A novel rapid detection for SARS-CoV-2 spike 1 antigens using human angiotensin converting enzyme 2 (ACE2)

2.2. Biolayer interferometry

Fc-tag tagged human ACE2 (ACE2, Cat. No. 10108-H05H), SARS-CoV-2 spike monoclonal antibody (S1-mAb, Cat. No. 40150-R007), SARS-CoV-2 spike protein (S1 subunit, Cat. No. 40591-V08H), and SARS-CoV-2 RBD (Cat. No. 40592-V08B) were purchased from Sino Biological. CR3022 and F26G19 were purchased from Abclon (Seoul, Korea). Plasmids encoding heavy and light chains of each antibody at a 1:6 ratio were transiently cotransfected into 293-F cells using PEI reagent (PolyScience, PA, USA). Six days after transfection, the supernatant was collected, and CR3022 and F26G19 were purified on Protein A columns (GE Healthcare, IL, USA). Binding affinities between SARS-CoV-2 spike antigens (S1 and RBD) and four different antibodies (ACE2, CR3022, F26G19, and S1-mAb) were analyzed by biolayer interferometry (BLI) on a BLITz instrument (ForteBio, CA, USA). After hydration of the biosensor tip for 10 min, BLI analysis began, followed by five steps: initial baseline (30 s), sample loading (300 s), baseline (120 s), binding (300 s), and dissociation (300 s). All of the initial baseline, baseline, and dissociation processes were performed using sample diluent buffer (0.02% Tween 20, 150 mM NaCl, and 1 mg/mL BSA in 10 mM PBS with 0.05% sodium azide, pH 7.4). In the sample loading step, each antibody (100 μg/mL) was immobilized on a biosensor tip coated with Protein A through the specific interaction between Protein A and the antibody Fc domain. After the baseline step, four different antigen concentrates were incubated with the antibody-conjugated biosensor tip to measure the association between antigen and antibody. From the four different binding curves obtained from each antigen concentrate, binding constants were calculated based on the 1:1 binding model.

2.3. Preparation of antibody-conjugated CNBs

A 1% stock solution of red cellulose nanobeads (CNBs) was purchased from Asahi Kasei Fibers Corporation (NanoAct, Cat. No. RE2AA; Miyazaki, Japan); the beads have an average diameter of 345 nm. A CNB conjugation kit, containing conjugation buffer, blocking buffer, and wash buffer, was purchased from DCN Diagnostics (CA, USA). Conjugation of the SARS-CoV-2 spike antigen-specific antibodies (CR3022, F26G19, S1mAb) and binding protein (ACE2) to the surface of CNBs was performed according to the manufacturer's instructions. Briefly, 0.12 mL of 0.5 mg/mL antibody was mixed with 0.060 mL of 1% stock CNB and 0.12 mL conjugation buffer, and the mixture was incubated for 2 h at 37 °C. Then, 7.2 mL blocking buffer was added to the mixture, which was incubated for an additional hour at 37 °C. The solution was centrifuged (14,400×g, 20 min, 4 °C), and the pellet was resuspended in 7.2 mL wash buffer. After centrifugation (14,400×g, 20 min, 4 °C), the washed pellet was resuspended in 0.5 mL wash buffer. The final concentration of antibody-conjugated CNB was approximately 0.1%. The exact concentration of the CNB was calculated by measuring the maximum absorbance at 554 nm by UV–vis spectrophotometry (Synergy H1; BioTek).

