Introduction

In recent decades, the high consumption of animal food in the human diet has boosted the excessive use of antibiotics to prevent and control diseases, promote growth, and improve the feed conversion efficiency in livestock farming1. The main classes of antibiotics used for these purposes include sulphonamides (mainly sulfamethazine, sulphadimidine, sulfamethoxypyridazine, sulfamethazine, and sulfathiazole), tetracyclines (including oxytetracycline, epi-oxytetracycline, chlortetracycline, doxycycline, among others), penicillins (such as amoxycillin, penicillin G, ampicillin, among others), macrolides (including gamithromycin, neospiramycin, erythromycin among others), quinolones and fluoroquinolones (mainly enrofloxacin, oxolinic acid, and ciprofloxacin), and aminoglycosides (such as neomycin or streptomycin)2. They are used in aquaculture, cattle, eggs, honey, milk, swine, poultry, rabbits, and sheep/goats3,4, and can be administered orally (through water or feed), as injectables, by transfusion, or topically.

The overuse of antibiotics (12 million kilograms worldwide each year1) can leave residues both in their original form and as metabolites or degradation products in foods5. It has been shown that antibiotic residues in food can lead to the development of bacterial resistance, making common infections difficult to treat6,7,8,9. In addition, the continuous ingestion of these residues can alter intestinal flora, causing digestive problems and favouring the growth of pathogenic bacteria. In some cases, prolonged exposure to low levels of antibiotics can trigger allergic reactions or toxicity, affecting organs such as the liver and kidneys10. The presence of antibiotic residues can also interfere with microbiota, weaken the immune system, and increase susceptibility to other diseases6. Specifically, diethylstilbestrol, nitrofurans, and chloramphenicol were banned because of their carcinogenic properties11,12.

To address these problems, it is essential to implement a combination of strategies involving the strict control of antibiotic use in agriculture and animal husbandry, sustainable agricultural practices to reduce reliance on antibiotics, and improved antibiotic residue detection techniques. This will ensure compliance with existing government regulations that can significantly reduce antibiotic contamination in the food chain and protect consumer health. The European Union (EU) has mandated that food undergo scientific assessment in line with Regulation (EC) No. 470/200913. Additionally, in collaboration with the FDA14 and the Codex Alimentarius organisation (which established the Codex Guidelines for the Establishment of a Regulatory Programme for Control of Veterinary Drug Residues in Food)15, the EU has implemented maximum residue concentration limits for pharmacologically active substances utilised in animals. To assess the presence of antibiotic residues in food, it is essential to use an extraction procedure to overcome the complexity of matrices and mitigate interference with target analyte quantification. Solid–liquid extraction (SLE), Solid-Phase Extraction (SPE), and liquid–liquid extraction (LLE) techniques have historically been the most widely used techniques, as reviewed by Barros et al.2. Combined with advanced analytical methods such as liquid chromatography (LC), extraction techniques such as SLE, d-SPE, SPE, and LLE have improved the detection and quantification of antibiotics, thus ensuring greater food safety and compliance with government regulations2,16,17,18,19,20. However, the current challenge is the development of miniaturised extraction procedures using minimal amounts of solvents, reagents, and samples to fulfil the goals of Sustainable Green Chemistry21. To this end, a μSPEed/UPLC-UV methodology was developed and validated for the simultaneous analysis of six veterinary antibiotics in environmental water: tetracycline (TC), chlortetracycline (CTC), oxytetracycline (OTC), doxycycline (DC), sulfamethoxazole (SMX), and trimethoprim (TMP). Overall, the developed methodology exhibits exceptional efficiency, with a total processing time of only 12 min and a sample volume of 500 μL, outperforming other methods in terms of speed and sample use while maintaining robust analytical performance and environmental friendliness. Moreover, this semi-automated approach reduces the workload of analysts and improves their reproducibility. The method’s adaptability and use of affordable, readily available equipment (μSPEed sorbent and DigiVol) make it easily accessible to other laboratories and suitable for analysing selected antibiotics in various matrices, including food and environmental samples.

