Pharmacogenomics of poor drug metabolism in Greyhounds: Cytochrome P450 (CYP) 2B11 genetic variation, breed distribution, and functional characterization

Greyhounds recover more slowly from certain injectable anesthetics than other dog breeds. Previous studies implicate cytochrome P450 (CYP) 2B11 as an important clearance mechanism for these drugs and suggest Greyhounds are deficient in CYP2B11. However, no CYP2B11 gene mutations have been identified that explain this deficiency in Greyhounds. The objectives of this study were to provide additional evidence for CYP2B11 deficiency in Greyhounds, determine the mechanisms underlying this deficiency, and identify CYP2B11 mutations that contribute to this phenotype in Greyhounds. Greyhound livers metabolized CYP2B11 substrates slower, possessed lower CYP2B11 protein abundance, but had similar or higher mRNA expression than other breeds. Gene resequencing identified three CYP2B11 haplotypes, H1 (reference), H2, and H3 that were differentiated by mutations in the gene 3′-untranslated region (3′-UTR). Compared with 63 other dog breeds, Greyhounds had the highest CYP2B11-H3 allele frequency, while CYP2B11-H2 was widely distributed across most breeds. Using 3′-UTR luciferase reporter constructs, CYP2B11-H3 showed markedly lower gene expression (over 70%) compared to CYP2B11-H1 while CYP2B11-H2 expression was intermediate. Truncated mRNA transcripts were observed in CYP2B11-H2 and CYP2B11-H3 but not CYP2B11-H1 transfected cells. Our results implicate CYP2B11 3′-UTR mutations as a cause of decreased CYP2B11 enzyme expression in Greyhounds through reduced translational efficiency.

Dog breed differences in hepatic CYP2B11 protein and mRNA. Microsomal CYP2B11 protein content and CYP2B11 mRNA abundance were measured in the same set of Greyhound, Beagle and mixed-breed dog liver samples (n = 5 livers per breed). As shown in Fig. 3a, significant breed associated differences in CYP2B11 content were observed (P < 0.001, ANOVA). Greyhound livers showed the lowest content, Beagle livers had the highest content, and mixed-breed livers were intermediate. On the other hand, CYP2B11 mRNA abundance in Greyhound livers was similar to Beagle livers (P > 0.05, Holm-Sidak test) and substantially higher than mixed-breed livers (P = 0.008; Holm-Sidak test) (Fig. 3b).
Identification of CYP2B11 genetic polymorphisms. Selected regions of the CYP2B11 gene, including the 5′-enhancer (to ~2,000 bp upstream), all 9 exons, and the complete 3′-untranslated region (UTR) were sequenced using DNA obtained from 13 Greyhounds, including the 5 Greyhounds used for liver samples. Sequence variants were identified by comparison to the current canine reference sequence (CanFam3.1) and compared to polymorphisms identified by analysis of publicly available whole genome sequence data from another 45 dogs representing 45 different breeds. Identified polymorphisms and the genotypes of individual dogs are given in Supplementary Table S1. These data are summarized as variant allele frequencies (with 95% confidence intervals) for the 13 Greyhounds and the 45 dogs from other breeds in Table 2. Nine genetic polymorphisms were identified, three of which were found in the dbSNP public database (rs21894687, rs852076551, and rs850924485). One polymorphism was located in the 5′-enhancer region (c.-489 G/A), one polymorphism was a synonymous SNP in exon 7 (c.966G/A), while the remaining 7 polymorphisms were clustered together in the 3′-UTR from cDNA positions 1913 to 2536. Allele frequencies for all but one of the 3′-UTR polymorphisms were more than 2-fold higher in the 13 Greyhounds compared to the 45 other dogs. One 3′-UTR polymorphism (c.2498G/T) was not found in any of the 13 Greyhounds evaluated. Breed differences in liver microsome CYP marker activities. CYP activities selective for the major drug metabolizing CYP enzymes were measured using liver microsomes obtained from Beagles (n = 5), mixedbreed dogs (n = 5) and Greyhounds (n = 5). Bars represent the mean and standard error of the activity values for individual Beagle and Greyhound liver microsomes expressed as a percentage of the mean activity of mixedbreed dog liver microsomes. *P < 0.05 by Student's t-test on log transformed data comparing Greyhound dog liver activities with Beagle and mixed-breed dog liver activities. Samples from Greyhound dogs were identified by their owners as dogs registered with the National Greyhound Association bred for racing. CYP2B11 haplotype analysis. Linkage disequilibrium analysis indicated strong linkage across the CYP2B11 gene (spanning about 16 kilobases) for most polymorphisms in both Greyhounds (Fig. 4a) and dogs from 45 other breeds (Fig. 4b). Exceptions were c.2498G/T, which was associated only with the exon 7 SNP and partially with the 5′-enhancer polymorphism, while the 3′-UTR SNP c.1952 C/T was not associated with any of the other polymorphisms.
