Parts of this work were published in abstract form (Gastroenterology 1987;92:1792)
Potential conflict of interest: Nothing to report.
Experiments were performed in 2 volunteers to define the biotransformation and physiological properties of norursodeoxycholic acid (norUDCA), the C23 (C24-nor) homolog of UDCA. To complement the in vivo studies, the biotransformation of norUDCA ex vivo using precision-cut human liver slices was also characterized. In the human studies, both a tracer dose given intravenously and a physiological dose (7.9 mmol, 3.0 g) given orally were excreted equally in bile and urine. By chromatography and mass spectrometry, the dominant biotransformation product of norUDCA in bile and urine was the C-23 ester glucuronide. Little N-acyl amidation (with glycine or taurine) occurred. The oral dose induced a sustained bicarbonate-rich hypercholeresis, with total bile flow averaging 20 μL/kg/min, a rate extrapolating to 2 L/d. The increased bile flow was attributed to cholehepatic shunting of norUDCA as well to the lack of micelles in bile. Phospholipid and cholesterol secretion relative to bile acid secretion decreased during secretion of norUDCA and its metabolites, presumably also because of the absence of micelles in canalicular bile. When incubated with human liver slices, norUDCA was glucuronidated, whereas UDCA was conjugated with glycine or taurine. In conclusion, in humans, norUDCA is glucuronidated rather than amidated. In humans, but not animals, there is considerable renal elimination of the C-23 ester glucuronide, the dominant metabolite. NorUDCA ingestion induces a bicarbonate-rich hypercholeresis and evokes less phospholipid and cholesterol secretion into bile than UDCA. Molecules that undergo cholehepatic shunting should be powerful choleretics in humans. (HEPATOLOGY 2005;42:1391–1398.)
The common natural C24 bile acids have an isopentanoic acid side chain attached to a C19 steroid nucleus.1 Bile acids with a 4-carbon (isobutanoic acid) side chain occur as trace constituents of human bile2 and as major biliary constituents in the sterol carrier protein 2 knockout mouse.3 Work in rodents indicates that such C23 nor- bile acids differ in their biotransformation and physiological properties from those of their corresponding natural C24 homologs.* Natural bile acids (with a C5 side chain) are efficiently N-acyl amidated (conjugated) with glycine or taurine.1 In contrast, nor- bile acids undergo little N-acyl amidation and, instead, are metabolized as drugs. The pattern of biotransformation depends on the structure of the nor- bile acid. Norcholate is secreted into bile unchanged.4–6 Nordihydroxy bile acids, such as norursodeoxycholic acid (norUDCA), are secreted into bile in part in unchanged form, in part as a trihydroxy derivative, and in part as a glucuronide or sulfate conjugate.7–13
The physiological properties of nor- bile acids also differ from those of their corresponding C24 homologs. Compared with cholic acid, norcholic acid induces more bile flow and less phospholipid secretion.5 Nordihydroxy bile acids induce a still greater choleresis, and bile becomes markedly enriched in bicarbonate.7–9, 12 To explain the hypercholeresis, a cholehepatic shunt hypothesis was proposed.7, 14 In this scheme, the nordihydroxy bile acid is absorbed passively in protonated (uncharged) form by cholangiocytes. Removal of a proton from ductular bile generates a bicarbonate anion, and there is no change in the osmolarity of bile. The absorbed bile acid returns to the hepatocyte via the periductular capillary plexus and is resecreted into the canaliculus, generating additional canalicular bile flow. Such cholehepatic shunting can occur multiple times, each cycle inducing canalicular bile flow and enriching ductular bile in bicarbonate.
In this report, we have defined the metabolism of norUDCA as well as its secretory properties in 2 human volunteers. The biotransformation of norUDCA was also defined in an ex vivo study using human liver slices. Human liver slices have proved useful for the study of human-specific drug metabolism, having drug metabolizing capacities that correspond to in vivo values.15–17 The chemical structure of norUDCA and UDCA are shown in Fig. 1.