Materials and methods

Reagents and materials

Analytical standards (≥99%) for tetracycline (TC), chlortetracycline (CTC), oxytetracycline (OTC), doxycycline (DC), sulfamethoxazole (SMX), and trimethoprim (TMP) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Louis, MO, USA. HPLC-grade acetonitrile (ACN), methanol (MeOH), and sodium hydroxide (NaOH) were obtained from Fischer Scientific (Loughborough, UK). Formic acid (FA), chloride acid (HCl, 37%) were purchased from Panreac Química (Barcelona, Spain). Purified water used in the experiments was produced from a Milli-Q ultrapure water purification system (Millipore, Milford, MA, USA). All reagents were of the highest available analytical quality.

The Digivol® syringe and μSPEed® cartridges (butyl silica (C4), octyl silica (C8), octadecyl silica (C18-RPS), octadecyl silica hydrophilic ODS (C18-ODS), unmodified silica (Sil), aminopropyl silane sorbent (WAX), polyfluoroalkyl substances (PFAS), porous PS/DVB normal phase (PS/DVB), porous PS/DVB reversed phase (PS/DVB-RP), PS/DVB anionic exchange (PS/DVB-SAX), PS/DVB cationic exchange (PS/DVB-SCX), and porous graphitic carbon (μCARB)) were supplied by EPREP (Mulgrave, Victoria, Australia).

Standard solutions were prepared at 1 mg mL−1 using ACN as the solvent (for CTC, MeOH was used). Lower concentrations were obtained by diluting standard solutions with an appropriate amount of ACN.

Instrumentation

Chromatographic analysis was performed using a Waters Ultra-High Pressure Liquid Chromatographic Acquity system (Acquity UPLC H-Class, Milford, MA, USA) equipped with a column heater, an Acquity sample manager (SM), a degassing system, a Water Acquity quaternary solvent manager (QSM), and a photodiode array (PDA) detector. The analytical column was a Acquity UPLC BEH C18 (130 Å, 2.1 mm × 50 mm, 1.7 μm, Waters, USA) operated at 30 °C with gradient elution at a flow rate of 0.40 mL min–1; the mobile phase for the chromatographic separation was a mixture of A (0.1% FA) and B (ACN). Separation was performed with the following gradient conditions: 95% A, 80% A at 3 min, 75% A at 3.5 min, 70% A at 4 min, 65% A at 4.5 min, and 60% A at 6 min. The system was re-equilibrated for 2 min with 95% A between injections. PDA detection was set at 280 nm during the analysis.

μSPEed procedure

Extraction was performed using a DigiVol® X-change® electronic automatic syringe (250 μL needle). The PS/DVB-RP cartridge was conditioned with 250 µL of MeOH (750 μL min−1 flow rate), equilibrated with 250 µL of H2O (750 μL min−1 flow rate), and 500 µL of the sample was loaded three times (500 μL min−1 flow rate). The elution step was performed with two cycles of 50 µL acidic MeOH (500 μL min−1 flow rate). The optimised extraction procedure required 6 min for each sample. Each cartridge was used for over 50 extractions without affecting its activity. The experimental procedure is illustrated in Fig. 1.

Fig. 1: Schematic workflow of the μSPEed optimised extraction procedure.
figure 1

TC tetracycline, CTC chlortetracycline, OTC oxytetracycline, DC doxycycline, SMX sulfamethoxazole, TMP trimethoprim.

Full size image

Results and discussion

Semi-automated solid-phase microextraction has been used to extract and purify four tetracyclines, SMX, and TMP, in aqueous samples. The selected antibiotics have different physicochemical properties and polarities. For this reason, full optimisation of the extraction and chromatographic conditions using a univariate protocol was performed. The starting conditions for the conditioning, equilibration, sample loading, and elution steps of µSPEed extraction were chosen based on previous experiments22,23.

