Dicloxacillin and Flucloxacillin Inhibit Hepatic Uptake Transporters—In Vitro Investigations and Physiologically Based Pharmacokinetic Modeling
Noora Sjöstedt, Ogochukwu U. Amaeze, Jeroen J. M. W. van den Heuvel, Tore B. Stage, Jan B. Koenderink, Nina Isoherranen, Heidi Kidron, Erkka Järvinen
TL;DR
This study shows that dicloxacillin and flucloxacillin can block liver and kidney transporters, which may affect drug interactions.
Contribution
The study provides new in vitro and PBPK modeling evidence of transporter inhibition by dicloxacillin and flucloxacillin.
Findings
Dicloxacillin and flucloxacillin inhibit OATP1B1, OATP1B3, and BCRP with IC50 values in the micromolar range.
PBPK models predict limited clinical impact on rosuvastatin and P-gp substrate pharmacokinetics when co-administered with these antibiotics.
Both antibiotics are transported by OATPs and OATs, suggesting potential for hepatic and renal uptake.
Abstract
Dicloxacillin and flucloxacillin are β‐lactamase‐resistant penicillin antibiotics that have been in clinical use for over 50 years. While both antibiotics are known to induce cytochrome P450 enzymes, there is limited information available regarding their interactions with drug transporters. Here, we investigated the in vitro transport and inhibition of hepatic organic anion transporting polypeptides (OATPs) and renal organic anion transporters (OATs) by these antibiotics in recombinant transporter overexpressing HEK293 cells. We also investigated the transport of these antibiotics by efflux transporters, as well as their inhibition of breast cancer resistance protein (BCRP) and P‐glycoprotein (P‐gp) using a HEK293 membrane vesicle transport assay. Dicloxacillin and flucloxacillin inhibited rosuvastatin transport by OATP1B1, OATP1B3, and OATP2B1, and the inhibition was strongest for…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5| Transporter | Dicloxacillin | Flucloxacillin | ||
|---|---|---|---|---|
| IC50 (95% CI), μM | Calculated | IC50 (95% CI), μM | Calculated | |
| OATP1B1 | 3.86 (2.61–5.64) |
| 30.7 (14.4‐69.8) |
|
| OATP1B3 | 6.68 (3.44‐13.34) |
| 20.7 (12.5‐35.0) |
|
| OATP2B1 | 35.5 (24.4‐51.7) |
| 64.2 (48.6‐84.5) |
|
| OAT3 | 19.5 (17.3‐21.8) | 0.07 | 26.6 (23.1‐30.6) |
|
| OAT4 | 7.23 (6.18‐8.45) |
| 32.7 (28.1‐38.0) |
|
| BCRP | 166 (97.6‐274) |
| 379 (227‐618) |
|
| P‐gp | 258 (173‐374) |
| ≥ 833 |
|
| Dicloxacillin | Flucloxacillin | |||
|---|---|---|---|---|
| Rosuvastatin | Dabigatran etexilate | Digoxin | Rosuvastatin | |
| AUC ratio | 0.98 (0.97–1.00) | 1.08 (1.07–1.08) | 1.02 (1.01–1.02) | 0.94 (0.93–0.94) |
| Cmax ratio | 1.10 (1.06–1.13) | 1.09 (1.09–1.10) | 1.03 (1.02–1.03) | 0.94 (0.92–0.95) |
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsDrug Transport and Resistance Mechanisms · Amino Acid Enzymes and Metabolism · Pharmacogenetics and Drug Metabolism
Introduction
1
Dicloxacillin and flucloxacillin are β‐lactamase resistant penicillin antibiotics used to treat staphylococcal infections, such as skin and soft tissue infections, but also endocarditis [1]. In Europe, flucloxacillin is among the top four antibiotics consumed [2]. Pharmacokinetics of both antibiotics are characterized by rapid clearance, high plasma protein binding and excretion via urine [3, 4, 5, 6, 7, 8, 9]. Because of the short elimination half‐lives of dicloxacillin and flucloxacillin, they are administered three to four times per day [1, 8, 10].
Dicloxacillin and flucloxacillin induce cytochrome P450 enzymes (CYPs) in vitro and in vivo in humans [8, 10, 11, 12]. Moreover, the reduction in warfarin efficacy when dicloxacillin or flucloxacillin therapy is initiated suggests a pharmacokinetic interaction between these antibiotics and warfarin [11]. Dicloxacillin weakly induces intestinal P‐glycoprotein (P‐gp) in humans and in vitro, but this does not translate to decreased efficacy of direct oral anticoagulants that are P‐gp substrates [13]. Only a few drugs are known to affect the pharmacokinetics of dicloxacillin and flucloxacillin. Probenecid slightly increases the plasma concentration of dicloxacillin and flucloxacillin, while cyclosporine increases and rifampicin slightly decreases the plasma concentration of dicloxacillin [6, 7, 9].