Six haplotypes (designated CYP2B11-H1 to -H6) could be inferred from genotype data for all dogs (listed in Table 3). Three haplotypes were found in Greyhounds (CYP2B11-H1, H2 and H3). CYP2B11-H1 and -H2 were the two most common haplotypes found in both Greyhounds and other dog breeds, although CYP2B11-H2 predominated (50% frequency) in Greyhounds, while CYP2B11-H1 predominated (62% frequency) in other breeds. The other haplotype found in Greyhounds (CYP2B11-H3) was much more common in Greyhounds (19% frequency) compared with other breeds (3% frequency). Apart from Greyhounds, CYP2B11-H3 was found in a Whippet (homozygous) and a Border Collie (heterozygous CYP2B11-H1/H3).   (Table 4) of haplotype frequencies between breeds that comprised the liver samples studied above (i.e. Beagles, NGA-registered Greyhounds, and mixed-breed dogs) showed similar H2 frequencies across the three breeds (21-26%), but a much higher H3 frequency in NGA-registered Greyhounds (18%) compared with mixed-breed dogs (2%). The H3 haplotype was not found in any genotyped Beagle dog samples. Interestingly, AKC-registered Greyhounds were quite different from the NGA-registered Greyhounds in that they lacked the H2 haplotype and had the highest H3 frequency of all breeds sampled (59%).
A broader evaluation of haplotype frequencies across Sighthound and non-Sighthound breed groups is shown in Fig. 5. The H2 haplotype was widely distributed across most breeds and was detected in all 19 (100%) of the Sighthound breeds sampled as well as in 41 of 45 (91%) non-Sighthound breeds. Furthermore, average (±SE) H2 frequency calculated for the breed groups was similar (P > 0.05, Mann-Whitney U test) in Sighthound (25 ± 6%) compared with non-Sighthound breeds (20 ± 3%). On the other hand, the H3 haplotype was more restricted in breed distribution, being found in 10 of 19 (53%) Sighthound breeds and only 10 of 45 (22%) non-Sighthound breeds. Furthermore, average haplotype frequency in Sighthound breeds (9 ± 3%) was over 4-fold higher (P = 0.003, Mann-Whitney U test) compared with non-Sighthound breeds (1.7 ± 0.7%). CYP2B11 mRnA splicing. To explore the mechanism underling CYP2B11 expression variability, whole transcriptome sequencing (RNA-seq) analysis was conducted using total RNA extracted from the same 5 Greyhound and 5 Beagles livers used for determining CYP activities and CYP2B11 mRNA quantitation to evaluate variation in mRNA splicing of CYP2B11 gene transcripts. Mapping with transcript analysis identified only a single transcript in all samples that was identical in mRNA length exon structure to the CYP2B11 reference sequence in Genbank (NM_001006652). No alternate splice forms were found. CYP2B11 mRnA allelic imbalance. A potential role for cis-acting regulatory genetic polymorphisms in CYP2B11 gene expression was evaluated by assessment of allelic imbalance using RNA-seq data for a subset of the previously studied liver samples that were found to be heterozygous with CYP2B11-H1 for the CYP2B11-H2 allele (3 Beagles and 3 Greyhounds) and for the CYP2B11-H3 allele (1 Greyhound). To account for mapping efficiency differences, RNA allelic ratios at each variant position were normalized using DNA allelic ratios obtained from whole genome DNA sequence data for 5 other dogs with the CYP2B11-H1/H2 genotype and one other dog with the CYP2B11-H1/H3 genotype. As shown in Fig. 6, dramatically lower RNA expression (mean ratios of 0.05 to 0.15) was observed for the CYP2B11-H2 allele relative to the CYP2B11-H1 allele for 2 of the 6 polymorphisms (c.2137 TG/CA and c.2166G/A) in both Greyhound and Beagle livers. CYP2B11-H3 expression was slightly lower (ratio of 0.7) than CYP2B11-H1 at the single SNP (c.1952 C/T) associated with this haplotype.  www.nature.com/scientificreports www.nature.com/scientificreports/ CYP2B11 3′-UtR transcript length variation. Reverse transcriptase PCR was used to determine the approximate length of the 3′ end of the CYP2B11-3′UTR reporter mRNA in MDCK cells transfected with each of the luciferase reporter constructs. PCR primers were designed to amplify the cDNA in 3 regions, from position c.1773 to c.1872 (Region 1), from c.1853 to c.2199 (Region 2) and from c.1853 to c.2312 (Region 3) (Fig. 8a). These regions were chosen to be upstream (5′) of the two polymorphisms (c.2137 TG/CA and c.2166G/A) that demonstrated significant allelic imbalance (Region 1), or to span these polymorphisms (Regions 2 and 3). Primers for GAPDH were also used to confirm RNA extraction and reverse transcription in each sample. Untransfected cells were assayed to exclude background CYP2B11 expression in the cell line.
As shown in Fig. 8b, the GAPDH primers resulted in bands of similar intensity in cells transfected with each CYP2B11-3′UTR reporter construct, as well as in untransfected cells. Strong bands were also detected with the Region 1 primers for all three CYP2B11-3′UTR reporter constructs, but not in untransfected cells. The Region 2 primers resulted in a strong band for CYP2B11-3′UTR-H1, but much weaker bands for CYP2B11-3′UTR-H2 and CYP2B11-3′UTR-H3. Furthermore, the Region 3 primers showed a strong band for CYP2B11-3′UTR-H1, but no bands for CYP2B11-3′UTR-H2 or CYP2B11-3′UTR-H3.
