The characterization of CN-EUs and coupled CN-EUs were analyzed using Zetasizer software (version 7.13, Macromedia Inc., San Francisco, CA, United States). The particle size and Zeta potential of CN-EUs are shown in Figures 1A,B. The average particle size of uncoupled CN-EUs and those conjugated with sheep IgG (SIgG) or anti-VAN polyclonal antibodies (PcAbs) were 192.7, 241.1 and 218.8 nm with polymer dispersity index (PDI) of 0.008, 0.024 and 0.025, and the Zeta potential of those were −65.0, −25.4, and −29.7, respectively. There were significant differences in particle size and Zeta potential between conjugated and uncoupled nanoparticles.

As shown in Figure 1C, using unconjugated protein as a blank control, the purified VAN conjugated antigen was identified by ultraviolet-visible spectroscopy (UV-VIS). The characteristic absorption peak of VAN is at 280 nm. The absorbance curves of VAN-human albumin (HSA) synthetic and the mixture of VAN and HSA were also showed an absorption peak at 280 nm, which indicated that VAN was linked with the carrier protein successfully.

The test strips are assembled and cut as shown in Figure 2A. The LFIA for testing VAN concentration was performed as a typical competitive time-resolved fluoroimmunoassay, illustrated in Figures 2B,C. When the sample buffer containing VAN was added to the sample pad, the mixture migrated to the conjugated pad by capillarity and partially combined with CN-EUs labeled anti-VAN PcAbs. After the complexes flow reached the NC membrane, the uncombined nanoparticles labeled anti-VAN PcAbs were captured by the test line once they had arrived, while the nanoparticles labeled SIgG were captured by control line, respectively. Subsequently, the superfluous fluorescent nanospheres migrated into the absorption pad. The test strip was then measured using a time-resolved fluorescence (TRF) reader to obtain the peak heights of the test line and the control line. In the competitive assay system, the VAN-HSA coating on the test line competed with the VAN in the sample for binding to the conjugates of CN-EUs with anti-VAN PcAbs, with the result that if there os more VAN in the sample, the lower fluorescent signal intensity appears on the test line. Therefore, an inverse relationship occurred between the fluorescence intensity of the test line and VAN concentration in the sample. On the other hand, as an internal control to confirm that the sample had migrated the lines and reacted correctly, the fluorescence intensity of the control line was almost constant, regardless of VAN concentration in the sample. Conclusively, the ratio of H_T/H_C was used as the assay result, which could counteract the intrinsic heterogeneity of the lateral flow test strip and the sample matrix, showing more reliability and reproducibility than when only H_T was used for signal quantitation.

The reaction time is a significant factor that heavily influenced the fluorescence intensity variation in the LFIA. In this case, a VAN standard sample of 0 ng*ml⁻¹ was used to evaluate the effect of reaction time by monitoring H_T, H_C, and H_T/H_C ratio. As illustrated in Figure 3A, over the range of 3–30 min of incubation, each value was based on ten replicated measurements and assessed per 3 min. The H_T and H_C sharply rose and reached their peak values in 12 and 6 min, respectively. Subsequently, the H_T stood at equilibrium for 3 min and then declined slowly, while the H_C fell immediately once reaching its peak. As shown in Figure 3B, the H_T/H_C ratio dropped continually during the reaction, and the minimum standard deviation presented at 15–18 min indicating that the H_T/H_C ratio was most stable. Considering both stability and timeliness, 15 min was selected as the incubation time for further studies.

The dilution test was performed to find the reliability of the assay using gradual serial dilution over the range from 1/2 to 1/32 with a sample buffer for three positive sera samples, and then analyzed according to the previous method. As shown in Figure 4, all three sample dilutions exhibited good linearity with an R square above 0.99, indicating little variation in the observed concentrations after correcting for sample dilution. This gives evidence that the proposed LFIA method was suitable for quantitative measurements.

The standard curve for the proposed LFIA was constructed based on the measurement of a series of different concentrations of VAN standards (0, 100, 1,000, 3,000, 5,000, 10,000, 30,000, 50,000, and 80,000 ng*ml⁻¹), which were accurately prediluted using a sample buffer. According to the fluorescence intensities recorded using a TRF strip reader as shown in Figure 5A, we obtained the standard curve by plotting the logit (Y) against the logarithm of the VAN concentration as represented by the equation: logit (Y) = 7.012–1.914*log(X), with a reliable coefficient of determination (R ² = 0.982). The dose-response curve was displayed throughout the whole range of VAN concentration, as shown in Figure 5B. The analytical sensitivity (limit of detection), defined as the concentration corresponding to the mean minus 2*SD (n = 15) of the H_T/H_C ratio of the zero standard, was calculated to be 69.2 ng*ml⁻¹.

