Beagle dogs are widely used in preclinical pharmacokinetic, safety, and formulation studies. However, little is known about intestinal and hepatic distribution of major enzymes and transporters involved in oral absorption and presystemic drug metabolism. We characterized mRNA levels of CYP3A12, CYP3A26, CYP2D15, UGT1A6, ABCB1 (MDR1), ABCC1 (MRP1), ABCG2 (BCRP), SLC15A1 (PEPT1), and SLC22A1 (OCT1) in dog liver and along the intestine by real-time quantitative reverse transcription-polymerase chain reaction. Tissue protein levels of CYP2D15, MDR1, and PEPT1 were obtained by Western blot. Gene distribution and expression variability was statistically described by a generalized additive mixed model smoothing function and correspondence analysis. Results were compared with the expression pattern known for the human orthologs. Hepatic mRNA levels for metabolic enzymes were generally higher than those for membrane transporters, whereas in the intestine the opposite was observed. Hepatic mRNA levels followed the order CYP2D15 > UGT1A6 ≈ CYP3A26 > ABCB1 ≈ SLC15A1 ≈ SLC22A1 > ABCG2 > ABCC1 ≈ CYP3A12. Along the gut, the genes were differentially distributed with greatest expression in duodenum/upper jejunum (ABCG2), middle jejunum (ABCB1 and SLC15A1), or in cecum/colon (ABCC1 and CYP2D15). CYP3A12, CYP3A26, SLC22A1, and UGT1A6 had a rather uniform expression. Intestinal mRNA profiles of CYP2D15, ABCB1, and SLC15A1 correlated with the respective protein levels. Canine CYP3A12/26, CYP2D15, and ABCB1 colonic distributions differed from those of human orthologs, whereas UGT1A6, ABCC1, ABCG2, SLC15A1, and SLC22A1 were comparable to those of humans in both small and large intestine. We aim to apply these data to better interpret pharmacokinetic studies in dogs with respect to their human relevance.
The assessment of absorption, distribution, metabolism, and excretion (ADME) and safety properties of candidate drugs in animal species and their translation to humans is an essential part of drug discovery and development. Nevertheless, species differences in xenobiotic metabolism between animals and humans complicate the use of animal data for prediction of human clearance and bioavailability, prodrug conversion, and formation of active or toxic metabolites (Krämer and Testa, 2008). The prediction of oral bioavailability in humans for those compounds undergoing significant intestinal presystemic metabolism is particularly challenging, because of the complex interplay between drug dissolution, permeability, intestinal transit time, and influence of metabolic enzymes and transporters.
The beagle dog is a commonly used model to assess oral bioavailability, evaluate new formulations, investigate the influence of food on the absorption of new drug compounds (Mealey et al., 2008), and as the standard species for preclinical safety evaluations (Keenan and Vidal, 2006). Although the gastric pH, motility, and emptying time in fasted dog and human stomach are reported to be comparable, a slower gastric emptying and a lower pH were observed postprandially in dogs compared with humans (Dressman, 1986). Furthermore, the species differences in expression level/distribution and substrate specificity of intestinal metabolic enzymes and active transporters make the translation of dog data to humans challenging (Bleasby et al., 2006; Martignoni et al., 2006; Komura and Iwaki, 2008).
Physiology-based modeling approaches have been proposed, which can account for known species differences and allow translation of dog data to humans (Parrott and Lave, 2008; Parrott et al., 2009; Dressman and Reppas, 2010). However, these methods rely on a thorough understanding of the physiologies and enzyme and transporter activities in both species, and this is currently not the case for the dog, making it a limiting factor.
To date, the distribution of metabolic enzymes and membrane transporters in beagle dogs has not been comprehensively investigated, particularly compared with the literature available for rats and humans. The reported studies in dogs mainly focus on the expression and catalytic activity of metabolic enzymes in the liver (Kyokawa et al., 2001; Mills et al., 2010). Less is known about the intestinal expression distribution of cytochromes P450 (P450s), UDP-glucuronosyltransferases (UGTs) (Bock et al., 2002; Mealey et al., 2008), and transporters (Conrad et al., 2001) or about P450 substrate specificity (Fraser et al., 1997; Turpeinen et al., 2007; Locuson et al., 2009) and intersubject variability in expression.