By combining the RT-PCR results with the RNA-seq allelic imbalance information, the approximate locations of the 3′ end of the mRNA for each CYP2B11 polymorphism were inferred (shown in Fig. 8c). For CYP2B11-3′UTR-H1, the data were consistent with the 3′end at c.2625 as given in the Genbank reference sequence  . Breed differences in CYP2B11 protein and mRNA. Microsomal CYP2B11 protein content (a) and CYP2B11 mRNA abundance (b) were measured in the same set of livers obtained from Beagles (n = 5), mixedbreed dogs (n = 5) and Greyhounds (n = 5). Data are expressed relative to the liver with the lowest value. Shown are box and whiskers plots summarizing data for individual dogs in each breed group. Significant differences between breed groups were identified by ANOVA on log transformed data (P < 0.05) for both CYP2B11 protein and mRNA. Shown for each set of data are the P-values for post hoc pairwise multiple comparisons testing (Holm-Sidak method). (2020) 10:69 | https://doi.org/10.1038/s41598-019-56660-z www.nature.com/scientificreports www.nature.com/scientificreports/ NM_001006652. For CYP2B11-3′UTR-H2, the 3′end is likely located between c.1913 and c.2138, while for CYP2B11-3′UTR-H3, it is likely between c.1952 and c.2199.
The CYP2B11 3′-UTR sequence was then evaluated for the presence of consensus polyadenylation signal sites. Two canonical polyadenylation signal sites (AAUAAA) were found. One site was located at c.2582, about 40 bp upstream of the predicted 3′ end of CYP2B11-3′UTR-H1, while the other site was at c.1715, about 200 bp upstream of the predicted ends of CYP2B11-3′UTR-H2 and CYP2B11-3′UTR-H3. None of the 3′-UTR polymorphisms appeared to create a novel consensus polyadenylation signal site or abolish an existing one. CYP2B11 diplotype association with activity, protein and mRnA. Differences in CYP2B11 enzyme activity, protein content, mRNA abundance, and protein/mRNA ratio (as an index of translation efficiency) between the 15 (previously studied) dog livers after grouping by CYP2B11 diplotype are shown in Fig. 9. Identified diplotypes included H1/H1 (2 Beagles and 3 mixed-breed), H1/H2 (3 Beagles, 2 mixed-breed, and 3 Greyhounds), H1/H3 (one Greyhound) and H3/H3 (one Greyhound). No dogs possessed the H2/H2 diplotype. Since there was only one dog liver with the H1/H3 diplotype and one dog liver with the H3/H3 diplotype, these data were grouped with the H1/H2 livers (10 livers total) for statistical comparison with the H1/H1 livers (5 livers). No differences in bupropion hydroxylation, CYP2B11 protein abundance or CYP2B11 mRNA expression were observed between H1/H1 and other diplotypes (P > 0.05, Mann-Whitney U test). However, CYP2B11 protein/mRNA values were significantly higher (P = 0.032, Mann-Whitney U test) in the H1/H1 group, compared with livers with other diplotypes with median (interquartile range) ratios of 5.6 (2.4-19) and 1.7 (1.1-3.2).

Discussion
Based on the results of the microsomal CYP activity marker assays and CYP2B11 immunoblotting, this study provides further evidence that CYP2B11 is deficient in Greyhounds. Furthermore, other CYPs involved in drug metabolism appear to be equally active, or, in the case of CYP2D15, perhaps even more active in Greyhounds compared with other dog breeds. Recombinant enzyme phenotyping indicated that propofol hydroxylation is largely mediated by CYP2B11, although with some involvement from CYP3A12. A role for CYP3A12 in propofol hydroxylation was confirmed by showing significant correlation of propofol hydroxylation activities with microsomal CYP3A protein content, although somewhat weaker than with CYP2B11 protein content. To rule out possible, perhaps additional, deficiency of CYP3A12 in Greyhounds, breed differences were evaluated using activity probes that were confirmed by recombinant enzyme phenotyping to be more selective than propofol hydroxylation for CYP2B11 (bupropion hydroxylation) and CYP3A12 (omeprazole sulfonation). Results using these latter probes suggest that CYP3A12 is not deficient in Greyhounds. These results have since been confirmed by us through quantitation of microsomal CYP protein concentrations using proteomic techniques that are more accurate and precise than immunoblotting 15 .
CYP2B11 mRNA concentrations in Greyhound livers were similar to Beagle livers and higher than mixed-breed livers indicating that low CYP2B11 activity and protein content in Greyhound livers was not a consequence of reduced gene transcription or mRNA instability, but could involve aberrant mRNA splicing or reduced translational efficiency. The most clinically important genetic polymorphism in human CYP2B6 Position a  112817078  112828499  112832580  112832619  112832805  112832834  112832951  112833166  112833204 Reference allele  Table 2. CYP2B11 genetic polymorphisms and allele frequencies. Genetic polymorphisms located in the CYP2B11 5′-enhancer (to ~2,000 bp upstream), exons 1-9, and 3′-UTR were identified by genomic PCR with Sanger sequencing (in 13 Greyhounds) or by analysis of publicly available whole genome sequence data by sampling one dog from each of 45 different breeds. Samples from Greyhound dogs were identified by their owners as dogs registered with the National Greyhound Association bred for racing. Shown are the locations of each polymorphism, predicted effect on the cDNA and protein, as well as the observed allele frequencies (95% confidence interval) in the Greyhounds and the dogs from the other breeds. Genotype data for each individual dog used to derive these allele frequencies are given in S1 Table. The genetic polymorphism labels used here (#1 to #9) correspond to the labels used in Fig. 4 and Table 3. a Position in base pairs in the CanFam 3.1 chromosome 1 sequence for the first nucleotide of the polymorphism.
mRNA allelic imbalance analysis has been used for a number of years in pharmacogenetic research and related disciplines to identify cis-acting polymorphisms that differentially alter expression levels of mRNA transcribed from different alleles 17 . Samples that are known to be heterozygous at polymorphic sites located within the transcript are typically used to enable direct comparison of the amount of variant transcript with the reference transcript within the same sample. Here, we used RNA-seq data from liver samples that were heterozygous for polymorphisms located within the CYP2B11 transcript. These polymorphisms had been identified by sequencing of genomic DNA extracted from the same liver samples and included 6 linked variants located in exon 7 and the 3′-UTR (CYP2B11-H2), and one SNP in the 3′UTR (CYP2B11-H3).
