Enhanced Enzymatically Amplified Metallization on Nanostructured Surfaces for Multiplexed Point‐of‐Care Electrical Detection of COVID‐19 Biomarkers

Abstract

Inexpensive yet sensitive and specific biomarker detection is a critical bottleneck in diagnostics, monitoring, and surveillance of infectious diseases such as COVID‐19. Multiplexed detection of several biomarkers can achieve wider diagnostic applicability, accuracy, and ease‐of‐use, while reducing cost. Current biomarker detection methods often use enzyme‐linked immunosorbent assays (ELISA) with optical detection which offers high sensitivity and specificity. However, this is complex, expensive, and limited to detecting only a single analyte at a time. Here, it is found that biomarker‐bound enzyme‐labeled probes act synergistically with nanostructured catalytic surfaces and can be used to selectively reduce a soluble silver substrate to generate highly dense and conductive, localized surface silver metallization on microelectrode arrays. This enables a sensitive and quantitative, simple, direct electronic readout of biomarker binding without the use of any intermediate optics. Furthermore, the localized and dry‐phase stable nature of the metallization enables multiplexed electronic measurement of several biomarkers from a single drop (<10 µL) of sample on a microchip.This method is applied for the multiplexed point‐of‐care (POC) quantitative detection of multiple COVID‐19 antigen‐specific antibodies. Combining a simple microchip and an inexpensive, cellphone‐interfaced, portable reader, the detection and discrimination of biomarkers of prior infection versus vaccination is demonstrated.

A multiplexed, microscale, sensitive, quantitative electronic biomarker detection technique is developed. Synergistic enhancement of enzymatic metallization on nanostructured surfaces is found. Localized metallization on microelectrodes directed by enzyme‐labeled immunoprobes enables a simple, dry‐state resistive readout of biomarker abundance. Applied to COVID‐19, a multiplexed viral antigen‐specific antibody measurement from a single drop of serum drops reveals unique convalescent and vaccinated antibody fingerprints.

1. Introduction

The current COVID‐19 pandemic and other recent outbreaks such as Ebola, Middle East Respiratory Syndrome, Severe Acute Respiratory Syndrome, and H1N1 have underscored the need for early detection and continued surveillance of emerging and re‐emerging infectious diseases. Outbreaks, many of which start with a zoonotic transmission event, often begin in remote or resource‐poor areas but threaten to rapidly spread globally if not brought under control promptly.^{[ 1} , ² , ³ , ^{4 ]} A critical bottleneck in achieving this is the lack of rapid and scalable yet accurate diagnostic, monitoring, and surveillance tools for infectious diseases.^{[ 5} , ^{6 ]} Diagnostic availability severely limited early efforts to contain the COVID‐19 pandemic in many countries globally. Even now, as the worldwide deployment of COVID‐19 vaccines progresses, concerns regarding the durability of vaccine efficacy, especially against newly emerging variants, have created a further need for the rapid monitoring of heterogenous and time‐varying individual vaccine responses as well.^{[ 7} , ^{8 ]} Multiplexed detection of various biomarkers such as multiple infection or vaccine‐elicited antigen‐specific antibodies in a single test, especially when coupled to machine‐learning based multivariate analytics to derive further clinical insight, can help fulfill the above critical needs.^{[ 9 ]}

ELISAs are the gold‐standard in laboratory‐based sensitive and quantitative detection of a range of biomarkers including serological testing for infectious diseases such as COVID‐19.^{[ 10} , ^{11 ]} For example, titers of neutralizing antibodies directed against the Spike antigen of SARS‐CoV‐2, as measured by quantitative ELISAs, are key correlates of vaccine‐induced protection against COVID‐19 and can thus be used to monitor individual vaccine efficacy.^{[ 12 ]} ELISAs are routinely performed using bulky and expensive but highly sensitive instruments which require highly trained personnel to operate.^{[ 13 ]} These instruments are usually based on optical detection of enzymatic probe‐catalyzed amplification products using, for example, lasers for illumination and photomultiplier tubes for detection. They remain too complex and expensive to scale and deploy globally. Even in resource‐rich settings they are currently used only in centralized diagnostics laboratories.^{[ 13 ]} On the other end of the complexity versus cost space, inexpensive lateral flow‐based assays (LFA) are easier to perform and deploy as point‐of‐care (POC) tests which offer binary (yes/no) readouts.^{[ 14} , ¹⁵ , ^{16 ]} These tests however lack both the sensitivity and quantitative ability of ELISAs.^{[ 16 ]} Additionally, neither ELISAs nor LFAs routinely offer multiplexing ability and require running separate tests for each separate biomarker further increasing cost and complexity.^{[ 16} , ¹⁷ , ^{18 ]} In blood‐based tests, the need for a larger sample volume for multiple tests can pose a significant practical challenge to POC testing. Large volumes of blood (>1 mL) can only be obtained via venipuncture‐based phlebotomy, instead of the finger‐prick based acquisition of small volume, droplet‐scale (<10 µL) samples. This necessitates additional equipment, expertise (e.g., trained phlebotomists), and results in higher biosafety and regulatory burdens as well.

