Marbofloxacin standard with a purity over 98% was received from Dr. Ehrenstorfer GmbH (Augsburg, Germany). MBF powder a purity over 97% (No. 201104006) was provided by Wuhan Huishen Biotechnology, Co., Ltd. The MBF solution (40 mg/ml) was prepared with normal saline added 0.1% acetum solution. The chemical agents were all high-performance liquid chromatography level and the other solvents were also analytical level in this study. Each isolate strain was subcultured for over three times to reach stable growth in tryptone soya broth (TSB) and tryptone soya agar (TSA; Qingdao Hai Bo biological Technology, Co., Ltd.) with 5% newborn calf serum (Zhejiang Tianhang Biotechnology Co., Ltd.).

A number of 8 healthy pigs with four females and four males were selected from 15 to 20 kg and from 8 to 10 weeks for this study and provided by the Hubei Provincial Laboratory of the Public Center for Animal Services. The animals received feed and water without antibiotics ad libitum in the separation fence during the experimental period, and the premixes were devoid of polyvalent cations affected by MBF uptake, such as aluminum, magnesium, calcium, iron, etc. In addition, pigs were fasted overnight and kept for over seven days to acclimate before the formal experiment study. This study was approved and permitted by the Ethical Committee of the Faculty of Veterinary Medicine from Huazhong Agricultural University. In addition, all of the animal care and experimental protocols from this study were performed in accordance with the Guide to the Care and Use of Laboratory Animals of Hubei Provincial Laboratory Animal Public Service Centre with a permit number SYXK 2013-0044.

One-hundred-thirty-four SS strains were isolated from Chinese pigs (from Henan, Sichuan, Hubei, Anhui, Hunan, Jiangxi, and so on) from 2016 to 2017. From these, we selected the SS HB2 strain (serotype 2) to detect the susceptibility effect of MBF in vitro. The SS HB2 strain was selected because its minimal inhibitory concentration (MIC) to MBF was similar to the MIC₉₀ of MBF against SS (134). These strains were authenticated with polymerase chain reaction (PCR). Before MIC detection, these isolates were subcultured at least three times in TSB and TSA. All isolates were kept at -80°C (Lei et al., 2017b).

Susceptibility detection (MIC) for MBF against SS was conducted with the agar dilution method which was based on the guidance of CLSI (Jorgensen, 1993; Lei et al., 2017c). The strains (2–4 μl) with 10⁸ CFU/ml were inoculated into TSA including 5% newborn calf serum, with two-fold serial dilutions of marbofloxacin (0.125–64 μg/ml). Those isolates with MIC values over 64 μg/ml were re-tested using a broader range of MBF dilutions. Inoculated plates were kept at 37°C for 48 h. The MIC value was considered the lowest drug concentrations that caused complete growth inhibition by the naked eye. E. coli (ATCC 25922) was performed as a quality control (QC) strain to check the results of the above susceptibility testing.

For MIC and minimal bactericidal concentration (MBC) determination of HB2 in vitro and ex vivo, it was performed using the microdilution technique following the guidelines of the CLSI, and MBC was tested by using inoculating the supplemented TSA with 100 μl of suspension received from the initial MIC detection with no distinct bacteria. The MBC value was considered the value at the lowest concentration of MBF with inhibiting bacterial density by 99.9%.

The mutant prevention concentration (MPC) of MBF was tested using the agar dilution method. The10¹⁰ CFU/ml of SS (HB2) was prepared to test the MPC on TSA plates (Blondeau et al., 2010). Then, the suspension of SS was unfolded onto TSA, containing serial dilutions of MBF (1–32 MIC); the MPC was the value at the lowest concentration with inhibiting bacterial growth at 37°C for 96 h.

The isolate of HB2 was selected to test the growth-time curve of SS by optical density (OD_{600 nm}). The growth curves of HB2 in serum and TSB were performed at 0, 2, 4, 6, 8, 10, 12, and 24 h using optical density.

According to the MIC value (1 μg/ml) of HB2 to MBF, TSA plates containing various MBF concentrations in the range of 1/4 to 32 MIC were prepared. For the vitro killing-time curve, 10⁶CFU/ml bacteria-containing suspension fluid was diluted to obtain with normal sterile saline. Then 100 μl dilute samples were spread onto the TSA plates at different time point including 0, 2, 4, 6, 8, 10, 12, and 24 h. Finally, these samples were cultivated at 37°C for over 48 h.

