed assays to ensure accurate decoding. The enzyme-mediated probe resetting improves efficiency by allowing batch testing—thereby reducing time and cost.

Fig. 4: Multiplexed detection and spatial identification using the SFB system.
Fig. 4: Multiplexed detection and spatial identification using the SFB system.
Full size image

a Generation of the mSFBC. The sequential introduction of Targets I-VII enables cyclic fluorescence imaging. TMSD reactions between targets and their corresponding Y-shaped probes on polystyrene beads transiently activate fluorescence, revealing the spatial positions of target-associated DNA beads. ExoIII-mediated target degradation resets the fluorescence signal, completing each cycle. Each type of DNA target (TLDBs) assigns a distinct pseudo-color. Merged fluorescence images of all SFBs corresponding to TLDBs generate the mSFBC reference, encoding spatial information for each target. b TLDB fluorescence images of an unknown sample SFB compared to the mSFBC, identifying Targets I, II, III, and VI. c Bead-by-bead comparison of TLDB fluorescence images in the unknown sample SFB (left) and the corresponding mSFBC (right). Data are mean ± S.D. (n = 3 independent experiments). The figure contains elements created by BioRender. d Colocalization analysis across three independent imaging fields. Each dot represents a PCC value between corresponding DNA beads in the unknown sample SFB and mSFBC reference. mPCC values for DNA beads I-VII are shown above each group. Error bars represent standard deviation. Scale bar: 5 μm. Source data can be found in the Source Data file.

Multiplex detection of pathogen-associated DNA targets

Antimicrobial resistance poses a critical global health threat, with resistant infections increasing at an alarming rate. Rapid gene detection is essential for guiding clinical decisions and mitigating the spread of resistant pathogens31,32. Many drug-resistant bacteria evade treatment by carrying species-specific marker genes and high-virulence factor genes, necessitating precise identification33,34,35. To address this challenge, the SFB system was utilized to detect multiple bacterial species, including Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa. To ensure specificity, conserved regions of resistance-associated genes were identified using NCBI BLAST, followed by multiple sequence alignments36. A systematic selection process determined species-specific marker genes (ddl, peg-344, gltA, and oprL) and hypervirulence genes (pvl, iucA, and hemO) (Fig. 5a). The selected genetic targets are critical for diagnosing high-risk, multidrug-resistant, and hypervirulent pathogens, focusing on species identification, antimicrobial resistance, and hypervirulence31,37,38,39,40. Detailed target information is provided in Supplementary Data 2. Based on these amplicon sequences of plasmid containing pathogenic sequences, PCR primers were designed using Primer3, while NUPACK41 assisted in designing seven Y-shaped probes for target detection (sequences listed in Supplementary Data 1). As shown in Fig. 5b, pathogen-associated plasmid samples (sequences listed in Supplementary Data 3) underwent PCR amplification to generate dsDNA, followed by λ Exo digestion to obtain ssDNA targets42. Supplementary Fig. 15 illustrates sequential fluorescence imaging of synthetic DNA targets (iucA, ddl, gltA, hemO, peg-344, oprL, pvl) using SFB cyclic imaging. Each target was assigned a pseudo-color. Merged fluorescence images of all TLDBs constructed the mSFBC, encoding the spatial distribution of these DNA targets. Following fluorescence reset, multiplexed detection was performed on unknown samples. TLDB fluorescence images were referred to the mSFBC reference. Colocalization analysis identified gltA in unknown sample SFB 1 and hemO in unknown sample SFB 2, demonstrating that a single mSFBC enables the detection of multiple unknown samples (Fig. 5c). Supplementary Fig. 16 presents bead-by-bead comparison of TLDB images between the unknown SFB and their corresponding mSFBC reference. The mPCC values for gltA and hemO were 0.88 and 0.78, respectively (Fig. 5d). These results confirm that the SFB system accurately decodes spatial fluorescence signals, supporting precise multiplexed pathogen detection. Figure 5e illustrates the workflow used to detect pathogens from clinical blood samples. DNA was extracted from samples, followed by PCR amplification to generate double-stranded DNA (dsDNA). Single-stranded DNA (ssDNA) targets were then produced, enabling identification using SFB system. Additionally, DNA sequencing was employed to validate the SFB system’s results. The SFB system identified pathogen-associated genes across all infected clinical samples (n = 10), matching the reference sequencing results (Table 1). Notably, up to 10 clinical samples were processed in batch within a single run using the same mSFBC reference, highlighting both the practical efficiency and the robust regeneration capability of the SFB system.