The purpose of this work was to characterize the mRNA distribution of four selected drug metabolism- and pharmacokinetics-relevant metabolic enzymes and five drug transport proteins along the intestine and in the liver of beagle dogs. mRNA levels in eight intestinal segments and in the liver were assessed by real-time quantitative reverse transcription-polymerase chain reaction (RTq-PCR). Individual target gene selection was based on their impact on drug absorption and metabolism in humans. In addition, transporters currently poorly characterized in the dog were selected to provide new insight on their intestinal expression and supportive information for physiology-based pharmacokinetic modeling. The following genes were investigated: ABCB1 [efflux multidrug resistance P-glycoprotein (MDR1, P-gp)], ABCC1 [efflux multidrug resistance-associated protein 1 (MRP1)], ABCG2 [efflux breast cancer resistance protein (BCRP)], SLC15A1 [influx peptide transporter-1 (PEPT1)], and SLC22A1 [influx organic cation transporter-1 (OCT1)] as transporters (Murakami and Takano, 2008), as well as CYP3A12, CYP3A26, CYP2D15 (P450), and UGT1A6 (UGT) as phase 1 and phase 2 metabolic enzymes, respectively (Testa and Krämer, 2007, 2008a). Several candidate housekeeping genes were evaluated for the normalization of the expression levels of the aforementioned target genes. These were glyceraldehyde-3-phosphate dehydrogenase (GAPDH), villin 2 (ezrin), hypoxanthine phosphoribosyl-transferase 1 (HPRT1), and ribosomal protein L32 (RPL32, component of the 60S ribosomal subunit).
To understand how the mRNA expression distribution along the intestine correlates with the respective protein expression, MDR1, PEPT1, and CYP2D15 were analyzed by Western blot, and the results were normalized to β-actin or GAPDH as internal controls.
Data analysis was performed with regard to the change in mean and individual mRNA levels along the intestine. mRNA distribution data and respective intersubject variability were analyzed and described by a generalized additive mixed model (GAMM), which fits a smoothing curve through the intestinal data of each dog and by correspondence analysis (CA) (Greenacre, 1992). Our data may contribute to the refinement of physiology-based pharmacokinetic models and to a better understanding of drug metabolism and pharmacokinetics in dogs, improving the translation of preclinical data from dogs to humans.
Materials and Methods
Collection of Liver and Intestinal Samples.
Dog necropsy was performed at F. Hoffmann-La Roche Ltd. (Basel, Switzerland). Four male and four female overnight-fasted beagle dogs (details in Table 1) were euthanized by intravenous injection of 150 mg/kg pentobarbital and were exsanguinated according to the institution guidelines in compliance with national and regional legislation. Liver and intestinal samples were immediately collected and rapidly rinsed three times with 0.9% aqueous NaCl at room temperature. A 2-cm-long sample from the duodenum, ileum, cecum, and colon was collected in the middle of the corresponding segment length, whereas the jejunum was sampled at ∼20% (jejunum segment J1), ∼40% (jejunum segment J2), ∼60% (jejunum segment J3), and ∼80% (jejunum segment J4) of the total jejunum length. Tissues for RNA extraction were placed in scintillation vials prefilled with 10 ml RNAlater solution (RNase inhibitor; Ambion, Austin, TX), whereas samples for Western blot analysis were shock-frozen in liquid nitrogen. All the collected tissues were stored at −80°C.
Approximately 30 mg of intestinal tissue were cut on ice, placed in a FastPrep-24 Lysing Matrix D tube containing ceramic beads (MP Biomedicals, Solon, OH) and 600 μl of “lysis” buffer (RLT buffer 1% β-mercaptoethanol from the RNeasy Mini kit, QIAGEN, Valencia, CA), and carefully homogenized with a Precellys-24 instrument (Bertin Technologies, Montigny le Bretonneux, France). After a protein digestion step with proteinase K (QIAGEN), total mRNA was extracted with the QIAGEN RNeasy Mini kit according to the manufacturer's instructions. For the extraction from the liver samples, 30 mg of tissue were homogenated in 1.2 ml of lysis buffer, and total mRNA was obtained as described above in this section but without protein digestion.
