Ultrabright nanoparticle-labeled lateral flow immunoassay for detection of anti-SARS-CoV-2 neutralizing antibodies in human serum.

The level of anti-SARS-CoV-2 neutralizing antibodies (NAb) is an indispensable reference for evaluating the acquired protective immunity against SARS-CoV-2. Here, we established an ultrabright nanoparticles-based lateral flow immunoassay (LFIA) for one-step rapid semi-quantitative detection of anti-SARS-CoV-2 NAb in vaccinee's serum. Once embedded in polystyrene (PS) nanoparticles, the aggregation-induced emission (AIE) luminogen, AIE490, exhibited ultrabright fluorescence due to the rigidity of PS and severe inhibition of intramolecular motions. The ultrabright AIE490-PS nanoparticle was used as a fluorescent marker of LFIA. Upon optimized conditions including incubation time, concentrations of coated proteins and conjugated nanoparticles, amounts of antigens modified on the surface of nanoparticles, dilution rate of serum samples, and so on, the ultrabright nanoparticles-based LFIA could accurately identify 70 negative samples and 63 positive samples from human serum (p < 0.0001). The intra- and inter-assay precisions of the established method are above 13% and 16%, respectively. The established LFIA has tremendous practical value of generalization as a rapid semi-quantitative detection method of anti-SARS-CoV-2 NAb. Meanwhile, the AIE490-PS nanoparticle is also promising to detect many other analytes by altering the protein on the surface.

Introduction

The coronavirus disease 2019 (COVID-19) caused by novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has become a global pandemic [1]. By the end of 2021, over 300 million people worldwide had been infected with SARS-CoV-2 and over 5 million people have died. At present, vaccination against SARS-CoV-2 is the key strategy to prevent this infectious disease [2]. Valid methods to evaluate the acquired immunity after vaccination are highly demanded. Testing of specific immunoglobulin M (IgM) and immunoglobulin G (IgG) against SARS-CoV-2 is commonly used to evaluate the effect of the vaccine [3]. However, only a small portion of the IgM and IgG can neutralize and resist SARS-CoV-2 [4], the positive detection of the specific IgM and IgG is not a reliable evaluation of acquired immunity against SARS-CoV-2. The viral entry of SARS-CoV-2 is mediated through the recognization of angiotensin-converting enzyme 2 (ACE2) on the human cell surface by the receptor-binding domain (RBD) in the spike protein of SARS-CoV-2 [5]. The acquired neutralizing antibodies (NAbs) after vaccination can competitively bind the RBD and neutralize SARS-CoV-2 [4]. Therefore, the direct detection of NAbs is a more accurate method to evaluate the effect of the vaccine [6,7]. Meanwhile, the detection of NAbs in human body could synergistically guide the vaccine strategies against new viral variants with particular reference [8].

The conventional virus neutralization tests require biosafety level 3 facilities due to the live virus operations, and the pseudovirus-based virus neutralization tests require biosafety level 2 facilities. The high-standard conditions cost much and require professional operator, which is unpractical for massive detection of NAbs. Therefore, several convenient methods, such as lateral flow immunoassay (LFIA) [9], enzyme-linked immunosorbent assay [10], surface plasmon resonance assay [11], and some emerging digital microfluidic systems [[12], [13], [14]], have been developed to detect SARS-CoV-2 NAbs recently. Among all of them, the LFIA method enjoys portability, rapidity, simplicity, and low cost, and it has been widely used in the point-of-care immunosensors. For example, the colloidal gold nanoparticles- [15] and colored cellulose nanobeads-based LFIA [16] were developed to detect the SARS-CoV-2 NAbs very recently. However, the sensitivity is insufficient and the linear range is usually narrow due to their restricted colorimetric signal. Fortunately, the fluorescent signal can provide higher sensitivity and the fluorescent marker-based LFIAs can improve the work linear range. Generally, the immunoassays contain competitive immunoassay and sandwich immunoassay. The sandwich immunoassay commonly has higher sensitivity and specificity. However, the NAb is a series of antibodies that can neutralize SARS-CoV-2, whose components are complex. It is hard to find two different binding sites on NAb. Therefore, the sandwich immunoassay is not suitable for NAb detection. Meanwhile, the direct competitive immunoassay is cheaper and more reliable than the sandwich immunoassay, the batch-to-batch variation is smaller, which is important for massive detection of NAbs. The direct competitive immunoassay was used in this study.

