SiO2 is a biocompatible monodisperse NP and has been used as a carrier for fluorescent and SERS materials in LFIA analysis. These composite NPs improve the sensitivity of LFIA, but their multi-scene application is limited by the need to equip them with additional instrumentation [41,42,43]. To address the need for multi-scene detection, we designed a colorimetric and fluorescence-enhanced NP for LFIA detection.

Dual-signal Si-Au/DQD NPs were synthesized using the PEI layer-by-layer (LbL) self-assembly method (Scheme 1a). The introduction of the dual-signal Si-Au/DQD NPs into the LFIA system to replace conventional tags exhibited the following features: (i) SiO2 as the core provides good dispersion and stability; (ii) the AuNPs and QDs adsorbed on the SiO2 NP surface provide good colorimetric and fluorescent signals, respectively, enabling dual-mode detection of the LFIA; (iii) the adsorbed second QD layer confers strong fluorescence to the Si-Au/DQD NPs and has more carboxylation sites, for antibody coupling, making it an ideal label for ultrasensitive LFIA.

The fabricated Si-Au/DQD NPs were characterized using TEM, SEM, energy-dispersive X-ray spectroscopy (EDS), and zeta potential measurements. Figure 1a-d show the TEM images of SiO2, Si-Au, Si-Au/QD, and Si-Au/DQD, respectively. The SiO2 NPs exhibited good homogeneity and dispersion (Fig. 1a), and Si-Au, Si-Au/QD, and Si-Au/DQD NPs with SiO2 as the core showed the same dispersion as the SiO2 NPs (Fig. 1b-d). Figure 1e-h show the high-resolution TEM images of SiO2, Si-Au, Si-Au/QD, and Si-Au/DQD, respectively. According to our previous work, many positively charged sites are formed on the surface of PEI-coated SiO2 NPs [12]. Therefore, the negatively charged AuNPs adhere tightly to the SiO2 surface by electrostatic adsorption, forming Si-Au NPs (Fig. 1f). As shown in Fig. 1f, due to the addition of a small amount of AuNPs, numerous exposed PEI sites remain on the SiO2 surface. The negatively charged QDs attach to the remaining exposed PEI sites through electrostatic interactions, forming Si-Au/QD NPs with mixed AuNPs and a QD layer (Fig. 1g). The co-distribution of AuNPs and QDs on the SiO2 surface indicates that AuNPs do not affect the subsequent adsorption of QDs. As presented in Fig. 1h, fluorescence-enhanced Si-Au/DQD NPs were obtained by repeating the adsorption process of PEI and QDs. With the adsorption of a mixed AuNP/QD layer and a QD shell, the Si-Au/DQD surface becomes rougher, providing many sites for the bio functionalization of antibodies. Figure 1i presents a partially enlarged TEM image of Si-Au/DQD. The magnified image shows that the 20 nm AuNP and 12 nm QD are closely fitted to the Si-Au/DQD surface. Figure 1j shows the EDS mapping of a single Si-Au/DQD NP. According to the distribution of the Si, O, Au, Cd, and Se elements, the Si-Au/DQD is a typical core-shell structure NP. The high-angle annular dark-field scanning TEM (HAADF) images confirm that AuNPs and QDs are distributed on the surface of the SiO2 (Fig. 1k). Meanwhile, the change in zeta potential also reveals the synthesis process of Si-Au/DQD NPs. As revealed in Fig. S1, the zeta potential of SiO2 increased sharply after PEI encapsulation, while the zeta potential decreased after the adsorption of negatively charged AuNPs and QDs, and this regular change indicates the successful construction of Si-Au/DQD NPs.

TEM images of (a) SiO2 core, (b) Si-Au, (c) Si-Au/QD, and (d) Si-Au/DQD NPs. High-resolution TEM images of single (e) SiO2, (f) Si-Au, (g) Si-Au/QD, and (h) Si-Au/DQD NP. (i) TEM image of Si-Au/DQD with local enlargement. (j) EDS elemental mapping images of a single Si-Au/DQD NP. (k) HADDF image of a single Si-Au/DQD NP.

