Abstract
Acyl glucuronides are intrinsically reactive metabolites of carboxylate drugs, capable of undergoing hydrolysis, intramolecular rearrangement (isomerization via acyl migration), and intermolecular transacylation reactions. Transacylation with nucleophilic groups located on protein molecules leads to covalent drug-protein adducts. Protein adducts can also form from the rearrangement isomers via a glycation mechanism. In this study, the isolated perfused rat liver preparation was used to separately trace the dispositions of the nonsteroidal anti-inflammatory drug diflunisal (DF), its reactive acyl glucuronide metabolite (DAG), and a mixture of DAG rearrangement isomers (iso-DAG), each administered at 30-μg DF equivalents/ml perfusate (four recirculating perfusions each group). After administration of DF, the drug was eliminated in a log linear manner over 3 hr, with apparent elimination half-life (t½) of 2.6 ± 0.4 hr. The sulfate conjugate (DS), excreted almost exclusively into perfusate, accounted for 14.2% of the dose, with the phenolic glucuronide (DPG) and DAG (11.1 and 7.9% of dose, respectively) excreted primarily in bile. Only a small portion (2.3%) of the dose was recovered as novel “diglucuronides” (D-2G, arising from phenolic glucuronidation of iso-DAG), excreted exclusively in bile. Covalent DF-protein adducts were found in both perfusate (0.98%) and liver (0.14%). After administration of DAG, rapid hydrolysis occurred (initial DAGt½ 17.3 ± 4.2 min). At 3 hr, recoveries (in comparison to DF-dosed perfusions) were similar for DF (51.7%) and DAG (8.3%), significantly decreased for DS (10.6%) and DPG (6.4%), and significantly increased for iso-DAG (0.8%), D-2G (9.1%), and covalent adducts in perfusate (1.49%) and liver (0.30%). After administration of iso-DAG, elimination from perfusate was slower (t½ 55 ± 15 min), and hydrolysis to DF was modest by comparison with DAG-dosed perfusions. Recoveries as iso-DAG and D-2G in bile were greatly enhanced (8.2 and 36.4%, respectively). Adduct formation was higher in liver (0.76% of dose) but not in perfusate (1.03%). Immunoblots of liver homogenates revealed drug-modified proteins at ca. 110 and 120 kDa. The results show that (a) DAG undergoes avid systemic deconjugation-conjugation cycling and isomerization to iso-DAG; (b) iso-DAG is more resistant to hydrolysis, is readily taken up by hepatocytes and undergoes novel metabolism (phenolic glucuronidation); and (c) the glycation pathway (i.e. using iso-DAG as substrate) plays a major role in formation of covalent DF-protein adducts in liver.
Acyl glucuronide conjugates result from the coupling of carboxylate substrates with glucuronic acid and, unlike other types of glucuronides, are chemically unstable in vitro and in vivo. This reactivity stems from the susceptibility of the ester group, which links the aglycone and glucuronic acid moieties to nucleophilic substitution reactions (Faed, 1984; Spahn-Langguth and Benet, 1992; Fenselau, 1994). It manifestsvia hydrolysis (regeneration of parent drug), rearrangement (intramolecular acyl migration yielding β-glucuronidase-resistant isomers), and transacylation reactions (fig.1). Transacylation may occur with small nucleophiles such as glutathione (Shore et al., 1995) or with nucleophilic groups located on macromolecules such as proteins (van Breeman and Fenselau, 1985; Williams and Dickinson, 1994; Dinget al., 1995). The latter pathway yields covalent drug-protein adducts in which the glucuronic acid moiety is not retained in the adduct. An alternative pathway of adduct formation involves rearrangement of the acyl glucuronide prior to interaction with the protein. The 2-, 3- and 4-O-β rearrangement isomers formed by acyl migration can undergo anomerization (mutarotation) via transient open-chain intermediates possessing an aldehyde group (fig. 1). The aldehyde group is reactive and can condense with amino groups on protein to yield (reversibly) a Schiff’s base, which can (theoretically) undergo Amadori rearrangement or cyclization reactions. In this “Schiff’s base” or “glycation” mechanism of covalent adduct formation, the glucuronic acid moiety is retained in the adduct (Ding et al., 1993;Smith et al., 1986; Williams and Dickinson, 1994). Irrespective of the mechanism, acyl glucuronide-mediated drug-protein adduct formation represents covalent modification of native proteins by foreign compounds and has attracted attention as a possible explanation for hypersensitivity (Spahn-Langguth and Benet, 1992; Worral and Dickinson, 1995; Zia-Amirhosseini et al., 1995) and certain other toxic responses to acidic drugs (Boelsterli et al., 1995; Smith and Liu, 1995).