Fig. 5: Multiplexed detection of drug-resistant pathogen genes using the SFB system.
Fig. 5: Multiplexed detection of drug-resistant pathogen genes using the SFB system.
Full size image

a Target genes associated with multiple antibiotic-resistant bacteria: Enterococcus faecium (ddl gene), Staphylococcus aureus (pvl gene), Klebsiella pneumoniae (peg-344 gene, iucA gene), Acinetobacter baumannii (gltA gene, hemO gene), and Pseudomonas aeruginosa (oprL gene). b Schematic of plasmid-derived ssDNA target extraction. Pathogenic sequences undergo PCR amplification, dsDNA digestion by λ Exo, and purification to yield ssDNA targets. The figure contains elements created by BioRender. c Identification of DNA targets in the unknown samples. Fluorescence images of two unknown SFB were referred to the mSFBC reference, identifying gltA in the unknown sample SFB 1 and hemO in the unknown sample SFB 2. Scale bar: 5 μM. The figure contains elements created by BioRender. d TLDB fluorescence images colocalization analysis between the unknown SFB and the mSFBC reference. The mPCC values for gltA and hemO are 0.88 and 0.78, respectively. e Procedure of the detection of pathogen genes in clinical blood samples. The figure contains elements created by BioRender. Source data can be found in the Source Data file.

Table 1 Pathogen gene detection results of clinical samples
Full size table

Detection of miRNA targets using the SFB system

Cancer remains one of the most significant global health challenges43,44. miRNAs regulate oncogenesis by modulating endogenous gene expression, making them valuable biomarkers for cancer diagnosis45,46. To enable multiplex miRNA detection using the SFB system, ExoIII was replaced with Ribonuclease H (RNase H) for miRNA digestion. RNase H selectively cleaves miRNA targets in DNA-RNA hybrids47,48. The reaction process remains the same as in the ExoIII-based SFB system (Fig. 6a). Computational analysis of The Cancer Genome Atlas (TCGA) database identified highly expressed miRNAs in breast cancer49. Four miRNAs—miR-182 (2.95 × 104), miR-143 (6.20 × 104), let-7a (2.05 × 104), and miR-183 (1.45 × 104)—were selected based on their cancerous tissue expression levels (see “Code Availability”). This analysis, which uses both example data and real TCGA datasets, enables reproducible identification of key miRNAs and their expression levels across cancer samples. Y-shaped probe sequences were designed using NUPACK41 (sequences listed in Supplementary Data 1). The interaction between Y-shaped probes and miRNA targets was verified via 12% PAGE, confirming successful TMSD reactions and RNase H-mediated miRNA degradation (Supplementary Fig. 17). Fluorescence cycling confirmed the stability of the Y-shaped probes in both individual and mixed systems (Fig. 6b and Supplementary Figs. 18, 19). Fluorescence kinetics validated the specificity of Y-shaped probes for their miRNA targets (Supplementary Fig. 20). All targets demonstrated high specificity. Furthermore, microscopy imaging confirmed the orthogonality of the four DNA beads, with minimal cross-reactivity observed (Fig. 6c and Supplementary Fig. 21). Fluorescence imaging of sequentially introduced miR-182, miR-143, let-7a, and miR-183 assigned pseudo-colors of purple, green, blue, and cyan, respectively (Supplementary Fig. 22). Merged fluorescence images of all TLDBs constructed the mSFBC reference, encoding the spatial distribution of these miRNA targets. Following fluorescence reset, multiplexed detection was performed on unknown samples. Comparison with the mSFBC reference identified miR-182 and miR-143 (Fig. 6d). Bead-by-bead colocalization analysis of TLDB fluorescence images of the unknown SFB and the corresponding beads in the mSFBC reference (Supplementary Fig. 23) yielded mPCC values of 0.67 for miR-182 and 0.89 for miR-143, verifying detection accuracy (Fig. 6e). Figure 6f illustrates the workflow for detecting miRNAs from clinical samples using the SFB system. miRNAs are extracted from tissue, enriched with magnetic beads, and identified using SFB. Figure 6g presents clinical validation of the SFB system in breast cancer and healthy samples. The table shows that miR-182, miR-143, let-7a, and miR-183 are consistently detected in breast cancer tissues, indicating their strong association with breast cancer. In contrast, only a few healthy samples exhibit these miRNAs, suggesting lower expression in non-cancerous tissues. Similarly, multiple samples in Fig. 6g can be profiled in batch within a single run using the same mSFBC reference. This differential detection highlights their potential as biomarkers for breast cancer diagnosis. These results validate the clinical applicability of the SFB system for distinguishing cancerous from healthy tissues based on miRNA expression.