For the killing-time curves in ex vivo (MBF in serum received at various time points from 0 to 24 h in PK study), the bacteria (10⁶ CFU/ml) were cultivated with serum samples containing various MBF concentrations. The ex vivo killing-time curve was administrated to accommodate an appropriate PD model using the hypothesis that a logarithmic decrease of bacteria amounts under MBF concentration according to the incubation time with the inhibitory sigmoid E_max model.

Eight pigs with four male and four female weighing 16–20 kg and aged 5–6 weeks were selected in this study. Each pig was administrated MBF (8 mg/kg) by oral administration. After a radical washout period (2 weeks), the pigs received MBF by intravenous injection (i.v, 8 mg/kg). The blood samples (5 ml) were collected at 0, 0.25, 0.5, 1, 2, 4, 6, 8, 10, 12, 24, 36, 48, 72 h after i.v and oral administrations.

The samples were collected with an anticoagulant and then centrifuged for 10 min at 3000 rpm to acquire the serum. 2 ml of dichloromethane was added to 0.5 ml of serum, the tubes vortexed for 2 min, and centrifuged for 10 min at 5000 rpm. The above procedures were repeated to perform twice. The dichloromethane fluid was taken to a clean tube and evaporated with nitrogen in a thermostat water bath at 60°C. A portion of the mobile phase (0.5 ml) was used to dissolve the dried sample. The dissolved samples were filtered through the membrane filters (0.22 μm) and analyzed by HPLC with UV for detection.

Serum protein binding of MBF was determined in triplicate on each of nine pooled blood samples, harvested at predetermined times from the eight pigs used in the PK study. The samples were centrifuged at 3000 rpm for 10 min to obtain the serum. For each sample (serum), the total concentration of MBF was determined as follows. Samples were centrifuged for 10 min at 4000 g using an Amicon Ultra Centrifugal Filter (Ultracel 10kD; Millipore Limited, Watford, Hertfordshire, United Kingdom) and MBF concentration re-determined on the ultrafiltrate.

The extraction samples in serum were analyzed using a C18 reverse-phase column (250 × 4.6 mm, i.d., 5 μm, Agilent, United States) and HPLC, which was performed with a detection wavelength of 299 nm at 30°C for the column. The mobile phase including 0.1% formic acid (phase A) and acetonitrile (phase B) (v:v, 18:82) with a mobile phase flow rate of 1 ml/min. In addition, the injection volume was 20 μl. The HPLC method validations of MBF in serum were analyzed by the standard external method. The linear range for the standard curve of MBF ranged from 0.05 to 10 μg/ml in serum was detected by HPLC, and the linear regression, curve recovery and coefficient of variation were calculated. The recovery ratios were the specific values of calculated peak area in serum to standard with different drug concentrations (0.05, 0.1, 0.5, 1, 5, 10 μg/ml). The LLOD was the lower detected concentration at the value of the signal to noise ratio (S/N) > 3 in serum. The LLOQ was the lower detected concentration at the value of S/N > 10 in serum.

Pharmacokinetics parameters were calculated from serum MBF concentrations with WinNonlin software (version 5.2.1, Pharsight Corporation, United States). The appropriate PK models were selected based on the MBF concentrations plotted on semi-logarithmic graphs. Moreover, the same models for PK analysis were selected after i.v and i.m administrations.

For PK-PD indices selection for an antibiotic, these three indices (fAUC/MIC, fC_max/MIC and fT > MIC) are standardized (Nielsen and Friberg, 2013). AUC is the area under the concentration-time curve, C_max is the highest concentration reached (the peak), and T > MIC is the cumulative percentage of a 24 h period that the concentration is above MIC, f is the free and unbound fraction of the drug. The best PK-PD index for a certain drug-bacteria combination is determined by plotting the value of an efficacy endpoint (typically log10 CFU/ml after 24 h of treatment) versus the magnitude of each of the three PK-PD indices. It has been reported that AUC/MIC could be the right and optimal PK-PD index for MBF based on the previously published study (Sidhu et al., 2010; Jian et al., 2015; Lei et al., 2017a; Zeng et al., 2017). The best PK-PD index of AUC/MIC was selected to use in this study. The applicable model equation for this study was shown as (Equation 1):

E is the summary PD endpoint, C is one of the three PK-PD indices as defined above, E_max is the maximum effect in Lg change after 24h compared to the initial inoculums, EC₅₀ is the magnitude of C that is needed to achieve 50% of E_max, N is the sigmoidicity factor. The figure of illustration relationship between E and C is provided, and the Y-axis means the effect of the antimicrobial agent measured as the log10 difference in bacterial number before and after 24 h incubation, and the X-axis means one of the three PK-PD indices as defined above.

