Development of an immunochromatographic lateral flow strip test for the rapid detection of diclofenac in medicinal wine

ABSTRACT

A rapid immunochromatographic lateral flow strip test (ILFST) of competitive format for the detection of diclofenac (DCF) in medicinal wine samples was developed. The half-inhibitory concentration (IC₅₀) values of the ILFST for DCF and the limit of detection (LOD) in DCF-spiked samples were 0.66 and 0.15 ng/mL, respectively. Recoveries of DCF from spiked medicinal wine samples were from 84.4% to 85.9% within an assay (intra-assay) and from 83.7% to 88.4% between assays (inter-assay). The coefficients of variation (CV) for intra-assay and inter-assay were 4.10–7.56% and 4.71–8.94%, respectively. These results indicated that the ILFST was highly sensitive and re-producible. The same authentic medicinal wine samples were tested by both ILFST and high-performance liquid chromatography (HPLC) respectively. The results showed no significant differences between the two methods. Therefore, the ILFST can be widely used as a convenient tool for both qualitative and quantitative detection of DCF residues in medicinal wine samples.

KEYWORDS:

• Diclofenac
• colloidal gold
• lateral flow immunoassay
• rapid detection

Introduction

Diclofenac [DCF, 1-(2,6-Dichlorophenyl)-2-indolinone], a nonsteroidal anti-inflammatory drug used to alleviate pain, fever, and inflammation associated with arthritis, rheumatoid arthritis, osteoarthritis, acute gout, and dysmenorrhea (Savaşer, Ozkan, & Işimer, 2005), was commonly used antipyretic analgesic for the rapid onset and low cost of treating rheumatic diseases. In China, some unscrupulous manufacturers have added diclofenac sodium to medicinal wine to enhance efficacy and gain illegitimate benefits. DCF is mainly absorbed in the gastrointestinal tract and long-term use of medicinal wine containing diclofenac causes adverse health effects in humans such as gastrointestinal ulceration, renal toxicity, and platelet dysfunction (Aygün, Kaplan, Odaci, Onger, & Altunkaynak, 2012). On the other hand, because of its polar nature, DCF cannot be effectively removed by wastewater treatment and subsequently continue to endanger the health of fish and birds (Zhang, Geissen, & Gal, 2008; Ziylan, Koltypin, Gedanken, & Ince, 2013).

There are many analytical methods for the determination of DCF in biological samples, such as high-performance liquid chromatography (HPLC) (Madikizela & Chimuka, 2017), HPLC with ultra-violet detection (Payán et al., 2009) or electrochemical detection (Chmielewska, Konieczna, Plenis, Bieniecki, & Lamparczyk, 2006), liquid chromatography-mass spectrometry (LC-MS) (Botitsi, Frosyni, & Tsipi, 2007), and gas chromatographic-mass spectrometry (GC-MS) (Shah, Barker, Naughton, Barton, & Ashraf, 2016). Although these methods are rather sensitive and specific for detecting DCF, they are time-consuming, laborious, and limited to laboratory use due to their dependency on sample pretreatment (separation, extraction, and derivatization), expensive instruments, and skilled professionals. In contrast, immunoassays, based on the specific reaction between antigen and antibody, can overcome some of these drawbacks. Thus, the enzyme-linked immunosorbent assays (ELISAs) (Huebner, Weber, Niessner, Boujday, & Knopp, 2015) and Upconversion-Linked immunosorbent assays (ULISAs) (Hlaváček et al., 2016) were developed for the detection of DCF in wastewater, livestock carcasses, and other biological samples (Saini et al., 2012). ELISAs eliminate sophisticated sample pretreatment and can tolerate certain matrix interferences because of washes between steps. Compared with ELISAs, immunochromatographic lateral flow strip test (ILFST) is simpler and rapid immunoassay that has been used in detecting pesticides (Byzova, Urusov, Zherdev, & Dzantiev, 2018; Ismail et al., 2016), veterinary drugs (Chen et al., 2017; Shelver & Smith, 2018), mycotoxins (Di Nardo et al., 2019; Majdinasab, Zareian, Zhang, & Li, 2019), bacteria (Canver et al., 2019; Saleh, Göttig, & Hamprecht, 2018), and viruses (Lyoo et al., 2017; Senthilkumaran et al., 2017). ILFST integrates all the required reagents into itself and can provide the results within 10 min that can be determined both qualitatively and quantitatively with unaided eyes and a strip reader, respectively. To our knowledge, no ILFST has been reported for the DCF residue so far.

