Enterohepatic circulation of glucuronide metabolites of drugs in dog

Abstract The enterohepatic circulation (EHC) of drugs is often the result of the direct glucuronidation, excretion of the metabolite into bile, followed by hydrolysis to the aglycone by the gut microbiome and finally reabsorption of drug into the systemic circulation. The aim of present study to identify key factors in determining the EHC in dog for canagliflozin and DPTQ, two compounds cleared by UDP‐glucuronosyltransferase (UGT) mediated O‐alkyl glucuronidation and cytochrome P450 (P450) mediated oxidation. The pharmacokinetic profiles of the drugs were compared between bile duct cannulated (BDC) and intact beagle dogs after a single intravenous administration. A long terminal elimination phase was observed for DPTQ but not for canagliflozin in intact dogs, while this long terminal half‐life was not seen in BDC animals, suggesting the EHC of DPTQ. Quantification of parent drugs and glucuronide metabolites in bile, urine and feces indicated low recovery of parent in bile and urine and low recovery of conjugated metabolites in urine for both drugs, while biliary excretion of these glucuronide metabolites in BDC dog were low for canagliflozin but much higher for DPTQ. The increased fecal recovery of parent drug in intact dog and the lack of glucuronide metabolites suggested the hydrolysis of DPTQ‐glucuronides by gut microbiome. Subsequent characterization of in vitro hepatic metabolism and permeability properties indicated the hepatic fraction metabolized by UGT, hydrolysis of metabolites, and reabsorption of the aglycone were key factors in determining the EHC of DPTQ.


| INTRODUC TI ON
Enterohepatic circulation EHC) is a process composed of a circuit of hepatic metabolism, biliary excretion, gut microbiome metabolism, followed by reabsorption from the gut back to systemic circulation. 1,2 Forty-five drugs were identified that undergo EHC in a recent review article. 3 Among these, the drug itself may be directly secreted into bile without undergoing metabolism; others undergo conjugation in the liver, such as glucuronidation and sulfation, then the metabolites are excreted in bile, stored in gallbladder, and released to the gut, where they undergo hydrolysis back to the parent drug by gut microbiome. 2 Pharmacokinetic parameters such as half-life (t 1/2 ), volume of distribution (V dss ), area under the concentration-time curve AUC) and bioavailability often are substantially affected by EHC with orally administered drugs, 2 demonstrating the importance of assessing the magnitude of EHC to model the pharmacokinetic/pharmacodynamics PK/PD) for a clinical candidate in drug discovery. This is especially true for compounds where a minimum concentration needs to maintained over a 24 hours period. A recent study investigated the fraction of hepatic metabolism fm) in dogs and suggested EHC will also affect the cumulative fm estimation of drugs that are metabolized by both P450s and UGTs in the liver, when the glucuronide metabolites are excreted into bile and hydrolyzed by gut microbiome during the process of EHC. 4 In the present study, the EHC of canagliflozin, known to be cleared by P450 mediated oxidation and UGT mediated O-glucuronidation in human, 5,6 and DPTQ 2-2,6-dichlorophenyl)-1-1S,3R)-5-2-hydroxy-2methylpropyl)-3-hydroxymethyl)-1-methyl-3, 4 dihydroisoquinolin-21H)-yl)ethan-1-one, 7 a discovery project compound primarily cleared by hepatic metabolism of UGT and P450, were investigated in dog. The dog was used since it is a preclinical species that is more physiologically relevant to human than rodents with respect to bile secretion, 8 biliary drug excretion 9 and colonic absorption. 10 The PK profiles were compared between bile duct cannulated (BDC) and intact dogs after a single intravenous administration of the drugs, and the distribution of O-glucuronide metabolites in plasma, bile, urine and feces was quantified to explore the fate of the metabolites during EHC. Subsequent in vitro hepatic metabolism and permeability studies were performed to characterize the physicochemical and metabolic properties that favor EHC. Given the disposition of canagliflozin has been reported for humans, 11 a comparison of the in vitro and in vivo metabolic and excretion data of canagliflozin in human to those obtained with dog was performed in order to identify the species differences in the process of EHC.

| Determination of pharmacokinetic parameters
Noncompartmental pharmacokinetic parameters were calculated using Watson version 7.4 (Thermo Scientific, Waltham, MA).
Statistical significance of differences for pharmacokinetic parameters between two groups was examined with Student's t test using Sigmaplot 12.5 (Systat Software Inc, San Jose, CA).