Most earlier studies of adduct formation in vitro andin vivo have focused on serum albumin/plasma protein (Spahn-Langguth and Benet, 1992), and progress continues in this areae.g. with acyl glucuronide-derived adducts of salicylic acid (Dickinson et al., 1994), mefenamic acid (McGurk et al., 1996), and gemfibrozil (Sallustio et al., 1997). Adduct formation with nonplasma proteins was first reported for clofibric acid in liver (Sallustio et al., 1991) and subsequently for diflunisal (DF)1in liver, kidney, intestine, skeletal muscle, and bladder (King and Dickinson, 1993; Dickinson and King, 1993) of rats dosed with these drugs. Such hepatic drug-protein adduct formation could be a causative or initiating factor in hepatotoxic responses (e.g.idiosyncratic hepatitis and cholestasis) to acidic drugs (Boelsterliet al., 1995). Studies directed at identification of hepatic protein targets have involved administration of the parent acidic drugs to rodent microsomes, hepatocytes, or live animals (Bailey and Dickinson, 1996; Hargus et al., 1994, 1995; Kretz-Rommel and Boelsterli, 1993, 1994a). Studies administering the acyl glucuronides or their isomers, which could offer considerable mechanistic insights, have been lacking.
We determined to address this deficiency by examining the detailed disposition of the nonsteroidal anti-inflammatory drug DF following its administration to the isolated perfused rat liver preparation in the form of parent drug, acyl glucuronide (DAG), and mixture of acyl migration isomers (iso-DAG). DF (fig. 2) is an interesting model carboxylate drug for such studies, as it forms stable phenolic glucuronide (DPG) and sulfate (DS) conjugates (Dickinson et al., 1991; Watt et al., 1991), as well as reactive DAG, as major metabolites. In addition, iso-DAG undergoes phenolic glucuronidation in rats to form novel quasi “diglucuronides” (D-2G) (King and Dickinson, 1991). The isolated perfused rat liver is a good biological system for studies of this type, as the degree to which acyl glucuronides form hepatic protein adducts is likely to depend on their intrinsic chemical reactivity and on their concentration and residence time in the liver.
Materials and Methods
Materials and Animals.
DF, clofibric acid, and bovine serum albumin (BSA, fraction V) were purchased from Sigma (St. Louis, MO). Authentic samples of DAG were purified from the urine of DF-dosed human volunteers as described previously (Watt et al., 1991). Iso-DAG was obtained by isomerization of DAG under slightly alkaline conditions prior to purification by preparative HPLC, also as described previously (King and Dickinson, 1991). Samples of iso-DAG were used as a 5:8:6 molar mixture of the 2-, 3-, and 4-isomers, respectively [this is the approximate equilibrium ratio of these interconverting isomers in buffer at pH 7.4 (Williams and Dickinson, 1994)], and contained negligible DAG and ≤2% DF. Enhanced chemiluminescence (ECL) reagents, nitrocellulose (ECL grade), and hyperfilm ECL were purchased from Amersham (Sydney, Australia). Goat anti-rabbit IgG (H and L chains)-horseradish peroxidase complex was purchased from Rockland (Gilbertsville, PA). Male Sprague Dawley-derived rats (300–350 g) were supplied by the Medical Faculty Animal House of the University of Queensland. Experiments were approved by the University’s Animal Experimentation Ethics Committee.
Isolated Perfused Liver Experiments.
