BS3 Bis(sulfosuccinimidyl)suberate (BS3), bovine serum albumin (BSA), sodium carbonate buffer, potassium phosphate buffer, phosphate buffered saline (PBS), 2-(N-Morpholino)ethanesulfonic acid (MES), N-Hydroxysuccinimide (NHS), 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), fetal bovine serum (FBS), donkey anti-sheep immunoglobulin G (IgG) polyclonal antibodies, Tris-HCl buffer, Tween-20, nitrocellulose membrane (NC membrane), porous glass fiber were bought from Sigma-Aldrich Chemical Co. (St. Louis, MO). Europium Chelate was bought from Thermo Fisher Scientific Inc. (Waltham, MA). S-PMA ELISA test kit was bought from Chemitrace Consultancy (Natland, United Kingdom).

AKTA Fast Protein Liquid Chromatograph (FPLC) System was from Cytiva (Marlborough, MA). Transmission electron microscopes (TEM) images were captured on an FEI Technai G2 20 Twin equipped with a 200 KV LaB6 electron source. Nanodrop 2000 was bought from Thermo Fisher Scientific Inc. (Waltham, MA). Fluorescent immunoanalyzer AFS-1000 was bought from Guangzhou Labsim Biotech Co., Ltd. (Guangzhou, China). BioDot BioJet BJQ 3000 dispenser, BioDot AirJet BJQ 3000 dispenser and BioDot paper cutter module CM4000 were bought from BioDot, Inc. (Irvine, CA, USA). A Safire-II multi-mode microplate reader from Tecan Group Ltd. (Zürich, Switzerland) was used for fluorescence and absorption analysis.

0.025 mmol of S-phenylcysteine, 0.025 mmol of crosslinker BS3, and 50 mg of BSA were dissolved in 5 mL of sodium carbonate buffer (pH = 9.6). The solution was stirred overnight. Subsequently, the conjugates obtained in this manner were dialyzed against 5 changes of 50 mM potassium phosphate buffer and washed with PBS for 3 days at 4 °C for subsequent use. Then, hapten-BSA conjugate was purified by FPLC. An ultraviolet (UV) detector was used to measure protein concentration at an absorption wavelength of 280 nm.

Conjugation of hapten-BSA to the resin was performed to develop an S-PMA column for the purification of S-PMA antibody. Subsequently, the S-PMA polyclonal antibody was purified using FPLC and then dialyzed. Protein concentration was measured using a UV detector at an absorption wavelength of 280 nm. The antibody was concentrated in a 50 kDa Centricon tube by centrifugation at 3000g for 10 min and filtered through a 0.22 μm filter. The optical density at A280 (OD@A280) of the antibodies, before and after dialysis, was measured using Nanodrop to determine the concentrations of S-PMA antibody.

Purified sheep anti-S-PMA antibody was covalently conjugated to the surface of carboxylate-modified EuNP as follows. First, 20 μL of EuNP (1 % w/v) and 990 μL of MES (50 mM, pH 5.5) were mixed in a 2 mL microcentrifuge tube. Then, 25 μL of 5 mg/mL NHS and 4.8 μL of 5 mg/mL EDC were added and mixed on a rocking shaker at room temperature for 30 min. Next, the solution was centrifuged at 41,657 g for 30 min, the supernatant was removed, and the resulting carboxyl-activated EuNP was resuspended in 1 mL of 10 mM PBS (pH 7.2). Afterwards, sheep anti-S-PMA antibody at a final concentration of 4 μg/mL was added into the above solution and incubated for 1 h at room temperature on a rocking shaker, to conjugate with EuNP. Then 100 μl of FBS was added and incubated for additional 15 min to block the excess binding sites. Finally, the supernatant was removed, and the remaining pellet was washed twice with PBSA (1 % BSA added) and resuspended in 10 % FBS and 10 % sucrose in 0.1 M Tris-HCl buffer (pH 8.2). The final EuNP-antibody conjugate (EuNP-Abs) (approx. 1 μg/mL) was stored at 4 °C before use.