RNA concentration for both liver and intestinal extracts was measured with a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. RNA integrity was assessed for all the tissue samples with an Experion microfluid capillary automated electrophoresis system and the Experion StdSens Analysis Kit (both from Bio-Rad Laboratories, Hercules, CA) according to the manufacturer's instructions. The RNA quality indicator (RQI) and the ratio of ribosomal RNA (28S/18S) were used to determine the integrity of the isolated RNA from each sample. The dog RNA extracted from 72 liver and intestinal samples showed high integrity, as indicated by an RQI value between 7.6 and 10 (RQI = 1 indicates degraded RNA, and RQI = 10 indicates intact RNA) and by a 28S/18S ratio > 1.25 (28S/18S between 1 and 2 denotes intact RNA).
RTq-PCR was performed for ABCB1 (MDR1), ABCC1 (MRP1), ABCG2 (BCRP), SLC15A1 (PEPT1), SLC22A1 (OCT1), CYP3A12, CYP3A26, CYP2D15, and UGT1A6, as well as for the housekeeping genes GAPDH, villin 2, HPRT1, and RPL32 with the corresponding gene expression assays available from Applied Biosystems (Foster City, CA). For GAPDH, primers and probes (dual-labeled TaqMan probes; Invitrogen, Carlsbad, CA) were designed as reported by Peters et al. (2007). Reverse transcription and RTq-PCR were performed with a LightCycler 480 instrument (Roche Applied Science, Mannheim, Germany) and a QuantiTect Probe RT-PCR Kit (QIAGEN). For transcribing mRNA into cDNA, 5 μl of RNA (200 ng/μl) were added in a 96-well plate (1 μg of total RNA per well) and were mixed with 20 μl of Master Mix [12.5 μl of QIAGEN QuantiTect Probe RT-PCR Master Mix 2×, 0.25 μl of QIAGEN QuantiTect RT Mix, 1.25 μl of primer probe (20×), and 6 μl of RNase-free water]. Reverse transcription was performed at 50°C for 30 min. The cycling conditions for PCR were the following: polymerase activation (15 min at 95°C) and 45 cycles alternating denaturation (15 s at 94°C) and annealing/extension (1 min at 60°C). Relative gene expression levels were calculated as 2−ΔCt, where Ct is the threshold amplification cycle number, ΔCt is the difference between the mean Ct value of the gene under study and the housekeeping gene. The limit of detection of mRNA was defined at a Ct value of 30.
Western Blot Analysis of Protein Expression.
Between 170 and 520 mg of frozen tissue were pulverized with a pestle in a precooled mortar containing liquid nitrogen and were stored at −80°C. The tissue powder was suspended in ice-cold buffer (200 mg/ml) consisting of 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, 50 mM HEPES (Invitrogen), and Complete EDTA-Free Protease Inhibitor (one tablet/50 ml; Roche Diagnostics, Mannheim, Germany), adjusted to pH 7.5, and supplemented with 2% NP40 (Calbiochem, San Diego, CA). The suspension was centrifuged (4°C, 16,000g for 15 min), and the protein concentration was measured with the Pierce BCA Protein Assay (Thermo Fisher Scientific). For the polyacrylamide gel electrophoresis, 30 μg of protein were mixed with 4 μl of NuPAGE LDS sample buffer (4×, pH 8.4; Invitrogen) and 1.6 μl of NuPAGE Sample Reducing Agent (10× with 500 mM dithiothreitol; Invitrogen) and were diluted with water to 16 μl. The mixture was heated for 5 min at 85°C and then centrifuged for 5 min at 10,000g. Approximately 30 μg of protein were loaded in each slot of a 4 to 12% bis-Tris polyacrylamide gel (Citerion XT; Bio-Rad Laboratories). Polyacrylamide gel electrophoresis was performed for CYP2D15, MDR1, PEPT1, GAPDH, and β-actin according to a standard protocol. Proteins were blotted on a nitrocellulose membrane (0.45 μm, 8.5 × 13.5 cm Criterion Blotting Sandwich; Bio-Rad Laboratories) and were detected with the following antibodies diluted in Tris-buffered saline (0.1% Tween, 5% bovine serum albumin): goat polyclonal anti-human CYP2D6 antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) 1:500, mouse monoclonal anti-human MDR1 antibody C219 (Merck, Darmstadt, Germany) 1:500, goat polyclonal anti-human PEPT1 antibody (Santa Cruz Biotechnology, Inc.) 1:200, mouse monoclonal anti-rabbit GAPDH antibody (AbD Serotec, Oxford, UK) 1:2000, mouse anti-β-actin antibody conjugated to horseradish peroxidase (Abcam, Cambridge, MA) 1:5000, and peroxidase-conjugated AffiniPure goat anti-mouse and anti-goat secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) both 1:5000. Proteins were visualized by chemiluminescence with Lumi-Light Western blotting substrate (Roche Diagnostics) or SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific). Detection was performed with an LAS-4000 camera (Fujifilm, Tokyo, Japan), and band intensity was quantified with the ImageQuant TL software (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK).