The rats were anesthetized with isoflurane, and the common bile duct and portal vein were cannulated. After ligation of the hepatic artery and severing of the hepatic vein, the livers were perfused with saline (ca. 100 ml), excised, and placed in a recirculating perfusion system maintained at 37°C. The perfusate (100 ml) consisted of 10% v/v washed human red blood cells, 1% w/v (10 mg/ml) bovine serum albumin, and 0.1% w/v (5.5 mM) D-glucose in a Krebs bicarbonate buffer gassed with 5% v/v CO2 in O2, pH 7.4, and was delivered by a peristaltic pump at a constant flow rate of 15 ml/min. Temperature (37 ± 1°C), perfusate pH (7.4 ± 0.1), portal vein pressure (≤40 mmHg), and bile flow (≥5 μl/min) were continuously monitored. If required, the pH was maintained in the stated range by addition of small volumes of 1 M NaHCO3. Preparations not meeting the viability criteria over the complete experimental period were discarded. After 40 min of stable perfusion, DF, DAG, or iso-DAG was administered as a bolus dose of 3-mg DF equivalents (1.5 ml of 2-mg DF equivalents/ml saline adjusted to pH 6.0) into the perfusate reservoir. Perfusate samples (0.5 ml) were collected predose and at 5, 10, 20, 30, 60, 90, 120, 150, and 180 min after dosing and immediately centrifuged. A 100-μl aliquot of each supernatant was added to 150 μl of internal standard solution (100 μg clofibric acid/ml acetonitrile containing 4% v/v acetic acid), and the samples were stored at −20°C until analysis. Bile was collected over ice into tared vessels containing 1 M acetic acid (100 μl) with frequent mixing to achieve mild acidification (pH 3.5–5.0). Bile samples (predose and hourly for 3 hr) were frozen (−20°C) until analysis. After the last sample collection at 3 hr, the perfusate was centrifuged, and the supernatant was frozen (−20°C) for analysis of covalently bound DF. The liver was rinsed, dried, weighed, homogenized in 2 volumes of 0.1 M phosphate buffer, pH 4.5, and frozen (−20°C) until analysis.
Analyses.
Analysis of noncovalently bound DF, DAG, iso-DAG, DPG, and DS in perfusate and bile samples was carried out as described earlier (Dickinson and King, 1989). In brief, the 100-μl perfusate supernatant samples to which 150 μl of internal standard solution had been added (above) were thawed, vortex mixed, and centrifuged. A 20-μl sample of supernatant was injected into the isocratic HPLC system. The bile samples were thawed, and a 20-μl aliquot mixed with 80 μl of 0.1 M phosphate buffer, pH 4.5, and 100 μl of internal standard solution (100 μg clofibric acid/ml acetonitrile). After centrifugation, a 20-μl sample was applied to the HPLC system. The D-2G in bile was analyzed collectively after alkaline hydrolysis to DPG (i.e. as the difference in DPG content of the bile samples before and after alkaline hydrolysis), also as described previously (King and Dickinson, 1991). For analysis of noncovalently bound DF and metabolites in liver homogenate, a 1-g sample was mixed with 2 volumes of internal standard solution (100 μg clofibric acid/ml acetonitrile). After centrifugation, a 20-μl sample was injected into the HPLC system.
Analysis of covalently bound DF was performed essentially as described earlier (King and Dickinson, 1993). In brief, 1-ml perfusate supernatant samples or 1-g liver homogenate samples were mixed with 4 volumes of acetonitrile containing 4% v/v acetic acid. After centrifugation, the supernatant was discarded, and the pellet was exhaustively washed eight times by resuspending in 5 ml of methanol/diethyl ether (3:1 v/v), followed by centrifugation, to remove noncovalently bound drug and metabolites. The pellets were then dried and digested in 1 ml of 1 M NaOH at 80°C overnight. Following acidification (200 μl of 10 M HCl), addition of internal standard solution (50 μl of 100 μg clofibric acid in water), and extraction into 5 ml of hexane/diethyl ether (1:1 v/v), the organic layer was separated and evaporated. The residue was reconstituted in 100 μl of methanol/water (1:1 v/v), and a 40-μl sample was analyzed by HPLC.
For all analyses, the HPLC equipment and operating conditions, described in detail elsewhere (Dickinson and King, 1989), were used. In brief, the column was a 4-μm Novapak C-18 cartridge contained in a RCM-100 radial compression module and was preceded by a Bondapak C-18 Corasil guard column (Water Associates, Milford, MA). The mobile phase was prepared by mixing 520 ml of methanol with 480 ml of an aqueous solution (0.01 M Na2HPO4adjusted to pH 2.7 with H3PO4, plus added Na2SO4.10H2O at 4% w/v). The flow rate was 2 ml/min with column eluent being monitored at 226 nm. Standard curves were prepared by spiking perfusate, bile, and liver homogenate samples from blank perfusions.
Immunoblotting of Drug-Modified Proteins.
