2.1. Single Drug Toxicity
The CCK-8 assay was performed on THLE-2 hepatocytes to determine the cytotoxicity of enrofloxacin (ENR), ciprofloxacin (CFX), florfenicol (FFC), and sulfadimidine (SMD). All the drugs exhibited a dose-dependent inhibition (
Table 2). The Dm values of ENR, CFX, FFC, and SMD were 13.11, 32.03, 392.5, and 358.6, respectively. ENR exhibited the greatest toxicity, with an inhibitory ratio of 58.78:84.25 within the dose range of 25 μg·L
−1 to 500 μg·L
−1. CFX revealed much higher toxicity than FFC and SMD at the same concentrations. These three drugs had an inhibitory ratio of 41.35:85.84, 17.93:54.75, and 18.39:54.55 within the dose range of 25 μg·L
−1 to 500 μg·L
−1, respectively.
The CalcuSyn2.0 software was used to generate dose–effect curves and median-effect plots for single drugs in
Figure 1A,B. All the (r) values of the median-effect plots were above 0.95, demonstrating that the experimental data agreed well with the median-effect equation of Chou. The dose–effect curves and median-effect plots did not fit well at very low concentrations since antimicrobials promote growth at low concentrations. The dose–effect curves of all the drugs had a flat sigmoidal shape (M < 1). Dm affords the toxic potency on THLE-2 cells, ENR > CFX > FFC > SMD. It was interesting that the ENR and CFX showed a cross point at a dose of about 180 µg·L
−1, which was near their MRLs in poultry liver (200 µg·kg
−1). The toxicity expression varied below and above this MRL for this two-drug combination, with the same primary mechanism, suggesting that they must have a different secondary toxicity mechanism.
2.2. Joint Toxicity of Three Binary Drug Combinations
The combined toxicity of ENR-CFX, ENR-FFC, and ENR-SMD was calculated. The drugs were initially mixed at the concentration ratio of 1:1 in six different dose groups. CalcuSyn2.0 software was again used to calculate the dose–effect curves, median-effect plots, and CI values for the binary combination in
Figure 1C–H. All the curves correspond well with the median-effect equation of Chou with a correlation rate of over 0.88.
CI values varied at different concentrations on THLE-2 cells, as shown in
Table 3. The joint toxicity of ENR-CFX showed synergism over certain dose ranges, and their CI values ranged from 0.264 to 0.651 within the dose range of (25,25) to (250,250) µg·L
−1. The joint toxicity of ENR-FFC was mutually enhanced at high concentrations, with CI values ranging from 0.383 to 0.831 within the dose range of (100,100) to (500,500) µg·L
−1. The joint toxicity of ENR-SMD exhibited synergism at each dose, except (25,25) µg·L
−1. The CI values of ENR-SMD ranged from 0.453 to 1.003. The dose–effect curves of all binary combinations showed a flat sigmoidal shape (M < 1). We conclude that on THLE-2 cells, ENR-CFX, ENR-FFC, and ENR-SMD exhibited dose-dependent synergistic toxicity, and the synergistic toxicity of ENR-SMD was the most notable.
Then, we determined whether the mixing ratio impacted joint toxicity. Three binary combinations were mixed at the ratios of 1:2, 1:4, 2:1, and 4:1 for at least four concentrations. Similarly, we performed another CCK-8 assay and calculated the CI value for each group via CalcuSyn software. The results are presented in
Table 4,
Table 5 and
Table 6, and
Figure 2.
A significant difference was observed in some drug combinations and mixing ratios. When the mixing ratio was 2:1 and 4:1, the joint effect of ENR-CFX could either exhibit synergism or antagonism. In contrast, when the mixing ratio was 1:2 and 1:4, their joint effect showed only synergism. In addition, for the binary combination ENR-SMD, their joint toxicity showed the strongest synergism when the mixing ratio was 1:4, but not 1:1 at the concentration of (5,20) µg·L−1 with the CI value of 0.245.
We sought to compare their CI values with the same effect to visualize the joint toxicity difference of each mixing ratio. The CI values of each mixing ratio were predicted at ED50, ED75, and ED90, respectively, using the previously obtained median-effect plots (
Figure 3).
