Potentiometric Biosensors Based on Molecular-Imprinted Self-Assembled Monolayer Films for Rapid Detection of Influenza A Virus and SARS-CoV-2 Spike Protein.

Rapid, yet accurate and sensitive testing has been shown to be critical in the control of spreading pandemic diseases such as COVID-19. Current methods which are highly sensitive and can differentiate different strains are slow and cannot be conveniently applied at the point of care. Rapid tests, meanwhile, require a high titer and are not sufficiently sensitive to discriminate between strains. Here, we report a rapid and facile potentiometric detection method based on nanoscale, three-dimensional molecular imprints of analytes on a self-assembled monolayer (SAM), which can deliver analyte-specific detection of both whole virions and isolated proteins in microliter amounts of bodily fluids within minutes. The detection substrate with nanoscale inverse surface patterns of analytes formed by a SAM identifies a target analyte by recognizing its surface nano- and molecular structures, which can be monitored by temporal measurement of the change in substrate open-circuit potential. The sensor unambiguously detected and differentiated H1N1 and H3N2 influenza A virions as well as the spike proteins of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and Middle-East respiratory syndrome (MERS) coronavirus in human saliva with limits of detection reaching 200 PFU/mL and 100 pg/mL for the viral particles and spike proteins, respectively. The demonstrated speed and specificity of detection, combined with a low required sample volume, high sensitivity, ease of potentiometric measurement, and simple sample collection and preparation, suggest that the technique can be used as a highly effective point-of-care diagnostic platform for a fast, accurate, and specific detection of various viral pathogens and their variants.

Introduction

The recent pandemic of COVID-19 caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has highlighted the importance of viral detection technologies, which currently are divided among rapid antigen detection[1] with limited sensitivity, the gold standard, reverse transcription-polymerase chain reaction (RT-PCR),[2] with high sensitivity but complicated detection apparatus, and recently clustered regularly interspaced short palindromic repeat (CRISPR)-based detection,[3,4] which is not yet in clinical use. Recent reports also demonstrated the detection of SARS-CoV-2 via surface-enhanced Raman scattering (SERS) using functionalized conductive nanostructures, which is highly sensitive and prompt but requires an elaborate experimental detection.[5−7]

Controversy has arisen as to the best detection strategy,[8] where the debate is centered upon the sensitivity of the method versus the speed of detection. RT-PCR, which is designed to identify specific genes of the viral ribonucleic acid (RNA), is the most sensitive method and can detect mutated versions and different strains. It, however, requires sophisticated equipment available only in testing laboratories and thus cannot be easily deployed at the point of care, adding to the delay of the results.[9,10] During this period, people become increasingly contagious and potentially spread the virus without realizing that they are carriers.[11] For example, the exposure to the SARS-CoV-2 results in a low viral particle titer initially, which, however, increases exponentially within the first three days to 106 particles and peaks around 5 days following the infection.[12] Transmission of the virus nevertheless can occur even from the beginning of the infection, while the patient is asymptomatic and unaware of being a carrier.[13] To address this issue of high transmissibility, rapid tests have been developed as they can provide faster feedback. However, sensitivity is a major drawback since they require high titers, nearly 1000-fold higher, to produce a positive diagnosis[14] while not being able to differentiate different strains or mutations.[15−17] Therefore, they are most useful in stopping the spread from a person already at the height of their infectivity but are not effective in identifying early signs of infection, which is critical for effectively stopping the spread of the disease.

Enveloped RNA viruses are a group of viruses that use RNA as genetic materials and often contain surface spike glycoproteins (i.e., S-proteins). Influenza viruses are a class of RNA viruses having transmission and infection mechanisms similar to those of SARS-CoV-2 and, thus, capable of rendering people infectious before symptoms are present. Particularly, the influenza A virus, such as H1N1 and H3N2, is a globular, single-stranded RNA virus with a diameter of 80–120 nm[18] and has two major S-proteins, hemagglutinin (HA) and neuraminidase (NA), which are responsible for the binding of the virus with sialic acid of respiratory tract cells and the release of viral progeny from infected cells, respectively.[19]

