Teaching an old dog new tricks: A lipid membrane-based electric immunosensor for real-time probing of the spike S1 protein subunit from SARS-CoV-2.

Fast, cheap, and easy to implement point-of-care testing for various pathogens constituted a game changer in past years due to its potential for early disease diagnosis. Herein, we report on the proof-of-concept of a simple method enabling in vitro detection of a structural spike protein subunit from the SARS-CoV-2 (S1 ) in aqueous samples. At the core of this discovery lies the well-known paradigm of monitoring the capacitive current across a reconstituted zwitterionic lipid membrane subjected to a periodic transmembrane potential, followed by the real-time spectral analysis enabling the extraction of the second harmonic of the capacitive current. Subsequent changes in the amplitude of this harmonic recorded during lipid membrane-S1 interactions were correlated with alterations induced in the inner membrane potential profile by the S1 protein subunit adsorption, and were shown to be augmented by ionic strength, the presence of a specific monoclonal antibody designed against the S1 subunit and the angiotensin-converting enzyme 2 (ACE2) protein receptor, and uninhibited by the presence of other human serum proteins.

coronavirus disease 2019

severe acute respiratory syndrome coronavirus 2

angiotensin‐converting enzyme 2

receptor‐binding domain

SARS‐CoV‐2 S1 subunit

Middle East Respiratory Syndrome

MERS‐CoV spike S1 subunit

potassium chloride

human serum

sodium dodecyl sulfate

hexadecane and 4‐(2‐hydroxyethyl)‐1‐piperazineethanesulfonic acid

1,2‐diphytanoyl‐sn‐glycerophosphocholine

bilayer lipid membrane

fast Fourier transform

European Bioinformatics Institute

INTRODUCTION

Coronavirus disease 2019 (COVID‐19) is a highly contagious infectious disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2), and since its first report in December 2019, it has spread worldwide reaching 154,815,600 cases by May 6, 2021 [1] and became a devastating pandemic [2]. Early accurate diagnosis is crucial for controlling the pandemic, and a wide number of assays ranging from molecular, serological, and immunological‐based approaches (e.g., reverse transcription–polymerase chain reaction, cutting‐edge CRISPR‐related methods) to classical diagnostic methods (e.g., chest CT scans) have helped scientists and physicians to screen and diagnose SARS‐CoV‐2 infection [3].

These testing methods still pose challenges in field practice mainly due to the high cost and intricacies related to sample‐preparation, expensive instrumentation facility, and skilled personnel needed. For the convenient, accurate, and rapid diagnosis of SARSCoV‐2 infection, the deployment of real‐time, rapid testing kits, and point of care tests freed from the need of expensive equipment remains critical. For such tasks and based on the cited examples, the four structural proteins of SARS‐CoV‐2, namely the spike surface glycoprotein (S), small envelope protein (E), matrix protein (M), and nucleocapsid protein (N), as well as their gene sequences and antibodies, could be used as targets for detection. The coronavirus entry into host cells and subsequent human pathogenicity is mediated by its 150 kDa transmembrane spike (S), heavily N‐linked glycosylated homotrimer glycoprotein, comprising two functional subunits responsible for binding to the host cell receptor (S1 subunit) with the receptor‐binding domain (RBD) and fusion of the viral and cellular membranes (S2 subunit) [4, 5, 6]. It is also known that for cell entry, SARS‐CoV‐2 specifically targets the transmembrane human angiotensin‐converting enzyme 2 (ACE2) [7, 8], with an affinity in the low nanomolar range [9]. As a viable alternative to the detection methods noted above, electrochemical biosensing platforms comprising bio‐recognition elements (e.g., specific antibodies) and signal transducers enabled the direct detection of various respiratory viruses [10] including SARS‐CoV‐2 [11].

