Advanced Drug Delivery Reviews 64 (2012) 930–942
Contents lists available at SciVerse ScienceDirect
Advanced Drug Delivery Reviews
journal homepage: http://www.elsevier.com/locate/addr
The transport of antiepileptic drugs by P-glycoprotein☆ Chunbo Zhang a, Patrick Kwan b,c,d, Zhong Zuo a, Larry Baum a,⁎
a School of Pharmacy, The Chinese University of Hong Kong, Shatin, Hong Kong
b Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong c Department of Medicine, The University of Melbourne, Royal Melbourne Hospital, Melbourne, Australia
d Department of Neurology, Royal Melbourne Hospital, Melbourne, Australia
article info
Article history:
Received 3 September 2011 Accepted 7 December 2011 Available online 16 December 2011
Keywords:
Antiepileptic drugs
Blood brain barrier
Drug resistance
Epilepsy
P-glycoprotein Structure–activity relationship
Contents
abstract
Epilepsy is the most common serious chronic neurological disorder. Current data show that one-third of pa- tients do not respond to anti-epileptic drugs (AEDs). Most non-responsive epilepsy patients are resistant to several, often all, AEDs, even though the drugs differ from each other in pharmacokinetics, mechanisms of ac- tion, and interaction potential. The mechanisms underlying drug resistance of epilepsy patients are still not clear. In recent years, one of the potential mechanisms interesting researchers is over-expression of P- glycoprotein (P-gp, also known as ABCB1 or MDR1) in endothelial cells of the blood–brain barrier (BBB) in epilepsy patients. P-gp plays a central role in drug absorption and distribution in many organisms. The ex- pression of P-gp is greater in drug-resistant than in drug-responsive patients. Some studies also indicate that several AEDs are substrates or inhibitors of P-gp, implying that P-gp may play an important role in drug resistance in refractory epilepsy. In this article, we review the clinical and laboratory evidence that P- gp expression is increased in epileptic brain tissues and that AEDs are substrates of P-gp in vitro and in vivo. We discuss criteria for identifying the substrate status of AEDs and use structure–activity relationship (SAR) models to predict which AEDs act as P-gp substrates.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction…………………………………………………….. 930

2. ThehypothesisthatAEDsactassubstratesforP-gp…………………………………….. 931
3. MethodsandcriteriatoidentifysubstratestatusofAEDs………………………………….. 931
4. TheoverexpressionofP-gpinepilepsypatientsandanimalmodels……………………………… 933
   4.1. Inepilepsypatients………………………………………………. 933
   4.2. Inanimalmodels……………………………………………….. 933
5. AEDsactassubstratesofP-gp……………………………………………… 934 5.1. Invitrocellmodels……………………………………………….. 934 5.2. Invivoanimalmodels……………………………………………… 935 5.3. Inepilepsypatients………………………………………………. 935 5.4. ThesubstratestatusofAEDs…………………………………………… 936
6. Structure–activityrelationship(SAR)betweenP-gpandAEDs………………………………… 936
7. AEDsinducetheoverexpressionofP-gp…………………………………………. 939
8. ProposedfurtherresearchtodeterminesubstratestatusofAEDs……………………………….. 940
9. Conclusions…………………………………………………….. 940
   References……………………………………………………….. 940
   ☆ This review is part of the Advanced Drug Delivery Reviews theme issue on “Antiepi- leptic Drug Delivery”.
   ⁎ Corresponding author. Tel.: +852 39436833; fax: +852 26035295. E-mail address: lwbaum@cuhk.edu.hk (L. Baum).
   0169-409X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.addr.2011.12.003
10. Introduction
    Epilepsy is the most common serious chronic neurological disorder, affecting more than 50 million people worldwide [1]. It is characterized by recurrent seizures, and is broadly classified into two types: focal or generalized epilepsy. Focal epilepsy causes

seizures originating within networks limited to one hemisphere; gen- eralized epilepsies cause seizures “originating at some point within, and rapidly engaging, bilaterally distributed networks” [2–5]. Al- though epilepsy is a major public health issue, we still understand lit- tle on many aspects of this disorder. More than 20 antiepileptic drugs (AEDs) are used clinically, and this number is still increasing. Despite the availability of so many AEDs, 30–40% of patients do not respond to pharmacotherapy [6,7]. The epilepsy in such patients are often resis- tant to a range of AEDs, even though the drugs differ from each other in pharmacokinetics, mechanisms of action, and interaction potential. Epilepsy drug-resistance has been defined as “failure of adequate trials of two tolerated and appropriately chosen and used AED sched- ules (whether as monotherapies or in combination) to achieve sustained seizure freedom” [8]. The mechanisms underlying drug re- sistance of epilepsy are still unclear but may include pathology at the epileptic focus or polymorphisms of drug transporters [9].
The blood–brain barrier (BBB) is a diffusion barrier, composed of endothelial cells, astrocyte end-feet, and pericytes (PCs) [10]. It limits the brain penetration of small molecules and protects the central ner- vous system (CNS) from xenobiotics. More than 98% of small mole- cules cannot penetrate the BBB. Thus, the entry of most drugs to the CNS is limited by the BBB [11,12]. One of the potential mechanisms of AED resistance is overexpression of multi-drug transporters in cap- illary endothelial cells of the BBB in non-responsive epilepsy patients. Since AEDs must traverse the BBB to enter the brain and exert their desired effects, the overexpression of multidrug transporters in the endothelial cells of the BBB may contribute to drug resistance [13,14]. One of the most intensively studied multidrug transporters is P-glycoprotein (P-gp, also known as ABCB1 or MDR1), which plays a central role in drug absorption and distribution in many or- ganisms. P-gp was discovered as a drug resistance gene in Chinese hamster ovary cells in 1976 by Juliano et al. [15]. P-gp functions as an ATP-driven efflux pump of substrates ranging from approximately 300 to 4000 Da in mass, including some HIV protease inhibitors, anti- biotics, immunosuppressive agents, and many other prescribed drugs [15–17].
Brain concentrations of many drugs increase in MDR1-null mice [18]. Similarly, P-gp inhibitors increase drug penetration into the brain [19,20]. For example, the P-gp inhibitor PSC833 can increase paclitaxel accumulation in the mouse brain [21]. These studies suggest that P-gp plays a remarkable role in restricting access of drugs to the CNS. The capillary endothelial cells of the BBB highly express P-gp, and this expression is increased in epilepsy patients [13,22–24]. In ad- dition, the expression is higher in drug-resistant than in drug- responsive patients [25,26]. Some studies indicated that several AEDs are substrates or inhibitors of P-gp, implying that P-gp plays an important role in refractory epilepsy, although other studies in various models provided conflicting evidence [9,27–29]. In general, evidence is greater for lipophilic AEDs as P-gp substrates [22,29,30]. However, there is still no consensus on whether (or which) AEDs are substrates of P-gp.
A wide spectrum of drugs, encompassing drugs of different clas- ses, have been found to be substrates of the P-gp efflux pump [31]. Identifying the substrate status of compounds is important for new drug design. Study of the common structural features of P-gp sub- strate compounds may help define their structure–activity relation- ship (SAR). Different models have been used to this end [31–33]. However, no review has focused on the SAR of AEDs acting as sub- strates for P-gp.
This review discusses the possible roles of P-gp in AED resistance, including the overexpression of P-gp in epilepsy and whether AEDs are substrates of P-gp. We examine the criteria for identifying the substrate status of AEDs, and use SAR models to predict whether AEDs act as P-gp substrates. Finally, we discuss the causal relation of AED resistance and P-gp overexpression in light of evidence that P- gp can be induced by some AEDs in vivo and in vitro [34,35].

2. The hypothesis that AEDs act as substrates for P-gp
   P-glycoprotein is one of the most intensively studied ABC family members. The ABC family is a large group of proteins comprised of membrane transporters, ion channels, and receptors. The transporters are conserved among species. In humans, two different 170 kDa pro- teins are called P-gp. They are encoded by the MDR1 (ABCB1) and MDR2 (ABCB4) genes, which are located near each other on chromo- some 7q21.1 [36]. In rodents, P-gp proteins are encoded by three genes, mdr1a, mdr1b, and mdr2 [37]. The MDR genes are classified into two groups. Human MDR1 and rodent mdr1a and mdr1b encode the transporters involved in multidrug resistance. MDR1 has the same functions as mdr1a and mdr1b. MDR2 and mdr2 encode phos- photidylcholine transporters in biliary canaliculi [37,38]. Most reports use the term P-gp to refer to the protein encoded by MDR1, mdr1a and mdr1b, and that convention will be followed here.
   The human P-gp is composed of two homologous halves. Each of them consists of an N-terminal, one hydrophobic transmembrane do- main (approximately 250 amino acid residues), one hydrophilic nu- cleotide binding domain (approximately 300 amino acid residues) and a C-terminal [39,40]. P-gp modulates drug distribution and dispo- sition in many organisms. There are hundreds of transport substrates of P-gp, including natural products, chemotherapeutic drugs, steroids, fluorescent dyes, linear and cyclic peptides, and ionophores. Most of them are hydrophobic, weakly amphipathic, and contain a heterocy- cle. However, the mechanism by which P-gp recognizes substrates is unclear. Many drugs used in our daily life are substrates of P-gp. The bioavailability of these drugs can be reduced by P-gp. In P-gp null (mdr1a−/− or mdr1b−/−) mice, plasma concentration of some drugs was increased compared with wild type mice [41].
   P-glycoprotein is not only highly expressed in multidrug resistant tumor cells, but is also highly expressed in barrier and excretory tissues, such as the epithelial cells of biliary hepatocytes, adrenal cortex, pancre- atic ductules, large intestine mucosal cells, proximal renal tubules, tes- tis, placenta, and blood–tissue barriers [30,42–46]. The capillary endothelial cells of the blood brain barrier (BBB) express P-gp at high levels as compared to other tissues. P-gp is located in the luminal plas- ma membrane (blood side) of brain capillary endothelial cells [47]. The pattern of distribution of P-gp is ideal to limit drug entry to the brain or to permit efflux of drugs from brain to blood, which suggests that P-gp plays an important role in protecting the brain against xenobiotics [18,30,46–51]. In general, the more lipophilic drugs penetrate more readily into the brain. However, there are many lipid-soluble drugs with lower brain permeability than would be predicted because they are pumped out of the brain by P-gp [52,53]. Most AEDs are very lipo- philic, but about one-third of epilepsy patients do not respond to them. One of the possible explanations is that lipophilic AEDs are pumped out of the brain by P-gp in the BBB, which decreases the con- centration of AEDs in the brain and affects drug efficacy. In order to sup- port the hypothesis, three criteria should be satisfied: first, AEDs would be substrates of P-gp; second, the P-gp level in the BBB would be higher in drug-resistant than in drug-responsive epilepsy patients; third, the concentration of AEDs in the brain would be lower in drug-resistant than in drug-responsive epilepsy patients. The evidence from patients and animal models indicates that refractory epilepsy is associated with the overexpression of P-gp and lower concentration of AEDs in brain [54]. In vivo and in vitro evidence also indicates that some AEDs act as substrates of P-gp, however there is some inconsistent evidence in this regard [13,30]. These observations are somewhat consistent with the hypothesis that lipophilic AEDs are pumped out of the brain by P-gp in the BBB, contributing partly to drug refractory epilepsy.
3. Methods and criteria to identify substrate status of AEDs
   Early studies of P-gp found that it was expressed in cancer cells and was a cause of drug resistance in cancer [15–17]. Many compounds
   C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942 931

