Escherichia coli is a frequent and crucial pathogenic bacteria which can cause septicaemia, enterocolitis and diffuse peritonitis, resulting in high mortality rates, increased morbidity and devastating economic losses in the livestock industry (Burt and Reinders, 2003; Splíchal et al., 2005; Laghari et al., 2013; Ijaz et al., 2015).

Marbofloxacin (MBF) is a synthetic third-generation fluoroquinolone antibiotic developed exclusively for veterinary treatment (EMEA, 1996). With a broad spectrum of bactericidal activity, marbofloxacin acts primarily as a bactericidal antibiotic for Gram-negative pathogens, some Gram-negative pathogens and Mycoplasma (Haritova et al., 2006; Vallé et al., 2012; Tohamy and El-Gendy, 2013). It is proposed for oral or parenteral administration for the treatment of respiratory and archenteric disease in pigs or bovines, with a high bioavailability close to 100% (Ding et al., 2010). Its extensive spectrum of activity also includes infected canine pathogens such as Staphylococcus spp., Proteus spp., Streptococcus spp., and E. coli, which has been approved for treatment in pets at a dosage of 2.0 mg/kg.bw once per day via oral administration (Spreng et al., 1995; Thomas et al., 1997; Paradis et al., 2001; Yu et al., 2007).

The pharmacokinetics (PK) of MBF has been investigated in various livestock such as goats, cows, cats, sheep, pigs, and dogs (Waxman et al., 2001; Schneider et al., 2004; Albarellos et al., 2005; Ding et al., 2010; Sidhu et al., 2010), showing a high concentration in plasma and peripheral tissue, with up to 100% bioavailability, and wide and rapid distribution in tissues. Previous studies have shown MBF to have excellent pharmacokinetic characteristics such as being well-absorbed after oral and parenteral administration, higher concentrations in tissues than in the plasma, and weakly bound to plasma proteins (<10%) (Haritova et al., 2006; Andraud et al., 2011; Sun et al., 2015). MBF is distributed widely throughout the animals’ bodies, which could reach 1.6 times higher drug concentrations in the skin than in the plasma in dogs, and MBF concentrations in the plasma can remain above the MIC (>24 h) longer than the dose density (Cester et al., 1996; Schneider et al., 1996). Although some studies have reported the pharmacokinetics of MBF in animals including pigs (Ding et al., 2010; Schneider et al., 2014; Yang et al., 2016), the pharmacokinetic data of MBF in pigs plasma and intestinal content are not sufficient enough to precisely predict the efficacy of this drug.

For the evaluation of antimicrobial drugs, it is essential to optimize the dose schedule to attain clinical cures and reduce the emergence of antimicrobial drug resistance (Burgess, 1999). The pharmacokinetics-pharmacodynamics (PK/PD) integration model can reveal the relationship between antibiotics and bacterium in specific animals, and quantify the potency and efficacy of antibiotics against bacterium. Moreover, the PK/PD integration model can also prevent resistance development and provide optimal dosage strategies (Drusano, 2007; Nielsen and Friberg, 2013). As an effective tool for assessing the optimal dosage regimens, PK/PD analysis has been recommended in the development of new antimicrobial compounds by the Food and Drug Administration (FDA) and European Medicines Agency (EMA) (Yan et al., 2017).

Few researches have investigated antibiotic concentrations at the ileum content in pigs. Due to the difficulty in determining free drug concentrations at the infected site, the PK data were obtained from the plasma in previously described reports. However, it has been demonstrated that the concentration in the plasma is significantly different from that in the target sites, such as ileum content, epithelial lining fluid and interstitial fluid, which could be observed in the results of previous studies (Messenger et al., 2012; Foster et al., 2016; Wang et al., 2016; Yan et al., 2017). Thus, it is recommended to determine the drug concentrations at the target site for PK-PD modeling to achieve more rational dosage regimens.

Since there are contradictory values in PK/PD indices which have correlated with the prevention of resistant mutant selection and efficacy, great attention has been paid to specific PK/PD indices for specific pathogens and antimicrobial agents. Prior to this study, the integration of PK data in pigs with the time course activity ex vivo for MBF against E. coli had not been performed. The antibacterial activity in vivo of MBF against E. coli and PK in the ileum content were evaluated and a typical method simple-T cannulation was used to obtain the MBF concentrations in the ileum content. Moreover, the rational dosage regimen of MBF against E. coli was established for veterinary clinical guidance based on PK-PD integration modeling.

