• View in gallery
    Figure 1—

    Gene expression profiling of ABCB1-1Δ mutant mice and ABCB1-WT mice following administration of P-gp substrates digoxin, doramectin, ivermectin, or moxidectin. Groups of 2 ABCB1-1Δ mutant mice and 2 ABCB1-WT mice were administered digoxin (1.53 mg/kg, SC), doramectin (10 mg/kg, SC), ivermectin (10 mg/kg, SC), or moxidectin (10 mg/kg, PO) and monitored hourly for 7 hours before euthanasia. A toxicogenomic approach based on DNA microarrays was used to examine whole-genome expression changes in the treated mice. A—Unsupervised hierarchical clustering analysis of whole-genome gene expression for ABCB1-1Δ mutant mice and ABCB1-WT mice administered P-gp substrates. The clustering revealed that all P-gp substrate–treated samples derived from ABCB1-1Δ mutant mice shared a similar gene expression pattern (regardless of the P-gp substrate administered) and formed a major cluster (cluster I). Similarly, all of the samples derived from ABCB1-WT mice formed another major cluster (cluster II), with the exception of 1 sample from the moxidectin-treated ABCB1-WT group. B—Results of principal components analysis of all samples. The 3 axes PC#1, PC#2, and PC#3 are the 3 principle components representing gene expression variations of the dataset. The first component explains the maximal amount of variance in the data. The second principal component is perpendicular to the first one and explains maximum residual variance. The third component is perpendicular to the first 2 and explains the maximum residual variance. The normalized gene expression value of the dataset is plotted along each axis, and the percentage of gene expression variance of the dataset is indicated along each axis. The identification number of each sample is indicated in the image.

  • View in gallery
    Figure 2—

    Venn diagram identifying common genes that were altered in ABCB1-1Δ mutant mice following administration of each of the P-gp substrates ivermectin, doramectin, moxidectin, and digoxin (2 mice/treatment). Genes with a ≥ 2-fold gene expression change, compared with findings for the ABCB1-WT mice, were used for comparison. Thirty-four genes were altered in all 4 treatment groups.

  • View in gallery
    Figure 3—

    Results of gene pathway analysis of altered gene signaling pathways and networks in ABCB1-1Δ mutant mice following treatment with ivermectin or digoxin. Genes with a ≥ 2-fold gene expression change, compared with findings for the ABCB1-WT mice, were used for pathway analysis. The schematics illustrate the top-ranked gene network associated with behavior from ivermectin-treated mice (A) and digoxin-treated mice (B). Data were analyzed through the use of QIAGEN's Ingenuity® Pathway Analysis (IPA®, QIAGEN Redwood City, www.qiagen.com/ingenuity).

  • 1. Schinkel AH, Mayer U, Wagenaar E, et al. Normal viability and altered pharmacokinetics in mice lacking mdr1-type (drug-transporting) P-glycoproteins. Proc Natl Acad Sci U S A 1997; 94: 40284033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2. Schinkel AH. The physiological function of drug-transporting P-glycoproteins. Semin Cancer Biol 1997; 8: 161170.

  • 3. Kitamura Y, Koto H, Matsuura S, et al. Modest effect of impaired P-glycoprotein on the plasma concentrations of fexofenadine, quinidine, and loperamide following oral administration in collies. Drug Metab Dispos 2008; 36: 807810.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. Schinkel AH, Smit JJ, van Tellingen O, 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 1994; 77: 491502.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5. Paul AJ, Tranquilli WJ, Seward RL, et al. Clinical observations in Collies given ivermectin orally. Am J Vet Res 1987; 48: 684685.

  • 6. Dadarkar SS, Deore MD, Gatne MM. Comparative evaluation of acute toxicity of ivermectin by two methods after single subcutaneous administration in rats. Regul Toxicol Pharmacol 2007; 47: 257260.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7. Orzechowski KL, Swain MD, Robl MG, et al. Neurotoxic effects of ivermectin administration in genetically engineered mice with targeted insertion of the mutated canine ABCB1 gene. Am J Vet Res 2012; 73: 14771484.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Mealey KL, Bentjen SA, Gay JM, et al. Ivermectin sensitivity in Collies is associated with a deletion mutation of the mdr1 gene. Pharmacogenetics 2001; 11: 727733.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Hugnet C, Bentjen SA, Mealey KL. Frequency of the mutant MDR1 allele associated with multidrug sensitivity in a sample of Collies from France. J Vet Pharmacol Ther 2004; 27: 227229.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Kawabata A, Momoi Y, Inoue-Murayama M, et al. Canine mdr1 gene mutation in Japan. J Vet Med Sci 2005; 67: 11031107.

  • 11. Neff MW, Robertson KR, Wong AK, et al. Breed distribution and history of canine mdr1–1Delta, a pharmacogenetic mutation that marks the emergence of breeds from the Collie lineage. Proc Natl Acad Sci U S A 2004; 101: 1172511730.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12. Swain MD, Orzechowski KL, Swaim HL, et al. P-gp substrate-induced neurotoxicity in an Abcb1a knock-in/Abcb1b knock-out mouse model with a mutated canine ABCB1 targeted insertion. Res Vet Sci 2013; 94: 656661.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13. Bloomquist JR. Toxicology, mode of action and target sitemediated resistance to insecticides acting on chloride channels. Comp Biochem Physiol C 1993; 106: 301314.

    • Search Google Scholar
    • Export Citation
  • 14. Bloomquist JR. Ion channels as targets for insecticides. Annu Rev Entomol 1996; 41: 163190.