In a recent effort made from our group toward with the established nanopore system [12, 13], we showed that the S1 subunit of the SARS‐CoV‐2 spike protein attaches to and permeabilizes lipid bilayer membranes even in the absence of a specific receptor protein [14]. Inspired from previous work on membrane‐active peptides [15, 16], we hypothesized that prior to permeabilization, the lipid membrane adsorption of the S1 subunit from the SARS‐CoV‐2 spike protein alters the distribution of mobile ions, surface, and polarization charges on the lipid membrane, all leading to changes in the electrical features the membrane, for example, the surface and/or dipole potentials. Here, by exploiting the dependence of the lipid membrane capacitance upon the potential difference across its hydrophobic core, we demonstrate that such changes are readily seen through monitoring the capacitive current across the lipid membrane substrate. Finally, we assemble our findings and propose a deceivingly simple, yet effective biosensing platform that allows for direct detection and recognition of the S1 protein, based on its specific interaction with a designed monoclonal antibody and the ACE2 receptor, through the response elicited by such complexes on the membrane potentials of a reconstituted planar lipid membrane.

EXPERIMENTAL METHOD AND REAGENTS

The specific SARS‐CoV 2 proteins used herein were SARS‐CoV‐2 Spike S1‐His Recombinant Protein (cat. #40591‐V08H), SARS‐CoV‐2 Spike S1 Antibody Rabbit Mab (cat. #40150‐R007), MERS‐CoV Spike S1 protein (cat. #40069‐V08H), and recombinant human ACE2 (His Tag) (cat. #10108‐H08H), all purchased from Sino Biological Europe GmbH (Germany). Potassium chloride (KCl), human serum (HS) (cat. #H4522), sodium dodecyl sulfate (SDS), ultrapure water, n‐pentane, hexadecane, and 4‐(2‐hydroxyethyl)‐1‐piperazineethanesulfonic acid (HEPES) were bought from Sigma–Aldrich (Germany). The 1,2‐diphytanoyl‐sn‐glycerophosphocholine (DPhPC) lipids were procured from Avanti Polar Lipids (Alabaster, AL, USA).

Significance Statement

A simple method enabling in‐vitro detection of a structural spike protein subunit from the SARS‐CoV‐2 (S1) in aqueous samples is highly desirable.

The amplitude of the second harmonic of the capacitive current through a lipid membrane substrate subjected to a periodic potential difference is correlated with alterations induced in the membrane potential profile by the S1 protein subunit adsorption.

Such changes are augmented by ionic strength, the presence of a specific monoclonal antibody designed against the S1 subunit and the human angiotensin‐converting enzyme 2 protein receptor (ACE2), and enable S1 detection in the nanomolar range.

Electrophysiology experiments

The symmetrical membrane system used herein was obtained from DPhPC lipids dissolved in HPLC‐grade n‐pentane (10 mg mL−1) as described before [14, 17]. In short, the lipid bilayer (BLM, bilayer lipid membrane) was formed on a pre‐treated with 10% v/v hexadecane in highly purified n‐pentane aperture of ∼120 μm diameter in a 25 μm thick Teflon film (Goodfellow, Malvern, MA, USA) clamped between two chambers each of 1 mL volume (filled with the electrolyte containing 0.1 M or 2 M KCl buffered at a pH value of ∼6.3 in 10 mM HEPES). The recording cell was housed in a Faraday cage (Warner Instruments, USA) and placed on a vibration‐free platform (BenchMate 2210, Warner Instruments, USA). All experiments were performed at a room temperature of ∼25°C, and all protein stock solutions were kept at −20°C when not used. The working principle used for the automated real‐time monitoring and evaluating of the capacitive current through the lipid bilayer was based on the well‐established inner field compensation (IFC) method, as previously described [18]. Briefly, we constructed a virtual instrument to collect the capacitive current via Ag/AgCl electrodes coupled with a Multiclamp 700B computer‐controlled amplifier or Axopatch 200B instrument (Molecular Devices, USA). Data acquisition was performed with an NI PCI 6221, 16‐bit acquisition board (National Instruments, USA) at a sampling frequency of 10 kHz within the LabVIEW 8.20 environment and within the same virtual instrument, we employed fast Fourier transform (FFT) spectral analysis on the recorded capacitive current (I C) through a lipid membrane subjected to a sinusoidal time‐varying potential difference superimposed on a dc bias u 0 (, with ʋ = 420 Hz and u 1 = 50 mV)—applied from the A/D D/A acquisition card, and digitally extracted and analyzed the power‐spectrum‐derived amplitude of the harmonic component at the 840 Hz (denoted herein as I 2). The values reflecting time changes in the I 2 amplitude as reported in the main paper were computed and reported as generated within the LabVIEW virtual instrument used throughout. To monitor how the difference in the dipole and surface potential of the membrane unfolds in time as a result of specific interactions with the molecules of interest (i.e., S1, ACE2, MERS S1, or S1–S1 antibody immunological complex added on the grounded cis chamber), the dc bias of the applied voltage signal to the membrane was kept constant () and the time‐evolution of the I 2 was later analyzed. The data handling and graphic representations of the recorded data were done using Origin 6 (OriginLab, Northampton, MA, USA).