Drug transporters involved in the disposition of dicloxacillin and flucloxacillin are poorly characterized. As negatively charged carboxylic acids with molecular weights of 450–470 Da, dicloxacillin and flucloxacillin are likely to interact with hepatic organic anion transporting polypeptides (OATPs) and renal organic anion transporters (OATs) [14, 15]. More importantly, in vitro reports suggest that these drugs may strongly inhibit human OATP1B1 and OATP1B3, and dicloxacillin may inhibit OAT3 [16, 17]. Similarly, knowledge on efflux transporters involved in the transport of dicloxacillin and flucloxacillin or efflux transporters inhibited by these drugs is limited. Dicloxacillin was identified as a P‐gp substrate in a cell assay, while it did not inhibit multidrug resistance‐associated proteins (MRP) 2, 3, or 4 in membrane vesicle assays [18, 19]. On the other hand, flucloxacillin did not inhibit breast cancer resistance protein (BCRP) or MRP2 in a cell assay [20].
In this study, we comprehensively characterized the in vitro transport and inhibition of human OATP1B1, OATP1B3, OATP2B1 and OAT1–OAT4 by dicloxacillin and flucloxacillin. We also investigated their transport by human efflux transporters and studied the inhibition of intestinal efflux transporters BCRP and P‐gp. Finally, we developed physiologically based pharmacokinetic (PBPK) models for both antibiotics to translate the in vitro inhibition data into human drug–drug interaction (DDI) predictions.
Methods
2
Chemicals and Reagents
2.1
Flucloxacillin and fluorescein sodium salt, dicloxacillin sodium salt monohydrate, sodium bicarbonate, sodium butyrate, HEPES, and analytical grade solvents methanol, acetonitrile, phosphoric acid and formic acid were acquired from Sigma‐Aldrich (St. Louis, MO, USA). Rosuvastatin calcium salt and rosuvastin‐d_6_ sodium salt were from Toronto Research Chemicals (North York, Ontario, Canada). N‐methyl‐quinidine was from Solvo Biotechnology (Szeged, Hungary). Tritium‐labeled methotrexate disodium salt was from Moravek (Brea, CA, USA). Dulbecco's Modified Eagle Medium (DMEM) supplemented with high glucose, GlutaMax, HEPES, phenol red and without sodium pyruvate, referred to hereinafter as DMEM, Hank's Balanced Salt Solution with calcium, magnesium, glucose and without sodium bicarbonate and phenol red, poly‐D‐lysine solution, fetal bovine serum, Nunc cell‐culture treated 48‐well plates, and Corning BioCoat poly‐D‐lysine 24‐well plates were from Thermo Fisher Scientific (Waltham, MA, USA).
Transient Expression of Transporters in HEK293 Cells
2.2
The human transporters OATP1B1, OATP1B3, OATP2B1, and OAT1‐OAT4 and a negative control protein (enhanced yellow fluorescent protein, eYFP) were transiently expressed in human embryonic kidney 293 (HEK293) cells by recombinant baculovirus‐mediated transduction. Baculoviruses carrying human genes for the transporters were prepared as reported previously [21, 22, 23]. HEK293 cells were cultured in DMEM containing 10% fetal bovine serum at 37°C and 5% CO_2_ and passaged two to three times per week. For the transport studies, 0.05 × 10^6^ cells for 48‐well plates (OATPs) or 0.06 × 10^6^ cells for 24‐well plates (OATs) were seeded onto cell‐culture treated well plates coated with poly‐D‐lysine. After 1 day, cells were transduced to produce recombinant proteins by changing media to DMEM containing 30 μL (48‐well plates) or 60 μL (24‐well plates) of recombinant baculovirus and 3 mM (OATs) or 5 mM (OATPs) sodium butyrate in a total media volume of 500 μL. The cells were cultured for a further 2 days before the transport assays.
Transport Assays in HEK293 Cells Overexpressing OATP and OAT Transporters
2.3
For the transport studies, media were removed and cells were washed once with warm transport uptake buffer (TP‐buffer—pH 7.4 HBSS with 25 mM HEPES and 4 mM sodium bicarbonate). Each experiment included triplicate (OATP) or duplicate (OAT) samples for every test condition. The uptake was started by removing liquid from wells and applying warm TP‐buffer (200 μL for 24‐well plates or 125 μL for 48‐well plates) containing a test compound to cells. Transport assays were terminated after a pre‐specified incubation time (reported in figure legends) by removing substrate solution and washing cells three times with cold TP‐buffer (300 μL for 48‐well plates and 400 μL for 24‐well plates). For the liquid chromatography mass spectrometry (LC–MS) analysis, compounds were extracted from cells with 125 μL (48‐well plates) or 250 μL (24‐well plates) of an ultrapure‐water solution containing 75% methanol and 0.1% formic acid, shaking at 250 RPM for 30 min. For the LC–MS analysis, internal standards (dicloxacillin for flucloxacillin, flucloxacillin for dicloxacillin, and rosuvastatin‐d_6_ for rosuvastatin) were spiked in the extraction solution before applying it to cells. Samples were transferred to well plates, centrifuged at 3000 g for 30 min, and transferred to a new well plate for the analysis. All substrate and inhibitor stocks were prepared in dimethyl sulfoxide (DMSO), and the DMSO concentration did not exceed 0.3% in the transport assays.