To assume an optimal regime, the following formula was used:

Where AUC/MIC represented the targeted endpoint for optimal effect in hours, MIC₉₀ represented the MIC value containing 90% isolates in the population of strains, CL represented clearance ration, fu represented the free drug in serum, and F represented the value of bioavailability.

The probabilities of distribution of predicted daily dosages were conducted for 100000 trails to achieve 50 and 90 % target attainment rates (TAR) for bacteriostatic, bactericidal and bacterial elimination effects by using Monte Carlo Simulations in Oracle Ball (Oracle Corporation, Redwood Shores, CA, United States).

Data were shown as mean ± SD and calculated from 6 to 8 independent experiments. The MIC₅₀ and MIC₉₀ were obtained by using SPSS software. The statistical analyses were done by the Student’s t-test and ANOVA using Prism software (Graphpad Software Inc., London, United Kingdom). P-values < 0.05 indicated statistically significant differences.

The MIC distribution and values of MBF against the 134 SS strains were shown in Figure 1. The MIC values ranged from 0.03125 to 8 μg/ml. The MIC₅₀ and MIC₉₀ of these strains were calculated to be 1 and 2 μg/ml respectively, suggesting MBF displayed a potent antibacterial effect against SS. According to the MIC₉₀ value and the result of serotype identification (not shown), we selected the HB2 serotype, with similar MIC and MIC₉₀ values, to study the in vitro and ex vivo antimicrobial effect of MBF.

The HB2 with the MIC close to the MIC₉₀ was chosen as the representative isolate for PD study in vitro and ex vivo. The MIC and MBC of HB2 were determined to be 1 and 4 μg/ml in TSB, and 2 and 4 μg/ml in pig serum, respectively. The MBC/MIC ratios were 4 and 2 in vitro and ex vivo, respectively. Moreover, the MPC of MBF against HB2 was 2.56 μg/ml. These result revealed that MBF might have a relatively gentle concentration-effect against SS.

The growth-time curves of HB2 in TSB and serum are illustrated in Figure 2. According to the growth-time curve profiles, the logarithmic phases of HB2 in TSB and serum were from 2–10 h and 2–12 h, respectively whose bacteria amount (OD values) were increased logarithmically. The total bacterial amount and growth rate were higher in TSB than in serum. The findings suggest that serum can inhibit bacterial growth.

The killing-time curves of MBF against HB2 in vitro and ex vivo are illustrated in Figures 3A,B. According to the profiles of curves in vitro and ex vivo, MBF displays a concentration-dependent bactericidal activity against SS as increasing drug concentrations induced more rapid and radical bactericidal effect. Moreover, both the in vitro and ex vivo curves show the same bactericidal tendency; after exposure to 1 MIC or less of MBF, the bacteria could recover growth after 12 h. However, after exposure to a greater concentration than 1 MIC of MBF for 24 h, the bacterial CFU values were markedly reduced (<30 CFU) (Figure 3A). The bacterial CFU values were also markedly reduced (<30 CFU) in the serum from pigs of the PK experiment whose MBF concentration was above 2 MIC (Figure 3B). The killing-time curves in vitro and ex vivo were analogical. These findings suggest that MBF has a concentration-dependent action against SS both in vitro and ex vivo.

