Hybridization chain reaction-based aptasensor for sensitive detection of Staphylococcus aureus in milk

Abstract

Staphylococcus aureus (S. aureus) is recognized as a high-risk pathogen responsible for severe food safety incidents. To address this concern, this study developed a fluorescent aptasensor based on fluorescence resonance energy transfer (FRET) for the rapid detection of S. aureus in milk. The proposed strategy employed high-specificity aptamers as recognition elements. Target bacteria were specifically captured, enriched, and isolated using aptamer-functionalized magnetic beads (MBs), thereby converting the concentration of S. aureus into a measurable fluorescence signal. To further enhance analytical sensitivity, a hybridization chain reaction (HCR)-based nucleic acid amplification strategy was integrated to achieve cascaded signal amplification. Under optimal conditions, the developed method achieved a limit of detection as low as 5.3 × 10¹ CFU/mL for S. aureus. This work provides a valuable reference for the development of biosensors with significant potential for diverse target detection in food safety and environmental monitoring.

Highlights

• •FRET-based aptasensor with HCR amplification for S. aureus detection.
• •Integrated magnetic beads and HCR for cascaded signal amplification.
• •Achieved detection limit of 53 CFU/mL.
• •Obtained recoveries of 94–110% and RSDs of 1.07–3.89% in spiked milk.
• •Showed excellent specificity against non-target bacteria up to 10⁶ CFU/mL.

1. Introduction

Staphylococcus aureus (S. aureus), as one of the important pathogenic bacteria causing food safety incidents, has long been a research hotspot and technical challenge in the field of food safety supervision for achieving rapid and accurate detection (Oniciuc et al., 2017; Rubab et al., 2018; van Asselt et al., 2016). While traditional methods like culture methods (Sun et al., 2025; Tenover et al., 2019), immunological assays (Hu et al., 2021; Zhao et al., 2023), and molecular biological techniques (Chen et al., 2025; Gan et al., 2025; Kou et al., 2024) have been widely utilized, each of these approaches presents certain limitations in meeting the demands for rapid detection. The conventional plate counting method has long been regarded as the “gold standard” for bacterial detection. Nevertheless, the detection procedure requires 48–72 h to complete, which substantially limits its application in rapid response to emergencies. Immunological techniques, such as enzyme-linked immunosorbent assay (ELISA) (Jiao et al., 2023; Li et al., 2024b; Tong et al., 2025) and lateral flow immunochromatography (LFIA) (Huang et al., 2023; Liu et al., 2024b; Yin et al., 2025a), are employed due to their operational simplicity. However, these methods are heavily dependent on specific antibodies, whose production is not only time-consuming (typically requiring 3 to 6 months) but also costly. More critically, in complex food matrices (e.g., milk), antibodies are prone to non-specific binding with non-target components, leading to an increased risk of false-positive results and compromising assay accuracy. Molecular methods, including polymerase chain reaction (PCR) (Xu et al., 2025b; Zhang et al., 2025a) and loop-mediated isothermal amplification (LAMP) (Dai et al., 2026; Trotter et al., 2024; Xu et al., 2012), offer high sensitivity but face practical limitations that impede their widespread adoption. Conventional PCR relies on precise thermal cycling instruments and necessitates post-amplification analysis via agarose gel electrophoresis. This additional step is time-consuming and involves the use of toxic reagents such as ethidium bromide, raising operational safety concerns and increasing waste disposal complexity. Although real-time quantitative PCR (qPCR) enables closed-tube detection and simplifies workflows, its reliance on expensive instrumentation and the need for skilled personnel restricts its application. Therefore, the development of novel rapid detection strategies that can overcome these limitations is of great significance for enhancing the prevention and control of foodborne risks associated with S. aureus.

In order to address the drawbacks of traditional detection methods, considerable research efforts have been directed toward the development of biosensor-based platforms for pathogenic bacteria detection. Over recent years, a variety of biosensors, such as colorimetric (Hu et al., 2025; Liu et al., 2024a; Shu et al., 2026), electrochemiluminescence (Gao et al., 2025b; Jiang et al., 2025; Song et al., 2024) and fluorescent biosensors (Gest et al., 2024; Hsiung & Zhang, 2024; Kamijo et al., 2025) have demonstrated considerable potential in bacterial detection. Among them, fluorescent biosensors have attracted particular interest owing to their ease of use, fast response time, and suitability for high-throughput applications. Recent advances in fluorescent sensing have further expanded their analytical capabilities, enabling highly sensitive and selective detection in complex biological matrices through innovative probe design and signal transduction strategies with broad applications spanning chemical and biological detection (Wang et al., 2022b). Liu et al. developed lanthanide nanoparticle‑gold (LNPs-Au) nanoprobes with robust NIR-II emission for fluorescence temperature sensing (Liu et al., 2023). The NIR-II emission enables deep tissue penetration and real-time non-contact temperature measurement in the biologically relevant range. Another work by Liu et al. developed a reusable multi-enzyme cascade bioreactor, achieving effective fluorescence discrimination and quantitation of amino acid enantiomers with enhanced cascade catalytic activity (Liu et al., 2022). Cao et al. reported a fluorescent linear polyurea which exhibits broad emission and enables low-cost Fe³⁺ detection and paper-based sensor applications (Cao et al., 2020). These examples highlight the versatility and practical potential of fluorescence-based platforms. The specificity of these sensors is ensured by the affinity between the recognition element and the target microorganism, enabling accurate and selective detection. In this context, aptamers have emerged as particularly promising recognition elements. As chemically synthesized single-stranded oligonucleotides, aptamers offer distinct advantages over traditional antibodies, including superior thermal stability, lower production cost, excellent batch-to-batch reproducibility, and facile chemical modification. These properties make aptamers ideal candidates for developing robust and reliable biosensing platforms. The versatility of aptamer-based biosensors has been well established across diverse applications, ranging from heavy metal ion and mycotoxin detection to antibiotic monitoring and pathogen identification (Niu et al., 2024; Xu et al., 2025a; Zhang et al., 2025b). Their integration with magnetic materials has further enabled efficient separation and enrichment in complex sample matrices (Li et al., 2024a), while dual-color fluorescent aptasensors have been developed for the simultaneous detection of multiple food contaminants, illustrating the broad utility of aptamer-based fluorescence strategies (Yin et al., 2025b). Additionally, to improve analytical sensitivity, these platforms are frequently integrated with signal amplification strategies. The emergence of isothermal nucleic acid amplification techniques has further expanded the functionality of fluorescent biosensors, enhancing their adaptability and broadening their application.

