• Open Access

Polymorphisms in the ABCB1 Gene in Phenobarbital Responsive and Resistant Idiopathic Epileptic Border Collies


  • Part of the results was presented as a poster at the 23rd Annual Symposium of the ESVN, Cambridge, UK, September 17–19, 2010.

Corresponding author: Lisa Alves, Department of Clinical Veterinary Medicine, Division of Clinical Neurology, Vetsuisse Faculty Bern, University of Bern, Länggassstrasse 128, 3001 Berne, Switzerland; e-mail: lisa.alves@itn.unibe.ch.


Background: Variation in the ABCB1 gene is believed to play a role in drug resistance in epilepsy.

Hypothesis/Objectives: Variation in the ABCB1 gene encoding the permeability-glycoprotein could have an influence on phenobarbital (PB) resistance, which occurs with high frequency in idiopathic epileptic Border Collies (BCs).

Animals: Two hundred and thirty-six client-owned BCs from Switzerland and Germany including 25 with idiopathic epilepsy, of which 13 were resistant to PB treatment.

Methods: Prospective and retrospective case-control study. Data were collected retrospectively regarding disease status, antiepileptic drug (AED) therapy, and drug responsiveness. The frequency of a known mutation in the ABCB1 gene (4 base-pair deletion in the ABCB1 gene [c.296_299del]) was determined in all BCs. Additionally, the ABCB1 coding exons and flanking sequences were completely sequenced to search for additional variation in 41 BCs. Association analyses were performed in 2 case-control studies: idiopathic epileptic and control BCs and PB-responsive and resistant idiopathic epileptic BCs.

Results: One of 236 BCs (0.4%) was heterozygous for the mutation in the ABCB1 gene (c.296_299del). A total of 23 variations were identified in the ABCB1 gene: 4 in exons and 19 in introns. The G-allele of the c.-6-180T > G variation in intron 1 was significantly more frequent in epileptic BCs resistant to PB treatment than in epileptic BCs responsive to PB treatment (Praw= .0025).

Conclusions and Clinical Importance: A variation in intron 1 of the ABCB1 gene is associated with drug responsiveness in BCs. This might indicate that regulatory mutations affecting the expression level of ABCB1 could exist, which may influence the reaction of a dog to AEDs.


antiepileptic drug


blood brain barrier


Border Collie


base pair


4 base pair deletion in the ABCB1 gene


central nervous system


idiopathic epilepsy


potassium bromide


minor allele frequency




polymerase chain reaction




single nucleotide polymorphism


3′-untranslated region

Epilepsy is the most common neurologic disorder in dogs and in humans.1 In most cases, seizure control requires long-term antiepileptic treatment, which often is successful in suppressing seizure frequency and severity. However, among epileptic dogs treated with standard antiepileptic drugs (AED) such as phenobarbital (PB) or potassium bromide (KBr), 25–30% are resistant to treatment despite adequate doses and serum concentrations of AEDs.2,3 This finding is similar to that in humans where one-third of epilepsy patients do not respond adequately to AEDs.4–6 Recently, a new definition of drug-resistant epilepsy in humans was proposed by the International League Against Epilepsy as failure of adequate trials of 2 AEDs to achieve sustained seizure control.7 Alternative AEDs such as levetiracetam, zonisamide, and gabapentine are only effective to a limited degree in managing PB- and KBr-resistant canine epilepsy.8,9

Pharmacokinetic theory proposes increased drug metabolism or altered expression of drug targets or drug transporter proteins in the blood-brain barrier (BBB) as possible mechanisms in resistance to AEDs.6,7 A prime candidate in this respect is the permeability-glycoprotein (P-gp), a transmembrane protein encoded by the ABCB1 gene, also called multidrug resistance 1 gene (MDR1). The ABCB1 gene in dogs is located on chromosome 14 and includes 28 exons.10,11 The protein, which functions as an ATP-dependent drug efflux pump in the BBB, supports a physiologic defense mechanism extruding a very wide variety of compounds that are potentially toxic to the central nervous system (CNS),10,12–15 including AEDs such as PB.16,17 Dysfunction of this neuroprotective protein may lead to either CNS intoxication or drug resistance. A notorious example of the former is a 4 base pair deletion (c.296_299del) in exon 4 of the ABCB1 gene of Collies and other herding breeds. In dogs carrying the deletion in a homozygous state, there is no functional P-gp and xenobiotics reach higher concentrations in the brain inducing severe neurological signs.18 Recently, Muñana et al19 reported that epileptic Collies homozygous for this ABCB1 deletion were significantly more likely to have better medical seizure control than epileptic Collies without the mutated ABCB1 gene.