932 C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942
were shown to be substrates or inhibitors of P-gp [30,55–57]. Evidence further indicated that P-gp is relevant to drug-resistant epilepsy, and that some AEDs are substrates or inhibitors for P-gp [28,29,58–61]. Several methods have been established to detect whether AEDs are sub- strates for P-gp in vivo or in vitro, including clinical research on patients, animal models, and cell based assays.
In clinical research, brain tissues isolated from drug-resistant epi- lepsy patients undergoing surgery to remove seizure foci are com- monly used to detect the location and expression level of P-gp. This is a direct way to study the association between refractory epilepsy and the drug efflux function of P-gp. However, there is a limitation in using human brain to identify whether an AED is a substrate of P- gp because, at present, there are not validated and readily feasible methods to detect the brain concentration of an AED and associate it with the P-gp level in live patients. Moreover, we lack correct con- trols to compare with drug-resistant epilepsy. Positron emission to- mography (PET) will be a useful tool to directly detect the function of P-gp in human brains. In monkeys, the brain distribution of [11C]- verapamil, a substrate of P-gp, increased in the presence of PSC833, an inhibitor of P-gp [62]. In a chronic rat model of temporal lobe ep- ilepsy which is resistant to phenobarbital (PB), a P-gp inhibitor, tari- quidar (TQD), increased the magnitude of [18F]MPPF (a substrate of P-gp) binding in hippocampus [63]. PET was also used in human brains to detect effects of MDR1 SNPs [64,65].
Animal models have played a key role in discovering and charac- terizing AEDs. They have advantages over clinical studies. First, animal models are used to detect associations among the level of P-gp in the BBB, the concentration of AEDs in the brain, and the drug responsive- ness status of epilepsy (drug responsive or non-responsive). Using rat epilepsy models, we can select subgroups which either respond or do not respond to AED treatment. It is possible to detect the level of P-gp in different brain regions and test the associated AED level. Then we can determine whether BBB P-gp expression differs between drug non-responders and drug responders and whether the P-gp level af- fects AED concentration in the brain. For example, in rats resistant to PB, P-gp was overexpressed in the hippocampus and other limbic brain regions compared with rats responsive to PB, and the anticon- vulsant activity of PB was restored by TQD [66].
Second, animal models are used to detect the function of P-gp. Using in vivo microdialysis in rats, it was shown that P-gp inhibitors can enhance the concentration of AEDs, such as phenytoin (PHT), PB, carbamazepine (CBZ), lamotrigine (LTG), and felbamate (FBM), in extracellular fluid of the cerebral cortex [67–69]. In mdr1a knock- out mice, which do not express the mdr1a isoform in the BBB, various drugs reached significantly higher brain concentrations than in wild type mice [70]. In contrast to human studies, animal models have the advantage that invasive methods may be used, allowing measure- ment of AEDs or P-gp expression. These models can provide evidence on whether P-gp is involved in controlling the brain distribution of AEDs. However, they may bear little resemblance to chronic epilepsy in humans.
However, in vivo studies with patients and animal systems may be too complex to prove whether AEDs are substrates for P-gp or that a decrease of AED levels is caused by the overexpression of P-gp. In vivo methods are also not very practical for rapid and large-scale screening of compounds. Compared with in vivo systems, in vitro as- says have advantages and are widely used. After detecting the activity of P-gp in vivo, in vitro systems can be used to optimize the discovery, or vice versa. An advantage of in vitro methods is the potential to dem- onstrate the molecular mechanism of action between AEDs and P-gp. Cell lines transfected with the MDR1 gene and thus overexpressing the P-gp protein are models to evaluate the potential of compounds to act as P-gp substrates/inhibitors. The results can be confirmed using P-gp inhibitors, which can reverse the effect of P-gp overexpres- sion on AED transport. Several AEDs, such as PHT, PB, oxcarbazepine (OXC), eslicarbazepine acetate (ESL), and levetiracetam (LVT), were
transported by P-gp in LLC and MDCK cell models. This directional transport of AEDs could be inhibited by selective P-gp inhibitors [29,71]. Moreover, through studying P-gp using cell biology, genetics, bioinformatics, and biochemistry, its mechanisms of action are being revealed [54,72–76]. However, in vitro systems cannot mimic physio- logical conditions. They cannot model the bioavailability, the brain accessibility, and the specific pharmacodynamic action in in vivo epi- leptic systems. Thus, it is necessary to validate activity using in vivo systems. Further, animal models differ from humans; it is necessary to confirm findings in the human.
The US Food and Drug Administration drafted guidance for decid- ing whether a drug is a P-gp substrate, recommending a bidirectional transport assay using cultured cells [77,78]. However, since most AEDs have low solubility and high permeability, diffusion across the cell layer may mask directional transport in a bidirectional transport assay; the assay may be started using equal concentrations of drug on both sides of the cell layer (concentration equilibrium transport assay, or CETA) in order to remove the concentration gradient and thus diffusion in order to reveal directional transport [29]. To evaluate the results of published studies, we need to establish criteria to define whether an AED acts as a substrate for P-gp and whether this action is associated with refractory epilepsy. Criteria to define the substrate status of AEDs should include in vivo and in vitro evidence: 1. In in vitro, or cell models, the AED is transported by P-gp, and this trans- port can be inhibited by P-gp selective inhibitors. Because different cell models may show inconsistent results (the P-gp expression level may differ among different models), a standard should be estab- lished to identify the reliability of evidence. Using more than one cell model to confirm that an AED acts as a P-gp substrate would increase the reliability. 2. In animal models, the brain concentration of an AED is decreased by overexpression of P-gp, and this decrease can be re- stored by P-gp inhibitors. Alternatively, the brain concentration of an AED is increased by adding P-gp inhibitors or by decreasing P-gp expression. 3. Evidence from patients may also be considered. PET, intraoperative microdialysis, and tissue isolated from epilepsy patients by surgery are common methods which can be used to inves- tigate the association between P-gp expression and AED brain con- centration. However, human samples are limited and may not be available to provide evidence of substrate status. Combining these three types of evidence, we suggest the following set of definitions of the substrate status of an AED:

1. If the AED is transported by P-gp in vivo (animal models or pa- tients) and in vitro (cell models), we define it as a definite substrate of P-gp.
2. If the AED is transported by P-gp in vitro but no in vivo evidence is available, or if the AED is transported by P-gp in vivo but no in vitro evidence is available, we define it as a probable substrate of P-gp.
3. If results of in vitro and in vivo transport studies of the AED are in- consistent, we define it as a possible substrate of P-gp.
4. If the AED is not transported by P-gp in vitro but no in vivo evi- dence is available, or if the AED is not transported by P-gp in vivo but no in vitro evidence is available, we define it as unlikely a sub- strate of P-gp.
5. If the AED is not transported in vivo and not transported in vitro, we define it as not a substrate of P-gp.
   These can be summarized in a decision tree (Fig. 1). Because of the
   limited availability of human data, evidence from patients weighs less heavily in these criteria than does evidence from animal models. On the other end of the spectrum, cell models are simple and thus weigh more heavily than in vivo models in these criteria.
   In order to define whether the P-gp substrate status of an AED is as- sociated with refractory epilepsy, we must consider evidence from humans, animals, and cellular models. We may use these criteria: First, the AED is a P-gp substrate. Second, in clinical studies, the ex- pression of P-gp is increased and the concentration of the AED is

decreased in drug-resistant epilepsy brains compared with drug re- sponsive brains, and the inhibition of P-gp is associated with clinical benefit. Third, the results in humans and animals are consistent. Com- bining the above criteria, we may be able to explain the relationship between P-gp and refractory epilepsy. However, the evidence reported to date is far from sufficient to give a definite conclusion now. More study of the relationship between AEDs and P-gp in epilep- sy will be required.

4. The overexpression of P-gp in epilepsy patients and animal models
   4.1. In epilepsy patients
   P-gp has the remarkable ability to restrict drug access to the CNS, which is demonstrated in the comparisons of P-gp gene knockout and wild type mice. Some evidence indicates that P-gp plays a role in the BBB of drug resistant epilepsy patients [22,30,79]. Does the expres- sion of P-gp differ between drug-resistant and drug-responsive epi- lepsy patients? If the level of P-gp is higher in drug-resistant patients than in drug-responsive patients, the resulting lower con- centration of AEDs in the parenchymal space may lead to the lack of drug response in such patients.
   Tishler et al. reported the enhancement of P-gp expression in the capillary endothelial cells of epileptic tissues compared with non- epileptic tissues. The brain mRNA level of MDR1 was increased more than 10 times in these patients [25]. Subsequent studies confirmed that the overexpression of P-gp was mainly located in the capillary endothelial cells [26]. Kwan et al. compared the P-gp level in capillary endothelial cells of temporal lobe tissues from epileptic patients after surgery. They demonstrated that P-gp expression in patients with re- current seizures was significantly higher than in seizure free patients [80]. The overexpression of P-gp in endothelial cells isolated from re- fractory patient brain specimens compared with cells from non- epileptic specimens was demonstrated at the mRNA and protein levels [26].
   Evidence exists that P-gp is overexpressed not only in endothelial cells but also in neurons and glial cells in drug resistant epilepsy patients (Table 1). Patients with mesial temporal lobe epilepsy (MTLE), tuberous sclerosis, malformations of cortical development (MCD) or focal cortical dysplasia (FCD) showed overexpression of P-gp in neurons and astrocytes [62,81,82]. The P-gp up-regulation in neurons and glial cells could lead to low concentrations of AEDs in the parenchymal space. The astrocytic foot processes around endo- thelial cells may pump drugs to endothelial cells. Neuroectodermal cells expressing MDR1 had lower intracellular PHT concentrations than did MDR1-negative cells [25]. P-gp-mediated drug extrusion by
   astrocytes of epileptic tissues was greater than by control astrocytes [83]. These results suggest that the upregulation of P-gp in astrocytes and neurons may also play a role in controlling the exchange of xeno- biotics between brain and plasma at the BBB [83]. The findings of overexpression of P-gp in the brain tissues of refractory epilepsy pa- tients are not complete because most of the studies lack the adequate controls, tissues from drug-responsive epilepsy patients, since these patients are not normally subjected to brain surgery. Thus, we need further study on whether P-gp overexpression in refractory epilepsy patients is associated with drug resistance or is just a general phe- nomenon of epilepsy irrespective of drug responsiveness.
   4.2. In animal models
   Animal models have provided evidence for P-gp overexpression in epileptogenic brain tissue from animals with refractory epilepsy (Table 2). Induced epileptic rat models have been widely used in studying P-gp expression and AED distribution. P-gp was overex- pressed in endothelial cells, neurons, and astrocytes in different kin- dled models [84–86]. In some cases, the overexpression of P-gp was transient and reversible. Kainate-induced epileptic rats showed tran- sient up-regulation of P-gp in multiple brain regions [87]. In electri- cally induced status epilepticus (SE) rats, P-gp was up-regulated within 1 week after SE to levels greater than those in chronic epileptic rats, and this increase was reversible [88].
   In some reports, overexpression of P-gp was investigated by mea- suring immediately after kindling [84–86,89]. However, measuring the P-gp level two weeks after seizures revealed no overexpression, suggesting the transient nature of P-gp overexpression by seizures in the temporal lobe [90]. Overexpression of P-gp may thus be a result of uncontrolled seizures but not of the processes underlying epilepsy [87,90]. However, a long-term increase of P-gp was reported in pen- tylenetetrazole kindled rats [84–86]. The differences in extent of P-
   C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942 933
   Fig. 1. Decision tree to determine P-gp substrate status of a compound.
   Table 1
   Overexpression of P-gp in human epileptic pathologies.
   Pathology P-gp overexpression in cell types References (Y/N)
   Endothelial Neurons Astrocytes cells
   Focal cortical dysplasia Y Hippocampal sclerosis Y Tuberous sclerosis Y Temporal lobe epilepsy Y Malformation of cortical –
   Y Y Y Y Y Y – – – Y
   [23,126–128] [24–26,83,127,129] [25,62,130] [80,120]
   [81]
   development
   Y/N: yes/no; –: not reported.