In this study, we aimed to develop an ILFST for the rapid detection of DCF in medicinal wines. DCF was conjugated with bovine serum albumin (BSA) and ovalbumin (OVA) through the activated ester (AE) method and the mixed anhydride (MA) method. Anti-DCF monoclonal antibody (mAb) was produced and then conjugated with colloidal gold nanoparticles for the probe. The ILFST was developed based on the competitive format and validated in terms of specificity, sensitivity, accuracy, and precision in spike recovery experiments, and compared with HPLC for detection in authentic medicinal wine samples.

Materials and methods

Chemicals and materials

DCF, N-Hydroxy succinimide (NHS), N,N-dimethylformamide (DMF), Na₂B₄O₇, tetramethylene oxide, 1-(3-(dimethylamino) propyl)-3-ethylcarbodiimide hydrochloricde (EDC), tributylamine, isobutyl chlorocarbonate, and a mouse monoclonal antibody isotyping kit were purchased from Sigma (St Louis, MO, USA). Ketoprofen, aspirin, indomethacin, ibuprofen, sulindac, naproxen, rofecoxib were purchased from Macklin Biochemical Co., Ltd (Shanghai, China), Bovine serum albumin (BSA) and ovalbumin (OVA) were purchased from Yuanye Biotechnology Co., Ltd (Shanghai, China). HRP conjugated goat-anti-mouse immunoglobulin G antibody was obtained from Sino-American Biotechnology Co., Ltd (Luoyang, China).

Preparation of immunogen and coating antigen

DCF-BSA and DCF-OVA conjugates were prepared using two methods: the AE method and the MA method. In the AE method, according to a previous report (Pinacho, Sánchez-Baeza, & Marco, 2012), 15 mg DCF was dissolved in 3 mL DMF and mixed with 6 mg NHS and 10 mg EDC at room temperature (RT) for 12 h. The mixture was centrifuged at 4000 rpm for 5 min to collect the supernatant. The active final products were harvested through rotary evaporation and dissolved in 1 mL DMSO before the addition of 2 mL BSA solution (42 mg dissolved in 0.01 M PBS) or OVA solution (28 mg dissolved in 0.01 M PBS) in a dropwise manner. In the MA method (Wainer, Fitch, Rothberg, & Fried, 1972), 15 mg DCF was dissolved in 2 mL tetramethylene oxide and stirred in an ice bath for 20 min after the addition of 27 μL tributylamine. Then the mixture was supplemented with 16 μL isobutyl chlorocarbonate and further stirred at RT for 3 h. Finally, 3 mL pre-cooled BSA solution (42 mg dissolved in 0.1 M Na₂B₄O₇) or OVA solution (28 mg dissolved in 0.1 M Na₂B₄O₇) was added dropwise to the mixture and stirred at RT for 24 h. In both methods, the final mixture was dialysed against PBS and centrifuged at 3000 rpm for 20 min at 4°C. The supernatant was collected and stored at–20°C. Ultra-violet scanning was used to judge the conjugation ratio between DCF with BSA or OVA, respectively. DCF-AE-BSA and DCF-MA-BSA, prepared by the two methods, were used as immunogen to immunize BALB/c mice respectively. Indirect and indirect competitive ELISAs were performed by microplate readers (Multiskan FC, Thermo Fisher Scientific Instrument Co., Ltd., Shanghai, China) using the corresponding DCF-OVA conjugates as coating antigens.

Production of anti-DCF mAb

BALB/c mice were subcutaneously immunized with 65 µg immunogen five times at intervals of 3 weeks. Serum antibodies were collected and detected by indirect and competitive indirect ELISAs as previously described (Sun et al., 2014). The mouse giving the best half-maximal inhibitory concentration (IC₅₀) in competitive indirect ELISA was selected to be vaccinated intraperitoneally with 100 µg immunogen before cell fusion. Hybridomas secreting anti-DCF mAbs were selected by competitive indirect ELISA and cultured to prepare ascites in paraffin-primed mice. Affinity and isotype of the mAbs were characterized using ELISA and a mouse mAb isotyping kit (Sigma, USA).