Electrical and electrochemical detection methods can transduce biochemical information such as analyte concentration directly to an electronic signal and thus offer an attractive opportunity to bypass expensive optical detection methods while retaining sensitivity and quantitative ability.^{[ 19} , ²⁰ , ²¹ , ^{22 ]} Interest in transforming diagnostics to an inexpensive POC format has led to a rapid expansion of research in electrochemical biosensing. Electronic detection principles are also more amenable to scaling down sample, sensor and system size, and cost via miniaturization, using microfabrication techniques, without the loss of sensitivity that some optical detection methods (e.g., absorbance) suffer upon length scaling.^{[ 19} , ²³ , ²⁴ , ²⁵ , ^{26 ]} Most current electrochemical biosensors however still remain too complex, specialized, and expensive compared to LFAs and thus have not found wide use in the clinic for POC diagnostics.

Silver reduction catalyzed by gold nanoparticles (AuNP) attached to target‐specific probes has been used earlier for POC optical detection of infectious disease biomarkers as well for electrical conductivity‐based DNA detection using passivated gold electrodes.^{[ 27} , ²⁸ , ²⁹ , ^{30 ]} Enzymatic silver metallization, developed earlier for contrast‐enhancement in electron microscopy of biological tissue sections, offers an attractive alternative for localized and rapid enzymatically enhanced silver deposition which can then be seamlessly coupled to a broad range of existing ELISA chemistries.^{[ 31} , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ^{39 ]} However, the electrical properties of enzymatically deposited silver layers have remained underexplored and difficult to control. Indeed it has been reported that background silver reduction on electrodes limits the use of these techniques in electronic bioassays, especially when using complex sample matrices such as blood or serum.^{[ 35 ]}

Here, we set out to develop a broadly applicable, miniaturized electronic detection principle for bioassays which can retain the simplicity of signal readout such as LFAs and yet can offer the sensitivity and quantitative detection ability of ELISAs while adding the ability for multiplexed detection of biomarkers. Using surface‐bound AuNPs to create a catalytic nanostructured surface and adding enzyme‐labeled target‐specific probes, we observed a novel synergistic effect on the probe‐directed enzymatic metallization which caused significantly enhanced selective silver deposition resulting in a highly conducting silver metal layer. We exploited this effect here to develop a novel, sensitive transduction principle that converts the binding events of biomarkers directly to the electrical properties of the amplified silver layer which can then be measured simply as a dry‐phase resistance using microelectrodes without the use of any bulky intermediate optics or expensive instruments. We demonstrate here that this method can be used for sensitive and quantitative detection of antibodies against SARS‐CoV‐2 viral antigens from convalescent COVID‐19 patient serum. Harnessing the miniaturized nature of this detection technique, we developed microchips which enable high‐throughput clinical screening bioassays in a portable format allowing the measurement of larger numbers (>50) of small volumes of (<0.1 µL) clinical samples on a single chip. Finally, we established that the localized nature of the metallization reaction and its dry‐phase readout technique enables specific detection of multiple biomarkers on nearby but electrically unconnected microelectrode pairs, from a single sub‐5 µL droplet of sample as well. We exploited this to detect multiple SARS‐CoV‐2 antigen‐specific antibodies from convalescent patient serum and vaccine‐recipient serum which could be distinguished from each other based on the multiplexed electronic antibody readout.

2. Results and Discussion

To explore the electrical properties of enzymatically amplified silver metallization on micro‐interdigitated electrodes (µIDEs), we ran initial assays of horseradish peroxidase‐conjugated streptavidin (HRP‐SA) binding with biotin‐conjugated bovine serum albumin (biotin‐BSA) immobilized on the µIDEs as a model target‐probe binding pair.^{[ 40 ]} For this, a microchip with an array of gold µIDEs on glass was microfabricated, which were verified prior to use to have very high electrical resistance (>10⁸ Ω) or effectively open circuits. These were then treated with a molecular adhesion layer (poly‐l‐lysine, PLL) and reversibly assembled with a thin laser‐cut polydimethylsiloxane (PDMS) film with an array of microwells. Biotin‐BSA was immobilized, as target, in the microwells (Figure  1a) incubated with the HRP‐SA probe solution alone. Silver enzymatic metallization substrate solution was added next (Figure 1b) and after a set metallization reaction time, the microchip was washed and dried (see Experimental Section and Figures S1 and S2, Supporting Information, for details). Enzymatic silver metallization was observed on the microchip but it was found, in optical micrographs (Figure 1f), to have a low density and electrical measurements of µIDE resistance showed a closed circuit but with a high (≈5 × 10⁵ Ω) and nonrepeatable (CV > 10%) measured resistance (Red bar in Figure 1e). Scanning electron microscopy (SEM) revealed a low number density of silver nanoparticles on the glass surface between electrodes (Figure 1g) and also a characteristic “desert rose” or “rosette” morphology of individual silver nanoparticles (Figure 1h), as has been described earlier in literature.^{[ 38} , ^{41 ]} Intriguingly, SEM imaging also showed a high density of silver nanoparticles on the gold surface of the µIDEs (Figure 1h). We inferred from this that the low density of silver nanoparticles on the glass was not enough to create a highly electrically conductive and repeatable path between two successive fingers of the µIDEs. We also hypothesized that the high density of silver nanoparticles on gold electrodes is due to the fact that the gold electrode surface itself can act as a competing nucleating and catalytic surface for silver reduction and deposition. As a control experiment, an identical assay was repeated in a microwell with only glass surface but no gold µIDEs. This resulted in a dark and dense silver metallization. This verified the interfering effect of the gold electrodes on enzymatic silver metallization,^{[ 35 ]} which we confirmed to occur on other metallic electrode surfaces as well (data not shown). We conjectured that this effect could be flipped and exploited as an advantage instead, from the electronic detection perspective, by providing an electrically unconnected nanoscale gold catalytic surface on the glass surface where it could help in eventually creating highly conducting path between the µIDE fingers after target‐probe driven enzymatic metallization.