SARS-CoV-2 is also an enveloped, single-stranded RNA virus surrounded by S-proteins with an average diameter of 80–120 nm,[20] similar to influenza A viruses. The outwardly projecting S-proteins are known to be crucial for mediating viral entry into host cells through interaction with the angiotensin-converting enzyme 2 (ACE2) receptor on the surface of human respiratory endothelial cells.[21] The binding between the two proteins triggers the cellular fusion of the virus and the subsequent release of its genetic material into the cytosol. Considering that the S-protein plays a crucial role in the induction of neutralizing antibodies and protective immune responses, it is a prime target for detecting live viruses. Seo et al. developed a field-effect transistor (FET)-based sensor by functionalizing graphene sheets with SARS-CoV-2 S-protein antibodies to detect the S-protein and the whole virion.[22] Pinals et al. detected the S-protein and SARS-CoV-2 virus-like particles using a single-walled carbon nanotube (SWCNT)-based optical sensor functionalized with ACE2.[23] The above examples demonstrate that viruses can be detected by applying a technology that can capture the S-protein of SARS-CoV-2 by intermolecular interactions.

Here, we demonstrate the potentiometric sensor using three-dimensional (3D) molecular imprinting as an effective method for viral testing, which can provide the sensitivity of RT-PCR[24,25] in detecting very low viral titers and differentiating different viral subtypes while producing results in less than 5 min. The technique is straightforward, and hence can be employed at the point of care for rapid communication with health care providers due to the simplicity of potentiometric detection that can be adopted for the development of rapidly deployable test kits, as recently demonstrated by Chaibun et al. for the rapid electrochemical detection of SARS-CoV-2 virions using the isothermal rolling cycle amplification (RCA) technique.[26] The molecular imprinting method has previously been shown to be very effective in the detection of whole particles of the Polio and Zika viruses,[27] where titers as low as 10 PFU/mL were measured in human saliva, without interference from other proteins. The method also could detect proteins such as fibrinogen, hemoglobin, and a cancer embryonic antigen (CEA), as cross-validated by enzyme-linked immunosorbent assay (ELISA).[28] In this study, we demonstrate the applicability of the 3D molecular imprinting technique for a rapid, specific, and highly sensitive detection of (1) enveloped RNA viruses using two different subtypes of influenza A virions, H1N1 and H3N2, and (2) the purified S-proteins of SARS-CoV-2 and Middle-East respiratory syndrome (MERS) coronaviruses.

Results and Discussion

Validation of the 3D Molecular Imprinting Mechanism

The mechanism of 3D molecular imprinting is shown in Figure . OH-terminated thiols are adsorbed onto a Au-coated Si chip together with template molecules, which are target analytes such as virions (Figure a,b). The template molecules, being larger than thiols, adsorb first onto the Au surface and fit conformally onto the surface structure (e.g., concave; with a roughness of ∼62.5 (±15.2) nm in the current study; Figure S2a) originating from the roughness of the underlying unpolished Si surface, which ranges from nano to micron in scale.[29] At the same time, the thiols adsorb around the template molecules, reacting with the Au surface via Au–sulfur bonds, and crystallize into a 2 nm thick self-assembled monolayer (SAM), wrapped tightly around the template. The template molecules are then selectively removed using a 3 M NaCl solution, leaving behind an imprinted thiol layer, an exact inverse replica of the surface molecular structure of the template itself (Figure c). The imprinted Au chip then serves as an electrode of potentiometric analyte detection (Figure d); specifically, the relative open-circuit potential (OCP) of a dPBS (Dulbecco’s phosphate-buffered saline) solution is being monitored using a three-electrode potentiometer consisting of an analyte-imprinted Au electrode, a reference electrode, and a counter electrode. A solution of an analyte is then added dropwise while the OCP is recorded. The change in OCP is generated when the analyte docks into the imprinted thiol layer. This docking process is possible only when the surface molecular structure of the introduced analyte matches that of the imprinted thiol layer, analogous to the lock-and-key mechanism of enzyme–protein interaction, thus enabling a specific detection of an analyte of interest.

Figure 1

Schematic illustration of the 3D molecular imprinting process: (a) Au-coated surface showing the underlying roughness of the unpolished Si wafer and (b) imprinting process where thiols and template molecules (target analytes; virus particles in this case) are adsorbed onto the Au surface. The thiols are permanently bound into a SAM that are crystallized around the template molecules. (c) Template molecules, which are weakly adsorbed on the Au surface, are removed by washing in a 3 M NaCl solution, leaving behind imprinted areas in the SAM with an inverse replica of the surface molecular structure of the template molecules. (d) Potentiometric analyte detection: When the analytes are readsorbed into the imprinted SAM layer, it creates a change in OCP. (e) Schematic example of the potentiometric response as a function of time when analytes are injected. Inset: The illustration of the potentiometric response occurring when the analyte comes in a direct contact with the unprotected imprinted area on the Au contact. (f) Schematic example of the normalized potentiometric response as a function of the analyte concentration for sensors imprinted with different analyte concentrations. Inset: The illustration of the density of the imprinted regions corresponding to the different concentrations.