934 C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942
Table 2
Overexpression of P-gp in rat epilepsy models.
Rat model P-gp overexpression in different regions (Y/N) References Cerebral cortex Hippocampus Limbic regions Amygdala Dentate gyrus
Pentylenetetrazole Y MAM N Pilocarpine – Kainic acid – Electrical –
– – Y – Y – Y Y Y Y
– – [86]
– – [85]
– – [84]
– Y [84,87,93,131] Y Y [66,88,92,106]
MAM: methylazoxymethanol. Y/N: yes/no; –: not reported.
gp induction among the various models may be due to the different durations of seizures [63,87,88,90].
The above studies did not demonstrate that the overexpression of P-gp was associated with drug-resistant epilepsy because they did not compare refractory animals with those that were responsive. Drug-resistant epilepsy rats were selected according to their response to PB. P-gp was overexpressed in the hippocampus and other limbic brain regions in the resistant rats compared with responsive rats, and P-gp was confined to the brain capillary endothelial cells [63,66,91]. Similarly, the P-gp level in capillary endothelial cells of PHT-resistant rats was twice that of PHT-responsive rats [92]. P-gp overexpression in the BBB resulted in decreases of brain levels of the P-gp substrates ondansetron, PHT, Rhodamine123 (Rho123) and PB [85,86,93]. The reduced level of ondansetron in MAM-induced seizure rats was reversed by adding a P-gp inhibitor [85]. The brain levels of some drugs increased 10–50 fold in P-gp null (mdr1a/b−/−) vs. wild type mice [18,94]. The hippocampal PHT concentration of mdr1a/b−/− mice was higher than that of wild type mice [93]. The above results sug- gest that the P-gp level plays an important role in the response to anti- epileptic drugs.

5. AEDs act as substrates of P-gp
   5.1. In vitro cell models
   In vitro cell models were used to study whether AEDs were sub- strates of P-gp (Table 3). There are three common detection methods based on cell lines: measure whether an AED affects the uptake of a substrate marker of P-gp, detect the permeability of an AED through polarized cell monolayers using transwells, or detect the uptake of an AED.
   Weiss et al. (2003) used the calcein acetoxymethylester (CAM) uptake assay to test the ability of AEDs to inhibit P-gp in LLC-MDR1 cells (LLC-PK1 cells transfected with human MDR1) and primary por- cine brain capillary endothelial cells. They found that CBZ, PHT, LTG, and valproate (VPA) inhibited the P-gp efflux function [9]. In OS2.4/ Doxo cells (canine osteosarcoma cells induced by exposure to doxo- rubicin to highly express P-gp), gabapentin (GBP), LTG, LVT, and PB decreased Rho123 efflux, while CBZ, FBM, PHT, topiramate (TPM), and zonisamide (ZNS) did not affect the uptake of Rho123 [95]. This method is fast and reproducible, but it poorly distinguishes substrates and inhibitors of P-gp. The compounds acting as substrates or inhibi- tors of P-gp may have different binding sites [96]. The relationships between AEDs and substrate markers may be competitive, noncom- petitive, or cooperative. A decrease in the uptake of calcein or Rho123 by an AED indicates that they have a competitive relationship and that the AED can be a substrate or inhibitor of P-gp. AEDs which do not inhibit the uptake may act as substrates of P-gp. Therefore, this is an indirect method to study the substrate status of AEDs, and a di- rect measurement is required for more reliable identification of sub- strate status.
   The uptake of AEDs was also tested, which is a more direct method and avoids the confusion of multiple mechanisms. In LoVo/dx cells (derived from the LoVo human colon adenocarcinoma cell line), which overexpress P-gp, the 10-OH-CBZ concentration was equal to that in LoVo cells but could be increased by the P-gp inhibitor XR9576 [97]. Yang et al. [122] used cultured rat brain microvascular endothelial cells (rBMECs) to test the uptake of PB. The uptake oc- curred in a time-, concentration-, and temperature-dependent man- ner which could be increased by cyclosporine A (CsA), ketoconazole, or the metabolic inhibitor dinitrophenol, all of which also decreased the efflux of PB [98].
   Table 3
   AEDs as substrates of P-gp in cell lines.
   Drug Evidence of AEDs as P-gp substrates in specific cell lines (Y/N) References Caco-2 MDCK LLC BMEC BREC Daoy/daoyar2 OS2.4/Doxo
   Carbamzaepine
   Ethosuximide
   Phenytoin
   Phenobarbital
   Sodium valproate Acetazolamide
   Topiramate
   Levetiracetam
   Vigabatrin
   Lamotrigine
   Gabapentin
   Felbamate
   Zonisamide
   Oxcarbazepine Eslicarbazepine acetate Carbamazepine-10,11-epoxide S-licarbazepine
   N – N – N N – – – – Y Y N Y N Y Y – – Y Y – – – – – – – – – ? – – – N Y – – – Y
   [9,28,71,95,99,100,132,133] [58,99] [9,25,28,29,58,70,95,99,104,132] [9,29,58,95,98]
   [9,27]
   [99] [9,95,99,59] [28,29,95] [99] [9,29,95,99] [9,95,99] [95]
   [95] [71] [71] [71] [71]
   N N
   N N
   N Y
   N Y
   – N
   Y –
   N –
   – N
   N –
   N –
   N –
   –––––– N –––––– N
   – Y – Y – Y – Y
   Y – Y – Y – Y –
   – – – – – – – – – – – –
   – – – – Y – – – Y N – – – Y
   Y/N: yes/no; –: not reported; ?: conflicting evidence. BREC: bovine retinal endothelial cells.
   BMEC: brain microvessel endothelial cells.

Monolayer systems have been used to test AED bi-directional transport by P-gp. This method uses a simplified model of the blood–brain barrier. Cells expressing P-gp are seeded on transwell membranes and grown to confluency. P-gp is expressed on the apical side. If the drug can be transported from the basolateral to the apical side of the monolayer, it is a substrate of P-gp. In a Caco-2 monolayer model, the transport rates of vigabatrin (VGB), GBP, PB, LTG, and CBZ were equivalent in both directions, while PHT, TPM, and ethosuxi- mide (ESM) displayed apical to basal transport. However, none of them were affected by P-gp inhibitors. Only acetazolamide (AZD) had a basolateral to apical Papp value much greater than apical to basolateral Papp value (3-fold greater), and the efflux was inhibited by a P-gp inhibitor [99,100]. In MDCKII or LLC cell monolayer models, VPA and CBZ did not exhibit directional transport [27,28]. PHT and LVT were directionally transported by mouse P-gp but not human P-gp, while CsA (a known substrate of P-gp) was transported by both types of P-gp [28].
However, monolayer directional transport assays suffer from a lack of sensitivity since many AEDs have high passive diffusion, which can overwhelm and thus mask directional transport by P-gp. Starting the assay with equal concentrations of drug on both sides of the monolay- er removes the concentration gradient, thus eliminating net diffusion and making transport by P-gp detectable. Using this concentration equilibrium transport assay system in LLC and MDCKII monolayer models, PHT, LVT, LTG, TPM and PB were found to be substrates of human P-gp, but CBZ was not a substrate [29,59,101,102]. The sub- strate status of PHT and PB was concentration dependent [58]. The major active metabolite of CBZ, carbamazepine-10,11-epoxide (CBZ- E), was a substrate of P-gp. OXC, ESL, and their active metabolite, S- licarbazepine (S-LC), were substrates of P-gp [71]. ESM and CBZ were not substrates of P-gp [58,71].
In the above experiments, different cell lines were used, but trans- port was always compared between cells overexpressing P-gp and cells not overexpressing P-gp. The MDCK and LLC cell lines have rela- tively low endogenous P-gp expression, yet can express relatively high levels of P-gp upon transfection. Cell monolayer assays are more laborious than uptake assays and should thus be employed judi- ciously. However, they provide the best data to support the P-gp sub- strate status of AEDs. A potential difficulty could occur when the parent cell lines of P-gp induced or transfected cell lines express a background level of P-gp, which can affect the assay. However, this problem can be removed by comparing transport by the P-gp overex- pressing cell lines with transport by the wild type cell lines.
The cell lines used above are limited in their ability to reproduce the drug-resistance mechanisms of the BBB. To partially overcome this limitation, primary brain microvascular endothelial cells isolated from brain tissue have been used. Using co-cultures of brain capillary endothelial and glial cells, Dehouck et al. developed an easy and re- producible method to study the BBB in vitro [103]. In the rat micro- vascular endothelial cell model, the transport of PB was significantly greater in the basal to apical direction than in the apical to basal di- rection, and CsA inhibited P-gp efflux [98]. In porcine brain capillary endothelial cells (pBCECs), CBZ significantly increased the intracellu- lar calcein concentration [9]. Using human microvascular endothelial cells (HBMECs), researchers found that the permeability of PHT was 10-fold less in BBB models using cells isolated from drug-resistant ep- ileptic patients than normal cells [104].
5.2. In vivo animal models
In vitro evaluations are usually followed by in vivo evaluations, whose results can be predicted to some extent from the in vitro ex- periments. In vivo experiments have more relation to the potential drug substrate status. Potschka et al. used a microdialysis method to test drug concentrations in the extracellular fluid (ECF) of the cere- bral cortex in rat brains. They found that the P-gp inhibitors, sodium
cyanide, verapamil, and PSC 833, can increase the ECF concentration of PHT, indicating that P-gp is involved in the BBB efflux of PHT to the plasma [67]. Then the AEDs, PB, LTG, and FBM, were studied in the same model; all were increased in the ECF by verapamil [69]. The concentration of CBZ in the ECF was also enhanced by verapamil and probenecid (an MRP inhibitor), indicating that P-gp and MRP are involved in the regulation of brain concentrations of CBZ [68]. How- ever, mdr1a/b(−/−) and wild-type mice did not exhibit significantly different brain concentrations of CBZ after its administration [100]. The ECF concentration of LVT was not regulated by verapamil and probenecid [105]. This may explain good efficacy of LVT in AED- refractory epilepsy.
The above results in animal models implicate P-gp in the BBB as playing an important role in the efflux of AEDs from the brain. How- ever, the animals were normal and thus do not represent the patho- logical conditions of epilepsy, especially refractory epilepsy. Animal epileptic models have the advantage of mimicking the pathological condition of epilepsy patients, and we can evaluate the association between the levels of P-gp and AEDs in brains. Brandt et al. used the temporal lobe epilepsy (TLE) rat model to test the effect of P-gp inhibitor on the distribution of AEDs in brain. In PB-resistant rats, P- gp was overexpressed in the hippocampus and other limbic brain re- gions compared with PB-responsive rats, and the anticonvulsant ac- tivity of PB was restored by TQD [66]. Thus, a P-gp inhibitor may counteract resistance to an AED in refractory epilepsy [106]. In brains of electrically induced epileptic rats, refractory rats had more P-gp and lower PHT concentration than did responsive rats. TQD increased the PHT concentration and efficacy in refractory rats [107]. Further- more, the pharmacokinetic profile of PHT was tested in brain and liver. The PHT concentration decreased 20–30% in the brain regions with P-gp overexpression, and the decrease was ameliorated by TQD. TQD increased PHT level in brain regions without P-gp overex- pression. However, the brain concentration of PHT was not affected by the overexpression of P-gp in the liver, indicating that P-gp only plays an efflux role in the BBB [108].
The P-gp knockout mice mdr1a−/− and mdr1a/1b(−/−) were also used to analyze the substrate status of AEDs. In mdr1a−/− mice, Schinkel et al. did not find a significant difference of [14C]-PHT level in brain compared with wild type mice, which was inconsistent with in vitro data [43,48,70,81]. Sills et al. measured the tissue distri- bution of eight drugs (PHT, PB, CBZ, VPA, LTG, VBG, GBP, and TPM) in mdr1a−/− mice and found that the brain/serum concentration ratios of CBZ, TPM, LTG, and GBP were higher in the knockout mice than the wildtype mice, indicating that these four drugs may be substrates of P-gp [109]. However, administration of CBZ did not result in signif- icantly different concentrations in brains of mdr1a/b(−/−) mice compared with wild type mice [100]. The results from knockout mice can be compared with those from P-gp transfected cells (LLC- MDR1 and MDCK-MDR1). Other transporters and enzymes may influ- ence the pharmacological results, and the confusing results need more investigation to explain. Mdr1a-deficient CF-1 mice were also used for detection of compounds that entered the CNS compartment [110,111]. These mice provide another model to evaluate the P-gp substrate status of AEDs.
5.3. In epilepsy patients
Some researchers analyzed the AED distributions in the brains of epilepsy patients to test whether AED concentrations were de- creased by the regional overexpression of P-gp. The ECF concentra- tions of several AEDs, such as CBZ, OXC, LTG and LVT, were significantly lower than their CSF concentrations in patients with in- tractable epilepsy [112]. However, it is not possible to judge whether these findings relate to overexpression of multidrug transporters in the brain because data from non-epileptic tissues were not included. 10,11-dihydro-10-hydroxy-5H-dibenzo(b,f)azepine-5-carboxamide
C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942 935