Figure 1.

Development of a novel transduction technique for converting the binding events of a biomarker to the electrical properties of an amplified silver metallization layer. a) Model target‐probe binding assay chemistry was performed on a gold microelectrode array (micro‐interdigitated electrodes, µIDEs) on glass assembled with laser‐cut polydimethylsiloxane (PDMS) microwells. Assay steps for investigation of the silver metallization using b) horseradish peroxidase‐conjugated streptavidin (HRP‐SA), c) gold nanoparticle (AuNP), and d) HRP‐SA with AuNP‐SA. e) Impedance values of silver metallization from assays with different probe solutions. Each data point is the mean of n = 2 replicates and error bars are standard deviations. f) Optical image of silver metallization with HRP‐SA probe only. g,h) SEM of silver metallization with HRP‐SA probe only. i) Optical image of silver metallization with AuNP‐SA probe only. j,k) SEM of silver metallization with AuNP‐SA probe only. l) Optical image of silver metallization with HRP‐SA and AuNP‐SA probes. m,n) SEM of silver metallization with HRP‐SA and AuNP‐SA probes.

To test this conjecture, streptavidin‐labeled AuNPs (AuNP‐SA) were included in the assays next. In a control assay, AuNP‐SA was used as the only probe (Figure 1c) to investigate the amount of silver metallization that can be catalyzed by AuNPs alone. Results of this AuNP‐only assay showed, however, an even lower density silver metallization (Figure 1i) on both glass surface and gold electrodes resulting in a very high measured resistance or effectively an open circuit (>10⁸ Ω, blue bar in Figure 1e and SEM image in Figures 1j–k). This shows that enzymatic metallization can create higher silver metallization than AuNPs alone, but neither is alone sufficient to create high conductivity paths between the electrodes. Surprisingly, however, in an assay where a mixture of AuNP‐SA and HRP‐SA was used as probe (Figure 1d), silver metallization was highly enhanced on the glass surface (Figure 1l) and this resulted in a repeatable and low measured resistance (<100 Ω, Green bar in Figure 1e and SEM image in Figures 1m,n). Electron microscopy revealed here a very high density of nanoparticles with similar rosette‐like morphology as observed earlier but now they formed a highly connected network of nanoparticles which were merging to also show an almost continuous layer‐like morphology. This high‐density, highly connected, and thus highly conductive silver metallization was clearly beyond what was observed from either of the above two metallization reactions catalyzed by HRP alone or AuNP alone. And indeed, it was even higher than what would be expected if those reactions were to be just occurring simultaneously but independently of each other here, i.e., the enhanced density of metallization is greater than what would be expected from a simple additive effect of the two independent reactions.

The results of this assay with combined AuNP and HRP probes thus showed that the surface‐bound AuNPs and the enzyme have a hitherto‐undescribed synergistic catalytic effect on the metallization which significantly enhances its density and conductivity. To the best of our knowledge, there is no mechanistic explanation currently available for this novel synergistic effect. We propose a hypothetical mechanism here that we think can explain our observations and corroborate other related prior observations. Silver metallization density on the surface depends on two key steps: reduction of silver ions and the attachment of the reduced silver to the surface. Both AuNPs and HRP are known to independently act as catalytic agents for reduction of silver ions in the presence of appropriate other reducing agents^{[ 42 ]} and for HRP, oxidizing substrates as well.^{[ 39 ]} Previous work has also shown that AuNPs act as a nucleating surface for the deposition of the reduced silver metal.^{[ 27 ]} On the other hand, it has also been shown earlier that HRP—and indeed many other proteins—act as nucleating surface for the deposition of reduced silver metal too. This forms the basis of the commonly used silver‐staining technique of visualizing proteins after gel electrophoresis. But this silver deposition has been observed to be a self‐limiting process.^{[ 38 ]} It has been hypothesized that this is due to blocking of the catalytic reduction site on the protein by the deposited silver itself. Here, we hypothesize that the synergistic effect of AuNPs and HRP emerges from the fact the HRP‐catalyzed reduced silver can, in this case, co‐deposit on nearby AuNPs, using them as the nucleating surface rather than the HRP alone. This can relieve the blocking and self‐limiting effect of the deposited silver on the catalytic effect of the HRP and allows the reduction reaction to proceed for much longer before the enzyme is rendered ineffective. Additionally, AuNPs likely continue to play a catalytic role as well in reducing more silver ions which adds to the total silver metallization on the surface. This novel synergistic effect can be used to transduce a biochemical binding event to an enzymatically amplified, dry‐stable, silver metallization layer and thus a simply measurable large change in electrical resistance.

To exploit this new transduction strategy in the electronic detection of disease biomarkers, we next aimed to test it for the detection of anti‐viral antigen‐specific antibodies in COVID‐19. Human immunoglobulin G (IgG) antibody directed against the SARS‐CoV‐2 Spike protein (anti‐S IgG) was selected as the initial target biomarker to pursue this. To obtain a direct electrical readout of anti‐S IgG, Spike protein (S) was immobilized on the µIDEs. Serially diluted clinical serum samples from convalescent COVID‐19 patients were added to the microwells (3 µL per well) and then probed with a mixture of HRP‐labeled and AuNP‐labeled anti‐human IgG antibodies followed by the silver metallization substrate solutions (Figure  2a). Dense silver metallization was observed on the µIDEs (Figures 2d–g) which were also found to have a serum‐dilution dependent measured resistance (red curve in Figure 2c) while corresponding pre‐pandemic healthy control serum and buffer only control showed little or no silver deposition (Figures 2h,i) and remained open circuits (data not shown). This demonstrated the ability of this transduction scheme to electronically detect and quantify viral antigen‐specific antibodies as COVID‐19 biomarkers from patient serum.