The sample pads (17 mm × 30 cm) composed of glass fiber were saturated with 0.1 M Tris-HCl buffer (pH 8.2) containing 0.25 wt% Tween-20. The pads were then dried at 37 °C for 2 h and stored in the desiccator. Both the test lines and the control line, having the same dispensing rate of 1 μL/cm, were prepared by dispensing hapten-BSA conjugate (1.0 mg/mL) and donkey anti-sheep immunoglobulin G (IgG) polyclonal antibodies (0.5 mg/mL) at different locations on a NC membrane (25 mm × 30 cm) using BioDot BioJet BJQ 3000 dispenser. The NC membranes were then dried at 37 °C for 30 min and stored in a dry room. Subsequently, the conjugate pad (8 mm × 30 cm) was pretreated with blocking buffer consisting of 0.1 M Tris-HCl with 1 wt% Tween-20, pH 8.2. The conjugation dilution buffer was dispensed onto the conjugate pad using BioDot AirJet BJQ 3000 dispenser. To achieve optimum performance and good repeatability of conjugate pad, the airjet dispenser should be set to a dispensing speed of 15 mm/s with a rate of 1 μL/cm. The dispenser pressure was set to 10 psi. After dispensing, the conjugate pads were then dried at 37 °C for 30 min and stored in dry room for further use. The different pads were assembled on a backing (60 mm × 30 cm) with an overlap of approximately 1–2 mm between them to ensure solution could migrate through the LFIA test strip. The LFIA test strip was cut at a width of 3.5 mm using BioDot paper cutter module CM4000.

In a typical assay, the sample was spiked into 200 μL of sample dilution solution (0.05 M Tris-HCl, 0.5 wt% BSA, 0.5 wt% Tween-20) to prepare the spiked buffer. The solution was loaded onto the sample pad of the LFIA test strip. The solution then migrated through the porous glass fiber pads and NC membrane towards the absorbent pad by capillary action. After 15 min, the LFIA test strip was inserted into the lateral flow strips reader, which recorded either the fluorescent intensity or the intensity of test line to the intensity of control line (T/C) ratio.

Serial dilutions of S-PMA were spiked into the sample dilution solution to achieve final concentrations of 0.1, 0.25, 0.5, 1, 1.25, 2.5, 5, 10, 15, 20, and 30 ng/mL in the spiked buffer. The sample solution was also used as the blank. In a typical test, 200 μL of sample dilution solution containing various concentrations of S-PMA was loaded onto the sample pad of LFIA test strip. The solution then migrated through the porous glass fiber pads and NC membrane towards the absorbent pad by capillary action. After 15 min, the LFIA test strip was inserted into the lateral flow strips reader, which recorded either the intensity or the T/C ratio. Additionally, standard curves were generated based on T/C ratio and S-PMA concentrations.

The stability test was performed according to the method of Song et al. [40], with slight modifications. In the stability test, the stability of the LFIA is demonstrated through aging experiments. Briefly, to accelerate aging study, the strip was stored in a sealed plastic bag which was filled with silica bead desiccant at 50 °C. The aging experiment lasted for nine weeks, and the T/C ratios of the samples and the recoveries of urine and serum samples were recorded weekly.

The LFIA for S-PMA analysis employs a competitive format (Fig. 2). The hapten-BSA conjugate serves as the capture antigen on the test line, while donkey anti-sheep IgG polyclonal antibodies are utilized as the secondary antibodies on the control line of LFIA strips. Before the test, capture antibody conjugates (purified sheep anti-S-PMA labelled with EuNP, noted as Eu) were preloaded onto the conjugate pad. In the absence of S-PMA in the running buffer, the sample migrates towards the absorbent pad through capillary action, carrying EuNP-antibody conjugates. These conjugates bind to the capture antigen on the test line, generating a strong fluorescent signal. Any surplus EuNP-antibody conjugates bind to the secondary antibody on the control line. However, in the presence of S-PMA, it competes with hapten-BSA conjugate on the test line for binding to the EuNP-antibody conjugates. This competition inhibits the binding of EuNP-antibody conjugates to the test line, resulting in reduced signal intensity. Consequently, the signal inhibition on the test line is indicative of S-PMA presence. In other words, the measured fluorescent signal strength is inversely proportional to the concentration of S-PMA.

