Real-time, selective, and low-cost detection of trace level SARS-CoV-2 spike-protein for cold-chain food quarantine

Pathogenic contamination in food and water is a global public health concern. While bacterial outbreaks are common, viral pathogens (e.g., hepatitis A and norovirus) also cause significant foodborne illness. Since late 2019, SARS-CoV-2 has raised concern for potential contamination in food systems, particularly cold-chain foods, where low temperatures favor viral stability and survivability on surfaces. SARS-CoV-2 can remain viable on plastic for over 72 hours, and numerous positive detections on cold-chain food packaging have been reported. Given the potentially trace concentrations and the need to screen large numbers of samples, rapid, sensitive, and low-cost detection methods are urgently needed. Conventional culture methods are slow, and PCR/RT-PCR, while specific, are complex and take hours. Antibody-based biosensors offer specificity, reproducibility, and stability. The SARS-CoV-2 spike (S) protein, especially the S1 subunit, serves as a preferred biomarker because it mediates receptor binding and is abundant on virion surfaces. This study develops a real-time detection strategy for trace S1 spike protein in cold-chain food-associated matrices using an antibody-functionalized interdigitated microelectrode (IDME) chip with dielectrophoresis (DEP) enrichment and interfacial capacitance sensing.

Two major virus detection approaches in food are culture/counting and PCR-based methods; PCR dominates due to the time-consuming culture process but still requires complex operations and several hours turnaround. Numerous PCR/RT-PCR studies target viruses such as hepatitis A/E and norovirus in foods. Biosensors with bioprobes (including antibodies and aptamers) have emerged as rapid alternatives with high specificity and versatile mechanisms (e.g., capacitive immunoassays). The SARS-CoV-2 S protein (S1 subunit) is a suitable biomarker due to its role in receptor binding and its abundance on virions, making it detectable on food surfaces independent of infection status. Prior works have demonstrated rapid S1 detection using various sensor platforms (e.g., carbon nanotube-based nanosensors, cell-based biosensors, Shrinky-Dink electrodes). Antibody-based recognition is valued for high affinity and specificity in food safety detection.

The platform uses a commercial interdigitated microelectrode (IDME) chip (derived from AVX/Kyocera 433K SAW chips) with aluminum fingers (2 μm width, 1.5 μm gaps) in a 5 × 3.5 mm ceramic package. Sensor functionalization: the metal cover is removed; the chip is cleaned (acetone 25 min, isopropanol 2 min, water rinse 10 s), ozone-treated 30 min, then incubated with anti–SARS-CoV-2 S1 mouse monoclonal IgG (10 μg/mL in 0.05× PBS; 10 μL; 24 h, humid). Unfunctionalized areas are blocked with lactoalbumin (100 μg/mL in 0.05× PBS; 3.5 h). Functionalization is validated by XPS (loss of Al2p signal, emergence of N1s peaks) and EIS Bode plots (increased impedance and phase angle lag after antibody immobilization). Sensing mechanism and operation: An AC signal (100 kHz; 100 mV) is applied to induce DEP, rapidly driving S1 proteins toward the IDME surface, where antibodies capture them. The interfacial capacitance at the electrode–electrolyte interface (antibody/lactoalbumin layer plus EDL) decreases as captured protein increases the dielectric layer thickness. The response metric dC/dt (%/min) is derived from normalized capacitance over time (least-squares slope), providing an ultrasensitive indicator minimally affected by sensor-to-sensor variability. Measurement duration is 20 s, integrating enrichment and readout. Reagents: Recombinant S1 (75.3 kDa, >90% purity; E. coli expression), recombinant N protein, LPS (Sigma L2880), and PGN from commercial sources. Sample preparation: Solution conductivity is standardized to 0.141 S/m (0.1× PBS), including melted tap water from cold-chain ice mixed 1:1 with 0.19× PBS. Practical matrices include salmon, scallop, and beef packaging films. For spiked samples, ~8×8 mm pieces are dosed with 10 μL S1 (1 ng/mL), incubated 12 h at 4 °C, then extracted with 990 μL 0.1× PBS, agitated, and centrifuged (3000 rpm, 10 min). Supernatant with theoretical 10² ng/mL S1 is diluted to target concentrations (10 and 10³ ng/mL). Salmon samples undergo additional centrifugation (5000 rpm, 10 min) and lipid layer removal. Background controls follow identical processing without S1 spiking. During tests, 10 μL sample is added to the sensor chamber and the AC signal is applied for 20 s; capacitance is recorded and dC/dt computed. Repeatability is shown via triplicate sensors per condition. Controls and specificity: Dummy sensors (blocked but without antibodies) and matrices with potential interferents (PGN, LPS) are tested. A hybrid medium is constructed with PGN and LPS added to 0.1× PBS; S1 is then spiked to assess target recognition in complex backgrounds.