936 C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942
(10-OHCBZ) is one of the main metabolites of OXC. The 10-OHCBZ brain-plasma concentration ratio and the mRNA level of MDR1 in brain are inversely linearly correlated in epilepsy patients [97]. The distribution of [11C]-verapamil was tested in epileptogenic and nonepileptogenic brain regions of patients with drug resistant uni- lateral TLE by using PET. However, there was no significant differ- ence between epileptogenic and nonepileptogenic regions [113]. PET has been used successfully in measurement of drug concentra- tions in the CNS [65]. PET may be widely used in the future to direct- ly analyze the function of P-gp in transport of AEDs in humans. Nuclear magnetic resonance (NMR) could also be used as a tool to measure drug concentrations in humans. 11C, 31P, or 19F radiola- beled drugs can be detected, and this method is suitable for head measurements [114]. Although human studies have some disadvan- tages, including multiple transporters in many organs, the difficulty of getting suitable controls, and other unexpected influences, they are the most direct method to test the P-gp substrate status of AEDs in relation to pathology.
5.4. The substrate status of AEDs
The preponderance of the above evidence indicates that several AEDs may be substrates of P-gp, while several are likely not (Table 3). However, the evidence is often inconsistent among the different models used, and the cell culture models do not fully represent the reality in vivo. In order to define the P-gp substrate status of AEDs, we need to consider the reliability of the evidence. Since in vitro cell models and in vivo animal models each have advantages—of simplicity and of mim- icking pathological and physiological conditions, respectively—we give them the same weighting as evidence of substrate status. In vitro, differ- ent cell lines and detection methods have different reliability: transwell experiments > uptake of AEDs > influence of P-gp substrate markers by AEDs. Clinical evidence has the advantage that the samples are patients, not just models, but suffers from the disadvantages of lacking suitable controls and of lacking a method to test P-gp transporter function in iso- lation from the function of other transporters. Furthermore, clinical ev- idence indicating the substrate status of AEDs is limited. Therefore, we give clinical data less weight than cell and animal models.
We combined the evidence from humans, animal models (in vivo) and cell models (in vitro) to conclude whether each AED acts as a sub- strate for P-gp (Table 4). LTG, OXC, PB, and PHT are definite substrates of P-gp since they exhibited consistent results with both in vivo and in vitro experiments. LVT, AZD, S-LC, CBZ-E and ESL are probable sub- strates of P-gp. LVT was found to be a substrate in cell models and
in humans but not in an animal model. In the human study, because P-gp expression was not measured and a specific inhibitor of P-gp was not used, it is difficult to judge whether LVT was transported by P-gp [112]. The species differences of P-gp in animal models and humans may also contribute to the conflicting results. AZD, S-LC, CBZ-E and ESL were found to be substrates for P-gp in a cellular model [71,99]. CBZ, TPM and FBM are possible substrates of P-gp be- cause the in vivo evidence is positive although in vitro evidence is negative. Human data indicated that CBZ is a substrate, but rat models and most of the cellular evidence indicated that CBZ is not a substrate. However, the human study did not use a P-gp inhibitor or control tis- sue [112]. Rat models supported FBM as a substrate, but cell models did not. TPM was tested in cell models. TPM can not inhibit P-gp ef- flux function in LLC and OS2.4/Doxo cells with P-gp overexpression [9,95]. However, it can be transported from the basolateral to the api- cal side of P-gp transfected cell monolayer models [59]. VPA and GBP are possible substrates of P-gp since the in vitro evidence was conflict- ing. ESM, VGB and ZNS are unlikely to be substrates of P-gp because all available evidence was negative; however, for some drugs, only one report has been published, thus the possibility remains that fu- ture studies may provide contradictory data. P-gp is not the only transporter involved in AED efflux. The multidrug resistance trans- porters (MRPs), such as MRP1 and MRP2, are also involved in the ef- flux of AEDs [13,29,30,115]. Thus, more studies are needed to determine which AEDs are substrates for P-gp and whether they also act as substrates of other transporters.

6. Structure–activity relationship (SAR) between P-gp and AEDs
   The goal of studying the structure–activity relationship (SAR) is to find a means of predicting the substrates of P-gp. In order to find the common set of structural and functional features required by the in- teraction between P-gp and its substrates, many approaches have been attempted. However, SAR for P-gp is complicated, and only a few published papers exist. There are limited reports on the SAR be- tween the substrate action of P-gp and AEDs. This section will collate the current reports on the SAR of compounds as substrates for P-gp and attempt to analyze the SAR of AEDs for P-gp. After identifying the common structural features required of substrates for P-gp, we analyzed the structures of AEDs using multiple models.
   There are several models to predict P-gp substrates. Hundreds of compounds which have been tested as substrates of P-gp were ana- lyzed using different models to find their common shared structures. Most models produce the same minimal requirements of molecules
   Table 4
   AEDs as substrates of P-gp in research models.
   Substrate Drug Evidence for AEDs as P-gp substrates References
   status
   Patients Animals Cell lines
   Definite Definite Definite Definite Probable Probable Probable Probable Probable Possible Possible Possible Possible Possible Unlikely Unlikely Unlikely
   Phenytoin
   Phenobarbital
   Lamotrigine
   Oxcarbazepine
   Levetiracetam
   Acetazolamide
   S-licarbazepine Carbamazepine-10,11-epoxide Eslicarbazepine acetate Carbamazepine
   Topiramate Felbamate Sodium valproate Gabapentin Ethosuximide Vigabatrin Zonisamide
   – Y Y – Y Y Y Y Y Y – Y Y N Y – – Y – – Y – – Y – – Y Y ? N – – ? – Y N – – ? – – ? – – N – – N – – N
   [9,25,28,29,58,70,95,99,104,132] [9,29,58,69,95,98] [9,29,69,95,99,112]
   [71,112]
   [28,29,95,105,112] [99]
   [71]
   [71]
   [71] [9,28,68,71,95,99,100,112,132,133] [9,95,99,59]
   [69,95]
   [9,27]
   [9,95,99]
   [58,99]
   [99]
   [95]
   Y: yes; N: no; –: not reported; ?: conflicting evidence.

C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942 937
Table 5
Structures and electron donor groups in AEDs.
AED Structure
Aromatic rings
H-bond donors (>N−, –OH,
=NH, –NH−)
H-bond acceptors
(>N−, –OH, =O)
Type I
Type II
SAR prediction for P-gp substrate
Experimental evidence for P-gp substrate
In vitro evidence for P-gp substrate status
Phenytoin
Primidone
Ethosuximide
Phenobarbital
Acetazolamide
Valproic acid
Carbamazepine
Oxcarbazepine
Felbamate
Pregabalin
Vigabatrin Tiagabine
Lamotrigine
Topiramate
Zonisamide
Levetiracetam
2 2 –NH− 2=O
1 2 –NH− 2=O
0 1 –NH− 2=O
1 2 –NH− 3=O
0 1 –NH− 3=O 2>N−
0 0 1=O 1−OH
2 1>N− 1=O
2 1>N− 2=O
1 0 2=O
0 0 1=O 1−OH
0 0 1=O 1−OH
0 1>N− 1=O 1 –OH
1>N− 1 3>N− 3>N−
0 0 2=O
1 1>N− 2=O 1>N−
0 1>N− 2=O 1>N−
1 –
2 –
2 –
1 1
1 – 1 –
– –
– –
2 –
**
**
–
**
–
–
*
*
–
Definite
–
Unlikely
Definite
Probable
Possible
Possible
Definite
Possible
–
Unlikely
–
Definite
Possible
Unlikely
Probable
Y: MDCK, LLC, BMEC, Daoy/daoyar2.
N: Caco-2, BREC, OS2.4/Doxo
N/A
N: Caco-2
Y: LLC, MDCK, BMEC, OS2.4/Doxo N:Caco-2
Y: Caco-2
Y: LLC
N: MDCK
N; Caco-2, MDCK, LLC, BREC, OS2.4/
Doxo
Y: LLC, MDCK
N: OS2.4/Doxo
N/A
N: Caco-2 N/A
Y: LLC, OS2.4/Doxo N: Caco-2
N: Caco-2, LLC, OS2.4/Doxo
Y: LLC
N: OS2.4/Doxo
Y: LLC, OS2.4/Doxo N: MDCK
1––
1 1
1
2
–
1
– –
1
–
–
–
– –
**
–
*
–
(continued on next page)