NorUDCA of high purity was synthesized by the Diamalt Company, Munich, Germany, and was a gift of the Falk Foundation, Freiburg, Germany. [23-14C]-norUDCA was synthesized in this laboratory.18 [22,23-3H]-UDCA was prepared by reductive tritiation of Δ22, 23–UDCA.19 The tritium label of 22,23-3H-cholic acid has been shown to be stable in vivo.20
Biotransformation of labeled bile acids was characterized by zonal scanning of thin-layer chromatography (TLC) plates11 after separation using acidic11 or alkaline21 solvent systems. Group-specific spray reagents were also used to detect glucuronides; for example, to detect carbohydrate-linked bile acids, a naphthol-resorcinol-phosphoric acid spray was used. To determine the chemical identity of the compound reacting as a carbohydrate conjugate, the adjacent zone with the identical Rf value was scraped from the plate, and its content eluted with methanol. Mass spectrometry was performed on the eluate using a Kratos VG-70SE instrument in the negative mode.22 To determine whether glucuronides were of the ethereal or ester type, samples were subjected to alkaline methanolysis, which hydrolyzes ester glucuronides but leaves ethereal glucuronides intact.23 Total bile acids were measured using the 3α-hydroxysteroid dehydrogenase method.24 This method measures C-23 ester glucuronides (with a reactive 3α-hydroxy group), but not ethereal glucuronides (with an ether linkage at C-3). Because greater than 90% of the glucuronides were ester glucuronides (see Results), the measured bile acid concentration using the enzymatic method is a very slight underestimate of the true bile acid concentration. Total phospholipid was determined enzymatically.25 Cholesterol was determined enzymatically using a kit (Sigma-Aldrich, St. Louis, MO). Biliary bicarbonate was determined gasometrically using a microgasometric device (Micro CO2 system, American Scientific Products, McGaw Park, IL).
In Vivo Studies.
Two volunteers were studied after giving written informed consent. The study protocol was approved by the University of California, San Diego Human Subjects Committee on January 9, 1986. NorUDCA was administered with permission of the Food and Drug Administration, IND 25,533.
The first study was performed in a 65-year-old man with a cholecystectomy many years before the study. A double-lumen tube was prepared with 2 balloons that were inflated to isolate a 5-cm segment of the second portion of the duodenum. A proximal lumen was used to infuse a nonabsorbable recovery marker, and a distal lumen was used to aspirate the recovery marker and biliary secretions.26 Tracer [23-14C]-norUDCA, 7 μCi, dissolved in normal saline, was administered intravenously. Bile was collected at 10-minute intervals for 160 minutes. Collections were stopped at this time because it was assumed erroneously that biliary secretion of the tracer was complete. Urine was collected from 0 to 3 hours and 3 to 12 hours.
The second study took place in a 62-year-old man with chronic obstructive pulmonary disease who had an indwelling cholecystostomy tube; the tube had been placed some weeks previously to treat acute cholecystitis because the patient was a poor operative risk. Liver tests and the leukocyte count were normal at the time of the study. NorUDCA, 7.9 mmol, was added to 200 mL of 100 mmol/L sodium bicarbonate. To this solution was added 3 μCi [23-14C]-norUDCA. The patient then drank the solution. Bile was collected for the next 4 hours, and urine was collected for the next 24 hours.
In Vitro Studies Using Human Liver Slices.
The metabolism of 23-14C-norUDCA and 22,23-3H-UDCA was examined in human liver slices at the University of Groningen. The protocol was approved by the Medical Ethical Approval Committee of the University Hospital Groningen, and informed consent was obtained. A liver sample was obtained from a patient undergoing partial liver resection because of colon cancer metastases. Precision-cut liver slices (8-mm diameter and 250-μm thickness) were prepared with a Krumdieck tissue slicer as described.15, 16 Slices were incubated in 3.2 mL Williams Medium E with 25 mmol/L glucose and gassed with 95%O2/5%CO2. After a preincubation of 60 minutes, 10 μmol/L UDCA containing tracer [22,23-3H]-UDCA or 10 μmol/L norUDCA containing tracer [23-14C]-norUDCA was added and the slices were incubated for an additional 3 and 16 hours. At the end of the incubation, medium and slices were homogenized, and the samples were frozen at −80°C.