The reproducibility of the developed LFIA was assessed according to the intra-assay (within a day) and inter-assay (between days) precision. As shown in Table 1, the intra- and inter-assay are 7.12–8.53% and 8.46–11.82%, respectively. All the CVs are around 10%, which indicates an acceptable level of precision for the VAN strip quantification. The specificity was evaluated by the examination of possible interferents at relatively high concentrations using the proposed method, including teicoplanin, penicillin, and cephalosporins. The cross-reactivity was calculated using the formula: cross-reactivity (%) = (measured concentration of VAN)/(expected concentration of interferent). As shown in Table 2, the results demonstrated that the developed polystyrene CN-EUs-based LFIA had a high specificity towards VAN.

In order to demonstrate the clinical application of the polystyrene CN-EUs-based LFIA system, 19 clinical serum samples were analyzed using the proposed method and LC-MS/MS simultaneously. The LC-MS/MS calibration curve is presented in Figure 6A, with the ratio of analyte peak area/IS peak area as the y-axis and the analyte concentration as the x-axis with an R square of 0.9988. The linear correlation between the two methods is shown in Figure 6B (R ² = 0.9713). The results indicate the LFIA has a satisfactory analytical performance and is comparable with the LC-MS/MS method for the determination of VAN concentration in human serum samples.

VAN (hydrochloride) (1404-93-9) was obtained from MedChem Express (Monmouth, NJ, United States). Anti-VAN PcAbs (TB2526741) were purchased from Invitrogen (Carlsbad, CA, United States). SIgG Fc fragment (013–0105) and a rabbit-anti-sheep IgG (H&L) (anti-SIgG) (613–4102) were purchased from Rockland Immunochemical Inc. (Limerick, PA, United States). Bovine serum albumin (BSA) was purchased from Roche Diagnostics (Indianapolis, IN, United States). Methanol (MS grade) and CN-EUs were purchased from Thermo Fisher Scientific Inc. (Waltham, MA, United States). Trehalose was obtained from Wako Pure Chemical Industries, Ltd. (Chuo-ku, Osaka, Japan). Sucrose was purchased from Macklin Biochemical Co., Ltd. (Shanghai, China). Sample pads (Ahlstrom 8964) and absorbent pads (H5015) were purchased from Jieyi Biotechnology (Shanghai, China). NC membranes (Hi-Flow Plus HFC13502), conjugate pads (GFCP203000), 0.22 μm syringe filters, and centrifugal filter units with an Ultracel-50 membrane were obtained from Millipore (Bedford, MA, United States). Triton-100, Proclin-300, Polyvinyl alcohol (PVA, average mol wt 30,000–70,000), Polyvinyl pyrrolidone (PVP, average mol wt 10,000), Casein-Na, 4-morpholineethanesulfonic acid (MES), 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysulfosuccinimide (sulfo-NHS), and human albumin (HSA) were purchased from Sigma-Aldrich (St. Louis, MO, United States). All other chemicals were of analytical reagent grade.

Ultrapure water used throughout the study was obtained through a Milli-Q water purification system (Millipore, Bedford, MA, United States). A probe sonicator, UP200S, was purchased from Hielsher (Teltow, Germany). A HulaMixer sample mixer was purchased from Invitrogen (Carlsbad, CA, United States). A zetasizer Nano-ZS90 was obtained from Malvern Panalytical Ltd. (Malvern, Britain). An ultra-microspectrophotometer was purchased from DeNovix Inc. (Wilmington, DE, United States). A BioJet Quant XYZ3060 dispenser was obtained from Biodot Ltd. (Irvine, CA, United States). A draught drying cabinet was purchased from Shelton Manufacturing, Inc. (Cornelius, OR, United States). A strip cutter was purchased from Kinbio Tech. Co., Ltd. (Shanghai, China). An aQcare TRF strip reader was obtained from Medisensor, Inc. (Daegu, Korea).