Figure e,f schematically illustrates typical potentiometric analyte detection data (OCP vs time) generated from a 3D molecular-imprinted electrode: Figure e shows the normalized temporal change in OCP (ΔV/Vo with Vo being the starting, baseline voltage) during analyte injections with increasing concentrations in a series of intervals, with the inset depicting the localized process responsible for the potential change, namely the analyte adsorption into its imprinted area, followed by the local equilibration of the charge. Figure e can be converted to a plot correlating the signal response with the analyte concentration (Figure f), wherein the OCP response initially increases linearly with the analyte concentration and saturates later as all available imprinted regions are readsorbed by the analytes. Meanwhile, the three different plots in Figure f illustrate the need for an optimized molecular imprint pattern density for maximizing the analyte detection response and sensitivity. It is noted that the 3D molecular-imprinted sensor shares similarities with the recently reported electrochemical biosensor based on a molecularly imprinted polymer (MIP), which could effectively detect SARS-CoV-2 S-protein and an antigen with high throughput.[30−32] The MIP sensor utilizes the molecular imprint of target molecules created on the polymer via in situ polymerization, rather than the crystallized SAM used in the current 3D molecular-printed sensor, which can detect both molecules and entire virions by controlling the roughness of the underlying substrate.[29]

Characterization of Molecular-Imprinted Electrodes

The contact angle goniometry provided evidence for the suggested 3D molecular imprinting of analytes and their readsorption processes (Figure a): The contact angle decreased sharply from 98.5° of the starting bare Au surface to 28.5° after adsorbing H1N1 virions to the Au surface with thiols due to the hydrophilic, hydroxyl end group of coadsorbed thiol SAM facing the air interface. This contact angle is slightly higher than that of a pure thiol SAM (25.7°), consistent with the incomplete surface coverage of SAM due the presence of codeposited H1N1 after the imprinting. On selectively removing the H1N1 virions by washing, the contact angle increased to 35.2° due to the freshly exposed bare Au surface localized in the imprinted pattern areas but was recovered upon the following readsorption of the analyte, confirming the explained analyte docking process. The whole processes are enabled by the formation of a strong Au–S covalent bonding of thiol SAM on the Au surface,[33] as supported by Raman spectroscopy (Figure b): The peak at ∼300 cm–1, originating from the Au–S stretching vibration mode, showed a negligible change in intensity, spectral shape, or position during the whole molecular imprinting, analyte washing, and readsorption processes, indicating that the SAM remained intact.

Figure 2

Surface characteristics of the Au surface during the molecular imprinting of H1N1 virions, following washing, and readsorption processes, compared with a pure thiol SAM probed by (a) contact angle goniometry and (b) Raman spectroscopy.

The feasibility of the potentiometric virus detection using a 3D molecular-imprinted electrode is confirmed by measuring the change in electrochemical properties of the Au surface during the steps of the molecular imprinting of influenza A viruses and their readsorption (i.e., detection) using electrochemical impedance spectroscopy (EIS). The influenza A virus was chosen as a model analyte considering its structural similarity with SARS-CoV-2 (i.e., enveloped RNA viruses with similar overall sizes and the presence of surface S-proteins) and the ease of their handling that does not require the biosafety level 3 (BSL-3) facility. Among characterizable EIS parameters, the charge transfer resistance (Rct) of the Au electrode well reflects the described steps of the molecular imprinting process and the analyte detection because the thiol SAM formed at the Au surface and the imprinting virus particles are electrically insulating, and the desorption and readsorption (detection) of the virus particles affect the overall interfacial resistance at the Au electrode surface.