As a promising dual-signal tag for the LFIA sensor, we systematically investigated the optical properties and colloidal stability of Si-Au/DQD. Figure 2a shows SiO2, AuNP, Si-Au, Si-Au/QD, and Si-Au/DQD photographs under visible and UV light. Under visible light, all the solutions, except the SiO2 solution, showed a purplish-red signal, because of the plasmon resonance excitation of the AuNPs. Under UV light, only the Si-Au/QD, and Si-Au/DQD solutions emitted bright fluorescence, which was caused by the fluorescence excitation of the QDs. The UV-vis spectra in Fig. 2b also confirm that the adsorption of QDs has no significant effect on the colorimetric properties of Si-Au.

The colorimetric performance of the Si-Au/QD is determined by the AuNPs adsorbed on the SiO2 surface, the variable QD-to-Au ratio on the SiO2 surface affects the fluorescence and colorimetric intensity of the dual-signal NPs. Thus, we optimized the inner layer QD-to-Au ratio. Figs. S2a-d show the TEM images of the Si-Au/QD NPs with different QD-to-Au ratios (4:1 to 1:1). The density of Au on the SiO2 surface increases with the Au ratio, and the intensity of the UV absorption peak increases, and the colorimetric performance is subsequently enhanced (Figs. S2e and f). However, Au and QD are co-adsorbed on the SiO2 surface, Au competes with QD for binding, and an increase in Au density reduces the binding of QD, weakening the Si-Au/QD fluorescence signal (Figs. S2e and g). Therefore, a balance between the colorimetric and fluorescent signals should be maintained. The Si-Au/QDs with different QD-to-Au ratios (4:1 to 1:1) were modified with antibodies and introduced into LFIA strips to detect the same MPXV samples to identify the best QD-to-Au ratios. As shown in Figs. S2 h-j, the fluorescence signal is strongest at the T line for a 4:1 ratio, but the colorimetric signal is weak. The fluorescence signal at the T-line gradually decreases and the colorimetric signal gradually increases as the ratio changes from 3:1 to 1:1, but the difference in the colorimetric signal is small. In addition, the corresponding fluorescence signal at the T-line confirms that the best results are obtained with a QD-to-Au ratio of 3:1. Therefore, the Si-Au/QD NP with a QD-to-Au ratio of 3:1 exhibits the best performance and is suitable as the core of a fluorescence-enhanced dual-signal Si-Au/DQD.

Next, we verified the fluorescence properties of each Si-Au/DQD component. Figure 2c shows no fluorescence peaks for SiO2, AuNP, and Si-Au, and strong fluorescence peaks for Si-Au/QD and Si-Au/DQD. The fluorescence peaks of Si-Au/DQD are stronger than those of Si-Au/QD, because more QDs are loaded on the Si-Au/DQD. As displayed in Fig. 2d-f, the fluorescence spectra, and insets under different conditions illustrate that the colorimetric properties and fluorescence intensity of Si-Au/DQD remain stable over a wide pH range (3–13) and under high salt environments and long-term storage. The Si-Au/DQD exhibited excellent stability because of the use of SiO2 as the core. In contrast, the ordinary AuNP (20 nm) solution was agglomerated rapidly in acidic (pH ≤ 3) solutions and highly concentrated NaCl (100–1000 mM) solutions (Fig. S3). The above results indicate that Si-Au/DQD has good colorimetric and fluorescence stability for biolabeling and is an ideal tag for LFIA.