Figure 2.

Two strategies for integration of gold nanoparticles (AuNPs) in assays of anti‐S IgG as a biomarker. a) Integration of AuNPs in the assay as a probe. b) Integration of AuNPs by creation of a nanostructured surface by co‐immobilization of AuNPs with S antigen. c) Serial dilution curves for electronic anti‐S IgG detection using the above two strategies. Each data point is the mean of n = 5 replicates and error bars are standard deviations. SEM images of silver metallization resulting from d,e) [1:40] dilution of COVID+ serum sample, f,g) [1:320] dilution of COVID positive serum sample, and h,i) phosphate buffer saline (PBS) control.

While the mixed AuNP‐labeled probe and HRP‐labeled probe‐based immunoassay design clearly demonstrated the same synergistic enhanced silver metallization effect as described earlier, we considered whether such mixed probes can create a competition between the two probes for binding to the captured anti‐S IgG molecules, which at high dilutions would act as a limiting reagent. This could thus eventually limit the overall sensitivity of this scheme. To avoid any such competitive effect and further boost the sensitivity of this transduction strategy, we explored a second variation of the assay scheme where we still integrate AuNPs in the assay but in an earlier assay step. In this second assay scheme, AuNPs were co‐immobilized on µIDEs with the antigen (i.e., S protein) in the first step (Figure 2b). This results in the creation of a nanostructured, catalytic AuNP‐bound surface on the µIDEs before the sample and probe binding and enzymatic metallization reaction steps occur (Figure S3, Supporting Information). The rest of the assay was same as earlier, except that the probe solution contained only HRP‐labeled anti‐human IgG probe now. Interestingly, this second assay scheme resulted in more than 100× further enhancement of sensitivity of detection (Blue curve in Figure 2c). In addition, it eliminated the need for addition of AuNP‐labeled probes and thus simplified the assay. Figure 2c shows the serum dilution‐dependent resistance obtained from both these assay schemes, obtained in each case starting with a total of <1 µL volume of serum. This assay can thus quantify anti‐S IgG from serum sample volumes lower than 1 µL, and with the second scheme, at a dilution up to ≈10 000‐fold. SEM images of silver metallization (Figures 2d–i) also showed that the density of silver metallization on glass increased with increasing concentration of serum sample (lowering dilution factor) and thus results in higher conductivity or lower resistance value as expected. We hypothesize that this enhancement is due to higher effective concentration of both AuNPs and HRP‐labeled probes in this scheme compared to when AuNP‐labeled probes and HRP‐labeled probes are used, where they have to compete to bind with surface‐bound analyte. Also, we tested whether the reaction kinetics change and found that silver metallization indeed occurs faster, in an analyte‐concentration dependent manner, with AuNP immobilization on the surface versus the use AuNP‐labeled probes (Figure S4, Supporting Information). We found that longer metallization times can result in increased sensitivity or lower limit of detection (LOD) up to a point. Interestingly, we also found that an eventual limit on increasing sensitivity via longer metallization times occurs via reduced contrast between positive and negative controls. At higher reaction times, nonspecific metallization, presumably via silver reduction by the surface‐bound AuNPs alone, starts to occur. Hence there is an optimum metallization time which provides the best combination of sensitivity and specificity. All further assays this work, here onwards, were carried out with this second higher sensitivity assay scheme with the optimum time found above (≈8 min). As described above, this makes use of antigens immobilized on a nanostructured, catalytic AuNP‐bound surface and its synergistic effect with enzyme‐labeled probes to generate an enhanced, high density silver metallization to transduce specific probe‐binding events as a change in electrical resistance.

While, clearly, this assay and detection technique provides more than enough sensitivity to detect anti‐S IgG as a biomarker from patient serum, in order to compare across sensing strategies and help expand the use of this technique to other applications, a quantitative sensitivity metric beyond the maximum serum dilution metric used here can be useful. The exact concentration anti‐S IgG in patient serum is, however, not known as it can vary based on stage of infection or host response. It is estimated that antigen‐specific IgG can vary between 0.1–1% of total serum IgG (6–16 mg mL⁻¹)^{[ 43 ]} and thus the 1:10 000‐fold dilution of serum could contain ≈0.6‐16 ng mL⁻¹ or equivalently ≈4–100 × 10⁻¹² m anti‐S IgG. In order to measure the LOD of our device and assay scheme and compare it with a standard optical ELISA, we also measured dose–response curves using a human monoclonal IgG antibody (mAb) directed against the SARS‐CoV‐2 S antigen as an analyte, using both platforms. Based on these results (Figures S5 and S6, Supporting Information), we find that the LOD on both platforms for this analyte is in ≈10–100 × 10⁻¹² m range. We thus also hypothesize that currently the LOD is likely limited by the analyte binding properties (K _d) rather than the detection platform itself. This indicates future optimization efforts maybe directed on improving analyte binding and more effective analyte capture to the surface.