The Nyquist plot obtained from the bare Au chip initially yielded a mass diffusion-limited electron transfer process, featuring a linear relationship between the real and imaginary parts of the overall impedance (Z′ vs Z″) and, thus, a negligible Rct (Figure a; black sphere). After adsorbing the template analyte, H1N1 virions, along with thiols on the Au surface (i.e., imprinting), the plot featured a well-defined semicircle (blue sphere), resulting in a measurable Rct, represented by the intercept at Z′ axis. This Rct value is relatively smaller than that of a pure SAM (Figure a,b; red sphere) because the virus-adsorbed surface has some exposed bare Au surface areas, which are caused by virions physically hindering thiol from evenly binding at the Au surface. After washing and removing the H1N1 virions from the Au surface, the Rct was significantly reduced (Figure a,b; green sphere), indicating that during the washing process, the virions became detached, creating a more exposed bare Au surface, thus increasing the electron transfer between the solution and Au. Now, when H1N1 virions were reintroduced (i.e., the analyte detection), the Rct increased to a level similar to that of the originally imprinted surface (Figure a,b; purple sphere) since the newly readsorbed H1N1 virions reduce the exposed bare Au surface as they dock into the inverse template pattern formed by thiols on the Au surface. Meanwhile, if a different type of virion, H3N2 in this case, was introduced to the H1N1-imprinited Au electrode, the Rct hardly changed (Figure a,b; dark yellow sphere), which indicates that the H3N2 virions could not be fitted to the molecular pattern formed by the H1N1 virions. These results clearly confirm that a specific, electrochemical detection of a target virion is possible based on the 3D molecular-imprinted Au substrate.

Figure 3

Comparison of the electrochemical characteristics. (a) Nyquist plots and (b) Rct of various surfaces representing each process step of 3D molecular imprinting of the target virions (H1N1) and their specific detection in chronological sequence: Bare Au (black) → after the adsorption of thiol SAM only (red) or the simultaneous adsorption of H1N1 virions and thiol SAM (blue) → after selectively removing H1N1 virions by washing, thus creating a H1N1-imprinted Au surface (green) → after re-exposing H1N1 virions (i.e., H1N1 readsorption; purple) or exposing H3N2 virions, which cannot lodge on the surface due to the incompatibility with the H1N1-specific surface thiol SAM pattern (yellow).

Detection of Influenza A Viruses (H1N1 and H3N2)

To use the molecular-imprinted Au electrode as a direct, real-time sensor for virions, we chose OCP detection since it requires the least amount of an analyte and can be sensitively calibrated for the quantitative determination of viral loads. Figure a shows the relative change in OCP (ΔV/Vo) as a function of time for an Au electrode imprinted for H1N1, where each point corresponds to the injection of the analyte of a specified concentration. The OCP response increased as the analyte concentration increased, and for each concentration, the response reached equilibrium within minutes. Overall, the temporal OCP response plot clearly indicates that the limit of detection (LOD) is ∼200 PFU/mL, which corresponds to ∼4000–100,000 particles/mL considering the reported particle-to-PFU ratio of ∼20–500 for H1N1.[34] The LOD is also comparable to that of RT-PCR for H1N1, ∼100 PFU/mL.[24,25] It is worth noting that the achieved LOD is several orders smaller than the typically known virial load of influenza A virus in a patient immediately after infection, which is in the range of ∼103–106 PFU/mL depending on the type of specimens (e.g., respiratory specimen, nasopharyngeal aspirate, and nasal swab).[35,36] This suggests that the molecular-imprinted sensor can provide a simple positive/negative detection. Meanwhile, the viral loads of SARS-CoV-2 in saliva, sputum, and nasal mucus from patients immediately after infection were reported to be ∼104–106 PFU/mL.[37] While being different types of viruses (orthomyxoviridae vs coronavirus), the influenza A virus and SARS-CoV-2 have similar size and morphology (e.g., presence of S-proteins), and, thus, the result suggests that the sensor likely has sufficient sensitivity for detecting SARS-CoV-2.

Figure 4

Temporal, analyte-concentration-dependent OCP response of molecular-imprinted Au electrode sensors to influenza A viruses of (a) H1N1 and (b) H3N2. The OCP response of the sensors for (c) H1N1 and (d) H3N2 as a function of the concentration of analytes (H1N1 or H3N2), demonstrating a virus-specific detection with negligible cross-sensitivity.

Similar to H1N1, various concentrations of H3N2 virions could be detected using an electrode imprinted for H3N2 (Figure b), featuring equally robust OCP signals with a similar LOD of ∼200 PFU/mL. For both types of viruses, there were virial load ranges that elicited a linear OCP response. As shown in Figure c,d, the OCP response increased linearly for both H1N1 and H3N2 as a function of the analyte concentration within a ∼200–1000 PFU/mL range, which can provide calibration data for determining an unknown virial load corresponding to a measured OCP response of the sensor. For a concentration greater than 1000 PFU/mL, the extent of the OCP response started to saturate. While an exact virus concentration cannot be determined in this region, a clear positive would be registered immediately within seconds.