Conjugation of anti-DCF mAb with colloidal gold and preparation of ILFST

Anti-DCF mAb 4B9 was labelled with colloidal gold produced via reduction of HAuCl₄ with 1% sodium citrate as reducing agent (Wang et al., 2019b). Antibody concentration that can prevent the aggregation of gold nanoparticles in the presence of 10% NaCl was selected for use and pH value was adjusted with 0.2 M K₂CO₃. In the labelling, 1 ml mAb 4B9 solution (2 μg/mL) and 5 mL colloidal gold solution (pH 9.0) were incubated for 20 min at RT. The mixture was incubated at RT for another 10 min with 1 mL of 10% BSA. The labelled mAb was then washed twice, and then centrifugated at 15,000 × g for 30 min at 4°C. The precipitant was finally resuspended in aqueous solutions of trehalose plus BSA and stored at 4°C before use.

Test procedure and principle

The ILFST was assembled as previously described (Khaemba et al., 2016; Wang et al., 2019a). The sample solution loads onto the sample pad, and then migrates toward the absorbent pad by capillary effect. If DCF is present in the sample, it may inhibit the binding of colloidal gold-labelled 4B9 with DCF-BSA on the test line (T line), thus, the colour intensity of the T line is reduced. Within the linear range, the concentration of DCF in the sample is negative correlated with the intensity of the red of the T line. One way or another, a visible red of the control line (C line) should always appear to indicate the validation of the ILFST. If no C line appear, the test is considered as invalid and should be performed again with another test strip (Figure 1).

Figure 1. Structure and schematic diagram of the ILFST. (a) Positive; (b) Negative

Sample pretreatment for the LFIST

Anti-rheumatic medicinal wine was purchased from a local market, and was tested to be negative by HPLC. The stock solution was prepared by diluting 2 mL medicinal wine in 8 mL PBS and storing at 4°C. Before the strip test, the stock solution was filtered by 0.22 um of microporous membrane.

Evaluation of the performance of LFIST

The sensitivity of the LFIST was judged by detecting DCF-free medicinal wine samples spiked with 0, 0.25, 0.5, 1.0, 2.0, 4.0 and 8.0 ng/mL of DCF. Each test was repeated in triplicate. Qualitative detection was realized by judging the presence of the T line by two people. For quantitative detection, relative optical density (ROD) of the T line was read by a TSR3000 membrane strip reader (Bio-Rad, USA). A standard curve was constructed by plotting the B/B₀ value toward the concentrations of the spiked samples. B and B₀ represented the ROD values obtained from testing spiked samples and the blank sample, respectively. The value of IC₅₀ was extrapolated by the linear regression equation.

Specificity of the LFIST was evaluated by testing structural analogues of DCF including ketoprofen, aspirin, indomethacin, ibuprofen, sulindac, naproxen, and rofecoxib. DCF and the structural analogues in the negative wine samples were tested at a concentration of 1000 ng/mL, respectively. IC₅₀ and cross-reactivity (CR) for each analyte were calculated and compared. CR was calculated using the equation: CR (%) = (IC₅₀ of DCF)/(IC₅₀ of the competitor) × 100.

The accuracy and precision of the ILFST were evaluated by detecting in medicinal wine samples containing 1.8, 3.6, and 7.2 ng/ml of DCF using the test strip, and were expressed as recovery and coefficient of variation (CV, %), respectively. For inter-assay precision, six batches of the ILFST were used to test samples of different concentrations in sextuplicate.

Authenticity of ILFST

Health wine samples, with five different DCF concentrations (2.4, 4.8, 9.6, 19.2, and 38.4 ng/ mL), were authenticated to be negative by LC-MS. These samples were detected in parallel by ILFST and HPLC, separately. HPLC was performed for instrumental analysis of DCF using Diamonsil C18 (250 × 4.6 mm × 5 μm) obtained from Dima Technology Co., Ltd. (Beijing, China). The mobile phases were an acetonitrile: 0.2% formic acid (60:40, v/v), flow rate of 0.8 mL/min, column temperature was 25°C, and DCF was monitored at 200 nm. Tested values obtained from the ILFST and HPLC were compared to the given concentrations using a one-sample T-test, and the values between the ILFST and HPLC, presented as means SE, were analysed by independent sample T-test. Differences between the test values were considered statistically significant at P < 0.05.

Results and discussion

Preparation of complete antigen

The immunogen DCF-AE-BSA and coating antigen DCF-AE-OVA were both synthesized using the AE method (Figure 2(a)), and DCF-MA-BSA and coating antigen DCF-MA-OVA synthesized by the MA method (Figure 2(b)). After UV scanning, the Coupling rates of DCF with BSA and OVA were calculated as 21.6:1 and 16.5:1, respectively, in the AE (EDC/NHS) preparation. The Coupling rates of DCF with BSA and OVA were calculated as 17.2:1 and 15.9:1, respectively, in the MA preparation. Upon comparison of serum antibody titre and IC₅₀ value (Table 1), the mouse that showed an antibody titre of 1:5.12 × 10⁴ and an IC₅₀ value of 7.89 ng/ mL was chosen for subsequent cell fusion, which was immunized with DCF-AE-BSA prepared using the AE method.