Separation of proteins from liver homogenates from control and drug-treated perfusions was carried out on 7.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) minigels under denaturing conditions. The gels were then equilibrated with transfer buffer for 20 min before electroblotting onto a nitrocellulose membrane (Tsang et al., 1983). Nonspecific binding was blocked with 5% skim milk powder in Tris-buffered saline (TBS; 20 mM Tris base, pH 7.5, containing 150 mM NaCl) for 1 hr at room temperature. Polyclonal DF antiserum (Bailey and Dickinson, 1996) and goat anti-rabbit IgG horseradish peroxidase complex were preabsorbed with control rat liver homogenate to minimize nonspecific staining before incubation with the blots. The blots were incubated with DF antiserum diluted 1/2000 in diluent (TBS containing 1% w/v skim milk powder, 1% w/v BSA, and 0.05% v/v Tween 20). The blots were then rinsed for 5 min followed by three 10-min washes in TBS containing 0.05% v/v Tween 20 and incubated with anti-rabbit IgG horseradish peroxidase complex (diluted 1/1000 in the diluent described above) for 1 hr at room temperature with shaking. After washing as before, modified proteins were visualized with the ECL system and compared with molecular weight standards.
Data Analysis.
Results throughout are expressed as means ± SD. Comparisons between groups were performed using unpaired Student’s ttest. Perfusate half-life (t½) values were determined by linear regression analysis.
Results
The concentration-time profiles of DF, DS, DPG, and DAG in perfusate after administration of DF to the isolated perfused rat liver preparation are shown in fig. 3. The elimination half-life (t½) of DF calculated over the entire study period (5–180 min) was 2.6 ± 0.4 hr. At 3 hr, 42.7% of the DF dose remained unchanged in perfusate, with 9.8% in the liver (table 1). The major metabolite was DS (14.2% of dose at 3 hr), excreted almost exclusively in perfusate. DPG accounted for 11.1% of the dose, the majority of which was excreted in bile (table 1). Although DAG appeared more rapidly than DPG in perfusate, its concentrations remained on a low plateau (fig. 3). DAG and its isomers accounted for 8.3% of the dose at 3 hr, the majority (as for DPG) being found in bile. Only a small portion (2.3%) of the dose was recovered as the novel D-2G in bile. Covalent DF-protein adducts were found in both perfusate and liver, accounting for 0.98 and 0.14%, respectively, of the dose.
Rapid hydrolysis of DAG occurred after its administration to perfused livers (fig. 4). Thet½ of DAG, calculated over the essentially linear 5–60-min period, was 17.3 ± 4.2 min. DF appeared rapidly in perfusate, achieving a peak concentration at 1 hr. Its elimination thereafter (90–180 min) seemed to be linear, with at½ value of 6.5 ± 0.7 hr. DF generated by hydrolysis of DAG is available for reconjugation to DS, DPG, and DAG. Such (partial) regeneration of DAG clearly accounts for the slowing in “elimination” rate of DAG observed after 90 min (fig. 4). Comparison of the 3-hr DAG dose recovery data with that for DF doses (table 1) revealed similar recoveries as DF and DAG, decreased recoveries as DS and DPG, and increased recoveries as iso-DAG, D-2G, and covalent adducts in perfusate and liver.
After administration of iso-DAG to perfused liver, the isomer mixture was eliminated in an essentially linear manner over the entire 5–180-min period (fig. 5), with at½ value of 55 ± 15 min. Hydrolysis to DF occurred but was modest by comparison with that obtained after DAG doses (fig. 4). Of DF metabolites resulting from reconjugation of liberated DF, only DS was measurable in perfusate. Recoveries of DF, DS, DPG, and DAG at 3 hr were considerably lower than after DF or DAG doses (table 1), confirming reduced hydrolysis of iso-DAG. On the other hand, recoveries of iso-DAG and D-2G were substantially increased. Covalent adduct formation was increased in liver but not in perfusate (table 1).
Covalent DF-protein adducts in perfusate and liver accounted for a total of 1–2% of DF, DAG, and iso-DAG doses after 3 hr of perfusion (table 1). The respective 3-hr adduct concentrations were 0.29 ± 0.04, 0.45 ± 0.03, 0.31 ± 0.04 μg DF/ml perfusate, and 0.39 ± 0.11, 0.71 ± 0.07, 1.85 ± 0.33 μg DF/g liver. A rabbit polyclonal antiserum previously raised against DF-modified keyhole limpet hemocyanin (Bailey and Dickinson, 1996) was used to identify drug-modified proteins in the liver homogenates after separation by SDS-PAGE and blotting onto nitrocellulose. Representative immunoblots are shown in fig. 6. Drug-modified proteins with apparent molecular weights of 110 and 120 kDa were observed after all three treatments. A weakly stained 140-kDa band was present in blots from control perfusions.
