Chemicals and materials

PCM, PCM sulfate, PCM glucuronide, orthocetamol, metacetamol, diclofenac, meloxicam, and tetracycline were purchased from Sigma-Aldrich (St. Louis, MO, USA). The stock solutions and dilutions were prepared in methanol from Actu-All Chemicals (Oss, the Netherlands). Mouse anti-PCM monoclonal antibody (mAb) (2 mg/mL) and PCM-bovine serum albumin (PCM-BSA) conjugate (4 mg/mL) were purchased from Ecalbio (Wuhan, China) and the goat anti-mouse IgG FcY (GAM)-specific polyclonal antibodies (1.3 mg/mL, 115–005–071) and donkey anti-goat IgG (DAG) polyclonal (H + L) antibodies (1.3 mg/mL, 705–005-003) from Jackson ImmunoResearch (Sanbio, Uden, The Netherlands). Spezial Schwartz 4 CNPs were purchased from Orion Engineered Carbons GmbH (Eschborn, Germany). Zeba™ spin desalting columns (7 K MWCO) were obtained from Thermo Scientific (Rockford, IL, USA). BSA, Tween-20, and trehalose were acquired from Sigma-Aldrich (Zwijndrecht, The Netherlands); boric acid and sodium tetraborate from VWR (Leuven, Belgium); and phosphate-buffered saline (PBS) tablets from Millipore Corporation (Cork, Ireland). The buffers were prepared in deionized water from a Milli-Q® system with a minimum resistance of at least 18.2 MΩ cm−1 (Millipore, Billerica, MA, USA). The 20-mm glass fiber-based pads (Grade 8950), proprietary fiber blend-based sample pad (Grade 1660), and 15-mm cotton fiber-based wicking pads (Grade 222) were obtained from Ahlstrom (Helsinki, Finland); 25-mm Unisart® CN 95 nitrocellulose membranes from Sartorius Stedim Biotech GmbH (Gottingen, Germany); and the silica absorbent pads from Sigma-Aldrich (Zwijnrecht, The Netherlands). Backing cards and aluminum foil pouch bags for storage of the produced icLFIAs were purchased from Kenosha (Amstelveen, the Netherlands) and the 96-well microtiter plates from Greiner bio-one (Alphen a/d Rijn, The Netherlands). Cassettes for the construction of the LFD were produced using in-house 3D printing facilities.

Preparation of CNPs functionalized with GAM antibodies

CNPs were functionalized with GAM antibodies as described previously by Sharma et al. [20]. Briefly, a 1% (w/v) aqueous CNP suspension was sonicated for 1 h at 40 kHz at room temperature. Next, a 0.2% (w/v) CNP suspension was prepared in 5 mM borate buffer (pH 8.8) containing boric acid and sodium tetraborate, and sonicated for 5 min. Before functionalization, the purified GAM antibodies were desalted and buffer exchanged with 5 mM borate buffer (pH 8.8) using 0.5-mL Zeba™ spin columns. Subsequently, the antibody concentration was measured with a DS-11 FX spectrophotometer for microvolumes (DeNovix, Wilmington, NC, USA). Afterwards, 0.35 mg GAM antibodies (e.g., 0.35 mL of a 1 mg/mL desalted and buffer exchanged GAM solution) were added dropwise to 1 mL of 0.2% carbon suspension. The content was gently mixed overnight with a magnetic stirrer at 200 rpm and 4 °C. After incubation, washing buffer (5 mM borate buffer, pH 8.8 containing 1% w/v BSA) was added dropwise to the suspension. The suspension was then stirred for 5 min and centrifuged at 13,600 × g for 15 min at 4 °C followed by collecting the pellet and discarding the supernatant. The washing procedure was repeated three times before the pellet was resuspended in storage buffer (100 mM borate buffer, pH 8.8 containing 1% w/v BSA) to a final concentration of 0.2% (w/v) CNPs. The GAM-CNP conjugate was stored at 4 °C until further use. The performance of the coupling procedure has been demonstrated by quantitative computer image analysis and was reported previously [23].