Fig. 3 illustrates the FPLC chromatogram of hapten-BSA conjugate after FPLC superdex purification. The narrow peak located at around 50 mL indicates the highly purified fraction of hapten-BSA conjugate, while the broader peak at 75 mL corresponds to the unreactive component.

Fig. 4 presents the FPLC curve of S-PMA antibody, with the peak at 100 mL denotes the large amount of antibody purified by S-PMA column, demonstrating excellent specificity and purification efficiency of S-PMA column. The OD@A280 value of S-PMA polyclonal antibody after dialysis was 5.56, revealing that the concentrated concentration of the polyclonal antibody obtained was 4.06 mg/mL.

The morphology of the EuNP observed using transmission electron microscopy (TEM) is depicted in Fig. 5. It reveals a homogeneous sphere structure with an approximate size of 101 ± 3 nm, consistent with findings in previous papers [42,43]. Additionally, TEM images show polystyrene shell structures with an approximate thickness of 15 nm. The incorporation of europium chelates into polystyrene shells is aimed at enriching the fluorescent signal and enhancing resistance to the quenching effects of the matrix. Similar core-shell nanocomposites have been reported using this approach, including Fe₂O₃@polystyrene [44], SiO₂@polystyrene [45] and encapsulated quantum dots [46]. Fluorescent analysis confirmed the large Stokes shift (~273 nm) with a narrow emission waveband (peak at ~617 nm) and a wide excitation waveband (peak at ~344 nm) (Fig. 6), indicating reduced background noise and higher achievable sensitivity in bioimaging settings [47].

Several factors influencing LFIA performance were investigated, as illustrated in Fig. 7. This study primarily investigated four factors: hapten-BSA conjugate concentration, EuNP-Abs concentration, reaction time and buffer pH. Previous studies have demonstrated that the ratio of the fluorescence intensity between the test line and control line (T/C ratio) on the test strip, measured simultaneously, can compensate for background factors [48]. Therefore, the T/C ratio was utilized to determine the results in both previous studies and our current research [49,50]. As shown in Fig. 7 (A), the T/C ratio significantly increased with the escalating dispensing hapten-BSA conjugate concentration from 0.2 to 1.0 mg/mL and when hapten-BSA conjugate concentration reached 1.0 mg/mL, the T/C ratio stabilized and was maintained to 1.4 mg/mL. Moreover, Fig. 7 (B) demonstrates that the T/C ratio increased with the rise in EuNP-Abs concentration on conjugate pads, reaching its maximum at an EuNP-Abs concentration of 1.0 μL/cm. Reaction time is also proved to be a critical factor affecting LFIA performance [51]. The T/C ratio gradually reached maximum around 15 min, indicating that the optimal readout time is 15 min (Fig. 7 (C)). To achieve the highest immunoreaction efficiency, the pH of detection buffer was optimized, as shown in Fig. 7 (D). Buffers with pH ranging from 5 to 10 were prepared. The T/C ratio was highest at a buffer pH of 8, indicating the optimal buffer pH for immunoreaction is 8. Subsequent studies were all conducted under optimal conditions.