Optimization of cartridge used for extraction

Cartridge selection is crucial for the μSPEed method. Different cartridges provide distinct retention mechanisms that are tailored to various target analytes22. Twelve cartridges (C4, C8, C18 RPS, C18 ODS, Sil, WAX, PFAS, PS/DVB, PS DVB, PS/DVB, PS DVB RP, PS/DVB, PS DVB SAX, PS DVB SCX, and μCARB) were tested for the extraction of the selected analytes. To assay the PS DVB SAX and PS DVB SCX cartridges, the sample pH was adjusted to 4, 7, and 9. This was essential to ensure that the sorbents were assayed under the best conditions. Moreover, the effect of pH on the extraction of antibiotics, including those used in this work, has been widely reported in the literature24,25,26. According to Mirzaei et al.24 and Zhou et al.26, the pH of water samples plays a crucial role in the extraction recovery of antibiotics, affecting both the stability of the antibiotics and their interactions with the extraction medium. To optimise the recovery, the sample pH is often adjusted using hydrochloric acid before extraction24. The recovery values for each cartridge are shown in Fig. 2. Overall, the PS DVB cartridges (normal phase and reverse phase, RP) were found to be the most suitable for extracting the selected antibiotics, outperforming the other sorbents assayed, including the ionic exchange sorbents mentioned above. PS DVB RP was selected for subsequent experiments because it is the most versatile PS DVB sorbent.

Fig. 2: Evaluation of the effect of sorbent on the extraction efficiency of selected antibiotics.
figure 2

TMP trimethoprim, OTC oxytetracycline, TC tetracycline, SMX sulfamethoxazole, CTC chlortetracycline, DC doxycycline. Errors bars represent the standard deviation (SD) of the data obtained.

Full size image

Sample conditions optimization: volume, loading cycles and pH

To improve the recovery of the selected analytes during extraction, further optimisation of the process was performed. In the third step of μSPEed extraction (see Fig. 1), the effects of the amount of sample, number of extraction cycles, and pH of the elution solvent on the recovery of the target analytes were assayed. While the sample volume and number of extraction cycles affect the amount of the selected compounds available in the solution to be retained in the sorbent during the extraction, the pH of the elution solvent was also assayed considering the dependence of the extraction efficiency on the pH mentioned above24,25,26. Accordingly, 3–5 sample loading cycles followed by MeOH or acidified MeOH elution were performed. The highest recoveries for all selected antibiotics (46.9–75.3%) were obtained for a 3 × 250 μL sample cycle loading at a flow rate of 500 μL min−1 using acidified MeOH as the elution solvent (Fig. 3a). The effect of the sample volume was further investigated by analysing the recovery of the target antibiotics at sample volumes of 500 and 800 μL. Optimum recoveries (45.9–85.9%) were obtained for a sample volume of 500 μL at a flow rate of 500 μL min−1 (Fig. 3b). Up to this point of extraction optimisation, the recoveries of some of the selected antibiotics, namely CTC and DC, were very low (not reaching 50%). This led us to thoroughly assess the influence of pH on the extraction of target analytes. We consider that tetracyclines are amphoteric compounds, with pKa values of 3.33, 7.55, 9.33 for CTC; 3.02, 7.97, 9.15 DC, 3.22, 7.46, 8.94 for OTC; and 3.32, 7.78, 9.58 for TC27,28; SMX is an acidic compound with a low pKa of 1.6 and 5.729, while TMP is an alkaline compound with a pKa of 7.1230. Therefore, it was necessary to select suitable pH conditions to ensure that the selected compounds were in the protonated/non-protonated forms. Accordingly, we considered a pH below all pKa values (pH 2.5, and not pH 4.0, as previously assayed), pH 7.0 as a neutral condition, and pH 8.5, and not 9.0 (basic conditions). The best performance (recoveries in the range 46.9–75.3%) was obtained with samples at pH 7 and elution with acidified MeOH (5% FA, Fig. 3c).

Fig. 3: Evaluation of the effects of the sample loading cycles, sample volume, and sample and elution solvent pH on the extraction efficiency of the selected antibiotics.
figure 3

Sample loading cycles (a), sample volume (b), and sample and elution solvent pH (c). TMP trimethoprim, OTC oxytetracycline, TC tetracycline, SMX sulfamethoxazole, CTC chlortetracycline, DC doxycycline. Error bars represent the standard deviation (SD) of the data obtained.