The specificity of MBF detection methods in plasma was good and suitable and there was no any interference observed in the chromatograms. Moreover, the linear range of the standard curve of MBF ranged from 0.02 to 10 μg/ml were shown in Figure 4. The coefficient of determination (R²) of standard curves from 0.05 to 10 μg/ml was 0.9999, and the inter-day and intra-day coefficient variation were <10% (Figure 4). Moreover, the recovery ratios values were in the range of 90 ± 2.23 to 94 ± 3.12 % in serum for MBF detection. The lower limit of detectability (LLOD) and lower limit of quantitation (LLOQ) were 0.02 and 0.05 μg/ml in plasma (Figure 4). In addition, the typical regression equation was y = 51.65 x – 0.087, R² = 0.9999 in serum, and y = 57.89 x + 0.132, R² = 0.9999 in standard (Figure 4).

For the pooled serum samples of the 8 pigs of the PK experiment, the percentage free drug concentration was 92 ± 4.3%. The mean MBF concentration-time profiles were exhibited in Figure 5 following oral and i.v administrations of MBF. In addition, the principal PK parameters were exhibited in Table 1 with using non-compartment models analysis after i.v and i.m administrations. The C_max, AUC_{0-24 h}, AUC_0-∞, Ke, t_1/2ke, T_max, Cl_b and MRT values were 6.28 μg/ml, 60.30, 88.94 μg.h/ml, 0.056 h^-1, 12.48 h, 0.75 h, 0.104 L/h and 16.27 h after oral administration, and 66.55, 94.40 μg.h/ml, 0.046 h^-1, 15.16 h, 0.085 L/h and18.93 h after i.v administration, respectively (Table 1). Furthermore, the bioavailability of MBF was calculated to be 94.21% after a single 8 mg/kg oral administration.

The PK-PD parameters of AUC_0-24h/MIC, AUC_0-24h/MPC, C_max/MIC and C_max/MPC received from PK and PD data in serum were exhibited in Table 2. Moreover, the estimated AUC_0-24h/MIC, AUC_0-24h/MPC, C_max/MIC, and C_max/MPC values of MBF against SS were 30.06, 23.47 h, 3.14 and 2.45, respectively (Table 2). Moreover, the time values for the MBF higher than the ex vivo MIC (T > MIC) and MPC (T > MPC) in serum were 11.02 and 7.56 h (Table 2). These results revealed that MBF could have high concentrations (above three times to MIC) and long antibacterial effect (over 24 h), and also reach the high concentration above MPC for 7 h. These indicated MBF had a strong antibacterial manner against SS in vitro and ex vivo.

Based on the previously described reports for quinolones and the profiles of the killing-time curve of MBF against SS, it could be obviously to estimate MBF had a concentration-dependent action. Thus, the AUC_{0-24 h}/MIC or C_max/MIC could be selected as PK-PD index. PK-PD index approach had become the gold standard for evaluation PK-PD of antibiotics and to guide the establishment of dosing regimens. These three indexes were highly correlated, making it important to have data from several different dosing regimens to be able to distinguish between them. According to the previously published study, it has been demonstrated that AUC/MIC could be regarded as the right and optimal PK-PD index for MBF (Sidhu et al., 2010; Jian et al., 2015; Lei et al., 2017a; Zeng et al., 2017). Therefore, the best PK-PD index of AUC/MIC was selected to use for proposing the optimal dose in this study. The antibacterial activity of MBF against HB2 was tested in the serum of those pigs used from the PK experiment (at 0.25–24 h after oral administration of 8 mg/kg MBF). Based on the inhibitory sigmoid E_max model, the relationship between antimicrobial efficacy and the ex vivo AUC_{0-24 h}/MIC ratios were simulated with an optimal R² (0.989) compared with others parameters, and the calculated model parameters are shown in Figure 6 and Table 3. The AUC_{0-24 h}/MIC ratios were calculated for bacteriostatic effect (E = 0, 25.23 h), bactericidal effect (E = -3, 35.64 h), and bacterial elimination effect (E = -4, 39.71 h) (Table 3).

Using the data generated here, we estimated the optimal once-daily doses of MBF (Table 4). The distribution of predicted daily population dosage (AUC_0-24h/MIC) values of MBF curing SS for 50 and 90% target were calculated and observed in Table 4 and Figure 7. Based on dose equations and Monte Carlo Simulations, we predicted the MBF dosages needed for bacteriostatic, bactericidal, and elimination activity curing SS HB2 as 5.88, 8.34 and 9.36 mg/kg.bw for 50% target, and 8.16, 11.31 and 12.35 mg/kg.bw for 90% target, respectively in Table 4.