RESULTS AND DISCUSSION

Physical principle of the detection system

As illustrated in Figure 1A and demonstrated previously [18, 19], in assessing the physical response of a lipid membrane to an applied potential difference (Figure 1B), two considerations are crucial: (i) due to electrostriction effects on an equivalent “elastic” capacitor that models to a first approximation a reconstituted lipid membrane, the membrane capacitance (C) depends quadratically upon the sensed voltage difference (ΔV H) across its hydrophobic core ( (C 0 represents for the minimum value of the membrane capacitance at ΔV H = 0 and α [V −2] is a constant (Supporting Information); (ii) the actual ΔV H embodies three main contributions, arising from the externally applied potential difference across the lipid membrane (ΔV ext), the difference on the lipid membrane surface potentials (V S) measured on both sides of the membrane, and respectively the difference in the membrane dipole potentials (ΔV D) [20, 21, 22, 23], as measured between the cis and trans monolayers of the membrane.

FIGURE 1

(A) Schematic diagram showing the Gouy–Chapman model for the planar lipid membrane surface (V S) and dipole potential (ΔV D) profiles [19]. In the symmetrical case, at ΔV ext = 0 (a), the potential difference across the hydrophobic core ΔV H = 0. (b) For the case of an asymmetrical membrane clamped at ΔV ext ≠ 0, the net ΔV H is also influenced by the asymmetry in the V S and ΔV D on the cis and trans monolayers. (c) Within the elastic capacitor description of the membrane, a periodic ΔV ext results in a time‐varying capacitive current (I C) comprising three harmonics (ω, 2ω, 3ω) of the ΔVex’s pulsation (ω) (see also Figure 2A). (B) Simplified schematics of the experimental setup used herein to apply voltages (ΔV ext) to the lipid membrane and record ensuing capacitive currents (I C). The dashed‐dotted rectangle illustrates the I–V head‐stage of the amplifier, fully controlled by the A/D D/A acquisition card via a personal computer (PC)

FIGURE 2

(A) Typical representation of the power‐spectra data on the capacitive current (I C) measured in the “open air” (no lipid membrane formed) and respectively after the successful formation of a lipid membrane, subjected to a periodic ΔV ext with zero dc bias (u 0) . In the latter case (“lipid membrane formed”), besides the fundamental harmonic measured at 420 Hz, the spectral analysis reveals the presence of two supplementary harmonics measured at 840 Hz (peak denoted herein by I 2) and 1260 Hz (peak denoted herein by I 3) (see the zoomed‐in inset, the dashed encircled areas). The fact that the second harmonic (I 2) is non‐zero at zero dc bias (u 0 = 0) (Supporting Information, “Theoretical account for the electric response of an elastic lipid membrane bilayer”), reflects among others an existing offset between the Ag/AgCl electrodes potentials, equivalent to applying a distinct from nil potential difference across the membrane. (B) The power‐spectrum amplitude changes of the second harmonic (I 2) from the capacitive current (I C) measured across a lipid membrane subjected to a periodic ΔV ext with variable dc bias (u 0) (at u 0 = −50, −100, and −150 mV, the percent increase of I 2 was 25%, 140%, and 308%, respectively), with dc bias u 0 = 0 but exposed asymmetrically (cis side only) to SDS (C) (440% increase of I 2 after addition SDS [25 μM] and respectively 740% after addition SDS [50 μM]), and at constant dc bias (u 0 = −150 mV) but exposed asymmetrically (cis side only) to human serum (HS 1%) (D). HS, human serum; SDS, sodium dodecyl sulfate