Full size image

Selection of elution solvent

In the last step (Fig. 1), the effect of the solvent on the elution of target analytes was assayed and fine-tuned. Tests were performed using MeOH, ACN, and acidic MeOH (with 5% FA and 1% FA). Other conditions such as acidic ACN were not assayed to preserve the integrity of the spare parts of the DigiVol® syringe. After evaluation, acidic MeOH with 5% FA was found to allow for a higher extraction of the target antibiotics, and this condition was selected for further analysis (Fig. 4). These results agree with other reports referring to MeOH as the most suitable solvent for desorbing compounds from SPE26. The volume of solvent required to elute all the analytes was also examined. Accordingly, 50, 100, and 150 μL of acidic MeOH were assayed. While increasing the elution volume from 50 to 100 µL yielded better results, there was no significant difference in the recoveries between the 100 μL and 150 μL elution volumes. We verified that the extraction of the target analytes was further enhanced with two consecutive 50 µL elution steps in comparison with a single 100 μL elution step (data not shown). All discarded solutions in the different steps of the µSPEed protocol were collected and analysed throughout the optimisation process to evaluate the effect of each condition on the extraction efficiency. This was essential to verify the efficiency of the µSPEed sorbent in retaining the target analytes during the sample loading step and the ability of the elution solvent to elute the analytes from the sorbent, thus avoiding carry-over effects. Specifically, in this case, a third consecutive elution step performed during the optimisation revealed that no antibiotics were retained in the sorbent. Overall, the best conditions obtained included 500 μL of the sample processed in a 3 × 250 µL sample loading cycle, followed by two consecutive elution steps with 50 μL of acidified MeOH (5% FA). The flow rate was maintained at 500 μL min−1 during the entire procedure.

Fig. 4: Evaluation of the effect of solvents on the extraction of the selected antibiotics.
figure 4

TMP trimethoprim, OTC oxytetracycline, TC tetracycline, SMX sulfamethoxazole, CTC chlortetracycline, DC doxycycline. Errors bars represent the standard deviation (SD) of the data obtained.

Full size image

μSPEed/UPLC-PDA validation

To allow appropriate validation of the methodology, the chromatographic separation of the target analytes was optimised using a gradient of acidified water (0.1% FA) and ACN (as described in “Materials and Methods”). This resulted in a fast chromatographic separation of only six min that, to the best of our knowledge, has not been previously reported for the selected antibiotics in this time range (Fig. 5).

Fig. 5: Typical UPLC-PDA chromatograms of the separation of selected antibiotics (standard solution, blue line; sample after extraction, red line; λ = 280 nm).
figure 5

CTC chlortetracycline, DC doxycycline, OTC oxytetracycline, SMX sulfamethoxazole, TC tetracycline, TMP trimethoprim.

Full size image

The analytical performance of the developed methodology was evaluated based on relevant parameters, such as the linear range, recovery, limit of detection (LOD), limit of quantification (LOQ), and precision (intra-day and inter-day). Calculations were performed using a calibration curve (plot of the peak area to the analyte concentration, seven concentration points, n = 3). The method was found to be linear in the range 0.92–10 μg L–1, r2 > 0.9914. The LOD values were calculated based on the equation LOD = 3.3(Sy/S), where Sy is the standard deviation of the response and S is the slope of the calibration curve, obtaining values in the range 0.30–1.23 μg L–1 (n = 6). The LOQ values were calculated based on the equation LOQ = 10(Sy/S), obtaining values in the range 0.92–3.73 μg L–1 (n = 6). The selectivity of the method was also verified, because no impurities were observed in the retention times (RTs) of the target analytes (Fig. 5). The intra- and inter-day precisions were expressed as the relative standard deviation (RSD) of the analyte content and were <20%. The recoveries (Rec, %) were calculated based on the following equation: Rec % = (amount of compound recovered/amount of compound originally present) ×100, and varied from 45.9 to –85.9%. The recoveries for CTC and DC (45.9% and 47.2%, respectively) were particularly low, but this was not totally unexpected. The optimized conditions used in this work, namely the neutral sample pH, do not favour the extraction of some tetracyclines, as already reported in the literature25,31,32.

Detailed data are shown in Table 1.

Table 1 Analytical performance of μSPEed/UPLC-PDA methodology
Full size table

Analysis of real samples

The μSPEed/UPLC-PDA methodology was applied to analyse water samples from Levadas, which are used by local farmers on Madeira Island to irrigate cultivated lands. Three samples were collected from different water channels, filtered, and subjected to μSPEed/UPLC-PDA analysis in triplicates. Replica samples were fortified with a standard mixture containing the target antibiotics at ML concentrations (Table 1) and analysed. No interference was detected at the retention times of the target antibiotics, suggesting that matrix interference can be disregarded. Moreover, none of the selected antibiotics were found in the analysed samples.

