A sex-dependent functional polymorphism in the canine multidrug resistance (MDR1) gene
Ute Philipp*, Silvia Fecht*, Anne Wöhlke, Ottmar Distl
*authors contributed equally to the work
Institute for Animal Breeding and Genetics, University of Veterinary Medicine Hannover, Bün-teweg 17p, 30559 Hannover, Germany (Philipp, Fecht, Wöhlke and Distl)
Corresponding author:
Ute Philipp
University of Veterinary Medicine Hannover Bünteweg 17p
30559 Hannover Germany
Phone +49 511-953-8866 Fax +49 511-953-8582
E-mail: ute.philipp@tiho-hannover.de
Abstract
MDR1 encodes P-glycoprotein which plays a significant role in the xenobiotic transport process.
In several dog breeds, the mdr1-1Δ mutation was associated with multiple drug sensitivity, but no further sequence variations in the canine MDR1 gene have been described. To analyse whether functional polymorphisms exist in exon 12, 21 and 26 of the canine MDR1 gene analogically to findings in the orthologous human gene, sequence analysis was performed. We detected a non-synonymous c.3439A>G transition in exon 26 which causes a Met1147Val substi-tution. Subsequently, we analysed 186 Elo dogs and 65 dogs of different breeds for the c.3439A
>G polymorphism. The presence of the G-allele in the analysed Elos with a frequency of 56.45%
was mainly influenced by the founder breed Samoyed. Furthermore, the mutant allele was found in six of 15 analysed breeds.
cDNA analysis performed in four Elos previously genotyped for the c.3439A>G SNP revealed two further synonymous SNPs c.504C>T and c.1663C>T in the coding region and a c.*230G>A in the 3`UTR. SNP frequencies and haplotypes for all identified mutations were determined in 186 Elos on genomic DNA.
Using quantitative RT-PCR, the relative expression levels of MDR1 in hair root samples depend-ing on the M1147V genotype were compared. We found genotype and gender specific expression differences. Homozygous male A/A dogs showed higher relative MDR1 expression levels than female Elos and male G/G animals. These in-vitro observations might reflect transcriptional dif-ferences in dogs and might have impact on the metabolizing of drugs using the MDR1 trans-porter system in-vivo.
The superfamily of ABC (ATP-binding cassette) transporters have been of particular interest for research work in humans and animals, i.e. mice and dogs, for several years now. The MDR1 gene (multidrug resistance gene, also known as ABCB1 gene, ATP-binding cassette sub-family B member 1) and its gene product P-glycoprotein are the most thoroughly analysed among the ABC transporters (1).
The physiological role of P-glycoprotein is the protection of the organism from toxic xenobiot-ics. P-glycoprotein is normally expressed in various mammalian tissues including brain capillary endothelial cells (2), the apical border of intestinal epithelial cells (3), biliary canalicular cells (4), renal proximal tubular epithelial cells (5), placenta (6), and testes (7). P-glycoprotein confers protection by limiting the uptake of compounds from the gastrointestinal tract and by contribut-ing to their excretion via the liver, kidneys, and intestine. Moreover, P-glycoprotein in the blood-brain-barrier and other blood-tissue barriers protects sensitive organs from exposure to toxic compounds that may have entered the bloodstream (8). P-glycoprotein actively extrudes selected xenobiotics from within the cell back into the lumen of brain capillary, intestine, bile canaliculus, or renal tubule.
More than 50 structurally different therapeutic drugs are known substrates for human and murine P-glycoprotein. Because of the high degree of homology of P-glycoprotein between species, the same drugs are expected to be substrates of the canine P-glycoprotein (9). Degree of expression and the functionality of the MDR1 gene product can directly affect the therapeutic effectiveness of such agents because they play an important role for the physiological cell protection during drug therapy (10).
The canine MDR1 gene attracted interest for research work after several descriptions of ivermec-tin neurotoxicity in Collies and the observation that affected dogs had elevated concentrations of ivermectin in the central nervous system indicating that ivermectin neurotoxicity was caused by a defect in the blood-brain barrier (11, 12 and 13).
