Carbamazepine is not a substrate for P-glycoprotein

Abstract

Aims

To determine whether the anticonvulsant carbamazepine (CBZ), a known CYP3A4 substrate, is also a substrate for the multidrug efflux transporter P-glycoprotein (Pgp).

Methods

The role of Pgp in the transport of CBZ was assessed in three systems: (a) in mdr1a/1b(−/−) and wild-type mice after administration of 2 mg kg⁻¹ and 20 mg kg⁻¹, which served as a model for brain penetration; (b) in Caco-2 cells, an in vitro model of the intestinal epithelium that is known to express high Pgp levels; and (c) by flow cytometry in lymphocytes using rhodamine 123, a fluorescent substrate for PgP.

Results

Brain penetration of both doses of CBZ at 1 h and 4 h was comparable in wild-type and mdr1a/1b(−/−) mice. Transport across the Caco-2 cell monolayer was Pgp-independent, and was not affected by the Pgp inhibitor PSC-833. CBZ had no effect on rhodamine 123 efflux from lymphocytes, in contrast to verapamil, which increased fluorescence intensity fivefold.

Conclusion

CBZ is not a substrate for Pgp. Its efficacy is unlikely to be affected by Pgp over-expression in the brain. Furthermore, the interaction of CBZ with drugs that modulate both CYP3A4 and Pgp function such as verapamil is probably due to inhibition of CYP3A4 and not Pgp.

Introduction

Epilepsy is a common condition affecting 1% of the UK population [1]. Approximately 30% of epileptics have inadequate control of seizures with drugs [2]. The mechanisms underlying such drug resistance are poorly understood. It has been suggested that over-expression of the multidrug transporter P-glycoprotein (Pgp) in the blood–brain barrier may increase drug efflux and limit access to the epileptic focus [3]. Clearly, this would only be important for drugs that are substrates for Pgp.

Carbamazepine (CBZ), a widely used anticonvulsant, is regarded as first line therapy for partial seizures [4]. It is also commonly used in combination with other antiepileptic drugs in the treatment of refractory epilepsy [2]. CBZ undergoes extensive metabolism, with the initial oxidative pathways being catalysed by CYP3A4 and CYP2C8 [5]. There is a well-known overlap between substrates for CYP3A4 and Pgp [6], but it is not known whether CBZ is also a substrate for Pgp. It is also interesting to note that CBZ neurotoxicity can be precipitated by concomitant administration of verapamil, erythromycin or grapefruit juice [7]. Although this has been ascribed to inhibition of CBZ metabolism by CYP3A4, these compounds are also known inhibitors of Pgp [8]. Thus, it is possible that the interaction with CBZ may be due to inhibition not only of CYP3A4, but also of Pgp.

In view of these concerns, in this study we have utilized several model systems to investigate whether CBZ is a substrate for Pgp.

Methods

Materials

Minimal Eagle's Medium (MEM), RPMI-1640, fetal calf serum, trypsin EDTA and Hanks balanced salt solution (HBSS) were purchased from Gibco BRL (Life technologies, Paisley, UK). Rhodamine 123, imipramine, verapamil and CBZ were obtained from Sigma (Poole, Dorset, UK). Ethyl acetate and acetonitrile were purchased from Fisher Scientific (Loughborough, UK).

Dosing of wild-type and transgenic knockout mice

Wild-type (fvb1) and mdr1a/1b(−/−) knockout mice (Taconic Farms, Germantown, USA) were administered CBZ (2 mg kg⁻¹ or 20 mg kg⁻¹ in 60% polyethylene glycol:H₂O; i.p.) 1 and 4 h before being sacrificed (n = 4 in each group). The doses used approximate to those used clinically, at the start of therapy, and in those on maintenance CBZ therapy. Blood (1.5 ml) was removed by cardiac puncture prior to removal of brain. The samples were frozen at −20 °C until analysed by h.p.l.c.

Determination of drug transport in Caco-2 cells

Caco-2 cells (1 × 10⁶), cultured in MEM supplemented with 20% (v/v) fetal calf serum were placed in 35 mm, six well-transwell (0.4 µm filters) culture plates (Costar, Bucks, UK) for approximately 2 weeks prior to transport experiments to allow the cells to adhere and reach confluence. Fresh medium was added to both apical (1.5 ml) and basolateral (2.5 ml) compartments every 2–3 days until the cells were 100% confluent. Cell monolayer integrity was assessed by measuring the transepithelial electrical resistance (TEER) using a Millicell-ERS (electrical resistance system). For the transport experiments, the cell monolayers were equilibriated in warm (37 °C) HBSS, following which the medium from all apical and basolateral compartments was removed, and replaced with CBZ (10 µm in HBSS) or HBSS alone. Transport was also assessed in the presence of the Pgp inhibitor PSC-833 (100 µm; a gift from Novartis, Basle), following a 10 min preincubation period. Incubations were performed for 1 h, after which the apical and basolateral compartments were sampled for analysis. Apparent permeability (Papp) values were then calculated for both apical to basolateral (Papp_AtoB), and basolateral to apical (Papp_BtoA) movement of compound [9]. A Papp_BtoA to Papp_AtoB ratio of greater than 1.0 was indicative of active efflux transport of compound by an apical transport protein such as P-gp.