Samples were shipped to San Diego, CA, and kept frozen until analysis. Cells were extracted with chloroform-methanol, 2:1, vol/vol. Bile acids were isolated from the medium by adsorption to C8 hydrophobic columns (Isolute, International Sorbent Technology, Glamoran, UK) followed by elution using methanol. The chemical form of radioactivity was determined by zonal scanning.
The apparent choleretic activity (ACA) of norUDCA and its metabolites was calculated in the second study when 7.9 mmol norUDCA was ingested. ACA is defined as Δ bile flow in hepatic bile/Δ bile acid recovered in hepatic bile.8 (The qualifier “apparent” is used because hepatic bile flow may differ from canalicular bile flow because of ductular absorption or secretion, and bile acid recovery may differ from canalicular bile acid secretion because of bile acid absorption in the biliary tree.) Total bile flow is the algebraic sum of canalicular bile secretion and ductular bile flow, the latter being the algebraic sum of ductular secretion or absorption. Canalicular bile flow can be divided into bile acid–dependent flow (BADBF) and bile acid–independent flow (BAIBF). Thus,
Bile was collected in six 15-minute intervals before the norUDCA was ingested (basal period). Bile acid output during this time is termed ”endogenous” bile acid secretion. The ACA of the endogenous bile acid was assumed to be 15 μL/μmol throughout the study.27 The product of endogenous bile acid output and the ACA gives the canalicular bile flow attributable to endogenous bile acid secretion. This value was subtracted from the observed bile flow to give non–bile acid–dependent bile flow [the sum of BAIBF (canalicular) and ductular bile flow].
During the period of induced choleresis, endogenous bile acid output continued but decreased from that observed during the basal periods. Its value was obtained by subtracting exogenous bile acid recovery (radioactivity divided by specific activity) from total bile acid output, as measured enzymatically. The remainder of bile flow was attributable to the choleretic effect of norUDCA and its metabolites in bile, permitting the ACA of the secreted norUDCA and its metabolites to be calculated. Thus,
where J is apparent bile acid secretion (bile acid recovery in duodenal bile).
When cholehepatic shunting of secreted (unmetabolized) norUDCA occurs, there is a stoichiometric generation of bicarbonate ion, signaled by enrichment of bile in bicarbonate concentration and a corresponding increase in biliary bicarbonate output. This value was used to calculate the fraction of secreted norUDCA that was absorbed in the biliary ductules.9 This value was added to recovered bile acid in bile to calculate secretion of norUDCA and its metabolites.
where BF denotes bile flow and the subscripts denote the steady-state period achieved after norUDCA was ingested and the basal (preingestion) periods.
Data are presented as mean ± standard deviation. Differences between steady-state and basal-state values of bile acid and bicarbonate concentration as well as biliary lipid coupling were tested for statistical significance using the Wilcoxon rank-sum test.
Biotransformation of norUDCA: In Vivo Studies in Healthy Volunteers.
The tracer dose of norUDCA administered intravenously was excreted slowly in bile, based on recovery of the duodenal aspirate. Only 38% of the infused dose was excreted by 160 minutes, and excretion was continuing at this time. Urinary recovery was 15% in 3 hours and 15% in the 3- to 12-hour collection, giving a total urinary recovery of 30%. Total recovery in bile and urine was 68%, of which 56% was in bile and 44% in urine. Radioactivity in the duodenal collection was entirely in the chemical form of norUDCA glucuronide, based on TLC mobility (Fig. 2). When subjected to alkaline methanolysis, 91% of radioactivity acquired the mobility of the methyl ester of norUDCA, indicating that the glucuronide was predominantly of the ester type.