Although there was no clear evidence for allelic imbalance with the CYP2B11-H3 SNP, we did observe almost complete loss of expression of the variant allele at two, but not all 6 of the CYP2B11-H2 variant sites. This finding was identical in all 6 liver samples with the H1/H2 diplotype (regardless of breed). Since these two variants www.nature.com/scientificreports www.nature.com/scientificreports/ (c.2137 TG/CA and c.2166G/A) were located within the middle 3′-UTR region, this finding is consistent with a shortened 3′-UTR in the CYP2B11-H2 mRNA at some position between c.1913 and c.2137, rather than at the expected 3′-UTR end at c.2625. This was confirmed by RT-PCR analysis of the transfected CYP2B11-H1 and CYP2B11-H2 3′-UTR luciferase reporters (Fig. 7b). Furthermore, in the initial study that cloned CYP2B11 cDNA, two CYP2B11 mRNA species (one long and one short) by Northern blot analysis of dog liver RNA were also reported 18 . The approximate sizes of those mRNA species (2.9 kb and 1.9 kb) are similar to the predicted sizes of the CYP2B11-H1 (2.6 kb) and CYP2B11-H2 (1.9 -2.1 kb) mRNA. Finally, sequence analysis of the 3′-UTR identified two canonical consensus polyadenylation signal sites, one located upstream of the CYP2B11-H1 3′-end and an alternate canonical polyadenylation signal site upstream of the predicted CYP2B11-H2 3′-end (Fig. 7C).
Surprisingly, we did not observe CYP2B11-H2 allelic imbalance in the RNA-seq data at the two polymorphic sites further downstream (c.2283A/G and c.2636G/C). This would have been expected if the shorter CYP2B11-H2 transcript was generated through early termination of transcription with polyadenylation close to the internal alternate polyadenylation signal site. Recently, a novel widely used mechanism has been identified that generates shorter transcripts from longer transcripts through post-transcriptional 3′-UTR cleavage 19 . This process results in two separate RNA fragments; the mRNA coding region with a shorter 3′-UTR tail and a stable uncapped autonomous RNA fragment. Our RNA-seq data provides preliminary evidence that such a mechanism may be involved in generating the shorter (final) CYP2B11-H2 transcript from a longer (precursor) CYP2B11-H2, as well as a separate stable RNA fragment containing the variant allele. Importantly, our data suggests that one or more of the CYP2B11-H2 3′-UTR polymorphisms may serve to enhance utilization of this process through mechanisms that do not involve altering the polyadenylation signal sequence.
Since the CYP2B11-H3 only consisted of a single 3′-UTR SNP located at c.1952, RNA-seq data was uninformative regarding the length of the CYP2B11-H3 3′UTR downstream of this position. However, RT-PCR of CYP2B11 3′-UTR luciferase reporters indicated that the CYP2B11-H3 3′UTR was also truncated relative to CYP2B11-H1 with a length that was similar to CYP2B11-H2. More precise mapping of the 3′UTR of the CYP2B11 mRNA variants could be done in futures studies using techniques such as 3′-rapid amplification of cDNA ends (3′-RACE) or single molecule real-time (SMRT) sequencing.
The main purpose of constructing the CYP2B11 3′-UTR luciferase reporters was to evaluate the functional effects of the H2 and H3 haplotypes on gene expression. Both haplotypes significantly reduced gene expression as measured by luciferase activity, although H3 had the greatest effect, more than twice that of H2. Truncation of the 3′-UTR in the H2 and H3 variants would be expected to decrease mRNA stability. However, no differences were observed in mRNA expression between H1, H2 and H3 luciferase constructs using primers targeting Region 1 (Fig. 8B). Furthermore, CYP2B11 mRNA abundance was not lower in dog livers with either of the H2 or H3 haplotypes compared to those with only the H1 haplotype (Fig. 9C). Genotyped dog liver data did suggest  Table 3. CYP2B11 haplotypes in Greyhounds and other dog breeds. Greyhounds (n = 13) and one dog from each of 45 other breeds were genotyped for 9 polymorphisms in the CYP2B11 gene. Samples from Greyhound dogs were identified by their owners dogs registered with the National Greyhound Association bred for racing. Details regarding the polymorphisms are given in Table 2. Six haplotypes (H1 to H6) could be inferred from these genotypes. The allele sequences are shown for each haplotype. Alleles that differ from the CanFam 3.1 reference sequence are indicated by bolding and underlining for each haplotype. Also shown are the frequencies of each haplotype. The genetic polymorphism labels used here (#1 to #9) correspond to the labels used in Fig. 4 and  www.nature.com/scientificreports www.nature.com/scientificreports/ that these haplotypes might reduce translational efficiency as reflected by lower CYP2B11 protein/mRNA ratios (Fig. 9D). Consequently, it is possible that the CYP2B11-H2 and CYP2B11-H3 variants create novel binding sites for microRNAs on the mature mRNA (c.1913 insert C and c.1952 C > T, respectively), which are known to regulate gene expression by repressing translation.  Sighthound breeds, 45 other (non-Sighthound) breeds, and 153 mixed-breed dogs. Breeds were designated by the dog's owner. Greyhounds were divided into two breed sub-groups based on whether they were identified by their owners as dogs registered with the National Greyhound Association (NGA*) bred for racing or were dogs registered with the American Kennel Club (AKC**) bred for other purposes. Haplotype frequencies are shown for individual breeds grouped into "Sighthound dog breeds" and "Other dog breeds" for comparison. Shown next to the breed name are the number of individual dogs that were sampled. At least 10 dogs were sampled per breed.   www.nature.com/scientificreports www.nature.com/scientificreports/ Although Greyhounds are the principle breed reported to experience anesthetic drug sensitivity, veterinarians, owners, and breeders suspect that some closely related breeds within the Sighthound group of dog breeds may also be sensitive [4][5][6] . Consequently, we determined and compared the prevalence of both the H2 and H3 haplotypes across diverse breeds. We hypothesized that any variant contributing to the slow metabolizer phenotype should be more prevalent in Greyhounds and possibly other related Sighthounds compared with non-Sighthound breeds. This hypothesis was not confirmed for CYP2B11-H2, which could reflect the milder effects of this haplotype on drug metabolism phenotype. However, we did find a significantly higher prevalence of CYP2B11-H3 among Sighthounds compared with non-Sighthounds, with AKC Greyhounds having the highest H3 frequency of all breeds sampled (nearly 60%).