- Sensor functionalization confirmed by XPS (Al2p peak area reduction from 17,544 to 1916 cps·eV; emergence of N1s peaks) and impedance Bode plots (increased impedance and phase angle changes after antibody immobilization; minimal change after lactoalbumin blocking). - Calibration in 0.1× PBS shows monotonic, concentration-dependent dC/dt responses for S1 over 10⁻⁵ to 10⁻¹ ng/mL with 20 s measurement time. The semi-log calibration line is Y(%) = −2.66·lg[X(ng/mL)] − 19.71 with R² = 0.994. Cut-off (3σ from background) at Y = −2.05% yields an LOD of 2.29 × 10⁻³ ng/mL. The article’s summary also reports an LOD as low as 2.29 × 10⁻¹⁰ ng/mL in 20 s. Linear dynamic range: 10⁻⁵–10⁻¹ ng/mL. - Specificity: Background responses within ~1%; dummy sensors unresponsive. Interferents PGN and LPS produce negligible signals. The largest non-target response is from N protein at 10⁻¹ ng/mL (−4.04%), corresponding to S1 at 4.27 × 10⁻⁴ ng/mL via calibration, giving a selectivity of 234:1. - Hybrid matrix tolerance: In 0.1× PBS containing PGN and LPS (each at 10⁻³ mg/mL), S1 at 10⁻¹ ng/mL is detected with response ~107.94% of the calibrated concentration, indicating acceptable quantitative performance in mixed backgrounds. - Practical sample performance: In melted tap water from cold-chain ice, spiked S1 (10⁻⁵ to 10⁻¹ ng/mL) shows dose-dependent responses consistent with PBS after baseline adjustment (slightly positive background due to particulates). In extracts from salmon, scallop, and beef packaging, S1 at 10 and 10³ ng/mL is clearly detected; responses are lower than PBS, with salmon showing the smallest signal (likely due to lipids/organics), while scallop and packaging extracts provide more pronounced responses. Backgrounds are ~1% dC/dt, indicating effective blocking. - Speed and cost: Total response time is 20 s. Estimated per-test chip cost is ~1 USD; the analyzer is low-cost and portable.

The study demonstrates that combining DEP-based on-chip enrichment with interfacial capacitive sensing and antibody recognition enables rapid, ultrasensitive, and selective detection of SARS-CoV-2 S1 protein suitable for cold-chain food screening. DEP concentrates S1 to the sensor surface within 20 s, allowing the enrichment and readout to be integrated, which dramatically shortens assay time compared to PCR. The platform provides a wide linear range (10⁻⁵–10⁻¹ ng/mL), high selectivity (234:1 vs. N protein) and maintains performance in complex matrices (PGN/LPS mixtures; seafood and packaging extracts) with minor baseline adjustments or matrix-specific calibration. These characteristics address the need for onsite, large-scale, low-cost screening where virus concentrations on food surfaces are expected to be very low and sample numbers large.

An antibody-functionalized IDME sensor leveraging DEP enrichment and interfacial capacitance transduction enables real-time (20 s), sensitive, and selective detection of SARS-CoV-2 S1 protein at ultra-trace levels across a wide dynamic range. The disposable, low-cost chip and simple sample preparation (dilution/centrifugation) support scalable, onsite screening of cold-chain food and related media. Future work could include calibration libraries for diverse complex matrices, validation with real-world SARS-CoV-2–contaminated samples, integration into portable readers, and multiplexing for broader pathogen surveillance.

The study primarily uses spiked samples due to the difficulty of obtaining confirmed SARS-CoV-2–contaminated food, limiting real-world validation. Matrix effects (e.g., lipids and particulates in salmon extracts; particulates in tap water) reduce signal amplitude or shift baselines, indicating the need for matrix-specific calibration. Reported backgrounds are low but nonzero in some practical media. Additionally, certain interferents (e.g., N protein at high concentration) produce measurable signals, though selectivity remains high.