Other variations of the ABCB1 gene may lead to overexpression of P-gp. In humans and rats, this mechanism appeared to prohibit AEDs from reaching therapeutic concentrations in epileptic foci.6,20,21 A silent ABCB1 gene single nucleotide polymorphism (SNP) has been linked to pharmacoresistance in human epilepsy, supposedly by altering the translation efficiency and a subsequently altered folding of the protein.22 Moreover, SNPs in the promoter region of the ABCB1 gene have been reported to be associated with difference in expression of ABCB1 messenger ribonucleic acid.23

In view of this central role of P-gp, we targeted the ABCB1 gene in a population of Border Collies (BCs) with idiopathic epilepsy (IE), which recently was described as a genetically mediated disease with a mode of inheritance resembling autosomal recessive transmission.24 BCs frequently present with severe epileptic seizures characterized by a high prevalence of cluster seizures and status epilepticus. Importantly, seizures often are poorly controlled with AEDs and drug-resistant epilepsy develops in up to 71% of the cases.24

Because the ABCB1 c.296_299del mutation has been reported in BCs and to be related with drug responsiveness, we sought to determine the frequency of this mutant ABCB1 allele in our BC population. Furthermore, the ABCB1 coding region and flanking sequences of all idiopathic epileptic BCs and a control group were sequenced to search for additional genetic variation and its association with drug responsiveness.

Materials and Methods

Data Collection

Blood samples were collected from 2005 to 2009 from 236 client-owned BCs from Switzerland and Germany and sent to the Institute of Genetics, University of Bern, Switzerland.

A total of 236 samples were collected from BCs, of which 25 were diagnosed with IE. From the 25 IE BCs, 13 dogs originated from Germany and 12 from Switzerland. The majority of the IE BCs came from a working lineage (17/25). For the remaining 8 BCs, lineage information was not provided.

The samples were organized into different groups. The epilepsy group consisted of BCs with IE (n = 25). A detailed description of IE in BCs recently has been reported.24 The diagnosis was based on standard physical and neurological examinations during the interictal period, absence of abnormal laboratory findings and a history of recurrent seizures for at least 1 year without any other clinical or neurologic signs according to previously published criteria.24 An owner questionnaire was used to record seizure phenotype and frequency before and during AED treatment in order to determine the seizure baseline frequency and monitor response to treatment, type of AED, and AED serum concentrations. Idiopathic epileptic BCs treated at least 3 months with PB as monotherapy were included in the study. If IE BCs failed to respond to PB treatment, a 2nd AED was used in several dogs, but this was not evaluated in our study. The 25 epileptic dogs were unrelated at the parental level. Considering these data and comparing the seizure baseline frequency for a minimal period of 3 months before and after initiating PB, the IE group then was further subdivided into PB-resistant IE BCs (n = 13) and PB-responsive IE BCs (n = 8). PB-resistant IE BCs were defined as dogs treated only with PB with serum concentrations within therapeutic ranges (>20 μg/mL), and showing no reduction or less than a 50% reduction in seizure frequency within a period of at least 3 months. PB-responsive IE BCs were defined as dogs with a 50% or greater reduction in seizure frequency after PB treatment within the first 3 months of treatment. PB responsiveness was assessed only in dogs in which serum concentrations were determined and classification as resistant or responsive was based on change in seizure frequency relative to a minimum of a 3-month baseline. Four IE BCs remained unclassified. The frequency of PB resistance was 52% (13/25) in our probands. An additional group consisted of 211 BCs with no history of seizures.

Genomic DNA was isolated from ethylenediaminetetraacetic acid anticoagulated venous blood and polymerase chain reaction (PCR) amplification was performed with 10 ng of DNA as described.25,26

Fragment Size Analysis for Detecting the ABCB1 c.296_299del Variant

Fluorescently labeled PCR primers were designed to amplify the region of the ABCB1 c.296_299del variant. The sizes of the PCR products were determined on an ABI 3,730 capillary sequencer.a Data were analyzed with GeneMapper 4.0 software.a A 138 base pair (bp) fragment from the wild type ABCB1 allele or a 134 bp fragment from the mutant ABCB1 allele with the 4 bp deletion was amplified.