The sensitivity of the fluorescent marker-based LFIA highly depends on the working fluorescent materials [17]. Bright nanomaterials such as quantum dots (QDs) [18] and lanthanide-based microspheres [19] are commonly used in LFIAs. Compared to the metallic luminophores, the color and fluorescence wavelength of organic luminogens can be easily tuned through engineering of chemical structures [20]. However, most luminogens undergo aggregation-caused quenching (ACQ) due to strong π−π stacking in aggregate state when they are introduced into nanoparticles or self-aggregation at the test line, which would reduce the detection performance of LFIA [21]. Recently, organic luminogens with aggregation-induced emission (AIEgens) have drawn broad attention because of their bright fluorescence in aggregate state or at high concentrations, which can overcome the drawback of traditional luminogens [22]. Since the bright fluorescence of AIEgens comes from the intramolecular rotation restriction and inhibition of nonradiative decay, the denser the packing of AIEgens, the brighter the fluorescence is [23]. Therefore, the AIEgens are generally used in dense packing formats such as surfactant matrix-AIEgen nanoparticles [24], amorphous or crystalline precipitates [25]. The surfactant matrix-AIEgen nanoparticles and amorphous precipitates can restrict intramolecular rotation and shield the influence of water, enabling bright fluorescence of the AIEgens [23]. But their packing is much looser than in crystalline state, in which the AIEgen molecules are orderly and densely packed with less intramolecular rotation and ultrabright fluorescence [25]. However, the nanocrystal of AIEgen is suffered from difficult conjugation with antibodies and difficult release from the pad, which restricts the utilization of the ultrabright AIEgen nanocrystal in LIFA. To introduce the ultrabright AIEgens into LIFA, the proper format of the AIEgens is desired.

In this work, we established an AIEgen-embedded polystyrene (PS) nanoparticles-based LFIA, and successfully detected the anti-SARS-CoV-2 NAb in serum from the vaccinees. PS is a rigid polymer with hydrophobic chains and steric phenyl rings. We hypothesized that the rigidity of PS particles could severely inhibit the intramolecular motions and trigger ultrabright fluorescence once the AIEgen is embedded into PS particles. Our study showed that the fluorescence signal of a green blue-emissive AIEgen (AIE490) was enhanced more than 10 times after being encapsulated into the carboxyl-modified PS nanoparticles (AIE490NP), which was also much brighter than QDs (Scheme 1 A). The AIE490NP was modified with ACE2 Fc chimera (ACE2-AIE490NP) as a fluorescence marker; the SARS-CoV-2 nucleocapsid S RBD fusion was coated on the nitrocellulose membrane as a test line. When the sample was negative (without NAbs), the test line exhibited bright fluorescence signal due to the strong ACE2-RBD binding; when the sample was positive (with NAbs), the test line exhibited a dim fluorescence signal due to the inhibition of ACE2-RBD binding through NAb-RBD binding (Scheme 1B). Benefiting from the ultrabright fluorescence of AIE490NP, 63 positive serum samples from vaccinees and 70 pre-SARS-CoV-2 serum samples were accurately identified by using the AIE490NP-based LFIA. Theoretically, the concentrations of NAbs could also be quantified by using the AIE490NP-based LFIA if the standard sample of NAbs is available. Meanwhile, one detection could be finished within 20 min, and the LFIA strip is portable and costless, which is important for widespread applications of SARS-CoV-2-related antibody detection and vaccination study.

Scheme 1

(A) Preparation of AIE490NP. (B) Schematic of the developed AIE490NP-based LFIA test strip for detection of anti-SARS-CoV-2 NAbs in human serum sample.

Experimental section

Synthesis of AIE490NP

The AIE490NP was assembled by encapsulating AIE490 into the 200 nm carboxyl-modified PS nanoparticles through the swelling method. First, 10 mg of PS nanoparticles were resuspended thoroughly in 1 mL of 2.5% SDS solution using a UP200S probe ultrasonic crusher (Hielscher, Teltow, Germany), and then 100 μg of AIE490 dissolved in 100 μL of THF was added. The mixture was then stirred for 4 h at room temperature. After centrifugation at 25,000g for 30 min and removing the supernatant, the AIE490NP were resuspended in 1 mL of ultrapure water and stored at 4 °C for future use.

Characterization of AIE490 and AIE490NP

The Nuclear Magnetic Resonance Spectra (NMR) of AIE490 were collected using an AscendTM400 spectrometer (Bruker Daltonics Inc., Germany). The UV–vis absorption and emission were obtained using a UV–vis spectrophotometer (Shimadzu, Japan) and a Lumina spectrophotometer (Thermo Fisher Scientific Inc., USA), respectively. The hydrodynamic diameter and zeta-potential were characterized using a Nano-ZS90 ZetaSizer (Malvern Panalytical Ltd., UK). The transmission electron microscopy (TEM) images were taken with an H-7500 transmission electron microscope (Hitachi Co., Ltd., Japan). The photoluminesce quantum yields (PLQYs) of AIE490 molecules aggregated in water, AIE490-DSPE-PEG, AIE490NP, QD-NP-525, and QD-NP-545 were measured using quinine sulphate (QY = 58%) in water as a reference.