| Determination of fraction of CYP-mediated oxidative metabolism in hepatocytes
The intrinsic clearance was determined in cryopreserved dog and human hepatocytes essentially as described previously. 4 The assay

| Bi-directional transport across MDCK cells
MDCKII cells stably expressing human wild-type P-gp were obtained from the Netherlands Cancer Institute Amsterdam, The Netherlands). MDCK cells were maintained and the assay was

| Liver and kidney microsomal glucuronidation assay
Intrinsic clearance of the compounds by glucuronidation was conducted in human and dog liver and kidney microsomes. Microsomes The microsomal unbound fraction both with and without 2% BSA was determined as described previously. 15

| LC-MS/MS quantification of drugs and metabolites
Canagliflozin and DPTQ in plasma, bile, urine and feces were quantified using LC-MS/MS. All samples plasma, urine and bile) were mixed with an organic internal standard solution essentially described previously, 15
Topological polar surface areas TPSA) were calculated based on the method described previously. 16 The number of hydrogen bond    Table 2), suggesting the enterohepatic circulation (EHC).

| Recovery of parent drugs and glucuronides in urine, feces and bile of intact and BDC dogs
The extent of renal excretion of canagliflozin, DPTQ and canagliflozin-glucuronides was negligible (<1% of dose) and the renal recovery of DPTQ-glucuronide was low (<5% of dose

| In vitro hepatocyte Cl int and metabolite profile
The intrinsic clearances (Cl int ) of canagliflozin and DPTQ characterized by parent loss in dog hepatocytes were inhibited 12.7 and 46.5% by ABT, respectively (  Table 4).
The subsequent dog hepatocyte metabolite profile indicated that oxidation and glucuronidation were major metabolic pathways for DPTQ in dog hepatocytes ( Figure 3 and Table 5). In the presence of ABT, the primary oxidative metabolite C was not detected, whereas glucuronide metabolite metabolite B) peak in the profile was not meaningfully reduced <20% change for metabolite/internal standard peak area ratio). Product A P+O+glucuronidation) was a secondary metabolite and the formation was inhibited by ABT. Note: Dogs were given canagliflozin and DPTQ at 0.5 mg/kg by IV bolus. Values were derived from plots shown in Figure 2. Results shown represent mean ± SD from four animals. a P < 0.01 compared to the parameters in intact animals.

| Liver and kidney microsomal glucuronidation
The unbound intrinsic clearances (Cl int,u ) determined by parent loss in liver and kidney microsomes were examined in order to understand whether the O-glucuronides detected in urine are produced by glucuronidation in the kidney or are formed in the liver, excreted into blood, and then eliminated from circulation by the kidney. The microsomal incubation was conducted with or without 2% bovine serum albumin (BSA), which is known to reduce K m values for substrates of UGT isoforms that are inhibited by long-chain unsaturated fatty acids released by incubation with human liver microsomes. 18 The Cl int,u was higher in the presence of BSA in human microsomes for canagliflozin (liver and kidney) and DPTQ (liver), as well as in dog liver microsomes for DPTQ ( Table 6). The substantial Cl int,u of glucuronidation in human kidney suggested the canagliflozin O-glucuronides detected in human urine could be locally produced in the kidney ( Table 6). The negligible glucuronidation in both liver and kidney microsomes of dog was consistent with negligible circulating and urinary recovery of canagliflozin-O-glucuronides. In contrast, the glucuronidation activity towards DPTQ was abundant in dog liver microsomes but not detected in kidney microsomes, suggested the urinary recovery of DPTQ-O-glucuronide in dog is mainly from circulation after hepatic metabolism. The lower glucuronidation activity in human liver microsomes was consistent with lower hepatocyte Cl int compared to dog for DPTQ, and glucuronidation intrinsic clearance by human kidney microsomes was not detected (Table 6).