Intestinal gene expression data were analyzed with GAMM. A GAMM fits a smoothing curve through the data for each dog separately. Changes in gene expression across segments were nonlinear, and there were no theoretical models that could be used to describe the functional form of the relationship. Treating intestinal segment as a categorical variable resulted in highly parameterized models, leading to difficulties with convergence when including the segment by dog interaction effect. GAMMs were, therefore, used because they provide the flexibility to model nonlinear relationships and use relatively few parameters, even if the p values are somewhat approximate (Zuur et al., 2009). Gene expression values were assumed Gaussian and were modeled using a penalized cubic regression spline, which allows the degree of smoothness of the regression curve to be estimated using cross-validation. The main independent variable was intestinal segment, which was treated as a continuous numeric variable (taking values 1 to 8). The effects of interest were 1) change in mean expression across segments, 2) whether the variability of expression was constant across segments, and 3) whether individual dogs had a different pattern of expression across segments (dog × segment interaction effect) or whether they all had a similar pattern. The influence of additional covariates (age, sex, and country of origin) was also examined.
The full GAMM was as follows: geneij = α + f(segmentij) + β1 × dogi + β2 × f(segmentij) × dogi + εij, where geneij is the expression of a particular gene for dog i in segment j, α is the overall mean, f(segmentij) is the smoothing function, β1 is the random effect of dog, and β2 is the random segment by dog interaction. The variance of the residuals was modeled as a power of a covariate: εij ∼ N(0,σ2 × fij 2δ), where fij are the fitted values.
CA was used to visualize the multivariate relationship between genes and the intestinal segment that a sample was derived from. CA is similar to principal component analysis, except that whereas principal component analysis gives the percentage of variation that each component (axis) accounts for, CA uses the related concept of “inertia” (the interpretation of percentage of variation and percentage of inertia is similar, and figures and text refer to variability, which is a more familiar concept). Analysis was performed in R (version 2.11.1) using the mgcv package (version 1.6–2; http://cran.r-project.org/web/packages/mgcv/), made4 (version 1.22.0; http://www.bioconductor.org/packages/2.9/bioc/html/made4.html), and ade4 (version 1.4–14; http://cran.r-project.org/web/packages/ade4/) packages. P values less than 0.05 were considered statistically significant.
Housekeeping Genes for RTq-PCR Data Normalization.
GAPDH is frequently chosen as the housekeeping gene for RTq-PCR data normalization. However, it is reported that the respective mRNA expression in dogs is not fully comparable across all tissues (Barber et al., 2005). In this work, GAPDH levels in dog liver and intestine were similar. Overall, the variability in GAPDH mRNA expression was varied within a factor of 2 across the different tissues (liver: mean Ct = 20.4 ± 0.4 from n = 16 measurements; intestine: mean Ct = 19.6 ± 0.4 from n = 128 measurements, samples from eight dogs). Hepatic mRNA expression of the investigated metabolic enzymes and transporters were, therefore, normalized to GAPDH. HPRT1 and RPL32 also showed stable mRNA expression along the gut. However, they were not considered suitable for data normalization, because of their very low expression levels compared with most target genes. Villin 2 is an enterocyte-specific protein, which showed ∼16-fold lower mRNA levels in the liver (mean Ct = 23.1 ± 0.4 from n = 16 measurements, samples from eight dogs) than in the intestine (mean Ct = 19.2 ± 0.6 from n = 128 measurements, samples from eight dogs). Villin 2 was considered an appropriate gene for normalizing the intestinal data because it showed comparable expression levels along the gut, and it is colocalized in the enterocytes with the genes under study. Expression levels of the tested housekeeping genes are reported in Table 2. For the Western blot analysis including liver and intestinal samples, GAPDH and β-actin were chosen for normalization, because of their relatively stable and high expression in both target organs.
mRNA Expression of Metabolic Enzymes and Membrane Transporters in Liver.