Fig. 6: Multiplexed detection of miRNA targets using the RNase H-based SFB system.
Fig. 6: Multiplexed detection of miRNA targets using the RNase H-based SFB system.
Full size image

a RNase H-based SFB system for miRNA detection. ExoIII was replaced with RNase H for miRNA recognition, while the reaction process remained the same as in the ExoIII-based SFB system. b Sequential fluorescence activation of four Y-shaped probes (miR-143, let-7a, miR-183, and miR-182). Four Y-shaped probes mixture was tested across eight cycles, with each type of probe triggered twice sequentially. The concentration of all DNA in the reaction system was 50 nM, RNase H concentration was 50 U mL-1, the reaction temperature was 37 °C, and the buffer composition consisted of 50 mM Tris-HCl, 75 mM KCl, 3 mM MgCl2, and 10 mM DTT (pH 8.3). Data are mean ± S.D. (n = 3 independent experiments), and the shaded areas represent error bars. c Orthogonality evaluation of four miRNA targets. The figure contains elements created by BioRender. d Identification of miRNA targets in the unknown sample. Fluorescence images of the unknown SFB were referred to the mSFBC reference, identifying miR-182 and miR-143. Scale bar: 5 μm. Data are mean ± S.D. (n = 3 independent experiments). The figure contains elements created by BioRender. e TLDB colocalization analysis. mPCC values of 0.67 for miR-182 and 0.89 for miR-143 confirmed colocalization accuracy. The set threshold of mPCC>0.5 validated the identification of both targets. f Workflow of miRNA extraction and detection using the SFB system. The figure contains elements created by BioRender. g Clinical validation of miRNA detection in breast cancer and healthy samples. Orange markers indicated the target miRNAs detected. Source data can be found in the Source Data file.

Discussion

In this study, we introduced an SFB platform based on TLDBs for multiplexed and reusable nucleic acid detection. By combining toehold-mediated strand displacement with enzyme-mediated probe resetting, the system enables repeated detection cycles using only single-color fluorescence. Targets are differentiated through spatial encoding, eliminating the need for complex dye combinations, spectral unmixing, or wash steps. The platform’s clinical applicability was demonstrated in the detection of pathogen-derived nucleic acids from infected blood and cancer-associated microRNAs from tissue samples, highlighting its potential utility in infectious disease diagnostics and oncology biomarker profiling.