(a) Photographs of SiO2, AuNP, Si-Au, Si-Au/QD, and Si-Au/DQD under visible and UV light. (b) UV-vis spectra of these NPs. (c) Fluorescence spectra of these NPs. Stability of Si-Au/DQD under pH (d), salt (e), and time (f) conditions

The Si-Au/DQD-based LFIA platform consists of immunochromatographic strips and a portable fluorescence reader for the qualitative or quantitative analysis of the MPXV (Scheme 1c). The constructed LFIA strip consists of four parts: a sample pad, an absorption pad, a conjugate pad coated with Si-Au/DQD-mAb, and NC membranes with C and T lines. The entire detection process was completed in one step. Briefly, the target solution was loaded onto the sample pad and the liquid migrated upward along the NC membrane through the capillary force. In the positive experiments, the A29L protein reacted preferentially with the immuno-Si-Au/DQD of the conjugated pad, and the immuno-Si-Au/DQD-A29L protein complex was subsequently captured by the immobilized secondary antibody at the T-line, forming a sandwich structure through antigen-antibody interactions. As the liquid continued to migrate, excess immuno-Si-Au/DQD was captured by the goat anti-mouse antibody in the C-line region, whereas the immuno-Si-Au/DQD was only captured by the C-line in the absence of the A29L protein in the sample solution. After a certain assay time, a qualitative result can be obtained by the visual observation of the colorimetric or fluorescence signal on the T-line. In addition, the quantitative analysis of the MPXV A29L protein was achieved using a portable fluorescence reading instrument.

The affinity of the antibody is one of the factors that affect the LFIA sensitivity. Therefore, we screened the different antibodies and selected a pair with the best affinity for the following experiments. W1 and W2 antibodies for MPXV dection were provided by Xiamen One Clone Biotech Inc (Xiamen, China). Additional antibodies were provided by Sino Biological Inc. As shown in Fig. 3a, different antibody pairs exhibit different performance when detecting the same concentration of A29L protein (10 ng/mL). As shown by the signal-to-noise ratio (SNR) of the different pairs, the M8/M9 pair showed the best detection performance (Fig. 3b). Therefore, the M8/M9 pair was selected as the detection/capture antibody for the LFIA platform. Notably, a weak fluorescence signal on the T-line of negative sample could be recorded by the portable fluorescence instrument, which derived from the residual immuno-Si-Au/DQD labels in the running solution and can be deducted as the background noise for target detection. Meanwhile, the interior morphology of the T-line region was observed by SEM to verify the reliability of our proposed LFIA platform. As shown in Fig. 3c and d, the T-line region of the positive assay had a considerable amount of Si-Au/DQD NPs, while no Si-Au/DQD NPs were found in the T-line region of the negative control. This phenomenon indicates that the colorimetric and fluorescent signals of the strips were from the Si-Au/DQD NPs, thus demonstrating the reliability of the dual-signal LFIA platform.

(a) Photographs of different antibody pairs for A29L protein detection. (b) SNR of different antibody pairs. SEM image from the (c) positive assay and (d) negative control

The composition of the running buffer, the size of the NC membrane, and the concentration of the capture antibody are parameters that affect the performance of the LFIA assay; hence, we optimize these parameters to achieve optimal LFIA performance [44]. Fig. S4 shows that the running buffer containing 1% PBST (10 mM, pH 7.4, 1% Tween 20), 1% BSA, and 10% FBS reduced the non-specific adsorption of the Si-Au/DQD tags and maximized the SNR on the T line. As indicated in Fig. S5, the SNR of the CN140 membrane is higher than that of CN95 because the pore size of the CN140 membrane is smaller than that of CN95, and its flow rate of the solution is slower, resulting in a more adequate immune response. Finally, the influence of the antibody concentration immobilized on the T-line on LFIA was evaluated. As shown in Fig. S6, the highest SNR was obtained with the capture antibody concentration of 0.9 mg/mL. Furthermore, the optimal response time for LFIA was explored by analyzing the SNR of the T-line over a time range. As shown in Fig. S7, a reaction time of 15 min was sufficient for the quantitative detection of MPXV.

