*3.3. QD-Based LFIA*

The scheme of QD-based LFIA is demonstrated in Figure 1b. For QD-based LFIA, the selection of working membranes aimed to decrease background fluorescence was carried out together with the optimization of specific reagen<sup>t</sup> concentrations described above. For this purpose, CNPC SS12 12/15 μ (Advanced Microdevices), HF120, and HF180 (Millipore) membranes differing in pore size and flow rates were tested. Figure 4 indicates that the use of CNPC SS12 (of a 12 and 15 μm pore size, respectively) leads to the formation of background coloration over its entire surface. Testing of Millipore membranes with different pore sizes and flow rates demonstrated that the application of the Millipore HF180 membrane facilitated achieving the maximum analytical signal intensity (as opposed to a 20% reduction in intensity when using a HF120 membrane), eliminating nonspecific binding, and ensuring uniform movement of samples.

**Figure 4.** Images of the test zones after the LFIA performed using CNPC SS12 and Millipore HF180 membranes. The LIN-BSA conjugate concentration was 0.15 mg/mL.

The next stage of assay optimization was to select the optimal reaction medium that would decrease nonspecific binding and provide a higher signal intensity. The use of PBST as a buffer solution for QD-based LFIA led to nonspecific binding of the antibody–QDs conjugate and background staining of the working membrane. For LIN detection, significantly higher signal intensities were obtained with BB. It is acknowledged that, for the effective elution of the antibody–QDs conjugate and its movement along the membrane, detergents must be added to the buffer [26]. It was shown that the addition of Tween-20 (0.05%), BSA (1%), and sucrose (0.1%) to BB eliminated the nonspecific sorption of QD-labeled antibodies in the test zone and increased the intensity of the analytical signal by 15%. BSA and sucrose were added to the buffer to reduce the flow rate (due to a viscosity increase) and, hence, to maximize the contact time of the sample with the labeled antibodies. Furthermore, the use of BSA allows blocking the sites of nonspecific sorption of the conjugate [34].

An antibody–QDs conjugate solution was applied to the interface of the sample pad and working membrane at a volume of 1 μL and a concentration range of 0.09 to 0.28 μM. The 0.09 μM conjugate concentration was shown to give the optimal fluorescence intensity. A further decrease in the concentration of the antibody–QDs conjugate led to a drop in the signal amplitude and a decrease in the reproducibility of the test results.

Figure 5 shows the calibration curve for LIN detection in the optimized LFIA. The instrumental LOD was 0.2 ng/mL and the dynamic linear range was 0.6–10.4 ng/mL. The visual LOD was 20 ng/mL. QD-based LFIA can provide results in 15 min.

**Figure 5.** Calibration curve of LIN in the QD-based LFIA and image of the test strips with increasing concentration of LIN ranging from 0.1 to 100 ng/mL under following optimal conditions: The LIN-BSA conjugate was applied at the test line at a concentration of 0.15 mg/mL; 1 μL of antibody–QDs conjugate (0.09 μM) was applied to the interface of the sample pad and nitrocellulose membrane. The error bars indicate the standard deviations for three measurements.

### *3.4. SERS-Based LFIA*

The principle of the SERS-based LFIA is illustrated in Figure 1c. In this study, the bioconjugates of the anti-LIN antibody and AuNPs functionalized with 4-MBA were applied as both a Raman reporter bioprobe and a detection probe. In this study, 4-MBA was chosen as the reporter molecule because of its ability to provide a strong binding to the AuNP surface and high SERS-signal while being in proximity with a metal surface (enhancement factor up to 1 × 107) [35]. To obtain the optimal characteristics of the test-system, the following parameters were optimized: The amount of the LIN–BSA conjugate immobilized on the test line; the amount of the Raman reporter bioprobe. After immersing the test strips in 100 μL of the LIN solutions, visual staining was detected on the test line after 15 min. According to the competitive format of assay, the color intensity and, consequently, the SERS signal provided by the AuNPs-MBA-Ab probe is inversely proportional to the LIN concentration. As shown in Figure 6, the SERS spectra of the Raman reporter bioprobe are characterized by two intense peaks at 1077 cm<sup>−</sup><sup>1</sup> and 1580 cm<sup>−</sup>1, which in conformity with spectral data for the MBA molecule [36] correspond to vibrations in the aromatic ring. Therefore, this confirms the specific binding of reporter bioprobe to the LIN-BSA conjugate on the test line. The highest peak at 1077 cm<sup>−</sup><sup>1</sup> was used further to quantify the antibiotic content. As follows from the spectra corresponding to different LIN concentrations, the SERS intensity gradually decreases at 1077 cm<sup>−</sup><sup>1</sup> with an increase in the concentration of antibiotics.