  • 15. Bloomquist JR. Chloride channels as tools for developing selective insecticides. Arch Insect Biochem Physiol 2003; 54: 145156.

  • 16. Fonnum F. Glutamate: a neurotransmitter in mammalian brain. J Neurochem 1984; 42: 111.

  • 17. Moriyama Y & Yamamoto A. Glutamatergic chemical transmission: look! Here, there, and anywhere. J Biochem 2004; 135: 155163.

  • 18. Chen J, Lipska BK, Weinberger DR. Genetic mouse models of schizophrenia: from hypothesis-based to susceptibility gene-based models. Biol Psychiatry 2006; 59: 11801188.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19. Kashani A, Lepicard E, Poirel O, et al. Loss of VGLUT1 and VGLUT2 in the prefrontal cortex is correlated with cognitive decline in Alzheimer disease. Neurobiol Aging 2008; 29: 16191630.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20. Nestler EJ, Barrot M, DiLeone RJ, et al. Neurobiology of depression. Neuron 2002; 34: 1325.

  • 21. Tsarovina K, Reiff T, Stubbusch J, et al. The Gata3 transcription factor is required for the survival of embryonic and adult sympathetic neurons. J Neurosci 2010; 30: 1083310843.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 22. Prajapati KD, Sharma SS & Roy N. Current perspectives on potential role of albumin in neuroprotection. Rev Neurosci 2011; 22: 355363.

  • 23. Fecht S & Distl O. Review of prevalence, genetic aspects and adverse effects of the mdr1–1Delta mutation in dogs. Dtsch Tierarztl Wochenschr 2008; 115: 212219.

    • Search Google Scholar
    • Export Citation

Advertisement

Identification of potential biomarkers of P-glycoprotein substrate neurotoxicity in transgenic mice expressing the mutated canine ABCB1 gene

Min ZhuOffice of Research, Center for Veterinary Medicine, US FDA, 8401 Muirkirk Rd, Laurel, MD 20708.

Search for other papers by Min Zhu in
Current site
Google Scholar
PubMed
Close
 DVM, PhD
,
Yi MingAdvanced Biomedical Computing Center, Information Systems Program, SAIC-Frederick Inc, National Institutes of Health, 1050 Boyles St, Frederick, MD 21702.

Search for other papers by Yi Ming in
Current site
Google Scholar
PubMed
Close
 PhD
,
Heidi SwaimOffice of Research, Center for Veterinary Medicine, US FDA, 8401 Muirkirk Rd, Laurel, MD 20708.

Search for other papers by Heidi Swaim in
Current site
Google Scholar
PubMed
Close
 BS
,
Marla D. SwainOffice of Research, Center for Veterinary Medicine, US FDA, 8401 Muirkirk Rd, Laurel, MD 20708.

Search for other papers by Marla D. Swain in
Current site
Google Scholar
PubMed
Close
 PhD
,
Michael J. MyersOffice of Research, Center for Veterinary Medicine, US FDA, 8401 Muirkirk Rd, Laurel, MD 20708.

Search for other papers by Michael J. Myers in
Current site
Google Scholar
PubMed
Close
 PhD
,
Christine DeaverOffice of Research, Center for Veterinary Medicine, US FDA, 8401 Muirkirk Rd, Laurel, MD 20708.

Search for other papers by Christine Deaver in
Current site
Google Scholar
PubMed
Close
 BS
,
Xiaolin WuNational Institutes of Health, SAIC-Frederick Inc, Laboratory of Molecular Technology, 915 Tollhouse Ave, Frederick, MD 21701.

Search for other papers by Xiaolin Wu in
Current site
Google Scholar
PubMed
Close
 PhD
,
Yolanda L. JonesOffice of Research, Center for Veterinary Medicine, US FDA, 8401 Muirkirk Rd, Laurel, MD 20708.

Search for other papers by Yolanda L. Jones in
Current site
Google Scholar
PubMed
Close
 MS, MBA
, and
Haile F. YancyOffice of Research, Center for Veterinary Medicine, US FDA, 8401 Muirkirk Rd, Laurel, MD 20708.

Search for other papers by Haile F. Yancy in
Current site
Google Scholar
PubMed
Close
 PhD

Abstract

Objective—To identify biomarkers of P-glycoprotein (P-gp) substrate neurotoxicity in transgenic mice expressing the mutant canine ABCB1 gene (ABCB1-1Δ).

Animals—8 ABCB1 knock-in and knock-out transgenic mice expressing the ABCB1-1Δ gene and 8 control mice expressing the wild-type canine ABCB1 gene (ABCB1-WT).

Procedures—Groups including 2 ABCB1-1Δ mutant mice and 2 ABCB1-WT mice were administered the P-gp substrates ivermectin (10 mg/kg, SC), doramectin (10 mg/kg, SC), moxidectin (10 mg/kg, PO), or digoxin (1.53 mg/kg, SC). A toxicogenomic approach based on DNA microarrays was used to examine whole-genome expression changes in mice administered P-gp substrates.

Results—Compared with control ABCB1-WT mice, ABCB1-1Δ mutant mice developed neurotoxic signs including ataxia, lethargy, and tremors similar to those reported for dogs with the ABCB1-1Δ mutation. Microarray analysis revealed that gene expression was altered in ABCB1-1Δ mutant mice, compared with findings for ABCB1-WT mice, following administration of the same P-gp substrates. Gene pathway analysis revealed that genes with a ≥ 2-fold gene expression change were associated with behavior and nervous system development and function. Moreover, 34 genes were altered in the ABCB1-1Δ mutant mice in all 4 drug treatment groups. These genes were also associated with behavior, which was identified as the top-ranked gene network.