Hybridization chain reaction (HCR), a typical enzyme-free isothermal amplification technique, has demonstrated significant advantages in the field of biosensing in recent years owing to its distinct advantages over other signal amplification strategies (Bai et al., 2024; Lv et al., 2025; Prinzen et al., 2020; Wang et al., 2023). HCR operates under mild thermal conditions (25–37 °C) without enzymatic catalysis, employing two well-designed hairpin probes to propagate a cascade of hybridization reactions that yield exponential signal gain. Beyond its amplification efficiency, HCR is distinguished by its remarkable specificity. The previous studies have demonstrated that a single-base mismatch can diminish signal intensity by more than 90%(Mo et al., 2023). Furthermore, the versatility of HCR is reflected in its compatibility with diverse detection readouts, including fluorescence (Xu et al., 2022), colorimetric (Chi et al., 2024), and electrochemical platforms (Xu et al., 2026), thereby enabling flexible integration into multifunctional biosensing systems.

Leveraging these advantages, HCR has been widely employed in the construction of detection platforms for foodborne pathogens. Gao et al. reported a dual-mode biosensor integrating the CRISPR/Cas12a system with enzyme-free isothermal amplification techniques (Gao et al., 2025a). This sensor responds to aptamer recognition events through dual amplification mechanisms involving HCR and catalyzed hairpin assembly (CHA), enabling highly sensitive detection of S. aureus. Li et al. developed an HCR amplification strategy activated by extracellular adenosine triphosphate (ATP) (Li et al., 2023). This approach combines aptamer-specific recognition with ATP-triggered HCR signal amplification, achieving sensitive quantitative detection of as few as 25 target cells in a 200 μL reaction system. Despite these advances, several limitations persist across different platforms. Aptamer-based fluorescence, colorimetric, and electrochemical methods without amplification still rely on complex multiple-step procedures, suffering from matrix interference, or low sensitivity that limit their practical application potential (Crivianu-Gaita & Thompson, 2016; Kang & Zhao, 2025; Li et al., 2026). Furthermore, while CRISPR-based methods offer high specificity, they often require additional Cas protein expression and guide RNA design, adding to system complexity and cost (Zhao et al., 2026). Similarly, LAMP-based methods demand precise thermal cycling (or sophisticated isothermal instruments) and are prone to non-specific amplification, necessitating meticulous primer optimization (Yang et al., 2026). Immunoassays, as noted before, are constrained by antibody production time, batch variability, and cross-reactivity in complex food matrices (Xiao et al., 2025). Even among HCR-based sensors, most operate solely in homogeneous solution, lacking effective target enrichment and separation, which limits their performance in real samples (Bi et al., 2017; Xia et al., 2023).

In contrast to these existing strategies, this study proposed a FRET-based fluorescent aptasensor that uniquely integrates aptamer-functionalized magnetic beads (MBs) with HCR amplification for rapid S. aureus detection in milk. Unlike antibody-dependent immunoassays, our method employs chemically synthesized aptamers, ensuring cost-effectiveness and batch-to-batch consistency. Different from CRISPR and LAMP approaches, our platform operates under isothermal, enzyme-free conditions without requiring thermal cyclers or additional protein components. More importantly, unlike previous HCR sensors that merely rely on signal amplification in solution, our design incorporates MB-mediated bacterial enrichment and target-triggered release of the HCR initiator, followed by FRET-based fluorescence readout, which collectively enhances sensitivity and effectively suppresses matrix interference. This work thus provides valuable insights for biosensors design and holds significant potential for applications in food safety and environmental monitoring.

2. Materials and methods

2.1. Materials

All DNA sequences (Table S1) used in this work were offered by Sangon Biotechnology Co., Ltd. (Shanghai, China) and dissolved with phosphate buffer. The DNAs stored at −20 °C were heated at 100 °C for 3 min and cooled to room temperature gradually before use. Sodium chloride (NaCl) and magnesium chloride (MgCl₂) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Ethylenediaminetetraacetic acid (EDTA), potassium chloride (KCl) and sodium hydroxide (NaOH) were obtained from Tianjin Fengchuan Chemical Reagent Co., Ltd. (Tianjin, China). Anhydrous ethanol (C₂H₅OH) and hydrochloric acid (HCl) were provided by Tianjin Beilian Fine Chemical Development Co., Ltd. (Tianjin, China). Ammonium sulfate ((NH₄)₂SO₄), agar and glucose (C₆H₁₂O₆) were offered by Fuchen Chemical Reagent Co., Ltd. (Tianjin, China). Yeast extract powder, pancreatic tryptone and beef extract were obtained from Guangdong Huankai Microbial Sci&Tech Co., Ltd. (Guangdong, China). Tween-20 was purchased from Aladdin Industrial Co., Ltd. (Shanghai, China). Phosphate-buffered saline (PBS) was obtained from Thermo Fisher Scientific Biochemical Products Co., Ltd. (Beijing, China). Tris-HCl buffer and epoxy magnetic beads (MBs) were obtained from Beijing Vokai Biotechnology Co., Ltd. (Beijing, China) and Biomag Biotechnology Co., Ltd. (Jiangsu, China), respectively. Ultrapure deionized water (18.2 MΩ.cm) was used throughout.