The performance of Si-Au/DQD-based LFIA was verified for various concentrations of the A29L protein with optimum parameters. In Fig. 4a, the photographs of the test strips under visible light and 365 nm UV light show that the samples with different concentrations of the MPXV A29L protein (0-100 ng/mL) produced different visual results. In visible light, a purple-red band appears in the region of the test strip’s C/T line, and the colorimetric signal at the T line decreased with the concentration. A faint purple-red band was observed when the concentration of the A29L protein decreased to 0.5 ng/mL, indicating a detection limit of 0.5 ng/mL in the colorimetric mode (Fig. 4ai). As shown in Fig. 4aii, the red fluorescent bands on the T line became fainter as the concentration of the A29L protein decreased, and they were positively correlated over a wide detection range of 0.005-100 ng/mL. Regardless of the mode, a reduction in the A29L protein concentration resulted in a significant decrease in the T-line signal of the test strip, verifying the feasibility of dual-signal quantitative assays. Meanwhile, the quantitative detection of the A29L protein was achieved by recording the fluorescence signal at the T-line zone and constructing the corresponding calibration curve (Fig. 4e). The obtained fluorescence signal calibration curve reveals that the fluorescence signal of the T-line is proportional to the concentration of the A29L protein in the range of 0.005 ng/mL to 100 ng/mL. The limit of detection (LOD) of the A29L protein was calculated to be 0.0021 ng/mL on the basis of the mean and triple standard deviation of the measured concentrations of the negative samples [45].

As a flexible POCT tool for MPXV detection, Si-Au/DQD-based LFIA was compared with other common immunodetection methods to evaluate its superiority. Theoretically speaking, the Si-Au/DQD tag contains a mixed AuNP/QD layer and a QD shell whereas the Si-Au/QD tag only has a mixed AuNP/QD layer, thus the Si-Au/DQD tag has more surface carboxyl groups and larger surface area for antibody modification. The maximum load capacity of antibody on the Si-Au/DQD and Si-Au/QD was investigated as shown in Fig. S8. The results suggested that Si-Au/DQD tag can load more MPXV antibody on its surface, thus possessing higher affinity to target MPXV antigen. The superior detection ability of the Si-Au/DQD labels was first verified by comparing them with monolayer Si-Au/QD tags and commercial quantum dot bead (QB) labels, using the same antibody pair. Figure 4b shows the results of the Si-Au/QD-based LFIA assay for the same A29L protein concentration. The limit of visual detection in its colorimetric mode is 0.5 ng/mL, which is the same as that of the Si-Au/DQD-based LFIA (Fig. 4bi). The fluorescence mode has a naked eye detection limit of 0.05 ng/mL, which is 10 times higher than that of the Si-Au/DQD-based LFIA (Fig. 4bii). The calibration curve constructed from the corresponding fluorescence signal shows that the LOD value of Si-Au/QD-LFIA was 11.4 times higher than that of Si-Au/QD-based LFIA (Fig. 4f). These results verified that using Si-Au/DQD with more QD layers as the fluorescent label can effectively improve the detection sensitivity of LFIA. Fig. S9a presents a TEM image of the commercially available QB, showing that the QB is slightly unstable and exhibits agglomeration. As depicted in Figs. S9b and c, the QB-based LFIA is only available in the fluorescence detection mode and has a naked eye detection limit of 0.1 ng/mL, which is 20 times lower than that of the Si-Au/DQD-LFIA. The low sensitivity is caused by fluorescence attenuation due to QB agglomeration. In addition, we carried out a comparison with the commonly used AuNP-based LFIA method to verify the colorimetric performance of the Si-Au/DQD-based LFIA. Figure 4c shows that the naked eye detection limit of the AuNP-based LFIA is 0.5 ng/mL under the same antibody pairing, which is on par with the colorimetric mode of the Si-Au/DQD-based LFIA. However, in the fluorescence mode, the Si-Au/DQD-based LFIA exhibited a 238-fold decrease in LOD compared with the conventional AuNP-based LFIA. Subsequently, we performed a comparison with the Si-Au/DQD-LFIA using a commercial ELISA kit purchased from Beijing Sino Biological Inc (Catalog# KIT40891). Figure 4d shows the results of the ELISA analysis of the A29L protein. The OD450 was recorded with a microplate reader to construct a calibration curve for the detection of the MPXV A29L protein. As shown in Fig. 4g, the LOD of the ELISA method for detecting the A29L protein was determined to be 7.1 pg/mL, which based on the IUPAC protocol. This result indicated the LOD of ELISA is about 3.3-fold higher than that of the Si-Au/DQD-based LFIA.