938 C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942
Table 5 (continued)
AED Structure
Aromatic rings
H-bond donors (>N−, –OH,
=NH, –NH−)
H-bond acceptors
(>N−, –OH, =O)
Type I
Type II
SAR prediction for P-gp substrate
Experimental evidence for P-gp substrate
In vitro evidence for P-gp substrate status
Flunarizine
Gabapentin
Eslicarbazepine acetate
Lacosamide
Carbamazepine 10,11-epoxide
S-licarbazepine
3 2>N−
0 0
2 1>N−
1 2 –NH−
1 1>N−
2 1>N−
2>N−
1 =O 1 –OH
2=O
1=O
1=O
1=O 1 –OH
– – *
1 – –
1 – **
2 – **
– – *
– – *
–
Possible
Probable
–
Probable
Probable
N/A
N: Caco-2, LLC Y: OS2.4/Doxo
Y: LLC, MDCK
N/A
Y: LLC, MDCK
Y: LLC, MDCK
Y/N: yes/no; –: not reported; N/A: not available.
to act as the substrates of P-gp: one or two hydrophobic centers, an aromatic ring, electron donor groups (> C = O, –OH, > N−), and/or H-bond donor (–OH, > N−, = NH, –NH−) [33]. Seelig indicated that carbonyl groups are the most efficient electron donor group, followed by alkoxy and tertiary amino groups [32]. The more H-bond donors and acceptors the compound contains, the more likely it acts as a sub- strate for P-gp. AEDs meeting these requirements include PHT, PB, CBZ, OXC, primidone (PRM), FBM, ZNS, LTG, ESL, CBZ-E, S-LC, lacosa- mide (LCM) and flunarzine (FNR) (Table 5). Evidence indicates that PHT, PB, ESL, CBZ-E, S-LC, and LTG are P-gp substrates (Table 4). VGB and ESM, which do not meet the minimal requirements, were not found to act as P-gp substrates (Table 4). However, FBM, ZNS, and CBZ, which meet the requirements, were not found to be sub- strates of P-gp (Table 4). Meanwhile some AEDs, including AZD, TPM, VPA, LVT, and GBP, which could not meet the minimal require- ments, acted as substrates for P-gp in in vitro experiments. Some of the predicted results and experimental results are inconsistent be- cause the molecular models are still not accurate, and the different models may give contradictory results, while the experimental evi- dence also used different models.
The minimal requirements of H-bond donors and accepters were not sufficient to identify the P-gp substrate status of compounds. See- lig analyzed the spatial relationship between the electron donor groups in substrates. There are two types of electron donor patterns in the substrate molecules: type I and type II. The type I unit contains two electron donors, and the distance between them is 2.5 ± 0.3 Å. The type II unit contains two or three electron donors, and the dis- tance between the two outer groups is 4.6 ± 0.6 Å (Fig. 2). Com- pounds containing at least one type I or type II unit could be predicted as substrates for P-gp [32,116]. The more type I and type II units, the stronger the substrate. PHT, PB, PRM, LTG, ESL, and LCM,
satisfying the minimal requirements of H-bond donors, also contain at least one type I or type II unit (Table 5). PHT, PB, ESL, and LTG were tested and shown to be P-gp substrates (Table 4). AZD, TPM, and LVT contain type I units and act as P-gp substrates, although they do not satisfy the minimal requirements of being H-bond do- nors. OXC, CBZ-E, and S-LC do not contain any type I and type II units, although they do satisfy the minimal requirements of being an H-bond donor. GBP and VPA are in the opposite situation in that they do have type I or type II units but do not meet the minimal requirements of H-bond donors. They are all P-gp substrates (Table 5). ZNS and CBZ, which are not substrates of Pgp, satisfy the minimal requirements of being an H-bond donor and do not contain any type I and type II units (Table 5).
A molecular scaffold was analyzed to determine the activity of chemicals acting as substrates of P-gp [33]. Over 100 analogues were tested on MDCK-MDR1 cell monolayers by bidirectional trans- port assays to determine whether the scaffold increased or decreased P-gp transport efficiency. The scaffolds affecting P-gp transport effi- ciency are listed (Fig. 3). AZD, TPM, and ZNS contain the P-gp trans- port enhancing scaffold, which may explain, in addition to its pattern of electron donors, why AZD and TPM act as P-gp substrates. P-gp pump efficiency is also affected by passive diffusion. An increase of passive diffusion can decrease pump efficiency [33].
The following groups appear in P-gp substrates from most to least frequently: > C = O, –O−, -NR2, –NRH, –OH, –N=, R-halide, –S−, –NH2, and >(Phe)2. Among them, an ester group formed by an alkoxy group combined with a carbonyl group or a carboxy group is seen most often in functional units [32]. According to the SAR models, we analyzed the structures of AEDs in order to predict their substrate status. If an AED satisfies the minimal requirements of H-bond donors and acceptors, we labeled it with one star. For

Fig. 2. Type I and type II patterns of electron donors recognized by P-gp. Modified from Ref. [32].
drugs satisfying the minimal requirements, we added one more star for type I or type II electron donor patterns or a molecular scaffold enhancing P-gp transport efficiency. The more stars an AED has, the more likely it is a Pgp substrate.
Among the currently available AEDs, PRM, PGB, TGB, and FNR have been measured neither in vitro nor in vivo. Considering the above el- ements required by substrates for P-gp, we predicted their P-gp sub- strate status. PRM and LCM contain aromatic rings and type I electron
Fig. 3. Scaffold affecting P-gp pumping efficiency. Modified from Ref. [33].
donor patterns, and may thus act as substrates for P-gp. FNR does not contain type I or type II electron donor patterns. The big LogP of FNR causes high passive diffusion, which may decrease the possibility for FNR to act as a substrate. There have been few published reports on the substrate status of ZNS, VGB, or AZD, therefore further research is needed to determine their substrate status.
The predicted substrate status for some AEDs was not completely consistent with experimental results (Table 5). Further study is need- ed to gain more information on the SAR of P-gp. More specific experimental assays and predictive approaches will be helpful in un- derstanding SAR. Future success in predicting P-gp substrates and the SAR of AEDs will provide a more efficient, molecular approach for the design of new AEDs.

7. AEDs induce the overexpression of P-gp
   Evidence has indicated that repeated seizures may induce the overexpression of P-gp in the epileptic brain. However, epilepsy pa- tients, especially those with refractory epilepsy, take AEDs for a long time. Could AEDs influence the expression of P-gp?
   Giessmann et al. isolated tissues from the duodenum of humans before and after they took drugs. They found that CBZ significantly in- duced the mRNA of MDR1 and MRP2 and the protein of MRP2 but not MDR1 [34].
   Researchers used rats to test the effect of AEDs on P-gp expression. The P-gp level significantly increased in cerebral cortex and hippo- campus of rats fed with CBZ for 21 days, compared with controls. Meanwhile, the brain/plasma concentration ratios of Rho123 in the same regions decreased in the drug treated rat brains [117]. These data are consistent with the results from the human duodenum.
   PB and PHT have the same effect on the expression of P-gp [117]. In the rat liver, VPA also induced mdr1a at the mRNA level [35]. Wang et al. examined a model of pharmacoresistant TLE, rats kindled by Coriaria Lactone, and found that CBZ, VPA and PHT can induce the overexpression of P-gp [118]. In the same model, Wang-Yilz et al. obtained a similar result: that the overexpression of P-gp in brain tis- sues was induced by CBZ and VPA. They also found that TPM and LTG did not significantly induce the expression of P-gp, and they signifi- cantly inhibited seizures [119]. However, Seegers et al. reported con- flicting results. Normal rats were treated with PB and PHT over 11 days, and the expression of P-gp in endothelium and parenchyma of several brain regions was measured by immunohistochemistry. No significant increase of P-gp was seen in any brain regions after pro- longed treatment with PB or PHT, except that PB-treated rats dis- played a moderate increase in P-gp expression in the piriform/ parietal cortex and cerebellum [120].
   In vitro, some AEDs induce P-gp. For rat astrocyte cells that were cultured and treated with AEDs, the overexpression of P-gp was in- duced by PHT, PB, CBZ, and VPA in a dose- and time-dependent man- ner. Notably, P-gp was induced by therapeutic concentrations of PB and PHT at 30 days [121]. VPA induced the overexpression and total activity of P-gp up to 4-fold in rats and in the tumor cell lines SW620, KG1a and H4IIE [35]. In primary or immortalized rat brain microvascular endothelial cells, PB, PHT, CBZ, and VPA can induce P- gp overexpression [122,123]. In primary cultured rat brain microvas- cular endothelial cells, the drug efflux function of P-gp was induced by PB, PHT, CBZ, and VPA [122]. Thus, P-gp was induced by some of the AEDs, such as PB, PHT, CBZ, and VPA, but not by other AEDs, such as LVT, LTG, tiagabine and TPM. The mechanism underlying the phenomenon is still unclear, and further study is needed to confirm the results. Interestingly, VPA can induce the overexpression of P-gp [35,122] although it was reportedly not a substrate of P-gp [27]. Fur- thermore, long term treatment with LVT can cause tolerance [124] but does not induce P-gp [35,122]. Further study is needed to uncover the molecular mechanisms behind the induction of P-gp by AEDs.
   C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942 939