2.2. Bacteria culture

The bacteria strains were inoculated by streaking onto solid culture medium plates and incubated at 37 °C for 24 h. A single colony was subsequently picked using an inoculation loop and transferred into 5 mL of sterile liquid medium contained in a 10 mL centrifuge tube. The culture was incubated at 37 °C with continuous shaking at 200 rpm for 12 h. For expansion, an aliquot of 200 μL of this pre-culture was inoculated into 20 mL of fresh liquid medium and further incubated under the same conditions (37 °C, 200 rpm) for 3.5 h.

2.3. Preprocessing of MBs

500 μL MBs solution (20 mg/mL) was sequentially washed three times with 500 μL coupling buffer (0.1 M PBS, 1 mM EDTA, pH 7.4) and washing buffer (containing 0.1% Tween-20, pH 7.4). The supernatant was removed via magnetic separation. Following the washing procedure, the MBs were resuspended in 500 μL coupling buffer, transferred to a low-retention centrifuge tube, and stored at 4 °C for further use.

2.4. Preparation of MBs-Apt complex

The MBs-Apt complex was prepared through covalently connecting epoxy MBs with amino-modified aptamer. Briefly, 1.5 μL, 2 μL, and 4.5 μL aptamer solution (20 μM) was added to 150 μL pre-treated MBs suspension (20 mg/mL) above, respectively. The mixture was then incubated at 25 °C and 250 rpm for 12 h, with continuous agitation to maintain uniform dispersion of the MBs. After incubation, magnetic separation was performed to remove the supernatant, followed by washing three times with washing buffer to eliminate nonspecific adsorption. Finally, the MBs-Apt solution was resuspended in 150 μL PBS (pH 7.4) to obtain the different concentrations (200 nM, 300 nM and 600 nM) of MBs-Apt complex and stored at 4 °C for further use.

2.5. Construction of MBs-Apt-Tri aptasensor

1.4 μL trigger solution (20 μM) was added to the prepared MBs-Apt solution (150 μL). After thorough mixing, the mixture was incubated at 25 °C with agitation at 250 rpm for 2 h. Following incubation, the mixture was subjected to magnetic separation, after which the complexes were washed three times with washing buffer to remove unbound components. Finally, the synthesized MBs-Apt-Tri aptasensor was resuspended in 150 μL PBS buffer (pH 7.4) containing 5 mM Mg²⁺.

2.6. Optimization of experimental conditions

2.6.1. Capture time of MBs-Apt complex toward S. aureus

For capture time optimization, 50 μL MBs-Apt (200 nM) was mixed with 50 μL S. aureus standard solutions (10² CFU/mL) and resuspended in 200 μL PBS (pH 7.4). The mixture was incubated at 25 °C with shaking at 250 rpm for 15, 30, 45, and 60 min, respectively. After incubation, magnetic separation was performed and the supernatant was removed. The pellet was resuspended in 1 mL PBS (pH 7.4) to obtain the MBs-Apt@SA complex. The suspension was spread evenly on agar plates, which, after surface drying, were inverted and incubated at 37 °C for 24 h.

2.6.2. Concentration of MBs-Apt complex

50 μL MBs-Apt (200 nM) and 50 μL of MBs-Apt (300 nM) were separately mixed with 50 μL 10² CFU/mL S. aureus suspension, and then diluted with 200 μL PBS (pH 7.4) and incubated at 25 °C with shaking at 250 rpm for 30 min, respectively. After incubation, magnetic separation was performed, the supernatant was removed, and the pellet was resuspended in 1 mL PBS to obtain the corresponding MBs-Apt@SA complexes. The suspensions were spread evenly on agar plates. After drying, the plates were inverted and incubated at 37 °C for 24 h. The capture efficiency was calculated based on colony counts.

To further assess the performance of MBs-Apt under higher bacterial loads, 50 μL MBs-Apt (300 nM) and 50 μL MBs-Apt (600 nM) were separately mixed with 50 μL 10² CFU/mL S. aureus suspension, followed by the addition of 200 μL PBS. The mixtures were incubated under the same conditions (25 °C, 250 rpm, 30 min). After magnetic separation and removal of the supernatant, the pellets were resuspended in 1 mL PBS and plated on agar plates. Following 24 h incubation at 37 °C, the colonies were enumerated to determine the capture efficiency.

2.6.3. Amount of MBs-Apt-Tri

10, 20, 30, and 40 μL MBs-Apt-Tri (600 nM) aptasensors were separately mixed with 30 μL S. aureus (10⁴ CFU/mL), and then adjusted to a final volume of 200 μL using PBS (pH 7.4, 5 mM Mg²⁺). After incubation at 25 °C with shaking at 250 rpm for 30 min, magnetic separation was performed, and the supernatants were collected. To initiate the signal amplification reaction, 2 μL HP1 (20 μM) and 4 μL HP2 (20 μM) were sequentially added to each supernatant, followed by incubation in a 37 °C water bath for 1 h. A negative control was prepared by adding equivalent amounts of HP1 and HP2 into 200 μL PBS (pH 7.4, 5 mM Mg²⁺), which was then incubated under identical conditions.