Preparation of AIE490NP conjugated with ACE2 and MIgG

The AIE490NP-ACE2 conjugation and AIE490NP-MIgG conjugation were prepared by the EDC-NHS method. Initially, 1 mg of AIE490NP was centrifuged to remove the storage solution and then resuspended in 500 μL of activating buffer containing 0.625 μmol of EDC and 5 μmol of sulfo-NHS. The interactant was gently shaken for 30 min at room temperature for activation, and then centrifuged to separate the activated AIE490NP. The activated AIE490NP was then washed two times using washing buffer and resuspended in 300 μL of binding buffer. Then, the activated AIE490NP was added with 50 μg of ACE2 protein which was purified and condensed into 200 μL of the binding buffer using a centrifugal filter unit with an Ultracel-10 membrane and stirred for 2 h to form a bipartite complex of AIE490NP-ACE2. The complex was then mixed with 500 μL of blocking buffer containing 5% BSA and incubated at room temperature for another 2 h to block the unreacted NHS ester on the surface of activated AIE490NP. After incubation, the AIE490NP-ACE2 was centrifuged to remove the unreacted reagent, unconnected ACE2 antigen, and binding buffer, and then redispersed in labeling antibody storage buffer and stored at 4 °C for further use. The conjugation procedure of AIE490NP-MIgG was similar to that of AIE490NP-ACE2, except the ACE2 protein was replaced with MIgG.

Treatment of sample pad, conjugate pad, and absorbent pad

The glass fiber was cut into 300 × 21 mm and 300 × 12 mm pieces to obtain untreated sample pads and conjugate pads. The pads were then soaked in the equivalent treatment buffer for 2 h at room temperature and dried at 37 °C for 24 h in a cabinet drier to gain sample pad and conjugate pad. The absorbent pad was 300 × 26 mm pieces cut from whole one without other treatment. All those pads were stored in a moisture-proof cabinet.

Preparation of AIE490NP-based LFIA test strip

The AIE490NP-based LFIA test strip is composed of five ingredients: sample pad, conjugate pad, absorbent pad, nitrocellulose membrane, and plastic adhesive backing plate. The RBD recombinant protein and anti-IgG were prediluted using coating buffer and then sprayed equably on the nitrocellulose membrane as test line and control line respectively. The nitrocellulose membrane was then stuck on the backing plate and dried overnight at 37 °C. The AIE490NP-ACE2 and AIE490NP-MIgG prediluted in labeling antibody dilution buffer were both spotted onto pretreated conjugate pad with a ratio of 10 μL/cm following by dried overnight at 37 °C. The five components were assembled sequentially to ensure a direct flow from sample pad to absorbent pad under capillarity. Last, the well-assembled plate was cut into 4 mm wide strips using a strip cutter. Each strip was packaged into a strip shell with a circular sample loading hole and a rectangular viewing window for further use.

Results

Synthesis and characterization of AIE490NP

AIE490 is a green blue-emissive AIEgen with a fluorescence peak at 490 nm. It is constructed by typical AIE units including tetraphenylethylene and triphenylamine. The synthetic pathway of AIE490 is detailedly described in the supporting information (Scheme S1), and the chemical structure and purification were characterized by NMR (Fig. S1) and mass spectrometry (Fig. S2). The simple synthetic procedure with high yield is advantageous for large-scale production. The absorbance and fluorescence spectra of AIE490 molecules aggregated in water (THF/water, v/v = 1/99, 10 μg/mL) were shown in Fig. S3. The absorbance maximum is 362 nm and the fluorescence maximum is 494 nm.

The AIE properties of AIE490 were examined by measuring the fluorescence intensity in THF/water mixtures with varied water fractions (f w, vol%). With the increasement of the water fractions, the fluorescence intensity of AIE490 increased gradually from 0 to 80% and increased significantly from 80 to 99%, exhibiting typical AIE properties (Fig. 1 A).

Fig. 1

(A) Fluorescence spectra of AIE490 in THF and THF/water mixture with different water fractions (fw). (B) Absorption and fluorescence spectra of AIE490NP, λex = 365 nm. The TEM images of PS nanoparticles (C) and AIE490NP (D), the insets show the image of PS nanoparticles and AIE490NP under visible light and UV light, respectively. (E) The fluorescence intensity of AIE490 aggregated in water, AIE490-DSPE-PEG, and AIE490NP, the insets show the fluorescence under UV light, λex = 365 nm (a is AIE490 aggregated in water, b is AIE490-DSPE-PEG, and c is AIE490NP). The amount of AIE490 aggregated in water was equal to that encapsulated in DSPE-PEG nanoparticles and PS nanoparticles to be 1 μg. (F) The fluorescence spectra of AIE490NP and two QPs, λex = 365 nm. The concentrations of the three nanoparticles were all 0.1 mg/mL.