940 C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942

8. Proposed further research to determine substrate status of AEDs
   The data collected so far generally support some AEDs as sub- strates for P-gp, but the evidence is mixed or missing for several AEDs. Thus, the status of AEDs as substrates for P-gp needs further study before firm conclusions can be drawn. We propose the follow- ing research directions:
9. Investigation of the overexpression of P-gp in epilepsy patient brains, comparing the expression level between drug-resistant and drug-responsive patients.
10. Detecting association between the overexpression of P-gp and AED concentrations in refractory epilepsy brains (or the clinical outcome).
11. Detecting pharmacoresistance of AEDs in human brain capillary
    endothelial cells in vitro.
12. Using consistent epilepsy models, from patients and animals in
    vivo to cells in vitro, to systematically define the substrate status
    of all AEDs.
13. Using biochemical and molecular methods to study the interaction
    between AEDs and P-gp in order to determine the mechanism by
    which P-gp transport AEDs.
14. Optimizing the SAR models to predict the P-gp substrate status of
    AEDs.
15. Conclusions
    Worldwide, the number of epilepsy patients is large and growing. Although many AEDs are used in blocking seizures, about one-third of patients do not respond to these drugs. This resistance to AEDs has been recognized as one of the greatest challenges in epilepsy treat- ment today [125]. Studies report overexpression of P-gp in drug- resistant epilepsy brains. Data from some animal models confirmed the clinical results and indicated that overexpression of P-gp is asso- ciated with decreased AED concentration in the brain. This decrease can be reversed by P-gp inhibitors. In vitro cellular models overex- pressing P-gp are widely used to investigate whether (or which) AEDs are substrates for P-gp. Some studies indicated that several AEDs are substrates or inhibitors of P-gp. These data support the hy- potheses that AEDs are substrates for P-gp and that overexpressed P-gp pumps AEDs out of the brain, which causes drug-resistance, al- though these findings were not consistent in all studies. However, there are no established criteria to define the P-gp substrate status of AEDs. We combined in vivo and in vitro systems to develop criteria to clarify whether or which AEDs act as substrates for P-gp. SAR models are widely used in drug discovery and prediction of molecular interaction. Combining multiple models, we analyzed the association between the structures and P-gp substrate status of AEDs and pre- dicted the substrate status of AEDs whose status has not yet been reported. In conclusion, more study is required in order to determine the causes of refractory epilepsy.
    References
    [1] S.P. Aiken, W.M. Brown, Treatment of epilepsy: existing therapies and future de- velopments, Front. Biosci. 5 (2000) E124–E152.
    [2] W.A. Hauser, J.F. Annegers, L.T. Kurland, Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota: 1935–1984, Epilepsia 34 (1993) 453–468.
    [3] S.D. Shorvon, Epidemiology, classification, natural history, and genetics of epi-
    lepsy, Lancet 336 (1990) 93–96.
    [4] A.D. Everitt, J.W. Sander, Classification of the epilepsies: time for a change? A
    critical review of the International Classification of the Epilepsies and Epileptic Syndromes (ICEES) and its usefulness in clinical practice and epidemiological studies of epilepsy, Eur. Neurol. 42 (1999) 1–10.
    [5] A.T. Berg, S.F. Berkovic, M.J. Brodie, J. Buchhalter, J.H. Cross, W. van Emde Boas, J. Engel, J. French, T.A. Glauser, G.W. Mathern, S.L. Moshe, D. Nordli, P. Plouin, I.E. Scheffer, Revised terminology and concepts for organization of seizures and ep- ilepsies: report of the ILAE Commission on Classification and Terminology, 2005–2009, Epilepsia 51 (2010) 676–685.
    [6] P. Kwan, M.J. Brodie, Epilepsy after the first drug fails: substitution or add-on? Seizure 9 (2000) 464–468.
    [7] G. Regesta, P. Tanganelli, Clinical aspects and biological bases of drug-resistant epilepsies, Epilepsy Res. 34 (1999) 109–122.
    [8] P. Kwan, A. Arzimanoglou, A.T. Berg, M.J. Brodie, W. Allen Hauser, G. Mathern, S.L. Moshe, E. Perucca, S. Wiebe, J. French, Definition of drug resistant epilepsy: consensus proposal by the ad hoc Task Force of the ILAE Commission on Thera- peutic Strategies, Epilepsia 51 (2010) 1069–1077.
    [9] J. Weiss, C.J. Kerpen, H. Lindenmaier, S.M. Dormann, W.E. Haefeli, Interaction of antiepileptic drugs with human P-glycoprotein in vitro, J. Pharmacol. Exp. Ther. 307 (2003) 262–267.
    [10] P. Ballabh, A. Braun, M. Nedergaard, The blood–brain barrier: an overview: structure, regulation, and clinical implications, Neurobiol. Dis. 16 (2004) 1–13. [11] W.M.Pardridge,Blood–brainbarrierdelivery,DrugDiscov.Today12(2007)54–61.
    [12] A.K. Ghose, V.N. Viswanadhan, J.J. Wendoloski, A knowledge-based approach in designing combinatorial or medicinal chemistry libraries for drug discovery. 1. A qualitative and quantitative characterization of known drug databases, J. Comb. Chem. 1 (1999) 55–68.
    [13] W. Loscher, H. Potschka, Drug resistance in brain diseases and the role of drug efflux transporters, Nat. Rev. Neurosci. 6 (2005) 591–602.
    [14] W. Loscher, G.J. Sills, Drug resistance in epilepsy: why is a simple explanation not enough? Epilepsia 48 (2007) 2370–2372.
    [15] R.L.Juliano,V.Ling,AsurfaceglycoproteinmodulatingdrugpermeabilityinChi- nese hamster ovary cell mutants, Biochim. Biophys. Acta 455 (1976) 152–162.
    [16] M.M. Gottesman, I. Pastan, The multidrug transporter, a double-edged sword, J. Biol. Chem. 263 (1988) 12163–12166.
    [17] M.Hennessy,J.P.Spiers,AprimeronthemechanicsofP-glycoproteinthemulti- drug transporter, Pharmacol. Res. 55 (2007) 1–15.
    [18] N. Mizuno, T. Niwa, Y. Yotsumoto, Y. Sugiyama, Impact of drug transporter stud- ies on drug discovery and development, Pharmacol. Rev. 55 (2003) 425–461.
    [19] E.M. Kemper, A.E. van Zandbergen, C. Cleypool, H.A. Mos, W. Boogerd, J.H. Beij- nen, O. van Tellingen, Increased penetration of paclitaxel into the brain by inhi- bition of P-glycoprotein, Clin. Cancer Res. 9 (2003) 2849–2855.
    [20] E.M. Kemper, M. Verheij, W. Boogerd, J.H. Beijnen, O. van Tellingen, Improved penetration of docetaxel into the brain by co-administration of inhibitors of P- glycoprotein, Eur. J. Cancer 40 (2004) 1269–1274.
    [21] S. Fellner, B. Bauer, D.S. Miller, M. Schaffrik, M. Fankhanel, T. Spruss, G. Bern- hardt, C. Graeff, L. Farber, H. Gschaidmeier, A. Buschauer, G. Fricker, Transport of paclitaxel (Taxol) across the blood–brain barrier in vitro and in vivo, J. Clin. In- vest. 110 (2002) 1309–1318.
    [22] W. Loscher, Drug transporters in the epileptic brain, Epilepsia 48 (Suppl. 1) (2007) 8–13.
    [23] E.Aronica,J.A.Gorter,G.H.Jansen,C.W.vanVeelen,P.C.vanRijen,S.Leenstra,M. Ramkema, G.L. Scheffer, R.J. Scheper, D. Troost, Expression and cellular distribu- tion of multidrug transporter proteins in two major causes of medically intrac- table epilepsy: focal cortical dysplasia and glioneuronal tumors, Neuroscience 118 (2003) 417–429.
    [24] E.Aronica,J.A.Gorter,M.Ramkema,S.Redeker,F.Ozbas-Gerceker,E.A.vanVliet, G.L. Scheffer, R.J. Scheper, P. van der Valk, J.C. Baayen, D. Troost, Expression and cellular distribution of multidrug resistance-related proteins in the hippocampus of patients with mesial temporal lobe epilepsy, Epilepsia 45 (2004) 441–451.
    [25] D.M.Tishler,K.I.Weinberg,D.R.Hinton,N.Barbaro,G.M.Annett,C.Raffel,MDR1 gene expression in brain of patients with medically intractable epilepsy, Epilep- sia 36 (1995) 1–6.
    [26] S.M. Dombrowski, S.Y. Desai, M. Marroni, L. Cucullo, K. Goodrich, W. Bingaman, M.R. Mayberg, L. Bengez, D. Janigro, Overexpression of multiple drug resistance genes in endothelial cells from patients with refractory epilepsy, Epilepsia 42 (2001) 1501–1506.
    [27] S.Baltes,M.Fedrowitz,C.L.Tortos,H.Potschka,W.Loscher,Valproicacidisnota substrate for P-glycoprotein or multidrug resistance proteins 1 and 2 in a num- ber of in vitro and in vivo transport assays, J. Pharmacol. Exp. Ther. 320 (2007) 331–343.
    [28] S.Baltes,A.M.Gastens,M.Fedrowitz,H.Potschka,V.Kaever,W.Loscher,Differ- ences in the transport of the antiepileptic drugs phenytoin, levetiracetam and carbamazepine by human and mouse P-glycoprotein, Neuropharmacology 52 (2007) 333–346.
    [29] C. Luna-Tortos, M. Fedrowitz, W. Loscher, Several major antiepileptic drugs are substrates for human P-glycoprotein, Neuropharmacology 55 (2008) 1364–1375. [30] P. Kwan, M.J. Brodie, Potential role of drug transporters in the pathogenesis of
    medically intractable epilepsy, Epilepsia 46 (2005) 224–235.
    [31] T.R. Stouch, O. Gudmundsson, Progress in understanding the structure–activity
    relationships of P-glycoprotein, Adv. Drug Deliv. Rev. 54 (2002) 315–328.
    [32] A. Seelig, A general pattern for substrate recognition by P-glycoprotein, Eur. J.
    Biochem. 251 (1998) 252–261.
    [33] T.J. Raub, P-glycoprotein recognition of substrates and circumvention through
    rational drug design, Mol. Pharm. 3 (2006) 3–25.
    [34] T. Giessmann, K. May, C. Modess, D. Wegner, U. Hecker, M. Zschiesche, P. Dazert,
    M. Grube, E. Schroeder, R. Warzok, I. Cascorbi, H.K. Kroemer, W. Siegmund, Car- bamazepine regulates intestinal P-glycoprotein and multidrug resistance pro- tein MRP2 and influences disposition of talinolol in humans, Clin. Pharmacol. Ther. 76 (2004) 192–200.
    [35] S. Eyal, J.G. Lamb, M. Smith-Yockman, B. Yagen, E. Fibach, Y. Altschuler, H.S. White, M. Bialer, The antiepileptic and anticancer agent, valproic acid, induces P-glycoprotein in human tumour cell lines and in rat liver, Br. J. Pharmacol. 149 (2006) 250–260.
    [36] D.F. Callen, E. Baker, R.N. Simmers, R. Seshadri, I.B. Roninson, Localization of the human multiple drug resistance gene, MDR1, to 7q21.1, Hum. Genet. 77 (1987) 142–144.