Mutation Analysis and Genotyping

The study population consisted of 41 BCs including 25 IE BCs, which were defined as a case group. There were 13 males (2 were neutered) and 12 females (1 was neutered). The mean age (range) of the IE dogs was 94 months (7–180 months). In addition, a sample of 16 BCs without epilepsy and unrelated at the parental level was selected as control group from the 211 nonepileptic BCs. There were 5 males (2 were neutered) and 11 females (3 were neutered). The oldest BCs were chosen as controls, reducing the odds of late-onset IE. The mean age (range) of the control dogs was 119 months (60–192 months).

To identify polymorphisms and other variations, we individually amplified the coding exons 2–28 with flanking sequences of the ABCB1 gene and generated PCR products of approximately 600 bp. After purification, the PCR products were directly sequenced with an ABI3730 capillary sequencer. The Sequencher 4.9 software was used for detecting and assessing SNPs and other polymorphisms (GeneCodes). The CanFam 2.0 dog genome assembly derived from a female Boxer was used as reference sequence.

Association Analysis

In the present work, 2 case-control studies were designed. The 1st case-control study included the 25 IE BCs as case group and the 16 nonepileptic BCs as control group. The 2nd case-control study included the 13 PB-resistant IE BCs as case group and the 8 PB-responsive IE BCs as control group. Statistical analysis for allele and genotype frequencies between case and control groups for both studies was conducted by the Pearson χ2-test by HAPLOVIEW version 4.2 statistical software.27 Only variants with a minor allele frequency (MAF) > 0.1 were analyzed. To correct for multiple testing, a Bonferroni correction was performed by multiplying the raw P-value with the number of markers tested to yield a corrected P-value (Pcorr). P-values <.05 were considered statistically significant.


Analysis of the ABCB1 c.296_299del Variant

A total of 236 BCs were screened for the c.296_299del variant. One BC was found to be heterozygous for the mutation. This dog originated from Germany and had IE resistant to PB treatment. Thus, from 236 BCs, 235 (99.6%) were homozygous for the wild type allele (normal) and 1 (0.4%) was a carrier for the mutant allele.

Search for Additional Variants in the ABCB1 Gene

The complete coding sequence of the ABCB1 gene was sequenced in 41 BCs in order to detect additional genetic variation. The sequences were inspected for deviations from the published Boxer reference sequence. The results are presented in the Table 1. A total of 23 genetic variants were detected. Three variations were in the coding parts of exons 3, 26, and 27, respectively. One variation was in the 3′-untranslated region (3′–UTR) of exon 28, and 19 variations were located in introns. The SNP in exon 3, c.72A > T, is predicted to result in a conservative amino acid exchange from lysine to asparagine at position 24 (p.K24N) and was found in only 1 BC with a PB-resistant IE phenotype. The SNP in exon 26, c.3439A > G, results in a conservative amino acid exchange from methionine to valine at position 1,147 (p.M1147 V). The variant allele was present in several dogs, including nonepileptic BCs. The variant in exon 27 was translationally silent and again present in only 1 BC with a PB-resistant IE phenotype.

Table 1.   Variants and minor allele frequencies (MAF) of the canine ABCB1 gene.
 CanFam2.1 CFA 14, PositionVariation (genomic)Variation (cDNA)ABCB1 PositionFunctionMAF
116,692,297g.16692297C > Tc.-6-203G > AIntron 1 0.15
216,692,274g.16692274A > Cc.-6-180T > GIntron 1 0.24
316,692,180g.16692180T > Ac.-6-86A > TIntron 1 0.38
416,687,044g.16687044A > Tc.72A > TExon 3p.K24N0.012
516,686,946g.16686946A > Gc.114 + 56T > CIntron 3 0.058
616,686,911g.16686911G > Ac.114 + 91C > TIntron 3 0.058
716,676,691g.16676691A > Gc.115−3T > CIntron 3 0.012
816,661,594g.16661594A > Cc.533 + 12T > GIntron 6 0.081
916,661,588g.16661588A > Tc.533 + 18T > AIntron 6 0.012
1016,647,856g.16647856G > Ac.1227 + 22C > TIntron 11 0.026
1116,647,009g.16647009A > Cc.1557 + 249T > GIntron 13 0.023
1216,635,599g.16635599C > Ac.2068−92G > TIntron 16 0.081
1316,623,052g.16623052C > Tc.2484 + 13G > AIntron 20 0.035
1416,611,901g.16611901_2insCAAATc.2689−151_150insGATTTIntron 21 0.14
1516,606,994g.16606994A > Cc.2931−406T > GIntron 23 0.43
1616,606,962g.16606962C > Gc.2931−374G > CIntron 23 0.14
1716,606,909g.16606909G > Cc.2931−321C > GIntron 23 0.14
1816,606,859g.16606859C > Tc.2931−271G > AIntron 23 0.070
1916,606,656g.16606656G > Ac.2931−68C > TIntron 23 0.24
2016,606,423g.16606423C > Tc.3087 + 9G > AIntron 24 0.14
2116,599,193g.16599193T > Cc.3439A > GExon 26p.M1147V0.11
2216,596,356g.16596356G > Ac.3567C > TExon 27silent0.012
2316,595,039g.16595039T > Cc.*230A > GExon 28, 3′-UTR 0.14