Inhibition Assays in HEK293 Cells Overexpressing OATP and OAT Transporters
2.4
In the OATP inhibition assays, inhibitors in warm TP‐buffer were directly applied to cells after media removal, followed by a 30‐min preincubation step in the presence of inhibitors. No preincubation step was included in the OAT inhibition assays. The inhibition assays were conducted as described above, except that inhibitors were included in the substrate solution. The rosuvastatin concentration was 2.5 μM in the OATP inhibition assays, and the transport assays were conducted for 2 min [24]. For the inhibition studies of OAT1 and OAT3, 5 μM fluorescein was assayed within 5 min, while 34 nM methotrexate within a 10 min assay was employed as the substrate for OAT2 and OAT4 [25, 26, 27, 28]. OAT1 and OAT3 cells were lysed with 200 μL of 1 M sodium hydroxide for 10 min, and 150 μL of these samples were transferred to a black well plate for fluorescein quantification. OAT2 and OAT4 cells were lysed with 200 μL of 0.1% Triton‐X100 for 10 min, and 150 μL of the samples were mixed with 4 mL of scintillation cocktail (Opti‐Fluor from PerkinElmer, Waltham, MA, USA). Each inhibition assay included a vehicle control without an inhibitor.
Efflux Transport and Inhibition Assays in HEK293 Membrane Vesicles
2.5
Membrane vesicles, 5 μL (7.5 μg of total protein), prepared as previously reported [23], were aliquoted to wells of a conical‐shaped 96‐well plate. The plate was incubated for 10 min at +37°C before the initiation of transport by adding 25 μL of TS‐buffer (10 mM Tris, 250 mM sucrose, pH adjusted to 7.4 with HEPES) containing magnesium chloride (10 mM final in the reaction), adenosine triphosphate (ATP) or adenosine monophosphate (AMP), 4 mM final in the reaction, and the transporter substrate and inhibitor. Each experiment included triplicate samples for both ATP and AMP samples. The concentrations of rosuvastatin and N‐methyl‐quinidine were 1 μM in the BCRP and P‐gp vesicle inhibition assays, respectively [29, 30]. The transport reaction times were 1 and 2 min for BCRP and P‐gp, respectively [29, 30]. The vehicle concentration (DMSO) was 1.1% in the efflux transporter inhibition assays. The transport reactions were quenched by adding 200 μL of cold TS‐buffer to each well and transferring the samples to a 96‐multiwell 1.0 μm glass fiber filter plate (Merck Millipore, Darmstadt, Germany) kept under vacuum. Wells were further washed four times with cold TS‐buffer, after which the plate was dried under vacuum. The compounds were eluted from the plate by applying 100 μL of 75% methanol containing 0.1% formic acid to each well and incubating at 250 RPM for 30 min before centrifuging the samples to an analysis plate at 2800 g for 2 min. For the rosuvastatin samples, the elution solvent contained rosuvastin‐d_6_ as the internal standard.
Analytical Methods
2.6
An LC–MS system consisting of Acquity I class UPLC connected to triple quadrupole Xevo TQ‐S mass spectrometer equipped with an electrospray ionization source, from Waters (Milford, MA, USA), was employed to quantify dicloxacillin, flucloxacillin, and rosuvastatin. N‐methyl‐quinidine was quantified with Agilent Technologies (Santra Clara, CA, USA) 1100 HPLC system equipped with a fluorescence detector. Each batch of LC–MS runs included a standard curve, and quantification was based on the analyte to internal standard peak area ratio plotted against the analyte concentration. An external standard curve calibration was employed for N‐methyl‐quinidine. Details of LC–MS and HPLC methods are reported in Supplementary Methods. Fluorescein was quantified with a fluorescence plate reader (Victor X3) by employing a standard curve and wavelengths 485 nm (excitation) and 535 nm (emission). An automatic liquid scintillation counter (Hidex 600 SL from Hidex, Turku Finland) was employed for the quantification of tritium‐labeled methotrexate.
PBPK Modeling
2.7
Full‐PBPK models for dicloxacillin and flucloxacillin were developed in Simcyp version 23 (Certara UK Limited, Sheffield, UK). Further details of the models are reported in Supporting Methods and Results (Tables S2 and S3, and Appendix S1 and S2). Published clinical pharmacokinetic data were divided into two datasets, one for model development and the other for model verification (Tables S4 and S5). For the DDI simulations, Simcyp built‐in library models for rosuvastatin, dabigatran etexilate, and digoxin were employed.