Conclusions and Clinical Relevance—These study data have facilitated understanding of the molecular mechanisms of neurotoxicosis in ABCB1-1Δ mutant mice following exposure to various P-gp substrates. Some genes appear to be potential biomarkers of P-gp substrate neurotoxicity that might be used to predict the safety of those drugs in dogs with the ABCB1-1Δ mutation.

Abstract

Objective—To identify biomarkers of P-glycoprotein (P-gp) substrate neurotoxicity in transgenic mice expressing the mutant canine ABCB1 gene (ABCB1-1Δ).

Animals—8 ABCB1 knock-in and knock-out transgenic mice expressing the ABCB1-1Δ gene and 8 control mice expressing the wild-type canine ABCB1 gene (ABCB1-WT).

Procedures—Groups including 2 ABCB1-1Δ mutant mice and 2 ABCB1-WT mice were administered the P-gp substrates ivermectin (10 mg/kg, SC), doramectin (10 mg/kg, SC), moxidectin (10 mg/kg, PO), or digoxin (1.53 mg/kg, SC). A toxicogenomic approach based on DNA microarrays was used to examine whole-genome expression changes in mice administered P-gp substrates.

Results—Compared with control ABCB1-WT mice, ABCB1-1Δ mutant mice developed neurotoxic signs including ataxia, lethargy, and tremors similar to those reported for dogs with the ABCB1-1Δ mutation. Microarray analysis revealed that gene expression was altered in ABCB1-1Δ mutant mice, compared with findings for ABCB1-WT mice, following administration of the same P-gp substrates. Gene pathway analysis revealed that genes with a ≥ 2-fold gene expression change were associated with behavior and nervous system development and function. Moreover, 34 genes were altered in the ABCB1-1Δ mutant mice in all 4 drug treatment groups. These genes were also associated with behavior, which was identified as the top-ranked gene network.

Conclusions and Clinical Relevance—These study data have facilitated understanding of the molecular mechanisms of neurotoxicosis in ABCB1-1Δ mutant mice following exposure to various P-gp substrates. Some genes appear to be potential biomarkers of P-gp substrate neurotoxicity that might be used to predict the safety of those drugs in dogs with the ABCB1-1Δ mutation.

P-glycoprotein is a transmembrane efflux protein in the ATP-binding cassette family of transport proteins. It is encoded by the MDR1 gene, which is currently termed ABCB1. Humans and dogs have only 1 form of the ABCB1 gene, whereas mice have 2 genes encoding slightly different Abcb1 isoforms: Abcb1a and Abcb1b.1 P-glycoprotein is present at the apical surface of epithelial cells in the liver, intestines, and kidneys and at the apical surface of brain capillary endothelial cells. It acts as a barrier for these organs, protecting the cells within them by actively extruding various xenobiotic compounds and endogenous metabolites.2 P-glycoprotein has very broad substrate specificity and affects drug disposition, including absorption, distribution, metabolism, and excretion.3 In the absence of a functional P-gp, disruption of the murine blood-brain barrier is known to cause increases in drug concentrations in the brain and other tissues, decrease drug elimination, and increase drug toxicity.4

A mutation in the ABCB1 gene has been reported in a subpopulation of dogs, including Collies, that were sensitive to the neurotoxic effects induced by ivermectin. Depending on the dosage of ivermectin given, Collies developed clinical signs ranging from mild (salivation, vomiting, ataxia, and tremors) to severe (seizure-like activity and coma).5 Ivermectin is a member of the avermectin class of P-gp substrates, and it is commonly used in veterinary medicine to treat and control parasitic infections.6,7 Sequence analysis of ABCB1 cDNA from ivermectin-sensitive dogs identified a 4-bp deletion, commonly referred to as ABCB1-1Δ, which results in a truncated, nonfunctional protein.8 Collies and other dogs harboring this mutation can develop severe and life-threatening neurotoxic effects because of an accumulation of ivermectin in brain tissue. In Collies, the frequency of the mutant allele is as high as 58%, and it is also present in several other breeds, including Australian Shepherds, Old English Sheepdogs, and Longhaired Whippets.9–11

Because ABCB1-1Δ mutant dogs have an increased risk of developing neurotoxic responses to avermectins, the safety of new drugs in this class can be tested in ivermectin-sensitive Collies. Owing to a scarcity of such Collies, it is becoming difficult to conduct such critical safety evaluations. Recently, Orzechowski et al7 reported on the development of a novel in vivo experimental knock-in and knock-out mouse model involving mice that have the mutant canine ABCB1 gene (ABCB1-1Δ), which can be used as a potential alternative means for assessment of avermectin toxicity in dogs. Experiments involving such mice may reduce the use of dogs in avermectin safety studies. In the mutant mice, the mouse Abcb1a isoform was replaced by the canine mutated ABCB1-1Δ cDNA via a gene-targeting knock-in approach. To avoid any artifactual results from the presence of the murine Abcb1b isoform, this isoform was further disrupted through a gene-targeting knock-out strategy. The resulting transgenic mice express the mutant canine ABCB1 gene under the control of the mouse Abcb1a promoter. By the same approach, a knock-in and knock-out mouse strain expressing the control wild-type canine ABCB1 gene, ABCB1-WT, was established by one of the authors (MDS). Following administration of P-gp substrates ivermectin, doramectin, moxidectin, or digoxin, transgenic ABCB1-1Δ mutant mice developed neurotoxic signs of ataxia, lethargy, and tremors similar to signs observed in ivermectin-sensitive Collies with the ABCB1-1Δ mutation.7,12 Therefore, this experimental mouse model has the potential to be used in lieu of ivermectin-sensitive Collies to assess the safety of avermectin and nonavermectin classes of P-gp substrates.