2.4. Preparation of LFIA strips

LFIA strips consisted of a sample pad (Ahlstrom, Helsinki, Finland), a conjugate pad (Ahlstrom, Helsinki, Finland), a nitrocellulose membrane (Advanced Microdevices, Haryana, India), and an absorbent pad (Ahlstrom, Helsinki, Finland) as shown in Fig. 1 b. The conjugate pad was treated with 0.1% Triton X-100 (Cat. No. T8787; Sigma-Aldrich, MO, USA) before CNB fixation. After complete drying, a 0.05% solution of antibody-conjugated CNB in stabilizing buffer (10 mM 2-amino-2-methyl-1-propanol (pH 9.0), 0.5% BSA, 0.5% β-Lactose, 0.05% Triton X-100, and 0.05% sodium azide) was sprayed on the conjugate pad, followed by incubation for 1 h at 37 °C in a vacuum oven (FDU-1200, EYELA, Tokyo, Japan; JSVO-30T, JSR, Gongju, Korea). A test line containing the capture probe and a control line were dispensed onto the nitrocellulose membrane using a line dispenser (BTM Inc., Uiwang, Korea) under the following conditions: dispensing speed, 50 mm/s; dispensing rate, 1 μL/cm. ACE2 (1 mg/mL) and 0.5 mg/mL anti-IgG antibody mixture [1:1:1 (v/v/v) anti-human IgG antibody (Cat. No. I2136, Sigma-Aldrich)/anti-rabbit IgG antibody (Cat. No. R5506, Sigma-Aldrich)/anti-mouse IgG antibody (Cat. No. A4416, Sigma-Aldrich)] in sample diluent buffer (Cat. No. ab154873; Abcam, Cambridge, UK) were used for the formation of the test line and control line, respectively. After line dispensing, the nitrocellulose membrane was dried for 1 h at 37 °C. To decrease the non-specific interaction between capture probes in test lines and detection probes, the nitrocellulose membrane was treated with the blocking solution (10 mM 2-amino-2-methyl-1-propanol (pH 9.0), 0.5% BSA, 0.5% β-Lactose, 0.05% Triton X-100, 0.05% sodium azide) for 1 h in a vacuum oven (37 °C). Each component of the LFIA strip was precisely assembled and cut to a 38 mm width, followed by integration into a housing for a single LFIA test.

Open in a separate window

Fig. 1

Cellular receptor (ACE2)-based LFIA. a) Schematic of ACE2 receptor recognition by SARS-CoV-2. ACE2, a type 1 membrane protein expressed in the lung, heart, kidney, and intestine, is the cellular receptor for SARS-CoV-2. b) Schematic of an ACE2-based LFIA consisting of a sample pad, conjugate pad, nitrocellulose membrane, and absorbent pad. The test line placed on the nitrocellulose membrane contains ACE2 for detection of the SARS-CoV-2 spike antigen. Anti-IgG antibody is used in the control line. The proposed LFIA can achieve sensitive and selective detection of SARS-CoV-2 spike antigen within 20 min.

2.5. Dot-blot assay for the discovery of sandwich pairs

To discover the optimum pair for detection of the SARS-CoV-2 spike antigen, a total of 12 capture probe–detection probe pairs (Supplementary Table 1) were tested. The affinity of the four different antibodies for SARS-CoV-2 S1 and RBD was confirmed through ELISA, Western blot, and BLI. Capture probe (1 mg/mL) and 0.5 mg/mL anti-IgG antibody mixture [1:1:1 (v/v/v) anti-human IgG antibody/anti-rabbit IgG antibody/anti-mouse IgG antibody] were used to form test dots and control dots, respectively. Dots were immobilized onto a nitrocellulose membrane (Advanced Microdevices) by loading 0.5 μL of the test or control solution. The nitrocellulose membrane was incubated with blocking solution (10 mM 2-amino-2-methyl-1-propanol (pH 9.0), 0.5% BSA, 0.5% β-lactose, 0.05% Triton X-100, 0.05% sodium azide) in a vacuum oven (1 h, 37 °C). The LFIA strip for the dot-blot assay was constructed as described in the previous section. For comparative analysis between the 12 pairs, 50 ng target antigen (SARS-CoV-2 Spike S1) was incubated with each detection probe in running buffer [10 mM adenosine monophosphate (AMP, pH 9.0), 5 mM EDTA, 200 mM Urea, 1% Triton X-100, 0.5% Tween 20, 500 mM NaCl, 1% PEG (MW 200)] for 10 min at RT, followed by loading of the mixed sample onto the LFIA strip. After 20 min, red signals indicating antigen detection were confirmed by the naked eye and smartphone camera. In addition, all dots were quantitatively analyzed using an image analyzer (Sapphire Biomolecular Imager, Azure Biosystems, CA, USA), and the relative intensity of each dot was calculated based on the average intensity of the dot minus the background level (Supplementary Table 1).