Data Analysis and Reporting
2.8
For the inhibition data analysis, the average of measured concentrations in the control cells was subtracted from the average of measured concentrations in transporter‐expressing cells for each inhibitor concentration and then normalized to the vehicle sample that represents 100% transport activity. Equation (1) was fit to the inhibition data. In Equation (1), Inhibition represents the percent inhibition of transporter‐mediated uptake at an inhibitor concentration of I and IC_50_ is the half‐maximal inhibitory concentration.
Equation (2) (the Michaelis–Menten equation) was fit to the transport kinetic assay data. In Equation (2), Uptake is the rate of uptake in transporter‐expressing cells from which the uptake in the control cells is subtracted at the concentration S of dicloxacillin or flucloxacillin, V max is the maximum rate of transport, and K m is the Michaelis constant.
The nonlinear least squares method (nls) in the R programming language environment (R version 4.3.1) was employed for fitting Equations (1) and (2) to the data.
To evaluate potential clinical DDIs for transporters, the FDA R values were calculated according to the FDA In Vitro Drug Interaction Studies guidance [31]. Equation (3) was used to estimate the maximum unbound plasma inhibitor concentration at liver inlet (I in,max) for OATPs, where f u,p is the unbound fraction in plasma, C max is the maximal total concentration in plasma, F a is the fraction absorbed after oral administration, F g is the fraction available after intestinal metabolism, k a is the first‐order absorption rate constant in vivo, dose is the oral dose, Q h is the hepatic blood flow, and R B is the blood‐to‐plasma concentration ratio.
The values of F a × F g and k a were set to 1 and 0.1/min to present a worst‐case estimation according to FDA's recommendation, since no experimental bioavailability data is available for dicloxacillin or flucloxacillin [31]. Hepatic blood flow (Q h) was set to 1.620 L/min [31]. A typical oral dose for dicloxacillin and flucloxacillin is 1000 mg. Experimental R B is 0.65 for flucloxacillin and was assumed to be the same for dicloxacillin [32]. The unbound fraction in plasma for dicloxacillin and flucloxacillin was set to 0.02 and 0.06, respectively [4]. The total C max values for dicloxacillin and flucloxacillin at a 1000 mg dose are reported as 67 μM and 46 μM, respectively [6, 8]. The resulting calculated I n,max,u values for dicloxacillin and flucloxacillin were 5.51 μM and 15.7 μM, respectively. For OATs, the maximal unbound plasma concentration of the interacting drug at steady state (I max,u) [31] was calculated by C max × f u resulting in 1.34 and 2.76 μM for dicloxacillin and flucloxacillin, respectively. The intestinal luminal concentration (I_gut_) [31] was estimated by dose/250 mL, resulting in 8505 and 8813 μM for dicloxacillin and flucloxacillin, respectively.
Results
3
Transporter Inhibition Studies
3.1
Dicloxacillin and flucloxacillin inhibited rosuvastatin transport into HEK293 cells expressing the human hepatic OATPs (Figure 1). We pre‐incubated the cells for 30 min in the presence of inhibitors, since this procedure is currently recommended for OATPs to fully reveal the strength of inhibition [31, 33, 34]. The IC_50_ values were lower for dicloxacillin than flucloxacillin (Table 1). We replicated the inhibition experiment by employing fluorescent substrates for OATPs (Figure S1), and the obtained IC_50_ values (Table S6) were comparable to the experiments with rosuvastatin as the OATP substrate (Table 1). We also tested whether the inhibition of OATP1B1 by dicloxacillin and flucloxacillin is reversible (Figure S2). We were able to wash out most of the inhibition, which indicates that the inhibition of OATP1B1 is reversible and is not caused by covalent binding of dicloxacillin and flucloxacillin to OATPs, as reported previously for these compounds with human serum albumin [35]. Moreover, we determined the K_i_‐value for the inhibition of OATP1B1 by dicloxacillin, since this was the strongest observed inhibition. The experimental K_i_‐value for the competitive inhibition of OATP1B1 by dicloxacillin was 3.65 μM (2.33–5.71, 95% CI), which is close to the experimental IC_50_‐value (Figure S3; Table 1).
In vitro inhibition of OATPs by dicloxacillin (A) and flucloxacillin (B). Rosuvastatin transport into OATP1B1‐, OATP1B3‐, OATP2B1‐ and eYFP‐ (control) expressing HEK293 cells was studied in the presence of seven different dicloxacillin or flucloxacillin concentrations and in the absence of inhibitors (vehicle) for 2 min. Rosuvastatin uptake into control cells was subtracted from the uptake into OATP expressing cells for each inhibitor concentration, and the results are presented relative to the uptake in the vehicle groups that were set to 100% transport. Solid lines show the sigmoidal fittings that were used to calculate the half‐maximal inhibitory concentrations (IC50 values are presented in Table 1). Data points represent mean values of triplicate samples in two independent experiments.