The GAPDH-normalized mRNA levels of the studied metabolic enzymes and transporters are reported in Fig. 1 and in Table 3. CYP2D15 showed the highest mRNA expression (10-fold over GAPDH), and CYP3A12 showed the lowest (0.01-fold over GAPDH). UGT1A6 and CYP3A26 mRNA levels were similar to each other (both 3.7-fold over GAPDH). Of the three investigated P450s, CYP2D15 represented 73.2%, CYP3A26 was 26.7%, and CYP3A12 was only 0.1% of the detected mRNA. For the membrane transporters, ABCB1 (MDR1), SLC15A1 (PEPT1), and SLC22A1 (OCT1) showed similar mRNA levels (1.07, 0.88, and 0.88-fold over GAPDH expression, respectively), whereas ABCG2 (BCRP) and ABCC1 (MRP1) were expressed at lower levels (0.11- and 0.02-fold over GAPDH expression, respectively).
CV in hepatic gene expression between the animals ranged from 15 (CYP2D15) to 100% (CYP3A12) (Table 3).
Metabolic Enzymes and Membrane Transporters along the Intestine.
Villin 2-normalized mRNA levels of the studied metabolic enzymes and membrane transporters in the intestine are reported in Table 4. The results of the fitted GAMM describing the gene expression trends along the intestine are shown in Fig. 2, and the respective statistical analysis is reported in Fig. 3 and Table 5. UGT1A6 mRNA level in the gut was 2 to 3 orders of magnitude higher than that of P450 enzymes. CYP3A12 and CYP3A26 expressions along the intestine were low, and CYP3A26 had unequal variances across the regions as well as a relatively high intersubject variability (p = 0.029), which might be related to the technical variability of the RTq-PCR assay associated with the low mRNA levels. In contrast to the situation in the liver, intestinal CYP3A12 mRNA expression was similar to the CYP3A26 level. CYP3A26 and UGT1A6 showed constant mRNA expression along the intestine. CYP2D15 mRNA level was highly variable in the duodenum, low in the jejunum and ileum, and increased in the cecum and colon (∼3-fold higher in the cecum than in the jejunum; p < 0.0001). The higher CYP2D15 mRNA expression in the large intestine was in agreement with the CYP2D15 protein level obtained by Western blot analysis (Fig. 4). Of the studied membrane transporters, PEPT1 showed the highest mRNA expression, followed by MDR1 ≫ BCRP and MRP1, whereas OCT1 had the lowest mRNA expression. The mRNA levels of ABCB1 (MDR1) increased along the small intestine, reaching the highest value in the ileum and decreasing in the cecum and colon. The observed expression trend was in line with the MDR1 protein levels along the gut obtained by Western blot analysis (Fig. 4) and was observed for all eight dogs. mRNA expression for SLC15A1 (PEPT1) was also high and constant along the small intestine (1.8–1.5-fold over villin 2 from the jejunum to the ileum, respectively) but was ∼15-fold lower in the large intestine. Again, the higher mRNA expression in the small versus large intestine was in agreement with the corresponding PEPT1 protein levels (Fig. 4). ABCG2 (BCRP) showed a higher expression in the small intestine ranging from 0.1- to 0.07-fold over villin 2 from the jejunum to the ileum, respectively, than in the large intestine (∼0.02-fold over villin 2). Intestinal mRNA levels for ABCC1 (MRP1) ranged from 0.02- to 0.06-fold over villin 2 levels, whereas for SLC22A1 (OCT1), no statistically significant changes in mRNA expression were observed across the intestinal segments. The SLC22A1 expression data include a few potential outliers (Fig. 2), possibly because the expression was close to the detection limit of the RTq-PCR. With these points removed, SLC22A1 showed a slight decrease in mRNA expression from the duodenum to the colon (p < 0.0001). CVs in intestinal gene expression between the animals ranged from 16% (ABCB1 in the segment “jejunum 3”) to 100% (CYP3A12 in the segments “duodenum” and “jejunum 2, jejunum 3”; SLC22A1 in “cecum” and “colon”; and ABCG2 in “cecum”) (Table 4). We found no statistically significant difference in gene expression related to age, gender, and country of origin of the animals.