From elementary electrostatic description and considering the case of an electrically asymmetric membrane, namely surface and dipole potentials assume different values on its cis and trans sides (i.e., and ), it is true that while clamped at a particular external potential difference (), the potential difference across the membrane's hydrophobic core writes (Figure 1Aa,b).

If for a such model lipid membrane, the externally applied potential difference consists of a constant term (u 0) and a sinusoidal component with amplitude (u 1) and pulsation (ω) , by virtue of elementary circuit analysis, it follows that the resulting time‐dependent capacitive current (I C(t)) embodies three harmonics of the fundamental pulsation (ω) (Supporting Information). For our analysis we focused solely on the second harmonic isolated from the power‐spectra of the capacitive current (Figure 2A), which in the time‐domain and with the notations employed above writes: . In this expression, the term δ stands for the lumped value quantifying the asymmetry in the surface and dipole potential of the lipid membrane (Supporting Information). It then follows that if the adsorption of certain membrane active molecules induces changes determines an asymmetry of the surface or dipole potentials (or both), the analyte detection can be easily achieved through monitoring the amplitude second harmonic in the capacitive current across the membrane. Moreover, the time course of this harmonic amplitude would correlate with the adsorption kinetics of the membrane active molecules under study.

To test our system, in a first set of control experiments we maintained the symmetry of a reconstituted lipid membrane and increased in a stepwise manner the value of constant term (u 0) of the ΔV ext. As expected, the FFT spectral analysis of the resulting I C clearly indicated a corresponding stepwise increase in the power‐spectral amplitude of the second harmonic of the capacitive current, called herein I 2 (Figure 2B).

Then, we maintained fixed values on the amplitude components of the ΔV ext but added on one side of the membrane SDS, an anionic amphiphile whose adsorption at the lipid membrane interface induces a negative surface charge density σ and alters the surface potential V S [mV] of the membrane only of the addition side of the amphiphile, according to [20, 24]. In this formula, C and z refer to the molar concentration and valence of the ith salt species in the bath, R is the general gas constant, F the Faraday's constant, T the absolute temperature, and and are the relative permittivity of the electrolyte and vacuum permittivity, respectively. By virtue of principles underlined above, increasing amounts of the SDS led to proportional augmentations in the amplitude of the I 2 (Figure 2C). Unlike the previous case, the time‐dependent unfolding of such changes is distinct from a stepwise‐like pattern and correlates with the time‐dependent, diffusion‐controlled SDS adsorption kinetics on the membrane interface. As a supplementary test for the detection selectivity in physiologically relevant samples, one‐side addition of 1 % HS under fixed ΔV ext left the electric characteristics of the lipid membrane largely unaffected, as judged from non‐significant changes in the I 2 (Figure 2D). At this point, we firmly concluded that a simple testing system as employed herein may indeed be capable of accurate real‐time monitoring of the presence of a soluble analyte, if its adsorption on the lipid membrane alters its electric properties.