Dicloxacillin and flucloxacillin inhibited < 50% of fluorescein transport into OAT1‐expressing HEK293 cells at 200 μM (Figure 2). Similarly, neither compound substantially inhibited OAT2‐mediated transport of methotrexate at 200 μM. In contrast, both compounds strongly inhibited fluorescein transport into OAT3‐expressing cells and methotrexate transport into OAT4‐expressing cells, resulting in similar IC_50_‐values for OAT3 and OAT4 as for the OATPs (Table 1). A preincubation step was omitted for the OAT inhibition assays since, currently, preincubation has not been reported to potentiate the inhibition of OATs similar to OATPs [34]. However, we wanted to further investigate whether preincubation potentiates the inhibition of OAT3 by dicloxacillin and flucloxacillin. An assay with 30 min preincubation resulted in the same IC_50_ values (Figure S4) as without the preincubation step (Table 1).
In vitro inhibition of OATs by dicloxacillin (A) and flucloxacillin (B). 5 μM fluorescein transport into OAT1, OAT3 and eYFP (control), and 34 nM methotrexate transport into OAT2, OAT4 and eYFP (control) expressing HEK293 cells were studied in the presence of 20 and 200 μM dicloxacillin and flucloxacillin (left side plots) for 5 min (fluorescein) or 10 min (methotrexate). Concentration‐dependent inhibition for OAT3 and OAT4 was derived from similar experiments with five different dicloxacillin or flucloxacillin concentrations, including a vehicle sample. For all plots, the uptake into control cells was subtracted from the uptake into OAT‐expressing cells for each inhibitor concentration, and the results are presented relative to the uptake in the vehicle groups that were set to 100% transport. Solid lines show the sigmoidal fittings that were used to calculate the half‐maximal inhibitory concentrations (IC50 values presented in Table 1). Data points represent mean values of duplicate samples in three independent experiments. Bars represent mean values of duplicate samples (presented as points) in an experiment.
Regarding the studied efflux inhibition, dicloxacillin and flucloxacillin inhibited BCRP at high concentrations in a concentration‐dependent manner (Figure 3A). For P‐gp, the highest concentration of dicloxacillin almost completely inhibited the transport, while flucloxacillin reduced P‐gp activity by 65% at the highest tested concentration of 2100 μM (Figure 3B). The calculated IC_50_ values were within the range of 166–379 μM (Table 1).
In vitro inhibition of BCRP (A) and P‐gp (B) by dicloxacillin and flucloxacillin. BCRP‐ or P‐gp‐containing membrane vesicles were incubated with 1 μM rosuvastatin (BCRP) or 1 μM N‐methyl‐quinidine (P‐gp) in the presence of five different concentrations of dicloxacillin and flucloxacillin. Rosuvastatin or N‐methyl‐quinidine transport in the presence of AMP was subtracted from the uptake in the presence of ATP for each inhibitor concentration, and the results are presented relative to the uptake in the vehicle groups that were set to 100% transport. Solid lines show the sigmoidal fittings that were used to calculate the half‐maximal inhibitory concentrations (IC50 values presented in Table 1). Data points represent mean values of triplicate samples in two independent experiments.
Transporter Uptake Studies
3.2
All OATPs and OATs transported dicloxacillin and flucloxacillin when assayed at 2 and 50 μM concentrations and 10 min transport (Figure 4A–D). Efflux transporters MRP2, MRP3, MRP4, BCRP, and P‐gp did not transport dicloxacillin or flucloxacillin in the membrane vesicle assay (Figure S5). We further investigated the time‐dependent transport by OATPs and OATs and found that the transport was linear up to 2 min (Figure S6). Because of the low transport activity of OAT1 and OAT2, we did not further characterize these transporters in the transport kinetic assays.
In vitro uptake of dicloxacillin and flucloxacillin into hepatic OATP and renal OAT expressing cells. Transport of 2 μM and 50 μM dicloxacillin and flucloxacillin into OATP1B1‐, OATP1B3‐, OATP2B1‐, OAT1‐, OAT2‐, OAT3‐, OAT4‐, and eYFP‐ (control) expressing HEK293 cells was investigated within 10 min incubation (panels A–D). The data points are from one experiment performed with triplicate samples. To study the kinetics for the transport (panels E–H), uptake of seven (OATs) or eight (OATPs) different dicloxacillin and flucloxacillin concentrations into OATP1B1, OATP1B3, OATP2B1, OAT3, OAT4, and eYFP (control) expressing HEK293 cells was studied within 2 min incubation. The uptake into control cells was subtracted from the uptake into transporter expressing cells for each concentration. Because of a substantial deviation from the Michaelis–Menten fittings (solid lines) at the higher concentrations, 75 μM and 250 μM (dicloxacillin, OATP1B1), 250 μM (flucloxacillin, OATP1B3), and 300 μM (dicloxacillin and OATs, flucloxacillin and OAT3) concentrations were excluded from the fittings and are shown with pale colors. Data are presented as mean values of triplicate samples in an experiment (OATPs) or as mean values of duplicate samples in two independent experiments (OATs) with error bars representing standard deviation.