We also confirmed the specific detection of a target virus. As shown in Figure c,d, when H3N2 virions were introduced to the H1N1-imprinted senor, there was a negligible OCP response, and vice versa. As shown in Figure S1, both H1N1 and H3N2 appear spherical in shape with insignificant differences in their mean diameters, 117.9 (±17.7) nm and 118.8 (±13.6) nm, respectively. However, these viruses differ in terms of the chemical/molecular structure of their S-proteins (i.e., H1 vs H3 and N1 vs N2 spikes).[38−41] The results, therefore, clearly show that the molecular-imprinted virus sensor is highly specific and sensitive to the surface molecular structure of a given type of virion.

Detection of the S-Proteins of SARS-CoV-2 and MERS

The SARS-CoV-2 particles are similar in size and structure (coronavirus) to those of H1N1 and H3N2,[42−44] and hence the molecular-imprinted sensor is expected to feature sensitivity and specificity comparable to those of the demonstrated H1N1 and H3N2 detection. Since we were unable to work directly with SARS-CoV-2 virions, the applicability was tested using S-proteins of SARS-CoV-2 virions. The detection of the spike proteins is a valuable assay as well since each SARS-CoV-2 virion is surrounded by about 100 outwardly protruding S-proteins,[45,46] which are major targets for the detection of live viruses. Figure shows the temporal, concentration-dependent OCP responses of S-proteins sensors for SARS-CoV-2 (Figure a), MERS coronavirus (Figure b), and heat-treated (HT)-S-protein of SARS-CoV-2 (Figure c), respectively. For all three types of S-proteins, the LOD was ∼100 pg/mL with saturation occurring at ∼2 × 106 pg/mL. As with the demonstrated whole virion detection, there was an immediate OCP response upon injection of the analyte, with the increase in OCP response for a given analyte concentration equilibrating in minutes. It is noted that the detection of these S-proteins was achieved using smooth, polished Si coated with Au having a roughness of 1.17 (±0.05) nm as a starting substrate (Figure S2b), considering the importance of matching the substrate roughness with the dimension of target analytes for efficient detection.[28]

Figure 5

Temporal, analyte-concentration-dependent OCP response of molecular-imprinted Au electrode sensors for S-proteins of (a) SARS-CoV-2 and (b) MERS, and (c) heat-treated (HT)-S-protein of SARS-CoV-2. The OCP response of the sensors for (d) SARS-CoV-2 S-protein, (e) MERS S-protein, and (f) SARS-CoV-2 HT-S-protein as a function of the concentration of various analytes, including S-proteins of SARS-CoV-2 and MERS, and HT-S-protein of SARS-CoV-2, demonstrating the analyte-specific detection with negligible cross-sensitivity.

The sensor also exhibited an S-protein-type-specific detection: For the sensor imprinted with the S-protein of SARS-CoV-2, the intensity of OCP response increased linearly with increasing concentration of S-proteins of SARS-CoV-2 above the LOD and saturated at ∼2 × 106 pg/mL (Figure d). However, the introduction of S-proteins of MERS or HT-S-protein of SARS-CoV-2 induced a negligible response within the tested concentration range, except for a weak response of HT-SARS-CoV-2 at the highest analyte concentration (2 × 106 pg/mL). Similarly, the sensor imprinted with the S-protein of MERS coronavirus featured a robust OCP response to the MERS virus S-protein only with negligible sensitivity to SARS-CoV-2 or HT-SARS-CoV-2 S-proteins (Figure e). Meanwhile, the HT-S-protein of the SARS-CoV-2 sensor had a weak but non-negligible cross-sensitivity to the S-protein of SARS-CoV-2 with no response to the MERS S-protein (Figure f). These results indicate that the S-proteins of SARS-CoV-2 and MERS coronavirus could be perfectly differentiated from each other by the molecular-imprinted sensor, which is consistent with the fact that the spike proteins of SARS-CoV-2 and MERS coronavirus structurally differ in terms of the orientation in the C-terminal domain of the S1 subunit.[47,48]