The stability of AIE490 was evaluated by monitoring the changes of fluorescence intensity under continuous irradiation of white light (100 mW/cm2, measured by LWP10W-A optical power meter (Beijing Laserwave OptoElectronics Technology Co., Ltd)), or under different storage temperatures and pH. The AIE490 exhibited a loss of only 8.0% in fluorescence intensity after 60 min irradiation (Fig. S4), indicating excellent photostability. The fluorescence intensity of AIE490 at different temperatures (10–50 °C) exhibited at most 1.6% of loss (Fig. S5 and Table S1), while those stored at different pH (5.5–9.0) exhibited at most 5.6% of loss (Fig. S6 and Table S2), demonstrating excellent stability against different temperature and pH. The high stability of AIE490 is beneficial to reliable detection.

For diagnostic applications, the AIE490 molecules were then encapsulated into carboxyl-modified PS nanoparticles using organic solvent swelling method to gain AIE490NP. AIE490 can dissolve well in THF at above 4 mg/mL. The well-solubility in THF allows AIE490 to be efficiently encapsulated into PS nanoparticles by swelling method. The optical properties of AIE490NP were characterized using UV–vis and fluorescence spectroscopy, the absorbance maximum and fluorescence maximum are similar to the aggregates in water. The maximum absorption and emission of AIE490NP were 365 nm and 490 nm, respectively (Fig. 1B). Fig. 1C and D showed the TEM images of PS nanoparticles and AIE490NP, whose sizes and morphologies were almost the same. The hydrodynamic diameter and zeta potential of AIE490NP were measured to be 180 nm and −53.7 mV, which were also similar to the PS nanoparticles (Figs. S7–S8). The negative zeta potential came from the carboxyl groups on the surface. The insignificant differences in hydrodynamic diameters and zeta potential indicated that the encapsulation of AIE490 would not damage the PS nanoparticles. The insets of Fig. 1C and D are the pictures of PS nanoparticles and AIE490NP under visible light and UV light, respectively. The strong fluorescence of the obtained AIE490NP indicated successful encapsulation of AIE490 into PS nanoparticles. By fitting a response curve of absorption at 365 nm, the number of AIE490 molecules in an AIE490NP was estimated to be 3.4 × 104 (Fig. S9).

To confirm the performance of AIE490NP, we observed the fluorescence intensity of AIE490NP and the equivalent AIE490 molecules aggregated in water or embed in DSPE-PEG nanoparticles (AIE490-DSPE-PEG). The FL signal of AIE490NP was 14 times as strong as the aggregate in water, 2 times as strong as AIE490-DSPE-PEG (Fig. 1E). The PLQYs of AIE490 aggregated in water, AIE490-DSPE-PEG and AIE490NP were measured as 31.4%, 26.9% and 33.6%, respectively. Since polystyrene also has an absorbance at the excitation wavelength, the PLQY of AIE490NP is only slightly higher than AIE490 molecules aggregated in water though AIE490NP is much brighter under the same photoexcitation. The ultrabright fluorescence of AIE490NP came from the shielding of water by the hydrophobic chains of PS and the severe inhibition of intramolecular motions by the rigidity of PS. Moreover, we compared the fluorescence intensity of AIE490NP with two QD-doped PS nanoparticles (QPs) that had been reported in our previous studies [26,27]. The AIE490NP also exhibited fluorescence intensity dozens of times that of QPs (Fig. 1F and Fig. S10). The PLQYs of QD-NP-525 and QD-NP-545 were 11.9% and 7.2%, respectively, which were much lower than the AIE490-based groups. The AIE490NP has the similar stability properties to AIE490 against continuous irradiation (Fig. S4), different temperatures (Fig. S5 and Table S1) and pH (Fig. S6 and Table S2). Meanwhile, the hydrodynamic diameters of AIE490NP at different temperatures or pH were maintained at 172.8–196.4 nm or 172.8–196.4 nm (Fig. S11), respectively, indicating excellent structural stability of AIE490NP against temperature and pH.