The buffer solutions used in this study are as follows: sample pad treatment buffer (100 mmol*L⁻¹ Na₂B₄O₇.10H₂O, 0.2% Casein-Na (wt/vol), 1% PVP (wt/vol), 0.1% NaN₃ (wt/vol) and 6% TritonX-100 (vol/vol)); conjugate pad treatment buffer (50 mmol*L⁻¹ Na₂HPO₄.12H₂O, 0.5% BSA (wt/vol), 0.5% PVA (wt/vol) and 1% TritonX-100 (vol/vol), pH 7.4); coating buffer (10 mmol*L⁻¹ Na₂HPO₄.12H₂O, 0.3% Trehalose (wt/vol), 0.9%NaCl (wt/vol) and 0.1%NaN₃ (wt/vol), pH 7.4); activating buffer (25 mmol*L⁻¹ MES, pH 6.1); binding buffer (25 mmol*L⁻¹ phosphate buffer, pH 7.0); blocking buffer (25 mmol*L⁻¹ phosphate buffer, 5% BSA (wt/vol), pH 7.4); washing buffer (25 mmol*L⁻¹ Tris, 0.9% NaCl (wt/vol), 0.05% Proclin-300 (vol/vol) and 0.2% Tween-20 (vol/vol), pH7.8); labeling antibody storage buffer (25 mmol*L⁻¹ Tris, 5% BSA (wt/vol), 1% Trehalose (wt/vol), 1% sucrose (wt/vol), 0.9% NaCl (wt/vol), 0.05% TWEEN-20 (vol/vol) and 0.05% Proclin-300 (vol/vol), pH 7.2); labeling antibody dilution buffer (20 mmol*L⁻¹ Tris, 1% BSA (wt/vol), 5% Trehalose (wt/vol), 20% sucrose (wt/vol) and 0.05% Proclin-300 (vol/vol), pH 9.0); sample buffer (10 mmol*L⁻¹ Na₂HPO₄.12H₂O, 1%BSA (wt/vol), 0.9%NaCl (wt/vol), pH 7.4); PBA (10 mmol*L⁻¹ Na₂HPO₄.12H₂O, 2 mmol*L⁻¹ KH₂PO₄.12H₂O, 0.8%NaCl (wt/vol) and 0.02%KCl (wt/vol), pH 7.4). All solutions were freshly prepared and filtered using 0.22 μm syringe filter before use.

The sample pads and conjugate pads were important ingredients of the CN-EUs-based LFIA, and both were made from glass fiber. The sample pads were cut into 300 × 17 mm pieces and saturated with sample pad treatment buffer for 2 h at room temperature. Then the pieces were dried at 37°C for 24 h in a draught drying cabinet and stored in an electronic moisture-proof tank at room temperature until use. The conjugate pads were sliced into 300*10 mm pieces and submerged in conjugate pad treatment buffer for 1.5 h at room temperature. The subsequent processing and storage conditions are the same as the sample pads.

The conjugates of nanoparticles and anti-VAN PcAbs/SIgG were prepared using the method described in our previous works, with only minor changes. Initially, 2 mg of CN-EUs (particle size: 199 nm) were centrifuged at 20,600 g for 15 min at 4°C to remove supernatant stock solution before being activated at room temperature in 500 μl activating buffer including EDC and sulfo-NHS with an ultimate concentration of 1.25 mmol*L⁻¹ and 10 mmol*L⁻¹, respectively. After 30 min gentle shaking at room temperature, the interactant was centrifuged at 29,700 g for 30 min at 4°C and washed twice with l ml washing buffer each time. The sediment was then resuspended in 400 μl binding buffer by sonication. Subsequently, 25 μg of specific anti-VAN PcAbs/SIgG was purified and concentrated into 100 μl binding buffer by centrifugal filter unit and an Ultracel-50 membrane was added, and then the coupled reaction proceeded for 2 h at room temperature with gentle upside-down mixing by sample mixer. The supernatant containing unlabeled antibodies was removed by centrifugation at 20,600 g for 15 min at 4°C and washed twice. Then the mixture was centrifuged again in the previous conditions and the supernatant was aspirated. The precipitate was resuspended with 1 ml blocking buffer and incubated for another 2 h at room temperature with mild shaking. It was then washed three times, and the conjugates that settled at the bottom were blended with labeling antibody storage buffer at a CN-EUs concentration of 2 mg*ml⁻¹. After verification of the conjugated CN-EUs, the conjugate solution containing the CN-EUs coupled with anti-VAN PcAbs and SIgG with a final concentration of 0.15 mg*ml⁻¹ and 0.03 mg*ml⁻¹ was sonicated with a probe at 0.5 cycle and 40% amplitude for 5 min and then distributed on a conjugate pad by the BioJet Quant XYZ-3060 dispenser at a rate of 10 μl*cm⁻¹. The pad was dried again and stored in a moisture-proof cabinet at room temperature.

The competitive antigens was designed to use HSA as a carrier protein and were prepared using the following method. First, 10 mg of HSA was dissolved in 500 μl of activating buffer including EDC and sulfo-NHS with a final concentration of 1.25 mmol*ml⁻¹ and 10 mmol*ml⁻¹, respectively, and uniformly mixed at room temperature for 30 min. The activated mixture was then centrifuged at 9,000 g for 10 min at 4°C by centrifugal filter unit with an Ultracel-50 membrane and washed using a binding buffer three times. After being collected and adjusted to 500 μl with the binding buffer, the supernatant was mixed with 1 mg of VAN dissolved in 500 μl of binding buffer and gently rotated overnight at 4°C. Subsequently, the synthesis was dialyzed against PBS at 4°C for 24 h twice. The competitive antigen, VAN-HSA, was identified by ultraviolet-visible spectroscopy (UV-VIS) and then stored at 4°C until use.