In the second study, a physiological dose of norUDCA (7.9 mmol, 3.0 g) containing a small amount of radioactivity was ingested by a patient with a biliary fistula. Radioactivity excretion in bile increased immediately, remained elevated for 3 hours, and then declined at 4 hours. Figure 3, bottom panel, shows the time course of radioactivity recovery. Biliary recovery was 30% of the ingested dose. Urinary recovery was 16% at 6 hours, 17% at 6 to 16 hours, 7% at 16 to 24 hours, and 3% at 24 to 40 hours. Total recovery in urine was 43% of the administered dose. Total recovery in bile (in 4 hours) and urine (in 40 hours) was 73% of the ingested dose.
By TLC, biliary radioactivity was predominantly (90%) norUDCA glucuronide based on its TLC mobility and its reactivity with a spray reagent for sugars (Fig. 4, left panel). Approximately 2% had the mobility of the glycine amidate of norUDCA; approximately 0.3% had that predicted for a sulfate of norUDCA; and approximately 5% had the mobility of unchanged norUDCA. After alkaline methanolysis, more than 90% of biliary radioactivity moved as the methyl ester of norUDCA, indicating that the norUDCA-glucuronide was predominantly the ester glucuronide. Urinary radioactivity had a similar profile with 93% of the label having the chromatographic properties of norUDCA glucuronide and 7% having that of unchanged norUDCA (Fig. 4, right panel).
Confirmation that the dominant metabolite of norUDCA was a glucuronide was shown by mass spectrometry of the eluate of the TLC zone adjacent (i.e., with the same Rf value) to the spot containing the major metabolite detected by the sugar spray reagent (Fig. 5). The major peak had the m/z value of 553, corresponding to the glucuronide of norUDCA. A small amount of norUDCA was also present (m/z = 377), but this was considered to be an artifact arising from post-chromatography hydrolysis of the ester glucuronide catalyzed by the acidic solvent system used for the TLC separation.
Biotransformation of norUDCA and UDCA by Human Liver Slices.
After incubation of norUDCA or UDCA with liver slices, most radioactivity (97%) was present in the medium. In the medium (Table 1), norUDCA was present predominantly as the unchanged compound, although its glucuronide and sulfate were present in low proportions; at 16 hours, a trihydroxy metabolite was also present. In contrast, UDCA was mostly conjugated with glycine or taurine, although small amounts underwent hydroxylation, glucuronidation, or sulfation.
Table 1. Biotransformation of norUDCA and UDCA in Precision-Cut Human Liver Slices: Distribution of Radioactivity Recovered From Medium, as Determined by TLC-Zonal Scanning
In the cells (Table 2) after 3 hours of incubation, norUDCA was metabolized predominantly to a glucuronide. UDCA, in contrast, was efficiently amidated; after 16 hours of incubation, 93% had been conjugated with glycine or taurine. The data also indicate that metabolites were rapidly excreted from the cells into the medium.
Table 2. Biotransformation of norUDCA and UDCA in Precision-Cut Human Liver Slices: Distribution of Radioactivity Present in Cells as Determined by TLC-Zonal Scanning
Data from the second human study were used to calculate the effect of absorbed norUDCA on bile flow and biliary lipid secretion. During the basal period, bile flow was relatively constant (6.6 ± 2.0 μL/kg/min, n = 6). After ingestion of norUDCA, bile flow increased at 15 minutes, continued to rise until 20 minutes and remained constant for the next 90 minutes (Fig. 3, upper panel). During the six 15-minute periods between 30 minutes and 120 minutes, exogenous bile acid output (norUDCA and metabolites) and bile flow remained relatively constant. Data from this steady-state period were used for calculation of the ACA value of norUDCA and its metabolites, as described in Materials and Methods.