Transport kinetics of dicloxacillin and flucloxacillin by OATP1B1, OATP1B3 and OATP2B1 followed Michaelis–Menten kinetics (Figure 4E,F; Table S7). Dicloxacillin and flucloxacillin had the highest affinity for OATP1Bs with K m‐values of 3 μM (dicloxacillin) and 10–18 μM (flucloxacillin). Similarly, the transport by OAT3 and OAT4 followed Michaelis–Menten kinetics (Figure 4G,H; Table S7) but with lower affinities in comparison to OATP1Bs with K m‐values between 22 and 53 μM (Table S7).
PBPK Modeling of Transporter Inhibition by Dicloxacillin and Flucloxacillin
3.3
To predict the clinical relevance of the observed in vitro inhibition, we calculated the risk for clinical DDIs based on the R‐values (Table 1) defined in the FDA in vitro guidance [31]. Both dicloxacillin and flucloxacillin showed a potential risk for clinical DDIs by inhibiting hepatic OATPs and intestinal BCRP, P‐gp, and OATP2B1 (Table 1). In addition, the FDA cut‐off value was exceeded for the inhibition of OAT4 by dicloxacillin and flucloxacillin, and for the inhibition of OAT3 by flucloxacillin (Table 1). The clinical relevance of OATP, BCRP, and P‐gp inhibition was further evaluated with PBPK modeling.
The developed PBPK models predicted the clinical pharmacokinetic profiles well for both compounds, with an absolute average fold error (Supporting Methods and Results) of 0.99 and 0.92 for dicloxacillin peak concentration (C max) and area under the curve (AUC) and 0.86 and 0.96 for flucloxacillin C max and AUC (Tables S8 and S9; Figures S7–S9). In addition, all simulated C max and AUC values were within the calculated drug and study‐specific acceptance ranges (Figure S10). Inhibition values derived from the in vitro studies (Table 1) were used as the model inputs for DDI simulations.
Simulations of oral dicloxacillin or flucloxacillin (1000 mg, q 8 h) administered simultaneously with 20 mg rosuvastatin (sensitive OATP and BCRP substrate) predicted only minor changes (≤ 10%) in rosuvastatin AUC or C max (Table 2; Figure 5B,D). Hepatic uptake of rosuvastatin was inhibited up to 27% by dicloxacillin (Figure S11A), as was the BCRP‐mediated apical efflux clearance in the jejunum (Figure S11C). OATP2B1‐mediated apical influx of rosuvastatin in the jejunum was predicted to be inhibited up to 62% by dicloxacillin (Figure S11C). On the other hand, flucloxacillin was predicted to have the largest effect on the intestinal apical uptake clearance of rosuvastatin with up to 43% inhibition (Figure S11D). In DDI simulations using sensitive P‐gp substrates, dabigatran etexilate (150 mg) or digoxin (0.5 mg) as object drugs and dicloxacillin as the precipitant, only minor increases in AUC and C max were predicted by P‐gp inhibition (Table 2; Figure 5A,C).
Simulated PK profiles for the object drugs in the DDI simulations in the presence and absence of dicloxacillin (panel A, C and D) or flucloxacillin (panel B). Solid black lines show the predicted mean systemic object drug concentrations in the absence of dicloxacillin or flucloxacillin, while dashed green lines represent the simulations when the object drugs are co‐administered together with dicloxacillin or flucloxacillin. The details of DDI simulations and predicted pharmacokinetic parameter changes are reported in Table 2.
To examine the effects of dicloxacillin and flucloxacillin on specific OATPs and sinusoidal hepatic uptake clearance in more detail, rosuvastatin DDI simulations were repeated with all the hepatic uptake clearance of rosuvastatin assigned to either OATP1B1, OATP1B3, or OATP2B1. Here, the strongest inhibition (up to 38%) was simulated for dicloxacillin when OATP1B1 accounted for all hepatic rosuvastatin uptake (Figure S11A). This simulation represents a case where a drug that is taken up into the liver solely by OATP1B1 and has low passive hepatic permeability (1%) is co‐administered with dicloxacillin. In this worst‐case scenario, the predicted rosuvastatin AUC was almost unchanged, but the C max ratio was 1.24‐fold. For flucloxacillin, all models predicted ≤ 11% changes in the hepatic uptake clearance of rosuvastatin (Figure S11B) and the AUC and C max ratios (0.95 and 0.96‐fold) were practically unchanged compared to the simulations with the original model (Table 2).
Discussion
4
Dicloxacillin and flucloxacillin have been in clinical use for over 50 years [5]. Recently, we reported that both antibiotics cause DDIs by inducing CYP enzymes [8, 10, 11, 36]. For many older drugs, comprehensive DDI investigations in vitro or in clinical trials were never conducted. Thus, it is unsurprising that we found dicloxacillin and flucloxacillin to inhibit several drug transporters in vitro. We assessed their in vivo DDI potential by employing static model calculations recommended by the FDA (R‐values) [31] and further by developing PBPK models to predict hepatic OATP inhibition and intestinal BCRP, P‐gp and OATP2B1 inhibition.