Discussion
The present results on the disposition of DF in the isolated perfused rat liver extend those obtained in bile-exteriorized rats (King and Dickinson, 1991) and confirm the pivotal role of the liver in metabolism of DF and selective transport of its conjugates. Thus, DF is taken up by hepatocytes and conjugated via sulfation, phenolic glucuronidation, and acyl glucuronidation. Whereas DS (330 Da) is transported predominantly across the sinusoidal membrane into perfusate, DPG and DAG (426 Da) are transported primarily (about two-thirds) across the canalicular membrane into bile. The novel D-2G (602 Da) is excreted almost exclusively in bile. Such preferential hepatic transport is in agreement with studies examining the influence of molecular size and chemical structure on preferential urinary or biliary excretion (Hirom et al., 1976; Klaassen and Watkins, 1984). Thus, in rats, compounds with molecular weight <ca. 350 Da are excreted predominantly in urine, and compounds > ca. 450 Da are excreted predominantly in bile. Compounds of intermediate molecular weight can undergo significant excretion by either route, usually in a complementary fashion.
Given that DS and DPG are chemically stable in vivo and under physiological conditions in vitro (Watt et al., 1991; Dickinson et al., 1991), a comparison of the kinetic (figs. 3-5) and recovery (table 1) data after administration of DF, DAG, and iso-DAG to the perfused liver clearly shows that DAG is disposed (a) primarily by hydrolysis to DF, i.e. systemic deconjugation-conjugation cycling, and (b) secondarily by isomerization to iso-DAG, which in turn leads to D-2G.
The identity and origins of D-2G have been investigated in detail earlier (King and Dickinson, 1991). D-2G is a “diglucuronide” mixture comprising the phenolic glucuronides of the 2-, 3-, and possibly 4-isomers of iso-DAG. There is no evidence to support appreciable phenolic glucuronidation of DAG itself (this would yield a true diglucuronide) or for acyl glucuronidation of DPG (an alternative route to the same true diglucuronide). D-2G is formed sequentiallyvia DF → DAG (enzymic), DAG → iso-DAG (nonenzymic), and iso-DAG → D-2G (enzymic). As there are other competing pathways for each of these steps, the D-2G yield increases with the proximity of the precursor substrate; in the present study, yields of D-2G were 2.3, 9.1, and 36.4% from substrates DF, DAG, and iso-DAG, respectively. We regard formation of D-2G as a novel metabolic pathway in that it utilizes the β-glucuronidase-resistant isomers of DAG, but apparently not DAG itself (or DPG), as substrates for further glucuronidation. There are some parallels here with early work on the comparative disposition of valproate glucuronide and its acyl migration isomers in rats (Dickinson et al., 1986), where certain phase I oxidative metabolism (4-hydroxylation) of the valproate moiety was strongly promoted when the isomers were available as substrates.
The kinetic and recovery data after iso-DAG dosing also clearly show that these polar products are more resistant to systemic hydrolysis (relative to DAG) and are taken up intact by hepatocytes. Such uptake is followed by excretion unchanged into bile (8.2% of the iso-DAG dose) or by metabolism via phenolic glucuronidation followed by excretion into bile (i.e. formation of D-2G, 36.4% of dose). The extent of hepatic uptake of intact DAG cannot be reliably estimated because of its avid systemic hydrolysis to lipophilic DF, which is then taken up and reconjugated in the liver.