2.6. Sensitivity and specificity analysis of LFIA

For the development of highly sensitive and specific LFIA for the detection of SARS-CoV-2, the concentration of CNB (Supplementary Fig. 1), the concentration of immobilized ACE2 (Supplementary Fig. 2), and the composition of the running buffer (Supplementary Fig. 3) were optimized. Finally, 0.05% of CNB for the detection probe, 1 mg/mL of ACE2 for the immobilized capture probes, and running buffer [10 mM AMP, (pH 9.0), 5 mM EDTA, 200 mM urea, 1% Triton X-100, 0.5% Tween 20, 500 mM NaCl, 1% PEG (MW 200)] were selected. The SARS-CoV-2 S1 and SARS-CoV-2 RBD antigens were serially diluted in sample diluent buffer (Cat. No. ab154873; Abcam), and diluted samples were mixed with running buffer at a ratio of 1:9 (v/v). The final concentrations of the diluted samples ranged from 500 to 5 ng/mL. One hundred microliters of running buffer containing each antigen concentrate were added to the inlet of the LFIA device. In this system, the sample flows along with the LFIA strip by capillary force and first encounters the antibody (S1-mAb)-conjugated CNB. The antigen in the sample is captured by the S1-mAb conjugated CNB, and this antigen-probe complex is detected by a pre-immobilized capture reporter (ACE2) on the nitrocellulose membrane. After 20 min of sample loading, the appearance of red color in the test and control lines is confirmed and analyzed on a Sapphire Biomolecular Imager.

For specificity testing, two different corona-related spike antigens (i.e., SARS-CoV S1, and MERS S1 antigens) were prepared at three different concentrations (100, 20, and 5 ng/mL). Three concentrates of each antigen were loaded to the LFIA device, and the line intensities were quantified with a portable line analyzer (Light-G; WellsBio, Seoul, Korea). A positive intensity at the test line is more than 50 by the manufacturer's recommendation of the portable analyzer. All experiments were performed three times, and the limit of detection in Fig. 5 c was calculated as the mean value of the negative control plus three times the standard deviation.

Open in a separate window

Fig. 5

Sensitivity and specificity of the ACE2-based LFA a) Results of ACE2-based LFA for the detection sensitivity of SARS-CoV-2 S1 antigen. Serially diluted antigen concentrates (concentration range: 500 ng/mL to 5 ng/mL) were tested by ACE2-based LFA. After 20 min, the LFA strips were photographed with a smartphone. Moreover, the intensity of the test and control lines was converted to a peak histogram by an image analyzer. b) Results of the comparative analysis for the detection selectivity: positive control – SARS-CoV-2 S1, negative control – SARS-CoV S1, MERS S1, and buffer solution. Using three different concentrates (1 μg/mL, 200 ng/mL, and 50 ng/mL) of the antigen sample, the detection performance of ACE2-based LFA was demonstrated. c) Bar graph of peak intensities for test lines. After 20 min for the sample flow, the intensity of the test lines was measured by a portable line analyzer. Inset) the detection intensity for the 5 ng antigen per reaction of each control. Limit of detection (LOD) was determined by the mean value of negative controls plus three times the standard deviation. P-values: ns > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

2.7. Purification and quantification of copy number of human coronavirus-OC43 (HCoV-OC43) and SARS-CoV-2

Purification of viral RNA from HCoV-OC43, γ-irradiated SARS-CoV-2 (Cat. No. NR-52287, BEI Resources, VA, USA) was performed using a QIAamp Viral RNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Serial dilution of viral RNA is targeted for nucleocapsid (N) gene of HCoV-OC43 (1.16 × 10¹²-10⁰ copies/μL), or envelope (Env) gene of SARS-CoV-2 (7.7 × 10⁶-10⁰ copies/μL) served as standards, and purified viral RNA was synthesized to complementary DNA, consistent with performing quantitative reverse transcription PCR (RT-qPCR) using the Luna® Universal Probe One-Step RT-qPCR Kit (New England BioLabs, MA, USA) according to manufacturer's protocol by following primers for HCov-OC43 gene: Forward 5′-AGC AAC CAG GCT GAT GTC AAT ACC Reverse 5′-AGC AGA CCT TCC TGA GCC TTC AAT, SARS-CoV-2 Env gene: Forward 5′-ACA GGT ACG TTA ATA GTT AAT AGC GT, Reverse 5′-ATA TTG CAG CAG TAC GCA CAC A, and specific probe (ACA CTA GCC ATC CTT ACT GCG CTT CG) labeled with FAM (6-carboxyfluorescein) and BHQ-1 (Back Hole Quencher-1). The single-step RT-qPCR was set for reverse transcription under 55 °C for 10 min, and amplification of 45 cycles under 95 °C for 10s, 60 °C for the 30s. The reaction was analyzed using the Bio-Rad CFX 96 Touch Real-Time PCR System (CA, USA).