The purpose of static models with pre‐defined cut‐off values for different transporters is to provide a qualitative assessment of potential DDI risk with minimal false negatives [37, 38]. This may, however, result in a high number of false positive signals and higher cut‐off values than those recommended by the FDA have been suggested for P‐gp and BCRP inhibition [39, 40, 41]. Here, the static model predictions clearly indicated a potential DDI for OATP1Bs and BCRP, whereas the PBPK modeling predicted no significant impact on the systemic exposure of OATP1B and BCRP probe rosuvastatin. The static evaluation likely overestimates the hepatic inlet concentration, which was estimated assuming complete absorption as to represent a worst‐case scenario. Unlike R values, mechanistic static modeling, and particularly PBPK modeling, can provide quantitative estimates for the magnitude of possible DDIs. Importantly, as the clearance of drugs is rarely dependent on a single pathway, they allow the simultaneous consideration of multiple pathways and the effect of precipitant drug on each of these pathways [37, 38]. For example, our object drug for the PBPK modeling, rosuvastatin, is transported by multiple transporters in the liver and intestine [39]. Despite the inhibition of several transporters, the dicloxacillin and flucloxacillin simulations predicted only minor changes in drug exposure (Figure 5; Table 2). This may be explained by the short half‐lives of dicloxacillin and flucloxacillin resulting from rapid absorption and clearance as well as their limited accumulation even after multiple dosing (Figure S12).
A novel finding here was that dicloxacillin and flucloxacillin are efficiently transported by OATP1B1, OATP1B3, and OATP2B1. This is unsurprising, as both compounds are anionic carboxylic acids with molecular weights falling within the typical range of OATP substrates [14]. Still, renal excretion is the major clearance pathway for both antibiotics [3]. Beringer et al. found that cyclosporine and probenecid increased the C max and AUC of dicloxacillin slightly (< 2‐fold) [7]. Furthermore, probenecid decreased the renal clearance of dicloxacillin by 4‐fold. Cyclosporine is a strong inhibitor of OATP1B1 and OATP1B3 but does not inhibit OAT1 or OAT3, while probenecid is a strong inhibitor of OAT1 and OAT3 and a weak inhibitor of OATP1B1 and OATP1B3 [42, 43, 44, 45]. These in vivo DDI findings can be explained by our results that dicloxacillin is transported by OATP1B1, OATP1B3, and OAT3 and that the inhibition of these transporters affects the pharmacokinetics of dicloxacillin, in particular the renal clearance.
Dicloxacillin and flucloxacillin cause rare but severe idiosyncratic drug‐induced liver injury (DILI) in humans [46, 47]. The risk of DILI is well established particularly for flucloxacillin [47]. Variants in human leukocyte antigen genes have been associated with flucloxacillin DILI, indicating that the DILI is immune‐mediated [47]. Flucloxacillin forms covalent adducts with hepatic proteins in hepatocytes in vitro, which may contribute to rare immune reactions targeting the liver [48]. Our novel finding that flucloxacillin and dicloxacillin are good substrates for hepatic OATPs may indicate a role for these transporters in the hepatic accumulation of these antibiotics, although no direct genetic associations have been reported between hepatic transporters and DILI for these antibiotics [47, 49].
Our findings may have implications for other structurally related compounds, such as oxacillin and cloxacillin, which share similar chemical properties with dicloxacillin and flucloxacillin, differing only at two positions of their phenyl rings [5]. Therefore, it is likely that these two antibiotics may also inhibit human OATPs, OATs, and BCRP. Indeed, a study found that cloxacillin inhibits human OAT1, OAT3, and OAT4, and it is transported by OAT3 similarly to what we found here for dicloxacillin and flucloxacillin [50]. Another study reported similar inhibition of OAT3 by dicloxacillin, cloxacillin, and oxacillin [17]. Furthermore, cloxacillin inhibited OATP1B1 and OATP1B3 transport around 50% at 10 μM in vitro [16]. The potential transporter DDI liability of oxacillin and cloxacillin should be further investigated.
There are some limitations to our study. We were unable to incorporate the transporter‐mediated distribution of dicloxacillin and flucloxacillin into our PBPK models due to the lack of in vitro transporter expression values for scaling and sufficient clinical data for verifying transport pathways. Inclusion of transport may help to better predict intracellular drug exposure, which could have implications for predicting efflux inhibition and liver accumulation. Furthermore, we focused here on short‐term inhibition by dicloxacillin and flucloxacillin and, therefore, did not include the previously observed weak P‐gp induction [13] in our dicloxacillin model. However, based on our simulations with dicloxacillin and dabigatran etexilate, induction is expected to have the dominant effect in a clinical scenario, as inhibition without induction was predicted to cause < 10% change in dabigatran exposure (Table 2).