However, the molecular mechanisms underlying the neurotoxic effects caused by P-gp substrates in transgenic mice are not known. The purpose of the study reported here was to identify biomarkers of P-gp substrate neurotoxicity in transgenic ABCB1-1Δ mutant mice. The intent was to evaluate, at the genomic level, the safety of the P-gp substrates ivermectin, digoxin, moxidectin, and doramectin when used in mice with either the canine ABCB1-1Δ mutant gene or the wild-type canine ABCB1 gene. To this end, the changes in gene expression in these transgenic mice following drug treatment were evaluated by use of a whole-genome gene expression microarray, and the altered genes were assessed for their association in gene pathways and networks.

Materials and Methods

Animals—The ABCB1 knock-in and knock-out mice with a targeted canine ABCB1-1Δ gene insertion were generated and characterized as previously reported.7 The ABCB1 knock-in and knock-out mice with a targeted canine wild-type ABCB1 gene insertion were established via the same approach (MDS, unpublished data). The C57BL/6J mouse strain was used to establish the ABCB1 knock-in and knock-out transgenic mouse colony.

Eight ABCB1-1Δ mutant mice and 8 ABCB1-WT mice were used in the study. These mice ranged in age from 17 to 30 weeks old and included both sexes. Previous studies7,12 found no correlation between time to onset or severity of clinical signs and adult mouse age or sex. All animal protocols were approved and monitored by the US FDA Center for Food Safety and Applied Nutrition Institutional Animal Care and Use Committee.

Chemicals and P-gp substrate administration—The following P-gp substrates were used in the study: ivermectin,a doramectin,b moxidectin,c and digoxin.d Four experiments were conducted, each of which involved 2 ABCB1-1Δ mutant mice and 2 ABCB1-WT mice. The mice in each group were administered a single dose of one of the P-gp substrates as described previously.7,12 Moxidectin (1 mg/mL) was administered by oral gavage to provide a dose of 10 mg/kg. Digoxin, doramectin, and ivermectin were administered SC. Doramectin (10 mg/mL) was diluted in propylene glycol (1, 2-propanediol)b to provide an injectable dose of 10 mg/kg in a total volume of 100 μL. Digoxin (0.25 mg/mL) was injected without dilution to provide an injectable dose of 1.53 mg/kg. The injectable ivermectin (10 mg/mL) contained solubilized 1% ivermectin formulated in 40% glycerol formal and 59% propylene glycol. Dilutions of the ivermectin stock were prepared in propylene glycol to generate concentrations that would provide doses of 10 mg/kg administered in a total volume of 100 μL.

Clinical sign observation—All mice were observed for clinical signs of toxicosis prior to drug administration (time 0) and at 1-hour intervals through 7 hours after the P-gp substrates were given. A predetermined scoring system for the clinical signs of ataxia, lethargy, and tremors was established previously.12 Scores of 0, 1, 2, and 3 were assigned for no, mild, moderate, and severe clinical signs, respectively. The 16 mice were euthanized 7 hours after drug administration by asphyxiation with carbon dioxide. Mouse brain tissue samples were then collected, and the tissues were stored in tissue containers.e Any mouse observed to be in undue pain or distress was euthanized prior to the 7-hour study endpoint. All clinical sign scores were recorded.

RNA extraction—Total RNA from mouse brain tissues was extracted with a tissue RNA extraction kitf following the manufacturer's instructions. The RNA quality and quantity were analyzed with a spectrophotometerg and bioanalyzer.h Equal amounts (1 μg) of RNA from 2 mice/genotype treated with each individual P-gp substrate were pooled together for microarray analysis. Separate RNA pools for each combination of genotype and drug treatment were generated.

Gene expression microarray and data analysis—Total RNA (1 μg) was reverse transcribed with T7-oligo(dT) primer and labeled with biotin by use of a labeling kiti following the manufacturer's protocol. The RNA was then labeled and hybridized to microarray gene assay arrayj and scanned on a gene assay array scanner.k Data were collected with the aid of commercial software.l

The microarray data (CEL files) were analyzed with commercial software,m with which data were processed and normalized by means of quantile normalization, and the probe set quantification was estimated by the iter-plier method and filtered by dabg (ie, detection above background) flags. The unsupervised hierarchical clustering, which is an unbiased and data-driven clustering of samples, and principal components analysis were performed with commercial software.n The differentially expressed genes were determined by means of the ANOVA method provided by commercial software.n,o

Pathway analysis of differentially expressed genes—Genes identified by gene expression microarray were analyzed with the aid of commercial softwarep to identify gene signaling pathways and networks. Genes with a ≥ 2-fold gene expression change (both up- and downregulated genes) were analyzed by means of the core analysis function included in the software to identify both direct and indirect relationships of genes in the context of biological processes, pathways, and networks. The top gene networks identified by the software were ranked by the computation of P scores, which were derived from P values derived with the Fisher exact test. The top biological functions of genes identified by the software were ranked by the computation of P values derived with the Fisher exact test and were grouped in the categories of disease and disorders, molecular and cellular functions, and physiologic system development and function. Data presented were selected on the basis of a value of P < 0.05.