After incubation, 100 μL each reaction solution was taken, and the fluorescence intensity was measured at an excitation wavelength of 480 nm and an emission wavelength of 518 nm.

2.6.4. Ratio of HP1/HP2

30 μL MBs-Apt-Tri (600 nM) was mixed with 30 μL S. aureus suspension (10⁴ CFU/mL). The mixture was then resuspended in 140 μL PBS (pH 7.4, 5 mM Mg²⁺). The resulting solution was incubated at 25 °C with shaking at 250 rpm for 30 min. After incubation, magnetic separation was performed, and the supernatant was collected. To different experimental groups, 2 μL HP1 (20 μM) and different volumes of HP2 (20 μM, 2, 3, 4, 5, 6, 7, and 8 μL, specifically) were added separately. All samples were subsequently incubated in a 37 °C water bath for 1 h. Following the reaction, 100 μL supernatant was taken, and the fluorescence intensity was measured at an excitation wavelength of 480 nm and an emission wavelength of 518 nm.

2.6.5. HCR reaction time

30 μL MBs-Apt-Tri (600 nM) was mixed with 30 μL S. aureus (10¹ CFU/mL or 10² CFU/mL) suspension. The mixture was resuspended with 140 μL PBS (pH 7.4, 5 mM Mg²⁺) and incubated at 25 °C in with shaking at 250 rpm for 30 min. After incubation, magnetic separation was performed, and the supernatant was collected. Then, 2 μL HP1 (20 μM) and 4 μL HP2 (20 μM) were added. The samples were subsequently reacted in a 37 °C water bath for three different time periods: 30 min, 45 min, and 60 min. Following reaction, 100 μL supernatant was taken, and the fluorescence intensity was measured at an excitation wavelength of 480 nm and an emission wavelength of 518 nm.

2.7. Detection of S. aureus

S. aureus standard solutions were prepared by serial dilution method. Briefly, sterile ultrapure water (900 μL) was aliquoted into sterilized low-adsorption microcentrifuge tubes (1.5 mL). Following homogenization, 100 μL amplified S. aureus solution was transferred into the first tube and mixed thoroughly. This process was repeated sequentially to yield standard samples at final concentrations of 10¹, 10², 10³, 10⁴, and 10⁵ CFU/mL.

Subsequently, different concentrations of S. aureus standard solutions (30 μL) were mixed with 30 μL MBs-Apt-Tri and 140 μL sterile PBS (pH 7.4, 5 mM Mg²⁺). Meanwhile, a control sample was prepared under identical conditions by substituting the bacterial solution with PBS. All samples were incubated at 37 °C with shaking at 250 rpm for 30 min. After incubation, magnetic separation was performed, and the supernatant was collected and supplemented with 2 μL HP1 (20 μM) and 4 μL HP2 (20 μM), followed by incubation in a 37 °C water bath for 45 min. After the reaction, 100 μL mixture was taken for fluorescence measurement at an excitation wavelength of 480 nm and an emission wavelength of 518 nm.

2.8. Specificity analysis

To evaluate the specificity of the detection method, four non-target bacterial strains were selected as controls, including methicillin-resistant Staphylococcus aureus (MRSA), Escherichia coli O157:H7 (E. coli O157:H7), Escherichia coli 8099 (E. coli 8099), and Salmonella Typhimurium (S. typhimurium). 1 mL fresh bacterial solutions were added to 2 mL sterile microcentrifuge tubes and diluted with 1 × PBS to a uniform concentration of 10⁶ CFU/mL. Each sample was then processed according to the detection procedure. After the reaction, 100 μL supernatant was collected, and fluorescence intensity was measured (ex: 480 nm, em: 518 nm).

2.9. Actual milk sample detection

To assess the applicability and accuracy of the detection method in real samples, commercially available UHT sterilized cow milk was used as the matrix. Simulated contaminated samples were prepared by spiking the milk with known concentrations of S. aureus. Prior to the experiment, the milk was exposed to ultraviolet light for 30 min to maintain sterile operating conditions. The initial concentration of S. aureus was determined by plate counting, followed by serial dilution with sterile milk to obtain spiked milk samples with bacterial concentrations of 1.0 × 10², 1.0 × 10³, and 1.0 × 10⁴ CFU/mL. Subsequently, 50 μL spiked sample was analyzed according to the detection procedure. A negative control was prepared using 50 μL sterile milk in place of the bacterial sample. Finally, the fluorescence intensity of each sample was measured under the corresponding fluorescence conditions.

2.10. Native polyacrylamide gel electrophoresis (PAGE) characterization

Different structures of DNA samples were characterized with 12% polyacrylamide gels within 5 × TBE buffer (pH 8.0, 1 mM EDTA, 45 mM Tris-Borate). Each sample (5 μL) was mixed with 1 μL 6× loading buffer and then loaded. The potential of all gels was 120 V, which were run for 40 min, and stained with SYBR Gold (1 μL) for 15 min and imaged in UVP Solo Elite lmaging System (Analytik Jena AG, USA).