Method setup by implementation of spot-based icLFIAs

The initial setup of the icLFIA for PCM was performed by manually spotting the bioreagents onto the nitrocellulose membrane. For this purpose, the nitrocellulose membrane was secured on a backing card as support and slightly overlaid with a wicking pad followed by cutting the card with all parts into 4-mm strips using the BioDot CM5000 Guillotine Cutter (Irvine, CA, USA). Subsequently, 0.5 μL of PCM-BSA (test dot) and 0.5 μL of DAG antibody (control dot) were pipetted on top of the cut nitrocellulose membrane and let dry for 2 h at room temperature. PCM-BSA conjugate was diluted to 0.125, 0.25, and 0.5 mg/L in PBS buffer, whereas the DAG antibody was diluted to 0.15 mg/mL in PBS buffer. Next, a dilution series of the anti-PCM mAb (1:250–1:2000) was prepared in running buffer composed of 0.01 M PBS (pH 7.4), 0.05% v/v Tween-20, and 1% w/v BSA. Additionally, PCM standard solutions in the concentration of 0, 0.01, and 1 mg/L were prepared. 1 μL anti-PCM mAb and 1 μL GAM-CNP conjugate were added to 98 μL diluted PCM standard solutions in a microtiter plate and gently mixed. The spot-based icLFIAs were put into the microtiter plate and developed for 10 min.

Production of line-based icLFIAs

For the line-based icLFIAs, the test line with the PCM-BSA conjugate (0.5 mg/mL) and the control line with the DAG antibody (0.15 mg/mL) were sprayed in PBS buffer onto a nitrocellulose membrane at a rate of 1 µL/cm using a BioDot XYZ3060 dispenser (Chichester, UK). The nitrocellulose membrane sheet was let dry for 2 h at room temperature.

Next, a mixture of 1000 times diluted anti-PCM mAb in running buffer (0.01 M PBS, pH 7.4 containing 0.05% v/v Tween-20 and 1% w/v BSA) was prepared. Subsequently, the anti-PCM mAb (1:1000) and the GAM-CNP conjugate were diluted 10 times in spraying buffer (0.01 M PBS, pH 7.4 containing 10% w/v trehalose and 1% w/v BSA). Glass fiber-based antibody and conjugate pads were prepared by pipetting the diluted mixtures of anti-PCM mAb and GAM-CNP on glass fiber pads, respectively. Finally, the pads were let dry overnight.

For the assembly of the complete icLFIA, the dried nitrocellulose membrane with the test and control lines was secured on a backing card as support. Next, the wicking pad was placed by the end of the icLFIA, and the glass fiber-based antibody pad was placed at the beginning of the icLFIA, both slightly overlapping with the nitrocellulose membrane for approximately 1 mm. Then, the conjugate pad was secured directly on the backing support at a 2-mm distance from the antibody pad. Finally, a proprietary fiber blend-based sample pad was placed on top of the antibody and conjugate pads (Fig. 1). The backing card with all parts was cut into 4-mm strips using the BioDot CM5000 Guillotine Cutter (Irvine, CA, USA). The cut icLFIAs were stored in special bags with a silica pad and closed by a sealing apparatus until further use.

Fig. 1

Construction of icLFIAs for PCM

Point-of-need LFD test for PCM in bovine urine

Before analysis, LFD was generated by placing the complete icLFIAs into a dedicated 3D printed cassette. Samples were prepared by pipetting 240 μL of running buffer and 60 μL bovine urine (ratio 80/20) in a 1.0-mL Eppendorf tube and the tube was vortexed for 5 s. Next, 100 μL of the homogeneous solution was pipetted onto the LFD and developed for 10 min.

Validation of the point-of-need LFD test Detection capability for screening (CCβ)

Endogenous PCM levels may be present in bovine urine samples. However, the maximal endogenous PCM level is not exactly known and neither is the minimal PCM level in the urine of cattle treated with PCM. Based on previous animal tests and other positive results found in bovine urines within WFSR, the screening target concentration of 5 mg/L was selected as the threshold from which PCM levels are expected in treated animals.