Multivariate Analysis of Intestinal Gene Expression.
CA (Fig. 3) shows that on the basis of the level of mRNA expression, the cecum and colon cluster together. The majority of the total variability (75%) was observed between the segments of the large intestine (cecum and colon) and those of the small intestine (duodenum, jejunum, and ileum), where the large intestine clearly separated along the x-axis (first principal axis) from the small intestine. This indicates that the small and large intestines are different in terms of gene expression levels. Segments from the duodenum to ileum were instead separated along the y-axis (second principal axis) following their anatomical order, showing that the small intestine segments anatomically adjacent to each other have similar patterns of gene expression.
Human CYP2D6 is involved in the metabolism of ∼25% of low-molecular-weight drugs. However, it represents only 2 to 4% of total P450 content in human liver (Zhou, 2009), and its mRNA level is 220- and 20-fold lower than CYP3A4 and CYP3A5, respectively (Rodríguez-Antona et al., 2001). In contrast, we found higher mRNA levels for CYP2D15 than for CYP3A26 in dog liver, which supports previous reports showing that CYP2D15 protein represents ∼20% of total P450 liver content (Sakamoto et al., 1995), whereas CYP3A represents ∼14% (Eguchi et al., 1996). Despite differences in CYP2D expression between human and dog, similar oxygenation kinetics were observed with the respective liver microsomes (Bogaards et al., 2000; Trepanier, 2006). Comparing the oxidation of CYP2D6 substrates in various preclinical species including mouse, rat, rabbit, dog, micropig, and monkey, the kinetics of human CYP2D6 were most similar to those of the canine ortholog CYP2D15 (Bogaards et al., 2000), which suggests the dog model is a better predictor of human CYP2D-mediated metabolism than other preclinical species. In our study, the intersubject variability of hepatic CYP2D15 expression was low (CV = 15%, n = 8 dogs), whereas in humans, the variability in CYP2D6 levels is noticeably higher (CV = 67%, n = 12 subjects) (Rodríguez-Antona et al., 2001). Intestinal mRNA levels of canine CYP2D15 were higher than those of CYP3A12 and CYP3A26 and tended to increase from the small to the large intestine. This pattern was confirmed by Western blot, where no significant CYP2D15 protein was detected with the anti-human CYP2D6 antibody in the duodenum, jejunum, and ileum, but a protein band at the expected molecular weight was seen in the cecum and colon. This is in contrast to the expression in human intestine, where CYP2D6 is present in the duodenum and jejunum but absent in the colon (de Waziers et al., 1990; Ding and Kaminsky, 2003).
Human CYP3A4 and CYP3A5 subfamilies represent ∼30% of total liver P450s and contribute to the metabolism of 45 to 60% of the currently used drugs (Testa and Krämer, 2008b). In dog liver, the level for CYP3A12 mRNA was markedly lower than the level for CYP3A26, which is in agreement with previous results reported by Mealey et al. (2008). However, it is uncertain to what extent the higher CYP3A26 expression translates into a larger impact of CYP3A26 on drug metabolism, because CYP3A26 shows generally lower hydroxylase activity than CYP3A12 (Fraser et al., 1997). The CYP3A mRNA levels that we measured for dog liver were lower than those known for human liver and were in agreement with the lower testosterone 6β-hydroxylation activity observed in dog versus human liver microsomes (Bogaards et al., 2000). The intersubject variability in hepatic CYP3A26 expression was low (CV = 22%, n = 8 dogs), whereas CYP3A12 had greater variability. A significant intersubject variability in expression was observed in humans, with CYP3A4 and CYP3A5 showing 128- and 24-fold difference in mRNA levels between subjects, respectively (Rodríguez-Antona et al., 2001).