(a) Colorimetric (i) and fluorescence (ii) photographs of Si-Au/DQD-based LFIA strip for MPXV A29L protein detection. (b) Colorimetric (i) and fluorescence (ii) photographs of Si-Au/QD-based LFIA strip for MPXV A29L protein detection. (c) Photographs of AuNP-based LFIA strips for MPXV A29L protein detection. (d) Photographs of ELISA kit for MPXV A29L protein detection. (ii) Corresponding calibration curves for MPXV A29L protein detection. (e-f) Corresponding calibration curves of Si-Au/DQD (e), Si-Au/QD (f), and ELISA (g) for the detection of MPXV A29L protein assay

Furthermore, the commercial AuNP-based LFIA can only detect 0.5 ng/mL of the target, which is 238 times higher than that of the Si-Au/DQD-LFIA (Fig. S10). The linear range of four detection methods (Si-Au/DQD-LFIA, Si-Au/QD-LFIA, AuNP-based LFIA and ELISA) were shown in Fig. S11. In addition, the main performance indexes of the four methods for for MPXV antigen detection were summarized in Table S1. These results fully demonstrate the flexibility and highly sensitive detection capability of dual-signal LFIA. Thus, the Si-Au/DQD-based dual-mode LFIA enables the flexible readout of signals and proves its suitability for POCT in resource-limited situations.

The reproducibility and specificity of the Si-Au/DQD-based LFIA were investigated to assess its feasibility in field applications. The reproducibility of the Si-Au/DQD-LFIA was verified by detecting the intermediate and low concentrations of the A29L protein. As shown in Fig. 5a, both colorimetric and fluorescence modes showed good fluorescence signal reproducibility on the T-line when detecting the intermediate concentrations of the target. In detecting low concentrations of targets, only the T-line of the fluorescence mode showed signals and exhibited good reproducibility, indicating that the proposed dual-signal LFIA has good reproducibility in different target concentrations. In addition, the specificity of Si-Au/DQD-LFIA was evaluated against other orthopoxviruses and respiratory viruses, including the vaccinia virus (VCAV, A27), the cowpox virus (CPXV, 162), VZV, FluA, FluB, SARS-CoV-2, and HAdV [46, 47]. Figure 5b shows that only the test strip with the A29L protein detection target showed distinct colorimetric and fluorescent bands in the T-line zone, and none of the test strips for the other targets showed colorimetric and fluorescent lines, demonstrating the high specificity of the method for the A29L protein.

(a) Reproducibility of Si-Au/DQD-based LFIA. (b) Specificity of Si-Au/DQD-based LFIA. The numbers 1–8 at the bottom of the strip represent the detection targets of MPXV (1ng/mL), VCAV (100 ng/mL), CPXV (100 ng/mL), VZV (100 ng/mL), SARS-CoV-2 (106 pfu/mL), FluA (106 pfu/mL), FluB (106 pfu/mL), and HAdV (106 pfu/mL)

We further evaluated the feasibility of Si-Au/DQD-based LFIA by analyzing the MPXV A29L protein spiked pharyngeal swab samples from healthy individuals due to the lack of clinical samples. Recovery experiments were performed on pharyngeal swab samples spiked with concentrations of 1, 0.5, 0.1, and 0.05 ng/mL and are summarised in Fig. 6; Table 1. For high concentrations, the samples can be quantified directly with the naked eye, and for low concentrations, they can be measured with the aid of a UV lamp or a portable fluorescence reader (Fig. 6). Furthermore, we calculated the recovery rate in the fluorescence mode. As listed in Table 1, the spiked recoveries in the fluorescence mode were 88.8-111.5% and the coefficients of variation were 4.22-10.28%, which are consistent with the requirements for quantitative assay of the actual samples.

(a) Colorimetric/fluorescent images of the strips and the corresponding T-line signal intensities for MPXV detection (1-0.05 ng/mL). (b) The linear relationship of the fluorescence signals with MPXV concentrations