To determine the CCβ, 22 different batches of in-house provided samples from bovine urine, previously confirmed to be blank with an in-house developed and validated confirmatory LC–MS/MS method, were spiked with PCM at 5 mg/L and were tested divided over three consecutive days. The technical details of the LC–MS/MS method are presented in the supplementary information file. The criterion for the detection capability is that at least 21 (95%) should be characterized as non-compliant, i.e., as a suspicious sample, by the LFD for PCM. When this is the case, the screening target concentration of 5 mg/L is the CCβ of the developed LFD test (although the exact CCβ might be lower).

Selectivity/specificity

To assess the selectivity, the aforementioned 22 in-house provided samples from bovine urine were tested as such divided over three consecutive days to assess the false positive rate in negative samples. Preferably, the matrix of the 22 samples from bovine urine does not influence the result of the test and only leads to negative screening results that would not require additional confirmational LC–MS/MS analysis. Secondly, the specificity, i.e., the ability of the LFD to discriminate between PCM and other structurally related PCM analytes, NSAIDs, and antibiotics, was investigated. Bovine urine samples were spiked with PCM sulfate, PCM glucuronide, orthocetamol, metacetamol, diclofenac, meloxicam, and tetracycline at 5 mg/L to conclude whether the aforementioned analytes interfere with the LFD and lead to false positive results (sample classified as suspect and marked for additional confirmational LC–MS/MS analysis). Each potential interference of structurally related PCM analytes, NSAIDs, and antibiotics was performed in three different batches of bovine urine.

Robustness

To determine the continued performance of the LFD under different experimental conditions, minor changes were applied to the method, which are most likely to occur during routine analyses and are expected to be critical. First, the reading time was varied between 5 and 15 min, instead of 10 min. Second, the dilution ratio for running buffer and bovine urine was varied between dilution ratios of 90/10 and 70/30 instead of 80/20. All the separate conditions were performed for blank and PCM-spiked (5 mg/L) samples in three different batches of bovine urine.

Stability of the icLFIAs

Since the test for PCM is intended for screening purposes and on-site analysis in fresh urine samples from farms, the stability of PCM in solution, extract, and matrix is not relevant for this application. However, the stability of the icLFIA is important as it consists of antibodies and conjugates, i.e., proteins whose functionality might be affected by light, time, and temperature, and thus storage shelf life. After production, the icLFIAs were stored in closed aluminum bags with a silica desiccant at room temperature. The bovine urine was stored at − 20 °C. The icLFIAs were tested at 1, 3, 7, 14, 28, and 56 days after production to prove that the result is not altered over time. Each time point was performed in triplicate for a blank and a PCM-spiked (5 mg/L) bovine urine sample.

Data assessment

The results of the icLFIAs were read by the naked eye and by a handheld optical detector, i.e., a Cube Reader (Chembio, Berlin, Germany). Images of the developed icLFIAs were recorded with a Samsung A50 smartphone in a Caruba Portable Photocube LED 70 × 70 × 70 cm (Caruba, Beilen, The Netherlands) to maintain similar lighting conditions. For the visual assessment, the appearance of both the control and the test line indicates a negative result, whereas only the appearance of the control line indicates a suspect result. The absence of the control line means that the test was not valid (Fig. 2).

Fig. 2

Assessment of LFDs (A) negative/complaint, (B) positive/suspect, (C) invalid. On the LFD cassettes, the c indicates the position the control line is expected, the t indicates the position the test line is expected and the s indicates the sampling site

The Cube Reader measures the intensity of the control and test line in arbitrary units (AU). Subsequently, the threshold (T) and the cutoff factor (Fm) could be determined based on the test line versus control line ratio as presented in Eqs. (1) and (2).

$$T=B-1.64 \times SDb$$

(1)

where B is the mean test/control line response of the blank samples and SDb is the standard deviation of the blank samples.

$$Fm=M-1.64 \times SDs$$

(2)

where M is the mean test/control line response of the spiked samples and SDs is the standard deviation of the spiked samples.