Preparation and characterization of AIE490NP-antigen conjugation

The preparation of AIE490NP-antigen conjugations was accomplished through activation of carboxyl on AIE490NP and coupling with amine on the antigens (Fig. 2 A). To ensure the successful conjugation of AIE490NP with ACE2 or MIgG, the hydrodynamic diameter and zeta potential of AIE490NP, AIE490NP-ACE2, and AIE490NP-MIgG were characterized. After modification of ACE2 and MIgG, the hydrodynamic diameters slightly increased from 180 nm to 230 nm and 250 nm, respectively (Fig. 2B). The zeta potentials also changed from −53.7 mV to −29.9 mV and −34.3 mV after the connection with the antigens (Fig. 2C). The increased zeta potentials were due to the block of negative carboxyl groups. Meanwhile, the ACE2 and MIgG could be directly observed in the TEM images of AIE490NP-ACE2 (Fig. 2D) and AIE490NP-MIgG (Fig. 2E), which further supported the successful conjugation. To further verify the protein modification, surface plasmon resonance (SPR) was used to detect the affinity of AIE490NP-ACE2/RBD and AIE490NP-MIgG/anti-IgG (Fig. S12). By linking RBD or anti-IgG to the surface of 3D Dextran chip as ligands, the dissociation equilibrium constants of AIE490NP-ACE2/RBD and AIE490NP-MIgG/anti-IgG were tested to be 101 pM and 87 pM (Table S3), respectively, indicating the successful modification of ACE2 and MIgG onto AIE490NP. Meanwhile, the conjugation with antigens hardly influenced the fluorescence intensity of AIE490NP (Fig. S13).

Fig. 2

(A) The conjugation process of AIE490NP-ACE2 and AIE490NP-MIgG. The hydrodynamic diameters (B) and zeta potentials (C) of AIE490NP, AIE490NP-ACE2, and AIE490NP-MIgG. The TEM images of AIE490NP-ACE2 (D) and AIE490NP-MIgG (E).

Working mechanism of the AIE490NP-based LFIA

Based on the recognition interaction between ACE2 and RBD, and the binding interaction between NAb and RBD, the AIE490NP-based LFIA was performed as a typical competitive immunoassay. The picture of the LFIA product was shown in Fig. 3 A. The RBD was coated as the test line, the anti-IgG was coated as the control line. The sample buffer containing human anti-SARS-CoV-2 NAb was dropped to the sample pad following by migration to the conjugate pad, and gradually carried the AIE490NP-ACE2 and AIE490NP-MIgG immobilized on the conjugate pad towards absorbent pad by capillarity. Once the mixture reached the test line, the anti-SARS-CoV-2 NAb would bind the SARS-CoV-2 RBD to prevent it from recognizing human ACE2, which would lead to a weaker fluorescent signal of the test line (HT). The higher the NAb concentration, the weaker the fluorescent signal would be. When the sample was negative, no competition happened to break the ACE2-RBD binding, and the AIE490NP-ACE2 would stay on the test line, exhibiting strong fluorescent signals. Meanwhile, the binding between anti-IgG and MIgG was not influenced so that the fluorescent signal of the control line (HC) was undisturbed by anti-SARS-CoV-2 NAb in the serum sample. Fig. 3B showed the pictures of the visual detection results upon 365 nm irradiation for several positive and negative samples. The test line of the negative samples was apparently brighter than that of the positive samples. After 20 min incubation, the test strip was measured using a portable fluorescence reader to obtain HT and HC (Fig. 3C).

Fig. 3

(A) The picture of the AIE490NP-based LFIA product. (B) The pictures of the visual detection results upon 365 nm irradiation for several positive and negative samples. (C) The fluorescence peak heights readout curve of positive and negative samples using.

To visually demonstrate the excellent performance of the ultrabright AIE490NP in LFIA, the QD-NP-525 and QD-NP-545 were also conjugated with ACE2 and MIgG to serve as fluorescent markers in the LFIA. As shown in Fig. S14, the LFIA test strip with AIE490NP exhibited a much higher fluorescence signal than QD-NP-525 and QD-NP-545 on the fluorescent band, indicating better potential in LFIA applications.

To counteract the intrinsic heterogeneity of the AIE490NP-based LFIA and the influence of serum matrix, the ratio of HT and HC was used as the final result of a test strip, which would make the results more reliable and reproducible. The inhibition rate was calculated to further confirm the inhibitory effect of the NAb in the serum sample on the RBD binding ability by the following equation:where R is the HT/HC ratio of sample buffer, R is the HT/HC ratio of the sample. Meanwhile, the signal-to-noise ratio was evaluated through the ratio of the inhibition rates between positive and negative samples, to confirm whether the AIE490NP-based LFIA could effectively distinguish the negative and positive serum samples.