3.2. Identification of the sandwich pair for detection of the SARS-CoV-2 S1 antigen

Successful detection of the target antigen requires two types of probes, the capture probe and detection probe, which recognize different sites of a specific antigen. Under unpredictable circumstances like the current COVID-19 outbreak, however, the development of two different antibody pairs would be too time-consuming and expensive, and therefore could not be applied to rapid diagnosis in the field with the goal of preventing the spread of disease. To overcome this drawback of antibody production, we used the cellular receptor, human ACE2, whose binding affinity for target antigen is comparable to those of antibodies, instead of one of the antibodies.

We used a dot-blot assay to discover the pair for sandwich antigen detection. ACE2 and three commercially available antibodies (CR3022, F26G19, and S1-mAb) were immobilized on a nitrocellulose membrane (NC membrane) as capture probes. Then, the other proteins (antibodies or receptor) not used in the immobilization were conjugated with cellulose nanobeads (CNB) for signal generation as detection probes, and the detection performance was evaluated. As shown in Fig. 4a and b, a total of 12 pairs were used for detection of SARS-CoV-2 S1 antigen, and non-specific interactions between capture and detection probes were also assessed. Furthermore, the intensity of each dot was analyzed on an image scanner (Fig. 4c and Supplementary Table 1). When ACE2 was used as the capture probe, sandwich detection of the S1 antigen was successfully achieved with all three detection probes. The result implies that the ACE2 binding site of the SARS-CoV-2 S1 antigen does not overlap with the epitopes of antibodies. In addition, we observed no false-positive signals in any of the negative controls (i.e., without antigen).

Open in a separate window

Fig. 4

Identification of the sandwich pair for detection of SARS-CoV-2 spike antigen. a) Schematic diagram of LFIA using ACE2 as the capture probe and sandwich analysis results obtained from paired antibodies (CR3022, F26G19, and S1mAb). SARS-CoV-2 S1 antigen (50 ng) was used as a positive control, and buffer containing no S1 antigen was used as a negative control. After 20 min, the strips were captured by a smartphone, and their peak intensities were analyzed. b) Schematic diagram of LFIA, using antibodies as the capture probe, and their sandwich analysis results. c) Peak intensities of capture probe (P_C)–detection probe (P_D) pairs. A total of 12 pairs of positive controls (50 ng S1 antigen) were tested, and their intensities were analyzed. Peak intensity was calculated by subtracting the background intensity of the strip from the average intensities of the dots.

On the other hand, when antibody was used as a capture probe, false-positive signals were observed due to a non-specific interaction between the capture and detection antibodies. Notably, non-specific interactions occurred between S1-mAb and CR3022 and between S1-mAb and F26G19 (the red circle of the negative control in Fig. 4b). In addition, the epitope of CR3022 appeared to overlap with the epitope of F26G19, resulting in a decrease in the signal of the test dot. The results of this experiment emphasize that discovery of suitable antibody pairs for the sandwich assay is complicated by several variables. The introduction of ACE2 as a replacement for antibodies could accelerate the development of antigen diagnostic kits and facilitate efficient management of outbreaks. Moreover, overall signals, in both test and control dots, were diminished in the case of ACE2-conjugated CNBs (Fig. 4b and Supplementary Table 1). Evaluation of 12 capture–detection probe pairs allowed selection of the most suitable pair for the sensitive detection of SARS-CoV-2 S1 antigen: capture probe, ACE2, and detection probe, S1-mAb.

3.3. Sensitivity and specificity of ACE2-based LFIA for SARS-CoV-2 S1

The LFIA sensor strip consists of a sample pad, a conjugation pad, a nitrocellulose membrane, and an absorbent pad (Fig. 1b). The sample containing the target antigen is introduced onto the sample pad and sequentially encounters the CNBs dried on the conjugate pad. The CNBs have already been coated with detection antibody (S1-mAb) through hydrophobic and/or electrostatic interactions. The detection antibody-coated CNB captures the target SARS-CoV-2 S1 antigen and transfers it to the nitrocellulose membrane. The test line containing the capture probe (ACE2) detects the SARS-CoV-2 S1 antigen that was previously captured by the detection probes (S1-mAb-conjugated CNB), allowing sandwich detection of the SARS-CoV-2 S1 antigen. Meanwhile, the control line serves to determine whether the sample has flowed through and the biomolecule on the conjugate pad is active. For this purpose, anti-IgG Ab was used to capture all antibodies that were already conjugated with the CNBs. The test and control lines were formed on the nitrocellulose membrane using a line dispenser. As the detection of the target antigen on the test lines of LFIA, the red signal from the CNBs makes it possible to visually confirm whether the sample contains the target antigen or not. To avoid non-specific interaction between capture probes in test lines and detection probes, the nitrocellulose membrane was adequately treated with blocking solution. Unbound detection probes pass through the nitrocellulose membrane and ultimately reach the absorbent pad located at the end of the strip, which serves to maintain the capillary force that drives sample flow.