The canine MDR1 gene is located on CFA (Canis familiaris autosome) 14 and composed of 28 exons. The human MDR1 (ABCB1) gene is located on HSA (Homo sapiens autosome) 7. The product of the human MDR1 gene is among all species documented in data banks the most simi-lar to that of the canine gene. Human and dog P-glycoprotein display 91 % overall homology, with non-consensus residues being located outside the functional segments and are composed each of 12 transmembrane domains and two nucleotide binding domains (14). A 4-bp deletion
ity in dogs (14, 15 and 16). This mdr1-1Δ mutation causes a frame-shift accompanied by multi-ple premature stop codons resulting in a severely truncated P-glycoprotein composed of < 10%
of the wild-type amino acid sequence. The remainder of the protein lost its protecting function, e.g. in the blood-brain-barrier. Neurotoxic side effects are provoked in dogs with the mdr1-1Δ mutation in the case of drug therapy with P-glycoprotein substrates because of accumulation of these substrates in brain tissue. Although both research groups screened the whole cDNA of the canine MDR1 gene, there was no evidence on further sequence variations in this gene in the ana-lysed Collies.
In humans, many studies were concerned with sequence variations in the MDR1 gene. DNA se-quence variations cause phenotypic changes by multiple mechanisms, e.g. by changing the en-coded protein sequence, or by affecting gene regulation, mRNA processing, and translation (17).
Many of the detected polymorphisms in humans do not show effects on expression or function of MDR1 but some implicate modified protein levels or functionality, e.g. the SNP in exon 26 (c.3435C>T) (10). This research group performed the first systematic screening of the MDR1 gene for the presence of polymorphisms by sequencing all 28 exons including the core promoter region and flanking intron-exon boundaries. Of the 15 identified SNPs, the research group iden-tified two SNPs at wobble positions with no amino acid changes [exon 12 (c.1236C>T) and exon 26 (c.3435C>T)] and found an association of the SNP c.3435C>T with a modified level of intes-tinal MDR1-expression. Although results are not always consistent, most studies suggest that the c.3435C>T transition is associated with decreased MDR1 function and reduced mRNA and/or protein expression in some tissues (1). The two synonymous SNPs [exon 12 (c.1236C>T) and exon 26 (c.3435C>T)] are in linkage disequlibrium (17) with a nonsynonymous SNP in exon 21 (c.2677G>T/A) which causes an amino acid change (899Ala>Ser/Thr). Furthermore, they stated that c.3435C>T is a functional SNP that decreases mRNA stability, thereby decreasing MDR1 mRNA and/or protein levels, by analysis of allele-specific expression in liver autopsy samples and in vitro expression experiments.
It is not yet known whether polymorphisms in the canine MDR1 gene exist like in the human gene. In addition to the known deletion mutation in the fourth exon, further sequence variations in the canine MDR1 gene might also be found which modify the structure or expression levels of the MDR1 mRNA and/or protein and thereby change the function of P-glycoprotein. Therefore, the purpose of this study was to search for functional polymorphisms by sequencing the canine MDR1 gene.
Results
Genotyping of c.3439 polymorphism. Sequence analysis of the exons 12, 21, 26 was performed in two Collies and six Elos. The analysed DNA sequences of the exons 12 and 21 perfectly matched with the reference sequence of the canine MDR1 gene (GenBank accession no.
NC_006596.2). We found an A to G transition in exon 26 (c.3439A>G) in the DNA sequences of four from six Elos. Three Elos were homozygous G/G, one Elo was heterozygous A/G and two Elos as well as the two Collies were homozygous A/A. The reference boxer sequence (dog ge-nome assembly 2.1) showed an A at this position and was therefore defined as wild-type.
Thereupon, an analysis of the c.3439A>G SNP in exon 26 was performed in 186 Elos and 65 dogs of different breeds using the RFLP developed. In Elos, the wild-type A-allele was found with a frequency of 43.55%, whereas the G-allele was prevalent with a frequency of 56.45%.