Comparison of μSPEed/UPLC-PDA with other extraction methods and greenness assessment

To obtain a clear insight into how the μSPEed/UPLC-PDA proposed here is compared with other equivalent methodologies, we performed a survey of methodologies previously reported using microextraction followed by chromatographic analysis with UV detection for the determination of tetracyclines, SMX, and TMP (Table 2). Many other methodologies employing MS analysis33,34,35 were not considered because although this configuration has the potential to allow better analytical performance, it is far more expensive and labourious. Similarly, promising approaches using custom-made sorbents or solvents and layouts that are not yet commercially available36,37,38,39 were also not discussed in this study. Table 2 summarises the main analytical features of the methods reported from 2018 to 2024. As can be easily observed, μSPEed/UPLC-PDA offers a faster extraction time (6 min per sample), requires much less sample (500 µL), and can simultaneously analyse six frequently used veterinary antibiotics, while delivering comparable analytical performance. In addition to these benefits, the system requires minimal laboratory effort beyond that of the semi-automatic syringe. Furthermore, both the µSPEed sorbent and DigiVol® are commercially available and highly cost-effective, and the electronic pipette can be utilised as a standard pipette, with the significant advantage of being easily programmable for automated tasks. This feature helps reduce the experimental errors typically associated with human operators performing repetitive actions.

Table 2 Comparison of the μSPEed-optimised extraction procedure with those of other studies
Full size table

The experimental setup of μSPEed/UPLC-PDA proposed herein can be readily modified for various matrices beyond environmental water samples, provided that they can be converted into a liquid state with minimal viscosity and complexity, thereby avoiding blockage of the sorbent chamber. This limitation can be overcome by the technique’s high preconcentration capability, which depends on the specific analytes and sample type under investigation.

To compare the proposed μSPEed/UPLC-PDA analysis with the other methodologies listed in Table 2, a detailed greenness assessment was performed. This includes the Analytical Eco-Scale assessment40, Analytical GREEnness (AGREE)41 tool, and BAGI tool42. The eco-scale measures the environmental impact by subtracting points from 100, with higher scores being greener. The μSPEed/UPLC-PDA method scored 76 points, indicating excellent green analysis, which is comparable to the reports presented in Table 2, ranging to from 72 to 79 points. The AGREE tool is based on 12 green chemistry principles and gave the μSPEed approach a score of 0.64, which is one of the highest of the 0.49–0.65 range obtained. The methodologies listed in Table 2 were further considered within the scope of practicality and assessed using the BAGI tool42. In simple terms, this tool measures, for instance, how easy it is to implement a given methodology and how many samples can be processed per unit time. The μSPEed/UPLC-PDA score was 67.5, one of the highest scores achieved in our comparison. Complementary analysis was performed using the green analytical procedure index (ComplexGAPI)43. This tool provides pictograms showing areas for improvement in terms of sampling, transport, solvents, and waste treatment. The data obtained, which are available as supplementary material, once again point to the greener profile of the μSPEed/UPLC-PDA analysis. Overall, the cumulative assessment using the most important green assessment tools available, namely the analytical eco-scale40, AGREE41 and GAPI tools43 and BAGI index42, clearly shows the higher greenness character of the μSPEed/UPLC-PDA analysis proposed here in comparison with other alternatives reported in the literature using similar methodologies.

Conclusion

In this study, a semi-automated solid-phase microextraction technique was validated for the quantification of six antibiotics (tetracycline, chlortetracycline, oxytetracycline, doxycycline, sulfamethoxazole, and trimethoprim) in aqueous samples with good analytical performance. The developed methodology exhibits exceptional efficiency, with a total processing time of just 12 min and a sample volume of 500 μL, outperforming other methods in terms of speed and sample use while maintaining robust analytical performance and environmental friendliness. Moreover, this semi-automated approach reduces the workload for analysts and improves reproducibility. The method’s adaptability and use of affordable, readily available equipment (μSPEed sorbent and DigiVol) make it easily accessible to other laboratories and suitable for analysing the selected antibiotics in various matrices, including food and environmental samples.