[37] J.A. Silverman, Multidrug-resistance transporters, Pharm. Biotechnol. 12 (1999) 353–386.
[38] P. Borst, Multidrug resistant proteins, Semin. Cancer Biol. 8 (1997) 131–134.
[39] P. Gros, J. Croop, D. Housman, Mammalian multidrug resistance gene: complete cDNA sequence indicates strong homology to bacterial transport proteins, Cell
47 (1986) 371–380.
[40] C.J. Chen, J.E. Chin, K. Ueda, D.P. Clark, I. Pastan, M.M. Gottesman, I.B. Roninson,
Internal duplication and homology with bacterial transport proteins in the mdr1 (P-glycoprotein) gene from multidrug-resistant human cells, Cell 47 (1986) 381–389.
[41] R.B. Kim, M.F. Fromm, C. Wandel, B. Leake, A.J. Wood, D.M. Roden, G.R. Wilkin- son, The drug transporter P-glycoprotein limits oral absorption and brain entry of HIV-1 protease inhibitors, J. Clin. Invest. 101 (1998) 289–294.
[42] A.T. Fojo, K. Ueda, D.J. Slamon, D.G. Poplack, M.M. Gottesman, I. Pastan, Expres- sion of a multidrug-resistance gene in human tumors and tissues, Proc. Natl. Acad. Sci. U. S. A. 84 (1987) 265–269.
[43] I. Sugawara, I. Kataoka, Y. Morishita, H. Hamada, T. Tsuruo, S. Itoyama, S. Mori, Tissue distribution of P-glycoprotein encoded by a multidrug-resistant gene as revealed by a monoclonal antibody, MRK 16, Cancer Res. 48 (1988) 1926–1929.
[44] C. Cordon-Cardo, J.P. O’Brien, J. Boccia, D. Casals, J.R. Bertino, M.R. Melamed, Ex- pression of the multidrug resistance gene product (P-glycoprotein) in human normal and tumor tissues, J. Histochem. Cytochem. 38 (1990) 1277–1287.
[45] C. Cordon-Cardo, J.P. O’Brien, D. Casals, L. Rittman-Grauer, J.L. Biedler, M.R. Mel- amed, J.R. Bertino, Multidrug-resistance gene (P-glycoprotein) is expressed by endothelial cells at blood–brain barrier sites, Proc. Natl. Acad. Sci. U. S. A. 86 (1989) 695–698.
[46] E. Beaulieu, M. Demeule, L. Ghitescu, R. Beliveau, P-glycoprotein is strongly expressed in the luminal membranes of the endothelium of blood vessels in the brain, Biochem. J. 326 (Pt 2) (1997) 539–544.
[47] G. Lee, S. Dallas, M. Hong, R. Bendayan, Drug transporters in the central nervous system: brain barriers and brain parenchyma considerations, Pharmacol. Rev. 53 (2001) 569–596.
[48] A.H. Schinkel, The physiological function of drug-transporting P-glycoproteins, Semin. Cancer Biol. 8 (1997) 161–170.
[49] R. Bendayan, G. Lee, M. Bendayan, Functional expression and localization of P- glycoprotein at the blood brain barrier, Microsc. Res. Tech. 57 (2002) 365–380.
[50] M. Demeule, A. Regina, J. Jodoin, A. Laplante, C. Dagenais, F. Berthelet, A. Mogh- rabi, R. Beliveau, Drug transport to the brain: key roles for the efflux pump P-
glycoprotein in the blood–brain barrier, Vasc. Pharmacol. 38 (2002) 339–348.
[51] W.M. Pardridge, P.L. Golden, Y.S. Kang, U. Bickel, Brain microvascular and astro-
cyte localization of P-glycoprotein, J. Neurochem. 68 (1997) 1278–1285.
[52] D. Schmidt, W. Loscher, Drug resistance in epilepsy: putative neurobiologic and
clinical mechanisms, Epilepsia 46 (2005) 858–877.
[53] D.J. Begley, ABC transporters and the blood–brain barrier, Curr. Pharm. Des. 10
(2004) 1295–1312.
[54] Y. Kimura, S.Y. Morita, M. Matsuo, K. Ueda, Mechanism of multidrug recognition
by MDR1/ABCB1, Cancer Sci. 98 (2007) 1303–1310.
[55] J.M. Ford, J.M. Yang, W.N. Hait, P-glycoprotein-mediated multidrug resistance:
experimental and clinical strategies for its reversal, Cancer Treat. Res. 87
(1996) 3–38.
[56] J.M. Ford, W.N. Hait, Pharmacologic circumvention of multidrug resistance, Cy-
totechnology 12 (1993) 171–212.
[57] C. Avendano, J.C. Menendez, Inhibitors of multidrug resistance to antitumor
agents (MDR), Curr. Med. Chem. 9 (2002) 159–193.
[58] C. Zhang, P. Kwan, Z. Zuo, L. Baum, In vitro concentration dependent transport of
phenytoin and phenobarbital, but not ethosuximide, by human P-glycoprotein,
Life Sci. 86 (2010) 899–905.
[59] C. Luna-Tortos, B. Rambeck, U.H. Jurgens, W. Loscher, The antiepileptic drug
topiramate is a substrate for human P-glycoprotein but not multidrug resistance
proteins, Pharm. Res. 26 (2009) 2464–2470.
[60] C.C. Hung, C.C. Chen, C.J. Lin, H.H. Liou, Functional evaluation of polymorphisms
in the human ABCB1 gene and the impact on clinical responses of antiepileptic
drugs, Pharmacogenet. Genomics 18 (2008) 390–402.
[61] P. Kwan, L. Baum, V. Wong, P.W. Ng, C.H. Lui, N.C. Sin, A.C. Hui, E. Yu, L.K. Wong,
Association between ABCB1 C3435T polymorphism and drug-resistant epilepsy
in Han Chinese, Epilepsy Behav. 11 (2007) 112–117.
[62] A. Lazarowski, G. Sevlever, A. Taratuto, M. Massaro, A. Rabinowicz, Tuberous
sclerosis associated with MDR1 gene expression and drug-resistant epilepsy,
Pediatr. Neurol. 21 (1999) 731–734.
[63] H. Bartmann, C. Fuest, C. la Fougere, G. Xiong, T. Just, J. Schlichtiger, P. Winter, G.
Boning, B. Wangler, A. Pekcec, J. Soerensen, P. Bartenstein, P. Cumming, H. Potschka, Imaging of P-glycoprotein-mediated pharmacoresistance in the hip- pocampus: proof-of-concept in a chronic rat model of temporal lobe epilepsy, Epilepsia 51 (2010) 1780–1790.
[64] M. Brunner, O. Langer, R. Sunder-Plassmann, G. Dobrozemsky, U. Muller, W. Wadsak, A. Krcal, R. Karch, C. Mannhalter, R. Dudczak, K. Kletter, I. Steiner, C. Baumgartner, M. Muller, Influence of functional haplotypes in the drug trans- porter gene ABCB1 on central nervous system drug distribution in humans, Clin. Pharmacol. Ther. 78 (2005) 182–190.
[65] A. Takano, H. Kusuhara, T. Suhara, I. Ieiri, T. Morimoto, Y.J. Lee, J. Maeda, Y. Ikoma, H. Ito, K. Suzuki, Y. Sugiyama, Evaluation of in vivo P-glycoprotein func- tion at the blood–brain barrier among MDR1 gene polymorphisms by using 11C-verapamil, J. Nucl. Med. 47 (2006) 1427–1433.
[66] H.A. Volk, W. Loscher, Multidrug resistance in epilepsy: rats with drug-resistant seizures exhibit enhanced brain expression of P-glycoprotein compared with rats with drug-responsive seizures, Brain 128 (2005) 1358–1368.
[67] H. Potschka, W. Loscher, In vivo evidence for P-glycoprotein-mediated transport of phenytoin at the blood–brain barrier of rats, Epilepsia 42 (2001) 1231–1240. [68] H. Potschka, M. Fedrowitz, W. Loscher, P-glycoprotein and multidrug resistance- associated protein are involved in the regulation of extracellular levels of the major antiepileptic drug carbamazepine in the brain, Neuroreport 12 (2001)
3557–3560.
[69] H. Potschka, M. Fedrowitz, W. Loscher, P-glycoprotein-mediated efflux of phe-
nobarbital, lamotrigine, and felbamate at the blood–brain barrier: evidence
from microdialysis experiments in rats, Neurosci. Lett. 327 (2002) 173–176. [70] A.H. Schinkel, E. Wagenaar, C.A. Mol, L. van Deemter, P-glycoprotein in the blood–brain barrier of mice influences the brain penetration and pharmacolog-
ical activity of many drugs, J. Clin. Invest. 97 (1996) 2517–2524.
[71] C. Zhang, Z. Zuo, P. Kwan, L. Baum, In vitro transport profile of carbamazepine, oxcarbazepine, eslicarbazepine acetate, and their active metabolites by human
P-glycoprotein, Epilepsia 52 (2011) 1894–1904.
[72] C. Kimchi-Sarfaty, J.M. Oh, I.W. Kim, Z.E. Sauna, A.M. Calcagno, S.V. Ambudkar,
M.M. Gottesman, A “silent” polymorphism in the MDR1 gene changes substrate
specificity, Science 315 (2007) 525–528.
[73] X.B. Chang, A molecular understanding of ATP-dependent solute transport by
multidrug resistance-associated protein MRP1, Cancer Metastasis Rev. 26
(2007) 15–37.
[74] S.V. Ambudkar, I.W. Kim, Z.E. Sauna, The power of the pump: mechanisms of ac-
tion of P-glycoprotein (ABCB1), Eur. J. Pharm. Sci. 27 (2006) 392–400.
[75] G.D. Leschziner, T. Andrew, M. Pirmohamed, M.R. Johnson, ABCB1 genotype and PGP expression, function and therapeutic drug response: a critical review and recommendations for future research, Pharmacogenomics J. 7 (2007)
154–179.
[76] A. Frelet, M. Klein, Insight in eukaryotic ABC transporter function by mutation
analysis, FEBS Lett. 580 (2006) 1064–1084.
[77] K.M. Giacomini, S.M. Huang, D.J. Tweedie, L.Z. Benet, K.L. Brouwer, X. Chu, A.
Dahlin, R. Evers, V. Fischer, K.M. Hillgren, K.A. Hoffmaster, T. Ishikawa, D. Kep- pler, R.B. Kim, C.A. Lee, M. Niemi, J.W. Polli, Y. Sugiyama, P.W. Swaan, J.A. Ware, S.H. Wright, S.W. Yee, M.J. Zamek-Gliszczynski, L. Zhang, Membrane transporters in drug development, Nat. Rev. Drug Discov. 9 (2010) 215–236.
[78] L. Zhang, Y.D. Zhang, J.M. Strong, K.S. Reynolds, S.M. Huang, A regulatory view- point on transporter-based drug interactions, Xenobiotica 38 (2008) 709–724.
[79] R.W. Robey, A. Lazarowski, S.E. Bates, P-glycoprotein—a clinical target in drug- refractory epilepsy? Mol. Pharmacol. 73 (2008) 1343–1346.
[80] P. Kwan, H.M. Li, E. Al-Jufairi, R. Abdulla, M. Gonzales, A.H. Kaye, C. Szoeke, H.K. Ng, K.S. Wong, T.J. O’Brien, Association between temporal lobe P-glycoprotein expression and seizure recurrence after surgery for pharmacoresistant temporal lobe epilepsy, Neurobiol. Dis. 39 (2010) 192–197.
[81] S.M. Sisodiya, J. Heffernan, M.V. Squier, Over-expression of P-glycoprotein in malformations of cortical development, Neuroreport 10 (1999) 3437–3441.
[82] S. Jozwiak, Contemporary opinions on classification, pathogenesis and treat- ment of drug-resistant epilepsy, Wiad. Lek. 60 (2007) 258–264.
[83] N. Marchi, K.L. Hallene, K.M. Kight, L. Cucullo, G. Moddel, W. Bingaman, G. Dini, A. Vezzani, D. Janigro, Significance of MDR1 and multiple drug resistance in re- fractory human epileptic brain, BMC Med. 2 (2004) 37.
[84] H.A. Volk, K. Burkhardt, H. Potschka, J. Chen, A. Becker, W. Loscher, Neuronal ex- pression of the drug efflux transporter P-glycoprotein in the rat hippocampus after limbic seizures, Neuroscience 123 (2004) 751–759.
[85] N. Marchi, G. Guiso, S. Caccia, M. Rizzi, B. Gagliardi, F. Noe, T. Ravizza, S. Bassa- nini, S. Chimenti, G. Battaglia, A. Vezzani, Determinants of drug brain uptake in a rat model of seizure-associated malformations of cortical development, Neu- robiol. Dis. 24 (2006) 429–442.
[86] X. Liu, Z. Yang, J. Yang, H. Yang, Increased P-glycoprotein expression and de- creased phenobarbital distribution in the brain of pentylenetetrazole-kindled rats, Neuropharmacology 53 (2007) 657–663.
[87] U. Seegers, H. Potschka, W. Loscher, Transient increase of P-glycoprotein expres- sion in endothelium and parenchyma of limbic brain regions in the kainate model of temporal lobe epilepsy, Epilepsy Res. 51 (2002) 257–268.
[88] E. van Vliet, E. Aronica, S. Redeker, N. Marchi, M. Rizzi, A. Vezzani, J. Gorter, Se- lective and persistent upregulation of mdr1b mRNA and P-glycoprotein in the parahippocampal cortex of chronic epileptic rats, Epilepsy Res. 60 (2004) 203–213.
[89] A. Nishimura, N. Honda, N. Sugioka, K. Takada, N. Shibata, Evaluation of carba- mazepine pharmacokinetic profiles in mice with kainic acid-induced acute sei- zures, Biol. Pharm. Bull. 31 (2008) 2302–2308.
[90] U. Seegers, H. Potschka, W. Loscher, Expression of the multidrug transporter P- glycoprotein in brain capillary endothelial cells and brain parenchyma of amygdala-kindled rats, Epilepsia 43 (2002) 675–684.
[91] Z.Y. Xiao, Yong, Xuefeng Wang, Development of PHT-PB-resistant amygdala- kindled rats and expression of MDR1, Zhonghua Shenjingke Zazhi 32 (6) (1999) 365–368.
[92] H. Potschka, H.A. Volk, W. Loscher, Pharmacoresistance and expression of multi- drug transporter P-glycoprotein in kindled rats, Neuroreport 15 (2004) 1657–1661.
[93] M. Rizzi, S. Caccia, G. Guiso, C. Richichi, J.A. Gorter, E. Aronica, M. Aliprandi, R. Bagnati, R. Fanelli, M. D’Incalci, R. Samanin, A. Vezzani, Limbic seizures induce P-glycoprotein in rodent brain: functional implications for pharmacoresistance, J. Neurosci. 22 (2002) 5833–5839.
[94] A.H. Schinkel, J.J. Smit, O. van Tellingen, J.H. Beijnen, E. Wagenaar, L. van Deemter, C.A. Mol, M.A. van der Valk, E.C. Robanus-Maandag, H.P. te Riele, et al., Disruption of the mouse mdr1a P-glycoprotein gene leads to a deficiency in the blood–brain barrier and to increased sensitivity to drugs, Cell 77 (1994) 491–502.
C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942 941