Results

Clinical signs of neurotoxicosis observed in ABCB1-1Δ mutant mice following administration of P-gp substrates—Two ABCB1-1Δ mutant mice and 2 ABCB1-WT mice were each successfully treated with a single dose of one of the P-gp substrates ivermectin, doramectin, moxidectin, or digoxin. After administration of the P-gp substrate, each mouse was monitored for any physiologic and behavioral changes at 1-hour intervals for up to 7 hours. Consistent with previous reports,7,12 the ABCB1-1Δ mutant and the ABCB1-WT mice had considerable differences in their response to the P-gp substrates. Although the ABCB1-WT mice remained unaffected after administration of ivermectin, doramectin, or moxidectin, the ABCB1-1Δ mutant mice developed mild to severe signs of ataxia, lethargy, and tremors. Notably, the ABCB1-WT mice had mild to severe clinical signs of ataxia, lethargy, and tremor after administration of digoxin, but to a lesser extent than did the ABCB1-1Δ mutant mice (data not shown). No mouse was observed to be in undue pain or distress at any time.

Gene expression profiling of ABCB1-1Δ mutant mice following administration of P-gp substrates—Unsupervised hierarchical clustering analysis of the whole-genome gene expression data separated the samples into 2 major clusters associated with the mouse genotype (Figure 1). The clustering revealed that all P-gp substrate-treated samples derived from ABCB1-1Δ mutant mice shared a similar gene expression pattern, regardless of the P-gp substrate administered. These samples formed a major cluster (cluster I). Similarly, all of the samples derived from ABCB1-WT mice formed another major cluster (cluster II), with the exception of 1 sample from the moxidectin-treated ABCB1-WT group (designated as a WT-moxidectin sample); however, 3-D mapping of all of the samples by means of principal components analysis revealed that this WT-moxidectin sample was still more closely related by distance to the ABCB1-WT group (cluster II) than to the ABCB1-1Δ mutant group (cluster I).

Figure 1—
Figure 1—

Gene expression profiling of ABCB1-1Δ mutant mice and ABCB1-WT mice following administration of P-gp substrates digoxin, doramectin, ivermectin, or moxidectin. Groups of 2 ABCB1-1Δ mutant mice and 2 ABCB1-WT mice were administered digoxin (1.53 mg/kg, SC), doramectin (10 mg/kg, SC), ivermectin (10 mg/kg, SC), or moxidectin (10 mg/kg, PO) and monitored hourly for 7 hours before euthanasia. A toxicogenomic approach based on DNA microarrays was used to examine whole-genome expression changes in the treated mice. A—Unsupervised hierarchical clustering analysis of whole-genome gene expression for ABCB1-1Δ mutant mice and ABCB1-WT mice administered P-gp substrates. The clustering revealed that all P-gp substrate–treated samples derived from ABCB1-1Δ mutant mice shared a similar gene expression pattern (regardless of the P-gp substrate administered) and formed a major cluster (cluster I). Similarly, all of the samples derived from ABCB1-WT mice formed another major cluster (cluster II), with the exception of 1 sample from the moxidectin-treated ABCB1-WT group. B—Results of principal components analysis of all samples. The 3 axes PC#1, PC#2, and PC#3 are the 3 principle components representing gene expression variations of the dataset. The first component explains the maximal amount of variance in the data. The second principal component is perpendicular to the first one and explains maximum residual variance. The third component is perpendicular to the first 2 and explains the maximum residual variance. The normalized gene expression value of the dataset is plotted along each axis, and the percentage of gene expression variance of the dataset is indicated along each axis. The identification number of each sample is indicated in the image.

Citation: American Journal of Veterinary Research 75, 12; 10.2460/ajvr.75.12.1104

For each of the P-gp substrate treatment groups, the gene expression profiles of ABCB1-1Δ mutant mice were further compared with those of the ABCB1-WT mice. Gene expression analysis revealed that genes were altered in ABCB1-1 mutant mice following administration of P-gp substrates. Genes with a ≥ 2-fold change in ABCB1-1Δ mutant mice (compared with findings in ABCB1-WT mice), including both up- and downregulated genes, were selected for further analysis. Among the 4 P-gp substrates, moxidectin had a greater effect on gene expression in the ABCB1-1Δ mutant mice than did ivermectin, digoxin, or doramectin. Moxidectin treatment resulted in 1,612 genes that had a > 2-fold gene expression change, which was approximately 5 times that of the other 3 P-gp substrates, with 272 genes for ivermectin, 363 genes for doramectin, and 372 genes for digoxin (Figure 2).

Figure 2—
Figure 2—

Venn diagram identifying common genes that were altered in ABCB1-1Δ mutant mice following administration of each of the P-gp substrates ivermectin, doramectin, moxidectin, and digoxin (2 mice/treatment). Genes with a ≥ 2-fold gene expression change, compared with findings for the ABCB1-WT mice, were used for comparison. Thirty-four genes were altered in all 4 treatment groups.

Citation: American Journal of Veterinary Research 75, 12; 10.2460/ajvr.75.12.1104

Gene pathway analysis of differentially expressed genes—Genes that were altered (≥ 2-fold gene expression change) in ABCB1-1Δ mutant mice, compared with findings for their counterpart ABCB1-WT mice, were further analyzed for their involvement in signaling pathways and networks in pathway analysis software. The pathway analysis revealed that altered genes from each drug treatment group were associated with behavior and nervous system development and function as the top-ranked gene networks. When the gene networks from all 4 drug treatment groups were compared, common genes were identified as the hub connecting different genes and gene signaling pathways. The Slc17a6 → Drd2 → Th and Gata3 → Pet → Th signaling pathways were involved in both ivermectin- and digoxin-associated gene networks (Figure 3).

