Sex-, feeding-, and circadian time-dependency of P-glycoprotein expression and activity - implications for mechanistic pharmacokinetics modeling

P-glycoprotein (P-gp) largely influences the pharmacokinetics (PK) and toxicities of xenobiotics in a patient-specific manner so that personalized drug scheduling may lead to significant patient’s benefit. This systems pharmacology study investigated P-gp activity in mice according to organ, sex, feeding status, and circadian time. Sex-specific circadian changes were found in P-gp ileum mRNA and protein levels, circadian amplitudes being larger in females as compared to males. Plasma, ileum and liver concentrations of talinolol, a pure P-gp substrate, significantly differed according to sex, feeding and circadian timing. A physiologically-based PK model was designed to recapitulate these datasets. Estimated mesors (rhythm-adjusted mean) of ileum and hepatic P-gp activity were higher in males as compared to females. Circadian amplitudes were consistently higher in females and circadian maxima varied by up to 10 h with respect to sex. Fasting increased P-gp activity mesor and dampened its rhythm. Ex-vivo bioluminescence recordings of ileum mucosae from transgenic mice revealed endogenous circadian rhythms of P-gp protein expression with a shorter period, larger amplitude, and phase delay in females as compared to males. Importantly, this study provided model structure and parameter estimates to refine PK models of any P-gp substrate to account for sex, feeding and circadian rhythms.


Figure S2: Co-localization of P-glycoprotein and beta-catenin in the membrane of colon mucosa cells in female mice at ZT15.
Double immunofluorescence staining of P-glycoproteein (green) and beta-catenin (red) in mouse ileum using confocal laser scanning microscopy. Co-localization was shown in yellow. Far

Mechanistic Model of talinolol pharmacokinetics
A mathematical model of talinolol PK was designed at the whole-body level, and ultimately aimed to quantify P-gp activity in the gastro-intestinal system, where physiological P-gp expression is usually highest. Only talinolol was considered as the drug metabolites appear in negligible quantities in mice (11). The model simulates talinolol blood and biliary transport in between eight compartments that represent blood, liver, and duodenum, jejunum and ileum membranes and lumens (Fig. 5A). Talinolol oral administration was modeled as an input forcing function in the duodenum lumen representing the exponential emptying of the stomach (12). The drug efflux from the intestinal cells to the intestinal lumen was modeled as a P-gp-mediated active transport. Similarly, the entero-hepatic circulation from the liver to the intestinal lumen towards the biliary system was modeled as an active efflux involving liver P-gp. All other drug transports were modeled as passive diffusion through the tissues. In particular, the drug efflux from the liver towards blood was considered as passive as no experimental evidence exists regarding any interactions between talinolol and the main liver transporters involved in drug efflux towards the blood, including Oat2, Oatp2b1, Mrp 1, 3, 4, 5, or 6. All passive transport rates are modelled using Fick's first law whereas active transport follows Michaelis-Menten kinetics.
The model further considers talinolol renal and intestinal clearance that are represented as first-order kinetics in the blood and the ileum lumen compartments respectively. Although P-gp is expressed in the kidneys, talinolol renal excretion is assumed to be P-gp independent as i) the co-administration of verapamil, a P-gp inhibitor, did not modify talinolol renal clearance in humans and ii) talinolol renal elimination was highly correlated to the urinary clearance of creatinine, an amino-acid that is eliminated through passive diffusion from blood to primary urine in both healthy volunteers and in patients with renal failure (13)(14)(15).
We here describe the equations of the physiologically-based model of talinolol whole body PK. TX denotes talinolol concentrations in the organ X, expressed in μM. The model equations are as follows: d T blood dt = −(3 * k inGut + k inLiver )T blood − k clear1 ( )T blood + k outLiver T liver d T liver dt = k inLiver T blood − k outLiver T liver + k portal ( T duodenum +T jejunum + T ileum ) The drug is administered into the stomach at an initial concentration C0, here 100mg/kg.
The exponential parameter λ(t) represents the rate of stomach emptying. For the sake of simplicity, it was set equal to the rate of drug progression along the duodenum, jejunum and ileum.
Overall, λ(t) represents the drug progression from esophagus -where it is experimentally administered-along through the intestine.
Circadian rhythms (terms in blue) are assumed in P-gp activities in the intestine (efflux towards the intestinal lumen) and in the liver (efflux towards the bile) in agreement with experimental results of this study and literature. Thus, P-gp intracellular concentrations, in the three intestinal membrane compartments and in the liver, are represented by a cosine function considering the main period 24h and the two first harmonics at 12h and 8h. In addition, 24-h rhythms of the renal clearance were also implemented to account for daily variations in renal excretion (16,17).
Similarly, the drug stomach emptying, progression along the intestinal lumen compartments and intestinal clearance are assumed to vary over the 24h span with the same circadian amplitude and phase to account for daily variations in the gastro-intestinal tract motility (18,19). Hence, in total, four circadian rhythms are considered in the model. Circadian rhythms in P-gp activity in the intestine and in the liver were modelled as:

Model parameters: organ volumes
Average male and female mouse weight were set as 0.025 kg and 0.023 kg respectively.
Relative blood volume was assumed to be 58.5 mL/kg (20). The volume of the liver was set to 1.39 mL for male mice and 1.12 mL for female mice, irrespective of feeding conditions (21).
Stomach volume was set to 0.37 mL for female mice (22) and at a 17% greater value for male mice. Duodenum volume was estimated as 0.2295 mL for male mice and 0.1905 mL for females (23). Jejunum volumes were set to 0.7541 mL for males and 0.6259 mL for females.
Ileum volumes were set to 0.0765 mL for males and 0.0635mL for females. The ileum mucosa volume was computed according to information from literature as follows (24). The ileum total radius in mouse was approximated to _ = 3500 and the wall thickness to =

200
. The ileum was assumed to have a cylinder shape so that the volume fraction corresponding to the ileum lumen was estimated by: where l is the ileum length. Hence, the volume of the lumen was set to 90% of the total ileum volume and that of the mucosa to the remaining 10%. The same fractions were kept for the duodenum and jejunum compartments.

Model Results: estimated parameters for passive talinolol transport
The physiologically based model of talinolol chronoPK was used to recapitulate both ileum and hepatic P-gp protein and talinolol PK datasets towards analysis of P-gp activity at the whole organism level. A set of model parameters was estimated for each mouse class, namely male/female, fed/fasted animals. This allowed to conclude on variations according to sex, feeding conditions and circadian timing of all model parameters (Fig. 5, S3). Talinolol stomach emptying and progression along the intestine were faster in male mice compared to females which is in agreement with human data (Fig. S3A, (26, 27)). As for ileum P-gp activity circadian mean, passive absorption rate from ileum lumen to mucosa cells was higher in males compared to females (Fig. S3B). In both sexes, fasting decreased talinolol transport rates from liver to blood and from blood to ileum ( Fig. S3D and F).