Recent barcoding strategies such as CCFB25 and FLUCO26 have demonstrated significant progress in multiplexed nucleic acid analysis. CCFB enables intuitive visual decoding via sequential strand displacement-driven color changes25, while FLUCO achieves high-density barcoding through combinatorial use of multiple fluorophores at varying intensities26. They addressed some drawbacks of conventional fluorescence barcoding, notably the limited coding capacity imposed by spectrally resolved dyes and the loss of specificity due to spectral crosstalk. However, they also present practical challenges. CCFB requires multi-step hybridization reactions and lacks reusability, which limits the stability and throughput. FLUCO relies on complex multi-channel imaging and precise dye calibration, increasing system complexity and reducing accessibility for point-of-care or resource-limited applications. In contrast, the SFB system offers clear practical advantages. It encodes each target by the bead’s physical position and reads it out with a single fluorescence channel, eliminating multicolor labeling and spectral unmixing. A simplified probe architecture together with an intensity-independent decoding procedure reduces synthesis, imaging, and analysis complexity. Enzyme-mediated resetting enables bead reuse without probe reloading, supporting high-throughput, single-color assays that avoid complex dye panels, computational unmixing, and extra wash steps. These features make SFB particularly suitable for scalable, cost-effective, and portable detection.

Despite its advantages, the SFB platform presents several limitations that suggest directions for further optimization. First, the current coding scale is constrained by imaging resolution, field-of-view size, and bead-immobilization density. While SFB presently supports seven targets, capacity can increase by packing beads more densely, acquiring multiple fields, or integrating orthogonal encodings such as multi-color50 or high-resolution imaging51. Second, repeated detection cycles may suffer from enzyme degradation, fluorophore photobleaching, or bead-surface wear, which can compromise signal stability. Future iterations may benefit from thermostable enzyme variants52 more photostable dyes53 or quantum dots54, and surface-protection strategies. Third, variations in strand-displacement kinetics across probes can introduce signal heterogeneity, affecting quantitation and reproducibility; rational probe design, kinetic optimization, and normalization during image analysis55 can mitigate these effects. Fourth, array-based assays, SFB requires a reference codebook. In practice, this overhead is amortized in SFB by batch testing: a single, validated codebook can be reused across multiple samples and runs, reducing time and cost16,56. Looking ahead, slide-level fiducial labeling and automated stage control could anchor long-lived, reusable codebooks and enable seamless multi-field mapping, further boosting throughput and scalability57,58. For sustained performance over repeated cycles, stabilization strategies such as bead-surface coatings and enzyme immobilization will be implemented to preserve bead integrity and maintain the consistent functionality of DNA probes and catalytic components across extended assays59,60. Addressing these aspects will strengthen the platform’s scalability, robustness, and diagnostic utility in clinical and research settings.

Methods

Ethical statement

In this study, blood and tissue samples were obtained from China-Japan Friendship Hospital and Tianjin Chest Hospital for the development of the multiplexing detection method. Participant demographic information was available (Supplementary Data 4). All participants provided written informed consent, including explicit consent to share potentially identifying individual-level data. All samples were handled in accordance with ethical guidelines for care and use of clinical samples of Beijing University of Chemical Technology (BUCT). The study was approved by the Ethics Review Board of BUCT. The assigned approval number is 2023-112-k21. A portion of the original samples was utilized for this research; the remainder has been stored at –80 °C to facilitate subsequent studies. Any request for the sharing of these samples must be formally submitted to the BUCT Ethics Review Board for review and approval.

Reagents and materials

All the oligonucleotides (sequences listed in Supplementary Data 1) and plasmids (sequences and related information are listed in Supplementary Data 3) used in this work were synthesized by Sangon Co. (Shanghai, China). The modified oligos were purified by HPLC, while the unmodified oligos were purified by PAGE. Exo III and other enzymes and their corresponding buffers were obtained from NEB (#M0262L). RIBOBIO (Guangzhou, China) provided the miRNA inhibitor, and APTES (# 440140) was purchased from Sigma-Aldrich. Additional purification was not performed for all chemicals as received. DNase/RNase-free deionized water from Tiangen Biotech Co. (#RT121-01) was used in all experiments.