In order to enable screening of large cohorts of clinical samples, as is often needed in the context of a pandemic, high throughput assay and detection techniques are critical. Next, we built the above transduction scheme into a high throughput microchip platform. Exploiting standard microfabrication techniques, we developed a high throughput chip with the dimension of a standard microscope slide (25 × 75 mm) that can be used for screening of 60 different serum samples. Also, as described earlier, a reversibly sealed thin PDMS film laser‐cut with microwells formed the wells around the electrodes with sample dispensing performed using standard multichannel pipettes, as commonly used in clinical laboratories (Figure  3a). The thin film wells also facilitated easy washing and other slide‐scale liquid handling by dipping the whole slide in wash buffers etc. enabling ease of use and automation using standard slide‐handlers. We term this microchip, the Electrode Array System for electronic ELISA, or the EASyELISA chip here. A cohort of 20 human serum samples (10 COVID+ serum samples and 10 healthy serum samples) with three replicates for each sample, were tested using the EASyELISA chip using the assay chemistry described above, with anti‐S IgG as the target biomarker. All COVID+ serum samples resulted in a dense spot of silver metallization while the healthy controls showed a significantly lower silver metallization (Figure 3b). SEM images of silver metallization also clearly showed a difference in density of silver metallization in COVID+ samples versus healthy controls (Figure 3c,e). The corresponding impedance values for COVID+ samples were more than six order of magnitude lower than the ones of the healthy controls, resulting in an area under the receiver–operator characteristic curve (AuRoC) of 1. This result showed that the EASyELISA chip can be used as a miniaturized platform for high throughput screening of clinical samples from serum samples as low as 0.1 µL. To compare performance of the EASyELISA with a gold standard technique, a standard microwell‐plate based ELISA with optical detection was conducted with the same cohort of clinical samples (Figure 3f). This resulted in a AuROC of 0.99 and showed perfect concordance between the EASyELISA and standard ELISA in distinguishing COVID+ versus healthy sera.

Figure 3.

EASyELISA chip for high throughput screeing of anti‐S IgG as a biomarker in clinical samples. a) Image of an EASyELISA chip including 60 pairs of micro‐interdigitated electrodes (µIDEs) for screening 60 different samples. b) Differentiation between 10 COVID+ serum samples and 10 healthy controls. SEM images of silver metallization resulting from a c) COVID+ sample and a e) healthy control. d) Impedance values for differentiation of COVID+ serum samples from healhy controls. f) Optical enzyme‐linked immunosorbent assays (ELISA) as a gold standard technique for testing the clinical samples. Each data point represents the mean of n = 3 replicates.

Finally, having established the ability for quantitative detection of single biomarkers and their high throughput screening from small volumes of clinical samples, we set out to investigate the possibility of multiplexed detection of biomarkers from a single small volume sample. We did this via the patterning of different capture antigens as spots on nearby but electrically unconnected µIDEs which can capture different targets from a single drop of sample, as is done in optically read microarrays. The key question here was whether the silver metallization, as it occurs from the substrate solution, is localized to where the enzyme‐labeled probe is bound or does it diffuse or spill over to nearby areas and hence cause cross‐talk in the electrical readout signal. A multiplexed microchip (15 × 25 mm) was fabricated that included four pairs of µIDEs for detection of four different biomarkers (Figure  4a). Two layers of PDMS were laser‐cut and reversibly sealed on the chip. The first layer included four microwells which were aligned on the µIDEs and used for immobilization of different antigens. The second layer included a larger well covering all the smaller wells for sharing sample and all further reagents between the smaller wells (Figure 4b). To investigate the possibility of multiplexed detection from a single sample drop, the model biotin‐SA chemistry was used again. Biotin‐BSA along with AuNPs was immobilized as the positive control on a different pair of µIDEs on each of four multiplexed microchips while the remaining three pairs of µIDEs on each microchip were coated with BSA and AuNPs as the negative controls. HRP‐SA probe and silver substrate solutions were then sequentially added to the bigger sample microwells on each microchip so as to cover all four µIDEs (Figure 4c). After metallization, washing and drying, metal deposition was observed only on positive control microwells, and no metallization was formed on negative control microwells (Figure 4d). All positive controls showed low measured resistance and negative controls remained as open circuits. This shows that although the sample, probe, and silver substrates were all shared as a single drop between smaller microwells, metallization occurred locally and remained on the surface of microwells which were coated with biotin‐BSA and did not diffuse away or attach to other microwells. Thus, unlike a conventional ELISA where converted colorigenic or fluorogenic signal due to surface‐bound enzyme‐labeled probes can diffuse in solution, in this transduction technique, the enzymatic silver metallization is localized to where the enzyme‐labeled probe is bound and further is dry‐stable after washing as well. This result establishes a unique ability of this transduction technique for multiplexed enzyme amplified electronic detection of biomarkers from a single small drop of sample without any concern for cross‐talk of signals from adjacent microwells.

Figure 4.

Multiplexed electronic biomarker detection. a) Multiplexed microchip for detection of four biomarkers. b) Design of two polydimethylsiloxane (PDMS) layers for different assays formats of multiplexing. c) Biotin‐SA assay for testing locallization of silver metallization. d) Localized property of silver metallization and impedance data for positive controls (biotin‐BSA) versus negative controls (BSA).

To leverage this unique property of this transduction technique that generates an amplified and localized signal, multiplexed assays were designed next for detection of multiple COVID‐19 antigen‐specific antibodies which can act as biomarkers of prior infection as well as vaccine response. The layout and reconfigurability of the multiplexed microchip and resealable thin‐film PDMS microwell layer design allow different formats of multiplexing using the same microchip design. For instance, one isotype of antibodies (e.g., IgG) against different viral antigens can be detected by immobilizing different antigens on the four different µIDEs and using a probe against the specific isotype of antibody. Alternatively, different isotypes of antibodies (e.g., IgG and IgM) against multiple antigens can also be detected using different isotype‐specific probes. Here, we demonstrate both formats of multiplexing and use them for distinguishing healthy control, COVID+ patients and vaccine recipients with no prior infection as well as for detecting two isotypes of human antibodies.