3. Results and discussion

3.1. Principle of S. aureus detection

This study developed a highly sensitive and specific detection method for S. aureus based on aptamer-functionalized magnetic beads and fluorescence-labeled hybridization chain reaction. The working principle of this method was illustrated in Scheme 1. 5′-amino-modified aptamer (Apt) was firstly covalently immobilized onto epoxy-functionalized MBs via ring-opening reaction to form a MBs-Apt complex. Additionally, the specifically designed trigger strand (Tri) was then introduced and hybridized with the immobilized aptamer to construct the MBs-Apt-Tri aptasensor. During target recognition, the aptasensor was incubated with S. aureus. Upon specific binding of the aptamer to S. aureus, the conformation of aptasensor changed, resulting in the release of the pre-hybridized trigger strand into the supernatant. This step converts the presence of the target into a detectable free nucleic acid signal. After magnetic separation, the supernatant containing the released Tri was collected. Based on the principle of fluorescence resonance energy transfer (FRET), two molecular beacon-based hairpins (HP1 and HP2) were added to the system. HP1 was labeled with FAM near 5′ end and BHQ-1 near 3′ end. The released Tri initiated a cascaded HCR, leading to the self-assembly of the hairpin into an extended fluorescent double-stranded polymer, thereby achieving signal amplification. Finally, quantitative analysis of the fluorescence signal enables highly sensitive detection of the target bacteria.

Scheme 1.

Schematic illustration of the fluorescent aptasensor construction and HCR signal amplification principle.

To validate the proposed sensing mechanism, we employed agarose gel electrophoresis to monitor the stepwise nucleic acid assembly processes. As shown in Fig. S1, the free Apt (lane 2) and Tri strand (lane 3) each exhibited distinct bands at their expected positions. Upon mixing the aptamer with the trigger strand, a retarded band corresponding to the Apt-Tri duplex appeared (lane 4), confirming successful hybridization between the two sequences. Subsequently, the target-triggered release of the Tri strand was verified by incubating the Apt-Tri complex with S. aureus (10⁵ CFU/mL) for 30 min. The released Tri strand (lane 10), was more apparent compared to that in the control sample without S. aureus (lane 4). This result directly demonstrates that the binding of S. aureus to Apt induces conformational changes leading to the specific release of the pre-hybridized Tri strand. The formation of HCR products was further characterized by incubating the released trigger strand with HP1 and HP2. In the presence of Tri strand, a high-molecular-weight smear characteristic of HCR polymers was observed (lane 8), whereas no such smear was detected in the absence of Tri strand (lane 11, lane 12) or in the presence of a mismatched trigger sequence (lane 15, lane 16). Furthermore, the omission of HP2 resulted in low-molecular-weight band (lane 13, lane 14), confirming that both hairpins are required for the formation of extended HCR products. These results collectively confirm that the HCR amplification proceeds as designed.

In addition to gel electrophoresis, control experiments were performed to exclude the possibility that the fluorescence signal arises from direct opening of HP1. As shown in Fig. S2, the fluorescence signal was negligible in the absence of the Tri strand, confirming that HP1 cannot be directly opened without the initiator. The control without HP2 showed no fluorescence recovery, confirming that the FRET signal is dependent on the cascade amplification process. Finally, a comparison between systems with and without HCR amplification revealed that the HCR-amplified system produced a fluorescence signal more than 5-fold higher than the non-amplified control, demonstrating the critical role of HCR in signal enhancement.

3.2. Characterization of nucleic acid sequences by NUPACK simulation

We performed secondary structure simulations of the HP1, HP2 and HCR amplification product using NUPACK software. The measured free energies for the two hairpins were −15.65 kcal/mol and −17.44 kcal/mol, respectively (Fig. S3). This thermodynamic profile ensures metastability, avoiding undesired unfolding or hybridization without the Tri. Concurrently, it provides sufficient lability for the Tri to efficiently open the hairpins and initiate the cascade amplification reaction via specific base-pairing. The cascade amplification product possessed a markedly low free energy further of −56.63 kcal/mol, indicating a conformation of high stability. This significant energy decrease is thus sufficient to thermodynamically drive the reaction forward and energetically realize the signal amplification process. This free-energy gradient design-from hairpins to amplification product-allows the detection system to achieve low background noise and high target responsiveness, thereby offering a theoretical basis for developing highly sensitive and specific detection methods. Furthermore, the design incorporates a FRET-based molecular beacon within the HP1 structure, such that precise free-energy modulation facilitates efficient fluorescence switching. As shown in Fig. S4, the HP1 structure, labeled with FAM and BHQ-1, forms a stable stem-loop with a free energy of −26.37 kcal/mol, positioning the fluorophore and quencher for efficient FRET-based quenching. The presence of Tri, which binds specifically to HP1, triggers a structural transition of HP1 to a linear conformation (free energy: 0.00 kcal/mol). This conformational change separates the fluorophore from the quencher, thereby leading to the recovery of fluorescence. These simulation results demonstrate the feasibility of a detection method based on the fluorescence signal generated by the HCR amplification-induced conversion of the HP1 structure.

3.3. Optimization of experimental conditions

To obtain higher detection efficiency, experimental conditions, including the capture time of the MBs-Apt complex toward S. aureus, the concentration of MBs-Apt complex, the amount of MBs-Apt-Tri, ratio of HP1/HP2, and the HCR reaction time were optimized.