Optimization of the antibody-antigen recognition conditions

The detection performance of an LFIA method also highly depends on the antibody-antigen recognition conditions, including recognition time, the concentration of protein coating on test/control line, the amount of protein modified on fluorescent nanoparticles, the usage amount of modified fluorescent nanoparticles, and so on. As shown in Fig. 4 A, over the range of 10–40 min of incubation after loading to test strip, the HT/HC ratios of sample buffer and negative sample reached a plateau at 20 min while that of the positive sample stayed stable. The inhibition rates of the negative and positive samples both reached a plateau after 20 min of incubation (Fig. 4B). The signal-to-noise ratio reaches a plateau at 15 min and its coefficient of variation (CV%) stabilizes at a low level after 20 min incubation (Fig. S15). According to the changing trend of inhibition rate and signal-to-noise ratio, it can be found that the detection requirements can be met at 10 min. However, the instability of positive results at 10 min may lead to unreliable outputs. Furthermore, since the established LFIA is for qualitative or semi-quantitative detection, consistency of results is more important. Therefore, we finally chose 20 min as the incubation time for the LFIA.

Fig. 4

(A) The HT/HC ratios of sample buffer, negative samples, and positive samples under different incubation time. The average and standard deviation were calculated through five replications. (B) The inhibition rates of negative samples, and positive samples under different incubation time. The average and standard deviation were calculated through five replications.

To figure out the optimized concentrations of RBD and anti-IgG coating onto nitrocellulose membrane, we selected 0.5, 1, and 2 mg/mL as alternatives. The results of the cross-pairing experiment showed that the HT/HC ratios of all sample buffer, negative sample, and positive sample heightened following the increase of RBD concentrations and decrease of anti-IgG concentrations (Fig. S16); the differences of inhibition rates between the negative and the positive samples reached a peak when the concentration of RBD is 1 mg/mL, and they are hardly related to the concentration of anti-IgG (Fig. S17). Comprehensively, the signal-to-noise ratios of all the samples were figured out by the values of the HT/HC ratios and the inhibition rates (Fig. 5 ). The signal-to-noise ratio reached the highest when the concentration of RBD was 1 mg/mL, the concentration of anti-IgG was 2 mg/mL, which were selected as the optimized coating concentrations.

Fig. 5

The signal-to-noise ratio of the AIE490NP-based LFIA under different concentrations of RBD coated on test line and anti-IgG coated on control line.

We also assessed whether the amount of ACE2 modified on the surface of AIE490NP would affect the detection results. As the ACE2 modified on the surface of each milligram of AIE490NP increased from 25 μg to 100 μg, the HT/HC ratio of sample buffer, negative samples, and positive samples all increased (Fig. S18) while the inhibition rates of all negative and positive samples kept stable (Fig. S19). The signal-to-noise ratio reached the highest when the surface-modified protein was 50 μg (Fig. 6 A).

Fig. 6

(A) The signal-to-noise ratio of the AIE490NP-based LFIA under different amount of ACE2 modified on the surface of AIE490NP. (B) The signal-to-noise ratio of the AIE490NP-based LFIA under different concentration of AIE490NP-ACE2 used in the LFIA.

Furthermore, we optimized the usage amount of the AIE490NP-ACE2. The HT/HC ratio increased significantly with the increase of AIE490NP-ACE2 antigen concentration from 62.5 to 1000 μg/mL (Fig. S20). At the same time, the inhibition rate of the positive sample reached the peak when the concentration of AIE490NP-ACE2 antigen was 250 μg/mL, while that of the negative sample decreased along with the increasing concentration of AIE490NP-ACE2 (Fig. S21). The signal-to-noise ratio reached the highest when the concentration of AIE490NP-ACE2 was 250 μg/mL (Fig. 6B).

Optimization of the dilution ratio of clinical serum samples

To achieve the highest signal-to-noise ratio, the dilution ratio of clinical serum samples was optimized. Both negative serum samples and positive serum samples were prediluted using sample buffer with the ratio of 1/5, 1/10, 1/20, and 1/40 before loading onto the test strip. After 20 min of incubation, each test strip was loaded into a fluorescence reader to obtain the fluorescent intensities of the test line (HT) and the control line (HC). The HT/HC ratios of the negative samples increased continuously accompanied by the increased dilution ratio from 1/5 to 1/40, which was mainly caused by the reduced influence of the serum matrix (Fig. 7 A). Meanwhile, due to the reduction of NAb concentration upon dilution, the HT/HC ratios of the positive samples gradually increased (Fig. 7A), which was solid evidence that the AIE490NP-based LFIA was promising for quantitative detection of NAbs. As shown in Fig. 7B, the inhibition rates of both negative and positive samples continued to decrease with the increased dilution ratio. Comprehensively, the signal-to-noise ratio reached the highest when the dilution rate was 1/20 (Fig. 7C).