Figure 3—
Figure 3—

Results of gene pathway analysis of altered gene signaling pathways and networks in ABCB1-1Δ mutant mice following treatment with ivermectin or digoxin. Genes with a ≥ 2-fold gene expression change, compared with findings for the ABCB1-WT mice, were used for pathway analysis. The schematics illustrate the top-ranked gene network associated with behavior from ivermectin-treated mice (A) and digoxin-treated mice (B). Data were analyzed through the use of QIAGEN's Ingenuity® Pathway Analysis (IPA®, QIAGEN Redwood City, www.qiagen.com/ingenuity).

Citation: American Journal of Veterinary Research 75, 12; 10.2460/ajvr.75.12.1104

The pathway analysis also searched for the top biological functions with which these altered genes were involved. Consistent with the top-ranked gene network with which these genes were associated, many of the altered genes from each P-gp substrate treatment group were critical for physiologic system development and function such as behavior and nervous system development and function. For instance, among the 272 altered genes from the ivermectin treatment group, 49 genes were associated with behavior and 58 genes were associated with nervous system development and function. In addition to the role in physiologic system development, many of the altered genes from the doramectin, moxidectin, and digoxin treatment groups were identified as being involved in neurologic diseases. For instance, 170 of the total 1,612 altered genes from the moxidectin treatment group were involved in neurologic diseases such as neuromuscular disease, Huntington's disease, seizures, and encephalitis.

The altered gene expression list (≥ 2-fold gene expression change) from all 4 drug treatment groups was further compared via Venn diagram analysis to identify commonly altered genes. Thirty-four genes were altered in all 4 treatment groups (Figure 2). The gene pathway analysis revealed that these genes are associated with behavior, endocrine system development, and molecular transport as the top-ranked gene network. Ten of the 34 genes were of particular interest for further validation by use of real-time RT-PCR (Table 1).

Table 1—

Ten of the 34 commonly altered genes in 8 ABCB1-1 mutant mice following administration (2 mice/treatment) of P-gp substrates digoxin (1.53 mg/kg, SC), doramectin (10 mg/kg, SC), ivermectin (10 mg/kg, SC), or moxidectin (10 mg/kg, PO).

   Fold change in gene expression (mutant vs wild-type)
Gene symbolAccession No.Gene ontology biological processIvermectinDigoxinDoramectinMoxidectin
Glra1NM_020492Transport, neuropeptide signaling pathway development−14.4−35.6−4.1−132.6
Mab21l2NM_011839Multicellular organismal−5.7−6.3−2.1−45.9
Ebf3NM_010096DNA-templated transcription−15.3−11.1−4.5−29.0
Slc10a4NM_173403Ion transport transport, sodium ion transport−15.7−8.4−10.8−22.7
Slc18a2NM_172523Drug transport transport, neurotransmitter transport−22.7−8.0−8.0−10.1
TtrNM_013697Transport2.4−11.510.3−48.1
DaoNM_010018Dopamine biosynthetic process−3.4−5.9−8.5−25.5
Klk6NM_011177Regulation of neuron projection development−4.8−9.8−4.5−24.0
GabrqNM_020488Transport, chloride transport, ion transport−15.1−3.3−13.4−10.7
AlbNM_009654Albumin, drug transport5.0110.6158.536.9

Each treatment group included 2 ABCB1-1Δ mutant mice. The 10 genes were selected from 34 genes that were altered by all 4 treatments by calculating the mean gene expression fold change of each of the 34 genes from all 4 drug treatment groups and ranking the mean gene expression fold changes and selecting the top 10 genes. Gene products were as follows: Glra1, glycine receptor, alpha-1 subunit; Mab2112, Mab-21-like 2 (Caenorhabditis elegans); Ebf3, early B-cell factor 3; Slc10a4, solute carrier family 10 (sodium/bile acid cotransporter family), member 4; Slc18a2, solute carrier family 18 (vesicular monoamine), member 2; Ttr, transthyretin; Dao, D-aminoacidoxidase; Klk6, kallikrein related-peptidase 6; Gabrq, γ-aminobutyric acid A receptor, subunit theta; and Alb, albumin.

Discussion

The present study, conducted with a toxicogenomic approach, focused on the canine ABCB1-1Δ gene mutation and the corresponding safety of administered P-gp substrates in mice with either that mutant gene or the wild-type canine gene. Results of the study confirmed findings from previous investigations7,12 in which the ABCB1-1Δ mutant mice were sensitive to the P-gp substrates ivermectin, doramectin, moxidectin, and digoxin and developed similar neurotoxic clinical signs (data not shown) of ataxia, lethargy, and tremors. By means of whole-genome gene expression microarray analysis, we distinguished the gene expression pattern of ABCB1-1Δ mutant mice from that of ABCB1-WT mice after administration of P-gp substrates, further demonstrating that ABCB1-1Δ mutant mice were more sensitive to the toxic effects of P-gp substrate exposure.