As a new fluoroquinolone, MBF had complete absorption, and high drug concentrations in ileum content, which was the target site infected by E. coli after oral administration, and high antimicrobial activity in vitro and ex vivo. However, due to the overuse and misuse of antimicrobial agents, the resistance of the pathogenic bacterium to fluoroquinolone or other drugs was increasingly more serious (Burgess, 1999). An increasing number of studies about E. coli resistance to fluoroquinolone including MBF have been reported in most countries, such as Europe, the United States, China, and so on (Liu et al., 2013, 2015; Cheng et al., 2014; El Garch et al., 2017). The cross-resistance to similar antibiotics was widespread, and the resistance mechanisms were complex. Thus, it was crucial to reasonably use and effectively manage these new generational antibacterial agents, and necessary to establish an optimal dosage for MBF curing E. coli in this study.

In this study, 218 isolated E. coli strains were selected for PD research, such MIC, MBC, MPC and time-killing curve evaluations when exposed to MBF. In total, 142 of all selected strains were screened as sensitive populations according to the CLSI breakpoint guidance document M100-S23 (MIC ≤ 2 μg/ml) and other published reports (Andraud et al., 2011). The MIC values were in the range from 0.03215 to 64 μg/ml in whole strain populations, and 0.03125 to 2 μg/ml in the sensitive populations, respectively. The MBF-resistant E. coli, which had been found in many other reports, were also identified (Pellet et al., 2006; Andraud et al., 2011); the results in this study were similar to those in published reports. The mechanism of resistance of E. coli to MBF may be the abuse of antibiotics and cross-resistance reduced by other earlier fluoroquinolone drugs (Huguet et al., 2013; Sato et al., 2013). In the sensitive E. coli populations, the MIC₅₀ and MIC₉₀ were 0.5 and 2 μg/ml, respectively (Figure 2). These results revealed that MBF had high antibacterial activity in vitro. A highly pathogenic clinical isolate, HB197, was selected in accordance with the pathogenicity experiment in mice (data not shown) and MIC₉₀. The MIC values and growth curves of HB197 in LB and ileum content showed no significant difference (Figure 1), and the time-killing curves in vitro showed that MBF was bactericidal against E. coli, with a concentration-dependent type (Figure 3). The 2 MIC concentration of MBF could completely eliminate E. coli after 8 h, with results similar to those previously described by Pellet et al. (2006) and Andraud et al. (2011), and other fluoroquinolone-like Enrofloxacins against E. coli in the reports (Wang et al., 2016). According to the profiles of Figure 3B, the drug concentration of MBF at the different time points from 0.25 to 4 h achieved from the plasma in PK study, had strong bactericidal effect within 8 h. Moreover, it also had bacteriostasis effect at the time points from 6 to 8 h. All of these results demonstrated that MBF had strong antibacterial activity against E. coli in vitro and in the ileum.

As far as we know, PK/PD integration modeling was an effect approach to dose titration studies for selecting rational dosage regimens in veterinary medicine (Toutain and Lees, 2004). Furthermore, the measurement at the infection site for PK and PD was a preferred method to analyze and correlate PK/PD modeling (Liu et al., 2002; Mouton et al., 2008). The PK of drug concentrations was investigated in serum as the ex vivo data in most published PK/PD modeling studies, and there were also studies for MBF against Haemophilus parasuis and Pasteurella multocida in beagle, sheep, calves, turtles, broiler chicken, pigs, and so on (Sidhu et al., 2010; Vallé et al., 2012; Potter et al., 2013; Qu et al., 2015; Sun et al., 2015; Yohannes et al., 2015). However, the PK of MBF in the ileum content which was the target infection site by E. coli of pigs was the first to investigate using ileac cannulation. The ileac cannulation in vivo could keep target animals in a normal physiological state; this state for animals had the advantage of obtaining a more accurate concentration for PK data (Garrison et al., 2002).

The PK data in plasma after oral administration of the recommended dose of 2 mg/kg for C_max, AUC_{0-24 h}, AUC, Ke, t_1/2ke, MRT and Cl_b were 0.55 μg/ml, 8.15, 14.67 μg⋅h/ml, 0.023 h^-1, 30.67 h, 34.83 h, 0.14 L/h, respectively (Table 1). The value of t_1/2ke (30.67 h) in this study was similar to that of a published report (23.14 h) both after orally administration (Ding et al., 2010), but higher than those in broilers (5.26 h), buzzards (4.11 h), beagle dogs (7.51 h), pigs (17.3 h), and turkeys (7.37 h) after intramuscular injection (Garciamontijano et al., 2001; Anadón et al., 2002; García-Montijano et al., 2003; Haritova et al., 2006). However, the value of C_max (0.55 μg/ml) after oral administration was lower than that after intramuscular injection, but similar to that (0.67 μg/ml) after oral administration reported by Haritova et al. (2006). The differences in these results were due to the routes of administration, species and dosing differences. Moreover, the value of AUC (14.67 μg.h/ml) was similar to that (11.37 μg.h/ml) reported in beagle dogs by Yohannes et al. (2015) and that reported in poultry (10.86 μg.h/ml) by Haritova et al. (2006).