CYP3A4 is also the most abundant and metabolically active P450 in human small intestine (Thörn et al., 2005; Paine et al., 2006). The few available canine data indicate that CYP3A intrinsic metabolic clearance is in general lower in dog than in human intestine (Komura and Iwaki, 2008). On the basis of our findings and data reported in the literature, one may expect an underestimation of presystemic metabolism for mutual CYP3A4/5-CYP3A12/26 substrates from dog studies. In contrast to liver, canine CYP3A12 and CYP3A26 isoforms had similar intestinal mRNA levels and showed constant (CYP3A26) or nearly constant (CYP3A12) expression along the gut. This pattern is different in humans, where a marked decrease in CYP3A mRNA level is observed in the colon (Thörn et al., 2005). The intersubject variability in CYP3A12 and CYP3A26 intestinal expression was within a factor of 6 (n = 8 dogs), with the lowest variation observed in the upper jejunum segments J1 and J2 (∼2-fold). Significant variability in CYP3A4 and CYP3A5 mRNA expression was also observed in human intestine (Thörn et al., 2005), resulting predominantly from transcriptional regulation (Martínez-Jiménez et al., 2007). For instance, ligand-activated pregnane X receptor (PXR) increases the expression of human CYP3A4 as well as canine CYP3A12 and CYP3A26 (Chen et al., 2009). Although all three enzymes are inducible by drugs binding to PXR, significant differences were observed comparing the effects of various drugs on canine versus human CYP3A expression, which may be attributed to the ligand specificity of PXR (Chen et al., 2009). The comparison of expression variability for the target genes between dogs and humans generally indicated a lower variability for the dogs. This may be related to the inbred nature of the dogs used in the study and may not be generalizable to other dogs. We therefore advise to treat this conclusion with caution.
UGTs are the second most relevant drug-metabolizing enzymes in humans, because of the number of drug substrates (Uchaipichat et al., 2006) and because they are the most important in phase 2 metabolism. UGT protein levels in human intestinal microsomes may be as high as those in liver microsomes (Tukey and Strassburg, 2000). In our study, UGT1A6 mRNA was detected at high levels in all the investigated dog tissues. This is qualitatively similar to humans, where UGT1A6 expression was observed in the liver, small intestine, and colon (King et al., 1999; Strassburg et al., 2000; Court et al., 2011). Moreover, along the gut of dogs, UGT1A6 mRNA levels were higher than CYP3A and CYP2D15, which is in line with observations in humans and different from the expression pattern in rats (Testa and Krämer, 2010).
Several ATP-binding cassette efflux transporters are expressed in the apical brush-border domains of enterocytes, where they may contribute to the reduction in oral absorption of drug substrates and can potentially induce transporter-mediated drug-drug interactions in humans (Murakami and Takano, 2008). Thus far, there are only a few studies on the role of membrane transporters on the intestinal absorption in dogs. McEntee et al. (2003) reported that the oral exposure of the MDR1 substrate docetaxel was 15-fold higher in dogs treated with MDR1 and CYP3A inhibitor cyclosporine than when the drug was administered alone. Likewise, coadministration of cyclosporine with grapefruit juice, another inhibitor of both MDR1 and CYP34A, increased the oral absorption of cyclosporine in dogs (Amatori et al., 2004). In the dogs investigated here, a high level of ABCB1 (MDR1) gene expression and MDR1 protein was detected in both the liver and intestine. Expression levels of both MDR1 mRNA and protein increased along the small intestine, followed by a strong decrease in the cecum and colon. This contrasts the expression pattern in humans, where ABCB1 mRNA levels in the gut increase from the duodenum to the colon (Thörn et al., 2005; Bruyere et al., 2010). Even though the highest degree of drug absorption occurs in the small intestine, the discrepancy between human and dog MDR1 expression in the large intestine might be relevant when studying the absorption of P-gp substrates with low permeability and/or in delayed release formulations.
Unlike MDR1, MRP1 (ABCC1) is predominantly expressed in the basolateral membrane of epithelial cells, showing pharmacological and toxicological functions and contributing to the inherent resistance of cell lines to a range of cytotoxic anticancer drugs (Wijnholds, 2002). In this work, canine ABCC1 expression correlated with that in human, with lower levels in the liver than in the intestine (Bleasby et al., 2006); however, in contrast to the other studied genes, the mRNA levels increased toward the large intestine (2.4-fold higher in the colon than in the jejunum). A similar gene expression trend along the gut was also observed in rats (Lu and Klaassen, 2008).