In conclusion, dicloxacillin and flucloxacillin inhibit human OATP1B1, OATP1B3, OATP2B1, OAT3, OAT4, BCRP, and P‐gp in vitro. Static modeling indicated a possible risk for DDIs caused by the inhibition of these transporters, but PBPK simulations did not predict clinically relevant DDIs between dicloxacillin or flucloxacillin and the OATP and BCRP substrate rosuvastatin. These findings may be further validated in a clinical study in humans using an appropriate probe substrate or endogenous transporter biomarkers.
Author Contributions
Erkka Järvinen, Noora Sjöstedt, Ogochukwu U. Amaeze, Tore B. Stage, Nina Isoherranen, Jan B. Koenderink, and Heidi Kidron wrote the manuscript. Erkka Järvinen, Noora Sjöstedt, Ogochukwu U. Amaeze, Tore B. Stage, Nina Isoherranen, Jan B. Koenderink, and Heidi Kidron designed the research. Erkka Järvinen, Noora Sjöstedt, Ogochukwu U. Amaeze, and Jeroen J.M.W. van den Heuvel performed the research. Erkka Järvinen and Noora Sjöstedt analyzed the data.
Funding
This study was supported by the Finnish Cultural Foundation, the Emil Aaltonen Foundation, and the Research Council of Finland (grant no: 332949). Ogochukwu Amaeze and Nina Isoherranen were supported by the Bill and Melinda Gates Foundation grant INV‐029091. Nina Isoherranen was supported in part by a grant from the NIH R01GM147947.
Conflicts of Interest
T.B.S. has given paid lectures for Pfizer and Eisai, consulted for Pfizer, been an expert witness for Sandoz A/S, and collaborated with Novo Nordisk A/S, all unrelated to the work reported in the present article. N.I. reports consultancy agreements with Merck and Boehringer‐Ingelheim, and honoraria from ASPET, Elsevier, and the US National Institutes of Health. All other authors declare no conflicts of interest.
Supporting information
Appendix S1: cts70487‐sup‐0001‐AppendixS1.xlsx.
Appendix S2: cts70487‐sup‐0002‐AppendixS2.xlsx.
Supporting Methods and Results: cts70487‐sup‐0003‐DataS1.pdf.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1C. Rayner and W. J. Munckhof , “Antibiotics Currently Used in the Treatment of Infections Caused by Staphylococcus aureus ,” Internal Medicine Journal 35 Suppl 2 (2005): S 3–S 16.16271060 10.1111/j.1444-0903.2005.00976.x · doi ↗ · pubmed ↗
- 2R. Bruyndonckx , N. Adriaenssens , N. Hens , et al., “Consumption of Penicillins in the Community, European Union/European Economic Area, 1997‐2017,” Journal of Antimicrobial Chemotherapy 76 (2021): ii 14–ii 21.34312657 10.1093/jac/dkab 173PMC 8314108 · doi ↗ · pubmed ↗
- 3M. Barza and L. Weinstein , “Pharmacokinetics of the Penicillins in Man,” Clinical Pharmacokinetics 1 (1976): 297–308.797501 10.2165/00003088-197601040-00004 · doi ↗ · pubmed ↗
- 4H. H. Thijssen and J. Wolters , “The Metabolic Disposition of Flucloxacillin in Patients With Impaired Kidney Function,” European Journal of Clinical Pharmacology 22 (1982): 429–434.7117355 10.1007/BF 00542548 · doi ↗ · pubmed ↗
- 5R. Sutherland , E. A. Croydon , and G. N. Rolinson , “Flucloxacillin, a New Isoxazolyl Penicillin, Compared With Oxacillin, Cloxacillin, and Dicloxacillin,” BMJ (Clinical Research Ed.) 4 (1970): 455–460.10.1136/bmj.4.5733.455PMC 18200865481218 · doi ↗ · pubmed ↗
- 6W. S. Putnam , J. M. Woo , Y. Huang , and L. Z. Benet , “Effect of the MDR 1 C 3435 T Variant and P‐Glycoprotein Induction on Dicloxacillin Pharmacokinetics,” Journal of Clinical Pharmacology 45 (2005): 411–421.15778422 10.1177/0091270004273492 · doi ↗ · pubmed ↗
- 7P. M. Beringer , J. Kriengkauykiat , X. Zhang , et al., “Lack of Effect of P‐Glycoprotein Inhibition on Renal Clearance of Dicloxacillin in Patients With Cystic Fibrosis,” Pharmacotherapy 28 (2008): 883–894.18576903 10.1592/phco.28.7.883 · doi ↗ · pubmed ↗
- 8D. B. Iversen , A.‐C. D. Dunvald , D. M. Jespersen , et al., “Flucloxacillin is a Weak Inducer of CYP 3A 4 in Healthy Adults and 3D Spheroid of Primary Human Hepatocytes,” Clinical Pharmacology and Therapeutics 114 (2023): 434–445.37235733 10.1002/cpt.2959 · doi ↗ · pubmed ↗