Fluorescence kinetics measurement of nucleic acid probes

The reaction system included 50 nM Y-shaped probe and varying concentrations of either 20 U mL⁻¹ ExoIII or 50 U mL⁻¹ RNase H, prepared in 200 μL PCR tubes. The ExoIII system contained its specific buffer composed of 10 mM Bis-Tris-Propane-HCl, 10 mM MgCl2, and 1 mM DTT (pH 7.0). The RNase H system utilized a buffer consisting of 50 mM Tris-HCl, 75 mM KCl, 3 mM MgCl2, and 10 mM DTT (pH 8.3). After 1 μL target (50 nM) addition, fluorescence intensity was immediately analyzed using a Rotor Gene Q real-time PCR system (QIAGEN, Germany). The system continuously recorded fluorescence at 37 °C with a gain value of 10 and a time interval of 10 s. The thermal cycling conditions were set as follows: 90 °C for 5 min, 90 °C for 1 min, then a temperature decrease of 1 °C per cycle for 65 cycles, 25 °C for 5 min, finally stored at 4 °C. The fluorescence kinetics data were baseline-corrected, and the AUC was calculated by integrating fluorescence intensity over time, reflecting both signal strength and persistence. The AUC was used to comprehensively evaluate probe performance, orthogonality, and stability across repeated detection cycles in a homogeneous solution.

Polyacrylamide gel electrophoresis (PAGE) of nucleic acid probes

12% native polyacrylamide gel (acrylamide/bis-acrylamide = 19: 1) was prepared in 0.5×TBE electrophoresis buffer (44.5 mM Tris, 44.5 mM boric acid, 1 mM EDTA, pH 8.0). After preparation, electrophoresis was performed at 120 V for 1 h. Probe samples at a concentration of 500 nM were mixed with 6×loading buffer at a 5: 1 ratio (v/v), and 5 μL of the mixture was loaded into each well. After electrophoresis, the gels were stained with SYBR Gold, and imaging was conducted.

DNA functionalized beads preparation and flow cytometry analysis

Biotin-modified Y-shaped probes (100 nM) were incubated with 3–4 μm streptavidin-coated polystyrene beads (0.8 μg mL⁻¹) at 4 °C for 12 h using a four-dimensional rotator. To prevent bead aggregation and cross-reactivity, non-magnetic, polystyrene-based DNA beads are used and immobilized on APTES-treated quartz slides through electrostatic adsorption. This design ensures stable bead positioning and prevents physical clustering. After incubation, the DNA beads were combined with 0, 50, 200, and 500 U mL⁻¹ ExoIII for 1 h to evaluate enzymatic activity. The 200-mesh filter was applied to remove aggregates before analysis. Flow cytometry (FACSVerse) was used 100,000 events were collected per sample. Fluorescence intensity data is processed and analyzed using FlowJo software.

Reaction chamber preparation and microscope setup for DNA bead fluorescence imaging

The quartz slides (24 × 30 mm) underwent APTES functionalization. After immersion in a 1: 50 APTES-acetone solution for 15 min, nitrogen drying followed. As shown in Supplementary Fig. 24 a 3D-printed conical microchamber (~47 µL) was bonded to a quartz slide and sealed with a transparent lid. A 20 µL calibration mark is engraved on the sidewall. A demonstration of microchamber assembly and sample loading operations on the motorized stage is shown in Supplementary Movie 1. The HILO microscopy system (Nikon Ti-U, 100× objective) integrated with an EMCCD camera (iXon 897, Andor) recorded fluorescence images. The 647 nm laser was coupled into a single-mode fiber, secured in a fiber launcher with an automated XY fiber holder for precise incidence angle adjustment, and mounted on a micrometer-driven optical guide. The 647 nm red laser was used to detect Cy5 (the acceptor) and capture the fluorescence images. As shown in Supplementary Fig. 25, the microscope can simultaneously monitor and switch between three fields of view, each containing 100–200 beads.