Figure 2. (a) Synthesis of DCF-BSA using activated ester method. (b) Synthesis of DCF-BSA using mixed anhydride method.

Table 1. Characterization of the optimal antiserum produced by MA and AE methods toward free DCF and coating antigens.

Coating antigen	Antiserum MA		Antiserum AE
	Titre (×10^3)	IC50 (ng/mL)	Titre (×10^3)	IC50 (ng/mL)
DCF-AE-OVA	6.4	51.13	51.2	7.89
DCF-MA-OVA	12.8	25.41	6.4	43.76

Characterization of mAb against DCF

After cell fusion and screening by direct and indirect competitive ELISAs, four hybridomas cell lines named 2F4, 3A7, 3D4, and 4B9 were selected. Affinity constant, antibody titre, and IC₅₀ value of each mAb were determined by ELISA (Table 2). mAb 4B9, which showed an antibody titre of 1:1,024,000 and an IC₅₀ value of 0.61 ng/mL, was selected to be labelled with colloidal gold nanoparticles. The affinity constant (Ka) and the subtype of 4B9 was determined to be 1.5 × 10¹⁰ L/mol and immunoglobulin G 1, respectively.

Table 2. Characterization of anti-DCF mAbs by ELISA.

mAb	Antibody titre	Ka (L/mol)	Linear regression equation	R^2	IC50 (ng/mL)
2F4	1:1.024 × 10^6	9.0 × 10^10	y = −0.3823x + 0.5091	0.9935	1.1
3A7	1:5.12 × 10^5	7.5 × 10^10	y = −0.4842x + 0.6562	0.9921	2.1
3D4	1:5.12 × 10^5	7.5 × 10^10	y = −0.4617x + 0.6177	0.9927	1.8
4B9	1:1.024 × 10^6	1.5 × 10^11	y = −0.3263x + 0.3986	0.9958	0.61

Sensitivity of the ILFST

Reference solutions of DCF at concentrations of 0, 0.25, 0.5, 1.0, 2.0, 4.0 and 8.0 ng/mL were analysed using the ILFST. The ROD value for each test was obtained by scanning the T line with a TSR3000 membrane strip reader (Figure 3) and are quantified in Table 3. A quantitative calibration curve was constructed by drawing the value of B/B₀ against the concentrations of DCF in medicinal wine samples (Figure 4). ROD values decreased in a linear manner as the final DCF concentration in spiked samples increased from 0.0 to 8.0 ng/mL. According to the linear equation, IC₅₀ was calculated as 0.66 ng/mL, with the coefficient of determination (R²) = 0.9933. The limit of detection (LOD) of the ILFST was determined to be 0.15 and 1.0 ng/mL with unaided visual assessment (Figure 5). High sensitivity of the ILFST reflected the high binding affinity between mAb 4B9 and DCF.

Figure 3. ROD curves from detections of different spiked health wine samples. Blank health wine samples were spiked with DCF to final concentrations of 0, 0.25, 0.50, 1.0, 2.0, 4.0 and 8.0 ng/ml and tested using the ILFST. Relative optical density (ROD) of the test line was read by a TSR3000 membrane strip reader (Bio-Dot).

Figure 4. The quantitative calibration curve drawn from testing different DCF standard solutions using ILFST. The X-axis represents the concentrations of DCF in spiked samples. B/B₀ represents the percentage of ROD values obtained from spiked samples divided by that of the blank sample.

Figure 5. Standard DCF samples were tested using the ILFST and the T lines with the naked eyes.

Table 3. Relative optical density from testing spiked health wine samples using ILFST

DCF concentration (ng/mL)	G/D × area-ROD (pixel)	G/Peak-ROD (pixel)
0	175.547	0.1044
0.25	127.632	0.0853
0.5	92.4924	0.0645
1.0	71.9189	0.0423
2.0	46.8408	0.0181
4.0	26.1388	0.0082
8.0	1.8859	0.0036

Specificity of the ILFST

To determine the specificity of the ILFST, CR was evaluated by testing structural analogues of DCF including ketoprofen, aspirin, indomethacin, ibuprofen, sulindac, naproxen, and rofecoxib. When these structural analogues were added at a final concentration of 1000 ng/ml, the colour of T lines remained negative. Determinations of the IC₅₀ values for each competitor and calculations of CR showed that the mAb 4B9 based ILFST had less than 0.1% CR with these competitors (Table 4).