To demonstrate the detection capability of the ACE2-based LFIA (ACE2-LFIA), we performed the sensitivity analysis using serially diluted samples of SARS-CoV-2 specific antigen (S1 and RBD, concentration range: 5–500 ng/mL) as shown in Fig. 5a and Supplementary Figs. 7 and 8. Twenty minutes after sample loading, the window of the LFIA device was photographed using a smartphone, and the intensity of the test and control lines was analyzed with an image scanner and analyzer that converts the line color intensity to signal peaks, as shown in Fig. 5a. In the case of S1 detection, the detection signal gradually decreased as the dilution factor increased, but even if only 5 ng antigen was present in the sample, the signal was still present. In addition, we confirmed that the detection signal was clearly higher for RBD than for S1 for all diluted samples, and even 1 ng RBD present in the sample could be successfully detected by our ACE2-LFIA. In addition, we observed no false-positive signals in the negative control (i.e., the absence of target antigen). Cellulose nanobeads (CNB), a colorimetric label used in LFIA, have been reported to have higher sensitivity than colloidal gold (Sakurai et al., 2014). Hence, we also compared the detection sensitivities of colloidal gold and CNBs (Supplementary Fig. 9). Colloidal gold-based LFIA detected RBD antigen at concentrations as low as 20 ng, and was thus significantly less sensitive than CNB-based LFIA. Because CNBs have higher color intensity and surface area than colloidal gold in the same volume, CNB-based LFIA was approximately ten times more sensitive than colloidal gold-based LFIA, consistent with recently reported results (Sakurai et al., 2014).

Next, we evaluated the specificity of ACE2-LFIA using S1 antigens from other coronaviruses (SARS-CoV and MERS-CoV). Three different concentrations (100, 20, and 5 ng/reaction) of each S1 antigen were introduced into the LFIA device, and the intensity of test lines was measured after 20 min using a portable line analyzer. It took less than 10 s to measure the line intensity of each LFIA device, enabling rapid and accurate diagnosis in point-of-care or laboratory settings. Real images of the LFIA device and portable line analyzer are shown in Supplementary Fig. 10. The SARS-CoV-2 S1 (<5 ng antigen) and RBD (<1 ng antigen) were successfully detected by the LFIA device, even at low antigen concentration (Fig. 5a and Supplementary Fig. 8). The sensitivity difference between SARS-CoV-2 S1 and RBD using ACE2-based LFIA might be associated with the K_D value of ACE2 and S1-mAb were lower for RBD binding than for S1 binding (Supplementary Table 2). Meanwhile, the MERS-CoV S1 antigen was not detected even at relatively high concentrations (100 ng antigen); however, 100 ng antigen of SARS-CoV S1 was slightly detected, as shown in Fig. 5b and c, and Supplementary Table 3. This means that more than 100 ng antigen of SARS-CoV S1 might not be distinguished using the ACE2-LFIA. The intensity slightly increased at a high concentration of SARS-CoV S1; however, the proposed ACE2-LFIA has the potential to distinguish SARS-CoV-2 from other coronaviruses due to the significant difference in intensity. The inset graph in Fig. 5c shows the detection intensity for the S1 antigens from three different coronaviruses at an antigen concentration of 5 ng. The detection signal of SARS-CoV-2 S1 was clearly higher than the limit of detection (LOD, mean value of negative controls + 3 × standard deviation), whereas those of SARS-CoV S1 and MERS-CoV S1 were similar to that of the negative control. Thus, the proposed ACE2-LFIA has the ability to detect the SARS-CoV-2 antigen with high sensitivity and without significant cross-reactivity from other coronaviruses.

We thank the National Culture Collection for Pathogens of the Korean CDC for providing clinical SARS-CoV-2 isolates. This work was supported by the National Research Council of Science and Technology (NST) and National Research Foundation (NRF) funded by the Ministry of Science and ICT, Republic of Korea (grant number: CRC-16-01-KRICT and NRF-2020M3E9A1043749) and Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant numbers: HI20C0033 & HI20C0363).