A recent genomic study indicates that most Sighthound breeds belong to one of two monophyletic groups 20 . Sighthound Group 1 contains the following breeds; Greyhound, Whippet, Scottish Deerhound, Irish Wolfhound, Borzoi, and Italian Greyhound. Sighthound Group 2 contains breeds from most of the other Sighthounds sampled in our study, as well as some breeds not considered Sighthounds, such as Great Pyrenees, Komondor, and Anatolian Shepherd. The CYP2B11-H3 haplotype was found in all of the Sighthound Group 1 breeds except Irish Wolfhound, while only one of the 7 Sighthound Group 2 breeds sampled (Ibizan Hound) had this haplotype. This finding suggests that CYP2B11-H3 may have arisen in a common ancestor of the Sighthound Group 1 breeds. Given the sporadic presence of CYP2B11-H3 in largely unrelated breeds outside of Group 1, it is likely that CYP2B11-H3 was dispersed from the Sighthound Group 1 breeds to other breeds through admixture and haplotype sharing, as was recently shown for other alleles by Parker et al. 20  www.nature.com/scientificreports www.nature.com/scientificreports/ AKC Greyhounds differed considerably from NGA Greyhounds in that they had a higher CYP2B11-H3 prevalence and lacked CYP2B11-H2. This difference may reflect a founder effect that occurred when these two populations were initially isolated. It might also be a consequence of selective breeding for different purposes, in that NGA Greyhounds are primarily bred for racing speed, while AKC Greyhounds are primarily bred for conformation.
Our results predict that some, but not all Greyhounds would have decreased CYP2B11 expression. Lowest CYP2B11 expression would be expected in dogs with the CYP2B11 H3/H3 diplotype, about 70% lower than for dogs with the H1/H1 diplotype. Although prior reports have shown lower clearance of propofol and thiobarbiturates in Greyhounds compared with mixed-breed dogs, all results were presented as aggregated data (i.e. mean ± SD) from 10 to 12 dogs per breed group 7,8 . Consequently, it is unclear whether there were differences between individual Greyhounds of a magnitude that would be consistent with the difference predicted by our in vitro data. It should also be pointed out that non-genetic factors such as enzyme induction and inhibition could contribute to variable CYP2B11 metabolism on top of genetic regulation. This is exemplified by enhancement of thiopental clearance by phenobarbital and inhibition of propofol clearance by chloramphenicol, respectively, in Greyhounds 9,10 . Figure 9. CYP2B11 diplotype-phenotype association analysis. Differences in (a) CYP2B11 activity, (b) protein, (c) mRNA, and (d) protein/mRNA ratio between the 15 previously studied dog livers are shown as box and whisker plots after grouping by CYP2B11 diplotype. Diplotypes included H1/H1 (2 Beagles and 3 mixed-breed), H1/H2 (3 Beagles, 2 mixed-breed, and 3 Greyhounds), H1/H3 (one Greyhound) and H3/H3 (one Greyhound). No dogs had the H2/H2 diplotype. Since only one dog liver had the H1/H3 diplotype and one liver had the H3/H3 diplotypes, these data were grouped with the H1/H2 livers (10 livers total) for statistical comparison by Mann-Whitney U test (P-values shown, N.S = not statistically significant) with the H1/H1 livers (5 livers). Samples from Greyhound dogs were identified by their owners as dogs registered with the National Greyhound Association bred for racing. (2020) 10:69 | https://doi.org/10.1038/s41598-019-56660-z www.nature.com/scientificreports www.nature.com/scientificreports/ In addition to detecting CYP2B11-H3 in 9 Sighthound breeds (other than Greyhounds), we also found this haplotype in 10 non-Sighthound breeds suggesting that the Sighthound CYP2B11 poor metabolizer phenotype might be found in non-Sighthound breeds. For most of these non-Sighthound breeds, the H3 haplotype frequency was relatively low (less than 10%). Therefore, the predicted frequency of the poor metabolizer CYP2B11 H3/H3 diplotype would be less than 1% (assuming we had a sufficiently representative sample of these breeds). However, we note that three of the breeds, including Labrador Retriever, Golden Retriever, and English Bulldog were ranked by the AKC in 2018 as the first, third, and fifth most popular dog breeds owned in the USA, respectively, based on annual AKC registration 21 . Consequently, the overall impact of this gene variant on these breeds could be substantial, at least in terms of the absolute numbers of dogs affected.
