Primary standards of Ampicillin trihydrate (99.0%, AMP), Benzathine penicillin G tretrahydrate (97.2%, PNG), Dicloxacillin sodium hydrate (99.0%, DCL), Oxacillin salt hydrate (99.0%, OXA), Tetracycline hydrochloride (97.7%, TC), Doxycycline hyclate (98.7%, DXC), Chlortetracycline hydrochloride (94.6%, CTC) and Oxytetracycline hydrochloride (95.0%, OTC) were supplied by Dr. Ehrenstofer GmbH (Germany). The 2,4-Dinitrophenylhydrazine, potassium phosphate, sodium azide, hydrochloric acid, sodium bicarbonate and calcium chloride were purchased from PanReac (Barcelona, Spain). Analytical grade methanol, sodium hydroxide EMSURE (99%), biotechnological grade sodium chloride, 2-mercaptoethanol, and coomassie blue brilliant G250 (CBB) were purchased from Merck (Darmstadt, Germany). Bovine serum albumin (BSA) free of fatty acids (98%) was supplied from Sigma Aldrich (San Luis, USA). Water was purified with a Milli-Q system (Millipore, Bedford, MA, USA).

Individual stock solutions of tetracyclines and β-lactams (100 μg·mL ^-1) were prepared with milli-Q water and methanol, respectively. For the sample’s contamination with the antibiotics of interest, serial dilutions of the stocks were made until obtaining an antibiotic concentration of 10 μg·mL ⁻¹ (working solution), in order to avoid sample protein precipitation during the sample preparation process. ¹³

Chicken breast samples were purchased from a local market in Cartagena, Colombia. They were stored in plastic polyethylene bags, transported to the laboratory maintaining cold chain (4°C), cut and cleaned to remove visible connective tissue and then, mechanically homogenized using a blender (Powergen by Fisher). Then, one-gram replicates of homogenized samples were collected in falcon tubes and stored at -20°C until further assays. This procedure was described by Marquez et al. (2020). ¹³

Chicken breast aliquots were randomly divided in control samples and samples to contaminate with tetracyclines and β-lactams. Then, samples were contaminated individually with 200 μL of each antibiotic working solutions until reaching final concentrations of 0.5, 1.0 and 1.5 MRL ( Table 1), followed by vortexing for 30s and then incubation for 1 h at room temperature in the dark. ¹⁴ All assays were carried out with three replicates.

To obtain the sarcoplasmic, myofibrillar and insoluble proteins, the method developed by Marquez et al. (2020) was employed. Briefly, homogenized samples were mixed with 10 mL of low ionic strength buffer (0.05 M K ₃PO ₄, 1 mM NaN ₃, 2 mM EDTA, pH 7.3), then, vortexing at 3,000 rpm for 3.5 minutes and centrifugation at 11,150 xg 10 minutes and 1 °C were carried out. The supernatant (sarcoplasmic proteins) was collected and refrigerated, and the pellet was resuspended with 5 mL of low ionic strength buffer, centrifuged and the supernatant was unified with the sarcoplasmic proteins fraction initially obtained. The remaining pellet was resuspended with 5 mL of high ionic strength buffer (0.55 M KCl, 0.05 M K ₃PO ₄, 1mM NaN ₃, 2 mM EDTA, pH 7.3) and the same vortexing and centrifugation steps were realized.). The new supernatant (myofibrillar proteins) was collected and also kept refrigerated at 4°C. The final pellet was resuspended in 0.15 M KCl to obtain the insoluble proteins. ¹³ Protein concentration was determined by the Bradford method (Coomassie blue brilliant/ethanol/phosphoric acid). For this, a calibration curve was realized using standard solutions of bovine serum albumin at concentrations between 0.0625 and 1.0 mg·L ⁻¹, which were put in microplates with the samples and the Braford reactive. Then, the absorbance was measured at 595 nm. ¹⁵