The CN-EUs-based LFIA is composed of five ingredients: sample pad, conjugate pad, absorbent pad, NC membrane, and backing plate. Initially, after being equilibrated to room temperature, the competitive antigen and anti-SIgG were concentrated by centrifugation for a final concentration of 3 mg*ml⁻¹ and 1 mg*ml⁻¹, respectively, and spotted on the NC membrane using the dispenser at a rate of 0.8 μl*cm⁻¹, being separated by a distance of 5 mm to serve as the test line and control line respectively. The membrane was then dried overnight at 37°C and stored in a moisture-proof cabinet. An absorption pad was cut into 300*22 mm strips without any other treatment. All four components were assembled on a 300*60 mm backing plate, so that the components overlapped sequentially to ensure a direct flow from the sample pad to the absorbent pad by capillarity. The whole plate was then cut into 3 mm wide strips using a strip cutter, as shown in Figure 2A. Each strip was packaged into a shell with a circular sample pad well and a rectangular viewing area marked with the test line and control line. Finally, the wrapped strips were sealed in ziplock plastic bags with desiccant and stored at room temperature for future use.

The VAN was dissolved in ultrapure water at a concentration of 100 μg*ml⁻¹ to obtain a working solution. The working solution was then diluted with a sample buffer to concentrations of 0, 100, 1,000, 3,000, 5,000, 10,000, 30,000, 50,000, and 80,000 ng*ml⁻¹ as VAN standards, and diluted with a blank serum to give final concentrations of 100, 1,000, and 10,000 ng*ml⁻¹ as quality control samples (QCs) of low, medium and high concentrations, respectively. A total of 19 patients’ serum samples was provided by Nan-fang Hospital (Guangdong, China). All samples were stored at −80°C after immediately after acquisition and unfrozen at 4°C only before use. The study was approved by the Ethical Committee of the Science and Technology Department of the Southern Medical University.

The proposed LFIA for assessing VAN was performed as a competitive time-resolved fluoroimmunoassay. First, 5 μl of the standards or samples was added to 300 μl of the sample buffer and blended thoroughly. Subsequently, 60 μl of the mixture was loaded onto the sample pad well and migrated towards the absorption pad by capillarity as shown in Figure 2B. The shell containing the test strip was then inserted into an aQcare TRF strip reader after 15 min of reaction and the fluorescence value at 613 nm wavelength of each line and the H_T/H_C ratio was measured under the excitation light of 333 nm wavelength. The fluorescence lateral flow procedure and performance measurement results are shown in Figures 2C, 5A.

The mean and standard deviations (SD) of the fluorescence assay were measured at the H_T/H_C ratio of zero standard on the response curve for 15 duplicates. Sensitivity was calculated as the concentration that corresponds to the value of mean minus double SD. The intra- and inter-assay precisions were obtained from analyzing three QC samples prepared using a negative serum with concentrations of 100, 1,000, and 10,000 ng*ml⁻¹, respectively. The QCs were tested ten times per day for intra-assay precision and five duplications for three sequential days for inter-assay.

The LC-MS/MS method was based on several previous reports with only minor adjustments, using tobramycin as an internal standard (IS) (Brozmanová et al., 2017). The VAN was dissolved and diluted using methanol for final concentrations of 0, 5, 10, 20, and 60 μg*ml⁻¹ as a calibration solution. Each calibration was spiked with IS to get a final concentration of 10 μg*ml⁻¹. Subsequently, 50 μl of calibration solution was aspirated and mixed with 50 μl methanol, 60 μl 33% trichloroacetic acid, 200 μl H₂O, 50 μl acetonitrile, and 50 μl 0.5 mol*L⁻¹ NH₄OH. After gentle oscillation blending, the mixture was centrifuged at 18,000 g for 10 min at 4°C. Then 50 μl supernatant was combined with 25 μl acetonitrile and analyzed on API 3200 triple quadrupole tandem mass spectrometers. The results were analyzed with ABI Analyst Software.

The VAN dose-response curve was obtained by plotting the logit-log against the logarithm of the VAN concentration (X) using GraphPad Prism 6. The H_T/H_C ratios of the zero VAN standard and the other VAN standards were defined as R₀ and R_x, respectively. A logit-log plot was acquired from the computational formula: ln(Y) = ln [(R_x/R₀)/(1-R_x/R₀)] and the fitting of the line of best fit was ln(Y) = A+ B * log(X). The data analysis was accomplished using SPSS 23.0 (SPSS Inc., Chicago, IL, United States). p < 0.05 was considered statistically significant.