The Caco-2 cells used were shown to express high levels of Pgp mRNA and protein by RT-PCR and Western blotting, respectively (data not shown). Furthermore, the same batch of cells had also been shown to have functional Pgp activity in our laboratories [9].

Development of an h.p.l.c. method for quantification of carbamazepine

Owing to lack of availability of radiolabelled CBZ, an h.p.l.c. method was developed to allow quantification in the animal and tissue culture studies. All experiments were carried out on whole brains, which were weighed following removal, and then homogenized, prior to extraction with ethyl acetate (3 ml × 3). In initial experiments, CBZ (0.8 µg ml⁻¹) was recovered quantitatively from spiked samples of brain and whole blood following the addition of an internal standard (imipramine) and extraction with ethyl acetate (3 ml, three times) with an efficiency of 96.4 ± 5.9% (n = 3). A rectilinear relationship was obtained between detector responses and CBZ recovered from spiked samples over the concentration range 0–1.6 µg ml⁻¹ (r² > 0.995, n = 3). CBZ and imipramine were separated from endogenous material at ambient temperature on a Lichrospher® 60 RP-select B column (Merck, Darmstadt, Germany, 125 × 4 mm, 5 µ). The mobile phase was composed of acetonitrile and ammonium acetate buffer (25 mm, pH 4.0) and was delivered at 1 ml min⁻¹ with a Kontron-325 gradient pump (Kontron Instruments, Massachusetts, USA) equipped with a Kontron-360 autosampler and Kontron-332 u.v. detector set at 285 nm. A gradient of 40%-80% acetonitrile over 15 min was used and CBZ and imipramine eluted at 3.5 and 8.5 min, respectively. Chromatograms were acquired and analysed using a Kontron PC integration package. The limits of detection and quantification were 30 ng ml⁻¹ and 75 ng ml⁻¹, respectively, with intra- and interday assay precision of 97.9% and 98.0%, respectively (n = 4). The data presented were all above the limit of quantification. The amount of CBZ in whole brain was calculated as:

where Y=Total volume of reconstituted extract (ml).

X=Concentration of CBZ in reconstituted extract (ng ml⁻¹).

A=Weight of murine brain (g).

Effect of carbamazepine on efflux of the fluorescent dye rhodamine 123

Lymphocytes were isolated from freshly drawn venous blood as described previously [10]. Rhodamine 123 efflux studies were carried out by the method of Profit et al. [9]. Briefly, lymphocytes were loaded with rhodamine 123 (1.5 µg ml⁻¹; 25min; 4 °C) before being washed twice with RPMI-1640 (1 ml; 1 °C). The cells were then incubated at 37 °C or 4 °C for 3 h in 1 ml of dye-free media to allow dye efflux, while parallel experiments were performed at 37 °C in the presence of either the positive control, verapamil (30 µm), or CBZ (10 and 100 µm). The cells were then harvested, washed twice in RPMI-1640, and fixed before flow cytometric analysis on an EPICS-XL flow cytometer (Beckman Coulter, Bucks, UK). The lymphocyte population was electronically gated to exclude debris. At least 5000 events were collected for each sample, with cellular rhodamine 123 fluorescence being plotted against the number of events. Data acquisition was performed using the computer program EXPO analysis software to determine median FL1 fluorescence values.

Statistical analysis

All results are presented as mean±s.d. with 95% confidence intervals for the difference between the means, where appropriate. Statistical analysis was performed by using the unpaired t-test, after confirming that the data were normally distributed. A two-tailed P value of < 0.05 was accepted as being significant.

Results

Studies in wild-type and mdr1a/1b(−/−) mice showed that there were no differences in the brain concentrations of CBZ after administration of low dose (2 mg kg⁻¹) at 1 h (95% CI for the difference −657–236), and high dose (20 mg kg⁻¹) at both 1 (95% CI −6104–5004) and 4 h (95% CI −357–282) (Figure 1).

Figure 1.

Brain concentrations of carbamazepine (CBZ) in wild-type (WT) and mdr1a/1b(−/−) knockout (KO) mice at 1 and 4 h after administration of either low (2 mg kg⁻¹) or high (20 mg kg⁻¹) doses of CBZ. Results represent mean ± s.d. of determinations in four mice within each group. No significant differences were found between WT and KO mice at the same doses and times (Student's t-test).

Within whole blood, CBZ concentrations in wild-type and knockout mice again did not show any difference and were as follows: 0.14 ± 0.08 µg ml⁻¹ and 0.40 ± 0.29 µg ml⁻¹ (95% CI for the difference −0.1–0.6) 1 h after administration of 2 mg kg⁻¹ CBZ in wild-type and knockout mice, respectively; and after administration of 20 mg kg⁻¹ CBZ, the concentrations were 7.61 ± 1.35 µg ml⁻¹ and 7.78 ± 3.05 µg ml⁻¹ (95% CI−3.9–4.3) at 1 h, and 0.56 ± 0.18 µg ml⁻¹ and 0.55 ± 0.16 µg ml⁻¹ (95% CI −0.3–0.3) at 4 h in wild-type and knockout mice, respectively.