Altogether, 144 from 186 Elos (77.4%) were heterozygous or homozygous for the G-allele (Ta-ble 1). In most of the analysed breeds the mutated allele could not be detected. In addition to the Elo breed, the G-allele was detected in Labrador Retrievers, Do-Khyis, Dalmatians, German Wirehaired Pointers, Hovawarts and Border Collies (Table 2). The G-allele could not be de-tected in Collies, German Shepherds, Dachshunds, Tibetan Terriers, English Cocker Spaniels, Irish Wolfhounds, Jack Russell Terriers, Boxers and Kromfohrlanders.
The subsequent regression analysis evaluated the influence of the gene contributions by the dif-ferent founder breeds on the presence of the allele A or the allele G. After the exclusion of seven from nine founder breeds due to very low and insignificant contributions to the variance ex-plained, the final model included the gene contributions by Samoyed and Dalmatian and was significant with a p-value of 0.0036. Table 3 shows the regression coefficients and error prob-abilities for the influence of gene contributions by Samoyed and Dalmatian on the presence of the alleles A and G in a random sample of 88 analysed Elo dogs. The regression coefficient for the gene contribution by Samoyed was at +0.0201 which indicated a major influence of the founder breed Samoyed for the G-allele. So, an increase by one percent of the gene contribution by Samoyed raises the frequency of the G-allele by 2.01%. The regression coefficient for the gene contribution by Dalmatian was at -0.0519 which indicated a major influence of the founder breed Dalmatian for the A-allele. Consequently, the frequency of the A-allele rises by 5.19% if the gene contribution by Dalmatian increases by one percent.
Effect of c.3439 polymorphism on protein sequence. The non-synonymous A to G exchange in exon 26 of the canine MDR1 gene causes an amino acid substitution from methionine to valine at position 1147 in the amino acid sequence (GenBank accession no. NP_001003215.1). For the analysis of possible impact of the modified amino acid sequence on P-glycoprotein, PolyPhen prediction and protein alignment for five selected species using ClustalW were performed. Poly-Phen analysis resulted in the prediction that this variant is benign with a PSIC (Position-Specific Independent Counts) score difference of 0.322. The protein alignment showed that only the dog displayed methionine at position 1147 in the amino acid sequence whereas in human, mouse and rat the amino acid valine was given at the corresponding position in the reference sequence (Fig-ure 2).
Polymorphisms in canine MDR1 trancript. To evaluate whether further sequence variations of the MDR1 transcript exist in the Elo breed, MDR1 cDNA of four Elo liver samples was se-quenced. The sequence data were compared to the Collie mRNA (NM_001003215.1) and the deduced mRNA from the genomic Boxer reference sequence (NC_006596.2).
Between the Elo dogs, three other SNPs (c.504C>T coding for Asp168 and c.1663C>T coding for Leu554 and the noncoding c.*230G>A, respectively) could be detected. The SNPs are local-ized in exons 6, 14 and 28. The c.504T and c.1663T alleles were not found in the Collie or Boxer reference sequence. Concerning the c.*230G>A polymorphism, the Boxer carries the A-allele while the Collie sequence exhibits the G-allele. The overall comparison to the deduced Boxer mRNA sequence revealed four polymorphisms (Table 4). Beside the c.3439A>G SNP, the se-quence differences between Elo and Boxer mRNA are synonymous or in the 3´UTR. Compared to the Collie mRNA sequence, 13 sequence differences including the c.3439 SNP were identified in the Elo mRNA leading to seven amino acid exchanges and insertion of asparagine (p.Lys24_
Glu25insAsn), respectively. The other sequence differences between Elo and Collie mRNA were synonymous or in the 3`UTR (Table 4).