Compared to plasma, the drug concentrations in the ileum content were significantly higher, with C_max and AUC of 11.28 μg/ml and 77.81 μg.h/ml, respectively, which were 20 and 5.3 times higher than those in plasma, respectively (Table 1). There were no other studies of the concentration of MBF in ileum content, and this was the first to evaluate the PK of MBF in the ileum content in this study. The difference in concentrations between plasma and intestinal contents could be due to the high lipophilicity of MBF; this could be the reason why the Cl_b (0.14 L/h) in plasma is much higher than that (0.026 kg/h) in intestinal contents in this study. These results revealed that MBF had a strong penetration ability for various tissues and easily accumulated in intestinal contents following oral administration. These characters were also similar to those of other studies of MBF in beagle dogs (0.17 L/h), pigs (0.21 L/h), and foals (0.34 L/h) (Vallé et al., 2012; Tohamy and El-Gendy, 2013; Yohannes et al., 2015). At the same time, MBF could also penetrate membranes and tissues, binding to solid parts of the ileum content. The drug concentration dissolved in the aqueous phase was the main activity and worth detecting. Due to the high drug concentrations in ileum content, the AUC (77.81 μg.h/ml) was much higher than that (22.56 μg.h/ml) in plasma by intravenous injection administration in the previously described report (Ding et al., 2010), the bioavailability was considered as 1 for calculation, and it also was reported in the published reports (Wang et al., 2016; Yan et al., 2017).

As a kind of concentration-dependent action for MBF, the parameters C_max/MIC > 10 and AUC_{0-24 h}/MIC > 125 were used as a threshold for the successful therapeutic outcome of fluoroquinolones against gram negative bacteria (Toutain et al., 2002). However, these thresholds may be different for different fluoroquinolones. There were differences in the immune status of target animals and pathogens. The published AUC_{0-24 h}/MIC was 46 h for bactericidal action in PK/PD study of MBF against Mannheimia haemolytica, and 88 h for the bactericidal action of MBF against Haemophilus parasuis (Aliabadi and Lees, 2002; Sun et al., 2015). Therefore, it was of great importance to study the PK/PD indices of fluoroquinolones including MBF individually. In this study, the PD data were obtained from ileum content to predict dosage regimens since it was more clinically relevant than those from broth. The ratios of ex vivo C_max/MIC, C_max/MPC, AUC_{0-24 h}/MIC, and AUC_{0-24 h}/MPC were 5.64, 1.37, 23.08 h and 5.63 h, respectively in ileum content (Table 2). The inhibitory sigmoidal E_max model was used for PK/PD integration model and dosage prediction, and it showed a favorable correlation (0.995) between the observed and predicted antibacterial efficacy of MBF against E. coli ex vivo (Figure 6). The ex vivo AUC_{0-24 h}/MIC ratios of MBF requiring bactericidal action and eradication of the clinical strain with MIC₉₀ of 2 μg/ml were 23.54 and 27.18 h, which were similar to in vivo AUC_{0-24 h}/MIC (23.08 h) achieved after oral administration (2 mg/kg). These results suggest that the recommended dosage of 2 mg/kg could guarantee clinical efficacy against infections associated with sensitive E. coli with an MIC₉₀ of 2 μg/ml. Based on the Monte Carlo simulations, the predicted daily dose for 50 and 90% targets to achieve bactericidal effect were 1.22 and 1.33 mg/kg, respectively. The Monte Carlo simulation to predict dosage for clinical use had the advantage of taking into account the PK/PD parameters based on bacteriological outcome. Furthermore, it could set target percentage such as 90 and 50% for simulation models for all data in relation to incidence in pigs (Dorey et al., 2017). However, due to the animals’ immune system also being an important factor contributing to bacterial eradication, the bacterial endpoint in vivo conditions may differ from the predicted dosages ex vivo (Garrison et al., 2002; Yan et al., 2017).

Most studies have demonstrated that the increasing antimicrobial resistance in gut flora was due to the misuse of antibiotics in unsuitable and irrespective administration routes (Nguyen et al., 2012; Zhang et al., 2013). The PK/PD model in the infection target site could provide more reasonable dosages. According to our PK/PD parameters ex vivo, the single doses required to reach bacteriostatic, bactericidal, and eradication activity for 90% target were 0.92, 1.33, and 1.53 mg/kg, respectively. These results demonstrated that the administered dosage (2 mg/kg) of marbofloxacin by oral administration daily could have an effective bactericidal effect against E. coli. Furthermore, it provided an alternative optimal dosage regimen (1.33 mg/kg for bactericidal and 1.53 mg/kg for eradication) and avoided the emergence of resistance for clinical veterinary use. However, the predicted dosage regimens should be validated in clinical practice to evaluate the treatment effect of infected pigs in future research.

JC conceived the study; JC and ZL designed the experiments; ZL, QL, and BY performed the experiments; ZL wrote the manuscript; QH, QZ, SY, KL, JX, and SZ improved the language. All authors reviewed the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.