After low dose CBZ administration, the whole blood and brain concentrations at 4 h were below the limit of sensitivity of the assay.

For studies with Caco-2 cells, all monolayers had a TEER greater than 200 Ω cm⁻², indicating confluence of the monolayer. The ratio of Papp_BtoA to Papp_AtoB was 0.78 ± 0.18 (n = 4 experiments in triplicate), indicating that CBZ was not actively transported from the basolateral to apical compartments. This was confirmed by the use of the Pgp inhibitor PSC-833, which failed to alter the ratio (0.87 ± 0.32; 95% CI for the difference −0.5–0.4).

The effect of CBZ on rhodamine 123 efflux in lymphocytes was determined by flow cytometry, and was compared with the effect of verapamil (30 µm) (Figure 2). In the presence of verapamil, the median fluorescence increased fivefold (P < 0.005). In contrast, CBZ had no effect on fluorescence intensity at either 10 µm or 100 µm (Figure 2).

Figure 2.

Ratios of median fluorescence intensity (FL1) of human lymphocytes in the presence of either verapamil or carbamazepine (CBZ) relative to control lymphocytes. Data are expressed as mean ± s.d. of four separate experiments (performed in duplicate). Statistical analysis performed by t-test: *P < 0.005 (when compared with control).

Discussion

P-glycoprotein (Pgp), an ATP-dependent membrane-bound drug efflux pump that is ubiquitously distributed, transports a large number of therapeutically and structurally disparate drugs [11]. It first came to prominence as a mechanism for conferring resistance to cancer chemotherapy [12]. More recently, it has been suggested that Pgp may also confer drug resistance in other diseases, including HIV [13] and epilepsy [3].

Approximately 30% of epileptics become refractory to drug treatment [2]. The finding of high levels of Pgp expression in surgically resected temporal lobe specimens has led to speculation of its role in the pathogenesis of refractory epilepsy [3]. It is therefore important to determine which of the anticonvulsants are Pgp substrates, as such knowledge would be of obvious therapeutic value. In this study, we have therefore investigated whether CBZ is a Pgp substrate using three different model systems that serve as surrogates for Pgp expression in different tissues.

Transgenic mdr1a/1b(−/−) mice are particularly beneficial when looking at brain penetration of drugs. These knockout mice have provided valuable information on the role of Pgp in drug disposition in vivo [11]. Our results showed that there were no differences in brain and whole blood concentrations of CBZ at both 1 and 4 h after administration of either high or low doses. This is an important observation since in patients with refractory epilepsy where Pgp has been shown to be over-expressed [3], the use of CBZ should lead to adequate brain concentrations, and hence no impairment of efficacy. Furthermore, many patients with refractory epilepsy are already treated with CBZ, and yet are still not adequately controlled [2]; this by itself indicates that factors other than (or as well as) Pgp expression are also determinants of treatment resistance in epilepsy.

The Caco-2 cell monolayer serves as a useful model to investigate the role of Pgp in determining drug absorption from the intestine. In accordance with the results obtained with the mdr1a/1b(−/−) knockout mice, no active transport of CBZ from the basolateral to apical membranes was demonstrated indicating that CBZ is not a substrate for intestinal Pgp. This was further confirmed by the use of PSC-833, a potent Pgp inhibitor [14], which failed to change the directional transport. This observation is of interest for two main reasons: first, the interaction of CBZ with verapamil and erythromycin [7] is therefore largely due to inhibition of CYP3A4, and not Pgp. Taken together with the fact that CBZ entry into the brain is not influenced by Pgp, it can be concluded that the interaction is occurring at the level of the intestinal wall and liver, and not brain. Second, it provides further evidence that not all CYP3A4 substrates are also Pgp substrates, and is in accordance with a recent study, which concluded that the overlap in substrate specificities of CYP3A and Pgp was nothing more than coincidental [6].

The third method used to assess Pgp-mediated transport of CBZ was with flow cytometry using the fluorescent dye rhodamine 123, which allowed assessment of function at the level of the individual cell [9]. This method gives an indication whether a compound is a Pgp substrate, but cannot by itself be used to conclusively prove that a compound is not a substrate since there are multiple binding sites within the Pgp molecule [15]. Furthermore, rhodamine 123 efflux may also partly be due to other transporters such as MRP1 [16]. In accordance with previous studies, verapamil, a known Pgp inhibitor [8], interfered with rhodamine 123 efflux from lymphocytes. However, CBZ had no effect on rhodamine 123 efflux indicating that it did not compete for transport by Pgp, or indeed by MRP1.

In conclusion, using three different systems, we have shown that CBZ is not a substrate for the efflux transporter Pgp. Thus, its efficacy is unlikely to be inadvertently affected by Pgp over-expression in the blood–brain barrier. Furthermore, the interaction of verapamil and other CYP3A4 substrates with CBZ is probably due to inhibition of CYP3A4 and not Pgp.