The capture time critically influences the sensitivity of the MBs-Apt complex. Inadequate capture time leads to incomplete target binding, thereby reducing capture efficiency and limiting the linear detection range. Conversely, prolonged incubation time contradicts the requirement for rapid detection and may affect bacterial viability, ultimately diminishing capture performance. Therefore, the capture time of MBs-Apt toward S. aureus was systematically optimized in this study. For 10² CFU/mL S. aureus, extending the capture time from 15 to 30 min significantly increased the number of colonies captured on the MBs-Apt. In contrast, further extension to 45 and 60 min led to a decrease in capture. As shown in Fig. 1A, the supernatant contained the fewest residual colonies (Y2) and the MBs-Apt carried the most captured colonies (Y1) at 30 min, suggesting an optimal binding equilibrium between the S. aureus and MBs-Apt. Consistent with this, the capture efficiency (captured colonies/total colonies×100%) was highest at 30 min (approximately 40%, Fig. 1B), which was significantly superior to other capture time. Despite the moderate capture yield, overall performance is not compromised. The fluorescence readout does not depend directly on the absolute capture amount but rather on the trigger strands released from the captured bacteria and subsequently amplified by HCR; this exponential amplification adequately offsets the modest yield. Moreover, 30 min optimally balances binding efficiency, detection speed, and bacterial viability, as longer incubation reduces capture efficiency, likely due to aggregation or desorption of the bacteria. Based on these results, 30 min was selected as the optimal capture time.

Fig. 1.

Capture effect of MBs-Apt at different incubation time. (A) Colonies present in magnetic fraction (Y1) and supernatant (Y2) after incubating MBs-Apt with 10² CFU/mL S. aureus for different time. (B) Capture rate of MBs-Apt toward S. aureus at different incubation time.

The detection performance is highly dependent on the MBs-Apt concentration. An insufficient concentration reduces sensitivity due to limit available binding sites, while an excessively high concentration may lead to the aggregation of MBs-Apt, thereby decreasing the recognition efficiency. Consequently, systematic optimization of MBs-Apt concentration is critical for attaining optimal capture efficiency. As shown in Fig. 2A, for S. aureus at 10² CFU/mL, increasing the MBs-Apt concentration from 200 nM to 300 nM improved capture efficiency from below 80% to over 90%. Subsequently, at a higher bacterial load of 10⁴ CFU/mL (Fig. 2B), the capture efficiency of 300 nM MBs-Apt dropped to 40%, whereas 600 nM MBs-Apt restored the efficiency to approximately 90%. These results indicate that capture efficiency is a function of both MBs-Apt concentration and target bacteria load. Efficient capture at a lower bacterial load (10² CFU/mL) was achieved with 300 nM MBs-Apt, whereas the higher bacterial load (10⁴ CFU/mL) required 600 nM MBs-Apt to attain a comparable efficiency level. To further characterize the immobilization efficiency and loading capacity at the selected concentration, varying amounts of epoxy-functionalized MBs (0.2, 2, 3, 5, and 6 mg) were incubated with 600 nM aptamer in a 150 μL reaction system. After magnetic separation, the supernatant absorbance at 260 nm was measured to determine the residual free aptamer (Fig. S5). The initial absorbance of the 600 nM aptamer solution was 0.10458. Upon incubation with increasing MBs amounts, the residual absorbance gradually decreased, reaching 0.05271 at 3 mg (the amount used throughout this study), indicating that approximately 45.36 pmol of aptamer remained unbound, corresponding to an immobilization efficiency of approximately 49.6%. When the MBs amount reached 5 mg, the supernatant absorbance further decreased to 0.0379, corresponding to an immobilization efficiency of approximately 63.8% and a loading capacity of 11.48 pmol/mg. Further increasing the MBs amount to 6 mg resulted in negligible additional reduction in absorbance, suggesting that aptamer binding sites on the MB surface approached saturation. Notably, while 5 mg of MBs yielded higher immobilization efficiency, 3 mg was selected for subsequent experiments as it provided sufficient aptamer loading for effective bacterial capture while maintaining economical material consumption. Increasing the amount of MBs-aptamer conjugates could also be a viable strategy to improve detection sensitivity, as higher loading may enhance the capture efficiency of target bacteria at low concentrations.

Fig. 2.

Capture effect of MBs-Apt at different concentration. (A-B) Colony images of magnetic fraction and supernatant for different concentrations of MBs-Apt against S. aureus at 10² CFU/mL (A) and 10⁴ CFU/mL (B).

The capture reproducibility of the MBs-Apt complex was also evaluated under the optimized conditions (600 nM MBs-Apt, 3 mg MBs, 10⁴ CFU/mL S. aureus). After magnetic separation, the supernatant and resuspended precipitate were separately plated for colony counting. The captured bacteria were predominantly recovered from the precipitate fraction, while only a small number of residual colonies were detected in the supernatant. The capture efficiencies for three independent replicates were determined to be 99.71%, 99.95%, and 99.66%, respectively, confirming effective target capture and good reproducibility of the capture step (Fig. S6). To realize maximum capture efficiency and minimum nonspecific binding, 600 nM was selected as the optimal working concentration for MBs-Apt complex. This condition enables stable and efficient capture across a broad range of target concentrations, providing a reliable foundation for subsequent detection process.

Additionally, the amount of the MBs-Apt-Tri aptasensor was systematically optimized. The results manifested that the fluorescence intensity increased continuously with increasing amounts of MBs-Apt-Tri, reaching a maximum at 30 μL. Conversely, add in beyond this volume resulted in a decline of fluorescence signal (Fig. S7). To further exclude the possibility of background interference arising from the spontaneous hybridization of HP1 and HP2, a trigger-free MBs-Apt control group was included. The MBs-Apt complex (without the trigger strand) was incubated under the same conditions as the MBs-Apt-Tri aptasensor, and the supernatant was collected after magnetic separation and subsequently reacted with HP1 and HP2. The resulting fluorescence intensity was low, comparable to that of the blank control (Fig. S8). In parallel, gel electrophoresis analysis (Fig. S1) showed no high-molecular-weight HCR products in the trigger-free group (lanes 7, 11 and 12), indicating that HP1 and HP2 remained predominantly in their hairpin conformations without the initiator. These results confirm that spontaneous hybridization between HP1 and HP2 is negligible under the experimental conditions and that the observed fluorescence signal is specifically attributed to the trigger strand released upon target recognition. Therefore, 30 μL was identified as the optimal dosage of MBs-Apt-Tri for subsequent experiments.