942 C. Zhang et al. / Advanced Drug Delivery Reviews 64 (2012) 930–942
[95] C.L. West, K.L. Mealey, Assessment of antiepileptic drugs as substrates for canine P-glycoprotein, Am. J. Vet. Res. 68 (2007) 1106–1110.
[96] S.V. Ambudkar, C. Kimchi-Sarfaty, Z.E. Sauna, M.M. Gottesman, P-glycoprotein: from genomics to mechanism, Oncogene 22 (2003) 7468–7485.
[97] N. Marchi, G. Guiso, M. Rizzi, S. Pirker, K. Novak, T. Czech, C. Baumgartner, D. Janigro, S. Caccia, A. Vezzani, A pilot study on brain-to-plasma partition of 10,11-dyhydro-10-hydroxy-5H-dibenzo(b,f)azepine-5-carboxamide and MDR1 brain expression in epilepsy patients not responding to oxcarbazepine, Epilepsia 46 (2005) 1613–1619.
[98] Z.H. Yang, X.D. Liu, P-glycoprotein-mediated efflux of phenobarbital at the blood–brain barrier evidence from transport experiments in vitro, Epilepsy Res. 78 (2008) 40–49.
[99] A. Crowe, Y.K. Teoh, Limited P-glycoprotein mediated efflux for anti-epileptic drugs, J. Drug Target 14 (2006) 291–300.
[100] A. Owen, M. Pirmohamed, J.N. Tettey, P. Morgan, D. Chadwick, B.K. Park, Carba- mazepine is not a substrate for P-glycoprotein, Br. J. Clin. Pharmacol. 51 (2001) 345–349.
[101] C. Zhang, L. Baum, Z. Zuo, P. Kwan, Transport of antiepileptic drugs by P-glycopro- tein in cell monolayer models, Am. Epilepsy Soc. Ann. Meet. (2009) abst. 2.201.
[102] W. Loscher, C. Luna-Tortos, K. Romermann, M. Fedrowitz, Do ATP-binding cas- sette transporters cause pharmacoresistance in epilepsy? Problems and ap- proaches in determining which antiepileptic drugs are affected, Curr. Pharm. Des. 17 (2011) 2808–2828.
[103] M.P. Dehouck, S. Meresse, P. Delorme, J.C. Fruchart, R. Cecchelli, An easier, repro- ducible, and mass-production method to study the blood–brain barrier in vitro, J. Neurochem. 54 (1990) 1798–1801.
[104] L. Cucullo, M. Hossain, E. Rapp, T. Manders, N. Marchi, D. Janigro, Development of a humanized in vitro blood–brain barrier model to screen for brain penetra- tion of antiepileptic drugs, Epilepsia 48 (2007) 505–516.
[105] H. Potschka, S. Baltes, W. Loscher, Inhibition of multidrug transporters by verap- amil or probenecid does not alter blood–brain barrier penetration of levetirace- tam in rats, Epilepsy Res. 58 (2004) 85–91.
[106] C. Brandt, K. Bethmann, A.M. Gastens, W. Loscher, The multidrug transporter hy- pothesis of drug resistance in epilepsy: proof-of-principle in a rat model of tem- poral lobe epilepsy, Neurobiol. Dis. 24 (2006) 202–211.
[107] E.A. van Vliet, R. van Schaik, P.M. Edelbroek, S. Redeker, E. Aronica, W.J. Wad- man, N. Marchi, A. Vezzani, J.A. Gorter, Inhibition of the multidrug transporter P-glycoprotein improves seizure control in phenytoin-treated chronic epileptic rats, Epilepsia 47 (2006) 672–680.
[108] E.A. van Vliet, R. van Schaik, P.M. Edelbroek, R.A. Voskuyl, S. Redeker, E. Aronica, W.J. Wadman, J.A. Gorter, Region-specific overexpression of P-glycoprotein at the blood–brain barrier affects brain uptake of phenytoin in epileptic rats, J. Pharmacol. Exp. Ther. 322 (2007) 141–147.
[109] G.J. Sills, P. Kwan, E. Butler, E.C. de Lange, D.J. van den Berg, M.J. Brodie, P-glyco- protein-mediated efflux of antiepileptic drugs: preliminary studies in mdr1a knockout mice, Epilepsy Behav. 3 (2002) 427–432.
[110] M. Yamazaki, W.E. Neway, T. Ohe, I. Chen, J.F. Rowe, J.H. Hochman, M. Chiba, J.H. Lin, In vitro substrate identification studies for p-glycoprotein-mediated trans- port: species difference and predictability of in vivo results, J. Pharmacol. Exp. Ther. 296 (2001) 723–735.
[111] D.F. Tang-Wai, S. Kajiji, F. DiCapua, D. de Graaf, I.B. Roninson, P. Gros, Human (MDR1) and mouse (mdr1, mdr3) P-glycoproteins can be distinguished by their respective drug resistance profiles and sensitivity to modulators, Biochem- istry 34 (1995) 32–39.
[112] B. Rambeck, U.H. Jurgens, T.W. May, H.W. Pannek, F. Behne, A. Ebner, A. Gorji, H. Straub, E.J. Speckmann, B. Pohlmann-Eden, W. Loscher, Comparison of brain ex- tracellular fluid, brain tissue, cerebrospinal fluid, and serum concentrations of antiepileptic drugs measured intraoperatively in patients with intractable epi- lepsy, Epilepsia 47 (2006) 681–694.
[113] O. Langer, M. Bauer, A. Hammers, R. Karch, E. Pataraia, M.J. Koepp, A. Abrahim, G. Luurtsema, M. Brunner, R. Sunder-Plassmann, F. Zimprich, C. Joukhadar, S. Gentzsch, R. Dudczak, K. Kletter, M. Muller, C. Baumgartner, Pharmacoresistance in epilepsy: a pilot PET study with the P-glycoprotein substrate R-[(11)C]verap- amil, Epilepsia 48 (2007) 1774–1784.
[114] [115]
[116] [117]
[118]
[119] [120] [121] [122]
[123] [124] [125]
[126] [127]
[128] [129]
[130] [131] [132]
[133]
A. Aszalos, Drug-drug interactions affected by the transporter protein, P- glycoprotein (ABCB1, MDR1) I. Preclinical aspects, Drug Discov. Today 12 (2007) 833–837.
K. Hoffmann, W. Loscher, Upregulation of brain expression of P-glycoprotein in MRP2-deficient TR(−) rats resembles seizure-induced up-regulation of this drug efflux transporter in normal rats, Epilepsia 48 (2007) 631–645.
K. Ueda, Y. Taguchi, M. Morishima, How does P-glycoprotein recognize its sub- strates? Semin. Cancer Biol. 8 (1997) 151–159.
T. Wen, Y.C. Liu, H.W. Yang, H.Y. Liu, X.D. Liu, G.J. Wang, L. Xie, Effect of 21-day exposure of phenobarbital, carbamazepine and phenytoin on P-glycoprotein ex- pression and activity in the rat brain, J. Neurol. Sci. 270 (2008) 99–106.
Y. Wang, D. Zhou, B. Wang, H. Li, H. Chai, Q. Zhou, S. Zhang, H. Stefan, A kindling model of pharmacoresistant temporal lobe epilepsy in Sprague–Dawley rats in- duced by Coriaria lactone and its possible mechanism, Epilepsia 44 (2003) 475–488.
Y. Wang-Tilz, C. Tilz, B. Wang, G.P. Tilz, H. Stefan, Influence of lamotrigine and topiramate on MDR1 expression in difficult-to-treat temporal lobe epilepsy, Epi- lepsia 47 (2006) 233–239.
U. Seegers, H. Potschka, W. Loscher, Lack of effects of prolonged treatment with phenobarbital or phenytoin on the expression of P-glycoprotein in various rat brain regions, Eur. J. Pharmacol. 451 (2002) 149–155.
Y. Lu, Y. Yan, X.F. Wang, Antiepileptic drug-induced multidrug resistance P- glycoprotein overexpression in astrocytes cultured from rat brains, Chin. Med. J. (Engl) 117 (2004) 1682–1686.
H.W. Yang, H.Y. Liu, X. Liu, D.M. Zhang, Y.C. Liu, X.D. Liu, G.J. Wang, L. Xie, In- creased P-glycoprotein function and level after long-term exposure of four anti- epileptic drugs to rat brain microvascular endothelial cells in vitro, Neurosci. Lett. 434 (2008) 299–303.
L. Lombardo, R. Pellitteri, M. Balazy, V. Cardile, Induction of nuclear receptors and drug resistance in the brain microvascular endothelial cells treated with antiepileptic drugs, Curr. Neurovasc. Res. 5 (2008) 82–92.
E.A. van Vliet, R. van Schaik, P.M. Edelbroek, F.H. da Silva, W.J. Wadman, J.A. Gor- ter, Development of tolerance to levetiracetam in rats with chronic epilepsy, Epilepsia 49 (2008) 1151–1159.
J.P. Stables, E.H. Bertram, H.S. White, D.A. Coulter, M.A. Dichter, M.P. Jacobs, W. Loscher, D.H. Lowenstein, S.L. Moshe, J.L. Noebels, M. Davis, Models for epilepsy and epileptogenesis: report from the NIH workshop, Bethesda, Maryland, Epi- lepsia 43 (2002) 1410–1420.
H. Ak, B. Ay, T. Tanriverdi, G.Z. Sanus, M. Is, M. Sar, B. Oz, C. Ozkara, E. Ozyurt, M. Uzan, Expression and cellular distribution of multidrug resistance-related pro- teins in patients with focal cortical dysplasia, Seizure 16 (2007) 493–503.
S.M. Sisodiya, W.R. Lin, B.N. Harding, M.V. Squier, M. Thom, Drug resistance in epilepsy: expression of drug resistance proteins in common causes of refractory epilepsy, Brain 125 (2002) 22–31.
S.M. Sisodiya, W.R. Lin, M.V. Squier, M. Thom, Multidrug-resistance protein 1 in focal cortical dysplasia, Lancet 357 (2001) 42–43.
H. Kubota, H. Ishihara, T. Langmann, G. Schmitz, B. Stieger, H.G. Wieser, Y. Yone- kawa, K. Frei, Distribution and functional activity of P-glycoprotein and multi- drug resistance-associated proteins in human brain microvascular endothelial cells in hippocampal sclerosis, Epilepsy Res. 68 (2006) 213–228.
A. Lazarowski, F. Lubieniecki, S. Camarero, H. Pomata, M. Bartuluchi, G. Sevlever, A.L. Taratuto, Multidrug resistance proteins in tuberous sclerosis and refractory epilepsy, Pediatr. Neurol. 30 (2004) 102–106.
R.F. Jin, R.P. Sun, X.P. Xu, Expression of multidrug resistance gene and topira- mate affect expression of multidrug resistance gene in the hippocampus of spontaneous epileptic rats, Zhonghua Er Ke Za Zhi 43 (2005) 733–737.
L.W. Maines, D.A. Antonetti, E.B. Wolpert, C.D. Smith, Evaluation of the role of P- glycoprotein in the uptake of paroxetine, clozapine, phenytoin and carbamaza- pine by bovine retinal endothelial cells, Neuropharmacology 49 (2005) 610–617.
K.M. Mahar Doan, J.E. Humphreys, L.O. Webster, S.A. Wring, L.J. Shampine, C.J. Serabjit-Singh, K.K. Adkison, J.W. Polli, Passive permeability and P- glycoprotein-mediated efflux differentiate central nervous system (CNS) and non-CNS marketed drugs, J. Pharmacol. Exp. Ther. 303 (2002) 1029–1037.