On the other hand, the observed weak but non-negligible cross-sensitivity between the S-proteins of SARS-CoV-2 and its counterpart heat-treated at 60 °C, particularly at high analyte concentration, suggests that some heat-treated S-proteins could fit into the imprinted thiol SAM pattern created by SARS-CoV-2 itself, and vice versa. We postulate that although most of the S-proteins had structural modifications during the heat treatment process, there could have been still an unaffected ensemble of S-proteins. Nevertheless, the overall data clearly showcase specific detection of S-protein (∼10 nm in size) by the imprinted sensor, differentiating their molecular-scale structural differences, and therefore this technique should be applicable to a specific detection of variants of SARS-CoV-2 by recognizing the molecular modification in S-proteins on a few nanometers scale occurring during structural mutations.

Detection of Viral Particles and S-Proteins in Human Saliva

We investigated whether the molecular-imprinted sensor could reliably detect target analytes in a human saliva solution, a more realistic sample type usable in clinical environments; to conduct a fast, point-of-care virus detection, it is preferred that samples are collected and used for testing without complex manipulation, such as cell culture typically used for highly sensitive SARS-CoV-2 detection.[49,50] The patient’s saliva can not only contain a high concentration of viral load but also is a body fluid easily collectible in a noninvasive manner, and thus it is suitable as a test sample. However, since nonpurified patient saliva samples contain many other protein molecules in addition to a virus, the highly specific and selective detection of only the target analyte is crucial. The selectivity was probed by injecting pure saliva or saliva mixed with a known concentration of analytes into the 3D molecular-imprinted sensor assembly. Figure shows the analyte-concentration-dependent OCP responses of various coronavirus analytes in a human saliva solution, including virions (H1N1, H3N2) (Figure a,b) and S-proteins (SARS-CoV-2 and MERS coronavirus) (Figure c,d). All of the types of sensors imprinted with different analytes exhibited an immediate response at an analyte concentration greater than 200 PFU/mL for virions or ∼200 pg/mL for S-proteins, with the extent of response increasing with increasing analyte concentration. In contrast, the injection of pure saliva elicited a negligible response for all of the sensors, which indicates that various molecules contained in saliva, such as electrolytes, enzymes, proteins, bacteria, and other viruses, could not adsorb onto analyte-specific molecular-imprinted patterns, which again confirms the detection mechanism enabled by the recognition of the surface molecular structure of analytes with feature scale smaller than 10 nm.

Figure 6

Potentiometric responses of molecular-imprinted sensors as a function of the volume of the added analyte concentration in human saliva. (a) H1N1 virion, (b) H3N2 virion, (c) S-protein of SARS-CoV-2, and (d) S-protein of MERS.

Finally, we also confirmed the specific detection of H1N1 virions against H3N2 (and vice versa) in human saliva with negligible cross-sensitivity (Figure ): when the sensor imprinted with H1N1 virions was challenged by the saliva containing H3N2, there was no response; the sensor responded only when the saliva containing H1N1 was introduced, with the OCP response intensity increasing as the analyte concentration increased (Figure a,c). Similarly, the H3N2 sensor responded only to the H3N2-containing saliva with no cross-sensitivity to N1H1-containing saliva (Figure b,d).

Figure 7

Temporal OCP response of (a) H1N1 and (b) H3N2 sensors to saliva solutions containing H1N1 or H3N2 virions. Corresponding analyte-concentration-dependent OCP response of (c) H1N1 and (d) H3N2 sensors.

Conclusions

We have shown that the 3D molecular imprinting could be used to realize a fast, specific, potentiometric detection of human influenza A viruses and S-proteins of SARS-CoV-2 and MERS coronavirus. Specifically, the technique could selectively detect H1N1 and H3N2 virions in minutes without cross-sensitivity, despite their similarity in size and overall morphology, with LOD down to 200 PFU/mL using a probe volume of less than 10 μL, recognizing the difference in the surface molecular structure of these virions (i.e., their S-proteins). Similarly, the S-proteins of SARS-CoV-2 and MERS coronavirus could be directly detected in minutes with specificity and a low LOD of 100 pg/mL in a test probe of 10 μL. The intentionally modified S-protein of SARS-CoV-2 via heat treatment, which simulates a different strain (i.e., variant), could be also differentiated from the original S-protein. Finally, we have demonstrated a realistic detection of H1N1 and H3N2 virions and S-proteins of SARS-CoV-2 and MERS coronavirus mixed in human saliva, a sample type best suited for a fast, point-of-care virus detection without any invasive sample collection and complex manipulation. The demonstrated speed and specificity of detection distinguishing different analytes via the recognition of the surface molecular structure (S-proteins in the current study), combined with a low required probe volume, high sensitivity, ease of potentiometric measurement (which can be readily miniaturized using an integrated circuit), and simple sample collection and preparation highlight a strong potential of the technique as a highly effective point-of-care diagnostic platform for a fast, accurate, and specific detection of various viral pathogens and their variants, including those related with SARS-CoV-2.