Human BCRP is predominantly expressed in the intestine, liver, and blood-brain barrier and is involved in multidrug resistance. In addition, there is evidence from human and rodent in vivo studies that BCRP can affect the intestinal absorption of drug substrates (Murakami and Takano, 2008), whereas no information is available for dogs. Here, the canine BCRP (ABCG2) intestinal mRNA distribution resembles that reported for humans (Seithel et al., 2006; Bruyere et al., 2010). However, the mRNA abundance for ABCG2 along the dog intestine was overall lower than that for ABCB1 (MDR1), which is in contrast to the human jejunum, where the ABCG2 mRNA level is higher than ABCB1 (Taipalensuu et al., 2001).
The mammalian proton-coupled di- and tri-peptide transporter PEPT1 (SLC15A1) is involved in the absorption of dietary nitrogen and is also a major route of drug delivery for several important classes of compounds, including several prodrugs (Nielsen et al., 2001, 2002; Nielsen and Brodin, 2003). In our study, the presence of PEPT1 in dog liver was confirmed by both RTq-PCR and Western blot analysis. Along the intestine, SLC15A1 levels were the highest among the studied genes. RTq-PCR and protein expression results correlated well, confirming high PEPT1 protein expression in dog small intestine, which is similar to the human situation (Ziegler et al., 2002; Rubio-Aliaga and Daniel, 2008).
The tissue distribution of the polyspecific OCT1 (SL22A1) is known to be species-specific with high expression levels in the liver of humans but predominant expression in the kidney in rats (Gründemann et al., 1994; Jonker and Schinkel, 2004). In our study, canine SL22A1 mRNA was largely expressed in liver and at a noticeably lower level along the gut, resembling more the expression distribution in humans (Jonker and Schinkel, 2004) than in rats (Gründemann et al., 1994).
A direct comparison of expression levels between the liver and the intestine was not attempted because the housekeeping genes are either differentially expressed in the two tissues (villin) or not strictly colocalized with the gene of interest (GAPDH is expressed in all cells of the intestinal wall, whereas the target genes are predominantly expressed in the enterocytes).
The gene expression pattern of four drug-metabolizing enzymes and five drug transporters in the liver and along the intestine of beagle dogs reveals a number of differences compared with the published human data. In particular, the tissue distribution of P450 isozymes and MDR1 appears to be markedly different in dogs compared with humans, whereas UGT1A6, PEPT1, OCT1, BCRP, and MRP1 more closely resemble the human tissue distribution. Such expression patterns, together with a broader knowledge of substrate specificity and catalytic activity of the dog enzymes, may be incorporated into physiology-based simulation tools to improve the mechanistic understanding of pharmacokinetics in dogs and, consequently, in humans.
Participated in research design: Schuler and Belli.
Conducted experiments: Haller, Schuler, Bachir-Cherif, and Belli.
Contributed new reagents or analytic tools: Haller and Bachir-Cherif.
Performed data analysis: Haller, Lazic, and Belli.
Wrote or contributed to the writing of the manuscript: Haller, Schuler, Lazic, Bachir-Cherif, Krämer, Parrott, Steiner, and Belli.
We thank A. Cetinsu, P. Schrag, M. Festag, and A. Roth for their expert technical support, and A. Augustin and D. Avila for their precious help with the Western blot analysis.
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
- absorption, distribution, metabolism, and excretion
- cytochrome P450
- multidrug resistance
- multidrug resistance-associated protein
- breast cancer resistance protein
- peptide transporter
- organic cation transporter
- glyceraldehyde-3-phosphate dehydrogenase
- hypoxanthine phosphoribosyl-transferase
- ribosomal protein L32
- generalized additive mixed model
- RNA quality indicator
- real-time quantitative reverse transcription-polymerase chain reaction
- correspondence analysis
- pregnane X receptor
- coefficient of variation.
- Received March 2, 2012.
- Accepted May 17, 2012.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics