Animals, reagents, and bacterial strains
Healthy lactating CD-1 mice (Charles River Laboratories, Beijing, China) weighing 30–45 g were maintained under specific pathogen-free environment in this study.
Rifaximin was obtained from Sigma Chemical Company, St. Louis, Missouri, USA. Dimethyl sulfoxide (DMSO) was used to dissolve the rifaximin standard at a concentration of 8,000 µg/ml. The working solution was prepared by diluting the stock solution to the appropriate concentration with DMSO.
Liquid chromatography-grade methanol, acetonitrile, and ammonium formate were obtained from Thermo Fisher Technology, China. The type 1 ultrapure water (18.2 MΩ.cm) was delivered by PALL Cascada Purification System, USA.
In 2017–2019, 45 clinical isolates were separately isolated from milk which was produced by 45 dairy cows with mastitis in Beijing. These isolates were identified by chromogenic medium, microscope, and PCR. The reference strain ATCC25922 was obtained from China Agricultural University.
Susceptibility tests
The minimum inhibitory concentrations (MICs) of rifaximin against 45 clinical E. coli strains were determined by broth microdilution method according to the recommendations of the Clinical and Laboratory Standards Institute [18]. E. coli ATCC25922 was used as control bacteria in all parallel experiments. The MIC50 and MIC90 values were calculated, which translate to inhibiting at least 50% and 90% the bacterial population, respectively.
The in vitro time-killing curve experiments of E. coli ATCC 25,922 were performed in two groups with an initial bacterial inoculum of 106 and 107 CFU/mL, respectively. These two groups of E. coli were exposed to rifaximin at concentrations of 0.5 × , 1 × , 2 × , 4 × , and 8 × MIC, respectively, and grew in a constant temperature incubator at 37 °C with a shaking at a speed of 200 rpm. The number of bacteria in each group were calculated by gradient dilution coating method. The MH agar plates were cultured in constant temperature incubator at 37 °C for 24 h before counting.
Pharmacokinetics
The mammary glands of the fourth pair are independent of each other and can be used as two independent samples for analysis. This method has been reported in our previous studies [17]. Four single-dose groups, with doses of 50, 100, 200, and 400 µg/gland, were used in the pharmacokinetics experiments. In all, 120 healthy lactating CD-1 mice were selected and randomly separated into four groups, with 30 mice in each group. The mice were anesthetized by intraperitoneal injection of pentobarbital and were injected with rifaximin using a micro-syringe into the gland under a dissecting microscope. The gland tissue samples were collected at 5 min, 10 min, 15 min, 30 min, 45 min, 1 h, 4 h, 8 h, 10 h, 12 h, and 24 h after administration. At each time point, six mammary gland tissues were collected. All mammary gland samples were processed, and rifaximin concentration in the mammary gland was detected by high-performance liquid chromatography (HPLC).
The processing method of mammary gland samples is similar to the method used in our previously published papers [17]. Briefly, mammary gland samples were homogenized, and 0.5 g of mammary gland samples were transferred into a 10-mL polypropylene centrifuge tube. Three milliliters of acetonitrile were also added into each centrifuge tube to extract rifaximin. The mixture was vortexed for 3 min and the samples were centrifuged at 7,104 × g for 5 min. The supernatant was transferred to another tube, and the remaining residue was re-extracted with 3 ml volume of acetonitrile. The two parts of supernatant were added to a solid phase extraction (SPE) cartridge (Oasis HLB 3 cc 60 mg, Waters Company, USA). Then, the SPE cartridge was eluted with 3 ml acetonitrile. The eluent was evaporated to dryness under a stream of N2 at 40 °C. Finally, the residue was reconstituted in 1 ml methanol and filtered through a 0.22-µm filter, and then was analyzed by HPLC.