The three newly identified SNPs were genotyped in 186 Elos (Table 1). The genomic Boxer se-quence (NC_006596.2) was set as wildtype sese-quence. For all SNPs, the Boxer alleles were more often observed. For polymorphisms c.504 and c.1663, the wildtype C alleles represented a fre-quency of 63.17 and 84.02 percent, respectively. And the c.*230 G-allele was even prevalent with a frequency of 91.94 percent. Linkage disequilibrium was observed between the loci c.504, c.3934 and c.*230 (Figure 3). For all four polymorphisms, haplotypes and their frequencies
were calculated. From 16 possible haplotypes, 15 were observed. Only nine showed a frequency over 0.01 percent (Table 5).
Expression of MDR1 using qRT PCR. To test whether the polymorphisms might influence the MDR1 expression hair root samples from 43 previously genotyped Elo dogs were taken and ana-lysed by qRT-PCR. The analysis showed, that only the c.3439A>G SNP affected the transcript level. The dogs carrying the A allele homozygous were considered as wild type, therefore the mean of the ΔCT of the AA dogs was used as calibrator. The relative expression levels of MDR1 in the analysed Elos by individual and genotype showed that A/A dogs have higher relative ex-pression levels than G/G and A/G dogs (Figure 3). Comparison of LS-means revealed a signifi-cant difference between the relative expression levels of all Elos with the A/A-genotype and the G/G-genotype. The expression level of the heterozygous animals showed an intermediate value (Table 6). A model which included the sex genotype interaction revealed that the expression level differences were mainly caused by the males while only slight differences were observed in fe-males (Table 7). Nonetheless, high individual differences of the expression levels of MDR1 in hair roots could be measured.
Discussion
A sample of 186 dogs of the Elo breed was analysed for a newly detected non-synonymous sin-gle nucleotide polymorphism in exon 26 (c.3439A>G) of the canine MDR1 gene. The mutated G-allele was prevalent in the Elo breed with a frequency of 56.45 % which is remarkable be-cause the A-allele is given in the reference sequence of the canine MDR1 gene. The regression analysis evaluated the influence of the gene contributions by the different founder breeds of the Elo on the presence of the A-allele or the G-allele in a random sample of 88 Elo dogs out of the 186 genotyped animals. The analysis resulted in the conclusion that the presence of the G-allele was mainly influenced by the founder breed Samoyed, whereas the presence of the A-allele in the analysed Elos was largely influenced by the Dalmatian founder dogs. Nevertheless, in the RFLP-PCR the G-allele was also found in the analysed Dalmatians, so this breed was also proved to carry both types of alleles. In addition, the G-allele was detected in the breeds Labra-dor Retriever, Do-Khyi, German Wirehaired Pointer, Hovawart and Border Collie. Due to the small number of dogs genotyped, the prevalence of the mutant allele in other dog breeds remains
Sequence analysis of the Elo MDR1 gene of four dogs which were genotyped for the c3439A>G SNP showed several sequence differences in and between breeds. The sequences were compared to the Collie reference mRNA and to genomic Boxer reference sequence. While the deduced pro-tein sequence between Boxer and Elo were identical except the Met1147Val polymorphism, seven amino acid exchanges were observed between the Elo and Collie protein and Boxer and Collie protein sequence, respectively. In human MDR1 mRNA, 61 SNPs are listed in dbSNP, from which are 39 missense mutations. In the orthologue mouse transcript 50 SNPs are reported, but only six non synonymous polymorphisms are described. These data implicates that the MDR1 Gene varies considerably within species and many functional isoforms may exist. The impact of the different isoforms on drug metabolizing has not been fully investigated as most studies have been focused on the c.3435C>T SNP in human, respectively.
Compairing the canine MDR1 amino acid sequence to four other mammal p-glycoprotein se-quences showed that the inserted Asn in the Boxer and Elo protein at position 25 is unique. Oth-erwise, the Elo and Boxer protein sequence seem to represent the more conserved isoform be-tween the species. For the surveyed amino acids, the 1147Val-Elo sequence matches best to the human orthologue (Table 8). Concerning the other amino acid positions, even if there were polymorphic sites between the other species, the Collie amino acid could not be observed. None-theless, all canine protein isoforms lead to functional receptor proteins which only differ slightly in molecular weight and amino acid composition (SI Table 4). The calculated isoelectric point shows no variation between the three canine MDR1 isoforms.