Association Analysis

Twelve of the 23 variants had MAFs of <10% and were regarded as rare alleles (Table 1). The other 11 variants had MAFs >10% in our probands and were used for association analyses. These 11 markers were in Hardy-Weinberg equilibrium (P-value cutoff >.01).

In the first case-control study, the allele frequencies of the markers were compared between the 25 IE BCs and the 16 nonepileptic BCs. No statistically significant difference was observed between the 2 groups (Praw= .0669, Pcorr= .7359).

In the 2nd case-control study, the allele frequencies were compared between 13 IE BCs that were PB-resistant and 8 IE BCs that were PB-responsive. A statistically significant association between the SNP c.-6-180T > G at intron 1 and drug responsiveness was detected. The G-allele of the SNP c.-6-180T > G was significantly more frequent in PB-resistant IE BCs than in PB-responsive IE BCs (Praw= .0025, Pcorr= .0275; Table 2). All responsive BCs were homozygous T/T at this marker. Most of the resistant BCs were heterozygous G/T, 1 was homozygous G/G and 3 were homozygous T/T (Table 3).

Table 2.   Allele frequencies of canine ABCB1 markers and case control association analyses.
c.-6-203 G > Ac.-6-180 T > Gc.-6-86 A > Tc.2689-151_150 insGATTTc.2931-406 T > Gc.2931-374 G > Cc.2931-321 C > Gc.2931-68 C > Tc.3087+9 G > Ac.3439 A > Gc.*230 A > G
  • *

    P < .05.

Epileptic versus nonepileptic
 BC (n = 41)
  Frequency epileptic (n = 25)0.160.300.420.880.500.880.880.760.880.120.88
  Frequency control (n = 16)0.0940.130.310.810.310.810.810.750.810.0620.81
  Praw value.3905.0669.3274.3989.0940.3989.3989.9181.3989.3920.4431
  Pcorr value1.7359111111111
Resistant versus responsive
 BC (n = 21)
  Frequency resistant (n = 13)0.230.420.460.960.580.960.960.810.960.150.96
  Frequency responsive (n = 8)
  Praw value.0379*.0025*.5821.0398*.2037.0398*.0398*.6581.0398*.0991.0510
  Pcorr value.4169.0275*1.43781.4378.43781.43781.5610
Table 3.   Genotype frequencies at the c.-6-180T > G SNP in responsive and resistant epileptic Border Collies.
Cases (resistant)391
Controls (responsive)8


AED resistance is a recurrent problem in the therapeutic management of canine and human IE. Because AED response is thought to be at least in part owing to genetic variability, pharmacogenomics and pharmacogenetics are used to investigate genes associated with the pharmacology of AED.28 In humans, the drug transporter hypothesis of pharmacoresistant epilepsy has been intensely investigated in recent years based on the hypothesis that localized overexpression of certain transporters proteins, such as P-gp would compromise the efficacy of AEDs by limiting their capacity to penetrate the BBB and reach their intended site of action.6,10,14,15,20,29–32 Typically, patients are resistant to multiple AEDs despite the fact that these drugs have different mechanisms of action. This suggests a nonselective mode of pharmacoresistance, which would very well fit a central role of P-gp in view of its extremely wide range of substrates.10,31,33 Altered interactions between P-gp and its substrates, including PB, were shown to be associated with ABCB1 gene polymorphisms in humans.16

A study using a canine osteosarcoma cell line with P-gp overexpression evaluated the affinity of canine P-gp for a number of commonly used AEDs and reported PB as a weak substrate for canine P-gp.34 Such findings may not necessarily be representative for the situation in epileptic CNS tissue, where effects on substrate affinity and expression of canine P-gp resulting from ABCB1 gene polymorphisms may be different. Variations in function and concentration of P-gp between anatomical locations (normal brain versus epileptic focus) have been reported in other species.35,36 The findings of Muñana et al19 showing that a mutated, nonfunctional P-gp is significantly associated with better medical seizure control strongly argue for an important role of this protein in canine AEDs. Supported by these data and in view of the high incidence of AED resistance in IE BCs,24 the ABCB1 gene in IE BCs seems to be particularly suited for pharmacogenetic studies.