Detection of the S1 subunit protein in buffer

In Figure 3, we present typical traces reflecting changes in the amplitude of the I 2 across a reconstituted lipid membrane, following asymmetric addition (cis side only) of nM amounts of the SARS‐CoV‐2 S1 subunit (S1), which we already established that adsorbs to the membrane [14]. As seen, incremental concentrations of the added S1 determine a correspondingly increase in the I 2 amplitude. According to the manufacturer's information, the pI value of the S1 subunit is 8.25, so that around pH 6.3 as used herein it carries a net positive charge. This suggests that increasing amplitudes of I 2 reflect incremental accumulation of a net positive charge on the membrane interface and a consequent cis side surface potential (V S( )) bias toward more positive values. Notably, time‐resolved traces shown in Figure 3 reflect in a real‐time manner the lipid membrane‐S1 adsorption kinetics. From a simple electrical perspective, at fixed ΔV ext and similar and unaltered membrane dipole potentials (ΔV D) on both membrane monolayers, positive changes in the V S manifested solely on the S1 addition side of the membrane interface, are equivalent with an increase in the net potential difference sensed across the membrane's hydrophobic core (ΔV H), equivalent with a membrane hyperpolarization (Figure 1Ab, red line). This is in agreement with the control experiments as shown in Figure 2A, in which larger potential differences across the membrane translate into correspondingly larger I 2 amplitudes.

FIGURE 3

(A) When a lipid membrane is subjected to a periodic (), incremental asymmetrical additions of the S1 subunit (cis side) led to corresponding increasingly amplitudes of the I 2 current harmonic, quantified in (B). (C) In the presence of S1 antibody, the amplitude of the I 2 current harmonic increased further, as quantified in (D) (red bars; I 2 measured with the S1–S1 antibody complex and blue bars I 2 measured with the S1 alone)

Detection of the S1 protein subunit in the presence of specific binding substrates and human serum

For probing the detection specificity, our strategy involved the use of specific molecular substrates for the S1 subunit, namely a synthesized monoclonal antibody and the ACE2 receptor. As shown in Figure 3, following incremental addition of S1 and assessment of I 2 changes the final injection of a monoclonal antibody (S1 antibody) as to achieve a 1:4 (S1–S1 antibody) molar ratio, determined a further augmentation in the I 2 amplitude. In a versatility test, a similar effect was seen in other experiments carried out in similar conditions, where the MERS‐CoV spike S1 subunit (MERS S1) was employed instead (Figure S1). One should note that such effects are solely attributable for the resulting S1–S1 antibody complex generated in the buffer, as by itself the S1 antibody does not generate any distinguishable change in the I 2 amplitude (Figure S2). To rationalize the lack of effect generated by the S1 antibody alone, one must recall that according to the manufacturer, the S1 antibody's pI is 6.47 rendering almost void of net electrical charge around pH 6.3 as used herein, thus being less effectively—as compared to the S1 subunit—engaged in an electrophoretic‐driven association with a negatively biased lipid membrane.

To mimic clinical conditions, we assessed the sensing capacity of the system described herein in the presence of commercial human serum added into the recording chamber. This is a vital step for any possible extension of the presented system to a field‐deployable biosensor, since real human serum samples contain up to 104 proteins [25], which may target the lipid membrane and affect non‐specifically the sensitivity and specificity of detection. As we discovered, S1 protein detection via S1–S1 antibody complex formation and I 2 monitoring was unaffected by the presence of other blood proteins, judged from the fact that supplementary addition of 1% HS did not interfere with the detection process (Figure S3).

In another set of experiments aimed at testing the detection specificity, addition of the more affine ACE2 receptor at a molar ratio of 1:4 (S1–ACE2) resulted in an even better S1 detection response (Figure 4).

FIGURE 4

In the presence of ACE2, the I 2 amplitude from the capacitive current through the membrane bathed in a buffer containing S1 [60 nM] and in response to (), increased by 43.1 ± 9% (A), whereas a similar concentration of the S1 antibody led to a smaller increase of 28.5 ± 6.1% (B). ACE2, angiotensin‐converting enzyme 2