A C18 reverse-phase column (Waters XBridge ShieldRP18 4.6 mm × 250 mm, 5.0 μm) was used to separate different substances. The mobile phase comprised methanol, acetonitrile, and ammonium formate (3.16 g/L, PH = 7.2) (V/V/V = 31.5:31.5:37). The flow rate of the mobile phase was 1.4 mL·min−1. The injection volume was 20 μl, and the detection wave length of UV was 276 nm. The limits of detection (LOD) and the limits of quantitation (LOQ) were determined by a known concentrations of rifaximin, whose lowest concentration met the requirement of a signal-to-noise ratio of ≥ 3 and ≥ 10, respectively. The accuracy and precision were evaluated by adding a known concentrations of rifaximin (10, 100, or 200 μg/g) to the blank mammary gland in five replicates over five consecutive days. The recoveries and relative standard deviations (RSD) were obtained to judge the accuracy and precision of the method, respectively. The recovery of rifaximin was calculated by rifaximin recovered from gland tissue dividing by the known concentration of the rifaximin standard. The precision was calculated based on the formula as RSD (%) = [SD/M] × 100%, where SD is the standard deviation and M is the average concentration of replicates.
Pharmacodynamics
The mouse mastitis model caused by E. coli was based on previous reports [16, 17, 19]. In brief, the offspring were removed from lactating mice at 2 h before the experiment. The selected mice were anaesthetized with pentobarbital. Then, a small cut was made at the far part of the fourth pair of mammary glands in the mouse abdomen, and 50 µl of 106 CFU/ml of bacterial fluid was injected into the gland through the nick in the mammary gland duct under a stereomicroscope using a 33G micro-syringe.
The pharmacodynamics experiments were divided into 12 therapeutic regimens covering various doses ranging from 25 to 800 µg/gland, and two dosing intervals of 12 h and 24 h per 24-h experiment cycle. A total of 36 lactating mice were used in the pharmacodynamics experiments with three mice in each dosing group. We selected 3 mice with successfully established mastitis model as the control group, which were not treated with rifaximin. The fourth pair of glands from mastitis mice were injected with 100 µl rifaximin. After 24 h of treatment, the mice were euthanized with CO2 and dissected to obtain gland samples. Then, the fourth pair of mammary glands were homogenized with a tissue homogenizer and diluted by 0.9% normal saline appropriately. The tissue diluent was inoculated on MH agar plates to count the bacteria. The antibacterial effect of rifaximin is expressed as the decrease in the number of bacteria.
PK/PD analysis
The inhibitory effect sigmoid Emax model of WinNonlin software (version 8.3; Pharsight, USA) was used to analyze the relationship between the antibacterial effect of different doses of rifaximin in the infected mice and the PK/PD parameters (AUC/MIC, T > MIC, Cmax/MIC). According to the simulation equation, the PK/PD target values for bactericidal effects were obtained. The simulation equation is as follows:
$$E = {E}_{\mathrm{max}} - \frac{\left({E}_{\mathrm{max}} - {E}_{0}\right) \times {C}_{e}^{N}}{{EC}_{50}^{N} + {C}_{e}^{N}}.$$
Here, E is the antibacterial effect of rifaximin, which is the bacterial decrease value (△log10CFU/ gland) in the mastitis mice group after 24 h of treatment; Emax is the log10CFU/gland in the drug-free control sample; EC50 is the value of PK/PD index of drug when the drug produces 50% the maximum antibacterial effect; Ce is the parameters of PK/PD model (including T > MIC, AUC/MIC and Cmax/MIC); N is the Hill coefficient, which describes the steepness of the dose–response curve.
Monte Carlo simulation
The three kinds of clinical regimens of rifaximin in the treatment of mastitis cows were computed by Monte Carlo simulation in Crystal Ball Professional V7.2.2 (developed by the U.S. Oracle company). The pharmacokinetic parameters AUC of rifaximin in the milk sample were assumed to be logarithmically distributed with a mean value and standard deviation of 340.73 ± 43.968 h·μg/ml according to our previous PK study of rifaximin in dairy cows (The results have not been made public, which will be published in another article.). The MIC obeys the custom distribution according to the distribution probability of MIC in the clinic susceptibility tests. The PK/PD target values were obtained from the mice mastitis model experiment. Monte Carlo simulation was performed for 10,000 sessions.