Protein carbonyl content was measured according to Mesquita et al. (2014) with some modifications described by Marquez et al. (2020). ^{13, 16} For this, 300 μL of 2,4-dinitrophenylhydrazine (DNPH, 10 mM in 0.5M H ₃PO ₄) was added to the same volume of protein solution (150 μg) and incubated in the dark for 10 minutes. Next, 160 μL of this solution was placed in a 96-well plate and 40 μL of 6M NaOH were added and incubation for 10 minutes was carried out. The change of colour was measured at 450 nm (FLUOstar Omega spectrophotometer, BMG-Lab Tech). The protein carbonyl content was calculated using the DNPH molar extinction coefficient corrected for microplates (ε = 11154 μM ^-1.cm ^-1). Protein carbonylation was expressed as nmol of carbonyls·mg ^-1 of protein.

Values are reported as mean ± SEM of three independent determinations. To assess the effect of antibiotics concentration on protein carbonylation, a unifactorial ANOVA with three levels and blank at 95% confidence was performed ( GraphPad Prism V5.01). Multiple comparisons of the means were made using the Tukey adjustment when the ANOVA was significant ( p < 0.05) for control and contaminated samples.

The oxidative capability of the assayed antibiotics was evaluated by determination of the carbonylation degree of exposed sarcoplasmic, myofibrillar and insoluble chicken breast proteins, and was expressed as the carbonyl index (CI) in nmol of carbonyls·mg ^-1 of proteins. CIs found in the control samples were on average 7.40 ± 0.85; 8.67 ± 1.03 and 12.79 ± 0.5178 nmol of carbonyls·mg ^-1 of proteins for sarcoplasmic, myofibrillar and insoluble, respectively. When those values were compared with the treated samples, it was observed that 16 of the 24 treatments assayed (66.7%) significantly increased their carbonyl groups, as shown in Figure 1.

The CIs induced by antibiotics on protein fractions were between 3.92 – 31.38; 6.72 – 50.07 and 11.77–61.49 nmol of carbonyls·mg ^-1 of proteins for sarcoplasmic, myofibrillar and insoluble, respectively. In addition, the developed Kruskal-Wallis test among the CIs from the three proteins groups showed that there were significant differences between their medians ( p<0.05), the insoluble proteins being the ones that showed the highest CIs (28.04 ± 1.61 nmol of carbonyls·mg ^-1 of proteins), followed by myofibrillar and sarcoplasmic proteins with mean values of 23.07 ± 1.42 and 17.28 ± 0.89 nmol of carbonyls·mg ^-1 of proteins, respectively. Next, we described the behavior of induced carbonylation by each group of antibiotics.

TCs’ oxidative capability is shown in Figure 2. In sarcoplasmic proteins, all assayed concentrations of CTC, OTC and DXC induced a significant carbonylation compared to the control ( p < 0.05); in contrast, the samples contaminated with TC, only showed a significant increase at 0.5 and 1.5 MRL ( p < 0.05). The maximum oxidant power of TC, CTC and DXC was observed at 1.5 MRL, while OTC was set at 0.5 MRL. Contrarily, in myofibrillar proteins, TC, CTC and OTC promoted a significantly higher carbonylation compare to control ( p < 0.05) at all concentrations assayed. DXC only induced a significant carbonylation at 1.0 MRL regarding to the control ( p < 0.05). Unlike to sarcoplasmic proteins, the maximum oxidant power of TC, OTC and DXC was observed at 1.0 MRL, while CTC’s continued at 1.5 MRL ( p < 0.05).

Finally, for insoluble proteins, all concentrations of tetracyclines induced significant carbonylation with respect to control ( p < 0.05). The maximum carbonylation promoted by TC and CTC was observed at 1.0 and 1.5 MRL ( p > 0.05), while OTC and DXC were at 0.5 and 1.5 MRL, respectively.