In the present study, as in others, formation of covalent drug-protein adducts makes up a small portion, in absolute terms, of acidic drug disposition. Although quantitatively minor, it may nonetheless assume considerable relevance toxicologically (Spahn-Langguth and Benet, 1992), although this is still largely speculative. The concentrations of DF-protein adducts found in perfusate 3 hr after dosing perfused livers with DAG were about 50% higher than after dosing with DF (table1). Although this result agrees in principle with DAG being a precursor of perfusate adducts, there is certainly no direct dependency of adduct formation on perfusate protein exposure to DAG, at least under these perfusion conditions. Thus, exposure of perfusate proteins to DAG, as measured by area under the DAG concentration-time curve from 5–180 min (figs. 3 and 4), was 79 ± 10 μg DF equivalents·min/ml after DF administration and 687 ± 157 μg DF equivalents·min/ml after DAG administration. This equates to an 8.7-fold higher exposure after DAG administration, though covalent adduct formation was only 1.5-fold higher. Clearly, factors in addition to DAG exposure [e.g. blocking of access to binding sites (Williams and Dickinson, 1994)] play a major role in modulating extent of adduct formation. Also of interest in the present study was the finding of lower adduct concentrations in perfusate after dosing with iso-DAG, even though exposure of perfusate proteins to iso-DAG, as measured by area under the perfusate concentration-time curve (2017 ± 348 μg DF equivalents·min/ml), was much greater than after exposure to DAG (above). This “exposure-normalized” comparison would suggest that iso-DAG is a much less reactive substrate for adduct formation with perfusate proteins than DAG. However, in earlier in vitro work, we found that the 4-, 3-, and 2-isomers of DAG (in that order) were better substrates for adduct formation with purified human serum albumin than was DAG itself (Dickinson and King, 1991). We surmised that the present result with perfusate adducts could be explained by the presence of glucose in the medium. Glucose can form protein adducts via the glycation pathway (fig. 1) (Shaklaiet al., 1984), potentially blocking glycation sites from other reactants such as iso-DAG. This question has been the subject of a separate investigation,2 which found that intrinsic differences in the protein (BSA was used in perfusate, human serum albumin was used in the earlier studies) rather than the effects of a competing glycating agent were responsible for the lower adduct yield from iso-DAG.
In liver, adduct concentrations were higher after dosing perfusions with iso-DAG than with DAG (table 1). This result clearly illustrates the capacity of and a role for the glycation pathway of adduct formation in the liver, as it could be concluded that iso-DAG, even though hydrophilic, was taken up intact by liver (shown by the enhanced formation and biliary excretion of D-2G, which utilizes iso-DAG as substrate, when iso-DAG was dosed to perfusions). Evidence of a role for the glycation pathway in hepatic tissue was reported earlier (Kretz-Rommel and Boelsterli, 1994a) for diclofenac administered to rat hepatic microsomes. The present data do not permit an estimate of the contribution that the transacylation pathway makes to hepatic adduct formation (fig. 1). Whereas the glycation pathway necessarily involves the isomers as substrates, the transacylation pathway should be strongly favored by the acyl glucuronide over its isomers, as the ester group of the former is more activated toward nucleophilic attack because of the proximity of the ring oxygen atom of the glucuronic acid moiety. Certainly, this scenario was demonstrated earlier for transacylation reactions of DAG (vs. iso-DAG) with the small chemical nucleophile methanol (Dickinson and King, 1991). It is worth noting that the adduct concentrations found in liver (1.85 ± 0.33 μg DF/g liver) after a modest exposure to iso-DAG (3-hr perfusion) were of the same order as those found in liver when rats were dosed twice daily for up to 7 days with DF at 50 mg/kg iv (maximum concentration 4.8 ± 1.2 μg DF/g liver) (King and Dickinson, 1993).
Apart from pointing to a role for the glycation pathway of adduct formation in liver, the present results do not encourage a more detailed analysis of the origin/disposition of the hepatic adducts. Obviously, the extent of exposure of biomolecules to reactive species (e.g. as measured by area under concentration-time curves) must influence the “yield” of adducts, and this has been usefully noted in earlier studies (Bailey and Dickinson, 1996; Benet et al., 1993; Smith and Liu, 1995). However, there seem to be no obvious correlations between hepatic adduct concentrations and measured parameters in the present study (e.g. perfusate concentrations or biliary excretion of DAG and iso-DAG), which could be used to throw more light on intrahepatic processes relevant to adduct formation. Thus, many factors, both known and unknown (e.g.intrinsic reactivity of glucuronide/isomers, duration of exposure to target proteins, access to and saturability of specific binding sites, stability of adducts in hepatic microenvironments), will contribute to measured hepatic adduct concentrations.