The synergistic toxicity for ENR-CFX was strongest at the 1:1 ratio, but weakest at the 4:1 ratio. The synergistic toxicity for ENR-FFC was the strongest at a ratio of 1:4 with an ED50, while at ED75 and ED90, the synergistic toxicities of ENR-FFC at 1:4 1:2, 1:1, and 2:1 were similar. The synergistic toxicity for ENR-SMD was stronger at 1:4 and 4:1 compared to other mixing ratios. Thus, we demonstrated that the joint toxicity of binary drug mixtures is mixing-ratio-dependent.
Previous studies on joint toxicity typically used a 1:1 mixing ratio, but our results argued that an experimental design with a single mixing ratio is inappropriate. The clinical dosage of drugs in this study was similar; thus, their residue levels in food are roughly the same. However, during actual use, there may be situations in which the dosage is privately changed, resulting in different residual proportions of drugs in food. Therefore, a pairwise testing of 1:1, 1:2, 1:4, 2:1, and 4:1 mixing ratios or other combinations of drugs is required to obtain proper results. ENR and CFX are frequently used in animals, and their chemical structures are similar; furthermore, the former can be metabolized in vivo to the latter [
13]. As a result, ENR and CFX were recognized as drugs with a similar mode of action. ENR and CFX reportedly inhibit CYP450 enzymes, which are responsible for drug metabolism in liver cells [
27]. The inhibition of CYP450 may be a possible mechanism for the cytotoxicity of these drugs. However, combining drugs with identical mechanisms can only lead to addictive effects or antagonism. The synergistic toxicity of ENR-CFX indicates that ENR potentially has a different toxic mechanism from that of CFX. FFC was recently shown to induce noticeable cytotoxicity by inhibiting mitochondrial protein synthesis [
28]. These two cytotoxicity mechanisms may combine, leading to synergistic toxicity. The mechanisms of combined toxicity are still uncertain and require further studies.
The median-effect equation of Chou is derived based on enzyme kinetic models of the law of mass-action, widely recognized in the field of medicine [
29]. One of the advantages of the Chou–Talalay method is that it does not require many experiments. For each mixing ratio of each drug combination, four to seven determining concentrations are sufficient for fitting median-effect plots. Data for calculating CI values at each Fa can be acquired by implementing a coefficient simulation, greatly reducing experimental costs. The Chou–Talalay method provides two Formulas (2) and (3) useful for calculating drugs with the same or different modes of action. However, there is generally no significant difference between the computed results of these two formulas. Since synergistic joint toxicity results are more useful to food safety risk assessment, ENR-FFC and ENR-SMD were treated as mutually non-exclusive drugs to ensure that CI < 1 indicates synergistic joint toxicity.
Organisms can be enriched by veterinary antimicrobials in the environment [
30]. As the detection technology developed, several residual antimicrobials in fresh food can be detected using one-time testing [
14]. Knowing the joint effects helps in food safety assessment, yet toxicology research for veterinary antimicrobials still focuses on single drugs. Taking enrofloxacin as an example, our study shows that its potential toxicity in fresh food is affected by other antimicrobials. Enrofloxacin’s toxicity is enhanced when combined compared to the toxicity from each drug residue alone—whether the drugs have the same or different mechanisms. Therefore, it is important to establish a database of joint toxicities of veterinary drugs. The original MRLs need to be adjusted, or a new MRL evaluating standard needs to be developed in which drug interactions are considered. In our opinion, it is important to evaluate and refine existing methodologies for assessing risks of exposure to two or more veterinary antimicrobials in combination, particularly in the context of setting MRLs in accord with government regulations such as EC 396/2005. Ideally, risk assessments in veterinary medicine should consider all possible residues (e.g., individually or in different ratios of combination) that influence pathways (e.g., fresh food, processed food, feeds) and routes of exposure (e.g., ingestion, dermal, inhalation) contributing to total exposure. However, appropriate data on levels of exposure in veterinary medicine from pathways and sources are not generally available, and further research is required. Therefore, the actual MRLs still need to be discussed while fresh foods face multiple sources of pollution of veterinary antimicrobials.
Nevertheless, the method in this study has limitations, such as it cannot evaluate the joint toxic effects that are difficult to quantify, such as neurotoxicity. Other approaches, such as animal studies, are required in this case.