Materials and Methods

Materials

11-Mercapto-1-undecanol (SH-(CH2)11-OH), potassium ferrocyanide (K4[Fe(CN6)]), potassium ferricyanide (K3[Fe(CN6)]), sodium chloride (NaCl; ACS reagent), potassium chloride (KCl), and acetone (high-performance liquid chromatography (HPLC) Plus, ≥99.9%) were all purchased from Sigma-Aldrich. Dimethyl sulfoxide (DMSO; certified ACS) was purchased from Fisher Scientific and ethyl alcohol (200 proof) was purchased from Pharmco-AAPER. Dulbecco’s phosphate-buffered saline (1× dPBS), without calcium chloride or magnesium chloride, was purchased from Life Technologies. Uranyl formate powder (UO2(CHO2)2·H2O) and carbon-coated 400 mesh copper grids were purchased from Electron Microscopy Sciences. Deionized water (DI H2O) was prepared in-house using a MilliporeSigma Direct-Q 3 UV-R water purification system at 23.0 ± 1.0 MΩ cm. Adhesive poly(tetrafluoroethylene) (PTFE) tape (S-14538) was purchased from ULINE, and Ag/AgCl reference electrodes (MW-2030 and MF-2052) and a coiled Pt wire auxiliary electrode (MW-1033) were purchased from BASi. Influenza A virus (H1N1/H3N2) was obtained from Codagenix, Inc. Spike proteins (SARS-CoV-2, MERS-CoV) were obtained from the Center for Systems and Synthetic Biology, University of Texas, Austin. HT- S-protein of SARS-CoV-2 was prepared by immersing the normal S-protein-sealed centrifuge tube in a water bath at 60 °C for 30 min. Human saliva was collected from healthy individuals by the institutional IRB-approved human subject research protocol (IRB 1212002). Informed consent was obtained from all subjects prior to collecting saliva samples. All handling of viral samples was done in a BSL-2 enclosure unless indicated otherwise.

Fabrication of Au-Coated Si Substrates

A 100 mm diameter single crystal (100) Si wafer was cleaned by oxygen plasma (20 W; 100 mTorr; 1 min) (March CS-1701). Ti/Au (7 nm/30 nm) layers were deposited onto the cleaned Si wafer on either a polished frontside (smooth surface) or unpolished backside (rough surface) via thermal evaporation (Kurt J. Lesker PVD75) under a base pressure of <10–6 mbar with a deposition rate of 1 ± 0.5 Ås–1. The roughnesses of the smooth and rough substrates were determined to be 1.17 (±0.05) nm and 62.5 (±15.2) nm, respectively, by atomic force microscopy (Figure S2). The Au-coated substrates were soaked in warm acetone for 30 min, followed by water-rinsing and final N2 blow-drying to remove any organic residue from the surface. The prepared substrate was covered with a PTFE tape with a 6 mm diameter hole made by a hole puncher. The exposed 6 mm diameter circular area was subjected to the molecular imprinting of an analyte of interest, as described in the next section. The PTFE tape remained on the substrate as a mask during the imprinting process as well as potentiometric detection.

Molecular Imprinting Process (SAM Formation)