The proposed LFIA approach was applied to determine S-PMA levels in the spiked buffer under the optimal conditions described above. Fig. 8A shows that LFIA strips images with spiked buffer of varying S-PMA levels. Due to the competitive binding method, the higher the S-PMA concentration in the sample, the weaker the fluorescence signal on the test line of the test strip. In Fig. 8B, the calibration curve shows the relationship between the T/C ratio of strips and the S-PMA concentrations of spiked buffer. As the concentration of S-PMA increases, a gradual attenuation of red fluorescence signal becomes evident under a UV (365 nm) lamp. This observation aligns the detection mechanism of competitive immunoassay. Quantitative measurement of the fluorescent signal from the test lines confirmed a reverse correlation. The linear range of the detection for S-PMA was from 0.25 to 30 ng/mL with a linear relationship of Y = −0.441 log (C_S-PMA) + 0.659 where Y is the T/C ratio and the C_S-PMA is the concentration of S-PMA. The coefficient of determination (R²) is 0.980. The calculated limit of detection (LOD) was determined to be 0.32 ng/mL, based on the three times signal-to-noise ratio standard. Previous study has shown that the concentration of S-PMA in human urine ranges from 0.934 to 14.0 ng/mL, which indicates that the LFIA introduced in this study can effectively detect S-PMA levels in human urine [21]. The American Conference of Governmental Industrial Hygienists recommends a biological exposure index for S-PMA of 25 μg/g creatinine, while European Chemicals Agency sets a biological limit values for S-PMA at 2 μg S-PMA/g creatinine [52]. Additional, the concentration of creatinine in human urine typically ranges from 80 to 200 μg/dL [53]. The LOD of LFIA is significantly lower than the above standards, which demonstrates the feasibility of its practical application. Certain urinary metabolites with structural similarities to PMA may interfere with its detection, potentially impacting data interpretation [24,55]. However, Aston et al. [24], reported that the PMA immunoassays are not affected by benzylmercapturic acid, a urinary metabolite of toluene. This finding is particularly relevant given the frequent co-exposure to benzene and toluene. Additionally, hippuric acid, a non-specific metabolite of toluene with low structural similarity to PMA, did not cause significant interference in PMA immunoassays.

The detection range and LOD of various approaches for detecting S-PMA are shown in Table 1. Previous studies indicate that conventional detection approaches such as LC-MS, LC-MS/MS GC-MS and HPLC are widely used for S-PMA detection [13,54,55]. However, the extraction and purification process prior to detection make these approaches time-consuming and the expensive equipment makes detection costly [56–58]. At the same time, many novel detection approaches, such as fluorescence quench and surface enhanced Raman spectroscopy (SERS), have been developed, but they struggle to match the detection accuracy of conventional approaches [26,59]. In comparison, the LFIA in this work reduced the LOD of S-PMA to levels comparable with conventional approaches, while greatly simplifying the operation steps, achieving rapid, cheap and accurate detection of S-PMA.

Product stability is a key indicator of LFIA performance andthe stability test results are shown in Fig. 9. According to the accelerated aging calculator, aging at 50 °C for 7 weeks is functionally equivalent to one year at room temperature. The accelerated aging method is based on the Arrhenius reaction rate function, which assumes that a 10 °C increase in temperature will cause the reaction rate to double. The results indicate that LFIA performance remained relatively stable over a nine-week period at 50 °C. The inhibition rate (B/B₀) is often used to reflect the detection performance of LFIA [61,62]. Here, the T/C ratio of the S-PMA spiked samples was defined as B, and the T/C ratio of the control sample is defined as B₀ [63]. In this study, B/B₀ is used to assess the stability of LFIA detection. The result of B/B₀ can not only demonstrate the LFIA’s ability to detect different levels of S-PMA, but also reflect the proportional relationship between samples with different S-PMA levels and the control sample. In stability testing, LFIA maintained performance above 90 % after nine weeks of aging. Specifically, when measuring S-PMA levels at 0.5 ng/mL, the LFIA’s detection performance decreased by only 6.9 % after nine weeks of storage, demonstrating stable performance. Furthermore, when testing various concentrations of S-PMA, LFIA produced relatively stable results at different aging times. These fundings indicate that LFIA can maintain relatively stable detection performance during the aging test period. In summary, the aging test results show that LFIA can be stored at room temperature for one year while maintaining reliable analytical detection performance. Additionally, the average relative standard deviation (RSD) of the test data from ten LFIA test strips in the same batch is 2.54 ± 0.763 %, while RSD of LFIA test strips across three batches is 2.40 ± 1.28 %. The intra- and inter-batch differences are well within the acceptable range for a biological assay (limit of 20 %) [64].