The c.3439A>G SNP in exon 26 causes an amino acid substitution in the MDR1 protein. Com-parison to a structural protein model of human p-glycoprotein indicates that the amino acid 1147 is part of an intracytoplasmic functional not conserved protein region (18). Protein alignment with four species and the PolyPhen prediction gave no further evidence that the substitution from methionine to valine has effects on structure or function of the protein. Valine is also found at the corresponding position in the reference amino acid sequence of other mammalian species like human, mouse or rat. Thus, valine in dogs may not have great impact on the functionality of P-glycoprotein since it is usually found at this position for functional proteins in other species. This location does not seem to be highly conserved because in the PolyPhen alignment output there are several amino acids listed for this position including valine, methionine, isoleucine and glu-tamic acid. It has to be assumed that a substitution from a nonpolar amino acid to another nonpo-lar amino acid does not have great impact on the structure of a protein. Nevertheless,
modifica-tions in amount or character of bonds in the case of an amino acid substitution can alter the sta-bility of the protein structure.
Moreover, the c.3439A>G SNP in exon 26 of the canine MDR1 gene is closely neighboured to the c.3435C>T SNP in the orthologous sequence of the human MDR1 gene. It is possible that the c.3439A>G transition has effects on mRNA and/or protein as it is the case for the c.3435C>T SNP in the human MDR1 gene. The c.3435C>T polymorphism is a synonymous substitution, so the amino acid sequence is not affected and there are no obvious structure modifications. Any-how, association of c.3435C>T with decreased MDR1 function and reduced mRNA or protein expression in some tissues was described (1). It was discussed that modification in MDR1 mRNA expression resulted from changing of mRNA stability (19).
Canine MDR1 expression was quantified in hair root cells. These samples are available without using invasive techniques. Prior to the assay, MDR1 expression could be detected in a measur-able quantity in hair root cells but showed a reduced abundance compared to the expression level in liver (data not shown). This is in concordance with a previous study (20) which demonstrated that MDR1 transcripts were in several human tissues, quantification showed that MDR1 was in liver more abundant than in skin.
In Elo dogs the individual MDR1 expression levels vary considerable in hair root cells. Such in-dividual variations were already observed in duodenal enterocytes of male Japanese in depend-ence of their c.3435C>T genotype (21). There might be several causes which contribute to the individual expression levels of the MDR1. In dogs, the estrogen concentration seems to influence the transcript level as gestating dogs showed increased expression levels. One female in the fourth week of gestation showed a 300-fold expression level, in a second dog known to be in an earlier gestation an eighteenfold increase of the expression level was observed. That would be in accordance with the observation that the expression of human MDR1 and its rodent orthologues is controlled by estrogens and progesterone in uteroplacental tissues (22, 23 and 24) and MDR1 mRNA were inducible by several female sex steroids in vitro (25). Contrary to this, no effect of estrogene or progesterone was found on mdr1 expression in rat liver (26). And, in lung paren-chyma individual expression variations of rat mdr1 was observed but no effect of pregnancy
In Elo dogs the individual MDR1 expression levels vary considerable in hair root cells. Such in-dividual variations were already observed in duodenal enterocytes of male Japanese in depend-ence of their c.3435C>T genotype (21). There might be several causes which contribute to the individual expression levels of the MDR1. In dogs, the estrogen concentration seems to influence the transcript level as gestating dogs showed increased expression levels. One female in the fourth week of gestation showed a 300-fold expression level, in a second dog known to be in an earlier gestation an eighteenfold increase of the expression level was observed. That would be in accordance with the observation that the expression of human MDR1 and its rodent orthologues is controlled by estrogens and progesterone in uteroplacental tissues (22, 23 and 24) and MDR1 mRNA were inducible by several female sex steroids in vitro (25). Contrary to this, no effect of estrogene or progesterone was found on mdr1 expression in rat liver (26). And, in lung paren-chyma individual expression variations of rat mdr1 was observed but no effect of pregnancy