There were some limitations to the current study. The numbers of available dog livers from different breeds and with different genotypes were somewhat limited and so the results utilizing those samples should be viewed with caution. Also, the numbers of available DNA samples for some dog breeds was limited by availability, with a minimal sample size of 10 dogs arbitrarily set by us, so extrapolation to the entire breed should be done with caution. Furthermore, genotyping of the over 2,000 dogs was carried out using single haplotype marker polymorphisms and so for H2, which consists of multiple SNPs, it remains possible that the selected marker is not unique to the variant haplotype within the larger population. Finally, our predictions concerning the impact of the H2 and H3 variants on CYP2B11 expression are entirely based on in vitro studies with extrapolation in vivo. Consequently, future studies are needed to confirm these findings such as through evaluation of CYP2B11 function in vivo using isoform specific drug phenotyping probes comparing dogs from different breeds and with different CYP2B11 genotypes. Studies are ongoing in our laboratory to further characterize the impact of CYP2B11 haplotypes in vivo.

Materials and Methods
Animal ethics statement. The collection and use of liver tissue employed in this study were considered exempt from review by the Institutional Animal Care and Use Committee at Washington State University since all tissues collected would have been normally discarded. The collection, storage, and use of the DNA samples employed in this study were approved by the Institutional Animal Care and Use Committee at Washington State University (protocols #04194 and #04539) and were collected in accordance with relevant guidelines and regulations. Informed owner consent was obtained for all dogs prior to DNA collection. Dog liver tissues and microsomes. Snap frozen liver tissue samples were obtained and stored at −80 °C from 15 untreated healthy adult dogs, including 5 Greyhounds (3 males and 2 females; all registered NGA dogs bred for racing), 5 male mixed-breed dogs, and 5 male Beagle dogs. Dogs were untreated (control) research animals that had been euthanized for reasons unrelated to this study. Liver microsomes were prepared from the liver tissue samples detailed above as previously described 22 and stored at −80 °C until use. Microsomal protein concentrations for liver microsomes were determined using the bicinchoninic acid assay (Thermo Fisher Scientific).
Dog breed DnA sampling. Stored DNA samples from client-owned dogs were retrieved from the Washington State University Veterinary Teaching Hospital Patient DNA Bank (n = 1,182) and the Comparative Pharmacogenomics Laboratory Sighthound DNA Bank (n = 875). DNA had been extracted from buccal swab samples obtained by the hospital staff or by the dog's owner. The majority of the hospital patient samples derived from dogs living in the Pacific Northwest of the United States, while the Sighthound samples were obtained primarily by mail from dogs living throughout the United States. A dog's breed was identified by the owner for the Hospital Bank whereas breed was identified by the owner along with accompanying breed registration identification for the Sighthound Bank. For the purposes of this study, the designations "mix", "mixed", "cross", "mutt", "mongrel" or similar by the owner was considered as a single group of "mixed-breed" dogs. The 2,057 DNA samples represented 64 different dog breeds including 19 Sighthound breeds, 45 non-Sighthound breeds, as well as 153 mixed-breed dogs. The designation of a breed as belonging to the 'Sighthound' group was based on the AKC's breed inclusion for Sighthounds 23 . Samples from Greyhound dogs were divided into two groups based on whether they were identified by their owners as dogs bred for racing and registered with the NGA (n = 180) or were dogs bred for other purposes and registered with the AKC (n = 61). All breed groups included samples from at least 10 different dogs. (2020) 10:69 | https://doi.org/10.1038/s41598-019-56660-z www.nature.com/scientificreports www.nature.com/scientificreports/ Recombinant canine CYPs. Recombinant canine CYP1A1, CYP1A2, CYP2B11, CYP2C21, CYP2C41, CYP2D15, CYP3A12 and CYP3A26, all co-expressed with canine P450 oxidoreductase (POR) as bactosomes, were purchased from Sekisui Xenotech LLC (Kansas City, KS, USA). Since recombinant canine CYP2A13, CYP2A25 and CYP2E1 were not commercially available, these enzymes were made in-house as follows.
cDNA sequences for canine CYP2A13, CYP2A25, CYP2E1 and POR (NCBI entries NM_001037345.1, NM_001048027.1, NM_001003339.1, and NM_001177805.1, respectively) were synthesized and cloned into the pFastBac1 ™ vector (Thermo Fisher Scientific) by GenScript (Piscataway, NJ, USA). CYP and POR recombinant baculoviruses were created using the Bac-to-Bac ® baculovirus expression system (Thermo Fisher Scientific) following the manufacturer's protocols. Briefly, recombinant baculoviruses were created by transforming DH10Bac competent Escherichia coli with the recombinant pFastBac1 ™ plasmids using heat shock. Recombinant bacmid DNA was isolated using a QIAprep ® Spin Miniprep Kit (Qiagen, Hilden, Germany) and transfected into Sf9 (Spodoptera frugiperda) insect cells through Cellfectin ® II reagent-mediated gene transfer to produce recombinant baculoviruses. Recombinant baculoviruses were clarified and amplified to create high-titer passage stocks. Gel electrophoresis and DNA sequencing confirmed the presence of the cDNA in recombinant baculovirus.