To further enhance the sensitivity, the ratio of HP1/HP2 was systematically optimized. As shown in Fig. 3A, the fluorescence signal increased along with the higher HP2 amounts. Increasing the HP1/HP2 ratio from 1:1 to 1:2 produced a significant enhancement in fluorescence intensity, while further increases HP1/HP2 ratio from 1:2.5 to 1:4, the fluorescence signal showed only a slightly enhancement, suggesting a saturating response. Thus, the optimal ratio of HP1/HP2 was established as 1:2 following a comprehensive evaluation of signal response and experimental cost, which provides high sensitivity and cost-effectiveness for reliable use in following experiments.

Fig. 3.

Fluorescence intensity produced by HCR amplification. (A-B) Fluorescence produced by HCR amplification under different HP1/HP2 ratio for 10⁴ CFU/mL S. aureus (A) and different HCR reaction time (B) for 10¹ CFU/mL and 10² CFU/mL S. aureus.

The HCR reaction time was systematically optimized to overcome the constraints of incomplete hybridization and weak signal output, improving the performance of the detection system. To determine the optimal incubation time, S. aureus at concentrations of 10¹ CFU/mL and 10² CFU/mL was used as the model target. The fluorescence signals under different HCR reaction time were compared to identify the best condition (Fig. 3B). After 45 min of reaction, the low bacterial counts can be sufficient to distinguish through fluorescence signal. In contrast to 30-min incubation, the signal intensity was significantly enhanced at 45 min, whereas it manifested comparable intensity observed at 60 min. Therefore, considering both sensitivity and time efficiency, the optimal HCR incubation time was determined to be 45 min.

3.4. Detection of S. aureus

Under the optimized conditions established above, the sensitivity of this method for S. aureus was evaluated. As shown in Fig. 4A, a concentration-dependent increase in fluorescence intensity was observed as the concentration of S. aureus increased from 10⁰ to 10⁵ CFU/mL. A good linear relationship was observed between the logarithm of bacterial concentration (x) and the fluorescence signal (y), fitting the regression equation y = 20.58× + 63.57 with a correlation coefficient of R² = 0.9895. Limit of detection (LOD) was calculated to be 5.3 × 10¹ CFU/mL, with a signal-to-noise of 3 (S/N = 3). Moreover, the aptasensor exhibits better detection performance compared to previously reported methods toward S. aureus detection (Table 1). To evaluate the practical applicability of the proposed aptasensor, the batch-to-batch reproducibility and storage stability were systematically investigated. Three independently prepared batches of the MBs-Apt-Tri aptasensor were tested against S. aureus at 10⁴ CFU/mL. The fluorescence responses from the three batches showed good consistency, with an RSD of 2.12% (n = 3) (Fig. S9), indicating that the fabrication protocol yields reproducible sensing performance.

Fig. 4.

Linear range and specificity of the aptasensor. (A) Linear responses of the aptasensor toward S. aureus from 10¹ CFU/mL to 10⁵ CFU/mL. (B) Responses of aptasensor toward S. aureus or other bacterial strains, E. coli 8099, E. coli O157:H7, MRSA, and S. Typhimurium.

Table 1.

Comparison of different methods with various analytical procedures for detection of S. aureus.

Method	Linear Range (CFU/mL)	LOD (CFU/mL)	Reference
microfluidic chip	102–104	63	(Quan et al., 2024)
Self-powered biosensing	2 × 103–2 × 106	2 × 103	(Wang et al., 2022a)
Cas14a1-aptamer biosensor	4 × 104–4 × 107	400	(Wei et al., 2022)
immunoassay	1.0 × 104-1.0 × 106	2.47 × 103	(Yan et al., 2017)
long-period fiber gratings	104–108	224	(Yang et al., 2019)
lateral flow biosensor	–	5.4 × 102	(Zhou et al., 2022)
HCR-based aptasensor	100–105	5.3 × 101	This work

The storage stability of the MBs-Apt probe was evaluated by storing it at 4 °C for 0, 3, 5, and 7 days, followed by magnetic separation and UV–vis analysis of the supernatant. As shown in Fig. S10, no significant aptamer absorbance was observed at any time point, confirming that the aptamer remained firmly attached to the MBs and that the probe could be prepared in advance. When the complete MBs-Apt-Tri aptasensor was stored under identical conditions and then used to detect S. aureus at 10⁴ CFU/mL, the fluorescence intensity progressively declined, retaining only 71%, 64%, and 39% of the initial signal after 3, 5, and 7 days, respectively (Fig. S11). This gradual loss likely results from degradation or conformational changes of the immobilized Apt-Tri during storage. These findings suggest that while MBs-Apt alone is stable enough for advance preparation, the complete aptasensor should be better freshly prepared to maintain optimal sensitivity or used it within 24 h to minimize signal loss.