β-Lts-induced carbonylation is shown in Figure 3. In sarcoplasmic proteins, all the assayed concentrations induced significant carbonylation on contaminated samples compared to control ( p < 0.05). For samples contaminated with AMP and PNG, maximum carbonylation was observed at 1.5 MRL, while in samples contaminated with DCL, maximum carbonylation was set at 1.0 MRL;for OXA the values were set at 0.5 and 1.0 MRL. There were no statistical differences between their CIs ( p > 0.05). In contrast, in myofibrillar proteins, DCL and OXA induced the same carbonylation in treated samples at all assayed concentrations ( p > 0.05), while AMP only induced a significant carbonylation at 0.5 and 1.5 MRL with respect to control ( p < 0.05), and PNG a 1.5 MRL carbonylation. Finally, for insoluble proteins, PNG, DCL and OXA promoted a significantly higher carbonylation than control at all concentrations assayed ( p < 0.05), while AMP only led to a 1.5 MRL carbonylation. For AMP and OXA, the maximum oxidant power was induced at 1.5 MRL, whereas that of DCL was similar for all three concentrations ( p > 0.05).

Figure 4 shows the oxidant power of TCs and β-LTs on sarcoplasmic, myofibrillar and insoluble proteins from chicken breast; oxidant power was calculated using the ratio between the CIs from treated samples and from the controls (Carbonyl index of protein fraction from contaminated samples/Carbonyl index of blank samples). As showed, among all antibiotics assayed, CTC and OTC promoted the highest carbonylation in myofibrillar proteins at all assayed concentrations (with average ratios of 7.9 and 5.4, respectively); these were followed by DCL in the insoluble (average ratio of 5.4) and myofibrillar fractions (average ratio of 4.0) and OXA in the sarcoplasmic and myofibrillar fractions (average ratios of 3.7 and 3 respectively). Oppositely, DXC, AMP and PNG induced the lowest carbonylation, mainly in myofibrillar (average ratios of 0.4, 0.5 and 0.5, respectively) and insoluble fractions (average ratios of 1.4, 0.7 and 1.4, respectively).

According to the carbonylation variability induced by the assayed antibiotics, we decided to assess individual and multiple correlations among CIs and independent variables assessed in the study: protein fraction, type of antibiotic, evaluated concentration of each antibiotic, and octanol-water partition coefficient (LogP). This last variable was included given that the antibiotics evaluated, by belonging to the same group, have similar chemical structures (pharmacophore) differing only in their radicals; therefore one of the most used parameters to evaluate differences in their toxicological behaviors in in vitro models are their solubility variations (LogP). ¹⁷

Simple regression analysis results between CIs induced by antibiotics at their respective concentration, and LogP revealed a significantly positive relationship for both variables combinations (R ² = 0.0327, p = 0.0044 for CIs and antibiotics concentrations, and R ² = 0.1235, p = 0.0019, for CIs and LogP). This indicates that they represent important parameters in the oxidant properties of antibiotics ( Figure 6a and 6b). When comparing adjusted determination coefficients obtained in correlation analyses, we were able to establish a hierarchy in the effect exercised by the evaluated independent variables (concentration and LogP) on the antibiotics-induced carbonylation, showing the following increasing order: LogP > concentration. Nevertheless, in order to evaluate the joint correlation level of independent variables assayed (concentration and LogP) vs the antibiotic-induced carbonylation, a multiple linear regression analysis was performed. Thus, the comparison between adjusted R ² obtained in the simple and multiple linear regression analysis showed that R ² in combined variables (14.48%) was higher than that of individual ones (3.27% and 12.35% for concentration and LogP, respectively), thus attributing the antibiotic-induced carbonylation effect to the synergic effect of the evaluated independent variables ( p < 0.0001).

Figshare: Raw data of chicken breast proteins exposure to antibiotics, https://doi.org/10.6084/m9.figshare.14799147.v3

This project contains the following underlying data: •

Data set 1: Raw data from carbonyl assay of tetracyclines.

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).