Amplified viral stocks were titered relative to the recombinant POR baculovirus stock using a TaqMan ® gene expression assay (Thermo Fisher Scientific) as described by Hitchman et al. 24 .
Sf9 shaking suspension cultures were grown in the dark at 27 °C in Sf-900 ™ II serum-free medium (Thermo Fisher Scientific) supplemented with 5% fetal bovine serum (HyClone Laboratories, Logan, UT, USA) to a cell density of 1.5 × 10 6 cells/mL. Cells were then co-infected with recombinant viruses encoding CYP and POR at optimal CYP:POR viral ratios determined in preliminary experiments. At 24 h post-infection, hemin (prepared by dissolving in 50% ethanol and 0.2 M NaOH) was added to the culture to achieve a final concentration of 2 µg/ mL 25 . Cells were harvested at 72 h post-infection by centrifugation and washed twice with 4 °C phosphate buffered saline (pH 7.4). Cells were stored at −80 °C until use.
Microsomes were prepared by homogenization using a pestle tissue grinder follow by 2-speed centrifugation (9,000 and 100,000 × g at 4 °C) and then reconstituted in 100 mM phosphate buffer (pH 7.4), 20% glycerol and 1 mM EDTA 26 . Functional CYP content of recombinant microsomes was measured by CO-difference spectrum using a microplate assay as described by Yang et al. 27 . For the CO-difference spectra, an extinction coefficient (Δ ε450-490 ) of 106,000 M −1 cm −1 28,29 was used. POR activity of the recombinant microsomes was assessed by the cytochrome c reduction assay as described by Guengerich et al. 29 but scaled to fit a microplate format. Functionality of the recombinant microsomes was assessed through 7-ethoxycoumarin metabolism to umbelliferone (7-hydroxycoumarin) as detailed by Waxman and Chang 30 . Microsomes were stored at −80 °C until use.
In vitro incubation assay conditions including substrate concentration, microsomal protein concentrations, incubation time, and analytical method details are given in Supplementary Table S2. Most metabolite concentrations were determined by HPLC with absorbance or fluorescence detection (700-series Satellite Wisp auto-injector, 500-series pump, 486 absorbance detector, 470 fluorescence detector; Waters, Milford, MA, USA). 6-Hydroxybupropion and omeprazole sulfone concentrations were determined using a liquid chromatography-triple-quadrupole (LC-MS/MS) system (Agilent 1100 liquid chromatography system; Agilent Technologies, Inc., Santa Clara, CA, USA connected to an API 4000 mass spectrometer, AB Sciex, Framingham, MA, USA). Preliminary studies were conducted for each biotransformation using pooled dog liver microsomes to ensure linear metabolite formation with respect to increasing time and microsomal protein concentration. The rate of metabolite formation was calculated by dividing the metabolite concentration in the sample by the incubation time and microsomal protein concentration. Experiments were conducted in duplicate and results for individual liver microsomes were averaged. Propofol 4-hydroxylation activity values reported previously for the same set of dog liver microsomes 13 were also used to compare to these newly generated data.
Propofol 4-hydroxylation, bupropion 6-hydroxylation, and omeprazole sulfonation activities were also measured using a panel of 11 recombinant canine CYP enzymes that included CYPs 1A1, 1A2, 2A13, 2A25, 2B11, 2C21, 2C41, 2D15, 2E1, 3A12, and 3A26. Propofol 4-hydroxylation activities were quantified as previously described 12,13 with slight modifications as follows. Recombinant CYP concentration in the incubation was 10 pmol/mL, propofol concentration was 5 μM, while the incubation time was 10 min. The HPLC column used was a The relative contributions of individual CYP isoforms to total liver microsome propofol hydroxylation, bupropion hydroxylation, and omeprazole sulfonation activities were estimated by adjustment of specific CYP activities using the average liver microsome abundance of each CYP. Abundance values determined by mass spectrometry in liver microsomes from 59 dogs of differing breeds were 2.8, 82, 11, 7.7, 79, 52, 1.8, 143, 72, 125, and 3 .8 pmoles  CYP per mg microsomal protein for CYPs 1A1, 1A2, 2A13, 2A25, 2B11, 2C21, 2C41, 2D15, 2E1, 3A12, and 3A26,  respectively 15 . CYP1A, CYP2B11, and CYP3A protein content by immunoblotting. Microsomal CYP1A, CYP2B11 and CYP3A protein content were determined by semi-quantitative immunoblotting using the same Greyhound, Beagle and mixed-breed dog liver microsomes (n = 5 per breed) described above. The technique was based on a method described previously with minor modifications 35 . Rabbit polyclonal antisera raised against rat CYP1A2 (AB1255) and rat CYP3A1 (AB1253) were purchased from Chemicon Millipore (Temecula, CA, USA). Rabbit polyclonal antisera raised against dog CYP2B11 was a generousgift from Dr. James Halpert (School of Pharmacy, University of Connecticut, Storrs, CT, USA) 36 . Briefly, 10 µg of microsomal protein was separated by sodium dodecyl sulfate acrylamide gel electrophoresis using a 26-well 5 to 15% gradient gel (Criterion, Biorad, Hercules, CA, USA). Proteins were then electrophoretically transferred using a semi-dry technique to polyvinyl difluoride membrane (Immobilon-P; Millipore Corporation). Membranes were blocked in 5% powdered nonfat milk in Tris-buffered saline-Tween (0.15 M NaCl, 0.04 M Tris, pH 7.7, and 0.1% Tween 20) for one hour at room temperature and then incubated overnight at 4 °C in Tris-buffered saline-Tween/5% milk containing the primary antibody at an appropriate dilution (1:500 for CYP1A2; 1:6,000 for CYP2B11; 1:1,000 for CYP3A). Blots were washed, reblocked, and then incubated at room temperature for one hour with a 1:10,000 dilution of a goat anti-rabbit IgG antibody conjugated to horse radish peroxidase (PerkinElmer, Inc., Waltham, MA, USA). After washing, chemiluminescence reagent (Super Signal; Pierce Chemical Cp., Dallas, TX, USA) was applied, and blots were imaged using the Kodak Image Station 440CF (Kodak, Rochester, NY, USA). Bands were quantified using Kodak 1D Image Analysis Software (Kodak) and net intensity values for each liver sample were expressed relative to the liver sample containing the lowest band intensity. Final results for each liver sample represent the average of 3 independent experiments. Preliminary studies were conducted using serial dilutions of pooled dog liver microsomes to ensure a linear relationship between the amount of microsomal protein loaded and band intensity up to 20 µg of loaded protein. To ensure equal protein loading, membranes were washed and total protein was visualized with Ponceau S reagent 37 .