Apparent biliary secretion (i.e., recovery in ductal bile) of norUDCA and metabolites averaged 0.24 ± 0.02 μmol/min/kg. Endogenous bile acid secretion declined during the steady-state period with the result that bile became progressively enriched in norUDCA and its metabolites. The proportion of bile acid output attributable to norUDCA and its metabolites rose from 53% to 76% during the 6 periods of the steady-state interval. Bile flow remained relatively constant at 20.2 ± 1.5 μL/kg/min. Bile flow caused by endogenous bile acid secretion was 2.3 ±1.0 μL/kg/min. and non–bile acid–dependent bile flow was assumed to be 3.9 μL/kg/min, based on the collections during the basal period. Thus, secretion of norUDCA and metabolites induced most canalicular bile flow during the steady-state period (14.2 ± 1.4 μL/min/kg), as shown in Fig. 3, top panel. The calculated ACA for norUDCA and metabolites was 58.5 ± 4.3 μL/μmol. Values of this magnitude are considered evidence of an induced hypercholeresis.14
The average bile acid concentration in bile collected during the steady-state period was 18.8 ± 2.8 mmol/L, a value well below that recorded during the basal period (41.8 ± 5.6 mmol/L, P < .02). During this interval, norUDCA and its metabolites were present at a concentration of 12.1 ± 0.9 mmol/L. This concentration is well below the critical micellization concentration (CMC) of norUDCA, which is 17 mmol/L.28 The CMC of the ester glucuronide of norUDCA should be considerably higher that that of norUDCA. Therefore, both norUDCA and its ester glucuronide were probably present in bile mostly in monomeric rather than in micellar form. The concentration of endogenous bile acids decreased during the choleresis period from 9.0 mmol/L to 4.0 mmol/L. The CMC of the bile acid mixture present in human bile in the absence of phospholipid is approximately 3 mmol/L,29 so that the concentration of micelles (total concentration minus CMC) in canalicular bile decreased from 39 mmol/L to 1 mmol/L during the steady-state period of choleresis.
Biliary bicarbonate averaged 28.0 ± 1.8 mmol/L during the basal period and increased to an average of 46 mmol/L (Fig. 2, middle panel, P < .02)). The bicarbonate enrichment (18.2 ± 2.7 mmol/L) was used to calculate the fraction of norUDCA secreted into canalicular bile that was absorbed in the biliary ductules.9 In this calculation, exogenous bile acid output was multiplied by 0.05 (the fraction of norUDCA in exogenous bile acid secretion as determined by zonal scanning). [Fractional ductular absorption is equal to “extra bicarbonate” divided by calculated canalicular secretion of norUDCA (recovery + extra bicarbonate)]. Calculated ductular absorption of norUDCA averaged 97%. Total canalicular secretion of norUDCA and its metabolites was 0.75 μmol/kg/min, a value approximately twice the bile acid secretion rate during digestion in healthy individuals.30
Induced Secretion of Phospholipids and Cholesterol.
The well-known induction of phospholipid secretion by bile acids may be expressed as a coupling coefficient, that is, Jphospholipid/Jbile acid where Jphospholipid is the recovery of phospholipids in bile and Jbile acid is the calculated canalicular secretion (recovery plus absorbed) of norUDCA and metabolites.31 This calculation assumes that phospholipid is neither absorbed nor secreted in the biliary ductules. The coupling coefficient was 0.24 ± 0.03 (n = 4) during the basal period and decreased to 0.089 ± 0.014 (n = 6) during the steady-state period (P < .02). Thus, secretion of norUDCA and its metabolites induced relatively little phospholipid secretion in bile.
Cholesterol secretion in bile is also induced by bile acid secretion. The coupling coefficient Jcholesterol/Jbile acid (× 103) decreased from 94 ± 13 (n = 4) during the basal period to 24 ± 10 (n = 6) during the steady-state period of induced choleresis by norUDCA and its metabolites (P < .02).