Fig. 7

The (A) HT/HC ratios and (B) inhibition rates of negative samples and positive samples under different dilution rates of serum samples. The average and standard deviation were calculated through five replications. (C) The signal-to-noise ratio of the AIE490NP-based LFIA under different dilution rates of serum samples.

Practical detection of anti-SARS-CoV-2 NAb in clinical serum samples

To determine the detection threshold of the AIE490NP-based LFIA for detecting anti-SARS-CoV-2 NAb in serum samples, 70 negative serum samples and 10 sample buffers were measured using this LFIA method to obtain the ratio of HT/HC (Fig. 8 A and Table S4). The mean values of HT/HC for the negative serum samples and sample buffers had no significant difference. 63 positive serum samples from vaccinees were measured to obtain the HT/HC ratios (Fig. 8A and Table S5), which were apparently decreased compared to the negative serum samples and sample buffers. All the positive serum samples were collected from vaccinees with twice vaccination of inactivated vaccine from China one month after the second vaccination. The inhibition rates of these samples were calculated with the assistance of the average of 10 repeated measurements of the sample buffer (Fig. 8B). According to the average plus 3-fold standard deviation of inhibition rates of the negative samples, the detection threshold of anti-SARS-CoV-2 NAb was calculated to be 28.35. The obvious difference between the inhibition rates of negative samples and positive samples indicated that the AIE490NP-based LFIA method could effectively distinguish whether a serum sample contains anti-SARS-CoV-2 NAb (p < 0.0001). The reproducibility was evaluated through testing five positive serum samples and five negative serum samples to determine the coefficient of variations (CVs) of intra- and inter-assay. As shown in Table 1 , the intra- and inter-assay of negative samples were around 9.37–12.99% and 12.69–15.03%, respectively, while those of the positive samples were 6.81–7.58% and 7.15–8.44%, respectively. All the CVs were below or around 15%, indicating an acceptable reproducibility of the established AIE490NP-based LFIA.

Fig. 8

(A) The HT/HC ratios of 10 sample buffer, 70 negative samples, and 63 positive samples. (B) The inhibition rates of 70 negative samples and 63 positive samples. The dashed line represents the threshold calculated as the average plus 3-fold standard deviation of inhibition rates of the negative samples to be 28.35. ***, p < 0.001.

Table 1

The reproducibility of the developed AIE490NP-based LFIA.

		Intra-assay (n = 5)		Inter-assay (n = 10)
		Inhibition rate (%)	CV (%)	Inhibition rate (%)	CV (%)
Negative samples	1	9.02 ± 0.85	9.37	9.38 ± 1.28	13.62
	2	14.00 ± 1.62	11.58	13.94 ± 2.09	15.03
	3	16.57 ± 1.75	10.62	16.43 ± 2.08	12.69
	4	12.35 ± 1.60	12.99	12.63 ± 1.82	14.44
	5	12.72 ± 1.61	12.69	12.90 ± 1.72	13.33
Positive samples	1	66.97 ± 4.56	6.81	67.44 ± 5.08	7.53
	2	70.22 ± 5.32	7.58	72.15 ± 5.80	8.04
	3	75.35 ± 5.56	7.37	75.97 ± 6.35	8.36
	4	65.88 ± 4.97	7.54	63.94 ± 4.57	7.15
	5	82.05 ± 5.62	6.85	79.99 ± 6.75	8.44

Discussion

We have proposed a well-applicable method that combined the AIE-PS nanoparticles and LFIA to achieve rapid and reliable detection of anti-SARS-CoV-2 NAb in human serums. The ultrabright AIE-PS nanoparticles were used as the marker to increase the sensitivity and shorten the detection time. The ultrabright nanoparticles-based LFIA has tremendous practical value of generalization as a rapid semi-quantitative detection method of anti-SARS-CoV-2 NAb.