Liver CYP2B11 mRNA quantitation. Total RNA was isolated using TRIZOL Reagent (ThermoFisher Scientific) from the same Greyhound, Beagle and mixed-breed dog livers (n = 5 per breed) used to isolate microsomes. CYP2B11 mRNA content relative to 18S rRNA content was determined by real-time PCR with Sybr Green-based detection (CFX96 Touch, Bio-Rad) as previously described 38 . Primers for CYP2B11 mRNA were Pri_459_forward: 5′-GGA TTC AGG AGG AGG CTC AGT GTC-3′ and Pri_460_reverse 5′-GAT GTT GGC GGT CAT GGA GTG G. Primers for 18S rRNA were Pri_127_forward: 5′-CCC CTC GCT GCT CTT AGC TGA GTG T-3′ and Pri_128_reverse 5′-CGC CGG TCC AAG AAT TTC ACC TCT. CYP2B11 sequencing and genotyping. Genetic polymorphisms located in the CYP2B11 5′-enhancer (to ~2,000 bp upstream), exons 1-9, and 3′-UTR were identified by Sanger sequencing of genomic PCR product using DNA obtained from 13 Greyhounds (5 from liver samples and 8 from buccal swab samples). Primers used for PCR and sequencing, as well as the gene region amplified and product size are given in Supplementary  Table S3 Table. Genotype data from the same CYP2B11 gene regions were also obtained from another 45 dogs (each of a different breed) by analysis of publicly available whole genome sequence data. Briefly, binary alignment files originally submitted by the Institute of Genetics, University of Bern, Switzerland were downloaded from the European Nucleotide Archive (Study ID PRJEB16012). Polymorphisms were identified and genotypes called on individual dog samples using the Freebayes bayesian genetic variant detector (arXiv:1207.3907) as implemented in Galaxy version 1.1.0 39 on a Bioteam Appliance (Bioteam, Middleton, MA, USA). The IDs of individual dogs that were sequenced and analyzed, as well as their nominal breed, are listed in Supplementary Table S1. Haploview 40 was used to evaluate the extent of linkage disequilibrium between identified polymorphic sites across the CYP2B11 gene and to resolve individual haplotypes.
Custom allele discrimination assays (Applied Biosystems TaqMan SNP Genotyping Assay, Thermo Fisher Scientific) were used to genotype DNA samples from 2,057 dogs for the CYP2B11 haplotype marker polymorphisms c.2137 TG/CA (CYP2B11-H2) and c.1952 C/T (CYP2B11-H3). Primer and reporter sequences are given in Supplementary Table S4. Assays were performed using a real-time PCR instrument (CFX96 Touch, Bio-Rad).
CYP2B11 RNA-seq. RNA-seq was conducted as described previously 41 . Total RNA was extracted from the same Greyhound and Beagle livers (n = 5 per breed) used for determining CYP activities and quantifying CYP2B11 mRNA. Briefly, cDNA libraries were prepared from total RNA from each liver using the Truseq Stranded Total RNA LT kit (Illumina, San Diego, CA, USA). Libraries were sequenced on an Illumina 2000 Instrument at the Columbia Genome Center (New York, NY, USA), generating 60 million 100-bp paired-end reads. After quality filtering, reads were mapped to the canine reference genome (CanFam3.1) using Tophat version 1.5.0 42 and transcripts assembled using Cufflinks version 0.0.7 42 , as implemented in Galaxy version 1.1.0 39 on a Bioteam Appliance (Bioteam). For allele expression analysis, mapped read depths of the variant and reference alleles in heterozygous H1/H2 and H1/H3 samples at the site of each polymorphism comprising the H2 and H3 haplotypes were obtained using GenomeBrowse version 2.1.2 (Golden Helix, Bozeman, MT, USA). Variant to reference ratios were obtained for each dog and averaged by breed and diplotype group. These raw average ratios were then corrected for possible