3.5. Specificity analysis

To evaluate the specificity of the sensing method, the fluorescence intensity was recorded under the presence of different kinds of bacterial strains, such as S. aureus, E. coli 8099, E. coli O157:H7, MRSA, and S. typhimurium at a concentration of 10⁶ CFU/mL, meanwhile, PBS sample was set as a negative control. Notably, although MRSA is a subtype of S. aureus, it was included as an interfering strain because the aptamer used in this work was selected against a specific methicillin-sensitive S. aureus strain, exhibiting strong affinity toward this target but negligible cross-reactivity with MRSA. This contrasts with the aptamer reported by Zhang et al., which was specifically selected for MRSA (Zhang et al., 2026). As shown in Fig. 4B, for 10⁶ CFU/mL bacterial strains, the fluorescence signals from all interfering bacterial strains (including MRSA) were comparable to the negative control, whereas the target S. aureus elicited a markedly strong response. This differential recognition confirms that our aptamer distinguishes the target strain from MRSA, and thus the sensor is intended for the detection of specific S. aureus strains rather than all strains within the species. Collectively, these results demonstrate high specificity and selectivity for the target S. aureus strain, effectively differentiating it from non-target bacteria. To more rigorously evaluate the selectivity of the assay, plate capture experiments were performed using the MBs-Apt complex against a panel of bacterial strains. The capture efficiency was determined by plate counting of the supernatant after magnetic separation. As shown in Fig. S12, the capture efficiency of MBs-Apt toward S. aureus was approximately over 90%, which was significantly higher than that toward other tested strains (all below 10%), consistent with the fluorescence specificity results presented in Fig. 4B. These results demonstrate that the method possesses high specificity and selectivity for S. aureus, effectively distinguishing it from non-target bacteria.

3.6. Detection of S. aureus in real samples

To validate the feasibility and accuracy of the proposed detection method in real samples, commercially available UHT sterilized cow milk was used as the sample matrix. Contaminated milk samples were prepared by spiking known quantities of S. aureus, followed by analysis using the developed method. The results are summarized in Table 2, the spiked recovery rates for S. aureus in milk ranged from 94% to 110%, with relative standard deviations (RSDs) values between 1.07% and 3.89%. Although UHT milk contains proteins and fat similar to raw milk, satisfactory recoveries (94%–110%, RSDs, 1.07%–3.89%) were achieved. This robustness can be attributed to the high amplification efficiency of the HCR system. According to Scheme 1, after magnetic separation, the supernatant containing the released trigger strand is collected for HCR. Trace amounts of residual proteins or fat may be co-transferred, yet the cascade HCR provides exponential signal amplification that effectively suppresses any background interference, ensuring a high signal-to-noise ratio.

Table 2.

Recovery rate of S. aureus in milk samples (n = 3).

Sample	Added (CFU/mL)	Detected (CFU/mL)	Recovery (%)	RSDs (%)
1#	1.0 × 102	1.10 × 102	110	3.89
2#	1.0 × 103	0.94 × 103	94	1.07
3#	1.0 × 104	1.06 × 104	106	1.56

RSDs: relative standard deviations.

In addition to the satisfactory accuracy, our method offers distinct practical advantages: the total assay time is approximately 15.5 h, substantially shorter than the 24–48 h required for the standard plate counting; and the sensitivity surpasses that of most qPCR assays (typically 10²–10³ CFU/mL), while avoiding thermal cyclers and nucleic acid extraction.

Nevertheless, we acknowledge that the current validation is conducted in sterilized milk. While the results are promising, the indigenous microflora and more intricate colloidal structures in naturally contaminated raw milk may introduce additional variables. Therefore, future work will focus on evaluating the method using naturally contaminated raw milk samples and cross-validating the results with standard plate counting or qPCR to further confirm its reliability in unprocessed food matrices.

Despite the satisfactory analytical performance demonstrated above, several limitations of this study should be acknowledged. The aptamer used herein was selected against S. aureus cells and may exhibit strain-dependent binding affinities due to variations in surface epitope expression; further evaluation using a broader panel of S. aureus strains is warranted. Additionally, because the aptamer recognizes surface components regardless of cellular metabolic activity, the method may not inherently distinguish viable from nonviable bacteria. However, for the intended application in sterilized dairy products, where S. aureus is strictly prohibited by Chinese national standards, this limitation does not compromise the screening utility, as the presence of either viable or nonviable S. aureus indicates potential contamination. The current validation was performed only in UHT milk; naturally contaminated milk samples (e.g., raw milk) should be tested in future studies to further confirm practical reliability. Addressing these limitations will further strengthen the method's applicability for routine food safety monitoring.

4. Conclusion

In summary, a fluorescent aptasensor integrating aptamer-functionalized magnetic beads with HCR signal amplification was successfully developed for the rapid detection of S. aureus. Notably, by combining efficient magnetic enrichment with enzyme-free isothermal amplification, this strategy circumvents the limitations of traditional methods that often require complex instrumentation or prolonged analysis time. Under optimal conditions, the method exhibited excellent specificity, enabling reliable discrimination against non-target bacteria even at concentrations as high as 10⁶ CFU/mL. Collectively, this work presents a rapid, sensitive, and highly specific detection strategy for S. aureus, highlighting its distinct advantages over conventional approaches and offering a promising platform for the analysis of diverse targets in complex food matrices and environmental samples.

CRediT authorship contribution statement

Xiaoli Wu: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Qun Zhao: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Yingying Wang: Writing – review & editing, Investigation, Formal analysis, Data curation. Lijun Liu: Writing – review & editing, Formal analysis. Alideertu Dong: Funding acquisition, Formal analysis, Data curation. Cong Wang: Writing – review & editing, Formal analysis. Yuchun Guo: Writing – review & editing, Supervision, Formal analysis, Conceptualization. Haixia Wu: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.