The anti-SARS-CoV-2 NAb level in the body is becoming an indispensable reference to evaluate the acquired protective immunity for SARS-CoV-2 [28]. Among all the NAb detection methods, LFIA has unique advantages, such as low cost, easy operation, and no need for complex pretreatment and sophisticated instruments [9]. The marker for signal output is the key component that concerns the performance of LFIA. Until now, the colloidal gold nanoparticle is still the most common marker for LFIA, but it is mostly used in qualitative LFIA [29]. Fluorescent markers such as lanthanide chelate nanoparticles, QPs and organic dyes can enable LFIAs with quantitative detection capability. However, the lanthanide chelate nanoparticles are confined to the fixed fluorescence wavelength; the QPs exhibited a lack of fluorescence intensities; the traditional organic dyes were limited with the quenched fluorescence in the aggregate state, poor photochemical stability, and narrow Stokes shift [30]. In the detection process by using LFIA, the markers are usually gathered in the test line or control line to give the output signals [31]. In our designs, we used an AIEgen as the fluorescent marker to improve the performance of LFIA. The AIE490 we synthesized exhibited typical AIE property. Its fluorescence was extremely enhanced when it turned to aggregate state in water. In this regard, AIE490 could overcome the quenched fluorescence of traditional organic dyes in aggregate state and satisfy the demand of LIFA better. Meanwhile, the excitation and emission wavelength of AIEgens are tunable. By designing the structure of AIEgen to change its emission wavelength, AIE nanoparticles can be better adapted to various LFIA detection instruments. The best excitation wavelength of AIE490 was 365 nm, which was a customized wavelength that matched the excitation light source of most portable detection instruments. As some AIEgens could be excited with the same laser but emit different colors of light, we will further try the multicolor markers-based LFIA to increase the detection sensitivity. The Stokes shift of AIE490 was 135 nm, which was large enough to distinguish the emission fluorescence from the excitation light. Therefore, the signal collector could easily and accurately read the outputs. All these advantages make AIE nanoparticles have broad prospects of application in LFIA even many other immunofluorescence sensors.

Though most hydrophobic AIEgens have a strong fluorescence in water, the aggregate format of an AIEgen does matter much about the fluorescence intensity. For example, the surfactant matrix-AIEgen nanoparticles are generally brighter than AIEgen precipitate in water [24]. The AIEgen crystals have much denser packing and exhibit ultrabright emission [25]. However, we found that neither the bright surfactant matrix-AIEgen nanoparticles nor the ultrabright AIEgen crystals could be used in LFIA due to the terrible release from the pad. In this work, we encapsulated AIE490 into the PS nanoparticles and found that the obtained AIE490NP had an ultrabright fluorescence. The fluorescence intensity of AIE490NP was twice that of the surfactant matrix-AIE490 nanoparticles, tenfold that of the precipitate of AIE490 in water. Furthermore, the fluorescence intensity of AIE490NP was also much stronger than QPs (Fig. S10). We proposed that the ultrabright emission of AIE490NP came from the rigidity of PS and severe inhibition of intramolecular motions of AIE490 in AIE490NP. On the benefit of the ultrabright emission of AIE490NP, the LFIA we designed could figure out the 70 negative serum samples and 63 positive serum samples without mistakes. Meanwhile, in the optimized conditions, the intra- (6.81–12.99%) and inter-assay (7.15–15.03%) of the AIE490NP-based LFIA indicated an acceptable reproducibility. The performance of the AIE490NP-based LFIA suggested much practical utility for the rapid detection of NAb levels in vaccinated sera.

The present AIE490NP-based LFIA was designed to be a quantitative detection of anti-SARS-CoV-2 NAb in human serum, but it just achieved semi-quantitative function due to the lack of a golden standard for anti-SARS-CoV-2 NAb calibration. Until now, the specific level of anti-SARS-CoV-2 NAb that can effectively prevent or alleviate symptoms has not been reported. But there is no doubt that the higher levels of NAbs mean greater protection. Therefore, the current semi-quantitative detection mode through the detection of fluorescence signal intensity can meet the evaluation needs of NAbs in the stage of see. Once standards were available, present AIE490NP-based LFIA could perform better.

Conclusions

In conclusion, an AIE490NP-based LFIA was successfully established for rapid semi-quantitative detection of anti-SARS-CoV-2 NAb, which could achieve the detection within 20 min with one step. To verify the performance of the present AIE490NP-based LFIA, 70 negative serum samples and 63 positive serum samples were measured. This present method could effectively distinguish the positive and negative serum samples. The intra- (6.81–12.99%) and inter-assay (7.15–15.03%) of the established AIE490NP-based LFIA indicate an acceptable reproducibility. The performance of the established AIE490NP-based LFIA suggests its potential as an alternative method for rapid detection of neutralizing antibody levels in vaccinated sera. At the same time, it also provides a basis investigation of multi-label AIE490NP-based LFIA. And the AIE490NP-based LFIA has great potential for various analytes detections, which deserve further research.

Credit author statement

Lun Bian: Methodology, Validation, Software, Writing – original draft, review, and editing. Zhaoyue Li: Investigation, manuscript revision. An He: Software, Data curation. Biru Wu: Investigation, Methodology, Software, synthesis and characterization of materials. Hui Yang: exploration of conditions for nanoparticles synthesis. Yingsong Wu: Conceptualization, Methodology. Fang Hu: Conceptualization, Methodology, Funding acquisition, writing, review, and editing. Guanfeng Lin: Conceptualization, Methodology, Funding acquisition, writing, review, and editing. Deqing Zhang: discussion and review.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.