In the context of covalently bound DF-derived species in liver, comment should be made about measured differences in the recoveries of noncovalently bound species in liver after the three treatments. After dosing perfusions with DF or DAG, only free DF (at 9.8 and 10.7% of dose, respectively; table 1) was measurable in liver at 3 hr. However, after dosing with iso-DAG, 7.3% of the dose was recovered as such in liver, with an additional 4.9% being recovered as DF. The origins of the 7.3% iso-DAG recovery should be considered in the light of earlier work (Smith et al., 1990) showing that glycation adducts (in this case of zomepirac and serum albumin) released glucuronide isomers upon incubation under acidic conditions. Thus, the question is raised as to whether the recovery of free iso-DAG in liver could be attributed to hydrolysis of glycation adducts of iso-DAG during analytical workup. We feel such a scenario is unlikely because of the very mild conditions (of pH, temperature, and time) employed in the present analytical workup. More likely, the recovery of 7.3% of the iso-DAG dose as such in liver at 3 hr reflects its actual presence there in relatively high concentrations, which in turn reflects (a) its presence in perfusate at relatively high concentrations and (b) its relative resistance to hydrolytic enzymes.
The identity of hepatic protein targets for acyl glucuronide-derived adduct formation, and the question of any biological consequences to this binding, are of considerable interest. Major protein targets at 110, 140, and 200 kDa (in order of intensity of staining of immunoblots) have been found in the livers of rats dosed with diclofenac (Hargus et al., 1994), zomepirac (Bailey and Dickinson, 1996), and DF (Bailey and Dickinson, 1996). In the case of diclofenac studies, the 110-kDa band has been identified as dipeptidyl peptidase, a plasma membrane enzyme whose activity was lowered after treatment of animals with diclofenac (Hargus et al., 1995). It should be noted, however, that other workers found a different pattern of hepatic protein targets for diclofenac acyl glucuronide-derived adduct formation (Kretz-Rommel and Boelsterli, 1993, 1994b). More work is required to clarify the reasons for these differences (e.g. are they due to differences in specificity of antibodies and/or differences between treatments: whole animals/organs vs. isolated/cultured hepatocytesvs. liver homogenates/subcellular fractions). In the present liver perfusion study, the 110-kDa protein was the major band stained with our polyclonal DF antibody. In addition, another band, not stained in control livers, was observed at ca. 120 kDa (fig. 6). This band was obscured in earlier immunoblots (Bailey and Dickinson, 1996) because of the intensity of staining of the 110-kDa protein. Bands at 140 and 200 kDa, attributable to the DF/DAG/iso-DAG treatments, were not observed. This may indicate that adducts with these proteins are formed more slowly than permitted during the rather acute exposure time (3 hr) of the perfused liver preparation. There were no major differences in the patterns of staining after the three treatments, and thus no additional mechanistic insights (e.g. differences in preferred protein targets for glucuronide vs. isomers) could be gleaned from these results.
In summary, DAG is a reactive acyl glucuronide metabolite that is disposed in the perfused rat liver by hydrolysis, rearrangement, and covalent binding to protein (in order of quantitative importance). Hydrolysis leads to avid deconjugation-conjugation cycling. Rearrangement yields isomers that undergo novel metabolism (via phenolic glucuronidation) and that contribute extensively (via glycation) to covalent drug-protein adduct formation. Contributions to adduct formation are also likely directly from DAG (via transacylation).
Acknowledgments
We thank Mr. Andrew King for technical advice concerning the isolated perfused rat liver preparation and the preparation and purification of diflunisal conjugates; we also thank Mr. Mark Bailey for technical advice concerning SDS-PAGE and immunoblotting.
Footnotes
-
Send reprint requests to: Dr. Ron Dickinson, Department of Medicine, Clinical Sciences Building, Royal Brisbane Hospital, Qld 4029 Australia.
-
This work was supported by a project grant from the National Health and Medical Research Council of Australia.
-
↵2 M. Wang and R.G. Dickinson, unpublished data.
- Abbreviations used are::
- DF
- diflunisal
- DAG
- diflunisal acyl glucuronide, iso-DAG, mixture of isomers of diflunisal acyl glucuronide formed by acyl migration
- DPG
- diflunisal phenolic glucuronide
- DS
- diflunisal sulfate
- D-2G
- mixture of phenolic glucuronides of isomers of diflunisal acyl glucuronide
- BSA
- bovine serum albumin
- ECL
- enhanced chemiluminescence
- TBS
- Tris-buffered saline
- SDS-PAGE
- sodium dodecyl sulfate-polyacrylamide gel electrophoresis
- Received May 27, 1997.
- Accepted October 7, 1997.
- The American Society for Pharmacology and Experimental Therapeutics