The mechanism of the surface molecular imprinting is shown in Figure . To prepare a well-ordered SAM with a targeted molecular imprint on the gold surface, an imprinting solution was prepared. The imprinting solution was obtained as follows: The template analyte was dispersed in 1× dPBS, and alkanethiols were dissolved in DMSO. Then, a mixture solution was made by blending them in the 19/1 (1× dPBS/ DMSO) volume ratio and continuously stirring for at least 10 min at 300 rpm. A compromised dissolution process was used due to the poor solubility of thiols in aqueous media and the denaturation of proteins in nonaqueous media or at extremely low pH. Following the literature value for the formation of a stable SAM, the concentration of the thiol was chosen to be 100 μM.[51] The final concentration of the imprinted analyte was 105 PFU/mL for virus and 2 mg/mL for S-protein, far less than 10 mol % of the thiol to restrain aggregation of the proteins and formation of the supramolecular structure. This also ensured a good solubility for thiol and no precipitation of biomolecules. Then, the fabricated Au-coated Si chip was immersed in 2 mL of the imprinting solution for 3 h at room temperature to coadsorb the analyte and thiol to the Au surface. During this process, thiol molecules are densely attached to the electrode surface through a Au–S bond, and templating biomolecules are adsorbed to the gold surface mostly through van der Waals force, hydrophobic, and/or electrostatic interactions, without strong chemical bonding.[52,53] The chips coadsorbed with template molecules and SAM were rinsed with DI water and then soaked in 10 mL of a 3 M NaCl solution at 37 °C for 3 h with gentle agitation, which was shown to be effective in selectively removing template molecules without affecting the integrity of SAM on the surface. Thereafter, the washed chips were rinsed with DI water and then soaked in 10 mL of DI water at 37 °C for 30 min with gentle agitation. By removing biomolecules from the Au surface through this process, cavities that are complementary to the template molecules are formed on the SAM matrix, creating the final imprinted sensor substrate. Due to its inherent complementarity, the prepared molecular imprint pattern and the target analyte behave like an enzyme–substrate complex, thus the readsorption being allowed only for the same type of biomolecules as the imprinted on the sensor chip surface.[54] As the target biomolecules are readsorbed into the imprinted cavity, the surface potential of the sensor is decreased and can be measured in real-time potentiometrically. Finally, the washed, molecular-imprinted chips were dried and stored at room temperature in a desiccator until tested within 10 h.

Potentiometric OCP Measurement

An open-circuit potentiometer (Lawson Laboratories, Model EMF16) was used to analyze the readsorption of target analytes on the molecular-imprinted surface by monitoring the change in OCP of the sensor chip surface without applying an external electric field. A three-electrode system was used, wherein a Ag/AgCl electrode was used as both reference and common electrodes, and a molecular-imprinted Au-coated Si chip was used as a working electrode. The three electrodes were all placed in a 10 mL beaker with 8 mL of 1× dPBS continuously stirred with a magnetic stirrer and a stir bar at 300–500 rpm. A known amount of an analyte was pipetted dropwise into the detection system in intervals after allowing the system to come to equilibrium, while OCP was recorded in real-time using L-EMF DAQ software. The detected OCP signal was normalized as [ΔV/V0] = (V – V0)/V0, where V is the detected real-time potential and V0 is the initial saturated potential.

Raman Spectroscopy

Raman scattering spectra were collected using visible excitation (785 nm, He–Ne laser; Renishaw inVia Raman microscope) in ambient air. The scattered photons were directed into a single monochromator and focused onto an air-cooled charge-coupled device. The Raman shift was calibrated with a built-in silicon standard sample. The spectral acquisition was accumulated 20 times (i.e., 20 scans) with 10 s/scan.

EIS Measurement

EIS measurements were performed using a Bio-Logic Science Instruments SP-200 potentiostat with the corresponding EC-Lab V11.10 software with a three-electrode system with a molecular-imprinted Au ship as a working electrode, a platinum wire as a counter electrode, and a Ag/AgCl electrode (BASi) as a reference electrode. The three electrodes were all placed in a 15 mL electrochemical glass cell vial with 15 mL of a KCl solution (0.1 M) containing [Fe(CN)6]4–/[Fe(CN)6]3– (2 mM each). EIS analyses were performed with an amplitude of 5 mV and a frequency range of 0.1–100,000 Hz.

Transmission Electron Microscopy (TEM)

Molecular structures of influenza A viruses were analyzed by TEM. A total of 5 μL of a dPBS solution mixed with influenza A virus was placed on a glow-discharge, carbon-coated Cu grid and adsorbed for 1 min. Then, the solution was removed by filter paper. The grid was then stained by a 0.75% uranyl formate droplet for 30 s. Morphology and particle size distribution were obtained from stained samples using a JEOL 2100 TEM system at 200 kV.

Atomic Force Microscopy (AFM)

AFM measurements were performed to analyze the surface topographies of two types of electrodes using a Park NX-20 atomic force microscope with a noncontact mode at room temperature and the scan rate was 0.7 Hz.