In the first format, the multiplexed assay was designed as follows: Four different proteins including two SARS‐CoV‐2 antigens—S and nucleocapsid protein (N), along with BSA (−ve CTRL), and Protein A/G (+ve CTRL) were immobilized with AuNPs on each of the µIDEs of a multiplexed microchip (Figure  5a). To examine specificity of the assay, samples containing human anti‐S IgG monoclonal antibody (S mAb) and human anti‐N IgG monoclonal antibody (N mAb) were assayed using this microchip. The S mAb sample resulted in a conductive and dense layer of silver metallization only on the S antigen and +ve CTRL spots with no cross‐reactivity with the BSA and N spots (Figure 5b). Similarly, the N mAb resulted in a conductive and dense layer of silver metallization only on the N antigen and +ve CTRL spots with no cross‐reactivity with the BSA and S spots (Figure 5c). The results from these mAb assays validate the specificity of the assays for multiplexed detection of antibodies specific to different antigens. Next, human sera including from healthy control, COVID+ and vaccine recipient (Pfizer BNT162b2, S mRNA) donor with no known prior COVID‐19 infection were tested using this microchip. As shown in Figure 5d, healthy serum showed dense silver metallization with low resistance only for the +ve CTRL spot while the others remained open circuits. COVID+ serum showed dense silver metallization with low resistance for the N, S, and +ve CTRL spots and open circuit for −ve CTRL (Figure 5e). This matches prior results showing development, upon SARS‐CoV‐2 infection, of IgG antibodies against a wide variety of viral antigens including N and S.^{[ 44} , ^{45 ]} The vaccine‐recipient but uninfected serum sample with no known prior infection with COVID‐19 showed dense metallization and low resistance for wells with S and +ve CTRL proteins and open circuit for wells with N and ‐ve CTRL (Figure 5f). This is as expected since the BNT162b2 vaccine is known to create an antibody response only against the S protein.^{[ 46} , ⁴⁷ , ⁴⁸ , ^{49 ]} Thus, this multiplexed microchip assay successfully measures, from a single drop of serum sample, an antigen‐specific antibody fingerprint that can differentiate between healthy, COVID+, and vaccinated samples using two different antigens and a probe against human IgG.

Figure 5.

Multiplexed detection of multiple SARS‐CoV‐2 viral antigen‐specific antibodies. a) Assay steps for detection of human IgG against S and N antigens. Testing specificity of the assay using b) S monoclonal antibody and c) N monoclonal antibody. Differentiation between d) healthy control, e) COVID+, and f) vaccinated serum samples with the multiplexed microchip. Each data point is the mean of n = 2 replicates and error bars are standard deviations.

Multiplexed detection of different antibody isotypes (e.g., IgG and IgM) can provide a more comprehensive insight on the status of infection. IgM antibodies are known to be the first antibodies produced by immune response and began to decline at week 3 of the illness while IgG antibodies are produced later and are detectable for longer periods.^{[ 50 ]} For instance, a test with a positive response for IgM and negative response for IgG can be related to an acute and active immune response to a recent infection and the subject maybe be more likely spread disease to others.^{[ 51 ]} Here, we used the reconfigurability of our multiplexed microchip platform to develop an assay to detect both IgG and IgM response to N and S proteins (Figure  6a). The same group of clinical samples as earlier were tested again with this second assay format. The results of this are shown in Figures 6b–d. Healthy serum showed a negative signal (no silver metallization—and hence open circuit) for all four combinations of antigens and probes, as expected (Figure 6b). The COVID+ serum resulted in a positive signal (dense silver metallization and low resistance) for all combinations of antigens and probes meaning the serum sample was likely collected close to onset of infection (Figure 6c). Notably, this result matched the data available from the rapid IgG/IgM test of the sample (data not shown). The vaccine‐recipient sample showed positive signal only for S antigen and IgG probe (Figure 6d). Given that the sample was collected 3 months post vaccination, this result is as expected as well.^{[ 52 ]} In summary, this multiplexed assay platform effectively measured a multi‐antigen multi‐isotype comprehensive antibody signature that not only differentiated between healthy, COVID+, and vaccine samples but also provided insight about the status of infection for the COVID+ sample.

Figure 6.

Second platform of multiplexing for detection of IgG and IgM antibodies against S and N antigens. a) Assay steps for multiplexed detection of IgG and IgM antibodies against S and N antigens. Differentiation between b) healthy control, c) COVID+, and d) vaccinated samples. e) Portable and hand‐held impedance analyzer for on‐site signal readout. f) Interior design of the hand‐held impedance analyzer. g) Smart‐phone application for data display and analysis of the multiplexed assay. Each data point is the mean of n = 2 replicates and error bars are standard deviations.

Finally, in pursuit of the goal for development of a portable POC diagnostic device, we developed a hand‐held impedance analyzer that can perform multifrequency impedance measurement of the multiplexed chip (Figure 6e and Figure S7, Supporting Information). A smartphone application, based on the Android platform, was developed which communicates with the impedance analyzer via Bluetooth and acquires, plots, stores, and communicates the data to cloud storage platforms (Figure 6f). This enables the user to perform measurements through the app while the multiplexed chip is connected to the impedance analyzer. The app also shows the measured impedance values as numerical values in a bar graph or as an impedance heat map overlaid on the layout of the multiplexed chip (Figure 6g).