**Figure 6.** SERS spectra arising from MBA on a test line for various LIN concentrations following the LFIA procedures under optimal conditions: The amount of AuNPs–MBA–Ab bioprobe was 2 μL; the amount of LIN–BSA conjugate was 0.2 mg/mL.

To investigate the impact of the Raman reporter bioprobe, AuNPs–MBA–Ab in amounts ranging from 1 to 5 μL were dotted on a sample pad of the strip (Figure 7a). For the coating antigen immobilized on the test line, LIN–BSA with concentrations ranging from 0.2 to 0.5 mg/mL were investigated, respectively (Figure 7b). The amount of these components is shown to have no significant influence on assay sensitivity (Figure 7).

**Figure 7.** Calibration curves of the SERS-based LFIA of LIN for various LIN–BSA conjugate and AuNPs–MBA–Ab bioprobe amounts. (**a**) The amount of AuNPs–MBA–Ab bioprobe was 1 μL (1) and 4 μL (2); (**b**) the amount of LIN–BSA conjugate was 0.2 mg/mL (1) and 0.5 mg/mL (2). B and B0 (B0 ≈ 1000 a.u.) correspond to the SERS intensities of MBA at 1077 cm<sup>−</sup>1, when standard and zero LIN solutions were applied to the sample pad, respectively. The error bars indicate the standard deviations for three measurements.

When the LIN–BSA concentration and a loading of an AuNPs–MBA–Ab bioprobe increased, the Raman intensities decreased. In contrast, a decrease in the amount of Raman reporter bioprobe and the concentration of the immobilized LIN–BSA conjugate allowed for the expansion of the dynamic range of the detected LIN concentrations. Therefore, the optimal amount of LIN–BSA conjugate was 0.2 mg/mL and the amount of AuNPs–MBA–Ab bioprobe added to the strip was 2 μL. The results shown in Figure 8 indicate that the dynamic linear range of the SERS-based LFIA varies from 2.8×10−<sup>6</sup> to 10 ng/mL with a detection limit of 1.4×10−<sup>6</sup> ng/mL. Such improved analytical characteristics can be explained by the high sensitivity of SERS detection toward a reporter molecule, which allows for simultaneous reduction in the amount of immobilized LIN-BSA conjugate and the Au-MBA-Ab reporter bioprobe. Not enough attention is being paid to the development of LFIA for LIN. According to the data shown in Table 1, most studies are devoted to the development of conventional colorimetric LFIAs for the detection of LIN, including instances where LIN is part of a panel of antibiotics (multiplex assay format).

**Figure 8.** The calibration curve of the SERS-based LFIA for LIN detection according to the following optimal assay conditions: The amount of AuNPs–MBA–Ab bioprobe was 2 μL; the LIN–BSA conjugate concentration was 0.2 mg/mL. The error bars indicate the standard deviation of the Raman intensities from MBA reporter molecule at 1077 cm<sup>−</sup><sup>1</sup> measured from 10 points along the middle of the test line.


**Table 1.** Lateral flow immunoassay (LFIA) tests for lincomycin detection.

Abbreviations: AuNPs—gold nanoparticles.

Replacing AuNPs with fluorescent microspheres made it possible to achieve a small gain in the assay sensitivity [23]. However, a significant decrease in the detection limit of LIN was achieved in our previous study (up to 8 pg/mL) when an indirect LFIA was implemented. In this study, the implementation of SERS readout technique in AuNP-based LFIA using the same immunoreagents revealed an approximately three orders of magnitude improvement in assay sensitivity. It should, however, be pointed out that such high sensitivity is resulted not only from the implementation of effective readout techniques but also from the excellent characteristics of the applied antibodies, the affinity of which was 1.15 × 10<sup>9</sup> M−<sup>1</sup> (according to information provided by the manufacturer). To date, the integration of immunoassay, in particular LFIA, with the SERS detection technique for the development of highly sensitive quantitative test systems is just getting started. Nevertheless, a number of studies prove the effectiveness of this method for the determination of target analytes at low concentrations [30,40–42]. The current study demonstrates excellent performance of the test system reached by the integration of the SERS technique with LFIA, and this integration may be considered as a potential tool for sensitive screenings of antibiotics. However, SERS technique, which is attractive because of its ability to detect extremely low analyte concentrations, is difficult to classify as point-of-care testing unless it is a handheld Raman reader format. As a result of the steps taken to design simple and portable SERS-based LFIA readers, the commercial availability of such devices may be expected in the future [43,44]. On the contrary, AuNP- and QD-LFIAs are fast and cheap out-of-laboratory techniques available today, the results of which can be quantified even using smartphones or handheld readers [45,46]. Therefore, the choice of appropriate technique is influenced by the object of study.