| MDCK permeability and gut reabsorption
Passive membrane permeability was high for DPTQ and moderate for canagliflozin (Table 7). DPTQ was not a P-gp substrate, whereas canagliflozin was determined to be a P-gp substrate in a P-gp-transfected Note: Dogs were given canagliflozin and DPTQ at 0.5 mg/kg by IV bolus. The bile, urine and feces samples were collected in BDC dogs, and the urine and feces was collected in intact dog over 48 hours post-dose. The bile, urine and feces samples after canagliflozin and DPTQ administration were separated into two aliquots, one for bioanalytical assay of parent drugs (1), and one treated with β-glucuronidase (2) and analyzed for parent drugs as described in Materials and Methods. The concentration of O-glucuronides was determined by subtracting the concentration of (1) from (2). Results shown represent means ± SD (n = 4). Abbreviations: NA, bile concentration in intact dog was not available; ND, the concentration was below the quantitative limit 1 ng/mL. a P < 0.05 compared to that in BDC dog. Note: Incubation contained 0.3 µmol/L compound and 10 6 /mL dog and human hepatocytes with and without pre-incubation of 1 mmol/L ABT for 30 minutes, as described in Materials and Methods. Results shown represent means ± SD (n = 4 or 5).  (Table 7). Compounds with relatively high numbers of hydrogen bond donors and acceptors as well as higher polar surface area are often P-gp substrates, 12 as is the case with canagliflozin (Table 1). This also agrees with a previous report that canagliflozin is a P-gp substrate. 6 The differences in the permeability and P-gp liability are likely to determine the extent of reabsorption to portal circulation and subsequent EHC. Note: Hepatocyte metabolite identification was conducted with and without pre-incubation of 1 mmol/L ABT as described in Materials and Methods. The peak area values were derived from plots in Figure 3. Abbreviations: P, parent drug; IS, internal standard; RT, retention time; ND, not detected. Note: Incubation contained 2 µmol/L compound and 0.5 mg/mL liver and kidney microsomes with and without 2% BSA, as described in Materials and Methods. The unbound Cl int (in the parenthesis) was calculated using Cl int divided by microsomal unbound fraction. Results shown represent average of two independent experiment. Abbreviations: ND, Cl int < 1.8 µL/min/mg protein, the quantitative limit.

Drugs
TA B L E 6 Hepatic and kidney UGT Cl int and unbound Cl int of canagliflozin and DPTQ

| D ISCUSS I ON
Enterohepatic circulation (EHC) can increase apparent volume of distribution (V dss ) and prolong half-life (t 1/2 ) of drugs. The more extensive the recycling, the more prolonged the t 1/2 and the greater the volume increase. 19 In the BDC dog, the EHC is blocked and the increase in apparent V dss and prolonged t 1/2 are no longer observed  (Table 8).
The EHC of canagliflozin in human was also reported to be negligible. 11 However, the underlying mechanism resulting in the same outcome was distinct between dog and human. Glucuronidation  Given the complexity in the EHC, the PK profile and parameters were less variable for canagliflozin in both animal groups and DPTQ in the BDC dog compared to DPTQ in intact dog. The dog is generally considered as the most biologically relevant preclinical species in investigating biliary excretion of xenobiotics. 8,9 In addition, the dog model is considered a suitable surrogate for the estimation of human colonic absorption of passively absorbed drugs. 10  It should be noted that the drugs were intravenously administered in fed condition in the present study, a simplified model to focus on identification of metabolic and physicochemical properties that influence EHC. Both canagliflozin 6 and DPTQ exhibited negligible intestinal glucuronidation in human intestinal microsomes data not shown). Extensive intestinal glucuronidation could complicate the assessment of fecal recovery of drugs and glucuronides for both intravenously and orally administered drug, such as ezetimibe, 26 but this was not the case with the two compounds in this study. In addition, the investigation of EHC was focused on O-alkyl-glucuronides in the present study, and EHC may not be as extensive in drugs that are metabolized to N-glucuronides or acyl-glucuronides, given the hydrolysis of N-glucuronides by β-glucuronidase by gut microbiome may not be as effective as for O-glucuronides 27 ; acyl-glucuronides could also undergo rearrangement and produce β-glucuronidase resistant species 28 that would complicate the assessment of EHC.
In conclusion, we demonstrated the hepatic UGT metabolism and/or permeability are crucial to EHC using canagliflozin and DPTQ as tool molecules. The models and the assays utilized in the present study, as summarized in Figure 4 and

ACK N OWLED G EM ENT
We are very grateful to Drs. Erik Hembre, Scott Monk and Nathan Yumibe for helpful discussions.

D I SCLOS U R E S
None declared.