3. Conclusion

We have presented here a sensitive, multiplexed electronic detection technique for biomarkers by developing a novel transduction method that converts an analyte concentration to an enhanced enzymatically amplified silver metallization on nanostructured catalytic surfaces which is electrically conductive. We discovered a novel synergistic effect of surface‐bound or probe‐bound AuNPs with enzymatic probe activity which enhances silver metallization significantly. We applied this novel transduction technique for quantitative and high throughput measurements of viral antigen‐specific antibodies from COVID‐19 convalescent patient sera. Further, we showed that the enhanced metallization was localized and can be applied, in an electronic microarray format, for multiplexed detection of comprehensive COVID‐19 antibody fingerprints specific to disease state or vaccine response from a single small volume (<1 µL) serum sample. We conducted this work with serum samples here primarily because of logistical reasons but we hypothesize that this platform can be easily extended to perform assays from whole blood as well. The noise signal caused by the nonspecific binding, if any, of cells to the interface is expected to be much lower compared to the conductive silver. In addition, the frequency of impedance measurement can also be tuned or impedance spectra over multiple frequencies can be measured in order to optimize the signal to background ratio.

While this technique already demonstrated enough sensitivity and specificity to compare favorably with the gold standard technique for this application, the novel interaction of nanostructured catalytic surfaces and enzymes reported here, bears further optimization for further enhancement of sensitivity and applicability of detection by investigating other sizes and shapes of AuNPs or other nanomaterials. In addition, the multiplexable nature of this technique, demonstrated here using a 4‐plex assay, also provides the opportunity for development of fully integrated POC automated electronic microarray systems for massively multiplexed detection of larger numbers of biomarkers. Multiplexed and quantitative biomarker measurements can enable more accurate diagnostic and prognostic monitoring in several contexts. In the context of COVID‐19 vaccine efficacy, it has been shown that neutralizing antibody titer levels, measuring which requires a quantitative assay, are a key correlate of protection due to the vaccine. The multiplexed assays can enable simultaneous measurement of vaccine efficacy against several SARS‐CoV‐2 variants. Also, we and others have recently shown that levels of specific subclasses of antibodies and their specific glycosylation and effector‐binding profiles, when measured in a multiplexed and quantitative fashion, can be used to predict the outcome in severe COVID‐19.^{[ 9 ]} Further, given the general applicability of the transduction technique to any binding‐based assay, detection of COVID‐19 neutralizing antibodies to monitor vaccine efficacy and detection of other biomarkers such as antigens, DNA, and even whole pathogens or cells for diagnosis and monitoring of infectious or other diseases are among a few of the many useful future applications possible.

4. Experimental Section

Materials

Sodium hydroxide (S5881), poly‐l‐lysine solution (P8920), bovine Serum Albumin (A7030), streptavidin−gold (S9059), and 10 nm gold nanoparticles (752 584) were obtained from Sigma Aldrich. Reagent alcohol (BDH1156) and Tween 20 (97062‐332) were obtained from VWR. Deionized (DI) water (Lab Chem LC267405) and phosphate buffer saline (PBS, 21‐040‐CVR) were obtained from Fisher Scientific. SARS‐CoV‐2 (2019‐nCoV) spike recombinant protein (IT‐002‐032p) was obtained from Immunetech. SARS‐CoV‐2 (2019‐nCoV) nucleocapsid‐His recombinant protein (40588‐V07E) was obtained from SinoBiological. Recombinant Protein A/G (6502‐1) was obtained from BioVision. Pierce Bovine Serum Albumin, Biotinylated (29 130), and HRP‐Conjugated Streptavidin (N100) were obtained from ThermoFisher. Goat Anti‐Human IgG H&L (10 nm gold, ab39596), Goat Anti‐Human IgG H&L (HRP, ab97175), and Recombinant Anti‐SARS‐CoV‐2 Spike Glycoprotein S1 antibody (CR3022, ab273073) were obtained from Abcam. Mouse Anti‐Human IgM‐HRP (9020‐05) and Mouse Anti‐Human IgG Fc‐HRP (9040‐05) were obtained from Southern Biotech. Anti‐covid‐19 Nucleocapsid‐mAb (AB02325‐10.0) was obtained from Absolute Antibody. EnzMet for General Research Applications (6010‐45ML) was obtained from CedarLane.

Clinical Samples

Serum samples were obtained from commercial vendors: Ray Biotech Inc and Innovative Research Inc. Both vendors obtained serum from donors with informed consent after IRB approval (PROTOCOL NO: SOP‐TF‐PH‐002 STERLING IRB ID: 8291‐BZhang). Details are on file and available with them.

Microfabrication of µIDEs

Glass wafers (100 mm diameter, 0.5 mm thickness) were cleaned in acetone and isopropyl alcohol for 5 min each, then cleaned in piranha solution for 20 min. Photoresist (NR‐9 1500PY, Futurrex) was spin coated and baked at 150 °C for 2.5 min. Standard photolithography process including UV exposure (Karl Suss MA‐6), baking (100 °C, 1 min) and development (RD‐6, Futurrex) was done. After 30 s of plasma de‐scum (Vision 320 RIE, Plasma‐therm), 10 nm of titanium followed by 100 nm of gold was deposited via e‐beam evaporation (Denton Explorer, Denton). Liftoff was performed by sonication in acetone for 5 min, and then the chips were rinsed with isopropyl alcohol and diced using a dicing saw (DISCO).