To interpret such increases in the I 2 amplitude through the Gouy–Chapman theory correlating surface potential (V S) changes with the adsorbed analyte‐induced surface charge density (σ) modifications [20], a paradox arises. Knowing that at pH 6.3 as used herein, S1 antibody is almost devoid of electric charge whereas the ACE2 receptor protein is negatively charged (S1 antibody's pI = 6.47, according to the manufacturer and ACE2 receptor's pI = 5.8, calculated with the EMBL‐EBI search and sequence analysis tools APIs), S1–S1 antibody and S1–ACE2 complexes are expected to bear an almost unchanged (S1–S1 antibody) to a lesser positive charge (S1–ACE2) as compared to the free S1 protein. By comparison to the situation when S1 was present alone in the electrolyte, it is then expected that S1–ACE2 would alter to a lesser extent the membrane surface charge density upon adsorption or leave it largely unchanged (S1–S1 antibody), thus entailing smaller to none changes in the resulting ΔV H and the recorded I 2 amplitude, respectively (Figure 1Ab). This on the other hand is in stark contrast with the experimental findings presented above (Figures 3 and 4). The apparent paradox is further deepened by another observation, according to which I 2 amplitude changes recorded following membrane interaction with either the S1 alone or in complexation with the S1 antibody are augmented in high salt electrolytes (Figure S4). This is again puzzling, as the mobile ions‐induced screening effect of the membrane electrostatics becomes prevalent in salt‐concentrated electrolytes. Namely, a high versus Low ionic strength buffer is expected to better shield changes in the surface potential and the ensuing intramembrane electric field caused by membrane adsorption of a similar number of charged analytes. In other words and contrary to our findings, the I 2 amplitude should in fact be augmented in a low salt electrolyte. At the present, we can only speculate with regard to the precise molecular mechanism through which such complexes alter the physical properties of the lipid membranes and in accord to our observations. A plausible explanation would be that S1–S1 antibody and S1–ACE2 complexes adsorption to the membrane will also increase the membrane dipole potential of the lipid monolayer where they adsorb to (ΔV D( ), Figure 1Ab, red lines). This is not unexpected, as previous data with other membrane insoluble analytes have demonstrated such an effect [20]. Moreover, such an effect would be facilitated in high ionic strength electrolytes—as shown herein, since the enhanced salt screening of the net electric charge on the complexes would in fact determine a more efficient accumulation of the complexes at the membrane surface or intercalation into the lipid matrix.

On the longer run, more experiments with different selections of lipid membrane‐forming lipids (e.g., charged lipids, cholesterol) and electrolyte pH may also shed more light into the physical mechanism(s) contributing to the effects seen, through additional possible changes involving membrane fluidity and packing.

CONCLUSION

In this report, we established a simple to operate and effective setup to specifically detect in a time‐resolved manner the SARS‐CoV‐2 S1 protein subunit in aqueous solution and presence of physiologically relevant molecules. The use of both monoclonal antibodies and the ACE2 receptor, as specific binding substrates, augmented the S1 detection. The method opens the perspective of fast and cheap detection of other S1 proteins containing receptor‐binding residues mutations, artificial or natural antibodies in the aqueous sample (e.g., IgG and IgM), testing for efficacy of therapeutics‐directed inhibitors (e.g., peptides) to the S proteins of SARS‐CoV or related viruses, quantification of the adsorption kinetics of the antigen‐antibody (receptor) complexes with a lipid membrane substrate of variable and controllable composition, or enable alternatives for monitoring of the interaction of viral antigens with selected protein targets, relevant for the discovery of decoy therapeutic proteins [26]. Despite its simplicity, we stress that the presented approach may evolve as a promising alternative, complementing established serological techniques [3]. In this context, each step toward the identification of target analytes should be optimized in terms of sensitivity, accuracy, and selectivity, starting from sample preparation from clinical sources, library construction of specific antibody/receptor candidates, and data processing. Prospective solutions to this end may include: (i) structure‐based design of highly efficacious antibodies/receptors directed at the targeted analyte; (ii) operating the detection in buffers with specific properties in terms of pH, ionic strength to increase both the propensity of target (antigen)‐receptor (antibody) complexation as well as subsequent interaction of the formed complex with the lipid membrane substrate. The extension of the method to a portable biosensing platform could be further facilitated by the use of lipid bilayer‐coated nanowires or nanofilms, as core sensing elements to detect the virus‐specific structural proteins or for screening suitable antibodies, and the implication of micro‐/nanoelectronics and integrated microfluidics.

Supporting Information

Click here for additional data file.