These studies describe the metabolism and physiological properties of norUDCA in humans. In addition, the study showed that the phenomenon of hypercholeresis can be induced by ingestion of norUDCA. The study also indicates that the metabolism of norUDCA in humans differs in two respects from that in experimental animals in which norUDCA pharmacology has been characterized previously.7, 12, 13
The first major difference is that norUDCA had considerable renal excretion. In animal studies of norUDCA metabolism, no urinary loss was observed.7 Because a considerable fraction of norUDCA (as its ester glucuronide) was eliminated in urine when a tracer dose was given, renal excretion cannot be explained by saturation of canalicular transport. Renal excretion is likely to have resulted from poor canalicular transport as well as efflux of the ester glucuronide from the hepatocyte into plasma by a basolateral efflux transporter in the MRP family.32
A second difference between the metabolism of norUDCA in humans and that in rodents is that the only major metabolite of norUDCA in humans was the ester glucuronide. In Fischer rats and hamsters, approximately half of norUDCA undergoes ethereal glucuronidation; in the guinea pig, almost all glucuronidation of norUDCA occurs at C-3.7
The in vivo studies were confirmed by the liver slice incubations, which showed that the major metabolite of norUDCA was a glucuronide, and that little N-acyl amidation occurred. In contrast, UDCA was conjugated with glycine or taurine, in agreement with in vivo human studies33
We explain the strong choleresis induced by norUDCA and its metabolites by two mechanisms. The first is the cholehepatic shunting of secreted norUDCA as evidenced by the enrichment of bile in bicarbonate. The second is the greater osmotic effect of secreted bile acids because norUDCA, its metabolites, and endogenous bile acids were present in mostly monomeric rather than in micellar form. Total bile flow during the hypercholeresis induced by norUDCA, if extrapolated to 24 hours, would approximate 2 L/d, a value approximately 3 times that of estimates of normal bile flow.34 The calculated absorption of norUDCA in the biliary tract (97%) was high, consistent with the known high membrane permeability of nordihydroxy bile acids.35
NorUDCA also induced less phospholipid and cholesterol secretion than endogenous bile acids. Synthetic bile acid epimers, per-oxo bile acids, and hydroxy oxo bile acids with CMC values well above those of natural bile acids have been shown to induce less phospholipid and cholesterol secretion than that induced by natural bile acids in numerous animal studies.31 Presumably because such molecules are less surface active, they adsorb less to the phospholipid hemi-vesicles protruding from the canalicular membrane36 and induce correspondingly less phospholipid secretion; the reduced proportion of bile acids in micelles also may contribute.
The current study suggests that developing bile acids that are potent, long-acting choleretic agents is possible. Such molecules must undergo little hydroxylation or conjugation during hepatocyte transport, be sufficiently hydrophobic to undergo efficient absorption in the biliary tree, and be efficiently re-secreted into canalicular bile. Cholehepatic shunting results in a continuing flux of molecules across the biliary ductular epithelium, and this might have therapeutic value, based on animal studies. In knockout mice lacking the cystic fibrosis gene product, deficient bicarbonate secretion by cholangiocytes occurs. Administration of norUDCA to such mice has been reported to restore bile alkalinity and bile flow.37 In the mdr2 knockout mouse, whose phenotype is absent biliary phospholipid secretion, peribiliary fibrosis is diminished by norUDCA but not by UDCA.38 Sulindac, a non-steroidal anti-inflammatory drug that also undergoes cholehepatic shunting in rats,39 has been shown to improve plasma aminotransferase levels in primary biliary cirrhosis patients when used as an adjunct to UDCA.40
Biotransformation patterns in humans may differ considerably from those in rodents, as shown by the current study as well as many others.41 The use of human liver slices permits human-specific biotransformation patterns to be detected early in the development phase of new drug candidates.
Micheal Scharl, University of Regensburg, provided technical assistance with the zonal scanning procedure.
The affix “nor” is defined by the IUPAC Compendium of Chemical Terminology to denote the elimination of one methylene group from a side chain of a parent structure. (The term is derived from an acronym in the German language: Nitrogen ohne Radikahlen, which was used to distinguish norepinephrine from epinephrine). For more complex molecules such as steroids or bile acids, indicating the location of the missing carbon (as per IUPAC Steroid Nomenclature rule 3S-6.2) is necessary. Thus, norUDCA, a C23 bile acid, is referred to as 24-nor, meaning that the side chain has been shortened by elimination of C-24 to produce the homolog with a C-23 carboxyl group.