Table 4. CR of ILFST with competitors.

Compounds	Structure	IC50 (ng/mL)	CR (%)
diclofenac		0.66	100
Ketoprofen		> 700	< 0.1
Aspirin		> 700	< 0.1
Indomethacin		> 700	< 0.1
Ibuprofen		> 700	< 0.1
Sulindac		> 700	< 0.1
Naproxen		> 700	< 0.1
2Rofecoxib		> 700	< 0.1

A DCF-thyroglobulin conjugate was synthesized by EDC/NHS method, and four mice were immunized to obtain antisera. An immunized mouse giving the highest antisera titre was used for hybridization, and its spleen cells were fused with myeloma cells. A generic mAb 12G5 against DCF was prepared. The LOD, IC₅₀ and the range of detection of competitive indirect ELISA were 7.8, 44 and 11–180 ng/L, respectively (Huebner et al., 2015). The CR of mAb 12G5 was 8.5–13% for DCF metabolites, but less than 1% for the structural analogues. The competitive indirect ELISA was applied for detection of DCF residues in river and lake water. Although this ELISA is highly sensitive and specific, it requires too many operation steps comparing to ILFST.

Accuracy and precision of the ILFST

The accuracy and precision of the ILFST were determined by detecting in medicinal wine samples containing 1.8, 3.6, and 7.2 ng/ mL of DCF using the test strip and expressed as recovery and CV, respectively. As shown in Table 5, recoveries ranged from 84.4 to 86.8% within an assay (intra-assay) and from 83.7 to 88.4% between assays (inter-assay), respectively. Both assay CV of the ILFST were less than 9.0%.

Table 5. Accuracy and precision of the ILFST.

Spiked DCF (ng/mL)	Intra-assay			Inter-assay
	Mean ± SD (ng/mL)	Recovery (%)	CV (%)	Mean ± SD (ng/mL)	Recovery (%)	CV (%)
1.8	1.52 ± 0.12	84.4 ± 6.7	7.56	1.51 ± 0.13	83.7 ± 7.2	8.94
3.6	3.12 ± 0.15	86.8 ± 4.2	4.84	3.05 ± 0.21	84.8 ± 5.8	6.79
7.2	6.18 ± 0.25	85.9 ± 3.5	4.10	6.36 ± 0.30	88.4 ± 4.2	4.71

Comparison of ILFST with HPLC

A comparison between ILFST and HPLC was performed by testing medicinal wine samples. The results showed that there was no significant difference between the ILFST with HPLC (Table 6), which means that the values tested by ILFST were reliable.

Table 6. Comparison between the ILFST and HPLC.

DCF (ng/mL) in health wine samples	ILFST (ng/mL)	HPLC (ng/mL)
2.4	2.04 ± 0.12^a	2.15 ± 0.09^a
4.8	4.18 ± 0.22^a	4.43 ± 0.15^a
9.6	8.45 ± 0.49^a	9.02 ± 0.31^a
19.2	16.74 ± 0.6 ^a	17.76 ± 0.56^a
38.4	32.85 ± 1.54^a	34.80 ± 1.26^a

^aThe superscript represents no statistical significance between the results given by ILFST and HPLC (P > 0.05).

Conclusions

An ILFST was established using a high-affinity, specificity mAbs 4B9 against DCF for detecting DCF residues. For DCF, the IC₅₀ was 0.66 ng/mL, and the LOD was 0.15 ng/mL in the reference solutions. The recoveries from medicinal wine samples were found to be from 84.4 to 85.9% within an assay (intra-assay) and from 83.7% to 88.4% between assays (inter-assay). The major advantages of the ILFST are relative cost-effectiveness, it is faster than chromatographic instrument analysis, and is easy to operate by professionals. Therefore, the ILFST is a convenient screening tool for detecting DCF residues in medicinal wine samples.

Ethical statement

The use of all BALB/c mice in this study was approved by the Animal Ethics Committee of Zhoukou Normal University and in accordance with all applicable institutional and governmental regulations concerning the ethical use of animals.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research was supported by grants from the National Science & Technology Pillar Program of the 12th Five-Year Plan & Technology (grant number 2014BAD13B05), the Natural Science Foundation of Henan Province (grant number 182300410093).