In view of the known role of the c.296_299del mutation in Collie dogs and its recently published association with seizure control,19 we sought to examine the incidence of this mutation in our samples. A very low occurrence of this 4 bp deletion in BCs has been previously detected in other countries.36,37 In our 236 samples, which included 11% IE dogs, the frequency of 0.4% of heterozygotes is in accordance with previously reported low frequencies in nonepileptic BCs and much lower than described for other herding breeds.37–40

In the present study, we focused on the coding regions of the ABCB1 gene, but also included flanking introns in order to ensure complete genotyping of the entire coding sequence and the splice sites. A case-control association revealed that the allele frequencies of an ABCB1 gene polymorphism in intron 1 are different between PB-resistant and PB-responsive IE BCs. PB-resistant IE BCs were more likely to have the variant G-allele at the SNP c.-6-180T > G. Thus, a G-allele at that location was associated with poor seizure control and may imply a worse prognosis. The genotype distribution indicates that the risk-increasing G-allele acts in a dominant fashion. Interestingly, this SNP is located at intron 1 near the 5′-end of the gene where the most important promoter elements are located.23

Regarding the putative functional effect of the promoter region, polymorphisms here can be responsible for different transcription activity of the ABCB1 gene, ultimately leading to different expression of P-gp. Regarding the phenotype PB-resistant IE BC, one could hypothesize that this promoter polymorphism might be related to an upregulation of the gene and an overexpression of P-gp in the brain of IE BCs. Functional studies are warranted to understand the regulatory effect of this promoter polymorphism and its potential clinical relevance.

Indeed, changes in expression, changes in functionality, or both of P-gp were shown to be associated with SNPs in the ABCB1 gene in other species.30,41–43 Overexpression of P-gp has been recognized in experimentally induced seizure foci and confirmed in surgically removed human epileptic brain tissue, potentially leading to focal areas where AEDs cannot penetrate.35,36,44 More complex mechanisms such as gene-gene interaction or interaction between different variants in the same gene also may interfere with protein expression and functionality.43

Regarding the coding region of ABCB1 gene, 3 variants were found. Two led to a conservative amino acid exchange in P-gp and the other was a synonymous substitution. Variants in the coding regions of the ABCB1 gene, even acting as silent polymorphisms, were reported to change substrate specificity.22,41,45 In our study, none of these SNPs were genetically associated with epilepsy in BCs or with PB responsiveness.

It has also been considered that increased expression of P-gp in drug-resistant epilepsy may only be an epiphenomenon that occurs in epileptic brain regardless of drug response6,46 or may depend on other factors such as hormones, and ingested plant compounds, certain food constituents, or supplements.6,7,10,35,42,47 Moreover, our results must be interpreted with caution. An association does not prove a cause and effect relationship, and false-positive association (type I error) because of population stratification is possible even if care is taken in choosing unrelated cases and controls at the parental level.

The underlying mechanism of drug-resistant epilepsy is complex and, although we assume a central role of P-gp, other membrane transporters could be involved. A recent genetic study screened 125 epileptic dogs representing 44 breeds for 384 SNP across 30 genes.48 All of these genes were related to PB pharmacology. Five genes were suggested to have an association with PB response although they did not reach statistical significance (KCNQ3, SCN2A2, GABRA2, EPOX HYD, and ABCB4). In the ABCB1 gene, 18 SNP were found that were not further described but showed no statistically significant association with PB response.48

The identification of the 4 bp deletion in the ABCB1 gene has been proven to be of clinical value to predict the response of an individual dog to a wide variety of xenobiotics and AEDs. Therefore, it would be valuable to identify polymorphisms in the ABCB1 gene that could be used as putative markers in order to make rational decisions concerning therapy. In view of the relatively small sample size, our results remain preliminary but warrant further association studies including all putative regulatory variants from the canine ABCB1 gene. Subsequent functional analyses will further enhance our understanding of the mechanisms underlying the effect of ABCB1 gene on epilepsy and drug responsiveness.


aApplied Biosystems, Rotkreuz, Switzerland


The authors thank Dr Daniela Gerber-Mattli, the European College of Veterinary Neurology Diplomates, and the pet owners for assistance with case recruitment. We thank Prof. M. Vandevelde for advice and for critical reading of the manuscript.