PLL Coating Process

Microfabricated chips were cleaned with 10% NaOH/60% reagent alcohol in DI water for 2 h followed by rinsing with DI water thoroughly. Next, chips were dipped in 30% PLL in 30 × 10⁻³ m PBS for 30 min. Then, chips were rinsed with DI water and dried with centrifuging for 1 min. PLL‐coated chips were stored under vacuum in a desiccator at room temperature.

PDMS Preparation

A PDMS film with a thickness of 0.1 mm was laser cut with a pattern of wells for each specific design of microfabricated chip and then soaked in a solution of 5% alconox in DI water for 30 min followed by rinsing with DI water. After air drying the PDMS films, scotch tape was used to remove any remaining dust or particles before visually aligning and reversibly sealing with chips.

Biotin‐BSA HRP‐SA Model Assay

An amount of 1 mg mL⁻¹ biotin‐BSA in PBS was added to each well and incubated for 1 h. All incubations were performed in a humidified chamber at room temperature unless otherwise specified. Next, the chip was blocked with 1% BSA in 0.1% Tween20 in PBS (0.1% PBST) for 30 min and washed with 0.1% PBST and PBS. All washing steps were performed by placing the chip in a 10 cm Petri dish filled with washing buffer on a plate shaker at 55 RPM for 10 min. Then, the chip was dipped in DI water and centrifuge‐dried for 1 min. Probe solutions, consisting of [1:400] HRP‐SA, or [1:3] GNP‐SA, or [1:400] HRP‐SA and [1:3] GNP‐SA were prepared in 1.5 mg mL⁻¹ BSA in 0.05% PBST. Next, probe added to each well and incubated for 1 h. After two washes with 0.1% PBST and one wash with PBS, the chip was dipped in DI water and centrifuge‐dried for 1 min.

Serum Sample Dilution and Addition for Immunoassays

The custom laser‐cut PDMS microwells used here hold between 1.5–3 µL each to ensure full surface coverage of the well and to avoid overflow. For all serum‐based assay results reported here (Figures 3, 6, 5), a minimum sample dilution of 1:100 was used here, except for the serial dilution curve shown in Figure 2c which starts at a 1:10 serum dilution. To create the 1:100 diluted sample, 0.1 µL of each serum sample was initially diluted with 9.9 µL of assay buffer (of 0.01% BSA in 0.05% PBST) in a separate tube. Then 3 µL each of this diluted serum sample was directly added into the microwells.

Silver Metallization and Impedance Measurements

Figure S2 (Supporting Information) describes steps for deposition of silver metal using the three‐part EnzMet substrate solutions (A, B, and C). Briefly, equal volumes of each substrate component A, B, and C were sequentially added and incubated for 4, 4, and 8 min, respectively. Upon completion of C development time (Figure S2, Supporting Information), chips were dipped in DI water to stop the silver metallization reaction and spin dry to record dry phase impedance measurement of deposited silver metal. Authors found that optimization of C development time was crucial (specially in presence of AuNPs) for increasing signal to background ratio (Figure S4, Supporting Information). Impedance measurements were performed by measuring impedance across each pair of IDEs using the custom‐built portable impedance analyzer (Figure 6e–g) which uses an applied voltage of 100 mV at a frequency of 10 000 Hz.

Integration of AuNPs in Immunoassay

AuNPs were integrated in the assay of COVID‐19 either by mixing HRP‐labeled probes with AuNP‐labeled probes or via direct immobilization of AuNPs along with the antigen on the surface. For the first scheme, a solution of 50 µg mL⁻¹ of each COVID‐19 antigen (S, N) or control proteins (BSA, AG) were prepared in PBS and added to each well and incubated overnight at 4 C. Next, the chip was blocked, washed, and dried as above. 0.1 µL of each serum sample was diluted by 9.9 µL of 0.01% BSA in 0.05% PBST in a sperate PCR tube. Then, 3 µL of each diluted serum sample was added to their corresponding wells.

Then, samples were added to the wells and incubated for 1 h, followed by washing and drying again. A probe solution consisting of [1:400] goat HRP‐anti human IgG and [1:6] goat AuNP‐anti human IgG in 1.5 mg mL⁻¹ BSA in 0.05% PBST was prepared and this mixture was added to each well. After 1 h incubation, the chip was washed, dried, and the silver metallization step was completed as above.

For the second scheme, [1:60] AuNPs in 0.5 × 10⁻³ m PBS was added to each well and incubated in for 1 h. Next, the chip was washed twice with PBS, dipped in DI water, and centrifuge‐dried. Antigen immobilization and sample incubation was performed as above. A probe solution consisting of [1:800] mouse HRP‐anti human IgG in 1.5 mg mL⁻¹ BSA in 0.05% PBST was prepared and added to each well. After 1 h incubation, the chip was washed and dried. Then, the silver metallization reaction was completed.

Statistical Analysis

Data presented in this work represent the mean and error bars represent the standard deviation of technical replicates (n = 2–5 technical replicates as indicated in the appropriate figure captions) conducted with separate drops of the same serum sample, on separate µIDEs. LOD was calculated by multiplying the standard deviation of the negative PBS control by three and divided by the slope of calibration curve. Prism software was used for plotting data and ROC analysis.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Rafat N., Zhang H., Rudge J., Kim Y. N., Peddireddy S. P., Das N., Sarkar A., Enhanced Enzymatically Amplified Metallization on Nanostructured Surfaces for Multiplexed Point‐of‐Care Electrical Detection of COVID‐19 Biomarkers. Small 2022, 2203309. 10.1002/smll.202203309

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.