mSFBC imaging

The as-prepared Y-shaped probe functionalized beads were added to the chamber and incubated for 30 min at room temperature. After electrostatic immobilization, the chamber was washed three times with PBS. 20 μL of imaging solution was added containing 500 U mL⁻¹ ExoIII or 1000 U mL⁻¹ RNase H and additional 500 nM dual-BHQ3 labeled Structure 2 which guarantees the rapid reset of Y-shaped probes. The imaging solution contains the corresponding buffer for ExoIII or RNase H. The quartz slide was placed under the microscope, and the HILO incidence angle was adjusted to excite the bead's surface. Before imaging, the camera cooled to –85 °C. Imaging was performed in HILO mode (exposure 0.01 s). Laser exposure was reported as per-frame fluence H (J·cm−2), with P = 3.2 mW measured at the sample plane and ROI area A = 1.11 × 10−5 cm2, giving H = 2.9 J·cm−2 per frame. Following this setup, 1 μL of target was introduced into the imaging solution, allowing for the capture of transient luminescence images. The Y-shaped probe was permitted sufficient time (10–20 min) to reset before the subsequent repetition of the process. Different targets were introduced sequentially to activate transient fluorescence signals on beads.

DNA extraction from blood and amplification

Pathogen DNA was isolated from whole blood samples using the QIAamp DNA Blood Mini Kit (Qiagen) in accordance with the provided instructions. In brief, 200 µL of blood was treated with proteinase K and lysis buffer to disrupt cells and release nucleic acids. Ethanol was then added to promote DNA binding to the silica membrane within the spin column. Following a series of wash steps to remove proteins and other contaminants, the DNA was eluted in 50 µL of nuclease-free buffer. For amplification, PCR was carried out using pathogen-specific primers. Thermal cycling was performed with an initial denaturation at 95 °C for 3 min, followed by 35 cycles of 95 °C for 30 s, 55–60 °C for 30 s, and 72 °C for 1 min, ending with a final extension at 72 °C for 5 min. After amplification, 1 μL of λ Exo (final concentration 100 U mL⁻¹) was added to generate ssDNA targets, which were then stored at –20 °C.

RNA extraction from tissues and enrichment by magnetic beads

Tissue samples were first homogenized in lysis buffer using a mechanical homogenizer, and total RNA, including small RNAs such as miRNAs, was extracted using a commercial kit (miRNeasy Mini Kit, Qiagen) according to the manufacturer’s guidelines. The homogenate underwent phase separation with phenol-chloroform, and RNA was subsequently purified through a silica membrane-based column. To selectively enrich miRNAs, magnetic beads conjugated with biotinylated capture probes complementary to target miRNA sequences were incubated with the extracted RNA at 4 °C for 0.5 h under gentle rotation. After thorough washing to eliminate non-specifically bound nucleic acids, the enriched miRNAs were eluted with nuclease-free water and stored at –80 °C for further analysis. According to our recently reported protocol for target miRNA enrichment61, the target miRNAs can reach 1–100 pM for SFB detection.

Target identification in clinical samples

After the generation of mSFBC images, all Y-shaped probes underwent resetting before the sample addition process was repeated. Subsequently, 1.5 μL of an unknown sample containing multiple ssDNA or miRNA targets was introduced into the fluorescence imaging solution. TLDB fluorescence images of the unknown sample were recorded.

Sanger sequencing

NGS was not used in this study because the analysis focuses on a limited number of targets (7 DNA), making Sanger sequencing a more cost-effective and straightforward method for precise identification. Sanger sequencing of the PCR amplicons of pathogen-specific genomic regions was performed. The resulting sequences were aligned to reference pathogen genomic sequences to confirm target identity.

Statistics and reproducibility

Sample sizes were determined based on prior studies and practical considerations, with no statistical method used to predetermine sample size. All experiments were independently reproduced at least three times, and no data were excluded from the analyses. Clinical samples were randomly assigned to minimize potential bias from batch effects or handling order, and sample analysis was performed in a blinded manner with coded identifiers to ensure objective assessment and minimize bias in the evaluation of the In Vitro Diagnostics method’s performance. Fluorescence intensity analysis was conducted using GraphPad Prism 10 (version 10.3.1, (464)). Data are reported as mean±S.D. Statistical significance was determined by one-way ANOVA. Fluorescence imaging data processing (e.g., colocalization analysis) was conducted using ImageJ (version 1.54 P).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.