Ivermectin, doramectin, and moxidectin belong to the avermectin family of P-gp substrates and are used as antiparasitic drugs in veterinary medicine.6,7 The avermectins block the electrical activity of nerve and muscle cells, which causes an influx of chloride ions into the cells, leading to hyperpolarization of the cells and subsequent paralysis and death of the parasites.13–15 Digoxin is a nonavermectin drug used for treating heart disease. The changes in gene expression found in ABCB1-1Δ mutant mice in response to the P-gp substrates have provided information regarding the molecular mechanisms underlying the neurotoxic clinical signs in the treated mice. Gene pathway analysis of the altered genes revealed that genes in mice from all 4 drug treatment groups were associated with behavior and nervous system development and function as the top-ranked gene network, which may explain the neurotoxic clinical signs of ataxia, lethargy, and tremors in the ABCB1-1Δ mutant mice following treatment. Slc17a6 and Gata3 were involved in the altered signaling pathways in mice in the ivermectin and digoxin treatment groups. Slc17a6 is expressed in glutamatergic neurons and acts to package glutamate into presynaptic vesicles. Glutamate is the most abundant excitatory neurotransmitter in the adult CNS,16 and glutamatergic neurons and glutamate-mediated excitatory signaling are implicated in all neuronal circuits of the CNS.17 Defective or altered glutamate signaling has been associated with several diseases, including Parkinson's disease, schizophrenia, Alzheimer's disease, amyotrophic lateral sclerosis, and depression.18–20 Gata3 encodes a transcription factor with important roles in the development of specific neuronal populations, including neurons in the inner ear, midbrain, hind brain, spinal cord, and sympathoadrenal system.21 Tsarovina et al21 demonstrated that the trans-acting T-cell–specific transcription factor GATA-3 is required for maintaining survival of embryonic and adult sympathetic neurons, which reveals a survival control mechanism in mature neurons, and is of particular interest for neuronal populations affected in neurodegenerative diseases.

Notably, the 34 genes altered in the mice in all 4 drug treatment groups were also associated with behavior, suggesting that these genes are essential for downstream neural signaling responses in the gene pathways. These genes are potentially useful as biomarkers of P-gp substrate neurotoxicity in ABCB1-1Δ mutant mice. Among the commonly altered genes, the expression of Gabrq was downregulated in ABCB1-1Δ mutant mice in all 4 drug treatment groups. Gabrq encodes a GABA A receptor; it is a multisubunit chloride channel that mediates the fastest inhibitory synaptic transmission in the CNS. The avermectins, such as ivermectin, act as GABA receptor antagonists. The blocking of GABA receptors by avermectins causes an influx of chloride ions into the cells, leading to hyperpolarization and subsequent paralysis of the neuromuscular systems.13–15 The toxicity of avermectins is related to their binding with GABA receptors in the CNS of the mammalian host.6

In all 4 drug treatment groups in the present study, the expression of Alb was highly upregulated in ABCB1-1Δ mutant mice, compared with findings in ABCB1-WT mice. Albumin is the most abundant plasma protein and is a major component of extracellular fluids including CSF, interstitial fluid, and lymph. Albumin functions primarily as a carrier protein for steroids, fatty acids, thyroid hormones, drugs, and metabolites in plasma. Albumin has been implicated in many neurologic diseases because of its ability to regulate hemodynamic properties of brain circulation and direct neuroprotective actions on neuronal and glial cells.22 In the absence of a functional P-gp, which acts as a pump for xenobiotic compounds and endogenous metabolites, there is an upregulation of Alb in the brain tissues of ABCB1-1Δ mutant mice. This increase of Alb gene expression in the brain tissue may lead to damage of the blood-brain barrier. The increased synthesis of albumin in mouse brains may help to mediate the neuroprotective actions on the damaged neuronal system.

However, the expression of these genes needs to be further validated by real-time RT-PCR analysis in individual ABCB1-1Δ mutant mice and ABCB1-WT mice that have been treated with P-gp substrates. Although ABCB1-1Δ mutant mice7 and ABCB1-WT mice were considered normal and the lineage was free of adverse effects on fertility, physiology, anatomy, or mortality rate, the expression of the selected genes needs to be further examined in ABCB1-1Δ mutant mice and ABCB1-WT mice not treated with P-gp substrates, to ascertain that the ABCB1-1 gene mutation itself does not affect gene expression in these mice.

The 4 P-gp substrates used in the present study have been documented to induce neurotoxic effects in dogs with the ABCB1-1Δ mutation.8,23 The clinical signs of neurotoxicosis observed in ABCB1-1Δ mutant mice in response to these drugs closely resembled the clinical signs observed in ABCB1-1Δ mutant Collies following treatment.7,12 As indicated by the results of the present study, the whole-genome gene expression profile of ABCB1-1Δ mutant mice may predict the gene expression changes in ABCB1-1Δ mutant dogs. Together, these features make this experimental mouse model a potential resource in lieu of the use of Collies for drug safety studies. The potential biomarkers of P-gp substrate neurotoxicity identified in ABCB1-1Δ mutant mice may help to predict the safety of P-gp substrates in dogs.

By use of a toxicogenomic approach to examine changes in whole-genome expression, the results of the present study indicated that gene expression is altered in ABCB1-1Δ mutant mice following administration of the P-gp substrates ivermectin, doramectin, moxidectin, or digoxin. Altered genes cause changes in signaling pathways and networks associated with behavior and nervous system development and function. These study findings should aid in understanding the molecular mechanisms of P-gp substrate–associated neurotoxicosis in ABCB1-1Δ mutant mice. Of the 34 common genes altered in ABCB1-1Δ mutant mice in response to ivermectin, doramectin, moxidectin, and digoxin, genes such as Gabrq, Dao, and Alb may be potentially useful as biomarkers of P-gp substrate neurotoxicity to predict the safety of such drugs in dogs with the ABCB1 gene mutation.

ABBREVIATIONS

GABA

γ-Aminobutyric acid

P-gp

P-glycoprotein

RT-PCR

Reverse transcription PCR

a.

Sigma Aldrich Corp, St Louis, Mo.

b.

Pfizer Inc, New York, NY.

c.

Fort Dodge Animal Health, Overland Park, Kan.

d.

Baxter International Inc, Deerfield, Ill.

e.

PAXgene Tissue Containers, Qiagen, Valencia, Calif.

f.

PAXgene Tissue RNA kit, Qiagen, Valencia, Calif.

g.

Nanodrop spectrophotometer, ThermoFisher Scientific Inc, Wilmington, Del.

h.

Agilent Technologies Inc, Santa Clara, Calif.

i.

One Cycle Target Labeling kit, Affymetrix Inc, Santa Clara, Calif.

j.

GeneChip Mouse Gene 1.0 ST array, Affymetrix Inc, Santa Clara, Calif.

k.

GeneChip scanner 3000, Affymetrix Inc, Santa Clara, Calif.

l.

GCOS software, Affymetrix Inc, Santa Clara, Calif.

m.

Power Tool APT, version 1.15.0, Affymetrix Inc, Santa Clara, Calif.

n.

Partek Genomic Suite, Partek Inc, St Louis, Mo.

o.

Limma data analysis package, Bioconductor, Fred Hutchinson Cancer Research Center, Seattle, Wash.

p.

IPA, Ingenuity Systems, Redwood City, Calif.

References

  • 1. Schinkel AH, Mayer U, Wagenaar E, et al. Normal viability and altered pharmacokinetics in mice lacking mdr1-type (drug-transporting) P-glycoproteins. Proc Natl Acad Sci U S A 1997; 94: 40284033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2. Schinkel AH. The physiological function of drug-transporting P-glycoproteins. Semin Cancer Biol 1997; 8: 161170.

  • 3. Kitamura Y, Koto H, Matsuura S, et al. Modest effect of impaired P-glycoprotein on the plasma concentrations of fexofenadine, quinidine, and loperamide following oral administration in collies. Drug Metab Dispos 2008; 36: 807810.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4. Schinkel AH, Smit JJ, van Tellingen O, 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 1994; 77: 491502.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5. Paul AJ, Tranquilli WJ, Seward RL, et al. Clinical observations in Collies given ivermectin orally. Am J Vet Res 1987; 48: 684685.

  • 6. Dadarkar SS, Deore MD, Gatne MM. Comparative evaluation of acute toxicity of ivermectin by two methods after single subcutaneous administration in rats. Regul Toxicol Pharmacol 2007; 47: 257260.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7. Orzechowski KL, Swain MD, Robl MG, et al. Neurotoxic effects of ivermectin administration in genetically engineered mice with targeted insertion of the mutated canine ABCB1 gene. Am J Vet Res 2012; 73: 14771484.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8. Mealey KL, Bentjen SA, Gay JM, et al. Ivermectin sensitivity in Collies is associated with a deletion mutation of the mdr1 gene. Pharmacogenetics 2001; 11: 727733.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9. Hugnet C, Bentjen SA, Mealey KL. Frequency of the mutant MDR1 allele associated with multidrug sensitivity in a sample of Collies from France. J Vet Pharmacol Ther 2004; 27: 227229.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10. Kawabata A, Momoi Y, Inoue-Murayama M, et al. Canine mdr1 gene mutation in Japan. J Vet Med Sci 2005; 67: 11031107.

  • 11. Neff MW, Robertson KR, Wong AK, et al. Breed distribution and history of canine mdr1–1Delta, a pharmacogenetic mutation that marks the emergence of breeds from the Collie lineage. Proc Natl Acad Sci U S A 2004; 101: 1172511730.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12. Swain MD, Orzechowski KL, Swaim HL, et al. P-gp substrate-induced neurotoxicity in an Abcb1a knock-in/Abcb1b knock-out mouse model with a mutated canine ABCB1 targeted insertion. Res Vet Sci 2013; 94: 656661.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 13. Bloomquist JR. Toxicology, mode of action and target sitemediated resistance to insecticides acting on chloride channels. Comp Biochem Physiol C 1993; 106: 301314.

    • Search Google Scholar
    • Export Citation
  • 14. Bloomquist JR. Ion channels as targets for insecticides. Annu Rev Entomol 1996; 41: 163190.

  • 15. Bloomquist JR. Chloride channels as tools for developing selective insecticides. Arch Insect Biochem Physiol 2003; 54: 145156.

  • 16. Fonnum F. Glutamate: a neurotransmitter in mammalian brain. J Neurochem 1984; 42: 111.

  • 17. Moriyama Y & Yamamoto A. Glutamatergic chemical transmission: look! Here, there, and anywhere. J Biochem 2004; 135: 155163.

  • 18. Chen J, Lipska BK, Weinberger DR. Genetic mouse models of schizophrenia: from hypothesis-based to susceptibility gene-based models. Biol Psychiatry 2006; 59: 11801188.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19. Kashani A, Lepicard E, Poirel O, et al. Loss of VGLUT1 and VGLUT2 in the prefrontal cortex is correlated with cognitive decline in Alzheimer disease. Neurobiol Aging 2008; 29: 16191630.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20. Nestler EJ, Barrot M, DiLeone RJ, et al. Neurobiology of depression. Neuron 2002; 34: 1325.

  • 21. Tsarovina K, Reiff T, Stubbusch J, et al. The Gata3 transcription factor is required for the survival of embryonic and adult sympathetic neurons. J Neurosci 2010; 30: 1083310843.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 22. Prajapati KD, Sharma SS & Roy N. Current perspectives on potential role of albumin in neuroprotection. Rev Neurosci 2011; 22: 355363.

  • 23. Fecht S & Distl O. Review of prevalence, genetic aspects and adverse effects of the mdr1–1Delta mutation in dogs. Dtsch Tierarztl Wochenschr 2008; 115: 212219.

    • Search Google Scholar
    • Export Citation

Contributor Notes

The authors thank Dr. Robert Stephens for support with microarray analysis.

Address correspondence to Dr. Zhu (min.zhu@fda.hhs.gov).